Glycoproteins in probiotic bacteria

Exploring the glycosylation potential of Lactobacillus rhamnosus GG by genomics and

glycoproteomics

Hanne TYTGAT

april 2015

Supervisors:Prof. J. Vanderleyden, KU LeuvenProf. S. Lebeer, UAntwerpProf. J. Winderickx, KU Leuven

Examination Committee:Prof. T. Niewold, ChairmanProf. D. Schols, KU LeuvenProf. B. Sels, KU LeuvenProf. K. Laukens, UAntwerpProf. W.M. de Vos, Wageningen University & UHelsinki

Dissertation presented in partial fulfilment of the requirements for the degree of Doctor in Bioscience Engineering

© 2015 KU Leuven, Science, Engineering & Technology & UAntwerp

Uitgegeven in eigen beheer, Hanne Tytgat, Leuven

Alle rechten voorbehouden. Niets uit deze uitgave mag worden vermenigvuldigd en/of openbaar gemaakt worden door middel van druk, fotokopie, microfilm, elektronisch of op welke andere wijze ook zonder voorafgaandelijke schriftelijke toestemming van de uitgever.

All rights reserved. No part of the publication may be reproduced in any form by print, photoprint, microfilm, electronic or any other means without written permission from the publisher.

Doctoraatsproefschrift nr. 1256 aan de faculteit Bio-ingenieurswetenschappen van de KU Leuven

voor oma.

Research was funded by the Agency for Innovation by Science and Technology in Flanders (IWT)

index.

inde

x.

I

index.

ii dankwoordvi abstractviii samenvattingx list of abbreviations

1 one. outline

11 two. the sweet tooth of bacteria: common themes in bacterial glycoconjugates

65 three. a network-based approach to identify substrate classes of bacterial glycosyltransferases 97 four. endogenous biotin-binding proteins: an overlooked factor causing false positives in streptavidin-based protein detection

107 five. systematic exploration of the glycoproteome of a beneficial gut microbe

133 six. glycosylated heterotrimeric pili of L. rhamnosus GG mediate DC-SIGN interaction

157 seven. general discussion

178 references202 addendum204 list of publications

II

ackn

owle

dgm

ents

DankwoorD

Het begon allemaal bijna tien jaar geleden, met de keuze om biochemie en biotechnologie te gaan studeren aan de KU Leuven. Een logische keuze: ik vond zowel biologie als chemie boeiend en leuk. Biochemie is gelukkig zoveel meer dan een som van de delen: een wereld van genen, cellen en pathways ging voor mij open. En wanneer het thuisfront vroeg wat ik daar dan later zou mee gaan doen, antwoordde ik stellig: forensisch onderzoeker. Hoewel dit me nog altijd boeit, ben ik vandaag blij dat het anders is gelopen. Dit werk is het resultaat van vier en een half jaar hard werk, met ontelbare ups en downs, bloed, zweet en tranen. Maar het was bovenal een periode waarin ik me volop kon wijden aan iets wat me boeit, met ontelbare leuke momenten, toffe ontmoetingen en leerrijke samenwerkingen.

Zonder de hulp van een heleboel mensen stond ik hier vandaag niet en ik wil van de gelegenheid gebruik maken om jullie te bedanken. Allereerst wil ik Jos en Sarah bedanken. Bedankt om voor jullie te mogen werken, me zoveel te leren en voor alle kansen die jullie me boden. Jos, bedankt voor je aanstekelijk enthousiasme voor onderzoek en om me de vrijheid te geven om mijn wilde plannen waar te maken. Je hebt een onschatbare rol gespeeld in het aanwakkeren van mijn passie voor wetenschap. Sarah, ik wil je graag bedanken voor al je tijd en werk die je in mijn project hebt gestoken. Ik weet dat ik zeker niet altijd de makkelijkste was, maar jouw aanmoedigingen en opmerkingen hebben me gedreven om telkens het onderste uit de kan te halen. Ik kan alleen maar hopen dat ik het onderzoek dat jij startte waardig heb voortgezet. Bedankt voor alles. Ik had het geluk om nog een derde promotor te hebben: Joris, een welgemeende merci! Het is uiteindelijk geen gistonderzoek geworden, maar ik weet dat ik altijd op je steun en hulp kan rekenen. Ik ben ongelooflijk trots op wat ik heb mogen realiseren voor o.a. iGEM en BioSCENTer. In deze reeks van mensen die van cruciaal belang geweest zijn voor mijn doctoraat, wil ik ook zeker Sigrid aanhalen. Sigrid, ik kijk ongelooflijk op naar jou en je werk. Je onuitputtelijke energie en daadkracht zijn een blijvende bron van inspiratie voor me. Ik hoop dat we ook na dit avontuur contact blijven houden.

Ik wil ook graag mijn voorzitter en de jury bedanken voor de interesse waarmee ze mijn werk gelezen hebben. Prof. Kris Laukens, bedankt voor de tips over de systematische assays, ze hebben me zeker geholpen om het één en ander in perspectief te plaatsen en beter te funderen. Prof. Dominique Schols, dank u wel voor uw enthousiasme en voor de tips ivm het pili werk. Prof. Bert Sels, wie had kunnen denken dat dit werk organosynthetische capaciteiten bezat? Ik alvast niet, bedankt om me dit te doen inzien. Last but not least, Prof. Willem M. de Vos, bedankt om in mijn jury te zetelen. Uw kennis over microbiologie lijkt onmetelijk te zijn, mijn hoofd tolt er soms een beetje van. Maar ik heb nog nooit van iemand zoveel bijgeleerd op zo’n korte termijn, het is dan ook een ware eer en genoegen om onder uw begeleiding aan een nieuw avontuur in Wageningen te beginnen. Bedankt voor uw kritische noten bij mijn

ackn

owle

dgm

ents

.

III

resultaten, de goede samenwerking, interessante discussies en de leuke mini sightseeing ritjes in Wageningen en Helsinki.

Glycobiologie is een relatief nieuw onderzoeksveld. Ik wil graag IWT Vlaanderen bedanken om potentieel te zien in dit project en dit financieel te ondersteunen. Zonder de steun van de Projectfinanciering (PF3M100234) had ik dit werk niet kunnen uitvoeren. Hier wil ik ook zeker Prof. Jan Balzarini, tevens assessor van mijn project, bedanken. De financiële steun van ASL, FWO en UAntwerpen hielp me bovendien om de halve aardbol af te reizen om bij te leren over glycobiologie.

Inzichten verkrijgen in bacteriële glycoconjugaten vergt een multidisciplinaire aanpak en ik heb tijdens dit doctoraat de hulp gekregen van fantastische mensen. Prof. Kathleen Marchal and Dr. Aminael Sanchez-Rodriguez, thank you very much for the collaboration on the in silico approach, I simply could not have done this without you and the results have really helped me throughout this work. Prof. Els Van Damme zonder uw lectines had dit werk er een pak magerder uit gezien. Prof. Ruddy Wattiez and Dr. Baptiste Leroy, I’m really happy that I could count on you for all the mass spectrometry. Baptiste, merci beaucoup pour la collaboration agréable! Also a large thanks to the laboratory of Prof. de Vos at UHelsinki. Especially to François Douillard and Pia Rasinkangas for the nice brainstorms and fruitful collaboration on the SpaCBA pili glycosylation and SecX story. I also want to thank Marcel Messing at the Haartman Institute for his help during my first stay in Finland. Prof. Paul Proost, bedankt om alle Edman-degradaties uit te voeren. Prof. Theo Geijtenbeek en Nienke van Teijlingen, ik wil jullie enorm hard bedanken voor de fijne en leerrijke tijd die ik mocht doorbrengen in jullie groep. Prof. Manfred Wuhrer en Kathrin Stavenhagen, I learnt some valuable lessons on glycan mass spectrometry and want to thank you for the nice brainstorms.

Intussen is het CMPG al sinds 2008 een plek waar ik met veel plezier kom werken. Bedankt aan alle collega’s van alle groepen voor de fijne momenten, closing times, laboweekends… Ook alle sportactiviteiten (frisbee, lopen, badminton, zwemmen, squashen) zorgden voor onvergetelijke momenten. Nooit had ik bij aanvang van mijn doctoraat durven denken dat ik nu zelfs zou genieten van 10 km te lopen. Bedankt iedereen voor de fijne jaren en alle mooie herinneringen. Het CMPG zou het CMPG niet zijn zonder Anita, JosD en André. Jullie leiden alles in goede banen en ik denk dat we allemaal niet genoeg beseffen hoe jullie ervoor zorgen dat wij zonder zorgen aan wetenschap kunnen doen. Merci!

Het werk in dit boekje was niet tot stand kunnen komen zonder de hulp van Geert en Tine. Ik waardeer jullie hulp enorm en wil jullie beiden van harte danken voor alles wat ik van jullie mocht bijleren. Jullie waren van onschatbare waarde voor dit werk. Bedankt om altijd te willen antwoorden op mijn soms ‘blonde’ vragen en te willen inspringen wanneer ik iets te veel hooi op mijn vork nam. I want to thank all current and former members of the Probio group for all the scientific and less scientific moments we shared over the years. Mariya, thanks for all your tips and all the

IV

ackn

owle

dgm

ents

fun moments we shared. Elke, het was echt leuk om met je samen te werken en om je er bij te hebben in Londen en Groningen. Cynthia, as your Rubia, I want to thank you for all the fun and nice talks. Elke, Mariya, Tine, Cynthia, Marijke and your husbandies (to be), thank you so much for making our India trip unforgettable! Shweta, thanks for being there, the talks inbetween work, the India trip, the friendship, our trips to NY and Seattle, London and Cork. Thanks for having us at your wedding. We will see each other soon in India or Belgium! That’s a promise. Ingmar, bedankt om mijn thesisbegeleider te zijn en de samenwerking voor de Kinderuniversiteitsessies. Op mijn beurt mocht ik ook drie thesisstudenten begeleiden: bedankt voor al jullie hulp en jullie interesse in mijn onderzoek. Ik wil vooral Anneleen bedanken om me zo goed en enthousiast te helpen bij alle experimenten en in te springen wanneer ik op buitenlandse stage was.

Marijke, ook ik vind het moeilijk om onze tijd in enkel zinnen samen te vatten. We hebben zoveel momenten, frustraties, tripjes en triomfen gedeeld, waar moet ik beginnen? Bedankt voor alle congrestripjes en alle momenten voor en na de congressen met de shoppingtrips, wijnbarretjes, etentjes, Madrileense jogging, sightseeing… Zonder jou waren deze vier jaren lang niet zo leuk geweest. Bedankt voor onze vriendschap en de ontelbare mooie herinneringen.

Geen Probio zonder Salmo: bedankt David, Ami en Kai voor alle hulp, tips en antwoorden op mijn vragen. Een oprechte dank u wel aan alle vroegere en huidige leden van S&P voor de leuke sfeer. Een speciale dank aan Ansie, Seppe, Akanksha, Hans, Sandra, Stijn en Elke. Het CMPG is echter meer dan S&P alleen. Maarten, bedankt om ons op te vangen in je éénmanslab, alle nuttige discussies en alle leuke momenten daar in ons lab beneden. Pietje, merci om mijn bureaupartner te zijn in het ‘kiekenkot’, je luisterend oor en alle steun, zeker toen ik de tweede keer naar het IWT moest. Merci! Mikey/Kim, dank je om er te zijn en te luisteren, de sportieve momenten, de nuttige discussies en de terrasjes. Veerle, bedankt voor alle leuke gesprekken en de fijne trip naar Rome. Dank je ook Bram, Annelies, Maarten, Natalie en Toon!

Tijdens mijn doctoraat kreeg ik ook de kans om me toe te leggen op wat minder wetenschappelijk getinte nevenprojecten. Sofie Boons, ik ben nog altijd heel trots op ons project. Ik heb ongelooflijk veel bijgeleerd en heb heel hard genoten van onze samenwerking. Wie weet doen we ooit nog eens iets samen? Ook een dank u wel aan Karen Verschooren om me mee te trekken in het ‘Alter Nature’ avontuur! Het was een leuke en verrijkende uitdaging voor me. Cheryl, thanks for getting me involved in DLI. I was not able to contribute as much as I would have wanted to, but I’m still a heavy supporter of your dream and our cause.

Zonder mijn vrienden waren het maar saaie jaren geweest. Bedankt allemaal om er altijd te zijn voor mij en alle onvergetelijke momenten. Lies, Sven, Isabelle, Jeroen, Jelle en Greet, zonder jullie was mijn tijd in Leuven niet half zo plezant geweest. Bedankt voor alle etentjes, tripjes in binnen- en buitenland en terrasjes. Bedankt voor alles. Nico, bedankt om er altijd voor me te zijn, voor Thailand, Amsterdam en onze Amerika roadtrip

ackn

owle

dgm

ents

.

V

door Amerika. Samen met Michiel, Jinne en Lorin zorgde je voor de nodige ontspannende feestjes, pokeravonden en etentjes. Merci mannen! Annelies, je kent me waarschijnlijk al het langst van iedereen. Ik vind het geweldig dat we er altijd zijn voor mekaar, zelfs al is dat fysiek niet zo. Bedankt voor alle telefoontjes, reisjes, etentjes, cocktails en shoppingsprees! Pieter, merci om me te blijven motiveren om te lopen en voor alle avonden dat we welkom waren bij jou thuis. Maar vooral, dit doctoraat en mijn papers hadden er niet half zo mooi uitgezien zonder jouw Photoshop kunsten. Bedankt dat ik altijd op je mag rekenen, dat je altijd paraat staat met tips en tricks en om mijn privéleraar ‘grafiek’ te zijn. Eline, het is heel nipt geweest, maar ik ben dan toch nog eerder Dr. geworden dan jij. Merci voor alle cocktails, onze mislukte Parijs trip, wandelingen in de sneeuw (ik beloof dat ik zal proberen Kerst terug te leren appreciëren) en zoveel meer. Wouter, Jef, Bieke, Samme, Ilse en Christiaan: merci voor alle Extravanganza-uitjes! De momenten met onze hele bende zijn altijd chaotisch, maar o zo leuk! Ik kijk al uit naar onze volgende avonturen! (Anne)Lies, we zien elkaar niet veel, maar je weet wat onze vriendschap voor me betekent.

Ik zou graag ook mijn familie en schoonfamilie bedanken. Bedankt voor jullie interesse in wat ik uitspookte hier in Leuven. Opa, Jan en tante Isabelle, bedankt om er altijd te zijn voor mij. Oma, pepe André, meme Paula en nonkel Roger, ik vind het ongelooflijk spijtig dat jullie er niet bij kunnen zijn, maar ik weet dat jullie trots zijn op jullie Hanneke. Ik mis jullie nog elke dag. Papa en mama, zonder jullie stond ik vandaag niet waar ik nu sta. Jullie hebben me gesteund en gepusht om het beste uit mezelf te halen. Ik hoop dat ik jullie trots heb kunnen maken.

Lim, voor ik je kende was alles heel zwart-wit, jij bracht grijs in mijn leven. Jij bent het allerbelangrijkste in mijn leven en ik ben TedxFlanders nog elke dag dankbaar dat we elkaar daar hebben mogen ontmoeten. Ik kan niet wachten om te beginnen aan onze avonturen samen en bedank je om me te steunen bij al mijn wilde plannen. Samen kunnen we alles aan. Bedankt dat ik je maske mag zijn, om er altijd te zijn, om me altijd te steunen. Ik hou van je.

Doctoreren, wat een ervaring. Ik heb ongelooflijk veel bijgeleerd en hard genoten van elk moment. Bedankt aan iedereen die er voor me was, me geholpen heeft en bijgedragen heeft aan al die leuke herinneringen. Bedankt ook aan iedereen die ik ben vergeten op te noemen of niet expliciet vermeld heb.

Op naar nieuwe avonturen…

Hanne

- Playing it safe always ends in disaster -

VI

abst

ract

abstract

Humans live in close contact with a huge amount of bacteria, i.e. the microbiota, of which the gut microbiota is the best documented. The better we get to know this bacterial load, the more imbalances of the microbiota are linked to a plethora of diseases. Probiotics, defined as ‘live microorganisms which, when administered in adequate amounts, confer a health benefit to the host’ can be used to prevent and influence such dysbiosis states. An underpinned use of probiotics requires a detailed understanding of the molecules underlying the specific interaction between these bacteria and their host. An important model probiotic strain and well-documented microbiota isolate is Lactobacillus rhamnosus GG, of which the clinical potential has already been widely studied and several key modulators of its probiotic function are known.

In this PhD work, we focus on an intriguing class of bacterial molecules, namely glycoproteins. In 2008, the first indications were found that the secreted Msp1 protein, a cell wall hydrolase and probiotic effector of L. rhamnosus GG, is glycosylated. These findings were confirmed by complementary techniques such as lectin blotting (i.e. (glyco)proteins specifically interacting with sugar moieties), specific glycostaining and mass spectrometry, and the first report on an L. rhamnosus GG glycoprotein was then published in 2012. Subsequently, we wondered whether L. rhamnosus GG glycosylates more of its proteins and whether protein glycosylation plays a biochemical role, is involved in bacterial physiology and/or governs host-microbe interaction functions. Glycans on proteins, but also on various other glycoconjugates, are generally used by bacteria to shield of their immunogenic molecules from the host immune system, but they can sometimes also function to enhance interactions with it, e.g. by targeting specific immune lectin receptors. Bacteria are found to produce an enormous diversity of these glycans, resulting in strain-specific bacterial barcodes. Moreover, glycosylation of proteins can modulate the biochemical properties of these substrates, like their activity and stability. Taken together, the study of glycoproteins of a Gram-positive beneficial microbiota isolate is thus both of high interest for the field of probiotics and microbiota and the field of bacterial glycobiology.

Glycosyltransferases (GTs) are the key enzymes of glycobiology, as they are involved in the build-up and attachment of glycans. Up till now, the exact factors that determine the substrate specificity of these enzymes remain to be elucidated, which often results in their flakey annotation in many bacterial genomes. Therefore, we designed an in silico workflow to screen bacterial genomes for genes encoding glycosyltransferases based on Hidden Markov Models typical for bacterial GTs (CAZy database and functionally characterized bacterial GTs). By relying on the functional network of the identified genes, we moreover could generate clues on their substrate specificity. Our results generated some interesting new insights and hypotheses, like the potential role of glycosylation in the regulation of the cell division machinery.

abst

ract

.

VII

Since the experimental substantiation of the glycosylation of bacterial proteins is very challenging, several technical difficulties were encountered and tackled during this PhD work. An important experimental approach was the use of lectin blotting experiments to detect sugars on proteins. Although biotinylated lectin probes that can be detected with streptavidin form the common standard for these analyses, we observed that streptavidin interacts with various bacterial proteins binding endogenous biotin, rendering false positive results. By applying an alternative detection method, using digoxigenin, we could drastically reduce the detection of false positive putative glycoproteins.

In addition, functional partition of the bacterial proteome in four experimental fractions, representing the cytosol, cell wall/membrane, cell wall-associated and secreted proteins, combined with 2D proteomics, lectin blotting and glycostaining, enabled us to map the putative glycoproteome of L. rhamnosus GG to different cellular locations. This resulted in the identification of 41 glycoproteins, of which most are involved in the carbohydrate metabolism.

To investigate the role of protein glycosylation on microbe-host interactions, we also studied the glycosylation status of the heterotrimeric SpaCBA pili, since these are key host interaction molecules of L. rhamnosus GG. Combining complementary techniques such as biochemical approaches with nanomechanical and immunological assays, we could show that these long flexible surface appendages are decorated with mannose and fucose. This unique multidisciplinary approach enabled not only the confirmation of the glycosylation of the SpaCBA pili, but also showed that the glycans present on the pili are key for the interaction between L. rhamnosus GG and the immune lectin receptor DC-SIGN present on dendritic cells.

Taken together, in this PhD work the protein glycosylation potential of L. rhamnosus GG was studied both relying on systematic and dedicated approaches. Several technical breakthroughs were achieved by proposing new robust workflows and optimizing known protocols. Both systematic approaches (one in silico and one proteomic) generated important new insights and hypotheses on the role of bacterial protein glycosylation, like the potential role of glycosylation in the role of multiprotein complexes (e.g. the cell division machinery). In addition, the more dedicated multidisciplinary approach on the elucidation of the glycosylation of the adhesive heterotrimeric SpaCBA pili of L. rhamnosus GG revealed important insights in the possible functional interaction roles of bacterial glycoproteins.

Conclusively, we believe that this work has enhanced both the knowledge on the importance of glycoproteins for the interaction of probiotics and members of the microbiota with the host and the knowledge on bacterial protein glycosylation in general.

VIII

sam

enva

ttin

g

samenvatting

De mens leeft in nauw contact met een enorme hoeveelheid bacteriën, met name de microbiota, waarvan de darmmicrobiota het best bestudeerd zijn. Hoe beter we deze bacteriën leren kennen, hoe duidelijker het wordt dat verstoringen van de microbiota kunnen leiden tot een rits aandoeningen. Probiotica, gedefinieerd als ‘micro-organismen, die wanneer ze in voldoende hoeveelheden worden toegediend, een gezondheidseffect kunnen hebben op de gastheer’, kunnen ingezet worden om deze dysbiose te voorkomen en te behandelen. Een onderbouwd gebruik van probiotica vereist een gedetailleerde kennis van de moleculen die aan de basis liggen van de specifieke interacties tussen de bacteriën en hun gastheer. Lactobacillus rhamnosus GG is een model organisme voor probiotica en werd oorspronkelijk geïsoleerd uit de microbiota. Het klinisch potentieel van deze bacterie is al uitgebreid bestudeerd en enkele belangrijke probiotische effectoren zijn al gekarakteriseerd.

In dit doctoraatsonderzoek focussen we op glycoproteïnen. In 2008 ontstonden de eerste vermoedens dat het gesecreteerde Msp1 proteïne, een celwandhydrolase en een probiotische effector van L. rhamnosus GG, geglycosyleerd is. De glycosylatie van Msp1 werd bevestigd met complementaire technieken zoals lectine blotting (dit zijn (glyco-)proteïnen die specifiek kunnen interageren met suikermonomeren), suikerkleuringen en massa spectrometrie. Die ontdekking deed ons o.a. afvragen of L. rhamnosus GG nog meer glycoproteïnen produceert en wat de functionele rol is van proteïnenglycosylatie. Dit laatste kan zowel gaan om de modulatie van de biochemische eigenschappen van het proteïne (bv. activiteit en stabiliteit), het beïnvloeden van de bacteriële fysiologie als het beïnvloeden van bacterie-gastheer interacties. Glycanen worden vaak door bacteriën gebruikt om immunogene moleculen af te schermen van het immuunsysteem van de gastheer of om net de interactie met de gastheer te bevorderen. Er is al aangetoond dat bacteriën een enorme diversiteit aan glycanen produceren, wat resulteert in een stam-specifieke barcode. Al deze factoren maken de studie van glycoproteïnen in een Gram-positieve gunstige stam waardevol, zowel voor het probiotica- en microbiota-onderzoek als in het kader van de verdere expansie van de bacteriële glycobiologie.

Glycosyltransferases (GTs), betrokken bij de opbouw en aanhechting van glycanen, zijn de sleutelenzymen van de glycobiologie. De factoren die de substraatspecificiteit van deze enzymen bepalen zijn tot nu toe onbekend, wat vaak resulteert in onbetrouwbare gen annotaties. Daarom ontwierpen we een in silico methode om bacteriële genomen te screenen voor genen die coderen voor glycosyltransferases gebaseerd op Hidden Markov Modellen typisch voor bacteriële GTs (CAZy database en functioneel gekarakteriseerde bacteriële GTs). Door het functioneel netwerk van de geselecteerde genen te bestuderen, konden we bovendien een zicht krijgen op hun mogelijke substraatspecificiteit. Dit resulteerde in enkele interessante nieuwe inzichten en hypotheses, zoals over de potentiële rol van glycosylatie in de regulatie van de celdelingsmachinerie.

sam

enva

ttin

g.

IX

Glycobiologie is, gezien het posttranslationele karakter ervan, een uitdagend onderzoeksveld. Tijdens dit doctoraatsonderzoek moesten oplossingen gevonden worden voor verschillende technische uitdagingen. Een belangrijke experimentele techniek is het gebruik van lectine blotting om suikers te detecteren op proteïnen. Hoewel gebiotinyleerde lectine probes die gedetecteerd kunnen worden met streptavidine de algemene aanpak vormen voor deze analyses, resulteerde dit in vals-positieve resultaten, aangezien streptavidine ook interageerde met proteïnen die endogeen biotine binden. Een alternatieve methode, gebruikmakende van digoxigenin, reduceert deze vals-positieve waarnemingen drastisch.

Bovendien konden we aan de hand van de functionele fractionatie van het bacteriële proteoom in vier fracties (cytosol, celwand/membraan, celwand geassocieerde en gesecreteerde proteïnen), gecombineerd met 2D proteomics, lectine blots en suikerkleuring, het glycoproteoom van L. rhamnosus GG in kaart brengen voor de verschillende fracties. Dit leidde tot de identificatie van 41 potentiële glycoproteïnen, waarvan de meeste een rol spelen in het suikermetabolisme.

Om de rol van proteïnenglycosylatie in bacterie-gastheer interacties te onderzoeken, bestudeerden we de glycosylatiestatus van de heterotrimere SpaCBA pili, aangezien dit belangrijke moleculen zijn voor de gastheerinteractie van L. rhamnosus GG. Een multidisciplinaire combinatie van biochemische, nanomechanische en immunologische technieken resulteerde in de detectie van mannose en fucose op deze SpaCBA pili. Bovendien konden we aantonen dat deze glycanen op de pili van belang zijn voor de interactie tussen L. rhamnosus GG en de immuun lectine receptor DC-SIGN, aanwezig op dendritische cellen.

In dit doctoraatsonderzoek werd het glycosylatiepotentieel van L. rhamnosus GG met behulp van systematische en gedediceerde methodes bestudeerd. Daarbij werden verschillende doorbraken geboekt op technologisch vlak, zoals het aanreiken van robuuste methodologieën en de optimalisatie van bestaande protocols. De beide systematische studies (in silico analyse en proteoomstudie) genereerden belangrijke nieuwe inzichten in de potentiële rol van glycosylatie van proteïnen in bacteriën. Bovendien toonde multidisciplinair onderzoek aan dat de SpaCBA pili van L. rhamnosus GG geglycosyleerd zijn, een belangrijk nieuw inzicht wat betreft de mogelijke functionele rol van bacteriële glycoproteïnen.

We kunnen stellen dat dit werk zowel de kennis van het belang van glycoproteïnen in probiotica en microbiota-gastheer interacties heeft verruimd, evenals de kennis over glycosylatie van bacteriële proteïnen in het algemeen.

X

list o

f abb

revi

atio

ns

List of abbreviations

AAL Aleuria aurantia lectinAFM Atomic force microscopyBSA Bovine serum albumineC CytoplasmicCAZy Carbohydrate Active enZymes databaseConA Concanavalin ACM Cytoplasmic membraneCPS Capsular polysaccharidesCWA Cell wall-associatedCWCM Cell wall/membraneCyt CytosolDC(s) Dendritic cell(s)DC-SIGN Dendritic Cell-Specific Intercellular adhesion molecule-3-Grabbing Non- integrinDIG DigoxigeninDMSO Dimethyl sulfoxideDSL Datura stramonium lectinDTT DithiothreitolEGTA Ethylene Glycol Tetra Acetic acidELLA Enzyme-Linked Lectin AssayEPS ExopolysaccharidesFITC Fluorescein isothiocyanateFuc FucoseFucNAc N-acetylfucosamineGADPH Glyceraldehyde-3-phosphate dehydrogenaseGal GalactoseGalNAc N-acetylgalactosamineGIT Gastrointestinal tractGlc GlucoseGlcNAc N-acetylglucosamineGNA Galanthus nivalis agglutininGRAS Generally Regarded As SafeGT(s) Glycosyltransferase(s)Hex HexoseHHA Hippeastrum hybrid agglutininHMM Hidden Markov ModelIAA Iodoacetic acidIM Inner mebrane

abbr

evia

tions

.

XI

LeX Lewis XLOS LipooligosaccharidesLPS LipopolysaccharidesLTA Lipoteichoic acidMAMP Microbe-Associated Molecular PatternMan MannoseMGL Macrofage galactose lectinmoDCs Monocyte-derived dendritic cellsOM Outer membraneOST OligosaccharyltransferasePAS Periodic Acid Schiff base PBS Phosphate-buffered salinePG PeptidoglycanPNA Peanut agglutininPVA PolyvinylalcoholPVDF Polyvinylidene difluorideRha RhamnoseRSA Rhizoctonia solani agglutininSBA Soybean agglutininSMFS Single-molecule force spectroscopySN SupernatantSP Secreted proteinsSRR Serine-rich repeatTA Teichoic acidTCA Trichloroacetic acidTM TransmembraneTBST Tris-buffered saline with TweenUDA Urtica dioica agglutininUndP Undecaprenyl phosphateUndPP Undecaprenyl pyrophosphateWGA Wheat germ agglutininWT Wild typeWTA Wall teichoic acid

one.

outline

a path through the maze

one.

outline

a path through the maze

2

outli

ne

1.1 backgrounD

1.1.1 Human after aLL? - microbiota anD probiotics

In this age we can start to ask ourselves how human we really are… The cells of our body are namely 10-fold outnumbered by the bacterial load we carry (1014). These bacteria occur in different niches in and on the body, like the skin, oral cavity, the nasopharynx, vagina, and their most important niche, the gastrointestinal tract (GIT) (The Human Microbiome Project Consortium, 2012). Currently, this bacterial load we all carry is being mapped, which has yet lead to the identification of several gut enterotypes (Arumugam et al., 2011), although the exact number and functional importance of these enterotypes is still debated on and continuously fine-tuned (Claes et al., 2014).

The microbiota and the host interact in general in a mutually beneficial way. The commensal bacteria in the gut for instance aid in the digestion of nutrients, protect against infections, promote angiogenesis and balance the immune system (Belkaid & Hand, 2014; Gareau et al., 2010). The microbiota thus plays an important role in host health. An imbalanced microbiota, or dysbiosis, is increasingly linked to several diseases, both gastrointestinal (e.g. inflammatory bowel diseases) and systemic (e.g. diabetes, obesitas) disorders, although the exact nature of this relation needs to be further determined (Belkaid & Hand, 2014; Cani & Delzenne, 2009; Tillisch, 2014).

This link between dysbiosis of the microbiota and diseases, lead to a peak in the interest in research on probiotics. Probiotics are ‘live microorganisms that, when administered in adequate amounts, confer a health benefit on the host’ (FAO/WHO, 2001). Rationale is to use microorganisms to modulate the microbiota in order to prevent and treat diseases linked to dysbiosis of the microbiota. Most probiotic strains are lactobacilli and bifidobacteria, i.e. endogenous members of a healthy microbiota, but also other strains are described, like Escherichia coli Nissle 1917 and the yeast Saccharomyces boulardii. Growing numbers of clinical trials support the view that certain bacterial strains or mixtures of strains can indeed influence the host health. But, not all results of intervention studies are evenly positive and the outcome of probiotic treatments in different pathologies is very difficult to predict (Segers & Lebeer, 2014).

In our groups, research is focused on probiotic effectors, mostly present on the bacterial surface or secreted by the bacteria, to get a better understanding of the molecules modulating the interaction with the host (Lebeer et al., 2008). We believe that a thorough molecular knowledge of the microbiota-host interaction is needed to enable a more substantiated usage of probiotics to prevent and treat diseases linked to dysbiosis.

1.1.2 LactobaciLLus rhamnosus gg: microbiota isoLate anD moDeL probiotic

In the labs of Prof. J. Vanderleyden (KU Leuven) and Prof. S. Lebeer (UAntwerpen), the

one.

3

molecules on the cell surface and secreted by Lactobacillus rhamnosus GG (ATCC 53103) are being investigated. This strain has been widely studied in clinical and intervention studies and is considered to be a model probiotic (Segers & Lebeer, 2014).

L. rhamnosus GG was originally isolated from a healthy microbiota by Gorbach and Goldin (hence the GG). L. rhamnosus GG was already proven to be efficient to prevent and treat a plethora of disorders like diarrhea (e.g. travellers diarrhea), atopic diseases (e.g. atopic dermatitis) and caries (Doron et al., 2005; Segers & Lebeer, 2014). However, the exact nature of the molecular interactions underlying its probiotic function remains to be fully understood.

Several key molecules present on the cell surface and secreted by L. rhamnosus GG have been studied in our labs to elucidate their structure and role in the interaction with the environment and host. The cell wall consists of a thick peptidoglycan layer, on which the SpaCBA pili (Kankainen et al., 2009; Lebeer et al., 2012a; Reunanen et al., 2012), exopolysaccharides (Francius et al., 2008; Lebeer et al., 2011; Lebeer et al., 2009) and specific (glyco)proteins are anchored. There are also lipoteichoic acids present, which are embedded in the cell membrane (Claes et al., 2010; Perea Velez et al., 2007). L. rhamnosus GG also produces two major secreted proteins, Msp1 and Msp2 (Claes et al., 2012; Lebeer et al., 2012b). In this PhD thesis, the focus lays on the glycoconjugates present on the cell wall of L. rhamnosus GG and more in particular on the glycoproteins produced by this strain.

1.2 scope

In 2012 our groups published the first report that the probiotic effector protein Msp1 of L. rhamnosus GG is glycosylated (Lebeer et al., 2012b). This formed the onset of the present PhD work, in which we aimed to explore the full protein glycosylation potential of this model probiotic strain.

Bacterial glycoproteins are widely explored in Gram-negative pathogens, but knowledge on their occurrence and role in Gram-positive beneficial bacteria is still scarce (Tytgat & Lebeer, 2014). Nevertheless, some important general trends strengthen the belief that glycoproteins and glycoconjugates in general are key molecules both modulating bacteria-host interactions and having special biochemical properties. These are thus important targets to study both in Gram-negatives and –positives and both in pathogens and beneficial species. Studies in pathogens have already shown that bacteria glycosylate immunogenic molecules to either enhance interaction or either to render them ‘stealth’ from the immune system. Moreover, bacteria were found to produce an overwhelming diversity of glycans, making them ideal candidates to establish these very specific interactions with their environment. And a last important finding, already well-substantiated for glycoproteins, is that the glycosylation of substrates modulates the biochemical parameters of these proteins, by influencing features like their activity and stability (Tytgat & Lebeer, 2014).

4

outli

ne

Taken these intriguing potential roles of glycoproteins together, it is key to elucidate the occurrence and role of these structures in beneficial bacteria. L. rhamnosus GG, being a microbiota isolate and a model probiotic strain, is an ideal organism to focus on, as it glycosylates one of its probiotic effectors, Msp1 (Lebeer et al., 2012b). Studying the glycosylation potential of this strain would thus not only push the field of microbiota-host interactions further, but would also widen the knowledge on bacterial glycosylation processes in beneficial, Gram-positive species.

1.3 experimentaL cHaLLenges in bacteriaL gLycoprotein anaLysis

In this work we will focus on gaining insights in the protein glycosylation potential of L. rhamnosus GG. In order to do so, a complementary set of tools is needed, as the non-template driven character, the enormous diversity of glycosylation processes in bacteria and the only fragmentary documentation of protein glycosylation processes in bacteria make this a particularly challenging field to study. Advances in (protein) glycobiology thus require a multidisciplinary approach and the combination of various systematic and dedicated approaches. This is exemplified in this work in which experimental and in silico analyses were combined. In this first chapter, we aim to briefly situate the current and future methodological possibilities, techniques and strategies to unravel the protein glycosylation potential and function in bacteria.

First, a few (mainly predictive) analyses can be done at the DNA level, both at the gene and genome level. Bioinformatics tools can offer valuable insights into the glycosylation potential of a strain, by predicting glycan-processing enzymes. Of special interest are the glycosyltransferases (GTs), which are the key enzymes of glycosylation (Lairson et al., 2008). The identification of these enzymes offers important leads for further analyses. But, as we illustrate in chapter 3, also other information can be inferred from these predictions: for instance, their genomic localization can generate clues on the nature of the glycosylation mechanism they are involved (Sanchez-Rodriguez et al., 2014). As we elaborate in more detail in chapter 2, the biosynthesis of glycoconjugates occurs via two main mechanisms: en bloc and sequential biosynthesis of glycoconjugates. The former is often characterized by a clustering of the genes for all involved enzymes, transporters and regulators in a dedicated operon, whilst the enzymes of the latter mechanism are rather scattered throughout the genome (Tytgat & Lebeer, 2014). This moreover renders insights in the potential underlying evolutionary processes.

The conservation amongst glycosyltransferase structures enables their classification into glycosyltransferase families, like in the Carbohydrate Active enZymes database (CAZy, www.cazy.org). This can be further exploited using bioinformatics tools, based on the use of Hidden Markov Models (HMM) representing these CAZy families, to mine genomes for genes encoding glycosyltransferases. Such a strategy has been exemplified in the model organism

one.

5

of prokaryotic glycosylation, Campylobacter jejuni (Kay et al., 2010) and for Arabidopsis plants (Hansen et al., 2010; Hansen et al., 2009). The power of in silico methods is uncannily, but these methods do come with some important caveats and limitations. Machine-learning techniques and other in silico tools are based on the extrapolation of current knowledge, which may stand in the way of new discoveries. This is of special importance in the still rather young field of glycobiology, where lots remains to be discovered. Also, the non-template driven character of this field might call for more specialized strategies, as bioinformatics tools often cannot make decisions on the importance of different kinds of input.

The identification of genes putatively encoding glycosyltransferases can also offer leads on the genes of interest to study via mutagenesis approaches. This enables the construction and study of knockout or overexpression mutants, which can provide clues on the functional role of the enzymes or entire pathway (for instance for operons involved in en bloc biosynthesis) and ideally on the substrate (e.g. by investigating which glycoconjugates are modulated in these mutants). Since standard bacterial cloning hosts such as Escherichia coli appear not to glycosylate proteins (Messner, 2004), an interesting approach to validate the link between a glycosyltransferase enzyme and its protein substrate is their heterologous expression in such a non-glycosylating host. This as was first illustrated in the benchmarking paper of Wacker et al., who transferred the N-protein glycosylation system of Campylobacter jejuni into E. coli (Wacker et al., 2002), which is especially relevant for Gram-negative bacteria. The analysis of GTs via mutagenesis comes with some important caveats: GTs might be essential for a bacterium, manipulation of a GT can affect the fitness of a cell drastically, not all bacteria can be manipulated genetically and the role of a mutated GT might be compensated by another GT (cf. chapter 2).

Once the genomic potential for glycosylation, i.e. a compendium of glycosyltransferases, is known, it also will be interesting to study the actual expression of these enzymes and proteins associated to glycosylation processes (e.g. transporters). Transcriptome analyses, like RNA-seq and microarrays are powerful tools to achieve this. Transcriptome data was therefore also implemented in the in silico method we designed in chapter 3: the STRING database is used there to provide information on the interaction partners of the predicted GTs of L. rhamnosus GG (Sanchez-Rodriguez et al., 2014). This STRING database gathers all information on predicted and proven interactions between proteins, including all available microarray data (von Mering et al., 2005). Moreover, the conditions in which certain enzymes, like GTs, are expressed can be monitored by differential transcriptomics. To the best of our knowledge, this approach has not yet been widely applied in the field of bacterial glycoprotein research, but might provide interesting clues to further elucidate the glycosylation potential of a bacterium.

Although DNA- and RNA-research can provide valuable information, proteomics and glycoproteomics are the methods of choice if one wants to investigate the glycoprotein repertoire directly. The two most commonly used techniques of proteomic analyses, i.e.

6

outli

ne

SDS-PAGE and Western blot, are also important workhorses in glycoproteome screenings. In chapter 2, an overview is given of the current knowledge on bacterial glycoconjugates illustrating the widespread use of SDS-PAGE and Western blot related techniques to study the glycosylation of proteins. An overview of many studies relying on these techniques is given in chapter 2, where all known glycoproteins relevant to human health are discussed (Tytgat & Lebeer, 2014).

Periodic Acid Schiff staining (PAS) of proteins separated by SDS-PAGE is an almost ancient but still commonly used method both in pro- and eukaryotes. This approach is the method of choice to visualize all glycoproteins in a sample (Fairbanks et al., 1971). The application of periodate oxidizes the glycol groups resulting in aldehyde formation. These aldehydes then can react with fuchsin to generate a pink coloring reaction. Drawbacks of this method are the low sensitivity (1 μg of glycoprotein) and false positives that can occur if the protein contains a lot of aromatic amino acids. To tackle the first issue, a commercial version of the PAS stain was marketed, in which the fuchsin was replaced by a fluorescent dye. The detection limit of this Pro-Q® Emerald dye (Life Technologies) is as little as 4 ng of glycoprotein (Hart et al., 2003).

Another important method to screen for glycoproteins after SDS-PAGE separation and blotting of protein samples, is the probing of Western blots with lectins. Lectins are (glyco)proteins that specifically recognize a sugar monomer in a specific configuration and conformation (Van Damme et al., 2011). This implicates that different lectins recognizing the same sugar monomer (e.g. GNA and HHA which both bind mannose), still will render different results when applied to the same sample. Lectins are often applied to generate clues on the sugar monomers present in a particular sample, but can thus also render false negative results, if a certain sugar monomer is not accessible in the right configuration or conformation. False positive results are also possible, as illustrated further in chapter 4, in which we report on some necessary tweaks to the design of a lectin experiment to prevent false positive reactions. These adaptations include the careful selection of the blocking agent and detection method (Tytgat et al., 2014). Lectins can also be used to generate functional clues on the role of glycans by the construction and screening of lectin microarrays. This technique was first described by the groups of Angeloni and co-workers (Angeloni et al., 2005) and by Hsu et al. who evaluated the potential of the technique to follow the glycome dynamically (Hsu et al., 2006).

Lectins can be further exploited to purify glycoproteins and glycopeptides, a method, which was for instance applied earlier in our lab to purify the Msp1 glycoprotein (Lebeer et al., 2012). Other purification techniques are HPLC (High-Performance Liquid Chromatography) techniques like HILIC (Hydrophilic Interaction Liquid Chromatography) and RP-HPLC (Reversed Phase-HPLC). These purification techniques are combined with mass spectrometry strategies to identify glycoproteins, investigate glycopeptides and determine the structure of the attached glycan, as was done for instance in C. jejuni (Ding et al., 2009). A good overview

one.

7

of this field is given in: (Hitchen & Dell, 2006; Marino et al., 2010; Schiel, 2012). The elucidation of glycan structures often relies on the release of the glycans from the peptide. To do so, both chemical and enzymatic techniques can be used. Chemical deglycosylation techniques, like the application of trifluoromethane sulfonic acid (Raju & Davidson, 1994) are often harsh, degrading both the protein and glycan, but they have the advantage that they can remove almost any glycan (both O- and N-linked) from any protein. Most enzymes used (e.g. PNGase F) only recognize eukaryotic glycoproteins or N-linked glycans (Marino et al., 2010), whilst the latter form of glycosylation is thought to be scarce in bacteria (Tytgat & Lebeer, 2014). To elucidate the 3D structure of a glycan or glycoprotein in detail, one can rely on NRM and X-ray crystallography (Kay et al., 2010; Marino et al., 2010).

At the moment information on the functional role of bacterial glycoproteins is still extremely scarce, explaining the lack of standard methods to investigate the functional role of protein glycosylation in a high-throughput manner. Nevertheless, as we illustrate in chapter 6, these studies ideally require the purification of intact glycoproteins and the availability of knockout or overexpression mutants affecting the glycosylation of the target protein (or chemically/enzymatically altered glycoproteins). Moreover, there is a need for validated and standardized phenotypic and biochemical assays to investigate the functional role of the glycan present on a certain protein. But some work already has been performed, mainly on the biochemical impact of glycans on their substrate, as illustrated for many in chapter 2, where the glycans for instance protect the protein from degradation by proteases (Tytgat & Lebeer, 2014). In our lab it was already shown that the Msp1 glycoprotein is more stable compared to its less glycosylated close homologue Msp2 (Lebeer et al., 2012). And in the closely related L. plantarum WCFS1 strain glycosylation modulates the activity of the Acm2 enzyme, which is homologous to Msp1 (Fredriksen et al., 2012).

We can conclude that the glycomics toolbox consists out of a variety of techniques, often requiring, especially in the case of MS-strategies, the expertise of a specialized group. For an extensive overview of all currently used techniques in glycosciences, the reader is referred to the following review: (Solis et al., 2015).

1.4 a guiDe tHrougH tHe cHapters

This PhD manuscript is a compilation of three peer-reviewed articles, a submitted manuscript and a manuscript in preparation. In this outline (chapter 1) on overview of these papers is given and how they fit in the general scope of this PhD, namely the elucidation of the protein glycosylation potential of the model probiotic L. rhamnosus GG.

To initiate this work, a thorough and comprehensive literature study of the current knowledge on bacterial glycoconjugates was carried out. Strikingly, only a few common general pathways are used by bacteria to synthesize the plethora of glycoconjugates they express. In this work we only focus on the important lessons that can be learnt from bacterial glycoprotein research

8

outli

ne

(chapter 2). For the extended work covering all existing glycoconjugates, the reader is referred to the complete version of the review (Tytgat & Lebeer, 2014).

Chapter 3 is dedicated to the mining of the genome of L. rhamnosus GG for genes encoding glycosyltransferases, the key enzymes of glycosylation processes. This bioinformatics approach was used to generate clues on interesting genes to focus on during later stages of this work by e.g. the construction of deletion mutants lacking glycosyltransferases. As glycosylation is a non-template driven process, the link between these enzymes and their substrates is still enigmatic. In this work we relied on a network-based in silico approach to elucidate potential targets of these glycosyltransferases. Moreover we present some interesting new hypotheses on various roles for glycosylation in bacteria (Sanchez-Rodriguez et al., 2014).

Since the experimental substantiation of the bacterial protein glycosylation is very challenging, several technical difficulties were encountered and tackled during this PhD work. An important glycobiology approach is the use of lectin blotting experiments to detect sugars on proteins. Although biotinylated lectin probes that can be detected with streptavidin form the common standard for these analyses, we observed that streptavidin interacts with various bacterial proteins binding endogenous biotin, rendering false positive results. In chapter 4, we describe that by applying an alternative detection method, using digoxigenin, we could drastically reduce the detection of false positive putative glycoproteins (Tytgat et al., 2014). Other optimizations of techniques are mentioned in chapter 5.

The optimization of techniques (cf. chapter 4) enabled the systematic study of the glycoproteome of L. rhamnosus GG after fractionation in different cellular fractions (cytosol, cell wall/membrane, cell wall-associated and secreted proteins), combined with 2D proteomics, lectin blotting and glycostaining. Chapter 5 reports on the design of a specific workflow to do so, which resulted in the identification of 40 new glycoproteins in L. rhamnosus GG, next to the confirmation of Msp1 glycosylation. Among the glycosylated hits several enzymes of the carbohydrate metabolism were found, together with transport- and protein metabolism related proteins. Our results give rise to new hypotheses on general trends that may occur across species, as our results substantiate data gathered in other species.

In chapter 6 a more dedicated approach was used to study the glycosylation of the adhesive SpaCBA pili of L. rhamnosus GG. Molecular biology techniques were combined with mechanical and immunological approaches to illustrate the glycosylation statues of these pili. Moreover, this multidisciplinary approach enabled us to prove that the glycans are important for the interaction between these pili, and thus L. rhamnosus GG, and dendritic cells via the immune lectin receptor DC-SIGN.

Chapter 7 gives a comprehensive overview the most intriguing results and breakthroughs of this PhD work and their potential importance for the field of microbiota and probiotics on the one side and the field of bacterial glycobiology on the other side.

one.

9

two.the sweet tooth of

bacteria:common themes in

bacterial glycoconjugates

everything you (n)ever wanted to know on bacterial glycoconjugates

two.the sweet tooth of

bacteria:common themes in

bacterial glycoconjugates

everything you (n)ever wanted to know on bacterial glycoconjugates

12

the

swee

t too

th o

f bac

teria

An extended version of this chapter was published:

Tytgat H.L.P., Lebeer S. (2014). The sweet tooth of bacteria: common themes in

bacterial glycoconjugates. MMBR, 78(3), 372-417.

two.

13

abstract

Humans have been increasingly recognized as being superorganisms, living in close contact with a microbiota on all their mucosal surfaces. Yet, most studies on the human microbiota focus on gaining comprehensive insights on the composition of the microbiota in different health conditions (e.g. enterotypes), while there is also a need for detailed knowledge on the different molecules that mediate interactions with the host. Glycoconjugates are an interesting class of molecules for detailed studies as they form a strain-specific barcode on the surface of bacteria, mediating specific interactions with the host.

Strikingly, most glycoconjugates are synthesized by similar biosynthesis mechanisms. Bacteria can produce their major glycoconjugates using a sequential or an en bloc mechanism, with both mechanistic options coexisting in many species for different macromolecules. In this chapter, these common themes are conceptualized and illustrated for all major classes of known bacterial glycoconjugates, with a special focus on the rather recently emerged field of glycosylated proteins. We describe the biosynthesis and importance of glycoconjugates in both pathogenic and beneficial bacteria and both in Gram-positives and –negatives. The focus lies on microorganisms important for human physiology. In addition, the potential of a better knowledge of bacterial glycoconjugates in the emerging field of glycoengineering and other perspectives are discussed.

14

the

swee

t too

th o

f bac

teria

2.1 introDuction to bacteriaL gLycobioLogy

Pathogenic and beneficial members of the microbiota establish an intimate interaction with the host mucosa in order to manipulate the host metabolism and immune system (Lebeer et al., 2010). Key to a better understanding of these interactions are molecules present on the bacterial cell surface and secreted in the environment. Often, these key molecules are glycoconjugates such as glycoproteins, exopolysaccharides (EPS), capsular polysaccharides (CPS), lipopolysaccharides (LPS), lipooligosaccharides (LOS), lipoglycans, peptidoglycan (PG), glycosylated teichoic acids (TA) and other glycosylated secondary cell wall polymers. The glycans present on these molecules show an enormous diversity in monosaccharide building blocks, anomeric configuration, conformation and stereochemistry (Marino et al., 2010), which largely exceeds the eukaryotic glycoconjugate repertoire. The resulting diversity is uncannily: two glucose residues can already be joined together in thirty different ways (Laine, 1994). Bacteria are also able to produce ‘exotic’ rare sugars like bacillosamine, present on glycoproteins of Campylobacter jejuni (Young et al., 2002), in contrast to the ten monosaccharides that are typically detected in mammalians (Marino et al., 2010). The prominent location of bacterial glycoconjugates on the cell wall and their enormous diversity suggest that they form a unique barcode on bacterial cell surfaces. This makes them ideal candidates to establish specific and ‘tight’ interactions with host cells and abiotic surfaces, ranging from adhesion to immunomodulation (Kay et al., 2010). Of particular interest are various lectin immune receptors with different specificity displayed by host cells to scan these bacterial barcodes and induce specific responses (Rabinovich et al., 2012). This is crucial in view of the abundance, importance and niche-specificity of the microbiota and pathogenic infections. Most studies of the microbiota focus on mapping the microbiota and microbiome in different health conditions (Qin et al., 2010; Qin et al., 2012; The Human Microbiome Project Consortium, 2012), with the work on enterotypes as one of the most widely discussed findings of this research (Arumugam et al., 2011). These general studies need to be complemented with dedicated studies on the bacterial mediators of specific interactions, such as glycoconjugates to generate a comprehensive view of our bacterial friends and foes. Currently, the studies of glycoconjugates in pathogens outnumber those in beneficial bacteria largely. Especially for glycoproteins, the discrepancy is apparent. Moreover, the field of (bacterial) glycobiology is enigmatic: understanding the ties between glycan structures and their biological function is hampered by the non-template nature of glycan biosynthesis and the resulting heterogeneity. In addition, their enormous diversity and versatility makes their study quite challenging (Zaia, 2011). Considering the large amount of energy cells dedicate to the build-up of glycans, their functional importance - from an evolutionary perspective - should be high. Also, a better fundamental knowledge on bacterial glycomes can open up new horizons in the discovery of new drugs, bioactive compounds and vaccines. A combination of existing and new emerging technologies is rapidly advancing the field of glycobiology. This chapter aims at giving an overview of the current knowledge on bacterial glycoconjugates and focuses on the

two.

15

commonalities of their biosynthesis mechanisms both in Gram-positive and –negative species, and functions in bacteria-host interactions both in pathogens and beneficial bacteria.

2.2 generaL tHemes in gLycoconjugate biosyntHesis

Strikingly, bacteria appear to use only two main pathways to synthesize a plethora of glycoconjugates. The two main modular mechanisms used for glycoconjugate biosynthesis are the sequential and en bloc mechanism. This chapter conceptualizes and illustrates these two biosynthesis mechanisms. The commonality of these mechanisms is also illustrated by the promiscuity of some glycosyltransferases (GTs), which are able to glycosylate different substrates in some bacterial species. For instance, in several pathogens, GTs are active both in LPS/LOS and glycoprotein biosynthesis pathways (Bernatchez et al., 2005; Hug & Feldman, 2011). More specific examples of these commonalities will be discussed later.

2.2.1 gLycosyLtransferases, tHe key enzymes of gLycosyLation

Glycoconjugates do not fit into a linear understanding of the flow of biological information, as the glycome is not a template-driven process such as DNA, RNA and protein biosynthesis. In contrast, the glycan structures that are present on glycoconjugates are primarily determined by the complement of glycan-modifying enzymes present in an organism (Zaia, 2011). GTs, glycoside hydrolases or glycosidases, transporters, flippases, polymerases and other enzymes are all required to synthesize and metabolize the glycome of an organism. The crucial enzymes in the build-up of glycans are the GTs, which is underlined by their presence in the genome: GTs and glycosidases are believed to constitute 1 to 3% of the gene products of bacterial, eukaryotic and archaeal organisms (Davies et al., 2005). This percentage lies higher in plants and bacterial members of the microbiota like Bacteroides and Bifidobacterium, which suggests an important role of glycans in interaction with the human host (Davies et al., 2005; Xu et al., 2007). Parasitic and symbiotic life forms, on the other hand, generally contain fewer glycosylation enzymes (Davies et al., 2005). It remains to be seen if this figure needs to be updated as the fields of (meta)genomics and glycomics further evolve.

GTs (EC 2.4.x.y.) strictu sensu catalyze the formation of glycosidic bonds between a sugar moiety from an activated donor molecule and a specific substrate molecule. GTs can use a range of donor molecules, both nucleotide and non-nucleotide ones, and can target a wide diversity of substrates, like saccharides, proteins, nucleic acids, lipids and small molecules (Lairson et al., 2008). Chemically, these enzymes catalyze the transfer of a glycosyl group to a nucleophilic group, usually an alcohol. The net result of this reaction is an O-, N-, S- or C-glycosylated acceptor molecule (Lairson et al., 2008). The glycosidic bond is established by retainment (e.g. UDP-glucose turns into an α-glucoside) or inversion of the stereochemistry of the anomeric carbon of the donor sugar monomer (which results in an β-glucoside for UDP-glucose). These two stereochemical outcomes of reactions catalyzed by GTs, is also used as a means to classify these enzymes.

16

the

swee

t too

th o

f bac

teria

Inverting GTs use a direct displacement SN2-like reaction to catalyze the formation of the glycosidic bond (Murray et al., 1996). The result is a net inversion of the stereochemistry of the anomeric reaction center of the donor substrate. Most of the inverting GTs require a divalent cation, typically Mg2+ or Mn2+, which plays the role of an acid catalyst, coordinated by the non-stringent D-X-D motif, although metal-independent GTs are also reported (Lairson et al., 2008; Ramakrishnan et al., 2004). For instance, SpsA, a GT involved in the biosynthesis of the spore coat of Bacillus subtilis, was one of the first characterized bacterial inverting GTs (Charnock & Davies, 1999a; Charnock & Davies, 1999b).

The catalytic mechanism used by retaining GTs is still debated and two hypothetical mechanisms are described in literature. Both can require a divalent cation coordinated by D-X-D for their catalytic activity. The first mechanism was extrapolated from knowledge on retaining glycosidases (cf. the elucidation of the inverting GT mechanism), which means that retaining GTs would use a double displacement mechanism with a short-lived glycosyl-enzyme intermediate (Zechel & Withers, 2000). However, despite numerous efforts, no such glycosyl-enzyme intermediate could be trapped (Lairson et al., 2004). The second proposed mechanism suggests a direct attack of the acceptor, by a SNi-like or ‘internal return’ mechanism (Jakeman, 2011). This hypothesis originated from research on the crystal structure of the LOS galactosyltransferase LgtC of Neisseria meningitidis (Persson et al., 2001). Recently, this SNi-like mechanism was further supported by mechanistic evidence, which makes this latter mechanism more plausible (Lee et al., 2011).

GTs can also be classified according to the donor substrates they recognize. GTs that use sugar mono- or diphosphonucleotides (e.g. UPD-Glc, GDP-Man) are termed Leloir GTs, in honor of Nobel prize winner Leloir who performed groundbreaking work on sugar nucleotides (Cabib & Leloir, 1954). GTs utilizing non-nucleotide donors, like polyprenol pyrophosphate and sugar-1-phosphate, are then called non-Leloir GTs (Lairson et al., 2008).

The biochemical characterization of GTs has proven to be hard, as the first reports on the X-ray crystal structures of GTs only started to emerge in the late 1990s (Unligil & Rini, 2000). Although the similarities between glycosidases and GTs are quite high, GTs appear to only exhibit a narrow spectrum of folds. The Leloir enzymes can be split in two large families based on their common 3D-fold: the GT-A en GT-B fold (Lairson et al., 2008). More recently, a third fold category was discriminated, the GT-C fold (Roychoudhury & Pohl, 2010). An important caveat hereby is that not all enzymes that represent a GT-A or GT-B fold are GTs. Also, non-Leloir GTs possess non-GT-A and non-GT-B folds. An example of such GTs is the transglycosylase, PBP2, of Staphylococcus aureus involved in peptidoglycan biosynthesis, which has a bacteriophage-lysozyme-like fold (Lovering et al., 2007).

As GTs are a very diverse group of enzymes and they can recognize a range of donor and acceptor molecules, a classification was proposed by the group of Henrissat based on the amino acid sequence of GTs, analogous to the classification of glycoside hydrolases (Campbell et

two.

17

al., 1997). These constantly updated GT families can be consulted online in the Carbohydrate-Active enZymes (CAZy) database (www.cazy.org) (Cantarel et al., 2009). Using algorithmic methods to subdivide protein sequences in distinct families, GTs are currently classified in 95 families (March 2014). Each GT family contains enzymes related by sequence and 3D-fold. As the mechanism of catalysis, and additionally their inverting or retaining character, is (largely) conserved within each family, this classification can be used to predict the mode of action of newly predicted GTs. Nevertheless, this does not mean that all enzymes of a family recognize the same donor and/or acceptor: polyspecificity is common among GT families (Coutinho et al., 2003). One should be prudent with the overinterpretation of predictions purely based on the classification of a GT in a certain family (Davies & Sinnott, 2008).

A few ‘special classes of GTs’ deserve some extra focus, i.e. the transglycosylases of peptidoglycan (PG) biosynthesis pathways, synthases catalyzing EPS/CPS formation and enzymes of the en bloc glycosylation pathway, such as priming GTs, polymerases and oligosaccharyltransferases (OSTs).

Transglycosylases are GTs involved in the polymerization of carbohydrate chains to PG. These enzymes follow the same mechanism as retaining glycoside hydrolases and their action can be described by a nucleophilic substitution reaction, resulting in inversion of the configuration (Welzel, 2005). Transglycosylases use lipid II as an acceptor and add these new PG subunits processively to the non-reducing end of the growing chain, which is the donor (Barrett et al., 2007; Perlstein et al., 2007; Yuan et al., 2007). Transglycosylases are classified in CAZy family 51 and have a lysozyme-like fold (Lovering et al., 2007). These enzymes occur as monofunctional or bifunctional enzymes, the latter combine their GT action with a transpeptidase domain (van Heijenoort, 2001).

Synthases are multifunctional processive GTs that catalyze the initiation, polymerization and transport of usually rather simple glycans. These enzymes are involved in the biosynthesis of e.g. the type 3 CPS of Streptococcus pneumoniae and EPS structures like alginate, cellulose and hyaluronan (Whitney & Howell, 2013; Yother, 2011). In Salmonella enterica serovar Borreze O:54, a synthase is responsible for O antigen biosynthesis (Keenleyside & Whitfield, 1996). Synthases are relatively small integral membrane proteins, consisting of 4 transmembrane domains and a large cytoplasmic loop. This loop harbors the catalytical residues and a Q-X-X-R-W motif involved in the processivity of the enzyme by retaining the growing polymer in the enzyme (Saxena et al., 2001; Weigel et al., 1997). The synthase of S. pneumoniae loosely associates with the growing glycan chain in the beginning, but interacts more tightly after synthesis of a short oligosaccharide to ensure processive elongation (Forsee et al., 2009).

Priming GTs are implicated in the en bloc biosynthesis of glycoconjugates like EPS, CPS, LPS O antigen and glycoproteins. These are special GTs, as they do not catalyze the formation of a glycosidic bond. Priming GTs are linked to the membrane and are involved in the transfer of a phosphorylated sugar monomer from an activated nucleotide to an undecaprenyl phosphate

18

the

swee

t too

th o

f bac

teria

(UndP) carrier, on which the glycan is then further assembled. These enzymes, also known as initiating GTs, catalyze the formation of an energy-rich phosphate bond (Bugg & Brandish, 1994; Lehrer et al., 2007; Price & Momany, 2005; Wang et al., 1996). This UndP-linked sugar monomer forms the substrate for the successive action of GTs that build up the glycan further (cf. infra).

OSTs transfer an oligosaccharide en bloc from the lipid carrier to the acceptor substrate, a protein (Musumeci et al., 2014; Wacker et al., 2002). The currently best-studied and characterized OST is the PglB OST of Campylobacter sp. (Ihssen et al., 2012; Kowarik et al., 2006b; Lizak et al., 2011b; Wacker et al., 2006; Wacker et al., 2002). This OST is strikingly similar to the central catalytic STT3 enzyme of the eukaryotic multicomponent OST counterpart (Knauer & Lehle, 1999b). Bacterial OSTs, like PlgB, however, are monosubunit enzymes (Feldman et al., 2005). PglB is involved as N-OST in the N-glycosylation of proteins in Campylobacter sp. and the factors determining its catalytic mechanism and specificity are extensively studied. The catalysis of the oligosaccharide transfer could be linked to the presence of an N-acetyl group in position 2 of the sugar directly linked to the UndP carrier. This C-2 acetamido group at the reducing end sugar is necessary for recognition and /or catalysis either as a partner in forming critical hydrogen bonds with the OST or as a stabilisator of the transition state (Wacker et al., 2006). The PglB enzyme was shown to recognize a D/E-Y-N-X-S/T (Y, X cannot be a P) motif located in flexible loops of folded proteins, which is an extension of the eukaryotic N-glycosylation motif (N-X-S/T) (Kowarik et al., 2006a; Kowarik et al., 2006b; Schwarz et al., 2011b). More specifically, D-Q-N-A-T was identified as being the optimal consensus sequence for C. jejuni (Chen et al., 2007a; Kowarik et al., 2006b).

In the biosynthesis of polysaccharides (EPS, CPS, LPS O antigen), the OST is replaced by a polymerase (Drummelsmith & Whitfield, 1999; Whitfield, 2006). Several oligosaccharide building blocks are, in the case of polysaccharide biosynthesis, linked together to form the glycan. This reaction is catalyzed by a ‘Wzy’ polymerase (Drummelsmith & Whitfield, 1999; Whitfield, 1995; Whitfield, 2006; Whitfield & Paiment, 2003; Whitfield & Roberts, 1999).

In most cases, the factors determining the specificity of GTs remain elusive and it is difficult to predict merely based on sequence analysis in which glycoconjugate biosynthesis they are involved. Thus, a crucial area for further research in bacterial glycobiology is the search for factors or motifs determining the substrate specificity of GTs, both in the GTs and in the substrates. This would significantly improve the annotation of GTs and glycoconjugate biosynthesis clusters and would also be highly relevant for the ongoing bacterial genome and metagenome projects. A complicating aspect is, however, that some GTs, and especially several OSTs, show promiscuity towards different substrates. PglB of C. jejuni for instance can both transfer O antigen polysaccharides and N-glycosylate proteins (Feldman et al., 2005). Another example is the O-OST of N. meningitidis, PglL, which can interfere with PG biosynthesis and even act as a Leloir GT by transferring monosaccharides (Faridmoayer et al.,

two.

19

2008; Faridmoayer et al., 2007; Musumeci et al., 2013).

2.2.2 Sequential glycosylation

Sequential glycosylation (Fig.2.1) in bacteria is the most basic form of glycosylation: GTs transfer the sugar moieties directly to the substrate and build up the glycan. Sequential glycosylation has been documented both in Gram-positive and –negative species (cf. infra). Essentially, GTs use nucleotide-activated sugars as donors to transfer sugar moieties to their substrate in the cytoplasm. This can occur directly, like in the case of protein glycosylation or glycosylation of TAs. Or, in the case of CPS and LPS O antigen biosynthesis, the glycan is synthesized whilst associated to the membrane (e.g. via diacylglycerol or UndP in rare cases). Once synthesized, the glycoconjugates are transported through the inner membrane (Gram-negative species) or cytoplasmic membrane (Gram-positives). Some peculiar GTs, i.e. synthases, are multifunctional and are both responsible for the biosynthesis and the transport of the glycan (cf. earlier) (Fig.2.1).

Figure 2.1 – The sequential mechanism of bacterial glycoconjugate biosynthesisIn this rather intuitive way of glycosylation, GTs transfer the sugar moieties from nucleotide-activated sugar donors directly to their substrate. This substrate can be a glycoprotein or a growing polysaccharide chain linked to the membrane. Once the glycoconjugate is fully synthesized, a transporter transfers the structure to the periplasm (Gram-negative species) or to the other side of the cytoplasmic membrane (Gram-positive species). In some cases, the GT also is responsible for the transport of the glycan across the membrane.

The glycans synthesized by this pathway are usually rather simple, meaning consisting of a single or only a few different sugar building blocks. In many cases, several GTs work in

20

the

swee

t too

th o

f bac

teria

concert, often even processively, i.e. without release of intermediate glycan products. Good illustrations of the sequential glycosylation mechanism are the ABC-transporter dependent pathway of CPS (e.g. E. coli group 2 and 3 capsules (Corbett & Roberts, 2008)) and LPS O antigen biosynthesis (e.g. E. coli 08 and 09a (Whitfield, 1995)). Also bacteria using a synthase for their CPS (e.g. type 3 CPS of S. pneumoniae (Dillard et al., 1995)), EPS (e.g. alginate produced by P. aeruginosa sp. (Rehm & Valla, 1997)) or LPS O antigen biosynthesis (e.g. in S. enterica serovar Borreze O:54 (Keenleyside & Whitfield, 1996)), follow the here-described sequential glycosylation mechanism. Substitution of proteins with simple glycans (both N- and O-linked), like flagellar glycosylation in several pathogenic species, also illustrates the sequential mechanism neatly (Grass et al., 2003; Logan, 2006). Furthermore, TA glycosylation in Firmicutes is thought to occur via the direct attachment of sugars to the backbone (Allison et al., 2011). The widespread nature of sequential glycosylation in bacteria is illustrated further in detail, where several type examples of bacterial glycoconjugate biosynthesis are discussed.

2.2.3 En bLoc gLycosyLation of gLycoconjugates

Figure 2.2 – The en bloc mechanism of bacterial glycoconjugate biosynthesisIn the en bloc mechanism, in addition to standard Leloir GTs, peculiar GTs come into play, i.e. the priming GT and OST (glycoproteins) or polymerase (polysaccharides). The priming GT forms an energy-rich bond between a sugar-phosphate moiety and the UndP carrier. Several GTs then consecutively glycosylate the lipid carrier further, till the full glycan (in the case of glycoproteins) or the building block (polysaccharides) is synthesized. A transporter (e.g. a flippase) then transfers the glycans to the other side of the inner membrane (Gram-negative species) or cytoplasmic membrane (Gram-positives). Glycosylated proteins are formed through the action of an OST, which transfers the glycan to the protein. In the case of polysaccharides, like EPS, CPS and the O antigen of LPS, the building blocks are assembled by a polymerase.

two.

21

The second important modular mechanism involved in the build-up of glycoconjugates is the en bloc mechanism (Fig.2.2). This mechanism is essentially dedicated to the biosynthesis of more complex glycoconjugates. En bloc biosynthesis of glycans starts in the cytoplasm, where a membrane-anchored priming GT utilizes a nucleotide-activated sugar residue to add a first glycan subunit via an energy-rich pyrophosphate bond to a recyclable lipid carrier (i.e. UndP) (Bugg & Brandish, 1994; Meier-Dieter et al., 1992; Price & Momany, 2005; Valvano, 2008). The glycan is then further assembled by the successive and concerted action of GTs that use nucleotide-activated sugar residues as donors. Once a building block of the glycan is fully synthesized, the lipid-linked oligosaccharide is flipped or transported across the inner membrane (Gram-negative species) or the cytoplasmic membrane (Gram-positive bacteria) (Bugg & Brandish, 1994). Here, a polymerase links several building blocks and forms a repetitive glycan polymer, which leads to the production of EPS, CPS and O antigen LPS structures (Whitfield, 1995; Whitfield & Roberts, 1999). The flippase is commonly denominated as Wzx and the polymerase as Wzy (Wzy dependent pathway) (Whitfield, 2006; Whitfield & Paiment, 2003). In the case of glycoprotein biosynthesis, the polymerase is ‘substituted’ by an OST (cf. supra), which transfers and ligates the glycan to a protein substrate in an en bloc manner (Kowarik et al., 2006a; Wacker et al., 2002). This mechanism has up till now only been reported in the O- and N-linked glycosylation of Gram-negative species, with N-protein glycosylation in C. jejuni being a type example (Nothaft & Szymanski, 2010; Szymanski et al., 2003a) (Fig.2.2). The en bloc mechanism and detailed illustrations of bacteria using this pathway to synthesize their glycoconjugates are further discussed below.

2.3 bacteriaL gLycoconjugates

Bacterial glycoconjugates display an enormous structural diversity. Bacteria can produce a variety of both intracellular and extracellular polysaccharides (Fig.2.3). The major glycoconjugates of the glycome of Gram-negative bacteria encompass PG, EPS, CPS, LPS, LOS, glycoproteins (both O- and N-linked) and intracellular storage polysaccharides. Gram-positives lack LPS and LOS, but can have lipoglycans and glycosylated TAs.

Intracellular polysaccharides are used as an energy reserve, like the well-known storage polysaccharide glycogen. An overview on the knowledge on glycogen and other bacterial intracellular storage polysaccharides is beyond the scope of this chapter and the reader is referred to existing reviews (Wang & Wise, 2011; Wilson et al., 2010).

For a detailed overview of the current knowledge on the biosynthesis mehanisms of PG, EPS, CPS, LPS, LOS, lipoglycan and teichoic acid the reader is referred to the original manuscript as published in MMBR (Tytgat & Lebeer, 2014). In this work the main focus lies on the current knowledge on protein glycosylation.

22

the

swee

t too

th o

f bac

teria

Figure 2.3 – Bacterial glycoconjugatesBacteria synthesize a plethora of glycoconjugates. Characteristic for Gram-negative species are LPS and LOS structures, while several Gram-positives are known to glycosylate their TAs (WTA and LTA). Both Gram-negative and -positive species can produce glycosylated flagella, pili or fimbriae, CPS, EPS and glycoproteins. The cell walls of both Gram-negative and –positive species also harbor a PG layer, which serves as a scaffold for the anchorage of other molecules. The round dots represent proteins, while the triangles are ribitol- or glycerolphosphate moieties.

2.4 protein gLycosyLation

In 1999, Apweiller et al. estimated based on analysis of the SWISS-PROT database, that about half of proteins in nature is glycosylated (Apweiler et al., 1999). Albeit their existence in prokaryotic species was refuted for a long time, glycoproteins were first discovered in Archaea in 1976 (Mescher & Strominger, 1976) and the occurrence of protein glycosylation in Bacteria is already well documented since the first reports in Clostridium bacteria that were published in 1975 (Sleytr, 1975; Sleytr & Thorne, 1976). The fact that bacterial protein glycosylation was discovered ‘quite late’ can be linked to the absence of glycoproteins in the most frequently used laboratory strains of E. coli and Salmonella sp. (Messner, 2004). Glycosylated bacterial proteins have mainly been characterized in pathogens and include surface layer proteins, pilins, flagellins and enzymes. Recent studies also report the occurrence of glycoproteins in beneficial bacteria like the gut symbiont Bacteroides fragilis (Fletcher et al., 2011; Fletcher et al., 2009) and Lactobacillus species (Fredriksen et al., 2012; Konstantinov et al., 2008; Lebeer et al., 2012b).

The glycosylation of proteins can occur via the linkage of glycans to the hydroxyl group of a serine or threonine residue and in some cases tyrosine residues, commonly referred to as O-glycosylation or the glycan can be linked to an asparagine residue, which results in an N-glycosylated protein. Occasionally, such as for the glycocin F protein of L. plantarum KW30, glycosylation is linked to a cysteine, resulting in S-glycosylation (Stepper et al., 2011; Venugopal et al., 2011). Protein glycosylation processes can also be distinguished based on

two.

23

the mechanism used to transfer glycans to the protein substrate, i.e. in a sequential or en bloc manner. Sequentially glycosylated substrates are directly targeted by GTs transferring sugar moieties (Fig.2.1). Often, these systems are dedicated, meaning that the genes encoding the protein substrate and the GT, as well as the necessary transporters, are clustered. If the glycan structure is fully synthesized on a lipid carrier (UndP carrier) prior to collective ligation on the substrate, this is called en bloc biosynthesis (Fig.2.2).

Protein en bloc glycosylation in prokaryotes resembles the eukaryotic N-glycosylation processes partially. In eukaryotes, protein glycosylation occurs in the lumen of the endoplasmic reticulum (ER), whilst in prokaryotes, which do not have organelles, the sugar nucleotide precursors are synthesized in the cytoplasm. The biosynthesis of the glycan then occurs in close proximity of the cell membrane, as priming GTs are membrane-bound enzymes. After biosynthesis, the glycan is flipped across the cell membrane. In Gram-negative species the transfer of the glycan to the substrate takes place in the periplasm. Whether Gram-positives also have a periplasm is still debated, but several studies point towards the presence of a small periplasm between the cell membrane and PG-layer (Matias & Beveridge, 2005; Matias & Beveridge, 2006; Takade et al., 1988; Umeda et al., 1992). Both eu- and prokaryotes utilize lipid carriers to assemble the glycans: dolichol pyrophosphate in eukaryotic species and UndP in bacteria (Jones et al., 2009). The en bloc glycosylation of proteins is catalyzed by an OST, which is typically a multimeric protein containing up to eight transmembrane subunits in eukaryotes, whilst bacterial OSTs appear to be monomeric proteins (Feldman et al., 2005). In eukaryotes, en bloc N-glycosylation occurs prior to folding, when the substrates are still flexible (Kowarik et al., 2006a). In contrast, data from C. jejuni suggests that en bloc glycosylation of proteins occurs on flexible, surface-exposed parts of folded proteins (Kowarik et al., 2006a). In addition, eukaryotes perform a quality control of the N-glycoproteins, while the glycans are generally also further modified or trimmed in the ER and Golgi apparatus. Quality control and trimming of glycans has –to the best of our knowledge- not (yet) been reported in bacteria.

Eu- and prokaryotes both can perform en bloc N-glycosylation and sequential O-glycosylation of proteins. However, in prokaryotes, sequential N-glycosylation and en bloc O-glycosylation have also been reported (Aebi et al., 2010; Dell et al., 2010; Larkin & Imperiali, 2011; Weerapana & Imperiali, 2006), highlighting that the glycosylation possibilities in prokaryotes appear to be larger. Current knowledge seems to indicate that N-glycosylation of proteins occurs almost exclusively in Gram-negative species, namely in epsilon and a few delta proteobacteria (Dell et al., 2010). Only one report has been published on a potentially N-glycosylated protein in Gram-positive bacteria, more in particular on the Platelet Aggregation-Associated Protein of S. sanguis (Erickson & Herzberg, 1993). Taken together, O-glycosylation is a widespread phenomenon in bacteria, whilst N-glycosylation of proteins seems to be more restricted. Glycosylation of proteins in bacteria has been shown to modulate the physicochemical properties of their substrates. Glycan modifications of proteins can protect them against proteolysis (e.g. AIDA-I of E. coli 2787 (Charbonneau et al., 2007)), confer extra stability (e.g.

24

the

swee

t too

th o

f bac

teria

C. jejuni proteins (Alemka et al., 2013)), enhance the activity (e.g. gingipains of P. gingivalis (Vanterpool et al., 2005a)) and change the surface properties (e.g. pili of P. aeruginosa 1244 (Smedley et al., 2005)). In addition, glycoproteins play a role in interactions between bacteria and their surroundings. Often, glycoproteins are important virulence factors and antigens, involved in adhesion events (e.g. AIDA-I of E. coli 2787 (Benz & Schmidt, 2001)), immune modulation (e.g. Apa glycoprotein of M. tuberculosis (Horn et al., 1999; Romain et al., 1999)) and evasion (e.g. the pili of N. meningitdis (Faridmoayer et al., 2008; Hamadeh et al., 1995; Marceau et al., 1998)). Protein glycosylation is often also heterogeneous and dynamic: one protein can carry more than 1 type of glycans and glycans can be incomplete or modified. In Clostridium botulinum this heterogeneity can manifest itself even between flagella of different isolates (Twine et al., 2008), which points towards a role for glycoproteins as a strain-specific barcode.

Genome sequencing of the food-borne intestinal pathogen Campylobacter jejuni revealed an unexpected capacity for this organism to produce a wide variety of glycoconjugates, including PG, LOS, CPS and N- and O-linked sugars (Guerry & Szymanski, 2008; Gundogdu et al., 2007; Parkhill et al., 2000). Glycobiologists started to focus their efforts on C. jejuni as a toolbox to unravel the mechanisms of protein glycosylation. Gradually, C. jejuni became a model organism of protein glycosylation, as it can both N- and O-glycosylate proteins by en bloc and sequential transfer resp. Especially the enzymes for protein N-glycosylation, clustered in the pgl locus, show remarkably high conservation (Nothaft & Szymanski, 2010; Szymanski & Wren, 2005), and can even be functionally transferred into Escherichia coli (Wacker et al., 2002), which was one of the early milestones of glycoengineering. In contrast, O-glycosylation genes are less conserved and phase variable (Karlyshev et al., 2002; van Alphen et al., 2008).

2.4.1 En bLoc n-gLycosyLation of proteins

N-protein glycosylation has been described only in a few species, and its occurrence in bacteria is believed to be rare. En bloc N-glycosylation has until now mainly been studied in detail in Campylobacter sp. (Nothaft & Szymanski, 2010), while H. influenzae and A. pleuropneumoniae harbor a sequential N-glycosylation mechanism (Grass et al., 2010; Schwarz et al., 2011a). Recently, genomic analysis revealed the presence of an en bloc N-glycosylation mechanism in some Helicobacter sp. (Loman et al., 2009). N-glycosylation has been suggested in Chlamydia trachomatis, but information on the underlying mechanisms is lacking (Swanson & Kuo, 1991a; Swanson & Kuo, 1991b). For a long time, it was also believed that Borrelia burgdorferi, the cause of Lyme disease, produced four N-glycoproteins (FlaA, FlaB, OspA, OspB) (Ge et al., 1998; Sambri et al., 1992). A later report by Sterba and co-workers refutes these findings and attribute the positive glycosylation signals to tightly bound culture medium glycoproteins (Sterba et al., 2008). This illustrates how important it is to document the occurrence of glycoproteins in bacteria by at least complementary methods.

two.

25

C. jejuni, a gastrointestinal pathogen to humans and commensal in avian colons, produces a wide variety of N-glycosylated proteins. First evidence on the existence of an N-protein glycosylation gene cluster came from the identification of LPS-like biosynthetic enzymes in C. jejuni 81-176. The glycans produced by the gene products of this locus, named the pgl locus (16 kb), seem to be immunodominant as their absence has been shown to result in reduced reactivity with antisera and a loss of immunogenicity (Szymanski et al., 1999). One of the first characterized glycoproteins of C. jejuni is PEB3 (CJ0289c), an important antigen of C. jejuni (Pei et al., 1991), which is glycosylated with a heptasaccharide of 1406 Da, that reacts with Soybean lectin (SBA) (Linton et al., 2002; Young et al., 2002). An asparagine residue of PEB3 was shown to be modified with one monomer of the special sugar D-Bacillosamine (Bac; 2,4-diacetamido-2,4,6-trideoxyhexose), five D-GalNAc residues and a D-Glc branch (Young et al., 2002). The en bloc nature of the process was confirmed by mutational analysis of the pgl locus, as interference at any stage of the oligosaccharide subunit assembly resulted in a total loss of glycoconjugate formation (Kelly et al., 2006). The only feature that seemed to be dispensable is the addition of the glucose branch of the heptasaccharide (Kelly et al., 2006). Gradually, more evidence emerged on the presence of other N-glycoproteins in C. jejuni strains (mostly in the 81-176 and NCTC 11168), confirming the general nature of the N-glycosylation process, i.e. targeting a range of proteins (Ding et al., 2009; Kakuda & DiRita, 2006; Linton et al., 2002; Rangarajan et al., 2007; Scott et al., 2009; Scott et al., 2011b; Wyszynska et al., 2008; Young et al., 2002). In total over 60 glycoproteins are already identified in C. jejuni sp. (Ding et al., 2009; Guerry & Szymanski, 2008; Scott et al., 2011b; Young et al., 2002).

The conserved pgl locus comprises all enzymes necessary for the N-glycosylation of proteins. Its role was confirmed by the functional transfer of the locus and the AcrA (CJ0367c) periplasmic glycoprotein into E. coli, which resulted in the production of glycosylated AcrA proteins (Wacker et al., 2002). The pgl cluster was also extensively studied in vitro, which -complementary to mutant analysis- enabled the characterization of all enzymes individually (Glover et al., 2005a).

PglF (CJ1120c) is a C6 dehydratase, which together with the UDP-4-keto-6-deoxy-GlcNAc C4-aminotransferase PglE (CJ1121c) (Vijayakumar et al., 2006) is involved in the biosynthesis of the Bac moiety (Schoenhofen et al., 2006c). PglD, an acetyltransferase modifying the UDP-4-aminosugar, is coupled to the reactions catalyzed by PglF and PglE (Olivier et al., 2006). Mutation of this enzyme resulted in monoacetylated Bac, which can still be transferred in mutant strains (Ding et al., 2009). A complementing enzyme was also found in the C. jejuni genome (Kelly et al., 2006). Later, PglD was found to be a multifaceted acetyltransferase as it can both target intermediates of O- and N-glycosylation processes (Demendi & Creuzenet, 2009) (Fig.2.4A.).

The priming GT or Bac 1-phosphorytransferase, PlgC, an integral membrane protein, then adds Bac to the lipid carrier (UndP). Its reaction is coupled to the action of the consecutive GTs

26

the

swee

t too

th o

f bac

teria

PglA, PglJ, PglH and PglI (Glover et al., 2006). PglA adds an α(1-3) linked GalNAc to Bac and this disaccharide serves as a substrate for the action of PglH, PglI and PglJ (Fig.2.4A.). These latter GTs add the other four GalNAcs in an α(1-4) manner (the first GalNAc is added by PglJ, the latter three by PglH) and a branching β(1-3) Glc (PglI) to the structure (Glover et al., 2005a; Kelly et al., 2006; Szymanski et al., 2003b; Young et al., 2002). PglH was found to have polymerase activity, as it can transfer three GalNAc residues consecutively (Glover et al., 2005a). This processive polymerase only harbors a single active site and product inhibition seems to limit the sequential GT reactions carried out by PglH to three. Regulation of this enzyme occurs thus via a ‘counting mechanism’ of product accumulation and inhibition (Troutman & Imperiali, 2009).

The protein encoded by CJ1130c (wlaB) transfers the resulting lipid-linked heptasaccharide from the cytoplasm to the periplasm and was renamed PglK. PglK is an ABC transporter using the hydrolysis of ATP as an energy source to flip the glycans (Alaimo et al., 2006) (Fig.2.4A.). Strikingly, expression of the pgl locus in E. coli revealed the presence of two distinct interchangeable mechanisms to flip the lipid-linked oligosaccharides across the membrane: one using PglK and the other one using the Wzx flippase of the LOS pathway. Moreover, PglK could even complement a defect in Wzx, albeit the two enzymes show very little similarity, which suggest a relaxed specificity of these enzymes (Alaimo et al., 2006). In H. pylori a homologue of PglK, WzK, was identified as a flippase involved in O antigen biosynthesis (Hug et al., 2010).

An OST termed PglB (CJ1126c) finally transfers the oligosaccharide from the lipid carrier to the substrate protein (Linton et al., 2002; Young et al., 2002) (see earlier) (Fig.2.4A.). Structurally, the enzyme contains 11 transmembrane segments and two relative large periplasmic regions apart from the periplasmic C-terminal domain (Li et al., 2010). PglB contains a strictly conserved W-W-D-Y-G motif, required for its activity in vivo (Ihssen et al., 2012; Wacker et al., 2002). Several X-X-D domains were also found: two in the C-terminal, one in the first periplasmic region and three more in the periplasm. These domains are typical for GT activity and are thought to be involved in catalysis (see before).

The PglB OST recognizes a motif in the acceptor substrate, similar to the eukaryotic N-X-S/T sequence (Glover et al., 2005b). Work by Kowarik and co-workers further extended the consensus sequence of PglB to D/E-Y-N-X-S/T, with Y and X being any amino acid except for proline (Kowarik et al., 2006b). They also showed the requirement of a negatively charged side chain at the -2 position of the glycoproteins. Chen and co-workers (Chen et al., 2007a) took these findings a step further and assessed variable amino acids at the Y and X position of the glycan sequon for their impact on glycosylation efficiency. The Y amino acid of the sequence is preferably a glutamine, asparagine or large hydrophobic moiety, while for the one on the X-position there is a preference for negatively charged amino acids, like lysine, arginine, alanine and serine. This resulted in the postulation of D-Q-N-A-T as the optimal

two.

27

acceptor sequence (Chen et al., 2007a). The lipid-linked oligosaccharides also need to be transferred to the periplasm where PglB is active (Linton et al., 2005; Nita-Lazar et al., 2005). In contrast to eukaryotic OSTs, PglB catalyzes with a high efficiency the glycosylation of folded bacterial substrates, but the glycosylation sites of bacterial proteins are predominantly present in the flexible, surface-exposed parts of folded protein (Kowarik et al., 2006a). The acceptor region appears to partially adopt an unstable α-helix conformation and form random coils (Slynko et al., 2009). These findings were confirmed experimentally via the crystal study of the PEB3 glycosylated adhesin (Rangarajan et al., 2007). Three key glycosylated residues are well exposed at the surface, which makes them accessible to PglB even in the folded state of the protein. In this way, PglB can target the glycosylation sites without the need for local protein structure rearrangements (Rangarajan et al., 2007).

Research on the specificity of PglB, PglC and PglJ towards the polyisoprenol diphosphate carriers on which the oligosaccharide is built, revealed the preference towards cis double bond geometry and α-unsaturation, while the precise length of the polyisoprene is less critical (Chen et al., 2007b). Li et al. later confirmed this relative relaxed specificity towards the lipid carriers (Li et al., 2010). Moreover, PglB recognizes an acetamido group at the C-2 position of the sugar directly linked to the UndPP carrier (Wacker et al., 2006). Bac, GlcNAc, GalNAc and FucNAc all have a C-2 acetamido group, which can partially explain the promiscuity of PglB (the enzyme can even transfer monosaccharides) towards the sugars it transfers (Li et al., 2010; Wacker et al., 2006).

Studies on the functional importance of N-glycans present on the proteins of C. jejuni are still rather scarce. These glycans seem to modulate the functions of several specific proteins, which are not the same for all proteins. Impaired glycosylation of the VirB10 glycoprotein, a type IV secretion system glycosylated at two sites, for instance, resulted in reduced natural transformation (Larsen et al., 2004). Mutations affecting the pgl locus resulted in altered phenotypes, such as decreased adhesion, invasion efficiency in vitro and abolishment of mouse and chicken colonization in vivo (Karlyshev et al., 2004; Szymanski et al., 2002; Vijayakumar et al., 2006; Young et al., 2002). The glycans were also hypothized to play a role in immune evasion by masking primary amino acid sequences, based on studies with human antisera (Szymanski et al., 1999). Moreover, the terminal GalNAc residues of the N-linked heptasaccharides were shown to interact with the human C-type Macrophage Galactose Lectin (MGL) receptor present on human dendritic cells and distinct macrophage subsets, thereby modulating immune responses (van Sorge et al., 2009). A mutant in the pgl locus also resulted in enhanced expression of the pro-inflammatory cytokine IL-6, again implicating a shielding role of the glycans (van Sorge et al., 2009).

In addition to a role in host interaction, recent work by Alemka et al. underlines the importance of N-glycosylation for bacterial fitness, as mutation of pglB resulted in a reduced growth in medium supplemented with chicken cecal content (Alemka et al., 2013). This result could be

28

the

swee

t too

th o

f bac

teria

linked to the presence of proteases in this medium as inhibition of the serine and metalloproteases resulted in restored viability and a partial rescue of bacteria growth. This further adds support for a role of N-glycans in protection against proteolysis of surface proteins.

A similar en bloc N-glycosylation system has been discovered in Helicobacter canadensis based on genome analysis (Loman et al., 2009). Further genome analysis showed the presence of PglB orthologues in H. canadensis, H. pullorum and H. winghamensis (Jervis et al., 2010). H. pullorum has two orthologues of pglB, pglB1 and pglB2, which are unrelated and not located in a large locus involved in protein glycosylation, although the colocalisation of pglD, pglE and pglF orthologues in a small separate cluster was reported. The PglB1 enzyme can transfer the C. jejuni heptasaccharide in E. coli, but results only in the modification of two sites of the CJ0114 protein instead of the four that are modified by PglB of C. jejuni. This indicates the OST character of this enzyme and the distinct specificity compared to PglB from C. jejuni. PglB2 seems to be essential, as no knockout can be made in this gene (Jervis et al., 2010). The character, activity and functionality of this glycosylation system remain to be further investigated. At the moment it seems, in contrast to the findings in C. jejuni, that the N-glycosylation of proteins is not a genus-wide feature of Helicobacter (Jervis et al., 2010).

The only report on a glycoprotein in Helicobacter sp. is on the glycosylation of RecA in H. pylori. Based on mutational analyses of recA and eno (an enolase homologue), glycosylation was claimed to be required for the full functionality of RecA in response to DNA damage (Fischer & Haas, 2004). However, glycosylation of RecA needs to be further substantiated, especially also in other bacterial species, given its universal role.

2.4.2 sequentiaL n-gLycosyLation of proteins

In H. influenzae, a distinct mechanism of N-glycosylation was reported, in which the glycosylation occurs in a sequential way (Fig.2.4B.). The high molecular weight HMW1 adhesin, which is loosely associated with the bacterial surface and involved in the attachment of the pathogen to human epithelial cells, is decorated with galactose, glucose and mannose residues (Grass et al., 2003). The glycan is believed to account for 5% of the total mass of HMW1 and is attached in the cytoplasm (Grass et al., 2003). 31 sites of the protein are modified with mono- or dihexoses (total of 47 hexoses), all harboring the N-X-S/T sequence, except one, which is glycosylated at an N-V-E sequon (Gross et al., 2008).

HMW1 is associated with two accessory proteins of which HMW1B is involved in the translocation of HMW1 from the periplasm to the cell surface and HMW1C was found to be required for HMW1 glycosylation (Grass et al., 2003). The HMW1C protein shows homology to GTs and can transfer glucose and galactose residues to HMW1. In addition, the GT can also generate hexose-hexose bonds, which suggests that this multifunctional GT belongs to a new family of bacterial GTs (Grass et al., 2010; Gross et al., 2008). HMW1C transfers glucose residues to all N-glycosylation sites, while it can transfer galactose to only a subset

two.

29

of the glucosylated N-residues. How the HMW1C GT makes the distinction between both types remains to be elucidated (Grass et al., 2010). The biosynthesis of the UDP-glucose precursors that are linked to HMW1 requires an enzyme of the LOS pathway, namely the phosphoglucomutase PgmB (Grass et al., 2003). Glycosylation of the HMW1 adhesin was shown to play a role in the tethering of the adhesin to the surface and stability of the protein, as glycosylation defects result in an HMW1 adhesin that is prematurely degraded and released from the surface (Grass et al., 2003).

A homologue of the HMW1C GT of H. influenzae was found by the group of M. Aebi in Actinobacillus pleuropneumoniae. This inverting cytoplasmic N-glycosyltransferase, termed NGT, targets the N-X-S/T consensus sequence and modifies this sequon with Gal and Glc residues. Elongation of the Glc residue can be performed by an α 1,6-glucosyltransferase (Schwarz et al., 2011a). The relaxed specificity of the NGT enzyme was illustrated by the reconstruction of this glycosylation pathway in E. coli. This general glycosylation system seems to have a preference for autotransporter adhesins, but can also glycosylate heterologous proteins (Naegeli et al., 2014).

In Chlamydia trachomatis, causing chlamydia in humans, three N-glycosylated proteins were reported. The analyses all date from the early 90s and rely on the use of lectins, glycosidases and blocking experiments. The 18 kDa and 32 kDa lectin binding proteins of the L2 serovar were found to be glycosylated and the glycans were suggested to be of importance for the interaction with HeLa cells (Swanson & Kuo, 1991a; Swanson & Kuo, 1994a). The best-documented glycoprotein is the Major Outer Membrane Protein (MOMP), which constitutes up to 60% of the chlamydial protein envelope. MOMP harbors the N-X-S/T sequence and its N-glycans mediate the attachment and infectivity of C. thrachomatis with HeLa cells (Kuo et al., 1996; Swanson & Kuo, 1991b; Swanson & Kuo, 1994b). This glycan is a high mannose type oligosaccharide of 8-9 mannose residues. The presence of other sugar monomers, like GlcNAc, was also suggested (Kuo et al., 1996). These findings remain, however, to be further substantiated to ascertain that the results are not origination from cross-contamination with human serum.

2.4.3 DeDicateD o-gLycosyLation of proteins: fLageLLa

C. jejuni has bipolar flagella involved in the colonization of mucus (Black et al., 1988; Logan et al., 1989). The structural proteins, FlaA and FlaB, of C. jejuni flagella are decorated with glycans. In C. jejuni 81-176, up to 19 sites of a surface-exposed region of FlaA are glycosylated (Thibault et al., 2001). This implicates the requirement of surface accessible S and T residues for the attachment of glycans (Thibault et al., 2001). The glycan modifying FlaA in many C. jejuni strains, like 81-176, is pseudaminic acid (Pse), a 9-carbon sugar similar to sialic acid, or derivatives thereof (Guerry et al., 2006). The flagellin structural protein FlaA of the related Campylobacter coli VC167 strain was found to be glycosylated on 16 sites, both with Pse and legionaminic acid (Leg) or its derivatives, which is a C-9 sugar related to sialic

30

the

swee

t too

th o

f bac

teria

acid like Pse (Logan et al., 2002; McNally et al., 2007). Whether a Campylobacter strain produces Pse, Leg or both and which derivatives it produces, is strain-specific and determined by the heterogeneous O-glycosylation loci they harbor. Flagellin glycosylation is a factor determining variety between strains and their serospecificity (Doig et al., 1996; Logan et al., 2009). FlaB is thought to be glycosylated similarly to FlaA in several C. jejuni strains (Logan, 2006).

The genetic locus responsible for the glycosylation of flagella in C. jejuni NCTC 11168 comprises nearly 50 genes and encompasses the genes involved in Pse biosynthesis, Maf (Motility accessory factor) genes, genes encoding the structural subunits FlaA and FlaB and lots of genes with unknown function (Gundogdu et al., 2007; Parkhill et al., 2000). Of note, the related strain C. jejuni 81-176 has only half the amount of genes, namely 24 genes, involved in flagellar O-glycosylation (Guerry et al., 2006).

Genes orthologous to sialic acid biosynthesis genes regulate Pse biosynthesis (neuA2, neuB3, neuB2, neuC2, neuA3) (Chou et al., 2005; Schoenhofen et al., 2006c). Leg biosynthesis is carried out by 11 ptm genes and uses particular GDP-linked sugar residues (Schoenhofen et al., 2009). The maf genes (maf1 – maf7) are members of the 1318 motility accessory factor family of flagellin-associated proteins and some (the identical maf1 and maf4 genes) harbor polymorphic G tracts, which enable motility variation by slipped-strand mispair (Karlyshev et al., 2002; van Alphen et al., 2008). The maf5 gene is involved in flagella formation and is therefore invariant (Karlyshev et al., 2002). Another four paralogous genes are members of the 617 gene family and contain intragenic single nucleotidic repeats, which are prone to phase variability. The exact role of the genes of both the Cj1318/maf and 617 gene family (i.e. families of highly homologous genes in the O-glycosylation cluster) remains to be elucidated (Szymanski & Wren, 2005). Strikingly, to none of the 50 genes of the O-glycosylation cluster, a GT activity is yet attributed. Up till now –to the best of our knowledge- no reports on the characterization of one or more O-GTs in C. jejuni strains were published. Recently, our group designed an in silico strategy to predict/annotate genes encoding GTs more accurately in bacterial species (Sanchez-Rodriguez et al., 2014). This study could identify four candidate genes involved in O-protein glycosylation in C. jejuni NCTC 11168, namely Cj1328, Cj1329, Cj1331 and Cj1333. Further experimental validation of these targets is needed to confirm that these are truly flagellar GTs.

The genetic variation displayed by the presence of both gene families (617 and Cj1318/maf) results in an alteration in the behavior of C. jejuni strains. For example, phase variation of the maf4 gene of C. jejuni strain 108 results in an altered glycosylation of the flagella and agglutination behavior (van Alphen et al., 2008). Other genes, like Cj1295 of C. jejuni NCTC 11168, also contain single nucleotide repeats and homopolymeric tracts, which can result in considerable structural changes in the flagellar glycoproteins (Hitchen et al., 2010).

two.

31

Figure 2.4 – Protein glycosylationFig. 2.4A. En bloc N-protein glycosylation in the model organism C. jejuni. The type example of en bloc protein glycosylation is the general N-protein glycosylation system of C. jejuni. Prior to the start of the glycosylation process, UDP-GlcNAC is modified to diNAcBac by the action of PglF, PglE and PglD. The priming GT PglC transfers this sugar moiety to and UndP carrier, followed by the successive actions of the PglA, J and H GTs adding GalNAc and Glc residues. Once finished, the glycan moiety is transferred across the inner membrane by the PglK flippase. Finally, the PglB OST then N-glycosylates the protein (Nothaft & Szymanski, 2010; Szymanski et al., 2003a; Szymanski & Wren, 2005) (and the references mentioned in the text). Fig. 2.4B. The peculiar sequential N-protein glycosylation system of H. influenzae. In H. influenzae the glycan moieties are directly attached to serines and threonines in the protein by HMW1C. Transportation across the inner membrane relies on the Sec pathway, followed by further export to the cell surface by HMW1B (Grass et al., 2003; Grass et al., 2010; Gross et al., 2008). Fig. 2.4C. Sequential O-protein glycosylation of pili, flagella and other proteins (illustrated for Fap1 of S. parasanguinis). The glycosylation of the Fap1 fimbrial protein of S. parasanguinis remains to be fully elucidated. It is currently known that the first step of the glycosylation process is catalyzed by a heterodimer of Gtf1 and Gtf2, whilst Gtf3 regulates the latter glycosylation steps. Especially the glycoprotein secretion mechanism is of note here, as it involves a SecA2-SecY2 heterodimer. The SecY2 forms a translocation channel, with SecA2 being an accessory protein. Gap1 and Gap3 are crucial accessory proteins for fimbrial biogenesis (Bu et al., 2008; Chen et al., 2004; Peng et al., 2008a; Peng et al., 2008b; Stephenson et al., 2002; Wu et al., 1998; Wu & Wu, 2011; Zhou & Wu, 2009; Zhou et al., 2010). Fig. 2.4D. The general en bloc protein glycosylation system of Neisseria sp. In Neisseria sp. the PglD, C and B or B2 enzymes modify a nucleotide-activated GlcNAc residue. The action of PglB results in the formation of DATDH, PglB2 in GATDH. These enzymes are also the priming GTs, adding these sugar residues to an UndP carrier. The successive action of PglA and PglE results in the addition of two Gal residues. The glycan is then transferred to the periplasm by a PglF flippase and ligated to the protein by the PglL/PglO OST (Borud et al., 2010; Borud et al., 2011; Faridmoayer et al., 2008; Faridmoayer et al., 2007; Hartley et al., 2011).

In C. jejuni 81-176, glycosylation was shown to be required for flagellar filament assembly as the glycans mediate interactions between subunits within the filament (Ewing et al., 2009). The flagellar glycans of the 81-176 strain also can establish interactions between different

32

the

swee

t too

th o

f bac

teria

filaments and modulate autoagglutination events (Ewing et al., 2009; Guerry et al., 2006). Differences in glycan modifications on flagella impact the formation of microcolonies on intestinal epithelial cells in vitro, which is critical for the virulence of the strains (Golden & Acheson, 2002). In C. jejuni NCTC 11168 flagellar glycosylation with Leg residues results in a change of the surface charge, important for the colonization of the colon of chickens (Howard et al., 2009). Of note, the flagellar glycans of C. jejuni NCTC 11168 are not implicated in the evasion of recognition of C. jejuni flagella by TLR-5 (de Zoete et al., 2010). Mutation of genes involved in flagellar glycosylation results in a loss of motility in C. jejuni and C. coli strains (Goon et al., 2003; Guerry et al., 1996; Logan, 2006).

H. pylori, a species closely related to C. jejuni, also produces complex polar flagella consisting of two structural glycosylated proteins, FlaA and FlaB. Both flagellins are modified with Pse: FlaA on seven sites and FlaB on ten sites (Josenhans et al., 1999; Schirm et al., 2003). The glycosylation sites lie in the central core region of the flagellin monomers, which are surface exposed in the filament (Schirm et al., 2003). In contrast to the remarkable heterogeneity of flagellin glycosylation in C. jejuni strains, the heterogeneity of isoforms and glycoforms is low in Helicobacter species. For instance, no derivatives of Pse are reported. The sheath surrounding the flagellar filaments of Helicobacter pylori probably negates the need for this variation (Schirm et al., 2003). This results in an analogous, but simpler glycosylation region compared to C. jejuni. This explains the large number of functional studies of enzymes involved in Pse biosynthesis in H. pylori, which facilitated the elucidation of the C. jejuni mechanism (Schirm et al., 2005; Schoenhofen et al., 2006a; Schoenhofen et al., 2006b; Schoenhofen et al., 2006c). The enzyme encoded by HP_0840 (PseB) was characterized as a bifunctional C6-dehydratase/C5-epimerase (Creuzenet et al., 2000; Schoenhofen et al., 2006c). The HP_0366 gene codes for the flagellar aminotransferase PseC (Schoenhofen et al., 2006a; Schoenhofen et al., 2006c). Further elucidation of Pse biosynthesis in H. pylori revealed the function of the other enzymes involved: HP_0326A or PseF is the CMP-Pse synthetase, HP_0326B or PseG the nucleotidase, PseH encoded by HP_0327 is an N-acetyltransferase and the HP_0178 gene codes for the Pse synthase (PseI) (Schoenhofen et al., 2006b). It was also shown in vitro that the modification of UDP-GlcNAc to form CMP linked Pse occurred in one step combining these six enzymes (Schoenhofen et al., 2006b). The enzyme encoded by HP_0114 is central to flagellar assembly, either via transfer of Pse to the flagellin monomer or via its function in later steps of flagellar assembly (Schirm et al., 2003). The modification of FlaA with glycans at seven sites was proven to be a general feature of H. pylori strains (Schirm et al., 2005).

Glycosylation of H. pylori flagellin is crucial for the filament assembly and bacterial motility, as mutants in the glycosylation region resulted in the absence of functional flagella (Asakura et al., 2010; Josenhans et al., 2002; Schirm et al., 2003). Flagellin is thought to be only glycosylated upon secretion (Josenhans et al., 2002). Study of a hypermotile mutant of H. pylori G27 resulted in the identification of HP_0518 as an enzyme involved in the deglycosylation of FlaA (Asakura et al., 2010). This mutant had an increased glycosylation of FlaA, which

two.

33

points towards a link between its glycosylation state and the pathogenicity of H. pylori: as this mutant can interact better with host cells and causes accelerated host cell responses, which is of course detrimental for a pathogen (Asakura et al., 2010). This also points out the need for a strict regulation of glycosylation. Recent papers report on the presence of other glycoproteins, of which many are cytoplasmic, in H. pylori strains besides the well known flagellar ones (Champasa et al., 2013; Hopf et al., 2011). An MS screening of enriched azide labeled proteins of H. pylori 26695 resulted in the identification of 125 candidate O-linked glycoproteins serving diverse functions and located in different fractions of the cell (Champasa et al., 2013).

In P. aeruginosa, two major serogroups (a and b) are discriminated, based on the expression of the flagellin structural protein, FliC. Totten and Lory reported an important discrepancy in molecular weight for FliC of P. aeruginosa type a strain PAK (52 kDa instead of the calculated 45 kDa) (Totten & Lory, 1990). Eight years later this molecular weight shift was attributed to the glycosylation of type a flagellin (Brimer & Montie, 1998). Genome comparison of the type a strain PAK and type b strain PAO1 revealed a flagellin glycosylation island of 14 genes (orfA-orfN) present in the PAK strain, organized in several operons (Arora et al., 2001). In the type b strain PAO1, this glycosylation island is replaced by three genes of unknown function and a putative GT. This putative GT shows low homology to the orfN gene present in PAK, which was recognized as a putative flagellin-GT (Dasgupta et al., 2003). Further research on the glycosylation island in type a strains revealed a high polymorphism for this cluster. Many strains encode a shorter version of the island, in which orfD, E and H are polymorphic and the orfI, J, K, L, M genes can be absent (Arora et al., 2004).

P. aeruginosa type a flagellins were shown to be modified at two sites (T189 and S260) on each monomer and produce unique glycan structures. The PAK flagellin is glycosylated with a heterogeneous O-linked glycan which can comprise up to 11 monosaccharides, linked to the protein backbone via a rhamnose residue. Flagellin of another strain, JJ692, has less complex modifications: single rhamnose subunits are linked to each site. In the PAK strain, a long glycosylation island is involved in the glycosylation of the PAK flagellin, encoding the OrfA enzyme responsible for the attachment of the heterogeneous glycan and the OrfN rhamnosyltransferase. In contrast, the JJ692 strain harbors a truncated glycosylation island, similar in length to the shorter glycosylation islands found in other Pseudomonas strains. Nevertheless, these other strains can show more extensive glycosylation of flagellin compared to the single glycan substitution in JJ692. Like in C. jejuni flagellar glycosylation, genetic variation of the glycosylation island can result in altered glycosylation profiles (Schirm et al., 2004a).

It was believed for a long time that b type flagella were non-glycosylated, until it was reported that the short genomic island (4 genes) of PAO1 modifies S191 and S195 with a simple glycan of 700 Da. This glycan was identified as a deoxyhexose with a unique modification of 209 Da, which harbors a phosphate moiety. In contrast to a type flagellar glycosylation, polymorphism

34

the

swee

t too

th o

f bac

teria

of the glycosylation island is low, which results in less heterogeneous flagellar glycosylation (Verma et al., 2006).

In contrast to the observations made in C. jejuni, the function of flagellin glycosylation seems to be not involved in flagellar motility or filament assembly in Pseudomonas (Schirm et al., 2004a). Also, the flagellar glycans are not the epitope recognized by the human anti-type a flagellin monoclonal antibody (Brimer & Montie, 1998). Nevertheless, flagellar glycosylation was shown to be involved in virulence of Pseudomonas in a burn wound model (Arora et al., 2005) and triggering of the inflammatory response, as was shown by the comparison of IL-8 production by A549 cells triggered by (un)glycosylated flagellin (Verma et al., 2005). The exact in vivo role of flagellar glycosylation in Pseudomonas strains remains unclear.

The first Gram-positive organism in which flagellar glycosylation was reported, is Listeria monocytogenes (Schirm et al., 2004b). The elucidation of flagellin glycosylation in L. monocytogenes resulted from earlier reports on the discrepancy in molecular weight of its flagellin, FlaA, which suggested a posttranslational modification (Dons et al., 1992; Peel et al., 1988). In 2004, the flagellins of the listerial serotypes 1/2a, 1/2b, 1/2c and 4b were shown to be β-O-glycosylated. Up to six sites of the central surface exposed region of the flagellin monomer are glycosylated (Schirm et al., 2004b). The enzyme encoded by Lmo0688 (GmaR) was later identified as a bifunctional GT protein that transcriptionally regulates the expression of its enzymatic substrate as it harbors both β-O-GlcNAc-transferase activity and is an anti-repressor of the MogR repressor, which represses flagellar production in a temperature dependent manner. As Listeria sp. only express their flagella at low temperature, their expression is repressed at 37°C (human body temperature) to evade recognition by the immune system (Shen & Higgins, 2006). Based on mutational analyses of the GmaR GT, flagellar glycosylation in Listeria is believed to be unnecessary for the secretion, stability and motility of the flagella, nor for adhesion and biofilm formation (Lemon et al., 2007; Shen & Higgins, 2006). Further research will have to shed light on the functional importance of flagellar glycosylation in this species.

Clostridium species are Gram-positive pathogens targeting the GIT and it is speculated that their glycosylated flagella might help them to establish an interaction with the host (Twine et al., 2009), although this remains to be documented. The first reports of flagellin glycosylation in Clostridium sp. were published on C. tyrobutyricum and C. acetobuytlicum ATCC824 (Arnold et al., 1998; Bedouet et al., 1998; Lyristis et al., 2000). C. botulinum was the first human pathogenic Clostridium species in which O-glycosylation of flagellin was studied. The genomes of these species harbor genes homologous to Leg biosynthetic genes of C. coli. The flagella are O-glycosylated with a new derivative of Leg on up to seven sites per monomer. Among C. botulinum isolates, a range of glycan structures could be found (Twine et al., 2008). Strains targeting infants appear to be glycosylated with Leg derivatives, whilst other strains that have no link to infections are generally decorated with hexuronic derivatives.

two.

35

This suggests a possible link between flagellin glycosylation and the pathogenic potential of Clostridium isolates.

The peritrichous flagella of C. difficile strains and isolates, well-known GIT pathogens, are modified with O-linked HexNAc residues. This modification seems to be conserved amongst most strains. In C. difficile 630, up to seven sites of the flagellin (FliC) are modified with a HexNAc residue on which a methylated aspartic acid is linked via a phosphate bond. C. difficile clinical isolates have more heterogeneous glycans consisting of a maximum of five monosaccharides, again starting with a HexNAc residue. In C. difficile 630, the GT CD0240 was further studied: inactivation of this gene led to a non-motile phenotype, which could only produce a limited amount of unglycosylated FliC. This finding illustrated the need for glycosylation of flagellin in the assembly and motility of flagella (Twine et al., 2009).

The bacterial agent causing syphilis, Treponoma pallidum, expresses flagella composed of core proteins, FlaB, and sheath proteins, FlaA. It was shown that the core protein, FlaB, is glycosylated (Wyss, 1998). The nature of its flagellar glycosylation has yet to be further investigated, but a role for FlaB glycosylation was speculated in the regulation of the assembly of FlaA on the core flagellin and in the transport and stability of flagellar components.

Taken together, flagellar glycosylation has been reported in many species. In species harboring simple flagella, like Pseudomonas, glycosylation appears a non-essential feature, while glycosylation appears to be essential for the assembly and motility of complex flagella, such as in Campylobacter, Helicobacter, Clostridium species. This latter fact seems to be typical for strains colonizing the human GIT. Nevertheless, more studies are needed to understand the functional importance of flagellin glycosylation.

2.4.4 DeDicateD o-gLycosyLation of proteins: piLi

Similar to glycosylated flagella, bacterial pili are often glycosylated. Some bacteria use a general system to do so, targeting both pili and other proteins (i.e. general O-glycosylation, see further), others harbor a mechanism dedicated to pili glycosylation.

We described earlier the glycosylation of flagella in certain Pseudomonas strains, while other strains express type IV pili and some decorate these pilins with glycans. The best-studied glycosylated pilins are from P. aeruginosa 1244. Each PilA subunit of the pilus fiber is decorated with a single trisaccharide (Castric et al., 2001). The glycan is evenly distributed over de fiber surface, at the seam between two adjacent pilins (Smedley et al., 2005). The trisaccharide consists of 5-N-β-OHC47NfmPse, xylose and FucNAc (Castric et al., 2001) linked to the C-terminal serine 148 (DiGiandomenico et al., 2002). This S148 lies adjacent to a disulfide loop, which is an important B-cell epitope (Comer et al., 2002). The terminal character of the pilin glycosylation is quite unusual, as proteins are mostly glycosylated on their core. Further comparison with a non-glycosylating strain PA103 resulted in the identification of this terminal serine as the major pilin glycosylation recognition feature. The

36

the

swee

t too

th o

f bac

teria

compatibility of the surface also plays a role, as a positive charge of the adjacent disulfide loop appears to enable glycosylation and the terminal serine needs to be at a minimum distance from the pilin subunit, probably to avoid steric hindrance of the disulfide loop (Horzempa et al., 2006a). These findings suggest that glycosylation occurs after pilin folding, early in the pilus biogenesis. Further investigation of pili glycosylation specificity revealed that only the structure of the sugar at the reducing end of the trisaccharide (FucNAc) is of importance, suggesting that the structural aspects of the remaining sugars are unimportant (Horzempa et al., 2006b).

The trisaccharide made of 5-N-β-OHC47NfmPse, xylose and FucNAc is similar to the O antigens of the LPS molecules of P.aeruginoa 1244, which suggests a common metabolic origin for both oligosaccharides. And indeed, a mutant affected in O antigen synthesis resulted both in the loss of O antigens and pilin glycosylation (DiGiandomenico et al., 2002). Heterologous expression of the O antigen producing machinery in E. coli further confirmed these findings. The fact that P. aeruginosa 1244 uses the same pathway to generate glycans for LPS formation and pilin glycosylation is probably an energy saving process.

The structural pilA gene was found to be part of an operon also harboring pilO, which was shown to be required for pilin glycosylation (Castric, 1995). This O-OST has a sequence pattern that is also found in WaaL (LPS ligase). The two enzymes compete for the same substrate, namely lipid linked monomeric O antigens, so it is no surprise that they are part of the same Pfam Wzy_C family (Musumeci et al., 2014; Qutyan et al., 2007). Nevertheless, the active site of PilO differs from the other enzyme. The Wzy domain of PilO and C-terminal proximal hydrophilic portion of PilO are the regions linked to enzyme activity (Qutyan et al., 2007). PilO has a relaxed glycan specificity towards short oligosaccharides, which prevents the addition of polysaccharides that would impede pilus assembly (Faridmoayer et al., 2007). The factors determining this remarkable preference for short oligosaccharides remain to be established. The only substrate specificity that could be found for PilO is the presence of a C-terminal serine or threonine residue, and the earlier mentioned minimum distance form the pilin surface and a compatible surface charge (Qutyan et al., 2010).

PilO was shown to be the only pilin specific glycosylation factor (Castric, 1995; DiGiandomenico et al., 2002). Mutational analysis of PilO resulted in non-glycosylated pili, which had a normal appearance but had somewhat reduced twitching motility and were more sensitive to pili-specific bacteriophages. Competition of this mutant with the wild type 1244 strain in a mouse model of acute pneumonia resulted in a greater survival for the strain carrying glycosylated pilins. This indicates that the glycosylation of the pilins is an important virulence factor that can interfere with infection establishment (Smedley et al., 2005). These findings were in concordance with earlier reports on the higher frequency of Group I Pseudomonas strains, such as 1244 and PAO1, in cystic fibrosis patients (Kus et al., 2004). There are in total five groups of Pseudomonas strains, which are defined based on the comparison of the DNA sequence

two.

37

comparison of the pilin clusters (Castric & Deal, 1994). The glycans on the pilin reduce the hydrophobicity of the pilin, which enables these strains to colonize other microenvironments (Smedley et al., 2005).

Another P. aeruginosa strain harboring glycosylated pilins is the Pa5196 strain of Group IV, later glycosylated proteins were also found in strain PA7 (Kus et al., 2008; Voisin et al., 2007). These pilins are decorated with a peculiar homo-oligomer of α(1-5)linked D-arabinofuranoses. This uncommon sugar also occurs in lipoglycans of Mycobacteria. The D-arabinofuranose biosynthesis of Pa5196 is also highly homologous to M. tuberculosis (Harvey et al., 2011). The arabinofuranoses are present on several positions in the flexible central part of the protein (Voisin et al., 2007). The T glycosylation sites are modified with trisaccharides and additional mono- and disaccharides can be found on serine residues, which add up to about 16 arabinoses per pilin subunit. The TfpW enzyme was identified as the pilin arabinosyltranferase and is thus a potential O-OST (Kus et al., 2008), which points towards an en bloc glycosylation mechanism. The glycan substitutions are required for the efficient assembly of the pilus and more in particular to establish subunit-subunit interactions (Harvey et al., 2011).

A peculiar case of glycosylation of pili (or fimbriae) can be found in Streptococcus parasanguinis FW213, where the fimbriae-associated protein, Fap1, which was later identified as the major constituent of its fimbriae, is O-glycosylated (Stephenson et al., 2002; Wu et al., 1998) (Fig.2.4C.). The Fap1 protein is a high molecular weight, serine-rich repeat protein of 200 kDa and is decorated with Rha, Glc, Gal, GlcNAc and GalNAc residues (Stephenson et al., 2002). The fimbriae of S. parasanguinis are important for the adhesion to teeth and the sugar residues present on Fap1 mediate this process (Stephenson et al., 2002; Wu et al., 2007b). Glycosylation is essential for biofilm formation of S. parasanguinis (Wu et al., 2007b). The glycans on Fap1 also protect the protein from degradation and stabilize its conformation (Chen et al., 2004).

Upstream of the fap1 gene, a biogenesis locus can be found comprising 3 orfs (gap1-3), genes coding for putative GTs (gtf1, gtf2, gly, nss, galT1 and galT2), a SecA2 and SecY2 system (Zhou & Wu, 2009). This SecA2 system is an accessory secretion system (next to the canonical SecA pathway) and allows the transport of glycosylated proteins, as glycosylation of Fap1 occurs intracellularly (Chen et al., 2004). The SecY2 protein not only forms the translocation channel, but is also thought to play a role in the complete glycosylation of Fap1 (Wu et al., 2007a). Also, the canonical SecA is thought to interact with the accessory SecA2 pathway (Zhou et al., 2011). The gap3 (orf3) gene of the locus upstream form fap1 plays a role in the glycosylation of Fap1 by linking this to the secretion of the protein. Gap3 is therefore of importance for fimbrial assembly, adhesion and in vitro biofilm formation (Peng et al., 2008a; Peng et al., 2008b). Gap1 also plays a role in the biogenesis of Fap1 and mediates glycosylation. To do this, the protein requires a conserved C-terminal 13 amino acid motif (Zhou et al., 2008). Taken together, Gap1 and Gap3 are accessory proteins interacting with

38

the

swee

t too

th o

f bac

teria

SecA2 and mediating Fap1 biogenesis in this manner (SecA2-Gap1-Gap3 complex) (Zhou et al., 2011) (Fig.2.4C.).

The Gtf1 GT catalyzes the first step in Fap1 glycosylation and adds GlcNAc (Bu et al., 2008; Wu et al., 2007a; Zhou et al., 2008). Gtf1 interacts via its N-terminal domain (DUF1975) with Gtf2 and the formation of this heterodimer complex was proven to be required for optimal GlcNAc transferase activity (Bu et al., 2008; Wu & Wu, 2011; Wu et al., 2010) (Fig.2.4C.). Gtf2 acts thus as a molecular chaperone via the stabilization, subcellular localization and modulation of the activity of Gtf1 (Wu & Wu, 2011).

The second step in Fap1 glycosylation is carried out by the product of the nss gene, which was renamed to Gtf3 (Zhou et al., 2010) (Fig.2.4C.). This Gtf3 forms a new subfamily of GTs and only shares homology to other Streptococcus sp. harboring similar glycosylated serine-rich proteins (Zhu et al., 2011).

The here-described system will be discussed further in the section on O-glycosylation of proteins in bacteria, as the association of a high molecular weight serine-rich repeat adhesion with an accessory SecA2 locus and GTs is highly conserved in Streptococcal and Staphylococcal species (Zhou & Wu, 2009).

In Streptococcus salivarius sp. K+ and K- strains can be distinguished, the former having long fibrils, the latter fimbriae. Antigen B and C, present on the fibrils of K+ cells are glycoproteins (Weerkamp et al., 1986) and the fimbriae of K- strains are assembled from glycoproteins that form a filamentous structure resistant to dissociation (Levesque et al., 2001).

2.4.5 DeDicateD o-gLycosyLation of otHer proteins

The inability to glycosylate proteins of the most studied E. coli strains is widely recognized. E. coli is therefore extensively used as a vehicle for the heterologous expression of glycosylation systems to confirm their ability to produce glycosylated proteins and in glycoengineering. Nevertheless, some glycoproteins were reported in some, mainly pathogenic, E. coli strains. Already in 1968, an envelope specific glycoprotein was identified in E. coli strain B (Okuda & Weinbaum, 1968), followed in 1975 by the publication of the glycosylation of the conjugative F pili (Tomoeda et al., 1975). The EDP208 and ColB2 F pili of E. coli K12 were later further investigated, the former were found to be unglycosylated and the glycosylation status of the latter is still uncertain (Armstrong et al., 1981). Strikingly, these early reports of E. coli glycoproteins were not investigated further. In 2003, a Gp45 membrane glycoprotein was identified in E. coli PCRO 1687 containing mannose, glucose, galactose and GalNAc residues (Kumar et al., 2003).

Better characterized are the self-associating autotransporters of E. coli, which are glycoproteins. Three autotransporters were identified in different strains: AIDA-I (Adhesin Involved in Diffuse Adhesion), TibA and Ag43. All are important virulence factors modified with

two.

39

heptose residues that are added by a single heptosyltransferase. These GTs are believed to be functionally interchangeable between strains as they have similar specificities (Charbonneau et al., 2007; Moormann et al., 2002; Sherlock et al., 2006).

In the enterotoxigenic E. coli (ETEC) strain H10407, a glycosylated TibA outer membrane protein was identified (Elsinghorst & Weitz, 1994; Lindenthal & Elsinghorst, 1999). Immediately upstream of the tibA gene, a tibC gene was identified, coding for a heptosyltransferase (Lindenthal & Elsinghorst, 1999). The TibA protein is decorated with single, surface exposed heptose residues, which is of importance for the binding of ETEC H10407 to HCT8 cells, a human ileocecal epithelial cell line in a specific and saturable manner. This suggests that TibA is an important factor in ETEC pathogenesis, involved in adhesion and invasion (Lindenthal & Elsinghorst, 2001). TibC homologues were found in several ETEC strains, but not in all (Moormann et al., 2002). In the same ETEC H10407 strain, an adhesin important for binding to intestinal epithelial cells, EtpA was also found to be glycosylated by an adjacent enzyme (EtpC). Both genes, together with an etpB gene encoding a transporter, are part of the same locus (Fleckenstein et al., 2006).

The AIDA-I adhesin was identified in the diarrhoeagenic E. coli 2787 clinical isolate and its glycosylation was found to be essential for the adhesion to human intestinal cells (Benz & Schmidt, 2001). The main role of AIDA-I glycosylation is thought to lie in the stabilization of the protein at the cell surface, offering protection against the action of the numerous proteases present in the gut (Charbonneau et al., 2007). Like in TibA glycosylation, the heptoses are added to AIDA-I by a heptosyltransferase, Aah (encoded by aidA), immediately upstream from the AIDA-I gene. AIDA-I is modified with up to 19 heptose residues, which come from ADP-glycero-manno-heptose precursors from the LPS pathway of E. coli (Benz & Schmidt, 2001). Glycosylation was found to occur at the extracellular domain, which is a β-helix and harbors imperfect repeats of 19 amino acids. Further investigation of Aah specificity, indicates that this enzyme rather recognizes a structural motif of β-helices, than a specific sequence (Charbonneau et al., 2012). Deletion of Aah resulted in a loss of adhesion, but retention of the biofilm formation and autoaggregation capacity.

The functional relation between TibC and Aah was further investigated, and indeed recombinant TibC can substitute for AIDA-I glycosylation (Moormann et al., 2002). Furthermore, AIDA-I specific antibodies can also recognize TibA. This strong homology suggests that Aah and TibC are members of a novel class of heptosyltransferases, which transfer heptose residues to multiple sites on the protein backbone (Moormann et al., 2002).

A third autotransporter is the Ag43 protein, which is widespread in entero- and uropathogenic E. coli strains, like UTI536. Ag43 is also modified with heptoses (18 glycosylation sites could be identified), but in contrast to TibC and Aah, no specific GT could be identified upstream from the flu gene (Knudsen et al., 2008; Sherlock et al., 2006). The homology with TibA and AIDA-I is high, as Aah and TibC can glycosylate Ag43. Glycosylation was found

40

the

swee

t too

th o

f bac

teria

to be important for the affinity of Ag43 for human cells (Sherlock et al., 2006). The glycans also increased the stability of Ag43 against chemical and thermal denaturation and increase refolding kinetics (Knudsen et al., 2008).

Several researchers report on the occurrence of glycoproteins in the periodontal pathogen Porphyromonas gingivalis. In the W50 strain, two proteases, RgpA and RgpB, are glycosylated (Curtis et al., 1999). These proteases are called Arg-gingipains and are extracellular cysteine proteases, degrading molecules like collagen, fibrin and fibronectin. RgpA occurs in three different forms: the heterodimer HRgpA containing an α-catalytic and β-adhesin domain, monomeric RgpA and membrane type mt-RgpA, both harboring only the α-catalytic domain. 2% of the molecular weight of the heterodimer could be attributed to glycosylation, whilst this is 14% in monomeric RgpA and even 30% in mt-RgpA. These percentages are similar for the monomeric RgpB, which can occur in a monomeric soluble form or as membrane-associated RgpB (Curtis et al., 1999; Rangarajan et al., 2005). The glycan is linked to the catalytic domain and cross-reacts with antibodies directed against Porphyromonas LPS. This indicated that there are similarities between the glycans present on the Rgp proteins and LPS structures (Slaney et al., 2002). Later, a similarity with special anionic cell surface polysaccharides (APS) was also reported (Paramonov et al., 2005). In sera of adult patients the glycans of RgpA are recognized as being immunogenic, while deglycosylation of the proteins abolishes this recognition (Slaney et al., 2002). Glycosylation of the proteins also seems to modulate their stability, as the less glycosylated HRgpA has a lower half time compared to RgpA (Curtis et al., 1999). Aberrant glycosylation had also dramatic effects on the stability of RgpA (Rangarajan et al., 2005). RgpB was found to have an effect on monomeric RgpA maturation (Aduse-Opoku et al., 1998). This is thought to occur via the activation of the glycosylating enzyme or via the induction of oligosaccharide extension (oligosaccharides of RgpA are consist of 7 to 35 sugar monomers) (Rangarajan et al., 2005). The secretion, processing, anchorage and/or activity of the gingipains are modulated by the activity of the vimA and vimE gene products, in which VimE is needed to establish proper glycan biogenesis, but is not a GT (Vanterpool et al., 2005b). Downstream of vimE, a vimF gene encodes a putative GT, which alters the glycosylation, anchorage and activity of the gingipains (Vanterpool et al., 2005a). The details of this mechanism remain to be studied further.

Another glycoprotein reported in P. gingivalis is the highly conserved outer membrane protein OMP85 of which the glycosylation was shown to influence biofilm formation (Nakao et al., 2008). The minor fimbriae of 67 kDa (encoded by the mfa1 gene) were also shown to be glycosylated and target DC-SIGN (Zeituni et al., 2010). Via this binding, P. gingivalis can bind and invade DCs and elicits an immunosuppressive response. A last reported glycoprotein is the hemin binding protein 35, HBP35 (40 kDa), an outer membrane protein involved in haem utilization. Like the arginine gingipains, this HbP35 is secreted by the Por secretion system (Shoji et al., 2011).

two.

41

Aggregatibacter actinomycetemcomitans serotype b strain VT1169, an oropharyngeal pathogen glycosylates a nonfimbrial extracellular matrix adhesion, EmaA. This protein trimerizes to form antenna-like structures that can bind collagen. EmaA is modified with the O PS moiety of LPS containing fucose, rhamnose and GalNAc (Tang & Mintz, 2010; Tang et al., 2007). Strikingly, the O PS is attached in the periplasm to the protein via the activity of the O Antigen ligase WaaL (Tang & Mintz, 2010; Tang et al., 2012). The glycan was proven to be crucial for collagen binding and the protection of EmaA against proteolytic degradation (Tang et al., 2012). Earlier, the potential glycosylation of the Flp fimbriae was reported (Inoue et al., 2000).

Two O-glycoproteins were reported in the bacterium causing ehrlichiosis, Ehrlichia chaffeensis: P120 and P156 (McBride et al., 2003; McBride et al., 2000). P120 is an immunodominant and surface exposed protein modified with glucose, galactose and xylose residues, attached to its Serine rich tandem repeat units (McBride et al., 2000).

Similar to the earlier described Fap1 glycosylation in Streptococcus parasanguinis, other Streptococcus sp. and Staphylococcus sp. O-glycosylate their major serine rich repeat adhesin via accessory GTs and secretion also occurs via an accessory SecA2 system (Zhou & Wu, 2009).

Streptococcus gordonii M99 glycosylates GspB (280 kDa), a serine rich repeat protein (SRR1 and SRR2) important for the adhesion to human platelets (Bensing & Sullam, 2002). Glycosylation occurs prior to export and 70 to 100 monosaccharides are added, probably sequentially, to several sites on a GspB protein (10% of its MW), of which Glc and GlcNAc residues are the most predominantly (Bensing et al., 2004a). Similarly to Fap1 (Fig.2.4C.), a SecY2A2 locus can be found in the adjacent to the gspB gene, comprising accessory proteins (Asp1-5), SecY2, SecA2, genes involved in GspB glycosylation (gtf, orf4, gly, nss) (Takamatsu et al., 2004a; Takamatsu et al., 2004b). The large glycoprotein is transported by the SecA2-SecY2 accessory system (Bensing & Sullam, 2002). Unglycosylated GspB can also be transported by the canonical Sec system (Bensing et al., 2005). Feltcher and Braunstein recently reviewed the accessory SecA2 system (Feltcher & Braunstein, 2012).

The accessory proteins are involved in the mediation of GspB secretion (Takamatsu et al., 2004b). Asp2 and 3 bind the serine rich repeats of GspB and are chaperones of the early phases of GspB transport (prior to full glycosylation) (Yen et al., 2011). Findings suggest that Asp2 is a bifunctional protein, which modulates, next to the transport, also the deposition of GlcNAc on the serine-rich repeats of GspB (Seepersaud et al., 2012). Asp4 and Asp5 are functional homologues of the canonical SecA pathway proteins SecE and SecG (Takamatsu et al., 2005b). Gtf and the protein encoded by orf4 are required for the glycosylation and translation or stability of GspB and were renamed respectively to GtfA and GtfB (Takamatsu et al., 2004a). These are homologous to the Gtf1 and Gtf2 initiating enzymes of S. parasanguinis and are most probably performing the same role in GspB glycosylation (Takamatsu et al.,

42

the

swee

t too

th o

f bac

teria

2004a; Wu & Wu, 2011). Gly adds the terminal sugars to GspB and its N-terminal domain shares homology to known GTs. Nss is thought to be involved in the biosynthesis of UDP-sugar precursors (Takamatsu et al., 2004a; Takamatsu et al., 2004b).

The DL1 strain of S. gordonii also expresses a serine-rich sialic acid binding adhesin (like GspB), termed Hsa (Bensing et al., 2004b; Takahashi et al., 2004). Both Hsa and GspB bind sialic moieties on the platelet membrane protein Ibα, the low molecular weight mucin MG2 and agglutinin present in saliva (Takamatsu et al., 2005a), showing that glycoproteins can also have lectin activity themselves.

In S. sanguis, adhesion to platelets is also mediated by a serine-rich protein, SrpA, homologous to Hsa and GspB from S. gordonii (Plummer et al., 2005). Other reported glycoproteins in this S. sanguis strain include a fibrillar glycoprotein in strain 12 (Morris et al., 1987) and a Platelet Aggregation-Associated Protein (PAAP), which is thought to be N-glycosylated with rhamnose residues (Erickson & Herzberg, 1993). This would be the first example of N-glycosylation in Gram-positive species, although more experiments are needed to confirm the N-glycosylation status of this protein.

A homologue of the Gtf1-Gtf2 system of S. parasanguinis and S. gordonii is also present in S. pneumoniae and glycosylates the Pneumococcal serine-rich repeat protein, PsrP (Shivshankar et al., 2009; Wu et al., 2010). This protein also has serine-rich repeat domains and is accompanied by the typical SecY2A2 locus (Shivshankar et al., 2009).

A last well-studied example of a serine-rich glycosylated adhesin and its accessory locus can be found in S. agalactiae NEM316 (i.e. Group B Streptococcus, GBS), where the Srr1 is glycosylated with GlcNAc and sialic acid residues prior to transport (Mistou et al., 2009). GtfA and/or GtfB are essential for the production of glycosylated Srr1, whilst the other accessory GTs are thought to play a role in the adaptation of GBS in vivo (Mistou et al., 2009).

The described conserved mechanism of the glycosylation and transport of serine-rich repeat adhesins is also conserved in Staphylococcus aureus strains. In S. aureus ISP479C the SraP platelet binding virulence factor is glycosylated (Siboo et al., 2005). An accessory SecA2 locus can be found, but Asp4, Asp5, Gly and Nss are missing. SecY2, SecA2 and Asp1-3 are involved in the surface expression of SraP and the GTs GtfA and B modulate its glycosylation (Siboo et al., 2008). Of 21 investigated isolates by Siboo and co-workers, 85% were shown to express SraP (Siboo et al., 2005).

The 60kDa immunodominant glycoprotein IDG-60 of S. mutans is modified with sialic acid, Man, Gal and GalNAc residues. This glycoprotein is essential for the maintenance of the cell wall integrity and cell shape, and thus for bacterial survival in stress conditions (Chia et al., 2001).

two.

43

Streptococcus faecium ATCC 9790 expresses a glycoenzyme that has autolytic N-acetylmuramoylhydrolase activity and is decorated with O-linked mono- and oligomers of glucose (Kawamura & Shockman, 1983). Also in L. rhamnosus GG, Lactobacillus plantarum WCFS1 and Lactococcus lactis MG1363, the glycosylation of a PG hydrolase was reported (Fredriksen et al., 2012; Huard et al., 2003; Lebeer et al., 2012b). In L. buchneri strains CD034 and NRRL, the glucosylation of putative glycosylhydrolases was reported recently (resp. LbGH25B and LbGH25N) (Anzengruber et al., 2013).

The Msp1 (p75) protein of L. rhamnosus GG is O-glycosylated with ConA-reactive sugars (Lebeer et al., 2012b). Its glycosylation was shown to have an impact on the stability of this PG hydrolase and protect against proteases. However, glycosylation was not essential for the PG hydrolyzing activity and Akt signaling capacity of Msp1 (Lebeer et al., 2012b). Acm2, the major autolysin of L. plantarum WCFS-1, is intracellularly glycosylated on its N-terminal AST domain, a domain that is rich in A, S and T residues (Fredriksen et al., 2012). Glycosylation was recently shown to be an autoregulating factor of the enzyme and have a negative impact on its N-acetylglucosaminidase activity. 21 mono GlcNAc substitutions were found in the AST domain of Acm2, which help to stabilize the enzyme against proteases (Fredriksen et al., 2012; Rolain et al., 2013). A broader screening of the glycoproteome of L. plantarum WCFS1 revealed 10 novel glycoproteins: 5 that have an intracellular nature, 2 extracellular PG hydrolases and one mucus-biding protein. Most glycoproteins are modified with 1 HexNAc residue, but in some cases hexose moieties were detected (Fredriksen et al., 2013). Based on the homology with the Gtf1 GT of Fap1 of S. parasanguinis (cf. earlier), the GtfA and GtfB GTs responsible for the O-glycosylation of Acm2 were recently elucidated (Lee et al., 2014). Taken together, these findings might point towards a general mechanism of glycosylation in this species, but this remains to be further substantiated.

2.4.6 generaL o-gLycosyLation systems

A few bacterial species contain a glycosylation system that targets a diverse set of proteins, often also pili and adhesins. In such ‘general systems’, the genes encoding the proteins substrates are not clustered with the genes encoding the GTs and transport functions, in contrast to dedicated glycosylation systems.

2.4.6.1 generaL o-gLycosyLation of neisseriaL proteins

The best-studied general O-glycosylation mechanisms are found in Neisseria gonorrhoeae or gonococcus and in N. meningitidis, also known as meningococcus (Fig.2.4D.). Both species are important Gram-negative human pathogens, the former causing gonnorhea, the latter meningitis.

Already in 1977, the first report on gonococcal pili glycosylation (strain P9) was published (Robertson et al., 1977). Dedicated research on the glycosylation of meningo- and gonococcal pili only emerged in the early 1990s (Jennings et al., 1998; Marceau et al., 1998; Marceau &

44

the

swee

t too

th o

f bac

teria

Nassif, 1999; Stimson et al., 1995; Virji, 1997; Virji et al., 1996) and has remained a hot topic in bacterial glycosylation with the discovery of the general character of the glycosylation machinery in 2009 (Vik et al., 2009).

N. gonorrhoeae strain MS II cells are covered with type IV pili and the S63 residue of this pilin subunit (PilE) is modified with a covalently bound O-linked GlcNAc-α(1-3)Gal dimer (Parge et al., 1995). The formation of the glycosidic bond between Gal and GlcNAc is catalyzed by a pilin galactosyltransferase, PgtA. In more than half of clinical N. gonorrhoeae isolates tested by Banerjee and co-workers (40 out of 70), this gene contains a poly G-tract, which renders PgtA prone to phase variation (Banerjee et al., 2002). Remarkably, the poly G-tract is present in all clinical isolates of patients with disseminated gonococcal infection (24 out of 24 isolates tested). The phase variation of PgtA is thought to be involved in the conversion of uncomplicated gonorrhea to a disseminated gonococcal infection, suggesting a clear link between PgtA and the glycosylated pili of N. gonorrhoeae as virulence factors (Banerjee et al., 2002). The phase variation of gonococcal pili glycosylation appears to enhance the immune evasion properties of the pilin glycans and thus the survival of N. gonorrhoeae in human blood (Ghosh et al., 2004; Parge et al., 1995).

In another N. gonorrhoeae strain, i.e. N400, pilin is glycosylated on S63 with a hexose linked to a proximal 2,4-diacetamido-2,4,6-trideoxyhexose (HexDATDH) (Hegge et al., 2004), with DATDH being diNAcBac (di-N-acylated-bacillosamine) (Hartley et al., 2011). Other glycoforms like DATDH, HexHexDATDH and O-acylated forms of the DATDH linked to hexoses are also found, which rounds up to a total of five different glycoforms in this N400 strain. The glycan is transferred en bloc to its substrate after assemblage on a lipid carrier (Aas et al., 2007) (Fig.2.4D.). Glycan biosynthesis is distantly related to the N-glycosylation pathway of C. jejuni (Hartley et al., 2011). DATDH is the product of the priming GT, PglB, a bifunctional acetyltransferase and phosphoglycosyltransferase, mainly characterized in N. meningitidis strains (Power et al., 2000). PlgB is in some strains and isolates replaced by PglB2, resulting in GATDH (2-acetamido-4-glyceramido-2,4,6-trideoxyhexose) sugar moieties (Kahler et al., 2001). PglC was identified as a transaminase, PglD as a dehydratase, PlgF as a flippase and PgtA was renamed to PglA. PglA is a GT that transfers the hexose moiety, i.e. galactose to DATDH. PglE is a GT involved in the transfer of the second hexose residue to HexDATDH. Both PglA and PglE are prone to phase variation, which results in the mono-, di- and trisaccharide glycoforms. Moreover PglE shows promiscuity towards Hex and HexNAc residues. The OST PglO shows a relaxed specificity towards mono-, di- and trisaccharides and mutation of this enzyme has been shown to result in the absence of glycan modifications (Aas et al., 2007). The glycan can be O-acetylated by the activity of the PglI acyltransferase enzyme, which uses AcetylCoA as a donor (Aas et al., 2007). Of interest, many N. gonorrhoeae and meningitidis strains were found to harbor another GT, PglH, which has a similar function to PglA. As this would result in a metabolic conflict with PglA, a conserved deletion inactivates PglH in most strains. This results in reduced glycan diversity

two.

45

as the action of PglH results in unique disaccharide products modified with glucose residues instead of galactose (Borud et al., 2011; Johannessen et al., 2012). Taken together, PglD, C and B synthesize the UDP-sugars and PglB, A, E and H form the core locus of glycan assembly, whilst the PglO OST transfers the finished glycan en bloc to the substrate (Borud et al., 2011; Hartley et al., 2011) (Fig.2.4D.).

N. gonorrhoeae strains can thus modulate their glycan diversity to enhance their virulence by the expression of a range of glycoforms resulting from their ability to express mono-, di- and trisaccharides, use both GATDH (PglB2) and DATDH (PglB), potentially acetylate (PglI) the sugars and the exploitation of PlgH and the phase variability of PglA and PglE (Hartley et al., 2011).

In 2009, the en bloc glycosylation mechanism of gonococcol pilins was shown to be a general system targeting -in addition to pili- also several membrane associated proteins (Vik et al., 2009). 11 glycoproteins were identified, sharing a S-A-P-A motif that was not sufficient, nor necessary for glycosylation. The glycoproteins include the disulfide oxidoreductase DsbA, various proteins involved in electron transport systems and redox components and proteins involved in solute uptake. These findings implicate the promiscuous character of PglO having various substrates (Vik et al., 2009). In 2012, another six new glycoproteins were reported, with PilQ, a secretin critical for the ability of pili to reach the cell surface, as most important example (Anonsen et al., 2012b; Collins et al., 2001).

The PilE pilin was not only shown to be glycosylated, but it is also modified with phosphocholine and phophoetanolamine residues on Ser 68 (Hegge et al., 2004). The clustering of both modifications on pilin peptides results in a dynamic interplay, and even a direct substrate competition between phosphoform modifications and O-linked glycans (Anonsen et al., 2012a).

Functionally, the glycan chain length modulates differential effects on gonococcal growth, while glycoform variation affects the intrinsic processes of pilus dynamics and interactions between subunits (Vik et al., 2012). Also the glycosylation of pilin has been linked to the activation of Complement Receptor 3 (CR3) of primary cervical epithelial cells by N. gonorrhoeae 1291 (Jennings et al., 2011) and adhesion of N. gonorrhoeae F62 (Gubish et al., 1982). Neisseria strains typically also produce truncated soluble pilins, so-called S-pilins, of unknown function (Marceau & Nassif, 1999). These S-pilins were found to be more abundant in gonococcal strains and glycosylation of the Ser 63 residue further increased, but was not necessary for, the S-pilin production. These findings are in contrast to these in N. meningitidis, where glycosylation of Ser 63 is crucial for the production of S-pilins. S-pilin production is thus divergent in both species (Marceau & Nassif, 1999).

The general O-glycosylation mechanism of N. meningitidis is also well studied and similar to the one in N. gonorrhoeae. In 2009, Ku and co-workers could demonstrate that the pilin

46

the

swee

t too

th o

f bac

teria

glycosylation mechanism of N. meningitidis is also a general system (Fig.2.4D.), with the discovery of the C-terminal glycosylation of the nitrite reductase, AniA, in the C311 strain (Ku et al., 2009).

Pilin glycosylation in N. meningitidis was first noted by the discrepancy in molecular weight of the pilins of N. meningitidis strains C311 and MC58 (Virji et al., 1993). Further investigation of the PilE pilin of the C311 strain revealed an O-linked trisaccharide consisting of a terminal β(1-4)-linked diGal covalently linked in an α(1-3) manner to a DATDH sugar which is directly attached to the pilin (Stimson et al., 1995; Virji et al., 1996), likely on the S63 residue (Marceau et al., 1998). Class I N. meningitidis strains show O-glycosylation of S63 with the Gal α(1-3)GlcNAc disaccharide and are highly similar to some gonococcal glycosylated pili (Marceau et al., 1998; Marceau & Nassif, 1999). Pili glycosylation is variable between meningococcal strains and isolates, but most have covalently linked galactose residues modifying their pili (Virji et al., 1999).

In N. meningitidis strain MC58, PglA was identified as the pilin specific galactosyltransferase responsible for the formation of the α(1-3) linkage between galactose and DATDH (Jennings et al., 1998). PglA also shows substrate flexibility towards both Gal α(1-3) DATDH (C311 N. meningitidis strain) and GalGlcNAc (N. gonorrhoeae and e.g. 8013SB N. meningitidis strain) (Warren et al., 2004). Mutation of PglA had no effect on pilus-mediated adhesion to human epithelial and endothelial cells. This enzyme has a homopolymeric G tract and is thus phase variable (Jennings et al., 1998). PglA of N. gonorrhoeae and N. meningitidis catalyze similar linkages using distinct substrates and are highly homologous (96% sequence identity). The N. gonorrhoeae enzyme can be constitutively expressed or show phase variability in some cases, the meningococcal enzyme in contrast is always phase variable (Banerjee et al., 2002). As mentioned earlier, also N. meningitidis strains can express PglH, a GT of which the activity results in unique glycans, but which is inactive in most strains (Borud et al., 2011; Johannessen et al., 2012).

PglE was identified in the C311 strain as a galactosyltransferase, catalyzing the transfer of the terminal galactose. The gene encoding this enzyme has up to 24 copies of a heptant repeat, which results in a phase variation between di- and trisaccharide structures. The number of heptant repeats varies between different strains (Power et al., 2003).

Also in C311, the clustered PglB, C and D enzymes were identified as being involved in the biosynthesis of DATDH, with PglB being the GT that transfers the sugar from the activated nucleotide directly to the lipid carrier. PglB was found to be bifunctional as it also acetylates the C2 or C4 of the diacetamido sugar. The PglC aminotransferase transaminates C2 or C4 and PglD is a dehydratase acting on C2, C4 or C6 (Power et al., 2000). PglB, C, D and A are present in all Neisseria species, which suggests a common pili glycosylation mechanism (Jennings et al., 1998; Power et al., 2000). PglI was also found to play a role in DATDH biosynthesis and shows homology to acetylases (Warren et al., 2004).

two.

47

After assembly of the glycan on the lipid carrier, the UndP-linked trisaccharide intermediate is transferred by the PglF flippase to a position where the glycan can be transferred to the pilin subunit prior to pilin polymerization (Kahler et al., 2001; Power et al., 2006). More in particular, the pyrophosphate linker between the glycan and the carrier needs to be adjacent to the OST (Faridmoayer et al., 2008). The PglF enzyme is similar to the WaaL enzyme involved in E. coli O antigen biosynthesis (Power et al., 2006). The involvement of a lipid-linked oligosaccharide intermediate in the glycosylation process was confirmed by mutants having a blockage in the pilin glycosylation pathway: these mutations also interfered with capsular transport and assembly (Kahler et al., 2001).

After translocation of the lipid-linked oligosaccharide in the periplasm, the PglL OST transfers the glycan to the substrate (Fig.2.4D.). Expression of PglL, pilin and diverse UndP-linked glycans of the MC58 strain in E. coli resulted in glycosylated pilins, which proved the relaxed specificity of PglL and its capability to transfer diverse oligo- and polysaccharides. The only requirement seems to be the translocation of the saccharides into the periplasm (Faridmoayer et al., 2007). PglL shows 95% identity to the gonococcoal OST PglL (Faridmoayer et al., 2007) and is also similar to the O antigen ligase of E. coli (Power et al., 2006). As mentioned earlier, these O-OSTs represent a novel family of which the members share a homologous domain of 30 amino acids, which is also present in the WaaL LPS enzyme (Faridmoayer et al., 2007).

The promiscuity of PglL was further tested and resulted in the finding that PglL can transfer virtually any glycan from an UndP carrier to pilin both in E. coli and Salmonella (Faridmoayer et al., 2008). PglL could even interfere with peptidoglycan biosynthesis (Faridmoayer et al., 2008). It was thought that the substrate specificity of PglL primarily lies in the lipid carrier. More in particular PglL recognizes the C1 and O1 of the monosaccharide at the reducing end of the glycan, the pyrophosphate linker and the first few carbon atoms of the lipid carrier (Faridmoayer et al., 2008). Later studies showed that the catalytic efficiency of PglL is linked to the length and conformation of the acyl chain of the glycan donor. PglL can even act as a Leloir enzyme and transfer sugar moieties directly from activated nucleotide donors. This means that the lipid carriers influence glycosylation, but are not essential for the PglL function (Musumeci et al., 2013). Taken together, this proves the extreme promiscuity of PglL, which opens new avenues in glycoengineering.

Nearly half of the clinical isolates have an insertion in the pglBCD operon, more particular in pglB, which results in a different glycosylation of the pili (Power et al., 2003). This insertion results in a different form of PglB, called PglB2, which encodes a glyceramidotransferase (Chamot-Rooke et al., 2007). PglB2 has a functional distinct C-terminal, resulting in a different modification of the acetamido sugar (Kahler et al., 2001). The net result is a GATDH sugar instead of DATDH (Chamot-Rooke et al., 2007). Moreover, PglB2 can be expressed by two different alleles, which results in extra glycan diversity (Viburiene et al., 2013).

48

the

swee

t too

th o

f bac

teria

Like in N. gonorrhoeae, the glycosylation of pilin is susceptible to variation, both by the gene complement expressed by strains and by the phase variability of the genes present (Power et al., 2003). For example, PglE and F are present in all strains studied, in contrast to PglG, H and B2, which are not present in C311, but are expressed in a large number of clinical isolates. Moreover PglG, H and B2 are potentially phase variable (Johannessen et al., 2012; Power et al., 2003; Viburiene et al., 2013). Together with the extreme promiscuity of the PglL OST, this results in an enormous glycan repertoire, which infers an evolutionary advantage to N. meningitidis (Borud et al., 2010; Faridmoayer et al., 2008). In analogy to N. gonorrhoeae, it was hypothesized that this glycan modification serves as a cloak for host immune responses (Faridmoayer et al., 2008; Hamadeh et al., 1995; Marceau et al., 1998). This is illustrated by the binding of anti-Gal antibodies to the pilin of N. meningitidis, which results in the blockage of complement-mediated lysis of the cells (Hamadeh et al., 1995). The diversification of the glycans present in N. meningitidis challenges the selection by the innate and adaptive immune system of the host (Borud et al., 2010). The pili of N. meningitidis are structurally variable and involved in adhesion processes (Virji, 1997; Virji et al., 1993). Nevertheless, glycan modification of the pili appears to play no major role in piliation and adhesion, but facilitates the solubility of pilin monomers (Marceau et al., 1998; Power et al., 2003).

2.4.6.2 General protein O-glycosylation in Bacteroidetes

Another important example of general O-glycosylation of proteins in bacteria can be found in the order of the Bacteroidales, where these systems were found across the whole Bacteroidetes phylum (Coyne et al., 2013), Flavobacterium meningosepticum (Plummer et al., 1995) and Tannerella forsythia (Posch et al., 2011).

B. fragilis is a common inhabitant of the human colon (10-20% of the colon microbiota) and uses a mammalian-like pathway to decorate numerous surface capsular polysaccharides and glycoproteins with L-fucose (Coyne et al., 2005; Fletcher et al., 2009). Fucose is an abundant surface modification of intestinal epithelial cells and the coordinated expression of similar surface molecules by the colonic symbiont and its host, i.e. molecular mimicry, result in a competitive colonization advantage for the B. fragilis symbiont (Coyne et al., 2005). It was found that microorganisms, and B. fragilis in particular, even stimulate the terminal fucosylation of glycoproteins and –lipids of intestinal epithelial cells (Bry et al., 1996). B. fragilis strain 9343 can cleave these L-fucose residues to internalize them and use as an energy source. More strikingly this strain can convert these exogenously acquired sugars to GDP-L-fucose to incorporate in its own surface capsular polysaccharide structures and glycoproteins (Coyne et al., 2005).

In 2009, a general O-glycosylation system in B. fragilis 9343 was characterized, glycosylating proteins with roles in chaperone requiring processes (BF0994 and BF0447), protein-protein interactions (BF2334 and BF2494), peptide degradation (BF0935 and BF3918) and surface lipoproteins (BF0522 and BF3567). The eight identified glycoproteins were all present in

two.

49

the periplasm or outer membrane and shared a common three residue motif, which was later identified as D-(S/T)-(M/I/A/L/T/V). The residues are thought to be involved both in catalysis of O-glycosylation and recognition (Fletcher et al., 2011; Fletcher et al., 2009). Engineering of this motif in naturally unglycosylated proteins resulted in site-specific glycosylation, which proves the glycoengineering potential of this strain (Fletcher et al., 2011).

An in silico screening of the proteome of B. fragilis based on the presence of this glycosylation motif and the fact that a signal peptidase I or II cleavage site or transmembrane spanning domain should be present (periplasmic and outer membrane localization of eight earlier confirmed glycoproteins), predicted 1021 candidate glycoproteins. Twelve of these were confirmed experimentally and four of them were predicted to localize at the inner membrane. Remarkably, four of the confirmed glycoproteins have roles in cell division and chromosomal segregation (BF0066, BF0252, BF0289 and BF0586), of which two are encoded by essential genes (FtsI, BF0252 and FtsQ, BF0259) (Fletcher et al., 2011). This may point towards a role for protein glycosylation in divisome regulation, which is in accordance with predicted glycoproteins in L. rhamnosus GG and C. jejuni (Sanchez-Rodriguez et al., 2014). Later, the number of essential candidate glycoproteins was augmented to 43 (Coyne et al., 2013).

The confirmation of the glycosylation status of all 12 extracytoplasmic potential glycoproteins carrying the glycosylation motif may implicate the presence of hundreds of extracytoplasmic glycoproteins, which would account for more than half of the total proteins present in the extracytoplasmic fraction, i.e. 365 out of 762 proteins (Fletcher et al., 2011).

The glycan added to the proteins is an oligosaccharide of nine units, which contains a fucose and is attached via a hexose (e.g. mannose) to the protein substrate (Posch et al., 2013). Further investigation of the glycan revealed the presence of a core and outer glycan, which are independently built. The genetic region (lfg gene locus, see below) encoding proteins involved in the biosynthesis of the outer glycan is conserved within a species, but divergent between species, whilst the genetic region for the core glycan is more conserved among species (core glycan is absent in an gmD-fclΔfkp) (Coyne et al., 2013). The enzymes (GTs and Wzx flippase) responsible for the en bloc glycosylation of proteins are clustered in the lfg gene locus. Strikingly, albeit the high homology between O-OSTs, no OST could yet be uncovered (Fletcher et al., 2009). The transcription of this locus is linked to metG (encoding a methyionyl-tRNA synthetase) transcription, which links protein synthesis and glycosylation (Coyne et al., 2013).

This general en bloc O-glycosylation system is central to the physiology of Bacteroides as a substantial amount of proteins in diverse locations and with a range of biochemical functions were found to be glycosylated. Moreover, no unglycoyslated forms of these proteins could be recovered. Glycosylation seems to be vital for protein synthesis and deletion of the glycosylation machinery was shown to result in a strain deficient in growth and unable to competitively colonize the GIT (Coyne et al., 2013; Fletcher et al., 2009). It is likely that this

50

the

swee

t too

th o

f bac

teria

machinery confers a competitive colonization advantage to these species (Coyne et al., 2005; Fletcher et al., 2011). Functionally, the glycosylation of proteins is thought to confer labels for binding and recognition and improve the stability, solubility and resistance of proteins against proteases (Fletcher et al., 2009).

The glycosylation machinery is conserved among Bacteroides species (Fletcher et al., 2009) and even phylum-wide: the four diverse classes of Bacteroidetes also harbor the O-glycosylation machinery, i.e. Bacteroidia (B. fragilis), Flavobacteria (F. meningosepticum), Sphingobacteria and Cytophaga (T. forsythia) (Coyne et al., 2013; Plummer et al., 1995; Posch et al., 2013). This points towards the early emergence of the general O-glycosylation system in evolution, before the divergence of the four Bacteroidetes classes. It is believed that the mechanism is maintained in view of its physiological importance to the diverse species of the phylum (Coyne et al., 2013).

Both B. fragilis and T. forsythia use an identical glycosylation motif, which allows the heterologous expression of glycoproteins of one species in the other (Posch et al., 2013). T. forsythia is a periodontal pathogen covered with an S-layer built up by two glycoproteins that show no homology to other known S-layer (glyco)proteins: TfsA and TfsB (Lee et al., 2006). Later also other glycoproteins like the BspA surface antigen were discovered (Posch et al., 2011; Veith et al., 2009). The oligosaccharide modifying the glycoproteins of T. forsythia contains mannose, L-fucose and Pse residues and is the result of genes encoded by an 6-8 kb conserved locus (TF2049-TF2055) (Posch et al., 2011). The S-layer of T. forsythia was identified as an important virulence factor as this helps the pathogen to evade recognition by the innate immune system of the host (Sekot et al., 2011; Settem et al., 2013).

The three amino acid glycosylation motif was also conserved in Flavobacterium meningosepticum, where several secreted proteins were found to be glycosylated at a D-S or D-T-T motif with a peculiar seven-residue oligosaccharide (Plummer et al., 1995; Reinhold et al., 1995).

2.4.6.3 FrancisELLa tuLarEnsis

F. tularensis is a pathogen causing tularemia, a severe potentially fatal zoonotic disease and is therefore classified as a potential bioterroristic agent. The bacteria expresses type IV pilin, PilA, which appears to be glycosylated based on aberrant migration on SDS-PAGE (Forslund et al., 2006). Expression of PilA in N. gonorrhoeae resulted in glycosylated pilins (Salomonsson et al., 2009). Later research confirmed pilin modification with a pentasaccharide consisting of hexoses and acylated hexoses. Glycoforms of the pentasaccharide were also found carrying unusual moieties linked to the distal sugar via a phosphate bridge (Egge-Jacobsen et al., 2011). Apart from the pilin, other proteins linked to virulence were also found to be glycosylated. Balonova and co-workers used a combination of techniques (Periodic Acid Schiff base stain, lectin blots and lectin affinity chromatography) to identify 15 new

two.

51

candidate glycoproteins, including proteins involved in cell division (FtsZ or FTSH_1830), a thioredoxin family protein DsbA (FTH_1071), OmpA family protein FTH_0323 and several chaperones (GroEL or FTH_1651 and DnaK or FTH_1167) (Balonova et al., 2010). Of note, many of these proteins were also identified by Janovska et al. (Janovska et al., 2007) as immunoreactive antigens in the live vaccine strain LVS of F. tularensis subsp. holarctica reacting with human tularemic sera. Interestingly, the glycosylated lipoproteins FTH_1071 and FTH_0414 were also previously found to interact with TLR-2/TLR-1 heterodimers of HeLa cell lines, resulting in the induction of a proinflammatory response (Thakran et al., 2008). Since interaction with these TLRs is generally mediated via acylation, glycosylation is thought to play a modulatory role, such as shown by Sieling et al. (Sieling et al., 2008). They demonstrated that glycosylation of the mycobacterial lipoglycoprotein LprG is required for the stimulation of innate immune responses via activation of MHC class II-restricted T-cells. The DsbA protein (FTH_1071), an essential virulence factor of Francisella, is modified with a hexasaccharide, which seems to be non-essential for the virulence in vivo in a murine model for tularemia (Thomas et al., 2011). Another important immunoreactive virulence factor encoded by FTH_0069 is glycosylated with the same hexasaccharide as DsbA and PilA. This oligosaccharide consists of two hexoses, 3 N-acetylhexosamines and an unknown monosaccharide containing a phosphate group (Balonova et al., 2012).

A gene cluster (FTT0789-FTT0800) was identified in the FSC200 strain, which contains three GT homologues, a GalE homologue (FTT0791), a transporter and three hypothetical proteins (Thomas et al., 2011). This cluster plays a role in glycan biosynthesis and the FTT0791 and FTT0798 gene products are implicated in DsbA glycosylation, with FTT0791 being the GalE homologue and thus a putative UDP-Gal-4-epimerase (Thomas et al., 2011).

The OST involved in the en bloc glycosylation of the protein substrates was also identified: PglA encoded by FTT0905 in FSC200. The expression of PglA is modulated by MglA, a virulence regulator, also required for pilus expression (Egge-Jacobsen et al., 2011; Zogaj et al., 2008). PglA is conserved in Francisella sp. and expression of PglA and PilA in E. coli proved to be sufficient for pilin glycosylation (Egge-Jacobsen et al., 2011). Homologues of PilA and PglA are found in all strains of the Francisella genus and the glycosylation machinery seems to be a general feature of these strains (Egge-Jacobsen et al., 2011).

2.4.6.4 acinEtobactEr baumanii

Only very recently, a general O-linked protein glycosylation system was discovered in the ‘superbug’ Acinetobacter baumanii, known as a causing agent of nosocomial infections. In A. baumanii ATCC 17978, seven glycoproteins were identified, including the OmpA protein A1S_1193, each modified with a branched pentasaccharide of GalNAc, Glc, Gal, GlcNAc and a triacetylated glucuronic acid (Iwashkiw et al., 2012b).

52

the

swee

t too

th o

f bac

teria

Strikingly, a common pathway synthesizes this pentasaccharide and the capsular polysaccharides. PglC (A1S_0061), the priming GT, initiates the production of the pentasaccharide via the transfer of a GlcNAc residue to the lipid carrier. Further assembly is carried out by four GTs encoded by A1S_0058, A1S_3482, A1S_0059 and A1S_0060. Once this oligosaccharide is finished, Wzx (A1S_0056) flips the structure to the periplasm. This pentasaccharide modifies glycoproteins and also serves as a building block for the capsular polysaccharides. The pathways diverge in the periplasm: glycoproteins are further processed by the PglL OST and capsular polysaccharides are polymerized by Wzy (A1S_3483) (Lees-Miller et al., 2013). The PglL OST involved in the biosynthesis of the glycans is encoded by A1S_3176, and shows homology to the other O-OSTs (Iwashkiw et al., 2012b). Furthermore, the maximum glycan modification found on glycoproteins is a decasaccharide, suggesting the inability of PglL to access polymerized glycans and use these as a substrate (Lees-Miller et al., 2013).

Abolishment of the glycosylation machinery in A. baumanii ATCC 17978 resulted in diminished biofilm formation, reduced virulence in two infection models (Dictyostelium discoideum and larvae of Galleria mellonella) and a reduced in vivo fitness in a mouse model of peritoneal sepsis (Iwashkiw et al., 2012b; Lees-Miller et al., 2013). Both glycoproteins and capsular polysaccharides are an important virulence factor of A. baumanii for infection of animal models.

Despite the genome plasticity of A. baumanii strains, the O-glycosylation machinery is present in all clinical isolates and all sequenced genomes. This indicates a strong evolutionary pressure to maintain this feature (Iwashkiw et al., 2012b).

2.4.6.5 pecuLiar gLycoproteins in actinomycEtEs

A peculiar general O-glycosylation mechanism is found in Actinomycetes sp., of which M. tuberculosis, M. leprae and M. bovis are important human pathogens. These species decorate their proteins with mannose residues following an evolutionary conserved mechanism similar to eukaryotic O-mannosylation processes (VanderVen et al., 2005). The identification of glycoproteins in these actinomycetes was mainly based on their reaction with the ConA lectin, which specifically binds to mannose residues (Dobos et al., 1995; Espitia & Mancilla, 1989; Garbe et al., 1993). Most of the identified glycoproteins are also acylated. Already in 1989, Espitia and co-workers identified a glycoprotein of 38 kDa and a doublet of 50-55 kDa in Mycobacterium tuberculosis H37Rv (Espitia & Mancilla, 1989). The glycosylation of the doublet was later confirmed by digestion of the glycans with the α-mannosidase from jack bean (Espitia et al., 1995). The 38kDa glycoprotein is involved in a phosphate specific transport system and is better known as PstS1 or PhoS1 (Rv0934) (Herrmann et al., 2000; Torres et al., 2001).

two.

53

A 19 kDa major membrane lipoprotein, and known virulence factor, LpqH (Rv3763), was also found to be ConA-reactive (Garbe et al., 1993; Herrmann et al., 2000) and its glycan substitutions seem to regulate the cleavage of a proteolytically sensitive linker region (Herrmann et al., 1996). The glycan confers thus resistance to proteases to LpqH and doing so retains the protein on the cell. Cleavage of the linker by contrast results in a non-acylated secreted antigen (Herrmann et al., 1996). LpqH is a major adhesin and TLR-2 agonist and was found to bind to mannose receptors of THP-1 monocytes. Doing so, LpqH was shown to promote the phagocytosis of mycobacteria (Diaz-Silvestre et al., 2005). Another adhesin, HBHA, or the Heparin-binding hemagglutinin, is also glycosylated and the glycan residues are thought to protect the lysine-rich C-terminal of the protein against proteolytic attack (Menozzi et al., 1998).

Apa (MPT32), encoded by Rv1860, is a ConA-reactive glycoprotein of 45 kDa rich in proline. 4% of its molecular weight can be attributed to the presence of glycan moieties (Dobos et al., 1995). Further research showed the presence of 5 glycopeptides in which a T residue is modified with a mannose, α(1-2) linked mannobiose or α(1-2) linked mannotriose (Dobos et al., 1996). The most common glycoforms are modified with 6, 7 or 8 mannose residues, Apa proteins modified with 3, 4 or 5 mannoses are also found, but glycoforms harboring 0, 1, 2 or 9 mannoses are seldom (Romain et al., 1999).

Recombinant expression of Apa in M. smegmatis and a non-glycosylating E. coli host resulted in a decreased capacity to stimulate T-lymphocyte responses in Guinea pigs (Horn et al., 1999). Moreover, deglycosylation of Apa generally decreased the capacity of the protein to elicit cellular immune responses in vivo and in vitro. Deglycosylated Apa is 10-fold less active in delayed type hypersensitivity reactions in immunized Guinea pigs and 30-fold less active in the stimulation of T-lymphocytes in vivo (Romain et al., 1999). These studies indicate mannose as an essential component of Apa as an antigen and in the modulation of T-cell dependent immune responses (Horn et al., 1999; Romain et al., 1999). Lara et al. showed that tuberculosis patients carry antibodies that mainly react against the carbohydrate part of Apa (Lara et al., 2004). M. tuberculosis targets the Pulmonary Surfactant Protein A (PSP-A) C-type lectin of the innate immune system mainly via its lipoglycans, but the Apa antigenic glycoprotein was also suggested to aid in the attachment to this receptor (Ragas et al., 2007).

Most of the here-mentioned glycoproteins are T-cell antigens, but the SodC superoxide dismutase (Rv0432), a B-cell antigen, was also found to be glycosylated at six residues in its N-terminal. Strikingly, apart from glycans attached to threonine residues, also glycosylated serine residues were identified, which represents the first evidence of serine linked O-glycosylation in Mycobacterium sp. (Sartain & Belisle, 2009).

In view of the evolutionary relatedness to O-mannosylation in yeast, mechanistic studies on O-protein glycosylation in M. tuberculosis were done and resulted in the identification of an O-mannosyltransferase encoded by Rv1002c, which initiates protein mannosylation

54

the

swee

t too

th o

f bac

teria

(VanderVen et al., 2005). This enzyme is a membrane protein and Sec translocation of the protein is required for its mannosylation, which also implicates the need for a lipid carrier for the glycan residues (VanderVen et al., 2005). The Rv1002c enzyme is crucial to the virulence of M. tuberculosis as mutation of the enzyme, and the subsequent loss of mannoproteins, severly impairs in vitro growth and attenuates the virulence of M. tuberculosis (Liu et al., 2013).

In the closely related strain M. bovis, a pathogen mainly targeting oxes, but also humans, some glycosylated antigens were identified (Fifis et al., 1991). The MPB83 antigen is the best-characterized one up till now and is modified at two threonine sites with mannose and α(1-3) mannobiose. This is peculiar, as the MPT32 mannose residues of M. tuberculosis are α(1-2) linked (Michell et al., 2003).

In M. leprae, no protein glycosylation mechanism has been described yet, but the LprG lipoprotein was found to be glycosylated and activates TLR-2 via its carbohydrate moiety. Moreover the glycan is required for the inhibition of IFNγ induced MHC class II T-cell activation and antigen representation by monocytes to T-cells by TLR-2 dependent mechanisms. A homologue of LprG is also present in M. tuberculosis (Rv1411c) and in the genome of M. leprae a homologue of Rv1002c was found (Sieling et al., 2008).

2.4.6.6 inDications of generaL o-gLycosyLation systems in otHer species

A conceptual study by Gebhart and co-workers revealed the presence of O-OSTs in Vibrio cholera N16961 (VC0393) and Burkholderia thailandensis E264 (BTH_I0650) (Gebhart et al., 2012). To the best of our knowledge- no glycoproteins are yet reported in V. cholera, but it was shown that it harbors the potential to express a glycosylation system in its genome (Gebhart et al., 2012). It remains to be established if this is functional in a native background.

In Burkholderia thailandensis and B. mallei species, flagellar O-glycosylation was reported earlier. In B. peudomallei, a gene of the LPS O antigen cluster (rmlB) was found to be involved in flagellar glycosylation (Scott et al., 2011a). Recently, it was hypothesized that the BimA autotransporter of B. mallei might be glycosylated by BimC (Schell et al., 2007). Taken together, Burkholderia species possess the genomic content to carry out protein glycosylation, but the nature of this system remains to be elucidated.

It is also not inconceivable that more general protein glycosylation systems yet have to be identified. The above described dedicated systems targeting flagella, pili and other proteins could target more proteins, since the elucidation of the general O-glycosylation system of Neisseria sp. was also initiated by the discovery of glycosylated pili (cf. supra). The discovery of multiple glycoproteins in L. plantarum is an example of a bacterium in which a general mechanism is very likely to be found (see earlier). Moreover, the protein glycosylation potential of a range of bacteria has not yet been studied.

two.

55

2.4.7 s-gLycosyLation of proteins

The modification of cysteine residues with glycans occurs regularly in eukaryotes. This phenomenon was only recently reported in a bacterium. Lactobacillus plantarum KW30 produces a bacteriocin, glycocin F, which is glycosylated on a serine with a GlcNAc residue and also on a cysteine residue with a HexNAc (Stepper et al., 2011; Venugopal et al., 2011). The serine-linked glycosylation was found to be essential for the bacteriostatic activity of the bacteriocin, whilst the cysteine-linked glycan enhances bacteriostasis. The glycans are also believed to protect this glycocin from proteolysis. The plantaricin ASM1 of L. plantarum A-1 and lantibiotic sublancin 168 of B. subtilis 168 are also believed to be (S-) glycosylated (Stepper et al., 2011; Venugopal et al., 2011).

2.4.8 s-Layer proteins

S-layer proteins form the outermost cell envelope of some bacterial species. These crystalline layers are built up of identical protein or glycoprotein subunits. The lattice-like S-layers are formed by an entropy-driven self-assembly process of their substituent (glyco)proteins. The mono- or multilayered S-layers confer an evolutionary advantage to the cells, but are not essential for their survival (Messner, 1997). Their self-assembling capacities make them interesting candidates to study in the field of biomimetics and glycoengineering of therapeutic targets (e.g. vaccine design, drug targeting and diagnostics) (Steiner et al., 2008). The research field dedicated to these intriguing molecules is called ‘nanoglycobiology’ (Messner et al., 2008).

It was long believed that only Gram-positive species could produce S-layers, but the first Gram-negative S-layer was discovered in 2006 in Tannerella forsythia (Lee et al., 2006). S-layers are common in Bacillaceae, like Clostridium sp. and Bacillus sp., with Bacillus stearothermophilus being a model organism (Messner et al., 2008). Many lactobacilli also produce this crystalline layer (Avall-Jaaskelainen & Palva, 2005). S-layers have been excellently reviewed (Messner, 1997; Messner et al., 2008; Schaffer & Messner, 2004), and we only provide a short overview of the current knowledge on their structure and biosynthesis mechanism in relation to their glycosylation.

Glycans on S-layer glycoproteins can be linear or branched long homo- or heteropolysaccharides composed of identical repeating units. In the bacteria studied to date, these oligosaccharides are O-linked to a serine or threonine, albeit also linkages to tyrosine residues are found. Commonly, one glycan is found per protein, which accounts for 1 to 10% of the molecular weight of the S-layer proteins. 50 to 150 sugar monomers (hexoses, amino sugars, exotic sugars…) can be found in 15-50 repeating units of 2-6 sugars.

S-layer glycans resemble O antigen carrying LPS molecules, as they consist of a tripartite structure in which the glycan chain of a variable number of repeating units (cf. O antigen) is linked via a variable core oligosaccharide to the S-layer protein backbone (versus lipid A in

56

the

swee

t too

th o

f bac

teria

LPS). The oligosaccharide is strain-specific, whilst the core is less variable. In some cases the core oligosaccharide is missing and the sugar is directly attached to the protein. The linkage sugar, i.e. the first repeating unit, can have an inverted anomeric configuration, in which case it is called a pseudocore. A non-carbohydrate constituent, like an O-methyl group, can cap the terminal sugar of the oligosaccharide. This constituent potentially serves as a termination signal, analogous to LPS biosynthesis. The proteins are water insoluble and have multiple domains: the cell wall targeting region (anchoring in the PG) and the self-assembly domain (Messner et al., 2008).

Nucleotide-activated sugars are used in the biosynthesis of S-layer glycoproteins and assembly of the oligosaccharide requires an UndP carrier. Genes involved in S-layer glycosylation are in some species clustered in a ‘S-layer glycosylation cluster’ (slg), as dedicated systems. This cluster harbors a priming GT, transporter, polymerase and OST, thus analogous to en bloc N-glycosylation. Some clusters contain a flippase, which would point towards a mechanism resembling the Wzy dependent O antigen biosynthesis.

2.5 commonaLities in gLycoconjugate biosyntHesis

The plethora of glycoconjugates produced by bacteria is, as elaborated extensively earlier, synthesized by only a limited amount of biosynthesis pathways. These pathways share important general themes, like homologous proteins, which point towards an evolutionary connectivity in glycoconjugate biosynthesis strategies. Several organisms even modify several of their glycoconjugates with the same glycans or have enzymes that are active in more than one biosynthesis pathway.

2.5.1 generaL tHemes in gLycoconjugate biosyntHesis patHways

The two most obvious general themes in glycoconjugate biosynthesis pathways have been conceptualized already earlier: the sequential (Fig.2.1) and en bloc pathway (Fig.2.2). Both were extensively illustrated using examples of several bacterial glycoconjugates (cf. Fig.2.4 and the extendend version of this chapter (Tytgat & Lebeer, 2014)). Here we zoom in on the common denominators of both biosynthesis pathways.

2.5.1.1 tHe LipiD carrier

The biosynthesis of almost all glycoconjugates requires the involvement of a lipid carrier on which the glycan moiety is built prior to its transport to its substrate. In most cases, this lipid carrier is an UndP carrier: O antigen of LPS, CPS, EPS, TA, PG and en bloc protein glycosylation (Bugg & Brandish, 1994). Even Mycobacteria sp. that have a modified PG layer (arabinogalactan) use a modified version of the UndP carrier for the assembly of their PG, being decaprenol phosphate (Wolucka et al., 1994). However, some sequential glycoconjugate biosynthesis pathways use membrane lipids such as diacylglycerol and phosphatidylinositol as carrier (e.g. CPS biosynthesis in E. coli K1) (Corbett & Roberts, 2008) and in the case of

two.

57

some membrane spanning synthases, no lipid carrier is needed (e.g. alginate biosynthesis) (Rehm & Valla, 1997; Whitney & Howell, 2013).

A priming GT transfers the first sugar moiety to the lipid carrier and catalyzes the formation of a high-energy pyrophosphate bond. This resulting undecaprenyl pyrophosphate (UndPP) lipid carrier serves as a scaffold for the further synthesis of the glycan. The lipid-linked oligosaccharide is then transported over the membrane, the phosphate-sugar bond is cleaved and the oligosaccharide is transferred to its substrate. The lipid carrier is then recycled to the cytoplasmic face of the membrane (Bugg & Brandish, 1994). A phosphatase (e.g. Wzb) removes the remaining extra phosphate after the transfer of the glycan (Valvano, 2008; Wugeditsch et al., 2001). The recycling of the UndP lipid carrier is essential for the viability of bacterial cells, because of its key role in PG biosynthesis (van Heijenoort, 2001).

As stated before, the factors determining the substrate (acceptor) and donor specificity of GTs remain largely unclear, but it is believed that priming GTs and OSTs recognize the lipid carriers (e.g. length, saturation…) (Hug & Feldman, 2011). An important exception is the PglL OST of Neisseria, which is known for its extreme promiscuity in view of lipid-linked oligosaccharides used (Faridmoayer et al., 2008).

2.5.1.2 priming or initiating gts

The activated lipid linked sugar monomer resulting from the action of the priming GT forms the acceptor for the action of other GTs that assemble the glycan further (Price & Momany, 2005).

As illustrated in the discussion of the biosynthesis of LPS structures (cf. (Tytgat & Lebeer, 2014)), there are currently two families of priming GTs characterized. The first family is the polyprenyl-P N-acetylhexosamine-1-P transferase family (PNPT), of which the E. coli WecA enzyme is a type example (Al-Dabbagh et al., 2008; Lehrer et al., 2007; Meier-Dieter et al., 1992; Price & Momany, 2005). The second family is the family of polyprenyl-P hexose-1-P transferases (PHPT) and its best-studied member is the WbaP priming GT of S. enterica serovar Typhimurium (Jiang et al., 1991; Wang et al., 1996). These families are evolutionary unrelated and have a different topology and primary sequence. The PNPT family is, besides in LPS O antigen biosynthesis, often implicated in TA (e.g. TagO of B. subtilis (Soldo et al., 2002)), PG (e.g. MraY (Ikeda et al., 1991)) and glycoprotein (e.g. WbpL of P. aeruginosa (Rocchetta et al., 1998)) biosynthesis. The PHPT family can initiate O antigen LPS (e.g. S. enterica) and EPS/CPS biosynthesis (e.g. various E.coli strains), but also glycoprotein biosynthesis (e.g. Neisseria sp.) (Power et al., 2000).

2.5.1.3 transport of tHe LipiD-LinkeD oLigosaccHariDe across tHe membrane

There are two main strategies used to transfer lipid-linked glycans across the membrane: flippases and ABC transporters. Both were discussed in detail in the extended version of

58

the

swee

t too

th o

f bac

teria

this chapter (Tytgat & Lebeer, 2014) for EPS/CPS and LPS O antigen biosynthesis, but are also involved in the biosynthesis of other glycoconjugates, like PG, TAs and glycoproteins (Fig.2.4A. and Fig.2.4D.).

The Wzx flippase system generally transfers short, complex and often branched oligosaccharides and exhibits low substrate specificity (Feldman et al., 1999). ABC transporters are involved in the translocation of longer glycan chains (Cuthbertson et al., 2005). The first system is believed to recognize the first sugar linked to the lipid carrier (Marolda et al., 2004), while ABC transporters recognize the stop elongation signal, like glycan caps or terminal methylation (Cuthbertson et al., 2007; Cuthbertson et al., 2005). The ABC transporters can consist of a single polypeptide chain (e.g. PglK of C. jejuni) or several polypeptide chains (e.g. Wzt and Wzm of E. coli) (Alaimo et al., 2006).

2.5.1.4 gLycan Ligation to tHe substrate

In the end, the glycan has to be conjugated to its substrate, being lipid A-core polysaccharide (O antigen), polysaccharide (EPS/CPS, O antigen) or protein (glycoprotein). Strikingly, these reactions are often catalyzed by evolutionary connected enzymes, all harboring a Wzy_C domain (Musumeci et al., 2014). These include WaaL O antigen ligases and O-OSTs involved in protein glycosylation. N-OSTs, like the PglB N-OST of C. jejuni, seem to be unrelated and of a different evolutionary origin (cf. earlier) (Hug & Feldman, 2011). These enzymes remove the glycan from its lipid carrier by cleaving the phosphate-sugar bond and form a new linkage with the substrate (Musumeci et al., 2014; Qutyan et al., 2007; Raetz & Whitfield, 2002). Nevertheless, it is currently very difficult to predict the substrate specificity, i.e. the glycoconjugates biosynthesis pathways in which these enzymes are involved merely from sequence information. This often results in miss annotations in full genome sequences (e.g. the annotation of GTs implicated in LPS biosynthesis in Gram-positive species) and thus probably also in metagenomic analyses.

2.5.1.5 common tHemes faciLitate new Discoveries

The fact that almost all biosynthesis pathways use homologous carriers, enzymes and transporters reflect the importance of the maintenance of these pathways from an evolutionary perspective. There is also an evolutionary connection to the eukaryotic glycobiosynthesis pathways, but an in-depth comparison of prokaryotic and eukaryotic glycosylation is beyond the scope of this work, as this is the topic of a previous review (Dell et al., 2010).

There are still many species of which the glycosylation potential has to be unraveled. New emerging techniques in MS (Ding et al., 2009; Scott et al., 2011b; Wuhrer et al., 2009), NMR (Slynko et al., 2009), analytical glycoscience (Balonova et al., 2009), single molecule force spectroscopy (Francius et al., 2008), shotgun glycomics (Song et al., 2011) and in silico glycobiology (Fletcher et al., 2011), boost the further in-depth study of new and already known glycoconjugates (Reid et al., 2010). The already known common themes can facilitate

two.

59

these studies and make it also easier to spot important exceptions and evidence which is still lacking. An example is found in B. fragilis, where a Wzx transporter was identified, but to the best of our knowledge, up till now, no OST (Fletcher et al., 2009).

2.5.2 overLap in gLycoconjugate biosyntHesis patHways

In several species the commonalities between the biosynthesis pathways of their glycoconjugates are even more elaborated. Several species use the same oligosaccharides to build up their LPS and glycoprotein structures. Other species harbor pathways encompassing shared enzymes. The most common overlapping glycoconjugate biosynthesis mechanisms are the polysaccharide (CPS, LPS, LOS) and glycoprotein biosynthesis pathways. There are two main trends, the first one being the attachment of the same glycan moieties to both glycoconjugates, the other being the competition for sugar precursors.

The first phenomenon, i.e. double usage of the same glycan, and thus biosynthesis pathway, to modify different glycoconjugates, is already illustrated in several species. A. baumanii has only one priming GT, PglC, to initiate CPS and O-glycoprotein biosynthesis (Lees-Miller et al., 2013). This results in the attachment of pentasaccharide CPS building blocks to glycoproteins (max. two linked subunits). A single locus thus is responsible for the production of both glycoconjugates (Cuccui & Wren, 2013; Lees-Miller et al., 2013). In C. jejuni NCTC 11168, both CPS and LOS consist of heptoses, which require the activity of the same GmhB phosphatase (Guerry & Szymanski, 2008; Karlyshev et al., 2005). The same glycan is also incorporated in the CPS and LPS glycans of E. coli O9a:K30 (Drummelsmith & Whitfield, 1999). But most examples of shared glycans are found between overlapping LPS and glycoprotein biosynthesis pathways. In P. aeruginosa 1244, the pili are glycosylated with the same building blocks as used for LPS O antigen biosynthesis (Castric et al., 2001; DiGiandomenico et al., 2002). Both pathways are distinguished later by the difference in specificity and active site of the competing enzymes Wzy, WaaL and PilO (Horzempa et al., 2006b). The EmaA adhesin of A. actinomycetemcomitans is modified with O antigen sugars and is a substrate of the WaaL enzyme (Tang & Mintz, 2010; Tang et al., 2007; Tang et al., 2012). The same is true in B. pseudomallei, where the glycan attached to the RgpA glycoprotein is similar to the anionic polysaccharides (Curtis et al., 1999; Paramonov et al., 2005). In E. coli 2787 the Aah GT uses heptose residues from the LPS core to glycosylate the AIDA-I adhesin (Benz & Schmidt, 2001).

As mentioned, a second trend is the competition for nucleotide activated sugar precursors between different glycoconjugate biosynthesis pathways. In P. aeruginosa strain PAO1, the LPS and glycoprotein biosynthesis pathways both use the TDP-L-Rha precursor (Lindhout et al., 2009). This is also the case in C. jejuni NCTC 11168, where the same Gne epimerase (CJ1131c) is involved in the production of Gal and/or GalNAc precursors, which are present on its N-glycoproteins, LOS and CPS structures (Bernatchez et al., 2005; Guerry & Szymanski, 2008). Glycosylation of the HMW1 adhesin of H. influenzae requires UDP-glucose and thus

60

the

swee

t too

th o

f bac

teria

the phosphoglucomutase PgmB, an enzyme also important in LOS biosynthesis (Grass et al., 2003). The same precursor is used in B. pseudomallei in LPS biosynthesis and flagellin glycosylation. The RmlB enzyme of the LPS O antigen cluster synthesizes this nucleotide activated sugar precursor (Scott et al., 2011a).

These overlapping enzymes and pathways are particular intriguing, as they render the bacteria with several advantages, but also disadvantages. As mentioned earlier, the production of glycoconjugates is an energy-costly process for bacteria. Commonalities in glycosylation pathways reduce the arsenal of precursors and enzymes needed and are thus energy saving. Bacteria also need a way to discriminate between these bifurcated biosynthesis pathways and have evolved several solutions to do so: they can express only a subset of the glycoconjugate repertoire (cf. H. pylori (Hug et al., 2010)), use downstream enzymes with different specificities (cf. P. aeruginosa (Qutyan et al., 2010)), have both pathways competing (e.g. same precursors for LOS, CPS and N-glycoprotein biosynthesis in C. jejuni (Bernatchez et al., 2005)) and their expression can be dependent on environmental stimuli (cf. expression of colonic acid in E. coli (Whitfield & Roberts, 1999)).

Common glycans can serve as a cloak to shield immunogenic molecules from the host immune system (cf. P. gingivalis (Slaney et al., 2002)). Some species exploit this by covering their surface with glycans resembling host sugars (e.g. fucose, sialic acid…). This strategy is used by commensals, like B. fragilis that harbors a surface fucosylation pathway to modify its CPS and O-glycoproteins (Coyne et al., 2005), but also pathogens like H. pylori use this to prolong their presence in the host (Hug et al., 2010). But the addition of similar antigenic glycans to more structures can also have a major disadvantage: it can enhance the recognition by the immune system. In P. aeruginosa 1244, for instance, O antigen subuntis are also added to the pilin, enlarging the pool of antigens and activating the cross reaction of antibodies (DiGiandomenico et al., 2002).

2.5.3 bifunctionaL anD promiscuous gts

Another common theme in glycoconjugate biosynthesis is the involvement of bifunctional and promiscuous GTs in several bacteria. Bifunctional GTs combine the catalysis of the formation of a glycosidic linkage with another catalytic role. Classic examples are the bifunctional PBPs used in PG biosynthesis, which have both a transglycosylase and –transpeptidase domain (van Heijenoort, 2001). In L. monocytogenes the GmaR (Lmo0688) is a bifunctional GT, which also transcriptionally regulates the expression of flagellar expression (cf. supra) (Shen & Higgins, 2006). The PglB priming GT of N. gonorrhoeae and N. meningitidis is both an acetyltransferase and a phosphoglycosyltransferase (Power et al., 2000). But the unrefuted champions of multitasking are ofcourse the synthases, which initiate, elongate and transport glycans (Yother, 2011).

Promiscuous enzymes can transfer a plethora of glycans and/or target several substrates. Most

two.

61

promiscuous enzymes are OSTs, with PglL of N. meningitidis MC58 being an exponent of this as this OST can transfer virtually any glycan (Faridmoayer et al., 2008; Faridmoayer et al., 2007). The enzyme can even interfere with the PG machinery and transfer incomplete PG subunits (muramoyl-GlcNAc tetra- and tripeptide) to pilin (Faridmoayer et al., 2008). PglL can even act as a Leloir enzyme and transfer monosaccharides (Musumeci et al., 2013). The same phenomenon could also be linked to the PglB N-OST of C. jejuni, of which the only requirement is the presence of a C-2 acetamidogroup in the non-reducing sugar (Li et al., 2010; Wacker et al., 2006). The PilO O-OST is also promiscuous, but to a lesser extent (cf. earlier) (Qutyan et al., 2007). An example of promiscuity of a non-GT is the fact that the PglK transporter of C. jejuni can complement a Wzx defect (Alaimo et al., 2006). Moreover, a homologue of PglK (WzK) was identified as the O antigen ligase in H. pylori (Hug et al., 2010).

2.6 concLusions

The field of bacterial glycosylation remains enigmatic, due to the complexity of the field. The here summarized current knowledge probably only scratches the surface of what there is to know on bacterial glycobiology. Bacterial glycoconjugates exhibit an enormous diversity, but their biosynthesis is also stooled on common themes and pathways. This common ground can enhance the study of the still unexplored glycosylation potential of several bacterial strains, but should be complemented with performant analytical methods, since sequence information does not adequately predicts the glycosylation potential.

Albeit glycoconjugates like EPS, CPS and LPS are studied for years now, many aspects remain unknown. In addition, especially research concerning the functional importance of glycoconjugates such as glycoproteins has been lagging behind. Currently, the focus lies mainly on the elucidation of their role in virulence, pathogenesis and bacteria-host interactions in general. But their functional importance can even be more vibrant than currently appreciated. For instance, several groups report on the (potential) glycosylation of proteins involved in the multienzyme complexes of the cell wall biosynthesis and division machinery (Balonova et al., 2010; Fletcher et al., 2009; Sanchez-Rodriguez et al., 2014). It is tempting to speculate that the yet unexplored glycosylation of major enzymes in these complexes could play a role in the regulation of their interaction and activity, but this remains to be elucidated. Only a few illustrations of the fascinating functional importance of bacterial protein glycosylation are the modulation of flagellin glycosylation in H. pylori to remain stealth from the immune system (Josenhans et al., 2002), the link between protein glycosylation, redox reactions and electron transport systems seen in Neisseria sp. (Ku et al., 2009; Vik et al., 2009) and the dynamic interplay between post-translational modifications in N. gonorrhoeae (Anonsen et al., 2012a).

Another important driver for the further expansion of the field of prokaryotic glycobiology is its glycoengineering potential. This technology is extremely interesting for industrial

62

the

swee

t too

th o

f bac

teria

applications like the amelioration of food quality by the engineering of EPS structures or the production of engineered glycoproteins, like vaccines and human therapeutic agents (e.g. insulin) (Campbell et al., 2007; Elliott et al., 2003; Farahmand et al., 2011; Ihssen et al., 2010; Iwashkiw et al., 2012a; Jain et al., 2003; Lindhout et al., 2011; Lizak et al., 2011a; Pandhal & Wright, 2010; Saxon & Bertozzi, 2001; Werner et al., 2007). A better knowledge of the glycosylation mechanisms in beneficial bacteria could cause a revolution in the biosynthesis of therapeutic molecules in prokaryotic vectors.

The current knowledge on bacterial glycosylation probably only scratches the surface of what there is to know. But the already known facts show that a further elucidation of the enigmatic world of bacterial glycoconjugates is worth pursuing as this can render important insights in bacteria-host interactions and our understanding of bacterial life. Without doubt, the future of bacterial glycobiology looks promising and exciting.

two.

63

three.a network-based

approach to identify substrate classes of

bacterial glycosyltransferases

a tale on microbiologists and bioinformaticians tackling one of the

mysteries of glycobiology

three.a network-based

approach to identify substrate classes of

bacterial glycosyltransferases

a tale on microbiologists and bioinformaticians tackling one of the

mysteries of glycobiology

66

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

This chapter was published:

Sanchez-Rodriguez A.#, Tytgat H.L.P.#, Winderickx J., Vanderleyden J., Lebeer S., Marchal K. (2014) A network-based approach to identify substrate relations of bacterial glycosyltransferases. BMC Genomics, 15 (349). #The authors contributed equally to this work

autHors’ contributions

ASR and HT contributed equally to this work.

HT and ASR conceived the study, together with their PhD supervisors KM and SL resp.. ASR performed the computational analysis. HT cured the in silico data, performed the experimental work and provided biological input for the computational analysis. ASR and HT interpreted the results and drafted the manuscript. KM participated in the design of the computational analysis and in the writing of the manuscript. SL participated in the design of the experimental work, provided biological input for the computational analyses and helped drafting the manuscript. JVDL, JW, SL and KM coordinated and managed the research.

thre

e.

67

abstract

Bacterial interactions with the environment- and/or host largely depend on the bacterial glycome. The specificities of a bacterial glycome are mainly determined by GTs, the enzymes involved in transferring sugar moieties from an activated donor to a specific substrate. Of these GTs their coding regions, but mainly also their substrate specificity are still largely unannotated as most sequence-based annotation flows suffer from the lack of characterized sequence motifs that can aid in the prediction of the substrate specificity.

In this chapter, we developed an analysis flow that uses sequence-based strategies to predict novel GTs, but also exploits a network-based approach to infer the putative substrate classes of these predicted GTs. Our analysis flow was benchmarked with the well-documented GT-repertoire of C. jejuni NCTC 11168 and applied to the probiotic model L. rhamnosus GG to expand our insights in the glycosylation potential of this bacterium. In L. rhamnosus GG we could predict 48 GTs of which eight were not previously reported. For at least 20 of these GTs a substrate relation was inferred.

We confirmed through experimental validation our prediction of WelI acting upstream of WelE in the biosynthesis of exopolysaccharides. We further hypothesize to have identified in L. rhamnosus GG the yet undiscovered genes involved in the biosynthesis of glucose-rich glycans and novel GTs involved in the glycosylation of proteins. Interestingly, we also predict GTs with well-known functions in peptidoglycan synthesis to also play a role in protein glycosylation.

68

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

3.1 backgrounD

The glycome, playing a crucial role in allowing bacteria to establish environment- and host-specific interactions (Kay et al., 2010; Upreti et al., 2003) consists of a wide variety of glycoconjugates, i.e. glycans being covalently linked to other macromolecules. In Gram-negatives, these glycoconjugates occur mainly in the outer membrane as a thin layer of peptidoglycan (PG) and lipopolysaccharides (LPS) or lipo-oligosaccharides (LOS). Across the outer membrane, exopolysaccharides (EPS) or capsular polysaccharides (CPS), glycoproteins and glycolipids can further decorate the cell surface (Upreti et al., 2003). In Gram-positives, which in contrast to Gram-negatives lack an outer membrane, complex polymers such as teichoic acids in Firmicutes and lipoglycans in Actinobacteria strengthen a thick layer of PG. CPS or EPS are also often found as most external layer in Gram-positive bacteria. Bacteria can also produce intracellular glycoconjugates, such as glycosylated secondary metabolites and storage polysaccharides like glycogen (Upreti et al., 2003).

Glycosyltransferases (GTs), transferring sugar moieties from an activated donor to a specific substrate (Lairson et al., 2008), are key enzymes in the biosynthesis of glycoconjugates. Depending on their specificity, the substrates of GTs range from lipids, proteins, saccharides, nucleic acids to small molecules (Lairson et al., 2008). In bacteria, two different glycosylation mechanisms have been described: sequential glycosylation, in which either soluble or membrane-associated GTs transfer glycan monomers directly to the final substrate and en bloc glycosylation, in which the sugar moiety is first assembled and only then transferred to the final substrate by an specialized GT (oligosaccharyltransferase (OST) or polymerase) (Guerry & Szymanski, 2008; Hug & Feldman, 2011). The latter mechanism is by far the best documented, and is involved in the biosynthesis of heteropolymeric EPS/CPS, O antigens in LPS, and even PG biosynthesis, highlighting the commonalities in the biosynthesis of these glycoconjugates (Hug & Feldman, 2011). Apart from their general role in glycosylation, the specificities of most of the GTs and the cellular role of their end products are still largely unknown. In addition, most of the substrate specificities of GTs involved in LPS, PG and glycoproteins have been described in Gram-negatives (Lerouge & Vanderleyden, 2002; Typas et al., 2012), while glycosylation in Gram-positives is much less studied.

Whereas sequence-based predictions have shown useful to identify potential GTs (Egelund et al., 2004; Hansen et al., 2010; Hansen et al., 2009), predicting the specificity of those identified GTs is less trivial, definitely for prokaryotes for which no clear sequence motifs determining substrate specificity have been described (Weerapana & Imperiali, 2006). In addition, many GTs and OSTs show substrate promiscuity (Faridmoayer et al., 2008; Wacker et al., 2006), hampering the identification of clear substrate motifs.

To improve the annotation of GTs in prokaryotes, we developed an analysis flow that uses a sequence-based strategy to predict GTs and a network-based approach (Szklarczyk et al., 2011) to identify links between these predicted GTs and other genes/proteins. Although such

thre

e.

69

links do not give insights into the precise biochemical mechanisms of a GT with its substrate, they aid in relating the GT to possible classes of molecules that could accept the sugar moieties from these GTs (referred to as substrate classes).

We tested our analysis flow on the genome of C. jejuni NCTC 11168, in which the important classes of glycoconjugates (N- and O-glycoproteins, PG, LOS, and CPS) are well characterized (Guerry & Szymanski, 2008).

Further applying our analysis flow on the probiotic bacterium Lactobacillus rhamnosus GG provided a comprehensive re-annotation of putative GTs in this species, the possible substrate classes of these GTs and their mode of action. These predictions are a very useful resource for experimentalists, predominantly because the study of (protein) glycosylation in lactobacilli and related organisms is not straightforward (Claes et al., 2012). Our predictions unveil putative novel mechanisms of (protein) glycosylation, involving the potential, promiscuous role of GTs with known function in PG biosynthesis.

3.2 metHoDs

3.2.1 bacteriaL proteomes

The proteomes and current genome annotations of Lactobacillus rhamnosus GG (NC_013198.1) and Campylobacter jejuni NCTC 11168 (NC_002163.1) were obtained from GenBank (http://www.ncbi.nlm.nih.gov/genbank/).

3.2.2 HiDDen markov moDeL profiLe searcHes

Hidden Markov Models (HMMs) describing known GT signatures were collected from SUPERFAMILY (http://supfam.cs.bris.ac.uk/SUPERFAMILY/), CAZy (http://www.cazy.org/) and Pfam (http://pfam.sanger.ac.uk/) and subdivided into three groups depending on their expected specificity for GTs (Table 3.1). For CAZy, a thorough search of this database was performed, and all the HMMs covering GT classes that had bacterial representatives were included in our analysis (see below).

The first and least specific group contains the HMM representing ‘Rossmann-fold domains’, which are known to resemble the GT-A and GT-B folds typical for GTs using sugar nucleotides as donor (Ha et al., 2001; Hansen et al., 2010; Lairson et al., 2008). A second group comprises the HMMs for ‘Sugar transferases’ and ‘UDP-Glycosyltransferases’ respectively, both HMMs of intermediate specificity covering a broad class of GTs (Egelund et al., 2004; Hansen et al., 2010). A last group combines a set of more GT-specific HMMs (10 in total), all of which are based on a small number of family-specific sequences (Baiet et al., 2011; Campbell et al., 1997; Di Guilmi et al., 2003; Knauer & Lehle, 1999a; Maldonado-Barragan et al., 2011; Mengin-Lecreulx et al., 1991; Provencher et al., 2003; Silberstein et al., 1995; Sun et al., 2012; Yoshida et al., 1998). This group combines HMMs extracted from CAZy (Coutinho et

70

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

al., 2003), representative for enzymes that catalyze glycosidic bonds (strictu-sensu GTs) with HMMs extracted from Pfam (Finn et al., 2010) that are representative for non-Leloir GTs that use non-nucleotide sugar donors or oligo/polysaccharides. Enzymes involved in the transfer of the sugar moiety to the final substrate (such as OSTs and priming GTs) are examples of this latter class of non-Leloir GTs.

The collected HMMs were used to screen entire proteomes (C. jejuni NCTC 11168 and L. rhamnosus GG) with hmmsearch from the HMMER package version 2.2 (Finn et al., 2011). Hits were filtered using an E-value cut-off of 0.1.

Table 3.1 - Summary of the Hidden Markov Models (HMMs) used to screen for glycosyltransferases in the proteomes of Campylobacter jejuni NCTC 11168 and Lactobacillus rhamnosus GG

HMM group

Description Database Reference

I Rossmann-fold domains SUPERFAMILY Ha et al., 2001

Egelund et al., 2004

Lairson et al., 2008

Hansen et al., 2010

II Sugar transferase SUPERFAMILY Egelund et al., 2004

Hansen et al., 2010

UDP-Glycosyltransferase SUPERFAMILY Egelund et al., 2004

Hansen et al., 2010

III Transglycosylase (PF00912) Pfam/CAZy Di Guilmi et al., 2003

Glycosyltransferase WecB/TagA/CpsF (PF03808)

Pfam/CAZy Maldonado-Barragán et al., 2011

Bacterial sugar transferase (PF02397) Pfam Yoshida et al., 1998

Provencher et al., 2003

Oligosaccharyltransferase STT3 subunit (PF02516)

Pfam/CAZy Baïet et al., 2011

DAD family (PF02109) Pfam Silberstein et al., 1995

OST3/OST6 family (PF04756) Pfam/CAZy Knauer et al., 1999

Glycosyltransferase family 25 (PF01755) Pfam/CAZy Campbell et al., 1997

Glycosyltransferase family 28 (PF04101) Pfam/CAZy Mengin-Lecreulx et al., 1991

Glycosyltransferase family 9 (PF01075) Pfam/CAZy Kadrmas et al., 1998

HMM group: HMMs were grouped according to their expected specificity for glycosyltransferase activity in an increasing order. Description: description of the HMM. The Pfam model id is also provided. Database: source of the model. Reference: bibliographic citation supporting the inclusion of the corresponding HMM in the analysis.

3.2.3 protein foLD recognition

The profile based fold recognition method pGenTHREADER (Lobley et al., 2009), accessible via the PSIPRED server (http://bioinf.cs.ucl.ac.uk/psipred/) was used to detect known GT-A/

thre

e.

71

GT-B folds in proteins predicted to be GTs by the HMM search. Each of the input sequences was aligned against a library of 3D folds based on CATH v3.3 (the Protein Structure Database, available at http://www.cathdb.info/) by pGenTHREADER. The library of 3D folds contains a total of 684 PDB structures of known GTs. Putative GTs were only retained if they predicted fold showed significant homology (net score > 46) to the one of a resolved 3D structure with known GT activity present in the library (refined set). We selected a cutoff > 46 on the net score of pGenTHREADER since any values higher than this threshold are categorized as HIGH to CERTIFIED confidence predictions (default conservative setting of the tool).

3.2.4 Detecting functionaL partners of gLycosyLtransferases

The STRING database (http://string-db.org/) was used as the source of functional networks (Szklarczyk et al., 2011; von Mering et al., 2005). We interrogated STRING using as queries our predicted GTs from both L. rhamnosus GG and C. jejuni NCTC 11168 to retrieve the network of functional partners associated to each query (query-based subnetwork). We only considered functional interactions with a score higher than 0.7, which is the default value in STRING for high confidence interactions. A total of 1112 functional interactions were retrieved for L. rhamnosus GG, supported by 2338 independent evidences distributed as follows: 1682 evidences based on the genomic context of the interacting partners (e.g. physical closeness, co-occurrence in closely related species, gene fusion events); 153 evidences based on the co-expression of the interacting partners; 28 evidences derived from high-throughput experiments (e.g. protein-protein interaction data); 465 evidences derived from the literature (text-mining). For C. jejuni NCTC 11168 a total of 1727 functional interactions were retrieved supported by 3190 independent evidences from the following data sources: 2520 evidences based on the genomic context of the interacting partners; 47 evidences based on co-expression; 37 evidences from high-throughput experiments; 584 evidences derived from the literature.

Gene Ontology annotation files for L. rhamnosus GG and C. jejuni NCTC 11168 were obtained from http://www.ebi.ac.uk/GOA/proteomes.html. To calculate which functional GO classes were enriched amongst interacting partners of a certain GT, we used the hypergeometric test, corrected for multiple testing using False Discovery Rate (Storey & Tibshirani, 2003).

We then created ‘consensus networks’ that combine the local network neighborhood of all GTs, predicted to belong to the same specificity class and of which the local subnetworks are enriched in the same GO terms. GT-specific subnetworks were merged in a consensus network by retaining the edges from all the composing subnetworks that either reflect GT-GT interactions, interactions between a GT and one or more transmembrane proteins (membrane associations) or interactions between GTs and proteins with predicted glycosylation signals (predicted protein substrate relation).

3.2.5 Detection of putative protein gLycosyLation sites

Glycosylation sites were predicted in the proteomes of C. jejuni NCTC 11168 and L.

72

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

rhamnosus GG using the GlycoPP webserver (http://www.imtech.res.in/raghava/glycopp/), specially developed for the analysis of prokaryotic protein sequences. Predictions were made using the hybrid approaches: BPP + ASA (for N-glycosites predictions) and PPP + ASA (for O-glycosites prediction) as suggested by the developers. A SVM threshold of 0.5 was used to reduce the probability of false positive predictions.

3.2.6 preDiction of transmembrane HeLices

Transmembrane helices were predicted using the TMHMM server version 2.0 (http:// www.cbs.dtu.dk/services/TMHMM/).

3.2.7 bencHmark

The available data on glycosylation in the paradigm organism C. jejuni NCTC 11168 was used for benchmarking purposes and helped us to fine-tune and evaluate our workflow. C. jejuni is considered as a model for bacterial glycosylation, since it can not only N- and O- glycosylate proteins by both sequential and en bloc transfer (Nothaft & Szymanski, 2010; Szymanski et al., 2003a), but also produces a wide variety of glycoconjugates, including PG, LOS and CPS. Because glycosylation is extensively studied in C. jejuni NCTC 11168 we used this model system to compile a literature benchmark dataset. We obtained information on 10 proteins with experimentally verified glycosyltransferase activity and known substrate specificity in C. jejuni (Cj1124c, Cj1125c, Cj1126c, Cj1127c, Cj1128c and Cj1129c involved in protein N-glycosylation and Cj1133, Cj1136, Cj1139c and Cj1148 involved in LOS biosynthesis). Proteins annotated in C. jejuni NCTC 11168 as GTs based on indirect evidence (e.g. through homology assignment) were omitted from the benchmark dataset.

3.2.8 reannotation of gts in c. jEjuni anD L. rhamnosus gg baseD on our preDictions anD Literature

For the GTs that were previously annotated with a GT-related function, a simplified annotation is proposed when the evidence on the exact GT activity is not available for L. rhamnosus GG (such as for LGG_00279, LGG_00280 and LGG_00281). In addition, gene names inferred from non-strong homology searches (i.e. BLASTn E-value > 0.01) were removed (e.g. LGG_00348). For GTs putatively involved in polysaccharide biosynthesis (LGG_00279-LGG_00283, see below), gene names were corrected in agreement with the correct gene nomenclature (Reeves et al., 1996).

3.2.9 experimentaL work

L. rhamnosus GG and its mutant derivatives were grown in MRS without agitation. A new ΔwelI::TcR gene deletion mutant, lacking the LGG_02047 gene, termed CMPG10811, was constructed as described earlier (Lebeer et al., 2012b), using the pro-7946 (5’-ATACTAGTTCTTATCATAGTTTCCAGACC-3’) and pro-7947

thre

e.

73

(5’-ATCCCGGGGTGGGGAACTTGCTG-3’) primers. As this is a gene deletion mutant in an operon, polar effects can not completely be ruled out. Total EPS determination, monomer analysis and adhesion assays were performed as previously described (Lebeer et al., 2009). Statistical analysis (One-way ANOVA) was performed using GraphPad Prism 6 on data corresponding to three technical repeats of three independent biological samples.

3.3 resuLts

3.3.1 annotating putative gLycosyLtransferases

To predict additional GTs, we used an HMM based screening (Figure 3.1A). To maximize the sensitivity of our screening, the heterogeneous functional family of GTs was represented by a collection of 12 different HMMs, each of which captures a different characteristic of known GTs (Table 3.1). These 12 HMMs were subdivided into three groups depending on their expected specificity for GTs, referred to as respectively I) ‘Rossmann-fold domains’, II) ‘Sugar transferase’ and ‘UDP-Glycosyltranferase’ and III) a set of nine more GT-specific HMMs.

As HMM-based screenings, definitely those performed with the least GT-specific HMMs, tend to also find many non-specific hits (false positives), predictions were further filtered using a protein fold recognition step: GTs predicted by the HMM profiling were only retained if they contained a three-dimensional fold with significant homology to folds present in experimentally confirmed GTs from any species (referred to as the refined set in Figure 3.1) (see Material and methods).

The results of the HMM based screening in both L. rhamnosus GG and C. jejuni NCTC 11168 before and after filtering with the fold based predictions are shown in Figure 3.2, together with the most abundant GO categories present amongst the predicted GTs. Filtering successfully reduced potential false positive predictions, for instance, a large fraction of oxidoreductases (all binding the cofactor NAD) obtained by screening with the least specific ‘Rossmann-fold domain’ HMM were removed after the fold recognition based filtering (Figure 3.2A). The three predictions in C. jejuni (Additional file 1: Table S1) and the five in L. rhamnosus GG (Table 3.2) made by the ‘Rossmann-fold domain’ HMM and retained after the fold recognition could not be retrieved by any of the other HMM models, showing the added value of also using this least specific class of HMMs. Screening with the ‘Sugar transferases’ and “UDP-glycosyltransferases’ HMMs in contrast resulted in predictions that were quite GT-specific, as indeed approximately 50% of the originally obtained predictions also contain a GT-like fold (Figure 3.2B and C). Fold-based filtering here removed mainly predicted DNA-binding proteins, as their mechanism of binding DNA is also based on recognizing the sugar moieties of the nucleotides. As expected, screening with the HMMs obtained from Pfam and CAZy resulted both in C. jejuni NCTC 11168 and L. rhamnosus GG in the highest fraction of hits that also displayed a GT-like fold (Figure 3.2D).

74

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

Figure 3.1 – Glycosyltransferase annotation flowA: Genome-wide annotation of glycosyltransferases (GTs). Glycosyltransferases are predicted by scanning the proteomes of the studied species for GT-specific signatures using Hidden Markov Models (HMM) from SUPERFAMILY, CAZy and Pfam. An additional fold recognition filtering step is applied to only retain those genes containing a three-dimensional fold (inferred by the PGenTHREADER algorithm) with significant homology to folds present in experimentally confirmed GTs (deposited in the SCOP database). B: Predicting GT substrate class and putative mode of action. The local network neighborhood of each query GT (black node) in a functional interaction network (STRING) is used to extract a GT-specific local subnetwork for each query GT. The local subnetwork of a GT comprises predicted functional partners (proteins being functionally related to the query GT). Based on the GO enrichment analysis of these genes in this local subnetwork, the substrate class of the query GT is derived. To gain information on the mode of glycosylation, the GT specific local subnetwork is further annotated with either membrane associations between a query GT and a predicted transmembrane protein (blue edge) and with relations indicative for protein glycosylation (yellow edge).

The performance of our GT prediction flow with and without the fold recognition filtering step was also evaluated in terms of the true-positive rate on the C. jejuni benchmark (containing 10 proteins with experimentally validated GT activity in C. jejuni NCTC 11168, see Materials and methods). To obtain a full recall of 100% (that is retrieving all 10 positives), we had to make 184 predictions before the filtering. After the filtering the true positive rate increased from 10/184 to 10/44 (Additional file 1: Table S1). In addition to recovering all benchmark GTs (those indicated with experimental validation in Additional file 1: Table S1), most other predictions corresponded to previously made GT related annotations in C. jejuni NCTC 11168 that were based on indirect evidence (e.g. through experimental validation in other closely related species), such as the loci comprising the GT genes responsible for the synthesis of LOS (CJ1133 – CJ1148) (Karlyshev et al., 2005), the GTs for N- (CJ1121c – CJ1129c) (Nothaft & Szymanski, 2010) and O-glycoprotein biosynthesis (CJ1311 – CJ1333) (Szymanski et al., 2003a) and the CPS biosynthesis cluster (CJ1416c – CJ1442c) (Gundogdu et al., 2007; Linton

thre

e.

75

et al., 2001). In addition, we made a total of 17 new predictions for yet unannotated genes in C. jejuni NCTC 11168 (Additional file 1: Table S1). Finally, we also retrieved four potential false positives (Additional file 1: Table S1).

Figure 3.2 – Annotated glycosyltransferasesResults for the model system Campylobacter jejuni are shown on the left panel and for L. rhamnosus GG on the right panel. Putative GTs were predicted using an HMM based screening. A: results obtained with an HMM recognizing ‘Rossmann-fold domains’, expected to be the HMM with the lowest specificity towards GTs (Table 1, class I). B: results obtained with a family of HMMs of intermediate specificity for GTs (Table 1, class II). C: results obtained with the class of HMMs, most specific for GTs (Table 1, class III). Pie charts indicate the extent to which different functional classes were enriched amongst the predictions obtained with the respective classes of HMMs. Slices indicated in red on the pie chart correspond to the functional classes of the predictions that were retained after the fold recognition filtering step. For each group of HMMs, the total number of predictions is denoted in black on top of every pie chart and the number of predictions retained after applying the fold recognition step is denoted in red.

76

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

Tabl

e 3.

2 - U

pdat

ed a

nnot

atio

n of

gly

cosy

ltra

nsfe

rase

s pr

edic

ted

in t

he g

enom

e of

Lac

toba

cill

us r

ham

nosu

s GG

Loc

us ta

gC

urre

nt A

nnot

atio

nPr

opos

ed A

nnot

atio

nH

MM

Evi

denc

eR

efer

ence

LGG

_002

79w

elA;

dTD

P-rh

amno

syl t

rans

fera

se rf

bFw

clA;

gly

cosy

ltran

sfer

ase

(put

ativ

e ce

ll w

all p

olys

acch

arid

e bi

osyn

thes

is)

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_002

80w

elB;

alp

ha-L

-Rha

alp

ha-1

,3-L

-rh

amno

syltr

ansf

eras

ew

clB;

gly

cosy

ltran

sfer

ase

(put

ativ

e ce

ll w

all p

olys

acch

arid

e bi

osyn

thes

is)

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_002

81w

elC

; alp

ha-L

-Rha

alp

ha-1

,3-L

-rh

amno

syltr

ansf

eras

ew

clC

; gly

cosy

ltran

sfer

ase

(put

ativ

e ce

ll w

all p

olys

acch

arid

e bi

osyn

thes

is)

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_002

83*

eps2

; hyp

othe

tical

pro

tein

wcl

D; p

utat

ive

glyc

osyl

trans

fera

se (p

utat

ive

cell

wal

l pol

ysac

char

ide

bios

ynth

esis

)U

DP-

Gly

cosy

ltran

sfer

ase

--

LGG

_002

95G

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_003

48yo

hJ; l

ipop

olys

acch

arid

e bi

osyn

thes

is p

rote

inPu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_003

49yo

hH; p

olyg

lyce

rol-p

hosp

hate

alp

ha-

gluc

osyl

trans

fera

sePu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_006

45G

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_006

95gt

rB; g

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_007

94pb

p1B;

pen

icill

in-b

indi

ng p

rote

in 1

Bpb

pb1B

; put

ativ

e gl

ycos

yltra

nsfe

rase

, pen

icill

in-b

indi

ng p

rote

in 1

B

(pep

tidog

lyca

n bi

osyn

thes

is)

Pfam

/CA

ZyC

onse

rvat

ion

Schw

artz

et a

l., 2

002;

K

anka

inen

et a

l., 2

009

LGG

_008

25rf

aG; g

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_008

26cp

oA; g

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_009

28*

yvcK

; tra

nspo

rter

Puta

tive

glyc

osyl

trans

fera

seR

ossm

ann-

fold

do

mai

ns-

-

LGG

_009

85*

Inte

gral

mem

bran

e pr

otei

nPu

tativ

e gl

ycos

yltra

nsfe

rase

Pfam

/CA

Zy-

-LG

G_0

0998

arbX

; lip

opol

ysac

char

ide

bios

ynth

esis

gl

ycos

yltra

nsfe

rase

Puta

tive

glyc

osyl

trans

fera

seSu

gar t

rans

fera

seC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_009

99ar

bY; l

ipop

olys

acch

arid

e bi

osyn

thes

is

glyc

osyl

trans

fera

sePu

tativ

e gl

ycos

yltra

nsfe

rase

Ros

sman

n-fo

ld

dom

ains

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_010

57G

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_010

62#

galU

; UTP

-glu

cose

-1-p

hosp

hate

urid

ylyl

trans

fera

seU

TP-g

luco

se-1

-pho

spha

te u

ridyl

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_010

69gt

rB; g

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_011

47G

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

Ros

sman

n-fo

ld

dom

ains

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_011

95*

met

Q; A

BC

tran

spor

ter

AB

C tr

ansp

orte

r, pu

tativ

e bi

func

tiona

l gly

cosy

ltran

sfer

ase

Pfam

--

LGG

_012

83m

urG

; und

ecap

reny

ldip

hosp

ho-

mur

amoy

lpen

tape

ptid

e be

ta-N

- ac

etyl

gluc

osam

inyl

trans

fera

se

mur

G; u

ndec

apre

nyld

ipho

spho

-mur

amoy

lpen

tape

ptid

e be

ta-N

-ac

etyl

gluc

osam

inyl

trans

fera

se (p

eptid

ogly

can

bios

ynth

esis

)U

DP-

Gly

cosy

ltran

sfer

ase

Con

serv

atio

nM

engi

n-Le

creu

lx e

t al.,

19

91; K

anka

inen

et a

l.,

2009

LGG

_014

12*

trm

FO; t

RN

A u

raci

l-5-m

ethy

ltran

sfer

ase

tRN

A u

raci

l −5-

met

hyltr

ansf

eras

e, p

utat

ive

bifu

nctio

nal

glyc

osyl

trans

fera

seR

ossm

ann-

fold

do

mai

ns-

-

thre

e.

77

Tabl

e 3.

2 - U

pdat

ed a

nnot

atio

n of

gly

cosy

ltra

nsfe

rase

s pr

edic

ted

in t

he g

enom

e of

Lac

toba

cill

us r

ham

nosu

s GG

Loc

us ta

gC

urre

nt A

nnot

atio

nPr

opos

ed A

nnot

atio

nH

MM

Evi

denc

eR

efer

ence

LGG

_002

79w

elA;

dTD

P-rh

amno

syl t

rans

fera

se rf

bFw

clA;

gly

cosy

ltran

sfer

ase

(put

ativ

e ce

ll w

all p

olys

acch

arid

e bi

osyn

thes

is)

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_002

80w

elB;

alp

ha-L

-Rha

alp

ha-1

,3-L

-rh

amno

syltr

ansf

eras

ew

clB;

gly

cosy

ltran

sfer

ase

(put

ativ

e ce

ll w

all p

olys

acch

arid

e bi

osyn

thes

is)

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_002

81w

elC

; alp

ha-L

-Rha

alp

ha-1

,3-L

-rh

amno

syltr

ansf

eras

ew

clC

; gly

cosy

ltran

sfer

ase

(put

ativ

e ce

ll w

all p

olys

acch

arid

e bi

osyn

thes

is)

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_002

83*

eps2

; hyp

othe

tical

pro

tein

wcl

D; p

utat

ive

glyc

osyl

trans

fera

se (p

utat

ive

cell

wal

l pol

ysac

char

ide

bios

ynth

esis

)U

DP-

Gly

cosy

ltran

sfer

ase

--

LGG

_002

95G

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_003

48yo

hJ; l

ipop

olys

acch

arid

e bi

osyn

thes

is p

rote

inPu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_003

49yo

hH; p

olyg

lyce

rol-p

hosp

hate

alp

ha-

gluc

osyl

trans

fera

sePu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_006

45G

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_006

95gt

rB; g

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_007

94pb

p1B;

pen

icill

in-b

indi

ng p

rote

in 1

Bpb

pb1B

; put

ativ

e gl

ycos

yltra

nsfe

rase

, pen

icill

in-b

indi

ng p

rote

in 1

B

(pep

tidog

lyca

n bi

osyn

thes

is)

Pfam

/CA

ZyC

onse

rvat

ion

Schw

artz

et a

l., 2

002;

K

anka

inen

et a

l., 2

009

LGG

_008

25rf

aG; g

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_008

26cp

oA; g

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_009

28*

yvcK

; tra

nspo

rter

Puta

tive

glyc

osyl

trans

fera

seR

ossm

ann-

fold

do

mai

ns-

-

LGG

_009

85*

Inte

gral

mem

bran

e pr

otei

nPu

tativ

e gl

ycos

yltra

nsfe

rase

Pfam

/CA

Zy-

-LG

G_0

0998

arbX

; lip

opol

ysac

char

ide

bios

ynth

esis

gl

ycos

yltra

nsfe

rase

Puta

tive

glyc

osyl

trans

fera

seSu

gar t

rans

fera

seC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_009

99ar

bY; l

ipop

olys

acch

arid

e bi

osyn

thes

is

glyc

osyl

trans

fera

sePu

tativ

e gl

ycos

yltra

nsfe

rase

Ros

sman

n-fo

ld

dom

ains

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_010

57G

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_010

62#

galU

; UTP

-glu

cose

-1-p

hosp

hate

urid

ylyl

trans

fera

seU

TP-g

luco

se-1

-pho

spha

te u

ridyl

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_010

69gt

rB; g

lyco

syltr

ansf

eras

e, g

roup

2Pu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_011

47G

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

Ros

sman

n-fo

ld

dom

ains

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_011

95*

met

Q; A

BC

tran

spor

ter

AB

C tr

ansp

orte

r, pu

tativ

e bi

func

tiona

l gly

cosy

ltran

sfer

ase

Pfam

--

LGG

_012

83m

urG

; und

ecap

reny

ldip

hosp

ho-

mur

amoy

lpen

tape

ptid

e be

ta-N

- ac

etyl

gluc

osam

inyl

trans

fera

se

mur

G; u

ndec

apre

nyld

ipho

spho

-mur

amoy

lpen

tape

ptid

e be

ta-N

-ac

etyl

gluc

osam

inyl

trans

fera

se (p

eptid

ogly

can

bios

ynth

esis

)U

DP-

Gly

cosy

ltran

sfer

ase

Con

serv

atio

nM

engi

n-Le

creu

lx e

t al.,

19

91; K

anka

inen

et a

l.,

2009

LGG

_014

12*

trm

FO; t

RN

A u

raci

l-5-m

ethy

ltran

sfer

ase

tRN

A u

raci

l −5-

met

hyltr

ansf

eras

e, p

utat

ive

bifu

nctio

nal

glyc

osyl

trans

fera

seR

ossm

ann-

fold

do

mai

ns-

-

LGG

_014

87pb

p1A;

pen

icill

in-b

indi

ng p

rote

in 1

Apb

p1A;

put

ativ

e gl

ycos

yltra

nsfe

rase

, pen

icill

in-b

indi

ng p

rote

in 1

A

(pep

tidog

lyca

n bi

osyn

thes

is)

Pfam

/CA

ZyC

onse

rvat

ion

Bar

rett

et a

l., 2

007;

K

anka

inen

et a

l., 2

009

LGG

_015

38Ph

age-

rela

ted

glyc

osyl

trans

fera

sePu

tativ

e gl

ycos

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_015

86yo

hH; g

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_015

87yo

hJ; g

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

Ros

sman

n-fo

ld

dom

ains

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_017

83pb

p2A;

mem

bran

e ca

rbox

ypep

tidas

e, p

enic

illin

-bi

ndin

g pr

otei

n 2A

pbp2

A; b

ifunc

tiona

l mem

bran

e ca

rbox

ypep

tidas

e, p

utat

ive

glyc

osyl

trans

fera

se, p

enic

illin

-bin

ding

pro

tein

2A

(pep

tidog

lyca

n bi

osyn

thes

is)

Pfam

/CA

ZyC

onse

rvat

ion

Di G

uilm

i et a

l., 2

003;

K

anka

inen

et a

l., 2

009

LGG

_019

91*

UD

P-N

-ace

tylg

luco

sam

ine

2-ep

imer

ase

Epim

eras

e, p

utat

ive

bifu

nctio

nal g

lyco

syltr

ansf

eras

eU

DP-

Gly

cosy

ltran

sfer

ase

--

LGG

_019

92*

UD

P-N

-ace

tylg

luco

sam

ine

2-ep

imer

ase

Epim

eras

e, p

utat

ive

bifu

nctio

nal g

lyco

syltr

ansf

eras

eSu

gar t

rans

fera

se-

-LG

G_0

1999

rmlA

; glu

cose

-1-p

hosp

hate

thym

idyl

yltra

nsfe

rase

rmlA

; glu

cose

-1-p

hosp

hate

thym

idyl

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nB

lank

enfe

ldt e

t al.,

200

0;

Kan

kain

en e

t al.,

200

9LG

G_0

2004

eps3

; sug

ar o

r lip

opol

ysac

char

ide

synt

hesi

s tra

nsfe

rase

Puta

tive

glyc

osyl

trans

fera

sePf

amC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_020

23#

glgP

; gly

coge

n st

arch

alp

ha-g

luca

n ph

osph

oryl

ase

glgP

, gly

coge

n al

pha-

gluc

an p

hosp

hory

lase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Yu e

t al.,

198

8;

Kan

kain

en e

t al.,

200

9LG

G_0

2024

glgA

; gly

coge

n sy

ntha

segl

gA; g

lyco

gen

synt

hase

(gly

coge

n bi

osyn

thes

is)

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kie

l et a

l., 1

994;

K

anka

inen

et a

l., 2

009

LGG

_020

25#

glgD

; glu

cose

-1-p

hosp

hate

ade

nyly

ltran

sfer

ase

glgD

; glu

cose

-1-p

hosp

hate

ade

nyly

ltran

sfer

ase

(gly

coge

n bi

osyn

thes

is)

Suga

r tra

nsfe

rase

Con

serv

atio

nB

allic

ora

et a

l., 2

003;

K

anka

inen

et a

l., 2

009

LGG

_020

26#

glgC

; glu

cose

-1-p

hosp

hate

ade

nyly

ltran

sfer

ase

glgC

; glu

cose

-1-p

hosp

hate

ade

nyly

ltran

sfer

ase

(gly

coge

n bi

osyn

thes

is)

Suga

r tra

nsfe

rase

Con

serv

atio

nB

allic

ora

et a

l., 2

003;

K

anka

inen

et a

l., 2

009

LGG

_020

40$

rmlA

1; g

luco

se-1

-pho

spha

te th

ymid

yl tr

ansf

eras

erm

lA1;

glu

cose

-1-p

hosp

hate

thym

idyl

tran

sfer

ase

Suga

r tra

nsfe

rase

Con

serv

atio

nB

lank

enfe

ldt e

t al.,

200

0;

Kan

kain

en e

t al.,

200

9LG

G_0

2042

rmlA

2; g

luco

se-1

-pho

spha

te th

ymid

ylyl

trans

fera

serm

lA2;

glu

cose

-1-p

hosp

hate

thym

idyl

yltra

nsfe

rase

Suga

r tra

nsfe

rase

Con

serv

atio

nB

lank

enfe

ldt e

t al.,

200

0;

Kan

kain

en e

t al.,

200

9LG

G_0

2043

wel

E; u

ndec

apre

nyl-p

hosp

hate

bet

a-gl

ucos

epho

spho

trans

fera

sew

elE;

prim

ing

glyc

osyl

trans

fera

se (g

alac

tose

-ric

h EP

S bi

osyn

thes

is)

Pfam

Expe

rimen

tal

valid

atio

nLe

beer

et a

l., 2

009

LGG

_020

44w

elF;

gly

cosy

ltran

sfer

ase,

gro

up 1

wel

F; p

utat

ive

glyc

osyl

trans

fera

se (g

alac

tose

-ric

h EP

S bi

osyn

thes

is)

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_020

45w

elG

; gly

cosy

ltran

sfer

ase,

ga

lact

ofur

anos

yltra

nsfe

rase

wel

G; p

utat

ive

glyc

osyl

trans

fera

se (g

alac

tose

-ric

h EP

S bi

osyn

thes

is)

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_020

46w

elH

; alp

ha-L

-Rha

alp

ha-1

,3-L

-rh

amno

syltr

ansf

eras

ew

elH

; put

ativ

e gl

ycoy

sltra

nsfe

rase

(gal

acto

se-r

ich

EPS

bios

ynth

esis

)Su

gar t

rans

fera

seC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_020

47w

elI;

glyc

osyl

trans

fera

se, g

roup

1w

elI;

glyc

osyl

trans

fera

se (g

alac

tose

-ric

h EP

S bi

osyn

thes

is)

UD

P-G

lyco

syltr

ansf

eras

eEx

perim

enta

l va

lidat

ion

This

wor

k

LGG

_022

84G

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

78

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

LGG

_022

85yo

hH; g

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

LGG

_023

47*

Hyp

othe

tical

pro

tein

Puta

tive

glyc

osyl

trans

fera

sePf

am-

-LG

G_0

2562

#gl

mU

; UD

P-N

-ace

tylg

luco

sam

ine

pyro

phos

phor

ylas

eU

DP-

N-a

cety

lglu

cosa

min

e py

roph

osph

oryl

ase

Suga

r tra

nsfe

rase

Con

serv

atio

nK

anka

inen

et a

l., 2

009

LGG

_028

69G

lyco

syltr

ansf

eras

e, g

roup

1Pu

tativ

e gl

ycos

yltra

nsfe

rase

UD

P-G

lyco

syltr

ansf

eras

eC

onse

rvat

ion

Kan

kain

en e

t al.,

200

9

Locu

s ta

g: g

ene

iden

tifier

of t

he p

redi

cted

GT.

Gen

es fo

r whi

ch a

GT

activ

ity w

as p

redi

cted

in th

is st

udy

that

was

not

pre

sent

in th

e cu

rren

t ann

otati

on a

re m

arke

d w

ith a

sta

r (*

). Po

tenti

al fa

lse p

ositi

ve re

sults

are

indi

cate

d w

ith a

has

h (# ).

Curr

ent a

nnot

ation

: fun

ction

al a

nnot

ation

as

in c

urre

nt g

enom

e re

leas

e of

Gen

Bank

(NC_

0131

98.1

). Pr

opos

ed

anno

tatio

n: n

ew a

nnot

ation

pro

pose

d ba

sed

on th

e re

sults

of o

ur a

naly

sis. H

MM

: des

crip

tion

of th

e Hi

dden

Mar

kov

Mod

el (H

MM

) with

whi

ch th

e in

dica

ted

GT w

as id

entifi

ed.

Evid

ence

: Lev

el o

f evi

denc

e fo

r the

GT

activ

ity. C

onse

rvati

on: s

how

s sig

nific

ant s

eque

nce

cons

erva

tion

with

an

expe

rimen

tally

valid

ated

GT

in a

clos

ely r

elat

ed sp

ecie

s. E

xper

imen

tal

valid

ation

: the

GT

activ

ity h

as b

een

expe

rimen

tally

val

idat

ed in

Lac

toba

cillu

s rha

mno

sus G

G. R

efer

ence

: ref

eren

ce to

the

publ

icati

on(s

) sup

porti

ng th

e ev

iden

ce.

thre

e.

79

The good agreement between our predictions and known information on glycosylation in C. jejuni NCTC 11168 (Nothaft & Szymanski, 2010), suggests that also for L. rhamnosus GG, the predictions summarized in Table 3.2 reflect true GTs. In addition, Table 3.2 provides a curated annotation update of GTs in L. rhamnosus GG: besides adding novel predictions, we removed potential erroneous annotations that originated through homology-based associations (indicated by conservation in Additional file 1: Table S1) as especially for GTs it is difficult to extrapolate the functional annotation without further experimental evidence (e.g. for LGG_00279). For GTs putatively involved in polysaccharide biosynthesis (LGG_00279-LGG_00283, see below), gene names were corrected in agreement with the conventional gene nomenclature (Reeves et al., 1996).

Of the total number of 48 final predictions in L. rhamnosus GG (Table 3.2), five correspond to the experimentally documented locus encoding the enzymes involved in the synthesis of the complex galactose-rich EPS of L. rhamnosus GG (Lebeer et al., 2011; Lebeer et al., 2009). We also recovered the conserved cluster of GTs involved in the production of the intracellular storage glycogen-like polysaccharides (Kiel et al., 1994) and the GTs necessary for the biosynthesis of PG (Di Guilmi et al., 2003). In 33 cases, our predictions were consistent with previously annotated GTs (supported either by sequence conservation or by experimental evidence in related species). In five cases, indicated in Table 3.2 with a hash, our predictions are likely false positives. Eight of the 48 predicted GTs in L. rhamnosus GG were completely novel (indicated with a star in Table 3.2).

Among the novel predictions, two resulted from the screening with the ‘Rossmann-fold domain’ (class I) (LGG_01412 and LGG_00928, see Table 3.2). The other novel predictions LGG_01195 (previously annotated as ‘ABC transporter’), LGG_00985 (previously annotated as ‘integral membrane protein’) and LGG_02347 (previously annotated as ‘hypothetical protein’ were all detected by screening with the dedicated HMMs of class III (Table 1), further confirming the added value of these HMMs to find additional GTs. The screening with the HMMs of class II predicted as potential GTs LGG_00283 (a yet unannotated protein), LGG_01991 and LGG_01992. Both latter enzymes exhibit a high similarity with experimentally validated GTs in E. coli of the UDP-glycosyltransferase/Glycogen phosphorylase superfamily (Kiel et al., 1994), further suggesting their GT activity. However, they also show high sequence homology with UDP-N-acetylglucosamine 2-epimerases. This would be in agreement with the work of Campbell et al. (2000) showing that UDP-N-acetylglucosamine 2-epimerase has homology to phosphoglycosyl transferases and shares the same catalytic mechanism (Campbell et al., 2000).

Despite the similar number of predicted GTs, the genomic organization of these predicted GTs is very different in C. jejuni NCTC 11168 and L. rhamnosus GG. In C. jejuni NCTC 11168, about 82% of the predicted GTs (corresponding to 36 GTs) are clustered into seven genomic regions, each of which contains at least two and on average five GTs that are physically

80

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

located next to each other. The remaining eight predicted C. jejuni NCTC 11168 GTs are scattered in the genome (i.e. with no other GT present immediately up- or downstream). For L. rhamnosus GG, a smaller fraction of the predicted GTs is organized in clusters: about 56% of the predicted GTs (corresponding to 28 GTs) are located in 9 clusters, that are on average slightly smaller (with a mean size of three GTs) than those found in C. jejuni NCTC 11168. The remaining 20 predicted GTs in L. rhamnosus GG are isolated in the genome. For both species, most of the well-studied experimentally verified GTs are localized in these clusters, e.g. in C. jejuni NCTC 11168 these clusters correspond to the genomic regions involved in the synthesis of LOS, CPS and N- and in O-protein glycosylation (Guerry & Szymanski, 2008), whereas in L. rhamnosus GG one of the predicted clusters correspond to the known region for galactose-rich EPS (Lebeer et al., 2011; Lebeer et al., 2009) and one to the cluster for the biosynthesis of intracellular storage glycogen-like polysaccharides (Kankainen et al., 2009; Kiel et al., 1994). The function of the remaining seven clusters in L. rhamnosus GG is yet unknown.

Compared to the ones organized in clusters in both genomes, most of the GTs found in isolation appear to be much less studied. A closer inspection of these isolated GTs showed that in L. rhamnosus GG (in 7 of the 20 cases (LGG_01057, LGG_01069, LGG_01147, LGG_01412, LGG_01487, LGG_01538, LGG_02004)), but not in C. jejuni NCTC 11168, these isolated GTs are flanked by DNA topoisomerases, tyrosine recombinases, Holliday junction-specific endonucleases, phage-related resolvases and transposases (according to the current genome annotation of L. rhamnosus GG (NC_013198.1)). In addition, overlaying our predictions with the results of a previous comparative analysis between L. rhamnosus GG and its close relative L. rhamnosus LC705 (Kankainen et al., 2009), indicates that many of the isolated GTs we identified are specific for L. rhamnosus GG (such as LGG_02004). These observations, together with the lower fraction of GTs occurring in large genomic clusters, indicates that in L. rhamnosus GG, much more than in C. jejuni NCTC 11168, the glycosylation potential has been shaped by horizontal gene transfer and intra-genomic rearrangements, similarly to what has been observed for GTs belonging to family 6 of GTs in bacteria and vertebrates (CAZy database) (Brew et al., 2010; Cantarel et al., 2009).

3.3.2 network-baseD strategy reLating gts to tHeir substrate cLasses

To relate the predicted GTs to their potential substrates, we exploit the ‘local neighborhood’ of these GTs in a functional network, hereby assuming that GTs should be connected to their substrates, either directly or indirectly, via other GTs or enzymes. For the network, we relied on STRING, of which the functional interactions are inferred from physical (genome-wide protein-protein interactions, literature) and functional data (genomic co-localization, co-expression, co-occurrences, gene fusion-fission events) (Szklarczyk et al., 2011; von Mering et al., 2005). The local neighborhood of a predicted GT (or local subnetwork) is here defined as the nodes that directly connect to the predicted GT (the latter of which is also referred to as

thre

e.

81

the query GT) in the STRING network. We could derive 44 subnetworks for C. jejuni NCTC 11168, and 48 for L. rhamnosus GG. For each GT-specific subnetwork, the GO categories that were most overrepresented amongst the members of the subnetwork were used to infer for the query GT of each subnetwork a putative substrate class. As such we could predict a substrate class for 30/44 GTs in C. jejuni NCTC 11168 and for 20/48 GTs in L. rhamnosus GG, which related to saccharides, PG, proteins and lipids (see Additional file 2: Table S2 for C. jejuni NCTC 11168 and Table 3.3 for L. rhamnosus GG).

The relation of the predicted GTs with their network neighbours was further specified using information on putative membrane associations or presence of glycosylation sites in the network members (Materials and methods): a query-GT being connected to a transmembrane protein is referred to as a ‘membrane association’ and is indicative for soluble GTs that exert their action by interacting with transmembrane proteins, e.g. a transporter of glycoconjugates (Charbonneau et al., 2012; Dell et al., 2010; Grass et al., 2010; Mohammadi et al., 2007; Wu & Wu, 2011). A query-GT being connected to proteins with putative glycosylation sites hints towards the glycosylation of those proteins by the query-GT (substrate relation).

To gain insight in the mutual interactions between GTs and of these GTs with other proteins involved in the same process, we created ‘consensus networks’ that combine the local network neighbourhood of all GTs, predicted to belong to the same specificity class and of which the local subnetworks are enriched in the same GO terms (Figure 3.3).

82

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

Tabl

e 3.

3 - P

ropo

sed

subs

trat

e cl

asse

ss o

f pr

edic

ted

glyc

osyl

tran

sfer

ases

in L

acto

baci

llus

rha

mno

sus

GG

Que

ry-G

T lo

cus

tag

Que

ry-G

T lo

caliz

atio

nE

nric

hed

GO

ca

tego

ries

Mem

bran

e as

soci

atio

nPa

rtne

r G

TsPr

opos

ed

subs

trat

e cl

ass o

f th

e qu

ery-

GT

Pote

ntia

l pro

tein

su

bstr

ate

Evi

denc

eR

efer

ence

LGG

_002

80C

EPS

bios

ynth

esis

LGG

_002

78 (h

ypot

hetic

al p

rote

in)

LGG

_020

43

LGG

_002

81

LGG

_002

83

LGG

_002

95

LGG

_002

79

LGG

_019

99

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

anka

inen

et a

l.,

2009

LGG

_002

81C

EPS

bios

ynth

esis

; PS

trans

port

LGG

_002

78 (h

ypot

hetic

al p

rote

in)

LGG

_002

80

LGG

_002

95

LGG

_010

57

LGG

_002

79

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

anka

inen

et a

l.,

2009

LGG

_002

95C

EPS

bios

ynth

esis

LGG

_002

96 (i

nteg

ral m

embr

ane

prot

ein)

LGG

_002

80

LGG

_020

43

LGG

_002

81

LGG

_028

69

LGG

_010

57

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

anka

inen

et a

l.,

2009

LGG

_010

62*

CEP

S bi

osyn

thes

is-

LGG

_020

26

LGG

_020

23

LGG

_020

25

Extra

cellu

lar

sacc

harid

es-

--

LGG

_020

40$

CEP

S bi

osyn

thes

is;

nucl

eotid

e-su

gar

met

abol

ism

-LG

G_0

2042

LG

G_0

2046

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nB

lank

enfe

ldt e

t al

., 20

00

LGG

_020

42C

EPS

bios

ynth

esis

; nu

cleo

tide-

suga

r m

etab

olis

m

-LG

G_0

2040

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nB

lank

enfe

ldt

et a

l., 2

000;

K

anka

inen

et a

l.,

2009

LGG

_020

43TM

Pept

idyl

-tyro

sine

de

phos

phor

ylat

ion,

re

gula

tion

of c

atal

ytic

ac

itivi

ty, E

PS

bios

ynth

esis

-LG

G_0

1992

LG

G_0

2047

Extra

cellu

lar

sacc

harid

es-

Expe

rimen

tal

valid

atio

nLe

beer

et a

l.,

2009

; Kan

kain

en

et a

l., 2

009

LGG

_020

45C

Poly

sacc

harid

e bi

osyn

thes

is;

poly

sacc

harid

e tra

nspo

rt

LGG

_002

82 (p

olys

acch

arid

e tra

nspo

rter)

LGG

_009

98

LGG

_009

99

LGG

_020

46

LGG

_020

47

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

anka

inen

et a

l.,

2009

thre

e.

83

Tabl

e 3.

3 - P

ropo

sed

subs

trat

e cl

asse

ss o

f pr

edic

ted

glyc

osyl

tran

sfer

ases

in L

acto

baci

llus

rha

mno

sus

GG

Que

ry-G

T lo

cus

tag

Que

ry-G

T lo

caliz

atio

nE

nric

hed

GO

ca

tego

ries

Mem

bran

e as

soci

atio

nPa

rtne

r G

TsPr

opos

ed

subs

trat

e cl

ass o

f th

e qu

ery-

GT

Pote

ntia

l pro

tein

su

bstr

ate

Evi

denc

eR

efer

ence

LGG

_002

80C

EPS

bios

ynth

esis

LGG

_002

78 (h

ypot

hetic

al p

rote

in)

LGG

_020

43

LGG

_002

81

LGG

_002

83

LGG

_002

95

LGG

_002

79

LGG

_019

99

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

anka

inen

et a

l.,

2009

LGG

_002

81C

EPS

bios

ynth

esis

; PS

trans

port

LGG

_002

78 (h

ypot

hetic

al p

rote

in)

LGG

_002

80

LGG

_002

95

LGG

_010

57

LGG

_002

79

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

anka

inen

et a

l.,

2009

LGG

_002

95C

EPS

bios

ynth

esis

LGG

_002

96 (i

nteg

ral m

embr

ane

prot

ein)

LGG

_002

80

LGG

_020

43

LGG

_002

81

LGG

_028

69

LGG

_010

57

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

anka

inen

et a

l.,

2009

LGG

_010

62*

CEP

S bi

osyn

thes

is-

LGG

_020

26

LGG

_020

23

LGG

_020

25

Extra

cellu

lar

sacc

harid

es-

--

LGG

_020

40$

CEP

S bi

osyn

thes

is;

nucl

eotid

e-su

gar

met

abol

ism

-LG

G_0

2042

LG

G_0

2046

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nB

lank

enfe

ldt e

t al

., 20

00

LGG

_020

42C

EPS

bios

ynth

esis

; nu

cleo

tide-

suga

r m

etab

olis

m

-LG

G_0

2040

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nB

lank

enfe

ldt

et a

l., 2

000;

K

anka

inen

et a

l.,

2009

LGG

_020

43TM

Pept

idyl

-tyro

sine

de

phos

phor

ylat

ion,

re

gula

tion

of c

atal

ytic

ac

itivi

ty, E

PS

bios

ynth

esis

-LG

G_0

1992

LG

G_0

2047

Extra

cellu

lar

sacc

harid

es-

Expe

rimen

tal

valid

atio

nLe

beer

et a

l.,

2009

; Kan

kain

en

et a

l., 2

009

LGG

_020

45C

Poly

sacc

harid

e bi

osyn

thes

is;

poly

sacc

harid

e tra

nspo

rt

LGG

_002

82 (p

olys

acch

arid

e tra

nspo

rter)

LGG

_009

98

LGG

_009

99

LGG

_020

46

LGG

_020

47

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

anka

inen

et a

l.,

2009

LGG

_020

46C

EPS

bios

ynth

esis

; po

lysa

ccha

ride

trans

port

LGG

_020

49 (p

olys

acch

arid

e tra

nspo

rter)

LGG

_020

45

LGG

_020

47

LGG

_019

99

Extra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

anka

inen

et a

l.,

2009

LGG

_020

47C

Poly

sacc

harid

e bi

osyn

thes

is;

poly

sacc

harid

e tra

nspo

rt

LGG

_020

43 (u

ndec

apre

nyl-P

-β-

gluc

osep

hosp

hotra

nsfe

rase

)LG

G_0

2043

LG

G_0

2045

LG

G_0

2046

LG

G_0

2869

LG

G_0

0295

LG

G_0

1057

Extra

cellu

lar

sacc

harid

es-

Expe

rimen

tal

valid

atio

nTh

is w

ork

LGG

_010

62*

CG

lyco

gen

bios

ynth

esis

-LG

G_0

2026

LG

G_0

2023

LG

G_0

2025

Intra

cellu

lar

sacc

harid

es-

--

LGG

_020

23C

Gly

coge

n bi

osyn

thes

is;

pyrim

idin

e nu

cleo

side

m

etab

olis

m

-LG

G_0

2026

LG

G_0

1062

LG

G_0

2024

LG

G_0

2025

Intra

cellu

lar

sacc

harid

es-

Con

serv

atio

nYu

et a

l., 1

988;

K

anka

inen

et a

l.,

2009

LGG

_020

24C

Gly

coge

n bi

osyn

thes

is;

resp

onse

to a

ntib

iotic

-LG

G_0

2023

LG

G_0

2025

LG

G_0

2026

Intra

cellu

lar

sacc

harid

es-

Con

serv

atio

nK

iel e

t al.,

199

4;

Kan

kain

en e

t al.,

20

09

LGG

_020

25C

Gly

coge

n bi

osyn

thes

is-

LGG

_020

23

LGG

_020

24

LGG

_020

26

Intra

cellu

lar

sacc

harid

es-

Con

serv

atio

nB

allic

ora

et a

l.,

2003

; Kan

kain

en

et a

l., 2

009

LGG

_020

26C

Gly

coge

n bi

osyn

thes

is-

LGG

_020

23

LGG

_020

24

LGG

_020

25

Intra

cellu

lar

sacc

harid

es-

Con

serv

atio

nB

allic

ora

et a

l.,

2003

; Kan

kain

en

et a

l., 2

009

LGG

_009

98C

Car

bohy

drat

e m

etab

olis

m; l

ipid

s m

etab

olis

m

LGG

_009

95 (h

ypot

hetic

al p

rote

in)

LGG

_020

45

LGG

_009

99Li

pid

-C

onse

rvat

ion

Kan

kain

en e

t al.,

20

09

LGG

_009

99C

Car

bohy

drat

e m

etab

olis

m; l

ipid

s m

etab

olis

m

LGG

_009

95 (h

ypot

hetic

al p

rote

in)

LGG

_020

45

LGG

_009

98Li

pid

-C

onse

rvat

ion

Kan

kain

en e

t al.,

20

09

LGG

_010

57*

CC

arbo

hydr

ate

met

abol

ism

; lip

ids

met

abol

ism

LGG

_020

04 (s

ugar

or L

PS

synt

hesi

s tra

nsfe

rase

)LG

G_0

2004

LG

G_0

0280

LG

G_0

2043

LG

G_0

2869

LG

G_0

0295

LG

G_0

2046

LG

G_0

2047

Lipi

d-

--

84

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

LGG

_007

94TM

PG-b

ased

cel

l wal

l bi

ogen

esis

--

Pept

idog

lyca

n-

Con

serv

atio

nSc

hwar

tz e

t al.,

20

02; K

anka

inen

et

al.,

200

9

LGG

_012

83C

PG-b

ased

cel

l wal

l bi

ogen

esis

LGG

_011

92 (r

od sh

ape-

dete

rmin

ing

prot

ein

Rod

A)

LGG

_014

87Pe

ptid

ogly

can

-C

onse

rvat

ion

Men

gin-

Lecr

eulx

et

al.,

199

1;

Kan

kain

en e

t al.,

20

09

LGG

_014

87TM

PG-b

ased

cel

l wal

l bi

ogen

esis

-LG

G_0

1283

Pept

idog

lyca

n-

Con

serv

atio

nB

arre

tt et

al.,

20

07; K

anka

inen

et

al.,

200

9

LGG

_015

38*

TMPG

bio

synt

hetic

pr

oces

s; re

gula

tion

of c

ell s

hape

; de

phos

phor

ylat

ion;

re

spon

se to

an

tibio

tics

-LG

G_0

0280

Pept

idog

lyca

n-

--

LGG

_017

83TM

PG-b

ased

cel

l wal

l bi

ogen

esis

--

Pept

idog

lyca

n-

Con

serv

atio

nD

i Gui

lmi e

t al.,

20

03; K

anka

inen

et

al.,

200

9

LGG

_007

94*

TMR

egul

atio

n of

cel

l sh

ape;

cel

l cyc

le-

-Pr

otei

nLG

G_0

1280

(cel

l di

visi

on p

rote

in

FtsI

)

--

LGG

_008

25*

CPr

otei

n tra

nsla

tion

LGG

_007

51 (S

NA

RE

asso

ciat

ed

golg

i pro

tein

)LG

G_0

0826

Prot

ein

LGG

_008

29

(Yku

J pro

tein

)-

-

LGG

_008

26*

CPr

otei

n tra

nsla

tion;

am

ino

acid

tran

spor

tLG

G_0

0751

(SN

AR

E as

soci

ated

go

lgi p

rote

in)

LGG

_008

25

LGG

_020

47Pr

otei

nLG

G_0

0829

(Y

kuJ p

rote

in)

--

LGG

_011

47*

CD

NA

met

abol

ic

proc

ess

LGG

_011

46 (p

redi

cted

OR

F)-

Prot

ein

LGG

_011

45

(DN

A-e

ntry

nu

clea

se)

--

LGG

_012

83*

CR

egul

atio

n of

cel

l sh

ape;

resp

onse

to

ant

ibio

tic, c

ell

divi

sion

LGG

_011

92 (r

od sh

ape-

dete

rmin

ing

prot

ein

Rod

A)

LGG

_014

87Pr

otei

nLG

G_0

1280

(cel

l di

visi

on p

rote

in

FtsI

)

--

thre

e.

85

LGG

_007

94TM

PG-b

ased

cel

l wal

l bi

ogen

esis

--

Pept

idog

lyca

n-

Con

serv

atio

nSc

hwar

tz e

t al.,

20

02; K

anka

inen

et

al.,

200

9

LGG

_012

83C

PG-b

ased

cel

l wal

l bi

ogen

esis

LGG

_011

92 (r

od sh

ape-

dete

rmin

ing

prot

ein

Rod

A)

LGG

_014

87Pe

ptid

ogly

can

-C

onse

rvat

ion

Men

gin-

Lecr

eulx

et

al.,

199

1;

Kan

kain

en e

t al.,

20

09

LGG

_014

87TM

PG-b

ased

cel

l wal

l bi

ogen

esis

-LG

G_0

1283

Pept

idog

lyca

n-

Con

serv

atio

nB

arre

tt et

al.,

20

07; K

anka

inen

et

al.,

200

9

LGG

_015

38*

TMPG

bio

synt

hetic

pr

oces

s; re

gula

tion

of c

ell s

hape

; de

phos

phor

ylat

ion;

re

spon

se to

an

tibio

tics

-LG

G_0

0280

Pept

idog

lyca

n-

--

LGG

_017

83TM

PG-b

ased

cel

l wal

l bi

ogen

esis

--

Pept

idog

lyca

n-

Con

serv

atio

nD

i Gui

lmi e

t al.,

20

03; K

anka

inen

et

al.,

200

9

LGG

_007

94*

TMR

egul

atio

n of

cel

l sh

ape;

cel

l cyc

le-

-Pr

otei

nLG

G_0

1280

(cel

l di

visi

on p

rote

in

FtsI

)

--

LGG

_008

25*

CPr

otei

n tra

nsla

tion

LGG

_007

51 (S

NA

RE

asso

ciat

ed

golg

i pro

tein

)LG

G_0

0826

Prot

ein

LGG

_008

29

(Yku

J pro

tein

)-

-

LGG

_008

26*

CPr

otei

n tra

nsla

tion;

am

ino

acid

tran

spor

tLG

G_0

0751

(SN

AR

E as

soci

ated

go

lgi p

rote

in)

LGG

_008

25

LGG

_020

47Pr

otei

nLG

G_0

0829

(Y

kuJ p

rote

in)

--

LGG

_011

47*

CD

NA

met

abol

ic

proc

ess

LGG

_011

46 (p

redi

cted

OR

F)-

Prot

ein

LGG

_011

45

(DN

A-e

ntry

nu

clea

se)

--

LGG

_012

83*

CR

egul

atio

n of

cel

l sh

ape;

resp

onse

to

ant

ibio

tic, c

ell

divi

sion

LGG

_011

92 (r

od sh

ape-

dete

rmin

ing

prot

ein

Rod

A)

LGG

_014

87Pr

otei

nLG

G_0

1280

(cel

l di

visi

on p

rote

in

FtsI

)

--

LGG

_014

87*

TMR

egul

atio

n of

cel

l sh

ape;

cel

l div

isio

n-

LGG

_012

83Pr

otei

nLG

G_0

1706

(cel

l di

visi

on p

rote

in/

peni

cilli

n-bi

ndin

g pr

otei

n 2)

; LG

G_0

1280

(cel

l di

visi

on p

rote

in

FtsI

); LG

G_0

0254

(D

-ala

nyl-

D-a

lani

ne

carb

oxyp

eptid

ase)

--

LGG

_017

83*

TMR

egul

atio

n of

cel

l sh

ape;

cel

l cyc

le-

-Pr

otei

nLG

G_0

1280

(cel

l di

visi

on p

rote

in

FtsI

)

--

Locu

s tag

: gen

e id

entifi

er o

f the

pre

dict

ed G

T us

ed a

s que

ry in

STR

ING

to o

btai

n a

quer

y-de

pend

ent s

ubne

twor

k. L

ocal

izatio

n: in

dica

tes w

heth

er th

e qu

ery-

GT w

as p

redi

cted

to b

e a

cyto

plas

mic

(C) o

r a tr

ansm

embr

ane

prot

ein

(TM

). En

riche

d G

O ca

tego

ries:

GO

cat

egor

ies e

nric

hed

amon

gst t

he m

embe

rs o

f the

que

ry-d

epen

dent

subn

etw

ork

of th

e in

dica

ted

quer

y-GT

. Onl

y ca

tego

ries

show

ing

an e

nric

hmen

t va

lue

of p

< 0

.05

are

show

n (a

ccor

ding

to a

hyp

erge

omet

ric te

st c

orre

cted

for

mul

tiple

tes

ting

usin

g Fa

lse D

iscov

ery

Rate

). M

embr

ane

asso

ciati

on: r

efer

s to

edg

es b

etw

een

the

quer

y-GT

and

mem

bers

of i

ts s

ubne

twor

k pr

edic

ted

to b

e tr

ansm

embr

ane

prot

eins

. Par

tner

GTs

: pre

dict

ed/e

xper

imen

tally

va

lidat

ed G

Ts t

hat

belo

ng t

o th

e su

bnet

wor

k of

the

que

ry-G

T. P

ropo

sed

subs

trat

e cl

ass

of t

he q

uery

-GT:

infe

rred

fro

m t

he G

O e

nric

hmen

t an

alys

is of

the

que

ry-d

epen

dent

su

bnet

wor

k of

the

indi

cate

d qu

ery-

GT d

eriv

ed fr

om S

TRIN

G. N

ovel

subs

trat

e pr

edic

tions

der

ived

from

this

stud

y ar

e in

dica

ted

by a

star

(*) n

ext t

o th

e lo

cus t

ag o

f the

corr

espo

ndin

g qu

ery-

GT. P

oten

tial p

rote

in s

ubst

rate

: it r

efer

s to

edg

es b

etw

een

the

quer

y-GT

and

mem

bers

of i

ts s

ubne

twor

k pr

edic

ted

to h

ave

N- o

r O-g

lyco

syla

tion

sites

. Suc

h pr

otei

ns a

re

ther

efor

e su

gges

ted

to b

e po

tenti

al su

bstr

ates

of t

he q

uery

-GT

in th

e ca

ses w

here

pro

tein

s are

the

prop

osed

subs

trat

e. E

vide

nce:

leve

l of e

vide

nce

for t

he p

redi

cted

subs

trat

e cl

ass

of th

e qu

ery-

GT. C

onse

rvati

on: s

how

s sig

nific

ant s

eque

nce

cons

erva

tion

with

a G

T fo

r whi

ch th

e su

bstr

ate

spec

ifici

ty h

as b

een

expe

rimen

tally

val

idat

ed in

clo

sely

rela

ted

spec

ies.

Ex

perim

enta

l val

idati

on: t

he s

ubst

rate

spe

cific

ity o

f the

GT

has

been

exp

erim

enta

lly v

alid

ated

in L

acto

baci

llus

rham

nosu

s GG.

Ref

eren

ce: p

ublic

ation

(s) s

uppo

rting

the

pred

icte

d su

bstr

ate

clas

s of t

he q

uery

-GT.

86

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

Figure 3.3 - Consensus networks derived for each of the predicted substrate classes of putative GTs in L. rhamnosus GG.Consensus networks show all GTs, having the same substrate class, together with their protein neighbors that are hypothesized to contribute to the same common glycosylation mechanism as the one the GTs are involved in. On the consensus networks, nodes are proteins than can either be GTs (green nodes), transmembrane proteins (orange nodes) or proteins containing glycosylation signals (violet nodes). Membrane associations established between GTs and transmembrane proteins are represented by blue edges while predicted substrate relations between GT and proteins containing glycosylation signals are represented by yellow edges. Black edges refer to interactions between predicted GTs. If the local network neighborhood of GTs (local subnetwork) belonging to the same substrate class shows enrichment in more than one GO category (e.g. both the GO terms of EPS and glycogen biosynthesis), the consensus network is shown for each of the enriched GO categories. A: consensus networks involving GTs, predicted to glycosylate saccharides. Note that here two independent consensus networks were derived corresponding to respectively extracellular and intracellular PS biosynthesis. B: consensus network involving GTs, predicted to glycosylate peptidoglycan (PG). C: consensus network involving GTs, predicted to glycosylate lipids. D: consensus networks involving GTs, predicted to glycosylate proteins. Three independent consensus networks were derived corresponding to respectively cell cycle regulation, protein translation and DNA metabolic processes. Our analysis suggests substrate promiscuity for MurG, PBP1A, PBP1B and PBPA, all of which were predicted to be involved in the glycosylation of both peptidoglycan and proteins.

3.3.3 inferreD substrate cLasses of preDicteD gts in tHe bencHmark

To assess the extent to which our network-based approach was able to correctly infer substrate classes, we used as benchmark again the 10 GTs in C. jejuni NCTC 11168 for which also the substrate specificity is known (see Material and methods). Our strategy was able to recover the known substrate class of all 10 GTs (sensitivity of 100%) on a total of 31 predicted substrate classes for GTs in C. jejuni (true positive rate of 10/31).

thre

e.

87

3.3.4 inferreD substrate cLasses of preDicteD gts in L. rhamnosus gg

The 20 GTs in L. rhamnosus GG for which we could predict their putative substrate class are summarized in Table 3.3.

3.3.4.1 gts preDicteD to gLycosyLate saccHariDes

In L. rhamnosus GG, the substrate class saccharides (Figure 3.3A) comprises the largest number of GTs, which is to be expected as saccharides are the most common substrates for GTs. The group of GTs that could be related to saccharides comprises two consensus networks: the first consensus network consists of GTs that, according to their GO annotation are involved in the biosynthesis of extracellular polysaccharides (WclC, WclB, WelE, WelG, WelH, WelI, RmlA2, LGG_00295) (Lebeer et al., 2011; Lebeer et al., 2009). The topology of this consensus network is indicative for en bloc glycosylation (Guerry & Szymanski, 2008; Hug & Feldman, 2011) because it contains several interconnected soluble GTs, all linked to a membrane-bound priming GT together with Wzx flippases that transfer the subunits en bloc (see below).

This consensus network (Figure 3.3A) can be further subdivided into two cliques of interconnected GTs. The first clique (welI-welG-welH-rmlA1-rmlA2) contains genes involved in the synthesis of galactose-rich EPS, such as amongst others WelE (LGG_02043), the priming GT, with an experimentally verified substrate (Lebeer et al., 2009). From the previously annotated gene cluster for galactose-rich EPS (Lebeer et al., 2011; Lebeer et al., 2009), our analysis only missed welJ, annotated as alpha-1,3-galactosyltransferase (LGG_02048), as this gene was not predicted as a GT in our analysis. This gene does not appear to contain any signatures of the currently known HMMs for GTs and might represent a false negative of our analysis or an erroneous annotation in the current release of the L. rhamnosus GG genome NC_013198.1. This last hypothesis is supported by the small gene size of welJ, which would be atypical for a GT.

Regarding the second clique (wclC-LGG_00295-wclB), it contains genes for which the substrate specificity towards saccharides is known from homology-based extrapolation only. As we know from previous work that L. rhamnosus GG contains, besides its galactose-rich EPS also shorter, glucose-rich polysaccharides structures, we would hypothesize that this clique contains the missing genes for those glucose-rich polysaccharides structures (Francius et al., 2008). The prediction of an independent Wzx flippase for each of the sets of interconnected GTs (cliques) (i.e. LGG_02049 for the galactose-rich clique and WclC and WclB for the clique putatively responsible for glucose-rich EPS synthesis), together with the known exquisite substrate specificity of Wzx flippases (Islam & Lam, 2013) further supports the hypothesis of each clique being responsible for the biosynthesis of another glycan type. Assuming that indeed the upper clique is involved in the synthesis of glucose-rich saccharide structures implies that the predicted link between WelE and this second clique (WclC, LGG_00295

88

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

and WclB) must be mere functional (i.e. not invoking a direct interaction), since knock-out experiments indicate that WelE is not the direct priming GT of the glucose-rich EPS structures (Lebeer et al., 2009).

The second consensus network (Figure 3.3A lower part, GlgA, GlgC, GlgD, GlgP, GalU) recapitulates all known members of the glycosylation system involved in glycogen synthesis except GlgB (LGG_02027), a conserved glycogen branching enzyme with transglycosylase activity, i.e. an enzyme that has both hydrolase and GT characteristics (Lim et al., 2003), which was not picked up by our HMM-based search step. From the predicted GTs in this network only GlgA, previously already known as a glycogen synthase, seems to be a genuine GT (Kiel et al., 1994; Wilson et al., 2010). For the other proteins GlgC, GlgD and GlgP, GalU -though related to glycan biosynthesis- enzyme activities other than GT activity have been documented (Ballicora et al., 2003). The consensus network of the glycogen enzymes is composed solely of soluble proteins, which is in agreement with the intracellular nature of the glycogen-like polysaccharides. The connectivity between only soluble GTs points towards a sequential glycosylation mechanism in which sugar monomers are directly transferred from activated sugar-nucleotide donors (probably produced by GalU) to the respective substrates.

3.3.4.2 gts preDicteD to gLycosyLate peptiDogLycans

Five GTs could be related to PG precursors (PBP1A, PBP1B, PBP2A, MurG and LGG_01538), an annotation that has previously been suggested based on sequence conservation of these GTs across species (Figure 3.3B). GO enrichment analysis of their functional subnetworks suggests, both in L. rhamnosus GG (Table 3.3) and C. jejuni NCTC 11168 (Additional file 2: Table S2), a link between PG biosynthesis and a diverse set of processes, such as the regulation of cell shape, cell cycle and response to antibiotics, in agreement with the well-known functions of PG. Compared to the genes involved in EPS biosynthesis, it is remarkable that the GT genes involved in PG biosynthesis and remodelling do not occur in genomic clusters. The diversity of the processes in which these PG GTs are involved, might imply their necessity to be expressed under different environmental stimuli, which in turn can explain their organization in individual transcriptional units rather than in operons.

The consensus network of this class of GTs (Figure 3.3B) shows that all of these GTs are predicted to have transmembrane domains except for the soluble protein encoded by murG. The network organization is consistent with the known two-stage mechanism of bacterial PG biosynthesis consisting of cytoplasmic glycosylation reactions mediated by soluble GTs, followed by membrane-bound transglycosylation activities (Peregrin-Alvarez et al., 2009; van Heijenoort, 2001).

3.3.4.3 gts preDicteD to gLycosyLate LipiDs

The group of GTs that could be related to lipids contains three predicted GTs (LGG_00998, LGG_00999, LGG_01057) (Figure 3.3C). For these three GTs, their respective functional

thre

e.

89

subnetworks showed enrichment for the terms ‘carbohydrate’ and ‘lipid metabolism’, suggesting that they are involved in the synthesis of lipoglycans present on the cell wall of the Gram-positive bacterium L. rhamnosus GG. This predicted role is more plausible than their homology based annotated role as ‘LPS biosynthesis glycosyltransferases’, as LPS molecules are absent in Gram-positives. The sparsity of the consensus network of these three GTs might be due to the incompleteness of the STRING network. So far, the existence of lipoglycans in L. rhamnosus GG has not yet been shown by biochemical studies.

3.3.4.4 gts preDicteD to gLycosyLate proteins

A final group of seven GTs could be related to protein substrates and contains both predicted transmembrane (PBP1A, PBP1B, PBP2A) and predicted soluble GTs (LGG_00825, LGG_00826, LGG_01147, MurG). The GTs in this class were classified as protein GTs because the putative protein substrates in their subnetworks carry glycosylation signals. The GTs fall apart in three consensus subnetworks related to respectively cell cycle regulation, protein translation and DNA metabolic processes (Figure 3.3D).

A first consensus network comprises three transmembrane GTs (PBP1A, PBP2A, PBP1B) and MurG all predicted to be involved in ‘cell cycle regulation’ (according to the GO enrichment analysis of their respective subnetworks). Their consensus network points towards a substrate relation between each of the four GTs MurG, PBP1A, PBP2A and PBP1B, and cell division proteins (between MurG, PBP1A, PBP2A and PBP1B and the cell division protein FtsI on the one hand and between PBP1A, LGG_01706 and LGG_00254 on the other hand). Two previous studies further support our predictions: in Bacteroides fragilis FtsI, and other cell cycle related proteins such as FtsX and FtsQ, have been shown to be glycosylated (Fletcher et al., 2011). In addition, a very recent study in L. plantarum WCFS1 (Fredriksen et al., 2013) provides experimental evidence for the glycosylation of the cell division proteins FtsY, FtsZ, and FtsK 1 (Fredriksen et al., 2013). Our results -on the other hand- indicate that the three transmembrane GTs and MurG, known to be involved in PG biosynthesis show substrate promiscuity and would also have relations with protein substrates in L. rhamnosus GG (Table 3.3). A link between PG biosynthesis and protein glycosylation is not completely impossible given the fact that these predicted ‘promiscuous’ GTs co-occur with their predicted protein substrates including FtsI in cell division multienzyme complexes (Figure 3.4).

This link between PG biosynthesis and protein glycosylation is further supported by the fact that the other predicted protein substrate of PBP1A (the D-alanyl-D-alanine carboxypeptidase (LGG_00254)), is also known to be directly involved in PG biosynthesis by introducing interpeptide cross-links. Although not yet reported for D,D transpeptidases, other PG remodeling enzymes such as the PG hydrolases Msp1 in L. rhamnosus GG (Lebeer et al., 2012b) and Acm2 in L. plantarum WCFS1 (Fredriksen et al., 2012) were recently shown to be glycosylated (Barinka et al., 2004).

90

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es Figure 3.4 - Protein glycosylation of the cell division machinerySchematic overview of the cell division machinery of L. rhamnosus GG. PBP1A, PBP1B, PBPB2A and MurG are predicted to be putative GTs. Our network-based analysis predicted PBP3, FtsI and PBP2B as putative substrates of the indicated GTs. The Msp1 cell wall hydrolase is the experimentally validated glycoprotein in L. rhamnosus GG (Lebeer et al., 2012b).

A second consensus cluster is composed of two soluble GTs predicted to be involved in ‘protein translation’ (LGG_00825-LGG_00826). Both of these GTs were predicted to participate in the glycosylation of YkuJ, a protein co-translated with CcpC, a repressor of the tricarboxylic acid cycle in Bacillus subtillis (Figure 3.3C) (Commichau et al., 2006). LGG_00825 and LGG_00826 also exhibit a membrane association mediated by LGG_00751, annotated in L. rhamnosus GG as a hypothetical protein with a pfam09335 domain typical for SNARE associated Golgi proteins in eukaryotes. The membrane association of both GTs via a protein involved in translation, together with the fact that the subnetwork of LGG_00825 is enriched in the function ‘protein translation’ is consistent with the existence of an eukaryotic counterpart of sequential co-translational glycosylation in bacteria (Dell et al., 2010).

A last consensus network comprises only one GT, LGG_01147, predicted to be involved in ‘DNA metabolic processes’. LGG_01147 shows a substrate relation with LGG_01145, encoding a putative DNA entry nuclease, while establishing a membrane association mediated by LGG_01146 (Figure 3.3C). Little is known about these interacting partners, but nucleases

thre

e.

91

are often glycosylated in eukaryotes (Pimkin et al., 2006). Although not specifically related to nucleases, glycosylation of extracellular enzymes has been reported in prokaryotes (Brechtel et al., 1999; Fredriksen et al., 2012; Huang et al., 1988; Lebeer et al., 2012b; Olsen & Thomsen, 1991) and is thought to promote their stability (Lebeer et al., 2012b). Whether this is also the case for LGG_01145 needs to be further substantiated.

3.3.5 experimentaL anaLysis of tHe gt network invoLveD in eps biosyntHesis

Figure 3.5 - Experimental validation of the EPS network hierarchyA: Total cell wall polysaccharides were extracted from respectively L. rhamnosus GG wild type, a ΔwelE::TcR gene deletion mutant (CMPG5351) and ΔwelI::TcR gene deletion mutant (CMPG10811). The total amount of EPS was measured. Error bars indicate standard deviations (of three repeats). One-way ANOVA statistical analysis rendered a p-value smaller than 0.05 for the variation of EPS across strains. B: Sugar monomer composition. The data are expressed as relative amounts, taking the total amount of detected monomeric sugars as 100%. Error bars indicate standard deviations (of three repeats). One-way ANOVA analyses (performed independently on each of the three datasets) rendered significant p-values (<0.05) for the variation of each sugar monomer across strains. C: Adhesion capacity. The adhesion capacity of wild type and mutants to Caco-2 cells is compared. Error bars indicate standard deviations (of three repeats). A One-way ANOVA analysis rendered a significant p-value (<0.05) for the variation of the adhesion capacity of the strains.

92

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

We experimentally validated the GT network hierarchy within the clique for galactose-rich EPS (Figure 3.3A) by constructing a gene deletion mutant in the welI gene and comparing its phenotype to the phenotypes of the wild type (WT) and the gene deletion mutant of the priming GT WelE. As phenotypes, we tested the amount and monomer composition of EPS, and the adhesion capacity to the intestinal epithelial cell line Caco-2 as an indirect measurement of the EPS level (Lebeer et al., 2009). According to our predictions, WelI would be one of the GTs that transfer sugar moieties to the sugar subunit initiated by the priming GT WelE. Based on these predictions, a gene deletion mutant of WelI would be expected to affect the amount of EPS, as in the absence of WelI less sugar moieties will be transferred to the subunit initiated by the WelE, but the effect of the WelI deletion on the phenotype should be less severe than the effect observed when deleting the priming GT WelE. A phenotype for the welI mutant intermediate between the WT and the welE gene deletion mutant is indeed observed for both assays confirming the predicted role of WelI upstream of WelE: the ΔwelI::TcR mutant displays a lower galactose-rich EPS content than the WT, but a higher content and more galactose than the gene deletion mutant of the priming GT WelE (Figure 3.5A and B). In agreement with EPS having a negative effect on adhesion, the adherence capacity is the highest for the welE mutant, intermediate for the welI mutant and lowest for the WT (Figure 3.5C).

3.4 Discussion

In this work we developed an analysis flow that uses sequence-based strategies to predict novel GTs, but also exploits a network-based approach to infer the substrate classes of these putative GTs. Using a broad definition of GT activity, including also HMMs for OSTs and other non-typical GTs, allowed covering a large part of the glycosylation potential. Applying our flow resulted in a careful revision of GTs in the current genome annotation of L. rhamnosus GG (NC_013198.1). We confirmed the identity of 33 GTs and predicted 8 novel ones. In contrast to what is observed in C. jejuni NCTC 11168, GTs appear to be much less clustered in genomic regions, but rather occur as isolated genes flanked by transposable elements. This points towards a key role of horizontal gene transfer in the acquisition of the glycosylation potential of L. rhamnosus GG.

Complementing the sequence-based with a network based-approach allowed us to also relate some of those GTs to their potential substrates. Most prior experimental studies focused on analyzing the specificity of GTs organized in clusters together with their auxiliary enzymes, as this allows for the straightforward extrapolation of known specificities of some members to all members in the cluster. By considering, next to the genomic organization, also links in a functional network, we could predict the substrate classes for the numerous, isolated GTs in L. rhamnosus GG. Exploiting membrane associations and substrate relations for the nodes in the GT-centered networks helped predicting the mutual relations between the GTs and between the GTs and their substrates.

thre

e.

93

Our analysis contributed to the annotation of GTs in L. rhamnosus GG. For instance, we hypothesize that one of the genomic regions that was previously annotated to be involved in EPS biosynthesis in general would contain the missing genes involved in the biosynthesis of short glucose-rich polysaccharides that are known to decorate the surface of L. rhamnosus GG (Francius et al., 2008). In addition, we uncovered several novel interactions. For instance, for the isolated GTs known to be involved in PG biosynthesis (PBP1B, PBP2A, PBP1A and MurG), our network-based approach suggests an additional role in the glycosylation of proteins that are either involved in the biosynthesis of the PG (LGG_00254) or in cell division (LGG_01280 or FtsI). Substrate promiscuity of GTs is not uncommon in bacteria as for instance in Gram-negative pathogens, enzymes with relaxed specificity are shared between different processes, such as LPS and glycoprotein biosynthesis (Guerry & Szymanski, 2008; Karlyshev et al., 2005). Validating the activity of GTs that were predicted to glycosylate proteins- is cumbersome, as in vitro enzymatic assays do not represent the cellular conditions that are relevant for the assembly of these GTs in multienzyme membrane-associated complexes (Zapun et al., 2012). However, because PG biosynthesis is a process involving multienzyme complexes for which the assembly is tightly regulated (Zapun et al., 2012), it is not unlikely that also protein glycosylation would act as an additional regulatory layer in this structural complex formation. Provided our hypothesis on their substrate specificity towards both proteins and PG would be true, these promiscuous GTs (PBP1B, PBP2A, PBP1A and MurG) are unlikely to be the priming GTs of their putative protein substrates, given their well characterized specificities towards PG precursors in both Gram-positives and negatives (Sauvage et al., 2008). We hypothesize that the priming GTs predicted to be involved in protein glycosylation must be (Lactobacillus) species- or strain-specific rather than generally conserved in prokaryotes. This is supported by the observation that the best documented glycoprotein in L. rhamnosus GG, i.e. Msp1, another protein associated to the divisome (Lebeer et al., 2012b) (see Figure 3.4), was no longer glycosylated after transfer to the Gram-negative E. coli (Claes et al., 2012) despite the fact that E. coli also has PBP1A, PBP1B, PBP2A and MurG homologs. In addition, the sugar monomers added on Msp1 (Lebeer et al., 2012b) and related PG hydrolases such as Acm2 (Rolain et al., 2013) show different sugar lectin specificities in L. rhamnosus, L. casei and L. plantarum.

3.5 concLusions

Our results show how combining sequence- and network-based computational predictions can unveil insights in the bacterial glycosylation potential, thereby providing novel links and interesting hypotheses for further investigation.

Future experiments are needed to confirm the predicted GT activity of these proteins, both by knockout mutagenesis and biochemical analyses after recombinant expression and purification. During this PhD work eleven knockout mutants of the predicted GTs were already constructed, with a focus on the ones predicted to target proteins (cf. addendum). Preliminary

94

a ne

twor

k-ba

sed

appr

oach

to id

entif

y ba

cter

ial g

lyco

syltr

ansf

eras

es

screening of these mutants did not result in a significant phenotype or the elucidation of their substrate. A more thorough (glyco)proteomic screening is needed, combined with biochemical assays underpinning the GT activity of these enzymes. Moreover, the known promiscuity and redundancy of GTs will require the construction of double and triple knockout mutants, which is currently ongoing.

3.6 aDDitionaL fiLes

The additional files of this work can be consulted in the published paper of this chapter (Sanchez-Rodriguez et al., 2014), which is open access (http://www.biomedcentral.com/1471-2164/15/349).

3.6.1 aDDitionaL fiLe 1: tabLe s1

List of glycosyltransferases predicted in the genome of C. jejuni NCTC 11168.

3.6.2 aDDitionaL fiLe 2: tabLe s2

Proposed substrate classes of glycosyltransferases in C. jejuni NCTC 11168.

thre

e.

95

four. endogenous biotin-binding

proteins: an overlooked factor causing false

positives in streptavidin-based

protein detection

the struggle with the technical challenges of studying non-template driven processes

four. endogenous biotin-binding

proteins: an overlooked factor causing false

positives in streptavidin-based

protein detection

the struggle with the technical challenges of studying non-template driven processes

98

endo

geno

us b

iotin

-bin

ding

pro

tein

s

This chapter was published:

Tytgat H.L.P., Schoofs G., Driesen M., Proost P., Van Damme E.J.M., Vanderleyden J., Lebeer S. (2015) Endogenous biotin-binding proteins: an overlooked factor causing false positives in streptavidin-based protein detection. Microbial Biotechnology, 8, 164-168.

autHors’ contributions

HT conceived the study and performed experimental procedures. Lectin experiments were done in collaboration with EV. PP performed the Edman degradation. HT drafted the manuscript.

four

.

99

summary

Biotinylation is widely used in DNA, RNA and protein probing assays as this molecule has generally no impact on the biological activity of its substrate. During the ‘high troughput’ streptavidin-based detection of glycoproteins in L. rhamnosus GG with biotinylated lectin probes, a strong positive band of approximately 125 kDa was observed, present in different cellular fractions. This potential glycoprotein reacted heavily with the Concanavalin A (ConA), a lectin that specifically binds glucose and mannose residues. Surprisingly, this protein of 125 kDa could not be purified using a ConA affinity column. Edman degradation of the protein, isolated via cation and anion exchange chromatography, lead to the identification of the band as pyruvate carboxylase, an enzyme of 125 kDa, which binds biotin as a cofactor. Detection using only the streptavidin conjugate resulted in more false positive signals of proteins, also in extracellular fractions, indicating biotin-associated proteins. Indeed, biotin is a known cofactor of numerous carboxylases.

The potential of false positive bands with biotinylated protein probes should thus be considered when using streptavidin-based detection, e.g. by developing a blot using only the streptavidin conjugate. To circumvent these false positives, alternative approaches like detection based on digoxigenin labeling can also be used.

100

endo

geno

us b

iotin

-bin

ding

pro

tein

s

4.1 introDuction

Biotin is a small molecule, which is widely used in molecular biology as a result of its extremely high affinity for streptavidin binding (Kd = 10-14 - 10-15M) (Green, 1975; Laitinen et al., 2006). The streptavidin-biotin interaction is one of the strongest non-covalent bonds known in nature, in strength almost matching covalent strength (Chaiet & Wolf, 1964; Laitinen et al., 2006). The rigid nature of the bond results in resistance of the complex against organic solvents, denaturants, detergents, proteolytic enzymes and extreme pH and temperature conditions. This is exploited in biological assays by coupling biotin to substrates like DNA, RNA and proteins, such as the sugar-binding lectins important in our study of bacterial glycoproteins. Detection and purification of these biotinylated substrates is then accomplished by application of a streptavidin conjugate tagged with an enzyme reporter (like alkaline phosphatase or horse radish peroxidase) or a fluorescent probe (Chapman-Smith & Cronan, 1999; Chevalier et al., 1997). Another important advantage of this detection method is the low molecular weight of the biotin, which enables linkage to its substrate without affecting the substrate activity. To further limit effects on substrate activity, the biotin is mostly linked via an extending long linker (Chapman-Smith & Cronan, 1999; Chevalier et al., 1997).

Biotin is an important cofactor of carboxylase enzymes in all domains of life. The biotin cofactor is involved in the transfer of carbon dioxide in oxidative metabolism reactions. Biotin is bound via a biotin carboxyl carrier protein, which can be part of a multidomain carboxylase or substitute a separate subunit of the carboxylase enzyme complex (Fugate & Jarrett, 2012; Tong, 2013; Wood & Barden, 1977). Some of these proteins can be moonlighting proteins, as described in yeast (Huberts et al., 2010).

Earlier reports already mentioned the presence of proteins binding biotin as a cofactor resulting from false positive reactions during gene probing in bacteria (Wang et al., 1993), and several reports have been published on the interference of endogenous biotin in eukaryotic assays (Chen et al., 2012; Horling et al., 2012; McKay et al., 2008). Nevertheless, many research groups still use biotinylated probes to detect specific DNA, RNA and protein structures. With the booming interest in glycobiology and bacterial glycoproteins, biotinylated lectins are now widely used to identify specific sugar residues (Coyne et al., 2005; Lebeer et al., 2012b; Lee et al., 2014; Wu et al., 2014). We report in this chapter on the occurrence of false positive hits due to endogenous biotin-binding proteins when biotinylated probes are used to probe (glyco)proteins.

As yet mentioned, an integral part of the experimental approach of this PhD research, is based on the detection of specific sugar modifications on proteins of L. rhamnosus GG using biotinylated lectins. Because we aimed here at an open approach to explore whether L. rhamnosus GG glycosylates other proteins than Msp1, it was important to investigate the robustness of the lectin-based screening approach. Intriguingly, an apparent strong positive protein band of approx. 125 kDa present in several proteome fractions could not be purified

four

.

101

using lectin affinity chromatography. The protein turned out to be LGG_01329, a pyruvate carboxylase, which harbors a domain that binds a biotin cofactor. This protein did not stain with the glycoprotein Periodic Acid Schiff (PAS) base staining method, and is thus not a glycoprotein. Probing the protein samples using only streptavidin, lead to the discovery of even more false positive hits. We therefore suggest the implementation of alternatives such as the digoxigenin detection method to circumvent these false hits.

4.2 resuLts anD Discussion

4.2.1 Detection of an intruiging 125 kDa banD using biotinyLateD Lectins

As yet mentioned, the discovery of a glycosylated protein in L. rhamnosus GG (ATCC 53103) (Lebeer et al., 2012b) raised the question if this bacterium is able to produce more glycoproteins. Therefore a Western blot screening was designed in which an array of biotinylated lectins was used to sample the glycosylation state of the proteome of L. rhamnosus GG. Lectins are (glyco)proteins which, similar to the use of antibodies in Western blot development, can bind specific carbohydrate structures. Their binding affinity depends on the configuration and accessibility of the specific sugar(s) that are recognized by the lectins under study (Van Damme et al., 2011).

We applied an array of biotinylated lectins with different specificities to Western blots of proteome samples of L. rhamnosus GG to screen for glycosylated proteins. We here depict the results (Fig. 4.1A) for Western blots of the extracellular proteome developed using Concanavalin A (ConA) (specific for glucose, terminal mannose) and a mix of lectins consisting out of ConA, the Galanthus nivalis lectin GNA (terminal mannose), the Hippeastrum hybrid HHA lectin (mannose), Wheat germ lectin (WGA) [N-acetylglucosamine, (GlcNAc)], the DSL isolated from Datura stramonium (GlcNAc), UDA originating from Urtica dioica (GlcNAc), Nictaba purified from Nicotiana tabacum leaves (GlcNAc), the Rhizoctonia solani lectin RSA [Galactose (Gal), N-acetylgalactosamine, (GalNAc)] and PNA purified from peanuts (Gal, GalNAc) lectins. Figure 4.1A shows the resulting lectin blots for the proteome of wild type L. rhamnosus GG and the ΔdltD::TcR mutant (CMPG5540) (Perea Velez et al., 2007). After incubation with the biotinylated lectins, the Western blots were probed with a streptavidin conjugate, which enabled the visual detection of bands (Fig. 4.1A).

To our surprise, both wild type and the ΔdltD::TcR mutant samples showed a strong band at approximately 125 kDa (Fig. 4.1A). All repetitions of the assays using single biotinylated lectins or mixtures of lectins confirmed these findings (results not shown). The same 125 kDa band was also observed on Western blots of the proteomes of other mutants of L. rhamnosus GG as well as in the probing of other cellular proteomic fractions (results not shown).

102

endo

geno

us b

iotin

-bin

ding

pro

tein

s

Figure 4.1 - Detection of glycoproteins on Western blot using biotinylated lectins and streptavidin results in false positive hits.(A) Biotinylated lectin blots - The exoproteome of wild type L. rhamnosus GG and its ΔdltD::TcR mutant were subjected to SDS-PAGE on NuPAGE® Novex® 12% Bis-Tris gels (Life Technologies) and subsequently blotted to PVDF membranes (Life Technologies). The prestained KaleidoscopeTM ladder (Bio-Rad) was added as a molecular weight marker. The Western blots were developed using biotinylated probes, in this case lectins. In the left panel the Western blot was probed with biotinylated ConA, which specifically binds glucose and terminal mannose. Incubation with streptavidin conjugated to alkaline phosphatase (Roche) enabled the visual detection of positive bands using NBT and BCIP. In the right panel the blot was developed using the same principle, but initial probing was performed using a mix of biotinylated lectins: ConA (Glc, Man), GNA (Man), HHA (Man), WGA (GlcNAc), DSL (GlcNAc), UDA (GlcNAc), Nictaba (GlcNAc), RSA (Gal, GalNAc) and PNA (Gal, GalNAc). In both Western blots and for both strains a peculiarly strong band appeared at approximately 125 kDa. (B) Using a streptavidin conjugate to sample the proteome for false positive hits – The blotted exoproteome of L. rhamnosus GG and the ΔdltD::TcR mutant were probed directly with a streptavidin conjugate. This resulted in the appearance of several bands, among which the strong 125 kDa band. Based on this result, we suggest that the band is not caused by a glycoprotein, but is a false positive. (C) PAS glycostain does not react with the 125 kDa protein – An SDS-PAGE gel of the proteome of L. rhamnosus GG was poststained with Periodic Acid Schiff base stain (PAS, Pro-Q® Emerald 488 stain, Life Technologies), a method to specifically stain glycosylated proteins in a gel. At 125 kDa, no band could be detected, which further supports our hypothesis that the 125 kDa signal perceived on the lectin blots is the result of a false positive hit.

4.2.2 purification of tHe 125 kDa protein

We set out to identify the protein eliciting this strong signal on all blots. Based on the positive signal of the protein with ConA, a ConA affinity chromatography approach (HiTrap ConA 4B column, GE Healthcare) was initiated. However, despite of numerous efforts, we were unable to purify the protein in this set-up. Finally, a combination of cation (SP Sepharose HP, GE Healthcare) and anion exchange chromatography (Q Sepharose HP, GE Healthcare) resulted in the successful purification of the protein. Identification by Edman degradation on a Procise 491 cLC protein sequencer revealed that the LGG_01329, a pyruvate carboxylase is responsible for the positive bands on the lectin blots. The theoretical molecular weight

four

.

103

of 125028 Da for this protein corresponds to the apparent molecular weight observed on SDS-PAGE (results not shown) and on the lectin blots (Fig. 4.1A). The molecular weight of glycosylated proteins typically differs from the mass calculated from the deduced amino acid sequence of the protein, which forms a first indication that the protein encoded by LGG_01329 is not glycosylated. This was confirmed with a gel stained with Periodic Acid Shiff base stain (PAS, Pro-Q® Emerald 488 stain, Life Technologies) a glycoprotein stain, on which the 125 kDa band is absent (Fig. 4.1B). Taken together with the fact that carboxylases bind endogenous biotin as a cofactor, we suggest that the LGG_01329 protein is a false positive hit resulting from the reaction of the endogenous biotin with the streptavidin detection method, even when the extracellular proteome is investigated. This also would explain why the 125 kDa protein could not be purified using ConA affinity chromatography, despite of its strong positive reaction with biotinylated ConA on Western blot.

4.2.3 confirmation tHat faLse positives are causeD by enDogenous biotin

Figure 4.2 - The digoxigenin – anti-digoxigenin detection as an alternative to avoid false positive hits caused by proteins binding endogenous biotin.

(A) DIG-labeled lectin blots – The wild type exoproteome of L. rhamnosus GG was Western blotted and developed using a mix of DIG-labeled lectins: ConA (Glc, Man), GNA (Man), HHA (Man), WGA (GlcNAc), DSL (GlcNAc), UDA (GlcNAc), Nictaba (GlcNAc), RSA (Gal, GalNAc) and PNA (Gal, GalNAc). These lectins were labeled using digoxigenin-3-O-methyl-ε-aminocaproix acid-N-hydroxysuccinimide ester (Roche). Anti-DIG Fab antibody fragments (Roche) were used to detect proteins that reacted positively with the lectin probes. Here we clearly see that the false positive band at 125 kDa is absent. (B) Negative control with anti-DIG – Direct application of the anti-DIG Fab antibody fragments (Roche) to the Western blotted proteome of L. rhamnosus GG results in a blot on which no bands can be perceived. This confirms that the DIG – anti-DIG detection method is a good alternative for the biotin-streptavidin system, without causing false positive hits.

104

endo

geno

us b

iotin

-bin

ding

pro

tein

s

To further confirm our hypothesis that the false positive 125 kDa band resulted from the binding of endogenous biotin to the pyruvate carboxylase encoded by LGG_01329, we performed a Western blot on the extracellular proteome fraction in which we probed the blot with streptavidin alone. This means that we did not apply any primary biotinylated protein (in this case lectins), but merely developed the blot after incubation with streptavidin conjugate (Fig. 4.1C). These results are depicted in figure 4.1C, in which a Western blot of the exoproteome of wild type L. rhamnosus GG and the ΔdltD::TcR mutant is shown, which was probed with streptavidin. One of the strongest signals on this blot could be found at 125 kDa (Fig. 4.1C) and represents the pyruvate carboxylase LGG_01329.

The streptavidin blot also reveals several other bands, which represent other false positives. Indeed, several genes turn up when screening the genome of L. rhamnosus GG for proteins binding biotin or related to biotin metabolism (Gene database of NCBI). Examples are the AccC (LGG_02112) and AccB (LGG_02114) proteins encoding an Acetyl-CoA carboxylase biotin carboxylase and its biotin carboxyl carrier protein subunit, respectively.

These findings implicate that the interpretation of ‘screening’ Western blots developed with biotinylated probes should be carried out with caution. We therefore suggest to always include a control Western blot only probed with streptavidin conjugate when using biotinylated probes (Fig. 4.1B). Complementary screening of the genome of the species under study can also reveal the presence of biotin-binding enzymes.

To circumvent the occurrence of false positive bands completely, we suggest the digoxigenin – anti-digoxigenin (DIG – anti-DIG) detection method as a good alternative. As digoxigenin is a steroid only produced by Digitalis plants, interference of endogenous material in other species is not an issue (Chevalier et al., 1997) (cf. Fig. 4.2B, for a confirmation in L. rhamnosus GG). The DIG label is recognized by an anti-DIG antibody, which binds with a high specificity to the small steroid molecule (only 390 Da). In practical set-ups, only the Fab fragments of the anti-DIG antibody are used to avoid (albeit rare) cross reaction with structurally related steroids (Kessler, 1991). An N-hydroxysuccinimide ester derivative with an 6-aminocaproate spacer is commercially available to label probes with DIG. Results for the L. rhamnosus GG wild type exoproteome probed with a DIG-labeled lectin mix show less bands and most importantly, at 125 kDa no band is visible, which leads to the presumption that false positives hits are lacking (Fig. 4.2A). This hypothesis is confirmed by a Western blot developed using only the anti-DIG antibody (Fig. 4.2B).

Our findings are corroborated by earlier reports on false positive results caused by endogenous biotin in eukaryotes (Chen et al., 2012; Horling et al., 2012; McKay et al., 2008) and in the gene probing of Streptococcus mutans (Wang et al., 1993). Interestingly, recent work by Lee et al. on the glycoproteome of L. plantarum WCFS-1 reported a recurring band of approx. 125 kDa on Western blots developed with the biotinylated WGA and the biotinylated lectins of Dolichos biflorus and Lens culinaris (Lee et al., 2014). Based on our results, we suggest that

four

.

105

this band could correspond to the pyruvate carboxylase Lp_2136 of 127 kDa.

4.3 concLusion

In this work we illustrated the occurrence of false positive results on Western blots developed with biotinylated probes. Biotin-binding proteins, such as carboxylases, which bind endogenous biotin as a cofactor, cause these false positives. These proteins also occur in extracellular fractions, possibly as moonlighting proteins. Based on the here-presented results we suggest that when using biotinylated probes, a control experiment only using a streptavidin probe should be included (Fig. 4.1C). A good alternative labeling method is DIG labeling and detection with anti-DIG Fab fragments.

�ve.systematic exploration of

the glycoproteome of a bene�cal gut isolate

2D gel electrophoresis vs. a PhD student: 34-8(but who is keeping score anyway)

�ve.systematic exploration of

the glycoproteome of a bene�cal gut isolate

2D gel electrophoresis vs. a PhD student: 34-8(but who is keeping score anyway)

108

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

This chapter was submitted for publication:

Tytgat H.L.P., Schoofs G., Vanderleyden J., Van Damme E.J.M., Wattiez R., Lebeer S., Leroy B.

autHors’ contributions

The study was conceived by HT, BL and SL. Experimental procedures were performed by HT. Lectin experiments were done in collaboration with EV. HT drafted the manuscript.

five.

109

abstract

The human body hosts a complex bacterial community that plays a major role in human health. Recently, massive parallel sequencing studies have resulted in major breakthroughs in the understanding of the diversity of this microbiota, but these compositional studies now need to be complemented with functional omic approaches at DNA, RNA and protein level. Glycoproteins form an interesting class of macromolecules involved in bacteria-host interactions, but they are not yet widely explored in Gram-positive and beneficial species.

Here, an integrated and widely applicable approach was followed to identify putative bacterial glycoproteins, combining proteome fractionation with 2D protein- and glycostained gels and lectin blots. This approach was validated for the model beneficial isolate of the microbiota Lactobacillus rhamnosus GG. The approach resulted in a list of putative glycosylated proteins receiving a ‘glycosylation score’. Ultimately, we could identify 41 unique glycosylated proteins in L. rhamnosus GG (6 top confidence, 10 high confidence and 25 putative hits, classification based on our glycosylation score). Most glycoproteins are associated with the cell wall and membrane but glycoproteins were also detected in the cytosol, a cellular location in which not many bacterial glycoproteins have been reported yet. Identified glycoproteins include proteins involved in transport, translation and sugar metabolism processes. Among them, we identified glyceraldehyde 3-phosphate dehydrogenase, GapA (LGG_00933), of which the glycosylation status was experimentally validated by a dedicated glycostaining.

A robust screening resulted in a comprehensive mapping of glycoproteins in L. rhamnosus GG. Fractionation of the proteome enabled us moreover to predict the cellular locations of the glycoproteins. Our results reflect the glycosylation of sugar metabolism enzymes, transporters and other proteins crucial for cell physiology. We hypothesize that protein glycosylation can confer an extra level of regulation, for e.g. by the affecting enzyme functions. This is the first systematic study of the glycoproteome of a probiotic and beneficial gut isolate.

110

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

5.1 introDuction

It is the era of the microbiome: a plethora of big data projects is mapping the bacterial load of eukaryotic species. The human intestine hosts a complex bacterial community important for human health and nutrition. Large sequencing studies have led to a better understanding of this bacterial community and have shown links between several dysbiosis states and susceptibility to diseases. Therefore, the exploration of ways to beneficially modulate this bacterial load to influence host health holds great promise for the treatment of dysbiosis-related diseases. An interesting strategy is the use of intrinsic members of the microbiota as probiotics. Probiotics are ‘live microorganisms which, when administered in adequate amounts, confer a health benefit on the host’ (FAO/WHO, 2001). Well-known probiotics are Lactobacillus sp., Bifidobacterium sp., beneficial strains of Escherichia coli and even Saccharomyces boulardii, although this yeast species does not seem to be an intrinsic member of the microbiota. To allow a targeted application of the best probiotic strains for specific dysbiosis conditions, in-depth knowledge of the nature of the (beneficial) bacteria-host interactions and the key molecules modulating these interactions is crucial.

From the bacterial-side, the molecules in the secreted protein fraction and on the cell wall are interesting candidates to investigate as these molecules can directly mediate contacts with host cells. In this context, glycosylated proteins are intriguing molecules. Research on mainly pathogenic glycoproteins has uncovered an enormous diversity of glycans present on bacterial proteins. These unique glycans might form, together with other surface saccharides, a species-specific barcode on bacterial cell surfaces and are ideal candidates to establish specific interactions with the environment (Tytgat & Lebeer, 2014). Glycoproteins can for instance interact with specific immune lectin receptors as was shown for Campylobacter jejuni glycoproteins that interact with the Macrophage Galactose Lectin receptor (van Sorge et al., 2009). On the other hand, glycans on bacterial surfaces can also be used to shield immunogenic surface factors from detection by the host immune system. Moreover, the attachment of glycans to proteins can modulate the biochemical properties of proteins, enhancing their activity, specificity and stability (Tytgat & Lebeer, 2014). Taken together, this makes glycoproteins potentially important microbiota-host interactions factors. Indeed, in the major Gram-negative commensal Bacteroides fragilis, the general protein O-glycosylation system was shown to be essential for the competitive colonization of the mammalian intestine (Coyne et al., 2013; Fletcher et al., 2011; Fletcher et al., 2009), although specific interactions have not been mapped yet. Also in several Gram-positive Lactobacillus strains, glycoproteins have been fragmentarily reported (Anzengruber et al., 2013; Fredriksen et al., 2012; Fredriksen et al., 2013), but -to the best of our knowledge- the only Gram-positive beneficial isolate of the gut microbiota that has been reported to produce at least one glycosylated protein is L. rhamnosus GG (Lebeer et al., 2012b). To facilitate more systematic analyses of bacterial glycoproteomes, we report here on an extended and widely applicable analysis of the glycoproteome of this model microbiota isolate and documented probiotic strain (Doron

five.

111

et al., 2005). This chapter complements our earlier work on the systematic screening of genes encoding glycosyltransferases in L. rhamnosus GG (Sanchez-Rodriguez et al., 2014).

In order to map the glycoproteome of the Gram-positive L. rhamnosus GG, the proteome was separated in four functional fractions: the secreted, cell wall associated, cell wall/membrane and cytosolic proteins. This protocol was designed as such to generate information on the localization of glycosylated proteins but also to improve detection of membrane and cell wall associated proteins among which most of the glycoproteins are expected. The subproteome fractions were subsequently analyzed using 2D electrophoresis and their glycosylation status was assessed using a combination of Periodic Acid Schiff glycostain (PAS) and lectin blotting with the Galanthus nivalis (GNA) lectin specific for mannose. After identification with mass spectrometry, the hits were evaluated using a ‘glycosylation score’ to assess their probability of being a genuine glycoprotein. Ultimately, this rendered a list of 41 putative glycoproteins. 6 glycoproteins were identified with top confidence, 10 with high confidence and 25 as putatively glycosylated. The list was further challenged and compared to earlier results and to known glycoproteins in other species. This resulted in the further confirmation of the efficacy of our workflow. To confirm the overall robustness of the workflow, glyceraldehyde 3-phosphate dehydrogenase (GapA, GAPDH), encoded by LGG_00933, was isolated from the four fractions and its glycosylation status was experimentally confirmed in all fractions by independent PAS staining experiments.

5.2 materiaL anD metHoDs

5.2.1 bacteriaL strains anD cuLture conDitions

L. rhamnosus GG was grown at 37°C in Lactobacilli AOAC medium (Difco) in non-shaking conditions.

5.2.2 gLycoproteome isoLation anD fractionation

L. rhamnosus GG was grown for 24 h in AOAC medium. The supernatant of the culture after centrifugation at 6000 g during 20 min resulted in the secreted proteins (i.e. fraction 1, SP). This fraction was filtered to remove any remaining bacteria (0.45 μm) followed by precipitation with trichloroacetic acid (TCA, 20% final concentration). The precipitated proteins were washed twice with cold acetone (100%). The protein pellet was air-dried and resuspended in lysis buffer (2 M thiourea, 7 M urea, 4% CHAPS, 2% DTT).

The pellet of the overnight culture, containing L. rhamnosus GG, was washed twice with PBS. The cells were then incubated in a solution of 1.5 M LiCl in 10 mM Tris-HCl (pH 8) for 1 h at 4°C. Centrifugation (6000 g, 15 min) resulted in a second fraction, i.e. the cell wall-associated proteins (CWA), which was precipitated with TCA (Bauerl et al., 2010). The remaining pellet was dissolved in PBS and sonicated at an amplitude of 20% (2x2’) to ensure cell lysis (Branson sonifier). The lysate was cleared by centrifugation (several runs at 4000 g

112

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

to remove cell debris). Cytosolic proteins (fraction 3, Cyt) could be found in the supernatant after centrifugation (22 000 g, 20 min). The pellet was washed three times with cold Tris-HCl 50 mM, pH 8, 500 mM NaCl. The resulting pellet was dissolved in PBS and contained the cell wall/membrane proteins (CWCM, 4th fraction). Similarly to the other fractions, both the Cyt and CWCM fractions were precipitated and dissolved in lysis buffer.

5.2.3 2D geL eLectropHoresis

2D gel electrophoresis is a method to separate proteins in two dimensions according to their intrinsic isoelectric point (i.e. isoelectric focusing) and their molecular weight (SDS-PAGE). 2D gel electrophoresis was performed as described earlier by Sonck et al. (Sonck et al., 2009).

Briefly, Immobiline DryStrips (24 cm) with a 3-11 pH range (GE Healthcare) were rehydrated overnight in reswelling buffer (6 M urea, 2 M thiourea, 2% CHAPS, 0.4% DTT, 0.002% Bromophenol blue) containing 0.5% 3-11 IPG buffer (GE Healthcare). Prior to anodic cup sample loading on the DryStrips, 0.8% 3-11 IPG buffer (GE Healthcare) was added to the samples dissolved in lysis buffer (cf. supra). Isoelectric focusing of the samples on the Immobiline DryStrips was performed on a Multiphor II instrument (GE Healthcare).

Prior to SDS-PAGE, the focused proteins were reduced using equilibration buffer (6 M urea, 30% glycerol, 2% SDS, 0.002% bromophenol blue in Tris-HCl 1.5 M, pH 8.8, 1% DTT. The same procedure was repeated with equilibration buffer containing 4.5 % IAA. Second dimension separation (SDS-PAGE) was performed by application of the DryStrips with the isofocused samples on fresh 15% polyacrylamide gels (Amresco®), using the EttanDALTsix instrument (GE Healthcare).

5.2.4 Detection of gLycosyLateD proteins

Glycosylated proteins were detected by PAS staining (Fairbanks et al., 1971). The 2D gels were fixed and stained with the Pro-Q® Emerald 488 Glycoprotein Gel Stain Kit (Molecular Probes®), a fluorescent and sensitive PAS stain (detection limit of 4 ng of glycoprotein) (Hart et al., 2003). After analysis of the PAS stained gels, the same gels were submerged overnight in Sypro® Ruby protein gel stain (Molecular Probes®). Scanning of the stained gels was performed on a Typhoon 9400 laser scanner (GE Healthcare).

5.2.5 image anaLysis anD spot picking

Gel images were analyzed using ImageMaster software (GE Healthcare). The positive hits with PAS stain were matched to the Sypro® stained spots. The relative intensity and volume percentage on glyco- and protein stained gels were compared for each spot. More in particular, the intensity of the spots was recalculated relatively to the spot with the highest intensity, both for the PAS and Sypro® stained gels. The same was done for the volume percentage. Then the ratio was calculated of the relative intensity values (value Pro-Q® Emerald 488/value Sypro®)

five.

113

and of the relative volume percentages. Spots that had a ratio higher than 1 were selected for further identification. In order not to miss any highly glycosylated spots, all hits with a relative intensity higher than 20% on the PAS stained gels were also picked for further identification. Spot picking was executed automatically with the Ettan SpotPicker (GE Healthcare).

5.2.6 2D Lectin bLot

2D gels of the four fractions of the proteome of L. rhamnosus GG were transferred to PVDF membranes (Millipore) using a TE77 Semi-Dry transfer unit (GE Healthcare). These Western blots were probed with digoxigenin-labeled GNA, which specifically binds mannose residues (Tytgat et al., 2014). To reduce background, polyvinylalcohol was used as a blocking agent (Thompson et al., 2011). Spots reacting with the lectin were matched to the Sypro® stained gels and picked for identification.

5.2.7 in-geL tryptic Digestion anD ms

Reactive spots were submitted to in-gel trypsin digestion. Spots were first washed twice in 50 mM NH4HCO3 for 15 min followed by dehydration during 25 min in 50% (v/v) acetonitrile in 50 mM NH4HCO3. 10 µL of 50mM NH4HCO3 containing 0.05 µg of trypsin was then added to the gel pieces and incubated on ice for 20 min. Trypsinolysis was carried on overnight at 37°C and stopped by addition of 2 µL of 1% (v/v) formic acid. Tryptic peptides were analyzed using an LC-MS/MS workflow as already described (Mastroleo et al., 2009). Briefly, peptides were separated by Reverse Phase chromatography using a 40 min ACN gradient (4-45%) and analyzed on an HCTultra Ion Trap (Bruker) in MS/MS mode. MS spectra were acquired across the 300-1500 m/z mass range and the four ions with the highest intensity were selected for MS/MS and excluded after one spectrum for 30 sec. Acquired data were analyzed with the Mascot search engine against the NCBI database (NCBInr_20130409) restricted to L. rhamnosus GG.

5.2.8 Data anaLysis anD gLycosyLation score

Proteins were first filtered through the identification p value: only proteins with p <0.05 were selected. Peptides with a Mascot ion score lower than 30 were also discarded. Finally only proteins identified with at least 2 unique peptides were further analyzed. A glycosylation score was attributed to each identified protein as follows: 1 point was given to a protein identified as a top hit in a spot with no other putative glycoproteins present (a putative glycoprotein is a protein with a glycosylation score > 2, top hit protein is defined as the protein with the highest Mascot score for a spot); 2 points were given each time a protein was identified as the sole protein in a reactive spot; 3 points were attributed if the protein has been detected as a putative glycoprotein in the repetition of the experiment.

5.2.9 purification of gapDH (Lgg_00933)

Protein fractions of L. rhamnosus GG were prepared as described above and GAPDH was

114

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

purified as described by Saad and co-workers (Saad et al., 2009). All fractions, except for the secreted proteins, were dissolved in NH4HCO3 25 mM pH 7.5, 5 mM EDTA, 1 mM PMSF (Buffer A) after TCA precipitation. The secreted protein fraction was concentrated and resuspended in buffer A using a tangential flow filtration system (KrosFlo, Spectrum Labs). All samples were dialyzed overnight (4°C) against buffer A. The dialyzed fractions were applied to a 5 mL Fast Flow Blue Sepharose CL 6B column (GE Healthcare) pre-equilibrated with buffer A at a flow rate of 2 mL/min. Elution was performed using a linear gradient of 0 to 10 mM nicotine amide adenine dinucleotide (NAD) in buffer A, pH 8.5 (buffer B) (Saad et al., 2009).

5.2.10 assessment of tHe gLycosyLation state of gapDH (Lgg_00933)

Glycosylation of GAPDH in the different proteome fractions of L. rhamnosus GG was tested using the Pro-Q® Emerald 488 Glycoprotein Gel Stain Kit (Molecular Probes®) on 1D gels, as described earlier (Lebeer et al., 2012b). Gels were imaged using a Typhoon 9400 laser scanner (GE Healthcare). Gels were post-stained with Sypro® to visualize the proteins. Confirmation that the bands corresponded to GAPDH was obtained by Western blots developed with GAPDH yeast antibody (Nordic Immunology). The high level of conservation among GAPDH proteins justified usage of this antibody in general and binding of this antibody to purified GAPDH of L. plantarum CMPG5300 ((Malik et al., 2014) & personal communication S. Malik) further confirmed this.

5.3 resuLts

5.3.1 Design of tHe workfLow

Our workflow to map the glycoproteome of a bacterium is depicted in figure 1. In short, the fractionated proteome was screened for glycosylation using a combination of 2D PAS gels and lectin probed 2D blots (Fig. 5.1).

As the proteome of bacteria is quite large (e.g. over 3000 proteins in L. rhamnosus GG (Kankainen et al., 2009)), 2D electrophoretic images can become (over)crowded with spots. In order to end up with interpretable gels and to generate clues on the cellular localization of glycosylated proteins, we designed a protocol to fractionate the proteome of L. rhamnosus GG in four subcellular fractions: cytosol (Cyt), cell wall/membrane (CWCM), cell wall-associated (CWA) and secreted protein (SP) fraction. After 2D electrophoresis, the gels were stained using PAS. This glycostain oxidizes the alcohol groups of the sugars, forming aldehydes that react with the Schiff base resulting in a color reaction (Fairbanks et al., 1971). The commercial version of this stain, the Pro-Q® Emerald 488 (Molecular Probes®) was applied, as this stain has a higher sensitivity (Hart et al., 2003). The same gels can then be post stained with non-specific Sypro® Ruby protein stain to visualize all proteins. An example is represented for the cytosol in figure 5.2.

five.

115

Figure 5.1 – Overview of the workflowThe here-presented workflow relies on the combination of proteome fractionation and 2D screening for potential glycoproteins. The proteome was fractionated in such a way that it was subdivided in four fractions. Centrifugation of an overnight culture of L. rhamnosus GG separated the secreted proteins (SP) from the cells, which were subjected to LiCl treatment resulting in the cell wall-associated proteins (CWA). The final sonication and ultracentrifugation steps result in the proteins present in the cell wall/membrane (CWCM) and the cytosol (Cyt). 2D gel electrophoresis resulted in a distinct 2D pattern for each fraction (cf. Sypro® stained gel pictures). The 2D separated proteomes were subjected to both PAS staining and lectin probing with GNA, which specifically binds mannose residues. The combination of positive hits from both assays was identified via LC-MS/MS, which resulted in a list of candidate glycoproteins. This list was further refined by scoring all hits with a ‘glycosylation scoring’ which rendered a list of 41 glycosylated proteins in L. rhamnosus GG. The glycosylation scoring system enabled the classification of the glycosylated proteins in top confidence hits (6), high confidence ones (11) and putative glycosylated proteins (25). Additionally, the initial fractionation of the proteome resulted in clues on the localization of these glycoproteins. The full list of identified glycoproteins and all additional generated data can be consulted in table 5.1.

116

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

Figure 5.2 - Glycoproteins are ubiquitously present in the cytosolic fraction of the L. rhamnosus GG proteome2D electrophoresis of the cytosolic fraction of the proteome of L. rhamnosus GG is shown. On these gels the proteins are first separated according to their isoelectric point in the first dimension, followed by molecular weight separation in the second dimension. The left panel shows the gel stained with the total protein Sypro® and thus depicts the cytosolic proteome. The same gel was screened for glycoproteins using the Pro-Q® Emerald 488 staining kit (PAS staining) (right panel). These gels illustrate the presence of glycoproteins in the cytosolic fraction of the proteome of L. rhamnosus GG. Similar gels were obtained for the other functional proteome fractions, i.e. secreted, cell wall/membrane and cell wall-associated proteins (available on request).

The robustness of the PAS screening was increased by the design of a set of rules to select PAS reactive spots for further identification by mass spectrometry. Abundantly present proteins might, for instance, cause an overinterpretation of the PAS signal of the spot. This problem was bypassed by calculating the ratio of the relative intensity and volume percentage of the glycostained versus the protein stained spot. These values are generated by the analysis software, and represent the intensity and distribution of this intensity relative to the area of the spot respectively. Only spots in which the ratio of the relative intensity or volume percentage was higher than 1, i.e. a significantly higher Pro-Q® Emerald 488 than Sypro® signal, were picked for further analysis. This list was further extended by addition of PAS stained spots with a relative intensity higher than 20%, to ensure not to miss heavily glycostained spots. This last step largely corroborated the earlier selected spots, but ensured the inclusion of spots that were heavily stained with both stains.

The same subproteome fractions were also subjected to lectin probing to complement the PAS screening. Lectins are known to bind glycans in a more selective manner, i.e. recognizing specific sugars in specific configurations and linkages, while the PAS staining method detects all glycoconjugates (Van Damme et al., 2011). The mannose specific lectin GNA was applied to the Western-blotted subproteome fractions. Mannose is a typical sugar incorporated in surface glycans by L. rhamnosus GG, i.e. the Msp1 glycoprotein and exopolysaccharides (Lebeer et al., 2012b; Lebeer et al., 2009). The lectin GNA was chosen over Concanavalin A (glucose and mannose specific), as GNA resulted in a lower background signal (results not shown). Positive spots were picked after matching them to Sypro® stained gels. Overall the GNA-reactive spots accorded well with the results of the PAS staining (results not shown).

five.

117

5.3.2 functionaL proteome fractionation

Application of the workflow to L. rhamnosus GG cells resulted in 2D gels with distinct patterns for the four fractions, which illustrates the efficacy of the fractionation process (results shown for the cytosol in Fig. 5.2.). The numerous spots visualized on the PAS stained gel illustrate the presence of glycoproteins in this cellular fraction. In total, eight glycoproteins were found exclusively in the cytosol (Table 5.1 & Fig. 5.2.). Moreover, the occurrence of the specific glycoproteins in each fraction was reproducible. In table 5.1, an overview is given of all identified glycoproteins and a dot (l) indicates the occurrence of a glycoprotein in the same fraction in each repetition of the experiment. The actual robustness of the proteome fractionation is probably even higher, as only hits with a glycosylation score higher than 2 were now included in the table.

A nice illustration of the robustness of the fractionation process is the fact that all five transport-related glycoproteins (Table 5.2) were identified in the cell wall/membrane fractions (Table 5.1). Another example is the three closely related glycoproteins Msp1 (LGG_00324), Msp2 (LGG_00031) and LGG_02016, being cell wall hydrolases. The function of Msp1 and Msp2 as cell wall hydrolases has previously been validated experimentally (Claes et al., 2012), in agreement with their main occurrence on the cell surface and secreted protein fractions of the proteome. Also consistent with their cellular function is the occurrence of glycosylated proteins involved in carbohydrate metabolism, transcription, translation and stress response in the intracellular fraction of the proteome (Cyt) (Table 5.1 and 5.2).

Quite peculiar though is the presence of Eno (LGG_00936), GapA (LGG_00933), Pgk (LGG_00934) and Tig (LGG_01351) in the extracellular fractions (CWCM, CWA and SP). Although the main functions of these proteins reside in the cytosol, their surface localization was already reported earlier (Deepika et al., 2012; Saad et al., 2009). This ‘aberrant’ localization suggests a moonlighting role for these proteins (Deepika et al., 2012; Izquierdo et al., 2009; Jeffery, 2003; Sanchez et al., 2009a), which has already been documented in other species for the glycolytic enzymes Eno, GapA and Pgk (Castaldo et al., 2009; Granato et al., 1999; Henderson, 2014; Kinoshita et al., 2008a; Kinoshita et al., 2008b; Sanchez et al., 2009a).

In total we were able to identify 15 cytosolic putative glycoproteins, 28 glycoproteins in the CWCM fraction, 17 cell wall-associated proteins and 5 secreted ones (Table 5.1). Nevertheless, although we illustrated that these fractions are quite robust and confirm earlier protein localizations, we are aware that some leakiness of fractions cannot be ruled out. Implementation of controls is hard, as proteins that were once thought to be typical for the cytosol, are now also discovered in other fractions and appear to be moonlighting proteins.

118

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

Tabl

e 5.

1 –

Ove

rvie

w o

f th

e gl

ycop

rote

ins

iden

tifi

ed in

L. r

ham

nosu

s G

G a

nd t

heir

cel

lula

r lo

caliz

atio

n

Loc

us ta

gA

cces

sion

num

ber

Ann

otat

ion

Mas

cot

scor

e

Prot

ein

MW

(Da)

#

pept

.

Frac

tion

Hig

h

glyc

.

scor

e

2nd

glyc

.

scor

e

Inde

p.

Obs

erv.

Ext

ra

evid

ence

Cyt

CW

CM

CW

ASP

Top

confi

denc

e gl

ycop

rote

ins

LGG

_003

24gi

|258

5073

19M

sp1

(p75

); ce

ll w

all-a

ssoc

iate

d gl

ycos

ide

hydr

olas

e (N

LP/P

60

prot

ein)

203

4976

64

ml

>10

5Ye

sO

ther

lact

obac

illi

LGG

_009

33gi

|258

5079

28G

apA

; gly

cera

ldeh

yde-

3-ph

osph

ate

dehy

drog

enas

e17

4536

929

32l

ll

l>1

0>1

0Ye

sTh

is w

ork

LGG

_009

34gi

|258

5079

29Pg

k; p

hosp

hogl

ycer

ate

kina

se26

5442

187

54m

mm

>10

Yes

LGG

_009

36gi

|258

5079

31En

o; p

hosp

hopy

ruva

te h

ydra

tase

1834

4709

852

ll

lm

>10

>10

Yes

LGG

_011

81gi

|258

5081

76A

tpA

; F0F

1 AT

P sy

ntha

se

subu

nit a

lpha

1470

5525

327

lm

>10

7Ye

s

LGG

_025

23gi

|258

5095

18Ld

h; L

-lact

ate

dehy

drog

enas

e69

335

481

15m

l>1

05

Yes

Hig

h co

nfide

nce

glyc

opro

tein

s

LGG

_000

31gi

|258

5070

26M

sp2

(p40

); su

rfac

e an

tigen

700

4263

115

mm

6Ye

sO

ther

lact

obac

illi

LGG

_000

78gi

|258

5070

73O

puC

a; g

lyci

ne/b

etai

ne/L

-pro

line

AB

C tr

ansp

orte

r ATP

-bin

ding

pr

otei

n

917

4643

513

ml

54

Yes

LGG

_013

75gi

|258

5083

70Py

k; p

yruv

ate

kina

se25

6762

866

56m

m9

Yes

LGG

_016

04gi

|258

5085

99D

naK

; mol

ecul

ar c

hape

rone

1342

6717

925

l6

5Ye

sL.

pl

anta

rum

W

CFS

1 H

.pyl

ori &

F. t

ular

ensi

s

LGG

_020

16gi

|258

5090

11Su

rfac

e an

tigen

NLP

/P60

1815

4094

739

mm

4O

ther

lact

obac

illi

LGG

_020

39gi

|258

5090

34R

mlC

; dTD

P-4-

dehy

dror

ham

nose

3,5

-epi

mer

ase

248

2280

74

mm

4Ye

s

LGG

_021

38gi

|258

5091

33G

pmA

; pho

spho

glyc

erom

utas

e39

925

938

5m

m4

LGG

_022

39gi

|258

5092

34G

roEL

; cha

pero

nin

Gro

EL31

057

357

6m

4F.

tula

rens

is &

H. p

ylor

i

LGG

_024

93gi

|258

5094

88Fu

sA; e

long

atio

n fa

ctor

G (E

F-G

)83

776

888

15l

m6

5Ye

sC

. jej

uni &

H. p

ylor

i

five.

119

LGG

_028

06gi

|258

5098

01H

trA; s

erin

e pr

otea

se12

2245

175

22m

ml

77

Yes

H. p

ylor

i

Puta

tive

glyc

opro

tein

s

LGG

_002

55gi

|258

5072

20Se

rA; 2

-hyr

oxya

cid

dehy

drog

enas

e11

033

861

2m

2

LGG

_002

57gi

|258

5072

52M

urE;

UD

P-N

-ac

etyl

mur

amoy

lana

lyl-D

-gl

utam

ate-

-2,6

-dia

min

opim

elat

e lig

ase

125

5676

32

m2

LGG

_004

18gi

|258

5074

13Ta

l; tra

nsal

dola

se59

425

533

7m

Yes

LGG

_006

34gi

|258

5076

29D

yp-ty

pe p

erox

idas

e fa

mily

pr

otei

n35

535

060

5m

2Ye

s

LGG

_009

51gi

|258

5079

46M

alE;

suga

r AB

C tr

ansp

orte

r su

bstra

te-b

idin

g pr

otei

n82

748

326

11m

2Ye

s

LGG

_013

51gi

|258

5083

46Ti

g; tr

igge

r fac

tor

874

4975

119

m2

LGG

_013

74gi

|258

5083

69Pf

kA; 6

-pho

spho

fruc

toki

nase

297

3417

66

m2

Yes

LGG

_014

74gi

|258

5084

69N

rdE;

ribo

nucl

eotid

e-di

phos

phat

e re

duct

ase

subu

nit

alph

a

143

8239

52

mm

22

Yes

LGG

_016

15gi

|258

5086

10N

usA

; tra

nscr

iptio

n el

onga

tion

fact

or

782

4454

616

mm

2Ye

s

LGG

_016

28gi

|258

5086

23R

psB

; 30S

ribo

som

al p

rote

in S

241

429

623

5m

2

LGG

_016

70gi

|258

5086

65Fm

t; m

ethi

onyl

-tRN

A

form

yltra

nsfe

rase

110

3422

62

m2

LGG

_017

17gi

|258

5087

12Ph

eT; p

heny

lala

nyl-t

RN

A

synt

heta

se su

buni

t bet

a34

288

016

11m

2H

. pyl

ori

LGG

_019

00gi

|258

5088

95H

ypot

hetic

al p

rote

in58

320

900

9m

2Ye

s

LGG

_019

97gi

|258

5089

92R

mlB

; dTD

P-gl

ucos

e 4,

6-de

hydr

atas

e 12

438

560

3m

2Ye

s

LGG

_020

09gi

|258

5090

04G

lnQ

; am

ino

acid

AB

C

trans

porte

r ATP

-bid

ing

prot

ein

307

2743

66

m2

Yes

LGG

_022

34gi

|258

5092

29M

utL;

DN

A m

ism

atch

repa

ir pr

otei

n39

871

723

9m

2Ye

s

120

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

LGG

_022

77gi

|258

5092

72R

plJ;

50S

ribo

som

al p

rote

in L

1038

318

237

5m

2Ye

s

LGG

_023

32gi

|258

5093

27G

ltX; g

luta

myl

-tRN

A sy

nthe

tase

245

5690

17

m2

LGG

_024

19gi

|258

5094

14M

tsA

; man

gana

se A

BC

tra

nspo

rter s

ubst

rate

-bin

ding

pr

otei

n

1172

3499

628

mm

2

LGG

_024

24gi

|258

5094

19Y

mjC

; NA

D-d

epen

dent

ep

imer

ase/

dehy

drat

ase

8123

701

2m

2

LGG

_024

72gi

|258

5094

67R

plF;

50S

ribo

som

al p

rote

in L

647

719

304

8m

2

LGG

_024

75gi

|258

5094

70R

plE;

50S

ribo

som

al p

rote

in L

514

620

198

2m

2

LGG

_024

98gi

|258

5094

93R

poB

; DN

A-d

irect

ed R

NA

po

lym

eras

e su

buni

t bet

a20

913

3938

6m

2

LGG

_025

75gi

|258

5095

70La

cD; t

agat

ose-

1,6-

diph

osph

ate

aldo

lase

1887

3633

437

mm

2Ye

s

LGG

_026

43gi

|258

5096

38R

ecA

; rec

ombi

nase

A68

642

775

13m

2H

. pyl

ori

For

each

gly

copr

otei

n th

e da

ta fo

r th

e hi

t w

ith t

he h

ighe

st c

onfid

ence

is s

how

n ou

t of

the

tw

o in

depe

nden

t re

petiti

ons

of t

he e

xper

imen

t. To

p co

nfide

nce

glyc

opro

tein

s ar

e att

ribut

ed a

gly

cosy

latio

n sc

ore

high

er th

an 1

0, h

igh

confi

denc

e hi

ts h

ave

scor

es ra

ngin

g be

twee

n 4

and

9. L

ower

gly

cosy

latio

n sc

ores

wer

e pu

t und

er th

e ‘p

utati

ve g

lyco

prot

ein’

de

nom

inat

or (>

2). L

ocus

tag

: gen

e id

entifi

er o

f the

gly

copr

otei

n. A

cces

sion

num

ber:

gi id

entifi

er o

f the

gly

copr

otei

n. A

nnot

ation

: ann

otati

on a

s in

cur

rent

gen

ome

rele

ase

of

GenB

ank

(NC_

0131

98.1

). M

asco

t sc

ore:

Hig

hest

mas

cot

scor

e ob

tain

ed fo

r th

e pr

otei

n. #

pep

tides

: Hig

hest

num

ber

of p

eptid

es id

entifi

ed. F

racti

on: A

dot

(l

) in

dica

tes

the

pres

ence

of t

he g

lyco

prot

ein

in b

oth

expe

rimen

ts in

this

frac

tion.

A c

ircle

(m) r

epre

sent

s th

e id

entifi

catio

n of

a g

lyco

prot

ein

in th

e fr

actio

n in

one

repe

tition

of t

he e

xper

imen

t. Hi

gh g

lyc.

sco

re: t

he h

ighe

st g

lyca

n sc

ore

that

cou

ld b

e att

ribut

ed to

a g

lyco

prot

ein

in o

ne o

f the

repe

tition

s of

the

expe

rimen

t. 2n

d gl

yc. s

core

: the

gly

cosy

latio

n sc

ore

of th

e gl

ycop

rote

in in

the

repe

tition

of t

he e

xper

imen

t. In

dep.

confi

rm.:Y

es w

as in

dica

ted

in th

is co

lum

n if

the

glyc

opro

tein

was

obs

erve

d at

leas

t onc

e in

a re

activ

e sp

ot, r

egar

dles

s of t

he

conc

omita

nt p

rese

nce

of o

ther

gly

copr

otei

ns in

this

spot

and

/or w

hen

the

prot

ein

was

iden

tified

in a

n in

depe

nden

t GN

A sc

reen

ing.

Ext

ra e

vide

nce:

ext

ra e

vide

nce

supp

ortin

g ou

r re

sults

, lik

e hi

ts a

nd tr

ends

pub

lishe

d in

oth

er sp

ecie

s.

five.

121

5.3.3 gLycoprotein iDentification

To circumvent the fact that most picked spots contain multiple proteins, a ‘glycosylation scoring’ system was implemented to identify genuinely glycosylated targets. Only proteins identified on the presence of at least two peptides were taken into account in our dataset. The ‘glycosylation score’ was conceived as a cumulative index: the confidence of the glycosylation status of a protein increases when a) the candidate glycoprotein is detected in both repetitions of the experiment (3 points), b) a protein is identified alone in a PAS-/GNA-reactive spot (2 points) and c) a protein is a top hit (highest mascot score) in a reactive spot in which no other glycoproteins are present (1 point). Based on this score, the putative glycoproteins were divided in three categories: top confidence hits (>10), high confidence glycoproteins (>4) and putative glycosylated proteins (>2). Proteins with a lower score were discarded from the dataset.

In total, our workflow (Fig. 1) resulted in the identification of one or more proteins in more than 230 PAS/lectin reactive spots through LC MS/MS (full datasets for each fraction are available as supplementary tables 1-8). The glycosylation score allowed filtering this dataset and selecting 41 glycosylated proteins in L. rhamnosus GG, which are presented in table 1. Among the six glycoproteins that were identified with top confidence, the earlier identified and experimentally validated Msp1 protein was found (Lebeer et al., 2012b). Moreover, 24 of the identified glycoproteins could be validated in independent experiments, i.e. confirmation of the glycosylation status of proteins in independently performed lectin blots and PAS screenings (results not shown, cf. ‘Independent observation’ column in table 1) or by other mass spectrometry based approaches.

We do acknowledge that our approach, i.e. the implementation of strict rules at several steps of the screening process, most probably neglects some glycoproteins, rendering some false negatives.

5.3.4 generaL trenDs in protein gLycosyLation

When looking at the list of identified glycoproteins, some major trends can be delineated. For instance, we already mentioned the high number of glycolytic enzymes that were found to be glycosylated (Table 5.1). The list of glycoproteins was mined for glycoproteins involved in similar processes (cf. KEGG database), which were grouped in functional classes. The result is depicted in figure 5.3 and table 5.2.

The identified glycoproteins can be organized in eight classes reflecting their functional role in L. rhamnosus GG. The largest set of glycoproteins (27%) comprises carbohydrate metabolism-related proteins (Fig. 3). The second largest subset of glycoproteins is involved in protein synthesis (22%), particularly in amino acid metabolism en translation processes. Other apparent trends are the glycosylation of transport- (12%) and stress-related (15%) proteins. Smaller classes contain cell division-related glycosylated proteins (10%), redox enzymes

122

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

(5%), members of the transcription machinery (5%) and an enzyme involved in nucleotide metabolism.

`

Figure 5.3 - General trends among glycoproteins identified in L. rhamnosus GGThis pie chart depicts the relative abundance of glycoproteins of each general trend that can be delineated in the glycoproteome of L. rhamnosus GG. For a more detailed view on the proteins present in these specific classes, the reader is referred to table 5.2.

5.3.4.1 carboHyDrate metaboLism

A striking 27% of the identified glycoproteins (11/41) can be linked to the metabolism of carbohydrates. Within this group, the largest portion of hits originates from glycosylated glycolytic enzymes (17%). Moreover, most of them were classified as high and even top confidence glycoproteins according to our glycosylation scoring. An additional set of glycosylated targets (10%) are involved in the general carbohydrate metabolism. This high number of glycosylated carbohydrate metabolizing enzymes might point towards a feedback loop in glycoprotein production and general sugar metabolism. Intuitively, the presence of such a level of ‘self’ regulation would make sense. Whether this hypothesis is indeed correct and if this trend can be validated in other species too, remains to be investigated.

Intriguingly, for some glycolytic enzymes (GapA, Eno and Pgk) an extra moonlighting function has been reported previously. Next to their glycolytic activity these enzymes also play role in binding to the extracellular environment (Castaldo et al., 2009; Granato et al., 1999; Henderson, 2014; Kinoshita et al., 2008a; Kinoshita et al., 2008b; Sanchez et al., 2009a). These proteins have potentially also a moonlighting role in L. rhamnosus GG as many observations were made in closely related lactobacilli (Castaldo et al., 2009; Granato et al., 1999; Kinoshita et al., 2008a; Kinoshita et al., 2008b). This hypothesis is substantiated by the occurrence of these glycoproteins in the extracellular fractions of the proteome of L. rhamnosus GG in this (Table 5.1) and previous studies (Deepika et al., 2012; Saad et al., 2009; Sanchez et al., 2009a; Sanchez et al., 2009b).

five.

123

Table 5.2 – General trends in the glycoproteome of L. rhamnosus GGLocus tag Accession n° Annotation ConfidenceCarbohydrate: glycolysis (-related enzymes)LGG_00933 gi|258507928 GapA; glyceraldehyde-3-phosphate dehydrogenase TopLGG_00934 gi|258507929 Pgk; phospholycerate kinase TopLGG_00936 gi|258507931 Eno; phosphopyruvate hydratase TopLGG_01374 gi|258508369 PfkA; 6-phosphofructokinase PutativeLGG_01375 gi|258508370 Pyk; pyruvate kinase HighLGG_02138 gi|258509133 GpmA; phosphoglyceromutase HighLGG_02523 gi|258509518 Ldh; L-lactate dehydrogenase TopCarbohydrate metabolism: generalLGG_00418 gi|258507413 Tal; transaldolase PutativeLGG_01997 gi|258508992 RmlB; dTDP-glucose 4,6-dehydratase PutativeLGG_02039 gi|258509034 RmlC; dTDP-4-dehydrorhamnose 3,5-epimerase HighLGG_02575 gi|258509570 LacD; tagatose-1,6-diphosphate aldolase PutativeProtein synthesis: amino acid metabolism and translationLGG_00255 gi|258507220 SerA; 2-hyroxyacid dehydrogenase PutativeLGG_01628 gi|258508623 RpsB; 30S ribosomal protein S2 PutativeLGG_01670 gi|258508665 Fmt; methionyl-tRNA formyltransferase PutativeLGG_01717 gi|258508712 PheT; phenylalanyl-tRNA synthetase subunit beta PutativeLGG_02277 gi|258509272 RplJ; 50S ribosomal protein L10 PutativeLGG_02332 gi|258509327 GltX; glutamyl-tRNA synthetase PutativeLGG_02493 gi|258509488 FusA; elongation factor G (EF-G) HighLGG_02472 gi|258509467 RplF; 50S ribosomal protein L6 PutativeLGG_02475 gi|258509470 RplE; 50S ribosomal protein L5 PutativeStress & chaperonesLGG_01351 gi|258508346 Tig; trigger factor PutativeLGG_01604 gi|258508599 DnaK; molecular chaperone HighLGG_02234 gi|258509229 MutL; DNA mismatch repair protein PutativeLGG_02239 gi|258509234 GroEL; chaperonin GroEL HighLGG_02643 gi|258509638 RecA; recombinase A PutativeLGG_02806 gi|258509801 HtrA; serine protease HighTransportLGG_00078 gi|258507073 OpuCa; glycine/betaine/L-proline ABC transporter ATP-binding

proteinHigh

LGG_00951 gi|258507946 MalE; sugar ABC transporter substrate-biding protein PutativeLGG_01181 gi|258508176 AtpA; F0F1 ATP synthase subunit alpha TopLGG_02009 gi|258509004 GlnQ; amino acid ABC transporter ATP-biding protein PutativeLGG_02419 gi|258509414 MtsA; manganase ABC transporter substrate-binding protein PutativeCell division related proteinsLGG_00031 gi|258507026 Msp2 (p40); surface antigen HighLGG_00257 gi|258507252 MurE; UDP-N-acetylmuramoylanalyl-D-glutamate--2,6-

diaminopimelate ligasePutative

LGG_00324 gi|258507319 Msp1 (p75); cell wall-associated glycoside hydrolase (NLP/P60) TopLGG_02016 gi|258509011 Surface antigen NLP/P60 HighRedoxLGG_00634 gi|258507629 Dyp-type peroxidase family protein PutativeLGG_02424 gi|258509419 YmjC; NAD-dependent epimerase/dehydratase PutativeTranscriptionLGG_01615 gi|258508610 NusA; transcription elongation factor PutativeLGG_02498 gi|258509493 RpoB; DNA-directed RNA polymerase subunit beta PutativeNucleotide metabolismLGG_01474 gi|258508469 NrdE; ribonucleotide-diphosphate reductase subunit alpha PutativeOther LGG_01900 gi|258508895 Hypothetical protein Putative

Locus tag: gene identifier of the glycoprotein. Accession number: gi identifier of the glycoprotein. Annotation: annotation as in current genome release of GenBank (NC_013198.1). Confidence: level of certainty with which the glycoprotein was identified (cf. Table 5.1).

124

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

5.3.4.2 protein syntHesis: amino aciD metaboLism anD transLation

Proteins involved in protein metabolism, more in particular in amino acid metabolism and translation processes, are largely represented among the glycosylated proteins identified (22%, Fig. 5.3). We could identify four glycosylated ribosomal proteins, as well as enzymes involved in the regulation of methionine, phenylalanine and glutamate metabolism (putative glycoproteins). The last protein in this class is the elongation factor G or FusA protein. Glycosylation of this elongation factor was already reported earlier in the model organism for bacterial glycosylation C. jejuni (Young et al., 2002) and in Helicobater pylori (Champasa et al., 2013). It is intriguing that members of the largest bacterial multienzyme complex, the ribosome, are glycosylated, since glycosylation of members of protein complexes has previously been observed or suggested for several other complexes (Tytgat & Lebeer, 2014).

5.3.4.3 stress-reLateD proteins anD cHaperones

Glycosylation of chaperones was already reported in several bacteria: L. plantarum WCFS1 (Fredriksen et al., 2013), B. fragilis (Fletcher et al., 2009), H. pylori (Champasa et al., 2013) and Francisella tularensis (Balonova et al., 2010). Our analysis resulted in the identification of no less than six glycosylated proteins that have a chaperone or stress response-related activity (Table 5.2). Glycosylation of the DnaK chaperone confirms the earlier reports in L. plantarum WCFS1 (Fredriksen et al., 2013), H. pylori (Champasa et al., 2013) and F. tularensis (Balonova et al., 2010). Another chaperonin, GroEL, was also found to be glycosylated in F. tularensis (Balonova et al., 2010) and H. pylori (Champasa et al., 2013) . Taken together, this might point towards a specific role or importance of the glycosylation of chaperones. It is conceivable that glycosylation of these proteins enhances their stability. Nonetheless, further confirmation of this emerging trend is needed with functional data in different bacteria.

RecA, encoding a recombinase important for DNA repair processes, was also found to be glycosylated in L. rhamnosus GG. The protein was isolated from the CWCM fraction and in view of this function, it can be envisaged that it is associated to the cytosolic fraction of the membrane. Our result corroborates a report from 2004 on the glycosylation of the RecA recombinase of H. pylori, which was one of the first reports on the occurrence of intracellular glycoproteins (Fischer & Haas, 2004). We nevertheless want to stress that leakiness of the fractions is a factor that needs to be taken into account and further validation is required.

5.3.4.4 transporters

12% (5/41) of the retrieved glycoproteins are transporters or transport-related proteins (Fig. 5.3 and table 5.2). Glycosylation of ABC transporters and their related proteins seems thus to be a general trend among our results. Two of the five glycosylated transporters are moreover linked to the transport of amino acids. This is in agreement with the reports on the glycosylation of membrane-associated transporters in C. jejuni (Scott et al., 2011) and ABC transporters in N. gonorrhoea (Vik et al., 2009) and H. pylori (Champasa et al., 2013). Moreover, some

five.

125

of the glycosylated transporters reported in these species are also linked to the amino acid metabolism of the cells.

5.3.4.5 ceLL Division-reLateD proteins

A common trend in lactobacilli, which are known to glycosylate proteins, is the glycosylation of cell wall hydrolases like the Msp1 glycoprotein of L. rhamnosus GG (Lebeer et al., 2012b) and related autolysins and cell wall hydrolases in L. plantarum WCFS1 (Fredriksen et al., 2012; Rolain et al., 2013), L. lactis (Huard et al., 2003) and L. buchneri (Anzengruber et al., 2013). In this work, the glycosylation of the cell wall hydrolase and probiotic effector Msp1 was further independently confirmed. Furthermore we report on the glycosylation of the closely related probiotic effector Msp2 surface antigen (p40) and LGG_02016, both identified with high confidence. Our group proved earlier that the rather extensive glycosylation of Msp1 improves its stability, but might hamper its activity compared to the closely related Msp2 protein (Lebeer et al., 2012b; Yan et al., 2007).

5.3.4.6 reDox-reLateD proteins anD otHer cLasses

Glycosylation of redox-related proteins has been proven already for a number of species, including C. jejuni (Ding et al., 2009; Scott et al., 2009; Young et al., 2002), N. gonorrhoea (Vik et al., 2009), H. pylori (Champasa et al., 2013) and F. tularensis (Balonova et al., 2010). Here we report on the glycosylation of two cytosolic redox-related proteins in L. rhamnosus GG (Table 5.2).

Smaller classes of glycosylated proteins in L. rhamnosus GG include transcription- (2 glycoproteins) and nucleotide metabolism-related proteins (1). One protein could not be attributed to a specific class (LGG_01900), as the role of this protein is currently unknown.

5.3.5 Confirmation of GAPDH glycosylation underlines the efficacy of the workflow

The LGG_00933 encoded glyceraldehyde 3-phosphate dehydrogenase (GAPDH), GapA, is a well-conserved glycolytic enzyme, catalyzing the conversion of glyceraldehyde 3-phosphate to D-glycerate 1,3-phosphate. Apart from this canonical function, several moonlighting roles have been attributed to this enzyme, especially in eukaryotes (Henderson, 2014; Henderson & Martin, 2014). In Prokarya, and more in particular in two L. plantarum strains, this protein has been implicated in binding of the bacterium to mucin, human blood antigens and the intestinal epithelial cell line Caco-2 (Kinoshita et al., 2008a; Kinoshita et al., 2008b; Ramiah et al., 2008). These findings might be valid in L. rhamnosus GG too, as L. plantarum bacteria are close relatives and earlier papers also report on the presence of GapA in the extracellular fractions of the L. rhamnosus GG proteome (surface and secretome) (Deepika et al., 2012; Sanchez et al., 2009a; Sanchez et al., 2009b). Nevertheless, further research is necessary to validate this hypothesis.

126

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

Our results reflect the ubiquitous presence of GapA, as this protein was identified in all four proteome fractions for both repetitions of the experiment, illustrating the robustness of the workflow (Table 5.1). These findings also corroborate earlier reports on the extracellular presence of GapA (Deepika et al., 2012; Saad et al., 2009; Sanchez et al., 2009a; Sanchez et al., 2009b), which further substantiates the moonlighting data (Henderson, 2014; Kinoshita et al., 2008a; Kinoshita et al., 2008b; Ramiah et al., 2008).

The GapA enzyme was subsequently purified from all four proteome fractions to confirm its glycosylation status as a validation of the used approach. Gel filtration chromatography resulted in GAPDH enriched samples for each of the four different fractions. This is exemplified for the cytosolic GAPDH enrichment depicted in figure 5.4: the Sypro® stained gel and GapA antibody blot show no other bands than the one of the LGG_00933 protein. Furthermore, this protein reacts strongly with the PAS stain. The GapA purified from the other three fractions generated similar results (results not shown).

Figure 5.4 - The LGG_00933 protein, GapA, is glycosylatedGlyceraldehyde 3-phosphate dehydrogenase (GapA) was isolated from all four fractions of the L. rhamnosus GG proteome. Here the results for the cytosol are depicted. The first panel shows the positive reaction of the protein with the GapA antibody. The protein stained (Sypro® stain) gel fragment shows the abundant presence of the enriched protein on 1D SDS-PAGE after successful purification of the protein from the proteome. Glycosylation of the GapA protein in the cytosolic fraction can be inferred from the strong positive signal of the protein with the Pro-Q® Emerald 488 dye, which is the commercial version of the Periodic acid Schiff base stain method for glycans. Glycosylation was also proven for GapA enriched from the other fractions of the proteome of L. rhamnosus GG (results not shown).

five.

127

Taken together, this independent analysis of purified GapA demonstrates the robustness and efficacy of the proposed workflow, as the glycosylation of this protein could be confirmed in all four fractions of the L. rhamnosus GG proteome. This is, to our knowledge, the first report on the glycosylation of this enzyme.

5.4 Discussion

Here we provide new insights in the glycoprotein repertoire of bacteria by implementing a robust and widely applicable workflow to screen glycoproteins (Fig. 5.1). This study complements our earlier work on the mapping of glycosyltransferases and their potential role (Sanchez-Rodriguez et al., 2014). The methodology makes use of well-established proteomic tools such as 2D electrophoresis and mass spectrometry and combines these techniques with glycoprotein identification assays, like the PAS glycostain method and lectin probing. More in particular the PAS glycostain was used as a general screening method, whilst the lectin probing was implemented as lectins bind selectively to sugar monomers in specific configurations. Here, the mannose specific lectin GNA was used, as prior work indicated that mannose is present in surface glycans of L. rhamnosus GG (Lebeer et al., 2012b; Lebeer et al., 2009), but the choice of the lectin will depend on the species/strains.

The vigorous assessment of the data resulted in the identification of 40 new glycoproteins in the microbiota isolate L. rhamnosus GG (Table 5.1). The glycosylation of these proteins was experimentally shown by two independent methods (PAS staining and lectin blotting), but should be confirmed by other complementary methods such as thorough MS analysis of the purified proteins and their attached glycans. Moreover, the way the workflow was conceived also made it possible to acquire data on the localization of these glycoproteins. While most glycoproteins were identified in the CWCM fraction, this particular way of breaking down the proteome in different fractions also allowed to confirm the presence of several glycosylated proteins in the cytosol (Fig. 5.2). This latter finding corroborates some scarce earlier reports (Fischer & Haas, 2004; Fredriksen et al., 2012), but is -to our knowledge- one of the first reports on the general occurrence of glycosylated proteins in the cytosol of bacteria. The fractionation process was found to be quite robust, since repetition of the experiment showed similar results (Table 1) and confirmed earlier reports on protein localization in L. rhamnosus GG (Deepika et al., 2012; Sanchez et al., 2009a; Sanchez et al., 2009b). For instance, transport-related glycoproteins were indeed identified from the CWCM fraction, while proteins implicated in translation and transcription processes are predominantly found intracellularly. A peculiar finding was the presence of some glycolytic enzymes, i.e. GapA, Eno and Pgk in the extracellular fractions of the proteome. However this result is in accordance with earlier localization studies (Deepika et al., 2012; Saad et al., 2009; Sanchez et al., 2009a; Sanchez et al., 2009b) and corroborates their moonlighting role (Henderson & Martin, 2014). Protein glycosylation might be a regulator of these proteins, i.e. it could act as a switch between their canonical and moonlighting role in the cell. Further research will shed light on the plausibility

128

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

of these hypotheses, but the observation of the glycosylation of these moonlighting proteins is intriguing.

The earlier experimentally validated glycosylated Msp1 protein of L. rhamnosus GG was found, as expected, among the top confidence hits of the analysis (Lebeer et al., 2012b). The glycosylation of similar cell wall hydrolases was already reported in closely related species, like L. plantarum WCFS1 (Fredriksen et al., 2012; Rolain et al., 2013), L. buchneri (Anzengruber et al., 2013), L. lactis (Huard et al., 2003) and Streptococcus faecium (Kawamura & Shockman, 1983). In this work we also report on the glycosylation of two closely related enzymes of Msp1, namely Msp2 and LGG_02016 (Table 5.1). Taken all these reports together, this might point towards a general trend, at least in lactobacilli. Glycosylation might be used by lactobacilli to fine-tune the functionality of these potentially autolytic enzymes related to cell division. This hypothesis is supported by the finding that L. plantarum WCFS1 utilizes glycosylation of the Msp1 homologue, Acm2, as a negative modulator of its enzymatic activity (Rolain et al., 2013). Another potential explanation is the usage of glycosylation as a switch between the canonical role of these enzymes (i.e. cell division) and a potential moonlighting role. These hypotheses can be valid for Msp1 (and Msp2). In addition to their bacterial role as cell wall hydrolases (Claes et al., 2012), Msp1 and Msp2 are also reported to modulate the survival of intestinal epithelial cells after secretion and are thus probiotic effectors of L. rhamnosus GG (Yan et al., 2011; Yan et al., 2007). In this respect, it was reported earlier that Msp1 is less active than its close Msp2 homologue, which was long thought to be non-glycosylated (Lebeer et al., 2012b; Yan et al., 2007). However, in this study, we used a more sensitive approach that resulted in the unambiguous identification of Msp2 glycosylation in independent samples. Probably, the glycan modifying Msp2 is much smaller than the one present on Msp1 causing a large molecular weight shift (ca. 75 kDa on SDS-PAGE vs. 48 kDa theoretically predicted) (Lebeer et al., 2012b), explaining the previous unsuccessful attempts to identify Msp2 as glycoprotein, but this needs to be further investigated by sophisticated glycan MS methods. Here we identified Msp2 as a glycoprotein in both repetitions of the experiment (one time with a significant glycosylation score, cf. table 1) and in an independent GNA-screening (results not shown). Further research is necessary to confirm if glycosylation of cell wall hydrolases is really a general trend in lactobacilli or even in bacteria in general and to delineate the functional importance of their glycosylation.

Another trend that could be inferred from the results was the high number of glycosylated enzymes of the carbohydrate metabolism, mostly glycolytic enzymes (7/11) (Table 5.2). Their glycosylation needs to be further substantiated by independent glycan assays, as it is possible that our observations result from the fact that these enzymes bind thier substrate, i.e. sugars. Neverthless, we envisage that loosely bound sugars will have been removed during the various washing steps of the protocol. Among them is the glyceraldehyde 3-phosphate dehydrogenase, GapA (LGG_00933), which we further purified to confirm its glycosylation. This enzyme is, albeit its intracellular

five.

129

glycolytic role, present in all four fractions of the glycoproteome (Table 5.1) and was found to be glycosylated in all four protein fractions (Fig. 5.4). The ubiquitous presence of this enzyme, even in the extracellular fractions of the proteome, corroborates earlier reports on its moonlighting role (Henderson, 2014; Kinoshita et al., 2008a; Kinoshita et al., 2008b; Ramiah et al., 2008). Another interesting glycosylated enzyme is the L-lactate dehydrogenase encoded by LGG_02523. Albeit the presence of other homologues of this enzyme in the L. rhamnosus GG genome, this seems to be –at least under the tested conditions– the only glycosylated one. Earlier research showed that this class of enzymes is regulated by different mechanisms in lactic acid bacteria (Feldman-Salit et al., 2013). Whether glycosylation can be added to this list, remains to be further substantiated.

Our results correlate well with earlier studies of bacterial glycoproteomes, with respect to the nature of the glycosylated proteins we report. For instance, screenings in C. jejuni (Ding et al., 2009; Scott et al., 2009; Young et al., 2002), B. fragilis (Fletcher et al., 2011), H. pylori (Champasa et al., 2013) and N. gonorrhoea (Vik et al., 2009) resulted in the identification of several transport-related proteins as glycoproteins, which corroborates our analysis in which 12% of the identified glycoproteins belong to this class in L. rhamnosus GG (Fig 5.3 & Table 5.2). Also the glycosylation of chaperones, like DnaK, is substantiated by results of L. plantarum WCFS1 (Fredriksen et al., 2013), H. pylori (Champasa et al., 2013) and F. tularensis (Balonova et al., 2010). Glycosylation of proteins involved in redox processes also seems to be a general trend, as this now has been reported in four distinct bacteria ((Balonova et al., 2010; Ding et al., 2009; Scott et al., 2009; Vik et al., 2009; Young et al., 2002) & Table 5.2). Apart from our report of their glycosylation in L. rhamnosus GG, similar findings emerged from glycoproteome screenings of C. jejuni (Ding et al., 2009; Scott et al., 2009; Young et al., 2002), F. tularensis (Balonova et al., 2010), H. pylori (Champasa et al., 2013) and N. gonorrhoea (Vik et al., 2009). Future analyses of glycoproteomes of a wider diversity of bacterial species will unveil if these trends can be further generalized for bacteria or a subset of species.

It is intriguing that members of the largest bacterial multienzyme complex, the ribosome, seem to be glycosylated. Maybe this glycosylation plays a role in the regulation of the ribosome assembly and the timing of translation. Further research is needed to further substantiate the glycosylation of these proteins and to shed light on the potential role of their glycan modifications. The identified glycoproteins are interesting targets to study in more detail for their immunogenic and host-interaction potential, especially for the hits in the extracellular sections of the proteome. Very peculiar targets to study further are the moonlighting proteins, like the glycolytic enzymes GapA, Pgk and Eno. Especially as their predicted/proven moonlighting role is related to host interaction, i.e. mucin binding, binding to epithelial cells, etc. (Castaldo et al., 2009; Granato et al., 1999; Henderson, 2014; Kinoshita et al., 2008a; Kinoshita et al., 2008b; Sanchez et al., 2009a).

130

syst

emat

ic e

xplo

ratio

n of

the

gly

copr

oteo

me

of a

ben

efici

al g

ut is

olat

e

5.5 concLusions

In this work, a new workflow is presented to map the glycoproteome of a bacterium in a way that provides information on the cellular localization of the glycosylated targets. We were able to identify 41 glycoproteins, which are important for the cellular physiology. Most glycoproteins were recovered from the cell wall/membrane fraction, but we also confirm the presence of glycoproteins in the cytosol. Moreover, our research generated some new insights in the role glycosylation can play in enzyme regulation. This analysis provided a list of targets that might play an important role in the interactions between the well-documented microbiota isolate and model probiotic strain L. rhamnosus GG on the one side and the human gut on the other side.

5.6 suppLementary information

PAS/lectin reactive spots identified with LC MS/MS for each fraction (SP, CWCM, CWA and Cyt) for each repetition of the experiment. Available on request.

five.

131

six.glycosylated

heterotrimeric pili of L. rhamnosus GG

mediate DC-SIGN interaction

pushing multidisciplinary collaborations to the next level

six.glycosylated

heterotrimeric pili of L. rhamnosus GG

mediate DC-SIGN interaction

pushing multidisciplinary collaborations to the next level

134

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

Manuscript in preparation.

Hanne L.P. Tytgat, Nienke H. van Teijlingen, Ruby M. Sullan, François P. Douillard, Marcel Messing, Pia Rasinkangas, Reetta Satokari, Jos Vanderleyden, Yves F. Dufrêne, Teunis B.H. Geijtenbeek, Willem M. de Vos, Sarah Lebeer

autHors’ contributions

HT conceived the study and performed most experiments. Experimental design was performed under guidance of SL, WDV, YD, FD and TG. Research was carried out by HT, NvT, RMS, MM, PR, FD and RS. The manuscript was drafted by HT.

six.

135

abstract

Pili are important proteinaceous cell wall appendages with a well-documented role in adhesion. While pili from Gram-positive bacteria are modified posttranslationally by sortase-specific cleavage reactions and intramolecular peptide bonds, we here report on a new posttranslational modification by glycosylation of pili and its role in the immunological signaling of these pili.

Molecular techniques were combined with nanoscale mechanical and immunological approaches to assess the glycosylation status of the SpaCBA pili of the model probiotic and beneficial microbiota isolate L. rhamnosus GG. Single-molecule force spectroscopy using AFM tips functionalized with different lectins showed the presence of mannose and fucose residues on live piliated L. rhamnosus GG cells but not on isogenic mutants lacking pili. Western blots probed with mannose-specific lectins and glycostained gels of purified pili corroborated these findings.

Intriguingly, the pili also interact with the immune lectin receptor DC-SIGN present on human dendritic cells (DCs). DC-SIGN typically binds fucose and mannose residues on microorganisms. This interaction was further studied using both purified pili and bacterial cells (including a non-piliated mutant) and both cell lines that express recombinant DC-SIGN and monocyte-derived DCs from different donors. The pili glycans were shown to be important for the physical and immunological interactions of L. rhamnosus GG with DCs, in a manner that could be blocked with anti-DC-SIGN agents.

significance This first report on the glycosylation of heterotrimeric pili in bacteria opens avenues to study pili glycosylation in other piliated bacteria. Our work corroborates the role of pili as crucial interaction partners in host-microbe interactions. Their glycosylation can infer an extra level of specificity to these interactions as we show in this work by evidencing the importance of pili glycans in immune interactions. These findings substantiate the few reports on the role of bacterial protein glycosylation and spur the research on the functional role of glycosylated pili. As these findings were gathered for the microbiota isolate and model probiotic L. rhamnosus GG, they are of importance in the further elucidation of the fine details of host-microbe/microbiota interactions.

136

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

6.1 introDuction

Host-microbe interactions are crucial to life: the microorganisms of the human microbiota are vital for our wellbeing. Currently, several big data projects, like the Human Microbiome Project (HMP) and the enterotypes study, are mapping this microbial diversity (Arumugam et al., 2011; The Human Microbiome Project Consortium, 2012). Concurrently, it is important to delineate the key molecules underlying this host-microbial symbiosis. An example of such crucial molecules, which also came out of the enterotype study, is the pili. These cell wall appendages are pivotal in microorganism colonization of the host epithelium, thus increasing their residence time (Arumugam et al., 2011; Krogfelt, 1991). The Arumugam study reported on the importance of these pili in Escherichia coli adhesion (Arumugam et al., 2011), but also other important microbiota members are piliated, like several Firmicutes (e.g. Lactobacillus sp. (Kankainen et al., 2009) and Streptococcus sp. (Brittan & Nobbs, 2015)) and Actinobacteria (e.g. Bifidobacterium sp. (Turroni et al., 2014) and Corynebacterium sp. (Rogers et al., 2011; Ton-That & Schneewind, 2003)) (Arumugam et al., 2011; Soriani & Telford, 2010; The Human Microbiome Project Consortium, 2012).

In view of the important role of these cell wall appendages, the understanding of their structure and its relation to their functional role is key to fully grasp the complexity of host-microbe interactions. The structure of Gram-positive pili is significantly different from Gram-negative species. Gram-positive pili are covalent structures assembled from multiple protein subunits (Kline et al., 2010; Mandlik et al., 2008; Proft & Baker, 2009). Regularly they are composed out of three subunits: a pilus backbone, a tip adhesin and a pilin decorating the pilus shaft (Mandlik et al., 2008; Proft & Baker, 2009). These subunits are posttranslationally linked by sortase-specific cleavage reactions and intramolecular peptide bonds. A housekeeping sortase then covalently binds the assembled pilus to the peptidoglycan backbone after recognition of an LPXTG motif (Hendrickx et al., 2011; Kang et al., 2007). Here we report on a new form of posttranslational modification of complex Gram-positive pili by glycosylation.

The beneficial microbiota isolate and model probiotic strain L. rhamnosus GG is an example of a bacterium producing such sortase-dependent heterotrimeric pili (Douillard et al., 2014; Segers & Lebeer, 2014). Immuno-EM images show that these bacterial cells are covered with 10 to 50 appendages of 1 mm in length (Kankainen et al., 2009). L. rhamnosus GG has two pili gene clusters in its genome, encoding two distinct pili appendages, the spaFED and spaCBA pili gene clusters (Kankainen et al., 2009). It is thought that only one of these gene clusters is actually expressed in standard laboratory conditions (Douillard et al., 2013; Reunanen et al., 2012). Structural studies showed that SpaA forms the pilus backbone and SpaC is the large tip adhesin. SpaB functions as a stop polymerization signal of the heterotrimeric pili and is mainly found at the pilus anchoring point, although some leaky expression across the pilus shaft is observed (Reunanen et al., 2012). These SpaCBA pili are important adhesive molecules: they are identified as the key modulators of the colonization and persistence of L.

six.

137

rhamnosus GG in mucosa-associated niches (Douillard et al., 2013; Kankainen et al., 2009; Lebeer et al., 2012a; von Ossowski et al., 2010). From an immunological perspective, their role is less unambiguous: the pili have been shown to interact with macrophages (Vargas Garcia et al., 2015), seem to impact TLR-2 signaling (Douillard et al., 2013; von Ossowski et al., 2013) and modulate several pro- and anti-inflammatory cytokines (Lebeer et al., 2012a; Vargas Garcia et al., 2015; von Ossowski et al., 2013). It is currently hypothesized that the pili do not play an immunogenic role as such, but rather facilitate immunogenic effects of other surface structures by promoting close contact with host immune cells (Lebeer et al., 2012a; Segers & Lebeer, 2014; Vargas Garcia et al., 2015). Indeed, earlier Single-Molecule Force Spectroscopy (SMFS) studies have shown the unique properties of the SpaCBA pili to control the adhesion of L. rhamnosus GG via a zipper-like mechanism in which (the broad specificity of) the tip adhesin SpaC plays a key role (Tripathi et al., 2013; Tripathi et al., 2012b). The location of SpaC on the tip facilitates the initial contact of L. rhamnosus GG with host cells, after which SpaC molecules dispersed along the pilus shaft establish the more intimate contact (Tripathi et al., 2013). These SMFS studies also provided some first, although indirect, indications on the potential glycosylation of the SpaCBA pili of L. rhamnosus GG in the sense that sheared pili were shown to strongly cluster together. These strong interactions were thought to be lectin-glycan bonds between SpaC pilins (Tripathi et al., 2012b), although no experimental evidence was yet provided.

In this work, we report on the glycosylation of the SpaCBA pili of L. rhamnosus GG and on the functional importance of its glycosylation. More in particular we established that the glycans present on these pili are key for the interaction between L. rhamnosus GG and dendritic cells via the immune lectin receptor DC-SIGN, which specifically recognizes fucose- and mannose- binding lectins. A unique multidisciplinary approach was used to describe the glycosylation of these pili combining nanomechanical, molecular, glycobiological and immunological techniques. This enabled us to present one of the few functional reports on the role of bacterial protein glycosylation and on the interaction of bacterial glycoproteins with immune receptors.

6.2 materiaL anD metHoDs

6.2.1 bacteriaL strains anD cuLture conDitions

L. rhamnosus GG was grown at 37°C in MRS medium (Difco) in non-shaking conditions. For some experiments the cells, dissolved in PBS, were heat killed by incubating them 15 minutes in a water bath at 60°C.

6.2.2 Data analysis

All data analyses were performed using GraphPad Prism® 6. Significant differences between two values were calculated using paired t-tests and the significance level was set at p <0,05.

138

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

6.2.3 singLe-moLecuLe force spectroscopy

Single-molecule force spectroscopy (SMFS) experiments were performed on live L. rhamnosus GG bacteria using the general procedure described previously (Francius et al., 2009; Francius et al., 2008; Lebeer et al., 2009). Two lectins were used, i.e. the fucose-specific Aleuria aurantia lectin (AAL, Vector Laboratories) and the Hippeastrum hybrid agglutinin (HHA) to detect mannose residues (kindly provided by Prof. Els Van Damme, UGhent). Wild type L. rhamnosus GG (ATCC 53103) and a pili-deficient ΔspaCBA::TcR mutant (CMPG5357) (Lebeer et al., 2012a) were analyzed, following single centrifugation at 4000 g for 2 min. AFM tips were functionalized with lectins using ~6 nm long PEG-benzaldehyde linkers (Ebner et al., 2007). Cantilevers were first washed with chloroform and ethanol, placed in an UV-ozone cleaner for 15 min, and immersed overnight in 5.6 M ethanolamine hydrochloride (in DMSO) to generate amino groups on the tip surface. The cantilevers then reacted with PEG linkers carrying benzaldehyde groups on their free-tangling end for 2 hours, were afterwards rinsed with chloroform and immersed further in 1% citric acid solution for 10 min. After rinsing with Milli-Q water, the cantilevers were covered with 200 mL PBS solution containing 0.2 mg/mL lectins, to which 2 mL of a 1 M NaCNBH3 solution was added. After 50 min of incubation, 5 μL of a 1 M ethanolamine hydrochloride solution (pH 9.5) was added to block unreacted aldehyde groups (10 min). Finally, the cantilevers were washed with PBS and stored in the same buffer with added 1% sodium azide (NaN3), until use (within 7 days). SMFS measurements were performed at 25°C in PBS buffer (pH 7.4) supplemented with 1 mM CaCl2 and 1 mM MgCl2 using a Nanoscope VIII Multimode AFM (Bruker Corporation) and oxide sharpened microfabricated Si3N4 cantilevers (Microlevers, Veeco Metrology Group). The spring constants of the cantilevers were typically in the range of 0.02–0.04 N/m, as determined by the thermal noise method (Hutter & Bechhoefer, 1993). Cells were filtered and immobilized onto 1.2 mm diameter isopore polycarbonate membrane (Millipore). The membrane filter was then gently rinsed 3 times with PBS, carefully cut into ~1x1 cm2 square and attached to a steel sample puck using a double-sided adhesive and mounted onto the AFM liquid cell. Unmodified cantilevers were first used to image and localize individual L. rhamnosus GG cells. The unmodified cantilevers were then replaced with the lectin-functionalized tip. Force mapping was performed by collecting a 32x32 array of force-distance curves on a 300–500 nm2 localized area of the cell and the two-dimensional adhesion maps were reconstructed using the custom analysis described previously (Sullan et al., 2014). All force curves were recorded at ~100 ms contact time, with a maximum applied force of 250 pN and 2000 nm/s approach and retraction speeds, unless specified otherwise.

6.2.4 purification of tHe spacba piLi of L. rhamnosus gg

L. rhamnosus GG cells (approx. 1011 CFU/mL) were grown in an industrial whey permeate based medium (Laakso et al., 2011) and were a kind gift of Dr. Tuomas Salojarvi and Dr. Ross Crittenden (Valio Oy Finland). Cells were concentrated by microfiltration, resulting

six.

139

in reatining of the pili (unpublished observation). This concentrate was centrifuged for 30 minutes at 20 000 g in an Avanti J-26 XPI high-performance centrifuge (Beckman Coulter), which resulted in the shearing of the pili (Tripathi et al., 2012b). The supernatant (i.e. Pili Raw Material or PRM) was lyophilised (Heto FD 2.5, Heto Lab Equipment) and the residue dissolved in 1/25 of its original volume of PBS. The resulting solution was filtered twice, first through a 0.45 µm Millex GS syringe-driven filter (Millipore) and then using a 0.22 µm Millex HA syringe-driven filter (Millipore). 3 mL aliquots of the filtrate were run through a 120 mL Superdex 200 gel-filtration column (GE Healthcare) on a FASTA system (Pharmacia) using PBS as the eluent at a speed of 1 mL/min. Previous experiments had shown that the pili elute in the first protein peak. Based on this result, twenty-four 1 mL fractions of the eluate were collected per run in a 96 deepwell plate. The registration of the absorbance at 280 nm was used to identify the fractions with a high pili content.

Two pools were made: Pool A, containing the six fractions of each run nearest to the top of the absorbance peak, and Pool B, containing the six fractions of each run following those of Pool A (cf. additional figure). Thus resulting in pure pili samples, with pool A having a higher molecular weight and purity (21 µg/mL) compared to pool B samples (9 µg/mL). Both pools were desalted by changing the buffer from PBS to 50 mM (NH4)2CO3/NH4HCO3 buffer, pH 8.0 using ZebaTM Spin desalting columns, 40K MWCO (Pierce Biotechnology). The resulting solutions were lyophilised (Heto FD 2.5) and the residues dissolved in 1/25 of their original volume of PBS and stored at -20 oC. The purity of the pili preparations was assessed by Coomassie blue staining, dot blot and probing of the blotted samples with a specific SpaC monoclonal antibody (kind gift of MicroDish BV and Podiceps BV) (results not shown). The purified pili samples (PRM, B and A) were aliquoted in small amounts, stored at -20°C and thawed prior to use.

6.2.5 purification of tHe L. rhamnosus gg msp1 protein

Native Msp1 was purified as described earlier (Lebeer et al., 2012b) using a combination of cationic exchange and ConA lectin affinity chromatography. Briefly, L. rhamnosus GG was grown for 24 h in AOAC medium. The supernatant of the culture (after centrifugation at 6 000 g for 20 min) was loaded on a SP Sepharose HighPerformance column (GE Healthcare), equilibrated with 60 mM lactate buffer (pH 4.0). The column was eluted using lactate buffer containing a NaCl gradient (100-1000 mM). The presence of Msp1 was verified using SDS-PAGE and the positive fractions were spin concentrated with Vivaspin filters with a MW cut off of 10 000 kDa (Sartorius Stedim biotech Gmbh). Msp1 was further purified from these positive fractions using a Hi-trap ConA 4B prepacked column (GE Healthcare).

6.2.6 isoLation of ceLL waLL-associateD proteins

Cell wall-associated proteins were isolated from wild type L. rhamnosus GG, the ΔspaCBA::TcR mutant lacking pili (CMPG5357) and the ΔwelE::TcR mutant (CMPG5351) (Lebeer et al., 2009)

140

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

on which the SpaCBA pili are overexposed due to the absence of the thick exopolysaccharides layer surrounding the cells. Kankainen and co-workers (Kankainen et al., 2009) used this approach earlier to confirm the presence of SpaCBA pili on L. rhamnosus GG cells. Here we followed the same protocol to isolate the cell wall-associated proteins. The obtained fractions were then separated on NuPAGE® Novex® 3-8% Tris-Acetate gels (Life Technologies), which are specifically designed to separate high molecular weight proteins. Western blotted samples were probed with Msp1 antibody (kind gift of Dr. Ingemar von Ossowski, UHelsinki), SpaC antiserum, the mannose-specific lectin HHA (kind gift of Prof. Els Van Damme, UGhent) and the fucose-specific lectin AAL (Vector Laboratories) (cf. below). This assay was repeated in triplicate.

6.2.7 perioDic aciD scHiff gLycostain

Purified pili fractions were separated by SDS-PAGE on commercial NuPAGE® Novex® 3-8% Tris-Acetate gels (Life technologies). Glycostaining was performed using the commercial Periodic Acid Schiff base stain Pro-Q® Emerald 488 (Molecular Probes®) following the manufacturer’s guidelines. Poststaining of the PAS stained gel with Sypro® Ruby stain (Molecular Probes®) enabled visualization of the protein content of the sample. The gels were scanned using a Typhoon 9400 laser scanner (GE Healthcare). Purified Msp1 was used as a positive control. The pili samples were also separated on Acrylamide Bis-Acrylamide gels (Amresco). The protein content of these samples was visualized using Sypro® stain and Silver stain (SilverQuestTM, Life Technologies). Glycosylation of the samples was tested using the Pro-Q® Emerald 488 kit. Sypro® and PAS stained gels were scanned with a Typhoon 9400 laser scanner. All assays were performed at least in triplicate for all purified pili fractions to verify the data.

6.2.8 Lectin anD antiboDy probing of western bLots

Purified pili samples and cell wall-associated proteins were separated by SDS-PAGE on Acrylamide Bis-Acrylamide gels (Amresco) and commercial NuPAGE® Novex® 3-8% Tris-Acetate gels. These gels were electroblotted to PVDF membranes (Life technologies) as previously described (Lebeer et al., 2012b) and using the here mentioned tweaks. The blots were blocked with 0.5% polyvinylalcohol (PVA, Sigma-Aldrich®) to reduce background noise (Thompson et al., 2011), dissolved in TBST (20 mM Tris-HCl, 500 mM NaCl, 0.1% Tween 20, pH 7,5) supplemented with 1 mM CaCl2 and 1 mM MgCl2 as cofactors for the lectins. All washing steps were performed using this TBST buffer with CaCl2 and MgCl2. The blots were probed with a plethora of eight digoxigenin-labeled lectins, such as the mannose-specific Galanthus nivalis agglutinin (GNA) and HHA (kindly provided by Prof. Els J.M. Van Damme, UGhent), as described in (Tytgat et al., 2014).

Both types of gels (Tris-Acetate gels and Acrylamide Bis-Acrylamide gels) were Western blotted and probed with SpaC antiserum to visualize their pili content as described by Kankainen et al. (Kankainen et al., 2009). Each purified pili fraction was tested as described

six.

141

at least in triplicate.

6.2.9 Enzyme-linked lectin assay

Purified pili (5 mg/mL, sample A) were dissolved in 50 mM sodium carbonate buffer (pH 9.6) and coated overnight onto ELISA plates (X50 Immunolon 4HBX plates, Thermo Scientific) at 4°C. For the lectin probing using the mannose-specific lectin HHA, mannan was used as a positive control (5 mg/mL). In the case of the fucose-specific lectin AAL, Lewis X was coated as a positive control (5 mg/mL) (kind gift of Prof. T. Geijtenbeek, AMC, UvA). Wells only containing 50 mM sodium carbonate buffer were used as negative controls for the assay.The next day the wells were washed trice with PBS, preceding the blockage of the wells with 0.5% PVA dissolved in TBS (Thompson et al., 2011) for 2-3h at 37°C. The wells were then washed three times with PBS prior to the addition of the digoxigenin-labeled lectins dissolved in TBS + 0.5% PVA (1h, 37°C). Unbound lectins were removed by washing the wells three times with PBST (PBS + 0.1% Tween 20). Then the secondary antibody, anti-digoxigenin (Roche) (Tytgat et al., 2014) was added to the cells for 1 hour at 37°C. Prior to development, the wells were washed three times with PBST. 4-nitrophenyl phosphate disodium salt (1mg/mL, Sigma-Aldrich®) dissolved in substrate buffer (carbonate bicarbonate buffer, pH 9.6) was added to each well. Color was allowed to develop for 15-30 min at room temperature and the absorbance was measured at 405 nm.

6.2.10 Dc-sign-fc eLisa

High-binding flat-bottom 96-well ELISA plates were coated overnight with 50 mL of 50 mg/mL ligand dissolved in 0.2 M coating buffer at room temperature. Mannan (1 mg/mL, Sigma-Aldrich®) was coated as a positive control, whilst plain coating buffer served as negative control. After two washes with TSM (20 mM TrisHCl, 150 mM NaCl, 1 mM CaCl2, 2 mM MgCl2) the wells were blocked with 1% BSA (Sigma-Aldrich®) for 30 minutes at 37°C. DC-SIGN-Fc fragments were 15 minutes preincubated with blocking agents in TSM: the D1 DC-SIGN specific antibody (20 mg/mL) and calcium chelator EGTA (10 mM). After two more washes with TSM, the plate was incubated with 50 L of DC-SIGN-Fc fragments (with block) for 2 hours at room temperature on a shaking platform. The wells were washed four times with TSM with 0.1% Tween 20 (TSMT), prior to the addition of goat anti-human Fc-PO (Sigma-Aldrich®) dissolved in TSMT (30 minutes). After six washes with TSMT, the ELISA could be developed using a 10 mL substrate solution of 0.1M NaAc, 100 mL TMB and 1 mL H2O2. 15 minutes later this reaction was stopped by the addition of 50 mL H2SO4. Experimental read-out was performed at OD450.

6.2.11 interaction between piLi-coateD beaDs, bacteria anD raji(Dc-sign) ceLL Lines

Beads were coated as described by Geijtenbeek and co-workers (Geijtenbeek et al., 1999). In short, streptavidin was covalently coupled to fluorescent beads (TransFluoSpheres,

142

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

1 μm, Alexa 488/Alexa647) and further coated by subsequent incubation with biotinylated Swine-anti-rabbit-antibody (20 mg/mL, 2h 37°C, Sigma-Aldrich®), SpaC antiserum (5 mg/mL, overnight 4°C) and purified pili (sample B, overnight 4°C). Bacteria were labeled with FITC (Sigma-Aldrich®) by mixing 50 mL of 1 mg/mL FITC (Sigma) in DMSO with 500 mL 2x108 bacteria in PBS, incubated for 1 hour at RT and washed extensively. Raji and Raji-DC-SIGN cells (Geijtenbeek et al., 2000a) were cultured in RPMI (Gibco) supplemented with 10% FCS, 2500 U/mL penicillin, 2500 μg/mL streptomycin, 100 mM L-Glutamine and passaged twice a week. 5x104 moDCs, Raji or Raji-DC-SIGN cells were preincubated for 30 min at 37°C with TSA (TSM supplemented with 0.5% Albumin), DC-SIGN specific antibody D1 (20 mg/mL), Lewis X (20 mg/mL) or EGTA (5 mM, Sigma-Aldrich®), respectively. Next, cells were incubated for 1 hour with FITC-labeled bacteria at MOI 20, washed extensively, fixed in 4% vol/vol PFA, and analyzed on FACSCantoTM II (BD Biosciences). With the fluorescent beads, 5x104 moDCs, Raji or Raji-DC-SIGN cells were preincubated for 30 min at 37°C with TSA, DC-SIGN specific antibody D1 (20 mg/mL), Lewis X (20 mg/mL), or EGTA (5 mM), respectively. Next, cells were incubated for 30 minutes with 3 mL beads, washed extensively, fixed in 4% PFA, and analyzed on FACSCantoTM II (BD, Biosciences).

6.3 resuLts

6.3.1 afm singLe-moLecuLe measurements inDicate tHat tHe L. rhamnosus gg spacba piLi are gLycosyLateD witH fucose anD mannose

SMFS (Dupres et al., 2005; Hinterdorfer & Dufrene, 2006) was previously used to assess the nanomechanical properties of the SpaCBA pili of L. rhamnosus GG (Tripathi et al., 2013). Here a similar approach was followed to assess the glycosylation status of these pili on living L. rhamnosus GG cells using AFM tips functionalized with mannose- and fucose-specific lectins. These lectins were chosen based on earlier knowledge on often occurring sugars in microbiota members, in view of the abundance of mannose and fucose in the gut mucus, like the sugars present on Msp1 (Lebeer et al., 2012b) and Bacteroides fragilis (Coyne et al., 2005; Fletcher et al., 2009).

Fig. 6.1 shows the adhesion and rupture length histograms of the interaction of wild type L. rhamnosus GG cells and the ΔspaCBA::TcR mutant (CMPG5357) lacking the SpaCBA pili (Lebeer et al., 2012a) with the two lectin probes. On wild type cells, a substantial fraction of the force curves exhibited adhesion forces for the two lectin probes, suggesting the detection of sugar residues. Close examination of the curves revealed two adhesive behaviors. A large fraction of the adhesive curves featured single adhesion peaks with moderate forces, mostly in the 50–200 pN range (lower curve in the insets in Fig. 6.1B, 6.1D). In view of earlier work (Francius et al., 2008), we ascribe these events to specific lectin-sugar interactions. In addition, a significant fraction of the curves (from 10% to 40% depending on the cell and on

six.

143

the probe) showed much larger adhesion forces, in the range of 500–1500 pN, and extended rupture lengths (500–2000 nm) (upper curve in the insets in Fig. 6.1B, 6.1D). Consistent with earlier work (Sullan et al., 2014; Tripathi et al., 2013), these force peaks showed linear shapes and constant force steps indicating that the L. rhamnosus GG pilus behaves as a nanospring, probably undergoing structural changes when subjected to high forces. This finding strongly suggests that the two lectin probes specifically bind to the L. rhamnosus GG SpaCBA pili, indicating that they are likely to be glycosylated with fucose and mannose residues. Potentially, also glucose could be present on the SpaCBA pili, as the ConA lectin also can specifically interact with these residues. Supporting these observations, pili-less cells (ΔspaCBA::TcR mutant, CMPG5357) showed no significant binding events (Fig. 6.1E) and a complete lack of nanospring profiles (Fig. 6.1F).

Figure 6.1 – Probing of fucose and mannose residues on L. rhamnosus GG pili using AFMFig. 6.1A and C depict the adhesion forces and Fig. 6.1B, D the rupture length histograms (n=1024) obtained in buffer, from the interaction between L. rhamnosus GG wild type and fucose- and mannose-binding lectin probes (AAL and HHA resp.). In Fig. 6.1E,F the force data for the interaction of a pili-deficient ΔspaCBA::TcR mutant (CMPG5357) with the two lectin probes are displayed. Insets show representative retraction force curves.

6.3.2 gLycostaining anD Lectin probing confirms gLycosyLation of purifieD spacba piLi witH mannose anD fucose resiDues

SpaCBA pili were purified from wild type L. rhamnosus GG cells to investigate their glycosylation status. Purification resulted in three fractions of pili of different purity, namely the Pili Raw Material (PRM), sample B (concentration of 9 μg/mL) and a pure sample of 21 μg/mL (A). Glycosylation of all these purified pili fractions was assessed using the Periodic Acid Schiff glycostain (PAS), of which the results are depicted in figures 6.2 and 6.3. All

144

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

purified pili fractions (fraction A, Fig. 6.2, and PRM and B fraction on Fig. 6.3), show a signal in the high molecular weight range (i.e. >200 kDa) of the gels, which is typically the region where SpaCBA pili can be found after SDS-PAGE (Kankainen et al., 2009). A SpaC antiserum probed Western blot corroborates that the SpaCBA pili are the cause of the PAS signal (Fig. 6.2 and 6.3). As a control of the efficacy of this method, purified Msp1 was also subjected to the same procedure resulting in the typical 75 kDa band (cf. previous results (Lebeer et al., 2012b)) and a higher glycostained band (complex or oligomer of Msp1). As shown in figure 6.2 and 6.3, the Msp1 protein is completely absent in sample B and A, whilst the glycoprotein is only present in very low abundance in the PRM fraction (cf. absence of bands at 75 kDa, Lebeer et al., 2012b). These results rule out any interference of Msp1 with the pili glycosylation signal. Protein content was visualized using Sypro® and silver stain (Fig. 6.2 and 6.3).

Figure 6.2 – Glycostaining of purified SpaCBA pili Application of the PAS glycostain to the purified pili sample (pool A) show the presence of glycoproteins in the high molecular weight area (> 200 kDa) of the SDS-PAGE separated samples. Purified Msp1 was used as a positive control, whilst probing of the Western blotted samples with SpaC antiserum showed that the PAS signal could be attributed to the SpaCBA pili as both show a similar pattern. Protein content is visualized using Sypro® staining. (LK = Precision Plus ProteinTM KaleidoscopeTM Standard, Bio-Rad)

All purified pili fractions were probed with a range of lectins with different specificities (Man, Fuc, GlcNAc, Gal, Glc, GalNAc), as described by Tytgat et al. (Tytgat et al., 2014). The results of the probing of the PRM and B fraction with the mannose-specific lectins GNA and HHA is depicted in figure 6.3. The lectins with other specificities, except for the fucose-specific lectin AAL, did not show any binding in the high-molecular weight area of the Western blotted samples (results not shown). All blots and gels were performed at least in triplicate to confirm the validity of the results.

six.

145

Figure 6.3 – Probing with mannose-specific lectins confirms SpaCBA pili mannosylationPurified pili fractions (PRM and pili pool B) were subjected to PAS staining, confirming their glycosylation in the high molecular weight area. Both Sypro® and Silver stain were used to show the protein content of these purified samples (first two lanes). The absence of bands at 75 kDa on PAS and lectin blots rules out the interference of this glycoprotein in the PAS signal (cf. sample B), whilst probing with SpaC antiserum confirmed the correspondence of the glycosylation signals to the presence of pili. Similarly, Western blots were also probed with the mannose-specific lectins HHA and GNA, further corroborating SpaCBA mannosylation.(LK = Precision Plus ProteinTM KaleidoscopeTM Standard, Bio-Rad)

The glycosylation of the pili with mannose and fucose residues was further substantiated with Enzyme-linked lectin assays (ELLA). In these assays the purified pili fraction A was probed with the mannose-specific lectin HHA (Fig. 6.4, left panel) and the fucose-specific lectin AAL (Fig. 6.4, right panel). Both lectins were shown to bind significantly to the purified pili, confirming the SMFS results that also indicated modification of the pili with mannose and fucose residues.

Figure 6.4 - Enzyme-linked lectin assays corroborate the presence of mannose- and fucose-residues on purified SpaCBA piliPurified pili significantly interact with the mannose specific lectin HHA (left panel) and fucose specific lectin AAL (right panel). Coating buffer was used as a negative control, whilst known ligands of both lectins, mannan and Lewis X respectively, served as a positive control. As a read-out absorbance was measured at 405 nm.

6.3.3 sampLing of ceLL waLL-associateD proteins of L. rhamnosus gg mutants confirms in vivo piLi gLycosyLation

To corroborate the SMFS results and glycan-probing assays performed on purified pili further, cell wall-associated proteins were isolated form L. rhamnosus GG cells as described

146

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

earlier (Kankainen et al., 2009). Cell wall-associated proteins were isolated from wild type L. rhamnosus GG, the pilus deficient ΔspaCBA::TcR mutant (CMPG5357) and the ΔwelE::TcR mutant (CMPG5351), which lacks the protective, thick exopolysaccharides layer, resulting in an overexposure of the SpaCBA pili (Lebeer et al., 2009).

The results are depicted in figure 6.5 and show that indeed the pili are absent in the ΔspaCBA::TcR

mutant (SpaC antiserum blot), causing also an absence of fucose- and mannose-specific lectin binding to cell wall-associated proteins of this strain (lectin probing with AAL and HHA resp.). Both wild type L. rhamnosus GG and ΔwelE::TcR cell wall-associated proteins do bind to these lectins. These results corroborate our earlier findings further on the presence of fucose and mannose on the SpaCBA pili of L. rhamnosus GG further. Interference of Msp1 signals in the high molecular weight area were ruled out by probing with Msp1 antibody. The presence of Msp1 and other glycoproteins can account for residual signals on the lectin blots for the ΔspaCBA::TcR mutant.

Figure 6.5 – SpaCBA pili are glycosylated on L. rhamnosus GG cellsCell wall-associated proteins of L. rhamnosus GG wild type (WT), the pilus-deficient ΔspaCBA::TcR mutant (CMPG5357, SpaC) and the pili-overexpressing ΔwelE::TcR mutant (CMPG5351, WelE) were isolated to probe their reactivity with mannose- and fucose-specific lectins (HHA and AAL resp.). The positive pattern resulting from lectin probing of wild type and ΔwelE::TcR cells could be linked to the presence of SpaCBA pili by a SpaC antiserum probed Western blot as both (most) lectin and SpaC signals are absent for the non-piliated L. rhamnosus GG strain samples. Interference of the Msp1 glycoprotein was ruled out by a Western blot probed with Msp1 antibody. (LK = Precision Plus ProteinTM

KaleidoscopeTM Standard, Bio-Rad)

6.3.4 purifieD spacba piLi interact witH Dc-sign-fc fragments

Both the SMFS and lectin probing experiments showed that SpaCBA pili of L. rhamnosus GG are modified with mannose and fucose residues. These results indicate that these pili are potential ligands for the immune lectin receptor DC-SIGN, which can detect a variety of microorganisms. This lectin receptor found on human dendritic cells (DCs) specifically recognizes fucose and mannose residues (Geijtenbeek et al., 2000b). Here, we used recombinantly produced DC-

six.

147

SIGN to test this hypothesis. These DC-SIGN Fc fragments consist of the extracellular portion of DC-SIGN (aa 64–404) fused at the C terminus to the human IgG1 Fc domain (Geijtenbeek et al., 2002).

Figure 6.6 – DC-SIGN-Fc fragments interact specifically with SpaCBA pili An ELISA with DC-SIGN-Fc fragments results in specific binding to SpaCBA pili. Coating buffer was used as a negative control and mannan as positive control (not shown). Binding was quantified by measuring the absorbance at 450 nm. The interaction of the SpaCBA pili with the DC-SIGN-Fc fragments could be blocked with the DC-SIGN specific antibody D1, mannan (M, a ligand of DC-SIGN) and the Ca2+ chelator EGTA.

Figure 6.6 shows the results of an ELISA-binding assay performed with purified pili (sample B) and recombinant DC-SIGN-Fc fragments (results of triplicate repetition are shown). There is indeed a significant interaction between the purified pili and the purified DC-SIGN receptor. Moreover, this interaction can be significantly blocked with a specific antibody directed against DC-SIGN, D1 (Geijtenbeek et al., 2000b). The calcium ion-chelating agent EGTA (ethylene glycol tetra acetic acid) also blocked the binding of the pili to DC-SIGN. This latter finding further corroborates the sugar-lectin nature of the interaction between the pili and DC-SIGN, as Ca2+ is an important cofactor enabling DC-SIGN binding (Geijtenbeek et al., 2000b). Mannan (1 μg/mL) was included as a positive control and showed a very strong interaction (OD450 = 0.9 ± 0.1) with the DC-SIGN-Fc fragments (result not shown). Compared to mannan, the interaction of the SpaCBA pili with the lectin receptor is more subtle (OD450 = 0.2 ± 0.03), but nonetheless significant. The occurrence of the fucose and mannose residues on the pili in a more complex context (conformation, configuration…) could explain the fact that their interaction is lower compared to that of a pure ligand, like mannan.

6.3.5 Dc-sign expressing ceLL Lines binD botH purifieD spacba piLi anD piLiateD ceLLs

To test the interaction between the SpaCBA pili and DC-SIGN in a more physiological context, purified pili were coincubated with the recombinant DC-SIGN expressing cell line (Raji-DC-SIGN) (van Gisbergen et al., 2005). To allow the quantitative flow cytometric assessment of this interaction, the purified pili were coupled to fluorescent beads (Geijtenbeek et al., 1999).

148

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

Figure 6.7 - Pili-beads interact with the Raji-DC-SIGN cell linePurified pili were conjugated to fluorescent beads enabling the assessment of their interaction to Raji cells transfected with the DC-SIGN receptor with flow cytometry. Pili-beads could not interact with untransfected Raji cells, but did show a strong interaction with the Raji-DC-SIGN cells. Binding of these beads to DC-SIGN could moreover be blocked with the D1 antibody, the Lewis X ligand and the chelating agent EGTA. Binding to Raji and Raji-DC-SIGN cells is shown relative to pili-bead Raji-DC-SIGN binding.

In figure 6.7 the binding between these pili-conjugated beads and the Raji-DC-SIGN cell line is shown relative to the interaction between bead-binding to untransfected Raji cells. In absolute values, a striking 60% of the pili-beads effectively bound the Raji-DC-SIGN cells. Using the DC-SIGN specific antibody D1, the interaction could be lowered to the background level, illustrating the specificity of the interaction between the pili and the DC-SIGN receptor. Similarly to the DC-SIGN-Fc ELISA results, EGTA also reduced the number of binding events significantly. Further corroborating the interaction between the glycosylated pili and the lectin receptor, is the blockage of the pili-bead interaction with the fucosylated DC-SIGN ligand Lewis X (LeX) (van Die et al., 2003).

Next, we aimed to check if the pili still function as DC-SIGN ligands when present in the more complex context of bacterial cells. Fluorescent wild type L. rhamnosus GG and ΔwelE::TcR CMPG5351 mutant piliated cells were tested (FITC-labeled). Non-piliated ΔspaCBA::TcR

CMPG5357 cells were used as a negative control.

The results depicted in figure 6.8 were obtained using heat killed cells, as independent experiments demonstrated no difference in binding efficacy compared to live cells (results not shown). The interaction between FITC-labeled bacterial cells and the cell lines was investigated using flow cytometry (Geijtenbeek et al., 2003). Wild type L. rhamnosus GG cells significantly interacted with the Raji-DC-SIGN cells when compared to their interaction with untransfected Raji cells. The non-piliated ΔspaCBA::TcR (CMPG5357) mutant however shows no specific interaction with the Raji-DC-SIGN cell line, as this interaction was not susceptible

six.

149

to blocking by DC-SIGN blocking agents and did not significantly differ from binding to untransfected Raji cells. The ΔwelE::TcR (CMPG5351) mutant, having more accessible pili, interacted even stronger with the DC-SIGN Raji cell line than wild type L. rhamnosus GG. When looking at the interference of specific blocking agents with the binding of the bacteria and the cell line, the same trends can be observed for the wild type and ΔwelE::TcR mutant bacteria. The results for the ΔwelE::TcR mutant seem to be a magnification of the wild type results. For both strains DC-SIGN recognition can be blocked almost to background level with the anti-DC-SIGN antibody D1 and Lewis X. Less blocking can be seen when the EGTA chelating agent is applied.

Figure 6.8 – Piliated strains can interact with DC-SIGN expressing cell linesWild type L. rhamnosus GG and ΔwelE::TcR cells could interact with the Raji-DC-SIGN cell line, whilst the non-piliated ΔspaCBA::TcR only showed aspecific binding to Raji cells. The interaction of piliated L. rhamnosus GG cells can be reduced and even prevented using DC-SIGN binding agents, like D1 and Lewis X, and the ion-chelator EGTA. All values represent the relative binding to Raji-DC-SIGN cells relative to the binding of wild type L. rhamnosus GG to untransfected Raji cells.

Taken together, these results suggest that the glycosylated pili are important ligands for the DC-SIGN C-type lectin receptor. A higher exposure and accessibility of the SpaCBA pili strongly increases (i.e. nearly doubles) the interaction between L. rhamnosus GG cells and the Raji-DC-SIGN cell line.

6.3.6 gLycosyLateD spacba piLi meDiate tHe interaction between L. rhamnosus gg anD monocyte-DeriveD Dcs

To challenge the interaction between the glycans of the SpaCBA pili of L. rhamnosus GG and DC-SIGN in in vivo mimicking conditions, adhesion of both pili-beads and live cells to monocyte-derived DCs (moDCs) was tested. Our goal was to analyze whether the interaction of the glycosylated SpaCBA pili is of importance in a more complex physiological context.

150

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

Figure 6.9 – Purified pili bind monocyte-derived dendritic cells via DC-SIGNBeads coated with purified pili could specifically interact with moDCs via DC-SIGN as the interaction could be inhibited by the addition of the D1 anti DC-SIGN antibody, the DC-SIGN ligand Lewis X and the Calcium chelator EGTA. Interference of the Mannose Receptor could be ruled out as an antibody directed against this receptor could not block binding of the pili to moDCs. Usage of the DC-SIGN antibody D1 was validated using an IgG1 antibody isotype, which was unable to interfere with DC-SIGN binding of the pili. Binding values are calculated relative to the binding of pili-beads to moDCs.

The same fluorescent pili-bead adhesion assay was performed as described earlier for the interaction with the Raji-DC-SIGN cells. moDCs of three independent donors were tested. The results clearly show a significant interaction between the pili-beads and the moDCs (Figure 6.9). Moreover this interaction could be blocked with the anti-DC-SIGN antibody D1, EGTA and Lewis X. In this assay we also included an isotype of the D1 antibody (IgG1), to illustrate that the DC-SIGN specific domains of the D1 antibody are responsible for the blocking effect. Inclusion of an antibody directed against the lectin Mannose Receptor (MR) further underpinned the specificity of the pili-DC-SIGN interaction. This also ruled out the involvement of the MR in the interaction between the pili and DCs as the MR antibody could not significantly diminish the interaction between the pili and the moDCs.

Finally, the role of glycosylated SpaCBA pili in the interaction between FITC-labeled live bacteria and moDCs was investigated. The complex context of these interactions makes these experiments quite challenging, resulting in milder interactions compared to the above described model systems (i.e. pili-beads, Raji-SIGN cell line, DC-SIGN Fc ELISA).

Nevertheless, the experiments reflect similar trends compared to the earlier assays, showing an interaction between the bacteria and the moDCs, which can be blocked with DC-SIGN ligands and blocking agents. In the upper panel of Fig. 6.10 a representative experiment is shown (two technical repeats in one moDC donor), equivalent results were obtained in independent repetitions of the experiment. A similar trend in binding could be perceived for wild type L. rhamnosus GG and ΔwelE::TcR cells. The interaction between these cells and the moDCs could be blocked with D1, LeX and to a minor extend also with EGTA. Also in this case, the increased accessibility (overexposure) of the pili on the ΔwelE::TcR cells results in a magnification of the trends perceived for wild type cells (results not shown). In the lower panel

six.

151

of figure 6.10, the interaction percentages are depicted for the ΔwelE::TcR cells, illustrating the specific blockage of moDC interaction in the presence of anti-DC-SIGN molecules. For instance, the D1 DC-SIGN antibody could reduce the binding of ΔwelE::TcR cells to moDCs with almost 30% (Fig. 6.10, lower panel). Incubation of moDCs with the negative control, i.e. ΔspaCBA::TcR cells, did not result in a specific interaction, as interaction levels could not be blocked. There is however significant aspecific interaction between the moDCs and these non-piliated cells, probably caused by the interaction of other cell wall molecules with other moDCs receptors or by aspecific agglutination of the bacteria to the moDCs.

Figure 6.10 – L. rhamnosus GG interaction with DCs is based on recognition of SpaCBA glycans by the DC-SIGN lectin receptorFig. 6.10 upper panel - Both piliated L. rhamnosus GG strains (i.e. wild type and ΔwelE::TcR) interact with moDCs. This interaction depends on the glycosylated pili on the one hand and the DC-SIGN lectin receptor of moDCs on the other hand. This latter fact is illustrated by the reduction of this interaction with the DC-SIGN binding blocking agents D1, EGTA and Lewis X. Pili-deficient L. rhamnosus GG bacteria (ΔspaCBA::TcR) could only interact aspecifically with the moDCs, as this interaction did not alter under the influence of anti-DC-SIGN agents. Binding of strains to moDCs is shown relative to the binding of wild type L. rhamnosus GG to moDCs without blocking agents. Fig. 6.10 lower panel – FACS plots of the interaction between ΔwelE::TcR cells and moDCs show that agents interfering with DC-SIGN binding actively reduce the interaction between the piliated bacteria and the moDCs. The effect is the most dramatic with the D1 antibody, which reduced moDC binding with almost 30%. On the x-axis: binding percentage of FITC-labeled bacteria to moDCs as measured in the FL1 channel of the FACs. On the y-axis: forward scatter.

152

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

6.4 Discussion

Pili are crucial molecules in host-microbe interactions (Arumugam et al., 2011). A good understanding of their structure is thus key to fully grasp the molecular basis of these specific interactions. Here we report on a second posttranslational modification of these sortase-dependent covalent structures in Gram-positive bacteria, namely their glycosylation. We illustrated via a unique multidisciplinary approach the modification of the heterotrimeric SpaCBA pili of the microbiota isolate and model probiotic L. rhamnosus GG. A combination of nanomechanical, glycobiological and immunological tools resulted in the proof of the modification of these pili with mannose and fucose residues. To strengthen this finding, assays with purified pili were complemented with experiments using bacterial cells. Doing so, we could rule out interference of the purification process (e.g. contamination) and evaluate the importance of the glycosylation of the SpaCBA pili in the presence of other cell wall molecules. A similar gradual build-up of assay-complexity was used in the assessment of the importance of pili glycosylation for the interaction with moDCs. Moreover, the inclusion of solid controls and blocking agents in our assays strengthen the power of our multidisciplinary approach.

The application of SMFS to probe bacterial cell surfaces for glycans was used earlier to detect the cell wall polysaccharides of L. rhamnosus GG (Francius et al., 2009; Lebeer et al., 2009) and we here show that this is also a useful tool to detect glycosylation of pili appendages. Earlier we reported on the peculiar nanospring properties of the SpaCBA pili when probed with an AFM tip functionalized with SpaC antiserum (Tripathi et al., 2013). In this work we detect the same nanospring behavior when probing the pili with lectin-functionalized tips. The fact that similar nanospring behavior was observed with the lectins as with the SpaC specific antiserum can be an indication that the SpaC pilins are the pilins carrying the sugar residues, but further research is still needed to substantiate this hypothesis. Peculiarly, sheared pili tend to aggregate in bundles, the formation of which may be potentially ascribed to lectin-sugar interactions of SpaC pilins (Tripathi et al., 2012b). It was suggested that these SpaC pilins harbor a von Willebrand factor domain with weak homology to a fucose-binding lectin domain (Kankainen et al., 2009). Our report on the presence of fucose residues on the SpaCBA pili underpins the plausibility of the hypotheses described by the studies of Kankainen et al. (Kankainen et al., 2009) and Tripathi et al. (Tripathi et al., 2012b), but further research is needed to proof their validity.

The fact that the SpaC pilin harbors a fucose-binding lectin-like domain, can also support an alternative explanation for our results. It can be the case that this fucose-binding motif strongly binds fucose, thus rendering a positive fucose signal in our assays. This hypothesis is currently being further investigated by growing both piliated and non-piliated L. rhamnosus GG cells in media complemented with fucose. As a read-out, Immuno-EM gold labeling of the fucose structures and SpaA structural pilin is used (P. Rasinkangas, F. Douillard, Prof. W.M.

six.

153

de Vos, UHelsinki). Furthermore, we will investigate the phenotype of a peculiar mutant of L. rhamnosus GG that lacks the genes of the fucose metabolism, but is still piliated. The results of these assays will clarify the plausibility of this alternative hypothesis.

We here present not only the first report on the glycosylation of complex heterotrimeric pili in bacteria, but also present important clues on the importance of its glycosylation for the establishment of specific interactions of L. rhamnosus GG with important immune cells, namely dendritic cells. On the latter cells, the DC-SIGN lectin receptor specifically recognizes fucose- and mannose-MAMPs (Microbe-Associated Molecular Patterns) on microorganisms. A significant interaction between the DCs and purified SpaCBA pili could be illustrated via several assays (DC-SIGN-Fc ELISA, binding to Raji-SIGN cell line and moDCs). All results could be confirmed on whole L. rhamnosus GG and ΔwelE::TcR mutant cells on which the SpaCBA pili are overexposed due to the lack of the thick protective exopolysaccharides layer (Lebeer et al., 2009). Non-piliated ΔspaCBA::TcR cells were used as a negative control and could indeed not interact with the moDCs in a DC-SIGN dependent manner. Moreover the interaction between both purified pili and piliated L. rhamnosus GG cells, and moDCs could be blocked with a DC-SIGN specific antibody, the Lewis X ligand of DC-SIGN and the calcium chelator EGTA, further illustrating the specific glycan-lectin nature of this interaction. Furthermore, we could rule out the involvement of the MR in the interaction between moDCs and L. rhamnosus GG. These results make this one of the few reports on the involvement of bacterial glycoproteins in immune interactions. Strikingly, DC-SIGN signaling depends on the presented MAMPs: fucosylated and mannosylated bacterial molecules elicit distinct immune responses (Gringhuis et al., 2009). It will be intriguing to elucidate the downstream DC-SIGN signaling provoked by binding of the SpaCBA pili, which are both fucosylated and mannosylated. Ongoing experiments include moDC maturation assays and cytokine assays, the latter probing for changes in IL-6, IL-10 and IL-12 mRNA levels. Earlier research established that the DC-SIGN receptor recognizes HIV and is involved in the virus transmission (Geijtenbeek et al., 2000a; Geijtenbeek et al., 2002). As both HIV and L. rhamnosus GG (pili) bind DC-SIGN, the potential of L. rhamnosus GG cells or purified pili to block transmission of HIV by dendritic cells is interesting to pursue.

Taken together we report on the glycosylation of the heterotrimeric adhesive pili of the microbiota isolate L. rhamnosus GG. Furthermore, we could show the importance of the fucosylation and mannosylation of the SpaCBA pili for the interaction of L. rhamnosus GG with the DC-SIGN lectin receptor of human DCs. This work provides new insights in the piliation of bacteria and wants to encourage the study of the potential impact of pili glycosylation in other important piliated microbiota members and bacteria in general. It will be intriguing to see if piliated pathogens, like Streptococcus pyogenes and Enterococcus faecalis, show a similar glycosylation and whether their glycosylation plays a role in immune signaling. Especially the latter pathogen and its close relative E. faecium, both expressing pili that show a striking homology (30-40%) to the SpaCBA pili of L. rhamnosus GG (Kankainen et al.,

154

glyc

osyl

ated

het

erot

rimer

ic p

ili m

edia

te D

C-SI

GN

inte

ract

ion

2009), are interesting targets to study via a similar approach. This homology to pathogenic pili holds promise for the application of glycosylated pili of GRAS species as adjuvants or therapeutic molecules since glycosylation of bacterial molecules has been implicated in the enhancement of the biochemical properties and binding specificities of their substrates.

6.5 aDDitionaL figure

This Coomassie Brilliant Blue stained gel (Acrylamide Bis-Acrylamide gel) shows the protein content of sample A (the purest purified pili sample) and B, which contains less pili material. These two samples result from the pooling of fractions during the purification process (cf. 6.2.4).

Additional figure – Pooled fractions of purified pili result in samples A and B The six fractions with the highest absorbance were pooled to generate sample A, whilst the next six fractions, still containing pure pili, were pooled generating sample B. Both samples were separated on an Acrylamide Bis-Acrylamide gel, which was stained with Coomassie Brilliant Blue stain to visualize the protein content.

six.

155

seven.

general discussion

so, what did we learn?

seven.

general discussion

so, what did we learn?

158

gene

ral d

iscu

ssio

n

7.1 tHe protein gLycosyLation potentiaL of L. rhamnosus ggThe scope of this PhD work was to unravel if, how and why L. rhamnosus GG glycosylates proteins. The discovery of the glycosylation of the cell wall hydrolase and probiotic effector Msp1 set off this study (Lebeer et al., 2012b). Glycoproteins and glycoconjugates in general are key molecules in bacteria-host interactions (Tytgat & Lebeer, 2014). The elucidation of their occurrence in L. rhamnosus GG paves the way for glycoengineering applications and a better knowledge of the molecules underlying microbiota- and probiotica-host interactions.

7.1.1 is msp1 aLone?

7.1.1.1 Glycoproteome screening

We could determine that Msp1 is not alone: L. rhamnosus GG produces a glycoproteome that is vaster and even comprises glycosylated proteins in the cytosol. We were able to identify 40 new potentially glycosylated proteins (cf. chapter 5), next to the confirmation of the glycosylation status of Msp1. Several interesting proteins were found to be glycosylated, of which most are involved in the carbohydrate metabolism. Other hits are proteins involved in protein metabolism, transporters and stress-related proteins (chapter 5). Homologues of several of these proteins were also found to be glycosylated in other species, which corroborates our results. Nevertheless, independent biochemical confirmation of their glycosylation status is needed.

Screening for glycoproteins was performed using 2D electrophoresis as this has already been proven to be a potent proteomics tool, but we combined it with PAS glycostain and lectin affinity screening. Glycosylated hits were challenged vigorously and attributed a glycosylation score to illustrate the level of confidence of their identification. Also, several strict rules were applied during the analysis to minimize the number of false positive hits. Even so, an independent assessment is still needed to unambiguously confirm the glycosylation status of these proteins. A good complementary approach to our 2D gel-based workflow would be a gel-free screening for glycosylated proteins. This would not only lead to the confirmation of the glycosylated proteins here identified, but would also enable the identification of extra glycoproteins which are not accessible by 2D analysis, like proteins present in very low abundance and poorly soluble proteins.

A gel-free approach relies on the enrichment of glycosylated peptides prior to their identification by LC MS/MS (Pan et al., 2011). Two methods are commonly applied to enrich glycopeptides: covalent linkage of glycopeptides to a hydrazine column after which the peptides are enzymatically or chemically released (Ding et al., 2009) and the second method using ion-paired zwitterionic HILIC (Hydrophilic Interaction Liquid Chromatography), but the latter has up till now only been applied for the enrichment of N-glycosylated proteins (Parker et al., 2011). After enrichment, the peptides are then digested using proteolytic enzymes, preparing

seve

n.

159

them for MS identification. Gel-free approaches are extremely challenging, as these methods have not yet been optimized for O-glycosylated bacterial proteins. Moreover this approach would require the isolation of the proteins from their various cellular locations, as we showed that glycoproteins are present in all cellular localizations. It is widely known that isolation of membrane-linked proteins is extremely hard. Even the purification of the Msp1 protein from the secreted protein fraction was not without bottlenecks, making the gel-free confirmation of glycosylated targets a hard endeavor. Extra technological advances are necessary to enable the application of this approach to a wide variety of bacteria. A more direct approach is the purification of glycosylated proteins using dedicated chromatographic techniques, as we did in chapter 5 for the GAPDH glycoprotein. Specific purification of proteins relies on the availability of specific antibodies or techniques, like the blue sepharose column used for the enrichment of GAPDH (chapter 5). Such tools are not readily available for all identified targets.

Enrichment of glycopeptides furthermore could enable the elucidation of the sugar monomeric composition of the attached glycan. Tandem mass spectrometry techniques result in the further breakdown of glycopeptides, rendering specific patterns. The second MS step namely results in the fragmentation of the glycan, resulting in several peaks, which are separated by a mass difference specific for a sugar monomer. The molecular weight of a fucose is for instance 147 Da. Linkage of these mass difference patterns to the known molecular weight of sugar monomers, enables the reconstruction of the glycan attached to a specific glycopeptide (Schiel, 2012). This technique requires a high purity and concentration of the protein sample. Moreover it is crucial that the glycan structure is not affected during the purification process. Combined with the challenging nature of glycopeptide enrichment, these glycomics experiments are hard and require the expertise of specialists in the field, like the group of Prof. Wuhrer (VU Amsterdam).

7.1.1.2 spacba piLi gLycosyLation

In a dedicated analysis, the glycosylation status of the heterotrimeric SpaCBA pili was investigated (chapter 6), in close collaboration with the group of Prof. de Vos (UHelsinki & Wageningen University). These pili complexes are too large to visualize on 2D gels, explaining why we did not identify them in our glycoproteome screening. Moreover, these structures shear off easily, albeit their covalent linkage to the thick peptidoglycan layer. In this context, we previously found that high-speed centrifugation of L. rhamnosus GG (8000 g) already can result in a loss of piliation (Tripathi et al., 2012b), thus explaining the absence of pili after sample preparation. Here we used a unique multidisciplinary approach, which relies on biochemical, nanomechanical (group of Prof. Dufrêne, UCL) and immunological (Prof. Geijtenbeek, AMC) techniques to show the modification of the SpaCBA pili of L. rhamnosus GG with mannose and fucose residues (chapter 6). This conclusion was based on all the results of the sampling of both purified pili and piliated strains of L. rhamnosus GG with a large array

160

gene

ral d

iscu

ssio

n

of techniques. We believe that this first report on the glycosylation of complex pili in bacteria will boost the study of other piliated species.

Nevertheless, definite mass spectrometric evidence is lacking. During a research stay in the group of Prof. Wuhrer (Analytical Chemistry group, VU Amsterdam), with whom we earlier worked together on the Msp1 glycosylation (Lebeer et al., 2012b), an attempt was made to check the glycans present on the pili. Unfortunately, the complexity of the pili hampered these attempts. Moreover, the low concentration in which bacterial pili can be purified is below the detection limit of the equipment commonly used for glycan and glycopeptide mass spectrometry. A further optimization of the pili purification process is needed to improve the yield of this process. An attempt to do so was made in the context of the PhD project of Marijke Segers, starting from 10 L L. rhamnosus GG cultures. Unfortunately we were unable to optimize this protocol during the time frame of her PhD, and up till now this did not result in a significant amount of purified pili. Another way in which the glycosylation status of the pili could be confirmed is their recombinant expression in a non-glycosylating host together with the GTs that are likely to modify these pili. This then should result in the heterologous production of glycosylated SpaCBA pili, similarly to the experiments performed by Wacker et al. (Wacker et al., 2002). However, we were not yet able to identify the candidate GTs that may be involved in pili modification (see also chapter 3). Another complementary method is the application of deglycosylating chemicals on purified protein samples. These assays were performed earlier to substantiate the glycosylation of Msp1 and rely on the usage of trifluoromethane sulfonic acid (Lebeer et al., 2012b). This is a harsh method and often also results in protein degradation. But this technique is likely to be improved, at the moment the potential of special metal complexes to hydrolyze O-glycosidic linkages are being investigated (personal communication).

Thus, albeit the high number of assays suggesting SpaCBA pili glycosylation, the evidence is mostly indirect. At the moment we thus cannot completely rule out that the glycosylation signal can originate from the strong binding of fucose by the SpaC pilin. This pilin harbors a fucose lectin-like domain (Kankainen et al., 2009), which would be an alternative explanation for the fucose-specific reactions we observe. Currently, we are trying to rule out this hypothesis via assays in which L. rhamnosus GG is grown in the presence of fucose, followed by the evaluation of the presence of a fucose-signal.

7.1.1.3 wHat about msp1?

Although we were able to elucidate that Msp1 is a glycoprotein, identify a glycosylation site, get some information on the attached glycan and establish the importance of its glycosylation for the stability of the protein (Lebeer et al., 2012b), still many open questions remained. During this PhD were able to show that the earlier discovered TVETPSSA glycosylation site (Lebeer et al., 2012b), is not the only glycosylated stretch in this protein. An msp1 mutant complemented with a vector carrying the Msp1 protein with a point mutation in the serines

seve

n.

161

of the glycosylation motif (CMPG10805) still was glycosylated (unpublished results, cf. addendum).

The quest for the GTs targeting Msp1, and now also of its homologues Msp1 and LGG_02016 is also still on. An important step in this search will be the experimental validation of the protein GTs identified by the in silico method (chapter 3) and the assays described further to confirm their GT activity and substrate specificity (cf. addendum).

This experimental validation, by both knockout mutagenesis and enzymatic analyses of recombinantly expressed and purified GTs is not straightforward (see below). We hypothesized that the glycosylation process of Msp1 is likely to involve a membrane step. Still we first focused on the construction of various knockout mutants in predicted cytosolic GTs (cf. addendum). This, because we realize that enzymatic assays with recombinantly expressed membrane-bound GTs, especially in multienzyme complexes, will be highly challenging. The L. rhamnosus GG GT knockout mutants that already could be constructed (cf. addendum) were screened for shifts in molecular weight of Msp1 (which is the major protein present in the supernatant). Unfortunately, none of these assays resulted in a clear difference in Msp1 glycosylation (cf. addendum). This can most likely be partially contributed to GT redundancy. Furthermore, various GT knockout mutants could not be constructed up till now, presumably because of their essential functions in L. rhamnosus GG. Therefore the search for the GTs targeting Msp1 is still open.

7.1.2 How Does L. rhamnosus gg gLycosyLates its proteins?

7.1.2.1 The quest for glycosyltransferases

The observation that Msp1 is glycosylated immediately gave rise to the question on how L. rhamnosus GG can glycosylate proteins. This question became even more pertinent now, since we detected several other glycoproteins in L. rhamnosus GG. A good way to investigate the glycosylation mechanism used by a bacterium to modify its substrates is to look at the GTs. As mentioned before, these are the key enzymes of glycosylation events. GTs are involved in the biosynthesis and attachment of glycans to other molecules, thus determining the glycome (chapter 2, (Lairson et al., 2008; Tytgat & Lebeer, 2014)). However, the factors determining the substrate specificity of these enzymes remain enigmatic, while the promiscuity (i.e. targeting of multiple substrates) of GTs hampers their substrate identification further (Tytgat & Lebeer, 2014; Weerapana & Imperiali, 2006). In many bacteria, the annotation of these enzymes is flakey: in L. rhamnosus GG there was for example even a GT attributed to the biosynthesis of LPS (LGG_00999) in the annotation (Kankainen et al., 2009). This does not make any sense for a Gram-positive bacterium and probably resulted from an amplification of annotation errors based on strong conservation.

We therefore designed a method that could identify genes potentially encoding GTs in bacterial species, without relying on annotation-related input. The in silico method relies on all currently

162

gene

ral d

iscu

ssio

n

available information on bacterial GTs and was tested by application both to the genomes of L. rhamnosus GG and C. jejuni, the latter being a model organism of prokaryotic glycosylation. In a second step we then looked at the functional network of the selected genes to get clues on their substrate specificity. This workflow enabled us to identify 48 GTs, of which 8 were never identified as such earlier in L. rhamnosus GG (chapter 3). Linking these genes to their potential substrate resulted in the prediction of the cluster that may be responsible for the biosynthesis of the glucose-rich polysaccharides of L. rhamnosus GG, namely the LGG_00278-LGG_00283 region (Francius et al., 2008). Together with technician Tine Verhoeven, who is an expert in the production of L. rhamnosus GG double homologous knockout mutants, several attempts were made to construct knockout mutants in this region. These attempts were so far –unfortunately– unsuccessful (cf. addendum). This may suggest that these glucose-rich polysaccharides and/or GTs perform an essential function in L. rhamnosus GG.

Furthermore, we could identify new GTs (17/44) in the well-studied bacterium C. jejuni, of which several might be involved in flaggellin glycosylation (chapter 3). This is interesting because the large region responsible for flagellin production and modification is known for quite some time now, but up till now the GTs responsible for its glycosylation were not identified (Nothaft & Szymanski, 2010). This is of special interest as the flagella of C. jejuni are important virulence factors. These results make this method an ideal tool to boost research on the glycosylation potential of ‘new’ species and to reassess species of which the glycome is already (partially) known (cf. C. jejuni results).

The list of genes potentially encoding GTs in L. rhamnosus GG combined with the information on their likely substrates provides a valuable list of targets for further experimental confirmation. This list does not render information on the precise biochemical mechanism of the GTs, but provides clues on the molecule classes that they might target. We attempted to underpin this with experimental evidence by knocking out GT encoding genes, which should then result in a shift in the molecular weight of its substrate resulting from the loss of glycosylation. Several GT knockout mutants were already constructed, with a focus on the ones predicted to target proteins (cf. addendum for a complete list), but none showed a clear (glyco)proteomic phenotype (results not shown). Probably this is partially due to the fact that GTs are often redundant and promiscuous, meaning that the role of a knocked out GT can be taken over by another one (chapter 2, (Tytgat & Lebeer, 2014)). The production of double and triple mutants might bypass this feature of GTs, but their construction seems to be even more challenging than the production of a single knockout mutant.

To merely confirm the actual GT activity of a GT, are enzymatic approaches. As GTs require sugars activated by a nucleoside phosphate (chapter 2), GT activity can be measured following the release of nucleoside phosphates using a phosphatase (Wu et al., 2011). Another strategy that was conceived in collaboration with Baptiste Leroy and Prof. Ruddy Wattiez (UMons) is to recombinantly express the GTs and purify them. The purified GTs could then be mixed with

seve

n.

163

crude bacterial extracts as a source of substrates and UDP, GDP or CMP activated sugars as substrates. This enables the stepwise fractionation and isolation of proteins with GT activity, which then can be identified by MS. The resulting list of GTs could then be compared to the predicted list of GTs, thus confirming their GT activity. Moreover, recombinantly expressed, thus non-glycosylated, substrates could be added to the GTs instead of the crude bacterial extracts to specifically mine for GTs targeting this particular substrate. This assay holds great promise, but its development is not without its challenges, as it will be difficult to maintain the enzymatic activity of the GTs along the fractionation procedure. Moreover, the recombinant expression of membrane-linked GTs will not be straightforward. However, the fact that we could identify several glycoproteins in the cytosol renders this a potentially valuable approach to study cytosolic GTs. Nevertheless, another important caveat is that enzymatic assays do not necessarily represent the cellular outcome.

Taken all these factors together, the experimental validation of GTs remains challenging. The experimental bottlenecks also explain why we were unable to pinpoint (a) GT(s) as the one(s) targeting Msp1, the SpaCBA pili and the other glycoproteins we identified in L. rhamnosus GG up till now.

7.1.2.2 mecHanistic HypotHeses

What we do know from the work on Msp1, is that the protein is O-glycosylated (Lebeer et al., 2012b). It is therefore very likely that the other identified glycoproteins of L. rhamnosus GG are also O-glycosylated. N-glycosylation seems to be a phenomenon exclusive to Gram-negative species (chapter 2, (Tytgat & Lebeer, 2014)), making it all the more likely that L. rhamnosus GG only harbors a protein O-glycosylation machinery. This hypothesis nevertheless needs further substantiation with experimental evidence, similarly as to the assays we performed for Msp1 in collaboration with Prof. Wuhrer (VU Amsterdam).

It is yet hard to give a conclusive answer on the question if L. rhamnosus GG uses a sequential or en bloc mechanism to glycosylate its proteins (chapter 2). Previous research has shown that the long, galactose-rich EPS are built in an en bloc manner. The EPS gene cluster namely harbors a priming GT, GTs to synthesize the glycan further and a polymerase to link all glycans together to form the final EPS structure (Lebeer et al., 2009). Potentially, L. rhamnosus GG uses a similar mechanism to target its proteins, as previously shown for e.g. the fucose-containing glycoproteins of B. fragilis (Fletcher et al., 2009).

On the other hand, a sequential glycosylation mechanism is also not unconceivable. Our in silico analysis showed that many GTs (20/48) of L. rhamnosus GG are not clustered in genomic regions, but are rather scattered throughout the genome as isolated genes flanked by transposable elements, suggesting their acquisition by horizontal gene transfer. And even if the GTs occur in clusters, these are generally smaller (mean size of 3 GTs) compared to the clusters occurring in C. jejuni (2-5 GTs) (chapter 3, (Sanchez-Rodriguez et al., 2014)).

164

gene

ral d

iscu

ssio

n

These findings would corroborate the presence of sequential mechanisms. Which of the two mechanisms or if L. rhamnosus GG utilizes both to produce glycosylated proteins remains to be determined.

7.1.2.3 export of gLycosyLateD proteins

In Streptococcus species, it was evidenced that glycosylated proteins are transported by an alternative Sec system, namely the SecA2 mechanism (as nicely reviewed in (Feltcher & Braunstein, 2012)). This accessory SecA2 mechanism harbors both proteins involved in glycosylation and transport of glycosylated substrates. Genome mining performed by François Douillard and Prof. Willem M. de Vos (UHelsinki & Wageningen University) resulted in the identification of a homologue of SecA2 in L. rhamnosus GG, which was called SecX (LGG_01063). Strikingly, this homologue only contains the specificity domain and not the motor domain that normally also is present in Sec proteins. So we are currently investigating if this SecX accessory transport system is actually functional in L. rhamnosus GG and if it is involved in the export of glycosylated proteins. We were able to construct a double homologous knockout mutant of the secX gene, which is currently being investigated. Based on their strong homology to the accessory GTs of the Streptococcus system, the involvement of the yohH-yohJ encoded GTs in this system is evaluated. L. rhamnosus GG harbors three homologues of these GT genes (LGG_00348-349, LGG_1586-87 and LGG_02284-85). Knockout mutants of the latter GT clusters were already successfully constructed (cf. addendum), but despite numerous efforts the knockout of the first homologous enzymes (LGG_00348-349) did not succeed up till now. The GT nature of these genes is further corroborated by their prediction as being GTs in our in silico analysis (chapter 3) and the fact that homologues of these enzymes are responsible for the Msp1 homologue Acm2 of L. plantarum WCFS1 (Lee et al., 2014). Potentially, these GTs are thus involved in the glycosylation of glycosylated proteins like Msp1. Nevertheless, the involvement of other GTs cannot be ruled out at the moment and also other GTs should be considered.

Currently we are in the process of assessing the phenotype and (glyco)proteome of these secX and yohH-yohJ mutants to validate their actual involvement in glycoprotein secretion in L. rhamnosus GG. Furthermore, we are constructing double and triple mutants targeting these genes of interest, to rule out redundancy and promiscuity of GTs and secretion systems. But most importantly, it remains to be elucidated if SecX is functional at all.

7.1.3 wHy Does L. rhamnosus gg botHer to gLycosyLate its proteins?

Glycosylation is an energy-costly process for a bacterium, so there must be an evolutionary benefit for the bacterium. Glycosylation in general appears to be key in bacterial host interactions (Kay et al., 2010). In pathogenic bacteria, in which protein glycosylation is already better studied, virulent proteins are often glycosylated. Strategies include the glycosylation of immunogenic molecules to remain stealth from the environment and host immune system

seve

n.

165

or in contrast to enhance the interaction with the immune system. But probably the best-documented function of protein glycosylation is the modulation of the biochemical properties of the targeted substrate. Glycosylation can enhance for instance activity of proteins and protect them against degradation by proteases (Tytgat & Lebeer, 2014).

The enormous diversity of glycans that can be produced by bacteria enlarges the glycan-repertoire immensely (Marino et al., 2010), enabling bacteria to establish very specific interactions with the environment. Often these glycans are even species-specific, functioning as a kind of ‘barcode’, for instance by interacting with specific immune lectins (Tytgat & Lebeer, 2014). Nevertheless, this specific barcoding function has not yet been widely documented for bacterial glycoproteins. For example, a role for N-glycosylated proteins and distinct lipooligosaccharides glycoforms in C. jejuni in interactions with the immune lectin MGL was documented (van Sorge et al., 2009), as well as binding of a mycobacterial cell wall glycoprotein SodC to Langerin (Kim et al., 2015).

In bacteria, the role of glycans in the interaction with the environment and host has mainly been studied in a dedicated way, i.e. by unraveling the role of a specific glycosylated protein for the interaction with a specific partner (cf. chapter 2 and (Tytgat & Lebeer, 2014)). The Consortium for Functional Glycomics (www.functionalglycomics.org) provides a platform where glycomics tools are bundled and can be consulted free of charge. Their services encompass glycan arrays, glycan profiling and gene microarrays. Up till now, most provided assays are dedicated to the study of eukaryotic glycosylation, but can sometimes be exploited to study the interaction of bacterial glycoproteins with mammalian tissues. It can be expected, with the current booming interest in bacterial glycoconjugates, that these services will be further expanded to enable functional high throughput screenings of prokaryotes too. The work of Hsu and co-workers is an example of a lectin microarray used to fingerprint the glycosylation of E. coli (Hsu et al., 2006).

7.1.3.1 tHe spacba piLi interact witH tHe immune Lectin receptor Dc-sign

In this PhD work, we discovered that the adhesive SpaCBA pili of L. rhamnosus GG are glycosylated (chapter 6). The fact that these heterotrimeric pili are modified with mannose and fucose moieties, made us wonder if they could interact with the immune lectin receptor DC-SIGN. This receptor present on human DCs specifically recognizes mannose and fucose residues. An initial ELISA test confirmed our hypothesis. Further tests relying both on the use of purified pili and (non-)piliated L. rhamnosus GG bacteria, and both on cell lines expressing DC-SIGN and monocyte-derived DCs were carried out during a research stay in the group of Prof. Geijtenbeek (AMC, UvA Amsterdam). These assays confirmed that the glycosylated SpaCBA pili are key for the interaction between L. rhamnosus GG and DCs (chapter 6). It will be intriguing to investigate if this is also true for other glycosylated proteins of L. rhamnosus GG, like Msp1.

166

gene

ral d

iscu

ssio

n

The downstream effect of this interaction is currently being further investigated. Preliminary results show that the pili influence the maturation of moDCs and potentially also affect cytokine expression (results not shown). Further corroboration of these results is needed to show that the L. rhamnosus GG-moDC interaction is of functional importance. This would make this one of the few reports on the role of glycans in the establishment of specific immunogenic interactions between bacteria and their host (manuscript in preparation).

7.1.3.2 gLycosyLation of ceLL waLL HyDroLases

In this work, we showed that the close homologues of Msp1, namely Msp2 and LGG_02016, are also glycosylated (chapter 5). Previous work showed that the glycosylation of Msp1 infers a larger stability to the protein compared to its close homologue Msp2, which was long thought to be unglycosylated (Lebeer et al., 2012b). In this work we showed in several independent assays that Msp2 is indeed glycosylated, but probably to a more minor extend than Msp1, explaining its late discovery (chapter 5). The modulation of the stability of Msp1 is a first functional clue why L. rhamnosus GG would glycosylate its proteins. This finding is in accordance to the role protein glycosylation plays in C. jejuni, where glycosylation protects the proteins against degradation by gut proteases (Alemka et al., 2013). This is further corroborated by findings in other gut bacteria, like B. fragilis (Fletcher et al., 2009), L. plantarum WCFS1 (Fredriksen et al., 2012) and E. coli 2787 (Charbonneau et al., 2007).

The fact that several homologues of these cell wall hydrolases are glycosylated in other closely related species raises the question if the glycosylation of these cell wall hydrolases is a common trait in lactobacilli. Research on the glycosylation of the Acm2 autolysin of L. plantarum WCFS1 showed that glycosylation acts as a negative regulator of its enzyme function (Fredriksen et al., 2012; Rolain et al., 2013). In view of the potentially ‘dangerous’ autolytic function of these enzymes it would make sense for the bacteria to fine-tune their activity by glycosylation. Indeed, earlier work showed that Msp2 is more active than its (heavier) glycosylated Msp1 homologue (Lebeer et al., 2012b). If this is indeed the case and if this is a feature shared by all lactobacilli or even by more distantly related species remains to be further substantiated.

A third plausible explanation for the glycosylation of Msp1 and Msp2 is the fact that glycosylation might be used by bacteria as a switch between the canonical role of an enzyme and its moonlighting function. Apart from being cell wall hydrolases, Msp1 and Msp2 are probiotic effectors that can modulate the survival of intestinal epithelial cells (Yan et al., 2007). It is conceivable, that in parallel to the well-known phosphorylation cycles used by organisms to switch between activity states, glycosylation is here used as a control mechanism between their enzymatic activities. This is corroborated by the finding of more glycosylated moonlighting proteins in our screening of the L. rhamnosus GG glycoproteome (chapter 5).

seve

n.

167

7.2 tecHnoLogicaL impact on bacteriaL gLycobioLogy

As illustrated, the field of bacterial glycobiology is challenging to study. The posttranslational and thus non-template driven character of these modifications inflicts several experimental difficulties (chapter 2, (Tytgat & Lebeer, 2014)). In the course of this PhD work we tackled some of these challenges by proposing new workflows and enhancing some existing protocols.

7.2.1 systematic approacHes to investigate tHe gLycosyLation potentiaL of bacteria

The field of bacterial protein glycomics is still rather young and glycoproteins have been identified only in a limited amount of species (chapter 2, (Tytgat & Lebeer, 2014)). To study the glycosylation potential of the model probiotic and microbiota isolate L. rhamnosus GG we relied on two systematic approaches: a first one to predict genes encoding GTs and a second one to screen the proteome for glycoproteins. In both cases we designed a new, robust workflow that can easily be applied (with some minor adaptations) to study new species.

7.2.1.1 annotation of gLycosyLtransferases anD tHeir substrate cLasses

The annotation of GTs in bacteria is often unreliable. This information is mostly inferred from data and gene annotations in other species, which leads to the accumulation of erroneous annotations across species (chapter 2, (Tytgat & Lebeer, 2014)). Thanks to a stimulating collaboration with the bio-informaticians of the group of Prof. K. Marchal (KU Leuven & UGhent) combined with our biological input on bacterial (protein) glycosylation, we were able to propose a new method to circumvent further propagation of errors. This sequenced-based method relies on all experimental validated data available about GTs, which is mostly bundled in the CAZy database and was further manually extended with all available literature on peculiar GTs and glycosylation mechanisms in bacteria (chapter 2 and 3). This workflow enables one to screen a bacterial genome for all genes potentially encoding GTs in a robust way.

Finding a link between a GT and its substrate has been proven to be extremely hard. There are currently no motifs described that link the two. The promiscuity of GTs moreover hampers the identification of substrate motifs (chapter 2, (Tytgat & Lebeer, 2014)). Currently, the ‘best’ way to identify the specificity of a GT was by extrapolation of information from clustered auxiliary enzymes and its potential substrate. But as seen in L. rhamnosus GG (chapter 3, (Sanchez-Rodriguez et al., 2014)) GTs can also occur isolated in the genome, rendering this approach not useful. To improve this, we start from the local neighborhood of a predicted GT. The STRING database, gathering all known information on protein-protein interactions (Szklarczyk et al., 2011), was mined to find functional partners of the identified GTs. This results in the drawing up of consensus networks, which enable the linkage of GTs to their potential target molecules. The more information (e.g. microarrays, RNA seq, proteomics) present in STRING, the better the output of the analysis. For instance, we were able to infer

168

gene

ral d

iscu

ssio

n

more substrate relations for the better-studied C. jejuni (30/44) than for L. rhamnosus GG (20/48) (chapter 3, (Sanchez-Rodriguez et al., 2014). Taken together, our analysis flow improves the attribution of plausible GT substrates largely, especially for isolated GTs.

Moreover, the current analysis was performed using all current information on bacterial GTs. As this field is still rather young and rapidly expanding, it will be of interest to redo the analysis in a couple of years. This will most probably result in better-substantiated predictions and might lead to the identification of extra GTs. Also the linkage of the predicted GTs to their substrate largely relies on the amount of data present in the STRING database. This is an important factor as illustrated for C. jejuni, in which more GTs could be linked to their potential substrate compared to L. rhamnosus GG. Initially, when we conceived the study, there was no data present in the STRING database for L. rhamnosus GG. Microarray data en other large interaction datasets were included upon our request by other groups, illustrating that the current information in the STRING database on L. rhamnosus GG is still limited. In a few years, based on the extra information on the interactome that will be generated by then, the substrate-relation inference step of our analysis could be largely improved. This might lead to better predictions of substrate specificities of the GTs, like the prediction of GTs potentially targeting Msp1 and other known glycoproteins of L. rhamnosus GG.

7.2.1.2 Mapping of bacterial glycoproteomes

The study of the glycoproteome of L. rhamnosus GG also led to the design of a broadly applicable workflow. To infer data on the localization of glycoproteins, a protocol was conceived that enables the separation of cytosolic, cell wall-associated, cell wall/membrane and supernatant proteins, resulting in a unique extra layer of information. Glycosylated proteins were then identified using a combination of both Periodic Acid Schiff base glycostain and lectin probing applied to the 2D separated glycoproteome fractions. Robustness of the workflow can be ensured by the performance of an independent repeat of the experiment, independent lectin probing experiments and the inference of a set of strict rules to select glycan-positive spots. MS identification then leads to a list of potentially glycosylated proteins in a species. This workflow can be easily applied to other species. The main caveat is to carefully select the lectin used for probing of the fractionated glycoproteomes, as this is likely to be species-specific (chapter 5).

As mentioned before, we strongly recommend the combination of this workflow with gel-free approaches. This would not only serve as an extra confirmation of the identified targets, but can also increase the number of identified glycoproteins, as insoluble or very large proteins are hard to detect in the here-proposed workflow. Moreover, the glycosylation of the here-identified hits, especially the putative ones, needs to be further experimentally substantiated. Purification and assessment of their glycosylation status using MS techniques are only few of the techniques that can be used to do so (see earlier).

seve

n.

169

7.2.2 caveats wHen performing Lectin probing

Already for several decades, lectin probing of Western blotted proteins is used by many researchers to detect glycosylated proteins (chapter 2, (Tytgat & Lebeer, 2014)). Also in this PhD work, lectin probing was one of the key experiments to identify and confirm the glycosylation of proteins. In this context we have a long-standing collaboration with lectin expert Prof. Els Van Damme (UGhent). She provided us with an array of lectins with different specificities, which were applied in several lectin screenings. Lectin probing proved to be extremely valuable both in dedicated approaches (cf. chapter 6 on pili glycosylation) and general open screening approaches (chapter 5). Albeit this technique has been universally used to detect glycoproteins, our experimental endeavors resulted in two important updates that can improve the reliability and resolution of lectin probing.

A first fact is that lectin probed blots often show a lot of background noise, meaning that the blot itself tends to color too. Generally, Western blots are blocked using a BSA solution or a solution of milk powder. Albeit this does not infer any problems for antibody-based detection of proteins, it does hamper the detection of glycoproteins. Both substances contain or are contaminated with glycosylated proteins, resulting in a high background signal when probing blots with lectin probes. We therefore, based on results of the group of Thompson et al. who reported this for ELISAs, recommend the use of the chemical polyvinylalcohol (PVA) instead of BSA or milk powder to reduce background noise (chapter 4, 5, 6, (Thompson et al., 2011)).

Most lectins used for glycan detection are linked to a biotin-molecule, enabling their detection with streptavidin. This resulted in L. rhamnosus GG in a recurring strong reactive band of 125 kDa with when screened with a plethora of lectins. However, this protein could not be purified using a lectin-affinity column. Ultimately, identification revealed that this band resulted from a pyruvate carboxylase, which are known to bind endogenous biotin. Probing of a Western blot with only a secondary streptavidin antibody, revealed the presence of more biotin-binding proteins, causing false positive results. We therefore recommend the glycobiology community to either include such a streptavidin-probed control blot or to shift to a digoxigenin-based detection system. This finding is not only valid for lectin probing, but for all biotin-streptavidin based probing systems (chapter 4, (Tytgat et al., 2014)). Especially for open screening approaches using lectins, as we performed in chapter 5, it is of high importance to take these potential false positives into account.

7.2.3 muLtiDiscipLinarity is key in gLycobioLogy

The challenging nature of bacterial glycobiology often demands a multidisciplinary approach. In this work collaborations were sought with specialists in glycan (Prof. Wuhrer, VUA) and protein mass spectrometry (Prof. Wattiez and B. Leroy, UMons), bioinformatics (Prof. Marchal, UGhent), pili (Prof. de Vos, UHelsinki & Wageningen University and F. Douillard), immunology (Prof. Geijtenbeek, AMC UvA), lectin (Prof. Van Damme, UGhent)… We

170

gene

ral d

iscu

ssio

n

strongly believe that the gathering of scientists with different backgrounds is essential to push new findings in this challenging field (chapter 2, (Tytgat & Lebeer, 2014)). These multidisciplinary collaborations were key in this PhD work and several research stays in other groups (e.g. 3 months in the lab of Prof. Geijtenbeek, AMC UvA and 2 short stays in the group of Prof. de Vos, UHelsinki) were crucial to obtain the here-presented results

A quite unique approach presented in this work, is the collaboration with the AFM-specialists of the group of Prof. Dufrêne (UCL). Previously, our lab has worked closely together with this group on the elucidation of the nanomechnical properties of the SpaCBA pili of L. rhamnosus GG (Francius et al., 2008; Lebeer et al., 2009; Sullan et al., 2014; Tripathi et al., 2013; Tripathi et al., 2012a; Tripathi et al., 2012b). The results were intriguing, as they could show that the pili act as a nanospring and support the adhesion of L. rhamnosus GG to its environment in a zipper-like manner. In chapter 6, we propose a new method to screen for glycosylated surface structures on bacteria, by probing live cells with AFM-tips modified with lectins. This led to the identification of fucose and mannose residues present on the SpaCBA pili of L. rhamnosus GG and provides a novel method to also screen other (piliated) bacteria for surface glycosylation (chapter 6). As shown before, cell wall polysaccharides can be detected in a similar method (Francius et al., 2009; Francius et al., 2008; Lebeer et al., 2009). We believe that single-molecule force spectroscopy combined with lectins provides a powerful method to screen surface glycosylation on intact bacterial cells.

7.3 new insigHts in bacteriaL gLycobioLogy

The results of this PhD also generated some new insights and hypotheses potentially relevant for a wider range of species or even to the broader field of bacterial glycobiology. These new ideas mainly resulted from the two systematic assays we performed (chapter 3 and 5). It will be interesting to see if these results and hypotheses can be further substantiated in the next coming years.

7.3.1 gLycosyLation as a reguLator of muLtienzyme compLexes

As mentioned earlier, research on the functional importance of protein glycosylation in particular has been lagging behind (Tytgat & Lebeer, 2014). Results from both our systematic analyses (chapter 3 and 5) now indicated that glycosylation of proteins might play a more important role than currently is appreciated. Both analyses showed that protein glycosylation may be involved in the regulation of multienzyme complexes.

In the course of the unraveling of the substrate of potential GTs, we found indications that some GTs are promiscuous and can be linked both to peptidoglycan biosynthesis and protein glycosylation. The proteins they target are moreover members of the cell division machinery (chapter 3). Results from our glycoproteome screening corroborated these findings, with the identification of three cell wall hydrolases as being glycosylated and the potential glycosylation

seve

n.

171

of the MurE PG biosynthesis enzyme (chapter 5). This hypothesis was underpinned further by similar reports on the glycosylation of members of the cell division machinery in B. fragilis (Fletcher et al., 2009) and L. plantarum WCFS1 (Fredriksen et al., 2013).

The glycoproteome screening of L. rhamnosus GG furthermore resulted in the identification of several proteins related to protein synthesis (chapter 5). More in particular, structural 30S and 50S ribosomal proteins were shown to be glycosylated, together with the elongation factor FusA. Of the latter, the glycosylation status was also evidenced in C. jejuni (Young et al., 2002). The ribosome represents another crucial multienzyme complex of the bacterial cell.

Taken together this might indicate that bacteria use protein glycosylation as an extra level of regulation of their crucial multienzyme complexes. The timing, assembly and stability of both the cell division and translation machinery is essential for the wellbeing of the cell, making their tight regulation key.

The experimental substantiation of the potential role of glycosylation as an extra regulation level of multienzyme complexes is however challenging. Purification of (membrane-associated) multienzyme complexes is cumbersome. For instance, some of them are only present in an assembled manner during specific moments during cell growth.

7.3.2 reguLation of enzymatic activity by gLycosyLation

A striking finding of the glycoproteome screening is the glycosylation of several moonlighting enzymes. Three moonlighting enzymes of the glycolysis (GapA, Eno and Pgk) were found to be glycosylated (chapter 5). Earlier we also discussed the potential implications of the glycosylation of the multifunctional cell wall hydrolases Msp1 and Msp2 (chapter 5). A possible explanation is that the cells might use glycosylation of these enzymes as a switch between their canonical role and their moonlighting function. The canonical role of all these enzymes is mostly situated in the cytosol or linked to the cell wall. Maybe the attachment of a glycan targets them for secretion, relocating them to the extracellular side of the bacterium, where they can exert their moonlighting role in the adhesion and modulation of the environment.

Another peculiar finding of chapter 5, is the high number of glycosylated carbohydrate-enzymes. A high number of enzymes involved in the glycolysis were identified as glycoproteins with high confidence. It is likely that their glycosylation functions as a kind of self-regulation mechanism. For instance, a top hit was the glycosylation of the L-lactate dehydrogenase encoded by LGG_02523 (chapter 5). This corroborates the observation that lactobacilli tend to regulate the activity of L-lactate dehydrogenases by different mechanisms (Feldman-Salit et al., 2013). Our results suggest that glycosylation can be added to this list.

7.4 generaL concLusions

In conclusion we give a short overview of the most important findings of this PhD work.

172

gene

ral d

iscu

ssio

n

• Design of a robust workflow that combines sequential and network-based computational predictions to unveil insights in the glycosylation potential of a bacterium. This unique method enables the identification of GT-encoding genes and moreover links them to their potential substrate molecules.

• Optimization of experimental techniques necessary to study glycoproteins.

• Design of a robust workflow to screen the functionally fractionated proteome of a bacterium for glycoproteins.

• Discovery of the glycosylation of the complex heterotrimeric SpaCBA pili of L. rhamnosus GG.

• Proof that the glycosylation of these pili is important for the interaction with the host immune system, making this one of the few detailed reports on the immunological importance of glycosylated structures.

• Generation of important new insights and hypotheses on protein glycosylation in bacteria and especially on the potential functional roles of glycosylation of proteins.

In general we believe that our results have enhanced the knowledge on glycoproteins as important molecules underlying the specific interactions between probiotics and microbiota members and their host. Moreover, our work has supported the bacterial glycobiology field in general, both by the design of new techniques and the proposition of important new hypotheses. However, the experimental validation of the here-proposed hypotheses promises to be challenging.

7.5 Lessons for gLycobioLogy

• Beware of false positive results: revise methods carefully to avoid them. It is advisable to include the appropriate controls to check for false positive and negative results. In the case of lectin blotting, the digoxigenin detection method is preferable over the biotin-streptavidin method.

• Multidisciplinarity is key. The non-template driven character of glycobiology demands the joining of investigative forces from a variety of disciplines. As illustrated in this work, the acquisition of results relies on the collaboration with mass spectrometrists, bio-informaticians, immunologists and many other experts.

• Glycosylation might play a role in the regulation of multienzyme complexes. Tight regulation of the assembly and activity of these complexes is crucial for the well being of the cell. We hypothesize on the role of protein glycosylation in the regulation of the cell division machinery and ribosome.

• Glycosylation might regulate the switch between the canonical and moonlighting function

seve

n.

173

of a protein. This hypothesis is corroborated by the glycosylation of several proteins with a reported moonlighting role such as the GAPDH enzyme which is both a glycolytic enzyme and is implicated to bind to molecules in the environment.

• Heterotrimeric pili structures can be glycosylated. Up till now only simple, monomeric pili have been reported to be glycosylated. Our results open new avenues in the study of the glycosylation of piliated species. Moreover, this pili glycosylation can be crucial for the establishment of specific immune reactions.

7.6 perspectives

Many fundamental mechanical questions concerning bacterial protein glycosylation remain unanswered. However, the importance of glycoproteins and of glycoconjugates in general in host-bacteria interactions becomes more and more apparent (Kay et al., 2010; Tytgat & Lebeer, 2014). Research on bacterial glycoproteins is a hot topic as the protein glycosylation potential of more and more species is studied. The past decades, most research focused on glycoproteins in pathogens, in view of their role in virulence, but nowadays more and more reports on the occurrence of glycoproteins in beneficial commensal bacteria are emerging. However, research on the functional importance of glycoproteins, and of glycoconjugates in general, is still lagging behind. The role of these glycosylated structures is likely to be even more important than currently appreciated, cf. the potential role of glycosylation in the regulation of multienzyme complexes. It can be envisaged that these gaps in knowledge will be gradually filled the next coming years, especially when taken into account that the field of bacterial protein glycosylation emerged only a few decades ago. The current knowledge only scratches the surface of what there is to know on the occurrence and function of bacterial glycoproteins. The future of bacterial glycobiology is promising and it will be fascinating to see which species are able to produce these structures and what their evolutionary motif is to do so.

As mentioned, glycosylation of proteins is of importance for the modulation of the biochemical properties of their substrates and to establish specific interactions with the environment. These features can be exploited by the attachment of glycans to specific molecules thus increasing their stability or activity and their interaction potential. Glycoengineering of molecules holds great promises for the production of therapeutics and bioactive molecules with enhanced properties (Werner et al., 2007). The basis for this glycoengineering field was formed by a key publication by Wacker and co-workers, showing the heterologous production of a glycoprotein of C. jejuni by transfer of the glycosylation machinery into E. coli (Wacker et al., 2002). This technology is already widely employed for the production of medical therapeutics in bacteria (Elliott et al., 2003; Farahmand et al., 2011; Iwashkiw et al., 2012a; Lindhout et al., 2011; Lizak et al., 2011a; Saxon & Bertozzi, 2001). Bacteria are the ideal production system for glycosylated agents, as they can be grown easily, production is cheaper, they can be genetically

174

gene

ral d

iscu

ssio

n

modified and the process can be accelerated compared to production in eukaryotes (Pandhal & Wright, 2010). Currently, much research is done on the production of glycoengineered human therapeutics in bacteria, like insulin (Jain et al., 2003). Also the production of vaccines both for humans (Glycovaxyn, Switzerland) as for animals (Group of Prof. Szymanski, Canada) is being pursued (Ihssen et al., 2010).

Most glycoengineering assays are based on the transfer of the glycosylation machinery of a pathogen and its target to E. coli, in which the glycan is further tailored to generate the therapeutic molecule of interest. But in the future it would be even more interesting to investigate the potential of new bacterial expression system for the production of glycosylated therapeutics that would also enable the in situ delivery of the agent. In terms of dosage and protection against degradation of the molecules during their passage through the harsh environments of the GIT, the in situ production of these molecules would be a large improvement. Lactic acid bacteria are envisaged to be ideal drug delivery vehicles (Wells & Mercenier, 2008). Combining these two improvements in disease treatment, i.e. the in situ delivery of glycosylated molecules, would revolutionize the way in which diseases are treated, by controlling the location of release and the efficacy of the therapeutic molecule.

Our work on the elucidation of the glycosylation potential of the original microbiota isolate and probiotic L. rhamnosus GG can aid in this revolution. The GRAS (Generally Regarded As Safe) status of this probiotic together with the fact that this bacterium can produce glycoproteins makes this an ideal candidate to further investigate. Moreover, in contrast to E. coli, L. rhamnosus GG does not produce endotoxins. Taken together, L. rhamnosus GG can very well be the probiotic of the future, delivering glycosylated therapeutic molecules or vaccines in situ in a safe and controllable way. Moreover, the fact that two probiotic effectors of L. rhamnosus GG were found to be glycosylated (i.e. Msp1 and the SpaCBA pili) also open avenues for the usage of these isolated glycoproteins as therapeutic molecules. The potential of glycoproteins of L. rhamnosus GG as safe new generation vaccine adjuvantia is currently being investigated.

Moreover, the results of this work further improve the knowledge on the molecules underlying probiotic- and microbiota-host interactions. The insights of this work can enhance an underpinned use of probiotics further. The bacterial glycan barcode of probiotics and microbiota is hypothesized to be a crucial part of the interactome with the environment and host. Furthermore, better insights on the key molecules present on probiotics can further substantiate the fundamental knowledge needed to support functional food and probiotic claims, as is now required by EFSA.

Another potential application of a better knowledge of bacterial glycobiology is the exploitation of enzymes to synthesize glycans in vitro. This organic synthesis of glycans holds many challenges such as the lack of control of the sterochemistry, low yield and difficulty to scale-up the process (Nicolaou & Mitchell, 2001). Chemical biologists try to tackle this by

seve

n.

175

purifying glycosyltransferases and other glycan-modifying enzymes in order to synthesize glycans in vitro. The benefits of using enzymes are wide-spread: the high catalytic activity, lack of side-reactions, mild reaction conditions and high region and stereoselectivity (Faijes & Planas, 2007). Especially bacterial GTs hold great promise, as they can be more easily cloned, overexpressed and purified compared to their eukaryotic counterparts (Merritt et al., 2013; Mutanda et al., 2014). Moreover, bacterial GTs are better accessible to engineering strategies to for instance enhance their substrate tolerance (Kiessling & Splain, 2010). These strategies are very promising, but are only starting to emerge as knowledge on bacterial glycosylation is growing. A benchmarking work was the report on the in vitro reconstruction of the pathway to synthesize the branched N-linked heptasaccharide of C. jejuni (Glover et al., 2005a). For the first time the biosynthesis of a bacterial glycan was completely reconstructed in vitro using purified enzymes. At the moment several groups are investigating this potential of bacterial GTs to perform one-pot synthesis of bacterial glycans (personal communication).

annex.references

addendumlist of publications

the extras

annex.references

addendumlist of publications

the extras

178

refe

renc

es

references

Aas, F. E., A. Vik, J. Vedde, M. Koomey, et al. (2007). Neisseria gonorrhoeae O-linked pilin glycosylation: functional analyses define both the biosynthetic pathway and glycan structure. Mol Microbiol 65(3): 607-624.Aduse-Opoku, J., M. Rangarajan, K. A. Young and M. A. Curtis (1998). Maturation of the arginine-specific proteases of Porphyromonas gingivalis W50 is dependent on a functional prR2 protease gene. Infect Immun 66(4): 1594-1600.Aebi, M., R. Bernasconi, S. Clerc and M. Molinari (2010). N-glycan structures: recognition and processing in the ER. Trends Biochem Sci 35(2): 74-82.Al-Dabbagh, B., D. Mengin-Lecreulx and A. Bouhss (2008). Purification and characterization of the bacterial UDP-GlcNAc:undecaprenyl-phosphate GlcNAc-1-phosphate transferase WecA. J Bacteriol 190(21): 7141-7146.Alaimo, C., I. Catrein, L. Morf, C. L. Marolda, et al. (2006). Two distinct but interchangeable mechanisms for flipping of lipid-linked oligosaccharides. EMBO J 25(5): 967-976.Alemka, A., H. Nothaft, J. Zheng and C. M. Szymanski (2013). N-glycosylation of Campylobacter jejuni surface proteins promotes bacterial fitness. Infect Immun 81(5): 1674-1682.Allison, S. E., M. A. D’Elia, S. Arar, M. A. Monteiro, et al. (2011). Studies of the genetics, function, and kinetic mechanism of TagE, the wall teichoic acid glycosyltransferase in Bacillus subtilis 168. J Biol Chem 286(27): 23708-23716. Angeloni, S., J. L. Ridet, N. Kusy, H. Gao, et al. (2005). Glycoprofiling with micro-arrays of glycoconjugates and lectins. Glycobiology 15(1): 31-41.Anonsen, J. H., W. Egge-Jacobsen, F. E. Aas, B. Borud, et al. (2012a). Novel protein substrates of the phospho-form modification system in Neisseria gonorrhoeae and their connection to O-linked protein glycosylation. Infect Immun 80(1): 22-30.Anonsen, J. H., A. Vik, W. Egge-Jacobsen and M. Koomey (2012b). An extended spectrum of target proteins and modification sites in the general O-linked protein glycosylation system in Neisseria gonorrhoeae. J Proteome Res 11(12): 5781-5793.Anzengruber, J., M. Pabst, L. Neumann, G. Sekot, et al. (2013). Protein O-glucosylation in Lactobacillus buchneri. Glycoconj J, 10.1007/s10719-013-9505-7.Apweiler, R., H. Hermjakob and N. Sharon (1999). On the frequency of protein glycosylation, as deduced from analysis of the SWISS-PROT database. Biochim Biophys Acta 1473(1): 4-8.Armstrong, G. D., L. S. Frost, H. J. Vogel and W. Paranchych (1981). Nature of the carbohydrate and phosphate associated with ColB2 and EDP208 pilin. J Bacteriol 145(3): 1167-1176.Arnold, F., L. Bedouet, P. Batina, G. Robreau, et al. (1998). Biochemical and immunological analyses of the flagellin of Clostridium tyrobutyricum ATCC 25755. Microbiol Immunol 42(1): 23-31.Arora, S. K., M. Bangera, S. Lory and R. Ramphal (2001). A genomic island in Pseudomonas aeruginosa carries the determinants of flagellin glycosylation. Proc Natl Acad Sci U S A 98(16): 9342-9347.Arora, S. K., A. N. Neely, B. Blair, S. Lory, et al. (2005). Role of motility and flagellin glycosylation in the pathogenesis of Pseudomonas aeruginosa burn wound infections. Infect Immun 73(7): 4395-4398.Arora, S. K., M. C. Wolfgang, S. Lory and R. Ramphal (2004). Sequence polymorphism in the glycosylation island and flagellins of Pseudomonas aeruginosa. J Bacteriol 186(7): 2115-2122.Arumugam, M., J. Raes, E. Pelletier, D. Le Paslier, et al. (2011). Enterotypes of the human gut microbiome. Nature 473(7346): 174-180.Asakura, H., Y. Churin, B. Bauer, J. P. Boettcher, et al. (2010). Helicobacter pylori HP0518 affects flagellin glycosylation to alter bacterial motility. Mol Microbiol 78(5): 1130-1144.Avall-Jaaskelainen, S. and A. Palva (2005). Lactobacillus surface layers and their applications. FEMS Microbiol Rev 29(3): 511-529.Baiet, B., C. Burel, B. Saint-Jean, R. Louvet, et al. (2011). N-glycans of Phaeodactylum tricornutum diatom and functional characterization of its N-acetylglucosaminyltransferase I enzyme. J Biol Chem 286(8): 6152-6164.Ballicora, M. A., A. A. Iglesias and J. Preiss (2003). ADP-glucose pyrophosphorylase, a regulatory enzyme for bacterial glycogen synthesis. Microbiol Mol Biol Rev 67(2): 213-225, table of contents.Balonova, L., L. Hernychova and Z. Bilkova (2009). Bioanalytical tools for the discovery of eukaryotic glycoproteins applied to the analysis of bacterial glycoproteins. Expert Rev Proteomics 6(1): 75-85.

refe

renc

es.

179

Balonova, L., L. Hernychova, B. F. Mann, M. Link, et al. (2010). Multimethodological approach to identification of glycoproteins from the proteome of Francisella tularensis, an intracellular microorganism. J Proteome Res 9(4): 1995-2005.Balonova, L., B. F. Mann, L. Cerveny, W. R. Alley, Jr., et al. (2012). Characterization of protein glycosylation in Francisella tularensis subsp. holarctica: identification of a novel glycosylated lipoprotein required for virulence. Mol Cell Proteomics 11(7): M111 015016.Banerjee, A., R. Wang, S. L. Supernavage, S. K. Ghosh, et al. (2002). Implications of phase variation of a gene (pgtA) encoding a pilin galactosyl transferase in gonococcal pathogenesis. J Exp Med 196(2): 147-162.Barinka, C., P. Sacha, J. Sklenar, P. Man, et al. (2004). Identification of the N-glycosylation sites on glutamate carboxypeptidase II necessary for proteolytic activity. Protein Sci 13(6): 1627-1635.Barrett, D., T. S. Wang, Y. Yuan, Y. Zhang, et al. (2007). Analysis of glycan polymers produced by peptidoglycan glycosyltransferases. J Biol Chem 282(44): 31964-31971.Bauerl, C., G. Perez-Martinez, F. Yan, D. B. Polk, et al. (2010). Functional analysis of the p40 and p75 proteins from Lactobacillus casei BL23. J Mol Microbiol Biotechnol 19(4): 231-241.Bedouet, L., F. Arnold, G. Robreau, P. Batina, et al. (1998). Evidence for an heterogeneous glycosylation of the Clostridium tyrobutyricum ATCC 25755 flagellin. Microbios 94(379): 183-192.Belkaid, Y. and T. W. Hand (2014). Role of the microbiota in immunity and inflammation. Cell 157(1): 121-141.Bensing, B. A., B. W. Gibson and P. M. Sullam (2004a). The Streptococcus gordonii platelet binding protein GspB undergoes glycosylation independently of export. J Bacteriol 186(3): 638-645.Bensing, B. A., J. A. Lopez and P. M. Sullam (2004b). The Streptococcus gordonii surface proteins GspB and Hsa mediate binding to sialylated carbohydrate epitopes on the platelet membrane glycoprotein Ibalpha. Infect Immun 72(11): 6528-6537.Bensing, B. A. and P. M. Sullam (2002). An accessory sec locus of Streptococcus gordonii is required for export of the surface protein GspB and for normal levels of binding to human platelets. Mol Microbiol 44(4): 1081-1094.Bensing, B. A., D. Takamatsu and P. M. Sullam (2005). Determinants of the streptococcal surface glycoprotein GspB that facilitate export by the accessory Sec system. Mol Microbiol 58(5): 1468-1481.Benz, I. and M. A. Schmidt (2001). Glycosylation with heptose residues mediated by the aah gene product is essential for adherence of the AIDA-I adhesin. Mol Microbiol 40(6): 1403-1413.Bernatchez, S., C. M. Szymanski, N. Ishiyama, J. Li, et al. (2005). A single bifunctional UDP-GlcNAc/Glc 4-epimerase supports the synthesis of three cell surface glycoconjugates in Campylobacter jejuni. J Biol Chem 280(6): 4792-4802.Black, R. E., M. M. Levine, M. L. Clements, T. P. Hughes, et al. (1988). Experimental Campylobacter jejuni infection in humans. J Infect Dis 157(3): 472-479.Borud, B., F. E. Aas, A. Vik, H. C. Winther-Larsen, et al. (2010). Genetic, structural, and antigenic analyses of glycan diversity in the O-linked protein glycosylation systems of human Neisseria species. J Bacteriol 192(11): 2816-2829.Borud, B., R. Viburiene, M. D. Hartley, B. S. Paulsen, et al. (2011). Genetic and molecular analyses reveal an evolutionary trajectory for glycan synthesis in a bacterial protein glycosylation system. Proc Natl Acad Sci U S A 108(23): 9643-9648.Brechtel, E., M. Matuschek, A. Hellberg, E. M. Egelseer, et al. (1999). Cell wall of Thermoanaerobacterium thermosulfurigenes EM1: isolation of its components and attachment of the xylanase XynA. Arch Microbiol 171(3): 159-165.Brew, K., P. Tumbale and K. R. Acharya (2010). Family 6 glycosyltransferases in vertebrates and bacteria: inactivation and horizontal gene transfer may enhance mutualism between vertebrates and bacteria. J Biol Chem 285(48): 37121-37127.Brimer, C. D. and T. C. Montie (1998). Cloning and comparison of fliC genes and identification of glycosylation in the flagellin of Pseudomonas aeruginosa a-type strains. J Bacteriol 180(12): 3209-3217.Brittan, J. L. and A. H. Nobbs (2015). Group B Streptococcus pili mediate adherence to salivary glycoproteins. Microbes Infect, 10.1016/j.micinf.2014.12.013.Bry, L., P. G. Falk, T. Midtvedt and J. I. Gordon (1996). A model of host-microbial interactions in an open mammalian ecosystem. Science 273(5280): 1380-1383.Bu, S., Y. Li, M. Zhou, P. Azadin, et al. (2008). Interaction between two putative glycosyltransferases is required for glycosylation of a serine-rich streptococcal adhesin. J Bacteriol 190(4): 1256-1266.Bugg, T. D. and P. E. Brandish (1994). From peptidoglycan to glycoproteins: common features of lipid-linked

180

refe

renc

es

oligosaccharide biosynthesis. FEMS Microbiol Lett 119(3): 255-262.Cabib, E. and L. F. Leloir (1954). Guanosine diphosphate mannose. J Biol Chem 206(2): 779-790.Campbell, C. T., S. G. Sampathkumar and K. J. Yarema (2007). Metabolic oligosaccharide engineering: perspectives, applications, and future directions. Mol Biosyst 3(3): 187-194.Campbell, J. A., G. J. Davies, V. Bulone and B. Henrissat (1997). A classification of nucleotide-diphospho-sugar glycosyltransferases based on amino acid sequence similarities. Biochem J 326 ( Pt 3): 929-939.Campbell, R. E., S. C. Mosimann, M. E. Tanner and N. C. Strynadka (2000). The structure of UDP-N-acetylglucosamine 2-epimerase reveals homology to phosphoglycosyl transferases. Biochemistry 39(49): 14993-15001.Cani, P. D. and N. M. Delzenne (2009). Interplay between obesity and associated metabolic disorders: new insights into the gut microbiota. Curr Opin Pharmacol 9(6): 737-743.Cantarel, B. L., P. M. Coutinho, C. Rancurel, T. Bernard, et al. (2009). The Carbohydrate-Active EnZymes database (CAZy): an expert resource for Glycogenomics. Nucleic Acids Res 37(Database issue): D233-238.Castaldo, C., V. Vastano, R. A. Siciliano, M. Candela, et al. (2009). Surface displaced alfa-enolase of Lactobacillus plantarum is a fibronectin binding protein. Microb Cell Fact 8: 14.Castric, P. (1995). pilO, a gene required for glycosylation of Pseudomonas aeruginosa 1244 pilin. Microbiology 141 ( Pt 5): 1247-1254.Castric, P., F. J. Cassels and R. W. Carlson (2001). Structural characterization of the Pseudomonas aeruginosa 1244 pilin glycan. J Biol Chem 276(28): 26479-26485.Castric, P. A. and C. D. Deal (1994). Differentiation of Pseudomonas aeruginosa pili based on sequence and B-cell epitope analyses. Infect Immun 62(2): 371-376.Chaiet, L. and F. J. Wolf (1964). The Properties of Streptavidin, a Biotin-Binding Protein Produced by Streptomycetes. Arch Biochem Biophys 106: 1-5.Chamot-Rooke, J., B. Rousseau, F. Lanternier, G. Mikaty, et al. (2007). Alternative Neisseria spp. type IV pilin glycosylation with a glyceramido acetamido trideoxyhexose residue. Proc Natl Acad Sci U S A 104(37): 14783-14788.Champasa, K., S. A. Longwell, A. M. Eldridge, E. A. Stemmler, et al. (2013). Targeted Identification of Glycosylated Proteins in the Gastric Pathogen Helicobacter pylori (Hp). Mol Cell Proteomics 12(9): 2568-2586.Chapman-Smith, A. and J. E. Cronan, Jr. (1999). Molecular biology of biotin attachment to proteins. J Nutr 129(2S Suppl): 477S-484S.Charbonneau, M. E., J. P. Cote, M. F. Haurat, B. Reiz, et al. (2012). A structural motif is the recognition site for a new family of bacterial protein O-glycosyltransferases. Mol Microbiol 83(5): 894-907.Charbonneau, M. E., V. Girard, A. Nikolakakis, M. Campos, et al. (2007). O-linked glycosylation ensures the normal conformation of the autotransporter adhesin involved in diffuse adherence. J Bacteriol 189(24): 8880-8889.Charnock, S. J. and G. J. Davies (1999a). Cloning, crystallization and preliminary X-ray analysis of a nucleotide-diphospho-sugar transferase spsA from Bacillus subtilis. Acta Crystallogr D Biol Crystallogr 55(Pt 3): 677-678.Charnock, S. J. and G. J. Davies (1999b). Structure of the nucleotide-diphospho-sugar transferase, SpsA from Bacillus subtilis, in native and nucleotide-complexed forms. Biochemistry 38(20): 6380-6385.Chen, M. M., K. J. Glover and B. Imperiali (2007a). From peptide to protein: comparative analysis of the substrate specificity of N-linked glycosylation in C. jejuni. Biochemistry 46(18): 5579-5585.Chen, M. M., E. Weerapana, E. Ciepichal, J. Stupak, et al. (2007b). Polyisoprenol specificity in the Campylobacter jejuni N-linked glycosylation pathway. Biochemistry 46(50): 14342-14348.Chen, Q., H. Wu and P. M. Fives-Taylor (2004). Investigating the role of secA2 in secretion and glycosylation of a fimbrial adhesin in Streptococcus parasanguis FW213. Mol Microbiol 53(3): 843-856.Chen, T., L. Hedman, P. S. Mattila, L. Jartti, et al. (2012). Biotin IgM antibodies in human blood: a previously unknown factor eliciting false results in biotinylation-based immunoassays. PLoS One 7(8): e42376.Chevalier, J., J. Yi, O. Michel and X. M. Tang (1997). Biotin and digoxigenin as labels for light and electron microscopy in situ hybridization probes: where do we stand? J Histochem Cytochem 45(4): 481-491.Chia, J. S., L. Y. Chang, C. T. Shun, Y. Y. Chang, et al. (2001). A 60-kilodalton immunodominant glycoprotein is essential for cell wall integrity and the maintenance of cell shape in Streptococcus mutans. Infect Immun 69(11): 6987-6998.Chou, W. K., S. Dick, W. W. Wakarchuk and M. E. Tanner (2005). Identification and characterization of NeuB3 from

refe

renc

es.

181

Campylobacter jejuni as a pseudaminic acid synthase. J Biol Chem 280(43): 35922-35928.Claes, I. J., S. Lebeer, C. Shen, T. L. Verhoeven, et al. (2010). Impact of lipoteichoic acid modification on the performance of the probiotic Lactobacillus rhamnosus GG in experimental colitis. Clin Exp Immunol 162(2): 306-314.Claes, I. J., G. Schoofs, K. Regulski, P. Courtin, et al. (2012). Genetic and Biochemical Characterization of the Cell Wall Hydrolase Activity of the Major Secreted Protein of Lactobacillus rhamnosus GG. PLoS One 7(2): e31588.Claes, I. J., C. E. Vargas Garcia and S. Lebeer (2014). Novel opportunities for the exploitation of host-microbiome interactions in the intestine. Curr Opin Biotechnol 32C: 28-34.Collins, R. F., L. Davidsen, J. P. Derrick, R. C. Ford, et al. (2001). Analysis of the PilQ secretin from Neisseria meningitidis by transmission electron microscopy reveals a dodecameric quaternary structure. J Bacteriol 183(13): 3825-3832.Comer, J. E., M. A. Marshall, V. J. Blanch, C. D. Deal, et al. (2002). Identification of the Pseudomonas aeruginosa 1244 pilin glycosylation site. Infect Immun 70(6): 2837-2845.Commichau, F. M., K. Forchhammer and J. Stulke (2006). Regulatory links between carbon and nitrogen metabolism. Curr Opin Microbiol 9(2): 167-172.Corbett, D. and I. S. Roberts (2008). Capsular polysaccharides in Escherichia coli. Adv Appl Microbiol 65: 1-26.Coutinho, P. M., E. Deleury, G. J. Davies and B. Henrissat (2003). An evolving hierarchical family classification for glycosyltransferases. J Mol Biol 328(2): 307-317.Coyne, M. J., C. M. Fletcher, M. Chatzidaki-Livanis, G. Posch, et al. (2013). Phylum-wide general protein O-glycosylation system of the Bacteroidetes. Mol Microbiol 88(4): 772-783.Coyne, M. J., B. Reinap, M. M. Lee and L. E. Comstock (2005). Human symbionts use a host-like pathway for surface fucosylation. Science 307(5716): 1778-1781.Creuzenet, C., M. J. Schur, J. Li, W. W. Wakarchuk, et al. (2000). FlaA1, a new bifunctional UDP-GlcNAc C6 Dehydratase/ C4 reductase from Helicobacter pylori. J Biol Chem 275(45): 34873-34880.Cuccui, J. and B. W. Wren (2013). Bacteria like sharing their sweets. Mol Microbiol 89(5): 811-815.Curtis, M. A., A. Thickett, J. M. Slaney, M. Rangarajan, et al. (1999). Variable carbohydrate modifications to the catalytic chains of the RgpA and RgpB proteases of Porphyromonas gingivalis W50. Infect Immun 67(8): 3816-3823.Cuthbertson, L., M. S. Kimber and C. Whitfield (2007). Substrate binding by a bacterial ABC transporter involved in polysaccharide export. Proc Natl Acad Sci U S A 104(49): 19529-19534.Cuthbertson, L., J. Powers and C. Whitfield (2005). The C-terminal domain of the nucleotide-binding domain protein Wzt determines substrate specificity in the ATP-binding cassette transporter for the lipopolysaccharide O-antigens in Escherichia coli serotypes O8 and O9a. J Biol Chem 280(34): 30310-30319.Dasgupta, N., M. C. Wolfgang, A. L. Goodman, S. K. Arora, et al. (2003). A four-tiered transcriptional regulatory circuit controls flagellar biogenesis in Pseudomonas aeruginosa. Mol Microbiol 50(3): 809-824.Davies, G. J., T. M. Gloster and B. Henrissat (2005). Recent structural insights into the expanding world of carbohydrate-active enzymes. Curr Opin Struct Biol 15(6): 637-645.Davies, G. J. and M. L. Sinnott (2008). Sorting the diverse: the sequence-based classifications of carbohydrate-active enzymes. Biochem J: c1-c5.de Zoete, M. R., A. M. Keestra, J. A. Wagenaar and J. P. van Putten (2010). Reconstitution of a functional Toll-like receptor 5 binding site in Campylobacter jejuni flagellin. J Biol Chem 285(16): 12149-12158.Deepika, G., E. Karunakaran, C. R. Hurley, C. A. Biggs, et al. (2012). Influence of fermentation conditions on the surface properties and adhesion of Lactobacillus rhamnosus GG. Microb Cell Fact 11: 116.Dell, A., A. Galadari, F. Sastre and P. Hitchen (2010). Similarities and differences in the glycosylation mechanisms in prokaryotes and eukaryotes. Int J Microbiol 2010: 148178.Demendi, M. and C. Creuzenet (2009). Cj1123c (PglD), a multifaceted acetyltransferase from Campylobacter jejuni. Biochem Cell Biol 87(3): 469-483.Di Guilmi, A. M., A. Dessen, O. Dideberg and T. Vernet (2003). The glycosyltransferase domain of penicillin-binding protein 2a from Streptococcus pneumoniae catalyzes the polymerization of murein glycan chains. J Bacteriol 185(15): 4418-4423.Diaz-Silvestre, H., P. Espinosa-Cueto, A. Sanchez-Gonzalez, M. A. Esparza-Ceron, et al. (2005). The 19-kDa antigen of Mycobacterium tuberculosis is a major adhesin that binds the mannose receptor of THP-1 monocytic cells and

182

refe

renc

es

promotes phagocytosis of mycobacteria. Microb Pathog 39(3): 97-107.DiGiandomenico, A., M. J. Matewish, A. Bisaillon, J. R. Stehle, et al. (2002). Glycosylation of Pseudomonas aeruginosa 1244 pilin: glycan substrate specificity. Mol Microbiol 46(2): 519-530.Dillard, J. P., M. W. Vandersea and J. Yother (1995). Characterization of the cassette containing genes for type 3 capsular polysaccharide biosynthesis in Streptococcus pneumoniae. J Exp Med 181(3): 973-983.Ding, W., H. Nothaft, C. M. Szymanski and J. Kelly (2009). Identification and quantification of glycoproteins using ion-pairing normal-phase liquid chromatography and mass spectrometry. Mol Cell Proteomics 8(9): 2170-2185.Dobos, K. M., K. H. Khoo, K. M. Swiderek, P. J. Brennan, et al. (1996). Definition of the full extent of glycosylation of the 45-kilodalton glycoprotein of Mycobacterium tuberculosis. J Bacteriol 178(9): 2498-2506.Dobos, K. M., K. Swiderek, K. H. Khoo, P. J. Brennan, et al. (1995). Evidence for glycosylation sites on the 45-kilodalton glycoprotein of Mycobacterium tuberculosis. Infect Immun 63(8): 2846-2853.Doig, P., N. Kinsella, P. Guerry and T. J. Trust (1996). Characterization of a post-translational modification of Campylobacter flagellin: identification of a sero-specific glycosyl moiety. Mol Microbiol 19(2): 379-387.Dons, L., O. F. Rasmussen and J. E. Olsen (1992). Cloning and characterization of a gene encoding flagellin of Listeria monocytogenes. Mol Microbiol 6(20): 2919-2929.Doron, S., D. R. Snydman and S. L. Gorbach (2005). Lactobacillus GG: bacteriology and clinical applications. Gastroenterol Clin North Am 34(3): 483-498, ix.Douillard, F. P., P. Rasinkangas, I. von Ossowski, J. Reunanen, et al. (2014). Functional identification of conserved residues involved in Lactobacillus rhamnosus strain GG sortase specificity and pilus biogenesis. J Biol Chem 289(22): 15764-15775.Douillard, F. P., A. Ribbera, H. M. Jarvinen, R. Kant, et al. (2013). Comparative genomic and functional analysis of Lactobacillus casei and Lactobacillus rhamnosus strains marketed as probiotics. Appl Environ Microbiol 79(6): 1923-1933.Drummelsmith, J. and C. Whitfield (1999). Gene products required for surface expression of the capsular form of the group 1 K antigen in Escherichia coli (O9a:K30). Mol Microbiol 31(5): 1321-1332.Dupres, V., F. D. Menozzi, C. Locht, B. H. Clare, et al. (2005). Nanoscale mapping and functional analysis of individual adhesins on living bacteria. Nat Methods 2(7): 515-520.Ebner, A., L. Wildling, A. S. Kamruzzahan, C. Rankl, et al. (2007). A new, simple method for linking of antibodies to atomic force microscopy tips. Bioconjug Chem 18(4): 1176-1184.Egelund, J., M. Skjot, N. Geshi, P. Ulvskov, et al. (2004). A complementary bioinformatics approach to identify potential plant cell wall glycosyltransferase-encoding genes. Plant Physiol 136(1): 2609-2620.Egge-Jacobsen, W., E. N. Salomonsson, F. E. Aas, A. L. Forslund, et al. (2011). O-linked glycosylation of the PilA pilin protein of Francisella tularensis: identification of the endogenous protein-targeting oligosaccharyltransferase and characterization of the native oligosaccharide. J Bacteriol 193(19): 5487-5497.Elliott, S., T. Lorenzini, S. Asher, K. Aoki, et al. (2003). Enhancement of therapeutic protein in vivo activities through glycoengineering. Nat Biotechnol 21(4): 414-421.Elsinghorst, E. A. and J. A. Weitz (1994). Epithelial cell invasion and adherence directed by the enterotoxigenic Escherichia coli tib locus is associated with a 104-kilodalton outer membrane protein. Infect Immun 62(8): 3463-3471.Erickson, P. R. and M. C. Herzberg (1993). Evidence for the covalent linkage of carbohydrate polymers to a glycoprotein from Streptococcus sanguis. J Biol Chem 268(32): 23780-23783.Espitia, C., R. Espinosa, R. Saavedra, R. Mancilla, et al. (1995). Antigenic and structural similarities between Mycobacterium tuberculosis 50- to 55-kilodalton and Mycobacterium bovis BCG 45- to 47-kilodalton antigens. Infect Immun 63(2): 580-584.Espitia, C. and R. Mancilla (1989). Identification, isolation and partial characterization of Mycobacterium tuberculosis glycoprotein antigens. Clin Exp Immunol 77(3): 378-383.Ewing, C. P., E. Andreishcheva and P. Guerry (2009). Functional characterization of flagellin glycosylation in Campylobacter jejuni 81-176. J Bacteriol 191(22): 7086-7093.Faijes, M. and A. Planas (2007). In vitro synthesis of artificial polysaccharides by glycosidases and glycosynthases. Carbohydr Res 342(12-13): 1581-1594.Fairbanks, G., T. L. Steck and D. F. Wallach (1971). Electrophoretic analysis of the major polypeptides of the human erythrocyte membrane. Biochemistry 10(13): 2606-2617.

refe

renc

es.

183

FAO/WHO (2001). Report on joint FAO/WHO expert consultation of health and nutritional properties of probiotics in food including powder milk with live lactic acid bacteria. Geneva, WHO.Farahmand, B., M. Khodabandeh, F. Mahboudi, F. Fotouhi, et al. (2011). Influenza Virus Hemagglutinin: A Model for Protein N-Glycosylation in Recombinant Escherichia coli. Intervirology, 000327784 [pii]10.1159/000327784.Faridmoayer, A., M. A. Fentabil, M. F. Haurat, W. Yi, et al. (2008). Extreme substrate promiscuity of the Neisseria oligosaccharyl transferase involved in protein O-glycosylation. J Biol Chem 283(50): 34596-34604.Faridmoayer, A., M. A. Fentabil, D. C. Mills, J. S. Klassen, et al. (2007). Functional characterization of bacterial oligosaccharyltransferases involved in O-linked protein glycosylation. J Bacteriol 189(22): 8088-8098.Feldman, M. F., C. L. Marolda, M. A. Monteiro, M. B. Perry, et al. (1999). The activity of a putative polyisoprenol-linked sugar translocase (Wzx) involved in Escherichia coli O antigen assembly is independent of the chemical structure of the O repeat. J Biol Chem 274(49): 35129-35138.Feldman, M. F., M. Wacker, M. Hernandez, P. G. Hitchen, et al. (2005). Engineering N-linked protein glycosylation with diverse O antigen lipopolysaccharide structures in Escherichia coli. Proc Natl Acad Sci U S A 102(8): 3016-3021.Feldman-Salit, A., S. Hering, H. L. Messiha, N. Veith, et al. (2013). Regulation of the activity of lactate dehydrogenases from four lactic acid bacteria. J Biol Chem 288(29): 21295-21306.Feltcher, M. E. and M. Braunstein (2012). Emerging themes in SecA2-mediated protein export. Nat Rev Microbiol 10(11): 779-789.Fifis, T., C. Costopoulos, A. J. Radford, A. Bacic, et al. (1991). Purification and characterization of major antigens from a Mycobacterium bovis culture filtrate. Infect Immun 59(3): 800-807.Finn, R. D., J. Clements and S. R. Eddy (2011). HMMER web server: interactive sequence similarity searching. Nucleic Acids Res 39(Web Server issue): W29-37.Finn, R. D., J. Mistry, J. Tate, P. Coggill, et al. (2010). The Pfam protein families database. Nucleic Acids Res 38(Database issue): D211-222.Fischer, W. and R. Haas (2004). The RecA protein of Helicobacter pylori requires a posttranslational modification for full activity. J Bacteriol 186(3): 777-784.Fleckenstein, J. M., K. Roy, J. F. Fischer and M. Burkitt (2006). Identification of a two-partner secretion locus of enterotoxigenic Escherichia coli. Infect Immun 74(4): 2245-2258.Fletcher, C. M., M. J. Coyne and L. E. Comstock (2011). Theoretical and experimental characterization of the scope of protein O-glycosylation in Bacteroides fragilis. J Biol Chem 286(5): 3219-3226.Fletcher, C. M., M. J. Coyne, O. F. Villa, M. Chatzidaki-Livanis, et al. (2009). A general O-glycosylation system important to the physiology of a major human intestinal symbiont. Cell 137(2): 321-331.Forsee, W. T., R. T. Cartee and J. Yother (2009). A kinetic model for chain length modulation of Streptococcus pneumoniae cellubiuronan capsular polysaccharide by nucleotide sugar donor concentrations. J Biol Chem 284(18): 11836-11844.Forslund, A. L., K. Kuoppa, K. Svensson, E. Salomonsson, et al. (2006). Direct repeat-mediated deletion of a type IV pilin gene results in major virulence attenuation of Francisella tularensis. Mol Microbiol 59(6): 1818-1830.Francius, G., D. Alsteens, V. Dupres, S. Lebeer, et al. (2009). Stretching polysaccharides on live cells using single molecule force spectroscopy. Nat Protoc 4(6): 939-946.Francius, G., S. Lebeer, D. Alsteens, L. Wildling, et al. (2008). Detection, localization, and conformational analysis of single polysaccharide molecules on live bacteria. ACS Nano 2(9): 1921-1929.Fredriksen, L., G. Mathiesen, A. Moen, P. A. Bron, et al. (2012). The Major Autolysin Acm2 from Lactobacillus plantarum Undergoes Cytoplasmic O-Glycosylation. J Bacteriol 194(2): 325-333.Fredriksen, L., A. Moen, A. A. Adzhubei, G. Mathiesen, et al. (2013). Lactobacillus plantarum WCFS1 O-linked protein glycosylation: An extended spectrum of target proteins and modification sites detected by mass spectrometry. Glycobiology, 10.1093/glycob/cwt071.Fugate, C. J. and J. T. Jarrett (2012). Biotin synthase: insights into radical-mediated carbon-sulfur bond formation. Biochim Biophys Acta 1824(11): 1213-1222.Garbe, T., D. Harris, M. Vordermeier, R. Lathigra, et al. (1993). Expression of the Mycobacterium tuberculosis 19-kilodalton antigen in Mycobacterium smegmatis: immunological analysis and evidence of glycosylation. Infect Immun 61(1): 260-267.

184

refe

renc

es

Gareau, M. G., P. M. Sherman and W. A. Walker (2010). Probiotics and the gut microbiota in intestinal health and disease. Nat Rev Gastroenterol Hepatol 7(9): 503-514.Ge, Y., C. Li, L. Corum, C. A. Slaughter, et al. (1998). Structure and expression of the FlaA periplasmic flagellar protein of Borrelia burgdorferi. J Bacteriol 180(9): 2418-2425.Gebhart, C., M. V. Ielmini, B. Reiz, N. L. Price, et al. (2012). Characterization of exogenous bacterial oligosaccharyltransferases in Escherichia coli reveals the potential for O-linked protein glycosylation in Vibrio cholerae and Burkholderia thailandensis. Glycobiology 22(7): 962-974.Geijtenbeek, T. B., D. S. Kwon, R. Torensma, S. J. van Vliet, et al. (2000a). DC-SIGN, a dendritic cell-specific HIV-1-binding protein that enhances trans-infection of T cells. Cell 100(5): 587-597.Geijtenbeek, T. B., R. Torensma, S. J. van Vliet, G. C. van Duijnhoven, et al. (2000b). Identification of DC-SIGN, a novel dendritic cell-specific ICAM-3 receptor that supports primary immune responses. Cell 100(5): 575-585.Geijtenbeek, T. B., G. C. van Duijnhoven, S. J. van Vliet, E. Krieger, et al. (2002). Identification of different binding sites in the dendritic cell-specific receptor DC-SIGN for intercellular adhesion molecule 3 and HIV-1. J Biol Chem 277(13): 11314-11320.Geijtenbeek, T. B., Y. van Kooyk, S. J. van Vliet, M. H. Renes, et al. (1999). High frequency of adhesion defects in B-lineage acute lymphoblastic leukemia. Blood 94(2): 754-764.Geijtenbeek, T. B., S. J. Van Vliet, E. A. Koppel, M. Sanchez-Hernandez, et al. (2003). Mycobacteria target DC-SIGN to suppress dendritic cell function. J Exp Med 197(1): 7-17.Ghosh, S. K., J. Zhao, M. C. Philogene, A. Alzaharani, et al. (2004). Pathogenic consequences of Neisseria gonorrhoeae pilin glycan variation. Microbes Infect 6(7): 693-701.Glover, K. J., E. Weerapana, M. M. Chen and B. Imperiali (2006). Direct biochemical evidence for the utilization of UDP-bacillosamine by PglC, an essential glycosyl-1-phosphate transferase in the Campylobacter jejuni N-linked glycosylation pathway. Biochemistry 45(16): 5343-5350.Glover, K. J., E. Weerapana and B. Imperiali (2005a). In vitro assembly of the undecaprenylpyrophosphate-linked heptasaccharide for prokaryotic N-linked glycosylation. Proc Natl Acad Sci U S A 102(40): 14255-14259.Glover, K. J., E. Weerapana, S. Numao and B. Imperiali (2005b). Chemoenzymatic synthesis of glycopeptides with PglB, a bacterial oligosaccharyl transferase from Campylobacter jejuni. Chem Biol 12(12): 1311-1315.Golden, N. J. and D. W. Acheson (2002). Identification of motility and autoagglutination Campylobacter jejuni mutants by random transposon mutagenesis. Infect Immun 70(4): 1761-1771.Goon, S., J. F. Kelly, S. M. Logan, C. P. Ewing, et al. (2003). Pseudaminic acid, the major modification on Campylobacter flagellin, is synthesized via the Cj1293 gene. Mol Microbiol 50(2): 659-671.Granato, D., F. Perotti, I. Masserey, M. Rouvet, et al. (1999). Cell surface-associated lipoteichoic acid acts as an adhesion factor for attachment of Lactobacillus johnsonii La1 to human enterocyte-like Caco-2 cells. Appl Environ Microbiol 65(3): 1071-1077.Grass, S., A. Z. Buscher, W. E. Swords, M. A. Apicella, et al. (2003). The Haemophilus influenzae HMW1 adhesin is glycosylated in a process that requires HMW1C and phosphoglucomutase, an enzyme involved in lipooligosaccharide biosynthesis. Mol Microbiol 48(3): 737-751.Grass, S., C. F. Lichti, R. R. Townsend, J. Gross, et al. (2010). The Haemophilus influenzae HMW1C protein is a glycosyltransferase that transfers hexose residues to asparagine sites in the HMW1 adhesin. PLoS Pathog 6(5): e1000919.Green, N. M. (1975). Avidin. Adv Protein Chem 29: 85-133.Gringhuis, S. I., J. den Dunnen, M. Litjens, M. van der Vlist, et al. (2009). Carbohydrate-specific signaling through the DC-SIGN signalosome tailors immunity to Mycobacterium tuberculosis, HIV-1 and Helicobacter pylori. Nat Immunol 10(10): 1081-1088.Gross, J., S. Grass, A. E. Davis, P. Gilmore-Erdmann, et al. (2008). The Haemophilus influenzae HMW1 adhesin is a glycoprotein with an unusual N-linked carbohydrate modification. J Biol Chem 283(38): 26010-26015.Gubish, E. R., Jr., K. C. Chen and T. M. Buchanan (1982). Attachment of gonococcal pili to lectin-resistant clones of Chinese hamster ovary cells. Infect Immun 37(1): 189-194.Guerry, P., P. Doig, R. A. Alm, D. H. Burr, et al. (1996). Identification and characterization of genes required for post-translational modification of Campylobacter coli VC167 flagellin. Mol Microbiol 19(2): 369-378.Guerry, P., C. P. Ewing, M. Schirm, M. Lorenzo, et al. (2006). Changes in flagellin glycosylation affect Campylobacter autoagglutination and virulence. Mol Microbiol 60(2): 299-311.

refe

renc

es.

185

Guerry, P. and C. M. Szymanski (2008). Campylobacter sugars sticking out. Trends Microbiol 16(9): 428-435.Gundogdu, O., S. D. Bentley, M. T. Holden, J. Parkhill, et al. (2007). Re-annotation and re-analysis of the Campylobacter jejuni NCTC11168 genome sequence. BMC Genomics 8: 162.Ha, S., B. Gross and S. Walker (2001). E. coli MurG: a paradigm for a superfamily of glycosyltransferases. Curr Drug Targets Infect Disord 1(2): 201-213.Hamadeh, R. M., M. M. Estabrook, P. Zhou, G. A. Jarvis, et al. (1995). Anti-Gal binds to pili of Neisseria meningitidis: the immunoglobulin A isotype blocks complement-mediated killing. Infect Immun 63(12): 4900-4906.Hansen, S. F., E. Bettler, A. Rinnan, S. B. Engelsen, et al. (2010). Exploring genomes for glycosyltransferases. Mol Biosyst 6(10): 1773-1781.Hansen, S. F., E. Bettler, M. Wimmerova, A. Imberty, et al. (2009). Combination of several bioinformatics approaches for the identification of new putative glycosyltransferases in Arabidopsis. J Proteome Res 8(2): 743-753.Hart, C., B. Schulenberg, T. H. Steinberg, W. Y. Leung, et al. (2003). Detection of glycoproteins in polyacrylamide gels and on electroblots using Pro-Q Emerald 488 dye, a fluorescent periodate Schiff-base stain. Electrophoresis 24(4): 588-598.Hartley, M. D., M. J. Morrison, F. E. Aas, B. Borud, et al. (2011). Biochemical characterization of the O-linked glycosylation pathway in Neisseria gonorrhoeae responsible for biosynthesis of protein glycans containing N,N’-diacetylbacillosamine. Biochemistry 50(22): 4936-4948.Harvey, H., J. V. Kus, L. Tessier, J. Kelly, et al. (2011). Pseudomonas aeruginosa D-arabinofuranose biosynthetic pathway and its role in type IV pilus assembly. J Biol Chem 286(32): 28128-28137.Hegge, F. T., P. G. Hitchen, F. E. Aas, H. Kristiansen, et al. (2004). Unique modifications with phosphocholine and phosphoethanolamine define alternate antigenic forms of Neisseria gonorrhoeae type IV pili. Proc Natl Acad Sci U S A 101(29): 10798-10803.Henderson, B. (2014). An overview of protein moonlighting in bacterial infection. Biochem Soc Trans 42(6): 1720-1727.Henderson, B. and A. C. Martin (2014). Protein moonlighting: a new factor in biology and medicine. Biochem Soc Trans 42(6): 1671-1678.Hendrickx, A. P., J. M. Budzik, S. Y. Oh and O. Schneewind (2011). Architects at the bacterial surface - sortases and the assembly of pili with isopeptide bonds. Nat Rev Microbiol 9(3): 166-176.Herrmann, J. L., R. Delahay, A. Gallagher, B. Robertson, et al. (2000). Analysis of post-translational modification of mycobacterial proteins using a cassette expression system. Febs Letters 473(3): 358-362.Herrmann, J. L., P. O’Gaora, A. Gallagher, J. E. Thole, et al. (1996). Bacterial glycoproteins: a link between glycosylation and proteolytic cleavage of a 19 kDa antigen from Mycobacterium tuberculosis. EMBO J 15(14): 3547-3554.Hinterdorfer, P. and Y. F. Dufrene (2006). Detection and localization of single molecular recognition events using atomic force microscopy. Nat Methods 3(5): 347-355.Hitchen, P., J. Brzostek, M. Panico, J. A. Butler, et al. (2010). Modification of the Campylobacter jejuni flagellin glycan by the product of the Cj1295 homopolymeric-tract-containing gene. Microbiology 156(Pt 7): 1953-1962. Hitchen, P. G. and A. Dell (2006). Bacterial glycoproteomics. Microbiology 152(Pt 6): 1575-1580.Hopf, P. S., R. S. Ford, N. Zebian, A. Merkx-Jacques, et al. (2011). Protein glycosylation in Helicobacter pylori: beyond the flagellins? PLoS One 6(9): e25722.Horling, L., W. L. Neuhuber and M. Raab (2012). Pitfalls using tyramide signal amplification (TSA) in the mouse gastrointestinal tract: endogenous streptavidin-binding sites lead to false positive staining. J Neurosci Methods 204(1): 124-132.Horn, C., A. Namane, P. Pescher, M. Riviere, et al. (1999). Decreased capacity of recombinant 45/47-kDa molecules (Apa) of Mycobacterium tuberculosis to stimulate T lymphocyte responses related to changes in their mannosylation pattern. J Biol Chem 274(45): 32023-32030.Horzempa, J., J. E. Comer, S. A. Davis and P. Castric (2006a). Glycosylation substrate specificity of Pseudomonas aeruginosa 1244 pilin. J Biol Chem 281(2): 1128-1136.Horzempa, J., C. R. Dean, J. B. Goldberg and P. Castric (2006b). Pseudomonas aeruginosa 1244 pilin glycosylation: glycan substrate recognition. J Bacteriol 188(12): 4244-4252.Howard, S. L., A. Jagannathan, E. C. Soo, J. P. Hui, et al. (2009). Campylobacter jejuni glycosylation island important in cell charge, legionaminic acid biosynthesis, and colonization of chickens. Infect Immun 77(6): 2544-2556.

186

refe

renc

es

Hsu, K. L., K. T. Pilobello and L. K. Mahal (2006). Analyzing the dynamic bacterial glycome with a lectin microarray approach. Nat Chem Biol 2(3): 153-157.Huang, L., C. W. Forsberg and D. Y. Thomas (1988). Purification and characterization of a chloride-stimulated cellobiosidase from Bacteroides succinogenes S85. J Bacteriol 170(7): 2923-2932.Huard, C., G. Miranda, F. Wessner, A. Bolotin, et al. (2003). Characterization of AcmB, an N-acetylglucosaminidase autolysin from Lactococcus lactis. Microbiology 149(Pt 3): 695-705.Huberts, D. H., H. Venselaar, G. Vriend, M. Veenhuis, et al. (2010). The moonlighting function of pyruvate carboxylase resides in the non-catalytic end of the TIM barrel. Biochim Biophys Acta 1803(9): 1038-1042.Hug, I., M. R. Couturier, M. M. Rooker, D. E. Taylor, et al. (2010). Helicobacter pylori lipopolysaccharide is synthesized via a novel pathway with an evolutionary connection to protein N-glycosylation. PLoS Pathog 6(3): e1000819.Hug, I. and M. F. Feldman (2011). Analogies and homologies in lipopolysaccharide and glycoprotein biosynthesis in bacteria. Glycobiology 21(2): 138-151.Hutter, L. J. and J. Bechhoefer (1993). Calibration of atomic-force microscope tips. Review of Scientific Instruments 64.Ihssen, J., M. Kowarik, S. Dilettoso, C. Tanner, et al. (2010). Production of glycoprotein vaccines in Escherichia coli. Microb Cell Fact 9: 61.Ihssen, J., M. Kowarik, L. Wiesli, R. Reiss, et al. (2012). Structural insights from random mutagenesis of Campylobacter jejuni oligosaccharyltransferase PglB. BMC Biotechnol 12: 67.Ikeda, M., M. Wachi, H. K. Jung, F. Ishino, et al. (1991). The Escherichia coli mraY gene encoding UDP-N-acetylmuramoyl-pentapeptide: undecaprenyl-phosphate phospho-N-acetylmuramoyl-pentapeptide transferase. J Bacteriol 173(3): 1021-1026.Inoue, T., H. Ohta, I. Tanimoto, R. Shingaki, et al. (2000). Heterogeneous post-translational modification of Actinobacillus actinomycetemcomitans fimbrillin. Microbiol Immunol 44(8): 715-718.Islam, S. T. and J. S. Lam (2013). Wzx flippase-mediated membrane translocation of sugar polymer precursors in bacteria. Environ Microbiol 15(4): 1001-1015.Iwashkiw, J. A., M. A. Fentabil, A. Faridmoayer, D. C. Mills, et al. (2012a). Exploiting the Campylobacter jejuni protein glycosylation system for glycoengineering vaccines and diagnostic tools directed against brucellosis. Microb Cell Fact 11: 13.Iwashkiw, J. A., A. Seper, B. S. Weber, N. E. Scott, et al. (2012b). Identification of a general O-linked protein glycosylation system in Acinetobacter baumannii and its role in virulence and biofilm formation. PLoS Pathog 8(6): e1002758.Izquierdo, E., P. Horvatovich, E. Marchioni, D. Aoude-Werner, et al. (2009). 2-DE and MS analysis of key proteins in the adhesion of Lactobacillus plantarum, a first step toward early selection of probiotics based on bacterial biomarkers. Electrophoresis 30(6): 949-956.Jain, S., D. H. Hreczuk-Hirst, B. McCormack, M. Mital, et al. (2003). Polysialylated insulin: synthesis, characterization and biological activity in vivo. Biochim Biophys Acta 1622(1): 42-49.Jakeman, D. L. (2011). Mechanisms of glycosyltransferases: the in and the out. Chembiochem 12(17): 2540-2542.Janovska, S., I. Pavkova, M. Hubalek, J. Lenco, et al. (2007). Identification of immunoreactive antigens in membrane proteins enriched fraction from Francisella tularensis LVS. Immunol Lett 108(2): 151-159.Jeffery, C. J. (2003). Moonlighting proteins: old proteins learning new tricks. Trends Genet 19(8): 415-417.Jennings, M. P., F. E. Jen, L. F. Roddam, M. A. Apicella, et al. (2011). Neisseria gonorrhoeae pilin glycan contributes to CR3 activation during challenge of primary cervical epithelial cells. Cell Microbiol 13(6): 885-896.Jennings, M. P., M. Virji, D. Evans, V. Foster, et al. (1998). Identification of a novel gene involved in pilin glycosylation in Neisseria meningitidis. Mol Microbiol 29(4): 975-984.Jervis, A. J., R. Langdon, P. Hitchen, A. J. Lawson, et al. (2010). Characterization of N-linked protein glycosylation in Helicobacter pullorum. J Bacteriol 192(19): 5228-5236.Jiang, X. M., B. Neal, F. Santiago, S. J. Lee, et al. (1991). Structure and sequence of the rfb (O antigen) gene cluster of Salmonella serovar typhimurium (strain LT2). Mol Microbiol 5(3): 695-713.Johannessen, C., M. Koomey and B. Borud (2012). Hypomorphic glycosyltransferase alleles and recoding at contingency loci influence glycan microheterogeneity in the protein glycosylation system of Neisseria species. J Bacteriol 194(18): 5034-5043.

refe

renc

es.

187

Jones, M. B., J. N. Rosenberg, M. J. Betenbaugh and S. S. Krag (2009). Structure and synthesis of polyisoprenoids used in N-glycosylation across the three domains of life. Biochim Biophys Acta 1790(6): 485-494.Josenhans, C., R. L. Ferrero, A. Labigne and S. Suerbaum (1999). Cloning and allelic exchange mutagenesis of two flagellin genes of Helicobacter felis. Mol Microbiol 33(2): 350-362.Josenhans, C., L. Vossebein, S. Friedrich and S. Suerbaum (2002). The neuA/flmD gene cluster of Helicobacter pylori is involved in flagellar biosynthesis and flagellin glycosylation. FEMS Microbiol Lett 210(2): 165-172.Kahler, C. M., L. E. Martin, Y. L. Tzeng, Y. K. Miller, et al. (2001). Polymorphisms in pilin glycosylation Locus of Neisseria meningitidis expressing class II pili. Infect Immun 69(6): 3597-3604.Kakuda, T. and V. J. DiRita (2006). Cj1496c encodes a Campylobacter jejuni glycoprotein that influences invasion of human epithelial cells and colonization of the chick gastrointestinal tract. Infect Immun 74(8): 4715-4723.Kang, H. J., F. Coulibaly, F. Clow, T. Proft, et al. (2007). Stabilizing isopeptide bonds revealed in gram-positive bacterial pilus structure. Science 318(5856): 1625-1628. Kang, D., E. S. Ji, M. H. Moon and J. S. Yoo (2010). Lectin-based enrichment method for glycoproteomics using hollow fiber flow field-flow fractionation: application to Streptococcus pyogenes. J Proteome Res 9(6): 2855-2862.Kankainen, M., L. Paulin, S. Tynkkynen, I. von Ossowski, et al. (2009). Comparative genomic analysis of Lactobacillus rhamnosus GG reveals pili containing a human- mucus binding protein. Proc Natl Acad Sci U S A 106(40): 17193-17198.Karlyshev, A. V., P. Everest, D. Linton, S. Cawthraw, et al. (2004). The Campylobacter jejuni general glycosylation system is important for attachment to human epithelial cells and in the colonization of chicks. Microbiology 150(Pt 6): 1957-1964.Karlyshev, A. V., J. M. Ketley and B. W. Wren (2005). The Campylobacter jejuni glycome. FEMS Microbiol Rev 29(2): 377-390.Karlyshev, A. V., D. Linton, N. A. Gregson and B. W. Wren (2002). A novel paralogous gene family involved in phase-variable flagella-mediated motility in Campylobacter jejuni. Microbiology 148(Pt 2): 473-480.Kawamura, T. and G. D. Shockman (1983). Purification and some properties of the endogenous, autolytic N-acetylmuramoylhydrolase of Streptococcus faecium, a bacterial glycoenzyme. J Biol Chem 258(15): 9514-9521.Kay, E., V. I. Lesk, A. Tamaddoni-Nezhad, P. G. Hitchen, et al. (2010). Systems analysis of bacterial glycomes. Biochem Soc Trans 38(5): 1290-1293.Keenleyside, W. J. and C. Whitfield (1996). A novel pathway for O-polysaccharide biosynthesis in Salmonella enterica serovar Borreze. J Biol Chem 271(45): 28581-28592.Kelly, J., H. Jarrell, L. Millar, L. Tessier, et al. (2006). Biosynthesis of the N-linked glycan in Campylobacter jejuni and addition onto protein through block transfer. J Bacteriol 188(7): 2427-2434.Kessler, C. (1991). The digoxigenin:anti-digoxigenin (DIG) technology--a survey on the concept and realization of a novel bioanalytical indicator system. Mol Cell Probes 5(3): 161-205.Kiel, J. A., J. M. Boels, G. Beldman and G. Venema (1994). Glycogen in Bacillus subtilis: molecular characterization of an operon encoding enzymes involved in glycogen biosynthesis and degradation. Mol Microbiol 11(1): 203-218. Kiessling, L. L. and R. A. Splain (2010). Chemical approaches to glycobiology. Annu Rev Biochem 79: 619-653.Kim, H. J., P. J. Brennan, D. Heaslip, M. C. Udey, et al. (2015). Carbohydrate-Dependent Binding of Langerin to SodC, a Cell Wall Glycoprotein of Mycobacterium leprae. J Bacteriol 197(3): 615-625.Kinoshita, H., H. Uchida, Y. Kawai, T. Kawasaki, et al. (2008a). Cell surface Lactobacillus plantarum LA 318 glyceraldehyde-3-phosphate dehydrogenase (GAPDH) adheres to human colonic mucin. J Appl Microbiol 104(6): 1667-1674.Kinoshita, H., N. Wakahara, M. Watanabe, T. Kawasaki, et al. (2008b). Cell surface glyceraldehyde-3-phosphate dehydrogenase (GAPDH) of Lactobacillus plantarum LA 318 recognizes human A and B blood group antigens. Res Microbiol 159(9-10): 685-691.Kline, K. A., K. W. Dodson, M. G. Caparon and S. J. Hultgren (2010). A tale of two pili: assembly and function of pili in bacteria. Trends Microbiol 18(5): 224-232.Knauer, R. and L. Lehle (1999a). The oligosaccharyltransferase complex from Saccharomyces cerevisiae. Isolation of the OST6 gene, its synthetic interaction with OST3, and analysis of the native complex. J Biol Chem 274(24): 17249-17256.Knauer, R. and L. Lehle (1999b). The oligosaccharyltransferase complex from yeast. Biochim Biophys Acta 1426(2): 259-273.

188

refe

renc

es

Knudsen, S. K., A. Stensballe, M. Franzmann, U. B. Westergaard, et al. (2008). Effect of glycosylation on the extracellular domain of the Ag43 bacterial autotransporter: enhanced stability and reduced cellular aggregation. Biochem J 412(3): 563-577.Konstantinov, S. R., H. Smidt, W. M. de Vos, S. C. Bruijns, et al. (2008). S layer protein A of Lactobacillus acidophilus NCFM regulates immature dendritic cell and T cell functions. Proc Natl Acad Sci U S A 105(49): 19474-19479.Kowarik, M., S. Numao, M. F. Feldman, B. L. Schulz, et al. (2006a). N-linked glycosylation of folded proteins by the bacterial oligosaccharyltransferase. Science 314(5802): 1148-1150.Kowarik, M., N. M. Young, S. Numao, B. L. Schulz, et al. (2006b). Definition of the bacterial N-glycosylation site consensus sequence. EMBO J 25(9): 1957-1966.Krogfelt, K. A. (1991). Bacterial adhesion: genetics, biogenesis, and role in pathogenesis of fimbrial adhesins of Escherichia coli. Rev Infect Dis 13(4): 721-735.Ku, S. C., B. L. Schulz, P. M. Power and M. P. Jennings (2009). The pilin O-glycosylation pathway of pathogenic Neisseria is a general system that glycosylates AniA, an outer membrane nitrite reductase. Biochem Biophys Res Commun 378(1): 84-89.Kumar, M., N. Mishra and R. K. Upreti (2003). A novel membrane glycoprotein of Escherichia coli. J Basic Microbiol 43(1): 28-35.Kuo, C., N. Takahashi, A. F. Swanson, Y. Ozeki, et al. (1996). An N-linked high-mannose type oligosaccharide, expressed at the major outer membrane protein of Chlamydia trachomatis, mediates attachment and infectivity of the microorganism to HeLa cells. J Clin Invest 98(12): 2813-2818.Kus, J. V., J. Kelly, L. Tessier, H. Harvey, et al. (2008). Modification of Pseudomonas aeruginosa Pa5196 type IV Pilins at multiple sites with D-Araf by a novel GT-C family Arabinosyltransferase, TfpW. J Bacteriol 190(22): 7464-7478.Kus, J. V., E. Tullis, D. G. Cvitkovitch and L. L. Burrows (2004). Significant differences in type IV pilin allele distribution among Pseudomonas aeruginosa isolates from cystic fibrosis (CF) versus non-CF patients. Microbiology 150(Pt 5): 1315-1326.Laakso, K., K. Koskenniemi, J. Koponen, M. Kankainen, et al. (2011). Growth phase-associated changes in the proteome and transcriptome of Lactobacillus rhamnosus GG in industrial-type whey medium. Microb Biotechnol 4(6): 746-766.Laine, R. A. (1994). A calculation of all possible oligosaccharide isomers both branched and linear yields 1.05 x 10(12) structures for a reducing hexasaccharide: the Isomer Barrier to development of single-method saccharide sequencing or synthesis systems. Glycobiology 4(6): 759-767.Lairson, L. L., C. P. Chiu, H. D. Ly, S. He, et al. (2004). Intermediate trapping on a mutant retaining alpha-galactosyltransferase identifies an unexpected aspartate residue. J Biol Chem 279(27): 28339-28344.Lairson, L. L., B. Henrissat, G. J. Davies and S. G. Withers (2008). Glycosyltransferases: structures, functions, and mechanisms. Annu Rev Biochem 77: 521-555.Laitinen, O. H., V. P. Hytonen, H. R. Nordlund and M. S. Kulomaa (2006). Genetically engineered avidins and streptavidins. Cell Mol Life Sci 63(24): 2992-3017.Lara, M., L. Servin-Gonzalez, M. Singh, C. Moreno, et al. (2004). Expression, secretion, and glycosylation of the 45- and 47-kDa glycoprotein of Mycobacterium tuberculosis in Streptomyces lividans. Appl Environ Microbiol 70(2): 679-685.Larkin, A. and B. Imperiali (2011). The expanding horizons of asparagine-linked glycosylation. Biochemistry 50(21): 4411-4426.Larsen, J. C., C. Szymanski and P. Guerry (2004). N-linked protein glycosylation is required for full competence in Campylobacter jejuni 81-176. J Bacteriol 186(19): 6508-6514.Lebeer, S., I. Claes, H. L. Tytgat, T. L. Verhoeven, et al. (2012a). Functional Analysis of Lactobacillus rhamnosus GG Pili in Relation to Adhesion and Immunomodulatory Interactions with Intestinal Epithelial Cells. Appl Environ Microbiol 78(1): 185-193.Lebeer, S., I. J. Claes, C. I. Balog, G. Schoofs, et al. (2012b). The major secreted protein Msp1/p75 is O-glycosylated in Lactobacillus rhamnosus GG. Microb Cell Fact 11(1): 15.Lebeer, S., I. J. Claes, T. L. Verhoeven, J. Vanderleyden, et al. (2011). Exopolysaccharides of Lactobacillus rhamnosus GG form a protective shield against innate immune factors in the intestine. Microb Biotechnol 4(3): 368-374.Lebeer, S., J. Vanderleyden and S. C. De Keersmaecker (2008). Genes and molecules of lactobacilli supporting

refe

renc

es.

189

probiotic action. Microbiol Mol Biol Rev 72(4): 728-764, Table of Contents.Lebeer, S., J. Vanderleyden and S. C. De Keersmaecker (2010). Host interactions of probiotic bacterial surface molecules: comparison with commensals and pathogens. Nat Rev Microbiol 8(3): 171-184.Lebeer, S., T. L. Verhoeven, G. Francius, G. Schoofs, et al. (2009). Identification of a Gene Cluster for the Biosynthesis of a Long, Galactose-Rich Exopolysaccharide in Lactobacillus rhamnosus GG and Functional Analysis of the Priming Glycosyltransferase. Appl Environ Microbiol 75(11): 3554-3563.Lee, I. C., S. van, II, S. Tomita, P. Morsomme, et al. (2014). GtfA and GtfB are both required for protein O-glycosylation in Lactobacillus plantarum. J Bacteriol 196(9): 1671-1682.Lee, S. S., S. Y. Hong, J. C. Errey, A. Izumi, et al. (2011). Mechanistic evidence for a front-side, SNi-type reaction in a retaining glycosyltransferase. Nat Chem Biol 7(9): 631-638.Lee, S. W., M. Sabet, H. S. Um, J. Yang, et al. (2006). Identification and characterization of the genes encoding a unique surface (S-) layer of Tannerella forsythia. Gene 371(1): 102-111.Lees-Miller, R. G., J. A. Iwashkiw, N. E. Scott, A. Seper, et al. (2013). A common pathway for O-linked protein-glycosylation and synthesis of capsule in Acinetobacter baumannii. Mol Microbiol 89(5): 816-830.Lehrer, J., K. A. Vigeant, L. D. Tatar and M. A. Valvano (2007). Functional characterization and membrane topology of Escherichia coli WecA, a sugar-phosphate transferase initiating the biosynthesis of enterobacterial common antigen and O-antigen lipopolysaccharide. J Bacteriol 189(7): 2618-2628.Lemon, K. P., D. E. Higgins and R. Kolter (2007). Flagellar motility is critical for Listeria monocytogenes biofilm formation. J Bacteriol 189(12): 4418-4424.Lerouge, I. and J. Vanderleyden (2002). O-antigen structural variation: mechanisms and possible roles in animal/plant-microbe interactions. FEMS Microbiol Rev 26(1): 17-47.Levesque, C., C. Vadeboncoeur, F. Chandad and M. Frenette (2001). Streptococcus salivarius fimbriae are composed of a glycoprotein containing a repeated motif assembled into a filamentous nondissociable structure. J Bacteriol 183(9): 2724-2732.Li, L., R. Woodward, Y. Ding, X. W. Liu, et al. (2010). Overexpression and topology of bacterial oligosaccharyltransferase PglB. Biochem Biophys Res Commun 394(4): 1069-1074.Lim, W. J., S. R. Park, M. K. Kim, C. L. An, et al. (2003). Cloning and characterization of the glycogen branching enzyme gene existing in tandem with the glycogen debranching enzyme from Pectobacterium chrysanthemi PY35. Biochem Biophys Res Commun 300(1): 93-101.Lindenthal, C. and E. A. Elsinghorst (1999). Identification of a glycoprotein produced by enterotoxigenic Escherichia coli. Infect Immun 67(8): 4084-4091.Lindenthal, C. and E. A. Elsinghorst (2001). Enterotoxigenic Escherichia coli TibA glycoprotein adheres to human intestine epithelial cells. Infect Immun 69(1): 52-57.Lindhout, T., U. Iqbal, L. M. Willis, A. N. Reid, et al. (2011). Site-specific enzymatic polysialylation of therapeutic proteins using bacterial enzymes. Proc Natl Acad Sci U S A 108(18): 7397-7402.Lindhout, T., P. C. Lau, D. Brewer and J. S. Lam (2009). Truncation in the core oligosaccharide of lipopolysaccharide affects flagella-mediated motility in Pseudomonas aeruginosa PAO1 via modulation of cell surface attachment. Microbiology 155(Pt 10): 3449-3460.Linton, D., E. Allan, A. V. Karlyshev, A. D. Cronshaw, et al. (2002). Identification of N-acetylgalactosamine-containing glycoproteins PEB3 and CgpA in Campylobacter jejuni. Mol Microbiol 43(2): 497-508.Linton, D., N. Dorrell, P. G. Hitchen, S. Amber, et al. (2005). Functional analysis of the Campylobacter jejuni N-linked protein glycosylation pathway. Mol Microbiol 55(6): 1695-1703.Linton, D., A. V. Karlyshev and B. W. Wren (2001). Deciphering Campylobacter jejuni cell surface interactions from the genome sequence. Curr Opin Microbiol 4(1): 35-40.Liu, C. F., L. Tonini, W. Malaga, M. Beau, et al. (2013). Bacterial protein-O-mannosylating enzyme is crucial for virulence of Mycobacterium tuberculosis. Proc Natl Acad Sci U S A 110(16): 6560-6565.Lizak, C., Y. Y. Fan, T. C. Weber and M. Aebi (2011a). N-Linked glycosylation of antibody fragments in Escherichia coli. Bioconjug Chem 22(3): 488-496.Lizak, C., S. Gerber, S. Numao, M. Aebi, et al. (2011b). X-ray structure of a bacterial oligosaccharyltransferase. Nature 474(7351): 350-355.Lobley, A., M. I. Sadowski and D. T. Jones (2009). pGenTHREADER and pDomTHREADER: new methods for improved protein fold recognition and superfamily discrimination. Bioinformatics 25(14): 1761-1767.

190

refe

renc

es

Logan, S. M. (2006). Flagellar glycosylation - a new component of the motility repertoire? Microbiology 152(Pt 5): 1249-1262.Logan, S. M., J. P. Hui, E. Vinogradov, A. J. Aubry, et al. (2009). Identification of novel carbohydrate modifications on Campylobacter jejuni 11168 flagellin using metabolomics-based approaches. FEBS J 276(4): 1014-1023.Logan, S. M., J. F. Kelly, P. Thibault, C. P. Ewing, et al. (2002). Structural heterogeneity of carbohydrate modifications affects serospecificity of Campylobacter flagellins. Mol Microbiol 46(2): 587-597.Logan, S. M., T. J. Trust and P. Guerry (1989). Evidence for posttranslational modification and gene duplication of Campylobacter flagellin. J Bacteriol 171(6): 3031-3038.Loman, N. J., L. A. Snyder, J. D. Linton, R. Langdon, et al. (2009). Genome sequence of the emerging pathogen Helicobacter canadensis. J Bacteriol 191(17): 5566-5567.Lovering, A. L., L. H. de Castro, D. Lim and N. C. Strynadka (2007). Structural insight into the transglycosylation step of bacterial cell-wall biosynthesis. Science 315(5817): 1402-1405.Lyristis, M., Z. L. Boynton, D. Petersen, Z. Kan, et al. (2000). Cloning, sequencing, and characterization of the gene encoding flagellin, flaC, and the post-translational modification of flagellin, FlaC, from Clostridium acetobutylicum ATCC824. Anaerobe 6(2): 69-79.Maldonado-Barragan, A., B. Caballero-Guerrero, H. Lucena-Padros and J. L. Ruiz-Barba (2011). Genome sequence of Lactobacillus pentosus IG1, a strain isolated from Spanish-style green olive fermentations. J Bacteriol 193(19): 5605.Malik, S., R. J. Siezen, B. Renckens, M. Vaneechoutte, et al. (2014). Draft Genome Sequence of Lactobacillus plantarum CMPG5300, a Human Vaginal Isolate. Genome Announc 2(6).Mandlik, A., A. Swierczynski, A. Das and H. Ton-That (2008). Pili in Gram-positive bacteria: assembly, involvement in colonization and biofilm development. Trends Microbiol 16(1): 33-40.Marceau, M., K. Forest, J. L. Beretti, J. Tainer, et al. (1998). Consequences of the loss of O-linked glycosylation of meningococcal type IV pilin on piliation and pilus-mediated adhesion. Mol Microbiol 27(4): 705-715.Marceau, M. and X. Nassif (1999). Role of glycosylation at Ser63 in production of soluble pilin in pathogenic Neisseria. J Bacteriol 181(2): 656-661.Marino, K., J. Bones, J. J. Kattla and P. M. Rudd (2010). A systematic approach to protein glycosylation analysis: a path through the maze. Nature Chemical Biology 6(10): 713-723.Marolda, C. L., J. Vicarioli and M. A. Valvano (2004). Wzx proteins involved in biosynthesis of O antigen function in association with the first sugar of the O-specific lipopolysaccharide subunit. Microbiology 150(Pt 12): 4095-4105.Mastroleo, F., B. Leroy, R. Van Houdt, C. s’ Heeren, et al. (2009). Shotgun proteome analysis of Rhodospirillum rubrum S1H: integrating data from gel-free and gel-based peptides fractionation methods. J Proteome Res 8(5): 2530-2541.Matias, V. R. and T. J. Beveridge (2005). Cryo-electron microscopy reveals native polymeric cell wall structure in Bacillus subtilis 168 and the existence of a periplasmic space. Mol Microbiol 56(1): 240-251.Matias, V. R. and T. J. Beveridge (2006). Native cell wall organization shown by cryo-electron microscopy confirms the existence of a periplasmic space in Staphylococcus aureus. J Bacteriol 188(3): 1011-1021.McBride, J. W., J. E. Comer and D. H. Walker (2003). Novel Immunoreactive glycoprotein orthologs of Ehrlichia spp. Ann N Y Acad Sci 990: 678-684.McBride, J. W., X. J. Yu and D. H. Walker (2000). Glycosylation of homologous immunodominant proteins of Ehrlichia chaffeensis and Ehrlichia canis. Infect Immun 68(1): 13-18.McKay, B. E., M. L. Molineux and R. W. Turner (2008). Endogenous biotin in rat brain: implications for false-positive results with avidin-biotin and streptavidin-biotin techniques. Methods Mol Biol 418: 111-128.McNally, D. J., A. J. Aubry, J. P. Hui, N. H. Khieu, et al. (2007). Targeted metabolomics analysis of Campylobacter coli VC167 reveals legionaminic acid derivatives as novel flagellar glycans. J Biol Chem 282(19): 14463-14475.Meier-Dieter, U., K. Barr, R. Starman, L. Hatch, et al. (1992). Nucleotide sequence of the Escherichia coli rfe gene involved in the synthesis of enterobacterial common antigen. Molecular cloning of the rfe-rff gene cluster. J Biol Chem 267(2): 746-753.Mengin-Lecreulx, D., L. Texier, M. Rousseau and J. van Heijenoort (1991). The murG gene of Escherichia coli codes for the UDP-N-acetylglucosamine: N-acetylmuramyl-(pentapeptide) pyrophosphoryl-undecaprenol N-acetylglucosamine transferase involved in the membrane steps of peptidoglycan synthesis. J Bacteriol 173(15): 4625-4636.

refe

renc

es.

191

Menozzi, F. D., R. Bischoff, E. Fort, M. J. Brennan, et al. (1998). Molecular characterization of the mycobacterial heparin-binding hemagglutinin, a mycobacterial adhesin. Proc Natl Acad Sci U S A 95(21): 12625-12630. Merritt, J. H., A. A. Ollis, A. C. Fisher and M. P. Delisa (2013). Glycans-by-design: engineering bacteria for the biosynthesis of complex glycans and glycoconjugates. Biotechnol Bioeng 110(6): 1550-1564.Mescher, M. F. and J. L. Strominger (1976). Purification and characterization of a prokaryotic glucoprotein from the cell envelope of Halobacterium salinarium. J Biol Chem 251(7): 2005-2014.Messner, P. (1997). Bacterial glycoproteins. Glycoconj J 14(1): 3-11.Messner, P. (2004). Prokaryotic glycoproteins: unexplored but important. J Bacteriol 186(9): 2517-2519.Messner, P., K. Steiner, K. Zarschler and C. Schaffer (2008). S-layer nanoglycobiology of bacteria. Carbohydr Res 343(12): 1934-1951.Michell, S. L., A. O. Whelan, P. R. Wheeler, M. Panico, et al. (2003). The MPB83 antigen from Mycobacterium bovis contains O-linked mannose and (1-->3)-mannobiose moieties. J Biol Chem 278(18): 16423-16432.Mistou, M. Y., S. Dramsi, S. Brega, C. Poyart, et al. (2009). Molecular dissection of the secA2 locus of group B Streptococcus reveals that glycosylation of the Srr1 LPXTG protein is required for full virulence. J Bacteriol 191(13): 4195-4206.Mohammadi, T., A. Karczmarek, M. Crouvoisier, A. Bouhss, et al. (2007). The essential peptidoglycan glycosyltransferase MurG forms a complex with proteins involved in lateral envelope growth as well as with proteins involved in cell division in Escherichia coli. Mol Microbiol 65(4): 1106-1121.Moormann, C., I. Benz and M. A. Schmidt (2002). Functional substitution of the TibC protein of enterotoxigenic Escherichia coli strains for the autotransporter adhesin heptosyltransferase of the AIDA system. Infect Immun 70(5): 2264-2270.Morris, E. J., N. Ganeshkumar, M. Song and B. C. McBride (1987). Identification and preliminary characterization of a Streptococcus sanguis fibrillar glycoprotein. J Bacteriol 169(1): 164-171.Murray, B. W., S. Takayama, J. Schultz and C. H. Wong (1996). Mechanism and specificity of human alpha-1,3-fucosyltransferase V. Biochemistry 35(34): 11183-11195.Musumeci, M. A., A. Faridmoayer, Y. Watanabe and M. F. Feldman (2014). Evaluating the role of conserved amino acids in bacterial O-oligosaccharyltransferases by in vivo, in vitro and limited proteolysis assays. Glycobiology 24(1): 39-50.Musumeci, M. A., I. Hug, N. E. Scott, M. V. Ielmini, et al. (2013). In vitro activity of Neisseria meningitidis PglL O-oligosaccharyltransferase with diverse synthetic lipid donors and a UDP-activated sugar. J Biol Chem 288(15): 10578-10587. Mutanda, T., M. P. Mokoena, A. O. Olaniran, B. S. Wilhelmi, et al. (2014). Microbial enzymatic production and applications of short-chain fructooligosaccharides and inulooligosaccharides: recent advances and current perspectives. J Ind Microbiol Biotechnol 41(6): 893-906.Naegeli, A., C. Neupert, Y. Y. Fan, C. W. Lin, et al. (2014). Molecular analysis of an alternative N-glycosylation machinery by functional transfer from Actinobacillus pleuropneumoniae to Escherichia coli. J Biol Chem 289(4): 2170-2179.Nakao, R., Y. Tashiro, N. Nomura, S. Kosono, et al. (2008). Glycosylation of the OMP85 homolog of Porphyromonas gingivalis and its involvement in biofilm formation. Biochem Biophys Res Commun 365(4): 784-789. Nicolaou, K. C. and H. J. Mitchell (2001). Adventures in Carbohydrate Chemistry: New Synthetic Technologies, Chemical Synthesis, Molecular Design, and Chemical Biology. Angew Chem Int Ed Engl 40(9): 1576-1624.Nita-Lazar, M., M. Wacker, B. Schegg, S. Amber, et al. (2005). The N-X-S/T consensus sequence is required but not sufficient for bacterial N-linked protein glycosylation. Glycobiology 15(4): 361-367.Nothaft, H. and C. M. Szymanski (2010). Protein glycosylation in bacteria: sweeter than ever. Nat Rev Microbiol 8(11): 765-778.Okuda, S. and G. Weinbaum (1968). An envelope-specific glycoprotein from Escherichia coli B. Biochemistry 7(8): 2819-2825.Olivier, N. B., M. M. Chen, J. R. Behr and B. Imperiali (2006). In vitro biosynthesis of UDP-N,N’-diacetylbacillosamine by enzymes of the Campylobacter jejuni general protein glycosylation system. Biochemistry 45(45): 13659-13669.Olsen, O. and K. K. Thomsen (1991). Improvement of Bacterial Beta-Glucanase Thermostability by Glycosylation. Journal of General Microbiology 137: 579-585.Pan, S., R. Chen, R. Aebersold and T. A. Brentnall (2011). Mass spectrometry based glycoproteomics--from a

192

refe

renc

es

proteomics perspective. Mol Cell Proteomics 10(1): R110 003251.Pandhal, J. and P. C. Wright (2010). N-Linked glycoengineering for human therapeutic proteins in bacteria. Biotechnol Lett 32(9): 1189-1198.Paramonov, N., M. Rangarajan, A. Hashim, A. Gallagher, et al. (2005). Structural analysis of a novel anionic polysaccharide from Porphyromonas gingivalis strain W50 related to Arg-gingipain glycans. Mol Microbiol 58(3): 847-863.Parge, H. E., K. T. Forest, M. J. Hickey, D. A. Christensen, et al. (1995). Structure of the fibre-forming protein pilin at 2.6 A resolution. Nature 378(6552): 32-38.Parker, B. L., G. Palmisano, A. V. Edwards, M. Y. White, et al. (2011). Quantitative N-linked glycoproteomics of myocardial ischemia and reperfusion injury reveals early remodeling in the extracellular environment. Mol Cell Proteomics 10(8): M110 006833.Parkhill, J., B. W. Wren, K. Mungall, J. M. Ketley, et al. (2000). The genome sequence of the food-borne pathogen Campylobacter jejuni reveals hypervariable sequences. Nature 403(6770): 665-668.Peel, M., W. Donachie and A. Shaw (1988). Physical and antigenic heterogeneity in the flagellins of Listeria monocytogenes and L. ivanovii. J Gen Microbiol 134(9): 2593-2598.Pei, Z. H., R. T. Ellison, 3rd and M. J. Blaser (1991). Identification, purification, and characterization of major antigenic proteins of Campylobacter jejuni. J Biol Chem 266(25): 16363-16369.Peng, Z., P. Fives-Taylor, T. Ruiz, M. Zhou, et al. (2008a). Identification of critical residues in Gap3 of Streptococcus parasanguinis involved in Fap1 glycosylation, fimbrial formation and in vitro adhesion. BMC Microbiol 8: 52.Peng, Z., H. Wu, T. Ruiz, Q. Chen, et al. (2008b). Role of gap3 in Fap1 glycosylation, stability, in vitro adhesion, and fimbrial and biofilm formation of Streptococcus parasanguinis. Oral Microbiol Immunol 23(1): 70-78.Perea Velez, M., T. L. Verhoeven, C. Draing, S. Von Aulock, et al. (2007). Functional analysis of D-alanylation of lipoteichoic acid in the probiotic strain Lactobacillus rhamnosus GG. Appl Environ Microbiol 73(11): 3595-3604.Peregrin-Alvarez, J. M., X. Xiong, C. Su and J. Parkinson (2009). The Modular Organization of Protein Interactions in Escherichia coli. PLoS Comput Biol 5(10): e1000523.Perlstein, D. L., Y. Zhang, T. S. Wang, D. E. Kahne, et al. (2007). The direction of glycan chain elongation by peptidoglycan glycosyltransferases. J Am Chem Soc 129(42): 12674-12675.Persson, K., H. D. Ly, M. Dieckelmann, W. W. Wakarchuk, et al. (2001). Crystal structure of the retaining galactosyltransferase LgtC from Neisseria meningitidis in complex with donor and acceptor sugar analogs. Nat Struct Biol 8(2): 166-175.Pimkin, M., C. G. Miller, L. Blakesley, C. A. Oleykowski, et al. (2006). Characterization of a periplasmic S1-like nuclease coded by the Mesorhizobium loti symbiosis island. Biochem Biophys Res Commun 343(1): 77-84.Plummer, C., H. Wu, S. W. Kerrigan, G. Meade, et al. (2005). A serine-rich glycoprotein of Streptococcus sanguis mediates adhesion to platelets via GPIb. Br J Haematol 129(1): 101-109.Plummer, T. H., Jr., A. L. Tarentino and C. R. Hauer (1995). Novel, specific O-glycosylation of secreted Flavobacterium meningosepticum proteins. Asp-Ser and Asp-Thr-Thr consensus sites. J Biol Chem 270(22): 13192-13196.Posch, G., M. Pabst, L. Brecker, F. Altmann, et al. (2011). Characterization and scope of S-layer protein O-glycosylation in Tannerella forsythia. J Biol Chem 286(44): 38714-38724.Posch, G., M. Pabst, L. Neumann, M. J. Coyne, et al. (2013). “Cross-glycosylation” of proteins in Bacteroidales species. Glycobiology 23(5): 568-577.Power, P. M., L. F. Roddam, M. Dieckelmann, Y. N. Srikhanta, et al. (2000). Genetic characterization of pilin glycosylation in Neisseria meningitidis. Microbiology 146 ( Pt 4): 967-979.Power, P. M., L. F. Roddam, K. Rutter, S. Z. Fitzpatrick, et al. (2003). Genetic characterization of pilin glycosylation and phase variation in Neisseria meningitidis. Mol Microbiol 49(3): 833-847.Power, P. M., K. L. Seib and M. P. Jennings (2006). Pilin glycosylation in Neisseria meningitidis occurs by a similar pathway to wzy-dependent O-antigen biosynthesis in Escherichia coli. Biochem Biophys Res Commun 347(4): 904-908.Price, N. P. and F. A. Momany (2005). Modeling bacterial UDP-HexNAc: polyprenol-P HexNAc-1-P transferases. Glycobiology 15(9): 29R-42R.Proft, T. and E. N. Baker (2009). Pili in Gram-negative and Gram-positive bacteria - structure, assembly and their role in disease. Cell Mol Life Sci 66(4): 613-635.

refe

renc

es.

193

Provencher, C., G. LaPointe, S. Sirois, M. R. Van Calsteren, et al. (2003). Consensus-degenerate hybrid oligonucleotide primers for amplification of priming glycosyltransferase genes of the exopolysaccharide locus in strains of the Lactobacillus casei group. Appl Environ Microbiol 69(6): 3299-3307.Qin, J., R. Li, J. Raes, M. Arumugam, et al. (2010). A human gut microbial gene catalogue established by metagenomic sequencing. Nature 464(7285): 59-65.Qin, J., Y. Li, Z. Cai, S. Li, et al. (2012). A metagenome-wide association study of gut microbiota in type 2 diabetes. Nature 490(7418): 55-60.Qutyan, M., M. Henkel, J. Horzempa, M. Quinn, et al. (2010). Glycosylation of pilin and nonpilin protein constructs by Pseudomonas aeruginosa 1244. J Bacteriol 192(22): 5972-5981.Qutyan, M., M. Paliotti and P. Castric (2007). PilO of Pseudomonas aeruginosa 1244: subcellular location and domain assignment. Mol Microbiol 66(6): 1444-1458.Rabinovich, G. A., Y. van Kooyk and B. A. Cobb (2012). Glycobiology of immune responses. Ann N Y Acad Sci 1253: 1-15.Raetz, C. R. and C. Whitfield (2002). Lipopolysaccharide endotoxins. Annu Rev Biochem 71: 635-700.Ragas, A., L. Roussel, G. Puzo and M. Riviere (2007). The Mycobacterium tuberculosis cell-surface glycoprotein apa as a potential adhesin to colonize target cells via the innate immune system pulmonary C-type lectin surfactant protein A. J Biol Chem 282(8): 5133-5142. Raju, T. S. and E. A. Davidson (1994). New approach towards deglycosylation of sialoglycoproteins and mucins. Biochem Mol Biol Int 34(5): 943-954.Ramakrishnan, B., E. Boeggeman, V. Ramasamy and P. K. Qasba (2004). Structure and catalytic cycle of beta-1,4-galactosyltransferase. Curr Opin Struct Biol 14(5): 593-600.Ramiah, K., C. A. van Reenen and L. M. Dicks (2008). Surface-bound proteins of Lactobacillus plantarum 423 that contribute to adhesion of Caco-2 cells and their role in competitive exclusion and displacement of Clostridium sporogenes and Enterococcus faecalis. Res Microbiol 159(6): 470-475.Rangarajan, E. S., S. Bhatia, D. C. Watson, C. Munger, et al. (2007). Structural context for protein N-glycosylation in bacteria: The structure of PEB3, an adhesin from Campylobacter jejuni. Protein Sci 16(5): 990-995.Rangarajan, M., A. Hashim, J. Aduse-Opoku, N. Paramonov, et al. (2005). Expression of Arg-Gingipain RgpB is required for correct glycosylation and stability of monomeric Arg-gingipain RgpA from Porphyromonas gingivalis W50. Infect Immun 73(8): 4864-4878.Reeves, P. R., M. Hobbs, M. A. Valvano, M. Skurnik, et al. (1996). Bacterial polysaccharide synthesis and gene nomenclature. Trends Microbiol 4(12): 495-503.Rehm, B. H. and S. Valla (1997). Bacterial alginates: biosynthesis and applications. Appl Microbiol Biotechnol 48(3): 281-288.Reid, C. W., K. M. Fulton and S. M. Twine (2010). Never take candy from a stranger: the role of the bacterial glycome in host-pathogen interactions. Future Microbiol 5(2): 267-288.Reinhold, B. B., C. R. Hauer, T. H. Plummer and V. N. Reinhold (1995). Detailed structural analysis of a novel, specific O-linked glycan from the prokaryote Flavobacterium meningosepticum. J Biol Chem 270(22): 13197-13203.Reunanen, J., I. von Ossowski, A. P. Hendrickx, A. Palva, et al. (2012). Characterization of the SpaCBA pilus fibers in the probiotic Lactobacillus rhamnosus GG. Appl Environ Microbiol 78(7): 2337-2344.Robertson, J. N., P. Vincent and M. E. Ward (1977). The preparation and properties of gonococcal pili. J Gen Microbiol 102(1): 169-177.Rocchetta, H. L., L. L. Burrows, J. C. Pacan and J. S. Lam (1998). Three rhamnosyltransferases responsible for assembly of the A-band D-rhamnan polysaccharide in Pseudomonas aeruginosa: a fourth transferase, WbpL, is required for the initiation of both A-band and B-band lipopolysaccharide synthesis. Mol Microbiol 28(6): 1103-1119.Rogers, E. A., A. Das and H. Ton-That (2011). Adhesion by pathogenic corynebacteria. Adv Exp Med Biol 715: 91-103.Rolain, T., E. Bernard, A. Beaussart, H. Degand, et al. (2013). O-glycosylation as a novel control mechanism of peptidoglycan hydrolase activity. J Biol Chem 288(31): 22233-22247.Romain, F., C. Horn, P. Pescher, A. Namane, et al. (1999). Deglycosylation of the 45/47-kilodalton antigen complex of Mycobacterium tuberculosis decreases its capacity to elicit in vivo or in vitro cellular immune responses. Infect Immun 67(11): 5567-5572.Roychoudhury, R. and N. L. Pohl (2010). New structures, chemical functions, and inhibitors for glycosyltransferases.

194

refe

renc

es

Curr Opin Chem Biol 14(2): 168-173.Saad, N., M. Urdaci, C. Vignoles, S. Chaignepain, et al. (2009). Lactobacillus plantarum 299v surface-bound GAPDH: a new insight into enzyme cell walls location. J Microbiol Biotechnol 19(12): 1635-1643.Salomonsson, E., A. Forsberg, N. Roos, C. Holz, et al. (2009). Functional analyses of pilin-like proteins from Francisella tularensis: complementation of type IV pilus phenotypes in Neisseria gonorrhoeae. Microbiology 155(Pt 8): 2546-2559.Sambri, V., C. Stefanelli and R. Cevenini (1992). Detection of glycoproteins in Borrelia burgdorferi. Arch Microbiol 157(3): 205-208.Sanchez, B., P. Bressollier, S. Chaignepain, J. M. Schmitter, et al. (2009a). Identification of surface-associated proteins in the probiotic bacterium Lactobacillus rhamnosus GG. International Dairy Journal 19: 85-88.Sanchez, B., J. M. Schmitter and M. C. Urdaci (2009b). Identification of novel proteins secreted by Lactobacillus rhamnosus GG grown in de Mann-Rogosa-Sharpe broth. Lett Appl Microbiol 48(5): 618-622.Sanchez-Rodriguez, A., H. L. Tytgat, J. Winderickx, J. Vanderleyden, et al. (2014). A network-based approach to identify substrate classes of bacterial glycosyltransferases. BMC Genomics 15: 349.Sartain, M. J. and J. T. Belisle (2009). N-Terminal clustering of the O-glycosylation sites in the Mycobacterium tuberculosis lipoprotein SodC. Glycobiology 19(1): 38-51.Sauvage, E., F. Kerff, M. Terrak, J. A. Ayala, et al. (2008). The penicillin-binding proteins: structure and role in peptidoglycan biosynthesis. FEMS Microbiol Rev 32(2): 234-258.Saxena, I. M., R. M. Brown, Jr. and T. Dandekar (2001). Structure--function characterization of cellulose synthase: relationship to other glycosyltransferases. Phytochemistry 57(7): 1135-1148.Saxon, E. and C. R. Bertozzi (2001). Chemical and biological strategies for engineering cell surface glycosylation. Annu Rev Cell Dev Biol 17: 1-23.Schaffer, C. and P. Messner (2004). Surface-layer glycoproteins: an example for the diversity of bacterial glycosylation with promising impacts on nanobiotechnology. Glycobiology 14(8): 31R-42R.Schell, M. A., R. L. Ulrich, W. J. Ribot, E. E. Brueggemann, et al. (2007). Type VI secretion is a major virulence determinant in Burkholderia mallei. Mol Microbiol 64(6): 1466-1485. Schiel, J. E. (2012). Glycoprotein analysis using mass spectrometry: unraveling the layers of complexity. Anal Bioanal Chem 404(4): 1141-1149.Schirm, M., S. K. Arora, A. Verma, E. Vinogradov, et al. (2004a). Structural and genetic characterization of glycosylation of type a flagellin in Pseudomonas aeruginosa. J Bacteriol 186(9): 2523-2531.Schirm, M., M. Kalmokoff, A. Aubry, P. Thibault, et al. (2004b). Flagellin from Listeria monocytogenes is glycosylated with beta-O-linked N-acetylglucosamine. J Bacteriol 186(20): 6721-6727.Schirm, M., I. C. Schoenhofen, S. M. Logan, K. C. Waldron, et al. (2005). Identification of unusual bacterial glycosylation by tandem mass spectrometry analyses of intact proteins. Anal Chem 77(23): 7774-7782.Schirm, M., E. C. Soo, A. J. Aubry, J. Austin, et al. (2003). Structural, genetic and functional characterization of the flagellin glycosylation process in Helicobacter pylori. Mol Microbiol 48(6): 1579-1592.Schoenhofen, I. C., V. V. Lunin, J. P. Julien, Y. Li, et al. (2006a). Structural and functional characterization of PseC, an aminotransferase involved in the biosynthesis of pseudaminic acid, an essential flagellar modification in Helicobacter pylori. J Biol Chem 281(13): 8907-8916.Schoenhofen, I. C., D. J. McNally, J. R. Brisson and S. M. Logan (2006b). Elucidation of the CMP-pseudaminic acid pathway in Helicobacter pylori: synthesis from UDP-N-acetylglucosamine by a single enzymatic reaction. Glycobiology 16(9): 8C-14C.Schoenhofen, I. C., D. J. McNally, E. Vinogradov, D. Whitfield, et al. (2006c). Functional characterization of dehydratase/aminotransferase pairs from Helicobacter and Campylobacter: enzymes distinguishing the pseudaminic acid and bacillosamine biosynthetic pathways. J Biol Chem 281(2): 723-732.Schoenhofen, I. C., E. Vinogradov, D. M. Whitfield, J. R. Brisson, et al. (2009). The CMP-legionaminic acid pathway in Campylobacter: biosynthesis involving novel GDP-linked precursors. Glycobiology 19(7): 715-725.Schwarz, F., Y. Y. Fan, M. Schubert and M. Aebi (2011a). Cytoplasmic N-glycosyltransferase of Actinobacillus pleuropneumoniae is an inverting enzyme and recognizes the NX(S/T) consensus sequence. J Biol Chem 286(40): 35267-35274.Schwarz, F., C. Lizak, Y. Y. Fan, S. Fleurkens, et al. (2011b). Relaxed acceptor site specificity of bacterial oligosaccharyltransferase in vivo. Glycobiology 21(1): 45-54.

refe

renc

es.

195

Scott, A. E., S. M. Twine, K. M. Fulton, R. W. Titball, et al. (2011a). Flagellar glycosylation in Burkholderia pseudomallei and Burkholderia thailandensis. J Bacteriol 193(14): 3577-3587.Scott, N. E., D. R. Bogema, A. M. Connolly, L. Falconer, et al. (2009). Mass spectrometric characterization of the surface-associated 42 kDa lipoprotein JlpA as a glycosylated antigen in strains of Campylobacter jejuni. J Proteome Res 8(10): 4654-4664.Scott, N. E., B. L. Parker, A. M. Connolly, J. Paulech, et al. (2011b). Simultaneous glycan-peptide characterization using hydrophilic interaction chromatography and parallel fragmentation by CID, higher energy collisional dissociation, and electron transfer dissociation MS applied to the N-linked glycoproteome of Campylobacter jejuni. Mol Cell Proteomics 10(2): M000031-MCP000201.Seepersaud, R., B. A. Bensing, Y. T. Yen and P. M. Sullam (2012). The accessory Sec protein Asp2 modulates GlcNAc deposition onto the serine-rich repeat glycoprotein GspB. J Bacteriol 194(20): 5564-5575.Segers, M. E. and S. Lebeer (2014). Lactobacillus rhamnosus GG - host interactions. Microbial Cell Factories.Sekot, G., G. Posch, P. Messner, M. Matejka, et al. (2011). Potential of the Tannerella forsythia S-layer to delay the immune response. J Dent Res 90(1): 109-114.Settem, R. P., K. Honma, T. Nakajima, C. Phansopa, et al. (2013). A bacterial glycan core linked to surface (S)-layer proteins modulates host immunity through Th17 suppression. Mucosal Immunol 6(2): 415-426.Shen, A. and D. E. Higgins (2006). The MogR transcriptional repressor regulates nonhierarchal expression of flagellar motility genes and virulence in Listeria monocytogenes. PLoS Pathog 2(4): e30.Sherlock, O., U. Dobrindt, J. B. Jensen, R. Munk Vejborg, et al. (2006). Glycosylation of the self-recognizing Escherichia coli Ag43 autotransporter protein. J Bacteriol 188(5): 1798-1807.Shivshankar, P., C. Sanchez, L. F. Rose and C. J. Orihuela (2009). The Streptococcus pneumoniae adhesin PsrP binds to Keratin 10 on lung cells. Mol Microbiol 73(4): 663-679.Shoji, M., K. Sato, H. Yukitake, Y. Kondo, et al. (2011). Por secretion system-dependent secretion and glycosylation of Porphyromonas gingivalis hemin-binding protein 35. PLoS One 6(6): e21372.Siboo, I. R., D. O. Chaffin, C. E. Rubens and P. M. Sullam (2008). Characterization of the accessory Sec system of Staphylococcus aureus. J Bacteriol 190(18): 6188-6196.Siboo, I. R., H. F. Chambers and P. M. Sullam (2005). Role of SraP, a Serine-Rich Surface Protein of Staphylococcus aureus, in binding to human platelets. Infect Immun 73(4): 2273-2280.Sieling, P. A., P. J. Hill, K. M. Dobos, K. Brookman, et al. (2008). Conserved mycobacterial lipoglycoproteins activate TLR2 but also require glycosylation for MHC class II-restricted T cell activation. J Immunol 180(9): 5833-5842.Silberstein, S., P. G. Collins, D. J. Kelleher and R. Gilmore (1995). The essential OST2 gene encodes the 16-kD subunit of the yeast oligosaccharyltransferase, a highly conserved protein expressed in diverse eukaryotic organisms. J Cell Biol 131(2): 371-383.Slaney, J. M., M. Rangarajan, J. Aduse-Opoku, S. Fawell, et al. (2002). Recognition of the carbohydrate modifications to the RgpA protease of Porphyromonas gingivalis by periodontal patient serum IgG. J Periodontal Res 37(3): 215-222.Sleytr, U. B. (1975). Heterologous reattachment of regular arrays of glycoproteins on bacterial surfaces. Nature 257(5525): 400-402.Sleytr, U. B. and K. J. Thorne (1976). Chemical characterization of the regularly arranged surface layers of Clostridium thermosaccharolyticum and Clostridium thermohydrosulfuricum. J Bacteriol 126(1): 377-383.Slynko, V., M. Schubert, S. Numao, M. Kowarik, et al. (2009). NMR structure determination of a segmentally labeled glycoprotein using in vitro glycosylation. J Am Chem Soc 131(3): 1274-1281.Smedley, J. G., 3rd, E. Jewell, J. Roguskie, J. Horzempa, et al. (2005). Influence of pilin glycosylation on Pseudomonas aeruginosa 1244 pilus function. Infect Immun 73(12): 7922-7931.Soldo, B., V. Lazarevic and D. Karamata (2002). tagO is involved in the synthesis of all anionic cell-wall polymers in Bacillus subtilis 168. Microbiology 148(Pt 7): 2079-2087. Solis, D., N. V. Bovin, A. P. Davis, J. Jimenez-Barbero, et al. (2015). A guide into glycosciences: How chemistry, biochemistry and biology cooperate to crack the sugar code. Biochim Biophys Acta 1850(1): 186-235.Sonck, K. A., G. Kint, G. Schoofs, C. Vander Wauven, et al. (2009). The proteome of Salmonella Typhimurium grown under in vivo-mimicking conditions. Proteomics 9(3): 565-579.Song, X., Y. Lasanajak, B. Xia, J. Heimburg-Molinaro, et al. (2011). Shotgun glycomics: a microarray strategy for functional glycomics. Nat Methods 8(1): 85-90.

196

refe

renc

es

Soriani, M. and J. L. Telford (2010). Relevance of pili in pathogenic streptococci pathogenesis and vaccine development. Future Microbiol 5(5): 735-747.Steiner, K., A. Hanreich, B. Kainz, P. G. Hitchen, et al. (2008). Recombinant glycans on an S-layer self-assembly protein: a new dimension for nanopatterned biomaterials. Small 4(10): 1728-1740.Stephenson, A. E., H. Wu, J. Novak, M. Tomana, et al. (2002). The Fap1 fimbrial adhesin is a glycoprotein: antibodies specific for the glycan moiety block the adhesion of Streptococcus parasanguis in an in vitro tooth model. Mol Microbiol 43(1): 147-157.Stepper, J., S. Shastri, T. S. Loo, J. C. Preston, et al. (2011). Cysteine S-glycosylation, a new post-translational modification found in glycopeptide bacteriocins. FEBS Lett 585(4): 645-650.Sterba, J., M. Vancova, N. Rudenko, M. Golovchenko, et al. (2008). Flagellin and outer surface proteins from Borrelia burgdorferi are not glycosylated. J Bacteriol 190(7): 2619-2623.Stimson, E., M. Virji, K. Makepeace, A. Dell, et al. (1995). Meningococcal pilin: a glycoprotein substituted with digalactosyl 2,4-diacetamido-2,4,6-trideoxyhexose. Mol Microbiol 17(6): 1201-1214.Storey, J. D. and R. Tibshirani (2003). Statistical significance for genomewide studies. Proc Natl Acad Sci U S A 100(16): 9440-9445.Sullan, R. M., A. Beaussart, P. Tripathi, S. Derclaye, et al. (2014). Single-cell force spectroscopy of pili-mediated adhesion. Nanoscale 6(2): 1134-1143.Sun, Y., M. Wang, Q. Wang, B. Cao, et al. (2012). Genetic analysis of the Cronobacter sakazakii O4 to O7 O-antigen gene clusters and development of a PCR assay for identification of all C. sakazakii O serotypes. Appl Environ Microbiol 78(11): 3966-3974.Swanson, A. F. and C. C. Kuo (1991a). The characterization of lectin-binding proteins of Chlamydia trachomatis as glycoproteins. Microb Pathog 10(6): 465-473.Swanson, A. F. and C. C. Kuo (1991b). Evidence that the major outer membrane protein of Chlamydia trachomatis is glycosylated. Infect Immun 59(6): 2120-2125.Swanson, A. F. and C. C. Kuo (1994a). The 32-kDa glycoprotein of Chlamydia trachomatis is an acidic protein that may be involved in the attachment process. FEMS Microbiol Lett 123(1-2): 113-117.Swanson, A. F. and C. C. Kuo (1994b). Binding of the glycan of the major outer membrane protein of Chlamydia trachomatis to HeLa cells. Infect Immun 62(1): 24-28.Szklarczyk, D., A. Franceschini, M. Kuhn, M. Simonovic, et al. (2011). The STRING database in 2011: functional interaction networks of proteins, globally integrated and scored. Nucleic Acids Res 39(Database issue): D561-568.Szymanski, C. M., D. H. Burr and P. Guerry (2002). Campylobacter protein glycosylation affects host cell interactions. Infect Immun 70(4): 2242-2244.Szymanski, C. M., S. M. Logan, D. Linton and B. W. Wren (2003a). Campylobacter--a tale of two protein glycosylation systems. Trends Microbiol 11(5): 233-238.Szymanski, C. M., F. S. Michael, H. C. Jarrell, J. Li, et al. (2003b). Detection of conserved N-linked glycans and phase-variable lipooligosaccharides and capsules from campylobacter cells by mass spectrometry and high resolution magic angle spinning NMR spectroscopy. J Biol Chem 278(27): 24509-24520.Szymanski, C. M. and B. W. Wren (2005). Protein glycosylation in bacterial mucosal pathogens. Nat Rev Microbiol 3(3): 225-237.Szymanski, C. M., R. Yao, C. P. Ewing, T. J. Trust, et al. (1999). Evidence for a system of general protein glycosylation in Campylobacter jejuni. Mol Microbiol 32(5): 1022-1030.Takade, A., A. Umeda, S. Yokoyama and K. Amako (1988). The substitution-fixation of Staphylococcus aureus. J Electron Microsc (Tokyo) 37(4): 215-217.Takahashi, Y., A. Yajima, J. O. Cisar and K. Konishi (2004). Functional analysis of the Streptococcus gordonii DL1 sialic acid-binding adhesin and its essential role in bacterial binding to platelets. Infect Immun 72(7): 3876-3882.Takamatsu, D., B. A. Bensing, H. Cheng, G. A. Jarvis, et al. (2005a). Binding of the Streptococcus gordonii surface glycoproteins GspB and Hsa to specific carbohydrate structures on platelet membrane glycoprotein Ibalpha. Mol Microbiol 58(2): 380-392.Takamatsu, D., B. A. Bensing and P. M. Sullam (2004a). Four proteins encoded in the gspB-secY2A2 operon of Streptococcus gordonii mediate the intracellular glycosylation of the platelet-binding protein GspB. J Bacteriol 186(21): 7100-7111.Takamatsu, D., B. A. Bensing and P. M. Sullam (2004b). Genes in the accessory sec locus of Streptococcus gordonii

refe

renc

es.

197

have three functionally distinct effects on the expression of the platelet-binding protein GspB. Mol Microbiol 52(1): 189-203.Takamatsu, D., B. A. Bensing and P. M. Sullam (2005b). Two additional components of the accessory sec system mediating export of the Streptococcus gordonii platelet-binding protein GspB. J Bacteriol 187(11): 3878-3883.Tang, G. and K. P. Mintz (2010). Glycosylation of the collagen adhesin EmaA of Aggregatibacter actinomycetemcomitans is dependent upon the lipopolysaccharide biosynthetic pathway. J Bacteriol 192(5): 1395-1404.Tang, G., T. Ruiz, R. Barrantes-Reynolds and K. P. Mintz (2007). Molecular heterogeneity of EmaA, an oligomeric autotransporter adhesin of Aggregatibacter (Actinobacillus) actinomycetemcomitans. Microbiology 153(Pt 8): 2447-2457.Tang, G., T. Ruiz and K. P. Mintz (2012). O-polysaccharide glycosylation is required for stability and function of the collagen adhesin EmaA of Aggregatibacter actinomycetemcomitans. Infect Immun 80(8): 2868-2877.Thakran, S., H. Li, C. L. Lavine, M. A. Miller, et al. (2008). Identification of Francisella tularensis lipoproteins that stimulate the toll-like receptor (TLR) 2/TLR1 heterodimer. J Biol Chem 283(7): 3751-3760.The Human Microbiome Project Consortium (2012). Structure, function and diversity of the healthy human microbiome. Nature 486(7402): 207-214.Thibault, P., S. M. Logan, J. F. Kelly, J. R. Brisson, et al. (2001). Identification of the carbohydrate moieties and glycosylation motifs in Campylobacter jejuni flagellin. J Biol Chem 276(37): 34862-34870.Thomas, R. M., S. M. Twine, K. M. Fulton, L. Tessier, et al. (2011). Glycosylation of DsbA in Francisella tularensis subsp. tularensis. J Bacteriol 193(19): 5498-5509.Thompson, R., A. Creavin, M. O’Connell, B. O’Connor, et al. (2011). Optimization of the enzyme-linked lectin assay for enhanced glycoprotein and glycoconjugate analysis. Anal Biochem 413(2): 114-122.Tillisch, K. (2014). The effects of gut microbiota on CNS function in humans. Gut Microbes 5(3): 404-410.Tomoeda, M., M. Inuzuka and T. Date (1975). Bacterial sex pili. Prog Biophys Mol Biol 30(1): 23-56.Ton-That, H. and O. Schneewind (2003). Assembly of pili on the surface of Corynebacterium diphtheriae. Mol Microbiol 50(4): 1429-1438.Tong, L. (2013). Structure and function of biotin-dependent carboxylases. Cell Mol Life Sci 70(5): 863-891.Torres, A., M. D. Juarez, R. Cervantes and C. Espitia (2001). Molecular analysis of Mycobacterium tuberculosis phosphate specific transport system in Mycobacterium smegmatis. Characterization of recombinant 38 kDa (PstS-1). Microb Pathog 30(5): 289-297.Totten, P. A. and S. Lory (1990). Characterization of the type a flagellin gene from Pseudomonas aeruginosa PAK. J Bacteriol 172(12): 7188-7199.Tripathi, P., A. Beaussart, D. Alsteens, V. Dupres, et al. (2013). Adhesion and nanomechanics of pili from the probiotic Lactobacillus rhamnosus GG. ACS Nano 7(4): 3685-3697.Tripathi, P., A. Beaussart, G. Andre, T. Rolain, et al. (2012a). Towards a nanoscale view of lactic acid bacteria. Micron 43(12): 1323-1330.Tripathi, P., V. Dupres, A. Beaussart, S. Lebeer, et al. (2012b). Deciphering the nanometer-scale organization and assembly of Lactobacillus rhamnosus GG pili using atomic force microscopy. Langmuir 28(4): 2211-2216.Troutman, J. M. and B. Imperiali (2009). Campylobacter jejuni PglH is a single active site processive polymerase that utilizes product inhibition to limit sequential glycosyl transfer reactions. Biochemistry 48(12): 2807-2816.Turroni, F., F. Serafini, M. Mangifesta, S. Arioli, et al. (2014). Expression of sortase-dependent pili of Bifidobacterium bifidum PRL2010 in response to environmental gut conditions. FEMS Microbiol Lett 357(1): 23-33.Twine, S. M., C. J. Paul, E. Vinogradov, D. J. McNally, et al. (2008). Flagellar glycosylation in Clostridium botulinum. FEBS J 275(17): 4428-4444.Twine, S. M., C. W. Reid, A. Aubry, D. R. McMullin, et al. (2009). Motility and flagellar glycosylation in Clostridium difficile. J Bacteriol 191(22): 7050-7062.Typas, A., M. Banzhaf, C. A. Gross and W. Vollmer (2012). From the regulation of peptidoglycan synthesis to bacterial growth and morphology. Nature Reviews Microbiology 10(2): 123-136.Tytgat, H. L. and S. Lebeer (2014). The Sweet Tooth of Bacteria: Common Themes in Bacterial Glycoconjugates. Microbiol Mol Biol Rev 78(3): 372-417.Tytgat, H. L., G. Schoofs, M. Driesen, P. Proost, et al. (2014). Endogenous biotin-binding proteins: an overlooked factor causing false positives in streptavidin-based protein detection. Microb Biotechnol, 10.1111/1751-7915.12150.

198

refe

renc

es

Umeda, A., S. Yokoyama, T. Arizono and K. Amako (1992). Location of peptidoglycan and teichoic acid on the cell wall surface of Staphylococcus aureus as determined by immunoelectron microscopy. J Electron Microsc (Tokyo) 41(1): 46-52.Unligil, U. M. and J. M. Rini (2000). Glycosyltransferase structure and mechanism. Curr Opin Struct Biol 10(5): 510-517.Upreti, R. K., M. Kumar and V. Shankar (2003). Bacterial glycoproteins: functions, biosynthesis and applications. Proteomics 3(4): 363-379.Valvano, M. A. (2008). Undecaprenyl phosphate recycling comes out of age. Mol Microbiol 67(2): 232-235.van Alphen, L. B., M. Wuhrer, N. M. Bleumink-Pluym, P. J. Hensbergen, et al. (2008). A functional Campylobacter jejuni maf4 gene results in novel glycoforms on flagellin and altered autoagglutination behaviour. Microbiology 154(Pt 11): 3385-3397.Van Damme, E. J., D. F. Smith, R. Cummings and W. J. Peumans (2011). Glycan arrays to decipher the specificity of plant lectins. Adv Exp Med Biol 705: 757-767.van Die, I., S. J. van Vliet, A. K. Nyame, R. D. Cummings, et al. (2003). The dendritic cell-specific C-type lectin DC-SIGN is a receptor for Schistosoma mansoni egg antigens and recognizes the glycan antigen Lewis x. Glycobiology 13(6): 471-478.van Gisbergen, K. P., L. C. Paessens, T. B. Geijtenbeek and Y. van Kooyk (2005). Molecular mechanisms that set the stage for DC-T cell engagement. Immunol Lett 97(2): 199-208.van Heijenoort, J. (2001). Formation of the glycan chains in the synthesis of bacterial peptidoglycan. Glycobiology 11(3): 25R-36R.van Sorge, N. M., N. M. Bleumink, S. J. van Vliet, E. Saeland, et al. (2009). N-glycosylated proteins and distinct lipooligosaccharide glycoforms of Campylobacter jejuni target the human C-type lectin receptor MGL. Cell Microbiol 11(12): 1768-1781.VanderVen, B. C., J. D. Harder, D. C. Crick and J. T. Belisle (2005). Export-mediated assembly of mycobacterial glycoproteins parallels eukaryotic pathways. Science 309(5736): 941-943.Vanterpool, E., F. Roy and H. M. Fletcher (2005a). Inactivation of vimF, a putative glycosyltransferase gene downstream of vimE, alters glycosylation and activation of the gingipains in Porphyromonas gingivalis W83. Infect Immun 73(7): 3971-3982.Vanterpool, E., F. Roy, L. Sandberg and H. M. Fletcher (2005b). Altered gingipain maturation in vimA- and vimE-defective isogenic mutants of Porphyromonas gingivalis. Infect Immun 73(3): 1357-1366.Vargas Garcia, C. E., M. Petrova, I. J. Claes, I. De Boeck, et al. (2015). Piliation of Lactobacillus rhamnosus GG promotes adhesion, phagocytosis and cytokine modulation in macrophages. Appl Environ Microbiol, 10.1128/AEM.03949-14.Veith, P. D., N. M. O’Brien-Simpson, Y. Tan, D. C. Djatmiko, et al. (2009). Outer membrane proteome and antigens of Tannerella forsythia. J Proteome Res 8(9): 4279-4292.Venugopal, H., P. J. Edwards, M. Schwalbe, J. K. Claridge, et al. (2011). Structural, dynamic, and chemical characterization of a novel S-glycosylated bacteriocin. Biochemistry 50(14): 2748-2755.Verma, A., S. K. Arora, S. K. Kuravi and R. Ramphal (2005). Roles of specific amino acids in the N terminus of Pseudomonas aeruginosa flagellin and of flagellin glycosylation in the innate immune response. Infect Immun 73(12): 8237-8246.Verma, A., M. Schirm, S. K. Arora, P. Thibault, et al. (2006). Glycosylation of b-Type flagellin of Pseudomonas aeruginosa: structural and genetic basis. J Bacteriol 188(12): 4395-4403.Viburiene, R., A. Vik, M. Koomey and B. Borud (2013). Allelic variation in a simple sequence repeat element of neisserial pglB2 and its consequences for protein expression and protein glycosylation. J Bacteriol 195(15): 3476-3485.Vijayakumar, S., A. Merkx-Jacques, D. B. Ratnayake, I. Gryski, et al. (2006). Cj1121c, a novel UDP-4-keto-6-deoxy-GlcNAc C-4 aminotransferase essential for protein glycosylation and virulence in Campylobacter jejuni. J Biol Chem 281(38): 27733-27743.Vik, A., F. E. Aas, J. H. Anonsen, S. Bilsborough, et al. (2009). Broad spectrum O-linked protein glycosylation in the human pathogen Neisseria gonorrhoeae. Proc Natl Acad Sci U S A 106(11): 4447-4452.Vik, A., M. Aspholm, J. H. Anonsen, B. Borud, et al. (2012). Insights into type IV pilus biogenesis and dynamics from genetic analysis of a C-terminally tagged pilin: a role for O-linked glycosylation. Mol Microbiol 85(6): 1166-1178.

refe

renc

es.

199

Virji, M. (1997). Post-translational modifications of meningococcal pili. Identification of common substituents: glycans and alpha-glycerophosphate--a review. Gene 192(1): 141-147.Virji, M., D. Evans, A. Hadfield, F. Grunert, et al. (1999). Critical determinants of host receptor targeting by Neisseria meningitidis and Neisseria gonorrhoeae: identification of Opa adhesiotopes on the N-domain of CD66 molecules. Mol Microbiol 34(3): 538-551.Virji, M., J. R. Saunders, G. Sims, K. Makepeace, et al. (1993). Pilus-facilitated adherence of Neisseria meningitidis to human epithelial and endothelial cells: modulation of adherence phenotype occurs concurrently with changes in primary amino acid sequence and the glycosylation status of pilin. Mol Microbiol 10(5): 1013-1028.Virji, M., E. Stimson, K. Makepeace, A. Dell, et al. (1996). Posttranslational modifications of meningococcal pili. Identification of a common trisaccharide substitution on variant pilins of strain C311. Ann N Y Acad Sci 797: 53-64.Voisin, S., J. V. Kus, S. Houliston, F. St-Michael, et al. (2007). Glycosylation of Pseudomonas aeruginosa strain Pa5196 type IV pilins with mycobacterium-like alpha-1,5-linked d-Araf oligosaccharides. J Bacteriol 189(1): 151-159.von Mering, C., L. J. Jensen, B. Snel, S. D. Hooper, et al. (2005). STRING: known and predicted protein-protein associations, integrated and transferred across organisms. Nucleic Acids Res 33(Database issue): D433-437.von Ossowski, I., T. E. Pietila, J. Rintahaka, E. Nummenmaa, et al. (2013). Using recombinant Lactococci as an approach to dissect the immunomodulating capacity of surface piliation in probiotic Lactobacillus rhamnosus GG. PLoS One 8(5): e64416.von Ossowski, I., J. Reunanen, R. Satokari, S. Vesterlund, et al. (2010). Mucosal adhesion properties of the probiotic Lactobacillus rhamnosus GG SpaCBA and SpaFED pilin subunits. Appl Environ Microbiol 76(7): 2049-2057.Wacker, M., M. F. Feldman, N. Callewaert, M. Kowarik, et al. (2006). Substrate specificity of bacterial oligosaccharyltransferase suggests a common transfer mechanism for the bacterial and eukaryotic systems. Proc Natl Acad Sci U S A 103(18): 7088-7093.Wacker, M., D. Linton, P. G. Hitchen, M. Nita-Lazar, et al. (2002). N-linked glycosylation in Campylobacter jejuni and its functional transfer into E. coli. Science 298(5599): 1790-1793.Wang, D., M. M. Waye, M. Taricani, K. Buckingham, et al. (1993). Biotin-containing protein as a cause of false positive clones in gene probing with streptavidin/biotin. Biotechniques 14(2): 209-212.Wang, L., D. Liu and P. R. Reeves (1996). C-terminal half of Salmonella enterica WbaP (RfbP) is the galactosyl-1-phosphate transferase domain catalyzing the first step of O-antigen synthesis. J Bacteriol 178(9): 2598-2604.Wang, L. and M. J. Wise (2011). Glycogen with short average chain length enhances bacterial durability. Naturwissenschaften 98(9): 719-729.Warren, M. J., L. F. Roddam, P. M. Power, T. D. Terry, et al. (2004). Analysis of the role of pglI in pilin glycosylation of Neisseria meningitidis. FEMS Immunol Med Microbiol 41(1): 43-50.Weerapana, E. and B. Imperiali (2006). Asparagine-linked protein glycosylation: from eukaryotic to prokaryotic systems. Glycobiology 16(6): 91R-101R.Weerkamp, A. H., H. C. van der Mei and R. S. Liem (1986). Structural properties of fibrillar proteins isolated from the cell surface and cytoplasm of Streptococcus salivarius (K+) cells and nonadhesive mutants. J Bacteriol 165(3): 756-762.Weigel, P. H., V. C. Hascall and M. Tammi (1997). Hyaluronan synthases. J Biol Chem 272(22): 13997-14000.Wells, J. M. and A. Mercenier (2008). Mucosal delivery of therapeutic and prophylactic molecules using lactic acid bacteria. Nat Rev Microbiol 6(5): 349-362.Welzel, P. (2005). Syntheses around the transglycosylation step in peptidoglycan biosynthesis. Chem Rev 105(12): 4610-4660.Werner, R. G., K. Kopp and M. Schlueter (2007). Glycosylation of therapeutic proteins in different production systems. Acta Paediatr Suppl 96(455): 17-22.Whitfield, C. (1995). Biosynthesis of lipopolysaccharide O antigens. Trends Microbiol 3(5): 178-185.Whitfield, C. (2006). Biosynthesis and assembly of capsular polysaccharides in Escherichia coli. Annu Rev Biochem 75: 39-68.Whitfield, C. and A. Paiment (2003). Biosynthesis and assembly of Group 1 capsular polysaccharides in Escherichia coli and related extracellular polysaccharides in other bacteria. Carbohydr Res 338(23): 2491-2502.Whitfield, C. and I. S. Roberts (1999). Structure, assembly and regulation of expression of capsules in Escherichia coli. Mol Microbiol 31(5): 1307-1319.

200

refe

renc

es

Whitney, J. C. and P. L. Howell (2013). Synthase-dependent exopolysaccharide secretion in Gram-negative bacteria. Trends Microbiol 21(2): 63-72.Wilson, W. A., P. J. Roach, M. Montero, E. Baroja-Fernandez, et al. (2010). Regulation of glycogen metabolism in yeast and bacteria. FEMS Microbiol Rev 34(6): 952-985.Wolucka, B. A., M. R. McNeil, E. de Hoffmann, T. Chojnacki, et al. (1994). Recognition of the lipid intermediate for arabinogalactan/arabinomannan biosynthesis and its relation to the mode of action of ethambutol on mycobacteria. J Biol Chem 269(37): 23328-23335.Wood, H. G. and R. E. Barden (1977). Biotin enzymes. Annu Rev Biochem 46: 385-413.Wu, H., S. Bu, P. Newell, Q. Chen, et al. (2007a). Two gene determinants are differentially involved in the biogenesis of Fap1 precursors in Streptococcus parasanguis. J Bacteriol 189(4): 1390-1398.Wu, H., K. P. Mintz, M. Ladha and P. M. Fives-Taylor (1998). Isolation and characterization of Fap1, a fimbriae-associated adhesin of Streptococcus parasanguis FW213. Mol Microbiol 28(3): 487-500.Wu, H., M. Zeng and P. Fives-Taylor (2007b). The glycan moieties and the N-terminal polypeptide backbone of a fimbria-associated adhesin, Fap1, play distinct roles in the biofilm development of Streptococcus parasanguinis. Infect Immun 75(5): 2181-2188.Wu, J., J. Zhu, H. Yin, R. J. Buckanovich, et al. (2014). Analysis of glycan variation on glycoproteins from serum by the reverse lectin-based ELISA assay. J Proteome Res 13(4): 2197-2204.Wu, R. and H. Wu (2011). A molecular chaperone mediates a two-protein enzyme complex and glycosylation of serine-rich streptococcal adhesins. J Biol Chem 286(40): 34923-34931.Wu, R., M. Zhou and H. Wu (2010). Purification and characterization of an active N-acetylglucosaminyltransferase enzyme complex from Streptococci. Appl Environ Microbiol 76(24): 7966-7971.Wu, Z. L., C. M. Ethen, B. Prather, M. Machacek, et al. (2011). Universal phosphatase-coupled glycosyltransferase assay. Glycobiology 21(6): 727-733.Wugeditsch, T., A. Paiment, J. Hocking, J. Drummelsmith, et al. (2001). Phosphorylation of Wzc, a tyrosine autokinase, is essential for assembly of group 1 capsular polysaccharides in Escherichia coli. J Biol Chem 276(4): 2361-2371.Wuhrer, M., A. R. de Boer and A. M. Deelder (2009). Structural glycomics using hydrophilic interaction chromatography (HILIC) with mass spectrometry. Mass Spectrom Rev 28(2): 192-206.Wyss, C. (1998). Flagellins, but not endoflagellar sheath proteins, of Treponema pallidum and of pathogen-related oral spirochetes are glycosylated. Infect Immun 66(12): 5751-5754.Wyszynska, A., J. Zycka, R. Godlewska and E. K. Jagusztyn-Krynicka (2008). The Campylobacter jejuni/coli cjaA (cj0982c) gene encodes an N-glycosylated lipoprotein localized in the inner membrane. Curr Microbiol 57(3): 181-188.Xu, J., M. A. Mahowald, R. E. Ley, C. A. Lozupone, et al. (2007). Evolution of symbiotic bacteria in the distal human intestine. PLoS Biol 5(7): e156.Yan, F., H. Cao, T. L. Cover, M. K. Washington, et al. (2011). Colon-specific delivery of a probiotic-derived soluble protein ameliorates intestinal inflammation in mice through an EGFR-dependent mechanism. J Clin Invest 121(6): 2242-2253.Yan, F., H. Cao, T. L. Cover, R. Whitehead, et al. (2007). Soluble proteins produced by probiotic bacteria regulate intestinal epithelial cell survival and growth. Gastroenterology 132(2): 562-575. Yasuda, E., H. Tateno, J. Hirabarashi, T. Iino, et al. (2011). Lectin microarray reveals binding profiles of Lactobacillus casei strains in a comprehensive analysis of bacterial cell wall polysaccharides. Appl Environ Microbiol 77(13): 4539-4546.Yen, Y. T., R. Seepersaud, B. A. Bensing and P. M. Sullam (2011). Asp2 and Asp3 interact directly with GspB, the export substrate of the Streptococcus gordonii accessory Sec System. J Bacteriol 193(13): 3165-3174.Yoshida, Y., Y. Nakano, Y. Yamashita and T. Koga (1998). Identification of a genetic locus essential for serotype b-specific antigen synthesis in Actinobacillus actinomycetemcomitans. Infect Immun 66(1): 107-114.Yother, J. (2011). Capsules of Streptococcus pneumoniae and other bacteria: paradigms for polysaccharide biosynthesis and regulation. Annu Rev Microbiol 65: 563-581.Young, N. M., J. R. Brisson, J. Kelly, D. C. Watson, et al. (2002). Structure of the N-linked glycan present on multiple glycoproteins in the Gram-negative bacterium, Campylobacter jejuni. J Biol Chem 277(45): 42530-42539.Yuan, Y., D. Barrett, Y. Zhang, D. Kahne, et al. (2007). Crystal structure of a peptidoglycan glycosyltransferase

refe

renc

es.

201

suggests a model for processive glycan chain synthesis. Proc Natl Acad Sci U S A 104(13): 5348-5353.Zaia, J. (2011). At last, functional glycomics. Nat Methods 8(1): 55-57.Zapun, A., M. Noirclerc-Savoye, N. Helassa and T. Vernet (2012). Peptidoglycan Assembly Machines: The Biochemical Evidence. Microbial Drug Resistance 18(3): 256-260.Zechel, D. L. and S. G. Withers (2000). Glycosidase mechanisms: anatomy of a finely tuned catalyst. Acc Chem Res 33(1): 11-18.Zeituni, A. E., W. McCaig, E. Scisci, D. G. Thanassi, et al. (2010). The native 67-kilodalton minor fimbria of Porphyromonas gingivalis is a novel glycoprotein with DC-SIGN-targeting motifs. J Bacteriol 192(16): 4103-4110.Zhou, M., Z. Peng, P. Fives-Taylor and H. Wu (2008). A conserved C-terminal 13-amino-acid motif of Gap1 is required for Gap1 function and necessary for the biogenesis of a serine-rich glycoprotein of Streptococcus parasanguinis. Infect Immun 76(12): 5624-5631.Zhou, M. and H. Wu (2009). Glycosylation and biogenesis of a family of serine-rich bacterial adhesins. Microbiology 155(Pt 2): 317-327.Zhou, M., H. Zhang, F. Zhu and H. Wu (2011). Canonical SecA associates with an accessory secretory protein complex involved in biogenesis of a streptococcal serine-rich repeat glycoprotein. J Bacteriol 193(23): 6560-6566.Zhou, M., F. Zhu, S. Dong, D. G. Pritchard, et al. (2010). A novel glucosyltransferase is required for glycosylation of a serine-rich adhesin and biofilm formation by Streptococcus parasanguinis. J Biol Chem 285(16): 12140-12148.Zhu, F., H. Erlandsen, L. Ding, J. Li, et al. (2011). Structural and functional analysis of a new subfamily of glycosyltransferases required for glycosylation of serine-rich streptococcal adhesins. J Biol Chem 286(30): 27048-27057.Zogaj, X., S. Chakraborty, J. Liu, D. G. Thanassi, et al. (2008). Characterization of the Francisella tularensis subsp. novicida type IV pilus. Microbiology 154(Pt 7): 2139-2150.

202

adde

ndum

aDDenDum: List of mutants

List of mutants constructed together with technician Tine Verhoeven in the framework of this PhD. All described mutants are double homologous recombinant knockout mutants, in which an antibiotic resistance cassette replaced the gene of interest.

gt mutants

Eleven mutants were produced in the framework of the results of our in silico screening for genes encoding GTs (chapter 3). Numerous attempts were made to produce mutants at least in all GTs predicted to target proteins, unfortunately, some of the targeted GTs seemed to be essential. This was the case for the predicted GTs encoded by LGG_00794, LGG_01487 and LGG_001283. Some mutants were already produced whilst the GT in silico screening was still in progress and later proved to be false positive hits of the analysis, which were discarded in the final optimized format of the workflow (LGG_01399 and LGG_02693).

Strain Genotype Target locus Fenotype

CMPG10800 ΔgtrB::TcR LGG_01069 None observed

CMPG10802 ΔyohH-yohJ::TcR LGG_01586-87 None observed

CMPG10803 Δhly::TcR LGG_01399 False positive GT

CMPG10804 ΔxylB::TcR LGG_02693 False positive GT

CMPG10806 ΔLGG_00825-26::TcR LGG_00825-26 None observed

CMPG10808 ΔarbX-arbY::TcR LGG_00998-999 None observed

CMPG10809 Δpbp2A::TcR LGG_01783 None observed

CMPG10811 ΔwelI::TcR LGG_02047 Less EPS compared to WT (chapter 3)

CMPG10815 ΔLGG_01147::TcR LGG_01147 None observed

CMPG10817 ΔyohH-yohJ::TcR LGG_02284-85 None observed

msp1 mutants

To investigate whether the identified glycosylation site in Msp1, namely TVETPSSA, was the only modified amino acid stretch in this protein, two new mutants were constructed. The existing msp1 single homologous recombinant mutant (CMPG10200) was complemented with plasmids carrying the msp1 gene in which the serines of the glycosylation motif were deleted via a point mutation (CMPG10805) and another one in which a large stretch rich in serines and alanines was deleted around the TVETPSSA region of Msp1 (CMPG10807). Both mutants still produced glycosylated Msp1 protein when stained with PAS glycostain and probing with lectins (results not shown).

secx mutants

To investigate whether L. rhamnosus GG harbors a special system to secrete glycosylated proteins, we constructed several mutants that can help us to elucidate the existence and

adde

ndum

.

203

functionality of such system.

Strain Genotype Target locus Fenotype

CMPG10810 Δabc::TcR LGG_02580-81 Analysis ongoing

CMPG10816 ΔsecX::TcR LGG_01063 Analysis ongoing

CMPG10818 ΔsecX::EryRΩabc::TcR LGG_01063 and LGG_002580-81 Analysis ongoing

CMPG10819 ΔsecX::TcRΩmsp1::EryR LGG_01063 and LGG_00324 Analysis ongoing

204

list o

f pub

licat

ions

List of pubLications

articLes in internationaLLy peer-revieweD journaLs

Tytgat H.L.P., Schoofs G., Driesen M., Proost P., Van Damme E.J.M., Vanderleyden J., Lebeer S. (2015) Endogenous biotin-binding proteins: an overlooked factor causing false positives in streptavidin-based protein detection. Microbial Biotechnology, 8, 164-168.

Tytgat H.L.P., Lebeer S. (2014). The sweet tooth of bacteria: common themes in bacterial glycoconjugates. MMBR, 78(3), 372-417.

Tytgat H.L.P.*, Sanchez-Rodriguez A.*, Winderickx J., Vanderleyden J., Lebeer S., Marchal K. (2014) A network-based approach to identify substrate relations of bacterial glycosyltransferases. BMC Genomics, 15 (349). (* equal contribution of first authors)

Burgain J., Gaiani C., Francius G., Revol-Junelles A.M., Cailliez-Grimal C., Lebeer S., Tytgat H.L., Vanderleyden J., Scher J. (2013). In vitro interactions between probiotic bacteria and milk proteins probed by atomic force microscopy. Colloids Surf. B. Biointerfaces, 104, 153-162.

Lebeer S., Claes I.J.J., Tytgat H.L.P., Verhoeven T.L.A., Marien E., von Ossowski I., Reunanen J., Palva A., de Vos W.M., De Keersmaecker S.C.J., Vanderleyden J. (2012). Functional analysis of the pili of Lactobacillus rhamnosus GG in relation to adhesion and immunomodulatory interactions with intestinal epithelial cells. Applied and Environmental Microbiology, 78 (1), 185-193.

Lebeer S., Claes I.J.J., Balog C.I.A., Schoofs G., Verhoeven T.L.A., Nys K., von Ossowski I., de Vos W.M., Tytgat H.L.P., Agostinis P., Palva A., Deelder A.M., De Keersmaecker S.C.J., Wuhrer M., Vanderleyden J. (2011). The major secreted protein Msp1/p75 is O-glycosylated in Lactobacillus rhamnosus GG. Microbial Cell Factories, 11 (15).

articLes in journaLs witHout peer review

Tytgat H.L., Sanchez-Rodriguez A., Schoofs G., Verhoeven T.L., De Keersmaecker S.C., Marchal K., Vanderleyden J., Lebeer S. (2012). A combined approach to study the protein glycosylation potential of Lactobacillus rhamnosus GG (LGG). Communic. Agric. Appl. Biol. Sci., 77 (1), 15-9.

articLes for a broaDer auDience

Tytgat H. (2014) ‘Duurzaam denken en doen met onze aarde - De rol van wetenschap en technologie in een duurzame ontwikkeling van de aarde’ Ethische perspectieven, 24 (3), 271-282, doi: 10.2143/EPN.23.2.0000000. Article on a debate on sustainability for a magazine called ‘Ethical perspectives’.

publ

icat

ions

.

205

Tytgat H. (2014). ‘Debat – Duurzaam denken en doen met onze aarde.’ Bioingenieus. Article on a debate on sustainability for the magazine of the faculty of Bioscience Engineering of KU Leuven.

Tytgat H., Marchal K., Vanderleyden J. (2013). ‘De maakbare natuur.’ Karakter. Article on synthetic biology and potential ethical concerns for the magazine of the Academische Stichting Leuven.

abstracts on internationaL meetings

Lebeer S., Tytgat H.L.P., Claes I.J.J., Schoofs G., Verhoeven T.L.A., De Keersmaecker S.C.J., Vanderleyden J. Glycoconjugates of Lactobacillus rhamnosus GG and their role in immunomodulation. 12th Gut Day Symposium. September 23, 2010. Ghent, Belgium. Poster presentation. Abstract book B4.

Lebeer S., Tytgat H.L.P., Claes I.J.J., Schoofs G., Verhoeven T.L.A., De Keersmaecker S.C.J., Vanderleyden J. Glycoconjugates of Lactobacillus rhamnosus GG and their role in immunomodulation. Joint Glycobiology Meeting. November 7-9, 2010. Ghent, Belgium. Poster presentation. Abstract book 29.

Tytgat H.L.P., Lebeer S., Claes I.J.J., Schoofs G., Verhoeven T.L.A., De Keersmaecker S.C.J., Vanderleyden J. Glycoconjugates of Lactobacillus rhamnosus GG and their role in immunomodulation. 16th PhD Symposium on Applied Biological Sciences. December 20, 2010. Ghent, Belgium. Poster presentation.

Tytgat H., Lebeer S., Claes I.J.J., Sánchez-Rodriguez A., Schoofs G., Verhoeven T.L.A., De Keersmaecker S.C.J., Marchal K., Vanderleyden J. Glycobiology of Lactobacillus rhamnosus GG: Focus on the glycosyltransferases. 10th Symposium on Lactic Acid Bacteria (LAB10). August 28 - September 1, 2011. Egmond Aan Zee, The Netherlands. Oral and poster presentation. Abstract book SL15.

Tytgat H.L.P., Sanchez-Rodriguez A., Lebeer S., Schoofs G., Verhoeven T.L.A., De Keersmaecker S.C.J., Vanderleyden J., Marchal K. A network-based approach to reveal the glycosylation potential of Lactobacillus rhamnosus GG. 13th Gut Day Symposium. October 20, 2011. Wageningen, The Netherlands. Oral presentation. Best presentation award

Tytgat H.L.P., Sánchez-Rodríguez A., Lebeer S., Schoofs G., Verhoeven T.L.A., De Keersmaecker S.C.J., Marchal K., Vanderleyden J. Glycobiology of the probiotic Lactobacillus rhamnosus GG (LGG): A network-based approach. 2011 Annual Conference of the Society for Glycobiology. November 9-12, 2011. Seattle, USA. Oral and poster presentation. Abstract book 145.

Tytgat H.L.P., Sanchez-Rodriguez A., Lebeer S;, Schoofs G., Verhoeven T., De Keersmaecker

206

list o

f pub

licat

ions

S.C.J., Vanderleyden J., Marchal K. A new network-based approach to reveal the protein glycosyltransferases of Lactobacillus rhamnosus GG. 22nd Joint Glycobiology Meeting. November 27-29, 2011. Lille, France. Oral presentation. Abstract book O5.

Tytgat H.L.P., Sanchez-Rodriguez A., Lebeer S., Schoofs G., Verhoeven T.L.A., De Keersmaecker S.C.J., Marchal K., Vanderleyden J. A combined approach to study the protein glycosylation potential of Lactobacillus rhamnosus GG (LGG). 17th PhD Symposium on Applied Biological Sciences. February 10, 2012. Leuven, Belgium. Oral presentation.

Tytgat H.L.P., Sanchez-Rodriguez A., Schoofs G., Verhoeven T.L.A., De Keersmaecker S.C.J., Marchal K., Vanderleyden J., Lebeer S. New insights in bacterial protein glycosylation: unraveling the glycosylation potential of Lactobacillus rhamnosus GG. Young Microbiologists Symposium on Microbe Signaling, Organisation and Pathogenesis. June 12-22, 2012. Cork, Ireland. Oral presentation.

Tytgat H.L.P., Schoofs G., Verhoeven T.L.A., Leroy B., Wattiez R., Lebeer S., Vanderleyden J. Exploring the glycoproteome of Lactobacillus rhamnosus GG. 14th Gut Day Symposium. November 9, 2012. Leuven, Belgium. Poster presentation.

Tytgat H.L.P., Schoofs G., Verhoeven T.L.A., Leroy B. Van Damme E.J.M., Wattiez R., Lebeer S., Vanderleyden J. Exploring the glycosylation potential of the probiotic Lactobacillus rhamnosus GG: focus on the glycoproteome. 18th PhD Symposium on Applied Biological Sciences. February 8, 2013. Ghent, Belgium. Poster presentation.

Tytgat H.L.P., Schoofs G., Verhoeven T.L.A., Leroy B. Van Damme E.J.M., Wattiez R., Lebeer S., Vanderleyden J. Exploration of the glycoproteome of Lactobacillus rhamnosus GG. Gut Microbiota for Health, 2nd World Summit, Danone. February 24-26, 2013. Madrid, Spain. Poster presentation.

Tytgat H.L.P., Verhoeven T.L.A., Schoofs G., Vanderleyden J., Lebeer S. The role of glycoconjugates in beneficial host-microbe interactions: focus on glycoproteins in Lactobacillus rhamnosus GG. 7th International Yakult Symposium. April 22-23, 2013. London, UK. Poster presentation.

Tytgat H.L.P., Schoofs G., Leroy B., Wattiez R., Vanderleyden J., Lebeer S. Is Msp1 alone? – Mapping the glycoproteome of Lactobacillus rhamnosus GG. 15th Gut Day Symposium. November 7, 2013. Groningen, The Netherlands. Poster presentation.

Tytgat H.L.P., Sánchez-Rodríguez A., Schoofs G., Verhoeven T.L.A., Leroy B., Wattiez R., Marchal K., Vanderleyden J., Lebeer S. Sweet old friends: exploring the protein glycosylation potential of the model probiotic Lactobacillus rhamnosus GG. PhD Interaction Day. December 19, 2013. Leuven, Belgium. Oral presentation. Recipient of the PhD award

publ

icat

ions

.

207

Tytgat H.L.P., Segers A., Schoofs G., Verhoeven T.L.A., Leroy B., Wattiez R., Vanderleyden J., Lebeer S. Sweet old friends – Novel insights in the glycosylation potential of the modle probiotic Lactobacillus rhamnosus GG. FASEB conference on Microbial Glycobiology. June 8-13, 2014. Itasca, USA. Poster presentation. Abstract book P46.

Tytgat H.L.P., Schoofs G., Verhoeven T.L.A., Leroy B., Vandamme E.J.M., Wattiez R., Vanderleyden J., Lebeer S. Mapping the glycoproteome of our sweet old friend Lactobacillus rhamnosus GG. 25th Joint Glycobiology Meeting. September 14-16, 2014. Ghent, Belgium. Poster presentation.