University of Groningen

The lytic transglycosylase family of Escherichia coliDijkstra, Arnoud Jan

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:1997

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Dijkstra, A. J. (1997). The lytic transglycosylase family of Escherichia coli: in vitro activity versus in vivofunction. [S.n.].

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 19-12-2020

85

Chapter 5A New Member of the Transglycosylase

Family of Escherichia coli Displays aGram-Positive Hydrolase Motif

Arnoud J. Dijkstra, Kenneth E. Rudd and Wolfgang Keck

86 Chapter 5

A New Member of the Transglycosylase Family of Escherichia coliDisplays a Gram-Positive Hydrolase Motif

Summary

The small Escherichia coli open reading frame YafG was one of the first lytictransglycosylase homologs that was identified on the basis of the presence ofconserved sequence fingerprints. These fingerprints were derived from the structureof the 70 kDa soluble lytic transglycosylase (Slt70). Based on their position and theoverall length of YafG, this protein seemed to be a homolog of the lysozyme-likecatalytic domain of Slt70. The results presented here show that YafG is, in fact, partof a much larger protein and that the sequence is extended on both the N-terminaland C-terminal side. The resulting open reading frame displays a consensuslipoprotein signal sequence and therefore was called membrane bound lytictransglycosylase D (MltD). The full length gene was cloned and induction of theexpression of MltD immediately provoked massive lysis, which could to a largeextend be prevented by osmotic protection of the cells. The peptidoglycan hydrolysisactivity of the overproduced protein could be demonstrated and it was shown thatthis activity is inhibited by the Slt70 inhibitor bulgecin. Interestingly, the C-terminalpart of the sequence contains two copies of a sequence motif that has been describedto be present in several gram-positive peptidoglycan interacting and hydrolyzingenzymes. The presence of this motif in MltD and in members of a family of gram-negative lysostaphin homologs, indicates that common principles may be involvedin the activity, cellular localization and function of peptidoglycan hydrolases ofdifferent specificity and origin.

Introduction

The taxonomic differentiation between gram-positive and gram-negative bacteria ishistorically based on the difference in cell-wall structure. The staining procedure asdescribed by Gram in 1884 relies on retention of a dye-complex in the cell-wall ofgram-positive bacteria, whereas it can be washed out from the cell-wall of gram-negative bacteria. The gram classification, albeit based on only one of severalfeatures that differ between gram-positive and gram-negative bacteria, proved to beevolutionary valid and underscores that the most prominent distinction betweenboth classes is in cell-wall morphology.In gram-negative bacteria the periplasmic space between the outer and innermembranes harbors a thin sheet of peptidoglycan. This peptidoglycan is aheteropolymeric macromolecule that encloses the cell and thus provides it with thenecessary mechanical stability. The basic architecture of the peptidoglycan isconserved between different bacteria, although the exact chemical compositionshows some variation (26). The polymer consists of relatively short glycan strands,composed of disaccharide subunits, that are cross-linked by peptides. As gram-positive bacteria lack an outer membrane, the peptidoglycan is located directly onthe outside of the cell, in some cases covered by an additional layer of S-proteins (3).The peptidoglycan of gram-positive bacteria can be over 10 times as thick as the

Peptidoglycan binding motif in MltD 87

peptidoglycan of gram-negatives, which is about 75-80% single layered and about20-25 % triple layered (25). In several gram-positives the multi-layered nature of thepeptidoglycan, in some cases combined with a higher flexibility of the peptide side-chain, allows a much higher degree of crosslinking (up to 90%) than in gram-negatives (25-40% in E. coli). Furthermore, the presence of teichoic acids renders thepeptidoglycan layer of gram-positives even more impermeable.The differences between the gram-positive and gram-negative peptidoglycanstructure are partly reflected in the metabolism of this polymer. Although penicillinbinding proteins are involved in this metabolism in both classes of bacteria and themajor peptidoglycan synthesizing activities are conserved, additional non-penicillinbinding proteins seem to be involved in the synthesis of the glycan strands in severalgram-positives (32). Whether these enzymes, that were proposed to be the majorglycan-synthesizing activities, prove to be homologs of the recently identifiedmonofunctional glycosyltransferase of E. coli remains to be seen (6).The peptidoglycan metabolism does not only involve synthetic activities but alsohydrolytic activities that are thought to be needed for growth and division of thecell. Divergence between gram-positive and gram-negative bacteria is also apparentin this hydrolytic part of the metabolism, and especially the autolytic enzymes thatare able to degrade the polymer are poorly conserved between both classes.Homologs of the largest family of glycan strand-degrading enzymes in E. coli, thelytic transglycosylases, have not been found in gram-positives so far and, vice versa,most gram-positive hydrolases do not have gram-negative homologs. A largenumber of these autolytic peptidoglycan hydrolases of different specificity has beenidentified in gram-positive bacteria, often by making use of the ability of theseenzymes to degrade gram-positive cell-walls in polyacrylamide gels (zymograms) orin agar plates. The major classes of activities seem to be the glucosaminidases,muramidases and amidases (39). The latter is the sole class for which members arepresent in both gram-positive and gram-negative bacteria. Of the three gram-positive amidase homologs in E. coli, the biochemical activity was characterized foronly one, AmiA (33, 48). This amidase turned out not to be a genuine autolysin, as itwas described only to accept lower molecular weight muropeptides as a substrate.On the other hand, induction of expression of the amiB gene leads to lysis of thebacterial cells. The peptidoglycan hydrolase activity of this proposed autolyticenzyme remains to be demonstrated though (49).Gram-positive bacteria do not have the possibility to topologically restrict proteinsto the outer membrane or the periplasm. Although it was recently proposed that theimpermeability of the peptidoglycan may, in fact, cause the formation of acompartment comparable to the periplasmic space (30), the display of proteins onthe surface or retention of proteins in the peptidoglycan layer calls for othermechanisms. These mechanisms seem to involve the covalent or non-covalentbinding to cell-wall components. In Staphylococcus aureus, a system was describedthat, dependent on the presence of a C-terminal recognition sequence, covalentlyattaches proteins to the peptidoglycan (31). Other proteins are thought to bind to thecell-wall through specific binding domains that are often characterized by sequencerepeats. The primary sequence of many peptidoglycan hydrolases show these repeatregions (1, 4, 5, 12, 18). Several types of repeats of different sequence and length havebeen identified and the number of repetitions varies between proteins. The repeats-containing domain of the major autolysin of Streptococcus pneumoniae, LytA, has been

88 Chapter 5

shown to bind specifically to the choline units of the wall teichoic acid (15, 38). Thebinding site of other repeat regions is more speculative.In this study we identified a new gram-negative peptidoglycan hydrolase that showsthe presence of a repeat sequence, that was first described to be present in a family ofgram-positive hydrolases and was proposed to mediate binding to thepeptidoglycan. This new member of the lytic transglycosylase family also contains aconsensus lipoprotein signal sequence and therefore is probably attached to themembrane. The resulting complex topological arrangement of this enzyme mayserve regulatory purposes. Single copies of the proposed peptidoglycan bindingfingerprint could be identified in members of a family of putative peptidoglycanhydrolases that are present in several gram-negative bacteria.

Materials and methods

General DNA manipulations

Established protocols were followed during DNA manipulations and cloningprocedures (37). Purification of PCR products and preparation of plasmid DNA wasperformed using Qiagen (Hilden, Germany) purification systems. Restrictionenzymes and ligase were obtained from New England Biolabs (Beverly,Ma.).Escherichia coli strain XL1_Blue (recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1lac [F’ proAB LacIqZ∆M15 Tn10 (Tetr)]c) from Stratagene (La Jolla, CA) served asgeneral cloning host. Polymerase chain reactions (PCR) were performed withExpand high-fidelity DNA-polymerase and reagents from Boehringer (Mannheim,Germany) in a Perkin Elmer (Norwalk, CT) thermocycler, with the following cyclingprotocol: melting at 94°C for 1 minute, annealing at 56°C for 2 minutes, and extensionat 72°C for 1 minute for a total of 30 cycles. To improve the quality of the PCRreactions, the hot-start system (Perkin Elmer) was used.

Cloning of the gene coding for MltD

Three different versions of the mltD gene were generated by PCR. A fragment of thegene corresponding to yafG was amplified using primers 5’-C GTCTC CCATGTACT-GGATAGC-3’ and 5’-G AAGCTT CAATGCCAGCATTTTAGG-3’, introducing aBsmAI site at the 3’-end and HindIII site at the 5’-end of the fragment. The full lengthmltD gene and a truncated version lacking the fragment coding for the putative N-terminal signal sequence were amplified using forward primers 5’-ACTATTGACA-CAC CCATGG AGGCAAAAGCG-3’ and 5’-GCTCGTGGGTT CCATGG GTACCGGC-AACGTTC-3', respectively, and reverse primer 5'-ATATCATCATGGT GGATCC AC-ATAAAGCGGC-3'. The forward primers used introduced an NcoI site at the startingmethionine codon or at Gln17 in the case of the version without signal sequence. Thereverse primers introduced a BamHI site downstream of the mltD stop codon.Chromosomal DNA from E. coli strain 122-1 (gal attλ bio deob-serb∆) (35) was used asthe template. The NcoI-SalI fragments of the PCR products were directly cloned intothe expression vector pET21d+ (Novagen, Madison, WI, USA), resulting in theexpression plasmids pETYafG, pETMltD (full length) and pETMltDns (withoutsignal sequence).

Peptidoglycan binding motif in MltD 89

Overproduction

The T7 promoter that is present in the pET21d+ vector was used to drive theexpression of the mltD constructs (42). For this purpose the expression vectors weretransformed into the host BL21(DE3) (F- dcm OmpT hsdS(rB- mB-) gal λ(DE3),obtained from Stratagene, carrying the specific T7 RNA polymerase under control ofthe Lac promoter. Expression of the different versions of MltD was induced by theaddition of 1 mM IPTG to cells that had been growing in Luria Bertani medium toan OD600 of about 0.6. Protection against the lytic effect of expression of the fulllength mltD was achieved by osmotically stabilizing the cells by the addition of 12%sucrose and 10 mM MgSO4 to the growth medium. Samples of induced cultureswere evaluated for protein expression by analysis on polyacrylamide gels. In orderto rule out the possibility that mutations had been introduced during PCR thatmight influence the expression or activity, three individual clones were evaluated.The solubility of the overproduced protein was determined by centrifugation at16000 xg of cell extracts that were prepared by sonification for 5 x 10 seconds at 140W, using a Heatsystems (Plainview, N.Y.) sonicator equipped with a micro tip.Peptidoglycan hydrolase activity of overproduced MltD was assessed usingzymogram analysis.

Zymography

Samples of whole cells or cellular extracts were analyzed for peptidoglycanhydrolase activity following a described procedure (2, 7). The samples wereseparated on 12% SDS-PAA gels, supplemented with 0.025 % purified E.colipeptidoglycan. In-gel renaturation was achieved by removal of the SDS by extractionwith Triton X-100. For this purpose the zymograms were incubated overnight at 37°Cin 20 mM Tris-maleate buffer pH 6.8, containing 1 mM MgCl2 and 0.1% Triton X-100. After renaturation the gels were washed for 15 minutes and stained with 0.1%methylene blue in 0.01% KOH for 30 minutes. Densitometric analysis was performedon a model GS-700 imaging densitometer (Bio-Rad Laboratories, Hercules, CA).Purified Slt70, PPB4, Slt35(MltB) and mutanolysin (Sigma Chemical Co., St. Louis,MS) were used as marker proteins. The sensitivity of MltD towards bulgecin wasdetermined by measuring the ability of the overproduced proteins to digestpeptidoglycan in an zymogram gel in the presence of the inhibitor. An extract ofcells that overproduced MltD in the cytoplasm, i.e. without the lipoprotein signalsequence, was supplemented with purified Slt70 and Slt35(MltB). Dilutions of thismixture were applied to a zymogram in duplo, and after electrophoresis two halvesof the zymogram were processed in the presence and the absence of 100 µg/mlbulgecin (Takeda Chemical Industries, Osaka, Japan), following the above describedprotocol.

Peptidoglycan hydrolysis assay

Two different methods were used to measure the rate of peptidoglycan hydrolysis ina liquid assay. The first procedure measured the solubilization of radiolabeledpeptidoglycan, that was prepared as described (13), by selective precipitation ofundigested material by the addition of N-cetyl-N,N,N-trimethylammoniumbromide(CTAB) (Merck, Darmstadt, Germany), followed by centrifugation and scintillationcounting of the supernatant (10). The second method was established to allow theprocessing of multiple samples in parallel and involves a filtration of the

90 Chapter 5

peptidoglycan hydrolysis reaction mixture. The reactions were performed in a 96-well plate and the reactions were started by adding 50 µl of 3H-labeledpeptidoglycan (15000 cpm) to 200 µl of a reaction master mix containing 10 mMphosphate pH 6.8 and the appropriate amount of enzyme or cellular extract. 50 µlsamples were removed from the reaction mixture at different time points and thereaction was stopped immediately by the addition of 5 µl 1 N HCl. The sampleswere transferred to a MultiScreen 96-well filter plate, type MADP NOB (Millipore,Bedford MA). After filtration by the application of a vacuum, the filters were washedwith 50 µl water. 40 µl of the combined filtrate was mixed with 160 µl of MicroScint-20 scintillation cocktail in LumaPlate microplates (Packard, Meriden, CT). Theamount of solubilized peptidoglcyan was measured in a TopCount microtiter platecounter (Packard).

Labeling of lipoproteins

For radiolabeling of lipoproteins, a 20 ml culture containing 12% sucrose was grownto an OD600 of 0.6 and was supplemented with 100 µCi of 3H-palmitate (Amersham,Buckinghamshire, UK), after which the cells were induced for expression of MltD bythe addition of 1 mM IPTG. A culture expressing the truncated form of MltD,without signal sequence, served as a negative control. After 1 hour the cells wereharvested by centrifugation and resuspended in 0.5 ml Tris-maleate pH 6.8. Thecellular proteins were solubilized by the addition of 55 µl 10% SDS. Aftercentrifugation at 20000 x g for 10 minutes, 5 ml ice-cold acetone was added to thesupernatant and the proteins were allowed to precipitate overnight at 4°C. Theprecipitated proteins were pelleted by centrifugation at 20000 x g for 30 minutes andthe pellet was resuspended in 50 µl 10% SDS. Samples corresponding to 106 cpmwere separated on a 15% SDS-PAA gel. Following electrophoresis, the gels werefixed in 10% acetic acid/ 30% methanol and impregnated with En3Hanceautoradiography enhancer (Du Pont, Boston, Ma.), after which they were exposed toBioMax film (Eastman Kodak, Rochester, NY).

Results

Identification of the mltD gene

Upstream of the dniR gene, that was reported to encode a repressor of nitratereductase, a putative open reading frame, yafG, had been identified (19). The YafGprimary sequence was found to contain the three fingerprints that were derivedfrom the structure of the 70 kDa soluble lytic transglycosylase from E. coli and thatwere proposed to be indicative of a muramidase activity (8, 9, 46). Furthermore,YafG was described to contain the bacterial transglycosylase motif, a motif thatcontains two of the above mentioned fingerprints (22). Based on these findingsseveral attempts were made to clone and express YafG, in order to study its potentialtransglycosylase activity. Although the yafG gene was successfully cloned, attemptsto express the protein remained unsuccessful and neither overproduction of YafG onCoomassie blue stained gels nor the induction of a lytic band on zymograms couldbe observed.

Peptidoglycan binding motif in MltD 91

MKAKAILLASVLLVGCQSTGNVQQHAQSLSAAGQGEAAKFTSQARWMDDG 50

TSIAPDGDLWAFIGDELKMGIPENDRIREQKQKYLRNKSYLHDVTLRAEP 100

YMYWIAGQVKKRNMPMELVLLPIVESAFDPHATSGANAAGIWQIIPSTGR 150

NYGLKQTRNYDARRDVVASTTAALNMMQRLNKMFDGDWLLTVAAYNSGEG 200

RVMKAIKTNKARGKSTDFWSLPLPQETKQYVPKMLALSDILKNSKRYGVR 250

LPTTDESRALARVHLSSPVEMAKVADMAGISVSKLKTFNAGVKGSTLGAS 300

GPQYVMVPKKHADQLRESLASGEIAAVQSTLVADNTPLNSRVYTVRSGDT 350

LSSIASRLGVSTKDLQQWNKLRGSKLKPGQSLTIGAGSSAQRLANNSDSI 400

TYRVRKGDSLSSIAKRHGVNIKDVMRWNSDTANLQPGDKLTLFVKNNNMP 450

DS

DniR

YafG

FIG. 1. Primary sequence of MltD. The consensus lipoprotein signal sequence isunderlined and its conserved cysteine is indicated by a flag. The start of the YafGand DniR open reading frames is shown. The black boxes contain the threefingerprints, on the basis of which the lytic transglycosylase activity was proposed.The two copies of the peptidoglycan binding motif (see text for further details) areindicated by the open boxes.

As the reported yafG open reading frame overlapped with the downstream dniRgene, the possibility that the reported sequence might contain a frame-shift mutationand that both open reading frames should, in fact, be fused, was investigated. Thishypothesis was confirmed by comparing the sequence of the yafG region with thatgenerated by the Japanese E. coli sequencing project (11). Moreover, another frame-shift mutation upstream of the yafG gene proved to be present, the correction ofwhich resulted in an N-terminal extension of 101 amino acids. This extensiondisplayed a consensus lipoprotein signal sequence (51), and therefore the correctedopen reading frame was called MltD, for membrane bound transglycosylase D. Thecorrected primary sequence of MltD is shown in Fig. 1.

Expression of MltD

The full length mltD gene was expressed using the T7 promoter system (42). Afterinduction of expression, massive lysis occurred within 15 minutes (Fig. 2). Based onthe proposed muramidase activity, this lysis was most likely caused by hydrolysis ofthe peptidoglycan by the expressed protein. Attempts were therefore made toprotect the cells from this effect by osmotic stabilization. Fig. 2 shows that thisapproach indeed prevented lysis. When these cells were observed using lightmicroscopy, the formation of spheroplast was apparent. This observation, combinedwith that of the failure of the protein to cause lysis when expressed without thesignal sequence, made a strong case for a peptidoglycan hydrolase activity.

92 Chapter 5

0

0.2

0 .4

0 .6

0 .8

1

1.2

0 .00 1.00 2.00 3.00

time (hours)

OD

60

0

no osmotic protection

osmotic protection

FIG. 2. Growth curves of cultures that were induced for the expression of MltD.The arrows indicate the timepoint of induction. The decrease in OD600 afterinduction without osmotic protection, was caused by lysis of the cells as observedby light microscopy. The less pronounced decrease in OD600 for the osmoticallystabilized cells represents a reduced level of lysis and the formation ofspheroplasts.

The proposed peptidoglycan hydrolytic activity was further studied usingzymography. Both periplasmically overproduced (under osmotic protection) andcytoplasmically overproduced MltD were able to degrade polymeric peptidoglycanafter in-gel renaturation. A clear halo was visible around the induced protein bandsafter combined methylene blue and Coomassie staining of the zymograms,indicating that the lytic zone was, in fact, larger than the protein band (Fig. 3). Thisobservation indicates that MltD is a genuine peptidoglycan hydrolase rather than apeptidoglycan-binding protein. Overproduction of the latter type of protein can alsoresult in the formation of a cleared zone on a methylene blue stained zymogram, aswas shown to be the case for the H. influenzae Skp (see chapter 6), but in this case nohalo is formed around the protein band.Further evidence that the formation of lytic zones in the zymogram indeed wascaused by a specific lytic transglycosylase activity, was obtained by in-gel refoldingof the proteins in the presence of bulgecin, an inhibitor of the soluble lytictransglycosylases (44, 47). The presence of bulgecin prevented peptidoglycanhydrolysis by overproduced MltD, indicating that the protein is indeed closelyrelated to the Slt70 type of lytic transglycosylases (Fig. 4).

Peptidoglycan binding motif in MltD 93

64

50

98

36

30

A B

1 2 31 2 3 4 5 6

FIG 3. Combined analysis of expression and activity of MltD. The displayedzymograms were processed as described in the materials and methods section,after which they were stained with both methylene blue and Coomassie brilliantblue R250. Panel A. Lane 1, prestained marker proteins of indicated sizes; lane 2 ,BL21(DE3)+pETMltD, grown under osmotic protection, before induction; lane 3,after induction; lane 4, soluble fraction of induced BL21(DE3)+pETMltD; lane 5,membrane fraction of induced BL21(DE3)+pMltD; lane 6, zymogram markerscontaining 1µg Slt70 and 2 µg mutanolysin. Panel B. Lane 1, prestained markers;lane 2, BL21(DE3)+pETMltDns before induction; lane 3, after induction.

0 2 10 50 1000

1

2

3

4

5

6

7

0 2 1 0 5 0 1 0 0

Dilution

OD

/mm

2

MltDMltD +bulgecin

A

B

C

Slt70

MltDMltD*

Slt70

MltDMltD*

FIG 4. Bulgecin sensitivity of MltD. Panel A, Zymogram analysis of dilutions ofa mixture of overproduced MltD(without signal sequence) and 4 µg Slt70.Dilution factors are indicated. Panel B. Analysis of the same dilutions, processedin the presence of 100 µg/ml bulgecin. C. Densitometric analysis of processedzymograms. Values were corrected for local differences in background stainingby normalization with respect to lytic zones generated by Slt35(MltB) (notdepicted in panel A/B). Slt35(MltB) is not sensitive to bulgecin (36).

94 Chapter 5

Lipoprotein character

As the sequence of MltD displays a consensus lipoprotein signal sequence, the invivo processing of this signal sequence was studied. Careful inspection of the majorlytic band produced by native MltD, showed that this band was composed of amajor band, that may correspond to unprocessed protein, and a minor band ofsmaller molecular weight that may correspond to processed protein and of which solittle protein is present that it hardly is visualized by Coomassie staining (Fig. 3 lane3 and 5). However, an extract containing cytoplasmically overproduced protein alsoshowed the presence of a smaller molecular weight band when analyzed onzymogram (Fig. 4 A). Therefore it cannot be excluded that the band that presumablycorresponds to the processed protein is an unspecific proteolytic degradationproduct or the product of a premature translational stop. Further testing of thelipoprotein character of MltD was therefore undertaken by 3H-palmitate labeling ofcells that were induced for the expression of full-length MltD. Based on the analysisof labeled proteins by fluorography, no specific labeling of overproduced MltDcould be concluded (data not shown). This inconclusive result could be caused by ageneral inefficiency of the labeling process, as signals from the endogenouslipoproteins were also only very weak.

Activity measurements of cellular extracts

Analysis of the solubility of the produced protein, showed that overexpression of thefull length MltD resulted in the formation of insoluble aggregates (Fig. 3A, lane 5).Of the version of MltD that was produced in the cytoplasm, about 50% proved to besoluble (data not shown). Although the in-gel refolding of both types of proteinresulted in peptidoglycan hydrolytic activity, attempts to measure the activitydirectly in cell lysates were unsuccessful. Using both the standard and the highthroughput assay that measure the solubilization of radiolabeled peptidoglycan, noactivity could be detected that was significantly higher than the backgroundpeptidoglycan hydrolysis caused by endogenous hydrolases present in the extracts.Also HPLC analysis of purified peptidoglycan that was incubated with crudeextracts of the cells before and after induction, did not show significant differences.Several attempts were made to optimize the conditions for detection of MltDactivity, including stepwise variation of the pH from 5.5 to 8.5 and predigestion orcodigestion of the substrate with endopeptidases such as purified PBP4. None ofthese approaches resulted in a significant increase in activity.

Presence of a putative peptidoglycan binding motif

The C-terminal part of MltD shows the presence of two copies of a short sequencemotif (Fig. 1). This motif is homologous to a motif that has been described to bepresent in one or more copies in cell-wall associated proteins from gram-positivebacteria. The peptidoglycan hydrolases of Enterococcus hirae, Streptococcus faecalis andLactococcus lactis contain this motif in 6, 5 and 3 copies, respectively (1, 4, 5).Furthermore, the motif has been described to be present in multiple copies in afamily of virulence related proteins from Listeria spp (21). These proteins arepossibly also involved in peptidoglycan hydrolysis, based on their described role incell separation (24). The lysins of several gram-positive phages contain the motif,two copies are found in a endopeptidase from Bacillus sphaericus and single copiesare present in Staphylococcal protein A and XlyA, a prophage endolysin from B.

Peptidoglycan binding motif in MltD 95

subtilis. The presence of this motif in all of these gram-positive cell wallprocessing/associated proteins has led to the proposal that it may be involved inpeptidoglycan binding (17, 18, 29).The presence of this proposed gram-positive peptidoglycan binding motif in E. coliDniR (4, 29), was noticed before but the implications of this observation were notdiscussed. Our results now show that these homologous motifs belong ,in fact, toMltD. Although this is the first demonstration of the presence of a gram-positivepeptidoglycan binding motif in a gram-negative peptidoglycan hydrolase, it may notbe an unique case though, as we were also able to identify single copies of the motifin a family of putative lysostaphin homologs. Lysostaphin is a peptidoglycanhydrolase that has been identified in Staphylococcus simulans and S. staphylolyticus(14, 34). It cleaves the pentaglycine crosslink in the staphylococcal peptidoglycan.Over the last couple of years, a number of sequences of which the C-terminal partshows strong homology to the active domain of lysostaphin have been identified ingram-negative bacteria. In E. coli, for instance, no less than four such homologs seemto present, NlpD, YebA, YibP (20, 27, 41), and an undescribed homolog that could beidentified from E. coli genomic sequence (and is called NlpE in Fig. 5). Furtherhomologs are present in Haemophilus spp., Vibrio cholerae, Pseudomonas aeruginosa andSalmonella spp. and Yersinia enterocolitica (23, 43, 45). These findings are puzzling asgram-negative peptidoglycan does not contain pentaglycine crossbridges. Theactivity and function of these gram-negative lysostaphin homologs thereforeremains elusive. Although the N-terminal part of the sequence is not overallconserved in this family, we found that a subset of these lysostaphin homologsshows the presence of one copy of the above mentioned peptidoglycan bindingmotif. Furthermore the H. influenzae AmiB homolog also displays one copy of themotif. An alignment of the motifs found in MltD, the lysostaphin homologs andAmiB is shown in figure 5.

Y T I K S G D T L N K I S A Q F G V S V A N L R S W N G I K G D L - - I F A G Q T I I V Y T V K K G D T L F Y I A W I T G N D F R D L A Q R N N I Q A P Y A - L N V G Q T L Q V Y I V R R G D T L Y S I A F R F G W D W K A L A A R N G I A P P Y T - I Q V G Q A I Q F Y K V N K G D T M F L I A Y L A G I D V K E L A A L N N L S E P Y N - L S L G Q V L K I Y T V K R G D T L Y R I S R T T G T S V K E L A R L N G I S P P Y T - I E V G Q K L K L Y V V S T G D T L S S I L N Q Y G I D M G D I S Q L A A A D K E L R N L K I G Q Q L S W

Y T V

T E G D T L K D V L V L S G L D D S S V Q P L I A L D P E L A H L K A G Q Q F Y W Y T V R S G D T L S S I A S R L G V S T K D L Q Q W N K L R G S - K - L K P G Q S L T I Y R V R K G D S L S S I A K R H G V N I K D V M R W N S - D T A - N - L Q P G D K L T L H I V K K G E S L G S L S N K Y H V K V S D I I K L N Q L K R K T - - L W L N E S I K I

Ef-AlysEc_NlpdPa_NlpdHi_LppBEc_NlpEEc_YebAHi_YebAEc_MltD_1Ec_MltD_2Hi_AmiB

FIG. 5. Alignment of the proposed peptidoglycan binding motifs as identified ingram-negative proteins with a representative copy from a gram-positivepeptidoglycan hydrolase. Ef_Alys, the fourth copy of five repeats from theautolysin of Streptococcus faecalis (amino acid 567-609); Ec_NlpD (123-166),Pa_NlpD (69-112), Hi_LppB (147-190), Ec_NlpE (50-93), Ec_YebA (77-121),Hi_YebA (131-175), lysostaphin homologs from E. coli, H. influenzae and P .aeruginosa; Ec_MltD_1 (343-384,) Ec_MltD_2 (402-442), two copies of the motif aspresent in MltD; Hi_AmiB (294-335), AmiB homologue from H. influenzae.

Discussion

The identification of YafG as a putative member of the lytic transglycosylase familyof E. coli was puzzling for two reasons. First the sequence did not display a signal

96 Chapter 5

sequence and therefore was not likely to be transported to the compartment whereits proposed substrate is located and, second, this sequence seemed to code for a truebacterial lysozyme, without an additional domain that could be involved inattenuation of this potentially very dangerous activity.The results of this study show that the originally reported sequence of YafG wastruncated due to the presence of sequencing errors, as was also shown to be the casefor another transglycosylase homolog, the E. coli MltC. In fact, the YafG sequence isextended on both the N-terminal and C-terminal side. This observation implies thatcare should be taken when evaluating open reading frames derived directly fromraw sequence data, this notion being underscored by our identification of a frame-shift mutation upstream of yet another member of the lytic transglycosylase family(unpublished data).The correct sequence of yafG, now called mltD, fully incorporates the tentative dniRgene. This gene has been described as a positive regulator of the E. coli hexahemenitrate reductase. DniR was identified during an attempt to clone the gene coding forthis nitrate reductase by means of immunological screening of a library with apolyclonal antiserum raised against the purified protein (19). As it is hard to imaginea physiological connection between this proposed function and peptidoglycanmetabolism, the most conceivable explanation for this paradox is, that the originalpreparation of nitrate reductase was contaminated with MltD, thus causing theantiserum to show cross-reactivity. This hypothesis is consistent with the fact thatthe only assay used to demonstrate the effect of the cloned dniR gene on theexpression of nitrate reductase, which has about the same size as MltD, isimmunological detection using the same antiserum as was used for the screening ofthe library. Furthermore, the dniR deletion that was constructed would most likelyresult in the expression of a truncated form of MltD and indeed a specific band ofsmaller molecular weight can be discriminated on the immunoblot of the deletionmutant that is displayed in the communication (19).The observation that no clear activity of MltD could be measured in solution, eventhough activity could be convincingly shown on zymograms, is puzzling. One has tobear in mind though that in both assay systems the degree of hydrolysis that isneeded to generate a signal is unknown. It is therefore imaginable that thehydrolysis of only a few bonds would suffice for the peptidoglycan to be releasedfrom the polyacrylamide matrix of a zymogram, whereas the conversion to productsthat cannot be precipitated by CTAB anymore, as measured in the solution assay,requires many more bonds to be hydrolyzed. Furthermore the zymograms areincubated for twelve hours or more, allowing even very weakly active proteins togenerate lytic zones. These long incubations are not feasible for the assays insolution, as the signal would be masked by the total degradation of the substrate byendogenous hydrolases that are present in the extracts. The apparent low activity ofthe full length protein was also observed for MltC, another member of the lytictransglycosylase family that was recently identified (8). The low activities of MltCand MltD may reflect a tight regulation of the lysozyme-like domain, possiblymediated by other domains of the protein, as seems to be the case for MltC. If themain function of these enzymes would be in overall peptidoglycan metabolism, onemight expect that their activities would have been detected during the numerousstudies that were aimed at identifying peptidoglycan hydrolase activities in cellextracts. As this is not the case, one may speculate that the function of these enzymesis in very specific processes, rather than in overall peptidoglycan metabolism.

Peptidoglycan binding motif in MltD 97

Alternatively, the complex topological arrangement of MltD may imply that theenzyme can only attain its full activity in the in vivo situation. Further studies on theactivity of MltD await purification of the protein which is currently underway.An exciting feature of this new member of the transglycosylase family is thepresence of two copies of a proposed peptidoglycan binding motif. This is the firstdemonstration of such a motif in a gram-negative peptidoglycan hydrolase andimplies that this specific principle of cell-wall interaction is not limited to gram-positive cell-wall associated proteins. As gram-negative peptidoglycan interactingproteins are trapped in the periplasm, the presence of this repeat as a means ofretention, as was discussed for some gram-positive repeat-bearing proteins (52),seems unlikely. Also a direct involvement in catalysis is improbable as the samerepeats are present in peptidoglycan hydrolases of totally different specificity andother members of the lytic transglycosylase family do not need these repeats foractivity. The proposed lipoprotein character of MltD complicates matters even more.Unfortunately, the in vivo lipidation could not be demonstrated in this study, whichmay be explained by the observation that the bulk of the overproduced proteinseems not to fold correctly and ends up in inclusion bodies. Moreover, thedevastating effect of the expression of MltD on the cell wall may interfere with theuptake of the 3H-palmitate or the processing of the preproteins, an assumption thatis supported by the inefficient labeling of endogenous lipoproteins. Based on thestructure of the signal sequence, it is, however, very likely that MltD is a lipoprotein.In this respect it will be interesting to study the effects of globomycin, an inhibitor ofthe lipoprotein-specific signal peptidase, on the induction of lysis by MltD Theresulting unique topological arrangement: an N-terminal domain that is linked tothe outer membrane and a C-terminal domain that binds to the peptidoglycan, isreminiscent of the peptidoglycan associated lipoproteins (28). Those proteins are,however, structural proteins. This arrangement of MltD may serve regulatorypurposes for controlling the activity of the central transglycosylase domain. The N-terminal and C-terminal domains could also be involved in interactions with otherproteins, as has been discussed for other members of the lytic transglycosylasefamily (16), thus localizing the resulting complex at a defined position in the cell-envelope. A comparable type of topological regulation may apply some of themembers of the family of gram-negative lysostaphin homologs. The NlpD subsetalso combines the presence of the peptidoglycan binding motif with a lipoproteincharacter. However, the organization is different as the motif is located in the N-terminal part of the sequence and therefore the membrane attachment andpeptidoglycan interaction seem to be mediated by the same domain. The function oreven the activity of these lysostaphin homologs remains unknown but maybe theconcept of lysostaphin being a gram-positive peptidoglycan hydrolase for whichthere are gram-negative homologs has to be reversed, as evidence is accumulatingthat this family of proteins seems to be much more abundant and conserved ingram-negative bacteria. Interestingly, results of a recent study suggest thatlysostaphin may have some effect on B. subtilis, the peptidoglycan of which ischemically related to E. coli and also devoid of pentaglycine cross-bridges (50).The presence of the peptidoglycan binding motif seems to be a common featureshared by a whole range of distinct peptidoglycan hydrolases. Figure 6 shows thedistribution of the motif in representative examples of established and putativepeptidoglycan hydrolases of different specificity, from both gram-positives andgram-negatives. From the large variance in the number of motifs and their position,

98 Chapter 5

the motif appears to have been incorporated into the different sequences as aseparate module. This hypothesis is supported by the observation that several pairsor families of conserved sequences differ with respect to the presence of the motif.For instance, the XlyA amidase shows homology over its full length with the CwlAtype amidases, except for a central stretch of sequence that contains thepeptidoglycan binding motif. Another interesting example is the AmiB amidasehomolog of H. influenzae that has a C-terminal extension which is absent in E. coliAmiB and that contains a copy of the motif. Furthermore, only a subset of the gram-negative lysostaphin homologs contains the motif and MltD is the only lytictransglycosylase identified so far in which it is found. The function of the motif isstill elusive. Although it is very likely that it is in some way involved inpeptidoglycan binding, this remains to be shown experimentally and the issue as towhy it is present in certain hydrolases whereas it is absent in close homologs ofthese, is very intriguing.

100 200 300 400 500 600 700

Alys muramidase

Enp1 endopeptidase

XlyA amidase

“lysostaphin”

MltD lytic transglycosylase

AmiB amidase

gram-positive

gram-negative

NlpD

YebA

FIG. 6. Distribution of the peptidoglycan binding motif in gram-positive and gram-negative hydrolases. : the motif; Alys, autolysin from S. faecalis, : region of homology to E. hirae and L. lactis muramidases; Enp1, endopeptidase from B. spaericus; XlyA, amidase from B. subtilis prophage PBSX, : region of homology to CwlA family of gram-positive amidases; NlpD and YebA, lysostaphin homologs from E. coli, : region of homology to lysostaphin; MltD from E. coli, : lytic transglycosylase domain; AmiB from H. influenzae, : homology to E. coli AmiB and B. subtilis CwlB/D, : homology only to E. coli AmiB. The presence of a lipoprotein signal sequence is marked by a flag. The ruler indicates sequence length.

The identification of MltD as described in this report adds another member to thefamily of lytic transglycosylases in E. coli, the modular built-up of which again isdifferent from that of the other members. Unfortunately, the physiological functionof MltD remains elusive, as is true for all other members of this family, which makesthe question as to why E. coli keeps such a large arsenal of lytic enzymes around,even more fascinating. The truncation of the open reading frame as performed in aneffort to delete the dniR gene, could indicate that the function of MltD is not essential

Peptidoglycan binding motif in MltD 99

for survival of the cell under laboratory conditions or, alternatively, that the C-terminal part of the protein is not essential for its function. No MltD homolog couldbe identified from H. influenzae genomic sequence but the sole transglycosylasehomolog that could be identified from the almost completed genome of Neisseriagonorrhoeae, is an MltD homologue (unpublished observation). As this organism isknown to release large amounts of the products of the lytic transglycosylases duringgrowth (40), the MltD homolog may have a role in peptidoglycan turn-over in thisorganism.

References

1. Beliveau, C., C. Potvin, J. Trudel, A. Asselin and G. Bellemare. 1991. Cloning,sequencing, and expression in Escherichia coli of a Streptococcus faecalis autolysin. J.Bacteriol. 173: 5619-23.

2. Bernadsky, G., T. J. Beveridge and A. J. Clarke. 1994. Analysis of the sodium dodecylsulfate-stable peptidoglycan autolysins of select gram-negative pathogens by usingrenaturing polyacrylamide gel electrophoresis. J. Bacteriol. 176: 5225-5232.

3. Beveridge, T. and L. Graham. 1991. Surface layers of bacteria. Microbiol. Rev. 55: 684-705.4. Buist, G., J. Kok, K. Leenhouts, M. Dabrowska, G. Venema and A. Haandrikman. 1995.

Molecular cloning and nucleotide sequence of the gene encoding the majorpeptidoglycan hydrolase of Lactococcus lactis, a muramidase needed for cell separation.J. Bacteriol. 177: 1554-63.

5. Chu, C., R. Kariyama, L. Daneo-Moore and G. Shockman. 1992. Cloning and sequenceanalysis of the muramidase-2 gene from Enterococcus hirae. J. Bacteriol. 174: 1619-25.

6. Di Berardino, M., A. J. Dijkstra, D. Stüber, W. Keck and M. Gubler. 1996. Themonofunctional glycosyltransferase of Escherichia coli is a member of new class ofpeptidoglycan-synthesising enzymes. FEBS Lett. 392: 184-188.

7. Dijkstra, A. J., F. Hermann and W. Keck. 1995. Cloning and controlled overexpression ofthe gene encoding the 35 kDa soluble lytic transglycosylase from Escherichia coli. FEBSLett. 366: 115-118.

8. Dijkstra, A. J. and W. Keck. 1996. Identification of new members of the lytictransglycosylase family in Haemophilus influenzae and Escherichia coli. Microb. DrugResist. 2: 141-145.

9. Dijkstra, B. W. and A. M. Thunnissen. 1994. 'Holy' proteins. II: The soluble lytictransglycosylase. Curr. Opin. Struct. Biol. 4: 810-3.

10. Engel, H., B. Kazemier and W. Keck. 1991. Murein-metabolizing enzymes fromEscherichia coli: sequence analysis and controlled overexpression of the slt gene, whichencodes the soluble lytic transglycosylase. J. Bacteriol. 173: 6773-82.

11. Fujita, N., H. Mori, T. Yura and A. Ishihama. 1994. Systematic sequencing of theEscherichia coli genome: analysis of the 2.4-4.1 min (110,917-193,643 bp) region. NucleicAcids Res. 22: 1637-9.

12. Ghuysen, J., J. Lamotte-Brasseur, B. Joris and G. Shockman. 1994. Binding site-shapedrepeated sequences of bacterial wall peptidoglycan hydrolases. FEBS Lett. 342: 23-8.

13. Harz, H., K. Burgdorf and J.-V. Höltje. 1990. Isolation and separation of the glycanstrands from murein of Escherichia coli by reversed-phase high-performance liquidchromatography. Anal. Biochem. 190: 120-8.

14. Heinrich, P., R. Rosenstein, M. Bohmer, P. Sonner and F. Gotz. 1987. The molecularorganization of the lysostaphin gene and its sequences repeated in tandem. Mol. Gen.Genet. 209: 563-9.

15. Höltje, J. and A. Tomasz. 1975. Specific recognition of choline residues in the cell wallteichoic acid by the N-acetylmuramyl-L-alanine amidase of Pneumococcus. J. Biol. Chem.250: 6072-6.

16. Höltje, J. V. 1996. Molecular interplay of murein synthases and murein hydrolases inEscherichia coli. Microb. Drug. Resist. 2: 99-103.

100 Chapter 5

17. Hourdou, M., M. Guinand, M. Vacheron, G. Michel, L. Denoroy, C. Duez, S. Englebert,B. Joris, G. Weber and J. Ghuysen . 1993. Characterization of the sporulation-relatedgamma-D-glutamyl-(L)meso-diaminopimelic-acid-hydrolysing peptidase I of Bacillussphaericus NCTC 9602 as a member of the metallo(zinc) carboxypeptidase A family.Modular design of the protein. Biochem. J. 292 ( Pt 2): 563-70.

18. Joris, B., S. Englebert, C. Chu, R. Kariyama, L. Daneo-Moore, G. Shockman and J.Ghuysen. 1992. Modular design of the Enterococcus hirae muramidase-2 andStreptococcus faecalis autolysin. FEMS Microbiol Lett 70: 257-64.

19. Kajie, S., R. Ideta, I. Yamato and Y. Anraku . 1991. Molecular cloning and DNA sequenceof dniR, a gene affecting anaerobic expression of the Escherichia coli hexaheme nitritereductase. FEMS Microbiol. Lett. 67: 205-11.

20. Karow, M. and C. Georgopoulos. 1992. Isolation and characterization of the Escherichiacoli msbB gene, a multicopy suppressor of null mutations in the high-temperaturerequirement gene htrB. J. Bacteriol. 174: 702-10.

21. Kohler, S., M. Leimeister-Wachter, T. Chakraborty, F. Lottspeich and W. Goebel. 1990.The gene coding for protein p60 of Listeria monocytogenes and its use as a specific probefor Listeria monocytogenes. Infect. Immun. 58: 1943-50.

22. Koonin, E. V. and K. E. Rudd. 1994. A Conserved Domain in Putative Bacterial andBacteriophage Transglycosylases. Trends Biochem. Sci. 19: 106-107.

23. Kovach, M., K. Hughes, K. Everiss and K. Peterson. 1994. Identification of a ToxR-activated gene, tagE, that lies within the accessory colonization factor gene cluster ofVibrio cholerae O395. Gene 148: 91-5.

24. Kuhn, M. and W. Goebel. 1989. Identification of an extracellular protein of Listeriamonocytogenes possibly involved in intracellular uptake by mammalian cells. Infect.Immun. 57: 55-61.

25. Labischinski, H., E. W. Goodell, A. Goodell and M. L. Hochberg. 1991. Direct proof of a"more-than-single-layered" peptidoglycan architecture of Escherichia coli W7: a neutronsmall-angle scattering study. J Bacteriol 173: 751-6.

26. Labischinski, H. and H. Maidhof. 1994. Bacterial peptidoglycan: overview and evolvingconcepts. p. 23-38. In J.-M. Ghuysen and R. Hakenbeck (ed.), Bacterial Cell Wall.Elsevier, Amsterdam.

27. Lange, R. and R. Hengge-Aronis. 1994. The nlpD gene is located in an operon with rpoSon the Escherichia coli chromosome and encodes a novel lipoprotein with a potentialfunction in cell wall formation. Mol. Microbiol. 13: 733-43.

28. Lazzaroni, J. and R. Portalier . 1992. The excC gene of Escherichia coli K-12 required forcell envelope integrity encodes the peptidoglycan-associated lipoprotein (PAL). Mol.Microbiol. 6: 735-42.

29. Longchamp, P., C. Mauel and D. Karamata. 1994. Lytic enzymes associated withdefective prophages of Bacillus subtilis: sequencing and characterization of the regioncomprising the N-acetylmuramoyl-L-alanine amidase gene of prophage PBSX.Microbiology 140 ( Pt 8): 1855-67.

30. Merchante, R., H. M. Pooley and D. Karamata. 1995. A periplasm in B. subtilis. J.Bacteriol. 177: 6176-6183.

31. Navarre, W., S. Daefler and O. Schneewind. 1996. Cell wall sorting of lipoproteins inStaphylococcus aureus. J Bacteriol 178: 441-6.

32. Park, W. and M. Matsuhashi. 1984. Staphylococcus aureus and Micrococcus luteuspeptidoglycan transglycosylases that are not penicillin-binding proteins. J. Bacteriol. 157:538-44.

33. Parquet, C., B. Flouret, M. Leduc, Y. Hirota and H. J. van. 1983. N-acetylmuramoyl-L-alanine amidase of Escherichia coli K12. Possible physiological functions. Eur. J. Biochem.133: 371-7.

34. Recsei, P., A. Gruss and R. Novick. 1987. Cloning, sequence, and expression of thelysostaphin gene from Staphylococcus simulans. Proc. Natl. Acad. Sci. U S A 84: 1127-31.

35. Roeder, W. and R. Somerville. 1979. Cloning the trpR gene. Mol. Gen. Genet. 176: 361-8.36. Romeis, T., W. Vollmer and J. V. Höltje. 1993. Characterization of Three Different Lytic

Transglycosylases in Escherichia coli. FEMS Microbiol Lett. 111: 141-146.37. Sambrook, R., E. Fritsch and T. Maniatis. 1989. Molecular Cloning: a Laboratory

Manual, 2nd ed. Cold Spring Harbor Laboratory, Cold Spring Harbor, N. Y.

Peptidoglycan binding motif in MltD 101

38. Sanz, J., E. Diaz and J. Garcia. 1992. Studies on the structure and function of the N-terminal domain of the pneumococcal murein hydrolases. Mol. Microbiol. 6: 921-31.

39. Shockman, G. D. and J.-V. Höltje. 1994. Microbial peptidoglycan (murein) hydrolases. p.131-166. In J.-M. Ghuysen and R. Hakenbeck (ed.), Bacterial Cell Wall. Elsevier,Amsterdam.

40. Sinha, R. K. and R. S. Rosenthal. 1980. Release of Soluble Peptidoglycan from GrowingGonococci: Demonstration of Anhydro-muramyl-containing Fragments. Infect. Immun.29: 914-924.

41. Sofia, H., V. Burland, D. Daniels, G. r. Plunkett and F. Blattner. 1994. Analysis of theEscherichia coli genome. V. DNA sequence of the region from 76.0 to 81.5 minutes.Nucleic Acids Res. 22: 2576-86.

42. Studier, F. W., A. H. Rosenberg, J. J. Dunn and J. W. Dubendorff. 1990. Use of T7 RNApolymerase to direct expression of cloned genes. Methods Enzymol 185: 60-89.

43. Tanaka, K. and H. Takahashi. 1994. Cloning, analysis and expression of an rpoShomologue gene from Pseudomonas aeruginosa PAO1. Gene 150: 81-5.

44. Templin, M. F., D. H. Edwards and J.-V. Höltje. 1992. A Murein Hydrolase is theSpecific Target of Bulgecin in Escherichia coli. J. Biol. Chem. 267: 20039-20043.

45. Theisen, M., C. Rioux and A. Potter. 1993. Molecular cloning, nucleotide sequence, andcharacterization of lppB, encoding an antigenic 40-kilodalton lipoprotein of Haemophilussomnus. Infect. Immun. 61: 1793-8.

46. Thunnissen, A.-M. W. H., N. W. Isaacs and B. W. Dijkstra. 1995. The catalytic domain ofa bacterial lytic transglycosylase defines a novel class of lysozymes. Proteins 22: 245-258.

47. Thunnissen, A. M. W. H., H. J. Rozeboom, K. H. Kalk and B. W. Dijkstra. 1995. Structureof the 70-kDa soluble lytic transglycosylase complexed with bulgecin A. Implicationsfor the enzymatic mechanism. Biochemistry 34: 12729-12737.

48. Tomioka, S., T. Nikaido, T. Miyakawa and M. Matsuhashi. 1983. Mutation of the N-acetylmuramyl-L-alanine amidase gene of Escherichia coli K-12. J. Bacteriol. 156: 463-5.

49. Tsui, H., G. Zhao, G. Feng, H. Leung and M. Winkler. 1994. The mutL repair gene ofEscherichia coli K-12 forms a superoperon with a gene encoding a new cell-wall amidase.Mol. Microbiol. 11: 189-802.

50. Ulijasz, A. T., A. Grenader and B. Weisblum. 1996. A vancomycin-inducible LacZreporter system in Bacillus subtilis: induction by antibiotics that inhibit cell wall syntesisand by lysozyme. J. Bacteriol. 178: 6301-6309.

51. von-Heijne, G. 1989. The structure of signal peptides from bacterial lipoproteins. ProteinEng. 2: 531-4.

52. Yother, J. and J. White. 1994. Novel surface attachment mechanism of the Streptococcuspneumoniae protein PspA. J. Bacteriol. 176: 2976-85.