Effect of Deletion of Genes Involved in Lipopolysaccharide ... · Truncating the LPS O-antigen and ceasing its synthesis in vivo are strategies for attenuating Salmonella for use

INFECTION AND IMMUNITY, Oct. 2011, p. 4227–4239 Vol. 79, No. 100019-9567/11/$12.00 doi:10.1128/IAI.05398-11Copyright © 2011, American Society for Microbiology. All Rights Reserved.

Effect of Deletion of Genes Involved in Lipopolysaccharide Core andO-Antigen Synthesis on Virulence and Immunogenicity

of Salmonella enterica Serovar Typhimurium�

Qingke Kong,1 Jiseon Yang,1 Qing Liu,1 Praveen Alamuri,1Kenneth L. Roland,1 and Roy Curtiss III1,2*

Center for Infectious Diseases and Vaccinology, The Biodesign Institute, Arizona State University, Tempe, Arizona 85287,1 andSchool of Life Sciences, Arizona State University, Tempe, Arizona 852872

Received 23 May 2011/Returned for modification 24 June 2011/Accepted 7 July 2011

Lipopolysaccharide (LPS) is a major virulence factor of Salmonella enterica serovar Typhimurium and iscomposed of lipid A, core oligosaccharide (C-OS), and O-antigen polysaccharide (O-PS). While the functionsof the gene products involved in synthesis of core and O-antigen have been elucidated, the effect of removingO-antigen and core sugars on the virulence and immunogenicity of Salmonella enterica serovar Typhimuriumhas not been systematically studied. We introduced nonpolar, defined deletion mutations in waaG (rfaG), waaI(rfaI), rfaH, waaJ (rfaJ), wbaP (rfbP), waaL (rfaL), or wzy (rfc) into wild-type S. Typhimurium. The LPSstructure was confirmed, and a number of in vitro and in vivo properties of each mutant were analyzed. Allmutants were significantly attenuated compared to the wild-type parent when administered orally to BALB/cmice and were less invasive in host tissues. Strains with �waaG and �waaI mutations, in particular, weredeficient in colonization of Peyer’s patches and liver. This deficiency could be partially overcome in the �waaImutant when it was administered intranasally. In the context of an attenuated vaccine strain delivering thepneumococcal antigen PspA, all of the mutations tested resulted in reduced immune responses against PspAand Salmonella antigens. Our results indicate that nonreversible truncation of the outer core is not a viableoption for developing a live oral Salmonella vaccine, while a wzy mutant that retains one O-antigen unit isadequate for stimulating the optimal protective immunity to homologous or heterologous antigens by oral,intranasal, or intraperitoneal routes of administration.

Live attenuated Salmonella vaccines can colonize the gut-associated lymphoid tissue (e.g., Peyer’s patches [PP]) and thesecondary lymphatic tissues to elicit immune responses againstboth Salmonella and the heterologous antigens delivered bythe vector during colonization of the host (9, 10, 12). Attenu-ation of Salmonella virulence can be achieved either by delet-ing the genes involved in central metabolic pathways such asaroA, aroC, purA, pabA, and pabB or by deletion or regulatedexpression of global regulators such as phoPQ, fur, and crp andother genes important for establishing an infection (reviewedin references 8, 10, 12, and 14).

Lipopolysaccharide (LPS) of Salmonella is a recognized vir-ulence determinant and is essential for functions, includingswarming motility (60), intestinal colonization (47), serum re-sistance (57), invasion/intracellular replication (43), and resis-tance to killing by macrophages, all of which are critical forsuccessful infection by the pathogen. LPS consists of threedifferent regions: a conserved lipid A, a short core oligosac-charide (C-OS), and the O-antigen polysaccharide (O-PS) as-sembled in a variable number of oligosaccharide repeatingunits (51). Lipid A is a ligand that stimulates the TLR4-MD2-CD14 pathway to activate the antimicrobial host defense dur-

ing bacterial infection (41, 50). The O-PS is Salmonella’s pri-mary defense against serum complement activation (42).Truncating the LPS O-antigen and ceasing its synthesis in vivoare strategies for attenuating Salmonella for use as live vac-cines.

Galactose is a component of both core and O-antigen, andmannose is a component of O-antigen. The galE and pmi genesencode enzymes that are critical for UDP-galactose and GDP-mannose synthesis, respectively. Thus, galE (19, 29) and pmi(13, 37, 58) mutants of Salmonella enterica serovar Typhimu-rium are conditionally rough (lacking O-antigen), dependingon the availability of galactose or mannose, respectively, re-sulting in strains that are attenuated and immunogenic. Foruse as vaccines, these strains are grown with their respectivesugars prior to immunization and produce complete O-anti-gen. RfaH, a transcriptional antiterminator, is required for thesynthesis of secreted and surface-associated cell components ofS. Typhimurium, including O-PS and core sugar componentsof LPS (2). An rfaH deletion mutant is described as “gentlyrough,” and exhibits deep-rough characteristics and stimulatesprotective immune response against Salmonella challenge inBALB/c mice (13, 28). The wzy (rfc) gene encodes the O-an-tigen ligase, required for polymerization of multiunit O-anti-gen molecules. A wzy (rfc) mutant produces semirough LPS,consisting of a complete core and a single O-antigen subunit.We have previously shown that a �pabA �pabB vaccine straincarrying a �wzy mutation is immunogenic and can effectivelydeliver the heterologous Streptococcus pneumoniae antigenPspA, resulting in protective immunity against Streptococcus

* Corresponding author. Mailing address: Center for Infectious Dis-eases and Vaccinology, The Biodesign Institute, Arizona State Uni-versity, P.O. Box 875401, 1001 S. McAllister Avenue, Tempe, AZ85287-5401. Phone: (480) 727-0445. Fax: (480) 727-0466. E-mail:[email protected].

� Published ahead of print on 18 July 2011.

4227

on June 30, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

http://iai.asm.org/

pneumoniae challenge (34). The immunogenicity of rfaH andwzy mutant strains can be further enhanced by programmeddownregulation of rfaH or wzy expression in a recombinantattenuated Salmonella vaccine (RASV) without compromisingtheir ability to colonize lymphoid tissues (34, 35). These pre-vious studies led us to consider manipulation of other genesinvolved in the synthesis of LPS, including the genes for coreoligosaccharide (C-OS) that have not been explored previouslyin the context of vaccine construction.

C-OS is assembled by the sequential transfer of sugars fromnucleotide mono-sugar or diphospho-sugar donors to a lipid Aacceptor by specific glycosyltransferases in the inner mem-brane (17, 63). The genes encoding these proteins are orga-nized in the waa operon (rfa operon) (53). The potential ofSalmonella mutants with truncated core and O-antigen as at-tenuated vaccines has not been systematically analyzed. In thiswork, we constructed strains with an ordered series of S. Ty-phimurium nonpolar deletion mutations causing stepwise trun-

cations of core sugars and evaluated these mutant strains forvirulence and ability to colonize BALB/c mice. The effect ofeach mutation in a well-characterized attenuated vaccine strainon its ability to deliver the Streptococcus pneumoniae antigenPspA and to affect an antigen-specific protective immune re-sponse was also determined.

MATERIALS AND METHODS

Bacterial strains, plasmids, media, and growth conditions. The strains andplasmids used in this study are listed in Table 1. S. Typhimurium cultures weregrown at 37°C in LB broth (4) or nutrient broth (Difco) or on LB agar with orwithout 0.1% arabinose. Selenite broth, with or without supplements, was usedfor enrichment of Salmonella from mouse tissues. Diaminopimelic acid (DAP)(50 �g/ml) was added for the growth of �asd strains (45). LB agar containing 5%sucrose was used for sacB gene-based counterselection in allelic exchange ex-periments. MOPS (morpholinepropanesulfonic acid) minimal medium (46) with/without 10 �g/ml p-aminobenzoic acid (pABA) was used to confirm the pheno-type of �pabA �pabB mutations.

TABLE 1. Bacterial strains and plasmids used in this study

Strain or plasmid Descriptiona Source

Suicide vectorspRE112 sacB mobRP4 R6K ori Cm� 15pYA4278 sacB mobRP4 R6K ori Cm�, pRE112-T-vector Lab collectionpYA4896 �waaG42 pRE112pYA4897 �waaI43 pRE112pYA4718 �rfaH49 35pYA4898 �waaJ44 pRE112pYA4899 �wbaP45 pRE112pYA4900 �waaL46 pRE112pYA4717 �wzy-48 34

Recombinant plasmidspYA3493 Plasmid Asd�; pBR ori �-lactamase signal sequence-based periplasmic secretion

plasmid (blaSS)32

pYA4088 852-bp DNA encoding the �-helical region of PspA (aa 3-285) intopYA3493; blaSS-pspA

65

S. Typhimurium�3761 UK-1 24�11308 �waaG42 In �3761�11309 �waaI43 In �3761�9945 �rfaH49 35�11310 �waaJ44 In �3761�11311 �wbaP45 In �3761�11312 �waaL46 In �3761�9944 �wzy-48 In �3761�9241 �pabA1516 �pabB232 �asdA16 �araBAD23 �relA198::araC PBAD lacI TT 65�11313 �pabA1516 �pabB232 �asdA16 �araBAD23 �relA198::araC PBAD lacI TT �waaG42 Derivation of �9241�11314 �pabA1516 �pabB232 �asdA16 �araBAD23 �relA198::araC PBAD lacI TT �waaI43 Derivative of �9241�9884 �pabA1516 �pabB232 �asdA16 �araBAD23 �relA198::araC PBAD lacI TT �rfaH49 35�11315 �pabA1516 �pabB232 �asdA16 �araBAD23 �relA198::araC PBAD lacI TT �waaJ44 Derivation of �9241�11316 �pabA1516 �pabB232 �asdA16 �araBAD23 �relA198::araC PBAD lacI TT �wbaP45 Derivation of �9241�11317 �pabA1516 �pabB232 �asdA16 �araBAD23 �relA198::araC PBAD lacI TT �waaL46 Derivation of �9241�9885 �pabA1516 �pabB232 �asdA16 �araBAD23 �relA198::araC PBAD lacI TT �wzy-48 Derivation of �9241

E. coli K-12�7232 endA1 hsdR17 (rK

� mK�) glnV44 thi-1 recA1 gyrA relA1 �(lacZYA-argF)U169 �pir

deoR (80dlac �(lacZ)M15)54

�7213 thi-1 thr-1 leuB6 glnV44 fhuA21 lacY1 recA1 RP4-2-Tc::Mu �pir �asdA4 �zhf-2::Tn10 54�7122 E. coli O78 Lab collection�3550 S. Enteritidis Lab collection

S. pneumoniae WU2 Wild-type virulent, encapsulated type 3 7

a The mutations introduced into �9241 background are indicated in bold.

4228 KONG ET AL. INFECT. IMMUN.


.org/D

ownloaded from

http://iai.asm.org/

Plasmids and mutant strain construction. The primers used in this study arelisted in Table 2. DNA manipulations were carried out as described elsewhere(55). Transformation of Escherichia coli and S. Typhimurium was done by elec-troporation. Transformants were selected on LB agar plates containing appro-priate antibiotics. Selection for Asd� plasmids was done on LB agar plates.

For the waaG42 deletion, two pairs of primers, DrfaG-1F/DrfaG-1R andDrfaG-2F/DrfaG-2R, were used to amplify approximately 350-bp fragments up-stream and downstream of gene rfaG, respectively, from the �3761 genome. Thetwo fragments were then joined by PCR using primers RfaG-1F and RfaG-2R.The terminal A was added at both ends to the resulting PCR product by using theGotaq enzyme (Promega). To facilitate convenient insertion of this PCR prod-uct, we constructed a T-cloning suicide vector, pYA4278, by inserting a blunt-endPtrc green fluorescent protein (GFP) cassette flanked by AhdI sites (66) intoSmaI and XbaI sites of pRE112. The PCR product was ligated to pYA4278 togenerate plasmid pYA4896, which carries a deletion of the entire waaG gene. Asimilar strategy was used to construct plasmids pYA4897 (�waaI43), pYA4898(�waaJ44), pYA4899 (�wbaP45), and pYA4900 (�waaL46) (Tables 1 and 2).The mutations were independently introduced into S. Typhimurium �3761 byallelic exchange using suicide vectors pYA4896, pYA4897, pYA4898, pYA4899,and pYA4900 to generate �11308, �11309, �11310, �11311, and �11312, respec-tively. The mutations were also introduced into S. Typhimurium vaccine strain�9241 to yield strains �11313, �11314, �11315, �11316, and �11317, respectively.�9944 and �9885 were generated by transforming pYA4717 (�wzy-48) into �3761and �9241, respectively. The presence of both pabA1516 and pabB232 mutationsin Salmonella strain �9241 and its derivatives was verified by the inability of thestrains to grow in MOPS minimal medium without p-aminobenzoate. The geno-type of the �asdA16 strain was confirmed by PCR, and the phenotype wasconfirmed by the inability of the strain to grow in media without DAP. The�araBAD23 mutation was verified by PCR and by its white colony phenotype onMacConkey agar supplemented with 1% arabinose. LPS profiles of Salmonellastrains were examined by the method of Hitchcock and Brown (25) using culturesstandardized based on the optical density at 600 nm (OD600).

MIC test. MICs of different antimicrobial substances were determined (64).Twofold serial dilutions of the bile salt deoxycholate (0.3 to 40 mg/ml), polymyxinB (0.075 to 4.7 �g/ml, Sigma), SDS (0.35 to 20 mg/ml, Sigma), and ox bile (1 to20 mg/ml, Sigma) were made across 96-well plates. Bacteria were grown to anOD600 of 0.8 to 0.9 in nutrient broth with or without 0.1% arabinose, harvested,

and washed in phosphate-buffered saline (PBS). Cells were diluted to 1 105 to5 105 CFU in nutrient broth with or without arabinose, and 0.1 ml of thediluted cell suspension was added to each well containing the antimicrobialsubstance. The microtiter plates were incubated overnight at 37°C. The opticaldensity of each culture was determined using a model SpectraMax M2e (Mo-lecular Devices, CA) plate reader. The threshold of inhibition was an OD600 of0.1. Actual titers were determined by spreading culture dilutions onto LB platesfollowed by overnight incubation at 37°C. Assays were repeated three times.

Motility assay. Swimming motility was assessed on LB plates solidified with0.3% agar and supplemented with 0.5% glucose. The plates were allowed to dryat room temperature for 2 h. Freshly grown bacteria were collected from LB agarplates without arabinose, washed, and diluted in saline. Six microliters of bacte-rial suspension (approximately 1 106 CFU) was spotted onto the middle of theplates and incubated at 37°C for 6 h. The diameter of the colonies was measured.Experiments were repeated three times.

Attachment and invasion assays. Attachment and invasion tests were per-formed as described previously (18). The human intestinal cell line INT407 wascultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with10% fetal calf serum, 2 mM L-glutamine, and 100 �g/ml penicillin and strepto-mycin. Cells were seeded 16 h prior to infection at a density of 5 105 cell/well(24-well plates). Bacteria were grown overnight statically in 3 ml LB. One ml ofthe overnight culture was added to 50 ml LB and grown at 37°C to an OD600 of0.95. The culture was diluted to 2 107 CFU/ml in DMEM without antibiotics.The tissue culture medium was replaced by DMEM supplemented with 10%fetal calf serum and 2 mM L-glutamine immediately before 100 �l of the bacterialsuspension was added to each well to achieve a multiplicity of infection (MOI)of 5:1. The plates were centrifuged at 200 g for 5 min to synchronize theinteraction between bacteria and monolayer cells and incubated at 37°C in a CO2

(5%) incubator to allow attachment for 1 h. Wells were washed 3 times with PBS,followed by the addition of 1% Triton X-100 to lyse the INT407 cells. Aliquotsof serial dilutions made from each well were spread onto LB plates and incu-bated overnight to determine the number of attached bacteria. Attachment wascalculated as follows: percent attachment � 100 (number of cell-associatedbacteria/initial number of bacteria added). The same procedure was used toevaluate invasion with the following modification. Prior to the lysis step, 1 ml ofDMEM containing 100 �g/ml gentamicin to kill extracellular bacteria was addedand the monolayer was incubated for an additional 1 h at 37°C. Invasion wascalculated as follows: percent invasion � 100 (number of bacteria resistant togentamicin/initial number of bacteria added). Both adherence and invasion as-says were performed three times.

Serum bactericidal assay. Pooled human serum (PHS) was purchased fromEquitech-Bio and stored at �70°C until used. Heat-inactivated serum (HIS) wasgenerated by incubating PHS for 2 h at 56°C. Bacteria grew statically in LB mediaovernight, and the next day they were grown to an OD600 of 0.9 to 0.95 under37°C at 180 rpm. Bacteria (1 ml) were resuspended in Dulbecco’s phosphate-buffered saline (DPBS) to a final concentration of 2 108 to 3 108 CFU/ml;50-�l portions of the bacterial suspensions were added to 50 �l PHS or HIS andincubated for 30 min at 37°C. After incubation, the samples were placed on iceto stop further bacteriolysis. Serial dilutions of the samples in PBS were platedon LB agar plates and incubated overnight at 37°C. Bacterial colonies werecounted to determine survival.

Determinations of LD50 and colonization in mice. All animal procedures wereapproved by the Arizona State University Animal Care and Use Committee.Seven-week-old, female BALB/c mice were obtained from the Charles RiverLaboratories (Wilmington, MA). Mice were acclimated for 7 days after arrivalbefore starting the experiments. Determination of the oral 50% lethal dose(LD50) followed our standard procedures (35).

To evaluate the effect of the administration route on colonization, mice wereinoculated orally with 20 �l of buffered saline with gelatin (BSG) containing 1 109 CFU or intranasally (i.n.) with 10 �l of BSG containing 1 109 CFU. Sixdays after oral inoculation, or 6 days for mutants and 3 days for wild-type afteri.n. inoculation, three animals per group were euthanized, Peyer’s patches (PP)were collected from all sites of the small intestinal surface, and bacterial colo-nization in PP represented the mean bacterial burden in pooled PP from eachmouse. Spleen and liver samples were collected, and their individual weightswere measured. Each sample was homogenized in a total volume of 1 ml BSG,and appropriate dilutions were plated onto MacConkey agar and/or LB agar todetermine the number of viable bacteria. The remaining tissue homogenate wasenriched for Salmonella by incubation in selenite cysteine broth overnight at37°C. Samples that were negative by direct plating and positive by enrichment inselenite cysteine broth were recorded as 1 CFU/organ.

Construction of vaccine strain derivatives. Each of the mutations under studywas introduced into the previously described attenuated Salmonella strain �9241

TABLE 2. Primers used in this work

Primer name Sequence 5�–3�

DrfaG-1F ..................GGATGCTATTGCACGCGGCTGDrfaG-1R..................TCACCTGCCTGCAGGCTCATAGCAGGCT

GTCCAAADrfaG-2F ..................CTATGAGCCTGCAGGCAGGTGATTTAG

ATGGTTGAGCDrfaG-2R..................TGATGTTCTGGTTAGCGGGTDrfaI-1F ....................GTATGGTTGGCAAAGCGCGCDrfaI-1R ...................TATGGGTTCCTGCAGGAGTGATCACTTT

TGTAATTTCDrfaI-2F ....................ATCACTCCTGCAGGAACCCATAGGTGAT

GTAATGDrfaI-2R ...................TTAGTTTGAGGCAATACCTCDrfaJ-1F....................AGTGAATAAAGCGCGTTTTGDrfaJ-1R ...................GAGGGGAACCTGCAGGTACATCACCTA

TGGGTTTTATDrfaJ-2F....................GATGTACCTGCAGGTTCCCCTCCCTAAA

TCTCCADrfaJ-2R ...................AGCGTAATGTTTTAATTTCCDrfbP1F ....................CAACTGATAAAAGTCAATCCDrfbP1R....................GTAAGCTTACCTGCAGGTTAATCCTCAC

CCTCTGADrfbP2F ....................GATTAACCTGCAGGTAAGCTTACCGAG

AAGTACTGADrfbP2R....................ATACGACGAGGCGTTTCGAGDrfaL-1F...................GTGGGCAACACCGGATTACGGDrfaL-1R ..................GCGTTTTTTTACTTTTCTCCACAATAGGT

TTGGDrfaL-2F...................TGGAGAAAAGTAAAAAAACGCGCTGAT

ACTTATDrfaL-2R ..................TTCTATCATATAGCCGATCTG

VOL. 79, 2011 LPS OUTER CORE MOIETIES AFFECT IMMUNITY AND VIRULENCE 4229


.org/D

ownloaded from

http://iai.asm.org/

(35, 65). Expression plasmid pYA4088, which specifies a Bla-SS-PspA secretedfusion protein, was introduced into each of the �9241 derivatives. The level ofBla-SS-PspA fusion protein produced by each strain was examined by usingWestern blots probed with anti-PspA antibody.

Immunogenicity of vaccine strains in mice. Strains were grown statically over-night in LB broth with 0.1% arabinose at 37°C. Arabinose was added to reduceantigen expression during in vitro growth (61). The following day, 2 ml of theovernight culture was inoculated into 100 ml of LB broth with 0.1% arabinoseand grown with aeration at 37°C to an OD600 of 0.8 to 0.9. Cells were harvestedby room temperature centrifugation at 4,000 rpm for 15 min, and the pellet wasresuspended in 1 ml of BSG. Groups of mice (6 or 7 mice per group) were orallyinoculated with 20 �l of BSG containing 1 109 CFU of each strain, or micewere i.n. or intraperitoneally (i.p.) inoculated with 10 �l or 100 �l of BSGcontaining 1 107 or 1 106 of each strain, respectively, on day 0 and boostedon day 35 with the same dose of the same strain. Blood was obtained every 4weeks by mandibular vein puncture. Blood was allowed to coagulate at 37°C for2 h. Following centrifugation, the serum was removed and stored at �80°C.

Antigen preparation and ELISA. Recombinant PspA (rPspA) was purified asdescribed previously (32). The rPspA clone, which encodes the �-helical regionof PspA (amino acids [aa] 1 to 302) in pET20b, was a kind gift from SusanHollingshead at the University of Alabama at Birmingham. S. Typhimurium LPSwas purchased from Sigma. Outer membrane proteins were prepared as de-scribed previously (32). An enzyme-linked immunosorbent assay (ELISA) wasused to assay serum antibodies against S. Typhimurium LPS, rPspA, and bacte-

rial outer membrane proteins, including those from Salmonella (SOMPs), aspreviously described (38). Color development (absorbance) was recorded at 405nm using a SpectraMax M2e automated ELISA plate reader (Molecular Devices,Menlo Park, CA). A405 readings 0.1 higher than PBS control values were con-sidered positive (serum dilution start at 1:50).

Statistical analyses. Statistical analyses were performed by using the Graph-Pad Prism 5 software package (Graph Software, San Diego, CA). Antibody titerswere expressed as means standard deviations. The means were evaluated withone or two-way analysis of variance (ANOVA) and Bonferroni’s multiple com-parison test for comparisons among groups of antibody titers.

RESULTS

Mutant construction and LPS phenotypes. We constructedan isogenic set of S. Typhimurium strains with nonpolar genedeletions that resulted in progressively shorter LPS, beginningwith a �wzy mutant whose LPS includes a single O-antigenunit, through a �waaG mutant with a core truncated to theheptose residues (Fig. 1A). Each deletion mutation resulted ina strain with the expected length of LPS (Fig. 1B). The wild-type parent strain, UK-1, produced a complete or smooth LPS

FIG. 1. Schematic representation of S. Typhimurium lipopolysaccharide molecule (inner core, outer core, and O-antigen) and LPS phenotypesof mutant and parent strains. (A) Dotted lines indicate the level of LPS truncation resulting from each mutation (Kdo, 3-deoxy-D-manno-octulosonic acid; PPEtN, pyrophosphorylethanolamine; Hep, Heptose; GlcNAc, N-acetylglucosamine; Glc, glucose; Gal, Galactose; P, phosphate).(B) Cells from indicated strains were grown and processed for LPS analysis as described in Materials and Methods. LPS was visualized by silverstaining polyacrylamide gel electrophoresis (PAGE) gels. The mutant strains (from left to right): �11308 (�waaG42), �11309 (�waaI43), �9945(�rfaH49), �11310 (�waaJ44), �11312 (�waaL46), �11311 (�wbaP45), and �9944 (�wzy-48). The expected location of O-antigen regions and coreis indicated on the right of the gel.



.org/D

ownloaded from

http://iai.asm.org/

(lane UK-1 in Fig. 1B). The LPS produced by the �waaG42mutant included only lipid A and the inner core (Fig. 1A) andmigrated the fastest in the gel as expected (Fig. 1B). The LPSproduced by the other mutants, as ordered from left to right inFig. 1B, included more core and/or O-antigen sugars, resultingin the expected reduced mobility in the gel. The �rfaH mutantproduced a truncated core with a molecular mass similar tothat of the �waaI mutant, as previously reported (35, 39).

The �waaL46 and �wbaP45 mutants synthesize completecore with no O-antigen attached. The �waaL46 mutant pro-duces an undecaprenyl-phosphate-O-unit (Und-PP-O-unit)through the action of WbaP (49), but it cannot be added to thelipid A core due to the absence of O-antigen ligase (WaaL) (5,33). Alternatively, the �wbaP mutant synthesizes O-antigenligase (WaaL) but lacks WbaP, the Und-P galactose phospho-transferase that produces an O-antigen ligase substrate by cat-alyzing the transfer of galactosyl-1-phosphate from UDP-ga-lactose to Und-P. The �wzy-48 mutant displayed the expectedsemirough phenotype with the addition of one O-antigen unit(Fig. 1B) (34).

To confirm the observed alterations in LPS profiles, weperformed infection studies with the O-antigen-specific phageP22. Strains grown in LB broth were used as recipients fortransduction assays. As expected, all mutants except the�rfaH49 and wild-type UK-1 strains were resistant to P22 in-fection (Table 3). The �rfaH49 strain was sensitive to P22 andwas transducible; however, the number of transductants ob-tained was much lower than that obtained with the wild-typestrain, as we observed previously (35), indicating that, althoughundetectable on our silver-stained gel (Fig. 1B), the �rfaH49mutant still produced minute amounts of the full-length O-an-tigen.

Phenotypic evaluation of mutant strains. The mutants’growth rates were evaluated in LB broth. All strains weresimilar to the UK-1 parent except strains �11308 (�waaG42)and �11309 (�waaI43), which had a slightly lower rate ofgrowth. Under similar conditions, it took these mutants 5 h toreach an OD600 of 1.3, compared to 4.5 h for the parent strainUK-1 (data not shown). We also noted that these two mutantstended to settle out of suspension more quickly than the otherstrains when held statically for 30 min.

The effect of LPS alterations on the resistance of these

mutants to detergents, such as deoxycholate (DOC) and oxbile, and to the antimicrobial peptide polymyxin B was deter-mined by measuring the MIC in each case (Table 3). Allmutants were more sensitive to DOC and polymyxin B thanwild-type UK-1, and the sensitivity varied based on the degreeof truncation of the outer core. The �waaG42 mutant, lackingthe entire outer core, was the most sensitive to DOC, with anMIC approximately 6-fold lower than that of the wild-typeparent strain. These results are in agreement with our previousreport that a �rfaH mutant is more sensitive to DOC than theparent strain (35). However, the strain with the �waaG42mutation exhibited at least 2-fold greater resistance to poly-myxin B than the �waaI43, �rfaH49 and �waaJ44 mutants,each of which contains more of its outer core. Strains carryingthe �waaL46, �wbaP45, and �wzy-48 mutations contain all thecore sugars and displayed intermediate sensitivity between thesmooth parent and the other mutants (Table 3). All strainswere resistant to the highest concentrations of ox bile tested.

Rough mutants are unable to swarm because of insufficientmoisture on the agar surface but can still swim in 0.3% softagar (60). Therefore, we determined the relative ability of eachmutant to swim on soft agar (Table 3). All mutants were ableto swim to various degrees on soft agar. Three mutants,�waaG42, �wbaP45, and �waaI43, each had substantially re-duced bacterial motility compared to other mutants (Table 3),whereas �rfaH49, �waaJ44, and �waaL46 strains showed2-fold reductions in motility as determined by the motilitydiameter on soft agar. Deletion of wzy-48 had no effect onswimming motility, which was similar to that measured forwild-type UK-1 (34).

Attachment/invasion and serum sensitivity assays. To ex-amine the effect of these LPS gene mutations on attachmentand invasion, we evaluated the ability of the mutants to attachto and invade INT407 human epithelial cells. All strains weresimilar to UK-1 in their ability to attach to INT407 cells,indicating that the outer core and O-antigen are not essentialfor this process (data not shown). In contrast, the �waaI43 and�waaG42 mutations resulted in a significant reduction in in-vasion compared to the other strains, indicating a requirementfor the inner glucose and the galactose residues of core for thisactivity (Fig. 2A). Deletion of wzy resulted in a small (2-fold)but significant increase in invasion compared to that of wild-

TABLE 3. MICs of antimicrobials, swimming motility, transduction efficiency, and virulence ofS. Typhimurium strain �3761 and its isogenic mutants

Strain P22 phagesensitivitya

MIC (DOCb bile)(mg/ml)

MIC (polymyxin B)(�g/ml)

MIC (ox bile)(mg/ml)

Motility (mm) onsoft agarc LD50

d

�11308 (�waaG42) � 1.3 0.15 �20 3.5*** �109

�11309 (�waaI43) � 2.5 0.08 �20 4*** �109

�9945 (�rfaH49) � 2.5 0.08 �20 17** �109

�11310 (�waaJ44) � 2.5 0.08 �20 15** �109

�11312 (�waaL46) � 2.5 0.3 �20 20** �109

�11311 (�wbaP45) � 2.5 0.3 �20 7*** �109

�9944 (�wzy-48) � 2.5 0.3 �20 45 �109e

�3761 �� 10 0.6 �20 45 1 104

a �, resistant; � and ��, degrees of sensitivity as determined by transduction efficiency.b DOC, deoxycholate.c ��, P � 0.01 compared with wild-type UK-1; ��, P � 0.001 compared with wild-type UK-1.d LD50, median lethal dose.e Some mice displayed symptoms (scruffy coat, lethargy) after inoculation with the �wzy-48 strain but recovered after 14 days.



.org/D

ownloaded from

http://iai.asm.org/

type UK-1 (P � 0.05), indicating that shortening the O-antigencontributes to bacterial invasion into epithelial cell. A similarphenomenon has been observed in Shigella, in which shorten-ing the LPS by glucosylation promotes bacterial invasion andevasion of innate immunity by enhancing TTSS function with-out compromising the protective properties of LPS (62).

Complete LPS is an important factor in serum resistance(57). We examined the resistance of our mutants to the bac-tericidal effects of human serum (Fig. 2B). All mutants weresignificantly more sensitive to human serum than the smoothparent strain. In general, the trend of sensitivity to the pooledhuman serum (PHS) followed the length of the LPS, withstrains producing longer LPS being more resistant. These dataare consistent with a previous study showing that a wbaP mu-tant was more sensitive to serum than a wild-type strain (30).However, one notable exception was that the �waaG42 mu-tant, producing the shortest core structure, was as resistant toserum killing as the �waaL mutant, which carries a completecore, and more resistant than mutants with intermediate corestructures. There were no differences in the sensitivity of any ofthe strains to heat-inactivated serum.

Virulence and colonization of mutant strains in mice. Weevaluated our mutants for virulence by determining the oralLD50 of each strain in BALB/c mice. All mutants were highlyattenuated, with no deaths at the highest dose tested, 1 109

CFU (Table 3). To determine how well each strain couldcolonize mouse tissues, groups of mice were orally inoculatedwith 1 109 CFU and the bacterial loads in Peyer’s patches(PP), spleen, and liver were determined 6 days after inocula-tion. The mean CFU counts of the wild-type UK-1 strainrecovered from PP (Fig. 3A), spleen (Fig. 3B), and liver (Fig.3C) were approximately 104, 106, and 106, respectively. Colo-nization levels of all mutants were significantly lower than theparent strain UK-1 (P � 0.05 to 0.001), and the relative dif-ferences observed among mutants were consistent for all or-gans examined. Notably, inactivation of waaG or waaI essen-tially eliminated the ability of S. Typhimurium to causesystemic infection. Both strains colonized PP poorly (�102

CFU/g), and their presence in spleen and liver was below thelimit of detection (Fig. 3B and C). The �rfaH49 mutant colo-nized both spleen and liver better than either the �waaG42 or�waaI43 mutants. Although the �waaJ44, �waaL46, �wbaP45,and �wzy-48 mutants were better able to colonize PP, spleen,and liver, the numbers of mutant bacteria isolated from eachorgan were significantly less than those of their UK-1 parent.Inflammation of the spleen, as determined by spleen weight,was also reduced significantly in mice inoculated with the mu-tants (P � 0.05 to 0.001) (Fig. 3D).

When Salmonella is administered orally, it must overcomethe low-pH barrier of the stomach, followed by the harshenvironment of the intestinal tract, including the presence oforganic acids, bile salts such as DOC, and antimicrobial pep-tides similar to polymyxin B, before interacting with host cells.While we did not test the acid resistance of our mutants,previous reports have indicated that rough strains of Salmo-nella are more sensitive to the presence of organic acids thansmooth strains (3). This, coupled with the observation that ourmutants exhibit increased sensitivity to DOC and polymyxin B,led us to examine alternate routes of immunization to bypassthe barriers imposed by the alimentary tract. Therefore, weevaluated the ability of the �waaI43, �waaL46, and �wzy-48mutants to colonize PP, spleen, and liver after i.n. inoculation.Mice receiving the mutants were evaluated after 6 days, andmice receiving the virulent parent were evaluated after 3 days(Fig. 4). All three mutants achieved levels approximately 10-fold greater for each tissue than the levels reached when ad-ministered orally. The �waaL and �wzy mutants were able tocolonize all tissues at levels comparable to those for UK-1,indicating that O-antigen is not required for colonization whenSalmonella is administered intranasally. There does appear tobe some benefit of core, since the �waaI mutant did not col-onize as well as the other strains.

Effects of the mutations on the immunogenicity of recombi-nant S. Typhimurium strains by oral administration. We werealso interested in determining the effect of core length on theability of an attenuated vaccine to deliver a heterologous an-tigen. To address this, we introduced each of the mutationsdescribed above into attenuated Salmonella strain �9241 (Ta-ble 1). Strain �9241 is attenuated due to the presence of the�pabA �pabB mutations and produces smooth LPS. We intro-duced plasmid pYA4088 encoding pspA, a pneumococcal gene

FIG. 2. INT407 epithelial cell invasion capability and serum resis-tance of S. Typhimurium wild-type UK-1 and its isogenic mutants.(A) Relative invasion of INT407 cells by S. Typhimurium wild-typeUK-1, �11308 (�waaG42), �11309 (�waaI43), �9945 (�rfaH49),�11310 (�waaJ44), �11312 (�waaL46), �11311 (�wbaP45), and �9944(�wzy-48). Values represent invasion of each strain relative to that ofthe wild type. Assays were performed in triplicate on three indepen-dent occasions. Averages standard errors (error bars) are shown.Values that are significantly different from that of the wild type by theone-way ANOVA test and Dunnett post test are indicated by asterisks(*, P � 0.05; **, P � 0.01). (B) Approximately 1 108 CFU/mlbacteria in DPBS (Dulbecco’s phosphate-buffered saline) were incu-bated with either HIS (heat-inactivated serum) or PHS (pooled humanserum) (final serum concentration, 50%) for 30 min at 37°C. Bacterialsurvival was enumerated by plating and counting colonies the followingday. Data from three independent experiments are shown and ex-pressed as means SD. Values that are significantly different fromthat of the wild type by the one-way ANOVA test and Dunnett posttest are indicated by asterisks (*, P � 0.05; **, P � 0.01; ***, P �0.001).



.org/D

ownloaded from

http://iai.asm.org/

encoding the protective antigen PspA, into each strain. Immu-noblot analyses of subcellular fractions from �9241(pYA4088)have shown that 51.5% of the Bla-ss-PspA fusion protein issecreted into the periplasm, 5.4% is secreted out of the cell,

and the remaining 43.1% is retained in the cytoplasm (65).Nine groups of mice were immunized on day 1 to 2 with 1 109 of each �9241 derivative and boosted with the same dose ofthe same strain 5 weeks later. Mice immunized with strain

FIG. 3. Colonization of mice by �3761 and its isogenic mutants 6 days after oral inoculation. Groups of mice (3 mice/group) were orally inoculatedwith approximately 1 109 CFU of the indicated strains. Log10 CFU/g tissue determined on day 6 postinoculation in Peyer’s patches (PP) (A), spleen(B), and liver (C) of inoculated BALB/c mice. (D) Day 6 spleen weights from inoculated mice. The horizontal lines represent the means, and the errorbars represent the standard errors of the means. ***, P � 0.001 compared with UK-1, **, P � 0.01 compared with UK-1; *, P � 0.05 compared withUK-1. Samples that were negative by direct plating and positive by enrichment in selenite cysteine broth were recorded as 1 CFU/organ.

FIG. 4. Colonization of mouse organs by �3761 and its isogenic mutants after intranasal (i.n.) administration with 109 CFU. Log10 CFU/g tissuewere determined on day 6 postinoculation for the mutants and on day 3 postinoculation for the wild-type strain in Peyer’s patches (A), spleen (B),and liver (C) of BALB/c mice. The horizontal lines represent the means, and the error bars represent standard errors of the means.



.org/D

ownloaded from

http://iai.asm.org/

�9241 carrying the empty vector pYA3493 served as negativecontrols. All mice survived immunization, and we did not ob-serve any signs of disease in the immunized mice during theexperimental period. We measured serum IgG antibody re-sponses to rPspA (Fig. 5A) and to Salmonella LPS (Fig. 5B) inimmunized mice. Parent strain �9241(pYA4088) and its iso-genic �wzy-48 counterpart induced the highest levels of IgGagainst PspA and LPS among all the vaccine strains. At 8weeks post-primary inoculation (3 weeks after the boost), the�wzy-48 and �wbaP45 vaccine strains induced anti-PspA IgGtiters similar to those of the parent strain �9241(pYA4088).Anti-PspA IgG titers in sera from mice immunized with theother vaccine strains were significantly lower (P � 0.05 to0.001). Serum IgG against Salmonella LPS induced by each ofthe �9241 derivatives was also significantly lower (P � 0.01)than that of the parent strain �9241(pYA4088).

The immune responses to PspA were further evaluated bymeasuring the levels of anti-PspA IgG isotype subclasses IgG1and IgG2a in each group of mice. At week 4, all strains eliciteda balanced Th1/Th2 response, based on the fact that the miceproduced similar titers of IgG1 and IgG2a (Fig. 5C and D),with the exception of the �wbaP strain, which elicited higher

titers of IgG2a than IgG1 compared to the other mutants (P �0.05), indicating a Th1-biased response. Interestingly, the wbaPstrain elicited significantly higher titers of IgG1 and IgG2athan the waaL strain (Fig. 5C and D), though both mutationsresulted in the same LPS structure (Fig. 1). By week 8, thetiters of IgG2a in mice immunized with �9241(pYA4088) andthe �waaJ, �waaL, �wbaP, and �wzy mutants were higher thanthe IgG1 titers, indicating a shift to a Th1-biased response. Theother mutants were poorly immunogenic, and the IgG1/IgG2ratio was the same at weeks 4 and 8.

Effects of the mutations on the immunogenicity of recombi-nant S. Typhimurium strains by i.n. or i.p. administration.Based on our observation that i.n. inoculation provided betteraccess of the �waaI, �waaL, and �wzy mutants to mouselymphoid tissues (Fig. 4), we wanted to assess the immunoge-nicity of �9241 derivatives when administered i.n. or i.p. Bothare routes that bypass the alimentary tract. LD50 values ofSalmonella vary based on the route of administration used;therefore, different doses were used to immunize the mice viadifferent routes of administration. Four groups of mice wereimmunized i.n. with 1 107 CFU or i.p. with 2 106 CFU ofstrain �9241 or its �waaI43, �waaL46, or �wzy-48 derivatives,

FIG. 5. Serum responses in orally immunized and control mice. Total serum IgG specific for rPspA (recombinant pneumococcal surface proteinA) (A), S. Typhimurium LPS (B), IgG1 specific for rPspA, (C) and IgG2a specific for rPspA (D) were measured by ELISA. Each group has 7 mice.Data represent reciprocal anti-IgG antibody titers in pooled sera from mice orally immunized with attenuated Salmonella carrying either pYA4088(pspA) or pYA3493 (control) at the indicated number of weeks postimmunization. The error bars represent variations between triplicate wells.Mice were boosted at week 5. ***, P � 0.001; **, P � 0.01; *, P � 0.05 compared to titers from mice immunized with �9241(pYA4088). #, the�wbaP45 derivative of �9241 was significantly (P � 0.05) different from the �waaL46 derivative. No immune responses were detected to PspA inmice immunized with �9241 containing the control plasmid (reciprocal titer, �1:50). Immune responses (IgG, IgG1, and IgG2a) against PspA werebelow the level of detection in mice immunized with waaG42 and rfaH49 at 4 weeks (reciprocal titers, �1:50). Immune responses against LPS werebelow the level of detection in mice immunized with the �waaG42, �waaI43, �rfaH49, �waaJ44, and �waaL46 strains at 4 weeks and with the�waaG42 strain at 8 weeks (reciprocal titers, �1:50).



.org/D

ownloaded from

http://iai.asm.org/

all carrying plasmid pYA4088, and boosted with the same doseof the same strain 5 weeks later. Serum IgG responses to PspAand to Salmonella LPS were measured by ELISA (Fig. 6A).The �wzy-48 mutant induced the highest titers of anti-PspAIgG in the i.n. immunized groups, significantly higher thanthose for the parent strain �9241 at week 8. Also at week 8, theanti-PspA titers in mice immunized with the parent strain�9241(pYA4088) and the isogenic �waaL46 mutant were iden-tical.

When administered i.p., the �wzy mutant also inducedhigher anti-PspA serum IgG titers than the other mutantsand titers were similar to those for the parent strain,�9241(pYA4088), by week 8 (Fig. 6B). The anti-PspA titerselicited by the �waaI43 and �waaL46 mutants were signifi-cantly lower than the titers observed in mice immunized withthe parent strain. The anti-LPS IgG titers induced by the threemutants were significantly lower than titers elicited by theparent strain by either immunization route (Fig. 6C and D).

Evaluation of immunogenicity against OMPs from differententeric bacteria. Previous work has shown that a complete lackof or regulated expression of dominant surface antigens, suchas O-antigen, increases the levels of serum antibodies that arecross-reactive to conserved surface epitopes of other entericbacteria (13, 35, 44). Therefore, we determined whether seraraised against our mutants that have truncated core and O-an-tigen had a similar effect. We evaluated the reactivity of pooled

immune sera (n � 7) from orally immunized mice described inthe previous experiment (Fig. 5) against purified OMPs fromhomologous and heterologous wild-type strains by ELISA (Fig.7). The results roughly correlated with the results we obtainedagainst PspA and LPS (Fig. 5). Antibodies raised against theparent vaccine �9241(pYA4088) reacted more strongly toOMPs from three enteric strains tested than those induced byany of its mutant derivatives. In all eight vaccine strains,�waaG and �waaL derivatives consistently induced the lowestreactivity against OMPs, while the �wzy derivative induced thehighest cross-reaction against the OMPs (Fig. 7). Similar re-sults were also achieved from the immunized mice by i.n. or i.p.route administration (Fig. 8), although the �wzy mutant deliv-ered i.p. elicited the highest level of antibodies cross-reactiveto Salmonella enterica serotype Enteritidis (Fig. 8B).

DISCUSSION

Attenuated salmonellae have the ability to stimulate bothcellular and humoral immunity against homologous or hetero-logous antigens after oral administration (9, 10, 12, 14). O-an-tigen and core sugar comprise key virulence determinants (47,57, 60). An incomplete core or O-antigen of LPS may resulteither from deficiency in glycosyltransferase activity, from theinability to synthesize sugars that are part of the core or O-an-tigen chain, or from deletion of LPS processing genes such as

FIG. 6. Serum responses in mice immunized by intranasal (i.n.) or intraperitoneal (i.p.) administration. Reciprocal serum IgG titers in miceimmunized i.n. or i.p. with the indicated Salmonella strains carrying pYA4088. Anti-PspA (pneumococcal surface protein A) responses in miceimmunized i.n. (A) or i.p. (B) and anti-LPS responses in mice immunized i.n. (C) or i.p. (D) were measured by ELISA. There were 6 mice in eachgroup. Mice were boosted at week 5. The error bars represent variations between triplicate wells. Mice were boosted at week 5. ***, P � 0.001;**, P � 0.01; *, P � 0.05 compared to titers from mice immunized with �9241(pYA4088).



.org/D

ownloaded from

http://iai.asm.org/

wzy, rfaH, or wzx (16, 20, 34, 35). The first can be achieved bydeleting either waa or wba genes (31), leading to nonreversibletruncation of LPS, and the second can be obtained by deletinggenes such as galE or pmi, responsible for dNDP-sugar syn-thesis, leading to reversible LPS truncation (11, 13, 29, 37, 58).In this study, we systematically investigated the relationshipbetween S. Typhimurium LPS core/O-antigen length and vir-ulence and immunogenicity.

Deletion of the waa or wba genes targeted in this studyresulted in truncated core structures (Fig. 1A and B) (31). rfaHmutants are reported to produce a mixed core structure con-taining the major Rb3, minor Ra, Rb1, Rb2, and some Rabeing capped by shorter or longer polymers of O-antigen units(39, 56). Our results with the �rfaH49 mutant are consistentwith the expected phenotype, as we observed primarily an Rb3truncated core (similar to results with the �waaI43 mutant)(Fig. 1A and B) in addition to O-antigen detectable only by ourP22 phage transduction assay (Table 3). This is likely to be due

to the fact that RfaH is a transcriptional anti-terminator thatreduces the polarity of long operons by binding to the opssequence located in the 5� regions of the waa, wba, and otheroperons (2). Therefore, the rfaH deletion does not fully abolishsynthesis of full-length transcripts (1, 2). As expected, all mu-tants with truncated outer core and O-antigen were more sen-sitive to DOC, polymyxin B, and pooled human serum thanwild-type UK-1, while all mutants were resistant to ox sourcebile (Table 3 and Fig. 2C). It is surprising that the �waaG42mutant that contains only the inner core was as resistant toPHS as the �waaL46 mutant that produces a complete coreand significantly (P � 0.01) more resistant to serum than the�waaJ44, �rfaH49, and �waaI43 mutants (Fig. 2), each ofwhich contains more sugar moieties attached to the terminalinner core (Fig. 1A). In addition, the waaG46 mutant was2-fold more resistant to polymyxin B than the �waaJ44,�rfaH49, and �waaI43 mutants (Table 3). One explanation forthis result is that the loss of O-antigen in our mutants may

FIG. 7. Cross-reactivity of serum IgG from orally immunized mice against OMPs (outer membrane proteins) from other enteric bacteria. IgGcross-reactivity of sera (1: 1,000 dilution) obtained from mice (n � 5) immunized with �9241(pYA4088) or its mutant derivatives to homologousand heterologous OMPs 8 weeks after the primary immunization by oral administration. Sera from mice immunized with �11313 (�waaG42),�11314 (�waaI43), �9884 (�rfaH49), �11315 (�waaJ44), �11316 (�wbaP45), �11317 (�waaL46), �9885 (�wzy-48), and �9241 (parent vaccinestrain) were individually tested against purified outer membrane proteins from enteric bacteria S. Typhimurium �3761 (A), an avian pathogenicE. coli O78 strain �7122 (B), or S. Enteritidis �3550 (C). The error bars represent variations from triplicate wells. ***, P � 0.001; **, P � 0.01;*, P � 0.05 compared to titers from mice immunized with �9241(pYA4088).

FIG. 8. Cross-reactivity of serum IgG from intranasal (i.n) and intraperitoneal (i.p.) immunized mice against OMPs (outer membrane proteins)from three enteric bacteria. IgG cross-reactivity of sera (1: 1,000 dilution) obtained from mice (n � 5) immunized with �9241(pYA4088) or itsmutant derivatives to homologous and heterologous OMPs 8 weeks after the primary immunization by i.n. route administration (A) and by i.p.route administration (B). Sera from mice immunized with �11314 (�waaI43), �11317 (�waaL46), �9885 (�wzy-48), and �9241 (parent vaccinestrain) were each individually tested against purified outer membrane proteins from S. Typhimurium strain �3761 (Typhimurium), avianpathogenic E. coli O78 strain �7122 (E. coli), and S. Enteritidis strain �3550 (Enteritidis). The error bars represent variations from triplicate wells.***, P � 0.001; **, P � 0.01; *, P � 0.05 compared to titers from mice immunized with �9241(pYA4088).



.org/D

ownloaded from

http://iai.asm.org/

expose and/or activate outer membrane proteases that partiallyrestore serum or polymyxin B resistance. It has been demon-strated that outer membrane proteins such as PgtE, PagC, andRck contribute to serum and/or antimicrobial peptide resis-tance (21, 23, 26, 48). It is known that PgtE, a Salmonella�-barrel surface protease, can function to activate plasminogenin rough but not smooth strains (21, 36) and also to inactivatecomplement components (52). In addition, pgtE is regulatedposttranscriptionally by PhoP (21, 36), a regulator known todirect modifications to LPS structure in vivo (22). It has beenpostulated that pgtE is activated in vivo, where it is known thatrfa gene expression is downregulated (40). Thus, it is possiblethat the activity of PgtE or some other relevant protease isincreased in our rough mutants and that in the �waaG42mutant, these postulated enzymes are more stable or moreactive than they are in some of the other rough mutants. In thisregard, results reported by Nishio et al., who examined theeffect of pagC expression in wild-type Salmonella entericaserovar Choleraesuis and an isogenic waaG mutant, demon-strated that the waaG mutant was more sensitive to serum thanthe wild-type strain (48), presumably due to the loss of O-an-tigen. However, deletion of pagC in the waaG mutant resultedin a further increase in sensitivity to serum (48), indicating thatPagC was active in that mutant.

All mutants with a single mutation were avirulent in mice byoral administration. In addition, the mutants had significantlyreduced abilities to colonize lymphoid organs of mice com-pared to those of the wild-type parent strain �3761 when ad-ministered orally. The ability to colonize was roughly in pro-portion to the length of the LPS; i.e., mutants with longer corestructures were better able to colonize. The �waaG and �waaImutants were unable to reach the spleen and liver when ad-ministered orally (Fig. 3). However, when administered i.n.,the �waaI mutant was able to colonize these organs (Fig. 4),suggesting that increased sensitivity of both �waaG and �waaImutants to bile in the intestinal tract (Table 3) and perhaps lowpH in the stomach led to reduced survival in the intestines.This, coupled with a decreased ability to invade epithelial cells(Fig. 2) (6), could account for their observed inability to crossthe intestinal epithelial barrier and reach the internal lymphoidorgans. In all, the virulence and colonization data indicate thatthe O-antigen and outer core are required for optimal invasionvia oral administration by Salmonella and are essential forresistance to bile, resistance to peptides, and immune evasionduring invasion to the intracellular niche (27). However, whenbypassing the intestinal tract, these mutants retained somecapacity to colonize host tissues (Fig. 4) and the retention ofsome, albeit suboptimal, immunogenicity (Fig. 6).

To evaluate the effect of each of the LPS alterations on theimmunogenicity of an attenuated S. Typhimurium vaccine,each mutation was moved into the vaccine strain �9241 thatcontains attenuating mutations �pabA and �pabB, which donot affect LPS structure. Mice immunized with �9241 deriva-tives expressing the heterologous antigen gene pspA producedanti-PspA serum IgG, although, with the exception of that ofthe �wzy mutant, the titers were significantly lower than thosein mice immunized with parent strain �9241 (Fig. 5A). Withthe exception of the �wzy mutant, the anti-LPS titers weresignificantly reduced for all mutants compared to the parentstrain (Fig. 5B). This is likely to be due to a combination of the

fact that the rough mutants do not produce O-antigen (Fig.1B) and the poor overall immunogenicity of the rough mutantsdue to their decreased colonizing ability. However, the �wbaPmutant elicited detectable anti-LPS titers (Fig. 5B). Althoughwe did not analyze the antibodies beyond what is shown in Fig.5, it is likely that the responses detected were directed towardthe lipid A and core regions of LPS, which were present in theELISA coating antigen. This kind of response could be desir-able if one is interested in producing a vaccine against otherbacteria that share the same or similar core epitopes, such as inother Salmonella and Shigella spp. and pathogenic Escherichiacoli (17, 51, 63).

Removal of the immunodominant O-antigen from the sur-face of Salmonella may result in enhanced immunogenicity ofouter membrane proteins and other surface antigens (13, 44).In a previous study, we described a strain �9241 derivative,�9852, with arabinose-regulated expression of rfaH (35). Thisstrain displayed a reversibly rough phenotype, dependent onarabinose availability. Mice orally immunized with strain�9852(pYA4088) developed serum IgG antibodies with en-hanced cross-reactivity against OMPs from a variety of entericbacteria compared with mice immunized with the smooth par-ent strain, �9241(pYA4088). This result has potential implica-tions for developing a multivalent vaccine to protect againstmultiple pathogens in a single vaccine platform. In the currentstudy, we found that mutations resulting in permanently rough�9241 derivatives did not enhance production of cross-reactiveantibodies to OMPs from any of the enteric bacteria tested(Fig. 7A to C), with the exception of the �wzy strain at week 8,although, while statistically significant, the enhancement wasmuch less than we observed previously with the regulated rfaHstrain �9852 (35), or even with an arabinose-regulated wzystrain (34). This most likely is due to the poor colonization andrapid elimination of the rough vaccine strains by the host,leading to their overall poor immunogenicity when adminis-tered orally.

In summary, we systematically examined the relationshipbetween S. Typhimurium LPS length and virulence, coloniza-tion potential, and immunogenicity. We conclude that mutantswith truncated core and O-antigen colonize mice poorly andthus elicit a suboptimal humoral response against a vectoredantigen. However, the wzy-48 mutant was roughly comparableto the smooth parent strain �9241 in most studies and there-fore may have potential to be used, along with other attenu-ating mutations, in developing RASVs capable of deliveringheterologous or homologous antigens to induce protective im-munity. However, these results, combined with our previousstudies, lead us to recommend the use of reversibly roughstrains as a preferred mode of driving the production of cross-reactive core and/or OMP antibodies, due to the inability ofpermanently rough strains to colonize and elicit high-titer an-tibody responses.

ACKNOWLEDGMENTS

The work was supported by NIH R01 grants AI056289 andAI060557 and the Bill and Melinda Gates Foundation grant 37863.

REFERENCES

1. Artsimovitch, I., and R. Landick. 2002. The transcriptional regulator RfaHstimulates RNA chain synthesis after recruitment to elongation complexesby the exposed nontemplate DNA strand. Cell 109:193–203.



.org/D

ownloaded from

http://iai.asm.org/

2. Bailey, M. J. A., C. Hughes, and V. Koronakis. 1997. RfaH and the opselement, components of a novel system controlling bacterial transcriptionelongation. Mol. Microbiol. 26:845–851.

3. Barua, S., et al. 2002. Involvement of surface polysaccharides in the organicacid resistance of Shiga toxin-producing Escherichia coli O157:H7. Mol.Microbiol. 43:629–640.

4. Bertani, G. 1951. Studies on lysogenesis. I. The mode of phage liberation bylysogenic Escherichia coli. J. Bacteriol. 62:293–300.

5. Billips, B. K., R. E. Yaggie, J. P. Cashy, A. J. Schaeffer, and D. J. Klumpp.2009. A live attenuated vaccine for the treatment of urinary tract infection byuropathogenic Escherichia coli. J. Infect. Dis. 200:263–272.

6. Bravo, D., et al. 2011. Different sugar residues of the lipopolysaccharideouter core are required for early interactions of Salmonella enterica serovarsTyphi and Typhimurium with epithelial cells. Microb. Pathog. 50:70–80.

7. Briles, D. E., et al. 1996. PspA, a protection-eliciting pneumococcal protein:immunogenicity of isolated native PspA in mice. Vaccine 14:858–867.

8. Bueno, S. M., P. A. Gonzalez, and A. M. Kalergis. 2009. Use of geneticallymodified bacteria to modulate adaptive immunity. Curr. Gene Ther. 9:171–184.

9. Cardenas, L., and J. D. Clements. 1992. Oral immunization using live atten-uated Salmonella spp. as carriers of foreign antigens. Clin. Microbiol. Rev.5:328–342.

10. Cheminay, C., and M. Hensel. 2008. Rational design of Salmonella recom-binant vaccines. Int. J. Med. Microbiol. 298:87–98.

11. Collins, L. V., S. Attridge, and J. Hackett. 1991. Mutations at rfc or pmiattenuate Salmonella typhimurium virulence for mice. Infect. Immun. 59:1079–1085.

12. Curtiss, R., III. 2002. Bacterial infectious disease control by vaccine devel-opment. J. Clin. Invest. 110:1061–1066.

13. Curtiss, R., III, et al. 2007. Induction of host immune responses usingSalmonella-vectored vaccines, p. 297–313. In K. A. Brogden, N. Cornick,T. B. Stanton, Q. Zhang, L. K. Nolan, and M. J. Wannemuehler (ed.),Virulence mechanisms of bacterial pathogens, 4th ed. ASM Press, Washing-ton, DC.

14. Curtiss, R., III, et al. 2010. New technologies in using recombinant attenu-ated Salmonella vaccine vectors. Crit. Rev. Immunol. 30:255–270.

15. Edwards, R. A., L. H. Keller, and D. M. Schifferli. 1998. Improved allelicexchange vectors and their use to analyze 987P fimbria gene expression.Gene 207:149–157.

16. Feldman, M. F., et al. 1999. The activity of a putative polyisoprenol-linkedsugar translocase (Wzx) involved in Escherichia coli O antigen assembly isindependent of the chemical structure of the O repeat. J. Biol. Chem.274:35129–35138.

17. Frirdich, E., and C. Whitfield. 2005. Lipopolysaccharide inner core oligo-saccharide structure and outer membrane stability in human pathogens be-longing to the Enterobacteriaceae. J. Endotoxin Res. 11:133–144.

18. Gahring, L. C., F. Heffron, B. B. Finlay, and S. Falkow. 1990. Invasion andreplication of Salmonella typhimurium in animal cells. Infect. Immun. 58:443–448.

19. Germanier, R., and E. Furer. 1971. Immunity in experimental salmonellosis.2. Basis for avirulence and protective capacity of galE mutants of Salmonellatyphimurium. Infect. Immun. 4:663–673.

20. Guard-Petter, J., C. T. Parker, K. Asokan, and R. W. Carlson. 1999. Clinicaland veterinary isolates of Salmonella enterica serovar Enteritidis defective inlipopolysaccharide O-chain polymerization. Appl. Environ. Microbiol. 65:2195–2201.

21. Guina, T., E. C. Yi, H. L. Wang, M. Hackett, and S. I. Miller. 2000. APhoP-regulated outer membrane protease of Salmonella enterica serovarTyphimurium promotes resistance to alpha-helical antimicrobial peptides. J.Bacteriol. 182:4077–4086.

22. Guo, L., et al. 1997. Regulation of lipid A modifications by Salmonellatyphimurium virulence genes phoP-phoQ. Science 276:250–253.

23. Haiko, J., M. Suomalainen, T. Ojala, K. Lahteenmaki, and T. K. Korhonen.2009. Breaking barriers—attack on innate immune defences by omptin sur-face proteases of enterobacterial pathogens. Innate Immun. 15:67–80.

24. Hassan, J. O., and R. Curtiss III. 1990. Control of colonization by virulentSalmonella typhimurium by oral immunization of chickens with avirulent�cya �crp Salmonella typhimurium. Res. Microbiol. 141:839–850.

25. Hitchcock, P. J., and T. M. Brown. 1983. Morphological heterogeneityamong Salmonella lipopolysaccharide chemotypes in silver stained polyacryl-amide gels. J. Bacteriol. 154:269–277.

26. Ho, D. K., H. Jarva, and S. Meri. 2010. Human complement factor H bindsto outer membrane protein Rck of Salmonella. J. Immunol. 185:1763–1769.

27. Holzer, S. U., M. C. Schlumberger, D. Jackel, and M. Hensel. 2009. Effect ofthe O-antigen length of lipopolysaccharide on the functions of type IIIsecretion systems in Salmonella enterica. Infect. Immun. 77:5458–5470.

28. Hone, D., R. Morona, S. Attridge, and J. Hackett. 1987. Construction ofdefined galE mutants of Salmonella for use as vaccines. J. Infect. Dis. 156:167–174.

29. Hone, D. M., et al. 1988. A galE-via (Vi-antigen-negative) mutant of Salmo-nella typhi Ty2 retains virulence in humans. Infect. Immun. 56:1326–1333.

30. Ilg, K., et al. 2009. O-antigen negative Salmonella enterica serovar Typhimu-

rium is attenuated in intestinal colonization but elicits colitis in streptomycin-treated mice. Infect. Immun. 77:2568–2575.

31. Kadam, S. K., A. Rehemtulla, and K. E. Sanderson. 1985. Cloning of rfaG,B, I, and J genes for glycosyltransferase for synthesis of the lipopolysaccha-ride core of Salmonella typhimurium enzymes. J. Bacteriol. 161:277–284.

32. Kang, H. Y., J. Srinivasan, and R. Curtiss III. 2002. Immune responses torecombinant pneumococcal PspA antigen delivered by live attenuated Sal-monella enterica serovar Typhimurium vaccine. Infect. Immun. 70:1739–1749.

33. Kaniuk, N. A., E. Vinogradov, and C. Whitfield. 2004. Investigation of thestructural requirements in the lipopolysaccharide core acceptor for ligationof O antigens in the genus Salmonella—WaaL “ligase” is not the sole de-terminant of acceptor specificity. J. Biol. Chem. 279:36470–36480.

34. Kong, Q., Q. Liu, A. M. Jansen, and R. Curtiss III. 2010. Regulated delayedexpression of rfc enhances the immunogenicity and protective efficacy of aheterologous antigen delivered by live attenuated Salmonella enterica vac-cines. Vaccine 28:6094–6103.

35. Kong, Q., Q. Liu, K. L. Roland, and R. Curtiss III. 2009. Regulated delayedexpression of rfaH in an attenuated Salmonella enterica serovar Typhimu-rium vaccine enhances immunogenicity of outer membrane proteins and aheterologous antigen. Infect. Immun. 77:5572–5582.

36. Kukkonen, M., et al. 2004. Lack of O-antigen is essential for plasminogenactivation by Yersinia pestis and Salmonella enterica. Mol. Microbiol. 51:215–225.

37. Li, Y., et al. 2009. Evaluation of new generation Salmonella enterica serovarTyphimurium vaccines with regulated delayed attenuation to induce immuneresponses against PspA. Proc. Natl. Acad. Sci. U. S. A. 106:593–598.

38. Li, Y., et al. 2008. A sopB deletion mutation enhances the immunogenicityand protective efficacy of a heterologous antigen delivered by live attenuatedSalmonella enterica vaccines. Infect. Immun. 76:5238–5246.

39. Lindberg, A. A., and C. G. Hellerqvist. 1980. Rough mutants of Salmonellatyphimurium: immunochemical and structural analysis of lipopolysaccharidesfrom rfaH mutants. J. Gen. Microbiol. 116:25–32.

40. Maurer, J. J., T. A. Doggett, L. Burns-Keliher, and R. Curtiss III. 2000.Expression of the rfa, LPS biosynthesis promoter in Salmonella typhimuriumduring invasion of intestinal epithelial cells. Curr. Microbiol. 41:172–176.

41. Montminy, S. W., et al. 2006. Virulence factors of Yersinia pestis are over-come by a strong lipopolysaccharide response. Nat. Immunol. 7:1066–1073.

42. Murray, G. L., S. R. Attridge, and R. Morona. 2006. Altering the length ofthe lipopolysaccharide O-antigen has an impact on the interaction of Sal-monella enterica serovar Typhimurium with macrophages and complement.J. Bacteriol. 188:2735–2739.

43. Nagy, G., et al. 2006. Down-regulation of key virulence factors makes theSalmonella enterica serovar Typhimurium rfaH mutant a promising live-attenuated vaccine candidate. Infect. Immun. 74:5914–5925.

44. Nagy, G., et al. 2008. “Gently rough”: the vaccine potential of a Salmonellaenterica regulatory lipopolysaccharide mutant. J. Infect. Dis. 198:1699–1706.

45. Nakayama, K., S. M. Kelly, and R. Curtiss III. 1988. Construction of anAsd� expression-cloning vector: stable maintenance and high level expres-sion of cloned genes in a Salmonella vaccine strain. Nat. Biotechnol. 6:693–697.

46. Neidhardt, F. C., P. L. Bloch, and D. F. Smith. 1974. Culture medium forenterobacteria. J. Bacteriol. 119:736–747.

47. Nevola, J. J., D. C. Laux, and P. S. Cohen. 1987. In vivo colonization of themouse large intestine and in vitro penetration of intestinal mucus by anavirulent smooth strain of Salmonella typhimurium and its lipopolysaccha-ride-deficient mutant. Infect. Immun. 55:2884–2890.

48. Nishio, M., N. Okada, T. Miki, T. Haneda, and H. Danbara. 2005. Identifi-cation of the outer membrane protein PagC required for the serum resis-tance phenotype in Salmonella enterica serovar Choleraesuis. Microbiology151:863–873.

49. Patel, K. B., S. E. Furlong, and M. A. Valvano. 2010. Functional analysis ofthe C-terminal domain of the WbaP protein that mediates initiation of Oantigen synthesis in Salmonella enterica. Glycobiology 20:1389–1401.

50. Poltorak, A., et al. 1998. Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: mutations in Tlr4 gene. Science 282:2085–2088.

51. Raetz, C. R. H., and C. Whitfield. 2002. Lipopolysaccharide endotoxins.Annu. Rev. Biochem. 71:635–700.

52. Ramu, P., et al. 2007. The surface protease PgtE of Salmonella entericaaffects complement activity by proteolytically cleaving C3b, C4b and C5.FEBS Lett. 581:1716–1720.

53. Reeves, P. R., et al. 1996. Bacterial polysaccharide synthesis and gene no-menclature. Trends Microbiol. 4:495–503.

54. Roland, K., R. Curtiss III, and D. Sizemore. 1999. Construction and evalu-ation of a �cya �crp Salmonella typhimurium strain expressing avian patho-genic Escherichia coli O78 LPS as a vaccine to prevent airsacculitis in chick-ens. Avian Dis. 43:429–441.

55. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: alaboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, ColdSpring Harbor, NY.

56. Sanderson, K. E., and B. A. D. Stocker. 1981. Gene rfaH, which affects



.org/D

ownloaded from

http://iai.asm.org/

lipopolysaccharide core structure in Salmonella typhimurium, is required alsofor expression of F-factor functions. J. Bacteriol. 146:535–541.

57. Shaio, M. F., and H. Rowland. 1985. Bactericidal and opsonizing effects ofnormal serum on mutant strains of Salmonella typhimurium. Infect. Immun.49:647–653.

58. Stevenson, G., and P. A. Manning. 1985. Galactose epimeraseless (GalE)mutant G30 of Salmonella typhimurium is a good potential live oral vaccinecarrier for fimbrial antigens. FEMS Microbiol. Lett. 28:317–321.

59. Sun, W., S. Wang, and R. Curtiss III. 2008. Highly efficient method forintroducing successive multiple scarless gene deletions and markerless geneinsertions into the Yersinia pestis chromosome. Appl. Environ. Microbiol.74:4241–4245.

60. Toguchi, A., M. Siano, M. Burkart, and R. M. Harshey. 2000. Genetics ofswarming motility in Salmonella enterica serovar Typhimurium: critical rolefor lipopolysaccharide. J. Bacteriol. 182:6308–6321.

61. Wang, S., et al. 2010. Salmonella vaccine vectors displaying delayed an-

tigen synthesis in vivo to enhance immunogenicity. Infect. Immun. 78:3969–3980.

62. West, N. P., et al. 2005. Optimization of virulence functions through glu-cosylation of Shigella LPS. Science 307:1313–1317.

63. Whitfield, C., N. Kaniuk, and E. Frirdich. 2003. Molecular insights into theassembly and diversity of the outer core oligosaccharide in lipopolysaccha-rides from Escherichia coli and Salmonella. J. Endotoxin Res. 9:244–249.

64. Wiegand, I., K. Hilpert, and R. E. W. Hancock. 2008. Agar and broth dilutionmethods to determine the MIC of antimicrobial substances. Nat. Protoc.3:163–175.

65. Xin, W., et al. 2008. Analysis of type II secretion of recombinant pneu-mococcal PspA and PspC in a Salmonella enterica serovar Typhimuriumvaccine with regulated delayed antigen synthesis. Infect. Immun. 76:3241–3254.

66. Zhang, X., W. Kong, S. Ashraf, and R. Curtiss III. 2009. A one-plasmidsystem to generate influenza virus in cultured chicken cells for potential usein influenza vaccine. J. Virol. 83:9296–9303.

Editor: B. A. McCormick



.org/D

ownloaded from

http://iai.asm.org/

Documents

Effect of Deletion of Genes Involved in Lipopolysaccharide ... · Truncating the LPS O-antigen and ceasing its synthesis in vivo are strategies for attenuating Salmonella for use