characterization of the caulobacter crescentus holdfast ... · but external appendages play...

JOURNAL OF BACTERIOLOGY, Nov. 2008, p. 7219–7231 Vol. 190, No. 210021-9193/08/$08.00�0 doi:10.1128/JB.01003-08Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Characterization of the Caulobacter crescentus Holdfast PolysaccharideBiosynthesis Pathway Reveals Significant Redundancy in the

Initiating Glycosyltransferase and Polymerase Steps�

Evelyn Toh,1 Harry D. Kurtz, Jr.,2 and Yves V. Brun1*Department of Biology, Indiana University, Bloomington, Indiana 47405-3700,1 and Department of Genetics and Biochemistry,

100 Jordan Hall, Clemson University, Clemson, South Carolina 296342

Received 21 July 2008/Accepted 25 August 2008

Caulobacter crescentus cells adhere to surfaces by using an extremely strong polar adhesin called the holdfast. Thepolysaccharide component of the holdfast is comprised in part of oligomers of N-acetylglucosamine. The genesinvolved in the export of the holdfast polysaccharide and the anchoring of the holdfast to the cell were previouslydiscovered. In this study, we identified a cluster of polysaccharide biosynthesis genes (hfsEFGH) directly adjacentto the holdfast polysaccharide export genes. Sequence analysis indicated that these genes are involved in thebiosynthesis of the minimum repeat unit of the holdfast polysaccharide. HfsE is predicted to be a UDP-sugarlipid-carrier transferase, the glycosyltransferase that catalyzes the first step in polysaccharide biosynthesis. HfsF ispredicted to be a flippase, HfsG is a glycosyltransferase, and HfsH is similar to a polysaccharide (chitin) deacety-lase. In-frame hfsG and hfsH deletion mutants resulted in severe deficiencies both in surface adhesion and inbinding to the holdfast-specific lectin wheat germ agglutinin. In contrast, hfsE and hfsF mutants exhibited nearlywild-type levels of adhesion and holdfast synthesis. We identified three paralogs to hfsE, two of which are redundantto hfsE for holdfast synthesis. We also identified a redundant paralog to the hfsC gene, encoding the putativepolysaccharide polymerase, and present evidence that the hfsE and hfsC paralogs, together with the hfs genes, areabsolutely required for proper holdfast synthesis.

Bacterial adhesion plays a crucial role in the establishmentof a community of microorganisms called a biofilm on hosttissues and nonbiological surfaces. The strategies employed bydifferent bacteria for mediating attachment to surfaces vary,but external appendages play important roles in adherence andcolonization. In enteropathogenic Escherichia coli and entero-hemorrhagic E. coli, the flagella can bind mucins on mucosalsurfaces (13), while enterohemorrhagic E. coli produce ad-hesive type IV pili that form the physical bridges betweenbacteria adhering to human and bovine host cells (69). Poly-saccharides also play important roles in adherence by bothgram-positive and gram-negative bacteria. For example, E. coliproduces an exopolysaccharide matrix composed of polymersof N-acetylglucosamine (GlcNAc) that influences the struc-tural integrity of its biofilm, particularly the transition fromtemporary polar attachment of cells to permanent lateral at-tachment of cells to a substrate (1). Staphylococcus epidermidisproduces a polysaccharide adhesin composed of a homopoly-mer of poly-�-1,6,GlcNAc (37) that is required for biofilmformation (65). A GlcNAc polysaccharide helps promote thetransmission of Yersinia pestis from the flea to the mammalianhost (21) and impacts bacterial colonization, virulence, andhost immune evasion (56, 63).

Surface adhesion and biofilm formation by the gram-nega-tive aquatic bacterium Caulobacter crescentus require thecontribution of three polar structures: the flagellum, pili, andholdfast (6, 12, 34), whose biosynthesis is regulated in a cell

cycle-dependent manner. C. crescentus produces two cell typesat every cell division, a motile swarmer cell with a flagellumand pili at the same pole and a stalked cell with a cell envelopeextension called the stalk tipped by the holdfast adhesin. Theswarmer cell differentiates into a stalked cell by shedding theflagellum, retracting the pili, and synthesizing a holdfast and astalk at the same pole (47). Although the C. crescentus pili andflagellum are important to optimize adhesion, the strong per-manent attachment of C. crescentus to a substrate requires theholdfast. The holdfast is an elastic material with gel-like prop-erties (35) and is strongly adhesive (61). Although the precisecomposition and structure of the holdfast material have notbeen determined, N-acetylglucosamine (GlcNAc) polymersare an important component of the holdfast and play a role inthe elastic properties (35) and the force of adhesion of theholdfast (61). Previous studies of C. crescentus holdfasts indi-cated that the holdfast is sensitive to lysozyme and chitinase,glycolytic enzymes specific for cleavage of �-1,4 linkages inoligomers of GlcNAc, suggesting that the holdfasts might con-tain oligomers of �-1,4-linked GlcNAc (40). Lectin bindingstudies also revealed that wheat germ agglutinin (WGA),which recognizes GlcNAc polymers, binds specifically to theholdfast (40).

Adhesion-deficient mutants from a C. crescentus transposonlibrary were previously grouped into three classes based onphenotypic characteristics, namely the holdfast biogenesis (hfsfor holdfast synthesis) and holdfast anchor (hfa for holdfastanchor) classes and a class of some pleiotropic developmentalmutants (podJ and pleC) (42, 46, 28, 29, 58). Mutations in thehfaABD operon result in various defects in surface adhesion,but most notably the holdfasts of these mutants have a reducedability to attach to the tip of the stalk, causing the shedding of

* Corresponding author. Mailing address: Department of Biology,Indiana University, Bloomington, Indiana 47405-3700. Phone: (812)855-8860. Fax: (812) 855-6705. E-mail: [email protected].

� Published ahead of print on 29 August 2008.

7219

on Decem

ber 10, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

https://crossmark.crossref.org/dialog/?doi=10.1128/JB.01003-08&domain=pdf&date_stamp=2008-11-01

http://jb.asm.org/

a fully adhesive holdfast polysaccharide (42, 46, 28, 29, 10).Insertional mutations in hfsDAB result in the complete abol-ishment of surface adhesion and holdfast production (58), in-dicating that these hfs genes are absolutely required for hold-fast synthesis. Sequence and phenotypic analyses indicate thathfsDAB are required for holdfast polysaccharide export; how-ever, the genes directly involved in holdfast polysaccharidebiosynthesis were not identified in previous genetic screens.

Here, we characterized a class of adhesion-deficient mutantsobtained in a previous transposon mutagenesis screen (42) andidentified a cluster of four genes directly adjacent to hfsDABCthat are involved in holdfast polysaccharide biosynthesis(hfsEFGH). We generated in-frame, nonpolar deletion mu-tants to assess the contribution of these hfsEFGH genes to C.crescentus surface adhesion and holdfast synthesis. We showedthat hfsE has two paralogs that can substitute for hfsE inholdfast synthesis. Single-deletion hfsE, pssY, and pssZ mutantscaused only a slight reduction in adhesion and holdfast synthe-sis, whereas a triple-deletion mutant was completely deficientin adhesion and holdfast synthesis. We identified a paralog tohfsC, which we call hfsI. An in-frame-deletion mutant with adeletion of the predicted polysaccharide polymerase gene hfsChad no surface adhesion phenotype (58). Just like an hfsCmutant, an hfsI deletion mutant does not display a holdfastsynthesis defect. Deletion of both hfsC and hfsI results in asevere holdfast synthesis defect. We hypothesized that theparalogs of hfsE may serve to add various sugars in the holdfastpolysaccharide repeat unit, while the paralog of hfsC may cat-alyze the formation of a different glycosidic linkage betweenthese different repeat units.

MATERIALS AND METHODS

Bacterial strains and growth conditions. All bacterial strains and plasmidsused in this study are listed in Table 1. All C. crescentus strains were cultured inpeptone-yeast-extract (PYE) medium (48) at 30°C. Kanamycin (20 �g/ml [plate];5 �g/ml [broth]), tetracycline (2 �g/ml [plate]; 1 �g/ml [broth]), nalidixic acid (20�g/ml [plate]), and 3% sucrose (plate) were used to supplement the C. crescentusmedia as necessary. E. coli strains were cultured at 37°C in Luria-Bertani (LB)medium. The LB medium was supplemented with kanamycin (50 �g/ml or 25�g/ml [plate]; 30 �g/ml [broth]) and tetracycline (12 �g/ml [plate and broth])when necessary.

DNA manipulations and sequencing. All restriction enzymes used for standardmolecular cloning in this study were purchased from New England Biolabs, Inc.(Ipswich, MA), and we followed standard molecular biology methods (3). Allprimers used in this study are listed in Table 2. Plasmid DNA was isolated byusing a QIAprep miniprep kit, and PCR products were purified by usingQIAquick spin columns (Qiagen, Valencia, CA) according to the proceduresrecommended by the manufacturer. Chromosomal DNA was isolated by using aPromega Magic miniprep DNA purification system (Promega, Madison, WI)according to the manufacturer’s instructions. Sequencing reactions were per-formed in the Indiana Institute for Molecular Biology at Indiana University onan Applied Biosystems 3730 automated fluorescence sequencing system, usingABI Prism BigDye Terminator cycle sequencing version 3.1 (Applied Biosys-tems, Foster City, CA). Sequence data were analyzed using Sequencher 4.7software (Gene Codes Corporation, Ann Arbor, MI) and the Codon Preferencemodule of the GCG Wisconsin package.

Analysis of hfs promoter by lacZ transcriptional fusions. Fragments werecloned upstream of the lacZ gene in pRKlac290 and analyzed for promoteractivity in the wild-type C. crescentus strain CB15. The levels of promoter activitywere determined by assaying for �-galactosidase activity as previously described(26), with the following modification: all measurements were done at 22°C. Thelevels of �-Galactosidase activity were expressed in Miller units and representthe averages of the results of five independent cultures done in triplicate, dis-played with standard deviations. �-Galactosidase activity conferred by the plas-mid alone was approximately 70 Miller units.

Generation of in-frame-deletion mutants. In-frame markerless deletions ofhfsE, hfsF, hfsG, hfsH, and hfsI were created by a two-step homologous recom-bination method, using upstream and downstream fragments of each gene clonedinto nonreplicating plasmid pNPTS138, which carries a kanamycin resistancegene cassette (nptI), along with the sacB cassette that confers sucrose sensitivity,as previously described (19, 60). Primers NewFuphfsE, NewRuphfsE, Fdw-HindIIIhfsE, and RdwPstIhfsE were used in a PCR to amplify 500 bp directlyupstream and downstream of the predicted start of the hfsE gene, leaving sixcodons at the start of the gene and seven codons at the end of the gene. PrimersNewFuphfsF, NewRuphfsF, FdwHindIIIhfsF, and RdwSphIhfsF were used inPCR to amplify 500 bp directly upstream and downstream of the start of the hfsFgene, leaving 34 codons at the start of the gene and 7 codons at the end of thegene. Primers FupSphIhfsG, RupHindIIIhfsG, FdwHindIIIhfsG, and RdwBam-HIhfsG were used in PCR to amplify 500 bp directly upstream and downstreamof the predicted start of the hfsG gene, leaving 6 codons at the start of the geneand 10 codons at the end of the gene. Primers FupBamHIhfsH, RupEcoRIhfsH,FdwEcoRIhfsH, and RdwHindIIIhfsH were used in PCR to amplify 500 bpdirectly upstream and downstream of the predicted start of the hfsH gene,leaving three codons at the start of the gene and four codons at the end of thegene. Primers F499HindIII, R499XbaI, Fm499XbaI, and Rm499EcoRI wereused in PCR to amplify 500 bp directly upstream and downstream of the pre-dicted start of the hfsI gene, leaving three codons at the start of the gene and fourcodons at the end of the gene. The deletion mutants were confirmed by colonyPCR, using the primers used to clone the upstream and downstream fragments,and were verified by sequencing.

Complementation of in-frame-deletion mutants. A series of pMR10-basedplasmids was constructed to complement all the deletion mutants. The individualgenes were PCR amplified separately from wild-type CB15 chromosomal DNAand introduced into pMR10, a medium-copy-number plasmid, together with itsnative promoter. Primers FupEcoRIhfsE and phfsEendNdeI were used to am-plify the hfsE promoter. Primer pairs FupEcoRIhfsE and HfsEendPstI,HfsFstartNdeI and HfsFendHindIII, HfsGstartNdeI and HfsGendBamHI, andHfsHstartNdeI and HfsHendBamHI were used to amplify the entire hfsE, hfsF,hfsG, and hfsH genes, respectively. The appropriate plasmids carrying the dif-ferent hfs genes under the control of the hfsE promoter were introduced to theindividual in-frame-deletion mutants and assayed for the restoration of surfaceadhesion and lectin binding.

Surface binding assay. Polystyrene binding assays were performed as previ-ously described (6), with the following modifications: the cells were allowed toadhere for 45 min, and 1.5 ml of a 1% crystal violet solution was used for staining.

Swarming motility assay. Exponential cultures (0.5 �l) grown in PYE mediumand normalized to an optical density of 0.3 at 600 nm were stabbed into 0.3%agar semisolid agar PYE medium. The plates were incubated at room temper-ature for 5 days. A nonmotile �pleC mutant (31) was used as a negative controlfor swarming.

Phage sensitivity assay. Phage sensitivity assays were performed as previouslydescribed (53), using the caulophage �CbK.

Microscopy. Fluorescent lectin binding assays were performed as previouslydescribed (24), except that Alexa Fluor 488-conjugated WGA (AF488-WGA)was used to label the holdfast (Invitrogen Molecular Probes). A Nikon EclipseE800 light microscope equipped with a 100� Plan Apo oil objective was used forphase-contrast microscopy, and a Nikon FITC-HyQ filter cube (Chroma Tech-nology) was used for epifluorescence microscopy. Images were captured by usinga Princeton Instruments cooled charge-coupled-device camera, model 1317, andMetaMorph imaging software, version 7.1.1 (Molecular Devices, Sunnyvale,CA). Transmission electron microscopy (TEM) was used to observe the appear-ance of the holdfast in the various mutants. Exponential-phase cells weremounted onto Formvar-coated, carbon film-stabilized copper grids (ElectronMicroscopy Sciences, Hatfield, PA) for 30 min. Each grid was washed in a dropof water, negatively stained with 7.5% uranyl magnesium acetate for 5 min, andwashed five times with water after being stained with uranyl magnesium acetate.The grids were examined with a Jeol JEM-1010 transmission electron micro-scope set to 80 kV.

RESULTS

Identification of a holdfast polysaccharide biosynthesis genecluster adjacent to the hfsDABC genes. To identify the genesinvolved in the biosynthesis of the minimum-repeat units of theholdfast polysaccharide, we revisited the Tn5 mutagenesisscreen for holdfast mutants of the freshwater C. crescentus

7220 TOH ET AL. J. BACTERIOL.

on Decem


.org/D

ownloaded from

http://jb.asm.org/

strain CB2A and examined an uncharacterized gene cluster(cluster IV in the previous study of Mitchell and Smit [42]). Allfour Tn5 insertion mutants in cluster IV, with the originaldesignations g3, g10, g4, and g6, were found to be adhesiondeficient, based on a quantitative surface binding assay.CB2Ag3, CB2Ag10, CB2Ag4, and CB2Ag6 exhibited 9%, 8%,

9%, and 7% binding to a polystyrene surface, respectively,compared to 100% for wild-type C. crescentus CB2A.

A fragment of Tn5 with a flanking chromosomal DNA frommutants g3, g10, g4, and g6 was cloned into pUC18, and theDNA sequence of the chromosomal DNA portion was used tosearch the genome sequence of C. crescentus CB15 (45), since

TABLE 1. Bacterial strains and plasmids used in this study

Strain or plasmid Description or construction Source orreference

StrainsE. coli

S17-1 E. coli 294::RP4-2(Tc::Mu)(KM::Tn7) 54DH5�F’ �80dlacZ�M15 �(lacZYA-argF)U169 endA1 recA1 hsdR17 (r m�) deoR thi-1 supE44 gyrA96 relA1 36Alpha-select deoR endA1 recA1 relA1 gyrA96 hsdR17(rK

mK�) supE44 thi-1 �(lacZYA-argFV169) �80�lacZ�M15

FBioline

C. crescentusCB15 Wild-type 48NA1000 syn-1000, previously called CB15N; a derivative of CB15 that does not synthesize holdfasts 14CB2A rsaA (S-layer) mutant of CB2 57CB2Ag3 Tn5 insertional mutant in cluster IV 42CB2Ag4 Tn5 insertional mutant in cluster IV 42CB2Ag6 Tn5 insertional mutant in cluster IV 42CB2Ag10 Tn5 insertional mutant in cluster IV 42YB4981 CB15 �hfsE This studyYB4982 CB15 �hfsF This studyYB2197 CB15 �hfsG This studyYB2198 CB15 �hfsH This studyYB3978 CB15 �hfsI This studyYB3979 CB15 �hfsC �hfsI This studyYB2840 CB15 �hfsC 58YB3779 CB15 �pleC 31YB4143 CB15 �pssY This studyYB4144 CB15 �pssZ This studyYB4976 CB15 �CC_1486 This studyYB5000 CB15 �pssY �pssZ �hfsE �CC_1486 This studyYB4997 CB15 �pssY �pssZ �hfsE This studyYB5001 CB15 �pssY �pssZ �CC_1486 This studyYB5701 CB15 �pssY �hfsE �CC_1486 This studyYB5702 CB15 �pssZ �hfsE �CC_1486 This studyYB4409 CB15 �pssY �pssZ This studyYB5696 CB15 �pssY �hfsE This studyYB5697 CB15 �pssZ �hfsE This studyYB5698 CB15 �pssY �CC_1486 This studyYB5699 CB15 �pssZ �CC_1486 This studyYB5700 CB15 �hfsE �CC_1486 This study

PlasmidspNPTS138 pLitmus 38 derivative with nptI, sacB, and RK2 oriT sequences and deleted bla gene M. R. K. Alley,

unpublished datapNPTS138�hfsE pNPTS138 parent vector containing 500-bp fragments upstream and downstream of hfsE This studypNPTS138�hfsF pNPTS138 parent vector containing 500-bp fragments upstream and downstream of hfsF This studypNPTS138�hfsG pNPTS138 parent vector containing 500-bp fragments upstream and downstream of hfsG This studypNPTS138�hfsH pNPTS138 parent vector containing 500-bp fragments upstream and downstream of hfsH This studypNPTS138�pssY pNPTS138 parent vector containing 500-bp fragments upstream and downstream of pssY (CC_0166) This studypNPTS138�pssZ pNPTS138 parent vector containing 500-bp fragments upstream and downstream of pssZ (CC_2384) This studypNPTS138�CC_1486 pNPTS138 parent vector containing 500-bp fragments upstream and downstream of CC_1486 This studypRKlac290 lacZ transcriptional fusion vector, Tetr, IncP-1 replicon, Mob 18plachfs3 1,663-bp XbaI-HindIII fragment containing an overlapping region of hfsE through the middle of hfsF

that was cloned into the XbaI-HindIII sites of pRKlac290This study

plachfs5 1,449-bp PstI-HindIII fragment containing an overlapping region of hfsG through the middle of hfsHthat was cloned into the PstI-HindIII sites of pRKlac290

This study

plachfs9 499-bp EcoRI-HindIII fragment containing a region upstream of hfsF that was cloned into the EcoRI-HindIII sites of pRKlac290

This study

plachfs12 539-bp EcoRI-HindIII fragment containing a region upstream of hfsE that was cloned into the EcoRI-HindIII sites of pRKlac290

This study

plachfs13 218-bp EcoRI-HindIII fragment containing an internal region of hfsF that was cloned into the EcoRI-HindIII sites of pRKlac290

This study

plachfs14 1,467-bp EcoRI-HindIII fragment containing an internal region of hfsE that was cloned into theEcoRI-HindIII sites of pRKlac290

This study

plachfs15 717-bp EcoRI-HindIII fragment containing an internal region of hfsE that was cloned into the EcoRI-HindIII sites of pRKlac290

This study

pMR10 Mini-RK2 cloning vector; carries RK2 replication and stabilization functions R. Roberts and C.Mohr

pMR10::Phfs-hfsE Complementation vector that carries the native hfsE promoter and the hfsE gene This studypMR10::Phfs-hfsF Complementation vector that carries the native hfsE promoter and the hfsF gene This studypMR10::Phfs-hfsG Complementation vector that carries the native hfsE promoter and the hfsG gene This studypMR10::Phfs-hfsH Complementation vector that carries the native hfsE promoter and the hfsH gene This studypMR10::PhfsA-hfsC Complementation vector that carries the hfsA promoter and the hfsC gene This study

VOL. 190, 2008 CAULOBACTER HOLDFAST SYNTHESIS PATHWAY 7221

on Decem


.org/D

ownloaded from

http://jb.asm.org/

the genome sequence of C. crescentus CB2A is not known.Sequence analysis of the junction of the Tn5 fragment revealedthat the g3, g10, g4, and g6 transposon insertion sites werefound in a region containing four open reading frames (ORFs)downstream from and convergent to the hfsC gene (Fig. 1A).Sequence analysis of these four ORFs revealed that thesegenes (CC_2425 to CC_2428), hereinafter called hfsEFGH, arehomologous to genes involved in polysaccharide biosynthesis(68). hfsE (CC_2425) encodes a predicted protein of 375amino acids (aa) that is homologous to a family of integralmembrane sugar transferases from the polyisoprenylphosphatehexose-1-phosphate transferase (PHPT) family (32, 64), in-cluding specialized glycosyltransferases like WsaP in Geobacil-lus stearothermophilus NRS 2004/3a (59), WbaP in E. coli (11),and WbaP in Salmonella enterica (64). These sugar transferasescatalyze the first step in polysaccharide biosynthesis by trans-ferring hexose-1-phosphate residues from UDP-hexoses to thelipid carrier molecule undecaprenol phosphate in the innermembrane (4, 8). HfsE possesses the three characteristic blocksof highly conserved amino acids (KFRSM, DELPQ, and PGITG)that are typically found in the C-terminal halves of prototypes ofthis family of PHPT sugar transferases (Fig. 1B). hfsF (CC_2426)encodes a predicted integral membrane protein of 480 aa with 14predicted transmembrane helices and a weakly conserved stretchof approximately 208 amino acids (polysacc_synt domain, ProteinFamilies database of alignments; hidden Markov models[HMMs], http://cmr.tigr.org/cgi-bin/CMR/HmmReport.cgi?hmm_acc�PF01943) that is often present in Wzx flippases (formerlyknown as RfbX) (38). Some of the Wzx proteins have been shownto catalyze the translocation of undecaprenol diphosphate-linkedK-repeating units formed at the cytoplasmic side of the inner

membrane across this membrane (8, 51, 15). hfsG (CC_2427)encodes a predicted cytoplasmic protein of 309 aa homologous tofamily 2 glycosyltransferases that participate in a wide range ofpolysaccharide synthesis systems (5, 27, 33, 43, 52), includingtransferring sugar units from UDP-glucose to cellulose (56) andUDP-GlcNAc to oligosaccharide Nod factors, (17), and isthought to catalyze the polymerization of GlcNAc. hfsH(CC_2428) encodes a predicted cytoplasmic protein of 257 aa thatbelongs to carbohydrate esterase family 4 (CE4). All members ofthis family catalyze the hydrolysis of either N-linked acetyl groupsfrom GlcNAc residues or O-linked acetyl groups from O-acetylxy-lose residues of their substrates (9). HfsH is similar to a NodBde-N-acetylase superfamily of polysaccharide deacetylases in-volved in the degradation and remodeling of GlcNAc substrateslike chitin (20, 41), chito-oligosaccharide rhizobial Nod factors,and peptidoglycan (49). We hypothesize that hfsE, hfsF, hfsG, andhfsH catalyze the biosynthesis, remodeling, and flipping across theinner membrane of the holdfast polysaccharide repeat unit(s).

Transcriptional and translational organization of the hfsEFGHgenes. The organization of the hfsEFGH genes suggests thatthese genes might be transcribed as an operon. The ORFs ofhfsF, hfsG, and hfsH overlap, with hfsF and hfsG sharing fournucleotides, whereas hfsG and hfsH share a single nucleotide;however, a 6-nucleotide space separates the end of hfsE andthe start of hfsF. Two results of our initial deletion experi-ments, however, suggested that careful examination of thepromoter organization in this region was required prior tophenotypic characterization of the deletion mutants. First, wewere unable to complement an initial in-frame hfsE deletionmutant (data not shown), and second, the initial deletion of thehfsF gene from codon 5 to codon 475 suggested that a pro-

TABLE 2. Oligonucleotides used in this study

Oligonucleotide Sequence (5 –3 ) Application

NewFuphfsE TCGCTTCGCGAATTCAACGGCGGATGTCGG hfsE in-frame deletion constructionNewRuphfsE CCCGCCGGTAAGCTTGCTGGCCCCGAT hfsE in-frame deletion constructionFdwHindIIIhfsE TGCCTGTTGAAGCTTCGTTGGGAA hfsE in-frame deletion constructionRdwPstIhfsE CTGGACCATCTGCAGCATAGCCGC hfsE in-frame deletion constructionNewFuphfsF GGGCGAGAATTCGATCGTCGGGCCCCG hfsF in-frame deletion constructionNewRuphfsF CCGACCGTAAAGCTTCCGCGTATGTTTGAT hfsF in-frame deletion constructionFdwHindIIIhfsF CTCTCCCTGAAGCTTCGTCGAAAG hfsF in-frame deletion constructionRdwSphIhfsF CGTGCATGCGCTCAACAGGGC hfsF in-frame deletion constructionFupSphIhfsG CTGCCCGCATGCGTCGGCCTC hfsG in-frame deletion constructionRupHindIIIhfsG CTCGAGTCGAAGCTTGTTGACGGG hfsG in-frame deletion constructionFdwHindIIIhfsG GCCATGAAGCTTTATCGCGCG hfsG in-frame deletion constructionRdwBamHIhfsG GTTGAGATCGGATCCGTAGGTGAT hfsG in-frame deletion constructionFupBamHIhfsH ACCCTGGACGGATCCGTCGGTGAT hfsH in-frame deletion constructionRupEcoRIhfsH GACCTTCTCGAATTCCATCGGCAT hfsH in-frame deletion constructionFdwEcoRIhfsH GAGGGCTCGCGGGAATTCGGGCTC hfsH in-frame deletion constructionRdwHindIIIhfsH GATCGCGCCAAGCTTTTCACGCTGTTC hfsH in-frame deletion constructionF499HindIII GACGAAGGTAAGCTTGTCCCAGAAGCC hfsI in-frame deletion constructionR499XbaI GCGATAGGGTCTAGACATGCCGAA hfsI in-frame deletion constructionFm499XbaI CGTCTTCTTTCTAGAGCCACAAACATTGGG hfsI in-frame deletion constructionRm499EcoRI CCTTCGGGATCCAGGCGCATTTCC hfsI in-frame deletion constructionFupEcoRIhfsE CCCAGTAGCGAATTCCCCAAGAGC Amplification of hfsE promoterphfsEendNdeI TGGGAAAACCATATGCAGCCTAGCGAT Amplification of hfsE promoterHfsEendPstI CCGCGCCAGCTGCAGCGCGGACTA Complementation construct of hfsEHfsFstartNdeI GCTAGGCCGCATATGTTCTGGCGC Complementation construct of hfsFHfsFendHindIII GCTCGTTGAAAGCTTCGTTCATGCGGA Complementation construct of hfsFHfsGstartNdeI GCAAGCCATATGAACGCGCCC Complementation construct of hfsGHfsGendBamHI CCATCGGCAGGATCCGGCCTCGCT Complementation construct of hfsGHfsHstartNdeI CGAGGACGTCATATGCCGATGGAA Complementation construct of hfsHHfsHendBamHI TCAGGACGAGGATCCGAGCCCGAT Complementation construct of hfsH


on Decem


.org/D

ownloaded from

http://jb.asm.org/

moter was present in its coding sequence (data not shown).Examination of the hfsE annotation at The Institute forGenomic Research Comprehensive Microbial Resource usedto design the gene deletion strongly suggested that the genewas misannotated. The annotation at the Comprehensive Mi-crobial Resource indicated that the gene spanned nucleotidecoordinates 2630635 to 2632173 in the C. crescentus genome,potentially encoding a protein of 512 aa. Amino acids 253 to460 had good sequence similarity to the PHPT family members(Fig. 1B), whereas the first 180 aa had no similarity to otherproteins. In addition, the third position GC bias and codon bias

analyses suggested that the first 137 codons were unlikely toencode a protein. Based on the GC bias analysis and the regionof similarity to PHPT family proteins, we hypothesized that thestart codon of hfsE is an ATG starting at nucleotide coordinate2631047; this coordinate was used in all our subsequent anal-yses. To determine the approximate location of the promot-er(s) responsible for transcribing this gene cluster in order todesign nonpolar deletions, we constructed a set of overlappingtranscriptional fusions to a promoterless lacZ reporter gene inthe plasmid pRKlac290 and assayed for �-galactosidase activ-ity as a reporter of promoter activity (Fig. 2).

FIG. 1. (A) Genetic organization of the hfsE, pssY, pssZ, and CC_1486 gene clusters. Each gene name and the gene’s predicted functions andcorresponding TIGR locus tags are listed below the arrows. GT stands for glycosyltransferase. Arrows indicate the direction of transcription foreach gene relative to another. The dashed lines depict regions of the gene deleted in the various deletion mutants. (B) Alignment of the C-terminalregion of HfsE and its paralogs, along with other PHPT family members. Highly conserved residues are shaded, while the asterisks indicate aminoacids conserved across all paralogs. The conserved blocks I, II, and III are indicated by lines above them and are conserved across all PHPThomologs. The sugar transferases compared are as follows: C. crescentus CB15 HfsE (accession number AAK24396), C. crescentus PssY (accessionnumber AAK22153), C. crescentus PssZ (accession number AAK24355), C. crescentus CC_1486 (accession number AAK23465), Burkholderiacenocepacia IST432 BceB (accession number ABC71344), E. coli WcaJ (accession number AP002647), Methylobacillus sp. strain 12S EpsB(accession number BAC41337), Xanthomonas campestris GumD (accession number AAA86372), Salmonella enterica serovar Typhimurium WbaP(formerly RfbP; accession number P26406), Erwinia amylovora AmsG (accession number Q46628), Sinorhizobium meliloti ExoY (accession numberQ02731), Streptococcus pneumoniae serotype 8 WchA (accession number AAK20699), S. pneumoniae Cps14E (accession number CAA59777),Streptococcus salivarius CpsE (accession number CAC18355), Streptococcus thermophilus EpsE (accession number AAC44012), Lactococcus lactisB35 EpsD (accession number AAD22526), and Geobacillus stearothermophilus NRS 2004/3a (accession number AAR99615). All accessionnumbers are from GenBank.


on Decem


.org/D

ownloaded from

http://jb.asm.org/

In order to verify that a promoter exists upstream of hfsE, weconstructed a transcriptional fusion, plachfs12, which yielded1,500 Miller units of �-galactosidase activity, indicating thatthere is a promoter residing in this 539-nucleotide fragmentupstream of hfsE. To determine whether there were any pro-moters in the hfsE coding region, two additional transcrip-tional fusions were constructed: plachfs15 and plachfs9. Nei-ther of these fusions had �-galactosidase activity, indicatingthat there are no promoters in the hfsE coding region. Theplachfs3 fusion, which included 1,000 nucleotides upstream ofthe probable start codon of hfsF and 45 codons of hfsF, had 930Miller units of �-galactosidase activity. Since the 5 end ofplachfs3 overlaps with plachfs15 and plachfs9 and there is nopromoter in the corresponding fragments, a promoter must liebetween the 3 end of plachfs9 and that of plachfs3. To delin-eate this promoter further, we constructed plachfs13. Theplachfs9 fusion had essentially no �-galactosidase activity,while the plachfs13 fusion had 500 Miller units of �-galactosi-dase activity, confirming the presence of a promoter in thisregion. These results indicate that there is a promoter in thecoding region of hfsF, which resides in a 218-bp nucleotidefragment contained within the plachfs13 transcriptional fusion.

To determine if there are additional promoters driving hfsGand hfsH gene expression, we constructed two other transcrip-tional fusions: plachfs14 and plachfs5, neither of which had�-galactosidase activity (Fig. 2). These results indicate thatthere is no promoter between hfsF and hfsG or between hfsGand hfsH. Based on the results of this promoter fusion analysis,we concluded that there are at least two promoters present inthe hfsEFGH gene cluster. Knowledge of the approximatelocation of these promoters was then used to construct non-polar deletions of the hfsEFGH genes.

Contribution of hfsEFGH to holdfast synthesis. To deter-mine the contribution of hfsEFGH to holdfast synthesis and toeliminate the possibility that the phenotypes of the CB2Atransposon insertion mutants were due to polar effects ondownstream genes, we generated in-frame-deletion mutantswith deletions of each of the four genes, hfsE (CC_2425), hfsF(CC_2426), hfsG (CC_2427), and hfsH (CC_2428), in strainCB15 and assayed for their ability to adhere to a polystyrenesurface and to bind a holdfast-specific lectin, WGA. Mutants

with deletions in hfsE and hfsF adhered to polystyrene withonly a small decrease in their binding abilities compared to thewild type, whereas the hfsG and hfsH mutants were severelyimpaired in their abilities to adhere to polystyrene (Fig. 3A).Lectin binding assays using AF488-WGA mimicked the adher-ence results: �hfsE and �hfsF predivisional cells had a slightlyreduced ability to bind to AF488-WGA, whereas �hfsG and�hfsH predivisional cells were deficient in binding to AF488-WGA (Fig. 3B).

In order to observe the holdfasts present in the hfs deletionmutants, we performed TEM on negatively stained whole-cellmounts of these hfs deletion mutants. In a wild-type CB15 cell,the holdfast appeared as a darkly stained amorphous materiallocalized at the tip of the stalk (Fig. 4, panel 1). In contrast,cells of NA1000, a spontaneous holdfast-deficient mutant ofCB15, completely lacked a detectable holdfast (Fig. 4, panel17). The hfsG and hfsH mutants were completely devoid ofholdfast material (Fig. 4, panels 3 and 4), whereas the hfsE andhfsF mutants had holdfasts that were comparable to that of thewild-type cell (Fig. 4, panels 5 and 2).

To confirm that the phenotype of each mutant was solelydue to the deletions, we introduced a series of complementa-tion plasmids carrying the native hfsE promoter fused to thehfsE, hfsF, hfsG, or hfsH gene into their respective deletionmutants and assayed for surface adhesion and lectin binding.All of the deletion mutants were complemented for adherenceto polystyrene (Fig. 3A) and their ability to bind AF488-WGA

FIG. 2. Promoter map of the hfsEFGH gene cluster. Transcrip-tional fusions to a promoterless lacZ were assayed for �-galactosidaseactivity. �-Galactosidase activities are expressed in Miller units withstandard deviations and are the averages of five independent assays.The approximate locations of the hfs promoters are indicated by thebent arrows. The dotted lines under the bent arrows represent thepositions of the promoters. All fusion fragments are drawn to scale anddepicted by solid horizontal lines.

FIG. 3. (A) Quantification of crystal violet-stained cells attached topolystyrene. CB15 is the wild type, and NA1000 is a nonadherentstrain. A minus () indicates an hfsEFGH in-frame-deletion mutant,and a plus (�) indicates the respective complemented deletion mutant.The results represent the means of independent experiments; themeans were calculated from a total of nine measurements derivedfrom three samples per experiment, three independent times. Errorbars represent standard deviations. (B) Quantitation of holdfastAF488-WGA labeling of predivisional (PD) cells. A total number of500 predivisional cells were counted, and the percent values reflect theproportion of AF488-WGA-labeled predivisional cells versus the totalof all predivisional cells counted. Error bars represent standard devi-ations.


on Decem


.org/D

ownloaded from

http://jb.asm.org/

(Fig. 3B), indicating that the phenotype of each mutant wascaused by the individual gene deletions.

Although the holdfast is a critical player for the optimalattachment to surfaces, other polar structures such as pili andthe flagellum have been shown to play a role in the attachmentof C. crescentus to surfaces (6). To verify that the attachmentdeficiencies of the hfsEFGH deletion mutants were solely due

to defects in holdfast synthesis and not the lack of pili ormotility, we assayed for the presence of pili in these mutants,using a phage sensitivity assay with the caulophage �CbK,which requires pili for infection (55), and performed a swarmmotility assay, which identifies swimming defects in semisolidagar. All of the hfsEFGH deletion mutants were sensitive to�CbK (data not shown), indicating that these deletion mutants

FIG. 4. TEM analysis of holdfasts of hfs deletion mutants and all the hfsE paralog combination mutants. Electron micrographs show holdfastsof representative stalked cells in the cell population at a magnification of �25,000. Labels: �E, �hfsE; �Y, �pssY; �Z, �pssZ; �1436, �CC_1486.Panels: 1, CB15; 2, �hfsF; 3, �hfsG; 4, �hfsH; 5, �hfsE; 6, �pssY; 7, �pssZ; 8, �CC_1486; 9, �hfsE �pssY �pssZ; 10, �pssY �pssZ; 11, �pssY�CC_1486; 12, �pssZ �CC_1486; 13, �hfsE �pssY �pssZ �CC_1486; 14, �hfsE �pssY; 15, �hfsE �pssZ; 16, �hfsE �CC_1486; 17, NA1000; 18,�hfsE �pssY �CC_1486; 19, �hfsE �pssZ �CC_1486; 20, �pssY �pssZ �CC_1486.


on Decem


.org/D

ownloaded from

http://jb.asm.org/

possess pili and were motile in semisolid agar (data notshown), which indicates that flagellum function is not affectedby these gene deletions.

We concluded that the hfsG and hfsH genes play a criticalrole in surface attachment and holdfast synthesis. Surprisingly,the deletion of hfsE and hfsF had only a minor effect on surfaceadhesion and holdfast synthesis.

hfsE has two paralogs that are redundant for holdfast syn-thesis. Based on amino acid sequence comparisons, HfsEshould catalyze the first step in polysaccharide biosynthesis bytransferring glucose-1-phosphate residues from UDP-GlcNActo the lipid carrier molecule, undecaprenol phosphate, in theinner membrane. The loss of function of such an enzymeshould render the cell incapable of manufacturing the holdfastpolysaccharide. Since the hfsE mutant exhibited negligible de-fects in adhesion and holdfast synthesis, we considered thepossibility that paralogs of hfsE might exist and provide redun-dant function. Analysis of the C. crescentus genome revealedthe presence of three paralogs to hfsE: the exopolysaccharideproduction protein encoded by pssZ (CC_2384), the exopoly-saccharide production protein encoded by pssY (CC_0166),and the protein encoded by the CC_1486 gene, which exhibit51%, 41%, and 37% amino acid identity to the protein en-coded by hfsE (CC_2425), respectively. All three predictedprotein products share significant amino acid similarity withother known sugar transferases that function as initiating en-zymes, including the three blocks of conserved amino acidresidues found in other prototypes of these enzymes, suggest-ing that these proteins may have similar functions (Fig. 1B).

To assess whether any of the paralogs of hfsE can compen-sate for the function of hfsE, we constructed in-frame-deletionmutants with deletions of each of the individual paralogousgenes and a comprehensive set of double-, triple-, and quadru-ple-combination mutants with hfsE and assayed for deficien-cies in surface adhesion and the ability to synthesize a holdfast.Single-deletion mutants with deletions of hfsE, pssY, and pssZhad similar, slightly reduced abilities to bind to a polystyrenesurface, whereas the CC_1486 mutant had a slightly increasedsurface adhesion phenotype (Fig. 5). The slight increase in surfaceadherence of the CC_1486 mutant is likely attributable to the

overproduction of an exopolysaccharidelike material in theculture (Fig. 6), resulting in a massive cell aggregation pheno-type exhibited by the single deletion of the CC_1486 mutant(Fig. 6). The large cell clumps made by �CC_1486 are oftenup to three times as wide as that depicted in the micrograph(Fig. 6).

Holdfast-specific lectin binding assays of the single-deletionmutants showed that they all bound lectin. In order to observethe holdfasts on these single-deletion mutants, TEM was per-formed on negatively stained whole-cell mounts. Each singlemutant possessed holdfast material at the tip of its stalk (Fig.4, panels 6 to 8). These results indicate that single mutations inhfsE, pssY, pssZ, and the CC_1486 gene alone do not preventholdfast synthesis (Fig. 5).

In the set of double-deletion hfsE mutants and its paralogs,the �pssY �CC_1486 mutant exhibited essentially wild-typelevels of surface adherence and the �pssY �pssZ and �pssZ�CC_1486 mutants had a similar, slightly reduced ability tobind to a polystyrene surface (Fig. 5). In contrast, the �hfsE�pssY and �hfsE �pssZ mutants exhibited a noticeable reduc-

FIG. 5. Quantification of crystal violet-stained cells of various paralogous mutant combinations of hfsE attached to polystyrene. The resultsrepresent the means of measurements derived from three samples from each of three biological replicates. Error bars represent standarddeviations.

FIG. 6. Biofilm formation in culture tubes (top panel) and phase-contrast micrographs (bottom panel) of three C. crescentus strains:wild-type strain CB15, a nonadherent NA1000 strain, and the�CC_1486 strain, respectively.


on Decem


.org/D

ownloaded from

http://jb.asm.org/

tion in surface adherence compared to the wild type (Fig. 5).Interestingly, the �hfsE �CC_1486 mutant exhibited a signif-icant increase in surface adherence compared to the wild type.All the double mutants bound the holdfast-specific lectin.TEM analyses on whole-cell mounts of these double mutantsindicated that although all of them possessed some holdfastmaterial at the tips of their stalks, there were some consistentdifferences in the holdfast appearance of a large proportion ofsome double mutants compared to that of the wild type (Fig. 4,panels 10 to 12 and 14 to 16). The holdfast of the �pssY �pssZmutant appeared to be positioned off to one side of the stalktip (Fig. 4, panel 10) compared to the holdfast of the wild-typeCB15 cell (Fig. 4, panel 1). The �hfsE �pssY, �hfsE �pssZ, and�hfsE �CC_1486 mutants appeared to have less holdfast ma-terial at the stalk tip (Fig. 4, panels 14 to 16) than the wild-typecell (Fig. 4, panel 1). The holdfast of the �pssZ �CC_1486mutant, on the other hand, appeared to be positioned only atthe very tip of the stalk (Fig. 4, panel 12), instead of envelopingthe entire stalk tip like that of the wild-type cell (Fig. 4, panel1). These results revealed that double-combination mutationswith hfsE, pssY, pssZ, or CC_1486 do not cause a dramaticreduction in surface adherence or in the inability to synthesizea holdfast but often cause holdfast misplacement at the tip ofthe stalk.

Of all the triple-combination mutants, only the �hfsE �pssY�pssZ mutant exhibited a severe deficiency in surface adher-ence (Fig. 5) and a complete loss of binding to AF488-WGAand lacked any holdfast material at the tip of the stalk, basedon TEM (Fig. 4, panel 9). In contrast, the �pssY �pssZ�CC_1486, �hfsE �pssY �CC_1486, and �hfsE �pssZ�CC_1486 mutants were all competent in surface adherence,and the �hfsE �pssY �CC_1486 and �hfsE �pssZ �CC_1486mutants exhibited an increase in surface adherence comparedto the wild-type (Fig. 5). The �hfsE �pssY �CC_1486, �hfsE�pssZ �CC_1486, and �pssY �pssZ �CC_1486 mutants allbound the holdfast-specific lectin and possessed holdfast ma-terial at the tips of their stalks (Fig. 4, panels 18 to 20), al-though the holdfast of the �hfsE �pssY �CC_1486 mutant alsoappeared to be positioned off to one side of the stalk tip, whilethere appeared to be less holdfast material at the stalk tips ofthe �pssY �pssZ �CC_1486 and �hfsE �pssZ �CC_1486 mu-tants. The phenotype of the �hfsE �pssY �pssZ �CC_1486quadruple mutant closely mimicked that of the �hfsE �pssY�pssZ triple mutant. The �hfsE �pssY �pssZ mutant did notbind the holdfast-specific lectin, but when complemented bythe introduction of a plasmid harboring hfsE fused to its nativepromoter, 70% of the predivisional cells bound the holdfast-specific lectin, compared to 78% of wild-type predivisionalcells. These results indicate that pssY, pssZ, or hfsE is redun-dant for holdfast synthesis.

hfsI is a paralog of hfsC that is redundant for holdfastsynthesis. Unraveling the existence of compensatory paralogsfor hfsE led us to reinvestigate the role of hfsC. HfsC is apredicted integral membrane protein with 11 transmembranehelices with sequence similarity to Wzy-like proteins. Wzy pro-teins are O-antigen polymerases that are involved in the poly-merization of high-molecular-weight exopolysaccharide (44)and succinoglycan (50). Previous work has shown that an hfsCmutant could still bind to surfaces and to AF488-WGA lectin,suggesting that hfsC does not play a role in holdfast synthesis,

despite the fact that it is located in the holdfast biosynthesisgene cluster or that there was a redundant gene (58). Sequenceanalysis revealed a possible redundant protein encoded byCC_0499 (hereinafter called hfsI) exhibiting 60% sequenceidentity to HfsC. We constructed an in-frame hfsI deletionmutant and assayed for any surface adherence and AF488-WGA lectin binding deficiencies. The �hfsI mutant had closeto wild-type levels of surface adherence (Fig. 7A), while the�hfsC mutant exhibited 78% surface adherence compared tothat of the wild type. The levels of AF488 lectin binding of the�hfsC mutant and the �hfsI mutant were similar to each otherbut were slightly reduced compared to that of the wild type(Fig. 7B). We constructed an hfsC and hfsI double mutant andassayed for surface adherence and holdfast-specific lectin bind-ing. The hfsC hfsI double mutant was completely incapable ofbinding to a polystyrene surface (Fig. 7A) and failed to bindthe holdfast-specific lectin AF488-WGA (Fig. 7B). In addition,the holdfast phenotype of the hfsC hfsI double mutant could becomplemented by the introduction of a plasmid harboring hfsCfused to the hfsA promoter. We concluded that hfsC and hfsIare redundant for holdfast synthesis.

DISCUSSION

In this study, we identified a set of polysaccharide biosyn-thesis genes that are involved in the biosynthesis (hfsEFGH,pssY, and pssZ) and polymerization (hfsC and hfsI) of therepeat unit of the holdfast polysaccharide. Together with the

FIG. 7. (A) Quantification of crystal violet-stained cells of �hfsC,�hfsI, and �hfsC �hfsI mutants attached to polystyrene. The resultsrepresent the means of independent experiments; the means werecalculated from a total of nine measurements each derived from threesamples per experiment, three independent times. Error bars repre-sent standard deviations. (B) Quantitation of holdfast AF488-WGAlabeling of predivisional (PD) cells. A total number of 500 predivi-sional cells were counted, and the percent values reflect the proportionof AF488-WGA-labeled predivisional cells versus the total of all pre-divisional cells counted. Error bars represent standard deviations.


on Decem


.org/D

ownloaded from

http://jb.asm.org/

previously identified hfsDAB (58) and hfaABD genes, involvedin the export and anchoring of the holdfast polysaccharide tothe stalk (28, 29, 10), we defined for the first time a pathwaythat unifies the distinct processes for holdfast polysaccharidesynthesis, export, and anchoring to the stalk.

Since the hfsEFGH and hfsCBAD gene clusters encode thefunctions required for polysaccharide synthesis, we expectedthat mutations in any of these genes would abolish surfaceadhesion and holdfast production; however, our results showedthat although hfsDAB, hfsG, and hfsH are essential for surfaceadhesion and holdfast synthesis, hfsE and hfsF are not. It is notsurprising to find that mutations in polysaccharide deacetylasearchetypes like HfsH result in a severe deficiency in surfaceadhesion and holdfast production, because the conversion ofan acetylated to a deacetylated form of the sugar may serve asa signal for polysaccharide secretion. For example, there isevidence that the N-deacetylation modification of the E. colipolysaccharide adhesin facilitates its secretion (23). In contrast,hfsE and hfsF are dispensable for surface adhesion and holdfastproduction, and the dispensability of hfsE is explained by theexistence of two paralogs of hfsE, pssY and pssZ. Our resultsrevealed that either hfsE, pssY, or pssZ, but not their paralog theCC_1486 gene, is sufficient for holdfast synthesis and surfaceadhesion. Intriguingly, strains in which only one of these threeredundant paralogs is present exhibit a slight reduction of surfaceadhesion, which suggests that each gene contributes to the bio-synthesis of the holdfast polysaccharide. PssZ is 51% identical toHfsE, and pssZ (CC_2384) is found in a gene cluster on the C.crescentus chromosome with a putative gene for a protein tyrosinephosphatase (CC_2385), a UDP-glucose-4 epimerase (CC_2383),and a UDP-glucose-6 dehydrogenase (CC_2382), suggesting thatthese gene products may synthesize an activated sugar precursorthat could be used in polysaccharide synthesis by the initiatingglycosyltransferase, PssZ. PssY exhibits 41% amino acid identityto HfsE, and pssY (CC_0166) is found in a gene cluster on the C.crescentus chromosome with a putative transglutaminaselike cysteine peptidase (CC_0167), an O-antigen polymerasegene (CC_0165), and a chain-length-determinant gene(CC_0164). This genetic organization of the pssY gene cluster,comprising a chain-length-determinant protein adjacent to apolymerase protein, is typical of gene clusters involved in thebiosynthesis of the O antigen of lipopolysaccharide, suggestingthat the primary function of PssY is in lipopolysaccharide syn-thesis but that pssY can substitute for hfsE and pssZ because ofsequence similarity. CC_1486 is unable to compensate for theloss of hfsE and its paralogs, indicating that this gene does notplay a role in holdfast polysaccharide synthesis. Compared tothe other paralogs, CC_1486 exhibits the lowest amino acidsequence identity to HfsE (37%). CC_1486 is found in a genecluster with tipN (CC_1485), a gene involved in establishingpolarity and the correct placement of polar organelles (22, 30)instead of being clustered with other polysaccharide biosyn-thetic genes, suggesting that CC_1486 may have other func-tions in the cell.

Similarly, we have identified hfsI (CC_0499), a paralog ofhfsC that is able to compensate for its loss. We observeddramatic defects in surface adhesion and holdfast synthesiscaused by an hfsC and hfsI double-deletion mutant. There isprecedence for redundancy in Wzy-like proteins. Two versionsof Wzy proteins exist in Pseudomonas aeruginosa and Salmo-

nella enterica serovar Anatum (25, 39). Each of the P. aerugi-nosa Wzy proteins catalyzes a different glycosidic linkage (� or�) of the O antigen, and mutations of one of the wzy genesresult in the production of an O antigen missing the specific �or � linkage. hfsI is not part of a gene cluster of polysaccharidebiosynthesis genes but is instead adjacent to genes encoding aputative RDD (arginine-aspartate-aspartate) family of trans-porter proteins (CC0498) and a phage tail fiber adhesin(CC_0500). We therefore hypothesized that hfsI primarilyfunctions in holdfast synthesis.

Although it is possible that there may be other flippasesthat can compensate for the loss of HfsF function, the iden-tification of flippases is made difficult because these proteinsare composed mostly of transmembrane helices (12–14) andhave no amino acid sequence similarity or an obvious signaturesequence except a weakly conserved stretch of approximately208 amino acids (polysacc_synt domain, Protein Families da-tabase of alignments; HMMs, http://cmr.tigr.org/cgi-bin/CMR/HmmReport.cgi?hmm_acc�PF01943) (38). Alternatively, wehypothesized that there may be another polysaccharide effluxsystem that can translocate the repeat units across the innermembrane, such as the ABC transporter superfamily of pro-teins that constitutes the second pathway of polysaccharidebiosynthesis (2), which is distinct and independent from theWzy-dependent pathway. A hallmark of the ABC transporter-dependent polysaccharide translocation system is the absenceof flippases (66).

Our current understanding of polysaccharide biosynthesisstems from years of research in capsular polysaccharide bio-synthesis (62, 66). Capsular polysaccharide biosynthesis andassembly occur by two main mechanisms. The first mechanismgoverns group 1 capsular polysaccharide synthesis and involvestwo integral inner membrane proteins in polymer translocationacross the inner membrane, namely the polymerase proteinWzy and the flippase protein Wzx (67). The presence of thesetwo proteins defines this pathway. Wzy-dependent pathwaystypically manufacture heteropolymeric polysaccharides (62). Inthe second mechanism, which governs group 2 capsular poly-saccharide synthesis, polymer translocation is facilitated by anATP binding cassette (ABC) transporter. The presence of theABC transporter defines this pathway. Homopolymeric poly-saccharides are usually synthesized in an ABC transporter-dependent fashion (7). Based on the presence of the Wzy(s)and Wzx homologs in C. crescentus, HfsC or HfsI, and HfsF,respectively, we propose a working model for the synthesis,export, and anchoring of the holdfast polysaccharide to thestalk that is analogous to group I capsular polysaccharide syn-thesis in E. coli (Fig. 8). An initiating glycosyltransferase(HfsE, PssY, or PssZ) adds the first sugar to the lipid carrierundecaprenol phosphate on the cytoplasmic side of the innermembrane. Subsequently, the glycosyltransferase HfsG elon-gates the growing polysaccharide repeat unit. Upon thede-O- or de-N-acetylation of substituted saccharides by thecarbohydrate esterase HfsH, the polysaccharide repeat unitis flipped across the inner membrane by the action of theflippase, HfsF, and polymerization of the polysacchariderepeat units is accomplished by the polymerases HfsC andHfsI. The polysaccharide chain undergoes continued poly-merization, and the polymerization of the polysaccharide iscontrolled by the combined action of HfsA and HfsB, rep-


on Decem


.org/D

ownloaded from

http://jb.asm.org/

resenting the periplasmic N-terminal and cytoplasmic C-terminal domains, respectively, of a polysaccharide exportprotein. The secretin HfsD then provides the channel forexport of the polysaccharide across the peptidoglycan andouter membrane. Subsequently, the holdfast polysaccharideis anchored to the cell surface by the concerted action of theholdfast attachment proteins, HfaABD (G. G. Hardy et al.,submitted for publication).

In summary, we showed that there are redundancies for twokey enzymes involved in the initiating step and polymerizationof the holdfast polysaccharide. To our knowledge, this level ofgene redundancy for the membrane glycosyltransferase thatinitiates the first step in polysaccharide biosynthesis is unprec-edented. Interestingly, we did not notice any redundancies inthe enzymes involved in translocation through the outer mem-brane. This finding supports the current paradigm that en-zymes involved in polysaccharide synthesis and polymerizationare substitutable but the polysaccharide translocation machin-eries are specialized (66). While hfsEFGH and hfsCBAD en-code the proteins required for the biosynthesis and export ofthe holdfast polysaccharide, the presence of redundant genesinvolved in the initiating and polymerization steps may providethe ability to add additional sugars and linkages, or both, to theholdfast polysaccharide. Future work will be needed to dissectthe precise roles that each of these redundant enzymes plays inpolysaccharide synthesis.

ACKNOWLEDGMENTS

We thank the members of the Brun lab and the Kurtz lab for criticalreadings of the manuscript. We also express special gratitude to Pam-ela Bonner-Brown, Patrick Curtis, Clay Fuqua, and Ellen M.Quardokus for helpful scientific discussions.

Early stages of this work were supported by grant GM51986 fromthe National Institutes of Health to Y.V.B and by funds from theCollege of Agriculture, Forestry and Life Sciences of Clemson Uni-versity to H.D.K. Most of this work was supported by grant GM077648from the National Institutes of Health to Y.V.B. and by the IndianaMETACyt Initiative of Indiana University, funded in part through amajor grant from the Lilly Endowment, Inc.

REFERENCES

1. Agladze, K., X. Wang, and T. Romeo. 2005. Spatial periodicity of Escherichiacoli K-12 biofilm microstructure initiates during a reversible, polar attach-ment phase of development and requires the polysaccharide adhesin PGA. J.Bacteriol. 187:8237–8246.

2. Alaimo, C., I. Catrein, L. Morf, C. L. Marolda, N. Callewaert, M. A. Valvano,M. F. Feldman, and M. Aebi. 2006. Two distinct but interchangeable mech-anisms for flipping of lipid-linked oligosaccharides. EMBO J. 25:967–976.

3. Ausubel, F. M., R. Brent, R. E. Kingston, D. Moore, J. G. Seidman, J. A.Smith, and K. Struhl. 1989. Current protocols in molecular biology. JohnWiley/Greene, New York, NY.

4. Becker, A., A. Kleickmann, H. Kuster, M. Keller, W. Arnold, and A. Puhler.1993. Analysis of the Rhizobium meliloti genes exoU, exoV, exoW, exoT, andexoI involved in exopolysaccharide biosynthesis and nodule invasion: exoUand exoW probably encode glucosyltransferases. Mol. Plant-Microbe Inter-act. 6:735–744.

5. Becker, A., S. Ruberg, H. Kuster, A. A. Roxlau, M. Keller, T. Ivashina, H. P.Cheng, G. C. Walker, and A. Puhler. 1997. The 32-kilobase exp gene clusterof Rhizobium meliloti directing the biosynthesis of galactoglucan: geneticorganization and properties of the encoded gene products. J. Bacteriol.179:1375–1384.

FIG. 8. Model of holdfast biosynthesis, export, and attachment in C. crescentus. IM, inner membrane; PG, peptidoglycan; OM, outermembrane, LPS, lipopolysaccharide. The protein names in parentheses are the E. coli homologs. (1) Activated precursors (nucleotide mono-phospho and diphospho sugars [denoted by gray diamonds]) in the cytoplasm are assembled on a lipid carrier, undecaprenyl phosphate (PP�),by the glycosyltransferase HfsE, PssY, or PssZ. (2) The glycosyltransferase HfsG participates in polysaccharide elongation by transferring sugarsubunits to the growing repeat unit. (3) The carbohydrate esterase HfsH deacetylates one or more sugar residues (denoted by black diamonds).The PP�-linked repeats are then flipped across the inner membrane by the flippase HfsF. (4) Repeat units are polymerized by the polymerasesHfsC and HfsI. (5) The polysaccharide is exported by the concerted actions of membrane periplasmic auxiliary protein 1, HfsA; a tyrosineautokinase, HfsB; and an outer membrane channel formed by HfsD (6). The polysaccharide is anchored to the cell by the secreted Hfa proteins.


on Decem


.org/D

ownloaded from

http://jb.asm.org/

6. Bodenmiller, D., E. Toh, and Y. V. Brun. 2004. Development of surfaceadhesion in Caulobacter crescentus. J. Bacteriol. 186:1438–1447.

7. Bronner, D., B. R. Clarke, and C. Whitfield. 1994. Identification of anATP-binding cassette transport system required for translocation of lipo-polysaccharide O-antigen side-chains across the cytoplasmic membrane ofKlebsiella pneumoniae serotype O1. Mol. Microbiol. 14:505–519.

8. Bugert, P., and K. Geider. 1995. Molecular analysis of the ams operonrequired for exopolysaccharide synthesis of Erwinia amylovora. Mol. Micro-biol. 15:917–933.

9. Caufrier, F., A. Martinou, C. Dupont, and V. Bouriotis. 2003. Carbohydrateesterase family 4 enzymes: substrate specificity. Carbohydr. Res. 338:687–692.

10. Cole, J. L., G. G. Hardy, D. Bodenmiller, E. Toh, A. Hinz, and Y. V. Brun.2003. The HfaB and HfaD adhesion proteins of Caulobacter crescentus arelocalized in the stalk. Mol. Microbiol. 49:1671–1683.

11. Drummelsmith, J., and C. Whitfield. 1999. Gene products required forsurface expression of the capsular form of the group 1 K antigen in Esche-richia coli (O9a:K30). Mol. Microbiol. 31:1321–1332.

12. Entcheva-Dimitrov, P., and A. Spormann. 2004. Dynamics and control ofbiofilms of the oligotrophic bacterium Caulobacter crescentus. J. Bacteriol.186:8254–8266.

13. Erdem, A. L., F. Avelino, J. Xicohtencatl-Cortes, and J. A. Giron. 2007. Hostprotein binding and adhesive properties of H6 and H7 flagella of attachingand effacing Escherichia coli. J. Bacteriol. 189:7426–7435.

14. Evinger, M., and N. Agabian. 1977. Envelope-associated nucleoid from Cau-lobacter crescentus stalked and swarmer cells. J. Bacteriol. 132:294–301.

15. Feldman, M. F., C. L. Marolda, M. A. Monteiro, M. B. Perry, A. J. Parodi,and M. A. Valvano. 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.

16. Freiberg, C., R. Fellay, A. Bairoch, W. J. Broughton, A. Rosenthal, and X.Perret. 1997. Molecular basis of symbiosis between Rhizobium and legumes.Nature 387:394–401.

17. Geremia, R. A., P. Mergaert, D. Geelen, M. Van Montagu, and M. Holsters.1994. The NodC protein of Azorhizobium caulinodans is an N-acetylglu-cosaminyltransferase. Proc. Natl. Acad. Sci. USA 91:2669–2673.

18. Gober, J. W., and L. Shapiro. 1992. A developmentally regulated Cau-lobacter flagellar promoter is activated by 3 enhancer and IHF bindingelements. Mol. Biol. Cell 3:913–926.

19. Gonin, M., E. M. Quardokus, D. O’Donnol, J. Maddock, and Y. V. Brun.2000. Regulation of stalk elongation by phosphate in Caulobacter crescentus.J. Bacteriol. 182:337–347.

20. Hekmat, O., K. Tokuyasu, and S. G. Withers. 2003. Subsite structure of theendo-type chitin deacetylase from a Deuteromycete, Colletotrichum linde-muthianum: an investigation using steady-state kinetic analysis and MS.Biochem. J. 374:369–380.

21. Hinnebusch, B. J., and D. L. Erickson. 2008. Yersinia pestis biofilm in the fleavector and its role in the transmission of plague. Curr. Top. Microbiol.Immunol. 322:229–248.

22. Huitema, E., S. Pritchard, D. Matteson, S. K. Radhakrishnan, and P. H.Viollier. 2006. Bacterial birth scar proteins mark future flagellum assemblysite. Cell 124:1025–1037.

23. Itoh, Y., J. D. Rice, C. Goller, A. Pannuri, J. Taylor, J. Meisner, T. J.Beveridge, J. F. Preston III, and T. Romeo. 2008. Roles of pgaABCD genesin synthesis, modification, and export of the Escherichia coli biofilm adhesinpoly-�-1,6-N-acetyl-D-glucosamine. J. Bacteriol. 190:3670–3680.

24. Janakiraman, R. S., and Y. V. Brun. 1999. Cell cycle control of a holdfastattachment gene in Caulobacter crescentus. J. Bacteriol. 181:1118–1125.

25. Kaluzny, K., P. D. Abeyrathne, and J. S. Lam. 2007. Coexistence of twodistinct versions of O-antigen polymerase, Wzy-� and Wzy-�, in Pseudomo-nas aeruginosa serogroup O2 and their contributions to cell surface diversity.J. Bacteriol. 189:4141–4152.

26. Kelly, A. J., M. J. Sackett, N. Din, E. M. Quardokus, and Y. V. Brun. 1998.Cell cycle-dependent transcriptional and proteolytic regulation of FtsZ inCaulobacter. Genes Dev. 12:880–893.

27. Krol, J., and A. Skorupska. 1997. Identification of genes in a Rhizobiumleguminosarum bv. trifolii whose products are homologues to a family ofATP-binding proteins. Microbiology 143:1389–1394.

28. Kurtz, H. D., Jr., and J. Smit. 1992. Analysis of a Caulobacter crescentus genecluster involved in attachment of the holdfast to the cell. J. Bacteriol. 174:687–694.

29. Kurtz, H. D., Jr., and J. Smit. 1994. The Caulobacter crescentus holdfast:identification of holdfast attachment complex genes. FEMS Microbiol. Lett.116:175–182.

30. Lam, H., W. B. Schofield, and C. Jacobs-Wagner. 2006. A landmark proteinessential for establishing and perpetuating the polarity of a bacterial cell.Cell 124:1011–1023.

31. Lawler, M. L., D. E. Larson, A. J. Hinz, D. Klein, and Y. V. Brun. 2006.Dissection of functional domains of the polar localization factor PodJ inCaulobacter crescentus. Mol. Microbiol. 59:301–316.

32. Leipold, M. D., N. A. Kaniuk, and C. Whitfield. 2007. The C-terminal

domain of the Escherichia coli WaaJ glycosyltransferase is important forcatalytic activity and membrane association. J. Biol. Chem. 282:1257–1264.

33. Lellouch, A. C., and R. A. Geremia. 1999. Expression and study of recom-binant ExoM, a �1-4 glucosyltransferase involved in succinoglycan biosyn-thesis in Sinorhizobium meliloti. J. Bacteriol. 181:1141–1148.

34. Levi, A., and U. Jenal. 2006. Holdfast formation in motile swarmer cellsoptimizes surface attachment during Caulobacter crescentus development. J.Bacteriol. 188:5315–5318.

35. Li, G., C. S. Smith, Y. V. Brun, and J. X. Tang. 2005. The elastic propertiesof the Caulobacter crescentus adhesive holdfast are dependent on oligomersof N-acetylglucosamine. J. Bacteriol. 187:257–265.

36. Liss, L. R. 1987. New M13 host: DH5�F competent cells. Focus 9 3:13.37. Mack, D., W. Fischer, A. Krokotsch, K. Leopold, R. Hartmann, H. Egge, and

R. Laufs. 1996. The intercellular adhesin involved in biofilm accumulation ofStaphylococcus epidermidis is a linear �-1,6-linked glucosaminoglycan: puri-fication and structural analysis. J. Bacteriol. 178:175–183.

38. Marolda, C. L., J. Vicarioli, and M. A. Valvano. 2004. Wzx proteins involvedin biosynthesis of O antigen function in association with the first sugar of theO-specific lipopolysaccharide subunit. Microbiology 150:4095–4105.

39. McConnell, M. R., K. R. Oakes, A. N. Patrick, and D. M. Mills. 2001. Twofunctional O-polysaccharide polymerase wzy (rfc) genes are present in the rfbgene cluster of Group E1 Salmonella enterica serovar Anatum. FEMS Mi-crobiol. Lett. 199:235–240.

40. Merker, R. I., and J. Smit. 1988. Characterization of the adhesive holdfast ofmarine and freshwater caulobacters. Appl. Environ. Microbiol. 54:2078–2085.

41. Mishra, C., C. E. Semino, K. J. McCreath, H. de la Vega, B. J. Jones, C. A.Jones, and P. W. Robbins. 1997. Cloning and expression of two chitindeacetylase genes of Saccharomyces cerevisiae. Yeast 13:327–336.

42. Mitchell, D., and J. Smit. 1990. Identification of genes affecting productionof the adhesion organelle of Caulobacter crescentus CB2. J. Bacteriol. 172:5425–5431.

43. Mukhopadhyay, P., J. Williams, and D. Millis. 1988. Molecular analysis of apathogenicity locus in Pseudomonas syringae pv. syringae. J. Bacteriol. 170:5479–5488.

44. Nakhamchik, A., C. Wilde, and D. A. Rowe-Magnus. 2007. Identification ofa Wzy polymerase required for group IV capsular polysaccharide and lipo-polysaccharide biosynthesis in Vibrio vulnificus. Infect. Immun. 75:5550–5558.

45. Nierman, W. C., T. V. Feldblyum, M. T. Laub, I. T. Paulsen, K. E. Nelson,J. A. Eisen, J. F. Heidelberg, M. R. Alley, N. Ohta, J. R. Maddock, I. Potocka,W. C. Nelson, A. Newton, C. Stephens, N. D. Phadke, B. Ely, R. T. DeBoy,R. J. Dodson, A. S. Durkin, M. L. Gwinn, D. H. Haft, J. F. Kolonay, J. Smit,M. B. Craven, H. Khouri, J. Shetty, K. Berry, T. Utterback, K. Tran, A. Wolf,J. Vamathevan, M. Ermolaeva, O. White, S. L. Salzberg, J. C. Venter, L.Shapiro, and C. M. Fraser. 2001. Complete genome sequence of Caulobactercrescentus. Proc. Natl. Acad. Sci. USA 98:4136–4141.

46. Ong, C. J., M. L. Y. Wong, and J. Smit. 1990. Attachment of the adhesiveholdfast organelle to the cellular stalk of Caulobacter crescentus. J. Bacteriol.172:1448–1456.

47. Pierce, D. L., and Y. V. Brun. 2008. Developmental control in Caulobactercrescentus: strategies for survival in oligotrophic environments, p. 10. In D. E.Whitworth (ed.), Myxobacteria multicellularity and differentiation. ASMPress, Washington, DC.

48. Poindexter, J. S. 1964. Biological properties and classification of the Cau-lobacter group. Bacteriol. Rev. 28:231–295.

49. Psylinakis, E., I. G. Boneca, K. Mavromatis, A. Deli, E. Hayhurst, S. J.Foster, K. M. Varum, and V. Bouriotis. 2005. Peptidoglycan N-acetylglu-cosamine deacetylases from Bacillus cereus, highly conserved proteins inBacillus anthracis. J. Biol. Chem. 280:30856–30863.

50. Reuber, T. L., and G. C. Walker. 1993. Biosynthesis of succinoglycan, asymbiotically important exopolysaccharide of Rhizobium meliloti. Cell 74:269–280.

51. Rick, P. D., K. Barr, K. Sankaran, J. Kajimura, J. S. Rush, and C. J.Waechter. 2003. Evidence that the wzxE gene of Escherichia coli K-12 en-codes a protein involved in the transbilayer movement of a trisaccharide-lipid intermediate in the assembly of enterobacterial common antigen.J. Biol. Chem. 278:16534–16542.

52. Saxena, I. M., R. M. Brown, Jr., M. Fevre, R. A. Geremia, and B. Henrissat.1995. Multidomain architecture of �-glycosyl transferases: implications formechanism of action. J. Bacteriol. 177:1419–1424.

53. Schoenlein, P. V., L. S. Gallman, and B. Ely. 1989. Organization of the flaFGgene cluster and identification of two additional genes involved in flagellumbiogenesis in Caulobacter crescentus. J. Bacteriol. 171:1544–1553.

54. Simon, R., U. Prieffer, and A. Puhler. 1983. A broad host range mobilizationsystem for in vivo genetic engineering: transposon mutagenesis in gram-negative bacteria. Bio/Technology 1:784–790.

55. Skerker, J. M., and L. Shapiro. 2000. Identification and cell cycle control ofa novel pilus system in Caulobacter crescentus. EMBO J. 19:3223–3234.

56. Sloan, G. P., C. F. Love, N. Sukumar, M. Mishra, and R. Deora. 2007. TheBordetella Bps polysaccharide is critical for biofilm development in themouse respiratory tract. J. Bacteriol. 189:8270–8276.


on Decem


.org/D

ownloaded from

http://jb.asm.org/

57. Smit, J., and N. Agabian. 1984. Cloning of the major protein of the Cau-lobacter crescentus periodic surface layer: detection and characterization ofthe cloned peptide by protein expression assays. J. Bacteriol. 160:1137–1145.

58. Smith, C. S., A. Hinz, D. Bodenmiller, D. E. Larson, and Y. V. Brun. 2003.Identification of genes required for synthesis of the adhesive holdfast inCaulobacter crescentus. J. Bacteriol. 185:1432–1442.

59. Steiner, K., R. Novotny, K. Patel, E. Vinogradov, C. Whitfield, M. A. Valvano, P.Messner, and C. Schaffer. 2007. Functional characterization of the initiationenzyme of S-layer glycoprotein glycan biosynthesis in Geobacillus stearother-mophilus NRS 2004/3a. J. Bacteriol. 189:2590–2598.

60. Stephens, C., A. Reisenauer, R. Wright, and L. Shapiro. 1996. A cell cycle-regulated bacterial DNA methyltransferase is essential for viability. Proc.Natl. Acad. Sci. USA 93:1210–1214.

61. Tsang, P. H., G. Li, Y. V. Brun, L. B. Freund, and J. X. Tang. 2006. Adhesionof single bacterial cells in the micronewton range. Proc. Natl. Acad. Sci. USA103:5764–5768.

62. Valvano, M. A. 2003. Export of O-specific lipopolysaccharide. Front. Biosci.8:s452–s471.

63. Vuong, C., J. M. Voyich, E. R. Fischer, K. R. Braughton, A. R. Whitney, F. R.DeLeo, and M. Otto. 2004. Polysaccharide intercellular adhesin (PIA) pro-

tects Staphylococcus epidermidis against major components of the humaninnate immune system. Cell. Microbiol. 6:269–275.

64. Wang, L., D. Liu, and P. R. Reeves. 1996. C-terminal half of Salmonellaenterica WbaP (RfbP) is the galactosyl-1-phosphate transferase domain cat-alyzing the first step of O-antigen synthesis. J. Bacteriol. 178:2598–2604.

65. Wang, X., J. F. Preston III, and T. Romeo. 2004. The pgaABCD locus ofEscherichia coli promotes the synthesis of a polysaccharide adhesin requiredfor biofilm formation. J. Bacteriol. 186:2724–2734.

66. Whitfield, C. 2006. Biosynthesis and assembly of capsular polysaccharides inEscherichia coli. Annu. Rev. Biochem. 75:39–68.

67. Whitfield, C., and A. Paiment. 2003. Biosynthesis and assembly of Group 1capsular polysaccharides in Escherichia coli and related extracellular poly-saccharides in other bacteria. Carbohydr. Res. 338:2491–2502.

68. Wilbur, W. J., and D. J. Lipman. 1983. Rapid similarity searches of nucleicacid and protein data banks. Proc. Natl. Acad. Sci. USA 80:726–730.

69. Xicohtencatl-Cortes, J., V. Monteiro-Neto, M. A. Ledesma, D. M. Jordan, O.Francetic, J. B. Kaper, J. L. Puente, and J. A. Giron. 2007. Intestinal ad-herence associated with type IV pili of enterohemorrhagic Escherichia coliO157:H7. J. Clin. Investig. 117:3519–3529.


on Decem


.org/D

ownloaded from

http://jb.asm.org/

characterization of the caulobacter crescentus holdfast ... · but external appendages play...

Documents