Chaperone-assisted pilus assembly and bacterial attachment

548

Bacterial pili assembled by the chaperone-usher pathway canmediate microbial attachment, an early step in theestablishment of an infection, by binding specifically to sugarspresent in host tissues. Recent work has begun to reveal thestructural basis both of chaperone function in the biogenesis ofthese pili and of bacterial attachment.

AddressesDepartments of *Molecular Microbiology and †Biochemistry andMolecular Biophysics, Washington University School of Medicine,St Louis, Missouri 63110, USA‡Department of Molecular Biology, Uppsala Biomedical Center,Swedish University of Agricultural Sciences, Uppsala, Sweden§e-mail: [email protected]

Current Opinion in Structural Biology 2000, 10:548–556

0959-440X/00/$ — see front matter© 2000 Elsevier Science Ltd. All rights reserved.

Abbreviationsdsc donor-strand-complementedUTI urinary tract infection

IntroductionGram-negative bacteria assemble a diverse array of com-plex surface organelles that play roles in many cellularprocesses, including motility, DNA uptake, secretion ofvirulence factors and attachment to and invasion of hosttissues. The construction of these organelles requires thatthe bacterium coordinate the folding, secretion andordered assembly of multiple distinct subunit proteins. Piliare fibrous surface organelles that can mediate attachmentto host tissues, a critical early step in the development of avariety of diseases [1]. Gram-negative bacteria assemblepili by various pathways, including the chaperone-usherpathway, the type IV pilus assembly pathway, the alterna-tive chaperone-usher pathway and the type IV secretionpathway [2–9]. Pili assembled by the chaperone-usherpathway contain adhesins that bind specifically to sugarspresent in host tissues. This review will focus on recentadvances in our understanding of the structural basis bothof the biogenesis of pili by the chaperone-usher pathwayand of the attachment to host tissues mediated by thesefibers. This work has begun to reveal the molecular mech-anisms of both pilus biogenesis and organelle assembly ingeneral. The insight gained into these critical pathogenicevents promises to aid the development of new methods toprevent and treat bacterial diseases.

Organelle assembly by the chaperone-usherpathwayThe chaperone-usher pathway participates in the assemblyof more than 30 pilus and nonpilus surface organelles [2–4](Table 1). Pili assembled via this pathway vary subtly intheir gross structures. For example, the P pilus, encoded by

the 11 genes of the pap (papA–papK) gene cluster, found inmany uropathogenic strains of Escherichia coli, is a compositefiber — a thick rod with a thinner tip fibrillum at its distalend [10,11]. The rod has a diameter of 7 nm, with a hollowcore, and is composed of primarily PapA subunits wound ina tight right-handed helix; the tip fibrillum consists of pri-marily PapE subunits wound in an open-helicalconformation [11–13]. The PapG adhesin, which binds togal(α1→4)gal-containing sugars found in the human kidneyand is necessary for the development of pyelonephritis in amonkey model, is located at the tip of the fibrillum[11,14,15]. Two adaptor subunits, PapF and PapK, arethought to link the adhesin to the tip fibrillum and the tipfibrillum to the rod, respectively [16]. The type 1 pilus,encoded by the fim gene cluster (fimA–fimH), has a similarcomposite structure, but its tip fibrillum is short and stubby.The rod is composed of predominantly FimA subunitswound in a tight right-handed helix; the tip fibrillum con-tains the mannose-specific adhesin FimH, as well as FimG.FimF is thought to link the tip fibrillum to the rod[10,17,18,19••]. Analysis of the mannose-binding activity offragmented type 1 pili suggested that FimH is also interca-lated in the rod, but with its mannose-binding activityburied. Breakage of the pili at these sites would then exposethe mannose-binding activity of these FimH molecules[20]. Type 1 pili are found in most E. coli strains, includingboth uropathogenic and commensal strains, as well asthroughout the Enterobacteriaceae family, and have beenshown to play a critical role in the pathogenesis of cystitis[21–23]. Hif pili, encoded by the hif gene cluster (hifA–hifE),found in pathogenic Haemophilus influenzae strains, have rods6–7 nm in diameter and a short, thinner tip differentiation.The rods have a cross-over repeat consistent with a double-stranded right-handed helical architecture. The rod iscomposed of primarily HifA, whereas the tip contains HifDand HifE [24–28]. Inhibition of hemagglutination by anti-serum raised against HifE indicates that this proteinmediates attachment to human cells [28].

The biogenesis of each organelle assembled by the chaper-one-usher pathway involves two dedicated proteins, aperiplasmic chaperone and an outer membrane usher(Figure 1). Subunits enter the periplasm through the Secapparatus. The chaperone — PapD in the pap system,FimC in the fim system, HifB in the hif system — theninteracts with each newly translocated subunit individually,facilitating its release from the cytoplasmic membrane in aprocess that may be driven by the folding of the subunitdirectly on the chaperone template [27,29–32,33•,34••].The chaperone remains bound to the folded subunit, form-ing a stable chaperone–subunit complex. In the absence ofthe chaperone, subunits are degraded. The formation of thechaperone–subunit complex thus serves to stabilize the

Chaperone-assisted pilus assembly and bacterial attachmentFrederic G Sauer*, Michelle Barnhart*, Devapriya Choudhury‡,Stefan D Knight‡, Gabriel Waksman† and Scott J Hultgren*§

Chaperone-assisted pilus assembly and bacterial attachment Sauer et al. 549

subunit. The chaperone also simultaneously caps an inter-active surface on the subunit, thereby preventing theformation of premature subunit–subunit interactions in theperiplasm [35–37]. The PapD chaperone can also bind toitself, using the same surface that binds to subunits [38•].Chaperone–subunit complexes are specifically targeted tothe outer membrane usher — PapC in the pap system andFimD in the fim system [39–41]. PapC has been shown toform a channel 2–3 nm in diameter, large enough to allowthe passage of pilus subunits [40]. In the fim and pap sys-tems, the chaperone–adhesin complex binds most rapidlyand tightly to the usher, an interaction that is thought to ini-tiate pilus assembly [39,41]. It has been shown that theformation of a chaperone–adhesin–usher ternary complexinduces a conformational change in the FimD usher to anassembly competent form that is maintained throughoutpilus assembly [41]. The usher is thought to facilitate thedissociation of the subunit from the chaperone, exposingthe subunit interactive surface and driving its assembly intothe pilus. The chaperone is thus not incorporated into theorganelle. The pilus is thought to grow through the usher asa linear fiber that packages into its final quaternary structureupon reaching the bacterial surface [19••,40].

Structural basis for chaperone function: donorstrand complementationThe crystal structure of the PapD chaperone and the NMRstructure of the FimC chaperone have been solved [42,43•].These chaperones consist of two immunoglobulin (Ig)-like

domains, with the overall shape of a boomerang. A con-served interdomain salt bridge maintains the relativeorientation of the two domains [4,43•,44]. The conservationof the hydrophobic core residues among chaperones in thePapD superfamily indicates that they most probably sharethe same overall structure [4,43•]. Two crystal structures ofPapD in complex with peptides derived from the C termi-nus of either the PapG adhesin or the PapK subunitrevealed that the peptides bound in an extended parallelβ-sheet conformation along the G1 β strand of the N-termi-nal domain of PapD [32,37]. The C-terminal carboxylate ofeach peptide was anchored in the cleft of the chaperone byinteractions with the conserved Arg8 and Lys112 residuesof the latter molecule. Alternating conserved hydrophobicresidues in the chaperone G1 β strand and in the peptidetogether formed a large hydrophobic area on the surface ofthe chaperone–peptide complex that was suggested tonucleate subunit folding on the chaperone [32].Subsequent transverse relaxation-optimized spectroscopy(TROSY) NMR measurements of the FimC–FimH chap-erone–adhesin complex mapped the surface of FimCinvolved in the interaction with FimH and indicated that itextended beyond the chaperone G1 β strand, but wasessentially limited to the N-terminal domain [45••].

Recently, the crystal structures of the FimC–FimH complexand the PapD–PapK complex were solved [46••,47••]. TheFimH adhesin consists of two domains, an N-terminal recep-tor-binding, or lectin, domain and a C-terminal pilin domain.

Figure 1

P pilus biogenesis. Single letters indicate thecorresponding pap gene. (i) Pilus subunitsenter the periplasm through the Secmachinery (YEG). In the absence of thechaperone, subunits misfold, aggregate andare degraded (ii). The PapD chaperone bindsto nascent subunits and is thought to facilitatetheir folding by donating its G1 strand tocomplete their Ig-like folds in a mechanismtermed donor strand complementation (iii).Donor strand complementation simultaneouslystabilizes the subunit and caps its interactivegroove. The soluble chaperone–subunitcomplexes are targeted to the PapC outermembrane usher. The PapD–PapGchaperone–adhesin complex binds mostrapidly and tightly to the usher (iv), initiatingpilus assembly and ensuring that the adhesin,which binds gal(α1→4)gal sugars, is at the tipof the pilus, which grows from the base.Incorporation of subsequent subunits isthought to occur by donor strand exchange, inwhich the N-terminal extension of a subunitdisplaces the chaperone G1 strand to occupythe groove of the subunit most recentlyincorporated in the pilus (v). The pilus isthought to move through the usher pore in alinear manner and adopt its final quaternarystructure once outside the cell (vi).

YEG

YEG

Outermembrane

Periplasm

Cytoplasmicmembrane

Tip fibrillum

Pilus rod

G

G

D

C

D

C

A

K

E

F

C

(i)

(ii)(iii)

(iv) (v)D

G

A

A

D

D

D

D

D

D D D

D

EE

F

K

(vi)

Current Opinion in Structural Biology

Both the single domain of PapK and the pilin domain ofFimH have Ig folds that lack the Ig fold C-terminal seventhβ strand (strand G). The conservation of their hydrophobiccore residues, as well as experimental evidence from mutage-nesis studies, indicates that subunits assembled into pili bythe chaperone-usher pathway share this C-terminally trun-cated Ig fold [2,36]. The lack of the C-terminal β strandproduces a cleft or groove on the surface of the subunit andexposes the hydrophobic core of the domain. In an interactiontermed ‘donor strand complementation’, the G1 β strand ofthe chaperone occupies the groove of the subunit, completingits Ig fold by providing the missing seventh β strand(Figure 2) [46••,47••]. The G1 β strand lies between the A′′(or A2) strand and the C-terminal F strand (corresponding tothe C-terminal peptides crystallized in complex with PapD)of the subunit (Figure 3a). The interaction between the chaperone G1 β strand and the subunit F strand is essentiallythe same as that seen in the PapD–peptide complex

structures [32,37]. The G1 β strand thus completes the Ig foldof the subunit in a noncanonical manner, as it runs parallel tothe F strand, rather than antiparallel, as it would in a canoni-cal Ig domain. The donor strand complementation interactionpresumably stabilizes the subunit by shielding its hydropho-bic core; indeed, the conserved alternating hydrophobicresidues in the G1 β strand of the chaperone become part ofthe hydrophobic core of the subunit (Figure 2). Mutagenesisand biochemical experiments have implicated the groove ofthe subunit in subunit–subunit interactions [35,36]. Thus,donor strand complementation also prevents premature subunit polymerization by capping the interactive groove.

Donor strand exchangePilus subunits have an N-terminal extension (residues1–13 in PapK) that does not contribute to the fold of thesubunit, but rather is generally disordered in thePapD–PapK crystal structure [47••]. The N-terminal

550 Carbohydrates and glycoconjugates

Table 1

Bacterial surface organelles assembled via the chaperone-usher pathway*.

Organelle Organism Chaperone Usher Disease

FibersP pilus E. coli PapD PapC Pyelonephritis, cystitisPrs pilus E. coli PrsD PrsC Cystitis?Type 1 pilus E. coli, Salmonella ssp., FimC FimD Cystitis

Klebsiella pneumoniaeF1C pilus E. coli FocC FocD Cystitis?S pilus E. coli SfaE SfaF UTI, NBMHif pilus H. influenzae HifB HifC Otitis media, meningitisHaf pilus H. influenzae biogroup HafB HafC Brazilian purpuric fever

aegyptiusType 2 and 3 pili Bordetella pertussis FimB (FhaD) FimC (FhaA) Whooping coughPef pilus S. typhimurium PefD PefC Gastroenteritis, salmonellosisLpf pilus S. typhimurium LpfB LpfC Gastroenteritis?, salmonellosis?MR/P pilus Proteus mirabilis MrpD MrpC Nosocomial UTIPMF pilus P. mirabilis PmfD PmfC Nosocomial UTIAft pilus P. mirabilis AftB AftC UTIAF/R1 pilus E. coli AfrC AfrB Diarrhea in rabbitsK99 pilus E. coli FanE FanD Neonatal diarrhea in calves,

lambs, pigletsK88 pilus E. coli FaeE FaeD Neonatal diarrhea in piglets987P pilus E. coli FasB FasD Diarrhea in pigletsF17 pilus E. coli F17D F17papC Diarrhea in pigletsMR/K (type 3) pilus K. pneumoniae MrkB MrkC Pneumonia

Nonfimbrial adhesinsNFA1–6 family E. coli NfaE NfaC UTI, NBMAfa-1 E. coli AfaB AfaC PyelonephritisDr/Afa-111 E. coli DraE DraD UTI, diarrheaM E. coli BmaB BmaC Pyelonephritis

Atypical structuresCS3 E. coli Cs3-1 Cs3-2 Traveler’s diarrheaCS31A pilus E. coli ClpE ClpD DiarrheaCS6 pilus E. coli CssC CssD? DiarrheaAAF/1 E. coli AggD AggC DiarrheaSef S. enteritidis SefB SefC Gastroenteritis, salmonellosisF1 antigen Yersinia pestis Caf1M Caf1A PlaguePH6 antigen Y. pestis, PsaB PsaC Plague

Y. pseudotuberculosisMyf Y. enteritidis MyfB MyfC Entercolitis

Unknown structures? E. coli RalE RalD Diarrhea in rabbits

*Adapted from Table 1 in [3]. See [3,4,49] for further details. NBM, newborn meningitis.

Chaperone-assisted pilus assembly and bacterial attachment Sauer et al. 551

extension has a conserved motif of alternating hydrophobicresidues reminiscent of that found in the G1 β strand of thechaperone. Thus, it has been proposed that, during pilusbiogenesis, the N-terminal extension of one subunit dis-places the G1 β strand of the chaperone bound to the

subunit most recently incorporated into the pilus(Figure 3b) [46••,47••]. Modeling of a pilus rod on the basisof its known helical structure and dimensions indicatesthat the N-terminal strand would run antiparallel, ratherthan parallel, as the chaperone G1 β strand does in the case

Figure 2

Donor strand complementation in thePapD–PapK structure. A stereo ribbondiagram of the PapD chaperone (cyan) incomplex with the PapK subunit (gray). Thechaperone G1 strand completes the Ig fold ofthe subunit, making β-strand interactions withthe A2 and F strands of PapK (see alsoFigure 3). The alternating hydrophobicresidues (red) of the G1 strand interact withthe hydrophobic core residues (yellow) ofPapK. The conserved Arg8 and Lys112residues of the chaperone anchor theC-terminal carboxylate group of the subunit inthe cleft of the chaperone. The N-terminalextension of PapK (upstream of the gray coilin the foreground) is disordered and notvisible in the structure. The FimH adhesin hasa receptor-binding domain here, rather thanan N-terminal extension (see also Figures 3and 4), consistent with its location at the tip ofthe pilus.

Figure 3

Topology diagrams. Dashes indicateadditional polypeptide not shown. (a) Donorstrand complementation. The chaperonedonates its G1 strand (red) to complete the Igfold of the subunit (white). The fold isnoncanonical, as the G1 strand runs parallelto the F strand. The N-terminal extension isshown as a blue strand. (b) After donor strandexchange. The N-terminal extension of onesubunit now completes the Ig fold of itsneighbor in a canonical manner, as theN-terminal extension runs antiparallel to theF strand. (c) Donor-strand-complementedFimH (dscFimH). DscFimH was constructedby fusing the N-terminal extension of FimG(blue), which is predicted to complete the foldof FimH in the pilus, to the C terminus of FimHwith a four amino acid linker (green). Unlikewild-type FimH, dscFimH is stable in vivo inthe absence of the chaperone.

C-term

Receptor-bindingdomain

C-term

N-terminalextension

N-terminalextension

C-term

C-term

BC

D"

D'

F E

A'

A"

G1

BC

D"

D'

F E

A'

A"

BC

D"

D'

E

A'

A"

F BC

D"

D'

E

A'

A"

F

(a) (c)(b)

Current Opinion in Structural Biology

of donor strand complementation, to the F strand of thesubunit whose fold it completes. Thus, this donor strandexchange mechanism of pilus assembly would produce amature organelle in which each subunit contributes astrand to complete the Ig fold of its neighbor in a perfectlycanonical manner. This switch in directionality of thedonor strand during donor strand exchange may be part ofthe mechanism that ensures that the transient chaper-one–subunit interactions are less stable than the morepermanent subunit–subunit interactions [48].

Chaperone-assisted subunit foldingIt has been hypothesized that one function of the chaper-one is to facilitate the folding of the pilus subunits [32,49].Alternatively, it is possible that, during pilus biogenesis,each subunit folds on its own in the periplasm and thenbinds to the chaperone. In this scenario, fully folded sub-units might also interact with other fully folded subunits inthe periplasm, setting up a competition between produc-tive chaperone–subunit interactions and potentiallynonproductive periplasmic subunit–subunit interactions.Alternatively, the chaperone could directly facilitate sub-unit folding by providing missing steric information in theform of its G1 β strand. In this model, the chaperone firstrecognizes a portion of the non-native subunit, perhaps theC-terminal carboxylate group and C-terminal F strand, inan interaction similar to that seen in the chaperone–pep-tide complexes [32,37]. The subunit would then fold onthe chaperone template to produce the fully formed chap-erone–subunit complex [32]. The chaperone would thuscouple the folding of the subunit to the capping of itsinteractive groove, ensuring that the groove is never free tointeract nonproductively with other subunits [32,49].Recent in vitro studies indicate that wild-type FimH dena-tured in urea does not refold upon dilution at theconditions tested. In contrast, dilution of the same materialin the presence of the FimC chaperone led to the produc-tive folding of FimH to yield a FimC–FimH complex[34••]. A donor-strand-complemented version of FimH(dscFimH) was then constructed by fusing the N-terminalextension of FimG, which is predicted to complete thefold of FimH in the mature type 1 pilus, to the C terminusof FimH to yield a single molecule that should form a com-pletely canonical Ig fold (Figure 3c). Unlike wild-typeFimH, dscFimH folded into a native structure in vivo andrefolded into a native mannose-binding conformation inthe urea dilution assay [34••]. These results indicate thatthe missing steric information required for subunit foldingat these conditions can be provided in cis and suggest thatthe chaperone does indeed facilitate subunit folding bydonor strand complementation.

It has been shown that the formation of β-strand secondarystructure is context-dependent, rather than sequence-dependent [50]. The G1 β strand may provide theappropriate context for the formation of the G1FC β sheet,which forms a portion of the Ig fold of the subunit in thecomplex (Figure 3a). The interaction between the subunit

C-terminal carboxylate group and the conserved arginine andlysine residues in the chaperone cleft may anchor the sub-unit C-terminal F strand to facilitate formation of this β sheet(Figure 2). Such an interaction would simultaneously posi-tion the subunit to allow the alternating hydrophobicresidues of the chaperone G1 β strand to contribute to theappropriate formation of the hydrophobic core of the sub-unit. This model suggests that the steric informationrequired for subunit folding resides not in a single amino acidchain, but rather in two distinct polypeptides [32,34••,49].

FimH receptor binding: structural detailsThe FimH adhesin of type 1 pili can bind both mono-mannose and oligomannose moieties associated with avariety of substrates. Naturally occurring E. coli FimHalleles bind oligomannose moieties with similar affinities,but can be classified on the basis of their affinity formonomannose. Those that bind monomannose with lowaffinity are generally associated with commensal intesti-nal strains, those that bind with high affinity withuropathogenic strains [51,52••]. The former phenotype isthought to provide an advantage in colonizing the oralcavity and thence the intestinal tract, the latter in colo-nizing the urinary tract. Thus, it has been proposed thatthe FimH polymorphism arises from selection mecha-nisms that balance these two advantages [52••].Expression of type 1 pili has been shown to enhance thevirulence of uropathogenic E. coli in a mouse urinary tractinfection (UTI) model by promoting both bacterial survival and the host inflammatory response to infection[22]. A clinical study revealed that the disease associatedwith UTI caused by type-1-expressing E. coli was moresevere than that associated with type-1-negative strainsof the same serotype [22] in children. High-resolutionelectron microscopy images show that type 1 pili mediateattachment to the uroplakin-coated surface of bladderepithelial cells in mice. Infection by type-1-expressingE coli, but not by an isogenic fimH– strain, leads to theexfoliation of these uroplakin-coated cells, presumably ahost defense mechanism designed to eliminate the infec-tion. Some bacteria, however, are able to persist in thehost by invading the now exposed underlying cells, amechanism that has been proposed to contribute to thefrequent recurrence of UTIs in many patients [53••].Recent results indicate that FimH mediates the bacterialinvasion of human bladder epithelial cells [54••]. FimHhas also been shown to mediate attachment to mast cellsand macrophages ([55–57]; see also Update). In the lattercase, the attachment allows the bacteria to bypass thenormal macrophage-killing pathway and to survive intra-cellularly for at least 24 hr [56]. FimH also binds tolaminin and collagen in vitro. Although collagen lacks ter-minal mannose residues, FimH binding was inhibited bymannose [58]. Finally, type 1 pili have been implicated inbacterial biofilm formation in one model system [59•].

A putative mannose-binding pocket has been identified atthe tip of the receptor-binding domain of FimH. A

552 Carbohydrates and glycoconjugates

molecule of cyclohexylbutanoyl-N-hydroxyethyl-D-glu-camide (C-HEGA), which is not a known inhibitor ofFimH–mannose binding, but which was required for theproduction of useful FimC–FimH crystals, was foundbound near the tip of the receptor-binding domain ofFimH in the crystal structure of the complex. Amino acidsPhe1, Ile13, Asn46, Asp47, Tyr48, Ile52, Asp54, Gln133,Asn135, Tyr137, Asn138, Asp140 and Phe142 line thisputative mannose-binding pocket and give it an overallnegative charge (Figure 4) [46••]. The pocket is largeenough to completely bury a monomannose residue, butcannot accommodate anything larger. The tight binding ofoligomannose moieties is thus expected to involve addi-tional contacts to subsites in the vicinity of themannose-binding pocket. A mutation of Pro49 near thepocket abolishes monomannose binding and significantlyreduces oligomannose binding as well [60••]. A doublemutation of Asn136 and Tyr137 inhibits hemagglutinationof guinea pig erythrocytes [61], an assay that measuresmannose-binding activity. Mutations that increase the abil-ity of FimH to bind mannose map to regions in thereceptor-binding domain away from the putative bindingpocket. It has been proposed that these mutations alter theconformational stability of the protein loops in the pocket,thus allowing more promiscuous binding [60••]. An alter-native explanation is that the lectin domain containsadditional weak carbohydrate-binding sites and mutationsin these regions fortuitously change these weak bindingsites to strong ones. Some plant lectins are known to con-tain multiple carbohydrate-binding sites — for example,the garlic lectin dimer contains seven mannose-binding

sites [62]. A more complete understanding of the moleculardetails of FimH–carbohydrate interactions, of the struc-tural basis of the differential binding of monomannoseand oligomannose moieties, and of the structural basis ofFimH-mediated bacterial invasion awaits further data.

Conclusions: a molecular model for organellebiogenesisThe binding of the pilus adhesin to its sugar receptor rep-resents a critical early step in the interaction between apathogen and its host. The studies of pili reviewed herehave begun to elucidate the molecular mechanisms of bac-terial attachment and the pathogenic consequences thatfollow. They also promise to shed light on general mecha-nisms of organelle biogenesis. For example, donor strandexchange produces a very stable interaction — subunitoligomers resist dissociation by SDS at room tempera-ture — allowing the construction of a very large, sturdyorganelle [48]. The contribution by one subunit of an ele-ment of the fold, be it a strand, helix or other component,to its neighboring subunit may represent one generalmechanism by which cells construct large organelles [49].Such an assembly mechanism would most probablyrequire coordination and regulation by the cell. Recentwork has identified a growing number of chaperonesinvolved in organelle biogenesis in Gram-negative bacte-ria. For example, the flagellar axial proteins each containan amphipathic helix at their C terminus. FlgN, a flagellarassembly protein, binds to the flagellar-hook-associatedproteins FlgK and FlgL via these C-terminal amphipathichelices and enhances their export [63••]. FliT, another

Chaperone-assisted pilus assembly and bacterial attachment Sauer et al. 553

Figure 4

Stereo ribbon diagram of a FimH receptor-binding domain. Sidechains that form theputative mannose-binding pocket at the tip ofthe domain are shown and labeled.

protein associated with flagellar assembly, binds to the fla-gellar filament cap protein FliD in a similar manner. Thus,it was proposed that FlgN and FliT act as chaperones forspecific flagellar subunits by binding to their helicaldomains and preventing premature subunit oligomeriza-tion [63••]. Other work revealed that the export of thehook-type proteins FlgD and FlgE required FliJ, suggest-ing that FliJ may also act as a chaperone [64••].Chaperones are also thought to play roles in the assemblyof the type III secretion apparatus, which shares homologywith the flagellum, and of the CS1 pilus family members(assembled by the alternative chaperone-usher pathway)[8,65,66••,67••]. By analogy to both the PapD superfamilyand the flagellar chaperones, one general function of thesemolecules may be to cap the strongly interactive surfacesof subunits until they reach the proper assembly site [49,63••,68], thus ensuring that organelle biogenesisproceeds smoothly and appropriately.

UpdateRecent work has demonstrated that caveolae, membranestructures involved in macromolecular import and trans-membrane signaling, play a role in bacterial entry intobone-marrow-derived mast cells (BMMCs) [69••]. CD48, areceptor for FimH-expressing E. coli [57], co-localized withcaveolae in BMMCs. Caveolae disrupting and usurpingagents inhibited bacterial entry into mast cells and caveolae-specific markers were recruited to sites of bacterial entry.Finally, intracellular bacteria co-fractionated with caveolae.It is proposed that this caveolae-mediated entry pathwayallows FimH-expressing E. coli to bypass the normal phagocytic pathway and remain viable inside mast cells.

AcknowledgementsSJH is supported by National Institutes of Health grants RO1DK5140604and R37AI2954910, GW by National Institutes of Health grantRO1GM54033, and SDK by grants from the Swedish Natural ScienceResearch Council and the Swedish Foundation for Strategic Research(Structural Biology Network). FGS was supported by a National ScienceFoundation predoctoral fellowship.

References and recommended readingPapers of particular interest, published within the annual period of review,have been highlighted as:

• of special interest•• of outstanding interest

1. Hultgren SJ, Jones CH, Normark S: Bacterial adhesins and theirassembly. In Escherichia Coli and Salmonella Cellular and MolecularBiology, edn 2. Edited by Neidhardt FC. Washington DC: ASM Press;1996:2730-2756.

2. Soto GE, Hultgren SJ: Bacterial adhesins: common themes andvariations in architecture and assembly. J Bacteriol 1999,181:1059-1071.

3. Thanassi DG, Saulino ET, Hultgren SJ: The chaperone/usherpathway: a major terminal branch of the general secretorypathway. Curr Opin Microbiol 1998, 1:223-231.

4. Hung DL, Knight SD, Woods RM, Pinkner JS, Hultgren SJ: Molecularbasis of two subfamilies of immunoglobulin-like chaperones.EMBO J 1996, 15:3792-3805.

5. Hung DL, Hultgren SJ: Pilus biogenesis via the chaperone/usherpathway: an integration of structure and function. J Struct Biol1999, 124:201-220.

6. Wall D, Kaiser D: Type IV pili and cell motility. Mol Microbiol 1999,32:1-10.

7. Manning PA, Meyer TF: Type-4 pili: biogenesis, adhesins, proteinexport and DNA import. Proceedings of a workshop. Gene 1997,192:1-198.

8. Sakellaris H, Scott JR: New tools in an old trade: CS1 pilusmorphogenesis. Mol Microbiol 1998, 30:681-687.

9. Burns DL: Biochemistry of type IV secretion. Curr Opin Microbiol1999, 2:25-29.

10. Hull RA, Gill RE, Hsu P, Minshew BH, Falkow S: Construction andexpression of recombinant plasmids encoding type 1 orD-mannose-resistant pili from a urinary tract infection Escherichiacoli isolate. Infect Immun 1981, 33:933-938.

11. Kuehn M, Heuser J, Normark S, Hultgren SJ: P pili in uropathogenicE. coli are composite fibers with distinct fibrillar tips. Nature 1992,356:252-255.

12. Bullitt E, Makowski L: Structural polymorphism of bacterialadhesion pili. Nature 1995, 373:164-167.

13. Gong M, Makowski L: Helical structure of P pili from Escherichiacoli. Evidence from X-ray fiber diffraction and scanningtransmission electron microscopy. J Mol Biol 1992, 5:735-742.

14. Lund B, Lindberg F, Marklund B-I, Normark S: The PapG protein isthe αα-D-galactopyranosyl-αα(1àà4)-ββ-D-galactopyranosyl-bindingadhesin of uropathogenic Escherichia coli. Proc Natl Acad SciUSA 1987, 84:5898-5902.

15. Roberts JA, Marklund B-I, Ilver D, Haslam D, Kaack MB, Baskin G,Louis M, Mollby R, Winberg J, Normark S: The gal(αα1-4)gal-specifictip adhesin of Escherichia coli P-fimbriae is needed forpyelonephritis to occur in the normal urinary tract. Proc Natl AcadSci USA 1994, 91:11889-11893.

16. Jacob-Dubuisson F, Heuser J, Dodson K, Normark S, Hultgren S:Initiation of assembly and association of the structural elementsof a bacterial pilus depend on two specialized tip proteins.EMBO J 1993, 12:837-847.

17. Brinton CC Jr: The structure, function, synthesis and geneticcontrol of bacterial pili and a molecular model for DNA and RNAtransport in gram negative bacteria. Trans NY Acad Sci 1965,27:1003-1054.

18. Jones CH, Pinkner JS, Roth R, Heuser J, Nicholes AV, Abraham SN,Hultgren SJ: FimH adhesin of type 1 pili is assembled into afibrillar tip structure in the Enterobacteriaceae. Proc Natl Acad SciUSA 1995, 92:2081-2085.

19. Saulino ET, Bullitt E, Hultgren SJ: Snapshots of usher-mediated •• protein secretion and ordered pilus assembly. Proc Natl Acad Sci

USA 2000, 97:9240-9245.High-resolution electron microscopy views of type 1 pili growing through theFimD outer membrane usher reveal that the pilus moves through the usheras a linear fiber and adopts its helical conformation once outside. The paperelucidates the roles of FimF and FimG in type 1 pilus assembly.

20. Ponniah S, Endres RO, Hasty DL, Abraham SN: Fragmentation ofEscherichia coli type 1 fimbriae exposes cryptic D-mannose-binding sites. J Bacteriol 1991, 173:4195-4202.

21. Thankavel K, Madison B, Ikeda T, Malaviya R, Shah AH,Arumugan PM, Abraham SN: Localization of a domain in the FimHadhesin of Escherichia coli type 1 fimbriae capable of receptorrecognition and use of a domain-specific antibody to conferprotection against experimental urinary tract infection. J ClinInvest 1997, 100:1123-1126.

22. Connell H, Agace W, Klemm P, Schembri M, Marild S, Svanborg C:Type 1 fimbrial expression enhances Escherichia coli virulence forthe urinary tract. Proc Natl Acad Sci USA 1996, 93:9827-9832.

23. Langermann S, Palaszynski S, Barnhart M, Auguste G, Pinkner JS,Burlein J, Barren P, Koenig S, Leath S, Jones CH, Hultgren SJ:Prevention of mucosal Escherichia coli infection byFimH-adhesin-based systemic vaccination. Science 1997,276:607-611.

24. Gilsdorf JR, Marrs CF, McCrea KW, Forney LJ: Cloning, expression,and sequence analysis of the Haemophilus influenzae type bstrain M43p+ pilin gene. Infect Immun 1993, 58:1065-1072.

25. McCrea KW, Watson WJ, Gilsdorf JR, Marrs CF: Identification ofhifD and hifE in the pilus gene cluster of Haemophilus influenzaetype b strain Eagan. Infect Immun 1994, 62:4922-4928.

554 Carbohydrates and glycoconjugates

26. Watson WJ, Gilsdorf JR, Tucci MA, McCrea KW, Forney LJ,Marrs CF: Identification of a gene essential for piliation inHaemophilus influenzae type b with homology to the pilusassembly platform genes of gram-negative bacteria. InfectImmun 1994, 62:468-475.

27. St Geme JW, Pinkner JS, Krasan GP, Heuser J, Bullitt E, Smith AL,Hultgren SJ: Haemophilus influenzae pili are composite structuresassembled via the HifB chaperone. Proc Natl Acad Sci USA 1996,93:11913-11918.

28. McCrea KW, Watson WJ, Gilsdorf JR, Marrs CF: Identification oftwo minor subunits in the pilus of Haemophilus influenzae.J Bacteriol 1997, 179:4227-4231.

29. Lindberg F, Tennent JM, Hultgren SJ, Lund B, Normark S: PapD, aperiplasmic transport protein in P-pilus biogenesis. J Bacteriol1989, 171:6052-6058.

30. Jones CH, Pinkner JS, Nicholes AV, Slonim LN, Abraham SN,Hultgren SJ: FimC is a periplasmic PapD-like chaperone thatdirects assembly of type 1 pili in bacteria. Proc Natl Acad Sci USA1993, 90:8397-8401.

31. Jones CH, Danese PN, Pinkner JS, Silhavey TJ, Hultgren SJ: Thechaperone-assisted membrane release and folding pathway issensed by two signal transduction systems. EMBO J 1997,16:6394-6406.

32. Soto GE, Dodson KW, Ogg D, Liu C, Heuser J, Knight S, Kihlberg J,Jones CH, Hultgren SJ: Periplasmic chaperone recognition motif ofsubunits mediates quaternary interactions in the pilus. EMBO J1998, 17:6155-6167.

33. Krasan GP, Sauer FG, Cutter D, Farley MM, Gilsdorf JR, Hultgren SJ, • St Geme JW: Evidence for donor strand complementation in the

biogenesis of Haemophilus influenzae hemagglutinating pili. MolMicrobiol 2000, 35:1335-1347.

Mutagenesis studies identify residues important in chaperone–subunit andsubunit–subunit interactions during Hif pilus biogenesis. This paper pre-sents evidence that donor strand complementation and exchange play rolesin the assembly of these pili, which, unlike other well-characterized pili, aredouble-stranded structures.

34. Barnhart MM, Pinkner JS, Soto GE, Sauer FG, Langermann S, •• Waksman G, Frieden C, Hultgren SJ: PapD-like chaperones provide

the missing information for folding of pilin proteins. Proc NatlAcad Sci USA 2000, 97:7709-7714.

The authors demonstrate that providing the missing strand of the pilindomain of FimH by fusing the N-terminal extension of FimG to the C termi-nus of FimH produces a molecule (donor-strand-complemented FimH ordscFimH) that, unlike wild-type FimH, is stably expressed in the periplasm inthe absence of the chaperone. In an in vitro folding assay, denatureddscFimH, but not wild-type FimH, refolded into a native mannose-bindingconformation in the absence of the chaperone. Denatured wild-type FimHdid refold in the presence of the chaperone to produce soluble FimC–FimHcomplexes. It is proposed that the chaperone can facilitate subunit foldingby providing steric information in the form of its G1 strand.

35. Kuehn MJ, Normark S, Hultgren SJ: Immunoglobulin-like PapDchaperone caps and uncaps interactive surfaces of nascentlytranslocated pilus subunits. Proc Natl Acad Sci USA 1991,88:10586-10590.

36. Bullitt E, Jones CH, Striker R, Soto G, Jacob-Dubuisson F, Pinkner J,Wick MJ, Makowski L, Hultgren SJ: Development of pilusorganelle subassemblies in vitro depends on chaperoneuncapping of a beta zipper. Proc Natl Acad Sci USA 1996,93:12890-12895.

37. Kuehn MJ, Ogg DJ, Kihlberg J, Slonim LN, Flemmer K, Bergfors T,Hultgren SJ: Structural basis of pilus subunit recognition by thePapD chaperone. Science 1993, 262:1234-1241.

38. Hung DL, Pinkner JS, Knight SD, Hultgren SJ: Structural basis of • chaperone self-capping in P pilus biogenesis. Proc Natl Acad Sci

USA 1999, 96:8178-8183.The PapD chaperone can form a dimer, capping its own subunit-interactingsurface. The authors present crystal structures of two PapD dimers.

39. Dodson KW, Jacob-Dubuisson F, Striker RT, Hultgren SJ: Outermembrane PapC molecular usher discriminately recognizesperiplasmic chaperone-pilus subunit complexes. Proc Natl AcadSci USA 1993, 90:3670-3674.

40. Thanassi DG, Saulino ET, Lombardo M-J, Roth R, Heuser J,Hultgren SJ: The PapC usher forms an oligomeric channel:implications for pilus biogenesis across the outer membrane.Proc Natl Acad Sci USA 1998, 95:3146-3151.

41. Saulino ET, Thanassi DG, Pinkner JS, Hultgren SJ: Ramifications ofkinetic partitioning on usher-mediated pilus biogenesis. EMBO J1998, 17:2177-2185.

42. Holmgren A, Branden C-I: Crystal structure of chaperone proteinPapD reveals an immunoglobulin fold. Nature 1989,342:248-251.

43. Pellecchia M, Guntert P, Glockshuber R, Wuthrich K: NMR solution • structure of the periplasmic chaperone FimC. Nat Struct Biol

1998, 5:885-890.The NMR structure of the FimC chaperone, which directs type 1 pilusassembly, reveals that it shares the same overall fold as PapD. Subtledifferences between the two structures are identified.

44. Hung DL, Knight SD, Hultgren SJ: Probing conserved surfaces onPapD. Mol Microbiol 1999, 31:773-783.

45. Pellecchia M, Sebbel P, Hermanns U, Wuthrich K, Glockshuber R: •• Pilus chaperone FimC-adhesin FimH interactions mapped by

TROSY-NMR. Nat Struct Biol 1999, 6:336-339.The mapping of the surfaces of the FimC chaperone that interact with FimHin the soluble chaperone–adhesin complex. The interacting surfaces areessentially limited to the N-terminal domain, but include regions beyondthose in the G1 strand that were already known.

46. Choudhury D, Thompson A, Stojanoff V, Langermann S, Pinkner J, •• Hultgren SJ, Knight SD: X-ray structure of the FimC-FimH

chaperone-adhesin complex from uropathogenic Escherichia coli.Science 1999, 285:1061-1066.

This paper, together with [47••], reveals the donor strand complementa-tion mechanism of chaperone–subunit interaction. In these two crystalstructures, the chaperone donates its G1 strand to complete the Ig foldof the subunit, simultaneously stabilizing it and capping its interactivesurface. The two papers present a donor strand exchange model of pilusbiogenesis in which the N-terminal extension of one subunit displacesthe chaperone G1 strand during assembly and completes the Ig fold ofits neighbor in the mature organelle. In the FimC–FimH structure, a puta-tive mannose-binding pocket was identified in the receptor-bindingdomain of FimH.

47. Sauer FG, Futterer K, Pinkner JS, Dodson K, Hultgren SJ, •• Waksman G: Structural basis of chaperone function and pilus

biogenesis. Science 1999, 285:1058-1061.See annotation to [46••].

48. Striker R, Jacob-Dubuisson F, Frieden C, Hultgren SJ: Stablefiber-forming and nonfiber-forming chaperone-subunitcomplexes in pilus biogenesis. J Biol Chem 1994,269:12233-12239.

49. Sauer FG, Knight SD, Waksman G, Hultgren SJ: PapD-likechaperones and pilus biogenesis. Sem Cell Dev Biol 2000,11:27-34.

50. Minor DL Jr, Kim PS: Context is a major determinant of ββ-sheetpropensity. Nature 1994, 371:264-267.

51. Sokurenko EV, Chesnokova V, Doyle RJ, Hasty DL: Diversity of theEscherichia coli type 1 fimbrial lectin. Differential binding tomannosides and uroepithelial cells. J Biol Chem 1997,272:17880-17886.

52. Sokurenko EV, Chesnokova V, Dykhuizen DE, Ofek I, Wu XR, •• Krogfelt KA, Struve C, Schembri MA, Hasty DL: Pathogenic

adaptation of Escherichia coli by natural variation of the FimHadhesin. Proc Natl Acad Sci USA 1998, 95:8922-8926.

Naturally occurring FimH variants bind monomannose with either high orlow affinity. These variants are found predominantly in uropathogenic andcommensal strains, respectively. The authors present evidence that high-affinity monomannose binding facilitates colonization of the urinary tract,whereas low-affinity monomannose binding facilitates colonization of theoral cavity. Thus, it is proposed that FimH polymorphism arises by aselection mechanism that balances these two advantages and, hence,that genetic variation in an originally commensal trait can produce a virulence factor.

53. Mulvey MA, Lopez-Boado YS, Wilson CL, Roth R, Parks WC, •• Heuser J, Hultgren SJ: Induction and evasion of host defenses by

type 1-piliated uropathogenic Escherichia coli. Science 1998,282:1494-1497.

The authors present high-resolution electron microscopy images of the inti-mate attachment of uropathogenic E. coli to bladder epithelial cells mediatedby type 1 pili. The pili make contact with the uroplakin array that coats thelumenal surface of these cells. Infection by uropathogenic E. coli leads to thedeath and exfoliation of the epithelial cells by an apoptotic-like mechanism;however, some bacteria circumvent this innate host defense mechanism byinvading the now exposed underlying cells.

Chaperone-assisted pilus assembly and bacterial attachment Sauer et al. 555

54. Martinez JJ, Mulvey MA, Schilling JD, Pinkner JS, Hultgren SJ: Type 1 •• pilus-mediated bacterial invasion of bladder epithelial cells.

EMBO J 2000, 19:1-10.This paper demonstrates that type 1 pili can mediate the invasion of humancultured bladder epithelial cells. The authors begin to elucidate the molecu-lar details of this process, which involves host actin polymerization, phos-phorylation by non-Src-family tyrosine kinases, phosphatidylinositol 3-kinase(PI 3-kinase) activation and the formation of complexes between PI 3-kinaseand focal adhesion kinase, and between vinculin and α-actinin.

55. Malaviya R, Ikeda T, Ross E, Abraham SN: Mast cell modulation ofneutrophil influx and bacterial clearance at sites of infectionthrough TNF-αα. Nature 1996, 381:77-80.

56. Baorto DM, Gao Z, Malaviya R, Dustin M, van der Merwe A, Lublin DM,Abraham SN: Survival of FimH-expressing enterobacteria inmacrophages relies on glycolipid traffic. Nature 1997, 389:636-639.

57. Malaviya R, Gao Z, Thankavel K, van der Merwe PA, Abraham SN: Themast cell tumor necrosis factor αα response to FimH-expressingEscherichia coli is mediated by the glycosylphosphatidylinositol-anchored molecule CD48. Proc Natl Acad Sci USA 1999,96:8110-8115.

58. Pouttu R, Puustinen T, Virkola R, Hacker J, Klemm P, Korhonen TK:Amino acid residue Ala-62 in the FimH fimbrial adhesin is criticalfor the adhesiveness of meningitis-associated Escherichia coli tocollagens. Mol Microbiol 1999, 31:1747-1757.

59. Pratt LA, Kolter R: Genetic analysis of Escherichia coli biofilm • formation: roles of flagella, motility, chemotaxis and type 1 pili.

Mol Microbiol 1998, 30:285-293.This paper presents evidence that type 1 pili are important in the initialattachment event in biofilm formation, whereas cell motility plays roles inboth initial attachment and movement along the surface during this process.

60. Schembri MA, Sokurenko EV, Klemm P: Functional flexibility of the •• FimH adhesin: insights from a random mutant library. Infect

Immun 2000, 68:2638-2646.This paper presents the effects of random mutagenesis on the monoman-nose and oligomannose binding properties of FimH. Monomannose bindingwas, in general, more sensitive to amino acid changes than oligomannosebinding, mirroring the situation seen in natural FimH variants (see [52••]). Amutation in Pro49, near the putative mannose-binding pocket, abolishedmonomannose binding, whereas mutations that increased monomannosebinding mapped to regions outside this pocket.

61. Schembri MA, Pallesen L, Connell H, Hasty DL, Klemm P: Linkerinsertion analysis of the FimH adhesin of type 1 fimbriae in anEscherichia coli fimH-null background. FEMS Microbiol Lett 1996,137:257-263.

62. Chandra NR, Ramachandraiah G, Bachhawat K, Dam TK, Surolia A,Vijayan M: Crystal structure of a dimeric mannose-specific

agglutinin from garlic: quaternary association and carbohydratespecificity. J Mol Biol 1999, 285:1157-1168.

63. Fraser GM, Bennett JQ, Hughes C: Substrate-specific binding of •• hook-associated proteins by FlgN and FliT, putative chaperones

for flagellum assembly. Mol Microbiol 1999, 32:569-580.Flagellar axial proteins, including hook-associated proteins (HAPs), are pre-dicted to have amphipathic helices at their C termini. The cytoplasmic FlgNprotein specifically binds the HAPs FlgK and FlgL by their C-terminalhelices and facilitates efficient flagellum assembly; FliT binds the filamentcap protein FliD by its C-terminal helix. It is proposed that FlgN and FliT arespecific chaperones that prevent premature oligomerization of HAPs duringflagellum biogenesis.

64. Minamino T, Macnab RM: Components of the Salmonella flagellar •• export apparatus and classification of export substrates.

J Bacteriol 1999, 181:1388-1394.Eight general components — FlhA, FlhB, FliH, FliI, FliO, FliP, FliQ andFliR — of the flagellar export apparatus are identified. The authors demon-strate that FliE is required for the export of some substrates and that FlgJ isrequired for rod assembly. The cytoplasmic protein FliJ is required for theexport of FlgD and FlgE. It is proposed that FliJ acts as chaperone for FlgDand FlgE during flagellum assembly.

65. Hueck C: Type III protein secretion systems in bacterialpathogens of animals and plants. Microbiol Mol Biol Rev 1998,62:379-433.

66. Jackson MW, Day JB, Plano GV: YscB of Yersinia pestis functions •• as a specific chaperone for YopN. J Bacteriol 1998,

180:4912-4921.YopN is required to prevent Yop secretion in Yersinia in the presence of cal-cium and before contact with the host cell. This paper presents evidencethat YscB acts a specific chaperone for YopN.

67. Neyt C, Cornelis GR: Role of SycD, the chaperone of the Yersinia•• Yop translocators YopB and YopD. Mol Microbiol 1999, 31:143-156.The authors demonstrate that SycD binds the putative pore-forming proteinYopB, as well as YopB and YopD together. It was previously known thatSycD binds the YopD translocator. The SycD-binding site on YopB could notbe localized to a discrete linear polypeptide segment. It is proposed thatSycD acts as a chaperone to prevent premature association of theYopB–YopD complex with LcrV.

68. Ellis RJ, Hemmingsen SM: Molecular chaperones: proteinsessential for the biogenesis of some macromolecular structures.Trends Biochem Sci 1989, 14:339-342.

69. Shin JS, Gao Z, Abraham SN: Involvement of cellular caveolae in •• bacterial entry into mast cells. Science 2000, 289:758-788.The authors present evidence that FimH-expressing E. coli gain entry intomast cells by a distinct, caveolae-mediated pathway. They suggest that thebacteria use this pathway to elude the normal phagocytic killing mechanismof mast cells and remain viable inside these cells.

556 Carbohydrates and glycoconjugates