type 1 fimbrial adhesin fimh elicits an immune response ... · type 1 fimbrial adhesin fimh elicits...

INFECTION AND IMMUNITY, Oct. 2011, p. 3895–3904 Vol. 79, No. 100019-9567/11/$12.00 doi:10.1128/IAI.05169-11Copyright © 2011, American Society for Microbiology. All Rights Reserved.

Type 1 Fimbrial Adhesin FimH Elicits an Immune Response ThatEnhances Cell Adhesion of Escherichia coli�†

Veronika Tchesnokova,1 Pavel Aprikian,1 Dagmara Kisiela,1 Sarah Gowey,1 Natalia Korotkova,1Wendy Thomas,2 and Evgeni Sokurenko1*

Department of Microbiology1 and Department of Bioengineering,2 University of Washington, Seattle, Washington 98195

Received 1 April 2011/Returned for modification 13 May 2011/Accepted 7 July 2011

Escherichia coli causes about 90% of urinary tract infections (UTI), and more than 95% of all UTI-causingE. coli express type 1 fimbriae. The fimbrial tip-positioned adhesive protein FimH utilizes a shear force-enhanced, so-called catch-bond mechanism of interaction with its receptor, mannose, where the lectin domainof FimH shifts from a low- to a high-affinity conformation upon separation from the anchoring pilin domain.Here, we show that immunization with the lectin domain induces antibodies that exclusively or predominantlyrecognize only the high-affinity conformation. In the lectin domain, we identified four high-affinity-specificepitopes, all positioned away from the mannose-binding pocket, which are recognized by 20 separate clones ofmonoclonal antibody. None of the monoclonal or polyclonal antibodies against the lectin domain inhibited theadhesive function. On the contrary, the antibodies enhanced FimH-mediated binding to mannosylated ligandsand increased by severalfold bacterial adhesion to urothelial cells. Furthermore, by natural conversion fromthe high- to the low-affinity state, FimH adhesin was able to shed the antibodies bound to it. When wholefimbriae were used, the antifimbrial immune serum that contained a significant amount of antibodies againstthe lectin domain of FimH was also able to enhance FimH-mediated binding. Thus, bacterial adhesins (or othersurface antigens) with the ability to switch between alternative conformations have the potential to induce aconformation-specific immune response that has a function-enhancing rather than -inhibiting impact on theprotein. These observations have implications for the development of adhesin-specific vaccines and may serveas a paradigm for antibody-mediated enhancement of pathogen binding.

Bacterial adhesion is the first step in the successful estab-lishment of infection by pathogens, and bacterial adhesins havefor a long time been prime candidates as targets for antibac-terial therapeutics such as specific-ligand-like inhibitors andvaccines. Type 1 fimbriae of Escherichia coli are filamentousappendages that confer bacterial binding to glycoproteins withterminally exposed mannose (9). The type 1 fimbrial adhesin isexpressed by more than 90% of uropathogenic strains of E. coliand has been demonstrated to be crucial for the establishmentof urinary tract infections (UTIs) in mice (18, 47). Its role inestablishing infection in humans is less clear (6, 14). Here, weshow that the adhesin’s capability to dynamically switch be-tween alternative conformations significantly affects its anti-genic properties and the functional impact of the antibodybinding.

Type 1 fimbriae are 0.5- to 1.5-�m-long structures that areassembled via the chaperone-usher pathway. Mannose-specificbinding is mediated by a 30-kDa adhesive protein, FimH, lo-calized in a fimbrial tip structure which also includes the minorsubunits FimG and FimF and is attached to the fimbrial rodcomposed of the polymerized major protein, FimA. FimHconsists of two domains connected by a short linker chain: thelectin domain (LD), with the mannose-binding pocket on itsdistal end, and the pilin domain (PD), connecting FimH to

FimG (11, 27). Both domains have Ig-like �-sandwich folds.Unlike the pilin subunit FimA, which is structurally highlyvariable, the primary structure of the adhesin FimH is 99%conserved (49).

It was found recently that FimH can exist in two alternativeconformations, with LD and PD either separated or closelyinteracting with one another (27). When the domains interact,the LD assumes a more twisted, compressed conformation.The conformational change in LD has a profound functionalimpact. In the interacting-domain conformation, the mannose-binding pocket is rather open (loose), while in the separated-domain conformation, the mannose-binding pocket closes (Fig.1). As a result, the affinity for mannose of the interacting-domainconformation is much lower than that of the separated-domainconformation (KD [equilibrium dissociation constant] � 300 �10�6 M versus KD � 1.2 � 10�6 M, respectively) (1). Becausethe association of PD with LD affects the mannose-bindingpocket located on the opposite part of LD, such regulation ofFimH affinity is allosteric. As allosteric regulation is reciprocalin nature, binding of the ligand to the loose pocket in theinteracting-domain, twisted conformation of LD facilitates thetightening of the binding pocket as well as untwisting of LDand separation of the domains. Thus, under equilibrium con-ditions and in the absence of the mannose ligand, the confor-mation of LD is heavily shifted toward the low-affinity, inter-acting-domain state (LAS). In contrast, under equilibriumconditions in the presence of mannose (soluble or surfaceattached), there is a shift in the conformation of LD toward thehigh-affinity, separated-domain state (HAS).

The allosteric properties of the FimH adhesin are the basis

* Corresponding author. Mailing address: 1705 Pacific St. NE, HSB,Box 357242, Seattle, WA 98195. Phone: (206) 685-2162. Fax: (206)543-8297. E-mail: [email protected].

† Supplemental material for this article may be found at http://iai.asm.org/.

� Published ahead of print on 18 July 2011.

3895

on October 12, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

of shear-enhanced adhesion of E. coli mediated by type 1fimbriae. Indeed, it has been shown that FimH-expressing bac-teria bind to mannosylated surfaces much more strongly underflow-induced shear than under static conditions (40). Single-molecule force spectrometry experiments have also demon-strated that the application of a sufficient level of tensile me-chanical force shifts FimH from a weakly to a strongly bindingmode to mannose (51). The structural basis of shear-enhancedand mechanically enhanced adhesion is the same—FimH do-mains are separated and sustained in this conformation undertensile force, thus switching LAS to HAS and preventing thelatter from shifting back. Such force-enhanced or so-calledcatch-bond properties of FimH could be relatively widespread,as they have been observed in other bacterial adhesins (e.g., Pand CFA/I fimbriae), as well as in eukaryotic adhesive pro-teins, such as P-/L-selectins, von Willebrand factor, and integ-rins (for a review, see reference 36).

While extensive studies have focused on comparing the ad-hesive properties of different conformational states of FimHor LD and on deciphering the role of mechanical force in themodulation of FimH function, the effects of the conforma-tional changes in LD on the immune response to fimbrialantigens has not been investigated. We have previously de-scribed a monoclonal antibody, MAb21, raised against purifiedLD, that has been shown to recognize a discontinuous (con-

formational) epitope in the HAS but not LAS conformation ofthe domain (27, 37). However, the extent to which the alter-native conformations of the adhesin affect its immunogenicityin general has not been investigated. Considering that FimHhas been a major candidate for the development of a vaccineagainst uropathogenic E. coli, such an investigation would havesignificant value for the development of preventive and ther-apeutic strategies for UTI.

In this study, we examined multiple polyclonal antibody (PAb)and monoclonal antibody preparations raised against LD ofFimH and determined that the immune response is heavilyskewed toward antibodies specific against only the HAS con-formation of FimH. Such antibodies, instead of inhibitingFimH function, significantly enhance its adhesive properties.This raises an important issue for understanding the physio-logical impact of the immune response against adhesins andother proteins that have the ability to dynamically shift theirconformations.

MATERIALS AND METHODS

Strains. The recombinant strains utilized here were described previously (35).The allele encoding the FimH:wt (wild-type) variant was derived from E. coliMG1655 and is identical to the allele encoding FimH in E. coli J96 that was usedto determine the X-ray crystallographic structure (11). For more details, seeMethods in the supplemental material.

Site-directed mutagenesis of fimH. Mutations were introduced into thefimH gene on the pGB2-24 plasmid by site-directed mutagenesis using theQuikChange kit as directed by the manufacturer (Stratagene, La Jolla, CA).Constructs containing the mutations were identified by sequencing the fimHgene.

Antibodies. The lectin domain of FimH:wt (amino acids 1 to 160) was used toraise 14 mouse polyclonal antisera, 6 rabbit polyclonal antisera, and 20 mousemonoclonal antibodies. Monoclonal antibody clones were selected as follows:multiple samples of hybridoma culture medium were sent to us by the providerand were tested in enzyme-linked immunosorbent assay (ELISA) for binding toimmobilized LD both in the absence and in the presence of antibodies. Polyclonalantiserum was obtained from all animals that were immunized. Polyclonal rabbitanti-FimHt serum raised against a naturally occurring mannose-binding truncateFimH (25) was kindly provided by Scott Hultgren (Washington University, St. Louis,MO). For more details, see Methods in the supplemental material.

Protein purification. Fimbriae were purified from recombinant E. coli strainsexpressing type 1 fimbriae with different FimH proteins as described previously(37). Lectin domain (FimH LD) expression and purification was performed asdescribed previously (1).

Antibody testing. Antibody binding to FimH was measured in ELISAs asdescribed previously (37) and in Methods in the supplemental material. MAbsand PAbs were screened for differential binding to immobilized fimbriae (0.4mg/ml) or LD (0.05 mg/ml) in the absence or presence of 1% �-methyl-D-mannopyranoside (hereinafter termed mannose) in ELISAs. Determination ofthe H/LR (HAS-to-LAS ratio) was performed by ELISA as well. Mapping ofepitopes for monoclonal antibodies was performed by ELISA as described pre-viously (37). The results of epitope scanning are summarized in Table S2 and Fig.S4 in the supplemental material. Horseradish peroxidase (HRP) binding tofimbriae in the absence and presence of antibodies was measured for all mono-clonal and all rabbit polyclonal antibodies as described previously (37). For moredetails, see Methods in the supplemental material.

Adhesion assays. Bacterial strains at an A600 of 0.3 were allowed to adhere toconfluent monolayers of T24 human bladder epithelial cells (ATCC HTB-4) for2 h at 37°C under 5% CO2 as described previously (23, 45) in the absence orpresence of 5 to 10 �g/ml of purified antibodies or Fab fragments or 1%mannose. Adhesion was evaluated by light microscopy of stained slides, countingthe numbers of bacteria and T24 cells in 10 to 20 fields of view and calculatingthe average number of bound bacteria per cell. For details, see Methods in thesupplemental material.

Protein structure analysis. Analysis of the spatial distribution of epitoperesidues on FimH LD (1TR7.pdb) and FimHGFFC tip (3JWN.pdb) was per-formed using PyMol software (DeLano Scientific LLC, San Francisco, CA). The

FIG. 1. Alternative conformations of FimH. On the left, FimHfrom FimHGFFC-TIP (3JWN.pdb) is in a low-affinity conformationwith LD and PD closely interacting and the linker chain buried in thedomain-domain interface; LD is shorter and wider (gray rectangleoverlay), and the mannose-binding site is wide open (triangle at top).On the right, FimH from FimCH complex (1QUN.pdb) is in a high-affinity, elongated form (gray rectangle overlaid over LD), with do-mains separated from each other and the linker chain extended; thebinding pocket is narrowed (triangle at top). The triangles represent-ing the binding pocket have Ile13, Tyr48, and Asp140 at the corners,with the sizes of the sides correlating with the actual distances betweenthe respective C� atoms.

3896 TCHESNOKOVA ET AL. INFECT. IMMUN.


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

structures were aligned using reference positions 36, 74, and 127 as describedpreviously (27).

The measurement of the root mean square deviation (RMSD) differences ofMAb epitopes is described in detail in the supplemental material under “Mea-suring RMSD differences of MAb epitope.” Briefly, the two instances of FimHlow-affinity structures are compared with 12 instances of FimH high-affinitystructures in VMD (Visual Molecular Dynamics software) to measure theRMSD of C� atoms of all epitope residues for each MAb group. The RMSD iscalculated for each LAS versus each HAS structure, as well for each pair ofstructures that are either both HAS or both LAS to calculate the “between-state”and “within-state” RMSD.

RESULTS

Polyclonal antiserum raised against isolated LD is primar-ily HAS specific. Rabbit polyclonal antiserum PAb280 raisedagainst LD was evaluated for the ability to recognize HASversus LAS conformations by analyzing graded dose-responsecurves of the antiserum in ELISAs. We employed purified type1 fimbriae immobilized on the plastic surface as the antigen. Infimbriae in the absence of mannose, FimH is in the native LASconformation, characterized by LD docked to PD. The HAScan be induced in fimbrial FimH by the addition of solublemannose. Thus, we compared antiserum binding in the absenceand presence of 1% �-methyl-D-mannopyranoside (hereinaftertermed mannose). In the absence of mannose, the effectivedilution of the antisera for half-maximal binding (50% effectivedose [ED50]) was �1:1,400, whereas in the presence of man-nose, the ED50 was �1:7,400 (Fig. 2A and Table 1), suggestingthat the polyclonal antibodies recognize HAS more than 5times better than LAS (HAS-/LAS-binding Ratio [H/LR] �5.4). The binding of the preimmune antiserum to fimbriae wasvery low and not affected by mannose (data not shown).

MAb21, a monoclonal antibody against purified LD specificfor the high-affinity state of FimH (37), displayed a bindingpattern to fimbriae similar to that of the PAb280 antiserum.MAb21 bound much better to fimbriae in the presence ofmannose than in its absence (Fig. 2B), with the difference

being even more pronounced than with PAb280 (H/LR �113.0). Both 1% D-mannose and �-methyl-D-mannopyranosideenhanced the MAb21 binding, whereas �-methyl-D-glucopy-ranoside lacked this effect on fimbrial FimH (see Fig. S1A inthe supplemental material). The sugar moieties of immuno-globulins did not interfere with the described effect since bothintact and mannosidase-treated MAb21 exhibited the samelevel of binding to FimH in the absence or presence of solublemannose (see Fig. S1B in the supplemental material). An ex-

FIG. 2. Polyclonal antiserum PAb280 recognizes the high-affinity form of FimH much better than the low-affinity form, similar to monoclonalantibody MAb21. PAb280 (A) and MAb21 (B) were titrated in ELISAs over immobilized purified fimbriae in the absence (open circles) orpresence (closed circles) of mannose (man). Each data point represents the average of triplicate repeats. Data were fitted in GraphPad softwareto a sigmoidal dose-response curve with variable slope. The half-maximal dilution (ED50) was determined both in the absence and in the presenceof mannose to calculate the HAS-to-LAS ratio (H/LR). PAb280 and MAb21 titrations were simultaneously carried out on immobilized LD todetermine the ED50 values and H/LRs (see Table S1 in the supplemental material). OD 650 nm, optical density at 650 nm.

TABLE 1. Polyclonal antisera against FimH LD preferentiallyrecognize FimH in the high-affinity state

Serum or MAb

ED50 in thea:

H/LRbAbsence of

mannosePresence of

mannose

PAb280 1,388 � 219 7,422 � 1,104 5.34PAb279 498 � 72 5,820 � 1,012 11.35PAb33 2,548 � 46 2,875 � 298 1.39c

PAb35 102 � 16 231 � 21 2.07anti-FimHt 18,188 � 514 36,652 � 2,951 2.01PAb14 183 � 38 2,474 � 348 13.74PAb15 259 � 83 4,411 � 938 17.68PAb16 1,052 � 426 5,518 � 873 5.87PAb17 10 � 1.8 138 � 33 13.97PAb18 30 � 3.9 378 � 81 12.41PAb19 34 � 5.1 239 � 50 6.92PAb20 148 � 42 1,349 � 234 9.46PAb49 82,229 � 10,343 65,095 � 5,780 0.8 NSPAb50 13,247 � 705 11,887 � 202 0.8 NSanti-PD 9,472 � 100 7,590 � 134 0.8 NS

a The ED50 values of polyclonal antisera recognizing isolated LD in the ab-sence and the presence of mannose were determined as described in Materialsand Methods.

b The high- versus low-affinity-state ratio (H/LR) was calculated for eachserum or MAb as the ratio of the ED50 in the presence versus the ED50 in theabsence of mannose. The H/LR was significantly greater than 1.0 unless indi-cated otherwise. NS, not significant.

c P � 0.29.

VOL. 79, 2011 ANTI-FimH ANTIBODIES ENHANCE ADHESION 3897


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

cess of nonspecific immunoglobulins did not have any effect onthe binding of polyclonal anti-FimH antiserum (PAb280) toFimH (see Fig. S1 in the supplemental material).

In contrast to the binding to fimbriae, the presence of man-nose did not increase (and actually slightly inhibited; see TableS1 in the supplemental material) the binding of both MAb21and PAb280 to purified LD, which is sustained in the HASconformation with or without mannose because of the absenceof PD (1, 27, 37).

Furthermore, we tested 7 mouse and 4 additional rabbitpolyclonal antisera raised against purified LD for the ability todifferentially recognize purified fimbriae in the presence andabsence of soluble mannose. The antisera were obtained usingdifferent commercial suppliers and different batches of purifiedLD. All antisera recognized purified LD similarly with or with-out mannose, again with slight inhibition of antibody bindingby mannose (see Table S1 in the supplemental material). Atthe same time, all but one of the anti-LD sera recognized themannose-induced HAS conformation of the fimbrial FimHsignificantly better than the LAS conformation, with H/LRsranging from 2.01 to 17.67 (Table 1). Though there was over100-fold variability in the absolute strength of the antibodyresponses of the different sera, no correlation between H/LRand titer level was observed (see Fig. S2 in the supplementalmaterial). The only serum sample with no significant differencein binding with or without mannose was rabbit antiserumPAb33, whose H/LR was 1.39. Similarly, fimbrial binding ofantibodies raised against purified PD also was not affected bymannose (Table 1).

Thus, a strong HAS-specific antibody response is inducedupon immunization with purified LD of FimH.

All monoclonal antibodies raised against LD are HAS spe-cific. We obtained 20 monoclonal antibodies (MAbs) raisedagainst purified LD and derived from six different mice. AllMAbs recognized purified LD equally well in the absence orpresence of mannose (see Fig. S3A in the supplemental ma-terial). At the same time, all MAbs recognized fimbrial FimHmuch better in the presence of mannose (Table 2; also see Fig.S3B in the supplemental material). Though the original titersof some MAbs were too low to reach saturation level, for thoseantibodies that were of sufficiently high concentration, theH/LRs ranged from 13.0 to 390.0 (Table 2; also see Fig. S3C inthe supplemental material). Thus, all of the MAbs obtainedappeared to be specific for the HAS conformation of LD.

We tested the antibodies on a panel of fimbriae with variousmutations in LD to identify epitopes (see Fig. S4 and Table S2in the supplemental material). We found that the MAbs fellinto four groups based on the locations of structural mutationsin FimH that severely diminished antibody binding.

The group A antibodies, which include 8 antibody clones(Table 2), recognize an epitope comprised of residues 26, 29,and 153 to 157 (Fig. 3A, low-affinity state, and B, high-affinitystate). This epitope is also recognized by the previously char-acterized MAb21 clone and is positioned at the bottom of thebeta-barrel-shaped LD facing the pilin domain, with most ofthe epitope buried in the interdomain interface (Fig. 3, darkerresidues). Thus, the group A epitope is far distant from man-nose-binding-site residues (Fig. 3, yellow). This epitope hasdistinctly different conformations in the high- and low-affinitystates of LD, with the between-state C� atoms root mean

square deviation (C� RMSD) being 3.25 Å, which is signifi-cantly higher than the within-state C� RMSD of 0.26 Å, basedon comparison of available structures of alternative FimH con-formations. The H/LRs for group A antibodies ranged from 31to 308.

The group B antibodies (5 clones) recognize an epitope thatincludes residues 59, 60, 88, 132, 141, and 143 (Fig. 3C and D).This epitope is located mostly on the side of the LD beta-barrelthat does not overlap with the interdomain interface. It isrelatively close to the mannose-binding site, though it does notshare any residues with the latter. The between-state C�RMSD is 0.82 Å, versus 0.21 Å for the within-state C� RMSDvalue. The H/LRs of the group B epitope antibodies rangedfrom 13 to 94.

The group C antibodies (6 clones) recognize an epitopepartially overlapping with the group B epitope, sharing resi-dues 59, 60, 141, and 143 but not 88 and 132 and having, inaddition, residues 57, 62, 89, 130, and 145 (Fig. 3E and F). Likethe group B epitope, the group C epitope is relatively close tobut does not overlap the mannose-binding site. The inter-stateC� RMSD is 1.42 Å, versus the within-state C� RMSD of 0.24Å. The H/LRs of the group C epitope ranged from 149 to 386.

Finally, group D is represented by only one antibody clone(MAb34) that recognizes an epitope including residues 55, 78,

TABLE 2. Monoclonal antibodies against FimH LD preferentiallyrecognize FimH in the high-affinity state

Group and MAb

ED50 in thea:

H/LRbAbsence of

mannosePresence of

mannose

AMAb4 52.2 � 16.4 3,308 � 325 63MAb5 93.0 � 5.5 5,876 � 306 63MAb10 1.1 � 0.12 34 � 1.5 31MAb11 1.0 32 � 2.6 33MAb21 486.1 � 27.2 57,867 � 2,359 119MAb24 116.3 � 43.3 35,803 � 4,066 308MAb35 167.9 � 25.5 16,509 � 3,659 98MAb36 163.6 � 42.7 16,662 � 805 102

BMAb1 38.1 � 4.7 503 � 36 13MAb12 49.7 � 4.6 2,069 � 820 42MAb13 96.0 � 18.9 9,047 � 2,037 94MAb25 70.3 � 14.9 1,779 � 508 25MAb37 182.0 � 73.6 6,656 � 1,919 37MAb28 5.0 � 0.5 738 � 53 149

CMAb29 5.9 � 1.3 1,040 � 86 176MAb30 1.4 � 0.16 533 � 74 386MAb31 2.5 � 0.39 619 � 51 252MAb32 1c 79 � 8.3 80MAb33 3.5 � 0.32 793 � 65 229

DMAb34 201.6 � 78.8 49,167 � 10,794 244

a The ED50 values of MAbs recognizing isolated LD in the absence and thepresence of mannose were determined as described in Materials and Methods.

b The high- versus low-affinity-state ratio (H/LR) was calculated for each MAbas the ratio of the ED50 in the presence versus the ED50 in the absence ofmannose. The H/LR was significantly greater than 1.0 unless indicated otherwise.

c This MAb did not reach half-maximum of the response even when usedundiluted.



http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

91, 92, and 93 (Fig. 3G and H). The epitope is located on theside of the LD beta-barrel, relatively distant from both theinterdomain interface and the mannose-binding site. It has arelatively low but still significant difference in the inter-stateversus the within-state C� RMSD (0.336 Å versus 0.177 Å,respectively; P � 0.0007). The H/LR of MAb34 is 244.

Thus, we have indentified the presence of at least 4 distinct3-dimensional conformational epitopes specific for the high-affinity form of LD. None of the epitopes overlapped with themannose-binding pocket of LD, and only the group A epitopeoverlapped with the interface between LD and PD.

Antibodies raised against LD do not inhibit but rather in-crease mannose-specific binding. None of the MAbs interfered

with the mannose-specific binding of HRP to isolated LDimmobilized on plastic surfaces (see Fig. S5A in the supple-mental material). However, all MAbs increased to various de-grees the mannose-specific HRP binding to fimbria-incorpo-rated FimH (Fig. 4A). The strongest increase in the bindingwas mediated by the group A epitope-specific antibodies.

Similar to MAbs, out of all polyclonal rabbit antisera, noneinhibited HRP binding to isolated LD (see Fig. S5B in thesupplemental material), including antisera with a predominantresponse against the high-affinity state (PAb280, H/LR 5.4) orwith no significant differences in the ability to recognize HASover LAS (PAb33). When HRP binding to fimbrial FimH wastested, a binding increase was noted for all antisera with an

FIG. 3. FimH adhesin contains four distinct LIBS epitopes that are recognized by monoclonal antibodies. Epitope residues (in red), binding-site residues (in yellow), and interdomain-bonding residues (in dark green for LD and dark cyan for PD) are mapped on crystal structures oftip-incorporated LD (A, C, E, G) and isolated LD (B, D, F, H). Positions 155 and 157 in the group A epitope participate in interdomain bondingas well (brown in panel A).



http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

H/LR significantly higher than 1, including PAb280 and anti-FimHt PAb (H/LR 2.01), shown in Fig. 4B. Only PAb33, whichshowed no preferential recognition of the high-affinity state ofLD, had no detectable effect on HRP binding by fimbriae(Fig. 4B).

We also tested E. coli adhesion to a monolayer of T24urinary bladder epithelial cells in the presence and absence ofpolyclonal antisera (PAb280 and PAb33) and MAb21 antibod-ies. PAb280 enhanced bacterial adherence to T24 uroepithelialcells, as did MAb21, both in a mannose-inhibitable manner(Fig. 4C). Furthermore, bacterial cell adhesion was enhancedby purified Fab fragments of PAb280. In contrast, PAb280preimmune antiserum did not affect bacterial adhesion. Poly-clonal antiserum PAb33 that does not have a measurable poolof high-affinity-specific antibodies also failed to enhance bac-terial adherence, although it did not inhibit the bacterial ad-herence either.

Thus, none of the antibodies raised against LD demonstratea notable inhibitory property against the mannose-specificbinding of LD. In contrast, most of the antibodies substantiallyincrease the binding property of FimH, including adhesion touroepithelium.

Antibodies raised against whole fimbriae enhance FimH-mediated binding. We immunized two rabbits with purifiedwhole fimbriae. Both immune antisera were able to recognizewhole fimbriae (Fig. 5A), although the response from one rabbit(PAb49) appeared to be stronger than the response from theother one (PAb50). PAb50, however, was able to efficiently rec-ognize isolated LD as well, unlike PAb49.

When both antisera were tested for the ability to influencethe FimH-mediated binding of soluble HRP, PAb50 but notPAb49 significantly increased the binding (Fig. 5B), i.e., it hadan effect similar to that of the anti-LD antibodies describedabove.

Thus, the immune response toward whole type 1 fimbriaecan elicit significant levels of anti-LD antibodies that have anenhancing effect on FimH binding.

FimH can shed the bound HAS-specific antibody. We de-termined whether the conditions favoring a shift from HAS toLAS would facilitate removal of the antibodies bound to LD inthe HAS conformation. Various MAbs were bound to fimbrial

FimH in the presence of mannose, and the unbound fractionwas removed by washing. Then, a buffer with or without solublemannose was added, followed by additional washes after dif-ferent periods of time and measurement of the MAb fractionremaining bound to FimH. All of the tested antibodies wereshed from fimbrial FimH at a significantly higher rate upon theremoval of mannose than in the presence of mannose (Fig. 6,open symbols), resulting in a more than 50% loss in the case ofC and D epitope-specific antibodies (MAb29 and MAb34, re-spectively) in 15 min. A B epitope-specific antibody (MAb25)required more than 60 min for 50% loss. The complex with anA epitope antibody (MAb21) was relatively stable but alsodropped almost 30% after 2 h upon the removal of mannose.In contrast, the amount of antibodies bound to isolated LDremained the same both in the presence and in the absence ofmannose (Fig. 6, closed symbols).

Thus, the removal of mannose from FimH results in “shed-ding” of the antibodies bound to the HAS conformation of LD.

FIG. 4. HAS-specific antibodies increase FimH-mediated binding to mannosylated substrates. HAS-specific monoclonal (A) and polyclonal(B) antibodies increased binding of HRP to immobilized FimH in whole fimbriae. Averages of four repeats are plotted on the graph. (C) HAS-specific antibodies increase binding of bacteria to uroepithelial cells. Average numbers of bound bacteria per T24 cell (ave/cell) counted in 10 fieldsof view on three slides are plotted on the graph. Error bars indicate standard deviations. Significant differences (P 0.0001) in binding in thepresence versus absence of antibodies are marked (***).

FIG. 5. Polyclonal serum against whole fimbriae which containsanti-LD antibodies enhances FimH-mediated HRP binding. (A) Rab-bit antifimbrial sera (PAb49 and PAb50) were tested for the ability torecognize immobilized fimbriae (1:2,000 dilution of PAb) and LD(1:200 dilution of PAb) in an ELISA. (B) Immobilized fimbriae boundsoluble HRP in the absence and presence of antifimbrial PAb (1:50dilution).



http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

DISCUSSION

Taken together, the data presented here show that immuni-zation with the mannose-binding domain of the FimH adhesininduces antibodies that predominantly (i) recognize only thehigh-affinity domain conformation assumed upon separationfrom the pilin domain; (ii) increase FimH-mediated adhesionrather than inhibiting it; and (iii) can be shed from FimH uponconformation shift. These observations reveal complex struc-tural and functional dynamics of the antigen-antibody interac-tion when the antigen can switch between alternative confor-mations.

Most antigenic epitopes in proteins are conformational (dis-continuous, or 3-dimensional) in nature (46), and there are alarge number of examples where MAbs recognize one confor-mation of a protein and not the other (3, 4, 21, 26). Thus, it wasexpected that upon immunization with purified LD (which is ina stable HAS conformation), a portion of the response wouldbe HAS specific. It was surprising, however, that in almost allsamples of polyclonal antisera, the level of antibody bindingto the mannose-induced HAS conformation of FimH wasso predominant. The high binding of antibodies to fimbrialFimH in the presence of mannose is unlikely to be due toantibody recognizing mannose in the pocket as a hapten,since the addition of mannose does not increase antibodybinding to immobilized LD, which would be expected if thiswere the case. It is not likely to be due to a relatively highavidity of HAS-specific antibodies either but, rather, is likely tobe due to the large pool of HAS-specific antibodies in theantisera. Indeed, all 20 hybridoma clones obtained againstisolated LD were HAS specific. Mapping of the HAS-specificepitopes for the monoclonal antibodies confirmed that they areligand-induced binding sites (LIBS), since all four of them arelocated away from the mannose-binding pocket and have, tovarious extents, different conformations in the HAS and LASforms of LD. Since most MAbs which recognized the samegroup epitope had different H/LRs (meaning different concen-trations of mannose required to induce clone-specific LIBS),they most likely belong to distinct clones rather than to the

same clone. Multiple origins of monoclonal antibodies recog-nizing overlapping discontinuous epitopes have been describedpreviously (17).

While the H/LRs for polyclonal antisera were generallywell above 1, all of them could also recognize fimbrial FimHin the absence of mannose relatively well, especially in com-parison with the monoclonal antibodies. This indicates thatbesides HAS-specific antibodies, the antisera contained anti-bodies which could bind to the LAS conformation, i.e., theywere specific to common epitopes. In any case, HAS-specificand common antibodies failed to inhibit the mannose-spe-cific interaction between LD and HRP when tested, even atincreasingly high concentrations. Evidently none of the pro-tein regions (linear or conformational) of the mannose-binding pocket of LD were sufficiently immunogenic to elicitstrong and/or high-affinity immune responses. Another expla-nation for the absence of antibodies against the binding site ofFimH which would be able to inhibit binding is the possibilitythat upon being injected into the animal, FimH antigen bindsto bulky glycoproteins in the host which sterically prevent theproduction of antibodies against this part of the protein. Toinvestigate this hypothesis, a similar study should be con-ducted, using for immunization FimH where either the bindingpocket is inactivated by point mutation (that presumably doesnot affect the epitope formation) or the binding pocket isblocked by a small, tightly attached ligand.

The observations that the anti-LD immune response doesnot elicit binding-inhibitory antibodies conflict with the resultsof previous studies where type 1 fimbriae-mediated adhesionof bacteria was inhibited by anti-FimH antibodies (24, 25, 30,39, 43, 44). One possible explanation could be differences inthe immunization or antigen preparation protocols. However,in addition to our own preparations of anti-LD antibodies, wealso examined antiserum obtained by another laboratory andused in previous studies (anti-FimHt, a generous gift of ScottHultgren, Washington University, St. Louis) (25). FimHt isa naturally occurring N-terminally truncated FimH proteinwhich is functionally active and contains the entire LD (19, 25).Thus, FimHt is likely to be in the HAS conformation, andindeed, the anti-FimHt antiserum has a significant pool ofanti-HAS antibodies and is able to increase FimH-mediatedbinding of HRP similarly to other anti-LD antiserum samples.We did not test antisera against the FimH/FimC complexwhich also have been proposed to exhibit FimH inhibitoryactivity (24, 30). However, in the FimH/FimC complex, LD isalso in the HAS conformation, identical to isolated LD (likelydue to the domain separation caused by FimC wedging).

Another reason for the apparent discrepancies of our bind-ing inhibition studies with previously published reports couldbe differences in the test conditions. In our study, we probedthe function of purified LD or fimbriae that are immobilizedon a plastic surface under nonsaturating conditions. In con-trast, in the inhibition experiments described previously, a sus-pension of type 1 fimbriae-expressing bacteria was mixed withantibodies and the mixture was added directly to the targetcells or mannosylated receptors (24, 25, 30, 43). Under thelatter conditions, an apparent binding inhibition effect could becaused by clumping of either the bacterium-expressed fimbriaeor the whole bacteria. When we conducted inhibition experi-ments under a similar experimental setup, we observed a de-

FIG. 6. Native FimH sheds high-affinity-specific antibodies in theabsence of ligand mannose. MAbs from all four groups were bound toimmobilized fimbriae (open symbols) or LD (closed symbols) in thepresence of mannose. Unbound MAbs and unbound mannose wereremoved by washing. Soluble mannose was added back to the wellsafter 0, 10, 60, and 120 min of incubation without mannose. Theamounts of MAbs remaining bound to fimbriae when incubated with-out mannose for various time periods were determined in an ELISA.



http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

crease in the number of bound bacteria in the presence of allantibodies tested when high concentrations were used, includ-ing antisera raised against whole fimbriae and purified pilindomain (not shown). Closer inspection revealed that micro-aggregates of bacterial cells formed under these conditionsand were removed easily during the washing step in adhe-sion assays, reducing the number of bacteria remaining at-tached. However, when we used purified Fab fragments in-stead of whole antibodies, we never observed any inhibition (oraggregation) under even the highest concentration used. Thus,under certain conditions, an apparent adhesion inhibition ef-fect could be due to nonspecific effects rather than directblockage of the binding pocket. However, the actual basis fordiscrepancies between our inhibition studies and those of oth-ers remains to be determined.

All but one of the anti-LD antibody samples led to func-tional activation of fimbria-incorporated FimH by increasingthe mannose-specific binding of soluble HRP. This counterin-tuitive effect was shown previously with MAb21 antibodies(37). In the present study, we also show that both monoclonaland polyclonal antibodies significantly increase the number ofbacteria adhering to uroepithelial cells. Importantly, the adhe-sion enhancement effect was seen with Fab fragments and atrelatively low concentrations of whole antibodies, where nobacterial aggregation could be noted. Such antibody concen-trations are likely to be more relevant to the natural immuneresponse to infection or even vaccine administration.

Antibodies able to stabilize a protein in one conformation,increasing its affinity toward the ligand, have been describedpreviously for several proteins, including integrins (4, 12), vonWillebrand factor (42), mouse macrophage galactose/N-acetyl-galactosamine-specific calcium-type lectin (mMGL) (21), andrhodopsin (3). We have previously proposed that the enhance-ment of mannose binding by the MAb21 antibody is due to itswedged position between the domains (like the FimC chaper-one in FimH/FimC complex). Based on the recently elucidatedcrystal structure of the domain-docked LAS conformation ofLD, the MAb21-specific A epitope indeed overlaps with theinterdomain interface, though the domains interact rather dif-ferently from what was computationally predicted previously(27). However, the three other LIBS epitopes identified hereare positioned away from the interdomain interface, and anti-body binding at these sites is unlikely to result in steric inter-ference with domain interaction (i.e., interdomain wedging).Thus, HAS-specific monoclonal and polyclonal antibodies arelikely to be able to sustain the HAS conformation, at least inpart, by stabilizing the HAS-specific conformation of LIBSepitopes.

The described effect of anti-FimH LD antibodies increasingbacterial adhesion raises the question of whether such immuneresponses could promote bacterial infection rather than pro-vide protection. In the last decade, there has been a risingnumber of reports describing antibodies which actually en-hance viral or bacterial infection and pathogenesis (5, 7, 16,28). One mechanism of this antibody-dependent enhancement(so-called ADE phenomenon) is based on the bound antibodyinteracting with the receptor on host cells directly or throughintermediary molecules, like complement components (5, 7,16). Another mechanism of ADE is based on antibody bindingto microbial antigens that results in sequestering important

targets from the immune defense mechanisms (28). Here, wedescribe an additional mechanism for possible enhancement ofpathogenesis via enhancement of adhesin function.

One should consider, however, that antibody-mediated en-hancement of microbial adhesion may not necessarily lead toincreased infection. An increased binding of bacteria to targetcells could lead to increased protection of the host (7). Forexample, FimH-mediated binding of E. coli and other entero-bacteria to the GP2 receptor of M cells is considered to be aprerequisite for the mucosal immune response against thesebacteria (15). Also, FimH locked in the HAS conformation ismuch more sensitive to inhibition by solubilized mannosylatedcompounds than FimH able to switch between LAS and HAS(31). Therefore, the effect of “activating” antibodies on thecourse of bacterial colonization could potentially be advanta-geous for clearing the organism from bacterial infection. In-deed, a protective effect against bladder infection by E. coli wasreported for immunization with FimHt and FimH/FimC inmurine models and, partially, in a primate model (8, 24, 25,33). However, using these antigens has not led to the develop-ment of a vaccine for clinical application. Thus, it would beimportant to determine the exact mechanism by which protec-tion may have occurred in the animal infection models.

One important issue in this regard is to consider which FimHvariant is expressed by the strain used for animal protection stud-ies. A large portion of uropathogenic strains, especially thosecausing cystitis or asymptomatic bacteriuria, express FimH that,under equilibrium conditions, is predominantly in the LAS con-formation. On the other hand, it has been shown that naturallyoccurring mutations in FimH can increase receptor bindingunder static conditions by promoting the HAS conformationvia interference with the domain-domain interaction or otherallosteric effects on the LD structure (1, 37). These FimHmutations were shown to be positively selected in some urinarytract isolates (predominantly those that cause kidney infec-tion), possibly to increase cell tropism under the low-shearconditions of urinary tract compartments above the urethra(34, 48). However, the increased predominance of the HASconformation in FimH mutants might represent a liability forE. coli in vaccinated animal hosts. Thus, immunization withFimH constructs in the HAS conformation could have differ-ent protective effects against bacteria expressing different nat-ural variants of FimH.

It will be interesting to determine functional as well as pro-tective effects of antibodies obtained against the LAS confor-mation of LD. One should consider, however, that unlike theHAS conformation that is stable in purified LD or the FimH/FimC complex, the native LAS conformation only exists whendomains are docked. It could be readily shifted into the HASconformation by binding of a mannosylated ligand, partial en-zymatic degradation of FimH, or tensile mechanical force ofany origin. It is likely that, similar to the HAS-specific epitopes,putative LAS-specific epitopes are primarily or exclusively of aconformational rather than linear nature. The immune re-sponse to conformational epitopes involves direct interactionof the entire antigen molecule with B cells. It is unclearwhether the LAS conformation is stable enough to be pre-served as such during the close B-cell interaction with thefimbrial tip-associated FimH.

The results of the experiment with antiserum raised against



http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

whole fimbriae hint that the above-described scenario mightwell be true. In this instance, two antifimbrial sera differ intheir ability to recognize isolated LD, which correlates withtheir ability to enhance FimH-mediated binding. We have pre-viously determined that FimH in the fimbria used for immu-nization is in the LAS conformation (in the absence of man-nose or shear) (27). Thus, it is possible that for immunization,one would need to obtain either purified LD or whole FimHprotein where the LAS conformation is stabilized by, for ex-ample, the introduction of disulfide bonds as reported previ-ously (27). Such studies are under way.

Dynamic interplay between the two alternative conforma-tions of FimH could have a significant role for the adaptiveimmune response during natural E. coli infections. It is neces-sary to evaluate what type of antibody response is naturallyelicited against FimH and how it affects E. coli pathogenesis.From these perspectives, the ability of FimH to shed antibod-ies bound to LD could be physiologically significant. It is ex-pected that transient binding and unbinding of natural man-nosylated ligands (e.g., soluble or cell surface glycoproteins) tothe fimbrial FimH would result in a dynamic shifting betweenFimH conformations. This phenomenon could hypotheticallywork both ways under physiological conditions, i.e., LAS canswitch to HAS and HAS to LAS. Continuous shifting back andforth between the conformational states of LD could causerepeated shedding by LD of antibodies specific to either HASor LAS conformations. This in turn could affect, for example,opsonic or functional effects of the anti-FimH antibodies, aswell as the adaptive immune response itself. Whether or notantibody shedding happens in vivo and has a physiologicalimpact, such as immune evasion, remains to be determined.Antibody shedding is a known phenomenon in parasitic nem-atodes that are able to slough off antibodies against their sur-face coat, but this involves complete shedding of the surfaceantigen and takes many hours of growth (10). To our knowl-edge, our study is the first to report the phenomenon of anti-body shedding mediated by a conformational switch of theprotein antigen.

It has been hypothesized that force-enhanced, catch-bondproperties are widespread among adhesive proteins of bothprokaryotic and eukaryotic cells. For example, P-selectins (29),integrins (2, 22), and Von Willebrand factor (20, 50) wereshown to strengthen interactions with ligands under increasedshear and mechanical tension. Shear-enhanced binding hasbeen demonstrated for P fimbriae (32) and the CFA/I adhesinof E. coli (38), as well as for Streptococcus gordonii. Like FimH,mechanical regulation of many of these adhesive proteins isknown or suspected to involve allosteric changes in the protein(41). Because conformational shift is an intrinsic feature ofallosteric proteins, such adhesive proteins are likely to exist inalternative conformations against which specific antibody re-sponses could be evoked. Indeed, antibodies specific to the“activated” integrin conformation are well characterized andthere are many similarities of their functional effects with theHAS-specific FimH antibodies (37). In addition to adhesiveproteins, allosteric properties and/or shifts between alternativeconformational states are attributes of many other types ofproteins, such as enzymes, toxins, or transport molecules (13).Thus, understanding the extent to which different conforma-tions are capable of inducing specific immune responses and

the functional impact of such responses will help us to betterunderstand the molecular pathogenesis of diseases and to de-sign more effective vaccines, antibody-based therapeutics, anddiagnostics, as well as a variety of low-molecular-weight allo-steric modulators of proteins that play crucial roles in sustain-ing health or in disease development.

ACKNOWLEDGMENTS

We sincerely thank Steve Moseley (University of Washington, Se-attle) for his support of this work and the valuable advice and criticalassessment he provided. We thank Florentina Perianu for her excellentwork as technician on this project.

This work was supported by the National Institutes of Health, grantsR01 AI050940 and R01 A1045820.

REFERENCES

1. Aprikian, P., et al. 2007. Interdomain interaction in the FimH adhesin ofEscherichia coli regulates the affinity to mannose. J. Biol. Chem. 282:23437–23446.

2. Astrof, N. S., A. Salas, M. Shimaoka, J. Chen, and T. A. Springer. 2006.Importance of force linkage in mechanochemistry of adhesion receptors.Biochemistry 45:15020–15028.

3. Bailey, B. W., et al. 2003. Constraints on the conformation of the cytoplasmicface of dark-adapted and light-excited rhodopsin inferred from antirhodop-sin antibody imprints. Protein Sci. 12:2453–2475.

4. Bazzoni, G., and M. E. Hemler. 1998. Are changes in integrin affinity andconformation overemphasized? Trends Biochem. Sci. 23:30–34.

5. Beck, Z., Z. Prohaszka, and G. Fust. 2008. Traitors of the immune system-enhancing antibodies in HIV infection: their possible implication in HIVvaccine development. Vaccine 26:3078–3085.

6. Bergsten, G., B. Wullt, M. A. Schembri, I. Leijonhufvud, and C. Svanborg.2007. Do type 1 fimbriae promote inflammation in the human urinary tract?Cell. Microbiol. 9:1766–1781.

7. Blanco, L. P., and V. J. Dirita. 2006. Antibodies enhance interaction ofVibrio cholerae with intestinal M-like cells. Infect. Immun. 74:6957–6964.

8. Bouvard, D., and M. R. Block. 1998. Calcium/calmodulin-dependent proteinkinase II controls integrin alpha5beta1-mediated cell adhesion through theintegrin cytoplasmic domain associated protein-1alpha. Biochem. Biophys.Res. Commun. 252:46–50.

9. Brinton, C. C. 1965. The structure, function, synthesis and genetic control ofbacterial pili and a molecular model for DNA and RNA transport in gramnegative bacteria. Trans. N. Y. Acad. Sci. 27:1003–1054.

10. Bruschi, F. 2002. The immune response to the parasitic nematodeTrichinella and the ways to escape it. From experimental studies to impli-cations for human infection. Curr. Drug Targets Immune Endocr. Metabol.Disord. 2:269–280.

11. Choudhury, D., et al. 1999. X-ray structure of the FimC-FimH chaperone-adhesin complex from uropathogenic Escherichia coli. Science 285:1061–1066.

12. Du, X., et al. 1993. Long range propagation of conformational changes inintegrin alpha IIb beta 3. J. Biol. Chem. 268:23087–23092.

13. Gunasekaran, K., B. Ma, and R. Nussinov. 2004. Is allostery an intrinsicproperty of all dynamic proteins? Proteins 57:433–443.

14. Hagan, E. C., A. L. Lloyd, D. A. Rasko, G. J. Faerber, and H. L. Mobley.2010. Escherichia coli global gene expression in urine from women withurinary tract infection. PLoS Pathog. 6:e1001187.

15. Hase, K., et al. 2009. Uptake through glycoprotein 2 of FimH(�) bacteria byM cells initiates mucosal immune response. Nature 462:226–230.

16. Hoffmann, E. M., and J. J. Houle. 1995. Contradictory roles for antibody andcomplement in the interaction of Brucella abortus with its host. Crit. Rev.Microbiol. 21:153–163.

17. Jin, L., B. M. Fendly, and J. A. Wells. 1992. High resolution functionalanalysis of antibody-antigen interactions. J. Mol. Biol. 226:851–865.

18. Johnson, J. R., and W. E. Stamm. 1987. Diagnosis and treatment of acuteurinary tract infections. Infect. Dis. Clin. North Am. 1:773–791.

19. Jones, C. H., et al. 1995. FimH adhesin of type 1 pili is assembled into afibrillar tip structure in the Enterobacteriaceae. Proc. Natl. Acad. Sci.U. S. A. 92:2081–2085.

20. Kim, J., C. Z. Zhang, X. Zhang, and T. A. Springer. 2010. A mechanicallystabilized receptor-ligand flex-bond important in the vasculature. Nature466:992–995.

21. Kimura, T., Y. Imai, and T. Irimura. 1995. Calcium-dependent conforma-tion of a mouse macrophage calcium-type lectin. Carbohydrate binding ac-tivity is stabilized by an antibody specific for a calcium-dependent epitope.J. Biol. Chem. 270:16056–16062.

22. Kong, F., A. J. Garcia, A. P. Mould, M. J. Humphries, and C. Zhu. 2009.Demonstration of catch bonds between an integrin and its ligand. J. CellBiol. 185:1275–1284.



http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

23. Korotkova, N., et al. 2006. Crystal structure and mutational analysis of theDaaE adhesin of Escherichia coli. J. Biol. Chem. 281:22367–22377.

24. Langermann, S., et al. 2000. Vaccination with FimH adhesin protects cyno-molgus monkeys from colonization and infection by uropathogenic Esche-richia coli. J. Infect. Dis. 181:774–778.

25. Langermann, S., et al. 1997. Prevention of mucosal Escherichia coli infectionby FimH-adhesin-based systemic vaccination. Science 276:607–611.

26. Legleiter, J., et al. 2009. Monoclonal antibodies recognize distinct confor-mational epitopes formed by polyglutamine in a mutant huntingtin fragment.J. Biol. Chem. 284:21647–21658.

27. Le Trong, I., et al. 2010. Structural basis for mechanical force regulation ofthe adhesin FimH via finger trap-like beta sheet twisting. Cell 141:645–655.

28. MacLennan, C. A., et al. 2010. Dysregulated humoral immunity to nonty-phoidal Salmonella in HIV-infected African adults. Science 328:508–512.

29. Marshall, B. T., et al. 2003. Direct observation of catch bonds involvingcell-adhesion molecules. Nature 423:190–193.

30. Meiland, R., et al. 2004. Fimch antiserum inhibits the adherence of Esche-richia coli to cells collected by voided urine specimens of diabetic women.J. Urol. 171:1589–1593.

31. Nilsson, L. M., W. E. Thomas, E. V. Sokurenko, and V. Vogel. 2006. Elevatedshear stress protects Escherichia coli cells adhering to surfaces via catchbonds from detachment by soluble inhibitors. Appl. Environ. Microbiol.72:3005–3010.

32. Nilsson, L. M., W. E. Thomas, E. Trintchina, V. Vogel, and E. V. Sokurenko.2006. Catch bond-mediated adhesion without a shear threshold: trimannoseversus monomannose interactions with the FimH adhesin of Escherichiacoli. J. Biol. Chem. 281:16656–16663.

33. Poggio, T. V., J. L. La Torre, and E. A. Scodeller. 2006. Intranasal immuni-zation with a recombinant truncated FimH adhesin adjuvanted with CpGoligodeoxynucleotides protects mice against uropathogenic Escherichia colichallenge. Can. J. Microbiol. 52:1093–1102.

34. Sokurenko, E. V., et al. 1998. Pathogenic adaptation of Escherichia coli bynatural variation of the FimH adhesin. Proc. Natl. Acad. Sci. U. S. A.95:8922–8926.

35. Sokurenko, E. V., H. S. Courtney, J. Maslow, A. Siitonen, and D. L. Hasty.1995. Quantitative differences in adhesiveness of type 1 fimbriated Esche-richia coli due to structural differences in fimH genes. J. Bacteriol. 177:3680–3686.

36. Sokurenko, E. V., V. Vogel, and W. E. Thomas. 2008. Catch-bond mechanismof force-enhanced adhesion: counterintuitive, elusive, but widespread? CellHost Microbe 4:314–323.

37. Tchesnokova, V., et al. 2008. Integrin-like allosteric properties of the catchbond-forming FimH adhesin of Escherichia coli. J. Biol. Chem. 283:7823–7833.

38. Tchesnokova, V., et al. 2010. Shear-enhanced binding of intestinal coloniza-tion factor antigen I of enterotoxigenic Escherichia coli. Mol. Microbiol.76:489–502.

39. Thankavel, K., et al. 1997. Localization of a domain in the FimH adhesin ofEscherichia coli type 1 fimbriae capable of receptor recognition and use of adomain-specific antibody to confer protection against experimental urinarytract infection. J. Clin. Invest. 100:1123–1136.

40. Thomas, W. E., L. Nilsson, M. Forero, E. V. Sokurenko, and V. Vogel. 2004.“Stick-and-roll” bacterial adhesion mediated by catch-bonds. Mol. Micro-biol. 53:1545.

41. Thomas, W. E., V. Vogel, and E. Sokurenko. 2008. Biophysics of catch bonds.Annu. Rev. Biophys. 37:399–416.

42. Ulrichts, H., et al. 2004. A monoclonal antibody directed against human vonWillebrand factor induces type 2B-like alterations. J. Thromb. Haemost.2:1622–1628.

43. Vandemaele, F., et al. 2005. Immunization with the binding domain of FimH,the adhesin of type 1 fimbriae, does not protect chickens against avianpathogenic Escherichia coli. Avian Pathol. 34:264–272.

44. Van Gerven, N., H. De Greve, and J. P. Hernalsteens. 2008. InactivatedSalmonella expressing the receptor-binding domain of bacterial adhesinselicit antibodies inhibiting hemagglutination. Vet. Microbiol. 131:369–375.

45. Van Loy, C. P., E. V. Sokurenko, R. Samudrala, and S. L. Moseley. 2002.Identification of amino acids in the Dr adhesin required for binding todecay-accelerating factor. Mol. Microbiol. 45:439–452.

46. Van Regenmortel, M. H. V. 1996. Mapping epitope structure and activity:from one-dimensional prediction to four-dimensional description of anti-genic specificity. Methods 9:465–472.

47. Warren, J. 1996. Clinical presentation and epidemiology of urinary tractinfections, p. 3–27. In H. L. T. Mobley and J. W. Warren (ed.), Urinary tractinfections: molecular pathogenesis and clinical management. ASM Press,Washington, DC.

48. Weissman, S. J., et al. 2007. Differential stability and trade-off effects ofpathoadaptive mutations in the Escherichia coli FimH adhesin. Infect. Im-mun. 75:3548–3555.

49. Weissman, S. J., et al. 2006. Clonal analysis reveals high rate of structuralmutations in fimbrial adhesins of extraintestinal pathogenic Escherichia coli.Mol. Microbiol. 59:975–988.

50. Yago, T., et al. 2008. Platelet glycoprotein Ibalpha forms catch bonds withhuman WT vWF but not with type 2B von Willebrand disease vWF. J. Clin.Invest. 118:3195–3207.

51. Yakovenko, O., et al. 2008. FimH forms catch bonds that are enhanced bymechanical force due to allosteric regulation. J. Biol. Chem. 283:11596–11605.

Editor: B. A. McCormick



http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

type 1 fimbrial adhesin fimh elicits an immune response ... · type 1 fimbrial adhesin fimh elicits...

Documents