Manganese Binds to Clostridium difficile Fbp68 and Is Essential for Fibronectin Binding*

Clostridium difficile is an etiological agent of pseudomembranous colitis and antibiotic-associated diarrhea. Adhesion is the crucial first step in bacterial infection. Thus, in addition to toxins, the importance of colonization factors in C. difficile-associated disease is recognized. In this study, we identified Fbp68, one of the colonization factors that bind to fibronectin (Fn), as a manganese-binding protein (KD = 52.70 ± 1.97 nm). Furthermore, the conformation of Fbp68 changed dramatically upon manganese binding. Manganese binding can also stabilize the structure of Fbp68 as evidenced by the increased Tm measured by thermodenatured circular dichroism and differential scanning calorimetry (CD, Tm = 58–65 °C; differential scanning calorimetry, Tm = 59–66 °C). In addition, enhanced tolerance to protease K also suggests greatly improved stability of Fbp68 through manganese binding. Fn binding activity was found to be dependent on manganese due to the lack of binding by manganese-free Fbp68 to Fn. The C-terminal 194 amino acid residues of Fbp68 (Fbp68C) were discovered to bind to the N-terminal domain of Fn (Fbp68C-NTD, KD = 233 ± 10 nm, obtained from isothermal titration calorimetry). Moreover, adhesion of C. difficile to Caco-2 cells can be partially blocked if cells are pretreated with Fbp68C, and the binding of Fbp68C on Fn siRNA-transfected cells was significantly reduced. These results raise the possibility that Fbp68 plays a key role in C. difficile adherence on host cells to initiate infection.

the N-terminal domain (NTD), whereas the fibronectin-binding site on Fbp68 resides in the C-terminal 194 amino acids (Fbp68C). Finally, Fbp68C-NTD interaction was able to mediate the adhesion of C. difficile to Caco-2 cells, indicating that Fbp68 is an important colonization factor contributing to clostridial virulence.

MATERIALS AND METHODS
Bacterial Strains and Cell Culture-C. difficile 630 was used in this study (29). C. difficile was cultivated in prereduced anaerobically sterilized peptone yeast extract broth with glucose (Anaerobe Systems, Morgan Hill, CA). Escherichia coli strains were cultured in Luria-Bertani broth (LB) with appropriate antibiotics (Table 1). Caco2 cells were cultured in Dulbecco minimum essential medium (DMEM) containing 10% fetal bovine serum (Invitrogen) and were grown at 37°C in a humidified atmosphere with 5% CO 2 (30).
Gene Knock-out and Characterization of the Mutants-The fbp68 was an insertion knock-out using the ClosTron gene knock-out system developed by Heap et al. (31)(32)(33). The intron target sites within fbp68 recognized by L1.LtrB-derived introns were identified by using intron target tool, one of which, 102 bp from the start codon, was used to generate a mutant designated CD⌬Fbp68 102 ( Table 1). The intron targeting region designated by the intron design tool was constructed synthetically by DNA 2.0 Inc. (Menlo Park, CA). The synthetic construct was inserting into the ClosTron plasmid pMTL007C-E2, and the resulting plasmid pMTL007C-E2-CDI-Fbp68 -102S (Table 1) was electroporated into the conjugative donor E. coli CA434 and then transferred via conjugation into C. difficile 630. Successful transconjugates were selected from a BHI plate supplemented with 250 g/ml cycloserine (Sigma) and 8 g/ml cefotaxime (Sigma) to select against the E. coli conjugal donor and 15 g/ml thiampheni-col (Sigma) to select for the pMTL007C-E2-CDI-Fbp68-102S retargeted C. difficile. Subsequent integrants were selected on BHI agar supplemented with lincomycin (20 g/ml) (Sigma). PCR using primers that flanked fbp68 and EBS universal primer was performed to demonstrate the integration of L1.LtrB-derived introns ( Table 2). PCR using Thio-F1 and ErmB-R1 primers ( Table 2) confirmed the Linco phenotype was caused by the splicing of the group I intron from the group II intron following integration and not a spontaneous Linco (32). Two primers, FliD-F and FliD-R, were used to demonstrate that the mutant is a C. difficile strain ( Table 2). The PCR product of fbp68 using primers Fbp68F and Fbp68R ( Table 2) was inserted into plasmid pMTL84151 (Table 1), electroporated into the conjugative donor E. coli CA434, and then transferred via conjugation into CD⌬Fbp68 102 for complementary testing as described previously (33).
zinc chloride, calcium chloride, protease K, and protein A-10 nm gold particle colloidal suspension were purchased from Sigma. 120-kDa cell binding domain (CBD) or 40-kDa domain of Fn, mouse anti-fibronectin, and mouse anti-␣-actin were purchased from Chemicon International (Temecula, CA). Fbp68 antibody was prepared by vaccination of rabbits. Briefly, intramuscular inoculation with 100 g of recombinant Fbp68 in Freund's complete adjuvant (first vaccination) was followed by 100 g intramuscularly in Freund's incomplete adjuvant (booster) 3 weeks later. Three weeks after the second booster, the rabbit was sacrificed, and serum was collected. Anti-C. difficile antibodies were previously prepared by vaccination of horses with whole killed cells of C. difficile. Mouse anti-histidine tag, rabbit anti-GST antibody, Texas Red-conjugated goat anti-rabbit IgG, FITC-conjugated goat anti-horse antibody, FITC-conjugated goat anti-mouse antibody, HRP-conjugated goat anti-mouse antibody, and HRPconjugated goat anti-rabbit antibody were ordered from Invitrogen. HRP-conjugated goat anti-horse antibody was purchased from KPL (Gaithersburg, MD). Plasmid Construction and Protein Purification-The construct for the expression of histidine tag fused with Fbp68, GST, and histidine tag fused with Fbp68N (amino acids 2-396) or Fbp68C (amino acids 397-591) was generated using vectors pQE30 (Qiagen, Valencia, CA) and pGEX4T2 (GE Healthcare) (Fig. 1A). To perform the PCRs, the primers shown in Table 2 were utilized based on the Fbp68 sequence (11). Primers were engineered to introduce a BamHI site at the 5Ј end of each fragment and a stop codon followed by a SalI site at the 3Ј end of each fragment. PCR products were sequentially digested with BamHI and SalI and then ligated into pQE30 or pGEX4T2 cut with BamHI and SalI, respectively. The soluble forms of all of recombinant proteins were purified from E. coli as described previously (34,35).
Isothermal Titration Calorimetry (ITC)-The experiments were carried out with a CSC 5300 microcalorimeter (Calorimetry Science Corp., Lindon, UT) at 25°C as described previously (26). Before the ITC experiment, traces of metal ions were removed from Fbp68 solution through incubation with EDTA and subsequent dialysis in Tris buffer (25 mM Tris, 150 mM sodium chloride (pH 7.0)). The cell contained 1 ml of a solution of Fbp68, and the syringe contained 250 l of a solution of various metal ions at different concentrations, as indicated in Table 3. For the NTD binding experiments, 1 ml of NTD (35 M) was in the cell, and 250 l of Fbp68C (325 M) with 100 M of MnCl 2 were in the syringe. All solutions were in Tris buffer (pH 7.0). The concentration of each species is presented in Table 3. The titration was performed as follows: 25 injections of 10 l with a stirring speed of 250 rpm with a delay time between injections of 5 min. Data were analyzed using BindWorks software (model CSC 5300, Calorimetry Science Corp.) fitting them to an independent binding model.
Fluorescence Spectrometry-Fbp68 was treated with 50 M of EDTA and then extensively dialyzed in Tris buffer (pH 7.0) to remove metal ions. Fluorescence emission spectra were measured on a Hitachi F7500 spectrofluorometer (Hitachi, San Jose, CA). All spectra were recorded in the correct spectrum mode of the instrument using excitation and emission band passes of 5 nm. The intrinsic Trp fluorescence of protein was recorded by exciting the solution at 295 nm and measuring the emission in the region from 305 to 400 nm. For manganese titration, 0, 25, 50, 100, 200, and 400 nM manganese chloride were mixed with 1 M Fbp68. and spectra were recorded after 3 min.
For the ANS fluorescence experiment, ANS binding was checked by adding 100 M ANS solution (10 mM stock in 100% methanol) to a protein solution (1 M) and incubated for 2 min, and spectra were recorded between 400 and 600 nm. Next, Fbp68 was treated with different concentrations of MnCl 2 (0, 25, 50, 100, 200, and 400 nM) for 5 min, and then spectra were measured at an excitation wavelength of 295 nm. All spectra were recorded in the correct spectrum mode with excitation and emission band passes of 5 nm each and corrected for volume changes before further analysis. All measurements were performed at 25°C.
CD Spectrometry-CD spectra were recorded on a Jasco J-815 spectropolarimeter under N 2 atmosphere at room temperature (25°C) in 0.02-and 0.5-cm path length quartz cells for far-and near-UV, respectively, with eight repeated measurements. The Fbp68 was treated with 50 M EDTA and then extensively dialyzed in Tris buffer (pH 7.0) to remove metal ions. Aliquots of manganese chloride solution (0, 25, 50, 100, 200, and 400 nM) were added to 1 M Fbp68 protein solution and incubated for 5 min. All spectra were recorded in Tris buffer (pH 7.0). In a melting temperature experiment, 10 M Fbp68 in the absence or presence of 100 M MnCl 2 was subjected to thermal unfolding, and data were collected at 1°C/ min increments from 25 to 100°C recording the ellipticity at 205 nm, with 30-s temperature equilibrations, followed by 30 s of data averaging. To measure the melting point, a first order derivative was applied to the results from the melting experiment. In all CD experiments, the background spectrum of Tris buffer (pH 7.0) alone was subtracted from the proteincontaining spectra.
Differential Scanning Calorimetry (DSC)-Excess heat capacity C p (T) of Fbp68C with or without MnCl 2 was measured using a DSC Q1000 microcalorimeter (Waters). The Fbp68 was treated with 50 M EDTA and then extensively dialyzed in Tris buffer (pH 7.0) to remove metal ions. Degassed sample containing 30 M Fbp68 with or without 100 M MnCl 2 in Tris buffer were heated at 0.1 K/min scan rate. Heat capacity, C p (T), data were recorded, corrected for buffer base line, and normalized to the amount of the sample. The TA Universal Analysis software (Waters) was used for the data analysis and display. All calorimetric experiments in this study were repeated three times to ensure reproducibility. Protease K Resistance Experiment-The recombinant histidine-tagged Fbp68 used in this study was treated with 50 M EDTA and subsequently dialyzed in Tris buffer (pH 7.0) to remove trace metal ions. One M Fbp68 was mixed with or without 100 M MnCl 2 and then dialyzed in Tris buffer to remove the unbound MnCl 2 to prevent unbound MnCl 2 from interfering with the activity of protease K. Then Fbp68 samples with and without MnCl 2 were analyzed for the sensitivity of protease K by treating with 0 -60 ng/l protease K (PK) at 37°C for 1 h. The reaction was stopped by adding 1 l of protease inhibitor without EDTA (Thermoscientific, Logan, UT) and then mixing with Laemmli sample loading buffer consisting of 50 mM Tris-HCl (pH 6.8), 100 mM dithiothreitol, 2% SDS, 0.25 mM PMSF, and 0.1% bromphenol blue in 20% glycerol. The digested Fbp68 was subjected to 10% SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes. The membranes were incubated in 5% skim milk in PBS/T overnight and then incubated with mouse anti-histidine tag antibody (1:1000). The immunocomplexes were detected with an HRP-conjugated goat anti-mouse IgG antibody (1:5000).
Binding Assays by ELISA-To determine the binding of untreated or manganese-treated Fbp68 to Fn (Fig. 4A), 100 l of 1 M Fn or BSA (negative control and data not shown) were coated on microtiter plate wells and blocked subse-quently as described previously (36). Various concentrations of 100 M MnCl 2 -treated histidine-tagged LigBCen2 (a known Fn-binding protein from Leptospira interrogans to serve as positive control) (36), untreated or 100 M MnCl 2treated histidine tagged Fbp68, or LigBCon4 (a protein from L. interrogans lacking Fn binding activity to serve as negative reference) (37) in 100 l of Tris buffer (pH 7.0) were added to microtiter plate wells and incubated for 1 h at 37°C. To map the binding sites of Fn on Fbp68 (Fig. 5A), different concentrations of 100 M MnCl 2 -treated GST-fused Fbp68 (positive control), LigBCon (a protein from L. interrogans lacking Fn binding activity to serve as negative control) (37), Fbp68N, or Fbp68C in 100 l of Tris buffer (pH 7.0) were added to microtiter plate wells and incubated for 1 h at 37°C. To localize the Fbp68C-binding sites on Fn (Fig. 5B), 100 l of 1 M Fn (positive control), NTD, gelatin binding domain, CBD, 40-kDa protein, or BSA (negative control) were coated on microtiter plate wells and blocked subsequently as described above. Serial concentrations of 100 M MnCl 2 -treated GST-fused Fbp68C or LigBCon (negative reference and data not shown) in 100 l of Tris buffer (pH 7.0) were added into microtiter plate wells and incubated for 1 h at 37°C. To detect the binding of Fbp68C to Caco-2 cells (Fig. 7A), 100 l of 10 M GSTfused Fbp68C or GST (negative control) was added to microtiter plate wells cultured with 10 5 Caco-2 cells for 1 h. To determine the inhibitory effect on C. difficile adhesion caused by pretreatment of Fbp68C (Fig. 7B), 10 5 Caco-2 cells were treated with 100 l of 10 M GST-fused Fbp68C or GST (negative reference) at 37°C for 1 h prior to incubation with 10 7 C. difficile at 37°C for 1 h.
To detect the binding of histidine-tagged Fbp68, LigBCen2, LigBCon, mouse anti-histidine tagged (1:200), and HRP-con- jugated goat anti-mouse IgG (1:1000) were used as primary and secondary antibodies, respectively (34,35). To measure the binding of GST-fused LigBCon, Fbp68, Fbp68N, Fbp68C, rabbit anti-GST (1:200), and HRP-conjugated goat anti-rabbit IgG (1:1,00) were used as primary and secondary antibodies, respectively. To determine the adhesion of C. difficile, horse anti-GST (1:200) and HRP-conjugated goat anti-horse IgG (1:1000) were used as primary and secondary antibodies, respectively. After washing the plates three times with TBST (0.05% Tween 20 in Tris buffer), 100 l of TMB (KPL) was added to each well and incubated for 5 min. The reaction was stopped by adding 100 l of 0.5% hydrofluoric acid to each well. Each plate was read at 630 nm using an ELISA plate reader (BioTek EL-312, Winooski, VT). Each value represents the mean Ϯ S.E. of three trials in triplicate samples. Statistically significant (p Ͻ 0.05) differences are indicated by an asterisk.
Bacterial Attachment Assay-To study the adhesion of C. difficile 630 (wild type), CD⌬Fbp68 102 (mutant), and CD⌬Fbp68 102 /pMTL84151-fbp68 (complemented mutant) to Caco-2 cells (Fig. 7E), an attachment assay was slightly modified from the method reported previously (38). Basically, icecold PBS buffer was used to wash the cells prior to the assay to prevent C. difficile invasion. A total of 10 8 CD630WT, CD⌬Fbp68 102 , or CD⌬Fbp68 102 /pMTL84151-fbp68 bacteria suspended in ice-cold medium was then incubated with 10 6 Caco-2 cells at 4°C for 1 h. The wells incubated without Caco-2 cells served as negative controls. Unattached bacteria were removed by five washes with PBS. PBS containing 1% (v/v) Triton X-100 was used to resuspend adherent bacteria, and the 10 4 -fold dilution of the adherent bacterial suspensions was spread on C. difficile agar plates (BD Biosciences) to determine the number of cell-associated bacteria per well.
Surface Plasmon Resonance (SPR)-The Fbp68-Fn or Fbp68C-NTD interaction was analyzed by an SPR technique using a Biacore 2000 instrument (GE Healthcare). Briefly, Fbp68 and Fbp68C were treated with 50 M EDTA and subsequently dialyzed in Tris buffer (pH 7.0) to remove trace metal ions. To determine the Fn binding activity of Fbp68 in the presence or absence of manganese (Fig. 4B), 100 M manganese chloride in Tris buffer (pH 7.0) was mixed with 1.5 M histidine-tagged Fbp68 when immobilized on an NTA chip (GE Healthcare). Then 10 l of 1000 nM Fn was injected into the flow cell at 10 l/min at 25°C. To measure the binding affinity of Fbp68C and Fn (Fig. 5E), 1.5 M histidine-tagged Fbp68C was incubated with Tris buffer with or without 100 M manganese chloride when immobilized on an NTA chip, and the NTA chip was conjugated to 500 M nickel sulfate prior to the immobilization of Fbp68 or Fbp68C. A control flow cell was injected with Tris buffer without Fbp68 or Fbp68C. Then 10 l of a serial concentration (0, 62.5, 125, 250, 500, and 1000 nM) of NTD was injected into the flow cell at 10 l/min at 25°C. All sensorgram data were subtracted from the negative control flow cell. To obtain the kinetic parameters of the interaction, the data of the sensograms were fitted by BIAevaluation software version 3.0 using the onestep biomolecular association reaction model (1:1 Langmuir model), which resulted in optimum mathematical fits with the lowest values.
Binding Assays by Confocal Laser-scanning Microscopy (CLSM)-To determine the binding inhibition of C. difficile to Caco-2 cells by Fbp68C by CLSM, 10 6 Caco-2 cells were preincubated with 50 M GST-Fbp68C or GST (negative control) in 100 l of PBS for 1 h at 37°C. Then 10 8 C. difficile 630 was added to each well and incubated for 1 h at 37°C (Fig. 7C). To measure the adhesion of C. difficile to Fn (Fig. 7D), 10 8 wild type C. difficile (CD630WT), fbp68 knock-out mutant (CD⌬Fbp68 102 ), or fbp68 knock-out mutant complemented with intact fbp68 (CD⌬Fbp68 102 /pMTL84151-fbp68) was added to glass slides in 24-well plates coated with 1 M Fn. BSA-coated slides served as negative controls. To determine the attachment of C. difficile and the binding of GST-Fbp68C or GST to Caco-2 cells (Fig. 7, C and D), rabbit anti-GST (1: 250) and horse anti-C. difficile antibodies (1:100) served as primary antibodies, and Texas Red-conjugated goat anti-rabbit IgG (1:250) and FITC-conjugated goat anti-horse IgG (1: 250) were used as secondary antibodies. Fixation and immunofluorescence staining were performed as described previously (36) with slight modifications. Briefly, C. difficile and Caco-2 cells were fixed in 2% paraformaldehyde for 60 min at room temperature. For antibody labeling, fixed bacteria were incubated in PBS containing 0.3% BSA for 10 min at room temperature. The primary and secondary antibodies in PBS containing 0.3% BSA were incubated sequentially for 60 min at room temperature. After incubation with primary and secondary antibodies, the glass slides were mounted with coverslips using Prolong Antifade (Molecular Probes, Eugene, OR) and viewed with a 60ϫ objective by CLSM (Olympus, America, Inc., Melville, NY). An Olympus Fluoview 500 confocal laser-scanning imaging system equipped with krypton, argon, and He-Ne lasers on an Olympus IX70 inverted microscope with a PLAPO 60ϫ objective was used. The settings were identical for all captured images. Images were processed using Adobe Photoshop CS2. For counting the attachment of C. difficile to Fn-coated wells, three fields (40ϫ objective) were selected at random to count the number of bound organisms. All studies were repeated three times, and the attachment of C. difficile to Caco-2 cells was scored by an operator who was blinded to the treatment group.
Small Interfering RNA (siRNA) Inhibition of Fbp68C Binding-siRNA duplexes directed against human Fn (Ambion AM121357) and negative siRNA duplex (Ambion AM4611) were purchased from Ambion (Foster City, CA). RNA duplexes were introduced into Caco-2 cells by the method of lipofection (37), and 8 ϫ 10 5 cells were transfected with 0.4 g of negative siRNA and Fn-siRNA. Adhesion assays were performed 72 h after lipofection (37). The knockdown efficiency of endogenous Fn expression was determined as described previously (37) with slight modification. The total protein content of 10 6 Caco-2 cells was analyzed using Western immunoblotting as described under "protease K resistance experiments." The protein bands of actin derived from Caco-2 cells were measured as a control using a mouse anti-actin antibody (1:5000). The band intensity was measured by densitometry using ImageJ software (National Institutes of Health, Bethesda) (39). Fbp68C binding assay was performed 72 h after lipofection. To determine the binding of Fbp68C to Fn, 100 l of 50 M GST-Fbp68C or GST was added to 10 6 Caco-2 cells transfected with Fn or negative siRNA. To determine the binding of Fbp68C and the expression of Fn on Caco-2 cells, rabbit anti-GST (1:250) and mouse anti-Fn (1:250) served as the primary antibodies, and FITCconjugated goat anti-mouse IgG (1:250) and Texas Red-conjugated goat anti-rabbit IgG (1:250) were used as secondary antibodies. Fixation, immunofluorescence staining, image detection, and processing were as described above. All the experiments were performed in triplicate.
Statistical Analysis-Each data point represents the means Ϯ S.E. for each sample tested in triplicate. Data were analyzed by Student's t test, and statistically significant differences were claimed at p values Ͻ 0.05.

Manganese Binds to Fbp68 and Induces Conformational
Change-Previously, Fbp68 was identified as a Fn-binding protein (11). Because metal ions can modulate the ECM binding activities of some ECM-binding proteins (26, 27, 40 -42), the binding activity of Fbp68 to Ca 2ϩ , Mg 2ϩ , Zn 2ϩ , and Mn 2ϩ was examined in this study through ITC. As shown in Fig. 1A and Table 3, Fbp68 bound to Mn 2ϩ (K D ϭ 52 Ϯ 1.97 nM) but not to Ca 2ϩ , Mg 2ϩ , or Zn 2ϩ , and the stoichiometric values indicated that five manganese ions were able to bind to one Fbp68 (Table 3). Furthermore, a significant quenching of tryptophan fluorescence spectra for Fbp68 (12%) upon Mn 2ϩ binding indicated manganese binding could also alter the conformation of Fbp68 (Fig. 2B). Interestingly, the Fbp68 conformational changes upon Mn 2ϩ binding were mainly in the random coiled region because the proportion of ␣-helix increased whereas that of random coil was reduced when Fbp68 titrated with Mn 2ϩ was measured through far-UV CD spectrometry (␣-helix from 56 to 69%, ␤-strand remaining at 25%, and random coil from 19 to 6%.) (Fig. 2C). In addition, the conformational change occurred in a hydrophobic area because the fluorescence of ANS-Fbp68 increased dramatically (20%) upon Mn 2ϩ binding (Fig. 2D).
Manganese Binding Enhances the Stability of Fbp68-The function of Mn 2ϩ binding is generally recognized as maintaining protein stability (43,44). To gain more insight about the function of manganese binding to Fbp68, DSC thermounfolding and thermo-denatured CD were performed. As shown in Fig. 3A, the CD profiles of heat-induced folding to unfolding transition in Mn 2ϩ -bound Fbp68 shifted to a significantly higher temperature compared with the apo-Fbp68, and the T m value of Mn 2ϩ -bound Fbp68 also dramatically increased (apo-Fbp68, T m ϭ 58.0 Ϯ 2.1°C; Mn 2ϩ -bound Fbp68, T m ϭ 65.9 Ϯ 1.3°C). A similar result was also observed in the DSC experiment (apo-Fbp68, T m ϭ 59.3 Ϯ 2.1°C; Mn 2ϩ -bound Fbp68, T m ϭ 66.1 Ϯ 3.2°C) (Fig. 3B). It has also been reported that Mn 2ϩ -bound proteins can better resist protease digestion compared with apoproteins. Therefore, protease K resistance assays were applied to Mn 2ϩbound or apo-Fbp68 to test their stability. As indicated in Fig.  3C, Mn 2ϩ -bound Fbp68 was able to tolerate a higher concentration of protease K because the decomposition of Mn 2ϩbound Fbp68 could not be observed until the addition of 60 ng/l protease K. However, apo-Fbp68 was vulnerable to protease K digestion at a comparatively lower concentrations of protease K (7.5 ng/l) (Fig. 3C).
Interaction of Fbp68 and Fn Requires Manganese-Fbp68 is a Fn-binding protein. To elucidate the effect of manganese on the Fbp68-Fn interaction, histidine-tagged Fbp68 was treated with EDTA and dialyzed in PBS buffer. The binding of Fn to untreated Fbp68 or EDTA-treated Fbp68 with or without Mn 2ϩ was then measured by ELISA. As shown in Fig. 4A, both untreated Fbp68 and Mn 2ϩ -bound Fbp68 could associate with Fn with similar affinities (Fbp68, K D ϭ 253 Ϯ 23 nM; Mn 2ϩ -bound Fbp68, K D ϭ 221 Ϯ 10 nM). Strikingly, EDTAtreated Fbp68 completely lost Fn binding activity on removal of divalent cation (Fig. 4A). These results clearly indicate that manganese is essential for Fbp68 binding to Fn. The SPR results (Fig. 4, B and C) support the same conclusion because Mn 2ϩ -bound Fbp68 was bound tightly to Fn, and EDTAtreated Fbp68 failed to bind to Fn.
Mapping the Fbp68 and NTD-binding Sites-To better define the Fn-binding site of Fbp68, Fbp68 was truncated into two fragments, Fbp68N (residues 2-396) and Fbp68C (resi-dues 397-591) (Fig. 1A). ELISA was performed to determine the interaction of Fn with Fbp68N or Fbp68C. As presented in Fig. 5A, GST-fused Fbp68C was strongly bound to immobilized Fn (K D ϭ 234 Ϯ 22 nM), whereas Fbp68N could not bind. In addition, the Fbp68-binding site on Fn was also mapped via ELISA, and only NTD of Fn was able to interact with GST-fused Fbp68C (K D ϭ 220 Ϯ 93 nM) (Figs. 1B and 5B). ITC and SPR studies yielded similar affinities (ITC, K D ϭ 233 Ϯ 10 nM; SPR, K D ϭ 216 Ϯ 56 nM) (Fig. 5, C and D), providing further support to the Fbp68C-NTD interaction determined by ELISA.
Generating fbp68 Mutant and Complemented Strains-To elucidate physiological roles of the Fbp68-Fn interaction, fbp68 mutant, and fbp68 complemented strains were generated. To obtain fbp68 mutant, the lincomycin-resistant gene was inserted between 102nd and 103rd base pair (Fig. 6A). The transconjugants of C. difficile 630 with the inserted intron would confer a lincomycin-resistant phenotype. To select the fbp68 knock-out mutant, PCR analyses with different primer pairs was performed. As shown in Fig. 6B, 3.6 kb of the DNA fragment could be amplified from the PCR with the primer pairs, Fbp68-F and Fbp68-R, in the selected fbp68 knock-out mutant, but only 1.7-kb PCR amplificon was obtained with the same primer pairs in wild type C. difficile 630 due to the lack of the inserted intron (Fig. 6B). In addition, the 0.9-kb lincomycin-resistant gene was amplified by primer pairs, ErmB-R1 and Thio-F1, and 0.3 kb of 5Ј adjacent fragment of inserted intron amplified by Fbp68-F and EBS-Uni- versal primer was observed in this fbp68 knock-out mutant (Fig. 6B). This mutation was complemented in trans with plasmid pMTL84151-fbp68 ( Table 1). The immunoblot analysis was also used to demonstrate that Fbp68 was absent in the mutant but restored in the complemented strain (Fig. 6C). This mutation was complemented in trans with plasmid pMTL84151-fbp68 ( Table 1).

Treatment of Fbp68C and the Transfection of Fn siRNA Inhibit C. difficile Binding on the Mammalian Cells-Bacterial
Fn-binding proteins contribute to host cell adhesion (36,37,45). Because Fbp68 is located on the surface of C. difficile (supplemental Fig. 1) (21), it is reasonable to hypothesize that Fbp68 might be one of the adhesins mediating C. difficile adhesion. To gain more understanding of the physiological relevance of Fbp68-Fn interaction, Caco-2 cells, a human epithelial colorectal cell line, were used to test the binding activity of Fbp68C. Compared with GST (negative control), GST-Fbp68C was able to bind to Caco-2 cells as shown in Fig. 7A. In addition, pretreatment of Caco-2 cells with Fbp68C decreased C. difficile adhesion by 51% (Fig. 7, B and C), in agreement with the binding affinity assay results. However, the indistinguishable Fn and cell binding activity of wild type C. difficile 630, fbp68 knock-out mutant, or fbp68 complemented strains suggest redundant adhesins appearing on C. difficile contribute to adhesion (Fig. 7, D and E). To further elucidate the receptor role of Fn on Caco-2 cells for putative binding partners such as Fbp68 of C. difficile, the binding of GST-Fbp68C or C. difficile to Fn or negative siRNA-transfected cells was examined. As shown in Fig. 8A, Fn siRNA duplex specifically reduced the expression of Fn. Moreover, the de-creased binding of GST-Fbp68C and a 35% reduction in adhesion of C. difficile to Fn siRNA-transfected cells (compared with the negative siRNA transfected cells) validated the conclusion that Fn plays a pivotal role in the adherence of C. difficile (Fig. 8B) (data not shown).

DISCUSSION
Two high molecular weight toxins (toxins A and B) of C. difficile play major roles in the pathogenesis of pseudomembranous colitis and antibiotic-associated diarrhea (1). However, not all C. difficile-associated disease is caused by toxigenic strains of C. difficile (5,8,14,46,47). Thus, other virulence factors, such as adhesins, are likely important in the pathogenesis of C. difficile-associated disease. Adhesion is a crucial first step that allows pathogenic bacteria, including C. difficile, to infect host cells. A group of virulence factors, MSCRAMMs, mediates adhesion of a wide variety of pathogenic bacteria, including Staphylococcus, Streptococcus, Enterococcus, Borrelia, Leptospira, and others (19,26,34,35,37,48,49). Generally, MSCRAMMs are located on the outer surface of bacteria thereby mediating attachment to host cells by interacting with fibrinogen or various ECM components such as Fn, laminin, collagen, elastin, and proteoglycan (19). A number of C. difficile MSCRAMMs that bind to either host cells or various ECM components have been described and include Fbp68 (11), SlpA (4), Cwp66 (14), and 27-kDa protein (50). These adhesins may be significant factors in the virulence of different C. difficile strains.

Manganese Binding to Fbp68
FEBRUARY 4, 2011 • VOLUME 286 • NUMBER 5 Staphylococcus aureus, are metalloproteins (26,27,41,51) In this study, we demonstrated that Fbp68 is a manganese-binding protein and also showed that it binds to manganese with high affinity and specificity (K D ϭ 52 Ϯ 1.97 nM). This is the first study to identify a bacterial manganese binding MSCRAMM. Because a higher concentration of free manganese would be an oxidative stress in the cells (52), the concentration of free manganese in vivo is extremely low (10 Ϫ7 M) (53). In addition, manganese is a trace element and is present at very low levels in the environment (10 Ϫ7 M) (52). To acquire environmental manganese, the binding affinity of most manganese-binding proteins must be high (e.g. E. coli manganese superoxide dismutase, K D ϭ 3.12 nM) (44). Thus, the high affinity of Fbp68 for manganese would help overcome the low concentration of manganese both in vivo and in the environment. Most metalloproteins undergo conformational changes upon metal ion binding. In certain extreme cases, apo-metalloproteins lose their conformation because metal ions stabilize the structure (54). In other cases, the geometry of the metal ion-binding sites is dominated by metalloproteins so the structure of apoproteins is able to be maintained although partial alteration of the conformation is still observed upon metal ion binding (55). Manganese binding changed the global conformation of Fbp68, but the structure of apo-Fbp68 can still be discerned in far-UV spectra. Thus, our results suggest that Fbp68 dominates the geometry of the manganese binding region.
Manganese usually enhances protein structural stability. For example, when manganese binds to manganese superoxide dismutase of E. coli, the protein structure is stabilized as evidenced by the increased T m of manganese-bound proteins in thermo-unfolding experiments (apo-superoxide dismutase, T m ϭ 52.5°C; manganese superoxide dismutase, T m ϭ 68.6°C) (44). Likewise, manganese-bound prion protein resists higher concentrations of protease K, which also indicates the protein structure is stabilized by manganese (apo-prion, protease K ϭ 2 g/ml; manganese prion, protease K Ͼ25 g/ ml) (43). In this study, the dramatically increased T m and the significantly decreased susceptibly of manganese-bound Fbp68 to protease K suggests that manganese binding stabilizes the structure of Fbp68, similar to other manganese-binding proteins. Enhanced Fbp68 stability upon manganese binding may aid C. difficile survival by preventing protease digestion, thereby providing a competitive advantage over other bacteria in maintaining a foothold in a highly competitive environment such as the gut.
Several recent studies show that metal ions modulate the functions of bacterial adhesins. Reportedly, calcium binding  (10 7 ). The adhesion of C. difficile to Caco-2 cells (10 5 ) was detected by ELISA. The reduced percentage of attachment was determined relative to the attachment of C. difficile to untreated Caco-2 cells. C, Fbp68C inhibits the binding of C. difficile to Caco-2 cells. Caco-2 cells (10 6 ) were pretreated with 50 M GST-Fbp68C or GST (negative control) prior to the addition of C. difficile (10 8 ). The adhesion of C. difficile or the binding of these proteins to Caco-2 cells were detected by CLSM. D and E, Fn and cell binding activity of wild type C. difficile (CD630WT), fbp68 mutant (CD⌬Fbp68 102 ), or fbp68 complemented strain (CD⌬Fbp68102/pMTL84151-fbp68). D, a total of 10 8 CD630WT, CD⌬Fbp68 102 , or CD⌬Fbp68102/pMTL84151-fbp68 were added to Fn-coated wells (1 M/well). The wells coated with 1 M BSA served as the negative control. Three CLSM fields were selected to count the number of bacteria that were bound to the Fn and BSA coated wells to determine the Fn binding activity of various genotypes of C. difficile as described under "Materials and Methods." E, a total of 10 8 cells of CD630WT, CD⌬Fbp68 102 , or CD⌬Fbp68102/pMTL84151-fbp68 were incubated in wells cultured with 10 6 Caco-2 cells. The wells incubated without Caco-2 cells served as the negative control. Cell binding activity was measured by the bacterial attachment assay as described under "Materials and Methods"; results are shown as the number of cell-associated bacteria per well. A, B, D, and E, each value represents the mean Ϯ S.E. of three trials in triplicate samples. Statistically significant differences (p Ͻ 0.05) are indicted by asterisk. C, CLSM settings were identical for all the captured images. Images were processed using Adobe Photoshop CS2. enhanced but was not essential for the Fn binding activities of LipL32 and Lig proteins from L. interrogans (26,27,41). However, ClfA binding to calcium reduces its fibrinogen binding affinity (51). Interestingly, we discovered that manganese not only improved Fbp68 binding to Fn but was required for Fn binding to Fbp68. Our results suggest that the mechanism by which manganese promotes the binding of clostridial Fbp68 to Fn differs from that of calcium bound to leptospiral LipL32 or Lig proteins. It is believed that metal ion binding dominates protein-protein interactions through several different mechanisms. In some cases, metal ion binding causes a conformational change, and the binding partner is able to selectively and specifically bind to holoproteins instead of apoproteins. A global structural transition was observed in Fbp68 upon manganese binding, which suggests a conformational change mediated the ability of Fbp68 to bind to Fn in the presence of manganese. Alternatively, metal ions can serve as a bridge to link a protein to its binding partner. For example, the general metal ion-dependent adhesion site on the ␣-subunit of integrin CR3 coordinates magnesium or manganese binding to its binding partners (56). However, Fn binding is entirely dependent upon the manganese within Fbp68, and Fbp68 lacks affinity for magnesium. Thus, manganese might form a cross-link between Fbp68 and Fn, but the binding motif should be different from the metal ion-dependent adhesion site of the ␣-subunit of integrin. As shown in Fig. 2 and Table  3, the stoichiometry of manganese binding to Fbp68 is five. Because Fbp68 possesses eight degenerated repeated sequences (11), it is possible that the manganese-binding motif of Fbp68 is located in these degenerated repeated sequences. However, an attempt to correlate these repeated amino acid sequences of Fbp68 with other known manganese-binding motifs was unsuccessful Thus, Fbp68 may utilize a novel manganese-binding motif.
C. difficile Fbp68 possesses high sequence similarity with other known Fn-binding proteins including Fbp54 of Streptococcus pyogenes (39% identity), PavA of Streptococcus pneumoniae (38% identity), FbpA of Streptococcus gordonii (30% identity), and Fbp of Bacillus subtilis (44% identity) (11,(57)(58)(59). The Fn binding region was mapped to the C-terminal 189 amino acids of PavA based on binding inhibition of full-length PavA or S. pneumoniae to immobilized Fn by truncated PavA without the C-terminal 189 amino acids (59). Similarly, Fn can only bind to Fbp68C, the C-terminal 194 amino acids, which show 42% sequence homology to the Fn binding region of PavA. Furthermore, the binding site for Fbp68 on Fn was localized to NTD, the heparin binding domain of Fn, consistent with PavA and Fbp54 (59). Taken together, it is likely that the mechanism of binding of PavA, Fbp54, and Fbp68 to Fn is similar. Recently, it was proposed that a group of similar NTD-binding motifs on diversified bacterial proteins such as FnbpA and FnbpB of S. aureus, FnBB of Streptococcus dysgalactiae, SfbI of S. pyogenes, and BBK32 of Borrelia burgdorferi possess a general binding mechanism called tandem ␤-zipper for bacterial MSCRAMMs binding to NTD (49, 60 -63). On the other hand, two NTD binding regions, unique Fn binding domain (UFbD) and repeated Fn binding domain (FbRD), were identified in Streptococcus PrtF1 and PrtF2 proteins (64). Neither of these Fn binding domains has sequence similarity with Fbp68, Fbp54, or PavA (data not shown). Thus, it is highly probable that Fbp68, Fbp54, and PavA possess a conserved but novel NTD-binding motif that utilizes an as yet undetermined Fn-binding mechanism.
ECM binding, including Fn binding, is regularly elicited by pathogens to adhere to host cells (19,48,65). Moreover, Fn can also serve as a mediator to induce endocytosis and initiate the entry of bacteria when it binds to bacterial Fn-binding proteins (65,66). Thus, the Caco-2 cell binding activity of Fbp68 and the reduced binding effects of C. difficile in Fbp68 or Fn siRNA treated Caco-2 cells suggest an adhesive role for Fbp68. However, the fbp68 knock-out mutant showed similar cell and Fn binding activities compared with wild type C. difficile 630, strongly suggesting multiple adhesins are present on the surface of C. difficile with redundant adhesive properties (5, 10 -11, 13-15). Furthermore, because C. difficile is not an intracellular pathogen, the rationale for C. difficile to bind to Fn is unlikely to be related to invasion as is the case for other bacteria. It was reported that the toxin of C. difficile could be digested by certain proteases in the cecum of mice (67) so adhesion of C. difficile may be able to target the toxin to the cell, thereby increasing toxin concentration at the cell surface and avoiding proteolysis, which would enhance toxin efficacy. On the other hand, another adhesin of C. difficile, SlpA, binds to ECM components and causes further epithelial damage following toxin-induced destruction of tissue (4). It has also been observed that the adherence of C. difficile to Caco-2 cells is enhanced through the addition of clostridial C. difficile transferase toxin (68). It is reasonable to hypothesize that Fbp68 and other adhesins of C. difficile might bind to ECM receptors that have been unmasked by tissue injury, resulting in more severe damage in the intestine and colon. Thus, clostridial toxins and adhesins may act synergistically in the pathogenesis of C. difficile-associated disease.
In conclusion, we have demonstrated that Fbp68 is a manganese-binding protein, manganese binding stabilizes the structure of Fbp68, and that Fbp68 binds to the NTD of Fn. In addition, this is the first study to identify a manganese-binding adhesin for Fn. Further studies to identify the binding motifs and define the mechanisms of Fn and manganese interaction with Fbp68 are ongoing in our laboratory.