INFECTION AND IMMUNITY, Feb. 1997, p. 609–616 0019-9567/97/$04.0010 Copyright q 1997, American Society for Microbiology
Vol. 65, No. 2
The Central Variable V2 Region of the CS31A Major Subunit Is Involved in the Receptor-Binding Domain PATRICK DI MARTINO,1 JEAN-PIERRE GIRARDEAU,2 MAURICE DER VARTANIAN,2 BERNARD JOLY,1 AND ARLETTE DARFEUILLE-MICHAUD1* Laboratoire de Bacte´riologie, Faculte´ de Pharmacie, Universite´ d’Auvergne, Clermont-Ferrand,1 and Laboratoire de Bacte´riologie, Institut National de la Recherche Agronomique, Saint Gene`s Champanelle,2 France Received 1 August 1996/Returned for modification 27 September 1996/Accepted 18 November 1996
CS31A is a K88-related capsule-like surface protein that mediates Escherichia coli and Klebsiella pneumoniae adhesion to the human Caco-2 and Intestine-407 cell lines. In this study, we demonstrate that ClpG, the major subunit of CS31A, contains the adhesive domain of the polymerized structure. We mapped this domain within the ClpG protein by performing adhesion inhibition experiments with Intestine-407 cells with nine synthetic peptides (CLP1 to CLP9) covering the dominant antigenic regions of ClpG and with the corresponding rabbit antipeptide antibodies. The peptides CLP1 (amino acid positions in parentheses) (5–18), CLP2 (44–56), CLP3 (82–96), CLP7 (174–190), CLP8 (185–199), and CLP9 (235–249) and corresponding antipeptide antibodies targeting parts of the N- and C-terminal regions of ClpG had no effect on the adhesion of the TCFF15 recombinant strain expressing CS31A. Only the CLP5 (132–146) peptide, corresponding to the central part of the protein, and relevant antibodies inhibited bacterial adhesion to intestinal epithelial cells. Anti-CLP4 (97–109) and anti-CLP6 (148–162) antibodies targeting regions surrounding the CLP5 sequence also inhibited bacterial adhesion. Site-directed mutagenesis experiments inducing changes in the amino acid sequence of the ClpG protein corresponding to the CLP5 peptide resulted in the expression of nonadhesive CS31A antigen. These findings indicate that the ClpG receptor-binding domain is located in the central variable V2 region. herence to mucosal surfaces of the intestinal tract (1). Three different K88 serotypes, termed K88ab, K88ac, and K88ad, have been described (12). The two most frequently isolated serotypes, K88ab and K88ac, promote different adhesive properties, since K88ab fimbriae agglutinate chicken, pig, and guinea pig erythrocytes, whereas K88ac fimbriae exhibit only weak hemagglutinating activity with guinea pig erythrocytes (3). The primary structure of the three corresponding major subunits has been determined and has been shown to contain both conserved and variable regions (16). The variable regions specifying the serotype-specific epitopes have been located within the FaeG major subunit and were found to be responsible for the differences observed in the adhesive properties between the serotypes (2). The three peptides, Ile-83–Ala– Phe-85, Ser-148–Leu–Phe-150, and Ala-156–Ile–Phe-158, derived from the K88 fibrillar major subunit, were found to inhibit the binding of K88 fibrillae to pig intestinal epithelial cells (14). In addition, site-directed mutagenesis and the construction of K88 hybrid proteins by exchange of variable regions of the fimbrial subunits demonstrated that residues at positions 134, 136, 147, 150, 155, 163 to 174, and 216 were involved in the constitution of the K88 receptor-binding domain (2, 14). Comparison of the predicted amino acid sequences of CS31A and K88 major subunits showed 46% similarity and revealed the presence of four regions of amino acid variability, called V1, V2, V3, and HV, interspersed among five conserved hydrophobic clusters (P1 to P5) of up to 20 residues (10). V2 region structure was found to be critical, since the insertions, deletions, or substitutions of amino acid residues completely suppressed CS31A biogenesis (4). Hydrophobic cluster analysis and secondary structure predictions revealed a common overall fold for the ClpG and FaeG proteins (25). Topological and epitope mapping of the CS31A major subunit has been achieved by using solid-phase peptide synthesis and polyclonal rabbit antibodies raised against native and denatured proteins (24). Peptides constituting linear antigenic epitopes on the
Bacterial colonization of epithelial surfaces of the host is often the first step before infection (7, 28). Enterobacterial strains can express fimbrial or nonfimbrial surface structures mediating bacterial adhesion to epithelial cells (13). CS31A is a capsule-like surface antigen harbored by bovine and human enterotoxigenic or septicemic Escherichia coli strains and by Klebsiella pneumoniae strains involved in nosocomial infections (9, 11, 15). Two different subtypes of the CS31A antigen, termed CS31A-H and CS31A-L, have been described (5). They differ by only one amino acid at position 89 of the mature protein, which results in different apparent molecular masses during sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and which results in the expression of an H-subtype-specific conformational epitope (9). These modifications do not alter adhesive ability in vitro, since both CS31A-L- and CS31A-H-producing bacteria adhere to the human intestinal cell lines Caco-2 and Intestine-407 and since their adhesion is inhibited by the same saccharides (6, 9). CS31A is encoded by large self-transmissible R plasmids which are thought to have facilitated the transfer of CS31A genetic determinants from E. coli to K. pneumoniae strains (6, 15). The CS31A adhesin belongs to the K88 adhesin family, because there is extensive nucleotide sequence homology throughout both gene clusters except in the subunit structural genes (22). Accessory proteins mediating expression of adhesins of the K88 family are functionally interchangeable (18). All of the members of this adhesin family share a similar genetic organization and are mainly composed of polymerization of one major subunit associated with minor components. K88 is a fimbrial structure produced by enterotoxigenic E. coli strains isolated from piglets which is involved in bacterial ad* Corresponding author. Mailing address: Laboratoire de Bacte´riologie, Faculte´ de Pharmacie, 28, Place Henri Dunant, 63001 Clermont-Ferrand, France. Phone: (33) 04 73 60 80 19. Fax: (33) 04 73 27 74 94. E-mail:
[email protected]. 609
610
DI MARTINO ET AL.
INFECT. IMMUN.
FIG. 1. Location of the nine CLP synthetic peptides and immunoreactive sites of ClpG. The CLP synthetic peptides used in bacterial adhesion inhibition experiments are boxed. The previously identified variable regions (V1, V2, V3, and HV) and conserved regions (P1, P2, P3, P4, and P5) between ClpG and FaeG major subunits are indicated by open and solid bars, respectively. The epitopes previously mapped on ClpG in peptide scanning experiments with polyclonal anti-ClpG antibodies are underlined in the ClpG sequence, and the epitopes recognized by antipeptide antibodies on ClpG are underlined in the CLP peptide sequences.
ClpG subunit were identified by examining the binding of antibodies raised against native CS31A antigen or denatured ClpG subunit to 249 overlapping nonapeptides covering the amino acid sequence of ClpG. Polyclonal antibodies raised against the immunogenic peptides identified were produced by rabbit immunizations with peptide-bovine serum albumin (BSA) conjugates. Anti-CLP2 (amino acid positions are in parentheses) (44–56), -CLP7 (174–190), -CLP8 (185–199), and -CLP9 (235–249) antibodies were shown to bind to the native CS31A protein in accessibility enzyme-linked immunosorbent assay (ELISA) experiments, indicating that regions 44 to 56, 174 to 190, 185 to 199, and 235 to 249 were surface exposed on CS31A (24). As previously shown for K88, variable and surface-exposed regions are thought to be involved not only in the antigenic specificity of CS31A but also in its adhesive properties. The aim of the present study was to map the receptorbinding domain of CS31A. For this purpose, we used the CLP synthetic peptides, which constitute linear antigenic epitopes on the ClpG subunit, and relevant antipeptide antibodies to inhibit adhesion of CS31A-producing bacteria to Intestine-407 cells. Site-directed mutagenesis experiments were also done to confirm the location of the binding domain. MATERIALS AND METHODS Bacterial strains and culture conditions. The E. coli recombinant strain TCFF15 expressing CS31A was used for bacterial adhesion inhibition tests (9). This strain harbored a transcomplementation system consisting of plasmid pDSPH524, which contains the CS31A accessory genes, and plasmid pCFF15, which contains the CS31A major subunit structural gene (8, 9). Recombinant plasmids were maintained in the E. coli host strain DH5a (Bethesda Research Laboratories, Inc.). Strains were grown at 378C overnight in Mueller-Hinton medium (broth or agar) (Institut Pasteur Production, Marnes-la-Coquette, France). Peptide ELISA. The CLP synthetic peptides covering ClpG immunoreactive sites and corresponding antipeptide sera used in this study have been described
elsewhere (23). Their position in the ClpG sequence is presented in Fig. 1. Immunoglobulins G (IgGs) were isolated from each antipeptide serum by protein A column chromatography (Pierce, Chicago, Ill.). After dialysis, the IgG concentrations were adjusted to that of the total IgG fraction of corresponding antipeptide sera. We determined the epitopes recognized by the antipeptide IgG by peptide-pin-based ELISA experiments. Duplicate sets of 249 overlapping nonapeptides covering the entire sequence of ClpG subunit protein were synthesized on solid polypropylene rods (Cambridge Research Biochemicals, United Kingdom). Successful peptide synthesis was monitored by the simultaneous synthesis of a positive and a negative control peptide and by subsequently testing the binding of these two peptides to the supplied corresponding monoclonal antibody. The synthesized peptides, on block pins configured to a 96-well microtiter plate, were then tested for binding with the different antipeptide IgGs in a peptide-pin-based ELISA. To prevent nonspecific absorption of antibodies, rods carrying the peptides were coated with 1% ovalbumin–1% BSA–0.1% Tween 20 in phosphate-buffered saline (PBS) for 1 h at room temperature. The pins were then incubated overnight at 48C with 100 ml of antipeptide IgG at an appropriate dilution per well. The pins were thoroughly washed with PBS-Tween 20 and incubated with 100 ml (per well) of peroxidase-labeled goat anti-rabbit antibodies at a dilution of 1:2,000 (Nordic Immunological Laboratories, Tilburg, The Netherlands) for 2 h at room temperature. After being washed with PBSTween 20, the pins were incubated with 100 ml (per well) of 2,29-azino-di-3-ethylbenzthiazoline-sulfonate hydrogen peroxide (ABTS) in phosphate citrate buffer for 20 min in the dark. Plates were read at 405 nm in a Dynatech MR5000 reader. These ELISA experiments were done twice with independent sets of synthesized peptides, and the results of the two experiments were compared. Extraction of bacterial surface proteins. The extraction of bacterial surface proteins was performed as previously described (6). Bacterial cultures grown overnight on Mueller-Hinton agar were harvested in PBS at pH 7.2. Bacterial surface proteins were separated from cells by heating the suspension at 608C for 20 min with gentle shaking. Cells and bacterial debris were removed by centrifugation at 12,000 3 g for 10 min and filtration at a pore diameter of 0.2 mm. The resulting supernatant was used for SDS-PAGE and Western blotting experiments. To test for adhesion inhibition by competing proteins, the supernatant was brought to pH 4.0 and stored overnight at 48C. The precipitated proteins were collected by centrifugation at 10,000 3 g for 30 min and suspended in 0.1 M PBS (pH 7.2). The final concentration of proteins was adjusted to 2 mg/ml. SDS-PAGE and immunoblotting. Crude protein extracts were fractionated by SDS-PAGE and stained with Coomassie brilliant blue R-250 or transferred onto nitrocellulose membranes as previously described (9). The membranes were blocked for 2 h at room temperature with Tris-buffered saline (10 mM Tris, 0.9%
VOL. 65, 1997
MAPPING OF THE CS31A RECEPTOR-BINDING DOMAIN
611
FIG. 2. Site-directed mutagenesis of the structural clpG gene in the V2 variable region. Nucleotides modified or deleted by mutagenesis and the resulting modified amino acid residues are underlined.
NaCl [pH 7.5]) and 2% (wt/vol) BSA. Filters were incubated with the antiCS31A serum and developed with peroxidase-labeled goat anti-rabbit antibodies (Nordic Immunological Laboratories). Bacterial adhesion to Intestine-407 cells. The Intestine-407 cell line, derived from human intestinal embryonic jejunum and ileum, was obtained from Flow Laboratories, Inc. (McLean, Va.). Monolayers of Intestine-407 cells were grown in an atmosphere containing 5% CO2, at 378C, in modified Eagle medium (Flow Laboratories, Les Ulis, France) containing 10% (vol/vol) fetal bovine serum, 1% nonessential amino acids, 200 U of penicillin per liter, and 50 mg of streptomycin per liter. Cell monolayers were grown to confluence on 24-well Falcon tissue culture plates (Becton Dickinson Labware, Oxnard, Calif.). Before adhesion tests, cells were washed once with PBS (pH 7.2). A suspension of 108 bacteria per ml of the cell line culture medium containing 1% (wt/vol) D-mannose was added to the tissue culture, and the mixture was then incubated for 3 h at 378C. After three washes with PBS, the cells were fixed in methanol, stained with 20% Giemsa stain, and examined microscopically under oil immersion. An adhesion index representing the average number of bacteria per cell was determined by examination of 100 cells. Bacterial inhibition experiments were done with synthetic peptides and with the corresponding antipeptide IgG. To test for adhesion inhibition by competing proteins, a suitable concentration of surface protein extracts was added to the Intestine-407 cell line. The mixture was incubated at 378C for 20 min, 108 bacteria per ml in the cell line culture medium containing 1% (wt/vol) D-mannose were added, and the adhesion test was performed as described above. To test for adhesion inhibition by antibodies, the specific polyclonal antipeptide antibodies were adsorbed against E. coli DH5a and adjusted to the same ELISA titer of 128. Serial dilutions of these antibodies were mixed with a suspension of 108 bacteria per ml in the cell line culture medium containing 1% (wt/vol) D-mannose and incubated together with shaking for 1 h at room temperature. Mixed bacterium-antibody solution was added to the tissue culture, and the adhesion test was performed as described above. Control experiments were done under the same conditions with normal whole IgG from preimmune rabbits. The ability of synthetic peptides to block the adhesion of the TCFF15 strain to Intestine-407 cells was determined as follows. Serial dilutions of synthetic peptides were preincubated with confluent Intestine-407 monolayers for 1 h at 378C. A suspension of 108 bacteria per ml in the cell line culture medium containing 1% (wt/vol) D-mannose was added to the cell-coated wells. The adhesion test was then performed as described above. Site-directed mutagenesis. In vitro site-directed mutagenesis in the clpG structural gene was performed on the plasmid pDEV41155 (8) by the method of Jung et al. (17) with a transformer site-directed mutagenesis kit purchased from Clontech Laboratories (Palo Alto, Calif.). Two oligonucleotides were used as primers. The selection primer mutated the single AflIII restriction site of the pUC18 plasmid vector into a single BglII site. With the changes at positions 406 to 414 in the clpG gene, the mutagenic primer created a EcoRI restriction site, and changed the amino acid sequence T136-G-L138 to R136-N-S-V139 (Fig. 2). The mutations were screened by the restriction of the plasmids by BglII and EcoRI enzymes. The incorporation of nucleotide exchanges was confirmed by DNA sequencing. The recombinant plasmid harboring the mutated clpG structural gene was termed pGV294. Nucleotide sequence determination. Sequencing of double-stranded DNA templates was performed by the dideoxy chain termination method of Sanger et al. (27) with Sequenase-modified T7 polymerase enzyme and a sequencing kit with [a-35S]dATP (U.S. Biochemical Corp., Cleveland, Ohio). Electron microscopy. E. coli recombinant strains producing wild-type or modified CS31A antigen were grown at 378C on Mueller-Hinton agar, harvested in
PBS (pH 7.2), and observed with a transmission electron microscope (HU 12A; Hitachi) after being stained with 1% phosphotungstic acid (pH 6.8). Gold immunolabeling was performed as described by Levine et al. (21). A washed bacterial suspension was placed on carbon-coated grids. Excess liquid was removed, and the grid was placed face down on a suitable dilution of adsorbed anti-CS31A serum for 15 min. After 10 washings, the grid was placed on a drop of gold-labeled antirabbit serum (Jansen Life Sciences Products, Olen, Belgium) for 15 min. After a further thorough washing, the grids were negatively stained with 1% ammonium molybdate. To prevent nonspecific labeling, 1% BSA and 1% Tween 20 were added to the wash solutions.
RESULTS Characterization of the antipeptide antibodies. Nine CLP synthetic peptides covering the dominant antigenic regions of ClpG were tested. The CLP2, CLP5, CLP6, and CLP8 peptide sequences overlapped the V1, V2, HV, and V3 variable regions observed between the ClpG and FaeG major subunits, respectively (Fig. 1). Polyclonal antipeptide antibodies have been previously obtained by rabbit immunizations with synthetic peptides conjugated to BSA (24). These antipeptide antibodies have been shown to react with their relevant unconjugated peptides and with the ClpG subunit in competitive ELISA and immunoblotting experiments (24). In this study, we determined the linear epitopes of the synthetic peptides by screening series of 249 overlapping nonapeptides in ELISA with the different antipeptide antibodies (Fig. 3). As judged by the reactivity pattern obtained with each of the nine antipeptide sera, peptide conjugates induced specific antibodies which recognized only overlapping peptides covering the sequence corresponding to their relevant immunogenic peptides. While anti-CLP1, -CLP3, -CLP4, -CLP5, -CLP8, and -CLP9 antibodies recognized only one epitope on peptide immunogens, antiCLP2, -CLP6, and -CLP7 antibodies produced a more complex response, reacting with at least two epitopes (Fig. 3). Except for anti-CLP9 antibodies, all of the antipeptide antibodies recognized epitopes that overlapped the epitopes previously mapped with anti-ClpG antibodies (Fig. 1). The differences observed between the epitopes recognized by antipeptide antibodies and those recognized by anti-ClpG antibodies may be due to differences between the conformation of the CLP synthetic peptides and that of corresponding amino acid sequences in the ClpG protein. Effect of synthetic peptides on CS31A-mediated bacterial adhesion to Intestine-407 cells. The nine synthetic peptides were assayed for their ability to inhibit the adherence of the recombinant strain TCFF15 expressing the CS31A antigen to
612
DI MARTINO ET AL.
INFECT. IMMUN.
FIG. 3. ELISA reactivities of antipeptide antibodies. The entire set of nonapeptides was subjected to epitope scanning with the antipeptide antibodies generated to peptides previously identified as linear epitopes of denatured ClpG subunit. Only the region of positive ELISA reactivity is shown in each panel. The value that identifies each nonapeptide is the amino acid position in the ClpG sequence corresponding to the amino acid end of a given peptide. Residues required for the binding of each antipeptide antibody to epitopes are shaded. OD, optical density.
Intestine-407 cells. As shown in Fig. 4A, only peptide CLP5 (132–146), covering part of the V2 variable region of the ClpG protein, was able to inhibit bacterial adhesion. The CLP5 peptide exhibited dose-dependent inhibition properties. Bacterial adhesion decreased by 8 and 46% in the presence of 50 and 200 mM CLP5 peptide, respectively (Fig. 4A). The CLP4 (97– 109) and CLP6 (148–162) peptides surrounding the CLP5 sequence in the ClpG protein did not have significant adhesion inhibition activity under the same conditions. No bacterial adhesion inhibition was obtained with peptides CLP2 (44–56), CLP7 (174–190), CLP8 (185–199), and CLP9 (235–249), corresponding to ClpG sequences previously shown to be surface exposed on native CS31A antigen. Effect of antipeptide IgG on the CS31A-mediated bacterial adhesion to Intestine-407 cells. All antipeptide IgGs were tested for their ability to inhibit the adhesion to the Intestine407 cell line of the recombinant strain TCFF15. Before use in inhibition experiments, the different antibodies were adjusted to the same ELISA titer of 128 by dilutions in PBS at pH 7.2. As shown in Fig. 4B, the adhesion of the recombinant strain TCFF15 to the epithelial cells was inhibited when bacteria were preincubated with anti-CLP4, -CLP5, and -CLP6 antibodies at a dilution of 2, by as much as 85, 70, and 65%, respectively. The adhesion inhibition was dose dependent, and
adhesion inhibition values were still 58.2, 34.5, and 28.8%, respectively, when IgGs were used at a dilution of 8. Under the same conditions, the anti-CLP1, -CLP2, and -CLP3 antibodies, corresponding to the N terminus of the ClpG protein; anti-CLP7, -CLP8, and -CLP9 antibodies, corresponding to the C terminus of the ClpG protein; and normal rabbit whole IgG from preimmune animals had no significant effect on the adhesion of the TCFF15 strain to Intestine-407 cells (Fig. 4B). The recombinant strain harboring mutated clpG structural gene expresses nonfunctional CS31A. Previous mutagenesis experiments to induce deletions or substitutions of amino acid residues in the V2 region of ClpG are shown in Fig. 2. None of the recombinant strains harboring these mutations produced CS31A (4). We performed oligonucleotide site-directed mutagenesis to induce changes in the amino acid sequence of the V2 region of the ClpG protein that did not interfere with the biogenesis of CS31A. As shown in Fig. 2, the sequence T136G-L138 was replaced by the sequence R136-N-S-V139 in the ClpG protein. The plasmid pGV294 harboring the mutated clpG gene was transformed into the DH5a strain containing the plasmid pDSPH524, which carries the CS31A accessory genes, to produce the modified CS31A antigen. The bacterial surface components of the TCFF294 E. coli strain, harboring
VOL. 65, 1997
MAPPING OF THE CS31A RECEPTOR-BINDING DOMAIN
FIG. 4. Inhibitory activity of synthetic peptides (A) and corresponding antipeptide IgG (B) on the CS31A-mediated bacterial adhesion to Intestine-407 cells. All specific polyclonal antipeptide antibodies were adsorbed against the E. coli recipient strain DH5a and adjusted to the same ELISA titer of 128 before use in inhibition experiments. The abilities of synthetic peptides and corresponding antipeptide IgG to inhibit bacterial adhesion are presented as the percent inhibition of the binding of the recombinant strain TCFF15 to Intestine-407 cells with respect to the level of TCFF15 adhesion to this cell line in the absence of competitors.
both pGV294 and pDSPH524 plasmids, were extracted as previously described (6) and fractionated by SDS-PAGE. The TCFF294 strain produced a 29-kDa surface protein reacting with the CS31A antiserum in Western immunoblotting experiments (Fig. 5). Moreover, immunogold labeling with a specific anti-CS31A serum showed that both TCFF15 and TCFF294 recombinant strains expressed morphologically similar CS31A antigen at the cell surface (Fig. 6). The TCFF15 strain expressing wild-type CS31A adhered to Intestine-407 cells with an adhesion index of 12.30 bacteria per eucaryotic cell (Fig. 7A), but the TCFF294 strain did not adhere; its adhesion index was 0.05 bacteria per cell (Fig. 7B). Surface protein extracts from TCFF15 and TCFF294 strains were used to inhibit the adhesion of the TCFF15 recombinant strain. The TCFF15 strain no longer adhered when the Intestine-407 cells were preincubated with wild-type CS31A antigen extract (adhesion index of 0.60 after pretreatment). When the Intestine-407 cells were pretreated with modified CS31A extract prepared from the TCFF294 strain, the adhesion index of the TCFF15 strain was 11.79, and that of the untreated control was 12.30.
613
its receptor-binding domain. CS31A is a surface antigen expressed by pathogenic E. coli and K. pneumoniae strains (9, 11, 15). Previous studies showed that CS31A is an adhesive structure implicated in bacterial attachment to human intestinal epithelial cells in vitro (6, 9, 19). Pretreatment of the bacteria with antibodies raised against the CS31A major subunit or pretreatment of eucaryotic cells with purified CS31A inhibited the bacterial adherence to Caco-2 cells (6). Moreover, E. coli recombinant strains harboring the cloned CS31A determinants were found to adhere to Caco-2 and Intestine-407 cells (9, 19). CS31A is a K88-related capsule-like surface protein (6, 11). The major subunits of the K88-related family do not possess any cysteine residue, but conserved glycine and proline residues are thought to be involved in maintaining the overall structure of the subunit. In the K88 fimbriae, the major subunit that makes up the polymerized structure is also the functional protein that mediates bacterial adhesion to epithelial cells. The residues involved in the constitution of the K88 receptor-binding domain were shown to be localized in the central part and in the C terminus of the protein (2, 14). The CS31A antigen belongs to the K88 adhesin family on the basis of extensive nucleotide sequence homology in the accessory genes (22). Nevertheless, CS31A and K88 structural genes encoding for major subunits shared weaker homology. There are four variable regions between FaeG and ClpG displayed all along the protein sequences, called V1, V2, V3, and HV (10). Peptides constituting linear antigenic epitopes on the ClpG subunit have been previously identified (24). These peptides covered both variable and conserved regions of the CS31A major subunit. In the present study, we mapped the epitopes recognized by antipeptide antibodies on the ClpG protein. The peptide scanning experiments with antipeptide antibodies and antiClpG antibodies showed that the different sets of antibodies did not recognize exactly the same epitopes. Thus, the conformation of the CLP synthetic peptides may be slightly different from that of the corresponding amino acid sequence in the ClpG protein. Nevertheless, we showed that the antipeptide antibodies recognized epitopes that overlapped the epitopes previously identified with anti-ClpG antibodies (24). In searching for the location of the receptor-binding domain on the CS31A adhesin, we first investigated the inhibitory activity of synthetic peptides derived from the ClpG sequence and of corresponding antipeptide antibodies. The synthetic peptide CLP5 localized in the V2 region was the only peptide to have significant inhibitory activity. Thus, the synthetic CLP5 peptide may mimic its conformation in the native CS31A antigen and may recognize the CS31A receptor at the surface of the Intestine-407 cells. This is consistent with the fact that most of the residues of FaeG that were demonstrated to be part of the K88 receptor-binding domain correspond to residues localized in or near the V2 and HV regions of the ClpG protein (2). Moreover, of the nine antipeptide sera tested in the ad-
DISCUSSION The results presented in this study clearly demonstrate that the CS31A major subunit ClpG is the adhesin of the polymerized structure and localize the region of the protein involved in
FIG. 5. SDS-PAGE (A) and Western blotting (B) analysis of surface protein extracts produced by recombinant strains expressing wild-type or mutated CS31A antigens. Lanes: 1, DH5a; 2, TCFF15; 3, TCFF294. The anti-CS31A serum was used at a dilution of 1:200.
614
DI MARTINO ET AL.
INFECT. IMMUN.
FIG. 6. Transmission electron micrographs after colloidal gold immunolabeling with antibodies raised against CS31A. (A) E. coli TCFF15. (B) E. coli TCFF294. Bars, 0.2 mm.
hesion inhibition assay, only antibodies raised against peptides localized in or near the central variable V2 region and the hypervariable HV region of ClpG had inhibitory activity against the CS31A-mediated bacterial adhesion to Intestine-407 cells. Residues surrounding the V2 region in the ClpG protein may also be involved in the formation of the CS31A adhesive domain, since anti-CLP4 and anti-CLP6 antibodies inhibited the CS31A-mediated bacterial adhesion to epithelial cells. However, the fact that peptides CLP4 and CLP6 did not inhibit bacterial adhesion suggests a steric hindrance by these antipeptide antibodies, which could physically recover the CS31A receptor-binding domain after their fixation in surrounding sequences. Anti-CLP4, -CLP5, or -CLP6 antibodies inhibited the CS31Amediated bacterial adhesion to epithelial cells, but previous accessibility and competitive ELISA experiments failed to detect a clear bond to the native form of the CS31A antigen (24). Leder et al. (20) reported a similar phenomenon of variable recognition of a native protein by corresponding antipeptide antibodies that was dependent upon the technique used. This
may be due to the existence of several conformational isomers of the synthetic peptides, with only one isomer simulating its natural conformation in the intact protein. Because this conformational isomer is in the minority, the antibodies raised against it are rare. The anti-CLP4, -CLP5, and -CLP6 antibodies may contain only a tiny fraction of antibodies that crossreact with native CS31A. Nevertheless, at the weaker concentrations used in adhesion experiments, there may have been enough antibody molecules reacting with CS31A antigen to prevent bacterial adhesion to epithelial cells. An alternative explanation is that recognition of the native CS31A adhesin by anti-CLP4, -CLP5, and -CLP6 antibodies may depend upon the interaction of this adhesin with its eucaryotic cell ligand. The CS31A adhesive domain could be inaccessible to antiCLP4, -CLP5, and -CLP6 antibody binding until the fixation to its eucaryotic cell ligand. Such observations have been previously reported for the interaction of the picornaviruses with eucaryotic cells (26). Indeed, there is some evidence that deep pits or “canyons” in the surface of the picornavirus virion contain the receptor-binding domain of the viruses (26). It has
FIG. 7. Micrographs showing the adherence to Intestine-407 cells of E. coli recombinant strains TCFF15 (A) and TCFF294 (B), producing wild-type or modified CS31A, respectively.
VOL. 65, 1997
MAPPING OF THE CS31A RECEPTOR-BINDING DOMAIN
been shown that conformational changes were induced in the canyon by the binding of small organic molecules (26). Thus, the binding of CS31A to its eucaryotic cell ligand could induce conformational changes in CS31A leading to the exposure of the V2 region (corresponding to peptides CLP4, CLP5, and CLP6) on the surface of the protein. This surface-exposed region could then be recognized by anti-CLP4, anti-CLP5, and anti-CLP6 antibodies. Such conformational changes could constitute a strategy for microorganisms to escape the host’s immune surveillance by protecting the receptor attachment site in a surface depression. Mutant strains harboring deletions or amino acid substitutions in or near the V2 region of the CS31A structural gene have been previously obtained (4). None of these mutant strains exported CS31A to the cell surface, which suggests that modifications in or near the V2 region of ClpG interfere with the biogenesis of CS31A. As for the CS31A antigen, deletion of regions coding for variable parts of the P fimbrillin has been shown to strongly reduce or abolish fimbriae production (30). It has been suggested that these variable regions are important in the biogenesis of adhesive factors because they impose correct spacing between conserved regions of the proteins. To confirm that it was involved in the formation of the CS31Abinding domain, we modified the V2 region without interfering with the integrity of CS31A by changing the amino acids at positions 136, 137, and 138 in the ClpG protein by site-directed mutagenesis experiments. The mutated ClpG protein was exported to the cell surface as shown by SDS-PAGE, Western blotting, and immunogold labeling experiments. Nevertheless, the modified CS31A antigen did not promote bacterial adherence to Intestine-407 cells and did not inhibit wild-type CS31A-mediated bacterial adhesion, which shows that the V2 region is essential for the adhesive functions of the CS31A antigen. The conformational constraints of the variable V2 region that is folding with a b-conformation (24, 25) may be important for the CS31A-mediated bacterial adhesion to the intestinal epithelial cell surface. A rigid structure consisting of two b-turns has also been shown to be critical for the binding of Pseudomonas aeruginosa PAK pili to pneumocyte cells (23). Bacterial adhesins like CS31A are involved in the microbial adherence to epithelial cells that leads to colonization of the host. CS31A is expressed by E. coli strains involved in bovine and human diarrhea and septicemia and by K. pneumoniae strains involved in nosocomial infections (5, 6, 15). E. coli and K. pneumoniae strains producing CS31A are multiresistant strains, and it is difficult to decontaminate patients and animals by antibiotherapy. The aim of studies of bacterial adhesins is to explore the possibilities of using microbial attachment inhibition as a means of preventing or curing infectious diseases without recourse to antibiotics. Strategies based on synthetic vaccines obtained by immunizations with peptide-carrier protein conjugates using synthetic peptides mimicking the receptor-binding domain of bacterial adhesins have already been successfully explored for P. aeruginosa type IV pili in a mouse model (29). Further progress in our understanding of the CS31A-binding domain may lead to a test for antiadhesive immunotherapeutic vaccine strategies.
ACKNOWLEDGMENTS This work was supported by the Ministe`re de la Recherche et de la Technologie. We thank the Microscopy Department of Michel Bourges for technical assistance in electron microscopy analysis.
615
REFERENCES 1. Anderson, M. J., J. S. Whitehead, and Y. S. Kim. 1980. Interaction of Escherichia coli K88 antigen with porcine intestinal brush border membranes. Infect. Immun. 29:897–901. 2. Bakker, D., P. T. J. Willemsen, L. H. Simmons, F. G. Van Zijderveld, and F. K. de Graaf. 1992. Characterization of the antigenic and adhesive properties of FaeG, the major subunit of K88 fimbriae. Mol. Microbiol. 6:247– 255. 3. Bijlsma, S. G. M., and J. F. Frik. 1987. Haemagglutination patterns of the different variants of Escherichia coli K88 antigen with porcine, bovine, guinea pig, chicken, ovine, and equine erythrocytes. Res. Vet. Sci. 43:122–123. 4. Bousquet, F., C. Martin, J. P. Girardeau, M. C. Me´chin, M. Der Vartanian, H. Laude, and M. Contrepois. 1994. CS31A capsule-like antigen as an exposure vector for heterologous antigenic determinants. Infect. Immun. 62:2553–2561. 5. Contrepois, M., Y. Bertin, J. P. Girardeau, B. Picard, and P. Goullet. 1993. Clonal relationships among bovine pathogenic Escherichia coli producing surface antigen CS31A. FEMS Microbiol. Lett. 106:217–222. 6. Darfeuille-Michaud, A., C. Jallat, D. Aubel, D. Sirot, C. Rich, J. Sirot, and B. Joly. 1992. R-plasmid-encoded adhesive factor in Klebsiella pneumoniae strains responsible for human nosocomial infections. Infect. Immun. 60:44– 55. 7. De Champs, C., M. P. Sauvant, C. Chanal, D. Sirot, N. Gazuy, R. Malhuret, J. C. Baguet, and J. Sirot. 1989. Prospective survey of colonization and infection caused by expanded-spectrum-b-lactamase-producing members of the family Enterobacteriaceae in an intensive care unit. J. Clin. Microbiol. 27:2887–2890. 8. Der Vartanian, M., M. C. Me´chin, B. Jaffeux, Y. Bertin, I. Fe´lix, and B. Gaillard-Martinie. 1994. Permissible peptide insertions surrounding the signal peptide-mature protein junction of the ClpG prepilin: CS31A fimbriae of Escherichia coli as carriers of foreign sequences. Gene 148:23–32. 9. Di Martino, P., Y. Bertin, J.-P. Girardeau, V. Livrelli, B. Joly, and A. Darfeuille-Michaud. 1995. Molecular characterization and adhesive properties of CF29K, an adhesin of Klebsiella pneumoniae strains involved in nosocomial infections. Infect. Immun. 63:4336–4344. 10. Girardeau, J.-P., Y. Bertin, C. Martin, M. Der Vartanian, and C. Boeuf. 1991. Sequence analysis of the clpG gene, which codes for surface antigen CS31A subunit: evidence of an evolutionary relationship between CS31A, K88, and F41 subunit genes. J. Bacteriol. 173:7673–7683. 11. Girardeau, J. P., M. Der Vartanian, J. L. Ollier, and M. Contrepois. 1988. CS31A, a new K88-related fimbrial antigen on bovine enterotoxigenic and septicemic Escherichia coli strains. Infect. Immun. 56:2180–2188. 12. Guine´e, P. A. M., and W. H. Janssen. 1979. Behavior of Escherichia coli K antigens K88ab, K88ac, and K88ad in immunoelectrophoresis, double diffusion, and hemagglutination. Infect. Immun. 23:700–705. 13. Hultgren, S. J., S. Abraham, M. Caparon, P. Falk, J. W. St. Geme III, and S. Normark. 1993. Pilus and nonpilus bacterial adhesins: assembly and function in cell recognition. Cell 73:887–901. 14. Jacobs, A. A. C., J. Venema, R. Leeven, H. Van Pelt-Heerschap, and F. K. de Graaf. 1987. Inhibition of adhesive activity of K88 fibrillae by peptides derived from the K88 adhesin. J. Bacteriol. 169:735–741. 15. Jallat, C., A. Darfeuille-Michaud, J.-P. Girardeau, C. Rich, and B. Joly. 1994. Self-transmissible R plasmids encoding CS31A among human Escherichia coli strains isolated from diarrheal stools. Infect. Immun. 62:2865– 2873. 16. Josephsen, J., F. Hansen, F. K. De Graaf, and W. Gaastra. 1984. The nucleotide sequence of the protein subunit of the K88ac fimbriae of porcine enterotoxigenic Escherichia coli. FEMS Microbiol. Lett. 25:301–306. 17. Jung, R., M. P. Scott, L. O. Oliveira, and N. C. Nielsen. 1992. A simple and efficient method for the oligodeoxyribonucleotide direct mutagenesis of double-stranded plasmid DNA. Gene 121:17–24. 18. Korth, M. J., J. M. Apostol, Jr., and S. L. Moseley. 1992. Functional expression of heterologous fimbrial subunits mediated by the F41, K88, and CS31A determinants of Escherichia coli. Infect. Immun. 60:2500–2505. 19. Korth, M. J., R. A. Schneider, and S. L. Moseley. 1991. An F41-K88-related genetic determinant of bovine septicemic Escherichia coli mediates expression of CS31A fimbriae and adherence to epithelial cells. Infect. Immun. 59:2333–2340. 20. Leder, L., H. Wendt, C. Schwab, I. Jelesarov, S. Bornhauser, F. Ackermann, and H. R. Bosshard. 1994. Genuine and apparent cross-reaction of polyclonal antibodies to proteins and peptides. Eur. J. Biochem. 219:73–81. 21. Levine, M. M., P. Ristaino, G. Marley, C. Smyth, S. Knutton, E. Boedeker, R. Black, C. Young, M. L. Clements, C. Cheney, and R. Patnaik. 1984. Coli surface antigens 1 and 3 of colonization factor antigen II-positive enterotoxigenic Escherichia coli: morphology, purification, and immune responses in humans. Infect. Immun. 44:409–420. 22. Martin, C., C. Boeuf, and F. Bousquet. 1991. Escherichia coli CS31A fimbriae: molecular cloning, expression and homology with the K88 determinant. Microb. Pathog. 10:429–442. 23. McInnes, C., F. D. So ¨nnichsen, C. M. Kay, R. S. Hodges, and B. D. Sykes. 1993. NMR solution structure and flexibility of a peptide antigen represent-
616
DI MARTINO ET AL.
ing the receptor binding domain of Pseudomonas aeruginosa. Biochemistry 32:13432–13440. 24. Me´chin, M.-C., E. Rousset, and J. P. Girardeau. 1996. Identification of surface-exposed linear B-cell epitopes of the nonfimbrial adhesin CS31A of Escherichia coli by using overlapping peptides and antipeptide antibodies. Infect. Immun. 64:3555–3564. 25. Me´chin, M. C., Y. Bertin, and J. P. Girardeau. 1995. Hydrophobic cluster analysis and secondary structure predictions revealed that major and minor structural subunits of K88-related adhesins of Escherichia coli share a common overall fold and differ structurally from other fimbrial subunits. FEBS Lett. 364:319–324. 26. Rossmann, M. G. 1989. The canyon hypothesis. Hiding the host cell receptor attachment site on a viral surface from immune surveillance. J. Biol. Chem. 264:14587–14590.
Editor: P. E. Orndorff
INFECT. IMMUN. 27. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74:5463–5467. 28. Selden, R., S. Lee, W. L. L. Wang, J. V. Bennett, and T. C. Eickhoff. 1971. Nosocomial Klebsiella infections: intestinal colonisation as a reservoir. Ann. Intern. Med. 74:657–664. 29. Sheth, H. B., L. M. G. Glasier, N. W. Ellert, P. Cachia, W. Kohn, K. K. Lee, W. Paranchych, R. S. Hodges, and R. T. Irvin. 1995. Development of an anti-adhesive vaccine for Pseudomonas aeruginosa targeting the C-terminal region of the pilin structural protein. Biomed. Peptides Proteins Nucleic Acids 1:141–148. 30. Van Die, I., M. Wauben, I. Van Megen, H. Bergmans, N. Riegman, W. Hoekstra, P. Pouwels, and B. Enger-Valk. 1988. Genetic manipulation of major P-fimbrial subunits and consequences for formation of fimbriae. J. Bacteriol. 170:5870–5876.
