Molecular Basis for the Enterocyte Tropism Exhibited bySalmonella typhimurium Type 1 Fimbriae*

Salmonella typhimurium exhibits a distinct tropism for mouse enterocytes that is linked to their expression of type 1 fimbriae. The distinct binding traits ofSalmonella type 1 fimbriae is also reflected in their binding to selected mannosylated proteins and in their ability to promote secondary bacterial aggregation on enterocyte surfaces. The determinant of binding in Salmonella type 1 fimbriae is a 35-kDa structurally distinct fimbrial subunit, FimHs, because inactivation of fimHs abolished binding activity in the resulting mutant without any apparent effect on fimbrial expression. Surprisingly, when expressed in the absence of other fimbrial components and as a translational fusion protein with MalE, FimHs failed to demonstrate any specific binding tropism and bound equally to all cells and mannosylated proteins tested. To determine if the binding specificity ofSalmonella type 1 fimbriae was determined by the fimbrial shaft that is intimately associated with FimHs, we replaced the amino-terminal half of FimHs with the corresponding sequence from Escherichia coli FimH (FimHe) that contains the receptor binding domain of FimHe. The resulting hybrid fimbriae bearing FimHes on aSalmonella fimbrial shaft exhibited binding traits that resembled that of Salmonella rather than E. coli fimbriae. Apparently, the quaternary constraints imposed by the fimbrial shaft on the adhesin determine the distinct binding traits of S. typhimurium type 1 fimbriae.

Salmonella typhimurium exhibits a distinct tropism for mouse enterocytes that is linked to their expression of type 1 fimbriae. The distinct binding traits of Salmonella type 1 fimbriae is also reflected in their binding to selected mannosylated proteins and in their ability to promote secondary bacterial aggregation on enterocyte surfaces. The determinant of binding in Salmonella type 1 fimbriae is a 35-kDa structurally distinct fimbrial subunit, FimHS, because inactivation of fimHS abolished binding activity in the resulting mutant without any apparent effect on fimbrial expression. Surprisingly, when expressed in the absence of other fimbrial components and as a translational fusion protein with MalE, FimHS failed to demonstrate any specific binding tropism and bound equally to all cells and mannosylated proteins tested. To determine if the binding specificity of Salmonella type 1 fimbriae was determined by the fimbrial shaft that is intimately associated with FimHS, we replaced the amino-terminal half of FimHS with the corresponding sequence from Escherichia coli FimH (FimHE) that contains the receptor binding domain of FimHE. The resulting hybrid fimbriae bearing FimHES on a Salmonella fimbrial shaft exhibited binding traits that resembled that of Salmonella rather than E. coli fimbriae. Apparently, the quaternary constraints imposed by the fimbrial shaft on the adhesin determine the distinct binding traits of S. typhimurium type 1 fimbriae.
Salmonella typhimurium is a highly evolved pathogen that has adapted to invading and surviving for long periods in its human and mouse hosts (1). The virulence of the organism has been attributed in part to its capacity to invade enterocytes and macrophages and to survive and grow intracellularly within these cells. A cluster of contiguous genes located in three pathogenicity islands, SPI-1, SPI-2, and the SelC locus in the chromosome of S. typhimurium, has been implicated these activities (2,3). Many of the genes mediating bacterial entry into host cells are located in SPI-1 (2). The entry mechanisms of Salmonella are quite complex, involving an intimate interaction between the bacterium and the host cell which results in the "exchange" of biochemical signals. As a consequence of this interaction, a complex set of signaling events is triggered in the host cell, leading to marked cytoskeletal rearrangements, membrane ruffling, and bacterial uptake by macropinocytosis (4). Central to these bacteria-host interactions is the activation in the Salmonella spp. of a specialized protein secretion system that directs the export of a variety of proteins and, in some cases, their direct translocation into the host cell (4,5). This bacterial secretion system, referred to as type III secretion system, does not appear to have any host cell specificity and has been shown to promote bacterial invasion of a wide range of host cells. Yet, in nature Salmonella exhibits a specific tropism for the gut epithelium of the host, and the primary targets for the invading Salmonella, at least in the initial stages of infection, are typically the cells lining the gastrointestinal tract.
S. typhimurium expresses several hair-like appendages (fimbriae) on its surface, whose function is as yet unclear. Since the fimbriae protrude peritrichously from the bacterial surface, it is highly likely that these structures will greatly influence the nature and intensity of any interactions that might occur between bacteria and host cells. The most commonly expressed fimbriae on Salmonella are the type 1 fimbriae, which are characterized by their capacity to bind mannose residues on host cell surfaces or in soluble compounds (6). That these structures contribute to the pathogenicity of S. typhimurium may be deduced from the findings that type 1-fimbriated Salmonella are significantly more virulent than their fim minus counterparts (7,41) and from the recent observations that type 1 fimbriae are expressed in vivo by S. typhimurium during an infection (8). The type 1 fim gene cluster is located at 15 min in the S. typhimurium chromosome (9); although not associated with any known pathogenicity islands, the fim cluster maps closely to a tRNA gene (10). Since most pathogenicity islands are located next to tRNA genes (2), the fim locus of Salmonella may be part of an as yet unrecognized pathogenicity island.
Type 1 fimbriae are expressed by Salmonella as well as most species of bacteria in the family Enterobacteriaceae (11). Although type 1 fimbriae are defined by their capacity to mediate mannose-specific binding reactions, it has become increasingly clear that there exists significant heterogeneity among type 1 fimbriae from different genera and even within the same species in their binding affinities for mannosylated compounds with defined oligomannose motifs. For example, Firon et al. (12) have reported a striking difference between the binding specificity of type 1 fimbriae of S. typhimurium and that of Escherichia coli and Klebsiella pneumoniae type 1 fimbriae. Madison et al. (13) demonstrated subtle but distinct differences in binding specificity between recombinant type 1 fimbriae of E. coli and K. pneumoniae. More recently, it has been reported that there is heterogeneity in binding to different mannosylated glycoproteins among type 1 fimbriae of different E. coli strains (14). These studies have also provided some clues to the underlying molecular basis for the heterogeneity in type 1 fimbrial binding. Whereas Sokurenko et al. (14) have linked heterogeneity in the binding among type 1 fimbriae of E. coli to differences in the covalent structure in their adhesin, FimHE, Madison et al. (13) have implicated the fimbrial shaft in defining the fine binding specificity between E. coli and K. pneumoniae type 1 fimbriae. In view of the tropism exhibited by Salmonella for gut tissue, the association of type 1 fimbrial expression with virulent strains of Salmonella, and the heterogeneity in binding specificity among type 1 fimbriae of Salmonella and other enterobacteria, we hypothesized that the type 1 fimbriae of S. typhimurium may play a critical role in modulating bacterial tropism to the gut of the host. By virtue of their distal location on the bacterium and their affinity for specific mannosylated residues, they could determine which tissues are ultimately invaded by Salmonella.
We have previously identified the corresponding mannosebinding adhesins on E. coli, K. pneumoniae, Serratia marcescens, and Enterobacter cloacae, and we have shown that their sizes ranged between 27 and 29 kDa and that they are antigenically conserved (15). Moreover, the gene encoding the adhesin is invariably at the distal end of the fim gene cluster. The complete DNA sequence of the Salmonella fim gene cluster has been determined by Clegg and Swenson (25) (GenBank TM accession number L19338), and it revealed some striking differences in composition and location of genes with that of other enteric fim gene clusters. Based on predicted amino acid sequences of the products of the Salmonella fim gene cluster, it was not possible to identify the fimbrial adhesin. Nevertheless, a gene located in the middle of the fim cluster was named fimH, because it exhibited limited homology with the adhesin of S fimbriae of E. coli. Thus, in addition to testing the contribution of Salmonella type 1 fimbriae to tissue tropism, we were interested in identifying and characterizing the putative adhesin.
Here, we demonstrate that Salmonella type 1 fimbriae mediate bacterial tropism for mouse gut epithelial cells (enterocytes) because, unlike type 1 fimbriae of E. coli, Salmonella type 1 fimbriae promoted strong bacterial binding to enterocytes but not to bladder epithelial cells. The distinct binding traits of Salmonella type 1 fimbriae were also reflected in their selective binding to particular mannosylated proteins and in mediating secondary aggregation of bacteria on the surface of enterocytes. Inactivation of a gene (fimHS) encoding a 35-kDa structurally distinct protein (fimHS) located in the middle of the fim cluster resulted in loss of binding activity without any loss of fimbrial expression. Surprisingly, when FimHS was expressed in the absence of other fimbrial proteins, the adhesin bound equally to enterocytes and bladder cells as well as to all mannosylated proteins tested. To see if the fimbrial shaft was responsible for the distinct binding specificity exhibited by the fimbriae, we replaced the amino-terminal half of the FimHS with the corresponding sequence from FimHE, which contained the receptor binding domain of FimHE. The resulting hybrid fimbriae expressed a chimeric adhesin (FimHES) on a shaft comprised of Salmonella fimbrial protein, and its binding traits resembled that exhibited by Salmonella fimbriae. Our findings suggest that although the adhesin is the structurally distinct FimH, the tropism exhibited by S. typhimurium type 1 fim-briae is determined by the fimbrial filament on which the adhesin is presented.

Bacterial Strains, Culture, and Plasmids
The bacterial strains and plasmids used in this work are described in Table I. All strains were cultured in brain heart infusion broth or in Luria broth (BBL Microbiology Systems, Cockeysville, MD).

Culture of Mouse Gut and Bladder Epithelial Cells
The immortalized mouse proximal small intestinal epithelial cell line SI-H10 was established from LFABP Ϫ596 to ϩ21 /tsA58 transgenic mice (16). These cells represent one of the first gut epithelial cell lines of mouse origin to be used to demonstrate adherence of enteric pathogens. These SI-H10 cells were cultured in Dulbecco's modified Eagle's medium (BioWhittaker, Walkersville, MD) containing 20% fetal calf serum (HyClone, Logan, UT) and 50 units of penicillin/50 g of streptomycin/ml (Life Technologies, Inc.) at 39°C in a humidified atmosphere of 95% air, 5% CO 2 . Immortalized mouse bladder epithelial cells MM45T.BL were obtained from American Type Culture Collection (ATCC, Rockville, MD). These cells were cultured in Dulbecco's modified Eagle's medium with 10% heat-inactivated fetal calf serum at 37°C in a humidified atmosphere of 95% air, 5% CO 2 .

3-[4,5-Dimethylthiazol-2-yl]-2,5-diphenyltetrazolium Bromide (MTT) 1 Bacterial Adherence Assay
This assay is based on the principle that live bacteria convert MTT intracellularly to purple formazan in direct proportion to viable bacterial cell number (17). The dye MTT (Sigma) is a yellow tetrazolium salt. MTT is reduced by dehydrogenases (expressed intracellularly by the bacteria) to form purple formazan with maximum absorbance at 570 nm. The crystals of formazan that are produced could be observed microscopically in individual bacterial cells and could be quantitated in a multiwell scanning spectrophotometer (enzyme-linked immunosorbent assay reader). This MTT bacterial adherence assay is a new, simple, inexpensive, and rapid colorimetric assay that we have developed in our laboratory to study the adherence of live, metabolically active bacteria. All the bacteria used in this study processed the MTT equally and produced the same amount of formazan. Increasing concentrations of bacteria produced corresponding increases in formazan as measured by A 570 . Dead bacteria and paraformaldehyde-fixed enterocytes or bladder epithelial cells do not react with MTT.
Cultures of bacteria were grown in static flasks at 37°C in LB until bacteria were in mid-log phase growth. Bacteria were harvested and washed twice in phosphate-buffered saline (PBS) by centrifugation and resuspended in PBS. The suspensions were adjusted to 1 ϫ 10 9 bacteria/ml (1 OD), and the suspensions were kept on ice until used.
Enterocytes (SI-H10 cells) or bladder epithelial cells (MM45T.BL cells) at a concentration of 0.15 ϫ 10 6 /ml were seeded in a 96-well tissue culture plate and allowed to grow in enriched medium (at 39°C in the case of SI-H10 cells or 37°C in the case of MM45T.BL cells) with 5% CO 2 . Under these conditions, most of the cells attached to the plastic surface of the dish and formed a monolayer. Confluent monolayers were washed three times with sterile PBS and were fixed overnight at 4°C in sterile PBS containing 1.5% paraformaldehyde. The monolayers were washed three times with sterile PBS and pretreated for 1 h at room temperature with blocking buffer (3% bovine serum albumin in PBS). Various amounts of bacteria were incubated with the monolayers for 1 h at 37°C in the presence and absence of ␣mM. Following adherence incubation, nonadherent bacteria were removed by washing the cell monolayer 3 times with PBS. Fifty l of LB was applied to each monolayer and incubated for 15 min at 37°C. MTT was dissolved in PBS at 2 mg/ml, filter-sterilized, and stored at 4°C until use. Fifty l of 2 mg/ml MTT was added to all wells, and the plates were incubated for 15 min at 37°C to allow reduction of MTT to formazan by live bacteria. The plates were centrifuged for 5 min at 1600 ϫ g to pellet the insoluble formazan-containing bacteria. Fluid from the wells was removed by inverting and gently blotting plates onto absorbent paper. 150 l of isopropyl alcohol was added to the wells, and the plates were sealed by Sealpette adhesive film (Midwest Scientific, St. Louis, MO) and stored in the dark for 15 min at 37°C to solubilize the formazan. The remaining formazan crystals were completely dissolved by mixing the well contents before measuring absorbance at 570 nm. All absorbance readings were obtained using an automated 96-channel microtiter plate reader (Bio-Rad) interfaced to a computer.

Mannosylated Glycoprotein-mediated Bacterial Aggregation Assay
The binding of type 1 fimbriated bacteria to the mannose-containing glycoproteins PAPM-BSA and HRP (Sigma) was studied spectrophotometrically. The cultures of bacterial cells were centrifuged, and the supernatants were removed, and each cell pellet was washed twice in sterile PBS before being resuspended in PBS. 990 l of each bacterial suspension (absorbance at 600 nm, 1.75) was transferred to a cuvette. Ten l of the glycoprotein solution at 5 mg/ml concentration in PBS was then added to each cuvette. The cuvettes were covered with Parafilm and shaken gently for 1 min, at which point they were placed into the spectrophotometer, and a reading at 600 nm was taken. As a control, 990 l of bacterial suspension (absorbance at 600 nm, 1.75) was mixed with 10 l of PBS. Subsequent readings were taken at 600 nm after a total elapsed time of 5, 15, and every 15 min thereafter for 1 h. The A 600 reading drops as the bacteria aggregate and settle at the bottom of the cuvettes. Aggregation was expressed as absorbance as a function of time.

Transmission and Scanning Electron Microscopy
For transmission electron microscopy, bacteria from overnight static liquid cultures were resuspended in PBS. A 25-l aliquot of the bacterial suspension was placed on top of a parlodion carbon-coated glow discharged grid for 30 s followed by staining with three drops of 1% phosphotungstic acid at pH 7.5. The grids were blotted dry and examined in a Joel 100B electron microscope.
For scanning electron microscopy, SI-H10 cells were grown on glass coverslips. The bacterial adherence experiment was carried out with live cells. The cells with adherent bacteria were fixed with 2% glutaraldehyde in PBS. The coverslips were washed twice with 0.1 M sodium cacodylate buffer, pH 7.3, at room temperature (RT) for 15 min. The cells were then postfixed with 1% OsO 4 in 0.1 M cacodylate buffer for 1 h at RT, followed by two 15-min washes at RT with cacodylate buffer. The cells were then dehydrated with ascending concentrations of ethanol at RT for 10 min at each step: 50, 70, 90, and 100%. The coverslips, in 100% ethanol, were then transferred to a Polaron E3000 Critical Point Drying Apparatus and critical point dried with CO 2 . Dried coverslips were mounted on aluminum specimen mounts, coated with 300 Å of gold in a Polaron E5000 Sputter Coating Unit, and viewed with a Hitachi S-450 Scanning Electron Microscope at 20 kV accelerating voltage.

DNA Manipulations
Restriction enzymes and DNA-modifying enzymes were used as recommended by the suppliers (Boehringer Mannheim; New England Biolabs, Beverly, MA). A complete list of plasmids that were generated as well as their gene products are described in Table I.
Linker Insertion Mutagenesis of fimHS-The fimHS gene on the pISF101 plasmid was inactivated by linker insertion mutagenesis. The S. typhimurium fim operon from pISF101 plasmid was released by digestion with SphI and subcloned into SphI-digested pUC 19 vector, and the resulting plasmid was named pKT301. pKT301 was digested with Bsu36I, and the linear DNA was gel-purified. The sequence of the primers that constitute the linker are primer Sty1, 5Ј TCA TAA TAA TGA TTA GGT GAG GAT CC 3Ј, and primer Sty2, 5Ј TG AGG ATC CTC ACC TAA TCA TTA TTA 3Ј. These oligonucleotides were phosphorylated and annealed and then ligated to the Bsu36I-linearized pKT301. E. coli DH5 ␣ FЈIQ competent cells were transformed, and the resultant colonies were screened for the introduced BamHI site present on the linker. DNA sequence analysis (18) using the sense strand primer Sty3, 5Ј TAT CGG CGC GTC GTT ATT TAG TC3Ј, and antisense primer Sty4, 5Ј AGC AAC CAC AAA GCC TAA AT3Ј, was carried out on the linker-inserted clones containing the BamHI site. One clone named pKT302, which contained an inactivated fimHS in S. typhimurium fim operon, was used to make the final construct in the pACYC184 plasmid. The mutant operon was subcloned into pACYC184 by digesting the pKT302 plasmid with SphI and by ligating the gel-purified 12-kb mutant operon fragment into SphI-linearized pACYC184. DH5 ␣ FЈIQ competent cells were transformed with the ligation mixture, and resultant colonies were screened by SphI and BamHI digestion and confirmed by DNA sequencing. The final construct was named pKT303. The plas-mid was introduced into E. coli ORN103 by a method outlined by Sambrook et al. (18).
Generation of malE/fimHS Gene Fusion Construct-The plasmid pISF101 that contains the entire type 1 fim gene cluster of S. typhimurium was used as the PCR template to generate fimHS DNA sequence. The PCR product was cloned into plasmid vector pMAL-p2 (New England Biolabs, Beverly, MA) using standard techniques to produce the plasmid pKT304. The vector pMAL-p2 encodes the affinity tag (MalE protein) and the (Asp)10 linker, which have a molecular mass of approximately 42 kDa. Oligonucleotides were synthesized and purified commercially (Integrated DNA Technologies, Inc., Coralville, IA). The following are the sequences of the forward and reverse oligonucleotide primers used to generate the fimHS DNA fragment employed in this study: primer Sty5, 5Ј GCC GGA ATT CTG CCG TAA TTC AAA CGG GAC G 3Ј, and primer Sty6, 5ЈGGG CAA GCT TCT ATT AAT CAT AAT CGA CTC GTA G 3Ј. Plasmids were introduced into E. coli ORN103 by heat shock method as outlined by Sambrook et al. (18). Expression vector pMAL-p2, amylose resin, antiserum against the maltose-binding protein (MalE), and purified MalE were obtained from the New England Biolabs (Beverly, MA).
Construction of the FimHES Chimera-The FimHES chimera-expressing clone was constructed by replacing in the recombinant plasmid pISF101 the S. typhimurium 5Ј end region of fimHS gene encoding amino acids 1-174 with corresponding E. coli 5Ј end of the fimHE gene encoding amino acids 1-158. The strategy used required introducing two restriction sites (NruI and HpaI sites) in the S. typhimurium sequence just upstream and downstream of the codons encoding amino acids 1 and 174, respectively, of the fimHS gene, and introducing the same restriction sites in the fimHE sequence flanking codons encoding amino acids 1-158. These sites were then used to replace the fimHS sequence with the fimHE sequence. The fimHES chimera bearing S. typhimurium type 1 fimbrial operon was reassembled in pACYC184 with multiple subcloning steps.
PCR Mutagenesis and Subcloning of S. typhimurium Fim Sequence to Introduce Internal NruI and HpaI Sites-In order to carry out PCR mutagenesis and to introduce internal NruI and HpaI sites in the fimHS gene, sequential PCR technology was used to generate a fragment of the fim operon from nucleotides 5286 to 6452 in which an NruI site and an HpaI site were introduced at nucleotides 5685 and 6224, respectively. Six PCR primers were designed and purchased from Life Technologies, Inc. Primer Sty7, CAGCGTAGTGCCGACAGAGG, is a sense primer located upstream of the AatII site at nucleotide 5379. Primer Sty8, GTCGCGAGCGCGGGATGGGTG, is an antisense primer located at the first site requiring mutation and includes the single base change at nucleotide 5684 to introduce an NruI site. Primer Sty9, CCGCGCTCGCGACGGTTTGCCG, is a sense primer also located at the first mutation site and contains the nucleotide change. These two primers overlap by 13 nucleotides in the region to be mutated. Primer Sty10, CCGGCGTTAACTTCACAGTTTTG, is an antisense primer. It contains two nucleotide changes necessary (nucleotides 6224 and 6227) to introduce an HpaI site. Primer Sty11, TGAAGTTAACGCCGGACAGGTCG, is a sense primer that contains the two nucleotide changes present in primer Sty10 to introduce the HpaI site. Primers Sty10 and Sty11 overlap by 15 bases in the region to be mutated. Primer Sty12, AG-CAACCACAAAGCCTAAAT, is an antisense primer, located just downstream of the unique Bsu36I site. Three separate PCR reactions were carried out on the pISF101 DNA template using the above described primers. Primers Sty7 and Sty8 were used in one reaction, primers Sty9 and Sty10 in the second, and primers Sty11 and Sty12 in the third. These three PCR reactions amplified products 404, 556, and 235 bp, respectively. The three PCR products were gel-purified and then all three were combined in a second round of PCR using primers Sty7 and Sty12. This reaction amplified a 1167-bp fragment that was cloned into the pGEM-T vector. Transformants were screened for the presence of NruI/HpaI fragments of approximately 500 bp. One clone with this phenotype was designated pKT305, and the sequence of the relevant regions was determined. The sequence analysis confirmed the introduction of the NruI and HpaI sites, as well as the integrity of the rest of the relevant S. typhimurium sequences.
PCR Cloning of fimHE Region to Introduce Flanking NruI and HpaI Sites-A DNA fragment of the fimHE encoding the amino acid residues 1-158 was generated from the pSH2 plasmid by PCR using primer Sty13, CTCGCGACGGTTTGTAAAACCGCCAATGGT, and primer Sty14, GTTAACTTCACAGCCGCCAGTAGGCAC, the former containing an NruI site and the latter containing an HpaI site. The 498-bp fragment generated in the PCR reaction was cloned into the pGEM-T vector. Transformants were screened by restriction analysis for the presence of the NruI/HpaI fragment. One clone containing the 498-bp NruI/HpaI fragment was sequenced. The sequence analysis confirmed the presence of the two introduced restriction sites as well as the integrity of the sequence of the entire PCR product. This clone was named pKT306.
Replacement of NruI/HpaI fimHS Fragment with NruI/HpaI fimHE Fragment-In the first of three subcloning steps, pKT306 was specifically digested with NruI and HpaI. The 498-bp fragment was gelpurified and cloned into the 3.6-kb fragment of pKT305, which had also been digested with NruI and HpaI and gel-purified. This construction was screened by NruI and HpaI digest for the presence of a 498-bp fragment. This plasmid was named pKT307. Additional restriction enzyme analysis also confirmed the presence of a KpnI site which was introduced from the E. coli sequence. The second step in the reconstruction involved subcloning 6727-bp PstI fragment from pISF101 into the PstI site of pUC21 (This recombinant plasmid was named pKT308) and replacing the AatII to Bsu36I fragment in pKT308 with the comparable fragment from pKT307 which contains the fimHE sequences. The plasmid pKT308 was initially linearized with Bsu36I and then partially digested with AatII. The 8.9-kb fragment was gel-purified. The plasmid pKT307 was digested to completion with AatII and Bsu36I, and the 975-bp fragment was gel-purified and cloned into the 8.9-kb fragment generated in the partial digest. The resultant clones were screened by AatII/Bsu36I digests and NruI/HpaI digest. One clone with the appropriate phenotype was chosen and named pKT309. The final subcloning step to assemble the chimeric fimHES-bearing operon in pACYC184 required digesting pKT309 with PstI and gel purifying the 6.7-kb fragment. This fragment was cloned into the 9-kb PstI fragment from pISF101. Transformants were screened by PstI digestion for the presence of two fragments, approximately 9 and 6.5 kb. There is a unique KpnI site in the E. coli sequence in the region cloned into the S. typhimurium fim operon. There are no KpnI sites in the pACYC184 vector or in the S. typhimurium fim operon. This construct was named pKT310.

Assays for Fimbriation and Isolation of Type 1 Fimbriae
Bacterial cell expression of type 1 fimbriae was confirmed by one or more of the following tests: ␣mM-sensitive agglutination of guinea pig erythrocytes and yeast (Candida albicans) cells, bacterial agglutination by anti-FimAS antiserum, and direct binding of fimbriae-specific antibodies in enzyme-linked immunosorbent assays (19). Type 1 fimbriae were isolated and purified by following the methods of Dodd and Eisenstein (20).

Extraction of Periplasmic Contents and Purification of MalE/ FimHS Fusion Protein
Once the expression of MalE/FimHS fusion was confirmed by immunoblotting, we induced cultures with 0.3 mM isopropyl-␤-D-thiogalactopyranoside to generate large amounts of the fusion protein in the periplasmic fraction (19). Periplasmic fluid was extracted from 2-liter culture suspensions as described previously (19) and passed over a Sepharose-amylose affinity column. The fusion protein was bound to the column via its affinity tag (the MalE portion). Bound fusion protein was subsequently eluted by the addition of 10 mM maltose, and the eluate was subjected to SDS-PAGE (19).

SDS-PAGE, Immunoblotting, and Densitometry
Purified FimHS ϩ wild type, FimHS Ϫ mutant, and FimHES ϩ hybrid fimbrial preparations (100-l volume each) were boiled for 5 min in 100 l of 0.125 N HCl. 100 l of 0.1 N sodium hydroxide (NaOH) was added to this mixture followed by the addition of 100 l of 4ϫ SDS-PAGE sample buffer and boiling for 8 min. These fimbrial samples were subjected to SDS-PAGE on a 15% slab gel by the method of Laemmli (21). Purified MalE/FimHS fusion protein was subjected to SDS-PAGE on a 10% slab gel. The proteins were subsequently electrophoretically transferred onto nitrocellulose membranes. The immunostaining of specific protein was performed essentially as described previously (19) using antisera raised against FimAS or FimHE or MalE. Densitometric scannings of protein bands were performed optically with the Eagle Eye II still video system (Stratagene, La Jolla, CA).

Overlay Assays Employing Mouse Enterocytes, Bladder Epithelial Cells, and Mannosylated Glycoproteins (PAPM-BSA and HRP)
Purified fusion protein comprising of MalE and FimHS (MalE/ FimHS) and MalE were subjected to SDS-PAGE and transferred onto nitrocellulose membranes. After blocking in 3% bovine serum albumin (BSA) in PBS for 1 h, the nitrocellulose blot was overlaid with 5 ϫ 10 6 biotinylated mouse enterocytes or bladder cells in the presence and absence of 100 mM ␣mM (19). The mouse cells were biotinylated as described previously (19). After 1 h of incubation at 22°C, the blot was rinsed several times with PBS, and the bound cells were probed with alkaline phosphatase-conjugated avidin (Sigma) followed by the substrate 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (BCIP/NBT) (19). In the glycoprotein overlay assays, after blocking the nitrocellulose blot in 3% BSA in PBS for 1 h, the membrane was overlaid with 100 g of HRP/ml of PBS for 1 h (in the presence and absence of 100 mM ␣mM) or 100 g of PAPM-BSA/ml of PBS for 1 h followed by incubation with 100 g of HRP/ml of PBS for 1 h. After 1 h of incubation at 22°C, the blot was rinsed several times with PBS. Bound HRP was directly detected by the addition of its substrate, 4-chloro-1-naphthol.

Generation of Antisera
Antisera to FimHE and to FimAS were raised as described earlier (19).

Protein Sequence Analysis
CLUSTAL X multiple sequence alignment program (22) was used to align amino acid sequences. MacBOXSHADE program (23) was used to shade identical amino acid residues.

Type 1 Fimbriae of S. typhimurium Promote Bacterial Binding to Murine Enterocytes but Not to Bladder Epithelial
Cells-We examined the adherence of two pairs of strains, the clinical S. typhimurium strain X4252 and its fim minus derivative, S. typhimurium X4253, and the fim minus laboratory E. coli strain, ORN103, and its recombinant derivative, ORN103(pISF101) harboring the recombinant plasmid pISF101 encoding S. typhimurium type 1 fimbriae. The assays were undertaken employing a previously described mouse enterocyte cell line, SI-H10 (16) that was grown as a monolayer on microtiter plates. The SI-H10 cells were rendered nonviable by paraformaldehyde fixation in order to exclude the possible contribution of post-contact-induced bacterial adherence events that typically require participation of a viable host cell. It was also necessary to use nonviable enterocytes because quantitation of adherent bacteria was dependent on assaying for MTT processing by live bacteria, and the presence of any host cell-derived formazan (MTT product) would have confounded the assay. Type 1 fimbriated S. typhimurium X4252 and E. coli ORN103(pISF101) bound to the gut epithelial cells in large numbers, whereas the nonfimbriated S. typhimurium X4253 and E. coli ORN103 strain bound only minimally (Fig.  1A). Type 1 fimbriae of Salmonella were specifically implicated in the binding reactions because the adherence of both type 1 fimbriated strains was markedly reduced in the presence of 100 mM ␣-methyl-D-mannopyranoside (␣mM).
To determine if the Salmonella type 1 fimbriae mediated a distinct binding reaction from other enterobacterial type 1 fimbriae, we compared enterocyte binding of the recombinant bacteria expressing type 1 fimbriae of E. coli with those expressing Salmonella fimbriae. We observed that the number of adherent bacteria expressing Salmonella type 1 fimbriae was considerably more than the number of adherent bacteria expressing E. coli fimbriae (Fig. 1A). Since Salmonella do not typically infect the urinary tract, whereas E. coli is a frequent pathogen, we compared the binding to mouse bladder epithelial cells of bacteria expressing Salmonella and E. coli type 1 fimbriae (Fig.  1B). Whereas E. coli type 1 fimbriae mediated high levels of bacterial binding to bladder cells and lower levels of binding to enterocytes (Fig. 1, A and B), bacteria expressing Salmonella type 1 fimbriae bound only enterocytes. Therefore, Salmonella and, perhaps, E. coli exhibit distinct binding tropisms in vitro reflecting their colonization sites in vivo (Fig. 1, A and B).
Characterization of the Binding Properties of S. typhimurium Type 1 Fimbriae-Heterogeneity in binding among different type 1 fimbriated bacteria may be manifested in differential binding to selected mannosylated glycoproteins (14,24). To confirm the distinct binding properties of Salmonella and E. coli type 1 fimbriae, we compared their ability to bind two mannosylated glycoproteins, horseradish peroxidase (HRP) and p-amino-phenyl-␣-D-mannopyranoside-BSA (PAPM-BSA). Fimbrial binding to the glycoproteins resulted in bacterial aggregation and sedimentation that can be readily monitored by a fall in the absorbance values of the mixture. Whereas bacteria expressing E. coli or Salmonella type 1 fimbriae bound and were aggregated by PAPM-BSA, only bacteria expressing E. coli type 1 fimbriae bound and were aggregated by HRP (Fig. 2,  A and B). Thus, compared with E. coli type 1 fimbriae, Salmonella fimbriae exhibited a more selective binding reaction implying that its mannose-binding adhesin possesses a distinct recognition motif from that on E. coli type 1 fimbriae.
Aggregative Bacterial Adherence Mediated by S. typhimurium Type 1 Fimbriae-Another remarkable feature of Salmonella type 1 fimbriae-mediated adherence is that secondary bacterial aggregates appeared to form around already adherent bacteria on the surface of enterocytes. When the adherence of S. typhimurium X4252 or E. coli ORN103(pISF101) to live SI-H10 enterocytes was examined by light microscopy, the bacteria appeared to be associated with the epithelial cells in distinct aggregates. The bacterial aggregates on the surface of enterocytes appeared after about 45 min of incubation, and the size of the aggregates increased with prolonged incubation (data not shown). Many of the bacteria in the aggregates seemed to be adhering to each other rather than to the enterocyte surface. Scanning electron micrographs illustrating the peculiar bacterial aggregation mediated by Salmonella type 1 fimbriae on the surface of the gut epithelium is shown in Fig. 3, A and C. None of the type 1 fimbriated bacteria appeared to bind each other prior to addition to the epithelial monolayer, indicating that this phenomenon is related to post-adhesion events. This phenomenon appeared to be unique to Salmonella type 1 fimbriae because E. coli ORN103(pSH2) bound enterocytes but failed to form bacterial aggregates (data not shown). None of the nonfimbriated S. typhimurium X4253 and E. coli ORN103 bound epithelial cells or formed aggregates on enterocytes (Fig. 3, B and D). This secondary adherence mediated by Salmonella type 1 fimbriae could potentially augment its tropic effects by amplifying the numbers of bacteria found associating with the enterocyte surface.
Identification of the Determinant of Adhesion on S. typhimurium Type 1 Fimbriae-We sought to identify the determinant on S. typhimurium type 1 fimbriae responsible for mediating specific adherence to enterocytes. The organization and make up of genes within the S. typhimurium fim cluster have been determined by Clegg and Swenson (25). However, because of the dissimilarities in the size, number, and location of various genes in the Salmonella fim cluster and other enterobacterial fim clusters, it was not possible to readily identify the determinant of binding. For example, the regulatory genes of Salmonella (fimZ, fimY, and fimW) are located at the 3Ј end of the gene cluster, whereas the E. coli type 1 fimbrial operon regulatory genes (fimB and fimE) are located at the 5Ј end of the gene cluster (Fig. 4A). Moreover, the DNA sequence of the S. typhimurium fim locus has recently been determined by Clegg and colleagues (GenBank accession number L19338 (25)), and none of the predicted protein sequences exhibited any homology with the type 1 fimbrial adhesins of E. coli or other enterobacteria.
It has previously been reported that the amino-terminal region of the FimH proteins of a wide range of enteric bacteria was antigenically conserved (15). Therefore, we used an antibody directed at a synthetic peptide corresponding to residues 1-25 of FimHE to probe for the Salmonella type 1 fimbrial adhesin. The fimbriae were prepared from S. typhimurium X4252, and the antisera reacted specifically with a band at 35 kDa on the Western blot (Fig. 4B). A similar band was detected when fimbriae from the clone ORN103(pISF101) was probed with FimHE antisera (data not shown). Thus, despite the lack of homology in the primary structure between FimHE and any of the gene products of Salmonella fim cluster, the antibody reacted with a 35-kDa protein. Presumably, the antibody was able to recognize conserved conformational epitopes. An open reading frame encoding a 35-kDa protein is present in the S. typhimurium gene cluster. This gene is located in the middle of the fim gene cluster which is distinct from the traditional location of enterobacterial fimH at the distal 3Ј end of the gene cluster.
If the gene encoding the 35-kDa protein was the Salmonella adhesin, we predicted that specific inactivation of this gene in the Salmonella fim cluster would result in loss of the adhesive phenotype. We introduced in-frame translation termination codons within the fimHS coding region in the fim operon of pISF101. The insertion of the linker containing the in-frame stop codons within the fimHS gene was verified by DNA sequencing. The newly generated plasmid, pKT303 (also described in Table I), was introduced into the E. coli host strain ORN103. The resultant transformant was examined for fimbrial expression by electron microscopy and found to express a comparable number of fimbriae per bacteria as E. coli ORN103(pISF101). Also, there appeared to be no obvious difference in morphology between FimHS Ϫ mutant and wild type fimbriae (Fig. 4C). The length of the mutant fimbriae also appeared to be comparable to that of the wild type.
We confirmed biochemically that the mutant fimbriae were deficient in the 35-kDa FimHS protein. We isolated fimbriae from the mutant ORN103(pKT303) and from ORN103-(pISF101) expressing wild type fimbriae and then probed them with an antibody raised against a synthetic peptide corresponding to residues 1-25 of FimHE on a Western blot. A second Western blot was also performed with an antibody to the 21-kDa FimA subunit of Salmonella type 1 fimbriae (FimAS). As shown in Fig. 4D, a 35 2). It is also noteworthy that comparative densitometric analyses of FimAS and FimHS bands in the wild type fimbrial preparation indicated that the ratio of the adhesin to the major subunit is 1:10. This contrasts with E. coli type 1 fimbriae where the ratio ranges from 1:100 to 1:60 (15).
We examined the ability of E. coli ORN103(pKT303) to bind mouse gut and bladder epithelial cells and the two mannosylated proteins (PAPM-BSA and HRP). We found that in contrast to ORN103(pISF101) expressing native Salmonella fimbriae, the mutant ORN103(pKT303) expressing FimHSdeficient fimbriae failed to mediate any appreciable binding to the monolayer of enterocytes or to bladder cells (Fig. 1, A and  B). Moreover, bacteria expressing the mutant fimbriae were not aggregated by either PAPM-BSA or HRP (Fig. 2, A and B). Together, the data suggest that the 35-kDa FimHS protein is the determinant on Salmonella type 1 fimbriae responsible for mediating adherence. However, these data do not preclude the contribution of other fimbrial components in the binding process.

Expression of FimHS as a Translational Fusion Protein with Maltose-binding Protein (MalE) and Examination of Its
Adhesive Capacity-We sought to investigate if FimHS could mediate its binding in the absence of other fimbrial proteins. We cloned fimHS but found that the recombinant gene product was highly unstable in the bacterial periplasm. We attempted to stabilize the FimHS by expressing it as a translational fusion protein with MalE as described previously for the stabilization of FimHE (19). The periplasm was isolated, and the MalE and MalE/FimHS proteins were affinity purified and analyzed by SDS-PAGE. The difference in migration between MalE (lane 1) and the MalE/FimHS fusion protein (lane 2) corresponded to the size of FimHS (Fig. 5A). A Western blot using antibodies to MalE (data not shown) and to FimHE confirmed that the 77-kDa protein was indeed a fusion of FimHS and MalE (lane 4).
To determine whether the MalE/FimHS fusion protein could bind enterocytes, we performed a cell blotting assay where MalE/FimHS were immobilized onto nitrocellulose membrane and then probed with biotinylated enterocytes in the presence and absence of 100 mM ␣mM. As shown in Fig. 5B, the enterocytes bound MalE/FimHS (lane 2) but not MalE (lane 1). Furthermore, this binding reaction was inhibitable by ␣mM (lane 4). Thus, FimHS alone is capable of mediating mannose-inhibitable binding to mouse enterocytes.
To see if the isolated FimHS also exhibited the characteristic binding specificity exhibited by Salmonella type 1 fimbriae, we tested the ability of the immobilized protein to bind bladder cells and the two mannosylated proteins, PAPM-BSA or HRP. The binding to bladder cells was undertaken as described for enterocytes. However, the assay for assessing binding to HRP is notable because it involves a single step that simultaneously takes advantage of the highly mannosylated nature of the glycoprotein (reflected in its ability to bind FimHS) and its intrinsic enzymatic properties (reflected in its ability to react with the Western blot substrate, 4-chloro-1-naphthol). Binding of PAPM-BSA to FimHS was assessed by the ability of the glycoprotein to block the binding of HRP to MalE/FimHS. Surprisingly, we found that FimHS bound bladder cells as well as it bound enterocytes, and it bound both mannosylated compounds equally well (Fig. 5C). This was a surprising finding and indicated that although Salmonella type 1 fimbriae appeared selective in its binding of mannosylated compounds, its FimHS adhesin did not exhibit this selective binding trait. In view of the fact that in the native fimbriae, the FimHS is intimately associated with the fimbrial shaft, it is likely that this structure may play a role in mediating some of the selec-tive binding traits of the fimbriae.
Creation and Characterization of Hybrid S. typhimurium Fimbriae Containing Chimeric FimHES-To explore further the relative contribution of FimHS and the Salmonella fimbrial shaft to fimbrial binding specificity, we replaced a stretch comprising amino acid residues 1-173 of FimHS with a corresponding stretch of residues 1-158 from FimHE containing the receptor binding domain of FimHE (19). We reasoned that if the shaft influenced the binding specificity, then the resulting hybrid fimbriae would function as Salmonella fimbriae. Alternatively, if the hybrid fimbriae behaved more as E. coli fimbriae, then we would assume that the primary structure of FimHS was responsible for the distinct binding traits of the fimbriae. The chimeric FimHES containing hybrid fimbriae expressed by ORN103(pKT310) bound to enterocytes but failed to bind bladder cells (Fig. 1, A and B). Moreover, FimHES containing fimbriae bound and aggregated PAPM-BSA but not HRP (Fig. 2, A  and B). Thus, the hybrid fimbriae comprising FimHES on a Salmonella shaft still retained the binding traits of native Salmonella fimbriae. Conceivably, the fimbrial shaft has an overriding influence on the binding specificity regardless of the primary structure of the adhesin.
Several aspects regarding the morphology and composition of the hybrid fimbriae were noted. Western blot analyses of the fimbriae employing FimAS-and FimHE-specific antisera followed by densitometry revealed that the amount of FimHES incorporated into the fimbrial filament was approximately 40% less than the level of FimHS in native Salmonella fimbriae (Fig.  6A). This observation may account for why the binding of the bacteria expressing FimHES to the enterocytes was less than the binding of bacteria expressing native Salmonella fimbriae (Fig. 1, A and B). Transmission electron microscopy revealed that the hybrid fimbriae were aberrant, with a curly and tangled appearance throughout its entire length (Fig. 6B). Type 1 fimbriae typically are highly stable organelles that require boiling in 0.125 N hydrochloric acid (HCl) to induce complete depolymerization. When we compared the stability of the hybrid fimbriae with that of native Salmonella fimbriae, we found that the hybrids were highly unstable. Whereas native Salmonella fimbriae required HCl treatment followed by boiling in SDS-PAGE sample buffer to be depolymerized, hybrid fimbriae were completely depolymerized merely by boiling in SDS-PAGE sample buffer at pH 6.8 (Fig. 6C). Thus, the incorporation of partially heterologous FimHES into the Salmonella fimbrial filament had a profound effect on both fimbrial morphology and stability. DISCUSSION Although it has been known for a relatively long time that S. typhimurium expresses type 1 fimbriae, the potential contribution of these organelles to infection has largely been ignored. In light of recent observations from several different laboratories on the role of E. coli and K. pneumoniae type 1 fimbriae in promoting bacterial infections and its harmful inflammatory sequellae (19,26,27), it is conceivable that these organelles also contribute significantly to the pathogenesis of Salmonella.
S. typhimurium invariably invades its hosts through the gut, reflecting a distinct adhesive preference for gut tissue. A battery of contiguous genes residing within SPI-1, a 40-kb pathogenicity island mapping at 63 min in the S. typhimurium chromosome, is largely responsible for mediating entry of Salmonella into epithelial cells (28). However, none of these genes appear to mediate Salmonella's tropism for enterocytes. Indeed, many of these "invasion-associated genes" are expressed only after the bacteria has already established intimate contact with the host cell. We reasoned that the determinant conferring tropism for gut tissue is an already expressed bacterial surface moiety and is likely to be one or more of its cell-surface fimbriae. Considering their length and peritrichous orientation, these organelles could be the first bacterial components to make contact with host tissue. Fimbriae have been implicated in mediating tissue tropism in several pathogens. For example, the fimbriae on pathogenic Neisseria gonorrheae (29) and pyelonephritogenic E. coli (30) mediate bacterial tropism for human genital cells and kidney tissue, respectively.
Salmonella type 1 fimbriae exhibit a distinct tropism for gut tissue because it mediates high levels of bacterial binding to mouse enterocytes but not to mouse bladder cells. The binding pattern of Salmonella type 1 fimbriae contrasts with that of E. coli type 1 fimbriae which mediated moderate bacterial binding to enterocytes and even higher levels of bacterial binding to bladder cells. The binding patterns displayed by these fimbriae correlate with the natural sites of infection of their respective It is known that fimbriae-mediated bacterial binding to mucosal cells can trigger various host cell responses that cumulatively can significantly affect the course of the infectious process (27,31). What is less well known is that the specific coupling of bacterial fimbriae with complementary receptors on host cells may activate genes within the bacteria, notably those that enhance bacterial virulence in vivo (32). One consequence of Salmonella type 1 fimbriae-mediated binding to enterocytes is the formation of discrete bacterial aggregates on the surface of the gut epithelial cell. The aggregates could be the combined effect of replication of attached bacteria and recruitment of proximate bacteria through their fimbriae. This phenomenon is seen with Salmonella type 1 fimbriae but not with E. coli type 1 fimbriae. However, it is distinct from bacterial aggregation mediated by the recently described thin fimbriae of S. typhimurium (16) because it was not seen in the absence of the host cell. That the Salmonella type 1 fimbrial adhesin is involved is indicated because aggregative adherence can be blocked by the addition of a D-mannose analog. The mechanism could involve binding of Salmonella fimbriae to mannosylated compounds on the surface of neighboring bacteria. Since the aggregation does not occur when bacteria are cultured alone, it appears that the expression of mannosylated compounds on the bacterial surface occurs only after contact with the host cell. The proposed mechanism is not unlike the aggregation of K. pneumoniae on human intestinal cells where the interaction of a newly synthesized capsule-like material and an, as yet, unidentified fimbriae have been suggested (33). Aggregative adherence could be beneficial to the bacteria because it enables them to increase their load on an epithelial cell even when the specific fimbrial receptor density is limiting.
Several lines of evidence allow us to conclude that the Salmonella fimbrial adhesin is the product of the fimHS gene as follows: (i) a protein in Salmonella fimbriae corresponding to the size of the fimHS gene product cross-reacted with anti-FimHE antibodies, (ii) inactivation of the fimHS gene resulted in abolishment of binding activity in the mutant ORN103-(pKT303) without concomitant loss of fimbriae as determined by electron microscopy and by biochemical analyses of mutant fimbriae, (iii) recombinant FimHS, when expressed in the absence of other fimbrial proteins bound mouse enterocytes in a D-mannose-inhibitable fashion. Our findings confirm and extend the earlier report of an association between the presence of 33-36-kDa proteins and functional Salmonella type 1 fimbriae (34) and the prediction of Clegg and Swenson (25) that the product of the fimHS gene was the adhesin. Previous attempts to inactivate fimHS have resulted in loss of fimbrial expression which has made it difficult to define precisely the functional role of the gene product (35). This is probably attributable to the polar effects of the fimHS mutation on the downstream gene, fimF, which is critical to fimbrial expression. Our studies have delinked fimbrial expression from FimHs expression and in doing so revealed that the adhesin is not required for the assembly of fimbriae.
Clegg and colleagues (25,(35)(36)(37) have reported on distinct aspects in the genetic organization, expression, and regulation of fim gene cluster of the S. typhimurium in relation to other enterobacterial fim gene clusters. Here, we have noted several differences between FimHS and its counterpart on other enterobacteria. First, its size is appreciably higher than the size of the FimH molecule of other enterobacteria which typically approximate 29 kDa (15). Second, fimHS maps at the center of the fim cluster and not at the distal end as is the case in other enterobacteria (25). Third, the amount of FimHS relative to FimAS in the fimbriae appears to be higher (1:10) rather than the 1:60 ratio found in E. coli type 1 fimbriae (15). Fourth, FimHS exhibits very limited amino acid sequence similarity with other enterobacterial FimH. Fig. 7 depicts the extent of homology in covalent structure even after progressive multiple sequence alignments of FimHS and FimHE proteins. Although the significance of these distinct traits of FimHS is currently unknown, they may reflect evolutionary advances over other FIG. 6. Generation and characterization of Salmonella hybrid fimbriae containing a chimeric FimHES. A, verification of the presence of the chimeric FimHES in hybrid fimbriae produced by strain ORN103(pKT310). WT, wild type; HB, hybrid. Western blots of native Salmonella fimbriae from ORN103(pISF101) (lanes 1 and 3) and chimeric FimHES containing Salmonella fimbriae from strain ORN103(pKT310) (lanes 2 and 4). Lanes 1 and 2 were reacted with antibodies to Salmonella FimA(FimAS). Lanes 3 and 4 were reacted with antibodies to FimHE. Notice that the amounts of the FimAS in lanes 1 and 2 are comparable but that the amount of the 32-kDa immunoreactive band (chimeric FimHES) in lane 4 is appreciably weaker than the 35-kDa immunoreactive band (FimHS) in native Salmonella fimbriae in lane 3. B, electron micrograph showing the irregular and flaccid appearance of fimbriae from ORN103(pKT310) containing chimeric FimHES. C, Coomassie Blue-stained preparations of isolated wild type Salmonella fimbriae (lanes 1 and 3) and fimbriae with chimeric FimHES (lanes 2 and 4). Lanes 1 and 2 were boiled only in SDS-PAGE sample buffer. Lanes 3 and 4 were boiled first in HCl and then in SDS-PAGE sample buffer. Unlike the native Salmonella type 1 fimbriae, hybrid fimbriae were readily depolymerized even in SDS-PAGE sample buffer. enterobacterial type 1 fimbrial adhesins.
A very surprising finding was that FimHS, when isolated from the fimbriae, failed to exhibit the selective binding specificity of native Salmonella fimbriae. Thus, in addition to binding enterocytes and to PAPM-BSA, MalE/FimHS exhibited strong binding to bladder cells and to HRP. This indicated that FimHS is intrinsically capable of recognizing the same mannose motifs bound by E. coli type 1 fimbriae. This is remarkable when one considers that there is little homology in primary structure between FimHS and FimHE. These findings led us to conclude that the binding pattern exhibited by the Salmonella fimbriae is not exclusively dependent on the primary structure of the adhesin. Since the adhesin is intimately associated with the fimbrial shaft, it is likely that the latter influences the binding traits of the former. We sought to assess the relative contribution of the shaft in modulating binding pattern of the fimbriae. Since it was reported that the receptor binding domain on FimHE is comprised of epitopes formed by amino acid residues in the first half of the molecule (19), we replaced the amino-terminal half of FimHS with the corresponding region on FimHE that contains the receptor binding domain on the E. coli adhesin. The resulting hybrid fimbriae contained a chimeric FimHES moiety with the receptor binding domain of FimHE but was presented on a Salmonella fimbrial shaft. The level of binding to enterocytes mediated by the hybrid fimbriae was less than that of native Salmonella type 1 fimbriae, and this is probably a reflection of reduced levels (by about 40%) of the adhesin in the fimbriae. Nevertheless, its binding pattern did not differ markedly from that of native Salmonella fimbriae as indicated by their binding to enterocytes and PAPM-BSA and their inability to bind bladder cells and HRP. Evidently, the loss of the amino-terminal half of FimHs was not accompanied by a concomitant loss of the distinct binding traits of Salmo-nella fimbriae. One interpretation of this finding is that the binding domain of FimHS is not contained in its amino-terminal half. This is unlikely in view of the recent finding that the only difference in primary structure between the FimH molecule of S. typhimurium type 1 fimbriae and the non-adhesive fimbriae of Salmonella pullorum are eight amino acid substitutions, all located within the amino-terminal halves of the molecules (35). The alternate and perhaps more plausible explanation is that the quaternary constraints imposed on the adhesin by the fimbrial shaft, which is comprised primarily of FimA and at least one minor subunit, FimF (35), may have an overriding influence on the binding pattern exhibited by the Salmonella fimbriae. That acquisition of functionally important FimHE epitopes still did not confer on the hybrid fimbriae the E. coli binding phenotype lends further support to this notion. Thus, to reconcile the broad binding reactions exhibited by FimHS in its isolated state and the limited binding reactions of the same molecule when incorporated in its fimbriae, we have proposed a model. In this model we postulate that the binding pocket on the isolated FimHS is promiscuous in that it is able to bind a broad range of mannosylated motifs enabling the molecule to bind different mannosylated moieties on various cells or soluble proteins. However, when incorporated into the fimbrial shaft, only few of the mannosylated motifs are recognized by the FimHS binding pocket because of the structural constraints imposed by the fimbrial shaft. This model provides the basis for the limited binding reactions exhibited by Salmonella type 1 fimbriae and for the overriding influence of the fimbrial shaft on the binding reactions exhibited by the adhesin moiety (Fig. 8).
Valuable clues on the structure of Salmonella type 1 fimbriae may be gleaned from the morphological and biochemical characterization of the hybrid fimbriae containing FimHES. That the hybrid fimbriae appeared wavy throughout its entire length and was unstable suggests that FimHS was intercalated longitudinally at multiple sites in the structure. Two models have been proposed for the location of FimHE in native E. coli fimbriae. In the first model, FimHE is located exclusively at the tips (38), which is analogous to the location of the PapG adhesin of the well characterized E. coli P fimbriae. In the second model, FimHE is located not only at the tips of the filament, but also intercalated longitudinally (15). The second model was supported by the subsequent finding that native E. coli type 1 fimbriae were significantly more prone to fragmentation than the adhesin-deficient mutant fimbriae when subjected to repeated cycles of freeze-thaw (39). Apparently, sites of FimHE incorporation were the preferred sites of fimbrial breakage during freeze-thawing since adhesion properties of fimbrial fragments had increased (39). In view of the extraordinary instability of the hybrid fimbriae bearing the heterologous Fim-HES, it is likely that the adhesin is incorporated more frequently into the structure than exclusively at the tips. More definitive studies are currently underway in this laboratory to localize FimHS more precisely in the fimbriae.
In summary, our studies indicate that the type 1 fimbriae on S. typhimurium is potentially critical to the infectious process because it determines the initial targeting of the bacteria to the gut. The type 1 fimbriae of S. typhimurium mediates high level of bacterial binding to enterocytes but not to bladder cells. The specific determinant of adhesion on the fimbriae is its structurally distinct 35-kDa subunit, FimHS. Surprisingly, in the absence of other fimbrial components, FimHS exhibited a broad range of mannose-sensitive binding reactions and did not display the selective binding associated with Salmonella type 1 fimbriae. We hypothesize that the tissue tropism exhibited by Salmonella type 1 fimbriae is determined at least in part by the fimbrial shaft which bears FimHS. It has previously been assumed that the fimbrial shaft is an innocuous structure on which the adhesin is strategically poised. Our studies indicate that these structures play a more active role in the pathogenic process by defining the specific receptor molecules and host cells that interact with the bacteria.
FIG. 8. Model to explain the differential binding traits of FimHS in its isolated state and when expressed on the fimbrial filament. It is hypothesized that in its isolated state, FimHS displays the ability to recognize various mannosylated motifs (as depicted in the figure) which allows the molecule to bind different mannosylated molecules on various cells or in soluble proteins. However, when incorporated at the tips of the Salmonella fimbrial shaft, FimHS is only able to bind a narrow range of mannosylated motifs because of the structural constraints imposed by the shaft, hence the narrow range of binding exhibited by S. typhimurium type 1 fimbriae. We have also depicted FimHS that are intercalated periodically along the fimbrial length. As indicated, the binding pocket in these intercalated FimHS are inaccessible in the quaternary conformation of the fimbriae, and thus these molecules are essentially nonfunctional until they are exposed by breakage. It is suspected that the sites of intercalation of FimHS are more susceptible to breakage from shear forces than other regions of the fimbriae. Thus even after the loss of a fimbrial fragment, the residual filament can still mediate attachment. It should be noted that the FimHs are likely to be located much further apart on the filament than is depicted in the figure.