INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Type 1 pili are filamentous proteinaceous appendages produced by several members of the Enterobacteriaceae. In Escherichia coli, type 1 pili have been studied extensively with regard to their genetics, biosynthesis, and ability to bind mannose-containing receptor molecules on a variety of eucaryotic cells (reviewed in reference 29). Although the pili are made principally of a single protein monomer, the product of the fimA gene, several minor protein components are also incorporated (12, 34). These are most often found at the ends of pili and are organized into fibrillar structures (15). One of the minor components, the product of the fimH gene (FimH), binds directly to the receptor (18). Whereas the specificity of the interaction of the fimH product can be influenced by other fimbrial components (23), several studies have linked certain naturally occurring fimH allelic types to the specificity of receptor binding (43, 46) and the strength of receptor binding (45). Additional experiments have suggested that some fimH allelic differences can contribute to tissue tropism (35, 43, 44). Gene fusion experiments have indicated that FimH binding capacity resides in the amino one-third to one-half of the protein (17, 48). However, point mutations in various parts of the coding region can effect a change in specificity for particular types of ligands (46). This is consistent with what is found with other bacterial adhesins (4). FimH has been used to express foreign antigens by inserting heterologous gene segments into the fimH gene (33) and used on its own as an effective immunogen in preventing experimental urinary tract infections in mice (20).

We have been particularly interested in the factors influencing the specificity of FimH binding and how FimH affects proper pilus structure (reviewed in 29). Some years ago, we isolated a number of fimH point mutants (13). These mutants were isolated following enrichment for individuals that formed pellicles when grown in static broth (a property associated with FimH) in the presence of a mannose analogue (-methyl-mannoside) that normally inhibits pellicle formation. All mutants isolated in this manner were defective in their ability to bind guinea pig erythrocytes. Additionally, several of the mutants produced pili with altered morphology.

In this report, we further characterize five of the fimH mutants isolated in the previous study. These five mutants represent all allelic classes of the 11 mutants initially isolated (13). In particular, we concentrate upon one of the mutants that was conditionally defective in erythrocyte agglutination, showing that erythrocyte and macrophage binding are differentially affected by the mutation. The results indicate that FimH-mediated attachment to different types of eucaryotic cells can occur through different mechanisms.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteria, bacteriophage, plasmids, and media. The bacterial strains, all E. coli K-12 derivatives, bacteriophage, and plasmids used in this investigation are listed in Table 1. Media consisted of L agar and L broth (28), -broth (28), tetrazolium agar (42), and MinA broth and agar (28). Antibiotics were added as previously described (30) unless otherwise noted.

Genetic techniques. Transformation with plasmid DNA followed the method described by Lederberg and Cohen (21). P1 transduction followed the method described by Miller (28).

Recombinant DNA techniques. Restriction endonuclease cleavage, plasmid and chromosomal DNA isolation, fragment purification, end filling, ligation, and subcloning were performed as previously described (36). DNA sequencing was performed on double-stranded plasmid DNA as previously described (50) using 20-oligonucleotide primer pairs bracketing various regions of the fimH and lacZ genes. PCR product sequencing of chromosomal and plasmid DNA was carried out as described by Russell and Orndorff (36). Bal31 exonuclease digestions were performed as described by Maniatis et al. (24).

Construction and induction of the lacZ-fimH fusion, and isolation of the fusion protein. A lacZ-fimH fusion was created by KpnI-PstI digestion of pORN148 followed by end filling and ligation to BamHI-digested, end-filled plasmid pUR288. This created plasmid pORN303. The fusion product contained approximately 92% of the mature FimH product. Pilot experiments revealed that the most efficient production of fusion protein was obtained by diluting an overnight culture of the strain containing the fusion plasmid (ORN206/pORN303) 1:50 in fresh warm L-broth containing ampicillin (100 µg/ml) and 5 mM isopropyl--D-thiogalactopyranoside (IPTG), followed by 4 h of growth with shaking at 37°C. After this period, bacteria were harvested by centrifugation (7,500 × g for 10 min), and the pellet was resuspended in 1× protein sample buffer (19) and boiled for 5 min. After a brief centrifugation to remove the insoluble material, various amounts of the supernatant were subjected to electrophoresis on a 10% discontinuous polyacrylamide gel (19). The fusion protein band was detected following Coomassie brilliant blue staining at the position expected from the predicted size of the fusion protein induced by IPTG. For preparative gels (prepared as above except 1.5-mm thick), the approximate location and amount of the fusion protein band were determined by staining marker strips containing high-molecular-weight protein standards (Bio-Rad) with Coomassie brilliant blue stain.

Immunological methods. Rabbit polyclonal antiserum against the LacZ-FimH fusion protein was raised by injecting a New Zealand White rabbit (ca. 2 kg) with a macerated acrylamide gel slice that contained approximately 100 µg of a LacZ-FimH fusion protein in complete Freund's adjuvant. Booster doses of 100 µg of fusion protein in incomplete Freund's adjuvant were administered approximately every 2 weeks with blood drawn beginning after the fourth boost and every 2 weeks thereafter, for a total of five bleedings. Antiserum was stored at 20°C, and aliquots were ammonium sulfate precipitated (7) and concentrated approximately fivefold in phosphate-buffered saline (PBS) prior to use.

Immunological and functional detection of FimH. The presence of FimH was determined immunologically by bacterial agglutination reactions performed in 96-well round-bottomed microtiter plates. Bacteria from overnight cultures were isolated by brief centrifugation (1 to 2 min in a microcentrifuge) and concentrated two- to fivefold in PBS (approximately 4 × 10⁹ to 10 × 10⁹ cells per ml) depending upon the experiment. Twenty-five microliters of antiserum was serially twofold diluted in microtiter wells, and 25 µl of the bacterial suspension was added to each well. Microtiter plates were incubated at room temperature for 1 h and then refrigerated (4°C) until the negative control wells (containing a fimH insertion mutant) had settled (typically 24 to 48 h).

Recombination mapping of fimH mutations. Transformants that contained a chromosomal fimH mutant allele received plasmids containing deletion derivatives of fimH via transduction. Transductants (800 to 1,000 colonies) were recovered from agar plates using a cotton swab. The material from each swab was expressed into approximately 1.0 ml of PBS, and 0.3 ml was added to a microcentrifuge tube containing 100 µl of settled fresh guinea pig erythrocytes. The bacteria and erythrocytes were mixed by inversion and incubated (to allow binding) for 10 min. The red blood cells were isolated by centrifugation for approximately 1 s in a microcentrifuge, and the supernatant was aspirated. The pellets were resuspended and washed with 1.0 ml of PBS five more times. The final pellet was resuspended in 0.5 ml of distilled water (to lyse the erythrocytes), and 2.0 ml of L-broth with chloramphenicol was added. This mixture was incubated overnight with shaking at 37°C to expand the population that remained bound to the erythrocytes. The following day, red blood cell debris was removed from 1.0 ml of the culture by a 2-s microcentrifugation step, and the bacteria in the supernatant were subsequently isolated by centrifugation for 2 min. The isolated bacteria were washed in 1.0 ml of PBS and resuspended in 0.75 ml of PBS. Fifty microliters of settled guinea pig erythrocytes was then added. The subsequent incubation, washing, and enrichment were repeated an additional two times. After the last outgrowth, cultures were streaked onto L-agar plates, and 20 individual colonies were scored for their ability to agglutinate erythrocytes.

Erythrocyte binding assay. Overnight cultures of fimH point mutants and positive and negative control strains (described below) were harvested by microcentrifugation, resuspended in PBS, and mixed 1:1 by volume to give a final concentration of approximately 5 × 10⁶ cells/ml. Sixty microliters of each mixture to be tested was added to an Eppendorf tube. Ten microliters was removed, diluted, and titered on maltose-tetrazolium plates to obtain the ratio of the two strains. To the bacteria remaining in the tube, 0.1 ml of settled guinea pig erythrocytes was added. The resulting suspension was then gently mixed and incubated at room temperature for 10 min. Adherent bacteria were removed by a brief (approximately 3 s) centrifugation to pellet erythrocytes and erythrocyte-bound bacteria. A portion of the supernatant was diluted and plated, and the resulting ratio was compared to the starting ratio.

Macrophage binding assay. Resident (unelicited) peritoneal macrophages from male BALB/c mice 8 to 12 weeks of age were used in these experiments. Macrophages were harvested, delivered into 48-well cluster culture plates, and incubated overnight as described previously (11). Macrophages (approximately 1 × 10⁵ to 2 × 10⁵ cells/well in 0.5 ml of tissue culture medium) were exposed to approximately 10⁶ E. coli (grown overnight in -broth, and harvested and mixed pairwise as described [11]) that were added in 25 µl of PBS for 10 min at 37°C. After incubation, wells were washed four times (each wash was with 0.5 ml of PBS). After the final wash, 0.5 ml of PBS containing 0.1% Triton X-100 was added to each well to lyse the macrophages. Approximately 5 min after the Triton X-100 additions, the contents of the wells were diluted and plated. The exposure of bacteria to Triton X-100 had no effect on bacterial viability. Control wells that contained no macrophages were used to assess nonspecific binding of bacteria. In no instance was the level of binding appreciable (>10% of that of macrophage-containing wells). When -methyl-mannoside (mm) was added to inhibit E. coli binding, a small volume of a 1.0 M mm stock solution (in PBS) was added to a final concentration of 50 mM. Control wells had an equivalent volume of PBS added. PBS solutions used to remove unbound E. coli also contained 50 mM mm.

Statistical and DNA sequence analysis methods. Standard deviation of the mean was calculated with the Microsoft Excel STDEV function. Standard error was calculated as the standard deviation divided by the square root of the number of experiments. The significance of mean differences was determined by Student's t test. Both tests were provided by the Microsoft Excel version 4 statistics package. Statistically significant differences were defined as a P of <0.05. Genetics Computer Group (version 7) SIGCLEAVE analysis using the cleavage rules of von Heijne (49) was used to assess possible alterations in the FimH signal cleavage site due to signal sequence mutations in fimH.

Nucleotide sequence accession numbers. The sequences of the five representative fimH mutant alleles have been deposited in GenBank with accession numbers as follows: fimH241, AF154925; fimH236, AF154926; fimH244, AF154927; fimH205, AF154928; and fimH208, AF154929.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Mapping point lesions in fimH. DNA sequencing of the plasmid-borne fimH alleles used to create the chromosomal mutants described by Harris et al. (13) provided the sequence of both strands of the entire fimH gene. The sequence of 11 alleles revealed that 5 were unique, each having a single lesion at the sites marked in Fig. 1. (The area sequenced included the fimH coding region plus a minimum of 20 bp on either side.) Sequencing of PCR amplicons from the chromosomal fimH alleles in all 11 strains revealed that they contained the same lesion as the plasmid-borne allele. (The entire gene was not sequenced in these cases [data not shown].) As a separate test that the lesions sequenced were responsible for the hemagglutination-negative phenotype, recombination mapping was carried out. This mapping involved enrichment for hemagglutination-positive individuals that were produced as a result of recombination between a chromosomal fimH allele having one of the point mutations and a set of in vitro-generated deletion derivatives of fimH residing on plasmids. The results of this mapping procedure (Fig. 2) were in good agreement with the lesion location indicated by DNA sequencing. Also, since the mutant alleles were not resequenced in their entirety after their introduction into the chromosome via homologous recombination (13), this mapping procedure indicated that unappreciated sequence differences in or around the chromosomal fimH gene were not responsible for the mutant phenotype.

View larger version (33K): [in this window] [in a new window]   FIG. 1.   Diagram of the fimH gene (diagonally striped rectangle) in the chromosome of strain ORN155. The entire fim gene cluster (fimBEACDEFGH) is located at approximately 98 min on the E. coli genetic map (more precise coordinates can be obtained from GenBank, accession no. AE000502, and reference 2). All genes in the cluster are transcribed left to right. The Tn5 insertion adjacent to fimH is not drawn to scale. Approximately one-third of the 5' end of the fimH gene is expanded to show the sequence. Nucleotide numbering follows the convention in GenBank. Amino acids are numbered, with positive numbers denoting residues in the mature protein. Negative numbers indicate the amino acids in the signal sequence. The caret (^) below the sequence denotes the signal processing site (12). Mutation sites are designated by asterisks (*) and allele number. Arrows show the base change involved. A minus sign indicates a deletion. Amino acid changes accompanying the mutations and the phenotypes conferred by the lesions are summarized in Table 2.

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Mapping of mutations shown in Fig. 1 by recombination. Mutants with chromosomal fimH alleles shown on the right side of the figure were incubated with the plasmids shown at the left. Recombinants capable of hemagglutination were recovered following enrichment as described in the text. The fimH gene is denoted by the diagonally striped rectangle. The dotted line indicates the region deleted. Specific deletions (codons deleted) from the 3' end of FimH are listed in Table 1. Deletions were created by removing either the restriction fragments indicated or by using Bal31 exonuclease digestions starting from the XhoI site in pORN118 as described in the text.

Variety of phenotypes displayed by fimH point mutants. Whereas all fimH mutants failed to agglutinate guinea pig erythrocytes, certain fimH point mutants displayed additional properties (summarized in Table 2). Two of these properties involved the generation of aberrant pilus morphologies (class II and class III), and a third involved a conditional agglutination phenotype, all of which were initially noted by Harris et al. (13). Class II mutants (represented here by strain ORN164) had longer than normal pili. Class III mutants (represented here by strain ORN158) had much longer and very sparse pili. The lesion associated with the class II phenotype predicted a change in the last amino acid at the signal sequence site from serine to leucine. The class III mutant had a frameshift lesion predicted to generate a truncated FimH product approximately 40% of the normal size. In addition to the aberrantly fimbriated mutants, a mutant with a conditional (temperature sensitive [ts]) erythrocyte agglutination phenotype was isolated (Table 2). This mutant (strain ORN157, carrying the fimH205 allele) had morphologically similar pili at both the permissive and restrictive temperatures but failed to agglutinate erythrocytes at the restrictive temperature (13).

Is the defect in the fimH products one of localization or function? In the initial characterization of fimH point mutants (13), it was inferred that the fimH product was being expressed and properly localized because the mutants formed pellicles in static broth, a property eliminated by fimH insertion mutations (13). However, in these tests, we could not conclude that parental amounts of mutant FimH were being expressed because the level of FimH expression needed for pellicle formation could have been lower than that needed for erythrocyte agglutination.

FimH expression in the ts mutant at the restrictive and permissive temperatures. Bacterial agglutination of ORN157 (carrying the ts fimH205 allele) using dilutions of FimH-specific antiserum revealed no noticeable difference between the parent and mutant in terms of antibody reactivity after growth at both the permissive and restrictive temperatures (summarized in Fig. 3A). In contrast, erythrocyte agglutination effected by the mutant was decidedly reduced (Fig. 3B).

View larger version (56K): [in this window] [in a new window]   FIG. 3.   Microtiter agglutination reactions of strain ORN157 bearing the fimH205 allele compared to the parental strain (ORN155) and strain ORN133, a fimH insertion mutant (fimH'-kan). (A) Bacterial agglutination of mutants grown at the temperatures indicated, 31 or 42°C, concentrated twofold from overnight cultures, and incubated with FimH antiserum (1:4 dilution) as described in the text. (B) Microtiter guinea pig erythrocyte agglutination by the same strains obtained under the same growth condition. Incubation conditions are described in the text.

Role of temperature in FimH-mediated erythrocyte agglutination in fimH205 mutants. Two ways that a temperature shift could effect a change in FimH function are (i) via a spontaneous (instantaneous) conformational change in existing FimH molecules and (ii) via a conformational change in nascent or newly synthesized fimH product (requiring time for new synthesis). We tested these two possibilities by examining the kinetics with which the hemagglutination phenotype (Hag) changed after a temperature shift in the presence and absence of sufficient chloramphenicol to halt protein synthesis. Our results (Fig. 4) revealed that the change from Hag⁺ to Hag (and vice versa) was not instantaneous and required new protein synthesis. Thus, it appeared that a temperature shift could not induce a structural change in existing FimH molecules.

View larger version (23K): [in this window] [in a new window]   FIG. 4.   Kinetics of the appearance of a hemagglutination-positive or -negative phenotype after a shift in temperature from 31 to 42°C and vice versa. (A and B) Emergence and disappearance, respectively, of the hemagglutination phenotype following a temperature shift. (C and D) Same shifts but in the presence of chloramphenicol (20 µg/ml). Diamonds represent points on growth curves taken from at least two experiments. Optical density at 600 nm values were normalized to the starting value in each experiment. Consequently, the units are arbitrary. The line shown through the points is a trend line drawn from a logarithmic regression analysis of the points. Hemagglutination was scored as a positive (+), weakly positive (+/), or no hemagglutination () depending on the strength of the agglutination reaction. To test hemagglutination, samples were removed at the times indicated, diluted (or concentrated) to a constant optical density, and tested for hemagglutination of guinea pig erythrocytes as described in the text. Error bars represent standard error of the mean.

Erythrocyte and macrophage binding by fimH205 mutants. Guinea pig erythrocyte binding was greatly influenced by temperature in the fimH205 mutant: fimH205 mutants bound erythrocytes with statistically the same effectiveness as a fimH insertion mutant at 42°C but had the binding effectiveness of the parental strain at 31°C (Fig. 5). In contrast, the ability of the fimH205 mutant to bind to macrophages was statistically the same as that of the parent at both 31 and 42°C (Fig. 5). For comparison, the results of this binding assay with the four other fimH mutants (the ones that did not agglutinate in FimH antiserum at parental levels) are shown. These four mutants bound with the same effectiveness (statistically) as the fimH insertion mutant.

View larger version (25K): [in this window] [in a new window]   FIG. 5.   Comparison of the binding ability of mutants having the fimH lesions indicated relative to a fimH insertion mutant (fimH'-kan strain ORN133, denoted as FimH in the legend box) and compared to the parental (ORN155) strain for their ability to bind guinea pig erythrocytes (RBCs) and resident BALB/c peritoneal macrophages. All mutant strains were grown at 37°C except for the fimH205 mutant, which was grown at the temperatures indicated. Parental binding values (ratio of the parental strain to the fimH insertion mutant; far right-hand bars were not significantly affected by the growth temperatures employed (31, 37, or 42°C), and for economy, a single average value is shown. Vertical bars represent the standard error of the mean. The dashed line indicates a 1:1 ratio between the test strain and the fimH insertion mutant. At least two experiments were averaged.

Macrophage binding by the ts mutant grown at the restrictive temperature was mannose insensitive. The ability of the nonmetabolizable mannose analog mm to inhibit macrophage binding was assessed with a fimH205 mutant (ORN183) grown at 31 and 42°C. The degree to which the addition of 50 mM mm inhibited binding was compared with binding by the parental strain (Fig. 6). Binding inhibition was quantitated by normalizing the reduction in binding effected by mm on the parent and mutant relative to a fimH insertion mutant present as an internal control in all assays. For the parental strain, mm inhibited binding approximately 3.5-fold regardless of the temperature (31, 37, and 42°C tested). For the mutant, however, mm failed to change the binding significantly in 42°C-grown cells. After growth at 31°C, mm inhibited macrophage binding of the mutant to a degree that was statistically the same as inhibition in the parent.

View larger version (34K): [in this window] [in a new window]   FIG. 6.   Degree of inhibition of macrophage binding by mm in the fimH205 mutant grown at the restrictive and permissive temperatures. Macrophage binding was carried out in microtiter wells containing either the parental strain (ORN115) or a fimH205 mutant strain (ORN183) each mixed with a fimH'-kan insertion mutant strain (ORN204) as described in the text. The degree to which mm inhibited binding of the parent or fimH205 mutant (ordinate values) was calculated from the change in ratio of these strains normalized to the change exhibited by the fimH'-kan insertion mutant. A value of 1 (dashed line) defines the value expected if there was no effect of mm addition. Results presented are the averages of three separate experiments, each performed in duplicate. Error bars denote standard error of the mean.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The results reported herein define the location of lesions and phenotypic properties of five fimH point mutants isolated after site-directed mutagenesis (13). Although 11 independently isolated fimH mutants were initially identified, pilot experiments revealed that some had identical fimH lesions. The five described here represent one of each allelic type isolated. One mutant with a novel conditional erythrocyte binding phenotype was further defined as having an altered binding specificity after growth under permissive and restrictive conditions.

The fimH mutants examined here were obtained by enriching for individuals that formed pellicles in static broth (a property associated with FimH) in the presence of mm (13). mm is a mannose analogue that inhibits pellicle formation as well as erythrocyte agglutination by type 1 pili. Since pellicle formation is eliminated in fimH insertion mutants (13), we expected that most, if not all, of the fimH mutants would express parental levels of FimH and that the lesions would thus define regions of FimH required for erythrocyte binding. However, in the present study, bacterial agglutination tests with polyclonal antiserum raised against a LacZ-FimH fusion protein did not confirm that parental amounts of FimH were being produced in four of the five mutants. Whereas it is possible that some of these mutants have lesions that define areas of FimH responsible for erythrocyte binding and are poorly reactive with antibody as a consequence, we could not rule out the possibility that their failure to agglutinate (or bind to) erythrocytes was due entirely to inadequate FimH exposure or expression.

Two of the four mutants that did not express normal amounts of FimH-cross-reacting material (those with the fimH236 and fimH208 alleles) displayed signs of defective pilus biogenesis (13). We assume that their phenotypes result from the aberrant routing of the defective products in the pilus biogenesis process. In the case of the fimH236 mutation, the lesion lies in an area of the gene encoding the signal sequence (12), and the altered amino acid would thus not be part of the mature protein. However, the fimH236 lesion may result in an alteration in the site at which the signal is cleaved (49), creating a defect in the amino-terminal portion of the mature protein. The nature of the fimH208 lesion (a frameshift lesion approximately one-third of the way into the gene produced a garbled sequence of 33 amino acids before terminating, leaving a product approximately 40% of normal size, 33% of which was garbled) produces a mutant with the most dramatically altered pilus morphologic phenotype (13). Such mutations may affect the ability of the product to properly interact with the FimC chaperone protein (14) or with minor pilus components needed for pilus assembly (15, 36, 39). The recently acquired crystalline structure of the FimC-FimH complex should prove helpful (6) in this regard.

By far the most straightforward mutant to characterize was the ts fimH205 mutant. This mutant had parental levels of FimH and had a very pronounced phenotype. The lesion defining this allele converted a leucine to an arginine approximately one-fifth of the way through the mature FimH protein (amino acid position 58). The location and possible importance of this particular amino acid have been previously noted in studies by Sokurenko et al. (45, 46), who examined naturally occurring fimH alleles. In one of their studies, the fimH allele from E. coli K-12 strain CSH50 had arginine at the 58 position (GenBank accession no. A36976). This allele conferred the ability to bind to periodate-treated fibronectin in a mannose-inhibitable fashion. In fact, only one additional amino acid change in the CSH50 fimH allele, at position 201 (a histidine in place of a threonine), kept the two alleles from being identical. (Our parental fimH allele was identical to that of E. coli K-12 strain MG1655 that has been completely sequenced [2]; GenBank accession number U14003.) Sokurenko et al. (46) commented on the similarity of the fimH205 mutants to carry out protein-protein interactions in pellicle formation and the ability of the CSH50 FimH to bind to periodate-treated fibronectin. It was thus reassuring to find, in the present study, that the two mutants have one very specific genotypic feature in common. However, in the report of Sokurenko et al. (46), no mention was made of a conditional nature of fibronectin binding. Also, the binding reported was subject to inhibition by mannose. (The protein-protein binding conferred by the fimH205 allele in forming a pellicle [13] and in binding to macrophages was insensitive to mannose inhibition.) A recent report by Pouttu et al. (35) has indicated that the collagen binding ability of FimH is related to an amino acid change at residue 62. Whether this binding was mediated through protein-protein interactions in collagen was not specifically addressed. In any case, it appears that residues 58 to 62 of the FimH protein are important for determining receptor binding specificity. In addition to the above, it has been previously reported that linker insertion mutagenesis that changed residue 56 abolished mannose-sensitive binding (41).

In the present study, the nature of the ts defect in products from the fimH205 allele was pointed out in experiments that measured phenotypic lag after a temperature shift. Since new protein synthesis was required to effect a change in phenotype after a temperature shift, presynthesized FimH evidently could not undergo a conformational change to assume the new activity. Rather, a new (or nascent) FimH molecule was needed to assume the new configuration responsible for the altered activity. Also, the differences in phenotypic lag when shifting from the permissive to the nonpermissive temperature and vice versa were consistent with the idea that comparatively few pili (containing functional FimH products) were sufficient for hemagglutination. That is, acquisition of the hemagglutination-positive phenotype after a temperature shift was much more rapid than the loss of the hemagglutinating ability after a reverse shift.

Of most interest to us was the finding that the ts mutant carrying the fimH205 allele had different eucaryotic cell binding specificities when grown at the permissive and restrictive temperatures. At the restrictive growth temperature, the fimH205 mutant, while not binding to erythrocytes, still bound to macrophages through a FimH-specific interaction that was insensitive to mannose inhibition. The most likely explanation for this phenotype involves the earlier observation (10) that at the restrictive growth temperature, this mutant forms pellicles that appear to involve FimH-FimH interactions that are insensitive to mannose inhibition. This protein-protein type of interaction may be employed here in binding to macrophages. That is, macrophages may have a protein(s) on their surface capable of interacting with the mutant version of FimH. This explanation implies that erythrocytes lack such a protein(s). In related experiments by Sokurenko et al. (43), a correlation was made between the degree of affinity of FimH products (from naturally occurring fimH alleles) for monomannose and their affinity for cultured uroepithelial cells. It would thus appear that FimH has the potential to display a number of binding specificities.

Factors that determine FimH binding specificity have the potential to be exploited for processes that require conditional attachment and release of microorganisms from specific ligands. Already, the natural binding affinity of FimH for mannose has been used in combination with other binding domains to create chimeric adhesins (40). Another use for adhesin mutants with altered receptor specificities is in experiments examining the effects of bacterium-eucaryotic cell interactions. The effects of bacterial attachment on eucaryotic cell physiology are just beginning to be understood (1, 8, 22). The effects of specific types of attachment on eucaryotic cells are almost always deduced from experiments in which bacterial mutants that lack the attachment organelle (or the adhesive part of the organelle) are used as negative controls (9, 11, 16). In the case of type 1 pili, such mutants do not bind eucaryotic cells with an efficiency high enough to serve as a truly appropriate control (i.e., one that eliminates the effects of nonspecific binding rather than just the elimination of binding entirely). Attachment specificity mutants, such as the one isolated here, may allow better-controlled assays of the effect of receptor-ligand interactions to be determined.