Ramifications of kinetic partitioning on usher‐mediated pilus biogenesis (original) (raw)

Introduction

A critical virulence mechanism common to most pathogenic bacteria is adherence to host tissues. Initial colonization events have been shown to be crucial in many different infectious processes, e.g. the binding of Escherichia coli to Galα(1–4)Gal sugar moieties in the kidney prior to development of pyelonephritis (Roberts et al., 1994). Bacterial adhesins mediate attachment by binding with stereochemical specificity to complementary receptors present on eukaryotic cells. Adhesins can be assembled onto the surfaces of bacteria as monomers, simple oligomers or as parts of supramolecular fibers (Hultgren et al., 1996). The most well‐characterized adhesive fibers are known as pili or fimbriae, which are rod‐like organelles 5–7 nm in diameter (Mitsui et al., 1973; Hultgren et al., 1996).

Type 1 pili are produced by the entire Enterobacteriaceae family. Their expression requires at least eight genes organized in the type 1 gene cluster (Hull et al., 1981; Orndorff and Falkow, 1984) (Figure 1). Type 1 pili consist of two distinct subassemblies: a thin tip fibrillum which is joined to the distal end of a pilus rod (Jones et al., 1995). The 7 nm wide, 5–7 μm long rod is comprised of FimA subunits arranged in a right‐handed helical conformation with 3.125 residues per turn (Mitsui et al., 1973). The tip fibrillum is ∼16 Å long and primarily contains the adhesive moiety of the pilus, FimH (Jones et al., 1995). In uropathogenic E.coli, FimH mediates binding to mannose‐containing receptors (Minion et al., 1989; Krogfelt et al., 1990) exposed on the luminal surface of both the human and mouse bladder epithelium (Langermann et al., 1997). This binding event has been shown to be critical in the ability of uropathogenic E.coli strains to colonize the bladder and cause cystitis in mice (Langermann et al., 1997). A FimH‐based vaccine has been shown to elicit IgG antibodies that block colonization in vivo and protect mice against a mucosal infection of the bladder and subsequent ascending urinary tract infections (Langermann et al., 1997). FimF and FimG have been suggested to play roles in initiation and termination of pilus assembly, respectively (Russell and Orndorff, 1992).

Figure 1

Schematic representation of the type 1 (fim) and P (pap) gene clusters listing the functions of the various gene products.

The assembly of >27 different adhesive organelles in diverse Gram‐negative bacteria requires a periplasmic chaperone and an outer membrane usher (Hung et al., 1996). FimC is the periplasmic chaperone required for type 1 pilus assembly (Klemm et al., 1992; Jones et al., 1993). It is one of the 27 members of the PapD chaperone family which are 30–60% identical to one another in amino acid sequence (Hung et al., 1996). The crystal structure of PapD, the chaperone involved in P‐pilus assembly, has been solved to 2.5 Å resolution. The structure revealed a two‐domain boomerang‐shaped molecule with each domain having an immunoglobulin‐like topology (Holmgren and Brändén, 1989). Molecular modeling revealed that all of the PapD‐like chaperones adopt similar immunoglobulin‐like two‐domain structures (Hung et al., 1996). PapD‐like chaperones bind to the C‐terminus of subunits via a beta zippering interaction (Kuehn et al., 1993; Bullitt et al., 1996). This interaction results in the partitioning of a subunit from a membrane‐associated state to a chaperone‐associated periplasmic complex (Jones et al., 1997). The maintenance of the C‐terminus of the subunit in an extended conformation in the context of the chaperone may predispose subsequent folding reactions of the subunit on the chaperone template (Jones et al., 1997). Chaperone–subunit complexes are targeted to outer membrane ushers where the chaperone is dissociated. This exposes the conserved C‐terminal region which is involved in mediating subunit–subunit interactions within the assembled organelle (Soto,G., Dodson,K., Liu,C., Jones,C.H., Ogg,D., Knight,S. and Hultgren,S.J., in preparation).

FimD is the outer membrane usher required for type 1 pilus assembly. When fimD is disrupted with a Tn_5_ insertion, bacteria are unable to produce type 1 pili (Klemm and Christiansen, 1990). Recently it was shown that the usher forms a channel, possibly providing for the translocation of subunits across the outer membrane (Thanassi et al., 1998); however, little is known about the structure, function or mechanism of action of usher proteins. PapC, the usher in the P‐pilus system (Figure 1), has been shown to differentially recognize chaperone–subunit complexes relative to the final position of the subunit in the pilus (Dodson et al., 1993). This information has led to the hypothesis that the final position of a subunit in the pilus is determined in part by the strength with which the subunits (complexed with the chaperone) bind to the usher assembly site. The usher is thus a multifunctional protein that presumably has a number of functional regions including membrane‐spanning, pore‐forming and periplasmic domains. The periplasmic domain presumably interacts with incoming chaperone–subunit complexes. Molecular modeling and topology studies suggest that the FaeD usher has a middle transmembrane portion, and that the N‐ and C‐terminal regions of the protein form periplasmic domains (Valent et al., 1995).

Despite the critical role of pili in colonization and virulence, the usher‐mediated mechanism by which bacteria are able to incorporate an adhesin into the pilus is poorly understood. The experiments reported here have uncovered critical chaperone–subunit–usher interactions that have revealed a kinetic partitioning mechanism by which the usher mediates ordered pilus assembly. Due to the almost ubiquitous use of membrane‐associated adhesive organelles and secreted proteins by pathogenic bacteria, understanding the events leading to the biogenesis of adhesive organelles and protein translocation across the outer membrane is of fundamental importance. Utilizing the tools we have developed in the type 1 pilus system, we have the unique opportunity to dissect the initial events in the process by which a virulence‐associated adhesin is translocated across the outer membrane and incorporated into an adhesive organelle.

Results

Purification and characterization of the usher and chaperone–subunit complexes

A histidine tag was genetically engineered onto the 3′ end of the fimD (usher) gene to facilitate purification. This gene product was fully functional in vivo as it was able to complement a fimD − type 1 operon (pETS6) using a standard hemagglutination assay (see Materials and methods). The plasmid containing His‐tagged fimD, pETS4, was transformed into the type 1 − strain, ORN103, and protein production was induced. The 92 kDa FimD protein was present in outer membrane preparations as determined by SDS–PAGE (Figure 2A) and N‐terminal sequencing. FimD was purified to >90% homogeneity by subjecting the outer membrane preparations to Ni2+ chromatography (Figure 2A, lane 5).

Figure 2

Purification of usher and chaperone–subunit complexes. (A) Coomassie Blue‐stained 12.5% SDS–PAGE of ORN103/pETS4 membrane fractions. FimD expression was induced with 20 μM IPTG (lanes 2 and 4). Lanes 1 and 3 are the uninduced controls. Lanes 1 and 2 are inner membrane fractions; lanes 3 and 4 are outer membrane fractions. FimD was only observed in the induced outer membrane fractions (lane 4, arrow). Lane 5 shows FimD purified from the outer membrane extracts by Ni2+ chromatography (see Materials and methods). Molecular weight markers in all the figures are in kilodaltons. (B) Coomassie Blue‐stained 12.5% SDS–PAGE showing the FimC chaperone alone (lane 5), and FimCH (lane 1), FimCG (lane 2), FimCF (lane 3) and FimCA (lane 4) complexes purified by Ni2+ chromatography and FPLC as described in Materials and methods.

A histidine tag was also genetically engineered onto the 3′ end of the fimC (chaperone) gene to facilitate purification of chaperone–subunit complexes. The FimC‐His tag protein was functional in assembling type 1 pili as it was able to complement a fimC − type 1 operon (pETS12) (see Materials and methods). We transformed the plasmid pETS1000 containing His‐tagged fimC into ORN103 alone or in combination with pETS2A (fimG), pETS1 (fimF), pHJ20 (fimH) or pETS5 (fimA). Following protein induction, periplasmic extracts containing chaperone–subunit complexes were subjected to Ni2+ chromatography and FPLC. Subunits bound to FimC were co‐purified as chaperone–subunit (CH, CG, CF and CA) complexes as determined by SDS–PAGE (Figure 2B). These results reveal that FimC binds to and forms chaperone–subunit complexes with minor subunits (FimG and FimF) and the major subunit (FimA). FimC and FimC–FimH complexes have been identified and purified previously (Jones et al., 1993).

Chaperone–subunit–usher interactions

We used the purified proteins shown in Figure 2 in an enzyme‐linked immunosorbent assay (ELISA) to investigate potential interactions between the FimD usher and the chaperone or the various chaperone–subunit complexes. A similar ELISA procedure was performed previously in the study of the PapC usher (Dodson et al., 1993).

FimC–FimH complexes had the highest affinity for FimD (Figure 3). Binding of all of the other complexes and of FimC alone to FimD also occurred, but at a very low level. The strong binding of FimC–FimH complexes to FimD may in part facilitate the localization of the FimH adhesin at the distal end of the tip fibrillum of type 1 pili. This is consistent with the model that Dodson et al. (1993) proposed for differential recognition of chaperone–subunit complexes in P‐pilus assembly. It is not apparent from the ELISA data what roles might be played by the low level of FimC, FimC–FimG, FimC–FimF and FimC–FimA binding.

Figure 3

Chaperone–subunit–usher interactions measured via ELISA. Microtiter plate wells were coated overnight with 5 μg/ml of pure FimD. Wells were blocked with 3% BSA/PBS and then various concentrations of pure FimCA (‐□‐), FimCF (‐⋄‐), FimCG (‐×‐), FimCH (‐●‐), or FimC alone (‐○‐) were incubated in the wells for 45 min at room temperature. Controls were done with no FimD and with no complexes added (see Materials and methods). Binding to FimD was detected with anti‐FimC antibody followed by alkaline phosphatase‐linked anti‐rabbit IgG. Absorption at 405 nm was read after addition of _p_‐nitrophenol phosphate. Points shown represent an average value from experiments done in triplicate.

Investigation of real time binding events

We reasoned that a study of association and dissociation events in real time would yield additional insight into the ordered mechanism of pilus assembly. Using surface plasmon resonance technology (Karlsson et al., 1991), we attempted to determine both kinetic and equilibrium association and dissociation constants. Using a Biacore 2000 (Biosensor) machine, we immobilized the FimD usher on the surface of a research grade CM5 Biosensor chip using a standard amide cross‐linking procedure. Then, we injected various concentrations of purified FimC chaperone or the various chaperone–subunit complexes over the immobilized usher. Representative binding curves for FimCH are shown in Figure 4A. Equilibrium evaluation showed that the FimC chaperone alone was unable to bind to FimD under these experimental conditions. FimCA, FimCG and FimCF bound relatively weakly to the usher, with calculated equilibrium constants (_K_D) of 176 nM (FimA), 670 nM (FimCG) and 1.37 μM (FimCF) (Figure 4B). In contrast, chaperone–adhesin (FimCH) complexes bound to FimD 20‐ to 150‐fold better than the other complexes, having a _K_D = 9.1 nM. The kinetic basis of the higher affinity of FimCH resided in the _k_on rate (4.3×105 M−1 s−1) which was ∼8‐ to 35‐fold faster than the _k_on rates of FimCG, FimCF and FimCA complexes (Figure 4C). In each interaction of the usher with the chaperone–subunit complexes, the _k_off was very slow, as only ∼1.0% of the total number of chaperone–subunit–usher complexes dissociated per second (Figure 4D). Complete pilus assembly occurs in minutes (Jacob‐Dubuisson et al., 1994); therefore, in physiological terms, individual chaperone–subunit complexes essentially remain attached to the usher once the initial interaction occurs. The differences in rates of dissociation would not necessarily affect the order of events in pilus assembly because before FimCH could dissociate from FimD, other subunits would presumably be added to FimH, resulting in the growth of a pilus. Accordingly, the ability of FimCH complexes to associate first or fastest with FimD appears to be a critical factor in insuring the tip localization of the FimH adhesin in the pilus.

Figure 4

Real‐time analysis of chaperone–subunit–usher interactions. (A) Representative binding curves from FimCH–D interactions. The curves (from top to bottom) represent decreasing concentrations of FimCH complex (from 1.0 to 0.075 μM) injected over immobilized FimD. Binding of FimCH to FimD is detected by the Biacore machine and converted to an arbitrary response unit (RU) over time (s) that can be analyzed to obtain kinetic and equilibrium constants. The regions representing real‐time association and dissociation events are labeled. (B) Log scale graphs showing equilibrium dissociation constants (_K_D in molar concentration units) for various chaperone–subunit–usher interactions in the type 1 (left graph) and P‐pilus (right graph) systems. For (B–D), the numbers listed above each bar in the graph represent the actual binding constant for that particular chaperone–subunit complex. The columns labeled PapDA* represent experiments done with PapAG150T (Bullitt et al., 1996) complexed with PapD. Standard errors shown graphically for _k_off and _k_on are an average of the standard errors from all measurements taken for a particular complex. The standard error for _K_D is a combination of the errors from the _k_off and _k_on (see Materials and methods). The chaperone–adhesin (FimCH or PapDG) complex in each system bound to the usher with a much greater affinity than any of the chaperone–subunit complexes. The chaperone alone (FimC or PapD) did not show any binding to the usher (FimD or PapC, respectively) when injected (up to 10.0 μM). The chaperone–adhesin complex (PapDG or FimCH) that was ‘mismatched’ (to FimD or PapC, respectively) also did not show any binding when injected (up to 2.0 μM). (C) Log scale graphs showing kinetic association constants (_k_on in M−1 s−1) for type 1 and P‐pilus systems. The chaperone–adhesin complexes in both systems bound fastest. (D) Log scale graphs showing kinetic dissociation constants (_k_off in s−1) for type 1 and P‐pilus systems. The ‘off’ rate is slow for all of the complexes (∼1% of the complexes dissociating per second).

Biacore binding experiments were also done using the P‐pilus system to determine if the principle of a rapid association and slow dissociation of chaperone–adhesin complexes to the usher is a general phenomenon. Analogous to the experiments with the type 1 system, the PapC usher was the immobilized ligand. The binding analytes were the PapD chaperone, a PapDG chaperone–adhesin complex and a chaperone–subunit complex (PapDAG150T). PapA subunits make up the pilus rod (Bullitt et al., 1996). The G150T PapA mutant was used because it does not self‐aggregate like wild‐type PapA (Bullitt et al., 1996). Self‐association could possibly interfere with the Biacore analysis. As was found in the type 1 system, the PapDG chaperone–adhesin complex had a 16‐fold greater affinity (_K_D = 90 nM) (Figure 4B) and a 29‐fold faster kinetic rate of association (_k_on = 5.7×104 M−1 s−1) (Figure 4C) with the usher in comparison with the PapDAG150T complex. Additionally, the kinetic dissociation rates (_k_off) remained slow (Figure 4D) and the PapD chaperone alone was unable to interact with the usher (Figure 4B). Therefore, analogous to what was seen in the type 1 pilus system, the faster association rate of the PapDG complex with the PapC usher may play a critical role in enabling the PapG adhesin to take its place as the first subunit incorporated into the pilus. In previous experiments (Dodson et al., 1993), the chaperone–major subunit complex (PapDA) was not shown to bind to the usher (PapC). The fact that we were able to detect binding of PapDA and of FimCA to their respective ushers (Figure 4B) is probably due to the use of pure protein and to the increased sensitivity of the surface plasmon resonance technique. The inability to detect binding of the chaperone alone or the mismatched chaperone–adhesin complex to the usher (Figure 4B) argues that even the weak interactions observed are biologically relevant.

Changes in trypsin susceptibility of FimD upon FimCH binding

Next, we wanted to determine if usher–chaperone–subunit interactions could be detected in vivo. We hypothesized that these interactions might result in conformational changes in the usher which could be detected by differences in sensitivity to trypsin. This assay was used successfully for studying surface‐exposed loops and mapping the topology of the K88 usher, FaeD (Valent et al., 1995). Trypsin digestion of whole cells expressing the FimD usher alone or co‐expressed with either FimC, FimCF, FimCG, FimCH or FimCA was followed by Western blotting of outer membrane preparations using anti‐FimD antiserum (Figure 5). When FimD was expressed alone or with FimC, FimCF, FimCG or FimCA individually, it was susceptible to complete proteolytic degradation by trypsin under the conditions used. It was somewhat surprising to find the outer membrane usher protein to be completely susceptible to externally added trypsin. This indicates that FimD must have a high number of surface‐exposed trypsin‐cleavable sites or that cleavage at one or a few sites leads to destabilization and further degradation of the protein. However, as seen in Figure 5, lane 13, a large portion of the FimD usher was protected from proteolytic digestion when the chaperone–adhesin complex, FimCH, was co‐expressed. The same results were obtained when whole cell suspensions were probed by Western blot analysis (data not shown), arguing that additional digestion did not occur during preparation of the outer membranes. The FimD fragment protected by co‐expression of FimCH was purified by Ni2+ chromatography (Figure 7) and N‐terminally sequenced, demonstrating that it represents a 40 kDa C‐terminal domain of FimD beginning with amino acid K477 of the mature protein. Cleavage occurred in an extracellular region of FimD, because the periplasm was not accessible to trypsin in these whole cell experiments. This was confirmed by showing that the periplasmic FimC chaperone, which is trypsin‐sensitive (unpublished data), was not cleaved by trypsin in this experiment (data not shown). Finally, if we lowered the concentration of trypsin or decreased the digestion time, a small amount of this protected 40 kDa fragment could be seen in an anti‐FimD Western blot even when FimD was expressed by itself or with any of the other Fim proteins (data not shown). This suggests that the FimD usher may exist in an equilibrium between protected and unprotected conformations, and that interaction of FimCH with FimD drives the usher into the protected form.

Figure 5

Susceptibility of FimD to cleavage by extracellular trypsin. The presence of FimD in outer membrane extracts was analyzed by Western blotting with anti‐FimD antiserum before (lanes 1–7) or after (lanes 9–15) the cells were treated with trypsin. Lane 8 shows molecular weight markers. FimD was expressed by itself (lanes 1 and 15) or co‐expressed with FimC (lanes 2 and 14), FimCH (lanes 3 and 13), FimCG (lanes 4 and 12), FimCF (lanes 5 and 11) or FimCA (lanes 6 and 10). As a negative control, FimCH alone was expressed (lanes 7 and 9). The N‐terminal sequence of the trypsin‐protected portion of FimD (lane 13) was FTDYYN. This placed the trypsin cleavage site at K477 of the mature protein.

We next investigated whether the trypsin‐protected form of FimD induced by interaction with the FimCH complex is maintained throughout pilus biogenesis. The state of type 1 pilus‐assembling FimD was investigated using ORN103 cells containing the plasmid pSH2, which carries the entire type 1 operon (Orndorff and Falkow, 1984), or pETS6, a FimD knock‐out of pSH2 that does not produce pili (unpublished data). We also used the clinical isolate NU14 (Hultgren et al., 1986) and a fimH − mutant of NU14, NU14‐1 (Langermann et al., 1997). The cells were grown under conditions that resulted in the expression of type 1 pili, as confirmed by their ability to hemagglutinate guinea pig erythrocytes (see Materials and methods), and then subjected to the trypsin treatment described above. In all cases, the 40 kDa portion of the FimD usher that was protected by the co‐expression of FimCH was also protected during expression of type 1 pili (Figure 6). This indicates that the conformation of the usher that is stabilized by its interaction with the chaperone–adhesin complex is maintained throughout pilus assembly. It should be noted that in these experiments a substantial portion of full‐length FimD remained undigested by trypsin. This could be due to steric inhibition of the protease by the 7 nm wide pilus rod.

Figure 6

Trypsin susceptibility of FimD in cells producing pili. Outer membrane preparations of piliated NU14 or NU14‐1 (fimH −) cells (left gel) or ORN103 cells containing the type 1 pilus operon [pSH2 or pETS6 (fimD −)] (right gel) were analyzed for the presence of FimD by Western blotting with anti‐FimD antiserum before (lanes 1, 3, 5 and 7) or after (lanes 2, 4, 6 and 8) the cells were treated with trypsin. The abnormally wavy pattern of FimD migration (lanes 1–4) is probably due to the presence of wild‐type lipopolysaccharide in the clinical strain NU14. The different mobility of FimD in NU14 (lane 1) versus FimD in NU14‐1 (lane 3) may be due to the assembly state of the usher as pili are being assembled in NU14 but not in NU14‐1 (Langermann et al., 1997).

Figure 7

Purification of the ternary FimC–FimH–FimD assembly intermediate. (A) Coomassie Blue‐stained 12.5% SDS–PAGE of outer membrane preparations after Ni2+ chromatography. The preparations were from cells in which His‐tagged FimD (from pETS4) was co‐expressed with non‐His tagged FimCG (lane 1), FimCF (lane 2), FimCA (lane 3) or FimCH (lane 5). Lane 4 shows that non‐His‐tagged FimCH alone does not stick to Ni2+ beads. Lane 6 shows that the 40 kDa C‐terminal fragment protected from extracellular trypsin digestion can be purified by Ni2+ chromatography as a ternary complex with FimCH. (B) Western blots of the FimD–FimC–FimH complex eluted as a single peak from a Superdex 200 column after Ni2+ chromatography and developed with anti‐FimD (lanes 1) or anti‐FimCH (lanes 2) antibodies. Western blot of an extract from trypsin‐treated cells identical to that shown in (A), lane 6, developed with anti‐FimD (lane 3) or anti‐FimCH (lane 4).

Isolation of a FimD–FimC–FimH assembly intermediate formed in vivo

The values derived from the Biacore experiments suggest a strong and long‐lasting interaction between FimCH and FimD, and the trypsin digestion results show that this interaction occurs in vivo. Therefore, we attempted to isolate a pilus assembly intermediate targeting the step at which the FimD usher interacts with FimC–FimH complexes. We co‐expressed His‐tagged FimD (pETS4) with non‐His‐tagged FimC and FimH (pETS1004) in ORN103 cells, and purified FimD using Ni2+ chromatography as discussed above. We found that FimC and FimH co‐purified with FimD, suggesting a stable, ternary, usher–chaperone–adhesin (FimD–FimC–FimH) complex (Figure 7A, lane 5). Non‐His‐tagged FimCH complexes expressed in the absence of FimD could not be purified by Ni2+ chromatography (Figure 7A, lane 4). This is the first demonstration of the in vivo formation of an usher intermediate in pilus assembly. When FimCA, FimCG or FimCF were co‐expressed with FimD, none of these chaperone–subunit complexes co‐purified with the usher (Figure 7A, lanes 1–3), consistent with their weak interactions as seen in the Biacore experiments and their inability to protect FimD from trypsin. The putative FimD–FimC–FimH complex was analyzed by FPLC gel filtration chromatography (Superdex 200) and was shown to elute as a single peak (Figure 7B, lanes 1 and 2).

The sensitivity of the ternary complex to trypsin was also analyzed. We co‐expressed the FimD usher with FimCH and subjected the cells to the tryptic digestion reaction as discussed above. The outer membrane fraction was then subjected to Ni2+ chromatography which resulted in the co‐purification of the 40 kDa C‐terminal fragment of FimD with FimCH (Figure 7A, lane 6, and B, lanes 3 and 4). These data demonstrates that the probable binding site for the FimCH chaperone–adhesin complex is in the C‐terminal portion of FimD.

Discussion

Presentation of a pilus‐bound adhesin protein for optimal binding to eukaryotic cells is a critical event in bacterial pathogenesis (Roberts et al., 1994; Langermann et al., 1997). The assembly of >27 diverse adhesive fibers in many different Gram‐negative bacteria occurs via a multistep chaperone–usher pathway (reviewed in Hultgren et al., 1996). In the terminal step, protein subunits are translocated across the outer membrane and assembled into composite heteropolymeric organelles in a defined order. The translocation event depends on the outer membrane usher. The usher forms an oligomeric assembly that spans the outer membrane and has an ∼2 nm channel (Thanassi et al., 1998). The pilus subunits are maintained in assembly‐competent native‐like conformations on the periplasmic chaperone template as they are escorted to the outer membrane ushers. Presumably, these chaperone–subunit complexes specifically interact with a periplasmic domain on the usher. Subsequent events result in the exchange of subunit–chaperone interactions for subunit–subunit interactions that may drive the assembly of the pilus.

Utilizing Biacore 2000 technology to examine ‘real time’ interactions, we discovered that the FimCH chaperone–adhesin complex bound to the FimD usher with a 20‐ to 150‐fold higher affinity and with an 8‐ to 35‐fold faster kinetic association rate (_k_on) than the other chaperone–subunit complexes. The kinetic dissociation rates (_k_off) of all of the complexes were found to be physiologically slow, arguing that the targeting of FimCH first or fastest to FimD is a critical factor in determining the position of the FimH adhesin at the tip of the pilus. The slow dissociation rates may also be part of a kinetic partitioning mechanism to ensure that subunits are incorporated into pili subsequent to their interaction with the usher. We found that the same kinetic partitioning existed in the P‐pilus system, and thus it may represent a conserved mechanism in many or all of the chaperone–usher pathways to regulate the ordered assembly of adhesive organelles. It is uncertain to what extent kinetic partitioning may regulate the later steps in ordered assembly.

The interaction between the FimCH chaperone–adhesin complex and the usher, FimD, evidently represents a critical event in the assembly of the type 1 pilus. In the NU14 clinical isolate, a fimH − mutation (NU14‐1) nearly abolishes pilus assembly (Langermann et al., 1997). This effect has also been observed in another clinical isolate (Connell et al., 1996). We show here that interaction of the FimCH complex with FimD most likely transduces a conformational change to surface‐exposed regions of FimD, making them trypsin resistant. This conformational change is maintained throughout pilus assembly. In the fimH − NU14‐1 strain, the inability of the usher to adopt a trypsin‐resistant conformation may explain the lack of pili produced by this and other fimH − mutant strains. Therefore, we suggest that the FimCH interaction primes the usher for pilus assembly by driving it into an assembly‐competent conformation. However, there are reports suggesting that ΔfimH laboratory strains can produce pili (although at reduced levels) (e.g. Maurer and Orndorff, 1985; Klemm and Christiansen, 1987). This could be explained by a dynamic usher protein which, particularly in laboratory strains, exists in a conformational equilibrium between assembly‐competent and ‐incompetent states. Indeed, under less stringent conditions, we found that in the ORN103 lab strain, a small amount of the trypsin‐protected form of FimD is present even in the absence of FimCH. Our data clearly show, however, that the FimC–H–D interaction stabilizes the protected conformation of FimD, locking the FimD usher into a conformation that is capable of mediating pilus assembly. In their topological study of the K88 pilus usher, FaeD, Valent et al. (1995) observed that an ∼45 kDa portion of the usher was protected from extracellular trypsin digestion even though no chaperone‐subunit complexes were co‐expressed. A more complete degradation of the FaeD usher may not have been seen for a number of reasons. For example, there may be differences in the lengths of surface‐exposed loops and/or trypsin cleavage sites. Valent et al. (1995) did not investigate the effects of individual chaperone–subunit interactions with the FaeD usher and, therefore, it is unknown whether these interactions would confer altered protease susceptibility to FaeD.

An alternative explanation for the trypsin resistance conferred to FimD by co‐expression of FimCH is that the FimH adhesin becomes translocated to the cell surface where it sterically inhibits trypsin cleavage of the FimD usher. However, this does not appear to be the case. Using anti‐FimH antibodies in an immunogold electron microscopy procedure, we were unable to detect FimH on the surface of ORN103‐containing pETS4 (fimD)/pETS1001 (fimCH) and these cells were hemagglutination negative (data not shown). This argues that under conditions when only FimC, FimH and FimD are expressed, FimH is not translocated across the outer membrane.

The data presented here suggest the following model. After the FimCH chaperone–adhesin complex binds to the periplasmic C‐terminal domain of the FimD usher, it forms a stable ternary complex that results in a conformational change in the usher. This ternary complex, which is formed more quickly than any other usher–chaperone–subunit complex, probably acts as an ‘assembly initiation complex’, providing a template for the next incoming chaperone–subunit complex (FimCG) and locking FimD into an assembly‐competent state. There is no evidence that the interaction of FimCH with FimD results in an uncapping of the chaperone (FimC) from the adhesin (FimH). Interaction of the incoming FimCG complex with the FimC–H–D ternary complex would then result in uncapping of the chaperone, via an exchange of a FimH–FimC interaction for perhaps a more favorable FimH–FimG interaction. In the P‐pilus system, it has been shown that the chaperone binds to a highly conserved beta zipper motif at the C‐terminus of subunits (Kuehn et al., 1993). This same motif participates in subunit–subunit interactions after the uncapping of the chaperone (Bullitt et al., 1996). FimC has a similar mechanism of action to that of PapD (Jones et al., 1993). Thus, a surface on the tip fibrillum subunit, FimG, not capped by the chaperone probably fits to a complementary surface on FimH; however, that surface on FimH remains capped by the chaperone in the ternary FimCH–FimD complex. Uncapping of the chaperone from the C‐terminal beta zipper of FimH allows the FimG–FimH assembly interaction to occur. The C‐terminal beta zipper of FimG would then remain capped by the FimC chaperone until the next subunit (FimF) is added to the growing pilus. FimH–FimG–FimF–FimC assembly intermediates have been identified previously (Jones et al., 1995), demonstrating the feasibility of such a model.

We feel that the experiments presented here elucidate critical interactions in the biogenesis of type 1 pili. By understanding the roles that the usher FimD, the chaperone FimC and the adhesin FimH are playing in type 1 pilus biogenesis, we believe that a more generalized understanding of the events responsible for translocation across the bacterial outer membrane may be realized. The requirement for translocation of a protein across the outer membrane is not unique to the 27 known chaperone–usher pathways. Instead, this represents a common theme in many other secretory machineries such as the type II secretion system which is involved in the secretion of virulence factors and the biogenesis of type IV pili (Salmond, 1996), the type III secretion system which is involved in the injection of virulence factors into host cells in a wide variety of bacteria (Lee, 1997) and even the production of filamentous phage (Russel, 1995). There is very little known about how any of the proteins from these various systems cross the outer membrane although all have requisite usher‐like outer membrane proteins to serve in translocation (Kazmierczak et al., 1994; Salmond, 1996; Lee, 1997). In fact, similarities in higher order structure between these outer membrane proteins and ushers have begun to emerge (i.e. Linderoth et al., 1997; Bitter et al., 1998; Thanassi et al., 1998). Understanding the details of secretion across the outer membrane in the type 1 pilus system may reveal strategies for targeting these various machineries in the design of novel antimicrobial agents that block the assembly or secretion of virulence factors.

Materials and methods

Plasmids and strains

The E.coli strain ORN103 was described previously (Orndorff and Falkow, 1984) as were strains NU14 and NU14‐1 (Hultgren et al., 1986; Langermann et al., 1997). All restriction and DNA‐modifying enzymes were from New England Biolabs. Plasmid pAN2 was made by taking a fragment of pSH2 containing fimD and placing it into pMMB91 (Dodson et al., 1993). In order to produce His‐tagged FimD, a portion of fimD was taken from pAN2 and cloned into pBR322 (pFimDtr). Then a PCR‐generated fragment containing the coding region from the _Pvu_I site of fimD to the 3′ end of the gene, along with the addition of six histidine codons and a _Pst_I site, was cloned into the _Pvu_I and _Pst_I sites in pFimDtr. The fimD gene with the six additional histidine codons was removed and ligated into pMMB66 (Furste et al., 1986) giving pETS4. pETS1000 was made by cloning a PCR‐generated fragment containing the coding region of fimC from the _Nsi_I site to the 3′ end with the addition of six histidine codons and an _Xba_I site into _Nsi_I–_Xba_I‐cut pHJ9205 (Jones et al., 1995). pETS1 was made by cutting fimF from pSH2 and cloning the fragment into pACYC184, making pFimF. After ensuring proper orientation, fimF was cut from pFimF and ligated into pMMB91, completing pETS1. The fimG gene was isolated by cutting fimG from pSH2 and ligating the fragment into pMMB66, making pETS2A. Inducible FimA production was ensured by using PCR to generate the fimA gene with an _Eco_RI site at the 5′ end and a _Bam_HI sites at the 3′ end of the gene. The PCR fragment was cloned into those sites in pMMB66, making pETS5. pETS1001, pETS1002, pETS1003 and pETS1005 were made by cutting pHJ20, pETS2A, pETS1 or pETS5 and ligating into pETS1000. pETS1004 was made in the same way as the other pETS100N constructs except that the fragment was ligated into pHJ9205 instead of pETS1000. Correct orientation was checked by restriction enzyme digest and protein production for all. All of the PCR fragments mentioned above were sequenced to ensure fidelity. The fimD knock‐out in the plasmid‐encoded type 1 operon, pETS6, was made by digesting pSH2 with _Pvu_I, blunting with T4 DNA polymerase and partially digesting with _Eco_RI. A gene fragment corresponding to the 3′ end of fimC was cut from pHJ9205 and the fragment was ligated with the fragment of pSH2, giving pETS6 which contains all of the type 1 genes but is missing ∼2.3 kb at the 5′ end of fimD. ORN103 transformed with pETS6 and pMMB66 could not hemagglutinate guinea pig red blood cells, while cells with pETS6/pETS4 gave HA titers that were equal to that of ORN103 containing pSH2 (1:128 compared with a titer of 0 with pETS6/pMMB66). A plasmid containing an isopropyl‐β‐D‐thiogalactopyranoside (IPTG)‐inducible fimC − operon was made using pETS5 with a fragment ligated in from pSH2 that contained fimFGH. A fragment from this new plasmid containing fimAFGH was ligated into pETS4, giving pETS12. The pETS12 plasmid in ORN103 gave no HA titer but could be complemented with either pETS1000 or pHJ9205 to give HA titers that were equivalent to pSH2.

Isolation and purification of FimD and FimC–subunit complexes

ORN103 cells transformed with pETS4 were grown at 37°C with aeration to an OD600 of ∼0.6, and FimD production was induced with 10 μM IPTG for 1 h. The cells were harvested by centrifugation, washed once with 20 mM HEPES (pH 7.5), passed twice through a French Press (American Instrument Co.), and the membrane fraction collected by centrifugation at 100 000 g for 1 h. The membrane fraction was solubilized by resuspending in 20 mM HEPES (pH 7.5) and Sarkosyl solution at room temperature, followed by centrifugation at 100 000 g for 1 h. This resulted in essentially only outer membrane components in the pellet. For Ni2+ chromatography purification, the outer membrane was solubilized by rocking at room temperature in a 1.0% Elugent (Calbiochem)/20 mM HEPES (pH 7.5) solution for 1 h followed by centrifugation at 100 000 g for 1 h. FimD was primarily in the soluble fraction which was passed over a HiTrap Column (Pharmacia) charged with Ni2+ following the manufacturer's protocols. FimD was eluted in 20 mM HEPES (pH 7.5), 20 mM EDTA, 0.1% Elugent, 500 mM NaCl and desalted in a Centricon 30 (Amicon) to 20 mM HEPES (pH 7.5), 0.1% Elugent, 75 mM NaCl. Variations on the purification included using Tris instead of HEPES and Zwittergent 3‐14 (Calbiochem) instead of Elugent.

Combinations of the different plasmids (pETS1, pETS2A, pETS5, pHJ20) and pETS1000 were transformed into ORN103 and protein production induced at OD600 ∼0.6 by adding 0.05% arabinose (for FimC) and 500 μM IPTG (for the subunits) for 1 h at 37°C. Cells were harvested by centrifugation and periplasmic preparations were done following published protocols (Jones et al., 1993). The various chaperone–subunit complexes were purified by Ni2+ chromatography using Ni‐NTA–agarose beads (Qiagen) following the manufacturer's protocols with the following exceptions: 20 mM Tris (pH 8.0) was used to wash beads, and proteins were eluted in 20 mM Tris (pH 7.5) and 0.1 M EDTA. FimCA and FimCG complexes were dialyzed into 20 mM Tris (pH 8.0), and excess FimC was removed on a Mono‐Q column using FPLC (Pharmacia). FimCH and FimCF complexes were dialyzed into 20 mM KMES (pH 6.5), and excess FimC was removed on a Mono‐S column using FPLC (Pharmacia). All complexes were then placed into 10 mM HEPES (pH 7.5), 150 mM NaCl by buffer exchange in a Centricon 10 (Amicon). Additional FimC, FimCH, PapD, PapDG and PapDAG150T complexes were purified using protocols described previously (Jones et al., 1993; Bullitt et al., 1996). FPLC gel filtration chromatography was done on the FimD–C–H complex using a Superdex 200 column (Pharmacia) under the following buffer conditions: 20 mM Tris (pH 8.0), 100 mM NaCl, 0.5% Elugent.

Electrophoresis and Western blotting

SDS–PAGE was done using standard laboratory protocols. For Western (immuno) blotting, proteins were transferred to PVDF (Biorad) or nitrocellulose (Schleicher & Schuel) and blocked overnight in TBS (150 mM NaCl, 10 mM Tris–HCl, pH 7.5)/3% bovine serum albumin (BSA). Blots were washed three times with TBST (500 mM NaCl, 0.05% Tween‐20, 20 mM Tris–HCl, pH 7.5), incubated with the appropriate antibody (anti‐FimD and anti‐FimCH were graciously provided by Medimmune Inc.) for 45 min, washed three times for 5 min with TBST, incubated with the secondary antibody [goat α‐mouse or rabbit IgG‐alkaline phoshatase (AP) conjugate (Pierce)], washed three times with TBST, and once with developer buffer (100 mM NaCl, 5 mM MgCl, 100 mM Tris–HCl, pH 9.5). The blots were developed in developer buffer with 0.04% NBT and 0.02% BCIP (Sigma).

ELISA

Briefly, purified FimD was allowed to adhere to microtiter plate wells overnight at 4°C in phosphate‐buffered saline (PBS). Following washing with PBS and a 2 h blocking procedure, the various combinations of purified FimC–subunit complexes (or FimC alone) were incubated at room temperature in the wells for 1 h. After washing, a 1:1000 dilution of anti‐FimC antibody was incubated in the wells for 45min. Following another washing, anti‐rabbit IgG–AP conjugate (Sigma) was incubated for 45 min. Finally, after additional washing, _p_‐nitrophenol phosphate was added. The OD405 was read after 1 h on a microtiter plate reader. Background absorbance and non‐specific chaperone–subunit binding to the plate were subtracted out using negative controls of FimD alone and the various chaperone–subunit complexes alone.

Trypsin experiments

ORN103 cells were transformed with plasmids coding for FimD (pETS4), FimD together with combinations of the chaperone (FimC) and the various type 1 subunits (pETS1001, pETS1002, pETS1003, pETS1004, pETS1005) or FimC alone (pETS1000). Negative controls were included with vector only (pMMB66). FimD production from the cells growing at 37°C was induced at OD600 ∼0.6 by addition of IPTG, and chaperone or chaperone–subunit complex production was induced by addition of arabinose. After 1 h, the cells were washed once with 20 mM HEPES (pH 7.5) and resuspended in 20 mM HEPES (pH 8.5). The resuspended cells were split into two equal portions. One half was digested with fresh 100 μg/ml trypsin (Sigma), and the other half mock digested (only the trypsin buffer was added). All were rocked at 37°C for 2 h. Then 10 mM phenylmethylsulfonyl fluoride (PMSF) was added, followed by harvesting by centrifugation, resuspending in HEPES and disruption by French Press. The outer membranes were isolated either by step sucrose gradient centrifugation (Nikaido, 1994) or by differential solublization (as described above for FimD purification). Outer membranes were isolated identically when using any of the type 1 pilus‐producing strains; however, the cells were grown in static broth with passages as described previously (Jones et al., 1995) until the bacteria gave an HA titer of 1:128. Protein sequencing was done by David McCourt at Midwest Analytical.

Biacore experiments

The Biacore 2000 machine, CM5 or NTA sensor chips and analysis programs were all from Pharmacia Biosensor. The detection principle for the Biacore is based on surface plasmon resonance (Karlsson et al., 1991). One of the proteins (the ligand) is immobilized on the surface of the analysis chip (Johnsson et al., 1991) and the other component (the analyte) is passed over this surface in a continuous flow. Binding of the analyte to the ligand results in a refractive index change near the surface, causing a change in the angle of light that is reflected off the gold‐linked dextran matrix upon which the ligand is immobilized. This change is proportional to the increase in mass and is converted to response units (RU) and plotted on an axis versus time in a graph called a sensorgram (Fagerstam et al., 1992) (see Figure 4A).

A sensorgram usually consists of an association phase when the analyte is first injected over the ligand surface, and a dissociation phase when the analyte is dissociating from the ligand and only buffer is flowing over the surface. The buffer used for the binding studies was 10 mM HEPES pH 7.5, 150 mM NaCl, 50 μM EDTA, 0.5% Elugent (Calbiochem). The kinetic association rate constant (_k_on) and kinetic dissociation rate constant (_k_off) can be calculated by direct analysis of the association and dissociation phase respectively, or by linear regression analysis. In order to minimize model fit or constant calculation errors due to analyte exhaustion (important in _k_on calculations) or substrate rebinding (important during _k_off calculations), the data actually used for determination of constants only represented 40 s—starting from 12 s following initiation of association/dissociation and continuing to the 52 s point. A lag of 12 s was used to allow detector stabilization following the bulk refractive index changes that occur with the injection of analyte (association) or buffer only (dissociation). All of the kinetic analyses assumed a pseudo‐first order interaction between the immobilized ligand and the analyte. Model fits were superior using linear, single‐site interaction models instead of non‐linear, multi‐site interaction methods of data evaluation based on χ2, standard error and residual values, and baseline offsets (as discussed in BIA Evaluation materials). Measured kinetic constant values were well within the range considered accurate for the Biacore 2000 machine and the BIA Evaluation software.

The _K_D standard error values were calculated by averaging the standard errors from all of the _k_on and _k_off values for a particular chaperone–subunit complex and performing a division of standard errors calculation. Results in Figure 4A–D are from experiments on a CM5 chip; however, identical experiments done using an NTA chip (which immobilized FimD or PapC via the His tag) and non‐His‐tagged FimC, FimCH, PapD and PapDG gave essentially identical results (data not shown). All of the calculations and graphic analyses were done using the equations provided in the BIA Evaluation software package.

References

Download references