Report

X-ray Structure of the FimC-FimH Chaperone-Adhesin Complex from Uropathogenic Escherichia coli

See allHide authors and affiliations

Science  13 Aug 1999:
Vol. 285, Issue 5430, pp. 1061-1066
DOI: 10.1126/science.285.5430.1061

Abstract

Type 1 pili—adhesive fibers expressed in most members of the Enterobacteriaceae family—mediate binding to mannose receptors on host cells through the FimH adhesin. Pilus biogenesis proceeds by way of the chaperone/usher pathway. The x-ray structure of the FimC-FimH chaperone-adhesin complex from uropathogenic Escherichia coli at 2.5 angstrom resolution reveals the basis for carbohydrate recognition and for pilus assembly. The carboxyl-terminal pilin domain of FimH has an immunoglobulin-like fold, except that the seventh strand is missing, leaving part of the hydrophobic core exposed. A donor strand complementation mechanism in which the chaperone donates a strand to complete the pilin domain explains the basis for both chaperone function and pilus biogenesis.

Type 1 pili are adhesive fibers expressed in E. coli as well as in most members of the Enterobacteriaceae family (1). They are composite structures in which a short-tip fibrillar structure containing FimG and the FimH adhesin (and possibly the minor component FimF as well) are joined to a rod composed predominantly of FimA subunits (1). The FimH adhesin mediates binding to mannose oligosaccharides (2,3). In uropathogenic E. coli, this binding event has been shown to play a critical role in bladder colonization and disease (4). Type 1 pilus biogenesis proceeds by way of a highly conserved chaperone/usher pathway that is involved in the assembly of over 25 adhesive organelles in Gram-negative bacteria (5). The usher forms an oligomeric channel in the outer membrane with a pore size of ∼2.5 nm (6) and mediates subunit translocation across the outer membrane. Periplasmic chaperones consist of two immunoglobulin-like domains with a deep cleft between the two domains (7–9). Chaperones stabilize pilus subunits and prevent them from participating in premature interactions in the periplasm by forming chaperone-subunit complexes (5). Here, the x-ray crystal structure of the FimC-FimH chaperone-adhesin complex from uropathogenic E. coliis described. The structure reveals a donor strand complementation mechanism that explains the basis of both chaperone function and pilus biogenesis.

The structure of the FimC-FimH complex was solved by means of multiwavelength anomalous dispersion (MAD) data to 2.7 Å collected from selenomethionyl FimC-FimH crystals, and subsequently refined to 2.5 Å (Table 1). Eight copies of the FimC-FimH heterodimer in the C2 asymmetric unit were arranged as two sets of four molecules related by approximate 41screw axes. Electron density was excellent for one set of molecules (Fig. 1A), allowing us to trace the entire complex. For the second set of molecules, electron density was poorer but allowed for unambiguous placement of a copy of the initially traced complex.

Figure 1

(A) A typical sample of the solvent-flattened experimental electron density map (contoured at 1.0σ) with the refined model superimposed. Arg8C and Lys112C anchor the COOH-terminus of FimH (Gln279H) in the subunit binding cleft of the chaperone through hydrogen bonds to the terminal carboxylate. (B) MOLSCRIPT (24) ribbon diagram of the FimC-FimH complex. FimH is colored yellow, except for the A" (green) and F (orange) strands of the pilin domain. FimC is colored blue, except for the G1 strand, which is cyan. The FimH pilin domain and the NH2-terminal domain of FimC form a closed superbarrel with a continuous core made from conserved residues in both proteins. A ball-and-stick representation of the C-HEGA molecule bound to the lectin domain of FimH indicates the position of the carbohydrate-binding site at the tip of the domain.

Table 1

Summary of data collection and MAD structure determination. Two seleno-methionated FimC-FimH crystals (space groupC2, a = 139.1 Å, b = 139.1 Å, c = 214.5 Å, β = 90.0 Å) exhibiting strong pseudo P41212 symmetry were used to collect MAD (22) data on BM14 of the European Synchrotron Radiation Facility. Data were recorded at each of three wavelengths corresponding to the peak of the Se white line, the point of inflexion of the K absorption edge, and a remote wavelength by using a MAR charge-coupled device detector. Data were reduced with the program HKL2000 (23), with further processing and scaling using the CCP4 processing package (23). An initial solution to the Patterson function was produced in the tetragonal pseudo space group both automatically with the program SOLVE (23) and manually with the program RSPS (23), and initial phases were calculated with SHARP (23). Density modification including fourfold noncrystallographic (NCS) averaging was done with the program DM (23). A model corresponding to the two copies of the complex in the pseudo asymmetric unit was built with the program O (23). Bulk solvent correction, positional, simulated annealing, and isotropic temperature factor refinement was carried out with X-PLOR (23) and REFMAC (23) with tight NCS restraints against a 2.5 Å native data set collected at Max II/BL711 in Lund. The current R factor andR free (on 5% of the data) are 24.0 and 26.8%, respectively. The root mean square deviations from ideal bond lengths and angles are 0.016 and 3.3 Å, respectively. No residues are in disallowed regions of the Ramachandran plot.

View this table:

FimH is folded into two domains of the all-beta class connected by a short extended linker (Fig. 1B). The NH2-terminal mannose-binding lectin domain comprises residues 1H to 156H, and the COOH-terminal pilin domain, which is used to anchor the adhesin to the pilus, comprises residues 160H to 279H (Fig. 2A). The overall structure of the FimC chaperone in the complex is essentially the same as that of the free chaperone (8, 9). The pilin domain of FimH binds in the cleft of the chaperone (Fig. 1B), although there is only limited contact between FimH and the COOH-terminal domain of FimC.

Figure 2

(A) Alignment of type 1 pilin sequences to the pilin domain of FimH. The end of the lectin domain and the start of the pilin domain in FimH are indicated by black arrowheads above the sequences. Clustal W (25) was used to align the sequences, which were manually adjusted to minimize gaps (indicated by dots) in secondary structure elements. Residue 1 of FimH is residue 22 in the precursor protein (26). Residues are coded as follows: identical (red); conserved character (blue); pilin NH2-terminal residues proposed to take part in donor strand complementation in the pilus (yellow); involved in chaperone binding (27) (open circle above the residue); carbohydrate binding pocket (boxed). The NH2-terminal extensions of the pilin subunits are in one large box. Limits and nomenclature for secondary structure elements are shown below the sequence. (B) β-sheet topology diagrams of the lectin (left) and pilin (right) domains of FimH. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

The lectin domain of FimH is an 11-stranded elongated β barrel with a jelly roll–like topology (Fig. 2B). Searches of the structural database (10, 11) did not reveal any significant structural homologs of this domain. The fold starts with a short β hairpin that is not part of the jelly roll. The final (11th) strand of the domain is inserted between the 3rd and 10th strands and thus breaks the jelly-roll topology. A pocket capable of accommodating a mono-mannose unit is located at the tip of the domain, distal from the connection to the pilin domain (Fig. 1B). A molecule of cyclohexylbutanoyl-N-hydroxyethyl-d-glucamide (C-HEGA) (12) is bound in this pocket (Fig. 3A). The glucamide moiety of C-HEGA is blocked at C1 and cannot form a pyranose, but is bent to approach the pyranose conformation. The C2, C3, C4, and C6 hydroxyl groups of C-HEGA are enclosed within the pocket, whereas the C5 hydroxyl and cyclohexylbutanoyl-N-hydroxyethyl groups point out from the pocket and are solvent exposed. Residues Asp54H, Gln133H, Asn135H, Asp140H, and the NH2-terminal amino group of FimH (Fig. 3A) are hydrogen bonded to the glucamide moiety of C-HEGA. FimH from a urinary tractE. coli isolate that has a lysine instead of asparagine at position 135H produces type 1 pili but is unable to mediate mannose-sensitive hemagglutination of guinea pig erythrocytes (13). Also, a mutation at residue 136H has been reported to completely block mannose binding (14).

Figure 3

(A) Stereo view of the carbohydrate binding pocket in FimH with a molecule of C-HEGA bound. Residues Phe1H, Ile13H, Asn46H, Asp47H, Tyr48H, Ile52H, Asp54H, Gln133H, Asn135H, Tyr137H, Asn138H, Asp140H, Phe142H line the surface of the pocket at the tip of the lectin domain. Residues that take part in hydrogen bonding to the glucamide moiety of C-HEGA are labeled. (B) (Left) Surface (28) of the FimH pilin domain showing the exposed hydrophobic core. Hydrophobic residues that are buried in the complex but solvent exposed upon removal of the chaperone are highlighted in yellow. (Right) Same as left but with FimC (blue ribbon) completing the immunoglobulin-like fold of the pilin domain. (C) Close-up of donor strand complementation interactions. The G1 strand of FimC (blue) donates hydrophobic residues to the core of the FimH pilin domain (yellow). The total solvent-accessible surface area that is buried between the pilin domain and the chaperone is roughly 1700 Å2 (on each domain). Donor strand complementation accounts for ∼60% of this area.

The pilin domain of FimH has the same immunoglobulin-like topology as the NH2-terminal domain of periplasmic chaperones, except that the seventh strand of the fold is missing (Fig. 2B). Two antiparallel β sheets (strands A′BED′ and D"CF) pack against each other to form a β barrel that is similar to, but distinct from, immunoglobulin barrels. As in the chaperones, strand switching occurs at the edges of the sheets. In the chaperones, the A1 strand of the NH2-terminal domain switches between the two sheets of the barrel (15). The first strand of the pilin domain exhibits a similar switch, but owing to the lack of a seventh strand, the second half of the A strand is not involved in main-chain hydrogen bonding within the domain. The D strand of the chaperones as well as of the FimH pilin domain also switches, but in the pilin domain the switch is an eight-residue loop instead of the cis-proline bulge found in the chaperones. The C-D loop and the D′-D" connection pack against each other and close the top of the barrel. The other side of the barrel, defined by the A and F edge strands, is open. Owing to the absence of a seventh strand, a deep scar is created on the surface of the domain. Residues that would be part of the hydrophobic core of an intact, seven-stranded fold instead line a deep hydrophobic crevice on the surface of the pilin domain (Fig. 3B).

In the complex, the seventh (G1) strand from the NH2-terminal domain of the chaperone is used to complement the pilin domain by being inserted between the second half of the A strand and the F strand of the domain (Fig. 3C). The final strand (F) of FimH forms a parallel β-strand interaction with the G1 strand of FimC and has its COOH-terminal carboxylate anchored at the bottom of the chaperone cleft through hydrogen bonding with the conserved residues Arg8C and Lys112C in FimC (Fig. 1A). This interaction is critical for chaperone function (16, 17).

The G1 strand of periplasmic chaperones contains a conserved motif of solvent-exposed hydrophobic residues at positions 103, 105, and 107 in FimC (15). In the complex, these residues are used to complete the unfinished hydrophobic core of FimH (Fig. 3C). The two residues Leu103C and Leu105C are deeply buried in the crevice created in the FimH pilin domain owing to the missing seventh strand. Ile107C is somewhat closer to the domain surface but makes van der Waals contacts with residues Val163H and Phe276H. Leu103Ccontacts residues Ile181H, Val223H, Leu225H, and Ile272H. Leu105C is in contact with Ile181H, Leu183H, Leu252H, Ile272H, and Val274H. We denote this mode of binding “donor strand complementation” to emphasize the fact that the pilin domain is incomplete and that the chaperone donates its G1 strand to complete the fold. Donor strand complementation has also been observed in the recent crystal structure of the PapD-PapK complex (18).

Genetic, biochemical, and electron microscopic studies have demonstrated that residues in two conserved motifs (the COOH-terminal F strand and an NH2-terminal motif) participate in subunit-subunit interactions necessary for pilus assembly (17). An alignment of the pilin sequences, based on the FimC-FimH crystal structure, revealed that the NH2-terminal motif was part of a 10- to 20-residue NH2-terminal extension that was missing in the FimH pilin domain (Fig. 2A) and disordered in the PapD-PapK complex (18). This region contains a pattern of alternating hydrophobic residues similar to the G1 donor strand of the chaperone. On the basis of molecular modeling, the NH2-terminal extension of a subunit is predicted to be able to take the place of the G1 strand of the chaperone, and fit into the pilin groove. Thus, during pilus assembly, alternating hydrophobic side chains in the NH2-terminal extension could replace the hydrophobic side chains donated to the pilin core by the G1 strand of the chaperone, through a donor strand exchange mechanism. Thus, every subunit would complete the immunoglobulin-like fold of its neighboring subunit.

The type 1 pilus is a right-handed helix with about three subunits per turn, a diameter of ∼70 Å, a central pore of about 20 to 25 Å, and a pitch of about 24 Å (19). To obtain this structure, insertion of the NH2-terminal extension must be antiparallel to strand F, in contrast to the parallel insertion observed for the G1 strand of the chaperone. Insertion in a parallel orientation would lead to rosettelike structures. Using the FimH pilin domain as a model for FimA, we constructed a model for the type 1 pilus that fit these data (Fig. 4). Each subunit was aligned to have its cleft facing toward the center of the pilus so that the height from the top to the bottom of the domain along the helix axis was ∼25 Å. By applying a rotation of 115° and a rise per subunit of 8 Å, a hollow helical cylinder is created. The outer diameter of this cylinder as measured across Cα atoms is 70 Å, and the inner diameter is 25 Å. FimA subunits from different strains of E. coli exhibit considerable allelic variation (13). The vast majority of the variable positions are on the outside surface of the pilus model proposed above (Fig. 4), which would account for the antigenic variability of type 1 pili.

Figure 4

Model of the type 1 pilus. The NH2-terminal extension participates in donor strand complementation between subunits as described in the text. Subunits one turn apart in the helix pack against each other through the sides of the pilin barrel. Charged residues located between the hydrophobic side chains in the NH2-terminal extension point into the solution on the inside of the hollow pilus rod. (A) The proposed interaction between two consecutive FimA molecules in the type 1 pilus rod. The modeled NH2-terminal extension is colored red. (B) View of the pilus from the top. Residue positions that are subject to allelic variation (shown in blue) map to the outer surface of the pilus. (C) Side view of the pilus.

The proposed head-to-tail interaction between subunits in a pilus is reminiscent of oligomerization through 3D domain swapping (20), in the sense that a part of one protein molecule is used to complement another. However, in this case, complementation occurs not only between identical protein chains (FimA in the pilus rod), but also between homologous but distinct chains (for example, FimG, FimF, and FimH in the pilus tip). Furthermore, because individual pilin protomers do not exist as stable monomers, there is no exchange of structural units between a monomeric and an oligomeric state. Instead, a different protein, the periplasmic chaperone, is needed to keep the monomeric subunits in solution by donating a unique part of its structure (the G1 strand) to the different subunit grooves.

On the basis of the structure of the FimC-FimH complex, we propose that the class of proteins known as pilins are missing necessary steric information needed to fold into a native three-dimensional structure. The information that is missing consists of the seventh-edge strand of an immunoglobulin fold. This strand, which is necessary for folding, is donated to the hydrophobic core of the pilin by the periplasmic chaperone in a donor strand complementation mechanism. A recent formulation of Anfinsen's classic postulate stated that “The steric information necessary for newly synthesized protein chains to fold correctly within cells resides solely in the primary structure of the initial translation product” (21). Here we provide an example of a case where some of that information is not inherent in the sequence of the protein to be folded but is instead transferred from another protein—the periplasmic chaperone.

  • * To whom correspondence should be addressed. E-mail: hultgren{at}borcim.wustl.edu (S.J.H.); stefan{at}xray.bmc.uu.se(S.D.K.).

REFERENCES AND NOTES

View Abstract

Navigate This Article