Structural Basis of Chaperone Function and Pilus Biogenesis

See allHide authors and affiliations

Science  13 Aug 1999:
Vol. 285, Issue 5430, pp. 1058-1061
DOI: 10.1126/science.285.5430.1058


Many Gram-negative pathogens assemble architecturally and functionally diverse adhesive pili on their surfaces by the chaperone-usher pathway. Immunoglobulin-like periplasmic chaperones escort pilus subunits to the usher, a large protein complex that facilitates the translocation and assembly of subunits across the outer membrane. The crystal structure of the PapD-PapK chaperone-subunit complex, determined at 2.4 angstrom resolution, reveals that the chaperone functions by donating its G1 β strand to complete the immunoglobulin-like fold of the subunit via a mechanism termed donor strand complementation. The structure of the PapD-PapK complex also suggests that during pilus biogenesis, every subunit completes the immunoglobulin-like fold of its neighboring subunit via a mechanism termed donor strand exchange.

Many pathogenic Gram-negative bacteria assemble hair-like adhesive pili on their surfaces that mediate microbial attachment by binding to receptors present in host tissues (1). P pili are adhesive organelles encoded by 11 genes (papA through papK) in thepap (pilus associated with pyelonephritis) gene cluster found on the chromosome of uropathogenic strains of Escherichia coli (2, 3). P pili are composite fibers consisting of a rod, 7 nm in thickness, joined to a thinner tip fibrillum (4). The rod consists of PapA subunits arranged in a right-handed helical cylinder with 3.28 subunits per turn (5,6). The fibrillum primarily consists of repeating PapE subunits arranged in an open helical configuration. The PapG adhesin is joined to the distal end of the tip fibrillum via the PapF adaptor subunit, and the fibrillum is joined to the rod via the PapK adaptor subunit (7). PapG is a critical virulence factor in pyelonephritic strains of E. coli (8).It binds specifically to the Galα(1-4)Gal digalactoside receptor found in the globoseries of glycolipids of the human kidney (9,10).

The biogenesis of P pili occurs via the highly conserved chaperone-usher pathway (11, 12). PapD is the prototypical periplasmic chaperone in a family that includes more than 30 members (13). It is a boomerang-shaped two-domain molecule with each domain having an immunoglobulin-like (Ig) fold (14). PapD has three primary functions: (i) It binds and caps interactive surfaces on pilus subunits to prevent them from participating in nonproductive interactions (15–19); (ii) it facilitates the import of subunits into the periplasm (17); and (iii) it may facilitate subunit folding (19). Chaperone-subunit complexes are targeted to the PapC outer membrane usher, which is an oligomeric pore-forming complex that facilitates chaperone uncapping and the subsequent translocation and assembly of subunits across the outer membrane (20, 21). To gain further insight into the processes of subunit capping and assembly in the chaperone-usher pathway of pilus biogenesis, we solved the structure of a complex of the subunit PapK bound to the chaperone PapD (Fig. 1A and Table 1) (22, 23).

Figure 1

(A) Density from a map calculated using MAD solvent-flattened phases. Atoms in PapK are yellow for carbon, red for oxygen, and blue for nitrogen; PapD is in green (28). (B) Stereo ribbon diagram of the PapD-PapK complex (29). PapK and PapD are in cyan and green, respectively. Secondary structures of PapK and PapD are labeled in magenta and black, respectively. Only secondary structures of PapD mentioned in the text are labeled. Subscripts 1 and 2 refer to domains 1 and 2 of PapD, respectively.

Table 1

Data collection and refinement statistics (22,23).

View this table:

PapK has the same overall variable-region Ig fold (24) as the NH2-terminal domain (domain 1) of PapD (Figs. 1B, 2A, and 2B). However, the Ig fold of PapK is incomplete: It lacks the COOH-terminal seventh strand, G, which in canonical Ig folds runs antiparallel to strand F and contributes to the hydrophobic core of the protein. In the PapD-PapK complex, this missing strand is provided by PapD, which donates its G1 β strand to complete the Ig fold of PapK (Figs. 1B and 2B), a mechanism we call donor strand complementation. However, an atypical Ig fold is produced because the donated strand runs parallel rather than antiparallel to strand F in PapK. Similar observations are made in the crystal structure of the FimC-FimH complex (25). The first eight NH2-terminal residues of PapK are disordered. The Ig fold of PapK (Figs. 1B and 2B) begins with β strand A, which is divided into strands A1 and A2 by a 310 helical turn. Strand A1 makes antiparallel β-sheet interactions with strand B while strand A2 makes antiparallel β-sheet interactions with the G1 strand of the chaperone. Strand B forms the edge of one of the two β sheets in the β sandwich of the Ig fold and runs antiparallel to strand E. Following strand B, the structure crosses over to the other side of the β sandwich through a short 310 helix to form strand C1, which runs antiparallel to strand F. The COOH-terminus of strand C1 deviates from the β-sheet arrangement to form a β meander (strands C′ and C"), which eventually returns, as C2, to make main-chain hydrogen bonds with strand F. Strand D constitutes an edge of the D, E, B, A1 β sheet and runs antiparallel to strand E. However, strand D is divided in the middle by an insertion that meanders toward strands C′ and C". Strand E is followed by a three-turn helix (αD) and a long loop structure that connects it to the COOH-terminal strand F. Finally, strand F forms a parallel β sheet with strand G1 of PapD. Hence, strand G1 of PapD is an integral part of the C, F, A2 β sheet of PapK.

Figure 2

(A) Sequence alignment of P pilus subunits (28). The secondary structural elements of PapK are indicated above the aligned sequences, with β strands and α or 310 helices in green and blue boxes, respectively. Residue numbers of PapK are indicated above the PapK sequence. Residues of PapK involved in contact with domain 1 or domain 2 of PapD are boxed in red and dark blue, respectively. Residues strictly conserved among pilus subunits are shown in pink. Conserved polar residues and conserved nonpolar residues are in light blue and yellow, respectively. (B) Topology of PapK. The nomenclature of secondary structures is according to (24). β strands are indicated as arrows; helices (either α or 310) are shown as cylinders. The G1 β strand of PapD is in pink.

The structure of apo-PapD has been described (14). The binding of PapK significantly alters the structure of PapD only in the F1-G1 loop [root mean square (rms) deviation in Cα atom positions, excluding the F1-G1 loop, of 0.65 Å]. The tip of this loop undergoes a flap motion of about 11 Å, such that residues 101 to 105 of PapD become part of the G1 β strand that inserts into the Ig fold of PapK (Fig. 3A). Two distinct sites on PapK, K1 and K2, interact with two corresponding sites on PapD, D1 and D2 (Fig. 3B). The K2-D2 interface (910 Å2) is less extensive than the K1-D1 interface (2524 Å2). Site K2 is formed primarily by residues in helix 310C and the COOH-terminal Arg157 side chain of PapK (Fig. 1B). Residues in site K2 interact with residues in the C2 and D2 strands and with the F2-G2 loop of domain 2 of PapD (site D2). Site K1 contains a deep groove that runs the length of the subunit and whose base consists of the hydrophobic core of PapK (Fig. 3C). This groove is the result of the missing G β strand in the Ig fold of PapK. Site D1 includes residues 101 to 112 of the G1 β strand of PapD, which insert into the K1 groove. The main-chain atoms of the G1 strand make β-sheet interactions with strands F and A2 of PapK on either side of the K1 groove; the alternating hydrophobic side chains of the G1 strand (Leu103, Ile105, and Leu107) interact with the hydrophobic base of the groove (Figs. 2B and 3C). Pilus subunits, in the absence of the chaperone, are unstable (26); the structure of PapK reveals that this instability is due to the exposure of the hydrophobic core of the subunit to the aqueous milieu of the periplasm. Thus, the chaperone, by donating its G1 β strand to complete the Ig fold of the subunit, shields the hydrophobic core of the subunit and contributes to its stabilization. The PapD-PapK structure also reveals how the chaperone prevents premature subunit-subunit assembly in the periplasm. Biochemical and genetic data have demonstrated that strand F residues participate in subunit-subunit interactions necessary for pilus assembly (16,19). In the structure presented here, strand F lines one edge of the K1 groove of the subunit, but its residues interact with (are “capped” by) the G1 strand residues of the chaperone. Thus, the insertion of the G1 β strand of the chaperone into the K1 groove prevents strand F residues from participating in interactions with other subunits.

Figure 3

(A) Superposition of the structures of apo-PapD (purple) and PapD complexed to PapK (green) (29). PapK is in cyan. The arrow indicates the conformational change in the F1-G1 loop upon subunit binding. (B) Definition of the binding sites in PapD and PapK. On the left, PapD is shown as a space-filling model and PapK as a ribbon. On the right, PapK is shown as a space-filling model and PapD as a ribbon. The binding sites defined in the text are labeled. (C) The G1 β strand of PapD as it inserts into the groove of PapK. The PapD G1 strand is represented as a stick model and PapK as a molecular surface [program GRASP (30) with color coding according to charge, blue for positive and red for negative]. PapD and PapK residues are labeled in yellow and cyan, respectively (28).

Finally, the structure of the PapD-PapK complex suggests a mechanism for pilus biogenesis at the outer membrane usher. In addition to strand F, a highly conserved motif at the NH2-terminus of the subunits has also been shown to participate in subunit-subunit interactions (19). This motif, located upstream of strand A1 and disordered in the structure, is highly homologous to the G1 strand of the PapD chaperone. Both regions consist of an alternating arrangement of hydrophobic residues: Leu103, Ile105, and Leu107 in the G1 strand of PapD, and Val11, Phe13, Gly15, and Val17 in the NH2-terminal region of PapA, for example (Fig. 2A). Thus, this region could fit into the groove of a neighboring subunit during pilus biogenesis. Because this NH2-terminal motif apparently protrudes away from the main body of the chaperone-subunit complex, it would be free to exchange with the G1 strand of the chaperone by a mechanism we term donor strand exchange. Thus, the Ig fold of every subunit would be complemented by the NH2-terminal motif of its neighboring subunit. Donor strand exchange differs from domain swapping, a well-characterized mechanism of protein assembly. Domain swapping implies that the swapped domains make identical interactions in both the monomeric and the oligomeric forms and that the monomeric form of the protein is stable (27). However, pilus subunits are unstable as monomers and can only exist either bound to a chaperone or as part of the pilus (26).

We used the PapK structure to model a PapA pilus rod. The disordered NH2-terminal motif of each pilus subunit was built as a β strand and inserted into the groove of each preceding subunit. Insertion parallel to strand F yielded a rod with a star-shaped cross section, inconsistent with electron microscopy data. Insertion antiparallel to strand F produced a pilus with a helical symmetry having dimensions similar to those experimentally observed (5,6) (Fig. 4). Thus, donor strand complementation with the chaperone results in an atypical Ig fold, whereas donor strand exchange between subunits produces a canonical variable-region Ig fold in the mature pilus (24). Stereochemical complementarity between the NH2-terminal motifs and grooves of the various subunits most likely restricts the order of subunit assembly. Thus, the molecular basis for the adaptor function of PapK may in part be a consequence of its NH2-terminal motif fitting the groove of PapE and its groove accommodating the NH2-terminal motif of PapA with stereochemical specificity.

Figure 4

(A) Molecular surface of a pilus rod [program GRASP (30)]. The disordered residues at the NH2-terminus of the subunit were modeled as a β strand that inserts into the groove of the preceding subunit antiparallel to strand F. (B) Stereo ribbon diagram of the rod model.

  • * These authors contributed equally to this report.


View Abstract

Navigate This Article