Molecular mechanism of P pilus termination in uropathogenic Escherichia coli

Abstract

P pili are important adhesive fibres that are assembled by the conserved chaperone–usher pathway. During pilus assembly, the subunits are incorporated into the growing fibre by the donor-strand exchange mechanism, whereby the β-strand of the chaperone, which complements the incomplete immunoglobulin fold of each subunit, is displaced by the amino-terminal extension of an incoming subunit in a zip-in-zip-out exchange process that is initiated at the P5 pocket, an exposed hydrophobic pocket in the groove of the subunit. In vivo, termination of P pilus growth requires a specialized subunit, PapH. Here, we show that PapH is incorporated at the base of the growing pilus, where it is unable to undergo donor-strand exchange. This inability is not due to a stronger PapD–PapH interaction, but to a lack of a P5 initiator pocket in the PapH structure, suggesting that PapH terminates pilus growth because it is lacking the initiation point by which donor-strand exchange proceeds.

Introduction

Gram-negative bacteria assemble adhesive fibres that mediate the initial step of various diseases (Thanassi et al, 1998a). The prototypical P pili are specifically required for uropathogenic Escherichia coli to bind to Gal-α(1–4)-Gal moieties in human kidney cells (Hoschutzky et al, 1989; Roberts et al, 1994; Dodson et al, 2001) and cause pyelonephritis (Roberts et al, 1994). P pili are encoded by the pap gene cluster and are assembled by means of the highly conserved chaperone–usher pathway, involving a periplasmic immunoglobulin-like chaperone and an outer membrane usher (Thanassi et al, 1998b; Soto & Hultgren, 1999). Six structural subunits make up a composite pilin fibre with a short-tip fibrillum joined to a rigid helical rod (Gong & Makowski, 1992). The crystal structure of the pilin protein PapK in complex with the chaperone PapD showed that the single domain of PapK adopts an immunoglobulin-like fold but lacks the seventh, carboxy-terminal G β-strand, producing a large hydrophobic groove on the side of the protein (Sauer et al, 1999). In a process termed ‘donor-strand complementation', strand G of the amino-terminal domain of the chaperone (β-strand G₁) inserts a conserved motif of four alternating hydrophobic residues (termed ‘P1–P4 residues') into four binding pockets in the hydrophobic groove of the pilus subunits (termed ‘P1–P4 binding pockets'). The G₁ strand makes main-chain/main-chain hydrogen-bonding interactions with strand F, and thus provides in trans the β-strand that is lacking in the pilus subunit, thereby completing its immunoglobulin fold (Choudhury et al, 1999; Sauer et al, 1999).

Pilus subunits are assembled in a hierarchical manner at the usher PapC, which is a large β-barrel protein that acts as a pore across the outer membrane (Henderson et al, 2004). The assembly involves the non-covalent polymerization of pilin subunits by a process called ‘donor-strand exchange' (DSE). All of the pilin subunits, apart from adhesin, have an extra N-terminal sequence that extends away from the subunit structure in a disordered state. This sequence, referred to as the N-terminal extension (Nte), comprises a highly conserved array of alternating hydrophobic residues, termed ‘P2–P5 residues' (Soto et al, 1998; Sauer et al, 1999). As chaperone–subunit complexes are differentially targeted to the usher (Dodson et al, 1993; Saulino et al, 1998), each subunit donates its Nte to complete the immunoglobulin fold of the preceding subunit, inserting its P2–P5 residues into the corresponding P2–P5 binding pockets in the groove of the pilus subunit (Jacob-Dubuisson et al, 1994; Sauer et al, 2002; Zavialov et al, 2003; Vetsch et al, 2006). Recently, a detailed mechanism was established for DSE both in vitro and in vivo. By using X-ray crystallography and real-time electrospray ionization-mass spectrometry, it was shown that DSE involves the formation of a transient ternary chaperone–subunit–subunit complex (Remaut et al, 2006). Ternary complex formation is critically dependent on the interaction of the P5 pocket of the subunit with the P5 residue of the Nte of the incoming subunit. After the ternary complex is formed, the DSE reaction proceeds through a β-strand displacement (or zip-in-zip-out) mechanism by which the P4, P3, P2 and P1 pockets are progressively occupied as the strand of the chaperone is displaced.

PapH has the ability to terminate pilus assembly (Baga et al, 1987). An in-frame deletion in the papH gene produces P pili that are very long and free from the cell (Baga et al, 1987), whereas overexpression of papH causes P pili that are short and bound to the cell. This has produced a model by which incorporation of PapH into the growing pilus both terminates growth and anchors the pilus to the cell membrane, making PapH the last subunit to be incorporated into the fibre. The molecular basis for the ability of PapH to terminate pilus assembly is however unclear. In this report, we investigate the molecular basis of pilus termination by PapH.

Results And Discussion

Incorporation and localization of PapH in the pilus

PapH is similar to other pilus subunits in sequence (Normark et al, 1986; Baga et al, 1987) but differs by having a longer Nte (Fig 1A). The first 36 amino acids of PapH comprise its Nte and consist of two regions—a proline-rich region and a motif of alternating hydrophobic residues (Ala 25, Phe 27, Gly 29, Val 31), presumably responsible for participating in DSE. The proline-rich region has been suggested to be important for membrane association (Baga et al, 1987); however, a number of other mechanisms could be predicted, as it is unknown whether PapH is localized to the outer membrane or is even incorporated into the pilus structure.

Figure 1.

PapH localizes to the outer membrane and pilus base. (A) The N-terminal extension (Nte) of PapH consists of Gly 1 through Cys 36. Deleting residues Pro 2–Asp 22 produced the N-terminal deleted PapH (PapH_Ntd1). A haemagglutinin (HA) epitope tag was inserted, replacing residues 14–22 (H-HA). (B,C) Western blots developed with anti-HA antiserum. (B) PapH-HA in periplasm extract (Per, lane 1) and purified pili (Pili; lane 2) when expressed with ΔpapH pap operon. (C) PapH-HA in outer membrane (OM) and cytoplasm/periplasm (Cyt) fractions when expressed with PapD and PapC only (DC/H-HA; lane 1) or with ΔpapH pap operon (ΔH/H-HA; lane 2). DSE, donor-strand exchange; Ntd1, N-terminal deleted.

Thus, we first set out to localize the function of PapH. We started by deleting the entire proline-rich region (Nte1) of PapH, producing the N-terminal deleted PapH (PapH_Ntd1; Fig 1A). This resulted in P pili that had similar length and distribution as in wild-type cells (supplementary Fig 1 online). The proline-rich region is therefore not responsible for the anchoring or truncation abilities of PapH. As there was no disruption in function, we chose to replace a portion of Nte1 with a haemagglutinin (HA) epitope tag (PapH-HA; Fig 1A). Bacterial cells expressing a ΔpapH pap operon were complemented with a plasmid expressing papH-HA with the chaperone papD (pDH-HA). Negative-stain electron microscopy showed good complementation by the PapH-HA construct, with the cells expressing pili levels similar to those in wild-type cells (supplementary Fig 1 online). PapH-HA was stable in the periplasm of these cells (Fig 1B, lane 1) and the pili purified by heat purification showed incorporation of PapH-HA into the pilus structure (Fig 1B, lane 2). If PapH were the last subunit incorporated into the growing pilus, then it should be localized to the outer membrane and dependent on the other pili subunits for this localization. To test this, bacterial cells expressing the ΔpapH pap operon or only the usher papC were complemented with pDH-HA. When expressed with only the chaperone and usher, PapH-HA was expressed stably but failed to localize to the outer membrane (Fig 1C, lane 1). However, when expressed with the entire pap operon, PapH–HA was both stable and localized to the outer membrane (Fig 1C, lane 2). Loading controls for the outer membrane fractions were analysed and confirmed by using SDS–polyacrylamide gel electrophoresis (SDS–PAGE; data not shown). These results show that PapH is incorporated into the pilus structure, is localized to the outer membrane and that this localization is dependent on the other pilin subunits. This strengthens the model of PapH being the last subunit incorporated into the growing pilus, whereby its incorporation into the pilus base truncates and anchors pili, leaving PapH localized to the outer membrane.

PapH is unable to undergo DSE in vitro

To provide an explanation for the termination of pilus biogenesis on incorporation of PapH, we suggested that a mechanism exists that prevents PapH from undergoing DSE with any other subunits. To test the hypothesis, a new biochemical in vitro assembly assay was set up to monitor DSE. PapD_His–PapA or PapD_His–PapH complexes—in which PapD is tagged at the carboxyl terminus with a His₆ tag—were incubated with increasing amounts of peptides derived from the Nte of either PapA (Nte_PapA; residues Ala 1–Asp 19) or PapH (Nte2_PapH; residues Gly 23–Arg 33)—both of which encompass their respective DSE regions. Subunit–peptide product formation was monitored after overnight incubation at 24°C by using a nickel-spin column, which retains the chaperone–subunit complex and the free chaperone. SDS–PAGE was then carried out to analyse the subunit–peptide product released in the flow-through samples. As some chaperone–subunit complexes have been shown to undergo spontaneous DSE-mediated polymerization (PapA) or nonspecific aggregation (PapH), and as these would compete with the process that we wished to monitor, non-aggregating mutants of PapA and PapH were used. To generate a non-polymerizing mutant of PapA, the Nte1 sequence preceding the DSE motif of PapA (Pro 2–Gln 8) was removed and Gly 15 was mutated to Asn (PapA_Ntd1G15N). Likewise, a stable mutant of PapH was generated by removing the Nte1 sequence (PapH_Ntd1; see text described above).

SDS–PAGE of the flow-through samples showed that, for PapA, the exchange reaction proceeded efficiently and the amount of subunit–peptide complexes released increased with the increase in Nte_PapA or Nte2_PapH peptide concentrations (Fig 2, left panel). Thus, we found that PapA undergoes DSE with the Nte of either PapA or PapH, forming Nte_PapA–PapA_Ntd1G15N or Nte2_PapH–PapA_Ntd1G15N products. However, the situation was very different when the PapD_His–PapH_Ntd1 complex was challenged by the Nte_PapA or Nte2_PapH peptides (Fig 2, right panel; only Nte_PapA is shown, as the results are identical to those of Nte2_PapH). No exchange was detected on the gel (only trace amounts with a 50:1 ratio). Similarly, in vitro DSE using the full-length PapH protein yielded no product formation (results not shown). Thus, in contrast with PapA, PapH is unable to undergo DSE.

Figure 2.

In vitro donor-strand exchange experiments. This figure shows the results of polyacrylamide gel electrophoresis analysis of flow-through samples from the Nickel-NTA spin columns after the complexes were incubated overnight at 24°C in the presence of Nte_Pap peptides. Left and middle left: increasing molar ratios of 1, 10, 25 and 50 (ratio of peptide to complex) of first Nte_PapA peptides, and second Nte2_PapH peptides, added to the PapD_His–PapA_Ntd1G15N complex. Middle right: increasing molar ratios of 1, 10, 25 and 50 (ratio of peptide to complex) of Nte_PapA peptide added to the PapD_His–PapH_Ntd1 complex. Similar results were obtained by using Nte2_PapH (not shown). MW indicates molecular weight markers (6, 14.4, 21.5 and 31.0 kDa from bottom to top). Right: the PapD_His–PapA_Ntd1G15N complex, the molecular weight markers (MW) and the PapD_His–PapH_Ntd1 complex, with no peptide added, eluted from the spin column.

Stability of the PapD–PapH complex

One possible reason for PapH not undergoing DSE might be that the interaction between PapD and PapH is stronger compared with the other subunits, and PapD cannot be uncapped by the Nte of any subunit when in complex with PapH. However, affinities between chaperone and subunits cannot be measured, as subunits are not stable on their own. Instead, the strength of the chaperone–subunit interaction is typically investigated by using urea (Barnhart et al, 2000). For these urea denaturation experiments, various subunit–chaperone complexes were incubated with increasing amounts of urea (0–8 M at 0.5 M intervals at pH 7.5) and analysed. The results (Fig 3) show that the PapD–PapH interaction is comparable with other subunits, except PapK which is stable even at 8 M urea. Thus, the inability of PapH to undergo DSE is not due to a stronger interaction with PapD.

Figure 3.

Urea denaturation experiments. The various complexes indicated above the four panels were incubated with increasing amounts of urea (0–8 M); eluates, after loading onto Ni-NTA spin columns, were analysed by using polyacrylamide gel electrophoresis. Only the range that leads to complete complex dissociation is shown (except for the PapD–PapK complex, which is stable even at 8.0 M urea). MW indicates molecular weight markers (6, 14.4, 21.5 and 31.0 kDa from bottom to top).

PapD–PapH_Ntd1 structure shows obstructed P5 pocket

PapH is the only pilin that does not undergo DSE, but shown earlier, its inability to undergo DSE is not due to an inherently more stable chaperone–subunit complex. Therefore, could PapH lack a hydrophobic groove or have crucial elements that would prevent DSE? To address this question, we determined the structure of the PapD–PapH_Ntd1 complex. Fig 4A is an overall representation of the structure of PapD–PapH_Ntd1, which is very similar to the already known structures of PapD–PapK or PapD–PapE_Ntd (Sauer et al, 1999, 2002). Similar to PapK and PapE, PapH lacks the G strand of its immunoglobulin fold and PapD complements this by donating its G1 strand. The groove at position P1–P4 is very similar to that of PapK and PapE (Fig 4–C). However, PapK and PapE have a clearly defined P5 pocket, whereas PapH has none (Fig 4B). In PapH, the P5 pocket is blocked by residue Thr 52 (Fig 4C). Although conserved in PapK (Thr 29; Fig 4C), Thr 52 obstructs the P5 pocket of PapH because it is forced into the observed conformation by a kink caused by Pro 53.

Figure 4.

Crystal structure of PapD–PapH_Ntd1 and P5 pocket of PapH and PapK. (A) Overall three-dimensional structure of the PapD–PapH_Ntd1 complex. PapD (green) and PapHNtd1 (blue) are shown in ribbon representations, with β-strand indicated with arrows and α-helices indicated as cylinders. Secondary structures of PapH and the G1 strand of the chaperone are labelled. (B) Surface representations of PapH_Ntd1 (left) and PapK (right) in complex with PapD. The position of the P5 pocket is indicated. For clarity, only residues in the G1 strand of the chaperone are shown (in a stick representation, colour-coded in red, blue and white for oxygen, nitrogen and carbon atoms, respectively). (C) Details of the P5 region of PapH (cyan) and the P5 pocket of PapK (red). Residues Thr 52 and Pro 53 of PapH are indicated in a stick representation, as are residues 29 and 30 of PapK. The locations of the P4 and P5 pockets in both proteins are indicated by an orange and a green circle, respectively.

The mechanism of DSE was recently examined in greater molecular detail and showed that the P5 pocket in the groove of assembling subunits has a crucial role in initiating DSE (Remaut et al, 2006). Using the Saf pilus system of Salmonella enterica, Remaut et al showed that (i) DSE is initiated by the interaction of the P5 residue on the Nte of the incoming subunit with the P5 pocket of the subunit that is already assembled, and (ii) obstructing the P5 pocket or mutating the P5 residues inhibits this interaction, leading to a marked decrease in ternary complex formation and a resulting inhibition of the DSE reaction. Thus, PapH does not undergo DSE, probably because it is lacking a P5 pocket. We propose that the lack of a P5 pocket explains the mechanism of PapH as a regulator of pilus length.

Conclusion

Using various techniques, the molecular basis by which a seemingly normal pilin subunit has such a marked role in controlling pilus length was investigated. This study shows that PapH acquires the same C-terminally truncated immunoglobulin-like fold as other pilin subunits and relies on the PapD chaperone for donor-strand complementation. PapH is then localized to the outer membrane where it is incorporated into the pilus structure, most probably by donating its Nte to complete the fold of PapA. However, we show that PapH is unable to undergo DSE. This failure was attributed to the lack of a P5 pocket in the groove of PapH. Indeed, a previous study has identified the P5 pocket as the initiator point for DSE (Remaut et al, 2006). Thus, the P5 residue of the Nte of PapH is able to insert into the P5 pocket of the groove of PapA, whereas no similar residue in any Nte can insert into the groove of PapH. Indeed, the absence of the P5 pocket in PapH is probably the single most important reason for its role as a terminator subunit. This study firmly establishes that PapH functions through a conserved assembly mechanism in pilus biogenesis. Most pilin subunits use their P5 pocket to promote assembly into the pilus structure through DSE, whereas PapH blocks this important pocket to truncate pilus assembly and regulate pilus length. This study emphasizes the P5 pocket as a conserved assembly mechanism of pilus biogenesis in Gram-negative pathogens, which, as shown by PapH, can provide a target for disrupting the assembly of these virulence factors.

Methods

Protein expression and purification. For details, see the supplementary information online.

PapH localization. Bacterial cells expressing pDH-HA in a ΔpapH pap background (pEMM13) or in the presence of papC (pYML113) were scraped from tryptic soy agar plates (grown at 37°C for 36 h with 0.1 μM isopropyl-β-D-thiogalactoside) and suspended in 5 mM Tris and 75 mM NaCl (pH 8.0). Resuspended cells were heated at 65°C to remove bound pili (Hoschutzky et al, 1989), and the supernatants containing pili were salt precipitated (Kuehn et al, 1992). The outer membranes of the cells were purified by using ultracentrifugation and made soluble in 1% Elugent and 20 mM HEPES (pH 7.5; Saulino et al, 1998). Purified pili, outer membrane and cytoplasm/periplasm samples were heated in SDS buffer and 4 M urea at 95°C, and analysed by SDS–PAGE followed by immunoblotting with a monoclonal HA antibody (Covance Inc., Berkeley, CA, USA) as the primary antibody and alkaline phosphatase-conjugated goat anti-mouse IgG (Sigma, St Louis, MO, USA) as the secondary antibody.

In vitro donor-strand exchange experiments. For each DSE experiment, 50 μg of the complex (PapD_His–PapA_Ntd1G15N or PapD_His–PapH_Ntd1) was mixed with the Nte_PapA or Nte2_PapH peptides in a peptide to complex molar ratio of 0 (control), 1, 10, 25 and 50 in 50 mM Tris–HCl (pH 7.5) and 50 mM NaCl (buffer A). The mixtures were then incubated overnight at 24°C in a shaker (150 r.p.m.). After incubation, 100 μl (50 μg) of the mixture was loaded onto the Nickel-NTA spin columns that were prewashed with buffer A. The columns were then centrifuged for 2 min at 700g and the flow-through samples re-loaded two more times. The flow-through samples were then analysed by using SDS–PAGE.

Urea denaturation experiments. For each urea experiment, 50 μg of PapD_His–PapA_Ntd1G15N, PapD_His–PapH_Ntd1, PapD_His–PapK and PapD_His–PapE_Ntd was incubated in urea solutions ranging from 0 to 8.0 M in 50 mM Tris–HCl (pH 7.5) and 300 mM NaCl (buffer A) for 15 min at 24°C. A control mixture of each complex without urea was also prepared. A 100 μl portion of the preincubated urea–protein (50 μg of complex) mixture was then loaded on pre-equilibrated Nickel-NTA spin columns and centrifuged for 2 min at 700g. The columns were washed with 300 μl of the corresponding urea solution. The urea was removed by adding 300 μl of buffer A. The remaining complexes were eluted with a buffer containing 250 mM imidazole (buffer B) and analysed by using SDS–PAGE.

Crystallization and structure determination. For details, see the supplementary information online. Coordinates of the PapD–PapH_Ntd1 complex have been deposited to the Protein Data Bank (entry code 2j2z).

Supplementary information is available at EMBO reports online (http://www.emboreports.org).

Supplementary Material

Supplementary Information

Supplementary Figure 1