The Activity of Escherichia coli Chaperone SurA Is Regulated by Conformational Changes Involving a Parvulin Domain

ABSTRACT

The periplasmic chaperone SurA is critical for the biogenesis of outer membrane proteins (OMPs) and, thus, the maintenance of membrane integrity in Escherichia coli. The activity of this modular chaperone has been attributed to a core chaperone module, with only minor importance assigned to the two SurA peptidyl-prolyl isomerase (PPIase) domains. In this work, we used synthetic phenotypes and covalent tethering to demonstrate that the activity of SurA is regulated by its PPIase domains and, furthermore, that its activity is correlated with the conformational state of the chaperone. When combined with mutations in the β-barrel assembly machine (BAM), SurA mutations resulting in deletion of the second parvulin domain (P2) inhibit OMP assembly, suggesting that P2 is involved in the regulation of SurA. The first parvulin domain (P1) potentiates this autoinhibition, as mutations that covalently tether the P1 domain to the core chaperone module severely impair OMP assembly. Furthermore, these inhibitory mutations negate the suppression of and biochemically stabilize the protein specified by a well-characterized gain-of-function mutation in P1, demonstrating that SurA cycles between distinct conformational and functional states during the OMP assembly process.

IMPORTANCE This work reveals the reversible autoinhibition of the SurA chaperone imposed by a heretofore underappreciated parvulin domain. Many β-barrel-associated outer membrane (OM) virulence factors, including the P-pilus and type I fimbriae, rely on SurA for proper assembly; thus, a mechanistic understanding of SurA function and inhibition may facilitate antibiotic intervention against Gram-negative pathogens, such as uropathogenic Escherichia coli, E. coli O157:H7, Shigella, and Salmonella. In addition, SurA is important for the assembly of critical OM biogenesis factors, such as the lipopolysaccharide (LPS) transport machine, suggesting that specific targeting of SurA may provide a useful means to subvert the OM barrier.

INTRODUCTION

The Gram-negative bacterial outer membrane (OM) is an essential, selectively permeable barrier between the cell and its external environment. This asymmetric bilayer affords Gram-negative bacteria resistance to a wide range of antimicrobial compounds and membrane perturbants due to strong lateral interactions between lipopolysaccharides (LPS) in the outer leaflet (1, 2). In order to maintain selective permeability, the OM is replete with integral β-barrel outer membrane proteins (OMPs), some of which serve as pores for nutrient diffusion. The transport and assembly of these β-barrel proteins comprise a complicated multistep process, from their secretion through the inner membrane translocase (Sec) machinery to transport across the aqueous periplasm and assembly into the OM by the β-barrel assembly machine (BAM) complex (3).

Prior to their assembly in the OM by BAM, these aggregation-prone, insoluble β-barrel substrates must be maintained in a folding-competent state as they traverse the aqueous, oxidizing periplasm. This maintenance is performed by a partially redundant network of chaperone proteins, proteases, and folding factors (4, 5). SurA has long been recognized as the primary OMP chaperone in Escherichia coli due to its significant contribution to the assembly of OMPs involved in processes ranging from virulence and membrane integrity to cellular viability (6–9). This importance is underscored by the fact that LptD, an essential OMP responsible for inserting LPS into the outer leaflet of the OM, is largely dependent on SurA for its assembly (8, 10). SurA is also thought to be the terminal chaperone in the assembly process, as it is the only chaperone known to interact with the periplasmic polypeptide-transport-associated (POTRA) domains of the BAM insertase BamA and, as the deletion of surA affects OMP maturation in the same manner as the loss of BamB, one of the four BamA-associated lipoprotein members of the BAM complex (4, 11, 12). Despite its central role in OMP assembly, SurA is not essential, because a redundant network of other chaperones, notably Skp, FkpA, and the chaperone/protease DegP, is sufficient to maintain viability in the absence of SurA (13, 14).

The functions of several periplasmic chaperones were better elucidated when it was discovered that their three-dimensional structures resemble those of well-known cytoplasmic chaperones and folding factors. The trimeric jellyfish structure of the functional Skp chaperone, for example, resembles that of cytosolic prefoldin, and Skp sequesters substrates in a manner mechanistically similar to that of this well-characterized chaperone (15, 16). SurA, on the other hand, resembles the cytosolic chaperone trigger factor, a ribosome-associated protein that contains a core chaperone domain, an FKBP-like peptidyl-prolyl isomerase (PPIase) domain, and a ribosome binding domain (17). SurA, like trigger factor, contains a core chaperone module (comprising the N and C termini) and an active parvulin-like PPIase domain (referred to here as P2) that is separated from the core by a linker (Fig. 1). Unlike trigger factor, SurA also contains an additional, inactive PPIase domain (P1) that, in the published structure, forms extensive noncovalent contacts with the core chaperone module, partially occluding the molecular cradle that is important for SurA and trigger factor chaperone function (17–19).

FIG 1.

SurA has a modular architecture. The structure of full-length SurA (PDB accession no. 1M5Y) is shown in surface (left) and cartoon (right) forms. Amino acid numbers reference the nonprocessed peptide sequence.

In spite of this unique domain architecture, PPIase activity is dispensable for SurA function (20, 21). In fact, a mutant lacking both PPIase domains is capable of complementing a ΔsurA mutation in vivo and is sufficient for chaperone function in vitro, suggesting that the SurA parvulin domains do not contribute significantly to SurA chaperone activity (20). Many proteobacterial SurA homologs contain at least one parvulin domain, although some harbor only the core chaperone module, further suggesting that the parvulin domains may only be necessary in specific circumstances or systems (22, 23).

Despite the apparent dispensability of PPIase domains for SurA chaperone activity, recent genetic and biochemical evidence has revealed a role for the enzymatically inactive parvulin domain (P1) in OMP biogenesis. First, the P1 domain was shown to confer specificity toward peptides with high aromatic amino acid content, a feature common in OMP substrates (24–26). The P1 domain is capable of binding directly to these peptides, and it has been suggested that P1 may confer specificity to the otherwise nonspecific binding of the core chaperone module to substrates (26). In addition, a recent genetic selection for mutations that rescue the OM permeability and OMP assembly defects of a bamA mutant termed bamA616 yielded mutations that alter SurA P1. The bamA616 mutation confers OMP assembly defects that resemble a ΔbamB mutant and also weakens the BamA-SurA interaction, suggesting that BamA is unable to modulate SurA activity in this strain. One suppressor obtained through this selection, termed surA10 (S220A), is a gain-of-function mutation in the P1 domain at the P1-core interface (27). Despite restoring OMP assembly in both a bamA616 and a ΔbamB background, surA10 confers instability to the SurA protein in cell extracts, indicating that the P1 domain may inhibit SurA function and regulate its substrate repertoire. Despite this recent progress, there exists little direct evidence that the P1 domain serves a regulatory role in SurA activity and OMP biogenesis.

Using the genetic framework established by Ricci et al. (27), we report here additional genetic and biochemical evidence that establishes the P1 domain as a regulator of SurA function and OMP assembly. We use synthetic phenotypes to show that the P1 domain inhibits OMP assembly when regulation of SurA by the BAM complex is compromised. Furthermore, we use intradomain cysteine tethering to clarify the structural impact of the gain-of-function mutation surA10 and demonstrate directly that the activity of SurA is correlated with its conformational state. These findings represent the first evidence in vivo of conformational changes within SurA during OMP assembly and clarify the relationship among BamA, BamB, and the SurA parvulin domains in SurA regulation. These analyses support a model wherein the inactive parvulin domain P1 regulates OMP biogenesis through conformational change in coordination with the BAM complex.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Strains and plasmids used in this work are described in Table S1 of the supplemental material. All strains were constructed using standard microbiological techniques. When necessary, media were supplemented with 125 mg/liter ampicillin, 25 mg/liter kanamycin, 20 mg/liter chloramphenicol, or 25 mg/liter tetracycline. When used for gentle lysis, the cultures were supplemented with 0.2% l-arabinose to induce the pSurA plasmid derivatives. Otherwise, all experiments were conducted in the absence of l-arabinose due to the uninduced expression from the pAER vector. All bacterial cultures were grown under aerobic conditions at 37°C.

Site-directed mutagenesis.

Mutagenic oligonucleotides used in this work are described in Table S1 of the supplemental material. Site-directed mutagenesis was performed using the Stratagene QuikChange site-directed mutagenesis kit per the manufacturer's instructions using platinum Pfx polymerase (Invitrogen) and the appropriate primers. PCR products were digested with DpnI (New England BioLabs) for 1 h at 37°C and then used to transform Mach1 competent cells and plated on LB agar supplemented with the appropriate antibiotic. Plasmids isolated from these transformed strains were then confirmed by DNA sequencing and used in subsequent experiments.

Quantification of small-molecule and antibiotic sensitivity.

Efficiency-of-plating assays were performed by growing overnight cultures in the appropriate medium and serially diluting the overnight culture 10-fold in 200-μl volumes in a 96-well plate. The volumes were then spotted onto LB agar plates containing the indicated antibiotics using a 48-pin replicator. All results presented are representative of at least three independent experiments.

Western blot analysis.

Then, 250-μl samples of culture were pelleted (13,000 × g, 1 min) and resuspended in sample buffer at a volume equal to the optical density at 600 nm (OD₆₀₀)/40. Oxidized samples were resuspended in the same volume of sample buffer without β-mercaptoethanol (β-ME). Samples were boiled for 10 min and subjected to electrophoresis with a 10% SDS-PAGE gel. Immunoblotting was performed using rabbit polyclonal antisera that recognize maltose binding protein (MBP) (1:30,000 dilution), LamB/OmpA (1:30,000 dilution), LptD (1:5,000 dilution), or SurA (1:8,000) dilution). Donkey anti-rabbit secondary antibody conjugated to horseradish peroxidase (GE Healthcare) was used at a 1:8,000 dilution for all immunoblots. Immunoblots were visualized using the ECL antibody detection kit (Amersham) and Hyblot CL film (Denville Scientific). All results presented are representative of at least three independent experiments.

Gentle lysis.

Gentle lysis was performed as previously described (27). Briefly, cultures were grown to an OD₆₀₀ of ≈0.8 in the presence of 0.2% l-arabinose, harvested by centrifugation, and resuspended in 100 μl of 20 mM potassium phosphate (pH 7.2) and 150 mM NaCl. BugBuster solution (Novagen) containing 5 μg/ml lysozyme, 50 μg/ml DNase I, and 50 μg/ml RNase I was added to a volume of 1 ml, and cells were allowed to lyse by rocking for 15 to 30 min at room temperature. Lysates were then centrifuged for 1 min at 13,000 × g to pellet cellular debris. SDS-PAGE buffer was added to the samples (either with or without β-ME), and samples were boiled for 10 min prior to Western blot analysis, as described above.

RESULTS

Deletion of the second parvulin domain results in inhibition of SurA activity.

To determine the importance of the SurA parvulin domains in OMP assembly, we first confirmed the ability of SurA deletion mutants lacking the second (ΔP2) or both (ΔP1ΔP2) parvulin domains to support OMP assembly in a ΔsurA background. Consistent with previous reports, all three mutants fully complement ΔsurA OMP assembly phenotypes, confirming that the core module of SurA is sufficient to effectively chaperone and facilitate the assembly of substrates (see Fig. S1 in the supplemental material) (20).

It had been previously demonstrated that many of the OMP assembly factors, including chaperones and BAM complex members, function in overlapping and redundant pathways (4, 14, 27). Indeed, a role for the SurA P1 domain in vivo was revealed only when surA10 was isolated in a background defective for BAM complex function (27). We thus suspected that a properly functioning BAM complex might mask any phenotype imposed by the deletion of the SurA parvulin domains. As such, we took advantage of the previously used bamA616 and bamB-null mutants and constructed double mutants to search for synthetic phenotypes that might reveal a role for the SurA parvulin domains in maintaining OMP assembly in this impaired system. Importantly, the levels of LamB, a substrate whose assembly is already impaired in either a bamB-null or bamA616 mutant background, were further reduced specifically when the P2 domain of SurA was absent (Fig. 2A and B). Accordingly, the surAΔP2 mutation exacerbated the existing detergent (SDS-EDTA) sensitivity of a bamA616 mutant strain, indicating that the OM permeability barrier is disturbed (Table 1). On the other hand, SurAΔP1ΔP2 facilitated LamB assembly as well as wild-type (WT) SurA. Intriguingly, SurAΔP1ΔP2 did not restore the relative impermeability of the OM, suggesting that the P1 domain, while detrimental to LamB assembly in the absence of P2, may contribute to a separable SurA function involved in maintenance of the OM barrier. Together, these data demonstrate that, in cells with compromised BAM function, the loss of the P2 domain allows the remaining P1 domain to inhibit SurA activity, as the subsequent deletion of the P1 domain in a bamB-null or bamA616 mutant background relieved this inhibition. Importantly, the surAΔP2 mutant was not phenocopied by a PPIase-null point mutant, suggesting that the P1 inhibition of SurA activity observed in the absence of the P2 domain is not due to a loss of PPIase activity (see Fig. S2 in the supplemental material) (20).

FIG 2.

SurA activity is inhibited by the P1 domain. Relative OMP levels were determined in MC4100 and in strains containing each of the surA mutations listed in either a bamA616 (A) or bamB::kan (B) background by SDS-PAGE and immunoblotting for the proteins indicated. (C) Gentle lysis (G. lysis; + lanes) was performed by releasing cellular contents with BugBuster extraction reagent and incubating extracts at 25°C prior to SDS-PAGE and immunoblotting for the periplasmic proteins SurA and maltose binding protein (MBP).

TABLE 1.

Growth of surA mutant strains on SDS-EDTA

MC4100 genotype	Plasmid	Growth on SDS-EDTAa
Wild type		++++
ΔsurA bamA616	pACYC	–
ΔsurA bamA616	pSurA	+++
ΔsurA bamA616	pSurAΔP2	++
ΔsurA bamA616	pSurAΔP1ΔP2	++
ΔsurA bamA616	pSurA10	++++
ΔsurA bamA616	pSurA10ΔP2	++++
ΔsurA bamA616	pSurA 30 + 425	+++
ΔsurA bamA616	pSurA10 30 + 425	++++
ΔsurA bamA616	pSurA 59 + 214	–
ΔsurA bamA616	pSurA10 59 + 214	–
ΔsurA bamA616	pSurA 61 + 218	–
ΔsurA bamA616	pSurA10 61 + 218	–

^a++++, growth similar to WT; +++, less growth than WT; ++, some growth; –, no growth.

surA10 alleviates inhibition by the P1 domain.

As previously mentioned, Ricci et al. isolated a gain-of-function S220A mutation in the P1 domain, termed surA10, that restored OMP assembly and OM integrity in either a bamB-null or bamA616 mutant background (27). If the surAΔP2 mutant reflects a P1-inhibited form of SurA, we reasoned that the gain-of-function surA10 mutation might relieve this inhibition. Indeed, when the surA10 mutation (S220A) was combined with the surAΔP2 mutation, the resulting protein improved the assembly of LamB in bamA616 and bamB-null strains, though not to the extent observed in a surA10 background (Fig. 2A and B; Table 1). This indicates that the surA10 mutation biases this P1-inhibited form of SurA toward a more active state.

One of the more intriguing phenotypes conferred by the surA10 mutation to the SurA protein is protease sensitivity that is evident only when cells are gently lysed or periplasmic contents are extracted (27). Ricci et al. argue that this is due to the disruption of noncovalent contacts between the closely associated P1 and core modules, as the Ser residue affected by the surA10 mutation resides in an α helix of the P1 domain near its interface with the core module. If this structural state were connected to the activity of SurA, it would be expected that the SurAΔP2 mutant protein, unlike SurA10, would be stable in gentle lysis. In accordance with this expectation, we found that SurAΔP2 is as stable as WT SurA when cells are gently lysed and that the combination of the surA10 mutation with the surAΔP2 mutation resulted in degradation similar to that of SurA10 alone (Fig. 2C). Together with the genetic phenotypes presented, these data demonstrate that SurAΔP2 is inhibited due to an association between the P1 domain and core module of SurA, stabilized in the absence of P2, which is only revealed when coordination with the BAM complex is impaired.

Direct interrogation of P1-core inhibition using cysteine tethering.

Although the parvulin deletion mutants provide strong genetic evidence for the inhibitory function of the SurA P1 domain, we sought to directly correlate the functional and conformational states of SurA. Since our data indicated that a stabilized P1-core interaction is the cause of SurAΔP2 inhibition, we expected that permanently tethering these two domains in this closed conformation would irreversibly inhibit SurA function. We reasoned, likewise, that covalent stabilization would negate the suppression caused by the gain-of-function surA10 mutation. To test this prediction, we first identified sites within the SurA structure capable of forming intramolecular disulfide bond tethers using DSDBASE (28) and then constructed cysteine substitutions at those sites. Specifically, we constructed two cysteine double mutants predicted to covalently tether the P1 domain to the core module of SurA and inhibit function (Fig. 3A). Cysteine tethering has been used to test the conformational flexibility and physical mechanisms of diverse systems, such as RNA polymerase (29) and the multidrug efflux pump AcrB (30), and was recently used to demonstrate that the core module of SurA acts as a rigid body and does not undergo large-scale conformational changes (31). As a control, we also constructed an internal (core-core) disulfide bond within the core module using one of the cysteine pairs previously shown by Zhong et al. to form efficiently (31). We expected that this core-core tether, unlike the P1-core tethers, would have no impact on SurA activity or stability in the presence and absence of the surA10 mutation.

FIG 3.

Disulfide bonds properly form between the SurA P1 domain and core module. (A) SurA disulfide tethers were designed using the disulfide database (DSDBASE) (28) and were depicted on the SurA crystal structure (PDB accession no. 1M5Y). The surA10 residue (S220) is represented by red spheres. In each mutant, the residues used for cysteine substitution are represented by black spheres and are boxed. (B) Whole-cell lysates were suspended in loading buffer with (+, left) or without (−, right) the reducing agent β-ME prior to SDS-PAGE and immunoblotting for the proteins indicated. MBP is shown as a control.

To detect the presence of disulfide bond formation, we examined the electrophoretic mobility of the double cysteine mutants using reducing and nonreducing SDS-PAGE. Although the core-core disulfide tether (in the SurA mutant with substitutions at positions 30 and 425 [SurA 30 + 425]) resulted in little alteration of the nonreduced molecular weight, the formation of this tether had been previously confirmed (Fig. 3B) (31). Importantly, both of the P1-core tethered mutants (SurA 59 + 214 and SurA 61 + 218) migrated at a significantly lower molecular weight than WT SurA, consistent with disulfide bond formation.

P1-core tethered mutants stabilize surA10.

Due to the instability the surA10 mutation confers in cell extracts, it has been proposed that the surA10 mutation increases SurA activity by disrupting noncovalent interactions between the P1 domain and the core chaperone module of SurA (27). To directly test if the physical and functional states of SurA are indeed correlated, we first examined the protease sensitivity of our three disulfide-tethered mutants and their cognate surA10 variants. If the surA10 mutation weakens the interaction between the core and P1 domains, P1-core tethered variants should be stable against protease degradation even in the presence of this mutation. Indeed, both of the P1-core tethered mutants were as stable as WT SurA upon gentle lysis and also stabilized SurA10 (Fig. 4A). Importantly, stabilization of SurA10 required P1-core intradomain tethering; the core-core tethered mutant did not stabilize SurA10. In addition, we found that the SurA10 protein stabilized by both P1-core tethers migrated at a lower molecular weight than WT SurA in nonreducing SDS-PAGE, indicating that the stabilized species is disulfide tethered (Fig. 4B). These data strongly argue that the surA10 mutation improves SurA function by interrupting noncovalent interactions between the P1 and core modules of SurA.

FIG 4.

Cysteine tethers between the SurA P1 domain and core module lock SurA in a stable conformation. (A) Gentle lysis (G. lysis; + lanes) was performed by releasing cellular contents with BugBuster extraction reagent and incubating extracts at 25°C prior to SDS-PAGE and immunoblotting for the periplasmic proteins SurA and MBP. (B) Extracts from gentle lysis were suspended in loading buffer with (top panel, SurA_Red) or without (bottom panel, SurA^Ox) the reducing agent β-ME prior to SDS-PAGE and immunoblotting for the proteins indicated.

The activity of SurA is correlated with its conformation.

Due to their drastic stabilization of SurA10 upon gentle lysis, we suspected that these engineered SurA tethering mutants might inhibit OMP assembly even in an otherwise WT background. In addition, we sought to confirm that any defects observed in these strains were specific to disulfide bond formation and not due to other effects the cysteine substitutions may have on the SurA protein. To test these possibilities, we examined the assembly of a number of OMP substrates in ΔsurA strains containing the P1-core tethered mutants. As a control, we substituted serine for the cysteine residues in each mutant to examine the impact of the substitutions alone on OMP assembly and OM permeability. Importantly, the P1-core tethered mutants, but not the core-core control mutant, displayed OMP assembly defects and increased OM permeability (Fig. 5). In particular, the assembly of LptD, a complex OMP substrate that depends exquisitely on SurA for its assembly (and, unlike LamB, is not downregulated in response to envelope stress), was significantly compromised in both P1-core tethered mutants but not in the control core-core tethered mutant. In addition, the substitution of serine for the cysteine residues, while destabilizing the mutant proteins, resulted in normal OMP assembly. The OM barrier of the SurA 59S + 214S mutant was also restored, but the SurA 61S + 218S mutant was unable to fully complement the ΔsurA permeability defect (Fig. 5B). We suspect this was likely due to the very low levels of SurA in this strain (Fig. 5A, bottom panel). Taken together, these data indicate that the P1-core disulfide tethers genuinely inhibit SurA activity and OMP assembly and that these defects are a specific consequence of disulfide bond formation.

FIG 5.

Cysteine tethers between the SurA P1 domain and core module inhibit SurA function. (A) Relative OMP and periplasmic protein levels were determined in MC4100 and in strains containing each of the surA mutations listed in a ΔsurA background by SDS-PAGE and immunoblotting for the proteins indicated. (B) Antibiotic sensitivity was assessed by spotting serial 10-fold dilutions of stationary-phase cultures of ΔsurA strains containing the mutations indicated on LB with or without vancomycin and incubated at 37°C.

P1-core disulfide tethers promote surA10.

Last, we sought to determine whether the stability conferred to P1-core tethered SurA10 upon gentle lysis correlated with negated suppression of surA10 in bamA616 and bamB-null mutant backgrounds. In accordance with their drastic effects on chaperone function, both P1-core disulfide tethers also conferred a significant LamB and OmpA assembly defect in bamA616 and bamB-null backgrounds, while the control core-core tethered mutant exhibited OMP levels comparable to those of WT SurA (Fig. 6). Importantly, the core-core tethered SurA10 protein also still suppressed both the OMP assembly and the envelope defects of bamB-null and bamA616 mutants, but both P1-core tethers abolished surA10-mediated suppression (Fig. 6; Table 1). These data show that the P1-core tethered mutants, like a surAΔP2 mutant, inhibit SurA activity. Moreover, the results demonstrate that this covalent inhibition, unlike the inhibition imposed by surAΔP2, cannot be reversed by the surA10 mutation. These phenotypes strongly correlate with the protease resistance conferred to SurA10 by the P1-core tethers, suggesting that the surA10 mutation improves SurA function by interrupting noncovalent interactions between the P1 and core domains of SurA and that the activity of SurA is diminished if the P1 and core domains are covalently associated.

FIG 6.

Disulfide bonds between the SurA P1 domain and core module negate surA10 suppression. Relative OMP levels were determined in MC4100 and in strains containing each of the surA mutations listed in either a ΔsurA; bamA616 (A) or ΔsurA; bamB::kan (B) background by SDS-PAGE and immunoblotting for the proteins indicated.

DISCUSSION

Despite significant efforts in vivo and extensive in vitro characterization, a general role for the SurA parvulin domains has been difficult to identify (20, 26). Taking into account the redundancy of the OMP assembly network, we used double mutants to uncover an inhibitory role for the SurA P1 domain and a regulatory role for P2. Furthermore, the activity of SurA, which is modulated by the parvulin domains, is correlated with distinct protease-sensitive and -resistant conformational states. We propose that wild-type SurA interconverts between these conformational and functional states in a manner that is mediated by the P1 domain and regulated by the P2 domain and the BAM complex.

Our finding that the deletion of P2 exacerbates the OMP assembly and OM permeability phenotypes of a bamA616 or a bamB-null strain offers some clarity as to the role of P2 in SurA function. Consistent with the observations of Ricci et al. (27), we argue that BamB and BamA share a function in the regulation of SurA activity and that this regulation is potentiated either directly or indirectly through the SurA P2 domain. In a bamA616 or a bamB-null background, the P2 domain is sufficient to facilitate partial regulation of SurA, but the absence of P2 leads to constitutive P1-mediated inhibition of SurA. Accordingly, inhibition is relieved either by surA10 or by the additional deletion of the inhibitory P1 domain. We propose that surA10 biases the protein toward a more open and active state that no longer requires regulation by the BAM complex and may deliver substrates in a manner that requires minimal interaction with the BAM complex. In agreement with this model, although SurA10 suppresses bamA616 mutant phenotypes, it binds the BamA616 mutant protein even more poorly than WT SurA does (27).

It remains to be elucidated if the P2 domain regulates SurA activity through contact with BamA or substrate or through an unknown mechanism. The bamA616 mutation reduces SurA binding to BamA, but bamB-null mutants exhibit no SurA-BamA binding defect (27). These results may be reconciled if BamA and BamB operate sequentially in the regulation of SurA. In this case, binding at the BamA POTRA domains is required for SurA regulation, but BamB activity is required specifically for the disinhibition of SurA. Intriguingly, some bacterial species (including Neisseria meningitidis) that do not have an inhibitory SurA P1 domain also lack bamB, implying that, in these organisms, BamB may not be important for OMP assembly, because it is not required for regulation of SurA (23, 32). While we think it unlikely that SurA regulation is the only function for BamB in most bacteria, important insight may come from an understanding of SurA regulation in these seemingly simpler systems.

Aside from its role in the regulation of SurA, it is also possible that the PPIase-active P2 domain contributes to substrate biogenesis. The PPIase domain of trigger factor is also nonessential for chaperone function (19) but makes nonpolar contacts with substrates (33, 34) and is involved in the proline-limited folding of specific substrates (35). Considering that the PPIase activity of P2 is not involved in regulating P1 inhibition, the P2 domain of SurA may serve a similar role, exhibiting accessory sequestration activity toward some substrates and enhancing the biogenesis of specific proteins (23). Indeed, the PPIase activity of P2 was recently shown to be important for the complete surface localization of Yersinia pseudotuberculosis invasin (36), and at least one SurA parvulin domain is required for complete novobiocin resistance and epithelial cell invasion in uropathogenic E. coli (37). A general role for P2 in the chaperoning of typical SurA substrates, however, remains elusive.

The P1 domain was first demonstrated to confer specificity to substrate binding, which is consistent with the accessory contribution of the P1 domain to chaperone activity but difficult to demonstrate in vivo (26). Taking advantage of synthetic phenotypes, we found that P1 also functions to inhibit and regulate SurA activity in vivo. We propose that P1 fulfills both of these functions in distinct steps. When not engaged with substrate, P1 may autoinhibit SurA and stabilize the core module. When binding substrate, however, the P1 domain instead contributes to assembly by providing substrate specificity.

Through the use of cysteine tethering and SurA10 mutant protein instability, we were able to directly correlate this P1-mediated inhibition and its relief by surA10 with the conformational state of the chaperone. We propose a model wherein interdomain interactions and conformational changes allow SurA to populate an equilibrium between an unstable, activated, open state and a stable, inhibited, closed state (Fig. 7). These shifts in conformation and activity are evident in autoinhibited signaling proteins (38) and are supported by the recent capture of SurA in a P1-core open conformation (26). Furthermore, the shifts in both activity and stability revealed here are interestingly similar to the functional transient disorder exhibited by another ATP-independent chaperone, Hsp33 (39). Although the cytoplasmic Hsp33 chaperone is regulated by oxidative stress, it is tempting to consider that SurA P1 too may become transiently unstable in response to activation and that this destabilization may promote substrate recognition and chaperone activity. In this way, the P1 domain may serve both regulatory and functional roles important for the proper assembly of OMPs.

FIG 7.

Model for the structural and functional equilibrium of SurA. SurA exists in equilibrium between two (less active and more active) states that are correlated with its structural stability (stable and unstable, respectively). The surA10 mutation biases SurA toward a more structurally disorganized, but more active, conformation through the loosening, or opening, of the P1-core domain interface. The surAΔP2 deletion (indicated by the dashed arrow and dashed P2 domain border) and P1-core disulfide tethers (labeled surA P1-core), however, bias SurA toward a more structured and stable but less active state, where the P1 and core domains are locked in a closed conformation. This equilibrium is regulated by the BAM complex, as the surA10 and surAΔP2 phenotypes are only clear when the BAM complex function is disrupted, such as in a bamA616 or bamB-null mutant.

Finally, the stabilization and autoinhibition imposed by the P1 domain may provide a protective function in response to changing conditions of the extracellular (and, thus, periplasmic) environment. Indeed, SurA is protected from aggregation by the acid-activated chaperone HdeA at low pH (40). In addition, the purified chaperone module of SurA is prone to aggregation and is unstable in vitro (26). P1 inhibition may provide stability like HdeA, namely, by protecting the critical chaperone activity of SurA during periods of sparse growth and OMP assembly or in response to changes in growth conditions. In contrast, in its active state, SurA provides critical chaperone activity at the expense of its stability.

Our work here does not differentiate between the various conformational states that SurA occupies when bound to BamA, substrate, or both. In addition, these data do not rule out the possibility of multiple rounds of conformational change and activation in the binding, chaperoning, and delivery of a single substrate. Nevertheless, our confirmation of the open and closed natures of the SurA10 and SurAΔP2 mutant proteins, respectively, argues strongly that the BAM complex is required to disinhibit or derepress SurA during the OMP delivery process. It remains to be elucidated how this newfound regulation of SurA relates to the recognition of and binding to substrate as well as the interplay between SurA and the other periplasmic members of the OM biogenesis network.