The sacrificial adaptor protein Skp functions to remove stalled substrates from the β-barrel assembly machine.

The biogenesis of integral β-barrel outer membrane proteins (OMPs) in gram-negative bacteria requires transport by molecular chaperones across the aqueous periplasmic space. Owing in part to the extensive functional redundancy within the periplasmic chaperone network, specific roles for molecular chaperones in OMP quality control and assembly have remained largely elusive. Here, by deliberately perturbing the OMP assembly process through use of multiple folding-defective substrates, we have identified a role for the periplasmic chaperone Skp in ensuring efficient folding of OMPs by the β-barrel assembly machine (Bam) complex. We find that β-barrel substrates that fail to integrate into the membrane in a timely manner are removed from the Bam complex by Skp, thereby allowing for clearance of stalled Bam-OMP complexes. Following the displacement of OMPs from the assembly machinery, Skp subsequently serves as a sacrificial adaptor protein to directly facilitate the degradation of defective OMP substrates by the periplasmic protease DegP. We conclude that Skp acts to ensure efficient β-barrel folding by directly mediating the displacement and degradation of assembly-compromised OMP substrates from the Bam complex.

The cell envelopes of gram-negative bacteria, mitochondria, and chloroplasts all contain an outer membrane (OM) consisting of integral transmembrane proteins that assume a β-barrel conformation (1, 2). In gram-negative bacteria such as Escherichia coli, β-barrel outer membrane proteins (OMPs) contribute to the selective permeability of the OM, protecting the cell from harmful molecules while still allowing for the uptake of nutrients (3). Structurally and functionally diverse OMPs serve a number of roles critical to cell viability, namely the selective passage of small molecules, efflux of toxins, insertion of lipopolysaccharide (LPS) onto the cell surface, and assembly of OMPs themselves (1, 4). Reflective of their importance in maintaining cellular integrity, defects in OMP biogenesis confer sensitivity to a wide array of toxic molecules including detergents, bile salts, and most importantly, antibiotics (5, 6). As such, considerable efforts have been made to identify agents that inhibit essential cellular processes performed by OMPs (7–12), with hopes of hastening the development of novel therapeutics to combat the ever-growing threat of antibiotic-resistant infections caused by gram-negative microbes (13, 14).

Ensuring efficient OMP biogenesis is a particularly challenging cellular feat. Newly synthesized OMPs must traverse the aqueous, oxidizing periplasm in an unfolded state, avoid self-aggregation, and subsequently complete proper assembly, all in an environment devoid of cellular energy such as adenosine triphosphate (15). A multitude of molecular chaperones and proteases function to overcome this challenge by minimizing unfolded OMP accumulation and facilitating OMP transport to the OM assembly machinery (16). Although more than a dozen chaperones and proteases with clear implications in OMP biogenesis have been identified (16–18), the most well-characterized and predominant proteins in E. coli are the chaperones SurA and Skp, as well as the chaperone protease DegP. Numerous genetic, biochemical, and proteomic studies have demonstrated that SurA is the primary periplasmic chaperone that facilitates transport of the bulk mass of OMP substrates to the OM (19–24). Skp and DegP, on the other hand, comprise a secondary, partially redundant OMP biogenesis pathway that primarily serves to minimize accumulation of unfolded OMPs, either by rescuing their assembly or promoting their degradation (19, 20).

Notably, Skp binds unfolded OMPs with dissociation constants in the low nanomolar range (25, 26), exceeding the binding affinities of either SurA or DegP (27–29), to form highly stable Skp–OMP complexes that display lifetimes on the order of hours (30). Given the substantial stability of Skp–OMP complexes, the precise mechanism of OMP release from Skp remains poorly understood. The rapid conformational dynamics of OMPs bound within the Skp cavity have been proposed to enable local substrate release that is ultimately driven by the recognition and folding of OMPs by the OM assembly machinery (30), thus coupling client release from Skp to the thermodynamic stability provided by OMP integration into a membrane (31). Indeed, substrate release and folding of OMPs from Skp–OMP complexes is enabled in vitro by incubation with OM folding machinery–containing liposomes (28, 32), demonstrating that Skp can facilitate productive OMP assembly. This mechanism of folding-driven substrate release has been similarly observed in genetic and biochemical studies indicating that Skp is capable of directly inserting OMPs into lipid bilayers in vitro (33), as well as the inner membrane in vivo (34), without assistance from the OM assembly machinery.

Whether OMPs are capable of being removed from Skp within physiological timescales in the absence of coupled folding, however, is not entirely clear. Under conditions of periplasmic stress, in which the burden of unfolded OMPs exceeds the rate at which they can be assembled, the activities of both Skp and DegP become crucial (19, 20, 24, 35). Given that Skp not only binds substrates with a higher affinity than DegP (29) but also does so several orders of magnitude faster (36), how unfolded OMPs are transferred from Skp to DegP for degradation under stress conditions is not obvious. Indeed, direct transfer of an OMP from Skp to DegP has yet to be demonstrated, and intriguingly, the formation of Skp–DegP–OMP ternary complexes has been reported in such experiments (29, 36).

Folding and insertion of nascent OMPs into the OM is catalyzed by the heteropentameric β-barrel assembly machine (Bam) complex, consisting of the BamA β-barrel and four accessory lipoproteins, BamBCDE (37, 38). Recent biochemical and structural studies have provided a relatively clear current model for the mechanism of β-barrel assembly. Following substrate recruitment to BamD (39), BamA catalyzes the sequential addition of β-hairpins in a C-to-N-terminal manner (40), with early folding occurring within the interior of the BamA barrel (41). Folding proceeds until membrane integration occurs, and subsequent stepwise hydrogen-bond formation between N and C substrate termini facilitates barrel closure and substrate release into the membrane (40).

One outstanding question concerns the fate of OMP substrates that have stalled while folding on the Bam complex. Protein misfolding in the periplasm, translational error, or impaired Bam complex function can result in substrates arresting on the assembly machinery, a condition that can ultimately be lethal if left unchecked (42–44). Until recently, investigations of stalled OMP substrates have been largely impeded by a lack of structurally defined folding intermediates and the absence of an established general mechanism of OMP assembly. Several studies to date have utilized mutant alleles of the large β-barrel LptD to probe Bam complex assembly (39, 41, 45, 46), and multiple proteases that degrade assembly-compromised LptD within distinct stages of its folding regime have been identified (46, 47). It is unclear, however, whether these stringent quality control mechanisms monitoring assembly of LptD are exerted on all β-barrel substrates or whether LptD represents a unique case given its remarkably complex folding trajectory (48). Given that OMP assembly by the Bam complex has evolved to be incredibly efficient—so efficient that unfolded OMPs cannot be detected at steady state—it stands to reason that quality control mechanisms ensuring the efficient assembly of all β-barrel substrates exist. Recently, it has been shown that extracellular loop deletions within the C-terminal half of the BamA β-barrel cause early folding defects and thus render stalled BamA susceptible to proteolysis by DegP (40). How DegP actively disengages a partially folded, stalled substrate from its folding on BamA, given the relatively weak and slow nature of DegP binding, is not obvious.

Here, we have utilized an assembly-defective variant of a slow-folding β-barrel OMP to investigate the fate of substrates that engage the OM assembly machinery but otherwise fail to undergo efficient folding and membrane integration. We identify a specific role for the periplasmic chaperone Skp in facilitating the degradation of defective OMP substrates by the protease DegP, thus imposing an active quality control mechanism that serves to remove assembly-compromised substrates from the Bam complex. Strikingly, we find that Skp is degraded alongside its bound substrate by DegP, thereby functioning as a sacrificial adaptor protein. By evaluating the requirement for Skp in degradation of a series of sequentially stalled β-barrel substrates, we find that Skp is only required to degrade substrates that have initiated folding on the Bam complex. Thus, β-barrel OMPs that have stalled during assembly specifically require Skp for their removal from the Bam complex and subsequent degradation by DegP. We conclude that Skp acts to ensure efficient β-barrel assembly by facilitating both the direct removal and degradation of stalled substrates from the Bam complex.

Results

Deletion of skp Suppresses the Assembly Defects of LptDY721D.

To gain insight into potential specialized roles of periplasmic chaperones during OMP biogenesis, we deliberately perturbed the OMP folding process by utilizing an assembly-defective mutant of an essential, slow-folding β-barrel substrate, LptD. LptD, together with its cognate lipoprotein LptE, forms the OM translocon of the Lpt (LPS transport) complex that facilitates the transport and insertion of LPS into the outer leaflet of the OM (49, 50). Folding of LptD is remarkably slow, orders of magnitude slower than other β-barrel substrates (51), taking roughly 20 min to fully complete (48). A mutant LptD substrate that harbors an amino acid substitution of a conserved tyrosine residue at position 721 for aspartate, LptDY721D, roughly triples the relatively long folding time of LptD by causing early assembly defects that are subsequent to substrate engagement (39). LptDY721D is not deficient in its ability to be recruited to the Bam complex, but rather is slow to complete folding on BamA and is thereby subject to proteolytic degradation (39, 46). Consequently, the early assembly defects of LptDY721D confer a recessive sensitivity to antibiotics due to reduced levels of functional LptDE translocons (39).

Degradation of LptDY721D is known to be mediated by a number of periplasmic and OM proteases, most notably the periplasmic protease DegP (39, 46). However, deletion of degP primarily affects total levels of LptDY721D without significant increases in the amount of fully folded, mature protein (46), suggesting that LptDY721D assembly is somehow largely prevented even in the absence of substantial proteolysis. Indeed, deletion of degP is insufficient to rescue the permeability defects conferred by LptDY721D in the presence of either bacitracin or vancomycin (46). Therefore, we reasoned that an additional periplasmic factor was responsible for preventing efficient LptDY721D assembly even in the absence of LptDY721D proteolysis.

Strikingly, we found that deletion of skp significantly suppressed the assembly defects of LptDY721D, as indicated by a restoration of resistance to both bacitracin and vancomycin (Fig. 1). Additionally, while deletion of degP in an lptD genetic background decreased viability in the absence of antibiotic (Fig. 1), as previously observed (46), further deletion of skp rescued cellular viability (Fig. 1). These results imply that even in the absence of substantial proteolysis by DegP, LptDY721D assembly is largely prevented by Skp. Moreover, in the absence of Skp, efficient degradation of LptDY721D by DegP must not occur.

Fig. 1.

Deletion of skp suppresses the assembly defects of LptDY721D. (A) Deletion of skp suppresses the OM defects conferred by LptDY721D. The indicated strains were grown overnight at 30 °C in LB medium, 10-fold serially diluted, and replica plated onto Mueller Hinton II agar supplemented with either 575 mg/L bacitracin or 90 mg/L vancomycin as indicated. Plates were incubated overnight at 37 °C. Results are representative of at least three independent experiments. (B) Deletion of skp improves LptDY721D assembly. The indicated strains were grown overnight at 37 °C, and assembly of LptDY721D was assayed by nonreducing (LptDOX) or reducing (LptDRED) SDS-PAGE followed by immunoblotting. RpoA levels are shown as a loading control. Blots are representative of at least three independent experiments.

To directly assess the effects of skp deletion on LptDY721D assembly, we examined the oxidation state of LptDY721D in the presence or absence of skp and/or degP. Upon completion of barrel folding, LptD undergoes a complex oxidative rearrangement to form two intramolecular disulfide bonds between nonconsecutive pairs of cysteine residues (48). Because only mature, properly assembled LptD adopts two nonconsecutive disulfide bonds (52), oxidation state thus serves as a reliable proxy for LptD assembly. Examination of LptDY721D assembly using both nonreducing and reducing sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) revealed that while deletion of either skp or degP increased the relative levels of total reduced LptDY721D (Fig. 1), only deletion of skp promoted the assembly of properly oxidized, mature LptDY721D (Fig. 1). These findings demonstrate that the beneficial effects observed upon deletion of skp are a direct result of improved LptDY721D assembly.

Taken together, these results suggest that LptDY721D is only able to reside on the Bam complex long enough to complete its lengthy assembly process if Skp specifically is absent. Given that LptDY721D is recruited normally to the OM assembly machinery (39), we conclude that Skp functions to remove assembly-defective LptDY721D from the Bam complex in order to directly or indirectly facilitate its degradation by DegP.

Skp Is Degraded with Bound Substrate by DegP.

Given that Skp was largely responsible for preventing efficient LptDY721D assembly, we next wondered whether Skp was directly facilitating the degradation of assembly-defective substrates such as LptDY721D by DegP. To address this hypothesis, we first utilized in vivo protein stability assays to monitor the degradation kinetics of LptDY721D in the absence of translation. Diploid strains expressing 3×FLAG-tagged LptD or LptDY721D from a low-copy plasmid were used in order to complement the recessive OM defects conferred by LptDY721D (39). In cells harboring wild-type LptD, total levels of LptD remained relatively constant following addition of excess spectinomycin (Fig. 2 ). However, in cells expressing LptDY721D, significant degradation of LptDY721D was observed over a period of 45 min following inhibition of protein synthesis. Importantly, the degradation of LptDY721D could be prevented upon deletion of either skp or degP (Fig. 2 ), suggesting that both Skp and DegP are required for efficient degradation of LptDY721D.

Fig. 2.

Skp is degraded with substrate by DegP. (A) LptDY721D is degraded in a Skp- and DegP-dependent manner. In vivo protein stability assays were performed to monitor LptDY721D degradation in the absence of translation. The indicated strains were grown to midlogarithmic phase at 37 °C and treated with 300 μg/mL spectinomycin to inhibit de novo protein synthesis. Aliquots of each culture were taken at the indicated timepoints and analyzed by SDS-PAGE and immunoblotting. OmpA immunoblots are shown as a loading control. Blots are representative of at least three independent experiments. (B) Quantification of immunoblotting data shown in A over three independent experiments. Plotted data represent protein abundance over time relative to the initial abundance at t = 0. Individual datapoints represent the mean ± SEM. Statistical significance was determined by two-way ANOVA with Tukey’s post hoc test. *P < 0.05; **P < 0.01. (C) Skp is degraded in an LptDY721D- and DegP-dependent manner. In vivo protein stability assays were performed as described in A and analyzed by SDS-PAGE and immunoblotting. The upper band is a cross-reactive protein (X) that serves as a loading control. Blots are representative of at least three independent experiments. (D) Quantification of immunoblotting data shown in C over three independent experiments. Plotted data represent protein abundance over time relative to the initial abundance at t = 0. Individual datapoints represent the mean ± SEM. Statistical significance was determined by two-way ANOVA with Tukey’s post hoc test. *P < 0.05.

Previously, our laboratory found that the periplasmic chaperone CpxP functions as a DegP adaptor protein to suppress the toxicity conferred by misfolded pilin subunits in the periplasm (53). Notably, CpxP’s enhancement of DegP-mediated proteolysis requires degradation of both the adaptor protein CpxP and its bound substrate by DegP (53). Given the requirement for Skp in efficient degradation of LptDY721D (Fig. 2 ) and the substantial affinity of Skp–OMP binding (25, 26, 29), we wondered whether Skp was similarly functioning as a sacrificial adaptor protein to directly mediate LptDY721D proteolysis by DegP. To evaluate this possibility, we monitored the degradation kinetics of Skp using the same in vivo protein stability assay conditions described above. Indeed, while Skp levels remained relatively constant in a wild-type strain, Skp was rapidly degraded in an LptDY721D-dependent manner (Fig. 2 ). Moreover, degradation of Skp could be prevented upon deletion of degP (Fig. 2 ), demonstrating that DegP was mediating the observed Skp proteolysis. Taken together, we conclude that the periplasmic chaperone Skp functions as a sacrificial adaptor protein that is degraded with bound LptDY721D by the periplasmic protease DegP.

Skp Functions to Remove Stalled OMPs from the Bam Complex.

Since the folding defects of LptDY721D are subsequent to engagement with the OM assembly machinery (39), we next wondered whether Skp functions generally to disengage OMPs that have stalled during folding on the Bam complex and are otherwise inaccessible for degradation by DegP. To test this hypothesis, we wished to design a series of OMP substrates that would stall at discrete points during the folding process and varied in the degree of assembly at the point of stalling. Folding of the BamA β-barrel by the Bam complex was recently shown to proceed in a C-to-N-terminal manner, with deletion of each extracellular loop within the BamA β-barrel causing substrates to stall at distinct points in assembly through slowed formation of β-hairpins (40). Extracellular loop deletions within the C-terminal half of the BamA β-barrel cause an early stalling in assembly and render BamA susceptible to degradation by DegP (40). We therefore reasoned that employing analogous C-terminal extracellular loop deletions for the model OMP LamB would allow us to evaluate the specific requirement for Skp in the degradation of stalled substrates by DegP. Moreover, because LamB exists functionally as a trimer, discrimination between the effects of skp deletion on both degradation and assembly of mutant substrates can be discerned by assaying levels of total monomeric and functional trimeric LamB, respectively.

A series of plasmid constructs harboring either full-length LamB or LamB that lacked one of nine extracellular loops (Fig. 3) was generated and expressed in wild-type, Δskp, ΔdegP, and Δskp ΔdegP genetic backgrounds. Levels of both total monomeric and functional trimeric LamB were then assayed to determine the requirement for Skp in either stabilization or assembly of mutant LamB proteins, respectively. Similarly to the BamA β-barrel (40), deletions of extracellular loops within the C-terminal half of LamB (ΔL9‒ΔL6) resulted in substantial LamB degradation in wild-type genetic backgrounds and showed significantly increased levels of total monomeric LamB upon degP deletion (Fig. 3), indicating early folding defects. Defects were similarly apparent in the amount of functional LamB trimer, where all mutant LamB proteins showed significantly reduced levels of trimeric LamB compared to wild-type protein (Fig. 3). For the C-terminal–most extracellular loop deletion (ΔL9), we observed increased levels of trimeric LamB upon deletion of skp or degP (Fig. 3), suggesting that LamBΔL9 is stalled at a point during folding at which Skp is not absolutely required for its degradation. However, for all subsequent extracellular loop deletions (ΔL8‒ΔL6), efficient assembly into trimeric LamB specifically required that Skp be absent (Fig. 3). Given that LamBΔL8, LamBΔL7, and LamBΔL6 are expected to have templated either three, five, or seven β-strands prior to the point of stalling, respectively (Fig. 3), these results suggest that Skp is specifically required to remove stalled substrates that have already initiated folding on the Bam complex.

Fig. 3.

Skp functions to remove stalled substrates from the Bam complex. (A) Structure of a LamB monomer depicting extracellular loops 6 through 9. The N- and C-terminal β-strands are shown in red and blue, respectively. Images were generated from the LamB crystal structure (Protein Data Bank ID: 1mal) using PyMOL. (B) Efficient assembly of LamBΔL8, LamBΔL7, and LamBΔL6 specifically requires that Skp is absent. The indicated strains were grown overnight at 30 °C, and levels of trimeric LamB were assayed by seminative SDS-PAGE followed by immunoblotting. For simultaneous analysis of total monomeric LamB levels, an aliquot of each sample was boiled (denatured) and analyzed by SDS-PAGE and immunoblotting. OmpA levels are shown as a loading control. Blots are representative of at least three independent experiments.

To further test this model, we wished to confirm that a Skp-dependent substrate (e.g., LamBΔL7), but not a Skp-independent substrate (e.g., LamBΔL9), was undergoing proteolysis in a Skp-dependent manner. As before, we utilized in vivo protein stability assays to monitor the degradation kinetics of both LamB and Skp in strains expressing either LamBΔL9 or LamBΔL7. In evaluating LamB stability, we found that both LamBΔL9 and LamBΔL7 underwent significant degradation over the course of 45 min posttreatment with translational inhibitor, with LamBΔL7 being degraded at a slightly slower overall rate than LamBΔL9 (Fig. 4 ). However, upon examining the stability of Skp, we observed significant Skp degradation only for LamBΔL7- and not LamBΔL9-expressing strains (Fig. 4 ). These data suggest that the degradation of LamBΔL7 by DegP requires prior removal of stalled LamBΔL7 from the Bam complex by Skp.

Fig. 4.

Skp is only required to degrade substrates that have initiated folding on the Bam complex. (A) Both LamBΔL9 and LamBΔL7 undergo degradation over the course of 45 min following inhibition of protein synthesis. The indicated strains were grown to midlogarithmic phase at 37 °C and treated with 300 μg/mL spectinomycin to inhibit de novo protein synthesis. Aliquots of each culture were taken at the indicated timepoints and analyzed by SDS-PAGE and immunoblotting. RpoA levels are shown as a loading control. Blots are representative of at least three independent experiments. (B) Quantification of immunoblotting data shown in A over three independent experiments. Plotted data represent protein abundance over time relative to the initial abundance at t = 0. Individual datapoints represent the mean ± SEM. Statistical significance was determined by two-way ANOVA with Sidak’s post hoc test. **P < 0.01. (C) Skp is only degraded in LamBΔL7-expressing strains. In vivo protein stability assays were performed as described in A and analyzed by SDS-PAGE and immunoblotting. The upper band is a cross-reactive protein (X) that serves as a loading control. Blots are representative of at least three independent experiments. (D) Quantification of immunoblotting data shown in C over three independent experiments. Plotted data represent protein abundance over time relative to the initial abundance at t = 0. Individual datapoints represent the mean ± SEM. Statistical significance was determined by two-way ANOVA with Sidak’s post hoc test. *P < 0.05; **P < 0.01.

Given the enormous stability of paired β-strands within a membrane environment, it seemed unlikely that the displacement activity of Skp would also extend to late-stage folding intermediates. To address this possibility, we first examined the effects of skp deletion on the late-stalled substrates LptD4213 (45, 54) and LamBΔL5, both of which would be predicted to have templated a substantial number of β-strands prior to the point of stalling. Deletion of skp and/or degP had no discernible effect on LptD4213 assembly (), in agreement with previous observations that LptD4213 is not a substrate of DegP (46, 55). In contrast to prior LamB loop deletions (ΔL9‒ΔL6), LamBΔL5 was largely stable and not subject to substantial proteolysis in a wild-type genetic background (), similar to what was previously observed for late-stalled BamA substrates (40). Additionally, deletion of skp and/or degP did not significantly alter LamBΔL5 trimerization (), suggesting that LamBΔL5 is largely inaccessible to removal by Skp.

We then wished to determine whether Skp activity was directly correlated with the ability of a substrate to be extracted from the membrane following treatment with urea. It has previously been shown that early-stage β-barrel folding intermediates are susceptible to membrane extraction by either sodium carbonate or urea, whereas late-stage folding intermediates are not (39, 40, 55). During initial experiments performed with cells grown at 37 °C, we observed significant down-regulation of LamB ΔL9‒ΔL6 expression when degP was deleted. This was likely due to σE-mediated translational repression via the small noncoding RNAs (sRNAs) MicA and RybB, of which LamB is a known regulatory target (56). To control for these effects, we performed two sets of experiments in parallel. First, membrane extractions were performed using the native LamB constructs with cells grown at 30 °C. At the same time, we ran identical extraction experiments with cells grown at 37 °C using LamB constructs in which the LamB signal sequences were replaced with the BamA signal sequence (BamA(SS)-LamB) in order to circumvent sRNA-mediated translational repression, as BamA expression is not down-regulated by either MicA or RybB (56). Following translocation across the inner membrane and cleavage of either the LamB or BamA signal sequence, the mature LamB proteins produced are identical and thus would be expected to undergo identical assembly processes upon being delivered to the OM.

Notably, both sets of membrane extraction experiments—with either native LamB constructs expressed at 30 °C or BamA(SS)-LamB constructs expressed at 37 °C—produced similar trends in substrate susceptibility to extraction by urea. Importantly, we observed a clear directionality in the substrate extraction profiles, with the C-terminal–most LamB loop deletions (LamBΔL9, LamBΔL8, and LamBΔL7) being sensitive to urea extraction (Fig. 5 ), whereas the more N-proximal loop deletions (LamBΔL6 and LamBΔL5) were largely resistant to extraction (Fig. 5 ), suggesting that LamB indeed undergoes C-to-N-terminal folding in a manner similar to BamA (40). Strikingly, LamBΔL6 was a substrate we observed to be markedly Skp-dependent in assembly (Fig. 3), demonstrating that Skp is even capable of displacing a substrate that has completed sufficient folding to resist extraction by urea (Fig. 5 ).

Fig. 5.

Skp activity is correlated with the degree of substrate assembly. (A) Urea extraction experiments were performed for LamB substrates ΔL9 through ΔL5 using cells grown at 30 °C. Total lysate, membrane prewash, and membrane postwash fractions were analyzed by SDS-PAGE and immunoblotting. BamA immunoblots are shown as a loading control. Blots are representative of three independent experiments. (B) Quantification of immunoblotting data shown in A over three independent experiments. Pre- and postwash abundances were calculated relative to the amount of wild-type (WT) substrate present and the percent of substrate remaining was determined by dividing the postwash abundance by the prewash abundance. Plotted data represent the mean ± SD. Statistical significance was determined by one-way ANOVA with Dunnett’s post hoc test. **P < 0.01. (C) Urea extraction experiments were performed for BamASS-LamB substrates ΔL9 through ΔL5 using cells grown at 37 °C. Total lysate, membrane prewash, and membrane postwash fractions were analyzed by SDS-PAGE and immunoblotting. BamA immunoblots are shown as a loading control. Blots are representative of three independent experiments. (D) Quantification of immunoblotting data shown in C over three independent experiments. The percent of substrate remaining was calculated as described in B. Plotted data represent the mean ± SD. Statistical significance was determined by one-way ANOVA with Dunnett’s post hoc test. *P < 0.05; ***P < 0.001. (E) Urea extraction experiments were performed for LptD4213 and LptDY721D using cells grown at 37 °C. Total lysate, membrane prewash, and membrane postwash fractions were analyzed by SDS-PAGE and immunoblotting. BamA immunoblots are shown as a loading control. Blots are representative of three independent experiments. (F) Quantification of immunoblotting data shown in E over three independent experiments. The percent of substrate remaining was calculated as described in B. LptD4213 and LptDY721D abundances were calculated relative to the matched WT and ΔdegP control strains, respectively. Plotted data represent the mean ± SD. Statistical significance was determined by one-way ANOVA with Tukey’s post hoc test. ***P < 0.001.

Lastly, we conducted similar experiments to determine whether the inability of Skp to affect LptD4213 assembly was correlated with the membrane extraction susceptibility of LptD4213. We utilized LptDY721D as a positive reference control, as LptDY721D has been previously shown to be susceptible to membrane extraction with sodium carbonate (39). As predicted, LptD4213 was largely resistant to membrane extraction with urea, in stark contrast with the pronounced extraction susceptibility of LptDY721D (Fig. 5 ). We conclude that Skp activity is correlated with the degree of substrate assembly, where substrates that proceed far enough in the assembly process (e.g., LamBΔL5 and LptD4213) are inaccessible to displacement by Skp.

Taken together, we conclude that Skp functions to ensure efficient OMP assembly by specifically removing stalled substrates from the Bam complex and subsequently serving as a sacrificial adaptor protein for their degradation.

Discussion

In this study, we have used the slow-folding β-barrel LptDY721D as a tool to glean mechanistic insight into the fate of substrates that engage the OM assembly machinery but cannot efficiently membrane integrate. Our results reveal an active quality control mechanism in which Skp directly facilitates the degradation of assembly-compromised OMP substrates from the Bam complex. We have demonstrated that following sequestration of defective OMPs, Skp is degraded alongside its bound substrate by DegP, thereby functioning as a sacrificial adaptor protein to enhance DegP-mediated proteolysis. We have established that the specific requirement for Skp in DegP-mediated proteolysis is largely restricted to substrates that have initiated folding but have not yet membrane integrated (Fig. 6). Given that the catalytic mechanism of OMP assembly by the Bam complex consists of C-to-N-terminal, stepwise templating of β-hairpins (40), substrates lacking an extracellular loop must engage the assembly machinery prior to being recognized as defective. Thus, these results demonstrate that Skp is specifically required to displace and degrade assembly-compromised substrates from the Bam complex.

Fig. 6.

Skp functions to remove stalled OMPs from the Bam complex and facilitates their degradation by DegP. Following translocation across the inner membrane and signal sequence cleavage, diffusion of nascent OMPs to the OM is predominantly mediated by SurA. Unfolded OMP substrates that accumulate in the periplasm or fail to initiate folding in a timely manner (e.g., LamBΔL9) are directly accessible to degradation by DegP. In contrast, degradation of substrates that have stalled while folding (e.g., LamBΔL7) requires prior removal from the Bam complex by Skp. Following OMP substrate displacement, Skp subsequently serves as a sacrificial adaptor protein to enhance DegP-mediated proteolysis and is degraded alongside its bound substrate. Upon completion of a substantial amount of folding (e.g., LamBΔL5), OMP substrates are largely inaccessible to displacement by Skp, as substrate thermodynamic stability exceeds that of Skp–OMP binding.

The high affinity (25) and remarkable stability (30) of Skp–OMP binding lends rationale as to why Skp must be degraded alongside its substrate by DegP. The rapid structural dynamics of the Skp–OMP complex comprise an equilibrium state that is both high in conformational entropy and low in enthalpy, enabling a conformational flexibility capable of permitting substrate release upon recognition of OMPs by downstream folding machinery (30). This substrate release from Skp is ultimately limited by differences in local affinity for any particular polypeptide segment, a criterion that is met by assembly machinery through recognition of the β-signal (30). Subsequent folding and insertion of substrates into the OM would enable rapid substrate release from Skp given the substantial thermodynamic stability afforded by OMP folding and integration into a membrane (31). In the absence of coupled folding, however, the relatively weak, slow nature of DegP–OMP binding (29, 36) is unable to facilitate sufficient substrate release from Skp in order to catalyze proteolysis in the absence of external energy. Consequently, the thermodynamic and kinetic limitations imposed by DegP-mediated degradation of a Skp-bound OMP require concerted proteolysis of both chaperone and substrate.

Notably, this observed proteolysis mechanism bears striking parallels to the concerted degradation of the periplasmic chaperone CpxP and misfolded pilin subunits by DegP (53). It is thus tempting to speculate that the requirement for adaptor proteins in DegP-mediated degradation might have evolved as a general mechanism to enable both rapid neutralization of particularly toxic substrates and their regulated proteolysis despite the kinetic and thermodynamic restraints of DegP binding. Indeed, mutations that stabilize the proteolytically active conformation of DegP and enhance the affinity of DegP binding confer dominant lethality under heat-shock conditions due to rogue protease activity (57), suggesting that additional regulatory mechanisms beyond simply controlling DegP proteolytic activity are necessary to viably combat periplasmic toxicity. The relatively rapid, high-affinity nature of Skp–OMP binding (25, 36) thereby ensures both timely neutralization of toxic substrates and enhancement of DegP protease activity in a manner that is regulated by the activity of Skp.

The high-affinity chaperone activity of Skp, which could represent a potential hazard to the cell if left uncontrolled, is carefully regulated by a disorder-to-order transition coupled to trimerization in response to substrate binding (58). In stark contrast, the activity of SurA is only weakly regulated, primarily through minor conformational rearrangements in response to substrate binding (59–61), effectively enabling largely constitutive chaperone activity. Importantly, these distinct differences in activation mechanism provide a potential means by which the chaperone activity of SurA could predominate over that of Skp under nonstress conditions despite its relatively weak binding to OMPs. Additionally, this work underscores the need for precise regulation of Skp chaperone activity, as rogue removal of OMPs that are in the process of folding on the Bam complex would almost certainly be deleterious. Rather, the activation mechanism of Skp (58) suggests that under conditions in which OMP assembly is compromised, prolonged dwelling of substrates on Bam likely promotes Skp monomer association, triggers subsequent trimerization, and allows for removal of stalled OMPs from assembly machinery, thus ensuring that the Skp-mediated displacement of OMPs from Bam is kinetically limited. Crucially, this mechanism of Skp-mediated disengagement of OMPs from assembly machinery implies that Skp—and not DegP, as previously suggested (44)—ultimately imposes the selective pressure for OMPs to rapidly membrane integrate, as substrates that fail to undergo membrane integration in a timely manner require removal from the Bam complex by Skp prior to their degradation by DegP. Moreover, the capability of Skp to mediate both displacement and subsequent degradation of folding-defective substrates explains the observed differential effects of surA and skp deletion on the assembly of OMPs harboring mutations in the β-signal (62).

Importantly, the ability to clear stalled substrates from the OM assembly machinery represents an additional role for Skp in OMP biogenesis and does not preclude its function as a chaperone under nonstress conditions. Rather, it suggests a more stringent mechanism of substrate release in which productive folding is likely necessary to remove OMPs from Skp in the absence of external energy. Only substrates that cannot complete proper folding are thereby subject to displacement from the Bam complex, thus imposing a quality control mechanism by which efficient OMP folding is ensured by the activity of Skp. Given that the direct interaction between Skp and DegP is relatively weak (29, 36)—likely to minimize premature degradation of Skp–OMP complexes—and that DegP binding is kinetically limited (36), it is possible that a single displaced OMP might undergo multiple attempts at productive folding before being subjected to proteolysis, thus ensuring that only terminally defective OMP substrates are ultimately degraded.

The results presented here solidify the notion that Skp must interact in close proximity, and perhaps even directly, with the Bam complex, especially under conditions in which OMP assembly is compromised. Given that the folding of β-barrel substrates by the Bam complex is catalyzed by the interior surface of the BamA barrel (41), Skp must interact within close proximity of the BamA lumen in order to actively disengage assembly-compromised substrates. Although previous attempts to productively cross-link Skp to the Bam complex have been unsuccessful (20), this is not particularly surprising given the wealth of information establishing SurA as the primary periplasmic chaperone facilitating transport of the bulk mass of OMPs to the OM (19–24). Additionally, because Skp-mediated removal of OMPs from the Bam complex is likely kinetically limited, Skp may only transiently interact with the assembly machinery under nonstress conditions. Despite these limitations, it is possible that utilization of a Skp-dependent, slow-folding OMP substrate similar to those identified in this work may indeed enable productive cross-linking of Skp to the Bam complex.

It may seem paradoxical that the removal of a quality control system designed to disengage assembly-compromised substrates from the Bam complex does not impose significant consequences for cellular viability. However, tolerance to reduced Bam complex function is surprisingly robust under laboratory conditions, as evidenced by a roughly 90% reduction in BamA expression producing negligible effects on growth and viability (63, 64). For this reason, we think it is unlikely that any single slow-folding OMP substrate would compromise Bam complex function sufficiently to produce Skp-dependent growth defects under laboratory conditions. Intriguingly, Skp is required for full virulence in Salmonella and Yersinia murine infection models despite being completely dispensable for growth under laboratory conditions (65, 66), suggesting a crucial role for Skp in vivo. Given that E. coli Skp function is similarly dispensable for growth under laboratory conditions (20, 24), we speculate that Skp-mediated degradation of stalled OMP substrates may be critical in environments where the assembly of many β-barrel substrates is likely to be less efficient, including in vivo contexts.

Given that the OMP quality control mechanisms imposed by Skp are restricted to substrates that have not yet membrane integrated, one critical outstanding question is what, if anything, monitors OMP assembly beyond the point of membrane integration. Consistent with previous observations for BamA (40), late-stalled LamB substrates are largely stable () and thus are likely not subject to proteolytic quality control. These observations lend the question of whether membrane integration represents a point of absolute commitment of assembly machinery to the folding of a particular substrate or whether additional quality control mechanisms capable of disengaging membrane-integrated substrates from the Bam complex exist. Additional studies investigating the quality control mechanisms that monitor late-stage folding by the Bam complex will be necessary to construct a complete understanding of the catalytic mechanism of β-barrel assembly.

Materials and Methods

Bacterial Strains and Growth Conditions.

All bacterial strains and plasmids used in this study are provided in the , respectively. Strains were constructed using standard microbiological techniques, as previously described (67). All plasmids were constructed using Gibson assembly or site-directed mutagenesis. Strains were grown in lysogeny broth (LB)supplemented with 20 mg/L chloramphenicol as appropriate. All strains were grown at either 30 or 37 °C as indicated. Deletion alleles originated from the Keio collection (68) and FLP recognition target–flanked kanamycin resistance cassettes were excised using the FLP recombinase as described previously (69). All LptD constructs contained a C-terminal 3×FLAG-tag (48), and all LamB constructs contained a 3×FLAG-tag flanked by 3× glycine residues (GGGDYKDHDGDYKDHDIDYKDDDDKGGG) inserted into extracellular loop four between codons 155 and 156 (70, 71). lptD-3×FLAG and lptD alleles were inserted at the native lptD locus through recombineering using Collinsella stercoris phage RecT (CspRecT) as described (72).

Efficiency of Plating Assay.

Strains expressing 3×FLAG-tagged LptD or LptDY721D from the native lptD locus were grown overnight in LB medium at 30 °C, 10-fold serially diluted, and replica plated onto Mueller Hinton II agar supplemented with 575 mg/L bacitracin or 90 mg/L vancomycin as indicated. Plates were incubated overnight at 37 °C.

SDS-PAGE and Immunoblotting.

SDS-PAGE was performed using 10% hand-cast polyacrylamide gels with Tris/glycine/SDS running buffer. For Skp blots, SDS-PAGE was performed using Novex 16% tricine precast gels (Invitrogen) and Tris/tricine/SDS running buffer. Proteins were transferred to nitrocellulose membranes (GE Healthcare). Immunoblotting was performed using rabbit polyclonal antisera that recognize OmpA (1:10,000 dilution), Skp (1:8,000 dilution), RpoA (1:50,000 dilution), and BamA (1:10,000 dilution) or a mouse monoclonal antibody recognizing FLAG (1:5,000 dilution; Sigma-Aldrich). Goat anti-rabbit IgG-peroxidase (1:10,000; Sigma-Aldrich) or goat anti-mouse IgG-peroxidase (1:3,000; Bio-Rad) secondary antibodies were used for all immunoblots.

In Vivo Protein Stability Assays.

Strains expressing plasmid-borne variants of 3×FLAG-tagged LptD or LamB were grown overnight at 30 °C and subcultured 1:100 into 10 mL LB medium. The resulting cultures were grown at 37 °C until an optical density at 600 nm (OD600) of about 0.5 to 0.7 was reached, at which point 300 μg/mL spectinomycin was added to each culture to inhibit de novo protein synthesis. 1 mL aliquots of each culture were taken at the indicated timepoints, collected via centrifugation (16,100 × g, 1 min, room temperature), and flash frozen in an ethanol–dry ice bath. Frozen cell pellets were resuspended in a volume (mL) corresponding to the initial OD600 divided by 7 of 2× Laemmli sample buffer (Bio-Rad) supplemented with 10% (vol/vol) β-mercaptoethanol. Samples were boiled for 10 min, and insoluble debris was pelleted by centrifugation (16,100 × g, 2 min, room temperature). Samples were analyzed by SDS-PAGE and immunoblotting as described above.

Analysis of Trimeric LamB Levels.

Strains expressing plasmid-borne variants of 3×FLAG-tagged LamB were grown overnight at 30 °C. Cells corresponding to an OD600 equivalent of 1 (5 × 108 cells) were collected by centrifugation (16,100 × g, 2 min, room temperature) and resuspended in 25 μL 1× BugBuster Protein Extraction Reagent (MilliporeSigma) supplemented with protease inhibitor mixture (1:100; Sigma-Aldrich), Benzonase nuclease (1:100; MilliporeSigma), and 1 M MgCl2 (1:100). Samples were lysed for 20 min at room temperature with gentle agitation. Then, 25 μL of 2× Laemmli sample buffer (Bio-Rad) supplemented with 10% (vol/vol) β-mercaptoethanol was added to dilute samples 1:2. Samples were centrifuged to pellet debris (16,100 × g, 2 min, 4 °C) and analyzed via seminative SDS-PAGE followed by immunoblotting as described above. For simultaneous analysis of total monomeric LamB levels, an aliquot of each sample was boiled for 10 min, centrifuged (16,100 × g, 2 min, room temperature), and analyzed by SDS-PAGE and immunoblotting as described above.

Analysis of LptD Oxidation State.

Strains were grown overnight at 30 or 37 °C as indicated. Cells corresponding to an OD600 equivalent of 1 (5 × 108 cells) were collected by centrifugation (16,100 × g, 2 min, room temperature) and resuspended in 80 μL 2× Laemmli sample buffer (Bio-Rad) that lacked β-mercaptoethanol. Samples were boiled for 10 min, and insoluble debris was pelleted by centrifugation (16,100 × g, 2 min, room temperature). For simultaneous analysis of total reduced LptD levels, an aliquot of each sample was supplemented with 10% (vol/vol) β-mercaptoethanol and boiled for 2 min. Samples were analyzed by SDS-PAGE and immunoblotting as described above.

Membrane Extraction with Urea.

Membrane extraction experiments were performed essentially as described (40) with minor modifications, as detailed in the .