Research article

Interaction between bacterial outer membrane proteins and periplasmic quality control factors: a kinetic partitioning mechanism

Si Wu, Xi Ge, Zhixin Lv, Zeyong Zhi, Zengyi Chang, Xin Sheng Zhao


The OMPs (outer membrane proteins) of Gram-negative bacteria have to be translocated through the periplasmic space before reaching their final destination. The aqueous environment of the periplasmic space and high permeability of the outer membrane engender such a translocation process inevitably challenging. In Escherichia coli, although SurA, Skp and DegP have been identified to function in translocating OMPs across the periplasm, their precise roles and their relationship remain to be elucidated. In the present paper, by using fluorescence resonance energy transfer and single-molecule detection, we have studied the interaction between the OMP OmpC and these periplasmic quality control factors. The results of the present study reveal that the binding rate of OmpC to SurA or Skp is much faster than that to DegP, which may lead to sequential interaction between OMPs and different quality control factors. Such a kinetic partitioning mechanism for the chaperone–substrate interaction may be essential for the quality control of the biogenesis of OMPs

  • DegP
  • fluorescence resonance energy transfer (FRET)
  • outer membrane protein
  • single-molecule detection
  • Skp
  • SurA


In the Gram-negative bacterium Escherichia coli, the biogenesis of the OMPs (outer membrane proteins) involves many steps, including the translocation of OMPs through the cytosol, the inner membrane and the periplasmic space, before the OMPs are folded and assembled into the outer membrane [13]. Among these steps, going across the periplasmic space is considered to be highly challenging owing to the aqueous environment and the fluctuating nature of the periplasm as a consequence of the high permeability of the outer membrane. At present, the translocation and protection of the nascent OMPs in the periplasmic space is not well understood. In vitro investigations have shown that many of the OMPs are capable of folding and inserting into the lipid bilayers spontaneously with different folding kinetics and efficiencies [46]. However, the efficient transportation of OMPs across the periplasmic space needs assistance from some facilitators in vivo [13].

Skp has been identified as a periplasmic chaperone that participates in the translocation of OMPs through the periplasmic space of E. coli [7,8]. The Skp protein binds to OMPs with remarkable selectivity [7] and interacts with the nascent OMPs at an early stage when OMPs are translocated through the inner membrane [9]. Skp prevents the nascent OMPs from aggregating in the periplasmic space by forming soluble periplasmic intermediates with them [8]. In vitro studies have shown that the Skp trimer forms a 1:1 complex with unfolded OmpA, OmpG and YaeT, and the dissociation constant was reported to be in the range of tens of nanomolar [10]. Skp was also found to accelerate the folding kinetics of OmpA and to increase its membrane insertion efficiency under in vitro conditions [11,12].

SurA, a major chaperone protein, is also located in the periplasm, assisting the folding of OMPs in the periplasmic space [13,14]. The level of OMPs in a strain with single deletion of surA is decreased in the outer membrane, demonstrating the participation of SurA in the translocation and folding of OMPs [14]. Although SurA has two peptidyl-prolyl domains, these domains are dispensable for the chaperone function of SurA [15]. SurA interacts with the Ar-X-Ar (Ar is an aromatic amino acid, and X is any amino acid) amino acid motif, which is prevalent in the OMPs, with affinities in the micromolar range as revealed by phage display and isothermal titration calorimetry studies [16,17].

The periplasmic protein DegP has been reported to exhibit both protease and chaperone activities [18], and is also suggested to be involved in the biogenesis of OMPs [1922]. Double-deletion of surA and degP leads to a lethal phenotype at normal temperature, suggesting that DegP may play a role in the translocation of OMPs [22,23]. Structural studies indicate that DegP alone exists as an inactive hexamer [24]. Previous work from our laboratory has shown that DegP is activated via formation of large cages, both 12- and 24-mers, upon its binding to the unfolded substrate proteins [25]. The formation of complexes of 12- and 24-meric DegP with substrates was independently reported by Krojer et al. [19], and suggestions on the mechanism for the protease-chaperone regulation of DegP were given based on the structural information.

These studies are undoubtedly very important in understanding the function of these quality control factors in the biogenesis of OMPs. However, the debate continues regarding the pathways in which SurA, Skp and DegP are involved, and the exact roles they play in safeguarding OMPs as they traverse the periplasmic space [13]. Revealing these aspects of the proteins will be greatly helped by looking into the interactions of these periplasmic quality control factors with OMPs. The present study was conducted in an attempt to reveal the interaction between OMPs and the three quality control factors involved in the translocation of OMPs across the periplasmic space. For this purpose, we used immunoblotting to analyse the level of OMPs in cells with two of these three genes deleted and probed the protein interactions, especially the binding kinetics between the unfolded OMP OmpC and SurA, Skp and DegP by using FRET (fluorescence resonance energy transfer) and single-molecule detection. Our observations reveal that OMPs may interact with these periplasmic quality control factors via a kinetic partitioning mechanism, in which SurA and Skp interact with OMPs in a fast step and protect the newly synthesized unfolded OMPs by forming protease-resistant structures and conveying them through subsequent quality control steps.


Bacterial strains, cell growth and immunoblotting analysis

All bacterial strains were grown in liquid LB (Luria–Bertani) medium at 37 °C with antibiotics added according to their resistance. For subculturing, the JGS276 [MC4100 Δskp ΔdegP] strain was diluted 1:1000-fold into LB medium after overnight culture. The JGS199 [MC4100 ΔsurA ΔdegP; λ(pBADsurA)], JGS200 [MC4100 Δskp ΔsurA; λ(pBADsurA)] and JGS272 [MC4100 Δskp ΔsurA degP(S210A); λ(pBADsurA)] strains were first grown in the presence of 0.2% L-arabinose overnight and washed twice with fresh LB medium before being diluted 1:2000, 1:10000 or 1:2000 into LB medium without arabinose to get ΔsurA ΔdegP and Δskp ΔsurA strains respectively. All cells were grown further for 6 h (MC4100 and JGS276) or 8 h (JGS199, JGS200, and JGS272) prior to immunoblot analysis. To overexpress DegP and DegP(S210A) in the MC4100 cells, the pACYC184-derived plasmids pACYC184p-DegP-His6 and pACYC184p-DegP(S210A)-His6 (see Supplementary Experimental section at for the plasmid construction) were transformed by electroporation. Samples containing an equal amount of cells with different genotypes were heated at 95 °C for 10 min in 2× loading buffer before being subjecting to SDS/PAGE analysis. The immunoblotting analysis was performed according to standard protocols [26]. Rabbit anti-OmpC, anti-OmpA, anti-DegP and mouse anti-OmpF polyclonal antibodies, and a mouse anti-GroEL monoclonal antibody were used as the primary antibodies. AP (alkaline phosphatase)-conjugated rabbit anti-mouse and goat anti-rabbit secondary antibodies were used and visualized by adding NBT (Nitro Blue Tetrazolium; Sigma) and BCIP (5-bromo-4-chloroindol-3-yl phosphate; Promega) according to the manufacturer's instructions.

Protein expression and purification

The pET-28a vector carrying the His-tagged DegP(S210A), Skp, SurA and OmpC were transformed into BL21(DE3) cells. Cells were grown in LB medium containing 50 μg/ml kanamycin at 37 °C. IPTG (isopropyl β-D-thiogalactopyranoside, 0.5 mM) was added after the D600 reached 0.6. All His-tagged proteins were purified by affinity chromatography using Ni-NTA (Ni2+-nitrilotriacetate) resin (GE Healthcare) according to the manufacturer's protocol. The proteins were finally dialysed into 50 mM sodium phosphate buffer containing 100 mM NaCl (pH 7.0). Urea (8 M) was included in the lysis buffer for the purification of OmpC in order to dissolve the inclusion bodies containing the unfolded OMP.

Fluorescence labelling

Fluorescence labelling was performed by incubating the proteins with a 5-fold molar excess of monoreactive NHS (N-hydroxysuccinimide) ester-modified Cy3 (indocarbocyanine; GE Healthcare), Cy5 (indodicarbocyanine; GE Healthcare), BODIPY 493/503 [boron dipyrromethene (4,4-difluoro-4-bora-3a,4a-diaza-s-indacene) 493/503; Invitrogen] or AF488 (Alexa Fluor® 488; Invitrogen) in a 0.1 M NaHCO3/Na2CO3 buffer (pH 8.3) at 25 °C for 3 h. Urea was also added to the reaction mixture to a final concentration of 8 M for labelling OmpC to prevent its aggregation. The free dye was removed using a PD-10 desalting column (GE Healthcare), using a 50 mM sodium phosphate buffer containing 100 mM NaCl (pH 7.0) to elute the labelled proteins (with 8 M urea added in the buffer for eluting unfolded OmpC). The dye-to-protein molar ratios of the eluted protein samples were all determined by dividing the molar concentration of the dye by that of the particular protein after correction according to the product instructions. The labelling ratios are 1.2 for the SurA–AF488 monomer, 1.5 for the Skp–AF488 trimer, 4.0 for the DegP(S210A)–AF488 hexamer, 2.7 for the OmpC–Cy3 monomer, 4.1 for the DegP(S210A)–Cy5 hexamer, 1.5 for the SurA–BODIPY monomer and 2.0 for Skp–BODIPY trimer. The secondary structures of the labelled proteins were found to be similar to the unlabelled proteins, as indicated by their highly comparable far-UV (195–260 nm) CD spectra (Supplementary Figure S1 at The labelled proteins retained their function of preventing the unfolded OmpC from precipitation, as was demonstrated by the light scattering assay and the size-exclusion chromatography (Supplementary Experimental section, and Supplementary Figures S2 and S3 at

Stopped-flow fluorescence measurement

The kinetics of interactions between chaperones and unfolded OmpC were measured by using stopped-flow equipment (Bio-logic) consisting of an SFM-300 mixer and a MOS450/AF-CD optical detection system equipped with a 150 W mercury/xenon lamp. The fluorescence donor (AF488) intensity was recorded with 488 nm excitation and the emission band pass filter was 510±15 nm. The AF488-labelled SurA, Skp or DegP(S210A) and Cy3-labelled OmpC in 8 M urea were mixed to a volume ratio of 20:1 to dilute the urea and to achieve a non-denaturing environment. The final concentration of each labelled protein component was 0.3 μM in the reaction system. In the three-component kinetic measurement, the unlabelled 5-fold excess of DegP(S210A) was first mixed with AF488–SurA or AF488–Skp before being loaded into the sample chamber. A similar experiment was carried out in order to directly monitor the interaction between OmpC–Cy3 and DegP(S210A) in the presence of SurA or Skp, in which AF488-labelled DegP(S210A) and unlabelled SurA or Skp were used. The normalized binding curves were then fitted by using a second-order reaction equation to obtain the kinetic parameters. The kinetic curves representing the slow binding process were fitted to a single exponential function to obtain the half time. The errors presented in the kinetics data are the fitting errors of the curves. The relative error was found to be 19% by repeated experiments on different days.

Single-molecule FRET measurement

Single molecule FRET measurements were carried out by using a home-built confocal microscope that has been described previously [27]. An inverted fluorescence microscope (TE2000-U, Nikon) was equipped with a ×100 objective [N.A. (numerical aperture)=1.4, oil; Nikon] and a 532 nm laser for excitation. The fluorescence emission from each sample was collected via the same objective and split into a donor and an acceptor channel by utilizing a Di650 dichroic mirror (Semrock) before being focused on to two avalanche photon diodes (SPCM-AQR-15; PerkinElmer Optoelectronics) with appropriate fluorescence filters. The sample was prepared by directly mixing Cy3-labelled DegP and Cy5-labelled OmpC, or by initially mixing Cy3-labelled DegP(S210A) and SurA before adding Cy5-labelled OmpC, or by initially mixing Cy3-labelled DegP(S210A) and Cy5-labelled OmpC before adding SurA, in which the final concentration of DegP(S210A)–Cy3 and OmpC–Cy5 were both 1×10−7 M, whereas that of SurA was 5×10−7 M. Such samples were subsequently diluted 2000-fold before being subjected to single-molecule FRET measurement. Surface absorption of proteins for the above measurements was prevented by adding 0.01% Tween 20 (Sigma) into the sample buffer (50 mM phosphate containing 100 mM NaCl, pH 7.0). The fluorescence of the labelled proteins or complex that freely diffused across the focal volume (approximately 1 fl) was simultaneously collected in the donor channel (565–650 nm) and the acceptor channel (670–735 nm) with the bin time chosen as 1 ms. The FRET efficiency was calculated as Iacceptor/(Idonor+Iacceptor) of each selected burst, and was used to generate the statistical FRET distribution histograms.

Single-molecule fluorescence coincidence detection

The setup for single-molecule fluorescence coincidence detection was similar to that of single-molecule FRET measurements, except that different fluorescence filters were used in each channel and a 488 nm laser (Spectra-Physics) was used as the excitation light source. For coincidence detection, BODIPY 493/503–SurA (or BODIPY 493/503–Skp), Cy5–DegP(S210A) and Cy3–OmpC were mixed at submicromolar concentrations, incubated for 10 min to reach equilibrium, and diluted to 10–50 pM. The fluorescence coincidence bursts of BODIPY and Cy5 were recorded in parallel by 510/30 filter and 692/40 filter (Semrock) after the fluorescence emission of the sample was split via the utilization of a Di555 dichroic mirror (Semrock). The simultaneous fluorescence bursts detected in both the BODIPY and Cy5 channels were counted. In this experimental design, fluorescence energy was also transferred from OmpC–Cy3 to DegP(S210A)–Cy5 under the direct excitation of Cy3 by the 488 nm laser. However, such a fluorescence burst in the Cy5 channel could be excluded if no signal was simultaneously detected in the BODIPY channel. The total number of real coincidence events, representing the formation of the SurA–OmpC–DegP(S210A) or Skp–OmpC–DegP(S210A) ternary complex, was calculated according to a method described previously [28]. The control experiment was performed by replacing BODIPY–SurA (or BODIPY–Skp) with BODIPY–BSA, or by using the two-component system in the above measurement.


SurA or Skp prevents the cleavage of the translocated OMPs by DegP

In an attempt to reveal the functional relationship of SurA, Skp and DegP in translocating OMPs through the periplasmic space, we started by examining via immunoblotting the level of OMPs in the whole cell of three different strains with double-deletion of the skp, surA and degP genes. The results showed that the level of OMPs (as represented by OmpA, OmpC and OmpF) was decreased in the skp surA double-deletion cells (Figure 1, lane 4), whereas those in the skp degP or surA degP double-deleted strains were highly comparable with that in the wild-type cells (Figure 1, lanes 1, 2 and 3). On the other hand, the strain with the surA skp double-deletion and with the mutation of the chromosomal degP gene to protease-deficient DegP(S210A) did not exhibit an obvious decrease in the level of OMPs (Figure 1, lane 5), indicating that it was the protease activity of DegP that caused the decrease in the level OMPs in the surA skp double-deleted mutant strain. To show that the decrease in OMPs in the surA skp double-deletion mutant was due to the lack of SurA and Skp instead of an up-regulated expression of DegP, we overexpressed DegP using a pACYCp-DegP vector in the wild-type cells, and no obvious decrease of OMPs was observed (Figure 1, lane 6), even though the DegP level in this strain was much higher than that in the strain with both surA and skp deleted (Figure 1, lane 4). As a control, OMPs in the wild-type cells overexpressing the protease-deficient DegP(S210A) by the pACYCp-DegP(S210A) vector were also analysed (Figure 1, lane 7). These results indicate that DegP exercises its protease activity to degrade the nascent unfolded OMPs that are not well protected by SurA and Skp during the translocation of OMPs across the periplasm.

Figure 1 The periplasmic chaperones SurA and Skp prevent the degradation of OMPs by DegP

Immunoblot analysis on the level of the OMPs OmpA, OmpC, and OmpF in whole cells with the indicated genotypes grown at normal temperature. The level of GroEL was monitored to indicate equal sample loading in each lane. Wt, wild-type.

The binding of unfolded OMPs to SurA or Skp exhibits a kinetic advantage

To discover the molecular mechanism for the interaction between OMPs and the chaperones, we carried out FRET measurements between OMPs (as represented by OmpC) and SurA, Skp or DegP(S210A) (a protease-defective mutant of DegP), in which SurA, Skp and DegP(S210A) were labelled with AF488, and unfolded OmpC was labelled with Cy3. The CD spectra, light scattering and the size-exclusive chromatography analyses indicated that the labelling process affected neither the secondary structure of the chaperones (Supplementary Figure S1) nor their chaperone activities (Supplementary Figures S2 and S3). We then performed stopped-flow studies to measure the kinetics of the interaction between OmpC and the three chaperones. The direct binding of Cy3-labelled OmpC to AF488-labelled SurA, Skp or DegP(S210A) was monitored via the time-dependent change of the donor (AF488) fluorescence. The results clearly demonstrated that unfolded OmpC bound to SurA (t1/2=58.5±0.7 ms) or Skp (t1/2=18.8±0.3 ms) at a rate approximately 1000-fold higher than to DegP (t1/2=28.3±0.1 s) (Figure 2A). When unfolded OmpC–Cy3 was added into a mixture containing SurA–AF488 and DegP(S210A), or Skp–AF488 and DegP(S210A), we observed a sudden initial decrease followed by a gradual increase in donor fluorescence. This most probably reflects the initial binding of unfolded OmpC–Cy3 to either SurA–AF488 or Skp–AF488 and the subsequent slow binding to DegP(S210A), with the half time of the processes measured to be 6.92±0.06 s and 26.1±0.3 s respectively (Figures 2B and 2C). Additional evidence supporting this conclusion is the observation of a time-dependent decrease in donor fluorescence when DegP(S210A), instead of Skp or SurA, was labelled with AF488 in the above experiment. We found that the reaction half times were 7.28±0.03 s and 27.7±0.3 s respectively, consistent with the above observations (Figure 2D). These results indicated that OmpC–Cy3 was captured by SurA or Skp and then bound to DegP(S210A), rather than dissociating from SurA/Skp into solution. These revelations imply that the binding of unfolded OMPs to SurA or Skp (rather than to DegP) is under kinetic control, and this kinetic advantage prevents their direct binding to and degradation by DegP.

Figure 2 Unfolded OmpC preferentially binds to Skp or SurA and is subsequently bound to DegP

(A) Time-dependent fluorescence curves of AF488-labelled SurA, Skp or DegP(S210A) proteins upon binding to the unfolded OmpC–Cy3. All of the fluorescence curves in this Figure were recorded utilizing a stopped-flow instrument. (B) Time-dependent fluorescence curves of AF488-labelled SurA upon binding to unfolded OmpC–Cy3 in the absence or presence of DegP(S210A). (C) Time-dependent fluorescence curves of AF488-labelled Skp upon binding to unfolded OmpC–Cy3 in the absence or presence of DegP(S210A). (D) Time-dependent fluorescence curves of AF488-labelled DegP(S210A) upon binding to unfolded OmpC–Cy3 in the presence of SurA or Skp. The fit is shown as dark grey lines, from which the reaction half times (t1/2) were obtained.

Ternary complexes containing OMP, SurA or Skp, and DegP are formed

The observations that unfolded OmpC captured by SurA or Skp can bind to DegP (Figure 2) and that there exists an interaction between DegP and SurA or Skp (Supplementary Figure S4 at imply the formation of SurA–OmpC–DegP or Skp–OmpC–DegP ternary complexes during this process. Therefore we designed a three-colour FRET experiment, in which SurA (or Skp), OmpC and DegP(S210A) were each labelled with BODIPY 493/503, Cy3 and Cy5 respectively. FRET between SurA–BODIPY (or Skp–BODIPY) and OmpC–Cy3, and between OmpC–Cy3 and DegP(S210A)–Cy5 were observed in the three-component system by ensemble fluorescence spectroscopy (Supplementary Figure S5 at To directly observe the ternary complexes, we then carried out FRET detection at the single-molecule level, envisioning that a coincident burst of BODIPY and Cy5 in the single-molecule coincidence detection would be possible only when the SurA–OmpC–DegP(S210A) or Skp–OmpC–DegP(S210A) ternary complex was formed (Figure 3A). We mixed the fluorescence labelled SurA (or Skp), OmpC and DegP(S210A) and counted the rate of coincidence occurrence in the BODIPY and Cy5 channels. The coincidence rate detected for the three-component samples (Figure 3B, bars 1 and 4) was revealed to be significantly higher than that of either the two-component systems, i.e. SurA–BODIPY/OmpC–Cy3, Skp–BODIPY/OmpC–Cy3 or Skp–BODIPY/DegP(S210A)–Cy5 (where coincidence signals came from the cross-talk of the two components) or the three-component negative control of BSA–BODIPY/OmpC–Cy3/DegP(S210A)–Cy5 (Figure 3B, bars 2, 3, 5 and 6). These observations demonstrated the existence of the SurA–OmpC–DegP(S210A) and Skp–OmpC–DegP(S210A) ternary complexes in the three-component system. In addition, we calculated the ratio factor R, which was defined as the coincidence burst intensity in the Cy5 channel divided by the total intensity in the Cy5 and BODIPY channels. The result in Figure 3(C) shows that, besides the lower coincidence occurrence, the R value distribution of the control systems were also lower than that of the three-component complex, further confirming that the coincidence signals came from the ternary complex.

Figure 3 Capture of ternary complexes of OmpC, DegP and SurA (or Skp) by single-molecule coincidence detection

(A) Schematic diagram indicating the three-colour single-molecule FRET detection. (B) Coincidence occurrence rate in: 1, SurA–BODIPY+OmpC–Cy3+DegP(S210A)–Cy5; 2, SurA–BODIPY+DegP(S210A)–Cy5; 3, BSA–BODIPY+OmpC–Cy3+DegP(S210A)–Cy5; 4, Skp–BODIPY+OmpC–Cy3+DegP(S210A)–Cy5; 5, Skp–BODIPY+DegP(S210A)–Cy5; and 6, Skp–BODIPY+OmpC–Cy3. (C) Distributions of the ratio factor R in the mixture of Skp–BODIPY, OmpC–Cy3 and DegP(S210A)–Cy5 (dark grey), the mixture of Skp–BODIPY and OmpC–Cy3 (light grey), and the mixture of Skp–BODIPY and DegP(S210A)–Cy5 (white).

Presence of SurA or Skp alters the interaction pattern between OMPs and DegP

We next examined whether there is any difference between the interaction of OmpC and DegP in the presence or absence of SurA (or Skp). For this purpose, the distribution of single-molecule FRET efficiency was measured, which can provide information about the conformational changes in protein–protein interactions. The histogram of FRET efficiency distribution of adding unfolded OmpC–Cy5 into a solution containing DegP(S210A)–Cy3 alone (with the peak at 0.53) (Figure 4A) differed significantly from that of adding unfolded OmpC–Cy5 into a mixture containing SurA and DegP(S210A)–Cy3 (with the peak at 0.39) (Figure 4B). By contrast, the distribution histogram of adding SurA protein into a pre-mixed solution of unfolded OmpC–Cy5 and DegP(S210A)–Cy3 (Figure 4C) was very similar to that of adding unfolded OmpC–Cy5 into DegP(S210A)–Cy3 alone (Figure 4A). The common peak seen here on the left of all of the histograms primarily resulted from the fluorescence signal of the unbound Cy3-labelled DegP(S210A), as well as that of the Cy3-labelled DegP(S210A) bound to the photobleached Cy5-labelled OmpC. A similar effect was also observed when SurA was replaced by Skp in the above analysis (results not shown). Our single-molecule detection demonstrated that the FRET efficiency distribution of unfolded OmpC directly bound to DegP differs from that of unfolded OmpC pretreated with SurA, indicating that SurA (or Skp) induces an alternation in the conformation of OmpC, which may allow OmpC to avoid being acted upon by the protease activity of DegP.

Figure 4 The distribution of single-molecule FRET efficiency for the interaction between DegP(S210A)–Cy3 and unfolded OmpC–Cy5

(A) In the absence of SurA. (B) SurA addition before unfolded Cy5–OmpC was added. (C) SurA addition after unfolded OmpC–Cy5 was added. The distribution histograms are fitted (grey lines) by a Gaussian function for the right peak and a γ-function for the left ‘zero’ peak in each panel, which results from donor-only-labelled protein complex or photobleached acceptor.


The translocation of newly synthesized OMPs through the periplasmic space in Gram-negative bacteria is regarded as an extremely challenging process [1,29,30]. Although several periplasmic chaperones have been identified as playing a role in the translocation of OMPs across the periplasmic space, little is known about how they work and collaborate in assisting the folding of OMPs. To understand more about their working mechanism, biophysical studies that probe the direct interactions in this process between the three related periplasmic quality control factors SurA, Skp, and DegP and their substrate OMPs are very helpful. One of the novel discoveries in the present study is that OmpC binds to SurA and Skp with dramatically different kinetics compared with its binding to DegP. In view of this novel observation, we suppose that the three chaperones interact with the OMPs in a sequential fashion, such that SurA and Skp function at an early stage when the unfolded OMPs are secreted into the periplasm, whereas DegP, as a bifunctional protease-chaperone, functions subsequently after SurA and Skp. Furthermore, the binding affinities between OmpC and the three chaperones, measured using FRET, reveal that the interaction between unfolded OmpC and DegP(S210A) is more stable than that between OmpC and SurA or Skp (Supplementary Figure S6 at The kinetic preference of OmpC binding to SurA and Skp and the thermodynamic advantage of OmpC binding to DegP together depicts a flow chart of sequential interactions that may be crucial for the translocation of OMPs across the periplasm.

The role of DegP in Gram-negative bacteria is known to be essential for cells to grow at high temperatures, and DegP exhibits both chaperone and protease activities in vitro [10]. Previously, DegP was found to form stable complexes with OMPs in their folded state, providing new clues for unveiling its quality control function [19]. Our immunoblotting analysis demonstrates that the OMPs in the surA skp double-deletion strain could be degraded by DegP via a direct interaction. Previous genetics studies have shown a decreased amount of the OMPs in the whole cell, even with a single deletion of either surA or skp [7,22]. It is conceivable that chaperones SurA and Skp may act as folding factors that bind OMPs immediately after the OMPs are secreted from the inner membrane through the Sec machinery. The kinetic partitioning of OMPs to the three quality control factors ensures the protection and partial folding of OMPs, to prevent their excessive degradation by DegP, facilitating effective OMPs biogenesis. Our single-molecule FRET results indicate that SurA may vary the interaction pattern between OMPs and DegP, and that DegP can form a ternary complex with SurA–OMPs or Skp–OMPs. These revelations suggest that, as a quality control factor, DegP may act upon OMPs subsequent to SurA and Skp, either to safeguard the partially folded OMPs for further folding and assembly, or to degrade the misfolded OMPs, depending on whether the OMPs have been properly protected by SurA/Skp or not.

One formerly proposed model regarding the roles of these three quality control factors in the biogenesis of OMPs revealed by genetic evidence is that they work as two parallel pathways: one involves Skp and DegP, whereas the other involves only SurA [22,23]. This hypothesis was primarily proposed in view of the fact that the skp surA and degP surA double-knockout strains show lethal phenotypes, whereas the skp degP double-knockout strain remains viable at normal temperatures [22,23]. However, this model has been questioned by other researchers [3]. The results of the present study imply that DegP not only collaborates with Skp, but also with SurA, which is demonstrated by the observation of an interaction between SurA and DegP, and by the formation of the SurA–OmpC–DegP ternary complex. Nevertheless, the interaction between SurA and DegP is weaker in comparison with that between Skp and DegP (Supplementary Figure S4), suggesting that the Skp protein somehow works more closely with DegP than does SurA, which is consistent with the previous genetic studies [22,23].

An issue that remains undefined is whether SurA and Skp have any preference in the OMPs that they bind. A broad range of substrate OMPs have been identified by proteomics methods to interact with Skp in vivo [31]. By comparing the abundance of OMPs in wild-type and surA-deleted strains, SurA was found to be involved in the biogenesis of FhuA and LptD, as well as in the biogenesis of the major OMPs OmpA, OmpC and OmpF [32]. There seems to be some overlap in OMP substrates for SurA and Skp, and the underlying purpose for having evolved two molecular chaperones upstream of DegP warrants further investigation. Undoubtedly, safeguarding the OMPs going through the periplasmic space is an extremely challenging and complicated task, exploration of which would greatly enhance our understanding of this paradigm of protein quality control.


Si Wu, Xi Ge, Zengyi Chang and Xin Sheng Zhao designed the experiments; Si Wu performed the FRET and single-molecule detection experiments; Xi Ge constructed the plasmids and carried out the biochemical experiments; Zhixin Lv contributed to the chaperone activity detection; Zeyong Zhi set up the instrument for single-molecule detection; and Si Wu, Xi Ge, Zengyi Chang and Xin Sheng Zhao analysed the data and wrote the paper.


This work was supported by the National Natural Science Foundation of China [grant numbers 20733001, 20973015 (to X.S.Z.), 30570355, 30670022 (to Z.Y.C.)]; and the National Key Basic Research Foundation of China [grant numbers 2010CB912302 (to X.S.Z.), 2006CB806508 (to Z.Y.C.), 2006CB910300 (to X.S.Z. and Z.Y.C.)].


We thank Professor Pengye Wang (Institute of Physics, Chinese Academy of Sciences, Beijing, China) and Professor Luhua Lai (Institute of Physical Chemistry, Peking University, China) for allowing us to use their stopped-flow instruments; Professor Thomas J. Silhavy (Department of Molecular Biology, Princeton University, Princeton, NJ, U.S.A.) for providing us with the knockout strains of JGS199, JGS200, JGS272 and JGS276; Professor Xuanxian Peng (School of Life Sciences, Sun Yat-Sen University, China) for providing us with the anti-OmpC antibody; and Professor Sarah Perrett (Institute of Biophysics, Chinese Academy of Sciences, Beijing, China) for the critical reading of our paper prior to submission.

Abbreviations: AF488, Alexa Fluor® 488; BODIPY, 493/503, boron dipyrromethene (4,4-difluoro-4-bora-3a,4a-diaza-s-indacene) 493/503; Cy3, indocarbocyanine; Cy5, indodicarbocyanine; FRET, fluorescence resonance energy transfer; LB, Luria–Bertani; OMP, outer membrane protein


View Abstract