Re-examination of PpiD function in the biogenesis of OMPs

To resurvey the role of PpiD in OMP maturation we analyzed the physiological consequences of both inactivation and overexpression of ppiD in wild-type cells and in the surA and skp mutants, respectively, using phenotypes known to report on OMP biogenesis and outer membrane integrity, such as σ^E activity, resistance of the cells to SDS/EDTA and to the antibiotic novobiocin, as well as the levels of major OMPs in their outer membranes. In contrast to previous work [18] we found that expression of multicopy ppiD from the IPTG-inducible P_trc promoter does not suppress the surA mutant phenotypes but rather interferes with cell growth (data not shown). We therefore used a plasmid (pPpiD) that carries ppiD under control of its natural promoter, which is positively regulated by the classical cytoplasmic σ³²-dependent heat-shock response and by the Cpx two-component system [18, 21].

Consistent with recent observations [20], the inactivation of ppiD in a surA strain did not cause lethality. Also, neither the inactivation nor the overexpression of ppiD in the tested strains had major effects on the activity of σ^E (Figure 1), on the plating efficiencies on 0.5% SDS/0.5 mM EDTA (Table 1), on the sensitivities of the cells to novobiocin, or on the levels of major OMPs in their outer membranes (data not shown). However, as is also the case for strains lacking Skp, PpiD-deficient strains showed slightly retarded growth on plates containing 0.5% SDS and 0.5 mM EDTA. At increased concentrations of SDS (2%) a ppiD skp double mutant even revealed a small (3-to 4-fold) plating defect (Table 1), but showed no major changes in the activity of σ^E and in the amounts of OMPs in the outer membranes of the cells relative to the Δskp single mutant (Figure 1 and data not shown). Thus, loss of PpiD appears to slightly interfere with outer membrane integrity without notably affecting the assembly of OMPs. Together these results suggest that PpiD plays only a minor role, if any, in the biogenesis of OMPs in the strain background used here.

Effects of inactivation and overexpression of ppiD on the Cpx envelope stress response

The σ^E signal transduction pathway partially overlaps with the CpxA/R pathway in sensing and responding to folding stress in the cell envelope [9]. Since ppiD is a member of the Cpx regulon [18] we asked whether the Cpx system would respond to inactivation or increased expression of ppiD. As shown in Figure 1B, inactivation of ppiD had no significant effect on Cpx activity in any of the tested strains, indicating that PpiD is not specifically involved in cell envelope functions that are monitored by the Cpx stress response pathway. In contrast, lack of SurA induced the Cpx response ~4-fold, as is consistent with the involvement of SurA in OMP and pilus biogenesis [20] and with misfolding pilus subunits being sensed by the Cpx signaling system [22]. The presence of ppiD in multicopy led to an about 2-fold induction of the Cpx response in all strains but the surA single and the surA ppiD double mutants. In the surA ppiD double mutant increased expression of ppiD from pPpiD slightly reduced Cpx activity, whereas it showed no significant effect on Cpx activity in the surA single mutant.

ppiD is a multicopy suppressor of the lethal surA skp phenotype

We also asked whether ppiD in multicopy would suppress the synthetic lethality of a surA skp mutant. SurA-depletion strains were constructed by placing the chromosomal surA gene under the control of the IPTG-inducible promoter P _Llac-O1 [23], so that expression of surA could be shut off in the absence of IPTG. As expected, P _Llac-O1 -surA Δskp cells grew poorly without IPTG but normal growth was restored by providing copies of either surA or skp on a plasmid (Figure 2B). Unexpectedly, growth in the absence of IPTG was also restored by ppiD in multicopy (pPpiD), although the colonies grew up slower and remained smaller than those grown in the presence of IPTG. The growth-promoting effect of pPpiD was abolished by the introduction of a frameshift mutation that results in a premature stop at codon 173 of the plasmid-borne ppiD gene (pPpiDfs601). Thus, suppression of surA skp lethality elicited by pPpiD requires the intact ppiD gene. Multicopy ppiD also restored viability of surA skp cells in liquid media (Figure 2C). The P _Llac-O1 -surA Δskp strain ceased growth approximately 3.5 h after transfer into non-permissive media (LB without IPTG) but continued to grow when it carried pPpiD, although with slower rates during the mid- to late logarithmic phase. Western blot analysis indicated that PpiD is present in these cells at levels greater than that of chromosomally encoded PpiD (Figure 2D, lane 3 versus lanes 1 and 2). Thus, increased production of PpiD restores viability of surA skp cells but it does not completely compensate for the growth defect caused by the simultaneous lack of the SurA and Skp chaperones.

Suppression of surA skp lethality does not require the parvulin domain but the membrane-localization of PpiD

The lethal phenotype of surA skp cells has been suggested to result from loss of periplasmic chaperone activity [10]. Consistent with this assumption, we found that the chaperone module of SurA (SurAN-Ct), which is devoid of any PPIase activity [2], is sufficient to fully complement the growth defect of the P _Llac-O1 -surA Δskp strain in the absence of IPTG (Figure 2B). To also dissect the activities and regions of PpiD required for complementation of surA skp lethality, we substituted amino acids G347 and I350 in its parvulin domain with alanine, generating the proteins PpiD^G347A and PpiD^I350A, respectively. These mutations had originally been reported to effectively eliminate PPIase activity of PpiD and to result in the loss of its reported surA complementing function [18]. In addition, we completely deleted the parvulin domain from the protein, resulting in PpiDΔParv (Figure 2A). Only most recently, while this manuscript was in preparation, PpiD and its isolated parvulin domain have been shown to be devoid of PPIase activity [19]. However, because G347 and I350 are located at the peptide binding site of the parvulin domain, it was suggested that substrate binding to this domain is important for the in vivo function of PpiD.

Both mutant proteins, PpiD^G347A and PpiD^I350A, complemented the growth defect of surA skp cells just as well as wild-type PpiD, whereas PpiDΔParv complemented slightly less well in these assays (Figure 2B and 2C). Western blot analysis indicated however, that PpiDΔParv was present in the cells at significant lower levels than plasmid-encoded wild-type PpiD (Figure 2D, lane 5 versus lane 3), suggesting that the protein is less stable. We have confirmed that all three mutant PpiD proteins also restore growth of a ppiD skp surA triple mutant (additional file 2), demonstrating that the surA skp complementing activity does not depend on some residual function provided by chromosomally encoded wild-type PpiD. Together, these results show that the parvulin domain is not required for PpiD to function in rescuing surA skp cells from lethality. Unfortunately, we were unable to assess meaningfully if the N-terminal region of PpiD which shows sequence similarity to a substantial portion of the chaperone domain of SurA ([16–18] and additional file 1) contributes to this function, as a protein lacking the respective region (PpiDΔ69-201, Figure 3A) was present in the cells at even lower levels than PpiDΔParv (Figure 3D, lanes 7 and 8).

Finally, we asked if PpiD must be anchored to the inner membrane to function in vivo. Neither production of soluble N-terminally His₆-tagged PpiD (PpiDΔTM) at a level similar to that of PpiDΔParv nor its production from pASKssPpiD at different inducer concentrations restored growth of surA skp cells (Figure 2, and data not shown). pASKssPpiD has also been used to produce and purify soluble His₆-PpiD from the periplasmic fraction of E. coli, thus confirming the periplasmic location of the protein. As soluble His₆-PpiD is functional in vitro (see below and [24]), these results suggest that the function of PpiD in vivo requires the protein to be anchored in the inner membrane.

Overproduction of PpiD lowers folding stress in the cell envelope of surA skp cells

Previous studies suggested that the lethal phenotype of a surA skp mutant is caused by severe protein folding stress in the periplasmic compartment of the cells [10, 25]. To determine whether increased PpiD levels restore viability of surA skp cells by counteracting folding stress in the cell envelope, we monitored the activities of the σ^E and Cpx stress pathways over time once growth of P _Llac-O1 -surA Δskp cells had leveled off in the absence of IPTG (time interval indicated in Figure 2C). At this time point, SurA was hardly detectable in the cells (Figure 3B), indicating that SurA had efficiently been depleted from the cells. During the course of the depletion of SurA in Δskp cells both the Cpx pathway and, as also reported previously [26], the σ^E-dependent pathway were strongly induced (Figure 3A). The σ^E and Cpx activities were 4- to 6-fold increased in SurA-depleted Δskp cells (surA skp pASK75) relative to those of SurA-depleted wild-type cells (surA pASK75). This is also reflected in further increased levels of DegP (Figure 3B, lane 4 versus lane 2), whose gene is positively controlled by the σ^E and Cpx stress responses [27, 28]. In Δskp cells that overproduced PpiD during the course of SurA depletion, σ^E and Cpx activities were significantly lower, being only 1.5- to 3-fold induced relative to the respective activities in surA cells. Consistent herewith, the level of DegP was lower in these cells than in surA skp cells that not overproduced PpiD but slightly higher than the DegP level in surA cells (Figure 3B, lane 5 versus lanes 4 and 2, respectively). Production of PpiDΔParv during the course of SurA depletion in Δskp cells reduced the σ^E and Cpx activities slightly less effectively and production of soluble His₆-PpiD (PpiDΔTM), which does not rescue surA skp cells from lethality, further induced both stress responses (Figure 3A). Thus, only increased levels of membrane-anchored PpiD proteins dampen the strong response of the σ^E and the Cpx envelope stress signal transduction pathways to the simultaneous loss of SurA and Skp chaperone activity.

Taking advantage of the fact that overproduction of the outer membrane lipoprotein NlpE specifically induces the Cpx response [29] and that the Cpx pathway negatively controls expression of the σ^E regulatory components [30] we also asked whether down-regulation of σ^E activity alone would be sufficient to restore growth of surA skp cells. Indeed, the presence of multicopy nlpE during the course of SurA depletion in Δskp cells led to a further induction of the Cpx response and down-regulated σ^E activity to a similar extent as overproduction of PpiD (see additional files 3 and 4). Overexpression of nlpE even slightly improved cell growth in liquid media but it did not restore growth of surA skp cells on solid plates. Thus, Cpx-mediated repression of σ^E alone is not sufficient to restore surA skp cell viability.

Effect of PpiD overproduction in surA skp cells on OMP biogenesis

The reduction of σ^E activity in surA skp cells elicited by higher levels of PpiD suggests that PpiD in these cells directly or indirectly affects OMP biogenesis. σ^E positively controls the production of small non-coding RNAs, which down-regulate OMP synthesis by translational repression [31], and decreased levels of OMPs in SurA-deficient cells therefore reflect defects in both OMP synthesis and assembly [6]. We asked if conversely, the decrease in σ^E activity in PpiD overproducing surA skp cells correlated with increased levels of the major OMP OmpA. Western blot analysis of crude cell extracts confirmed a slight increase in the level of OmpA in these cells as compared to surA skp cells (Figure 4A lane 5 versus lanes 4 and 6, respectively), suggesting that in the absence of SurA and Skp increased levels of PpiD stimulate OmpA synthesis and/or stability. To substantiate this result and to explore a possible influence of PpiD on OmpA folding in surA skp cells, we examined the consequence of PpiD overproduction on the OmpA folding state during the course of SurA depletion in Δskp cells. The OmpA folding state can be conveniently followed by a shift in the apparent mass on SDS polyacrylamide gels. The folded β-barrel domain of OmpA is stable in 2% SDS and migrates faster than unfolded OmpA if not heat-denatured prior to electrophoresis [32]. OMPs were prepared by gentle lysis to preserve their native conformation [33] and OmpA folding intermediates were detected by western blotting (Figure 4B). In contrast to previous work showing that unfolded OmpA accumulates in surA skp double null cells [26], we found the conditional surA skp mutant to contain significantly reduced levels of both, folded and unfolded forms of OmpA (lanes 4 and 5). This difference may reflect the use of a different SurA depletion strategy or the presence of higher levels of DegP protease activity in the strain used here, or both. In any case, the amount of folded OmpA was clearly increased in surA skp cells that overproduced PpiD (lane 3) and was almost as high as that in surA cells (lane 1). Thus, in surA skp cells both synthesis and folding of OmpA is stimulated by increased PpiD levels. Together, these results suggest that in the absence of both chaperones, SurA and Skp, overproduction of PpiD can, at least in part, counteract the defects in the biogenesis of OmpA and possibly of other OMPs.

PpiD has in vitro chaperone activity

The above findings suggest that suppression of the lethal surA skp phenotype by overproduction of PpiD does not simply result from regulatory events in response to increased PpiD levels but rather from functional complementation of the surA skp caused deficiency. As the defects of the surA skp double mutant are thought to result from lack of periplasmic chaperone activity [10], we asked whether the PpiD and PpiDΔParv proteins provide such an activity by examining their capability to prevent aggregation of thermally denatured citrate synthase, a classic in vitro assay for chaperone function [34]. SurA had previously been shown to possesses this activity [2] and was used as a control. When citrate synthase was thermally denatured in the presence of an 8-fold molar excess of SurA (based on citrate synthase monomer) aggregation was significantly reduced (Figure 5). Chymotrypsinogen A, which served as a negative control, showed no or only minor effects at this concentration. In contrast, an 8-fold excess of PpiD reduced aggregation of citrate synthase significantly, although less effectively than SurA, requiring 2-fold higher concentrations to have roughly the same effect. PpiDΔParv finally, which lacks the PPIase domain (Figure 2A), protected citrate synthase about 2-fold more effectively from aggregation than intact PpiD, being almost as effective as SurA. The observed chaperone effects are not caused by protease contamination, as the same amount of citrate synthase was present in samples taken throughout a 1200 s incubation period with a 16-fold excess of PpiD proteins (as judged by SDS-PAGE and silver stain; data not shown). Thus, PpiD exhibits a chaperone activity that is carried in the non-PPIase regions of the protein. The finding that PpiDΔParv complements the growth defect of a surA skp mutant less well than full-length PpiD (Figure 2C) although it exhibits stronger in vitro chaperone activity (Figure 5) likely relates to the presence of lower levels of PpiDΔParv than of plasmid-encoded intact PpiD in these cells (Figure 2D). The overall chaperone activity provided by PpiDΔParv in the cells may thus be weaker than that provided by the overproduced intact PpiD.

PpiDΔParv complements the growth defect of an fkpA ppiD surA triple mutant

To provide further in vivo evidence for the existence of a chaperone activity of PpiD we took advantage of a phenotype that has previously been shown to be associated with inactivation of ppiD. Such a phenotype is exhibited by an fkpA ppiD surA triple mutant, which displays growth defects during mid- to late exponential phase in liquid culture, while all double mutant combinations including these genes grow normally [20]. The fkpA gene codes for the periplasmic folding factor FkpA, which like SurA exhibits PPIase and chaperone activity [35, 36]. Our complementation analysis showed that both the SurAN-Ct protein, which only exhibits chaperone activity [2], and PpiDΔParv restore growth of the fkpA ppiD surA mutant as well as intact PpiD (Figure 6). This demonstrates that the growth phenotype of the triple PPIase mutant is not due to loss of PPIase activity but to loss of chaperone function. It also shows that PpiD shares this function with SurA and FkpA. As in SurA, the chaperone activity is carried solely in the non-parvulin regions of the protein (PpiDΔParv).

Lack of PpiD confers increased temperature-sensitivity in a degP mutant

The periplasmic protease DegP also acts as a chaperone [15, 37] and the simultaneous lack of DegP and SurA gives a synthetically lethal phenotype [10]. We therefore asked whether similarly a chaperone function of PpiD may be disclosed by the combined deletion of ppiD and degP. DegP-deficient strains display a temperature-sensitive phenotype at temperatures above 37°C [38]. Accordingly, we compared the growth of degP and ppiD single mutants with that of a degP ppiD double mutant at 30, 37, and 42°C (Figure 7). As expected, lack of DegP compromised cell growth above 37°C. In contrast, the ppiD single mutant showed wild-type growth at all temperatures. However, the degP ppiD double mutant was more temperature sensitive than the degP single mutant and grew normally only at 30°C. Thus, degP ppiD mutants show a synthetic conditional phenotype at temperatures greater than 30°C.

PpiD is a SurA-like multidomain chaperone

To date, four representatives of the three major families of PPIases are known to exist in the periplasm of E. coli: the cyclophilin PpiA [39], the FKBP-like protein FkpA [35], and the parvulin-like proteins PpiD [18] and SurA [6–8]. In addition to PPIase activity, SurA and FkpA also exhibit prolyl isomerase-independent chaperone activity [2, 36] and the major function of SurA in the maturation of the integral β-barrel OMPs actually is that of a chaperone [2]. While PpiD has also been implicated in OMP biogenesis, the biochemical activity required for this function was reported to be a PPIase activity carried in its parvulin domain [40]. A chaperone activity has so far not been demonstrated for either PpiD or PpiA.

In this study we for the first time directly demonstrate, both in vitro and in vivo, that PpiD exhibits a PPIase-independent chaperone activity that resides in the N- and/or C-terminal regions of the protein. The parvulin domain of PpiD is neither required for function in vivo nor for chaperone activity in vitro, as a PpiD protein lacking this domain fully complements the growth defect of an fkpA ppiD surA triple mutant and protects citrate synthase from thermal aggregation even more effectively than wild-type PpiD. In addition, these results show that a catalytic prolyl isomerase activity plays no major role for the function of PpiD in vivo. This conflicts with previous results [40] but is consistent with most recent data showing that the parvulin domain of PpiD is devoid of detectable PPIase activity in vitro[19].

The chaperone function of PpiD is most likely carried in its N-terminal region, which shares sequence similarity with the N-terminal region of SurA (see additional file 1A; [16–18]) and thus with a substantial part of the SurA chaperone module [2]. Model structures of this region of PpiD generated by alignment based as well as by automated three-dimensional homology modeling (see additional file 1, C and D) show some deviation from the crystal structure of the SurA chaperone module mainly in the helix 1-helix 2 and the helix 3-helix 4 interconnecting loop regions. Interestingly, whereas the chaperone module of SurA lacks sequence similarity with the structurally homologous chaperone domain of the E. coli cytosolic Trigger factor (TF) [41], the predicted helix 1-loop-helix 2 region of PpiD shows similarity on the amino acid level with the corresponding region of TF (24.1% identity between regions 43-121 and 295-371 of PpiD and TF, respectively; see additional file 1, B and E).

The similarities in sequence and predicted structure between PpiD, SurA and TF suggest that PpiD contains a conserved SurA-like chaperone module. However, for a complete chaperone active module the region of PpiD that would correspond to the C-terminal helix of SurA still needs to be identified. As an integral element of the conserved module structure this helix is indispensable for the stability and activity of SurA [2, 42] and presumably also of other members of this family of chaperones. The C-terminal helix of SurA was originally identified as the stabilizing region of the protein as it is very basic (predicted isoelectric points of 10.5) as compared to the rather acidic N-terminal region (predicted isoelectric point 5.3) [2]. Similarly, the corresponding helix in the chaperone domain of TF is rather basic as opposed to the rest of the module (predicted isoelectric points of 8.4 and 4.7, respectively). Finally, the N-terminal region of PpiD is acidic too (predicted isoelectric point of 4.7) and therefore the single basic region of the protein which is located in the C-terminal domain (amino acids 511-560, predicted isoelectric point of 10) and is predicted to be rich in α-helical secondary structure, would be a primary candidate for the stabilizing region. Taken together, all indications are that PpiD is a membrane-anchored SurA-like multidomain chaperone, which like SurA combines a conserved chaperone module with an inactive parvulin domain. Different from SurA however, PpiD lacks a second active parvulin domain and instead contains a C-terminal domain, whose function remains to be determined.

Role of PpiD in the periplasm

PpiD was previously reported to be redundant in function with SurA in the maturation of OMPs [18]. Our results however, establish that PpiD plays no major role in the biogenesis of OMPs and that it cannot compensate for lack of SurA in the periplasm. In addition, PpiD differs from SurA in that it requires to be anchored in the inner membrane to function in vivo whereas SurA is functional both in a soluble and in a membrane-anchored state (S. Behrens-Kneip, unpublished results). Then again, ppiD in multicopy suppresses the surA skp caused deficiencies. The strong induction of the σ^E and Cpx stress pathways during the course of depletion of SurA from Δskp cells is significantly reduced by simultaneous overproduction of PpiD. This suggests that increased levels of PpiD rescue surA skp cells from lethality by counteracting the severe folding stress in the cell envelope which results from the loss of periplasmic chaperone activity. We cannot formally eliminate the possibility that the observed effects result from a response of (an)other known or still to be identified regulatory pathway(s) to increased levels of PpiD in the inner membrane. Such regulatory mechanisms may, for instance, induce periplasmic protease activity that reduces folding stress by protein degradation. However, they would not readily explain our observation that PpiD overproducing surA skp cells contain higher levels of folded forms of OmpA even though they lack two of three chaperones critical for OMP folding. The third OMP chaperone, DegP, appears to interact preferentially with OMPs that already contain substantial levels of folded structure [15] and would thus be expected to predominantly assist in late steps of OMP folding. Moreover, since DegP levels in surA skp cells are reduced by overproduction of PpiD it seems implausible that DegP is responsible for the observed effect on OmpA folding. This, together with our finding that PpiD has chaperone activity in vitro leads us to suggest that PpiD, when present at sufficient levels, is able to partially compensate for the simultaneous loss of SurA and Skp chaperone function.

But why would PpiD promote the folding of OmpA in a surA skp double mutant but have no discernable impact on OMP folding in the respective surA and skp single mutants? We believe that this effect is due to overlapping substrate specificities but yet distinct roles of these chaperones in the periplasm, as has also been suggested for the SurA and Skp chaperones [5, 26]. Both SurA and Skp interact with unfolded major OMPs [2, 43] and facilitate their biogenesis, yet they cannot functionally substitute one another in the cell (Figure 1 and our unpublished data) and are thought to act in parallel pathways of OMP folding [5, 26]. The peptide binding specificity of PpiD has been shown to overlap with that of SurA but to be less specific [44], suggesting that PpiD is capable of interacting with a broader range of substrates. Thus, while unfolded major OMPs obviously are no preferred substrates of PpiD, they may still effectively interact with PpiD for folding in the absence of the competing chaperones SurA and Skp. In this context it is important to mention, that overproduction of PpiD does not restore viability of a surA degP double mutant (S. Behrens-Kneip, unpublished results). This suggests that, when overproduced in surA skp cells, PpiD compensates for the lack of Skp upstream of DegP in the proposed Skp/DegP branch of protein folding rather than for the lack of SurA. The magnitude of suppression of the surA skp phenotypes elicited by multicopy ppiD and the additive phenotypes of the ppiD degP and skp ppiD double mutants described in this work are in support of this notion. Moreover, direct evidence for partially overlapping functions of the membrane-anchored PpiD and the soluble Skp comes from the recent finding that both proteins interact with OMP polypeptides very early during translocation and promote their release from the SecYEG translocation channel [12, 14, 24].

It has recently been proposed that PpiD is a periplasmic gatekeeper of the Sec translocon responsible for newly translocated OMPs [24]. Our work agrees with and refines this assumption, as it shows that PpiD exhibits the requisite chaperone activity for such a function, that this function is not preferentially directed at folding of OMPs, and that PpiD cooperates with SurA, Skp, FkpA and DegP in mediating protein folding in the periplasmic compartment of the cell. We suggest that the role of PpiD is to assist in the initial periplasmic folding events of many newly secreted envelope proteins.

In the cytosol, the folding of newly synthesized proteins is initiated by the ribosome-associated chaperone TF [45, 46]. Of note, PpiD and TF show some interesting analogies. First, similar to PpiD TF is composed of three domains: an N-terminal ribosome-binding domain, a central FKBP-like PPIase domain, and a C-terminal chaperone module which is structurally homologous to the chaperone module of SurA [41, 47] and, as outlined above, shows sequence similarity with the N-terminal putative chaperone region of PpiD. Second, TF associates with the ribosome to sequester and protect polypeptides just as they emerge from the peptide exit tunnel [46] and this association is crucial for its in vivo function [48]. PpiD on the other hand, is anchored in the inner membrane and interacts with newly translocated polypeptides that emerge from the periplasmic exit site of the Sec translocon [24] and according to our data, the anchoring of PpiD in the membrane is required for its function in vivo. Third, TF is dispensable for cell viability and a deletion of the tig gene confers a discernable phenotype only in combination with a mutation of the dnaK gene for the cytosolic chaperone DnaK [45]. Likewise, lack of PpiD gives a discernable phenotype only in cells with already compromised periplasmic chaperone activity, such as in the fkpA ppiD surA triple mutant and in the degP ppiD and ppiD skp double mutants. Finally, the amino acid sequence pattern of known PpiD binding peptides [44] resembles that of the peptide binding motifs identified for the cytosolic chaperones TF and DnaK, consisting of a central patch of hydrophobic amino acids flanked by positively charged amino acids [49]. Altogether, we speculate that PpiD may represent the periplasmic counterpart of TF. Its fixed localization in the inner membrane not necessarily conflicts with such a function, as it may provide a local enrichment of the binding partners but still allows PpiD to dynamically interact with and cycle on and off its interaction partners by lateral diffusion in the membrane, just as it is the hallmark of TF function on translating ribosomes [50]. Further detailed studies on the interaction of PpiD with both the Sec translocation machinery and translocating polypeptides as well as the analysis of the structure-function relationship in PpiD are needed to ascertain its real function in the periplasm.

Media and growth conditions

Luria-Bertani (LB) media were prepared as described [51]. Ampicillin (Ap), chloramphenicol (Cm), kanamycin (Kan), spectinomycin (Spec), and tetracycline (Tc) were used at final concentrations of 100, 20, 30, 50 and 10 μg ml^-1, respectively. For assaying β-galactosidase activity in cpxP-lacZ reporter strains the medium was buffered with 100 mM sodium phosphate to a pH of 7.0, at which cpxP transcription, which is affected by extracellular alkaline pH, is induced to a medium level [57]. Strains were grown at 37°C with aeration unless noted otherwise.

Strains

Plasmids

Assay of susceptibility to SDS/EDTA

The sensitivity of the strains to SDS/EDTA was determined in plating assays as previously described [2]. The efficiency of plating was calculated from the colony count after incubation at 37°C for 24-48 h. A minimum of three experiments were performed for each strain and condition.

Spot dilution assays

SurA-depletion strains were freshly transformed with the required plasmids and were grown overnight at 37°C in selective LB containing 1 mM IPTG. Overnight cultures were adjusted to an optical density at 600 nm (OD₆₀₀) of 4.0 and 10-fold serially diluted with IPTG-free LB. Ten microlitres of the 10^-1, 10^-3, 10^-5, and 10^-7 dilutions were spotted on LB ± 1 mM IPTG plates supplemented with the appropriate antibiotics and incubated at 37°C for 16-24 h. To test for temperature sensitivity, strains were grown overnight at 30°C in LB and were diluted and spotted on LB plates as described above.

SurA depletion in vivo

SB44452 or SB44997 were freshly transformed with the appropriate plasmids and grown overnight at 37°C in LB/Ap/Kan/Spec (buffered at a pH of 7.0, if required) supplemented with 1 mM IPTG and 0.2% (w/v) maltose to induce expression of the maltoporin LamB. Two milliliters of each overnight culture were pelleted in a microcentrifuge and were washed three times in 2 ml of LB to remove IPTG from the cells. The washed cultures were then diluted to an OD₆₀₀ of 0.01 into 50 ml of LB/Ap/Kan ± 1 mM IPTG. These pre-cultures were grown for 4-5 cell generations with shaking in a gyratory water bath at 37°C and diluted into fresh LB/Ap/Kan ± 1 mM IPTG to an OD₆₀₀ of 0.005. Aliquots were sampled for β-galactosidase assays, for western blot analysis, and for the preparation of OmpA folding intermediates at the indicated time points after the second sub-culturing and processed as described below.

β-galactosidase assays

Eσ^E and Cpx pathway activities were assayed by monitoring the β-galactosidase activity resulting from the expression of the chromosomal σ^E- and Cpx-dependent reporter fusions rpoH P3::lacZ[56] and cpxP-lacZ[57], respectively, in cells growing in LB medium (buffered at pH 7.0 for Cpx assays) at 37°C. Overnight cultures were diluted to an OD₆₀₀ of 0.005 into fresh media and grown with shaking in a gyratory water bath at 37°C. Duplicate samples (0.5 ml) were taken throughout the early exponential phase of the growth curve (OD₆₀₀ = 0.08-0.4) and β-galactosidase activity was measured by the standard assay [53]. Eσ^E and Cpx activities shown in Figure 1 were determined from the slope on the line of a differential plot of β-galactosidase activity in 0.5 ml of culture versus OD₆₀₀ and normalized to the wild-type case. In Figure 3, the average β-galactosidase activity/OD₆₀₀ (Miller Units) was calculated and normalized to that of wild-type. Statistical analysis was performed using a Student's t-test.

Western blot analysis

Whole cell extracts were prepared by resuspending cells in urea protein sample buffer (8 M urea, 200 mM Tris-Base, 200 mM DTT, 2% SDS, 0.02% bromphenol blue) followed by short sonication and heating of the sample to 95°C for 10 min. Extracts from equal numbers of cells were run on SDS-polyacrylamide gels and transferred to nitrocellulose membranes. The membranes were probed with dilutions of rabbit polyclonal antisera raised against SurA (1:10 000), PpiD (1:10 000), DegP (1:20 000), Hsc66 (1:20 000), LamB (1:3000), and with mouse monoclonal antibodies raised against OmpA (1:500), respectively. Alkaline phosphatase conjugated goat anti-rabbit and anti-mouse IgGs (Sigma, 1.10 000 dilutions), respectively, served as secondary antibodies. They were visualized by incubating the blots in reaction buffer (100 mM Tris-HCl, pH 8.8, 100 mM NaCl, 5 mM MgCl₂, 37.5 μg/ml nitro blue tetrazolium, 150 μg/ml 5-bromo-4-chloro-3-indolyl phosphate). Signal intensities were quantified using ImageJ software http://rsb.info.nih.gov/ij/. Hsc66 and MalE were used as the internal standard for each lane. Experiments were repeated a minimum of two times for each strain and condition, and data for one representative experiment are shown.

Preparation of OmpA folding intermediates

During the course of SurA depletion, samples corresponding to an equal number of cells were harvested by centrifugation and immediately frozen in a dry ice/ethanol bath. Folded and unfolded OmpA folding intermediates were isolated by gentle lysis as previously described [33]. Samples were mixed with protein sample buffer (3% SDS, 10% glycerol, 5% β-mercaptoethanol in 70 mM Tris, HCl, pH 6.8), heated to 37°C for 10 min and loaded onto 12.5% SDS-polyacrylamide gels. Electrophoresis was performed at 50 V and OmpA intermediates were detected by Western blot analysis as described above.

Protein purification

N-terminally His₆-tagged PpiD proteins and C-terminally His₆-tagged SurA were produced in E. coli CAG44102 from pASKssPpiD, pASKssPpiDΔParv and pASKSurA, respectively, and purified from the periplasmic fraction by affinity chromatography on Ni²⁺-chelating sepharose as previously described [2]. Protein concentrations were determined by UV absorbance at 280 nm using absorption coefficients of 36 270 M^-1 cm^-1, 26 740 M^-1 cm^-1 and 29 450 M^-1 cm^-1 for His₆-PpiD, His₆-PpiDΔParv and SurA-His₆, respectively, as calculated according to [58].

Analysis of chaperone function in vitro

Effects of PpiD proteins on the thermal aggregation of citrate synthase were determined according to [34]. Aggregation was monitored on a Hitachi F-4500 spectrofluorometer with both excitation and emission wavelengths set to 500 nm at a spectral bandwidth of 2.5 nm. Data points were recorded every 0.5 s.