The Escherichia coli σE extracytoplasmic stress response monitors and responds to folding stress in the cell envelope. A protease cascade directed at RseA, a membrane-spanning anti-σ that inhibits σE activity, controls this critical signal-transduction system. Stress cues activate DegS to cleave RseA; a second cleavage by RseP releases RseA from the membrane, enabling its rapid degradation. Stress control of proteolysis requires that RseP cleavage is dependent on DegS cleavage. Recent in vitro and structural studies found that RseP cleavage requires binding of RseP PDZ-C to the newly exposed C-terminal residue (Val148) of RseA, generated by DegS cleavage, explaining dependence. We tested this mechanism in vivo. Neither mutation in the putative PDZ ligand-binding regions nor even deletion of entire RseP PDZ domains had significant effects on RseA cleavage in vivo, and the C-terminal residue of DegS-processed RseA also little affected RseA cleavage. Indeed, strains with a chromosomal rseP gene deleted for either PDZ domain and strains with a chromosomal rseA V148 mutation grew normally and exhibited almost normal σE activation in response to stress signals. We conclude that recognition of the cleaved amino acid by the RseP PDZ domain is not essential for sequential cleavage of RseA and σE stress response in vivo.
Regulated intramembrane proteolysis (RIP) is ubiquitously conserved mechanism of transmembrane signalling, in which target proteins are processed within or around their transmembrane segments, leading to liberation of their soluble domains from the membrane. The released domains are biologically active molecules such as transcription factors and growth factors (Brown et al., 2000). The unique proteolytic process in RIP is catalysed by a novel class of membrane proteases called intramembrane cleaving proteases (I-CLiPs) (Weihofen and Martoglio, 2003; Wolfe, 2009), which can be classified into four families: S2P (site-2protease) (metalloprotease), rhomboid (serine protease), signal peptide peptidase (SPP) (aspartyl protease) and γ-secretase (Presenilin) (aspartyl protease). Among I-CLiPs, S2P family proteases have been identified in a wide variety of organisms from bacteria to human. Eukaryotic S2P proteases have been known to play pivotal roles in signal transduction pathways for sterol and fatty acid metabolism and ER stress responses. Bacterial S2P proteases are involved in diverse cellular processes including environmental stress responses, cell differentiation, production of pheromones or expression of virulence (Chen and Zhang, 2010). S2P family proteins share conserved HEXXH and LDG zinc metalloprotease active-site motifs located in the membrane-cytoplasm boundary regions of their transmembrane segments.
RseP, the Escherichia coli orthologue of the S2P protease (Brown et al., 2000; Kanehara et al., 2001), plays a critical role in regulation of the σEextracytoplasmic stress response (ESR) (Ades, 2008). σE is an alternative sigma factor responsible for transcription of genes encoding proteins and small RNAs involved in biogenesis and quality control of the cell envelope. In normal growth conditions, σE is kept inactive by tight binding to the N-terminal cytoplasmic domain of RseA (Campbell et al., 2003), a membrane-bound anti-σ protein with an Nin–Cout (type II) transmembrane configuration. Accumulation of misassembled outer membrane proteins (OMPs) is the best-characterized stress cue for the σE stress response (Mecsas et al., 1993; Rouviere et al., 1995). Exposed C-terminal residues of malfolded OMPs directly bind to and activate DegS, a membrane-anchored serine protease, to cleave RseA between Val148 and Ser149 in the periplasmic region (Walsh et al., 2003). This first cleavage (site-1 cleavage) triggers the RseP-catalysed second cleavage (site-2 cleavage) of DegS-processed RseA between Ala108 and Cys109 in the transmembrane segment, resulting in release of the cytoplasmic fragment (1–108) of RseA complexed with σE from the membrane (Alba et al., 2002; Kanehara et al., 2002; Akiyama et al., 2004; Flynn et al., 2004). Finally, the RseA cytoplasmic domain is degraded by the cytoplasmic proteases such as ClpXP (Flynn et al., 2004; Chaba et al., 2007), allowing σE to act in transcription of stress-responsive genes (Dartigalongue et al., 2001; Bury-Mone et al., 2009). RseP normally cleaves RseA only after prior site-1 cleavage. This design prevents stress-independent cleavage of RseA, and thus de-regulated expression of σE-controlled genes. Several lines of evidence suggest that in vivo, DegS (Grigorova et al., 2004), the periplasmic protein RseB (Grigorova et al., 2004; Chaba et al., 2011), the Gln-rich regions in the RseA periplasmic domain (Kanehara et al., 2003) and the periplasmic PDZ domains of RseP (see below) all negatively regulate RseP to prevent unregulated cleavage of intact RseA, either directly or indirectly.
RseP spans the membrane four times with both its N- and C-termini exposed to the periplasmic space (Kanehara et al., 2001; Drew et al., 2002) and has two tandemly positioned PDZ domains [PDZ-N (or PDZ1) and PDZ-C (or PDZ2)] in its periplasmic region between the second and third transmembrane segments (Inaba et al., 2008; Li et al., 2009). The PDZ (named after PSD-95, DLG and ZO-1) domain is one of the most widely distributed protein modules and participates in a variety of cellular processes by binding specific ligands, generally at their C-terminal three to five residues (Jelen et al., 2003). PDZ domains are not completely conserved among the S2P family proteases; some possess one or more PDZ domains whereas others do not, and their functional roles are poorly understood.
Structural studies of the isolated PDZ domains of RseP (Inaba et al., 2008; Li et al., 2009) showed that they assume a canonical PDZ fold, although their primary sequences are circularly permutated (Kinch et al., 2006), which is quite uncommon in eukaryotic PDZ domains (te Velthuis et al., 2011). PDZ-N is implicated in regulating proteolytic function of RseP. First, mutations or deletions in PDZ-N enabled RseP to cleave full-length RseA in a DegS-independent manner (Kanehara et al., 2003; Bohn et al., 2004). Second, strongly de-regulated RseP mutants able to cleave full-length RseA in the absence of DegS were identified in a reporter-based screening method and mapped around the putative ligand-binding region of PDZ-N. Finally, in a reconstituted assay of sequential cleavage of RseA by DegS and RseP using purified, detergent-solubilized components, an RseP PDZ-N mutant protein cleaved full-length RseA independent of DegS (Inaba et al., 2008). These studies suggest that PDZ-N is involved in a regulation of the RseP function.
Recently Li et al. (2009) proposed a new model for sequential cleavage: binding by RseP PDZ-C to the newly exposed C-terminal residue (Val148) of RseA, generated by DegS-catalysed site-1 cleavage, is essential for site-2 cleavage by RseP. Using an in vitro reconstituted system, they showed that the mutational alteration of Val148 of RseA to several charged or dissimilar amino acids crippled site-2 cleavage. Furthermore, they showed that single missense mutations in the putative ligand-binding region of the PDZ-C (I304A) or PDZ-N (I215A) abolished the protease activity of RseP against DegS-cleaved RseA. The crystal structures of isolated PDZ-C showed that it forms a homodimer in a manner that the C-terminal residue of one PDZ-C molecule interacted with the ligand-binding pocket of a neighbouring molecule (Inaba et al., 2008; Li et al., 2009). Li et al. (2009) further demonstrated that the PDZ-C domain possessing the I304A mutation no longer interacts with the C-terminal residue of the neighbouring molecule and that, using a PDZ-C derivative having an RseA peptide (G143KASPV148) at its C-terminus, the C-terminal Val148 residue could bind to the ligand-binding pocket of PDZ-C. It has been speculated that the binding of C-terminal Val148 of the DegS-processed RseA either directly activates RseP by inducing some conformational changes of RseP or by releasing possible inhibitory effects of PDZ-C, or recruits RseA to the proteolytic active site of RseP.
The model of Li et al. (2009) is attractive because it provides a plausible answer to the question of why site-2 cleavage by RseP requires the preceding site-1 cleavage. However, it does not dovetail with published results from in vivo experiments. For instance, RseP variants having alterations of conserved residues in the putative ligand-binding regions of PDZ-N or PDZ-C and even those having a deletion of the entire PDZ-C or PDZ-N domain possess a significant proteolytic activity against a model substrate (Inaba et al., 2008). Also, various C-terminally truncated RseA derivatives having different C-terminal residues still exhibited regulated cleavage by RseP (Kanehara et al., 2003). In this study, we thus evaluated the role of the RseP PDZ domains in the substrate recognition and proteolysis in vivo. Our results suggest that the recognition of the exposed C-terminal residue of DegS-processed RseA by the PDZ domains of RseP makes little, if any, contribution to RseA cleavage in vivo and to OMP-induced σE activation.
RseP PDZ domain mutants retain in vivo proteolytic activity
To examine the roles of the RseP PDZ domains in substrate proteolysis in vivo, we examined the ability of mutations in the putative ligand-binding region of PDZ-N and PDZ-C (I215A of PDZ-N and I304A of PDZ-C), as well as deletion of the entire PDZ-C domain (ΔPDZ-C) to cleave RseA. The I304A mutation substantially reduces the size of the putative ligand-binding pocket of PDZ-C and interfered with ligand binding (Li et al., 2009); the I215A mutation is the corresponding substitution in PDZ-N. Both mutations almost completely block RseA cleavage by RseP in vitro (Li et al., 2009). In vivo examination of RseA cleavage by RseP requires us to be able to manipulate DegS and RseP, both of which are essential. However, as their essential activity is to cleave RseA, both genes can be deleted in a ΔrseA strain background (Kanehara et al., 2002). Additionally, these strains tolerate low-level expression of HA-RseA from a plasmid enabling the following experiments (Kanehara et al., 2001). We expressed the wild type and above mutant RseP's from a plasmid together with N-terminally haemagglutinin (HA) tagged-RseA (HA-RseA) (see Fig. 1A) in a strain deleted for rseP and rseA (Fig. 2A). When HA-RseA was expressed without concomitant expression of RseP, DegS-cleaved RseA accumulated in addition to a small amount of the full-length protein (Fig. 2A, lane 2), as expected for a strain carrying DegS but lacking RseP. Coexpression of wild-type RseP (RseP-His6-Myc: C-terminally His6-Myc-tagged RseP, hereafter referred to as RseP-HM; Fig. 2A, lane 1), but not RseP(E23Q) (Fig. 2A, lane 6), a proteolytically inactive variant, resulted in almost no accumulation of RseA confirming RseP-dependent degradation of RseA. Importantly, DegS-cleaved RseA was hardly detectable when the I304A, I215A and ΔPDZ-C mutant forms of RseP were coexpressed, supporting the ability of theses PDZ mutants to degrade DegS-cleaved RseA (Fig. 2A, lanes 3–5). Anti-Myc immunoblotting showed (Fig. 2A–C, lower panels) similar levels of accumulation of the wild-type and mutant forms of RseP in this and following experiments. Similar results were obtained when we used HA-RseA148 as a substrate, a derivative of HA-RseA that mimics the DegS-cleaved RseA (RseA1–148) (Fig. 2B). We also examined the cleavage of a model substrate, HA-MBP-RseA(LY1)148, in which the cytoplasmic and transmembrane domains of HA-RseA148 had been replaced with the tightly folded maltose-binding protein (MBP) domain and the first transmembrane segment of LacY (lactose permease) respectively (Fig. 2C). We previously showed that a similar model substrate, HA-MBP-RseA(LY1)140 having a shorter periplasmic tail, was cleaved in an RseP-dependent manner and generated a stable cytoplasmic HA-MBP domain-containing fragment (Akiyama et al., 2004). The accumulation of the stable cleavage product enabled us to estimate the cleavage efficiency of the site-2 cleavage. HA-MBP-RseA(LY1)148 was also converted to a smaller fragment that was reactive with anti-HA and anti-MBP (data not shown), dependent on the proteolytic activity of coexpressed RseP. Coexpression of the RseP PDZ mutants also resulted in similar, efficient cleavage of the model substrate. Then, kinetics of HA-MBP-RseA(LY1)148 cleavage was examined by pulse-chase experiments (Fig. 2D). While pulse-labelled HA-MBP-RseA(LY1)148 was stable in the ΔrseP strain (Fig. 2D, lanes 4–6), it was rapidly converted to the CL (cleaved) form (within 6 min) when expressed with the wild-type (Fig. 2D, lanes 1–3) or PDZ mutant forms of RseP (Fig. 2D, lanes 7–12), although slightly slower RseP cleavage was observed with the I304A mutant.
We also examined the effects of the I215A/I304A double mutations and deletions of PDZ-N (ΔPDZ-N) as well as the entire PDZ domain (ΔPDZ-NC) on in vivo proteolysis of HA-MBP-RseA(LY1)148 by RseP (Fig. 2E and F). The results showed that the I215A/I304A (Fig. 2E, lane 3) and the ΔPDZ-N variants (Fig. 2F, lane 4) promoted cleavage of the substrate in a similar manner to wild-type RseP. Coexpression of the RseP variant lacking both PDZ domains (ΔPDZ-NC) was almost ineffective in promoting substrate cleavage. However this RseP variant was very unstable, showing markedly decreased cellular accumulation (Fig. 2F, lane 5). When we expressed RseP(ΔPDZ-NC) from a high-copy-number plasmid to achieve significant cellular amounts of the mutant protein (Fig. 2G), significant cleavage of the substrate was observed. These results suggest that the mutations in the putative ligand binding regions or even the deletions of the PDZ domains have little effect on in vivo proteolysis of RseA by RseP.
The exposed C-terminal residue of the site-1 cleaved form of RseA does not critically affect the RseP cleavage in vivo
Li et al. (2009) proposed that the identity of the C-terminal residue of DegS-processed RseA (Val148 in full-length RseA) is critical for the RseP cleavage. This conclusion derives from their finding that several conserved mutations (e.g. V148A) at this position allowed efficient cleavage whereas other mutations to dissimilar residues (e.g. V148S and V148R) prevented cleavage in vitro, a proposition we evaluate here in vivo. We first examined the behaviour of V148A, V148S and V148R derivatives of HA-RseA in ΔrseA strains additionally carrying ΔrseP and/or ΔdegS (Fig. 3A). Behaviour of full-length wild-type HA-RseA was consistent with previous experiments (Kanehara et al., 2002): it accumulated in the ΔdegS strains (Fig. 3A, lanes 3 and 4); was mostly converted to the DegS-cleaved form of RseA in the strain with DegS and lacking RseP (ΔrseP degS+; Fig. 3A, lane 2), and did not accumulate in strains with both DegS and RseP (Fig. 3A, lane 1), as it is subsequently degraded in the cytoplasm (Flynn et al., 2004; Chaba et al., 2007). Importantly, all of the V148A, the V148S and the V148R mutant RseA proteins were degraded similarly to HA-RseA in the presence of both DegS and RseP, indicating that neither substitution impeded RseP cleavage (Fig. 3A, lanes 5–16). Interestingly, all the RseA substitutions impeded cleavage by DegS, as less of the DegS cleaved band was formed than when the wt RseA protein was cleaved [Fig. 3A, compare lane 2 with lanes 6, 10 and 14; this phenotype was also observed for V148S and V148R in Li et al., (2009)]. In addition to V148, Li et al. (2009) examined the possible role of P147 in proteolysis of RseA and found that mutational alterations of P147 did not affect RseP cleavage. We also found that the P147A and P147S mutations did not affect in vivo DegS and RseP cleavage (Fig. 3A, lanes 17–24). These results show that, in contrast to the previous in vitro observations, the V148S or V148R mutation of RseA does not interfere with RseP cleavage in vivo.
To clarify the contribution of a residue at position 148 of RseA to in vivo RseP cleavage, we prepared a set of variants of a model substrate, HA-MBP-RseA148, with each of the 20 amino acids as their C-terminal residue. HA-MBP-RseA148 is a derivative of HA-RseA148, in which an HA-MBP domain replaces the RseA cytoplasmic domain. Pulse-chase experiments showed that HA-MBP-RseA148 and HA-RseA148 exhibited comparable RseP susceptibility (Kanehara et al., 2002; Akiyama et al., 2004 and data not shown). HA-MBP-RseA148 is a mimic of DegS-cleaved RseA (RseA1–148), thus cleavage of this model substrate is dependent solely on RseP (Fig. 3B, lanes 1, 2). All 20 variants of HA-MBP-RseA148 were expressed in cells lacking chromosomal RseA (ΔrseA), and either with or without RseP (rseP+ and ΔrseP respectively). Variants were examined for RseP-dependent cleavage by anti-HA immunoblotting (Fig. 3B), and results were quantified (Fig. 3C). The CL band corresponding to the HA-MBP-containing RseP-cleaved product is unstable and degraded by some cytoplasmic protease (Akiyama et al., 2004; Koide et al., 2008 and data not shown). Thus, the total amount of UC (uncleaved) plus CL in the RseP+ strain is much less than that of UC in the ΔrseP strain. We thus estimated the cleavage efficiency by comparing the UC amount in the RseP+ strain with that in the ΔrseP strain. Note that, although HA-MBP-RseA received some non-specific cleavage and generated a small amount of a CL-like fragment even in the ΔrseP strain, comparison of the UC bands in the rseP+ and ΔrseP strains allowed evaluation of the RseP-dependent cleavage. All variants were cleaved by RseP at efficiencies almost comparable to the wild-type protein [ranging from 98% for V148 (wild type) to 86% for V148D], indicating that the substitutions of the Val148 residue had only a marginal impact on in vivo RseP cleavage. These results strongly suggest that the newly exposed C-terminal residue of DegS-processed RseA has little contribution to the subsequent proteolysis by RseP in vivo.
The C-terminal residue of the site-1 cleaved form of RseA is important when RseP is solubilized
Our in vivo results differ significantly from the in vitro results of Li et al. (2009) regarding the role of the C-terminal residue of the DegS-cleaved form of RseA in the RseP cleavage. One major difference between experimental conditions lies in whether the enzymes and the substrate are integrally associated with membrane or solubilized in detergent extracts. We therefore investigated the RseP cleavage of the HA-MBP-RseA148 variants with different C-terminal residues in detergent-solubilized conditions. Membranes prepared from the ΔrseP strain expressing these proteins were solubilized with 1% n-dodecyl-β-d-maltoside (DDM) and incubated with or without purified RseP (Fig. 4A). Whereas the wild-type HA-MBP-RseA148 (V148) substrate was converted to the CL from in a time- and RseP-dependent manner, no CL fragment was detected for V148D in the presence or absence of RseP even after 14 h incubation (Fig. 4A). A less complete time-course of three additional variants (V148I, V148L and V148K) indicated that no cleavage was observed for the V148D and V148K variants at 3 h, and that cleavage is RseP-dependent as no cleavage was observed for the proteolytically dead mutation [RseP(H22F), Fig. 4B]. Taken together, these results suggest that the C-terminal residue of the site-1 cleaved form of RseA significantly affects RseP cleavage only in membrane-solubilized conditions.
The PDZ domains of RseP are dispensable for OMP-dependent σE activation
To gain insights into the roles of the PDZ domains in the RseP function in physiological conditions, we constructed strains carrying a chromosomal rseP mutant gene encoding a PDZ-deleted variant protein. In this case, we could not use the ΔrseA strains employed above, as they do not have an authentic stress signalling pathway. Instead, we took advantage of the finding that rseP is not essential in a strain lacking two outer membrane proteins, OmpA and OmpC (Douchin et al., 2006). We used a ΔompA ΔompC strain carrying a σE-dependent lacZ reporter gene (CAG16037) and replaced its chromosomal rseP gene by an rseP(ΔPDZ-C) or rseP(ΔPDZ-N) mutant gene by making use of homologous recombination. Anti-RseP immunoblotting analysis showed that the ΔPDZ-C and ΔPDZ-N mutant strains expressed the respective mutant proteins instead of wild-type RseP (data not shown and see Fig. 5D). The accumulation levels of the ΔPDZ-C and ΔPDZ-N protein appeared to be much lower than that of RseP in the parental rseP+ strain, although quantitative estimations are difficult as our anti-RseP antibodies recognize the PDZ region of RseP.
The ΔrseP strain exhibited slightly slower growth at 42°C while it grew almost normally at 30°C and 37°C as compared with the isogenic rseP+ strain (wt) (Fig. 5A). Interestingly, we found that growth of the ΔrseP strain was severely impaired at 20°C, suggesting that some function of RseP is critical for cell growth at this temperature even in the presence of the ΔompA ΔompC mutations. On the other hand, cell growth of the ΔPDZ-C and ΔPDZ-N mutant strains was comparable to that of the wild-type strain at any temperatures. These results indicate that the PDZ domains of RseP are dispensable to support normal cell growth at least under these conditions.
We then examined the stress-dependent σE activation in the ΔPDZ-C and ΔPDZ-N mutant strains, by overexpressing OmpC. Periplasmic OmpC overproduction is a well-characterized stress cue that strongly induces the σE-pathway stress response (Mecsas et al., 1993). Cells of the rseP+, ΔrseP, ΔPDZ-C and ΔPDZ-N mutant strains carrying OmpC- (pYH114) and LacI- (pSTD343) expressing plasmids were grown at 30°C to a mid-log phase and induced for OmpC expression with IPTG. Growth of the ΔrseP strain ceased about 2 h after OmpC induction (Fig. 5B). In contrast, growth of the other three strains was apparently not affected by OmpC overexpression. The σE activity was assayed by monitoring expression of LacZ from the σE-dependent lacZ reporter gene (rpoHP3-lacZ) at 1 h intervals after OmpC induction (Fig. 5C). The OmpC overexpression in the rseP+ and PDZ mutant strains resulted in similar, substantial increase in the σE activity, while the σE activity was unchanged in the ΔrseP strain. We also examined accumulation of the RseA and RseP proteins in these strains before (3 h) and 1 h after (4 h) OmpC induction by anti-RseA and anti-RseP immunoblotting analysis (Fig. 5D). The DegS-cleaved RseA accumulated in the ΔrseP strain as expected, but this species was hardly detectable in the rseP+ and the PDZ mutant strains for pre-OmpC-induction samples. Similar results were obtained with the post-OmpC-induction samples. The apparent accumulation levels of RseA in the RseP-expressing strains did not change significantly before and after the OmpC induction, while σE was activated in response to the stress. This could be explained by the previous findings that the RseA level is both positively (through the σE-mediated stimulation of transcription of rseA) and negatively (through proteolysis of RseA) regulated under a stressed condition, and the σE activity is not simply correlated with the level of RseA (Ades et al., 2003). These results are consistent with those of the above σE activity assays and indicate that both ΔPDZ-C and ΔPDZ-N mutant proteins has comparable proteolytic activities to wild-type RseP, as suggested from the previous results (Inaba et al., 2008). Note that we observed slightly higher levels of accumulation of DegS-cleaved RseA in the PDZ mutant strains as compared with in the rseP+ strain after OmpC induction (Fig. 5D and data not shown). This might be ascribable to apparently decreased accumulation of the ΔPDZ-C and ΔPDZ-N mutant proteins (Fig. 5D, lower panels). We found that the accumulation level of full-length RseA in the ΔrseP strain was much lower than the strains expressing RseP or its derivatives (Fig. 5D, lanes 2 and 6). One reason for this would be that the RseP deletion caused low cellular σE activity, and thus higher extracytoplasmic stresses such as elevated accumulation of misfolded OMPs, which promoted DegS cleavage of full-length RseA. These results collectively suggest that neither of the PDZ-C and PDZ-N domains of RseP is essential for RseA cleavage in the conventional OMP-dependent σE activation process.
We also constructed strains possessing the rseA(V148S) or rseA(V148A) mutant gene on the chromosome and examined the effects of the RseA V148 mutations. Consistent with the above results, neither mutation affected the cell growth or RseA cleavage by RseP (Fig. 6A and C). Also, the V148A mutation had little influence on the OmpC-dependent σE activation, while the V148S mutation slightly lowered it (Fig. 6B). The lower σE activity in the rseA(V148S) mutant could be due to a somewhat increased accumulation level of the full-length RseA(V148S) protein, especially after prolonged OmpC induction (data not shown), which was probably caused by inefficient DegS cleavage (Fig. 3A). These results further support that recognition of the C-terminal residue of DegS-processed RseA by RseP PDZ is not a prerequisite for RseA cleavage by RseP.
Li et al. (2009) recently proposed, mainly based on their in vitro and structural studies, that the second PDZ domain of RseP, PDZ-C (PDZ2), plays an essential role in the site-2 cleavage of RseA by recognizing the newly exposed C-terminal residue of DegS-processed RseA. Although this model nicely explains the requirement of site-1 cleavage for the RseP-catalysed site-2 cleavage, previous in vivo results (Kanehara et al., 2003; Inaba et al., 2008) are apparently inconsistent with this model. We thus investigated whether this model applies in vivo. Our results indicate that neither the RseP PDZ domain nor the C-terminal residue of the DegS-cleaved RseA is significant for sequential proteolysis in vivo.
The I304A (in PDZ-C) and I215A (in PDZ-N) mutations in the predicted ligand-binding regions of the PDZ domains, which severely impaired the site-2 cleavage of DegS-processed RseA in vitro (Li et al., 2009), had almost no effect, individually or in combination, on the cleavage of RseA and its derivatives in vivo. Moreover, even the deletion of the PDZ-C or PDZ-N domain minimally affected the substrate cleavage. Although the simultaneous deletion of PDZ-N and PDZ-C (ΔPDZ-NC) substantially destabilized RseP, a significant cleavage of a model substrate was observed upon expression of the ΔPDZ-NC mutant proteins from a high-copy-number plasmid. Thus far, direct interaction between PDZ-C and the RseA Val148 residue has not been observed by biochemical methods including an isothermal titration calorimetry assay (Li et al., 2009). It is thus conceivable that the interaction is, if at all, very weak. Taken together, our results strongly suggest that ligand binding by the PDZ domains plays no essential role for the proteolytic activity of RseP in vivo.
Li et al. (2009) suggested that the identity of the exposed C-terminal residue (at position 148) of DegS-processed RseA is critical for the site-2 cleavage as some residues (uncharged residues with similar side-chain length or hydrophobicity to Val) at this position allowed in vitro RseP cleavage whereas others (residues with a positively charged or dissimilar polar side-chain) severely inhibited it. Although they showed that expression of the ‘RseP-uncleavable’ mutants of RseA significantly lowered stress-induced σE activation as compared with that of the wild type and ‘RseP-cleavable’ mutants, how they exerted such effects is unclear as neither the levels of their expression nor efficiencies of their DegS/RseP cleavages had been examined. In this study, by using an RseA-derived model substrate (HA-MBP-RseA148) that mimics the DegS-processed form, we directly examined the effects of systematic substitutions of the C-terminal residue (at position 148) of RseA on the site-2 cleavage and found that the identity of the residue at position 148 little affected substrate cleavage efficiencies in vivo.
Our findings collectively indicate that, in contrast to the previous proposal, binding of the RseA C-terminus to RseP PDZ is not the essential step for the RseP-catalysed site-2 cleavage. Importantly, we found that strains having a mutant chromosomal rseP gene lacking either of the two PDZ domains and strains having a mutant chromosomal rseA gene with an alteration of V148 grew normally and exhibited almost normal RseA cleavage and σE activation in response to an extracytoplasmic stress. Although our results do not rule out the possibility that RseP PDZ binds RseA V148, they show that this binding, if any, would exert very small effects on RseA cleavage and make little contribution to stress-induced σE activation in vivo.
Interestingly, when we used detergent solubilized RseP, we substantially reproduced the Li et al. (2009) findings that the C-terminal residue of the DegS cleaved substrate plays a critical role. It is very likely that interaction between RseP and substrates is more robust in their natural milieu. Our previous results indicate that RseP–substrate interaction is altered under a solubilized condition (Koide et al., 2008), and an apparent lower enzymatic activity of the solubilized RseP protein could also be partly ascribed to its structural alterations. Additionally, integral association of both the enzyme and substrate with the two-dimensional membrane in correct topology restricts their movement and enhances proper interaction. If PDZ-C domain has a weak ability to bind the RseA 148 residue in detergent extracts, it could facilitate interaction between solubilized RseP and substrates although this interaction does not seem to significantly contribute to in vivo RseA cleavage. Although characterizing the behaviour of solubilized membrane-bound enzymes is important for initial studies, our results suggest that it is important to also investigate their behaviour in a membrane-integrated state to understand their physiological functions, especially when both enzyme and substrate are integral membrane proteins, as is the case for intramembrane proteolysis.
Our current results suggest that the possible ligand binding to the PDZ domains is not required for activation of RseP because the PDZ missense and deletion mutants retained almost the full proteolytic activity. It could thus be involved in modulation of the RseP function. Strains carrying the ΔPDZ-N or ΔPDZ-C mutant form of the rseP gene on the chromosome exhibited nearly normal RseA cleavage and σE reporter expression in response to OmpC overproduction, indicating that neither of the PDZ domains is required for the OMP-induced σE activation. One tempting possibility is that the PDZ domains have a role in regulation of a non-canonical signalling pathway. We have previously isolated a number of the RseP mutants that can cleave full-length RseA in a DegS-independent manner (Kanehara et al., 2003; Inaba et al., 2008) and found that many of strong mutations affected putative ligand-binding region of PDZ-N. The existence of these de-regulated mutants led us to assume that the ligand binding to PDZ-N modulates the RseP function in a non-conventional, DegS-independent σE activation process. It would be possible that binding of some still unidentified ligands to PDZ-N induces conformational changes of RseP that enable the full-length substrate to make access to the otherwise recessed protease active site of RseP. The recent finding that Salmonella RseP can cleave full-length RseA in a DegS-independent but RseP-PDZ-dependent manner under an acid stress condition (Muller et al., 2009) also suggest that the PDZ domain is required for a novel signalling pathway. Crystal structures of the RseP PDZ domains showed that, while PDZ-C has an open conformation that allows a ligand binding, the predicted ligand-binding pocket of PDZ-N is partially covered by a short α-helix (residue 208–213), which would interfere with ligand binding (Inaba et al., 2008; Li et al., 2009). However, we found that an RseP derivative having a deletion of this short α-helix cleave full-length RseA independently of the preceding site-1 cleavage like the PDZ-N ligand-binding site mutants (Inaba et al., 2008; Y. Hizukuri and Y. Akiyama, unpubl. data), suggesting a potential role of this short helix in regulation of ligand binding to PDZ-N. Further understanding of the roles and functional mechanism of the RseP PDZ domains would require identification of putative PDZ ligands and structural information on the whole RseP protein including both the PDZ domains and the membrane-embedded domain.
We found that the ΔrseP derivative of a ΔompA ΔompC strain exhibited severe growth defect at 20°C while its growth was only slightly slower at higher temperatures as compared with the wild-type (rseP+) strain. Because the rseP disruption causes reduced cellular σE activity, the observed growth defects could be ascribed to defective σE stress response. However, we found that additional disruption of rseA, which causes constitutive activation of σE, only partially improved growth at 20°C while growth at higher temperatures was almost completely restored (Y. Hizukuri and Y. Akiyama, unpubl. results). Retardation of cell growth by the rseP disruption even in the absence of RseA was also documented previously (Saito et al., 2011). These observations suggest that RseP has an important function other than σE activation at least at a low temperature. Thus far, RseA has been the only established physiological substrate of RseP, but we have recently proposed that RseP is a protease with unexpectedly wide substrate specificity and catalyses cleavage of not only RseA but also various remnant signal peptides (Saito et al., 2011). The observed cold sensitivity of the ΔrseP strains might result from defects in degradation of other physiological substrates including signal peptides. It would be possible that the PDZ-mediated regulation and/or substrate recognition are involved in their cleavage.
Bacterial strains, plasmids and media
Escherichia coli K-12 strains and plasmids used in this work are listed in Tables 1 and 2 respectively. Construction of the individual strains and plasmids are described in Supporting information.
Cells were grown in L medium (containing 10 g of bacto-tryptone, 5 g of yeast extract and 5 g of NaCl per litre; pH was adjusted to 7.2 by NaOH) or M9 medium (Miller, 1972) (with omission of CaCl2). Ampicillin (50 μg ml−1), chloramphenicol (20 μg ml−1), spectinomycin (50 μg ml−1), kanamycin (25 or 12.5 μg ml−1) and/or tetracycline (25 μg ml−1) were added for selecting transformants and transductants and for growing plasmid-harbouring strains.
Immunoblotting, pulse-chase and immunoprecipitation experiments
Immunoblotting using anti-HA (HA-probe (Y-11), Santa Cruz Biotechnology) or anti-Myc [c-Myc (9E10), Santa Cruz Biotechnology] antibodies was carried out as described previously (Inaba et al., 2008). Rabbit polyclonal anti-RseP and anti-RseA antibodies were raised against the PDZ-NC domain of RseP (Inaba et al., 2008) and the His6-tagged RseA cytoplasmic domain respectively. For purification of His6-tagged RseA cytoplasmic domain, 200 ml of pre-cultured BL21(DE3)/pSTD1202 cells was inoculated into 1l of L medium containing 1 mM IPTG and grown at 37°C for 4 h. Cells were collected by centrifugation and disrupted by French Press at 4°C. After removal of membranes by ultracentrifugation, supernatants were loaded onto a Ni-NTA agarose column and bound proteins were eluted with linear 20–500 mM imidazole gradient essentially as described previously (Akiyama and Ito, 2003) except that no detergent was included in the buffers. Before using anti-RseA or anti-RseP antibodies in immunoblotting, they were pre-treated with whole-cell lysates of AD1840 (the ΔrseA ΔrseP ΔdegS strain) for 1 h at 4°C to reduce background. For preparation of the AD1840 cell lysates, cells of AD1840 were collected from an overnight culture, disrupted by French Press and mixed with a 1/10 volume of 10× phosphate-buffered saline containing 1% Tween-20. Antibody-decorated proteins were visualized using Chemi-Lumi-One L kit (Nacalai Tesque) and Fuji LAS3000 mini lumino-image analyser. For pulse-chase and immunoprecipitation analysis, cells were grown in M9 medium supplemented with 20 μg ml−1 each of 18 amino acids (other than methionine and cysteine), 2 μg ml−1 thiamine and 0.4% glucose to exponential phase and induced with 1 mM IPTG and 1 mM cAMP for 30 min. Cells were then pulse-labelled with [35S] methionine for 1 min followed by chase with unlabelled methionine for the indicated periods. Immunoprecipitation was carried out essentially as described (Akiyama et al., 2004).
Analysis of the in vivo proteolytic activities
For in vivo proteolytic activity assays of RseP-HM and its derivatives, cells that had been pre-cultured in L medium containing 0.4% glucose at 30°C were inoculated into M9 medium supplemented with 20 μg ml−1 each of 20 amino acids, 2 μg ml−1 thiamine, 0.4% glucose, 1 mM IPTG and 1 mM cAMP, and grown at 30°C for 3 h. Proteins were trichloroacetic acid-precipitated and analysed by SDS-PAGE and immunoblotting using anti-HA or anti-Myc antibodies as described previously (Akiyama et al., 2004).
In vitro cleavage of HA-MBP-RseA148 and its derivatives under a solubilized condition
Cells were grown in L medium supplemented with 0.4% glucose, 1 mM IPTG and 1 mM cAMP at 30°C for 3 h and mixed with spectinomycin (final concentration of 100 μg ml−1) to stop protein synthesis. Harvested cells were washed with 10 mM Tris-HCl (pH 8.1) and suspended in buffer containing 20% sucrose, 30 mM Tris-HCl (pH 8.1), 1 mM DTT (dithiothreitol) and 1 mM Pefabloc SC (Merck Chemicals). Cells were converted to spheroplasts by incubating at 0°C for 30 min after addition of 1/10 volume of 0.1 mg ml−1 lysozyme (dissolved in 0.1 M EDTA). After sonical disruption of spheroplasts, EDTA (pH 8.0) was added (final concentration, 1.5 mM) and cell debris was removed by low speed centrifugation. Membranes were collected by centrifugation (at 138 k g, 60 min) and suspended in buffer containing 10 mM Tris-HCl (pH 8.1) and 16% glycerol. Subsequently, membranes were solubilized by incubation with 1% DDM at 0°C for 1 h with occasional mixing. After clarification by centrifugation (at 138 k g for 30 min), solubilized proteins were mixed with or without purified RseP-HM or RseP(H22F)-HM (Akiyama et al., 2004) and incubated at 37°C in buffer containing 100 mM NaCl, 50 mM Tris-HCl (pH 8.1), 0.02% DDM, 2.5% glycerol, 5 μM zinc-acetate, 10 mM 2-mercaptoethanol and 1× Protease Inhibitor Cocktail (EDTA-free) (Nacalai Tesque). The RseP proteins were suspended in the buffer containing 10 mM Tris-HCl (pH 8.1), 300 mM KCl, 5% glycerol, 0.02% DDM and this buffer was added in no enzyme control. Samples were mixed with an equal volume of 2× SDS sample buffer and incubated at room temperature for 10 min with vigorous mixing and then at 37°C for 5 min. Cleavage of substrate by RseP was analysed by SDS-PAGE and immunoblotting using anti-HA antibodies.
The σE activity was assayed by monitoring β-galactosidase activity expressed from a chromosomal σE-dependent lacZ reporter gene (rpoHP3-lacZ). The enzymatic activity was measured as described by Miller (Miller, 1972) and expressed in Miller units.
We appreciate National BioResource Project E. coli for the Keio strains (JW0940, JW2203 and JW0198). We are grateful to H. Mori, S. Chiba, S.-i. Narita, T. Nogi and C.A. Gross for helpful advice and stimulating discussion, to C.A. Gross and M.S. Guo for critically reading and editing the Manuscript. We also thank K. Yoshikaie and M. Sano for technical supports. This work was supported by a research grant from the Institute for Fermentation, Osaka (to. Y.A.), JSPS KAKENHI Grant Number 24370054 (to Y.A.), 24·5188 (to Y.H.) and MEXT KAKENHI Grant Number 19058007 (to Y.A.). Y.H. was supported by JSPS Research Fellowships for Young Scientists.