Identification of a New Member of the Phage Shock Protein Response in Escherichia coli, the Phage Shock Protein G (PspG)*[boxs]

The phage shock protein operon (pspABCDE) of Escherichia coli is strongly up-regulated in response to overexpression of the filamentous phage secretin protein IV (pIV) and by many other stress conditions including defects in protein export. PspA has an established role in maintenance of the proton-motive force of the cell under stress conditions. Here we present evidence for a new member of the phage shock response in E. coli. Using transcriptional profiling, we show that the synthesis of pIV in E. coli leads to a highly restricted response limited to the up-regulation of the psp operon genes and yjbO. The psp operon and yjbO are also up-regulated in response to pIV in Salmonella enterica serovar Typhimurium. yjbO is a highly conserved gene found exclusively in bacteria that contain a psp operon but is physically unlinked to the psp operon. yjbO encodes a putative inner membrane protein that is co-controlled with the psp operon genes and is predicted to be an effector of the psp response in E. coli. We present evidence that yjbO expression is driven by σ54-RNA polymerase, activated by PspF and integration host factor, and negatively regulated by PspA. PspF specifically regulates only members of the PspF regulon: pspABCDE and yjbO. We found that increased expression of YjbO results in decreased motility of bacteria. Because yjbO is co-conserved and co-regulated with the psp operon and is a member of the phage shock protein F regulon, we propose that yjbO be renamed pspG.

The phage shock protein operon (pspABCDE) was first characterized in Escherichia coli (1) and is highly conserved in many Gram-negative bacteria including several pathogens. There is good evidence that the psp genes are involved in protecting the bacterial cell during infectious processes. For example, pspC mutants of Yersinia enterocolitica are severely attenuated for virulence during infection (2) and exhibit growth defects when the type III secretion system is expressed (3).
Significantly, the psp genes are among the most highly upregulated genes in Salmonella typhimurium during macrophage infection (4). The psp operon is also up-regulated during swarming in S. typhimurium (5) and during biofilm formation in E. coli (6).
Expression of the psp operon in E. coli is induced by protein IV (pIV), 1 a secretin from filamentous phage f1 (1). pIV forms a pore in the bacterial outer membrane that is required for the assembly and export of filamentous phage (7,8). The pIV protein is the founding member of a large family of bacterial secretins, all of which form large multimeric export channels in the outer membrane. Overexpression of several secretins, often components of the type II and type III bacterial secretion systems, has also been shown to induce expression of the psp operon (e.g. Refs. 7 and 9) establishing that the response is not restricted to a phage protein. Expression of the psp operon can also be induced following overexpression of mutant forms of the outer membrane protein PhoE that are not efficiently secreted (10). PspA synthesis is switched on under conditions that block or reduce the efficiency of the export apparatus, for example, mutants in secA, secD, and secF (10) and depletion of YidC (11,12). Mutations in components of the twin-arginine translocation pathway also leads to PspA induction under anaerobic conditions (Ref. 13 and see also Ref. 12). Other more general stresses including extreme heat shock (50°C), hyperosmotic shock, ethanol treatment (10%), and uncouplers of proton-motive force induce psp (reviewed in Ref. 14). The common factor that may link psp-inducing stresses is their effect in dissipating proton-motive force. Indeed, it is significant that PspA, an effector protein of the phage shock response, is known to be involved in maintaining proton-motive force under stress conditions (15). In addition to Psp protein homologues in other Gram-negative bacteria, a PspA homologue (VIPP1) has been found in Synechocystis, which is thought to be important in thylakoid formation, consistent with a role of PspA in sustaining membrane function (16).
Psp proteins mediate regulation of the psp operon (17,18). Transcription of the psp operon is driven by the 54 -RNA polymerase ( 54 -RNAP) (17), which is activated by the enhancer binding protein PspF (19) and facilitated by integration host factor (IHF) (20). The expression of PspF is negatively autogenously controlled (21). PspA negatively regulates psp transcription by binding to the activator protein PspF (22,23). Conversely, PspB and PspC act as positive regulators of psp operon transcription by overcoming the negative regulation imposed by PspA under specific inducing conditions (e.g. pIV) (17,24,25). Phenotypes of cells lacking the psp operon are very subtle and include reduced survival in stationary phase at alkaline pH and changed motility (14).
Here we have used whole genome transcriptional profiling to determine the global effect of pIV synthesis in E. coli. In the highly restricted response we have identified one new gene associated with the psp system, pspG (previously yjbO). pspG is physically unlinked with the psp operon but is co-conserved and co-regulated with the psp operon genes by 54 , PspF, IHF, and PspA. Several lines of evidence suggest that PspG is an effector of the phage shock system and not a regulator of psp expression.
Microarray Analysis-Growth of cultures was halted with 1/10 volume of 5% phenol in ethanol and RNA was extracted with hot phenol/ SDS (36). RNA was treated with DNase I for 1 h at 37°C. For the initial microarray experiments, RNA was fluorescently labeled during reverse transcription and cDNA was hybridized to E. coli PCR product microarrays according to S. Kustu and co-workers (37). Hybridization, scanning, and normalization were carried out as described (38) and genome images were prepared (37). Experiments were performed in duplicate with a dye swap. The microarray data for the pIV experiments were generated with fluorescently labeled genomic DNA as a reference channel in each experiment using E. coli and S. typhimurium PCR product microarrays printed at IFR (39 -41). Experiments were performed in quadruplicate, consisting of two biological replicates and two technical replicates. Microarray slides were scanned with a Genepix 4000B scanner (Axon Instruments). Fluorescent spot and local background intensities were quantified using Genepix Pro software. For labeling, hybridization, and data analysis protocols and details of statistical filtering procedures, see the online site (ifr.bbsrc.ac.uk/Safety/Microarrays/#Protocols). Further statistical analysis was carried out using Cyber-T (visitor.ics.uci.edu/genex/cybert/).
Western Blotting-Western blotting was carried out as described (23) using primary antibodies to PspA (12) and pIV (a gift from Marjorie Russel). pIV antibodies were used at a 1:10,000 dilution with donkey anti-rabbit secondary antibodies (Amersham Biosciences).
Bioinformatics Methods-Fuzzpro (EMBOSS programs) was used to search for consensus sequences in regions of DNA by allowing small numbers of mismatches to be introduced to the search.
Motility Assay-Motility assays were carried out using motility agar (1% tryptone, 0.5% NaCl and 0.3% agar) plus the appropriate antibiotic. 2 l of a fully grown LB overnight culture was pipetted into the motility agar, plates were incubated at 37°C for 6 h, and zones of motility were measured in millimeters.

RESULTS
pIV Secretin Stress Results in the Up-regulation of pspAB-CDE and pspG-To examine the transcriptional response to pIV-induced stress in E. coli wild type MG1655 cells containing the plasmids, pPMR129 (pIV) or pGZ119EH (vector control) were grown to mid-log phase, expression from the plasmids was induced with IPTG for 1 h, and cells were harvested for RNA extraction. The synthesis of pIV reached high levels after 1 h (see Supplementary Materials and Fig. 1) indicating that it should elicit a full cellular response but did not lead to reduced growth rates, or reduced yields of cells, indicating a lack of toxicity. Microarray analyses showed increased levels of psp operon transcripts in pIV-expressing cells compared with the vector control (Fig. 1A). We correlated activation of the pspA promoter with increased levels of the PspA protein using Western blotting (data not shown). Other than the psp operon genes, only a single gene, yjbO, showed a significant and sizeable up-regulation in response to pIV secretin stress ( Fig. 1A; see Supplementary Materials Tables I and III). This data indicates that large transcriptional responses of E. coli to pIV are very rare and identify a new gene involved in the phage shock response, yjbO. We propose to rename this gene pspG.
To confirm that transcript levels of pspG are increased in wild type MG1655 cells expressing pIV, RT-PCR was carried out on the RNA samples used for the microarray experiments. RT-PCR clearly demonstrates that pspG transcription is up-regulated, along with pspA transcription, in pIV-expressing MG1655 cells compared with the vector control ( Fig. 2A). ␤-Galactosidase assays using a translational reporter for PspG (pLL1) confirm that PspG is produced in response to pIV in MC1061 cells (Fig. 2B).
To determine whether the response to pIV detected in E. coli is conserved in other bacteria that contain the psp operon, a pIV expression experiment was carried out in Salmonella enterica serovar Typhimurium LT2. There is significant upregulation of psp operon and pspG transcripts in the pIVexpressing S. typhimurium cells (Fig. 1B; see Supplementary  Materials Tables I and III). The transcriptional response to pIV in S. typhimurium resembles that of E. coli in that the response to pIV secretin stress is highly restricted. Our comparative transcriptomic analysis of responses of E. coli and S. typhimurium to pIV shows that the common core of up-regulated genes are pspABCDE and pspG. Transcription of pspF does not show any change in response to pIV expression, consistent with control of PspF being exclusively at the level of activity (21). Such a limited and specific response to pIV stress resembles the response of E. coli cells to IPTG, a gratuitous inducer of the E. coli lac operon. We performed a microarray experiment to show that IPTG only causes significant increased expression of lac operon genes in MG1655, no other transcriptome changes occur (see Supplementary Materials Table II). As with lac promoter activity induced by IPTG, the effect of pIV inducing stimulus under our growth conditions in E. coli appears to be close to gratuitous (44).
pspG Transcription Is Regulated by psp-encoded Proteins-To examine the effect of overexpression of the psp genes on the transcriptome, transcripts from cells lacking the negative regulator PspA (MG1655⌬pspA) were compared with transcripts from cells lacking the positive regulator PspF (MG1655⌬pspF). In MG1655⌬pspA, the psp operon is expressed at high levels because the negative regulator of its transcription has been removed. Conversely, in MG1655⌬pspF, expression of the psp operon is completely absent because the activator protein required for 54 -RNAP driven transcription has been removed. Levels of psp expression in the MG1655⌬pspA strain are therefore close to levels in wild type cells expressing pIV, but without the production of PspA. It is clear from Fig. 3 that pspBCDE is transcribed at high levels in MG1655⌬pspA compared with MG1655⌬pspF. As with the re- sponse of wild type MG1655 cells to pIV stress, there is very little transcriptional change across the whole genome in response to overexpression of psp genes in MG1655⌬pspA, with the clear exception of the gene pspG, which is strongly upregulated. The level of expression from the pspA promoter and the pspG promoter was similar when we compared wild type cells expressing pIV to cells lacking PspA, the only known negative regulator of psp expression. This result establishes that pIV is a strong and effective inducing signal.
We considered that the synthesis of pIV in mutants unable to mount a wild type Psp response might result in additional changes in the transcriptome to compensate for the inability of the cell to adapt to stress arising through the failure to express the psp genes. MG1655⌬pspA, MG1655⌬pspBC, and MG1655⌬pspF cells containing pPMR129 (pIV) and pGZ119-EH (vector control) were grown to mid-log phase and induced with IPTG. The synthesis of pIV reached high levels after 1 h (see Supplementary Materials Fig. 1), which is consistent with the observation that filamentous phage grow normally in psp mutants (1). The psp mutant strains did not show any growth defects on expression of pIV. Microarray analysis showed that synthesis of pIV in MG1655⌬pspA, MG1655⌬pspBC, and MG1655⌬pspF does not cause any pIV-dependent changes in the transcriptome attributable to the loss of PspA, PspBC, or PspF, respectively (see Supplementary Materials Table I). pspG was not further up-regulated in MG1655⌬pspA, MG1655⌬pspBC, or MG1655⌬pspF cells when expressing pIV, probably because the psp operon, and therefore pspG, were constitutively on in MG1655⌬pspA and always off in MG1655⌬pspBC and MG1655⌬pspF. This data shows that expression of pspG is negatively regulated by PspA and its transcription may be activated by PspF via a 54 promoter and positively regulated by PspBC.
The psp Operon and pspG Are Co-conserved and Co-regulated-pspG and pspFpspABCDE are not physically linked on the chromosome, but pspG is highly conserved among bacteria in which the psp operon is conserved. Furthermore, all bacteria containing a recognizable psp operon carry a pspG homologue, and pspG homologues are not present in bacteria lacking a psp operon. This shows that the pspG and psp loci are co-conserved.
Because our experiments indicated that pspG transcription is regulated by the same elements that regulate psp operon transcription, we used a bioinformatic approach to search the pspG promoter region for the control elements that are present in the psp operon promoter, which are binding sites for 54 , PspF, and IHF. Using the program fuzzpro (EMBOSS programs) and the consensus sequence WWWTCAA[N4]TTR for IHF binding (45) and sequences GGCACGCAAATTGT for 54 binding and TAGTGTAATTCGCTAACT for PspF binding (based on the 54 and PspF binding sites in the pspA promoter) (20, 46) we found potential binding sites for 54 , IHF, and PspF (Fig. 4).
Using the translational fusion for pspG (pLL1) we found that 54 -RNAP and activation by PspF are required in vivo for pIV-induced PspG expression. In the wild type strains the basal level of PspG expression is extremely low, but is upregulated upon induction with pIV. In mutant strains for 54 and PspF, pspG expression is abolished both before and after induction (data not shown). DNase I protection assays using purified components confirmed that the pspG upstream DNA region was bound by PspF (Fig. 5A, lane 3) and by the 54 -RNAP (Fig. 5B, lane 3) at the promoter sequence predicted by bioinformatics. In vitro transcription assays established that a transcript originated from the predicted pspG promoter region, dependent upon 54 , PspF, and ATP (Fig. 6A). Transcripts from the pspG promoter increased with increasing concentrations of PspF (Fig. 6B) and it appears that in the absence of IHF the pspG promoter is much more sensitive to PspF concentration than the pspA promoter. Addition of IHF to the in vitro transcription assay increases the level of transcripts from the pspG promoter indicating that the predicted IHF binding site in the pspG promoter region is functional (Fig. 6C) and that the binding of IHF to the pspG promoter facilitates pspG transcription. IHF enhances psp operon transcription (20,46) and facilitates binding of PspF to its upstream activation sequences in the psp operon regulatory region, autogenously down-regulating pspF transcription (20,21). Consistent with this data and the increased sensitivity of the pspG promoter to PspF in comparison to the pspA promoter (Fig. 6B), PspG expression in an IHF mutant (MC1068) is increased both before and after induction by pIV in vivo (data not shown). Combined, these results provide strong evidence that pspG is tightly co-regulated with the psp operon and is a member of the PspF-dependent regulon.
PspG Is Not a Regulator of the PspF Regulon-To test whether PspA can be induced by pIV, extreme heat shock, or ethanol shock in cells lacking PspG function, we carried out ␤-galactosidase assays using the transcriptional reporter for pspA (pSJ1) in JWK5716_1 (Kmϩ) (⌬pspG) and its parent strain, BW25113. Before induction, basal levels of PspA were equally low (ϳ50 Miller units) in both wild type and ⌬pspG strains. PspA can be induced by pIV stress (ϳ10-fold), extreme heat shock (ϳ8-fold), and ethanol shock (ϳ5-fold) in cells that cannot produce PspG to the same level as in wild type cells. This demonstrates that PspG is not essential for psp operon transcription when the operon is induced using either a specific secretin stimulus or general membrane stress stimuli. Because psp operon transcription is not affected by the absence of PspG in the cell, then PspG is clearly not acting as a regulator of psp operon transcription. Using the translational reporter for PspG expression (pLL1), we show that PspG expression is unchanged in the ⌬pspG strain compared with wild type both before induction (ϳ150 Miller units) and after induction (ϳ700 Miller units). Therefore we conclude that PspG is not involved in controlling the PspF regulon.
PspG Is an Effector of the PspF Regulon-Considering the putative membrane location of PspG, and that PspG is not involved in psp regulation per se, it is likely that this protein is an effector of the psp system. It has been shown that the psp operon is up-regulated during swarming in Salmonella (5) and psp mutants have altered motility (14) therefore we employed a motility assay to compare wild type cells to strains mutant for various psp genes to explore a possible effector function of PspG (Fig. 7). Cells lacking the negative regulator (⌬pspA and ⌬psp-ABC) (therefore with increased PspG expression) show decreased motility. Note that in the presence of PspBC, motility is less decreased. As a control, strains deleted for pspBC show no change in motility. Double mutants for the activator PspF and the negative regulator PspA (therefore no PspG expression) show slightly increased motility. Cells deleted for pspG or for both pspA and pspG also show slightly increased motility. Strains mutant for the activator PspF (no PspG and no Psp operon expression) show unchanged motility. When PspG is expressed from pLL8, motility in both wild type and ⌬pspF cells is greatly decreased (Fig. 7). Induction of PspG expression by 0.4% arabinose decreased motility to a higher extent compared with non-inducing conditions (data not shown). PspG up-regulation does not induce the PspF regulon response, 2 therefore we assume that PspG can function independently of the inducing signal and the Psp response. In summary, these results establish that increased expression of PspG causes decreased motility, whereas the lack of PspG or abolished expression of PspG results in slightly increased motility implying that PspG is an effector of the psp system. DISCUSSION Our data show that expression of pIV in both E. coli and S. typhimurium results in the significant up-regulation of psp-ABCDE and pspG (yjbO) transcripts. Previously it has been shown that the pspABCDE operon and pspG are among the 25 most highly up-regulated genes in S. typhimurium infecting macrophages (4). Similarly high levels of expression of the psp operon and pspG are also seen in S. typhimurium infecting epithelial cells and in Shigella flexneri infecting macrophages and epithelial cells. 3 Therefore, the psp response linked with pspG is observed in a range of enteric bacteria, with potentially important roles for these genes in bacterial virulence. In contrast to the response of E. coli to stresses such as nitrogen limitation (37) and specialized growth conditions (47)(48)(49), which cause substantial changes on a transcriptional level, the synthesis of pIV causes a very restricted change in gene transcription. Restricted responses in microarray experiments have been reported previously, for example, the limited transcriptional response to cell division inhibitors (50). It is possible that a range of pIV-dependent changes in the cell do occur, but only at a translational or post-translational level. Identifying gene regulators through expression profiling may well prove to be a generally challenging problem (51).
As it has been shown for the pspA promoter, both the transcriptional activator, PspF, and the 54 -RNAP physically interact in vitro with the pspG promoter and activation is dependent on PspF and 54 -RNAP in vivo. pspG expression is also subject to negative regulation imposed by PspA. This is in agreement with the results of Green and Darwin (52) in Y. enterocolitica.
Here we have shown that the pspG promoter is more sensitive to PspF activation than the pspA promoter implying that under stress conditions and release of PspA negative regulation, pspG responds rapidly. The transcription of pspG in vitro is enhanced by IHF in a concentration-dependent manner. However, in vivo the basal level of pspG expression in IHF mutants is increased. IHF works to enhance transcriptional activation of the psp operon (20,46) and facilitates the binding of PspF to upstream activation sequences I and II in the psp operon regulatory region. Thus IHF enhances both the activation of psp transcription and the negative autogenous control of pspF keeping the PspF concentration at a low level (20,21). Strains lacking functional IHF will therefore under normal growth conditions have increased concentrations of the activator PspF and decreased concentrations of the negative regulator PspA. Hence, the increased level of pspG expression in IHF mutants should be because of the high sensitivity of the pspG promoter to PspF activation and diminished negative regulation by PspA. To summarize, our data show that pspG is a member of the PspF regulon in E. coli and Salmonella and is tightly regulated in concert with the psp operon. Because the expression of the activator PspF is constant under all growth conditions, the key regulatory point under normal growth conditions is strong negative regulation imposed by PspA, whereas under inducing conditions this regulation is lifted leading to the coordinated expression of the psp operon and pspG. In fact, PspF specifically regulates only pspABCDE and pspG. pspG is physically separated from the psp operon on the chromosome, but is conserved in all bacteria harboring the psp operon and there are no obvious PspG homologues in bacteria that lack a recognizable psp operon. Therefore, the co-regulated expression of the psp operon and pspG by PspF and PspA is likely to be a widely conserved and important feature of cellular adaptation to secretin-induced stress. It is striking that the psp operon and the pspG 54 promoters, which are physically unlinked, are both regulated by PspF, PspA, and IHF in exactly the same fashion. Darwin and Miller (3) suggest that in Y. enterocolitica, another genetic locus is involved in the psp response to secretin stress because a double pspF/psp operon mutant showed a more severe growth defect than the psp operon mutant alone. Further evidence to support this has been reported (52). Our data suggests that the separate genetic locus could be the Y. enterocolitica homologue of pspG. pspG is dispensable for basal level expression and induction of the PspF regulon by pIV, ethanol shock, or extreme heat shock under normal growth conditions and so PspG is not acting to regulate the expression of psp genes. PspG was not toxic, judged by growth rates and yields. In this study we demonstrate that increased expression of PspG results in decreased motility, whereas the lack of PspG expression causes slightly increased motility. It has been shown that swarming in Salmonella induces the psp operon (5), and therefore up-regulation of PspG (this paper). PspA has been implicated in maintaining proton-motive force under stress conditions (15) and proton-motive force is proportional to cell motility (53)(54)(55). The precise function of PspG in motility is not clear from our data, but because PspA upon induction is an effector involved in proton-motive force maintenance we assume that PspG could play an additional role as an effector of the Psp response. The greatest reduction in motility is seen in strains that overproduce PspG but it is likely that under different growth conditions the reduction in motility might be because of synergistic actions of PspA and PspG. The major difference between PspA and PspG could be that PspA requires induction to switch between being a negative regulator to being an effector, whereas PspG, according to our results, is constantly in an effector state. Because PspA and the previously unknown PspF regulon member, PspG, are so tightly co-regulated, our results raise issues about previously described phenotypes attributed to PspA, in particular the contribution of PspG to these phenotypes. We propose that stimuli that induce the psp operon disrupt the integrity of the inner membrane and affect the proton-motive force of the cell. To test the role of PspG as an effector of the phage shock response it will be important in future work to study, using microarray analysis, the response of wild type, pspABCDE, and pspG mutant strains to stresses that change or uncouple the proton-motive force.