Down-regulation of Porins by a Small RNA Bypasses the Essentiality of the Regulated Intramembrane Proteolysis Protease RseP in Escherichia coli*

Adaptation to extracytoplasmic stress in Escherichia coli depends on the activation of σE, normally sequestered by the membrane protein RseA. σE is released in response to stress through the successive RseA cleavage by DegS and the RIP protease RseP. σE and proteases that free it from RseA are essential. We isolated a multicopy suppressor that alleviated RseP and DegS requirement. The suppressor encodes a novel small RNA, RseX. Its activity required the RNA-binding protein Hfq. We used the property that small RNAs are often involved in RNA-RNA interactions to capture RseX putative partners; ompA and ompC mRNA, which encode two major outer membrane proteins, were identified. RseX activity was shown to confer an Hfq-dependent coordinate OmpA and OmpC down-regulation. Because RseP is shown to be no longer essential in a strain lacking OmpA and OmpC, we conclude that RseP, which is required for normal σE activation, prevents toxicity due to the presence of two specific outer membrane proteins that are down-regulated by RseX.

Regulated intramembrane proteolysis (RIP) 2 refers to a biochemical activity by which membrane-bound proteins are cleaved to release factors into the bacterial cytoplasm or into the eukaryotic cytosol (19 -21). These factors often activate gene transcription and regulate different processes such as spore formation in bacteria or cell differentiation and lipid biosynthesis in eukaryotes. RseP is an integral membrane protein with a zinc metalloprotease active site motif and an ortholog of the eukaryotic S2P protease involved in RIP (18,22). As such, the release of E from the membrane-associated protein RseA by the successive activity of DegS and RseP is an example of the RIP process.
Fine tuning of bacterial gene expression is often performed by small noncoding RNA, frequently called sRNA (23). Biocomputing based analysis suggested that E. coli have about a hundred of them (24 -26). The sRNAs contribute to many processes including transcriptional and translation regulation, DNA replication, and even protein degradation and translocation (23). A remarkable example of their role in gene expression is given by their regulation of rpoS, which encodes S , itself a regulator of stationary phase-induced genes (27). rpoS is regulated posttranscriptionally by at least three sRNAs. Two of them (DsrA and RprA) stimulate RpoS translation, whereas the other (OxyS) represses RpoS translation. Each of these sRNAs is transcribed in response to specific stress conditions, allowing adaptation to changing environments. Recent observations also reveal that expression of outer membrane proteins (Omps) is also under the control of several sRNAs (28 -30).
Here, we report the identification of RseX, a new small noncoding RNA, selected on the basis of its suppressor activity on RseP essentiality. The suppression mechanism requires the RNA-binding protein Hfq. Using RseX as bait, two of its targets, ompA and ompC mRNA were captured. RseX overproduction results in reduced amounts of ompA and ompC mRNA and corresponding outer membrane proteins. The absence of both OmpA and OmpC itself allows survival to RseP depletion independently of RseX. Our results show that the sRNA RseX allows survival in the absence of the RIP protease RseP by preventing ompA and ompC expression.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, Oligonucleotides, and Culture Conditions-The E. coli strains, plasmids, and oligonucleotides used in this work are described in supplemental Tables S1-S3, respectively. P1 vir-mediated transduction was carried out as described (31). We previously reported the construction and use of an E. coli genomic DNA library (32). The chromosomal rseX gene was deleted (see supplementary tables) by targeted gene substitution using a combination of two published protocols (33,34) as described before (35). The rseX deletion * This work was supported in part by the program "Puce à ADN" (CNRS) and the Programme de Recherche Fondamentale en Microbiologie, Maladies Infectieuses et Parasitaires (Ministè re de L'Education Nationale, de la Recherche et de la Technologie) and in part by Actions Concerté es Incitatives Microbiologie Grant MIC 0324 (Ministè re de L'Education Nationale, de la Recherche et de la Technologie). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Tables S1-S5. 1  was confirmed by PCR and Southern blot analysis. Excision of antibiotic-resistant cassettes flanked by flippase recognition targets from several strains (36) were removed using pCP20 (34). Cells were grown in NZ broth or on solid NZ containing 15 mg ml Ϫ1 agar (37). When necessary, antibiotics were added at the following concentrations: 100 g ml Ϫ1 ampicillin, 20 g ml Ϫ1 chloramphenicol, 30 g ml Ϫ1 kanamycin, and 10 g ml Ϫ1 tetracycline. Expression from the P LtetO-1 promoter was obtained by the addition of 1 M anhydrotetracycline (aTc), and from the P lac/ara-1 promoter it was obtained by the addition of 1 mM IPTG and 0.2% L-arabinose.
Molecular Biology and Protein Techniques-The ␤-galactosidase assay was described previously (31,35), with the modification that cultures were grown in NZ broth. Plasmid preparations, DNA cloning and ligation, PCR amplification, and DNA transformation were carried out according to standard protocols (37) or to manufacturer instructions. DNA sequencing was performed on an ABI310 automatic DNA sequencer (Applied Biosystems Inc., Foster City, CA). Histidine-tagged Hfq was purified from strain PhB3247 (containing pTE607) as described (38).
Outer membrane protein analyses were performed as follows. MG1655Z1 overnight cultures were diluted 500-fold and grown at 37°C in medium with or without aTc, as indicated. The bacteria were harvested (at A 600 of about 0.2 and 1.5) by centrifugation at 4°C and concentrated to a growth equivalent of A 600 100. Pellets washed once in 10 ml Tris-HCl (50 mM) EDTA (5 mM) buffer, pH 8.0 (TE), and suspended in 2.5 ml of TE were lysed by lysozyme (1 mg/ml) and by sonication. Unlysed bacteria and cell debris were removed by centrifugation (5000 rpm, 10 min at 4°C). The supernatants were then ultracentrifuged at 70,000 rpm for 45 min at 4°C. The pellets were incubated overnight at 4°C in 2 ml of Tris-HCl (50 mM), pH 8, MgCl 2 (25 mM), Triton X-100 (1%). Solubilized proteins and Triton X-100-insoluble materials were then separated by ultracentrifugation as above. 10 l of outer membrane proteins (in pellet) were run on 12% SDS-PAGE containing 8 M urea. The proteins were visualized by Coomassie Blue staining. The band intensities were quantified by integrating corresponding signals obtained from four gel scans using GelScan 1.2 software (developed by Dr. Yvan Zivanovic). The intensity of each lane was normalized to an observed invariant band. Samples containing pZE11-RSEX were compared from cultures growing in the presence or absence of RseX inducer (aTc). The reported differences of OmpA and OmpC signals between the two conditions are possibly underestimated because other invariant proteins might compose the signal.
RNA Techniques-5Ј RACE mapping was previously described (25,39). Northern blots were performed as described (37,40), except that membrane prehybridization and hybridization were performed at 65°C, and for identification of RseX RNA, the samples were electrophoresed under denaturing conditions on 10% polyacrylamide gels containing 8 M urea. Primers used for PCR ␣ 32 P-labeled products used are described (supplemental Table S3).
Gel mobility shift assays were performed in as follows: 5Ј end-labeled RseX (0.05 pmol) were incubated with increasing amounts of Hfq, unlabeled RseX, yeast tRNA, or ompA mRNA in 10 l of RNA binding buffer (10 mM Tris-HCl, pH 7.5, 10% glycerol, 60 mM KCl, 2 mM MgCl 2 , 0.2 mM EDTA, 1 mM dithiothreitol, 50 ng/l bovine serum albumin, 100 ng/l tRNA) at 30°C for 20 min. RNAs were obtained by T7 RNA polymerase-driven transcription (using MEGAshortscript kit; Ambion Inc., Austin, TX) performed on PCR products having the rseX or truncated ompA mRNA genes under control of the T7 promoter (for primers, see supplemental Table S3). The binding reactions were mixed with 2.5 l of native loading dye and resolved on native PAGEs (8 or 10%) at 4°C.

Selection of rseX, a Multicopy Suppressor of RseP Depletion-The rseP
gene is an essential gene in E. coli (18,22). To understand the reason for RseP essentiality, we selected for multicopy suppressors permitting survival of RseP-depleted strains. We previously reported the construction of a strain with a deletion of chromosomally encoded rseP and carrying an ectopic copy of rseP under the control of a P LtetO-1 (PhB2387) promoter (35). In this strain, viability depended upon rseP expression obtained by the addition of a gratuitous inducer (aTc) to the media. This strain was transformed with a confirmed E. coli genomic DNA library (32). 200 clones that grew in the absence of RseP expression were isolated among ϳ10 6 transformants. Analysis of 12 of these multicopy suppressors showed that the selected plasmids carried inserts from the 4-or 44-min (min) region of the E. coli chromosome (4 and 8 clones, respectively). Inserts from the 4-min region contained rseP and were not further analyzed. Using a similar selection, Akiyama and co-workers (14) selected plasmids encoding E , allowing them to demonstrate that RseP is involved in RseA degradation. None of the 12 plasmids that we characterized encoded E . Subcloning of inserts from the 44-min region led to a plasmid containing solely a 170-bp insert (pRSEX), which was sufficient to permit survival to an RseP depletion. Presence of pRSEX in the wild-type strain allowed the establishment of an rseP disruption mutant by P1-vir-mediated transduction (supplemental Table S1), whereas no clones were obtained in the corresponding strain having the control plasmid; the selected plasmid thus compensated for the complete absence of RseP.
The cloned insert was a DNA fragment corresponding to the yedR-yedS intergenic region. yedR gene encodes a putative integral membrane protein, and yedS encodes a putative homolog of OmpS1, an outer membrane protein from Salmonella typhi (41); we will refer to YedS as OmpS1. The cloned insert did not encode any putative open reading frame, and we therefore considered, as we will show later, that suppression could be due to activity of a small RNA. The putative corresponding gene and sRNA were named rseX and RseX for RNA suppressor of extracytoplasmic stress protease, respectively.
Identification of RseX sRNA-The existence of more than 200 putative sRNA genes in the E. coli genome was predicted using a bioinformatics approach (42). Among them, it was proposed that the yedR-ompS1 (yedS) intergenic region expresses IS096, a 122-nucleotide sRNA, although authors failed to detect any transcript by Northern blot hybridization probing for IS096 and thus had no proof of its existence. We did detect an RseX-specific transcript of about 100 nucleotides by Northern blot analysis in a strain containing pRSEX (Fig. 1). RACE mapping of RseX in a strain containing pRSEX showed that the 5Ј end initiates likely at base pair 2031671 on the E. coli genetic map (supplementary Table S4). Because the RACE was more efficient after a tobacco acid pyrophosphatase treatment, we concluded that the observed amplified fragment originates from an RNA having its transcriptional promoter rather than an RNA issued from an endonucleolytic cleavage. A putative 70 promoter for this initiation point was detected as well as a strong putative rho-independent terminator (finishing at position 2031761 on the E. coli genetic map). This observation shows that the length of RseX is of 91 nucleotides, with no apparent open reading frame. Regulatory sRNAs are characterized by the presence of stable secondary structures; indeed, a stem loop in the 5Ј extremity of RseX is predicted ( Fig. 1) by mfold software (43). Thus far, the RseX sequence is conserved among enterobacteria closely related to E. coli.
Expression of RseX was placed under transcriptional control of the P LtetO-1 promoter on multicopy plasmid (pZE11-RSEX). When transformed in a strain having the TetR repressor, expression of RseX was conditional upon the addition of aTc (see Fig. 5A). In this strain, the ⌬rseP allele was introduced in the presence of aTc, and the strain remained viable as long as the inducer was present, showing that rseX transcription is necessary for the suppression phenotype (data not shown).
RseX Activity Is Hfq-dependent-Hfq, initially identified as required for the replication of Q␤ phage RNA in E. coli, is an abundant pleiotropic regulator that controls expression of many proteins by affecting mRNA translation, stability, or polyadenylation (reviewed in Ref. 44). Its inactivation results in a wide variety of phenotypes caused by the loss of its RNA chaperone activity, which is proposed to facilitate interactions between sRNAs and their targets. Because RseX was putatively an sRNA acting through an RNA/RNA interaction, we hypothesized that RseX suppression of RseP depletion could be Hfq-dependent. To test this, we used a strain (PhB2364) in which RseP expression was conditionally dependent on the presence of inducers (arabinose/IPTG) (35). As described above, the presence of pRSEX permitted growth of this strain in the absence of inducers. However, introduction of an hfq deletion in this strain prevented its growth in the absence of arabinose/IPTG ( Fig.  2A). This hfq-dependent phenotype was not due to a polar effect of the deletion, because it was complemented by an Hfq encoding plasmid (pTX381). The suggested putative binding activity of Hfq with respect to RseX was tested in vitro by gel mobility shift assays. Migration of a constant amount of labeled synthetic RseX mRNA was determined in the presence of increasing amounts of purified histidine-tagged Hfq, using physiological pH and salt concentrations and a large excess of tRNA from yeast to reduce nonspecific RNA binding to Hfq. RseX Hfqdependent mobility retardation was observed, showing that Hfq binds to RseX (Fig. 2B). Binding between RseX and Hfq is specific, because an 8000-fold molar excess of total tRNA displaced only a minor fraction of RseX from a preformed RseX⅐Hfq complex, whereas a 20-fold excess of cold RseX released labeled RseX from the complex (Fig. 2, C and D).
We concluded that Hfq binds to RseX and is required for its suppressive activity on ⌬rseP. This observation suggested that the activity of RseX occurred through an RNA/RNA interaction.
Multicopy rseX Is a Suppressor of ⌬degS but Not of the Absence of E -The viability of a strain containing the degS gene solely under control of the P lac promoter depends on the presence of IPTG in the medium (17). Because inactivation of RseA suppresses a ⌬degS muta-tion, it was concluded that DegS is essential because it contributes to the release of E by degrading RseA (17). When transformed by a plasmid containing rseX, growth of the ⌬degS strain became IPTG-independent (Fig. 3A), suggesting that a high level of RseX can bypass the need for DegS. To confirm this result, we used a strain in which RseX was conditionally expressed upon the addition of aTc (pZE11-RSEX). The ⌬degS arg::Tn5 allele was introduced in these strains in the presence of inducer (supplemental Table S1); removal of inducer was deleterious to strain growth (data not shown). We concluded that accumulation of RseX, in addition to its suppressive effect on rseP, was also a suppressor of degS.
DegS and RseP degrade the E -sequestrating protein RseA; increasing the quantity of E proficient for transcription compensates for DegS  The RseX Hfq-dependent phenotype is mediated via RseX/Hfq interactions. A, 5-l spots on NZ plates (with and without IPTG/arabinose as indicated) of 10-fold serial dilutions (from left to right) of overnight cultures grown in NZ supplemented with IPTG and arabinose (Ara). The strains are derivatives of PhB2364 (attHK022 [P lac/ara-1 ::rseP] ⌬rseP), containing pRSEX as indicated. The plasmid pRSEX is a suppressor that allows growth of PhB2364 in the absence of inducer. However, introduction of the hfq deletion prevents the pRSEX suppressive effect. The phenotype observed in the hfq mutant is reversed by pTX381 encoding Hfq. B, autoradiogram of a native gel retardation assay using purified labeled RseX (RseX*) with increasing amounts of purified His-tagged Hfq. The observed interaction between RseX and Hfq (RseX*-Hfq) is specific because the interaction between RseX and Hfq is not affected by the presence of large amounts of unlabeled total yeast tRNA (C), and it is displaced by the addition of small amount of unlabeled RseX (D). and RseP deficiency (13,14,17,45). It was therefore proposed that RseP, like DegS, is essential because it frees E from its sequestration by RseA. We therefore investigated whether the RseP suppressor effect of RseX was due to a recovery of E activity. E -dependent transcription can be determined by measuring ␤-galactosidase activity in a strain carrying an rpoH-P3::lacZ gene fusion (in CAG16037) (9). CAG16037 derivatives containing either a control plasmid, pRSEX, ⌬degS pRSEX, or ⌬rseP pRSEX were constructed and tested for ␤-galactosidase activity. In the absence of DegS or RseP, pRSEX did not restore normal levels of E activity (Fig. 3B), showing that RseX is not involved in the recovery of E activity.
In the parental strain (degS ϩ rseP ϩ ), expression from the rpoH P3::lacZ promoter was slightly but reproducibly lower when pRSEX was present rather than the control plasmid (Fig. 3B). This is compatible with the hypothesis that accumulation of RseX results in a reduction of extracytoplasmic stress and possibly the requirement for E . The above results suggested that the E pathway could be dispensable in the presence of multicopy rseX. However, using selection of a 95% rpoE (encoding E )-linked marker (nadB::tet), the ⌬rpoE::Cm allele (46) could not be introduced in a pRSEX-containing strain. We concluded that although it cannot fully bypass the requirement for E , overexpression of RseX can bypass the need for DegS and RseP without activating the E pathway. Capture of RseX Targets: ompA and ompC mRNA-The Hfq dependence of the RseX phenotype for suppression of ⌬rseP lethality led us to hypothesize that RseX acts through an RNA/RNA interaction with a target RNA. Identification of RseX putative targets using sequence alignment software by searching for complementary stretches on the E. coli genome did not reveal any obvious candidates. To identify the RseX target(s), we developed an in vitro screen to isolate RNAs that bind to RseX (Fig. 4A). The strategy that we used has common features to one previously reported (47). Synthetic RseX sRNA expressed from a T7 promoter was purified, biotinylated, fixed via streptavidin-associated magnetic beads, and incubated with RNAs extracted from E. coli strains grown in different conditions (cf. supplementary data). cDNAs produced from eluted RNAs were Cy3-labeled and then hybridized on pangenomic E. coli DNA chips to identify RNAs captured by RseX. Two spots gave a strong hybridization signal; they corresponded to ompA and ompC PCR products (Fig. 4B). Weaker signals were also detected (supplemental Table S5). The two ompA and ompC genes are genetically unlinked. Both encode outer membrane proteins forming passive diffusion pores, which allow small molecular weight hydrophilic materials across the outer membrane. Our experiment suggested that the ompA and ompC mRNAs are likely to interact directly with RseX and that by affecting the amount of OmpA and OmpC, RseX would bypass the need for RseP.
RseX Interacts with 5ЈompA mRNA-Interaction of RseX with ompA mRNA was tested by gel mobility shift assays. Samples of labeled RseX were incubated with increasing concentrations of mRNA corresponding to the first 167 nucleotides of ompA mRNA. The labeled RseX band shifted to higher molecular weight as the ompA mRNA concentration increased (Fig. 4C). This experiment confirmed the interaction between RseX and ompA mRNA, as observed by our microarrray screening for RseX targets. Because the RseX mobility shift was obtained with a truncated mRNA, we concluded that the interaction involves the first 167  . The resulting complex was incubated with pooled RNA (wavy lines) preparations extracted from exponential and stationary phase cells (step 3). RseX-unbound RNAs (thin wavy lines) were removed by several washes, whereas specific complexes remain associated to the streptavidin, which is kept in the tube using a magnet. RseX-bound RNAs were then eluted (step 4) and used as matrices for reverse transcription performed with random hexameric primers. cDNAs were Cy3-labeled (step 5). A control experiment in which RseX was omitted and Cy3 was substituted by Cy5 was performed to discriminate nonspecific RNA binding. Both preparations were mixed and hybridized on E. coli DNA chips. Images of fluorescent spots were obtained by scanning at wavelengths of 532 nm (Cy3) and 635 nm (Cy5). For more details, see the supplementary data. B, the relevant sections of an array obtained as described in A are presented. Each section of the array contained a positive control corresponding to genomic E. coli DNA as indicated. Two spots gave a strong Cy3 fluorescent signal. They correspond to positions having PCR products corresponding to ompA and ompC, as indicated. The experiments were fully repeated twice, giving equivalent results. C, autoradiogram of a native gel retardation assay using purified labeled RseX (RseX*, 0.05 pmol) with increasing amounts (lanes from left to right: 0, 1, 5, 10, 20, 30, and 60 pmol) of purified truncated ompA mRNA (167 nucleotides (Ϫ135 to ϩ32)). The RseX/ompA mRNA interaction is confirmed by mobility shift of labeled RseX in the presence of ompA mRNA.
nucleotides of ompA mRNA that contains mainly the 5Ј-untranslated region.
RseX Reduces OmpA and OmpC Expression-The above results suggested that the presence of multicopy rseX could affect the quantity of expressed outer membrane proteins. Because RseX interacts with ompA and ompC mRNA, we first analyzed whether the accumulation of RseX would result in a decrease of the targeted mRNA. Indeed, amounts of both ompA and ompC mRNA were drastically decreased when RseX was overproduced (Fig. 5A). We also made similar observations by comparing transcriptomes of a strain overexpressing RseX and its isogenic control (containing pUC18) using a E. coli microarray (data not shown). This RseX mRNA-dependent reduction in ompA and ompC mRNAs is likely to be due to mRNA instability driven by RNaseE-based ribonucleoprotein complexes (48). Outer membrane protein extracts from strains containing pRSEX or a control plasmid were also examined; as expected, the presence of high levels of RseX led to decreased amounts of OmpA and OmpC (Fig. 5B). Quantification on four protein gels indicates that induction of RseX expression leads to a reduction of OmpA and OmpC band intensities by 0.5-and 0.3-fold, when compared with normalized levels in the absence of RseX expression (strain containing pZE11-RSEX without aTc).
Because complementarities between RseX and the 5Ј-untranslated region of ompA and ompC mRNA were detected (Fig. 5C), we consider it likely that pairing takes place with mRNA 5Ј-untranslated regions. We conclude that these interactions between RseX and its targets result in a significant reduction of OmpA and OmpC expression.
RseP Is Not Essential When OmpA and OmpC Are Absent-Our data suggested that the absence of the OmpA and OmpC could be a reason for survival of E. coli in the absence of RseP. The ⌬rseP allele was directly introduced by P1-vir-mediated transduction only into the double ⌬ompA ⌬ompC mutant (MG1655Z1 derivative) but not in the parental strain or single mutant ⌬ompA and ⌬ompC strains. Conversely, in a ⌬rseP pRSEX strain, expression of OmpA or OmpC under the control of an arabinose/lactose promoter (using pZS14-OmpA and pZS14-OmpC, respectively) was lethal, canceling the effect of RseX on ⌬rseP survival.
The absence of porins can result in strains expressing low level of E -regulated genes (9). The fact that RseP is dispensable in strains deficient for certain outer membrane proteins could be attributed to the correlated low level of expression of E -regulated genes. We measured E -dependent transcription by means of rpoH-P3::lacZ gene fusion in outer membrane protein deficient strains (Fig. 6B). The simultaneous absence of OmpC and OmpF leads to a very low levels of E activity. However, the double mutation ⌬ompC ⌬ompF was not a suppressor of ⌬sreP. We therefore concluded that the low level of E -regulated gene expression is not per se a suppressive condition of ⌬rseP.
Using selection of nadB::tet, as described above, the ⌬rpoE::Cm allele could not be introduced in ⌬ompA ⌬ompC or ompA ⌬ompC ⌬ompF strains. The absence of the three outer membrane proteins cannot fully bypass the requirement for E , suggesting that residual activity of E is still required.
Because Hfq is required for the survival of a ⌬rseP strain in the presence of pRSEX and is necessary to stimulate the sRNA interactions with their targets, we tested the Hfq involvement in the E response (Fig. 6B). Its absence resulted in a more than 3-fold activation of the pathway as judged by the expression from rpoH-P3::lacZ gene fusion. We concluded that likely by its activity on sRNAs affecting outer membrane protein expression, Hfq is necessary for the down-regulation of E response.
Altogether, the above results show that specific reduction in the amounts of both OmpA and OmpC outer membrane proteins, which can be triggered by RseX in an Hfq-dependent manner, permits growth in the absence of the essential RIP protease RseP. Therefore, this imply that in a ⌬rseP strain, which has a low E activity because of RseA stabilization, the presence of OmpA and OmpC generates a lethal toxicity.

DISCUSSION
RseP is an essential inner membrane protease that degrades RseA, a transmembrane anti-E factor. We isolated a multicopy suppressor of an RseP-depleted strain. We demonstrated that it expresses RseX, an sRNA of 91 nucleotides. Combining the property of sRNAs to bind RNA targets with the use of microarrays, we developed a screen that allowed us to identify ompA and ompC mRNA as putative partners of RseX. Analysis of transcripts and of outer membrane protein fraction show unambiguously that multicopy rseX down-regulates OmpA and OmpC.
The observation that a plasmid encoding OmpC was responsible for a down-regulation of OmpF previously led to the identification of MicF, an sRNA encoded by a gene upstream and adjacent to ompC (28,49). Similarly, MicC, an sRNA encoded by a gene upstream and adjacent to ompN (encoding the OmpN porin) regulates OmpC (29). Recently, it was shown that MicA is an antisense sRNA that regulates OmpA expression by specific inhibition of ribosome binding (30,50). RseX is a newly characterized sRNA, which in contrast to the porin-interfering sRNAs cited above, modulates two outer membrane proteins, OmpA and OmpC. A generalized bacterial strategy of using sRNAs to regulate outer membrane protein expression emerges from these examples (Fig. 7).
The rseX gene is adjacent to ompS1 (yedS), which encodes a putative porin. This organization is similar to that of micF with respect to ompC or that of micC with respect to ompN. These examples reveal a correlation between the genetic location of genes expressing sRNAs that regulate outer membrane proteins and outer membrane protein coding genes (illustrated in Fig. 7). Curiously, in all these cases, the genes of the regulated outer membrane proteins are genetically unlinked to the gene of the corresponding regulatory sRNAs. Because the excess of outer membrane proteins can be deleterious, one can propose that the establishment of a new outer membrane protein, possibly providing an evolutionary advantage, is favored by the simultaneous down-regulation of another outer membrane protein already present in the cell.
Many sRNAs are induced during specific growth conditions. For example, MicA accumulates upon entry into stationary phase (30,50), and MicF, which is under the control of at least four regulators, increases at low temperature and in minimal medium (reviewed in Ref. 51). The natural conditions of RseX induction have not yet been found. Nevertheless, one can speculate that certain stress conditions requiring the coordinate down-regulation of both OmpA and OmpC for adaptation would activate RseX expression. The presence of many sRNAs that are differentially regulated contributes to the fine tuning of outer membrane protein ratios according to growth conditions.
Outer membrane proteins play a key role in the regulation of extracytoplasmic stress response. Overproduction of OmpF, OmpC, OmpT, or OmpX causes an increase in E activity, whereas inactivation of several outer membrane proteins leads to lower expression of E -regulated genes (9,45). Porins such as OmpC act as regulators of the adaptation response by stimulating DegS activity against RseA (11). It was proposed that by lowering the E response, the absence of porins also reduced the E requirement. However, lowering the extracytoplasmic stress response is not in itself sufficient to bypass the requirement for the E activator RseP, because: (i) deletions of porins that result in low E response (e.g. ⌬ompC ⌬ompF) do not suppress the absence of RseP and (ii) RseX overexpression, which compensates the ⌬rseP strain, had only a moderate effect on the E response. It is likely that suppression by RseX is due to the loss of toxicity that is mainly OmpA-dependent.
Recent results indicate that two Hfq-binding sRNAs under transcriptional regulation of the two component system EnvZ/OmpR are also negative regulators of outer membrane proteins (52). So far, all sRNAs directly involved in porin regulation are also Hfq-dependent negative regulators. This places Hfq as a central element of outer membrane  . Schematic view of sRNAs involved in regulation of outer membrane proteins. Name, orientation, and position in min on the E. coli genetic map of sRNAs and outer membrane protein genes are indicated (left). The arrows point from genes to RNAs (dashed arrows indicate sRNAs), mRNAs to outer membrane proteins, and porins to DegS indicating transcription, translation, and E pathway activation, respectively. Short thick wavy lines represent sRNAs, whereas thinner long wavy lines represent mRNAs. Hfq is essential for the activity of the shown sRNAs, as symbolized by the column below Hfq. The absence of Hfq prevents MicA, MicC, MicF, and RseX negative regulatory activities on outer membrane protein expression, which can explain the induction of E response in ⌬hfq strains. protein expression. Here, we show that its absence results in a strong activation of a E regulated promoter. This induction is likely due to the loss of at least MicA, MicC, MicF, and RseX functionality, which in turn up-regulates outer membrane protein expression; as accumulation of outer membrane proteins activates proteolytic degradation of RseA, the E pathway is stimulated. Therefore, Hfq appears to be a general regulator of outer membrane protein expression, and consequently, it is also a key factor of E regulation.
The negative regulation of OmpA and OmpC by RseX has consequences on outer membrane permeability. Modulation of entry of small molecules into the cell could lower the requirement for E and therefore bypass the requirement for RseP. Alternatively, RseP or E -regulated genes are necessary for biogenesis of OmpA and OmpC, and the absence of these two outer membrane proteins are sufficient to remove the toxicity induced by the absence of factor necessary for proper folding of outer membrane proteins. In such a case, activation of the E response would be more necessary for outer membrane protein biogenesis than for adaptation to extracytoplasmic stress. It is interesting to notice that rseP is predicted to be in an operon with yaeT (encoding YaeT alias Omp85), a gene recently shown to be essential in outer membrane protein assembly (53)(54)(55). It is therefore possible that RseP also contributes to OmpA and OmpC biogenesis outer membrane proteins via a role in YaeT activity.