and Poul Valentin-Han
and Molecular Biology,University of SouthernDenmark, Campusvej 55,DK-5230, Odense M, Denmark
doi:10.1016/j.jmb.2006.09.004*Corresponding authorKeywords: RybB RNA; Hfq; outer membrane proteins; regulon; non-coding RNAs
Bacteria respond to changes in their environmentby global changes in transcription. These changes intranscription are often accomplished by the induc-tion of alternative sigma factors, which direct RNApolymerase to specific promoters. In enteric bacteriaone of the key pathways involved in maintaining
cell envelope integrity during stress and normalgrowth is controlled by the alternative sigma factorE. The rpoE gene encoding E is essential inEscherichia coli, and an unknown suppression hasto occur to allow growth of rpoE strains.1 Activa-tion of E is triggered by various stress signals,which are sensed in the envelope and communicatedto the cytoplasmic compartment by a complex signalAbbreviations used: sRNA, smallOM, outer membrane; OMP, outer mIPTG, isopropyl--d-thiogalactopyraguanosine 3,5-bispyrophosphate; phase sigma factor in E. coli; N, thecontrolling nitrogen use in E. coli.E-mail address of the correspondi
0022-2836/$ - see front matter 2006 EEnteric bacteria respond to misfolded proteins by activating the transcrip-tion of heat shock genes. These genes are arranged in two major regulonscontrolled by the alternative sigma factors H and E. The two transcriptionfactors coordinate the stress response in different cellular compartments; theH regulon is induced by stress in the cytoplasm whereas the E regulon isactivated by stress signals in the cell envelope. In Escherichia coli E plays acentral role in maintaining cell envelope integrity both under stressconditions and during normal growth. Previous work established that E
is essential for viability of the bacterium and up-regulates expression ofapproximately 100 protein-encoding genes that influences nearly everyaspect of the cell envelope. Moreover, the expression of several outermembrane proteins is down-regulated upon E activation. Here, we showthat two Hfq-binding small RNAs, MicA and RybB, are under positivecontrol of E. Transient induction of RybB resulted in decreased levels of themRNAs encoding OmpC and OmpW. E-mediated regulation of ompC andompW expression was abolished in strains lacking RybB or Hfq. RecentlyMicA was shown to act in destabilizing the ompA transcript when rapidlygrown cells entered the stationary phase of growth. Also, the alternativesigma factor down-regulates this message in a small non-coding RNA-dependent fashion. These findings add the E regulon to the growing list ofstress induced regulatory circuits that include small regulatory RNAs andprovide insight in a homeostatic loop that prevent a build-up ofunassembled outer membrane proteins in the envelope.
2006 Elsevier Ltd. All rights reserved.Department of Biochemistrysen
Jesper Johansen, Anders Aamann Rasmussen, Martin OvergaardConserved Small Non-codinE Regulon: Role in Down-Membrane Proteinsnon-coding RNA;embrane protein;noside; ppGpp,S, the stationarysigma factor
lsevier Ltd. All rights reserveCATION
RNAs that belong to thegulation of Outer
J. Mol. Biol. (2006) 364, 18transduction pathway. The primary point of regula-tion of E is its interaction with the inner membrane-spanning protein RseA and the regulated proteoly-sis of this anti-sigma-factor by the transmembraneproteases DegS and RseP (formerly known as YaeL).Under non-stress conditions E is sequestered byRseA, thereby decreasing the cytoplasmic availabil-ity of E for transcription initiation. DegS, the
E-dependent Small Regulatory RNAsprimary sensor of the signal transduction pathway,becomes activated when its periplasmic PDZ do-main recognizes specific unfolded C-terminal motifsof outer membrane porins.2 Activated DegS initiatesthe cleavage of RseA which triggers a proteolysiscascade that leads to release of E and in turn toinduction of the envelope stress response genes.35
Several lines of evidence, however, indicate thatthe E system can be activated by other stresssignalling pathways. First, it has been observed thatoverexpression of full-length OmpC can still induceE in a strain that should not be capable of sensingunfolded porins due to a deletion of the PDZdomain of DegS.3 Second, the periplasmic proteinRseB, encoded together with E and RseA in therpoE-rseABCoperon, binds to RseA and the interac-tion appears to modulate both activity and stabilityof RseA. Since RseB also binds to misfoldedperiplasmic proteins, it has been proposed that thepresence of such proteins might titrate RseB awayfrom RseA and lead to an increase in active unboundE.6 Third, E activity increases upon entry intostationary phase.7 This growth phase dependentactivation of E is dependent upon the metabolicsignalling molecule guanosine 3,5-bispyrophos-phate (ppGpp) and does not require RseA.8
Various genomic strategies have been used tounravel the E. coli E regulon.912 Taken together theapproaches have led to the identification of 47 E-dependent promoters, which control the expressionof approximately 100 genes. The majority of thesegenes participate in the synthesis, assembly andhomeostasis of outer membrane proteins (OMPs)and lipopolysaccharide, the key constituents of theouter membrane. Also E up-regulates the expres-sion of itself, its negative regulators RseA and RseBand the positive regulator of the cytoplamic stressresponse H. An important feature of the rpoE-rseABC operon is the presence of two conserved E
promoters: one upstream of rpoE and a secondupstream of rseA.11 The autoregulation of E andRseA ensures a rapid increase in proteins needed forenvelope homeostasis under stress conditions and,moreover, provides an efficient mechanism forshutting off E activation when stress signals arereduced or removed. In addition, the expression ofseveral OMPs (i.e. OmpA and the porins OmpC,OmpF, OmpWand OmpX) is down-regulated uponinduction of E.11,12 How this down-regulation maybe accomplished is not well understood.Here we report that two conserved small non-
coding RNAs (sRNAs), MicA (alias SraD) and RybB,belong to the E regulon. Transcription of the micAand rybB genes is strictly dependent on E. We usedmicroarrays to identify putative targets for the twosRNAs and found that induction of RybB stronglydown-regulated ompC and ompW (yciD)13,14 and, inkeeping with previous work, that a transientexpression of MicA led to a strong reduction of theompAmRNA level. The effect of RybB expression on
2ompC and ompW mRNA levels was confirmed byNorthern blot analysis. Moreover, E-mediatedregulation of ompA and ompC/ompW expressionwas dependent on MicA and RybB, respectively, aswell as on the Sm-like chaperone Hfq. Takentogether, these findings indicate that E-dependentsRNAs are important players in a homeostaticregulatory loop that prevent a build-up of unas-sembled OMPs in the envelope during stressconditions.
Transcriptional regulation of MicA and RybBRNAs
The 72 nt MicA RNA was identified by acomputational approach that focused on promoterelements recognized by the E. coli house keepingRNA polymerase (E70) and on Rho-independentterminators present in empty intergenic regions.The transcript of MicA turned out to be shorter thanpredicted and 5 end mapping data assigned thetranscription initiation site to a promoter with a lessperfect match to the consensus sequences for 70-RNA polymerase. The synthesis of MicA is tightlyregulated; the RNA transcript is present at lowlevels during exponential growth in LB medium at37 C and exhibits a strong increase in abundanceupon entry into stationary phase.15 The regulatoryfactor(s) that mediates this growth phase variationin MicA expression has not been identified. Recentwork demonstrated that E. coli MicA RNA is anantisense regulator, which down-regulates the ompAmRNA level when rapidly grown cells enterstationary phase. The MicA-mediated decay of theompA mRNA depends on Hfq and in vitro studiessuggest that binding of MicA specifically interfereswith ribosome binding to the translational initiationregion of the messenger.16,17
In an attempt to gain insight into transcriptionalregulation of micA we inspected the DNA segmentupstream of the 5 end of the transcript. No obvioussequences that resemble the 70 consensus sequenceseem to be present in the immediate upstreamregion. Rather, this region contains 10 and 35sequences with a perfect match to E-dependentpromoters (i.e. GAACTT-N16-TCTNA).
18 This find-ing prompted us to search for additional sRNAs thatmight be under positive control of E. For this weused pattern searches at the Colibri database andfocused on DNA sequences inside intergenic regionsthat contain conserved E and rho-independentterminator motifs separated by 60140 base-pairs.The search resulted in an additional candidate, theRybB RNA, which was identified in an approachusing comparative genomics and microarrays.19 ThesRNAwas found to co-immunoprecipitate with Hfq,suggesting that it might act via base-pairing to targetmRNAs. The expression pattern of RybB is similar tothat found for MicA20 and the sequence of the twosRNA genes, as well as the immediate upstreamregion, is conserved in a wide range of Enterobac-teriaceae (sequence alignments of the promoterregions are presented in Figure 1). These findingshttp://www.genolist.pasteur.fr/Colibri/
suggested that expression of MicA and RybB mightbe under positive control of E. Indeed, such aregulation would be consistent with the growthphase-dependent transcription activation of the twosRNAs.To test this possibility, we assayed the expression
pattern of the MicA and RybB RNAs in isogenicwild-type and rpoE strains.21 Total RNA sampleswere isolated from the strains grown to log phaseand early stationary phase and subjected toNorthern analysis. The experiment is presented inFigure 2(a). In accordance with previous studies theRNA species increased in abundance during tran-sition from vegetative cells to stationary cells in thewild-type strain.15,17,20 In contrast, MicA and RybBexpression could not be detected in the rpoEstrain.To verify this result, we constructed a low-copy-
number plasmid (pNDM-rpoE) that expressed E
under the control of the inducible lac promoterderivative PA1/O4.
22 A wild-type strain was trans-formed with the plasmid, and as a control withplasmid pNDM220 (empty vector).23 The resultingstrains were grown to exponential phase (A6000.4),rpoE expression was induced with increasingamounts of IPTG for 10 min, and the RNA levelsof MicA and RybB were determined by Northernblotting of total RNA extracts. In Figure 2(b), theupper panel shows the induction of rpoE transcrip-
the expression of the two sRNAs and that the MicAand RybB levels correlated with the mRNA levels ofrpoE. Also we determined the ompA mRNA levelfrom total RNA isolated from wild-type and micAstrains expressing rpoE from the lowcopy numberplasmid. The results are presented in Figure 2(c) andshow that the ompA mRNA steady-state level inwild-type cells dropped rapidly as rpoE synthesisincreased, whereas induction of rpoE synthesis in themicA mutant strain had no detectable effect on theompA mRNA level. Thus, micA and rybB transcrip-tion requires E and the alternative sigma factorindirectly down-regulates ompA mRNA in a MicA-dependent fashion.Hfq binding sRNAs are believed to act as anti-
sense regulators and have in many cases beenshown to affect the stability of target mRNAs.24,25
In an attempt to identify putative targets for RybB aswell as additional targets for MicA, we thereforeexamined the effects of controlled transient expres-sion of the two RNAs on mRNA profiles by DNAmicroarrays. To this end we constructed micA/pNDM-micA and rybB/pNDM-rybB strains car-rying an ectopic copy of the sRNA gene under thecontrol of the inducible PA1/O4 promoter. Thestrains, along with control strains carrying theempty vector pNDM220, were grown in LBmediumto late log phase and induced with IPTG for 10 min.Total RNAwas extracted from the two set of strains,
3E-dependent Small Regulatory RNAstion as a function of added IPTG. The lower panelsshow that a short-term expression of rpoE inducedtreated with DNase I and converted into fluorescentlabelled cDNA by reverse transcription. cDNA from
Figure 1. Sequence conserva-tion analysis of micA and rybBpromoters in different enterobac-
Eteria. promoter elements areframed and arrows indicate initia-tion points of RNA synthesis.
control samples and sRNA induced samples werelabelled with Cy3 (green) and Cy5(red), respective-ly, and relative mRNA levels were determined bytwo-colour hybridization to E. coli glass slide cDNAmicroarrays (purchased from Ocimum Biosolu-tions). The induction of MicA and RybB underthese conditions was six- to eightfold higher than theexpression from chromosomal copies. For RybBmarked changes were observed for two genesencoding the outer membrane proteins OmpC andOmpW (down-regulated 11- and sevenfold, respec-tively, upon rybB induction) whereas only one spotcorresponding to OmpA was strongly affected byMicA expression (down-regulated 12-fold).The microarray data suggested that RybB, like
MicA, is a regulatory RNA that acts in modulatingthe expression of outer membrane proteins (OMPs).To examine such a role of the sRNA we first testedwhether the ompC mRNA levels are affected bydeleting the rybB gene. Wild-type and rybB strainswere cultivated in LB, cells were harvested atvarious time points after inoculation, RNA wasextracted, and Northern analysis was performed onequal amounts of total RNA to determine therelative ompC mRNA levels at each point duringgrowth. The results are presented in Figure 3(a) andshow that the ompCmRNA levels were significantlyelevated in the strain deleted for rybB (three- tofourfold during transition from exponential phase to
ompC mRNA levels. As for the arrays describedabove, RybB was induced from a low copy numberplasmid in a strain deleted for the chromosomal copyof the sRNA. In Figure 3(b), the upper panel showsthe induction of the RybB RNA as a function ofadded IPTG. The middle panel shows that theinduction of RybB led to a rapid decrease inompCmRNA and the RybB levels inversely correlatedwith the levels of ompCmRNA. Finally, we askedwhether RybB and Hfq are also required for E-mediated regulation of ompC expression. The resultspresented in Figure 3(c) show that the steady-stateompC mRNA level remained unaffected of elevatedexpression of rpoE when RybB or Hfq was absentfrom the cell. Similar results were obtained with theompW mRNA (Figure 3). Consistent with the micro-arrays, however, the effect of RybB was stronger onthe ompC mRNA than on the ompW mRNA.
Over the past decade, it has become clear thatcells contain a wealth of small, non-coding RNAs
Figure 2. Regulation ofmicA and rybB synthesis by thetranscriptional factor E. (a) Effects of rpoE deletion onMicA and RybB RNA levels. Isogenic rpoE+ (BW25113)41