SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Note added in proof
  9. References
  10. Supporting Information

The small RNA, ArcZ (previously RyhA/SraH), was discovered in several genome-wide screens in Escherichia coli and Salmonella. Its high degree of genomic conservation, its frequent recovery by shotgun sequencing, and its association with the RNA chaperone, Hfq, identified ArcZ as an abundant enterobacterial ‘core’ small RNA, yet its function remained unknown. Here, we report that ArcZ acts as a post-transcriptional regulator in Salmonella, repressing the mRNAs of the widely distributed sdaCB (serine uptake) and tpx (oxidative stress) genes, and of STM3216, a horizontally acquired methyl-accepting chemotaxis protein (MCP). Both sdaCB and STM3216 are regulated by sequestration of the ribosome binding site. In contrast, the tpx mRNA is targeted in the coding sequence (CDS), arguing that CDS targeting is more common than appreciated. Transcriptomic analysis of an arcZ deletion strain further argued for the existence of a distinct set of Salmonella loci specifically regulated by ArcZ. In contrast, increased expression of the sRNA altered the steady-state levels of > 16% (> 750) of all Salmonella mRNAs, and rendered the bacteria non-motile. Deep sequencing detected a dramatically changed profile of Hfq-bound sRNAs and mRNAs, suggesting that the unprecedented pleiotropic effects by a single sRNA might in part be caused by altered post-transcriptional regulation.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Note added in proof
  9. References
  10. Supporting Information

Genome-wide searches involving diverse methodologies have established that bacteria express a large number of small RNAs (sRNAs) (Altuvia, 2007; Livny and Waldor, 2007; Sharma and Vogel, 2009). Functional analyses have identified a subset of these sRNAs as post-transcriptional regulators of gene expression that are involved in a wide range of physiological circuits, including outer membrane biogenesis (Guillier et al., 2006; Vogel and Papenfort, 2006; Valentin-Hansen et al., 2007), quorum sensing (Lenz et al., 2004), iron homeostasis (Masséet al., 2007), long-term survival (Wassarman, 2007), the SOS response to DNA damage (Vogel et al., 2004), sugar stress and metabolism (Vanderpool, 2007; Görke and Vogel, 2008), or the control of photosynthesis (Duhring et al., 2006), secondary metabolism (Kay et al., 2005), carbon storage (Weilbacher et al., 2003), or virulence (Romby et al., 2006; Toledo-Arana et al., 2007), to name just a few.

The recent surge in sRNA identification was triggered by several pioneering studies in Escherichia coli which began in 2001 (Argaman et al., 2001; Rivas et al., 2001; Wassarman et al., 2001; Chen et al., 2002; Vogel et al., 2003; Zhang et al., 2003); within 2 years, close to 100 E. coli sRNAs had been described. These studies usually relied upon sequence conservation of candidate sRNAs in other enterobacterial genomes, either for initial biocomputational prediction or for the subsequent selection of candidate sRNAs for experimental validation. As a result, a considerable number of sRNAs that are highly conserved among enterobacteria, including the model pathogen Salmonella enterica serovar Typhimurium, have been identified (Hershberg et al., 2003; Papenfort et al., 2008). The study of this ‘enterobacterial core set’ of sRNAs is particularly interesting as it might identify functions that extend beyond individual bacterial species.

ArcZ is one of these ‘core sRNAs’, and was identified in both of the first genome-wide sRNAs screens in E. coli as either SraH (Argaman et al., 2001) or RyhA (Wassarman et al., 2001). Based on its genomic location and on the results of our present study, the sraH/ryhA gene was here renamed to arcZ (for arc-associated sRNA Z). 5′ and 3′ RACE mapping in E. coli showed that arcZ transcription initiates ∼100 bp upstream of the yhbL start codon, and that it ends at a ρ-independent terminator (Argaman et al., 2001). On Northern blots, however, the ∼120 nt precursor transcript was barely visible; instead, a ∼50 nt processed species corresponding to SraH 3′ end was detected (Argaman et al., 2001). An updated alignment shows that arcZ genes are consistently located in the intergenic region (IGR) between yhbL and arcB, and transcribed in the opposite direction to these two genes (Fig. 1B).

image

Figure 1. Conservation and expression of ArcZ among diverse enterobacterial species. A. Alignment of enterobacterial arcZ sequences including upstream promoter region (LT2: Salmonella typhimurium LT2; K12: Escherichia coli K12; Cko: Citrobacter koseri; Ent: Enterobacter Sp.638; Kpn: Klebsiella pneumoniae; Spr: Serratia protemaculans; Ype: Yersinia pestis; Plu: Photorhabdus luminescens). The putative −10 and the −35 boxes of the arcZ promoter are indicated; ‘1’ denotes the transcriptional start site. Red nucleotides are conserved in all 8 genera. STOP codons of the arcB and yhbL genes are underlined; the arcB CDS overlaps with the transcriptional terminator of the arcZ gene. B. Gene synteny analysis of the arcZ gene. The genomic context of the arcZ gene in diverse enterobacterial species is shown in relation to homologues of the yhbL and arcB genes. C. ArcZ expression in WT and ΔarcZ cells detected by Northern blot analysis. Total RNA extracts of Salmonella WT and ΔarcZ cells collected at several stages of growth (OD600 of 0.5, 1.0, 2.0, and 3 h or 6 h after cells had reached an OD600 of 2.0) were subjected to Northern blot analysis and probed with a radioactively labelled oligonucleotide specific to the ArcZ 3′ end. Co-migration of a radioactively labelled DNA marker was used for size determination. Probing for 5S rRNA served as loading control.

Download figure to PowerPoint

The 3′ end of the arcZ gene is highly conserved among enterobacterial species, and this includes the ρ-independent terminator, which partially overlaps the 3′ coding sequence (CDS) of arcB located downstream on the opposite strand (Fig. 1A). In every case, the 5′ border defined by the yhbL gene, remained in close proximity to arcZ. These genomic features may indicate functional relatedness within the yhbL-arcZ-arcB gene cluster. yhbL (a.k.a. elbB) encodes a protein whose function is largely unknown, except that it cross-reacts with the σ70 subunit of E. coli RNA polymerase (Ueshima et al., 1992), and that it impacts upon lycopene biosynthesis (Jin and Stephanopoulos, 2007 and references therein). ArcB, encoded by the downstream gene, is a well-investigated aerobic respiration control sensor protein that transmits alterations in oxygen availability to the global transcriptional regulator, ArcA (Iuchi et al., 1990; Bekker et al., 2006).

Co-immunoprecipitation (coIP) experiments in E. coli (Wassarman et al., 2001; Zhang et al., 2003) and Salmonella (Sittka et al., 2008; 2009) have shown that ArcZ associates with the bacterial RNA chaperone, Hfq. Specifically, using deep sequencing technology we found that ArcZ was ∼10% of the Hfq-bound Salmonella sRNAs (Sittka et al., 2008); only InvR sRNA, a pathogenicity island-encoded repressor of outer membrane protein synthesis (Pfeiffer et al., 2007), showed a higher rate of recovery with Hfq.

Hfq-associated sRNAs commonly use antisense mechanisms to either positively or negatively regulate trans-encoded target mRNAs; the function of Hfq is to aid the formation of the typically weak and imperfect RNA duplexes between regulator and target (Valentin-Hansen et al., 2004; Aiba, 2007). Although Hfq-dependent sRNAs modulate translational initiation by interfering with 30S ribosome binding, alterations of target mRNA levels are also often observed (Vogel and Wagner, 2007; Waters and Storz, 2009).

Investigations in various model bacteria predicted a central role for Hfq and its associated sRNAs in bacterial physiology and virulence (Tsui et al., 1994; Ding et al., 2004; Sonnleitner et al., 2006; Guisbert et al., 2007). The phenotypes of hfq mutants have been intensely studied in Salmonella (Brown and Elliott, 1996; Bang et al., 2005; Figueroa-Bossi et al., 2006; Sittka et al., 2007; 2008; Wilson et al., 2007; Yoon et al., 2009). Hfq controls nearly a fifth of all Salmonella genes (Sittka et al., 2008), consistent with the predicted global function of Hfq in post-transcriptional regulation (Valentin-Hansen et al., 2004). To achieve a thorough understanding of this layer of bacterial gene expression control, a systematic analysis of abundant and well-conserved Hfq-dependent sRNAs must be pursued (Vogel, 2009).

The present study addressed the expression and cellular targets of the ArcZ sRNA in Salmonella. We found that increasing the level of ArcZ has an exceptionally pleiotropic effect that alters the expression of > 750 Salmonella genes without causing marked changes in growth rate. In contrast to these global changes, we also identified specific mRNA targets of ArcZ, including the sdaC and STM3216 mRNAs which are repressed at their Shine-Dalgarno (SD) sequence by pairing of the conserved 3′ region of ArcZ. In addition, we discovered that ArcZ acts to repress the tpx mRNA within the CDS, downstream of the ribosome binding site.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Note added in proof
  9. References
  10. Supporting Information

Expression of ArcZ in Salmonella

To study expression of ArcZ in Salmonella, we examined wild-type (WT) and isogenic ΔarcZ mutant strains at different stages of growth (Fig. 1C). ArcZ-specific transcripts were detected with a radioactively labelled oligonucleotide recognizing the conserved 3′ end of ArcZ (Fig. 1A). We identified three ArcZ-specific bands: two larger species (about 120–130 nt) corresponding to primary ArcZ transcripts and an abundant smaller transcript (∼50 nt), representing the processed 3′ fragment of ArcZ (Fig. 1C). This confirms that the processed form of ArcZ is predominantly present within the cell as previously observed in E. coli (Argaman et al., 2001).

When E. coli is grown in rich media, the processed ArcZ RNA only accumulates when cultures enter stationary phase (Argaman et al., 2001; Wassarman et al., 2001). Levels of Salmonella ArcZ are also highest at stationary phase, with a peak about 3 h after cells had reached an OD of 2.0 (Fig. 1C); unlike in E. coli, Salmonella ArcZ is readily detectable at early stages of growth.

To correlate transcript abundance with promoter activity and to monitor ArcZ-specific expression at different stages of growth we fused the arcZ promoter (ParcZ) to a gfp reporter gene. We used the constitutive promoter, PLtetO-1 (Lutz and Bujard, 1997), as a control to distinguish arcZ promoter activity from growth-dependent variation in the accumulation of green fluorescent protein (GFP) (Fig. S1). Transcriptional activity of PLtetO-1 was higher than ParcZ under all growth conditions. The activity of the ParcZ reporter correlated with accumulation of the 120 nt primary transcript detected in Fig. 1C; expression of both promoters increased towards stationary phase. These data are consistent with primary and processed ArcZ transcripts having different stabilities (Vogel et al., 2003; Sittka et al., 2009): the primary transcript is rapidly processed after transcription (half-life ∼3 min), which correlates well with arcZ promoter activity, and the processed version of ArcZ is stable (half-life > 32 min) and accumulates in the cell during growth.

Expression of ArcZ requires Hfq but is independent of arcB, yhbL or rpoS

The arcB and yhbL genes show conserved synteny with arcZ (Fig. 1B), which might reflect important functional relationships. Specifically, because arcB and yhbL encode proteins predicted to control transcription (see Introduction), this situation resembled the gcvB, oxyS or sgrS sRNA loci in which one of the flanking genes encoded the cognate transcription factor (Altuvia et al., 1997; Urbanowski et al., 2000; Vanderpool and Gottesman, 2004). To test whether ArcZ was controlled by a similar mechanism, arcB or yhbL were disrupted, and ArcZ expression was determined in these mutant strains after growth to late stationary phase (Fig. 2). Both mutants exhibited ArcZ RNA levels comparable to the parental wild-type strain, suggesting that arcB and yhbL did not impact upon ArcZ expression, at least under the tested growth condition.

image

Figure 2. Expression of ArcZ in various Salmonella mutant backgrounds. Northern blot detection of ArcZ as above but comparing Salmonella WT, ΔarcZ, ΔyhbL, ΔarcB, Δhfq and ΔrpoS strains as indicated above the lanes. Samples were taken following growth to late stationary phase (6 h after cells had reached an OD600 of 2.0).

Download figure to PowerPoint

The intracellular accumulation of many sRNAs is critically dependent on the RNA chaperone Hfq (Wassarman et al., 2001; Urban and Vogel, 2007). Probing for ArcZ in an hfq mutant background revealed complete absence of both primary and processed ArcZ (Fig. 2). Importantly, the absence of Hfq did not impair arcZ transcription as judged by the unchanged activity of the ParcZ reporter in Δhfq (Fig. S2). Thus, Hfq facilitates ArcZ accumulation at the level of transcript stability.

The alternative sigma factor, σS (encoded by rpoS), plays a major role in the expression of stationary phase genes in enterobacteria (Klauck et al., 2007). However, despite the maintenance of the cytosine residue at position −13 upstream of the transcriptional start site (Fig. 1A), which is one hallmark of σS-dependent promoters (Weber et al., 2005), σS was not required for stationary phase accumulation of ArcZ (Fig. 2).

The 3′ overlapping positions of the arcZ and arcB genes (Fig. 1A), which is reminiscent of the E. coli gadY-gadX or E. faecalis par loci (Weaver et al., 1996; Opdyke et al., 2004), prompted us to examine a potential cis-regulatory relationship between their transcripts. As shown above, mutation of arcB did not significantly alter ArcZ abundance (Fig. 2); likewise, abrogation of arcB transcription did not alter the position of ArcZ termination as determined by 3′ RACE (Fig. S3). To monitor the reverse case, that is, a potential cis-effect of ArcZ on the arcB messenger, we collected RNA samples from WT and arcZ mutant cells grown to late stationary phase and determined arcB transcript levels by quantitative real-time PCR. Abundance of arcB mRNA was not significantly altered in ΔarcZ in both the RT-PCR (Fig. S3) and in the microarray analyses presented further below (Table 1). In addition, expression of ArcA/B-controlled genes (Liu and De Wulf, 2004) did not change upon arcZ deletion (Fig. S3). In summary, under the condition most relevant for the present work (stationary phase), the potential cis-regulatory relationship of ArcZ and arcB might be marginal.

Table 1.  Genes differentially regulated in the absence of arcZ.
GeneIDFold-regulationaDescriptionb
  • a.

    Fold-regulation obtained by transcriptomic analysis of ΔarcZ cells at OD = 2 + 6 h on Salmonella whole genome microarrays. Genes that were at least twofold differentially regulated and had a P-value ≤ 0.05 are listed.

  • b.

    Description was based on the annotation found at Colibase (http://xbase.bham.ac.uk/colibase/).

sdaCSTM2970+9.1Putative HAAAP family serine transport protein
tpxSTM1682+4.3Thiol peroxidase
fimASTM0543+4.2Major type 1 subunit fimbrin (pilin)
yhbLSTM3327+3.3Sigma cross-reacting protein (SCRP-27A)
sdaBSTM2971+3.2l-serine dehydratase

Identification of the mRNAs directly targeted by ArcZ

To determine the biological functions of ArcZ in Salmonella we first took a transcriptomic approach: wild-type and ΔarcZ strains were grown to late stationary phase (6 h after cells had reached an OD600 of 2.0) and expression profiles were compared on whole genome microarrays. Only 5 of 4716 Salmonella genes showed altered transcript abundance greater than twofold in the absence of ArcZ (Table 1). The sdaC gene, encoding a serine transport protein was most strongly upregulated (9.1-fold); interestingly, sdaB, the second gene in the dicistronic sdaCB operon, showed only a 3.2-fold increase in expression. In addition, tpx, a periplasmic thiol peroxidase (4.3-fold), and fimA, a major type-1 fimbrial subunit (4.2-fold), were upregulated. Note that we consider the 3.3-fold increase in yhbL mRNA levels to reflect the ΔarcZ mutant construction, which removed parts of the potential yhbL transcription control region.

Analyses of sRNA deletion strains successfully identified direct mRNA regulations, e.g. ompX mRNA repression by CyaR sRNA (Papenfort et al., 2008), but can be complicated by secondary effects resulting from the altered expression of a direct target gene on unrelated genes. To minimize such indirect effects, we examined global mRNA changes following short-term expression (Masséet al., 2005; Papenfort et al., 2006) of ArcZ from an arabinose-inducible PBAD promoter. Salmonella carrying a pBAD-ArcZ expression plasmid or a pBAD control vector were induced for 10 min at early stationary phase, and the resulting changes in mRNA abundance were scored on microarrays. Eight transcripts showed significantly altered transcript abundance (> threefold) upon ArcZ pulse-expression (Table 2). Importantly, the sdaCB and tpx genes, which were both upregulated in ΔarcZ (Table 1), were repressed upon ArcZ overexpression. The Salmonella-specific STM3216 gene was also strongly downregulated (−6.2-fold), making it another potential candidate for ArcZ-mediated repression (Table 2). In addition, the STM1281 and STM3601 genes were repressed by 3.3-fold and 3.1-fold, respectively, whereas two other genes (STM0893 and gcd) were upregulated by ArcZ in these experiments.

Table 2.  Genes differentially regulated upon ArcZ overexpression.
GeneIDFold-regulationaDescriptionb
  • a.

    Fold-regulation obtained by transcriptomic analysis after transient induction of ArcZ on Salmonella whole genome microarrays. Genes that were at least threefold differentially regulated and had a P-value ≤ 0.05 are listed.

  • b.

    Description was based on the annotation at Colibase (http://xbase.bham.ac.uk/colibase/; accessed February 2009).

STM0893STM0893+4.0Putative Fels-1 prophage integrase
gcdSTM0169+3.1Glucose dehydrogenase
STM3601STM3601−3.1Putative phosphosugar isomerase
STM1281STM1281−3.3Putative inner membrane protein
sdaBSTM2971−4.8l-serine dehydratase
tpxSTM1682−5.7Thiol peroxidase
STM3216STM3216−6.2Putative methyl-accepting chemotaxis protein
sdaCSTM2970−7.6Putative HAAAP family serine transport protein

Validation of ArcZ target genes

To confirm the negative effect of ArcZ on the expression of sdaC, tpx and STM3216 genes, we C-terminally tagged the chromosomal copies of these genes with a triple FLAG epitope to compare protein abundance in wild-type or ΔarcZ cells. Western blot analysis at various growth phases (Fig. 3) showed that all proteins accumulated to higher levels in ΔarcZ compared with the WT strain, albeit at different stages of growth. Specifically, the increase in levels of SdaC::FLAG and STM3216::FLAG proteins were visible in exponential (Fig. 3A and C; lanes 1–3 versus 6–8) but not in stationary phase (lanes 4–5 versus 9–10), suggesting that ArcZ controls SdaC and STM3216 synthesis in rapidly growing cells. In contrast, levels of Tpx::FLAG increased by about threefold in ΔarcZ cells at late stationary phase (Fig. 3B, lanes 4–5 versus 9–10). The observed growth phase-dependent regulations indicate that ArcZ effects might depend on the availability of potential target mRNAs, and raises the possibility that ArcZ regulates additional genes under other environmental conditions.

image

Figure 3. SdaC, Tpx and STM3216 protein expression in WT and sraH mutant cells. Left panel: Graphical presentation of the sdaC (A), tpx (B) and STM3216 (C) genomic localization in Salmonella, drawn to scale. Grey bars indicate sequences present in the GFP-reporter system. Right panel: Western blot analysis of SdaC::FLAG (A), Tpx::FLAG (B) and STM3216::FLAG (C) proteins in WT and sraH mutant strains. Whole-cell protein samples were collected at various stages of growth (OD600 of 0.5, 1.0, 2.0, 6 h after cells had reached an OD600 of 2.0 and 24 h after inoculation) and subjected to Western blot analysis using a monoclonal anti-FLAG antibody. GroEL was used as loading control.

Download figure to PowerPoint

ArcZ represses the sdaC and STM3216 mRNAs by RBS sequestration

Bacterial sRNAs generally recognize the 5′ regions of target mRNAs at the ribosome binding site to repress translational initiation (Vogel and Wagner, 2007; Waters and Storz, 2009). To identify the binding sites of ArcZ within our three major targets, we searched the 5′ regions of sdaC, STM3216 and tpx mRNAs for potential RNA-duplex formation using the RNAhybrid and TargetRNA algorithms (Rehmsmeier et al., 2004; Tjaden et al., 2006). This analysis did not reveal convincing results for sdaC or tpx, but the RNAhybrid algorithm did predict stable ArcZ pairing with the entire SD sequence of STM3216, and which would be likely to cause translational repression (Fig. 4A).

image

Figure 4. Post-transcriptional targeting of STM3216, sdaC and tpx by ArcZ. Left panel: Schematic presentation of proposed duplex formation between STM3216 (A), sdaC (B) or tpx (C) mRNAs and ArcZ sRNA. RNAhybrid (available at http://bibiserv.techfak.uni-bielefeld.de/rnahybrid) was used to predict potential interactions. Numbers indicate relative position to the translational start site of STM3216, sdaC and tpx or position downstream of the transcriptional start site of arcZ. Two arrows denote base-pair mutations inserted into one of the sequences to generate STM3216*, sdaC*, tpx* or ArcZ*. SD and AUG sequences are marked in bold letters. Right panel: Western blot analysis of Salmonella cells harbouring plasmids pBAD-ArcZ or mutant plasmid, pBAD-ArcZ*, in combination with either wild-type or mutant reporter plasmids, as indicated. Whole-cell samples were collected 2 h after induction of ArcZ by 0.2% l-arabinose, and STM3216::GFP (A), SdaC::GFP (B) or Tpx::GFP (C) protein levels were determined by Western blot analysis. GroEL was used as loading control.

Download figure to PowerPoint

To test the prediction, we used an established two-plasmid system that involves coexpression of the sRNA with a translational target gene fusion to the amino terminus of GFP (Urban and Vogel, 2007). The 5′ region of STM3216, spanning 31 bp of the 5′ UTR and the first 20 codons, was cloned into the pXG10 low-copy gfp fusion vector (Fig. 3C). Transcription of the fusions was driven by the constitutive PLtetO-1 promoter to specifically assay post-transcriptional regulation, while ArcZ was expressed from the inducible pBAD-ArcZ plasmid. The SalmonellaΔarcZ strain transformed with both the pXG10-derived STM3216::gfp fusion and pBAD-ArcZ plasmid was grown to an OD600 of 0.1, split into two cultures, while only one was treated with l-arabinose. Cultivation was continued for additional two hours, before ArcZ-dependent regulation was determined by Western blot analysis of STM3216::GFP (Fig. 4A). ArcZ severely reduced STM3216::GFP expression by approximately 12-fold (Fig. 4A; lanes 1–2), but did not regulate gfp alone (pXG-1 control plasmid; Fig. S4), arguing that repression by ArcZ was specific and involved the cloned 5′ region of STM3216. To validate the predicted ArcZ–STM3216 interaction, we introduced point mutations in ArcZ (G70 to C) and STM3216 (C−12 to G), yielding the ArcZ* and STM3216* alleles (Fig. 4A). Mutation of the sRNA drastically reduced its ability to repress STM3216::GFP (to ∼1.5-fold; Fig. 4A, lanes 3–4). Likewise, the STM3216* mutation upstream of the SD clearly reduced ArcZ-dependent target repression (Fig. 4A, lanes 5–6). Repression was restored by combination of the compensatory ArcZ* and STM3216*::gfp alleles (Fig. 4A, lanes 7 and 8), supporting our prediction that ArcZ inhibits STM3216 via a ∼14 bp duplex around the SD sequence.

Irrespective of our initial failure to predict ArcZ binding sites, the rapid downregulation of sdaC and tpx following ArcZ expression strongly argued that these mRNAs were also direct targets. To verify direct regulation, we constructed sdaC::gfp and tpx::gfp fusions as above, cloning the 5′ region of sdaC from the previously predicted CRP-dependent transcriptional start site [position −123 in Salmonella (Hollands et al., 2007)] through codon 10, or the 5′ region of tpx from its transcriptional start site (position −32 in Salmonella;Kim et al., 1996) through codon 70 (Fig. 3A and B). Both fusions were specifically downregulated by pBAD-ArcZ (sdaC::gfp, 3.6-fold; tpx::gfp, 6.4-fold; Fig. 4A and B; lanes 1–2), confirming that ArcZ repressed sdaC and tpx at the post-transcriptional level. In addition, regulation was abrogated by the ArcZ* mutation, suggesting that the same sequence of ArcZ that targets STM3216 was critical for sdaC and tpx regulation, too.

Given that the conserved 3′ region of ArcZ contains a validated anti-SD sequence (Fig. 4A), we searched again for potential ArcZ duplexes with the tpx or sdaC mRNAs, this time querying the 3′ end of ArcZ against the cloned sdaC and tpx sequences. RNAhybrid predicted an unusually short ArcZ duplex of 5 + 5 base-pairs with the SD of sdaC (Fig. 4B) which would be formed exclusively by Watson-Crick base-pairs, interrupted by an adenosine bulge at the 6th position in sdaC.

To validate this interaction in vivo, we mutated the sdaC::gfp reporter by changing C−12 to G (designated sdaC*), i.e. in a compensatory fashion with respect to ArcZ* (Fig. 4B). The target mutation increased the basal expression of the reporter by about twofold (Fig. 4B; compare lanes 1 and 5), likely because the C to G change at this position increases interaction with the 16S ribosomal subunit by one base pair and so enhances translation. Importantly, however, sdaC*::gfp was fully resistant to ArcZ (Fig. 4B; lanes 5–6), but repression was restored when sdaC*::gfp was combined with ArcZ* (∼3.5-fold; Fig. 4B, lanes 7–8). Therefore, ArcZ represses both sdaC and STM3216 by masking their SD sequence, inhibiting translational initiation and likely accelerating the decay of the untranslated mRNAs (Deana and Belasco, 2005; Dreyfus, 2009).

ArcZ targets the coding sequence of tpx mRNA

Unlike for sdaC and STM3216, no ArcZ interaction with the ribosome binding site was predicted for tpx mRNA. In contrast, the RNA hybrid algorithm predicted an imperfect RNA duplex of the 3′ region of ArcZ with the +10 to +26 region in the CDS of tpx (Fig. 4C). In favour of this prediction, a tpx*::gfp fusion (C to G point mutation at tpx position +22) was observed to be resistant to ArcZ (Fig. 4C, lanes 5–6), as compared with 6.4-fold repression by ArcZ (lanes 1–2). We note that tpx*::gfp produces considerably less fusion protein than tpx::gfp, despite the fact that the transversion at position +39 did not alter the amino-acid sequence of Tpx*::GFP. We tentatively explain this effect by the rare ACG (Thr) codon in tpx*::gfp instead of the parental AGG (Thr) codon. Most importantly, however, regulation was restored by combination of the compensatory tpx*::gfp and pBAD-ArcZ* plasmids (lanes 7–8), validating the formation of the ArcZ-tpx duplex in vivo. This experiment proves that ArcZ binds tpx mRNA within the coding sequence itself, downstream of known translational control elements.

Hfq is required for ArcZ mediated target repression

To evaluate the role of Hfq in the post-transcriptional control of the STM3216, sdaC and tpx target genes, we cotransformed a SalmonellaΔhfq strain with the pBAD-ArcZ and the STM3216::gfp, sdaC::gfp or tpx::gfp plasmids. Fusion regulation was examined by induced ArcZ expression as above, and found to be fully abrogated in Δhfq (Fig. S5). Interestingly, several E. coli sRNAs that require Hfq for intracellular stability similarly to ArcZ retain residual regulatory activity when overexpressed in E. coliΔhfq (Sledjeski et al., 2001; Urban and Vogel, 2007). In contrast, Hfq seems absolutely essential for ArcZ-mediated target repression.

Pleiotropic effects of increased ArcZ expression

Pioneering discoveries in the field of sRNA-mediated gene regulation were based on fortuitous overexpression of afore unknown sRNA genes in E. coli, e.g. the discoveries of the micF or dsrA sRNA genes through their multicopy effects on OmpF or H-NS protein synthesis (Mizuno et al., 1984; Sledjeski and Gottesman, 1995). To further understand biological functions of ArcZ, we constructed two different expression plasmids with a ColE1 origin: Plasmid parcZ carries arcZ with its native promoter, and plasmid pPL-ArcZ expresses the sRNA at a higher level, from the constitutive PLlacO-1 promoter (Lutz and Bujard, 1997).

Under standard conditions in Luria–Bertani (LB) media, neither plasmid caused a growth defect (Fig. S6). However, SDS-PAGE of whole-cell proteins collected throughout growth to late stationary phase revealed drastic changes in the abundance of major Salmonella proteins in the presence of both arcZ expression plasmids. These included OmpC and OmpD, the two most abundant outer porins of Salmonella, and FliC, the main structural component of the flagellar complex, which were depleted in stationary phase cells (Fig. 5A). We confirmed ArcZ-dependent flagellar repression at the phenotypic level; severe motility defects were observed in strains carrying parcZ and pPL-ArcZ, compared with the WT strain (Fig. 5B).

image

Figure 5. Characterization of ArcZ overexpression in Salmonella. A. Upper panel: Whole-cell protein patterns of WT and ArcZ overexpression strains. Salmonella proteins derived from total cell lysates of several growth conditions [OD600 of 0.5 (lanes 1, 6, 11); 1.0 (lanes 2, 7, 12); 2.0 (lanes 3, 8, 13); 3 h (lanes 4, 9, 14) and 6 h (lanes 5, 10, 15) after cells had reached an OD600 of 2.0] were separated on a 12% SDS page and stained for abundant proteins with coomassie blue. Co-migration of a prestained marker was used for size estimation. B. Swimming motility of the same strains on soft agar. Overnight cultures were used to inoculate soft-agar plates and checked for motility after 6 h. C. Determination of in vivo copy number of ArcZ. Levels of full-size and processed ArcZ from wild-type Salmonella or ΔarcZ cells carrying parcZ or pPL-ArcZ plasmids were compared with serial dilutions of in vitro synthesized ArcZ transcripts (0.4, 2, 10 and 20 ng). Total RNA isolated from three stages of growth [OD600 of 0.5 (lanes 5, 8, 11), 2.0 (lanes 6, 9, 12) and 6 h after cells had reached an OD600 of 2.0 (lanes 7, 10, 13)] as well as in vitro transcripts, were subjected to Northern blotting and probed with a radioactively labelled oligonucleotide specific to the ArcZ 3′ end. Probing for 5S rRNA served as loading control.

Download figure to PowerPoint

To evaluate the degree of ArcZ expression, we determined in vivo copy numbers of the RNA in Salmonella wild-type, or in ΔarcZ carrying either plasmid parcZ or pPL-ArcZ (Fig. 5C). Total RNA was prepared at three representative phases of Salmonella growth and compared with serial dilutions of in vitro-synthesized ArcZ RNA. Both plasmids caused accumulation of full-length and processed ArcZ (Fig. 5C), yielding substantial build-up of the processed sRNA, which was most pronounced in exponentially growing cells. At OD600 of 2.0, parcZ and pPL-ArcZ increased levels of processed ArcZ by 8-fold and ∼15-fold respectively. Under the same conditions, the level of the primary ArcZ transcript was increased by ∼16-fold or ∼38-fold in the parcZ or pPL-ArcZ harbouring strains, respectively, suggesting a rate-limiting step in ArcZ processing. Comparison of in vitro and in vivo transcript abundance determined ∼18, ∼150 or ∼260 molecules of processed ArcZ per WT, parcZ or pPL-ArcZ cells, respectively, at an OD600 of 2.0. Copy numbers at other growth conditions are given in Table S1.

Transcriptomic analysis reveals a pleiotropic effect of ArcZ overexpression in Salmonella

To corroborate the global regulations observed upon ArcZ overexpression, we performed microarray analysis of ΔarcZ harbouring a control plasmid or the parcZ plasmid at late stationary phase (6 h after cells had reached an OD600 of 2.0). Strikingly, ∼16% (757/4716) of the Salmonella genes showed differential expression (≥ twofold change; P-value ≤ 0.05), 386 of which were upregulated and 371 downregulated (Table S2).

Classification of the 757 Salmonella genes with altered expression according to KEGG database (http://www.genome.jp/kegg/) identified particular functional categories (Table S3). In line with the observed loss of motility (Fig. 5B) pathways containing genes for flagellar assembly or chemotaxis were most strongly affected by ArcZ (87–95% of genes in each pathway; Table S3). Intriguingly, loss of Hfq function caused a similar motility defect in Salmonella (Sittka et al., 2007), which prompted us to compare transcriptomic profiles from the ArcZ overexpression and the Δhfq mutant strains. Of the total of 734 Hfq-dependent genes identified at early stationary phase in S. typhimurium (Sittka et al., 2008), more than half (381 genes) were also affected by increased expression of ArcZ (Fig. S7), indicating a functional overlap.

A third group of genes significantly affected by ArcZ were involved in Salmonella pathogenicity and type III secretion. Notably, 82% of the Salmonella pathogenicity island 1 (SPI-1) genes, which are essential for invasion of host cells (Ellermeier and Slauch, 2007), displayed strong downregulation following ArcZ overexpression (Table S3), suggesting global suppression of virulence gene expression. Again, these results mirrored the transcriptomic analysis of a Δhfq strain where 90% of all SPI-1 genes were downregulated in the absence of Hfq (Sittka et al., 2008).

ArcZ overexpression alters the repertoire of Hfq-bound sRNAs

To test our hypothesis that ArcZ overexpression had a direct effect upon Hfq-dependent gene regulation, we analysed Hfq-bound RNAs by 454 pyrosequencing of cDNA (Sittka et al., 2008; Sittka et al., 2009). Specifically, we compared Hfq-bound RNAs in WT, carrying control vector or pPL-ArcZ, in late stationary phase. From a total number of 118 255 cDNA sequences obtained from both strains, 85 688 (∼73%) mapped to the Salmonella chromosome, and 210 mapped to one of the Salmonella virulence plasmids. A total number of 13 707 sequences matched known mRNAs, with 7720 derived from WT and 5987 from ArcZ overexpression samples (Fig. S8). The largest number of sequence matches was found for known sRNAs (26 868 sequences; 31% of all matched sequences), and revealed a drastically different sRNA association pattern in the two strains. Specifically, 20 628 sequences of known sRNAs were identified in the ArcZ overexpression strain, but only 6240 sequences were seen in the WT strain. ArcZ represented ∼23% of all Hfq-bound sRNAs from the WT strain (1454 matches), making it the most abundant Hfq-bound sRNA under the tested growth condition (Fig. S8). In the ArcZ overexpression strain, ArcZ itself constituted 61% (12 547 sequences) of all matched sRNAs (Fig. 6A, Fig. S9).

image

Figure 6. Effects of ArcZ overexpression on Hfq-bound sRNAs. A. Northern blot analysis of Hfq::FLAG co-immunoprecipitated sRNAs. RNA samples co-immunoprecipitated from a chromosomally tagged hfq::FLAG strain harbouring the pJV300 control plasmid or the pPL-ArcZ expression plasmid were subjected to Northern blot analysis with sRNA-specific oligonucleotide probes; extracts of the WT strain carrying the pJV300 control vector (expressing non-FLAG-tagged Hfq) served as control. Total RNA equivalent to 1/10 of extract used in the immunoprecipitations was run in parallel (lanes ‘total’). B. Schematic representation of changes in Hfq-bound sRNAs induced by increased ArcZ expression. Fold-changes in sRNAs recovered from Hfq::FLAG were calculated by dividing normalized numbers of sequences matching known Salmonella sRNA genes from WT (hfq::FLAG + pJV300) and a strain with increased levels of ArcZ (hfq::FLAG + pPL-ArcZ).

Download figure to PowerPoint

As well as increasing the total number of bound sRNAs by more than threefold, ArcZ overexpression caused a re-profiling of the Hfq-bound sRNAs (Fig. 6B), and increased numbers of Hfq-bound stress-related sRNAs, including RprA (cell surface stress; Majdalani et al., 2001), RyhB (iron starvation; Massé and Gottesman, 2002), MicA and RybB (envelope stress; Papenfort et al., 2006; Valentin-Hansen et al., 2007), MicF (response to antimicrobial compounds; Delihas and Forst, 2001) and SgrS (sugar stress; Vanderpool and Gottesman, 2004) (Fig. 6B). High levels of these sRNAs suggested pronounced activation of stress regulons. Induction of the σE-dependent RybB and MicA sRNAs was consistent with approximately threefold increased level of the σE encoding rpoE mRNA (Table S2). The upregulation of RybB may account for some of the depletion of OmpD and OmpC proteins observed upon ArcZ overexpression (Figs 5A and 6; Table S2), because RybB represses the ompC and ompD mRNAs (Papenfort et al., 2006).

Our analysis also identified a number of sRNAs that showed reduced Hfq binding upon overexpression of ArcZ (Fig. 6B). These include the CRP-dependent sRNA, CyaR (Papenfort et al., 2008), which was decreased most drastically (∼14-fold). In addition, recovery of InvR sRNA encoded on the major virulence island, SPI-1 (Pfeiffer et al., 2007), was approximately fivefold decreased, which might reflect the observed repression of SPI-1 virulence gene upon increased ArcZ production (Table S3).

Taken together, the deep sequencing data revealed a profound change in the range of sRNAs bound by Hfq caused by ArcZ overexpression, indicating that a potential competition for Hfq binding might contribute to the observed pleiotropy of ArcZ action by perturbing post-transcriptional regulations.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Note added in proof
  9. References
  10. Supporting Information

The systematic screens for bacterial sRNAs identified an abundant class of ‘core’ sRNAs, conserved since the origin of the enterobacterial clade. Clearly, the functional analysis of such sRNAs will contribute to the unraveling of the cellular pathways that are controlled by sRNAs (and Hfq) in Gram-negative species. Our study has revealed that Salmonella ArcZ silences genes of several pathways, including amino acid uptake, oxidative stress and chemotaxis. This gene silencing is mediated by direct interaction of ArcZ with targets either at the conventional RBS or a CDS binding site. However, ArcZ can also cause unprecedented global regulation by a single sRNA. Potential causes of indirect regulations by ArcZ will be discussed.

The ArcZ regulon

The identification of the direct target mRNAs of a particular sRNA is an important step in defining the regulatory circuits that involve post-transcriptional control. In this study, we used two complementary approaches for target mRNA determination: analysis of pulse-expression data identified seven mRNAs as potential targets of ArcZ (Table 2), three of which (sdaC, tpx and STM3216) we have confirmed to be directly repressed by ArcZ antisense pairing (Fig. 4). All three genes show increased protein synthesis, and two (sdaC, tpx) elevated mRNA levels in ΔarcZ (Fig. 3, Table 1).

At first glance, a functional overlap between these three top ArcZ targets is not apparent. The sdaC gene encodes a serine-specific transporter in E. coli and is co-transcribed with sdaB, which encodes the related serine deaminase. A potential role of the sdaCB operon in the catabolism of serine was suggested (Shao et al., 1994), and supported by observations that the sdaCB locus is subject to catabolite repression via direct binding of the Crp protein to the sdaC promoter (Hollands et al., 2007). Two additional studies identified sdaC as an essential component for bacteriophage C1 adsorption (Likhacheva et al., 1996) and action of the antibacterial colicin V (Gerard et al., 2005); accordingly, we speculate that post-transcriptional control by ArcZ may help to counteract phage and colicin toxicity.

The tpx mRNA encodes a ∼20 kDa periplasmic protein that confers altered antioxidant capabilities to E. coli; specifically, loss-of-function of this thioredoxin-linked thiol peroxidase renders E. coli more sensitive to diverse oxidative compounds (Cha et al., 1995). The functions of the Salmonella-specific STM3216 locus are discussed further below, but these do not identify obvious links between ArcZ expression and the other two target genes.

In contrast, we suggest that the cellular function of ArcZ may be linked to the transcriptional regulator ArcA, which controls both sdaC and tpx in E. coli (Kim et al., 1999; Liu and De Wulf, 2004), and regulates STM3216 in Salmonella (Evans, 2008). Intriguingly, the shared gene synteny suggests an intimate connection between arcZ and arcB, which codes for the cognate sensor kinase of the ArcA response regulator. A link with the arc regulon would fit with other bacterial sRNAs that are – physically or functionally – associated with transcription or sigma factor genes, and belong to distinct bacterial stress response pathways (Altuvia et al., 1997; Massé and Gottesman, 2002; Vanderpool and Gottesman, 2004; Guillier and Gottesman, 2006; Papenfort et al., 2006).

An increasing number of functional studies have revealed that ‘core’ sRNAs often directly and simultaneously act on multiple mRNAs, to command both repression and activation of protein synthesis (Majdalani et al., 2005; Papenfort and Vogel, 2009). Some of these sRNAs recognize a large number of structurally unrelated mRNAs of genes with related functions; e.g. GcvB controls multiple ABC transporter mRNAs (Sharma et al., 2007); RybB and OmrAB each recognize a distinct set of messengers of outer membrane proteins (Papenfort et al., 2006; Guillier and Gottesman, 2008), and RNAIII of Staphylococcus aureus regulates multiple virulence factor mRNAs (Boisset et al., 2007; Huntzinger et al., 2005; Novick and Geisinger, 2008). E. coli RyhB sRNA has a similar ability, and directly interacts with multiple mRNAs that encode proteins that are associated with iron metabolism and starvation (Geissmann and Touati, 2004; Prevost et al., 2007; Vecerek et al., 2007; Desnoyers et al., 2009). In contrast, ArcZ appears to be a regulator of multiple target genes with apparently unrelated functions, and resembles the properties of CyaR sRNA, which not only controls OmpX synthesis (Johansen et al., 2008; Papenfort et al., 2008) but also the luxS gene involved in quorum sensing (De Lay and Gottesman, 2009).

The conserved 3′ region of ArcZ targets the coding sequence of tpx mRNA

Bacterial sRNA regulators commonly target mRNAs in the 5′ UTR, at the RBS or at other sites that control the translational activity of the mRNAs (Wagner, 2009; Waters and Storz, 2009). The ArcZ interactions with sdaC and STM3216 conform to this rule, and because in both cases the SD sequence is masked, it is safe to predict that these mRNAs are inhibited at the earliest step of translational initiation, i.e. 30S ribsome binding.

Strictly speaking, the RBS is constituted by the Shine-Dalgarno (SD) sequence and the start codon (AUG), i.e. the two key elements of mRNA that anchor the initiating ribosome. However, initiating ribosomes occupy a wider region on mRNA, ranging from residue −20 in the 5′ UTR to residue +19 in the CDS (Beyer et al., 1994; Hüttenhofer and Noller, 1994), suggesting that sRNAs could also control translation by sequence sequestration in the CDS. Indeed, we recently defined a ‘five codon window’ in which short base-pairing of sRNAs to the mRNA CDS inhibits 30S binding without any RBS contacts (Bouvier et al., 2008). The ArcZ–tpx interaction proposed here involves tpx codons 4–9 (Fig. 4C), and is therefore still likely to inhibit 30S ribosome association. In addition, we recently showed that MicC sRNA represses ompD mRNA in the deep CDS, by binding to codons 23–26 (Pfeiffer et al., 2009). Instead of impacting on translational initiation, MicC was shown to repress ompD by accelerating RNase E-dependent mRNA decay. In analogy, ArcZ might additionally control tpx by mRNA destabilization. Because both ArcZ is associated with Hfq (Wassarman et al., 2001; Sittka et al., 2008), and Hfq forms complexes with RNase E, either directly (Morita et al., 2005) or by RNA bridges (Worrall et al., 2008), the silencing might involve a direct recruitment of RNase E to tpx mRNA, as previously proposed for SgrS and RyhB mediated mRNA destabilization in E. coli (Morita et al., 2005). While the precise molecular mechanism of tpx repression remains to be established, our discovery of an additional example of targeting downstream of AUG supports the hypothesis that CDS targeting is common and should be considered in target site predictions (Pfeiffer et al., 2009).

Importantly, even though the positions of ArcZ binding vary within target mRNAs, the ArcZ sRNA utilizes a single conserved element to mediate target repression. Similar conservation of a subregion of sRNA is found in other regulators, including GcvB, OmrAB, RNAIII and CyaR (reviewed by Papenfort and Vogel, 2009). However, unlike these examples, ArcZ is processed after transcription, and relevant nucleotides for target regulation almost exclusively reside in the processed 3′ fragment. It is therefore possible that ArcZ requires processing to produce the active molecule that base-pairs with its target mRNAs. Endonucleolytic maturation of the primary sRNA to leave an active 3′ fragment has been demonstrated for MicX sRNA of Vibrio cholerae (Davis and Waldor, 2007). Consistent with this, we have recently found that ArcZ is processed by RNase E, and that the 3′ end of ArcZ is sufficient to repress the STM3216, sdaC and tpx targets (K. Papenfort, N. Said and J. Vogel, unpublished results).

ArcZ controls STM 3216, a horizontally acquired gene with a function in chemotaxis

Comparative genomic analysis showed that sdaC and tpx are present throughout the enterobacterial tree. However, the integration of STM3216 within the aer-yqjI locus is specific to the Salmonella genus (Fig. S10). In fact, STM3216 represents a horizontally acquired Salmonella‘signature’ gene that was present in all 22 analysed Salmonella species, but absent from closely related E. coli, Klebsiella and Yersinia species (Porwollik et al., 2002).

Computational predictions suggested that the STM3216 contains a chemoreceptor-associated signalling domain found in methyl-accepting chemotaxis proteins (MCP), plus several transmembrane domains and a HAMP (histidine kinase, adenylyl cyclases, methyl binding proteins and phosphatases) domain (Frye et al., 2006). STM3216 was also shown to be FhlDC-dependent (Frye et al., 2006), and its predicted role in motility was confirmed in a Salmonella mutant background which lacked the five major MCPs (tsr, tar, tap, tcp and aer); deletion of STM3216 abrogated swimming and swarming motility (Wang et al., 2006), favouring a model in which different MCPs share redundant functions, at least under standard lab conditions. While the precise biological role of STM3216 remains to be determined, the predictions might link ArcZ to chemotaxis functions.

Intriguingly, ArcZ represents the first example of a conserved Hfq-dependent sRNA that modulates the expression of a horizontally acquired gene. Reciprocally, the conserved ompD gene was previously shown to be silenced by the horizontally acquired Hfq-dependent InvR sRNA (Pfeiffer et al., 2007). We note that a high proportion of the horizontally acquired Salmonella genes are, directly or indirectly, controlled by Hfq (Figueroa-Bossi et al., 2006; Sittka et al., 2008; Ansong et al., 2009). Based on these observations, we postulated that the existing pool of Hfq-dependent sRNAs play a role in the control of newly acquired Salmonella genes at the post-transcriptional level (Sittka et al., 2008; Vogel, 2009). Such a role would be analogous to the function of H-NS protein, which silences foreign genes at the transcriptional level (Lucchini et al., 2006; Navarre et al., 2006). The ArcZ-mediated repression of STM3216 suggests a new paradigm involving the recruitment of enterobacterial core sRNAs to control the important class of horizontally acquired genes, and raises the possibility that such core sRNAs may also control Salmonella virulence factors.

Hfq and implications for target discovery and regulation

Numerous studies have described Hfq-dependent phenotypes and regulation in a wide range of bacterial species (Tsui et al., 1994; Ding et al., 2004; Sonnleitner et al., 2006; Guisbert et al., 2007). Even though post-transcriptional control was assumed to be responsible for the observed regulation, it was not clear whether particular mRNAs of interest were directly targeted by Hfq, or if an sRNA was involved. We combined Hfq coIP studies with deep sequencing analysis (Sittka et al., 2008) to assemble a catalogue of Hfq-associated mRNAs, which have the potential to be direct sRNA targets. Specifically, we identified ∼730 distinct mRNA species that were enriched by Hfq coIP (Sittka et al., 2008). Several of the most highly enriched mRNAs, e.g. ompX (∼23-fold) or ybfM (15-fold), have since been allocated with the enterobacterial ‘core’ sRNAs, CyaR or MicM respectively (Johansen et al., 2008; Papenfort et al., 2008; De Lay and Gottesman, 2009; Rasmussen et al., 2009). Similarly, the direct ArcZ targets identified here (the sdaC, tpx and STM3216 mRNAs) were enriched between fourfold and 16-fold by Hfq coIP (Sittka et al., 2008). Given that none of these mRNAs were predicted to be ArcZ targets by available biocomputational algorithms, our results emphasize the need to create global experimental data sets such as Hfq association of transcripts, and to integrate these into target predictions.

Pleiotropy of increased ArcZ expression

Increasing the level of expression of sRNAs has been a valuable tool for the study of potential functions of post-transcriptional regulators (Vogel and Wagner, 2007). Specifically, in the case of the peroxide-inducible OxyS sRNA, this approach identified 40 differentially regulated genes, making OxyS a major player in bacterial oxidative stress response (Altuvia et al., 1997).

Similarly, increased expression of ArcZ suggested a pleiotropic effect on the cellular proteome of Salmonella, and resulted in lack of motility (Fig. 5A and B). In particular, the depletion of the FliC protein, the core component of the bacterial flagellar apparatus and depletion of several abundant outer membrane proteins shows a global regulatory function of ArcZ. The molecular basis responsible for these observations remains to be determined: ArcZ could act globally via several layers of regulation, including transcription, mRNA stability, protein stability or translation. Analysis of the Salmonella transcriptome revealed differential expression of > 750 genes upon ArcZ overexpression (Table S2), representing > 16% of the genome. Intriguingly, ∼50% of these genes are also Hfq-dependent and share similar functional categories (Fig. S7, Table S3; Sittka et al., 2008).

The observed pleiotropic deregulation by ArcZ may reflect direct and indirect effects. First, indirect effects may result from the direct deregulation of the mRNA of a transcription factor; we note that of the > 750 deregulated genes, rpoE and rpoS encode alternative sigma factors that themselves govern the transcription of several hundred Salmonella genes (Bang et al., 2005; Skovierova et al., 2006). Work from the Gottesman group showed that multicopy expression of ArcZ (RyhA) significantly increased expression of an rpoS::lacZ fusion construct in E. coli (Wassarman et al., 2001), and that ArcZ directly stimulates rpoS mRNA translation (P. Mandin and S. Gottesman, pers. comm.) similar to the action of the DsrA and RprA sRNAs (Repoila et al., 2003). We have confirmed that ArcZ increases σS levels in rapidly growing cells (Fig. S11); however, the effect is mild in late stationary phase, i.e. at the growth phase in which the massive gene regulation by ArcZ was observed on microarrays. In addition, comparison of the genes affected by ArcZ or σS shows that only 25% of the genes that are globally regulated by ArcZ are also known to be σS-regulated (Fig. S12), thus an increase in σS activity might not fully explain the pleiotropy of ArcZ action.

Second, the pleiotropic effects might originate from direct regulation of non-cognate mRNAs by ArcZ. Given that ArcZ can target one of the most conserved elements of bacterial messengers (the SD), its increased intracellular concentration upon overexpression may force certain mRNAs into complexes with ArcZ that would not form in wild-type cells. This model of ‘off-targeting’ builds on insights from the mammalian post-transcriptional machinery that describe the ability of small RNAs to silence non-target genes by relaxed binding of mRNAs (Jackson and Linsley, 2004). In mammals, this phenomenon can cause the development of severe disease (Ma et al., 2007), supporting the idea that tight control of the expression of small RNA is essential to balance cellular systems.

Third, our deep sequencing-based profiling showed a significant change in the pattern of Hfq-bound RNAs (Fig. 6B). Intriguingly, even though the expression of Hfq remained constant (GEO accession no. GSE17771), increased expression of ArcZ led to the recovery of more sRNAs (Fig. S8), suggesting that higher levels of ArcZ could displace other RNA molecules from Hfq, and deregulate their function. Indeed, the total number of Hfq-bound mRNAs was significantly reduced upon increased expression of ArcZ (Fig. S8). This suggests that another direct consequence of increased ArcZ levels might be the titration of available Hfq. This concept was originally proposed by Storz and co-workers (Zhang et al., 1998), who suggested that the OxyS sRNA would regulate rpoS mRNA by competing with Hfq-binding and thereby decrease σS synthesis at the post-transcriptional level. OxyS accumulates to ∼4500 copies in peroxide-treated cells (Altuvia et al., 1997), yielding approximately one OxyS molecule per two Hfq hexamers; the total number of Hfq hexamers was estimated to be ∼10 000 hexamers/E. coli cell (Kajitani et al., 1994; Ali Azam et al., 1999). On the face of it, the relevant ∼260 ArcZ molecules/Salmonella cell determined here (Table S1) upon sRNA overexpression do not seem to support a titration of the abundant Hfq protein. However, the cellular copy number of Hfq is a matter of controversy, because an earlier study set it at ∼400 Hfq hexamers/E. coli cell (Carmichael et al., 1975). Our unpublished work in Salmonella also indicates a lower copy number of Hfq hexamers, indicating that the ArcZ may indeed be able to usurp Hfq at the expense of other sRNAs or mRNAs.

Interestingly, such Hfq titration could account for the motility defect observed for increased ArcZ expression. Global determination of Hfq binding partners identified the fliC mRNA as a potent Hfq binder (Sittka et al., 2008); we found that overexpression of ArcZ decreased the level of Hfq-bound fliC mRNAs by approximately fivefold (data not shown). Taken together, our data reinforce the previous postulate (Zhang et al., 1998) that sRNAs can be potent global regulators that act through Hfq titration. The pleiotropic impact of ArcZ action is likely to be a combination of the factors discussed above, with similarities to the CsrB-like RNAs that titrate CsrA-like proteins (Babitzke and Romeo, 2007; Lapouge et al., 2008), and to 6S RNA that sequesters RNA polymerase (Wassarman, 2007).

Perspective

We have discovered that the core ArcZ sRNA directly controls three mRNAs, one of which (STM3216) is encoded by a horizontally acquired gene that is specific to Salmonella species. Unlike most other bacterial sRNAs, ArcZ is expressed at all stages of growth. The finding that ArcZ-mediated repression of SdaC and STM3216 synthesis is most pronounced in rapidly growing cells, and that Tpx is most dramatically regulated at stationary phase (Fig. 3) indicates that the target profile of ArcZ changes according to the intracellular availability of mRNAs. Our findings have important implications for target identification. For example, repression of the STM3216 was identified by pulse expression of ArcZ at early stationary phase (Table 2); however, STM3216 mRNA levels were not elevated in stationary phase ΔarcZ cells (Table 1), suggesting that successful target identification by transcriptomic approaches is strongly dependent upon the repertoire of the available messengers at a given time point.

Our observation that increased expression of ArcZ causes such a fundamental change in gene expression patterns and the repertoire of Hfq-bound sRNAs suggests that the level of ArcZ never rises substantially within the bacterial cell, as this would have major consequences. It was recently shown that Vibrio cholarae uses a sophisticated system involving two feedback loops to orchestrate the proper expression of Qrr1–4 sRNA during quorum sensing (Svenningsen et al., 2009), highlighting the need to quantitatively and temporally control sRNA expression to safeguard cellular homeostasis.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Note added in proof
  9. References
  10. Supporting Information

Oligodeoxynucleotides

Table S4 lists all oligodeoxynucleotides used in this study.

Strains

The mouse-virulent strain SL1344 of Salmonella enterica serovar Typhimurium is referred to as the wild-type strain throughout this study. Its derivates were constructed using the lambda red recombinase method (Datsenko and Wanner, 2000). The arcZ mutant strain was previously published (Papenfort et al., 2008), as well as strains JVS-0255 (Δhfq; Pfeiffer et al., 2007), JVS-00748 (ΔrpoS; Kowarz et al., 1994) and JVS-1338 (hfq::3xFLAG; Pfeiffer et al., 2007). Strain JVS-3791 (ΔarcB::kan) was constructed using primer pair JVO-3485/-3486, replacing nucleotides +1 to 2276 (corresponding to the translational start site of arcB). Strain JVS-4826 (ΔyhbL::kan) was constructed using primer pair JVO-4269/-4270, replacing nucleotides −10 to 653 (corresponding to the translational start site of yhbL). Obtained mutants were verified by PCR using primers JVO-0179/3487 and JVO-0177/4271 respectively. C-terminal flagged tagged versions of target genes, i.e. sdaC::FLAG (JVS-1519), STM3216::FLAG (JVS-1500) and tpx::FLAG (JVS-4060), were constructed by a modified lambda red approach based on (Uzzau et al., 2001) using primers JVO-1792/1793, JVO-1088/1089 and JVO-0177/4271 respectively. Strains with chromosomally integrated Ptet::gfp (JVS-3858) or ParcZ::gfp (JVS-4027) constructs were established as previously described (Hautefort et al., 2003). Amplification of promoter fragments was achieved by primer pairs JVO-3404/3409, 1941/2151 respectively.

Plasmids

A complete list of all plasmids used in this study can be found in Table 3. To express a plasmid-borne arcZ gene from the tightly controlled, l-arabinose-inducible PBAD promoter (Guzman et al., 1995), the following strategy was used to clone the arcZ gene under PBAD control such that transcription would start precisely at their native +1 site. Plasmid pBAD-His-myc was PCR amplified [cycling parameters: 95°C/30′′, 25 × (95°C/10′′, 59°C/30′′, 72°C/2′), 72°/10′] with primers JVO-0900/-0901 (JVO-0901 introduces an XbaI restriction site upstream of the rrnB terminator sequence) and Phusion DNA polymerase (Finnzymes, Finland), the PCR product digested with XbaI and DpnI, and purified. To PCR amplify the arcZ insert, the sense primer (JVO-0898) was designed such that it starts with the sRNA +1 site and that it carries a 5′ phosphate modification. The antisense primer (JVO-0177) binds downstream of the arcZ terminator, and adds an XbaI site to its sequence. Following amplification with Phusion DNA polymerase, the PCR product was digested with XbaI and gel-purified. Vector- and sRNA-derived PCR products were ligated with T4 DNA ligase (5′ blunt end/3′ XbaI site) and transformed, yielding plasmids pBAD-ArcZ (pKP-4-13). Correct inserts were confirmed by sequencing of the plasmids with vector primers, pBad-FW and pBad-REV. The very same insert was used for construction of plasmid pPL-ArcZ (pJU-19); here, an XbaI-digested PCR product obtained with primers PLlacoB and PLlacoD on plasmid pZE-12-luc served as vector backbone. Plasmid pKP-4-13 served as template for establishment of pBAD-ArcZ* (pKP-181-3) harbouring a single nucleotide exchange which was introduced by primers JVO-4273/4274. Details of the cloning procedure can be found elsewhere (Urban and Vogel, 2007). To construct the arcZ complementation plasmid, parcZ (pKP-48-1), the sRNA gene was amplified from genomic DNA (primers JV0-0177/-0179), the PCR product digested with XhoI and XbaI, and inserted into vector pZE-12-luc treated with the same enzymes. Construction of psdaC::gfp fusion plasmid (pKP-102-1) was achieved by amplification of a DNA fragment spanning from −123 (corresponding to the sdaC translational start site) to 30 bp of sdaC coding region using primers JVO-3367 and JVO-3368. Inserts for plasmids ptpx::gfp (pKP-125-1) and pSTM3216::gfp (pKP-127-1) were amplified by amplification with primer pairs JVO-3688/3689 and JVO-3737/1863 respectively. The PCR products were digested with BrfBI and NheI, gel-purified and ligated into pXG-10 (Urban and Vogel, 2007) backbone digested with the same enzymes. Plasmids pKP102-1, pKP125-1 and pKP-127-1 were used as templates for establishment of psdaC*::gfp (pKP187-2), ptpx*::gfp (pKP188-3) and pSTM3216*::gfp (pKP-183-1) harbouring a single nucleotide exchange, which was introduced by primers JVO-4276/4277, JVO-4552/4553 or JVO-4554/4555 respectively. Competent E. coli TOP10 or TOP10 F′ cells (Invitrogen) were used for all cloning procedures.

Table 3.  Plasmids used in this study.
Plasmid trivial namePlasmid stock nameRelevant fragmentCommentOrigin, markerReference
pJV300  Control plasmid, expresses a ∼50 nt nonsense transcript derived from rrnB terminatorColE1, AmpRSittka et al. (2007)
pBADpKP-8-35 pBAD control plasmid, expresses the same ∼50 nt nonsense RNA as pJV300pBR322, AmpRPapenfort et al. (2006)
pBAD-ArcZpKP-4-13ArcZArcZ under the control of the inducible pBAD promoterpBR322, AmpRThis study
pPL-ArcZpJU-19ArcZColE1 plasmid, based on pZE12-luc, expresses ArcZ from a PLlacO promoterColE1, AmpRThis study
parcZpKP-48-1ArcZColE1 plasmid, based on pZE12-luc, expresses ArcZ from its endogenous promoterColE1, AmpRThis study
pBAD-ArcZ*pKP-181-3ArcZ*Same as pKP-4-13 but carries a single nt exchange in its conserved 3′ endpBR322, AmpRThis study
psdaC::gfppKP-102-1sdaC 5′UTRGFP reporter plasmid. Carries the Salmonella sdaC 5′UTR and 30 bp of coding sequencepSC101*, CmRThis study
ptpx::gfppKP-125-1tpx 5′UTRGFP reporter plasmid. Carries the Salmonella tpx 5′UTR and 210 bp of coding sequencepSC101*, CmRThis study
psdaC::gfp*pKP-187-2sdaC* 5′UTRSame as pKP-102-1 but carries a single nt exchange in putative ArcZ interaction sequencepSC101*, CmRThis study
ptpx::gfp*pKP-188-3tpx* 5′UTRSame as pKP-125-1 but carries a single nt exchange in putative ArcZ interaction sequencepSC101*, CmRThis study
pSTM3216::gfppKP-127-1STM3216 5′UTRGFP reporter plasmid. Carries the Salmonella STM3216 5′UTR and 60 bp of coding sequencepSC101*, CmRThis study
pSTM3216::gfp*pKP-183-1STM3216* 5′UTRSame as pKP-127-1 but carries a single nt exchange in putative ArcZ interaction sequencepSC101*, CmRThis study
pKD-4 Template for KmR mutant constructionoriRγ, AmpRDatsenko and Wanner (2000)
pKD-46ParaB-γ-β-exoTemperature-sensitive lambda red recombinase expression plasmidoriR101, AmpRDatsenko and Wanner (2000)

Media and bacterial growth

Growth in LB broth (220 r.p.m., 37°C) or on LB plates at 37°C was used throughout this study. SOC medium was used to recover transformants after heat-shock or electroporation and prior to plating. Antibiotics (where appropiate) were applied at the following concentrations: 100 μg ml−1 ampicillin, 50 μg ml−1 kanamycin, 20 μg ml−1 chloramphenicol. For induction from pBAD-derived plasmids, cultures were treated with a final concentration of 0.2% l-arabinose. Bacterial motility was assayed on LB plates containing 0.3% agar (final concentration) at 37°C.

RNA isolation and Northern detection

Overnight cultures were diluted 1/100 in fresh medium and grown to the indicated cell densities (OD600). Culture aliquots were removed and mixed with 0.2 vol. of stop solution (5% water-saturated phenol, 95% ethanol), and snap-frozen in liquid nitrogen. After thawing on ice, bacteria were pelleted by centrifugation (2 min, 16 000 rcf, 4°C), and RNA was isolated using the Promega SV total RNA purification kit as described at http://www.ifr.ac.uk/safety/microarrays/protocols.html (Kelly et al., 2004) or using the Trizol Method. The purified RNA was quantified on a Nanodrop machine (NanoDropTechnologies). RNA samples (∼5 μg) were denatured for 5 min at 95°C in RNA loading buffer [95% (v/v) formamide, 0.1% (w/v) xylene cyanole, 0.1% (w/v) bromphenol blue, 10 mM EDTA], separated on 8.3 M urea/5% polyacrylamide gels, and transferred to Hybond-XL membranes (GE Healthcare) by electroblotting (1 h, 50 V, 4°C) in a tank blotter (Peqlab, Germany). Following prehybridization of the membranes in Rapid-hyb Buffer (GE Healthcare), 32P-labelled gene-specific probes were added. Radioactively labelled oligodeoxynucleotide JVO-0157 (complementary to nucleotides 80–95 of ArcZ RNA) was used for detection of arcZ transcripts. Detection of RybB (radio-labelled oligonucleotide JVO-1205) and InvR (radio-labelled oligonucleotide JVO-222) sRNAs was previously described (Papenfort et al., 2006; Pfeiffer et al., 2007). Hybridization was carried out at 42°C. Equal RNA loading was verified by hybridization with radioactively labelled oligodeoxynucleotide JVO-0322, which is complementary to Salmonella 5S rRNA.

Microarray experiments

ArcZ overexpression.  Strain JVS-0082 (ΔarcZ) was transformed with plasmids pJV300 (control) and pKP48-1 (parcZ), and grown in liquid culture inoculated 1:100 from an overnight culture to 6 h after cells had reached an OD600 of 2.0. For sample preparation, aliquots equivalent to 4 OD were removed and treated with 0.2 vol. of stop solution (5% water-saturated phenol, 95% ethanol), and snap-frozen in liquid nitrogen. RNA was isolated as described before (Promega SV total RNA purification kit). We considered genes to be differentially expressed if they displayed ≥ twofold changes in all replicates and were statistically significantly different (Student's t-test; P-value ≤ 0.05). A list of raw values used for this analysis can be found under accession number GSE17771. Statistical analysis, data visualization and data mining were performed using GeneSpring 7.3 (Agilent).

arcZ mutant strain:  Strains JVS-007 (WT) and JVS-0082 (ΔarcZ) were transformed with plasmid pJV300 (control) and grown in liquid culture inoculated 1:100 from an overnight culture to 6 h after cells had reached an OD600 of 2.0, and analysed as above (raw values listed under accession no. GSE17771). We considered genes to be differentially expressed if they displayed ≥ twofold changes in all replicates and were statistically significantly different (Student's t-test; P-value ≤ 0.05).

ArcZ pulse expression.  Strain SL1344 was transformed with plasmids pKP8-35 (control) and pKP4-13 (pBAD-ArcZ), and grown in liquid culture inoculated 1:100 from an overnight culture to an OD600 of 1.5. Expression of the insert was induced with l-arabinose (0.2% final concentrations) for 10 min. For sample preparation, aliquots equivalent to 3 OD were removed, and RNA was analysed as above (microarray raw values listed under accession no. GSE17771). We considered genes to be differentially expressed if they displayed ≥ threefold changes in all replicates and were statistically significantly different (Student's t-test; P-value ≤ 0.05).

The microarrays used in this study include PCR products of all the genes present in the sequenced S. typhimurium strain LT2. In addition, we added 229 genes specific to S. typhimurium strain SL1344. Details of all the amplicons can be found at http://www.ifr.ac.uk/Safety/MolMicro/pubs.html. Our experimental design involves the use of Salmonella enterica serovar Typhimurium genomic DNA as the cohybridized control for one channel on all microarrays. This method has the advantage of allowing the direct comparison of multiple samples. Total RNA and chromosomal DNA were labelled by random priming according to the protocols described at http://www.ifr.bbsrc.ac.uk/safety/microarrays/protocols.html. Briefly, 16 μg RNA were reverse transcribed and labelled with Cy3-conjugated dCTP (Pharmacia) using 200 U of Stratascript (Stratagene) and random octamers (Invitrogen). Chromosomal DNA (400 ng) was labelled with Cy5-dCTP using the Klenow fragment. After labelling, each Cy3-labelled cDNA sample was combined with Cy5-labelled chromosomal DNA and hybridized to a microarray overnight at 65°C. After hybridization, slides were washed and scanned using a GenePix 4000 A scanner (Axon Instruments). Fluorescent spots and the local background intensities were identified and quantified using Bluefuse software (BlueGnome, Oxford). To compensate for unequal dye incorporation, data centring to zero was performed for each block (one block being defined as the group of spots printed by the same pin).

Whole-cell protein fractions and Western blot

Culture samples were taken according to 1 OD600, and centrifuged 2 min at 16 100 g at 4°C. The cell pellet was resuspended in 1× sample loading buffer (1× SLB; Fermentas) to a final concentration of 0.01 OD ml−1. For coomassie staining a total of 0.1 OD were loaded on a 12% SDS gel and stained over night. Western blotting and detection of GFP fusion proteins and GroEL was performed as described (Urban and Vogel, 2007). FLAG-tagged proteins were detected using a monoclonal anti-FLAG antibody (Sigma; #F1804) and α-rabbit-horseradish peroxidase (HRP; 1:5000 in 3% BSA, TBST20) following visualization using Western Lightning reagent (PerkinElmer), and signals detected with a Fuji LAS-3000 CCD camera.

Hfq co-immunprecipitation

Strain JVS-007 (WT) was transformed with plasmids pJV300 (control) or pJU-19 (pPL-ArcZ), and grown in liquid culture inoculated 1:100 from an overnight culture to 6 h after cells had reached an OD600 = 2.0. A protocol for recovery of Hfq-bound sRNA was published previously (Pfeiffer et al., 2007). Note that different from the original protocol cells were lysed by passing them 3 times in a ‘French Press’. A detailed protocol for data cDNA synthesis, HTPS and data processing was published elsewhere (Sittka et al., 2008).

Determination of in vivo abundance of ArcZ

This method was previously described by Pfeiffer et al. (2007). The ArcZ full-size transcript was established by amplification of the Salmonella ArcZ sequence (stretching from the transcriptional start site to the 3rd uridine residue of the terminator sequence) using primer pair 4045/4046 and cloning of this sequence into plasmid pHDV-for via EcoRI and NcoI (Walker et al., 2003), carrying a 5′ located T7 promoter sequence and a 3′ located HDV ribozyme. Following digestion with HindIII, 200 ng of this plasmid served as template for in vitro transcription of ArcZ full-lenght transcript using the Megascript kit (Ambion). The processed version of ArcZ was chemically synthesized (Ribotask; Table S5). Total RNA (corresponding to ∼2.2 × 109 cells) of wild-type Salmonella or cells carrying either parcZ or pPL-ArcZ were compared with serial dilution of both in vitro transcripts (20, 10, 2, 0.4 ng) by Northern blotting and subsequent hybridization with radioactively labelled oligonucleotide JVO-0157. A detailed description of in vivo copy determination can be found in Supporting information.

Sequence retrieval and alignments

Information for sequence alignments was collected using BlastN searches (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi) of the following genome sequences (accession numbers are given in parentheses): Salmonella typhimuriumLT2 (NC_003197), E. coliK12 (NC_000913), Citrobacter koseri ATCC BAA-895 (NC_009792), Enterobacter sp. 638 (NC_009436), Yersinia pestisCO92 (NC_003143), Photorhabdus luminescens ssp. laumondii (NC_005126), Serratia proteamaculans 568 (NC_009832), Shigella flexneri 2a str 301 (NC_004337), Salmonella bongori 12149 and Klebsiella pneumoniae MGH 78578 (NC_009648).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Note added in proof
  9. References
  10. Supporting Information

We thank Cynthia. M. Sharma for computational analysis, and Johannes Urban for plasmid pJU-19. K.P. was supported by a stipend from the Boehringer Ingelheim Fonds, Germany. This work was supported by the Core Strategic Grant from the BBSRC to the Hinton lab, and by funds from the DFG Priority Program SPP1258 Sensory and Regulatory RNAs in Prokaryotes to J. Vogel.

Note added in proof

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Note added in proof
  9. References
  10. Supporting Information

The microarray data discussed in this study have been deposited in the NCBI Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo) under accession number GSE17771.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Note added in proof
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Note added in proof
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
MMI_6857_sm_Figures_and_Tables.pdf867KSupporting info item
MMI_6857_sm_Table_S2.pdf76KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.