CsrA Represses Translation of sdiA, Which Encodes the N-Acylhomoserine-l-Lactone Receptor of Escherichia coli, by Binding Exclusively within the Coding Region of sdiA mRNA

Abstract

The RNA binding protein CsrA is the central component of a conserved global regulatory system that activates or represses gene expression posttranscriptionally. In every known example of CsrA-mediated translational control, CsrA binds to the 5′ untranslated region of target transcripts, thereby repressing translation initiation and/or altering the stability of the RNA. Furthermore, with few exceptions, repression by CsrA involves binding directly to the Shine-Dalgarno sequence and blocking ribosome binding. sdiA encodes the quorum-sensing receptor for N-acyl-l-homoserine lactone in Escherichia coli. Because sdiA indirectly stimulates transcription of csrB, which encodes a small RNA (sRNA) antagonist of CsrA, we further explored the relationship between sdiA and the Csr system. Primer extension analysis revealed four putative transcription start sites within 85 nucleotides of the sdiA initiation codon. Potential σ⁷⁰-dependent promoters were identified for each of these primer extension products. In addition, two CsrA binding sites were predicted in the initially translated region of sdiA. Expression of chromosomally integrated sdiA′-′lacZ translational fusions containing the entire promoter and CsrA binding site regions indicates that CsrA represses sdiA expression. The results from gel shift and footprint studies demonstrate that tight binding of CsrA requires both of these sites. Furthermore, the results from toeprint and in vitro translation experiments indicate that CsrA represses translation of sdiA by directly competing with 30S ribosomal subunit binding. Thus, this represents the first example of CsrA preventing translation by interacting solely within the coding region of an mRNA target.

INTRODUCTION

Bacteria utilize a variety of regulatory networks that sense and respond to changing environmental conditions, resulting in global changes in gene expression. Disruption of these global regulatory networks can have severe consequences for bacterial growth and survival (13, 21, 28). The conserved carbon storage regulation system (Csr), also referred to as Rsm (repressor of secondary metabolites) in some bacterial species, globally controls gene expression at the posttranscriptional level. Depending on the particular bacterial species, Csr regulates virulence, quorum sensing, motility, biofilm development, peptide uptake, and carbon metabolism (reviewed in references 1, 29, and 39).

The Csr system of Escherichia coli consists of several known components. (i) CsrA is an RNA binding protein that represses or activates gene expression by repressing translation initiation and/or by destabilizing or stabilizing target transcripts (2, 19, 43). CsrA directly represses its own translation; however, CsrA also indirectly activates its transcription by affecting the transcription of a σ^S-dependent promoter, one of five promoters that drive csrA transcription (46). Because CsrA is a homodimeric protein with two identical RNA binding surfaces, it is capable of simultaneously binding two sites within a target transcript (8, 23, 24, 32). (ii) CsrB and CsrC are two small RNA (sRNA) antagonists of CsrA. Both sRNAs contain several CsrA binding sites and function by sequestering multiple CsrA dimers. CsrB and CsrC have short half-lives, which permits rapid adjustments to their intracellular concentrations (20, 44). (iii) CsrD is a protein that specifically targets CsrB and CsrC for degradation by the endonuclease RNase E and polynucleotide phosphorylase (PNPase), a 3′-to-5′ exonuclease. Unlike other GGDEF-EAL domain proteins, CsrD does not metabolize the secondary messenger cyclic di-GMP (c-di-GMP) (37). (iv) The BarA-UvrY two-component signal transduction system (TCS) activates transcription of csrB and csrC in response to end products of carbon metabolism (6, 38). As CsrA indirectly activates transcription of csrB and csrC by an unknown mechanism, the cellular level of CsrA is tightly controlled. A recent study using RNA-seq identified 721 transcripts that bind to CsrA, suggesting that CsrA directly affects the expression of >15% of the E. coli genome (11). Since >40 of the transcripts identified by RNA-seq encode regulatory proteins, several of which are global regulators themselves, it is apparent that Csr directly or indirectly affects the expression of a substantial fraction of the E. coli genome.

Although considerable sequence variation exists among the known E. coli CsrA binding sites, GGA is a highly conserved motif that is often present in the loop of short RNA hairpins (9, 20, 44). CsrA has been shown to repress translation initiation of several genes in various organisms by binding 1 to 6 sites, one of which overlaps the cognate Shine-Dalgarno (SD) sequence, so that bound CsrA competes with 30S ribosomal subunit binding (2, 3, 8, 11, 18, 22, 46, 48). There is one example in which CsrA represses translation by binding to two sites that straddle the SD sequence. In this instance, the downstream binding site overlaps the translation initiation codon (17). Another variation of translational repression was identified in which RsmA of Pseudomonas aeruginosa binds to and stabilizes an RNA hairpin that sequesters the SD sequence (15). In the last two examples, it does not appear that CsrA directly contacts the SD sequence.

A common quorum-sensing system in Gram-negative bacterial species involves the synthesis and detection of N-acyl-l-homoserine lactones (AHLs). Although E. coli does not synthesize AHLs, it contains SdiA, which senses and responds to AHLs produced by other bacterial species (reviewed in references 25, 34, and 35). Since we previously showed that sdiA indirectly stimulates transcription of csrB via the BarA-UvrY TCS (38), we were interested in further exploring the relationship between sdiA and the Csr system. A position weight matrix (pwm) search tool was previously used to identify hfq and csrA as CsrA-regulated genes (3, 46). Using this bioinformatics approach, we identified two putative CsrA binding sites in the initially translated region of sdiA. In contrast to all other known CsrA-regulated genes, a potential CsrA binding site was not identified in the untranslated region of sdiA. As CsrA has never been shown to control gene expression by binding exclusively to the coding sequence of a target transcript, we carried out experiments to determine whether CsrA regulates sdiA expression.

MATERIALS AND METHODS

Bacterial strains and plasmids.

All E. coli strains used in this study are listed in Table 1. The DNA oligonucleotide primers used in this study are listed in Table S1 in the supplemental material. The plasmid cloning vector pMLB1034 was published previously (33). Plasmid pME3 was constructed by cloning a PCR fragment containing the sdiA promoter, leader, and translation initiation region extending from −121 to +100 relative to the start of translation into the EcoRI and BamHI sites of the pMLB1034 polylinker. The primer pair sdiA-F1 and sdiA-R1 and chromosomal DNA were used in the PCR.

Table 1.

E. coli strains used in this study

Strain	Descriptiona	Source
CF7789	F− λ− ΔlacI-lacZ (MluI)	M. Cashel
PLB1732	CF7789/sdiA′-′lacZ	This study
PLB1733	CF7789/sdiA′-′lacZ BS2 mutationsb	This study
PLB1734	CF7789/sdiA′-′lacZ BS1 mutations	This study
PLB1735	CF7789/sdiA′-′lacZ BS1-BS2 mutations	This study
PLB1736	CF7789/sdiA′-′lacZ csrA::kan	This study
PLB1737	CF7789/sdiA′-′lacZ BS2 mutations csrA::kan	This study
PLB1738	CF7789/sdiA′-′lacZ BS1 mutations csrA::kan	This study
PLB1739	CF7789/sdiA′-′lacZ BS1-BS2 mutations csrA::kan	This study
TRCF7789	CF7789/csrA::kan	30

^aAll sdiA′-′lacZ fusions containing −186 to +97 relative to the start of sdiA translation were integrated into the λ att site via the CRIM system (12).

^bCsrA binding site mutations: BS1, G6C, G7C, G12C, and G13C; BS2, G56C, G57C, G60C, and G61C. The BS1-BS2 mutant fusion contained all eight of these substitutions.

The conditional-replication, integration, and modular (CRIM) system, which allows integration of lacZ fusions into the λ att site, has been described previously (12). Plasmid pLFT is a CRIM-based translational fusion vector that confers ampicillin resistance on integrated fusions (11). Plasmid pAG1 was constructed by cloning a PCR fragment containing the sdiA promoter, the leader region, and the initially translated region (−186 to +97 relative to the start of sdiA translation) into the PstI and EcoRI sites of the pLFT polylinker, resulting in an sdiA′-′lacZ translational fusion with the 33rd codon of sdiA fused to the 10th codon of lacZ. The primer pair sdiA-F2 and sdiA-R2 and chromosomal DNA were used in the PCR. Plasmids pYH190, pYH189, and pYH191 are derivatives of pAG1 containing mutations in the CsrA binding sites BS1 (G6C, G7C, G12C, and G13C), BS2 (G56C, G57C, G60C, and G61C), and both (G6C, G7C, G12C, G13C, G56C, G57C, G60C, and G61C), respectively. These mutations were introduced by PCR using the primer pairs BS1-1 and BS1-2 and/or BS2-1 and BS2-2. CF7789 cells were transformed with the CRIM helper plasmid pINT (12). The resulting strain was used to integrate the wild-type (WT) and mutant sdiA′-′lacZ fusions into the λ att site. Verification of single-copy integration by PCR followed a published procedure (12). The proper DNA sequence was verified by sequencing PCR-generated DNA fragments. A csrA::kan allele (30) was subsequently introduced into each of these strains by P1vir transduction.

Primer extension (PE) assay.

E. coli strain CF7789 was grown in LB at 30°C to late exponential phase. Cells were harvested by adding 10 ml of culture to an equal volume of frozen buffer (8.5 mM Tris-HCl, pH 7.2, 5 mM MgCl₂, 25 mM sodium azide, and 500 μg/ml choramphenicol), followed by centrifugation. The cell pellets were suspended in 500 μl of a 2:1 mixture of RNA Protect bacterial reagent (Qiagen)-Tris-EDTA (TE) buffer and then placed on ice for 10 min. Cells were collected by centrifugation, and the cell pellets were frozen at −80°C. RNA was isolated using the RNeasy bacterial protocol (Qiagen), and DNA was removed using 2 U of Turbo DNase (Ambion). The RNA was extracted with phenol-chloroform and precipitated. Four micrograms of RNA was annealed to 0.9 pmol of a 5′-end-labeled oligonucleotide complementary to either −14 to +15 (primer sdiA-PE1) or +66 to +100 (primer sdiA-PE2) of sdiA in TE buffer by heating to 85°C and cooling to room temperature. Reverse transcription reaction mixtures (10 μl) contained the hybridization mixture and 1× avian myeloblastosis virus (AMV) reverse transcriptase buffer, 3.35 mM deoxynucleoside triphosphate (dNTP), 1.6 U RNasin, 20 μg/ml acetylated bovine serum albumin (BSA), 1 mM dithiothreitol (DTT), and 25 U of AMV reverse transcriptase (New England BioLabs). The reaction mixtures were incubated for 1 h at 42°C, stopped by adding 6 μl of gel loading buffer (70 mM EDTA, 85% formamide, 0.1× Tris-borate-EDTA [TBE], 0.025% xylene cyanol, and 0.025% bromphenol blue), and placed on ice. Samples were fractionated through a 6% sequencing gel. Sequencing reactions were performed using the same end-labeled primers and pME3 as a template. Radioactive bands were visualized using a phosphorimager.

β-Galactosidase assay.

Bacterial cultures were grown in LB medium at 37°C. Cells were harvested at various times during growth, washed with 10 mM Tris-HCl (pH 7.5), and frozen as cell pellets at −20°C. Cell extracts were prepared by suspending frozen cell pellets in 0.5 ml of BugBuster (Novagen). After 30 min of incubation at 37°C, 0.3 ml of Z buffer (27) containing 0.2 mg/ml lysozyme was added, and incubation was then continued for 30 min at 37°C. Following removal of cell debris, protein concentrations were determined by the Bio-Rad protein assay. β-Galactosidase assays were performed using the cell extracts as described previously (2).

Gel shift assay.

His-tagged CsrA (CsrA-H6) was purified as described previously (9). DNA templates for in vitro transcription were synthesized by PCR using various plasmids as DNA templates and the primer pair sdiA-F-T7 and sdiA-R3. RNA extending from −26 to +97 relative to the start of sdiA translation was synthesized in vitro using the RNAMaxx kit (Stratagene) and PCR fragments as DNA templates. Gel-purified RNA was treated with calf intestinal alkaline phosphatase and subsequently labeled at the 5′ end with [γ-³²P]ATP and T4 polynucleotide kinase. RNA suspended in TE buffer was heated to 85°C for 3 min, followed by slow cooling to room temperature. Binding reactions (10 μl) contained 10 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 100 mM KCl, 200 ng/μl yeast RNA, 0.2 mg/ml BSA, 7.5% glycerol, 20 mM DTT, 10 or 100 pM sdiA RNA, CsrA-H6 (various concentrations), and 0.1 mg/ml xylene cyanol. Competition assay mixtures also contained unlabeled competitor RNA. The reaction mixtures were incubated for 30 min at 37°C to allow CsrA-RNA complex formation. Samples were fractionated through native 15% polyacrylamide gels using 0.5× TBE as the gel running buffer. Radioactive bands were visualized with a phosphorimager and quantified using ImageQuant 5.2 software (Molecular Dynamics). The apparent equilibrium binding constants (K_d) of CsrA-sdiA RNA interaction were calculated as described previously (45).

Footprint assay.

The RNA footprint assay followed a published procedure (46). The DNA template for in vitro transcription was synthesized by PCR using plasmid pME3 as the template and the primer pair sdiA-F-T7 and sdiA-R1. Labeled RNA extending from −26 to +100 relative to the start of sdiA translation was generated as described for the gel shift assay. RNA suspended in TE buffer was heated to 85°C for 3 min, followed by slow cooling to room temperature. Binding reaction mixtures (10 μl) contained 10 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 100 mM KCl, 27.5 ng/μl yeast RNA, 0.2 mg/ml acetylated BSA, 7.5% glycerol, 20 mM DTT, 4 nM sdiA RNA, CsrA-H6 (various concentrations), and 0.1 mg/ml xylene cyanol. The reaction mixtures were incubated for 30 min at 37°C to allow CsrA-RNA complex formation. After the initial binding reaction, 0.075 U RNase T1 (Roche) was added, and incubation was continued for 15 min at 37°C. The reactions were terminated by the addition of 10 μl of gel loading buffer and placed on ice. Partial alkaline hydrolysis and RNase T1 digestion ladders were prepared as described previously (4). Samples were fractionated through an 8% polyacrylamide sequencing gel. Radioactive bands were visualized with a phosphorimager.

Toeprint assay.

Toeprint assays were performed using a modification of a published procedure (48). E. coli 30S ribosomal subunits were purified as described previously (3). DNA templates for in vitro transcription were generated by PCR using pME3 as the template and the primer pair sdiA-F-T7 and pME3plk-R. Gel-purified sdiA RNA (−26 to +100 relative to the start of sdiA translation) in TE buffer (150 nM) was hybridized to a 5′-end-labeled DNA oligonucleotide (150 nM) complementary to the 3′ end of the transcript by heating for 3 min at 85°C followed by slow cooling to room temperature. Toeprint reaction mixtures (10 μl) contained 2 μl of the hybridization mixture (30 nM final concentration), 2 μM CsrA-H6, and/or 1 μM 30S ribosomal subunits ± 5 μM tRNA^fMet, 375 μM each dNTP, and 10 mM DTT in toeprint buffer (10 mM Tris-HCl, pH 7.4, 10 mM MgCl₂, 60 mM ammonium acetate, and 6 mM 2-mercaptoethanol). Previously frozen 30S ribosomal subunits were thawed and activated by incubation for 15 min at 37°C. Mixtures containing CsrA or 30S ribosomal subunits were incubated with RNA for 30 min at 37°C to allow complex formation. When mixtures contained both CsrA and 30S ribosomal subunits, the RNA was first incubated with CsrA for 30 min at 37°C, at which time 30S ribosomal subunits were added and incubation was continued for 30 min. The reaction mixture was incubated for 15 min at 37°C following the addition of 1 U of AMV reverse transcriptase. The reactions were terminated by the addition of 6 μl of gel loading buffer. Samples were heated for 2 min at 90°C prior to fractionation through 6% polyacrylamide sequencing gels. Radioactive bands were visualized with a phosphorimager.

In vitro translation assay.

csrA′-′gfp translational fusion and control bla transcripts were synthesized in vitro using the RNAMaxx kit. CsrA-deficient E. coli S-30 extract was prepared from TRCF7789 (csrA::kan). In vitro translation reactions followed previously published procedures (46). The S-30 extract was incubated with RNase-free DNase I for 15 min at 37°C. Reaction mixtures (24 μl) contained 60 mM Tris-HEPES, pH 7.5, 60 mM NH₄Cl, 10 mM MgCl₂, 50 mM KCl, 0.5 mM EGTA, 5 mM DTT, 2 mM ATP, 0.6 mM GTP, 0.08 mM calcium folinate, 4 mg/ml of aprotinin, 4 mg/ml of leupeptin, 4 mg/ml of pepstatin A, 12 μg S-30 extract, 800 U/ml of DNase I, 500 U/ml of RNasin, 10 mM phosphoenolpyruvate, 35 U of pyruvate kinase, 0.4 mg/ml of E. coli tRNA, 130 nM mRNA, 10 μCi [³⁵S]methionine, 0.5 mM each of the other amino acids, 0.8 mM spermidine, and CsrA-H6 (various concentrations). The reaction mixtures were incubated for 45 min at 37°C and terminated by adding 6 μl of stop buffer (125 mM Tris-HCl, pH 6.8, 5% SDS, 25% glycerol, 2% 2-mercaptoethanol, and 12.5 mg/ml of bromophenol blue). Samples were heated at 100°C for 2 min prior to fractionation by 14% SDS-PAGE. Radioactive bands were visualized with a phosphorimager and quantified using ImageQuant 5.2 software.

RESULTS

CsrA represses sdiA expression.

CsrA represses translation initiation of several E. coli genes by binding 1 to 6 sites in the untranslated leaders of target transcripts so that bound CsrA blocks 30S ribosomal subunit binding. In all but two cases, a binding site overlaps the cognate SD sequence. Whereas the CsrA binding sites are located exclusively in the untranslated region of most genes that are repressed by CsrA, two examples exist in which the downstream binding site overlaps the translation initiation codon, while in another example, two of the four binding sites are present within the first 11 codons of the initially translated region (2, 3, 8, 11, 17, 41, 46). GGA is the most highly conserved sequence motif within CsrA binding sites (1, 9). A pwm search tool used to predict CsrA binding sites was previously developed to search for CsrA-regulated genes. This program assigns values from 0 to 100% according to the minimum and maximum scores calculated by the pwm (3). Using this program, we predicted the presence of two CsrA binding sites (BS1 and BS2) that were centered on the 4th and 20th codons of the sdiA coding sequence. Notably, no binding site was identified in the untranslated sdiA leader (Fig. 1). While both of the putative binding sites contained two GGA motifs, the pwm predicted that the second GGA was the authentic motif in each case. As CsrA has never been shown to control gene expression by binding exclusively to the coding sequence of a target transcript and, as found previously, SdiA indirectly activates transcription of csrB, experiments were performed to determine whether CsrA regulated sdiA expression.

Fig. 1.

Sequence of the sdiA promoter and CsrA binding site regions. The transcription start sites mapped by primer extension are marked by bent arrows. Putative −35 and −10 promoter elements for each promoter are shown. P1 contains a putative extended (ext) −10 promoter element. The csrA SD sequence and the CsrA binding sites (BS1 and BS2) are in boldface. The position weight matrix score for BS1 and BS2 are shown in parentheses. The identities of four point mutations in both BS1 and BS2 are shown below the wild-type sequence. The positions of CsrA and 30S ribosome toeprints (Rib) are marked with arrows. Numbering is with respect to the start of sdiA translation.

Because the location of the sdiA promoter(s) had never been reported, we first conducted primer extension experiments to identify potential transcription start sites in vivo. Total cellular RNA was extracted in late-exponential-phase growth from strain CF7789 grown in LB. Purified RNA was subjected to primer extension analysis using two different primers that hybridized to positions −14 to +15 or +66 to +100 of the sdiA transcript. A total of four different PE products were detected (Fig. 2, PE1 to PE4). Potential σ⁷⁰-dependent promoters were identified that could give rise to each of the PE products (Fig. 1).

Fig. 2.

Primer extension analysis of the sdiA promoter region. Total RNA was isolated from late-exponential-phase cultures. RNA was hybridized to a 5′-end-labeled DNA primer complementary to either −14 to +15 (top) or +66 to +100 (bottom) of sdiA RNA and subsequently extended with reverse transcriptase. Sequencing reactions were performed using the same end-labeled primers and plasmid pME3 as a template. Reverse transcriptase products PE1 to PE4 identified four putative transcription start sites (indicated by asterisks). Potential promoter sequences (P1 to P4) corresponding to PE1 to PE4 are shown in Fig. 1. The contrast of the gel files was uniformly modified in Adobe Photoshop CS3.

To determine whether CsrA regulated sdiA expression, a chromosomally integrated sdiA′-′lacZ translational fusion was generated that contained the entire promoter and CsrA binding site region. Expression of the integrated fusion increased severalfold during exponential-phase growth and subsequently decreased as the cells entered stationary phase (Fig. 3A). Expression of the sdiA′-′lacZ fusion was also examined in a CsrA-deficient strain. While the same temporal expression pattern was observed, expression of the fusion was 2-fold higher in the csrA::kan genetic background (Fig. 3A), indicating that CsrA represses sdiA expression.

Fig. 3.

CsrA-dependent regulation of sdiA expression. Cells were grown in LB at 37°C. Representative growth curves are shown as dashed lines. The β-galactosidase values are averages of two independent experiments with standard deviations. (A) Effect of csrA on expression of a wild-type sdiA′-′lacZ translational fusion. (B) Effect of mutating BS1 on CsrA-dependent regulation of sdiA expression. (C) Effect of mutating BS2 on CsrA-dependent regulation of sdiA expression. (D) Effect of mutating BS1 and BS2 on CsrA-dependent regulation of sdiA expression. Solid circles, csrA⁺; open circles, csrA::kan.

CsrA binds specifically to the two predicted CsrA binding sites.

Quantitative gel mobility shift assays were performed to determine whether CsrA bound to an sdiA transcript containing the two predicted CsrA binding sites. A distinct band was observed between 1 and 64 nM CsrA, indicating that CsrA formed a tight complex with the transcript with an apparent K_d value of 4 nM (Fig. 4A). The specificity of CsrA-sdiA RNA interaction was investigated by performing competition experiments with specific (sdiA and csrA) and nonspecific (phoB) unlabeled RNA competitors (Fig. 4B). Unlabeled sdiA and csrA were effective competitors, whereas phoB was not. We conclude that CsrA binds to sdiA with high affinity and specificity.

Fig. 4.

Gel mobility shift analysis of CsrA-sdiA RNA interaction. (A) Determination of the apparent equilibrium binding constant (K_d) of CsrA binding to WT sdiA transcripts and to transcripts containing mutations in CsrA binding sites, BS1 and/or BS2. 5′-end-labeled sdiA RNA (10 pM WT RNA or 100 pM mutant [mut] RNA) was incubated with various concentrations of CsrA (0 to 64 nM for WT RNA or 8 to 1,000 nM for mutant RNA). The positions of bound (B) and free (F) RNA are shown. The simple binding curves for these data are shown on the right. (B) Competition assay. Labeled WT sdiA RNA (10 pM) was incubated with CsrA in the absence or presence of specific (sdiA and csrA) or nonspecific (phoB) competitor RNA. The positions of bound (B) and free (F) RNA are shown.

CsrA-sdiA RNA footprint experiments were also performed to verify that CsrA binds to the two predicted sites. CsrA prevented RNase T1-mediated cleavage of G6, G7, G12, and G13 within BS1 (Fig. 5). Similarly, CsrA prevented cleavage of G54, G56, G57, G60, and G61 of BS2, although G60 and G61 were poorly cleaved even in the absence of CsrA (Fig. 5). Taken together with the gel shift results, we conclude that BS1 and BS2 constitute authentic CsrA binding sites.

Fig. 5.

CsrA-sdiA RNA footprint analysis. 5′-end-labeled sdiA RNA was treated with RNase T₁ in the presence of the concentration of CsrA shown at the top of each lane. Partial alkaline hydrolysis (OH) and RNase T₁ digestion (T1) ladders, as well as a control lane in the absence of RNase T₁ treatment (C), are shown. The RNase T₁ ladder was generated under denaturing conditions so that every G residue in the transcript could be visualized. The positions of the CsrA binding sites (BS1 and BS2) and the sdiA translation initiation codon (Met) are shown. Numbering is with respect to the start of sdiA translation. The contrast of the gel file was uniformly modified in Adobe Photoshop CS3.

CsrA-dependent repression of sdiA expression requires CsrA binding to the sdiA coding sequence.

We next tested the effect of mutating the highly conserved GGA motifs in BS1 and/or BS2 on CsrA binding. Although the pwm predicted that the second GGA sequence in both BS1 and BS2 constituted the authentic GGA motif, all four of these G residues in BS1 and BS2 were changed to C residues to ensure that each binding site would be destroyed (Fig. 1). Gel mobility shift results indicated that the affinity of CsrA was reduced ∼10-fold when BS2 was mutated and ∼100-fold when BS1 was mutated, indicating that CsrA binds more tightly to BS1. Importantly, binding was lost when the BS1 and BS2 mutations were combined (Fig. 4A).

As mutations in BS1 and BS2 eliminated CsrA binding in vitro, the effect of mutating BS1 and/or BS2 on the expression of integrated sdiA′-′lacZ fusions was examined in WT and CsrA-deficient strains (Fig. 3A to C). Expression of the mutant fusions retained the temporal expression pattern observed for the WT fusion. However, compared to the WT fusion, the difference in expression between the WT and csrA strains was greatly reduced for the BS1 mutant fusion. While the BS2 mutation had only a small effect on CsrA-dependent regulation by itself, when combined with the BS1 mutation, CsrA-dependent regulation was essentially eliminated. These results indicate that CsrA binding to BS1 is most important for repression and are consistent with a model in which CsrA bound at BS1 competes with ribosome binding.

CsrA represses translation of sdiA by competing with ribosome binding.

As the results described above indicate that CsrA represses expression of sdiA by binding to two sites in the initially translated region of the sdiA transcript, toeprint experiments were performed to determine whether bound CsrA affected ribosome binding. A CsrA-dependent toeprint band was observed at position 16U, which is near the 3′ edge of BS1 (Fig. 1 and 6, compare first and second lanes). The lack of a toeprint corresponding to BS2 might reflect the low affinity of CsrA observed for this site in the gel shift analysis (Fig. 4A). Note that the band between the SD sequence and the initiation codon does not constitute a CsrA toeprint, as this band was observed in the absence of protein (Fig. 6, first lane). Toeprint experiments were also performed to identify the positions of bound 30S ribosomal subunits. A weak tRNA^fMet-dependent 30S ribosomal subunit toeprint was observed at position 17U, which is at the expected position 16 nucleotides (nt) downstream from the A of the initiation codon (Fig. 6, compare third and fourth lanes). The weak ribosome toeprint might reflect the unusually large distance (14 nt) between the sdiA SD sequence and initiation codon. Importantly, when CsrA was bound to sdiA mRNA prior to the addition of tRNA^fMet and 30S ribosomal subunits, the toeprint pattern was identical to that of CsrA alone (Fig. 6, compare second, fourth, and fifth lanes). We conclude that CsrA competes with ribosome binding to sdiA mRNA.

Fig. 6.

CsrA and 30S ribosomal subunit toeprint analysis of sdiA RNA. The absence (−) or presence (+) of CsrA, tRNA^fMet, and/or 30S ribosomal subunits (30S Rib) is shown at the top of each lane. CsrA was added prior to tRNA^fMet and 30S ribosomal subunits when they were present in the same reaction. The positions of CsrA and 30S ribosomal subunit (Rib) toeprint bands are marked. The positions of the sdiA SD sequence and the translation initiation codon (Met) are shown. Sequencing lanes to reveal A, C, G, or U residues are labeled. Numbering is with respect to the start of sdiA translation. The contrast of the gel file was uniformly modified in Adobe Photoshop CS3.

The in vitro gel shift, footprint, and toeprint results described above indicate that bound CsrA blocks ribosome binding. Thus, in vitro translation experiments were performed with S-30 extracts produced from a CsrA-deficient strain of E. coli to determine whether CsrA represses translation of sdiA mRNA. Our initial attempt to examine the effect of purified CsrA on SdiA synthesis was problematic, as the newly synthesized protein was rapidly degraded (not shown). Since we previously found that using gfp translational fusions minimized similar problems in cell-free translation experiments (3, 46, 47), an sdiA′-′gfp translational fusion transcript was tested. The fusion transcript produced an apparent doublet of SdiA-GFP fusion proteins, as well as a doublet of what appears to be SdiA-GFP degradation products. Addition of increasing concentrations of CsrA to the reaction led to a corresponding decrease in SdiA-GFP synthesis. Translational repression is particularly evident at CsrA concentrations of ≥500 nM. At the highest concentration of CsrA tested (2 μM), translation of the fusion protein was reduced by 65% (Fig. 7). In vitro translation experiments were also performed with the bla transcript as a negative control. In this case, slight CsrA-dependent inhibition of Bla synthesis was observed at the higher CsrA concentrations (Fig. 7). We conclude that CsrA represses translation of sdiA by blocking ribosome binding.

Fig. 7.

Effect of CsrA on in vitro translation of sdiA′-′gfp mRNA. The E. coli S-30 extract was prepared from a CsrA-deficient (csrA::kan) strain. Reactions were carried out with the concentration of CsrA indicated at the top of each lane in the absence (−) or presence (+) of sdiA′-′gfp or bla control transcripts. SdiA-GFP and Bla translation products were analyzed by SDS-PAGE. The relative level of SdiA-GFP (black bars) and Bla (gray bars) synthesis as a function of the CsrA concentration is shown at the bottom. The level of polypeptide synthesis in the absence of CsrA was set to 1.0 for each transcript. The contrast of the gel files was uniformly modified in Adobe Photoshop CS3.

DISCUSSION

Bacteria have the ability to regulate gene expression in response to population density by a process known as quorum sensing. A common quorum-sensing system in Gram-negative bacterial species involves the synthesis and detection of AHLs by LuxI and LuxR homologs, respectively. Since AHLs can freely diffuse across cell membranes as the bacterial population increases, the concentration of AHLs in the cellular environment increases. Once a critical level of AHL is reached, AHL reenters the cell and is detected by a LuxR-type response regulator, thereby altering the expression of target genes (25, 34, 35). While E. coli does not contain a LuxI homolog and is unable to synthesize AHL signaling molecules, it does contain SdiA, a LuxR homolog that is capable of sensing and responding to AHL produced by other intestinal microbes (10, 14, 40). For example, it was recently shown that SdiA-AHL signaling aids in the colonization of enterohemorrhagic E. coli O157:H7 in cattle (14). SdiA and AHL have also been shown to activate a gene encoding an EAL domain protein involved in c-di-GMP metabolism in E. coli K-12 (49), as well as genes within the glutamate-dependent acid resistance system in E. coli K-12 and O157:H7 (10). SdiA and AHL were also shown to repress a component of the flagellar basal body in E. coli O157:H7 (10). Several additional studies identified a variety of genes that were either activated or repressed when sdiA was overexpressed (reviewed in reference 34). However, it was recently shown that the expression of several of these genes was unaffected by chromosomally encoded sdiA, suggesting that the sdiA overexpression results should be interpreted with caution (10).

While considerable effort has been invested in determining how SdiA-AHL signaling affects E. coli gene expression, relatively little is known about how sdiA itself is regulated. Most notably, sdiA expression is activated by a second quorum-sensing system mediated by autoinducer 2 (AI-2) (7, 36, 49). In this study, we found that CsrA binds to two sites in the initially translated region of the sdiA transcript and represses SdiA synthesis by blocking ribosome binding. Although CsrA has been shown to regulate the translation of a variety of genes in several organisms (reviewed in references 1 and 39), this is the only known example of CsrA regulating gene expression by binding exclusively in the coding sequence. Moreover, our results establish that CsrA is capable of repressing translation initiation without binding to or otherwise occluding the sdiA SD sequence.

Posttranscriptional regulation by a factor binding exclusively to the coding sequence is not unique to CsrA-mediated repression of sdiA expression. The MicC sRNA inhibits expression of ompD in Salmonella enterica serovar Typhimurium by pairing with codons 23 to 26 of the ompD coding sequence. However, in this case, MicC-ompD mRNA pairing accelerates RNase E-dependent decay of the ompD message rather than repressing translation initiation (26). Thus, although most of the known posttranscriptional control mechanisms involve binding of a regulatory factor to the 5′ untranslated region of target transcripts, these two examples suggest that factor binding to coding sequences could be a common regulatory strategy.

In addition to the relationship between CsrA and quorum sensing identified in this study, CsrA participates in controlling several processes in E. coli, including motility, peptide uptake, carbon metabolism, and biofilm formation (1, 29, 39). The link between CsrA and biofilm formation is especially interesting because at least three mechanisms converge on CsrA-dependent inhibition of this process. First, as SdiA activates biofilm formation (38), CsrA-mediated repression of SdiA synthesis would inhibit the process. Second, CsrA represses translation of the first gene in the pgaABCD operon, which is required for the synthesis of poly-β-1,6-N-acetyl-d-glucosamine (poly- GlcNAc; PGA). This polysaccharide functions as an adhesin in biofilm formation (16, 41, 42). Third, as c-di-GMP activates PGA production, CsrA-mediated inhibition of c-di-GMP production represses biofilm formation (5, 17, 31).

The regulatory circuitry of the Csr system contains three known negative feedback loops that affect CsrA activity, implying that maintenance of optimal CsrA activity is critical (Fig. 8). First, CsrA represses its own translation (46). Second, CsrA indirectly stimulates transcription of csrB and csrC via the BarA-UvrY TCS (38). Finally, CsrA represses expression of CsrD, which is required for degradation of CsrB and CsrC by RNase E. Thus, CsrA indirectly stabilizes its own sRNA antagonists (37). CsrA also indirectly activates its own transcription, although the underlying mechanism is not known (46). Our results here reveal another positive feed-forward loop for CsrA activity (Fig. 8). In this regulatory loop, sdiA stimulates transcription of csrB via the BarA-UvrY TCS (38). Thus, CsrA-mediated repression of SdiA synthesis would reduce transcription of the CsrA antagonists CsrB and CsrC, leading to increased CsrA activity. As CsrA appears to regulate hundreds of genes in E. coli (11), it is not surprising that its activity is tightly regulated. Thus, it is likely that additional regulatory loops that contribute to modulating the level of CsrA activity in the cell will be identified.

Fig. 8.

Regulatory circuitry of the Csr system. See the text for details.