CsrA Participates in a PNPase Autoregulatory Mechanism by Selectively Repressing Translation of pnp Transcripts That Have Been Previously Processed by RNase III and PNPase

Bacterial strains and plasmids.

All bacterial strains used in this study are listed in Table 1. All numbering throughout the manuscript is with respect to the start of pnp translation. E. coli strain S17-1 λpir (27) was used for conditional-replication, integration, and modular (CRIM) plasmid construction (30). CRIM translational fusions using plasmid pLFT (19) were generated as follows. Plasmid pHP4 contains the pnp promoter, leader, and initially translated regions (positions −245 to +182 relative to the start of pnp translation) cloned between the PstI and BamHI sites of pLFT, thereby generating a pnp′-′lacZ translational fusion (where ′-′ indicates that pnp was truncated at the 3′ end and lacZ was truncated at the 5′ end). A GGA-to-GAG mutation in CsrA binding site 1 (BS1) was introduced using the QuikChange protocol (Stratagene), resulting in plasmid pHP11. This construct was generated so that base pairing in the stem was maintained (Fig. 1, left panel). The wild type (WT) and the mutant fusion were integrated into the chromosomal λ att site of strain CF7789 as described previously (30), resulting in strains PLB2176 and PLB2198, respectively. The csrA::kan (29) and pnpΔ683::(St^r Sp^r) (28) alleles were introduced into PLB2176 by P1 transduction using TRMG1655 and SK10019 as donor strains, resulting in strains PLB2184 and PLB2415, respectively. Similarly, strain PLB2418 was constructed by transferring the pnpΔ683::(St^r Sp^r) allele into PLB2184.

The rnc::Cm^r knockout PLB2409 strain was constructed by following a published procedure (31). Briefly, the pKD46 helper plasmid and a PCR product in which the first 210 codons of rnc were replaced with the cat gene from pKD3 were used to disrupt rnc while maintaining expression of the essential downstream era gene. The resulting rnc::Cm^r allele was confirmed by PCR. The rnc::Cm^r allele was subsequently transduced into strains PLB2176 and PLB2184, resulting in strains PLB2416 and PLB2419, respectively.

β-Galactosidase assay.

Bacterial cultures containing pnp′-′lacZ translational fusions were grown in Luria-Bertani (LB) broth supplemented with 50 μg/ml ampicillin at 37°C and harvested at various times throughout growth. β-Galactosidase activity was determined as described previously (32). At least three independent experiments were performed for each strain.

Gel mobility shift assay.

Quantitative gel mobility shift assays followed a published procedure (21). His-tagged CsrA was purified as described previously (33). RNAs containing the full-length pnp leader (pnp_Fl) or one that began at the 3′ RNase III cleavage site at position −81 (pnp_Pr) were synthesized using a RNAMaxx high-yield transcription kit (Agilent Technologies) and PCR-generated DNA templates. In addition, mutant pnp_Pr RNAs were generated in which the GGA motif in CsrA BS1 and/or BS2 was changed to CCA. Gel-purified RNA was dephosphorylated and then 5′ end labeled using T4 polynucleotide kinase (New England BioLabs) and [γ-³²P]ATP (7,000 Ci/mmol). Labeled RNAs were renatured by heating for 1 min at 80°C and slow cooling to 25°C. Binding reaction mixtures (10 μl) contained 10 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 100 mM KCl, 32.5 ng of yeast RNA, 7.5% glycerol, 20 mM dithiothreitol, 4 U of RNase inhibitor (Promega), 0.1 nM labeled RNA, purified CsrA-H6 (various concentrations), and 0.1 mg/ml xylene cyanol. Competition assays also contained unlabeled RNA competitors. Reaction mixtures were incubated for 30 min at 37°C to allow CsrA-RNA complex formation and then fractionated through 15% polyacrylamide gels. Free and bound RNA species were visualized with a Typhoon 9410 phosphorimager (GE Healthcare), and the apparent equilibrium binding constants (K_d) of CsrA-RNA interactions were calculated as described previously (11).

Enzymatic footprinting.

Labeled RNAs were prepared as described above. Binding reaction mixtures (10 μl) were identical to those used in the gel mobility shift assay except that the concentration of labeled pnp RNA was increased to 50 nM and 200 μg/ml acetylated bovine serum albumin (BSA) was included in the reaction mixture. Following a 15-min incubation at 37°C to allow CsrA-RNA complex formation, RNase T1 (0.12 U) was added and incubation was continued for 15 min at 37°C. Reactions were stopped by addition of 10 μl of stop buffer (95% formamide, 0.025% SDS, 20 mM EDTA, 0.025% bromophenol blue, 0.025% xylene cyanol, 760 μg/ml yeast RNA), and the reaction mixtures were placed on ice. RNase T1 and base hydrolysis ladders were prepared as described previously (34). Samples were fractionated through a 6% (vol/vol) polyacrylamide–8 M urea sequencing gel. Cleavage products were visualized using a phosphorimager.

mRNA half-life analysis.

Strains MG1655 (WT) and TRMG1655 (csrA::kan) (29) were grown at 37°C in LB broth to the exponential phase prior to the addition of rifampin (200 μg/ml) to prevent transcription initiation. After 1 min following rifampin addition, 0.8-ml aliquots were removed at various times and total cellular RNA was prepared by the hot phenol method as described previously (35). For Northern analysis, RNA samples were fractionated through 1.0% denaturing formaldehyde-agarose gels. RNA was transferred to a Hybond N+ membrane (Amersham Biosciences) by capillary blotting with 10× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) (pH 7.0) for 16 h and fixed to the membrane by UV cross-linking at a wavelength of 302 nm for 3 min. Oligonucleotides pnpTL+11R (5′-GGATTAAGCAATGTAATATCCTTTCTC-3′) for pnp mRNA hybridization and 16SrRNAR (5′-GCAGGTTCCCCTACGGTTACCT-3′) for 16S rRNA hybridization, which served as a loading control, were 5′ end labeled with [γ-³²P]ATP and T4 polynucleotide kinase (New England BioLabs). Hybridization was performed according to the manufacturer's instructions (Amersham Biosciences). Labeled RNA species were visualized with a phosphorimager and quantified using ImageJ software (36).

In vitro coupled transcription-translation assay.

Plasmid pT7-pnpFl′-′lacZ contains a T7 promoter driving transcription of a pnp′-′lacZ translational fusion (−156 to +182). This plasmid was designed such that transcription initiated at the natural pnp transcription start site but with two additional G residues for initiation by T7 RNA polymerase (RNAP). Plasmid pT7-pnpPr′-′lacZ was constructed in a similar manner except that transcription began at −81, thereby mimicking a pnp transcript that was fully processed by RNase III and PNPase. Plasmid pT7-pnpPrM′-′lacZ is identical to pT7-pnpPr′-′lacZ except that it contains a GGA-to-GAG mutation in BS1. These three plasmids were used as the templates for coupled transcription-translation reactions using a PURExpress in vitro protein synthesis kit according to the instructions of the manufacturer (New England BioLabs). Reaction mixtures containing 20 nM plasmid DNA template and various concentrations of purified His-tagged CsrA were incubated for 2 h at 37°C. β-Galactosidase activity was determined according to the manufacturer's instructions.

CsrA represses pnp expression.

PNPase is a 3′-to-5′ exoribonuclease and a component of the E. coli degradosome (23, 24). PNPase participates in an autoregulatory mechanism that also involves RNase III. RNase III-mediated cleavage of a pnp leader RNA structure removes the top portion of the stem such that the transcript containing the downstream pnp coding sequence remains base paired to the upstream (5′) cleavage fragment (Fig. 1). PNPase then degrades the 5′ fragment from the newly generated 3′ end, resulting in a pnp transcript that is regulated by an unknown translational repression mechanism (25, 26). Our previously published RNA-seq studies identified pnp as a CsrA target (19). Visual inspection of the untranslated leader region of pnp mRNA identified two putative CsrA binding sites, each containing the highly conserved GGA motif. The downstream site (BS2) overlaps the pnp SD sequence, suggesting that CsrA might be capable of repressing pnp translation. Prior to the action of RNase III and PNPase, the upstream site (BS1) would be sequestered in the large secondary structure, perhaps inhibiting CsrA binding (Fig. 1). Thus, we tested a model in which CsrA participates in the pnp regulatory mechanism by repressing translation following the action of RNase III and PNPase. Translational repression could then lead to accelerated decay of pnp mRNA.

We first examined expression of a pnp′-′lacZ translational fusion in wild-type and CsrA-deficient strains. Expression in the wild-type background was low throughout growth. In contrast, expression of the fusion in the csrA mutant background increased dramatically in exponential phase and remained high in stationary phase, indicating that csrA represses pnp expression throughout growth, with maximal 20-fold repression occurring during stationary phase (Fig. 2). We also examined expression of a fusion in which the GGA motif in BS1 was replaced with GAG. This mutant leader was designed to prevent CsrA binding to BS1 while maintaining base pairing of the long stem recognized by RNase III (Fig. 1, left panel). Expression of the mutant fusion was ∼5-fold higher than expression of the wild-type fusion, suggesting that CsrA is unable to fully repress pnp expression when CsrA is not bound to BS1. We did not test mutations in BS2 since mutations in this site would also disrupt the pnp SD sequence.

CsrA binds to fully processed pnp mRNA.

To determine whether processing of the pnp leader transcript by RNase III and PNPase affects CsrA binding, we performed quantitative gel mobility shift assays with two in vitro-generated transcripts, both of which contained BS1 and BS2. One transcript contained the full-length pnp leader (pnp_Fl), while the 5′ end of the other transcript began at the 3′ RNase III cleavage site (pnp_Pr), thereby mimicking the transcript produced in vivo following RNase III and PNPase processing (Fig. 3A). CsrA bound tightly to fully processed pnp_Pr RNA with an apparent K_d of 20 nM. In contrast, only weak binding to full-length pnp_Fl RNA was observed at the highest CsrA concentration tested (Fig. 3B and E). The specificity of CsrA-pnp_Pr RNA interaction was investigated by performing competition experiments with specific (pnp_Pr) and nonspecific (phoB) unlabeled RNA competitors. Unlabeled pnp_Pr was an effective competitor, whereas phoB was not, indicating that the CsrA-pnp RNA interaction is specific (Fig. 3C). We also tested CsrA binding to pnp_Pr RNA containing GGA-to-CCA mutations in BS1 and/or BS2 (Fig. 3D and E). CsrA bound to the transcript containing BS1 only (BS2 mutant) with an apparent K_d of 57 nM, while the affinity of CsrA for the transcript containing BS2 only (BS1 mutant) was about 2-fold lower. Binding was not observed to the transcript containing mutations in both binding sites.

CsrA-pnp_Pr RNA footprint experiments were performed to verify CsrA binding to BS1 and BS2. Bound CsrA reduced RNase T1-mediated cleavage of G residues at positions −73, −71, −70, −67, and −65 within BS1. Similarly, CsrA prevented cleavage of positions −10 and −9 within BS2 (Fig. 4). In addition to protection of BS1 and BS2, bound CsrA caused increased cleavage of G3 within the UUG start codon and at positions G16 and G20 further downstream, indicating that CsrA binding alters the RNA structure in the translation initiation region. The finding that CsrA protected BS1 at the lowest concentration of CsrA used in this analysis, while protection of BS2 required a higher CsrA concentration, is consistent with our gel shift analysis demonstrating that CsrA has higher affinity for BS1 (Fig. 3D). In addition, our structure mapping data in the control lane without CsrA (Fig. 4) are consistent with the two CsrA binding sites being single stranded in the fully processed RNA, as depicted in our RNA structural model (Fig. 1, right panel). Taken together, our binding studies indicate that BS1 and BS2 constitute authentic CsrA binding sites. Moreover, our results demonstrate that CsrA binds to fully processed pnp mRNA with high affinity and specificity but not to the unprocessed pnp leader transcript.

CsrA represses pnp translation.

There are two mechanisms in which CsrA has been shown to repress gene expression posttranscriptionally. First, bound CsrA can decrease the stability of pnp mRNA. Second, CsrA can repress translation by blocking ribosome binding. These two possibilities are not mutually exclusive, as destabilization of mRNA can be an indirect consequence of translation inhibition. We first examined the stability of pnp mRNA in WT and CsrA-deficient strains by Northern blotting. The pnp half-life durations in the wild-type and csrA mutant strains were 4.3 and 6.3 min, respectively (Fig. 5). This difference is too small to account for the large difference in expression observed in the pnp′-′lacZ translational fusion studies (Fig. 2).

We next used the in vitro coupled transcription-translation PURExpress system to determine whether CsrA inhibits pnp translation. Three different plasmids carrying pnp′-′lacZ translational fusions, all of which were driven by identical T7 RNA polymerase promoters, were used in this analysis. One plasmid gave rise to full-length pnp leader RNA (pnp_Fl) identical to that used in our in vivo expression studies (Fig. 2), while a second was designed to generate a transcript mimicking the fully processed pnp transcript (pnp_Pr). These two transcripts were essentially identical to those used in our gel shift analysis (Fig. 3A and B). The third construct contained the BS1 GGA-to-GAG mutation in pnp_Pr. Expression from each construct was measured by determining β-galactosidase activity. Addition of CsrA to the system caused a slight decrease in expression of the pnp_Fl construct; expression was reduced only 3% and 19% at 2.5 and 10 μM CsrA, respectively (Fig. 6). In contrast, expression of the pnp_Pr construct was reduced 54% and 91% at the same two CsrA concentrations. Expression of the pnp_Pr construct containing the BS1 mutation exhibited an intermediate level of CsrA-mediated repression. We conclude that bound CsrA represses translation of pnp but only after the RNA has been fully processed by RNase III and PNPase. These data also suggest that the small effect of CsrA on pnp mRNA stability (Fig. 5) was an indirect consequence of altered translational repression or RNA secondary structure.

CsrA-mediated repression of pnp expression requires prior processing by RNase III and PNPase.

Our data described above are consistent with a model in which CsrA represses translation of pnp transcripts following processing by RNase III and PNPase (Fig. 1). In the absence of processing by these two ribonucleases, the CsrA binding sites would be inaccessible, since both binding sites are sequestered in the RNA secondary structure. However, in the unprocessed transcript, the pnp SD sequence would be sequestered in an RNA structure, which would inhibit translation (Fig. 1, left panel). Following processing by these two nucleases, the pnp leader RNA refolds such that the SD sequence is single stranded and available for ribosome binding. However, this processed and refolded transcript would then be subject to CsrA-mediated translational repression since both of the CsrA binding sites are single stranded. Since our model predicts that CsrA functions downstream of RNase III and PNPase, we expected that disrupting rnc or pnp would suppress the effect of the csrA mutation on pnp expression. Thus, epistasis studies were carried out to test this model by examining expression of a pnp′-′lacZ translational fusion containing the entire leader region. Since processing of the pnp transcript is initiated by RNase III-mediated cleavage, we first compared expression levels of the pnp′-′lacZ fusion in WT, csrA, rnc, and csrA rnc strains (Fig. 7A). As we had previously observed (Fig. 2), pnp expression was much higher in the csrA mutant, which reflects the loss of CsrA-mediated translational repression of the fully processed transcript. We also found that expression was elevated in the rnc strain, indicating that sequestration of the pnp SD sequence in the unprocessed mRNA is less effective at repressing translation than CsrA-mediated repression of processed transcripts (Fig. 7A). Notably, as predicted by our model, the large increase in expression of the csrA mutant strain was completely suppressed in the RNase III-deficient background. However, we cannot account for the unexpected result that expression in the csrA rnc double-mutant strain was actually 2-fold lower than expression in the rnc strain.

We next compared expression levels in the WT, csrA, pnp, and csrA pnp strains (Fig. 7B). Expression was elevated in the pnp mutant strain but to a much smaller extent than in the csrA strain. Importantly, as predicted by our model, the large increase of expression in the csrA mutant strain was completely suppressed in the PNPase-deficient background; expression levels in the pnp and pnp csrA strains were essentially identical. Taken together with the results of our in vitro studies, we conclude that CsrA is effective only at binding to and repressing translation of pnp transcripts that are fully processed by RNase III and PNPase.