Initial genome sequencing projects revealed three csrA-like genes in the carbon storage regulator family (32). By performing a bioinformatic search of the L. pneumophila Philadelphia-1 genome, we identified five csrA-like genes, including the well-characterized canonical csrA (Fig. 1A) (13). The amino acid sequences of the four paralogs exhibit 28% to 58% identity and 59% to 80% similarity.

The amino acid residues determined to be critical for CsrA function in E. coli are retained in CsrA proteins across myriad genera surveyed, including the canonical L. pneumophila CsrA protein (1). By aligning the amino acid sequences of the csrA paralogs of L. pneumophila Philadelphia-1, we confirmed that several key residues, including the arginine residue that is essential for the RNA binding activity, are retained in all five (1) (Fig. 1A). Based on the conserved amino acid sequence, the five L. pneumophila Philadelphia-1 loci are likely valid csrA paralogs.

To determine if the Philadelphia-1 strain is unique in encoding multiple csrA paralogs, the in silico search was expanded to include 13 more fully sequenced strains of legionellae: 8 additional independent isolates of L. pneumophila, 2 laboratory strains derived from Philadelphia-1, 2 strains of L. longbeachiae, and 1 strain of L. drancourtii (see Table S1 in the supplemental material). Altogether, 71 csrA-like genes were identified, and each retained several residues predicted to be key for RNA binding (1). In particular, the arginine residue critical for RNA binding was conserved in 70 of the 71 proteins; in the one exception, another cationic residue, lysine, was in place of the arginine (see Table S2). Each of the 14 surveyed strains contained between 4 and 7 csrA-like genes. Without exception, all 14 encoded not only the canonical csrA gene but also a copy of a previously uncharacterized paralog, lpg1593, which we name here csrR. Accordingly, both the csrA and csrR genes appear to belong to the core genome of Legionella species. In stark contrast, the remainder of the csrA-like genes are located within previously characterized or putative integrative conjugative elements (ICEs), large genomic islands known or predicted to transfer horizontally between strains of legionellae (see Table S3) (33–35).

To compare the sequences of the 63 Legionella csrA genes found in the 12 strains, excluding the two laboratory-derived strains Lp02 and JR32, we generated a heat map that displays the amino acid similarity of each protein to each of the other paralogs. By this approach, we recognized three clusters: two that correspond to the presumptive core CsrA and CsrR proteins and a third that encompasses a more diverse group of ICE-related CsrA-like proteins (Fig. 1B). Because csrR is encoded by all surveyed strains, is highly conserved, and retains key RNA binding residues, we hypothesized that this gene would regulate traits fundamental to L. pneumophila’s life style, as CsrA does.

Since csrR is a paralog of the essential regulator csrA, we first tested whether the two genes regulate the same developmental stage. To do so, we either deleted or induced expression of the csrR gene and then assayed phenotypes known to be regulated by the canonical csrA gene (13). Analyzed under standard laboratory culture conditions (37°C in rich ACES-buffered yeast extract with thymidine [AYET] medium), neither loss nor gain of csrR function altered expression of any L. pneumophila replication-phase or transmission-phase traits tested. These included, for broth cultures, growth, pigmentation, motility, and sensitivity to sodium, heat, and osmotic stress, as well as infection and intracellular growth in primary mouse macrophages and amoebae (data not shown) (13). Since csrR appeared unlikely to regulate traits that are hallmarks of the commonly studied replicative and transmissive cell types, we next considered conditions that L. pneumophila likely encounters in the environment. Indeed, a dual-CsrA system (RsmA/RsmF) equips P. aeruginosa to modulate biofilm formation (36).

The legionellae naturally thrive in freshwater and soil habitats, and L. pneumophila can survive for long periods in water in a VBNC state (29, 30). Therefore, we tested whether csrR coordinates differentiation to a resilient environmental cell type. Employing the method of Garduño and colleagues (30), we inoculated autoclaved tap water at 45°C with wild-type (WT) L. pneumophila, a csrR mutant, or a csrR mutant strain in which a wild-type csrR allele had been integrated into the chromosome. After 24 days, we quantified bacterial viability by fluorescence microscopy using a LIVE/DEAD BacLight stain and culturability on rich nutrient agar (37). The wild-type culture retained 79% viability, which was significantly higher than the mutant culture’s 50% viability (P value = 0.049; Fig. 2A). Likewise, the level of colony forming units (CFU) recovered at 24 days relative to day 0 was significantly higher for the wild-type culture than for the mutant culture (5.6% of WT, 0.3% of mutant; P value = 0.0027) (Fig. 2B). Although the merodiploid mutant strain exhibited an intermediate level of viability as measured by microscopy assay (Fig. 2A) (P > 0.05 compared either to the wild type or to the mutant), genetic complementation was incomplete. Two control experiments indicated that the lack of full complementation was likely due to poor expression of csrR. First, quantitative reverse transcriptase PCR (qRT-PCR) revealed that the merodiploid strain contained approximately 5% of the wild-type levels of csrR RNA (data not shown). Second, we quantified reduced viability relative to wild-type L. pneumophila by both microscopy and CFU assays after isolating and analyzing nine independent csrR mutant strains constructed by three different strategies.

Because knowledge of a protein’s regulation can provide insight into its function, we next analyzed the upstream sequences of csrR. Bioinformatic analysis identified motifs typical of CsrA binding sites in the 5′ noncoding region of csrR (11). As a first step to analyze csrR regulation, we mapped by 5′ rapid amplification of cDNA ends (5′-RACE) its transcriptional start site to −52 relative to the translational start (data not shown). Three independent sequencing events revealed that the transcript starting with the −52 start site was predominant, although rare transcripts started at −53 or −54. On the basis of the data indicating the predominant species, we made several predictions about the csrR mRNA. Putative −10 and −35 sites as well as the likely SD sequence approximately 10 bases 5′ of the translational start codon were identified (Fig. 3A). RNA-fold software predicted a secondary structure for the csrR transcript in which the SD sequence is positioned in a loop extending from a long stem (Fig. 3B), a common binding motif for CsrA protein (38). Accordingly, we postulated that the csrR transcript is bound and its stability regulated by CsrA.

To test directly whether CsrA protein binds to csrR mRNA, we performed an in vitro assay. Purified CsrA protein was incubated with 5′-end-labeled csrR RNA, and then binding was analyzed by gel electromobility shift assay. As predicted by the in silico analysis, CsrA protein bound tightly to csrR RNA, with a calculated dissociation constant of 32 ± 9 nM (Fig. 4A). This physical interaction was specific, since incubation with an excess of unlabeled csrR RNA effectively competed with CsrA protein binding to labeled csrR RNA (Fig. 4B). In contrast, incubation with even higher concentrations of a heterologous competitor RNA, E. coli phoB, did not inhibit CsrA binding to csrR. Therefore, CsrA protein binds csrR RNA tightly and specifically in vitro.

To verify our observation that CsrA protein binds to the csrR transcript at the predicted stem-loop structure surrounding the SD sequence, we performed RNA footprinting experiments. Interestingly, although CsrA did bind at the predicted SD site (“BS2” in Fig. 4C), the protein also protected a GGA motif within the first two codons of the csrR coding region (“BS3” in Fig. 4C). Although binding in the 5′ untranslated region of its target gene is typical, in some cases CsrA does bind in coding regions (12). The RNA footprinting assay also revealed a potential third binding site (BS1); however, because this third site was protected only weakly, we focused next on the stronger BS2 and BS3 binding sites.

To analyze in live cells whether either or both BS2 and BS3 binding sites contribute to csrR posttranscriptional regulation, we constructed a series of green fluorescent protein (GFP) reporters (Fig. 5A). For the BS2-only transcriptional reporter, gfp expression is driven by a sequence corresponding to approximately 300 bp immediately 5′ of the csrR coding region that includes BS2. Although this reporter contains the csrR SD sequence, its gfp SD sequence likely controls translation; thus, the BS2-only construct is referred to as a transcriptional reporter. The BS2/BS3 translational reporter includes both the BS2 and BS3 directly 5′ and in frame with gfp coding sequences, eliminating the gfp SD sequence and including the first 5 codons of csrR. For the BS2 stem-loop-disrupted (BS2-sld)/BS3 reporter, the BS2 stem-loop structure was disrupted on the translational reporter by changing the six consecutive adenines immediately 5′ of the SD sequence to uracils, thereby preventing stem formation with the eight uracils 3′ of the SD sequence.

L. pneumophila strains carrying each reporter were cultured in broth, and then aliquots were assessed for GFP fluorescence. As a marker for the growth phase, we also analyzed a reporter of the flaA flagellin gene, which is strongly induced in post-exponential (PE) phase (39). The BS2-only reporter generated a high level of fluorescence in exponential (E) phase (~15,000 relative fluorescence units [RFU]) that then decreased (to ~5,000 RFU) in PE phase (Fig. 5B), the period when transmission traits are induced (16). In comparison to the BS2-driven expression, in E phase, the BS2/BS3 and the BS2-sld/BS3 translational reporters demonstrated 3-fold and 6-fold less fluorescence, respectively (Fig. 5B). Therefore, the BS3 site appears to destabilize the mRNA and/or inhibit translation of the transcript. In addition, on the basis of the reduced fluorescence seen when the stem-loop was disrupted, the BS2 site may somehow stabilize the mRNA.

To assess whether CsrA protein destabilizes the csrR transcript in vivo, we utilized a previously characterized conditional csrA mutant (13). In this strain, the chromosomal csrA locus is deleted, and a plasmid carries a functional copy of the gene under IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible control. Thus, we measured the impact of CsrA protein on csrR transcript levels by culturing cells in the presence and absence of IPTG.

To determine whether CsrA alters the stability of csrR transcripts, the conditional csrA mutant was cultured with or without IPTG for 9 h, which corresponded to E phase, a period when csrR promoter activity is high (Fig. 5B) (40). Next, transcription was inhibited by treatment with rifampin. RNA was isolated 0 and 15 min later, and then the relative levels of csrR mRNA were quantified by qRT-PCR. There was an inverse correlation between the presence of CsrA protein and csrR mRNA levels (Fig. 6A), indicating that the presence of CsrA decreased csrR transcript stability. A similar pattern has been reported for other CsrA-mRNA interactions (41, 42).

We then tested in vivo whether CsrA destabilization of csrR transcript depends on its BS3 motif (Fig. 4 and 5). To do so, we performed qRT-PCR on wild-type L. pneumophila strains that carried plasmids in which an IPTG-inducible promoter drives either a wild-type or BS3 mutant csrR allele that also encodes a 6-histidine epitope. To ensure the presence of active CsrA, RNA was isolated from strains cultured with or without IPTG for 9 h in E phase. Induction of the BS3 mutant generated >8-fold more csrR transcript than the corresponding uninduced control, whereas the wild-type allele showed a <2-fold increase (P value = 0.0327; Fig. 6B). Thus, in cultures of wild-type L. pneumophila, csrR RNA is destabilized by a mechanism that requires its BS3 binding site for the CsrA repressor protein.

Finally, we tested the prediction that, by decreasing csrR transcript stability, CsrA repressor binding reduces CsrR protein levels. For this purpose, we performed Western analysis on the strains that encode His-tagged CsrR on mRNA that either contained or lacked the BS3 RNA binding site for CsrA protein. CsrR protein was detected only in L. pneumophila that carried the csrR allele that lacked the BS3 binding site for CsrA (Fig. 6C). Therefore, CsrR protein expression is inhibited by the BS3 sequence positioned at the start of the csrR opening reading frame, most likely by direct binding of csrR mRNA by the CsrA repressor protein (Fig. 3 to 5).

Strains, plasmids, and primers used in this study are listed in Table S4 in the supplemental material. L. pneumophila Philadelphia-1 laboratory-derived strain Lp02, a thymidine auxotroph, was cultured at 37°C in AYE broth and on ACES-buffered charcoal (Fisher)-yeast extract (Becton, Dickinson) (CYE) agar supplemented with 100 µg/ml thymidine (Sigma), 400 µg/ml cysteine (Fisher), and 135 µg/ml ferric nitrate (J. T. Baker) (51). Thymidine was omitted when culturing thymidine prototroph strains. When necessary for antibiotic selection of mutants or plasmids, media were supplemented with kanamycin (Sigma) (10 µg/ml); chloramphenicol (Fisher) (5 µg/ml); or gentamicin (Gibco) (10 µg/ml). Where indicated, gene expression was induced by adding IPTG (Gold Biotechnology) to reach a final concentration of 200 µM. For all experiments, colonies were first inoculated into broth, incubated overnight, and diluted to an optical density at 600 nm (OD₆₀₀) of 0.05 to 0.2 and then cultured to the E phase (OD₆₀₀ of 1.0 to 2.0) or PE phase (OD₆₀₀ of 3.7 to 4.0), as indicated.

To identify the csrA-like genes in the Legionella pan-genome (see Table S1 in the supplemental material), the amino acid sequence of CsrA from L. pneumophila Philadelphia-1 was submitted to BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi). The heat map was generated by first entering the percent amino acid identity of each protein pair into an Excel spreadsheet and then assigning a grayscale to the values. The proteins were then manually clustered by similarity. The order of the proteins from top to bottom (and left to right) is listed in Table S2.

A chloramphenicol resistance cassette was amplified from pKD3 (52) using primers ZA44 and ZA45. Flanking homology to ~500 bp 5′ and 3′ of csrR was generated from Lp02 genomic DNA using primers ZA37 and ZA43 and primers ZA46 and ZA38, respectively. The three PCR products were then joined by splicing by overhang extension (SOE) PCR using outermost primers ZA37 and ZA38 to generate the csrR::camR allele used to replace the csrR gene by natural transformation (53), producing strain MB1369.

To insert a wild-type csrR locus into a neutral site on the chromosome of the csrR mutant, the csrR gene and its native promoter were integrated into the 154-bp intergenic region between lpg2528 and lpg2529 (54). To do so, we first constructed plasmid pSU-pZLKm. The multicloning site of pSU2719 (55) was amplified using primers SC1 and SC2 and cloned into the SalI restriction site of pZL790 (54) to create vector pZL790-MCS. Next, a kanamycin resistance cassette flanked by FLP recombination target (FRT) sites was amplified from pKD4 (52) with primers Sc3 and Sc4 and cloned into the SacI site of pZL790-MCS, creating vector pZL790-MCS-Km. Then, the lpg2528-MCS-Km-lpg2529 fragment from pZL790-MCS-Km was amplified with primers SC5 and SC6 and ligated into pSU2719 at the Nco1 and HindIII sites, after the HindIII site was made blunt with Klenow fragment, producing vector pSU-pZLKm. csrR with its native promoter (~300 bp 5′ of the translational start) was amplified from Lp02 genomic DNA with primers ZA49 and ZA60 and ligated into the EcoRI site of pSU-pZLKm. This plasmid was then integrated by natural transformation into the 154-bp intergenic region between lpg2528 and lpg2529 in csrR mutant MB1369 to generate strain MB1372. Merodiploid mutants were verified by growth on CYETcam+kan and by PCR.

L. pneumophila is naturally a thymidine prototroph, but the Lp02 laboratory strain was made a thymidine auxotroph in 1993 to exploit “thymineless death” as a strategy to identify intracellular growth mutants (56). To avoid the potentially confounding phenotype of thymineless death under conditions of prolonged starvation, thymidine prototrophy was restored to strains MB110, MB1369, and MB1372 by replacing by natural transformation their mutant thyA allele with the wild-type thyA gene from L. pneumophila Philadelphia-1 encoded on pJB3395 as described previously (53), generating strains MB1368, MB1370, MB1371, MB1373, and MB1374.

Thymidine prototroph strains MB1368, MB1370, MB1371, MB1373, and MB1374 were analyzed to avoid thymidineless death during the prolonged incubation. The wild type and two independent isolates of each mutant and merodiploid strain were cultured in AYET medium to the E phase, diluted to an OD₆₀₀ of 0.2, and then cultured to the PE phase. The cultures were then washed twice by suspension in 50 ml of autoclaved tap water and centrifugation at 4,500 × g for 15 min. Cells were then resuspended in 10 ml of autoclaved tap water in 15-ml conical tubes with the caps loosely affixed to allow air exchange and incubated statically in a 45°C incubator. The viability of duplicate samples was assessed by LIVE/DEAD BacLight (Life Technologies) staining. A 165-μl volume of culture was mixed with 165 µl of autoclaved tap water, 1 µl of each stain was added, and the samples were incubated at room temperature in the dark for 15 min. Aliquots (10 µl) were mounted on slides pretreated for 5 min with 5 µl poly-l-lysine, and 100 bacteria were scored for each replicate of each strain. Culturability of duplicate samples was quantified by plating 10-fold serial dilutions in autoclaved tap water on CYE agar.

For all experiments analyzing RNA, 2 ml of liquid culture was pelleted by centrifugation at 10,000 × g for 5 min at 4°C, the cells were resuspended in 1 ml of Trizol reagent (Life Technologies), and RNA was isolated following the manufacturer’s protocol. Residual DNA was degraded using a Turbo DNA-free kit (Life Technologies) following the manufacturer’s protocol, and RNA was stored at −20°C.

5′-RACE was performed using a ligation-anchored PCR strategy described previously (57). Briefly, RNA was isolated as previously described, and first-strand cDNA synthesis was performed using SuperScript II reverse transcriptase (Life Technologies), the manufacturer’s protocol, and ZA54 as the csrR gene-specific primer. The ZA80 adaptor oligonucleotide, with 5′ phosphorylation for the ligation and a 3′ amino modifier to prevent its ligation, was ligated to cDNA using T4 RNA ligase (NEB) following the manufacturer’s protocol. PCR amplification was performed using ZA81 and ZA54 with platinum Taq DNA polymerase (Invitrogen), and ligation into pGEM T-easy (Promega) was performed following the manufacturer’s protocols. Sequencing from the pGEM plasmid was performed on three independent clones using primer SL3. mRNA fold prediction was done using the IDT DNA UNAFold tool, inputting the first 70 bp of the csrR mRNA sequence by using the experimentally determined transcriptional start.

csrA with a His-tagged epitope was amplified from Lp02 genomic DNA using primers ZA33 and ZA34, ligated into plasmid p206gent, and transformed into E. coli Rosetta(DE3) pLysS. The strain was cultured in Terrific Broth media containing 20 µg/ml chloramphenicol and 10 µg/ml gentamicin at 37°C to an OD₆₀₀ of ~0.5, supplemented with 500 µM IPTG, and cultured for 16 h at 22°C. Cells were then harvested by centrifugation and flash frozen with liquid N₂. Protein was purified using a 5-ml Hi-Trap metal affinity column (GE Healthcare) following the manufacturer’s instructions. Protein was loaded in a His buffer (25 mM NaH₂PO₄, 500 mM NaCl, 20 mM imidazole) and eluted using an imidazole gradient of 20 to 300 mM. Protein-containing fractions were then diluted 1:6 in the elution buffer (due to low solubility) and dialyzed overnight at 4°C in freezing buffer (10 mM Tris-HCl [pH 7.4], 10 mM MgCl₂, 100 mM KCl, 10% glycerol).

A csrR fragment (−168 to +103) was amplified from Lp02 genomic DNA using primers HY2 and HY4 and cloned into the EcoRI and HindIII sites of the pTZ18U polylinker (Stratagene).

RNA was synthesized using a MEGAscript kit (Life Technologies) and a PCR fragment containing a T7 promoter derived from primer HY1. A DNA fragment containing csrR sequence from +3 to +84 relative to the start of transcription was used as the template. Gel-purified RNA was 5′ end labeled with [γ-³²P]ATP. RNA suspended in Tris-EDTA (TE) buffer was heated to 90°C for 1 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, 200 ng/µl yeast RNA, 0.2 mg/ml bovine serum albumin (BSA), 7.5% glycerol, 20 mM dithiothreitol (DTT), 0.1 nM csrR RNA, CsrA-H6 (various concentrations), and 0.1 mg/ml xylene cyanol. Competition assay mixtures also contained unlabeled competitor RNA. To allow CsrA-RNA complex formation, reaction mixtures were incubated for 30 min at 37°C. Samples were then fractionated through native 15% polyacrylamide gels using 0.5× Tris-borate-EDTA (TBE) running buffer. Radioactive bands were visualized with a phosphorimager and quantified using ImageQuant 5.2 software. The apparent equilibrium binding constant (K_d) of CsrA-csrR RNA interaction was calculated as described previously (58).

RNA was synthesized using a MEGAscript kit and a PCR fragment containing a T7 promoter derived from primer HY3. The template was a DNA fragment containing csrR sequence extending from 32 to +84 relative to the start of transcription. Binding reactions (10 µl) containing 5 nM csrR RNA and the indicated concentrations of CsrA-H6 were otherwise identical to those in the gel shift assay. After the initial binding reaction, 0.075 U RNase T1 (Roche) was added to the reaction mixtures, and incubation continued for 15 min at 37°C. Reactions were terminated by addition of 10 µl of gel loading buffer, and reaction mixtures were placed on ice. Partial alkaline hydrolysis and RNase T1 digestion ladders of each transcript were prepared as described previously (59). After samples were fractionated through the use of standard 6% polyacrylamide sequencing gels, radioactive bands were visualized with a phosphorimager.

To construct the csrR-GFP BS2-only transcriptional reporter, ~300 bp immediately 5′ of the csrR coding region was amplified from Lp02 genomic DNA using primers ZA49 and ZA50, ligated directly 5′ of GFPmut3 in vector pBH6119, a derivative of pJB98 that encodes a promoterless GFPmut3 locus (39), and then electroporated into Lp02 to generate strain MB1375. For the csrR-GFP BS2/BS3 translational reporter, the sequence incorporated into the BS2 reporter was extended to include the first 5 codons of csrR. After PCR amplification using primers ZA49 and ZA73, the fragment was fused directly 5′ and in frame with the second codon of GFPmut3 amplified from pBH6119 using primers ZA72 and ZA84 via the SOE PCR strategy using primers ZA49 and ZA84. This product was then ligated into pBH6119, replacing the promoterless GFPmut3, and electroporated into Lp02 to generate strain MB1376. To construct the csrR-GFP BS2-sld/BS3 translational reporter, the stem-loop structure of BS2 predicted to favor CsrA binding was disrupted by changing the six consecutive adenines immediately 5′ of the SD sequence to thymidines in the DNA sequence (resulting in uracils in the RNA). To do so, the mutation was inserted into two overlapping PCR products using pcsrR-GFP_BS2/BS3 as the template, primer pair ZA49 and ZA75, and primer pair ZA74 and ZA84. The two products were then joined via SOE PCR using primers ZA49 and ZA84 to generate the BS2 mutant allele, which was then ligated into pBH6119, replacing the promoterless GFPmut3, and electroporated into Lp02 to produce strain MB1377. For fluorescence experiments, colonies of MB1375 to MB1377 were cultured in broth to the E phase and then diluted to an OD₆₀₀ of 0.1. Aliquots were collected at the times shown and normalized to an OD₆₀₀ of 1, and their fluorescence was quantified. Strain MB355, containing pflaA-GFP, served as a marker of the entry into the PE phase, and strain BH006 carrying pBH6119, encoding GFP but no promoter, was the negative control (39). Growth levels of all strains were similar, as assessed by OD₆₀₀ readings throughout the experiment (data not shown).

Wild-type csrR was amplified from Lp02 genomic DNA using primers ZA59 and ZA60 and the PCR product ligated into p206cam (60). This plasmid was electroporated into Lp02 to generate strain MB1378.

The BS3 mutant allele of csrR was constructed by SOE PCR. Two PCR products were generated. Product 1 started ~500 bp 5′ of the transcriptional start generated by forward primer ZA37 and included a GAT-to-TTG mutation in the second codon of csrR that was inserted using reverse primer ZA83. Product 2 encoded the csrR coding region, included the GAT-to-TTG mutation in the second codon that was inserted by the forward primer ZA82, and extended ~500 bp 3′ of the coding region by using reverse primer ZA38. One microliter of each of these products then served as the template in a SOE PCR using outermost primers ZA37 and ZA38. Next, this PCR product containing the BS3 mutation was used as the template to amplify a His-tagged version of only the coding region, using primers ZA59 and ZA60. Finally, the product was ligated into p206cam. This plasmid was electroporated into Lp02 to generate strain MB1379.

The conditional csrA mutant strain MB464 was cultured on CYE agar supplemented with 200 µM IPTG and then inoculated into AYET medium containing 200 µM IPTG. After culture to the E phase, cells were collected by centrifugation, resuspended in AYET medium without IPTG, and then divided between two tubes that contained or lacked 200 µM IPTG. These samples were then cultured to an OD₆₀₀ of ~1 and aliquots collected for RNA isolation. Next, 100 µg/ml rifampin was added and cultures were incubated for 15 min before a second aliquot was collected for RNA isolation.

cDNA was generated using 1 µl of isolated RNA in a 20-µl reaction mixture with an iScript cDNA synthesis kit from Bio-Rad following the manufacturer’s protocol. A 1:50 dilution of cDNA served as the template for qRT-PCR in 20-µl reaction mixtures using iQ SYBR green Supermix (Bio-Rad). Primers ZA53 and ZA54 were used to assess csrR mRNA levels, which were normalized to the internal control 16S rRNA amplified using primers SL1 and SL2 and threshold cycle (ΔΔC_T) analysis.

Strains MB1378 and MB1379 were inoculated from colonies and cultured to the E phase, diluted to an OD₆₀₀ of 0.05, and divided into two tubes containing or lacking 200 µM IPTG. After incubation for ~15 h to the E phase, cultures were normalized to an OD₆₀₀ of 10 by centrifugation of aliquots at 13,000 rpm for 2 min, and then cells were resuspended in 100 µl of Laemmli buffer (2% SDS, 10% glycerol, 5% 2-mercaptoethanol, 0.005% bromophenol blue, 62.5 mM Tris-HCl, pH 6.8). Samples were then lysed by boiling for 5 min and debris pelleted at 13,000 rpm for 2 min. Proteins were separated on a 12% mini-Protean TGX precast gel (Bio-Rad); a Precision Plus Kaleidoscope protein standard ladder (Bio-Rad) was used as a size marker and to verify transfer. To determine relative loading levels and transfer of each sample, Ponceau S (Fisher) staining was performed on a duplicate membrane. CsrR was detected via the 6×His epitope that was inserted in frame at its carboxy terminus using a 1:5,000 dilution of anti-His (C-terminal)-horseradish peroxidase (HRP) antibody (Novex; Life Technologies) and SuperSignal West Pico chemiluminescent substrate (Thermo Scientific). A His-tagged derivative of CsrA paralog Lpg2094 was the positive control.