The global regulatory Csr (carbon storage regulator) and the homologous Rsm (repressor of secondary metabolites) systems of Gram-negative bacteria typically consist of an RNA-binding protein (CsrA/RsmA) and at least one sRNA that functions as a CsrA antagonist. CsrA modulates gene expression post-transcriptionally by regulating translation initiation and/or mRNA stability of target transcripts. While Csr has been extensively studied in Gram-negative bacteria, until now Csr has not been characterized in any Gram-positive organism. csrA of Bacillus subtilis is the last gene of a flagellum biosynthetic operon. In addition to the previously identified σD-dependent promoter that controls expression of the entire operon, a σA-dependent promoter was identified that temporally controls expression of the last two genes of the operon (fliW-csrA); expression peaks 1 h after cell growth deviates from exponential phase. hag, the gene encoding flagellin, was identified as a CsrA-regulated gene. CsrA was found to repress hag′–′lacZ expression, while overexpression of csrA reduces cell motility. In vitro binding studies identified two CsrA binding sites in the hag leader transcript, one of which overlaps the hag Shine–Dalgarno sequence. Toeprint and cell-free translation studies demonstrate that bound CsrA prevents ribosome binding to the hag transcript, thereby inhibiting translation initiation and Hag synthesis.
The central component of the Csr (carbon storage regulator) and the homologous Rsm (repressor of secondary metabolites) systems of Gram-negative bacteria is an RNA-binding protein (CsrA or RsmA) that regulates gene expression post-transcriptionally by affecting ribosome binding and/or mRNA stability. The Csr system of Escherichia coli is involved in the repression of several stationary-phase processes and in the activation of some exponential-phase functions (reviewed in Romeo, 1998; Babitzke and Romeo, 2007). CsrA of E. coli represses gluconeogenesis, glycogen metabolism, peptide transport and biofilm formation (Romeo, 1998; Jackson et al., 2002; Dubey et al., 2003; Wang et al., 2005; Babitzke and Romeo, 2007), whereas CsrA activates glycolysis, acetate metabolism and flagellum biosynthesis (Romeo, 1998; Wei et al., 2000; 2001). The mechanism of CsrA-mediated repression in E. coli has been characterized in considerable detail. CsrA typically binds to multiple conserved sequences in target transcripts, one of which overlaps the cognate Shine–Dalgarno (S-D) sequence. Thus, bound CsrA inhibits translation initiation by blocking ribosome access to target mRNAs. Translation inhibition is thought to contribute to the observed rapid degradation of the transcripts (Romeo, 1998; Wang et al., 2005). While the mechanism of CsrA-mediated positive control has not been elucidated, it is known that bound CsrA can lead to mRNA stabilization (Wei et al., 2001). Two non-coding small RNAs (sRNAs), CsrB and CsrC, also participate in the Csr regulatory circuitry of E. coli. Both of these sRNAs contain multiple CsrA binding sites and function as antagonists of CsrA by sequestering this protein (Liu et al., 1997; Weilbacher et al., 2003).
In addition to E. coli, Csr systems have been characterized in several other Gram-negative bacterial species. In each case, at least one CsrA homologue and one sRNA function in an analogous manner as for E. coli. As is the case for E. coli, Csr appears to constitute a global regulatory system in many Gram-negative organisms. For example, the Csr system of Salmonella enterica regulates epithelial cell invasion, the production of flagella and certain pathways of carbon metabolism (Lawhon et al., 2003; Fortune et al., 2006). In the root-colonizing biocontrol species, Pseudomonas fluorescens, the Rsm system regulates expression of extracellular antifungal compounds that protect plant roots from fungal pathogens (Heeb et al., 2002; Kay et al., 2005; Reimmann et al., 2005). In Pseudomonas aeruginosa, Rsm regulates quorum sensing, exoproduct production, motility and biofilm formation (Heurlier et al., 2004; Kay et al., 2006). Csr also plays an important role in controlling quorum sensing in Vibrio cholerae (Lenz et al., 2005). Finally, in the plant pathogen Erwinia carotovora, the Rsm system regulates plant pathogenesis, exoproduct production, motility and quorum sensing (Chatterjee et al., 1995; 2002; Liu et al., 1998).
Despite the intensive study of Csr (Rsm) in Gram-negative organisms, investigation of the analogous system in Gram-positive bacteria has not been reported. While bioinformatic approaches identified genes that appear to be homologues of csrA and the sRNA-encoding genes in Gram-positive organisms (White et al., 1996; Kulkarni et al., 2006), the functions of these genes have not been verified. We cloned and characterized csrA from Bacillus subtilis as a first step in examining the function of Csr in this model Gram-positive organism. Our results indicate that expression of csrA is temporally controlled by a newly identified σA-dependent promoter. Moreover, we found that CsrA represses translation initiation of hag, the gene encoding the flagellin protein of B. subtilis.
Identification of a temporally regulated σA-dependent fliW-csrA operon
The predicted amino acid sequence of B. subtilis CsrA is 42% identical to the E. coli protein (White et al., 1996). csrA of B. subtilis is the last gene of an operon that contains several genes involved in flagella biosynthesis. σD promoters control transcription initiation of this operon, and the single-gene hag operon just downstream (Fig. 1A). Expression of B. subtilis csrA was examined as a first step in characterizing the Csr system of this organism.
Expression of csrA was examined by monitoring β-galactosidase expression from a csrA′–′lacZ translational fusion that had been integrated into the amyE locus of the B. subtilis chromosome (Table 1 and Fig. 1A, PLBS178). The csrA′–′lacZ translational fusion was constructed such that it contained 1158 bp of DNA upstream from the csrA coding sequence, including the yviE gene of unknown function and fliW (yviF), a gene that may encode a flagellum assembly factor (Titz et al., 2006). Although this fusion did not include any known promoters, expression was observed throughout the B. subtilis growth cycle (Fig. 1B). Expression of this fusion was low during exponential-phase growth but exhibited a rapid fivefold increase in expression as the culture entered the transition state and peaked at T1. T0 is defined as the point in which the growth rate deviates from that observed during exponential growth, while T1 is 1 h later (Fig. 1B). These results suggested that a promoter within the 1158 bp region upstream of csrA was responsible for the observed temporal expression of csrA. To demonstrate that a promoter within the cloned region was responsible for the observed expression, all but 30 nt upstream from the csrA translation initiation codon was deleted from the csrA′–′lacZ translational fusion. In this case, β-galactosidase expression was not observed, indicating that a previously unidentified promoter was located within the deleted region (Fig. 1, PLBS226).
pheA1 trpC2 csrA::tet (Tcr) pMK3 (Kmr) amyE::Phag(−182 to +316)hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) pCSB7 (Kmr) amyE::Phag(−182 to +316)hag′–′lacZ Cmr
rpoB18 (Rifr) csrA::tet (Tcr) pMK3 (Kmr) amyE::Phag(−182 to +316)hag′–′lacZ Cmr
rpoB18 (Rifr) csrA::tet (Tcr) pCSB7 (Kmr) amyE::Phag(−182 to +316)hag′–′lacZ Cmr
rpoB18 (Rifr) csrA::tet pMK3 (Kmr)
rpoB18 (Rifr) csrA::tet pCSB7 (Kmr)
amyE::(−3 to +574)fliW-csrA′–′lacZ Cmr
amyE:: PfliW (−53 to +574)fliW-csrA′–′lacZ Cmr
pheA1 trpC2 amyE::Phag(−128 to +166)hag′–′lacZ Cmr
pheA1 trpC2 amyE::Phag[−128 to +166(Δ+53 to +55)]hag′–′lacZ Cmr
pheA1 trpC2 amyE::Phag[−128 to +166(A72C:T73A:C75T)]hag′–′lacZ Cmr
pheA1 trpC2 amyE::Phag[−128 to +166(Δ+53 to +55, A72C:T73A:C75T)]hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) amyE::Phag(−128 to +166)hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) amyE::Phag[−128 to +166(Δ+53 to +55)]hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) amyE::Phag[−128 to +166(A72C:T73A:C75T)]hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) amyE::Phag[−128 to +166(Δ+53 to +55, A72C:T73A:C75T)]hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) pMK3 (Kmr) amyE::Phag(−128 to +166)hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) pCSB7 (Kmr) amyE::Phag(−128 to +166)hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) pMK3 (Kmr) amyE::Phag[−128 to +166(Δ+53 to +55)]hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) pCSB7 (Kmr) amyE::Phag[−128 to +166(Δ+53 to +55)]hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) pMK3 (Kmr) amyE::Phag[−128 to +166(A72C:T73A:C75T)]hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) pCSB7 (Kmr) amyE::Phag[−128 to +166(A72C:T73A:C75T)]hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) pMK3 (Kmr) amyE::Phag[−128 to +166(Δ+53 to +55, A72C:T73A:C75T)]hag′–′lacZ Cmr
pheA1 trpC2 csrA::tet (Tcr) pCSB7 (Kmr) amyE::Phag[−128 to +166(Δ+53 to +55, A72C:T73A:C75T)]hag′–′lacZ Cmr
Primer extension experiments were carried out using total cellular RNA to map the 5′ end of transcripts originating from the newly identified promoter. Two closely spaced primer extension products were observed just upstream from the fliW coding sequence (Fig. 2A). A likely σA-dependent promoter (PfliW) was identified just upstream that included an extended −10 sequence. In vitro transcription reactions using purified B. subtilis RNA polymerase confirmed that this was a σA-dependent promoter (Fig. 2B).
Two additional csrA′–′lacZ translational fusions were constructed and integrated into the amyE locus to further characterize expression from PfliW. These two fusions contained either 53 or 3 nt upstream from the transcription start site (Fig. 1A). Both the temporal expression pattern and expression level from the fusion beginning at −53 (PLBS520) was similar to the fusion beginning at −968 (PLBS178), whereas expression was not observed from the fusion beginning at −3 (PLBS519) (Fig. 1B). These results demonstrate that fliW and csrA constitute a two-gene σA-dependent operon that is embedded within the 3′ end of the larger flagellar operon. Thus, both fliW and csrA are transcribed by RNA polymerase containing σA (EσA) and σD (EσD); however, the relative contribution of these two promoters to expression of these genes has not been explored.
CsrA regulates expression of hag
As an initial attempt to identify CsrA-regulated genes, a comparative analysis of protein synthesis in wild-type (W168) and CsrA-deficient (csrA::tet, PLBS177) strains was performed. Protein synthesis was also examined in a csrA::tet strain that contained either a complementing plasmid [pCSB7 (csrA+), PLBS383] or an empty vector (pMK3, PLBS382). Crude cell extracts were prepared from cultures harvested at various stages of the growth cycle, and analysed by one-dimensional SDS-PAGE. One abundant protein of ∼45 kDa exhibited enhanced expression in csrA mutant strains during late exponential- and early stationary-phase growth (Fig. 3A and data not shown). The level of this protein was reduced in the csrA-overexpressing strain, suggesting that CsrA negatively regulated synthesis of this protein. The N-terminal amino acid sequence of this protein was determined to be MRINHNIAAL. A search of the protein sequence database revealed a 100% identity with the N-terminal sequence of Hag, the flagellin protein of B. subtilis. Interestingly, hag is a σD-controlled gene that is located immediately downstream from csrA in the B. subtilis chromosome (Fig. 1A).
To determine whether csrA influenced cell motility, wild-type (W168) and csrA mutant (PLBS177) strains were inoculated onto motility plates. The diameter of the wild-type halo was 32 mm, while the diameter of the csrA mutant halo was 24 mm, suggesting that csrA positively affects motility despite having an inhibitory effect on hag expression (data not shown). Motility assays were also carried out with wild-type (1A96) and csrA mutant strains that either contained pCSB7 (csrA+) or an empty vector (pMK3). As before, the diameter of the halo from the wild-type strain (WT/pMK3, PLBS188) was approximately 30% larger than that of the csrA mutant (csrA::tet/pMK3, PLBS191) (Fig. 3B). However, expression of csrA from pCSB7 caused a dramatic reduction in motility in wild-type (PLBS189) and csrA mutant (PLBS192) backgrounds, indicating that overexpression of csrA represses motility (Fig. 3B). The latter results are consistent with the one-dimensional SDS-PAGE analysis, which suggested that CsrA represses Hag synthesis (Fig. 3A).
To further examine CsrA-dependent regulation of hag expression, a hag′–′lacZ translational fusion was integrated into the amyE locus of wild-type, csrA::tet, csrA::tet/pCSB7 (csrA+) and csrA::tet/pMK3 (empty vector) strains in both W168 and 1A96 genetic backgrounds (Table 1). Expression of the hag′–′lacZ fusion was examined throughout the growth cycle in each strain when grown in LB. In the W168 genetic background, hag′–′lacZ expression was slightly higher in the csrA mutant compared with the wild-type strain, while overexpression of csrA from pCSB7 resulted in an approximate fivefold reduction in expression of the fusion (Fig. 4A). CsrA-dependent repression of hag′–′lacZ expression was more substantial in the 1A96 genetic background. In this case, expression was about twofold higher in the csrA mutant compared with the wild-type strain, while overexpression of csrA resulted in approximately eightfold lower expression (Fig. 4B). CsrA-mediated repression was even more pronounced when the 1A96 strains were grown in minimal-0.2% acid casein hydrolysate (ACH) media; expression was about threefold higher in the csrA mutant (Fig. 4C), while overexpression of csrA resulted in > 10-fold lower hag′–′lacZ expression (Fig. 4D). These results are consistent with the one-dimensional protein studies and indicate that CsrA inhibits hag expression.
CsrA binds to two sites in the hag leader transcript
CsrA was purified with a histidine tag at its C-terminus (CsrA-H6). Previous glutaraldehyde cross-linking studies established that E. coli CsrA is a homodimer in solution (Dubey et al., 2003). The NMR structure of the E. coli protein confirmed the subunit organization of CsrA (Gutiérrez et al., 2005). Glutaraldehyde cross-linking studies of CsrA-H6 (referred to as CsrA from here on) demonstrated that B. subtilis CsrA is also a homodimer in solution (data not shown).
The hag transcript contains a 91 nt untranslated leader. To characterize the interaction of CsrA with hag RNA, quantitative gel mobility shift assays were performed with a transcript containing +1 to +100 relative to the start of hag transcription. CsrA binding to this transcript was detected as a distinct band in native gels at ∼100 nM CsrA (Fig. 5A). Non-linear least-squares analysis of these data yielded an estimated Kd value of 115 ± 10 nM with a cooperativity coefficient (n) of 3.1. The positive cooperativity suggested that binding of CsrA to one binding site stimulated CsrA interaction with an additional binding site(s). As the concentration of CsrA was increased further, additional shifted species of higher molecular weights were observed. This gel shift pattern suggested that multiple CsrA dimers were bound to each hag transcript at these higher CsrA concentrations. Although the stoichiometry of these complexes has not been examined, we presume that the first shifted species contained one hag transcript and one CsrA dimer, and that the additional shifted species contained multiple CsrA dimers. The specificity of CsrA–hag RNA interaction was investigated by performing competition experiments with specific (hag) and non-specific (trpL) unlabelled RNA competitors (Fig. 5B). Unlabelled hag RNA was an effective competitor, whereas B. subtilis trp leader RNA was not. These results establish that CsrA binds specifically to hag RNA.
CsrA–hag RNA footprint experiments were carried out to identify the CsrA binding sites in the hag transcript. Three single strand-specific RNases were used as probes for these studies. As the concentration of CsrA was increased from 0 to 500 nM, protection of several nucleotides from RNase T1 (G specific), RNase T2 (A preference) and/or RNase A (C and U specific) cleavage was observed (Fig. 6). CsrA protected G53, G54, G77, G78, G79, G81 and G82 from RNase T1 cleavage. In the case of RNase T2, CsrA protected residues A51 through A55, C75, A76, G78 and G79 from cleavage. CsrA also protected residues C69, C75 and G79 from RNase A cleavage, although RNase A is reported to be specific for pyrimidines. The composite footprint indicates that CsrA protected two RNA segments, A51 to A55 (binding site 1, BS1) and C75 to G82 (binding site 2, BS2) (Fig. 6). Importantly, BS2 overlaps the hag S-D sequence, suggesting that bound CsrA could be capable of inhibiting translation initiation of hag.
The relative affinity of CsrA for each binding site in the hag leader transcript was compared using transcripts that contained either BS1 or BS2. For each of the transcripts containing single binding sites, a gradual transition from free to bound RNA was observed (Fig. 5C and D). Non-linear least-squares analysis of these data yielded estimated Kd values of 1.2 μM and 2.6 μM for BS1 and BS2 respectively. The Kd values for the single binding sites were at least an order of magnitude higher than the Kd value determined for a transcript containing both binding sites, suggesting that the high degree of cooperativity observed with the full-length hag leader transcript containing both binding sites is critical for high-affinity CsrA–hag RNA interaction (Fig. 5A).
To determine the influence of the two CsrA binding sites on CsrA-dependent regulation of hag expression, hag′–′lacZ translational fusions were engineered such that they contained a wild-type hag leader, or leaders containing mutations in BS1, BS2 or both. The BS1 mutation consisted of a GGA motif deletion (Δ+53 to +55). A more subtle change was required for BS2 as the GGA motif is a component of the hag S-D sequence. In this case we altered three residues upstream of the hag S-D sequence but within BS2 (A72C:U73A:C75U). Both of these mutant binding sites, alone or in combination, essentially eliminated CsrA-dependent regulation of hag expression; expression of the three mutant fusions in a csrA+ background was comparable to expression of the wild-type fusion in a csrA mutant background (Fig. 4C). Furthermore, introduction of the csrA::tet allele into strains containing the mutant fusions did not cause a further increase in expression (data not shown). However, overexpression of csrA from a plasmid suppressed the regulatory defects associated with the binding site mutations (Fig. 4D), suggesting that CsrA had a lower affinity for mRNA containing the mutations in BS1 and/or BS2 and that overproduction of CsrA compensated for the reduced affinity.
CsrA regulates translation initiation of hag by blocking ribosome binding
The position of the CsrA binding sites in the hag transcript suggested a model in which bound CsrA could block ribosome binding. CsrA and 30S ribosomal subunit toeprint experiments were performed to test this prediction. The presence of bound CsrA or a 30S ribosomal subunit should block primer extension by reverse transcriptase, typically resulting in a toeprint band just downstream of the bound macromolecule. The toeprint results are presented in Fig. 7 and summarized in Fig. 6B. Prominent CsrA toeprints were observed at positions G59 and A84, corresponding to positions just downstream from the previously identified CsrA footprints.
Similar toeprint experiments were performed to identify the position of bound 30S ribosomal subunits. A prominent tRNAfMet-dependent toeprint band was observed 15 nt downstream from the A of the AUG initiation codon (Fig. 7), which is the same distance from the translation initiation codon that was previously observed for several mRNAs (e.g. Baker et al., 2002; Dubey et al., 2003; Yakhnin et al., 2004; 2006a). Toeprint experiments were also carried out to determine if CsrA could inhibit ribosome binding. When CsrA was bound to the hag transcript prior to the addition of 30S ribosomes and tRNAfMet, both of the CsrA-dependent toeprint bands were observed, while the ribosome toeprint was considerably reduced or absent (Fig. 7). These results demonstrate that bound CsrA inhibits ribosome binding to the hag transcript.
As the toeprint results indicated that bound CsrA competes with ribosomes for binding to the hag transcript, RNA-directed cell-free translation experiments using a CsrA-deficient B. subtilis S-30 extract were performed to determine whether CsrA could inhibit synthesis of a Hag–GFP fusion peptide from a pre-existing mRNA template. The GFP fusion was used because we previously found that GFP fusion proteins are stable in B. subtilis S-30 extracts, thereby simplifying the analysis of cell-free translation experiments (Schaak et al., 2003). A major protein species that was dependent on the addition of the hag′–′gfp transcript was produced (Fig. 8). No translation product corresponding to the fusion peptide was observed without the addition of the hag′–′gfp transcript. Addition of increasing concentrations of purified CsrA to the translation system resulted in a corresponding decrease in the level of Hag–GFP synthesis. Similar cell-free translation experiments were carried out with a ycbK′–′gfp fusion transcript as a control. While translation of this fusion transcript was inhibited by TRAP (Yakhnin et al., 2006a), CsrA did not inhibit YcbK–GFP synthesis (Fig. 8). In conjunction with the footprint and toeprint results described above, the cell-free translation experiments demonstrate that CsrA binding to the hag message inhibits Hag synthesis by competitively blocking ribosome binding.
Csr (Rsm)-mediated global regulation has been under intensive study in Gram-negative organisms; however, until now this global regulatory system had not been investigated in any Gram-positive organism. csrA of B. subtilis is located in a chromosomal region that contains several genes involved in flagella biosynthesis that are under control of EσD (Fig. 1A). flgK and flgL encode flagellar hook-associated proteins (Mirel et al., 1994; http://genolist.pasteur.fr/SubtiList/), whereas flgM encodes an anti-σD factor (reviewed in Hughes and Mathee, 1998; Helmann, 1999). FlgM inhibits σD activity until completion of the hook and basal body (HBB) of the flagellum. Perhaps FlgM of B. subtilis is exported through the completed HBB, as is the case for E. coli and Salmonella (Macnab, 1996). In any event, once HBB formation is completed in B. subtilis, FlgM-mediated inhibition of σD activity is abrogated such that σD is available for binding to core RNA polymerase. Because transcription of flgM is dependent on EσD, this mechanism constitutes an autoregulatory loop for flgM expression. Furthermore, in the absence of FlgM, σD-dependent transcription of hag is activated. As csrA is the last gene in the flg operon, it is assumed that EσD is capable of transcribing this gene as well.
During the course of these studies we found that fliW-csrA constitutes a temporally controlled two-gene EσA-dependent operon (Figs 1 and 2). It was recently shown that FliW interacts with the C-terminus of Hag and that deletion of fliW leads to reduced Hag accumulation and a substantial reduction in motility. Thus, it was proposed that FliW functions as a flagellum assembly factor (Titz et al., 2006). The finding that expression of csrA from PfliW is growth phase dependent and highest at T1 provides circumstantial evidence that expression of csrA might be under control of one or more transition state regulators. It is also possible that CsrA participates in regulating the transition state in B. subtilis, a period of time in which critical decisions are made in determining the developmental fate of the organism (reviewed in Phillips and Strauch, 2002).
In E. coli and related species, motility is regulated by several global regulatory circuits, which converge to modulate the expression of the master operon of flagellum biosynthesis, flhDC (Wei et al., 2001). In the case of E. coli, motility and chemotaxis genes have been grouped into three classes that constitute a hierarchy of gene expression (reviewed in Macnab, 1996). flhDC constitutes the class 1 genes, which encodes a DNA-binding protein that is required for activation of the class 2 genes. The class 2 genes encode components of the HBB, as well as fliA, a gene that encodes an alternative sigma factor (σF). σF is responsible for activating transcription of the class 3 genes, which encode the anti-σF factor (FlgM), as well as the flagellin, motor and chemotaxis proteins. CsrA-mediated control of flagella biosynthesis and/or motility has been observed in several Gram-negative organisms (e.g. Wei et al., 2001; Lawhon et al., 2003; Heurlier et al., 2004). In E. coli, CsrA post-transcriptionally regulates flhDC expression by binding to the flhDC operon leader transcript and stabilizing the mRNA, although the precise mechanism of message stabilization has not been elucidated (Wei et al., 2001; C.S. Baker, T. Romeo and P. Babitzke, unpublished results).
The hierarchy of gene expression differs in B. subtilis in that class 1 genes analogous to flhDC have not been identified and that the majority of the fla/che genes and sigD are transcribed by EσA (reviewed in Aizawa et al., 2002). σD, in turn, is responsible for activating expression of the class 3 genes, which includes the flg operon, hag, and genes encoding motor and chemotaxis proteins. Flagellin is one of the most abundant proteins in motile B. subtilis (Fredrick et al., 1995). As hag expression is absolutely dependent on σD (Helmann et al., 1988; Mirel and Chamberlin, 1989), it is unlikely that transcripts originating from PfliW contribute to substantial hag expression as a consequence of transcriptional readthrough past the csrA terminator. While hag expression requires EσD, interaction of the α-subunits of RNA polymerase with an A+T-rich upstream promoter (UP) element leads to a 20-fold increase in hag expression (Fredrick et al., 1995). In addition, CodY represses expression of hag in nutrient-rich conditions by binding to its promoter region (Bergara et al., 2003). Interestingly, the expression pattern of csrA (Fig. 1B) and hag (Fig. 4) reflects the extent of motility throughout the growth cycle in B. subtilis; motility increases during the transition from exponential growth into stationary phase (transition state) and then declines later in stationary phase (Nishihara and Freese, 1975). Thus, it appears that CsrA functions as a governor to prevent excess Hag synthesis, rather than as an on-off switch. The finding that CsrA inhibits translation initiation of hag demonstrates that CsrA participates in controlling flagella biosynthesis in B. subtilis. Given that CsrA inhibits Hag synthesis, it is somewhat surprising that motility is slightly reduced in a csrA mutant strain (Fig. 3B). Perhaps inappropriate flagellin production interferes with flagellar assembly and/or function. Alternatively, CsrA might regulate other genes involved in flagella synthesis, although these possibilities are not mutually exclusive. While the reason for the small motility defect in csrA mutant strains is not known, it is clear that overexpression of csrA leads to a dramatic reduction in motility (Fig. 3B), consistent with our findings that CsrA represses hag′–′lacZ expression (Fig. 4) and Hag synthesis (Figs 3A and 8).
Our footprinting studies identified two CsrA binding sites in the hag leader transcript (Fig. 6). The most highly conserved feature of CsrA binding sites in several Gram-negative organims is a GGA motif, which is often located in the loop of short hairpins (Babitzke and Romeo, 2007). A SELEX (systematic evolution of ligands by exponential enrichment) analysis with E. coli CsrA identified a binding site consensus of RUACARGGAUGU with the underlined residues being 100% conserved. In each case, the GGA motif was located in the loop of a short predicted hairpin (Dubey et al., 2005). Both of the CsrA binding sites in hag contain a GGA motif. In each case, the positions of the CsrA toeprints were 4 nt downstream from the GGA sequence (Figs 6 and 7). Using the SELEX-derived E. coli binding site consensus as a guide, we can predict that BS1 and BS2 are GCACAAGGACGU and AUUCAGGGAGGA respectively. In addition to the striking similarity to the E. coli consensus sequence, the GGA motifs of both hag binding sites are present within the loops of short predicted RNA hairpins (not shown). Thus, it is surprising that the affinity of B. subtilis CsrA for individual hag binding sites was only 1–3 μM (Fig. 5C and D). Perhaps the high degree of cooperativity observed for the hag transcript containing both CsrA binding sites is crucial for high-affinity interaction (Fig. 4A). As CsrA appears to have two RNA-binding surfaces per dimer (Mercante et al., 2006), it is possible that CsrA of B. subtilis exhibits increased avidity for RNA that contains at least two binding sites.
The general mechanism of CsrA-mediated translation inhibition has been elucidated in considerable detail for E. coli. glgC, cstA and pgaA correspond to the first genes of operons involved in glycogen synthesis, peptide transport and synthesis of a biofilm adhesin respectively (Baker et al., 2002; Dubey et al., 2003; Wang et al., 2005). Each of these mRNAs contains four to six CsrA binding sites. Biochemical characterization of CsrA interaction with these mRNAs indicates that one of the CsrA binding sites overlaps each cognate S-D sequence such that bound CsrA prevents ribosome binding. In the case of B. subtilis hag, CsrA binds to two sites, one of which overlaps the hag S-D sequence. Furthermore, the toeprint and cell-free translation data indicate that CsrA inhibits Hag synthesis by blocking ribosome binding to the hag transcript (Figs 7 and 8). Thus, it is apparent that the mechanism of CsrA-mediated repression of hag translation in B. subtilis is similar to the mechanism in E. coli. Despite the striking similarity of the regulatory mechanisms in these two organisms, B. subtilis csrA did not complement a csrA null mutant of E. coli (data not shown).
In addition to a CsrA homologue, at least one sRNA containing multiple CsrA binding sites participates in the Csr regulatory circuitry by sequestering the cognate RNA-binding protein. For example, E. coli has two sRNAs (CsrB and CsrC), while P. fluorescens (RsmX, RsmY and RsmZ) and E. carotovora (RsmB) contain three and one sRNA respectively (reviewed in Babitzke and Romeo, 2007). A recent bioinformatic approach identified one likely sRNA gene in B. subtilis (CsrB), although this has not been experimentally verified (Kulkarni et al., 2006).
The Csr (Rsm) system has been shown to be a global regulator of several biochemical pathways and cellular processes in a number of Gram-negative bacterial species. A common process controlled by this system is flagella biosynthesis/motility. Thus far the only Csr-regulated gene that has been identified in Gram-positive bacteria is hag of B. subtilis. It is intriguing that this gene encodes the major flagellin protein of this organism. Whether Csr constitutes a global regulatory system in B. subtilis and other Gram-positive organisms remains to be determined.
Bacterial strain and plasmids
All B. subtilis strains used in this study are listed in Table 1. The cloning vectors pTZ18U and pTZ19R contain a T7 RNA polymerase promoter upstream from their polylinker (U. S. Biochemical). The B. subtilis integration vector ptrpBGIPLK was described previously (Merino et al., 1995). Plasmid pMG601 contains the chromosomal region flanking the σD-dependent hag promoter (Mirel and Chamberlin, 1989). Plasmid pMGM1 contains the 2.2 kb EcoRI fragment from pMG601 subcloned into the pTZ18U polylinker. In addition to the σD promoter and the N-terminal coding sequence of hag, this chromosomal fragment contains the complete coding sequences for yviE, fliW and csrA. Plasmid pCSB3 contains the 1.6 kb EcoRI–HindIII fragment from pMGM1 subcloned into the ptrpBGIPLK polylinker, thereby generating a fliW-csrA′–′lacZ translational fusion (−968 to +574 relative to fliW transcription). pCSB3 also contains the σA-dependent PfliW promoter and the entire yviE coding sequence upstream of fliW. Plasmid pTP2 contains a PCR fragment (+454 to +574 relative to fliW transcription) cloned into ptrpBGIPLK, thereby producing a fliW-csrA′–′lacZ translational fusion without PfliW. Both pCSB3 and pTP2 were linearized with ScaI and integrated into the chromosomal amyE locus of B. subtilis strain W168, resulting in strains PLBS178 and PLBS226 respectively. Plasmid pYH103 contains a PCR fragment (−53 to +574 relative to fliW transcription) cloned into the EcoRI–HindIII sites of pTZ18U, while plasmid pYH104 contains a PCR fragment (−3 to +574 relative to fliW transcription) cloned into the EcoRI–HindIII sites of pTZ19R. The PCR-derived fragments from pYH103 and pYH104 were subcloned into the ptrpBGIPLK polylinker, resulting in fliW-csrA′–′lacZ translational fusions with (pYH107) or without (pYH106) PfliW. Plasmids pYH106 and pYH107 were linearized with ScaI and integrated into the chromosomal amyE locus of B. subtilis strain PLBS338 (Yakhnin et al., 2004), resulting in strains PLBS519 and PLBS520 respectively.
The csrA knockout allele (csrA::tet) was constructed by subcloning the tet gene from pHY300PLK into the XmnI site of the csrA gene in pMGM1, resulting in plasmid pCSB1. Chromosomal gene replacement of WT csrA with the csrA::tet allele was carried out by linearizing pCSB1 with KpnI and subsequently transforming W168 and 1A96 with the linearized plasmid, resulting in PLBS177 and PLBS183 respectively. Selection was for tetracycline resistance. Proper allelic replacement was confirmed by Southern blotting.
Plasmid pCSB4 contains a hag′–′lacZ translational fusion driven by the endogenous σD-dependent hag promoter (−182 to +316 relative to hag transcription) in ptrpBGIPLK. This construct contains the 75th hag codon fused in frame with the 9th codon of lacZ. pCSB4 was linearized with PstI and subsequently integrated into the amyE locus of W168, PLBS177, 1A96 and PLBS183 resulting in strains PLBS179, PLBS180, PLBS184 and PLBS185 respectively. Plasmid pYH119 contains a hag′–′lacZ translational fusion driven by the endogenous σD-dependent hag promoter (−128 to +166 relative to hag transcription) in ptrpBGIPLK. This construct contains the 25th hag codon fused in frame with the 9th codon of lacZ. Plasmids pYH120, pYH121 and pYH122 are mutant derivatives of pYH119; each mutation was generated using the QuikChange II protocol (Stratagene). The GGA residues (+53 to +55 relative to hag transcription) of CsrA BS1 were deleted to generate pYH120. Plasmid pYH121 contains three nucleotide substitutions in CsrA BS2 (A72C:T73A:C75T). The mutations in pYH120 and pYH121 were combined to generate pYH122. Plasmids pYH119, pYH120, pYH121 and pYH122 were linearized with PstI and integrated into the amyE locus of strain 1A96, resulting in strains PLBS552, PLBS553, PLBS554 and PLBS555 respectively. Similarly, plasmids pYH119, pYH120, pYH121 and pYH122 were linearized with PstI and integrated into the amyE locus of strain PLBS183, resulting in strains PLBS560, PLBS561, PLBS562 and PLBS563 respectively. Plasmid pMK3 is an E. coli–B. subtilis shuttle vector (Sullivan et al., 1984). pMK3 was used to transform 1A96, PLBS183, PLBS177, PLBS180, PLBS185, PLBS560, PLBS561, PLBS562 and PLBS563, resulting in strains PLBS188, PLBS191, PLBS382, PLBS380, PLBS378, PLBS564, PLBS567, PLBS569 and PLBS571 respectively. Plasmid pCSB7 was constructed by subcloning the 2.2 kb EcoRI fragment containing csrA from pMGM1 into pMK3. pCSB7 was used to transform 1A96, PLBS183, PLBS177, PLBS180, PLBS185, PLBS560, PLBS561, PLBS562 and PLBS563, resulting in strains PLBS189, PLBS192, PLBS383, PLBS381, PLBS379, PLBS565, PLBS568, PLBS570 and PLBS572 respectively.
Plasmid pCSB9 contains csrA cloned into the NdeI–XhoI sites of pET21a+ (Inovagen). This plasmid was designed to allow purification of CsrA containing six additional His residues at the C-terminus (CsrA-H6). Plasmid pYH42, containing the ycbK′–′gfp translational fusion under control of a T7 promoter, was described previously (Yakhnin et al., 2006a). The plasmid pYH108 is identical to pYH42 except that the ycbK-specific sequences were replaced with hag DNA (+7 to +124 relative to hag transcription), thereby generating a hag′–′gfp fusion (11th hag codon fused in frame with gfp) under control of the same T7 promoter. These two plasmids were used to generate run-off transcripts for cell-free translation studies. Plasmid pYH110 contains nt +7 to +181 relative to hag transcription cloned into the EcoRI–BamHI sites of pTZ18U. This plasmid was used as a template for sequencing reactions in the toeprint analysis.
Bacillus subtilis cultures containing fliW-csrA′–′lacZ translational fusions were grown at 37°C in minimal-ACH medium. B. subtilis cultures containing hag′–′lacZ translational fusions were grown at 37°C in LB broth or in minimal-ACH medium. When appropriate growth media also contained 5 μg ml−1 chloramphenicol, 10 μg ml−1 kanamycin, 12.5 μg ml−1 tetracycline and/or 200 μM l-tryptophan. Cells were grown for 8–10 h and β-galactosidase expression was monitored throughout the growth cycle as described previously (Du and Babitzke 1998; Baker et al., 2002).
SDS-PAGE analysis of total cellular proteins
Bacillus subtilis W168 (wild type), PLBS177 (csrA::tet), PLBS382 (csrA::tet pMK3 vector) and PLBS383 [csrA::tet pCSB7 (csrA+)] were grown in minimal-ACH medium at 37°C. Five millilitres of culture samples were harvested every 30 min during exponential growth and during the transition from exponential- to stationary-phase growth. Cell pellets were washed with 10 mM Tris-HCl (pH 7.5). Cell extracts were prepared by re-suspending pellets in 200 μl of Z buffer (Platt et al., 1972) containing 3 μg ml−1 lysozyme followed by incubation for 1 h at 37°C with shaking. Incubation with shaking was continued for 45 min at 37°C following the addition of 200 μl of Bug Buster protein extraction reagent (Novagen). Following centrifugation, the protein concentration of the crude cell extracts were estimated using the Bio-Rad assay with BSA as a standard. Aliquots containing 20 μg of protein were fractionated by 10% SDS-PAGE and proteins were visualized by staining with Coomassie blue.
Semisolid tryptone medium (1% tryptone, 0.5% NaCl and 0.35% agar) was inoculated from single colonies and incubated for 16 h at 30°C with 100% humidity (Wei et al., 2001). Growth media for strains containing pMK3 or pCSB7 also contained 10 μg ml−1 kanamycin.
Primer extension reaction
Total RNA was isolated from a late exponential-phase culture of B. subtilis W168 grown in minimal-ACH medium using the RNeasy kit (Qiagen). Twenty micrograms of total RNA was hybridized to 150 nM 32P-end-labelled DNA oligonucleotide complementary to nt +111 to +128 (relative to fliW transcription) for 3 min at 90°C. Reaction mixtures (10 μl) containing 4 μl of hybridization mixture, 375 μM of each dNTP, 10 mM DTT, 1× MMLV reaction buffer and 20 U μl−1 MMLV reverse transcriptase (USB) was incubated for 10 min at 42°C. Reactions were terminated by the addition of 6 μl of stop solution (95% formamide, 20 mM EDTA, 0.025% SDS, 0.025% xylene cyanol, 0.025% bromophenol blue). Samples were fractionated through standard 6% polyacrylamide sequencing gels. Sequencing reactions were performed using pYH103 as template and the same end-labelled DNA oligonucleotide as a primer. Radiolabelled bands were visualized by phosphorimagery.
In vitro transcription
Single-round in vitro transcription reactions were performed as described previously (Yakhnin and Babitzke, 2002). B. subtilis EσA was purified as described (Yakhnin et al., 2006b). Stable transcription elongation complexes were formed in a reaction containing 10 μM ATP, 10 μM UTP, 2 μM GTP and 2 μCi [α-32P]-UTP and DNA templates. Transcription elongation was halted following incorporation of nt 7 due to the absence of CTP. Elongation of halted transcription complexes was resumed by the addition of all four NTPs together with heparin. The final concentrations were 150 μM of each NTP and 100 μg ml−1 heparin. Elongation reactions were stopped after 10 min by the addition of an equal volume of the same stop solution used in the primer extension analysis. Samples were fractionated through 6% polyacrylamide sequencing gels. DNA templates used in this analysis included a 187 bp PCR fragment with PfliW (−53 to +134 relative to fliW transcription) and a 137 bp PCR fragment without PfliW (−3 to +134 relative to fliW transcription). A DNA template giving rise to 120 nt transcript was used as a size marker.
CsrA-H6 protein purification
BL21 E. coli cells containing pCSB9 were grown at 37°C in LB broth and 100 μg ml−1 ampicillin to an OD600 of 0.6, at which time 1 mM IPTG was added to the culture and incubation was continued for 5 h. Cells were harvested by centrifugation and the cell pellet (10 g) was suspended in lysis buffer containing 100 mM Tris-HCl (pH 8.0) and 400 mM NaCl (5 ml of buffer per gram of cells). Cell lysate was prepared using a French press, followed by centrifugation at 20 000 g for 30 min at 4°C. The resulting supernatant was diluted with an equal volume of water and loaded onto a 50 ml DEAE 52 (Whatman) column that had been pre-equilibrated with 50 mM Tris-HCl (pH 8.0) and 200 mM NaCl. The flow-through from the DEAE column was mixed with 2 ml of pre-washed Ni-NTA resin (Qiagen) for 1 h at 4°C and packed into a column. The column was washed successively with buffer A (50 mM Tris-HCl, pH 8.0, 200 mM NaCl, 10% glycerol), buffer A containing 5 mM imidazole and buffer A containing 15 mM imidazole. CsrA-H6 was eluted with buffer A containing 500 mM imadazole. Column fractions were analysed by 15% SDS-PAGE and Coomassie blue staining. Fractions containing pure CsrA-H6 were combined and dialysed against 20 mM Tris-HCl (pH 8.0), 200 mM NaCl, 10% glycerol and 1 mM EDTA. The concentration of CsrA-H6 was estimated using the Bio-Rad protein assay with BSA as a standard.
Gel mobility shift assay
Quantitative gel mobility shift assays used to examine CsrA–hag RNA interactions followed a published procedure (Dubey et al., 2003). Three different hag RNAs were synthesized in vitro with the Ambion MEGAscript kit and PCR-derived fragments as templates. Each template contained a T7 promoter and hag-specific sequences (+1 to +100, +1 to +69 or +59 to +151 relative to hag transcription). Gel-purified RNAs were dephosphorylated with calf intestinal alkaline phosphotase and 5′ end-labelled with [γ-32P]-ATP. Binding reactions (10 μl) contained 40 mM Tris-HCl (pH 7.5), 4 mM MgCl2, 100 mM KCl, 32.5 ng of yeast RNA, 20 mM DTT, 10% glycerol, 4 U of RNase inhibitor (Promega), 0.25 nM 5′ end-labelled hag RNA, various concentrations of purified CsrA-H6 and 0.1 mg ml−1 xylene cyanol. Competition assay mixtures also contained unlabelled hag or trpL RNA. Reaction mixtures were incubated at 37°C for 30 min and subsequently fractionated on native 15% polyacrylamide gels. Radiolabelled bands were visualized, quantified, and Kd values were calculated as described (Yakhnin et al., 2000).
RNA footprint assay
RNA footprint assays were performed by modifying published procedures (Dubey et al., 2003; Yakhnin et al., 2006a). RNA was synthesized in vitro with the MEGAscript kit using a PCR-generated DNA template containing a T7 promoter and hag-specific DNA (+1 to +151 relative to hag transcription). Gel-purified RNA was labelled as described for the gel mobility shift assay. Binding reactions (10 μl) contained 40 mM Tris-HCl (pH 7.5), 4 mM MgCl2, 100 mM KCl, 32.5 ng of yeast RNA, 5 mM DTT, 10% glycerol, 100 μg ml−1 BSA, 2 nM 5′ end-labelled hag RNA, various concentrations of purified CsrA-H6, 2 mM DTT and 10% glycerol. CsrA–hag RNA complexes were allowed to equilibrate at 37°C for 30 min prior to the addition of RNase T1 (5 × 10−3 U μl−1), RNase T2 (0.04 U μl−1) or RNase A (5 × 10−8 U μl−1). Incubation was then continued for 15 min at 37°C. Reactions were terminated by adding 5 μl of the same stop solution used in the primer extension assay. Samples were fractionated through 6% sequencing gels. Radiolabelled bands were visualized by phosphorimagery.
Toeprint assays were performed by modifying published procedures (Hartz et al., 1988; Yakhnin et al., 2006a). Linearized plasmid pYH110 was used to produce hag RNA with the MEGAscript kit. Gel-purified RNA (150 nM) in TE was hybridized to a 32P-end-labelled DNA oligonucleotide (150 nM) complementary to the 3′ end of the transcript by heating to 85°C for 3 min and slow cooling. Toeprint assays were carried out with 1 μM CsrA-H6 and/or 200 nM E. coli 30S ribosomal subunits and 5 μM E. coli tRNAfMet (Sigma). Toeprint reaction mixtures (10 μl) contained 15 nM RNA:DNA oligonucleotide hybrid, 375 μM each dNTP, 10 mM DTT and 100 μg ml−1 BSA in AMV reverse transcriptase buffer (Roche). CsrA toeprint reactions were incubated for 30 min at 37°C to allow CsrA–hag RNA complex formation. 30S ribosomal subunits were activated by incubation for 15 min at 37°C before being added to the toeprint reaction mixture. Toeprint reaction mixtures containing CsrA and/or tRNAfMet and 30S ribosomal subunits were incubated for 10 min at 37°C prior to the addition of 0.5 U of AMV reverse transcriptase. Incubation was then continued for 15 min at 37°C. Reactions were terminated by the addition of 6 μl of the same stop solution used in the primer extension assay. Samples were fractionated through standard 6% sequencing gels. Sequencing reactions were performed using pYH110 as the template and the same end-labelled DNA oligonucleotide as a primer. Radiolabelled bands were visualized by phosphorimagery.
RNA-directed cell-free translation
hag′–′gfp and ycbK′–′gfp transcripts were synthesized using the MEGAscript kit. A CsrA-deficient B. subtilis S-30 extract was prepared by following published procedures (Chambliss et al., 1983; Yakhnin et al., 2004). Cell-free translation reactions were carried out by modifying published procedures (Du and Babitzke, 1998; Yakhnin et al., 2004). The S-30 extract was incubated with RNase-free DNase I for 15 min at 37°C to remove chromosomal DNA and to allow time for endogenous mRNA to be degraded by cellular RNases. Reaction mixtures (24 μl) contained 60 mM Tris-HEPES (pH 7.5), 60 mM NH4Cl, 15 mM MgCl2, 12 mM KCl, 0.5 mM EGTA, 5 mM DTT, 2 mM ATP, 0.6 mM GTP, 0.08 mM calcium folinate, 4 µg ml−1 aprotinin, 4 µg ml−1 leupeptin, 4 µg ml−1 pepstatin A, 4 μl of S-30 extract (12 mg of total protein), 800 U ml−1 DNase I, 500 U ml−1 RNasin, 10 mM phosphoenolpyruvate, 35 U ml−1 pyruvate kinase, 0.4 mg ml−1E. coli tRNA, 0.04 μg ml−1 mRNA, 10 μCi of [35S]-methionine, 5 mM potassium glutamate, 5 mM glutamine, and 0.1 mM of each of the other amino acids. Reaction mixtures were incubated for 30 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−1 bromophenol blue). Aliquots (10 μl) were heated at 95°C for 5 min and proteins were analysed by 14% SDS-PAGE. Radiolabelled bands were visualized by phosphorimagery.
The authors thank Michael Chamberlin for plasmid pMG601, as well as Carla Fisher and Matthew Meyer for technical assistance. N-terminal sequencing of Hag was performed at The Pennsylvania State University College of Medicine macromolecular core facility in Hershey, PA. This work was supported by Grants GM52840 and GM59969 from the National Institutes of Health.