Salmonella pathogenicity islands 1 and 2 (SPI-1 and SPI-2) play key roles in the pathogenesis of Salmonella enterica. Previously, we showed that when Salmonella grows in Luria–Bertani medium, HilD, encoded in SPI-1, first induces the expression of hilA, located in SPI-1, and subsequently of the ssrAB operon, located in SPI-2. These genes code for HilA and the SsrA/B two-component system, the positive regulators of the SPI-1 and SPI-2 regulons respectively. In this study, we demonstrate that CsrA, a global regulatory RNA binding protein, post-transcriptionally regulates hilD expression by directly binding near the Shine–Dalgarno and translation initiation codon sequences of the hilD mRNA, preventing its translation and leading to its accelerated turnover. Negative regulation is counteracted by the global SirA/BarA two-component system, which directly activates the expression of CsrB and CsrC, two non-coding regulatory RNAs that sequester CsrA, thereby preventing it from binding to its target mRNAs. Our results illustrate the integration of global and specific regulators into a multifactorial regulatory cascade controlling the expression of virulence genes acquired by horizontal transfer events.
Horizontal transfer of DNA fragments, encoding for a variety of virulence factors, has been of paramount importance in the evolution of pathogenic bacteria by providing the ability to survive in and exploit different host niches and thus to cause different diseases (Ochman et al., 2000; Schmidt and Hensel, 2004). To take advantage of the genetic information contained in the acquired genes, bacteria had to adapt regulatory mechanisms to control their potentially deleterious expression under non-permissive conditions, and to induce their appropriate spatiotemporal expression inside the hosts to establish an infection.
Salmonella are Gram-negative bacteria comprising around 2500 serotypes, many of which are important human and animal pathogens. Two major clinical manifestations characterize Salmonella infections: enteritis (salmonellosis), a self-limiting local intestinal inflammatory response, and enteric fever (typhoid), a systemic infection characterized by the dissemination of the bacteria throughout the reticuloendothelial system via infected macrophages to the lymph nodes, liver and spleen (Ohl and Miller, 2001; Haraga et al., 2008). Salmonella enterica serotype Typhimurium (S. Typhimurium) can produce enteritis in humans, calves or chickens, whereas in mice it causes a systemic infection similar to human typhoid fever produced by serotype Typhi (Ohl and Miller, 2001; Haraga et al., 2008); thus it is frequently used as a model to study Salmonella pathogenesis. Most of the Salmonella virulence genes are located within distinct genomic regions denominated Salmonella pathogenicity islands (SPIs), which have been acquired by independent horizontal transfer events (Groisman and Ochman, 1997; Marcus et al., 2000). SPI-1 and SPI-2, composed by 39 and 44 genes, respectively, play a key role in Salmonella pathogenesis (Hansen-Wester and Hensel, 2001). Each pathogenicity island encodes a type three secretion system (T3SS), different effector proteins, which are translocated into the host cells by their cognate T3SS, chaperones and transcriptional regulators that control the expression of the genes within each island (Hensel, 2000; Marcus et al., 2000; Hansen-Wester and Hensel, 2001). SPI-1 genes are necessary for the invasion of epithelial host cells, thus for the intestinal colonization leading to enteritis; whereas SPI-2 genes are required for bacterial replication/survival inside macrophages, and thus for the systemic disease (Hensel et al., 1998; Santos et al., 2003; Haraga et al., 2008). However, different studies have revealed that SPI-2 genes also play a role in the development of the intestinal inflammatory disease and, accordingly, they are expressed in the intestinal lumen (Bispham et al., 2001; Brown et al., 2005; Coburn et al., 2005; Coombes et al., 2005; Hapfelmeier et al., 2005; Jones et al., 2007).
Here we demonstrate that the SirA/BarA two-component system and the Csr post-transcriptional regulatory system regulate the expression of both the SPI-1 and SPI-2 genes in Salmonella growing in LB medium. SirA activates the expression of CsrB and CsrC, which counteract the post-transcriptional repression exerted by CsrA, whose binding sites overlap the SD sequence and translation initiation codon of hilD mRNA. HilD then activates expression of hilA and ssrAB, the genes encoding the central regulators of SPI-1 and SPI-2 respectively. This study integrates a regulatory cascade constituted by global and horizontally acquired regulatory elements that controls the expression of the SPI-1 and SPI-2 regulons.
SirA is involved in the expression of the SPI-2 regulon
Recently, we demonstrated that the SPI-2 genes are expressed when S. Typhimurium grows in the nutritionally rich medium LB (Bustamante et al., 2008). Under this growth condition the SPI-1-encoded regulator HilD activates the sequential expression of the SPI-1 and SPI-2 regulons at the early and late stationary phases of growth respectively. In this study, we investigated the possibility that, in addition to HilD, other reported SPI-1 regulators could be involved in controlling the expression of SPI-2 genes. Previous studies have shown that the global SirA/BarA two-component system positively regulates the expression of SPI-1 (Johnston et al., 1996; Ahmer et al., 1999; Altier et al., 2000a; Teplitski et al., 2003) and SPI-4 genes (Gerlach et al., 2007).
To investigate if SirA also regulates the SPI-2 genes, we analysed the expression of a transcriptional fusion of the ssaG (SPI-2) promoter to the cat (chloramphenicol acetyl transferase) reporter gene, in wild-type (WT) S. Typhimurium strain SL1344 and its ΔsirA derivative, as well as in the ΔhilC, ΔhilA, ΔhilD, ΔompR and ΔssrB mutants. Each strain was grown in LB or in N-minimal medium (N-MM), two different in vitro growth conditions shown to sustain the expression of SPI-2 genes (Deiwick et al., 1999; Miao and Miller, 2000; Bustamante et al., 2008). Expression of ssaG-cat showed a similar reduction in the ΔsirA and ΔhilD mutants grown in LB with respect to the WT strain (Fig. 1A), but was not affected when they were grown in N-MM (Fig. 1B). These results are consistent with our previous observations indicating that HilD is required for the expression of the SPI-2 genes when Salmonella grows in LB but not in N-MM (Bustamante et al., 2008). Also in accordance with our previous results (Bustamante et al., 2008), expression of ssaG was reduced in both LB and N-MM in the ΔompR and ΔssrB mutants, which lack the SPI-2 regulators OmpR and SsrB. In contrast, expression was not affected in the ΔhilA and ΔhilC mutants that lack the SPI-1 regulators HilA and HilC (Fig. 1A and B). As a control, we determined the expression of a hilA-cat transcriptional fusion in the same strains grown in LB. In agreement with previous studies showing that both SirA and HilD positively regulate hilA (SPI-1) expression (Johnston et al., 1996; Ahmer et al., 1999; Schechter et al., 1999; Schechter and Lee, 2001; Schechter et al., 2003; Teplitski et al., 2003; Ellermeier et al., 2005; Olekhnovich and Kadner, 2006), expression of the hilA-cat fusion was reduced in the ΔsirA and ΔhilD mutants, but was not affected in the ΔompR, ΔssrB and ΔhilC mutants (Fig. 1C). Furthermore, expression of the hilA-cat fusion increased in the ΔhilA mutant (Fig. 1C), consistent with a previous study showing that HilA negatively regulates its own expression (De Keersmaecker et al., 2005). Taken together, these results indicated that SirA, in addition to HilD, OmpR and SsrB, is required for the expression of SPI-2 genes when Salmonella grows in LB.
To determine if SirA controls ssaG expression, and thus of other SPI-2 genes, by regulating the ssrAB operon, we analysed the expression of SsrB in WT and ΔsirA strains, as well as in the ΔhilA, ΔhilD, ΔhilC, and ΔompR mutants, which were tested as controls. To follow the expression of SsrB, the ssrB chromosomal gene was tagged with the sequence encoding a 3×FLAG epitope, rendering strains expressing a C-terminal 3×FLAG-tagged SsrB protein (SsrB-FLAG). Total protein extracts were obtained from culture samples of these strains grown in LB and analysed by Western blotting. In agreement with the results shown in Fig. 1A, expression of SsrB-FLAG was drastically reduced in the ΔsirA, ΔhilD and ΔompR mutants, but not in the ΔhilA and ΔhilC mutants (Fig. 1D). In contrast, when these strains were grown in N-MM, SsrB-FLAG expression was not affected in the ΔsirA and ΔhilD mutants, but was still highly reduced in the ΔompR mutant (data not shown). Together, these results indicated that expression of the SsrA/B system and, as a consequence, the SPI-2 genes, requires the HilD, OmpR and SirA proteins when S. Typhimurium is grown in LB.
SirA controls the expression of SPI-1 and SPI-2 genes by regulating HilD expression
HilD and OmpR positively regulate ssrAB expression by directly binding to its regulatory region (Lee et al., 2000; Feng et al., 2003; 2004; Bustamante et al., 2008). Because the expression of OmpR was not affected in the absence of SirA or HilD (data not shown), we hypothesized that SirA could regulate the expression of the ssrAB operon directly, acting along with HilD and OmpR, or indirectly by controlling the expression of HilD in a cascade fashion. To distinguish between these possibilities, we first analysed the expression of SsrB-FLAG by Western blot in the ΔhilD and ΔsirA mutants containing plasmids pBAD-HilD1 or pBAD-SirA1, expressing HilD and SirA respectively. Expression of HilD from plasmid pBAD-HilD1 restored the expression of SsrB-FLAG in both the ΔhilD and ΔsirA mutants to a level that was similar to the WT strain (Fig. 2A). In contrast, expression of SirA from plasmid pBAD-SirA1 only restored the expression of SsrB-FLAG in the ΔsirA mutant (Fig. 2A). Similar results were obtained when the expression of a C-terminal 3×FLAG-tagged HilA protein (HilA-FLAG) or of the hilA-cat fusion was determined in the same genetic backgrounds containing plasmids expressing HilD or SirA (Figs 2B and S1).
These results show that HilD is able to activate the expression of ssrAB and hilA in the absence of SirA, whereas SirA requires the presence of HilD, suggesting that SirA controls HilD expression, which in turn directly regulates hilA and ssrAB. In agreement with this notion, the presence of pT3-HilD1, but not of pK3-SirA1, activated expression of hilA-cat, hilC-cat and rtsA-cat transcriptional fusions in the absence of any other Salmonella-specific regulatory protein, as it was determined in an E. coliΔuvrY mutant lacking UvrY, the E. coli orthologue of SirA (Fig. S2). In contrast, expression of a csrB-cat transcriptional fusion was activated in the presence of pK3-SirA1, but not of pT3-HilD1 (Fig. S2). These results are in agreement with previous studies showing that HilD directly regulates hilA, hilC and rtsA (Schechter et al., 1999; Schechter and Lee, 2001; Ellermeier et al., 2005) and that SirA directly controls the expression of csrB (Teplitski et al., 2003; Teplitski et al., 2006).
The results described above suggest that the SirA/BarA system controls HilD expression. To further determine the extent to which the SirA/BarA system controls hilD expression, we analysed the activity of a hilD-cat transcriptional fusion and measured hilD mRNA levels by q-RT-PCR in WT, ΔsirA and ΔbarA strains. As HilD has also been reported to positively regulate its own expression (Olekhnovich and Kadner, 2002; Ellermeier et al., 2005), the expression of hilD was also tested in the ΔhilD mutant. Expression of hilD was similarly reduced in the ΔsirA and ΔbarA mutants and to a greater extent in the ΔhilD mutant (Fig. 3). These results, together with those from previous studies (Lucas and Lee, 2001; Ellermeier et al., 2005; Mizusaki et al., 2008), demonstrate that the SirA/BarA system positively regulates the expression of hilD.
SirA binds to the csrB and csrC regulatory regions, but not to those of hilD, hilA, hilC or ssrAB
The results described above are consistent with the notion that SirA acts through HilD to indirectly control the expression of hilA and ssrAB. In this regard, results from a previous study suggested that SirA most likely acts along with HilD or through HilD to control hilA expression and thus SPI-1 regulation (Ellermeier et al., 2005). Moreover, it is known that SirA regulates the expression of CsrB and CsrC and that these non-coding RNAs positively regulate Salmonella invasion and expression of SPI-1 genes (Altier et al., 2000a; Teplitski et al., 2003; Fortune et al., 2006; Teplitski et al., 2006). However, it has also been shown that SirA binds to the regulatory regions of hilA and csrB, but that it does not bind to the regulatory regions of hilD and csrC (Teplitski et al., 2003; Teplitski et al., 2006).
To further dissect and clarify the mechanism by which SirA regulates the expression of SPI-1 and SPI-2 genes, we analysed the interaction of this protein with the regulatory regions of genes under the control of the SirA cascade. SirA was purified as an MBP-SirA fusion protein expressed from plasmid pMAL-SirA1 (as described in Experimental procedures section in Supporting information). SPI-1 protein secretion was restored in the ΔsirA mutant carrying plasmid pMAL-SirA1 (data not shown), confirming the functionality of MBP-SirA. Binding of phosphorylated and non-phosphorylated purified MBP-SirA to a 32P-5′-end-labelled PCR product spanning the regulatory region of csrB was tested by electrophoretic mobility shift assays (EMSAs) to further verify the functionality of the fusion protein. MBP-SirA similarly shifted the csrB fragment whether it was phosphorylated or not, starting at a concentration of 1.5 µM (Fig. S3A), indicating that MBP-SirA phosphorylation is not required for DNA binding in vitro, which is consistent with previous results showing that phosphorylation only increases the DNA binding affinity of SirA-His6X by approximately twofold (Teplitski et al., 2003). SirA phosphorylation in vivo seems to be critical for its function, because when its predicted phosphorylated residue, Asp54, is substituted for an alanine, or in the absence of its cognate sensor kinase BarA, csrB expression is not induced (data not shown). Therefore, it is likely that the in vitro conditions used to test SirA DNA binding, or the fusion of SirA to the MBP or to the His6X tag, rendered a SirA protein that specifically binds to its target DNA sequences regardless of its phosphorylation state.
In contrast, a DNA fragment containing the regulatory region of sigD, which was used as a negative control, was not shifted even at a concentration of 3 µM (Fig. S3A). Specific binding was confirmed by competing MBP-SirA binding to the labelled csrB fragment with 10- to 50-fold excess of unlabelled specific (csrB) and non-specific (sigD) fragments by EMSA. As shown in Fig. S3B, only the unlabelled csrB fragment was able to compete with the interaction of MBP-SirA with the labelled fragment of csrB.
We next performed EMSAs with phosphorylated MBP-SirA and labelled DNA fragments containing the regulatory regions of hilD, hilA, hilC, ssrAB and csrC. The DNA fragments used for the EMSAs were those used to construct the corresponding transcriptional fusions that showed SirA-mediated regulation (Figs 5 and S4). The fragments spanning the hilD, hilA, hilC, ssrAB and sigD regulatory regions were not shifted by phosphorylated MBP-SirA up to a concentration of 3 µM (Fig. 4A–D). Similar results were obtained using non-phosphorylated MBP-SirA (data not shown). At higher concentrations (≥ 5 µM), phosphorylated or non-phosphorylated MBP-SirA bound to all of these fragments, including sigD, suggesting that MBP-SirA binds non-specifically at these concentrations (data not shown). In contrast, csrC was bound by MBP-SirA, starting at a concentration of 2 µM (Fig. 4E). Specific binding to the csrC regulatory region was also confirmed by competitive EMSAs, as described above for csrB. Only a 10- to 50-fold excess of unlabelled csrC fragment, but not of the unlabelled non-specific sigD fragment, was able to disrupt the interaction of MBP-SirA with the csrC labelled fragment (Fig. 4F), confirming the specific binding of MBP-SirA to csrC. SirA-mediated regulation of csrB and csrC in the strain used in this study (S. Typhimurium SL1344), was confirmed by analysing the expression of csrB-cat and csrC-cat transcriptional fusions in the WT and ΔsirA strains (Fig. S4).
The lack of binding of MBP-SirA to the hilD, hilA and ssrAB regulatory regions is in agreement with the genetic evidence indicating that the SirA/BarA system regulates the expression of hilD, and thus of ssrAB and hilA, indirectly. In contrast, specific binding of MBP-SirA to both the csrB and csrC regulatory regions supports the notion that the SirA-mediated expression of hilD is regulated by the Csr system.
SirA controls the expression of HilD through the CsrB and CsrC RNA molecules
Therefore, to determine if SirA regulates expression of hilD and the SPI-1 and SPI-2 genes through CsrB and CsrC, single ΔcsrB and ΔcsrC mutants, as well as a ΔcsrBΔcsrC double mutant, were constructed as described in Experimental procedures. Expression of the hilD-cat (SPI-1), hilA-cat (SPI-1), ssrAB-cat (SPI-2) and ssaG-cat (SPI-2) transcriptional fusions was determined in these mutants and compared with their expression in WT and ΔsirA strains; additionally, the level of the hilD mRNA was measured by q-RT-PCR. Expression of hilD was not appreciably affected in the single ΔcsrB or ΔcsrC mutants; however, expression in the ΔcsrBΔcsrC double mutant strain was reduced to the level observed in the ΔsirA mutant (Fig. 5A and B). As expected, expression of the hilA, ssrAB and ssaG fusions was also reduced in the ΔcsrBΔcsrC double mutant and in the ΔsirA mutant, but not in the single ΔcsrB and ΔcsrC mutants (Fig. 5C and data not shown). Furthermore, the secretion of the SPI-1-encoded proteins SipA, SipB, SipC and SipD (Fig. 5D, upper), as well as the expression of the SPI-2-encoded effector protein SseB (Fig. 5D, lower), were considerably reduced in the ΔcsrBΔcsrC double mutant and, as expected, in the ΔsirA mutant, but not in the single ΔcsrB and ΔcsrC mutants. In agreement with these results, a previous study reported that the expression of hilA and sipC (a SPI-1 gene), decreased in a ΔcsrBΔcsrC double mutant, while it was not affected in ΔcsrB and ΔcsrC single mutants (Fortune et al., 2006). It has been demonstrated that the loss of CsrB causes a compensatory increase in CsrC levels and vice versa (Weilbacher et al., 2003; Fortune et al., 2006), which probably explains why the ΔcsrBΔcsrC double mutant, but not the single mutants, affected expression of the SPI-1 and SPI-2 genes.
To further investigate if SirA acts only through CsrB/C or has an additional CsrB/C-independent regulatory role on the expression of hilD or the SPI-1 and SPI-2 genes, we analysed the expression of the hilD-cat, hilA-cat and ssaG-cat fusions in the ΔsirA mutant containing plasmids pMPM-K3 (empty vector) or pK3-CsrB1, expressing CsrB from a constitutive promoter. Expression of hilD, hilA and ssaG was restored in the ΔsirA mutant, but not in the ΔhilD mutant, in the presence of pK3-CsrB1 (Fig. 6). In agreement with the notion that SirA controls the expression of hilD and, as a consequence, of the SPI-1 and SPI-2 genes through CsrB/C, a plasmid expressing SirA was not able to restore the secretion of the SPI-1-encoded proteins (Fig. 7, upper), or the expression of the SPI-2-encoded SseB protein (Fig. 7, lower), in a ΔsirAΔcsrBΔcsrC triple mutant. In contrast, SPI-1-mediated protein secretion and the expression of SseB, were restored by a plasmid expressing HilD in this triple mutant (Fig. 7). Taken together, these results support the model in which SirA indirectly regulates hilD expression, and thus the SPI-1 and SPI-2 genes, by directly activating transcription of csrB and csrC.
CsrA represses expression of HilD by binding to the hilD leader transcript
In E. coli, CsrB and CsrC regulate gene expression by binding to and sequestering the RNA binding protein CsrA, thus counteracting the negative post-transcriptional control exerted by CsrA on its target mRNAs (Babitzke and Romeo, 2007; Lucchetti-Miganeh et al., 2008; Timmermans and Van Melderen, 2010). To evaluate the regulatory role of CsrA on the expression of the SPI-1 and SPI-2 genes, we first analysed the expression of these genes when CsrA is overexpressed. As CsrB and CsrC antagonize CsrA activity, the phenotypes associated with CsrA overexpression were predicted to mimic the phenotypes observed in the ΔcsrBΔcsrC double mutant strain. The transcriptional activity of the hilD-cat, hilA-cat and ssrAB-cat, as well as of the sirA-cat, csrB-cat and csrC-cat transcriptional fusions was determined in a WT strain containing plasmids pMPM-K3 (empty vector) or pK3-CsrA1. Additionally, the level of the hilD mRNA was measured by q-RT-PCR and the expression of HilA-FLAG and SsrB-FLAG proteins was determined by Western blot, in these same genetic backgrounds. Overproduction of CsrA drastically reduced the expression of hilD, hilA and ssrAB (Fig. 8A, B and E); while the expression of sirA, csrB and csrC was not affected (Fig. 8C and D). Furthermore, the presence of pK3-CsrA1 drastically reduced the expression of HilA-FLAG and SsrB-FLAG, as well as the secretion of the SPI-1-encoded proteins in WT S. Typhimurium (Fig. 8F and data not shown).
To further confirm the role of CsrA, we analysed the expression of SPI-1 and SPI-2 genes in a csrA deletion mutant (Table 1) carrying the hilD-cat, hilA-cat, ssrAB-cat, sirA-cat, csrB-cat and csrC-cat reporter fusions. The level of the hilD mRNA was also measured by q-RT-PCR in this mutant. Consistent with the observed negative regulatory role of CsrA overexpression, expression of hilD, hilA and ssrAB significantly increased in the ΔcsrA mutant (Fig. 8A, B and E), whereas the expression of the sirA, csrB and csrC was not affected (Fig. 8C and D).
In agreement with these results, it was previously shown that overproduction of CsrA reduced the expression of hilD and hilA (Altier et al., 2000b). This study also reported the generation of a csrA mutant in S. Typhimurium strain 14028s, which surprisingly showed a reduction in the expression of SPI-1 genes; however, this mutant presented a severe growth defect, which may have affected the growth phase-dependent induction of the SPI-1 genes. Our ΔcsrA mutant showed a less severe growth defect, which was restored when CsrA was expressed in trans (Fig. S5). It remains to be seen if the different effects on growth and on the expression of SPI-1 and SPI-2 genes caused by the absence of CsrA could also be attributed to distinct features of the strains used, SL1344 (this study) and 14028s (Altier et al., 2000b), or to random secondary suppressor mutations acquired as a consequence of eliminating a pleiotropic regulatory protein.
The results described above, indicating that SirA/BarA and CsrB/C activate expression of hilA and ssrAB through HilD, also suggested that CsrA could directly repress hilD expression. To test this possibility, we performed EMSAs to determine whether CsrA binds to hilD, hilA, ssrAB or csrB transcripts. The addition of 1 µM CsrA led to reduced mobility of the hilD, ssrAB and csrB transcripts, indicating that CsrA bound to these RNAs (Fig. 9A). In contrast, CsrA did not bind to hilA RNA (Fig. 9A). We next performed experiments to determine the affinity of CsrA for hilD, ssrAB and csrB mRNAs. CsrA bound to the hilD transcript as a distinct band (Fig. 9B). A non-linear least-squares analysis of these data yielded an apparent Kd value of 5 nM CsrA for hilD mRNA. The specificity of CsrA–hilD RNA interaction was investigated by performing competition experiments with specific (hilD) and non-specific (E. coli phoB) unlabelled RNA competitors. Unlabelled hilD was an effective competitor, whereas phoB did not compete with the CsrA–hilD RNA interaction (Fig. 9B). These results establish that CsrA binds specifically to the hilD transcript. Similar binding experiments were performed with the ssrAB and csrB transcripts. The affinity of CsrA for ssrAB mRNA (Kd = 340 nM) was much lower than for hilD. In addition, while unlabelled ssrAB RNA was able to compete with the labelled transcript for CsrA binding, it did so poorly (Fig. 9C). As expected, CsrA bound specifically to csrB mRNA with high affinity (Kd = 20 pM). Moreover, multiple shifted species were observed (Fig. 9D). While the stoichiometry of the shifted complexes was not examined, the first shifted species likely contain one CsrA dimer bound per transcript, with the slower migrating species containing multiple CsrA molecules bound to each transcript.
Footprint experiments were then performed to identify the CsrA binding site(s) in the hilD and ssrAB transcripts using RNase T1 as a probe (cleaves following single-stranded G residues). CsrA protected five G residues in the hilD transcript that overlapped the hilD SD sequence and translation initiation codon (Fig. 9E). As both of these regions contain the highly conserved GGA motif (Fig. 9E), it is apparent that CsrA binds to two sites in this transcript. These results indicate that CsrA represses hilD expression, presumably by blocking 30S ribosomal subunit binding. CsrA also protected several G residues in the ssrAB transcript that overlapped its SD sequence (data not shown). Thus, it appears that CsrA is also capable of directly repressing ssrAB translation by binding to a single site in the ssrAB transcript, although the reduced affinity of CsrA for this RNA suggests that repression would be less extensive than for hilD.
By blocking ribosome binding and translation, CsrA-mediated translational repression often leads to accelerated degradation of target transcripts (Babitzke and Romeo, 2007; Lucchetti-Miganeh et al., 2008, Timmermans and Van Melderen, 2010). Therefore, we tested whether bound CsrA accelerated hilD mRNA decay. The half-life of hilD mRNA was determined in WT and ΔcsrA strains. To facilitate detection of hilD mRNA, the two strains were transformed with plasmid pT3-HilD1, which carries the hilD gene under the control of a constitutive promoter, to increase the cellular concentration of the hilD transcript. These strains were grown to mid-exponential phase in LB; then, transcription was inhibited by adding rifampicin. Culture samples were collected at different time points after addition of the drug for RNA extraction and analysis by Northern blotting. The hilD mRNA was detected as two unique bands, which showed similar intensity in the WT and ΔcsrA mutant strains immediately after the treatment with rifampicin (Fig. 9F). However, the hilD mRNA decayed faster in the WT strain (0.79 ± 0.05 min half-life) compared with the ΔcsrA mutant (1.54 ± 0.05 min half-life) (Fig. 9F).
Taken together, it can be concluded that CsrA-mediated translational repression leads to more rapid degradation of hilD mRNA. In addition, by reducing HilD synthesis, CsrA would also affect HilD positive autoregulation (Olekhnovich and Kadner, 2002; Ellermeier et al., 2005), thus reducing hilD transcription. Furthermore, our results suggest that CsrA indirectly represses hilA and ssrAB transcription under growth conditions that favour HilD-dependent activation of the SPI-1 and SPI-2 regulons.
In recent years it has become more evident that the mechanisms regulating virulence gene expression in Salmonella are more complex than previously thought. A myriad of regulatory factors and environmental cues have been shown to be part of the molecular events leading to the appropriate spatiotemporal expression of the genes that integrate the SPI-1 and SPI-2 regulons (Altier, 2005; Jones, 2005; Ellermeier and Slauch, 2007; Fass and Groisman, 2009). In this regard, we recently provided evidence showing that HilD is a key regulatory protein that establishes transcriptional cross-talk between the SPI-1 and SPI-2 regulons, by sequentially activating expression of the SPI-1 and SPI-2 genes when Salmonella grows in LB medium (Bustamante et al., 2008). HilD directly binds to the regulatory regions of hilA and the ssrAB operon, thereby interfering with H-NS-mediated repression of their promoters. HilA and the SsrA/B two component system, in turn, activate the SPI-1 and SPI-2 regulons respectively (Schechter et al., 2003; Olekhnovich and Kadner, 2006; Bustamante et al., 2008). Interestingly, HilD is not required for expression of the ssrAB operon when Salmonella grows in minimal media containing low concentrations of magnesium and phosphate (Bustamante et al., 2008). Under these growth conditions ssrAB expression instead depends on other regulators such as OmpR, PhoP and SlyA (Lee et al., 2000; Feng et al., 2003; 2004; Bijlsma and Groisman, 2005; Linehan et al., 2005; Fass and Groisman, 2009).
First, we demonstrated that SirA positively controls expression of the SPI-1 and SPI-2 genes through HilD. A previous study reported that SirA-His6 bound to a region located downstream of the hilA promoter and the regulatory region of hilC (Teplitski et al., 2003), suggesting a direct SirA regulation of these SPI-1 genes. In contrast, we found that MBP-SirA specifically bound to the regulatory regions of csrB and csrC, but not to the regulatory regions of hilA and hilC containing the fragments used by Teplitski et al. 2003 (Figs 4 and S3). Differences in these results could be attributed to different DNA binding specificities between SirA-His6 and MBP-SirA, or other technical differences; however, the genetic evidence presented here supports the notion that SirA does not directly activate hilA. Briefly, we demonstrated that SirA cannot activate expression of SsrB and HilA proteins in the absence of HilD (Figs 2, S1 and S2), in agreement with results from a previous study (Ellermeier et al., 2005). In contrast, HilD activated expression of ssrAB and hilA in the absence of SirA, indicating that HilD does not require the presence of SirA to directly activate expression of the SPI-1 and SPI-2 regulatory genes. In agreement with our results, mathematical models considering that SirA regulates SPI-1 genes through HilD are able to predict results comparable with experimental data. However, they fail to predict the level of participation of intermediate components of the SPI-1 regulatory cascade, when considering a direct regulation of SirA on hilA or hilC (Ganesh et al., 2009). Furthermore, CsrB was able to restore the expression of the SPI-1 and SPI-2 genes in the absence of SirA, but not in the absence of HilD (Fig. 6), indicating that SirA acts through CsrB to regulate these genes, and that HilD is downstream of SirA and CsrB in the regulatory cascade controlling them. This is consistent with previous studies indicating that SirA orthologues regulate gene expression through the corresponding CsrB and CsrC orthologues (Babitzke and Romeo, 2007; Lucchetti-Miganeh et al., 2008; Brencic et al., 2009; Timmermans and Van Melderen, 2010). In this regard, we demonstrated that SirA specifically binds to the csrC regulatory region (Fig. 4), an interaction previously demonstrated only for csrB (Teplitski et al., 2003; Teplitski et al., 2006). Moreover, the demonstration that CsrA binds to the SD sequence and translation initiation codon of hilD mRNA, but not to hilA mRNA (Fig. 9), strongly correlates with the genetic and biochemical data indicating that regulation of the SPI-1 and SPI-2 regulons require direct activation of csrB and csrC by SirA, which in turn overcomes CsrA-dependent repression of HilD synthesis. HilD is then responsible for the direct activation of hilA and ssrAB and, indirectly, of the SPI-1 and SPI-2 regulons.
Additionally, we found that CsrA binds to ssrAB mRNA, but with a much lower affinity than to hilD mRNA (Kd = 340 nM vs. 5 nM) (Fig. 9). Whether this weak CsrA–ssrAB RNA interaction is capable of regulating ssrAB expression remains to be determined. However, as expression of ssrAB requires HilD when Salmonella grows in LB medium, CsrA-mediated regulation of HilD synthesis appears to be the primary mechanism by which CsrA controls ssrAB expression. Moreover, SPI-2 expression was not significantly affected in the ΔsirA mutant growing in N-MM (Fig. 1), which favours HilD-independent activation of ssrAB.
Our results together with those from other groups, illustrate how multiple global and Salmonella-specific regulators have been integrated to form a complex regulatory circuitry that allow the appropriate spatiotemporal expression of genes that have been acquired by horizontal transfer events. In addition, they further expand our understanding about the molecular mechanisms co-ordinating the expression of the SPI-1 and SPI-2 genes.
Bacterial strains, plasmids and primers used in this work are listed in Tables 1, 2 and S1 respectively.
Table 2. Plasmids used in this study.
Reference or source
The co-ordinates for the cat fusions are indicated with respect to the transcriptional start site for each gene.
pBR322 derivative containing a promoterless chloramphenicol acetyltransferase (cat) gene, ApR
Bacterial cultures were grown at 37°C in LB medium, containing 1% tryptone, 0.5% yeast agar and 1% NaCl, at pH 7.5; or in N-MM containing 5 mM KCl, 7.5 mM (NH4)2SO4, 0.5 mM K2SO4, 1 mM KH2PO4, 100 mM Tris-HCl (pH 7.5), 10 µM MgCl2, 0.5% glycerol and 0.1% casamino acids. When necessary, media were supplemented with the following antibiotics: ampicillin (200 µg ml−1), streptomycin (100 µg ml−1), tetracycline (10 µg ml−1) or kanamycin (20 µg ml−1). Bacterial cultures grown overnight in LB medium were concentrated and resuspended in fresh LB or N-MM to an OD600 of 1. Then, 250 ml flasks containing 50 ml of LB or N-MM were inoculated with 1 ml of the bacterial suspensions and incubated at 37°C in a shaking water bath at 200 r.p.m. (Gyromax 902; Amerex Instruments). Culture samples for CAT activity determination or for Western blot analysis, were taken after 5 or 10 h of growth in LB under the conditions described above. These time points were previously optimized for the analysis of SPI-1 and SPI-2 regulation respectively (Bustamante et al., 2008).
The CAT assays and protein quantification to calculate CAT-specific activities were performed as described previously (Puente et al., 1996).
Construction of mutant strains and strains expressing FLAG-tagged proteins
Non-polar gene-deletion mutant strains (Table 1) were generated by the λRed recombinase system, as reported previously (Datsenko and Wanner, 2000). The genes sirA, barA, csrB, csrC or csrA were replaced with a selectable kanamycin resistance cassette in the S. Typhimurium SL1344 strain generating strains ΔsirA::kan (JPTM23), ΔbarA::kan (JPTM39), ΔcsrB::kan (JPTM40), ΔcsrC::kan (JPTM41) and ΔcsrA::kan (JPTM44) respectively. The gene uvrY was replaced with the kanamycin resistance cassette in E. coli MC4100 to generate the ΔuvrY::kan (JPMC46) strain. When required, the kanamycin resistance cassette was excised from the respective mutant strain, by using helper plasmid pCP20, expressing the FLP recombinase, as described previously (Datsenko and Wanner, 2000), generating S. Typhimurium mutant strains ΔhilA (JPTM24), ΔhilD (JPTM25), ΔhilC (JPTM26), ΔsirA (JPTM27), ΔompR (JPTM28), ΔssrB (JPTM29) and ΔcsrB (JPTM42) and E. coliΔuvrY (JPMC47). The ΔcsrBΔcsrC::kan double mutant (JPTM43) was generated by introducing the ΔcsrC::kan deletion into the ΔcsrB mutant (JPTM42) by P22 transduction. The sirA::Tn10d ΔcsrBΔcsrC::kan triple mutant (JPTM45) was generated by introducing the sirA::Tn10d deletion into the ΔcsrBΔcsrC::kan double mutant (JPTM43) by P22 transduction.
The chromosomal ssrB gene was FLAG-tagged in S. Typhimurium SL1344, using a modification of the λRed recombinase system for gene replacement, as described previously (Uzzau et al., 2001), generating strain JPTM30. The ssrB::3XFLAG-kan allele from strain JPTM30 was transferred to strains ΔhilA (JPTM24), ΔhilD (JPTM25), ΔhilC (JPTM26), ΔsirA (JPTM27) and ΔompR (JPTM28) by P22 transduction, generating strains JPTM31, JPTM32, JPTM33, JPTM34 and JPTM35 respectively. The hilA::3XFLAG-kan allele from strain JPTM36 was transferred to strains ΔhilD (JPTM25) and ΔsirA (JPTM27) by P22 transduction, generating strains JPTM37 and JPTM38 respectively. All mutant strains were verified by PCR amplification and sequencing.
SDS/PAGE and Western blot assays
Whole-cell extracts were prepared from bacterial samples collected at the indicated time points from LB cultures. Ten micrograms of each extract were subjected to electrophoresis in SDS-12% polyacrylamide gels, and then transferred to 0.45 µm pore size nitrocellulose membranes (Bio-Rad), using a semidry transfer apparatus (Bio-Rad). The membranes containing the transferred proteins were blocked in 5% non-fat milk for 1 h. Immunoblots were performed with anti-FLAG M2 (Sigma) or anti-DnaK (StressGen) monoclonal antibodies, or anti-SseB polyclonal antibody (Coombes et al., 2004), at 1:1000, 1:20 000 and 1:2000 dilutions respectively. Horseradish peroxidase-conjugated anti-mouse or anti-rabbit (Pierce), at a dilution of 1:10,000, were used as the secondary antibodies. Bands on the blotted membranes were developed by incubation with the Western Lightning Chemiluminescence Reagent Plus (Perkin-Elmer) and exposed to KodaK X-Omat films.
Protein secretion analysis
S. Typhimurium strains were grown for 9 h in LB. Then, 1.5 ml samples were harvested from each culture and concentrated by centrifugation for 5 min at 16 000 g. Proteins contained in culture supernatants were precipitated by the addition of 10% (v/v) trichloroacetic acid and overnight incubation at 4°C. Precipitated proteins were concentrated by centrifugation at 16 000 g for 30 min at 4°C. Pellets were resuspended in SDS-PAGE loading buffer containing 10% (v/v) saturated Tris base. Samples were subjected to SDS-PAGE analysis, using 12% polyacrylamide gels and stained with Coomassie Brilliant Blue R-250.
The regulatory regions of hilD, hilC, csrB, csrC, hilA, ssrAB and sigD were amplified by PCR using primer pairs hilDFBamHI/hilDRHindIII, hilCFSmaI/hilCRSalI, csrB-Fw1HindIII/csrB-Rev1, csrC-BFw1/csrC-HRv1, hilA1FBamHI/hilA2RHindIII, ssaBFBglII/ssrBRS6E (Bustamante et al., 2008) and sigDBH1F/sigDH3R, respectively, and S. Typhimurium SL1344 chromosomal DNA as template. PCR products were labelled during the reaction using 32P-5′-end-labelled primers and then purified using the QIAquick PCR purification kit (Qiagen). DNA binding reactions were performed in binding buffer [10 mM Tris-HCl (pH 8), 50 mM KCl, 1 mM DTT, 0.5 mM EDTA, 5% glycerol and 10 µg ml−1 BSA] by mixing ≈ 100 ng of each PCR product with ≈ 100 ng of the sigD fragment (negative control) and increasing concentrations of purified phosphorylated or non-phosphorylated MBP-SirA, in a total volume of 20 µl. Phosphorylated MBP-SirA was generated by incubation with 100 mM acetyl phosphate (Sigma) for 180 min at room temperature, as previously reported (Kenney et al., 1995). Protein-DNA binding reactions were incubated at room temperature for 20 min and then electrophoretically separated in 6% non-denaturing polyacrylamide gels in 0.5× Tris-borate-EDTA buffer at room temperature. The unlabelled DNA fragments were stained with ethidium bromide and visualized with an Alpha-Imager UV transilluminator (Alpha Innotech). The 32P-5′-end-labelled DNA fragments were detected by exposing to KodaK X-Omat films.
Quantitative real-time RT-PCR (q-RT-PCR) assays
Total RNA of S. Typhimurium SL1344 and its isogenic mutants was purified from samples of cultures grown in LB medium at 37°C for 5 h, using the RNeasy Mini Kit (Qiagen). Chromosomal DNA was removed by incubating 2 µg of RNA from each strain with DNase I (Invitrogen) according to the manufacturer's instructions. cDNA was synthesized in a reaction containing 0.5 µg of each DNase-treated-RNA and 5 pmol of primers hilD-RT-R (for hilD) and dnaK-RT-R (for dnaK), using the Revert Aid H Minus First Strand cDNA Synthesis kit (Fermentas). The obtained cDNA was used as template for real-time PCR assays, with 5 pmol of primer pairs for hilD or dnaK (hilD-RT-F/hilD-RT-R or dnaK-RT-F/dnaK-RT-R, respectively) and the SYBR Green PCR Master Mix (Perkin-Elmer/Applied Bio-systems). Real-time PCR reactions were performed with the ABI Prism 7000 Sequence Detection System (Applied Bio-systems) set in the standard run mode, and data collected using the Rotor Gene 6000 series software 1.7. Reaction conditions were 10 min at 95°C, and 40 cycles at 95°C for 15 s and 60°C for 60 s. The level of dnaK mRNA was used as an internal control to normalize the results obtained for the hilD mRNA. The quantification technique used to analyse data was the 2 –ΔΔCT method (Livak and Schmittgen, 2001). All real-time PCR reactions were performed in triplicate and were repeated using RNA purified from three different bacterial cultures.
Electrophoretic mobility shift assays followed published procedures (Baker et al., 2007; Yakhnin et al., 2007). His-tagged CsrA (CsrA-H6) from E. coli was purified as described previously (Dubey et al., 2005). Note that the amino acid sequence of CsrA from S. enterica and E. coli are identical. RNA was synthesized in vitro using the RNA Maxx Transcription kit (Stratagene). PCR fragments used as templates in transcription reactions contained a T7 promoter and hilA, hilD, ssrAB or csrB sequences extending from +1 to +365, +1 to +89, +1 to +183 or +1 to +371 relative to the start of transcription respectively. Gel-purified RNA was 5′-end-labelled with [γ-32P]-ATP. RNA suspended in TE buffer was heated to 90°C for 1 min followed by slow cooling to room temperature. Binding reactions (10 µl) contained 10 mM Tris-HCl, pH 7.5, 10 mM MgCl2, 100 mM KCl, 200 ng µl−1 yeast RNA, 0.2 mg ml−1 BSA, 7.5% glycerol, 20 mM DTT, 10–500 pM RNA, CsrA-H6 (various concentrations) and 0.1 mg ml−1 xylene cyanol. Competition assay mixtures also contained unlabelled competitor RNA. Reaction mixtures were incubated for 30 min at 37°C to allow CsrA-RNA complex formation. Samples were then fractionated through native 10% polyacrylamide gels using 0.5× TBE as the gel running buffer. Radioactive bands were visualized with a phosphorimager and quantified using ImageQuant 5.2 software. Apparent equilibrium binding constants (Kd) of CsrA–RNA interaction were calculated as described previously (Yakhnin et al., 2000).
RNA footprint assays
RNA footprint assays followed published procedures (Baker et al., 2007; Yakhnin et al., 2007). Binding reactions (10 µl) containing various concentrations of CsrA-H6 and 5 nM hilD or ssrAB RNA were otherwise identical to those described for the EMSAs. After the initial binding reaction, 0.03 U RNase T1 (Fermentas) was added to the reaction mixtures, and incubation was continued for 15 min at 37°C. Reactions were terminated by the addition of 10 µl of gel loading buffer (0.37% EDTA, pH 8.0, 0.3% bromophenol blue and xylene cyanol in 95% formamide) and placed on ice. Partial alkaline hydrolysis and RNase T1 digestion ladders of each transcript were prepared as described previously (Bevilacqua and Bevilacqua, 1998). Samples were fractionated through standard 6% polyacrylamide sequencing gels. Radioactive bands were visualized with a phosphorimager.
RNA extraction, Northern blot and half-life determination
To determine hilD mRNA half-lives, S. Typhimurium WT and ΔcsrA mutant strains containing plasmid pT3-HilD1 were grown in LB to an OD600 ∼ 0.6, at which time 200 µg ml−1 rifampicin was added to the cultures to prevent transcription initiation. Total RNA was purified from samples taken at 0, 0.5, 1, 2 and 4 min after rifampicin addition by the hot phenol extraction method, as described previously (Georgellis et al., 1992).
Northern blot analysis was performed by fractionation of the purified RNA samples (20 µg) on a 1% agarose-formaldehyde gel, which were then transferred to a nitrocellulose membrane by capillary transfer. Membranes were cross-linked using a cross-linking device (Stratalinker, Stratagene), and prehybridized for 3 h at 42°C in a buffer containing 5× Denhardt's solution, 5× SSC, 0.2% SDS, 50% formamide and 250 µg of sheared salmon sperm DNA per ml. Subsequently, a radio-labelled hilD specific DNA probe, denatured at 90°C for 5 min, was added to the prehybridization buffer and the membranes were incubated at 42°C overnight. The hilD specific probe was obtained by digesting plasmid pT3-HilD1 with SalI and BamHI, separation of the fragments on agarose gels and purification of the hilD specific band by using the Qiagen Agarose Purification Kit. Probe labelling was performed by using [α-32P]dCTP and the Rediprime II Kit (Invitrogen), according to the manufacturer's instructions. Membranes were washed twice with 50 ml of SSC 2× and SDS 0.1% at 42°C, and twice with SSC 0.2× and SDS 0.1% at 42°C. Images were obtained using the phosphoimager screens and analysed using the Typhoon image scanner (Amersham). Determination of the hilD mRNA half-life was performed as previously described (Von Gabain et al., 1983).
Construction of plasmids and Expression and purification of MBP-SirA sections are described in Experimental procedures in Supporting information.
We thank A. Vázquez and F.J. Santana for technical assistance, E. Calva for critical reading of the manuscript, S. Miller and F. Heffron for providing Salmonella strains, and B.B. Finlay for providing the anti-SseB polyclonal antibody. This work was supported by grants from Dirección General de Asuntos del Personal Académico (DGAPA) (IN227306-3 and IN210309-3 to V.H.B., IN224107-3 to J.L.P., and IN219709-3 to D.G), from Consejo Nacional de Ciencia y Tecnología (CONACYT) (83277 to V.H.B and 6079 to J.L.P.), and from the National Institutes of Health (Public Health Service grant GM059969 to P.B.), and from the International Centre for Genetic Engineering and Biotechnology (ICGEB) (CRP/MEX08-02) (to D.G.). L.C.M is supported by a pre-doctoral fellowship from CONACYT (No. 169380).