The two-component system AfsQ1/Q2 of Streptomyces coelicolor was identified in our previous work as a pleiotropic regulator for antibiotic biosynthesis and morphological differentiation under the condition of a minimal medium supplemented with 75 mM glutamate. In this work, we report the dissection of the mechanism underlying the function of AfsQ1/Q2 on antibiotic production and also the identification of the AfsQ1/Q2 regulon. The results showed that AfsQ1/Q2 stimulated antibiotic ACT, RED and CDA production directly through the pathway-specific activator genes actII-ORF4, redZ and cdaR respectively. In addition, expression of sigQ that encodes a sigma factor and is divergently transcribed from afsQ1 was also subject to direct regulation by AfsQ1/Q2. The precise AfsQ1 binding sites in the upstream regions of these target genes were determined by DNase I footprinting assays coupled with site-directed DNA mutagenesis. By computational prediction and functional analysis, at least 17 new AfsQ1 targets were identified, including pstS gene encoding a high-affinity phosphate-binding protein and two developmental genes whiD, bldM. For the AfsQ1/Q2 regulon, an AfsQ1 binding motif comprising the sequence GTnAC-n6-GTnAC has been defined. Interestingly, we found from electrophoretic mobility shift assays and transcriptional analysis that AfsQ1/Q2 can also function as a repressor for nitrogen assimilation, and AfsQ1 can compete with GlnR for the promoter regions of glnA and nirB, suggesting the cross-regulation between AfsQ1/Q2 and GlnR in nitrogen metabolism. These findings suggested that AfsQ1/Q2 is important not only for antibiotic biosynthesis but also in maintaining the metabolic homeostasis of nutrient utilization under the stress of high concentration of glutamate in S. coelicolor.
Two-component systems (TCSs) are key signal transduction pathways employed by microorganisms to adapt to complex environments (Stock et al., 2000). Typically, TCS comprises a membrane-bound histidine kinase that senses a specific environmental stimulus, and a cognate response regulator that mediates the cellular response, mostly through transcriptional regulation of target genes (Mascher et al., 2006). We are especially interested in the TCSs from streptomycetes, one of the most important microorganisms for medical applications that produce a vast array of secondary metabolites with important biological activities (Kieser et al., 2000). A better understanding of TCS functions in Streptomyces, especially in its secondary metabolism, is of great importance. To date, the regulation of TCSs on secondary metabolism has been studied most intensively in the model strain of streptomycetes, Streptomyces coelicolor A3(2) (van Wezel and McDowall, 2011).
Streptomyces coelicolor A3(2) is a Gram-positive soil bacterium that undergoes a complex life cycle involving mycelial growth and spore formation. It produces multiple antibiotics, such as actinorhodin (ACT), undecylprodigiosin (RED) and calcium-dependent antibiotic (CDA) (Hopwood et al., 1995). Bioinformatic analyses of the S. coelicolor genome revealed the presence of a multitude of TCSs, including 67 typical TCSs, 13 orphan response regulators and 17 unpaired histidine kinases (Hutchings et al., 2004). During the past decades, increasing efforts have been made in understanding of the function of TCSs on primary and secondary metabolism in S. coelicolor, which led to the identification of several TCSs mainly involved in secondary metabolism, including AbsA1/A2, CutR/S, RapA1/A2, AbrA1/A2, AbrC1/C2/C3, AfsQ1/Q2 and an orphan histidine kinase (OhkA) (Ishizuka et al., 1992; Chang et al., 1996; Lu et al., 2007; 2011; McKenzie and Nodwell, 2007; Shu et al., 2009; Yepes et al., 2011) and two TCSs as key regulators for primary metabolism, including GlnR and PhoR/P (Fink et al., 2002; Sola-Landa et al., 2005; Apel et al., 2007; Tiffert et al., 2008; 2011). PhoR/P is a global regulator of phosphate metabolism and GlnR is an OmpR-type orphan response regulator that plays a central role in the regulation of nitrogen assimilation. In addition, the cross-regulation between different TCSs was also revealed recently (Rodriguez-Garcia et al., 2009; Wang et al., 2009). For example, Rodriguez-Garcia et al. (2009) reported the direct negative control on the transcription of glnR and its targets glnA, glnII and amtB by PhoP under phosphate limitation, suggesting the cross-regulation of PhoR/P and GlnR in nitrogen metabolism.
AfsQ1/Q2 was previously identified for its ability to stimulate ACT and RED production in Streptomyces lividans. However, deletion of afsQ1/Q2 in S. coelicolor resulted in no obvious phenotype changes under the tested conditions (Ishizuka et al., 1992). In our previous study, it was found that the effects of afsQ1/Q2 deletion on antibiotic biosynthesis can be detected on a minimal medium (MM) supplemented with 75 mM glutamate (Glu) as the sole nitrogen source. In addition, sigQ gene that encodes a putative sigma factor and is divergently transcribed from afsQ1/Q2 was shown to be under the control of AfsQ1/Q2 and function together with afsQ1/Q2 in the regulation of antibiotic biosynthesis (Shu et al., 2009).
In this study, we report the identification of the AfsQ1/Q2 regulon in S. coelicolor. The results showed that in addition to its role on antibiotic biosynthesis, AfsQ1/Q2 is also involved in the regulation of carbon and phosphate metabolism in S. coelicolor. Moreover, we presented evidences for the cross-regulation between AfsQ1/Q2 and GlnR in the control of nitrogen metabolism.
Comparison of growth and antibiotic production of the S. coelicolor parental strain M145 and its mutant ΔafsQ1/Q2
Our previous study demonstrated that the mutant ΔafsQ1/Q2 (with the deletion of TCS genes afsQ1/Q2) of S. coelicolor exhibited significantly impaired ACT, RED and CDA biosynthesis when grown on a MM supplemented with 75 mM L-glutamate (Glu) as the sole nitrogen source (Shu et al., 2009). Here, we quantitatively determined the effects of afsQ1/Q2 inactivation on ACT and RED production in detail. The parental strain M145 and ΔafsQ1/Q2 were cultivated on the MM plates supplemented with 75 mM Glu as described previously, and then the growth curves and antibiotic production were measured at different time points.
We found that, throughout the time-course, ΔafsQ1/Q2 was found to grow rapidly and accumulate a higher biomass than that of the original strain M145 (Fig. 1A). ACT biosynthesis was significantly impaired in the ΔafsQ1/Q2 mutant as shown in Fig. 1B. For RED production, we found that on the MM supplemented with 75 mM Glu, both M145 and the ΔafsQ1/Q2 mutant produced little RED, and deletion of afsQ1/Q2 led to obviously reduced RED production (Fig. 1C). Both rapid growth and reduced ACT and RED biosynthesis of the ΔafsQ1/Q2 mutant were restored by complementation with plasmid pSETafsQ harbouring the afsQ1/Q2 genes and their promoter regions (Fig. 1). These results confirmed the positive role of AfsQ1/Q2 in antibiotic biosynthesis in S. coelicolor on the MM supplemented with 75 mM Glu.
AfsQ1/Q2 activates antibiotic biosynthesis via direct interaction with pathway-specific regulatory gene promoters
We have previously demonstrated at the transcriptional level that AfsQ1/Q2 affected the biosynthesis of ACT, RED and CDA through positive control of the pathway-specific regulatory genes actII-ORF4, redD and cdaR respectively (Shu et al., 2009). Here, the function of AfsQ1/Q2 on another pathway-specific regulatory gene redZ responsible for RED biosynthesis was also analysed by quantitative real-time RT-PCR (qPCR). It was found that, consistent with the reduced RED production in ΔafsQ1/Q2, redZ expression was downregulated two to threefold throughout the tested time-course, as shown in Fig. 2A.
To determine whether AfsQ1/Q2 directly regulates these pathway-specific regulatory genes which then activate antibiotic biosynthesis, electrophoretic mobility shift assays (EMSAs) were conducted. AfsQ1 was overexpressed as StrepII-tagged protein (Strep-AfsQ1) in Escherichia coli and then purified. The probes containing the corresponding promoter regions of actII-ORF4, cdaR, redD and redZ were tested (Table S3). To verify the specificity of the AfsQ1–DNA interaction, EMSAs with an excess of unlabelled specific or non-specific competitor DNA were used as controls. The results showed that purified Strep-AfsQ1 bound specifically to the promoters of actII-ORF4, cdaR and redZ; however, no shifted band was observed for the redD promoter, even at a very high Strep-AfsQ1 protein concentration (1.6 μM) (Fig. 2B). It therefore can be concluded that the regulation of AfsQ1/Q2 on ACT, CDA and RED biosynthesis is mediated directly through the respective pathway-specific regulatory genes, actII-ORF4, cdaR and redZ.
AfsQ1/Q2 directly regulates the transcription of sigQ, a negative regulator for antibiotic biosynthesis
sigQ encodes a putative sigma factor (Bentley et al., 2002) and is divergently transcribed from afsQ1/Q2, which are clustered with afsQ3 in the S. coelicolor genome (Fig. 3A). afsQ1, afsQ2 and afsQ3 feature overlapping start and stop codons (GTGA and A), suggestive of a putative operon, which was confirmed by RT-PCR (data not shown). Our previous work showed that sigQ and afsQ1/Q2 work together to regulate the production of ACT, RED and CDA, and the transcription of sigQ is positively controlled by AfsQ1/Q2 (Shu et al., 2009). To determine whether AfsQ1 activates the transcription of sigQ directly, EMSAs were conducted with Strep-AfsQ1 and the probe (sigQ_afsQ1_int, −349 to −1 nt relative to translational start site) covering the intergenic region (349 bp) between sigQ and afsQ1 (Table S3, Fig. 3A). As shown in Fig. 3B, Strep-AfsQ1 interacted specifically with the probe sigQ_afsQ1_int. It was found that, at lower protein concentrations (0.02–0.05 μM), two shifted bands were observed in the gels, and by increasing the Strep-AfsQ1 concentration to 0.2 μM, an increase in the abundance of the upper band accompanied with the disappearance of the lower band was seen. The results indicated that AfsQ1 activates sigQ transcription possibly via direct interaction with the intergenic region between afsQ1 and sigQ.
To check whether AfsQ1 was autoregulated by binding to the intergenic region (sigQ_afsQ1_int), ΔafsQ1 with an in-frame deletion of 561 nt (from 55 to 615 nt) within afsQ1 open reading frame was constructed and the expression of afsQ2 and afsQ3 was determined using qPCR. As controls, the transcription of afsQ1 and sigQ was also checked. Little difference in the transcription of afsQ2 and afsQ3 was found between M145 and ΔafsQ1 across the time-course we examined (Fig. 3C). In contrast, in the ΔafsQ1 mutant, sigQ transcription was drastically reduced and as a negative control, no afsQ1 transcript was detected (Fig. 3C). These results demonstrated that AfsQ1 regulates only the transcription of sigQ, but not that of the afsQ1/Q2/Q3 operon.
Identification of the precise AfsQ1 binding sites for regulation
To precisely determine the AfsQ1 binding sites, DNase I footprinting experiments were performed using [γ-32P]-labelled probes. The DNA probes harbouring the respective upstream regions of two AfsQ1 target genes (redZ and sigQ) were chosen for DNase I footprinting assays. The results revealed a protected 40-nt stretch in the upstream region of redZ in the sense strand, extending from nucleotide −213 to −174 nt with respect to the first nucleotide of the translational start site of redZ (Fig. 4A and C). The protection area in the intergenic region between afsQ1 and sigQ covered the nucleotides from −242 to −209 nt relative to the first nucleotide of the sigQ translational start site (Fig. 4B and C).
The identification of two protection regions allowed us to determine the putative consensus sequence for AfsQ1 binding. Comparison of the protected regions revealed the presence of a 16-nt signature sequence comprising two 5-nt direct repeats separated by 6-nt, GAAAC-n6-GTATC and GTGAC-n6-GTGAC in the upstream regions of redZ and sigQ, respectively, which might function as the AfsQ1 binding sites (Fig. 4C).
To assess the importance of the signature sequence for AfsQ1 binding, the two 5-nt repeats, the intervening 6-nt spacer and also the flanking regions were respectively mutated as illustrated in Table S4. The binding activities of AfsQ1 with the mutated sequences were determined by EMSAs (Fig. 5A and B). The original DNA probes were used as positive controls. The results showed that the affinity of AfsQ1 to the mutated sites sigQmutC, sigQmutD and sigQmutE was abolished completely, whereas strong shifted signals can still be observed between AfsQ1 and the mutated sites sigQmutA and sigQmutB, respectively, suggesting that both of 5-nt direct repeats are crucial for the interaction with AfsQ1, while the 6-nt spacer and the flanking sequences are dispensable for AfsQ1 binding (Fig. 5A). For the protected sequence in the redZ promoter region, we found that, similar to that for sigQ promoter region, mutations in the 6-nt spacer and the flanking regions (redZmutA and redZmutB) have little effect on the affinity of AfsQ1 binding; however, mutations in both 5-nt direct repeats or the second repeat (5′-GTATC-3′) completely demolished the specific AfsQ1 binding (Fig. 5B, redZmutC and redZmutE). Interestingly, when only the first 5-nt repeat (5′-GAAAC-3′) was mutated, strong shifted bands can still be seen; however, the pattern of shifted bands changed as compared with the original probe (Fig. 5B, redZmutD), implying that only the second 5-nt sequence (5′-GTATC-3′) is indispensable for AfsQ1 binding and the initial binding of AfsQ1 to the second repeat could lead to its cooperative binding to the first 5-nt repeat.
Normally, transcription factors recognize similar DNA motifs in the promoter regions of different targets, which prompted us to suggest that the promoter regions of the other two AfsQ1 targets, cdaR and actII-ORF4 should also have the 16-nt AfsQ1 binding sites identified above. Inspection of the nucleotide sequences of the promoter regions of cdaR and actII-ORF4 revealed the presence of several putative AfsQ1 binding sites (two conserved 5-nt sequences with a spacer of six variable nucleotides) (Table S4). To evaluate whether these DNA motifs are indeed AfsQ1 binding sites, we introduced mutations (5′-TGGAG-3′) into both of the 5-nt sequences of the DNA motifs and EMSAs were carried out (Table S4). As shown in Fig. 5C, AfsQ1 bound strongly to the probe containing the wild-type cdaR promoter region and to the mutated probe cdaRmutB, in which both of 5-nt sequences located at −24 to −9 nt relative to translational start site were mutated. In contrast, the affinity of AfsQ1 to cdaRmutA (in which two 5-nt located at −105 to −90 nt were mutated) was completely abolished, suggesting that the 16-nt motif (−105 to −90 nt) is responsible for the AfsQ1 binding. For the region upstream of actII-ORF4, surprisingly, mutations of three potential AfsQ1 target sequence (actmutA, actmutB and actmutC) have little effect on the affinity of AfsQ1 binding in comparison with the wild-type sequence (Fig. S2A, Table S3), indicating that these three 16-nt motifs are dispensable for AfsQ1 binding.
In order to pin down the AfsQ1 recognition motif upstream of actII-ORF4, three small overlapping fragments covering the original probe used above (extending from −170 to +70 nt with respect to the transcriptional start point) were generated and EMSAs were performed. As shown in Fig. S2B, AfsQ1 bound to the fragments actF1 and actF2 that have a 45-nt overlap, but not to the actF3 fragment. These results suggested that the AfsQ1 binding site in the actII-ORF4 promoter region is probably located in the overlapping region (45-nt) of actF1 and actF2. As described above, the 16-nt sequence (5′-GGAGATCGCTTGTGAC-3′, −63 to −48 nt relative to the transcriptional start point) at the 3′ end of the overlapping region was shown to be not required for AfsQ1 binding (actmutA). Thus, the AfsQ1 binding site in actII-ORF4 promoter region was narrowed down to the 29-nt sequence (−92 to −64 nt relative to the transcriptional start point). To prove this 29-nt sequence is required for AfsQ1 binding, EMSAs were carried out using a Cy5-labelled synthetic probe of the 29-nt sequence (actF4). As expected, AfsQ1 bound to actF4 as that of the original full-length probe (Fig. S2B).
To further determine the precise AfsQ1 binding site upstream of actII-ORF4, DNase I footprinting assay was performed with the 5′-[γ-32P]-labelled DNA fragments containing the upstream region of actII-ORF4 (−170 to +70 nt relative to the transcriptional start point). The analysis revealed two protected regions separated by 12 nt, which are located between −67 and −62, and between −87 and −80 nt relative to the transcriptional start point respectively (Fig. S2C). The protection regions coincided well with the results generated by the size-down assay described above. These results clearly demonstrated that the binding site of AfsQ1 in the promoter region of actII-ORF4 lacks a typical identified AfsQ1 binding motif, revealing the diversity of the binding motifs needed for AfsQ1 regulation.
In vivo verification of AfsQ1 binding site upstream of sigQ by mutagenesis
As the mutation of the AfsQ1 binding site was severely impaired in the affinity of AfsQ1 binding in vitro, it is suggested a corresponding role in vivo. To test this hypothesis, we constructed a S. coelicolor mutant PSS carrying the mutation as that of sigQmutC in the sigQ promoter region through homologous recombination. The strain PSS and M145 were grown on MM supplemented with 75 mM Glu and tested for antibiotic production as well as transcript level of sigQ. As expected, the mutation in the sigQ promoter region resulted in remarkably decreased sigQ expression (Fig. 6B), and also increased ACT and RED production compared to the parental strain M145 (Fig. 6A), in good agreement with the phenotype upon sigQ deletion as we reported previously (Shu et al., 2009).
To exclude the possibility that the effect of the mutation in PSS inactivates the binding site for RNA polymerase instead of the AfsQ1 binding site, the promoter structure of sigQ, including the transcriptional start point and −35, −10 regions, was analysed using high-resolution S1 nuclease protection assay. S1-mapping result showed that sigQ has a single putative transcriptional start point, which is localized at the nucleotide T or C at position of 178 or 177 nt with respect to the first nucleotide of the translational start site (ATG) (Fig. S1). Two hexameric sequences (5′-CGGACA-3′ and 5′-CAGGCT-3′) separated by 20 nt were found upstream of the identified putative transcriptional start point, similar to the consensus −35 and −10 regions reported for the Streptomyces promoters (Strohl, 1992). As the AfsQ1 binding site is situated 6 nt upstream of the −35 region of the sigQ promoter, it is possible that the binding of AfsQ1 will help recruit RNA polymerase and thus enhance the transcriptional level of sigQ. It was therefore concluded that the two 5-nt direct repeats upstream of sigQ indeed functioned as the AfsQ1 binding site in vivo.
Prediction and verification of the AfsQ1 regulon in S. coelicolor
The 16-nt sequences of the identified AfsQ1 binding sites upstream of sigQ, redZ and cdaR were used to build a position weight matrix (Table S5) and to scan the S. coelicolor genome with PREDetector program (Hiard et al., 2007). The regulation bounds were set by default from −350 to +20 nt relative to the first nucleotide of gene translational start site. One hundred and seventy-two putative AfsQ1 binding sites were identified with the cut-off score above 7.6 (Table S6). Among them, 23 putative targets were selected based on two criteria as follows and validated experimentally by EMSAs: (i) candidate AfsQ1 binding sites should contain relatively conserved cis-element of GT/AnAC-n6-GtnAC; (ii) genes with upstream regions containing a putative AfsQ1 binding site should be well annotated. The EMSAs analysis showed that AfsQ1 interacted specifically with 17 of the 23 putative binding sits (Fig. 7A).
The newly identified AfsQ1 binding sites were found located in the intergenic regions between the following three sets of two divergently transcribed genes, including cpkA/cpkD (SCO6275/SCO6276, encoding a putative type I polyketide synthase and a possible secreted protein respectively), leuA/SCO2529 (SCO2528/SCO2529, encoding a 2-isopropylmalate synthase and a possible metalloprotease respectively) and bldM/whiD (SCO4768/SCO4767, encoding two regulators associated with aerial mycelium formation and sporulation), and also in the upstream regions of abrC3 (SCO4596, encoding a TCS response regulator, involved in regulation of antibiotic production and development), xysA (SCO0674, encoding a secreted endo-1,4-beta-xylanase), gap1 (SCO1947, encoding a glyceraldehyde-3-phosphate dehydrogenase), SCO2978 (encoding a putative ABC transporter), pstS (SCO4142, encoding a phosphate-binding protein precursor), amfC (SCO4184, encoding an aerial mycelium-associated protein) and other genes with unknown function (Table S6). Comparison of all identified AfsQ1 binding sites mentioned above allowed the identification of a consensus motif GTnAC-n6-GTnAC by WebLogo (Fig. 7B).
To study the effect of AfsQ1/Q2 on the expression of the newly identified targets, qPCR was performed using RNA from the parental strain M145 and the ΔafsQ1/Q2 mutant grown on the MM supplemented with 75 mM Glu at two time points (36 and 48 h). As shown in Fig. 7C, transcription of SCO2978, pstS, abrC3, cpkA, cpkD and bldM was clearly reduced, while the enhanced expression of whiD and SCO2529 was detected in the ΔafsQ1/Q2 mutant compared with M145, indicating that AfsQ1/Q2 has a dual activator/repressor function. Similar dual function has also been described for other response regulators in S. coelicolor, such as GlnR and PhoP (Sola-Landa et al., 2005; Tiffert et al., 2008; Rodriguez-Garcia et al., 2009). For the transcription of gap1, amfC and leuA, no significant difference between M145 and ΔafsQ1/Q2 was observed. No transcription of xysA was detected in M145 or ΔafsQ1/Q2 (data not shown).
Interestingly, AfsQ1 target genes cpkA and cpkD are located within a type I polyketide synthase gene cluster (cpk, SCO6273-SCO6288) responsible for the biosynthesis of a yellow pigment named as yCPK, whose production was enhanced on Glu-supplemented medium, prompting us to propose that AfsQ1/Q2 plays a role in the regulation of yCPK biosynthesis. To verify this possibility, yCPK production was compared between M145 and ΔafsQ1/Q2 grown on the MM supplemented with 75 mM Glu at 36 h. As expected, deletion of afsQ1/Q2 led to substantial reduction in yCPK production (Fig. S3), consistent with the decreased transcription of cpkA and cpkD in the ΔafsQ1/Q2 mutant.
AfsQ1 functions as a transcriptional repressor of nitrogen assimilation
AfsQ1/Q2 function only when cells are grown on the MM supplemented with 75 mM Glu as the sole nitrogen source, and the identified AfsQ1 binding sites are very similar to two copies of ‘a-site’ (GTnAC) of the previously reported GlnR box (Tiffert et al., 2008). In addition, computational analysis using PREDetector (described above) identified two putative AfsQ1 binding sites located upstream of glnA and amtB genes, encoding glutamine synthetase I and ammonium transporter respectively (Table S6). We therefore speculated that AfsQ1/Q2 might also be involved in the regulation of nitrogen metabolism in S. coelicolor. To address this possibility, as an initial approach, EMSAs were performed to determine the in vitro interaction between Strep-AfsQ1 protein and the respective promoter regions of seven nitrogen assimilatory genes (including glnA, glnII, gdhA, amtB, ureA, nirB and nasA) as well as two nitrogen regulator genes, glnR and glnRII. The specificity of AfsQ1–DNA interaction was checked by a competition assay with an excess of unlabelled specific and non-specific competitors. The results revealed that Strep-AfsQ1 bound specifically to the promoter regions of seven structural genes (Fig. 8A), but not to those of glnR and glnRII under the same reaction conditions, or even when a very high concentration of Strep-AfsQ1 protein (1.6 μM) was used (data not shown).
To determine the effect of AfsQ1/Q2 on the transcription of seven target genes involved in nitrogen assimilation, qPCR was performed using RNA isolated from the parental strain M145 and the ΔafsQ1/Q2 mutant grown on the MM medium supplemented with 75 mM Glu at different time points (24, 36, 48 and 60 h). It was found that deletion of afsQ1/Q2 led to higher transcription of glnA, glnII, amtB and ureA compared with M145 at some time points (over twofold) as shown in Fig. 8B, indicating that AfsQ1/Q2 acts as a transcriptional repressor of these nitrogen metabolism genes. For gdhA expression, no significant difference between M145 and ΔafsQ1/Q2 was observed (Fig. 8B). We did not detect the transcription of nirB and nasA in either strain under the condition we employed (data not shown).
AfsQ1 and GlnR compete for the promoter regions of nitrogen assimilatory genes
Because of the similarity between the binding sites of AfsQ1 and GlnR, we tested whether AfsQ1 binding to the promoter regions of nitrogen metabolism genes interferes with GlnR binding. Strep-AfsQ1 and His6-tagged GlnR (His-GlnR) were applied together and separately in EMSAs, respectively, with either glnA or nirB promoter region. As shown in Fig. 9A, when Strep-AfsQ1 and His-GlnR were used separately (glnA, lane 8 and lane 2), both of them bound to the Cy5-labelled glnA promoter, forming distinct protein–DNA complexes. When keeping the concentration of His-GlnR (0.2 μM) constant (glnA, lanes 2 to 5), increasing the concentration of Strep-AfsQ1 (0.1–0.4 μM) resulted in the appearance of AfsQ1–DNA complexes accompanied by decreased GlnR–DNA complex formation. When an equal concentration (0.2 μM) of both proteins was used, the GlnR–DNA complexes predominated over AfsQ1–DNA (glnA, lane 4). However, if a 10-fold molar excess of Strep-AfsQ1 (0.4 μM) over His-GlnR (0.04 μM) was applied, binding of His-GlnR to the glnA promoter was greatly affected, resulting in formation of more AfsQ1–DNA than GlnR–DNA complexes (glnA, lane 7). A similar trend was observed in the competitive EMSAs with the nirB promoter region. However, the binding activity of AfsQ1 with the nirB promoter is higher than that of GlnR, as a molar ratio of 1:4 (Strep-AfsQ1 to His-GlnR) applied simultaneously in the binding reaction resulted in nearly equal molar DNA–protein complexes (nirB, lane 4). The competitive EMSAs clearly demonstrated that AfsQ1 and GlnR competed for the occupancy of glnA and nirB promoter regions in vitro, implying that AfsQ1 might function as a repressor for the expression of nitrogen metabolism genes at the transcriptional level by blocking GlnR activation, at least for glnA.
To determine the precise AfsQ1 binding sites in the glnA and nirB promoter regions, the four 5-nt sequences bound by GlnR were replaced by random nucleotides (TGGAG) (Table S4), and EMSAs of Strep-AfsQ1 with the mutated promoter regions of glnA and nirB were performed. The affinity of AfsQ1 to the mutated sites glnAmutB and glnAmutC was completely abolished in comparison with the original target, whereas mutation of the first and fourth 5-nt sequences (glnAmutA and glnAmutD) had little effect on AfsQ1 binding affinity, but resulted in the disappearance of the upper shifted band, revealing that the central two 5-nt sites (GAAAC-n6-GTCAC) bound by GlnR are required for AfsQ1 binding, and two boundary 5-nt sites may be bound by AfsQ1 in a cooperative manner (Fig. 9B). For the four mutations in the promoter regions of nirB, AfsQ1 cannot bind to nirBmutB, nirBmutC, nirBmutD except for nirBmutA under the identical reaction condition, demonstrating that the last three 5-nt sites of GlnR binding box in the nirB promoter are indispensable for AfsQ1 binding (Fig. 9C). These results indicated that the AfsQ1 binding site overlaps with the GlnR box in the promoter regions of glnA and nirB. Moreover, complex AfsQ1 binding sites were also revealed, which consist of two to four conserved 5-nt direct repeats with 6-nt as a spacer.
Molecular mechanism underlying the function of AfsQ1/Q2 on secondary metabolism in S. coelicolor
In the present work, we described the dissection of the molecular basis of AfsQ1/Q2 on antibiotic biosynthesis in S. coelicolor. It was revealed that AfsQ1/Q2 activates ACT, RED and CDA biosynthesis directly through actII-ORF4, cdaR and redZ respectively. The precise AfsQ1 binding sequences in the promoter regions of these three genes were determined. A characteristic AfsQ1 binding motif (with a length of 16 nt) comprising two 5-nt direct repeats separated by six variable nucleotides was identified in the respective upstream regions of cdaR and redZ. However, the AfsQ1 binding sequence in actII-ORF4 promoter region lacks similar binding motif, indicating that there exists other binding mode for AfsQ1 regulation, which yet to be discovered in the future. The diversity of the binding motifs for AfsQ1 regulation was also described for other TCS response regulator. Sola-Landa et al. (2008) identified three types of PhoP binding motifs which are composed of two to six 11-nt direct repeat units (DRus). The most complex binding motif (Class III) contains six DRus with poor conservation (Sola-Landa et al., 2008).
Besides AfsQ1, several other activators as well as repressors have also been identified involved in regulation of ACT, RED and CDA production through the pathway-specific activator genes in S. coelicolor (Uguru et al., 2005; McKenzie and Nodwell, 2007; Rigali et al., 2008). For instance, AbsA2, the response regulator paired with histidine kinase AbsA1, represses the biosynthesis of ACT, RED and CDA by directly interfering with the expression of actII-ORF4, redZ and cdaR, respectively, in a phosphorylation-dependent manner (McKenzie and Nodwell, 2007). However, the precise binding site for AbsA2 in these promoter regions is still unclear. DasR, a GntR-family member, is another key regulator of antibiotic biosynthesis, which can bind to the dre site (DasR binding box: 5′-TGGTCTAGACCA-3′) presented in the promoters of actII-ORF4 and redZ to repress their transcription (Colson et al., 2007; Rigali et al., 2008). AtrA, a TetR-family transcriptional regulator, functions as an activator of ACT production through a direct interaction with the regions flanking the actII-ORF4 promoter and two binding sites of AtrA were identified within the coding region of actII-ORF3 and actII-ORF4 respectively (Uguru et al., 2005). Interestingly, the identified AfsQ1 binding site (5′-GAAAC-N6-GTATC-3′) upstream of redZ gene overlaps with the DasR binding dre site. However, the AfsQ1 binding sequence in the actII-ORF4 promoter region is located separately from all of the binding sites identified so far. It will be interesting to decipher how AfsQ1/Q2 and other global regulators interact with each other in regulation of the biosynthesis of these antibiotics in S. coelicolor.
AfsQ1/Q2 might also play a positive role in yCPK biosynthesis, a yellow-pigmented secondary metabolite encoded by cpk gene cluster (SCO6273-SCO6288). Intriguingly, the function of AfsQ1/Q2 on yCPK production may be achieved through directly regulating the expression of two structural genes, including cpkA encoding a putative type I polyketide synthase subunit and cpkD encoding a putative secreted protein but not through pathway-specific activator gene kasO (data not shown). It was previously reported that there are other regulators involved in the control of cpk genes in S. coelicolor. For instance, we showed that DraR-K repressed the transcription of cpk genes via interaction with the promoter region of the pathway-specific activator gene kasO (Yu et al., 2012). Similarly, two γ-butyrolactone receptors ScbR and ScbR2 were identified to function as repressors directly via regulating the transcriptional level of kasO (Takano et al., 2005; Xu et al., 2010). In addition, intriguingly PhoP/R has recently been found to enhance cpk transcription possibly via binding to three specific regions internal to two polyketide synthase genes (cpkB and cpkC) in cpk gene cluster (Allenby et al., 2012). Moreover, it was also reported that DasR and ArgR may influence the transcription of cpk genes, although their exact mechanism is still to be determined (Rigali et al., 2008; Perez-Redondo et al., 2012). The data described above clearly reveal the complicated regulatory network of yCPK biosynthesis in S. coelicolor, which allows the bacteria to integrate different environmental stress signals in the regulation of yCPK biosynthesis.
One of the AfsQ1 target genes, abrC3, encodes a response regulator of an atypical TCS SCO4596/4597/4598 (AbrC1/C2/C3) consisting of two histidine kinases and a response regulator that was reported to have a positive role in both antibiotic production and morphological differentiation (Yepes et al., 2011). AfsQ1 positively regulates its expression, suggesting that the function of AfsQ1/Q2 might be partly mediated by AbrC1/C2/C3. The role of AbrC1/C2/C3 in AfsQ1/Q2-mediated signal transduction system is yet to be determined.
AfsQ1/Q2 has a pleiotropic role in primary metabolism
By genome screening using the known AfsQ1 binding sites upstream of sigQ, redZ and cdaR, in combination with EMSAs and transcriptional analysis, we identified a new set of genes within the AfsQ1/Q2 regulon. Functional analysis of these target genes pointed to the new roles of AfsQ1/Q2 in carbon, nitrogen and phosphate metabolism in S. coelicolor.
Three putative genes involved in carbon metabolism were identified as AfsQ1 direct targets, xysA (SCO0674), gap1 (SCO1947) and SCO2978. However, inactivation of afsQ1/Q2 resulted in the reduced expression of only SCO2978 that encodes a putative ABC transporter sugar binding protein, but did not affect the transcription of xysA and gap1. Nevertheless, binding of AfsQ1 to the presumptive promoter regions of these three carbon metabolism genes suggests a connection of AfsQ1/Q2 with carbon metabolism in S. coelicolor.
As reported previously, genes involved in nitrogen assimilation in S. coelicolor, including amtB, glnA, glnII, gdhA, ureA, nirB and nasA, are governed by the central nitrogen regulator GlnR under nitrogen-limited condition (Tiffert et al., 2008; Wang and Zhao, 2009). In addition, it was discovered that TCS PhoP/R exerts a negative control on the transcription of glnR, glnA, glnII and amtB upon phosphate limitation (Rodriguez-Garcia et al., 2009). Here, we revealed that AfsQ1 can bind specifically to the promoter regions of these seven nitrogen assimilatory genes and repress the transcription of amtB, glnA, glnII and ureA under the condition of MM supplemented with 75 mM Glu. This finding complements well the recent study of Nieselt et al. (2010), in which they showed that in a medium with an excess of glutamate, genes for nitrogen assimilation in S. coelicolor are repressed by other regulators or mechanisms other than that mediated via PhoR/P. Although nearly no effect on the expression of gdhA upon afsQ1/Q2 deletion, and no transcription of nirB and nasA in both the parental strain and the mutant was detected under the tested condition, we can not rule out the possibility that the effect of AfsQ1/Q2 on these nitrogen metabolism genes might be exerted under other conditions.
Competitive EMSAs revealed that the AfsQ1 binding site overlaps with that of GlnR in the promoter regions of glnA, indicating that AfsQ1 may function as a repressor of glnA transcription by blocking the GlnR activation. Further analysis of the AfsQ1 and GlnR binding sites upstream of glnA and nirB allowed us to distinguish the significant differences in the binding motif between AfsQ1 and GlnR. The sequence containing four 5-nt (a-b-a-b) required for GlnR recognition, only three (b-a-b) in the upstream region of nirB or two (the middle two 5-nt) in glnA promoter region are indispensable for AfsQ1 binding, further demonstrating the diversity of the AfsQ1 binding motif. In addition, we found that cross-regulation between AfsQ1 and GlnR is probably reciprocal, as GlnR could bind to the promoter regions of actII-ORF4, redZ and cdaR respectively (He et al., unpubl. data). Interestingly, we also found that the AfsQ1 binding site upstream of glnA also overlaps with that of PhoP; it is therefore still a possibility that the differential expression of glnA as well as other nitrogen metabolism genes in the ΔafsQ1/Q2 mutant might be partly ascribed to the function of PhoP.
It should be noted that regulation of nitrogen metabolism is quite complex; besides GlnR, AfsQ1/Q2 and PhoP/R, NnaR and AfsR are also involved (Rodriguez-Garcia et al., 2009; Amin et al., 2012; Santos-Beneit et al., 2012). NnaR, a new GlnR target, was identified recently as a new player in nitrogen metabolism. Four nitrate/nitrite assimilation genes, narK, nirB, nirA and nasA, were subject to direct regulation of NnaR, and a cooperative binding with GlnR to the nirB promoter was identified (Amin et al., 2012). In addition, AfsR, the global regulator of antibiotic biosynthesis in S. coelicolor (Lee et al., 2002), has recently been identified to bind to the promoter region of glnR and the AfsR binding site overlaps with that of PhoP (Santos-Beneit et al., 2012). In this study, we also showed that AfsQ1 positively controls the expression of pstS, which is a member of the PhoP regulon and also the direct target of AfsR (Santos-Beneit et al., 2009), indicating the possibility of cross-talk between these global regulators in phosphate metabolism. Overall, the cross-regulation in nitrogen and phosphate metabolism (also secondary metabolism) by global regulators, such as GlnR, PhoP, AfsR and AfsQ1, enables microbes to rapidly adapt their metabolism to the changing environments. The interaction between AfsQ1 and the other two regulators (PhoP and AfsR) will be the subject of our future research.
Growth curve analysis revealed that AfsQ1/Q2 has a negative role in S. coelicolor growth; the mutant with inactivation of afsQ1/Q2 accumulated a higher biomass in comparison with the original strain M145. Two possible reasons are: one is that the negative control of AfsQ1/Q2 on nitrogen assimilation relieved in ΔafsQ1/Q2 probably accelerated the primary metabolism; another is that reduced ACT and RED in the mutant would result in increased malonyl-CoA and possibly also acetyl-CoA supplies being available to primary metabolism, such as lipid biosynthesis and ATP production by TCA cycle, which would lead to better growth of the ΔafsQ1/Q2 mutant. However, such a hypothesis still needs further proof.
Plausible mechanism for the function of AfsQ1/Q2 on morphogenesis
Four genes involved in morphological differentiation were identified as the targets of AfsQ1, including bldM, whiD, amfC and SCO2529. whiD is required for the late stages of sporulation in S. coelicolor (Molle et al., 2000). Its transcriptional divergent gene bldM encodes a response regulator required for aerial mycelium formation (Bibb et al., 2000; Molle and Buttner, 2000). SCO2529 encodes for a possible metalloprotease with a peptidase family M4 (thermolysin family) conserved domain. Enzymes of the thermolysin family are secreted by both Gram-positive and Gram-negative bacteria to degrade extracellular proteins and peptides for bacterial nutrition, especially prior to sporulation. Experiments with proteinase inhibitor have revealed the importance of extracellular proteases in sporulation in Bacillus subtilis (Dancer and Mandelstam, 1975). amfC encodes an aerial mycelium-associated protein which is important in development (Kudo et al., 1995). Thus, we suggested that the remarkably increased expression of SCO2529 and whiD, and decreased bldM expression at 48 h (cells enter the stage of sporulation) in the ΔafsQ1/Q2 mutant might account for the rapid spore formation of the ΔafsQ1/Q2 mutant (Shu et al., 2009). Interestingly, bldM is also a member of the BldD regulon (den Hengst et al., 2010). BldD functions to repress its expression during vegetative growth through interaction with a predicted BldD binding site which is separated from the identified AfsQ1 binding site. It was predicted that they may control bldM expression by responding to different signals.
Proposed AfsQ1/Q2-mediated signal transduction system in S. coelicolor
A tentative regulatory model of AfsQ1/Q2 in S. coelicolor is proposed (Fig. 10). Under the MM condition supplemented with high concentration of glutamate as the sole nitrogen source, signals that might be an intermediate of nitrogen metabolism or the ratio of C/N/P ratio could be generated as we proposed previously (Yu et al., 2012). In response to the signals, the sensor histidine kinase AfsQ2 autophosphorylates and then activates its cognate response regulator AfsQ1. Activated AfsQ1 then issues its roles on antibiotic biosynthesis through binding to the promoter regions of its target genes, including actII-ORF4, redZ, cdaR and cpkA/cpkD directly, thereby stimulating the biosynthesis of ACT, RED, CDA and yCPK, respectively, directly controlling the expression of whiD/bldM and SCO2529 to affect aerial mycelium formation and sporulation. As for its role in primary metabolism, we speculated that AfsQ1/Q2 might be involved in the overall co-ordination of S. coelicolor metabolism. Under the MM condition supplemented with 75 mM Glu (with sufficient nitrogen supply), the cells are possibly required to repress the transcription of genes responsible for nitrogen assimilation (such as amtB, glnA and glnII), meanwhile channel more energy and resources (exerted by AfsQ1/Q2) to carbon and phosphate metabolism by activating the expression of genes involved in phosphate and carbon uptake (such as pstS and SCO2978), thus keeping the C/N/P ratio in an equilibrium state.
In conclusion, we report the comprehensive analysis of the AfsQ1/Q2 regulatory system involved in morphogenesis, primary and secondary metabolism and also the identification of AfsQ1/Q2 regulon in S. coelicolor. For the regulon, a conserved AfsQ1 binding motif (GTnAC-n6-GTnAC) comprising two 5-nt direct repeats separated by six variable nucleotides was defined. In addition, cross-regulation between AfsQ1/Q2 and GlnR was established, which further reveals the complex regulatory network of nitrogen metabolism in S. coelicolor. It is worth noting that for the target genes in the AfsQ1/Q2 regulon, not all in vitro AfsQ1 binding events could result in alterations in gene expression. This phenomenon has also been reported for other regulators in S. coelicolor, such as GlnR (Tiffert et al., 2008). There are two possible explanations for this. One is that other transcriptional regulators also influence the expression of these genes and the overall roles may lead to no obvious difference in transcription, and another is the specific experimental condition we employed here which might account for this result. It is likely that in other media, these genes might also show significant differential expression in the ΔafsQ1/Q2 mutant. Our future research will exploit high-throughput technologies, such as chromatin immunoprecipitation (ChIP) with microarray technology (ChIP on chip) or sequencing (ChIP sequencing) (Allenby et al., 2012) to fully identify the direct AfsQ1 targets, which would help to shed more light on the biological significance of the broader roles of AfsQ1/Q2 on morphological and physiological differentiation in S. coelicolor.
Bacterial strains, plasmids and growth conditions
Bacterial strains and plasmids used in this study are listed in Table S1. Primers used are listed in Table S2. S. coelicolor M145 and its derivatives were grown on MS agar at 30°C for spore suspension preparations (Kieser et al., 2000). Solid MM (Kieser et al., 2000) in which ammonium is replaced by 75 mM L-glutamate (Glu) as the sole nitrogen source was used for determination of antibiotic production (ACT and RED) and growth curves as well as for RNA preparation. E. coli strains were cultivated at 37°C in Luria–Bertani medium. If necessary, the media were supplemented with antibiotics (100 μg ml−1 for ampicillin, 50 μg ml−1 for kanamycin, 50 μg ml−1 for apramycin and 50 μg ml−1 for thiostrepton).
Determination of growth curves and antibiotic production
Determination of antibiotic production was performed as described by Kieser et al. (2000). For the determination of growth curves, spores of S. coelicolor M145 and its derivative strains with the same amount (OD450 = 1.0, 100 μl) were grown on MM plates (diameter 7.5 cm) supplemented with 75 mM Glu covered with sterile plastic cellophane and harvested at different time points (24, 36, 48, 60, 84 and 120 h). Cells were dried at 65°C for 2 days.
Construction of the in-frame deletion mutant ΔafsQ1
ΔafsQ1 with an in-frame deletion of 561 nt (from 55 to 615 nt) within afsQ1 open reading frame was constructed by PCR-targeting system as described previously (Datsenko and Wanner, 2000; Gust et al., 2003) with some modifications. Briefly, the disruption cassette was amplified using primers afsQ1-tar-fw and afsQ1-tar-rev (Table S2) and the plasmid pIJ773loxP (in which two flp sites were replaced by loxP sequence) (Table S1) as template. The amplified cassettes were introduced by electroporation into E. coli BW25113/pIJ790 that harbours the cosmid 3–65 carrying the afsQ1 gene (Table S1), and then apramycin-resistant recombinants were selected. The resulting mutant cosmid was transferred into E. coli ET12567/pUZ8002 and then introduced into M145 by conjugal transfer. The mutants with afsQ1 disruption were selected from exconjugants that were apramycin-resistant and thiostrepton-sensitive. The correct disruption mutants were confirmed by colony PCR using primer pair afsQ1-che-fw and afsQ1-che-rev (Table S2), resulting in ΔafsQ1::acc(3)IV mutants. To remove the apramycin-resistant gene acc(3)IV flanked by loxP sites from the chromosome, a plasmid pALCRE containing the synthetic cre(a) gene (Herrmann et al., 2012) was introduced into ΔafsQ1::acc(3)IV by conjugation. Exconjugants with thiostrepton resistance were selected and replicated on MS agar with or without apramycin. The colonies grown on MS without apramycin, but not on MS with apramycin, were expected mutants, in which the aac(3)IV-oriT cassette should have been removed. The obtained strains were streaked on non-selective MS plates (without thiostrepton) continuously for three passages to remove pALCRE. Finally, the correct in-frame deletion mutant ΔafsQ1 was confirmed by PCR and DNA sequencing.
Mutation of AfsQ1 binding site in the sigQ promoter region in vivo
The mutation of AfsQ1 binding site upstream of the sigQ gene in S. coelicolor genome was performed by gene replacement via homologous recombination. Two homologous arms (440 bp and 588 bp) containing the upstream and downstream regions of the mutation site were amplified from S. coelicolor M145 genomic DNA using the primers sigQu-fw and sigQu-rv (harbouring the reverse complement sequence of the mutation site) (Table S2), sigQd-fw (harbouring the mutation site) and sigQd-rv (Table S2) respectively. The two arms were ligated using overlapping PCR and then inserted between the HindIII and EcoRI sites of the temperature-sensitive plasmid pKC1139 to create pKCsigQ (Table S1). The resulting plasmid pKCsigQ was passed through E. coli ET12567/pUZ8002 and introduced into S. coelicolor M145 by conjugation. Exconjugants with apramycin resistance were selected on MS medium at 30°C. The obtained colonies were streaked on MS plates with apramycin resistance continuously at 37°C for three rounds to create single cross-over mutants. The correct mutants were confirmed by PCR and DNA sequencing, and then picked and grown in non-selective YEME liquid medium at 30°C for three rounds before screening for apramycin-sensitive double cross-overs. The correct mutant strains were confirmed by PCR and DNA sequencing, generating PSS.
RNA preparation and real-time RT-PCR
RNA preparation was performed as described previously (Lu et al., 2007). Fresh spores of M145 and its derivatives were harvested and pre-geminated, and then spread on MM plates (with 75 mM Glu) covered with plastic cellophane and incubated at 30°C. Mycelia were collected at different time points, frozen immediately in liquid nitrogen and ground into powder. RNA isolation was carried out with TrizolTM (Invitrogen) following the procedures recommended by the manufacturer. Contaminating DNA was removed by digestion with DNase I (Takara) and verified by PCR. Reverse transcription of total RNA was performed using the reverse transcriptase kit (Invitrogen) and random hexamers (Takara).
Real-time RT-PCR (qPCR) was conducted using iQTM SYBR Green Supermix (Bio-Rad) with primers listed in Table S2. The reactions were carried out in MyiQ2 two-colour real-time PCR machine, using the following conditions: 95°C for 2 min, followed by 40 cycles of 95°C for 20 s, 60°C for 20 s and 72°C for 20 s. Three PCR replicates were performed in parallel for each transcript. The 16S rRNA was used as an internal control. The relative transcript levels of tested genes were normalized to 16S rRNA and determined using the 2−ΔΔCT method (Livak and Schmittgen, 2001). The values were presented as fold change in comparison with the relative expression levels for each gene at the first test time point (24 or 36 h) in the original strain M145, which was arbitrarily assigned as value 1. Error bars indicate the standard deviation from three independent biological replicates. Data are analysed by t-test, *P < 0.05, significant difference.
Overexpression and purification of AfsQ1 and GlnR proteins
AfsQ1 overexpression in E. coli were carried out according to the method described by Tiffert et al. (2008). The afsQ1 coding region was amplified using primers afsQ1-fw and afsQ1-rv, adding the sequence (5′-TGGAGCCACCCGCAGTTCGAAAAA-3′) encoding an N-terminal StrepII-tag using Taq polymerase (Qiagen), and cloned into the pDRIVE cloning vector (Qiagen). After digestion with NdeI and HindIII, Strep-afsQ1 was transferred into vector pJOE2775 under the control of the Prham promoter, resulting in the recombinant plasmid pSQ1. Protein overexpression in E. coli host strain Rosetta was induced with 0.2% rhamnose overnight at 16°C. Cells were harvested and resuspended with a solution of 100 mM Tris, 150 mM NaCl 1 mM EDTA and 10 mM β-mercaptoethanol, pH 8.0, and then sonicated on ice, followed by centrifugation (1 h, 13 000 g, 4°C) to remove cell debris and membrane fractions from the soluble fraction. Purification of StrepII-tagged proteins (Strep-AfsQ1) from the soluble fraction was performed at 4°C with Strep-Tactin Superflow gravity flow columns (IBA), according to the manufacturer's instructions. His6-tagged GlnR protein (His-GlnR) was purified from E. coli BL21 (DE3) harbouring pEXSCR as described by Wang et al. (2009). The purity of the eluted Strep-AfsQ1 and His-GlnR was checked by 10% SDS-PAGE.
Electrophoretic mobility shift assays
DNA probes containing the putative promoter regions of the tested genes (Table S3) were generated by PCR using the primers listed in Table S2 and labelled at the 5′ end with two different methods using either [γ-32P] ATP or Cy5. For radioactive labelling, PCR product was labelled with [γ-32P] ATP using T4 polynucleotide kinase (NEB). Cy5-labelled probes were obtained via PCR using the primer 5′-AGCCAGTGGCGATAAG-3′, which was labelled with Cy5 at the 5′ end, according to the method described previously (Tiffert et al., 2008). 32P-labelled DNA probes (1000 cpm) or 10 ng of Cy5-labelled probes were incubated individually with varying amounts of Strep-AfsQ1 at 25°C for 20 min in a buffer of 20 mM Tris (pH 7.9), 1 mM dithiothreitol (DTT), 30 mM KCl, 10 mM MgCl2, 0.04 mg ml−1 calf BSA, 5% glycerol and 100 μg ml−1 sonicated salmon sperm DNA (Sangon) (total volume 20 μl). As controls, unlabelled specific probe (200-fold) or non-specific competitor DNA (200-fold, the sonicated salmon sperm DNA) was preincubated with Strep-AfsQ1 for 20 min at 25°C, followed by the addition of labelled probe and incubation for another 20 min at 25°C. In the competitive EMSAs, Strep-AfsQ1 and His-GlnR were applied together and separately with 10 ng of Cy5-labelled probes. The resulting DNA–protein complexes were subjected to electrophoresis on non-denaturing 8% polyacrylamide gels with a running buffer containing 40 mM Tris-HCl (pH 7.8), 20 mM boric acid, and 1 mM EDTA at 140 V and 4°C for 1 h. After electrophoresis, gels were dried and used for autoradiography or directly scanned for fluorescent DNA using a FLA-9000 phosphorimager (Fujifilm).
DNase I footprinting assays
DNase I footprinting assays were performed according to the procedures described by Yang et al. (2007). Briefly, DNA fragments containing the promoter region of redZ (366 bp, −350 to +16 nt relative to translational start site) and the intergenic region between afsQ1 and sigQ (349 bp, −349 to −1 nt relative to translational start site of sigQ) were prepared by PCR with primer pairs listed in Table S2 (redZ-F-fw/redZ-F-rv and sigQ_afsQ1_int-F-fw/sigQ_afsQ1_int-F-rv). Sense primers (redZ-F-fw and sigQ_afsQ1_int_F-fw) were 5′ end-labelled with T4 polynucleotide kinase (NEB) and [γ-32P] ATP, as described by the manufacturer. The footprinting reactions were identical to those used in EMSAs, but in a total volume of 50 μl and approximately 400 000 cpm of the DNA probe was used. After incubation of the mixture at 25°C for 20 min, 5.5 μl of RQ1 RNase-free DNase I buffer and 0.3 U DNase I (Promega) were added, and the mixture was incubated for 75 s. The reaction was stopped by adding 50 μl of stop solution (20 mM EGTA, pH 8.0) and extracted with 100 μl of phenol. After precipitation in ethanol, the pellet was washed with 70% (v/v) ethanol and resuspended in 5 μl of loading buffer [80% formamide, 10 mM NaOH, 1 mM EDTA, 0.1% (w/v) xylene cyanol, 0.1% (w/v) bromophenol blue]. The sequence ladder was prepared using an fmolTM DNA cycle sequencing kit (Promega) with the labelled primers. Samples were separated on a 6% polyacrylamide-urea gel and visualized by autoradiography.
High-resolution S1 nuclease mapping
High-resolution S1 nuclease mapping assay was carried out as described previously (Kieser et al., 2000). Briefly, RNA sample was isolated from cultures of M145 grown on MM plate with 75 mM Glu at 36 h. The hybridization probe for S1 nuclease protection analysis was generated by PCR from M145 genomic DNA using 32P 5′ end-labelled primer sigQ-S1-rv and an unlabelled primer sigQ-S1-fw (Table S2). For the S1 nuclease reaction, 40 μg of RNA was used for hybridization with the labelled probe in NaTCA buffer at 45°C for 15 h. S1 nuclease (Promega) digestions were performed at 37°C for 1 h according to the manufacturer's instructions. The sequence ladder was prepared using an fmolTM DNA cycle sequencing kit (Promega) with the labelled reverse primer (sigQ-S1-rv). Samples were separated on a 6% polyacrylamide-urea gel and visualized by autoradiography.
Mutation of the AfsQ1 binding sits in the upstream regions of sigQ, redZ, cdaR, actII-ORF4, nirB and glnA
Site-directed mutagenesis method was used to generate DNA probes with the mutated sequences for examining the importance of specific nucleotides for AfsQ1 binding. The original upstream regions of sigQ, redZ, cdaR, actII-ORF4, nirB and glnA were cloned into pMD-18T simple vector (Takara) respectively. The mutated DNA fragments were generated by PCR with the primers carrying the mutations (listed in Table S2) and the plasmids harbouring the wild-type DNA fragments as templates respectively. The PCR products were phosphorylated by T4 polynucleotide kinase (Takara), and circularized by ligation using T4 DNA ligase (Takara) and used to transform E. coli DH5α component cells. The resulting mutant plasmids were confirmed by DNA sequencing and used as templates to amplify and label mutated DNA probes for EMSAs.
We wish to thank Prof. Keith F. Chater for providing the PCR-targeting system. This work was supported by National Basic Research Program of China (2011CBA00806 and 2012CB721103), National Natural Science Foundation of China (31121001, 30970033 and 30830002) and Natural Science Foundation of Shanghai (11ZR1442700).