A novel two-component system (TCS) designated as DraR-K (sco3063/sco3062) was identified to be involved in differential regulation of antibiotic biosynthesis in Streptomyces coelicolor. The S. coelicolor mutants with deletion of either or both of draR and draK exhibited significantly reduced actinorhodin (ACT) but increased undecylprodigiosin (RED) production on minimal medium (MM) supplemented separately with high concentration of different nitrogen sources. These mutants also overproduced a yellow-pigmented type I polyketide (yCPK) on MM with glutamate (Glu). It was confirmed that DraR-K activates ACT but represses yCPK production directly through the pathway-specific activator genes actII-ORF4 and kasO, respectively, while its role on RED biosynthesis was independent of pathway-specific activator genes redD/redZ. DNase I footprinting assays revealed that the DNA binding sites for DraR were at −124 to −98 nt and −24 to −1 nt relative to the respective transcription start point of actII-ORF4 and kasO. Comparison of the binding sites allowed the identification of a consensus DraR-binding sequence, 5′-AMAAWYMAKCA-3′ (M: A or C; W: A or T; Y: C or T; K: G or T). By genome screening and gel-retardation assay, 11 new targets of DraR were further identified in the genome of S. coelicolor. Functional analysis of these tentative targets revealed the involvement of DraR-K in primary metabolism. DraR-K homologues are widely spread in different streptomycetes. Interestingly, deletion of draR-Ksav (sav_3481/sav_3480, homologue of draR-K) in the industrial model strain S. avermitilis NRRL-8165 led to similar abnormal antibiotic biosynthesis, showing higher avermectin while slightly decreased oligomycin A production, suggesting that DraR-K-mediated regulation system might be conserved in streptomycetes. This study further reveals the complexity of TCS in regulation of antibiotic biosynthesis in Streptomyces.
Streptomyces produce a vast array of secondary metabolites with important biological activities; many of them have been widely used in medicine and agriculture (Challis and Hopwood, 2003). Streptomyces coelicolor, the best known representative of the Streptomyces genus, harbours a variety of secondary metabolite biosynthetic gene clusters for multiple antibiotics, such as red-pigmented undecylprodigiosin (RED), blue-pigmented polyketide actinorhodin (ACT), the calcium-dependent antibiotic (CDA), and yellow-pigmented type I polyketide (yCPK) (Bentley et al., 2002; Gottelt et al., 2010). Dissecting secondary metabolism and its regulation has been a major focus in the research of S. coelicolor, and so far the information collected has provided valuable clues for better understanding and rational manipulation of other industrially or medically important streptomycetes (Noh et al., 2010; Lu et al., 2011).
Regulation of secondary metabolism involves complex interactions of both pathway-specific and global regulators that trigger or repress expression of antibiotic biosynthetic genes. One of the best characterized pathway-specific regulators are the Streptomycesantibiotic regulatory proteins (SARPs), which normally possess a C-terminal winged helix–turn–helix (wHTH) DNA-binding domain and consequently have the most direct impact on antibiotic production through activating transcription of the biosynthetic genes (Wietzorrek and Bibb, 1997). In S. coelicolor, the SARPs elucidated so far include actII-ORF4 (Arias et al., 1999), redD/redZ (Takano et al., 1992; White and Bibb, 1997), kasO (also called cpkO) (Takano et al., 2005) and cdaR (Ryding et al., 2002), responsible for the biosynthesis of ACT, RED, yCPK and CDA respectively. In the case of global regulators, many of them are members of two-component system (TCS), which is the predominant signal transduction system employed by bacteria to monitor and respond to changing environments (Hakenbeck and Stock, 1996). Bioinformatics analysis of the complete genome sequence of S. coelicolor (Bentley et al., 2002) reveals the presence of 67 typical TCSs, however, to date, only a few of them have been identified to play roles in secondary metabolism, such as phoR-P (Sola-Landa et al., 2003; Santos-Beneit et al., 2009), absA1-A2 (McKenzie and Nodwell, 2007), afsQ1-Q2 (Ishizuka et al., 1992; Shu et al., 2009), rapA1-A2 (Lu et al., 2007), abrA1-A2 and abrC1-C2-C3 (Yepes et al., 2011). PhoR-P is involved in both phosphate metabolism and antibiotic biosynthesis (ACT and RED) in S. lividans (Sola-Landa et al., 2003) and S. coelicolor (Santos-Beneit et al., 2009). Its effects on antibiotic production were not directly mediated by the pathway-specific regulators, but through interaction between PhoP and another important transcriptional activator, AfsR, by competing for the afsS promoter region (Santos-Beneit et al., 2009; Martín and Liras, 2010). AbsA1-A2, located within the CDA biosynthetic gene cluster, was shown to repress the biosynthesis of ACT, RED and CDA, by directly interfering with the expression of actII-ORF4, redZ and cdaR respectively (McKenzie and Nodwell, 2007); however, the AbsA2 binding sites in these promoters remain unknown. AfsQ1-Q2 exhibited positive regulation on antibiotic biosynthesis in a medium-dependent manner (Shu et al., 2009).
To identify more TCSs involved in secondary metabolism and establish further links between global and pathway-specific regulators, we have screened the established S. coelicolor TCS gene-deletion mutant library using MM supplemented with different nitrogen sources. Here, we describe the identification and characterization of a novel TCS, DraR-K (sco3063/sco3062) in S. coelicolor and its homologue DraR-Ksav (sav_3481/sav_3480) in S. avermitilis. Our results show that DraR-K is involved in medium-dependent regulation of secondary metabolism in S. coelicolor and exhibits differential roles in biosynthesis of various antibiotics.
DraR-K plays differential roles in antibiotic biosynthesis in a medium-dependent manner
Bioinformatics analysis revealed that DraK (sco3062) shows all characteristics of a membrane-bound sensor histidine kinase (HK), whereas DraR (sco3063) is predicted to be an OmpR-type response regulator (RR) with a wHTH motif in the C-terminal output domain (Fig. S1). Three different mutants with deletion of either or both of the entire draK and draR ORFs (ΔdraR-K, ΔdraK and ΔdraR) were generated in the model strain S. coelicolor M145 using a PCR-targeting system, and confirmed by colony PCR (Fig. S2) and DNA sequencing (data not shown).
During growth on complex media (MS and R2YE), no obvious phenotypic change was observed for the ΔdraR-K, ΔdraK and ΔdraR mutants when compared with the parental strain M145 (Fig. S3). However, when grown on minimal medium (MM) supplemented separately with different concentration (1, 10 and 75 mM) of various nitrogen sources, such as glutamine (Gln), glutamate (Glu), threonine (Thr), glycine (Gly), KNO3 and (NH4)2SO4 as described in our previous study (Shu et al., 2009), these three mutants exhibited clearly reduced pigment production in the presence of higher concentration of the tested nitrogen sources (especially 75 mM) (Fig. S4A). Interestingly, further antibiotic assays comparing ACT and RED production between M145 and ΔdraR-K revealed that deletion of draR-K resulted in completely opposite effects on these two coloured antibiotics, showing drastically decreased ACT while increased RED biosynthesis, as shown in Fig. S4B. We noticed that with the increasing concentration of nitrogen sources, the mutant phenotype of decreased ACT but enhanced RED production tends to become more and more evident. At the concentration of 75 mM, ΔdraR-K displayed this phenotype on all of the tested nitrogen sources, except that on MM with (NH4)2SO4, there was no RED production in M145 and ΔdraR-K. The most evident phenotype of ΔdraR-K was observed when grown on Gln-based MM (Fig. S4B). ΔdraK and ΔdraR mutants exhibited a similar phenotype as that of ΔdraR-K (data not shown). In addition, for both the parental strain M145 and ΔdraR-K, increasing concentration of nitrogen sources can trigger ACT production under the tested conditions (Fig. S4B), suggesting that higher concentration of nitrogen sources could serve as an environmental or nutritional inducing factor for ACT production in S. coelicolor, consistent with results from other streptomycetes (Aharonowitz, 1980). It should be noted that, when grown on MM supplemented separately with three nitrogen sources [Gln, Thr and (NH4)2SO4] (Figs 1A and S4A), especially on Gln-based MM, S. coelicolor M145 produced only purple pigment, which was considered as intracellular ACT and turned blue under alkaline conditions (Kieser et al., 2000). The spontaneous blue colour produced under other conditions is actually the lactone form of the intracellular ACT, and several genes in the act gene cluster are responsible for the conversion of ACT to the lactone form and its export (Kieser et al., 2000).
We also found that on MM with 75 mM Glu, production of a yellow pigment was obviously enhanced in ΔdraR-K at the early growth stage, especially at 24 h (Fig. 1A), indicating that DraR-K might function as a repressor for its biosynthesis in a glutamate-dependent manner. This yellow pigment was previously determined to be the product of a type I polyketide synthase gene cluster (cpk, sco6273–6288), named as yCPK, and its production was enhanced on Glu-supplemented medium (Gottelt et al., 2010). It was reported that the precursor of yCPK (abCPK), a colourless compound, possesses antibacterial activity and can be converted into yCPK extracellularly (Gottelt et al., 2010). However, the chemical structures of these two CPK compounds are yet to be determined.
Given that the mutants displayed the most evident phenotypic alterations on Gln-based MM (Figs 1A and S4) and the enhancement of yCPK formation in ΔdraR-K was only observable on 75 mM Glu-based MM (Fig. 1A), we chose MM with 75 mM of either Gln or Glu to further investigate the mechanism of DraR-K involved in the biosynthesis of these three antibiotics.
As an initial step, to exclude the possibility that the phenotypic changes were caused by a polar effect, genetic complementation was performed and draR-K genes with the upstream region (903 bp upstream of the translation start site of draR, covering the entire intergenic region between draR and sco3064) were introduced into ΔdraR-K using the integrative vector pSET1521 (Table 1). The results showed that complementation with pSETdraR-K restored the pigment production of ΔdraR-K grown on MM with 75 mM of Gln or Glu (Fig. 1B).
pSET1521 with 2983 bp DNA fragment containing the ORF of draR-K and its putative promoter region
Expression vector, Kanr
draR gene of S. coelicolor M145 cloned in pET28a between EcoRI and HindIII
draRsav gene of S. avermitilis NRRL-8165 cloned in pET28a between EcoRI and HindIII
Routine TA cloning vector, derived from pUC18 vector
A 324 bp DNA fragment containing putative promoter region of actII-ORF4 inserted into pMD 18-T simple
pMD-act with the mutated site actII-ORF4 Ia
pMD-act with the mutated site actII-ORF4 Ib
A 454 bp DNA fragment containing putative promoter region of kasO inserted into pMD 18-T simple
pMD-kasO with the mutated site kasO II
Next, growth and antibiotic production (ACT and RED) of ΔdraR-K (with control vector pSET1521), the parental strain M145 (with pSET1521) and the complemented strain (ΔdraR-K/pSETdraR-K) were quantitatively compared on these two defined MM. As shown in Fig. 1C and D, compared with M145/pSET1521 and the complemented strain, growth of ΔdraR-K was much better throughout the entire time-course on MM with 75 mM Gln; in contrast, ΔdraR-K growth was better than that of the parental and complemented strains when cultured on 75 mM Glu only before 60 h. These data revealed that DraR-K acts as a repressor of S. coelicolor growth under the tested conditions.
ACT and RED formation in M145 has quite different profiles on these two defined MM (Fig. 1C and D). On Gln-based MM, ACT production initiated at late stages (after 60 h) and only low RED level was observed throughout the tested period. In contrast, on Glu-based MM, both ACT and RED synthesis initiated around the transition phase (24–36 h); afterwards, ACT formation remained at a higher level, while RED production declined quickly to a low level.
Deletion of draR-K resulted in severely reduced ACT and markedly enhanced RED production. As shown in Fig. 1C and D, on Gln-based MM, ΔdraR-K produced almost no ACT throughout the entire time-course but produced 3, 13, 7 and 6 times more RED than did the parental strain M145 at 60, 72, 96 and 120 h respectively. However, on Glu-based MM, ACT production at 36, 48, 60, 72, 96 and 120 h were reduced to 11.6%, 20.7%, 35%, 31.6%, 39.8% and 49.2% of the levels in M145, respectively, and RED levels were increased only as a burst at the early stage around 36 h. Complementation of the ΔdraR-K mutant with the plasmid pSETdraR-K restored ACT and RED production.
As the S. coelicolor strains produced both intra- and extracellular ACT when grown on Glu-based MM, a quantitative analysis of ACT production in liquid culture was also performed. Compared with the parental and complemented strains, ΔdraR-K produced little intra- and extra-cellular ACT throughout the entire time-course (Fig. S5), which was consistent with the phenotype changes observed on solid Glu-based MM. The results as described above clearly indicated that DraR-K plays a differential role in the regulation of ACT, RED and yCPK production in a medium-dependent manner. To our knowledge, this is the first identified TCS that functions differentially in antibiotic biosynthesis in S. coelicolor.
To determine whether DraR-K also exerted a role in morphological differentiation in S. coelicolor, scanning electron microscopy (SEM) was conducted. The samples were collected from the S. coelicolor strains, including M145/pSET1521, ΔdraR-K/pSET1521 and the complemented strain, after incubation on Gln or Glu-based MM plates covered with cellophane for 4 days. We found that, in contrast to the parental and complemented strains, which were covered with abundant straight or loosely coiled spore chains, ΔdraR-K formed many abnormal coiled and short spore chains on Gln-based MM. However, on Glu-based MM where S. coelicolor tends to produce non-sporulating aerial hyphae, the ΔdraR-K mutant generated a significant number of aerial hyphae that were short and branched compared with the parental and complemented strains, which harboured long and non-branched aerial hyphae (Fig. S6). These results conclusively demonstrated that DraR-K is also involved in the regulation of morphological differentiation.
Negative autoregulation of DarR-K
The genomic localization of genes draK and draR is shown in Fig. 2A. A short intergenic region (34 bp) between the two genes suggested a possible operon, which was confirmed by reverse transcriptase polymerase chain reaction (RT-PCR) (Fig. S7). To better understand the regulatory mechanism of DraR-K in antibiotic biosynthesis, we analysed the promoter structure of draR using high-resolution S1 nuclease protection assay. The probe used was a 655 bp DNA fragment corresponding to the nucleotide positions ranging from −564 nt to +91 nt relative to the draR translation start site. A single S1-mapping band was obtained, which showed an adenine residue (A) to be the draR transcription start point (Tsp), coinciding with the first nucleotide of the start codon (ATG) of draR (Fig. 2B). Two hexameric sequences (5′-TTGCCC-3′ and 5′-TACCAT-3′) separated by 21 nt (Fig. 2B) were found upstream of the identified putative Tsp, similar to the consensus −35 and −10 region sequences reported for Streptomyces promoters (Strohl, 1992). Considering the promoter-like structure, draR is most likely transcribed as a leaderless mRNA and translated in the absence of a conventional 5′ mRNA leader and ribosome binding site (Jones et al., 1992). Leaderless mRNA transcripts are rare in most bacteria, but a large number of such mRNAs have been identified in Streptomyces (Hong et al., 2004), some of which are TCS genes, such as afsQ1 (Ishizuka et al., 1992), vanR (Hong et al., 2004) and phoR (Sola-Landa et al., 2005) in S. coelicolor.
The fact that two genes are co-transcribed does not exclude the possibility that the downstream gene has its own promoter. Indeed, this is the case for the TCS phoR/phoP of S. lividans that constitutes an operon but phoP is also transcribed from a promoter located within phoR (Ghorbel et al., 2006). Thus, to determine whether the draK gene has its own promoter, S1 nuclease protection assay was also carried out with a probe (423 bp) extending from −245 nt to +178 nt relative to the draK translation start site, and transcription originating from upstream of draR was used as a positive control. As expected, a single mapping band was clearly detected upstream of draR; however, no specific signal was detected with the draK probe (data not shown), suggesting that transcription of draK and draR originates from the same promoter upstream of draR.
Since autoregulation is common for transcriptional regulators in bacteria (Wang et al., 2009), the effect of draR deletion on the transcription of draK was determined by qRT-PCR. As shown in Fig. 2C, draK transcription was increased over twofold at the late stage(s) of the tested time-course when draR was deleted, although with different degrees in the presence of 75 mM Gln or Glu. These results suggested that DraR might function as a negative autoregulator.
To assess whether DraR autoregulates directly, DraR was overexpressed in Escherichia coli as an N-terminal His6-tagged protein (His6-DraR) and purified to near homogeneity by a single chromatography on Ni-NTA resin. Electrophoretic mobility shift assays (EMSAs) were performed with His6-DraR protein using the promoter region of draR (324 bp, containing nucleotides −234 to +90 with respect to the translation start site of draR). However, no retardation of the probe was observed (data not shown), suggesting that DraR might be autoregulated indirectly or that DraR binding to its own promoter involves the phosphorylated form of DraR or/and additional as-yet unidentified factors (Goyal et al., 2011).
DraR-K regulation of ACT and yCPK production mediated by pathway-specific activators
To determine the effect of draR-K deletion on the expression of pathway-specific regulatory genes, qRT-PCR was performed. As shown in Fig. 3A, consistent with decreased ACT biosynthesis in ΔdraR-K relative to M145, transcription of actII-ORF4 in ΔdraR-K was significantly reduced under the tested conditions (either Gln or Glu-based MM); however, the extent of transcriptional alterations in actII-ORF4 was different. On MM with Gln, actII-ORF4 transcription was substantially decreased throughout the tested time-course, suggesting that DraR-K plays a predominant role in the regulation of actII-ORF4 transcription under this condition. In contrast, on MM with Glu, its expression was decreased only at the first tested time point (24 h). Transcription of kasO was significantly enhanced in ΔdraR-K at the early stages (24 and 36 h) only on Glu-based MM compared with M145, which is identical to the phenotype of increased yCPK biosynthesis in the ΔdraR-K mutant. These results clearly suggested that DraR-K regulates ACT and yCPK production via pathway-specific activators actII-ORF4 and kasO respectively.
In qRT-PCR analysis, we also found that, in contrast to higher RED production in ΔdraR-K, there was no obvious increase in the transcription of redD and redZ in ΔdraR-K, as shown in Fig. 3A. Thus, it is speculated that the effect of draR-K deletion on RED production may be independent of redD/Z.
Cooperation of DraR-K with another TCS AfsQ1-Q2 in the control of ACT biosynthesis
As described above, on MM with Glu, actII-ORF4 expression was significantly decreased only at 24 h (Fig. 3A), indicating that transcriptional regulator(s) other than DraR-K may also be involved in the activation of actII-ORF4 expression under the condition of Glu-based MM. In our early study, TCS AfsQ1-Q2 was found to regulate ACT, RED and CDA biosynthesis positively only on Glu-based MM (Shu et al., 2009). The similar pattern of medium-dependent regulation prompted us to speculate that AfsQ1-Q2 and DraR-K may function co-ordinately in the activation of actII-ORF4 transcription under the condition of Glu-based MM. To prove this hypothesis, the double mutant (ΔdraR-afsQ1) with in-frame deletion of the entire ORFs of both afsQ1 and draR was constructed, and transcription of actII-ORF4 was compared between ΔdraR and ΔdraR-afsQ1 using qRT-PCR. The results demonstrated that in contrast to ΔdraR, the actII-ORF4 transcript was significantly decreased in ΔdraR-afsQ1 across the tested time-course (24, 36, 48 and 60 h), as shown in Fig. 3B, indicating that AfsQ1 may be one of the activators that cooperate with DraR in the regulation of actII-ORF4 transcription on Glu-based MM.
Binding of DraR to the upstream regions of actII-ORF4 and kasO
Electrophoretic mobility shift assays (EMSAs) were performed with purified His6-DraR protein to determine whether DraR controls the biosynthesis of ACT and yCPK directly through the pathway-specific activators. In the EMSAs, probes containing the corresponding upstream regions of actII-ORF4 (PactII-ORF4, 324 bp from nucleotide position −183 to +141 relative to the Tsp of actII-ORF4) (Gramajo et al., 1993) and kasO (PkasO, 454 bp from position −332 to +122 with respect to the Tsp of kasO) (Takano et al., 2005) were tested. As shown in Figs 4A and 5A, strong gel-shift of His6-DraR with both probes was observed. These observations suggested that DraR regulates ACT and yCPK biosynthesis directly through interaction with the respective upstream regions of actII-ORF4 and kasO. Up to now, DraR-K is the first TCS characterized in S. coelicolor that regulates yCPK biosynthesis directly through the pathway-specific activator kasO.
Detection of the DraR binding sites
To identify the specific DraR binding sites in the upstream regions of its target genes, DNase I footprinting assay was carried out using 32P-labelled probes of actII-ORF4 (the same with PactII-ORF4) and kasO (a 349 bp DNA fragment from −171 to +178 relative to the Tsp of kasO) respectively. DraR protected a region of the actII-ORF4 probe indicated as site I in Fig. 4B, covering the nucleotides from −124 to −98 nt with respect to the actII-ORF4 Tsp (Fig. 4C). As for the DraR-protected region in the kasO promoter, we found that, as indicative of site II, it extended from −1 to −24 nt with respect to the kasO Tsp, which overlapped with the −10 and −35 regions of the kasO promoter (Takano et al., 2005) (Fig. 5B and C), suggesting that the repression of DraR on the transcription of kasO might be caused by steric hindrance of RNA polymerase binding to the promoter of kasO.
Comparison of the two protected sequences yielded an 11 bp consensus sequence, 5′-AMAAWYMAKCA-3′ (M: A or C; W: A or T; Y: C or T; K: G or T) (Table 2). To assess the importance of the consensus sequence for DraR-binding, random mutations were introduced as illustrated in Table 2, and the binding activity of DraR with the mutated sequences was determined by EMSAs. The results revealed that the binding of DraR to the mutated sites of actII-ORF4-Ia, actII-ORF4-Ib and kasO-II was completely abolished in comparison with their corresponding wild-type sites (Figs 4D and 5D), indicating the importance of these consensus sequences for DraR binding.
Table 2. DraR putative binding sites and their mutants used in EMSAs.
Orientation of DraR binding sites in either of the two promoter regions is indicated by +1 (sense strand) or −1 (antisense strand).
Location in base pair with respect to the transcription start point of actII-ORF4 or kasO.
Identification of new DraR target genes
The 11 bp consensus sequence was applied to in silico screening of the genomic sequence of S. coelicolor M145 using the PREDetector tool (Hiard et al., 2007). The analysis identified 34 upstream regions containing potential DraR binding site(s) with high similarity to the consensus sequence obtained above (Table S3), using the position weight matrix shown in Table S2. In order to determine if the results generated in silico are of biological significance, 22 genes with annotations were selected for EMSA analysis. The probes, containing the upstream regions of the 22 genes (Table S3), were 5′-labelled with Cy5. In subsequent EMSAs, DraR interacted specifically with 11 probes (Fig. 6A). The putative target genes included those encoding a putative enoyl-CoA hydratase (sco6748), a probable aminotransferase (sco6222), a possible peptidase (sco4883), a glutamate synthase gltB (sco2026), a pyruvate kinase (sco2014), a putative sodium: solute symporter (sco3139), a probable ABC transporter ATP-binding subunit (sco5393), a possible ABC transporter transmembrane subunit (sco1147), two putative secreted proteins (sco6198 and sco0072) and a putative integral membrane protein (sco1753).
To investigate the regulatory role of DraR in the expression of these new target genes, 8 target genes were selected for qRT-PCR analysis using RNA isolated from S. coelicolor M145 and ΔdraR-K. The strains were grown on MM supplemented separately with 75 mM of Gln (36, 48, 60 and 72 h) and Glu (24, 36, 48 and 60 h). On Gln-based MM, DraR repressed the transcription of sco6748, sco6222, sco5393 and sco4883, as the transcript levels of these genes were increased over twofold in ΔdraR-K compared with M145 at some of the tested time points. In contrast, the sco1147, sco2014 and sco3139 transcripts were reduced in the ΔdraR-K mutant. For transcription of gltB, no significant difference between M145 and ΔdraR-K was detectable (Fig. 6B). However, on Glu-based MM, draR-K deletion only resulted in enhanced expression of sco6748 and gltB, and had little effect on expression of the other six genes that were tested (Fig. 6B).
draR-K homologues in S. avermitilis play a differential regulatory role in the production of avermectin and oligomycin A
BLAST analysis revealed that draK and draR homologues are widely distributed in streptomycetes. The amino acid sequences of DraK and DraR homologues from six Streptomyces strains available in the database StrepDB (http://streptomyces.org.uk/) were analysed with Bioedit (Fig. S8). We found that DraK, DraR and their homologues display high amino acid sequence identities (DraK 78–82% and DraR 94–98%) among the indicated streptomycetes, including S. coelicolor, S. griseus, S. venezuelae, S. clavuligerus, S. scabies and S. avermitilis. In particular, the wHTH regions of DraR-like proteins necessary for DNA binding are all the same except for a single amino acid difference (E or D) at the first position of wHTH (Fig. S8A), suggesting that the DraR-K-mediated regulation system of antibiotic biosynthesis might be highly conserved among Streptomyces strains.
To test this hypothesis, the function of the DraR-K homologues (DraR-Ksav) in the industrial model strain S. avermitilis NRRL-8165 was investigated and the draR-Ksav disruption mutant (ΔdraR-Ksav) was also generated using the PCR-targeting system. Through flask fermentation, we found that inactivation of draR-Ksav resulted in abnormal antibiotic production, including enhanced avermectin (AVE) B1a while slightly reduced oligomycin A production (olmA), as shown in Fig. 7A. The AVE B1a production in the parental strain S. avermitilis 8165 was too low to be detected in the tested fermentation medium; however, its production in ΔdraR-Ksav was significantly increased up to more than 100 mg l−1 at 5 and 10 days under the same condition. In contrast, olmA production in ΔdraR-Ksav was only slightly decreased when the mutant was cultivated for 10 days. These results confirmed our hypothesis that DraR-K-mediated differential regulation of antibiotic biosynthesis may be conserved in streptomycetes, and draR-K homologues could be useful targets for metabolic engineering of industrially important Streptomyces strains.
To study the differential effects of DraR-Ksav on avermectin and oligomycin A production in more detail, transcript levels of the pathway-specific activator genes olmRI/RII and a polyketide synthase gene olmA4 for oligomycin A biosynthsis (Ōmura et al., 2001; Guo et al., 2010), as well as a pathway-specific regulatory gene (aveR) and a structural gene (aveC) for AVE B1a production (Stutzman-Engwall et al., 2003; Guo et al., 2010), were analysed in the parental strain S. avermitilis 8165 and ΔdraR-Ksav. RNA samples were isolated from cultures cultivated in the fermentation medium for 10 days. qRT-PCR experiments confirmed that DraR-Ksav activated the transcription of olmRI, olmRII and olmA4, as transcript levels of these three genes were significantly reduced in ΔdraR-Ksav compared with the parental strain (Fig. 7B). In contrast, transcription of aveR and aveC were increased by more than twofold in ΔdraR-Ksav relative to S. avermitilis 8165, confirming that DraR-K represses the biosynthesis of AVE B1a (Fig. 7B).
To determine whether DraRsav regulates the biosynthesis of olmA and AVE B1a via direct regulation of pathway-specific activators, EMSAs were subsequently carried out. DraRsav was overexpressed in E. coli as an N-terminal His6-tagged protein (His6-DraRsav) and purified. We found that His6-DraRsav interacted specifically with the upstream region of olmRI (325 bp, containing the nucleotides −266 to +59 with respect to the translation start site of olmRI) (Fig. 7C). However, His6-DraRsav did not bind to the promoter region of olmRII (265bp, containing the nucleotides −231 to +34 with respect to the translation start site of olmRII) or aveR (262 bp, containing the nucleotides −90 to +172 with respect to the transcription start point of aveR). Two 11 bp direct repeats, similar to the consensus binding sequence of DraR, exist in the upstream region of olmRI (Fig. 7D). The results implied that activation of oligomycin A production by DraR-Ksav was mediated directly by the pathway-specific activator gene olmRI, while repression of avermectin production by DraR-Ksav either requires additional factors yet to be identified for binding of DraRsav to the aveR promoter or might be mediated indirectly through aveR.
DraR-K is a distinctive two-component system in Streptomyces
Our results indicate that DraR-K functions as both an activator of ACT biosynthesis and a repressor of yCPK and RED production. The effects on ACT and yCPK production occur at the transcriptional level, by the direct action of DraR-K on pathway-specific activator genes. However, the role of DraR-K in RED production was found to be independent of the pathway-specific activator genes redD/redZ. It is well known that biosynthesis of ACT and RED involves common precursor metabolites, such as acetyl-CoA and malonyl-CoA (Tahlan et al., 2007), thus it is speculated that increased precursor supply might account for increased RED production in the ΔdraR-K mutant. A similar phenomenon had been described in previous reports (Ou et al., 2009). However, whether competition between different biosynthetic pathways for common precursor metabolites plays a role in these effects is still to be determined.
The promoter regions protected by DraR were located at the nucleotide positions from −124 to −98 nt and −24 to −1 nt with respect to the transcription start points of actII-ORF4 and kasO respectively. The positions of the DraR binding sites could account for the dual role of DraR-K on ACT and yCPK biosynthesis, which is in good agreement with the mechanism of regulation by transcription factors (TFs) in bacteria (Rodionov, 2007). As described, the binding sites of repressors are normally positioned between −60 and +60, sterically blocking access of RNA polymerase (RNAP) to the target promoter. In contrast, activators promote transcription by binding to sites located either upstream of or adjacent to the promoter −35 element to recruit RNAP to the promoter (Rodionov, 2007).
The genome sequence of S. coelicolor reveals at least 21 clusters responsible for the biosynthesis of secondary metabolites (Bentley et al., 2002). We are further investigating the effects of DraR-K on the biosynthesis of other secondary metabolites. Our initial transcriptomic analysis showed that deletion of draR-K also resulted in altered transcription of several other gene clusters for secondary metabolite biosynthesis, including increased transcription of gene clusters encoding coelibactin (sco7681–7691) and isorenieratene (sco0185–0191), and reduced expression of clusters encoding desferrioxamines (sco2782–2785) and tetrahydroxynaphthalene (sco1206–1208) respectively (Yu et al., unpubl. data), suggesting that DraR-K functions as a global regulator of secondary metabolism in S. coelicolor.
A similar differential regulatory role of DraR-Ksav in antibiotic biosynthesis (oligomycin and avermectin) in S. avermitilis was identified and the effects occurred both at the transcriptional level, supporting the hypothesis that the DraR-K regulation system for antibiotic production might be highly conserved among streptomycetes.
DraR-K regulation of antibiotic biosynthesis exerted in a medium-dependent manner
The ΔdraR-K mutant exhibits markedly reduced ACT and enhanced RED production only when grown on MM with high concentration of different nitrogen sources, whereas little alteration was observed on R2YE and MS media. There are two possible explanations. First, the regulation of antibiotic biosynthesis involves numerous transcription factors in S. coelicolor, and in most cases these regulators are likely to function together and form a complex regulatory network (Chang et al., 1996; McKenzie and Nodwell, 2007; Lu et al., 2011; van Wezel and McDowall, 2011). As such, under certain circumstances, in the absence of DraR-K function, participation of an alternative regulator(s) is possible, resulting in functional redundancy of DraR-K (Chang and Jaehning, 1997; Simon et al., 2001). However, under MM conditions, other regulators might not be activated due to lack of specific signals; therefore, loss of DraR-K function cannot be compensated, resulting in obvious phenotypic changes of ΔdraR-K in antibiotic production. Second, it is possible that signals sensed by DraK are missed or counteracted by other unknown factors when cells are grown on MS or R2YE medium, but might be formed under the condition of MM supplemented with higher (but not lower) concentrations of nitrogen sources. This kind of signal might be a common intermediate (small molecules) generated during nitrogen metabolism or the ratio change of C/N (Martín et al., 2011) under the stress of higher concentration of nitrogen sources. For instance, acetyl phosphate (acP), an intermediate of the phosphotransacetylase (Pta)-acetate kinase (AckA) pathway, has been found to act as a global signal (McCleary et al., 1993; Wolfe, 2005) and likely contributes to OmpR (RR) phosphorylation together with its cognate HK EnvZ (Park and Forst, 2006; Park et al., 2009a). Hence, further investigations are needed to compare the phosphorylation status of DraK between rich medium (MS, R2YE) and MM with 75 mM Glu or Gln, which might provide clues for the mechanism underlying the medium-dependent antibiotic regulation of DraR-K.
Involvement of DraR-K in primary metabolism
Eleven new DraR target genes in the S. coelicolor genome, including gltB, sco6748, sco6222, sco6198, sco5393, sco4883, sco3139, sco2014, sco1753, sco1147 and sco0072, were identified. Transcriptional analysis revealed that DraR exerts its function on its target genes mainly on Gln-based MM. We found that it apparently repressed the transcription of sco6748 while activated sco2014 expression. sco6748 encodes a putative enoyl-CoA hydratase (ECH) involved in fatty acid elongation and storage lipid biosynthesis (Seubert et al., 1968). sco2014 encodes a pyruvate kinase and is associated with the generation of pyruvate. Thus, it is postulated that in S. coelicolor grown on Gln-based MM, TCS DraR/K might positively control the expression of genes (e.g., sco2014) involved in the generation of pyruvate that can yield either acetyl-CoA or oxaloacetate and negatively control the expression of genes (e.g. sco6748) encoding enzymes involved in storage lipid biosynthesis. Consequently, in the parental strain M145, acetyl-CoA pool generated by the catabolism of Gln will be available for ACT biosynthesis and not for storage lipid biosynthesis. In contrast, in ΔdraR-K, the higher expression of sco6748 might cause increased storage lipid biosynthesis, which would result in less acetyl-CoA being available for ACT formation. This enhanced acetyl-CoA storage as lipids could promote Gln catabolism at a higher rate and thus better growth of the ΔdraR-K mutant as compared with the parental strain in the presence of Gln.
Binding of DraR to the presumptive promoter region upstream of gltB, which encodes a glutamate synthase, suggests a possible involvement of DraR-K in nitrogen metabolism. However, enhanced gltB transcription in ΔdraR-K was only detectable when cells were grown on Glu-based MM. The increased expression of gltB and sco6748 might account for the higher biomass of the ΔdraR-K mutant in the presence of Glu before 60 h.
Interaction of DraR-K and other global regulators in regulating antibiotic biosynthesis
It was revealed in this study that DraR-K co-ordinates with another TCS, AfsQ1-Q2, in the activation of actII-ORF4 transcription during growth on Glu-based MM. The regulation of actII-ORF4 is complex and its promoter is the direct target for several other regulators in addition to DraR and AfsQ1, such as the TetR-family transcription factor AtrA (Uguru et al., 2005), GntR-family global regulator DasR (Rigali et al., 2008), ROK family regulator ROK7B7 (sco6008) (Park et al., 2009b), and TCS response regulator AbsA2 (McKenzie and Nodwell, 2007). The binding sites of AtrA in the upstream region of actII-ORF4 were located at the nucleotide positions from −170 to −152 nt and +75 to +100 nt, while the region protected by DasR spanned the nucleotide positions from −27 to −12 nt (Uguru et al., 2005; Rigali et al., 2008), with respect to the Tsp of actII-ORF4. As the binding sites for both AtrA and DasR in the promoter region of actII-ORF4 did not overlap with that of DraR, it is suggested that they may co-ordinate with each other in the regulation of ACT biosynthesis. One important question to be addressed is how all of these regulators interact with each other, and at the same time ensure a fine-tuned regulation of ACT biosynthesis in response to different nutritional signals.
DraR-K is the first identified TCS that regulates yCPK biosynthesis by interacting with the kasO promoter region. Previously, ScbR, a γ-butyrolactone-binding protein, was shown to repress the transcription of kasO directly via interaction with the operator sites OA (position −35 to −3 nt, relative to the Tsp) and OB (position −222 to −244) (Takano et al., 2005). Interestingly, the DraR binding site (TGCTGGTTTGT) is completely contained within the ScbR site OA in the kasO promoter region (Fig. 5). The overlapping of the binding sites for ScbR and DraR implies that they might compete with each other in the repression of kasO, however, in the presence of both of the sensing signals. It should be noted that regulators other than DraR and ScbR have also been identified to be involved in the control of kasO transcription in S. coelicolor. For instance, DasR was also reported to repress kasO transcription (Rigali et al., 2008); however, no DasR binding site (dre) has been found as yet in the upstream region of kasO and its exact regulatory function remains to be determined. In addition, ScbR2, a homologue of ScbR, encoding ‘pseudo’γ-butyrolactone receptor, has recently been reported to bind to the kasO promoter region and the ScbR2 operator sequence was precisely identified (Xu et al., 2010; Wang et al., 2011). Interestingly, the core of the ScbR2 binding site overlaps completely with that of the ScbR site OB (Takano et al., 2005; Wang et al., 2011). Thus, further investigation is needed to elucidate the possible interactions between these regulators.
A predicted model for DarR-K-mediated regulation of antibiotic biosynthesis in S. coelicolor
A possible working model for signal transduction and differential regulation of antibiotic biosynthesis by DraR-K in S. coelicolor is proposed (Fig. 8). Under the stress of high concentrations of nitrogen sources, signals that might be a common intermediate of nitrogen metabolism or the ratio change of C/N can be generated. Upon receiving the signals, the DraK sensor kinase becomes autophosphorylated at His and this phosphate is then transferred to Asp in the DraR response regulator. Phospho-DraR enhances ACT biosynthesis but represses yCPK biosynthesis by interacting with the promoter regions of the pathway-specific activator genes actII-ORF4 and kasO respectively (Figs 1, 3–5). It is intriguing that DraR-K-mediated repression of RED biosynthesis was independent of the pathway-specific activator genes redD/redZ (Figs 1 and 3A), indicating that its role in RED biosynthesis might be exerted at a non-transcriptional level and ascribed to the abnormal metabolic flux of common precursors shared with other secondary biosynthetic pathways. Under the stress of 75 mM Glu, another TCS, AfsQ1-Q2, could also be activated and cooperate with DraR-K in the activation of ACT production (Fig. 3B). Furthermore, DraR-K is also involved in the repression of S. coelicolor growth (Fig. 1C and D), possibly directly by affecting the transcription of sco6748, sco2014 and gltB, encoding a putative enoyl-CoA hydratase (ECH), a pyruvate kinase and a glutamate synthase respectively (Fig. 6). The differential effects of DraR/K on its target genes involved in primary and secondary metabolism under the conditions of Gln and Glu-based MM might account for the different phenotypes of the parental strain M145 and ΔdraR-K grown on these two defined MM. The mechanism underlying the differential functions of DraR/K under these two different conditions is still to be elucidated.
In summary, we characterized a novel TCS, DraR-K, with differential roles in regulating antibiotic biosynthesis and primary metabolism in S. coelicolor. Our results have established that DraR-K regulates antibiotic production via pathway-specific activators. Our future research goals include determining how DraR-K regulates primary metabolism, identifying the other direct targets of DraR regulation, and addressing whether antibiotic production might also be influenced by changes in primary metabolism. Nevertheless, this study has revealed a novel regulatory system and further exhibited the complexity of the TCS regulatory network in industrially important Streptomyces.
Bacterial strains, plasmids and growth conditions
Bacterial strains and plasmids used in this study are listed in Table 1. Primers are shown in Table S1. S. coelicolor M145, S. avermitilis NRRL-8165 and their derivatives were cultivated on MS agar as described in Kieser et al. (2000) for spore suspension preparation and conjugal transfer. Solid MS, R2YE, and minimal medium (MM) (Kieser et al., 2000) supplemented separately with 1 mM, 10 mM and 75 mM of glutamine (Gln), glutamate (Glu), threonine (Thr), glycine (Gly), KNO3 and (NH4)2SO4 were used for S. coelicolor phenotypic screening. MM supplemented with 75 mM Gln or Glu was used for the quantitative determination of S. coelicolor cell growth and antibiotic production (ACT and RED). For the quantitative analysis of intra- and extracellular ACT production, liquid MM with 75 mM Glu was used. For determination of oligomycin A and avermectin B1a production in S. avermitilis, BioK seed medium and BioK fermentation medium were employed with some modifications (Chen et al., 2009). YEME (Hopwood et al., 1985) was used to grow Streptomyces cells for the genomic DNA extraction. E. coli strains were grown at 37°C in Luria–Bertani (LB) medium. When 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, 200 µg ml−1 for hygromycin B, and 50 µg ml−1 for thiostrepton).
DNA manipulation and conjugal transfer
Plasmids, cosmids and chromosomal DNA preparations were performed according to the standard techniques as described before (Sambrook et al., 1989; Kieser et al., 2000). Plasmids or cosmids with oriT fragment were firstly introduced by transformation into the methylation-deficient E. coli ET12567 (pUZ8002) and then transferred to S. coelicolor or S. avermitilis by intergeneric conjugation as described previously (Kieser et al., 2000).
Construction of gene deletion mutants in S. coelicolor and S. avermitilis
The in-frame deletion mutants of ΔdraR, ΔdraK, ΔdraR-K and ΔdraR/afsQ1 in S. coelicolor M145 were constructed using PCR-targeting system described by Datsenko and Wanner (2000) and Gust et al. (2003) with some modifications. The cosmids 1–71 (carrying the draR-K genes) and 3–65 (harbouring the afsQ1 gene) were constructed in our laboratory using vector pHAQ31 with two antibiotic resistant genes (amp and tsr) and oriT fragment (Table 1). The template pIJloxPΔoriT without oriT fragment and replacement of the flp sites by loxP sequences was modified from pIJ773 as below and used for amplification of disruption cassettes. For construction of pIJloxPΔoriT, the DNA fragment containing the acc(3)IV gene was amplified using primers loxPp1-EcoRI/p2-HindIII (Table S1) and pIJ773 as the template. Then the EcoRI/HindIII fragment was cloned into pIJ773 to replace the original EcoRI/HindIII fragment (1382 bp), obtaining pIJloxPΔoriT. The insert of the plasmid was verified by DNA sequencing.
To construct the in-frame deletion mutants of ΔdraR, ΔdraK and ΔdraR-K (deletion of both or either of the entire draK and draR ORFs), the disruption cassettes were amplified respectively using primers DdraR-fw/rev, DdraK-fw/rev and DdraRK-fw/rev (Table S1), and the plasmid pIJloxPΔoriT as template. Subsequently, the amplified cassettes were introduced respectively by electroporation into E. coli BW25113/pIJ790 harbouring cosmid 1–71 and then apramycin-resistant recombinants were selected. The resulting correctly mutated cosmids were introduced into E. coli ET12567/pUZ8002, and then transferred into S. coelicolor A3(2) M145 by conjugation respectively. The knockout mutations in draR, draK and draR/K were selected from exconjugants that were apramycin resistant and thiostrepton-sensitive. All of the mutants were confirmed by colony PCR using primer pairs JdraR-fw/rev, JdraK-fw/rev and JdraRK-fw/rev (Table S1) respectively. For the removal of the apramycin resistance gene flanked by loxP sites from the chromosome of the mutants, a plasmid pALCRE containing the synthetic cre(a) gene (Herrmann et al., 2012) was individually transferred into ΔdraR::acc(3)IV, ΔdraK::acc(3)IV and ΔdraR-K::acc(3)IV by conjugation. Those exconjugants with thiostrepton resistance were selected and replicated on MS agar in the presence or absence of apramycin. The colonies grown on MS without apramycin, but not on MS with apramycin, were expected mutants, in which aac(3)IV cassette was removed. Finally, in order to remove pALCRE from the mutants, the obtained strains were streaked on non-selective (without thiostrepton) MS agar continuously for three passages. The ΔdraR, ΔdraK and ΔdraR-K mutants with deletion of both or either of the entire draK and draR ORFs were confirmed by colony PCR and DNA sequencing.
The in-frame double deletion mutant ΔdraR-afsQ1 was constructed on the basis of ΔdraR with the method described above using primer DafsQ1-fw/rev and confirmed by colony PCR using primer JafsQ1-fw/rev (Table S1) and DNA sequencing.
The disruption mutant of draR-Ksav (ΔdraR/Ksav) in S. avermitilis with the entire draK and draR ORFs replaced by aac(3)IV-oriT cassette was generated with the method described above. The cosmid 10–22 containing the draR-Ksav genes was constructed in our laboratory using pSuperCos1 vector (Xia et al., 2009) (Table 1). The template pIJloxP for amplification of disruption cassettes was modified from pIJ773, with the replacement of the flp sites by loxP sequences. To generate pIJloxP, the DNA fragment containing the acc(3)IV-oriT was amplified using primers loxPp1-EcoRI/p3-HindIII (Table S1) and pIJ773 as the template. Then the EcoRI/HindIII fragment was cloned into pIJ773 to replace the original EcoRI/HindIII fragment, generating pIJloxP and checked by DNA sequencing. Subsequently, the disruption cassette was amplified using primers DdraRKsav-fw/rev (Table S1). ΔdraR/Ksav mutants were selected from exconjugants that were apramycin resistant and kanamycin-sensitive, and further confirmed by colony PCR using primer JdraRKsav-fw/rev (Table S1) and DNA sequencing.
Complementation of the in-frame deletion mutant of draR-K
The primer pair CdraRK-fw/rev (Table S1) was used to amplify the DNA fragment containing the draR/K ORFs and their upstream region (903 bp, covering the entire intergenic region between draR and sco3064) from the genomic DNA of S. coelicolor M145. The DNA fragment was first cloned in pMD18-T simple vector for DNA sequencing and then inserted into the BamHI site of pSET1521 (Shu et al., 2009). The resulting complementation vector pSETdraR-K was introduced into ΔdraR-K by conjugation. The plasmid pSET1521 was used as a negative control.
Determination of cell growth and antibiotic production
Suspensions of 7 × 107 pregerminated spores from M145/pSET1521, ΔdraR-K/pSET1521 and the complemented strain were grown on MM plates (7 cm-diameter) supplemented with 75 mM Glu or Gln covered with cellophane at 30°C. For dry weight determination, cultures were harvested at the time-course (36, 48, 60, 72 and 84 h) and then dried for 4 h at 80°C. ACT and RED production were determined as described previously (Kieser et al., 2000). The mycelial cells were harvested at five time points (48, 60, 72, 96 and 120 h) from MM agar with 75 mM Gln, and at the time-course of 36, 48, 60, 72, 96 and 120 h from MM agar with 75 mM Glu. For ACT, KOH was added to the samples at a final concentration of 1 M. After 1–2 h incubation at room temperature, the cultures were centrifuged at 13 000 g for 5 min, and then the optical density at 640 nm was determined. For RED, the mycelial pellets that were solubilized in KOH as above were washed by water until the suspension colour is not blue, then followed by extraction with 1 ml methanol (adjusted to pH 2.0) overnight at room temperature. The optical density at 530 nm was determined. Measurements were obtained from three independent cultures and repeated three times.
As for the quantitative analysis of intra- and extracellular ACT production in liquid MM with 75 mM Glu, M145 and its derivatives were incubated at 30°C, 200 r.p.m., for 36, 48, 60, 72 and 84 h. Samples (2 ml) were taken and ACT production in the supernatants (extracellular ACT) and pellets (intracellular ACT) were measured as described above respectively.
Avermectin B1a and oligomycin A production in S. avermitilis were determined by high-pressure liquid chromatography (HPLC) as described previously (Lu et al., 2011). Briefly, spores of different S. avermitilis strains (with the same OD450 ≈ 0.5) were inoculated into 30 ml BioK fermentation seed medium in 250 ml baffled flasks. After incubation for 40 h at 30°C on a rotary shaker at a speed of 200 r.p.m., cultures were trans-inoculated into three parallel 250 ml baffled flasks with 30 ml BioK fermentation medium by 8% inoculation and grown at 30°C, 200 r.p.m. Samples were taken at two time points, 5 and 10 days. For avermectin B1a and oligomycin extraction, 7 ml methanol was added to 3 ml fermentation broth in a 20 ml tube and treated in an ultrasonic cleaner for 20 min. After centrifugation at 8000 g for 10 min, the supernatants were analysed using an Agilent Eclipse XDB-C8 column (4.6 by 150 mm) maintained at 30°C, with a solvent system of methanol-water (85:15, v/v) and a flow rate of 0.9 ml min−1. Commercial oligomycin A (98%, w/w) and avermectin B1a (95.9%, w/w) (Hisun Pharmaceutical Co. Ltd, Zhejiang, China) were used to make standard curves for quantitative determination.
Scanning electron microscopy (SEM)
Suspensions of 1 × 107 pregerminated spores from M145/pSET1521, ΔdraR-K/pSET1521 and the complemented strain were grown on MM supplemented with 75 mM Glu or Gln covered with cellophane at 30°C for 4 days. A piece of cellophane covered with mycelia was cut off and fixed with fresh 2% glutaraldehyde (pH 7.2) and 1% osmium tetroxide for at least 24 h, followed by dehydration in graded ethanol and complete drying in HCP-2 (Hitachi, Japan); coated with gold with a Fine Coater JFC-1600 (JEOL, Tokyo, Japan); and then examined with scanning electron microscopy (JSM-6360LV; JEOL, Tokyo, Japan).
RNA procedures, including RT-PCR and real-time RT-PCR
Suspensions of 7 × 107 pregerminated spores from M145, ΔdraR, ΔdraR-K and ΔdraR-afsQ1 mutants were spread on MM plates with 75 mM Gln or Glu covered with sterile cellophane, and incubated at 30°C. Samples were harvested at 24, 36, 48, 60 and 72 h, and immediately flash frozen and ground into powder in liquid nitrogen. For RNA sample of S. avermitilis NRRL-8165 and ΔdraR-Ksav, the strains were grown in 250 ml baffled flasks with 30 ml BioK fermentation medium for 10 days as the method described for avermectin and oligomycin A production assay. Samples were collected by centrifugation and immediately flash frozen and ground into powder in liquid nitrogen. RNA isolations were performed with Trizol (Invitrogen, USA) following the procedures recommended by the manufacturer. Subsequently, RNA samples were digested with DNase I (Takara, Japan) to eliminate the contamination of chromosomal DNA. Quality and quantity of RNAs were examined by UV spectroscopy and agarose gel electrophoresis.
The RT-PCR was performed using the method described previously (Shu et al., 2009). RT-PCR was used to determine whether draK and draR are cotranscribed using primers RTdraRKco-fw/rev (Table S1). The reaction products were subjected to electrophoresis on 2% agarose in TAE buffer and stained with ethidium bromide (EB).
Quantitative real-time RT-PCR (qRT-PCR) was performed using the method described previously (Lu et al., 2011). The primers used are listed in Table S1. The reactions were carried out in a CFX96 thermal cycler (Bio-Rad, USA) using the following conditions: 95°C for 2 min, followed by 40 cycles of 95°C for 20 s, 60°C 20 s and 72°C 20 s. Three PCR replicates were performed for each transcript, and hrdB (Gottelt et al., 2010) and rpoD gene (Kitani et al., 2009) were used as internal control of S. coelicolor and S. avermitilis respectively. The relative fold change of RNA transcript (mutant/WT) was determined using the 2−ΔΔCt method, in which ΔΔCt = (Cttested gene − CthrdB)mutant − (Cttested gene − CthrdB)WT. The values are averages of two independent qRT-PCR analyses and error bars represent standard deviation.
S1 nuclease mapping
S1 nuclease mapping assay was performed to determine the transcription start point(s) of draR and draK according to the method as described by Kieser et al. (2000). Total RNA samples were isolated from cultures grown on cellophane-covered MM plates with 75 mM Gln as described above. Hybridization probes of draR and draK were prepared by PCR from M145 chromosomal DNA using 5′[γ-32P] end-labelled antisense oligonucleotide primers (S1draR-rev and S1draK-rev) (Table S1) and unlabelled sense oligonucleotides S1draR-fw and S1draK-fw, which anneal to the upstream regions of gene draR and draK respectively. For S1 nuclease reaction, about 40 µg of RNA was used to hybridize with the end-labelled probe in 20 µl NaTCA buffer [3 M NaTCA (sodium trichloracetate, Aldrich, USA), 50 mM PIPES, 5 mM EDTA, pH 7.0] and denatured at 65°C for 15 min, followed by incubation at 45°C for 15 h. S1 nuclease (Promega, USA) digestion was performed at 37°C for 1 h as described by Kieser et al. (2000). The reaction products were separated on 6% (w/v) polyacrylamide/urea sequencing gel, along with sequencing reactions prepared with the labelled primer S1draR-rev using the fmol DNA cycle sequencing kit (Promega, USA).
Overexpression and Purification of His6-DraR and His6-DraRsav protein
The draR of S. coelicolor M145 and draRsav of S. avermitilis NRRL-8165 were amplified respectively using primer pairs EdraR-fw/rev and EdraRsav-fw/rev (Table S1), then inserted into pET28a vector between EcoRI and HindIII sites, generating the recombinant plasmid pETdraR and pETdraRsav. After confirmation by DNA sequencing, pETdraR and pETdraRsav were introduced individually into E. coli BL21 (DE3) for protein expression. E. coli BL21 (DE3) harbouring pETdraR or pETdraRsav was grown in 100 ml LB with 50 mg ml−1 kanamycin at 37°C to an OD600 of about 0.6. Then the cultures were induced by IPTG at a final concentration of 0.5 mM and followed by incubation at 20°C overnight. For the purification of His6-DraR and His6-DraRsav, the cells were harvested by centrifugation at 5000 g, 4°C for 10 min, washed twice with binding buffer [20 mM Tris-HCl, 500 mM KCl, 10 mM imidazole, 10% glycerol (pH 7.9)], re-suspended in 15 ml of the same buffer with Triton X-100 (final concentration 1%) and PMSF (final concentration 0.1 mM). The cell suspension was treated by sonication on ice. After centrifugation (13 000 g for 40 min at 4°C), the supernatant was recovered and loaded on the Ni-NTA agarose chromatography (GE healthcare, Sweden). After extensive washing with buffer [20 mM Tris base, 500 mM KCl, 50 mM imidazole, 10% glycerol (pH 7.9)], His6-DraR or His6-DraRsav was specifically eluted from the resin with 4 ml elution buffer [20 mM Tris base, 500 mM KCl, 500 mM imidazole, 10% glycerol (pH 7.9)] and concentrated to about 3 mg ml−1 by ultrafiltration (Millipore membrane, 10 kDa cut-off size) according to the protocol provided by the manufacturer. The purity of the eluted His6-DraR protein or His6-draRsav was checked on 10% SDS-PAGE.
Electrophoretic mobility shift assays (EMSAs)
The EMSAs were performed as described before (Yang et al., 2007). For EMSAs of the promoter regions of actII-ORF4 and kasO with His6-DraR, the forward primers (up_actII-ORF4-fw and up_kasO-fw1) (Table S1) were labelled with radioactive 32P using T4 polynucleotide kinase (Promega, USA), and then combined with the unlabelled reverse primers (up_actII-ORF4-rev and up_kasO-rev1) (Table S1) to amplify 5′ end 32P-labelled probes using M145 genomic DNA as the template. The PCR products were purified by QIAquick PCR Purification Kit (QIAGEN, Germany).
In the case of mutational analysis of DraR binding sites within actII-ORF4 and kasO promoters, and EMSAs of His6-DraR with the promoter regions of draR (nt −234 to +90 with respect to transcription start site) and 22 predicted new target genes (upstream regions of probes were shown in Table S3), the labelling of probes was performed via PCR using Cy5-labelled primer (5′-AGCCAGTGGCGATAAG-3′) according to the method described by Tiffert et al. (2008). The same Cy5-labelling method was used for EMSAs of His6-DraRsav with the promoter regions of olmRI, olmRII and aveR.
In each EMSA reaction, about 10 ng of Cy5-labelled probes or 1000 c.p.m 32P-labelled probes was incubated individually with various quantities of His6-DraR or His6-DraRsav at 25°C for 30 min in the buffer containing 1 mg ml−1 of poly (dI–dC) (Roche, USA), 20 mM Tris-base (pH 7.5), 1 mM dithiothreitol (DTT), 10 mM MgCl2, 0.5 mg ml−1 calf BSA and 5% glycerol. As controls, unlabelled probes (50-fold, 100-fold of the labelled probes) of specific and non-specific were added respectively. After incubation, protein-bound and free DNA were separated by electrophoresis on non-denaturing 5% polyacrylamide gels with 0.5× TBE running buffer at 10 V cm−1 and 4°C. For 32P-labelled probes, gels were dried and exposed to Biomax radiographic film (Kodak, USA). For Cy5-labelled probe, gels were directly scanned by FujiFilm FLA-9000 (FUJIFILM, Japan). In order to quantify the Cy5-labelled probes, probe DNA concentrations were detected by ultraviolet spectrophotometer Nanodrop 2000 (Thermo Scientific, USA) at the wavelength of 260 nm.
DNase I footprinting assays
DNase I footprinting assays were performed as described previously (Yang et al., 2007). The primer pair up_actII-ORF4-fw/rev and up_kasO-fw/rev2 (Table S1) were used to amplify the promoter regions of actII-ORF4 and kasO respectively. The reaction mixture (50 µl volume) contained about 60 000 c.p.m. of 32P-labelled DNA probe, different amounts of His6-DraR and 20 mM Tris-HCl (pH 7.5), 1 mM DTT, 10 mM MgCl2, 0.5 mg ml−1 calf BSA, 5% (v/v) glycerol. After incubation at 25°C for 30 min, 5.5 µl of RNase-free DNase Buffer and 1 µl of 0.3 U DNase I (Promega, USA) were added to the mixture and incubated for 1 min at room temperature. The reaction was stopped immediately by adding 50 µl of stop solution (20 mM EGTA, pH 8.0), and extracted by 100 µl of phenol/CH3Cl (1:1, v/v) twice. After precipitation in 2–3 volumes of ethanol with the addition of 0.1 volumes of 3 M sodium acetate (pH 4.8) and 2 µl glycogen (Sigma-Aldrich, USA) for 2 h in −20°C, the pellet was washed with 75% (v/v) ethanol and re-suspended in 5 µl of H2O and 5 µl of formamide/dye mixture, followed by electrophoresis on a 6% (w/v) polyacrylamide/urea gel. The labelled sequencing ladders were prepared using the fmol DNA cycle sequencing kit (Promega, USA) with the labelled sense primer for PactII-ORF4 (up_actII-ORF4-fw), and antisense primer for PkasO (up_kasO-rev2) respectively.
Mutational analysis of the DraR-binding sequence
To assess the importance of the 11 bp consensus sequence 5′-AMAAWYMAKCA-3′ for DraR binding, the sites actII-ORF4Ia, actII-ORF4Ib and kasOII were replaced by random nucleotides using primers muactII-ORF4Ia-fw/rev, muactII-ORF4Ib-fw/rev and mukasOII-fw/rev respectively (Table S1). For amplification of the mutated fragments, the DNA fragments of the upstream regions of actII-ORF4 and kasO cloned in pMD18-T simple vector (Takara, Japan) were used as templates. The recombinant vectors (pMD18-actmuIa, pMD18-actmuIb and pMD18-kasOmuII) were verified by DNA sequencing and used as templates for amplifying the 5′ end Cy5-labelled mutated probes. The binding of His6-DraR to the mutated probes was measured by EMSAs.
This work was supported by National Basic Research Program of China (2011CBA00806), National Natural Science Foundation of China (31121001, 30970033 and 30830002), and Natural Science Foundation of Shanghai (11ZR1442700). We are grateful to Professor Keith F. Chater for providing the PCR-targeting system.