Introduction

Microorganisms must modulate their gene expression repertoire in order to adapt to changing environments. One of the predominant signal transduction mechanisms employed by microbes is the phosphotransfer pathway commonly referred to as “two-component” signal transduction systems (TCS), which typically consist of a sensor histidine kinase (HK) and a response regulator (RR) and have been found across all three domains of life, the Bacteria, Archaea, and Eukarya (Hoch 2000). The concept of TCS was initially introduced from extensive studies on the model prokaryotic bacteria, such as Escherichia coli in which 62 two-component proteins have been identified and their regulatory functions are involved in diverse cellular processes, including chemotaxis, osmoregulation, metabolism, and transport (Mizuno 1997, 2005). So far, more than 4,000 TCSs have been detected from 145 sequenced bacterial genomes (Ulrich et al. 2005; Zhang and Shi 2005). Results of recent studies on the model streptomycete, Streptomyces coelicolor, have led us to learn a new scenario as to the versatility of TCSs in mycelial prokaryotes, which are involved in stress-induced differentiation, such as sporulation and secondary metabolism.

S. coelicolor is a genetically well-studied strain in streptomycetes, which differ conspicuously from most other bacteria for their fungal-like developmental cycles and synthesis of multiple antibiotics. The complete genome sequence of S. coelicolor revealed 84 putative sensor HK genes and 80 putative RR genes, among them 67 HK–RR are located adjacently on the chromosome (Bentley et al. 2002; Hutchings et al. 2004). Currently, the functions and possible downstream regulatory targets of these TCSs are largely unknown, although several have been found involved in regulation of very diverse cellular processes such as osmoregulation, metabolism, cell growth, and differentiation in S. coelicolor (Hutchings et al. 2004). Several well-studied TCSs include PhoP-R system mediating phosphate limitation response (Sola-Landa et al. 2003, 2005, 2008; Rodriguez-Garcia et al. 2007), CseB-C and VanR-S systems activating genes involved in integrity of cell envelope when S. coelicolor A3 (2) was exposed to vancomycin and some cell-wall-specific antibiotics (Paget et al. 1999b; Hong et al. 2002; Hutchings et al. 2006a), and TCS systems involved in antibiotic production, such as CutR-S (Chang et al. 1996), RapA1-A2 (Lu et al. 2007), AbsA1-A2 (Adamidis et al. 1990; Brian et al. 1996; Aceti and Champness 1998; Anderson et al. 1999, 2001; Ryding et al. 2002; Sheeler et al. 2005; McKenzie and Nodwell 2007), and AfsQ1-Q2 (Ishizuka et al. 1992). Among them, AbsA1-A2 system is the most well studied and has been found to negatively regulate biosynthesis of the four antibiotics synthesized by S. coelicolor, actinorhodin (ACT), undecylprodigiosin (RED), methylenomycin, and the calcium-dependent antibiotic (CDA), by directly interfering with the expression of pathway-specific regulatory genes such as actII-ORF4 (Arias et al. 1999), cdaR (Ryding et al. 2002), and redZ (White and Bibb 1997).

Another TCS, the AfsQ1-Q2 system was identified initially for its ability to stimulate production of ACT and RED in Streptomyces lividans. However, disruption of either afsQ1 or afsQ2 on the S. coelicolor chromosome by use of phage ΦC31KC515 led to no detectable change in secondary metabolite formation or morphogenesis when the mutants were grown under rich YMPG medium. In addition, the afsQ1 gene on pIJ922 suppressed the S. coelicolor absA mutation and caused ACT production, suggesting its possible roles in regulating antibiotic production in S. coelicolor (Ishizuka et al. 1992). In this work, to seek further evidence for the function of AfsQ1-Q2 in S. coelicolor, we constructed the afsQ1-Q2 deletion mutant and studied its phenotype under various growth conditions. Consistent with previous report (Ishizuka et al. 1992), no obvious change of any phenotype was observed when cultured on complex media. However, when grown on defined minimal medium (MM) with glutamate as sole nitrogen, the mutant almost lost its ability to generate ACT.

In a previous study, an extracytoplasmic function sigma factor, SigE, that is located upstream of the CseB-CseC two-component system was found to be required for normal cell wall integrity and its expression was controlled by CseB-CseC in response to signals from the cell envelope in S. coelicolor (Paget et al. 1999a, b; Hong et al. 2002). Sequence analysis showed that a gene encoding putative sigma factor, sigQ, was also located upstream of afsQ1-Q2 (Bentley et al. 2002). The similar genetic organization tempted us to investigate the possible function of sigQ in this study. Similar to the case in SigE-CseB-CseC, the expression of sigQ was also under control by afsQ1-Q2. However, to our surprise, deletion of sigQ led to the precocious hyperproduction of antibiotic and delayed formation of sporulating aerial mycelium in the same glutamate-based defined MM. The results suggested that afsQ1-Q2 and sigQ worked together in the regulation of both antibiotic biosynthesis and morphological development in a medium-dependent manner in S. coelicolor.

Materials and methods

Strains, plasmids, and culture conditions

Bacterial strains and plasmids are listed in Table 1. MS agar medium (Kieser et al. 2000) was used to make spore suspensions. GYM (Chakraburtty et al. 1996), R2YE, and solid glucose-minimal medium (Kieser et al. 2000) in which ammonium was replaced by glutamate were used to determine ACT and RED production. RNA was isolated from cultures grown on MM medium supplemented with 75 mM sodium l-glutamate as the sole nitrogen source. 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, 100 μg ml−1 for thiostrepton).

Table 1 Bacterial strains and plasmids used in this study
Full size table

Transformation and conjugation

Plasmids or cosmids with oriT fragment are introduced by transformation into the methylation-deficient E. coli host ET12567 containing the RP4 derivative pUZ8002 and then transferred to S. coelicolor A3(2) M145 and its derivatives by intergeneric conjugation as described previously (Kieser et al. 2000).

Construction of ΔafsQ, ΔsigQ, and ΔafsQsigQ mutants

The mutants were constructed by replacing the entire coding region of the target genes with an apramycin resistance oriT cassette using the polymerase chain reaction (PCR)-targeting system described by Datsenko and Wanner (2000) and Gust et al. (2003). Primers used for construction of the deletion mutants in this study are listed in Supplementary Table 1. All mutants generated were confirmed by PCR, reverse-transcription (RT)-PCR, and DNA sequencing.

Complementation of the deletion mutants

The integrative plasmid pSET1521 was modified from pSET152 (Kieser et al. 2000) by adding a thiostrepton resistance gene (tsr) and constructed as follows. Plasmid pSET152 was digested with SphI and treated with Klenow polymerase fragment first and then alkaline phosphatase. The BclII fragment containing the tsr gene from pWHM3 was cloned in pSET152, generating pSET1521. The primers AQc1-c2, A1SQc1-c2, SQc1-c2, and ASQc1-c2 (listed in Supplementary Table 1) were used to amplify the fragments containing the open reading frames and the putative promoter regions of afsQ1-Q2, afsQ1-sigQ, sigQ, and afsQ1-Q2-sigQ, respectively, from the genomic DNA of S. coelicolor M145. The fragments were first cloned on pMD18-T vector and then moved into pSET1521. The resulting complemented vectors were introduced into S. coelicolor mutant strains by conjugation. The plasmid pSET1521 was used as a negative control in complementation assay.

Assay of the calcium-dependent antibiotic

CDA were assayed according to the method described by Kieser et al. (2000). Briefly, the tested strains were cultured on MM agar with glutamate for 24–48 h and then overlaid with soft LB agar containing Staphylococcus aureus and Ca(NO3)2 at a final concentration of 12 mM throughout the plate. A zone of inhibition, which is absent when Ca(NO3)2 is omitted, is diagnostic for CDA.

RNA preparation and RT-PCR analysis

RNA preparation and RT-PCR assays were performed as described previously (Lu et al. 2007). RT-PCR primers used in this study are listed in Supplementary Table 2. MM with 75 mM glutamate covered with cellophane disks was used for RNA isolations. For each RT reaction, 1 μg of RNA was denatured together with 500 ng of random hexamers and 10 mM deoxynucleotide triphosphates at 75°C for 5 min and quick-chilled on ice. 5× RT buffer, RNase inhibitor (Takara), and ReverTra Ace® (Toyobo) were added to each RNA–primer mixture to give a total volume of 20 μl. The reaction mixtures were incubated at 37°C for 10 min, followed by 42°C for 60 min. The reactions were stopped at 85°C for 5 min and chilled on ice. The PCR program was as follows: 30 s at 94°C (denaturation), 30 s at 62°C (annealing), and 30 s at 72°C (elongation). The reaction was completed by incubating for 5 min at 72°C. Samples were separated on 1.5% agarose gels in Tris–borate–ethylenediaminetetraacetic acid (TBE) buffer and stained with ethidium bromide. 16S rRNA gene in S. coelicolor was used as internal control. To exclude the possibility of DNA contamination in the RNA isolations, PCR amplification was performed on all RNA samples without prior reverse transcription.

egfp fusions and expression

Plasmid pIJ8660 (Sun et al. 1999) was digested with NheI and treated with Klenow polymerase fragment and alkaline phosphatase, respectively. The BclI fragment containing the tsr gene from pWHM3 was cloned in pIJ8660, generating pIJ86601. For egfp fusions, the upstream regions of hrdB (896 bp) and sigQ (906 bp) were amplified by PCR with the primers PHrdB1-2 and PSQ1-2, respectively (listed in Supplementary Table 1). Both fragments were sequenced in pMD18-T vector, and the correct clones were digested with KpnI and BglII and transferred to vector pIJ86601, which was digested with KpnI and BamHI generating pIJ86601-hrdBp and pIJ86601-sigQp, respectively. The constructs were introduced into S. coelicolor M145 by conjugation from E. coli. Microscopic images were obtained by a Zeiss LSM 510 META confocal microscope (argon laser at 488 nm, objective 40 × 0.75 NA dry PL Fluotar).

Results

AfsQ1-Q2 system regulates antibiotic production and morphological development in a medium-dependent manner in S. coelicolor

The afsQ genes of S. coelicolor were identified for their ability to stimulate ACT and RED production in S. lividans when afsQ1-Q2 or just afsQ1 was introduced on a plasmid (Ishizuka et al. 1992). However, disruption of afsQ1 or afsQ2 on the S. coelicolor genome had no effect on either antibiotic production or morphological differentiation when grown in rich media, such as YEME and YMPG liquid medium (Ishizuka et al. 1992). In this study, we constructed a mutant with afsQ1-Q2 gene cluster deleted by PCR-targeting system. When compared to wild-type M145, the afsQ mutant established no visible change on pigmented antibiotic synthesis and morphogenesis when grown on MS, R2YE, and GYM media (data not shown), consistent with early study (Ishizuka et al. 1992). We then tested the mutant in an MM medium supplemented with various nutritional factors in a plate assay. It was found that deletion of afsQ1-Q2 genes impairs the production of ACT, RED, and CDA when grown on MM with 75 mM sodium l-glutamate as the sole nitrogen source (Fig. 1a–c). Even cultivated more than 4 days, the afsQ mutant showed almost no extracellular ACT production. Meanwhile, growth of aerial mycelium in the mutant was found to be faster than M145 (Fig. 2). To exclude the possibility that the effects were caused by sodium, we also replaced sodium l-glutamate by several other sodium salts; in those cases, no phenotype change between the mutant and M145 was observed (data not shown), indicating that the effects were indeed caused by glutamate. In addition, to exclude the possibility that the phenotype changes in the afsQ mutant were caused by polar effect on downstream genes, afsQ1 and afsQ2 with their possible promoter region were introduced into the mutant by using the integrative vector pSET1521 (modified from pSET152 with a thiostrepton-resistant gene; Kieser et al. 2000). Introduction of the complementary constructs into this mutant restored antibiotic production, and the aerial mycelium formation was back to the same level as wild-type M145, indicating that the phenotype changes in the mutant resulted solely from the afsQ mutation (Fig. 1c,d). However, pSET1521 carrying the afsQ1 without afsQ2 did not restore ACT production (Fig. 1d), suggesting that kinase AfsQ2 might be the sole phosphorus group donor of response regulator AfsQ1.

Fig. 1
figure 1

Phenotypes of S. coelicolor M145 and its derivatives. a ACT production by the wild type (M145), the afsQ mutant (ΔafsQ), sigQ mutant (ΔsigQ), and afsQ-sigQ double mutant (ΔafsQsigQ) on MM supplemented with 75 mM glutamate was observed at day 2. b RED production by the wild type (M145), the afsQ, and sigQ and afsQsigQ mutants on MM with 10 mM glutamate at day 2. c CDA production by the wild type (M145) and the afsQ, sigQ, and afsQsigQ mutants and their complementary transformants on MM with 75 mM glutamate for 31 h. d The phenotype of S. coelicolor M145a (M145 with pSET1521), afsQa (ΔafsQ with pSET1521), sigQa (ΔsigQ with pSET1521), afsQsigQa (ΔafsQsigQ with pSET1521), and their complementary transformants afsQb (ΔafsQ with pSET1521-afsQ), afsQd (ΔafsQ with pSET1521-afsQ1sigQ), sigQb (ΔsigQ with pSET1521-sigQ), and afsQsigQb (ΔafsQsigQ with pSET1521-afsQsigQ) on MM with 75 mM glutamate

Full size image
Fig. 2
figure 2

Morphology of mycelium formation by M145 and afsQ and sigQ mutants. Cells were incubated at 30°C for 28 and 40 h on MM plates with 75 mM glutamate. In contrast to M145 (a), ΔafsQ (b) was covered with abundant aerial mycelia after 28-h growth. However, the sigQ-null mutant (c) exhibited a severe delay in formation of aerial hyphae even incubated for 40 h when M145 (d) was covered with abundant aerial hyphae

Full size image

SigQ regulates antibiotic production and morphological development in a medium-dependent manner in S. coelicolor

Right upstream of afsQ1-Q2 genes on the S. coelicolor chromosome is a sigQ gene which encodes a putative sigma factor (Bentley et al. 2002). The sigQ gene is transcribed in the opposite direction from afsQ1-Q2, so they were not in the same operon. The similar genetic organization was also found previously for the TCS cseB-cseC system and sigma factor sigE, which worked together for normal cell wall integrity (Paget et al. 1999b; Hong et al. 2002). The similarity in terms of putative functions (i.e., AfsQ2 has ∼40% identity to CseC at the amino acid level; Hutchings et al. 2006b) and genetic organization between the sigQ-afsQ1-Q2 and sigE-CseB-CseC systems prompted us to study whether SigQ is also involved in regulating antibiotic production and morphological differentiation in S. coelicolor. The sigQ gene was disrupted by replacing the sigQ coding sequence with the apramycin resistance gene and the phenotype of the mutant was studied. Although the sigQ mutant appeared identical to the wild-type strain on all rich media tested, it produced ACT and RED earlier and in greater amounts than M145 on an MM medium supplemented with glutamate (Fig. 1a,b). In addition, formation of aerial mycelium of the mutant was also obviously delayed (Fig. 2). The overproduction of antibiotic and the poor sporulation could be complemented in trans by integrating a functional copy of sigQ gene into M145 strain by using pSET1521 (Fig. 1c,d). The results demonstrated that SigQ also regulates antibiotic production and morphological development in a medium-dependent manner, and its function is antagonizing from that of AfsQ1-Q2 in S. coelicolor.

SigQ is secondary to AfsQ1-Q2 in the signal transduction cascade

Although the afsQ1-Q2 and sigQ seemed to function differently in regulating antibiotic production and morphological differentiation in S. coelicolor, our next question is whether they are functionally coupled, like in the case of SigE-CseB-CseC (Paget et al. 1999a; Hong et al. 2002). We then determined the transcript level of afsQ and sigQ in the wild-type M145 strain and the afsQ or sigQ mutant using semiquantitative RT-PCR. RNA samples were isolated from cells grown on cellophane on the solid MM containing 75 mM glutamate at 36, 48, and 60 h, respectively. 16S rRNA was used as a negative control for this assay. The results showed a significant reduction of the sigQ transcript in the afsQ mutant (Fig. 3a), while deletion of sigQ had no effect on afsQ1-Q2 transcription (data not shown). The reduction can be restored by complementing the afsQ mutant with the afsQ1-Q2 cluster (Fig. 3b). In addition, introduction of plasmid pSET1521-sigQ and pSET1521-afsQ1sigQ in the afsQ mutant could not restore the transcription of sigQ (Fig. 3c), suggesting that the afsQ1-Q2 genes are required for the transcription of sigQ under the condition we tested. To further confirm the regulation of sigQ by afsQ1-Q2, in vivo transcription level of sigQ was compared between M145 and the afsQ mutant using an enhanced green fluorescence reporter (egfp) system. The promoter region of the sigQ gene (sigQp) was fused to the egfp gene in the integrative plasmid pIJ86601 (modified from pIJ8660 with a thiostrepton-resistant gene; Sun et al. 1999). The resulting plasmid was introduced into the wild-type M145 and the afsQ mutant, respectively, and the expression of EGFP was investigated using a confocal microscopy. The results showed that in M145, EGFP was clearly induced from the sigQ promoter on the MM supplemented with glutamate, whereas the expression of EGFP was not observed in the afsQ mutant (Fig. 4). In a control egfp experiment, the promoter of the housekeeping gene hrdB was shown to be insensitive to disruption of afsQ gene in M145 (Fig. 4). The results demonstrated that the transcription of sigQ is controlled by AfsQ1-Q2.

Fig. 3
figure 3

Transcription analysis of sigQ and afsQ gene in S. coelicolor M145 and its derivatives grown on MM with 75 mM glutamate. RNA was extracted from mycelium harvested at different time points. PCR products were separated on 1.5% gel in TBE 1× buffer. 16S rRNA gene was used as a control. a RT-PCR analysis of the sigQ gene in S. coelicolor M145 and the afsQ mutant (ΔafsQ). RNA was extracted from mycelium harvested 36, 48, and 60 h. b RT-PCR analysis of the afsQ and sigQ gene in S. coelicolor M145 (M145a), the afsQ mutant (ΔafsQa), and the complemented strain (ΔafsQb). RNA was extracted from mycelium harvested at 54 h. c RT-PCR analysis of the sigQ gene in afsQa (ΔafsQ with pSET1521), afsQb (ΔafsQ with pSET1521-afsQ), afsQc (ΔafsQ with pSET1521-sigQ), and afsQd (ΔafsQ with pSET1521-afsQ1sigQ). RNA was extracted from mycelium harvested at 36 h

Full size image
Fig. 4
figure 4

Analysis of sigQp-egfp and hrdBp-egfp transcriptional fusions in whole colonies of M145 and the afsQ mutant grown on minimal agar in the presence of 75 mM glutamate. EGFP fluorescence was visualized using a Zeiss LSM 510 META confocal microscope after 42-h incubation at 30°C as described in experimental procedures. a Fluorescence microscope, b light microscope

Full size image

We also checked the dependence of SigQ on AfsQ1-Q2 by genetic studies. The results showed that, although the deletion of afsQ or sigQ caused different phenotypes in the conditions we tested, the afsQ-sigQ double mutant had the same phenotypes as the afsQ mutant (Fig. 1a), suggesting that the disruption of sigQ in afsQ mutant cannot reverse the decreased antibiotic production resulting from the deletion of afsQ1-Q2. Moreover, introduction of both afsQ1-Q2 and sigQ genes (pSET1521-afsQsigQ) into the afsQ-sigQ double mutant resulted in the phenotype similar to M145 (Fig. 1d), and introduction of only afsQ1-Q2 genes (plasmid pSET1521-afsQ) generated a phenotype similar to the sigQ mutant (i.e., overproduction of ACT in the afsQ-sigQ mutant; Fig. 1d). The results further suggested that, while AfsQ1-Q2 and SigQ may function together in the regulation of antibiotic biosynthesis and morphological development, AfsQ-Q2 may be located at a higher level than SigQ in the signal transduction cascade and sigQ may be involved in antagonizing the function of afsQ1-Q2.

AfsQ-SigQ system regulates antibiotic production through pathway-specific activators

To obtain a better understanding of the possible regulatory mechanism of the AfsQ1-Q2-SigQ on antibiotic production, we determined the effects of afsQ1-Q2-sigQ gene expression on the transcription of pathway-specific factors for ACT, RED, and CDA biosynthesis, including actII-ORF4 (Arias et al. 1999), redD (Takano et al. 1992; White and Bibb 1997), and cdaR (Ryding et al. 2002). RT-PCR studies were conducted using RNA samples isolated from M145, ΔafsQ, and ΔsigQ at 36 or 42 h in solid MM media supplemented with 75 mM glutamate as the sole nitrogen. The results showed that transcription of actII-ORF4, redD, and cdaR were significantly reduced in the afsQ mutant (Fig. 5a), while it increased in the sigQ mutant (Fig. 5b). The results were consistent with the physiological measurements and suggested that AfsQ and SigQ regulate antibiotic production of S. coelicolor by affecting the transcription of pathway-specific activators.

Fig. 5
figure 5

Transcription analysis of pathway-specific regulatory genes in S. coelicolor M145 and its derivatives grown on MM with 75 mM glutamate. Expression of actII-ORF4, redD, and cdaR in ΔafsQa, b and ΔsigQc. RNA was extracted from mycelium harvested 36 or 42 h. PCR products were separated on 1.5% gel in TBE 1× buffer. 16S rRNA gene was used as a control

Full size image

Discussion

One well-established mechanism to conduct signal transduction in streptomycetes is the two-component system (Horinouchi and Beppu 1992), through which environmental signal recognition results in the autophosphorylation of sensor histidine protein kinase, followed by transfer of the phosphoryl group to the second component protein, a response regulator (Parkinson and Kofoid 1992). Analysis of the S. coelicolor genome has revealed evidences of abundant two-component regulatory systems (Bentley et al. 2002). Although the functions of these regulators are still largely unknown, the existence of so many two-component regulatory systems in the S. coelicolor genome clearly implies their important physiological roles. Thus, the elucidation of their functions will certainly benefit the understanding of complex regulatory networks involved in the production of secondary metabolites in these organisms. In the past several years, our laboratory has been studying functions of various TCS systems in Amycolatopsis mediterranei and S. coelicolor using molecular genetics approach (Wang et al. 2004; Wei et al. 2004; Lu et al. 2007). As part of these continuous efforts, we reported here the functional characterization of afsQ1-Q2 and its upstream sigQ gene. Instead of using the rich media in which disruption of either afsQ1 or afsQ2 in S. coelicolor led to no detectable change in secondary metabolite formation or morphogenesis (Horinouchi and Beppu 1992), in this study, we checked the effects of afsQ1-Q2 gene deletion on antibiotic biosynthesis and morphological development in an MM medium supplemented with various nitrogen sources in a plate assay. The results showed that, when cultivated on defined MM with glutamate as the sole nitrogen source, the afsQ mutant exhibited significantly decreased production of antibiotic and rapid growth of aerial mycelium, suggesting that afsQ1-Q2 is a pleiotropic TCS system involved in both antibiotic production and morphological differentiation in S. coelicolor, and its regulatory effects are in a medium (glutamate)-dependent manner. The detailed mechanism of this medium (glutamate)-dependent regulation is still yet to be discovered in Streptomyces; however, it has been reported in E. coli that the P (II) signal transduction proteins GlnB and GlnK are uridylylated–deuridylylated in response to the intracellular glutamine level, and the cellular glutamine signal is transduced by uridylyltransferase and GlnB to modulate TCS NtrB-NtrC-dependent gene expression (Maheswaran and Forchhammer 2003).

Two-component signal system genes characterized so far in Streptomyces fall into two classes according to their pleiotropic effects on secondary metabolism and morphological differentiation: (1) those that affect only antibiotic production, such as absA-absB (Anderson et al. 2001) and cutR-cutS (Chang et al. 1996), and (2) those that affect only morphological differentiation, such as ramR (Ma and Kendall 1994). We found here that AfsQ1-Q2 system not only regulates antibiotic production but also is involved in the formation of aerial mycelium in S. coelicolor. These two phenotypes are closely related. For example, it has been reported that during carbon- and phosphate-limited cultures, Bacillus subtilis attempts a number of adaptive approaches (motility, chemotaxis, and synthesis of extracellular degradative enzymes and antibiotics), any of which, if successful, would allow vegetative growth to resume before it begins sporulation (Fisher and Sonenshein 1991; Fisher 1999). If it is also the case for S. coelicolor, it may explain the early formation of aerial mycelium formation in the afsQ mutant that has stopped producing antibiotic.

In early studies, a CseB-CseC two-component system has been found activating its upstream sigE gene in response to signals from the cell envelope in S. coelicolor (Paget et al. 1999b). Consistently, a constructed sigE-null mutant has the same phenotype as a cseB-null mutant. In addition, analysis of the expression of 12 putative target genes controlled by sigE (the cwg operon) showed a medium-dependent manner: in low-Mg2+ medium, transcription of the cwg operon was induced by vancomycin in a sigE-dependent manner but, in high-Mg2+ medium, there was substantial cwg transcription even in a sigE-null mutant (Paget et al. 1999b; Hong et al. 2002). A putative sigma factor, sigQ is also located upstream of afsQ1-Q2 genes (Bentley et al. 2002). We thus wonder whether the set of them behaves the same way as SigE-CseB-CseC. Consistent with our expectation, RT-PCR and egfp analyses showed that sigQ expression is under control by AfsQ1-Q2; however, to our surprise, deletion of sigQ led to the precocious hyperproduction of antibiotic and delayed formation of sporulating aerial mycelium in a medium (glutamate)-dependent manner, contrary to the decreased antibiotic production and early formation of aerial mycelium by afsQ1-Q2 deletion, and the double mutant ΔafsQsigQ has the same altered phenotypes as the afsQ mutant. The results provided evidence for a novel medium-dependent regulatory mechanism in which a sigma factor SigQ, under the transcriptional regulation of by afsQ1-Q2, might be responsible for antagonizing the function of AfsQ1-Q2. In addition, we also showed that ACT, RED, and CDA pathway-specific activator genes, actII-ORF4, redD, and cdaR, are among the possible target genes regulated by afsQ1-Q2-sigQ.

Although, to our best knowledge, no similar type of antagonizing mechanism has been reported for Streptomyces, it was found that KbpA, which is located just upstream of a Ser/Thr protein kinase afsK, inhibited autophosphorylation of AfsK by binding the catalytic domain of the unphosphorylated form of AfsK and put a brake on the unlimited production of the pigments under the control of the AfsK-AfsR system, thus playing a role in a negative-feedback system (Umeyama and Horinouchi 2001). Although more proof is still needed, it is speculative that a similar mechanism may be present that σQ is recruited by core RNA polymerase to transcribe some genes encoding regulatory protein (s) as KbpA, which may antagonize the function of AfsQ1-Q2 by interacting directly with the sensor domain of AfsQ2 to modulate its activity and then control the level of phospho-AfsQ1. When sigQ is disrupted, this repression disappears and the function of phospho-AfsQ1 was enhanced, resulting in precocious overproduction of antibiotic. However, it needs further investigations.

Taken together, we can conclude that afsQ1-Q2-sigQ is a pleiotropic but conditionally required signal transduction system for both secondary metabolism and morphological development in S. coelicolor. This highly regulated process, in which different regulatory mechanisms are involved, could provide benefits in fine-tuning the timing and strength of the expression of the genes involved in antibiotic biosynthesis and morphological development in S. coelicolor. Finally, as early Southern blot hybridization experiments showed that sequences homologous to afsQ1-Q2 are present in almost all of the actinomycetes examined (Ishizuka et al. 1992) and our recent examination of the Streptomyces avermitilis and Streptomyces griseus genome has revealed that the sigQ and afsQ1-Q2 homologs (SAV_3351, SAV-3352, and SAV_3353; SGR_2634, SGR_2635, and SGR_2636, respectively) were organized in an identical way (Ikeda et al. 2003; Ohnishi et al. 2008), it is worth further investigation whether the novel regulatory mechanism reported here for afsQ1-Q2-sigQ represents a commonly used regulatory strategy in other actinomycetes.