Regulation of the Streptomyces coelicolor Calcium-Dependent Antibiotic by absA, Encoding a Cluster-Linked Two-Component System

Abstract

The Streptomyces coelicolor absA two-component system was initially identified through analysis of mutations in the sensor kinase absA1 that caused inhibition of all four antibiotics synthesized by this strain. Previous genetic analysis had suggested that the phosphorylated form of AbsA2 acted as a negative regulator of antibiotic biosynthesis in S. coelicolor (T. B. Anderson, P. Brian, and W. C. Champness, Mol. Microbiol. 39:553–566, 2001). Genomic sequence data subsequently provided by the Sanger Centre (Cambridge, United Kingdom) revealed that absA was located within the gene cluster for production of one of the four antibiotics, calcium-dependent antibiotic (CDA). In this paper we have identified numerous transcriptional start sites within the CDA cluster and have shown that the original antibiotic-negative mutants used to identify absA exhibit a stronger negative regulation of promoters upstream of the proposed CDA biosynthetic genes than of promoters in the clusters responsible for production of actinorhodin and undecylprodigiosin. The same antibiotic-negative mutants also showed an increase in transcription from a promoter divergent to that of absA, upstream of a putative ABC transporter, in addition to an increase in transcription of absA itself. Interestingly, the negative regulation of the biosynthetic transcripts did not appear to be mediated by transcriptional regulation of cdaR (a gene encoding a homolog of the pathway-specific regulators of the act and red clusters) or by any other recognizable transcriptional regulator associated with the cluster. The role of absA in regulating the expression of the diverse antibiotic biosynthesis clusters in the genome is discussed in light of its location in the cda cluster.

Members of the bacterial genus Streptomyces synthesize the majority of known microbial antibiotics, as well as a number of other products that have been shown to exhibit useful biological activities. Streptomyces coelicolor has been used as a model organism for the study of the regulation of antibiotic biosynthesis (7). This species produces four chemically distinct antibiotics: actinorhodin (Act), undecylprodigiosin (Red), calcium-dependent antibiotic (CDA), and methylenomycin (Mmy). The pigmentation properties of Act (blue) and Red (red) have greatly facilitated genetic screens designed to identify genes that either affect the synthesis of those products individually or affect the coordinated regulation of antibiotics in S. coelicolor during the life cycle. The genes responsible for the synthesis of each of the four antibiotics have been found to be clustered in distinct locations (15, 39). The biosynthetic genes within the act and red clusters are regulated at the transcriptional level by the cluster-linked regulators actII-ORF4 and redD, respectively (6, 23, 42), which form the core of the streptomycete antibiotic regulatory protein (SARP) family of regulators (28, 44), which in turn are found in antibiotic clusters in diverse streptomycetes. Many of the SARP regulators have been shown to function as activators of biosynthetic promoters in the associated antibiotic clusters (21, 23, 42). Transcription of these SARP activators shows growth phase regulation, increasing during the transition period between exponential growth in liquid medium and the stationary phase (23, 42).

The onset of antibiotic production in Streptomyces cultures grown on solid medium is largely coincident with the development of aerial mycelium (16). Several genes have been identified in S. coelicolor that are required for both antibiotic production and the development of the aerial mycelium (bld genes) (13, 14, 17). However, a number of genes have also been isolated that affect only antibiotic production. Of these, absA was identified initially in a screen for mutants that showed loss of the Act and Red pigments but retained normal sporulation (the Abs⁻ phenotype [2]). Such mutants were subsequently found to be CDA⁻ and Mmy⁻ as well. The point mutations that caused the nonproduction phenotype were located in the sensor kinase of a two-component signal transduction system (4), although null mutations in either the sensor kinase (absA1) or the adjacent response regulator (absA2) caused the precocious hyperproduction of antibiotics (the pha phenotype [8]). Changes in the specific residues predicted to be critical for phosphorylation of AbsA2 also caused the overproduction phenotype, indicating that the phosphorylated form of the putative DNA-binding protein AbsA2 acted as a global negative regulator of antibiotic biosynthesis (5). The Abs⁻ mutants behave as though phosphorylation is deregulated, locking AbsA2 into the negatively regulating phosphorylated form (5). The antibiotic phenotypes of absA mutants were correlated with changes in the levels of actII-ORF4 and redD transcripts, implicating absA as a regulator of expression of those genes (1). The absA transcript is also subject to autoregulation, with phosphorylated AbsA2 mediating positive transcriptional autoregulation (5).

Data released by the Sanger Centre (Cambridge, United Kingdom) as part of the effort to sequence the entire S. coelicolor genome (http://www.sanger.ac.uk/Projects/S_coelicolor/) showed that absA may be part of the gene cluster responsible for the synthesis of CDA. Chemical analysis of CDA had shown that it was a cyclic lipopeptide composed of 11 amino acid residues linked to a six-carbon fatty acid chain (32), which in previous studies had been shown to require calcium ions for its mode of action (33). In experiments based on the premise that CDA was likely to be synthesized by a nonribosomal peptide synthetase, degenerate primers designed against conserved regions of nonribosomal peptide synthetases from other species were used to amplify two sections of S. coelicolor chromosomal DNA using PCR (18). These fragments of the CDA peptide synthetases were found to hybridize to the cosmids E8 and E69 from the minimal set of ordered cosmids for the S. coelicolor genome (18, 39). This location correlated well with the previously determined genetic map position for the cda cluster (26), and confirmation that part of the CDA peptide synthetases had been cloned was provided by the fact that gene disruptions directed by the cloned fragments yielded strains that were unable to produce CDA. The genomic sequence data provided by the Sanger Centre showed that absA was separated from the peptide synthetase genes by only three open reading frames (ORFs). In this paper we briefly describe the genetic organization of the cda cluster, based on our analysis of the genomic sequence. We define many of the transcriptional start sites within the cda cluster and present data showing the effect of absA mutations on transcription within the cda cluster. In order to assess the role of absA in regulation of CDA biosynthesis, we define promoters that are positively or negatively regulated by absA and evaluate potential regulation by absA of other possible transcriptional regulators in the cda cluster.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The S. coelicolor strains used in this study were J1501 (hisA1 uraA1 strA1 pgl) C542 (hisA1 uraA1 strA1 pgl absA1-542 [2]), and C500 (hisA1 uraA1 strA1 pgl absA::pTBA500 [5]). Spores of S. coelicolor strains were harvested from cultures grown on SpMR agar (5), in the case of C500 including hygromycin at 50 μg/ml. S. coelicolor cultures for isolation of RNA were grown on cellophane discs laid on top of SpMR, inoculated at a density of approximately 10⁶ spores per plate, and incubated at 30°C.

Plasmid and DNA manipulations.

Oligonucleotide primers (Tables 1and 2) were prepared by the Macromolecular Structure Facility at Michigan State University (East Lansing, Mich.). Escherichia coli plasmid preparations were carried out using Wizard Miniprep columns (Promega) or by a standard alkaline lysis method (40).

TABLE 1.

Oligonucleotide primers used to prepare probes for S1 nuclease protection assays

Primer	Transcript detected	Cosmid (location)	Primer sequence	Complement
WC20	p4 (absA)	E8 (21284–21302)	CATCGACGGCCACAGGTTC	WC116
WC94	p6 (cdaR)	E8 (13484–13462)	CCGCCCACCGTAAGACCTCGGCC	WC106
WC106	NDa	E8 (12900–12922)	GGCGCACTGACGAAAGCAAGGGC	WC94
WC114	p3 (SCE63.04c)	E63 (2420–2440)	GTCGACGAGGAAGCCGGAGGC	WC123
WC116	p5 (SCE8.17c)	E8 (20811–20832)	AACGAGAGGTCCGAGACGGCCG	A1A3
WC117	p7 (SCE8.08c)	E8 (9838–9858)	TCGGCCAGGGCGTGCACCTCC	WC118
WC118	ND	E8 (10280–10260)	GCGACCTCCGCCGCGCTCAGG	WC117
WC119	p1 (SCE29.18c)	E29 (24628–24648)	GGTCCACTTCCAGATCCTCGG	WC120
WC120	ND	E29 (25117–25096)	CCCGCGTTGATGTCAGAAGTGG	WC119
WC123	p2 (cda PSI)	E63 (2950–2930)	CGGCTCAGCACCGCGTGGTCC	WC114
WC127	p9 (SCE22.24c)	E22 (25340–25360)	CGGGCCACCTTGCGCTGATCG	WC128
WC128	p8i (SCE8.01)	E8 (398–376)	CGGTCTGCGCGCGCGTATGCGGC	WC127
WC129	ND	E8 (1371–1391)	CGCGAGGGCAACGACGAGACG	WC130
WC130	p8ii (SCE8.02)	E8 (1761–1779)	CGCGGCCGAGTACGGCGATGC	WC129
WC134	p13 (SCE22.17)	E22 (17140–17163)	TGATCCACTGGATCTGGCTGAAGG	WC135
WC135	ND	E22 (17529–17522)	GCCTGCACCGGATAGCCCAGTTCG	WC134
WC136	p12 (hrdD)	E22 (20737–20760)	CTGGTTGCCGAGCCGGAGGTCTCG	WC137
WC137	ND	E22 (21050–21073)	TGGCTTGAGATCCTCCTCGACTCC	WC136
WC138	ND	E22 (22160–22129)	TAGCGCACCTGGTAGTAGTGCTCC	WC139
WC139	p10 (SCE22.22)	E22 (22487–22510)	GCCTCGATGTAGGAGCGCCAGACG	WC138
WC165	ND	5B8 (11253–11276)	CGAGGTGGAACGCGTGGTGCCCGC	WC166
WC166	hrdB	5B8 (11665–11688)	CCGGCCCCGTCCACGGCGGACAGG	WC165

^aND, transcripts not detected.

TABLE 2.

Oligonucleotide primers used for RT-PCR assays

ORF name	Forward primer	Reverse primer	Cosmid (location)	No. of cycles
SCE22.17	CCCTCCTCTCCCTTCTCCAGACGC	AGGGCACCTGCTCCAGCTTGG	E22 (17423–17734)	35
SCE22.22	ACTACGGCCTGCTGGTCAAGCTGG	TGAGCCGGTCGAAGACCGCCTGGC	E22 (22576–22839)	35
SCE63.04c	CCAGTGACATCGCGTACGCCGAGC	CTGAAGCGCCTCGGCACCGGCGAG	E63 (2180–2487)	25
trpE	CACCGACCGGTGCCGCTCGAGCCG	GACGACGAGGTCGTCGGCGGCGAGG	E8 (8377–8604)	25
SCE8.16c	AGAAGGGCGTCGACGTCACGATGCC	CAGGGTCTTGGAGCGCAGCATCCG	E8 (19550–19731)	25

Sequence analysis.

The sequences of the ORFs from the cosmids E22 (GenBank accession no. AL355832), E8 (accession no. AL035654), E63 (accession no. AL035640), and E29 (accession no. AL035707) provided by the Sanger Centre were used in BLAST searches to detect proteins in the databases that showed similarity to the cda ORF. In cases where protein sequences with a significant level of homology were detected, a number of the protein sequences from the databases were aligned with the cda sequence using the PILEUP program to show that regions of similarity were conserved across the whole group. Putative functions for the ORFs from E22, E8, E63, and E29 were assigned on the basis of similarity to proteins that had previously been characterized biochemically.

Disruption of PSI in JR123.

A fragment internal to the PSI gene was amplified by PCR using Taq polymerase and the oligonucleotide primers WC121 (5′GATCCGGCGCGTCTTCGAGGC3′) and WC122 (5′GTCAGGTCGAAGGTGTGCCGG3′). The 416-bp product was cloned using pCR2.1 from a TA Cloning Kit (Invitrogen). A plasmid containing the correct insert was used to transform E. coli strain DM1. Plasmid DNA from this strain was then prepared by a standard alkaline lysis procedure (40) and used to transform protoplasts of S. coelicolor J1501, selecting for kanamycin resistance. The arrangement of the chromosomal DNA around the site of integration of pNJR123 was confirmed by Southern analysis.

Bioassay for CDA.

Assays for CDA were carried out using the method previously described by Adamidis et al. (2).

RNA isolation.

Isolation of RNA from S. coelicolor strains was carried out as described by Hopwood et al. (27). The first set of RNA was isolated from the strains C542 and J1501 grown on SpMR over cellophane, with mycelium harvested at 24, 48, and 72 h after inoculation. A second set of RNA was isolated from the strains C542, J1501, and C500 grown on SpMR and harvested at 18, 24, 42, 48, and 72 h after inoculation. Production of CDA during the time course was tested by the method described previously (2). The RNA was examined by agarose gel electrophoresis and quantified spectrophotometrically using a Pharmacia Biotech GeneQuant II spectrophotometer.

S1 nuclease protection assays.

Double-stranded DNA probes for the S1 nuclease protection assays were generated by PCRs in which one primer had been labeled at the 5′ end using T4 polynucleotide kinase (PNK). The primers used to generate each probe are shown in Table 1. In each case, 50 pmol of the appropriate primer was added to 4.0 μl of spermidine (100 mM), made up to 32 μl with deionized water, and placed in a water bath at 70°C for 10 min before being placed on ice. Each primer was then labeled using 5 U of PNK (Promega) in the buffer specified by the manufacturer and 12.5 μl of 3,000-Ci/mmol [γ-³²]P in a final volume of 50 μl. After 20 min of incubation at 37°C, a further 5 U of PNK was added, followed by a further incubation period of 20 min at 37°C. Labeling reactions were diluted to 220 μl with deionized water, extracted with a phenol-chloroform mixture (1:1) and then chloroform alone, and precipitated in ethanol at −70°C after the addition of glycogen (2 μl) and sodium acetate (pH 6.8) to a final concentration of 0.3 M. Half of the labeled primer was added to a PCR mixture that included 100 ng of S. coelicolor chromosomal DNA, 0.2 mM deoxynucleoside triphosphates (dNTPs), 1.5 mM MgCl₂, 5% glycerol, 2.5% dimethyl sulfoxide, 1% formamide, 0.5 U of Taq polymerase (Gibco BRL), and the buffer specified by the manufacturer in a total volume of 100 μl. Thermocycling conditions for PCR were 95°C for 1 min, followed by 30 cycles of 95°C (1 min), 65°C (1 min), and 72°C (1 min), followed by a 10-min incubation at 72°C. The S1 nuclease protection assays were performed as previously described (1) using 20 or 40μg of RNA and a quantity of probe that gave a total activity of approximately 100,000 cpm. After precipitation in isopropanol and incubation at −70°C for at least 3 h, the reactions were separated on a 6% acrylamide gel, using ³²P-end-labeled φX174 HinfI DNA as a size standard. Each set of S1 nuclease protection assays shown in Fig. 3 and 4 was repeated at least once.

FIG. 3.

S1 nuclease protection assays of RNA isolated from the antibiotic-deficient absA strain C542, the parental strain J1501, and the strain in which absA2 had been truncated (Materials and Methods), C500. Assays are shown for transcription from p3, the promoter upstream of the hydroxymandelate synthase gene; p4, the absA promoter, p5, the promoter upstream of the putative ABC transporter transcribed divergently to absA; and p6, the promoter from which cdaR is transcribed. Primers used for each probe are given in Table 2. Each set of assays was repeated at least twice. For comparison, reactions using probes for transcription of redD, the pathway-specific regulator of the Red cluster, and hrdB, the major sigma factor of S. coelicolor, used to provide a positive internal control, are also shown. The numbers of hours are shown along the top.

FIG. 4.

Quantitative RT-PCRs separated on 3% agarose gels, stained with ethidium bromide, and visualized under UV. The reactions included RNA samples shown in Fig. 3. Oligonucleotide primers were designed to detect internal fragments for SCE8.16c, the putative ABC transporter transcribed divergently from absA; SCE63.04c, the putative hydroxymandelate synthase gene; and trpE2, the putative anthranilate synthase component I. The primers used are given in Table 3. The names given in parentheses represent the promoters examined in the S1 nuclease protection experiments shown in Fig. 3 and 4 that lie upstream of the regions amplified in the RT-PCRs shown here. The numbers of hours are shown along the top.

Quantitative RT-PCR.

The quantitative reverse transcriptase (RT)-PCR assays were conducted using separate RT and PCRs. Each RT reaction mixture contained 1 μg of RNA (heated to 70°C for 10 min and cooled on ice, prior to the addition of the other constituents), 0.5 U of avian myeloblastosis virus RT (Promega), 2.5% (vol/vol) RNAsecure (Ambion), 1 mM dNTP mixture, 1× buffer (supplied by the manufacturer of the enzyme), 25 pmol of reverse primer for each transcript (Table 2), and deionized water to give a final volume of 20 μl. Each reaction was incubated at 45°C for 30 min, 50°C for 30 min, and 55°C for 30 min in an MJ Research Minicycler. For the PCR, 4 μl of the RT reaction was added to 0.2 mM dNTPs, 1.5 mM MgCl₂, 5% glycerol, 2.5% dimethyl sulfoxide, 1% formamide, 0.5 U of Taq polymerase (Gibco BRL), the appropriate forward and reverse primers (Table 2), and the buffer specified by the manufacturer in a total volume of 100 μl. Thermocycling conditions for PCR consisted of an incubation at 95°C for 1 min, followed by a certain number of cycles of 95°C (1 min), 65°C (1 min), and 72°C (1 min), followed by an incubation at 72°C for 10 min. Each set of PCRs for each transcript that was assayed was repeated using different numbers of cycles to ensure that amplification had not been restricted by depletion of a reactant. The number of cycles used to detect the transcripts described in Fig. 5 is given in Table 2. After completion of the thermocycle program, 12 μl of each reaction mixture was subjected to electrophoresis in a 2% agarose gel using Tris-borate-EDTA buffer, and DNA was visualized by ethidium bromide staining and UV illumination.

FIG. 5.

S1 nuclease protection assays for transcription of the putative efflux protein, from a point upstream of the TetR homolog, SCE22.24c, designated p9 in Fig. 1. The RNA samples used were the same used in the reactions shown in Fig. 3, as is the set of reactions used to quantify glk transcription, shown as a positive control. nt, nucleotides. The numbers of hours are shown along the top.

RESULTS

absA regulation of antibiotic gene clusters.

Previous work on the role of absA in regulation of antibiotics in S. coelicolor had shown that antibiotic-nonproducing absA mutants were deficient in transcription of the act and red antibiotic loci (1). In addition, mutations of the absA locus that enhanced antibiotic production (pha) were found to increase transcription from the act and red clusters (1, 8). Moreover, absA appeared to regulate Act and Red production through transcriptional regulation of the actII-ORF4 and redD pathway-specific regulatory genes. Because absA mutations affect CDA production as well as Act and Red, it was of interest to conduct similar studies to examine the transcription of genes of the CDA biosynthetic cluster in absA mutants.

To assess cda transcription, we used S1 nuclease protection assays, supplemented with RT-PCR assays, on RNA isolated from S. coelicolor cultures grown on cellophane-covered solid medium (SpMR). Selection of potential promoter sites within the cda cluster was based on the locations of the ORFs described by the Sanger Centre, the direction of their transcription, and the size of intergenic regions, as discussed in more detail below. Individual probes and primers for the transcription assays were designed using the sequence information provided by the Sanger Centre. The oligonucleotide primers used to generate each probe are given in Table 1. The S. coelicolor strains used to make RNA were C542 (Materials and Methods), an Abs⁻ strain (1, 2, 4); J1501, the Abs+ parent of C542; and C500 (Materials and Methods), a Pha absA disruptant (5).

Arrangement of genes in region of cda locus and physical association of absA with cda gene cluster.

Figure 1 shows a 98-kb region that surrounds the DNA that was previously predicted to encode the peptide synthetases involved in CDA biosynthesis 19; Sanger Centre database). The sizes of ORFs and directions of gene transcription that are shown are based on the sequence information and annotations provided to the public by the Sanger Centre Streptomyces coelicolor Genome Sequencing Project (http://www.sanger.ac.uk/Projects/S_coelicolor/). Figure 1 also shows the locations of promoters that we have predicted to function in the transcription of the cda gene cluster.

FIG. 1.

Organization of the ORFs surrounding the CDA peptide synthetase genes (PSI, PSII, and PSIII) from data supplied by the Sanger Centre. Solid arrows denote the direction of promoters detected by S1 mapping, while dotted arrows denote positions where attempts were made to detect transcription using S1 mapping but no signal was detected. Open boxes covering the junctions of two ORFs denote an overlapping of predicted start and stop codons for the adjacent ORFs.

From the sequencing data we discovered that the absA1/absA2 genes lay 4 kb from the peptide synthetase genes. ORFs in the intervening region between absA and the peptide synthetase encoding DNA, left of absA and right of the peptide synthetase DNA, were analyzed by comparing their amino acid sequences with previously characterized sequences present in the databases (Materials and Methods). In most cases the deduced functions for the ORFs (Table 3) agreed with those given in the annotation of the sequence by the Sanger Centre, with exceptions given in the text below.

TABLE 3.

Putativefunctions of ORFs surrounding CDA peptide synthetases

ORF name	Putative function
E22.16c	Efflux protein; MFS
E22.17	Transcriptional regulator; DeoR family
E22.18c	Transcriptional regulator; TetR family
HrdD	Sigma factor
BarA	Phosphinothricin resistance
E22.21c	Unknown
E22.22	Transcriptional regulator; MarR family
E22.23c	Efflux protein; MFS
E22.24c	Transcriptional regulator; TetR family
E8.01	Secreted protein
E8.02	Transcriptional regulator; IclR family
E8.03	DAHP synthasea
TrpC2	Indole-3-glycerol phosphate synthase
TrpD2	Anthranilate phosphoribosyltransferase
TrpG	Anthanilate synthase component II
TrpE2	Anthranilate synthase component I
E8.08c	Possible methyltransferase
E8.09	Cation-transporting ATPase
CdaR	Transcriptional regulator; SARP family
E8.11c	Unknown
E8.12c	Possible lipase
E8.13c	Unknown
E8.14c	Prephenate dehydrogenase
E8.15c	Unknown
E8.16c	ABC transporter integral membrane protein
E8.17c	ABC transporter ATP-binding protein
AbsA1	Sensor kinase
AbsA2	Response regulator
E8.20c	p-Hydroxyphenylglycine transaminase
E63.05c	Hydroxymandelate oxidase
E63.04c	Hydroxymandelate synthase
CDA PSI	Peptide synthetase
CDA PSII	Peptide synthetase
CDA PSIII	Peptide synthetase
E29.02	Hydrolase
E29.03	Unknown
E29.04c	ABC transporter
E29.05c	Oxygenase
E29.06c	Unknown
E29.07c	Unknown
E29.08c	Unknown
E29.09c	Unknown
E29.10c	Unknown
E29.11c	Unknown
E29.12c	Myo-inositol-1-phosphate synthase
E29.13c	Unknown
E29.14c	Salicylate hydroxylase
E29.15c	Beta-ketoacyl-acyl carrier protein synthase III
E29.16c	Acyl coenzyme oxidase
E29.17c	Beta-ketoacyl synthase II
E29.18c	Acyl carrier protein

^aDAHP, 3-deoxy-d-arabinoheptulosonic acid 7-phosphate.

An Abs⁻ mutant that fails to produce CDA fails to express the transcript encoding the peptide synthetases.

Figure 1 shows the DNA region encoding the peptide synthetase enzymatic activity involved in CDA biosynthesis. The sequence information released by the Sanger Centre defined three ORFs, PSI (7,463 amino acids [aa]), PSII (3,670 aa), and PSIII (2,416 aa), with similarity to previously characterized peptide synthetases (12). To confirm the involvement of this DNA region in CDA production, we created a disruption mutation in the PSI ORF. We transformed J1501 with a plasmid containing a section of DNA that was internal to the PSI coding region (pNJR123; see Materials and Methods), selecting for kanamycin resistance and hence the disruption of PSI by integration of the plasmid through insert-directed homologous recombination. The resulting strain (NJR123) was shown to be unable to synthesize CDA using bioassays (data not shown).

We predicted that a start point for transcription of PSI (designated p2 in Fig. 1) would lie within the 269-bp region that separated the predicted start codon for PSI from the predicted start codon for SCE63.04c, the divergently transcribed ORF. We therefore used the oligonucleotides WC123 (radiolabeled at the 5′ end) and WC114, both designed against the genomic sequence, to create a PCR product suitable for S1 nuclease protection assays. The RNA used in these assays was isolated from the antibiotic-deficient absA mutant (C542) and the parental strain (J1501) 24, 48, and 72 h after inoculation of SpMR agar covered with cellophane. Under these conditions CDA production in J1501 commenced between the 24- and 48-h time points, while red pigment production began a few hours before aerial mycelium formation that occurred at around 48 h after inoculation. Quantitative S1 nuclease protection assays using this RNA showed that transcripts from p2, originating from a point approximately 70 bp upstream of the predicted start codon for PSI, were readily detectable in RNA isolated from J1501 but were undetectable in RNA isolated from C542 (Fig. 2A). As a control to evaluate the quantity and integrity of RNA used in this experiment, a probe for glucose kinase (Glk) transcription was used (Fig. 2A), as has been described previously (1, 5). Replicate experiments using independently isolated RNA time courses showed very similar results, as will be discussed in more detail below. The lack of expression of a transcript required for CDA production explains the CDA⁻ phenotype of C542. This result was also consistent with a previously proposed hypothesis (5) that, in Abs⁻ absA mutants, the transcriptional regulator AbsA2 is locked into a negatively regulating state (possibly directly repressive).

FIG. 2.

S1 nuclease protection assays of RNA isolated from C542 and J1501 24, 48, and 72 h after inoculation of SpMR solid medium. Each set of assays was repeated at least twice. Assays of transcription from a Glk probe are given as an internal positive control. Nucleotide (nt) lengths are given for the radiolabeled φX174 HaeIII DNA used as a size standard. Each panel represents a separate experiment to detect transcription from the promoter given (see Fig. 1 for location). (A) p2, the promoter upstream of the peptide synthetase genes; (B) p1, the promoter upstream of the putative fatty acid biosynthesis genes; (C) p3, upstream of the hydroxymandelate synthase, and p7, the promoter upstream of the trp genes; (D) p4, the absA promoter, and p5, the promoter upstream of the putative ABC transporter, transcribed divergently to absA. Primers used to make each probe are given in Table 2.

An Abs⁻ mutant fails to express several putative CDA biosynthetic transcripts.

To determine whether absA regulated additional transcripts beside the peptide synthetase-encoding transcript described above, we examined the sequence data for ORFs that might also be involved in CDA biosynthesis. Right of PSI, PSII, and PSIII (as shown in Fig. 1) lay four ORFs that showed similarity to genes involved in fatty acid metabolism, suggesting that they may be required for synthesis of the lipid portion of CDA. SCE29.18c, SCE29.17c, and SCE29.15c were similar to acyl carrier proteins, beta-ketoacyl synthase II enzymes, and beta-ketoacyl–acyl carrier protein synthase III enzymes from a large number of bacteria, respectively. In contrast, SCE29.16c was found to share homology with a number of putative acyl coenzyme A oxidases, the vast majority of which were from eukaryotic species. S1 nuclease protection assays using the RNA samples mentioned above and a probe designed to detect transcriptional start points upstream of the start codon for SCE29.18c (p1 in Fig. 1) showed that transcription from a promoter in this location was readily detectable in RNA isolated from J1501 but was undetectable in RNA isolated from C542 (Fig. 2B).

Two regions of the cluster appear to be involved in the synthesis of amino acid precursors that are themselves incorporated into CDA. The three ORFs divergent to the peptide synthetases (SCE8.20c, SCE63.05c, and SCE63.04c), downstream of the promoter region designated p3 in Fig. 1, encoded proteins homologous to a p-hydroxyphenylglycine transaminase (HpgT; 56.0% identity), hydroxymandelate oxidase (Hmo; 55.5% identity), and 4-hydroxymandelate synthase (HmaS; 59.4% identity), respectively, which are encoded in the biosynthetic cluster for chloroerenomycin in Amycolatopsis orientalis (30, 43). These enzymes are predicted to act in conjunction with prephenate dehydrogenase (encoded in the CDA cluster in SCE8.14c and in the chloroerenomycin cluster by ORF1) (30, 43) to synthesize p-hydroxyphenylglycine, a nonproteinogenic amino acid that is found in the final structure of CDA (32), as well as chloroerenomycin and other members of the vancomycin group of antibiotics.

A series of ORFs on the leftward side of absA, downstream of the promoter region designated p7 in Fig. 1, included homologs of enzymes required for the synthesis of tryptophan, an amino acid incorporated twice into the structure of CDA (32). These enzymes included TrpE, TrpG, TrpD, and TrpC, which would be expected to carry out three of the four reactions required to convert chorismate to indole-3-glycerol phosphate, with the exception being the reaction carried out by N-phosphoribosylanthranilate isomerase (TrpF). There are, however, additional trp genes at other locations around the chromosome that may represent the set required for synthesis of tryptophan during growth, many of which have been previously characterized (25, 29).

S1 nuclease protection assays of the set of RNA isolated from C542 and J1501 (mentioned in the previous section) showed that transcripts from a promoter upstream of SCE63.04c (p3 in Fig. 1) and a promoter upstream of SCE8.08c (p7 in Fig. 1) could be readily detected in RNA isolated from J1501 but could not be detected in RNA from C542 (Fig. 2C). This indicated that the genes in the cluster that we propose are required for the synthesis of tryptophan and p-hydroxyphenylglycine, two of the predicted substrates for the CDA peptide synthetases, are also subject to negative regulation by absA.

The transcript encoding absA and a divergent transcript possibly encoding an ABC transporter are positively regulated by absA.

Previous work (5) showed that the absA two-component system is transcribed from a single promoter, designated p4 in Fig. 1, and that transcription of absA increased during growth. An Abs⁻ mutant (C542) exhibited increased levels of absA transcript relative to the wild type, while nonfunctional mutations in absA (resulting in the production of an unphosphorylable form of AbsA2) caused a significant reduction in absA transcription, leading to the hypothesis that absA positively autoregulates, although whether this autoregulation was direct or indirect was not determined (5). This pattern was also seen following analysis of absA transcription in the set of RNA described in the previous section (Fig. 2C) and in a second set that was isolated from C542, J1501, and a strain in which absA2 had been truncated, C500, at 18, 24, 30, 42, and 48 h after inoculation of SpMR agar (see Fig. 4). The Abs⁻ mutation used (absA1-542) caused greatly increased expression of absA relative to the wild type, while the disruption of absA2 in C500 caused a failure to activate absA transcription above a basal level (see Fig. 4). In S1 analysis of the second set of RNA, quantification of hrdB transcript from a promoter previously described (9) was used as a positive control to assess the amount and quality of RNA in each sample (Fig. 3).

The first two ORFs transcribed from a promoter divergent to that of absA (from p5, shown in Fig. 1) are predicted to encode an ABC transporter. One of the ORFs, SCE8.17c, showed close similarity to ATP-binding proteins of ABC transporters responsible for self-immunity to lantibiotics, antimicrobial peptides produced ribosomally by a number of gram-positive bacteria that are modified posttranslationally to contain the residue lanthionine (e.g., 38% identity to MrsF from the mersacidin cluster of Bacillus sp. strain HIL Y-85,54728 [3]; and 42% identity to NisF from the nisin cluster of Lactococcus lactis [20]). SCE8.16c showed little similarity to characterized proteins in the databases, but analysis using the TMPred program (http://www.ch.embnet.org/software/TMPRED_form.html) indicated the presence of six membrane-spanning regions, an arrangement typical for such integral membrane domains (35).

Transcription from a promoter (p5 in Fig. 1) upstream of the predicted ABC transporter (SCE8.17c, and SCE8.16c) was quantified in both sets of RNA mentioned in the previous section using S1 nuclease protection assays. This showed that, in contrast to the effects of absA mutations on promoters upstream of the predicted biosynthetic genes (positions p1, p2, p3, and p7 in Fig. 1), transcription of SCE8.17c was greatly increased in C542 relative to J1501 (Fig. 2D) and was significantly reduced in C500, the strain in which AbsA2 had been disrupted (Fig. 3), as was the case for transcription of absA itself. A similar result was obtained using quantitative RT-PCR analysis of the second set of RNA, amplifying a section of transcript from SCE8.16c (Fig. 4). This suggested that AbsA2 was also required for activation of transcription from the promoter divergent to that of absA and that the overphosphorylation of AbsA2 (that we believe takes place in C542) may therefore cause the overexpression of one of the putative ABC transporters in the cluster (SCE8.17c and SCE8.16c).

absA effects on transcription of cdaR, a SARP family homolog of pathway-specific regulators.

One potential transcriptional regulator associated with the cda cluster was an ORF named CdaR, which showed significant homology to members of the SARP protein family. This family includes the pathway-specific regulators from the Act and Red clusters (actII-ORF4 and RedD, respectively) (44) from S. coelicolor, in addition to a number of regulators found within antibiotic gene clusters from a variety of Streptomyces species. Experiments on regulation of many of these clusters, in particular the well-characterized act and red clusters, have shown that the SARP family regulators act as activators of transcription of the biosynthetic genes in the cluster. Therefore, one possible explanation for the absence of CDA biosynthetic transcripts in the Abs⁻ C542 strain (as noted in Fig. 2) was that the absA1-542 mutation affected transcription of the cdaR gene.

To assess cdaR transcription, we first determined by S1 nuclease protection assays that the gene was transcribed from a single transcriptional start site, designated p6 in Fig. 1, using a probe that covered 439 bp upstream of the proposed translational start point for cdaR. We then evaluated expression from p6 in absA mutants. Figure 3 shows that the cdaR transcript accumulated slightly earlier in the strain C542 versus J1501 versus C500, but overall the differences between the strains were slight in comparison with the effect on transcription from the putative biosynthetic promoter p3. Similar experiments on independently isolated RNA time courses have revealed little evidence for regulation of cdaR by absA. Thus, the negative regulation of CDA production seen in C542 does not result from a negative effect of the absA1-542 mutation on cdaR expression. These results suggested that the AbsA2 protein directly regulates the cda biosynthetic promoters and in its negatively regulating form overrides any positive regulation that may be exerted by cdaR.

We note that in this work we have not examined the effects of cdaR per se in the regulation of cda transcripts. However, the expression profiles of the cdaR transcript and that of the putative biosynthetic genes expressed from p3 are similar in J1501 RNA (see Fig. 3), a result that is consistent with an activation role for cdaR. We also noted that there is a substantial lag after biosynthetic transcripts are detected before CDA biosynthetic activity is detected.

Figure 3 includes an assessment of the expression of the SARP pathway-specific activator of the red gene cluster, RedD, as well as production of the Red antibiotic in the culture. In J1501 the redD transcript is just detected at 24 h, as is the case for cdaR, and then increases considerably in the later time points. The Red antibiotic itself is detected somewhat later than CDA. In the C542 mutant, the Red antibiotic was not detectable, as has previously been reported (2). Also consistent with previous results (1) was the reduced but detectable level of RedD transcription in C542. In RNA from C500, the quantity of transcript detected in the 24- and 30-h time points was similar to that in samples from J1501, but redD transcription in C500 had greatly increased by the 42- and 48-h time points (Fig. 3). It had previously been inferred that the reduction in act and red biosynthetic transcription in C542 was due to reduced expression of the pathway-specific activators, actII-ORF4 and redD. The results of this work therefore indicate that absA effects on regulation of cda differ from absA effects on regulation of act and red.

Timing of transcripts in cda cluster.

In Fig. 2, all of the putative biosynthetic promoters, p1, p2, p3 and p7, show similar profiles of expression in J1501 RNA, all being present from 24 h onwards. In this particular experiment, a time point earlier than 24 h had not been obtained, so we did not have evidence as to the onset of expression of these promoters. However, examination of the second set of RNA (Fig. 3) from C542, J1501, and C500 showed that the onset of expression from p3, the promoter upstream of the predicted hydroxyphenylglycine genes, occurred in J1501 between 18 and 24 h, as was the case for cdaR. This set of RNA samples was subjected to RT-PCR analysis to compare the timing of expression of an additional putative biosynthetic transcript. Figure 5 shows RT-PCR analysis of an ORF likely transcribed from p7 (trpE2) as well as the ORF likely transcribed from p3 (SCE63.04c). In both cases, transcript was detected at 24 h but not at the earlier time point. Another interesting feature of this set of results was that expression of trpE2 and SCE63.04c appeared to decline in J1501 by the 42- and 48-h time points, but in C500 expression of these ORFs slightly increased at 42 and 48 h relative to the earlier time points. This suggested that the loss of AbsA2 allowed continued expression of these ORFs. Transcription of the absA regulatory system itself differs noticeably from the transcripts for the biosynthetic genes and cdaR. In the RNA time courses shown in both Fig. 2 and 3, transcription from the absA promoter p4 and from the divergent promoter p5 was detected in J1501 RNA later than transcript from the biosynthetic promoters. Expression of the putative integral membrane protein (SCE8.16c) downstream from p5 was assessed in the second set of RNA by RT-PCR (Fig. 5), showing more clearly than in the S1 protection assays a temporal delay in J1501 in its onset relative to the two biosynthetic transcripts examined. Overall this result suggests that absA activation, which is dependent upon absA itself (5), occurs after CDA biosynthesis has commenced. In other words, the negative regulatory effect of AbsA2 on CDA synthesis may serve to modulate cda expression in the wild type, perhaps in competition with cdaR putative activation (see Discussion).

Transcription of a putative efflux protein is moderately affected by mutations in absA.

The protein encoded by SCE22.23c showed homology to a large number of predicted members of the major facilitator superfamily (MFS), in particular the drug:H+ antiporter family containing 14 transmembrane domains (DHA14) (44). Transcription from a promoter upstream of SCE22.24c (p9 in Fig. 1), the ORF encoding a putative TetR homolog that lay upstream of the putative MFS transporter (SCE22.23c), was noticeably reduced in C542 relative to J1501 (Fig. 5). Further experimentation will be required to determine whether SCE22.23c is involved in the export of CDA and whether its expression is regulated by the product of SCE22.24c, as is the case for actII-ORF1/ORF2 in S. coelicolor (11) and tcmR/tcmA in Streptomyces glaucescens GLA.0 (24).

Additional candidates for regulatory genes associated with the cda cluster are not regulated by absA.

Other potential regulatory genes in the cluster included SCE8.02 (for which we predicted a transcriptional start site at p8ii in Fig. 1), which showed significant homology with members of the IclR family of transcriptional regulators. However, the structure of SCE8.02 was unusual, in that both the N-terminal half and the C-terminal half showed similarity to members of the IclR family and in both cases more strongly to some of the “outside” members of the IclR family than to each other. This duplication may indicate that SCE8.02 is the site of a genetic rearrangement and perhaps is not an active gene.

The region from SCE22.24c to SCE22.16c (the top line of Fig. 1) contained five possible transcriptional regulators. Two of these, SCE22.24c and SCE22.18c, encoded small proteins that showed homology to members of the TetR superfamily, although both of these ORFs were located upstream of potential efflux proteins of the MFS. Examination of the homologs of the putative MFS genes associated with the cda cluster (SCE22.24c and SCE22.18c) showed that approximately half the ORFs in the S. coelicolor genome that are similar are associated with an ORF encoding a TetR-like protein. SCE22.22 (for which we predicted a transcriptional start site at p10 in Fig. 1) encoded a member of a family of transcriptional regulators typified by MarR, the repressor of the multiple antibiotic resistance locus (marRAB) of E. coli (41). In particular, SCE22.22 showed striking similarity (65% identity at the amino acid level) to MmcW from the biosynthetic cluster for mitomycin C in Streptomyces lavendulae (34). SCE22.17 (for which we predicted transcriptional start site p13 in Fig. 1) encoded a protein with 51% identity to DnrO, a regulator of daunorubicin biosynthesis in Streptomyces peucetius (37, 38).

Attempts to detect transcription from the regions upstream of the ORFs encoding the putative MarR family member (SCE22.22), the homolog of DnrO (SCE22.17), and the apparently duplicated IclR family member (SCE8.02; line 2 of Fig. 1) using S1 nuclease protection assays all failed to reveal detectable transcript in multiple repeated attempts (data not shown). Transcription of SCE22.22 was detected using quantitative RT-PCR (when 35 temperature cycles were used), but the quantity of product did not vary significantly across the time courses of RNA from C542, J1501, and C500 (data not shown). A product internal to SCE22.17 was amplified under similar conditions, but in this case a product was also detected in controls where RT had been omitted, suggesting that the product amplified in the test samples may have derived from a trace amount of contaminating DNA present in the samples (data not shown). Together these results suggest that the strong regulation of cda transcripts observed in Fig. 2 to 4 is not mediated through any of the possible regulatory genes discussed in this section.

The sigma factor HrdD encoded adjacent to the cda cluster does not regulate CDA biosynthesis.

The previously cloned and sequenced gene hrdD is one of four sigma factors of the ς⁷⁰ family that were detected by hybridization of the rpoD gene of Myxococcus xanthus to S. coelicolor DNA (9). Although RNA polymerase holoenzyme containing HrdD was found to be capable of initiating transcription in vitro from the promoters of both actII-ORF4 and redD (22), a derivative of J1501 in which hrdD had been disrupted (J1958) was not affected in the production of Act, Red (9, 10), or CDA (this study; data not shown). S1 nuclease protection assays of transcription from the hrdD promoter previously described (9) in the set of RNA isolated from C542, J1501, and C500 showed that transcription of this gene was not significantly affected by mutations in absA (data not shown) and therefore that regulation of cda transcripts by absA was not through regulation of hrdD expression.

DISCUSSION

We identified potential targets for absA regulation in the cda cluster through examination of the genomic sequence around the CDA peptide synthetases. This sequence analysis suggested that the cda cluster was composed of a relatively small number of transcriptional units, although we have not attempted to show that all of the constituent ORFs in each case were cotranscribed. The study of the effects of absA on cda genes was based on the quantification of cda transcripts in RNA isolated from the Abs⁻ mutant C542, the pha mutant C500, and the parental strain J1501. Previous genetic study of absA had formed the basis for a model predicting that the absA1-542 mutations (located in the sensor kinase) caused the overall inhibition of antibiotics through the constitutive phosphorylation of AbsA2 (4, 5). Therefore, analysis of cda transcripts in RNA isolated from C542 was intended to model the effects of an increase in the concentration of the phosphorylated from of AbsA2, while analysis of RNA from C500 showed the effects of loss of AbsA2 activity. S1 nuclease protection assays showed that transcripts from promoters upstream of the likely biosynthetic operons (p1, p2, p3, and p7 in Fig. 1) were reduced in quantity in RNA isolated from C542 to the point of being undetectable by this method. This took place without a commensurate effect on transcription of the putative pathway-specific regulator for the cda cluster, cdaR, a gene that shows sequence similarity to the pathway-specific regulators of the act and red clusters. We should note that the work described in this paper has not been directed at demonstrating the role of CdaR in regulation of the cda cluster, a relationship that will require investigation elsewhere. Reports of negative regulation of antibiotic biosynthesis genes by pathway-linked genes are uncommon, being limited largely to the examples of mmyR from the Mmy cluster (15), dnrO from the daunorubicin cluster (36, 37), and jadR2 in the jadomycin cluster of Streptomyces venezuelae (45).

To examine the possibility that negative regulation of the cda promoters by absA was mediated by transcriptional control of a putative regulator in the cda cluster other than cdaR, we attempted to quantify transcription of the ORFs nearby that also showed sequence similarity to characterized transcriptional regulators. Assays for transcription of the IclR family regulator (SCE8.02), the MarR family regulator (SCE22.22), and the homolog of DnrO (SCE22.17) using both S1 nuclease protection methods and RT-PCR did not allow the detection of significant quantities of transcript of any of these three ORFs, and so we may conclude that they are not active regulators under the conditions tested. It seems quite unlikely that the strong negative regulation of the cda biosynthetic promoters by absA would be mediated by transcriptional control of a regulator outside the cluster, and so we may presume at this point that the effect of AbsA2 on those promoters is direct, although this point will obviously need to be clarified by direct examination of binding of AbsA2, probably in its phosphorylated state, to the biosynthetic promoters of the cda cluster.

While absA1-542 caused negative regulation of the cda biosynthetic transcripts, we also noticed a high level of transcription from the absA promoter itself, p4, and the divergent promoter p5 (upstream of a putative ABC transporter, SCE8.17c, and SCE8.16c) throughout the time course of RNA from C542. In the mutant in which AbsA2 had been truncated (C500), there was a loss of activation of transcription in both directions in the divergent absA promoter. Overall, this suggested that the phosphorylated form of AbsA2 activates transcription of SCE8.17c as well as absA itself. The sequence similarity of SCE8.17c to resistance proteins in bacteriocin clusters raises the possibility that it is required for protection of S. coelicolor against the effects of CDA. Further experiments are required to investigate the significance of p5 regulation, including genetic experiments to examine the role of the ABC transporter in CDA biosynthesis and/or resistance.

Examination of the timing of transcription from the promoters upstream of the predicted p-hydroxyphenylglycine and tryptophan biosynthesis genes (p3 and p7 in Fig. 1) and from the divergent absA promoter provided some clues as to the normal role of absA in regulation of the cda cluster. One observation was that transcription from p3 and p7 in J1501 (in the second set of RNA) was initiated earlier than transcription from the divergent promoters between absA and SCE8.17c (Fig. 3 and 4). This effect was also seen in an S1 protection assay analysis of the first set of RNA, shown in Fig. 2. As AbsA2 appears to be the limiting factor in activation of absA transcription, the quantity of transcripts from the absA promoter can perhaps be used as an indicator for the presence of the active form of AbsA2, presumably generated following the binding of signal molecule by AbsA1. Therefore, it seems that AbsA2 became active in both time courses after expression of the CDA biosynthetic genes had already begun. The timing of the final export of CDA from J1501 and absA knockout mutants, shown by bioassays reported in a previous paper (5), had suggested that absA served to delay the onset of CDA biosynthesis, since the loss of AbsA2 activity allowed earlier production of CDA. Overall this may suggest that AbsA2 exerts a negative influence on production of CDA after the onset of transcription from the putative biosynthetic promoters examined in this study (p1, p2, p3, and p7) but before the final export of active CDA product. The issue of the timing of onset of biosynthetic transcripts from the act, red, and cda clusters, as well as the pathway-specific regulators in each case and absA, should perhaps be the focus of a much more detailed time course experiment.

A surprising aspect of this work was the discovery that absA was embedded within the cda cluster. The mutant screen that led to the isolation of absA mutants did not employ an effect on CDA as a primary criterion for mutant isolation. Rather it was a tight Act⁻ Red⁻ phenotype that was the primary isolation criterion, and the CDA⁻ phenotype was observed in a secondary screen. Yet the Abs⁻ mutation absA1-542 blocks cda transcription more strongly than it blocks act and red transcription. The results of the S1 nuclease protections assays presented here on cda transcription contrasted with the results of previous experiments in which it was shown that the quantity of transcripts of both act biosynthetic genes and the pathway-specific regulators (actII-ORF4 and redD) were reduced significantly but not abolished in C542 (1), underlining an asymmetry in the relationship between absA and cda and between absA and the other clusters in S. coelicolor. In this work we have proposed that AbsA2 directly represses cda promoters, perhaps in competition with cdaR, if cdaR does indeed function as a positive regulator. These results raise the question of whether AbsA2 regulates Act and Red biosynthesis through an indirect mechanism, rather than through a direct transcriptional regulation of act and red genes, in particular the actII-ORF4 and redD activators.

The effect of a regulator from one antibiotic gene cluster on the regulation of the other clusters in the same genome is novel, particularly in light of the possibility that the four clusters may have not originated in S. coelicolor. There is some evidence to suggest that antibiotic biosynthesis clusters have been disseminated among different species over an evolutionary period through horizontal gene transfer (20, 31). The clustering of the antibiotic genes themselves is perhaps an indication that the genes required for those pathways have already been subjected to a selective pressure for organization into discrete packages of DNA by genetic transfer. The possible movement of clusters between species over an evolutionary time frame is highly relevant to the consideration of the molecular interactions that may have evolved to link the regulation of those clusters to the regulatory hierarchy of the “host” organism.