A chromosomal DNA fragment from Streptomyces antibioticus ATCC11891 was isolated by its ability to stimulate actinorhodin and undecylprodigiosin biosynthesis in Streptomyces lividans TK21. This fragment includes two open reading frames, whose deduced translated products resemble enzymes involved in sulfur metabolism (ORF1) and LysR-type transcriptional regulators (ORF2). The cloning of the promoter region of ORF2 (abaB) in high copy number led to overproduction of both antibiotics suggesting that this phenotype might well be due to titration by this region of one or more proteins. Southern blot analysis revealed that abaB gene is highly conserved among all streptomycetes tested.
The great industrial importance of the streptomycetes for their ability to produce secondary metabolites has focused the efforts of many scientists on the study of antibiotic biosynthesis and its regulation in these bacteria. Streptomyces coelicolor A3(2) has been used for many years as a model for morphological and physiological differentiation studies, including aspects of the regulation of antibiotic biosynthesis . S. coelicolor produces four antibiotics: actinorhodin, undecylprodigiosin, methylenomycin and calcium-dependent antibiotic (CDA). With the exception of CDA, for which information is not yet available, the biosynthetic genes involved in each particular pathway are known to be clustered along with a pathway-specific regulatory gene. In addition to these linked regulatory elements, antibiotic production is also controlled by pathway-non-specific regulators, which are located outside the clusters and may pleiotropically control the production of more than one antibiotic.
Several pleiotropic regulators have been described in Streptomyces that affect morphological differentiation (bld genes)  and/or secondary metabolite production. Some of the latter regulators control antibiotic production globally (absA, absB) [2, 3], or regulate only some of the antibiotic biosynthetic pathways (abaA, afsR, afsR2, afsQ1/afsQ2, a putative antisense regulator) [4–8]. The diversity of these regulatory elements makes it likely that several mechanisms might be involved in regulating antibiotic production; the combination of all these regulatory mechanisms, and the still unknown connections between different levels of regulation in Streptomyces, make this topic challenging for further understanding of the control of secondary metabolite production.
In the present study, we attempted to contribute to the general understanding of the complex network of antibiotic biosynthesis regulation in Streptomyces by looking for new heterologous DNA sequences capable of stimulating antibiotic production in S. lividans TK21. As DNA donor we chose S. antibioticus ATCC11891 for its ability to produce polyketide antibiotic (oleandomycin) and S. lividans TK21 as recipient strain. As a result of our study, we describe a novel pleiotropic regulator, which, unlike other previously characterized genes, seemed to stimulate antibiotic production when its promoter region is propagated on a high copy number plasmid.
2Materials and methods
2.1Bacterial strains and plasmids
Escherichia coli JM101 was used for cloning DNA fragments on plasmid pUC19 and propagation of phage M13mp18/19 for DNA sequencing . S. antibioticus ATCC11891 was used as DNA donor for making the library and S. lividans TK21 as recipient; S. coelicolor J802 and S. coelicolor J1501 were used to study stimulation of methylenomycin biosynthesis . Micrococcus luteus ATCC 10240 was used to detect CDA production as an alternative to the Bacillus strain used previously .
2.2Media, culture conditions, microbiological procedures and enzymatic determinations
Streptomyces strains were grown on R2YE agar medium or in YEME liquid medium, and transformations were carried out as described previously . For actinorhodin and undecylprodigiosin production and total RNA extractions, strains were grown until complete sporulation on a cellophane sheet on the surface of R2YE agar medium . For xylanase and arabinofuranosidase determinations cultures were propagated in liquid media and enzymes determined essentially as described previously . When required for selection thiostrepton was used at a concentration of 50 μg/ml in agar medium and 5 μg/ml in liquid cultures. E. coli strains were grown on L agar or L broth and the microbiological procedures were as described elsewhere .
2.3DNA sequencing and computer analysis
DNA sequencing was carried out by the dideoxy-chain termination method . Suitable subcloning in the M13mp18 and mp19 derivatives allowed 100% DNA sequencing of both strands. DNA sequence and open reading frame analysis, protein alignments and secondary structure predictions were carried out using the software programs of the University of Wisconsin Genetics Computer Group (UWGCG; Version 7.3-AXP) . Comparisons of sequences were made against the EMBL nucleic acid data base (updated daily) and the Swissprot data base (updated weekly).
2.4DNA and RNA manipulations
For isolation, cloning and manipulation of nucleic acids, methods were those described previously [9, 10]. High-resolution S1 mapping was carried out following the method described previously .
2.5Antibiotic production assays
Production of the pigmented antibiotics actinorhodin and undecylprodigiosin was detected spectrophotometrically, and analyzed by HPLC and mass spectroscopy. After complete sporulation of S. lividans TK21, with or without the DNA activating sequences cloned in pIJ486, was reached in R5 solid medium , the agar was homogenized, acidified with HCl to pH 3.0 and extracted with ethyl acetate. The mycelium was extracted separately with methanol. For further antibiotic analysis, the organic fractions were vacuum dried and resuspended in methanol, diluted with either 11.6 M HCl or 10 M NaOH up to a final concentration of 500 mM. Spectra were carried out between 350 and 700 nm.
For HPLC analysis samples were chromatographed using a Shimadzu LC-9A HPLC. Spectral scans of selected peaks were conducted using a Waters-996 photodiode array detector and Millenium 2010 for processing. Samples extracted from agar plates were applied to a Nucleosil 120/5 C18 125 mm×4.6 mm column (Scharlau Science) and eluted with mobile phase I containing buffer (A) 0.02 M phosphoric acid/0.02 M NaOH, pH 3–(B) methanol, with a programmed gradient from 45% (A):55% (B) (0–2 min) to 5% (A):95% (B) (12 min). Samples extracted from mycelia were separated with a Nucleosil 120/9 C8 150 mm×4.6 mm column (Scharlau Science) and eluted with mobile phase II containing buffer (A) 0.02 M phosphoric acid/0.02 M NaOH, pH 3–(B) acetonitrile and a gradient from 50% (A):50% (B) (0–4 min) to 20% (A):80% (B) (21 min). All were run at a flow rate of 1 ml/min.
Mass spectrometry was carried out on a Hewlett Packard 5988A spectrometer linked to a gas chromatograph, solid probe (DIP) (for actinorhodin) or HPLC (for undecylprodigiosin) (through an HP `Particle Bean' 59980A interphase for liquids). Samples were exposed to an ionization energy of 70 eV at a source temperature of 250°C and 200°C for HPLC and DIP, respectively. Accurate masses were obtained at a resolution of 1000.
CDA production and stimulation of methylenomycin production were checked as described previously .
3.1Cloning of a DNA fragment capable of stimulating antibiotic production in S. lividans
A library of S. antibioticus ATCC11891 chromosomal DNA fragments, generated by partial Sau3AI digestion, was prepared by ligation into the BamHI site of pIJ486 . The ligation mixture was introduced by transformation into S. lividans TK21 and the recombinants selected with thiostrepton. Among the transformants, an intensely blue colony was isolated, and found to harbor a plasmid carrying a 4.3 kb DNA fragment from the S. antibioticus chromosome (see later). Its phenotype was shown to be due to the cloned fragment by retransformation of S. lividans TK21 with the recombinant plasmid; the plasmid was named pCNB2000 (Fig. 1).
The 4.3 kb insert was trimmed down and it was shown that only the 1.8 kb EcoRI-SphI fragment (pCNB2008) yielded blue colonies after transformation of S. lividans. One of the blue pigmented recombinants carrying pCNB2008 was used for further characterization. A more detailed restriction map was obtained for pCNB2008 (Fig. 1). pCNB2000 and pCNB2008 failed to complement actII-ORF4 and bldA mutations in S. coelicolor, suggesting that the cloned gene might well be a pleiotropic activator rather than a pathway-specific regulator.
3.2Southern blot analysis
Southern blotting against DNA of other regulatory genes (actII-ORF4, abaA, afsR or redD), using as probe the cloned DNA fragment of pCNB2008 on plasmid pUC19 (named pCNB2004) confirmed the lack of relatedness of the cloned DNA to these genes. Southern blots of BamHI-digested chromosomal DNAs from several Streptomyces species (S. coelicolor A3(2) (J1501), S. lividans TK21, S. ambofaciens ATCC 15154, S. violaceolatus 2910 and S. glaucescens ETH 22794) probed with pCNB2004 revealed a single hybridizing band in all chromosomal digests (data not shown). Bands of similar size and intensity were observed in S. coelicolor, S. lividans and S. glaucescens, and smaller bands were observed in S. violaceolatus and S. ambofaciens, the hybridizing band in the latter being much more intense than in the other. These results clearly indicated the existence of similar sequences well conserved in different Streptomyces species. The SphI-SmaI digest of chromosomal DNA of S. antibioticus ATCC11891 showed a band of identical size to that in the SphI-SmaI digest of pCNB2008, suggesting that the cloned DNA was colinear with the original S. antibioticus chromosome and did not result from rearrangement during the cloning experiments.
3.3DNA sequence of the 1.8 kb antibiotic-activating fragment
The DNA sequence of the 1.8 kb EcoRI-SphI fragment from pCNB2008 (Fig. 1) was determined (Fig. 2) and analyzed for open reading frames (ORFs) using the program CODONPREFERENCE  with the codon usage table for Streptomyces. Two possible ORFs, separated by 226 base pairs, were deduced from the DNA sequence and named ORF1 and ORF2. Both ORFs seemed to be incomplete in the sequenced fragment. Further DNA sequencing of pCNB2000 beyond the SphI site allowed the sequence of ORF2 to be completed.
ORF1 lacks the 5′ end of its DNA sequence; comparison of its deduced amino acid sequence with the data bases showed significant similarities with the carboxy-terminal end of two enzymes involved in sulfur metabolism, cystathionine γ-lyase (Cgl) from rat (P18757) and the cystathionine γ-synthase (MetB) from E. coli (P00935), reaching 73% and 71% similarity, respectively.
The most likely start codon for ORF2 was located at position 926–928 (see Fig. 2), as deduced from considerations of codon usage and the presence of a good putative ribosomal-binding site at an appropriate distance from the potential start codon. This putative start codon is preceded by an AG-rich sequence, which extends up to 30 nt. The deduced molecular mass of the ORF2 product would be 32 kDa.
Comparison of the deduced amino acid sequence of ORF2 with sequences in the data bases showed significant similarities with the amino-terminal ends of several LysR-type transcriptional regulators (Fig. 3), which include the well conserved helix-turn-helix motif known to be involved in the DNA-binding activity of members of this family . The calculated score obtained with the Dodd-Egan matrix  for the ORF2 product was SD=5.147 and the prediction of its secondary structure by the Chou-Fasman algorithm obtained using the PEPTIDESTRUCTURE program  makes it very likely that a helix-turn-helix motif exists at this position in the ORF2 product. Furthermore, 11 out of 11 conserved amino acid residues described for the helix-turn-helix motif of the LysR regulators  could be found at the amino-terminal region of ORF2. The predicted size of the translated ORF2 product (301 aa) is very close to the average size described for LysR regulators (300 aa) . The most relevant similarities were found with the regulator of the operon for ribulose bisphosphate carboxylase (RbcR) from Chromatium vinosum (P25544) and the regulator of the β-lactamase (AmpR) from Rhodobacter capsulatus (P14145), with up to 60% and 63% similarity, respectively; as in OxyR protein from E. coli and Salmonella typhimurium the AbaB protein contains five cysteine residues which are similarly arranged along the amino acid sequence.
3.4Identification of the antibiotic-activating sequence
Since the blue pigmented phenotype of S. lividans transformants was elicited by the 1.8 kb EcoRI-SphI fragment which does not code for a complete ORF, it can be suggested that the antibiotic-producing phenotype would very likely be induced by a DNA sequence rather than a gene product. Location of the activating sequence within a smaller fragment was achieved by digesting the cloned DNA with different restriction enzymes to give overlapping fragments which were ligated into pIJ486. The different ligation mixtures were used to transform S. lividans TK21 protoplasts. These experiments made it possible to reduce the antibiotic-activating DNA sequence to a 700 bp fragment, and showed that the intergenic region between ORF1 and ORF2 and a small part of the amino-terminal of ORF2 (at least 34 nt within the N-terminus of the coding region of ORF2) are essential for the blue pigmented phenotype.
3.5Transcriptional organization of the ORF2
The transcription start points of ORF2 were determined by high resolution S1 mapping. Total RNA from S. antibioticus ATCC11891 and S. lividans TK21 (carrying either pCNB2008 or the vector pIJ486) were hybridized with the 960 bp fragment (EcoRI-ApaLI) as probe from pCNB2004. The fragment was previously labelled (5′) at the ApaLI site with [γ-32P]ATP. After S1 digestion, two protected fragments of 120 and 84 bp respectively could be detected in RNA samples from S. antibioticus and S. lividans carrying pCNB2008 (but not when carrying only pIJ486) (Fig. 4), implicating nt 840 and 876 as transcription start points of ORF2 (Fig. 4). The 120 bp protected fragment was weaker than the 84 bp fragment, suggesting that the latter transcription start point showed more transcriptional activity under the culture conditions used.
3.6Characterization of the activating sequences on antibiotic production
In order to explore the role of the 1.8 kb fragment on antibiotic production, recombinants of S. lividans carrying pCNB2008 were tested for actinorhodin, undecylprodigiosin and CDA biosynthesis, while stimulation of methylenomycin biosynthesis was tested in S. coelicolor J802 carrying pCNB2008.
As shown by spectrophotometrical analysis, actinorhodin as well as undecylprodigiosin biosynthesis was greatly activated in S. lividans recombinants carrying pCNB2008. Samples extracted from agar plates and separated by HPLC showed several peaks with actinorhodin-like spectral patterns, suggesting the presence of different actinorhodin-like substances in the agar medium (Fig. 5A). Similar studies carried out with mycelial samples made it possible to identify up to three prodigiosin-like pigments with different retention times (Fig. 5B). Mass spectra confirmed the presence of actinorhodin (ion m/z=634.7), most probably the γ-actinorhodin, and undecylprodigiosin (ion m/z=393.5) produced by S. lividans TK21 carrying pCNB2008.
CDA biosynthesis was not activated as shown by bioassay against Micrococcus luteus ATCC10240. No stimulation of methylenomycin production could be observed in S. coelicolor J802.
Furthermore, the antibiotic activating DNA fragment did not show any effect either on sporulation in S. lividans TK21 or on the production of some extracellular enzymes tested (xylanase, arabinofuranosidase).
These results suggest that the DNA sequence of the 1.8 kb DNA fragment includes a putative pleiotropic antibiotic biosynthesis activator and thus we named this new pleiotropic regulator abaB ( ntibiotic iosynthesis ctivator).
From a genomic library of S. antibioticus ATCC11891 (as a polyketide producer) a DNA fragment was isolated that stimulates heterologously the polyketide antibiotic actinorhodin production in S. lividans TK21. Within the cloned fragment we have identified a new gene (abaB). The significant similarity of AbaB with LysR-type regulators would suggest an inducer-mediated transcriptional regulation mechanism based on DNA binding of this putative regulator to its target gene(s). Unfortunately, S. antibioticus could not be transformed in vitro with the standard procedure, so that the possible function of abaB in the donor strain could not be studied, but the presence of similar sequences in other Streptomyces suggests a general function. Moreover, the control of LysR-type regulators on the transcription of an adjacent gene, makes ORF1, whose amino acid sequence is highly similar to enzymes involved in sulfur metabolism, a putative target for AbaB.
Nevertheless, the antibiotic biosynthesis activation triggered by the subcloned DNA sequence rather than by the ORF1/ORF2 gene product, might be coincident with the autoregulatory mechanism of LysR-type regulators by the binding of the regulator to its own promoter or a nearby region. Thus, the 700 bp fragment from S. antibioticus capable of activating antibiotic biosynthesis in S. lividans would probably offer such DNA-binding region for an AbaB equivalent or other regulators operating in the host S. lividans and competing for the same DNA sequence.
Efforts are being made to clone the abaB equivalent from S. coelicolor, which would make it possible to study the role of such gene in a more suitable Streptomyces strain for a genetic work.
We thank J.M. Arribas from the Central Unit for Analytical Services of the University of Alcalá de Henares for kindly analyzing samples by mass spectroscopy and D.A. Hopwood for helpful discussion of the manuscript. This research was supported by grants from the Spanish CICYT (BIO 93-1181), European Union, (BIO2-CT94-2067) and Proyecto Concertado SmithKline and Beecham, S.A./CSIC (CDTI, No. 930139).