Pleiotropic functions of a Streptomyces pristinaespiralis autoregulator receptor in development, antibiotic biosynthesis, and expression of a superoxide dismutase.

In Streptomyces, a family of related butyrolactones and their corresponding receptor proteins serve as quorum-sensing systems that can activate morphological development and antibiotic biosynthesis. Streptomyces pristinaespiralis contains a gene cluster encoding enzymes and regulatory proteins for the biosynthesis of pristinamycin, a clinically important streptogramin antibiotic complex. One of these proteins, PapR1, belongs to a well known family of Streptomyces antibiotic regulatory proteins. Gel shift assays using crude cytoplasmic extracts detected SpbR, a developmentally regulated protein that bound to the papR1 promoter. SpbR was purified, and its gene was cloned using reverse genetics. spbR encoded a 25-kDa protein similar to Streptomyces autoregulatory proteins of the butyrolactone receptor family, including scbR from Streptomyces coelicolor. In Escherichia coli, purified SpbR and ScbR produced bound sequences immediately upstream of papR1, spbR, and scbR. SpbR DNA-binding activity was inhibited by an extracellular metabolite with chromatographic properties similar to those of the well known gamma-butyrolactone signaling compounds. DNase I protection assays mapped the SpbR-binding site in the papR1 promoter to a sequence homologous to other known butyrolactone autoregulatory elements. A nucleotide data base search showed that these binding motifs were primarily located upstream of genes encoding Streptomyces antibiotic regulatory proteins and butyrolactone receptors in various Streptomyces species. Disruption of the spbR gene in S. pristinaespiralis resulted in severe defects in growth, morphological differentiation, pristinamycin biosynthesis, and expression of a secreted superoxide dismutase.

In Streptomyces, a family of related butyrolactones and their corresponding receptor proteins serve as quorum-sensing systems that can activate morphological development and antibiotic biosynthesis. Streptomyces pristinaespiralis contains a gene cluster encoding enzymes and regulatory proteins for the biosynthesis of pristinamycin, a clinically important streptogramin antibiotic complex. One of these proteins, PapR1, belongs to a well known family of Streptomyces antibiotic regulatory proteins. Gel shift assays using crude cytoplasmic extracts detected SpbR, a developmentally regulated protein that bound to the papR1 promoter. SpbR was purified, and its gene was cloned using reverse genetics. spbR encoded a 25-kDa protein similar to Streptomyces autoregulatory proteins of the butyrolactone receptor family, including scbR from Streptomyces coelicolor. In Escherichia coli, purified SpbR and ScbR produced bound sequences immediately upstream of papR1, spbR, and scbR. SpbR DNA-binding activity was inhibited by an extracellular metabolite with chromatographic properties similar to those of the well known ␥-butyrolactone signaling compounds. DNase I protection assays mapped the SpbR-binding site in the papR1 promoter to a sequence homologous to other known butyrolactone autoregulatory elements. A nucleotide data base search showed that these binding motifs were primarily located upstream of genes encoding Streptomyces antibiotic regulatory proteins and butyrolactone receptors in various Streptomyces species. Disruption of the spbR gene in S. pristinaespiralis resulted in severe defects in growth, morphological differentiation, pristinamycin biosynthesis, and expression of a secreted superoxide dismutase.
Diffusible pheromones often coordinate expression of specific genetic programs within a population of bacteria as they reach high cell density. Pioneering studies leading to the discovery of ␥-butyrolactone signaling molecules were made in Streptomyces, Gram-positive filamentous bacteria characterized by a densitydependent developmental program that includes aerial mycelium formation and the biosynthesis of secondary metabolites often having antibiotic activity. Khokhlov et al. (1,2) demonstrated that in Streptomyces griseus, these developmental programs could be coordinated by nanomolar concentrations of a molecule they identified as a ␥-butyrolactone and named "A-Factor." The chemical structures of many species-specific Streptomyces butyrolactones have since been determined, along with a growing family of putative receptor proteins (3)(4)(5)(6)(7). To facilitate subsequent discussions of the ␥-butyrolactone signaling system, its components and their functions are diagrammed in Fig. 1.
Many ␥-butyrolactone receptors (GABRs) 1 bind to a conserved nucleotide motif (autoregulatory element) (ARE)) and thus act as repressors of transcription (8). The butyrolactones that accumulate in culture media are thought to act as quorum signaling molecules by releasing their corresponding GABR proteins from operator sites, thus activating gene expression (6).
Developmental systems under butyrolactone control have been best characterized in S. griseus and Streptomyces virginiae. Studies of Horinouchi, Beppu, and co-workers (9) support a model describing how A-Factor and its receptor in S. griseus, ArpA, mediate pleiotropic effects on development.
Binding of A-Factor to ArpA derepresses expression of a transcriptional activator, AdpA (10). AdpA promotes expression of strR, the activator of streptomycin biosynthetic genes, and other unknown genes that control aerial mycelium formation. In S. virginiae, butyrolactones and a corresponding receptor (BarA) take part in regulating synthesis of a streptogramin complex called virginiamycin (11) via unknown regulatory pathways. Related Streptomyces antibiotic regulatory proteins (SARPs) commonly activate expression of biosynthetic gene clusters. Thus, SARPs are potentially the ultimate target for some quorum-sensing signaling pathways that switch on antibiotic biosynthesis (12).
Gram-negative bacteria employ quorum-sensing systems based on homoserine lactones, structurally related to ␥-butyrolactones, to control a diverse array of density-dependent phenotypes (13,14). In Pseudomonas, quorum-sensing systems control synthesis of virulence factors as well as enzymes such as catalases that detoxify reactive oxygen species (ROS) and superoxide dismutase (15).
Superoxide dismutases are ubiquitous parts of cellular defenses against oxidative stress that catalyze dismutation of the toxic superoxide anion into hydrogen peroxide (H 2 O 2 ), thus preventing the spontaneous formation of more toxic forms of ROS by the Haber-Weiss reaction. The two classes of superoxide dismutases described in Streptomyces utilize either Ni 2ϩ (SodN) or Fe 2ϩ /Zn 2ϩ (SodF) as cofactors. Ni 2ϩ represses the constitutive expression of sodF expression and induces sodN (16).
ROS may come from endogenous metabolism or exogenous sources. Primary metabolic conversions that generate ROS are largely limited to flavin and flavoproteins that activate molecular oxygen (17). Such monooxygenases often participate in the respiratory chain within the cytoplasmic membrane. In contrast, pathogenic bacteria probably employ superoxide dismutase to defend themselves against external ROS they may encounter as part of the host antimicrobial response. The SodF released by Mycobacterium tuberculosis is thought to protect the organism from oxidative attack by macrophages (18).
Our studies involved S. pristinaespiralis, a saprophytic soil organism that produces pristinamycin, a clinically important streptogramin antibiotic complex. Like other streptogramins, it is a mixture of compounds based on two structurally dissimilar synergistic antibiotics. The streptogramin B component, pristinamycin I (PI), is a cyclic hexadepsipeptide; the streptogramin A compound, pristinamycin II (PII), is a polyunsaturated cyclic peptolide (19).
PI and PII biosynthetic genes are clustered together with papR1 (putative regulator of pristinamycin antibiotic production), the SARP gene described here. We purified SpbR (S. pristinaespiralis butyrolactone-responsive transcriptional repressor), a GABR protein that bound to a site upstream of the papR1 promoter, and showed that the corresponding gene was required not only for colonial development and antibiotic biosynthesis, but also for expression of a leaderless superoxide dismutase found as the major protein in the medium.

Transformation
Streptomyces protoplasts were transformed, spread on R2YE medium, and allowed to regenerate for 20 h at 30°C (20). Transformants were selected by overlaying the plates with 1 ml of aqueous apramycin (1 mg) or thiostrepton (0.3 mg). To enhance integration into the chromosome via homologous recombination, plasmids were alkali-denatured before protoplast transformation (21). E. coli cells were transformed by calcium shock or electroporation (22).

Growth Conditions
S. pristinaespiralis NRRL2958 spores (5 ϫ 10 8 ) grown on HT7T medium were inoculated in a 250-ml baffled flask containing 100 ml of minimal inoculum medium. After incubation on a rotary shaker at 200 rpm for 26 h at 30°C, the culture served as inoculum for a 2-liter pilot fermenter containing minimal inoculum medium (pH 6.8). The fermentation was carried out at 28°C with constant aeration. For strains containing plasmid pNL5 or pIJ904, the media were supplemented with thiostrepton (3 g/ml).

Gel Retardation Assay for DNA-binding Proteins
DNA fragments were end-labeled by filling in with Klenow fragment or fully labeled by PCR in the presence of [␣-32 P]dATP. This generated probes of different specific activities for use in gel retardation assays.
Crude or semipurified DNA-binding proteins (5 g) were incubated with radiolabeled DNA fragment (0.006 pmol of the filled-in probe and 0.06 pmol of the PCR-generated probe) in the presence of 1-5 g of competitor DNA (poly(dI-dC)⅐(dI-dC), Amersham Pharmacia Biotech) in 20 l of tris buffer containing additives (TA; 10 mM Tris (pH 7.5), 10 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, 0.1% Triton X-100, and 10% glycerol) supplemented with 250 mM NaCl. After incubation, the reaction mixture was resolved on 5% polyacrylamide gels in 90 mM Tris, 90 mM borate, and 2 mM EDTA (pH 8), run at room temperature and constant voltage (7 V cm Ϫ1 ) for 2 h. Gels were dried, and radioactive bands were recorded by autoradiography or on a PhosphorImager.
FIG. 1. The butyrolactone signaling cascade. ␥-Butyrolactones (q) and their regulators (GABR) can act as transcriptional repressors by binding to AREs. SARPs are needed for the expression of most antibiotic biosynthetic pathways. Studies of spbR and the papR1 promoter established a robust consensus sequence for GABR binding and that GABR proteins bind directly SARP promoters as well as undefined promoters needed for growth, development, and expression of the central oxidative stress adaptive enzyme superoxide dismutase (SOD) SodF.

SpbR Purification
Preparation of Crude Extracts-Stationary phase mycelia grown in HT7T medium were harvested by centrifugation, washed twice with 250 ml of TA buffer supplemented with 10 mM NaCl (TAN buffer), and stored at Ϫ80°C. All steps of the purification were carried out at 4°C. The mycelia (ϳ100 g), harvested from 15-liter cultures, were disrupted by sonication for 10 min at 90 watts (Sonifer B12) in 200 ml of TA buffer supplemented with 100 mM NaCl. Protease inhibitors (benzamidine, pepstatin, and leupeptin; 1 g/ml final concentration; Sigma) were added prior to sonication. After sonication, phenylmethylsulfonyl fluoride (10 mM) was added. Cell debris was removed by centrifugation for 45 min at 9000 rpm in a Sorvall GSA rotor (13,000 ϫ g).
Ammonium Sulfate Precipitation-Ammonium sulfate (Schwarz/ Mann) was slowly added to the cooled protein extract to a final concentration of 28% (w/v). The extract was clarified by centrifugation at 13,000 ϫ g for 60 min at 4°C in a Sorvall GSA rotor. Soluble proteins were precipitated using ammonium sulfate (60% (w/v) final concentration) and collected by centrifugation at 13,000 ϫ g for 60 min in a Sorvall GSA rotor. The pellet was resuspended in 150 ml of TAN buffer, dialyzed against 10 liters of the same buffer, and then clarified by centrifugation for 30 min in a Sorvall SS34 rotor at 12,000 rpm (17,300 ϫ g). The protein pellet was redissolved in TA buffer.
DEAE Anion-exchange Chromatography-The protein sample was then loaded onto a 150-ml DEAE-Sepharose column (XK 26/50 Amersham Pharmacia Biotech) equilibrated with TAN buffer. The column was washed with TAN buffer, and proteins were eluted in a 600-ml salt gradient of 10 -450 mM NaCl in TA buffer. An aliquot (4 l) of each 5-ml fraction was tested by gel mobility shift assays. The active fractions were pooled and supplemented with NaCl to make the conductivity equivalent to that of 100 mM NaCl in TA buffer.
Heparin Chromatography-Active DEAE fractions were loaded onto a 50-ml heparin-Sepharose CL-6B column (XK 26, Amersham Pharmacia Biotech) pre-equilibrated with 100 mM NaCl in TA buffer. The proteins were eluted in a 300-ml NaCl gradient (100 mM to 1 M) in TA buffer. Active fractions were concentrated using an Amicon YM-10 filter with a 10-kDa cutoff.
MonoQ Anion-exchange Chromatography-The active heparin fraction was loaded on a 1-ml MonoQ column (HR5/5, Amersham Pharmacia Biotech). SpbR activity was detected in the early fractions of a 60-ml NaCl gradient (100 mM to 1 M) in TA buffer. Fractions were stored in the elution buffer at 4°C.
DNA Affinity Chromatography (23)-pMF1, containing the region upstream of papR1 (nucleotides 113-435; accession number A37840), was digested with HindIII and end-labeled by filling in the overhanging ends with biotinylated dATP. The insert was released by cleavage with SmaI, purified (ϳ50 g), and mixed with streptavidin-coated magnetic beads. The beads were incubated with the active MonoQ fractions (5 ml, ϳ10 mg) for 30 min at room temperature and then separated using a magnet and sequentially washed with 2 ml of TA buffer supplemented with 50 mM NaCl. Nonspecific DNA-binding proteins were removed in washes using the same buffer containing 300 g of poly(dI-dC)⅐(dI-dC). The activity was eluted stepwise in TA buffer (10-ml aliquots) containing increasing NaCl concentrations (0.1, 0.25, 0.5, 0.8, and 1 M).

N-terminal Amino Acid Sequence Determination
Proteins were precipitated by the addition of 2 volumes of acetone, resuspended in Laemmli sample buffer, and separated on SDS-poly-acrylamide gels (23) or two-dimensional protein gels (24). Proteins were transferred from the gel to a nitrocellulose membrane (Immobilon P, Millipore Corp.) by electroblotting in CAPS buffer (10 mM CAPS (pH 11) in 10% methanol). Proteins were visualized on the membrane by staining with a solution of 5% Ponceau S red in 10% acetic acid. The protein band was excised for N-terminal sequence analysis by Edman degradation.

Cloning the spbR Gene
Degenerate oligonucleotides were designed based on N-terminal (MARQERAV) and internal (LTVFQGAL) sequences of the purified SpbR protein to PCR-amplify the 5Ј-region of the gene (forward primer, RTGGCSCGICAGGARCG; and reverse primer, STYSCGSGGGAC-GAGGTGSCASTC). A 350-bp PCR fragment generated using S. pristinaespiralis genomic DNA as a template was cloned and sequenced. The predicted protein sequence had strong homology to the helix-turn-helix motif of bacterial transcriptional regulators belonging to the GABR family. This fragment was used to probe Southern blots of S. pristinaespiralis genomic DNA. A 4.1-kb hybridizing BclI band was subcloned into the BamHI site of pUC18 (pHG1). The spbR gene, along with flanking regions of 1.4 and 1.9 kb, was subcloned on an EcoRI/MscI fragment into the EcoRI/EcoRV sites of pUC21 (pNL4). The plasmid was cleaved at its unique MluI site in the spbR ORF, blunt-ended by T7 DNA polymerase, and ligated to a SmaI fragment containing an apramycin resistance gene cassette (aaC(3)IV) (24). Cleavage of this plasmid at sites in the adjacent polylinker (EcoRI/XbaI) released a fragment containing the disrupted spbR gene and allowed it to be subcloned into pSET151 (25), a non-replicative plasmid that has the thiostrepton resistance marker. The construct (pNL6) was alkali-denatured and used to transform S. pristinaespiralis NRRL2958 protoplasts (21). Among 90 apramycin-resistant transformant colonies, only one was thiostreptonsensitive (spbR25).
To show by complementation in trans that the phenotypes observed were due to the disrupted spbR gene, a plasmid containing only spbR and its promoter was constructed. The spbR gene was removed from pHG1 by cleavage at BamHI (within insert)/EcoRI (vector) sites and ligated with pUC21 cleaved by the same enzymes. The fragment containing spbR was excised by BamHI and BglII (pUC21-encoded) and cloned into pIJ904 at the BamHI site (pNL5).

Disruption of the spbR Gene
A fragment containing the disrupted spbR gene and its flanking regions was cloned into a non-replicative plasmid (pSET151; construction described in the Fig. 2 legend) with the selectable thiostrepton resistance marker. This plasmid (pNL4) was used to transform S. pristinaespiralis. Among 100 apramycin-resistant transformants, only one was thiostrepton-sensitive. Southern hybridization (see Fig. 2B) showed that this clone (spbR25) contained the expected disruption of spbR resulting from a double crossover event. A low copy number plasmid containing the intact spbR gene (pNL5) (see Fig. 2A) was able to restore this activity in trans.

Recombinant SpbR Protein Produced in E. coli
The coding region of the spbR gene was amplified by PCR using oligonucleotides NeuNT (5Ј-ACAACACATATGGCGCGGCAGGAGCGGG-3Ј) and NeuCT (5Ј-GGTAAGCTTTGGTGGGGTGGGTCAGT-3Ј). The amplified fragment was inserted into the NdeI/HindIII sites of the expression plasmid pDS56 (pHG2). pHG2 was used to transform E. coli M15 carrying a plasmid that supplies Lac repression (pRep4). This transformant was grown in LB medium supplemented with 100 mg/liter ampicillin and 25 mg/liter kanamycin. Synthesis of the protein was induced by the addition of 2 mM isopropyl-␤-D-thiogalactopyranoside. Purification of the recombinant FIG. 2. Construction of an spbR mutant. A, restriction map and mutagenesis of spbR. SpbR was mutagenized by the insertion of the aaC(3)IV gene, conferring apramycin resistance (see "Experimental Procedures"). The BclI/BamHI fragment subcloned into pNL5 is indicated by the arrow. B, confirmation of the spbR25 locus by Southern hybridization. FspI/ MscI-digested genomic DNA from NRRL2958 or spbR25 was probed using the FspI/MscI fragment (A). wt, wild-type strain; ⌬, spbR25.
protein was done according to the procedure used for the isolation of SpbR from Streptomyces cell extracts, omitting the DNA affinity chromatography step.

Isolation and Expression of the scbR Gene
An S. coelicolor homolog of spbR was isolated from genomic DNA. A PCR fragment was amplified using degenerate primers based on conserved amino acid motifs within the DNA-binding domains of GABR genes barA, arpA, and farA: ALYFHFA (SGCGAAGTGGAAGTAS-RRSGC) and AAAEVFDE (GCSGCSGCSGARGTSTTCGACGA). This 150-bp fragment was used as a hybridization probe to clone a 5-kb region of the locus. The sequence of the scbR ORF was later found to be identical to that recorded by E. Takano (accession number AJ007731) and by the Sanger Center S. coelicolor Genome Sequence Project (accession number AL132824 http://www.sanger.ac.uk/Projects/S_coelicolor/). The scbR gene was subcloned as a 5-kb BclI fragment into the BamHI site of pUC18 (pMF10).

Superoxide Dismutase Assays
Crude cell extracts separated on SDS-polyacrylamide gels were transferred onto a nitrocellulose membrane and probed with rabbit anti-M. tuberculosis superoxide dismutase antibodies provided by M. A. Horwitz (UCLA) and then with swine peroxidase-conjugated anti-rabbit antibodies. Crude cell extracts were also separated electrophoretically on nondenaturing acrylamide gels, and superoxide dismutase was stained in situ (26).

Southern Blot Hybridization
Digested genomic DNAs were separated on 1% agarose gels, transferred by vacuum blotting to nitrocellulose membranes (Hybond N ϩ Amersham Pharmacia Biotech), and probed by Southern hybridization under standard conditions (22) or under low stringency (0.2ϫ SSC and 0.5% SDS at room temperature).

Polymerase Chain Reaction
PCRs were carried out using 100 pmol of primer, buffer supplied by PerkinElmer Life Sciences, and a Protocol Thermocycler (AMS Biotechnology). Standard hot-start PCR and touchdown PCR were performed in the presence of 10% Me 2 SO.

DNase I Footprinting
The PpapR1 fragment was removed from pMF1 by HindIII/SmaI digestion and end-labeled by filling in the HindIII overhang with [␣-32 P]dATP. The probe was incubated with a saturating amount of purified SpbR (determined by gel retardation assay) in the presence of 4 g of competitor DNA (poly(dI-dC)⅐(dI-dC)) and 0.5 units of DNase I. The digestion products were resolved on a 6% sequencing gel.

Analysis of Secondary Metabolites
Secondary metabolites in the media were extracted in ethyl acetate, resolubilized in Me 2 SO, and assayed directly for antibiotic activity (pristinamycin) or separated by HPLC. The disc antibiotic assay utilized B. subtilis ATCC6633 growing on NE agar as an indicator lawn. To identify individual components of the pristinamycin complex, ethyl acetate extracts were applied to a reverse-phase column (rpc c 2 /c 18 Sc2,1/1.0 SMART, Amersham Pharmacia Biotech) in 0.1% trifluoroacetic acid and eluted in a linear gradient of H 2 O and acetonitrile. Fractions were assayed for compounds that inhibited SpbR DNA-binding activity.

Protein Quantification
Protein content was measured using a kit supplied by Bio-Rad with bovine serum albumin as a standard.

An SARP Gene Is Present in the Pristinamycin Biosynthetic
Cluster-papR1, encoding a protein (284 amino acids) homologous to the SARP family, was identified in the course of sequencing the pristinamycin biosynthetic gene cluster (nucleotides 431-1283; accession number A37840). BLASTP protein data base searches showed that PapR1 has the highest similarity (73% identity) to TylS and significant matches with a family of SARPs required for expression of various Streptomyces antibiotic biosynthetic gene clusters, including DrrR1 (56% identity; daunomycin), MtmR (45% identity; mithramycin), and ActII-ORF4 (37% identity; actinorhodin). Other SARPs tested have proven to be essential for antibiotic biosynthesis; gene disruption of papR1 reduced both PI and PII yields by only 30%. 2 This may reflect the activity of a second SARP gene recently identified in the pristinamycin gene cluster. 2 We assumed that these redundant (papR2) genes were involved in the control of pristinamycin biosynthesis, and studies of papR1 might identify higher level regulators coordinating synthesis of PI and PII with developmental signals. Interestingly, visual inspection of the sequence upstream of the papR1 transcriptional start site identified a potential ARE (see below).
Pristinamycin Biosynthesis during Stationary Phase Is Associated with Alterations in the DNA-binding Activity of a Putative Regulator of the papR1 Promoter-Samples of a growing culture (Fig. 3A) were assayed for pristinamycin biosynthesis as well as for proteins that potentially regulate the papR1 promoter (PpapR1) (Fig. 3B). Bioassays showed that pristinamycin antibiotic activity accumulated in the medium beginning shortly after the maximum mycelial mass was attained (40 -100 h) (Fig. 3A). Gel retardation assays using a fragment encoding the papR1 promoter region (PpapR1) detected a potential GABR transcriptional regulatory protein (SpbR) in crude cytoplasmic extracts prepared from these cultures. PpapR1 gel shift activity, not detected in growing cultures ( A, S. pristinaespiralis NRRL2958 was grown in a fermenter as described under "Experimental Procedures." Arrows indicate samples whose representative HPLC profiles are shown in Fig. 6. Samples were taken at the numbered time points and assayed for pristinamycin and SpbR activity (see Fig. 6B). B, SpbR activity was measured by mobility shift assays using a SmaI/HindIII fragment (0.006 pmol) containing 322 bp of the papR1 promoter region. Lane P, the PpapR1 probe. In 3B), increased shortly after the cultures entered stationary phase, coincident with the activation of pristinamycin biosynthesis (Fig. 3A).
Purification of SpbR from S. pristinaespiralis-SpbR was enriched from crude extract (7 g of protein) of stationary phase mycelia (100 g) using sequential DEAE, heparin, and MonoQ ion-exchange columns (see "Experimental Procedures"). The final step of the purification employed biotinylated PpapR1 fragment fixed to streptavidin-coated beads (23). The fact that SpbR remained fixed to the matrix after exposure to either nonspecific competitor DNA or high salt concentrations suggested specific interactions and allowed a 100-fold purification. SDS-polyacrylamide gel electrophoresis analysis (Fig. 4A) of the activity (Fig. 4B) specifically eluting at high salt concentrations (Ͼ250 mM) (lanes 6 -9) suggested that it corresponded to a protein with an apparent mass of 28 kDa. This protein (3 g, 93 pmol) was blotted onto a nitrocellulose filter, and its N-terminal sequence (determined by Edman degradation) was MARQERAV. Four internal peptides generated by Staphylococcus aureus V8 endoproteinase had the following N-terminal sequences: LTVEQGAL, VADLY, DFSP, and VLAYEEAVRR.
Cloning and Sequencing of the S. pristinaespiralis spbR Gene-Oligonucleotides based on N-terminal (MARQERAV) and internal (LTVFQGAL) SpbR amino acid sequences were used to amplify the 5Ј-region of spbR, thereby facilitating the cloning and sequencing of the corresponding locus (described under "Experimental Procedures"). The locus was cloned on a 4.1-kb fragment in pUC18 (pHG1). Extracts of this strain of E. coli retarded migration of the PpapR1 fragment, whereas strains containing the vector alone were inactive.
DNA sequence analysis predicted SpbR to be a 228-amino acid protein (25.9 kDa) with similarity (40 -60% identity) to bacterial The arrowhead indicates the 28-kDa band that copurified with the gel retardation activity (estimated to be ϳ0.3 g by comparison with similarly Coomassie Blue-stained serum albumin). Lanes 1 and 10 are molecular mass markers (in kilodaltons). B, PpapR1-binding activity (SpbR) was monitored during purification steps. Gel retardation experiments were done using the PpapR1 probe, isolated as a HindIII/SmaI fragment. Lane 1, free PpapR1 probe (0.006 pmol); lanes 2-9, probe incubated with aliquots (0.002% of the total fraction volume) of the samples described for A.
transcriptional regulators belonging to the GABR family, including FarA, BarA, ArpA, JadR1, and TylP. SpbR is most similar to TylP in the Streptomyces fradiae tylosin gene cluster.
Expression of spbR in E. coli and purification of recombinant SpbR (see "Experimental Procedures") provided definitive proof that SpbR is the PpapR1-binding protein. Recombinant SpbR had chromatographic characteristics (on DEAE, heparin, and Superdex 200 columns) similar to those of the purified native protein. Purified recombinant SpbR migrated as a single molecular species corresponding to ϳ50 kDa on a size-exclusion column (Superdex 200) (Fig. 5A). These data established the purity (Ͼ95%) of the recombinant protein and suggested that both native and recombinant SpbR formed dimers in solution. Finally, recombinant SpbR retarded migration of the PpapR1 fragment in a manner indistinguishable from that detected in S. pristinaespiralis (Fig. 3B, lane 8). The binding curve determined an approximate K D of 3 ϫ 10 Ϫ8 M (Fig. 5B) for the monomer or 1.5 ϫ 10 Ϫ8 M based on its apparent dimeric form (Fig. 5A).
A Butyrolactone-like Compound Inhibits SpbR Binding-An inhibitor of SpbR (purified recombinant protein) gel shift ac-tivity was detected in ethyl acetate extracts (note that ethyl acetate extracts amphipathic compounds such as PI, PII, and butyrolactones) of the culture medium either before (32 h) (Fig.  6A) or after (64 h) (Fig. 6B) antibiotic activity appeared in the medium. Although very little A 215 -adsorbing material (Fig. 6A) was detected in the 32-h culture, fraction 21 inhibited formation of the SpbR-PpapR1 complex. This activity was also present in late stationary phase cultures (Fig. 6B) and likewise eluted from the HPLC column in fraction 21. In both cases, the inhibition was concentration-dependent, suggesting specific inhibition as has been reported for other butyrolactone-binding proteins: FarA (27), BarA (27), and ArpA (28). Although molecules belonging to the pristinamycin complex (a mixture of compounds representing biosynthetic intermediates or derivatives of PI and PII) were also present in stationary phase cultures, formation of the SpbR-PpapR1 complex (ϳ0.1 M) was not inhibited by molar excesses (ϳ0.5 M) of PI or PII. The S. griseus ␥-butyrolactone (A-Factor) also eluted in this part of the HPLC gradient (fraction 22).
Gel retardation assays showed that several structurally related butenolides inhibited binding of SpbR (0.15 M) to the  Fig. 1) and separated by HPLC (absorption was recorded at 210 nm). HPLC fractions were resuspended in Me 2 SO and added to SpbR-PpapR1 complexes (0.06 pmol of a PCR-generated radiolabeled fragment, 0.17 pmol of recombinant SpbR protein, 5 g of bovine serum albumin, and 2 g of competitor DNA). A compound able to dissociate the SpbR-PpapR1 complexes was detected in fraction 21 of both cultures. In B, all fractions were also assayed after dilution to 1:2 and 1:10. Although PII comigrated with the activity in fraction 21, PII was inactive. Five g of PII was not able to dissociate the complex (the A 210 of material in fraction 21 was much less than that of 5 g of PII).
These results indicated that SpbR interacted most specifically with an S. pristinaespiralis ligand that was similar to Streptomyces butyrolactone quorum-sensing autoregulators.
DNA-binding Motifs for Autoregulatory Proteins-Purified recombinant SpbR protein protected a 31-bp sequence of the PpapR1 fragment against DNase I digestion (Fig. 7A). This FIG. 7. SpbR recognizes a sequence motif found upstream of SARP and GABR genes. A, DNase I footprinting of an SpbR-binding site in the papR1 promoter region. A DNA fragment (nucleotides 113-435) encoding the nucleotide sequence upstream of papR1 was partially digested with DNase I in the presence or absence of sufficient purified recombinant SpbR to fully saturate its binding site (SpbR was titrated using the same DNA fragment and assayed by gel retardation). These digests were separated on a 6% acrylamide gel and compared with a sequencing ladder. The sequencing ladder was generated using a primer corresponding to the HindIII end of the probe fragment (AGCTTCGAACACGGCTCCTACCAA). B, data base searches identify operator motifs belonging to the ARE family. A matrix of conserved nucleotide residues from seven ARE sequences that have been verified by DNase I footprinting or gel retardation (listed in the first sequence series) was compiled to search the Streptomyces sequences (Matrix 1) in the GenBank TM /EBI Data Base using Mat-ind and Mat-Inspector (kindly provided by K. Quandt). This analysis identified six additional motifs that were all located upstream of genes belonging to the GABR or SARP family (second sequence series). These 13 sequences were used to compile Matrix 2, which identified the six genes of the third series and the relative abundance of nucleotides at each position (the consensus "IUPAC string"). The relative similarity index of each sequence to the consensus was defined in a final search all 19 sequences (Matrix sequence had strong homology to all experimentally verified AREs, including those preceding arpA, barA (BARE3), barB (BARE1 and BARE2), and farA.
Two ARE motifs were visually identified upstream of the spbR translational start codon at bp Ϫ19 to Ϫ42 (PspbR-ARE1) and bp Ϫ289 to Ϫ317 (PspbR-ARE2). To confirm that these sequences were indeed SpbR-binding sites, fragments containing each putative ARE were tested separately by gel retardation assay. Titration of purified recombinant SpbR demonstrated that its affinity for these motifs was similar to that of the PpapR1 fragment (data comparable to those shown in Fig.  5B; not shown).
A nucleotide sequence matrix compiled from seven experimentally verified AREs (30) was used to search the Streptomy-ces nucleotide sequences in the complete GenBank TM /EBI Data Bank and the Sanger Center S. coelicolor Database (Ͼ90% complete). Nineteen ARE-like sequences were detected and used to determine a set of matrix similarity indices that represented the relative adherence of each sequence to the most probable motif and the relative frequency of nucleotides at each position (IUPAC string) (Fig. 7B). These putative ARE-regulated genes include GABRs (scbR and tylP), SARPs (ccaR and tylS), proteins involved in butyrolactone biosynthesis (afsA and farX), and other genes within antibiotic biosynthetic clusters (jadR1, jadR2, vmsR, and tylQ).
Of the three putative ARE-binding sites identified in the S. coelicolor genome, two were experimentally verified by gel retardation assay (data not shown) using both SpbR (purified) and crude extracts of E. coli producing recombinant ScbR (M15/ pMF10) (see "Experimental Procedures") (data not shown). These were located immediately upstream of scbR or between the 5Ј-sequences of adhC, a putative alcohol dehydrogenase, and an unidentified ORF. A putative site upstream of a histidine kinase paralog was located only 3 kb downstream of scbR.
A Constructed spbR Mutant Has Pleiotropic Defects in Pristinamycin Biosynthesis, Growth, and Aerial Mycelium Formation-To study spbR function, the gene was inactivated by insertion of an apramycin resistance cassette (aaC(3)IV) into its unique MluI site ( Fig. 2A). Diverse spbR-determined phenotypes were observed by comparing wild-type cultures with this mutant (spbR25) in various liquid and solid media. All phenotypes described below were suppressed by a plasmid containing spbR (pNL5) ( Fig. 2A) and were therefore attributed to inactivation of spbR rather than polar effects on transcription of downstream genes or mutations in other loci. SpbR DNA-binding activity, assayed by gel retardation of the papR1 fragment, was not detected in extracts from this mutant.
spbR25 grown in HT7T liquid medium did not produce any antibiotic activity detected with disc assays or the major secondary metabolite HPLC peaks characteristic of the wild-type strain. These included PI and PII as well as all of the other unidentified compounds (Fig. 6B), most of which belong to the pristinamycin complex. In addition, an unidentified dark pigment produced by the wild-type strain was not produced by spbR25.
Although spbR25 grew like its parent in HT7T liquid medium, colony growth and morphological development were very slow on corresponding agar-based solid media. Closer microscopic examination (Fig. 8) showed that germination and the earliest phase of colony development (the first 24 h) were FIG. 8. Developmental defects of the spbR25 mutant. S. pristinaespiralis NRRL2958 (wild-type (WT)) and spbR25 (spbR mutant) were streaked out on HT7T agar plates and photographed at the same magnification after various periods of growth (20,26,40, and 60 h) at 30°C under a light microscope and after 60 h under a dissecting microscope. similar in the wild-type and spbR25 strains. However, as the colony became barely visible, the mutant did not maintain wild-type rates of growth. Although the wild-type strain matured into much larger sporulating colonies, the growth and morphological development of spbR25 colonies were severely retarded. The mutant strain had similar defects on all solid media tested (mannitol-soya, minimal production medium, NE, HT7T, and R2YE).
SpbR Is Needed for Expression of an Extracellular Superoxide Dismutase-SDS-polyacrylamide gel electrophoresis analyses ( Fig. 9) showed the progressive accumulation of a major 23-kDa protein in NRRL2958 mycelia growing in liquid cultures. This band was not detected in spbR25 unless the wildtype spbR gene was supplied in trans (pNL5) (data not shown). The N-terminal sequence of the 23-kDa protein eluted from either SDS-polyacrylamide or two-dimensional gels was the same: GTYALPDLPYDYSALAPAITPEILE. The sequence was identical to that of S. coelicolor A3(2) superoxide dismutase SodF at 19 (underlined) of 25 positions, suggesting that it is the N-terminal sequence of S. pristinaespiralis superoxide dismutase SodF.
This protein was independently identified as SodF by Western blotting using an antibody raised against M. tuberculosis SodF (Fig. 10A). The antibody detected a 23-kDa protein in cell extracts of NRRL2958 that was not present in spbR25. As previously reported for S. coelicolor (16), expression of sodF in S. pristinaespiralis was suppressed by the addition of Ni 2ϩ to the medium, but was not affected by chelation of divalent cations (Fig. 10A).
In situ detection of superoxide dismutase enzymatic activities resolved on native acrylamide gels confirmed these results. In the absence of supplemented NiCl 2 , a single weak superoxide dismutase activity band, present in NRRL2958 cultures, was missing in the mutant. NiCl 2 suppressed accumulation of this protein, presumed to be SodF, and induced a slower migrating superoxide dismutase isoenzyme in both strains (Fig.  10B), presumed to be SodN.
Finally, SpbR shared an unusual feature with M. tuberculo-sis SodF. Western blotting using anti-M. tuberculosis SodF antibody detected an spbR-dependent 23-kDa protein in the medium (data not shown). SDS-polyacrylamide gel electrophoresis analysis of the total protein composition of the medium revealed that it was the only major band detectable by Coomassie Blue staining (Fig. 9, lane 8). Its N-terminal sequence (the first 5 residues were determined by Edman degradation) was identical to cytoplasmic S. pristinaespiralis SodF. Thus, both M. tuberculosis and S. pristinaespiralis accumulated SodF in the medium without the apparent N-terminal processing that characterizes type II protein secretion systems.

DISCUSSION
Our studies of S. pristinaespiralis SpbR extended several unifying concepts and established the principal functions of Streptomyces quorum-sensing signals and their receptors. Data base searches for corresponding operator sites revealed that, in addition to autoregulating their own expression and that of other genes involved in butyrolactone synthesis, GABR proteins may control SARPs, the primary class of antibiotic regulatory proteins in Streptomyces. However, other targets may play a different role related to pleiotropic growth defects outlined below.
Reinforcing previous reports, we concluded that SpbR-related GABRs play alternative physiological roles involving species-specific regulatory systems. In S. griseus, disruption of the A-Factor receptor leads to early differentiation and increased streptomycin biosynthesis. Inactivation of the barA gene in S. virginiae leads to precocious virginiamycin biosynthesis, but does not affect morphological differentiation (11). Although these data suggest inhibitory functions for butyrolactone receptors in other Streptomyces species, our results revealed positive roles for spbR in S. pristinaespiralis in maintaining growth, regulating antibiotic biosynthesis, and allowing a normal response to oxidative stress.
Pleiotropic Effects of the spbR Mutation on Growth and Antibiotic Biosynthesis-spbR25 had a developmental growth defect on solid media. Just as the mycelial mass became barely visible on solid media, the mutant failed to maintain growth rates comparable to those of its non-defective parent. This may be interpreted as an inability to carry out a developmentally controlled metabolic transition allowing continued growth. Thus, the SpbR quorum sensor protein may facilitate recovery from a growth arrest that occurs during differentiation of several Streptomyces species (31)(32)(33).
Curiously, although growth of spbR25 cultures were not impaired in liquid cultures with the same composition as the HT7T solid medium, pristinamycin biosynthesis was still blocked. This probably resulted from distinct developmental programs reflected most obviously in morphological differences between mycelia grown on liquid versus solid media. For example, free circulation of the medium and fragmentation of mycelia in liquid cultures may allow better nutrient exchange or prevent localized accumulation of negatively acting waste products. These results suggest that the inability of spbR25 to produce pristinamycin is not a simple function of its conditional vegetative growth defect, an interpretation further supported by the fact that SpbR bound upstream of an SARP gene located within the pristinamycin biosynthetic cluster.
Regulatory Targets of spbR-This first demonstration that an autoregulator receptor protein interacted with an SARP gene promoter region led to the conclusion that AREs are primarily found upstream of genes involved in butyrolactone or antibiotic biosynthesis. DNase I footprinting showed that SpbR protected a sequence similar to those protected by BarA, and FarA (AREs) and provided the first experimental evidence that certain GABR proteins recognize heterospecific ARE motifs. FIG. 10. The spbR25 mutant is defective for regulation of S. pristinaespiralis SodF synthesis. A, NRRL2958 (wild-type (wt)) or spbR25 (⌬) was inoculated into HT7T medium and harvested after 20 or 40 h. Experimental cultures were exposed to 200 M NiCl 2 or 200 M EDTA from 10 to 20 h or from 30 to 40 h. Western blotting detected a protein band that cross-reacted with M. tuberculosis extracellular SodF. B, the mycelial extracts from the cultures described for A before and after exposure to NiCl 2 were subjected to native gel electrophoresis. Superoxide dismutase activity was detected in situ (26). First lane, wild-type strain (40 h); second lane, spbR25 (40 h); third lane, wild-type strain exposed to 200 M NiCl 2 ; fourth lane, spbR25 exposed to 200 M NiCl 2 . The left arrow indicates an SodF isoenzyme, and the right arrow indicates the Ni 2ϩ -induced superoxide dismutase.
Only three AREs were clearly detected by the sequence matrix search of the S. coelicolor Database, which included at least 90% of the 8.7-megabase pair(s) chromosome. In addition to scbR, two uncharacterized genes of unknown function (one between adhC and an unidentified ORF (SCD95A) and the other upstream of a putative histidine kinase) were detected. This suggested that autoregulator receptor proteins do not interact directly with known SARPs controlling undecylprodigiosin (redD) or actinorhodin (actII-orf4) antibiotic biosynthesis. However, the data cannot absolutely rule out the possibility that the matrix specifically identified only a subset of the ARE sequences under the control of GABR proteins in vivo.
Although it is not known whether spbR is genetically linked to papR1, these genes have several interesting similarities to the S. fradiae tylosin biosynthetic cluster (34). In both systems, AREs were identified upstream of SARP genes (papR1 and tylS) and autoregulatory proteins (spbR and tylP). Furthermore, there was a syntenous arrangement of cytochrome P-450, the autoregulatory receptor (SpbR or TylP), and acyl-CoA oxidase genes. These observations indicate that the two antibiotic regulatory systems utilize a conserved mechanism to coordinate host metabolism with antibiotic biosynthesis.
SpbR Is Needed for SodF Expression-N-terminal sequence, in situ activity staining, and immunoblot analyses showed that spbR25 lacks a major cytoplasmic protein identified as SodF. The S. pristinaespiralis 23-kDa superoxide dismutase was maintained at rather constant levels during growth and was repressed by NiCl 2 , as previously reported for sodF in S. coelicolor and Streptomyces lividans (16). This, along with its size, cross-reactivity with antibodies raised against M. tuberculosis SodF, and N-terminal sequence, showed that it corresponds to the S. pristinaespiralis sodF gene.
SodF was also the major extracellular protein that accumulated in the medium of S. pristinaespiralis NRRL2958 cultures. The fact that both internal and external proteins had the same size and N-terminal sequences indicated that superoxide dismutase might well be autotransported as a leaderless protein (35). The extracellular SodF of M. tuberculosis is thought to defend the pathogen from macrophage oxidative attack. Our observation of the same phenotype in a related saprophytic bacteria suggests a more generic metabolic function.
Extracellular superoxide dismutase may also protect externally exposed macromolecules from oxidants present at the cell-environment interface. Flavodoxin enzymes in respiratory chains are believed to be the primary metabolic sources of superoxides in bacteria (36). It is conceivable that respiratory enzymes or associated diffusible flavanoid electron carriers might generate ROS able to oxidize externally exposed macromolecules. Interestingly, in Caulobacter crescentus, superoxide dismutase may serve to protect essential cytoplasmic membrane proteins exposed to the environment (37). In support of this model, catalase, another oxidative repair enzyme, is exported in Streptomyces (38). CatB, a developmentally controlled catalase studied in S. coelicolor, is required for aerial mycelium formation (38), which may depend on activation of oxidative metabolism (33,39). Thus, an increased requirement for adaptation to oxidative stress, generated by metabolic shifts, may be an integral part of Streptomyces colonial development.