H2O2-sensitive fur-like repressor CatR regulating the major catalase gene in Streptomyces coelicolor.

Streptomyces coelicolor produces three distinct catalases to cope with oxidative and osmotic stresses and allow proper growth and differentiation. The major vegetative catalase A (CatA) is induced by H(2)O(2) and is required for efficient aerobic growth. In order to investigate the H(2)O(2)-dependent regulatory mechanism, an H(2)O(2)-resistant mutant (HR40) overproducing CatA was isolated from S. coelicolor A3(2). Based on the genetic map location of the mutated locus in HR40, the wild type catR gene was isolated from the ordered cosmid library of S. coelicolor by screening for its ability to suppress the HR40 phenotype. catR encodes a protein of 138 amino acids (15319 Da), with sequence homology to ferric uptake regulator (Fur)-like proteins. Disruption of catR caused CatA overproduction as observed in the HR40 mutant, confirming the role of CatR as a negative regulator of catA expression. The levels of catA and catR transcripts were higher in HR40 than in the wild type, implying that CatR represses transcription of these genes. Transcripts from the catA and catR genes were induced within 10 min of H(2)O(2) treatment, suggesting that the repressor activity of CatR may be directly modulated by H(2)O(2). A putative CatR-binding site containing an inverted repeat of 23 base pairs was localized upstream of the catA and catR gene, on the basis of sequence comparison and deletion analysis. CatR protein purified in the presence of dithiothreitol bound to this region, whereas oxidized CatR, treated with H(2)O(2) or diamide, did not. The redox shift of CatR involved thiol-disulfide exchange as judged by modification of free thiols with 4-acetamido-4'-maleimidylstilbene-2,2'-disulfonate. From these results we propose that CatR regulates its downstream target genes as a repressor whose DNA binding ability is directly modulated by redox changes in the cell.

All aerobically growing organisms come into contact with reactive oxygen species, generated as a by product of normal respiratory processes or from encounter with exogenous oxi-dants. To counter the damaging effect of reactive oxygen species, cells have evolved anti-oxidant defense systems, whose expression is usually induced by reactive oxygen species and/or oxidants. Cells possess regulators that sense and transduce the oxidative signals to allow coordinated gene expression.
Streptomyces coelicolor, a Gram-positive aerobic soil bacterium that undergoes complex cycle of morphological and physiological differentiation, responds to H 2 O 2 by inducing about 100 proteins as judged by two-dimensional gel analysis (9). It produces two monofunctional catalases (CatA and CatB) and a catalase peroxidase (CatC), among which only CatA is induced by H 2 O 2 . CatA is the major catalase required for efficient growth of mycelium and resistance against H 2 O 2 (10). CatB is induced by osmotic stress or at the stationary phase, and is required for proper differentiation and osmoprotection of the cell (11). CatC is expressed transiently at late exponential to early stationary phase, and its function in vivo has not been * This work was supported by International Collaborative Research Grants from Korea Science and Engineering Foundation (KOSEF) and the Biotechnology and Biological Sciences Research Council (BBSRC). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  revealed yet (12). In addition to these, it has been reported that S. coelicolor possesses two kinds of superoxide dismutases (Ni-SOD and FeZn-SOD encoded by sodN and sodF, respectively), an alkyl hydroperoxide reductase system (encoded by ahpCD), and a thioredoxin system (encoded by trxBA) as oxidative defense proteins (13)(14)(15)(16). Three transcriptional regulation systems controlling the expression of a subset of these oxidative defense proteins have been found: an anti-factor RsrA which binds a cognate factor, R , in a redox-dependent manner and thus regulates R -dependent transcription of the trxBA and sigR genes (16,17); OxyR which activates the ahpCD gene but not the catalase genes (15); and FurA which regulates catalase peroxidase gene (catC; Ref. 12). A soxR-like gene has been predicted from the sequence information (S. coelicolor Genome Project, The Sanger Center).
The redox-dependent activity modulation is best described for RsrA. It binds and sequesters R from transcribing trxBA and its own gene under reducing conditions. Upon oxidation by H 2 O 2 or a thiol-oxidizer diamide, it loses its R binding activity, releasing R to transcribe its target genes. The redox-dependent modulation of RsrA activity is mediated via thiol-disulfide exchange. The induced thioredoxin, in turn, reduces RsrA, forming a negative feedback loop. This allows the homeostatic control of intracellular redox conditions by RsrA. Not much is yet known about the activity modulation of OxyR in S. coelicolor nor the mechanism of H 2 O 2 -dependent induction of the major catalase. In this study, we report on the isolation and characterization of another novel redox-sensitive regulator CatR, which serves as a transcriptional repressor regulating the catA gene.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-S. coelicolor A3(2) J1501 was mutagenized with UV to isolate H 2 O 2 -resistant mutants (18). J1915, a glkA derivative of M145 (19), was used to disrupt the catR gene. Streptomyces cells were grown in YEME medium for liquid culture, and on R2YE, nutrient agar (NA), or minimal medium plates for surface culture (20). E. coli ET12567, a non-methylating strain, was used to prepare DNA to transform S. coelicolor (21).
Cloning catR from the Ordered Cosmid Library-Based on genetic mapping of the mutated locus in the H 2 O 2 -resistant mutant HR40, ordered cosmids from K13 to 2E1 (kindly provided by H. Kieser), encompassing the mthB2 locus (22), were tested to isolate catR. Cosmid DNAs were prepared from E. coli ET12567, and introduced into HR40 protoplasts following denaturation (23). Cosmid-integrated clones were selected on R2YE medium containing 200 g/ml kanamycin, and tested for their ability to repress CatA overproduction in HR40. The amount of catalase production was initially assessed by dropping 30% H 2 O 2 on colonies and observing the intensity of bubbling caused by O 2 evolution, and then confirmed by Coomassie staining of proteins in SDS-PAGE.
Two overlapping cosmids, 6F2 and 7E4, which repressed catalase overproduction in HR40 were further analyzed. Restriction fragments in the overlapping region of the two cosmids were cloned into pSET152 (24), a conjugation vector containing the ⌽C31 phage attachment site. Each plasmid was used to transform E. coli ET12567, which harbors a mobilizing plasmid pUZ8002 (a kind gift from D. H. Figurski), and then transferred into HR40 by conjugation based on the method of Mazodier et al. (25). Exconjugants were selected with 50 g/ml apramycin, and tested for catalase production. S1 Mapping Analysis-RNA was isolated from S. coelicolor cells grown in YEME as described previously (20). Cells were resuspended in Kirby mixture and disrupted by sonication with a micro tip (Sonics and Materials Inc.) at 25% of the maximum amplitude (600 W, 20 kHz) for 10 s. For mapping 5Ј ends of catR transcripts, a 270-base pair HindIII/ BglII fragment of the PCR-amplified catR gene uniquely labeled with [␥Ϫ 32 P]ATP at the BglII site (148 nucleotides from the start codon) was prepared. S1 nuclease protection assay was done as described previously (26). For mapping catA transcript, a 0.6-kb SalI/BglII fragment of the catA gene uniquely labeled at the BglII site (244 nucleotides from the start codon) was used as a probe.
Disruption of catR-catR was disrupted by inserting a hygromycinresistance cassette (hyg, provided by J. Ainsa) in the middle of the gene. Plasmid pDH5 containing hyg was digested with SphI, and cloned into the same site of pIJ2925 (provided by M. Bibb). The cassette was excised as an 1.9-kb BglII fragment, and inserted into the BglII site of pJH3013 which contains a 1.2-kb HincII/BamHI fragment of catR ( Fig.  2) in pUC18. The resulting plasmid pJH3013H was digested with EcoRI and HindIII, and the 3.1-kb insert was cloned into EcoRI/HindIII-cut pIJ6650 (provided by M. Paget), a pKC1132 derivative containing the glkA gene marker, resulting in pJH3083H. The plasmid pJH3083H was introduced into E. coli ET12567 containing pUZ8002, and then transferred by conjugation into J1915, a glkA Ϫ derivative of M145. The exconjugants were selected on SFM media containing 150 g/ml hygromycin and 50 g/ml apramycin. The selected exconjugants were then grown under hygromycin selection, and the spores were plated on MM containing 200 mM deoxyglucose plus 150 g/ml hygromycin. The surviving colonies, expected to have lost the vector including the counterselectable glkA by second crossover, were confirmed for apramycin sensitivity by replica plating. The catR disruption was then confirmed by Southern hybridization.
Gel Mobility Shift Assay-The catR promoter region from Ϫ111 to ϩ71 nucleotides from the start codon was generated by PCR with primers CR1 (5Ј-CTCTTGGCCAATGCCGCCCCGG-3Ј) and CRS1 (5Ј-TCGGCCACGACCCGCCGCTGCG-3Ј). The catA promoter region from Ϫ84 to ϩ254 nucleotides from the start codon was generated by PCR with primers CAD84 (5Ј-AGCAGGCTTCGGCATCAATT-3Ј) and CAC (5Ј-TCGGAGAAGATCTTCGCGCTGG-3Ј). The PCR product was endlabeled with [␥-32 P]ATP using T4 polynucleotide kinase. Unincorporated isotopes were removed by centrifugation through a Sephadex G-50 spun column. The end-labeled probe (about 30,000 cpm for less than 0.1 pmol per each reaction) was incubated with 200 ng (13 pmol) of purified CatR in 20 l of binding buffer (10 mM Tris-HCl, pH 7.5, 1 mM MgCl 2 , 40 mM KCl, 100 g of poly(dI-dC) per ml, and 5% glycerol) at 30°C for 10 min. The DNA/protein mixture was electrophoresed on a 5% native PAGE in 20 mM Tris borate buffer, and analyzed by autoradiography.
Western Blot Analysis-Polyclonal antibody against purified CatR was raised in mice. The reacting signal was detected by goat anti-mouse immunoglobulin G conjugated with horseradish peroxidase using the Western ECL detection system (Amersham Pharmacia Biotech).

Characterization of an H 2 O 2 -resistant Mutant of S. coeli-
color-An H 2 O 2 -resistant mutant HR40 was generated by UV mutagenesis of spores from S. coelicolor A3(2) J1501 as described previously (18). It grows as rapidly as the wild type and differentiates normally, although it produces less of the blue antibiotic, actinorhodin, than the wild type. The protein profile of HR40 was compared with that of its parent (J1501) on SDS-PAGE (Fig. 1A). A prominent difference in HR40 was the overproduction of catalase A. No other significant changes in protein profile were observed. HR40 overproduced about 50fold more CatA than the wild type as judged by Western blot analysis (Fig. 1B). Measurement of the catalase activity in cell extracts also produced similar results (data not shown). Expression of other antioxidant enzymes such as catalase peroxidase (CatC) and alkyl hydroperoxide reductase system (Ah-pCD) was lowered in HR40 (Fig. 1B). The level of catalase B (CatB), a stationary phase-specific catalase, did not change significantly (Fig. 1B). Although HR40 was more resistant to H 2 O 2 than the wild type, it was slightly sensitive to cumene hydroperoxide, consistent with the reduced expression of AhpCD. Sensitivity of HR40 against superoxide generating agents such as paraquat, plumbagin, and menadione was similar to that of J1501 (data not shown).
Cloning of catR-The mutation (catR1) in HR40 had been previously mapped close to the mthB2 locus by genetic crosses (18). This locus is near the center of the chromosome, being separated by more than half a genome distance from catA, which has been physically mapped to the AseI F fragment near one end of the linear chromosome (22). 2 Therefore, catR is likely to encode a regulator for catalase A production, the catR1 allele giving rise to either a repressor with lost function or an activator with gained function. It has been previously suggested that catA expression may be regulated by a repressor system on the grounds that the presence of catA promoter fragments on a multicopy plasmid increased the production of catalase A in Streptomyces lividans. Therefore, assuming that the catR1 is a recessive loss-of-function mutation, we attempted to clone the wild type catR by introducing into HR40 genomic DNAs from ordered cosmids spanning the mthB2 locus, screening for wild type levels of catalase production as follows: since HR40 overproduces catalase A, application of 30% H 2 O 2 solution on top of its colonies generates the explosive evolution of O 2 in contrast to mild bubbling on wild type colonies. Using this handy screening tool, we were able to select two overlapping cosmids (6F2 and 7E4) out of 23 cosmids (from K15 to 2E1) tested, as candidates harboring the catR gene. Cosmids 6F2 and 7E4 overlaps by about 21 kb. To localize catR more precisely, various restriction fragments of the overlapping region were subcloned into pSET152, a conjugation vector containing a phage integration site (att) (24). The recombinant plasmids were introduced into HR40 and tested for phenotype suppression. We found that a 1.2-kb HincII-BamHI fragment (in pJH3113) was able to repress catalase overproduction in HR40, whereas its subfragment (0.8-kb HincII-BglII fragment in pJH 3115) was not (Fig. 2, A and B, lanes 3 and 4). This enabled us to narrow down the position of the catR gene.
Nucleotide sequence analysis revealed that the insert DNA in pJH3113 contained two open reading frames, one coding for a homologue of inosine-phosphate dehydrogenase and the other coding for a Fur homologue ( Fig. 2A). Since plasmid pJH3115 failed to repress mutant phenotype, it appeared that the downstream open reading frame encoding a Fur homologue is the catR gene regulating catA expression.
To confirm the role of CatR as a repressor of catA gene expression, the catR gene in strain J1915 was disrupted as described under "Experimental Procedures." The ⌬catR mutant (JH11) showed no defect in growth or differentiation, except that it produced a reduced amount of actinorhodin on R2YE plates as observed for HR40. As expected, JH11 overproduced CatA to an extent comparable to HR40 (Fig. 2B, lane 6).
The other bacteria (Fig. 3). The closest match was found with FurV of Streptomyces venezuelae, whose gene is transcribed divergently from the bca gene encoding bromoperoxidase catalase (27). CatR also matched closely with PerR of B. subtilis.
To identify the mutated residue in HR40, the catR1 allele was amplified from HR40 genome by PCR and sequenced. A single nucleotide change from G to A at nucleotide 152 was detected, causing a codon change from Arg to Gln at residue 51. Since this residue is within a putative helix-turn-helix motif of Fur proteins (Fig. 3), we postulate that the mutated CatR protein in HR40 may be deficient in binding to the catA promoter.
Negative Regulation of catA and catR Transcription by CatR-We analyzed the transcripts from the catA and catR genes in wild type and HR40 mutant by S1 mapping (Fig. 4). HR40 produced catA transcripts drastically more than the wild type, suggesting that catA transcription is under the negative control of CatR. HR40 also produced more catR transcripts than the wild type, especially the longer one (catRp2), whose 5Ј end lies at about 70 nucleotides upstream from the translation start site. The 5Ј ends of the shorter transcripts (catRp1) were located at multiple sites immediately downstream of the ATG codon. A putative promoter (TCGGGA-N 18 -TAGGCT) resembling the sequences recognized by the major factor HrdB was identified upstream of the translation start codon, separated from ATG by 7 residues. Therefore, it is conceivable that the shorter transcripts were generated from degradation of transcripts initiated at the translation start site (catRp1). catRp2 start sites were located more precisely by high-resolution S1 mapping at 74 and 66 nucleotides upstream from the ATG codon (data not shown). A putative HrdB -type promoter (TTG-GCC-N 17 -TACAAT) was found upstream of the 74 site. The observation that catRp2 transcript was highly expressed in HR40, like the catA transcript, suggests that transcription from both catRp2 and the catA promoter is negatively regulated by CatR. The catRp1 promoter may be partially regulated by CatR. The production of CatR protein was also elevated in HR40 as judged by Western analysis (data not shown).
H 2 O 2 Induction of the catA and catR Transcripts-We then examined whether catR expression is induced by H 2 O 2 as is the catA expression. The changes in both catR and catA transcripts following H 2 O 2 treatment were monitored by S1 mapping (Fig.  5). Within 10 min of H 2 O 2 treatment, the catR transcripts increased to a maximum level and decayed to the basal level within 30 min. Both the catRp1 and catRp2 transcripts exhibited similar behavior even though catRp2 transcripts were induced more dramatically. The catA transcript increased sharply within 10 min, reaching a maximal level at around 20 min, and decayed within 60 min to the basal level. The similar kinetics of rapid induction by H 2 O 2 of both genes suggests that the transcriptional regulation is mediated by activity modulation of an already existing regulator, most likely CatR, by H 2 O 2 . The delay observed for catA transcript in reaching and decaying from the maximum level might reflect the higher stability of catA transcript than catR. The changes in the level of CatA and CatR proteins followed the changes in mRNAs as judged by Western analysis (Fig. 5). The CatA protein began to increase from 20 min, reaching a maximum value between 40 and 50 min following H 2 O 2 treatment. The CatR protein also began to increase from 20 min reaching a maximum value at about 40 min.
Identification of the Putative Binding Site of CatR-On the basis of the assumption that CatR binds directly to the promoter region of catA and catR genes we examined the nucleotide sequences of these promoter regions. The nucleotide sequences of the two promoter regions were compared with the similar sequence region of the furV-bca genes from S. venezuelae (Fig. 6). furV is divergently transcribed from the neighboring bca gene, whereas the transcription units of the catR and catA gene in S. coelicolor are distantly located. We identified an inverted repeat of 23 base pairs in all these genes; between Ϫ32 and Ϫ54 nucleotides from the start codon for catR and furV, between Ϫ70 and Ϫ92 for catA, and between Ϫ73 and Ϫ95 for bca genes. Counting from the catA transcription start site (ϩ1), FIG. 4. S1 mapping analysis of catR and catA transcripts in J1501 and HR40. RNAs were prepared from J1501 (lanes 1-3) and HR40 cells (lanes 4 -6) grown in YEME for 12, 22, and 40 h. The 5Ј ends of catR and catA mRNAs were mapped by S1 nuclease protection assay as described under "Experimental Procedures." The transcripts from catRp1, catRp2, and catA promoters produce protected bands as designated. the inverted repeat box lies between Ϫ43 and Ϫ65. The bca gene contains putative promoter elements almost identical with the catA promoter (catAp). A consensus CatR binding sequence (TCcNRARYNR-N3-YNRYTYNgGA) was deduced. In order to verify the role of this region in vivo, we generated recombinant reporter plasmids containing the catAp region up to Ϫ69 or Ϫ54 nucleotides from the transcription start site in front of the reporter gene in pXE4 plasmid (28). catAp-driven transcripts were determined by S1 mapping in S. lividans. The strain containing the shorter upstream sequence produced a greater amount of catAp-driven transcripts than the longer one, suggesting that the inverted repeat sequence does indeed function as a negative regulatory site (data not shown).
Redox-dependent Binding of CatR to the catR and catA Promoter Fragments-We then examined the binding of CatR protein to the catA and catR promoter fragments by gel mobility shift assay. CatR protein overproduced in E. coli was purified as described under "Experimental Procedures" in the presence of 1-2 mM DTT throughout the preparation. Purified CatR protein specifically bound to both catR and catA promoters in the presence of 1 mM DTT (Fig. 7, lane 2). When the CatR protein was oxidized by treatment with 10 mM H 2 O 2 or diamide, it did not bind to the promoters (Fig. 7, lanes 5 and 6). Addition of extra metals did not enhance the DNA binding activity. However, addition of 5 mM EDTA inhibited DNA binding of CatR, implying that metals are required for DNA binding (data not shown).
Involvement of Thiol Oxidation in Redox Modulation of CatR-Since CatR contains four cysteine residues, which are conserved in most of the Fur-like proteins, we tried to find out whether the redox modulation of CatR binding activity is mediated via thiol-disulfide exchange between the cysteine residues. To this end, purified CatR in the absence or presence of oxidants (10 mM H 2 O 2 or 10 mM diamide) was treated with 20 mM 4-acetamido-4Ј-maleimidylstilbene-2,2Ј-disulfonate (AMS) to alkylate sulfhydryl groups of cysteines. As presented in Fig.  8, in the absence of oxidants and AMS, unmodified monomeric CatR migrated as a 17-kDa band in SDS-PAGE, in close agreement with its predicted size (lane 1). AMS-treated CatR in the presence of 1 mM DTT migrated as a retarded band of 24 kDa (lane 2). Considering the molecular mass of AMS (500 Da), the extent of mobility retardation far exceeds that predicted from alkylation of all four available cysteines in CatR. This could have resulted from the conversion of CatR protein into a more relaxed conformation by AMS attachment. CatR oxidized by H 2 O 2 or diamide did not exhibit any retardation upon AMS treatment, implying that all the cysteine residues were oxidized and formed disulfide bonds under the conditions employed (lanes 4 and 6). In the diamide-treated sample, a small amount of dimer-sized band appeared, as predicted from the formation of intermolecular disulfide bonds. DISCUSSION A number of bacteria contain multiple Fur homologues, whose role is specified in metal assimilation or oxidative stress response. Among three Fur-like proteins found in B. subtilis, Fur and Zur regulate the uptake of iron and zinc, respectively (29,30), whereas PerR acts as a repressor for peroxide regulon (8). Campylobacter jejuni also contains PerR responsible for iron-dependent repression of catalase and alkyl hydroperoxide reductase, in addition to Fur which represses iron uptake genes (31).
S. coelicolor contains at least four Fur homologues; FurA which controls catalase peroxidase (encoded by catC) in a metaldependent manner (12), CatR which controls catalase A production in response to H 2 O 2 , and two Fur-like proteins of unknown function predicted from the genome sequencing data base (S. coelicolor Genome Project, The Sanger Center). Mycobacterium tuberculosis, a close relative of S. coelicolor in being a Gram-positive bacterium of high G ϩ C content, contains two fur-like genes, furA and furB. The furA gene is located upstream of the katG gene encoding catalase peroxidase (32,33), similar to the equivalent gene organization in Streptomyces reticuli (furS-cpeB) (34) and S. coelicolor (furA-catC) (12). This group of Fur homologues seems to regulate the catalase peroxidase gene in a manner sensitive to metals but insensitive to H 2 O 2 , as observed in the regulation of mycobacterial katG or S. coelicolor catC (33,12). On the other hand, CatR seems closely related with B. subtilis PerR in H 2 O 2 -dependent regulation of downstream target genes. Closer sequence match among CatR, PerR, and FurV of S. venezuelae suggests that these regulators may share similar characteristics of peroxide sensing (Fig. 3).
We postulate a model for the rapid adaptation response against H 2 O 2 in S. coelicolor as demonstrated in Fig. 9. Under a normal reducing intracellular environment, CatR is bound to the catA and catR genes and represses their transcription. Upon exposure to H 2 O 2 , the free cysteine thiols of CatR are oxidized to form disulfide bonds, causing loss of DNA binding activity and thus derepression of catA and catR. The induced catalase A efficiently removes H 2 O 2 , whereas the overproduced CatR represses both genes as soon as peroxide is removed, thus forming an efficient negative feedback loop.
In contrast to B. subtilis PerR which is a global regulator of peroxide regulon involving several anti-oxidant genes, CatR in S. coelicolor seems to regulate only CatA production among the known peroxide-degrading enzymes including CatB, CatC, and AhpCD (15). Therefore, the induction of antioxidant enzymes in S. coelicolor seems to require more specified regulators for individual enzymes. In addition, unlike B. subtilis PerR regulon, whose induction at post-exponential phase is inhibited by Mn(II) or Fe(II) (7), catA expression in S. coelicolor is not induced at post-exponential phase and is relatively insensitive to the amount of metals used in our experiments. Addition of excess metal salts or the iron chelator dipyridyl to minimal  2-7). S1 mapping analysis was carried out for catR and catA transcripts as dscribed in the legend to Fig. 4. The same cell extracts were analyzed for CatA, CatC, and CatR proteins by Western blot. media elicited little change in CatA levels (data not shown). Only a high concentration of Ni(II) (1 mM) in Nutrient agar plate medium repressed CatA expression (data not shown). However, in DNA binding assays, deprivation of metals by EDTA inhibited CatR binding, whereas addition of extra metals did not enhance the DNA binding. This indicates that not only reduced thiols of cysteine residues but also appropriate metals are required for CatR to bind to its specific target DNA. Further investigation is needed to reveal the specific cysteine residues and the type of metals, if any, involved in the redox modulation of CatR binding activity.
The four cysteines of CatR in Cys 92 -X 2 -Cys 95 and Cys 132 -X 2 -Cys 135 are conserved in most Fur-like proteins (Fig. 3). If we assume that Fur homologues share basically similar tertiary structures, the cysteine pair Cys 92 and Cys 95 in CatR may be involved in structural zinc binding like its equivalent cysteine pairs in Fur proteins from E. coli and B. subtilis (29,35,36). Disulfide bond formation between these two cysteines may cause release of the coordinating zinc, and loss of DNA binding activity as a result. A similar mechanism has been postulated in the activation mechanism of E. coli heat shock protein Hsp33 (37). Hsp33 is a member of a newly discovered family of heat shock proteins, whose chaperone activity is induced by disulfide bond formation with concomitant release of coordinating zinc (37). Zinc transfer from eukaryotic zinc metallothionein to zinc-depleted protein is also affected by the oxidation state of the thiolate ligands. Oxidation of the thiolate ligands to form disulfide bond causes release of coordinating zinc from the metallothionein (38). We do not rule out the possibility that the C-terminal Cys 132 -X 2 -Cys 135 is also involved in redox-mediated control of CatR activity. The CX 2 C motif is conserved in active sites of disulfide oxidoreductases such as thioredoxin, disulfide isomerase, and Dsb proteins of E. coli (39 -41), participating in reversible disulfide bond formation. Therefore it is quite possible that the C-terminal cysteine pair of CatR could form a disulfide bond and affect the conformation of CatR. Further research is required to address the role of these conserved cysteines and metal binding in regulating the DNA binding The location of the CatR-binding site in catR and catA promoters suggests that CatR repress transcription of these genes via different modes. The transcription from the catRp2 promoter is likely to be blocked at the elongation stage whereas the catA transcription could be blocked at the initiation stage (see Fig. 6). The transcription from the catRp1 promoter, which is partially repressed by CatR, may be blocked at the initiation stage. The binding of CatR near the Ϫ35 promoter element may be effective for repression as observed in catA promoter, whereas binding at closer proximity as observed in catRp1 may be less effective. The similarity between CatR and FurV (from S. venezuelae) in the coding region as well as in the nucleotide sequences of the regulatory region suggests that FurV may act similarly to CatR in H 2 O 2 -dependent regulation of downstream target genes.
In addition to the previously described OxyR and R /RsrA systems, we have here identified another redox-sensitive transcriptional regulator controlling oxidative defense genes in S. coelicolor. Although not much is known about the detailed mechanism of activity modulation, all three regulators seem to involve thiol-disulfide exchange. The rich reservoir of potential transcriptional regulators predicted from the sequence data base (S. coelicolor Genome Project, The Sanger Center) leads us to expect a multitude of regulators responding to myriads of environmental cues. Along with detailed specificity, we expect to discover some unity in regulatory mechanisms in further investigations of the action mechanism of these regulators.