|Home | About | Journals | Submit | Contact Us | Français|
Streptomyces coelicolor mutants lacking the zinc-responsive Zur repressor are conditionally defective in sporulation, presumably due to the overexpression of one or more Zur target genes. Gene disruption analyses revealed that deregulation of previously known Zur targets was not responsible for the sporulation phenotype. We used microarrays to identify further Zur targets and discovered that Zur controls a cluster of genes predicted to direct synthesis of an uncharacterized siderophore-related non-ribosomally encoded peptide designated coelibactin. Disruption of a key coelibactin biosynthetic gene suppressed the Zur sporulation phenotype, suggesting that deregulation of coelibactin synthesis inhibits sporulation.
Zinc is an essential element in all organisms, generally acting as a metalloenzyme cofactor or to stabilize protein folds. Several uptake systems have been characterized for bacteria, including the widely distributed high-affinity ZnuABC ABC transporter (16). In addition, ribosome remodeling has emerged as a novel zinc homeostasis mechanism in recent years (1, 12, 15). This involves the production of zinc-free ribosomal proteins (R proteins) to replace up to seven different zinc-binding R proteins. Zinc is usually coordinated by cysteine residues in the zinc R proteins, and these are therefore termed C+ proteins, whereas the alternative R proteins that lack cysteines are termed C− proteins (11). The genes that encode ZnuABC systems and alternative C− R proteins are usually controlled by the zinc-responsive transcriptional repressor Zur (4, 15, 16). In the antibiotic-producing actinomycete Streptomyces coelicolor, Zur negatively regulates seven C− R proteins, including five located in a single “RPC−” cluster, as well the znuACB operon (13, 18). Interestingly, when grown as a confluent lawn on solid agar, the S. coelicolor Δzur::apr mutant S121 (13) displayed a white phenotype, as opposed to the normal gray appearance that derives from mature spore pigment (Fig. (Fig.1A).1A). Conversely, single dispersed colonies fully develop the gray spore pigment (data not shown). The sporulation defect was also seen on zinc-depleted minimal media and on MS agar containing excess zinc (100 μM) (data not shown). Scanning electron microscopy revealed poorly septated aerial hyphae in white confluent areas of growth, whereas normal spore chains were detected in gray colonies (data not shown). Therefore, the deletion of zur conditionally inhibits sporulation, probably through the overexpression of one or more target genes.
We constructed a series of double deletion mutants to test whether the deregulation of the znuACB operon or the C− R proteins was responsible for the ΔZur phenotype. A standard PCR targeting and λ-Red-mediated recombination approach was used to construct mutant alleles in cosmid-borne genes, which were then transferred to S. coelicolor by conjugation and homologous recombination (5, 6). First, an ΔRPC−::apr allele was constructed in cosmid StE9 (17). The DNA encoding rpsR2 (SCO3425) to SCO3431 (chromosomally equivalent to nucleotide [nt] 3789250 to nt 3792451) was replaced with an apramycin (apr) resistance cassette from pIJ773 (5). The FLP recombinase recognition target (FRT) sequences that flank the apr cassette were exploited to recombine out the cassette using Escherichia coli EL250 (araC-PBADflpe ). Following this, the apr cassette from pIJ773 was reintroduced into the cosmid, this time replacing the amp gene, and the markerless ΔRPC− allele was introduced into S. coelicolor M145 by conjugation from the donor strain ET12567(pUZ8002) (14). Following the isolation of single-crossover recombinants (Aprr), putative double-crossover mutants (Aprs) were isolated and then verified by Southern analysis (data not shown). The resulting strain, S153 (ΔRPC−), had no obvious phenotype even when grown under zinc depletion, confirming that the five C− R proteins encoded by this cluster are nonessential. A Δzur::apr allele (13) was recombined into the resulting strain to generate S154 (ΔRPC− Δzur::apr). S154 retained the white ΔZur phenotype (Fig. (Fig.1A),1A), indicating that overexpression of RPC− cluster genes is not the underlying cause of the morphological defect. Similarly, a mutant allele that replaced the entire znuACB-zur cluster (nt 2704007 to nt 2707218) with the apr cassette from pIJ773 in cosmid StC121 was constructed (17). The ΔznuACB-zur::apr mutant, designated S151, also maintained the white ΔZur phenotype (Fig. (Fig.1A).1A). The ΔznuACB-zur::apr mutant grew poorly on defined minimal media lacking added zinc but formed normal-sized colonies when supplemented with 5 to 25 μM ZnCl2. This aspect of the phenotype was similar to that of the ΔznuA mutant S150, which was constructed in a similar manner to S153, in which the entire znuA coding sequence was deleted (chromosomally equivalent to nt 2704007 to nt 2704988), and confirmed that the ZnuABC system plays a key role in zinc uptake in S. coelicolor.
In order to identify additional candidate genes that might underlie the ΔZur sporulation defect, we investigated the Zur regulon by using a microarray approach. Triplicate cultures of M145 (wild type [wt]) and S121(Δzur::apr) were grown to mid-log phase (optical density at 450 nm [OD450] of ~0.8) in NMMP liquid medium (8), which includes 3.5 μM added ZnCl2. Cultures were treated with RNAprotect (Qiagen) to prevent further RNA synthesis and degradation, and then RNA was isolated as described in the supplemental material. M145 or S121 total RNA was labeled with Cy5-dCTP or Cy3-dCTP, respectively, and then fluorescent cDNAs were cohybridized to whole-genome-PCR-spotted microarray slides (University of Surrey Microarray Group) and scanned using an Axon Genepix 4000B microarray scanner. Following filtering procedures to remove bad spots and poorly expressed genes, we identified 52 genes that were >2-fold upregulated in S121 compared to levels for M145 (P value of <0.05) (see Table S1 in the supplemental material). We screened these genes and potentially cotranscribed genes for upstream Zur operator sequences by using the consensus ATGnnnnTCnTTTT (where n is any nucleotide) (13), and along with previously identified Zur targets, we identified two new gene clusters that might be controlled by Zur (Fig. (Fig.2;2; Table Table1).1). The SCO472-77 cluster encodes a likely ABC metal uptake system that might encode an alternative route for zinc acquisition. Indeed, SCO0473 is homologous to ZnuA (42% similarity), SCO0475 is homologous to ZnuB (46% similarity), and SCO0476 is homologous to ZnuC (50% similarity). SCO0473 and SCO475 are also related to Rv2059 and Rv2060, respectively, which were shown to be controlled by Zur in Mycobacterium tuberculosis (10). The SCO7676-92 cluster is predicted to be involved in the production of an uncharacterized non-ribosomally encoded peptide with siderophore characteristics (2). The 17-gene cluster is arranged as at least three transcriptional units, with transcription initiating upstream of genes encoding ferredoxin (SCO7676), an adenylate-forming enzyme (SCO7681), and a nonribosomal peptide synthetase (SCO7682) (Fig. (Fig.2A).2A). Although only four genes in this cluster exhibited >2-fold-increased levels of expression, most of the remaining genes were significantly overexpressed in the Δzur mutant, suggesting that the whole cluster is controlled by Zur. To confirm that Zur binds to the SCO0474/75 and SCO7681/82 intergenic regions, as well as upstream from SCO7676, we performed electromobility shift assays using ~200-bp PCR products that included the putative Zur binding sites and purified Zur (Fig. (Fig.2B),2B), as described previously (13). In each case, Zur bound specifically to the probes, exhibiting a concentration-dependent laddering effect, as previously observed at other Zur targets (13).
Quantitative real-time PCR (Q-PCR) was used to confirm that genes in these clusters were modulated by zinc. S. coelicolor M600 (8) was grown in NMMP medium lacking the addition of minor-elements solution to mid-exponential phase and then treated with 25 μM ZnSO4. RNA was prepared as described previously (7) from mycelium sampled immediately before and 30 min after zinc addition. Following this, the culture was treated with the nonspecific chelator EDTA (2 mM), and RNA was sampled 30 min later. Each gene tested was negatively regulated by zinc in a similar manner to znuA (SCO2505) (Fig. (Fig.3).3). Although SCO7681 and SCO7682 did not appear to be induced significantly in the microarray experiment, each was zinc responsive. This apparent discrepancy suggests an additional regulatory mechanism; indeed, it was recently discovered that the pleiotropic regulator AbsC negatively controls SCO7681 and SCO7682 by binding to a site that overlaps the Zur binding site (A. Hesketh and M. J. Bibb, personal communication).
We hypothesized that the overexpression of the coelibactin cluster might be the underlying cause of the white ΔZur phenotype. To test this, a Δzur::hyg allele was constructed by replacing the zur coding sequence with a hygromycin resistance cassette in cosmid StC121 (A. Hesketh and M. J. Bibb, personal communication). The resulting strain, M1016, has a ΔZur phenotype identical to that of the Δzur::apr mutant S121. This allele was also introduced into M904, an M145 derivative in which SCO7682 was disrupted 279 nt into the coding region by Tn5062. M904 displays a wild-type sporulation phenotype and grows normally on media deprived of added zinc (data not shown; A. Hesketh and M. Bibb, personal communication; 3). Strikingly, the Δzur::hyg SCO7682::Tn double mutant also sporulated normally (Fig. (Fig.1B),1B), suggesting that the morphological phenotype detected in Δzur mutants is largely caused by the overexpression of the coelibactin cluster.
The morphological defect observed in Δzur mutants appears to be due to the overexpression of a cluster of genes predicted to direct the production of coelibactin. Coelibactin has not been isolated or characterized in any way, and yet bioinformatic approaches suggest that it is a siderophore-related compound. By definition, siderophores bind extracellular iron in order to increase bioavailability of this metal. However, our data suggest that coelibactin might be a zinc-chelating compound and, importantly, reveal approaches to overexpress the biosynthetic genes enabling its isolation and characterization. It is unclear why coelibactin overexpression should impede the sporulation process.
The microarray data described in this paper can be accessed at ArrrayExpress (accession number E-MAXD-56).
We thank Colin Smith and the other members of the University of Surrey Microarray group for help and advice, Mervyn Bibb and Andrew Hesketh for communication of unpublished data and supply of strains, and Greg Challis for useful discussions. We acknowledge the assistance of Kassimatis Dimitrios and David Randall in the construction and analysis of mutants.
The work was supported by BBSRC grant BBC5038541 and a studentship to B.P. and by a University of Sussex studentship to G.A.O.
Published ahead of print on 13 November 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.