Feed-Forward Regulation of Microbisporicin Biosynthesis in Microbispora corallina

Abstract

Lantibiotics are ribosomally synthesized, posttranslationally modified peptide antibiotics. Microbisporicin is a potent lantibiotic produced by the actinomycete Microbispora corallina and contains unique chlorinated tryptophan and dihydroxyproline residues. The biosynthetic gene cluster for microbisporicin encodes several putative regulatory proteins, including, uniquely, an extracytoplasmic function (ECF) σ factor, σ^MibX, a likely cognate anti-σ factor, MibW, and a potential helix-turn-helix DNA binding protein, MibR. Here we examine the roles of these proteins in regulating microbisporicin biosynthesis. S1 nuclease protection assays were used to determine transcriptional start sites in the microbisporicin gene cluster and confirmed the presence of the likely ECF sigma factor −10 and −35 sequences in five out of six promoters. In contrast, the promoter of mibA, encoding the microbisporicin prepropeptide, has a typical Streptomyces vegetative sigma factor consensus sequence. The ECF sigma factor σ^MibX was shown to interact with the putative anti-sigma factor MibW in Escherichia coli using bacterial two-hybrid analysis. σ^MibX autoregulates its own expression but does not directly regulate expression of mibA. On the basis of quantitative reverse transcriptase PCR (qRT-PCR) data, we propose a model for the biosynthesis of microbisporicin in which MibR functions as an essential master regulator and the ECF sigma factor/anti-sigma factor pair, σ^MibX/MibW, induces feed-forward biosynthesis of microbisporicin and producer immunity.

INTRODUCTION

Microbisporicin is a potent bactericidal lantibiotic produced by strains of the genus Microbispora (1, 18, 20). In addition to one methyl-lanthionine and three lanthionine bridges and a C-terminal S-[(Z)-2-aminovinyl]-d-cysteine (1, 18), modifications found in other lantibiotics, microbisporicin contains the following two unusual modified amino acids: 5-chlorotryptophan and 3,4-dihydroxyproline (1, 18). These modifications are unique to microbisporicin within the lantibiotic class of compounds and are rare in other ribosomally synthesized peptides (although both chlorination and hydroxylation are often found in other natural products [4]). Microbisporicin is active against a wide range of Gram-positive bacterial pathogens, probably through binding to the immediate precursor for cell wall biosynthesis, lipid II (1, 18). Under the commercial name NAI-107, microbisporicin is in preclinical trials and displays efficacy in animal models of multidrug-resistant infections superior to the those of the drugs of last resort, linezolid and vancomycin (10).

A variety of regulatory mechanisms are employed to control lantibiotic production. This may reflect the wide range of input signals that induce lantibiotic biosynthesis in the different producing organisms. Many lantibiotics regulate their own production in a cell density-dependent manner, frequently mediated by a two-component regulatory system (2), but this is often integrated with other external signals, such as developmental cues, pH, and cell stress responses (9, 14, 16, 28). Deletion of mibA, encoding the microbisporicin prepropeptide, from Microbispora corallina NRRL 30420 markedly reduced mib gene expression, including that of the putative regulatory genes mibX and mibR, suggesting that microbisporicin similarly regulates its own biosynthesis (4).

Our previous analysis revealed several genes predicted to encode proteins with possible regulatory functions in microbisporicin production, including, uniquely for an antibiotic biosynthetic gene cluster, an extracytoplasmic function (ECF) σ factor/anti-σ factor complex (4). σ^MibX belongs to the extracytoplasmic function (ECF) family of RNA polymerase σ factors, a subgroup that responds to extracellular signals (e.g., cell envelope stress) by influencing transcription initiation through the recruitment of RNA polymerase core enzyme at relevant promoter sequences (25). ECF sigma factors differ from other σ⁷⁰ proteins (the “vegetative” σ factors) in possessing only the conserved σ₂ and σ₄ regions for DNA binding and interaction with RNA polymerase (21), respectively. Furthermore, the consensus motif found at many ECF-dependent promoters differs from those of the σ⁷⁰ proteins, containing an “AAC” motif in the −35 region and “CGT” nucleotides clustered in the −10 region (8, 17). Many ECF sigma factors autoregulate their own expression (27). A putative ECF sigma factor consensus motif was identified in five of the six likely promoter regions present in the microbisporicin gene cluster, including that of mibX (Fig. 1) (4).

Fig. 1.

Microbisporicin gene cluster. mibA encodes the microbisporicin prepropeptide and is cotranscribed with genes encoding biosynthetic enzymes for formation of lanthionine bridges (mibBC) and S-[(Z)-2-aminovinyl]-d-cysteine (mibD) and with genes encoding a putative exporter (mibTU) and a gene of unknown function (mibV). A two-component ABC transporter encoded by mibEF and a lipoprotein encoded by mibQ may confer immunity to microbisporicin (4). Further modifications are mediated by a tryptophan halogenase-flavin reductase couple (mibHS) and likely by a cytochrome P450 (mibO). Putative regulatory proteins are encoded by mibXW and mibR. mib genes of unknown function are mibJ, mibYZ (possibly encoding an ABC transporter), and mibN (encoding a sodium/proton antiporter). For a detailed description of mib gene function, see reference 4. The positions of the predicted ECF sigma factor promoter consensus sequences (full arrows) and the expected position of the promoter for mibA (broken arrow) are shown.

For the majority of characterized ECF sigma factors, a second protein, usually coordinately expressed, is involved in regulating ECF sigma factor activity. This anti-σ factor is able to sequester the σ factor away from its promoter binding sites, thus preventing transcription initiation until the receipt of a specific signal (27). This signal would probably be sensed by the anti-σ factor or a protein regulating its activity, leading to the release of the σ factor and subsequent expression of its target genes. This release can be mediated via a conformational change (for example, regulation of σ^R by RsrA in Streptomyces coelicolor [25]) or via proteolytic degradation (7). mibW, downstream of and apparently translationally coupled to mibX, encodes a protein predicted to function as an anti-sigma factor protein for σ^MibX (4).

We previously suggested a model in which σ^MibX is required for high-level expression of the microbisporicin gene cluster (including its own gene) and in which its activity depends on low levels of production of microbisporicin (4). We proposed that low levels of microbisporicin (potentially produced by expression from a starvation-induced promoter) or cell envelope stress (induced by the likely interaction of microbisporicin with lipid II) prevents MibW from interacting with σ^MibX, thus releasing the ECF σ factor and resulting in high-level expression of the entire microbisporicin gene cluster. Interestingly, the operon beginning with mibA lacks the ECF sigma factor consensus motif, suggesting that regulation by σ^MibX might be mediated through a gene in one of the other σ^MibX-regulated operons. We suggested that a likely candidate for this is mibR, which encodes a protein with a predicted helix-turn-helix DNA binding domain between amino acids 173 and 214 (Pfam) (3).

In this study, we characterize the functions of σ^MibX, MibW, and MibR and their roles in regulating microbisporicin biosynthesis in M. corallina.

MATERIALS AND METHODS

Strains and general methods.

Oligonucleotides and plasmids are described in Table S1 in the supplemental material. M. corallina NRRL 30420 was grown and manipulated, and microbisporicin was detected, as described previously (4).

High-resolution S1 nuclease mapping.

RNA was purified from exponential-phase wild-type M. corallina as described previously (4). Probes for S1 nuclease protection analysis were generated by PCR using the oligonucleotide pairs shown in Table S1 in the supplemental material. Nonhomologous tails (underlined) were incorporated into some probes to distinguish between full-length protection and probe/probe reannealing. Asterisks mark the oligonucleotides, internal to the respective protein-coding sequences, which were labeled at the 5′ end with [γ-³²P]ATP. The oligonucleotides were labeled using T4 polynucleotide kinase (Epicentre) by following the manufacturer's instructions. Hybridizations were carried out using 30 μg wild-type M. corallina RNA in sodium trichloroacetic acid buffer at 45°C overnight after denaturation at 70°C for 10 min (15). G+A sequencing ladders were generated from the end-labeled probes by chemical sequencing (22).

Bacterial two-hybrid analysis.

pT25 contains a multiple-cloning site (MCS) preceded by a fragment of cya of Bordetella pertussis that encodes half of the catalytic domain (T25) of adenylate cyclase (CyaA) (12). XP461 is pT25 containing the leucine zipper fragment (zip) (13). The other half of CyaA (T18) is encoded by pUT18 and is preceded by a MCS. pUT18C is similar to pUT18 except that the MCS follows the cya gene fragment, allowing the generation of C-terminal rather than N-terminal fusion proteins (13). The leucine zipper fragment from XP461 was removed by KpnI digestion and cloned in the KpnI sites of pUT18 and pUT18C to generate the positive-control plasmids pUT18-zip and pUT18C-zip, respectively.

mibX was amplified by high-fidelity PCR from pIJ12125 using the primers LF101F and LF101R (no stop codon included) and primers LF101F and LF101R2 (stop codon included), which introduced BamHI and KpnI sites (5′ and 3′, respectively, with respect to mibX) into the resulting PCR products. The PCR products were gel purified, digested with BamHI and KpnI, and ligated into pUT18 and pUT18C cut with BamHI and KpnI to generate pIJ12367 and pIJ12368, respectively. mibW was amplified by high-fidelity PCR from pIJ12125 using primers LF102F and LF102R, which introduced BamHI and KpnI sites into the resulting PCR product. The PCR product was gel purified, digested with BamHI and KpnI, and ligated into XP458 (pT25) cut with BamHI and KpnI to generate pIJ12369. All constructs were confirmed by Sanger sequencing.

Combinations of the resulting plasmids were introduced into Escherichia coli BTH101 by cotransformation and screened on MacConkey medium (Difco) containing 1% maltose, 0.5 mM IPTG (isopropyl-β-d-thiogalactopyranoside), 25 μg/ml chloramphenicol, and 100 μg/ml carbenicillin. Plates were incubated at 30°C for several days until the color of the colonies had developed fully. Quantification of β-galactosidase activity was carried out as previously described (26). Two independent clones from each interaction plate were assayed for β-galactosidase activity, and the values were averaged.

Luciferase assays.

The intergenic region between mibX and mibA, with and without mibX, was amplified by PCR using primer pairs containing EcoRI and BamHI sites and in both orientations, with respect to the restriction sites. The primers used were LF096F and LF096R (pIJ12341), LF096F2 and LF096R (pIJ12342), LF097F and LF097R (pIJ12343), and LF097F and LF097R2 (pIJ12344). The resulting PCR fragments were cloned into EcoRI-BamHI-cut pIJ5972, an integrative Streptomyces promoter-probe plasmid carrying TTA codon-free derivatives of the luxAB reporter genes (19). The resulting constructs and pIJ5972 (negative control) were transferred by conjugation into S. coelicolor M1146 (5). Plasmid-containing strains were grown on Difco nutrient agar in single wells of a 25-well plate (10 cm × 10 cm; Sterilin) for 2 days. Each well was inoculated with approximately 5 × 10⁶ spores. Plates were exposed to filter paper impregnated with n-decanal for 5 min, and luciferase activities were observed using a NightOwl camera (Berthold) equipped with WinLight software (Berthold) using a 1-min exposure time.

Construction of M. corallina deletion mutants.

Genes carried by pIJ12125 (4) were replaced with an apramycin-resistant (apr)-oriT cassette amplified from pIJ773 using the primer pairs listed in Table S1 in the supplemental material, as described in reference 6. Mutations were confirmed by PCR using flanking primers and by restriction digests to confirm the integrity of the targeted cosmid. Conjugations between E. coli ET12567/pUZ8002 carrying the oriT-containing cosmid and M. corallina were carried out as previously described (4). Deletion mutants of M. corallina were constructed as described previously (4).

Complementation of ΔmibR::apr.

To complement the mibR deletion mutant, a PCR product was generated by high-fidelity PCR using primers LF117F and LF112R that contained 32 bp of the 3′ end of mibQ (including the stop codon), the intergenic region between mibQ and mibR (including the promoter P_mibR identified by S1 mapping), and mibR. This fragment was blunt-end cloned into the EcoRV site of pIJ10706 (4) to generate pIJ12376, which was confirmed by Sanger sequencing. pIJ12376 and pIJ10706 (the empty vector control) were transferred into the ΔmibR mutant by conjugation from E. coli ET12567/pUZ8002.

qRT-PCR.

M. corallina wild-type and mutant strains were cultured in triplicate as described previously (4). Mycelial growth was followed by measuring the optical density at 450 nm of 1 ml of culture taken from each of three replicate cultures at different time points. RNA was extracted from 2.5 ml of mycelium of each sample at 48 h of growth as described previously (4). A total of 2.5 μg RNA was treated with DNase I in a 25-μl volume by following the manufacturer's instructions (amplification grade; Invitrogen). A total of 8 μl of each of the resulting RNA samples was converted to cDNA (4). To control for DNA contamination in the quantitative reverse transcriptase PCR (qRT-PCR), a duplicate set of cDNA synthesis reactions were performed but with the reverse transcriptase enzyme omitted. Following RNase H treatment, the samples were diluted 1:100 with nuclease-free water, and 2.5 μl was used in the quantitative PCR with SYBR Greener qPCR Supermix (Invitrogen), according to the manufacturer's instructions. Each 25-μl reaction mixture contained 200 nM forward and reverse primers and 5% dimethyl sulfoxide (DMSO). PCR cycling was performed in a Chromo4 machine (Bio-Rad, CA) at 50°C for 2 min and 95°C for 10 min, followed by 40 cycles of 95°C for 15 s, 58°C for 60 s, and 72°C for 1 s. Parallel reactions were performed in the same 96-well plate using different dilutions of M. corallina genomic DNA to generate a standard curve for each gene analyzed. All determinations were performed in triplicate, and the results were analyzed using Opticon 2 Monitor software (MJ Research, Waltham, MA). Values were normalized to an endogenous control gene, hrdB, encoding the homolog of the vegetative sigma factor of S. coelicolor (4). All the control samples from cDNA synthesis lacking reverse transcriptase gave values comparable to that of the background, indicating that the RNA samples were not contaminated with genomic DNA.

RESULTS

High-resolution S1 nuclease mapping of transcriptional start sites in the microbisporicin gene cluster.

Bioinformatic analysis of the mib gene cluster revealed five likely intergenic regions, with one of them being divergent. The mib gene cluster was thus predicted to consist of six operons (Fig. 1), and ECF sigma factor consensus motifs were identified upstream of five of them (4).

³²P-labeled probes homologous to the predicted 5′ ends of the transcripts of mibJ, mibQ, mibR, mibX, mibA, and mibE were used to map the transcriptional start site of each gene by S1 nuclease protection analysis (Fig. 2; see also Fig. S1 in the supplemental material). S1 nuclease-protected fragments were sized by polyacrylamide gel electrophoresis and by comparison to DNA size markers. Fragments that could not be accurately sized by this method were further analyzed by comparison to a GA sequencing ladder derived from the respective 5′ end-labeled probe (Fig. 2; see also Fig. S1).

Fig. 2.

High-resolution S1 nuclease mapping of the 5′ end of the mibA and mibE transcripts using PCR-generated probes and RNA from wild-type M. corallina NRRL 30420. The most likely transcription start site is indicated by an asterisk. The full-length probe was run as a control for the identification of full-length protection. The Maxam-Gilbert GA chemical sequencing ladder was generated from the full-length probe (22). The marker was ³²P-labeled HinfI-digested ΦX174 DNA. For mibE, the length of the protected fragment (determined from the size markers) and the sequencing ladder were used to unambiguously assign the transcriptional start site.

The transcriptional start sites of mibJ, mibQ, mibR, mibX, and mibE lie 8 to 10 nucleotides downstream of the identified ECF sigma factor consensus −10 sequences (Fig. 3). This suggests that these genes and operons are subject to direct transcriptional control by the ECF sigma factor σ^MibX, which is essential for microbisporicin biosynthesis (4).

Fig. 3.

Summary of the results of high-resolution S1 nuclease mapping of the 5′ ends of transcripts from mibJ, mibX, mibR, mibQ, and mibE using PCR-generated probes and RNA from wild-type M. corallina NRRL 30420. The predicted −35 and −10 elements are conserved and are highlighted with gray boxes. The determined transcriptional start sites (+1) are highlighted in boldface.

Nonhomologous sequences at the 5′ end of the probes for mibJ, mibQ, and mibR were used to distinguish between transcriptional read-through from upstream promoters and probe/probe reannealing. Fragments approximately 15 nucleotides shorter than the respective probes for mibQ and mibR indicated the additional presence of transcriptional read-through into these genes (see Fig. S1 in the supplemental material).

The transcriptional start site of mibA was not preceded by an ECF sigma factor consensus sequence (Fig. 4). Instead, −35 and −10 sequences closely matching those of the consensus sequence of the housekeeping sigma factor of Streptomyces coelicolor (29) were identified upstream of the mibA transcriptional start (Fig. 4). The predicted −35 element of mibA matches exactly those of several S. coelicolor and E. coli σ⁷⁰ promoters, and the −10 element includes the strictly conserved “T” at position 6 as well as the preferred nucleotides “T” at position 1 and “GA” at positions 3 and 4, respectively (29). The spacing of 17 nucleotides between these two elements is also characteristic of this group of promoters (29). This suggests that mibA transcription, unlike that of the other mib genes, is not mediated by σ^MibX but probably by the housekeeping sigma factor of M. corallina. However, transcription of mibA might be influenced by other transcriptional regulators acting at the mibA promoter.

Fig. 4.

Predicted promoter site of mibA. The transcription start site indicated by S1 nuclease mapping is shown (+1) in boldface. The predicted −35 and −10 regions are underlined.

Analysis of the interaction between σ^MibX and MibW.

MibW was proposed previously to function as an anti-sigma factor that regulated the activity of σ^MibX (4). MibW would bind to σ^MibX, possibly through its predicted N-terminal cytoplasmic domain, and thus tether it to the membrane, where MibW would be embedded via its predicted transmembrane helices. To determine if an interaction occurred between MibW and σ^MibX, a bacterial two-hybrid (BACTH) experiment was performed.

mibX was fused to the gene encoding the T18 fragment of adenylate cyclase in the following two vectors: pUT18, in which σ^MibX would be at the N terminus of the fusion protein (pIJ12367), and pUT18C, in which σ^MibX would be at the C terminus of the fusion protein (pIJ12368). mibW was introduced into pT25, resulting in fusion of the putative anti-sigma factor to the C terminus of the T25 fragment of adenylate cyclase. pT25 and pUT18 or pUT18C, all containing the leucine zipper fragment (zip) from GCN4 (yeast protein), were used as positive controls. The leucine zipper domain interacts strongly with itself and provides a reliable positive control for bacterial two-hybrid studies (12). Negative controls used were the empty vectors.

Interacting pairs of proteins were screened initially by transforming E. coli BTH101 with the appropriate plasmids and monitoring restoration of adenylate cyclase activity on MacConkey-maltose indicator agar plates incubated at 30°C for 4 days (12). Interaction was readily observed between T25-MibW and both T18-σ^MibX and T18C-σ^MibX (data not shown), indicating that the position of the T18 fragment did not affect the ability of MibW to interact with the ECF sigma factor. Two clones of each interaction pair were assayed for β-galactosidase activity. Interaction between MibW and σ^MibX was confirmed, with β-galactosidase activity shown to be higher than that of the positive-control leucine zipper fragments (Fig. 5).

Fig. 5.

Bacterial two-hybrid experiment to investigate the interaction between MibW and σ^MibX. The listed pairs of constructs were transferred into the BACTH reporter strain E. coli BTH101 by transformation. The resulting transformants were selected on MacConkey-maltose agar containing 100 μg/ml carbenicillin and 25 μg/ml chloramphenicol and incubated at 30°C for several days before two independent clones were picked and subjected to β-galactosidase assays. The histogram was plotted using the average activity levels in Miller units of the two clones. Error bars represent the spread of values between clones tested.

σ^MibX induces transcription from the mibX promoter in Streptomyces coelicolor.

A series of constructs were derived from the reporter plasmid pIJ5972, which contains a promoterless luxAB cassette encoding the enzyme luciferase and which integrates in the ΦC31 attachment site of the chromosome of S. coelicolor (Fig. 6) (19). pIJ12341 contained the intergenic region between mibX-mibA (with the mibA promoter driving expression of luxAB) and mibX (with expression driven from its own promoter), whereas pIJ12342 contained only the intergenic region. pIJ12343 contained the intergenic region between mibX-mibA (with the mibX promoter driving expression of luxAB) and mibX, whereas pIJ12344 contained only the intergenic region (Fig. 6). These constructs, along with pIJ5972, were introduced into S. coelicolor M1146 by conjugation (5). Three independent clones were selected from each conjugation and grown in duplicate for 2 days on Difco nutrient agar (on which S. coelicolor does not make aerial hyphae and spores, which can interfere with light emission). The mycelium was exposed for 5 min to n-decanal that had been spotted onto filter discs, and light emission was detected using a NightOwl camera. No light production was detected from the vector-only control strain or from strains containing pIJ12341 or pIJ12342, where luxAB were located downstream of the mibA promoter. A very low level of light production from the strains containing pIJ12344 was observed, indicative of a low level of transcriptional activity from the mibX promoter in the absence of σ^MibX, possibly mediated by one of the other ECF sigma factors of S. coelicolor (25). In contrast, high levels of light emission from the pIJ12343 clones containing mibX and with the luxAB genes transcribed from the mibX promoter were observed. This suggests that σ^MibX positively autoregulates its own expression but does not direct transcription from the mibA promoter.

Fig. 6.

Luciferase reporter analysis of σ^MibX activity in S. coelicolor M1146. pIJ5972 (bottom) contains promoterless luxAB. In pIJ12341, the mibA promoter (P_mibA) transcribes luxAB in the presence of σ^mibX. pIJ12342 differs by the absence of σ^mibX. In pIJ12343, the mibX promoter (P_mibX) transcribes luxAB in the presence of σ^mibX. pIJ12344 differs by the absence of σ^mibX. Three independent clones of S. coelicolor M1146 (shown side by side) containing these constructs were grown in duplicate (indicated by numbers 1 and 2 to the left of the images) for 2 days on Difco nutrient agar, and light production was visualized using a NightOwl camera (Berthold) after applying the substrate n-decanal on filter paper discs for 5 min. The images shown are representative examples obtained from three independent experiments. E, EcoRI; B, BamHI.

mibR is essential for production of microbisporicin in M. corallina.

pIJ12125, containing the entire mib gene cluster but with the apramycin cassette from pIJ773 replacing mibR, was mobilized into M. corallina by conjugation. Two clones were identified that gave rise to single colonies that were apramycin resistant yet sensitive to kanamycin (resistance conferred by the cloning vector), indicating the occurrence of double crossing-over. Replacement of mibR in these clones by the apramycin resistance cassette was confirmed by PCR. Since mibR lies in a monocistronic operon (Fig. 1) and is adjacent to convergently transcribed mibW, its deletion should not have polar effects on the expression of other mib genes. The two clones were grown in VSPA liquid medium (4) for 7 days along with the wild-type strain. Unlike the wild-type control, supernatants from the mutant strains did not generate a zone of inhibition when spotted onto a lawn of Micrococcus luteus (Fig. 7, showing one representative clone). Bioactivity and microbisporicin production (confirmed by matrix-assisted laser desorption ionization–time of flight [MALDI-TOF] analysis) were restored by complementation in trans, with mibR expressed from its own promoter. Thus, mibR is essential for microbisporicin biosynthesis.

Fig. 7.

Analysis of the effect of replacing mibR with the apr cassette from pIJ773. Mutant (ΔR) and wild-type (WT) strains of M. corallina NRRL 30420 were grown for 7 day in VSPA medium, and 40 μl of culture supernatant was assayed for activity against M. luteus.

Deletion of mibX or mibR affects transcription of the mib gene cluster.

Wild-type M. corallina and the ΔmibX::apr and ΔmibR::apr mutants were grown in triplicate in VSPA liquid medium for RNA isolation and in triplicate for the assessment of mycelial growth by measuring the optical density at 450 nm. The three strains grew initially at similar rates, although the ΔmibX::apr mutant exhibited a slower growth rate at later time points but accumulated to the same optical density in stationary phase as the other strains (see Fig. S2 in the supplemental material). Samples of supernatant were taken at intervals to assess microbisporicin biosynthesis by bioassay against M. luteus. The earliest time point at which bioactivity could be detected in the wild type was 40 h (see Fig. S2). There was no bioactivity with the supernatants from the ΔmibX::apr or ΔmibR::apr mutants even after 7 days of growth (the latest time point assayed) (data not shown).

To assess the effect of deletion of mibX and mibR on transcription of the mib gene cluster, RNA was isolated from 2.5 ml of mycelium from each of the three replicate cultures of each strain after 48 h of growth, i.e., after microbisporicin biosynthesis had begun in the wild-type strain. mib gene expression was assessed by quantitative reverse transcriptase PCR using the M. corallina homolog of hrdB, the vegetative sigma factor gene of S. coelicolor, as an internal control (4). Representative genes were chosen from each predicted operon of the mib cluster, and their expression levels were assessed using the primers listed in Table S1 in the supplemental material.

Expression of all of the mib genes tested was strongly dependent on both mibX and mibR (Fig. 8). Interestingly, the mibA and mibD expression levels were 4- and 5.4-fold lower in the ΔmibR mutant than in the ΔmibX mutant, respectively. Conversely, levels of expression for mibJ, mibQ, and mibE were 1.84-, 6.95-, and 2.4-fold lower in the ΔmibX mutant than in the ΔmibR mutant, respectively. Compared to wild-type levels, expression of mibR was reduced to a far lesser extent (8.8-fold) in the ΔmibX mutant than mibJ, mibQ, or mibE (2,561-, 1,398-, and 3,124-fold, respectively). These results are consistent with direct regulation of mibJ, mibQ, and mibE by σ^MibX and direct regulation of the mibA operon by MibR.

Fig. 8.

Analysis of the effect of deletion of mibX or mibR on mib gene expression. The expression of each gene displayed on the vertical axis was assessed by qRT-PCR in three biological replicates of M. corallina NRRL 30420 ΔmibX::apr (gray histograms) and ΔmibR::apr (black histograms). The number of copies of each transcript was normalized to the number of copies of hrdB in each sample. The normalized copy numbers for the three biological replicates in each sample were averaged and are displayed as percentages of wild-type expression.

DISCUSSION

Regulation of microbisporicin biosynthesis is complex and appears to rely on a strict interplay between three regulatory proteins, MibR, σ^MibX, and MibW, and production of microbisporicin itself. S1 nuclease protection analysis confirmed that the previously identified putative ECF sigma factor consensus sequences lie within appropriate distances from transcriptional start sites to act as promoter sequences for all of the operons in the mib cluster except mibA. This strongly suggests that σ^MibX, an ECF sigma factor encoded within the cluster and essential for microbisporicin production, directs the transcription of these operons. Furthermore, σ^MibX directed its own expression in a heterologous host, presumably through recruitment of RNA polymerase to the consensus motif in its own promoter region.

Interaction between full-length σ^MibX and MibW was confirmed in E. coli. This is consistent with MibW functioning as a σ^MibX-specific anti-sigma factor, sequestering it at the membrane (MibW is a predicted transmembrane protein) in the absence of an activating signal. MibW, by regulating σ^MibX activity, would thus play a crucial role in regulating microbisporicin production. Our attempts to delete mibW in M. corallina have proved unsuccessful, resulting in the isolation of an apparent suppressor mutation in mibX that is likely to abolish sigma factor function (data not shown). Thus, in the absence of MibW, constitutive production of microbisporicin might overwhelm the organism's immunity system, potentially mediated by MibEF (4), resulting in cell death. Attempts to complement the ΔmibX::apr mutation with mibXW expressed in trans from their native promoter restored microbisporicin production but not to wild-type levels, potentially reflecting a surfeit of MibW in the complemented strain and highlighting the importance of a 1:1 stoichiometry for the two proteins for regulation.

Our results suggest a model for the regulation of microbisporicin biosynthesis in which MibR functions as a master regulator to promote low-level microbisporicin biosynthesis, which subsequently induces a feed-forward mechanism mediated by σ^MibX that results in high-level microbisporicin production (Fig. 9). Prior to the onset of detectable microbisporicin production, a basal level of expression of mibXW leaves the system poised for activation, with σ^MibX sequestered at the membrane. We propose that an unknown signal, possibly nutrient limitation, results in activation of transcription of mibR mediated through a growth rate-dependent promoter. This would result in expression of the mibABCDTUV operon and production of a form of microbisporicin that lacks chlorination of tryptophan at position 4 and hydroxylation of proline at position 14, which may therefore possess reduced antibiotic activity (18). Export, potentially mediated by MibTU, would permit either direct interaction of the peptide with MibW or a low level of inhibition of peptidoglycan biosynthesis that may be perceived by the anti-sigma factor. Either signal could then result in inactivation of MibW, release of σ^MibX, and high-level expression of the entire mib gene cluster, including genes we predict to confer immunity to microbisporicin, namely, those encoding the ABC transporter MibEF (4) and the putative lipoprotein MibQ (LanI lipoproteins confer immunity in other lantibiotic-producing organisms [2]). Furthermore, induction of expression of mibHS, required for tryptophan chlorination, and of mibO, encoding a cytochrome P450 probably involved in proline hydroxylation, would result in the formation of fully processed and active microbisporicin (4).

Fig. 9.

Model for the regulation of microbisporicin biosynthesis. Prior to detectable microbisporicin production, MibW sequesters σ^MibX at the membrane, preventing its interaction with target promoters. An unknown signal (indicated by the question mark) activates transcription of mibR at an unidentified promoter (P_?). MibR then activates transcription of the mibABCDTUV operon, leading to production of microbisporicin. Interaction of the peptide with MibW, or a low level of inhibition of peptidoglycan biosynthesis that may be perceived by the anti-sigma factor, results in inactivation of MibW, release of σ^MibX, and high-level expression of the entire mib gene cluster.

Deletion of mibX had a less severe effect on expression of mibA and mibD than deletion of mibR, consistent with direct regulation of mibABCDTUV by MibR and not by σ^MibX. This is also consistent with the results of the luciferase assays in which mibA expression was not induced by σ^MibX. Conversely, expression of mibJ, mibQ, and mibE was more severely affected by deletion of mibX than by deletion of mibR, consistent with their direct regulation by σ^MibX and with the presence of the ECF sigma factor consensus sequence upstream of their respective transcriptional start sites.

Feed-forward regulation of antibiotic biosynthesis, in which biosynthetic pathway intermediates as well as the final product are postulated to activate export and potentially immunity mechanisms, was suggested previously for actinorhodin production in Streptomyces coelicolor (30), and other examples also exist (11, 23–24).

Our results not only contribute to our knowledge of the regulatory mechanisms used to control lantibiotic biosynthesis in actinomycetes but also provide a basis for rational attempts to improve the level of microbisporicin production for pharmaceutical development and application by manipulation of the regulatory genes mibR and mibXW.