Extracellular signalling, translational control, two repressors and an activator all contribute to the regulation of methylenomycin production in Streptomyces coelicolor

Authors


*E-mail keith.chater@bbsrc.ac.uk; Tel. (+44) 1603 450297; Fax (+44) 1603 450045.

Summary

Bioinformatic analysis of the plasmid-linked gene cluster associated with biosynthesis of methylenomycin (Mm) suggested that part of the cluster directs synthesis of a gamma-butyrolactone-like autoregulator. Autoregulator activity could be extracted from culture fluids, but differed from gamma-butyrolactones in being alkali resistant. The activity has recently been shown to comprise a series of novel autoregulator molecules, the methylenomycin furans (termed MMF). MMF autoregulator activity is shown to account for the ability of certain Mm non-producing mutants to act as ‘secretors’ in cosynthesis with other ‘convertor’ mutants. Three genes implicated in MMF biosynthesis are flanked by two regulatory genes, which are related to genes for gamma-butyrolactone-binding proteins. Genetic evidence suggests that these two genes encode components of a hetero-oligomeric repressor of MMF and Mm biosynthesis. The Mm biosynthetic genes themselves depend on the activator gene mmyB, which appears to be repressed by the putative MmyR/MmfR complex until enough MMF accumulates to release repression. The presence of TTA codons in mmyB and the main MMF biosynthetic gene causes Mm production to be dependent on the pleiotropically acting bldA gene, which encodes the tRNA for the rarely used UUA codon.

Introduction

Streptomycetes are the richest source of antibiotics and other secondary metabolites used in human and veterinary medicine and agriculture (Hopwood, 2007). An unusual epoxycyclopentenone antibiotic, methylenomycin A (Mm), is one of several antibiotics produced by the model strain Streptomyces coelicolor A3(2), in which the Mm biosynthetic genes are located on the 365 kb linear plasmid SCP1 (Vivian, 1971; Kirby et al., 1975; Wright and Hopwood, 1976; Kirby and Hopwood, 1977; Bentley et al., 2004). Mm is also produced by Streptomyces violaceoruber SANK 95570 (Haneishi et al., 1974), in which the relevant genes are carried on a large circular plasmid, pSV1 (Aguilar and Hopwood, 1982; Yamasaki et al., 2003). The two gene sets are more than 99% identical at the nucleotide level (Chater and Bruton, 1985; Yamasaki et al., 2003; Bentley et al., 2004).

In this paper, we present a molecular genetic analysis of the regulation of Mm production in S. coelicolor. Previous studies using phage-mediated mutational cloning (Chater and Bruton, 1985) and deletion analysis (Fisher et al., 1987) indicated that at least 17.5 kb of SCP1 DNA were needed for the biosynthesis of Mm and its proper regulation. Furthermore, a 1.5 kb XhoI fragment overlapping the right-hand end of the 17.5 kb segment also led to Mm non-production when used to integrate an attP-deleted phage vector into the Mm biosynthesis gene cluster (S. Angel and K.F. Chater, unpublished; Fig. 1), showing that the largest biosynthetic operon encompasses nine genes extending from mmyB to mmyK (Fig. 1). The limits of the Mm biosynthetic cluster were further extended when we found that in-frame deletions of two more genes, mmyY and mmyF, diverging from mmyB, both eliminated Mm production (Fig. 1 and Table 1). The known Mm biosynthetic gene cluster thus consists of 21 genes in about 19 kb of SCP1 DNA. Table S1 relates these 21 genes to likely functions.

Figure 1.

Features of the gene cluster for methylenomycin biosynthesis. Broad functional attributions are based on the results of Chater and Bruton (1985), and the orientation of the cluster in that publication is retained here, although it is the opposite of that given in the later SCP1 sequence, in which the individual genes were identified (Bentley et al., 2004). The two XhoI sites shown in bold flank a restriction fragment shown to be internal to a transcription unit by mutational cloning (see Results). The extent of the deletions of the right and left ends of the cluster in the previously described secretor and convertor mutants 2438 and 2425, respectively, were determined by Southern blotting using the probes indicated and by PCR using primers specific for the underlined genes. The four fragments shown above the restriction map were used in the phage-mediated construction of transcriptional fusions to the xylE reporter gene.

Table 1.  Phenotypes of PCR-targeted mutants (J1506 SCP1+ background).
StrainGene deletionMm productionMMF productionMMF responsive
J1506 +n/an/a
J2629ΔmmyR::aac(3)IV+++n/an/a
J2650ΔmmfP::scar++n/an/a
J2642ΔmmfH::scar+
J2643ΔmmfL::scar+
J2635Δ(mmfP→mmfL)::aac(3)IV+
J2636ΔmmfR::aac(3)IV–/+n/an/a
J2637Δ(mmyR→mmfR)::aac(3)IV+++n/an/a
J2638Δ(mmyR→mmfL)::aac(3)IV+ +
J2639Δ(mmfP→mmfR)::aac(3)IV–/+
J2641ΔmmyR::aac(3)IV,ΔmmfR::scar+++n/an/a
J2644ΔmmyB::scar+
J3994ΔmmyY::aac(3)IV+
J3995ΔmmyF::aac(3)IV+

In S. coelicolor, Mm production is eliminated by mutation of bldA (Chater and Merrick, 1976; Merrick, 1976), one of many chromosomal genes with pleiotropic regulatory effects on secondary metabolism (Bibb, 2005). The bldA gene product is the tRNA for the UUA (leucine) codon (Lawlor et al., 1987; Leskiw et al., 1991a,b: see Chater and Chandra, 2008; for a comprehensive review). The corresponding codon TTA is very rare in Streptomyces genes, which have very high GC content: indeed, no primary metabolism gene of S. coelicolor contains a TTA codon (Bentley et al., 2002; Li et al., 2007), and the deletion of bldA does not impede vegetative growth (Leskiw et al., 1993), but causes major defects in colony differentiation and in the production of several antibiotics (Hopwood, 1967; Merrick, 1976). Pathway-specific regulatory genes actII orf4 and redZ for the blue antibiotic actinorhodin and the red prodiginine complex, respectively, both contain a TTA codon, providing the main route by which bldA mediates its effects on these pathways (Fernandez-Moreno et al., 1991; White and Bibb, 1997; Guthrie et al., 1998).

In this paper, we show that the Mm gene cluster includes three regulatory genes and the determinants for an extracellular signal, and that these all participate in a multi-step, multilevel cascade to activate the biosynthetic genes. This prompted a parallel metabolomics-based study in which we characterized the extracellular signal as a complex of novel furans, collectively termed MMF (methylenomycin furans; Corre et al., 2008). We further demonstrate that two key genes in the regulatory cascade contain TTA codons that are responsible for the bldA dependence of Mm production.

Results

Analysis of the segment of SCP1 required for Mm biosynthesis leads to several regulatory predictions

Earlier functional analyses (Chater and Bruton, 1983; 1985; Fisher et al., 1987) coupled with more recent DNA sequencing (Bentley et al., 2004) indicated that the five genes at the left end of the cluster as shown in Fig. 1 are involved in regulating Mm biosynthesis. The outer two of the five, mmyR and mmfR, encode likely DNA-binding regulatory proteins of the TetR superfamily. They are separated by a probable operon of three genes (mmfL,H,P), the first of which (mmfL) encodes a protein related to AfsA, the key enzyme for the biosynthesis of the gamma-butyrolactone signalling molecule A-factor that is involved in activating streptomycin biosynthesis in Streptomyces griseus (Kato et al., 2007) (Fig. S1). This suggested that MmfL, MmfH and MmfP are the biosynthetic enzymes for an A-factor-like molecule that may regulate Mm production. In a parallel study, it has been shown that the expression of these enzymes leads to the production of a series of novel furans which act as the autoregulator for Mm production, and have collectively been designated MMF (Corre et al., 2008).

MmfR is a paralogue (32% amino acid identity) of the repressor protein ArpA that, in S. griseus, binds A-factor (Fig. S2). A-factor binding causes ArpA to be released from the promoter region of adpA, which encodes a global activator of secondary metabolism (Horinouchi, 2002; Chater and Horinouchi, 2003). It therefore seemed possible that MmfR fulfils a similar role in mediating the effects of MMF. The second regulatory protein, MmyR, is also related to ArpA (32% amino acid identity), but lacks two conserved residues: a proline at residue 115 (the equivalent residue in ArpA influences DNA-binding, probably indirectly: Onaka et al., 1997), and a tryptophan at residue 119 that is required in ArpA for binding of gamma-butyrolactone (Sugiyama et al., 1998) (Fig. S2). To date, two distinct classes of ArpA-like proteins have been recognized on the basis of their pI value and ability to bind autoregulator. Typically, proteins having an acidic nature (pI ∼5) have been shown to bind autoregulator, whereas those having a basic nature (pI ∼10) have not (Choi et al., 2004). Given the acidic nature of MmfR (pI 6.26), it seems likely that it binds autoregulator, while MmyR, which has a pI of 8.02, probably does not. Deletion of most of mmyR resulted in Mm overproduction (Chater and Bruton, 1985; Fisher et al., 1987; Fig. 1), pointing to a role for MmyR in repressing Mm production. The regions of MmfR and MmyR corresponding to the DNA-binding helix–turn–helix of other ArpA-like proteins are particularly highly conserved, suggesting that the DNA sequences recognized by these proteins might be similar.

A central part of the 21-gene cluster comprises the resistance/export gene mmr and a diverging arsR-like regulatory gene, mmyJ, which probably functions to regulate mmr (perhaps providing the means by which mmr is induced in the presence of Mm: Hobbs et al., 1992). This part of the cluster has been studied previously in some detail (Neal and Chater, 1991), and we do not deal with it further here. It is flanked on both sides by converging clusters of mmy genes (mmyTOG and mmyBQEDXCAPK), shown to be operons necessary for Mm biosynthesis by relating their organization to earlier mutational cloning analysis (Fig. 1; Chater and Bruton, 1983; 1985). These two clusters mostly encode proteins related to enzymes of known function in secondary metabolism, and most of the genes have been allocated likely roles in the proposed biosynthetic pathway (Challis and Chater, 2001; Corre and Challis, 2005). Two genes diverging from mmyB (SCP1.228, mmyF, and SCP1.229, mmyY, possibly forming an operon) are also involved in Mm biosynthesis (see Introduction). It seemed likely that the expression of these (at least) three operons of biosynthetic genes would depend in some way on mmyR, mmfR and the mmfL,H,P genes.

Interestingly, the first gene (mmyB) in the largest mmy operon appears to be regulatory, rather than biosynthetic: it encodes a member of the Xre family of transcriptional activators. As a working hypothesis that guided some of the experiments described below, we considered it possible that the activation of mmyB was the main role of the MMF/MmyR/MmfR regulatory system, and that MmyB might directly activate transcription of the mmyTOG and mmyBQEDXCAPK operons, and of mmyY and mmyF.

A Mm non-producing mutant previously defined as a ‘secretor’ in cosynthesis experiments is deleted for nearly all the predicted Mm biosynthetic genes, but retains the putative regulatory segment

Kirby and Hopwood (1977) used cosynthesis tests to identify secretors and convertors in a collection of Mm non-producing mutants. It was thought likely at the time that secretors had a relatively late block in Mm biosynthesis, and produced an intermediate that could be used as the substrate for Mm production by a convertor strain that had a relatively early block. However, the sequence of the cluster suggested an alternative possibility: the convertor might in fact be unable to produce MMF, while secretor strains might retain the ability to produce MMF, but be directly defective in Mm biosynthesis. We analysed DNA from the secretor mutant 2438 and the convertor mutant 2425, using PCR, with primers for nine of the genes in the cluster (Table S2), and Southern blotting of XhoI- and PvuI-digested DNA (Fig. 1). It turned out that both strains had large deletions. The secretor had lost nearly all the genes of the Mm biosynthetic operons and the resistance segment, but retained the left end of the cluster (mmyR, mmfLHP, mmfR and mmyT); while the convertor had lost DNA from the left end (mmyR, mmfLHP), but retained intact Mm biosynthetic operons, as well as mmfR (Fig. 1). These results suggested that the material produced by the secretor strain was not a precursor of Mm, but was MMF, the putative signalling molecule; and that the synthesis of MMF from readily available metabolites required no more than the mmfLHP genes (possibly with the addition of mmyT, which also remained in the secretor strain).

Genetic analysis of the roles of the mmfLHP genes in MMF production

To examine the roles in MMF production of the mmyR, mmfLHP, mmfR and mmyT genes remaining in the secretor strain 2438, the genes were introduced, individually or in various combinations, into an SCP1-free strain, J1501, using integrating vectors possessing the int-att region of bacteriophage φC31 (Bierman et al., 1992). The abilities of the various resulting strains to act as secretors to the convertor strain 2425 were tested (Fig. 2). J1501 derivatives carrying the six leftmost genes in the Mm gene cluster (plasmids pIJ6580, 6581) acted as secretors, and three of these six genes (mmyR, mmfR and mmyT) could be eliminated without affecting secretor ability (plasmids pIJ6582, 6584), proving that the presence of just the mmfLHP genes was indeed sufficient to lead to production of MMF, confirming and extending the conclusions of the molecular analysis of 2438 described above.

Figure 2.

The ability of strains carrying segments of the mmf region to produce MMF. J2621, an SCP1 strain carrying mmfLHP on the pSET152 integrating vector, had no effect by itself on the Mm-sensitive indicator strain (A), but could elicit Mm production by the convertor strain 2425 (B). J2622, which resembled J2621 but contained only mmfL, did not elicit Mm production by 2425 (C, D). (E) MMF production resulting from the introduction of various DNA segments using the pSET152 vector. MMF production was assessed by the cosynthesis assay exemplified in A–D. (+) MMF production; (–) no MMF production.

As the S. griseus afsA gene is sufficient to permit the synthesis of a gamma-butyrolactone from primary metabolites (Kato et al., 2007), it was possible that the afsA-like gene mmfL might by itself permit the production of a compound capable of stimulating Mm production by the convertor strain. However, pIJ6566, in which mmfL was the only intact inserted gene, did not confer secretor activity. This indicated that mmfH and/or mmfP were also needed for MMF to be produced in biologically significant amounts. To analyse the roles of the mmfLHP genes further, we used PCR-targeted mutagenesis to eliminate them, either individually or all together, from the otherwise intact Mm gene cluster located on SCP1 in J1506 (Table 1). Where it was necessary, unmarked in-frame deletions were made to avoid polarity effects. The deletion of the three genes together (in J2635) caused the loss of Mm production, giving a strain that acted as a convertor in cosynthesis tests with strain 2438 (Fig. 3). This was the phenotype expected of a strain specifically unable to make MMF. The separate deletion of mmfL (J2643) or mmfH (J2642) had the same phenotypic effect as the triple deletion (Table 1), consistent with the evidence above (Fig. 2) that each of these two genes played an essential role in the production of bioactive MMF. However, the mmfP mutant (J2650) produced Mm, at levels somewhat higher than those of J1506. We further investigated the role of mmyP by making an in-frame deletion of mmfP from plasmid pIJ6581, which contained the mmyR, mmfLHP, mmfR and mmyT genes (see above), and introduced the mmfP-deleted plasmid (pIJ10215) into the SCP1-free strain J1501. In repeat experiments, the resulting strain showed little or no secretor activity. To reconcile the contrasting results on the role of mmfP obtained by Mm production and cosynthesis tests, we suggest that, in those cells expressing mmfLH but not mmfP, the action of some MmfP-like activity from another gene permits MMF to accumulate intracellularly at a level high enough to interact with the MMF receptor in the same cytoplasm and thereby activate Mm production in those cells; but that the level of MMF diffusing to the extracellular environment is insufficient to stimulate Mm production by the whole population. Alternatively, the substrate of MmfP may be able to bind to MMF receptor, but be less readily accumulated in the extracellular environment.

Figure 3.

Deletion of mmfLHP from J1506 results in the ‘convertor’ phenotype. J2635 is a J1506 derivative with a constructed deletion of mmfLHP from the resident copy of SCP1. Strain 2438 is a secretor strain. A pigment-free zone, indicating Mm production, was seen only when 2438 and J2635 were grown close together (B). No interaction was seen when J2635 was grown on its own (A) or close to the convertor strain 2425 (C).

Using the convertor strain 2425 as indicator, we found that MMF was extractable with ethyl acetate from both surface and liquid complete medium (CM) cultures of the secretor strain 2438 after incubation for 48, 72 or 96 h, although production was not detectable in extracts from 24 h cultures (Fig. S3A–D). Further tests (Fig. S3E–H) showed that MMF, like known gamma-butyrolactones, was resistant to heat, protease and acid treatment. However, unlike most gamma-butyrolactones, MMF was not sensitive to alkali treatment. This indicated that MMF might not have a typical gamma-butyrolactone structure, an inference confirmed by the demonstration reported elsewhere that MMF comprises a series of 4-alkyl 2-hydroxymethylfuran-3-carboxylic acids (Corre et al., 2008).

Biosynthesis of MMF is independent of chromosomal genes for the biosynthesis of γ-butyrolactones

Streptomyces coelicolor A3(2) and its derivatives have previously been shown to make gamma-butyrolactone autoregulators, including SCB1 (Anisova et al., 1984; Efremenkova et al., 1985; Kawabuchi et al., 1997; Takano et al., 2000), and the key genes for production, scbA and scbR, have been identified on the chromosome (Takano et al., 2001). Inactivation of either of these two adjacent genes eliminated detectable production of gamma-butyrolactones by the SCP1-free strain M145 (Takano et al., 2001). In order to find out whether these genes play any significant part in the biosynthesis of MMF, we introduced pIJ6580 (carrying the mmyR, mmfLHP, mmfR and mmyT genes) into M145 and its scbA and scbR mutants (M571 and M572 respectively: Takano et al., 2001). All three strains (J2626, J2627 and J2628) acted as secretor strains with convertor strain 2425 (not shown). Thus, MMF biosynthesis from primary metabolites appears to be independent of known chromosomal genes for gamma-butyrolactone biosynthesis. We also analysed the same strains to find out whether the genes for MMF production could restore SCB1 production to the scbA and scbR mutants, by spotting ethyl acetate extracts from 72 h CM cultures onto lawns of M145, and scoring for the halo of red pigment production caused by the presence of SCB1 (Takano et al., 2000). The presence of pIJ6580 did not restore SCB1 production to the mutants (not shown), further demonstrating the independence of the two pathways.

The MMF signal is transduced by the MmyR and MmfR proteins

Clearly, MMF accumulation provides a signal for the onset of Mm production. To find out whether the signal transduction route leading from MMF to Mm production involved MmyR and/or MmfR, we first examined the phenotypes of mmyR and mmfR null mutants generated by PCR targeting of the autonomous SCP1 present in strain J1506 (Table 1). In confirmation of earlier studies (Chater and Bruton, 1985; Fisher et al., 1987), deletion of mmyR (in J2629) resulted in marked overproduction of Mm, consistent with a role of MmyR in repression of Mm biosynthesis, and making it unlikely that it is an activator (unless it activates an as yet unidentified repressor). Deletion of mmfR (in J2636) had the opposite effect, causing substantial loss of production. Hence, MmfR plays a role either directly or indirectly in the activation of Mm biosynthesis. A direct activator role was ruled out, because strain J2641, in which mmyR and mmfR were both deleted, showed Mm overproduction.

To clarify the roles of MmfR and MmyR further, we deleted these genes together with the mmfLHP genes, so that any effects of MMF were eliminated. The presence of mmfR (J2639), mmyR (J2638) or both genes (J2635) in the absence of mmfLHP resulted in greatly reduced Mm production, indicating that MmfR and MmyR can each act as a repressor (and be able to bind DNA) in the absence of the other. However, Mm production in an mmfR-containing strain lacking the mmfLHP genes could be induced by MMF supplied by the secretor strain 2438, whereas the repression seen in a strain carrying only mmyR could not be relieved by growth close to the secretor strain (compare J2639 with J2638). Thus, sensitivity to MMF appears to be supplied by MmfR. The simplest interpretation of these observations is that MmyR and MmfR, two paralogous proteins, may together form a hetero-oligomeric repressing complex in which MmyR provides strong DNA binding, and MmfR binds the inducing ligand, MMF, as well as binding to DNA.

In some systems, mmfL-like genes, and hence autoregulator biosynthesis, are themselves subject to regulation by the autoregulator binding protein (for example, in the case of the scbA–scbR system of S. coelicolor: Takano et al., 2001). We therefore also introduced the same mmyR and mmfR null mutations into plasmid pIJ6581, which contained only the mmyR to mmyT segment of SCP1. These deleted plasmids were conjugated into the SCP1-free strain J1501, and the cosynthesis assay was used to evaluate the effects of the mutations on MMF biosynthesis in the absence of Mm production (Fig. 4). The strain carrying the mmfR-deleted plasmid was ineffective in cosynthesis, suggesting that MmfR is necessary for efficient MMF biosynthesis in the presence of MmyR, while the mmyR-deleted plasmid gave rise to abundant MMF production. As earlier experiments had shown that a strain carrying only the mmfLHP operon, without either mmyR or mmfR, was a good secretor of MMF (Fig. 2, pIJ6584), we conclude that an MMF-sensitive complex of MmfR and MmyR represses MMF production as well as production of Mm.

Figure 4.

Phenotypes of J1501 carrying PCR-targeted deletion derivatives of pIJ6581. Deletion of mmyR (in J2645) resulted in hyper-elicitation of Mm production by the convertor strain 2425 (A, B), whereas deletion of mmfR (in J2649) resulted in loss of elicitation of Mm production by 2425 (C, D). (E) Summary of effects on MMF production of deleting particular genes from pIJ6581. Blanks in place of genes indicate replacement by the ‘scar’ sequence. MMF production was assessed by comparison with the strain carrying pIJ6581, using the cosynthesis assay exemplified in A–D. (++) MMF overproduction; (+) MMF production; (–) no MMF production.

Evidence that mmyB encodes a transcriptional activator of Mm biosynthesis

Many clusters of antibiotic biosynthetic genes are dependent on pathway-specific transcriptional activators encoded by genes within the clusters. There are two apparent regulatory genes other than mmyR and mmfR in the mmy cluster. One of these, mmyJ, is probably involved in the regulation of the resistance gene mmr (Neal and Chater, 1991; Hobbs et al., 1992), and is not further investigated here, though we do not rule out the possibility that it could influence Mm production as well as resistance. The other, mmyB, encodes a protein with a likely DNA-binding motif related to that of the Xre family of transcriptional regulators. The in-frame deletion of mmyB from J1506 caused the loss of Mm production, though the mutant retained the ability to produce MMF as judged by cosynthesis (Fig. 5). We therefore propose that MmyB is a pathway-specific activator of Mm biosynthesis, and that mmyB may be the direct target for repression of Mm biosynthesis by the putative MmyR/MmfR complex.

Figure 5.

MmyB is required for the production of Mm but not for that of MMF. J2640 (ΔmmyB::aac(3)IV) did not produce Mm (A), but elicited Mm production by the converter mutant 2425 (B).

Analysis of transcription of genes involved in the production and sensing of M-factor

Our regulatory model requires that there should be recognition sequences for the MmyR/MmfR complex associated with promoters upstream of mmfL and mmyB, and that potential target genes in the regulatory cascade should not be expressed before the genes implicated in regulating them. We therefore examined the transcription of the regulatory genes and of some representative Mm biosynthetic genes, using S1 nuclease protection analysis of RNA isolated from J1506 liquid cultures during growth and Mm production. These experiments, although mostly not at the nucleotide level of resolution, allowed the approximate localization of transcription start points (tsp).

In these experiments, Mm production showed a clear association with stationary phase (Fig. 6), as observed by Hobbs et al. (1992). On entry into stationary phase, a first Mm-related product was detectable by thin-layer chromatography (TLC). It was presumed to correspond to desepoxy-4,5-didehydromethylenomycin A (dMm, also known as methylenomycin C) (Hornemann and Hopwood, 1978; Corre and Challis, 2005). Mm itself appeared after a slight lag. The eventual disappearance of dMm as Mm accumulated was consistent with previous suggestions that dMm was a precursor of Mm (Hornemann and Hopwood, 1978). It was noticeable that dMm/Mm production immediately followed a sharp drop in medium pH, which was previously found to induce Mm production (Hayes et al., 1997).

Figure 6.

Growth characteristics and methylenomycin (Mm) production of J1506 (SCP1+). Cultures of J1506 (50 ml, liquid CM) were harvested at the indicated time points and evaluated for biomass (dry cell weight) and pH (upper panel) and Mm production (by TLC; lower panel). A second spot, apparently corresponding to the Mm precursor desepoxy-4,5-didehydromethylenomycin A (dMm, methylenomycin C) (Hornemann and Hopwood, 1978, Corre and Challis, 2005), accumulated slightly earlier than Mm.

All the transcripts studied had a fairly similar profile in relation to growth and Mm production (with some subtle differences that we discuss below) (Fig. 7): they all showed a very sharp increase in abundance at 48 h, the time at which Mm production was first detectable, indicating that transcriptional regulation plays a major role in the onset of both MMF and Mm biosynthesis, and that there is a fairly rapid transition from the upshifted expression of the MMF biosynthetic genes to the expression of the Mm biosynthetic genes. The mmyR, mmfL and mmfR transcripts were also detectable at the first three time points, although at low levels compared with later time points, and mmyB transcript could be detected at a low level at 36 h. In contrast, the transcripts of the biosynthetic genes were not detected before 48 h. Thus, the pattern of mRNA accumulation was consistent with the regulatory model.

Figure 7.

S1 nuclease protection analysis of transcription of regulatory and structural genes for Mm and MMF biosynthesis. RNA samples were isolated from cultures grown in liquid CM for the stated times. Arrows above the panels indicated time points at which desepoxy-4,5-didehydromethylenomycin A (dMm, methylenomycin C) and Mm could be detected by TLC (see Fig. 6). The approximate lengths of protected species were estimated by comparison with ΦX174 DNA/HinfI Dephosphorylated Markers (Promega). All probes included a non-homologous sequence at the 3′-end to allow discrimination between probe-probe reannealing and FLP (full-length protection indicating that the probe is internal to a transcript).

The 5′-ends of transcripts could be detected upstream of mmyR, mmfL, mmfR, mmyT and mmyB, but only full-length probe protection was seen for mmyA and mmyQ, indicating that they were transcribed as internal parts of longer transcription units – a conclusion consistent with sequence-informed interpretation of genetic data from mutational cloning, implying the existence of a polycistronic mmyBQEDXCAPK mRNA (Chater and Bruton, 1983; 1985; see above). Here we briefly summarize basic features of the transcription initiation regions that were revealed.

Transcription from the divergent mmfL, mmfR promoter region.  This appears to be complex. The two mmfR tsps were separated by about 27 bp, with the single mmfL tsp apparently being located between them. Even though the 5′-end positions are approximate, either there is a short overlap, possibly of about 15 nt, between the longer mmfR transcript and the mmfL transcript, or at the very least there is considerable interplay between leftward and rightward transcription in this complex promoter region. All the transcripts from this region have substantial untranslated lengths at their 5′-ends (between c. 70 and c. 120 nt).

Transcription of mmyR. It appeared that mmyR is transcribed both from its own promoter and by readthrough from a remote upstream promoter (conceivably the mmfL promoter, because the time-course of transcription of the readthrough mmyR transcript was very similar to that of the mmfL transcript). The mmyR-specific transcript has an untranslated leader sequence of about 47 nt.

Transcription of mmyT. The single tsp for mmyT was located about 105 bp downstream of the mmfR stop codon, and 119 bp upstream of the mmyT start codon. There was no evidence of readthrough transcription from mmfR.

Transcription of mmyB. This appeared particularly complex, both in terms of possible tsps and in relation to the presence of three in-frame potential start codons. To clarify this we used DNA sequence ladders as size standards in S1 mapping of the mmyB promoter region to permit the precise allocation of tsps (Fig. 8). The most abundant mRNA species (tsp2) initiated 17 bp downstream of the start codon allocated by Bentley et al. (2004), 1 bp upstream of a possible GTG start codon, and 19 bp upstream of another. The least abundant species (tsp1) initiated 20 bp upstream of tsp2, and might permit the use of the annotated start codon, with a leader of 2 nt (leaderless or near-leaderless transcripts are not uncommon in Streptomyces; Strohl, 1992): alternatively, this species might also be translated from either of the GTG codons suggested for the tsp2 mRNA. The 3′-end of another relatively low-abundance species was 15 nt downstream of tsp2, and therefore possessed only the most downstream of the three possible start codons. We do not consider this mRNA species further because it could well be a processing product from one of the larger species.

Figure 8.

High resolution S1 mapping of mmyB mRNA. Accurate transcription start points (tsp) were determined by comparison with a sequence ladder generated using the probe used for S1 mapping of mmyB. For clarity, the sequence ladder is read as the plus-strand sequence. The positions of tsp are indicated to the left of the autoradiograph, and presumptive translation start codons are shown to the right. The sequence of the mmyB promoter and the interpretation of the transcription mapping are shown in the lower part of the figure. Arrows indicate tsp positions, and dotted underlining indicates likely translation start codons. A potential MARE sequence is boxed, and B-boxes are underlined (see Fig. 9 for further details of MARE sequences).

Analysis of these tsp regions for conserved sequences (using RSAT consensus: http://rsat.ulb.ac.be/) identified a completely conserved 9 bp sequence (CGGGAAGGT) in the upstream regions of mmyR and mmyB and within the mmfLmmfR intergenic region (Figs 8 and 9). This core sequence was embedded in an 18 bp palindrome (consensus ATACCTTCNNGAAGGTAT) that showed 14/18 identities in each pairwise comparison. A sequence somewhat similar to the core sequence was also detected in the mmyT promoter region, corresponding to just one half of the palindrome (Fig. 9). The core sequence was similar to the CYNNNCGGGT sequence for recognition by a small selection of gamma-butyrolactone binding proteins (Folcher et al., 2001). Such elements were called ‘AREs’ (autoregulator-responsive elements), so we here term these putative MMF-autoregulator responsive elements ‘MAREs’. As with AREs from other Streptomyces species, and in general agreement with the expectations of the regulatory model, the MAREs are located in positions in which they might be expected to interfere with the initiation of transcription of the gene to be regulated (Figs 8 and 9). Thus, the mmfL, mmfR MARE appears to overlap the mmfL tsp, and to be centred 13 bp downstream of mmfR tsp1, and 14 bp upstream of mmfR tsp2; the mmyR MARE is centred 11 bp upstream of the tsp; the mmyB MARE is centred 21 bp upstream of the weak tsp1, and 42 bp upstream of the very strongly used tsp2; and the mmyT possible MARE is centred 27 bp upstream of the tsp.

Figure 9.

Candidate sequences (MAREs) for recognition by MMF-sensing regulatory proteins. Sequences upstream of mmyR, and in the mmfL–R and mmyB–Y bidirectional promoter regions, are aligned with the proposed ARE consensus sequence proposed by Folcher et al. (2001), which is shown at the top of the figure. The consensus core sequence identified using RSAT is boxed, and additional conserved nucleotides are underlined. An 18 bp palindromic sequence derived from the alignment is shown. A sequence related to one half of the palindrome was also found in the mmyT promoter region, and is shown at the bottom of the figure.

In searching for repeated sequences, we also found two copies of an almost perfect, directly repeated 16 bp sequence (CA CCA/G G/AGG CCC CGC CG) centred 131 bp and 101 bp upstream of mmyB tsp2 (Fig. 8). A further 9/10 bp match to part of this 16 bp sequence was present on the other strand, somewhat closer to the diverging mmyY gene (CTG AGG CCC C). When these sequences and their complementary sequences were aligned, a consensus sequence showing considerable inverted symmetry was revealed (CGCCGGGGCCCCGXXG). Thus, in broad agreement with the regulatory model, the regions around the transcription start sites of likely MmyB-activated genes do contain related sequences that are not found in other promoters. On the tentative proposition that these sequences are binding sites for MmyB, we term them ‘B-boxes’, but we emphasize that further work will be needed to find out whether any of the sequences do in fact bind MmyB.

A bldA mutant is interrupted at two points in the regulatory cascade leading to Mm biosynthesis

Transcriptional fusions of the xylE reporter gene (Ingram et al., 1989) to the mmfLHP, mmyTOG and mmyBQEDXCAPK operons in J1507 gave rise to readily detected catechol oxygenase activity, which rose steeply in every case upon entry into stationary phase (Fig. 10A). This corresponded closely with the results of direct analysis of mRNA (see above). However, catechol oxygenase activity from these fusions was abolished in a bldA mutant, as shown both in plate tests (Fig. 10B) and by quantitative assays of liquid CM cultures (less than 0.05 mU mg−1 protein in samples taken at 48 h). This showed that bldA is required for transcription of all of these genes, and suggested that bldA might affect the pathway-specific genes involved in regulating production. Indeed, the cluster contains two TTA codons, one at codon 8 of mmfL, and the other at the annotated codon 23 of mmyB (the latter would also be translated from the shorter mRNAs initiated at tsp2 and ‘tsp3’ of mmyB– see above) (Bentley et al., 2004). If the effects of bldA on Mm production were mediated directly through these codons, then changing the codons to alternative leucine codons might restore the ability to produce Mm to a bldA mutant.

Figure 10.

Expression of xylE transcriptional fusions to genes involved in Mm biosynthesis is abolished in a bldA mutant.
A. Time-courses of catechol oxygenase activity in liquid CM cultures of J1507 carrying transcriptional fusions of xylE to mmyG, mmyK and mmfH.
B. Effect of bldA mutation on xylE expression from mmyG::xylE, mmyK::xylE, mmyD::xylE and mmfH::xylE transcriptional fusions. The cultures were both streaked and spotted on CM, then incubated for 4 days and sprayed with catechol. Catechol oxygenase activity encoded by the xylE reporter gene was visualized by the development of a yellow pigment.

To test this, we used two strategies, both of which gave the same results. Each TTA codon was changed to a CTC codon, and the altered genes (including promoter sequences, and part of the gene immediately downstream) were inserted into either an integrative vector (pSET152) that introduced the codon change into the S. coelicolor chromosome by site-specific recombination at a phage attachment site, or suicide vectors (pSET151 and pKC1132) that can be rescued by integration into S. coelicolor by homologous recombination involving sequences cloned into the vector. The latter route allowed allele replacement of the TTA-containing genes by CTC-containing versions. The Streptomyces host was strain J1703, a bldA mutant containing SCP1 stably integrated into its chromosome. Derivatives in which either one, or both, of the TTA codons had been converted to CTC were tested for Mm production and, as appropriate, for cosynthesis with the secretor and convertor mutants 2438 and 2425 (Table 2).

Table 2.  Results of codon conversion experiments in bldA mutant (J1703) background.
StrainGenotypeMm productionMMF productionMMF responsive
J2600mmfLTTA/mmfLTTA pSET151
J2601mmfLTTA/mmfLTTA pSET152
J2602mmfLTTA/mmfLCTC pSET151+
J2603mmfLTTA/mmfLCTC pSET152+
J2604mmfLCTC+
J2605mmyBTTA/mmyBTTA pKC1132
J2606mmyBTTA/mmyBTTA pSET152
J2607mmyBTTA/mmyBCTC pKC1132+
J2608mmyBTTA/mmyBCTC pSET152+
J2609mmyBCTC+
J2610mmfLCTC, mmyBTTA/mmyBTTA pKC1132+
J2611mmfLCTC, mmyBTTA/mmyBTTA pSET152+
J2612mmfLCTC, mmyBTTA/mmyBCTC pKC1132+n/an/a
J2613mmfLCTC, mmyBTTA/mmyBCTC pSET152+n/an/a
J2614mmfLCTC, mmyBCTC+n/an/a

The double codon conversion restored Mm production, confirming that no other gene was interpolated between bldA and the pathway-specific genes for Mm biosynthesis (J2612, J2613, J2614). Moreover, although neither single codon conversion restored production, the mmyB codon conversion mutants were able to produce Mm when grown next to the secretor mutant 2438 (J2607, J2608, J2609), and the mmfL codon conversion mutants produced MMF, as judged by their ability to induce Mm production by the convertor strain 2425 (J2602, J2603, J2604). The strains listed in Table 2 include derivatives heterozygous for the codon conversion (J2602, J2603, J2607, J2608, J2612 and J2613). In these strains, as expected, the CTC version of the gene was dominant in the phenotypic tests.

In relation to the regulatory model therefore these results showed that the TTA codon in mmfL is the direct cause of the inability of a bldA mutant to make MMF, and confirmed that the absence of MMF synthesis prevents the production of Mm. The ability of the mmyB codon-changed version to act on the cluster in trans was consistent with the positively acting role proposed for it in the switching on of the Mm biosynthetic genes. Moreover, the ability of each of the two codon-converted versions of the genes to act in trans suggested that the presence of an untranslated UUA codon near the start of the mmfLHPR and mmyBQEDXCAPK transcripts did not have severe polarity effects on the expression of downstream genes.

Discussion

A likely evolutionary conjunction of two quite distinct, separately regulated biosynthetic pathways to form the present-day gene set for Mm biosynthesis

The gene cluster for Mm biosynthesis can be considered to have two parts. The five leftmost genes include three that are needed for the biosynthesis and release from the cell of ethyl acetate-soluble material, MMF, which was detected by its ability to activate Mm biosynthesis. MMF has some features in common with gamma-butyrolactone autoregulators found in many streptomycetes, but differs in being alkali-stable, suggesting some unusual feature in its structure – an inference that has been verified by our recent detailed chemical characterization of it as a series of structurally novel compounds, the methylenomycin furans (Corre et al., 2008). Recently, Bunet et al. (2008) have found that a somewhat gamma-butyrolactone-like autoregulator of alpomycin biosynthesis in Streptomyces ambofaciens is also alkali-stable, raising the possibility that MMF-like molecules may not be unusual.

Although our analysis indicated that the synthesis of biologically detectable levels of MMF required mmfL, mmfH and mmfP,Corre et al. (2008) found by more sensitive LC-MS analysis that all five MMF molecular species could be detected in a strain of S. coelicolor containing just mmfL in the absence of the rest of the MMF-Mm gene cluster. Corre et al. (2008) suggested that the functions of MmfH and MmfP can be carried out at low efficiency by paralogous enzymes encoded by the chromosome of S. coelicolor, and identified candidate genes (actVA and SCO3558 respectively).

The three MMF biosynthetic genes are flanked by two regulatory genes, mmyR and mmfR, whose protein products both resemble various gamma-butyrolactone-sensitive repressors. These two genes are involved in the autoinduction of MMF biosynthesis. The much larger right-hand part of the cluster is mainly occupied by three (or possibly four) operons encoding 13 enzymes of Mm biosynthesis. Thus, the converging mmyTOG and mmyBQEDXCAPK (mmyB is a regulatory gene) operons are separated by two genes involved in Mm resistance/Mm export (mmr, encoding an integral membrane transporter, and mmyJ, encoding an ArsR-like protein believed to regulate mmr in response to the accumulation of Mm; Neal and Chater, 1987; Hobbs et al., 1992); while, to the right of mmyB, and diverging from it, two further genes (mmyY, mmyF) are also involved in Mm biosynthesis (here we consider mmyY and mmyF as a likely operon, because they are in the same orientation and are separated by only 30 bp; but their operon status remains to be experimentally demonstrated).

The parts of the cluster responsible for the biosynthesis of MMF, on one hand, and of Mm on the other, each contain a single TTA codon, which makes the production of both metabolites dependent on bldA.Chandra and Chater (2008) found that most gene clusters for antibiotic biosynthesis in streptomycetes contain at least one TTA codon, and that this is often in a pathway-specific regulatory gene. We think it likely that the two parts of the Mm cluster originally evolved independently, each being bldA-dependent, and came together through horizontal gene transfer aided by their presence on plasmids [Chater and Kinashi (2007) discussed the possible role of plasmids in bringing about the linkage of independent secondary metabolic gene sets with synergistic effects]. It seems that the only biochemical connection between the two parts is the ability of the presumed MmyR/MmfR complex to repress transcription of genes in the other part of the cluster. Thus, on this model, the evolution of the mmyB promoter region to contain a binding site for the repressor complex was all that was needed to connect the two parts biochemically. It is possible that, before the establishment of the connection between the mmf and mmy clusters, Mm production was under the simpler pathway-specific control of MmyB alone.

Regulation of MMF biosynthesis

Early in growth there is low-level transcription of three transcription units: mmfLHPmmyR, potentially allowing the synthesis of low levels of MMF and MmyR; mmfR; and mmyR. Overall, this level of transcription may provide enough MmyR/MmfR repressor complex to feedback-limit expression of the mmfLHPmmyR, mmyR and mmfR promoters, possibly by binding to the MAREs upstream of these transcription units. In response to the nutritional limitation that signals the advent of stationary phase, the level of bldA tRNA increases in some cells (Leskiw et al., 1993; Trepanier et al., 1997), which would permit relatively free translation of the UUA codon in mmfL mRNA, and the intracellular level of MMF would consequently increase sufficiently to bind to the repressor complex, probably via the MmfR component, and derepress the autoregulatory circuit within those cells. The resulting rapid extracellular accumulation of MMF then causes all physiologically competent cells in the population to become derepressed for MMF biosynthesis (in morphologically differentiated colonies, it is possible that some hyphal cell types may be debarred from production).

Role of MmyB in regulating the structural genes for Mm biosynthesis

The regulation of at least two, and possibly all three, of the biosynthetic operons for Mm biosynthesis is apparently through the action of MmyB as a transcriptional activator. Bioinformatic analysis indicates that MmyB contains an N-terminally located DNA-binding domain of the Xre helix–turn–helix type. Related proteins, which include the bacteriophage λ CI protein, often act as transcriptional activators, by contact with the σ component of RNA polymerase holoenzyme (for a recent example, see McGeehan et al., 2006). Interestingly, mmyB appears to be the first gene in an operon of otherwise biosynthetic genes: no tsp could be detected between mmyB and the next gene, mmyQ, from which it is separated by 86 bp; and transcription from the mmyB promoters is comparable in kinetics to transcription into mmyQ and mmyA, which are in the same operon (Fig. 7). In addition, a DNA fragment extending from a XhoI site in mmyB to another in mmyE eliminated Mm production when used to integrate a phage vector in a mutational cloning experiment (S. Angel and K.F. Chater, unpublished), providing genetic evidence that the fragment fell entirely within a transcription unit.

The 231 bp of DNA separating the mmyB and mmyY coding sequences contains two near-identical direct repeats of a 16 bp sequence that may well be a binding site for MmyB. A further copy of 9/10 bp of this ‘B-box’ sequence is also present closer to mmyY, but on the opposite strand. Interestingly, a segment (CCAGGG) with similarity to one half of the palindromic sequence, notably showing a 6/6 match with one of the boxes upstream of mmyB, was present just upstream of the (approximate) mmyT transcription start site. This sequence is thus found near the promoters of two, and possibly all three, of the Mm biosynthetic operons, but in different positions in relation to the likely −35 and −10 sequences. Further biochemical study will be needed to confirm that mmyB is autoregulatory, and that MmyB can activate transcription using B-boxes in different configurations in transcription initiation complexes.

The initial upregulation of mmyB expression requires release from repression by MmyR/MmfR complexes bound to the mmyB promoter (probably at its MARE sequence). This is presumably achieved as the extracellular, and hence the intracellular, concentration of MMF reaches a critical point. As one consequence of this, mmyB transcript will accumulate synchronously in all physiologically prepared cells in the population. If these cells meet a further physiological condition – the availability of sufficient levels of charged bldA tRNA – MmyB will be made, and any limitations on the transcription, translation or abundance of the downstream part of the mRNA caused by non-translation of the UUA codon will be lifted to permit rapid accumulation of the early biosynthetic enzymes, further accelerated by the increasing abundance of the autoactivating MmyB. This will lead to the strong activation of the mmyYF and (possibly) mmyTOG operons, permitting the later steps in Mm biosynthesis to be completed.

This multiply self-reinforcing model is based on genetic studies, transcription analysis and interpretation of DNA sequence. An important next stage will be the in vitro reconstruction of the multi-component interactions predicted to take place at the promoters believed to be the targets of the regulatory elements. We note that our model does not explain why Mm production can be activated by a rapid drop in pH (Hayes et al., 1997).

Experimental procedures

Bacterial strains and culture conditions

Strains derived from S. coelicolor A3(2) are listed in Table 3. The general conditions for maintenance and growth of these strains, and the media CM, YEME (Yeast extract, Malt extract medium), TSB (tryptone soy broth), SMSS (supplemented minimal medium) and DNA (Difco Nutrient Agar), were as in Kieser et al. (2000). For work done in Escherichia coli, maintenance and growth on medium Luria–Bertani were as in Sambrook and Russell (2001). The general hosts for cloning in E. coli were DH5α (Sambrook and Russell, 2001) and XL1-Blue (Stratagene).

Table 3. Streptomyces strains used in this work.
StrainPrecursorTransformed withRelevant characteristics
J1501  SCP1-, SCP2-hisA1 uraA str1 pgl-1
J1506  SCP1+, SCP2-hisA1 uraA str1 pgl-1
J1507  SCP1NF., SCP2-hisA1 uraA str1 pgl-1
J1703  SCP1NF., SCP2-hisA1 uraA str1 pgl-1
2425  SCP1, SCP2-hisD3 cysB6 strA1
2438  SCP1, SCP2-hisD3 cysB6 strA1
M145  SCP1-, SCP2-
M571  SCP1-, SCP2-ΔscbA
M572  SCP1-, SCP2-ΔscbR
J2600J1703pIJ6565mmfLTTA. in pSET151
J2601J1703pIJ6566mmfLTTA. in pSET152
J2602J1703pIJ6567mmfLCTC. in pSET151
J2603J1703pIJ6568mmfLCTC. in pSET152
J2604J2602 mmfLCTC.
J2605J1703pIJ6572mmyBTTA. in pKC1132
J2606J1703pIJ6573mmyBTTA. in pSET152
J2607J1703pIJ6574mmyBCTC. in pKC1132
J2608J1703pIJ6575mmyBCTC. in pSET152
J2609J2607 mmyBCTC.
J2610J2604pIJ6572mmyBTTA. in pKC1132
J2611J2604pIJ6573mmyBTTA. in pSET152
J2612J2604pIJ6574mmyBCTC. in pKC1132
J2613J2604pIJ6575mmyBCTC. in pSET152
J2614J2612 mmfLCTC. and mmyBCTC.
J2617J1501pIJ6580SCP1-251mmyT
J2618J1501pIJ6581mmyRmmyT
J2619J1501pIJ6582mmfPmmyT
J2620J1501pIJ6583J2618 Δ(mmfP-mmfL)
J2621J1501pIJ6584mmfPmmfL
J2622J1501pIJ6566mmfL
J2626M145pIJ6580SCP1-251mmyT
J2627M571pIJ6580SCP1-251 →mmyT
J2628M572pIJ6580SCP1-251mmyT
J2629J1506pIJ6586ΔmmyR:: aac(3)IV
J2635J1506pIJ6592Δ(mmfPmmfL)::aac(3)IV
J2636J1506pIJ6593ΔmmfR::aac(3)IV
J2637J1506pIJ6594Δ(mmyRmmfR)::aac(3)IV
J2638J1506pIJ6595Δ(mmyRmmfL)::aac(3)IV
J2639J1506pIJ6596Δ(mmfPmmfR)::aac(3)IV
J2641J1506pIJ6598ΔmmyR::aac(3)IV, ΔmmfR::scar
J2642J1506pIJ10202ΔmmfH::scar
J2643J1506pIJ10203ΔmmfL::scar
J2644J1506pIJ10208ΔmmyB::scar
J2645J1501pIJ10214ΔmmyR::scar
J2646J1501pIJ10215ΔmmfP::scar
J2647J1501pIJ10216ΔmmfL::scar
J2648J1501pIJ10217Δ(mmfPmmfL)::scar
J2649J1501pIJ10218ΔmmfR::scar
J2650J1506pIJ10201ΔmmfP::scar
J3994J1506pIJ10243ΔmmyY::aac(3)IV
J3995J1506pIJ10244ΔmmyF::aac(3)IV
J1703::KC134J1703KC134BglII/SstI fragment from pIJ519
J1507::KC134J1507KC134BglII/SstI fragment from pIJ519
J1703::KC135J1703KC135C2.18 fragment in KC861
J1507::KC135J1507KC135C2.18 fragment in KC861
J1703::KC140J1703KC140A4.2 fragment in KC861
J1507::KC140J1507KC140A4.2 fragment in KC861
J1703::KC142J1703KC142A3.13 fragment in KC861
J1507::KC142J1507KC142A3.13 fragment in KC861

Plasmids and phages

Cosmid 73, which contains a DNA insert of SCP1 DNA that includes the entire Mm biosynthetic gene cluster, was used for PCR targeted mutagenesis to generate specific mutations in the gene cluster present in S. coelicolor strain J1506 (autonomous SCP1). The conjugative vectors pSET151, pKC1132 and pSET152 were used to introduce the codon conversions of mmfL and mmyB into S. coelicolor (Bierman et al., 1992). The attP-deleted phiC31 phage vector KC860 was used for mutational cloning and for constructing transcriptional fusions of the xylE reporter gene (Bruton et al., 1991). For routine subcloning, pIJ2925 (Janssen and Bibb, 1993) and pBluescript vectors (Stratagene) were used. Plasmids constructed in this work are listed in Table S6 and illustrated in Figs 2 and 4.

DNA manipulation and cloning

Basic procedures for DNA work were as in Sambrook and Russell (2001) for E. coli and Kieser et al. (2000) for Streptomyces. PCR-targeted mutagenesis on DNA cloned in E. coli, to generate both marked deletions and their unmarked derivatives (after yeast ‘FLP’-recombinase-mediated excision of marker resistance genes), and the mobilization of mutated DNA into Streptomyces, were as in Gust et al. (2004), using the primers listed in Table S3. All mutants were confirmed by PCR and Southern blotting as in Gust et al. (2003). The Quickchange kit (Stratagene) was used for site-directed mutagenesis. Mutations were confirmed by DNA sequencing.

S1 nuclease protection analysis

The isolation of RNA and the protocol for S1 mapping were as in Kieser et al. (2000).

Bioassays for Mm, MMF and SCB1

The detection of Mm used the SCP1-free strain J1501 as indicator. The agar piece method of Kirby and Hopwood (1977) was adopted. Cosynthesis of Mm was tested by placing agar pieces inoculated with a secretor or convertor strain next to an agar piece inoculated with the strain to be tested, as in Kirby and Hopwood (1977), on CM spread with J1501. SCB1 was detected by its ability to stimulate red/blue pigmentation on an SMMS plate spread with spores of M145 (Takano et al., 2000).

Extraction of Mm, MMF and SCB1

The extraction of Mm from CM liquid culture filtrates and its detection by TLC were as in Kirby et al. (1975). MMF or SCB1 were extracted from culture filtrates or surface culture by mixing the liquid or macerated agar with an equal volume of ethyl acetate and leaving overnight before removing the ethyl acetate and drying it in a rotary evaporator. The resulting material was dissolved in 200 μl methanol.

Construction and analysis of transcriptional fusions to the xylE reporter gene

Transcriptional fusions to the xylE reporter (Ingram et al., 1989) were constructed by inserting fragments of the Mm gene cluster next to xylE in the polylinker region of the phage vector KC861 as described by Bruton et al. (1991). In brief, this vector contains the thiostrepton-resistance gene tsr to select for lysogeny, but it lacks integration functions. The inserted fragments permitted it to integrate by homologous recombination into S. coelicolor strains carrying the mmy cluster as part of plasmid SCP1. Such lysogens could be selected on thiostrepton-containing medium. The fragments used corresponded to the PstI fragments C2.18 (ends in mmyG), A4.2 (ends in mmyK) and A3.13 (ends in mmyD), and a BglII–SstI fragment from pIJ519 (ends in mmfH): these fragments were previously described in Chater and Bruton (1985), and are shown in Fig. 1. They were all converted into BglII fragments by subcloning into pIJ2925, and ligated into the BamHI site of KC861. Ligation mixtures were used to transfect protoplasts of Streptomyces lividans 1326 (Kieser et al., 2000). The resulting plaques were arrayed on nitrocellulose filters, which were hybridized with non-radioactively labelled probes. The orientation of inserts was determined by restriction analysis. After characterization, representative phages were used to lysogenize J1507 and J1703 (Table 3). Southern blots of DNA extracted from the lysogens confirmed that integration was at the predicted locations in the mmy cluster. The catechol oxygenase activity of lysogens was displayed by spraying 42 h and 72 h cultures on CM agar medium with catechol (Ingram et al., 1989). The plates were photographed 90 min after catechol addition. Activity, revealed by yellow coloration, was detected only with correctly orientated insertions into J1506 and J1507. Quantitative measurements of catechol oxygenase activity were carried out on cell-free extracts of mycelium scraped from cultures grown for different lengths of time on cellophane discs on CM agar (Ingram et al., 1989).

Acknowledgements

This work was initiated on the basis of DNA sequence information provided by Celia J. Bruton and Nigel Hartley (Accession number AJ276673). The mutational cloning experiment using an XhoI fragment from mmyB to mmyF was performed by Sue Angell. S. O'Rourke was supported by a studentship from the John Innes Foundation, and Andreas Wietzorrek by the John Innes Foundation. Keith Chater is a John Innes Foundation Emeritus Fellow.

Ancillary