Different alleles of the response regulator gene bldM arrest Streptomyces coelicolor development at distinct stages

Authors


Abstract

whiK was one of five new whi loci identified in a recent screen of NTG-induced whi mutants and was defined by three mutants, R273, R318 and R655. R273 and R318 produce long, tightly coiled aerial hyphae with frequent septation. In contrast, R655 shows a more severe phenotype; it produces straight, undifferentiated aerial hyphae with very rare short chains of spores. Subcloning and sequencing showed that whiK encodes a member of the FixJ subfamily of response regulators, with a C-terminal helix–turn–helix DNA-binding domain and an apparently typical N-terminal phosphorylation pocket. Unexpectedly, a constructed whiK null mutant failed to form aerial mycelium, showing that different alleles of this locus can arrest Streptomyces coelicolor development at very distinct stages. As a consequence of the null mutant phenotype, whiK was renamed bldM. The bldM null mutant fits into the extracellular signalling cascade proposed for S. coelicolor and is a member of the bldD extracellular complementation group. The three original NTG-induced mutations that defined the whiK/bldM locus each affected the putative phosphorylation pocket. The mutations in R273 and in R318 were the same, replacing a highly conserved glycine (G-62) with aspartate. The more severe mutant, R655, carried a C-7Y substitution adjacent to the highly conserved DD motif at positions 8–9. However, although BldM has all the highly conserved residues associated with the phosphorylation pocket of conventional response regulators, aspartate-54, the putative site of phosphorylation, is not required for BldM function. Constructed mutant alleles carrying either D-54N or D-54A substitutions complemented the bldM null mutant in single copy in trans, and strains carrying the D-54N or the D-54A substitution at the native chromosomal bldM locus sporulated normally. BldM was not phosphorylated in vitro with either of the small-molecule phosphodonors acetyl phosphate or carbamoyl phosphate under conditions in which a control response regulator protein, NtrC, was labelled efficiently.

Introduction

A special feature of the developmental cycle of the filamentous bacteria Streptomyces is the formation, at the start of differentiation, of an aerial mycelium, a structure consisting of hyphae that have to grow out of the aqueous environment of the substrate mycelium into the air. These multigenomic aerial hyphae, which impart a characteristic fuzzy appearance to the developing colonies, subsequently differentiate to form chains of exospores (Chater, 1998; Kelemen and Buttner, 1998).

The development of aerial hyphae in Streptomyces has been dissected genetically by the isolation of bld mutants that lack aerial mycelium and therefore have a shiny, ‘bald’ appearance. Unlike a second class of developmental mutations, whi mutations (which appear only to affect the differentiation of aerial hyphae into spores), bld mutations have pleiotropic effects, which often cause defects in carbon catabolite repression and in cell–cell signalling and sometimes prevent antibiotic production, in addition to blocking differentiation (Merrick, 1976; Champness, 1988; Willey et al., 1993; Pope et al., 1996; Kelemen and Buttner, 1998; Nodwell et al., 1999).

It seems clear from the behaviour of bld mutants on different media that aerial mycelium formation can occur by at least two different pathways. When Streptomyces coelicolor is grown on rich media such as R2YE, aerial mycelium is associated with the production of a small, hydrophobic peptide called SapB, a putative morphogen that coats the surface of aerial hyphae (Willey et al., 1991). SapB is a surfactant that allows aerial hyphae to break the surface tension of the aqueous environment of the substrate mycelium and grow into the air (Tillotson et al., 1998). On R2YE, SapB production and aerial mycelium formation depend on bldA, bldB, bldC, bldD, bldF, bldG, bldH, bldI, bldJ (formerly bld261;Nodwell et al., 1999) and bldK. However, aerial mycelium formation in these mutants can be restored by the exogenous addition of purified SapB protein, or by growing them close to SapB-producing colonies (Willey et al., 1991). On the other hand, aerial mycelium formation and sporulation can be restored to most bld mutants (an exception being bldB) simply by growing them on a minimal medium containing mannitol as the sole carbon source and yet, under these conditions, no SapB is detectable (Willey et al., 1991).

What are the signals that initiate the erection of the aerial hyphae? On rich media, there is evidence for the involvement of a complex extracellular signalling cascade in S. coelicolor (Willey et al., 1993; Nodwell et al., 1996; 1999; Nodwell and Losick, 1998). When some pairs of bld mutants are grown on R2YE in close proximity, but without actual contact, one mutant induces the other both to synthesize SapB and to erect aerial hyphae and sporulate. If this ‘extracellular complementation’ is observed, it is always unidirectional, with one bld mutant acting as a ‘donor’ and the other as a ‘recipient.’ Experiments with the whole range of bld mutants showed that a subset falls into the following hierarchy, in which each mutant can rescue the developmental defect in all the mutants to the left, but not to the right:

[bldJ] < [bldK, bldL] < [bldA, bldH] < [bldG] < [bldC] < [bldD].

These data have been interpreted to mean that aerial mycelium formation on rich media is initiated by a signalling cascade involving at least five different extracellular signals. Each signal causes the synthesis and/or release of the next signal, eventually causing the bldD-dependent production of SapB, and perhaps other morphogens, which allow aerial hyphae to overcome surface tension and grow into the air (Willey et al., 1993; Nodwell et al., 1996; 1999; Nodwell and Losick, 1998). Currently, there is biochemical evidence only for ‘signal 1’ (see Fig. 5A in Results;Nodwell and Losick, 1998).

Figure 5.

Positioning of bldM in the extracellular signalling cascade proposed by Willey et al. (1993). The bldM null mutant restored a fringe of aerial mycelium formation to representative bldK (B) and bldJ, bldA, bldH, bldG and bldC mutants (not shown). bldM failed to restore aerial mycelium formation to a bldD mutant, and the bldD mutant also failed to restore aerial mycelium formation to bldM (C). bldM is therefore a member of the bldD extracellular complementation group (A; adapted from Kelemen and Buttner, 1998). Extracellular complementation was scored after 4 days of growth.

whiK is one of five new whi loci identified in a recent screen of NTG-induced whi mutants (Ryding et al., 1999). It was defined by three mutants, R273, R318 and R655. All three strains formed white colonies with wild-type levels of aerial mycelium. R273 and R318 produced long, tightly coiled aerial hyphae with frequent septation. In contrast, R655 showed a more severe phenotype; it produced straight, undifferentiated aerial hyphae with very rare short chains of spores (Ryding et al., 1999).

Here, we show that whiK encodes an apparently typical member of the response regulator family, proteins whose activity is usually regulated by phosphorylation by a cognate sensor kinase. However, we show that aspartate-54 (D-54), the predicted site of phosphorylation in WhiK, is not required for function. Unexpectedly, we found that a constructed whiK null mutant was bald, showing that different alleles of this locus can block S. coelicolor development at very distinct stages. As a consequence of the null mutant phenotype, whiK was renamed bldM. bldM fits into the extracellular complementation cascade proposed by Losick and coworkers (Willey et al., 1993; Nodwell et al., 1996; 1999; Nodwell and Losick, 1998), and is a member of the bldD extracellular complementation group.

Results

whiK encodes a putative response regulator

Ryding et al. (1999) isolated a single SCP2*-based clone, pIJ6700, that complemented the three whiK mutants R273, R318 and R655. The nucleotide sequence of the 3.8 kb insert carried by pIJ6700 was determined, and potential protein-coding sequences were predicted with the aid of the frame program (Bibb et al., 1984). One incomplete (orf4) and three complete potential protein-coding sequences (orf1, orf2 and orf3) were identified (Fig. 1). orf1, orf2 and orf3 were each subcloned into the conjugative vector pSET152, which integrates site specificity into the S. coelicolor chromosome at the phage ΦC31 attB site, generating pIJ6602, pIJ6603 and pIJ6604 respectively (Fig. 1). Each orf was introduced in turn into the three whiK mutants. Neither the pSET152 vector alone, nor orf2 or orf3, had any effect on the phenotypes of R273, R318 and R655. In contrast, pIJ6602, carrying orf1, restored wild-type levels of sporulation to all three strains. As a consequence, orf1 was designated whiK.

Figure 1.

Genetic organization of the 3.8 kb clone carrying whiK/bldM. The positions of the four protein-coding regions are indicated by arrows, and restrictions sites referred to in the text are marked. The extent of the subclones used in the complementation tests is shown below. The location of whiD, immediately upstream of whiK/bldM, is also shown.

Global similarity searches of the EMBL databases showed that the incomplete open reading frame, orf4, encodes the N-terminal 82 residues of a homologue of a 38.9 kDa protein of unknown function found in the genome of the related actinomycete Mycobacterium tuberculosis. orf2 encodes an ECF (Extracytoplasmic Function) RNA polymerase sigma factor, and orf3 is guaB, encoding inosine 5′ monophosphate dehydrogenase, an enzyme involved in GMP biosynthesis.

WhiK is 203 amino acids long and is an apparently typical member of the FixJ subfamily of two-component response regulators (Fig. 2) (Da Re et al., 1994; Pao and Saier, 1995; Hackenbeck and Stock, 1996), showing, for example, 32% identity with DegU, involved in the control of competence and degradative enzyme biosynthesis in Bacillus subtilis (Dahl et al., 1992). Two-component regulatory systems represent a conserved signal transduction system through which bacteria accomplish diverse signalling tasks. In prototypical systems, signals are perceived by a sensor kinase that autophosphorylates on a conserved histidine residue using the γ-phosphoryl group of ATP, at a rate proportional to the magnitude of the stimulus. The phosphate is then transferred from the histidine of the sensor kinase to a conserved aspartate in the response regulator, thereby modulating its activity (Hackenbeck and Stock, 1996; Egger et al., 1997). The genes encoding response regulator–sensor histidine protein kinase cognate pairs are often, but not always, adjacent in the bacterial chromosome. Analysis of the ongoing S. coelicolor genome sequence (http:// www.sanger.ac.uk/Projects/S_coelicolor/) showed that there is no sensor kinase gene within 40 kb of whiK on the S. coelicolor chromosome. Interestingly, the gene lying immediately upstream of whiK is another recently characterized developmental regulatory gene, whiD (Fig. 1; Molle et al. 2000). The gene order whiD, ECF sigma factor gene, guaB, orf4 is preserved in the chromosome of M. tuberculosis, but no homologue of whiK is present between the homologues of whiD and the ECF sigma factor gene.

Figure 2.

Alignment of the predicted amino acid sequence of WhiK/BldM with other members of the FixJ subfamily of response regulators and with CheY. Conserved lysine and aspartate residues of the phosphorylation pocket are marked by open triangles, the putative site of phosphorylation is marked with an asterisk, the C-terminal helix–turn–helix motif is underlined, the site of insertion of the hyg cassette in the whiK::hyg null mutant is shown, and the amino acid substitutions carried by the original whiK mutants, R273, R318 and R655, are indicated. The proteins and their corresponding amino acid sequence accession numbers are: WhiK_Strco, S. coelicolor WhiK/BldM (CAA09263); DnrN_Strpe, S. peucetius DnrN (AAD15247); RamR_Strco, S. coelicolor RamR (AAA21390); UhpA_Ecoli, E. coli UhpA (P10940); UvrY_Ecoli, E. coli UvrY (P07027); NarL_Ecoli, E. coli NarL (P10957); NarP_Ecoli, E. coli NarP (P31802); AbsA2_Strco, S. coelicolor AbsA2 (AAB08053); DegU_Bacsu, B. subtilis DegU (P13800); BvgA_Borpe, Bordella pertussis BvgA (P16574); RedZ_Strco, S. coelicolor RedZ (CAA69209); FixJ_Rhim, Rhizobium meliloti FixJ (P10958); WhiI_Strco, S. coelicolor WhiI (CAA19241) CheY_Ecoli, E. coli CheY (P06143).

Like most response regulators, WhiK has an N-terminal receiver domain carrying four conserved aspartate and lysine residues, which are known to form an acidic phosphorylation pocket in other members of the family. These four residues include Asp-54, the putative site of phosphorylation (Stock et al., 1989) (Fig. 2). In addition, in common with other members of the FixJ subfamily of response regulators, WhiK has a putative helix–turn–helix (H–T–H) DNA-binding motif at its C-terminus (Fig. 2).

A constructed whiK null mutant is unable to erect aerial hyphae

A whiK null mutant allele was constructed in vitro by inserting a hygromycin resistance gene (hyg) at a unique SphI restriction site internal to whiK(Figs 1 and 3). This mutant allele was used to replace the wild-type allele in J1915 as described in Experimental procedures. The chromosomal structure of a representative whiK mutant was confirmed by Southern blot analysis (Fig. 3C), and the strain was designated J2151.

Figure 3.

Construction of a whiK/bldM null mutant.

A and B. Restriction maps of the wild-type and whiK::hyg null mutant alleles respectively; the black arrow represents the whiK gene.

C. Southern blot analysis of chromosomal DNA from J1915 (whiK+ lanes 2 and 4) and J2151 (whiK::hyg; lanes 3 and 5) digested with ApaI (lanes 2 and 3) or BamHI and HindIII (lanes 4 and 5). The size markers (lane 1) are the 1 kb ladder (Bethesda Research Laboratories). The probe used was the 3.5 kb BamHI–HindIIl fragment (A).

Unexpectedly, J2151 was found to have a bald phenotype, being unable to erect aerial hyphae on MS agar or R2YE (Figs 4A, 5B and C). Unlike some S. coelicolor bld mutants (Merrick, 1976; Nodwell et al., 1999), J2151 was not blocked in the production of actinorhodin or undecylprodigiosin, the two pigmented antibiotics made by this strain. For most of the bld mutants (an exception being bldB), mycelium formation and sporulation can be restored simply by growing them on MM containing mannitol as the sole carbon source. However, J2151 showed no signs of aerial mycelium formation on MM containing mannitol. As a consequence of the phenotype of the null mutant, whiK was renamed bldM. The bldM null mutant was fully complemented by pIJ6602, the pSET152 derivative carrying bldM (Fig. 4B). The ΔglkA119 deletion, present in the genetic background of J2151, causes ectopic sporulation in the substrate hyphae (the Esp phenotype) of the parent strain, J1915, on certain media (Kelemen et al., 1995). However, introduction of the DNA deleted in the ΔglkA119 background into J2151 on the integrative plasmid pSET152 had no morphological consequences, showing that the ΔglkA119 deletion did not contribute to the bald phenotype of J2151.

Figure 4.

Scanning electron micrographs showing (A) the bald phenotype of the constructed bldM null mutant, J2151, and (B) its complementation by pIJ6602, the pSET152 derivative carrying bldM (see Fig. 1). Colonies were grown on MS agar for 4 days.

bldM is a member of the bldD extracellular complementation group

To see whether bldM could be positioned in the extracellular complementation hierarchy proposed by Willey et al. (1993), we grew the bldM null mutant J2151 on R2YE close to each bld mutant in turn and found that it restored a fringe of aerial mycelium formation to representative bldJ, bldK, bldA, bldH, bldG and bldC mutants (e.g. Fig. 5B). In contrast, bldM failed to restore aerial mycelium formation to a bldD mutant, and the bldD mutant also failed to restore aerial mycelium formation to bldM (Fig. 5C). bldM therefore fits into the proposed hierarchy and is a member of the bldD extracellular complementation group (Fig. 5A).

The NTG-induced whiK/bldM mutations affect the putative phosphorylation pocket

The three NTG-induced mutations that originally defined the whiK/bldM locus were amplified from the chromosomes of R273, R318 and R655 and sequenced. R273 and R318 were found to carry the same mutation, suggesting that these two strains may be clonal in origin. The R273/R318 mutation is a GGC to GAC change, giving rise to a G-62D substitution, and the more severe mutant, R655, carries a TGC to TAC change, giving rise to a C-7Y substitution (Fig. 2). Both these mutations fall in the putative phosphorylation pocket of BldM.

The crystal structures of two response regulators, NarL and CheY, have been determined (Stock et al., 1989; Volz and Matsumura, 1991; Baikalov et al., 1996). The levels of amino acid similarity of BldM with NarL and CheY were sufficient to allow us to model the structure of the phosphorylation pocket of BldM using the swissmodel program (Fig. 6). The R655 mutation introduces a bulky aromatic tyrosine residue adjacent to the highly conserved pair of aspartates at positions 8–9 (Fig. 6). These aspartate residues are known to be important in Mg2+ binding and catalysis in other response regulators (Mg2+ is required for the phosphoryl group transfer reactions of CheY and, presumably, all other response regulators; Volz, 1995). The less severe R273/R318 mutation introduces an additional aspartate into the putative phosphorylation pocket in place of a highly conserved glycine (Fig. 6). In CheY, this glycine lies at the start of the α3 helix after a tight 180° turn of only three amino acids (Volz and Matsumura, 1991). As noted by Volz (1995), this geometry places severe steric restrictions on the nature of the side-chain at this position, suggesting that substitution of the glycine with a larger residue such as aspartate would necessarily distort the overall structure of the phosphorylation pocket.

Figure 6.

Hypothetical three-dimensional structure of the phosphorylation pocket of BldM, showing the positions of the highly conserved aspartate (D-8, D-9 and D-54) and lysine (K-104) residues and the amino acids substituted in the R273/R318 (G-62D) and R655 (C-7Y) point mutants. The BldM phosphorylation pocket was modelled on the crystal structures of NarL and CheY, using the program swissmodel. The phenotypes of the R273/R318 (G-62D) and R655 (C-7Y) mutants, as revealed in the scanning electron microscope, are shown. Colonies were grown on MS agar for 4 days.

Asp-54 is not required for BldM function

Given that the NTG-induced whiK/bldM mutations affected the putative phosphorylation pocket, we investigated the role of D-54 in BldM function. Two amino acid substitutions were made at D-54, replacing the putative site of phosphorylation with either an asparagine (D-54N) or an alanine residue (D-54A). Both these substitutions render position 54 non-phosphorylatable, and Asp → Asn substitutions in particular have been widely used to block phosphorylation of many other response regulators, thereby rendering them inactive. The D-54N and D-54A alleles, carried on 1.3 kb PvuII fragments, were cloned into the integrative vector pSET152 to create pIJ6614 and pIJ6615, respectively, and were introduced into the bldM null mutant strain J2151 by conjugation from Escherichia coli. Unexpectedly, both alleles restored wild-type levels of aerial mycelium formation and sporulation. One possible explanation for this result could have been high-frequency gene conversion between the bldM::hyg allele and the D-54N or D-54A allele integrated at the ΦC31 attB site. However, the D-54 codon lies only 10 bp from the site of insertion of the hyg cassette in the bldM null mutant allele (Fig. 2), and polymerase chain reaction (PCR) amplification and sequencing confirmed that the D-54N and D-54A mutations were still present in the complementing alleles at the ΦC31 attB site. In other two-component systems, overexpression of the non-phosphorylated form of the protein can sometimes mimic the activity of the phosphorylated form (Hackenbeck and Stock, 1996). To eliminate possible vector-driven overexpression of the D-54N or D-54A alleles as an explanation for complementation, we constructed two strains carrying either the D-54N substitution (strain J2153) or the D-54A substitution (strain J2154) at the native chromosomal bldM locus, as described in Experimental procedures. Both J2153 and J2154 made wild-type levels of aerial mycelium and sporulated normally, showing that Asp-54 is not required for BldM function.

BldM could not be phosphorylated in vitro by acetyl phosphate or carbamoyl phosphate

Some response regulators can be phosphorylated in vitro in the absence of their cognate kinases using low-molecular-weight phosphodonors such as acetyl phosphate, phosphoramidate or carbamoyl phosphate (Feng et al., 1992; Lukat et al., 1992; McCleary et al., 1993; Gu et al., 1994; McCleary and Stock, 1994; Reyrat et al., 1994). To determine whether BldM could be phosphorylated in vitro, we overexpressed N-terminally His-tagged derivatives of the wild-type, D-54N and D-54A forms of the protein in E. coli. All three forms remained soluble and were purified using nickel chelate affinity chromatography. We could not detect phosphorylation of any form of BldM using either [32P]-acetyl phosphate or [32P]-carbamoyl phosphate under conditions in which a control response regulator, NtrC, was phosphorylated efficiently by both phosphodonors (Fig. 7 and data not shown). The same negative results were obtained after removal of the BldM N-terminal His-tags using thrombin (data not shown).

Figure 7.

Attempted in vitro phosphorylation of BldM, BldM D-54N, BldM D-54A and NtrC. [32P]-carbamoyl phosphate was generated and incubated with BldM (lane 3), BldM D-54N (lane 4), BldM D-54A (lane 5), NtrC (lane 6) or no additional protein (lane 2). Controls lanes were BldM (lane 1) and NtrC (lane 7) before exposure to [32P]-carbamoyl phosphate.

A. Coomassie brilliant blue-stained SDS gel.

B. Autoradiograph of the same gel.

The positions of carbamate kinase (CK), NtrC and BldM are indicated. In this experiment, the N-terminal His-tags were still present on the three BldM proteins, but identical results were obtained after their removal with thrombin.

Discussion

bldM encodes a putative response regulator

In the simplest model, all bld genes might encode direct components of the signalling cascade. From the bld gene sequences available to date, this is clearly not the case. The best-characterized bld gene is bldA of S. coelicolor, which encodes the only tRNA that can efficiently translate the rare leucine codon UUA (Lawlor et al., 1987; Leskiw et al., 1991a,b). Loss of production of the two S. coelicolor pigmented antibiotics, actinorhodin (Act) and undecylprodigiosin (Red), in bldA mutants can be accounted for by the presence of TTA codons within actII-ORF4 (Fernández-Moreno et al., 1991) and redZ (White and Bibb, 1997; Guthrie et al., 1998), encoding pathway-specific activators for Act and Red biosynthesis respectively. However, the UUA-containing gene(s) that account for the morphological defects of bldA mutants have yet to be identified. Similarly, bldB and bldD encode small transcription factors (Elliot et al., 1998; Pope et al., 1998; Elliot and Leskiw, 1999) that must also have indirect roles in signal detection or synthesis. On the other hand, bldK is a complex locus that encodes homologues of the subunits of known oligopeptide permeases, and mutations in bldK confer resistance to the toxic tripeptide bialaphos, implying that BldK is an oligopeptide importer (Nodwell et al., 1996). Here, we show that bldM encodes a response regulator, a class of proteins involved in signal transduction, typically in conjunction with a membrane-localized, cognate sensor kinase.

Is BldM regulated by phosphorylation?

BldM has an apparently typical N-terminal phosphorylation domain, and the whi mutations that originally defined the locus fall in the putative phosphorylation pocket. However, Asp-54 is not required for BldM function, as substitution of this residue with alanine or asparagine has no phenotypic consequence.

When the normal site of phosphorylation (Asp-57) of the chemotaxis response regulator CheY is replaced with an asparagine, the resulting protein is active in vivo as a result of phosphorylation of the adjacent residue, Ser-56 (Bourret et al., 1990; Appleby and Bourret, 1999). (This is true only in the absence of CheZ, a protein that greatly enhances dephosphorylation of CheY; Appleby and Bourret, 1999.) Introduction of an S-57A substitution has no effect on the wild-type CheY protein, but an S-57A/D-57N double mutant is inactive. These results show that phosphorylation at an alternative site can cause activation of a mutant response regulator lacking the conventional site of phosphorylation. However, there is no evidence that alternative site phosphorylation occurs in wild-type CheY and, hence, that it normally plays a role in chemotaxis (Appleby and Bourret, 1999). Two other response regulators are known to be phosphorylated at an alternative site when the natural site has been substituted. FixJ D-54N from Rhizobium meliloti is similar to CheY D-57N, in that it shows constitutive partial activity associated with phosphorylation of an alternative site (Reyrat et al., 1994), and E. coli NtrC D-54N is weakly phosphorylated in vitro (Moore et al., 1993). In neither of these two cases has the alternative site of phosphorylation been identified but, as Appleby and Bourret (1999) point out, like CheY, both these proteins have a hydroxyamino acid immediately adjacent to the aspartate on the N-terminal side (threonine-53 in the case of FixJ and serine-53 in the case of NtrC). A strictly analogous alternative phosphorylation event in BldM D-54N is not possible because the residue at position 53 of BldM is methionine (Fig. 2). It should also be noted that, while substitution of CheY Asp-57 with asparagine allows phosphorylation of Ser-56, other reported substitutions at position 57 of CheY do not. In contrast, not only is BldM D-54N fully active, but so is BldM D-54A.

Two response regulators with unusual phosphorylation pockets have been described in S. coelicolor. The early whi locus, whiI, required for normal sporulation septum formation in the aerial hyphae, encodes a member of the response regulator family of proteins (Aínsa et al., 1999). However, WhiI lacks one of the aspartate residues and the lysine residue strongly conserved in the conventional phosphorylation pocket (Fig. 2; Aínsa et al., 1999). The RedZ response regulator protein is a pathway-specific activator encoded within the red cluster of genes in S. coelicolor, and RedZ is absolutely required for production of the red antibiotic undecylprodigiosin (White and Bibb, 1997; Guthrie et al., 1998). RedZ lacks the two conserved phosphorylation pocket aspartate and lysine residues that are absent from WhiI, and RedZ also lacks the conserved aspartate residue corresponding to the site of phosphorylation in conventional response regulators (Fig. 2; Guthrie et al., 1998). Nevertheless, there is experimental evidence to suggest that the activity of RedZ is regulated post-translationally (J. White and M. J. Bibb, personal communication).

Several explanations for our observations on BldM are possible. Conceivably, the D-54A and D-54N forms of BldM could simply mimic the phospho-aspartate protein, but as Asp → Asn substitutions have been widely used to inactivate response regulators, this seems highly unlikely. It is also conceivable that wild-type BldM is regulated by phosphorylation on Asp-54 but that, in the absence of Asp-54, BldM can be activated by phosphorylation on an alternative residue in a manner analogous to CheY, although not at position 53, which is a methionine in BldM. Appleby and Bourret (1999) were able to identify Ser-56 as the alternative site of phosphorylation in the D-57N protein because CheY can be efficiently phosphorylated in vitro. Unfortunately, we could not phosphorylate the wild-type, D-54N or D-54A forms of BldM using either [32P]-acetyl phosphate or [32P]-carbamoyl phosphate under conditions in which a control response regulator, NtrC, was phosphorylated by both phosphodonors. If BldM is part of a conventional two-component system, a possible way forward is to try to develop a genetic screen for the putative cognate sensor kinase gene. The genes encoding response regulator–sensor kinase cognate pairs are often, but not always, adjacent, but there is no sensor kinase gene close to bldM on the S. coelicolor chromosome. The gene lying immediately upstream of bldM is another developmental regulatory gene, whiD (Fig. 1; Molle et al. 2000). Whether the proximity of bldM to whiD has any significance remains to be seen.

A further possibility is that the wild-type protein is simply not regulated by phosphorylation, at least on Asp-54. There is at least one clear precedent for our results with BldM in the literature, also arising from the streptomycetes. Otten et al. (1995) identified DnrN as a protein essential for daunorubicin biosynthesis in Streptomyces peucetius. Like BldM, DnrN has an apparently typical phosphorylation domain, but Furuya and Hutchinson (1996) refer to this protein as a ‘pseudo-response regulator’ because a D-55N mutant of DnrN is active in vivo, and the mutant and wild-type DnrN proteins bind to a target promoter with equal affinity (Kd = 50 nM) in vitro (Otten et al., 1995; Furuya and Hutchinson, 1996). Because there is evidence that RedZ activity is regulated post-translationally, even though RedZ lacks the conserved aspartate residue corresponding to the site of phosphorylation in classical response regulators, it has been suggested that it could work in conjunction with a co-regulator interacting with the phosphorylation pocket (J. White and M. J. Bibb, personal communication). If the same were true for BldM, this could explain why the whi mutations that originally defined the whiK/bldM locus fall into the phosphorylation pocket, yet Asp-54 is not required for BldM function.

Does BldM have two separate functions during differentiation?

The three original NTG-induced whiK mutants, R273, R318 and R655, had classical whi mutant phenotypes, being apparently defective solely in the differentiation of aerial mycelium into mature spore chains. In particular, all three strains produced wild-type levels of aerial mycelium. It was therefore completely unexpected when a constructed whiK null mutant proved to be bald (in the original screen, we specifically looked for white mutants, so we would not have isolated null mutants at this locus). Different alleles of this single locus can therefore arrest S. coelicolor at very distinct developmental stages, which is unprecedented in Streptomyces. The meaning of these observations is unclear. Perhaps the original point mutants simply represent weak alleles, and these somehow permit wild-type levels of aerial mycelium formation while preventing maturation of this mycelium into spores. Alternatively, there is the intriguing possibility that BldM/WhiK has two separate functions during differentiation, and that these functions can be dissected genetically.

Experimental procedures

Bacterial strains, plasmids, growth conditions and conjugal plasmid transfer from E. coli to Streptomyces

S. coelicolor strains used are summarized in Table 1 and were cultured on R2YE, MM containing 0.5% (w/v) mannitol as a carbon source or MS agar (mannitol plus soya flour) (Kieser et al., 2000). To bypass the methyl-specific restriction system of S. coelicolor during conjugation from E. coli, unmethylated plasmids were conjugated from dam dcm hsdS E. coli strain ET12567 (MacNeil et al., 1992), as described by Ryding et al. (1999). Plasmids used were pKC1132 (Bierman et al., 1992), pIJ2925 (Janssen and Bibb, 1993) and pSET152 (Bierman et al., 1992).

Table 1. Derivatives of S. coelicolor A3(2) used in this work.
StrainGenotypeReference 
J1915 ΔglkA119 SCP1 SCP2 Kelemen et al. (1995)  
J2151 ΔglkA119 bldM::hyg SCP1 SCP2This work 
J2153 ΔglkA119 bldM-D54N SCP1 SCP2This work 
J2154 ΔglkA119 bldM-D54A SCP1 SCP2This work 
R273 bldM273 SCP1 SCP2 Ryding et al. (1999)  
R318 bldM318 SCP1 SCP2 Ryding et al. (1999)  
R655 bldM655 SCP1 SCP2 Ryding et al. (1999)  
J1700 bldA39 hisA1 uraA1 strA1 SCP1 SCP2 Lawlor et al. (1987)  
J660 bldC18 mthB2 cysD18 agaA7 SCP1NF SCP2* Merrick (1976)  
J774 bldD53 cysA15 pheA1 mthB2 strA1 SCP1NF SCP2* Merrick (1976)  
WC103 bldG103 hisA1 uraA1 strA1 Pgl SCP1 SCP2 Champness (1988)  
WC109 bldH109 hisA1 uraA1 strA1 Pgl SCP1 SCP2 Champness (1988)  
HU261 bldJ261 hisA1 uraA1 strA1 Pgl SCP1NF SCP2* Willey et al. (1993)  
NS17 bldK::aadA SCP1 SCP2 Nodwell et al. (1996)  

Nucleotide sequencing

The 3.8 kb fragment carrying whiK was sequenced with the universal primer after the generation of a set of nested deletions. Deletions were generated with exonuclease III according to the Erase-a-Base kit protocol (Promega), based on the method of Henikoff (1984). The nucleotide sequence has been deposited with the DDBJ/EMBL/GenBank databases (accession no. AJ010601).

Construction of a whiK null mutant

A whiK null mutant derivative of J1915, a plasmid-free, glkA derivative of the wild-type strain, was constructed using the method of Buttner et al. (1990). This method makes use of the counterselectable glucose kinase gene (glkA), which allows a positive selection to be made for gene replacement, provided that the mutations are constructed in a strain carrying a deletion of glkA. A 1.3 kb BamHI–SphI fragment carrying the 5′ end of whiK was cloned into pUC19 digested with BamHI and SphI, generating pIJ6626. A 1.7 kb SphI–BstXI fragment carrying the 3′ end of whiK was cloned into pIJ6626 digested with SphI and HindIII to create pIJ6608, and a whiK null mutant allele was generated by blunt-end cloning a 1.8 kb hyg cassette (Zalacain et al., 1986) into the SphI site internal to whiK, yielding pIJ6609 (Figs 1 and 5). This reconstructed a contiguous 3 kb segment of the chromosome, but carrying the 1.8 kb hyg insertion in whiK. The entire 4.8 kb insert was removed from pIJ6609 as a PvuII fragment and cloned into the EcoRV site of pKC1132. Finally, a 1.3 kb BglII fragment carrying the counterselectable glkA gene was ligated into the unique BglII site in the vector to create pIJ6610.

pIJ6610 was introduced into S. coelicolor J1915 (ΔglkA119) by mating from E. coli, and exconjugants in which the plasmid had presumptively integrated at the whiK locus by single-crossover homologous recombination were selected with apramycin. After one round of non-selective growth, putative whiK::hyg null mutants in which the delivery plasmid had been lost were selected on MM containing 100 mM 2-deoxyglucose and 50 µg ml−1 hygromycin.

Construction of D-54N and D-54A alleles of bldM

A 1 kb XbaI–ApaI fragment carrying bldM was blunt-end cloned into pUC19 digested with XbaI to create pIJ6605. Aspartate to asparagine (D-54N) and aspartate to alanine (D-54A) substitutions were engineered by polymerase chain reaction (PCR)-based site-directed mutagenesis of pIJ6605. In this method, two abutting oligonucleotides were used to amplify the entire pIJ6605 plasmid, simultaneously introducing a single basepair change in the D-54 codon of bldM. The PCR programme was 10 cycles of 30 s at 95°C, 45 s at 60°C and 3 min at 68°C, followed by 10 cycles of 30 s at 95°C, 45 s at 60°C and 4 min at 68°C, before a final 10 min incubation at 68°C. The primers used for the D-54N substitution were 5′-AACGTGCGCATGCCCGGCCTGG-3′ and 5′-CATCAGAATCAGGTCCGAGCGG-3,′ and for the D54A substitution 5′-GCCGTGCGCATGCCCGGCCTGGGCG-3′ and 5′-CATCAGAATCAGGTCCGAGCGGTCG-3′ (the N-54 and A-54 codons are underlined). The PCR reactions were performed using high-fidelity Pfu DNA polymerase (Promega). The PCR products were self-ligated to regenerate the replicon, and the resulting alleles of bldM were sequenced over their entire length to ensure that only the desired mutation had been introduced. Finally, 1.3 kb PvuII fragments carrying the constructed bldM mutant alleles were cloned into the EcoRV site of the conjugative vector pSET152 to create pIJ6614 (D-54N) and pIJ6615 (D-54A).

Construction of J2153 and J2154, strains carrying D-54N or D-54A alleles, respectively, at the native bldM locus

A counterselectable version of the conjugative vector pKC1132 was created by cloning a 1.3 kb BglII fragment carrying glkA into the unique BglII site in the vector to generate pIJ6607. Subsequently, 1.3 kb PvuII fragments carrying constructed bldM D-54N and D-54A mutant alleles (see above) were cloned into the EcoRV site of pIJ6607 to create pIJ6616 (D-54N) and pIJ6619 (D-54A). The D-54N and D-54A mutant alleles carried on pIJ6616 and pIJ6619 were used to replace the bldM::hyg null allele in the chromosome of J2151, taking advantage of glkA to counterselect the delivery vector. The D-54 codon was only 10 bp from the site of insertion of the hyg cassette in the bldM::hyg null mutant allele, minimizing any possibility of recombination in that interval. In addition, the D-54N/D-54A mutation was almost exactly central to the 1 kb of Streptomyces DNA within the 1.3 kb PvuII fragment, thus promoting crossovers on either side of the mutation in approximately equal measure.

pIJ6616 and pIJ6619 were introduced individually into J2151 (bldM::hyg ΔglkA119) by mating from E. coli, and exconjugants were selected with apramycin. After one round of non-selective growth, putative isolates from which the delivery plasmid had been lost were selected on MM containing 100 mM 2-deoxyglucose. All these isolates were either bald, hygromycin-resistant colonies in which the plasmid had excised leaving the bldM::hyg mutant allele in the chromosome, or sporulating, hygromycin-sensitive colonies in which the plasmid had excised leaving the D-54N or D-54A mutant allele in the chromosome. The presence of the D-54N or D-54A mutation (and the absence of any unintended changes) in 10 sporulating, hygromycin-sensitive isolates from each experiment was confirmed by PCR amplification and sequencing of the entire bldM gene, and representative isolates were designated J2153 (D-54N) and J2154 (D-54A).

Overproduction and purification of BldM, BldM D-54N and BldM D-54A

A 1 kb XbaI–ApaI fragment carrying bldM was blunt-end cloned into HincII-cut pIJ2925 (Janssen and Bibb, 1993) such that, in the resulting plasmid, pIJ6611, the EcoRI site in the polylinker was upstream of bldM. The 400 bp of DNA between the EcoRI site and a unique AvaI site 40 bp downstream of the bldM ATG start codon was replaced with two complementary oligonucleotides (5′-AATTCCATATGACCTCTGTTCTGGTTTGCGACGACTCTCCGCTGGC-3′ and 5′-TCGGGCCAGCGGAGAGTCGTCGCAAACCAGAACAGAGGTCATATG-3′) that introduced an NdeI site overlapping the ATG start codon and also replaced the third, fourth, fifth and sixth codons with synonymous codons commonly associated with genes expressed at high levels in E. coli. The cassette replacement was verified by sequencing the resulting plasmid, pIJ6612. Parallel constructs for BldM D-54N and BldM D-54A were created by subjecting pIJ6612 to PCR mutagenesis as described above for pIJ6605. Finally, the bldM alleles were excised as 0.6 kb NdeI–BglII fragments and cloned into the expression vector pET15b (Novagen) cut with NdeI and BamHI, generating pIJ6613 (BldM), pIJ6617 (BldM D-54N) and pIJ6618 (BldM D-54A).

pIJ6613, pIJ6617 and pIJ6618 were each introduced into E. coli BL21λDE3(pLysS) (Studier and Moffat, 1986), and bldM expression was induced in exponentially growing cells (OD 0.5 at 600 nm) by the addition of 0.1 mM IPTG. The resulting N-terminally His-tagged BldM proteins were purified by nickel affinity chromatography, and their molecular weights were determined by electrospray ionization mass spectrometry (ES-MS). In each case, the observed molecular weight of the protein matched (±5 Da) the predicted molecular weight, and showed that the N-terminal N-formylmethionine had been removed post-translationally. The His-tags were subsequently removed using thrombin.

In vitro phosphorylation assays

[32P]-acetyl phosphate was generated by incubating acetate kinase (2 units; Sigma A6781) in 30 µl of 25 mM Tris-HCl, pH 7.5, 60 mM potassium acetate, 10 mM MgCl2 containing 50 µCi of [γ-32P]-ATP (6000 Ci mmol−1) at 25°C (Gu et al., 1994). [32P]-carbamoyl phosphate was generated by incubating carbamate kinase (2 units; Sigma C5408) in 30 µl of 25 mM Tris-HCl, pH 8.3, 60 mM ammonium carbamate, 10 mM MgCl2 containing 50 µCi of [γ-32P]-ATP (6000 Ci mmol−1) at 37°C. After 15 min of incubation, the [32P]-acetyl phosphate or [32P]-carbamoyl phosphate solution was mixed with 20 µl of 40 mM Tris-HCl, pH 7.9, 200 mM NaCl, 0.2 mM dithiothreitol (DTT), 0.2 mM EDTA containing BldM, BldM D-54N or BldM D-54A (final concentration 10 µM) or NtrC (final concentration 5 µM; kindly provided by Sara Austin, John Innes Centre). The labelling reactions were stopped after 5 min by mixing with 20 µl of loading buffer, and 20 µl of each reaction was applied directly to an SDS-polyacrylamide gel, omitting the usual heat denaturation step. After electrophoresis, the gel was dried and exposed to a Kodak X-OMAT film for 2 h.

Scanning electron microscopy

Scanning electron microscopy was performed as described by Molle et al. (2000).

Acknowledgements

We thank Tracy Nixon, Jeff Stock, Dick Hutchinson, Keith Chater and Mark Paget for helpful discussion, Justin Nodwell, Keith Chater, David Hopwood and Allan Downie for their comments on the manuscript, Sara Austin for the gift of NtrC protein, Rick Evans-Gowing for help with the electrospray ionization mass spectrometry, and Justin Nodwell for independent confirmation of the extracellular complementation experiments. This work was supported by BBSRC grant 83/P07658 (to M.J.B.), by a John Innes Foundation studentship (to V.M.), by a Lister Institute Research Fellowship (to M.J.B) and by a grant-in-aid to the John Innes Centre from the BBSRC.

Ancillary