Keith F. Chater. E-mail email@example.com; Tel. (+44) 1603 452571; Fax (+44) 1603 456844.
whiI is one of several loci originally described as essential for sporulation in Streptomyces coelicolor A3(2). We have characterized whiI at the molecular level. It encodes an atypical member of the response regulator family of proteins, lacking at least two of the residues strongly conserved in the conventional phosphorylation pocket. It is not adjacent to a potential sensor kinase gene. Fifteen mutant alleles of whiI were sequenced, revealing, among others, six mutations affecting conserved amino acids, several frameshift mutations and one mutation in the promoter. The whiI promoter is specifically transcribed by the sporulation-specific σWhiG-containing form of RNA polymerase. Transcription of whiI is temporally controlled, reaching a maximum level coincident with the formation of spores. Further transcriptional studies suggested that WhiI is involved directly or indirectly in repressing its own expression and that of another σWhiG-dependent sporulation-specific regulatory gene, whiH.
The genus Streptomyces comprises mycelial soil bacteria that undergo complex morphological differentiation to permit effective dissemination (Chater, 1998). Long chains of unigenomic spores are produced on aerial hyphae, which grow out of the substrate mycelium after 1 or 2 days of laboratory culture. In the apical part of each aerial hypha, a specific multiple and synchronous cell division process leads to the formation of prespore compartments, which round up, develop a thick wall and acquire a characteristic pigment, which is grey in the case of the genetic model species Streptomyces coelicolor A3(2).
S. coelicolor mutants defective in sporulation were identified because they failed to produce the spore-specific grey pigment and remained white on prolongated incubation (Hopwood et al., 1970). Many of these whi mutants failed to form spore compartments at all, and the genetic analysis of these mutants revealed six ‘early’whi genes necessary for full sporulation septation (Chater, 1972; McVittie, 1974). Some early whi genes have been characterized at the molecular level: whiG (Méndez and Chater, 1987; Chater et al., 1989; Tan et al., 1998) encodes a sporulation-specific sigma factor, σWhiG; the whiH product (Ryding et al., 1998) has homology to some metabolism-related repressors; and whiB (Davies and Chater, 1992) encodes the prototype of a new family of proteins characteristic of actinomycetes (J. Soliveri et al., manuscript in preparation). All these early sporulation genes are necessary for the complete expression of the late sporulation genes, including those of the whiE locus, encoding the polyketide synthase responsible for spore pigment production (Davis and Chater, 1990; Kelemen et al., 1998) and sigF (Potúckováet al., 1995; Kelemen et al., 1996). sigF encodes a sigma factor, σF, specific for some late sporulation events, including the expression of one operon of the whiE genes (Kelemen et al., 1998). This suggests that a network of gene interactions and regulatory events is required for the completion of the sporulation process.
The recent demonstration that the early sporulation gene whiH depends directly on the early sporulation-specific sigma factor σWhiG was an advance in understanding the link between the early and late sporulation events (Ryding et al., 1998). In addition, cytological studies of single and double early whi mutants have suggested that the main influence of whiH on the sequential decisions made en route to sporulation septation is at a rather late stage (Chater, 1975; Flärdh et al., 1999). However, the fact that expression of the late sporulation genes sigF and whiE is not completely abolished in whiH null mutants (Kelemen et al., 1996; 1998) indicates that at least one other regulatory element must be closely involved in activating these late genes. A candidate gene for such a role is whiI. Mutants in whiI have a somewhat whiH-like phenotype, with a significant amount of fragmentation of aerial hyphae, apparently resulting from the formation of occasional sporulation septa (Chater, 1972; McVittie, 1974), and an early analysis of constructed double mutants indicated that whiG, whiA, whiB and whiH are all epistatic to whiI in general morphological phenotype (Chater, 1975). Here, we report the cloning of whiI using a map-based strategy, leading to new evidence of a complex regulatory network for the early stages of sporulation.
Cloning the whiI gene and complementation of 15 putative whiI mutants
In previous experiments, libraries of S. coelicolor DNA in various vectors failed to reveal any clones that complemented whiI mutants [notably the library of Ryding et al. (1998), which did contain clones for several other whi genes]. In case this was because of some unexpected aspect of whiI mutations, such as dominance of the mutant alleles tested, we adopted a new approach that would permit the recognition of whiI-containing DNA irrespective of dominance. This involved the use of a cosmid library (Redenbach et al., 1996) in map-based cloning, taking advantage of the close genetic linkage of whiI to a previously cloned locus cysCD (Lydiate et al., 1988). The library had been constructed in the cosmid Supercos1, which contains a kanamycin resistance determinant selectable in S. coelicolor, allowing direct transformation of S. coelicolor. Homologous recombination between the cloned DNA and the chromosome should give strains diploid for particular chromosomal segments of average length 37.5 kb. The considerable length of the diploid region would make further recombinational interaction between the duplicated sequences quite likely, including the generation of homogenotes. Thus, in the event of whiI−/whiI+ heterogenotes being white because of possible dominance of whiI mutations, whiI+/whiI+ grey homogenotes could be formed. Additionally, loss of an integrated whiI-containing cosmid by a second cross-over during subculture on kanamycin-free medium would be frequent and should often result in replacement of the mutated allele by whiI+.
Accordingly, cosmids from the S. coelicolor cosmid library were used to transform the whiI mutants C17 and C40 in a ‘chromosome walking’ strategy, starting from cosmid 9D7, which contains cysCD, and going in both directions (at the time of these experiments, the orientation of cosmids in this region had not been determined). Primary KanR transformants were patched to MM with and without kanamycin to permit segregation of KanS derivatives. Six overlapping cosmids completely restored both sporulation and grey pigment biosynthesis to the majority of transformants of both whiI mutants. Cosmid 1C3 was selected for further analysis. Several overlapping fragments, subcloned in the integrative vector pDH5, restored wild-type levels of sporulation and grey colour to the C40 mutant. When the smallest complementing subclone, pIJ6404, which contained a 2.2 kb SphI–BamHI fragment, was used to transform other whiI mutants, more than 90% of the transformants were grey and sporulated, indicating that the complete whiI gene was present in the insert of pIJ6404. The remaining 10% of the transformants were white, possibly because of homogenotization.
The insert in pIJ6404 was then cloned in the phage vector KC210, producing phage KC1002. All the 15 whiI mutants isolated in the original collection were fully complemented by KC1002. This result confirmed the provisional allocation of all the mutations to the whiI locus, which was based on linkage analysis in conjugational crosses (Chater, 1972), and showed that all the mutations were recessive. Thus, the failure of previous whiI cloning attempts was probably the result of chance rather than the dominance of mutant alleles.
Sequencing the extremities of the insert in pIJ6404, allowed its localization in the emerging sequence data from the S. coelicolor genome project (http://www.sanger. ac.uk/Projects/S_coelicolor/; cosmid 1C3, accession number AL023702, nucleotides 21438–23630). This fragment comprises the 5′ region of open reading frame (ORF) SC1C3.16c, one complete centrally located ORF (SC1C3.17c) and the 3′ part of ORF SC1C3.18c, all three ORFs being located in the complementary strand of the annotated sequence and orientated towards the origin of replication of the chromosome. Therefore, whiI was assigned to ORF SC1C3.17c.
whiI encodes a response regulator-like protein with some atypical features
Annotated sequence information about the non-coding DNA flanking whiI is shown in Fig. 1A. The putative start codon is a GTG codon located at position 23012, preceded by a purine-rich sequence AGGAGG that could act as a ribosome binding site. Other putative start codons (ATG at position 22949 and a GTG at position 22946) are not preceded by a purine-rich sequence. The stop codon at nucleotide 22352 defined an ORF of 660 nucleotides with a G + C content of 75% and a third position bias of 95%, both values typical of Streptomyces genes (Bibb et al., 1984; Wright and Bibb, 1992). This codon bias starts at the GTG codon postulated to be the translation initiation codon. The region immediately after the coding sequence shows a strong strand bias between purines and pyrimidines: 56 out of 66 nucleotides of the coding strand are pyrimidines. In this region, a heptamer sequence closely resembling the consensus CCG(A/C)TCC is completely repeated eight times, and the first five nucleotides are repeated twice more. A 38-nucleotide sequence consisting of a perfect inverted repeat was found 120 nucleotides downstream of the stop codon.
The putative WhiI protein is similar to members of the response regulator family of transcriptional activators (Fig. 2) (Hakenbeck and Stock, 1996). The two well-characterized response regulators most similar to WhiI are DegU from Bacillus subtilis, which plays an important role in decisions between alternatives on entry into stationary phase (Msadek et al., 1995), and NarL from Escherichia coli, which is involved in nitrate and nitrite regulation of anaerobic gene expression (Stewart and Rabin, 1995) and has been characterized structurally (Baikalov et al., 1996). These proteins belong to the FixJ subfamily of response regulators (Hakenbeck and Stock, 1996), and sequence features characterizing this family can be found in the C-terminal region of WhiI (L-159xxRE, G-173, I-179 and T-189).
Typically, response regulators have an N-terminal domain in which several residues are conserved, forming a phosphorylation pocket directly involved in regulating the transcriptional activator activity of the C-terminal domain, which normally possesses a helix–turn–helix DNA-binding motif (Hakenbeck and Stock, 1996). The conserved residues of the phosphorylation pocket are two adjacent aspartates near the N-terminus of the protein (typically in positions 11 and 12), another aspartate in the middle of the N-terminal domain (typically near position 55), a hydroxylated residue at position 82 (normally threonine, but serine in DegU and NarL) and a lysine residue close to the end of the N-terminal domain, near position 105. In WhiI, after a rather long N-terminus, only one aspartate is present at position 27 and although the central aspartate, which is normally phosphorylated, is conserved at position 69, and there is a serine residue at the position of the conserved threonine or serine, the position normally occupied by the universally conserved lysine is replaced by a threonine residue. The genes flanking whiI do not appear to encode sensor protein kinases.
Sequencing the mutant alleles of whiI
In confirmation of the complementation tests, DNA sequencing revealed mutations in all of the 15 whiI mutants tested (Fig. 1A and Fig. 2). (Mutant allele numbers correspond to the numbers used to denominate the original whiI mutants, i.e. whiI6 is the whiI allele of mutant C6.)
Seven mutants have single amino acid substitutions: whiI122 (Val-24Asn) and whiI62 (Leu-38Arg) have polar residues in positions occupied by hydrophobic amino acids; a highly conserved glycine residue near the conserved Asp-69 is changed in whiI225 (Gly-77Asp); whiI36 has a substitution (Ser-82Pro) in a non-conserved position of the N-terminal domain; in whiI95, the mutation introduces an arginine residue in a position that is normally occupied by a hydrophobic residue (Leu-93Arg); whiI6 (Ser-186Phe) is altered in a highly conserved Ser residue in the putative DNA-binding motif (a second mutation, C→T at position 22 734, which does not change the amino acid sequence, was also found in this mutant); and whiI17 (Val-190Gly) is changed in a highly conserved valine residue in the putative DNA-binding motif.
Six mutants contained frameshifts eliminating the whole C-terminal domain and a variable length of the N-terminal domain: whiI46 has a single nucleotide deletion, which causes a frameshift from Pro57, leading to a shorter product; in whiI229 and whiI244, the sequences GGACGTCC and GGACG, respectively, containing the codon for the conserved and normally phosphorylatable Asp69 (shown in bold), are repeated, causing a frameshift that puts in frame a stop codon nearby; in whiI80, the deletion of one nucleotide and the change of C→T in the adjacent nucleotide cause the immediate termination of the protein at Gly-77; a single nucleotide insertion was found in whiI226, causing a frameshift from the position of Gly-110 onwards and a putative longer aberrant product; in whiI87, the deletion of 10 nucleotides causes a frameshift from His-143 and a shorter aberrant protein.
One mutation, whiI235, is an in frame deletion within the N-terminal domain, with a C→T change at the point of deletion, which removes the residues from Ala-68 to Lys-113, i.e. including the conserved Asp-69 (Fig. 2). The mutation in whiI40 (T→A) was located in the non-coding DNA upstream of whiI (Fig. 1A). Further analysis (see below) showed this to be a critical position in the whiI promoter.
Morphology of the whiI mutants and tentative genotype–phenotype relationships
A tentative molecular classification of the whiI mutants identifies four groups. First, those that lack the WhiI C-terminal putative DNA-binding domain (C46, C80, C87, C226, C229 and C244) should be unable to recognize target sequences in DNA and show the most severe phenotypes. The second class consists of C40: this mutant would have a normal WhiI protein, but the mutation in the promoter is predicted to reduce the levels of WhiI (see below), so it should perhaps show a leaky phenotype. Thirdly, the C6 and C17 mutants, which have single amino acid substitutions in the putative DNA-binding domain, could have an altered affinity for target genes. The fourth group would comprise the mutants altered in the N-terminal domain, C36, C62, C95, C122, C225 and C235, which are expected to be affected either in modulation of WhiI by other putative regulatory factors (kinases or other proteins, small ligands, etc.) or in tertiary or quaternary structure. The mutations in the last two groups would have unpredictable phenotypes.
Using phase-contrast microscopy, the length of the aerial hyphae of the whiI mutants generally appeared to be similar to that of wild-type colonies. Different mutants, however, differed in the structure of the aerial hyphae. All the mutants lacking the C-terminal putative DNA-binding domain (class 1: C46, C80, C87, C226, C229 and C244) and one of the point mutants for this domain (class 3: C17) have a similar phenotype (loosely coiled aerial hyphae with no spores, as illustrated by C46 in Fig. 3C), which we can therefore assume represents the most severe phenotype for the whiI mutants (the null mutant J2450 described in the next section also shows this phenotype). The same phenotype is also shown by some mutants affected in the N-terminal domain (class 4: C95, C122, C225 and C235), suggesting that, in these mutants, the mutation probably locked the protein in an inactive conformation, and indicating that the N-terminal domain of WhiI plays an important role in the activity of the protein. C6 (class 3) and C62 (class 4) had short stems bearing a tightly coiled knot-like apical region (Fig. 3A), indicating that the point mutations in these strains do not completely inactivate WhiI function. C36 (class 4) and C40 (the class 2 promoter mutant) were the only whiI mutants that produced some spore chains (albeit relatively short), and their aerial hyphae were loosely coiled (Fig. 3B). As C40 probably produces a reduced amount of wild-type protein, we tentatively suggest that the whiI36 point mutation, which alters a non-conserved residue, may affect WhiI abundance rather than its regulatory function, perhaps by shortening the mRNA or protein half-life.
Construction of an insertional null mutant of whiI
Plasmid pIJ6405 is a pIJ6404 derivative in which the SmaI site of the polylinker has been removed, so that two adjacent SmaI sites within the whiI gene were the only ones left in the plasmid. A hygromycin resistance cassette (Zalacaín et al., 1986) was inserted between these sites, and the resulting plasmid was introduced into the wild-type S. coelicolor M145. Out of about 200 HygR transformants, seven were ThioS, indicating the probable loss of the original vector (pDH5) and the replacement of the wild-type whiI gene by the whiI::hyg construction. When tested on MM, the HygR ThioS colonies all developed white aerial mycelium, and no spores could be seen. In Southern blotting, DNA samples from all these candidates showed a 4.5 kb PvuII hybridizing fragment, in contrast to the 2.8 kb fragment of the wild-type M145, indicating that effective allelic replacement had taken place.
One of the HygR ThioS candidates, J2450, was analysed further. Protoplasts of J2450 were transformed with pIJ6404 or the cosmid 1C3 (both containing the wild-type whiI gene). In both cases, the transformants recovered the grey pigmentation and the sporulation levels typical of wild-type colonies (Fig. 3E). Phase-contrast microscopy (Fig. 3D) or scanning electron microscopy (SEM) (Fig. 3F) of J2450 colonies revealed examples of some aerial hyphae with moderately long regular coils and others with structures like rings or knots at the tips. Few, if any, constrictions or fragments indicative of sporulation septation were seen in the aerial hyphae. This phenotype resembled that of the class 1 mutants, which lack the C-terminal domain of WhiI (the predicted DNA-binding domain). Subtle differences between the class 1 mutants and the deletion mutant may be caused by differences in genetic background.
Chromosomal condensation and partitioning in the whiI mutants
During sporulation of the multigenomic aerial hyphae, proper chromosome condensation and partitioning have to take place before division into unigenomic spore compartments. It has been shown recently by DAPI staining that, in whiG, whiA and whiB mutants, the chromosome remains in an uncondensed state, with continuous distribution of DNA all along the aerial hyphae. On the other hand, clear deficiencies in chromosomal partitioning were detected in whiH mutants, as the DNA condensed forming well-separated but irregular patches (Flärdh et al., 1999). Because of the somewhat similar phenotypes of whiI and whiH mutants observed by phase-contrast microscopy and SEM (see above; Fig. 3; Ryding et al., 1998) and the regulatory connection between these two genes (see next section), we analysed DNA distribution in the aerial hyphae of different whiI mutants by DAPI staining. In J2450 (Fig. 4), the DNA was continuous along the hyphae, as described by Flärdh et al. (1999) for whiG, whiA and whiB mutants, and in contrast to whiH mutants. Similar patterns of DNA distribution were observed in other whiI single mutants (data not shown). In the case of the C40 promoter mutant, the pattern of the chromosomal distribution in the short spore chains was similar to the wild type (Fig. 4).
Transcriptional analysis of whiI
RNA was isolated from surface cultures of S. coelicolor M145 at different times of cultivation (and therefore at different stages of development), and transcription of whiI was investigated by S1 nuclease mapping (Fig. 5). A major transcription start point was detected. The main band was occasionally accompanied by a weak signal (see Figs 5–7), ≈ 10 nucleotides shorter. We cannot eliminate the possibility that this indicates a secondary transcription start point for whiI. The period of maximum whiI transcription coincided with the first signs of sporulation of the aerial mycelium, consistent with a specific role for WhiI at this developmental stage. This pattern of expression was similar to that observed for whiH (Fig. 5; Ryding et al., 1998), except that no whiI mRNA was detected in those samples containing only vegetative mycelium, even when the films were overexposed.
Using high-resolution S1 mapping and a corresponding sequence ladder as the standard, the major, and probably the only, transcription start point of whiI was localized to the T complementary to the A23087 of the sequence of cosmid 1C3 (EMBL: AL023702), 75 nucleotides upstream of the proposed start codon (Fig. 1A). This T residue is four nucleotides downstream of the sequence GCCGA AGA, which in turn is 16 nucleotides downstream of the sequence TAAA. These two sequences resemble the −10 and −35 regions, respectively, of S. coelicolor promoters dependent on the sporulation-specific σWhiG form of RNA polymerase.
To investigate the possibility that whiI is whiG dependent, S1 nuclease protection assays were carried out with RNA samples isolated from J2400, an M145 derivative carrying a disrupted whiG gene (Fig. 5). No whiI transcript could be detected in any RNA sample from the whiG disruption mutant, even when the autoradiographs were overexposed, whereas transcription of the constitutive hrdB control gene was essentially normal. This showed that σWhiG is necessary for the expression of whiI in vivo and that a whiG mutant contains no other sigma factor capable of directing transcription of whiI.
Purified σWhiG (Tan et al., 1998) and E. coli core RNA polymerase were used in an in vitro run-off transcription assay on two different DNA templates containing the whiI promoter. With both templates, a transcript of the expected size was detected (Fig. 6), demonstrating that the dependence of whiI on whiG is a consequence of direct interaction of σWhiG RNA polymerase with the whiI promoter.
The −10 and −35 sequences of σWhiG-dependent promoters are similar to those recognized in other species by σ factors homologous to σWhiG, i.e. σD in B. subtilis and σFliA in Salmonella typhimurium (Tan and Chater, 1993). In these promoters, the T nucleotide in the −35 box is completely conserved, suggesting that it is a critical position. Therefore, the mutation in C40 (affecting this T) probably decreases the level of expression of whiI40. Recently, the whiH promoter was also shown to be σWhiG dependent (Ryding et al., 1998) and, interestingly, there is extended similarity between the promoters of whiI and whiH, including an inverted repeat (Fig. 1B) centred on position −13 in the whiI promoter.
Transcription from the whiI and whiH promoters is increased in a whiI disruption mutant
Ryding et al. (1998) have shown that whiH expression was detectable much earlier and was stronger and continued for longer in a whiH point mutant than in a whiH+ control strain. This raised the question of whether whiI expression might similarly be subject to autoregulation, and the extended sequence similarity of the whiI and whiH promoters also raised the possibility of cross-regulation between WhiI and WhiH. Transcription analysis was therefore carried out with whiI and whiH mutants (Fig. 7), using hrdB as a control.
Indeed, whiI mRNA was more abundant, detectable earlier and remained detectable for longer in a whiI disruption mutant (J2450). Thus, just as with whiH, there appears to be some direct or indirect autoregulatory mechanism. Interestingly, disruption of whiH also affected whiI transcript levels to a moderate extent. In J2210 (whiH::hyg), fairly abundant whiI mRNA was detected when aerial mycelium was still sparse, and levels then declined slowly over 4 more days. Therefore, it appears that WhiH has a modest repressing effect on whiI transcript abundance.
It also appears from Fig. 7 that there is a reciprocal, and indeed somewhat stronger, cross-regulation of whiH by WhiI. Transcription of whiH in J2450 (whiI::hyg) is earlier, stronger and more persistent than in the wild type (e.g. M145, Fig. 5) and is very similar to that in the whiH::hyg disruption mutant J2210, as well as to that originally described by Ryding et al. (1998) for the whiH119 point mutant.
Characterization of the whiI sporulation-specific regulatory locus
Fifteen out of 50 S. coelicolor whi mutants from the original collection (Hopwood et al., 1970) were mapped to the whiI locus (Chater, 1972). Molecular analysis, described here, showed that a fragment of DNA containing only one complete gene restored the wild-type phenotype to the 15 whiI mutants, and that there are sequence differences in this gene (or its promoter) in all of them. Moreover, a constructed disruptant of this gene produced a phenotype similar to that of some of the whiI mutants, confirming that the gene is indeed whiI. The genes immediately upstream and downstream of whiI do not encode proteins of known function. Interestingly, their only known close homologues are in mycobacteria: these two genes are immediately adjacent to each other in Mycobacterium tuberculosis (Rv2680 and 2681, Cole et al., 1998) and Mycobacterium leprae (EMBL entry U15181, ORFs U1764v and U1764u), reinforcing the absence of direct homologues of whiI from these non-sporulating actinomycetes and, therefore, its likely specificity for sporulation.
whiI encodes a somewhat atypical member of the response regulator family of proteins. Generally, response regulators are associated with sensor protein kinases in prokaryotic two-component systems (Hakenbeck and Stock, 1996) and, in response to some signal, the sensor kinase modulates the activity of the associated response regulator via phosphorylation–dephosphorylation. Very often, the genes for sensor kinases are immediately adjacent to the genes for the cognate response regulators. Neither the genes immediately adjacent to whiI nor any other gene in an interval of 14 kb around whiI showed homology to genes encoding protein sensor kinases. Other genes for response regulators that are not located next to a gene for a cognate sensor kinase include redZ, which encodes a pathway-specific regulator of undecylprodigiosin biosynthesis in S. coelicolor (Guthrie et al., 1998), and spo0A, which encodes a key sporulation regulatory protein in B. subtilis (Hoch, 1995; Stragier and Losick, 1996).
Study of the mutant alleles of whiI may help to understand the mode of action of WhiI, although further experiments will doubtless be needed. Specially significant has been the finding of several point mutations and one in frame deletion in the N-terminal domain. Most of these single amino acid substitutions affect conserved residues, reinforcing the likelihood that the N-terminal domain, which is associated with phosphorylation in typical response regulators, retains functional importance in WhiI. The finding of two mutations in highly conserved residues of the predicted DNA-binding motif in the C-terminal domain also confirms the functional importance of this region.
As the structure of the unphosphorylated NarL has been determined (Baikalov et al., 1996), and NarL falls in the same subfamily of response regulators as WhiI, we attempted some structure comparisons, using a computer-based prediction of WhiI structure generated by the psa program (Stultz et al., 1997). The prediction for NarL coincided with that determined crystallographically. The main observations about the N-terminal region of WhiI were: no β-strand was predicted in the position of amino acids 43–49, equivalent to the β2-strand (residues 32–38) of NarL; and the helices α5 and α6 of NarL (residues 113–125 and 132–141) were replaced by a single shorter α-helix in WhiI (indeed, the WhiI sequence in this region shows no perceptible resemblance to other response regulators and contains three potentially helix-destructive proline residues; see Fig. 2, residues 118–140). Neither of these differences would be strongly predicted to interfere with the formation of a normal phosphorylation pocket. The C-terminal region of WhiI was predicted to be a helix–β-strand–helix, whereas in NarL, a helix–turn–helix motif has been described and was correctly predicted by the psa analysis. However, strong primary sequence similarity to NarL in this region makes it highly likely that it has a DNA-binding function.
How might WhiI activity be regulated?
Most of the genes necessary for Streptomyces sporulation identified so far encode regulatory proteins rather than enzymatic or structural proteins, indicating that the control of sporulation in S. coelicolor is complex. The gene products include sigma factors [whiG, (Chater et al., 1989); and sigF (Potúckováet al., 1995; Kelemen et al., 1996)], a GntR-like protein (whiH, Ryding et al., 1998) and the response regulator WhiI described in this work. Moreover, these proteins are all members of families of proteins that are generally regulated post-translationally.
The absence of some of the conserved residues in the phosphorylation pocket make it possible that the activity of WhiI is not regulated via aspartate phosphorylation–dephosphorylation and may instead be implicated directly in sensing some intracellular signal, either by binding a small ligand, whose concentration could change during development, or by interacting with another regulatory protein. Alternatively, WhiI may indeed be regulated by phosphorylation either at the conserved aspartate 69 or in an unconventional way. Many response regulators can be phosphorylated by small phosphodonors, such as acetylphosphate (Lukat et al., 1992), even when the specifically associated sensor kinase is present and fully functional. This represents an alternative mechanism of in vivo control of the activity of response regulators. Spo0A in B. subtilis (Hoch, 1995; Stragier and Losick, 1996) receives the phosphate group through a phosphorelay signal transduction cascade, in which several sensor kinases integrate different signals by transferring a phosphate group to a common response regulator (Spo0F), which phosphorylates Spo0A via the Spo0B phosphotransferase. Ultimately, the concentration of phosphorylated Spo0A is the factor that decides whether or not the cell becomes committed to sporulation. Perhaps WhiI has a similar role in integrating multiple signal inputs, some of which might involve the products of other whi genes as equivalents of other spo0 genes of B. subtilis.
The σWhiG regulon in the sporulation of S. coelicolor
One interesting finding of this work is that whiI, like the recently characterized whiH gene (Ryding et al., 1998), depends directly on the sporulation-specific sigma factor σWhiG. This indicates that, in S. coelicolor, sporulation may be regulated by a complex network of genetic interactions, rather than by a simple linear cascade of gene activation. Most probably, WhiI and WhiH each control several further genes necessary for the completion of sporulation. Sporulation genes have been classified as ‘early’ (those needed for the formation of the sporulation-specific septa) and ‘late’ (necessary for the maturation of spores and the synthesis of the spore pigment), but the molecular mechanism that links the early and the late genes is poorly understood. One possibility for this connection is whiH because, in whiH mutants, the expression of the late genes sigF and part of the whiE genes is very considerably reduced, although it is not completely abolished. whiI is a second alternative, in view of the total absence of sigF and whiE transcription in the whiI mutant C17 (G. H. Kelemen, personal communication). Future studies will include examination of the sigF and whiE promoter regions as putative targets for WhiH and WhiI.
A σWhiG-dependent gene dispensable for sporulation has also been described: orfTH4, encoding a ProX-like protein (Tan et al., 1998). The promoters of whiI, whiH and orfTH4 share homology in the −35 and −10 boxes, but the additional similarity in the spacer region in whiI and whiH promoters (Fig. 1B) is not present in the orfTH4 promoter. We have also shown that transcription of whiI and whiH in the wild type follows a similar timecourse, both genes being switched on and off at about the same time. This could reflect the binding of a sporulation-specific protein to these sequences (below, we discuss whether this protein could be WhiH or WhiI).
The apparent cross-regulation between WhiI and WhiH
In view of the similar expression pattern and whiG dependence of whiI and whiH, we analysed the effects of mutations in each gene on the expression of both. In addition to confirming and extending the previous evidence that WhiH shows direct or indirect autorepression, the new results can be summarized as three points: first, WhiI seems to autorepress its own expression slightly; secondly, the absence of WhiI considerably derepresses the expression of whiH; and thirdly, the expression of whiI is affected to some extent by the absence of WhiH. There is a possible trivial explanation of these observations: aerial mycelium was detectable earlier in the mutant cultures used for RNA preparation than in the wild type and does not change into spore chains, so that we should expect to detect whiH and whiI mRNA earlier, and perhaps later as well, in the mutants. However, the comparatively high abundance of whiH mRNA in the early samples of both kinds of mutants and of whiI mRNA in the whiI mutant, when aerial mycelium was still very sparse, makes this trivial explanation inadequate, except perhaps in the rather weak effects of whiH mutations on whiI expression.
Assuming, then, that the regulatory effects are real, the two gene products, WhiI and WhiH, might directly mediate these regulatory effects (Fig. 8) in a pattern somewhat reminiscent of the interplay between the response regulator Spo0A and the repressor AbrB in sporulation of B. subtilis (Fujita et al., 1998). We further propose (enlarging on the suggestions of Ryding et al., 1998 for whiH ) that both proteins are modified by different signals activated during aerial mycelium development (probably around the time when aerial growth stops). In response, autorepression and cross-repression would be relieved and more of both proteins made. As both proteins are required for sporulation, we propose that their modified forms (i.e. resulting from interaction with the proposed cognate signals) activate certain target genes, either by concerted action at particular promoters or by independent action on different promoters.
What might be the targets for WhiI?
The mutually repressive effects of WhiI and WhiH suggest that their promoters may well contain binding sites for one or both proteins. The inverted repeat sequences overlapping the −10 regions of both promoters and further extended similarities, as well as another inverted repeat around the −30 region of whiHp, are candidates for such binding sites. These promoters, as well as those of sigF and whiE (see above), will provide material for investigation of WhiI and WhiH DNA–protein interactions. It will also be of interest to find out whether constructed whiI–whiH double mutants show still greater transcription of the two promoters (as would be predicted if they function independently and additively) or whether the single and double mutants have the same levels of expression.
Strains, plasmids, phages and culture conditions
S. coelicolor A3(2) strains used were: 1147 (A3(2) wild type, prototrophic, SCP1+ SCP2+ Pgl+) and M145 (prototrophic, SCP1− SCP2− Pgl+) (Hopwood et al., 1985). The 15 whiI mutants (C6, C17, C36, C40, C46, C62, C80, C87, C95, C122, C225, C226, C229, C235 and C244) were originally derived from strain 1147 (Hopwood et al., 1970; Chater, 1972). The following disruptant mutants were derived from strain M145: J2450 carries a whiI::hyg disruption in the chromosome (this work); J2210 carries a whiH::hyg disruption (Ryding et al., 1998); and J2400 contains a whiG::hyg disruption (Flärdh et al., 1999). S. lividans 1326 (S. lividans 66, Lomovskaya et al., 1972) was used for phage propagation. The non-methylating E. coli strain ET12567 (MacNeil et al., 1992) was used to prepare DNA that could bypass the methyl-sensing restriction system of S. coelicolor (Kieser and Hopwood, 1991), and E. coli DH5α (Hanahan, 1983) was used for routine subcloning.
Several cosmids that hybridized with the AseI-B fragment of the S. coelicolor chromosome (Redenbach et al., 1996) were used in attempts to complement whiI mutants, and one of the positives (cosmid 1C3) was selected to localize the whiI gene more precisely using the integrative plasmid pDH5 (Hilleman et al., 1991) as a vector for subcloning and complementation experiments. A pDH5 derivative, pIJ6404, containing whiI as a 2.2 kb SphI–BamHI fragment, was used in complementation tests.
Media and conditions for Streptomyces culture and protoplast preparation were as described by Hopwood et al. (1985). S. coelicolor whiI mutants were cultivated on MM [minimal medium containing 0.5% (w/v) mannitol as carbon source]. Protoplasts were transformed with cosmids and integrative plasmids using alkali-denatured DNA (Oh and Chater, 1997) and regenerated on the rich medium R2YE. Transformants were restreaked onto MM containing the appropriate antibiotics. For DNA extraction, Streptomyces strains were cultivated in YEME liquid medium supplemented with 34% sucrose, 0.5 g l−1 MgCl2, 5 g l−1 glycine, l-proline, l-arginine and l-cysteine (each at 75 mg l−1), 100 mg l−1l-histidine and 15 mg l−1 uracil.
The 2.2 kb SphI–BamHI fragment containing whiI was cloned in the vector pIJ2925 (Janssen and Bibb, 1993) and excised as a BglII fragment before cloning in the phage KC210 (att+, thiostrepton resistant; C. J. Bruton, personal communication) to produce the phage KC1002. A dense mycelial suspension of each of the whiI mutants was inoculated onto an R2YE plate and, after drying, two 15 μl spots of a KC1002 high-titre suspension were placed on the plate. Plates were incubated for 5 days, then replicated onto MM containing thiostrepton and restreaked in order to observe the phenotype of the transformants.
E. coli strains were transformed by electroporation. Cells were grown at 37°C in LB to an optical density of 0.5 U at 600 nm, then kept on ice for 30 min before being collected by centrifugation at 4°C, washed twice in cold water, once in cold 10% glycerol and finally resuspended in 10% glycerol; samples of 50 μl of competent cells were quickly frozen in a dry ice–ethanol bath and stored at −70°C. For each transformation, a sample was allow to thaw in ice, mixed with the DNA or ligation mix and transferred to a 0.2 cm electroporation cuvette (Bio-Rad). A 2.5 kv pulse was applied to the cuvette in a Bio-Rad gene pulser. Cells were regenerated in LB at 37°C and plated on appropriate selective media.
For phase-contrast microscopy of Streptomyces, strains were cultivated on MM for up to 5 days at 30°C. Then, coverslips were touched against the top of the colonies, and the impression preparations obtained were observed in a Zeiss Axiophot microscope. Photographs were taken using Kodak Technical pan film. For fluorescence microscopy of 4′,6-diamidino-2-phenylindole (DAPI)-stained material and for scanning electron microscopy, cultures or colonies were processed and examined as described by Flärdh et al. (1999).
DNA techniques and sequence analysis
Streptomyces DNA was extracted as described in procedure 3 of Hopwood et al. (1985), modified for small-scale preparation. Phage cloning was performed as described in procedure II of Hopwood et al. (1985). E. coli plasmid preparation, DNA digestion, ligation, agarose gel electrophoresis and DNA blotting were as described by Sambrook et al. (1989). For DNA hybridization, probes were radiolabelled with [α-32P]-dATP using random hexanucleotides and the Klenow fragment of E. coli DNA polymerase (Boehringer Mannheim). For sequencing, the mutant alleles of the 15 original whiI mutants, the templates containing the whiI gene and its promoter were prepared and sequenced as described by Ryding et al. (1998).
Inocula of 108 cfu of spores of the wild-type M145 strain or a dense mycelial suspension of the whi mutants were spread on cellophane membranes placed on MM plates and incubated at 30°C. Mycelium was harvested at different time points (normally 18, 24, 36, 48, 72, 96 and 120 h), frozen quickly in liquid nitrogen and ground until a fine white powder was obtained. Ground mycelial samples were kept frozen at −20°C until the time course was completed. Then, RNA was extracted using the acid phenol procedure described by Tsui et al. (1994). S1 protection assays were performed using the hrdB probe as a control, as described by Kelemen et al. (1996), except that the probe was generated by polymerase chain reaction (PCR) as described by Ryding et al. (1998). The probe to detect whiI transcripts was generated by PCR using the radiolabelled oligonucleotide 5′-GACGGTGGAA CGGACGCGCG-3′ (nucleotides 22911–22930) and the unlabelled oligonucleotide 5′-GGGTCCGCACGTCCGGAGGA-3′ (complement of nucleotides 23251–23270). The first of these oligonucleotides was used to produce the sequence ladder used as a standard to locate the transcription start site. The probe to detect whiH transcripts was generated using the upstream oligonucleotide 5′-TCGTACCGCTCGTACAGAAG GTCTGG-3′ and the radiolabelled downstream oligonucleotide 5′-GCGTACGGGTAGCGGTCGAGTTCGGG-3′. In the experiments shown in Figs 5 and 7 (except those involving J2400), whiI and whiH probes were added in a single tube.
In vitro run-off transcription on DNA templates using purified σWhiG and E. coli core RNA polymerase was performed as described by Tan et al. (1998). The template containing the whiI promoter was generated by PCR, using the oligonucleotides described above, and digested with NcoI or StuI to produce templates of 320 and 254 nucleotides, respectively, in which the extreme of the fragment is located 137 and 71 nucleotides downstream of the putative transcription start site.
This work was supported by the Commitment and Development programme of the BBSRC (grant CAD 04380). J.A.A. was also partially supported by a postdoctoral grant form the Spanish Ministerio de Educación y Cultura (PF 98). We acknowledge the Sanger Centre Streptomyces coelicolor genome sequencing team for making their data available to the whole research community. We are grateful to Celia J. Bruton for providing the phage KC210, Helen M. Kieser for providing cosmids and whiI mutants, Eriko Takano for providing some subclones of cosmid 1C3, and Kim Findlay for invaluable instruction and help with the electron microscope. We acknowledge Gabriella H. Kelemen for thoughtful discussions, David Hopwood, Tobias Kieser, Martin Drummond and Ray Dixon for critical reading of the manuscript, and three anonymous reviewers for comments on the relationship of WhiI to other response regulators.
*Present address: Department of Physiological Sciences, University of Newcastle upon Tyne, Newcastle NE2 4HH, UK.