Molecular Microbiology (1992) 6(19), 2797-2804

Transcriptional regulation of the redD transcriptionalactivator gene accounts for growth-phase-dependentproduction of the antibiotic undecylprodigiosin inStreptomyces coelicolor A3(2)

E. Takano, H.C. Gramajo, E.Strauch,†N. Andres,‡J.White and M.J. Bibb*John Innes Institute, John Innes Centre, Colney Lane,Norwich NR4 7UH, UK.

Summary

Transcription of redO, the activator gene required forproduction of the red-pigmented antibiotic undecyl-prodigiosin by Streptomyces coelicolor A3(2),showed a dramatic increase during the transitionfrom exponential to stationary phase. The increase inredO expression was followed by transcription ofredX, a biosynthetic structural gene, and the appear-ance of the antibiotic in the mycelium, and coincidedwith the intracellular appearance of ppGpp. However,ppGpp production elicited either by nutritional shift-down of, or addition of serine hydroxamate to, expo-nentially growing cultures had no stimulatory effecton redO transcription. The presence of redO on amulticopy plasmid resulted in elevated levels of theredO transcript and production of redX and undecyl-prodigiosin during exponential growth; the normalgrowth-phase-dependent production of undecyl-prodigiosin appeared to be mediated entirely throughthe redO promoter, which shows limited similarity tothe consensus sequence for the major class ofeubacterial promoters.

Introduction

Membersof the genus Streptomyces are notable for theirability to produce a vast array of secondary metabolites.Many of these compounds are used as antibiotics orpharmacologically active agents. In liquid medium, anti-biotic production is generally limited to stationary phase,or to cultures growing at low growth rates (Demain et al.,1983); in agar-grown cultures, it appears to be temporally

Received 22 April, 1992; revised and accepted 15 June, 1992. Presentaddresses: †Robert Koch Institut des Bundesgesundheitsamtes, Nor-dufer 20, 1000 Berlin 65, Germany; ‡Hoechst AG, Pharma Qualitätskon-trolle Bioanalytik, 6230 Franfurt-Main, Germany. *For correspondence.Tel. (0603) 52571; Fax (0603) 56844; E-mail BIBB@UK.AC.AFRC.JII.

correlated with the onset of morphological differentiation,and some genes are required for both processes, imply-ing at least some common elements of genetic control(Chater, 1989; Hopwood, 1988). Although repression byexcess phosphate or easily assimilated sources of carbonor nitrogen appears to limit production of some antibiotics(Demain, 1989), growth-rate, and potentially growth ratecontrol, may playa general role in determining the onsetof secondary metabolism. In this context, it is interestingto note the correlation, albeit incomplete, between anti-biotic production and the appearance of [p]ppGpp in sev-eral streptomycetes (Ochi, 1986; 1987a,b; 1990; Strauchet al., 1991; Kelly et al., 1991), and the apparent ability of[p]ppGpp to reduce growth rate in Escherichia coli(Sarubbi et al., 1988; Schreiber et al., 1991).

Streptomyces coelicolor A3(2) is by far the most stud-ied streptomycete in terms of genetics. It produces atleast four chemically distinct antibiotics: actinorhodin,undecylprodigiosin, methylenomycin and a calcium-dependent antibiotic (Hopwood, 1988). We have charac-terized the stringent response and the production ofppGpp in this species (Strauch et al., 1991); we noted apartial correlation between ppGpp synthesis and tran-scription of actIII, one of the biosynthetic genes for acti-norhodin (Hallam et al., 1988). Here, we focus onundecylprodigiosin, a red-pigmented tripyrrole antibioticmade by the same species (the red pigmentation of S.coelicolor A3(2) is due to a mixture of biosyntheticallyrelated prodigionines, of which undecylprodigiosin is themajor component (Tsao et al., 1985); we refer to the mix-ture as undecylprodigiosin). Genetic studies indicate thatthere is a cluster of at least 18 genes (probably more)involved in undecylprodigiosin production (Rudd andHopwood, 1980; Feitelson etal., 1985; Coco etal., 1991),and the entire biosynthetic cluster has been cloned on aDNA fragment of approximately 35.7 kb (Malpartida et al.,1990). Towards, if not at, one end of the cluster lies a pre-sumptive pathway-specific activator gene, redO. Thefunction of this gene was predicted from the inability ofredO mutants to co-synthesize undecylprodigiosin withother red mutant classes (Rudd and Hopwood, 1980; Fei-telson et al., 1985); from the lack of expression of theredE and/or redF genes required for O-methyltransferase

2798 E. Takano et al.

activity in redO mutants (Feitelson et al., 1985); and fromthe ability of extra cloned copies of redO to elicit overpro-duction of undecylprodigiosin in S. coelicolor A3(2)(Narva and Feitelson, 1990) and Streptomyces lividans(Malpartida et al., 1990), a close relative of S. coelicolorA3(2) in which little undecylprodigiosin is usually made.The nucleotide sequence of redO has been determined(Narva and Feitelson, 1990).Here, we define the in vivo redO transcriptional start

site, and study the temporal and growth-phase transcrip-tion of redO and redX, a structural gene for an early stepin the biosynthetic pathway (Guthrie and Chater, 1990;Malpartida et al., 1990). We assess the possible correla-tion between transcription of redO and redX, and the pro-duction of ppGpp, which was elicited in exponentiallygrowing cultures either by amino acid depletion (nutri-tional shiftdown) or by addition of serine hydroxamate, aseryl tRNA synthetase inhibitor that induced lower levelsof ppGpp. We also investigated the effect of increasedlevels of redO expression during exponential growth onredX transcription and on undecylprodigiosin biosynthe-sis.

Results

Undecylprodigiosin production in liquid culture

In minimal medium supplemented with 0.2% w/vcasamino acids, S. coelicolor A3(2) strain M145 grewexponentially with a doubling time of 2.2 h, and made arapid transition into stationary phase, approximately 12 hafter inoculation, at an OD450 nm of about 2 (approximately1 mg ml-1 dry weight). Undecylprodigiosin production wasreadily detected as soon as the culture entered stationaryphase. The low number of doublings (approximately five)limited the size of mycelial clumps, and may have servedto minimize physiological heterogeneity that might haveresulted from nutritional or oxygen limitation. Increasingthe amount of glucose (fivefold) or phosphate (10-fold)had no effect on growth, but removal of the (NH4)2S04 ledto a reduction in final biomass equivalent to the growthobserved in nitrogen-limited minimal medium, indicatingthat the casamino acid-supplemented cultures were alsonitrogen-limited. Supplementation with higher concentra-tions of casamino acids did not increase the growth rate,but did increase the final yield of mycelium. In minimalmedium lacking casamino acids, M145 grew with adoubling time of about 4 h, and entered stationary phasein a less synchronous manner at an OD450 nm of about0.6; replacement of glucose with the same molar concen-tration of arabinose, fructose or galactose resulted in alower growth rate and lower final OD450 nm (about 0.2 forgalactose and fructose). In the absence of casaminoacids, undecylprodigiosin production occurred upon entry

into stationary phase, regardless of the carbon sourceused, although the amount produced was much lowerthan in the presence of casamino acids. Given the morerapid transition from exponential to stationary phase, thehigher levels of synthesis of undecylprodigiosin, and ourdesire to carry out nutritional shiftdowns, all subsequentexperiments used minimal medium supplemented with0.2% w/v casamino acids.

S1 nuclease mapping of the redD transcriptional start site

An RNA-protected fragment was readily detected withRNA isolated from cultures of S. coelicolor A3(2) M145producing undecylprodigiosin (Fig. 1A). Transcripts corre-sponding in size and orientation to this apparent pro-moter, and to the four oppositely orientated promoters(pr1-4) identified by Narva and Feitelson (1990), werealso observed in in vitro transcription studies using theNael fragment shown in Fig. 1B and RNA polymerase iso-lated from a transition phase culture of M145 (data notshown), providing further evidence that the RNA-pro-tected fragment represented an in vivo transcriptionalstart site. It is preceded by sequences in the -35(ATGACG) and -10 (CACGAT) regions (Fig. 1B) thatshow limited similarity to the consensus sequences forthe major class of eubacterial promoters (TTGACA andTATAAT, respectively: Hawley and McClure, 1983). Thisstart site lies downstream of the previously assignedtranslational start codon of redO (Narva and Feitelson,1990); based on these results, as well as FRAME analysis(Bibb et al., 1984) of the redO sequence and the similarityof the putative redO product to the activators for the acti-norhodin and daunorubicin pathways (the products ofactII-ORF4 (Fernandez-Moreno et al., 1991) and dnrl(Stutzman-Engwall et al., 1992), respectively), the GTGlocated at position 700 (Fig. 1B) probably represents theredO translational start codon, yielding a predicteduntranslated leader sequence of 199 nucleotides and aprotein of 29.6 kDa. The GUG codon would be precededby a sequence (GGGGG) that is complementary, allowingone G:U base pair, to that found close to the 3' end of the16S rRNA of S. coelicolor A3(2) (CCUCC; (Baylis andBibb, 1987)), and that is presumed to be involved in ribo-some binding.Analysis of sequences upstream of redO had identified

the 5' end of an oppositely orientated open reading frameof unknown function that appeared to be transcribed fromfour promoters, pr1-4 (Narva and Feitelson, 1990;EMBL/GenBank/DDJB Nucleotide Sequence DataLibraries Accession Number M29790). Recent searchesof the translated EMBL Nucleotide Sequence DataLibrary indicate that the product of this gene is homolo-gous (32% and 56% amino acid sequence identity andsimilarity, respectively, over the available 183-amino-acid

Undecylprodigiosin production in Streptomyces coelicolor A3(2) 2799

Fig. 1. A. High-resolution S1 nuclease mapping of the transcriptional start site of redO. redO, the S1-nuclease protected fragment derived from hybridizationof the 1.3 kb NdeI-ClaI probe uniquely labelled at the 5' end of the ClaI site to RNA isolated from a culture producing undecylprodigiosin; T+C, A>C, G areMaxam and Gilbert sequence ladders derived from the same end-labelled fragment; SM, end-labelled HpaII-digested pBR322 size marker. The most likelytranscriptional start point is indicated by the asterisk adjacent to the nucleotide sequence of the ladders.B. Nucleotide sequence of the redO promoter region. The sequence of a 768 bp NaeI fragment (Narva and Feitelson, 1990) is shown. Probable trans-lational start codons and potential ribosome-binding sites are underlined. The 5' ends of the redO mRNA identified in these studies, and of the oppositelyorientated transcripts (pr1-4) identified by Narva and Feitelson (1990), are indicated by open circles, and the associated arrows indicate the direction oftranscription; putative -1 0 and -35 regions for redO are underlined. Inverted repeats are indicated by converging arrows. The ClaI site (5'-A T'CGAT-3')used for S1 nuclease mapping of redO is shown at position 726.

sequence) to the product of the trkA gene of E. coli(EMBL Release 29, Accession Number X52114), which isapparently a membrane protein involved in K+ transport.Whether the S. coelicolor A3(2) gene plays a role in unde-cylprodigiosin production, or possibly its export, remainsto be determined.

Transcriptional activation of redD and redX during thetransition from exponential to stationary phase

To assess the growth-phase expression of redO andredX, S1 nuclease protection experiments were carriedout using RNA from exponential and stationary phase cul-tures. Although a low level of redO transcription wasobserved during exponential growth, it showed a markedincrease during the transition into stationary phase (Fig.2B); transcription of redX, undetectable during exponen-tial growth, occurred on entry into stationary phase, corre-lating with the appearance of the antibiotic (undecylpro-digiosin is strongly hydrophobic and remains cell-associated; Fig. 2C).

ppGpp and transcriptional activation of red D and redXduring exponential growth

ppGpp was detected towards the end of exponential

growth of M145, and showed maximal levels on entry intostationary phase, correlating with the appearance ofundecylprodigiosin in the culture (Fig. 2A). To assesswhether ppGpp might playa role in triggering the onset ofantibiotic production, ppGpp production was induced inexponentially growing cultures, either by nutritional shift-down or by addition of serine hydroxamate, and the effecton redO and redX transcription was monitored by S1nuclease mapping. Nutritional shiftdown at an OD450 nm ofabout 0.5 led to the production of large intracellular con-centrations of ppGpp (a peak level of approximately200 pmol mg-1 dry weight). The culture continued to growat a much reduced rate, presumably utilizing (NH4)2S04as nitrogen source, before entering stationary phase6-7 h after shiftdown (Fig. 3A). No increase in redO tran-scription was detected until 4 h after shiftdown (Fig. 3B),and even then it was considerably less than that observedin a normal growth curve (Fig. 2B). Transcription of redXand undecylprodigiosin production were not detected,perhaps because of the lower levels of redO transcription.Addition of serine hydroxamate at 25 mM and 50 mM atan OD450 nm of 0.35 resulted in lower intracellular ppGppconcentrations (peak levels of 47 and 75 pmol mg-1 dryweight, respectively) and continued exponential growth,but at much reduced rates (doubling times of 5.1 hand

2800 E. Takano et al.

Fig. 2. A. Growth curve of, and ppGpp production by, S. coelicolor A3(2) strain M145.to, doubling time during exponential growth.Band C. S1 nuclease mapping of the redO (B) and redX (C) transcripts in RNAsamples isolated from the culture shown in Fig. 2A at the times indicated.For A, Band C, the shaded boxes labelled RED denote the presence of undecyl-prodigiosin in the mycelium. For Band C: EXP and STAT indicate the exponentialand stationary phases of growth, respectively, and the shaded area between themdenotes the transition phase; SM, end-labelled HpaII-digested pBR322 size marker.

17 h, respectively). Transcription of redO and redX, andundecylprodigiosin synthesis, occurred in the culturetreated with 25 mM serine hydroxamate as soon as itentered stationary phase, 11 h after maximal ppGpp pro-duction (data not shown),

Multiple copies of redD result in undecylprodigiosinproduction during exponential growth

The stationary phase production of undecylprodigiosinmight reflect transcriptional regulation of redO alone, oradditional limitations imposed on the expression of thebiosynthetic structural genes during exponential growth,either at the level of their transcription and/or translation,or by the inhibition or inactivation of the enzymes theyencode. To assess this possibility, a 2.4 kb BglII fragmentcontaining redO was cloned into the multicopy (approxi-mately 150 copies per chromosome) plasmid pIJ487(yielding pIJ6014) in M145, and the effects on transcrip-tion of redO and redX, and on undecylprodigiosin produc-tion, were assessed. The growth characteristics ofM145(p1J6014) did not differ significantly from those ofM145; however, redO transcripts were readily detected at

the earliest point in exponential growth from which suffi-cient RNA could be isolated for analysis, and were pre-sent in amounts similar to those observed in stationaryphase cultures of M145 (Fig. 4A). The level of redOtran-script increased during exponential growth and continuedto rise in stationary phase. Elevated levels of the redXtranscript were also observed (Fig. 4B), and undecyl-prodigiosin was readily detectable in mid-exponentialphase. After 65 h, the yield of undecylprodigiosin fromM145(p1J6014) (6.1 µg ml-1) was 5.5 times higher thanthat from M145 (1.1 µg ml-1).

Discussion

The production of antibiotics in stationary phase, or at lowgrowth rates, might reflect the consumption of a nutrientthat represses or inhibits synthesis during growth. We donot believe this to be the case for undecylprodigiosin inthese studies for the following reasons. (i) The stationaryphase production of undecylprodigiosin in minimalmedium severely limited in growth by different carbonsources (e.g. galactose and fructose) strongly suggeststhe lack of ammonium or phosphate repression and/or

Undecylprodigiosin production in Streptomyces coelicolor A3(2) 2801

Fig. 3. A. Growth curve of S. coelicolorA3(2) strain M145 with (,0.) andwithout (0) nutritional shiftdown. The shaded box labelled 0 REDdenotes the presence of undecylprodigiosin in the mycelium of the controlculture (0), which grew exponentially with a doubling time of 2.2 h; SO,point of nutritional shiftdown; ▲, time points at which samples were takenfrom the culture subjected to nutritional shiftdown for RNA isolation.B. S1 nuclease mapping of the redO transcript in RNA samples from theculture subjected to nutritional shiftdown (,0. in Fig. 3A) at the timesindicated (▲ in Fig. 3A); STAT, stationary phase; SM, end-labelled HpaII-digested pBR322 size marker.

inhibition. (ii) The initial levels of ammonium and phos-phate were well below those observed by Hobbs et al.(1990) to reduce undecylprodigiosin production signifi-cantly. (iii) Since undecylprodigiosin was not made duringexponential growth in minimal medium, there is no reasonto suspect that supplementation ,with casamino acids,necessary for studies of nutritional shiftdown, repressedor inhibited production. (iv) Increasing the amount ofglucose (fivefold) or phosphate (10-fold) had no signifi-cant effect on growth or on the stationary phase onset ofundecylprodigiosin production, further suggesting that thelatter is not repressed or inhibited by these metabolites atthe concentrations normally used. These observationsare consistent with the notion that growth rate per se, orthe cessation of growth, is important in determining theonset of undecylprodigiosin production, at least underthese conditions.Transcription of redD in S. coelicolor A3(2) M145

increases dramatically as the culture makes the transitionfrom exponential to stationary phase, and transcription ofredX and the production of undecylprodigiosin clearlyoccur after growth has finished. This differs from anearlier report (Hobbs et al., 1990) in which undecylpro-digiosin synthesis occurred during growth; however, inthat work, growth does not appear to have been exponen-tial, but rather was linear, and was much slower than inour experiments, again consistent with a potential role forgrowth rate in determining the onset of antibiotic biosyn-thesis. The results obtained with redX are similar,although not identical, to those obtained by Guthrie andChater (1990), who used redX::xyIE fusions to detect

Fig. 4. S1 nuclease mapping of the redO (A) and redX (B) transcripts inRNA samples isolated from a culture of S. coelicolor A3(2) M145(pIJ6014) at the times indicated. The shaded boxes labelled RED denotethe presence of undecylprodigiosin in the mycelium; EXP and STATindicate the exponential and stationary phases of growth, respectively,and the shaded area between them denotes the transition phase; SM,end-labelled HpaII-digested pBR322 size marker; M145-STAT indicatesthe use of RNA from a stationary phase culture of M145 (equivalent to the18 h time point in Fig. 2B)

2802 E. Takano et al.

redX transcription during a relatively long period of transi-tion into stationary phase after inoculating cultures withhomogenized mycelium; the use of such inocula, ratherthan freshly germinated spores, may have led to largermycelial pellets and therefore to increased physiologicalheterogeneity, perhaps explaining the departure fromexponential growth towards the end of rapid growth(although it should be noted that a different medium wasused that might equally be responsible).Increasing the copy number of redO resulted in higher

levels of the redO and redX transcripts, undecylpro-digiosin production in exponential growth, and a consider-able increase in the yield of the antibiotic. Thus the onlylimitation to undecylprodigiosin synthesis, at least underthese conditions, appears to be the availability of enoughof the RedO activator. Increases in undecylprodigiosinproduction after introduction of cloned copies of redO intoS. coelicolor A3(2) were noted also by Malpartida et al.(1990) and Narva and Feitelson (1990), although it wasnot possible to deduce whether this reflected earlierexpression of the pathway genes, or their expression athigher levels once growth had ceased; in our studies, theformer clearly occurs. Similar increases in antibiotic pro-duction were observed with extra copies of the pathway-specific activator genes for actinorhodin (Fernandez-Moreno et al., 1991), daunorubicin (Stutzman-Engwall etal., 1992), and streptomycin (Ohnuki et al., 1985); asnoted by Chater (1990), extension of this principle toindustrial fermentations could potentially achieve equallydramatic improvements in antibiotic titre, particularly withunimproved strains. The reason for the continuousincrease in the level of the redO transcript during growthof M145(pIJ6014) (Fig. 4A) is unknown, but may reflect apossible increase in plasmid copy number, or possiblychanges in supercoiling that might result in elevatedlevels of expression of the now plasmid-borne redO pro-moter.Our results are consistent with the notion of a role for

growth rate, or perhaps more specifically the cessation ofgrowth, in determining the onset of antibiotic production.Although antibiotic biosynthesis has been observed inchemostats at low growth rates (see Bushell (1989) forreferences), this might reflect physiological heterogeneitywithin mycelial clumps, and production by cells in the inte-rior that have stopped growing. In this context, it is inter-esting to note the apparent ability of ppGpp to reducegrowth rate in Escherichia coli (Sarubbi et al., 1988;Schreiber et al., 1991), and the apparent correlation ofppGpp and antibiotic biosynthesis, albeit incomplete, pre-viously observed in several streptomycetes (Ochi, 1986;1987a,b; 1990; Strauch et al., 1991; Kelly et al., 1991).Although we saw a correlation between ppGpp synthesisand transcription of redO during the transition of a normalculture into stationary phase, this was not observed after

nutritional shiftdown or after addition of serine hydrox-amate. These results, and the observations of Bascaránet al. (1991), suggest that an increase in the level ofppGpp is not a sufficient physiological signal for the acti-vation of transcription of antibiotic biosynthetic genes.It is difficult to know whether the low level of redO tran-

scription during exponential growth represents a basallevel of expression from all cells, or reflects physiologicalheterogeneity within clumps of mycelium, with transcrip-tion exclusively from those that are growth-limited; theresults obtained with the multi-copy construct pIJ6014 areconsistent with either interpretation. Although the ele-vated levels of the redO transcript observed with this multicopy plasmid might reflect titration of a putative redOrepressor, there is no genetic evidence to suggest theexistence of a negatively acting regulatory gene.Narva and Feitelson (1990) failed to detect a redDtran-

script by S1 nuclease mapping in experiments with RNAfrom 2-3 d cultures. Albeit with a different medium, thiswould correspond to a point in our experiments at whichthe level of the redO transcript had declined to barelydetectable levels (data not shown). redO is transcribedfrom a promoter with limited similarity to the consensussequence for the major class of eubacterial promoters.Although the preliminary experiments with RNA poly-merase might indicate a role for a minor holoenzyme con-taining an alternative sigma factor in redO activation (seeButtner (1989) for a review of RNA polymerase hetero-geneity in Streptomyces), the transcripts we observed invitro might represent a basallevel of redO transcription bythe major holoenzyme in the absence of a positively act-ing regulatory molecule that might be required for ele-vated activity in vivo.

redO is homologous to the pathway-specific activatorgene (actII-ORF4) required for actinorhodin biosynthesisin the same host (Fernandez-Moreno et al., 1991), to theactivator gene (dnrl) for daunorubicin production in Strep-tomyces peucetius (Stutzman-Engwall et al., 1992), andto the 5' end of the putative pleiotropic regulatory geneafsR of S. coelicolor A3(2) (Horinouchi et al., 1990). Align-ment of the amino acid sequences they encode fails toreveal a convincing DNA-binding motif that is common toall four proteins, and previously suggested possiblemotifs (Stutzman-Engwall et al., 1992; Horinouchi et al.,1990) fall outside the regions of homology; perhaps thesegenes encode a family of regulatory proteins with a novelmeans of recognizing specific nucleotide sequences.

Experimental procedures

Bacterial strains and culture conditions

High-density spore preparations (about 1010 colony-formingunits ml-1) of S. coelicolor A3(2) strain M145 (SCP1-, SCP2-,

Undecylprodigiosin production in Streptomyces coelicolor A3(2) 2803

prototrophic) were obtained and pre-germinated as previouslydescribed (Strauch et al., 1991). Aggregated germlings weredispersed by brief sonication and inoculated into 50 ml ofmedium in 250 ml siliconized flasks containing coiled stainlesssteel springs (about 25 cm flask-1, with 2.5 turns cm-1). Unlessotherwise stated, the medium contained 55 mM (1% w/v)glucose, 25 mM TES buffer (pH 7.2), 15 mM (NH4)2S04,0.5 mM NaH2P04, 0.5 mM K2HP04, 5 mM MgS04, 5% w/vPEG 6000, 1% v/v Rhodorsil Antimousse 426R antifoam, and0.1% v/v trace element solution (0.1 g l-1 of each of ZnS04.7H20, FeS04.7H20, MnCI2.4H20, CaCI2.6H20 and NaCI))supplemented with 0.2% w/v casamino acids. The inoculumwas adjusted to give an OD450nm against water of about 0.1(0.05 of which reflected medium components (largelyantifoam)); this corresponded to a spore inoculum of 4 x 106

ml-1. Flasks were incubated at 30°C and 300 r.p.m., andgrowth was monitored at OD450nm against water. The condi-tions used were modified from those adopted and developedby Hodgson (1982); the high spore inoculum, coiled springsand PEG 6000 all contributed to dispersed growth. Nutritionalshiftdown, treatment with serine hydroxamate, and assess-ment of undecylprodigiosin production were performed as pre-viously described (Strauch et al., 1991).

Plasmids

For analysis of redO transcription, a 2.4 kb BamHI-PstI frag-ment from pIJ2341 (Malpartida et al., 1990) that contains all ofredO (Narva and Feitelson, 1990) was cloned in BamHI-PstI-cleaved pIJ2926 (a derivative of pUC18 (Yanisch-Perron et al.,1985), with BglII sites flanking a modified polylinker; G. R.Janssen, personal communication) to yield p1J6013.For analy-sis of redX transcription, the 1.4 kb BamHI fragment frompIJ2341 that is internal to a redbiosynthetic transcript (Malpar-tida et al., 1990; Guthrie and Chater, 1990) was cloned inpSPT18 (obtained from Boehringer Mannheim) with the pro-moter-proximal BamHI site adjacent to the SP6 promoter of thevector; the resulting plasmid (pIJ6000) was subjected torestriction mapping and a Smal site was found approximately220 bp from the promoter-proximal end of the insert. pIJ6014 isa derivative of pIJ487 (Ward et al., 1986) with the 2.4 kb BglIIredO fragment from pIJ6013 inserted in the BglII site of thepolylinker.

S 1 nuclease mapping

For each S1 nuclease reaction, 30 µg of RNA was hybridized inNaTCA buffer (Murray, 1986) to about 0.02 pmol (approxi-mately 104 Cerenkov c min-1) of labelled probe. For redO, a1.3 kb ClaI-Ndel fragment from pIJ6013 that contains the redOpromoter region was uniquely labelled with 32p (Maxam andGilbert, 1980) at the 5' end of the Clal site within the redO cod-ing region (Fig. 1B, nucleotide position 726) and used asprobe. The results reported were confirmed using a probeuniquely end-labelled at a position 147 nucleotides down-stream of the redO translational start codon and generatedusing the polymerase chain reaction (Ehrlich, 1989) and syn-thetic oligonucleotides; no additional transcriptional start siteswere detected. For redX, a 990 bp SmaI-AatII fragment frompIJ6000 uniquely labelled (Maxam and Gilbert, 1980) at the 5'

end of the Smal site that lies an undetermined distance down-stream of the redX promoter was used as probe. All subse-quent steps were as described in Strauch et al., (1991).Nucleotide sequence ladders were derived as described byMaxam and Gilbert (1980). Before assigning a precise RNA ini-tiation site for redO, one nucleotide was subtracted from thelength of the protected fragment to account for the difference in3' ends resulting from S1 nuclease digestion and the chemicalsequencing reactions (Hentschel et al., 1980).

Amino acid sequence comparisons

The amino acid sequences encoded by redO and by the 5' endof the oppositely orientated open reading frame were used tosearch Release 29 of the EMBL Data Library using TFASTA;

alignments of the redO product with those of actII-ORF4, dnrland afsR, and of the protein encoded by the oppositely ori-ented open reading frame with the trkA product, were achievedusing BESTFIT; both programs were accessed via the UWGCGPackage (Devereux et al., 1984).

Reproducibility

Each experiment was performed in its entirety at least twice,and usually three times. The results shown in this paper weretypical of the repeated experiments.

Acknowledgements

We thank Paco Malpartida for the gift of pIJ2341 , Mark Buttner,Keith Chater, David Hopwood and Tobias Kieser for their com-ments on the manuscript, and David Hodgson for many usefulconversations. The work was supported by grant-in-aid fromthe Agricultural and Food Research Council and the JohnInnes Foundation. E.S., H.C.G. and N.A. were recipients ofpostdoctoral fellowships from the Deutscher AkademischerAustauschdienst, CONICET, and The Royal Society, respec-tively.

ReferencesBascaran, V., Sanchez, L., Hardisson, C., and Braña, A.F.(1991) Stringent response and initiation of secondarymetabolism in Streptomyces clavuligerus. J Gen Microbiol137: 1625-1634.

Baylis, H.A., and Bibb, M.J. (1987) The nucleotide sequence ofa 16S rRNA gene from Streptomyces coelicolor A3(2). NuclAcids Res 15: 1716.

Bibb, M.J., Findlay, P.R., and Johnson, M.W. (1984) The rela-tionship between base composition and codon usage in bac-terial genes and its use in the simple and reliable identifica-tion of protein coding sequences. Gene 30: 157-166.

Bushell, M.E. (1989) The process physiology of secondarymetabolite production. In Microbial Products: NewApproaches. Baumberg, S., Hunter, I., and Rhodes, M.(eds). Cambridge: Cambridge University Press, pp. 95-120.

Buttner, M.J. (1989) RNA polymerase heterogeneity in Strepto-myces coelicolor A3(2). Mol Microbiol 3: 1653-1659.

Chater, K.F. (1989) Aspects of multicellular differentiation inStreptomyces coelicolor A3(2). In Genetics and MolecularBiology of Industrial Microorganisms. Hershberger, C.L.,Queener, S.W., and Hegeman, G. (eds). Washington D.C.:American Society for Microbiology, pp. 99-107.

2804 E. Takano et al.

Chater, K.F. (1990) The improving prospects for yield increaseby genetic engineering in antibiotic-producing strepto-mycetes. Bio/Technology 8: 115-121.

Coco, E.A., Narva, K.E., and Feitelson, J.S. (1991) Newclasses of Streptomyces coelicolor A3(2) mutants blocked inundecylprodigiosin (Red) biosynthesis. Mol Gen Genet 227:28-32.

Demain, A.L. (1989) Carbon source regulation of idiolitebiosynthesis in actinomycetes. In Regulation of SecondaryMetabolism in Actinomycetes. Shapiro, S. (ed.). BocaRaton: CRC Press, pp. 127-134.

Demain, A.L., Aharonowitz, Y., and Martin, J.-F. (1983)Metabolic control of secondary biosynthetic pathways. InBiochemistry' and Genetic Regulation of CommerciallyImportant Antibiotics. Vining, L.C. (ed.). London: Addison-Wesley, pp. 49-72.

Devereux, J., Haeberli, P., and Smithies, O. (1984) A compre-hensive set of sequence analysis programs for the VAX.Nucl Acids Res 12: 387-395.

Ehrlich, H.A. (1989) PCR Technology. New York: StocktonPress.

Feitelson, J.S., Malpartida, F., and Hopwood, D.A. (1985)Genetic and biochemical characterization of the red genecluster of Streptomyces coelicolor A3(2). J Gen Microbiol131: 2431-2441.

Fernandez-Moreno, M.A., Caballero, J.L., Hopwood, D.A., andMalpartida, F. (1991) The act cluster contains regulatory andantibiotic export genes, direct targets for translational controlby the bldA transfer RNA gene of Streptomyces. Cell 66:769-780.

Guthrie, E.P., and Chater, K.F. (1990) The level of a transcriptrequired for production of a Streptomyces coelicolor antibi-otic is conditionally dependent on a transfer RNA Gene. JBacteriol 172: 6189-6193.

Hallam, S.E., Malpartida, F., and Hopwood, D.A. (1988)Nucleotide sequence, transcription and deduced function ofa gene involved in polyketide antibiotic synthesis in Strepto-myces coelicolor. Gene 74: 305-320.

Hawley, D.K., and McClure, W.R. (1983) Compilation and anal-ysis of Escherichia coli promoter DNA sequences. NuclAcids Res 11 : 2237-2255.

Hentschel, C., Irminger, J.-C., Bucher, P., and Birnsteil, M.L.(1980) Sea urchin histone mRNA termini are located in generegions downstream from putative regulatory sequences.Nature 285: 147-151.

Hobbs, G., Frazer, C.M., Gardner, D.C.J., Flett, F., and Oliver,S.G. (1990) Pigmented antibiotic production by Strepto-myces coelicolor A3(2) - kinetics and the influence of nutri-ents. J Gen Microbiol 136: 2291-2296.

Hodgson, D.A. (1982) Glucose repression of carbon sourceuptake in Streptomyces coelicolor A3(2) and its perturbationin mutants resistant to 2-deoxyglucose. J Gen Microbiol 128:2417-2430.

Hopwood, D.A. (1988) Towards an understanding of geneswitching in Streptomyces, the basis of sporulation andantibiotic production. Proc R Soc Lond B 235: 121-138.

Horinouchi, S., Kito, M., Nishiyama, M., Furuya, K., Hong, S.K.,Miyake, K., and Beppu, T. (1990) Primary structure of AfsR,a global regulatory protein for secondary metabolite forma-tion in Streptomyces coelicolor A3(2). Gene 95: 49-56.

Kelly, K.S., Ochi, K., and Jones, G.H. (1991) Pleiotropic effectsof a relC mutation in Streptomyces antibioticus. J Bacteriol173: 2297-2300.

Malpartida, F., Niemi, J., Navarrete, R., and Hopwood, D.A.(1990) Cloning and expression in a heterologous host of thecomplete set of genes for biosynthesis of the Streptomycescoelicolor antibiotic undecylprodigiosin. Gene 93: 91-99.

Maxam, A.M., and Gilbert, W. (1980) Sequencing end-labelledDNA with base specific chemical cleavages. Meth Enzymol65: 449-560.

Murray, M.G. (1986) Use of sodium trichloroacetate and mungbean nuclease to increase sensitivity and precision duringtranscript mapping. Anal Biochem 158: 165-170.

Narva, K.E., and Feitelson, J.S. (1990) Nucleotide sequenceand transcriptional analysis of the redO locus of Strepto-myces coelicolor A3(2). J Bacteriol 172: 326-333.

Ochi, K. (1986) Occurrence of the stringent response in Strep-tomyces sp. and its significance for the initiation of morpho-logical and physiological differentiation. J Gen Microbiol132: 2621-2631.

Ochi, K. (1987a) A rei mutation abolishes the enzyme inductionneeded for actinomycin synthesis by Streptomyces antibioti-cus. Agric Bioi Chem 51 : 829-835.

Ochi, K. (1987b) Metabolic initiation of differentiation and sec-ondary metabolism by Streptomyces griseus: Significance ofthe stringent response (ppGpp) and GTP content in relationto A factor. J Bacteriol 169: 3608-3616.

Ochi, K. (1990) A relaxed (rei) mutant of Streptomyces coeli-color A3(2) with a missing ribosomal protein lacks the abilityto accumulate ppGpp, A-factor and prodigiosin. J GenMicrobiol 136: 2405-2412.

Ohnuki, T., Imanaka, T., and Aiba, S. (1985) Self cloning inStreptomyces griseus of a str gene cluster for streptomycinbiosynthesis and streptomycin resistance. J Bacteriol 164:85-94.

Rudd, B.A.M., and Hopwood, D.A. (1980) A pigmentedmycelial antibiotic in Streptomyces coelicolor: control by achromosomal gene cluster. J Gen Microbiol 119: 333-340.

Sarubbi, E., Rudd, K.E., and Cashel, M. (1988) Basal ppGpplevel adjustment shown by new spoT mutants affect steadystate growth rates and rrnA ribosomal promoter regulation inEscherichia coli. Mol Gen Genet 213: 214-222.

Schreiber, G., Metzger, S., Aizenman, E., Roza, S., Cashel,M., and Glaser, G. (1991) Overexpression of the relA genein Escherichia coli. J Bioi Chem 266: 3760-3767.

Strauch, E., Takano, E., Baylis, H.A., and Bibb, M.J. (1991)The stringent response in Streptomyces coelicolor A3(2).Mol Microbiol 5: 289-298.

Stutzman-Engwall, K.J., Otten, S., and Hutchinson, C.R.(1992) Regulation of secondary metabolism in Strepto-myces spp. and overproduction of daunorubicin in Strepto-myces peucetius. J Bacteriol 174: 144-154.

Tsao, S.W., RUdd, B.A.M., He, X., Chang, C., and Floss, H.G.(1985) Identification of a red pigment from Streptomycescoelicolor A3(2) as a mixture of prodigiosin derivatives. JAntibiot 38: 128-1 30.

Ward, J.M., Janssen, G.R., Kieser, T., Bibb, M.J., Buttner,M.J., and Bibb, M.J. (1986) Construction and characterisa-tion of a series of multi-copy promoter-probe plasmid vectorsfor Streptomyces using the aminoglycoside phosphotrans-ferase gene from Tn5 as indicator. Mol Gen Genet 203:468-478.

Yanisch-Perron, C., Vieira, J., and Messing, J. (1985)Improved M13 phage cloning vectors and host strains:nucleotide sequences of the M13mp18 and pUC19 vectors.Gene 33: 103-119