The relA/spoT-Homologous Gene in Streptomyces coelicolor Encodes Both Ribosome-Dependent (p)ppGppSynthesizing and -Degrading Activities

Abstract

Streptomyces coelicolor (p)ppGpp synthetase (Rel protein) belongs to the RelA and SpoT (RelA/SpoT) family, which is involved in (p)ppGpp metabolism and the stringent response. The potential functions of the rel gene have been examined. S. coelicolor Rel has been shown to be ribosome associated, and its activity in vitro is ribosome dependent. Analysis in vivo of the active recombinant protein in well-defined Escherichia coli relA and relA/spoT mutants provides evidence that S. coelicolor Rel, like native E. coli RelA, is functionally ribosome associated, resulting in ribosome-dependent (p)ppGpp accumulation upon amino acid deprivation. Expression of an S. coelicolor C-terminally deleted Rel, comprised of only the first 489 amino acids, catalyzes a ribosome-independent (p)ppGpp formation, in the same manner as the E. coli truncated RelA protein (1 to 455 amino acids). An E. coli relA spoT double deletion mutant transformed with S. coelicolor rel gene suppresses the phenotype associated with (p)ppGpp deficiency. However, in such a strain, a rel-mediated (p)ppGpp response apparently occurs after glucose depletion, but only in the absence of amino acids. Analysis of ppGpp decay in E. coli expressing the S. coelicolor rel gene suggests that it also encodes a (p)ppGpp-degrading activity. By deletion analysis, the catalytic domains of S. coelicolor Rel for (p)ppGpp synthesis and degradation have been located within its N terminus (amino acids 267 to 453 and 93 to 397, respectively). In addition, E. coli relA in an S. coelicolor rel deletion mutant restores actinorhodine production and shows a nearly normal morphological differentiation, as does the wild-type rel gene, which is in agreement with the proposed role of (p)ppGpp nucleotides in antibiotic biosynthesis.

Actinomycetes are very well-known for their ability to produce several secondary metabolites with important biological properties (10). In general, antibiotic formation in Streptomyces is developmentally regulated, with production in liquid cultures being associated with the stationary phase, while in agar-grown cultures it is coupled to morphological differentiation (11, 14, 19). While the global regulatory mechanisms of antibiotic production are still poorly understood, a role for growth rate or growth cessation in triggering synthesis of secondary metabolites has been suggested (10, 12).

Attempts at elucidating some of the steps in the mechanism(s) controlling antibiotic production by Streptomyces coelicolor have been conducted in several laboratories. Many putative regulatory genes have been isolated and characterized. Such a gene, whose deduced product shows strong similarities to proteins of the RelA and SpoT (RelA/SpoT) family (implicated in ppGpp metabolism), has recently been cloned from S. coelicolor (7, 8, 30). Deletion of this gene gave an actinorhodine-nonproducing phenotype (7, 30), suggesting a role for ppGpp in antibiotic production. This gene, originally called ORF1, will be referred to as rel, while relA and spoT will be exclusively used for E. coli.

In members of the family Enterobacteriaceae, the highly phosphorylated guanine nucleotides [(p)ppGpp] are pleiotropic effectors involved in the adaptation of bacterial cells to nutritional and environmental changes. The transient (p)ppGpp accumulation upon nutrient depletion is characterized by several features: reduction in stable RNA accumulation, stimulation of certain amino acid operons, and induction of specific stationary-phase gene expression (for reviews, see reference 6). In Escherichia coli, (p)ppGpp formation depends on two coordinated pathways, which are catalyzed by two gene products: RelA (34) [(p)ppGpp synthetase I] and SpoT (50) [ppGpp 3′ pyrophosphohydrolase, or (p)ppGpp synthetase II]. Under amino acid deprivation, when codon-specific uncharged tRNA is bound to the ribosomal acceptor site, a relA-dependent (p)ppGpp-synthesizing activity occurs, allowing the cell to adapt to the reduced amino acid availability. Biochemical studies of the RelA protein reveal that it is a ribosome-associated enzyme, whose activity involves the pyrophosphoryl transfer from ATP to GTP (or GDP). During carbon source starvation, the (p)ppGpp accumulation seems to be largely dependent on the spoT gene product (18, 35, 59) (despite the possible partial contribution of RelA, due to the transient amino acid limitation that follows carbon source deprivation [9]). Recent in vivo functional studies of several spoT deletions support the notion that the protein is a bifunctional enzyme with (p)ppGpp-synthesizing and -degrading activities, although SpoT 3′ pyrophosphate transferase activity has not been demonstrated in vitro (17). E. coli RelA/SpoT sequence comparison has revealed that both proteins are closely related (36). The similarities might well be a consequence of a gene duplication, following which relA has evolved to a (p)ppGpp-synthesizing activity, while spoT has become more specialized for (p)ppGpp degradation (36).

Among the RelA/SpoT homologs so far studied, the Streptococcus equisimilis Rel protein has been reported to function as SpoT when expressed in E. coli, but S. equisimilis disruptants behave similarly to relA mutants (32). For the S. equisimilis enzyme, both weak ribosome-independent (p)ppGpp-synthesizing and -degrading activities have been demonstrated in vitro (32). These experimental findings suggest that the S. equisimilis rel gene performs both relA and spoT activities; this suggestion has been reinforced by recent in vivo studies of this bacterium (33) and Bacillus subtilis (58).

We have reported that an S. coelicolor rel-deleted mutant loses its ribosome-associated (p)ppGpp-synthesizing activity (30). A similar rel-deleted mutant of a different S. coelicolor strain (M600) has been reported to be unable to accumulate ppGpp upon amino acid deprivation (7, 8). Taken together, these results suggest that S. coelicolor rel might function as E. coli relA. However, sequence comparison of S. coelicolor Rel with other RelA/SpoT proteins has shown higher similarities to E. coli SpoT (8, 30), which, unlike E. coli RelA, is directly involved in (p)ppGpp degradation; this opens up a crucial question: does S. coelicolor Rel have some functions similar to those of SpoT?

In addition, an unrelated gene (polynucleotide phosphorylase) apparently involved in (p)ppGpp metabolism has recently been reported in Streptomyces antibioticus; when assayed in vitro, both a ribosome-independent (p)ppGpp synthetase and a polynucleotide phosphorylase activity are found (22).

In this paper, we describe the functional characterization of S. coelicolor Rel, determined by using complementation analysis of well-defined E. coli relA and spoT mutants.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains used are listed in Table 1. The E. coli vectors and recombinant plasmids used are shown in Table 2. E. coli M13mp19-derivative phage (60) was used for in vitro mutagenesis. The Streptomyces plasmids pIJ486 (57), pIJ941 (27), and pSCNB080 (30) were used. The Streptomyces φC31-derivative phages used were PM1 (29) and KC859 (4). pCNB3033, a Streptomyces integrative vector, was constructed by cloning the 3.6-kbp NdeI-KpnI fragment from phage KC859, previously made blunt ended by T4 DNA polymerase treatment, into SmaI-digested pIJ2921, yielding plasmid pMF2058; the hygromycin resistance gene from PM1 was finally cloned into BamHI-PstI-digested pMF2058. Similarly, the Streptomyces integrative thiostrepton-derived vector pMF2024 was obtained.

TABLE 1.

Bacterial strains used in this study

Strain	Relevant genotype	Reference
E. coli
JM101	supE thi Δ(lac-proAB) F′(traD36 proAB+lacIqlacZΔM15)	60
CF1652	MG1655 relA251	59
CF1693	MG1655 spoT207 relA251	59
ET12567	F−dam-13::Tn9 dcm-6 hsdM hsdR recF143 zjj-201::Tn10 galK2 ara-14 lacY1 xyl-5 leuB6 thi-1 tonA31 rpsL136 hisG4 tsx-78 mtl-1 glnV44	28
S. coelicolor
J1501	hisA1 uraA1 strA1 pgl SCP1− SCP2−	13
18J	J1501 rel-deleted mutant	30

TABLE 2.

Plasmids used in this study

Plasmid	Relevant features and construction [reference]a
pUC18	pBR322-derived E. coli vector, bla [60]
pSU21	pACYC184-derived E. coli vector, cat [1]
pIJ2921	pUC-derived E. coli vector, bla [21]
pET19b	pBR322-derived E. coli T7 polymerase-expression vector; (Novagen, Madison, Wis.)
pVLT31	RSF1010-lacIq/Ptac hybrid broad-host-range expression vector, tet [15]
pGG21	relA plasmid, colE1 replicon, aphII [34]
pCNB3033	Streptomyces nonreplicating pIJ2921-derived vector carrying bacteriophage φC31 integration functions, bla, hyg
pCNB080	Original S. coelicolor 4.1-kbp fragment; rel and its adjacent regions cloned in BamHI-digested pSU21
pCNB0115	relA under its natural promoter control; relA SalI (12,390)-BamHI (9,151) fragment cloned in pSU21
pCNB0118	As pCNB0115, but NdeI-engineered site (11,512) at relA start codon
pCNB0121	rel Sau3AI (1)-BamHI (805), NdeI-engineered site (308) at rel initiation codon cloned in pSU21 (BamHI site)
pCNB0127	rel NdeI-engineered (308) BamHI (4,068) fragment and relA SalI (12,390)-NdeI-engineered (11,512) fragment (PrelA) simultaneously cloned into pUC18 (EcoRI-HindIII sites), expressing native Rel protein
pCNB0128	rel first 500 bp in frame to lacZ under relA promoter control; pCNB0127 partially digested BamHI, followed by self-ligation
pCNB0129	relA with rel promoter; relA NdeI-engineered (11,512) BamHI (9,151) fragment and rel Sau3A (1)-NdeI-engineered (308) fragment (Prel) cloned in pUC18 (EcoRI-HindIII sites)
pCNB0131	rel Sau3A (1)-BstEII (3,284) fragment cloned in pSU21 (EcoRI-XbaI sites)
pCNB0163b	rel NdeI-engineered (584) BstEII (3,284) fragment cloned under PrelA control, expressing Rel(93–847); Met…Stop
pCNB0165	pCNB0127 EcoRI-HindIII insert cloned in pVLT31 (EcoRI-HindIII sites)
pCNB0166	pCNB0163 EcoRI-HindIII insert cloned in pVLT31 (EcoRI-HindIII sites)
pCNB0173	pCNB0129 BamHI insert cloned into pCNB3033 (BamHI site)
pCNB0182	pCNB0131 EcoRI-HindIII insert cloned into pCNB3033 (BamHI site)
pCNB0187b	rel NdeI-engineered (308) SmaI (1,775) fragment cloned under PrelA control, expressing Rel(1–489)
pCNB0190b	rel NdeI-engineered (308) BstEII (3,284) fragment cloned under PrelA control with the in-frame deletion StyI (1,665)-PvuII (1,742), expressing Rel(1–453, 479–847)
pCNB0812b	rel BalI (993)-BstEII (3,284) fragment cloned under PrelA control, expressing Rel(230–847); MetLeuGluAspPro…Stop
pCNB0815b	rel BglII (1,383)-SmaI (1,775) fragment cloned under PrelA control, expressing Rel(359–489); MetLeuAspArgGlySer…Stop
pCNB0817b	rel BalI (993)-PstI (1,574) fragment cloned under PrelA control, expressing Rel(230–422); MetLeuGluAspPro…SerAsnAsnStop
pCNB0818b	rel BglII (1,107)-StyI (1,665) fragment cloned under PrelA control, expressing Rel(267–453); MetLeuAspArgGlySer…Stop
pCNB0819b	rel NdeI-engineered (584) NruI (1,500) fragment cloned under PrelA control, expressing Rel(93–397); Met…ValSerAsnAsnStop
pCNB0820b	rel NdeI-engineered (584) BglII (1,383) fragment cloned under PrelA control, expressing Rel(93–359); Met…Stop
pCNB0821b	rel NdeI-engineered (584) BglII (1,107) fragment cloned under PrelA control, expressing Rel(93–267); Met…Stop

^abla, ampicillin resistance gene; cat, chloramphenicol acetyltransferase gene; tet, tetracycline resistance gene; aphII, aminoglycoside phosphotransferase gene from Tn5; hyg, hygromycin resistance gene. Numbers in parentheses after a restriction enzyme name indicate the nucleotide position, which corresponds to either rel or relA reported in the databases (EMBL Data Bank accession no. X92520 and U29580, respectively). Plasmids were constructed by cloning appropriate fragments either directly or through end-filling with Klenow DNA polymerase I (except for PstI digestion, which was made blunt ended by using T4 DNA polymerase). The amino acids incorporated by the cloning strategy are indicated by the three-letter code. 

^bConstructed by simultaneous cloning with the relA promoter (PrelA) into pUC18 (similar to that described for pCNB0127). 

General techniques.

Isolation, cloning, and manipulation were performed as previously described for E. coli (48) and Streptomyces (20). NdeI restriction sites were generated by using the Sculptor in vitro mutagenesis system (Amersham; RPN 1526). Previously, suitable restriction fragments from either rel or relA were cloned in M13mp19, and mutagenesis was performed according to the manufacturer’s recommendations, with the following synthetic oligonucleotides: P01, 5′-ACTTCTTACCGCAACCATATGTCCTCTCCT-3′; P02, 5′-TGGGCCTCGTCTGGCATATGGACTCCTCGTGCGCGAT-3′; and P03, 5′-CCGCCTGGCCCATATGGGCGTGCAGCGC-3′.

Construction of recombinant plasmids.

The recombinant plasmids used and their construction are summarized in Table 2. Hybrid plasmids containing the relA and rel genes, including their promoter regions, were constructed by previously generating an NdeI restriction site by in vitro mutagenesis as described above; primers P01 and P02 were used for relA and rel, respectively. From the resulting NdeI-engineered pCNB0118 and pCNB0121 (for the relA and rel genes, respectively), the hybrid constructions were obtained, in which the rel gene is under the control of the relA promoter (pCNB0127) and the relA gene is under the control of the rel promoter (pCNB0129). To construct pCNB0163, expressing the truncated Rel(93–847) protein, an NdeI restriction site was previously generated with primer P03.

To localize the Rel catalytic domains, several recombinant plasmids carrying various deletions of the gene were constructed (Table 2). In some cases, the rel coding fragments were previously cloned into pET19b, creating suitable restriction sites for further subclonings and generating translation start and stop codons. By this procedure, some of the resulting recombinant gene products have additional amino acids, which are indicated in Table 2.

The rel expression plasmids have been cloned under the control of the relA promoter and possess the original relA ribosome-binding site.

Media and growth conditions.

E. coli strains were grown on either liquid or solid 2YT medium (48). For in vivo ³²P_i labeling, MOPS (morpholinepropanesulfonic acid) glucose minimal medium (38) containing 0.2 mM phosphate, 0.4% glucose, amino acids (50 μg/ml each), and bases (20 μg/ml each) was used. Growth on serine-methionine-glycine (SMG) medium (56) was used to test the rel-dependent (p)ppGpp response. Growth of E. coli CF1693 was also checked on minimal glucose M9 medium (48). Glycogen assays were performed as described elsewhere (17), with glucose-rich medium (0.5% tryptone, 0.5% yeast extract, 2% glucose).

Determinations of intracellular (p)ppGpp concentrations.

Cell labeling was performed as described previously (49). Changes in the intracellular (p)ppGpp pool upon amino acid starvation were quantitated after dl-serine hydroxamate treatment (1 mg/ml, final concentration) (55); cells were uniformly ³²P_i labeled in MOPS medium with the composition described above, but with serine omitted. The glucose concentration in MOPS medium was lowered to 0.04% when glucose starvation was induced, either with or without amino acids, by α-methylglucoside addition (2.5%, final concentration). Amino acid (isoleucine) deprivation in MOPS medium was achieved with valine at 500 μg/ml and relieved by addition of isoleucine (100 μg/ml, final concentration). Acid extraction of nucleotides was done as described previously (2) and resolved by one-dimensional polyethylenimine thin-layer chromatography developed with 1.5 M KH₂PO₄ (pH 3.4) as described by Cashel (5). Nucleotides were quantified with a Fujix Bas 1000 imaging system with TINA 2.08 software (Raytest, Straubenhardt, Germany). The amounts of (p)ppGpp were expressed as fractions of the total nucleotide pool [GTP plus (p)ppGpp]; correction was done as previously described (30), with acid extracts from either CF1652 or CF1693. In some cases, quantification of (p)ppGpp was also performed by liquid scintillation counting as described previously (30).

Ribosome purification.

Ribosomes were prepared essentially as described by Krohn and Wagner (25) with minor modifications: cells were disrupted by sonication, and buffer A consisted of 50 mM Tris acetate (pH 8), 15 mM magnesium acetate, 60 mM potassium acetate, 30 mM ammonium acetate, 1 mM dithiothreitol, 0.2 mM EDTA, and 0.5 mM phenylmethylsulfonyl fluoride. High-salt-washed ribosomes were obtained by resuspension of low-salt-washed ribosomes in buffer A containing 0.7 M ammonium acetate and centrifuged as previously described (25); the resulting pellet was resuspended in buffer A and used directly as high-salt-washed ribosomes, while the supernatant, which constituted the high-salt wash, was dialyzed against buffer A. The crude extract is defined as the supernatant after centrifugation of the disrupted cells at 25,000 × g for 30 min at 4°C.

Measurement of (p)ppGpp synthetase activity.

(p)ppGpp synthetase assays were carried out essentially as described previously (30), except that the incubation time was 30 min. To measure this enzyme activity in crude extracts, either with or without 18% methanol, 12-μl aliquots of the reaction mixture were taken at 0, 5, 10, and 30 min, and reactions were stopped with 2 μl of 4 M formic acid. Nucleotides were resolved and quantified as described above; the activity was calculated within the linear range.

Preparation of (p)ppGpp.

Preparative scale synthesis of ppGpp and pppGpp was performed essentially as described elsewhere (25), with low-salt-washed ribosomes from E. coli CF1652 carrying the relA plasmid (pCNB0115). Labeled 3′-[α-³²P]ppGpp and -pppGpp were prepared according to the method of Sy and Lipmann (53) with some modifications: the reaction proceeded for 1 or 2 h at 30°C with [α-³²P]GTP and was stopped by two phenol-chloroform extractions and then a final chloroform extraction. The aqueous phase was diluted at least six times with 50 mM triethylammonium acetate (pH 7.7) and applied to a 1-ml DEAE-Biogel column that had been equilibrated with the same buffer. Stepwise elution was performed as described previously (53), and ppGpp- and pppGpp-containing fractions were lyophilized to remove excess salt and dissolved in a small amount of water.

Preparation and assay of (p)ppGpp phosphohydrolase activity.

Crude extracts were obtained as described above, except that buffer B (50 mM Tris acetate [pH 8], 1 mM EDTA, 0.3 M KCl, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 20% glycerol) was used. Because of high nonspecific nucleotidase activities, 1.5 ml of the crude extract was previously fractionated with ammonium sulfate up to 20%. After centrifugation at 25,000 × g for 20 min at 4°C, the pellet was suspended in 0.5 ml of buffer B and dialyzed against buffer B. The insoluble protein aggregates were collected by centrifugation at 9,000 × g for 30 min at 4°C, dissolved in 0.1 ml of buffer B containing 1 M KCl, and stored at −20°C until use. (p)ppGpp-degrading activity was carried out in 30 μl of reaction mixture in buffer B containing 45 μg of protein, 0.2 mM [α-³²P](p)ppGpp (about 1 μCi/μmol), and 10 mM MnCl₂. The final KCl concentration was 0.3 M. Six-microliter aliquots were taken at 0, 5, 10, 15, and 20 min (in which the reaction was shown to be linear), and the reaction was stopped with 1 μl of 4 M formic acid. Nucleotides were resolved and quantified as described above.

Miscellaneous methods.

Protein was measured as described elsewhere (3) with bovine serum albumin as a standard. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was carried out with the buffer system described by Laemmli (26) in either a 10 or 15% polyacrylamide gel, and protein bands were visualized by staining with Coomassie brilliant blue R250.

RESULTS

Expression of S. coelicolor Rel in E. coli. (i) In vitro assay of S. coelicolor (p)ppGpp synthetase.

As a first approach to analyze the functional properties of S. coelicolor rel gene, we examined the expression of the S. coelicolor rel gene in the E. coli relA deletion CF1652 strain. To this end, this strain was transformed with plasmid pCNB080, which contains the S. coelicolor rel gene under the control of its own promoter. However, neither the presence of a recombinant protein by SDS-PAGE analysis nor (p)ppGpp synthetic activity was detected (data not shown). Clearly, this could be due to a failure to obtain expression of rel gene. Consequently, a construct was made in which the native rel gene-initiating UUG codon was changed to AUG, and the coding sequence was cloned under the control of the relA promoter, including the original relA ribosome-binding site (pCNB0127). Crude extract from CF1652 strain transformed with pCNB0127 showed a soluble protein in SDS-PAGE, having the expected apparent molecular mass (94 kDa) for the S. coelicolor Rel protein (Fig. 1). The expected 84-kDa RelA protein could be detected when strain CF1652 is transformed with pCNB0115, which harbors the relA gene. By using crude extracts from this strain expressing either the rel or relA gene, the (p)ppGpp synthesizing activities were calculated to be 2.8 and 22.5 nmol of (p)ppGpp formed/min/mg, respectively; as with the wild-type RelA, the S. coelicolor Rel activity was detected in the ribosomal fraction (Fig. 2).

FIG. 1.

SDS-PAGE of crude extracts from E. coli CF1652. Lanes: 1, plasmid harboring the rel gene (40 μg); 2, plasmid carrying the relA gene (52 μg); 3, with no plasmid (46 μg). The overproduced recombinant S. coelicolor Rel and E. coli RelA of 94 and 84 kDa, respectively, are indicated by arrowheads. The positions and molecular masses of marker proteins are indicated on the right of the gel. The smaller bands observed in either lane 1 or 2 correspond to β-lactamase and chloramphenicol acetyltransferase, respectively.

FIG. 2.

Synthesis of polyphosphorylated guanine nucleotides by purified ribosomes. Reaction conditions were as specified in Materials and Methods, with low-salt-washed ribosomes from E. coli CF1652 with no plasmid (lane 1); plasmid harboring the E. coli relA gene (pCNB0115), either with or without methanol lanes 2 and 4, respectively); plasmid carrying the S. coelicolor rel gene (pCNB0127), either with or without methanol (lanes 3 and 5, respectively); and without ribosomes added (lane 6). Final protein concentrations were 0.90 (lane 1), 1.05 (lanes 3 and 5), and 0.95 (lanes 2 and 4) mg/ml.

Whether or not this ribosome-associated (p)ppGpp synthetase is also ribosome dependent in vitro was also investigated. Thus, supernatants of high-salt-washed ribosomes from either S. coelicolor J1501 (carrying additional copies of rel in pSCNB080) or E. coli JM101 (carrying additional copies of relA in pGG21) were obtained as described in Materials and Methods. No (p)ppGpp formation occurred with either preparation. Nevertheless, when either supernatant was analyzed for its (p)ppGpp-synthesizing activity in the presence of high-salt-washed ribosomes from either S. coelicolor 18J or E. coli CF1652 [which are completely devoid of ribosome-associated (p)ppGpp formation] or 18% methanol, the (p)ppGpp synthesis was detected (Table 3). Thus, in both cases, the enzymatic activity appeared to be ribosome dependent. The resulting activities of S. coelicolor Rel were almost identical, irrespective of the source of ribosomes, suggesting that its interactions with E. coli ribosomes might well be similar to that of RelA.

TABLE 3.

Relative synthesis of (p)ppGpp by supernatants of high-salt-washed ribosomes derived from either S. coelicolor J1501/pSCNB080 or E. coli JM101/pGG21^a

Addition to reaction mixture	% (p)ppGpp synthesis
	J1501/pSCNB080b	JM101/pGG21c
None	NDd	ND
High-salt-washed ribosomes from 18J	100	75
High-salt-washed ribosomes from CF1652	92	100
18% methanol	25	10

^aConditions for reaction and (p)ppGpp quantification were as specified in Materials and Methods. Preincubation of high-salt washes and high-salt-washed ribosomes was performed for 30 min at 4°C. Reaction mixtures contained 19, 4.8, 90, and 68 μg of protein from either high-salt washes from S. coelicolor and E. coli or high-salt-washed ribosomes from S. coelicolor and E. coli, respectively. 

^b100% = 1.6 nmol of (p)ppGpp formed/50 μl of assay mixture. 

^c100% = 6.9 nmol of (p)ppGpp formed/50 μl of assay mixture. 

^dND, not detectable. 

Since the S. coelicolor Rel protein is ribosome associated and its activity is ribosome dependent in vitro, it was of interest to assess if the Rel activity is also functionally ribosome dependent. As a result, we undertook in vivo functional studies.

(ii) In vivo functional analysis of the rel gene in E. coli.

Because E. coli CF1652 lacks a functional relA gene and, consequently, has no ribosome-dependent (p)ppGpp formation, this strain shows a relaxed response and is unable to grow on SMG medium. To explore if the S. coelicolor rel gene is able to complement the relaxed phenotype of E. coli CF1652, this strain was transformed with several plasmids: pCNB080, pCNB0115, pCNB0127, pCNB0128, and pCNB0129 (Table 2). As shown in Fig. 3, pCNB0127 transformants behave like those containing pCNB0115 and pCNB0129 (carrying the relA gene), while no growth of either pCNB080 or pCNB0128 (a rel-lacZ fusion) transformants was observed. Because pCNB0127 transformants grew on SMG medium, a functional rel-dependent (p)ppGpp response may be acting.

FIG. 3.

Growth of E. coli CF1652 transformants on SMG medium. The constructs used to transform E. coli CF1652 are shown on the left-hand side, and growth analysis is shown on the right. CF1652 transformants were grown in 2YT medium with the selective antibiotics until the A₄₆₀ was 0.6 to 0.8, washed twice with minimal glucose M9 medium, spread on SMG agar plates, and incubated for 16 h at 37°C. pSU21, control plasmid. Sc, S. coelicolor; Ec, E. coli.

In vivo synthesis of (p)ppGpp by the S. coelicolor rel gene product was estimated after amino acid starvation in response to the addition of dl-serine hydroxamate. (p)ppGpp accumulated in E. coli CF1652 carrying pCNB0127 (Fig. 4A), as is observed for the native relA gene; the maximal production is detected 20 min after the addition of serine hydroxamate, which represents an eightfold increase over the basal levels. Upon chloramphenicol addition, the (p)ppGpp levels dropped within a few minutes (Fig. 4A). This result might be expected, since chloramphenicol is known to be an inhibitor of the ribosome-dependent (p)ppGpp formation (46), and the (p)ppGpp might be rapidly degraded by the (p)ppGppase activity of the intact SpoT protein from CF1652. The basal (p)ppGpp levels in CF1652 transformed with pCNB0127 were the same as those in JM101 (35 and 21 pmol/A₄₆₀ units, respectively) and were 20-fold less than that for CF1652 carrying pCNB0115 (662 pmol/A₄₆₀ units) (Fig. 4B). Notably, the (p)ppGpp levels in strain CF1652 transformed with the rel gene after 20 min of amino acid deprivation were 2- and 10-fold lower than those observed in the presence of either single or multiple copies of the relA gene (268 versus 495 and 2,580 pmol/A₄₆₀ units, respectively) (Fig. 4B).

FIG. 4.

Changes in (p)ppGpp levels upon amino acid deprivation. (A) Relative (p)ppGpp accumulation of strain CF1652 containing the S. coelicolor rel plasmid pCNB0127 after serine hydroxamate addition (○). Chloramphenicol (at a final concentration of 100 μg/ml and indicated with arrow [Cm]) was added to culture after 20 min (•). (B) Amounts of (p)ppGpp upon serine hydroxamate treatment of E. coli JM101 (single-copy E. coli relA) (columns 1 and 4) and E. coli CF1652 transformed with either pCNB0127 (S. coelicolor rel) (columns 2 and 5) or pCNB0115 (multicopy E. coli relA) (columns 3 and 6). Results for cultures prior to the addition of serine hydroxamate and after 20 min are shown by columns 1 to 3 and 4 to 6, respectively. At each incubation period, samples of 5 μl of acid extracts were analyzed for their (p)ppGpp content.

Taking these results together, we inferred that the S. coelicolor Rel protein behaves like E. coli RelA. In order to investigate additional functions of the S. coelicolor Rel protein, we explored its properties in the well-defined E. coli CF1693 relA spoT double mutant.

Functional analysis of S. coelicolor Rel in an E. coli ΔrelA ΔspoT background. (i) The S. coelicolor rel gene suppresses the phenotypic defects of the E. coli relA spoT double mutant.

The wild-type relA gene is lethal in E. coli CF1693 because of the resultant high intracellular (p)ppGpp concentration, which presumably arises from the failure of (p)ppGpp degradation normally catalyzed by SpoT (17, 59). Because S. coelicolor Rel appears to function in the same manner as E. coli RelA, it was of interest to examine whether the rel gene could be introduced into the CF1693 strain. Viable transformants were obtained when the rel gene on plasmid pCNB0127 was introduced in this strain. Nevertheless, the transformants have a severely slow growth rate; well-defined colonies formed only after 36 to 48 h on 2YT plates at 30°C.

S. coelicolor rel was cloned in a low-copy-number plasmid under the relA promoter (pCNB0165) for further analysis of strain CF1693 because it grows slowly when it harbors pCNB0127. The ability of the rel gene to reverse several features associated with the ppGpp-deficiency phenotype of CF1693 was explored. Growth on minimal medium of CF1693 transformed with pCNB0165 plasmid was detected, but not with the vector as a control, indicating that the amino acid requirements of this strain (59) are suppressed by the S. coelicolor rel gene. Because glycogen synthesis is believed to be strongly dependent on (p)ppGpp (47), indirect evidence of (p)ppGpp synthesis by the S. coelicolor rel product could be obtained by evaluating the accumulation of glycogen in CF1693 and its derivative when grown on glucose-rich medium. CF1693 fails to accumulate glycogen and, consequently, strains yellow with iodine vapors, while pCNB0165 transformants showed an intense brown stain (data not shown). Additionally, when CF1693 carrying either pVLT31 (control plasmid) or pCNB0165 was grown in 2YT medium, their respective doubling times were 66 and 114 min, respectively. This difference is expected if the levels of (p)ppGpp are elevated because of S. coelicolor (p)ppGpp synthetase activity. Interestingly, when pCNB0165 transformants of CF1693 were plated on solid 2YT medium, the colonies exhibited a morphology different from those of this strain containing the vector plasmid. While the latter displayed fuzzy edges and tended to lyse upon prolonged incubation in the stationary phase, those harboring the S. coelicolor rel gene were characterized by a translucent appearance and remained viable for several days. The cell morphology of the CF1693 strain under the light microscope has been reported as being more elongated than that of the wild type (59). This phenotype was observed either during exponential growth in liquid 2YT medium or during both exponential and stationary phases in M9 medium containing all amino acids, while an apparently normal cellular morphology was restored by transformation with pCNB0165 (data not shown). These results suggest that the rel gene can partially complement some deficiencies of the relA spoT double deletion mutation.

(ii) Characterization of the rel gene in E. coli relA spoT double mutant.

In order to investigate if Rel protein has some functional features similar to those of SpoT, we analyzed the response of CF1693 transformed with pCNB0165 to (p)ppGpp upon carbon source deprivation. It is well known that upon glucose starvation, (p)ppGpp production is largely dependent on SpoT when amino acids are present in the medium (17). After the addition of α-methylglucoside, the ppGpp pool (only this compound was detected) in CF1693 transformants increased only in the absence of added amino acids (Fig. 5), suggesting that glucose depletion leads first to amino acid starvation, and this subsequently causes activation of ribosome-dependent S. coelicolor rel activity.

FIG. 5.

Accumulation of ppGpp upon glucose starvation of E. coli CF1693 containing the S. coelicolor rel plasmid pCNB0165. Relative ppGpp levels are shown at various times after α-methylglucoside treatment. (□) MOPS minimal medium containing all amino acids. (■) MOPS minimal medium with no amino acids. Samples of 20 μl (delivered in 5-μl intervals) of the acid extracts were analyzed for the estimations of amounts of (p)ppGpp. Basal ppGpp levels were either 38.5 or 28.1 pmol/A₄₆₀ units with and without amino acids, respectively.

To analyze if S. coelicolor Rel had some (p)ppGpp-degrading activity, we measured the decay of ppGpp following direct reversal of stringency conditions. Thus, isoleucine was added to valine-inhibited cultures of strain CF1693 transformed with either pCNB0165 or pCNB0166, expressing the wild-type Rel and the truncated Rel(93–847), respectively. The results are shown in Fig. 6. Basal ppGpp levels of CF1693 transformed with pCNB0166 appeared to be 1.5-fold higher than those with pCNB0165, which might well correlate with its even lower growth rate (doubling time of 122 min in 2YT medium). Upon amino acid starvation, ppGpp accumulation occurred in the presence of both full-length Rel and truncated Rel; interestingly, after addition of isoleucine, ppGpp disappears in both strains within a few minutes. From the equation k = ln2/half-life, their first-order decay constants were estimated, which were 0.11/min and 0.25/min for native Rel and Rel(93–847), respectively. These values approximate those observed for E. coli SpoT and S. equisimilis Rel (16, 17, 32) and are shorter than the ppGpp half-life for the spoT-deleted mutant used here, which has been reported to be at least 30 min (17). Thus, either the full-length Rel or the N-terminally deleted Rel is involved in the ppGpp degradation. Furthermore, in vitro (p)ppGpp hydrolysis was also detected in fractionated cell extracts from CF1693 transformed with either pCNB0165 or pCNB0166, but not from the strain containing the control plasmid; their respective activities were calculated to be 3.2 and 3.9 nmol/min/mg. ppGpp and pppGpp were similarly degraded to GDP and GTP, respectively, and this activity required Mn²⁺ ions, which could not be substituted for by Mg²⁺.

FIG. 6.

Reversal of a valine-imposed stringent response by isoleucine addition in E. coli CF1693 harboring S. coelicolor rel plasmids. (A and B) ppGpp accumulation upon amino acid (isoleucine) deprivation by valine addition of strain CF1693 transformed with either pCNB0165 (expressing native Rel) (○) or pCNB0166 [expressing Rel(93–847)] (□), respectively. After 20 min, a valine-induced stringent response was reversed by isoleucine addition, and ppGpp decay was analyzed for either the pCNB0165 (•) or pCNB0166 (■) strain. (C) Rate of ppGpp decay in either strain. The amounts of ppGpp shown are relative to the time of isoleucine addition.

E. coli relA restores actinorhodine production in an S. coelicolor rel mutant.

The S. coelicolor rel-deleted strain 18J does not produce actinorhodine, but its synthesis is restored by the presence in trans of the rel gene (30). Because the Rel protein appears to share several of the properties of E. coli RelA, it was of interest to test if the relA gene was able to restore actinorhodine production in the streptomycete mutant.

Attempts to clone the relA gene in S. coelicolor 18J under the control of the rel promoter were unsuccessful. Neither in a high- nor a low-copy-number plasmid nor in an integrative vector was it possible to obtain a recombinant strain carrying the relA gene, whereas transformants carrying the rel gene were easily isolated. The inability to obtain recombinants even with an integrative vector (pCNB0173) could well be the result of the lethal accumulation of (p)ppGpp as a consequence of the synthetic activity of the RelA, which might be higher than that of the Rel protein. Therefore, an inducible promoter was chosen for complementation analysis in order to avoid the possible deleterious effects of the relA gene being expressed in S. coelicolor under the control of the rel promoter. Thus, the relA gene was cloned under the control of the tipA promoter (37), with the Streptomyces integrative plasmid pIJ8600 (52) used as the vector. Several transformants of S. coelicolor 18J were obtained with this construction. Plasmid integration into the chromosome was confirmed by Southern analysis of their respective total DNAs. Interestingly, when grown on R5 solid medium, even without thiostrepton induction, both actinorhodine production and a nearly normal morphological differentiation were observed in the 18J mutant strain carrying the E. coli relA gene, while no phenotypic changes were detected in this strain harboring the integrated vector as a control. Thus, the relA-dependent activity appears to be sufficient to restore actinorhodine production in the heterologous host, as does the wild-type rel gene.

Localization of the catalytic domains of Rel protein.

From the previous results, it can be deduced that the Rel protein, like RelA, has a ribosome-dependent (p)ppGpp-synthesizing activity, which is activated upon amino acid deprivation. Nevertheless, unlike RelA, the Rel protein is also capable of degrading (p)ppGpp, a function it shares with SpoT. Additionally, the Rel protein has been shown to contain a sequence that matches the consensus pattern of a putative ATP-GTP binding motif (amino acids 458 to 465) (30), a particular feature which is not present in any other known protein involved in (p)ppGpp metabolism. In order to define the functional domains of the Rel protein, expression analyses of truncated recombinant Rel proteins (including an internal 25-amino-acid in-frame deletion of the Rel protein, by which the putative ATP-GTP binding motif has been removed) were undertaken with either E. coli CF1652 or CF1693.

(i) (p)ppGpp-synthesizing domain.

To localize more precisely the protein region involved in (p)ppGpp-synthesizing activity, several constructs were generated with various Rel deletions (Table 2 and Fig. 7). The resulting plasmids were tested with the E. coli relA mutant strain (CF1652) for their ability to induce (p)ppGpp synthesis in vivo. The viability and growth rate of CF1693 (the relA spoT strain) when transformed with the recombinant plasmids might be used as an additional indication of (p)ppGpp synthesis. By SDS-PAGE analysis, all recombinant proteins were expressed at a similar level to that detected in Fig. 1 for the full-length Rel protein, except for the truncated proteins Rel(359–489), Rel(230–422), and Rel(267–453), which were almost undetectable (data not shown). The results are summarized in Fig. 7.

FIG. 7.

Phenotypic characterization of the recombinant plasmids expressing truncated S. coelicolor Rel proteins. Recombinant pUC18 derivatives containing the indicated rel deletions are shown. NdeI-engineered restriction sites are indicated by asterisks. Restriction sites from pET19b are shown in nonitalic style. In vivo (p)ppGpp activity was determined as specified in Materials and Methods. The ribosome dependence was estimated by subjecting the recombinant strains to amino acid starvation by the addition of serine hydroxamate; samples taken at 0, 7, 15, and 30 min were analyzed for their (p)ppGpp content. +, (p)ppGpp detected (a) or (p)ppGpp accumulation (b) in response to the addition of serine hydroxamate; −, no (p)ppGpp detected (a) or no change in (p)ppGpp levels (b) upon the addition of serine hydroxamate. t, viable transformants; u, unable to select transformants; n.a., not applicable. Transformants of CF1693 carrying pCNB0127 and pCNB0163 showed a very slow growth rate.

While the in vivo basal (p)ppGpp concentration of CF1652 expressing either the native Rel or the truncated Rel proteins Rel(93–847), Rel(1–489), and Rel(267–453) were very similar (32 to 47 pmol/A₄₆₀ units), those observed for the truncated Rel proteins Rel(230–847) and Rel(1–453, 479–847) were 200 and 170 pmol/A₄₆₀ units, respectively. The inability to obtain CF1693 transformants by using some of the truncated Rel plasmids (Fig. 7) might be due to the high (p)ppGpp levels because of either the deregulated (p)ppGpp synthesis or the lack of some (p)ppGppase activity; another possible explanation is that these peptides would be toxic to this particular strain.

In addition, by in vitro measurements, (p)ppGpp-synthesizing activity was detected only when crude extracts from the recombinant strains were used, in which (p)ppGpp was present in vivo (data not shown).

From these results, it is suggested that the (p)ppGpp synthesizing domain of S. coelicolor Rel can be allocated between residues 267 and 453.

(ii) Ribosome dependence.

In order to analyze the ribosome dependence of Rel (p)ppGpp-synthesizing activity, transformants of strain CF1652 with several recombinant plasmids were subjected to amino acid deprivation by addition of serine hydroxamate (Fig. 7). As expected, transformants expressing the truncated Rel(93–847) protein showed a (p)ppGpp response similar to that of native Rel. The absence of changes in the (p)ppGpp levels for the truncated Rel(1–489) protein is in agreement with the results for a truncated RelA protein (51) and supports the observed ribosome dependence of rel activity. In vivo analysis of Rel(1–453, 479–847) revealed no incremental effect on the (p)ppGpp levels during amino acid starvation; thus, this region of Rel protein, which contains the putative ATP-GTP binding domain, might be important in the modulation of the ribosome-dependent activity of the Rel protein. Interestingly, transformants expressing the truncated Rel(230–847) protein showed no change in (p)ppGpp concentration upon amino acid starvation, suggesting that a portion of the Rel N terminus might well participate with the C-terminal fragment in the ribosome-dependent activity.

(iii) (p)ppGpp-degrading domain.

To localize the region of the Rel protein which is involved in (p)ppGpp degradation, we focused our attention on the Rel region, homologous to that of SpoT, known to contain this function (17). Recombinant plasmids (pCNB0819, pCNB0820, and pCNB0821) carrying several rel deletions (Table 2) were used to transform CF1693 carrying pCNB0165. Plasmid pCNB0165 was used as a source of (p)ppGpp synthesis; the role of the corresponding recombinant proteins on ppGpp degradation was analyzed after reversion of the valine-imposed stringent response by addition of isoleucine. Only with plasmid pCNB0819, expressing Rel(93–397), were we able to detect a higher level of ppGpp disappearance (k = 0.19/min), indicating that the (p)ppGpp-degrading activity is contained within this 305-amino-acid fragment. It should be stressed that the recombinant proteins expressed by these plasmids were not detected by SDS-PAGE analysis (data not shown).

DISCUSSION

We have investigated the potential functions of S. coelicolor Rel, providing evidence that it behaves similarly to E. coli RelA; however, unlike this protein, S. coelicolor Rel is capable of degrading (p)ppGpp, a function it shares with SpoT. Conversely, S. coelicolor Rel might not have a (p)ppGpp-synthesizing function similar to SpoT, as deduced by carbon source starvation analysis. Still, the presence of some (p)ppGpp synthesis in the Streptomyces background of a Rel protein like SpoT could be argued and constitutes an interesting feature that remains to be investigated.

Notably, the ppGpp decay rate for the native Rel protein (k = 0.11/min) is lower than the values observed for the E. coli SpoT (16, 17) and S. equisimilis Rel (32) proteins but is still shorter than the reported ppGpp half-life for the E. coli strain, CF1693, used here (k = <0.02/min) (17). This result might reflect either some insensitivity of S. coelicolor Rel to the reversal of the stringent response or the fact that the (p)ppGpp-degrading activity is somehow slower, probably because of changes in the modulation of its catalytic properties in the heterologous host. Of interest in this context is the higher ppGpp decay constant obtained for the Rel(93–847) protein (k = 0.25/min), which is within the range of those found for other (p)ppGppases.

Taking the results of deletion analysis together, the Rel catalytic region appeared to be contained in a 361-amino-acid fragment within its N terminus (Fig. 8). While 305 amino acids (from 93 to 397) are required for (p)ppGpp degradation, the overlapping region containing residues 267 to 453 is sufficient to yield some (p)ppGpp-synthesizing activity. In addition, a regulatory region is described at its C terminus, which might be involved in the regulation by ribosome. This regulatory domain might include the putative ATP-GTP binding motif; nevertheless, whether or not this region is able to bind ATP or GTP requires further biochemical studies. Remarkably, the 187-amino-acid peptide involved in (p)ppGpp synthesis is shorter than the corresponding homologous sequence of either the C-terminally truncated RelA or the 291-amino-acid SpoT protein, which has been reported to be able to synthesize (p)ppGpp (17, 51). Conversely, the 305-amino-acid peptide that retains the (p)ppGpp-degrading activity of S. coelicolor Rel is larger than the corresponding domain determined for SpoT (17). Attempts to further shorten this fragment failed to obtain functional recombinant truncated peptides.

FIG. 8.

Schematic representation of the domain structure of S. coelicolor Rel. The identified domains are drawn to scale. The ATP-GTP binding motif is indicated in black.

It is interesting to note that the truncated Rel(230–847) protein showed a ribosome-independent (p)ppGpp synthesis; thus, there might exist an additional modulatory domain within the N terminus of the native protein, in addition to that contained at its C terminus, or these regions are both needed for its ribosome-dependent activity. Alternatively, the (p)ppGpp levels might be regulated by the (p)ppGpp-degrading activity of S. coelicolor Rel; this function is apparently not present in this truncated Rel protein (31) and, hence, might account for the failure of the ribosome-dependent (p)ppGpp response. This observation is of great interest when trying to understand the mechanisms of this (p)ppGpp synthetase regulation and its activation by ribosome.

Unlike native RelA, the in vitro and in vivo (p)ppGpp synthesizing activities of S. coelicolor Rel when expressed in E. coli appeared to be lower. This might be attributed to the presence of nonfunctional protein. Another possible explanation is that the Streptomyces ribosome-associated (p)ppGpp activity might have some distinct modulation of its catalytic properties. The fact that the in vivo (p)ppGpp levels upon amino acid deprivation in Streptomyces have been reported to be significantly lower than those observed in E. coli (45, 54) might be in agreement with the latter hypothesis. In addition, its closer similarity to SpoT rather than to RelA within its N terminus might reflect the fact that S. coelicolor Rel and SpoT share some common features within the proposed (p)ppGpp-synthesizing domain. Such a view is supported by reports of weak (p)ppGpp synthetase activity of SpoT (17).

Interestingly, relA restores actinorhodine production in the S. coelicolor rel-deleted 18J strain, in the same way as a native rel gene does. Moreover, relC mutants of Streptomyces have been recently characterized (23, 44). These mutants showed a deficiency in antibiotic production and a defective ribosomal (p)ppGpp synthesis (23, 42–44), which were restored by introducing the rplK (= relC) gene in trans (23, 44). In E. coli, a functional 50S ribosomal protein, L11, encoded by the relC gene has been shown to be required for the activation of the RelA protein and, consequently, for the synthesis of (p)ppGpp (6). Although additional work is required, these observations provide support for the proposed significance of (p)ppGpp in the onset of antibiotic production in Streptomyces (24, 39–43). It is remarkable that RelA, unlike S. coelicolor Rel, is apparently completely devoid of (p)ppGppase activity (6); a stable S. coelicolor strain, 18J, transformed with relA indicates that an additional gene for (p)ppGppase activity may exist in Streptomyces or that the activity might result from nonspecific (p)ppGpp degradation by other nucleotidases. These features are of particular interest and are currently being investigated.

In Streptomyces, a more distantly related isofunctional gene for (p)ppGpp synthesis has been reported (22), providing additional opportunities for functional specialization and the incorporation of independent regulatory mechanisms associated with specific metabolic roles. For that reason, definition of the catalytic properties of S. coelicolor Rel, particularly those responsible for regulating its activity, and mutational analysis constitute a useful tool for analyzing Rel’s role in Streptomyces (p)ppGpp metabolism. Studies combining these methods are in progress.