Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Novel Two-Component Systems Implied in Antibiotic Production in Streptomyces coelicolor

  • Ana Yepes,

    Affiliation Instituto de Biología Funcional y Genómica/Departamento de Microbiología y Genética, Consejo Superior de Investigaciones Científicas (CSIC)/Universidad de Salamanca, Edificio Departamental, Campus Miguel de Unamuno, Salamanca, Spain

  • Sergio Rico,

    Affiliation Instituto de Biología Funcional y Genómica/Departamento de Microbiología y Genética, Consejo Superior de Investigaciones Científicas (CSIC)/Universidad de Salamanca, Edificio Departamental, Campus Miguel de Unamuno, Salamanca, Spain

  • Antonio Rodríguez-García,

    Affiliation Instituto de Biotecnología de León, INBIOTEC, Parque Científico de León, León, Spain

  • Ramón I. Santamaría,

    Affiliation Instituto de Biología Funcional y Genómica/Departamento de Microbiología y Genética, Consejo Superior de Investigaciones Científicas (CSIC)/Universidad de Salamanca, Edificio Departamental, Campus Miguel de Unamuno, Salamanca, Spain

  • Margarita Díaz

    Affiliation Instituto de Biología Funcional y Genómica/Departamento de Microbiología y Genética, Consejo Superior de Investigaciones Científicas (CSIC)/Universidad de Salamanca, Edificio Departamental, Campus Miguel de Unamuno, Salamanca, Spain


The abundance of two-component systems (TCSs) in Streptomyces coelicolor A3(2) genome indicates their importance in the physiology of this soil bacteria. Currently, several TCSs have been related to antibiotic regulation, and the purpose in this study was the characterization of five TCSs, selected by sequence homology with the well-known absA1A2 system, that could also be associated with this important process. Null mutants of the five TCSs were obtained and two mutants (ΔSCO1744/1745 and ΔSCO4596/4597/4598) showed significant differences in both antibiotic production and morphological differentiation, and have been renamed as abr (antibiotic regulator). No detectable changes in antibiotic production were found in the mutants in the systems that include the ORFs SCO3638/3639, SCO3640/3641 and SCO2165/2166 in any of the culture conditions assayed. The system SCO1744/1745 (AbrA1/A2) was involved in negative regulation of antibiotic production, and acted also as a negative regulator of the morphological differentiation. By contrast, the system SCO4596/4597/4598 (AbrC1/C2/C3), composed of two histidine kinases and one response regulator, had positive effects on both morphological development and antibiotic production. Microarray analyses of the ΔabrC1/C2/C3 and wild-type transcriptomes revealed downregulation of actII-ORF4 and cdaR genes, the actinorhodin and calcium-dependent antibiotic pathway-specific regulators respectively. These results demonstrated the involvement of these new two-component systems in antibiotic production and morphological differentiation by different approaches. One is a pleiotropic negative regulator: abrA1/A2. The other one is a positive regulator composed of three elements, two histidine kinases and one response regulator: abrC1/C2/C3.


Antibiotics are highly valuable secondary metabolites that are broadly produced in different species of the genus Streptomyces, a filamentous soil bacterium with a complex life cycle. In fact, this genus produces about half of all known microbial antibiotics [1]. The onset of antibiotic production depends on the growth stage of the microorganism and usually takes place contemporaneous with differentiation of the aerial mycelium into spores. Both differentiation and antibiotic production can be triggered by many environmental changes (physical and chemical), such as nutrient deprivation, pH, temperature, etc. These changes must be sensed and integrated in a cell response to promote rapid adaptation to the new growth conditions. The quickest and most efficient bacterial responses to extracellular stimuli occur via histidine- aspartate (His-Asp) phosphorelay cascades. These systems are made up of inner membrane-spanning protein kinases, which sense the external environment, and their respective (cognate) cytoplasmic response regulator partners, which generally exhibit DNA-binding properties. Most of these signal transduction systems only require a single sensor (HK: histidine kinase) and a cognate response regulator (RR) and are thus referred to as two-component systems (TCSs) [2]. Recently, some atypical systems have been described, such as a kinase phosphorylated by GTP instead of ATP [3] and the phosphorylation independent activation response regulators, named PIARR [4][6].

S. coelicolor A3(2) is the best genetically studied Streptomyces strain and has become the model organism for these species. The complete sequence of its 8.7 Mb linear chromosome is available ( [7]) [8] and contains 84 sensor kinase and 80 response regulator genes, 67 of which lie adjacent on the chromosome and are predicted to form TCSs [9]. The mean HK/RR (TCS) content of S. coelicolor (considering the whole 7825 ORFs) is 0.86% as compared with 0.65% for other free-living microorganisms studied or 0.26% for pathogenic bacteria (25% and 70% more in Streptomyces, respectively) [10]. This abundance of TCSs could reflect the complexity of the regulatory network of Streptomyces that would allow this genus to adapt and survive in multiple and adverse environmental conditions.

S. coelicolor A3(2) produces at least four chemically distinct antibiotics: actinorhodin (ACT), undecylprodigiosin (RED), calcium-dependent antibiotic (CDA) and methylenomycin, all of whose biosynthetic genes are located in clusters. The antibiotic production responds to a hierarchy of different levels of decision, distinguishing global or pathway-specific regulators [11]. Pathway-specific regulators are part of the biosynthetic clusters (i.e., actII-ORF4 for ACT [12]; redD for RED [13] and cdaR for CDA [14]). Global regulators are located elsewhere and have the ability to regulate operons that belong to different metabolic pathways, and as a consequence mutants in these genes usually show pleiotropic phenotypes. Among the global regulators there are some of which are affecting different process such as differentiation and antibiotic production (i.e. BldA [15], RelA [16], AbsB [17]) and others reported just as global antibiotic regulators (i.e. AbsA1/A2 [18]). TCSs usually act as global regulators that mediate the response from external/internal stimuli to the final target genes.

The function of most of the 67 S. coelicolor TCSs is unknown; only a few have an assigned role. Six of them have been reported to modulate the antibiotic production and the best studied, absA1/A2, is involved in their global regulation [14], [18][21]. The aim of this study was to determine the role of other five TCSs of S. coelicolor, whose functions are as yet unknown. Four of them are annotated as homologues to absA1/A2 in the Streptomyces Annotation Server ( [22]). The fifth TCS, although it showed less similarity with absA1/A2, is an interesting system because it is composed of two HKs and one RR and may be considered a three-component system. A sequence comparison of this system with the available databases indicated its conservation in almost all the species of Streptomyces that are being sequenced by the Broad Institute (

In this study, the knockouts of the TCSs encoding genes (sensor and regulator at the same time) were generated and the changes in antibiotic production and morphological differentiation were monitored in several medium cultures. Two of the five TCSs selected (ΔSCO1744/1745 and ΔSCO4596/4597/4598) showed significant differences in both antibiotic production and morphological differentiation, and have been renamed as abr (antibiotic regulator). No changes in antibiotic production were detected in the deletion mutants of the other three systems encoded by the ORFs SCO3638/39, SCO2165/66 and SCO3640/41 in any of the conditions tested. The mutant ΔSCO1744/45 showed a pleiotropic phenotype. The ACT, RED, and CDA productions on some media were triggered, suggesting a negative role of this system in the antibiotic production. In addition, the morphological differentiation was accelerated. An opposite pleiotropic phenotype was revealed for the ΔSCO4596/97/98 mutant (TCS formed by two HKs and one RR). This mutant showed a decrease in ACT, RED, and CDA antibiotic productions and a delay in differentiation, which indicates that this system is a positive global regulator of the antibiotic production and differentiation. Microarray analyses of the ΔSCO4596/97/98 and wild-type transcriptomes were performed.


Construction of null mutant strains

According to the annotations of the S. coelicolor database genome ( [22]), five TCSs were selected. Four of them, SCO1744/45, SCO2165/66, SCO3638/39, and SCO3640/41, shared about 30% identity between their corresponding HKs and that of the well-known global antibiotic regulator absA1/A2, AbsA1. Additionally 50% identity was found between their RRs and the AbsA2 regulator, which are considered to be homologues (Table S1 and Table S2). The fifth one, composed of two HKs (SCO4597 and SCO4598, which share 57% identity) and one RR (SCO4596), presented less similarity to absA1/A2 (25% HKs-AbsA1 and 33% RR-AbsA2) but both HKs were predicted to be functionally associated to AbsA1 using STRING application (Search Tool for the Retrieval of Interacting Genes/Proteins) ( [23] (AbsA1-SCO4598 association score of 0.726 just below the AbsA1–AbsA2 and AbsA1-RedZ scores, 0.949 and 0.923, respectively; AbsA1-SCO4597 association score of 0.691). To determine the relevance of these five TCSs in antibiotic production, null mutant strains of each system were obtained from the S. coelicolor M145 strain by the REDIRECT procedure (see Material and Methods). The correct replacement of the genes by the cassette was confirmed by Southern blot hybridization using appropriate DNA probes (data not shown).

To detect putative alterations in antibiotic production and/or development of cells in the mutant strains compared to the wild type, all of them were grown on several solid media at 30°C. The media used were a minimal medium (NMMP) and different complex media (NA, YEPD, R2YE, PGA and MSA).

Two of the five TCSs selected null mutant strains (ΔSCO1744/45::accIV and ΔSCO4596/97/98::accIV) consistently displayed significant differences in antibiotic production and differentiation compared to the wild type (Figure 1). No differences were observed in any conditions for the mutants of the systems SCO3638/39 and SCO3640/41 (data not shown). The absence of the system SCO2165/66 in the mutant seemed to slightly increase production of the three antibiotics (ACT, RED, and CDA) in R2YE, PGA, and NA media respectively but these results were difficult to replicate and need further study (data not shown).

Figure 1. Antibiotic production and differentiation of the different strains.

Wild-type strain: S. coelicolor M145. Mutant strains: S. coelicolor ΔabrA1/A2 and S. coelicolor ΔabrC1/C2/C3. A: ACT production on NMMP solid (top) and liquid (bottom) medium; B: ACT production on NA solid (top) and NB liquid (bottom) medium; C: RED production on PGA solid (top) and liquid (bottom) medium; D: CDA production bioassay against B. subtilis on NA solid medium (top) and inhibition halo diameter quantification (bottom); E: differentiation assay on YEPD (two days' growth). S. coelicolor M145 (black columns), ΔabrA1/A2::aacIV (grey columns), ΔabrC1/C2/C3::aacIV (white columns). Error bars correspond to standard deviation of four independent experiments.

Clearly, the effect of mutations ΔSCO1744/45::accIV and ΔSCO4596/97/98::accIV was medium-dependent, especially in the production of ACT. Although differences could also be seen on R2YE and YEPD media (data not shown), the strongest effects in ACT production were on NMMP and on NA (Figure 1A, 1B). Both mutants showed different phenotypes on both solid media: the mutant ΔSCO1744/45::accIV displayed an ACT overproduction after three days' growth, while ΔSCO4596/97/98::accIV strain produced significantly less of this antibiotic molecule compared to the wild type (Figure 1A, 1B). To quantify these observations, liquid cultures were performed determining the rate growth and ACT production of each strain in both media at different times. As shown in Figure 1A (lower panel), the ACT production in the mutant ΔSCO1744/45::accIV in liquid NMMP was increased more than sixfold compared to the wild type at 96 h and the production in the mutant ΔSCO4596/97/98::accIV was about the half of the wild type. On the other hand, ACT production of ΔSCO4596/97/98::accIV strain in NB was about 40% of that of the wild type but less production of ACT in the mutant ΔSCO1744/45::accIV in NB was also observed showing a more complex nutritional behaviour of this mutant. The growth rates of the strains were similar in both liquid media (Figure S1).

Differences in undecylprodigiosin production were also observed on R2YE but mainly on PGA solid media (Figure 1C). Quantification of RED production in PG liquid medium showed that ΔSCO1744/45::accIV produces 67% more and ΔSCO4596/97/98::accIV approximately 50% less compared to the wild-type strain. As mentioned, this is not due to a growth defect because the growth curves of the three strains were almost identical in the culture conditions used (Figure S1).

CDA production, measured as the inhibition halo against Bacillus subtilis, was evaluated on NA plates in the presence or absence of calcium (see Materials and Methods) (Figure 1D). Once again, ΔSCO1744/45::accIV had higher CDA production than the wild-type strain (8.5%, the average of seven independent assays), and ΔSCO4596/97/98::accIV presented a decrease of 32% of the inhibition halo.

Finally, YEPD was the medium used to document the morphological development. ΔSCO1744/45::accIV mutant presented an accelerated formation of aerial mycelium, while ΔSCO4596/97/98::accIV showed a clear delay in the differentiation in these culture conditions (Figure 1E).

In summary, mutant ΔSCO1744/45::accIV overproduced the three antibiotics and also showed a positive role in differentiation (the aerial mycelia and spores appeared sooner than in the wild-type strain). In contrast, strain ΔSCO4596/97/98::accIV showed a decreased production of the antibiotics ACT, CDA, and RED, and the differentiation was delayed.

From these results we can conclude that the two-component systems composed by the ORFs: SCO1744/45 and SCO4596/97/98, acted as antibiotic production regulators, and thus they were called abrA1/A2, and abrC1/C2/C3, respectively.

Genetic complementation of TCSs null mutants

To make sure that the null mutant phenotypes observed were due to the absence of TCSs genes and not to mutagenesis polar effects, the genetic complementation was carried out. First of all, the mutagenesis apramycin cassette of the each null mutant strain (ΔabrA1/A2::accIV and ΔabrC1/C2/C3::accIV) was eliminated to avoid possible polar effects (see Materials and Methods). The resulting strains harboured a small scar (83 bp) in place of the former antibiotic resistance sequence (ΔabrA1/A2 and ΔabrC1/C2/C3) and displayed the same phenotypes as the original mutants (data not shown).

The reverting strains were obtained by ectopic integration of plasmids derived from pKC796Hyg in the ΦC31 attachment site: pHabrA (whole system), pHabrC1/2/3 (whole system), pHabrC1/3 (with a deletion in the gene encoding kinase AbrC2), and pHabrC2/3 (with a deletion in the gene encoding kinase AbrC1) (see Materials and Methods). Wild type and mutant strains with the integrated pKC796Hyg plasmid were used as controls. It is worth mentioning that integration of any plasmid in the ΦC31 site provokes a decrease in the antibiotic production [24], especially on NMMP medium. As shown in Figure 2, both ΔabrA1/A2 (pHabrA) and ΔabrC1/C2/C3 (pHabrC1/2/3) restored the phenotypes of ACT production and differentiation of wt (pKC796Hyg), although partially in the case of ACT production in ΔabrC1/C2/C3 (pHabrC1/2/3) strain (Figure 2B). The ΔabrC1/C2/C3 mutant phenotype could also be reverted by complementation with pHabrC2/3 but not with pHabrC1/3 (Figure 2C) suggesting a more important role of HK AbrC2 (SCO4597) in the signalling network in this medium.

Figure 2. Mutant complementation.

A: Complementation of ΔabrA1/A2 phenotypes by the integrative plasmid pHabrA derived from pKC796Hyg on NMMP. Top: morphological differentiation. Bottom: ACT production. B: Complementation of ΔabrC1/C2/C3 phenotypes by the integrative plasmid pHabrC1/2/3 derived from pKC796Hyg on NA (2 days). Top: morphological differentiation. Bottom: ACT production. C: Complementation of ΔabrC1/C2/C3 phenotypes by the integrative plasmid pHabrC1/3 and pHabrC2/3 derived from pKC796Hyg on NA (3 days). Top: morphological differentiation. Bottom: ACT production.

The reversion of the mutant phenotypes was also analysed using multicopy plasmids derived from pN702GEM3 (high copy number: 40–100 copies/genome) harbouring either abrA1/A2 (plasmid pNXabrA) or abrC1/C2/C3 genes (plasmid pNabrC) (see Materials and Methods). When abrA1/A2 genes were expressed in the multicopy plasmid the mutant phenotype was not only reverted (Figure S2), but also antibiotic production (ACT, RED and CDA) was even lower than in the wt (pN702GEM3). Additionally, the strain ΔabrA1/A2 (pN702GEM3) had an accelerated aerial mycelium formation, as opposed to ΔabrA1/A2 (pNXabrA) and wt (pN702GEM3) strains.

Unexpectedly, the ΔabrC1/C2/C3 (pNabrC) strain had even less antibiotic production (ACT, RED, and CDA) than the mutant ΔabrC1/C2/C3 (pN702GEM3) strain (Figure S3). However, when the genes were expressed from a low copy number plasmid pAbrC (derived from pHJL401 5–10 copies/genome see Materials and Methods), both phenotypes, antibiotics production and morphological differentiation, were reverted (Figure S3).

Our results confirm that both systems have different roles in regulation; while both affect antibiotic production and morphological differentiation pathways, the AbrA1/A2 is a negative pleiotropic regulator and AbrC1/C2/C3 is a positive pleiotropic regulator.

Microarray analysis of the ΔabrC1/C2/C3 strain

In order to determine the genes whose expression could be affected by the lack of the three-component system, microarrays assays comparing gene expression levels between ΔabrC1/C2/C3 and wild-type strains were performed. Total RNA preparations were obtained from cultures (four replicates) grown for 50 h on NA solid medium (see Materials and Methods). Statistical analysis of the microarray results using limma provided a differential expression value and an associated p-value for each gene. After correction of these p-values for multiple testing (FDR or pdf, see Materials and Methods), only a few genes were statistically significant (p<0.05) (see Table 1). Most of them, however, encoded either hypothetical proteins or proteins of putative functions, which were not easily correlated with the phenotype observed. If uncorrected p-values were considered (p<0.05), 201 genes appeared to be upregulated and 202 genes downregulated in the mutant strain. This set of genes should be taken with caution since it might contain false positives. Nevertheless, certain genes showed expression changes that could be correlated with phenotypic observations or with a shared function. Thus, the lower antibiotic production of the mutant ΔabrC1/C2/C3 was reflected in the expression changes of structural and regulatory genes. Particularly, the ACT and CDA pathway-specific regulators actIIORF4 and cdaR were slightly downregulated (see Table 1). Semiquantitative RT-PCR (see Materials and Methods) confirmed this (see Materials and Methods). When compared to the wild-type strain M145 the transcript levels of these genes in the mutant strain decreased to 60% and 16%, respectively (Figure 3).

Figure 3. RT-PCR assays.

S. coelicolor M145 and ΔabrC1/C2/C3 RNAm amplification of actIIORF4 and RNA 16S (25 cycles) and cdaR (40 cycles) by RT-PCR. Ribosomal RNA 16S amplification (25 cycles) was used as control. Quantification of signal intensities is shown at the left.

Table 1. Selected genes differentially expressed in the Microarray assay ΔabrC1/C2/C3 vs M145 by their p-value<0.05 and FDR/pfp<0.05 or their biological meaning (italics).

Therefore, the downregulation of the mentioned SARPs encoding genes causes, at least partially, a decrease in ACT and CDA production in ΔabrC1/C2/C3 as indeed the phenotypic assays showed.

Expression differences in translation-related genes were also found (Table 1). Some genes encoding ribosomal proteins and amino acid transporters proteins were downregulated in the mutant strain, while the ribosomal recycling factor encoding gene (frr, SCO5627) was upregulated.


In this paper, we reported the study of five new TCSs from S. coelicolor M145 and the involvement of two of them, named AbrA1/A2 and AbrC1/C2/C3, in antibiotic production. Notoriously, the phenotype of both knockout strains was conditional. This fact is not surprising since the TCSs are frequently aimed to respond to specific environmental signals (i.e. AfsQ1-Q2-sigQ [25]), which can be easily missed in some culture media or conditions.

Additionally, our data show how both TCS systems studied played pleiotropic roles in bacteria since not only affected different antibiotic pathways but also different biological processes such as morphological differentiation. Up to date most of the characterized TCSs in S. coelicolor have been reported to have an effect on antibiotic production (i.e. CutR/S [26], EcrA1/A2 [27], PhoR/P [28], AbsA1/A2 [18], RapA1/A2 [29]). However, just one among them, (AfsQ1-Q2-sigQ) has been described to be involved in both secondary metabolism and morphological development [25].

As detailed in the results section, the null mutant strain ΔabrA1/A2 (SCO1744/45) overproduced the three antibiotics tested in a medium dependent manner. This fact makes this system extremely interesting since it could be used to overproduce clinical useful antibiotics by expressing abrA1/A2 alleles in heterologous streptomycetes as has been recently reported for the system AbsA1/A2 [30]. Interestingly, this system only has an orthologue in S.lividans being absent in all the other Streptomyces species sequenced to date. However, the S. lividans knockout does affect neither antibiotic production nor morphological differentiation (data not shown). Therefore, this system seems to represent a S. coelicolor specific antibiotic regulator.

The system, AbrC1/C2/C3, must be considered special because it has two kinases and one regulator. Besides, each gene is separated from the upstream ORF by a DNA sequence long enough to have its own promoter (286, 112, and 171 nt, respectively). Therefore, each gene might be expressed independently in order to suit its own needs. This system is conserved in all the Streptomyces species sequenced so far as well as in the ones those are in the process of being sequenced. Furthermore, the response regulator protein SCO4596 shares about 80% identity at the amino acid level in all the species. This consistently indicates an important role for this special system.

Our data demonstrate that the deletion of the three genes originates a strain with reduced capacity to produce the three antibiotics studied, ACT, RED, and CDA. Similar phenotypes were obtained with the expression of these three genes in a high copy number plasmid but not in a low copy number where the phenotypes were reverted to the wild type ones (Figure S3), showing that this effect was dose dependent. On the contrary, the mutant phenotype with respect to morphological differentiation was reverted even in multicopy number plasmid (Figures 2 and S3). This suggests that separate mechanisms underlie the effects of AbrC1/C2/C3 on antibiotic production and differentiation, as was found with AbsB protein [31].

Microarray analysis and RT-PCR studies demonstrated the role of AbrC1/C2/C3 over antibiotic production was at least partly through transcription of pathway-specific regulator genes actIIORF4 and cdaR. However, with the data obtained to date, we cannot determine whether this is a direct regulation due to the binding of AbrC3 to the specific promoters of the pathway regulators or an indirect effect through a complex regulatory network. Therefore, deeper studies will be performed to understand the role of this TCS in the regulation of antibiotic production in the pigmented streptomycete S. coelicolor. Expression differences between ΔabrC1/C2/C3 mutant and wild-type strains have also been found in genes associated with translation machinery. We hypothesized that a lower expression of some ribosomal protein genes (9 out of 62) in the mutant may affect the synthesis of proteins needed for the production of antibiotics, and in response cells try to compensate this by increasing the ribosomal recycling factor. The relation between enhanced protein synthesis during the stationary phase and the expression of regulatory proteins governing antibiotic production has been suggested previously [32], [33]. In addition, previous work has correlated the ribosomal proteins and the frr overexpression with ACT production [34] and more recently with avermectin overproduction [35].

It is widespread known that antibiotic production in S. coelicolor is a complex process that is regulated by a broad network of genes. In this paper two new two-component global regulators in this network have been identified. It is noteworthy that, they are among the very few TCSs identified on S. coelicolor that are affecting two different but related processes: the antibiotic production and developmental differentiation. One, abrA1/A2, is a negative regulator; the other, abrC1/C2/C3, a three-component system composed by two HKs and one RR, is a positive regulator.

Materials and Methods

Strains, media and culture conditions

Escherichia coli strains growth was accomplished as described previously [36]. BW25113 (pIJ790) (containing the λRed system) is an E. coli K12 (ΔaraBAD, ΔrhaBAD) derivative [37]; non-methylating ET12567 (pUZ8002) is dam, dcm, hsdS, cat, tet containing the atra genes [38] and E. coli DH5α (pBT30) is recA, cat, bla containing flp gene [37]. For CDA bioassays a wild-type strain of Bacillus subtilis (CECT 4522) was grown as an overlay on NA medium. S. coelicolor M145 (prototroph, SCP1, SCP2, methylenomycin) and its mutant strain derivatives were grown on R2YE, NA, MSA, PGA, YEPD, and NMMP [39]. Liquid cultures were performed in 100 ml baffled flasks with 15 ml medium each. When necessary, the medium was supplemented with antibiotics (E.coli media: 100 µg ml−1 for ampicillin, 50 µg ml−1 for apramycin, 50 µg ml−1 for kanamycin, 34 µg ml−1 for chloramphenicol, and 25 µg ml−1 for nalidixic acid. S. coelicolor media: 20 µg ml−1 for neomycin and 20 µg ml−1 for hygromycin).

Isolation and manipulation of DNA

Plasmid isolation, restriction enzyme digestion, ligation, and transformation of E. coli and S. coelicolor were carried out by methods of Sambrook et al [40] and Kieser et al [39], respectively. The plasmids and cosmids used are listed in Table 2. Total genomic DNA from S. coelicolor (gDNA) was isolated from a 24–36 h cultures in TSB medium following the procedure described in Hopwood et al [41], but scaled to 1–2 g of mycelium.

Deletion of the TCSs selected

REDIRECT PCR-targeting technology [42] was used to replace the genes of the entire coding region of each TCS (comprising histidine kinase and response regulator) to an apramycin (aac(3)IV gene) resistance cassette. Mutagenic cassettes were flanked by the recognition sequence of E. coli Flipase (FRT) and contained the conjugation transfer origin oriT (FRT-aac(3)IV-oriT-FRT) and were amplified using the High-Fidelity Expand PCR system (Roche Co.) with the primers listed in Table S3 using plasmid pIJ773 as template. The generated cassettes were introduced into E. coli BW25113 (pIJ790) harbouring the appropriate cosmid for each studied system (Table 2: SCI11, SC5F7, SCH10 and SCD20; [22]) and preinduced for λRed functions, by the addition of arabinose, to obtain a target gene-disrupted version of the mutant cosmids. The disrupted cosmids, confirmed by restriction analysis, were isolated and transferred from E. coli ET12567 (pUZ8002) to S. coelicolor M145 by conjugation. Exconjugants were selected on MSA medium containing apramycin (50 µg ml−1), and the double crossover products identified by screening their sensitivity to kanamycin (50 µg ml−1). The disruptions were confirmed by Southern hybridization and the DIG DNA labelling and detection kit (Roche Co.) was used for probe preparation (obtained with primers of Table S3).

To avoid putative polar effects of the mutagenesis cassette gene replacement in S. coelicolor M145, the antibiotic resistant marker and the oriT region were eliminated in two steps. In a first step, the corresponding disrupting cosmids were introduced in E. coli DH5α (pBT30) strain (harbouring the Flipase gene, FLP) in which, the recombination between both FRT mutagenesis cassette-flanking regions takes place. In these new cosmids only 81 base pairs (SCAR) remained in frame with the adjacent ORFs. Afterwards, in the second step, the SCAR cosmids were transferred to the Streptomyces apramycin-resistance mutant strains by protoplast transformation, selecting neomycin-resistance clones in the first place (unique recombination). Finally the strains were apramycin and neomycin-sensitive (double recombination). PCR assays confirmed the correct recombination in the new Streptomyces mutant strains.

Plasmid constructions

All the plasmids used in this work are listed in Table 2. Integrative plasmid pHabrA was obtained by cloning PCR-amplified abrA1/A2 genes and their own promoter in the shuttle Streptomyces integrative plasmid pKC796Hyg [43]. In the intermediate pAY001 plasmid, the promoter PCR fragment was amplified with primers AY-033 (adding an EcoRI site), and AY-034 (adding an NdeI site) (Table S3), using SCI11 as a template, was cloned in pXHis1 plasmid [44]. pAY002 derivative plasmid harbours the pair of genes amplified by PCR with primers AY-035 (additional NdeI site) and AY-036 (adding an XhoI site) in the NdeI/XhoI sites of pAY001 plasmid. Fragment BglII/BglII from pAY002 was finally cloned in pKC796Hyg plasmid, yielding pHabrA plasmid.

pNXabrA plasmid was obtained by cloning the fragment NdeI/HindIII from pAY002 plasmid in the sites of the pNX24 plasmid (pN702GEM3 derivative [45]). In this shuttled (E.coli-Streptomyces) multicopy plasmid the xylanase promoter xysAp controls abrA1/A2 gene expression.

The three genes abrC1/C2/C3, and their intergenic regions were cloned in a pN702GEM3 plasmid yielding a multicopy plasmid in several steps. An intermediate E. coli monofunctional plasmid called pSCD20 was constructed by cloning a BspEI/PmII fragment from a subclone of SCD20 cosmid in the BspEI/Ecl136II sites of pHJL401 [46]. Afterwards, the fragment EcoRV/HindIII from pSCD20 was introduced in pN702GEM3 to get pNabrC. The low copy number pAbrC plasmid derived from pHJL401 [46] was obtained by cloning the BglII/HindIII fragment from pNabrC in the BamHI/HindIII sites of pHJL401.

To obtain the integrative plasmid for mutant ΔabrC1/C2/C3 complementation, pNSCD20 was digested with HindIII, filled with Klenow polymerase, and BglII digested. This fragment was inserted into the BlgII/EcoRV sites of pKC796Hyg to get pHabrC1/2/3. Plasmid with abrC1 gene disrupted was got by digesting pHabrC1/2/3 with XhoI and religated (eliminating a fragment of 260 nt containing the promoter and the 5′ end of the gene), yielding pHabrC2/3. To disrupt abrC2 gene an inner fragment of 1180 nt was eliminated from pHabrC1/2/3 using SfiI/AgeI sites and by treatment with T4 DNA polymerase before ligation, the plasmid got was named pHabrC1/3.

The new plasmids were introduced into the corresponding strains by protoplast transformation as previously described [39].

Antibiotic determination

Antibiotic production was assayed on solid media as described below. Plates were inoculated with 103 spores streaked or added to a 5 µl drop. For CDA production the strains were grown on NA medium at 30°C for 2 days. Afterwards, the plates were overlaid with 5 ml of soft agar plus 60 mM Ca(NO3)2 inoculated with B. subtilis as the test microorganism (0.2 ml, 0.25 DO) and incubated at 30°C for 20 h. A replica plate without calcium was used as a negative control. For ACT production on solid media, the strains were grown on different media (YEPD, R2YE, NMMP, NA) at 30°C for at least 3 days to observe the blue halo around the colonies. RED production was detected on PGA medium after 2 days as the red colour of colonies.

The ACT and RED antibiotic productions were quantified in liquid cultures using the standard spectrophotometric method [39] with minor modifications. 15 ml of medium were inoculated with 4×106 spores/ml. Culture samples were mixed with 1N KOH overnight at 4°C, centrifuged (15000 g, 10 min), and A640 of supernatants were determined to quantify ACT (ε640 = 25320). To quantify RED, pellets were washed twice in 0.5 M HCl and extracted in 0.5 M HCl-methanol for 2 h. After centrifugation (15000 g, 5 min), supernatant' absorbance were measured (ε530 = 100500). Dry weight of samples at different times was measured to monitor culture growth.

Microarrays assays

For RNA extraction from S. coelicolor wild-type and ΔabrC1/C2/C3 mutant strains, NA plates covered with a cellophane sheet were inoculated with 7.5×106 spores and incubated at 30°C for 50 h. Prior to RNA isolation using a RNeasy Midi Kit (Quiagen) the mycelia was harvested and suspended in RNA-protect Bacteria Reagent (Qiagen). An additional step with RNase free DNase (Qiagen) was incorporated to remove any contaminating DNA. The quality and concentration of RNA were assayed using gel electrophoresis and spectrophotometer assays (Q-bit and Agilent bioanalizer). Four biological replicates were used.

cDNA versus gDNA microarrays experiments were chosen due to the advantages described elsewhere [47], [48]. The S. coelicolor SCo40 microarrays used were obtained from the Functional Genomics Laboratory of Surrey University (UK) [49]. The Pronto! Universal Microarray Hybridization kit (Corning, # 40026) was used for pretreatment and prehybridization. Cy3-cDNA and Cy5-gDNA labelling reactions were performed according to the recommendations described by [49]. Hybridization assays were done as in Rodríguez-García et al. [28] and TIFF images were generated by Genepix DNA Microarray Scanner 4000B and processed with Genepix Pro 4.0 software. Bioconductor software package limma (linear models for microarray analysis) and rank products were used to analyse and assess the statistical significance of the data [28], [50]. Background correction was applied using the normexp function. Then, the log of Cy3/Cy5 intensities were normalized using block-weighted medians and global loess. The different p-values of the contrast between both strains were corrected for multiple testing FDR (false discovery rate) or by the rank products pfp method (proportion of false positives). To consider a gene differentially expressed, it should have passed at least one of these criteria: limma FDR-corrected p-value<0.05 or rank products pfp value<0.05. All data is MIAME compliant and the raw data has been deposited in a MIAME compliant database (ArrayExpress, accession number E-MEXP-2841)

Semiquantitative RT-PCRs

RT-PCR assays were performed with 200 ng RNA in a final volume of 20 µl with the Superscript™ One-Step RT-PCR with Platinum® Taq System Kit (Invitrogen). The primers used are specified in Table S3. Reactions were made as follows: 30 min at 55°C (cDNA synthesis); 2 min at 95°C; 20–40 cycles: 45 sec at 94°C, 30 sec at 65°C and 40 sec at 65°C; 10 min at 72°C. To check the DNA absence in the RNA samples, similar reactions avoiding the cDNA synthesis step were done in parallel. 2 µl of each reaction were run in 1.6% agarose gel buffered with TAE 1×. Each set of reactions was repeated varying the number of cycles to ensure that the PCR had not reached the plateau phase. As a positive internal control RT-PCR of 16S RNA was used. RT-PCR band images were quantified using Quantity One Analysis software 4.6.6 (Bio-Rad).

Supporting Information

Table S1.

Identity percentages among the sensor kinases by a local alignment (Emboss).


Table S2.

Identity percentages among the response regulators by a local alignment (Emboss).


Figure S1.

Growth curves of the different strains in NMMP (A), NB (B) and PGA (C) S. coelicolor M145 (triangles), S. coelicolor ΔabrA1/A2 (circles) and S. coelicolor ΔabrC1/C2/C3 (squares). Error bars correspond to standard deviation of two independent experiments measured by duplicate.


Figure S2.

Phenotypes of strains expressing abrA1/A2 in multicopy plasmid. A: Effect of the expression of abrA1/A2 genes by the high copy number plasmid pNXabrA derived from pN702GEM3 on NMMP medium. Top: morphological differentiation. Bottom: ACT production. B: CDA bioassays on NA medium, RED production on PGA medium, and morphological differentiation on YEPD medium (2 days), in the different strains.


Figure S3.

Phenotypes of strains expressing abrC1/C2/C3 in multicopy plasmids. A: Effect of expression of abrC1/C2/C3 genes by the high copy number plasmid pNabrC derived from pN702GEM3: ACT production on NA medium, CDA bioassays on NA medium, RED production on PGA medium, and MD morphological differentiation on YEPD medium (2 days), by the different strains. B: Effect of expression of abrC1/C2/C3 genes by the low copy number pAbrC plasmid derived from pHJL401: ACT production on NA medium, CDA bioassays on NA medium, RED production on PGA medium, and MD morphological differentiation on YEPD medium (3 days), by the different strains.



We thank Jose Manuel Fernández-Ábalos for his technical advice and support. Thanks are also due to MJ Jiménez Rufo for her excellent technical work. Thanks to D. Posner for English language revision.

Author Contributions

Conceived and designed the experiments: MD RIS AY . Performed the experiments: AY SR. Analyzed the data: AY AR-G MD. Contributed reagents/materials/analysis tools: AR-G. Wrote the paper: MD RIS AR-G.


  1. 1. Challis GL, Hopwood DA (2003) Synergy and contingency as driving forces for the evolution of multiple secondary metabolite production by Streptomyces species. Proc Natl Acad Sci U S A 100: Suppl 214555–14561.
  2. 2. Flamez C, Ricard I, Arafah S, Simonet M, Marceau M (2008) Phenotypic analysis of Yersinia pseudotuberculosis 32777 response regulator mutants: new insights into two-component system regulon plasticity in bacteria. Int J Med Microbiol 298(3–4): 193–207.
  3. 3. Scaramozzino F, White A, Perego M, Hoch JA (2009) A unique GTP-dependent sporulation sensor histidine kinase in Bacillus anthracis. J Bacteriol 191(3): 687–692.
  4. 4. Wang L, Tian X, Wang J, Yang H, Fan K, et al. (2009) Autoregulation of antibiotic biosynthesis by binding of the end product to an atypical response regulator. Proc Natl Acad Sci U S A 106(21): 8617–8622.
  5. 5. Ruiz D, Salinas P, López-Redondo ML, Cayuela ML, Marina A, et al. (2008) Phosphorylation-independent activation of the atypical response regulator NblR. Microbiology 154(Pt 10): 3002–3015.
  6. 6. Kato H, Chibazakura T, Yoshikawa H (2008) NblR is a novel one-component response regulator in the cyanobacterium Synechococcus elongatus PCC 7942. Biosci Biotechnol Biochem 72(4): 1072–1079.
  7. 7. The Wellcome Trust Sanger Institute website. Available: Accessed 2011 Apr 25.
  8. 8. Bentley SD, Chater KF, Cerdeño-Tárraga AM, Challis GL, Thomson NR, et al. (2002) Complete genome sequence of the model actinomycete Streptomyces coelicolor A3(2). Nature 417(6885): 141–147.
  9. 9. Hutchings MI, Hoskisson PA, Chandra G, Buttner MJ (2004) Sensing and responding to diverse extracellular signals? Analysis of the sensor kinases and response regulators of Streptomyces coelicolor A3(2). Microbiology 150(Pt 9): 2795–2806.
  10. 10. Kim D, Forst S (2001) Genomic analysis of the histidine kinase family in bacteria and archaea. Microbiology 147(Pt 5): 1197–1212.
  11. 11. Martínez-Antonio A, Collado-Vides J (2003) Identifying global regulators in transcriptional regulatory networks in bacteria. Curr Opin Microbiol 6(5): 482–489.
  12. 12. Gramajo HC, Takano E, Bibb MJ (1993) Stationary-phase production of the antibiotic actinorhodin in Streptomyces coelicolor A3(2) is transcriptionally regulated. Mol Microbiol 7(6): 837–845.
  13. 13. Takano E, Gramajo HC, Strauch E, Andres N, White J, et al. (1992) Transcriptional regulation of the redD transcriptional activator gene accounts for growth-phase-dependent production of the antibiotic undecylprodigiosin in Streptomyces coelicolor A3(2). Mol Microbiol 6(19): 2797–2804.
  14. 14. Ryding NJ, Anderson TB, Champness WC (2002) Regulation of the Streptomyces coelicolor calcium-dependent antibiotic by absA, encoding a cluster-linked two-component system. J Bacteriol 184(3): 794–805.
  15. 15. Chater KF, Chandra G (2008) The use of the rare UUA codon to define “expression space” for genes involved in secondary metabolism, development and environmental adaptation in Streptomyces. J Microbiol 46(1): 1–11.
  16. 16. Hesketh A, Chen WJ, Ryding J, Chang S, Bibb M (2007) The global role of ppGpp synthesis in morphological differentiation and antibiotic production in Streptomyces coelicolor A3(2). Genome Biol 8(8): R161.
  17. 17. Xu D, Seghezzi N, Esnault C, Virolle MJ (2010) The Over-expression of a Transcriptional Regulator of the TetR family Represses Antibiotic production and Sporulation in Streptomyces coelicolor. Appl Environ Microbiol.
  18. 18. McKenzie NL, Nodwell JR (2007) Phosphorylated AbsA2 negatively regulates antibiotic production in Streptomyces coelicolor through interactions with pathway-specific regulatory gene promoters. J Bacteriol 189(14): 5284–5292.
  19. 19. Sheeler NL, MacMillan SV, Nodwell JR (2005) Biochemical activities of the absA two-component system of Streptomyces coelicolor. J Bacteriol 187(2): 687–696.
  20. 20. McKenzie NL, Nodwell JR (2009) Transmembrane topology of the AbsA1 sensor kinase of Streptomyces coelicolor. Microbiology 155(Pt 6): 1812–1818.
  21. 21. Anderson TB, Brian P, Champness WC (2001) Genetic and transcriptional analysis of absA, an antibiotic gene cluster-linked two-component system that regulates multiple antibiotics in Streptomyces coelicolor. Mol Microbiol 39(3): 553–566.
  22. 22. Streptomyces database website. Available: Accessed 2011 Apr 25.
  23. 23. Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) website. Available: Accessed 2011 Apr 25.
  24. 24. Vicente CM, Santos-Aberturas J, Guerra SM, Payero TD, Martín JF, et al. (2009) PimT, an amino acid exporter controls polyene production via secretion of the quorum sensing pimaricin-inducer PI-factor in Streptomyces natalensis. Microb Cell Fact 8: 33.
  25. 25. Shu D, Chen L, Wang W, Yu Z, Ren C, et al. (2009) afsQ1-Q2-sigQ is a pleiotropic but conditionally required signal transduction system for both secondary metabolism and morphological development in Streptomyces coelicolor. Appl Microbiol Biotechnol 81(6): 1149–1160.
  26. 26. Chang HM, Chen MY, Shieh YT, Bibb MJ, Chen CW (1996) The cutRS signal transduction system of Streptomyces lividans represses the biosynthesis of the polyketide antibiotic actinorhodin. Mol Microbiol 21(5): 1075–1085.
  27. 27. Li YQ, Chen PL, Chen SF, Wu D, Zheng J (2004) A pair of two-component regulatory genes ecrA1/A2 in S. coelicolor. J Zhejiang Univ Sci 5(2): 173–179.
  28. 28. Rodríguez-García A, Barreiro C, Santos-Beneit F, Sola-Landa A, Martín JF (2007) Genome-wide transcriptomic and proteomic analysis of the primary response to phosphate limitation in Streptomyces coelicolor M145 and in a DeltaphoP mutant. Proteomics 7(14): 2410–2429.
  29. 29. Lu Y, Wang W, Shu D, Zhang W, Chen L, et al. (2007) Characterization of a novel two-component regulatory system involved in the regulation of both actinorhodin and a type I polyketide in Streptomyces coelicolor. Appl Microbiol Biotechnol 77(3): 625–635.
  30. 30. McKenzie NL, Thaker M, Koteva K, Hughes DW, Wright GD, et al. (2010) Induction of antimicrobial activities in heterologous streptomycetes using alleles of the Streptomyces coelicolor gene absA1. J Antibiot (Tokyo) 63(4): 177–182.
  31. 31. Xu W, Huang J, Lin R, Shi J, Cohen SN (2010) Regulation of morphological differentiation in S. coelicolor by RNase III (AbsB) cleavage of mRNA encoding the AdpA transcription factor. Mol Microbiol 75(3): 781–791.
  32. 32. Wang G, Inaoka T, Okamoto S, Ochi K (2009) A novel insertion mutation in Streptomyces coelicolor ribosomal S12 protein results in paromomycin resistance and antibiotic overproduction. Antimicrob Agents Chemother 53(3): 1019–1026.
  33. 33. Tanaka Y, Komatsu M, Okamoto S, Tokuyama S, Kaji A, et al. (2009) Antibiotic overproduction by rpsL and rsmG mutants of various actinomycetes. Appl Environ Microbiol 75(14): 4919–4922.
  34. 34. Hosaka T, Xu J, Ochi K (2006) Increased expression of ribosome recycling factor is responsible for the enhanced protein synthesis during the late growth phase in an antibiotic-overproducing Streptomyces coelicolor ribosomal rpsL mutant. Mol Microbiol 61(4): 883–897.
  35. 35. Li L, Guo J, Wen Y, Chen Z, Song Y, et al. (2010) Overexpression of ribosome recycling factor causes increased production of avermectin in Streptomyces avermitilis strains. J Ind Microbiol Biotechnol.
  36. 36. Hanahan D (1983) Studies on transformation of Escherichia coli with plasmids. Journal of Molecular Biology 166(4): 557–580.
  37. 37. Datsenko KA, Wanner BL (2000) One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci U S A 97(12): 6640–6645.
  38. 38. MacNeil DJ, Gewain KM, Ruby CL, Dezeny G, Gibbons PH, et al. (1992) Analysis of Streptomyces avermitilis genes required for avermectin biosynthesis utilizing a novel integration vector. Gene 111(1): 61–68.
  39. 39. Kieser T, Hopwood DA, Bibb JM, Chater KF, Buttner MJ (2000) Practical Streptomyces genetics. Norwich, UK: John Innes Foundation.
  40. 40. Sambrook J, Fritsch E, Maniatis T (1989) Molecular cloning: a laboratory manual. Cold Spring Harbor, N. Y.: Cold Spring Harbor Laboratory.
  41. 41. Hopwood DA, Bibb JM, Chater KF, Kieser T, Bruton CJ, et al. (1985) Genetic manipulation of Streptomyces: A laboratory manual. Norwich, UK: John Innes Foundation.
  42. 42. Gust B, Challis GL, Fowler K, Kieser T, Chater KF (2003) PCR-targeted Streptomyces gene replacement identifies a protein domain needed for biosynthesis of the sesquiterpene soil odor geosmin. Proc Natl Acad Sci U S A 100(4): 1541–1546.
  43. 43. Díaz M, Esteban A, Fernández-Ábalos JM, Santamaría RI (2005) The high-affinity phosphate-binding protein PstS is accumulated under high fructose concentrations and mutation of the corresponding gene affects differentiation in Streptomyces lividans. Microbiology 151(Pt 8): 2583–2592.
  44. 44. Adham SA, Campelo AB, Ramos A, Gil JA (2001) Construction of a xylanase-producing strain of Brevibacterium lactofermentum by stable integration of an engineered xysA gene from Streptomyces halstedii JM8. Appl Environ Microbiol 67(12): 5425–5430.
  45. 45. Fernández-Abalos JM, Reviejo V, Díaz M, Rodríguez S, Leal F, et al. (2003) Posttranslational processing of the xylanase Xys1L from Streptomyces halstedii JM8 is carried out by secreted serine proteases. Microbiology 149: 1623–1632.
  46. 46. Larson JL, Hershberger CL (1986) The minimal replicon of a streptomycete plasmid produces an ultrahigh level of plasmid DNA. Plasmid 15(3): 199–209.
  47. 47. Talaat AM, Howard ST, Hale Wt, Lyons R, Garner H, et al. (2002) Genomic DNA standards for gene expression profiling in Mycobacterium tuberculosis. Nucleic Acids Res 30(20): e104.
  48. 48. Gadgil M, Lian W, Gadgil C, Kapur V, Hu WS (2005) An analysis of the use of genomic DNA as a universal reference in two channel DNA microarrays. BMC Genomics 6(1): 66.
  49. 49. Faculty of Health and Medical Sciences website. Available: Accessed 2011 Apr 25.
  50. 50. Smyth GK, Michaud J, Scott HS (2005) Use of within-array replicate spots for assessing differential expression in microarray experiments. Bioinformatics 21(9): 2067–2075.
  51. 51. Paget MS, Leibovitz E, Buttner MJ (1999) A putative two-component signal transduction system regulates sigmaE, a sigma factor required for normal cell wall integrity in Streptomyces coelicolor A3(2). Mol Microbiol 33(1): 97–107.
  52. 52. Redenbach M, Kieser HM, Denapaite D, Eichner A, Cullum J, et al. (1996) A set of ordered cosmids and a detailed genetic and physical map for the 8 Mb Streptomyces coelicolor A3(2) chromosome. Molecular Microbiology 21: 77–96.