Streptomyces coelicolor M145 (prototroph; SCP1⁻, SCP2⁻) (Hopwood et al., 1985) was the original strain used to obtain all of the XRE/DUF397 genes and to generate the deletion mutants obtained in this study (S. coelicolor ∆SCO2246/45, S. coelicolor ∆SCO2253/52, S. coelicolor ∆SCO4678/79, S. coelicolor ΔSCO6236/35, S. coelicolor ΔSCO7615/16). This strain and the subsequent mutants were grown on LB, PGA, MSA, NMMP, R2YE, and YEPD solid media for carrying out the phenotypic assays, R2YE was used for transformation, MSA for sporulation, and YEPD for spore quantification. Escherichia coli DH5α and E. coli ET12567 (a dam⁻ strain) were used when constructing the various plasmids and for obtaining the DNA used to transform S. coelicolor (Sambrook et al., 1989; McLean et al., 2019). Transformation and manipulation of S. coelicolor and E. coli DNA were performed using the method described in Green and Sambrook (2012) and Kieser et al. (2000). Bacillus subtilis CECT 4522 was used in the antibiogram experiments. When necessary, the medium was supplemented with the antibiotics: ampicillin (100 μg mL⁻¹), kanamycin (50 μg mL⁻¹), chloramphenicol (25 μg mL⁻¹), or nalidixic acid (25 μg mL⁻¹) for E. coli; and neomycin (20 μg mL⁻¹) and apramycin (50 μg mL⁻¹) for S. coelicolor.

Total genomic DNA was extracted using the modified CTAB method (hexadecyltrimethylammonium bromide) (Cafaro et al., 2011). In total, 1.5 mL of mycelium in liquid medium was collected and transferred to a 1.5 mL screw cap tube and centrifuged at 13000 rpm for 5 min and the cells were washed with a 10.3% sucrose solution. The cells were then macerated using glass beads in a lysis solution (0.3 M Sucrose; 25 mM EDTA pH 8.0; 25 mM Tris–HCl pH 8.0) and digested with 500 μL of lysozyme (3 mg mL⁻¹) and RNAse (50 μg mL⁻¹) at 37°C. DNA extraction was performed in 500 μL of CTAB buffer, with a chloroform extraction step, and isopropanol precipitation was carried out for 30 min at −80°C. The DNA pellet was washed with 70% ethanol, and resuspended in TE buffer, and the integrity of the genomic DNA was verified by agarose gel electrophoresis.

Plasmid isolation, restriction enzyme digests, ligation, and transformation of E. coli and S. coelicolor were carried out using the methods developed by Green and Sambrook (2012) and Kieser et al. (2000), respectively. The oligonucleotide used to amplify the genes individually or together and those used to obtain the different mutations are listed in the Supplementary Table S1. Total DNA of the S. coelicolor M145 was used as the template to obtain all the studied genes. For each system, three intermediate E. coli plasmids were constructed using plasmid pXHis1 as the backbone (Adham et al., 2001) and named pXHisXXXX and pXHisXXXX/XX (where XXXX or XXXX/XX means the SCO or SCOs amplified). Constructs were made for each gene separately as well as constructs carrying both genes (Supplementary Table S2). All genes were under the control of the xylanase promoter isolated from S. halstedi (Rodriguez et al., 2005). Then, for overexpression in Streptomyces, the gene or genes were introduced into the shuttle plasmid E. coli-Streptomyces pNX4 (Rodriguez et al., 2005), generating the final plasmids named pNX-XXXX or pNX-XXXX/XX (XXXX or XXXX/XX means the SCO or SCOs amplified) (Supplementary Table S2).

Overexpression of the constructed plasmids was performed in S. coelicolor using five culture media: LB, NMMP, PGA, and R2YE. Phenotypic studies were carried out on solid media plates inoculated with 5 × 10⁵ spores added in a 5-μL drop and incubated at 30°C for several days to monitor changes in the color of the colony, in the culture medium, and morphology. Undecylprodigiosin (RED) production (red color) can be detected after 2 days as red colonies. Actinorhodin (ACT) production (blue color) can be observed as a blue halo around the colonies after 3 days of culture. All cultures were monitored photographically on each of the culture media used for 7 days. All experiments were performed in triplicate.

Antibiotic production was quantified from cultures in liquid LB medium inoculated with 4 × 10⁶ spores mL⁻¹. ACT and RED antibiotic production were quantified using the spectrophotometric method described by Yepes et al. (2011). All experiments were performed in triplicate.

For the bioassay, patches of the corresponding S. coelicolor strains were grown during 10 days at 30°C on solid R2YE, LB, NMMP, and PGA. Then, circular sections of the patches were collected and placed on LB plates inoculated with a lawn of B. subtilis. The plates were first incubated at 4°C for 5 h (to allow diffusion of any produced molecules) and then at 37°C for 24 h. The inhibition halos were measured. The error bars correspond to triplicate assays.

The CRISPR/Cas9 system, developed by Tong et al. (2015), was used to delete the two genes of the 6 systems selected (see results). The primers used to amplify and clone in the pCRISPR-Cas9 plasmid (Tong et al., 2015) the whole sgRNA guide by PCR are indicated in Supplementary Table S1, generating pCRISPR-Cas9-sgXXXX (XXXX corresponds to the SCO genes to delete in each system) (Supplementary Table S2). One kilobase (kb) on the left and one on the right of the corresponding XRE/DUF397 genes were amplified using genomic DNA. The final DNA fragment used as a template was obtained through an overlapping PCR method using the primers indicated (Supplementary Table S1). This fragment was digested with restriction enzyme NheI and introduced into the XbaI site of pCRISPR-Cas9-sgXXXX plasmids to generate the final plasmids pCRISPR-Cas9-XXXX. The plasmids generated were verified by sequencing them. Supplementary Table S2 is a list of all plasmids generated.

For each one of the XRE/DUF397 systems, the two plasmids generated, the one harboring only the sgRNA (pCRISPR-Cas9-sgXXXX) and the final plasmid with the guide and the template (pCRISPR-Cas9- XXXX), were introduced into E. coli ET12567 pUZ8002. Then, the plasmids were transferred to S. coelicolor by interspecific conjugation and by selecting Apra-resistant colonies. An empty plasmid (pCRISPR-Cas9) was used as the negative control. To induce plasmid loss, which was thermosensitive, and get the integration of the amplified DNA into the genome, the colonies obtained were grown in liquid TSB under agitation (200 rpm) at 37°C for 2 days. Apra-sensitive colonies were selected on R2YE solid media.

Genomic DNA was obtained from the S. coelicolor wild-type strain and from the putative M145 mutants and the correct deletion of XRE/DUF397 genes were verified by PCR using the corresponding oligonucleotides for each gene pair (Supplementary Table S1).

For the complementation analysis, both genes under their own promotor were introduced into the pKC796 integrative plasmid generating the pKCXXXX/XX plasmids (XXXX/XX corresponds to the SCOs genes of each system) (Supplementary Table S2).

As mentioned before, the S. coelicolor genome has 15 genes that encode proteins with an XRE domain that are clustered with genes encoding a small protein with a DUF397 domain (Tables 1, 2 and Supplementary Figures S1, S2). Throughout the text, these systems are referred to as XRE/DUF397.

The conservation of these S. coelicolor systems, at the protein level, among other species of Actinobacteria, was analyzed using the Actinoblast database that contains more than one hundred completely sequenced actinobacterial genomes and other outsiders such as Enterobacteriaceae and Bacillaceae² (Chandra and Chater, 2014). In Figure 1, it can be observed that the most conserved pair was SCO4441/4442, which was only absent in Micrococcineae and some of the Streptosporangineae species included in the comparison. Other pairs were less represented, such as systems SCO2245/2246 and SCO2252/SCO2253, which were only present in S. coelicolor and its relative S. lividans. In general, the degree of conservation was higher among Streptomyces spp. than with other actinomycetes. Two phylogenetic trees comprising the orthologous genes included in Figure 1 are included in Supplementary Figures S3 S4. In the XRE proteins, 6 robust groups were obtained with a frequency of ≥98 replicates while in the DUF397 only groups 5 and 6 were found with these characteristics (Supplementary Figures S3, S4). Conservation of the 15 Xre proteins and the 15 DUF397 proteins is shown in Supplementary FIgure S5.

The pair Scr1/Scr2 had been previously studied in our laboratory and was found to not behave as a toxin/antitoxin system as predicted, but rather as a potent activator of antibiotic production when overexpressed in the strain S. coelicolor M145 and other Streptomyces species (Santamaría et al., 2018). To analyze whether some of the other 14 members of this XRE/DUF397 family behaved the same as Scr1/Scr2, each gene was individually and jointly overexpressed in strain S. coelicolor M145.

Three multicopy plasmids were generated for each system with each gene under the control of the strong xysAp promoter (see Materials and Methods) (Rodriguez et al., 2005; Supplementary Table S2). The S. coelicolor M145 transformant strains of each construction were obtained and the effects of the overexpression of the three plasmids generated for each system was studied on solid R2YE medium and compared with the strain harboring the empty plasmid pN702Gem3 as the control (Fernandez-Abalos et al., 2003).

None of the S. coelicolor strains overexpressing the 14 putative toxins (DUF397 proteins) showed a clear negative effect on S. coelicolor viability, suggesting that these proteins did not produce toxic effects under the conditions used (Figures 2A–N). These results contradict the possibility of these pairs functioning as TA systems in the same way as Scr1/Scr2. However, on this medium, almost all systems exhibited differences, compared with the control, of greater or less intensity in the patterns of differentiation and/or production of ACT when at least one of their genes was overexpressed. Only systems SCO4543/42 and SCO6629/30 did not present appreciable differences with the control strain (Figures 2H–M).

Since the main interest of this study was to identify those systems involved in the regulation of antibiotic production, we selected for further study six of the systems showing higher or lower production of ACT compared to the control. Those with a drastic phenotypic alteration in morphological differentiation were also selected. The systems chosen for further study were: SCO2246/45, SCO2253/52, SCO4176/77, SCO4678/79, SCO6236/35, and SCO7615/16. So, the overexpression of the DUF gene (SCO2245) in pair SCO2246/45 showed a developmental delay and decreased ACT production. However, the overexpression of the XRE gene (SCO2246) showed an increase in ACT production (Figure 2B). In system SCO2253/52, the overexpression of the XRE (SCO2253) gene and both XRE/DUF genes showed a drastic developmental delay and the overexpression of the elements separately showed a slight increase in ACT production (Figure 2C). In system SCO4176/77, the separate overexpression of both genes, as well as their join overexpression, showed a clear bald phenotype but ACT production was the same as the control (Figure 2F). As for SCO4678/79, the individual overexpression of the XRE gene (SCO4678) and the two genes together, XRE and DUF, produced a bald phenotype and higher ACT production (Figure 2I). Conversely, for SCO6236/35, the overexpression of the XRE gene (SCO6236) produced a drastic reduction in ACT production. Additionally, a small delay in differentiation was also observed with the overexpression of this gene separately and the two together (Figure 2L). For system SCO7615/16, the overexpression of the XRE gene (SCO7615) led to lower ACT production, and the differentiation was delayed when both genes were overexpressed (Figure 2N).

To explore if the overexpression of the six selected systems (SCO2246/45, SCO2253/52, SCO4176/77, SCO4678/79, SCO6236/35, and SCO7615/16) had different phenotypes depending on the culture medium used, their effect was also studied on solid LB, NMMP, and PGA (specific for RED production) media (Figure 3).

The results showed that overexpression of the individual genes SCO2245 and SCO2246 reduced the production of ACT (blue color) on LB medium while no effect was detected for the complete SCO2246/45 system. On NMMP, no differences, compared with the control, were observed and, on PGA, only the XRE gene produced a small increase in RED production at 7 days of culture (Figure 3A). For the SCO2253/52 system, the reduction in ACT production was stronger with the overexpression of the gene SCO2253, or both genes SCO2253/52, than with SCO2252 on LB. Reduced ACT production was also observed on NMMP when the XRE gene alone or the complete system were overexpressed. On PGA, only the DUF397 gene produced a small increase in RED production at 5 and 7 days of culture. Also, a bald phenotype was observed on LB when gene SCO2253 was overexpressed (Figure 3B). The analysis of genes SCO4176/77 on LB revealed that the strongest reduction in ACT production occurred when gene SCO4176 was overexpressed. Reduced ACT production was observed on NMMP when the XRE or the DUF397 genes were expressed and on PGA only the DUF397 gene produced a small increase in RED production at 5 and 7 days of culture (Figure 3C). The overexpression of system SCO4678/79 led to a drastic reduction in ACT production when individual genes SCO4678 or SCO4679 were overexpressed on LB medium. When the gene SCO4679 (DUF397 gene) or both SCO4678/79 genes were overexpressed on NMMP, a clear increase in ACT production was observed. However, the overexpression of the SCO4678 (XRE gene) was associated with a reduction in ACT production on this medium. On PGA, only the expression of both genes led to a small increase in RED production at 5 and 7 days of culture (Figure 3D). The overexpression of gene SCO6236 or the complete system, SCO6236/35, induced a clear reduction in ACT production when the experiment was done on LB. Reduced ACT production was observed on NMMP when the XRE or DUF397 genes were expressed individually. On PGA, only the DUF397 gene showed a small increase in RED production at 5 and 7 days of culture (Figure 3E). The overexpression of system SCO7615/16 presented a slight decrease in ACT production when both elements were expressed individually or together. A reduction in ACT production was observed on NMMP when the XRE or DUF397 genes were expressed separately and, on PGA, no effect was observed (Figure 3F).

In summary, the overexpression of the six XRE/DUF systems acted as pleiotropic regulators influencing ACT and RED production and/or morphological differentiation. Therefore, in all of the systems analyzed, the overexpression of one gene or both of the genes in the pair provoked a reduction in ACT production on LB medium. Only one of the systems, SCO4678/4679, acted as a positive regulator on NMMP when the DUF gene (SCO4679) or the whole system were overexpressed. The rest of the systems provoked, in general, a minor reduction in ACT production on NMMP medium. Conversely, five of the systems had a slightly positive effect on RED production when cultured on PGA.

To complete the phenotypic study of the overexpression of the six selected XRE-DUF397 systems, the production of bioactive secondary metabolites induced under the conditions assayed was analyzed. Antibiograms against Bacillus subtilis of each of the constructions were performed. The inhibition halos were an indirect quantification of the antibiotics produced due to the overexpression of the transformants grown on each medium used: R2YE, LB, NMMP, and PGA. As shown in Figure 4, variations in biological activity were observed and were culture media dependent.

In system SCO2245/46, only a decrease in the size of the halo was observed when mutants overexpressing gene SCO2245 (DUF) were grown on LB or SCO2246 (XRE) on NMMP (Figures 4B,C).

System SCO2253/52 exerted a negative regulation over antibiotic production on three of the media used: R2YE, LB, and PGA. This reduction was observed with the overexpression of the individual gene and the gene pair (Figures 4A,B,D). However, on NMMP, the individual overexpression of the two elements of this system produced more biological activity than the strain carrying the empty plasmid. However, no biological activity was observed when both genes were overexpressed together (Figure 4C). In order to corroborate the result obtained on LB medium, liquid cultures of the control strain and the strains with the overexpression of the individual gene and the gene pair were made (Supplementary Figure S6). Clearly, the strains that individually overproduced each protein had a strong reduction in ACT production, which was less intense when both genes were expressed. These data are in agreement with the results obtained on solid media as represented in the Supplementary Figure S6. In fact, there was an increase in RED production in liquid (at 4 days) and on solid media (over time) with the strain carrying the plasmid pNX2252.

For the system SCO4176/77, only a small decrease in the inhibitory halo was observed when the overexpressing strains harboring the DUF gene were grown on R2YE and PGA (Figures 4A,D).

For SCO4678/79, the increase in the bioactive products on NMMP was remarkable when the DUF gene or both genes were overexpressed (Figure 4C). This corresponded to the increase in ACT production previously observed on this medium (Figure 3D). In addition, an increase was observed when both genes were overexpressed on PGA.

A general negative effect was observed in the transformants carrying one or two genes of system SCO6235/36 on R2YE medium (Figure 4A), which was also the same on LB and NMMP when XRE or both were overexpressed (Figures 4B,C) and on PGA with the XRE gene (Figure 4D). On the other hand, on NMMP, there was an increase when the DUF gene was the one that was overexpressed (Figure 4C).

Finally, when the SCO7615/16 system was overexpressed either individually or as a pair, a small decrease in antibiotic production was observed (Figures 4A,B,D). This decrease was more evident on NMMP medium, which lacked biological activity when the genes were individually overexpressed (Figure 4C).

To further investigate the role of the six XRE/DUF397 systems selected (SCO2246/45, SCO2253/52, SCO4176/77, SCO4678/79, SCO6236/35, and SCO7615/16) both genes of each system were deleted in S. coelicolor M145 using the CRISPR-Cas9 system (Tong et al., 2015; see Materials and methods).

Deletions were obtained in all systems except for the gene pair SCO4176/77, as it was not possible to obtain deletion mutants even after several attempts. The mutants of the other five systems were verified by PCR (Supplementary Figures S7–S11) and their phenotypes were analyzed. Drops containing 5 × 10⁵ spores were inoculated on five different solid media (R2YE, LB, NMMP, YEPD, and MSA), and their development and the production of colored antibiotics were monitored for 7 days (Figure 5).

The ∆SCO2246/45 mutant did not show any relevant phenotypic differences on any of the culture media tested (Figure 5A).

The ∆SCO2253/52 mutant showed a drastic reduction in ACT production when grown on R2YE and YEPD. Also, on YEPD there was a clear delay in the development of aerial mycelium. Conversely, this mutant presented a small increase in ACT production on NMMP (Figure 5B).

On the other hand, the ∆SCO4678/79 mutant did not show any clear phenotypic differences on R2YE, while on LB and NMMP more ACT was produced after 5 days. This increase in ACT production was also observed on YEPD on day 7 (Figure 5C).

Likewise, the ∆SCO6236/35 mutant did not show any clear phenotypic differences on the R2YE, LB, YEPD, or MSA media; however, on NMMP it produced more ACT than the wild type, the difference being much clearer on day 7 (Figure 5D).

Finally, the ∆SCO7615/16 mutant showed a different phenotype on each of the media used. On R2YE and YEPD, a bald phenotype was observed, even after 7 days, with little or no ACT production, as compared to the wild-type strain. On LB medium, lower production of ACT was also observed although the spores produced aerial mycelium. However, on NMMP medium, the mutant produced a greater amount of ACT. Finally, on MSA medium, the mutant produced more ACT and pigments than the wild-type strain (Figure 5E).

The complementation of these mutant strains was carried out using the integrative plasmid pKC796 harboring the corresponding deleted genes for each case. The phenotypic assays were carried out on those mutants which had the most dramatic differences compared to the wild-type strain; ∆SC2253/52, ∆SCO4678/79 and ∆SCO7615/16. The cultures were done on the medium on which was observed the strongest differences between the mutant and wild type strain. As shown in Figure 6, there were complementation of the phenotypes presented by the mutants ∆SCO2253/52 on YEPD medium, and those of the ∆SCO4678/79 and ∆SCO7615/16 on R2YE. Based on these results, it was concluded that the deletion of these genes was responsible for the phenotypes observed.