All Streptomyces and Escherichia coli strains used in this work are listed in Table 1. Spores were harvested from SFM solid plates (Kieser, 2000) after growth at 30°C for 7 days. The differentiation analyses were carried out on GYM plates covered with cellophane and on minimal medium (MM) plates supplemented with glucose (10 g/l), both inoculated with 10⁷ spores from a fresh water suspension and cultured at 30°C. The samples for quantification of actinorhodin and undecilprodigiosin production were obtained from 100-ml sucrose-free R5A (Fernández et al., 1998) cultures grown at 30°C and 200 rpm in 500-ml flasks. Calcium-dependent antibiotic (CDA) production was measured on nutritive agar from Oxoid. E. coli strains were cultured in LB and 2xTY media at 37°C. The following antibiotics were added to select plasmid-bearing and mutant strains: ampicillin (100 μg/ml), apramycin (100 μg/ml for E. coli, 25 μg/ml for S. coelicolor), chloramphenicol (25 μg/ml), hygromycin (100 μg/ml for E. coli, 200 μg/ml for S. coelicolor), kanamycin (50 μg/ml), and nalidixic acid (25 μg/ml).

Genomic DNA isolation and conjugation were performed following standard methods (Kieser, 2000; Sambrook and Russell, 2001).

Total RNA samples were isolated as previously described (Rioseras et al., 2016) using RNeasy Mini spin columns and treated with DNase I (Qiagen). The quantity and integrity of the RNA samples were measured with Nanodrop 2000 (Thermo Scientinfic) and 2100 Bioanalyzer (Agilent).

The SCO4117 ORF was replaced with the apramycin resistance cassette. The fragments upstream (UP-1944pb) and downstream (DW-2234pb) of SCO4117 were amplified by PCR from S. coelicolor DNA using the primers SCO159-SCO160 and SCO163-SCO164, respectively. The PCR products were cloned and sequenced in pCR™-Blunt II-TOPO® obtaining pTOPO-UP and pTOPO-DW. The DW fragment was liberated from pTOPO-DW by EcoRV/PstI digestion and subcloned in pTOPO-UP digested with the same restriction enzymes. The resulting plasmid, pTOPO-UPDW, was linearized with PstI and ligated with a PstI-fragment from pIJ773 containing an apramycin resistance cassette with an oriT, obtaining pΔSCO4117. This vector was introduced in the wild-type strain by conjugation and double recombinants were selected as kanamycin-sensitive and apramycin-resistant. Southern hybridization and PCR amplification with SCO134 and 4117R (Table 1) primers were performed to check the ΔSCO4117 mutant.

To discard polar effects in the expression of the SCO4117 neighbor genes due to the apramycin gene insertion, we recreated the SCO4117 knockout by CRISPR-Cas9. We used the system designed by Tong et al. (2015). The 20-nt target sequence was selected inside the SCO4117 and amplified by PCR with the primers SCO215 and sgRNA-R. The product of 110 bps was digested with NcoI/SnaBI and cloned in NcoI/SnaBI-digested pCRISPR-Cas9, obtaining pCRISPR-SgSCO4117. The SCO4117-surrounding regions were amplified by PCR with SCO216/SCO217 and SCO218/SCO219 primers. The couple of DNA fragments were combined by overlap extension PCR (Lee et al., 2010) with the primers SCO216/SCO219. The PCR product was cloned and sequenced in pCR™-Blunt II-TOPO®. The insert was released with EcoRV and cloned into pCRISPR-SgSCO4117 digested with StuI. The final vector pCRISPR-SCO4117A was introduced into the S. coelicolor wild-type strain by conjugation. The conjugants harboring the plasmid were selected using apramycin resistance, plated in SFM and grown at 37°C for 3 days for plasmid clearance. Mutations were confirmed by PCR using the primers Mut4117F and Mut4117R (Table 1).

The SCO4117 mutant harboring the nucleotides encoding for the first 305 amino acids of SCO4117, those including the sigma factor domain (Figure 1), was created by CRISPR-Cas9. The 20-nt target sequence was selected inside the SCO4117, but outside the first 305 amino acids (Figure 1) and amplified by PCR with the primers SCO215/sgRNA-R. The SCO220/SCO221 and SCO222/SCO219 primers were used (primer SCO222 contain the SCO4117 stop codon in frame with the first 305 amino acids) to amplify the SCO4117 sigma factor-surrounding regions. DNA fragments were combined by overlap extension PCR with the primers SCO220/SCO219. The amplicon was cloned and sequenced in pCR™-Blunt II-TOPO®. The insert was released with EcoRV and cloned into pCRISPR-SgSCO4117 digested with StuI, obtaining vector pCRISPR-SCO4117B. Conjugation, mutagenesis, and mutant confirmation was performed as described above (Table 1).

The complementation of S. coelicolor ΔSCO4117 was performed via the integration of plasmid pMS82 (Gregory et al., 2003), harboring the SCO4117 ORF and an upstream region large enough to include the two promoter regions identified by Jeong et al. (2016). The SCO4117 was amplified from the S. coelicolor chromosome using the primers SCO134 and 4117R. The 2171 bp-fragment was cloned in pCR™-Blunt II-TOPO® and sequenced to check the absence of mutations. A SpeI fragment obtained from TOPO-4117 was ligated into pMS82 digested with SpeI resulting in pMS82-SCO4117. This plasmid was transferred by conjugation to the S. coelicolor ΔSCO4117 mutant. The conjugants harboring the plasmid were selected by hygromycin resistance and the correct plasmid insertion at the SCO4848 attB site was verified by PCR using the primers SCO4848F and pMS82R (Table 1).

Restoration of the wild-type genotype of the ΔSCO4117 mutant was achieved using the CRISPR-Cas9 system for Streptomyces designed by Tong et al. (2015). The 20-nt target sequence was selected inside the apramycin resistance cassette, replacing SCO4117 and amplified by PCR with the primers SCO197 and sgRNA-R. The 120-bps product was digested with NcoI/SnaBI and cloned in NcoI/SnaBI-digested pCRISPR-Cas9 obtaining pCRISPR-120. A single fragment of 3.7 kb, containing the complete SCO4117 sequence and the surrounding regions, was amplified by PCR from S. coelicolor chromosome using SCO4117-FA and SCO4117-R primers. The product was then cloned and sequenced in pCR™-Blunt II-TOPO®. The resulting plasmid was digested with EcoRV/SpeI to liberate a 3.7 kb fragment. The SpeI-end was filled with the Klenow enzyme and the product was cloned into pCRISPR-120 linearized with StuI. The final vector pCRISPR-4117, was introduced in S. coelicolor ΔSCO4117 by conjugation. Conjugant selection and plasmid clearance were performed as described above. Restoration of the wild-type genotype was checked by the loss of apramycin resistance (Table 1).

Samples were obtained from GYM plates covered with cellophane at different developmental stages. The bacteria were stained with SYTO 9 and propidium iodide (LIVE/DEAD Bac- Light Bacterial Viability Kit, Invitrogen, L-13152) and observed under a Leica TCS-SP8 confocal laser-scanning microscope at wavelengths of 488 and 568 nm excitation and at 530 (green) or 640 nm (red) emissions.

Undecylprodigiosin and actinorhodin were quantified spectrophotometrically according to Tsao et al. (1985) and Bystrykh et al. (1996). For actinorhodin quantification, KOH was added to the culture samples at a final concentration of 1N. Cellular pellets were discarded by centrifugation and actinorhodin concentration was spectrophotometrically determined at 640 nm applying the linear Beer–Lambert relationship (ε₆₄₀ = 25,320). The culture samples for undecylprodigiosin quantification were vacuum-dried, resuspended in methanol, acidified with 0.5N HCl and spectrophotometrically assayed at 530 nm, using the Beer–Lambert relationship to estimate concentration (ε₅₃₀ = 100,500).

CDA production was determined via a bioassay against Bacillus subtillis. Oxoid nutritive agar (ONA) plates (90 mm in diameter) were inoculated with 5 μl of a Streptomyces spore suspension at 1 × 10⁵ spores/ml and incubated at 30°C. After 2 days, the plates were overlayed with 5 ml of soft ONA (0.75% agar), inoculated with B. subtilis (OD = 0.25) and supplemented with Ca(NO₃)₂ (60 mM). Negative controls were performed in parallel without adding calcium. Inhibitory halos were measured after 15 h at 30°C.

Growth was determined by measuring the protein concentration with the Bradford assay (Biorad) and a bovine serum albumin standard (Sigma). Total protein extracts were obtained mixing a volume of culture with a volume of 1 M NaOH, boiling for 5 min, and removing cell debris by centrifugation at 7740 g.

Next-generation sequencing (NGS) was performed by Stab Vida (Caparica, Portugal) from two biological replicates using the ΔSCO4117 mutant and the S. coelicolor wild-type strain. Ribosomal RNA was depleted with the Ribo-Zero Bacteria Kit (Illumina), and the cDNA library construction was carried out using the TruSeq Stranded mRNA Library Preparation Kit (Illumina). The DNA was sequenced in the Illumina HiSeq 2500 platform using 100-bp paired-end sequencing reads (at least 20 M reads per sample). Raw data are available via the Gene Expression Omnibus database (accession GSE107661).

Bioinformatic analysis of the sequenced data was performed under the Linux operative system using the following software: FastQC to check the quality of the sequences, Cutadapt for trimming sequences, Bowtie2 for mapping with the Streptomyces coelicolor genome and Cuffdiff for differential expression test analysis (Trapnell et al., 2012). Variations in transcript abundances were considered significant if they had a p-value < 0.03 (Supplementary Table 1). Transcript abundances without significant variations (p-values higher than 0.03) are shown in Supplementary Table 2.

The transmembrane topology of the SCO4117 gene was analyzed using the Phobius software (http://phobius.sbc.su.se/). The SCO4117 orthologs were obtained from the Strepdb database (http://strepdb.streptomyces.org.uk/): SLI_4349 (S. lividans), SAV_3491 (S. avermitilis), SVEN15_3779 (S. venezuelae), SGR_3904 (S. griseus), and SCLAV_3143/ SCLAV_3144 (S. clavuligerus). Amino acid similarities were estimated using the software package Lalign (http://www.ch.embnet.org/software/LALIGN_form.html).

The 46 genes significantly down-regulated in the ΔSCO4117 mutant were grouped into 33 putative operons (Supplementary Table 1). A library of putative regulatory sequences has compiled from the 250 nucleotide-long sequences located upstream of the putative operons. BioProspector (Liu et al., 2001) was used to search for bipartite overrepresented motifs in these sequences. Searches were performed only in the forward strand, and the following parameters were varied iteratively: the lengths of the −35 and −10 motifs were varied between 5 and 7 nucleotides; the spacer length was varied between 15 and 20 nucleotides in 1-nucleotide intervals. From all generated motifs, the highest scoring bipartite motif was selected.The motif logo was created using Weblogo (Crooks et al., 2004).

A High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) was used to synthetize cDNA from 0.5 μg of RNA from two biological replicates. Real-Time PCRs of the SCO4117 gene were carried out on an ABI PRISM 7900 HT thermocycler (Applied Biosystems). The reactions were performed in triplicate, containing 2 μl of 2-fold diluted cDNA, 10 μl of SYBR Green PCR Master Mix (Applied Biosystems), and 300 nM of specific primers (RT-SCO4117-F/R) (listed in Table 1) in a final volume of 20 μl. The hrdB (SCO5820, amplified using primers hdrB-F/R) was used as reference since its expression showed no variation between strains in our RNA-seq results (Supplementary Table 2). The DNA contamination and primer dimer amplification were tested in negative controls replacing cDNA by RNA or water. Amplification conditions were as follows: 2 min at 50°C, 10 min at 90°C, 40 repetitions of 15 s at 95°C, and 1 min at 60°C. Primer efficiencies were measured using serial dilutions of genomic DNA as template and the relative quantification of gene expression was performed by the ΔΔCt method (Livak and Schmittgen, 2001).

The expression of the most differentially expressed genes between the SCO4147 mutant and the wild strain was validated by qRT-PCR: SCO5077, actVA2; SCO2478, reductase activated by actinorhodin; and SCO1178 epimerase. We also analyzed key genes from the actinorhodin (actII-4, SCO5085) and undecylprodigiosin (redF, SCO5898) clusters; one of the bald genes (bldN, SCO3323); two of the genes participating in the aerial hyphae formation (chpA, SCO2716 / chpH, SCO1675); and SCO0761, a hypothetical protein (primers are listed in Table 1). The correlation between transcript abundances quantified by RNA-seq and qRT-PCR was adequate (regression coefficient of 0.71) (Supplementary Figure 1).

Gene SCO4117, the only member of the ECF52 family present in S. coelicolor, encodes a conserved multidomain protein (average protein similarity of 68.4% among S. griseus, S. avermitillis, S. lividans, S. clavuligerus, and S. venezuelae). The SCO4117 harbors a putative ECF sigma factor domain (conserved domain database accession TRIGR02937, 86% average similarity), a putative zinc finger domain (pfam13490, 86% average similarity), a putative transmembrane domain (79% average similarity), a putative carbohydrate-binding module (smart00776, 74% average similarity), and a proline-rich region separating the transmembrane and the carbohydrate module (Figure 1). Surprisingly, in the case of the S. clavuligerus ortholog, there is a stop codon separating the carbohydrate-binding domain module ORF (SCLAV_3143) from the other domain ORFs (SCLAV_3144) (http://strepdb.streptomyces.org.uk/), perhaps due to an error in the sequence at the proline-rich region. Gene SCO4117 is localized downstream the actinorhodin positive regulator atrA (SCO4118), but both genes were described as being expressed independently (Uguru et al., 2005). This synteny is maintained in all Streptomyces species analyzed, with the exception of Streptomyces avermitilis, in which the SCO4117 and atrA orthologs (SAV_3491 and SAV_4110 respectively) are separated.

The S. coelicolor SCO4117 knockout mutant (ΔSCO4117) do not show a significant variation in growth (Figure 2A), or in the maximum amount of actinorhodin and undecylprodigiosin production (Figures 2B,C), in sucrose-free R5A liquid cultures. However, the mutant shows a delay in actinorhodin production compared to the wild-type strain (Figure 2B). The SCO4117 knockout mutant showed lower CDA production (Figure 2D) and a delay in MII differentiation (notice the discontinuities characterizing the MI hyphae in the mutant at 24 h) (Figure 2F) (Manteca et al., 2006) and sporulation (Figures 2E,F) in solid GYM solid cultures. The delay in development was not complemented when SCO4117 and an upstream region large enough to include the two promoter regions identified by Jeong et al. (2016) were introduced into the mutant using the integrative plasmid pMS82-SCO4117 (Figure 2G). As discussed below, the SCO4117 gene expression was not restored in the complemented mutant (the SCO4117 transcript abundance was 4-fold less than in the wild-type strain) (Figure 2H). Interestingly, the ΔSCO4117 phenotype was only restored when a copy of the SCO4117 ORF was introduced at the native position in the ΔSCO4117 mutant, using the CRISPR-Cas9 methodology (Figure 2I). This result discards the possibility of artifactual mutations at chromosomal positions different to SCO4117, generating the phenotypes observed.

As discussed below, the above data suggest a complex regulation of the expression of the SCO4117 gene. Polar effects in the expression of SCO4117 neighbor genes due to the apramycin-resistance gene transcription, are unlikely, because, as detailed below, we did not detect significant variations in the expression of the SCO4116 and SCO4118 genes in the ΔSCO4117 mutant compared to the wild-type strain (Supplementary Table 2). In order to further discard a possible effect of the apramycin-resistance gene in the phenotypes observed, we recreated the knockout mutant by CRISPR/Cas9. We eliminated the SCO4117 ORF, obtaining the CRISPR/Cas9-ΔSCO4117 mutant (strain “1” in Figures 3A,B). This mutant, shows a phenotype similar to the ΔSCO4117 mutant (strain “2” in Figures 3A,B).

The expression of SCO4117 was analyzed at three key developmental stages (MI, 17 h; aerial mycelia, 48 h; sporulating mycelia, 72 h) by qRT-PCR, in solid GYM cultures of the wild-type strain. As reported previously in our transcriptomic work (Yagüe et al., 2013b), SCO4117 is overexpressed during the sporulating stage in the S. coeliolor wild-type strain (Figure 3C).

A mutant expressing the sigma factor domain from the SCO4117 ORF was created by CRISPR/Cas9 (CRISPR/Cas9-σ mutant) (Figure 1). The CRISPR/Cas9-σ mutant shows a delayed actinorhodin production and lower CDA production than the wild-type strain (Supplementary Figure 2). Sporulation is also delayed, but it is faster than in the CRISPR/Cas9-ΔSCO4117 knockout mutant (Figure 4A). Interestingly, the expression of SCO4117 is slightly up-regulated (1.3-fold) in the CRISPR/Cas9-σ mutant compared to the wild strain (Figure 4B), suggesting that the sigma factor domain regulates the expression of its own gene.

The differences between the ΔSCO4117 mutant and the wild-type transcriptomes were analyzed at 44 h on solid GYM cultures, the developmental conditions preceding the differences observed in sporulation (Figures 2E,F). Fifty-six genes showed altered expression in the ΔSCO4117 mutant compared to the wild-type strain (Supplementary Table 1). As expected, the SCO4117 transcript was absent in the ΔSCO4117 knockout (Supplementary Table 1). The expression of the SCO4117 neighbor genes (SCO4116 and SCO4118) was not significantly affected (Supplementary Table 2). Particularly interesting is that 46 of 56 transcripts, 82% of the total, are down-regulated in the ΔSCO4117 mutant compared to the wild strain (green bars in Figure 5, Supplementary Table 1), suggesting a pleiotropic activator effect of the SCO4117 ECF which is absent in the knockout mutant. 7,679 transcripts did not show significant variations (p-values higher than 0.03) (Figure 5).

Several key developmental and physiological genes were differentially expressed in the ΔSCO4117 mutant compared to the wild-type strain (Figure 5; Table 2). The expression of genes involved in secondary metabolism (deoxysugar synthases and actinorhodin biosynthetic genes) was highly down-regulated (up to 0.003, i.e., up-regulated 333-fold in the wild-type strain). The expression of genes involved in aerial mycelium formation and sporulation (chaplins, rodlins, nepA, ssgB, SCO7449) was also highly down-regulated in the mutant (up to 0.005, i.e., up-regulated up to 200-fold in the wild-type strain) (Figure 5; Table 2). SCO4684 (encoding the ScoF3 cold shock protein) was down-regulated in the mutant. The other 22 transcripts down-regulated in the ΔSCO4117 mutant, included, in addition to SCO4117, transcripts of genes encoding enzymes, stress and secreted proteins, as well as uncharacterized proteins (Supplementary Table 1). The 10 transcripts up-regulated in the ΔSCO4117 mutant included: SCO0194, encoding a putative sigma factor; SCO2162, encoding a quinolinate synthetase; SCO6102, encoding a nitrite/sulphite reductase; and 6 transcripts from uncharacterized genes (Figure 5; Table 2).

The 46 genes encoding transcripts down-regulated in the ΔSCO4117 mutant (Figure 5) (Supplementary Table 1) might harbor SCO4117 motif bounds in their promoters. These genes were grouped into 33 putative operons (Supplementary Table 1) whose promoter regions were analyzed (see Materials and Methods). Twenty-nine promoters harbored the putative “CSGYN-17bps-SRHA” SCO4117 motif bound at their promoter regions (Figure 6) (Supplementary Table 1).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.