Escherichia coli strain BW25113 (pIJ790) (containing the λRed system) (Datsenko and Wanner, 2000) and non-methylating ET12567 (pUZ8002) (harboring the tra genes in the non-transmissible RP4-derivative plasmid pUZ8002) (MacNeil et al., 1992) were used for PCR-targeted mutagenesis of S. coelicolor M145. E. coli DH5α and ET12567 (a dam^- strain) were used to obtain the DNA to transform S. coelicolor. For CDA bioassays, a parent strain of Bacillus subtilis (CECT 4522) was grown as an overlay on NA medium (Hopwood et al., 1985). S. coelicolor M145 (prototroph; SCP1^-, SCP2^-) and the Δaor1 mutant strain were grown on R2YE, MSA, YEPD, NMMP, PGA, R5 and LB solid media for transformation, sporulation, spore quantification, and phenotypic assays (Fernández et al., 1998; Kieser et al., 2000). LB medium (Sambrook et al., 1989) was used for RNA-Seq experiment. When necessary, the medium was supplemented with antibiotics: E. coli media - ampicillin (100 μg mL^-1), apramycin (50 μg mL^-1), kanamycin (50 μg mL^-1), chloramphenicol (25 μg mL^-1), or nalidixic acid (25 μg mL^-1) and S. coelicolor media – neomycin (20 μg mL^-1) or thiostrepton (10 μg mL^-1).

Plasmid isolation, restriction enzyme digestion, ligation, and transformation of E. coli and S. coelicolor were carried out using the methods of Green and Sambrook (2012), and Kieser et al. (2000), respectively. The plasmids and cosmids used are listed in Supplementary Table S1.

REDIRECT PCR-targeting technology was used to replace the coding region of the SCO2281 gene with an apramycin resistance cassette [aac(3)IV gene] in cosmid SCC30 (Gust et al., 2003). The primers used to amplify the mutagenesis cassette (SRG-28 and SRG-30) with pIJ773 as the template are listed in Supplementary Table S2. The mutated cosmid SCC30 ΔSCO2281::acc(3)IV (ΔSCC30-1) obtained in E. coli BW25113(pIJ790) were demethylated in the ET12567(pUZ8002) strain and transferred by conjugation to S. coelicolor M145. The desired double recombinants carrying apramycin resistance and sensitive to kanamycin (the selection marker for the vector sequences) were selected. Southern blotting and PCR assays confirmed the deletion of the SCO2281 gene in S. coelicolor M145.

ACT and RED antibiotic production was assayed on solid LB medium plates inoculated with 10⁵ spores added in a 5 μL drop at 30°C. RED production was detected after 2 days as the red color of colonies. ACT production was observed after 3 days of growth as a blue halo around the colonies. For CDA, after 2 days of growth on NA, the plates were overlaid with 5 mL of soft agar plus Ca(NO₃)₂ (70 mM) inoculated with B. subtilis as the test microorganism (0.2 mL, 0.25 DO) and incubated at 30°C for 24 h. A replica plate without calcium was used as a negative control. These experiments were all performed in triplicate.

Antibiotic production was assayed from cultures in liquid LB medium inoculated with 10⁶ spores mL^-1 at 30°C. ACT and RED antibiotic production were quantified using the spectrophotometric method described in Yepes et al. (2011). For CDA, plates of NA with and without Ca(NO₃)₂ (70 mM) were inoculated on the surface with 200 μL of B. subtilis (DO₆₀₀ 0.25) using sterile glass balls. Afterward, a sterile cork borer was used on these plates to make wells with a diameter 0.8 cm. 150 μL of the supernatant of a 48 h Streptomyces culture were added into the wells and incubated at 30°C for 24 h.

The integrative plasmid pSETaor1 (Supplementary Table S1) was constructed by cloning the aor1 gene and its promoter region in the shuttle Streptomyces integrative plasmid pSET152t. First, the aor1 sequence was cloned in the bifunctional E. coli–Streptomyces plasmid pXHis1 in the NdeI and XhoI sites, remaining under the control of the xysA promoter and yielding the intermediate plasmid pXHisaor1. Second, the PCR-amplified aor1 promoter, cloned with the oligonucleotides SRG-37 and SRG-38 and the NdeI and EcoRI sites, replaced the xysA promoter, obtaining the pHisaor1 vector. This promoter corresponds to the 545 bp upstream region of SCO2282 that forms an operon with aor1 gene. Finally, the BglII–BglII fragment of pHisaor1, containing the aor1 gene plus its promoter, was cloned into the previously de-phosphorylated BamHI site in pSET152t, obtaining the integrative plasmid pSETaor1.

The multicopy plasmid pNXaor1 was obtained by cloning the NdeI/HindIII fragment from the plasmid pXHisaor1 into the same sites of pNX24 (pN702GEM3 derivative). To construct the multicopy plasmid pNaor1, its own promoter (upstream region of SCO2282) previously amplified by PCR was cloned as an EcoRI/NdeI fragment into the same sites of pNXaor1 (Supplementary Table S1).

Specific primers for the genes tested were designed using the Primer3 web-based tool² (Supplementary Table S2). Five microgram of the RNA samples were treated with RNAase-free DNaseI (Promega) according to the manufacturer’s instructions. One microgram of the resulting RNA was used as template for cDNA synthesis using iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad) in 20 μL reaction volumes. The samples were diluted 1:1 with distilled water and 2 μL was used in the quantitative PCR reaction with 10 pmol of forward and reverse primer and 5 μL of SsoAdvanced SYBR Green Supermix (Bio-rad).

Each assay was performed in duplicate using the CFX96 Touch^TM Real-Time PCR Detection System (Bio-Rad). Control PCRs were included to check that there was neither DNA (without RT, enzyme negative control) nor environmental contamination. Absolute quantification was performed using decimal serial dilutions of S. coelicolor genomic DNA with a known number of copies as standard.

Streptomyces coelicolor M145 and S. coelicolor Δaor1 were grown in LB medium inoculated with 10⁶ sp mL^-1 for 36 h at 28°C. The cells from three biological replicates were collected and treated with RNA protect, and afterward, RNA was extracted with RNeasy Mini Kit (Qiagen) following the company instructions. RNA samples quantity and quality were performed by Nanodrop and Bioanalyzer (Agilent). All samples passed the quality threshold of RIN > 7.

Macrogen Inc. (South Korea) provided the RNA-Seq data. The rRNA was removed with Ribo-Zero rRNA Removal Kit (Bacteria) and one library per sample was made using the Illumina TruSeq Stranded Total RNA kits. An independent library was constructed and sequenced for each triplicate. Sequencing was carried out in a HiSeq 2500 (Illumina) instrument with 100 bases PE reads. Reads were quality trimmed at Q35 with the script prinseq-lite, and only those that exceeded 50 bases in length were kept. Singletons were also excluded from the clean read set. For each library, the number of reads pairs obtained ranged between 21.2 and 24.4 millions, and around 99% passed the quality filter.

Mapping was carried out with the program Bowtiew2 using default settings and the Streptomyces coelicolor A3 (2) reference genome (GenBank NC_003888). SAM file was converted to BAM format using samtools, then sorted and indexed. Between 95 and 98% of the cleaned-read set was successfully mapped to the reference genome.

Reads for rRNA origin were detected counting those that mapped to the genome coordinates of the genes 16S, 23S and 5S. In all libraries the proportion of ribosomal reads was below 8%.

For the differential expression analysis, strand specific reads were counted for each CDS feature annotated in the genome reference using HTSEQ v0.6.1p1. A table containing the raw read counts for each CDS feature was generated and used for the subsequent comparative analysis using the BIOCONDUCTOR package EDGER v3. The analysis was performed using the protocol described for RNA-Seq experiments in the user’s guide revised in April 20th 2016.

The RNA-Seq data analyzed here has been deposited in the Sequence Read Archives (SRA, Accession code: PRJNA380047).

The gene aor1 (SCO2281) from S. coelicolor has 687 nucleotides and 73.36% GC content and encodes a putative NarL orphan RR containing two domains: a REC domain (position 9–128) containing the phosphoaceptor Asp from the HK (CheY-homologous receiver) and an HTH-LuxR DNA binding domain (position 156–218) (SMART database³). The encoded protein contains 228 amino acids, with an isoelectric point of 5.47 and a molecular weight of 24.68 kDa. It is highly conserved in all of the Streptomyces spp. sequenced to date, with identities ranging from 99 to 75%. Since Aor1 is an orphan RR, there is no HK codifying gene within its genomic context (Figure 1A). The phylogenetic circular dendrogram of S. coelicolor RRs (P2CS database) shows that aor1 (SCO2281) and abrC3 (SCO4596) are closely related and share 59% protein sequence identity (Figure 1B). This fact points to the possibility of crosstalk between the HKs AbrC1 and/or AbrC2 of the AbrC system and Aor1. In addition, the other closely related RR gene SCO2358 shares 55.7% protein sequence identity with Aor1, and forms part of an operon with the HK SCO2359. Remarkably, the P2CS database also considered the three genes comprising the AbrC TCS as orphans because the intergenic space between the genes is large enough to contain a promoter region. Moreover, the Prediction of Interaction Specificity in Two-Component Systems website of Biozentrum⁴ showed that there were 20 orphan kinases and 29 orphan regulators in S. coelicolor, including the AbrC genes. The existence of a common RNA has been shown for the two HKs AbrC1 and AbrC2, however, no common RNA has been obtained for the HK AbrC2 and the RR AbrC3 (Rodríguez et al., 2015). Using this prediction tool, the HKs AbrC1 (SCO4598) and AbrC2 (SCO4597) were in forth and fifth place as being probable partners with Aor1 (after SCO3750, SCO6424, and SCO0211).

The functional role of Aor1 was analyzed replacing the gene by an apramycin resistance cassette, using the REDIRECT technology (Material and Methods). This mutant strain exhibited a phenotype of an hypo-production of the antibiotics ACT, RED, and CDA, and a delayed morphological development in different media (Supplementary Figure S1), with more extreme phenotypes than those reported for the ΔabrC mutant strain (Rico et al., 2014a). Due to the drastic phenotype (Figure 2A), and highly reproducibility found in LB medium, complementation assays and RNA-seq analysis (see below) were performed in that medium. These results suggest that Aor1, as previously shown for AbrC3, acts as a positive RR of antibiotic production and differentiation. The Δaor1 mutant phenotypes were complemented by the single copy integrated plasmid pSETaor1 (Supplementary Table S1), which discarded the possibility of polarity effects caused by the mutation as the origin of the phenotypes (Figure 2B).

As shown in Figure 1A, aor1 is located in the vicinity of three genes; two of them (SCO2279 and SCO2282) have been annotated as hypothetical proteins, while SCO2280 has been annotated as a TetR family transcriptional regulator.

In order to determine if all four genes were co-transcribed and participants of a common regulation, qRT-PCR experiments were carried out to detect transcripts within the intergenic regions between the genes. It was found that the intergenic regions were not large enough to contain promoters regions, except for the region between SCO2279 and SCO2280 and SCO2280 and SCO2281. qRT-PCR experiments using primers designed to these intergenic regions (Supplementary Table S2) were positive and showed the amplification of a transcript in all cases (Figure 3), revealing the existence of a transcript that included all of the genomic region from SCO2279 to SCO2282. This result suggested the existence of an operon with the promoter region located upstream of SCO2282.

Overexpression of the aor1 gene was performed under the control of the SCO2282 upstream region, which corresponded to the promoter region of the aor1 operon. The plasmid pNaor1 was obtained (Material and Methods) and introduced into the M145 wild-type strain. Antibiotic production was compared to the strain transformed with the empty plasmid used as control on LB agar plates and in liquid LB medium. The high copy number plasmid used to overexpress of the aor1 gene did not provoke an increase in the production of the antibiotic ACT, as would be expected if it were a positive regulator. This result suggests that Aor1 could possibly be exerting negative control over other negative regulator/s, a situation that will be addressed in the “Discussion” section of this paper (Supplementary Figure S2).

To study the role of the Aor1 RR in a gene regulatory cascade in S. coelicolor, we compared the transcriptome of S. coelicolor M145 with S. coelicolor Δaor1. Stranded RNA-Seq was performed using RNA from samples of each strain, in triplicate, at 36 h (Material and Methods). The principal component analysis (PCA) and the sample distance matrix showed a distinct gene expression signature between M145 and the Δaor1 mutant strains (Supplementary Figures S3, S4). According to the PCA, the first principal component accounted for 82.5% of the variance, and clearly separated both groups. Moreover, the global analysis of the data showed that the absence of the aor1 gene provoked a huge impact on the transcriptome (Figure 4 and Supplementary Figures S3, S4).

To facilitate the analysis of the RNA-Seq data, an additional analysis was performed but only on those transcripts whose levels changed at least three-fold in the Δaor1 mutant compared with the levels in M145. Five hundred and four genes showed high differential expression (287 down-regulated and 217 up-regulated) between both strains in the conditions studied, with a significant p-value ≤ 0.002 and FDR ≤ 0.01. A complete list of the genes corresponding to these transcripts can be found in Supplementary Tables S3, S4. These selected differentially expressed genes were categorized into seven general clusters of functional categories, using David Gene Ontology analysis terms as the reference (Huang da et al., 2009a,b), and are listed in Table 1. The relevance of the differentially expressed genes and their correlation with the phenotypic changes observed in the Δaor1 mutant were analyzed (see below). Several genes were chosen to perform q-RT-PCR and validate the RNA-Seq results (Supplementary Table S5).

In addition, samples were collected in parallel at 36, 48, 62, and 134 hours (h) to quantify antibiotic production in the tested growth conditions. ACT production was significantly detected at 48 h in the wild-type strain M145, and this production increased exponentially at 62 and 134 h. However, the Δaor1 mutant strain presented very small production even at 62 h, and had a delayed trigger in ACT production at 134 h, yielding slightly lower levels than that of the M145 strain at this time (Figure 5A). Regarding RED production, a high level of this antibiotic was produced by M145 at the first time point of 36 h, with a maximum of production at 48 h. RED production in the mutant strain was reduced in all of the times analyzed (Figure 5B). Bioassays against Bacillus subtilis were used to test the CDA production using the supernatants of 48 h cultures (Material and Methods). There was a small inhibition halo in the M145 strain without calcium that could be due to the production of the other antibiotics (RED and ACT). This halo was much more considerable with plates containing calcium, corresponding to CDA production. In the mutant strain, no inhibition halo was observed in the two conditions used (+Ca and –Ca), indicating that the mutant strain produced less of this antibiotic (Figure 5C).