The strains and plasmids used in this study are listed in Table 1. Cultivation conditions, sporulation, two-parental conjugation and solid fermentation of S. chartreusis strains were performed as described previously (Gou et al., 2013). For isolation of chromosomal DNA and RNA, S. chartreusis NRRL3882 was grown in liquid TSBY medium at 30°C. For spore collection and two-parental conjugation, S. chartreusis NRRL3882 was cultivated at 30°C on solid SFM. TSBY medium was also used for growth curve analysis. Genomic DNA isolation of S. chartreusis strains were performed according to Kieser’s group (Kieser et al., 2000). Total RNA was isolated with an RNA isolation kit (Tiangen, China) according to the method of Hopwood’s group (Hopwood et al., 2010). Growth conditions, plasmid isolation, and manipulation of Escherichia coli strains were carried out as described by Sambrook and Russell (2001).

The complete sequence of the calcimycin biosynthetic gene cluster can be found in GenBank (accession No. HM452329). Similarity and conserved domain analyses were performed using NCBI BlastP (Johnson et al., 2008) and CD-search (Marchler-Bauer et al., 2015), respectively.

The calR3 gene in S. chartreusis was disrupted by inserting the apramycin resistance gene aac(3)IV with REDIRECT^R technology, as described previously (Gou et al., 2013), and the aac(3)IV-oriT cassette was amplified from the pIJ773 plasmid using primers calR3-F1 and calR3-F2 (Supplementary Table 1). The obtained PCR product (1.38 kb) was introduced into BW25113/pKD46 harboring the fosmid p16F9 target against the calR3 gene, which generated the mutant plasmid pJTU3791 (ΔcalR3). The introduction of pJTU3791 into the S. chartreusis strains by conjugation and selection of double-crossover mutant strains through apramycin resistance were performed according to Kieser’s group (Kieser et al., 2000). Validation of the recombinant strain was performed by PCR analysis using primers calR3-F3 and calR3-F4 (Supplementary Table 1).

For the complementation of strain GLX26(ΔcalR3), the intact calR3 gene was amplified from S. chartreusis NRRL 3882 genomic DNA with the primers calR3-F5 and calR3-F6 using a high-fidelity DNA polymerase (KOD-plus, TOYOBO). The resulting PCR fragment was cloned into an integrative plasmid, pJTU2170, that was derived from plasmid pIB139 (Huang et al., 2011), resulting in pJTU3795. The calR3-complemented strain GLX29 (ΔcalR3:calR3) was generated by introducing pJTU3795 into GLX26 (ΔcalR3) via conjugation.

The S. chartreusis NRRL 3882 and GLX26 strains were inoculated into TSBY media and incubated at 30°C for 72 h, then 2% of the seed cultures were inoculated into the liquid SFM media to incubate for an additional 5 days. The culture broths were filtered using gauze and the residues were washed with 50 ml TSBY medium to remove extracellular calcimycin adhered to the medium particles. The collected filtrate and the mycelia were extracted with ethyl acetate separately. The ethyl acetate was evaporated, redissolved in 1 ml methanol and analyzed using HPLC.

The S. chartreusis NRRL 3882 WT strain, GLX26 (ΔcalR3), and GLX29 (ΔcalR3:calR3) were grown on solid SFM medium at 30°C for 7 days. Culture pretreatment and HPLC analysis conditions were performed as described previously (Gou et al., 2013), with a slight modification: the linear gradient elution of methanol was from 65 to 100%.

Mycelia of S. chartreusis NRRL 3882 and GLX26 strains grown in TSBY medium were collected after 96 h of culturing and individually frozen in liquid nitrogen. Total RNA extractions were performed with a total RNA purification kit (Tiangen, China) following the operating manual. Total RNA samples were digested by DNase I to remove contaminating genomic DNA, and the purity and quantity of the RNA samples were measured using a ScanDrop 200 spectrophotometer and agarose gel electrophoresis. The DNA-free RNA samples (2 μg) were reverse-transcribed with the Fast Quant RT kit (Tiangen, China) according to the manufacturer’s instructions. Transcription analysis of intergenic regions of the cal genes was performed to determine polycistrons according to the methods of Xu’s group (Xu et al., 2013). Genomic DNA was used as a template to amplify all regions as positive controls, and DNase I-treated RNA was used as a negative control (primers in Supplementary Table 1). Gene transcriptional analysis was performed by real-time quantitative PCR, with the obtained cDNAs as templates and using the listed primers. The 16S rDNA gene was used as an internal reference. Experiments were performed using the SuperReal qPCR PreMix (SYBR Green) and analyzed with an ABI7900HT Real-Time PCR System using 384-well plates. Each reaction system (20 μl) contained template cDNA, primer pairs (each 300 nM), and 10 μl SuperReal SYBR Green PreMix. The PCR reaction conditions were the following: 95°C for 15 min followed by 40 cycles of 95°C for 10 s, and 60°C for 30 s.

A DNA fragment containing intact calR3 was amplified using KOD-plus DNA polymerase from S. chartreusis NRRL 3882 genomic DNA with primers 28aR3-F1 and 28aR3-F2. The obtained PCR fragment was double-digested using NdeI and EcoRI and then ligated with linearized vector pET28a(+) (Novagen) at the corresponding restriction sites to generate pET28a-calR3. After sequencing validation, pET28a-calR3 was introduced into E. coli BL21(DE3)/plysE (Stratagene) for protein over-expression. Cell growth, induction, and harvest conditions were performed as previously described (Wu et al., 2013). Collected cells were re-suspended in binding buffer (20 mM Tris-Cl, pH 8.0, 300 mM NaCl) and disrupted by ultrasonication. After centrifugation, the supernatant was loaded onto a Ni²⁺-NTA column (Qiagen, Germany) that had been pre-equilibrated with binding buffer. The elution condition was a linear-gradient of imidazole from 0 to 500 mM. The purified protein was desalted by a HisTrap desalination column with buffer C (100 mM Tris-HCl, pH 8.0) and measured using SDS-PAGE. Finally, the purified His₆-tagged CalR3 protein was stored in stock buffer containing 10% glycerol at -80°C.

For the electrophoretic mobility shift assay (EMSA), target DNA probes were amplified from S. chartreusis NRRL 3882 genomic DNA with the primers shown in Supplementary Table 1 and incubated with purified His₆-tagged CalR3 protein at 37°C for 30 min in 20μl reaction buffer (20 mM Tris-Cl, 100 mM KCl, 10 mM DTT, pH 8.0). Electrophoretic conditions used by Marchler-Bauer et al. (2015) were applied. Gels were stained in 0.5 μg/ml ethidium bromide and analyzed with a gel electrophoresis imaging system.

The DNase I footprinting assays to determine CalR3 protein-binding sites were performed as described by Zianni’s group (Zianni et al., 2006). The 111-bp T-R3L DNA fragment was PCR amplified with primers P17-F1/P17-F2 and inserted into pGM-T vector (TIANGEN) at the EcoRV site, and then the plasmid was PCR-amplified using FAM labeled T7 primer and M13R primer to generate the probe. Probes were quantified using a NanoDrop 2000C after gel-purification. For each reaction volume (40 μl), 500 ng of labeled probes were incubated with various amounts of CalR3 protein at 25°C for 30 min. For partial digestion, about 0.6 units DNase I and 400 nmol CaCl₂ were added to the reaction mixture. Partial digestion, the electrophoresis conditions and data analysis were performed by the methods reported by Wang et al. (2014).

At 96 h of culturing of S. chartreusis, NRRL 3882 was used for total RNA extraction, and then total RNA (2 μg) was reverse-transcribed with 2.5 pmol of the gene-specific primer calTSP1 or calR3SP1 using a 5′-Full RACE kit (TAKARA). The generated cDNAs were recovered from agarose gels prior to addition of a poly(dC) tail at the 3′-end of the cDNA by treatment with terminal deoxynucleotidyl transferase. Tailed cDNA was directly amplified with the poly(dG) anchor primer AAP and a second internal gene-specific primer calTSP2 or calR3SP2. The PCR product was diluted 100-fold and then nest-amplified with an additional round of PCR using an AUAP anchor primer and a nested primer, calTSP3 or calR3SP3. Final PCR products were inserted into the cloning vector pMD18-T followed by sequence validation. The first base after the poly(dG) sequence was considered to be the TSP (Transcriptional Start Point).

Streptomyces chartreusis NRRL 3882 fermentation cultures grown in SFM liquid medium for 7 days were used for preparation of fermentation supernatant and purified cezomycin. After centrifugation, the culture supernatants were extracted twice with equal volumes of ethyl acetate. The solvent was then concentrated through rotary evaporation, and residues were redissolved in methanol and subjected to EMSAs. The calcimycin intermediate cezomycin used for EMSAs was isolated from S. chartreusis NRRL 3882 fermentation cultures as described previously (Gou et al., 2013).

The calR3 gene encodes a putative TetR family transcriptional regulator (TFR). BlastP analysis shows that CalR3 displays distinct similarity, especially at the N-terminus, to some TFRs such as MonRII in S. cinnamonensis (51% identity), AcrR in E. coli (22% identity), TylP and TylQ in S. fradiae (22 and 19% identities, respectively), and TetR (19% identity) (Supplementary Figure 1A). Conserved-domain prediction indicates that CalR3 contains a TetR-N-type helix-turn-helix (HTH) DNA binding domain at the N-terminus (Supplementary Figure 1B). There is no conserved domain predicted at the C-terminus, indicating that the induction mechanism for CalR3 is unknown.

To elucidate the role of CalR3 in calcimycin biosynthesis, the calR3 gene was inactivated using a PCR-targeted method (Kieser et al., 2000), and the calR3 disruption vector pJTU3791 was introduced into S. chartreusis NRRL 3882 via conjugation. For inactivation of calR3 in S. chartreusis NRRL 3882, an internal 455-bp DNA sequence in calR3 was replaced with the aac(3)IV-oriT cassette (Figure 2A). Candidate mutants with apramycin resistance were confirmed by PCR amplification. A 1,643-bp fragment was produced from the ΔcalR3 mutant strain GLX26 in contrast to a 714-bp DNA fragment from WT (Figure 2B). HPLC analysis showed that the calcimycin and cezomycin yields of GLX26 fermentative samples were improved by 8-fold and 30-fold, respectively, compared with WT strains (Figure 2C).

To demonstrate that these changes were due to deletion of calR3, pJTU3795, which harbors the intact calR3 gene, was introduced into GLX26. The resulting complemented strain GLX29 (ΔcalR3:calR3) restored the production of calcimycin and cezomycin (Figure 2C). Growth curve analysis of the WT, GLX26, and GLX29 strains in TSBY medium indicated that disruption of calR3 did not significantly affect cell growth (Supplementary Figure 2). Calcimycin and cezomycin yields from GLX26 were not due to altered cell growth, so CalR3 may function as a negative regulator in the calcimycin biosynthetic pathway.

Gene transcription analysis was performed to elucidate the role of CalR3 in calcimycin biosynthesis. The ORFs of calB2 to calB4, calN1 to calN3 and calA4 to calA5 were close neighbors (calB3-B2: 34 bp; calB4-B3: 39 bp; calN2-N3: 21 bp; calA5-A4: 27 bp) or slightly overlapped (calN1-N2: 4 bp overlaps), indicating that calB4-B2, calN1-N3 and calA5-A4 constitute transcriptional units, respectively. Amplifications of intergenic regions between other cal genes were performed to identify polycistrons. Data show that all the tested intergenic regions except positive control (calB4-B3), lacked positive PCR signals, indicating that all cal genes, except calB4-B2, calN1-N3, and calA5-A4, were located in independent transcriptional units (Supplementary Figure 3).

Real-time PCR was performed with cDNAs of GLX26, and WT strains were cultivated for 96 h (4 days, stationary phase, see Supplementary Figure 2) to investigate the possible effect of calR3 disruption on the transcription of cal genes. The PKS genes calA1, calA5 (co-transcripts with calA4), the benzoxazole gene calB4 (co-transcripts with calB3 and calB2), the α-ketopyrrole gene calN1 (co-transcripts with calN2 and calN3), the regulatory genes calR1, calR2, and calR3, the post-modification gene calM (Wu et al., 2013), and the resistance gene calT were selected for different locations and putative functions. Real-time PCR analysis showed that the transcription of calT and calR3 were dramatically increased by 171- and 30-fold, respectively, in GLX26 (ΔcalR3) compared with the WT strain (Figure 3).

calT encodes a putative antibiotic resistance protein located upstream of calR3 and is transcribed in the opposite direction (Figure 1B). Similar to AcrB, a multidrug efflux protein in E. coli, CalT is predicted to harbor two MMPL family domains to alter transmembrane transport (Supplementary Figure 1C). In addition, the gene arrangement of calR3 and calT is almost the same as that of the tetracycline-resistance genes tetR and tetA; therefore, CalT is hypothesized to export intracellular calcimycin and/or its derivatives out of cells. It intensively hints that CalR3 could regulate the transcription of calT and calR3 and further affect calcimycin and cezomycin yield.

To investigate the regulation site of CalR3, EMSAs were performed using DNA fragments containing intergenic regions in the gene cluster as substrates. Soluble full-length recombinant N-terminal His₆-tagged CalR3 protein with a theoretical molecular weight of 25.4 kDa was over-expressed in E. coli and purified (Figure 4A). According to transcriptional unit analysis, the 16 intergenic regions from calC to calU4 were amplified using their corresponding primers (Supplementary Table 1). Contiguous genes from calB4 to calB2, calN1 to calN3 and calA5 to calA4 were not tested since they are supposed to be co-transcripted, respectively, thus contain no regulation site at DNA level. The results showed that recombinant CalR3 protein had binding activity with the calT–calR3 intergenic region (T–R3) but not for any other of the 15 DNA probes investigated (Supplementary Figure 4 and Figure 4C). Because calT and calR3 are divergently transcribed, there may be bidirectional promoters located in the 156-bp intergenic region of T–R3. Thus, we redesigned two fragments: T–R3L and T–R3R (Figure 4B). Each fragment covers half of the enlarged T–R3 region and shares 18-bp overlap regions with adjacent fragments. EMSA assays on T–R3L and T–R3R showed that only T–R3L was bound by CalR3 with two retarded bands (Figure 4B). EMSA using increasing concentrations of CalR3 for T–R3L confirmed that there were two distinctly shifted bands, and that the T–R3L region might contain two CalR3 binding sites (Figure 4D). Different EMSA results between T–R3 and T–R3L may be due to the extended sequence at the 5′-terminus of T–R3L, which could contain a CalR3 recognition site. Precise recognition sites must be known to determine the mechanism of the CalR3 regulation of calT and/or calR3.

To illuminate the regulatory sites of CalR3, the 111-bp T–R3L region was PCR-amplified using FAM-labeled primers (P17-F1 and P17-F2, Supplementary Table 1) and analyzed using a DNase I footprinting assay in the presence/absence of His₆–CalR3. Two protected regions (sites I and II) were identified in the T–R3L region, and the shorter site, I, overlapped with the calT translational start codon (Figures 5A,B). This explained why T–R3L had two shifted bands, whereas T–R3 had only one site I in T–R3 that was at the leftmost end without any protective bases essential for efficient binding of CalR3. The next reasonable question is the accurate TSPs as well as the relative core promoter regions such as -10 and -35 sites of calT and calR3, according to which the detailed regulation mechanism of CalR3 would be illuminated.

TetR-family transcriptional regulator proteins typically form symmetric dimers and bind to palindromic sequences (Ramos et al., 2005). Palindromic analysis of the CalR3-binding regions revealed a 14 bp palindromic sequence in the site II region (Figures 5A,B). To assess the contributions of site I and the palindromic sequence in the site II region to CalR3 binding, EMSAs were performed with a WT probe m0 or a mutated sequence, as shown in Figure 5C, and the molar ratio of probe to CalR3 was properly determined to exhibit two shifted bands on the gel. The affinity of CalR3 for mutated probe m1 which lacked inverted repeats, was abolished completely. For probe m2, base changes were introduced into the site I region and only one retarded band was observed. The affinity of CalR3 for mutated probe m1&2, which was mutated in both inverted repeats and site I, was totally abolished (Figure 5D). These findings indicate that the 14 bp palindromic sequence in the site II region is essential for CalR3 binding, and CalR3 binding activity to site I depends on the existence of site II.

To find the TSPs of calT and calR3, 5′ RACE was performed using a commercial reverse transcription and cDNA amplification kit. For each gene, 10 clones were sequenced, followed by sequence alignment to determine the TSP site (Figure 5B and Supplementary Figure 5). The calR3 TSP is 104 bp upstream of the calR3 translational start codon, and the calR3 TSP is close to the CalR3 binding site II with only a one-base interval; its -10 and -35 regions, as well as the calT TSP and its -10 region, were completely covered by the CalR3 binding site II. Thus, CalR3 directly blocks its own transcription and that of the calT gene. Additionally, the CalR3 binding site I is located in the calT start codon region, indicating tighter regulation for calT transcription compared with calR3. These data are consistent with the observation that transcription of calT and calR3 was significantly increased after calR3 was knocked out, and calT is increased to a greater extent than calR3 (Figure 3).

Liquid TSBY medium was used to separate the extracellular and intracellular calcimycin. However, the solubility of calcimycin is very low in liquid medium, thus extracellular calcimycin would be trapped together with the medium particles during filtration. Therefore, we washed the mycelia adhere to the gauze to separate extracellular and intracellular calcimycin. HPLC analysis showed that the extracellular and intracellular calcimycin yields of GLX26 were increased 11- and 7-fold, respectively, compared with WT strains (Supplementary Figure 6). The ratio of extracellular to intracellular calcimycin of WT and GLX26 strains were 1:5 and 1:3, respectively, suggesting that disruption of calR3 enhanced the export of calcimycin.

Small-molecule compounds act as TFR ligands, typically play roles in the regulation of antibiotic biosynthesis and are related to target regulatory gene(s). Some CalR3 ligands were found to be involved in the regulation of calcimycin biosynthesis. CalT, a putative transmembrane efflux protein, was directly regulated by CalR3; it may pump out the CalR3 ligand(s) during fermentation. To verify this hypothesis, the extracted fermentation supernatant from WT strains was used to evaluate its effects on the affinity of CalR3 for the T–R3L region by EMSA. The results showed that the WT fermentation supernatant inhibited the DNA-binding ability of CalR3 in a concentration-dependent manner (Figure 6, left panel), indicating that there were some CalR3 ligands produced by S. chartreusis NRRL 3882 in the fermentation broth. EMSAs were performed (Figure 2C) using a calcimycin standard, and a purified cezomycin which differs from calcimycin because it lacks an N-methyl group on its benzoxazole ring (Figure 1A). The data showed that calcimycin-induced dissociation of CalR3 from the T–R3L region (1.9 mM) occurred in a concentration-dependent manner (Figure 6, middle panel). Moreover, cezomycin prefers dissociating CalR3 from target DNA regions even at 0.95 mM (Figure 6, right panel). Thus, the N-methyl group at the C-3 position of calcimycin can suppress the binding of calcimycin to CalR3, but more studies are needed.