Coelimycin inside out — negative feedback regulation by its intracellular precursors Authors: Magdalena Kotowska, Mateusz Wenecki, Bartosz Bednarz, Jarosław Ciekot, Wojciech Pasławski, Tomasz Buhl, Krzysztof J. Pawlik Published: 2024 ABSTRACT AbstractCoelimycin (CPK) producer Streptomyces coelicolor A3(2) is a well-established model for the genetic studies of bacteria from the genus Streptomyces, renowned for their ability to produce a plethora of antibiotics and other secondary metabolites. Expression regulation of natural product biosynthetic gene clusters (BGCs) is highly complex, involving not only regulatory proteins, like transcription factors, but also the products of the biosynthetic pathway that may act as ligands for some regulators and modulate their activity. Here, we present the evidence that intracellular CPK precursor(s) (preCPK) is involved in a negative feedback loop repressing the CPK BGC. Moreover, we provide a characterization of the cluster-encoded efflux pump CpkF. We show that CpkF is essential for the extracellular CPK production. In order to track down which CPK compounds — intra- or extracellular — are the ones responsible for the feedback signal, a luciferase-based reporter system was applied to compare the activity of 13 CPK gene promoters in the wild-type (WT) and two mutated strains. The first strain, lacking the CPK-specific exporter CpkF (ΔcpkF), was unable to produce the extracellular CPK. The second one did not produce any CPK at all, due to the disruption of the CpkC polyketide synthase subunit (ΔcpkC). All tested promoters were strongly upregulated in ΔcpkC strain, while in the ΔcpkF strain, promoter activity resembled the one of WT. These results lead to the conclusion that the CPK polyketide acts as a silencer of its own production. Supposedly this function is exerted via binding of the preCPK by an unidentified regulatory protein.Key points•Intracellular coelimycin precursor takes part in a negative cpk cluster regulation•CpkF exporter is essential for the extracellular coelimycin production•Simple method for the analysis of coelimycin P2 production in agar mediumGraphical abstract Supplementary InformationThe online version contains supplementary material available at 10.1007/s00253-024-13366-1. Introduction Bacteria from the genus Streptomyces remain undefeated leaders in their ability to produce antibiotics and other secondary metabolites (SM) of great importance to human and veterinary medicine. The majority of the SM biosynthetic gene clusters (BGCs) remain “silent” until the specific signals associated with the environmental conditions and the colony development state trigger their expression. Coelimycin (CPK) cluster of Streptomyces coelicolor A3(2) is a perfect example of such a “silent” BGC. The expression of CPK BGC is tightly controlled by a number of pleiotropic and cluster-situated regulatory proteins (Pawlik et al. 2010; Bednarz et al. 2019). It is switched on by a quorum sensing signalling via γ-butyrolactones (GBLs), which releases the ScbR repressor from the DNA and allows for the expression of CpkO and CpkN — cluster-situated activators. ScbR2 repressor, activated by CpkO, turns off the expression of the cluster. SMs and their biosynthetic intermediates, for example auricin (Kutas et al. 2013), sansanmycin (Li et al. 2013) and lidamycin (Li et al. 2015), can act as ligands controlling the expression of their cognate BGCs. Results of our previous work suggested the existence of a negative regulatory feedback loop involving the polyketide product of the CPK BGC (Bednarz et al. 2021). In the current work, we provide the evidence of the existence of such a feedback loop and show that the intracellular CPK precursor(s) participate in this mechanism. The scheme of CPK biosynthesis is shown in Fig. 1. The unsaturated 12-carbon chain synthesized from six malonyl residues by the type one modular polyketide synthase Cpk, composed of three multi-enzymatic subunits CpkA/B/C, is released from the megasynthase in the form of an aldehyde by the C-terminal thioester reductase domain of the CpkC subunit (Fig. 1, compound 1). It is converted into an amine derivative (Fig. 1, compound 3) by the ω-transaminase CpkG (Gomez-Escribano et al. 2012; Awodi et al. 2016). We propose the term preCPK for the 12-carbon intracellular coelimycin precursors. Further post-polyketide modifications are performed by the extracellular enzymes CpkD, ScF and CpkH. They include the formation of two epoxide rings and give rise to the compound with weak antibacterial activity (antibiotic CPK, abCPK) (Gottelt et al. 2010). Nonenzymatic reactions of the bis-epoxide coelimycin A (CPK A) with N-acetylcysteine or glutamate present in the medium result in the formation of coelimycins P1 or P2, respectively, which are both yellow pigments (yCPK) (Gomez-Escribano et al. 2012).Fig. 1Biosynthesis of coelimycin Biosynthesis of coelimycin The CPK biosynthetic gene cluster from S. coelicolor A3(2) includes the cpkF gene encoding a protein with high homology to multidrug efflux pumps from the major facilitator superfamily (MFS) (Pawlik et al. 2007). The MFS is the largest superfamily of transporters, consisting of more than 100 distinct families. It was recently expanded to include five more distantly related families that do not have known transport functions (Wang et al. 2020). MFS transporters use chemiosmotic energy, often from the transmembrane electrochemical proton gradient. They operate by the mechanism of symport, where two or more solutes (one of them being H+ or Na+) are transported in the same direction: antiport, where two or more solutes are transported in opposite directions; or uniport, where a single solute is transported. MFS members have 12 or 14 transmembrane α-helical segments (TMSs), with an overall pseudo-twofold symmetry of two distinct six-transmembrane helix bundles and a substrate accommodating cavity between them (Wang et al. 2020; Drew et al. 2021). Conserved sequence motifs were identified in MFS proteins (Paulsen et al. 1996b). Some of them are shared by all MFS members, while others are family specific. Interestingly, there is no clear motif distinguishing between substrate-specific and multidrug transporters. A number of studies have investigated the roles of single amino acids in substrate recognition and transport across membranes (reviewed in Kumar et al. (2020)). Another recent extensive review (Drew et al. 2021) provided a detailed mechanistic explanation of the alternating-access mechanisms of transport, based on the available MFS structures. Genomic analyses provide interesting data on the abundance of different transporter classes in industrially important bacteria (Lorca et al. 2007; Getsin et al. 2013; Zhou et al. 2016). As shown by a comparative analysis of 11 Steptomyces genomes, transporter genes constitute 10–13% of all ORFs (990 in the case of S. coelicolor A3(2)), and approximately one-third of them belong to the MFS (Zhou et al. 2016). Membrane transporters are often encoded within biosynthetic gene clusters, where they participate in the final steps of natural product synthesis and protect host cells producing toxic compounds (Severi and Thomas 2019). BGCs usually include at least one predicted transmembrane efflux protein. Transporter-encoding genes from biosynthetic gene clusters are often similar to those conferring antibiotic resistance in pathogenic bacteria. Transporter engineering has the potential to increase specialized metabolite production and is particularly important in hybrid biosynthetic systems construction (Severi and Thomas 2019). Despite their abundance, experimental data on the specific roles of secondary metabolite transporters are limited. The first part of the present study focuses on the characterization of CpkF transporter from the coelimycin BGC. In the second part, addressing the feedback mechanism, CpkF (transporter)- and CpkC (main polyketide synthase subunit)-deficient strains are used as tools for tracking down the coelimycin-dependent transcriptional changes in the CPK biosynthetic cluster. Materials and methods DNA manipulation and bacterial strain growth conditions Standard protocols were used for DNA manipulations (Sambrook and Russell 2001). Fragments amplified by PCR were first cloned into the p-GEM-T Easy vector (Promega), verified by DNA sequencing and cloned into appropriate plasmids. The oligonucleotides, plasmids and bacterial strains used in this work are listed in Tables S1, S2 and S3 in the supplementary material, respectively. The culture conditions, transformation and conjugation methods followed the general procedures for E. coli (Sambrook and Russell 2001) and Streptomyces (Kieser et al. 2000). S. coelicolor was cultivated on the following media: 79NG (Bednarz et al. 2021), Glu-MM (Kotowska et al. 2014) and MS (Kieser et al. 2000). Construction of cpkF deletion and complementation strains The cpkF gene (SCO6278) was replaced on the St1G7 cosmid with a hygromycin resistance cassette by PCR targeting (Gust et al. 2004). The modified cosmid St1G7-cpkFDM was then used for homologous recombination with the S. coelicolor M145 strain chromosome. The cpkF deletion mutant strain was named P112. The plasmids pWP4 and pIJ10257apra-cpkF were introduced into the P112 strain via conjugation with the E. coli ET12567/pUZ8002 strain to generate the complemented strains P127 and P321, respectively. The empty plasmids pWP3 and pIJ10257apra were introduced into the P112 strain to generate the control strains P126 and P322, respectively. HPLC analysis of coelimycin production Petri plates containing 20 ml of Glu-MM solid medium were inoculated with approximately 108 spores of S. coelicolor strains spread evenly on the agar surface. The plates were incubated at 30 °C for 27 h. The whole agar with biomass was frozen overnight at − 20 °C, thawed and centrifuged (10 min, 27,000 × g, 4 °C). The cleared supernatant was collected for HPLC analysis. HPLC analysis was performed as described previously (Kotowska et al. 2014). Reporter system assay A set of promoter regions of genes from the cpk biosynthetic cluster was amplified from pFLUXH derivatives (Bednarz et al. 2021) with the primers GFLUX-F and GFLUX-R and cloned into the pFLUX reporter plasmid linearized with HindIII and NdeI via Gibson assembly. The promoter region of the cpkK gene was amplified with the primers ksTF and p6284_Nde, ligated with pUC18 digested with SmaI and recloned into pFLUXH digested with NdeI and BamHI. Amplification with GFLUX-F and GFLUX-R primer pair and Gibson assembly was used to clone the pcpkK fragment into the pFLUX. The promoter regions of the cpkI, cpkJ and accA1 genes were amplified with respective primer pairs (Table S1) from the genomic DNA of S. coelicolor A3(2) wild-type strain M145 and cloned by Gibson assembly into either pFLUX or pFLUXH. pFLUX derivatives with promoter regions (Table S2) were individually introduced into the M145 and P112 strains by conjugation. pFLUXH derivatives were introduced into the P100 strain. Luciferase activity was monitored for 72 h in cultures grown on solid 79NG medium in a ClarioStar Plus microplate reader (BMG Labtech) as described previously (Bednarz et al. 2021). Antimicrobial activity test The antimicrobial activity of mono- and divalent cationic lipophilic compounds against S. coelicolor A3(2) strains was tested by the disc diffusion method. Two hundred microlitres of spore suspension in water (OD600 = 0.3) was spread on solid 79NG medium. Ten microlitres of water solutions of acriflavine hydrochloride (5 mg/ml), benzalkonium chloride (5 mg/ml), chlorhexidine diacetate (0.5 mg/ml) and pentamidine isethionate salt (5 mg/ml) was applied to Whatman GF/F sterile discs with a diameter of 6 mm. The discs were then placed on the surface of the inoculated plates. The inhibition zone diameter was measured after 48 h of incubation at 30 °C. Bioinformatic tools An amino acid sequence similarity search was performed with BLAST (Altschul et al. 1990). A phylogenetic tree of CpkF and its homologues was generated with Clustal Omega (Sievers et al. 2011) and visualized with iTOL (Letunic and Bork 2021). Consensus prediction of the CpkF topology was performed by TOPCONS (Tsirigos et al. 2015). The presence of signal peptides in protein sequences was analysed with SignalP 5.0 (Almagro Armenteros et al. 2019). Results The CpkF transporter is necessary for efficient yCPK production According to the biosynthetic pathway proposed by Gomez-Escribano et al. (2012), the precursor of CPK, (3Z,5S,6E,8E,10E)−1-aminododeca-3,6,8,10-tetraen-5-ol (Fig. 1, compound 3), in the form of a monovalent cation, is the predicted native substrate transported by CpkF. Further biosynthetic steps performed by extracellular enzymes lead to the formation of the colorless bis-epoxide CPK A, which can then form adducts with culture medium components. The use of the minimal medium containing glutamate (Glu-MM) allows the conversion of virtually all the precursor to CPK P2 and allows for convenient detection of the product by HPLC (Kotowska et al. 2014). To investigate the role of the CpkF transporter in the CPK production, the cpkF gene was replaced with a hygromycin resistance cassette, yielding the cpkF deletion strain P112. A lack of the transporter CpkF abolished yellow CPK production. Complementation of the deletion by the cpkF gene under the constitutive promoter ermEp* either on an integrating plasmid (strain P321) or on a high copy number plasmid (strain P127) restored yCPK production (Fig. 2A). This was confirmed by HPLC analysis (Fig. 2B, Table 1). A single peak of A427, which eluted at 9 min, was observed. The 9-min peaks from the M145, P127 and P321 samples were confirmed by mass spectrometry as CPK P2.Fig. 2The production of coelimycin by S. coelicolor A3(2) strains. The strain names are the same as in Table 1. (A) Cultures on 79NG medium incubated for 24 h. CPK is visible as a yellow pigment. (B) Representative HPLC chromatograms monitoring the absorbance at 427 nm from 27-h cultures on solid Glu-MM medium. The 9-min peaks from the M145, P127 and P321 samples were confirmed by mass spectrometry as coelimycin P2. (C) Close-up of the residual CPK P2 peaks in the HPLC chromatogramsTable 1The production of coelimycin P2 by S. coelicolor A3(2) strains after 27 h of culture on solid Glu-MM medium measured as the area under the 9-min peak. The data are shown as the means from three plates ± standard deviation. Figure 2 shows the chromatogramsStrainStrain descriptionPeak area (mAU*min)% of wild typeM145Wild type7.26 ± 1.06100P112ΔcpkF0.09 ± 0.011.3P127ΔcpkF + pWP4 (complemented)3.39 ± 1.3446.6P126ΔcpkF + empty pWP30.03 ± 0.010.4P321ΔcpkF with pIJ10257apra-cpkF (complemented)3.72 ± 0.5351.3P322ΔcpkF + empty pIJ10257apra0.13 ± 0.051.9P100ΔcpkC0.00.0 The production of coelimycin by S. coelicolor A3(2) strains. The strain names are the same as in Table 1. (A) Cultures on 79NG medium incubated for 24 h. CPK is visible as a yellow pigment. (B) Representative HPLC chromatograms monitoring the absorbance at 427 nm from 27-h cultures on solid Glu-MM medium. The 9-min peaks from the M145, P127 and P321 samples were confirmed by mass spectrometry as coelimycin P2. (C) Close-up of the residual CPK P2 peaks in the HPLC chromatograms Table 1The production of coelimycin P2 by S. coelicolor A3(2) strains after 27 h of culture on solid Glu-MM medium measured as the area under the 9-min peak. The data are shown as the means from three plates ± standard deviation. Figure 2 shows the chromatogramsStrainStrain descriptionPeak area (mAU*min)% of wild typeM145Wild type7.26 ± 1.06100P112ΔcpkF0.09 ± 0.011.3P127ΔcpkF + pWP4 (complemented)3.39 ± 1.3446.6P126ΔcpkF + empty pWP30.03 ± 0.010.4P321ΔcpkF with pIJ10257apra-cpkF (complemented)3.72 ± 0.5351.3P322ΔcpkF + empty pIJ10257apra0.13 ± 0.051.9P100ΔcpkC0.00.0 The production of coelimycin P2 by S. coelicolor A3(2) strains after 27 h of culture on solid Glu-MM medium measured as the area under the 9-min peak. The data are shown as the means from three plates ± standard deviation. Figure 2 shows the chromatograms Careful investigation of the HPLC plots revealed a low level of residual CPK P2, below 2% of the wild-type level, in the deletion strain P112 and its derivatives with empty plasmids (strains P126 and P322), as opposed to the P100 strain, which is unable to complete the synthesis of the polyketide backbone of the CPK precursor due to the knockout of cpkC gene encoding one of the main polyketide synthase subunits (Fig. 2C, Table 1). CpkF is a cluster-specific efflux pump dedicated to coelimycin CpkF, a 543 aa protein encoded by the SCO6278 gene from the CPK BGC of S. coelicolor A3(2), belongs to the major facilitator superfamily of membrane transport proteins according to its predicted amino acid sequence similarity. A BLAST search of the Transporter Classification Database (TCDB) (Saier et al. 2021) revealed high similarity to proteins from the Drug:H+ Antiporter-2 (14 Spanner) (DHA2) family (TC# 2.A.1.3) (Table S4). The closest homologues of CpkF are found within Streptomyces BGCs of unsaturated polyketides such as enediynes, angucyclines, anthracyclines and polyenes (Table S5, Fig. 3).Fig. 3A phylogenetic tree of CpkF and its homologues listed in Tables S4 and S5 generated with Clustal Omega (Sievers et al. 2011) and visualized with iTOL (Letunic and Bork 2021). The color labels denote the classes of the main compounds produced by the respective BGCs: red, enediyne; green, angucycline and anthracycline; blue, polyene macrolactam; black, other. MFS pumps not associated with BGCs are marked with a pink background A phylogenetic tree of CpkF and its homologues listed in Tables S4 and S5 generated with Clustal Omega (Sievers et al. 2011) and visualized with iTOL (Letunic and Bork 2021). The color labels denote the classes of the main compounds produced by the respective BGCs: red, enediyne; green, angucycline and anthracycline; blue, polyene macrolactam; black, other. MFS pumps not associated with BGCs are marked with a pink background Structurally, CpkF consists of 14 transmembrane α-helical segments, as predicted by the web server TOPCONS (Tsirigos et al. 2015), for membrane protein topology consensus prediction. The locations of the TMSs together with the hydropathy and amphipacity plots (Zhai and Saier 2001) are shown in Fig. S1. Seven conserved sequence motifs shared by MFS members and the DHA2 family can be found in the CpkF sequence (Table S6, Fig. S1). CpkF has a leucine residue (L320) in the position corresponding to D323 of QacA, an efflux pump from Staphylococcus aureus that confers resistance to a variety of both mono- and divalent lipophilic cationic antimicrobial compounds. Aspartic acid in TMS10 (D323) was shown by mutational studies to be necessary for QacA to accommodate divalent cations (Paulsen et al. 1996a). CpkF does not contain any acidic residues within TMS10 or TMS12, which could potentially facilitate the binding of divalent cations in the absence of the crucial Asp (Hassan et al. 2007). Accordingly, we expect CpkF to be able to transport monovalent cationic compounds. This is in line with the structure of its predicted native substrate from the coelimycin biosynthetic pathway (Fig. 1, compound 3). The susceptibilities of the wild-type S. coelicolor strain M145 and the cpkF deletion strain P112 to several compounds that are substrates of the QacA efflux pump, similar to CpkF, were compared by the disc diffusion method. Complementation strains P127 and P321, as well as their respective control strains with empty plasmids (P126 and P322), were also included in the experiments. Examples of monovalent (acriflavine and benzalconium) and divalent (chlorhexidine and pentamidine) lipophilic cationic compounds were tested. No significant differences were found between the strains (Fig. 4). This result indicated the high specificity of CpkF despite its similarity to multidrug efflux pumps.Fig. 4The susceptibility of S. coelicolor A3(2) strains to biocides (acriflavine, benzalconium, chlorhexidine and pentamidine) measured as the diameter of the growth inhibition zone. Dashed line represents the disc diameter (6 mm). The mean values and standard deviations of triplicate samples are shown. The strain names are the same as in Table 1 The susceptibility of S. coelicolor A3(2) strains to biocides (acriflavine, benzalconium, chlorhexidine and pentamidine) measured as the diameter of the growth inhibition zone. Dashed line represents the disc diameter (6 mm). The mean values and standard deviations of triplicate samples are shown. The strain names are the same as in Table 1 The intracellular coelimycin precursor(s) takes part in a negative regulatory feedback loop To assess the effect of the lack of extracellular CPK production on the CPK BGC expression in the transporter-deficient strain, the activities of 13 promoter regions were compared between the wild-type (M145) and the cpkF deletion (P112) strains. A luciferase reporter system was used as described previously (Bednarz et al. 2021). Figure 5 shows that the transcription of most of the biosynthetic genes in the strain lacking CpkF remained at the same level as that in the parental strain. Similarly, the production of the activators CpkO and CpkN, the repressors ScbR and ScbR2, and the GBL synthase ScbA was not significantly affected. The only observed difference was an approximately twofold reduction in the transcription of cpkK, accA1 which are involved in the substrate (malonyl-CoA) supply, and scoT, which encodes a type II editing thioesterase.Fig. 5Transcription profiles of cpk cluster biosynthetic (A) and regulatory (B) genes. Promoter activities were measured in a luciferase-based reporter system. The wild-type (M145), cpkF deletion (P112) and cpkC disruption (P100) strains were grown on solid 79NG medium. For clarity, the standard deviation is shown for every 10 h. The legend and axes descriptions refer to all charts Transcription profiles of cpk cluster biosynthetic (A) and regulatory (B) genes. Promoter activities were measured in a luciferase-based reporter system. The wild-type (M145), cpkF deletion (P112) and cpkC disruption (P100) strains were grown on solid 79NG medium. For clarity, the standard deviation is shown for every 10 h. The legend and axes descriptions refer to all charts To address the question of whether the intracellular precursor(s) of CPK could affect the cpk gene cluster transcription, the activities of the same promoters were measured in the cpkC disruption strain (P100), which is unable to finish the synthesis of the polyketide backbone of CPK due to the lack of the functional CpkC subunit of the polyketide synthase (Fig. 1). Transcription of the whole CPK BGC was enhanced in this strain (Fig. 5). Discussion The investigation of SM exporters can reveal the details of BGCs regulation by their cognate products. For example, the deletion of drrAB genes of the daunorubicin/doxorubicin exporter providing the antibiotic resistance to the producer, Streptomyces peucetius, uncovered the existence of a feedback transcriptional regulatory mechanism, where the lack of the transporter leads to the reduction of antibiotic production. Daunorubicin remaining in the cell intercalates in the DNA in a specific site and thus breaks the activatory cascade (Srinivasan et al. 2010). In S. coelicolor A3(2), two transporters, ActA and ActB, play a key role in a sophisticated two-step feed-forward mechanism of actinorhodin (ACT) production control. ACT biosynthetic intermediates induce actAB expression to make sure that the efflux pumps are in place to remove the antibiotic as soon as it is produced. The extracellular ACT is a signal by which the producing subpopulation of cells induces sustained actAB expression in the rest of mycelium and protects the whole culture from antimicrobial effects of the accumulating product (Xu et al. 2012). The present study focuses on investigating the role of coelimycin polyketide in the autoregulation of its BGC. Our study model consisted of two S. coelicolor A3(2) mutant strains: P112 (ΔcpkF) deficient in extracellular CPK production due to the lack of the transporter CpkF and P100 (ΔcpkC) with completely abolished production of the metabolite due to the lack of the functional CpkC subunit. This allowed us to track down which CPK compounds, present in the cytoplasm — preCPK, or outside of the cell — abCPK/yCPK, could possibly have a signalling role in the cluster autoregulation. The existence of a negative feedback loop exerted by the Cpk synthase product(s) was proposed in our previous article (Bednarz et al. 2021). It was based on the observed upregulation of CPK BGC, including the main activator CpkO, in the cpkN disruption strain (ΔcpkN). We considered it unlikely for CpkN to act as a repressor of the cpkO gene. Both CpkN and CpkO belong to the SARP (Streptomyces antibiotic regulatory proteins) family of regulators acting mostly as activators (Yan and Xia 2024) and they are both required for CPK production. CpkN is a key activator of scoT gene (Bednarz et al. 2021) encoding the editing thioesterase, necessary for CPK production, which works as a “cleaner” removing non-reactive acyl residues blocking the “assembly line” (Kotowska et al. 2014). Thus, ΔcpkN cannot produce CPK due to the lack of ScoT thioesterase. Indeed, the inability to produce CPK by this strain was reversed by providing ScoT expression from the ermEp* promoter. In the cpkO deletion strain (ΔcpkO), transcription of biosynthetic genes was obviously silenced due to the lack of the main activator — CpkO, but interestingly transcription from the native cpkO promoter was strongly enhanced. Therefore, we postulated that it was the lack of coelimycin synthesis that resulted in cpkO gene upregulation in both ΔcpkO and ΔcpkN strains, and that the activation of the remaining cpk genes in ΔcpkN was a consequence of the increased level of CpkO. In the current study the in vivo luciferase reporter system was utilized to measure the activity of most of the CPK BGC promoters in P100 (ΔcpkC) and P112 (ΔcpkF) strains. Obtained results let us conclude that the CPK cluster is negatively regulated by the intracellular preCPK compound(s), while the extracellular CPK versions do not exert this feedback role (Fig. 6). Upregulation of all tested promoters was observed in the P100 strain unable to synthesize preCPK. The lack of the transporter CpkF did not change markedly the transcription of cpk genes, even though the extracellular yCPK was not produced. Synthesis of preCPK in the P112 strain seems to be undisturbed.Fig. 6Autoregulation of the coelimycin biosynthetic gene cluster by its intracellular polyketide product. X represents an unknown receptor of preCPK ligand. (A) In native conditions the preCPK compound(s) acts as a negative feedback autoregulator. (B) Without the specific exporter CpkF intracellular precursors remain in the cytoplasm. The cluster transcription is similar to the wild-type strain with few exceptions of half-downregulated promoters indicating higher preCPK levels in the cell. (C) Loss of the CpkC synthase subunit leads to complete elimination of the CPK production. The cluster transcription is strongly upregulated in comparison to the WT strain Autoregulation of the coelimycin biosynthetic gene cluster by its intracellular polyketide product. X represents an unknown receptor of preCPK ligand. (A) In native conditions the preCPK compound(s) acts as a negative feedback autoregulator. (B) Without the specific exporter CpkF intracellular precursors remain in the cytoplasm. The cluster transcription is similar to the wild-type strain with few exceptions of half-downregulated promoters indicating higher preCPK levels in the cell. (C) Loss of the CpkC synthase subunit leads to complete elimination of the CPK production. The cluster transcription is strongly upregulated in comparison to the WT strain The observation of residual ~ 2% of CPK P2 production by the ΔcpkF strain indicates that a small amount of the precursor found its way outside the cells and was converted into the final product. This means that the whole machinery of initial biosynthetic steps is working properly. The preCPK compounds are present in the cell and their regulatory roles are fulfilled, which is reflected in the results of the reporter assay of the cpk gene promoters. We have shown that the cpkF gene deletion does not significantly affect the expression of biosynthetic and regulatory genes of CPK BGC (Fig. 5). An approximately twofold reduction in cpkK, accA1 and scoT promoters activity is unlikely to drastically reduce the production of preCPK. It is not clear if the precursor accumulates in the cell or is directed to other metabolic pathways, yet the observation of lowered cpkK, accA1 and scoT expression levels may suggest increased preCPK level in the cytoplasm and the negative regulation of the cluster. These three promoters could be the most susciptible to the metabolite-dependent regulation. There are two possible explanations of preCPK release without CpkF. The CPK production time is correlated with extensive mycelium rebuilding (Yagüe et al. 2013). The cellular envelopes integrity could have been disturbed by naturally occurring mechanical tensions or the damage was done during the sample preparation for the HPLC measurements. Some portion of preCPK could have simply leaked out of the cell. The other possibility is that the precursor was released by another efflux pump. A BLAST search against S. coelicolor A3(2) proteins revealed several MFS transporters with high sequence homology to CpkF (Table S7). However, given the rather undisturbed supply of preCPK and negligible CPK P2 production in the P112 strain, these transporters cannot efficiently replace CpkF. This may be caused by their different substrate specificities and/or different time and level of expression. The CpkF transporter seems to be dedicated to its specific substrate, despite its sequence similarity to multidrug efflux pumps such as QacA and QacB from S. aureus. Its deletion did not alter the susceptibility to biocides (Fig. 4), as opposed to other MFS efflux pumps, for example SCO4121 from S. coelicolor A3(2) (Nag and Mehra 2021). We suspect that preCPK can influence the transcription activity of the cpk cluster by acting as a ligand for a yet unidentified transcription factor denoted in Fig. 6 as an “X” protein. There are two groups of cluster-encoded regulators that hypothetically could exert the role of a CPK intermediate sensor. The first ones are TetR-family repressors ScbR and ScbR2, and the second ones are SARP activators CpkO and CpkN. It is well known that TetR-family proteins can bind low molecular ligands including products of their own BGCs (Takahashi et al. 1986). ScbR2 was shown by EMSA to dissociate from its target DNA in the presence of unecylprodigiosin and ACT (Xu et al. 2010). Considering the possibility that ScbR2 would be the preCPK-binding protein working in a high concentration of the metabolite(s), when a certain level of it is produced, it would have to increase its DNA-binding after interaction with the ligand(s), to make this regulation model work. Recently, it has been reported for the first time that SARPs can interact with biosynthetic intermediates or final products. Nosiheptide (NOS) BGC of Streptomyces actuosus is activated by the only cluster-situated regulator — NosP — belonging to SARPs. It has been shown that NosP dissociates from its target promoter sequence in the presence of NOS and its intermediate (Li et al. 2018). Moreover, upon the deletion of NosP, all biosynthetic genes were silenced, while transcription from the native nosP promoter was enhanced. This effect resembles cpkO upregulation in ΔcpkO and ΔcpkN strains (Bednarz et al. 2021). In S. actuosus mutant strain unable to produce the raw peptide precursor of NOS, upregulation of biosynhetic genes was observed, similarly as in the ΔcpkC strain (P100) described here. CPK cluster-situated SARP activators, CpkO and CpkN, could be possible sensors of preCPK. PreCPK could disturb SARP binding to DNA, thus autorepressing cpk cluster. CpkN, just as NosP, belongs to small SARPs composed of only two domains: N-terminal DNA-binding domain and C-terminal bacterial transcriptional activation domain (BTAD). CpkO belongs to medium SARPs which contain additionally a highly conserved nucleotide-binding domain NB-ARC (Yan and Xia 2024). It cannot be excluded that the true preCPK-binding protein is none of the so far mentioned ones. CPK BGC regulation is a highly intricate network, which was presented by our group in the past years (Bednarz et al. 2019, 2021). Regulators encoded outside of the cluster, such as response regulators from two-component systems could potentially bind preCPK as well. The cpk cluster encodes also an orphan histidine kinase CpkM (SCO6268), formerly OrfB, the expression pattern of which correlates with the other cpk genes. CpkM is a positive regulator of the cluster (Wenecki, unpublished), yet its interactome is unknown. It demonstrates that there might be plenty of so-far unidentified regulators, the presence of which is yet to be unraveled. The transporters encoded within SM BGCs are usually assumed to export the SM out of the cell; however, their exact roles in the final steps of the natural product synthesis, as well as the effects of export on the BGC regulation and on the host self-resistance, remain underexplored (Severi and Thomas 2019). One of the strategies to avoid suicide by antibiotic-producing cells is to use a biosynthetic intermediate to induce transcription of the transporter gene so that the efflux pump is in place to remove the toxic compound as soon as it is produced (Van Keulen and Dyson 2014). Although it is generally assumed that secondary metabolite biosynthesis is completed within the cell, there is evidence for the extracellular location of some biosynthetic enzymes (Chater et al. 2010). This is the case in coelimycin biosynthesis, where the polyene intermediate must be exported to make it available to the extracellular flavin-dependent epoxidases/dehydrogenases ScF, CpkD, and CpkH, which perform post-polyketide modifications, including double epoxidation, which is responsible for the toxicity of the compound (Gomez-Escribano et al. 2012). Microarray transcriptomic analysis of S. coelicolor A3(2) (Nieselt et al. 2010) revealed a strong transient peak in the expression of CPK BGC at approximately 24 h. Later, the genes directly related to the synthesis of the CPK polyketide backbone (cpkA/B/C) are switched down to low level expression, while the transcription of genes associated with post-polyketide modifications remains constantly elevated. The common transcription profile of cpkF and other biosynthetic genes (particularly co-transcription with cpkE, cpkG and one of the extracellular enzymes, cpkD) suggests its involvement in the export of the polyketide produced by Cpk synthase and fine synchronization of the intermediate export with further enzymatic steps of biosynthesis (Fig. S2). We have shown here that the so far univestigated MFS efflux pump CpkF is indeed involved in CPK export from the cell and is necessary for the production of the final compound (Fig. 2). To find clues about the roles of CpkF homologues in secondary metabolism, an amino acid sequence similarity search was performed against all proteins belonging to 2500 BGCs collected in the MIBiG repository (Medema et al. 2015) . The transporters most similar to CpkF were found to be involved in the biosynthesis of enediynes and other polyketides, such as angucyclines, anthracyclines and polyene macrolactams (Fig. 3). Although the final structures of the found compounds differ markedly from each other and from that of coelimycin P1, the presence of multiple conjugated double (and triple, in the case of enediynes) bonds is their common feature, which resembles the predicted polyene CPK precursor transported by CpkF. The unsaturated bonds of polyketide backbones are often involved in cyclization, oxidation and other reactions (Nett and Moore 2009; Gomez-Escribano et al. 2012; Ma et al. 2017). We consider it likely that CpkF homologues are reponsible for the export of biosynthetic intermediates, which are then modified by extracellular enzymes to yield the final product. The extracellular location of proteins can be predicted by the presence of signal peptides, short N-terminal sequences that direct newly synthesized proteins to secretion machinery and are subsequently cleaved off by specialized proteases (Almagro Armenteros et al. 2019). Indeed, analysis by SignalP software revealed the presence of proteins with signal peptides in the majority of BGCs containing CpkF homologues. A CpkD homologue was found in the enediyne sporolide biosynthetic cluster (BGC0000150, protein Strop_2710, ABP55154). ScF and CpkH homologue was found in the tetrahydroisoquinoline alkaloid naphthyridinomycin biosynthetic cluster (BGC0000394, protein AGD80628). In addition, we recommend the use of solid Glu-MM medium and the “freeze/thaw/centrifuge” procedure as a simple and reproducible method of sample preparation for HPLC analysis of CPK P2 production. Supplementation of the minimal medium with glutamate has been shown previously to promote the conversion of CPK precursors to CPK P2 and to facilitate its direct detection in the culture medium (Kotowska et al. 2014). The use of plate cultures allowed us to overcome the notorious variability of antibiotic production rates in liquid shake cultures, which is a problem well known to investigators of Streptomyces physiology (Siebenberg et al. 2010). Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 592 KB) Supplementary file1 (PDF 592 KB)