A Regulator Based “Semi-Targeted” Approach to Activate Silent Biosynthetic Gene Clusters

By culturing microorganisms under standard laboratory conditions, most biosynthetic gene clusters (BGCs) are not expressed, and thus, the products are not produced. To explore this biosynthetic potential, we developed a novel “semi-targeted” approach focusing on activating “silent” BGCs by concurrently introducing a group of regulator genes into streptomycetes of the Tübingen strain collection. We constructed integrative plasmids containing two classes of regulatory genes under the control of the constitutive promoter ermE*p (cluster situated regulators (CSR) and Streptomyces antibiotic regulatory proteins (SARPs)). These plasmids were introduced into Streptomyces sp. TÜ17, Streptomyces sp. TÜ10 and Streptomyces sp. TÜ102. Introduction of the CSRs-plasmid into strain S. sp. TÜ17 activated the production of mayamycin A. By using the individual regulator genes, we proved that Aur1P, was responsible for the activation. In strain S. sp. TÜ102, the introduction of the SARP-plasmid triggered the production of a chartreusin-like compound. Insertion of the CSRs-plasmid into strain S. sp. TÜ10 resulted in activating the warkmycin-BGC. In both recombinants, activation of the BGCs was only possible through the simultaneous expression of aur1PR3 and griR in S. sp. TÜ102 and aur1P and pntR in of S. sp. TÜ10.


Introduction
Actinomycetes produce more than two-thirds of the antibiotics used in medicine and agriculture, as well as numerous antitumour agents, and several eukaryotic cell differentiation effectors, such as apoptosis inducers and inhibitors [1]. To date (10 June 2021), more than two thousand genomes of filamentous actinomycetes have been sequenced and annotated (https://jgi.doe.gov/) [2]. Their analysis using genome mining tools like anti-SMASH [3][4][5][6][7][8], PRISM [9] or NaPDoS [10] revealed that each producer has a much higher potential to synthesise natural products than was assumed for many years. This assumption was because strains cultivated under standard laboratory conditions produced only a limited number of compounds.
Thus, to activate the so-called "silent" biosynthetic gene clusters (BGCs) and to improve their biosynthetic potential various strategies have been applied, including culture A frequently observed regulatory protein family in Streptomyces are the SARPs [39]. SARPs are cluster-situated activators mostly found within BGCs encoding the biosynthesis of antibiotics. The first studied and characterised SARP regulator was ActII-ORF4 controlling the actinorhodin (Act) biosynthesis in S. coelicolor [40]. SARPs include a N-terminal winged HTH motif, which is crucial for binding to a conserved DNA sequence. The binding motif of SARPs consists of heptameric repeats with 4 bp spacers located stringently 8 bp upstream of -10 promoter region [41].
Additional regulators commonly present in Streptomyces that influence antibiotic production are two-component systems (TCS) consisting of a sensor histidine kinase and a response regulator. TCSs are also essential e.g., in communication or in adaptation to the environment. The sensor histidine kinase senses a specific signal from the environment (ligand). Upon ligand binding, it gets autophosphorylated and subsequently transfers the phosphoryl group to the cognate response regulator, which then activates or represses the transcription of target genes [28]. AbsA1/AbsA2 is an extensively studied TCS of S. coelicolor, which is involved in activating calcium-dependent antibiotic (CDA) transcription [42].
On the other hand, atypical response regulators also exist (ARR). They display homologies to the response regulators of TCS, but lack the important residues for phosphorylation, which are essential for signal transmission in TCS. Accordingly, ARRs use a different, not fully understood signal response mechanism [43]. Aur1P, the activator of the angucyclinelike antibiotic auricin, is a member of ARRs family regulators. The direct binding of Aur1P to the auricin promoter region regulates the transcription of auricin biosynthesis. DNase-I footprint experiments revealed that the Aur1P binding sequence is a tandem repeat sequence TCCCTTG separated by a 24 bp spacing region accompanied by sequence regions CCTTG and CCT [44,45].
Although the regulators can be classified according to their sequences, in most cases, it is not possible to deduce from this neither their function as activator or repressor, nor the type of product of the respective BGC they are regulating [46].
In our "semi-targeted" approach we used the activating nature of regulators to initiate the transcription of silent clusters. For this purpose, representative CSR genes of the protein families LuxR, StrR, ARR, MarR, SARPs were introduced into different strains of the Tübingen strain collection. We demonstrated the development of a very promising method that circumvents the limitations of classical regulatory approaches to discover new natural compounds.

Selection and Cloning of Regulatory Genes for Activation of Silent Clusters
To exploit the full biosynthetic potential of actinomycetes, a variety of methods to active silent clusters has been developed in many scientific groups worldwide. These include, on the one hand, methods for untargeted biosynthetic gene cluster (BGC) activation, such as the introduction of pleiotropic activators [47], the use of chemical elicitors [48] or transcription factor decoys [49]. Thus, activating silent clusters is more of a random event. On the other side, targeted approaches need specialised constructs for each BGC of interest, i.e., methods, such as promoter refactoring or manipulation of pathway-specific transcriptional regulators [50,51]. This procedure requires specific knowledge of the BGC and is time-consuming.
We chose a "semi-targeted" approach that takes an intermediate position between the untargeted and targeted one (Figure 1). Our work is focusing on activating silent clusters by introducing heterologous regulatory genes in strains of the Tübingen strain collection, which have previously been evaluated for their biosynthetic potential. We reasoned that the production of heterologous cluster situated regulators (CSRs) from various protein families would enable the detection of metabolites that would otherwise not be produced. It is feasible that the binding of these regulators to the promoter region of genes encoding the biosynthesis of bioactive compounds could activate their transcription without prior knowledge of the specific BGC sequences and by overcoming the necessity of an environmental signal.
the untargeted and targeted one (Figure 1). Our work is focusing on activating silent clusters by introducing heterologous regulatory genes in strains of the Tübingen strain collection, which have previously been evaluated for their biosynthetic potential. We reasoned that the production of heterologous cluster situated regulators (CSRs) from various protein families would enable the detection of metabolites that would otherwise not be produced. It is feasible that the binding of these regulators to the promoter region of genes encoding the biosynthesis of bioactive compounds could activate their transcription without prior knowledge of the specific BGC sequences and by overcoming the necessity of an environmental signal. Figure 1. Scheme of the "semi-targeted" approach to activate silent BGCs in Actinomyces. pEM1, plasmid carrying different cluster situated regulator genes (CSRs); pEM2, plasmid carrying genes encoding different Streptomyces antibiotic regulatory proteins (SARPs).
First, we selected CSRs regulator genes from various BGCs, which were experimentally characterised and validated. The genes were cloned as a single operon under the control of the strong constitutive promoter ermEp* into the integrative vector pRM4 [52]. Two plasmids (pEM1 and pEM2) were constructed by a classical cloning approach. These plasmids are carrying different sets of regulatory genes amplified of the genes from the respective Streptomyces strains (Table 1). Plasmid pEM1 contained four different genes aur1P-pntR-strR-fkbN. Particular focus was taken on including one representative from each type of CSR (Table 1).
Plasmid pEM2 carries five different Streptomyces antibiotic regulatory protein (SARP) genes, actIIORF4-red-griR-aur1PR3-papR2 (Table 1). Our aim was to apply SARPs, which are highly distinctive, and which regulate the biosynthesis of structurally diverse compounds. Based on all small SARPs available in the NCBI database, five SARPs were selected, which regulate the synthesis of structurally different products. Using phylogenetic analyses, we confirmed that the regulators do not display any homologies to each other (SI. Figure S1A, SI. Figure S1B). First, we selected CSRs regulator genes from various BGCs, which were experimentally characterised and validated. The genes were cloned as a single operon under the control of the strong constitutive promoter ermEp* into the integrative vector pRM4 [52]. Two plasmids (pEM1 and pEM2) were constructed by a classical cloning approach. These plasmids are carrying different sets of regulatory genes amplified of the genes from the respective Streptomyces strains (Table 1).  (Novakova et al., 2005) aur1P and aur1PR3 Plasmid pEM1 contained four different genes aur1P-pntR-strR-fkbN. Particular focus was taken on including one representative from each type of CSR (Table 1).
Plasmid pEM2 carries five different Streptomyces antibiotic regulatory protein (SARP) genes, actIIORF4-red-griR-aur1PR3-papR2 (Table 1). Our aim was to apply SARPs, which are highly distinctive, and which regulate the biosynthesis of structurally diverse compounds. Based on all small SARPs available in the NCBI database, five SARPs were selected, which regulate the synthesis of structurally different products. Using phylogenetic analyses, we confirmed that the regulators do not display any homologies to each other (SI. Figure S1A, SI. Figure S1B).

Activation of Silent BGCs in Strains of the Tübingen Strain Collection
Three strains isolated from soil in Switzerland, Streptomyces sp. TÜ17, Streptomyces sp. TÜ10 and Streptomyces sp. TÜ102, were selected to activate silent clusters by applying the "semi-targeted" approach. According to internal data of the Tübingen natural products database, S. sp. TÜ17 is producing grisein, a sideromycin antibiotic active against Gramnegative and Gram-positive bacteria, already discovered in 1947 [53]. S. sp. TÜ10 is known as nonactin producer [54], whereas no biologically active metabolite has been isolated from S. sp. TÜ102 so far.
Both plasmids pEM1 and pEM2 were introduced into the three strains, and the integration into their genomes was confirmed by PCR.

Metabolites Produced by Recombinant Streptomyces sp. TÜ17 Strains
The integration of pEM1 in S. sp. TÜ17 resulted in activating at least of one BGC. Ethylacetate extracts of S. sp. TÜ17::pEM1 showed antibiotic activity against B. subtilis and E. coli in contrast to the recombinant strain S. sp. TÜ17::pEM2 and the wild type S. sp. TÜ17 ( Figure 2A). The subsequent HPLC/MS analyses revealed a peak with a retention time (Rt) of 9.7 min and a molecular mass of m/z 464.3 [M+H] + . This peak was absent in both, S. sp. TÜ17 and S. sp. TÜ17::pEM2 extracts ( Figure 2B). The molecular mass and the Rt of this peak did not match with data from the internal natural product database. To further characterise the compound, a high-resolution LC-MS analysis of the S. sp. TÜ17::pEM1 extract was performed. The identified mass m/z 464.1703 [M+H] + was compared with the data of the molecules from the Dictionary of Natural Products (CRC Press, DNP 28.2). The masses were in accordance with the masses of mayamycin, an angucycline antibiotic also produced by a Streptomyces strain isolated from the marine sponge Halichondria panicea [55]. In addition, the MS/MS fragmentation pattern of the substance produced by S. sp. TÜ17::pEM1 also coincided with the pattern of the maymycin standard (SI. Figure S2) (kindly provided by J. Wiese, Helmholtz-Zentrum für Ozeanforschung Kiel, Germany). Mayamycin, a compound of the angucycline type, shows activity against several antibiotic-resistant strains, such as Klebsiella pneumoniae, Pseudomonas aeruginosa, Staphylococcus aureus and others, as well as against cancer cell lines, such as GXF251L (gastric cancer) LXF529L (non-small-cell lung cancer) MAXF401NL (mammary cancer) and many others [55].
Marine actinomycetes are still considered to be promising natural producers of new antibiotics with clinical relevance. They usually live associated with other organisms, such as sponges, but are difficult to isolate and cultivate. To activate the production of mayamycin, subinhibitory concentrations of the antibiotics bacitracin and tetracycline were added to the culture [55]. Applying our "semi-targeting" approach, we demonstrated that it is not necessary to use exotic sources to isolate interesting metabolites, but rather to exploit the potential of already isolated actinomycetes.
To confirm the specific activation of the cluster by the introduced regulators, we analysed the published mayamycin BGC, and the regulator gene may18 contained therein [56]. Multiple sequence alignment of May18 and all CSRs encoded on plasmid pEM1 revealed the highest homology of May18 to Aur1P (SI. Figure S3). Moreover, BlastP analysis of Aur1P showed 70% identity and 81% similarity to May18. To verify the hypothesis that Aur1P indeed initiates activating the maymycin BGC, the individual CSR regulator genes were introduced into S. sp. TÜ17. The silent maymycin BGC could only be activated in the recombinant strain Streptomyces sp. TÜ17::pEM-aur1P, carrying the may18 homologous gene ( Figure 2C). From these results, it can be concluded that for activating BGC, a plasmid is needed that covers a broad spectrum of potential activator genes as comprehensively as possible. In our case, this was achieved by cloning CSR genes with high diversity. After showing that the method generally works, we aimed to investigate the molecular mechanism of silent cluster activation by the introduced regulatory genes in more detail. Therefore, we selected two sequenced strains from the Tübingen strain collection. This has the advantage that in the case of successful activation of a silent BGC, the corresponding cluster could be identified more easily. We focused on a strain from which no compound has been isolated so far (S. sp. TÜ102, Accession Nr.: JAHDTK000000000) and on one from which several metabolites have already been characterised (S. sp. TÜ10, Accession Nr.: JAHDTJ000000000).
Ethylacetate extracts of the recombinant strain S. sp. TÜ102::pEM2 showed antibacterial activity against B. subtilis, whereas no antibiotic activity was observed in extracts of the recombinant strain S. sp. TÜ102::pEM1 and the wild type S. sp. TÜ102 ( Figure 3A).
Thus, the introduction of the plasmid pEM2, carrying various SARP regulator genes, led to activating at least one silent cluster encoding the biosynthesis of a biologically active substance. HPLC-MS analyses of the extracts revealed a peak with a retention time (Rt) of 9.4 min and a molecular mass m/z 1302. 4 [M+H] + , which could not be detected in the S. TÜ102::pEM1 and S. sp. TÜ102 extracts ( Figure 3B). To evaluate which substance was produced, we compared the Rt and molecular mass of this peak with the data in our internal database. This matched with the data of chartreusin, a type II polyketide antibiotic (PK) with a mixed PK-carbohydrate structure. In addition to the antibiotic activities, chartreusin displays substantial chemotherapeutic activities [57]. Its mode of action (MoA) is based on the inhibition of topoisomerase II during DNA synthesis [58].
This has the advantage that in the case of successful activation of a silent BGC, the cor sponding cluster could be identified more easily. We focused on a strain from which compound has been isolated so far (S. sp. TÜ102, Accession Nr.: JAHDTK000000000) a on one from which several metabolites have already been characterised (S. sp. TÜ10, A cession Nr.: JAHDTJ000000000).
Ethylacetate extracts of the recombinant strain S. sp. TÜ102::pEM2 showed antib terial activity against B. subtilis, whereas no antibiotic activity was observed in extracts the recombinant strain S. sp. TÜ102::pEM1 and the wild type S. sp. TÜ102 (Figure 3 Thus, the introduction of the plasmid pEM2, carrying various SARP regulator genes, to activating at least one silent cluster encoding the biosynthesis of a biologically act substance. HPLC-MS analyses of the extracts revealed a peak with a retention time (Rt 9.4 min and a molecular mass m/z 1302. 4 [M+H] + , which could not be detected in the TÜ102::pEM1 and S. sp. TÜ102 extracts (Figure 3 B). To evaluate which substance w produced, we compared the Rt and molecular mass of this peak with the data in our ternal database. This matched with the data of chartreusin, a type II polyketide antibio (PK) with a mixed PK-carbohydrate structure. In addition to the antibiotic activities, ch treusin displays substantial chemotherapeutic activities [57]. Its mode of action (MoA based on the inhibition of topoisomerase II during DNA synthesis [58].  In the subsequent step, in silico analyses were performed to determine whether a potential chartreusin BGC is present in the genome of S. sp. TÜ102. For this reason, the genome sequence of S. sp. TÜ102 was evaluated by using the genome mining tool antiSMASH [7]. Thereby 34 potential BGC were predicted. One of them showed high similarities to the type II PKS BGC that codes for the synthesis of chartreusin. The chartreusin BGC has been identified and characterised in Streptomyces chartreusis HKI-249 [57]. The nucleotide sequence of the published chartreusin BGC (GenBank: MH540322.1) displays 99% nucleotide identity to the BGC in strain S. sp. TÜ102 (~35 kb of both BGCs were aligned). Both BGCs contain the same number of genes in an identical orientation. In addition to the structural genes responsible for the stepwise biosynthesis of the compound, both BGCs are coding for three regulatory genes: chaR1, encoding a putative transcriptional regulator of the SARP family, chaR2, encoding a putative sensor histidine kinase, and chaR3, encoding a putative transcriptional regulator of the LacI family [57]. Although both BGCs are nearly identical, the BGC in S. sp. TÜ102 remained silent under the cultivation conditions used; only the introduction of heterologous SARP regulatory genes in S. sp. TÜ102::pEM2 resulted in activating the BGC.
For a detailed understanding of these findings, phylogenetic analyses were performed using the Phylogeny.fr tool. This involved comparison of the protein sequences of six SARPs encoded in the Streptomyces sp. TÜ102 genome and the protein sequences of SARPs introduced heterologously via plasmid pEM2. These analyses revealed the clustering of ActIIORF4 and Aur1PR3 with the chartreuse ChaR1-like SARP regulator (SI. Figure S4). The additional blastP analyses confirmed 51% identity of Aur1PR3 and 42% identity of ActIIORF4 with the ChaR1 SARP regulator.
Screening the promoter regions of the BGC genes for SARP binding motifs revealed two putative SARP binding motifs in chartreusin BGCs of both producer strains: The first (CTGCTCGAAATGTGCTCGAAGAGG) upstream of the chaA-R genes and the second (CGACTCGAGAAATCCTCGACCCGC) upstream of the chaJ-Q genes (Figure 4). Furthermore, the SARP binding site upstream of the chaJ gene was identical to the binding motif of the ActIIORF4 regulator within the Act BGC [40].
dition to the structural genes responsible for the stepwise biosynthesis of the compo both BGCs are coding for three regulatory genes: chaR1, encoding a putative trans tional regulator of the SARP family, chaR2, encoding a putative sensor histidine ki and chaR3, encoding a putative transcriptional regulator of the LacI family [57]. Altho both BGCs are nearly identical, the BGC in S. sp. TÜ102 remained silent under the c vation conditions used; only the introduction of heterologous SARP regulatory gen S. sp. TÜ102::pEM2 resulted in activating the BGC.
For a detailed understanding of these findings, phylogenetic analyses were formed using the Phylogeny.fr tool. This involved comparison of the protein sequenc six SARPs encoded in the Streptomyces sp. TÜ102 genome and the protein sequenc SARPs introduced heterologously via plasmid pEM2. These analyses revealed the clu ing of ActIIORF4 and Aur1PR3 with the chartreuse ChaR1-like SARP regulator (Sl. Fi 4). The additional blastP analyses confirmed 51% identity of Aur1PR3 and 42% identi ActIIORF4 with the ChaR1 SARP regulator.
Screening the promoter regions of the BGC genes for SARP binding motifs reve two putative SARP binding motifs in chartreusin BGCs of both producer strains: The (CTGCTCGAAATGTGCTCGAAGAGG) upstream of the chaA-R genes and the se (CGACTCGAGAAATCCTCGACCCGC) upstream of the chaJ-Q genes ( Figure 4). thermore, the SARP binding site upstream of the chaJ gene was identical to the bin motif of the ActIIORF4 regulator within the Act BGC [40]. To subsequently specify which of the five introduced heterologous SARP regula genes was responsible for the activation, the regulatory genes were introduced indiv ally into S. sp. TÜ102. Surprisingly, none of the individual regulatory genes was ab activate the putative chartreusin BGC. Consequently, we constructed the pla pRM4_actIIORF4-aur1PR3, bearing the genes actIIORF4 and aur1PR3. Remarkably presence of these two SARPs, ActIIORF4 and Aur1PR3 in S. sp. TÜ102 resulted in the duction of chartreusin, demonstrating that both regulators are required for activatin silent chartreusin BGC (Figure 3 B).
In contrast, the second investigated strain S. sp. TÜ10 exhibited a different behav Here, the ethylacetate extract of the wild-type strain S. sp. TÜ10 already revealed ant terial activity against B. subtilis. In the HPLC/MS chromatograms of the extracts o recombinant strain S. sp. TÜ10::pEM1, two additional peaks with an Rt of 9.2 min a molecular mass m/z of 1006.4 [M+H] − and Rt 9.8 min and a molecular mass m/z 10 To subsequently specify which of the five introduced heterologous SARP regulatory genes was responsible for the activation, the regulatory genes were introduced individually into S. sp. TÜ102. Surprisingly, none of the individual regulatory genes was able to activate the putative chartreusin BGC. Consequently, we constructed the plasmid pRM4_actIIORF4-aur1PR3, bearing the genes actIIORF4 and aur1PR3. Remarkably, the presence of these two SARPs, ActIIORF4 and Aur1PR3 in S. sp. TÜ102 resulted in the production of chartreusin, demonstrating that both regulators are required for activating the silent chartreusin BGC ( Figure 3B).
In contrast, the second investigated strain S. sp. TÜ10 exhibited a different behaviour. Here, the ethylacetate extract of the wild-type strain S. sp. TÜ10 already revealed antibacterial activity against B. subtilis. In the HPLC/MS chromatograms of the extracts of the recombinant strain S. sp. TÜ10::pEM1, two additional peaks with an Rt of 9. [M+H] − , respectively, were present. Both peaks were missing in the extracts of the WT the recombinant strain S. sp. TÜ10::pEM2 ( Figure 5). To determine the structural properties of the compounds associated with these pe the data were compared with the data of the internal natural product database. Both, Rt and the masses corresponded to the already known natural products deacetyl-w mycin (Rt 9.2 min, m/z 1006. 4 [MH] − ) and warkmycin (Rt 9.8 min, m/z 1048. 4 [MH] − ). W mycin is a glycosylated angucycline antibiotic with activity against gram-positive bact and tumour cell lines [59]. Warkmycin and deacetyl-warkmycin were extracted fr To determine the structural properties of the compounds associated with these peaks, the data were compared with the data of the internal natural product database. Both, the Rt and the masses corresponded to the already known natural products deacetyl-warkmycin (Rt 9.2 min, m/z 1006. 4 [MH] − ) and warkmycin (Rt 9.8 min, m/z 1048. 4 [MH] − ). Warkmycin is a glycosylated angucycline antibiotic with activity against gram-positive bacteria and tumour cell lines [59]. Warkmycin and deacetyl-warkmycin were extracted from Streptomyces strain Acta 2930 isolated from an embryonic sand dune in Warkworth Northumberland, UK [59]. Recently, two new members of the warkmycin family, warkmycin CS1 and CS2, were identified in a Streptomyces strain isolated from the integument of ants of the Tribe Attini [60]. To support the hypothesis that S. sp. TÜ10::pEM2 produced a warkmycin-like compound, we performed genome mining analyses to screen for a corresponding BGC. Using antiSMASH, 40 putative BGCs were identified. One of the predicted BGC showed a high similarity to the warkmycin cluster. It contains all structural genes described to encode the synthesis of warkmycin, as well as three genes coding for regulators (warR1, warR2 and warR3). Blast analyses and multiple sequence alignment performed by Clustal Omega (SI. Figure S5) revealed similarities (56% identity, 73% positivity) of WarR1 to the Aur1P regulator, which acts as an activator of angucycline type auricin biosynthesis [61]. WarR2 belongs to the TetR family of regulators, and, like most regulators, in this family, could act as a repressor [60]. WarR3 is a putative regulator of the ArsR family and shows similarity (35% identity, 61% positivity) to PntR, a pentalenolactone BGC regulator [35]. Known representatives of this family are transcriptional repressors with an HTH domain that dissociates from DNA in the presence of metal ions (e.g., Zn 2+ , As 2+ , As 3+ , Sb 3+ , Cd 2+ ) [62].
Again, we were interested in determining the regulator that is responsible for activating warkmycin. Therefore, the regulatory genes of the plasmid pEM1 were separately introduced into the strain S. sp. TÜ10. Based on the similarities of the two regulators encoded in the warkmycin cluster to the regulator genes aur1P and pntR that we introduced into S. sp. TÜ10, we hypothesised that both genes are necessary to activate the warkmycin cluster. For this reason, the plasmid pEMaur1P_pntR was constructed, including the corresponding regulator genes. In the recombinant strain S. sp. TÜ10::pEMaur1P_pntR the production of the warkmycin-like substance was again restored ( Figure 5).
To further verify this result, we performed RT-PCR analyses. For this, we isolated RNA from the S. sp TÜ10 WT and the recombinant strains S. sp TÜ10::pEM1, S. sp TÜ10::pEMaur1P, S. sp TÜ10::pEMpntR and S. sp TÜ10::pEMaur1P_pntR after 72 h cultivation under production conditions. We demonstrated that the warkmycin genes b9w61_07295 (annotated as beta-ACP synthase), b9W61_07275 (annotated as cyclase) and b9W61_07205 (annotated as glycosyl transferase) were only transcribed when both regulator genes aur1P and pntR were also transcribed. The presence of only one of the two regulatory genes, on the other hand, did not lead to the transcription of the biosynthetic genes. The regulator genes warR1, warR2 and warR3 were not transcribed, at all (SI. Figure  S6). Based on these results, we concluded that only the presence of both regulatory genes aur1P and pntR under the control of the constitutive promoter ermEp* enabled activating the warkmycin cluster.

Conclusions
We have developed a powerful tool to activate silent clusters in Streptomyces strains. By using a "semi-targeting" approach, metabolites are produced under standard cultivation conditions that are otherwise only synthesised by using specific techniques. These results indicate that new compounds can be identified when this approach is applied in an automated high-throughput procedure with many different strains.
Our results illustrate the reasons why the induction of gene expression often fails: In many cases, multiple activators are needed. By introducing plasmids with a broad spectrum of regulator genes into the strains, we circumvent the problem and activate silent clusters that otherwise can only be activated using adapted methods.

Bacterial Strains and Cultivation Conditions
Different Streptomyces strains were used for the amplification of the regulator genes (Table 1).

DNA Manipulations, Plasmids
All DNA manipulations in E. coli strains were done according to [63]. Streptomyces manipulations were done according to [64]. Restriction enzymes were used according to the recommendations of suppliers (Thermo Fisher Scientific, Waltham, MA, USA) The integrative pRM4 vector [49] was used for cloning of regulatory genes under the control of the strong constitutive ermEp* promoter. The recombinant plasmids were integrated via the ΦC31 att site into the chromosome of the Streptomyces strains.

Constructions of Plasmid pEM1 and pEM2
Standard cloning techniques were used to construct plasmids pEM1 and pEM2. Primers for amplification are listed in Table S1. All cloned genes are under the control of the strong ermEp* promoter, and each gene carries RBS for its translation. All plasmids were isolated by QIAGEN Plasmid Plus Midi Kit (Qiagen, Hilden, Germany) and sequenced.
Genomic DNA (gDNA): Genomic DNA from Streptomyces strains was isolated using the VWR peqGOLD Bacterial DNA kit (VWR international GmbH, Pennsylvania, USA). PCR with gDNA as a template was used to confirm the integration of plasmids.

Multiple Sequence Alignment
An omega cluster tool was used to perform multiple sequence alignments with standard settings using HMM profile-profile techniques [66].
For conjugation, spores of Streptomyces were harvested from the SFM or HA medium, and SFM medium was used to plate mating.

Bioassay
Extracts of the recombinant strains and the corresponding WT strains were spotted on antibiotic discs and tested against B. subtilis ATCC 6633 and E. coli K12. Reference strains (B. subtilis or E. coli) were grown in liquid LB medium to OD = 0.6 and inoculated into agar plates (OD = 0.2). Plates were stored open for drying for 20 min under the sterile bench. The spotted antibiotic discs were placed on plates. Strains were incubated overnight by 28 • C or 37 • C.