Streptomyces belongs to the soil-dwelling actinomycetes and produces the most plentiful bioactive natural products (Quinn et al., 2020). It accounts for about two-thirds of the clinically used antibiotics and also is the main producer of many pharmaceutical agents, including immuno-suppressors, anti-cancer drugs, and enzyme inhibitors, etc. Moreover, the genome sequencing project reveals that Streptomyces usually contain 30–50 biosynthetic gene clusters (BGCs) to produce natural products (Nicault et al., 2020). However, only about 5%–10% of BGCs have been scientifically studied, and most of them are transcriptionally silenced in laboratory, which has prompted scientists to re-investigate Streptomyces for this rich repertoire of natural products by awakening these cryptic BGCs (Liu et al., 2021). There has always been an urgent demand to develop effective and efficient strategies to improve productivity, which is particularly important for the full understanding of biosynthetic mechanisms and the pharmaceutical development of these bioactive small molecules (Onaka, 2017).

The biosynthesis of these complex natural products are sequentially catalyzed by a cascade of enzymes, which are encoded by the genes within the corresponding BGCs. Thus, the efficient production of natural products in Streptomyces is extensively dependent on the catalysis of biosynthetic enzymes, expression of gene clusters, supply of building blocks, coenzymes, and energy (such as ATP, NAD(P)⁺/NAD(P)H) (Hertweck, 2009; Walsh, 2016), etc. A full understanding and transcriptional engineering of the regulatory networks are promising ways of improving the expression of biosynthetic enzymes, potentially enhancing the production of natural products in Streptomyces (Rigali et al., 2018). Nevertheless, the total regulatory networks might be more complex than we expect. For example, it has been found that cell morphology significantly affects the production of natural products (Khushboo et al., 2023). Moreover, target deciphering of the regulatory mechanisms for a certain BGC is highly technology-dependent and time-consuming. Heterologous expression of the whole BGCs in a universal cell chassis, along with promoter reconfiguration is emerging as an alternative promising way to visualize the cryptic natural products from Streptomyces. However, the efficient cloning and expression of giant BGCs from Streptomyces are challenges currently faced by the scientific community (Liu et al., 2018; Nepal and Wang, 2019; Xu and Wright, 2019). Alternatively, a cell-free system of Streptomyces is emerging as a new opportunity for natural product discovery and derivation, while some bottlenecks still need to be optimized (Moore et al., 2023). Moreover, the productivity of natural products from Streptomyces is not always satisfactory for downstream research and development (Kang and Kim, 2021).

Due to the regulatory complexity and ingenious balance of the whole cell metabolism, a small disturbance of the network will not always lead to ideal productivity. In contrast to the rational but partial reconfiguration of these regulatory networks, the random mutagenesis occurs through more overall alterations in the genome and is more straightforward as it is only based on the ultimate phenotypic outputs, for example, the productivity of natural products. It can be carried out with biological methods, such as insertion of mobile genetic elements transposons and random evolution by high resistance to antibiotics (Xu et al., 2017; Luo et al., 2018; Yu et al., 2019a), chemical agents, such as N-methyl-N′-nitro-N-nitrosoguanidine (MNNG) (Mo et al., 2013; Rubio et al., 2021) and more convenient physical measures mainly based on high-energy radiation from ultraviolet (UV) and the recently emerging atmospheric and room temperature plasma (ARTP) (Mo et al., 2013; Yu et al., 2019b; Ye et al., 2021), as well as the combinatory applications of above strategies (Mo et al., 2013; Yan et al., 2021). However, the biggest challenge of the mutagenesis strategy is the high throughput but efficient screening of strains for hyper-production of the desired natural products. To overcome this obstacle, many reporter systems have been developed in Streptomyces, such as chromogenic strategies with enzymes IdgS and XylE to produce the blue and yellow pigments (Li et al., 2015; Li et al., 2018), respectively, and antibiotic resistance to kanamycin (Li et al., 2018; Gongerowska-Jac et al., 2021). However, false positives frequently arise in single-reporter systems, because different disturbances might result in the same reporter phenotypes. For example, ribosome mutations might also lead to high resistance to kanamycin (Luo et al., 2018). Thus, dual reporter systems have been preferentially used to more efficiently obtain the desired mutants (Xiang et al., 2009). The reporter systems are usually designed linked to the target promoters for screening of improved transcriptional strength, but mostly on the replicative plasmids or integrated in the genome distant from the native promoters of the biosynthetic gene cluster (Gongerowska-Jac et al., 2021). Probably due to the influence of genomic locations or the incompleteness of transcriptional elements on the cloned promoter (Bilyk et al., 2017), the strength of the reporter system does not always correlate well with the productivity of natural products.

Daptomycin is a lipopeptide antibiotic produced by Streptomyces roseosporus. It is biosynthesized through the activation of decanoic acid and attachment to a cyclic peptide, which is catalyzed by DptEF and the non-ribosomal peptidyl synthetases (NRPSs) DptABCD. Genes dptEFABCD comprise a large operon to produce a giant polycistronic transcript for daptomycin biosynthesis (Miao et al., 2005; Baltz, 2009). Based on the genome mining of transcription factors and deciphering of regulatory mechanisms, the removal of the negative regulators (Mao et al., 2015; Mao et al., 2017; Luo et al., 2018) and stabilization of the positive pleiotropic regulator AtrA (Xu et al., 2023) have significantly increased daptomycin production. Meanwhile, targeting multi-level rational metabolic engineering by precursor supply, regulatory pathway reconstruction, byproduct removal, and gene cluster multiplicity have improved the yield by about 6.6 folds (Lyu et al., 2022). To more comprehensively disturb the regulatory networks for the efficient increase of daptomycin production, the present study developed a dual-reporter system at the native daptomycin gene cluster and efficiently obtained a strain with improved productivity of 6.4 folds with simple UV-induced mutagenesis. This strategy is highly effective for efficient screening and would be widely applicable in other Streptomyces species for the hyper-production of natural products.

Bacterial strains and growth conditions. The bacterial strains used in this study are listed in Table 1. Streptomyces roseosporus L30 (wild type) (Sun et al., 2020) was used as the parental strain in this study. Escherichia coli DH5α was the host for plasmid construction, and ET12567/pUZ8002 was used for the conjugation of plasmids from E. coli into S. roseosporus. The wild type L30 and its derivative strains were cultured at 30°C with shaking at 250 rpm in 2% tryptic soy broth (TSB) liquid medium plus 5% PEG6000. Streptomyces strains were cultured at 30°C on the R5 solid medium for sporulation and the MS medium (2% mannitol, 2% soybean flour, 2% agar) for conjugation. Daptomycin production was carried out in the YEME medium containing 4% glucose, 0.5% peptone, 0.3% yeast extract and 0.3% malt extract for fermentation, and 0.1% decanoic acid (V: V = 1: 1 methyl oleate) was added every 12 h from 72 h to promote the biosynthesis of daptomycin (Sun et al., 2020). Kanamycin (Km) resistance screening of the dual reporter strains was performed on the yeast malt glucose (YMG) medium with 0.4% yeast extract, 1% malt extract, 0.4% glucose, 0.2% CaCO_3, and 2% agar. E. coli DH5α and E. coli ET12567/pUZ8002 were cultured in Luria-Bertani (LB) medium (1% tryptone, 0.5% yeast extract, 0.5% NaCl) at 37 °C.

Plasmid construction. All plasmids and primers were provided in Table 2 and Supplementary Table S1, respectively. Plasmid pKC1139-trAI Cas9 is a fine-tuned CRISPR-Cas9 genome editing tool. It contains a TRAI-Cas9 cassette to precisely and efficiently remove genomic fragments and is suitable for genome editing of industrial Streptomyces. cas9 expression is repressed under the non-induction condition and its highly expressed cognate inhibitor AcrIIA4. The high expression of Cas9 can be induced with thiostrepton and theophylline for genome editing (Jiang et al., 2021). To construct the plasmid pKC1139-trAI Cas9-PKSII-melC for deletion of PKSII (orf3388-orf3390, totally 2.5 kb) and melC1/C2 (orf0284-orf0285, totally 1.4 kb), the fragments containing PKSII and melC spacers were inserted into the BamHI/XbaI site of pKC1139-trAI cas9 plasmid. The left and right homologous fragments of PKSII were cloned by primers 1 and 2, primers 3 and 4, respectively. And the left and right homologous fragments of melC were amplified from L30 genomic DNA with primers 5 and 6, and primers 7 and 8, respectively. Then we inserted DNA fragments as above into the EcoRI site of pKC1139-trAI cas9-spacer plasmid with the ClonExpress MultiS One step Cloning Kit (Vazyme, China). Plasmid pSUC01 is a suicide plasmid containing the indigoidine synthetase gene from S. albidoflavus J1074 for efficient gene deletion. When the indigoidine synthetase gene exists after a single-crossover, the mutants of conjugants are blue with apramycin resistance. After multiple replicates with non-selection, the double-crossover mutants can be readily obtained with the loss of blue pigments and apramycin resistance (Dong et al., 2021). The plasmid pSUC01-rppA was used for the deletion of rppA (orf0223, 1.0 kb in total). The left and right homologous fragments of rppA were amplified from the L30 genomic DNA with primers 9 and 10, primers 11 and 12, respectively, and cloned into the EcoRI/KpnI site of pSUC01. pKC1139 is a temperature-sensitive plasmid for gene deletion (Bierman et al., 1992). For construction of the pKC1139-idgS-sfp plasmid to insert idgS-sfp into the dpt gene cluster after dptF, the left and right homologous fragments of dptF were amplified with primers 13 and 14, primers 15 and 16, respectively, from the L30 genomic DNA, while the fragment of idgS-sfp was amplified with primers 17 and 18 from the plasmid pCIMt002 (Li et al., 2015), and cloned into the EcoRV/HindIII site of pKC1139. The fragment containing neo fragment was amplified from the plasmid pET28a with primers 19 and 20. The left and right homologous fragments of dptD were amplified from the L30 genomic DNA with primers 21 and 22, primers 23 and 24, respectively, and cloned into the EcoRI/XbaI site of pSUC01 for pSUC01-neo by the ClonExpress MultiS One step Cloning Kit (Vazyme, China). The PCR amplification conditions involved pre-denaturation at 94°C for 4 min, denaturation at 98°C for 10 min, annealing at 65°C for 15 s, and extension at 68°C (time was calculated according to 1 kb/min) with 30 cycles.

Strain construction for the dual reporter. To avoid chromogenic interference, genes PKSII, melC1/C2 (melC), and rppA were knocked out by in-frame deletion. The recombinant plasmid pKC1139-trAI cas9-PKSII-melC was transformed into ET12567/pUZ8002 and conjugated with S. roseosporus L30 wild-type (WT) strain. The recombinants were selected on the R5 agar medium containing 50 μg/mL apramycin. After culturing at 30°C for 5 days, the mycelia were transferred to the TSB liquid medium containing 15 μg/mL thiostrepton and 10 mM theophylline to induce the simultaneous deletion of PKSII and melC. The bacterial suspension was diluted 10² and plated on the R5 agar plate. PCR was then performed to identify PKSII-melC double knock-out strains (Jiang et al., 2021). The plasmid pSUC01-rppA was introduced into the ΔPKSIIΔmelC dual knockout strain by intergeneric conjugation. The recombinants were selected on the R5 agar medium containing 50 μg/mL apramycin and the blue single-crossover strains were transferred to a new R5 agar medium. After growth at 30°C for 3 days, the mycelia of blue strains were inoculated into the TSB liquid medium without apramycin, and the bacterial suspension was diluted 10³ and cultured on the R5 agar plate. The white double-crossover strains were picked and PCR was performed to identify the ΔPKSIIΔrppAΔmelC triple knock-out strains (Dong et al., 2021), which was designated L31. To insert promoter-less neo gene into the dpt gene cluster after dptD in the ΔPKSIIΔrppAΔmelC strain, the recombinant plasmid pSUC01-neo was conjugated into the triple knock-out strain through ET12567/pUZ8002. And the neo insertion mutant was developed as the same procedure to delete rppA described above. The white double-crossover strains were picked and PCR was performed to identify the neo-ΔPKSIIΔrppAΔmelC strain. Finally, the recombinant plasmid pKC1139-idgS-sfp was introduced into the strain from the previous step by ET12567/pUZ8002. The recombinants were selected on the R5 agar medium containing apramycin. After growth at 37°C for 5 days with 50 μg/mL apramycin, the mycelia from different single colonies were collected and PCR was performed to select single-crossover mutants, which were then inoculated on a new R5 agar medium without apramycin at 37°C for 3 days for double cross-over and plasmid loss. The spores were collected and spread to single colonies. The mycelia from different single colonies were then collected and PCR was performed to select double-crossover mutants with insertion of idgS-sfp fragment in the neo-ΔPKSIIΔrppAΔmelC strain into the dpt gene cluster after dptF (Sun et al., 2020). The dual-reporter system strain was designated L32.

Optimization of UV-based random mutagenesis. To test whether the medium was suitable for the dual-reporter strain (L32) and produce the blue pigment, mycelia of L32 were fermented in the TSB liquid medium for 2 days to reach the logarithmic growth, and an appropriate amount of bacterial solution was transferred on ISP4, YMG, MS, and R5 plates, and incubated at 30°C for 2 weeks. To test the Km resistance of the reporter strain before UV mutagenesis, mycelia of L32 were fermented in a shake flask with the liquid YEME medium. After the bacteria grew to the logarithmic growth phase, 20 μL of mycelia were spread on the YMG plates with Km gradient concentration from 0 to 50 μg/mL, and cultured at 30°C for 5 days to determine the minimum tolerance concentration of kanamycin after UV mutagenesis. Then UV radiation time by conducting a lethality curve and dilution of UV treated spores were optimized. The spores of the dual-reporter strain, which was incubated in TSB liquid medium for 2 days, were diluted with water to OD₆₀₀ = 0.4–0.6, and exposed to ultraviolet radiations with the wavelength of 254 nm at a distance of 37 cm (8 W UV-Lamp) for 0, 1, 2, 3, 4, 5, 6, 7, and 8 min, respectively. Three parallel groups for each mutagenesis were set up. The UV treated spores were diluted with water to 100 and 1000 folds (1× UV treated spores as a control) and transferred to the R5 agar plate containing 50 μg/mL kanamycin and incubated in the dark, respectively. After 7 days, the number of colonies were counted to draw the lethality curve. Lethal rate (%) = (number of plate colonies before mutagenesis - number of plate colonies after mutagenesis)/number of plate colonies before mutagenesis × 100%. The dilution concentration with uniform distribution of single colonies and the appropriate number was used as the concentration standard for the next dilution. The time of UV radiation with a lethal rate of more than 90% was selected as the time standard for the next mutagenesis.

Screening of random mutagenesis. The spores of the reporter strain were exposed to ultraviolet radiations, and grown on R5 plates with 50 μg/mL Km, concentration for about 5 days in the dark. The single colonies with the dark blue in each plate were picked and inoculated in a YEME liquid medium for fermentation. 20 μL of the bacterial suspension was patched on the Km resistance gradient plates from 50 to 100 μg/mL and from 100 to 200 μg/mL km. At the same time, the starting reporter strain L32 without UV treatment was used as a control. Next, mutants with higher Km resistance than L32 obtained from the above mutagenesis were picked and fermented in a shake flask in the YEME medium. 1 mL of fermentation broth in 36, 48, 60, 72, 84, and 96 h, respectively, was collected to quantitatively evaluate the indigoidine production on high performance liquid chromatography (HPLC).

Production and analysis of indigoidine and daptomycin. In total, 1 mL of the fermentation broth was collected and terminated by adding an equal volume of methanol. The supernatant was harvested by centrifugation and analyzed on HPLC using a Shimadzu LC-20AP with an Agilent ZORBAX SB-C18 column (4.6 × 150 mm, 5 µm) (Agilent Technologies). The HPLC condition of daptomycin began with 10% acetonitrile for 5 min, followed by a linear gradient from 15% to 55% acetonitrile over 20 min, and subsequently, the condition was switched to 95% acetonitrile for 5 min and then restored to the initial composition of 10% acetonitrile for another 5 min. A constant flow rate of 0.5 mL/min was maintained. Detection of the analyte was performed at 215 nm with pure daptomycin as a standard. Detection of the analyte was performed at 215 nm with pure daptomycin as a standard. The HPLC condition of indigoidine began with 10% acetonitrile for 5 min, followed by a linear gradient from 10% to 50% acetonitrile over 30 min, the condition was then switched to 95% acetonitrile for 5 min, and then restored to the initial composition of 10% acetonitrile for another 5 min. A constant flow rate of 1 mL/min was maintained. Detection of the indigoidine was performed at 560 nm.

HR-LC-MS analysis of indigoidine. We collected 1 mL of the fermentation broth and terminated by adding an equal volume of methanol. The fermentation broth was centrifugated at 13,000 rpm at 4°C for 10 min. The supernatant was transferred to a new centrifuge tube and subjected to LC-MS on an Agilent Eclipse TC-C18 column (5 μm, 4.6 × 250 mm, Agilent Technologies), at a flow rate of 0.3 mL/min. Solution A (pure water) and solution B (pure acetonitrile) were employed to isolate the components with ultraviolet detection set at 560 nm. Elution was initiated with 10% acetonitrile for 5 min, then followed by a constant gradient from 10% to 55% acetonitrile over 10 min, and subsequently, the condition was switched to 95% acetonitrile for 5 min and then restored to the initial composition of 10% acetonitrile for another 5 min. The MS system was operated with an electrospray ionization source in positive ion mode.

Preparation of Streptomyces RNA and real-time quantitative PCR (qRT-PCR). RNA was prepared from mycelia using an RN43-EASY spin Plus kit (China Aidlab Biotechnologies Co., Ltd.) to ensure the ratio of A₂₆₀/A₂₈₀ at 1.9–2.0. Genomic DNA was removed with RNase-free DNaseI (TaKaRa) and cDNA was prepared with HiScriptII Reverse Transcriptase (Vazyme). The reverse transcriptional reaction involves 50°C for 15 min and 85°C for 5 s. Then qRT-PCR was performed with R222-HiScriptII QRT SuperMix for qPCR (Vazyme) of genes neo (primers 27 and 28), idgS-sfp (primers 29 and 30), dptA (primers 31 and 32), dptBC (primers 33 and 34), dptD (primers 35 and 36) and dptE (primers 37 and 38), with a house-keeping gene orf0041 (primers 25 and 26) as an internal control. The qRT-PCR reaction procedures included 95°C for 30 s, 95°C for 5 s, 56°C for 5 s, and 72°C for 20 s. A total of 40 cycle numbers were taken from the second to the fourth step. The relative fold changes in gene expression were calculated based on the following formula 2^−ΔΔCT.

For the efficient screening of the desired mutants after random mutagenesis for the hyper-production of natural products in Streptomyces, the present study developed a native-locus dual reporter system by integrating both the antibiotic-resistant gene neo and the endogenous chromogenic gene idgS at the native gene cluster for the biosynthesis of natural products (eg. daptomycin), ideally fused with backbone genes as a whole operon. Compared to the traditional single reporter system, the desired mutants could be readily obtained based on our dual reporter. The false positives could be significantly reduced, and the productivity was highly correlated with the effectiveness of both reporters. Moreover, the productivity of the target natural product can also be remarkably improved. The increased folds from our one-dose mutagenesis are comparable to the time-consuming strategy based on mechanism deciphering and network refactoring. This strategy could be widely and readily applied in Streptomyces to efficiently ramp up the productivity of drugs and activate the cryptic gene clusters, to speed up the pharmaceutical development in Streptomyces.