Enhancing microbial metabolic capacity through high-energy electron beam-induced intense structural variations

Abstract

Small-molecule metabolites are key pharmaceutical resources embedded in complex organismal metabolomes. Scalable microbial production depends on metabolic activation capacity, which in turn requires efficient genetic variation. Structural variants (SVs), key drivers of phenotypic diversity, are pivotal for organism evolution, yet their highly efficient induction remains challenging. While DNA double-strand breaks (DSBs) facilitate SVs formation, existing mutagenesis technologies struggle to balance high DSB efficiency with cellular preservation, particularly in microbial strain improvement for metabolite production. Conventional irradiation methods suffer from low SVs induction rates, making strain enhancement a lengthy and labor-intensive process. Here, we systematically compare six irradiation technologies in Streptomyces lividans 1326 and identify high-energy pulsed electron beams (HEPE) as an approach which effectively induces strong DSBs while preserving cellular integrity. This results in extensive SVs that reshape genome sequences and 3D chromatin structure, leading to activation of secondary metabolite production. By integrating HEPE with high-throughput metabolomics (HEPE-HiTMS), we discover two secondary metabolites with unusual C-N linkage, respectively. Applied across various microorganisms, HEPE enables record-high clavulanic acid and microcin J25 production, and markedly increases lovastatin yields. With its ability to induce SVs with minimal cytotoxicity, HEPE represents a powerful tool for cryptic metabolite discovery and industrial strain development.

The majority of microbial metabolites remain cryptic or minimally expressed under laboratory conditions. Here authors show that high-energy pulsed electron beams (HEPE) can be used to induce extensive structural variation, activating secondary metabolite biosynthesis.

Introduction

Microbial metabolites have profoundly influenced human civilization-from traditional brewing and vinegar making to the discovery of antibiotics-and remain central to industries such as pharmaceuticals, energy and agriculture¹. However, under natural conditions, the biosynthesis of metabolites is tightly regulated, with over 80% being produced at limited production levels or even undetectable². These insufficiently produced metabolites, especially the fully silent secondary metabolites (SMs), are normally difficult to access through cultivation optimization or simple regulatory engineering, and require extensive and profound genetic rewiring to activate. This challenge contributes to the slow pace of drug discovery and extended timelines in industrial strain development. At the core lies a major bottleneck: the lack of efficient, scalable tools to boost metabolic output³.

Gene mutations, particularly structural variations (SVs >50 bp)^4–6, drive microbial evolution by disproportionately altering genomic architecture, transcriptional networks, and metabolic output^7–9. While DNA double-strand breaks (DSBs) catalyze SV formation^10,11, spontaneous DSBs in microbes under natural conditions are extremely rare. This limits natural mutation rates and hampers the efficiency of strain improvement for metabolites production^12–14. Irradiation mutagenesis accelerates evolution by introducing severe DNA damage, thereby inducing intense mutation¹⁵. Traditional approaches such as UV and γ-ray irradiation, as well as newer methods like atmospheric room-temperature plasma (ARTP), have long been used in microbial mutagenesis¹⁶. However, these techniques are rarely used for activating cryptic metabolites, and they fail to achieve both robust DSB induction and high cellular integrity, which are essential for maximizing SV formation and mutant survival¹⁷.

To address this challenge, we systematically compare six different types of irradiation technologies using the model host Streptomyces lividans 1326 (S. lividans 1326): UV and microwave (MW; low-energy non-ionizing radiation)¹⁸; X-ray and γ-ray (high-energy ionizing radiation)¹⁹; ARTP²⁰ and HEPE (particle radiation)²¹. We evaluate their capacity to induce DSBs, preserve cellular integrity, and enhance the production of SMs with diverse expression levels²². Among them, HEPE, previously unreported in microbial mutagenesis, emerges as superior, generating intense SVs that remodel genome sequences and 3D chromatin architecture, significantly boosting metabolic capacity.

By coupling HEPE with high-throughput metabolomics (HEPE-HiTMS), we rapidly identify and characterize two structurally novel, previously undetectable compounds—griseobrucins and fradibactin with rare C-N linkage, respectively. Extending HEPE to diverse microorganisms (actinobacteria, bacteria, fungi), we achieve yield enhancements in the production of clavulanic acid²³ (2×; record-level), lovastatin²⁴ (7.3×), and antimicrobial peptide microcin J25²⁵ (3×; highest reported). Our findings establish HEPE as a transformative mutagenesis tool that enhances metabolic output by driving SV-based genomic remodeling with minimal cytotoxicity. This approach provides a powerful tool for the discovery of bioactive compounds and applications in industrial biotechnology.

Results

DNA damage induction and cellular integrity across six irradiation technologies

To evaluate and compare the DNA double-strand break (DSB) induction efficiency of six irradiation technologies, we performed TdT-mediated dUTP nick end labeling (TUNEL) assays and plasmid linearization analyses. Given the heterogeneity in irradiation types and energy levels among the six technologies, direct comparison of their DNA-damaging effects using identical radiation doses was precluded. We thus irradiated samples with four escalating dose levels, covering the commonly used upper and lower detection limits of irradiation instruments to enable cross-platform analysis²¹ (see Materials and Methods).

Mycelia of S. lividans 1326 were irradiated with UV, MW, X-rays, γ-rays, ARTP, and HEPE, after which the cells were harvested and subjected to TUNEL assays to detect single- and double-strand DNA breaks (SSBs and DSBs). Results from fluorescence microscopy clearly revealed no detectable signals in UV-, MW-, or ARTP-treated samples at their respective maximum irradiation durations of 600 s, 180 s and 300 s. Samples treated with X-ray and γ-ray exhibited faint signals at low doses but pronounced fluorescence at the highest doses of 100 Gy and 3 kGy respectively. Whereas HEPE generated strong signals even at minimal doses of 1 kGy, with intensity increasing in a dose-dependent manner (Fig. 1a). These findings indicate that X-ray, γ-ray, and HEPE possess robust penetration capacity to induce intracellular DNA damage, with HEPE causing the most severe DNA damage, followed by γ-ray and X-ray. In contrast, UV, MW, and ARTP showed limited penetration capacity for DNA breaks.

Fig. 1. Comparison of the DNA (a) and cell structure (b) damage of S. lividans 1326 induced by six irradiation technologies.

a TUNEL-based fluorescence micrographs showing DNA strand breaks. Scale bar is 5 µm. b Scanning electron micrographs revealing cellular damage. Scale bar is 1 µm. The irradiation doses are specified in the upper left of each image. Microscopy images shown are representative of findings observed in at least three independent experiments. The wild-type (WT) strain S. lividans 1326 was subjected to six irradiation technologies: ultraviolet (UV), microwave (MW), X-ray, γ-ray, atmospheric and room-temperature plasma (ARTP) and high-energy pulsed electron beam (HEPE).

To further investigate DSB induction by each method, plasmid pIJ773 (4.3 kb) was irradiated and then analyzed by plasmid linearization assay. Supercoiled-to-linear conformational shifts, indicative of extracellular DSBs, were negligible in UV- and MW-treated samples even at their maximum doses. Conversely, HEPE, γ-ray, ARTP, and X-ray induced progressive plasmid linearization, with efficiency ranked as HEPE > γ-ray > ARTP > X-ray. Notably, HEPE almost fully linearized pIJ773 at minimal doses, indicating its most powerful DSBs induction effect (Supplementary Fig. 1). These results demonstrate that DSB induction correlates with radiation energy levels²⁶, and HEPE, the energy of which can easily reach industrial safety limits (10 MeV)-exhibits unparalleled efficiency in DSB generation.

Next, scanning electron microscopy (SEM) of irradiated S. lividans 1326 cells revealed severe lysis and structural damage in UV-, MW-, X-ray-, and γ-ray treated samples at subminimal doses (90 s, 60 s, 50 Gy, 0.5 kGy), respectively. In contrast, ARTP (90 s) and HEPE (6 kGy) caused observable damage at submaximal doses, with negligible effects observed at subminimal doses (Fig. 1b). This highlights the superiority of particle radiation (HEPE, ARTP) in preserving cellular integrity over ionizing and non-ionizing radiation.

Based on these results, HEPE shows the strongest DSB induction for generating SVs while causing minimal cell damage, an optimal balance for enhancing microbial metabolite production. Given its limited application in mutagenesis breeding, we selected two competitive controls for further evaluation: ARTP, which also moderately induces DSBs with minimal cell damage but has limited penetration, and UV, which is the most cost-effective and accessible.

Evaluating metabolite production in S. lividans 1326 by UV, ARTP and HEPE

The general host S. lividans 1326 contains the biosynthetic gene clusters (BGCs) of pigments ACT (actinorhodin) and RED (undecylprodigiosin). These two BGCs are lowly expressed or cryptic under certain culture conditions, and exhibits blue and red colors upon their activations²². We compared mutagenic efficiencies of HEPE, UV, and ARTP by screening pigmented mutants. S. lividans was irradiated and the resulting positive colorful mutants were screened, counted, and evaluated for phenotypic stability and the production of ACT and RED (Fig. 2a). A total of 308,672 surviving colonies (with and without blue or red color) were counted, ensuring that each irradiation dose contributed over 5000 colonies for subsequent mutagenic efficiency analysis (Fig. 2b). The blue and red pigmented colonies, identified as positive mutants with activated ACT and RED production, were used to calculate the positive mutation frequency. At optimal HEPE dose (6 kGy), the positive mutation frequency reached 10.7%, which is 8.9 and 6.7 times higher than that of UV and ARTP, respectively, despite comparable mortality rates ( > 97%, Fig. 2c–e).

Fig. 2. Comparison of mutagenic efficiency among UV, ARTP and HEPE in S. lividans 1326.

a Workflow of radiation mutagenesis with UV, ARTP and HEPE. b Number of clones obtained across different irradiation doses of UV, ARTP and HEPE. Positive mutation frequency (bar chart, left y-axis) and lethality rate (line graph, right y-axis) for UV (c), ARTP (d) and HEPE (e) under different doses. Data are presented as mean ± SEM from six independent experiments, each performed with three replicates (n = 6). f Production stability of ACT (actinorhodin) and RED (undecylprodigiosin) in mutants induced by UV, ARTP and HEPE over ten rounds of subculturing. g Comparison of average ACT and RED yields produced by the mutants of UV, ARTP and HEPE irradiation. Data are presented as mean ± SEM from three independent samples, each performed with three replicates (n = 3). Significance was assessed by one-way ANOVA. *P < 0.05; **P < 0.01; ***P < 0.001. Source data are provided as a Source Data file.

To further assess the production stability of the mutations induced by the three technologies, totally 96 dual-colored mutants, 32 each from UV, ARTP and HEPE were randomly selected and subcultured for ten generations. The proportion of mutants that retained pigmentation in each generation was calculated (Supplementary Fig. 2). After ten rounds of subculturing, 60.9% of HEPE-induced mutants stably produced ACT/RED, exceeding the rates for UV (29.7%) and ARTP (42.2%), which confirms the superior genetic stability of the HEPE-derived mutants (Fig. 2f). Meanwhile, HEPE mutants exhibited higher average ACT/RED yields. Analysis of ACT/RED production at generations 3, 5, 7 and 10 in three stably pigments-producing mutants from each method showed that HEPE not only outperforms UV and ARTP in mutation efficiency but also in production stability and yield enhancement (Fig. 2g). This suggests that mutations induced by HEPE are less readily repaired than those of UV and ARTP.

Investigation of genomic and 3D chromatin alterations in S. lividans 1326 induced by HEPE

To investigate the mechanism underlying the superior mutagenic efficiency of HEPE irradiation, we sequenced all the genomes of 96 pigmented mutant strains previously obtained through UV, ARTP, and HEPE irradiation. Compared to the original S. lividans strain, the total number of genetic variants in the UV, ARTP, and HEPE groups was 2817, 2523, and 2545, respectively, with SNVs being the predominant variant type (Fig. 3a, Supplementary Table 1). Although HEPE irradiated mutants did not show increase in SNVs and indels compared to UV and ARTP irradiated mutants (Supplementary Table 1-2), they exhibited a marked increase in the SVs (151), surpassing those induced by UV (70) and ARTP (111). This confirmed our expectation that HEPE induces more DSBs, leading to a higher number of SVs^10,27,28 (Fig. 3a). Notably, among all SV types, 56 translocations (TRAs) in HEPE mutants exceeded UV/ARTP by nearly 6-fold, indicating frequent occurrence of non-homologous end joining (NHEJ) in the genomes of HEPE mutants^28–30 (Fig. 3a).

Fig. 3. The genomic alterations induced by HEPE irradiation mutagenesis.

a The variant type and number of mutants irradiated by UV, ARTP or HEPE. Variant types include single nucleotide variation (SNV), small insertion and deletion (indel), and large-scale structural variation (SV) such as inversions (INV), insertions (INS), deletions (DEL), duplications (DUP), and translocations (TRA). b Genomic distribution of variants sites in mutants irradiated by UV, ARTP or HEPE. Scatterer dots represent SV type. The colored links represent TRA type. Circles 1, 2 and 3 (from outer to inner), genomic distribution of mutation sites in UV, ARTP and HEPE group, respectively. Circle 4, BGCs of S. lividans 1326 (BGC of ACT and RED labeled as act and red). Circle 5, representation of the S. lividans 1326 genome, scaled in megabases. Its colors correspond to the genetic partition into the left arm (bright blue), core region (gray) and right arm (bright green), as demarcated by the first and last rRNA coding gene²⁸. c Comparison of the significant interactions between WT S. lividans 1326 and its mutant (H_D1) derived from HEPE mutagenesis. From outer to inner in the Circle plot: Circle 1, identical to Circle 5 in panel (b). Circle 2, gene intensity. The significant interactions are shown as blue line links, with the color gradient (dark to light) indicating increasing P-values. d Comparison of the 3D chromosome conformation between WT S. lividans 1326 and its mutant (H_D1) derived from HEPE mutagenesis. Its colors correspond to the genetic partition in panel (b). Source data are provided as a Source Data file.

A comparison of the genomic distributions of SVs induced by UV, ARTP, and HEPE irradiation in S. lividans was performed using a dispersion analysis with genome binning (see Methods). The result revealed that HEPE group had a more uniform genomic distribution of SVs than the UV and ARTP groups, as evidenced by its lower CV value, although all groups significantly deviated from a uniform distribution (Supplementary Table 3). Circos plot visualization further illustrated the SVs distribution patterns (Fig. 3b, Supplementary Table 4). In the UV and ARTP groups, SVs were predominantly located at the arm ends, with TRA occurring between spatially proximate left and right arms. In contrast, the HEPE group exhibited SVs also abundantly in the core regions. Accordingly, TRA events in the HEPE group were observed between arms as well as between arm ends and the core region.

SVs can exert a greater impact on phenotype changes. This impact is not solely attributed to larger alterations in DNA sequence, dramatic changes in the three-dimensional (3D) chromosome conformation could also play a key role compared to SNVs and indels³¹. To investigate potential alterations in 3D conformation in HEPE mutant, we compared the chromosome architecture of the wild type (WT) and the mutant strain (H_D1) using high-throughput chromosome conformation capture (Hi-C). First, circus plot (Fig. 3c) of the significant interactions (see Methods) revealed a substantial increase in interaction number in mutant (8830) compared to WT (458). Furthermore, the 3D architecture (Fig. 3d) showed that WT maintain a compact chromosome organization, whereas the mutant exhibited a more relaxed conformation. This difference can be elucidated by the differential interaction and compact analysis (see Methods). As shown in the differential interaction heatmap and the compact matrix (Supplementary Fig. 3), WT is enriched in long-range interactions, maintaining its tighter spatial connectivity. In contrast, mutant showed enhanced short-range interactions, corresponding to a looser spatial organization. This result indicated that mutant’s reinforced local interactions disrupt the native compact 3D genome architecture. This structural relaxation likely facilitates the binding of RNA polymerase and other transcriptional machinery to DNA by increasing accessibility to regulatory regions and gene promoters, ultimately modulating the expression of secondary metabolite BGCs³².

Transcriptome sequencing was subsequently performed on the S. lividans 1326 WT strain and HEPE mutants exhibiting activation of the act or red gene cluster. Differential expression analysis confirmed substantial transcriptional alterations, including the significant upregulation of multiple genes within the red or act gene cluster (Supplementary Fig. 4). HEPE treatment also induced widespread transcriptional perturbations across numerous other BGCs (Supplementary Fig. 5, Supplementary Table 5). Additionally, random SVs can have possible unintended effects, such as growth defects. To assess off-target consequences of HEPE mutagenesis, three mutants were selected for phenotype and growth curve analysis. Compared to WT strain, mutants exhibited reduced sporulation and delayed growth (Supplementary Fig. 6). Transcriptome data linked this phenotype to a marked downregulation of the sporulation-associated gene sapA and the regulatory factors whiB and whiD, which control cell division and spore septum formation^33,34 (Supplementary Table 6).

Collectively, this dual disruption of DNA sequence and 3D chromatin architecture likely drives enhanced metabolite production^35,36. These results support our hypothesis that HEPE generates more SVs, which induce more intense dual alterations of DNA sequence and 3D chromosome conformation, thereby leading to the higher production of ACT and RED, as previously reported in S. coelicolor ³¹.

Application of HEPE for cryptic metabolites discovery

Given the demonstrated efficiency of HEPE in inducing extensive genomic alterations, we developed the HEPE-HiTMS workflow, a strategy which couples HEPE irradiation with high-throughput metabolomics to evaluate its potential in activating fully cryptic SMs, potentially uncovering novel structural metabolites. HEPE-HiTMS integrates BGC activation with metabolomic profiling as readout³⁷ (Fig. 4a). Resulting MS data were computationally amalgamated into a 3D plot, correlating metabolite intensity and m/z of the compounds detected across mutants to enable algorithm-driven metabolomes analysis³⁸.

Fig. 4. Application of HEPE for SM discovery.

a Workflow of HEPE-HiTMS approach for cryptic metabolite activation. b The 3D plot of secondary metabolome of S. griseobrunneus after HEPE irradiation, showing MS profiles in the m/z range of 300–500. c HPLC–MS analysis confirming the production of griseobrucin A and B in flask cultures. d Structures of griseobrucins elucidated by X-ray crystal and NMR spectroscopy. e The 3D plot of secondary metabolome of S. fradiae after HEPE treatment, showing MS profiles in the m/z range of 800–1000. f HPLC-MS analysis verifying the production of fradibactin A and B in flask cultures. g BGC and structure for fradibactins. Source data are provided as a Source Data file.

We randomly selected two streptomycete species, Streptomyces griseobrunneus CGMCC 29023 and Streptomyces fradiae CGMCC 29024. S. griseobrunneus was firstly irradiated with HEPE; approximately 120 mutagenic clones were cultured for MS acquisition using a high-throughput approach. The metabolomes were visualized in a 3D plot, revealing three significantly induced metabolites (m/z 386.1, 404.1, and 418.2) in the 300–500 m/z range, which were absent in the parental strain (Fig. 4b, c, Supplementary Fig. 7). The activated metabolome was further annotated using FUNEL, which confirmed that compound m/z 404.1 appears to be novel, with no matches in NPAtlas or COCONUT database (Supplementary Data 1)³⁹. Scaling fermentation of high-yield mutants enabled purification and structural elucidation of these compounds, designated griseobrucins A-C (1-3).

Analysis of 1D/2D NMR revealed griseobrucin A as a methylated diketopiperazine (DKP) conjugated to histidine via an unusual C-N bond, rather than the canonical amide linkage⁴⁰, while griseobrucins B and C are hydrolytic derivatives (Fig. 4d, Supplementary Fig. 8–10, Supplementary Table 7–9). X-ray diffraction resolved griseobrucin C’s absolute configuration (CCDC number 2432012), which is retained in variants A and B (Fig. 4d, Supplementary Fig. 11, Supplementary Table 10).

We next applied HEPE-HiTMS pipeline to S. fradiae. About 120 obtained clones were subjected to generate a 3D plot metabolomic profiling, revealing two strongly activated metabolites with identical molecular weight (m/z 890.5), undetectable in the parental strain within the 800–1000 m/z range (Fig. 4e–f, Supplementary Fig. 7). FUNEL analysis confirmed their novelty, with no match in public databases (Supplementary Data 1). These two compounds, named fradibactin A (4) & B, were purified from scaled fermentation for structural characterization.

1D/2D NMR analyses identified fradibactins as two enantiomers of hybrid siderophores merging hydroxamate and carboxylate groups, containing a unique C-N double bond with Z configuration, a linkage unreported in known hydroxamate-type siderophores (Fig. 4g, Supplementary Fig. 12, Supplementary Table 11). This unique structural element prompted us to investigate its BGC. AntiSMASH mining uncovered two hydroxamate-associated BGCs⁴¹ (Supplementary Fig. 13). Targeted gene inactivation localized fradibactins biosynthesis to the fdn cluster, which harbors an atypical aminobutyrate-2-oxoglutarate aminotransferase (fdnA) adjacent to conserved hydroxamate formation genes (fdnB-F), implicating fdnA in C-N double bond formation (Supplementary Fig. 14, Supplementary Table 12, 13). The successful discovery of the cryptic griseobrucins and fradibactins demonstrates HEPE’s exceptional capacity to activate metabolites, particularly those fully undetectable metabolites through targeted structural variations, positioning HEPE-HiTMS as a robust platform for discovering cryptic microbial metabolites in the future.

HEPE-induced yield increasing of clavulanic acid in industrial Streptomycete

Having demonstrated HEPE’s effectiveness in activating cryptic metabolites, we next applied it to boost the production of targeted metabolites in industrial strains. Unlike wild-type strains, these industrial strains had already undergone conventional mutagenesis and extensive screening, resulting in optimized metabolite production. Our objective was to leverage HEPE mutagenesis to achieve further yield improvement that are unattainable through conventional methods. We firstly selected Streptomyces clavuligerus 70116-an industrial clavulanic acid overproducer derived from WT S. clavuligerus 3585 through 6 years of conventional mutagenesis (non-HEPE) and strain selection (Fig. 5a). Approximately 10⁴-10⁵ spores of S. clavuligerus were subjected to HEPE irradiation on ISP2 agar plates following S. lividans mutagenesis protocol. A total of 110 surviving colonies were selected and fermented in an optimized medium. Comparison of clavulanic acid yields revealed that 27 out of 110 mutant clones exhibited at least a 10% increase in titer (Fig. 5b). The high mutation rate indicates that HEPE mutagenesis is not compromised by prior conventional mutations.

Fig. 5. Application of HEPE for increasing clavulanic acid yield in industrial Streptomycete.

a Workflow for screening higher-yield clones in industrial strains through HEPE mutagenesis. b Mutagenic efficacy of HEPE for clavulanic acid. FC (fold change), the ratio of product yield from the mutant strain compared to the starting strain. c (Left) The titer of clavulanic acid produced by WT, industrial progenitor 70116 and HEPE-derived mutant H_M32 in 5 L benchtop bioreactor. (Right) Genome variants analysis of each strain. Data are presented as mean ± SEM (n = 3). Significance was assessed by one-way ANOVA. *P < 0.05; ***P < 0.001. Source data are provided as a Source Data file.

The highest-yielding mutant H_M32 reached a clavulanic acid titer of 5.5 g/L in flasks after 6 d fermentation, representing a 60% increase over the original strain (3.5 g/L). After confirming its productions stability over five rounds of subculturing (Supplementary Fig. 15a), H_M32 was advanced to large-scale cultures. It reached a yield of 6.3 g/L in a 5 L benchtop bioreactor, which signifies an additional 15% increase over the starting strain and is comparable to the record-high of 6.11 g/L²³ (Fig. 5c).

To investigate the mechanism underlying the progressive increase in clavulanic acid production among these producers, we performed whole-genome sequencing on the WT, the starting strain 70116, and the highest-yielding mutant H_M32. Analysis of genome variations revealed that strain 70116, which underwent 5 rounds of mixed mutagenesis from WT (Fig. 5c), accumulated a large number of variations compared to the wild type, including 265 SNVs, 29 indels, and 7 SVs, collectively resulting in the 2.3 g/L increase in clavulanic acid production compared to WT strain. In contrast, the HEPE-irradiated mutant H_M32 achieves 2.8 g/L over 70116 with fewer total mutations (36 SNVs, 16 indels, 5 SVs) but a significantly higher proportion of SVs (Fig. 5c). This result underscores that SV frequency, rather than the total mutation count is the key driver of phenotypic enhancement, aligning with SVs’ powerful impact on transcriptional regulation and 3D genome architecture.

Application of HEPE for enhancing metabolites production in diverse industrial species

With the success in streptomyces, we extended the application of HEPE to diverse microbial taxa to evaluate its broad mutagenic potential across other bacteria and fungi. For comparative purpose, UV and ARTP were included alongside HEPE. Moreover, to apply HEPE to plasmid-dependent heterologous expression, which requires unstable multicopy plasmids and can compromise long-term strain performance, we selected an engineered Escherichia coli CGMCC 16349 strain designed for heterologous production of the antimicrobial peptide microcin J25 (mccJ25)⁴². Following UV, ARTP, and HEPE irradiation, 16.3% of HEPE-treated mutants exhibited >2-fold increase in mccJ25 yield, outperforming UV and ARTP (Fig. 6a–d). Notably, the top 3 mutants resulted in >3-fold yield enhancements, and remained stable after 10 rounds of subculturing in 100 mL flasks (Supplementary Fig. 15b). In scale-up 40 L bioreactors, the best mutant produced mccJ25 at a titer of 7.5 g/L, surpassing the previous literature yield of 4 g/L²⁵ (Fig. 6e, f).

Fig. 6. Application of HEPE for increasing mccJ25 yield in industrial E.coli and lovastatin yield in Monascus purpureus.

Yield changes of mccJ25 within mutants irradiated by UV (a), ARTP (b) and HEPE (c). Fold change (FC) indicates the yield ratio of the mutant relative to the starting strain. Each dot represents an individual mutant sample. The lines and error bars depict the mean ± SEM (n = 20). d Mutagenic efficacy of HEPE for mccJ25. e The yield of mccJ25 produced by top three HEPE-derived mutants in flask cultures. Data are presented as mean ± SEM (n = 3). f The titer of mccJ25 produced by the best-performing mutant (H_9B) in 40 L bioreactor. Yield changes of lovastatin within mutants irradiated by UV (g), ARTP (h) and HEPE (i). Fold change (FC) indicates the yield ratio of the mutant relative to the starting strain. Each triangle represents an individual sample. The lines and error bars depict the mean ± SEM (n = 15). j Mutagenic efficacy of HEPE for lovastatin. k The yield of lovastatin in 4 L solid-phase fermentation container. Data are presented as mean ± SEM (n = 3). Source data are provided as a Source Data file.

Next, we evaluated HEPE, UV, and ARTP irradiation in filamentous fungi, which are prolific producers of pharmaceutically relevant metabolites⁴³. We selected the lovastatin-producing strain Monascus purpureus MY2, which was generated by 3 rounds of irradiation on a WT strain Y2, for this study²⁴. Each irradiation treatment gave 60 surviving colonies for solid-phase fermentation and lovastatin analysis. HEPE irradiation (2 kGy) generated 13.3% mutants with over 2-fold yield increases, far exceeding UV and ARTP (Fig. 6g–j). Only 48.3% of HEPE mutants exhibited reduced production, compared to 68.3% (UV) and 75.0% (ARTP). The top-performing HEPE mutant was subcultured for five generations in 1 L solid-phase fermentation system to assess lovastatin production stability. The results indicated that lovastatin yield remained generally consistent across generations (Supplementary Fig. 15c). Furthermore, this mutant exhibited a more than 6-fold yield enhancement in 4 L solid-phase fermentation container, reaching a titer of 29 mg/g - nearing industrial benchmarks (Fig. 6k).

Collectively, the marked yield improvements of clavulanic acid, mccJ25, and lovastatin across actinobacteria, proteobacteria, and fungi underscore the remarkable superiority of HEPE irradiation over UV and ARTP, positioning it as a broadly applicable tool for industrial metabolite enhancement.

Discussion

Wild-type microbes are typically inefficient producers, with tightly repressed bioactive metabolite synthesis. Enhancing their metabolic output has long relied on DNA mutagenesis. However, conventional irradiation technologies-whether ionizing (e.g., X/γ-ray), non-ionizing (e.g., UV, MW) or particle radiation (e.g., ARTP)-struggle to balance high mutation frequency with cell viability. Here, we demonstrate that HEPE overcome this trade-off, inducing intense DSBs while preserving cellular integrity. This approach maximizes the formation of heritable SVs and significantly activates the expression of metabolome profile. Crucially, HEPE-induced DSBs activate non-homologous end joining (NHEJ), a repair pathway prone to translocations^28,44. Our Hi-C data suggests that HEPE-induced SVs lead to alteration in 3D genome architecture. Such dual changes in both the linear genome sequence and spatial genome organization synergistically perturb transcriptional regulation³¹. Transcriptomic data further confirm widespread changes in the expression levels of BGCs in HEPE-treated strains. These observations explain the higher metabolic outputs and stability of HEPE compared to UV/ARTP.

HEPE’s superiority stems from its unique physical properties. As a particle radiation source easily operating at 10 MeV (within industrial safety limits) or higher, HEPE enables deep tissue penetration and clustered DSBs without thermal denaturation^45,46. This contrasts with UV/MW (surface-level damage), γ-ray/ X-ray (diffuse ionization) or ARTP (poor penetration capacity). HEPE irradiation can induce extensive SVs in the Streptomyces genome while preserving cellular viability. This resilience can be attributed to its inherent genomic instability and stress response strategies. The Streptomyces genome appears architecturally organized to tolerate restructuring. Essential genes for primary metabolism and growth are clustered in the core region of the chromosome, while secondary metabolite BGCs and other adaptive features are often located on more plastic chromosomal arms. This compartmentalization allows the arms to undergo HEPE-induced SVs while buffering the essential core from fatal disruption. Furthermore, Streptomyces possesses linear chromosomes with terminal inverted repeats (TIRs), creating “open ends” that facilitate end-joining recombination and chromosomal rearrangements⁴⁷. Such genomic instability drives Streptomyces’ rapid evolution, aiding adaptation to competitive soil microenvironments. Under stresses like HEPE irradiation, genomic instability increases, triggering an SOS-like response. Streptomyces also harbors diverse two-component systems and mobile genetic elements that mediate DNA recombination and repair upon sensing stress signals^47,48. Robust DNA repair systems allow it to withstand radiation-induced DNA damage and even convert such damage into genetic diversity. This genomic instability coupled with robust stress resistance enables Streptomyces not only to survive tenaciously but also to evolve into a vast natural repository of medicinal compounds.

To cope with irradiation-induced DNA DSBs, cells kickstart the error-prone NHEJ pathway for rapid repair. This survival mechanism directly contributes to the generation of SVs. The mutational outcome depends not only on the number of DSBs but also on the nature of the DNA damage and the subsequent cellular repair response across different irradiation sources¹⁴. For instance HEPE, produce complex and clustered DNA lesions almost instantaneously that are challenging to repair and might be preferentially handled by rapid but error-prone mechanisms like NHEJ¹⁵. In contrast, UV radiation mainly generates cyclobutane–pyrimidine dimers, which are commonly repaired by photolyase-mediated photoreactivation⁴⁹. Plasma irradiation, however, induces oxidative DNA damage or strand breaks through reactive species intracellularly⁵⁰. Such oxidative DNA lesions are mainly repaired by two mechanisms: base excision repair (BER) and nucleotide excision repair (NER)⁵¹. SSBs are typically repaired by BER, whereas DSBs are generally repaired by homologous recombination (HR) or NHEJ mechanisms. Differences in the nature of DNA damage and repair mechanisms explain why HEPE generates more SVs. A more thorough understanding of the mechanisms underlying SVs formation will be advanced through continued characterization of SVs breakpoints and the development in vitro and in vivo model systems that can generate chromosomal SVs.

The application potential of HEPE is underscored by its success across bacteria (genetic engineered), actinobacteria, and fungi. Industrial strains—often pre-optimized through decades of mutagenesis—achieved record yields of clavulanic acid and mccJ25, and significant yield improvement of lovastatin after a single HEPE treatment. This scalability, combined with the HEPE-HiTMS pipeline for discovering fully cryptic, structurally novel metabolites griseobrucins and fradibactins, positions HEPE as a versatile tool for drug development. Regrettably, neither griseobrucins nor fradibactins demonstrated no discernible antimicrobial or cytotoxic activity against the tested human cancer cell lines. Further systematic biological activity assays are warranted to fully explore their medicinal potential^52–55.

Despite a high load of SVs constantly reshaping the genomic landscape of BGCs, the observed chemical diversity may represent only a fraction of their genomic potential. This discrepancy arises from multi-layered biological constraints. Many secondary metabolite BGCs are governed by dedicated pathway-specific regulators, yet such regulatory networks frequently exhibit functional redundancy or hierarchical control⁵⁶. Consequently, even if genomic rearrangements activate silent BGCs, their expression may still be repressed by upstream global regulators or remain inactive due to the absence or dysfunction of required pathway-specific activators. Furthermore, secondary metabolites biosynthesis depends critically on precursor molecules supplied by primary metabolism. Newly activated BGCs often impose a metabolic burden, leading to insufficient precursor flux and ultimately aborted product synthesis. Under evolutionary pressure, activated gene clusters may undergo secondary mutations that rapidly silence these resource-consuming BGCs to restore growth fitness. This silencing-driven selection further explains why compounds frequently escape detection at the final stage of analyses.

While HEPE requires specialized accelerators, its mutagenic efficiency significantly reduces screening time, workload and costs, offsetting initial investments. Compared to synthetic biology approaches such as CRISPR/Cas9 or BGC refactoring, HEPE offers distinct advantages as well as certain limitations. CRISPR/Cas9 enable targeted genomic edits but heavily relies on genetic engineering manipulation. In contrast, HEPE operates without the need for genetic manipulation or the introduction of exogenous systems, making it particularly suitable for applications subject to strict regulatory constraints and for use in wild-type industrial strains with limited genetic tools. In terms of precision, HEPE introduces random mutations across the genome, differing fundamentally from the single-nucleotide accuracy of CRISPR/Cas9. Although this limits its utility for specific edits, it allows for broad mutagenesis and discovery of unexpected functional mutations-thereby serving as a valuable complement to targeted approaches. Regarding scalability, HEPE is highly scalable for generating diverse mutant libraries in a single step, bypassing the need for complex DNA construct design and transformation procedures required for CRISPR-based methods or BGC refactoring. This makes it particularly efficient for high-throughput mutagenesis in non-model or genetically recalcitrant strains. For strain traceability, CRISPR/Cas9 leaves precise and trackable edits, whereas HEPE-induced mutations are stochastic and require whole-genome sequencing to establish mutational profiles and track strain.

Beyond laboratory efficacy, the industrial-scale application of HEPE-mutagenized strains warrants consideration of regulatory and biosafety. Unlike genome editing, which often falls under genetically modified organism regulations due to the use of foreign DNA or targeted genetic modifications, HEPE is a non-transgenic method that avoids exogenous DNA integration. However, the extensive and random mutations induced by HEPE demand rigorous stability and safety assessments to ensure consistent industrial performance. Although HEPE circumvents transgene-related concerns, the substantial genomic alterations still require thorough evaluation to guarantee product safety and environmental safety before large-scale use.

In conclusion, HEPE redefines irradiation mutagenesis by bridging the gap between genomic variation and cellular integrity. Its ability to generate diverse, stable SVs offers a path toward metabolite discovery and rapid strain development, which may have significant implications for biotechnology and medicine.

Methods

Bacterial strains

The Streptomyces lividans 1326 was kindly provided by the laboratory of Prof. Wenjun Guan at Zhejiang University. The Streptomyces griseobrunneus CGMCC 29023 and Streptomyces fradiae CGMCC 29024 (GenBank accession numbers SAMN45944959 and SAMN45945006) were kindly provided by the laboratory of Prof. Min Wu at Zhejiang University. Streptomyces clavuligerus 3585, 70116 and H_M32 was stored in the laboratory of Prof. Weishan Wang at Institute of Microbiology, Chinese Academy of Sciences. Strain 70116 was derived from 3585 with 5 rounds mutagenesis of UV-Nitrosoguanidine (NTG)-Diethyl sulfate (DES)-UV-NTG. Escherichia coli CGMCC 16349, H_11F, H_4B and H_9B was stored in the Sinagri Yingtai Bio-peptide Co.,Ltd. Monascus purpureus Y2, Monascus purpureus M_Y2 and Monascus purpureus H_MY2 was stored in the laboratory of Prof. Wei Liu at China Pharmaceutical University. Strain M_Y2 was derived from Y2 with 3 rounds mutagenesis of UV-MW-UV.

Mutagenesis for S. lividans 1326

The cryo-preserved cultures of S. lividans 1326 were plated on SFM agar (2% soybean, 2%mannitol, 2% agar, pH 7.0). From the agar plate, a single colony was picked and inoculated into YEME medium (0.3% yeast extract, 0.3% malt extract, 0.5% peptone, 1% glucose, 10.7% sucrose, pH 7.0) and incubated at 30 °C for 1-2days. The cultured cells were collected and washed twice with distilled water, after which they were resuspend in sterile water to a concentration of 50 mg/mL. Alternatively, the cultures were spread on SFM agar and incubated at 30 °C for 5–7days. After sporulation, spores were harvested and washed twice with distilled water to remove the retained aerial mycelia and cell debris. The purified spores were resuspended in sterile water, and the optical density of the spore suspension was adjusted to 0.4 at 595 nm. The resulting culture or spore suspension was subsequently used for mutagenesis experiments.

UV mutagenesis: 20 μL culture or spore suspension was spread in a petri dish, which was placed under the UV lamp (15 W, 254 nm, PHILIPS) at a distance of 30 cm. The light intensity of the ultraviolet lamp used was 48 μW/cm². The irradiation time were 30, 90, 180 and 600 s.

MW mutagenesis: 20 μL culture or spore suspension was transferred into 1.5 mL Eppendorf tube, which was placed in microwave oven. The irradiation time were 30, 60, 90 and 180 s. During the experiment, the centrifuge tube was removed from the microwave oven every 10 seconds and placed on ice to minimize thermal effects.

X-ray mutagenesis: 20 μL culture or spore suspension was put into 1.5 mL Eppendorf tube, which was placed in Biological X Ray Irradiator (RS2000pro, Rad Source Technologies). The irradiation doses were 20, 50, 80 and 100 Gy.

γ-ray mutagenesis: The γ-ray device is a wet static cobalt-60 irradiation system with a source strength of 18,600. 20 μL culture or spore suspension was put into 1.5 mL Eppendorf tube, and irradiated with ⁶⁰Co γ-ray source. The irradiation doses were 0.2, 0.5, 1 and 3 KGy.

ARTP mutagenesis: 20 μL culture or spore suspension was coated on the surface of a stainless-steel slide and placed in the chamber of ARTP mutation system (ARTP-M, Tmaxtree Biotechnology Co., Ltd., Wuxi, China). Ultra-high purity helium was used as the plasma working gas and the operating parameters were set as follows: radiofrequency power, 120 W; helium gas flow rate, 10 L/min; distance between the plasma torch nozzle exit and the sample plate, 2 mm. The irradiation time were 15, 45, 90 and 300 s.

HEPE mutagenesis: 20 μL culture or spore suspension was put into 1.5 mL Eppendorf tube and placed on the conveyor of the high-energy pulsed electron accelerator (CIAE-DZ-10/20, China Institute of Atomic Energy). The operating parameters were as follows: electron beam energy, 10 MeV; electric current, 1.5 mA; scan frequency, 335 Hz; absorbed dose rate, 4 kGy/s. The irradiation doses were 1, 3, 6 and 20 KGy.

TdT mediated dUTP nick end labeling (TUNEL) assays

To directly observe DNA damage in mutants after irradiation, fluorescein isothiocyanate (FITC)-12-dUTP was used to label damaged DNA at 3′-OH ends exposed by DNA breakage, catalyzed by Terminal Deoxynucleotidyl Transferase (TdT)⁵⁷. The culture suspension of S. lividans 1326 was irradiated with UV, MW, X-ray, γ-ray, ARTP, or HEPE, as described in the above method. Subsequently, the cells were treated with 4 mg/mL lysozyme for 1 h at 37 °C, followed by incubation with 2% Triton X-100 for 10 min at room temperature. After centrifugation, cells were resuspended in equilibration buffer prior to staining. Cells were incubated with FITC-12-dUTP for 1 h at 37 °C in the dark, then washed twice with sterile water to remove excess labeling mix.

Cell preparations were mounted on glass slides and imaged using a Nikon Eclipse Ti inverted microscope with a FITC fluorescence filter (excitation/emission 470/520 nm). Equivalent exposure times and image adjustments were used for all samples.

Mutagenesis for plasmid pIJ773

The genetic effects are presented at the molecular level by irradiating a loopful plasmid (pIJ773) with UV, MW, X-rays, γ-ray, ARTP or HEPE. The irradiation mutagenesis process and doses were consistent with the S. lividans 1326. The plasmid concentration was approximately 70 ng/µL, with a volume of 20 µL for each irradiation dose. After irradiation, the DNA was analyzed on 1.0% agarose gel electrophoresis in 1×TAE (Tris-Acetate-EDTA) buffer at 140 V for 30 min. An untreated plasmid sample served as the negative control, and the plasmid digested with HindIII-restriction enzyme was used as the positive control.

Scanning electron microscopy

Scanning electron microscopy (SEM) was employed to observe the membrane damage of mutants after irradiation treatment. Spore suspensions of S.lividans 1326 were treated with MW, UV, X-rays, ⁶⁰Co, ARTP or HEPE respectively following the dose escalation described above. After irradiation, the suspensions were centrifuged at 2375 g for 10 min to collect spore pellets. The pellets were fixed with glutaraldehyde for 2 h at room temperature and then at 4 °C overnight. Subsequently, the samples were rinsed with PBS for three times (10–15 min each time). Following this, the samples were fixed with 1% osmium solution for 1.5 h and rinsed again three times with PBS (10-15 min each time). The samples were then dehydrated through a graded series of ethanol (50, 70, 90 and 100%) and dried by Critical Point Dryer CPD 300 (Leica). The dried specimens were sputter coated with Pt-palladium (8 mA, 60 s) by high vacuum ion sputtering apparatus (Quorum Q150T ES plus). Finally, the samples were observed under a scanning electron microscope Nova Nano 450 (Thermo FEI) at 5kv.

Calculation of Positive Mutation Frequency Induced by UV, ARTP and HEPE

20 μL spore suspension of S. lividans 1326 was irradiated with UV, ARTP or HEPE at varying doses, following the irradiation mutagenesis process described previously. The UV irradiation time were 0, 30, 60, 90, 120, 150, 180 and 210 s. Using an ultraviolet lamp with a light intensity of 48 μW/cm², the resulting light energies at irradiation times of 30, 60, 90, 120, 150, 180, and 210 seconds are calculated to be 1.44, 2.88, 4.32, 5.76, 7.20, 8.64, and 10.08 mJ/cm², respectively. For ARTP, the irradiation time were 0, 15, 30, 45, 60, 75, 90, 105, and 120 s. The HEPE irradiation doses were 0, 0.5, 1, 2, 3, 4, 5, 6, 7, and 8 KGy.

After irradiation, spores were resuspended in 500 μL sterile water, serially diluted, and plated on SFM agar. The plates were incubated at 30 °C for 7 days. Colonies exhibiting red, blue, or mixed pigmentation were classified as positive mutants. Control samples were untreated spore suspensions (20 μL). The positive mutation frequency was calculated as the ratio of pigmented colonies to the total number of colonies in the control sample. The lethal rate was determined as 1 minus the survival rate. All experiments were conducted in triplicate, with sextuplicate samples for each condition.

The genetic stability of positive mutants

To assess genetic stability, the frequency of positive mutants was evaluated across 1-10 subgenerations. Cryopreserved mutants were streaked on SFM agar and incubated for 7 days (first subculture). Clones were then subcultured onto new plates for another 7-day incubation (second subculture), and this process was repeated up to the tenth subculture. Thirty-two positive mutants were randomly selected for each irradiation treatment, and their color phenotype was recorded for each subgeneration. Mutants showing mixed red and blue were considered genetically stable (scored as 1), while those showing predominantly red, blue, or a lighter mix were considered relatively stable (scored as 0.5). Mutants with wild-type coloration were deemed unstable (scored as 0). The genetic stability frequency was calculated by summing the scores and dividing by 32.

The production of ACT and RED for genetic stability assessment in 1-10 subgenerations was evaluated as follows: Three randomly selected positive mutants from each treatment were subcultured and fermented in MS medium (2% mannitol, 2% soybean) for 5 days. ACT production was measured by centrifuging 1 mL of fermentation broth (10 min, 13,500 g) and recording the absorbance of the blue supernatant at 600 nm. RED production was quantified by resuspending 10 mg of mycelial pellet in CH₃OH–HCl (pH 1.5), vortexing until biomass disruption, and centrifuging the suspension (10 min, 13,500 g). The absorbance of the supernatant was measured at 530 nm. Each fermentation was performed in triplicate for each mutant.

Whole genome re-sequencing and variation analysis

Ninety-six positive mutants, treated with UV, ARTP, or HEPE, were selected for whole genome re-sequencing. These mutants, previously used for genetic stability assessment, were compared with the wild-type S. lividans 1326 as the reference strain. All strains were cultured in TSBY medium (3% tryptic soy broth medium, 0.5% yeast extract) for 1-2 days, and genomic DNA was extracted using the CTAB method⁵⁸. Sequencing was performed on the Illumina HiSeq (NovaSeq 6000) platform at Wuhan Benagen Technology Co., Ltd. The genome was used for subsequent variation analysis.

Strains with over 60X genome coverage were used for variant detection, including single nucleotide variants (SNVs), insertion-deletions (indels), and structural variations (SVs). Paired-end raw reads were quality-controlled and filtered using fastp (v0.23.4), and the clean reads were mapped to the S. lividans 1326 reference genome with BWA (v2.2.1). Variants were called using the Genome Analysis Toolkit (GATK, v4.6.0.0). For indels detection, insertions and deletions <50 bp were considered. SV detection was carried out using smoove (v0.2.6) to identify large indels ( ≥ 50 bp), tandem repeats, and translocations.

Dispersion analysis utilizing genome binning was incorporated to characterize the distribution of SVs induced by UV, ARTP or HEPE. Based on the SVs data, a 10 kb sliding window approach was employed to assess the dispersion of SVs across the genome. Using the SVs count distribution obtained from the sliding-window analysis, we further calculated the coefficient of variation (CV) and performed a chi-square goodness-of-fit test to evaluate whether the SVs distribution in each group significantly deviated from a uniform pattern. The CV serves as a measure of dispersion: a lower value indicates a more uniform distribution, while a higher value reflects greater variability and clustering.

High-throughput chromosome conformation capture (Hi-C) of S. lividans 1326

Hi-C libraries were prepared as previously described⁵⁹. Briefly, bacterial cells were cross-linked with 3% formaldehyde for 1 h at 4 °C, followed by quenching with 0.375 M glycine for 15 min. Cells were lysed with lysozyme at 37 °C for 10 min, and endogenous nuclease was inactivated with 0.5% SDS. Chromatin was digested with 100U Sau3AI (NEB) and labeled with biotin-14-dCTP (Invitrogen), followed by ligation with 50U T4 DNA ligase (NEB). After reversing cross-links, the ligated DNA was purified using the QIAamp DNA Mini Kit (Qiagen). DNA was sheared to 300–500 bp, blunt-end repaired, A-tailed, and adapters were added. The DNA was purified through biotin-streptavidin pull-down and PCR amplification. Hi-C libraries were quantified and sequenced on the MGI-seq platform (BGI, China).

After quality filtering, the clean Hi-C data was iteratively mapped to the S. lividans 1326 genome using the ICE software. The valid pairs after filtering were binned into 5 kb nonoverlapping genomic intervals to generate a two-dimensional contact matrix. The contacts between 5 kb bins of interactions of each sample were transferred to Fit-Hi C software (v2.0.0) to calculate the corresponding cumulative probability P-value and Q-value. After calculation, the interactions in which both the P-value and Q-value were less than 0.01 and the contact count >2 were identified as significant interactions. To more intuitively compare and visualize the interaction differences between the samples, the contact matrix of each sample was first converted into a Z-score matrix. The Z-score matrices of the two samples were then subtracted to obtain a differential interaction matrix, which was subsequently displayed as a heatmap. The compact analysis reflects the local chromosome packing state by calculating, for each genomic bin, the proportion of its total interactions that occur within a defined local genomic window relative to its interactions with the entire chromosome. These values are then visualized as a matrix. Pastis (v0.1)⁶⁰ was used to construct the 3D chromosomal structure based on a 5 kb contact map, utilizing a multidimensional scaling (MDS) model. The origin and terminus regions were annotated using Centurion software⁶¹.

Transcriptome analysis

Total RNA was extracted from the tissue using Trizol reagent. RNA-seq transcriptome library was prepared following the NEBNext^® Ultra II™ Directional RNA Library Prep Kit. Libraries that pass quality control was sequenced on the Illumina high-throughput sequencing platform using paired-end 150 bp reads. Bowtie 2 was used to filter the sequencing data to make clean data. Then the cleaned reads were mapped to the S. lividans 1326 genome. Transcript expression levels were quantified using RSEM. The significant differentially expressed genes were identified by edgeR with definition of |log₂(fold change)| ≥ 1, false discovery rate (FDR) ≤ 1, and P < 0.05.

Screening of strains Streptomyces griseobrunneus CGMCC 29023 and Streptomyces fradiae CGMCC 29024

Mutagenesis of Streptomyces griseobrunneus CGMCC 29023:

Streptomyces griseobrunneus CGMCC 29023 was irradiated with a 6 kGy dose as described for S. lividans 1326. After mutagenesis, 120 clones were inoculated into 96-well plates, each containing 400 μL TSBY medium (3% tryptic soy broth, 0.5% yeast extract). After 24 h of incubation at 30 °C, 600 rpm, 20 μL from each well was transferred to a new 96-well plate with 600 μL GYB medium (2% glucose, 1% soluble starch, 0.3% corn extract, 1% yeast extract, 0.1% beef extract, 0.05% MgSO₄·7H₂O). The cultures were incubated for 7 days at 30 °C, 600 rpm. After centrifugation, 300 μL of supernatant from each well was subjected to solid-phase extraction using a C18 column (Phenomenex, 55 μm, 50 mg/well) in a 96-well format. Elution was performed with 500 μL 50% MeCN (in water), and the eluates were dried in vacuo, then resuspended in 30 μL 50% MeCN for HR-ESI-MS analysis.

Mutagenesis of Streptomyces fradiae CGMCC 29024:

The mutagenesis procedure for S. fradiae CGMCC 29024 followed the same protocol as for S. griseobrunneus CGMCC 29023.

Analysis of HR-ESI-MS data

High-resolution ESI-MS data were acquired using a Sciex ZenoTOF 7600 mass spectrometer (Sciex, Framingham, MA, USA), coupled to a Shimadzu Nexera LC-40DX3 HPLC system (Shimadzu, Kyoto, Japan). A Luna C18 column (Phenomenex, 1.6 μm, 30 × 2.1 mm) was used, and the mobile phases A and B consisted of water and MeCN, both with 0.1% formic acid. A 1 μL sample was injected into the mass spectrometer, with the following gradient elution program: 5% MeCN for 1 min, followed by 10–40% MeCN over 7 min, and 40–100% MeCN over 10 min, at a flow rate of 0.3 mL/min.

The mass spectrometer was operated in positive ion mode with the following parameters: spray voltage 5500 V, declustering potential 80 V, collision energy 10 V, accumulation time 0.5 s, and mass range 100-2000 Da. Source gas parameters were: ion source gas 1 and 2 at 50 psi, curtain gas at 35 psi, and gas temperature at 500 °C.

For visualization, LC-MS data were converted to the mzXML format using ProteoWizard-MSConvert software (ver 3.0). The data were binned in 1 m/z units and medium signals were excluded from 3D plots. Signals above an intensity threshold were plotted in MATLAB using the bar3 function. Visual inspection of the 3D plots allowed identification of metabolites whose production was induced or enhanced by HEPE mutagenesis.

Purification and structural elucidation of griseobrucins A-C

Griseobrucin A-C were purified from Streptomyces griseobrunneus CGMCC 29023 as follows:

The strain was inoculated from an SFM agar plate into 3 mL TSBY medium in a sterile 14 mL culture tube. After overnight growth, the culture was diluted to 0.5% (v/v) into 1 L flasks containing 200 mL GYB medium. A total of 10 L of fermentation culture was used for compound isolation. After 7 days of cultivation at 30 °C, 210 rpm, the supernatant was repeatedly loaded onto a prepacked C18 column (YMC, 10 µm, 20 g). Step-wise elution was carried out using 150 mL H₂O, followed by 15% and 30% MeOH (in water). Fractions containing griseobrucin A, B, or C (as determined by HR-LC-MS) were combined, dried in vacuo, resuspended in 15% MeOH, and further purified by loading onto a Sephadex LH-20 column (Cytiva). Elution was performed with 15% MeOH, and the fractions were again combined and dried.

For griseobrucin A purification, the sample was resuspended in 15% MeOH and purified by HPLC (Thermo Scientific UltiMate 3000) on a BEH Phenyl column (Waters, 5 µm, 10 × 250 mm), using a gradient from 3% to 8% MeOH over 30 min, then increased to 100% MeOH at 31 min, maintaining at 100% until 35 min. Fractions from 31–35 min were pooled, dried, and resuspended in 15% MeCN for further purification on an Amide C16 column (Supelco, 5 µm, 4.6 × 250 mm) with a 3–8% MeCN gradient over 20 min. Fractions collected at 5 min, containing griseobrucin A (1), were lyophilized to yield 3 mg.

Griseobrucin B and C were purified similarly, using an Amide C16 column (Supelco, 5 µm, 10 × 250 mm) with an isocratic elution of 3% MeCN for 20 min. Fractions collected from 10–18 min were pooled, dried, resuspended in 15% MeCN, and purified by HPLC on a Luna C18 column (Phenomenex, 5 µm, 4.6 × 250 mm) with isocratic 3% MeOH for 10 min. Fractions containing griseobrucin B and C (at 5.5 min and 6.5 min, respectively) were lyophilized to dryness, yielding 12 mg of griseobrucin B (2) and 10 mg of griseobrucin C (3).

NMR spectra were recorded using a Bruker Ascend 600 MHz (Bruker, Rheinstetten, Germany) in H₂O:D₂O (9:1). Data were analyzed using MestReNova software (MestreLab Research).

X-ray diffraction

Crystal of griseobrucin C (3) was obtained by slow evaporating of its methanol solution at room temperature over four days. A suitable crystal was selected and mounted on a Bruker D8 Venture diffractometer (Bruker, Karlsruhe, Germany). The data was collected at 170.0 K. The structure was solved by Intrinsic Phasing with SheLXT and refined by Least Squares Minimization with the SheLXL within the Olex2 program package.

Purification and structural elucidation of fradibactin A &B

Fradibactin was purified from Streptomyces fradiae CGMCC 29024 following a similar protocol: A total of 20 L of fermentation culture was processed as described for griseobrucin. After 7 days of cultivation, the supernatant was loaded onto a prepacked C18 column (YMC, 10 µm, 20 g) and eluted with 150 mL H₂O, 15% and 30% MeCN. The 15% fraction containing fradibactin was dried, resuspended in 15% MeCN, and purified by Preparative HPLC on a Triart C18 column (YMC, 5 µm, 20 × 250 mm) at 10 mL/min using a gradient of 5%-40% MeCN over 40 min. Fractions from 20–30 min were pooled, dried, resuspended in 15% MeCN, and further purified by HPLC on a BEH Phenyl column (Waters, 5 µm, 10 × 250 mm) at 2 mL/min, using a gradient from 8% to 20% MeCN over 30 min. Fractions from 19–24 min were pooled, dried, resuspended in 15% MeCN, and purified on an Amide C16 column (Supelco, 5 µm, 4.6 × 250 mm) with a 5-13% MeCN gradient over 35 min. Fractions at 29 min and 31.5 min containing fradibactin A and fradibactin B, respectively, were collected and lyophilized, yielding 2 mg of fradibactin A (4) and 0.8 mg of fradibactin B.

NMR spectra were recorded using a Bruker Ascend 600 MHz (Bruker, Rheinstetten, Germany) in DMSO-d₆. Data were analyzed in MestReNova (MestreLab Research).

Genetic inactivation in strain Streptomyces fradiae CGMCC 29024

Construction of Knockout Plasmids:

For the Δfdn insertional mutant, 2.1 kb upstream and downstream fragments of BGC fdn were amplified by PCR using primers 1/2 and 3/4. The 773 cassette, containing the apramycin resistance gene (apr) fused to oriT, was obtained by EcoRI/HindIII digestion of the vector pIJ773. Vector pJTU1289, carrying ampicillin and thiostrepton resistance markers (AmpR, TsrR), was digested with XbaI and KpnI. All fragments were gel-purified and assembled using the One Step Cloning Kit (Vazyme Biotech Co. Ltd) to create plasmid pΔfdn. The assembled plasmid was transformed into E. coli DH5α cells and verified by Sanger sequencing.

Conjugation and Mutant Selection:

The pΔfdn plasmid was transformed into E. coli ET12567/pUZ8002. E. coli ET12567 carrying the plasmid was cultured in LB with Amp (100 μg/mL), Apr (50 μg/mL), Kan (50 μg/mL), and Cm (25 μg/mL) to an OD600 of 0.4–0.6. Cells were centrifuged (4000 g, 10 min), washed twice with LB, and resuspended in 500 μL LB. Fresh spores of S. fradiae CGMCC 29024 were collected from SFM agar, washed with sterile water, and resuspended in TES buffer (50 mM, pH 8.0). The spore suspension was heat-shocked at 50 °C for 10 min, followed by a 1 min ice bath and incubation with 2×spore activation buffer (1% yeast extract, 1% casein hydrolysate, 5 mM CaCl₂) at 37 °C/200 rpm for 2 hours. The spores were collected (1000 g, 5 min), resuspended in LB, and mixed with E. coli ET12567 suspension. The mixture was plated on SFM agar and incubated at 30 °C for 18–20 h. Exconjugants were selected on SFM agar containing Apr (50 μg/mL) and trimethoprim (50 μg/mL). After 3–5 days of incubation, single colonies were picked, grown on SFM agar with Apr, and further validated. Exconjugants were cultured in TSBY medium (Apr, 50 μg/mL) for 2 days at 30 °C/210 rpm, then subcultured without antibiotics for 1 day, followed by a second subculture for 1 day. Surviving cells were plated on SFM agar, and single colonies were tested for thiostrepton (Tsr) susceptibility (10 μg/mL) and apramycin resistance (50 μg/mL). Double-crossover mutants resistant to Apr and susceptible to Tsr were confirmed by PCR using primers 5/6 and 7/8.

A successful Δfdn double-crossover mutant was obtained, which was unable to produce fradibactin.

Test of in vitro antimicrobial activity

Antimicrobial activity was tested against Staphylococcus aureus ATCC 29213, Enterococcus faecalis ATCC 19433, Pseudomonas aeruginosa ATCC 27853, Escherichia coli ATCC 25922, Stenotrophomonas maltophilia ATCC 13637, Candida albicans ATCC 90028, Candida krusei ATCC 6258, Candida glabrata ATCC 15126, and Candida parapsilosis ATCC 22019. Bacteria were cultured in LB medium at 37 °C, while fungi were cultured in potato dextrose broth (PDB) at 30 °C. Minimum inhibitory concentration (MIC) was determined using the microtiter method according to CLSI guidelines. Overnight cultures of the test strains were diluted to an initial OD₆₀₀ of 0.005 in 100 μL per well in a 96-well plate. Compounds were tested at concentrations of 0, 0.5, 1, 2, 4, 8, 16, 32, and 64 μg/mL. Plates were incubated at 37 °C for 16 hours, and OD₆₀₀ was measured before and after incubation. MIC values were defined as the compound concentration that reduced OD₆₀₀ by >95% compared to the untreated culture. Controls included a negative control (90 μL bacterial culture + 10 μL saline), positive controls (nisin and vancomycin), and a blank (100 μL MH-II broth without bacteria and drugs). All experiments were performed in triplicate.

Test of cytotoxic activity

The cytotoxic activity screening was conducted by experimental staff at Qingdao AC biotechnology co., Ltd using a commercially available Cell Counting Kit-8 (CCK-8) assay. The compounds were evaluated against a panel of twenty human cancer cell lines, including A549 (BNCC337696), MKN-45 (CL-0292), HCT 116 (BNCC287750), HeLa (CL-0101), HepG2 (BNCC 338070), K-562 (CL-0130), 786-O (CL-0010), TE-1 (CL-0231), 5637 (CL-0002), GBC-SD (CL-0085), L-02 (BNCC338358), MCF-7 (CL-0149), SF126 (CL-0435), DU145 (BNCC1000143), CAL-62 (CL-0618), PATU8988T (CL-0579), HOS (CL-0360), A-375 (CL-0014), A-673 (CL-0017), HEK-293T (CL-0005). All cells were obtained from BeNa Culture Collection or Wuhan Procell. All cell lines were authenticated by short tandem repeat (STR) profiling within 6 months prior to experimentation and were confirmed to be free of mycoplasma contamination.

Screening of strain Streptomyces clavuligerus 70116

Mutagenesis of Streptomyces clavuligerus 70116:

Streptomyces clavuligerus 70116 was irradiated with HEPE (6 kGy) as described for S. lividans 1326.

Mutant Screening:

After irradiation treatment, cells were resuspended in 200 μL sterile water, serially diluted, and spread on SFM agar plates. Plates were incubated at 28 °C for 5–7 days. One hundred and ten clones were randomly selected and inoculated into 500 mL Erlenmeyer flasks containing 100 mL seed medium (3% soybean, 2% dextrin, 0.5% yeast extract, 0.1% K₂HPO₄; pH 7.0), cultured at 28 °C, 250 rpm for 72 h. For small-scale flask cultivation, 250 μL of the cultures were transferred to 500 mL Erlenmeyer flask containing 30 mL fermentation medium (2% glycerol, 0.5% olein, 2% dextrin, 2.5% peptone, 2.1% morpholine propyl sulfonic acid, 0.2% K₂HPO₄, 0.01% NaCl, 0.02% MgCl₂, 0.02% CaCl₂2H₂O, 0.003% FeCl₃, 0.0005% CuCl₂, 0.005% MnSO₄, 0.0005% ZnCl₂; pH 7.0), followed by 144 h incubation at 25 °C, 250 rpm.

Clavulanic acid production analysis:

Cells were removed by centrifugation (10,000 g, 10 min). The clarified supernatant containing clavulanic acid was filtered through 0.22 μm membranes prior to HPLC. Analytical separation employed an SB-C18 column (4.6 mm × 100 mm) column under thermostatic control (33 ± 0.5 °C). The injection volume was 20 μL, and detection wavelength was set at 210 nm. The mobile phase consisted of 10 mmol/L tetrabutylammonium hydrogen sulfate and methanol (85:15, v/v), adjusted to pH 6.0.

Screening of strain Escherichia coli CGMCC 16349

Mutagenesis of Escherichia coli CGMCC 16349:

Cryopreserved E. coli CGMCC 16349 cultures were plated on LB agar (1% Tryptone, 0.5% yeast extract, 0.5% sodium chloride, 2% agar) supplemented with 100 µg/mL Ampicillin and 50 µg/mL Kanamycin. A single colony was inoculated into LB broth and incubated overnight at 37 °C. Subsequently, the cultures were washed twice with distilled water and resuspended in sterile water, adjusting the optical density to 0.4 at 595 nm. For irradiation mutagenesis, 20 μL of the cell suspension were subjected to three different treatments: UV (60, 120, 180, and 240 s), ARTP (30, 60, 90, and 120 s), or HEPE (1, 2, 3, and 4 kGy). The mutagenesis process was the same as described for S. lividans 1326.

Mutant Screening:

After irradiation treatment, cells were resuspended in 200 μL sterile water, serially diluted, and spread on LB agar plates supplemented with 100 µg/mL Ampicillin and 50 µg/mL Kanamycin. Plates were incubated overnight at 37 °C. Eighty clones per irradiation treatment were randomly selected and inoculated into 12 mL sterile tubes containing 5 mL TSBY medium (100 µg/mL Ampicillin, 50 µg/mL Kanamycin), cultured overnight at 37 °C, 210 rpm. For small-scale fermentation, 100 μL of the cultures were transferred to 100 mL Erlenmeyer flask containing 20 mL LB medium, and incubated for 24 h at 37 °C, 210 rpm.

Fed-batch fermentation in a 40 L bioreactor:

The fed-batch fermentation of high-yield mutant was performed in a 40 L bioreactor (Sinagri Yingtai Bio-peptide Co.,Ltd.). The medium was composed of 3.2% corn flour, 2% soybean meal, 1.3% peptone, 0.2% KH₂PO₄ and 2% (NH₄)₂SO₄. Dissolved oxygen (DO) and pH were measured by a DO probe and a pH electrode, respectively. The temperature was maintained at a constant 37 °C, and pH at 7.0. The fed-batch process was initiated after 5 h of cultivation, using a concentrated glucose stock solution to maintain a glucose concentration of 0.15 g/L in the bioreactor. Fermentation broth was collected every 4 h during the fermentation process for monitoring cell growth (OD₆₀₀) and the titer of microcin J25.

Microcin J25 Production Analysis:

Cells were removed by centrifugation (16,000 g, 10 min), and 1 mL of clarified supernatant from each flask was loaded onto a prepacked C18 column (YMC, 10 µm, 0.5 g). Elution was performed with 1 mL 50% MeCN in water, and the eluates were directly analyzed by high-resolution electrospray ionization mass spectrometry (HR-ESI-MS).

The titer of microcin J25 in bioreactor was analyzed by HPLC. Fermentation broth was centrifuged at 13,500 g for 10 min. The resulting supernatant was filtered through a 0.22 µm membrane for subsequent HPLC detection. A serial dilution of microcin J25 standard stock solutions was prepared for quantitative analysis. An Agilent 1260 series HPLC system equipped with a ZORBAX 300SB-C18 column (5 µm, 4.6 × 250 mm) was used for detection. The mobile phase consisted of 0.1% trifluoroacetic acid in water-acetonitrile (950:50, v/v, phase A) and 0.85% trifluoroacetic acid in water-acetonitrile (100:900, v/v, phase B), flowing at a rate of 1 mL/min. The detection wavelength was set at 214 nm.

Screening of strain Monascus purpureus M_Y2

Mutagenesis of Monascus purpureus M_Y2:

Cryopreserved M. purpureus M_Y2 cultures were plated on PDA medium (4.8% Potato leach powder, 8% Glucose, 5.6% Agar, pH5.6 ± 0.2). After 7–10 days of incubation at 30 °C, spores were harvested, transferred to a 50 mL flask with 10 mL sterile water, and filtered through a 20 mL sterile syringe. The spore concentration was adjusted to 10⁶ CFU/mL. 20 μL of the spore suspension was irradiated with UV (20, 30, 40, and 50 s), ARTP (25, 50, 70, and 90 s), or HEPE (1, 2, 3, and 4 kGy) as described for S. lividans 1326.

Mutant Screening:

After irradiation treatment, spores were suspended in 200 μL sterile water, serially diluted, and spread on PDA plates. Plates were incubated at 30 °C for 7–10 days. Sixty clones from each treatment were randomly selected and inoculated into 250 mL Erlenmeyer flask containing 100 mL seed medium (8% glucose, 1% peptone, 0.2% NH₄H₂PO₄, 0.05% MgSO₄·7H₂O, 0.01% CaCl₂), followed by 48 h incubation at 30 °C, 140 rpm. For solid-phase fermentation, the seed liquid was transferred to a 4 L glass fermenter with 400 g of indica rice at a 10% (v/m) ratio and incubated at 30 °C and 60% humidity for 14 days. After incubation for 48 h, the rice medium was turned twice daily.

Lovastatin production analysis:

After fermentation, 0.5 g of the culture was extracted with 30 mL 70% acetonitrile, 30% phosphoric acid aqueous solution (0.01%) in a shaking bath at 50 °C for 1 h. The solid residue was removed by centrifugation (13,500 g, 5 min), and the supernatant was filtered through a 0.22 µm microporous membrane. Lovastatin production in the filtered supernatant was quantified by HPLC on Shimadzu LC-2010AT system using a ZORBAX Eclipse XDB-C18 column (250 mm × 4.6 mm, 5 µm). An isocratic elution of 70% MeCN-H₂O (0.01% phosphoric acid) over 15 min was used, with a flow rate of 1 mL/min. The ultraviolet detection wavelength was 238 nm.

Statistics and reproducibility

Data are presented as the mean ± SEM of three independent biological replicates unless otherwise stated. No statistical method was used to predetermine sample size. No data were excluded from the analyses. Comparisons among groups were performed using one-way ANOVA. Statistical analyses were performed using IBM SPSS Statistics (version 25), and statistical significance was defined as *P < 0.05, **P < 0.01, ***P < 0.001. n and P values are described in the figures or figure legends or Source Data file. All experiments were independently repeated at least three times with similar results.

Author contributions

X.F., W.W. and F.X. designed the research; X.F., S.Z. and Z.Y. evaluated irradiation methods; X.F. and Z.X. conducted genomic analysis; X.F. and Y.Z. conducted metabolites activation in S. lividans; Z.L. and W.W conducted production enhancement of clavulanic acid; X.F., Y.Z., D.Q., T.H. and J.W. conducted production enhancement of mccJ25; J.X., J.Y. and W.L. conducted production enhancement of lovastatin; X.F. and M.R.S. performed 3D plot analysis; X.F. conducted genetic inactivation, fermentation and compound isolation of novel metabolites; J.Y. and M.W. elucidated compounds structures; M.R.S., W.W. and F.X. wrote the manuscript.

Peer review

Peer review information

Nature Communications thanks Jean Molinier, Gerard Wright, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Genome and Hi-C sequencing data for S. lividans 1326 and its mutants have been deposited in the NCBI under BioProject PRJNA1264485. RNA sequencing data are available from the NCBI database under accession number GSE315268. Genome sequencing data for S. griseobrunneus and S. fradiae are available via NCBI under accession numbers SAMN45944959 and SAMN45945006. The mass spectrometry-based metabolomics raw data have been deposited in the MassIVE database under accession code MSV000100599. The X-ray crystallographic coordinates for structure reported in this study have been deposited in CCDC under accession number 2432012 [https://www.ccdc.cam.ac.uk/structures/search?id=doi:10.5517/ccdc.csd.cc2mmq00] Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-69723-3.