Streptomyces sp. MG010 was isolated from the marine sponge Mycale sp. located off the coast of Fujian, China. Gram-positive (Staphylococcus aureus ATCC 6538 and Bacillus subtilis ATCC 6633) and Gram-negative bacteria (Escherichia coli ATCC 25922 and Vibrio alginolyticus ATCC 33783) were used as indicators for the antibacterial activity assays.

ISP2 medium and LB medium were used for strain screening, fermentation, and antimicrobial bioassays. All media were sterilized at 115°C for 30 min before use (Bertani, 1951; Shirling and Gottlieb, 1966).

Streptomyces sp. MG010 strain was inoculated into ISP2 solid medium and incubated at 28°C for 10 days. The mature spores were harvested using sterile cotton swabs scraped over the medium surface, and then washed twice with sterile saline. The spore pellets were then transferred to an Eppendorf tube with glass beads, and the spore chains were disrupted by shaking for 10 min. After centrifuging at 12,000 × g for 10 min and discarding the supernatant, the spore precipitate was carefully resuspended in 1 mL of sterile saline. The monospore suspension was counted using a blood counting chamber and the concentration was adjusted to 10⁷–10⁸ CFU/mL.

The operating parameters were as follows: an input voltage of 110 W, distance of 2 mm between the plasma torch nozzle exit and sample plate, pure helium used as the working gas, a gas flow rate of 10 slpm, and the temperature of the plasma jet was below 40°C. The monospore suspension (10 μL) was poured into a sterilized stainless-steel plate (6 mm diameter) and exposed to the ARTP plasma jet (SiQingYuan Inc., Wuxi, China) for different treatment times: 0, 30, 45, 60, 75, 90, 105, 120, 150, and 180 s.

To obtain the optimum mutation, the lethality and positivity rates of spores under different mutant treatment times were determined according to the following equations:

where A is the total colony count of the sample without treatment and B represents the total colony count after the different mutant treatment times with ARTP.

where C is the total colony count of the mutants with increased antibacterial activity and D represents the total colony count of mutants with varying antibacterial activity.

After treatment, the steel plates were placed into new 1.5 mL Eppendorf tubes and washed with 1 mL sterile saline. Then, 100 μL spore suspension was separately transferred onto ISP2 solid medium and cultured at 28°C for 5 days. Pure mutant colonies from the solid medium were selected and inoculated onto LB solid medium with typical indicative bacteria at 37°C for 1 day using the replica plating method. The diameters of the inhibitory zones were observed and measured. Next, 100 μL of the cell-free supernatants of the fermentation broths from potentially active mutants was drawn out for verification of the antibacterial activity using the Oxford Cup assay. The top mutants with increased antimicrobial activity were selected as the original isolates for the next round of ARTP mutagenesis. A total of three mutagenesis iterations were performed along with the screening methods (Figure 1). In addition, the genetic stability of the identified mutant isolates was evaluated by eight rounds of serial sub-cultivations.

The 16S rRNA gene sequences (∼1500 bp) of the MG010 strain were amplified and analyzed by the National Center of Biotechnology Information (NCBI) website¹ to select the most related reference strains. The nucleotide sequence data showed that Streptomyces sp. MG010 (GenBank accession number: MW186238) in this study shared 99% homology with Streptomyces variabilis strain HNS054 (GenBank accession number: ASM104418v1). Therefore, the S. variabilis strain HNS054 was chosen as the reference genome for subsequent experiments. Antibiotics and secondary metabolite analysis shell (antiSMASH) version 3.0.5 was used to analyze the potential to produce secondary metabolites.

Total RNA was extracted and purified from the wild-type strains (labeled NA1, NA2, and NA3) without testing their bioactivity against indicative bacteria, weakly bioactive mutants (labeled I1, I2, and I3), or the strongest bioactive mutants (labeled M1, M2, and M3) in the three parallel samples using the TIANGEN RNAprep Pure Micro Kit (TianGen, China) and RNA purification kit (TianGen, China), according to the manufacturer’s instructions. The purity and concentration of the RNA (rRNA 28s/18s) were evaluated using a NanoDrop 2000 spectrophotometer (NanoDrop, Thermo Scientific, United States), and the integrity of the RNA was assessed using agarose gel electrophoresis. Then, all the RNA samples were subjected to RNA-seq library preparation and directed sequencing on the Illumina HiSeq platform (Meiji, China).

After filtering the low-quality sequences, a high content of unidentified nucleotides N, and adaptor-polluted sequences, clean reads were assembled using the software provided by sequencing platform Illumina. Clean reads were mapped to the reference Streptomyces sp. and the HNS054 genome (accession number: NZ_LDZX00000000.1, available at: https://www.ncbi.nlm.nih.gov/nuccore/NZ_LDZX00000000.1) using the software Bowtie2². The abundance of gene transcripts from RNA-seq was quantified using the software RSEM package³. RPKM (reads per kilobase of exon model per million mapped reads) was used to indicate the abundance of a transcript. The edge R⁴ was used to analyze differential expression analysis (fold change) and the related significance level between the wild and mutant strains. The false discovery rate (FDR) values below 0.05 or log₂ | fold change| ≥ 1 were considered to be significantly differentially expressed transcripts.

Gene functions were annotated by online programs and databases, including Gene Ontology (GO), Clusters of Orthologous Groups of protein (COG), and KEGG. Among these, GO terms enriched differentially expressed genes (DEGs) were analyzed using the Blast2GO software package⁵. The GO terms with significant expression (p-value < 0.05) were also enriched in KEGG pathways⁶; COG⁷ terms were used as tools for functional annotation, classification, and protein evolution analysis. Other databases such as Nt (NCBI non-redundant nucleotide sequences), Nr (NCBI non-redundant protein sequences), and Pfam (Protein family) were also used to annotate gene functions.

A total of nine RNA samples were collected for quantitative real-time (qRT)-PCR analysis to verify the transcriptome data. The primers used for qRT-PCR were designed using the Premier 5 software. The detailed information for the primers is listed in Supplementary Table S3. The qRT-PCR reaction system was performed following the manufacturer’s instructions using the SYBR Premix EX Tag kit (Takara, Japan), containing 10 μL of 2 × SYBR Premix EX Tag, 2 μL of each cDNA sample, 0.5 μL of Primer (10 μM), and added RNase Free H₂O up to a final volume of 20 μL. The reaction conditions were as follows: 95°C for 2 min; 40 cycles (95°C for 5 s; 60°C for 30 s) and 72°C for 30 s. The melting curve conditions were as follows: 95°C for 15 s; 60°C for 60 s; and 95°C for 15 s. The 16S rRNA of the wild strain was used as the reference gene in this study. QuantStudio 6 (Thermo Fisher Scientific, United States) was used to carry out all qRT-PCR reactions by detecting the SYBR fluorescence signal strength. Primer Quest Tool software was used to design primers with a product length of 100–300 bp (Invitrogen, ShangHai). The 2^−ΔΔCt method was used for data processing.

The wild-type strain (NA), weakly mutant strains (IA), and maximum antimicrobial activity of the mutant strains (MA) were first grown on ISP2 medium at 28°C for 5 days, after which the spores were inoculated into a 2.5 L Erlenmeyer flask containing 2.5 L ISP2 medium at 220 rpm and 28°C for 7 days. The crude extracts were obtained by extracting the fermentation broths with equal volumes of ethyl acetate and then butanone three times. The clarified supernatant was obtained by centrifuging the concentrates at 12,000 × g for 5 min. Both the ethyl acetate and butanone layers were collected and evaporated on a rotatory evaporator, and the concentrates were resolved in methanol (500 μL). The samples were adjusted to 5 mg/mL and analyzed by LC-MS with a linear gradient of 5–95% solvent B (solvent B: 0.1% formic acid + 100% CH₃CN; solvent A: 0.1% formic acid in H₂O) over 40 min at a rate of 1 mL/min with a UV length of 254 nm.

The free software MZmine2⁸ was used to assess chromatograms by detecting, deconvoluting, aligning, and normalizing the data. Only those peaks with a high-quality mass intensity were retained, while the rest were discarded.

The major predicted active fractions were collected by semi-preparative HPLC with a C₁₈ column using a CH₃CN–H₂O elution system as previously described to identify the relatively pure peaks. The fractions were then dissolved in DMSO and adjusted to a uniform concentration of 100 μg/mL. Bacterial indicators were selected for antibacterial activity assays using the disk paper method. After cultivation for 24 h, the diameters of the bacteriostatic circles were measured.

The survival rate of Streptomyces sp. decreased with an increase in the mutagenesis time (Figure 2A). The lethal rates sharply increased when strains were exposed to ARTP before 75 s, and were observed to plateau at a treatment time of 75 s, with the death rate reaching 94% for the protection of the spore cell wall. Meanwhile, no spores survived treatments of 90 s or longer. Therefore, 75 s was chosen as the optimal exposure time for all subsequent experiments.

After three mutagenesis iterations, a total of 813 mutants were selected for antimicrobial activity screening. The mutation rates decreased as the positive mutation rates increased from 8.97 to 40.94%, indicating that interactive mutagenesis increased the possibility of emerging active strains (Supplementary Table S1). The antibacterial activities were then compared among the optimal mutants in each round (Figure 2B). Compared with the wild-type strain, the inhibitory diameter increased by 68% (first round), 71% (second round), and 75% (third round) after each round of ARTP treatment under the same culture conditions.

To detect the genetic stability of the mutant strains, eight rounds of subculture shake flask fermentation of the optimal mutants were conducted. The results indicated that the inhibitory activity of the identified mutant strains retained effective genetic stability (Figure 2C).

Partial characteristics, including the thermal stability and protease degradation activity, of one selected representative mutant were analyzed. The specific experimental methods are listed in Supplementary Figure S5. The bioactivity of the fermentation broth against indicative bacteria did not significantly differ when placed at 80°C for 4 h; however, no activity was observed above 80°C (Supplementary Figures S5A,B). The active substances from the fermentation broth were not sensitive to trypsin or chymotrypsin (Supplementary Figure S5C). Briefly, the results showed that the active substances of the mutant strains in the fermentation broth had excellent thermal stability and resistance to protease degradation.

The transcriptome of nine mutants (NA, IA, and MA in three parallels, respectively) were sequenced using the RNA-seq Illumina platform. In total, 177,898,864 clean reads were obtained from the NA, IA, and MA groups, among which 164,556,861 reads had a mapped ratio larger than 80%. The correlation coefficient between the wild strain and the mutants showed good quality of the collected cDNA reads (Supplementary Figure S1A). Compared with the wild strain, the clean reads mapping to the unique genes demonstrated that 1712 transcript genes of the IA mutants were significantly changed, including 956 upregulated and 756 downregulated genes, while 1812 transcript genes of the MA mutants were significantly changed, including 1145 upregulated and 676 downregulated genes (Supplementary Figure S1B). The significance of these gene transcriptions was determined by both FDR < 0.05 and | log₂FC| ≥ 1.

Seven of the total 825 unigenes were annotated to three major functional groups using the GO categories, including biological process (4584 unigenes), molecular function (1900 unigenes), and cellular component (1341 unigenes; Supplementary Figure S2A). Comparing the NA and MA mutant strains, the most DEGs were found to be associated with biological processes, with equal numbers of upregulated and downregulated genes. Between the IA and MA mutants, remarkable changes were observed, with more upregulated than downregulated genes (Supplementary Table S2). Based on the DEGs, five of the total 123 unigenes were assigned to 23 function units based on the COG function annotation, with the largest number assigned to transcription (387 unigenes), accounting for 7.55% (Supplementary Figure S2B). A total of 5051 unigenes were also annotated to KEGG pathways involved in metabolic pathways (815 unigenes) and biosynthesis of secondary metabolites (370 unigenes), among which 20 pathways were significantly enriched, including ABC transporters, carbon metabolism, and biosynthesis of amino acids (Supplementary Figure S2C).

Comparative analysis using antiSMASH version 3.0.5 of functional gene clusters between the wild-type strain and mutants revealed a total of 21 gene clusters related to secondary metabolites that were significantly differentially expressed (Table 1). Using these DEGs, we conducted pairwise comparisons (Figure 3). The transcription results responding to ARTP mutagenesis illustrated that certain metabolic pathways, such as ribosomal metabolic processes, peptide and antibiotic biosynthesis and metabolism processes, transcription factors, and protein binding transport, became more active with an increase in antibacterial activity. Several gene clusters related to PKS or NRPS compounds were also differentially expressed between the NA and MA groups, such as cluster 11 and 12 (T1 PKS-NRPS), cluster 23 (T3 PKS), cluster 37 (NRPS), and cluster 51 (T2 PKS). Additionally, type I or IIIPK synthase and peptide synthase were significantly increased, although the numeric values of these transcriptome data (FPKM) ranged from 10 (at minimum) to 100 (at maximum) on average. Cluster 42, which was predicted to produce Lantipeptide, also showed higher gene expression levels and fold change differences in ABC transporter, membrane protein, and nitroreductase family deazaflavin-dependent oxidoreductase.

The expression levels of each gene in cluster 17 showed a trend toward significant upregulation, with FPKM ranging from 10 to 3000 on average. This gene cluster, 22,493 bp in length, located in 11,658–34,178 nt, was predicted to encode and synthesize lasso peptide compounds with a molecular weight of 2164 Da using antiSMASH software (Figure 4A). The gene cluster included the DNA-binding response regulator, sensor histidine kinase, export ABC transporter ATP-binding protein, lasso RiPP family leader peptide-containing protein, lasso peptide isopeptide bond-forming cyclase, lasso peptide biosynthesis PqqD family chaperone lasso peptide biosynthesis B2 protein, and ABC transporter ATP-binding protein (Table 2). Among these, AC003_RS25355 contained the complete leader peptide and core peptide encoding the lasso peptide compounds (Figure 4B).

To validate the confidence of the RNA-seq data, five genes were selected randomly from the NA, IA, and MA mutant strains for qRT-PCR analysis (Supplementary Table S3). The linear regression relationship of gene expression levels was effective, with R² of the NA, IA, and MA mutants reaching 0.9993, 0.998, and 1, respectively, while R² of the reference 16S rRNA genes reached 0.99 (Supplementary Figure S3). The fold changes in the gene expression differences between the qRT-PCR results and transcriptomic data showed that although there was little deviation in these quantitative results, the up- and downregulated trends of these DEGs were the same (Supplementary Figure S4 and Supplementary Table S4). The other eight genes (AC003_RS25345, AC003_RS25350, AC003_RS25355, AC003_RS35325, AC003_RS25365, AC003_RS25370, AC003_RS25375, and AC003_RS25380) in cluster 17 that predicted for lasso peptide also underwent several rounds of qRT-PCR analysis. Although the low gene expression level limited these results, significant differences were observed in the gene expression fold changes between the MA and NA mutants, thereby verifying that the changes in gene expression of the lasso peptide cluster were relatively accurate.

The butanone-extracted samples of the wild-type strain (NA) and mutants (IA, MA) were analyzed by HPLC-MS. The results of metabolite profiling showed that the MA strain produced a higher yield of the putative lasso peptide (RT = 17.35 min) compared to the wild-type strain (NA) or IA (Figure 4C). LC-MS data further revealed a relative molecular mass of 1082.43 ([M + 2H]²⁺ at m/z = 2164.86) (Figure 4D). The genomics and metabolomics analyses of this signal were both targeted to the molecule aborycin (Frechet et al., 1994). Therefore, ARTP mutagenesis might awaken the silent BGC and, to some extent, improve the production of this lasso peptide.

Additionally, the metabolic profiles of ethyl acetate extracted crudes between the wild strain (NA) and mutant strain (MA) were compared by LC-MS, determining their TIC (total ion count) and XIC (extracting ion current) (Figure 5). MA produced a series of low-molecular-weight compounds that did not contain in the NA, such as [M + H]⁺ at m/z = 261.1231 (RT = 7.45 min), [M + H]⁺ at m/z = 374.1642 (RT = 16.22 min), [M + H]⁺ at m/z = 386.1256 (RT = 17.85 min), and [M + H]⁺ at m/z = 304.2993 (RT = 20.25 min) (Figures 5A,B). To further predict these molecules, several online databases were used to identify the putative molecular formulas and structures (Figure 5C). Moreover, the spectroscopic data indicated that several unknown peaks with strong base peak intensities potentially represent new molecules with intriguing bioactivities. A panel of 12 putative fractions was used to detect potent antibacterial activities against the indicators (Figure 5D). Fractions B, D, E, and F were significantly different between the NA and MA in liquid chromatograms, among which fractions D, E, and F were against S. aureus with an MIC of 100 μg/mL. Meanwhile, component G and H did not exhibit antibacterial activities here, and the corresponding compounds could not be obtained from online databases, implying identification of novel compounds.