Fish T. bernacchii (66.8 g) was collected by ice fishing from the Antarctic China Zhongshan Station (69°22′27”S, 76°22′20″E) in February 2020. After the morphological identification, the fish was characterized using DNA barcoding of cytochrome oxidase subunit I (COI). The fresh fish was sealed and stored in sterile plastic bags immediately at 4°C for subsequent experiments. The intact fish abdominal contents were aseptically dissected from the fish, and 2 ml were taken into a centrifuge tube to investigate the composition and function of symbiotic bacteria.

The fish abdominal contents were diluted with 3.4% NaCl to make dilutions. Each dilution (200 μl) was plated on the surface of the isolation medium. The media used for isolation were luria-bertani agar (LB), nutrient agar (NA), tryptose soya agar (TSA), reinforced clostridium medium (RCM), modified artificial seawater medium (MASM), potato dextrose agar (PDA), yeast malt glucose (YMG), minimal medium (MM; Supplementary Table S1), brain heart infusion agar (BHIA, Hopebio, Qingdao, China), 2216E agar (Acmec, Shanghai, China). The plates were incubated at 20°C for 3 days and bacterial growth were detected. The distinct colonies were picked up and purified via repeated streaking (at least three times) on the isolation medium, which then were maintained in 50% glycerol and preserved at −80°C for further study.

Genomic DNA was extracted from bacteria cells using the Rapid Bacterial Genomic DNA Isolation Kit (Sangon, Shanghai, China) according to the manufacturer’s instructions. The 16S rDNA genes were amplified by PCR using 27F (5’-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5’-GGTTACCTTGTTACGACTT-3′) primers. The amplified DNA sequences were analyzed by Azenta Life Science Co., Ltd. for 16S rDNA sequencing, and the 16S sequences were aligned and identified against existing sequences in the NCBI database using the BLAST program. Further, the nucleotide sequences of the isolates were aligned with closely related sequence using Mega software (Kumar et al., 2008) and a phylogenetic tree was constructed to show the relationship between the isolates and the reference strains.

Genomic DNA was extracted using the E.Z.N.A.^® soil DNA Kit (Omega Bio-tek, Norcross, GA, United States) based on the manufacturer’s protocol. The hyper-variable regions V3-V4 of the 16S rRNA gene were amplified using a universal primer set 338F (5’-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5’-GGACTACHVGGGTWTCTAAT-3′). Sequencing of the 16S amplification was performed by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China) on an Illumina MiSeq PE300 platform (Illumina, San Diego, United States) with two paired-end read cycles of 300 bases each. The raw 16S rRNA gene sequencing reads were demultiplexed, quality-filtered by fastp version 0.20.0 (Chen et al., 2018) and merged by FLASH version 1.2.7 (Magoč and Salzberg, 2011). Operational taxonomic units (OTUs) with 97% similarity cutoff (Stackebrandt and Goebel, 1994; Edgar, 2013) were clustered using UPARSE version 7.1 (Edgar, 2013), and chimeric sequences were identified and removed. The taxonomy of each OTU representative sequence was analyzed by RDP Classifier version 2.2 (Wang et al., 2007) against the 16S rRNA database using confidence threshold of 0.7.

The Wizard® Genomic DNA Purification Kit (Promega) was used for genomic DNA extraction according to the manufacturer’s protocol. We utilized the Nanopore (Oxford Nanopore) and Illumina HiSeq X Ten sequencing platforms to perform whole-genome shotgun sequencing. The sequencing service was provided by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China).

The original image data was transferred into sequence data via base calling, which was defined as raw data or raw reads and saved as FASTQ file. A statistic of quality information was applied for quality trimming, by which the lower quality data was removed to form clean data. The reads were assembled into a contig using hierarchical genome assembly process (HGAP) and canu. The last circular step was checked and finished manually, generating a complete genome with seamless chromosomes and plasmids. Finally, error correction of the assembly results was performed using the Illumina reads. Glimmer (Delcher et al., 2007) was used for CDS prediction, tRNA-scan-SE (Chan and Lowe, 2019) was used for tRNA prediction and Barrnap was used for rRNA prediction. The predicted CDSs were annotated from NR, Swiss-Prot (Bairoch and Apweiler, 2000), Pfam (Finn et al., 2013), GO (Consortium G O, 2004), COG (Lars Juhl et al., 2008) and KEGG (Ogata et al., 1999) database using sequence alignment tools such as BLAST (Ye et al., 2006), Diamond (Neshich et al., 2005) and HMMER (Potter et al., 2018). Briefly, each set of query proteins were aligned with the databases, and annotations of best-matched subjects (e-value <10⁻⁵) were obtained for gene annotation.

Colonies of L7-1 were inoculated into 250 ml of sterile LB medium and then incubated at 20°C for 24 h on a rotary shaker (180 rpm) to produce seed broth. The seed broth (12 ml) was transferred into a 500 ml Erlenmeyer flask, containing fresh rice solid medium (rice 40 g, sterile seawater 60 ml). The flasks were in a state for cultivation at 20°C for 30 days. A total of 50 flasks were prepared for this study. The rice solid culture of L7-1 was extracted with ethyl acetate three times and evaporated dry by rotary evaporator to give a crude extract for subsequent analysis.

The crude extract (16.7 g) was fractionated by a silica gel column eluting with mixed solvents of PET/EtOAc (2/1, 1/1, 1/2, 1/5, 1/10, v/v), and EtOAc/MeOH (5/1, 1/1, 1/5, v/v) to furnish eight fractions (Fr-1-Fr-8) based on the results of TLC analysis. Fr. 4 (3.4 g), was further purified by Octadecyl-functionalized silica gel column (ODS, Cosmosil 75C₁₈-Prep) with MeOH/H₂O (3,7, 1,1, 7:3, 1:0, v/v) to afford four subfractions (Fr. 4.1–4.4). Compound 1 (61 mg, t_R 20 min, MeOH/H₂O, 30/70) was purified from Fr. 4.3 by Agilent 1,260 HPLC using a column (Agilent Zorbax SB-C₁₈, 250 × 9.4 mm, 5 μm; flow rate: 1 ml/min; UV detection: 210 nm). HPLC and analytic grace solvents used for this study were purchased from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China).

Optical rotation was recorded on a RUDOLPH Autopol I Automatic polarimeter (Rudolph Research Analytical). NMR spectra were acquired on a Bruker 500 spectrometer or a JEOL 600 spectrometer using the standard programs and acquisition parameters and the chemical shifts were expressed in δ (ppm) relative to DMSO-d6 (δC 39.5 and δH 2.50).

The indicator bacteria used in the bacteriostatic circle experiment were provided by Shanghai Rainbowfish Company, including Chromobacterium violaceum ATCC 12472, Salmonella choleraesuis, Mycobacterium smegmatis MC2155, Escherichia coli MG1655, Enterococcus faecalis FA2-2, Staphylococcus aureus ATCC25923. Cultures of six indicator bacteria (with an OD of approximately 0.5) were inoculated in LB medium in an amount of 100 μl. The sterilized circular filter paper sheets were attached to the LB. Dissolve 10 mg of the crude extract in 1 ml of methanol. Then 6 μl of the crude extracted metabolites and negative control (methanol) were dropped onto the circular filter paper sheet, after incubated at 37°C for 12 h. Antibacterial bioactivity was evaluated by measuring the area of the zones inhibition and calculating the inhibition rate, which was measured with Image J.

The antibacterial activities of serranticin inhibiting the growth of C. violaceum, S. choleraesuis, M. smegmatis, E. coli, E. faecalis, S.aureus E. tarda, and A. hydrophila were determined by the micro-broth dilution method (Xin et al., 2012). Gentamicin was used as a positive control and DMSO was used as a negative control.

Human lung cancer NCI-H460, human breast carcinoma MCF-7 and MDA-MB-231 cell lines were obtained from the National Collection of Authenticated Cell Cultures (Shanghai, China). All cells were incubated at 37°C in a humidified incubator with 5% CO₂ incubator. Cells after the third generation were used for experiment.

The assay using 3-(4,5-dimethylthiazole-2-yl)-2,5-dip-. henyltetrazolium bromide (MTT) against the MDA-MB-231, NCI-H460, and MCF7 cell lines (Huang et al., 2008). Cells (3 × 10³ per well for cancer cell lines) were seeded with 200 μl of medium in 96-well plates. we dissolved the test sample in dimethyl sulfoxide (DMSO) to final concentrations of 0.78, 1.56, 3.13, 6.25, 12.5, 25, 50, and 100 μM in each well in assays against the MDA-MB-231, NCI-H460 and MCF7 cell lines. The blank control treatment used an equivalent volume of DMSO.

To find the gene cluster responsible for serranticin biosynthetic, we first download the whole genome sequence of Serratia plymuthica V4 (Accession number NZ_CP007439.1) and searched for the genome sequence of synthetic compound serratiochelin.

Drug sensitivity tests were performed on seven common clinical antibiotics via the K-B agar diffusion method. Cultures of S. myotis L7-1 (with an OD of approximately 0.5) were inoculated in LB medium in an amount of 100 μl. The sterilized circular filter paper sheets were attached to the LB. Then 6 μl of the antibiotics (1 mg/ml) and negative control (LB medium) were dropped onto the circular filter paper sheet, after incubated at 37°C for 18–24 h. Antibacterial assay was evaluated by measuring the diameter of the zone inhibition.

The symbiotic microbiota is typically beneficial or neutral during the multi-species interactions (Moreau, 2020). To explore the role of symbionts, we designed and performed two assays to identify the composition of the fish symbiotic microbiota (Figure 1A). The bacteria enrichment samples of Antarctic fish Trematomus bernacchii were harvested. Symbiotic bacteria of T. bernacchii were isolated using the culture method (TB-I). A total of 47 distinct bacterial isolates were selected for 16S rDNA sequence identification (Figure 1B). Phylogenetic analysis revealed that a few taxa were predominated among these isolated strains, and the dominant bacterial species of T. bernacchii sample belonged to Serratia. Whether it was a nutrient-rich medium or a low-nutrient medium, Serratia was the abundant species (36.2%) and maintained growth under different nutrient conditions (Supplementary Figure S1A; Supplementary Tables S1, S2).

We also extracted DNA from T. bernacchii abdominal contents and deep sequencing of 16S rRNA (TB-S) revealed a total of 11 bacterial species (Supplementary Figure S1B). There were large differences in the bacterial compositions displayed by the two methods. Despite the differences in bacteria composition, the two methods shared five bacterial genera, Serratia, Vagococcus, Carnobacterium, Psychrobacter, and Brochothrix (Figure 1C). These 3 unique bacteria genera in TB-I included Pseudomonas, Micrococcus, and Bacillus. These 6 unique bacteria genera in TB-S included norank-f-Bacillaceae, Photobacterium, Corynebacterium, Jeotgalicoccus, Aerococcus, and Paracoccus (Figure 1C). Notably, the relative abundances of genus Serratia varied considerably among the shared bacterial symbionts in the two compositions. Serratia (0.02%) was the least abundant bacteria in the 16S rRNA sequencing (TB-S). Whilst Serratia (43.6%) was the most abundant bacteria in the culture method (Figure 1C). Moreover, results of the different culture plates showed that Serratia species had the highest proportion among all the isolated strains from the various media. The growth assay determined that Serratia kept the highest growth viability in minimal nutrient media (MN; Supplementary Figure S2). The obtained results led us to assume that Serratia sp. was highly adaptive, and able to tolerate a range of nutrient conditions.

A total of 17 strains of symbiotic bacteria were identified as Serratia myotis based on 16S rDNA sequence (Supplementary Table S2), with similarity between 99.50 to 99.86%. To determine the relationship between the 17 strains of S. myotis, we identified the homologous sequence between them using DNA star, with interspecific similarity between 99.40 to 100% (Supplementary Table S2). Based on the high similarity of sequences, S. myotis L7-1 was selected as a representative randomly for further studies (Figure 2A). The following description of S. myotis L7-1 was based on data reported previously. S. myotis is gram-stain-negative, facultative anaerobic, non-spore-forming, rod-shaped cells, 1–1.3 mm in length and 0.5–0.6 mm in diameter, and motile by subpolar flagella (García-Fraile et al., 2015). In our cultivation, S. myotis L7-1 colony morphology on LB agar medium was smooth, circular, convex and transparent, with a diameter of 1.5 to 2.0 mm after 48 h of growth at 20°C (Figure 2B). Optimal growth was observed at 25–31°C (Figure 2C; Supplementary Figure S3A), pH 6–9 (Figure 2D; Supplementary Figure S3B) and with 0–3% NaCl (Figure 2E; Supplementary Figure S3C). Type strain is 12 T (=CECT 8594^T = DSM 28726^T). 16S rDNA sequence accession number (GenBank) is KJ739884 (García-Fraile et al., 2015).

Chemical agents in the living environment of symbiotic microbiota exhibit a broad range of biological activities (Lu et al., 2016). Our next question was whether the natural products of Serratia sp. were of biological meaning as the antibiotics. To confirm their bioactivities, crude extracts from the isolates were tested for antimicrobial activity toward six bacterial species with the disk diffusion method. We found that the crude extract of S. myotis could consistently inhibit the growth of pathogenic bacteria (Figure 2F). Whereas, these of other isolated bacterial species from the same sample showed no significant antibacterial activity (Supplementary Figure S4). Next, we further investigated the effectiveness of S. myotis L7-1. The crude extract of S. myotis exhibited antibacterial activity against Enterococcus faecalis, Mycobacterium smegmatis, Salmonella choleraesuis, Chromobacterium violaceum, Staphylococcus aureus and Escherichia coli at 10 mg/ml, with inhibition rate of 71.8, 64.6, 66.9, 64.7, 63.0, 55.9% (Figure 2G), respectively. It seems that, S. myotis L7-1 could be a natural reservoir of bioactive secondary metabolites.

The whole genomic annotation was performed to decipher the functional genes of S. myotis in the Antarctic fish (Figure 3A). The sequences obtained from S. myotis L7-1 were assembled into one chromosome and four plasmids of size 5,503,952 bp (Chr 1; Figure 3B), 110,419 bp (pA; Supplementary Figure S5A), 49,145 bp (pB; Supplementary Figure S5B), 38,106 bp (pC; Supplementary Figure S5C), 3,223 bp (pD; Supplementary Figure S5D), with a GC content of 55.06%, and 5,272 genes with coding sequences. A total of 85 tRNA, 22 rRNA and 120 sRNA were detected. A total of 5,357 genes were predicted to display COG functional annotation, including 2,096 transport and metabolism genes, 2,550 cellular process and signal transduction genes, 526 information storage and processing genes, and 185 functionally unknown genes.

Functional classification of the microbial genes revealed the prevalence of genes for the metabolic system, including the putative biosynthetic gene clusters (BGCs) and carbohydrate-active enzymes (CAZymes). We identified a total of 285 BGCs associated genes and assigned secondary metabolic pathways using antiSMASH (Kai et al., 2017; Figure 3C). The results revealed the presence of 10 gene clusters encoding biosynthesis of secondary metabolites (SMs), including three non-ribosomal peptide-synthetase (NRPS; 37.5%), a hserlactone (5.3%), a NRPS-like (12.3%), a betalactone (7%), an arylpolyene, siderophore (15.8%), a thiopeptide (7%), a Ribosomally synthesized and post-translationally modified peptides-like (RiPP-like; 4.9%), and other (10.2%) gene clusters (Supplementary Figure S6). The percentage values mean the proportion of the related genes in the total BGCs genes. The abundant biosynthetic genes and related clusters indicated the important roles of versatile SMs production in S. myotis L7-1, as an untapped resource of undiscovered molecules. CAZymes analysis (Cantarel et al., 2009) identified 122 proteins belonging to the families of structurally-related catalytic and carbohydrate-binding modules (or functional domains) of enzymes that degrade, modify or create glycosidic bonds. Glycoside hydrolase (GH) family (47.5%) and glycosyl transferases (GTs; 35.2%) were the most abundant proteins in the CAZymes (Figure 3D).

Next, we showed S. myotis L7-1 harbored 85 genes associated with the “stress response” and linked to adaptation to extreme environments. Among these genes, universal stress genes (9.4%), multiple stress resistance genes (5.9%), dioxygenase genes (49.4%) and genes related to “oxidative stress” (9.4%), might assist the strain to cope with oxidative stress in seawater caused by the Antarctic environment. Cold shock protein genes (14.1%) and heat shock protein genes (11.8%) would facilitate the strain to survive in Antarctic temperatures (Figure 3E). Another interesting finding related with environmental adaptation was that several antibiotic-resistance genes (ARGs) and multiple antibiotic resistance mechanisms were predicted in the genome of S. myotis L7-1, by exploring the pan-genome analysis from the CARD database (Jia et al., 2016). The majority of ARGs contribute to encode the cephalosporin (5.9%), cephamycin (4.7%), carbapenem (4.3%), peptide antibiotic (4.1%), aminoglycoside antibiotic (3.4%), tetracycline antibiotic (12%), macrolide antibiotic (10.3%), fluoroquinolone antibiotic (10.1%), penam (7.1%) and others (38%; Figure 3F; details in Supplementary Table S3). Furthermore, antibiotic resistance mechanism included the reduced permeability to antibiotic (1.2%), antibiotic target replacement (0.9%), antibiotic efflux (70.9%), antibiotic target alteration (10%), antibiotic inactivation (7.1%), antibiotic target protection (6.8%), and antibiotic efflux, antibiotic target alteration (3.2%; Figure 3G). These results suggest that S. myotis had the ability to resist several antibiotics, and the ARGs possibly facilitated the competitiveness and colonization of bacteria in the extreme Antarctic environments.

Serratia sp. are ubiquitous to different environments and show highly competitive fitness in versatile niches (Hover et al., 2016). The available genomic data of symbiotic fish bacteria is limited. To decipher the highly adaptive features of S. myotis, we next compared S. myotis L7-1 genome with four closely related bacterial genomes, including Serratia quinivorans 13,188 isolated from soil (Ashelford et al., 2003), Serratia grimesii BXF1 isolated from the pinewood nematodes (Nascimento et al., 2018), Serratia liquefaciens ATCC 27592 isolated from the spacecraft assembly facilities (Schuerger et al., 2013), and Serratia proteamaculans 336x isolated from the rinds of cheeses (Zhang et al., 2018). A total of 118 BGCs genes and 109 CAZyme genes were common to the five Serratia genomes (Supplementary Figure S5E). The unique gene clusters associated with SMs production in S. myotis L7-1 included betalactone, RiPP-like, arylpolyene, and siderophore. The unique CAZymes included glycosyl transferases and polysaccharide lyases (Supplementary Figure S5F). A total of 2,479 KEGG genes and 2,087 COG genes were common to the five Serratia genomes (Figures 4A,B). These unique genes in S. myotis L7-1 include those encoded for trafficking proteins, membrane fusion proteins, intracellular multiplication proteins, transport proteins, multiple enzymes and the others (Supplementary Tables S4, S5). Moreover, there was one notable feature that attracted our attention, the complete type IV secretion system with the particular plasmids.

L7-1 contained one chromosome and four plasmids, while other four reference bacterial genomes harbored one chromosome and no plasmids (Supplementary Figure S5). S. myotis L7-1 and all of the reference Serratia strains contained multiple secretion system genes, but only S. myotis L7-1 possessed a complete type IV secretion system (T4SS) located on the plasmids (Figure 4C). Many bacterial species deploy T4SSs to inject toxic effectors into target bacteria, thus inducing the death of rival cells (Rambow-Larsen and Weiss, 2004; Kutter et al., 2008; Juhas et al., 2010; Arslan-Aydogdu and Kimiran, 2018; Sgro et al., 2019). The type IV secretion system in S. myotis L7-1 included mainly type IV secretion system protein genes (virB1, 2, 4, 5, 6, 8, 9, 10, 11 and virD4) and a series of intracellular multiplication protein genes (icmB, E, G, J, K, L, O, P, T; Supplementary Table S4).

The versatile biosynthetic gene clusters and preliminary antimicrobial activity of the crude extract showed the great bioactive potential in S. myotis L7-1. To further determine the competitive roles of natural products, the bioactivity-guided fractionation assay against various pathogenic bacteria was conducted (Figure 5A). In total, 16.7 g of crude extract was obtained from a large-scale culture in rice solid medium. Bioactivity-guided purification and isolation strategy led to the isolation of bioactive fraction, compound 1 (61 mg; Figure 5B). It is proved that compound 1 displayed significant anti-bacteria activity as a leading product, compared with that of the other compounds from S. myotis L7-1. The results demonstrated that bioactivity-guided fractionation was an attractive and effective approach for profiling and screening of bioactive compounds.

Compound 1 was isolated as a white powder with a UV absorption at 210 nm (Figure 5B) and a specific optical rotation value of [α]_D²⁰–26.40^o (c 1.05, acetone; Supplementary Table S6). Its molecular formula is C₂₁H₂₃N₃O₇ (429.15). Its ¹H NMR spectrum (600 MHz, in DMSO-d₆, Supplementary Figure S7A) showed the following signals: δ_H 2.23 (3H, d, J = 6.4 Hz, H-12), 3.28 (2H, m, H-4′), 3.97 (2H, m, H-5′), 4.08 (2H, m, H-3′), 5.24 (1H, d, J = 7.4 Hz, H-9), 5.64 (1H, m, H-10), 7.45 (1H, t, J = 8.0 Hz, H-5″), 7.52 (1H, t, J = 7.9 Hz, H-5), 7.68 (1H, dd, J = 7.9, 1.4 Hz, H-4), 7.75 (1H, dd, J = 7.9, 1.4 Hz, H-4″), 7.85 (1H, dd, J = 7.9, 1.4 Hz, H-6), 8.04 (1H, dd, J = 8.0, 1.4 Hz, H-6″), 9.09 (1H, t, J = 6.0, NH-2′), 9.58 (1H, t, J = 5.9 Hz, NH-6′), 10.00 (2H, s, OH-3, 3′), 12.52 (s, OH-2) and 13.54 (s, OH-2′). The ¹³C NMR spectrum (150 MHz, in DMSO-d₆, Supplementary Figure S7B) showed 21 signals: δ_C 20.83 (q, C-12), 29.08 (t, C-4′), 36.74 (t, C-5′), 36.85 (t, C-3′), 73.88 (d, C-9), 78.97 (d, C-10), 110.49 (s, C-1), 115.09 (s, C-1″), 117.21 (d, C-6″), 118.01 (d, C-4), 118.08 (d, C-5″), 118.82 (d, C-5), 118.89 (d, C-4″), 119.58 (d, C-6), 145.91 (s, C-3), 146.42 (s, C-3″), 148.45 (s, C-2), 149.92 (s, C-2″), 165.85 (s, C-7), 169.75 (s, C-1′), and 169.94 (s, C-7′). Based on its ¹H and ¹³C NMR data, HRESIMS data, specific optical rotation value, and comparison with previous literature data (Kuo et al., 2011, 2012), compound 1 was proven to be identical to serranticin. The ¹H and ¹³C NMR data of serranticin (1) were also summarized in Supplementary Table S7.

To further determine the bioactivity of serranticin in pharmaceutical potential, human pathogenic bacteria and tumor cells were chosen to conduct the antibiotic assay. Serranticin was tested against several bacterial strains, including C. violaceum, S. choleraesuis, M. smegmatis, E. coli, E. faecalis, and S. aureus. Gentamicin (an antibiotic against both Gram-positive and negative bacteria) was used as the positive control. The results showed that serranticin exhibited strong antibacterial activity against both C. violaceum and S. choleraesuis at 50 μg/ml, with inhibition rate of 89.5 and 72.8%, respectively (Figure 5C). In addition, serranticin also displayed moderate activity toward M. smegmatis, E. coli, E. faecalis, and S. aureus with inhibition rate of 41.2%, 55.45%, 57.9% and 52.6%, respectively. The antimicrobial activity was studied further to evaluate the minimum inhibitory concentration (MIC) against C. violaceum and S. choleraesuis by broth microdilution method. Serranticin showed excellent antimicrobial activities against C. violaceum and S. choleraesuis with MIC values of 6.25 and 12.50 μg/ml, while the MIC value of gentamicin were 6.25 and 6.25 μg/ml (Figure 5D). Serranticin approximately rivaled the antimicrobial activity of the positive control gentamicin.

Antitumor activities of serranticin were tested by the MTT method using the human breast carcinoma MDA-MB-231, MCF7 and human lung cancer NCI-H460 cell lines. Serranticin inhibited the growth of three tested cell lines significantly and the cell viability values at 100 μM were given below, 48.5% (MDA-MB-231), 33.4% (NCI-H460), and 47.2% (MCF7; Figure 5E), respectively. The half-inhibitory concentration (IC50) of serranticin on MDA-MB-231, NCI-H460 and MCF7 cell lines were determined to be 2.8, 3.8 and 12.1 μM (Figure 5F), respectively. The data indicated that serranticin showed strong inhibitory activity against human cancer cell lines, suggesting that it can be a potential candidate for pharmaceutical application.

It seems that S. myotis L7-1 harbored the ARGs to resist several antibiotics. To further determine the multiple resistance of S. myotis L7-1, the drug sensitivity test was conducted. Using the Kirby-Bauerdisca (K-B) agar diffusion method (Figure 6A), we tested the drug susceptibility with common clinical antibiotics at 1 mg/ml, including ampicillin, spectinomycin, vancomycin, kanamycin, tetracyclines, gentamycin and amphotericin. LB medium was used as negative control. The results indicated that the activities of ampicillin, spectinomycin, amphotericin and vancomycin were equivalent to the negative control for S. myotis L7-1 in terms of inhibition, with no significant zone of inhibition shown (Figure 6B). The diameters of the inhibition circle were 13.3, 11.3 and 13.7 cm for kanamycin, tetracyclines and gentamycin. From these results, it is clear that S. myotis can tolerate ampicillin, spectinomycin, amphotericin and vancomycin, while it is sensitive to kanamycin, tetracyclines and gentamycin.

The BGCs analyses of S. myotis reveal the potential production of bioactive compounds. The preliminary antagonism assays indicated that S. myotis L7-1 inhibits the growth of bacteria pathogens (Figure 2F). Next, the bioactivity-guided fractionation against aquatic pathogen was conduct, presenting the excellent active agent, serranticin. The antimicrobial activity of serranticin was determined in vitro for against Edwardsiella tarda and Aeromonas hydrophila, the serious bacteria pathogens of fish. The results showed that serranticin was active against E. tarda and A. hydrophila at 50 μg/ml, with inhibition rate of 83.3 and 70.5% (Figure 6C). The MIC values of the serranticin were determined by the broth microdilution method. The results indicated that serranticin had strong anti-aquatic pathogenic bacteria activity in inhibiting the growth of E. tarda (MIC: 6.25 μg/ml), and A. hydrophila (MIC: 6.25 μg/ml; Figure 6D). Positive control gentamicin had anti-bacteria activity with an MIC value of 12.5 μg/ml against E. tarda and 12.5 μg/ml against A. hydrophila. Serranticin displayed even stronger antimicrobial activity than positive control gentamicin, that contributes to the development of probiotics and potential utility in fishery industry.

Unless specifically noted, each experiment was repeated three or more times independently. Data were collected from three biological and three technical replicates, unless otherwise noted. Data shown in column graphs or line chart represent mean ± the standard deviation (SD), as indicated in the figure legends. Statistical analysis was performed with Origin 2022.¹ Graph preparation was performed,² itol.embl.de and ChemDraw 19.0. More details were given in the figure legends and methods.