The cultivation and plant protection laboratory of the Tea Research Institute of the Hunan Academy of Agricultural Sciences isolated Br. brevis HNCS-1 from tea garden soil and deposited it in China Center for Type Culture Collection (No. M 2022713). The cultivation and plant protection laboratory isolated and preserved Gloeosporium theae-sinensis, Elsinoe leucospira, Phyllosticta theaefolia, Fusarium sp., and Cercospora theae.

The fungi were routinely cultivated at 26°C on a PDA medium (6 g potato extract, 20 g dextrose, 20 g agar, 1,000 ml H₂O, pH 5.6 ± 0.2) and stored at 4°C on the same medium. Before using them in an experiment, they were activated on a PDA medium for 24 h at 26°C and then transferred by streaking. The HNCS-1 strain was stored in 20% glycerol at −80°C. It was activated by streaking at 37°C for 12 h on LB agar medium (10 g tryptone, 5 g yeast extract, 10 g NaCl, 15 g agar, 1,000 ml H₂O, pH 7.0 ± 0.1).

Br. brevis HNCS-1 was inoculated into 500 ml Erlenmeyer flasks containing 200 ml of the LB liquid medium and cultivated at 37°C and 200 rpm for 12 h. Then, 10 ml of the seed cultures were transferred to 500 ml Erlenmeyer flasks containing the NB medium (10 g peptone, 3 g beef extract, 5g NaCl, 1,000 ml H₂O, and pH 7.2 ± 0.2) and cultivated at 37°C and 180 rpm for 3 days. Br. brevis HNCS-1 fermentation broth was centrifuged at 4°C and 10,000 rpm for 10 min, and the supernatant was collected and filtered through a 0.22 μm filter.

On a PDA medium containing a 10% cell-free supernatant of Br. brevis HNCS-1, spore germination was used to determine the bacteriostatic activity against fungi. A fungus cake was made using a 5 mm diameter sterile cork borer and placed in the center of the plate. Plates of P. theaefolia, Fusarium sp., E. leucospira, and C. theae were incubated at 26°C for 12 days, 11 days, 22 days, 22 days, and 7 days, respectively. The experiment was performed thrice. The antagonistic activity of Br. brevis HNCS-1 was evaluated by inhibition rate (IR), which is calculated using the following formula:

where A is the diameter of tea fungus disks in the control treatment, B is the diameter of tea fungus disks with Br. brevis HNCS-1, and 5 is the diameter of the disks inoculated with tea fungus.

The entire genome of Br. brevis HNCS-1 was sequenced using the Illumina and Nanopore platforms. Guangdong Magigene Technology Co., Ltd. performed this study in China. Following the standard protocol provided by Oxford Nanopore Technologies (ONT), the sequencing procedure includes sample quality detection, library construction, library quality detection, and library sequencing. SMRT Link v5.1.0 was chosen for the assembly of third-generation data alone, and Unicycle was chosen for the assembly of second-generation and third-generation data. tRNAscan-SE v1.3.1, rRNAmmer v1.2, and Rfam databases briefly predicted Glimmer, tRNAs, and rRNAs, respectively. PHAST identified prophage sequences within the genome assemblies. Island viewer predicted Genomic Islands (GIs).

The complete nucleotide sequence was also searched against Non-Redundant (NR) Protein Database, Swiss-Port, Cluster of Orthologous Groups (COGs), Kyoto Encyclopedia of Genes and Genomes (KEGG), Gene Ontology (GO), Pfam, Carbohydrate-Active enZYmes (CAZy) database, Pathogen–Host Interactions (PHIs), Virulence Factor Database (VFDB), and Comprehensive Antibiotic Resistance Database (CARD) for functional annotation and further function assignment. Basic Local Alignment Search Tool (BLAST) and Diamond were used to compare the gene sequence with the reference database sequences. The finest matching results were selected based on identity, e-value, and score, which were the results of genome annotation. Circular maps of the genome were generated using Circos (Krzywinski et al., 2009).

Br. brevis HNCS-1 and 17 other Br. brevis genomes were used to construct a phylogenetic tree, with B. subtilis ATCC 13952 serving as an out-group. Except for Br. brevis HNCS-1, all sequences were obtained from the National Center for Biotechnology Information (NCBI) (Table 1). OrthoFinder software (Emms and Kelly, 2019) was used to generate a maximum likelihood (ML) phylogenetic tree, which was visualized using Figtree v1.4.4. To evaluate the phylogeny of the Br. brevis strains, average nucleotide identity (ANI) was calculated using the FastANI software (Jain et al., 2018). The thermal map was then visualized using the TBtools software (Chen et al., 2020).

To identify the core and strain-specific genes, a pan-genome analysis was performed on 18 Br. brevis isolates using the BPGA v1.3 software, which implemented functional ortholog clustering using the amino acid sequences based on Gene Family (GF) approach (Chaudhari et al., 2016). The collection of genes shared by all 18 strains was referred to as the pan-genome, while the set of shared genes within the test genomes was referred to as their core genomes. For pan-genome genetic contexts, Heap's Law can be represented by the following formula:

where y represents the extent of the pan-genome, n represents the number of genomes, and k and γ are fitting parameters. According to this law, γ can be calculated as α = 1–γ, so when α < 1 (0 < γ < 1), the extent of the pan-genome increases unboundedly with the successive addition of new genomes and is considered open. In contrast, if α > 1 (γ < 0), the pan-genome trajectory approaches a plateau and can be considered closed as additional genomes are added (De Jesus et al., 2022).

The NR database was chosen as the protein sequence database for our annotation system based on a balance of quality, comprehensiveness, and our practical needs. Using the complete nucleotide sequence of the Br. brevis HNCS-1 genome, the online tools antiSMASH 6.0 (Blin et al., 2021) were able to identify clusters of secondary metabolite genes.

One thousand milliliters of Br. brevis HNCS-1 supernatant was frozen at −80°C for 24 h and then concentrated to 20 ml by vacuum freeze drying. The concentrated sample (2 ml) was separated by column chromatography on Sephadex G-75 using an H₂O-CH₃OH gradient [H₂O, H₂O-CH₃OH = 4:1 (v/v), H₂O-CH₃OH = 3:2 (v/v), H₂O-CH₃OH = 2:3 (v/v), H₂O-CH₃OH = 1:4 (v/v), CH₃OH]. The eluent was collected using an EP tube, ranging from 0.7 ml to 1.0 ml per tube. The cell-free eluate's antibacterial activity against G. thea-sinensis was determined using agar diffusion on a PDA medium after it was filtered through a 0.22 μm filter membrane. In total, 1 ml of G. thea-sinensis spore suspension was distributed across the agar surface of the plates. Then, a sterilized cork borer was used to create 5 mm diameter wells with equal spacing in the agar. Overall, 3 days were spent incubating the dishes at 26°C. The experiment was performed thrice.

For the separation, ultra-high-performance liquid chromatography (Exion, SCIEX) was utilized. The process utilized an Agilent ZORBAX SB-C18 column (2.1 mm × 150 mm, 2.7 μm) with acetonitrile (A) and 0.1% formic acid in water (B) as gradient elution mobile phases at a flow rate of 0.3 ml/min. The gradient elution protocol was as follows: 5% B from 0–5 min, 5–50% B from 5–20 min, 50–100% B from 20–30 min, and 100% B from 30–35 min. The temperature of the column was set to 35°C.

The X500R Q-TOF mass spectrometer (SCIEX, Framingham, MA, USA) was used to acquire untargeted mass spectral data. The LC effluent was injected into the mass spectrometer. ESI parameters were as follows: temperature: 500°C; ion source gas 1 and 2: 50 psi; curtain gas: 35 psi; CAD gas: 7 psi. Collision-induced dissociation at 35 ± 15 V in IDA mode was utilized to capture the MS and MS/MS spectra. Maximum candidate ions: 10; threshold for intensity: 600 cps; full scan mass range: 500–1,000 Da; ion discharge voltage: 5,500 V; collision energy: 35 V; collision energy spread: 15 V.

Br. brevis HNCS-1 was evaluated for its antifungal activity against pathogenic fungi. The results demonstrated that HNCS-1 inhibited the mycelia proliferation of five tea fungal diseases significantly (Figure 1). Br. brevis HNCS-1 was isolated from tea garden soil and demonstrated broad-spectrum antagonistic activity. The mycelial growth of Phyllosticta theaefolia, Fusarium sp., Cercospora theae, and Gloeosporium theae-sinensis was completely inhibited by 10% Br. brevis HNCS-1 supernatant. Additionally, it significantly inhibits the mycelial proliferation of Elsinoe leucospira. Br. brevis HNCS-1 is a beneficial microbe with biotechnological application potential.

The genome of the HNCS-1 strain consists of a single circular chromosome measuring 6,353,630 bp with a GC content of 47.15%; no plasmids were detected (GenBank CP128411). This chromosome genome consisted of 6,342 coding DNA sequences (CDS), which accounted for 87.50% of the genome. In addition, 15 sRNA, 44 rRNA, and 127 tRNA were predicted based on the chromosome sequence. In the genome, 31 Genomic Islands were identified, but no CRISPR repeat regions or prophages were found (Supplementary Figure 1).

In total, 5,796 and 4,080 identified genes were annotated as NR and Swiss-port, respectively, and 2,916 and 4,714 genes were classified into functional categories based on GO and COG designations, respectively (Supplementary Figures 2, 3). Overall, 5,810 KEGG pathways have been assigned (Supplementary Figure 4). Moreover, 4,808, 2,906, 1,190, 173, and 6 genes, respectively, were annotated in Pfm, PHI, VFDB, CAZyme, and CARD (Supplementary Figures 5, 6).

The phylogenetic tree of 18 Br. brevis genomes and B. subtilis ATCC 13952 was constructed using the ML method and all orthogroups, with the ATCC 13952 strain serving as the root. It was found that HNCS-1, Br. brevis strains 17, and ATCC 13952 formed a large cluster, and different strains of Br. brevis revealed two groups (Figure 2A). In the meantime, strain HNCS-1 was a member of the same clade as strains G25-137, Ag35, ATCC 35690, GZDF3.1, NBRC100599, and LABIM17, but it was not a sister group to any other strains.

To substantiate the results of the phylogenetic analysis, we also determined the ANI values of various strains. Strains with ANI values above 95% are regarded as belonging to the same species (Richter and Rossello-Mora, 2009). The results showed that the genetic relationship between ANI and phylogenetic trees was basically consistent, but ANI was more precise, dividing 18 strains of Br. brevis into 5 species (Figure 2B).

According to the results, they were not distinguished by the mixed trend of geographical origin and phylogenetic clustering of strains isolated from similar environments. The genome's phylogenetic clustering differed substantially from the strains' specific habitat classification.

The BPGA software determined a pan-genome for the strain HNCS-1 and 17 sequenced Br. brevis strains by comparing pan-genome analyses of bacterial species by utilizing protein clustering data in the total pan-genome. There were 10,359 CDS in the total pan-genome of the 18 Br. brevis strains. Among the 10,359 protein-coding genes, 3,742 core genes represented 36.12% of the genes in the pan-genome of Br. brevis. The number of accessory gene families (3,961 genes) was greater than the number of core gene families (3,742 genes). Moreover, Br. brevis HNCS-1 contains 1,586 accessory genes. In addition, the strain HNCS-1 also contained the greatest number of distinct transcripts (560 genes) and has greater potential for gene exploration. Br. brevis strains NCTC2611 and DSM30 encoded the fewest number of specific strain genes with 4 and 6, respectively (Figure 3A).

The number of Br. brevis genomes was plotted against the size of the pan-genome and the core genome. The pan-genome curve exhibited an asymptotic trend, indicating that 18 genomes were inadequate to characterize the complete gene repertoire of Br. brevis. According to the curve generated for these 18 genomes based on Heap's Law and least-square fit of exponential regression decay, the number of gene families in the pan-genome increases with the addition of each additional genome (γ = 0.22), indicating that the pan-genome of Br. brevis strains is currently open but may be closed soon (Figure 3B).

It is noteworthy that the cell-free supernatant fermentation of Br. brevis inhibited numerous bacterial and fungal diseases. Additionally, the core genome sequence of Br. brevis was mined for AMP-encoding genes. Only one gene cluster implicated in edeine NRPS was identified in the core genome after NR annotation. The gene cluster likely contributes significantly to antimicrobial activity. The sequence similarity between the genes and the predicted ede BGC of Br. brevis Vm4 was high (Westman et al., 2013). The identified ede BGC in the Br. brevis HNCS-1 genome was structured identically to the Br. brevis Vm4 genome. In brief, the ede BGC (43.57 kb) in Br. brevis HNCS-1 contained 17 open reading frames, designated edeA through edeQ based on the homologous sequences in Br. brevis Vm4 (44.12 kb) (Figure 4A).

Additionally, the genome sequence of Br. brevis HNCS-1 was mined by antiSMASH software for the presence of gene-encoding AMPs. A total of 13 putative biosynthetic gene clusters were identified by the HNCS-1 strain genome (Supplementary Figure 7). Four putative gene clusters shared a high degree of similarity (>70% of genes shared similarity) with the petrobactin, tyrocidine, bacillopaline, and gramicidin gene clusters. Three putative gene clusters displayed modest similarity (<30% of genes displayed similarity) to the previously reported zwittermicin A, aurantinin B-D, and pacidamycin (1–7, D) gene clusters. Six putative gene clusters were not conserved in comparison with known clusters.

The antimicrobial peptide should be edeine, according to the results of genome annotation analysis. Edeines are a class of linear pentapeptides generated by Brevibacillus, a soil bacterium. Edeines A, B, D, and F exist as two isomers of α and β; however, only the α isomer possesses remarkable antibiotic properties (Figure 4B).

The antibacterial potential of eluent extracts containing antibacterial compounds was evaluated. Using an agar well diffusion assay, the antibacterial potential of eluent extracts from distinct collection tubes against G. theae-sinensis was determined. With an increase in eluent volume, the bacteriostatic activity initially increased steadily and then decreased dramatically (Figure 5A). The sample with the highest antibacterial activity was analyzed by UPLC-MS/MS for additional confirmation. According to the total ion chromatogram, the majority of the ultraviolet absorption is of short wavelength.

The theoretical molar masses of edeine A (C₃₃H₅₈N₁₀O₁₀), edeine B (C₃₄H₆₀N₁₂O₁₀), edeine D (C₃₃H₅₈N₁₀O₉), and edeine F (C₃₄H₆₀N₁₂O₉) are 754.4337, 796.4555, 738.4388, and 780.4606. Compound 1 (m/z 755), Compound 2 (m/z 797), Compound 3 (m/z 739), and Compound 4 (m/z 781) were extracted from the total ion current based on the ion mass spectral (m/z). Compound 1 (m/z 755) produced a [M+H]⁺ ion peak m/z 755.4132 (−1.7 ppm, C₃₃H₅₈N₁₀O₁₀) at a retention time of 1.182 min (Figure 5B-a). The electrospray ionization-mass spectrometry results indicated that its fragmentation pattern was identical to that of edeine A, including m/z 738, 737, 720, 667, 553, 536, 401, 390, and 203 (Figure 5C-a, Supplementary Figure 8). Compound 2 (m/z 797) produced a [M+H]⁺ ion peak m/z 797.4612 (−2.6 ppm, C₃₄H₆₀N₁₂O₁₀) at a retention time of 1.165 min (Figure 5B-b) with observable signals. The electrospray ionization-mass spectrometry results indicated that its fragmentation pattern was identical to that of edeine B, including m/z 780, 779, 761, 443, and 245 (Figure 5C-b, Supplementary Figure 9). Compound 3 (m/z 739) had faint signals at the retention time of 1.165 min (Figure 5B-c), but its peak value was too low to be analyzed further based on the available mass spectrum data. Compound 4 (m/z 781) produced a [M+H]⁺ ion peak m/z 781.4667 (−2.2 ppm, C₃₄H₆₀N₁₂O₉) at the retention time of 1.708 min (Figure 5B-d). The electrospray ionization-mass spectrometry results indicated that its fragmentation pattern was identical to that of edeine B, including m/z 780, 779, 761, 443, and 245 (Figure 5C-c, Supplementary Figure 10). To sum up, edeines were a class of highly polar AMPs with limited retention on reverse phase chromatography, which was consistent with the chromatographic retention property of the compounds discussed previously. Compounds 1, 2, and 4 were edeine A, B, and F, respectively. The chromatographic peak area suggested that edeine B had the maximum concentration, followed by edeine B and F had the lowest concentration.

In addition, to confirm the existence of secondary metabolites predicted by antiSMASH software, we extracted relevant data from the total ion current based on the theoretical molecular weight of petrobactin, tyrocidine, bacillopaline, and gramicidin, but they were not detected (Supplementary Figures 11–14).