The tested pathogens were obtained from the Research Center for Engineering Technology of Kiwifruit at Guizhou University, including B. dothidea (OM802621), Phomopsis sp. (OP740383), Rhizoctonia solani (OK287282), Monilinia fructicola (MW281764), Fusarium fujikuroi (OL774567), A. alternata (MG189596.1), Colletotrichum spaethianum (OL347722), Pseudomonas syringae pv. actinidiae (NZ_CP032631.1), and Pantoea agglomerans (ON955285.1). The pathogenic fungi were cultured in potato dextrose agar (PDA: 250 g potato, 20 g glucose, and 15 g agar/L) for 5 days at 28°C in the dark. The pathogenic bacteria were cultured in beef extract peptone agar (NA: 5 g beef extract, 10 g peptone, 5 g NaCl, and 15 g agar/L) for 2 days at 28°C in the dark.

The kiwifruit were picked in an orchard in Xiuwen County, Guizhou Province (26°45′N, 106°65′E) in October 2021, packed in sterile sampling bags, and stored at 4°C. The kiwifruit surface was washed with sterile water to remove impurities, followed by soaking in 75% ethanol for 60 s, and subsequent washed with sterile water to remove the residual alcohol. Whole kiwifruit were crushed and 1 ml of crushed sample was diluted 1,000 times, spread on NA plates, and incubated at 28°C for 48 h. Subsequently, single colonies with uniform colour, glossiness, and other characteristics were selected and re-inoculated on new NA plates for 48 h. The dual-culture assay for assessing growth inhibition of pathogenic fungi was performed as described by Yan et al. (2020). Preliminarily, antagonistic bacteria that inhibited the growth of pathogenic fungi were screened out. The with obvious inhibition zones were selected for further study.

Antagonistic bacterial strains were preliminarily identified by referring to Buchanan and Gibbons (1984) and Dong and Cai (2001). Single bacterial colonies were picked using an inoculating needle and cultured in NA for 5 days at 28°C under 75% relative humidity. The morphological characteristics of the single colonies were observed with the naked eye and the results of Gram staining were observed using a DM500 binocular microscope (Leica Microsystems, Wetzlar, Germany). Simultaneously, single bacterial colonies were inoculated in NA and cultured at 28°C for 24 h with shaking at 180 rpm. The ultrastructure of the bacterial cells was observed using an SU8010 scanning electron microscope (SEM; Hitachi, Tokyo, Japan) with reference to the method previously reported by Hasan et al. (2014).

To further identify the taxonomy of the antagonistic bacterial strains, phylogenetic trees of the 16S ribosomal RNA, DNA gyrase subunit A (gyrA), RNA polymerase subunit B (rpoB) and phosphoribosylaminoimidazolecarboxamide formyltransferase (purH) genes were constructed using multilocus gene sequence analysis (Rooney et al., 2009; Kubo et al., 2011; Dhruw et al., 2020). DNA was extracted from bacterial cells using the TIANamp Bacteria DNA Kit (Tiangen Biotech Co., Ltd., Beijing, China) according to the manufacturer’s instructions. The PCR reaction was performed in a total volume of 20 μL containing 1 μL of 100 ng/μL DNA, 0.4 μL each of 10 μmol forward and reverse primers (Table 1), 10 μL of 2 × Taq Master Mix, and 8.2 μL ddH₂O. The thermal cycling program included a denaturation step at 94°C for 30 s, an annealing step at 55 or 60°C for 30 s, and a final extension at 72°C for 10 min for 35 cycles. The obtained PCR products were separated on 1.0% w/v agarose gel at 150 V for 20 min to confirm the DNA amplification of a single product. The PCR products were sequenced by Sangon Biotech Co., Ltd., (Shanghai, China). BLAST was used to align the sequences with the National Center for Biotechnology Information (NCBI)¹ nucleotide sequence database and the reference sequences were downloaded. Construction of multi-locus phylogenetic trees was performed using maximum likelihood, maximum parsimony, and Bayesian inference methods (Chen et al., 2022a). The sequencing data for 16S rRNA, gyrA, rpoB, and purH are deposited under GenBank accession numbers ON076886, ON086734, ON086735, and ON086736, respectively.

The in vitro inhibitory activity of the antagonistic bacterial strains against pathogenic bacteria or fungi was assessed with reference to the methods described by Yan et al. (2020) and Chen et al. (2022b). Briefly, a pathogen colony with a diameter of 6 mm was placed in the centre of a PDA plate. The bacterial strains were cultured in 50 ml of NA at 28°C for 48 h with shaking at 130 rpm. The bacterial suspension was adjusted to an OD₆₀₀ reached 1.5–2 using a NanoPhotometer-N50 instrument (Implen, Munich, Germany). Subsequently, 100 μL aliquots of bacterial suspension were applied to four disc of filter paper (diameter of 6 mm), which were symmetrically placed 25 mm away from the pathogen colony. The control group contained only pathogen colonies without bacterial strains. All treatments were repeated three times and incubated at 28°C for 5 days.

The effects of the VOCs produced by the antagonistic strains were evaluated using the double-plate pair-button method (You et al., 2021). Briefly, 150 μL of bacterial suspension (OD₆₀₀ reached 1.5–2) was spread evenly on an NA plate, the pathogenic fungus plug (diameter = 6 mm) was placed in the centre of a PDA plate, and the two plates were sealed together. The pathogen colony diameter was measured using a ruler and hyphal growth inhibition was calculated as follows: A = (a1–a2)/a1 × 100, where A is the growth inhibition of the hyphae, a1 is the colony diameter of the control pathogen, and a2 is the colony diameter of the treated pathogen. The control group used 150 μL of sterile distilled water on the NA plate instead of 150 μL of bacterial suspension.

Hyphae at the edge of the antifungal zone were selected to evaluate the antagonistic effects of the isolated bacterial strains. Morphological changes were observed using a DM500 binocular microscope with magnification multiples of 400 (Leica) and SU8010 SEM (Hitachi) (Mo et al., 2021).

The genomic DNA of CTXW 7-6-2 was extracted using the Rapid Bacterial Genomic DNA Isolation Kit (Sangon Biotech Co., Ltd.) according to the manufacturer’s instructions. Whole-genome sequencing was performed on the Illumina HiSeq sequencing platform (San Diego, CA, USA) and established using the NEB Next^® Ultra™ DNA Library Prep Kit for Illumina^® (New England Biolabs, Ipswich, MA, USA). Trimmomatic v. 0.36 was used to filter the sequencing data to obtain high-quality data, and FastQC v. 0.11.2 was used to visually assess quality, reads, trimming, and de novo assembly. The next-generation sequencing data were assembled using SPAdes v. 3.5.0. GapFiller v. 1.11 was used to complement the contigs obtained by splicing, and PrInSeS-G v. 1.0.0 was used to correct editing errors and small insertions and deletions during the splicing process. The completed whole-genome sequence was submitted to NCBI under GenBank accession number JAMKCC000000000.1.

Prokka v. 1.10 was used to predict the gene elements of the assembled results, using Prodigal to predict coding genes, Aragorn to predict tRNAs, RNAmmer to predict rRNAs, and Infernal to predict miscRNAs. The Bacterial Pan Genome Analysis tool (BPGA) (Chaudhari et al., 2016) pipeline was used with default parameters, and a phylogenetic tree was constructed for the pan-/core-genome analysis of 25 representative Bacillus strains with sequenced genomes.

National Center for Biotechnology Information Blast+ v. 2.2.28 was used to align the protein sequences with those in the Pathogen-Host Interactions Database (PHI-base),² along with those in the Clusters of Orthologous Groups of proteins (COG), Conserved Domain Database (CDD), NCBI non-redundant protein sequences (nr), and protein family (Pfam) databases. Gene Ontology (GO³) analysis was performed using the protein annotation results from the Swiss-Prot and TrEMBL databases and annotation information from the UniProt database. Pathway enrichment analysis was conducted using the Kyoto Encyclopedia of Genes and Genomes (KEGG) Automatic Annotation Server.⁴ HMMER3 v. 3.1b1 was used to align the gene set protein sequences with the Carbohydrate-Active enZYmes database (CAZy),⁵ thereby obtaining the corresponding carbohydrate-active enzyme annotation information (E-value < 10 ^––5). The Rapid Annotations using Subsystem Technology (RAST)⁶ (Overbeek et al., 2013) approach was used to categorise the gene sets of proteins involved in specific biological processes.

The secondary metabolite synthesis gene clusters of B. subtilis CTXW 7-6-2 were predicted and analysed for potential bacteriostatic ability using antiSMASH v. 4.0 (Blin et al., 2017).⁷ B. subtilis NCIB 3610 (produces red pigment, GenBank: CP020102.1) and B. subtilis subsp. subtilis 168 (does not produce red pigment, Gen Bank: NC_000964.1) were used as control strains for analysis of secondary metabolite synthesis gene clusters.

The enzymes encoded by yvmC [cyclo(L-leucyl-L-leucyl) synthase] and cypX (pulcherriminic acid synthase) are known to be involved in the production of the red pigment pulcherriminic acid by Bacillus spp. (Li et al., 2017). Hence, the amino acid sequences of these two enzymes were selected from the whole gene sequence of the antagonistic bacteria. ProtParam⁸ was used to analyse the physicochemical properties of the proteins. Meanwhile, Predict Protein⁹ and SWISS-MODEL¹⁰ were used to predict the secondary and tertiary protein structures, respectively.

Quantitative real-time PCR (qRT-PCR) was used to determine the relative expression levels of yvmC and cypX in the antagonistic bacteria cultured at 28°C for 1 and 5 days with shaking at 180 rpm, respectively. B. subtilis GUMT323, which cannot produce red pigment, was used as the control strain (provided by Professor Ding Haixia of Guizhou University, CCTCC no. M 2018873). Sangon Biotech Co., Ltd., designed and synthesised the primers for yvmC and cypX, the sequences of which were submitted to their online platform (Table 1).¹¹ The rpoB gene was used as the internal control (Jiang et al., 2021). Total RNA extraction and reverse transcription were performed using the RNAprep Pure Bacteria Kit (Tiangen Biotech) and First Strand cDNA Synthesis Kit (Tiangen Biotech), according to the manufacturer’s instructions. The cDNA was diluted to 200 ng/μL for the qPCR assay with each gene-specific primer. qRT-PCR was performed using Talent qRT-PCR SYBR Green PreMix (Sangon Biotech Co., Ltd.) on the Bio-Rad CFX96 qRT-PCR system (Bio-Rad, Hercules, CA, USA). Four biological replicates were analysed. Relative gene expression was calculated using the 2 ^{–Δ Ct} method.

Statistical analysis was conducted using SPSS Statistics v. 13.0 software (SPSS Inc., Chicago, IL, USA). All treatments were performed using three biological replicates. Data are expressed as the mean ± standard deviation (SD). Differences between treatments were determined using one-way ANOVA and Tukey’s honest significant difference test. A p-value < 0.05 was considered to be statistically significant.

Strain CTXW 7-6-2 was a rod-shaped gram-positive bacterium with dimensions of 0.62–0.75 μm × 1.5–2.0 μm (Figure 1A). After culturing for 5 days, the strain formed a single medium-sized, round, flat, and dry colony that was opaque, grey-white in colour, and later produced a light red pigment. Following PCR amplification and sequence alignment analyses of 16S rRNA, gyrA, rpoB, and purH, the constructed phylogenetic tree showed that strain CTXW 7-6-2 clustered with B. subtilis, with a self-sustaining rate of 100% maximum likelihood and 1.00 Bayesian inference. Bacillus cereus ATCC 14579 was shown to be an outgroup (Figure 1B).

A total of 25 bacterial strains were isolated, among which strain CTXW 7-6-2 displayed strong antagonistic effects using a double-layer double-culture assay. Table 2 lists the antagonistic effects of strain CTXW 7-6-2 against kiwifruit soft rot-causing pathogens, with growth of Phomopsis sp., A. alternata, and B. dothidea inhibited by 81.76 ± 3.554%, 69.80 ± 1.242%, and 32.03 ± 0.608%, respectively. Additionally, strain CTXW 7-6-2 inhibited the growth of F. fujikuroi, which causes leaf spot disease in kiwifruit, by 38.37 ± 2.265%. No inhibitory effect was observed against P. syringae pv. actinidiae, which causes bacterial canker in kiwifruit.

Strain CTXW 7-6-2 also exhibited antifungal effects against pathogenic fungi that cause diseases in tobacco, plums, and Chinese ground orchids. The most significant antagonistic effects were observed against R. solani, which causes target spot disease in tobacco, with the colony diameter reduced to 0.36 ± 0.191 cm and growth inhibited by 93.36 ± 3.529%. The antagonistic effects of strain CTXW 7-6-2 were also observed against C. spaethianum, which causes anthracnose in Chinese ground orchids, and M. fructicola, which causes brown rot in plums, inhibiting growth by 72.38 ± 0.673% and 56.87 ± 3.801%, respectively. Strain CTXW 7-6-2 displayed no inhibitory effect against P. agglomerans, which causes bacterial shot-hole disease in plums.

The VOCs of strain CTXW 7-6-2 had differing inhibitory effects against soft rot-causing pathogens in kiwifruit. The VOCs displayed no inhibitory effect against B. dothidea growth (Figure 2), but the colony diameter of A. alternata was reduced to 1.46 ± 0.25 cm, growth was inhibited by 76.76 ± 3.93%, and no melanin was produced (Figure 2B). The VOCs of strain CTXW 7-6-2 inhibited the growth of Phomopsis sp. by 54.48 ± 7.96% and reduced the colony diameter to 3.10 ± 0.54 cm.

Botryosphaeria dothidea, Phomopsis sp., and A. alternata displayed obvious growth inhibition zones near CTXW 7-6-2, with slow mycelial growth. The inhibited B. dothidea colonies were grey-black in colour, the tops and internodes of the hyphae were swollen, and their growth was hindered. The inhibited Phomopsis sp. colonies had neat edges, distorted hyphae, and showed dissolution. The inhibited A. alternata colonies were dense and whitish in colour, and the hyphae were swollen and became bead-like at each internode (Figure 3).

The SEM images revealed that the hyphae of the three pathogenic fungi were deformed to varying degrees. The untreated hyphae of each fungal species appeared smooth and plump. In comparison, the hyphae near CTXW 7-6-2 were rough, shrunken, twisted, and otherwise deformed, indicating severe damage (Figure 3).

The genomes of strain CTXW 7-6-2 and 25 representative Bacillus strains contained the same number of core genes and differing numbers of accessory, unique, and exclusively absent genes (Figure 4A). The phylogenetic analysis based on core and pan-genome data demonstrated that strain CTXW 7-6-2 was clearly clustered with B. subtilis and distinct from other Bacillus species (Figures 4B,C), with B. cereus ATCC 14579 as an outgroup. The B. subtilis CTXW 7-6-2 genome was finally assembled into 15 contigs with a total length of 4,221,676 bp (N50 contig size of approximately 1,046,756 bp), guanine and cytosine (GC) content of 43.41%, 4,428 protein-coding genes, 75 tRNAs, 10 rRNAs, and 45 unknown genes. A total of 3,079 annotations from the COG database were divided into 20 functional groups, mainly related to amino acid transport and metabolism, carbohydrate transport and metabolism, and transcription (Figure 4D).

The whole-genome of B. subtilis CTXW 7-6-2 was functionally annotated using eight gene annotation databases (Figure 5A). A total of 4,240 annotations were found in the nr database, accounting for 99.72% of all annotations. Additionally, 4,006 (94.21%), 3,583 (84.27%), 3,449 (81.11%), and 4,228 (99.44%) annotations were found in the Swiss-Prot, CDD, Pfam, and TrEMBL databases, respectively. A total of 1,775 KEGG annotations were associated with 34 metabolic pathways (Figure 5B). The genome of B. subtilis CTXW 7-6-2 had the highest number of genes associated with metabolism (2,104); carbohydrate, amino acid and cofactors, vitamin metabolism, and environmental information processing (412); genetic information processing (253); and cell processes (84). A total of 222 annotations were found in the CAZy database, including glycoside hydrolases (67), glycosyl transferases (61), carbohydrate esterases (46), auxiliary activities (11), carbohydrate-binding modules (27), and polysaccharide lyases (10) (Figure 5C). Glycosyl transferases, glycoside hydrolases, and carbohydrate esterase-related enzymes are involved in the synthesis of secondary metabolites via non-ribosomal pathways. Furthermore, B. subtilis CTXW 7-6-2 genome had 4 count proteins chitinase (GH18), 1 count proteins chitosanase (GH46), 13 count proteins endoglucanase (GH 5,51,16,26,11), and 8 count proteins lysozyme (GH23,73) encoding for possible antifungal CAZymes. AlsD (Alpha-acetolactate decarboxylase) and alsS (acetolactate synthase) genes encoding proteins are involved in the production of VOC 3-hydroxy-2-butanone. GO analysis revealed the main molecular functions of B. subtilis CTXW 7-6-2 were related to catalytic activity and binding function, while the main cellular components were related to cell and membrane parts (Figure 5D). In addition, 92 annotations were found in the PHI-base, including 10 proteins associated with loss of pathogenicity and 50 proteins associated with reduced virulence.

Rapid annotations using subsystem technology (RAST) subsystem analysis (Figure 6) revealed that the genome of B. subtilis CTXW 7-6-2 contained 245 genes involved in the metabolism of amino acids and their derivatives, 201 genes involved in carbohydrates, 161 genes involved in protein metabolism, 55 genes involved in RNA metabolism, and 132 genes involved in cofactors, prosthetic groups, vitamins, and pigment synthesis. The above annotation results indicated that various material and energy metabolism pathways provided the necessary nutrients and energy supply for the growth of B. subtilis CTXW 7-6-2. In addition, 27 genes were involved in iron acquisition and metabolism, 39 genes were involved in the stress response, 38 genes were involved in membrane transport, 40 genes were involved in motility and chemotaxis, 60 genes were involved in the cell wall and capsule, 6 genes were involved in secondary metabolism, 52 genes were involved in fatty acids, lipids, and isoprenoids, suggesting beneficial traits for potential biocontrol capabilities.

Predicting the metabolites produced by B. subtilis CTXW 7-6-2 is conducive to further analysis of its antifungal compounds. The antiSMASH analysis showed that B. subtilis CTXW 7-6-2 had 14 secondary metabolite biosynthetic gene clusters (BGCs), including five NRPS (non-ribosomal peptide synthetase) clusters, two putative NRPS + polyketide clusters, a polyketide cluster, and a thiopeptide cluster. The bioactive compounds with a predicted similarity of 100% were identified as bacillaene, bacillibactin, subtilosin A, bacilysin, and luminmide (Table 3). B. subtilis CTXW 7-6-2 also had four gene clusters with unknown functions, including one CDPS (tRNA-dependent cyclodipeptide synthase) cluster, one T3PKS (Type III PKS) cluster, and two terpene clusters (Figure 7), suggesting that some B. subtilis CTXW 7-6-2 gene clusters may synthesise new antibacterial substances. Schedule 1 and Table 3 show that CTXW 7-6-2 shared seven and six metabolic species classes with B. subtilis NCIB 3610 and 168, respectively. Luminmide, plipastatin, and basiliskamide A/B were unique to B. subtilis CTXW 7-6-2. Both B. subtilis CTXW 7-6-2 and NCIB 3610 had a predicted a gene cluster of unknown function containing yvmC as the core gene, in contrast to B. subtilis subsp. subtilis 168 (Schedule 1). Bacillus sp. that produce pulcherrimin are speculated to contain yvmC.

The ProtParam analysis indicated that the molecular weights of the yvmC and cypX proteins of B. subtilis CTXW 7-6-2 were 28,506.74 and 45,499.08 Da and contained 248 and 405 amino acids, respectively. Additionally, the yvmC and cypX proteins of B. subtilis CTXW 7-6-2 had 33 and 41 positively charged residues (Arg + Lys) as well as 35 and 55 negatively charged residues (Asp + Glu), respectively. Moreover, their theoretical isoelectric points were 6.57 and 5.42 and their grand average of hydropathicity index values were −0.268 and −0.128, respectively. The calculated instability index values indicated that the yvmC (46.24) and cypX (50.78) proteins of B. subtilis CTXW 7-6-2 were negatively charged, unstable, and hydrophilic.

Predict Protein analysis of the secondary protein structure of yvmC showed that alpha helices, beta turns, extended strands, and random coils accounted for 48.39, 4.03, 12.90, and 34.68% of the total structure, respectively. Alpha helices, beta turns, extended strands, and random coils accounted for 50.37, 4.20, 10.86, and 34.57% of the secondary protein structure of cypX, respectively. The secondary structures of both proteins were dominated by alpha helix and random coil structures, followed by extended strands, and then beta turns (Figures 8A,B). The protein tertiary structure homology was predicted using SWISS-MODEL, as shown in Figures 8C,D.

The relative expression levels of yvmC and cypX in B. subtilis CTXW 7-6-2 increased with culture time, increasing 6.6938 ± 0.517-fold and 0.983 ± 0.042-fold on day 5 compared to that on day 1, respectively (Figure 9). yvmC was more highly expressed than cypX in B. subtilis CTXW 7-6-2. In contrast, yvmC and cypX were relatively less expressed in B. subtilis GUMT323, with expression levels of lower than 0.06-fold.