Comparative genomics of Bacillus velezensis and Bacillus subtilis reveals distinction and evolution of lipopeptide antimicrobial gene clusters

Abstract

Species belonging to the genus Bacillus are recognized as important biocontrol agents, especially the Bacillus subtilis and Bacillus velezensis exhibit the excellent antifungal activity, being found in a variety of habitats and demonstrating significant metabolic versatility. However, knowledge regarding the genetic diversity of different Bacillus species is limited. In this study, we employed comparative genomics to elucidate the genetic diversity and evolutionary relationships between B. velezensis and B. subtilis. Our results indicated that the antibacterial activity and colonization features, including biofilm formation and swarming, of B. velezensis strains were significantly greater than those of B. subtilis strains. We conducted a comprehensive genomic analysis of various Bacillus group strains and found that the genome size of B. velezensis was larger than that of B. subtilis, while the GC content of B. subtilis was higher than that of B. velezensis. The Average Nucleotide Identidy (ANI) value and phylogenetic analysis revealed ambiguous classifications among some Bacillus strains. Furthermore, the 20 Bacillus strains examined yielded a pangenome size of 7068 genes, with strain-specific genes ranging from 24 to 305. The core and specific genome of B. velezensis strains, annotated for secondary metabolite biosynthesis, transport and catabolism, were significantly more abundant than those of B. subtilis. The most pronounced difference between B. velezensis and B. subtilis strains was observed in the gene cluster encoding the iturin family of lipopeptides. Evolutionary analysis suggested that the iturin gene cluster of Bacillus may have been transferred from Paenibacillus spp. via horizontal gene transfer (HGT) events during the evolution. Additionally, functional analysis demonstrated that the iturin gene cluster effectively inhibits Fusarium pathogens. Collectively, these findings provide a foundation for a deep understanding of the evolution of different Bacillus strains and establish a theoretical basis for the application of Bacillus strains in agricultural production.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-025-12234-0.

Introduction

As reported in a comprehensive review, pathogenic microorganisms that affect plant health pose a significant and persistent threat to global food production and ecosystem stability [1]. These pathogens diminish plant productivity, resulting in billions of dollars in annual losses due to diseases [2]. Over the past centuries, agrochemicals have played a crucial role in enhancing both crop quality and quantity. However, concerns regarding environmental pollution and the residual effects of pesticides on human health and ecosystem have emerged [3]. Consequently, scientists are increasingly focusing on biopesticides. The genera Bacillus, Streptomyces, and Pseudomonas have been extensively studied and are increasingly marketed as biological control agents [4–6]. Biological preparations derived from spore-forming Bacillus are preferred due to their long-term viability, which facilitates the development of commercial products [7]. Bacillus is classified as a Gram-positive, rod-shaped, spore-forming bacterium that can be aerobic or facultative anaerobic, producing highly resistant dormant endospores in response to nutritional or environmental stresses [8]. Bacillus-based biocontrol agents play a fundamental role in the field of biopesticides, with many Bacillus species demonstrating effectiveness against a broad range of plant pathogens and serving as beneficial agents for plant growth promotion [9–12]. However, our understanding of the evolution and adaption of Bacillus species remains limited.

With the advancement of sequencing techniques and a significant reduction in sequencing costs, the number of sequenced genomes has increased dramatically over the past 10 years [13]. We can determine the genetic traits involved in biocontrol mechanisms based on genome sequences. For instance, the genome sequence analysis of the biocontrol strains B. altitudinis GLB197, which exhibits strong inhibition of grape downy mildew, reveals a gene cluster associated with nonribosomal peptide synthetase and plant interactions [14]. Additionally, a secondary metabolite gene cluster was identified in the genome sequence of biocontrol strain Bacillus velezensis JLU-1, confirming the antagonistic activity of the secondary metabolites [15]. Numerous genomes are now available in public repositories (e.g., GenBank), providing greater opportunities to compare different strains and understand their evolution and adaption. For example, by selecting various Novosphingobium strains isolated from diverse habitats and conducting comparative genomic analysis, researchers identified habitat-specific genes and regulatory hubs that influence habitat selection for Novosphingobium spp. Notably, sulfur acquisition and metabolism were the only core genomic traits that significantly differed in proportion among these ecological groups [16]. In the case of Klebsiella pneumoniae, a comparison between the genome of the mutualistic, nitrogen-fixing endophyte K. pneumoniae 342 and the clinical isolate MGH78578 revealed major difference in metabolism, surface attachment, and secretion. These findings suggest that strain 342 is well adapted to evade plant defense reactions and successfully establish itself within a plant [17]. It is well-established that bacterial adaption to diverse environments during the evolutionary process leads to differentiation, as evidenced by distinct genomic and physiological characteristics [18]. However, the genome-based diversity and evolution of species within the genus Bacillus have not been thoroughly investigated to date.

B. subtilis and B. velezensis have garnered significant attention as biological agents against plant pathogens [19]. B. velezensis, along with its sister species B. subtilis, constitutes an evolutionary compact yet physiologically versatile group of bacteria, with strains isolated from a diverse array of ecological habitats, including agricultural soil, aquatic environments, and human food [18, 20]. However, there is limited knowledge regarding the genomic distinctions among different Bacillus species. The Bacillus amyloliquefaciens group encompasses B. velezensis, B. siamensis, and B. amyloliquefacines. Pangenome analysis reveal that the core genome of B. velezensis is enriched in genes related to secondary metabolism when compared to B. siamensis and B. amyloliquefaciens, particularly the fengycin gene cluster [21]. Additionally, it has been reported that the gene cluster associated with macrolactin synthesis is present in B. velezensis but absent in B. siamensis and B. amyloliquefaciens [22]. A detail genomic analysis of the type strains DSM7 and FZB42 indicates that DSM7 does not produce the polyketides difficidin and macrolactin and is impaired in its ability to synthesize lipopeptides other than surfactin [23]. Comparative genomic analysis of B. paralicheniformis and B. licheniformis demonstrated that specific genome segments, including insertion sequence, prophage-like elements, and secondary metabolism synthases in B. paralicheniformis, are involved in adaptation to environmental niches [24]. Furthermore, phylogenomic analysis suggests that strains isolated from plant-associated (PA) habitats can be distinguished from those originating from non-plant-associated (nPA) niches. Functional analyses indicate that the core genomes of PA strains are more abundant in genes relevant to intermediary metabolism and secondary metabolites biosynthesis compared to those of nPA strains, and they also possess additional specific genes involved in the utilization of plant-derived substrates and the synthesis of antibiotics [18].

To gain a better understanding of the genomic diversity and evolutionary dynamics of different species within the genus Bacillus, we utilized two strains, B. velezensis PG12 and B. subtilis 9407, which have been identified as promising biocontrol agents against apple ring rot from our lab [25, 26]. Additionally, we incorporated 18 previously published genomes of B. velezensis and B. subtilis for comparative genomic analysis. The objectives of this study are: (1) to identify genomic and phenotypic differences between B. velezensis and B. subtilis; (2) to determine the specific genes responsible for phenotypic distinctions among these species; and (3) to explore the evolution and functions of these specific genes. Our results indicated significant genomic differences between B. velezensis and B. subtilis strains, with the most notable difference being the iturin gene cluster associated with lipopeptide production. Evolutionary analysis suggested that the iturin gene cluster in Bacillus may have been transferred from Paenibacillus spp. through horizontal gene transfer. Furthermore, functional analysis demonstrated that the iturin gene cluster exhibits strong inhibitory effects against Fusarium pathogens. Collectively, these findings provide a foundational understanding of the evolutionary dynamics among different Bacillus strains and offer a theoretical basis for their application in agricultural production.

Materials and methods

The bacterial and fungal strains used in this study

The biocontrol strains PG12 and 9407 were isolated from apple and demonstrated a stronger inhibitory effect against apple ring rot [25, 26]. All plant pathogens including Botryosphaeria berengeriana, Botrytic cinerea, Colletotrichum spp., Fusarium spp., Alternaria solani, Botryosphaeria dothidea, Cryptosporella viticola, Monilia sp., Exserohilum turcicum, Fusarium oxysporum, and Valsa mali used in this were stored at the College of Plant Protection, China Agricultural University. Fungal strains were preserved in paraffin at 4℃ prior to use, while bacterial strains were routinely cultured in Luria–Bertani (LB) media (1% tryptone, 0.5% yeast extract, and 1% NaCI) or on LB supplemented with 1.5% (w/v) agar at 37℃. The pathogens were cultivated at 25℃ on potato dextrose agar plate (PDA, 200 g potato, 20 g glucose, and 15 g agar per liter). Erythromycin (5 µg/mL) was added as an antibiotic for culturing deficient mutant strains (∆ituD).

In vitro antifungal experiment

The plant pathogens were activated on a PDA plate for 5 to 7 days. Subsequently, a 5 mm diameter mycelial plug was excised from the edge of the plate and placed at the center of a fresh PDA plate. In the plate confrontation assay, 2 μL of overnight cultures grown in LB medium was spotted on the PDA plate, positioned 2.5 cm from the center. The antifungal activities were measured by measuring the diameter of the mycelium after 5 to 7 days of incubation at 28℃. Each fungal strain was tested in triplicate.

Bioactivity assay

The fermentation broths of bacterial strains PG12 and 9407 were prepared in LB medium at 37℃ with shaking at 180 rpm and subsequently diluted to an optical density of OD₆₀₀ = 1 using sterile water. Healthy tomato leaves were selected for the bioactivity assay. Initially, the tomato leaves were surface-disinfected with 75% ethanol and rinsed three times with sterile water. The leaves were then immersed in 100 mL of the prepared bacterial suspensions for 30 min. Control samples consisted of tomato leaves soaked in sterile water. After allowing the water to evaporate from the tomato leaves, a 5 mm diameter mycelial disc was inoculated onto the leaves, which were then placed on 1% water agar and incubated in a 12 h light/dark cycle at 25℃. The diameters of the lesions on the tomato leaves were measured separately after 5 d.

Biofilm formation and swarming motility assay

The biocontrol strains PG12 and 9407 were prepared by shaking at 37℃ until an optical density (OD₆₀₀) of 0.8 was achieved using LB liquid medium (5 mL). Subsequently, 1 mL of the cell culture was collected via centrifugation at 6000 g for 5 min, washed with phosphate-buffered saline (PBS), and then resuspended in 100 μL of PBS. The swarming motility of the aforementioned strains was assessed using the protocols described by Chen et al., [27]. LB plates containing 0.7% agar were dried in a laminar flow hood for 20 min, after which 3 μL of the cell suspension of strains PG12 and 9407 were added to the center of each plate. The plates were incubated at room temperature for 6 h to allow for sufficient cell growth to visualize the swarming zone. Then, the swarming agar plates were dried for another 2 h in a laminar flow hood, after which the diameter of the swarming zone was measured. The experiment was conducted with three independent assays.

The biofilm formation experiment was conducted in MSgg medium. The strains PG12 and 9407 were cultured overnight in LB medium at 37℃. Subsequently, 4 μL of culture was inoculated into 4 mL of MSgg medium in 12-well plates and incubated at 28℃ for up to 72 h. To quantify biofilm formation, the culture beneath the biofilm was carefully removed, and the biofilm in each well was washed with 3 mL sterile saline. The biofilm was then fixed with 2 mL of 99% (v/v) methanol for 15 min and air-dried. The dried pellicles were stained with 2 mL of crystal violet (1%) solution for 10 min. The excess staining solution was cautiously removed. Finally, the pellicle bound by the staining solution was dissolved in 5 mL of acetic acid solution (33%, v/v) and diluted 100–200 times for determination of absorbance at OD₅₇₀ [28]. For colony architecture, 2 μL of the culture was added onto the surface of an MSgg plate containing 1.5% agar. The plates were then incubated at 28℃ for 72 h. The experiments were conducted with three independent replicates. The procedure for the type strain B. velezensis FZB42 and B. subtilis 168 was identical to the aforementioned method.

Phylogenomic analyses

A maximum-likelihood phylogenetic tree of Bacillus species was constructed based on single-copy core proteins shared among Bacillus genomes and the genome of Paenibacillus polymyxa M1. The construction methods were as follows: (1) Multiple alignment of amino acid sequences was performed using mafft version 7.310 [29]; (2) Conserved blocks from the multiple alignment of test protein were selected using Gblocks [30]; (3) The maximum-likelihood tree was constructed using RAxML version 8.2.10 software with the PROTGAMMALGX model and 100 bootstrap replicates. The tree was visualized using iTOL (http://itol.embl.de/). The ANI values were calculated using JSpecies software with MUMmer (NUCmer) alignment [31]. Single gene alignments were conducted using Molecular Evolutionary Genetics Analysis (MEGA). Neighbor-joining trees were constructed using MEGA software with 1000 bootstrap replicates [32]. The phylogenetic tree was visualized using iTOL (https://itol.embl.de/).

Pan-genome analysis

The Bacillus strains studied were obtained from the National Center for Biotechnology Information (NCBI). The Pan Genome Analysis Pipeline (PGAP) was employed to identify all orthologous pairs among the tested Bacillus genomes [33]. Core orthologs, defined as those with at least 50% protein sequence identity to one another, were clustered with a minimum of 50% overlap with the longest sequence and an e-value of 1e-5. The common dataset of shared genes among the tested strains was designated as their core genome, while the total set of genes within the tested genomes was referred to as the pangenome. Unique genes were defined as the set of genes present in each individual strain. The core genome and strain-specific genomes were extracted from the pangenome table using a custom Perl script [34]. Functional annotation of the core and specific genomes was conducted using the Cluster of Orthologous Groups of proteins (COG) database. All gene clusters for antibiotic synthesis were predict using the antiSMASH 7.01 online prediction software with default parameters [35].

Construction of ituD deletion mutant

The ituD deletion mutant was generated from PG12 through homologous recombination. Primers ituD-U-F/ituD-U-R and ituD-D-F/ituD-D-R were utilized to amplify the upstream region and downstream region (1000 bp) of ituD, respectively. The erythromycin resistance gene was amplified from pAX01-gp35 plasmid using primer pair Em-F and Em-R. To construct the knockout vector, the upstream, downstream, and resistance gene regions were ligated using gene splicing by overlap extension (3200 bp). The resulting product was then ligated into pMD19-T to create the knockout plasmid pTITUD, which was subsequently transferred into PG12 via electroporation. Erythromycin-sensitive clones were isolated, and the mutants were identified by PCR using primer pair ituD-out-F/ituD-out-R and confirmed by Sanger sequencing (Table S1).

Results

The biocontrol strain PG12 has a stronger antagonist against plant pathogens than strain 9407

We examined the antifungal activities of biocontrol strain PG12 and 9407 on agar plates. A total of twenty-one plant pathogens, sourced from different host plant, were randomly selected for this study. The results indicated that strain PG12 effectively inhibited the mycelia growth of the plant pathogens, as evidenced by a clear inhibition zone, whereas strain 9407 exhibited reduced antifungal activity under the same experimental conditions (Figure S1). These findings confirmed that both strains are viable biocontrol agents. Notably, strain PG12 demonstrated significantly stronger antifungal activity compared to strain 9407 against most pathogenic organisms (p-value < 0.05, Figure S1). Furthermore, we selected B. cinerea as the model pathogen and performed biocontrol experiment in tomato leaf to confirm the ability of antifungal activities of PG12 and 9407. The lesion area in leaves treated with strains PG12 and 9407 was significantly smaller than that in the control group treated solely with LB broth (Fig. 1). The results from the leaf assays revealing that the lesions in leaves treated with strain PG12 were significantly reduced compared to those treated with strain 9407. This indicates that strain PG12 is more effective in controlling plant diseases.

Fig. 1.

Comparative analysis of the inhibitory effect on tomato gray mold between biocontrol strain PG12 and 9407. A Representative photographs of the inhibitory effect of biocontrol bacteria (PG12 and 9407) on tomato gray mold. B Data presented are the inhibitory effect of biocontrol bacteria (PG12 and 9407) on tomato gray mold. The data are expressed as the mean ± SD (n = 3). * represented p-value < 0.05 and ** represented p-value < 0.01.CK represented the treatment of LB

Colonization is a crucial step in facilitating disease control activities. Therefore, it is imperative to investigate the processes associated with colonization. Firstly, strains PG12 and 9407 were examined for their swarming motilities. The results indicated that strain PG12 exhibited excellent swarming motility, colonizing more than half of the plate after 5 h of growth. Interestingly, the swarming motility of strain 9407 was significantly decreased compared to that of the PG12 strain (Fig. 2A, B). Subsequently, the biofilm formation of strains PG12 and 9407 was also investigated. Strain PG12 formed a thick and wrinkled floating biofilm in MSgg broth and developed colonies characterized by dense wrinkled structures on the MSgg plate (Fig. 2C, D, E). In contrast, strain 9407 formed a thin and fragile floating layer along with flat colonies in both the MSgg plate and broth (Fig. 2B, C, E). Additionally, we compared the colonization features of strain B. velezensis FZB42 and B. subtilis 168, and the results was consistent with those obtained from strain PG12 and 9407 (Figure S2). Strain PG12 likely possesses superior colonization abilities compared to strain 9407. Overall, the biocontrol strain PG12 demonstrated stronger antifungal activities than strain 9407.

Fig. 2.

Comparative analysis of phenotype features of biocontrol strain PG12 and 9407. A Swarming ability assays of PG12 and 9407. Swarming assays were assessed in LB plate containing 0.7% agar incubated at room temperature for 5 h. B Comparative analysis of the swarming ability for PG12 and 9407. C Pellicle formation of strain PG12 and 9407 on Msgg liquid media. D The biofilm formation of PG12 and 9407 based on OD readings from 12-well-plate experiments. E The colony morphology of strain PG12 and 9407 on MSgg plates. Biofilm assays were detected in MSgg liquid media for 72 h at 28℃. Data presented as mean ± SD (n = 6). The statistical analysis was performed using GraphPad Prism 7 software by one-way ANOVA. ** indicated P < 0.01

General genomic organization of B. velezensis and B. subtilis strains

The genome of PG12 and 9407 were sequenced previously [21, 36]. According to the assembled data, the scaffolds for the biocontrol strains PG12 and 9407 are 22 and 16, respectively. The genome sizes of PG12 and 9407 are found to be similar, with PG12’s genome sequence measuring 3,995,119 bp in length and a G + C content of 46.45%. In contrast, 9407’s genome sequence is 4,068,075 bp with G + C content of 43.66% [10, 21]. The predicted number of protein-coding genes for PG12 and 9407 are 3,884 and 4,033, respectively. Compared to PG12, strain 9407 exhibits a smaller average length of protein-coding genes. Furthermore, we compared the basic genomic features of members from the two groups of strains. The average genome size of B. subtilis is larger than that of B. velezensis strains, while the GC content of B. velezensis is higher than that of B. subtilis strains (p < 0.0001, Fig. 3).

Fig. 3.

Different features of the two Bacillus species. A The genome size. B The GC content

Core and specific genome of B. velezensis strains are more abundant in genes involved in secondary metabolism compared with B. subtilis strains

To gain insight into the core and variable gene pool among Bacillus isolates from diverse phylogenetic backgrounds, we conducted pan-genomic studies. We utilized 20 strains to construct the pangenome for B. velezensis and B. subtilis, which can be employed to describe these two bacterial species (Table 1). The analysis of the 20 Bacillus strains revealed a total pangenome size of 7068 genes. The core genome comprises 2679 genes, the accessory genome consists of 2294 genes, and the unique genome contains 2095 genes. Additionally, the strain-specific genes for the selected Bacillus strains ranged from 24 to 305 (Fig. 4). Furthermore, we computed the set of orthologous proteins shared among the test genomes from each species to elucidate the genomic features. Notably, B. velezensis possesses a pangenomes of 5285 genes, which includes 3135 core genes, 864 accessory genes, and 1286 unique genes. The number of unique genes per strain varies from 43 to 314. In contrast, the B. subtilis strains exhibit a pangenomes of 5092 genes, comprising 3286 core genes, 834 accessory genes, and 990 unique genes. The unique genes for each strain range from 25 to 185 (Figure S3). Functional annotation of the genes in the pangenome, performed using the COG database, revealed a similar distribution of functional categories across the different pangenome set. We compared the core genome and specific genome of B. velezensis and B. subtilis strains using COG assignments to ascertain whether there were differences in the proportions of the core and specific genomes attributable to a particular cellular process. The results indicated that the secondary metabolites biosynthesis, transport and catabolism in B. velezensis are significantly higher than those in B. subtilis (p-value < 0.05, Table 2).

Table 1.

The Bacillus used in this study

Species	Strain	Genome size (Mbp)	Status	Source	Target diseases for biological control
B. velezensis	Bs-916	3.93	Complete	Paddy soil, China	Rhizoctonia solani
	CC178	3.92	Complete	Cucumber phyllosphere, Korea	Fusarium oxysporum, Phytophthora capsici, Rhizoctonia solani, and Sclerotinia sclerotiorum
	B15	4.01	Complete	Grape skin, China	Several grape fungal pathogens
	J-5	4.12	Complete	Tomato rhizosphere, China	Botrytis cinerea
	L-S60	3.90	Complete	Turfy soil, China	Rhizoctonia solani, Fusarium oxysporum and so on
	UCMB5036	3.91	Complete	Inner tissues of cotton plant, Tajikstan	Diseases in oilseed rape and Arabidopsis thaliana
	FZB42	3.92	Complete	Sugar beet, rhizosphere	A broad spectrum of pathogens
	NJN-6	4.05	Complete	Banana rhizosphere, China	Fusarium oxysporum
	CAU B946	4.02	Complete	Rice rhizosphere, China	Tobacco black shank, rice sheath blight, cotton fusarium wilt, cotton verticillium wilt, and wheat scab
	PG12	4.00	Draft	Apple fruit, China	Apple ring rot
B. subtilis	BAB-1	4.02	Complete	Cotton rhizosphere, China	Several pathogens including tomato gray mold
	PY79	4.06	Complete	-	-
	BSn5	4.09	Complete	Amorphophallus konjac calli tissue, China	Erwinia carotovora
	GQJK2	4.07	Complete	Lycium barbarum rhizosphere, China	Fusarium solani
	SG6	4.08	Complete	Anthers of luffa	Fusarium graminearum
	UD1022	4.03	Complete	Natural Root	-
	BSD-2	4.03	Complete	Inner of cotton stem, China	Verticillium dahlia Kleb and Botrytis cinerea
	XF-1	4.06	Complete	Chinese cabbage rhizosphere, China	Many fungal pathogens of plants
	HJ5	4.01	Complete	Cotton rhizosphere, China	Verticillium wilt of cotton
	9407	4.07	Draft	Apple fruit, China	Apple ring rot

Fig. 4.

The genome diversity of B. velezensis and B. subtilis strains. Each strain represented by an oval that is colored: B. velezensis (blue) and B. subtilis (red). The number of orthologous coding sequences shared by all strains (the core genome) is in the center. Numbers in non-overlapping portions of each oval showed the strain-specific genes. The total number of protein coding genes within each genome is listed below the strain name

Table 2.

Comparison of COG assignments between B. velezensis and B. subtilis strains

	Core genome			Specific genes
Individual functional categories	B. velezensis	B. subtilis	P- value	B. velezensis	B. subtilis	P- value
J: Translation, ribosomal structure and biogenesis	142	149	0.8099	1	1	1
K: Transcription	183	191	0.8306	19	36	0.7686
L: Replication, recombination and repair	101	100	0.8857	3	2	0.3716
D: Cell cycle control, cell division, chromosome partitioning	31	28	0.6954	1	0	0.3774
V: Defense mechanisms	47	50	0.8384	10	13	0.6609
T: Signal transduction mechanisms	79	77	0.8078	5	9	1
M: Cell wall/membrane/envelope biogenesis	163	159	0.7306	10	13	0.6609
N: Cell motility	21	21	1	0	1	1
U: Intracellular trafficking, secretion, and vesicular transport	25	25	1	1	2	1
O: Posttranslational modification, protein turnover, chaperones	90	90	0.9396	3	8	0.549
C: Energy production and conversion	166	158	0.5673	12	12	0.2816
G: Carbohydrate transport and metabolism	168	193	0.2766	16	44	0.1171
E: Amino acid transport and metabolism	241	257	0.6391	8	23	0.249
F: Nucleotide transport and metabolism	77	75	0.8054	0	0	-
H: Coenzyme transport and metabolism	166	166	0.9099	17	23	0.6115
I: Lipid transport and metabolism	71	67	0.6673	2	2	0.6345
P: Inorganic ion transport and metabolism	144	150	0.8575	7	15	0.6578
Q: Secondary metabolites biosynthesis, transport and catabolism	59	38	0.03113	21	3	0.00000122
R: General function prediction only	302	316	0.7659	22	42	0.6805
S: Function unknown	299	312	0.7975	21	48	0.2859

Distribution of lipopeptides gene clusters reflects the species phylogenetic

Lipopeptides are small compounds synthesized by various microorganisms as secondary metabolites. Bacillus species produce different types of lipopeptides derived from secondary metabolites that exhibit specific activities against plant pathogens [37]. Antimicrobial lipopeptides from Bacillus species are primarily categorized into three main families: iturin, surfactin, and fengycin. These three types of lipopeptides were identified in the PG12 genome, whereas only surfactin and fengycin were detected in the genome of strain 9407 (Figure S4). Furthermore, we utilized the aforementioned genome sequences to investigate the differences between B. velezensis and B. subtilis. The most significant distinction between the B. velezensis and B. subtilis strains was observed in the gene cluster encoding the iturin family. In contrast, the surfactin and fengycin gene clusters are highly conserved across B. velezensis and B. subtilis strains, with only minor variations noted among some strains (Fig. 5).

Fig. 5.

The gene clusters for lipopeptide antibiotics in Bacillus strains

To elucidate the evolution of the iturin gene cluster in Bacillus strains, we constructed phylogenetic trees based on the protein sequences of ituA, ituB, ituC, and ituD (Figure S5-S8). The sequences of Bacillus clustered with Paenibacillus spp. suggesting a transfer of the iturin gene cluster from Paenibacillus spp. However, we found that there are complete iturin gene clusters in the genome of Paenibacillus. The iturin gene cluster was probably transferred as a complete entity. The iturin gene clusters are highly similar among different B. velezensis strains, with the homology of the gene cluster reaching 97%−99%. By analyzing the upstream and downstream genes of the iturin genes in the B. velezensis strains, we identified that the upstream and downstream genes of the iturin gene cluster include CoA transferasesubunit A and B, 3-hydroxybutyrate dehydrogenase, glucronoxylanase, etc., among different strains (Fig. 6). The high similarity genes indicates that the upstream and downstream genes are highly conserved. In contrast to the B. velezensis strain, the results of collinear analysis of B. atrophaes strains show that the upstream and downstream genes of the iturin antibiotic in the B. atrophaes strain include putative transporter YoaB, putative sugar kinase YoaC, glutamate 5-kinase, LysR family transcriptional regulator, which also exhibit certain conservatism among different strains (Figure S9). This suggests that B. velezensis and B. atrophaes strains may have acquired their related gene clusters through different evolutionary processes.

Fig. 6.

The recombination hot spot of the iturin gene cluster in B. velezensis strains

The iturin gene cluster is a functional unit for suppression Fusarium

To investigate whether the iturin gene cluster functions as a unit for antifungal activities, we conducted a comparative analysis of the antibacterial activity among PG12, 9407, and ∆ituD. We selected pathogenic fungi that exhibited significant differences in mentioned study (Figure S1). The results indicated that the inhibitory effects on plant pathogens, such as Fusarium moniliforme, Fusarium sp. (sugarbeet), Fusarium sp. (cotton), Botryosphaeria dothidea, Fusarium sp. (potato), were significantly different between PG12 and ∆ituD. In contrast, no significant differences were observed between 9407 and ∆ituD. Therefore, the iturin antibiotic family may demonstrate effective control over Fusarium strains (Fig. 7). Additionally, while there were some differences in the inhibitory effects of strain 9407 compared to ∆ituD against other pathogens, these differences were not significant, suggesting that other types of antibiotics may also contribute to the observed effects. Overall, the acquisition of iturin antibiotic gene cluster is crucial for the inhibitory effects against Fusarium pathogens.

Fig. 7.

Comparative analysis of antibacterial activity between wild type strain PG12, 9407, and the mutant strain ∆ituD. The left strain is the biocontrol strain 9407, the right strain is the biocontrol strain PG12 and the below strain is ∆ituD. The data are expressed as the mean ± SD (n = 3). * represented p-value < 0.05 and ** represent p-value < 0.01

Discussion

In this study, we found that the antibacterial activity and colonization features of B. velezensis were significantly higher than those of B. subtilis. Phylogenetic analysis revealed that certain Bacillus strains exhibit ambiguous classification. Furthermore, the core and specific genome of B. velezensis, annotated with respect to secondary metabolite biosynthesis, transport and catabolism, are significantly more extensive than those of B. subtilis, particularly regarding the iturin gene cluster. Evolutionary analysis suggested that the iturin gene cluster of Bacillus may have been transfer from Paenibacillus spp. through HGT events. Function analysis demonstrated that the iturin gene cluster effectively inhibits Fusarium pathogens. These findings provide comprehensive evidence for the evolutionary of different Bacillus species and valuable insights for engineering Bacillus strains for agricultural applications.

The taxonomy of Bacillus should require more detailed research

Traditional molecular and biochemical identification methods for bacterial strain often leads to inaccuracies and misidentifications. However, with the advent of sequencing technologies, an increasing number of genome sequences have been published, providing an ideal opportunity to enhance our understanding of bacterial phylogeny. The biocontrol strains PG12 and 9407 were identified as B. velezensis and B. subtilis, respectively, based on the whole genome sequencing. Moreover, the classification of many Bacillus strains remains contentious. For instance, the phylogenetic and taxonomic relationships within the B. cereus group are still debated [38], and the current taxonomy of B. pumilus group strains also requires vertification [14]. ANI value was higher than 95% could be determined to be the same species, while some research recommend 92.5% as the threshold for the B. cereus group [39, 40]. Notably, the ANI value between strains Bs-916 and FZB42 was recorded at 98.86%. Furthermore, strain B. subtilis J-5 clustered with B. velezensis in evolutionary analyses, suggesting that these two strains should indeed be classified as B. velezensis (data not shown). We will distinguish closely related B. cereus group strains based on the reported ANI values of 92.5%. This approach also enhances the accuracy of identifying environmental isolates. Furthermore, it enables us to conduct the evolutionary studies within this group. During our evolutionary analysis of Bacillus strains, we also identified numerous strains of uncertain classification in the database. The classification of these uncertain species can be refined through ANI values analysis. Additionally, the classification of many Bacillus strains remains ambiguous. The ANI value serves as a robust method for studying the classification of bacterial strains, highlighting the need for increased attention to the classification issues of Bacillus strains. By integrating whole genome sequences of various Bacillus strains, we can establish a solid foundation for future research.

The pan-genome of B. velezensis strains are more abundant in genes involved in secondary metabolism compared with B. subtilis strains

The phenotypic analysis demonstrated that the antifungal activities of PG12 are superior to those of 9407, corroborating the results obtained from the type strain of B. velezensis and B. subtilis (Fig. 1, 2 and Figure S1, S2). To elucidate the specific traits of B. velezensis and B. subtilis strains, we selected several biocontrol Bacillus strains for comparative genome analysis. Firstly, the genome size of B. velezensis is larger than that of B. subtilis, whereas the GC content of B. subtilis exceeds that of B. velezensis (Fig. 3). It has been reported that bacteria from complex environmental habitats, characterized by diverse metabolic activities, tend to possess larger genomes and higher GC content [41]. For instance, the genome size and GC content are correlated with the ecological strategies of different marine bacteria, with free-living bacteria exhibiting lower GC content and smaller genomes compared to path-associated bacteria [42]. Microbes may evolve into auxotrophs by losing certain genes, thereby adapting to reliance on external resources [43]. The pangenome of B. velezensis and B. subtilis is considered open, indicating that there are frequent gene gain and loss events during the evolution and adaption. Furthermore, research on the pangenomes is critical for elucidating the functional differences and similarities between strains, providing molecular evidence for phenotypic variations and commonalities [44]. The pangenome analysis revealed an increasing number of secondary metabolites gene clusters identified in B. velezensis (Table 2). Additionally, we observed that the iturin family gene cluster is not annotated in the B. subtilis strains when analyzed through anti-SMASH and comparative genomic analysis (Fig. 5 and Figure S4). Marine B. pumilus strains exhibited a great abundance of genes associated with defense, as well as replication, recombination and repair, compared to those isolated from land [45]. Furthermore, the pangenome analysis of B. altitudinis, B. pumilus and B. safensis indicated that B. altitudinis is enriched in non-core genes related to DNA recombination and repair, which may aid this species in copying with various stresses. In contract, B. safensis possesses a higher number of genes involved in carbohydrate transport and metabolism, while B. pumilus displays the highest quantity of non-core genes and strain-specific genes, contributing to the genetic diversity of this species [45]. This analysis provides insights into the evolutionary of specific strains. However, the phylogenetic trees exhibit inconsistencies among different environment isolates, suggesting potential functional variations across these environments. Further in-depth analysis was warranted to confirm these findings. Despite the observed genomic differences, particularly concerning the iturin gene cluster, we must conduct HPLC analysis to ascertain differences in antibiotic production in the future studies.

The evolutionary analysis for the iturin family gene cluster

Gene acquisition methods primarily encompass horizontal gene transfer and gene replication. Horizontal gene transfer is particularly prevalent in bacteria. The principal methods employed to detect horizontal gene transfer include parametric methods and evolutionary tree [46]. In this study, we found no significant difference between the GC content of the iturin gene cluster and the genomic GC content. It has been observed that regions acquired through horizontal transfer often exhibit lower GC content compared to the host background genome, even after prolonged retention [47]. Evolutionary analyses of the iturin gene clusters revealed that genes derived from Bacillus strain cluster with those from Paenibacillus strain (Figure S5-S8). However, it remains essential to confirm the presence of the iturin family antibiotic gene cluster in Paenibacillus strains and to elucidate its form, as this will enhance our understanding of the evolution of this antibiotic gene cluster. We identify the complete iturin gene cluster in the Paenibacillus strains. The antibiotics belonging to the iturin family probably undergo complete horizontal transfer at the gene cluster level resulting in the current antibiotic gene cluster. Functional analysis demonstrated that the iturin gene family cluster exhibited strong inhibitory effects against Fusarium (Fig. 7). However, further experiments are required to verify its efficacy against other plant pathogens. Further studies should also include greenhouse experiments and mechanistic investigations. The number of biocontrol strain in the B. amyloliquefaciens group was found to be higher than that of the B. subtilis strains. This may be attributed to the presence of more extensive gene cluster for secondary metabolites in B. velezensis compared to B. subtilis. The results indicated that most gene clusters responsible for fengycin, bacitracin, and lantipeptide were exclusively present in B. paralicheniformis and were acquired through HGT. These clusters may be served as genetic markers for distinguishing B. paralicheniformis from B. licheniformis [48]. In the collinearity analysis, the upstream and downstream genes associated with the iturin gene cluster in B. velezensis strains were identified as CoA transferase subunit A and B, 3-hydroxybutyrate dehydrogenase, glucronoxylanase. In contrast, the upstream and downstream genes of the iturin antibiotic in the B. atrophaes strain included the putative transporter YoaB, putative sugar kinase YoaC, glutamate 5-kinase, LysR family transcriptional regulator (Fig. 6 and Figure S9). Evolutionary analysis, along with other analyses, was integrated with the whole genome for comprehensive evolutionary assessment. Antibiotics from the iturin family may have emerged through two distinct evolutionary processes, with the B. subtilis strain evolving later and subsequently losing the iturin gene cluster (Figure S10). Therefore, further research is necessary to elucidate the evolutionary process.

Conclusion

The antibacterial activity of the B. velezensis strains was significantly higher than that of the B. subtilis strains. Additionally, the B. velezensis strains exhibited superior biofilm formation and cell mobility compared to the B. subtilis strain. Furthermore, the B. velezensis strain possessed a more comprehensive iturin lipopeptide gene cluster than the B. subtilis strains, as indicated by lipopeptide analysis. This gene cluster may have been transferred from Paenibacillus strains. Both B. velezensis and B. subtilis strains acquired the iturin gene cluster through two ways; it is plausible that during evolution, the B. subtilis strains lost this gene cluster. Functional analysis revealed that the iturin gene cluster effectively inhibits Fusarium pathogens.

Abbreviations

LB

Luria-bertani

COG

Cluster of orthologous groups of proteins

nPA

Non-plant-associated

PDA

Potato dextrose agar

PBS

Phosphate-buffered saline

NCBI

National center for biotechnology information

PGAP

Pan genome analysis pipeline

HGT

Hhorizontal gene transfer

ANI

Average nucleotide identidy

PA

Plant-associated

Authors’ contributions

QW, YL, and ZW designed the experiments. QZ, YZ, and LZ performed the experiments. QZ and YZ wrote the manuscript and analyzed the data. WJ, L, JZ, LW, HG, and YL prepared the figures. All authors reviewed the results and approved the final version of the manuscript.

Funding

This study was supported by National Key R&D Program of China (2023YFD1401400), National Natural Science Foundation of China (32172474 & 32102273), and Potato Germplasm Innovation and New Variety Breeding Challenge Program (2022JBGS0037).

Data availability

The genome sequence of PG12 and 9407 has been deposited in NCBI number the GenBank accession number PIWI00000000.1 and PISO00000000.1, respectively.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

All authors of the manuscript have reviewed and consented to the publication of identifiable details, including photographs and any information within the text in the journal of BMC Genomics.

Competing interests

The authors declare no competing interests.