Bacillus velezensis R22 inhibits the growth of multiple fungal phytopathogens by producing surfactin and four fengycin homologues

Abstract

Significant agricultural losses are caused by the phytopathogenic fungi Botrytis cinerea and Phytophthora infestans, as well as bacteria of the Ralstonia solanacearum species. The present work aimed to isolate rhizobacteria for simultaneous biocontrol of these three phytopathogenic species and to suggest the mechanisms of their antagonistic action. Among 120 Bacillus spp. isolated from soils, Bacillus velezensis and Bacillus licheniformis strains displayed the highest activity against all three phytopathogens. A rapid, polymerase chain reaction-based method for detecting nonribosomal peptide synthetase genes was developed to elucidate the genetic basis of these traits. The presence of fenA, srfAA, ppsA, and lchAA genes, encoding fengycin/surfactin/plipastatin synthetases and lichenysin synthase, was revealed in the strains’ genomes. The whole genome sequencing (WGS) of B. velezensis R22 showed that it contains 4,081,504 bp (with G + C content 46.35%), 4087 genes for 3935 proteins, 72 tRNAs, 14 rRNAs, and 5 ncRNAs. WGS allowed the prediction of 10 complete clusters for secondary metabolites with putative antimicrobial activity: difficidin, fengycin, bacillaene, butyrosin, bacillibactin, bacilysin, surfactin, macrolactin H, macrolactin R22, and velezensin. LC-MS and high-sensitivity UHPLC-Q-TOF LC-MS/MS analysis were used to search for the predicted metabolites in cell-free supernatants of B. velezensis R22. The compounds with the strongest antifungal activity are surfactin with a C15 β-OH fatty acid chain; two homologous forms of fengycin A; and two fengycin B homologues containing C16 and C17 β-hydroxy fatty acid chains. The broad antimicrobial spectrum of B. velezensis R22 and its molecular characterization provide a good basis for the future development of plant protection preparations.

Keywords:

• Bacillus velezensis
• fengycin
• surfactin
• Botrytis cinerea
• Phytophthora infestans
• LC-ESI-MS

Introduction

The damage caused by plant diseases, or by the so-called biotic stress, leads to a drastic reduction in agricultural yields. Two-thirds of all diseased plants are infected with fungi, diminishing the annual harvest by at least 30% worldwide [1]. Fungicides are the most commonly used pesticides to control mildew, rust, and mold, as well as to prevent various other fungal infections. The forecast for sales revenue of fungicides in 2027 is USD 22.92 billion, with a production volume of 1675.97 kilotons [2]. The majority of these fungicides are synthetic compounds including triazoles, diazoles, benzimidazoles, and dithiocarbamates. However, the widespread application of fungicides inevitably leads to the accumulation of toxic residues in fruits, the development of resistance in target pathogens, and a negative impact on human health and the environment [3,4]. Moreover, the use of fungicides is not permitted in bio-food production because contamination of the final product is virtually certain, including cereals and grains, oilseeds, pulses, fruits, vegetables, and even poultry fat [5]. Therefore, the share of biofungicides has been gradually increasing in the last decade and is expected to show an extremely rapid annual growth rate of 6.7% in terms of production volume until 2027 [2].

Biofungicides contain live microorganisms that can control fungal plant pathogens. These are soil or rhizobacteria including Bacillus amyloliquefaciens, Bacillus velezensis, Bacillus subtilis, Bacillus nakamurai, Bacillus siamensis, and the no less effective Bacillus licheniformis [6]. The following plant protection preparations containing Bacillus strains have been commercially available to date: RhizoVital® and RhizoPlus® (based on B. velezensis FZB42 and produced by ABiTEP GmbH, Germany), Amylo-X® WG (B. amyloliquefaciens subsp. plantarum D747 of Certis Europe BV, The Netherlands), Sonata® (containing Bacillus pumilus QST2808 and manufactured by AgraQuest Inc., Davis, CA), Taegro® (containing B. velezensis FZB24 and produced by Novozymes Biologicals, Inc., Salem, VA). Other strains of the B. velezensis group, which are also protected by patents, are Bacillus methylotrophicus strain CECT8662 (patent ES2639375B1), B. amyloliquefaciens subsp. plantarum BS89 (patent BG67257B1), B. velezensis NRRL B 50150 (US20100008893A1) and B. velezensis RTI301 (US20180020676A1).

The fungicidal effect exerted by Bacillus spp. is based on the synthesis of unique secondary metabolites: lipopeptides and polyketides with a molecular weight between 1.6 and 10 kDa [7,8]. Some of them are cyclic peptides, others are macrolactones of an amphiphilic and surface-active nature, for example, bacillomycin, iturin, fengycin, plipastatin, and mycosubtilin. Iturins, fengycins, and surfactins are lipopeptide families known for their proven antifungal effect. Their chemical structure shows that fengycin and plipastatin are lipodecapeptides containing a lactone ring in their 𝛽-hydroxy fatty acid chain with 14 to 17 carbon atoms. Surfactin and lichenysin are heptapeptides with the chiral sequence LLDLLDL linked with 𝛽-hydroxy fatty acid with a chain length of 12 to 16 carbon atoms. The difference between lichenysin A and surfactin is in the first and seventh amino acids of the cyclic peptide, where lichenysin contains glutamine (Gln) instead of glutamate (Glu), and isoleucine (Ile) instead of leucine (Leu). These compounds are synthesized non-ribosomally, and the key enzymes in their formation are the nonribosomal peptide synthetases (NRPS).

In all cases, the mechanism of action of lipopeptides against fungal cells involves the insertion of the hydrophobic lipid tail into the cytoplasmic membrane of the fungal hypha, autoaggregation, perforation of the cell membrane, and leakage of cytoplasmic contents. These properties are especially valuable in the fight against oomycetes of the genus Phytophthora infestans, destructive pathogens that cause rot disease, wilting of plants, and brown necrosis of vascular tissues, sometimes resulting in total loss of the crop [9]. Another harrowing pathogen for various plants is the ascomycete necrotrophic fungus Botrytis cinerea, the cause of gray mold, an infectious disease known on all continents and in all temperature zones, leading to 80% crop loss [10]. Other mold species that affect the Solanaceae family are the genera Alternaria and Neocosmospora [11]. Regarding bacterial pathogens, the most significant agricultural losses are caused by Cupriavidus necator (Ralstonia eutropha) and Ralstonia solanacearum (Smith 1896), also known as Pseudomonas solanacearum or Burkholderia solanacearum. These bacteria are the cause of brown rot and bacterial wilt, which affect more than 250 plant species from 50 families [12].

The aim of the present work was the isolation and molecular characterization of new promising strains with application as biofungicides against multiple phytopathogens. A specific goal was also the development of a rapid method for detection of polyketide synthase genes and demonstration of the in vivo production of the relevant fungicidal metabolites.

Materials and methods

Soil source, media, isolation, and storage of the strains

Some of the samples were taken from soil, including the rhizosphere, from remote areas in Bulgaria. Three samples were a gift from the Bulgarian Antarctic Expedition made in 2015 and contained soil (grass rhizosphere) from Livingstone Island, Antarctica. The isolation method of microorganisms belonging to Bacillus spp. was previously described [13]. The strains were maintained on slant Nutrient Broth agar at 4°С or as a frozen liquid culture supplemented with 20% glycerol at −70 °C. The strain B. velezensis R22 was deposited in the National Bank for Industrial Microorganisms and Cell Cultures (NBIMCC) under registration no. 9096.

Test microorganisms

Fungal pathogens Botrytis cinerea ATCC^®28985 (NBIMCC 120), Phytophthora infestans ATCC^®36608, Alternaria alternata NBIMCC 110, Aspergillus fumigatus NBIMCC 1415, Aspergillus niger NBIMCC 3350, and Neocosmospora keratoplastica CBS 490.63 were purchased from the respective microbial culture collections. Bacterial plant pathogen R. solanacearum 2436 was purchased from NBIMCC.

Antimicrobial activity assay

The strains were cultured for 48 h in Nutrients Broth, after which the biomass was separated by centrifugation and the cell-free supernatants were purified through a syringe filter (45 μm pore size) before the tests. The activity against B. cinerea and R. solanacearum was assayed by the agar diffusion method [14]. Antifungal activity against P. infestans was tested by two different methods, the resazurin-based microtiter assay [15] and a giant colony-growth observation assay [16]. Briefly, the test substance was included in potato-glucose agar at a final concentration of 1%, and a piece of 7 days fungal mycelium (9 mm in diameter) was placed in the center of the petri dish and incubated at a temperature of 28° C for 7 days. The diameters of the test control colonies were measured at certain time intervals and the antifungal index in % was calculated by the following formula: Antifungal index = (1 − Da/Db) × 100, where Da is the diameter of the colony in the test and Db is the diameter of the colony in the control.

DNA isolation, PCR, and sequencing

Total DNA was isolated from 24-h pure cultures using GeneMATRIX Bacterial and Yeast Genomic DNA Purification Kit (EURx, Gdansk, Poland). Strains’ identification was done after polymerase chain reaction (PCR) with universal primers targeting the bacterial 16S rRNA gene (27F and 1492R). PCR products were separated in 1% agarose (Alfa Aesar GmbH & Co KG, Karlsruhe, Germany), dissolved in 0.5 × Tris-Borate-EDTA Buffer (Sigma-Aldrich Chemie GmbH, Darmstadt, Germany). The gel electrophoresis was carried out using a Midi Horizontal Electrophoresis System (Thistle Scientific Ltd, Rugby, UK) set at 100 V for 30 min.

PCR amplification of 16S rRNA gene fragments was done using universal primer pair 27F and 1492R. The obtained fragments were sequenced by Macrogen Inc. (Amsterdam, Netherlands). The 16S rRNA gene sequences of the isolates were compared with those of the type strains through maximum-likelihood methods with 1000 bootstrap replications using MEGA version 6.0.

PCR amplification of the fragments of lipopeptide-synthetase genes was carried out using the primer pairs designed for this study and listed in Table 1.

Table 1. Primers used for the amplification and sequencing of gene fragments from nonribosomal peptide synthetases.

Primer	Sequence (5’-3’)	PCR product	Position	T, °C
fenA_F fenA_R	Gcggtcaagctttccgctcggtgtgac ctgtttatggcttcgttatacatatgcatcagc	2103 bp	4815 − 4842^a 6879 − 6912^a	63.0
sur_F sur_R	ggctgagagcatcgcaccggcggccgagcg Aacggatggtgcttcagctcaattccttcc	1110 bp	6263 − 6293^a 7343 − 7373^a	55.8
lic_F lic_R	cgatagtgatggagcggtcagagcgggtca Cttgttcagctccgagcacttcctgccata	1440 bp	1540 − 1570^b 2950 − 2980^b	60.0

T, °C is the used temperature of annealing.

^a Position in the genome of B. velezensis FZB42 (Region 1,961,346–1,963,824; GenBank acc. no. CP000560).

^b Position in the genome of B. licheniformis ATCC 14580 (Region 370,023–408,881; GenBank acc. no. AE017333.1).

PCR experiments were carried out in 25 μL reaction volume containing 2× TaKaRa Taq Version 2.0 mix (Clontech Laboratories, Inc., A Takara Bio Company, Mountain View, CA), 50 ng DNA template, and 0.4 μmol/L primers, for 35 cycles in a QB-96 Satellite Gradient Thermal Cycler (LKB Vertriebs GmbH, Vienna, Austria) under the following temperature profile: initial denaturation, 94 °C for 3 min; denaturation, annealing, and elongation (10 s at 98 °C, 40 s at 55.8-63 °C, and 2 min at 72 °C, respectively); final elongation at 72 °C for 5 min. The optimal temperature of annealing was determined by gradient PCR.

WGS and bioinformatic analyses

De novo sequenced genome of B. velezensis strain R22 was registered in NCBI GenBank under the following WGS accession: JASUZW000000000.1, BioSample SAMN35683875. All procedures on genome library preparation, whole genome sequencing, and genome annotation were described by Petrova et al. [17]. AntiSMASH 7.0 platform predicted secondary metabolites synthesis [18].

Preparation of lipopeptide extract fromB. velezensis strain R22

The cell-free active supernatant of B. velezensis strain R22 was acidified to pH 2 with 6 mol/L HCl and after incubation at 4 °C overnight, the sample was centrifuged at 14,000g for 30 min for lipopeptide precipitation. Lipopeptides were extracted from the pellet by methanol (≥99.9%, LC-MS Chromasol^TM, Honeywell, Germany), and insoluble impurities were removed by centrifugation at 10,000g. Methanol was removed under vacuum at 40 °C as described by Pathak et al. [19]. The obtained lipopeptide extract was dissolved in 70% acetonitrile containing 0.1% formic acid (≥98.0%, Sigma-Aldrich Chemie GmbH) and filtered through a syringe filter (0.2 μm).

LC-ESI-MS and LC-ESI-MS/MS experiments

LC-ESI-MS experiments were performed on a UHPLC-Q-TOF mass spectrometer incorporating a Bruker Impact HD Q-TOF system equipped with a CaptiveSpray nanoBooster ionization source and Compass 1.7.5 software. The LC gradient of water-acetonitrile (with 0.1% formic acid) for 30 min was used, which was adjusted as follows: 0–1.5 min 70% acetonitrile; 1.5–20 min, 70% to 85% acetonitrile; 20–23 min from 85% to 100% acetonitrile, 23–25 min to 100% acetonitrile; 25–27 min from 100% to 70% acetonitrile, 27–30 min. A total of 70% acetonitrile, at a flow rate of 0.1 mL/min. The separation was performed on a C18 RP analytical column (2.1 × 100 mm, 2.7 μm particle size).

Mass spectra were obtained at a 1 Hz scan rate. For automatic MS/MS, the collision frequency was increased from 800 to 400 V p-p. Baseline CE values for the precursor, with an m/z value in this range, were automatically calculated from these values linearly. The source temperature and voltage were set at 525 K and 4500 V, respectively. The dry gas flow and nebulizer pressure were set at 8.0 L/min and 27 psi, respectively. The eluted metabolites were detected at λ 226 nm by high-precision mass scanning in the 150–3500 m/z range, in positive ion mode. Ten-microliter aliquots from the prepared sample were injected into the C18 RP analytical column. The mixture containing standards of surfactin (with purity ≥98%, Sigma-Aldrich Co, St. Louis, MO, made in Israel) and fengycin (with purity ≥90%, Sigma-Aldrich Co, made in France) in a 1:1 ratio was prepared in a concentration of 4 µg/mL. LC-ESI-MS analysis of this mixture was conducted under the same experimental conditions as those for the lipopeptide extract from B. velezensis strain R22.

Data analysis

Data are presented as mean values from two independent fermentations of B. velezensis strain R22 and two independent lipopeptide extraction experiments. The ESI-MS data were analyzed by Data Analysis software ver. 4.0 (Bruker Daltonics, Bremen, Germany).

Results

Taxonomic position of the Bacillus spp. isolates with antifungal activity

From different geographical areas and different habitats, 120 strains with morphological and physiological characteristics of the genus Bacillus were isolated. Rod-shaped, gram-positive, catalase-positive bacteria capable of spore formation and with typical bacilli colony morphology were selected. The initial screening for strains with antifungal activity was performed as the isolates were cultivated for 48 h, and the cell-free supernatants were tested for activity against Botrytis cinerea ATCC^®28985. Then, 14 strains selected as the most active were identified at species level by 16S rRNA gene sequencing (Figure 1).

Figure 1. Phylogenetic tree of Bacillus spp. based on 16S rRNA sequences through maximum-likelihood methods with 1000 bootstrap replications using MEGA version 6.0. Escherichia coli BL21(DE3) (AM946981.2) was used as an outgroup in the analysis. NCBI GenBank accession numbers used in the comparison reference strains are the following: B. velezensis FZB42, OR485707; B. amyloliquefaciens ATCC 23350, NR_118950; B. licheniformis ATCC 14580, NC_006270.3; B. aerius MN-1, MN252912; B. safensis ATCC BAA-1126^T, AF234854; B. aryabhattai B8W22, NZ_JYOO01000023. NCBI acc. no. of the newly isolated strains are OR482395, OR482394, OR482396, OR482397, OP554433, MK461937, MK461933, MK461938, MK461934, MK461936, OR482398, OR482400, OR482399, MK461947.

The strains possessing antifungal activity clustered into two main groups: B. amyloliquefaciens/B. velezensis and B. licheniformis group (Figure 1). These two bacterial groups were the most abundant in both Bulgarian and Antarctic rhizosphere samples. The nucleotide sequences of 16S rRNA of the strains R7, R19, R22, and R23 were 100% identical to those of B. velezensis strain FZB42 (OR485707), formerly known as B. amyloliquefaciens subsp. plantarum type strain. The species affiliation of B. velezensis R22 was confirmed also by whole-genome sequence analyses. The genome similarity estimated by Average Nucleotide Identity showed that the genome of B. velezensis UA0241 (NZ_CP097598.1) was the closest with 98.92%. In silico obtained DNA-DNA hybridization value with the genome of the type strain of B. velezensis KACC 13105T (CP014838) was 90.1%. The strain R10 was determined as B. amyloliquefaciens due to the sequence identity with the type strain ATCC 23350 (NR_118950). Five of the isolates (55-1, 13, 24, 16-1, and 39) belonged to the species Bacillus lichenifomis with 100% identity of 16S rRNA with the type strain ATCC 14580. Bacillus aerius, Bacillus aryabhattai, and Bacillus safensis representatives were also isolated; however, they were not among the investigated rhizosphere samples.

Antifungal activity

Since the fungus B. cinerea is devastating to the crops of many edible plants, the initial screening of the strains was performed for activity against this plant pathogen. Analyzed by the well-diffusion method, 14 strains showed the highest activity against B. cinerea. All B. velezensis strains, B. aryabhattai 17M, B. aerius 1p/11 and 22M, and B. licheniformis 39 were able to inhibit the development of this fungus significantly (Table 2, Supplemental Figure S1).

Table 2. Inhibition of Botrytis cinerea growth by cell-free supernatant of Bacillus spp. cultures obtained after 48 h of cultivation.

Species	Strain	Inhibition zone (mm)
B. velezensis	R22	34 ± 3
B. aryabhattai	17М	29 ± 3
B. velezensis	R23	25 ± 3
B. aerius	1p/11	25 ± 3
B. velezensis	R19	25 ± 2
B. amyloliquefaciens	R10	25 ± 2
B. velezensis	R7	24 ± 2
B. licheniformis	39	24 ± 2
B. licheniformis	24	21 ± 2
B. licheniformis	13	21 ± 2
B. aerius	22M	20 ± 2
B. licheniformis	55-1	20 ± 2
B. licheniformis	16-1	19 ± 1
B. safensis	14-A	16 ± 1
Control	Nutrient broth	0

Antifungal activity assay against P. infestans (performed by the resazurin microtiter method) showed that the cell-free supernatants of the strains can substantially inhibit spore formation by this phytopathogenic fungus (Table 3).

Table 3. Inhibition of Phytophthora infestans growth by the cell-free supernatants of 48-h Bacillus spp. cultures.

Species	Strain	Inhibition (%)
		24 h	48 h
B. licheniformis	39	93.64	91.8
B. velezensis	R22	88.06	86.02
B. aerius	22M	87.03	84.89
B. aerius	1p/11	88.28	84.25
B. velezensis	R7	87.16	84.64
B. safensis	14-A	87.28	83.87
B. aryabhattai	17М	85.65	81.46
B. amyloliquefaciens	R10	84.65	80.58
B. licheniformis	13	84.19	79.32
B. velezensis	R23	81.45	76.42
B. velezensis	R19	80.69	76.01
B. licheniformis	55-1	78.18	75.05
B. licheniformis	24	75.94	71.63
B. licheniformis	16-1	74.31	70.24
Control	Nutrient broth	–	–

Eight strains, B. licheniformis 39, B. velezensis R22 and R7, B. aerius 22M and 1p/11, B. safensis 14-A, B. aryabhattai 22M, and B. amyloliquefaciens R10, inhibited the growth by more than 80%. The antifungal activity of the most active strains, B. velezensis R22 and B. licheniformis 39 was also evaluated by a second technique, the antifungal growth index of a giant colony of the pathogen P. infestans in the presence and absence of supernatant of the strains (Table 4). The comparison showed that the strains had a very close antifungal index and it was very high: over 80% after 7 days of cultivation with the phytopathogen.

Table 4. Comparison of the antifungal activity against Phytophthora infestans displayed by B. velezensis R22 (BV R22) and B. licheniformis 39 (BL 39).

Strain	24 h		48 h		72 h		7 days
	D (cm)	AI (%)	D (cm)	AI (%)	D (cm)	AI (%)	D (cm)	AI (%)
BV R22	0.9 ± 0.1	43.8	1.5 ± 0.1	63.4	1.5 ± 0.1	76.9	1.5 ± 0.1	82.35
BL 39	0.9 ± 0.1	43.8	1.7 ± 0.1	58.5	1.6	75.4	1.6 ± 0.1	81.18
Control	1.6 ± 0.1	–	4.1 ± 0.1	–	6.5 ± 0.2	–	8.5 ± 0.1	–

D, colony diameter; AI, Antifungal index = (1 − D_sample/D_control) × 100.

B. velezensis R22 showed the highest antifungal activity against both B. cinerea and P. infestans, therefore, its fungicidal spectrum against other pathogenic fungi was also investigated. In a co-inoculation experiment performed in 100 mm Petri dishes, B. velezensis R22 was able to limit the growth of the phytopathogens A. alternata, A. fumigatus, A. niger, and N. keratoplastica for 7 days (Figure 2). R22 did not show any activity against fungi of the genus Mucor.

Figure 2. Antifungal activity of B. velezensis R22 against phytopathogenic fungi. (A) Alternaria alternata; (B) Aspergillus fumigatus; (C) Aspergillus niger; (D) Neocosmospora keratoplastica. The co-inoculation experiments were performed in 100 mm Petri dishes for 7 days.

Antibacterial activity

Regarding bacterial plant pathogens, serious crop losses of the Solanaceae family (potatoes, tomatoes, eggplants, pepper) are caused by C. necator (R. eutropha) and R. solanacearum, also known as P. solanacearum or B. solanacearum. To elucidate the potential of the newly isolated Bacillus strains to control bacterial infections in addition to fungal ones, the fourteen strains active against fungi were tested for activity against R. solanacearum (Smith 1896). Notably, when tested by the agar-diffusion method, five strains possessed high activity, reaching more than 2 cm sterile halo. However, as with antifungal activity, there was strain specificity in the degree of inhibition and, despite being of the same species, different B. velezensis strains showed different activity. R22 was the most active against R. solanacearum with approximately 3 cm bactericidal zone (Table 5, Supplemental Figure S2).

Table 5. Inhibition of R. solanacearum growth by cell-free supernatant of Bacillus spp. cultures obtained after 48 h of cultivation.

Species	Strain	Inhibition zone (mm)
B. velezensis	R22	29 ± 3
B. velezensis	R23	25 ± 3
B. velezensis	R19	25 ± 2
B. velezensis	R7	20 ± 2
B. licheniformis	39	20 ± 2
Control	Nutrient broth	–

Molecular screening for NRPS genes in the genomes of the newly isolated Bacillus spp. strains

To elucidate the genetic basis, genome mining of the respective species was performed, and primers specific for fengycin, surfactin, and lichenysin were generated (Table 1). The operon encoding fengycin synthetase is about 38 kb in length and consists of five open reading frames for fenC, fenD, fenE, fenB, and fenA; the last gene was targeted by the designed primers. The obtained PCR products with the expected size of 2103 bp confirmed the presence of fenA. Five of the strains yielded the proper PCR product (Figure 3).

Figure 3. PCR amplification of fenA gene encoding fengycin synthetase A. Designations: M, molecular weight marker (Perfect Plus^TM 1 kb DNA Ladder, EURx); 1, B. velezensis R7; 2, B. amyloliquefaciens R10; 3, B. velezensis R19; 4, B. velezensis R22; 5, B. velezensis R23; 6, negative control (E. coli DH5α).

The fenA sequences of B. velezensis R7, R10, R19, and R23 were deposited in the GenBank under accession nos. OR568612, OR568613, OR568614 and OR568615, respectively. They have 99% identity with fenA genes of B. velezensis isolates (CP053717.1, CP055160.1, CP040881.1), and 92% identity with fenD of the type strain of B. amyloliquefaciens DSM7 (FN597644).

A PCR product with a similar size was obtained with the fen_F/fen_R primer pair and DNA of B. licheniformis 16-1. However, sequencing of the PCR product (OR568616) showed 97% identity with the ppsA gene encoding non-ribosomal plipastatin synthetase PpsA in B. subtilis strain RS10 (CP046860).

The newly isolated strains were also screened for the presence of genes for the surfactin synthetase and lichenysin synthase (Figure 4).

Figure 4. PCR amplification of specific fragments of srfAA gene encoding surfactin synthetase A (A) and lchAA gene for lichenysin synthase (B). Lanes and samples: M, molecular weight marker (Perfect Plus^TM 1 kb DNA Ladder, EURx); (A): 1, B. velezensis R7; 2, B. amyloliquefaciens R10; 3, B. velezensis R19; 4, B. velezensis R22; 5, B. velezensis R23; 6, negative control (E. coli DH5α); (B) 1, B. licheniformis 16-1; 2, B. licheniformis 13; 3, B. licheniformis 24; 4, B. licheniformis 39; 5, B. licheniformis 55-1; 6, negative control (E. coli DH5α).

The results showed that the five strains belonging to the species B. velezensis and B. amyloliquefaciens all possessed the surfAA gene in their genomes, which was proved by sequencing of the 1110 bp PCR fragments (OR591497, OR591498, OR591499, OR591500, and OR591501, for R7, R10, R19, R22, and R23, respectively). Nucleotide sequences possessed between 98.8% and 99.1% identity with surfAA genes found in the chromosome of B. amyloliquefaciens R8-25 (CP054479.1), and those of B. velezensis isolates SWUJ1 (CP077672.1), K203 (CP092185.1), YJ0-1 (CP128184.1).

The entire lchAA gene encoding the enzyme lichenysin synthase is 10,746 bp in size and encodes a protein of 3,582 amino acids; we targeted to amplify a 1440 bp fragment of it. PCR fragments suggesting the presence of lchAA were obtained with genomic DNA of B. licheniformis strains 13, 16-1, 24, 39, and 55-1, and the sequenced fragments were deposited in the GenBank under the following accession numbers: OR591502 (strain 16-1), OR591503 (strain 13), OR591504 (strain 55-1), OR591505 (strain 39), and OR591506 (strain 24). Although the PCR product (1440 bp) from B. licheniformis 16-1 was obtained in a limited amount, its nucleotide sequence showed 100% identity with lchAA of B. licheniformis ATCC 14580 (CP034569.1), as did the others.

WGS of B. velezensis R22 and determination of clusters for the synthesis of secondary metabolites with antimicrobial activity

Exhibiting high antifungal activity against a wide range of phytopathogens, B. velezensis R22 was subjected to whole-genome sequencing. Its genome was assembled into 14 contigs, being a circular chromosome, 4,081,504 bp in size (with G + C content of 46.35%) without plasmids, and containing 4087 genes that encode 3935 proteins, 72 tRNAs, 14 rRNAs, and 5 ncRNAs. AntiSMASH 7.0 platform predicted the synthesis of eleven secondary metabolites (Table 6).

Table 6. Secondary metabolites predicted to be synthesized by B. velezensis R22 using AntiSMASH 7.0 platform.

Contig^a	Cluster	Region	Start (5′-3′)	End (5′-3′)	Compound	Type	Similarity (%)
1	1	1.1	543594	635972	Difficidin	Polyketide	100%
		1.2	809338	848988
		1.3	901349	921477
1	2	1.4	988641	1123173	Fengycin	Lipopeptide	100%
1	3	1.5	1193075	1293617	Bacillaene	Polyketide/lipopeptide	100%
1	4	1.6	1512777	1600983	Macrolactin Н	Polyketide	100%
1	5	1.7	1922542	1943282	Butyrosin А/В	Saccharide	7%
2	6	2.1	142895	194687	Bacillibactin	Lipopeptide	100%
2	7	2.2	720547	761965	Bacilysin	–	100%
3	8	3.1	191350	256757	Surfactin	Lipopeptide	82%
4	9	4.1	143567	194775	Macrolactin R22	Polyketide	40%
5		5.1	46224	58365	Phosphonate	–
5	10	5.2	86835	116957	Velezensin	Polyketide	–

^a Contigs’ numbering is according to the genome of B. velezensis R22 deposited in NCBI GenBank (accession no. JASUZW000000000.1, BioSample SAMN35683875).

Importantly, Contigs 1 to 5 (where all eight complete clusters encoding polyketides/lipopeptides synthesis fall) showed over 99% similarity to the chromosome of B. velezensis strains YJ11-1-4 (CP011347.1) and GYL4 (CP020874.1), both known for their strong fungicidal activity. The R22 genome contains also 218 genes related to carbohydrate metabolism, including eight genes encoding glycoside hydrolases (amyE, malL, sacA, xynA, xynB, xynD, xynC, and eglS), enabling the strain to convert plant polysaccharides such as cellulose, hemicellulose, and starch, as do many other B. velezensis strains [20]. The possible plant-growth-promoting action can be helped by the synthesis of the volatile compounds 2,3-butanediol and acetoin, as the genome of B. velezensis R22 contains the genes ilvB, alsS, and ilvH (for acetolactate synthase), alsD (acetolactate decarboxylase), and bdhA encoding (R, R)-2,3-butanediol dehydrogenase.

Lipopeptide extract from B. velezensis R22 and its characterization by mass spectrometry

LC-MS and LC-MS/MS analyses by a UHPLC-Q-TOF mass spectrometer proved the presence of several lipopeptides produced by B. velezensis R22 and found in the cell-free supernatant. The data from LC-ESI-MS analyses showed that ions with the highest intensities were determined as [M + H]⁺ at m/z 1036.698 and m/z 1058.679 (Figure 5A, B) and as the series double charged ions [M + 2H]²⁺ at m/z 732.405, 739.412, 746.419, and 753.427, presented also as [M + H]⁺ ions at m/z 1463.804, 1477.818, 1491.832, and 1505.849 (Figure 5A, C). These results are in good accordance with data from LC-ESI-MS analyses of the mixture containing standards of surfactin and fengycin conducted under the same experimental conditions (Supplemental Figure S3). That way the major active compounds in the lipopeptide extract from B. velezensis R22 were identified as surfactin with molecular mass of 1035.698 and 1057.7 Da eluted between 14.7 and 16.3 min and fengycin with molecular masses 1462.8, 1476.8, 1490.8, and 1504.8 Da eluted between 4.4 and 5.9 min.

Figure 5. Analyses of lipopeptides of B. velezensis strain R22 by UHPLC-Q-TOF mass spectrometer. (A) The total ion chromatogram of 10 μL lipopeptides extracted from the cell-free supenatant obtained after B. velezensis strain R22 growth, identified at λ 226 nm in the range m/z 150–3500; (B) LC-MS of a peak eluted between 14.7 and 16.3 min, predominant [M + H]⁺ ion at m/z 1036.698 corresponding to surfactin; (C) LC-MS of peak eluted between 4.4 and 5.9 min dominated by double charged ions [M + 2H]²⁺ at m/z 732.405, 739.412, 746.419, and 753.427 corresponding to [M + H]⁺ ions at m/z 1463.804, 1477.818, 1491.832, and 1505.849, identified as different forms of fengycin.

The LC-MS/MS analysis revealed the structures of the bioactive compounds corresponding to the molecular masses determined by LC-MS analyses. An LC-MS/MS spectrum of precursor ion [M + H]⁺ at m/z 1036.698, corresponding to surfactin containing a C15 β-hydroxy fatty acid chain, is shown in Figure 6. The determined series of fragment ions at m/z 923.613 (b6, C15), 810.527(b 5, C15), 792.517 (b5-H₂O, C15), 695.499 (b4, C15), 677.454 (b4-H₂O, C15) and 370.259 (b3, C15) and 352.248 (b3-H₂O, C15), revealed the loss of Leu-Leu-Asp-Val from the C terminus (Figure 6). Furthermore, a typical set of y fragment ions at m/z 685.454 (y6), 572.365 (y5), 459.282 (y4), 360.212 (y3) suggested a loss of Leu-Leu-Val-Asp in the middle of the peptide chain. The peak at m/z 810.527 is formed by the loss of Leu-Leu from the molecule, which was confirmed by the peak at m/z 792.517 (b5-H₂O, C15). This is a result of the loss of Leu-Leu-OH₂, and the formation of an ester bond between the carboxyl group of Leu and the hydroxyl group of the aliphatic moiety in the surfactin molecule. Thus, the precursor ion at m/z 1036.698 corresponded to surfactin containing a C15 β-OH fatty acid whose peptide sequence was Glu-Leu-Leu-Val-Asp-Leu-Leu.

Figure 6. LC-MS/MS spectrum of the surfactin precursor [M + H]⁺ at m/z 1036.698, containing a C15 β-hydroxy fatty acid chain with interpretation.

The analysis of doubly charged ions [M + 2H]²⁺ at m/z 732.405, m/z 739.412, m/z 746.419, and m/z 753.427 by LC-MS/MS led to the identification of different forms of fengycin in the lipopeptide extract from B. velezensis R22 (Figure 7). The ESI-MS/MS spectra of [M + 2H]²⁺ ions at m/z 734.405 and m/z 739.412 (Figure 7A, B) showed typical fragmentation ions of m/z 966 and 1080, which in this case were also detected as double-charged ions [M + 2H]²⁺ at m/z 540,769 and m/z 483.732. This is a result of the loss of β-hydroxy fatty acid-Glu and β-hydroxy fatty acid-Glu-Orn moieties, respectively, from the N-terminal segment with an Ala residue at position 6 of the fengycin A cyclic decapeptide. Similar results from mass spectrometric analyses by MALDI-TOF-MS/MS were shown in the study of Yang et al. [21]. The interpretation of MS and MS/MS spectra of ions [M + 2H]²⁺ at m/z 732.405 and m/z 739.412 (respectively [M + H]⁺ ions at m/z 1463.804 and 1477.818) showed a 14 Da (CH2–) difference in the molecular mass, while the peptide sequences were the same. Therefore, two homologous forms of fengycin A with the same peptide sequence but different C16 and C17 β-OH fatty acids, respectively, were established. The profile of the LC-MS/MS spectra shown in Figure 7C, D reveals typical fragmentation ions at m/z 994 and m/z 1108, which are the product of lost β-hydroxy fatty acid-Glu and β-hydroxy fatty acid-Glu-Orn moieties, respectively, from the N-terminal segment with a Val residue at position 6 and C17. The characteristic fragment ion at m/z 1108 was deduced in both cases as a double-charged ion at m/z 554.78 with high intensity. A similar approach to prove different forms of fengycin by mass spectrometry was used by Zhang et al. [22]. Overall, the data from the LC-MS and LC-MS/MS spectra confirm the presence of surfactin with molecular mass 1035.7 Da (precursor [M + H]⁺ at m/z 1036.698) and C15 β-OH fatty acid chain. Four different fengycin compounds were detected: fengycin A with molecular masses 1462.8 and 1476.8 Da (with C16 and C17 β-OH fatty acid chains), and fengycin B with molecular masses 1490.8 and 1504.8 Da, with C16 and C17 β-OH fatty acid chains, respectively.

Figure 7. LC-MS/MS spectra of different forms of fengycin: (A) the precursor [M + 2H]²⁺ at m/z 732.405, containing Ala of position 6 and C16 β-hydroxy fatty acid chain; (B) the precursor [M + 2H]²⁺ at m/z 739.412, containing Ala of position 6 and C17 β-hydroxy fatty acid chain; (C) the precursor [M + 2H]²⁺ at m/z 746.419, containing Val of position 6 and C16 β-hydroxy fatty acid chain; (D) the precursor [M + 2H]²⁺ at m/z 753.427, containing Val of position 6 and C17 β-hydroxy fatty acid chain.

A summary of the results for lipopeptides proven in the extract obtained from B. velezensis strain R22 cell-free cultural broth is presented in Table 7.

Table 7. Lipopeptides established by mass spectrometry in cell-free fermentation broth (48-h culture) of B. velezensis strain R22.

No	[M + H]^+ (m/z)	Intensity	[M + 2H]^2^+ (m/z)	Lipopeptide family	Type	Sequence
1	1036.698	6 × 10^6	518.849	Surfactin	C15, Leu7
2	1463.804	3 × 10^5	732.405	Fengycin A	C16, Ala6
3	1477.804	2.5 × 10^5	739.412	Fengycin A	C17, Ala6
4	1491.832	1.5 × 10^5	746.419	Fengycin B	C16, Val6
5	1505.849	1 × 10^5	753.427	Fengycin B	C17, Val6

Discussion

The search for new rhizobacterial strains that exhibit antifungal activity and improve the growth of crops is considered a ‘green’ alternative to the chemical fungicides used so far, being a sustainable and environmentally friendly solution to the pressing problem of plant diseases [23,24]. About 15% of the root surface is covered by a microbial population, of which the species belonging to the genus Bacillus are the most important because of their symbiotic relationship with the plant [25]. Plant photosynthetic products (sugars) are secreted by the roots and are used by bacteria as a substrate. Rhizobacteria, in turn, stimulate the transport of mineral nutrients and their absorption by plant roots, improve growth through the synthesis of substances with antimicrobial action and volatile compounds, activate plant systemic resistance to diseases, and mitigate environmental stresses [26–28]. In many cases, a combination of Bacillus strains that act synergistically against pathogens is used for plant protection. For example, US patent US6589524B1 reveals that a combination of strains B. cereus NRRL B-30517 and NRRL B-30519, B. amyloliquefaciens NRRL B-30518 and B. subtilis NRRL B-30520, act synergistically when included in the composition of formulae targeted against fungi of the genus Phytophthora spp. [29]. The highest fungicide activity is reported for B. velezensis strains. B. velezensis SDTB038 is known to inhibit P. infestans in potatoes [30], KOF 112 in grapes [31], and QSE-21 in harvested tomatoes and apples [32]. B. velezensis strain JRX-YG39 [33] inhibits the mycelia growth of B. cinerea and induces the defense of Arabidopsis thaliana and grapes.

Here, along with the high activity of B. velezensis R22, we describe antagonism against the combination of three phytopathogens: B. cinerea, P. infestans, and the bacterium R. solanacearum, which is an exceptional property that has not been reported so far. In the present work, the selected strains with fungicidal activity confirmed the enormous potential of the B. amyloliquefaciens/B. velezensis group for the development of biopesticide formulae. Moreover, several strains with antifungal activity from other Bacillus spp. were also isolated (Figure 1): B. aryabhattai, B. aerius and B. safensis; as their molecular characterization revealed their potential as biofungicides for the first time. B. velezensis strains carry genes encoding the synthesis of fengycin and surfactin, while all studied representatives of B. licheniformis contained a gene for lichenisyn synthase, a key enzyme in the synthesis of lichenisyn.

It has to be noted that besides the ability to produce a large number of antimicrobials, Bacillus spp. are particularly promising in biopesticide formulations for several additional reasons: (1) they form endospores that are resistant to external harsh conditions, and thus amenable to successful spreading in the fields; (2) most Bacillus species are safe for humans and animals and have a generally recognized as safe status; and (3) bacilli are unpretentious to cultivation conditions and assimilate a wide spectrum of inexpensive substrates including sugars of cellulose and lignocellulose content [34,35]. The last property is particularly important in the development of profitable biotechnology.

B. velezensis R22 should be singled out as particularly promising for prospective biopesticide production. In addition to its antifungal activity, the strain also exhibits impressive antibacterial activity. The WGS analysis and the following annotation revealed that a large part (∼340 kb) of the genome of B. velezensis R22 is involved in the non-ribosomal synthesis of lipopeptide and polyketide antimicrobial molecules, siderophores, and bacteriocins. A set of ten giant gene clusters is responsible for the synthesis of a wide range of bioactive secondary metabolites through modularly organized mega-enzymes known as non-ribosomal peptide synthetases and polyketide synthases. Part of the gene clusters are involved in the synthesis of cyclic lipopeptide molecules such as fengycin, surfactin and the iron siderophore bacillibactin. Three other gene clusters (mln, bae, and dfn, 199 kb) are responsible for the synthesis of antibacterial polyketides, such as macrolactin, bacillaene, and difficidin, and the bac cluster (6.9 kb) encodes genes for synthesis of the antibacterial dipeptide bacilysin.

Although the genome of B. velezensis R22 is generally similar to the genomes of other strains of the species, it also contains some unique regions responsible for the synthesis of antimicrobial metabolites. Two gene clusters encoding a new lipopeptide and a new polyketide antibiotic were found in the genome. Gene cluster 9 has some similarity (40–44%) to the genes encoding the polyketide macrolactin H in B. amyloliquefaciens subsp. plantarum FZB4 and the halophilic bacterium Jeotgalibacillus marinus, but also has 17% homology with the genes encoding myxovirescin in Myxococcus xanthus. We therefore consider that cluster 9 encodes a novel polyketide. Cluster 10 shows 34% similarity to the genes responsible for the synthesis of the polyketide pulvomycin in the genus Streptomyces, 35% similarity to mycalamide encoded by an assembled marine microbiome, 27% homology to the calimantacin genes of Pseudomonas fluorescens and 25% similarity to the aurantinine genes of B. subtilis. The presence of this gene cluster of 30,123 nucleotide pairs is a unique feature of B. velezensis R22 and its most distinguished feature. Similar genes (for new antimicrobial compounds not previously discovered in bacilli genomes) are evidence of possible horizontal genetic transfer between different rhizobacterial genera inhabiting the same ecological niche and the possibility of regrouping genes into clusters with new traits [36]. The described in silico prediction of gene clusters encoding secondary metabolites such as lipopeptides and polyketides in the genome of B. velezensis R22, together with experimental data showing its antagonistic action against different fungal species (Oomycetes and Ascomycetes), suggest that this strain may have biotechnological potential. Several clusters found in the strain R22 (for fengycin, bacillibactin, macrolactin, surfactin, and bacilysin) share similarities with those present in bacterial isolate B. velezensis FZB42 with known antifungal activity [37,38]. To prove whether the putative antifungal metabolites are present in the cell-free fermentation broth of B. velezensis R22, LC-MS and LC-MS/MS experiments on a UHPLC-Q-TOF mass spectrometer were performed. The results demonstrated the presence of five active metabolites with the highest intensity: surfactin with precursor [M + H]⁺ at m/z 1036.698 containing a C15 β-OH fatty acid chain; two homologous forms of cyclic decapeptide fengycin A with an Ala residue in position 6, with C16 and C17 β-OH fatty acids, detected as [M + 2H]²⁺ at m/z 732.405 ([M + H]⁺ at m/z 1463.804) and m/z 739.412 ([M + H]⁺ at m/z 1477.804), respectively, as well as two homologous forms of fengycin B containing C16 and C17 β-hydroxy fatty acid chain ([M + 2H]²⁺ at m/z 746.419 and m/z753.427 and [M + H]⁺ at m/z 1491.832 and m/z 1505.849). The presence of these lipopeptides is the most probable reason for the high antimicrobial activity against phytopathogenic fungi, which agrees with the observations of Wang et al. [39] about fengycin homologs with a higher number of carbon atoms in the side chain fatty acids.

Although there is a huge effort to develop biofungicides worldwide, chemical fungicides will continue to play a vital role in the agricultural industry, as they are associated with already established low-cost chemical production and employment. Therefore, it is important to compare the effectiveness of chemical and biofungicides in terms of minimum inhibitory concentration (MIC) for different phytopathogens. The recent study by Sharma et al. [40] evaluated the effectiveness of chemical fungicides in controlling Phytophthora blight of pigeon peas. MIC was determined for several fungicides and the results showed that the most effective fungicides were the combination of methyram + dimethomorph, which showed the lowest MIC value of 0.5 µg/mL, followed by metalaxyl-M + mancozeb, 35 µg/mL; cimoxanil + mancozeb, 60 µg/mL; and famoxadone + cimoxanil, 140 µg/mL [41]. Another study evidenced the activity of carbendazim, mancozeb, and thiram against gray mold B. cinerea with MIC values of 1, 8, and 4 µg/mL, respectively [42]. For comparison, Wang et al. [39] reported that iturin exhibits activity against B. cinerea at MIC concentration 62.50 μg/mL; C₁₆ fengycin A possesses MIC 31.25 μg/mL, and C₁₈ fengycin A, MIC 0.49 μg/mL. These data indicate that lipopeptides have similar MICs to chemical fungicides. Spontaneous production of lipopeptides in Bacillus spp. supernatants range between 247.3 mg/L (fengycin) [42] and 400 mg/L (iturin) [43]. Therefore, if a profitable biotechnological production can be established, biofungicides have an obvious advantage in the field, since crops can be treated directly with a product obtained from the diluted bacterial culture. Moreover, along with the individual contribution of each of the lipopeptide fungicidal compounds in these formulae, the possibility of a synergistic mode of action exists. In our case, several putatively novel metabolites with antifungal activity synthesized by B. velezensis strain R22 need future confirmation.

Conclusions

New rhizosphere Bacillus spp. with antifungal activity from different geographical areas and habitats were isolated. Considering the geographical location of the sampling, it may be noted that the strains isolated from Bulgaria and Antarctica showed high and similar fungicide activity. B. velezensis R22 and B. licheniformis 39 demonstrated the highest activity against the phytopathogens B. cinerea and P. infestans. A novel, rapid PCR-based approach for detecting genes implicated in the non-ribosomal synthesis of fengycin/plipastatin, surfactin, and lichenisyn was developed. The whole genome sequence of B. velezensis R22 revealed the presence of ten complete lipopeptide synthesis clusters, some of which possessed very low similarity to those known so far. The presence of five lipopeptides was established by highly sensitive mass spectrometry: surfactin with molecular mass 1035.7 Da, two homologous forms of cyclic decapeptide fengycin A with molecular masses 1462.8, and 1476.8 Da, as well as two homologous forms of fengycin B with molecular masses 1490.8 and 1504.8 Da. Further investigation of these strains under field conditions needs to be performed to reveal their actual and full antagonistic potential, thus pointing out exciting prospects for the development of new biofungicides.

Authors’ contribution

Conceptualization, P.P. and K.P.; methodology, M.G., E.K., N.A., D.K., L.V., and P.D.; software, P.P.; validation, M.G.; investigation, N.A., M.G., E.K., D.K., and L.V; resources, K.P.; data curation, K.P.; writing—original draft preparation, P.P., L.V., P.D.; writing—review and editing, K.P., L.V., and P.D.; funding acquisition, P.P. and K.P. All authors have read and agreed to the published version of the manuscript.

Disclaimer statement

The opinions based on the results obtained and expressed in the article are solely those of the authors and do not reflect the opinions or beliefs of the company that provided the funding.

Disclosure statement

The authors declare no conflict of interest.

Data availability

All data concerning nucleotide sequences are available in the NCBI GenBank. De novo sequenced genome of B. velezensis strain R22 has been deposited in DDBJ/ENA/GenBank under accession no. JASUZW000000000.1, BioSample SAMN35683875. The version described in this article is the first version. The Sequence Read Archive is deposited under accession no SRP441980.

Additional information

Funding

This study was supported by AGRIA AG by Contract no. 1/2021.