Antifungal Activity of Bacillus Species Against Fusarium and Analysis of the Potential Mechanisms Used in Biocontrol

1Numerous species of Bacillus have been identified as plant-growth promoting bacteria (PGPB;, ) and/or biocontrol agents (BCA). The most commonly studied PGPB/BCA are B. amyloliquefaciens, B. licheniformis, and B. subtilis. PGPB employ a variety of strategies to enhance plant growth and survival by both direct and indirect mechanisms. The most common direct mechanisms are phytohormone production, the acquisition of nutrients such as phosphorous and nitrogen, and the control of pathogens through various means, for example, through the synthesis of hydrolytic enzymes, antifungal compounds, lipopeptides, or antibiotics. The indirect mechanisms include protection from abiotic stress brought about by drought, salinity, etc., the triggering of specific defense-related pathways, particularly the induction of systemic resistance (ISR) against pathogens and pests; and the release of volatile organic compounds [VOCs; also called bacterial volatile compounds (BVCs; ) as well as the suppression of reactive oxygen species (ROS)]. Numerous articles and reviews about the variety of mechanisms PGPB use to promote plant growth have been published.

2Recently, B. simplex has been added to the growing list of PGPB. Prior to these studies, B. simplex was known mostly for its slope-specific adaptation and incipient sympatric speciation in “Evolution Canyons I and II” in Israel. Earlier, we described several of the traits required for a number of the mechanisms for both plant-growth promotion and biocontrol in B. simplex 30N-5, including evidence for the synthesis of polyamines, the production of siderophores, and the synthesis of antimicrobial peptides and antibiotics. B. simplex 30N-5 also has ACC (1-aminocyclopropane-1-carboxylic acid) deaminase activity, based on the ability of B. simplex 30N-5 to grow on ACC. In this report, we compared B. simplex 30N-5 and B. subtilis 30VD-1, both isolated from the UCLA Mildred E. Mathias Botanical Garden. The other two B. simplex strains, 237 and 11, were isolated from the Negev Desert in Israel. Our earlier study showed that B. simplex 30N-5 inhibits the growth of two members of the F. oxysporum complex and F. solani, formerly Nectria haematococca. However, still little is known about the effectiveness of B. simplex strains in terms of their usefulness as PGPB/BCA.

3Mechanisms for both direct and indirect PGPB functions are described through the use of physiological and biochemical assays, and more recently by in silico analyses. Although not all of the genomes of B. simplex strains are sequenced, the genome sequences of eight are available including those analyzed here: B. simplex strain 30N-5 and also strains II3b11, P558, and BA2H3. Having genome sequences from a diverse set of B. simplex strains from different habitats presents an excellent opportunity to learn more about the potential BCA of this species.

4Fusarium is an ascomycete that causes plant infections known as head blights, vascular wilts, patch diseases, root rots, and yellowing disease. Although Fusarium fungi can invade plants through seeds and wounds, they also enter plants through their roots, especially if root cap border cells are damaged. The disease is managed by crop rotation, planting resistant crops, letting the land lie fallow, and/or applying chemical fungicides. Because fungicides have negative effects on both the environment and human health and, in addition, can promote the development of resistant strains of certain disease-causing fungi, efforts to utilize plant-associated bacteria to manage Fusarium-induced pathogenesis have been pursued. Most of the bacteria studied for biocontrol activity are pseudomonads and species of Bacillus, which use diverse mechanisms for the biocontrol of Fusarium, including siderophore-mediated competition for iron, antibiotic production, and induced systemic resistance.

5In this report, we tested the potential PGP activity of an understudied Bacillus species and a B. subtilis strain for demonstrable functions that correlated with the in silico analysis of plant-growth promotion-related and biocontrol-associated genes. B. simplex 30N-5, 11, and 237, and also B. subtilis 30VD-1, were compared to the well-known PGP bacilli, such as B. pumilus, B. amyloliquefaciens, and B. licheniformis. Employing a comparative approach between in silico studies and a demonstrable chemical product or activity is likely to lead to a more rational choice of probiotic microbes for promoting plant growth and inhibiting disease.

6We compared in vitro antagonism of B. simplex 30N-5 and B. subtilis 30VD-1 on Fusarium spp. to that elicited by the Negev-desert isolated B. simplex strains, 11 and 237. All four bacterial strains prevented F. oxysporum f. sp. conglutinans (FOC), F. oxysporum f. sp. matthioli (FOM), and F. solani (FS) from forming dense hyphal mats with similar efficiency (Figure 1). After 7 days of incubation, the four different bacteria had inhibited fungal growth by 60–70% in comparison to the controls (Figure 1). Although B. simplex 30N-5, 11, and 237 each caused extensive hyphal thinning and a comparatively less dense hyphal network than the fungus in the control set, they were unable to halt FOM, FOC, and FS growth as markedly as did B. subtilis 30VD-1. Phase contrast microscopy of the fungal mycelia in confrontation assays with B. subtilis 30VD-1 revealed distortions and abnormal swelling resulting in bulbous structures in the hyphal cell wall. No such changes were detected in control mycelia (Supplementary Figure S1).

7Because of the hyphal thinning phenotype, we hypothesized that growth inhibition might be brought about by the activity of hydrolytic enzymes. Although the genome of B. simplex 30N-5 is in permanent draft stage (IMG genome ID: 2516653086), it allowed us to detect genes, which when expressed, could be involved in inhibiting fungal growth. The fact that additional B. simplex genomes are available made them useful as scaffolds to find missing genes in one or more of the species.

8Fusarium cell walls are composed of chitin, α-1,3-glucans, and β-1,3-glucans, suggesting that chitin-degrading enzymes and/or glucanases are likely to be responsible for B. subtilis 30VD-1’s effects on hyphal thinning. Because we lack a complete genome sequence for B. subtilis 30VD-1, we used the B. simplex genomes to elucidate the identity of the genes responsible for cell wall degradation. The 30VD-1 genome sequence is in draft stage and undergoing analysis. However, preliminary studies strongly suggested that many of the genes detected in the B. simplex genomes are found in B. subtilis 30VD-1.

9Two distinct mechanisms are known for degrading chitin, a linear polymer of β-1,4-linked N-acetylglucosamine (NAG) residues, which shares structural similarities with cellulose. One mechanism employs endochitinases, also known as 1,4-β-poly-N-acetylglucosaminidases (EC 3.2.1.14). Exochitinases, cellobiase/β-glucosidases (EC 3.2.1.21), or exoglucanases, now classified as β-N-acetylhexosaminidases (nagZ) (EC 3.2.1.52), release NAG dimers and monomers from chitin. Although we found no evidence for genes encoding 1,4-β-poly-N-acetylglucosaminidase or any other chitinase (chiC or chiA) or similar genes in the sequenced B. simplex strains except for B. simplex VanAntwerpen02 (data not shown), the nagZ gene and two genes designated as 6-phospho-β-glucosidase (celF) were detected in the B. simplex genomes, with high identity among them (orange shading, Supplementary Figure S2). The B. simplex nagZ genes are only 62–69% identical to nagZ in the more typical PGP bacilli (blue shading, Supplementary Figure S2), and show 80–98% identity among the eight sequenced B. simplex strains. Furthermore, neither B. simplex 30N-5 nor B. subtilis 30VD-1 were positive in tests using the chitin azure assay, which detects chitinase activity, whereas Paenibacillus tundrae 47 was positive.

10The identity of the two B. simplex celF genes compared to the B. firmus/B. cereus/B. megaterium group of Bacillus spp. in our survey (purple shading, Supplementary Figure S2), varied from 71 to 74% for B. megaterium and 90% to the B. cereus JM-mgvxx-63 gene. A celF ortholog in B. kribbensis DSM 17871 was 68–72% identical to the B. simplex genes, but no celF ortholog was detected in B. firmus DS1. A third 6-phospho-β-glucosidase, annotated in several Bacillus species as a broad-specificity cellobiase (bglA), was 86% identical to a gene in B. cereus JM-mgvxx-63 (Figure 2), but we saw no evidence in the B. simplex genomes for genes encoding cellobiase or cellulase except for IMG2517085743, which was annotated as Cellulase M (an endoglucanase; EC:3.2.1.4). Evidence that this gene could encode cellulase is suggested by the hydrolytic enzyme assay, in which the B. simplex and the B. subtilis strains formed halos on CMC plates (Table 1 and Supplementary Figure S2). Of the four strains, however, B. subtilis 30VD-1 produced the largest halo on CMC plates compared to its colony size (Supplementary Figure S2). A cellulase (glycosyl hydrolase family 5) gene was detected in the incomplete 30VD-1 genome, but it exhibited very little sequence similarity with the B. simplex genes.

11The second mechanism of chitin breakdown requires deacetylation of chitin through the activity of chitin deacetylases (CDA; EC 3.5.1.41) followed by subsequent hydrolysis of the deacylated chitin by chitosanase (EC 3.2.1.132). Seven different sequences in the B. simplex 30N-5 genome were annotated as xylan/chitin- or peptidyl/chitin-deacetylases and six of them are shown in Figure 2. A gene sequence annotated as endoglucanase Y/chitosanase was detected in B. simplex 30N-5 and the other B. simplex genomes (Figure 2). This same gene, called endoglucanase Y in B. cereus JM-Mgvxx-63, is only 38% identical to the B. simplex 30N-5 gene. Similarly, a chitosanase gene was detected in B. subtilis 30VD-1, albeit at a very low identity to the B. simplex genes.

12To test whether B. simplex could utilize chitin, the bacteria were grown on colloidal chitin-containing plates. Figure 3A shows that the positive control P. tundrae 47 grew robustly on colloidal chitin-containing medium and not only produced a halo indicating chitinolytic activity, but also developed a central clearing of the colony as seen in studies on chitinolytic and chitin deacetylase-producing bacteria. Although neither the B. simplex strains nor B. subtilis 30VD-1 developed obvious halos around the peripheries of their colonies, they all grew on the colloidal chitin-containing medium, especially B. subtilis 30VD-1. Moreover, the central part of the colony cleared (Figures 3B,C,E,F) as described for chitin deacetylase producers. In contrast, Kocuria rosea 43 did not grow on the colloidal chitin-containing medium. Only an outline of the inoculum and distinct particles of colloidal chitin were observed even after 3 weeks of incubation (Figure 3D).

13Genes encoding chitin-degrading enzymes such as lysozyme are also present in the B. simplex 30N-5 genome. Two genes, one annotated as lysozyme, and the other as lysozyme-like, were detected in all B. simplex strains studied (Figure 2). A number of protease encoding sequences are also present, and all of the Bacillus strains exhibited protease activity on plates (Supplementary Figure S2).

14Two adjacent genes (IMG2517085627 and -628), one of which, gene -627, was annotated as endopolygalacturonase, are present in the B. simplex 30N-5 genome. No pectin methyl esterase encoding genes were found in the B. simplex genomes. However, the presence of pectinolytic activity in three B. simplex strains and B. subtilis 30VD-1 on pectin-containing plates strongly suggests that the gene annotated as endopolygalacturonase is active (Supplementary Figure S2). Moreover, B. subtilis 30VD-1 produced a much larger halo for its colony size than the B. simplex strains (Table 1 and Supplementary Figure S2), suggesting that a similar gene sequence is present.

15Both B. simplex 30N-5 (IMG251083929) and P558 possess a gene annotated as xylA, which is 89.7% identical and orthologous to a gene with the same name in B. solisalsi CGMCC 1.6854. A gene encoding a presumed beta-xylosidase (IMG2517086168) was detected in the B. simplex 30N-5 genome and the other B. simplex genomes. This gene was 100% identical to a gene from Bacillus sp. Soil745 annotated as alpha-N-arabinofuranosidase, which is known to act on arabinoxylan. In spite of the lack of more definitive genomic data, evidence for xylanase activity by the B. simplex strains and B. subtilis 30VD-1 was detected in a plate assay (Table 1 and Supplementary Figure S2). Thus, it is likely that one or more of the genes annotated as xylanase are expressed.

16We used several different methods to study biocontrol. In a pilot experiment using a protocol described by, we tested scented stock seeds with FOM and FS, and found that both sets of seeds were quickly overrun with fungal hyphae, but more so when treated with B. simplex 30N-5 than with B. subtilis 30VD-1. Another test was performed on scented stock seedlings inoculated only with B. subtilis 30VD-1 and challenged with the fungi. After 14 days, the FOM- and FS-treated plants were severely stunted whereas the FOM- or FS-infected plants treated with B. subtilis 30VD-1 were greener and larger, although not as well developed as the untreated controls. When the experiment was terminated and the dry weights measured, the dry weights of B. subtilis 30VD-1 treated FOM- and FS-seedlings were statistically higher than the diseased plants albeit not as high as the non-Fusarium infected seedlings (Supplementary Figure S3). The B. subtilis 30VD-1-treatment of the FOM- and FS-infected plants resulted in a 34.6 and 26.3% increase in dry weight compared to the respective FOM and FS-infected control plants. Using the disease severity scale developed by, the reduction in disease severity was 33.3% for both treatments.

17Because of its larger seed size compared to scented stock, we tested the antagonistic efficacy of the 30VD-1 strain for its ability to control Fusarium wilt infection in pea (Pisum sativum). The plants raised from 30VD-1 bacterized seeds exhibited reduced disease development and better plant health in FOM-infested soil compared to the non-bacterized controls growing under similar conditions. This observation was supported by a 35.6% increase in dry plant biomass of the 30VD-1+FOM treatment in comparison to FOM inoculation alone (Figure 4).

18Many PGP bacilli produce BVCs that inhibit fungal growth. We employed a dual plate assay based on previous studies in the literature and variations thereof, but the bacteria did not remain confined to their half of the dual plate assay in the time needed to finish the experiment. We next tried the two-Petri dish assay described by Bernier et al. (201l), and first tested pure compounds, which are known to inhibit fungal growth, namely phenol, benzaldehyde, 1-decanol, and 1-dodecanol, in the center plate. The most effective pure chemical tested against FS and FOM was phenol, which completely inhibited fungal growth; the least effective was 2-propanone, which had no effect. Repeated experiments showed that phenol, 1-decanol, 1-dodecanol, and benzaldehyde were inhibitory to fungal growth. These results showed that the fungi responded to the pure chemicals, but when the bacilli alone were used in the experiments, no fungal growth inhibition was observed even after 5 days. The dual plate assay results demonstrated B. subtilis 30VD-1 as the most effective biocontrol agent in comparison to the rest of the 3 B. simplex strains included in the study. The biocontrol efficacy of the 3 B. simplex strains was transient and did not persist for long, so B. subtilis 30VD-1 owing to its persistent and effective antagonism toward Fusarium spp. was selected for all further in vitro and in vivo studies.

19In the split plate assay, where we grew the fungal cells (FOM/FS) in a liquid medium on one side of the split plate and B. subtilis 30VD-1 on the other side, we saw a significant decrease in growth of FOM and FS in response to volatiles. The FOM and FS dry biomass (g) in response to B. subtilis 30VD-1 was recorded as 0.496 ± 0.20 and 0.705 ± 0.05 g, respectively, in comparison to the control FOM and FS counterparts, which were 0.935 ± 0.03 and 0.819 ± 0.09 g, respectively. The reduction in FOM and FS biomass following B. subtilis 30VD-1 inoculation and incubation was 46.9 and 16.5%, respectively, in comparison to the FOM/FS controls.

20To determine whether or not a component of the bacterial culture filtrate (CF) was inhibitory to FOM and FS proliferation, we extracted the B. subtilis 30VD-1 CF with 1-butanol. B. subtilis 30VD-1 CF was mixed with equal amount of 1-butanol, and the partitioned butanol extract on drying was resuspended in sterile distilled water and used as crude butanol extract for fungal growth bioassay. Measurements of fungal growth were made after the fungus in control plate attained full growth. The crude butanol extract caused 40–42% inhibition of FOM and FS growth at 100 μg/ml, and was effective even at a dilution of 1:100 (Supplementary Figure S4). Further characterization of the extracted active fraction is in progress.

21Plant-growth promoting bacteria have numerous strategies, both direct and indirect, to enhance plant growth and survival. Previously, we observed fungal hyphal thinning in response to coinoculation between B. simplex 30N-5 and F. solani. In this study, we examined this B. simplex strain and strains 11 and 237 in greater detail and analyzed additional B. simplex genomes to establish the genetic support for the results we observed. Evidence for genes that encode enzymes in the known chitin degradation pathways in the B. simplex and B. subtilis 30VD-1 genomes and that are orthologous to genes in the well-known PGP bacilli were found in the B. simplex 30N-5 genome. Support for their activity was discovered when the three B. simplex strains and B. subtilis 30VD-1 grew and exhibited a colony phenotype identical to that described by on colloidal chitin-containing plates. This result strongly suggests that the mechanism predicted by the genome analysis – deacetylation of chitin through chitin deacetylase activity and hydrolysis of the deacylated chitin by chitosanase – is functional. Thus, the in silico and experimental studies agreed, demonstrating a direct correlation between genotype and phenotype.

22We also detected several genes potentially encoding hydrolytic enzymes and found evidence for the synthesis of these enzymes through specific plate assays. The ability to break down fungal cell walls is a defining characteristic of many BCAs, including Streptomyces galilaeus CFFSUR B-12, which exhibits antifungal activity against Mycosphaerella fijiensis, a plant pathogen, and F. oxysporum f. sp. lycopersici, which causes tomato wilt. Other cell wall-degrading enzymes produced by bacilli include cellulases, pectinases, and xylanases, and genes encoding these hydrolytic enzymes were detected in the B. simplex genomes. Bacteria employ hydrolytic enzymes to break down plant cell wall material for utilizable carbon, but endophytes also use these enzymes to enter plant tissues. B. simplex 11 is an endophyte based on its isolation from dead root material, and other B. simplex strains are known to be endophytes in maize. In addition, endophytic bacteria have been shown to control Fusarium-induced plant diseases.

23The BVC 2,3-butanediol and other volatiles promote growth in Arabidopsis in response to pathogens. Both B. subtilis GB03 and B. amyloliquefaciens IN937a synthesize 3-hydroxy-2-butanone (acetoin), which gives rise to 2,3-butanediol, the proposed inducer. In our previous study, we could not find in the genome of B. simplex 30N-5 the gene alsD, which encodes alpha-acetolactate dehydrogenase, the enzyme required for the decarboxylation of acetolactate into acetoin. Moreover, B. subtilis 30VD-1 and the three strains studied here do not produce acetoin, which was also the case for B. simplex M3-4. However, other BVCs produced by bacteria protect plants from biotic stress, and previous work demonstrated that B. subtilis strains produce a number of them. B. simplex and B. subtilis are known to synthesize 2-propanone, phenol, 1-decanol, 2-undecanone, 1-decanol, 1-dodecanol, methyl-tetradecanoate, and 2-ethyl-3,5-dimethylpyrazine, and many other microbial VOCs. found that B. simplex BVCs from six distinct isolates killed almost 100% of the parasitic nematodes Panagrellus redivivus and Bursaphelenchus xylophilus. When pure compounds were employed, these authors further determined that the BVCs that elicited the highest nematode mortality were phenol, 1-decanol, 2-nonanone, 2-undecanone, phenylethanone, cyclohexene, and benzaldehyde.

24Based on the fact that several of the BVCs resulting in nematode death overlap with the antifungal BVCs used by B. amyloliquefaciens NJN-6, we expected that the Bacillus strains in our study would synthesize antifungal BVCs and inhibit the growth of FOM and FS. Our experiments using the dual plate assays at first suggested that the B. simplex strains and B. subtilis 30VD-1 inhibited FOM growth, but repetitions under similar conditions showed no effect on fungal colony diameter. Only in the cases where pure chemical compounds were employed, did we see a dramatic effect. However, because the cultures were touching after 7 days, we decided to employ more restrictive experimental regimes to separate the bacteria and the fungi. The split plate method, when the cultures were grown in liquid instead of solidified medium, led to a reduction in dry fungal biomass in comparison to the control plates. noted that BVCs were less effective overall than actual bacteria in inhibiting the growth of certain fungi including F. oxysporum. Our results suggested that contact was necessary for complete FOM- and FS-growth inhibition leading us to test whether CF extracts were involved in mycelial growth inhibition, which they were.

25In nature, many factors influence inter-organism communication. Soil bacteria that associate with plants often live in nutrient-poor soil, but most studies on plant-growth promotion and biocontrol are performed under close-to-ideal nutrient conditions. Also, bacteria in nature live in consortia communicating with one another as well as with eukaryotic organisms and other bacteria. Data show that either combining PGPB or co-inoculating PGPB and non-pathogenic fungi serve as more effective BCA than a single bacterial strain (;;; and references therein). Also, as indicated for BVCs, it is likely that cooperation is common between BCA and plants, but as yet few of the factors involved in this cooperativity and/or host specificity have been thoroughly investigated. As a consequence, whether the transition occurs with the same efficacy from laboratory and greenhouse conditions, where positive responses are observed, to field conditions is a major challenge.

26More challenges will arise as attempts are made to change from the use of synthetic fertilizers and pesticides to sustainable agricultural practices whereby beneficial soil microbes promote plant growth and protect plants from predator and pathogen attack. In addition, PGPB enable plants to obtain limiting nutrients, such as Fe, P, and N, which also improve plant health. This effort will require a combination of genomic, biochemical, and experimental analyses as well as testing pairwise and more complex combinations.

27Several of the bacteria and fungi examined in this study were described previously. B. simplex 237, isolated from soil under the canopy of Zygophyllum dumosum, and B. simplex 11, isolated from surface-sterilized, dried Z. dumosum root tissue, were also investigated (Table 2). B. amyloliquefaciens subsp. plantarum FZB42 and B. subtilis GB03, obtained from the Bacillus stock center, were used as controls in several of the physiological assays. For the chitin-degradation assay, the procedure used by was followed. P. tundrae 47 and Kocuria rosea 43 were used as positive and negative controls, respectively. Details about the strains are presented in Table 2.

28The B. simplex, B. subtilis, and B. amyloliquefaciens strains were grown overnight at 30°C on agar-solidified Tryptone Yeast extract (TY) or Luria-Bertani (LB) medium. The fungi, Fusarium oxysporum f. sp. matthioli (FOM), F. oxysporum f. sp. conglutinans (FOC), and F. solani (FS) were maintained on Potato dextrose agar (PDA), and based on experimental requirements were grown on PDA or solidified V8 medium at 30°C for a week prior to their use in the anti-fungal assays.

29The testing of biocontrol efficiency due to diffusible compounds was performed by growing the pathogen (FOM/FOC/FS) and test bacteria on V8 agar using a dual culture technique. Briefly, the bacterial strains were grown overnight in TY liquid medium and then after centrifugation to eliminate the culture medium, were diluted with sterilized Millipore water to an OD600 value between 0.35 and 0.40. Fifty microliters of the bacterial suspensions were streaked in a line on V8 agar in conventional Petri dishes 1 cm from the plate edge. Fungal plugs 5 mm in diameter were then placed on the agar 2.5 cm away from the bacterial streak either immediately after the bacteria were plated (for B. subtilis 30VD-1 and the water control) or after 24 h (for the slower growing B. simplex strains). The plates were incubated at 30°C. The fungal radii were measured both toward and away from the bacterial streak every 24 h for 7 days at 15°, 10°, 4°, 0°, −4°, −10°, and −15° from the perpendicular formed by the plug and bacterial line. The radii were measured from the center of the fungal plug until the point where the region of highest hyphal density began to break down. The percent fungal growth inhibition was calculated using the equation: [Growth Inhibition = (FBacteria–FWater)/FWater], where FBacteria is the radius of fungal growth toward the bacteria and FWater is the radius of fungal growth toward the water in the control. Each experiment was repeated three or more times, with 10 replicates for each treatment.

30Fifteen to twenty microliters of an overnight bacterial culture were spotted onto LB, TY, or specialized media for testing hydrolytic enzyme activity. After 3–4 days of incubation, the three B. simplex strains and B. subtilis 30VD-1 were examined on the assay plates. For detecting cellulase and xylanase activity, a procedure modified from was used in which the overlay medium was eliminated and carboxymethylcellulose (CMC) was incorporated directly in the plate medium. Pectinase activity was detected on a pectin (either apple or citrus peel)-containing medium as described by after 5 days of incubation. The pectin plates were left unstained or visualized with 0.05% ruthenium red staining for 20 min followed by several rinses with distilled water.

31To detect the activity of chitin-degrading enzymes, the colloidal chitin-containing medium described by was used. Clear zones around or within the colonies are considered evidence of chitinase activity in a plate containing colloidal chitin whereas growth of the colony along the edges on the medium coupled with a central clearing are diagnostic of chitin deacetylase activity. The experiment was repeated twice with 4–5 replicates per experiment.

32Protease activity was determined using skim milk agar medium, which contained (per liter): 5 g pancreatic digest of casein, 2.5 g yeast extract, 1 g glucose, 7% skim milk solution, and 15 g of agar. Bacterial cells were spot-inoculated and after 2 days incubation at 28°C, the plates were flooded with Coomassie Brilliant Blue (0.25% w/v) dissolved in methanol:acetic acid:water (5:1:4 v/v) for 10 min to enhance the visibility of halos around the bacterial colonies. Destaining with methanol:acetic acid:water was performed to ensure better contrast before taking pictures of the assay plates. Photographs were taken on an Illuma System Light Control platform with a Canon PowerShot ELPH350HS camera. Each experiment was repeated three times with 3 or more colony replicates.

33The draft and finished genomes used for the analysis of the PGPB were utilized in this study, except that the genomes of B. simplex BA2H3 (NCBI Accession number: NZ_KN360955) and B. simplex 558 (NCBI Accession number: NZ_CCXW01000001) were also included, as well as B. subtilis 30VD-1 (IMG ID number 2786546174) and the two Paenibacillus genomes from the earlier paper were eliminated. A group of manually annotated genes from the B. simplex 30N-5 genome, which were likely to be involved in biocontrol based on the literature, served as the reference genes (set at 100%) to query known PGP Bacillus species in the Joint Genome Institute IMG/ER or NCBI1 databases for blastp homologs. As in our previous study, the K-means clustering algorithm established the species order shown in Figure 2. The blastp searches were filtered to include conservative alignments of e ≥ 10−5 and sequence identities of 50% or higher. Identities greater than 50% were indicated by different shades of orange in Figure 2 whereas those less than 50%, which were retained because of potential gene homology, were shaded blue. Genes that were missing either because the genome is unfinished or because the gene was not found are indicated as “n.d.” (not detected) in unshaded cells of the table.

34Bacterial strains were evaluated for their volatile-mediated antifungal activity against FOM and FS using the centrally partitioned Petri dishes (85 mm × 15 mm) with or without physical contact between the two microorganisms grown on either side. The larger partition was filled with 10 ml LB agar for the growth of bacteria and the smaller one with 10 ml potato dextrose broth (PDB) for the fungal counterpart. One hundred microliters of a 48 h grown bacterial suspension containing approximately 109 CFU/mL were spread onto the LB agar and a fungal spore suspension containing 106 spores/mL resuspended in water was inoculated into the PDB medium. The plates were then sealed with Parafilm and incubated in the dark for 7 days. The controls were uninoculated LB agar and PDB in the separate partitions.

35Following incubation at room temperature in the dark for 7 days, the antifungal effect of bacterial volatiles on the fungal pathogen was recorded in terms of reduction in dry fungal biomass as compared to the control. For fungal dry biomass observations, the PDB compartment was emptied onto a pre-weighed Whatman filter paper and dried at 60°C. Reductions in dry fungal biomass readings were measured and calculated as percentages.

36Seeds of Matthiola longipetala ssp. bicornis (scented stock), obtained from Baker Creek Heirloom Seed Company, Mansfield, MO, United States, were surface-sterilized by immersing them in 95% ethanol for 1 min, followed by three rinses with sterile distilled water. For the treatments, seeds were bacterized for 3 h by imbibing them in a bacterial suspension, which had been grown for 48 h to contain approximately 109 CFU/ml. Seeds imbibed in LB medium served as the control. The seeds were placed into sterile OPTILUXTM Petri dishes (100 mm × 20 mm) containing complete Hoagland’s medium solidified with 1% agar until germination. At that time, healthy seedlings were gently uprooted from the agar and dipped for 30 min into distilled water, or into an aqueous suspension of B. subtilis 30VD-1, or inoculated with a spore suspension of FOM or FS (106 spores/ml), or both B. subtilis 30VD-1 and FOM or FS. The control and treated seeds were then replanted onto 0.5% Hoagland’s agar. The seedlings were incubated for 3 days in the dark at 28°C and then exposed to light until 15 days after the initiation of experiment.

37Based on an assay developed by for Arabidopsis, the disease severity index was modified for the stock and pea experiments. Healthy plants classified as 0 or 1 were green, but plants rated with an index of 1 had smaller leaves than the healthy 0-ranked plants. Stock plants showing moderate yellowing and wilting with less developed leaves were given a disease severity index of 2 whereas plants displaying additional disease symptoms such as acute yellowing, wilting, and an even greater reduction in leaf area were considered as having an index level of 3. Plants with a disease severity index of 4 were highly stunted whereas dead or near dead plants were categorized as index 5. Plant dry weight measurements were taken. The plants were gently uprooted and placed into pre-weighed microcentrifuge tubes, air-dried and then measured on an analytical balance. There were 10 replicates for each experiment, which was repeated four times.

38Because FOM has a broad host range, pea (Pisum sativum) was used as a test plant for pathogenicity and protection by B. subtilis 30VD-1. Seeds of pea (var. Little Marvel) were surface-sterilized and bacterized as described earlier. Seeds imbibed in bacteria-free medium were used as a control. Pathogen challenge was performed by mixing the spore suspension in soil such that the spore count was 105 spores/g of soil. Thus, treatments utilized in the study were the control, 30VD-1 treated, FOM-treated, and 30VD-1+FOM-treated. Six replicates of each treatment, with five plants in each pot, were maintained. The experiment was conducted in a Conviron plant growth chamber. Fusarium wilt assessment was conducted after 45 days according to the scale described above. Plant dry weight measurements were made after drying the plants in a 60°C oven.

39Photographs of the plates and the plants were taken with a Canon PowerShot ELPH350HS camera.

40The statistical analysis of the data was performed using SPSS software, version 21 (SPSS Inc., Chicago, IL, United States). After collection of the fungal growth parameters, or distances to the bacterial culture, the data were processed with one-way analysis of variance (ANOVA). A post hoc Fisher’s LSD test was used to identify treatments with means significantly different from the water control at P ≤ 0.05.

41AH, NK, and PM-H were involved in planning and execution of the research, analysis and interpretation of the data, and writing the manuscript. TI optimized and performed the biocontrol assays along with NK, PM-H, and NN. MM performed the in vitro assays and helped in writing and editing the manuscript. AH did the genome analysis and ES and DK contributed strains and intellectual support. EH verified the identity of all the strains and edited the manuscript.

42The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.