Selecting lipopeptide‐producing, Fusarium‐suppressing Bacillus spp.: Metabolomic and genomic probing of Bacillus velezensis NWUMFkBS10.5

Abstract The results of this study indicate that the maize rhizosphere remains a reservoir for microbial strains with unique beneficial properties. The study sought to provide an indigenous Bacillus strain with a bioprotective potential to alleviate maize fusariosis in South Africa. We selected seven Bacillus isolates (MORWBS1.1, MARBS2.7, VERBS5.5, MOREBS6.3, MOLBS8.5, MOLBS8.6, and NWUMFkBS10.5) with biosuppressive effects against two maize fungal pathogens (Fusarium graminearum and Fusarium culmorum) based on 16S rDNA gene characterization and lipopeptide gene analysis. The PCR analysis revealed that lipopeptide genes encoding the synthesis of iturin, surfactin, and fengycin might be responsible for their antifungal activities. Few of the isolates also showed possible biosurfactant capability, and their susceptibility to known antibiotics is indicative of their eco‐friendly attributes. In addition, in silico genomic analysis of our best isolate (Bacillus velezensis NWUMFkBS10.5) and characterization of its active metabolite with FTIR, NMR, and ESI‐Micro‐Tof MS confirmed the presence of valuable genes clusters and metabolic pathways. The versatile genomic potential of our Bacillus isolate emphasizes the continued relevance of Bacillus spp. in biological management of plant diseases.

as Fusarium head blight (FHB) or Fusarium ear rot (FER) and many more, is caused by fusarium members such as F. graminearum and F. verticillioides. These diseases affect maize production in South Africa and other areas of the world (Boutigny et al., 2012(Boutigny et al., , 2011Summerell & Leslie, 2011). Maize, a staple crop in South Africa, is consumed daily in most households and used in the production of animal feeds (Janse van Rensburg, Mclaren, Flett, & Schoeman, 2015;Lamprecht, Tewoldemedhin, Botha, & Calitz, 2011); therefore, efforts to reduce loss due to preharvest and postharvest contamination by F. graminearum infection have recently gained significant attention (Boutigny et al., 2011;Mngqawa et al., 2016). The presence of mycotoxins zearalenone and deoxynivalenol found in maize grains infected by F. graminearum is also a cause for concern (Wang, Ndoye, Zhang, Li, & Liao, 2011).
Reports show that BCAs used for crop protection perform better in their native geographical regions due to increased survival rate compared to the use of imported commercial BCAs (Abiala, Odebode, Hsu, & Blackwood, 2015;Bardin et al., 2015;Grzywacz, Stevenson, Mushobozi, Belmain, & Wilson, 2014;Pereira, Nesci, Castillo, & Etcheverry, 2010). Our goal in this present work was to select indigenous Bacillus strains from the maize rhizosphere, evaluate their anti-phytopathogenic potentials in vitro against Fusarium spp., molecularly characterize the Bacillus isolates, and identify the likely mechanisms they employ in their anti-phytopathogenic activities. Bacillus spp. secrete lipopeptide compounds such as surfactin, fengycin, and iturin that they utilize in antibiosis. The presence of these cyclic lipopeptides in our maize root-associated strains would be valuable if they are to be considered for in planta studies and subsequently for the management of F. graminearum infections in South Africa.
Strains within the genus have also been reported to synthesize structurally diverse secondary metabolites that exhibit broad-spectrum antibiotic activities, and the genomic basis for the synthesis of these secondary metabolites has been attributed to the presence of polyketide synthases (PKSs) and non-ribosomal peptide synthetase (NRPS) in their genomes (Raaijmakers, Bruijn, & Kock, 2006;Roongsawang, Washio, & Morikawa, 2011;Tyc, Song, Dickschat, Vos, & Garbeva, 2017). The amphipathic structure, the hydrophilic peptide portion, and a hydrophobic fatty acid portion of these peptides show resemblances. These peptides also exhibit a cyclic nature due to the linkage of their C-terminal peptide residue either indirectly to a β-hydroxy fatty acid or directly to a β-amino acid (Mnif & Ghribi, 2015;Ongena & Jacques, 2008). These antimicrobial peptides have been isolated, quantified, purified, and characterized using various approaches and techniques that ensure the chemical components responsible for their bioactivity are well understood. The majority of the current approaches employed involve the combination of chromatographic techniques, mass spectrometry, nuclear magnetic resonance (NMR), and Fourier transform infrared spectroscopy (FTIR) (Biniarz, Łukaszewicz, & Janek, 2017;Jasim, Sreelakshmi, Mathew, & Radhakrishnan, 2016).
Reports have shown that expression of biosynthetic genes and secretion of secondary metabolites may be difficult during laboratory culture of potential BCAs due to growth conditions (Laureti et al., 2011). The non-expression of genes or secretion of secondary metabolites can hinder the identification or detection of the specific metabolite or gene responsible for the antimicrobial activities of a BCA (Michelsen et al., 2015). To fully understand beneficial bacterial species, genomes of multiple independent isolates are required for comparison . Comparing the total repertoire of genes for a group of genomes from close bacterial species is an instrumental approach for the development of novel beneficial compounds and for the functional characterization of important genetic determinants in significant microbial strains (Medini, Donati, Tettelin, Masignani, & Rappuoli, 2005). The bacterial pangenome can be defined as the complete repository of genes located in the genome of closely related bacterial species. This includes the "core genome" (genes identified in two or more strains) and the "dispensable genome" (genes peculiar to single strains) Rouli, Merhej, Fournier, & Raoult, 2015;Tettelin et al., 2005).
The core and dispensable genes are crucial signatures for recognizing species diversity.
Researchers now use a combinational approach such as genome mining, pan-genome analysis, structural data elucidation, and metabolomic characterization to identify biosynthetic products secreted by important microbes (Dunlap, Bowman, & Schisler, 2013;Van Der Voort et al., 2015). Often, genomic data offer predictions that lead to the detection of novel biosynthetic pathways, genes, and enzymes which then enables experimental isolation, structural elucidation, and chemical characterization of novel compounds (Challis, 2008). The combinational approach ensures that in silico or theoretically predicted biosynthetic products correlate with structurally or chemically identified metabolites (Ziemert, Alanjary, & Weber, 2016). Here, we examined the antimicrobial potential of the lyophilized extract of the secondary metabolites secreted by Bacillus velezensis NWUMFkBS10.5 while employing electro-spray ionization mass spectrometry (ESI-Q-TOF MS), FTIR, and NMR to confirm that its secondary metabolites were functionally active.

Lastly, the genomic information of B. velezensis and other related
Bacillus strains having strong potential for the control of phytopathogens including F. graminearum has been made available in recent years (Dunlap et al., 2013;Dunlap, Schisler, Bowman, & Rooney, 2015;Lee et al., 2015;Palazzini, Dunlap, Bowman, & Chulze, 2016;Pan, Li, & Hu, 2017). In light of this, we sequenced the genome of our best isolate NWUMFkBS10.5, to determine its phylogenomic association and its unique antimicrobial trait.

| Description of sampling sites and sample collection from rhizosphere
Depending on the width of the maize plot and in no particular order, 20-30 g of rhizospheric soil was collected randomly from four maize rows, 15-25 m, apart in each plot from 10 maize farms in the North West Province of South Africa at harvest time. The geographic location of the sampling sites covers 28,206 km 2 area. The temperature ranges between 17°C and 31°C during the summer and between 3°C and 21°C during the winter, with an average rainfall of 360 mm. Harvested maize plants were shaken gently at the roots to manually remove the loosely attached soil. The adhering root soil was considered as the rhizosphere soil, and these samples were pooled for each location. We obtained 10 different soil samples from the different maize plots.  Chen, Chen, Zhang, and Zhu (2014) were slightly modified. A 5-mm-diameter plug from an actively a growing (7-day-old) mycelial culture of F. graminearum was placed in the center of freshly prepared PDA plates (90 mm). From the 200 Bacillus isolates initially selected, six fresh colonies from 24 hr LB agar culture were circularly streaked (equidistance 1.5 cm) along each PDA plate at a distance of 1.5 cm from the edge of the plate using a sterile inoculating loop. Control plates consisted of F. graminearum placed on PDA alone. The plates were further incubated at 28°C for 7 days. Thereafter, only 11 isolates (BS1.1, BS2.7,BS3.5,BS4.3,BS4.6,BS5.5,BS6.2,BS6.3,BS8.5,BS8.6,and BS10.5) exhibiting strong inhibition were selected for further antifungal confirmatory tests.

| Confirmatory in vitro antifungal test
Plates were prepared as described above; however, antagonism was carried out against two fungal pathogens (F. graminearum and where PGI is the percentage of growth inhibition, C1 is the control mycelia area of uninhibited fungi, and C2 is the distance between the bacterial colony and the growing edge of the fungal mycelia. Experiments were repeated three times, and the values were recorded as the means of three replicates.

| Drop collapse test and microplate assay
To determine the production of biosurfactant compounds, a modified "drop collapse test," applied according to Yanes, Fuente, Altier, and Arias (2012)

| Detection of lipopeptide genes and molecular characterization of Bacillus isolates
Identification of the selected Bacillus isolates was by 16S rDNA gene sequencing (Garbeva, Veen, & Elsas, 2003), and the presence of lipopeptide genes in the DNA extracts of the Bacillus isolates was determined with a 25 μl reaction mixture containing 1.5-2.5 μg of template DNA; 1 μl of primer, 12.5 μl OneTaq Quick-Load 2× master mix with standard buffer (New England Biolabs NEB), and 9.5-10.5 μl nuclease-free water in PCR thermocycler. All the primers utilized in PCR amplification protocols were synthesized by Whitehead Scientific, Integrated DNA Technologies (Supporting Information Table S3). The PCR amplicons were analyzed by electrophoresis in 1% (w/v) agarose gel, and the sizes of the bands were determined using 1-kb molecular marker. The gel containing 10 μg/ ml ethidium bromide (Bio-Rad) was visualized using a gel documentation system (Gel Doc 2000, Bio-Rad) to confirm the expected size of the PCR products. NucleoSpin Microbial DNA Purification Kit (Macherey-Nagel) was used to purify the PCR products which were then sent to Inqaba Biotec (Pretoria, South Africa) for sequencing.
16S rDNA sequences were blast searched on the NCBI GenBank and ENA database (default settings). Aligned sequences were analyzed using MEGA 7.0 software (Tamura et al., 2011), and phylogenetic trees were reconstructed based on the 16S rDNA gene using the neighbor-joining methods (Saitou & Nei, 1987). Topological robustness was evaluated by bootstrap analysis (Felsenstein, 1985) based on 1,000 replicates.

| Extraction, collection of cell-free supernatant, and purification of secondary metabolites
From the seven antagonistic subsets, isolate BS10.5 (NWUMFkBS10.5) was the most effective with the highest number of encoding genes, and it was chosen for further analysis.
The production and purification of BS10.5 active metabolites were done according to  with slight modification. Cell-free antimicrobial substances from BS10.5 were collected after the rhizobacteria were grown in 1 L LB broth at 30°C with continuous shaking at 200 g for 72 hr. The cells were harvested by centrifugation at 13,000 g for 15 min, and the culture supernatant was filter sterilized through 0.22 µm nitrocellulose membranes (Millipore Corporation) filters to obtain cell-free supernatants. About 100 ml of cell-free supernatant was stored for anti-pathogen test. Isolate BS10.5 was further grown in 1 L LB broth at 30°C with constant shaking at 200 g for 4 days.
After fermentation, the cell filtrate was collected by centrifugation at 6,000 g for 15 min at 4°C, and the supernatant was acid-precipitated by adjusting to pH 2.0 with 6 M HCl. After an overnight incubation at 4°C, the precipitate was centrifuged at 8,000 g at 4°C for 15 min and the pellet was dissolved in methanol-water (50:50) and then filtered through 0.22 µm PTFE membrane filter to remove larger particles and cell components. The mixture was then concentrated by vacuum evaporator at 45°C and then finally lyophilized.

| Antibacterial activity
The activity of the cell-free supernatants was determined by disk diffusion assay. Sterile filter paper disks were impregnated and 28°C. Antibacterial activity was observed as inhibition zones around the disk, and the experiment was conducted twice in triplicates.

| Antifungal activity
Sterile filter paper disks impregnated with 60 µl of the cell-free supernatant of BS10.5 were placed at the far edge of a PDA plate, and 5-mm agar plugs of the Fusarium pathogens (F. graminearum and F. culmorum) were transferred to the opposite edge of the PDA plates. Nystatin (30 μg/disk) was used as control, and after 7 days of incubation and observation of plates at 25°C, zones of inhibition were recorded.

| Dose-dependent anti-pathogenic activity of the BS10.5 lyophilized extract
Disk diffusion was employed for the antimicrobial assay follow-

| Effect of the BS10.5 lyophilized extract and commercial fungicides on fungal mycelia growth
Using well diffusion, the activity of the BS10.5 extract powder was also compared to the activity of commercial fungicides using

| Fourier transform infrared spectroscopy
Fourier transform infrared (FTIR) spectroscopy identifies the types of chemical bonds and functional groups in compounds, and it is useful in elucidating the components of an unknown sample. Ten microgram of the lyophilized extract of BS10.5 was analyzed with transform alpha (FTIR) KBr-integrated spectrometer (Bruker).
Spectrum reading was at 400-4,000 wavenumbers (cm −1 ) with an average of 32 scans. Spectrum was viewed and collated with the OPUS spectroscopy software.

| Nuclear magnetic resonance spectroscopy (NMR)
Twenty microgram of the lyophilized extract of BS10.5 was dissolved in 0.5 ml of deuterated DMSO (dimethyl sulfoxide), and 1 H and 13 C NMR spectrum were acquired on the specific signal assignment.
NMR spectra were recorded using a Bruker Avance III 500 MHz spectrometer at room temperature with chemical shifts () recorded against the internal standard tetramethylsilane (TMS).

| Mass spectrometry analysis by ESI-Q-TOF MS
A high-resolution mass spectrum was obtained for the lyophilized extract of BS10.5 with an Applied Biosystems 4800 Plus MICRO-TOF/TOF analyzer (AB Sciex, USA) operated in the positive ion mode with an accelerating voltage of 20 kV, 337 nm nitrogen laser for ionization and α-cyano-4-hydroxycinnamic acid for matrix. Bruker compass data analysis was used to process the mass spectrometry data while molecular weights and formulae were characterized by mass spectrum smart formula tools. The FTIR, NMR, and mass spectrometry analyses were carried out at the Chemical Resource

Beneficiation/Laboratory Analytical Services of the North-West
University (Potchefstroom, South Africa).

| Data mining and in silico analysis of NWUMFkBS10.5 genome
Anti-smash (4.0.0rc1) and PRISM were used to predict the biosynthetic products and genetic clusters present in BS10.5. A Pan-genome was created on the Kbase platform to run a comparison of the total genes present in BS10.5 and other established biocontrol Bacillus strains.

| Statistical analysis
A multivariate general linear model was used to analyze treatment means and inhibition rates. Least significant difference test (LSD), Duncan multiple test, and Student-Newman-Keuls (SNK) test were used to compare observed means, pathogen-antagonist relationship, treatment effects, and effect of conditions of inoculation using SPSS statistical software (version 22) at the significance level of 5%.

| Presumptive selection and identification of bacterial isolates
All the 200 isolates selected from the HiChrome Bacillus agar according to manufacturer's descriptions (Supporting Information

| Bacillus inhibition of Fusarium mycelia
Antagonistic activity of the Bacillus isolates against the two Fusarium pathogens showed that lower inhibition rates were seen in condition 1 (antagonist inoculated on PDA 3 days before pathogens) and condition 2 compared to condition 3, where the fungal mycelia plug was inoculated on the PDA 3 days before the antagonists were inoculated (Tables 1 and 2). Overall, F. culmorum was more resistant to the antagonists during condition 1 and condition 2 when compared with susceptibility of F. graminearum during those conditions of treatment. However, the inhibition of F. graminearum was less during condition 3.
The stability of the biosurfactants produced under different growth conditions, however, might need to be determined because the production of biosurfactants is dependent on conducive pH, temperature, and salinity concentration (Rivardo, Turner, Allegrone, Ceri, & Martinotti, 2009).

| Molecular characterization of Bacillus isolates
The PCR analysis carried out against the seven isolates revealed that each isolates harbored at least one lipopeptide gene. PCR amplification was carried out according to previously established protocols.
The targeted genes and primers utilized during the PCR amplification are shown in Supporting Information Table S3. Multiple lipopeptide antibiotic genes were detected in BS10.5 amplicon ( predominant amplicon bands were produced from the PCR products from the gel electrophoresis (Supporting Information Figure S1). The presence of the detected lipopeptide antibiotics could be responsible for the antagonistic activities of the Bacillus isolates against fungal pathogens.

| Phylogenetic analysis
Based on the blast search of the partial 16S rDNA gene sequences and submission to NCBI, GenBank accession numbers were assigned to the seven isolates (Supporting Information  Figure S2).

| Antimicrobial activity of cell-free supernatants
The results of the antimicrobial activity of the cell-free supernatants showed that the cell-free culture filtrate from BS10.5 Note. •: (Positive) a PCR amplicon of expected size was seen; ○: negative.
TA B L E 4 Genes detected in the antagonistic Bacillus isolates using specific primers sets contains strong bioactive substances with inhibitory potentials against fungal and bacterial pathogens (Supporting Information   Table S5). The cell-free filtrates inhibited F. graminearum and B. cereus at the same rate, however, but inhibited K. pneumoniae and P. aeruginosa moderately. In comparison with antibiotics used against the fungal pathogens, the cell-free supernatant exhibited an inhibition level close to that of nystatin. However, it showed higher inhibitory potential than tetracycline and ciprofloxacin against the bacterial pathogens based on the concentration of the antibiotics used.

| Antipathogenic activity of the BS10.5 lyophilized extract at different concentrations
During the lyophilized extract antipathogenic test, the effect of the extracts decreased relative to an increase in dilution with PBS.
Among the bacterial pathogens, EF was the most susceptible to the lipopeptides extracts of BS10.5, while MC was the least susceptible. F. graminearum was more sensitive to the extract concentrations than F. culmorum (Figure 2). At 30 µl, three of the pathogens (PA, MC, and KP) showed no sensitivity.

| Comparative antifungal activity of the BS10.5 lyophilized extract and commercial fungicides
The BS10.5 extract showed a higher inhibition rate against F. graminearum than the commercial fungicides (nystatin and amphotericin) (Figure 3).
These wave numbers show characteristics similar to lipopeptides.
The stretching and vibration mode of the absorbance is indicative of aliphatic chains, alkyl chains, peptide bonds, and two amide bonds, signifying the presence of a compound with ester and amino groups.
This result, which is consistent with the report of Romero et al.

| NMR analysis
The proton analysis of the compound showed NH and OH proton at >8.00 ppm, some -CH 3 and CH 2 signals at <2.00 ppm and CH 2 -COO>2.00 ppm but <4.00 ppm (Figure 4). These are suggestive of a peptide backbone of secondary amide, aliphatic chains, and ester linkages, respectively. Chakraborty et al., (2014), gave a similar report from the proton spectra of the lipopeptide extract of B. vallismortis JB201 and B. subtilis SJ301. The signals of the carbon analysis were inconclusive despite several runs (data not shown).
In Supporting Information Figure S5a-k, the colors symbolize different functional gene types: blue (transport-related genes), green (biosynthetic genes), red (regulatory genes), and gray (additional genes).
In the antismash data, the highest number of biosynthetic genes was found in clusters 5, 6, 7, and 14 (10 each), whereas the PRISM result predicted 18 clusters in the NWUMFkBS10.5 genome. In Table 7, the functions of the predicted biosynthetic compounds are given.

| Whole genome nucleotide/NCBI biosynthetic gene blast and Pangenome comparison
Firstly, we carried out a genome blast search of NWUMFkBS10.5 to determine its genetic relatedness, and then, we blast searched the genome against selected lipopeptide gene clusters and constructed the phylogenetic trees (Figure 7 and Supporting Information Figure   S6). Thirdly, the pangenome comparisons of four established agriculturally and industrially relevant Bacillus strains with NWUMFkBS10.5

| D ISCUSS I ON
When in search of beneficial microbial strains for commercialization, key screening approaches and sequential steps have been identified that lead to selecting robust and effective candidates (Köhl, Postma, Nicot, Ruocco, & Blum, 2011). A comprehensive in vitro analysis starting from the culturing stage to molecular identification of action mechanisms and backed up by in silico genome exploration of potential candidates provides strains with better in planta viability and efficacy (Adeniji & Babalola, 2018). In this study, out of the 200 isolates initially screened for antagonism against the fusarium pathogens, 11 isolates showed consistent antagonism and among 11 isolates, isolate BS10.5 exhibited the strongest and most consistent antagonism considering the conditions. Applying biocontrol agents as a bioprotective coating prior to planting of crop seedlings has been demonstrated in previous reports (Yang et al., 2015). The antifungal result of this study demonstrates that our Bacillus isolates are candidates for preplanting and post-harvest bioprotective inoculants. The reduction in fungal mycelia growth (Tables 1 and 2) shows they could protect plants against the onset of fusariosis. Members of the Bacillus sp. exhibit remarkable biocontrol activity against phytopathogens due to the production of the lipopeptide group of antibiotics (surfactin, fengycin, bacillomycin, and iturin) Vitullo, Pietro, Romano, Lanzotti, & Lima, 2012). We detected these groups in some of our Bacillus isolates.
Culture-free extracts from B. velezensis antagonists have been used in bioprotection assays through several modes of application such as seed treatments and pour plate mixtures (Cao et al., 2018).  growth (Leclère et al., 2005). The antimicrobial activity shown by the cell-free supernatants of isolate BS10.5 suggests that the secondary metabolite of the isolate can also be of immense industrial benefit when purified in sufficient quantity.
Having employed a dose-dependent assay to determine the concentrations at which the pathogens were most sensitive to BS10.5 extracts, we report that the growth of the pathogens was retarded available (Koumoutsi et al., 2004;Xu et al., 2013), and Bacillus spp.
The presence of surfactin, bacillomycin, iturin D, and fengycin D biosynthetic genes, detected in the BS10.5 amplicons during PCR analysis, was confirmed from the WGS data. The BS10.5 genome analysis showed similarity identity above 85% for some biosynthetic genes clusters (BGC). Technically, BGC or biosynthetic genes are considered to be present when the similarity index is 65% and above (Van Der Voort et al., 2015). However, our report here also shows F I G U R E 6 Subsystem summary of the genome Bacillus velezensis NWUMFkBS10.5 predicted by SEED Viewer v2.0. Genomic features are colored according to their functional classification types (Overbeek et al., 2014) that a low percentage similarity of a BGC does not signify the absence of the predicted BGC in the genome being analyzed and this correlates with our identification of iturin gene during PCR, despite its predicted similarity identity in cluster 6 (BGC0001098) being below 65%. Coding region specific for fengycin is predicted to be present in cluster 6 at 86%, bacillomycin at 66%, and iturin at 53% similarity (Table 6). These genes were also detected by specific primers during the PCR-gel electrophoresis and ESI-Q-TOF MS analysis.  .
We applied the pangenomic analysis to locate genes that were responsible for metabolic activity and genes that were dispensable to the survival of NWUMFkBS10.5. The analysis showed that the strain shared core resemblance with the other analyzed Bacillus isolates.
Remarkably, biosynthetic clusters predicted to synthesize compounds such as sessilin, bananamides, cichopeptin, taiwachelin, tolaasin, basiliskamides, kalimantacin, and mersacidin were however identified in strain NWUMFkBS10.5 ( this is a rare report describing the genomic and biocontrol potential of a native B. velezensis strain peculiar to Africa. Although we did not corroborate our results with in planta experiments in this study, the exhaustive nature of this combinatorial in vitro study affirms the in planta viability of B. velezensis NWUMFkBS10.5. Lastly, the information gathered on NWUMFkBS10.5 will be valuable for its biotechnological manipulation and its probable development into a biofungicide in South Africa.

ACK N OWLED G M ENT
This work is based on the research supported by National Research Foundation, South Africa (Grant UID: 81192).

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interests.

AUTH O R S CO NTR I B UTI O N
AAA designed and performed the in vitro assays/experiments, molecular analysis, in silico genome analysis, analyzed the data, and drafted the manuscript. OSA contributed to the chemical characterization and its data interpretation. OOB participated in the experimental design, supervised the work, and provided funding. All authors contributed to the drafts, final version of the paper, and approved submission.

E TH I C S S TATEM ENT
This article does not contain any studies with human participants or animals performed by any of the authors.

DATA ACCE SS I B I LIT Y
The accession numbers of the 16S rDNA nucleotide sequences have been deposited at the NCBI GenBank (Supporting Information