Screening of Antimicrobial Activities and Lipopeptide Production of Endophytic Bacteria Isolated from Vetiver Roots

The exploration of certain microbial resources such as beneficial endophytic microorganisms is considered a promising strategy for the discovery of new antimicrobial compounds for the pharmaceutical industries and agriculture. Thirty-one endophytic bacterial strains affiliated with Bacillus, Janthinobacterium, Yokenella, Enterobacter, Pseudomonas, Serratia, and Microbacterium were previously isolated from vetiver (Chrysopogon zizanioides (L.) Roberty) roots. These endophytes showed antifungal activity against Fusarium graminearum and could be a source of antimicrobial metabolites. In this study, in particular, using high-throughput screening, we analyzed their antagonistic activities and those of their cell-free culture supernatants against three species of Fusarium plant pathogens, a bacterial strain of Escherichia coli, and a yeast strain of Saccharomyces cerevisiae, as well as their capacity to produce lipopeptides. The results showed that the culture supernatants of four strains close to B. subtilis species exhibited antimicrobial activities against Fusarium species and E. coli. Using mass spectrometry analyses, we identified two groups of lipopeptides (surfactins and plipastatins) in their culture supernatants. Whole-genome sequencing confirmed that these bacteria possess NRPS gene clusters for surfactin and plipastatin. In vitro tests confirmed the inhibitory effect of plipastatin alone or in combination with surfactin against the three Fusarium species.


Introduction
Microorganisms are an important resource for the mining of bioactive ingredients, which can then be used as drugs or leading compounds for the pharmaceutical and agrochemical industries [1,2]. For example, a number of bioactive compounds have been characterized in soil-derived microorganisms such as Streptomyces sp. [3]. Currently, discovering new bioactive compounds and/or new modes of action appears crucial to control multidrug-resistant microbes, which threaten the sustainable developmental goals set by the United Nations regarding human health and food security. In addition, due to societal expectations for more sustainable agriculture around the world and in Europe in particular, the characterization of bioalternatives to chemical inputs that are assumed to be less toxic to the environment and human health is vital [4,5]. Hence, exploring microbial diversity in less investigated environments could offer opportunities to discover new bioactive compounds and/or new modes of action.
In this context, endophytic microorganisms could be an attractive reservoir [6]. Indeed, endophytic microorganisms that occupy different ecological niches in the internal tissues of the plant are known to produce a high diversity of secondary metabolites, which can have pharmaceutical and agricultural applications [7,8]. In particular, these microorganisms produce several chemical classes of biomolecules with antimicrobial properties, such as ecomycin, kakadumycin, and munubicins [7,9,10].
Among the antimicrobial biomolecules, lipopeptides can be distinguished [11,12]. These biosurfactants can be produced by different microorganisms. For example, bacterial strains affiliated with the Bacillus genus were reported to produce a mixture of lipopeptides of the fengycin (plipastatin), iturin, and surfactin families [12]. These antimicrobial compounds are thought to play a key role in the antagonistic activities of these Bacillus strains against various fungal phytopathogens or foodborne pathogens [11,[13][14][15][16]. In addition to such direct effects, lipopeptides could aid in plant tissue colonization. Finally, some lipopeptides are also known to induce systemic resistance in plants, which is promising for the development of biocontrol solutions that carry a lower risk of resistance emergence [17,18]. To date, lipopeptides have been characterized in only limited numbers of endophytic bacteria [9][10][11]19].
Since the development of whole-genome sequencing, genome mining has become a powerful tool with which to investigate the potential of a strain for the synthesis of new natural compounds [6]. In particular, non-ribosomal peptide synthetase (NRPS) genes involved in the biosynthesis of lipopeptides are considered to be an indicator for the prediction of the biological activities or functions of certain bacteria [20,21]. NRPS operons often consist of regularly arranged modules, where one module is responsible for the adenylation, condensation, and thiolation of one amino acid, leading to an extension of peptide chains. Due to the modular structure of NRPS, homology analysis of predicted NRPS genes identified thanks to bioinformatic tools can help to predict peptide sequences [22]. A database dedicated to non-ribosomal peptides, named NORINE [23], was developed to help to identify such peptides and could therefore be used to assess the potential of endophytic bacteria to produce lipopeptides.
Vetiver (Chrysopogon zizanioides (L.) Roberty) is a perennial grass species used for its essential oil, mainly in the production of perfume. In our previous work, we identified 31 endophytic bacterial isolates derived from vetiver roots, showing strong inhibitory activity toward the phytopathogenic fungus Fusarium graminearum in in vitro culture [24]. These strains affiliated with Bacillus, Janthinobacterium, Yokenella, Microbacterium, Pseudomonas, Enterobacter, and Serratia genera could be a source of bioactive molecules with antimicrobial activities such as lipopeptides.
In this work, given the large number of bacterial candidates, we developed a new highthroughput workflow for the evaluation of the spectrum of antagonistic activities of these strains against reference yeast, bacterial strains, and phytopathogenic Fusarium species, as well as for the detection of lipopeptides that might be responsible for these activities. This workflow involves an automatic liquid handler for antimicrobial activity characterization and sample preparation and two mass spectrometry analysis techniques for molecule identification. Finally, to confirm the accuracy of the mass analysis, the genomes of the best lipopeptide-producing bacterial strains were sequenced and analyzed by bioinformatics, revealing the presence of NRPS gene clusters responsible for the production of antifungal lipopeptides.

Bacterial and Fungal Strains
A collection of 31 endophytic bacterial strains previously isolated from vetiver roots was used [24]. These strains were selected for their growth inhibition activity against F. graminearum in in vitro dual cultures and characterized taxonomically (Table 1). They were stored in a 15% glycerol solution at −80 • C. The 16S rDNA sequences of the 31 strains were deposited in GenBank, and the accession numbers are shown in Table 1.
One strain of Escherichia coli K12 (ATCC, Manassas, MA, USA) and one strain of Saccharomyces cerevisiae DSM1333 (DSMZ, Braunschweig, Germany) were also used. Both were stored at −80 • C.

Dual Culture Test of Selected Vetiver Bacterial Endophytes against F. oxysporum and F. culmorum
Bacterial strains were cultured in 2 mL of nutrient broth (NB) (Sigma Aldrich, St. Louis, MO, USA) inoculated with a single colony for approximately 24 h at 28 • C. Fifty µL of the liquid culture adjusted to an optical density at 600 nm of 1.0 was placed at 2 points 3 cm from an eight-mm-diameter mycelial plug from a 5-10-day-old culture of F. oxysporum or F. culmorum on PDA at 25 • C, placed at the center of a Petri dish (9 cm in diameter) containing 20 mL of PDA (Sigma-Aldrich, St. Louis, MO, USA). Three replicates were prepared for each bacterial strain. The plates were incubated at 25 • C. The diameters of the fungal colonies were measured at 6 and/or 11 days, according to the radial growth of the fungal species. The inhibition rate (%) was calculated with the following formula: 100 × (1 − dt/dc), where dt is the mean diameter of the Fusarium strains in the test plates and dc in the control plates.

Preparation of Supernatant Samples of Bacterial Endophytes
For evaluation of the inhibitory activity of cell-free bacterial supernatants in diffusion assays, bacteria were cultured at 28 • C 180 rpm for 24 h in 2 mL of King's B (KB, proteose peptone 20 g/L, KH 2 PO 4 1.5 g/L, MgSO 4 , (H 2 O) 7 1.5 g/L, glycerol 10 mL/L, pH 7.2) or NB (Sigma Aldrich, St. Louis, MO, USA) medium. Bacterial cultures were then centrifuged at 11,000× g and 4 • C for 10 min and the supernatant was filtered through a 0.2 µm polyethersulfone membrane (Acrobatic ® Syringe Filter with Supor Membrane). KB and NB media were chosen because they are often used for the growth of heterotrophic bacteria such as Pseudomonas and Bacillus [25,26]. Moreover, KB medium was also studied for the production of antifungal compounds by Serratia strains [27].
For mass spectrometry analysis, a single colony of each endophytic bacterial strain was cultured in 2 mL of KB medium at 28 • C 180 rpm for approximately 24 h. Then, 200 µL of the preculture was transferred into 10 mL of KB in 50 mL sterile tubes and cultured under the same conditions. After 24, 48, and 72 h, 1 mL of each culture was placed in 2 mL sterile tubes and the pH was adjusted to approximately 7 by dropping 1 M KOH. The tubes were then centrifuged at 10,000× g at 4 • C for 10 min, and 500 µL of the supernatant was collected in new sterile tubes. As a control sample, we used medium treated in the same way. The samples were kept at 4 • C for 4 to 5 days until analysis by MALDI-TOF.

Antagonism Test of Cell-Free Supernatants of the Selected Vetiver Bacterial Endophytes against Fusarium Species
Cell-free supernatants from a bacterial culture on KB or NB were tested against F. graminearum, F. oxysporum, and F. culmorum in 9-cm-diameter dishes containing 20 mL PDA. One hundred µL of the supernatant sample was placed in 8-mm-diameter wells 2 cm distant from the center of the plate. An 8-mm-diameter fungal plug from a 5-10-day-old culture on PDA at 25 • C of one of the 3 Fusarium species was inoculated at the center of the plate. The Petri dish was incubated at 28 • C in darkness. Each plate had three different samples and one control sample. The control sample corresponded to a non-inoculated medium filtered through a 0.2 µm membrane. Three independent replicates were prepared. The radii of the fungal colonies were measured after 3 days (F. graminearum and F. culmorum) and 8 or 11 days (F. oxysporum) of incubation at 28 • C. The inhibition rate was calculated with the following formula: Inhibition rate (%) = 100 × (1 − rt/rc). Herein, rt is the radius of test samples, and rc is the radius of the control sample in the same plate as the test samples.

Fully Automated Workflow for Cell-Free Supernatant Screening for Antimicrobial Metabolites
The workflow used in this study is presented in Figure 1. This workflow requires a liquid handler (Biomek FX p , Beckman Coulter, Brea, CA, USA). This robot is used first for the realization of antimicrobial activity in the plates. For this, the cell-free supernatants are placed in contact with the target strains at different dilutions, including the required control experiments, following the plate map described in Figure S1. The liquid handler was used to first distribute the samples into the different plates required for the liquid assays and MS analysis. This included the direct deposition of the samples onto the MALDI target, after their mixing with the matrix, again using the liquid handler. The plates were then incubated inside an off-robot incubation chamber, and then placed back on the liquid handler for mixing and reading of the plate after the incubation time. Finally, the liquid handler was used to prepare the samples that were analyzed by HPLC-TOF. After incubation at the proper temperature, the prepared plates were read for their OD using a spectrophotometric plate reader (Filtermax F5, Beckman Coulter, 600 nm) present on the robot deck. The liquid handler was also used in parallel for the conditioning of the cell-free supernatants and their automated deposition onto the MALDI target for the first step of mass spectrometry analyses. For this, 15 µL of sample was mixed with 15 µL of matrix (10 mg/mL HCCA matrix in 50/47.5/2.5: acetonitrile/MS-grade water/trifluoroacetic acid) and 2 µL of the mixture was precisely deposited onto each spot of the MALDI target. These analyses were then performed with a matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (MALDI-TOF, Autoflex Speed, Bruker) to rapidly detect high-molecular-weight secondary metabolites, with a special focus on the lipopeptides that are released by the strains into the culture supernatant. Finally, the samples were gathered into 96-well microtiter plates and then dried using a rotative vacuum concentrator, before being taken in half a volume of 100% methanol. The most promising samples were then analyzed by UPLC-Q-TOF mass spectrometry (Synapt G2Si, Waters) to detect the other potentially produced metabolites and perform high-resolution analyses of the lipopeptides detected with MALDI-TOF. The protocols used for the mass spectrometry analyses are detailed below in their respective paragraphs.
2.6. Antagonism Test of Cell-Free Supernatant of the Endophytes against E. coli and S. cerevisiae E. coli and S. cerevisiae were precultured in Mueller-Hinton (MH, Sigma-Aldrich, St. Louis, MO, USA) and YPG (yeast extract 10 g/L, bacto-peptone 20 g/L, glycerol 20 mL/L, pH 7.2) media, respectively. The optical density at 600 nm was adjusted to 0.1 before inoculation. The cell-free supernatants (1 mL) from the bacterial cultures in KB medium were transferred manually into a 2 mL DeepWell plate for further use. The supernatant was then mixed with the target strains using a Biomek FXp liquid handler (Beckman Coulter, Pasadena, CA, USA) in a polystyrene 96-well microtiter plate under sterile conditions to achieve the following plate design ( Figure S1).
The plates were lidded and then incubated at 37 • C for E. coli and 30 • C for S. cerevisiae for 24 h. The growth inhibition (GI) was calculated with the following formula: 100 × (1 − (ODt/ODg), where ODt is the mean OD600 value of the test samples and ODg is the mean OD600 value of the growth control samples (MH or YPG + inoculum). For yeast, because filtered KB medium also caused a certain reduction in yeast growth, the inhibition values were normalized to the values of the control samples (KB + YPG + inoculum).

Lipopeptide Detection with MALDI-TOF
For lipopeptide detection in the supernatants, MALDI-TOF mass spectrometry analyses were performed on an Autoflex Speed TM (Bruker Daltonics, Billerica, MA, USA). The molecular mass measurements were performed in the range of 700 to 3500 m/z and in automatic reflectron mode using FlexControl TM 3.4 software. The equipment parameters were as follows: voltage values of ion sources #1 and #2 set as 19.00 and 16.50 keV, respectively; voltage values of reflectrons #1 and #2 set as 21.00 and 9.50 keV, respectively; lens tension 8.00 keV; pulsed extraction 120 ns; laser intensity between 60 and 70%; smartbeam parameters set to ultra and sample rate; and digitizer settings set to 4.00 GS/s. The MS signals were acquired by summing 10,000 laser shots per spectrum. Prior to each analysis, the spectrometer was calibrated using the monoisotopic values of the manufacturer's Peptide Calibration Standard II calibrant (Bruker Daltonics, Billerica, MA, USA), containing Bradykinin Fragment 1-7, Angiotensin II, Angiotensin I, Substance P, Bombesin, Renin Substrate, ACTH clip 1-17, ACTH clip 18-39, Somatostatin 28. The calibrant was prepared by mixing 5 µL of calibrant-diluted mixture according to the manufacturer's specification with 5 µL of a 10 mg/mL HCCA matrix in 50/47.5/2.5: acetonitrile/MS-grade water/trifluoroacetic acid, and 2 µL was then spotted on a Polished Steel 384 MALDI target (Bruker Daltonics, Billerica, MA, USA). Samples were prepared by mixing 5 µL of supernatant with 5 µL of a 10 mg/mL HCCA matrix in 50/47.5/2.5: acetonitrile/MS-grade water/trifluoroacetic acid, and 2 µL was then spotted on a Polished Steel 384 MALDI target (Bruker Daltonics, Billerica, MA, USA). The plate was dried prior to analysis under a laminar airflow hood at room temperature. Mass spectra were visualized using FlexAnalysis software (version 3.4; Bruker Daltonics, Billerica, MA, USA).

Accurate Mass Measurement of Lipopeptides by Liquid Chromatography Coupled to High-Resolution Mass Spectrometry (LC-HRMS)
Ten µL of the collected supernatant was chromatographically separated at 30 • C in reversed-phase ultrahigh-performance liquid chromatography (RP-HPLC) using an Uptisphere Strategy C18-RP column (250 × 3.0 mm, 5 µm particles, Interchim, Montluçon, France) on a biocompatible ACQUITY UPLC system (Waters, Manchester, UK) and an acetonitrile gradient (flow rate 0.6 mL/min) with solvent A (0.1% formic acid/99.9% H 2 O (v/v)) and solvent B (0.1% formic acid/99.9% acetonitrile (v/v)) as follows: 5% solvent B for 5 min, then 5 to 100% solvent B for 35 min, followed by washing and equilibrating procedures with 100 and 5% solvent B for 5 and 10 min, respectively. The HPLC eluent was then ionized into the electrospray source of the SYNAPT-G2-Si mass spectrometer (Waters) at a voltage of 3 kV, using a desolvation gas (N 2 ) at a flow of 900 L.h-1, a nebulizer gas flow of 6.5 bar, and source and desolvatation temperatures of 120 and 500 • C, respectively. The mass spectrometer was previously calibrated using a sodium formate solution and measurements were performed in sensitivity, positive, and data-dependent modes using the proprietary MassLynx software (version 4.1, Waters). A maximum of 5 precursor ions with an intensity threshold of 1000 were selected for the fragmentation (MSMS) by collision-induced dissociation (CID) with specified voltages ranging from 10 to 15 V and from 20 to 100 V for the lower-and higher-molecular-mass ions, respectively. Mass data were collected in the measurement range of 200 to 2000 m/z for MS and of 10 to 2000 m/z for MSMS using lock mass correction with 556.632 m/z, corresponding to simply charged leucine enkephalin.
Genomic DNA libraries were prepared using the Nextera XT Library Prep Kit (Illumina, San Diego, CA, USA) following the manufacturer's protocol, with the following modifications: input DNA was increased 2-fold, and PCR elongation time was increased to 45 s. DNA quantification and library preparation were carried out on a Hamilton Microlab STAR automated liquid handling system (Hamilton Bonaduz AG, Rapperswil-Jona, Switzerland). Pooled libraries were quantified using the Kapa Biosystems Library Quantification Kit for Illumina. Libraries were sequenced with Illumina sequencers (HiSeq/NovaSeq) using a 250 bp paired-end protocol. Reads were trimmed using Trimmomatic 0.30 with a sliding window quality cutoff of Q15 [28]. De novo assembly was performed on the samples using SPAdes version 3.7 [29], and the contigs were annotated using Prokka 1.11 [30].
The genomes of four B. subtilis strains were submitted to antiSMASH (version 6.0.1) [31] to predict the biosynthesis gene clusters of the secondary metabolites.

Effects of Surfactin and Plipastatin on the In Vitro Growth of Fusarium Species
Surfactin and plipastatin (synonym fengycin) were purchased from Sigma-Aldrich (ref. S3523, SMB00292, respectively). Surfactin was dissolved in dimethyl sulfoxide (DMSO) at 100 mg/mL and then diluted in sterile water to 1 mg/mL. Plipastatin was dissolved in 0.1% DMSO at 1 mg/mL and then filtered through a 0.2 µm pore membrane. A mixture of surfactin and plipastatin was prepared at a ratio = 1:1 (w/w). The lipopeptides were then diluted to their final concentrations of 5, 10, 25, 50, 100, 200, and 400 µg/mL. As a control, 1, 0.1, and 0.55% DMSO were used for the surfactin, plipastatin, and mixed solutions, respectively. An in vitro assay was performed with the plate layout described in Section 2.4., applying 40 µL lipopeptide solution per well (diameter 6 mm).

Data Analysis
The figures and statistical tests were performed with R studio with R version 3.6.3. To compare the growth inhibition rate of surfactins, plipastatins, and mixed solutions, the Tukey-Kramer test was performed for each concentration, using the multcomp package.
In all the inhibition assays against plant pathogens, E. coli, and yeast, the negative inhibition values were replaced with 0%.

Antifungal Activities of Vetiver Endophytic Bacterial Strains against Plant Pathogens
In this study, 31 selected endophytic bacteria from vetiver roots were investigated (Table S1). These bacteria, affiliated with the Bacillus, Enterobacter, Janthinobacterium, Pseudomonas, Serratia, Microbacterium, and Yokenella genera, have previously shown high inhibition rates against F. graminearum in in vitro cultures (Table S1, [24]). In this study, we evaluated their ability to inhibit the mycelial growth of two other well-known Fusarium plant pathogenic species, F. culmorum and F. oxysporum (Figure 2 and Figure S2). Depending on the bacterial genus, the inhibitory rate varied from 1.1 to 53.4% and 0 to 55.3% against F. culmorum and F. oxysporum, respectively. Bacteria affiliated with the genera Serratia and Bacillus showed the highest inhibitory rates against F. culmorum and F. oxysporum (Figure 2). Serratia strains, closely related to S. grimesii species, showed similar inhibition rates against F. culmorum (approximately 47%) and F. oxysporum (approximately 48%). On the other hand, except for the strain affiliated with B. cereus, Bacillus strains belonging to B. subtilis and B. tequilensis showed a higher inhibitory rate against F. culmorum (approximately 45%) than F. oxysporum (approximately 39%). It should be noted that although bacterial strains affiliated with the Serratia or Bacillus genus showed similar capabilities to inhibit the mycelial growth of F. culmorum, the appearance of the fungal colonies differed, depending on the bacterial strains with which the fungus interacted ( Figure S2). For other bacterial strains, the strain affiliated with genus Enterobacter had a good inhibitory rate against F. culmorum (47%), while bacteria affiliated with Janthinobacterium, Pseudomonas, Yokenella, and Microbacterium genera showed overall lower inhibitory rates against F. culmorum and F. oxysporum.

Antimicrobial Activities of the Cell-Free Crude Supernatants of the Vetiver Bacterial Endophytes
The antimicrobial (antifungal and antibacterial) activities of filtered culture supernatants from the endophytic bacteria were then evaluated. For this purpose, KB and NB media were used to grow the bacteria.
The inhibitory activities of the bacterial supernatants against Fusarium species were tested by diffusion assays (Figure 3). The results obtained showed contrasting inhibitory effects of the culture supernatants depending on the bacterial strains, with inhibition rates ranging from 0% to almost 43% (Figure 3, Table S1). Only supernatants from strains affiliated with B. subtilis (strain numbers 22, 23, 26, and 28) and, specifically, those from liquid culture in KB medium displayed a significant inhibitory rate against the three Fusarium pathogenic species (Figure 3, Table S1). The inhibitory effects of the supernatants from bacterial culture in NB medium varied depending on the B. subtilis strain and the Fusarium species ( Figure S3). The inhibitory activities of supernatants of the bacterial cultures in KB medium were also tested against E. coli and S. cerevisiae in 96-well plates. The supernatants were serially diluted 2 to 8 times. For E. coli, only three strains affiliated with Enterobacter (strain number 29) and Bacillus (strains number 26 and 28) showed little to moderate antibacterial activity (Table 1, Figure S4). The Enterobacter-affiliated strain showed the highest inhibition rate (37.6%, observed at the 2-fold dilution), followed by the two B. subtilis-affiliated strains, which showed inhibition rates of 16 and 10% at the 8-fold dilution, respectively ( Figure S4). Regarding the test with S. cerevisiae, no strain showed a growth inhibition effect ( Figure S4, Table 1).

Screening for Lipopeptides in Supernatant Samples of Vetiver Endophytes by MALDI-TOF
In a preliminary approach, to screen lipopeptide production by endophytic bacteria, cell-free supernatants from 24 h cultures of the 31 endophytic strains were analyzed using MALDI-TOF, and the main results are presented in Figure 4. KB medium was used as a control and subjected to MALDI-TOF analysis to remove molecules derived from the medium. The peaks at m/z = 909.0, 925.0, 1019.0, 1020.8, 1034.9, 1036.4, 1173.1, 1188.9, and 1190.3 were, therefore, eliminated. Of the 31 strains analyzed, five belonging to the Bacillus genus yielded interesting profiles, as shown in Figure 4. Four strains affiliated with B. subtilis (strain numbers 22, 23, 26, and 28) and whose supernatants showed significant antifungal activities against Fusarium species had similar peak profiles within the mass ranges of m/z = 1020 to 1550 (Figure 4). The mass spectra of these strains showed two distinct groups of peaks within the mass ranges of m/z = 1020 to 1080 and m/z = 1450 to 1550. The appearance of this group of peaks appears to be similar to that of lipopeptides. Furthermore, we observed that 3 of the 4 strains affiliated with B. subtilis (strain numbers 23,26,28) and the strain affiliated with B. tequilensis (strain number 25) also shared a group of peaks in the mass range of m/z 3400 to 3480. The latter mass seems to correspond to the production of subtilosin A, a bacteriocin produced by certain strains of Bacillus. In-depth analysis of the culture supernatants at 48 and 72 h was therefore undertaken for these five strains. Compared to strains affiliated with B. subtilis (strain numbers 22, 23, 26, and 28), some peaks (such as the ones at m/z = 1030, 1088, 1485, and 1515) were not detected or were found only after 72 h of cultivation (peaks at m/z =1044 to 1074) in the supernatant of the strain belonging to B. tequilensis (strain number 25) ( Table 2). The detected peaks were considered to potentially correspond to two lipopeptide families, surfactins and plipastatins. The peaks observed in the supernatants of all Bacillus strains after 72 h of culture at m/z = 1030.72, 1044.74, and 1058.75 differed by 14 Da, suggesting a group of compounds with different fatty chain lengths. These peaks were attributed to sodium adducts of surfactin, with fatty acid chains of C13, C14, and C15, respectively ( Table 2). As this analysis did not allow us to identify the amino acid at position 7 of the peptidic cycle, it is also possible that these masses correspond to the C14, C15, and C16 isoforms of surfactin A with a valine at position 7. Moreover, the peaks at m/z = 1060.71, 1074.74, and 1088.73 also differed by 14 Da and were attributed to potassium adducts of surfactin C14, C15, and C16 or to the C14, C15, and C16 isoforms of surfactin A with a valine at position 7. These compounds were found in the supernatants of strains 22, 23, 26, and 28 related to B. subtilis after 48 h of culture, whereas they were only detected after 72 h of culture in strain 25 belonging to B. tequilensis.
The peaks observed at m/z = 724.44, 746.43, and 762.42 were identified as the dicharged form of plipastatin A C15 or plipastatin B C13 with hydrogen, sodium, and potassium adducts, respectively. The peaks at m/z 781.35 were identified as the dicharged form of the plipastatin A C16 or plipastatin B C18 isoform with unsaturated fatty acid chains and sodium adducts. All plipastatin isoforms were observed in the supernatants of the five different strains at 48 and 72 h of culture time. The observed peaks at m/z = 1485.86, 1515.89, and 1571.63 might be assigned to potassium or sodium adducts of plipastatin A or B ( Table 2)

Accurate Mass Measurement by LC-HRMS
The culture supernatants of the five Bacillus strains (22, 23, 25, 26, and 28) were subjected to LC-HRMS analysis to obtain more resolutive mass signals and to assign the correct masses to the lipopeptides. Different masses of surfactins and plipastatins were clearly detected in the five Bacillus strains, confirming the assignments proposed in Table 2. Figure 5 represents, as an example of the results, the isotopic clusters of surfactin A C13 ( Figure 5A) and C15 ( Figure 5B) and plipastatin A C17 or B C15 and A C18 or B C16 ( Figure 5C) and A C19 or B C17 ( Figure 5D) extracted from the analysis of supernatant of the B. subtilis strain 23. MSMS analysis of the different forms of surfactin confirmed the predominant presence of surfactin A C15 with leucine at position 7 ( Figure 6, fragmentation pattern in blue). However, MSMS spectra revealed the presence, with a lower intensity, of surfactin B C16 with valine at position 7 ( Figure 6, fragmentation pattern in red). Regarding plipastatin, MSMS analysis revealed no informative spectra due to the poor ionization of the different forms of this lipopeptide.

Genome Sequencing and NRPS Cluster Characterization
Strains affiliated with Bacillus and closely related to B. subtilis were sequenced. Analysis of draft genomes of the four sequenced Bacillus strains showed that these genomes are composed of 4,042,990 to 4,269,420 bp, included in 14 to 26 contigs, with L50 ranging from 377,686 to 2,301,381 bp. GC content ranged from 43.2 to 43.7%. Draft genome annotation identified 4036 to 4338 coding sequences and 83 to 84 tRNA genes. Potential secondary metabolite clusters were screened using the genome mining tool antiSMASH. The Bacillus strain close to the B. tequilensis species (strain 25) was excluded from this analysis as the culture supernatant of this strain did not show any antifungal activity against Fusarium sp. The results showed that the genomes of the four strains might harbor four NRPS gene clusters (Table 3). These predicted clusters included genes encoding lipopeptides, surfactin, and plipastatin (Figure 7), already detected in the supernatants. Notably, the known peptide sequences of plipastatins have glutamine at the eighth position, while antiSMASH affiliated glutamic acid at this position of the putative plipastatin B ( Table 3). The difference between glutamine and glutamic acid gives an additional one to the monoisotopic mass of known plipastatin B. The masses obtained during the HRMS analysis are precise enough to confirm the presence of glutamine at position 8 of the plipastatin molecule. Concerning the surfactin, both isoforms with leucine and valine at position 7 are produced by Bacillus strains belonging to B. subtilis species.  The other clusters might encode the antimicrobial dipeptide bacilysin, the siderophore bacillibactin, and the bacteriocin subtilosin A.

Antagonistic Activities of Surfactins and Plipastatins against Fusarium Species
We analyzed the effects of surfactin and plipastatin identified in the supernatant samples of B. subtilis (strain numbers 22, 23, 26, and 28) and B. tequilensis strains (strain number 25) alone or combined against F. graminearum, F. culmorum, and F. oxysporum. Surfactin had no clear inhibitory activity against the in vitro mycelial growth of F. graminearum and F. culmorum at any concentration ( Figure S5). In contrast, plipastatin exhibited antifungal activities from the 25 µg/mL concentrations for the two Fusarium species ( Figure S5). Thus, for F. graminearum, plipastatin showed an inhibition rate of approximately 15% at a concentration of 25 µg/mL, which increased to approximately 30% for higher concentrations. For F. culmorum, plipastatin displayed better antifungal activity, with an inhibition rate of approximately 40% from the concentration of 25 µg/mL. For F. oxysporum, inhibition effects were not clear up to concentrations of 100 µg/mL for all solutions. In contrast, plipastatin showed an inhibition rate of approximately 40% from the concentration of 200 µg/mL ( Figure S5).
Higher antifungal activities were recorded when plipastatin was tested in combination with surfactin, with growth inhibition of F. graminearum and F. culmorum observed from 10 µg/mL. These effects were not observed with F. oxysporum ( Figure S5).

Discussion
There is an increasing need to develop original biocontrol solutions in agriculture and to identify new drugs for use as pharmaceuticals to address various social concerns. Microorganisms and/or the associated bioactive compounds that they can synthesize might constitute a promising reservoir for this purpose.
In the present study, we investigated in greater detail the potential of 31 endophytic bacterial strains isolated from vetiver, which showed strong growth inhibitory activity against F. graminearum [24]. To better investigate the potential of these strains, we set up an automated high-throughput screening workflow combining antimicrobial testing on microtiter plates and mass spectrometry analysis using MALDI-TOF and LC-HRMS. Thus, these various bacteria were characterized for their antimicrobial activities against microorganisms, including phytopathogenic fungi affiliated with the genus Fusarium, yeast, and bacteria. They were also analyzed for their ability to produce lipopeptides that might be involved in any antagonistic activities.
The results that we obtained showed that strains affiliated with S. grimosii, B. subtilis, and B. tequilensis displayed the strongest growth inhibition in an in vitro dual culture against F. culmorum and F. oxysporum. These results confirm that S. grimosii, B. subtilis, and B. tequilensis strains are useful antagonists of plant fungal pathogens [27,33]. Their antagonistic activities are known to be related to the synthesis and release of a broad range of secondary metabolites with antimicrobial activities. Analysis of the antimicrobial activities of the cell-free culture supernatants of these bacterial strains showed that only the supernatants of B. subtilis (strain numbers 22, 23, 26, 28) retained antagonistic activity toward Fusarium species but also against E. coli. These results may suggest different antibiosis mechanisms between Serratia and Bacillus strains, whose metabolism is highly variable, especially with regard to the type of antimicrobial compounds produced (soluble and/or volatile compounds) [34]. It should also be noted that the culture supernatants were obtained from mono-cultivated bacteria under non-limiting resource conditions. As a result, some of the compounds involved in the antagonistic activities observed in dual cultures may not have been produced or produced only weakly. Indeed, it has been shown that resource competition and inter-and intraspecific microbial interactions are often implicated in the induction of the biosynthesis of antimicrobial compounds [35][36][37].
By using MALDI-TOF analyses, we characterized inhibitory compounds in the cell-free supernatants of Bacillus strains. We highlighted the very probable mass of two lipopeptide families. Surfactins and plipastatins were detected in supernatants of the four strains belonging to B. subtilis (strain numbers 22,23,26,28) and the strain belonging to B. tequilensis (strain number 25). These results are consistent with the presence of NRPS gene clusters detected in the genomes of Bacillus subtilis strains. Both molecules are already described to be produced by B. subtilis species and the closely related species B. tequilensis [38]. Although surfactins are rarely reported as antibacterial molecules except against Legionella species [39,40], they could also have antifungal activities and act synergistically with plipastatins (synonyms of fengycins) to potentiate the antagonistic activities of some Bacillus strains against phytopathogenic fungi such as F. graminearum or F. oxysporum [41,42]. In our study, these molecules seem to explain the antagonistic effects of the B. subtilis strains toward the three phytopathogenic Fusarium species. Other lipopeptides, such as iturins and bacillomycin, have been thought to play a role in the antagonistic activities of some Bacillus strains against Fusarium species [11,14,43]. In our case, these compounds were not detected in the supernatant samples of B. subtilis strains (strain numbers 22,23,26,28) by MALDI-TOF, LC-HRMS, or by whole-genome analysis. This reflects the metabolic variations between bacterial strains belonging to the same species that may result from the different habitats in which these microorganisms have grown before being isolated.
Finally, it is interesting to mention that our B. subtilis and B. tequilensis strains, except for B. subtilis strain number 22, showed similar MS peak profiles at m/z over 3000. These compounds could correspond to some bacteriocins, such as a subtilosin A or a variant of subtilosin, which have a molecular weight at m/z of approximately 3400. Our chromatogram showed minor species at m/z 3438 that correspond to potassium adducts, as previously described [44,45]. These compounds could explain the activity of B. subtilis strains against E. coli.

Conclusions
This paper revealed that four B. subtilis strains isolated from vetiver roots originating from Reunion Island have interesting antagonistic activities against several Fusarium plant pathogenic species in in vitro cultures. These inhibitory activities could partly be due to the production of lipopeptides, surfactins, and plipastatins. These strains are promising for the protection of plants against Fusarium diseases. Further studies are now needed to evaluate the crop protection activity of the four B. subtilis strains and to study their mode of action in planta. It would also be interesting to analyze, for the other strains in this study, such as those affiliated with Serratia or Pseudomonas, the metabolic traits that may participate in their antagonistic activities in in vitro dual cultures.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/microorganisms10020209/s1, Figure S1: The layout of 96-well plates for growth inhibition test against E. coli and S. cerevisiae; Figure S2: Dual culture test of vetiver endophytes and F. culmorum (a) and F. oxysporum (b); Figure S3: Growth inhibition rate of the cell-free supernatants of vetiver endophytic bacterial strains in NB against F. graminearum, F. culmorum, and F. oxysporum; Figure S4: Growth inhibition rate of the cell-free supernatant of vetiver endophytic strains against E.coli (a) and S. cerevisiae (b); Figure S5: Growth inhibition rate of the solutions of commercial surfactins, plipastatins, and mixtures of the two lipopeptides against F. graminearum, F. culmorum, and F. oxysporum; Figure S6: Liquid handler protocol for the high-throughput screening assays; Figure S7: Liquid handler protocols for the MALDI-TOF target preparation; Table S1: Summary of the in vitro growth inhibition activities of the 31 vetiver endophytes against Fusarium plant pathogens with the bacterial cells or the cell-free supernatants.