Soil samples were collected from the drought-affected region of rhizospheric soil of Arhar [Cajanus cajan (L.)], maize (Zea mays L.), and wheat (Triticum aestivum L.) plants growing in different states, viz Uttar Pradesh (UP), Bihar, and Madhya Pradesh (MP) of India, and stored at 4°C for further analysis. The soil of the collection sites had good permeability and was silty loam in nature.

Bacteria were isolated from rhizospheric soil samples by adopting the technique of serial dilution. Required dilution was spread on yeast extract mannitol agar (YEMA) media containing per liter of distilled water: yeast extract, 1.00 g; KH₂PO₄, 0.50 g; mannitol, 10.00 g; MgSO₄.7H₂O, 0.20 g; NaCl, 0.10 g; and agar, 15.00 g, adjusting pH 6.8–7.0. Plates were kept in an incubator at 28°C for 24–48 h. Colonies were handpicked from the master plates and sustained as pure cultures in YEMA media with regular transfer to fresh media and stocked for further analysis. The isolates obtained were characterized for Gram staining, colony morphology, and biochemical characters and screened for their antimicrobial activity.

Among all the 53 bacterial isolates, the strain JRBHU6, screened for its potential biocontrol activity, was subjected to molecular characterization. The DNA was isolated using a spin column kit (Himedia, India). The universal bacterial primers, forward: −530F (5′-GCTCTAGA GCTGACTGACTGAGTGCCAGCMGCCGCGG-3′); reverse: −800R (5′TACCAGGGTATCTAATCC3′) were used for the amplification of 16S ribosomal RNA (rRNA) gene. The PCR was carried out in 50 μl PCR mixtures including genomic DNA (gDNA) (∼50 ng), forward primer (100 ng), reverse primer (100 ng), 10 mM deoxyribonucleotide triphosphates (dNTPs) mix (2 μl), 10 × AMTaq Pol. buffer (5 μl), and AMTaq polymerase enzyme (3 U). Cycling conditions were as follows: initial denaturation at 94°C for 5 min, 35 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 1.30 min, and a final extension of 5 min at 72°C. Purified amplicons were sequenced by Sanger method in an ABI 3500xL genetic analyzer (Life Technologies, United States). Sequencing files (.ab1) were edited using CHROMASLITE (version 1.5) and further analyzed by Basic Local Alignment Search Tool (BLAST) with closest culture sequence retrieved from the National Centre for Biotechnology Information (NCBI) database that finds regions of local similarity between sequences. The 16S rRNA gene sequence obtained for the isolated bacterial strain JRBHU6 was compared with other bacterial sequences using NCBI mega BLASTn¹ for their pairwise identities. Multiple alignments of these sequences were carried out by ClustalW 1.83 version of EBI² with 0.5 transition weight. The 16S rRNA sequences of the bacteria was submitted to NCBI Genbank for obtaining the accession number.

Cultures of bacterial isolates were spot inoculated on colloidal chitin agar medium and kept in an incubator at 30°C for 5 days to check the activity of chitinase. The test was marked as positive if there was a zone of clearance around the colony (Sharma et al., 2020).

The culture was streak inoculated on cellulose agar containing carboxymethyl cellulose (CMC) or cellulose as sole carbon source at pH 7.0 and incubated at 30°C for 48 h. The isolate showing growth on cellulose agar was considered cellulase positive. After incubation, plates having colonies were repeatedly treated with 1% Congo Red with intermittent washing by 1 M NaCl solution. Cellulase producers developed a clear zone around the colony (Apun et al., 2000).

To extract the different bioactive compounds, the isolated bacterial cultures were first inoculated into minimal media broth and incubated at 37°C for 5 days on orbital shaker. The inoculum media comprised of (g/L) 20 g glucose, 10 g peptone, 0.5 g NaCl, 2 g CaCO₃, 2 g KNO₃, 0.8 g K₂HPO₄, 0.7 g MgSO₄.7H₂O, 0.5 g KCl, 1 g yeast, 0.002 g MnCl₂, and 0.002 g FeSO₄ and was adjusted at pH 7. The bacterial cells were then separated from the liquid medium by using a refrigerated centrifuge at 12,000 g for 10 min. The light brown colored supernatant was used to extract the bioactive compounds in 350 ml of different organic solvents, viz methanol, ethanol, ethyl acetate, and acetone each by liquid-liquid extraction method, and the extract was concentrated using a rotary evaporator. The bioactive components present in the different organic crude extracts were then tested for their antibacterial activity by Kirby–Bauer agar well diffusion assay against diverse bacteria and by dual culture plate assay for fungi, which were previously available in our lab. The microbes selected to determine the antimicrobial activity of the crude extracts of Burkholderia seminalis JRBHU6 were Gram-negative bacteria (Pseudomonas aeruginosa, Escherichia coli DH5α, Klebsiella pneumonia, and Shigella boydii), Gram-positive bacteria (Staphylococcus aureus) and fungi, viz Fusarium oxysporum, Aspergillus niger, Microsporum gypseum, Trichophyton mentagrophytes, and Trichoderma harzianum. Nutrient broth and nutrient agar media were used throughout the experiments for bacteria. For fungus, Sabouraud Dextrose broth and Sabouraud Dextrose agar (SDA) were used during the experiments. Amphotericin B for fungus and streptomycin for bacteria were used as positive control. Stock solutions were prepared by dissolving 100 mg of dried crude extract in dimethyl sulfoxide (DMSO). From these stock solutions, 50 μg/ml of working concentration was prepared. Thirty microliters (10⁹ CFU ml^–1) of test microorganism was spread on nutrient agar plates. Four wells (8 mm in diameter) were cut into the agar media with a sterilized cork borer; 50 μl of crude extract of each solvent containing 50 μg/ml was poured into the respective well. Fifty microliters of antibiotic (50 μg/ml) and DMSO (50 μl per well) were also poured into one well each per plate as positive and negative control, respectively. Inoculated plates were then incubated in biological oxygen demand (BOD) incubator at 28°C for 24 h, and the zone of inhibition was measured in millimeters. The effect of methanolic crude extract on cell viability was further confirmed by epifluorescence microscopy (Nikon INTENSILIGHT C- HGFI, JAPAN) following Tawakoli et al. (2013) in different pathogenic bacteria. The number of cells was observed in randomized microscopic ocular grid fields (0.0156 cm² grid area) at 40x magnification. Two different light filters were used for discriminating the dead and living cells (TRICT filter for the visualization of dead cells: excitation 528–553 nm, emission 590–650 nm; FITC filter for live cells: Excitation 465–495 nm, emission 515–555 nm).

All the antimicrobial activities were performed thrice and the values represented are mean of three replications ± standard error (SE). The data was subjected to analysis of variance (ANOVA) using SPSS Ver. 16 (SPSS Inc., Chicago, IL). The treatment mean values were compared with Duncan’s multiple range tests at P ≤ 0.05 significance level.

The methanolic crude extract was fractionated on silica gel thin layer chromatography (TLC) plates. Different mobile phases were prepared separately using mixtures of solvents, methanol, and water (2:1 v/v) used as standard mobile phase. Methanolic crude extract was applied as a spot on TLC plate and left for complete separation using the previously prepared mobile phases. The plates stood for solvent evaporation, and the R_f value of spots was recorded using ordinary and UV lamps, respectively.

High-performance liquid chromatography (HPLC), a very popular and widely used method to separate, identify, and quantify compounds from a mixture, was used to identify the antimicrobial compounds in the bacterial extract. The compounds separated through TLC were curettaged and was subjected to reverse-phase HPLC (RP-HPLC) (Rao et al., 2017). The purity of the antimicrobial compounds was analyzed by analytical HPLC supported with Waters Spherisorb 5 μm ODS2 4.6 mm × 250 mm analytical cartridge (C-18 column) on a Waters 515 pump, isocratic reverse phase system with a 2,998 photodiode array detector at 210 nm, and the range given was 190–600 nm (Elshafie et al., 2017a, c). The flowrate was 1.0 ml/min, and additional UV detector was measured at 254 nm using Empower 2 software. Methanol and water (2:1) were used as the mobile phase. The purified bacterial methanol extract was mixed with HPLC-grade methanol and filtered by using a 0.22-micromillipore membrane filter before injecting into the injection port. The samples were run for 15 min, and the retention time was noted; based on the percentage of the peak area, the purity of each compound was determined.

Gas chromatography–mass spectrometry (GC/MS) analysis was conducted as explained by Gohar et al. (2010) with slight differences in analytical conditions. HPLC-eluted compounds were analyzed using Thermo Scientific gas chromatography coupled with ITQ 1100 mass detector and X-Caliber software and National Institute of Standards and Technology (NIST) spectral data (GCMS, Thermo Fisher Scientific) with DB-5 MS capillary column (30 μm × 0.25 μm internal diameter and 0.25 μm film thickness). Analytical chromatographic conditions were as follows: 2 μl sample injected; carrier gas helium at 1 ml/min; injector and transfer line temperatures used were 220 and 280°C, respectively; over temperature program included 50°C–2 min hold, 150°C/min rate; increased to reach 160°C and remained at temperature for 0 min hold, 50°C/min rate; increased to reach 200°C and remained at temperature for 1 min hold, 2°C/min rate; increased to reach 230°C and remained at temperature for 1 min hold, 80°C/min rate; increased to reach 285°C and remained at temperature for 6 min hold; split ratio = 1:50; ionization energy, 70eV; mass range, 50–650. The composition was determined by comparing peak retention times with those of reference standards.

Using JASCO Fourier-transform infrared (FTIR) spectrometer, the molecular structure of the compound was partially identified. HPLC eluted compounds were dissolved in dry KBr, which was later subjected to FTIR spectroscopy. The measurement was carried out at infrared spectra in the region 600–4,000 nm cm^–1, using ATR accessories.

¹H NMR spectra were acquired by dissolving the purified compound eluted from HPLC in deuterated (d₆) DMSO solvent at a concentration of 1 mg ml^–1 and analyzed on a JEOL-500 MHz NMR spectrometer.

This target prediction method is based on the principle that the similar chemical structures tend to perform similar biological functions (Willett et al., 1998; Bender and Glen, 2004). Thus, the chemical structures that share structural or substructural similarity will interact with similar protein target.

PharmMapper³ (Wang et al., 2017) web service was used for pharmacophore-based target predictions. In order to derive a dataset, pharmacophores derive structures complexed with small molecules from Binding DB, DrugBank, PDB Bind, and the PDTD databases. Results are discussed in “In silico Target Identification.”

Burkholderia seminalis JRBHU6 showed a clear halo zone around colony-confirming solubilization of colloidal chitin in nutrients agar media supplemented with 1% colloidal chitin after incubation at 30°C for 5 days (Figure 1A).

Burkholderia seminalis JRBHU6 showed a zone of clearance around colony indicating solubilization of carboxymethylcellulose (CMC) in nutrients agar media, confirming its cellulose-degrading ability (Figure 1B).

Burkholderia seminalis JRBHU6 showed strong antifungal activity against all the fungal strains, viz F. oxysporum, A. niger, M. gypseum, T. mentagrophytes, and T. harzianum incubated at 25°C for 7 days (Figure 1C). The high inhibitory activity of B. seminalis JRBHU6 against all the selected soil fungi indicated the secretion of antifungal compounds into the agar medium in dual culture plate assay which restricted all the selected fungi to less than 50% area of the plate. The antagonistic behavior of B. seminalis JRBHU6 was evident from the suppressed mycelial growth along with the arrested spore/sclera formation in the fungi.

The inhibitory action of B. seminalis JRBHU6 crude extract obtained in different organic solvents was observed in terms of diameter of the inhibition zone formed around the agar well by diffusion of antimicrobial substances following Kirby–Bauer agar well diffusion method. The zone of clearance around the well indicated the positive antibacterial activity given by the test sample. Maximum inhibition zone formed by the derived crude extracts of B. seminalis JRBHU6 was observed in methanolic crude extract tested against S. aureus, which was 22.4 mm and in ethanolic crude extract (20.4 mm). The methanolic crude extract proved better than other organic extract in exhibiting a broad spectrum inhibition activity against the gram-negative P. aeruginosa (18.8 mm), E. coli DH5α (17.8 mm), S. boydii (15.4 mm), and K. pneumonia (14.8 mm) strains. The diameter of the inhibition zone formed by the methanolic crude extract against each microorganism was found to be equal or greater than that of the standard antibiotic streptomycin (22.2 mm) used in the assay (Figure 2). The different bacterial pathogenic cultures exposed to 18 hrs in methanolic crude extract of B. seminalis JRBHU6 were also examined by epifluorescence microscopy to check their viability. Epifluorescent microscopic images clearly confirmed the bactericidal nature of the crude extract (Supplementary Figure 2).

Separation of compounds present in B. seminalis JRBHU6 methanolic crude extract was best observed in methanol and water (2:1 v/v) solvent system as compared to other tested solvent systems. A single brown spot was separated. The Rf values for the colored spot was 0.71.

The bioactive compounds in methanolic extract was purified by HPLC, which showed a characteristic peaks with retention time 1.961 min, 2.245 min, 5.103 min, and 6.121 min at PDA 210 nm (Figure 3). The eluted active compounds at retention time 2.245 min having the highest peak area was subjected to further characterization by GC-MS, FTIR, and ¹H-NMR studies.

GC-MS analysis was performed to identify the bioactive molecules present in the methanolic extract. GC-MS analysis showed the presence of multiple compounds in the sample based on molecular weight, retention time, and molecular formula when compared with the NIST library. The peak area corresponds to the quantity of the compound present in the sample. Based on GC-MS spectra (as depicted in Figures 4, 5), the two most intense peaks at 13.46 min corresponding to PPDH present in the sample showing mass of 154 Da with molecular formula C₇H₁₀O₂N₂ [pyrrolo(1,2-a)pyrazine-1,4-dione,hexahydro] and at 16.80 min corresponding to PPDHMP present in the sample analyzed showing mass of 210 Da with molecular formula C₁₁H₁₈N₂O₂ [pyrrolo (1,2-a)pyrazine-1,4-dione, hexahydro-3(2-methylpropyl)] were identified.

The major peak obtained for Pyrrolo[1,2-a]pyrazine-1,4-dione,hexahydro (PPDH) was further studied through FTIR spectroscopy. FTIR spectrum of PPDH showed the characteristic peaks at 3,400–3,140 cm^–1 corresponding to amide N–H group, 2,940 cm^–1 corresponding to alkyl C–H group, and at 1,638 cm^–1 indicating C = O stretching of amide, respectively (Figure 6A).

The ¹H-NMR spectrum (JEOL 500 MHz, solvent; DMSO-d₆) showed signals at δ = 0.8 (2H, q), 1.205 (2H, m), 1.809 (2H, t), 3.108 (2H, s), and at 3.411(1H, t) ppm corresponding to aliphatic protons and peak at 7.236 ppm corresponding to N–H proton (Figure 6B).

The isolated bioactive PPDH fully characterized by GC-MS, HPLC, FTIR, and NMR techniques was confirmed as pyrrolo(1,2-a)pyrazine-1,4-dione,hexahydro (Figure 6B).

The target prediction method was based on the principle that the similar chemical structures tend to perform similar biological functions (Willett et al., 1998; Bender and Glen, 2004) and therefore will interact with similar protein target. Keeping this in view, the smile structure of PPDHMP [(CC(C)CC1C(= O)N2CCCC2C(= O)N1)] was submitted to PubChem database (with similarity ≥ 70%) for chemical similarity search. Results showed the chemical similarity with our studied PPDH. In this regard, chitinase B could also be the likely target for PPDHMP.

All the predicted targets along with their functions are given in Table 1. Interestingly, all the identified targets were enzymes. Results were analyzed based on their Z-score value (Table 1). A large positive Z-score indicates high significance of the target to a query compound. All predicted targets with a Z-score > 1.0 were considered likely targets. Interestingly, 1W1P was one of the predicted targets along with human and bacterial proteins (Table 1). All likely targets were further subjected to docking analysis.

Ligand-based docking technique was implemented in this work to study the binding interactions between identified bioactive compounds PPDH and PPDHMP and their protein targets. All molecular docking was carried out using SwissDock web service. Results were analyzed based on binding free energy (ΔG) and full fitness of compounds to their targets. The ΔG and full fitness of PPDH was calculated as −6.43 and −1,584.6 kcal/mol, respectively. The values of ΔG and full fitness for PPDHMP are given in Table 2. The energy above the threshold values of ΔG and full fitness ≥ −6.42 and −1,583.48 kcal/mol, respectively, for PPDHMP was considered for further analysis. Interestingly, PPDHMP has shown strong full fitness to human proteins: cell division protein kinase7 (−1,561.97 kcal/mol) and mitogen-activated protein kinase 8 (−1,964.97 kcal/mol) with significant ΔG values of −6.81 kcal/mol. Besides human proteins, it has also shown good binding affinity and full fitness with bacterial proteins: choloylglycine hydrolase, camphor 5-monooxygenase, chitinase B, and tyrosine phenol-lyase.

Docking of chitinase B protein with both PPDH and PPDHMP was observed. The PPDHMP has also shown almost the same binding free energy (−6.42 kcal/mol) and full fitness (−1,583.48 kcal/mol) with chitinase B (Table 2). Residues of chitinase B protein, viz Tyr10, Phe51, Asp142, Glu144, Tyr214, and Trp403, were found to be actively involved in binding interaction with both PPDH and PPDHMP (Figures 7A,B). Residue Asp142 of chitinase B made an H-bond interaction with the PPDH (Figure 7A), while residue Tyr98 made an H-bond interaction with PPDHMP (Figure 7B). Docking results of PPDHMP with its other targets are enumerated in Figure 8 and Table 2. While PPDHMP showed good binding energy with all its targets, it showed poor full fitness toward tautomerase (−491.25 kcal/mol), divalent-cation tolerance protein (−562.42 kcal/mol), streptavidin (−475.12 kcal/mol), and subtilisin BPN proteins (−631.26 kcal/mol). However, it showed good binding affinity and full fitness with bacterial proteins: choloylglycine hydrolase, camphor 5-monooxygenase, and tyrosine phenol-lyase.

Docking results indicate that both compounds have shown good binding affinity toward their targets. Strong binding of these compounds with proteins and enzymes may inhibit the enzymatic activities in the pathogen leading to growth inhibition, which can be the reasons for the observed antimicrobial effects. The observed in silico studies, however, needs experimental validation to decipher the exact mechanism involved in growth inhibition. Interestingly, PPDHMP has also shown strong binding affinity to some of human proteins: cell division protein kinase 7 (−1,561.97 kcal/mol) and mitogen-activated protein kinase 8 (−1,964.97 kcal/mol) with significant ΔG values of −6.81 kcal/mol. Molecular docking results completely support the multitargeting nature of PPDHMP. Our in silico docking results provide a strong platform to extend the studies further in this direction.