Pyrrolnitrin Biosynthesis From Rhizospheric Serratia Spp. With Antifungal Activity and Binding Interactions of PrnF With Ligands

Pyrrolnitrin (PRN) from rhizobacteria displays a key role in biocontrol of phytopathogenic fungi in rhizospheric soil. Therefore, different rhizospheric soils were investigated for the prevalence of PRN producer in minimal salt (MS) medium containing tryptophan (0.2 M NaCl; pH 8) using three successive enrichments. Of 12% isolates, only ve bacterial strains had shown PRN secretion, screened with Thin Layer Chromatography (R f 0.8) and antifungal activity (27 mm) against phytopathogen. The phenetic and 16S rRNA sequence revealed the close aliation of isolates (KMB, M-2, M-11, TW3, and TO2) to Stenotrophomonas rhizophila (KY800458), Enterobacter spp. (KY800455), Brevibacillus parabrevis (KY800454), Serratia marcescens (KY800456) and Serratia nemtodiphila (KY800457). Puried compound from isolates was characterised using UV, IR, HPLC, LCMS and GCMS as PRN. However, BLASTn hit of prn gene sequences from both Serratia species showed 99% similarity with NADPH dependent FMN reductase component (prnF). The homology protein model of prnF was developed from translated sequence of S. marcescens TW3 with chromate reductase of Escherichia coli K-12. Docking with FMN and NADPH was performed. The study demonstrated the possible role of prnF NADPH dependent FMN reductases in prnD for supply of reduced avin in rhizobacterial strain of Serratia spp. which may pave a way to understand PRN production. both Serratia strains and compared it with PrnF protein models of reference strain in order to nd out structural similarity. Finally, PRN was extracted from ve selected rhizobacterial isolates and characterized for (a) chemical identity with analytical tools, and (b) bioactivity against reference phytopathogen. of mycelial growth; conidial germination of F. oxysporum was undetected even after 10 days in the present study, indicating cessation of sporulation possibly due to more active phenyl-pyrrole component in methanolic extracted samples (Kilani and Fillinger 2016). These results eventually suggest key role of rhizobacterial isolates in suppressive of phytopathogen by PRN in alkaline and saline soil. To investigate the structural basis of the reaction mechanism of PrnF, a homology model of the PrnF was built using crystal structure of the avin reductase (3SVL) from Escherichia coli K12 that showed maximum identity in BLAST (Fig. 7a). The developed model had shown a dual binding cleft for the NADPH and FMN substrate, like the previous reports (van den Heuvel et al. 2004; Okai et al. 2006; Kim et al. 2008; Tiwari et al. 2012). The result from ProSA as Z-Score, Verify 3D and ERRAT plot displayed − 6.07, 81.48% and 87.4016, respectively (Fig. S2a-b). The calculated Ramachandran’s plot suggested that 91.0%, 7.2%, 0.9% and 0.9% of the residues in derived model were in the most favored, additional allowed, generously allowed and disallowed regions, respectively. Altogether 98% of the residues were placed into the combined favored and allowed groups. Thus, PROCHECK and other tools validated the folding integrity of the PrnF model and indicated that the protein structure derived from the 3SVL template was of higher quality in terms of protein folding. Figure 7b shows the distribution of (cid:0) and φ from the Ramachandran’s plot for the non-glycine, non-proline residues. Molecular docking was performed to predict the molecular interactions between NADPH, FMN and amino acid residues of PrnF protein using Autodock, Autodock Vina™ combined with PyRx ™. The molecular docking study revealed that FMN molecule was bound with a groove of the PrnF protein. A single hydrogen bond interaction of FMN with the amino acid residues was observed (Fig. S3a). The pyrophosphate moiety of the FMN molecule was bonded with Isoleucine25 by carbon-hydrogen bond at 6.38 Å. While ribose and adenine moiety of the FMN forms (i) pi-sigma bond interaction (shown in dark pink colour) with Leucine39 (3.96), (ii) alkyl bond (interpreted in oxygenase (FMN), (FAD), oxygenases. oxygenase Pf-5 PrnD Pf-5 prnD prnD FMN (prnF) role in PRN secretion. prnD specic primers conserved sequences of Serratia spp. for prnD, (ii) amplication of fragments with Serratia produce PRN and, (iii) gene loci its Serratia spp. distinct reaction of PRN biosynthesis NADPH dependent FMN reductase avin Serratia spp. antibiotic functional of a studied protein sequence facilitated by developing three-dimensional (3D) structure of protein using comparative or homology modelling which provided a structure related to one known protein. PrnF is superimposed with reductase part of the Escherichia coli K12 (3SVL). alkyl isoalloxazine deep wide conformation nicotinamide ring protein FMN − 5.7 − 5.0, on component arylamine oxygenase in PRN

Presently, PRN is biosynthesised by rhizobacterial strains of Burkholderia cepacia, B. pseudomallei, Corallococcus exigus, Cystobacter ferrugineus, Enterobacter agglomerans, Myxococcus fulvus, Serratia spp., uorescents and non-uorescent Pseudomonas spp. (Gerth et al. 1982; Roitman et al. 1990; Chernin et al. 1996; Roberts et al. 2007; Costa et al. 2009) and reported for the presence of prn operon (Hamill et al. 1967;Gerth et al. 1982; Roitman et al. 1990; Chernin et al. 1996; El-Banna and Winkelmann 1998; Roberts et al. 2007; Costa et al. 2009;Parry et al. 2011). The gene cluster (prnA through prnD) for PRN was reported in Bulkholderia cepacia, Pseudomonas pyrrocinia and Serratia spp. ) and newly discovered prnF ( avin reductase) in close proximity to prnD constituted part of the gene cluster (Lee and Zhao 2007) for PRN biosynthesis from tryptophan through chlorination, followed by rearrangement, regioselective chlorination, oxidation of amino group in the presence of reduced avin by avin reductase ; van Pee 2001). Among these, prn gene sequences revealed more conservation between Burkholderia, Pseudomonas, Serratia derived sequences and hence, prnD gene constitute speci c detection system and de Souza (2003) developed speci c primers located in the prnD gene to assess phylogenetic relationship in PRN producing rhizobacteria. prnD (aminopyrrolnitrin oxygenases) in the pathway catalyse a 6-electron oxidation of amine group of aminopyrrolnitrin to a nitro group to form PRN in the presence of reduced avin (Nakatsu et al. 1995). The evidences of prnF and prnD to form a two component NADPH dependent monooxygenase was recently delineated from prnD-catalyzed arylamine oxidation in Pseudomonas uoresces Pf-5 (Hohaus et al. 1997;Tiwari et al. 2012).
Besides, PrnF was also demonstrated to stimulate (i) chorinaton by PrnA and PrnC activity in PRN synthesis (Richard 2003), (ii) halogenation of aromatic compounds by RebH and ThdH in rebecamycin, thienodolin (Sanchez et al. 2002;Sebold et al. 2006) and (iii) bromination by BrvH in marine metagenomes (Neubauer et al. 2018) and reported to catalyse the reduction of avin such as avin mononucleotide, avin adenine dinucleotide and ribo avin by NAD(P)H to form reduced avin which is required to activate oxygen by the terminal oxygenase ). At present, prnF was identi ed via sequence analysis only from Pseudomonas uorescence Pf-5 and characterized to supply reduced avin to prnD oxygenase component to function in PRN synthesis . On this premise, we initiated our study with (i) isolation and selection of rhizospheric bacteria from eco-habitats for extracellular secretion of PRN, (ii) identi cation of rhizobacterial strains with phenetic and 16S rRNA gene sequencing, (iii) investigation of prnD gene in PRN secretion from both Serratia spp with prnD speci c primers (Souza and Raaijmakers 2003) (iv) characterization of prnF component encoding avin reductase based on bioinformatic studies, and (v) analysis of structural model of prnF protein from both Serratia strains and compared it with PrnF protein models of reference strain in order to nd out structural similarity. Finally, PRN was extracted from ve selected rhizobacterial isolates and characterized for (a) chemical identity with analytical tools, and (b) bioactivity against reference phytopathogen.

Materials And Methods
Growth media, chemicals, phyto-pathogens Various media viz. King's B (KB), Trypan Blue Tetracycline (TBT) and Mineral salt (MS) medium (pH 8.0) were used for isolation of bacteria and screening for secretion of PRN. Plant pathogen Fusarium oxysporum MTCC 9913 was grown and maintained on Potato Dextrose Agar (PDA) for antimicrobial assay and stored at 4 °C. All chemicals as well as solvents used were of reagent grade and procured from M/s S.D. Fine Chemicals, Mumbai and culture media were purchased from Hi-Media Pvt. Ltd., Mumbai.

Soil analysis and isolation, enrichment of bacteria for halo-metabolite secretion
The rhizospheric soil was taken from radish (RJ) grown in nearby area (21. The IR spectra of each extracted sample was recorded for the presence of functional group using Fourier transform infrared spectroscopy (FTIR) (88522 Spectrum Two TM, Perkin-Elmer, USA) in the range of 4000-350 cm − 1 spectral region. Each sample was analysed using % transmission with devoid of sample as control.
High performance liquid chromatography (HPLC) (Younglin (S.K) Gradient System; Software: Autochro-3000) of each sample was conducted at Shree Analytics, Jalgaon (El-Banna and Winkelmann 1998). The HPLC column used was a C18 (4.6 mm x 250 mm (YMC; particle size packing 5 µm). The sample was detected with isocratic system using acetonitrile: water (20 to 100%) as solvent phase. Flow rate was 0.7 mlmin − 1 with ambient temperature and absorbance was checked at 220 nm (UV Detector-730D). Molecular mass spectrum was determined by LC coupled with mass spectrophotometer (LC-MS) (Waters Micromass Q-Tof Micro, USA) at Punjab University, Chandigarh. MS analysis was conducted to screen LC eluent for PRN.
Further, characterization of extracted sample from two bacterial isolates (TW-3 and TO-2) were also conducted by gas chromatography (GC) at NCML, Antagonistic activity of halometabolites towards phytopathogen Each extracted sample was examined for in vitro antagonistic activity towards test phytopathogen Fusarium oxysporum MTCC 9913 using agar-well diffusion assay (Ahmad et al. 2008). About 20 µl of each extracted sample was applied against 10 4 fungal spores ml − 1 into 20 ml of PDA and incubated at 28 °C for 240 h. The growth of phytopathogen in the absence of active compound was monitored and used as a control for the antagonist assay. The zone of inhibition around wells (mm) was estimated after 8 days of incubation. Each experiment was repeated twice, and data was analysed for standard deviation.

Results
Isolation and screening of halometabolite producing bacteria from rhizospheric soil The rhizospheric soil samples namely, RJ and RD were analysed for physico-chemical characteristics. The analysis of both soil (RJ and RD) are revealed in Table 1. The soil type of both RJ and RD samples appeared to be Loam and Sandy loam with high alkalinity and salinity, respectively. The saline soil likely harbour halometabolites (e.g. PRN) secreting bacteria in rhizosphere ( van Pee 2001). Besides, the abundance of tryptophan in the root exudates of radish was reported earlier (Kravchenko et al. 2004) ( Table 2), possibly a precursor for PRN biosynthesis in rhizospheric region. Hence, both salinity and tryptophan were incorporated in the selection medium for isolation of halometabolite secreting rhizobacteria in the present study. The evidences about prevalence of PRN secreting rhizobacteria from rhizospheric soil of grasslands and arable soil was reported earlier with PCR based approach (Garbeva et al. 2004). Total 111 bacterial strains (40 and 71 bacterial morphotypes) were isolated from rhizospheric soil (radish and rice). The preliminary analysis of methanolic extracts of cell free broth from all isolates for PRN was detected with (i) UV-Vis spectrum (Fig. 1a) and (ii) Ehrlich's test using Ehrlich's reagent. Among these, only 40 isolates showed characteristic Ehrlich's reagent spectra (Kessler et al. 1990) of pyrrole ring at 520-570 nm (Fig. 1b). Analogous to UV-Vis spectra of 40 samples, TLC plate developed violet coloured spot after Ehrlich's reagent reaction with R f value of 0.8 (Fig. 2a)   . Among these, only 05 cell free methanol extract of each had shown comparatively signi cant percent area of peak obtained at 27.5 min from bacterial strains namely, TW-3, TO-2, M-11, M-2 and KMB and hence, selected further for extracellular production of PRN. The methanolic extracts of two bacterial strains, TW-3 and TO-2 were characterised by mass spectrophotometry. The LCMS analysis showed a main peak at m/z = 256.232 and 256.121 for TW-3 and TO-2 samples, respectively ( Fig. 3a-b), thus, indicated the presence of PRN. The GC analysis of both samples from both isolates TW-3 and TO-2 also showed highest peak at RT 10.6 min (Fig. 4)  Antagonistic activity at PRN extracts against F. oxysporum MTCC 9913 The samples (40) (Table S2b). The results also showed 99% probability and excellent con dence level with S. marcescens. Table S2a summarizes  The bacterial isolates were investigated for the presence of characteristic prnD coding gene by PCR based ampli cation using primer that have been derived from conserved region of prnD gene identi cation by de Souza and Raaijmakers (2003). The electrophoretic gel image of PCR ampli ed product (Fig. 6c) showed distinct bands in Lane 1 and 2 whereas, lane 3 and 5 showed multiple bands and Lane 4 exposed weak single band. Hence, the gel extraction of PCR products was carried out for the band present in the range of 700-800 bp and taken for DNA sequencing (Fig. S1b). However, these gained gene sequences have shown 99% similarity with putative NADPH dependent FMN reductases present in complete genome of S. marcescens after BLASTn which didn't match with targeted prnD gene. Hence, translated nucleotide query sequence of both isolates was checked using BLASTx against protein data bank and showed Initially, PSI BLAST of protein sequence from isolates TW-3 and TO-2 was performed to get position speci c scoring matrix for secondary structure of related protein sequences. The BLAST hit of protein sequences from isolates TW-3 and TO-2 showed 100 and 99% identities with class of NAD(P)H-dependent oxidoreductase protein of Serratia spp. Therefore, gene sequence analysis unambiguously con rmed the capacity of the strains TO-2 and TW-3 to secrete PRN and indicate that the prn gene may be contributing to PRN production.
Comparative PrnF structure modelling using MODELLER and molecular dynamic simulation and docking To investigate the structural basis of the reaction mechanism of PrnF, a homology model of the PrnF was built using crystal structure of the avin reductase (3SVL) from Escherichia coli K12 that showed maximum identity in BLAST (Fig. 7a) The result from ProSA as Z-Score, Verify 3D and ERRAT plot displayed − 6.07, 81.48% and 87.4016, respectively ( Fig. S2a-b). The calculated Ramachandran's plot suggested that 91.0%, 7.2%, 0.9% and 0.9% of the residues in derived model were in the most favored, additional allowed, generously allowed and disallowed regions, respectively. Altogether 98% of the residues were placed into the combined favored and allowed groups. Thus, PROCHECK and other tools validated the folding integrity of the PrnF model and indicated that the protein structure derived from the 3SVL template was of higher quality in terms of protein folding. Figure 7b shows the distribution of and φ from the Ramachandran's plot for the non-glycine, non-proline residues.
Molecular docking was performed to predict the molecular interactions between NADPH, FMN and amino acid residues of PrnF protein using Autodock, Autodock Vina™ combined with PyRx ™. The molecular docking study revealed that FMN molecule was bound with a groove of the PrnF protein. A single hydrogen bond interaction of FMN with the amino acid residues was observed (Fig. S3a) (Fig. S4).

Discussion
The secretion of PRN in soil by few rhizospheric bacteria has contributed potential role in the (i) suppression of soil to plant pathogens and, (ii) organically driven agricultural practices (Costa et al. 2009). Rhizospheric RJ soil appeared strong brown coloured, with intense dried cracks, while RD soil seems to be dark red to black in colour, possibly due to high iron content. The alkalinity and salinity probably served as key determinants for the occurrence of rhizobacteria in soil to produce PRN (Garbeva et al. 2004) and hence, rhizobacteria application to crop has emerged out as preferred eco-friendly strategy to control phytopathogens compared to the synthetic chemical fungicides.
The rhizobacteria belonging to the genus Serratia was found in root exudates of plants that had a nity towards tryptophan. The root exo-metabolite contains tryptophan in detectable amount in radish root exudates, but, the in vivo concentration of PRN could be produced from available tryptophan is within the range required for fungal inhibition. Hence, secretion of PRN is always in limited quantity than other bacterial metabolites. Interestingly, the previous report had shown that bacterial cell retains PRN inside the cell more than exterior (Roitman et al. 1990) and hence, in vitro antibiosis was examined with supernatant to select e cient rhizobacteria against a test phytopathogen F. oxysporum. Antagonistic activity towards targeted phytopathogen was demonstrated in 23. necessitate to optimize PCR or RT PCR with new speci c primers for the occurrence of prnD coding genes in rhizobacteria KMB, M-2 and M-11 strains. These results agree with earlier report (Costa et al. 2009) where two Serratia strains 5.1R and 5.3R were not ampli ed with prnD. In the present study, both isolates TO-2 and TW-3 have demonstrated PRN production with chemical tools and, shown distinct band after PCR ampli cation but, the gene sequences (KY867430 and KY867431) (Fig. S1B) were not matching with prnD gene after BLASTn analysis. Although, rhizobacteria TO-2 and TW-3 have shown to secrete PRN by analytical tools but undetected for prnD may be due to polymorphic nature of prnD sequences (Costa et al. 2009). However, these prn gene sequence have shown 99% similarity with putative NADPH dependent FMN reductases present in complete genome of S. marcescens after BLASTn. Additionally, BLASTx revealed that the gene sequences belong to FMN reductase superfamily and thus, suggest the presence of prn gene which possibly codes for the supply of reduced avin during enzymatic reactions.
The earlier reports suggested the role of FAD dependent prnF to supply the reduced avin for functioning of prnD as product in PRN secretion Tiwari et al. 2012). However, functional or structural differences between the FMN-dependent and FAD-dependent enzyme systems were undetected (Ellis 2010). Hence, the query sequences (KY867430 and KY867431) in the present study were considered as prnF gene. On the contrary, earlier study had shown that prnF is non-speci c and not directly involved in PRN biosynthesis (van Pee 2012) but, it is required for prnD to function and enhance PrnDs activity  The functional characterization of a studied protein sequence was facilitated by developing three-dimensional (3D) structure of protein using comparative or homology modelling which provided a structure related to one known protein. PrnF is superimposed with reductase part of the Escherichia coli K12 (3SVL). The objective to perform molecular docking is to (i) gain optimized conformation of PrnF, FMN and NADPH i.e. ligand-receptor complex with less binding energy and (ii) predict binding parameters of ligand-protein complex. The docking simulation with modelled PrnF has shown that several amino acids are in close contacts with the ligands i.e. FMN and NADPH (Fig. 7c). Overall, FMN is bounded with protein through hydrogen bond along with van der Waals, carbonhydrogen, Pi-Sigma and alkyl bonds (Fig. S3b). The isoalloxazine ring of FMN probably xed in the deep groove of the protein whereas NADPH restricted at wide groove with compact bonded conformation with nicotinamide ring where, binding capacity of protein-NADPH and protein FMN was − 5.7 and − 5.0, respectively.
Theoretical indications of the PrnF reductase in the two-component aryl amine oxygenase system allows to continue more detailed investigation of this

AVAILABILITY OF DATA AND MATERIALS
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.        Supplementary Files This is a list of supplementary les associated with this preprint. Click to download.