ORIGINAL RESEARCH

published: 22 October 2018

doi: 10.3389/fphar.2018.01038

In silico Prediction, Characterization,

Molecular Docking, and Dynamic
Studies on Fungal SDRs as Novel
Targets for Searching Potential
Fungicides Against Fusarium Wilt in
Tomato
Mohd Aamir 1 , Vinay Kumar Singh 2 , Manish Kumar Dubey 1 , Mukesh Meena 1,5 ,
Sarvesh Pratap Kashyap 3 , Sudheer Kumar Katari 4 , Ram Sanmukh Upadhyay 1 ,
Amineni Umamaheswari 4 and Surendra Singh 1*
1
Laboratory of Mycopathology and Microbial Technology, Centre of Advanced Study in Botany, Institute of Science, Banaras
Hindu University, Varanasi, India, 2 Centre for Bioinformatics, School of Biotechnology, Institute of Science, Banaras Hindu
University, Varanasi, India, 3 Division of Crop Improvement and Biotechnology, Indian Institute of Vegetable Research, Indian
Council of Agricultural Research (ICAR), Varanasi, India, 4 Bioinformatics Centre, Department of Bioinformatics,
Sri Venkateswara Institute of Medical Sciences University, Tirupati, India, 5 Department of Botany, University College of
Science, Mohanlal Sukhadia University, Udaipur, India
Edited by:
Vivek K. Bajpai,
Dongguk University, South Korea
Vascular wilt of tomato caused by Fusarium oxysporum f.sp. lycopersici (FOL)
Reviewed by:
is one of the most devastating diseases, that delimits the tomato production
Ashutosh Bahuguna, worldwide. Fungal short-chain dehydrogenases/reductases (SDRs) are NADP(H)
Daegu University, South Korea
dependent oxidoreductases, having shared motifs and common functional mechanism,
Dinesh Yadav,
Deen Dayal Upadhyay Gorakhpur have been demonstrated as biochemical targets for commercial fungicides. The
University, India 1,3,6,8 tetra hydroxynaphthalene reductase (T4HNR) protein, a member of SDRs
*Correspondence: family, catalyzes the naphthol reduction reaction in fungal melanin biosynthesis.
Surendra Singh
surendrasingh.bhu@gmail.com
We retrieved an orthologous member of T4HNR, (complexed with NADP(H) and
pyroquilon from Magnaporthe grisea) in the FOL (namely; FOXG_04696) based
Specialty section: on homology search, percent identity and sequence similarity (93% query cover;
This article was submitted to
Predictive Toxicology,
49% identity). The hypothetical protein FOXG_04696 (T4HNR like) had conserved
a section of the journal T-G-X-X-X-G-X-G motif (cofactor binding site) at N-terminus, similar to M. grisea
Frontiers in Pharmacology
(1JA9) and Y-X-X-X-K motif, as a part of the active site, bearing homologies with two
Received: 09 December 2017
fungal keto reductases T4HNR (M. grisea) and 17-β-hydroxysteroid dehydrogenase
Accepted: 27 August 2018
Published: 22 October 2018 from Curvularia lunata (teleomorph: Cochliobolus lunatus PDB ID: 3IS3). The catalytic
Citation: tetrad of T4HNR was replaced with ASN115 , SER141 , TYR154 , and LYS158 in the
Aamir M, Singh VK, Dubey MK, FOXG_04696. The structural alignment and superposition of FOXG_04696 over
Meena M, Kashyap SP, Katari SK,
Upadhyay RS, Umamaheswari A and
the template proteins (3IS3 and 1JA9) revealed minimum RMSD deviations of
Singh S (2018) In silico Prediction, the C alpha atomic coordinates, and therefore, had structural conservation. The
Characterization, Molecular Docking,
best protein model (FOXG_04696) was docked with 37 fungicides, to evaluate
and Dynamic Studies on Fungal SDRs
as Novel Targets for Searching their binding affinities. The Glide XP and YASARA docked complexes showed
Potential Fungicides Against Fusarium discrepancies in results, for scoring and ranking the binding affinities of fungicides.
Wilt in Tomato.
Front. Pharmacol. 9:1038.
The docked complexes were further refined and rescored from their docked poses
doi: 10.3389/fphar.2018.01038 through 50 ns long MD simulations, and binding free energies (1Gbind ) calculations,

Frontiers in Pharmacology | www.frontiersin.org 1 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

using MM/GBSA analysis, revealed Oxathiapiprolin and Famoxadone as better

fungicides among the selected one. However, Famoxadone had better interaction of
the docked residues, with best protein ligand contacts, minimum RMSD (high accuracy
of the docking pose) and RMSF (structural integrity and conformational flexibility of
docking) at the specified docking site. The Famoxadone was found to be acceptable
based on in silico toxicity and in vitro growth inhibition assessment. We conclude that
the FOXG_04696, could be employed as a novel candidate protein, for structure-based
design, and screening of target fungicides against the FOL pathogen.
Keywords: THN reductase, fungicide, melanin, protein–fungicide interaction, homology modeling, MD
simulations, MM/GBSA analysis

INTRODUCTION Singha et al., 2011; Anand et al., 2013; Khan et al., 2014),
flutolanil (Moncut WP 30%), tolclofos-methyl/thiram (Rhizolex
Tomato (Lycopersicon esculentum Mill.) is one of the most 50% WP) and carboxin-thiram (Vitavax 200 WP) (Mohamed
widespread vegetable crops grown across the globe. However, and Amer, 2014), mancozeb + carbendazim (0.125 + 0.05%)
the growth and economic productivity of tomato crop are (Barhate et al., 2015), mancozeb + copper sulfate + copper
well constrained by various biotic and abiotic stress conditions oxychloride (Ramaiah and Garampalli, 2015), metiram (55%)
(Bergougnoux, 2014; Gupta and Rashotte, 2014). Vascular wilt and pyraclostrobin (5%) (Yeole et al., 2016), thiophanate
disease caused by Fusarium oxysporum f.sp. lycopersici (FOL) methyl (La Torre et al., 2016), propiconazole, thiabendazole,
(Sacc.) W. C. Snyder and H. N. Hans (FOL) is one of the most benomyl, fuberidazole, thiophanate, myclobutanil triadimefon,
destructive diseases (Amini and Sidovich, 2010; Prihatna et al., difenoconazole, tebuconazole, epoxiconazole, methoxy-acrylates,
2018), that affects the growth and economic production of tomato ethyl phosphonates (de la Isla and Macías-Sánchez, 2017), Nativo
(Yeole et al., 2016; Prihatna et al., 2018). The wilt pathogen FOL 75% WG, Cordate 4WP, fluopyram 20% + tebuconazole 20%,
is the most common soil-borne Ascomycetous fungus that infects and tebuconazole 50% + trifloxystrobin 50% (Patón et al., 2017).
through roots and develops symptoms leading to vascular wilt Short-chain dehydrogenases/reductases (SDRs) are
in tomato (Park et al., 2013; Rongai et al., 2017). It invades the NADP(H)-dependent oxidoreductases characterized by
xylem vessels resulting in wilting and death of the plant (Swarupa conserved catalytic tetrad (N-S-Y-K) and cofactor binding
et al., 2014). The high-frequency incidence (25–55%) of Fusarium site (TGxxxGxG) (Jörnvall et al., 1995; Filling et al., 2002)
wilt disease in tomato has been reported from various regions of with having common α/β-folding pattern, and characterized
India (Asha et al., 2011; Pandey and Gupta, 2014; Nirmaladevi by presence of a central β-sheet typical to Rossmann-fold
et al., 2016). The infection and disease development of the fungus with helices on either side (Kavanagh et al., 2008). The fungal
leads to devastating agricultural losses, which may cover up to 1,3,6,8-tetrahydroxynaphthalene reductase belongs to SDR
80% under the favorable weather conditions. family mediates the naphthol reduction reactions in melanin
The vascular wilt disease of tomato is characterized by biosynthetic pathway (Liao et al., 2001). The protein Blast
vascular browning, that involves the deposition of melanin-like results at NCBI revealed that M. grisea T4HNR (SDR) showed
compounds on the walls of xylem vessel and other neighboring high sequence similarity with other fungal keto reductases,
parenchymatous cells (Mace et al., 2012). The control of vascular involved in the biosynthesis of fungal melanin and mycotoxins,
wilt disease is difficult and mainly achieved through the use of that includes versicolorin reductase from Magnaporthe oryzae
chemical fungicides (Minton, 1986; DeVay et al., 1988; Swarupa (99%), Verticillium alfalfae (77%), Verticillium dahliae (76%),
et al., 2014). The most commonly used chemical fungicides that Colletotrichum graminicola (79%), versicolorin reductase (VerA)
have been used up to till date against the Fusarium sp. either alone from Emericella nidulans (52%), and 17β-hydroxysteroid
or in combination with other integrated approaches includes dehydrogenase (17β-HSDcl) of Cochliobolus lunatus (52%). The
iprodione (Amany and Ellil, 2005) (Rovral) (dithiocarboxamide) crucial role of the fungal SDR gene in M. oryzae is required for
benomyl (Benelate) carbendazim, prochloraz, fludioxonil, infection related development and pathogenicity (Kwon et al.,
bromuconazole, azoxystrobin (Amini and Sidovich, 2010; 2010). The function of a novel fungal SDR gene (adh1) encoding
for alcohol dehydrogenase has been reported to play a crucial
Abbreviations: DHN, 1,8-dihydroxynaphthalene; DOPE, discrete optimized role in virulence of Fusarium wilt pathogen in tomato (Corrales
protein energy; GLIDE, grid-based ligand docking with energetics; MM/GBSA, et al., 2011).
molecular mechanics generalized Born surface area; PMDB, protein modeling
database; ProSA, protein structural analysis; ProTSAV, protein structure analysis Fungal melanins are high molecular weight dark brown to
and validation; RAMPAGE, Ramachandran plot analysis; RMSD, root mean black colored pigments synthesized via the pentaketide pathways
square deviation; RMSF, root mean square fluctuation; SDR, short-chain in the cell wall of fungal groups belonging to Ascomycotina
dehydrogenase/reductases; T3HNR, 1,3,8-trihydroxynaphthalene reductase;
and Deuteromycotina (Bell and Wheeler, 1986). The DHN
T4HNR, 1,3,6,8-tetrahydroxynaphthalene reductase; UniProtKB, Universal
Protein Resource Knowledgebase; VADAR, volume area dihedral angle reporter; melanin biosynthetic route is the most common among
OPLS, (optimized potentials for liquid simulations). fungi where melanins are synthesized through the acetate

Frontiers in Pharmacology | www.frontiersin.org 2 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

via the polyketide synthase pathway (Chiewchanvit et al., and scytalone dehydratase (VDAG_03393) (Xiong et al.,
2017). DHN melanin pathway has been investigated in many 2014). Recently, the gene clusters and enzymes, involved in
filamentous plant pathogenic fungi including Cochliobolus melanin and other pigment biosynthesis, were explored in
heterostrophus (Eliahu et al., 2007), Alternaria spp. (Kheder Ascomycota including Aspergillus spp. based on transcriptomic
et al., 2012), Colletotrichum spp. (Ludwig et al., 2014), genera and gene expression studies. The studies revealed that the
Gaeumannomyces (Frederick et al., 1999), Phyllosticta musarum core polyketide synthase (PKS) gene clusters have crucial
(Kubo and Furusawa, 1991), and V. dahliae (Wheeler et al., role in biosynthesis of DHN type of pigment (Palonen
1978). During the biosynthesis of fungal melanin through et al., 2017). The phylogenetic analysis of the extended PKS
pentaketide pathway, tetrahydroxynaphthalene reductase revealed striking similarities with group of known pigments
(T4HNR) catalyzes the NADP(H)-dependent reduction of of Fusarium spp., which predicts the similar function for
1,3,6,8-tetrahydroxynaphthalene (THN) into (+)-scytalone and this PKS (Palonen et al., 2017). Some chemical fungicides
1,3,8-trihydroxynaphthalene into (−)-vermelone (Figure 1). that inhibit the biosynthesis of melanin have been used in
The DHN pathway-based classification depends on their controlling plant pathogenic fungi (Kurahashi, 2001). In the
preference for the Naphthoquinon precursors or on the effect last few years, many melanin biosynthesis inhibitors have been
of inhibitors such as phthalide or tricyclazole, which binds with used against rice blast pathogen such as triazoloquinoline,
hydroxynaphthalene reductases having classical short-chain pyroquilon, tricyclazole, and coumarin (Yamaguchi and Kubo,
dehydrogenase/reductase (SDR) with Rossmann-fold domains 1992; Kimura and Tsuge, 1993). The formation of melanin by
(Palonen et al., 2017). It has been reported that the polyketide the members of Fusarium genus has been recently reported
pathway in filamentous fungi is an important metabolic process as it was found that F. graminearum accumulates melanins in
that regulates their growth, development, and pathogenicity a process dependent on polyketide synthase PGL1 (Frandsen
(Xiong et al., 2014). Fungal melanin is an important polyketide et al., 2016). Furthermore, F. keratoplasticum, a significant
and genes responsible for the biosynthesis of melanins have causing agent of fusariosis produces melanin or melanin-like
been reported in V. dahliae including hydroxynaphthalene compounds during in vitro cultivation and also inside the
reductase (VDAG_03665), polyketide synthase (VDAG_00190), growing tissues as confirmed through immunofluorescence

FIGURE 1 | General mechanism of DHN melanin biosynthesis pathway in fungi. The tetrahydroxynaphthalene reductase (T4HNR) catalyzes the NADP(H)-dependent
reduction of 1,3,6,8-tetrahydroxynaphthalene (THN) into (+)-scytalone and 1,3,8-trihydroxynaphthalene into (–)-vermelone. 1,8-dihydroxynaphthalene (DHN) is the
immediate precursor of the polymer.

Frontiers in Pharmacology | www.frontiersin.org 3 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

labeling with anti-melanin monoclonal antibody (MAb) better fungicidal action [for example, T4HNR complex with
(Chiewchanvit et al., 2017). The fungus FOL forms brown pyroquilon and NADP(H) used successfully against M. grisea]
colored melanin that is insoluble in water and organic solvents (Singh et al., 2014). The objective of the present study is
but soluble in alkaline medium (1 M KOH) (Amany and to evaluate the efficacy of potential inhibitor (fungicides)
Ellil, 2005). Dicarboxamide produces antimicrobial oxidants that could bind to the crucial residues of the FOXG_04696.
using ROS molecules, thus inhibiting the growth of many Furthermore, the protein–fungicide docking studies with target
pathogenic fungi, including F. oxysporum (Abo Ellil and protein could be useful to evaluate the comparative efficacy of
Sharaf, 2000). The sensitivity of some potent phytopathogenic an individual fungicide over each other against vascular wilt
fungi such as Sclerotium cepivorum, Alternaria alternata, pathogen.
and FOL pathogen against melanin biosynthesis inhibitor
(fungicides having dicarboxamide group) have been well
evaluated (Amany and Ellil, 2005). Furthermore, fungicides that MATERIALS AND METHODS
inhibit the biosynthesis of melanin (tricyclazole, pyroquilon, and
iprodione) could be employed as a useful tool for controlling Database Search, Comparative
plant pathogenic fungi that utilize polyketide metabolites as
intermediates (Motoyama and Yamaguchi, 2003; Singh et al., Phylogeny, and Functional Domain
2014). The 17-β-hydroxysteroid dehydrogenase (SDR) was Analysis
recently used as a molecular target for fungicide tricyclazole The protein sequence available for the crystal structure of the
against Cercospora canescens, causing Leaf spot disease in T4HNR protein complexed with NADP(H) and pyroquilon
mung bean (Vigna radita) (Singh et al., 2014). In a recent fungicide, and solved through X-ray diffraction in Magnaporthe
study, the inhibitors for F. oxysporum copper nitrite reductase grisea was selected for searching all the sequential homolog
(NirK), involved in the fungal denitrification process were and orthologs using NCBI Blast server1 (Altschul et al., 1997)
searched using hierarchical in silico screening approach that keeping the default values, and against the non-redundant
consists of pharmacophore modeling and molecular docking protein sequences, with searching the organism as FOL 4287
(Matsuoka et al., 2017). The ranges of the molecular target (taxid: 426428). The sequences were also retrieved, checked,
for currently used fungicides are narrow, and therefore, and confirmed from the JGI genome portal for FOL with
the threat of resistance development necessitates the need having transcript ID 13950. The Blast-p annotations were further
for the discovery of novel targets for fungicides (Foster, checked across several databases. The FOXG_04696 homolog and
2018). orthologous sequences to the T4HNR protein were identified
In the last few years, several studies have been done on in silico using Blast-p and collected for multiple sequence alignment
characterization of an unknown hypothetical proteins/essential using ClustalW (Thompson et al., 1994). Multiple sequence
genes from pathogenic microbes, that might have a possible role alignment was done to represent the consensus and conserved
in regulation of metabolic process, or play an indispensable role residues present in the T4HNR protein across the different
in microbial pathogenicity (Ravooru et al., 2014; Silva et al., members using CLC BIO workbench. The alignment results
2015; Marklevitz and Harris, 2016; Kumar et al., 2017; Prava were further checked using the BioEdit tool (Hall, 1999). The
et al., 2018). Recently, a hypothetical protein (FcRav2) with phylogenetic relationship between the different homolog and
ROGDI such as leucine zipper domain, and homologous to yeast orthologs were established using the neighbor-joining (NJ)
Rav2 was reported in F. culmorum. It was demonstrated that and maximum parsimonious method using the MEGA6 suite2
FcRav2 protein may become a suitable target for new antifungal (Tamura et al., 2013) at 1000 replication bootstrap values.
drug development or the plant−mediated resistance response in The similarities and differences in the T4HNR proteins in
filamentous fungi of agricultural interest (Spanu et al., 2018). between different homologous and orthologous fungal partners
In this study, we have predicted and characterized a fungal were visualized based on the comparison of their protein
SDR (the FOXG_04696) as a putative receptor protein, and sequences retrieved through the conservation of genomic
a novel target, for structure-based protein–fungicide complex positions (segments) using circos visualization tool3 (Krzywinski
interactions. The predicted protein was found to be good et al., 2009) at 50% cutoff filter values. The functional domain
enough based on qualitative and quantitative parameters and of the identified protein was searched using ExPASy-PROSITE
was further docked with 37 known commercial fungicides, scan4 (de Castro et al., 2006; Sigrist et al., 2010). The identified
frequently used against different phytopathogens, to find the FOXG_04696 protein sequences were further searched for
best fungicide/agrochemicals (among the selected) that could finding the functional signature sequences against the InterPro
target the FOXG_04696 and therefore, useful for controlling protein signature database using InterProScan 5.05 (Jones et al.,
vascular wilt fungi. The environmental toxicity assessment could 2014).
be used to decide the dosage formulations that could be used
safely without having any loss to the non-target organism. 1
http://blast.ncbi.nlm.nih.gov/Blast.cgi
With this view, the selected fungicides were further evaluated 2
http://www.megasoftware.net/
based on in silico toxicity assessment tools. It was found the 3
http://circos.ca/
fungicide that binds with crucial residues forming active site 4
http://prosite.expasy.org/scanprosite/
of the receptor protein (disrupt the protein function) have a 5
https://www.ebi.ac.uk/interpro/interproscan.html

Frontiers in Pharmacology | www.frontiersin.org 4 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

Gene Prediction and Chromosomal interactions were analyzed at their high and highest confidence
Mapping level.
Sequence of the protein tetrahydroxynaphthalene reductase
(T4HNR) complexed with NADP(H) and pyroquilon (1JA9) was Structural Modeling
searched to find its sequential orthologs in the FOL pathogen, The homology modeling of the protein FOXG_04696 (T4HNR
using Blast-p against the non-redundant database. Furthermore, like) was performed using Modeller v9.19. The protein sequence
the two protein sequences (1JA9 and the FOXG_04696) were was queried against the PDB database14 (Berman et al.,
aligned using BL2 seq (Blast-p). The FOXG_04696 protein 2000) with having sequence similarities >90% using Blast-
sequence was also aligned with the protein sequence of 17-β- p to identify the closely related structural homologs for the
hydroxysteroid dehydrogenase [other closely related structural FOXG_04696. The first hit obtained on Blast-p annotation was
homolog (3IS3)]. Furthermore, the FOXG_04696 protein was found to 17-β-hydroxysteroid dehydrogenase (SDR enzyme)
searched using the tBlastn against the Refseq (reference protein) from Cochliobolus lunatus was taken as a template (PDB ID: 3IS3;
genome database, searching for F. oxysporum f.sp. lycopersici. 46% identity, 96% of query coverage; E-value of 2e–75). The PDB
The first hit obtained was further scanned with the gene file of the template (3IS3) was retrieved from the Protein Data
prediction tool Fgenesh6 . Furthermore, the chromosomal map Bank (PDB). The alignment file was generated using CLUSTALX.
was generated to identify and locate the position of the gene, The target sequence file, alignment file, and template’s PDB
that encodes the hypothetical protein FOXG_04696 using the file, PDB file (3IS3) was initialized in the Modeller script file
Ensembl-BLAST tool. (script.py). The script file (script.py) was executed using Modeller
command prompt. Twenty-five models were generated for the
CATH Analysis FOXG_04696, each with having a DOPE score. Furthermore,
The functional annotation of the predicted FOXG_04696 the protein model with least DOPE score was selected for final
protein was done using CATH server. The FOXG_04696 validation.
protein sequence was submitted to CATH database7 (Sillitoe
et al., 2015) for structural classification, based on domains Model Validation
organization, and folding patterns that belong to homologous The stereochemical stability of the predicted models were further
protein superfamilies. The FunFHMMer8 (Das et al., 2016) was verified using various protein quality based parameters such
used for functional classification of the identified CATH super as percentage residues lying in favored and allowed regions,
families. The ReviGO webserver9 (Supek et al., 2011) was used the number of glycine and proline residues and orientation of
for plotting the functional annotation in terms of molecular dihedral angles including phi (ϕ) and psi (ψ) and backbone
function and biological processes involved using scattered plot conformation using PROCHECK module of the PDBSum
diagram. The CELLO2GO webserver10 (Yu et al., 2014) was used server15 (Laskowski et al., 2005), and also confirmed using
for finding the probable subcellular localization of the predicted the RAMPAGE server16 (Lovell et al., 2003). The qualitative
protein. The possible functional role of the FOXG_04696 was assessment methods were based on ProSA analysis (probable
predicted in terms of gene ontology enrichment analysis. residues lying at a specific distance and interactions observed
between the model and the solvent i.e., solvation17 (Wiederstein
Identification of Functional Sites and Sipp, 2007). The VERIFY3D (Eisenberg et al., 1997) server
The functional sites of the identified protein were searched using was used to check the compatibility of atomic models (3D) with
CD search on CDD webserver11 (Marchler-Bauer et al., 2015, its own primary amino acid sequences (1D). The quality was
2017) at three interfaces including protein active site, substrate verified using the ERRAT score values18 (statistics of non-bonded
binding site, and chemical binding (NADP binding site). The atomic interactions and distribution of atoms) (Colovos and
meta-pocket server12 (Huang, 2009) was used for the prediction Yeates, 1993). The overall quality assessment of predicted model
of three prominent binding sites in the FOXG_04696 protein. was done through ProTSAV score values19 (Singh et al., 2016).
The quantitative evaluation of the model was done through
Protein–Protein Interaction Network the VADAR20 (Willard et al., 2003). The modeled FOXG_04696
The FOXG_04696 protein was submitted to the STRING (Search protein was superimposed over the template T4HNR (SDR)
Tool for the Retrieval of Interacting Genes/Proteins database protein of M. grisea (1JA9) to compare their structural alignment
version 10.0)13 (Szklarczyk et al., 2007) server for the functional and similarities using the Automated Structural Alignment
interaction associative network between the partners, and the Server (AuStrAlis)21 . The final model was submitted to an online

6 14
http://www.softberry.com/ http://www.rcsb.org/pdb/
7 15
http://www.cathdb.info/ http://www.ebi.ac.uk/pdbsum/
8 16
http://www.cathdb.info/search/by_funfhmmer http://mordred.bioc.cam.ac.uk/~rapper/rampage.php
9 17
http://revigo.irb.hr/ https://prosa.services.came.sbg.ac.at/prosa.php
10 18
http://cello.life.nctu.edu.tw/cello2go/ http://services.mbi.ucla.edu/ERRAT/
11 19
https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi http://www.scfbio-iitd.res.in/software/proteomics/protsav.jsp
12 20
https://projects.biotec.tu-dresden.de/metapocket/ http://vadar.wishartlab.com/
13 21
http://string-db.org/ http://eds.bmc.uu.se/eds/australis.php

Frontiers in Pharmacology | www.frontiersin.org 5 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

repository protein modeling databases (PMDB)22 (Castrignano Molecular Mechanics and Binding
et al., 2006). Energy Assessment
The protein–fungicide docked complexes were further analyzed
Preparation of Protein and Ligands for Molecular Mechanics/Generalized Born Surface Area
The ligands were retrieved from the PubChem database. (MM/GBSA) analysis to predict the free binding energies of
The FOXG_04696 protein was selected as a target receptor the protein–fungicide docked complexes. The binding energy
protein and was imported to the Maestro v11. The structure calculated through MM/GBSA was more accurate than the XPG
was prepared using protein preparation wizard of the Score (Lyne et al., 2006). The binding free energy 1Gbind was
Schrödinger. Optimization of protein was done at neutral calculated by the following equations (Liang et al., 2017; Zhang
pH and then the structure was minimized by applying optimized et al., 2017).
potentials for liquid simulations (OPLS-3) force field for
all atoms (Umamaheswari et al., 2010). A receptor grid of 1Gbind = 1Gcomplex − (1Greceptor + 1Gligand )
10Å × 10Å × 10Å was generated on defined binding site
residues of the FOXG_04696 using Glide v7.1 (Grid-based 1G = 1Egas + 1Gsol − T1Sgas
Ligand Docking with Energetics, Schrödinger, LLC, New York, 1Egas = 1Eint + 1EELE + 1EVDW
NY, United States, 2017) (Friesner et al., 2006). The ligand
was prepared through adjusting the chemical correctness 1Gsol = 1GGB + 1GSurf
(protonation), stereochemical and ionization variation using
Epik and LigPrep modules. The energy minimization was done These energy contributions are computed from the atomic
at neutral pH 7.0 ± 2.0. coordinates of the protein, ligand and complex using the (gas
phase) molecular mechanics energy function (or force field).
The solvation free energy term Gsolv contains both polar and
Protein–Fungicide Docking
non-polar contributions. The binding free energy (1Gbind )
The Glide XP ligand docking protocol was employed to predict
could be dissociated into various energy terms. Since the
the scoring and binding interactions between the FOXG_04696
same trajectory was selected for extraction of receptor protein,
and the ligand Famoxadone. The prepared ligand was docked
ligand, and protein–ligand complex, we neglected the internal
into the binding site of the FOXG_04696. The van der Waals
energy change (1Eint ). Therefore, the gas–phase interaction
radii of non-polar regions of the T4HNR were limited to
energy (1Egas ) between the receptor and the ligand was the
1.0Å with the partial atomic charge of 0.25 in the receptor
sum of electrostatic (1EELE ) and van der Waals (1EVDW )
grid generation. Rigid receptor docking was utilized to dock
interaction energies. The solvation free energy (1Gsol ) could be
each ligand into every refined low-energy conformation of the
distributed into non-polar and polar energy terms, and the polar
T4HNR produced from the earlier phases (high-throughput
solvation energy (1GGB ) is calculated by using the VSGB2.1 GB
virtual screening and standard precision methods). XP docked
model, and was default parameter for Prime calculations using
complexes were evaluated using Xtra precision Glide score (XPG
the OPLS2.1/3/3e force field. The Post-docking MM/GBSA is
Score). The XPG score optimized the ligand binding energy
implemented in Schrödinger software using the program Prime,
on the behalf of the force field parameters, and penalties that
with options to include receptor and ligand flexibility; the entropy
had significant influences over the receptor-ligand binding. The
term is neglected by default. Simulations were performed using
following equation denotes the formulae for XPG calculations.
GBSA continuum model. The Gaussian surface area model
Score = a∗ vdW + b∗ Coul + Lipo + Hbond + Metal instead of vdW was employed for denoting the solvent accessible
surface area.
+ BuryP + RotB + Site

where vdW, Coul, Lipo, H bond, metal, BuryP, Rot B, and Molecular Dynamics (MD) Simulations
Site denote van der Waals energy, Coulomb energy, lipophilic The receptor–ligand interactions for fungicides having
contacts, hydrogen-bonding, metal-binding, penalty for buried minimum binding energy (stronger binding) were further
polar groups, penalty for freezing the rotatable bonds, and polar evaluated using molecular dynamics simulations analysis. MD
interactions with the residues in the active site, respectively; simulations studies were performed up to 50 ns through
a = 0.065 and b = 0.130 are coefficient constants of van der Waals Desmond v 4.2 to analyze the conformational stability
energy and Coulomb energy, respectively. of the FOXG_04696–Famoxadone, and the complexes in
The molecular docking of the FOXG_04696 with fungicides the solvated model system, embedded with ordered water
was also performed through YASARA (Yet Another Scientific molecules (ordered water molecules may involve in protein
Artificial Reality Application) (Krieger and Vriend, 2014; Chen binding sites and influence protein ligand binding by
et al., 2015). The YASARA docked protein–fungicide complexes bridging protein–ligand interactions and can make large
were analyzed for the comparative binding energies and contributions to the binding affinity). The Desmond supports
dissociation constant (K d ) of the docked molecular complexes algorithms typically used to perform fast and accurate MD
(Yadav et al., 2017). simulations. Long-range electrostatic energy and forces were
calculated using particle-mesh-based Ewald techniques. The
22
http://www.caspur.it/PMDB FOXG_04696–ligand docked complexes were solvated, using

Frontiers in Pharmacology | www.frontiersin.org 6 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

orthorhombic simple point charge (SPC) water model. The statistical data were expressed in mean ± SEM values of three
solvated system was neutralized with counter ions and independent replications data ± SD, and the average data of
physiological salt concentration was limited to 0.15 M. The one experiment was interpreted through one-way analysis of
receptor–ligand complex system was assigned with optimized variance (ANOVA), while the comparison of mean separations
potentials for liquid simulations-AA (OPLS-AA) 2005 force was performed with Duncan’s multiple range test (DMRT) with
field (Madhulitha et al., 2017). The system was specified on P ≤ 0.05 of significance level.
periodic boundary conditions, the particle mesh Ewald (PME)
(Maragakis et al., 2008) method was applied for electrostatics. In silico Toxicity Assessment
Lennard-Jones interactions cutoff was set to 10Å and SHAKE The in silico toxicity assessment of the selected fungicide
algorithm (Friesner et al., 2006) was employed for limiting was made with different online tools and software including
movement of all covalent bonds involving hydrogen atoms. FAF-Drugs 4.023 (Lagorce et al., 2017). Furthermore, the
The solvated model system, prior to MD simulationss study, environmental toxicity hazard assessments were also evaluated
was passed through a six-step relaxation protocol for energy through admetSAR24 (Cheng et al., 2012). The drug-likeness of
minimization (Katari et al., 2016). At first, only solvent molecules the selected fungicide was evaluated through Lipinski Rule of Five
were allowed for energy minimization which then followed by using the server given in the web link25 (Lipinski, 2004; Jayaram
minimization of the entire system using the Broyden–Fletcher– et al., 2013).
Goldfarb–Shanno (LBFGS) algorithm (Chiranjeevi et al., 2016).
The minimized system was further analyzed with NVT ensemble
for 12 picoseconds (ps) simulationss at 10 K temperature. The RESULTS
non-hydrogen solute atoms were restrained at 300 K temperature
for 24 ps. Furthermore, the system was simulated for 24 ps in the Database Search, Comparative
NPT ensemble at 300 K temperature without restrains in order to Phylogeny, and Functional Domain
attain an equilibrium state (Chubb et al., 2006). The minimized Analysis
system without any restrains was further subjected to 50 ns
The Blast-p results against the non-redundant (nr) database
NPT simulations production (Cichero et al., 2013; D’Ursi et al.,
with organism Fusarium oxysporum f.sp. lycopersici 4287
2016). Berendsen thermostats and barostat were used to control
(taxid: 426428) revealed the homology of the query protein
the temperatures and pressures during the initial simulations
sequence (1JA9) with the target protein FOXG_04696. The
(Pradeep et al., 2015). For MD simulations, Desmond was
query sequences showed 93% query coverages with 49% identity
utilized as constraints, which are enforced using a variant of the
with the target protein FOXG_04696 (XP_018239507.1). The
SHAKE algorithm, allowed the time step to be increased. These
PDB Blast-p annotation revealed the queried sequence of
approaches can be used in combination with time-scale splitting
the FOXG_04696 had more than one structural homologs
(RESPA-based) integration schemes. The purpose was to find
like 1JA9 (47% identity), 3IS3 (44% identity), and therefore,
the interactions between protein and ligand in protein–ligand
could be used as a template protein for homology modeling
complex during MD simulations.
of our target protein. The Uniprot results identified the
queried protein sequence as an uncharacterized/hypothetical
In vitro Inhibition Test protein of F. oxysporum f.sp. lycopersici (strain 4287/CBS
The selected fungicide (Famoxadone) was used for in vitro 123668/FGSC 9935/NRRL 34936) (A0A0D2XL72). Interestingly,
assessment against the FOL pathogen. The pathogenic culture both T4HNR (1JA9) and 17-β-hydroxysteroid dehydrogenase
was obtained from Laboratory of Mycopathology and Microbial (3IS3) query sequences when searched against the reference
Technology, Department of Botany, Institute of Science, Banaras protein (Ref seq) database, with searching for Fusarium
Hindu University, Varanasi, India and the fungicide Famoxadone (taxid: 5506) the first and significant hit obtained showed
(Sigma-Aldrich, St Louis, MO, United States) was used for an orthologous relationship of the queried protein with
evaluating its in vitro efficacy. Four separate concentrations 50, the hypothetical protein FOXG_04696 [XP_018239507.1; 49%
100, 150, and 200 µL were employed along with the control identity (1JA9): 93% query coverages; E-value: 3e−74 and
solution (having only PDA) and amended in 20-mL PDA 46% identity (3IS3): 94% query cover; E-value: 1e−71],
medium. A 5-mm culture disc was extracted from the freshly which further confirms the existence of similar T4HNR and
inoculated pathogen culture in each of the four plates. The plates 17-β-hydroxysteroid dehydrogenase-like protein (FOXG_04696)
were further incubated at 27 ± 2◦ C under observation and the in the FOL pathogen. The phylogenetic tree was constructed
radial growth of the hyphae was recorded at even (2, 4, 6, and 8) based on the neighbour-joining end (NJ) method revealed the
days interval. polyphyletic origin of the FOXG_04696 protein (Figure 2). The
The percent growth inhibition (PI) was calculated using the evolutionary conservation and functional diversification of the
following formula [(C − T)/C × 100] where I = inhibition fungal SDRs across the related taxonomic group have been shown
percentage; C = radial growth of the pathogen in control, and through maximum parsimonious method based phylogenetic
T = radial growth of the pathogen fungicide treatment (Suneeta tree (Supplementary Figure S1). The PROSITE results revealed
et al., 2016). The percentage inhibitions measured in the form
of radial growth were subjected to statistical analysis. All the 23
http://fafdrugs3.mti.univ-paris-diderot.fr/
experiments were executed in triplicates and repeated twice 24
http://lmmd.ecust.edu.cn/admetsar1/predict/
employing a completely randomized design. The representative 25
http://www.scfbio-iitd.res.in/software/drugdesign/lipinski.jsp

Frontiers in Pharmacology | www.frontiersin.org 7 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 2 | Phylogenetic relationships between the different fungal taxa showing the evolution of short-chain dehydrogense/reductases (T4HNR like) protein. The
tree was constructed based on distance-based neighbour-joining (NJ) method with 1000 bootstrap relications using MEGA6.0. The tree showed the existence of
several clades for fungal short-chain dehydrogenases/reductases (SDRs) between the evolutionarily related taxa. The hypothetical protein (FOXG_04696) lacks
common ancestor and therefore predicts the polyphyletic evolution of SDR in Fusarium oxysporum f.sp. lycopersici. The bootstrap values are mentioned below the
tree.

the presence of common functional domain with signature 8e−77). This confirms that T4HNR (1JA9) is closely related
sequences characteristic to the SDR family (IPR002347) and the with 1JA9 based on percentage identity and query cover values.
NADP binding domain superfamily (InterproID: IPR036291) The Fgenesh results located the position of the FOXG_04696
(Supplementary Figure S2). The circos results revealed the encoding gene along with transcriptional start sites (TSS) and
polyphyletic ancestry of the FOL with other Ascomycetous fungal poly A tail across the full-length genome (Figure 4). The
taxa at highest filter cutoff values. However, at the medium FOXG_04696 gene was found to be located on chromosome7
scale (50% cutoff score) we found similarity index at their low (NC_030992.1 with 87% identity; 99% query coverages; E-value
percentage values, with the other homolog and orthologous 2e−150). The chromosomal map represented the position
members (Figure 3). The multiple sequence alignment results of the gene (FOXG_04696) on chromosome 7 (genomic bp
showed the strong conservation of core residues (red square) 7: 22061–22974; 100% identity; E-value 0.0) (Supplementary
occupied within the functional domain, with the substitution Figure S4).
of some residues at consensus positions (Supplementary
Figure S3). It has been reported that the aldo–keto reductase
superfamily might have been evolutionarily diverged from an Structural Modeling, in silico
ancestral multifunctional oxidoreductases (Jez et al., 1997). Characterization, and Model Validation
However, the presence of similar and identical active sites, The modeler generated 25 predictive models for protein
across the distantly related fungal taxonomic group, explained FOXG_04696 with different discrete optimized potential energy
their convergent evolution as SDRs superfamily (Jez et al., (DOPE) score values. The model with least values for DOPE score
1997). The conserved domain database alignment results for the (21st model; −28563.03 kcal/mol) was selected as a final model
queried protein identified the conserved functional sites that for in silico characterization and docking studies. The predicted
include (both active site and substrate binding site) across the model was visualized through the visualization module of the
evolutionary diverged fungal partners. Discovery Studio 3.0 (Figure 5A). The three putative prominent
binding sites identified in the target protein structure have been
Gene Prediction and Chromosomal shown (Supplementary Table S1). It was found that most of the
Mapping residues involved in binding to ligand (fungicide) were occupied
The BL2seq (Blast-p) results revealed that (XP_018239507.1) from first the major binding, site (binding site 1; metapocket
protein was found to have (97% query coverages; 50% results) of the predicted model. The major catalytic sites inside
identity; E-value 3e−78) with the T4HNR protein of M. grisea the protein occupying all the potential residues that get involved
(XP_003715430.1). By contrast, the BL2seq query with 3IS3 in binding with ligands have been shown in Figure 5B with
resulted into (94% query coverages; 46% identity; E-value red balls showing active sites (Figure 5C). The residues that

Frontiers in Pharmacology | www.frontiersin.org 8 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 3 | The circos visualization map showing the similarities and differences for the SDRs (T4HNR like) among the five major phytopathogenic fungi, retrieved
from genome comparision (based on sequential alignment). The circos map was generated at 50% cutoff score values and drawn using percentage identity
matrices, calculated and obtained during phylogenetic clustering of the protein sequences using ClustalW, and represented the positional conservation and
relationship between the genomic intervals.

constitute the functional ligand-binding sites have been shown with the G factor value 0.12 (Supplementary Figure S5). This
(Figure 5D). confirms the predicted model quality had good stereochemical
The selected model was verified for their stereochemical quality and was close to the template structure. The ProSA
quality assessment. Furthermore, in each case of qualitative results in finding the potential error in the predicted model
assessment, a comparative study was done with experimentally revealed the Z score value −8.11 (Supplementary Figure S6)
solved crystal structures, to check the quality, reliability, accuracy, against the template (3IS3) score value −8.97. The Z score
stability and compatibility of the computationally predicted of other template (1JA9) was found to be −9.67. The ProSA
protein. The Ramachandran plot obtained through RAMPAGE evaluates the qualitative values of the modeled structures based
server revealed that the predicted model FOXG_04696 had on atomic coordinates. The energy plots represent the potential
99.3% residues [97.7% (favored) + 1.6% (allowed)] lying in problems spotted in protein structures. The Z score revealed the
favored region [compared to the experimentally solved and protein structures could be correlated well with crystal structures
X-ray resolved template protein structure (3IS3) where we found of similar lengths, where the positive value corresponds to
98.0% (favored) and 1.6% (allowed) residues against the expected problematic or erroneous parts of the input structure. ProQ is a
values 98.0% (favored) and 2.0% (allowed) regions]. The other neural network-based predictor based on a number of structural
sequential homolog 1JA9 had similar results like 3IS3 98.0% features predicts the quality of protein model. The ProQ result
(favored) and 1.6% (allowed). The PROCHECK module of the showed LG score of 5.677 which represents that the structure
PDBSum server, further justified the stereochemical goodness is of very good quality. The ERRAT score for the modeled
of the predicted model, with 94.2% residues accommodating in structure was found to be 90.40% against the template (99.20%)
the most favored regions (A, B, and L) and only 4.9% residues (Supplementary Figure S7). The Verify 3D evaluated that the
occupied in the additionally allowed regions (a, b, l, and p) predicted protein has 91.30% residues had an average 3D-1D

Frontiers in Pharmacology | www.frontiersin.org 9 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 4 | (A) The comparison of the query protein (FOXG-04696) with the protein sequences of the T4HNR (M. grisea). The two protein sequences (1JA9 with the
FOXG-04696) were aligned using the BL2seq (Blast-p). The hypothetical protein FOXG_04696 show more sequence similarity based on the percentage identities
(49%) and query coverages (97%) with the 1JA9. The other structural homolog the 17-β-hydroxysteroid dehydrogenase (3IS3) had percentage identities (46%) and
query coverages (96%). (B) Prediction of the FOXG_04696 protein encoding gene with coding sequences, transcription start sites and Poly A tail.

FIGURE 5 | (A) Predicted structure of the FOXG_04696 modeled through homology modeling using Modeller v9.19 and visualized through the Discovery Studio 3.0
visualization tool. (B) The big red sphere represents the cavities surrounding the active sites and was visualized using the visualization module of the Discovery studio
3.0. The three binding sites were explored through the meta-pocket server. (C) The three putative binding sites as shown through three different colored red balls.
(D) General view of protein-ligand interaction showing the residues from the active site (FOXG_04696) residues involved in making interaction with ligand (fungicide).

score ≥0.2 (Supplementary Figure S8 and Supplementary (50%) and coil (31%) with interspersed beta sheets (18%) with
Table S1). The quality assessment at various interfaces has extensive H bonding groups [donor and acceptor; with the
resulted in a combined ProTSAV score value which revealed that observed value of 83% against the expected 75% score values,
the predicted protein was stable and had RMSD values in the and mean H = bond energy −1.7; sd = 1.0 (expected −2.0
range of a good model (at green–yellow interface) (Figure 6). sd = 0.8)]. We superimposed the full length predicted protein
The VADAR statistics for quantitative evaluation of the predicted FOXG_04696 (258 residues) over both the template 3IS3 (260
model revealed that the model structurally composed of helical residues) and the 1JA9 (259 residues) to perform structural

Frontiers in Pharmacology | www.frontiersin.org 10 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 6 | The qualitative assessment of the predicted model FOXG_04696 and its comparative evaluation with the X-Ray resoluted template proteins (1JA9) using
the ProtSAV score. (A) The qualitative assessment of the modelled protein (FOXG_04696) based on ProTSAV score. (B) The ProTSAV score for the template protein
1,3,6,8-tetrahydroxynaphthalene reductase (T4HNR) complexed with NADPH and pyroquilon (1JA9). The ProTSAV evaluated the predicted model structures, based
on some popular online servers and standalone tools, and furnishes with a single quality score in case of individual protein structure, along with a graphical
representation and ranking in case of multiple protein structure assessment. In our results, the ProTSAV score was found close to 1JA9 which predicts the model
has reasonable stability and accuracy in terms of qualitative and quantitative parameters.

alignment, using AuStrAlis server. The RMSD deviations on (Supplementary Figure S10). The other substrate binding site in
superposition along the protein carbon backbone were 0.49Å the FOXG_04696 showed the extensive conservation of serine,
(3IS3) and 1.51Å (1JA9) with the FOXG_04696. This further isoleucine and tyrosine residues SER141 , ILE142 , TYR154 , and
confirms the results of the qualitative assessment, and structural TYR196 (represented as SER164 , ILE165 , TYR178 , and TYR223 in
conservation of SDRs proteins among the closely related group 1JA9) (Supplementary Figure S11 and Table 1). However, at
and therefore, their crucial role in the fungal biosystem. These some positions in the FOXG_04696, even the glycine residues
results indicated that the two proteins had a similar structural were found to be extensively conserved, which reflects their
assignment and topological orientation (functional domain and crucial role in NADP binding including GLY13 , GLY17 , and
folds) that predicts their indispensable role. The final predicted GLY93 (represented by GLY36 , GLY40 , GLY116 , GLY209 , and
models were submitted to an online repository, protein modeling GLY210 in 1JA9) which might play an indispensable role and
database (PMDB) under the name SDRs (T4HNR: organism imparts specificity to FOXG_04696.
name: Fusarium oxysporum f.sp. lycopersici) and were provided
with having accession number PM0081606. CATH Results
The structural classification through CATH server revealed
Active Site Prediction that the predicted model belongs to (α+β) type (3), (A) three
The putative ligand binding sites (both major and minor) for layer (aba) sandwich type architecture (3.40), having Rossmann
the predicted protein were identified through Meta-pocket 2.0 fold (3.40.50) and bearing to NADP binding Rossmann fold
server. The conserved domain databases (CDD) server prediction (alpha/beta folding pattern with a central beta-sheet) like
revealed the conservation of the catalytic tetrad (NSYK) ASN115 , domain family protein (30.40.50.720). The functional annotation
SER141 , TYR154 , and LYS158 in the FOXG_04696 which was using Funfam (functional families) revealed the possible
found to be conserved in 1JA9, and were replaced with biological role of the characterized protein based on three
ASN138 , SER164 , TYR178 , and LYS182 with the presence of ontological terms that include biological process, molecular
canonical glycine-rich NADP-binding sites (Supplementary function, and cellular component. The first five significant GO
Figure S9 and Supplementary Table S2). By contrast, the terms in biological processes included secondary metabolite
C-terminal residues providing specificity for substrate binding biosynthetic process (GO: 0044550), secondary metabolite
(NADP) had conserved active site residues (GLY13 , ARG16 , process (GO: 0019748), pigment biosynthetic process (GO:
GLY17 , ILE18 , ARG36 , TYR37 , VAL38 , SER39 , SER40 , ALA63 , 0046148), and sterigmatocystin biosynthetic process (GO:
ASP64 , VAL65 , ASN91 , SER92 , GLY93 , VAL114 , ILE139 , SER140 , 0045461). The significant terms in molecular processes found
SER141 , TYR154 , LYS158 , PRO184 , THR16 , ASP187 , and MET188 ) were versicolorin reductase activity (GO: 0042469), tropinone
compared to the active site residues for template (1JA9) reductase activity (GO: 0050358), NAD+ binding (GO: 0070403),

Frontiers in Pharmacology | www.frontiersin.org 11 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

TABLE 1 | Comparative evaluation of the active sites and other binding site residues for T4HNR Magnaporthe grisea (1JA9) and the predicted protein FOXG_04696.

T4HNR complexed with

NADP(H) and pyroquilon T4HNR Magnaporthe oryzae FOXG_04696 active site Common residues
(1JA9) active site residues (1JA9) predicted active site residues (NCBI-CDD results) (FOXG_04696 and 1JA9)
(X-ray diffraction) (NCBI-CDD results)

GLY36 ARG39 GLY40 ILE41 Catalytic tetrad Catalytic tetrad GLY13 ARG16 GLY17 ILE18
GLY61 SER62 SER63 ALA86 ASN138 SER164 TYR178 LYS182 ASN115 SER141 TYR154 LYS158 SER39 SER40 ALA63 ASP64
ASP87 ILE88 ASN114 SER115 ASN91 SER92 GLY93 GLU95
GLY116 LEU137 THR162 NADP-binding residues NADP-binding residues SER140 SER141 TYR154 LYS158
SER163 , SER164 TYR178 (substrate) (substrate) PRO184 THR186 ASP187
LYS182 PRO208 GLY209 GLY210 GLY36 ARG39 GLY40 ILE41 GLY13 ARG16 GLY17 ILE18 MET188 TYR196
VAL211 THR213 ASP214 MET215 ASN59 TYR60 GLY61 SER62 VAL38 SER39 SER40
PHE216 SER220 TYR223 ILE282 SER63 ALA86 ASP87 ILE88 ALA63 ASP64 VAL65 ASN91
ASN114 SER115 GLY116 LEU137 SER92 GLY93 GLU95 VAL114
THR162 SER163 SER164 LYS182 ILE139 SER140 SER141 TYR154
PRO208 GLY209 GLY210 VAL211 LYS158

THR213 ASP214 MET215 PRO184

Chemical (fungicide) binding LYS185 THR186 ASP187
residues MET188
SER164 ILE165 TYR178 GLY210 TYR189 ALA193 TYR196
MET215 PHE216 SER220
TYR223

The active sites for 1JA9 were retrieved through the X-ray-crystal structure and those for FOXG_04696 were retrieved from the NCBI conserved domain database (CDD)
server. The common residues were obtained from structural alignment. The common residues present in both have been shown in a separate column.

(S, S)-butanediol dehydrogenase activity (GO: 0047512), simulations results into unstable trajectories that finally lead
and alcohol dehydrogenase (NAD) activity (GO: 0004022). into disruption of the complex. By contrast, the realistic
The scattered plot diagram was generated through the ReviGO complexes provide stable behavior (Yunta, 2016). Furthermore,
web server was based on non-redundant GO terms with based on obtained MD trajectories, 1Gbind was computed
scoring values higher is better. The first five significant terms by using MM/GBSA calculations. In many studies, it has
structured around three ontologies, which discussed biological been demonstrated that binding free energies predicted by
processes, molecular function and a subcellular component of MM/GBSA-based rescoring of the docked complexes are in good
predicted protein has been shown (Supplementary Figure S12). agreement with experimental binding affinities (Suenaga et al.,
The subcellular localization and function annotation were 2012; Shen et al., 2013). The Oxathiapiprolin had the least
further revealed through the CELLO2GO server discussed the 1Gbind of –75.50 (±0.54) kcal/mol and XPG docking score of
queried protein sequence, was found to be associated with −1.86 kcal/mol with 17 binding site residues (LEU100 , VAL103 ,
biosynthetic and secondary metabolism processes, with having ILE108 , LEU112 , VAL116 , TRP146 , GLY147 , VAL148 , PRO149 ,
an oxidoreductase activity (88.4%) (Figure 7A). The tag cloud ARG150 , HIS151 , ALA152 , LEU153 , SER155 , ALA156 , SER157 , and
diagram describes the frequent keywords associated with the ALA160 ) of the T4HNR were found to involve in van der
assigned GO terms, and therefore, represents the functional Waals interactions with Oxathiapiprolin. The Famoxadone had
relevance of the proteins and the other associated processes in the 1Gbind of −66.90 (±0.47) kcal/mol and lower XPG score
which their function have been elucidated (Figure 7B). (than Oxathiapiprolin) of −3.30 kcal/mol, it displayed two
hydrogen bonds with key binding site residues TYR154 and
Protein–Fungicide Interaction THR186 and 27 residues were found to be involved in making
The modeled protein FOXG_04696 was docked with all the van der Waals interactions GLY13 , SER15 , ARG16 , GLY17 , ILE18 ,
37 fungicides to generate their binding mode and dynamic GLY19 , TYR37 , VAL38 , ASN91 , SER92 , GLY93 , ILE94 , GLU95 ,
simulations was done to refine the best pose with allowed ILE139 , SER140 , SER141 , ILE142 , SER143 , TYR154 , LYS158 , PRO184 ,
conformational change in the FOXG-04696 (Rachman et al., LYS185 , THR186 , ASP187 , MET188 , TYR189 , ALA192 , ALA193 , and
2018). We have evaluated the protein–fungicide interaction TYR196 ) within 4Å binding site region of Famoxadone with
through YASARA and Glide-based molecular docking program. T4HNR. The 3D surface view of the docked Famoxadone–
It was found that both the tools have discrepancies in results FOXG_04696 complex has been shown to represent the putative
for accurate pose prediction among the various putative docking H bond acceptor and donor group (Figure 8A). The functional
poses, revealed through scoring functions, which might leads H bond acceptor and donor group from protein major binding
into conclusion that, docking scores are not sufficiently precise sites of proteins have been shown in Figure 8B. The 3D
to represent the protein ligand binding affinity (Suenaga et al., structure of two effective ligands (fungicides) has been shown in
2012). MD simulations analysis of the docked complexes Figures 8C,D.
discriminated the correct docking poses from decoy poses, The protein–fungicide docking was further analyzed through
as the unstable and incorrectly docked structures during MD the YASARA, an auto dock based tool for molecular docking

Frontiers in Pharmacology | www.frontiersin.org 12 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 7 | (A) Functional annotation of the FOXG_04696 measured in the form of gene ontology enrichment analysis. The three ontological terms used were the
molecular function, biological process involved, and cellular location. The sub-cellular localization of the protein was shown as a big pie chart and were retrieved
through the Cello predictor. (B) The tag cloud diagram showing the keywords that are frequently associated with the FOXG_04696 protein and indicates its probable
function in biosynthetic mechanisms and metabolism.

and virtual screening to calculate the docking score (kcal/mol) (SDR) protein. We have investigated the X-Ray determined
and dissociation constant (K d ) µM. The maximum YASARA crystal structure of the T4HNR (1JA9) complexed with
score was found to be associated with the Oxathiapiprolin fungicide pyroquilon to find out the residues that were
(7.81 kcal/mol) with least dissociation constant K d value involved in binding with T4HNR in an accurate and flexible
1.86 (µM) followed by the Famoxadone (7.65 kcal/mol; K d docking poses (Supplementary Figure S13). The investigation
value 2.43 µM). The protein–ligand docking through the revealed that pyroquilon docked with maximum residues that
YASARA showed the efficient, stronger, and stable binding with constituted the major binding sites (active site). In this way,
positive YASARA score∗ (YASARA scoring∗ , where positive one could predict that the fungicides that target the active
energy means stronger binding and negative energy means no site residues of T4HNR protein with maximum interacting
binding) (Chen et al., 2015) with Oxathiapiprolin followed by residues (more accurate docking pose) and better protein
Famoxadone. The putative H bond acceptor and donor group ligand contacts (flexible docking) could have better binding
in ligand Famoxadone were shown through a receptor mesh efficiency, and therefore, would be useful for disrupting
diagram (Figure 9A). The 3D diagram of the Famoxadone that the functional mechanism of T4HNR. In our results, we
interacted with crucial residues from the major binding site has have evaluated the comparative docking efficiency (pyroquilon
been shown (Figure 9B). as control) to investigate the binding affinity measured in
The YASARA based soring, dissociation constant, and contact the form of YASARA-based docking score, and dissociation
receptor residues involved in binding with the FOXG_04696 constant of the docked complexes. The residues involved
have been shown in Supplementary Table S3. Since the in making feasible and accurate docking of T4HNR with
fungicide pyroquilon is an efficient fungicide used against pyroquilon were GLU118, SER164 , ILE165 , ALA166 , TYR178 ,
the rice blast pathogen (M. grisea PDB ID: 1JA9), and PRO208 , GLY209 , GLY210 , MET215 , PHE216 , ASN219 , SER220 ,
targets the residues, forming active sites of the T4HNR TYR223 , LEU240 , and ILE282 . The computational screening

Frontiers in Pharmacology | www.frontiersin.org 13 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 8 | (A) The overall 3D surface view of the modeled protein FOXG_04696 represented to display all the possible H-bond donor and acceptor group when
complexed with ligand (Famoxadone). (B) The interaction of the ligand (Famoxadone) with protein FOXG_04696 with the possible H-bond donor and acceptor
groups, near the ligand interacting or binding sites (active sites). (C) The 3D representation of the ligand Oxathiapiprolin. (D) The 3D structure of the ligand
Famoxadone.

and docking studies of 37 fungicides with the FOXG_04696 TRP146 , GLY147 , VAL148 , PRO149 , ARG150 , HIS151 , ALA152 ,
revealed that Oxathiapiprolin followed by the Famoxadone LEU153 , SER155 , ALA156 , SER157 , and ALA160 rather than
binds with maximum YASARA score and least dissociation the specified docking sites. The molecular complexes formed
constant (K d ). after protein–fungicides interaction for different fungicides
The molecular docking and virtual screening through Glide has been visualized through the visualization tool of Discovery
XP ranked ligands based on an accurate pose prediction(the studio 3.0 and have been represented (Supplementary Figure
ligand ability to bind for a specific receptor conformation) for S14).
each-protein—fungicide complex in order to separate those
ligands that don’t bind, in a ranked list. Furthermore, analysis MD Simulations
of the YASARA results for the two top scored docked protein– In MD simulations analysis, the FOXG_04696–Oxathiapiprolin
fungicide complexes (Oxathiapiprolin and Famoxadone) complex had an average potential energy of −113166.16 kcal/mol
revealed that Famoxadone docked with FOXG_04696 in an which disclosed the steadiness of the complex. The average
accurate and flexible docking pose with residues that constituted RMSD for the FOXG_04696 backbone and the Oxathiapiprolin
the major binding site (active site including the catalytic tetrad) were 2.49 and 2.42Å, respectively (Supplementary Figure S15A).
GLY13 , SER15 , ARG16 , VAL38 , SER39 , SER40 , ASP64 , VAL65 , The average RMSF for backbone and side chain for the
SER66 , SER92 , GLY93 , ILE94 , LYS110 , and VAL114 . Since, we did FOXG_04696 accommodating with the Oxathiapiprolin were
not find any stable docking conformation for stable binding of 1.54 and 1.70Å, respectively (Supplementary Figure S15B).
the Oxathiapiprolin at that particular specified docking site (like Oxathiapiprolin–FOXG_04696 complex exhibited five hydrogen
Famoxadone). The lower ranking of the Oxathiapiprolin (higher bonds with water (SER143 , ALA144 , VAL145 , GLY147 , and
XPG score) binding over the Famoxadone (lower XPG score) SER155 ), three water mediated hydrogen bonds (VAL148 , ARG150 ,
could be interpreted from the fact that the Oxathiapiprolin was and LYS166 ), four hydrophobic and water-mediated hydrogen
found to docked in an alternative conformation docked with bonds (TRP146 , ALA152 , ALA156 , and ALA159 ) with seven
the residues that were either absent from any major or minor hydrophobic interactions (LEU100 , VAL103 , ILE108 , LEU112 ,
binding site, or were present beyond the limit required for PRO149 , LEU153 , and ALA160 ) with the key binding site residues
an accurate docking pose prediction. The visualization of the to form a stable complex (Supplementary Figure S15C). We
Glide XP docked complexes revealed that Oxathiapiprolin have shown the protein–ligand interaction 2D diagram as
bounded with LEU100 , VAL103 , ILE108 , LEU112 , VAL116 , visualized through the Discovery Studio 3.0 (Figure 10A).

Frontiers in Pharmacology | www.frontiersin.org 14 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 9 | (A) The ligand Famoxadone represented through the wire mesh diagram to show the probable H-bond donor or acceptor groups that could have crucial
role in protein–fungicide interaction. (B) 3D representation of the ligand molecule when docked with the FOXG_04696 showing the crucial residues of protein that
have important contribution in binding with the ligand donor or acceptor group.

By contrast, the FOXG_04696–Famoxadone complex has an among all the docked 37 fungicides for the specified docking
average potential energy of −112628.96 kcal/mol disclosed the site (active site or major binding site). The 3D representations
steadiness of this complex (Figure 10B). The average RMSD for the protein–fungicide interaction for both Oxathiapiprolin
for the T4HNR backbone and the Famoxadone were found to (Figure 11A) and Famoxadone have been shown (Figure 11B).
be 2.83 and 1.20Å, respectively (Supplementary Figure S10C). The correlation plot showing the values of correlation coefficient
The average RMSF for the backbone and side chain of the R2 = 0.335 based on binding affinities (kcal/mol) and MM/GBSA
T4HNR to accommodate the Famoxadone were reported 1.30 binding free energy (1Gbind ) calculations showing the strong
and 1.86Å, respectively (Supplementary Figure S15D). The correlation between the predicted binding free energies and
Famoxadone exhibited seven hydrogen bonds (ASN91 , SER92 , ranking affinities/scoring of the fungicides for the docked
GLY93 , GLU95 , SER141 , THR186 and ASP187 ) and 15 hydrophobic complexes (Figures 11C,D).
interactions (GLY13 , ARG16 , ILE18 , VAl38 , VAL65 , ILE94 , ILE139 , We have compared our docking results, both from the
ILE142 , TRP146 , LYS158 , PRO184 , MET188 , TYR189 , ALA192 , and Glide XP docking and the YASARA protein–ligand docking
TYR196 ) with the key binding site residues in forming a stable tool. Furthermore, the docked complexes were rescored through
complex (Supplementary Figure S15E). Both the complexes MM/GBSA free energy binding calculations to validate the
were relatively stable with the lesser average potential energy docking based results for ranking the binding affinities of
but Famoxadone displayed more interactions with FOXG_04696 docked ligands (fungicides). The ultimate goal for MM/GBSA
compared to Oxathiapiprolin in 50 ns MD simulations, with calculations was to estimate the enthalpy change on ligand
lesser RMSD and RMSF values and best protein ligand contacts binding, through comparison of an average enthalpy change

Frontiers in Pharmacology | www.frontiersin.org 15 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 10 | (A) The 2D representation for the docked complex of the FOXG_04696. The figure showed the putative residues involved in interaction with
Famoxadone. The different colors have been used for showing the different types of molecular interactions involved. (B) MD simulations trajectories for the
FOXG_04696–Famoxadone complex showing the average potential energies of the docked complexes during the 50-ns MD simulations. (C) Plot of the root mean
square deviation (RMSD) of Cα of T4HNR (protein) and the FOXG_04696–Famoxadone (complex). RMSDs were calculated using the initial structures as templates.

for the bound and unbound states. The MM/GBSA results to have maximum interacting residues for sites that constitute the
re-ranked the docked complexes in terms of their accurate minor binding sites or second probable binding site (metapocket
pose prediction and efficacy for binding affinities (Table 2). results; Supplementary Table S1). The MD simulations and
We did not find any significant docking pose for the the YASARA based docking for the Famoxadone-protein was
Oxathiapiprolin–FOXG_04696 complex, particularly on the found to have residues from a major binding site that include
specified docking site (active site including catalytic tetrad) of GLY13 SER15 , ARG16 , VAL38 , SER39 , SER40 , ASP64 , VAL65 ,
(FOXG_04696) even at minimized grid space. However, MD SER66 , SER92 , GLY93 , ILE94 , LYS110 , and VAL114 . By contrast, the
simulations of the FOXG_04696–fungicide docked complexes YASARA based Oxathiapiprolin–protein complex was found to
revealed the stable binding of the Famoxadone over the have residues like GLY13 , ARG16 , VAL38 , ASP64 , VAL65 , SER66 ,
Oxathiapiprolin with all crucial residues occupying interactions LYS67 , SER92 , GLY93 , ILE94 , ASP109 , LYS110 , LEU112 , GLY113 ,
with ligand in MD simulations. The 2D diagram of the and VAL114 , whereas the MD simulations analysis covered the
protein–fungicide complexes, when visualized through the residues not lying in major binding site (meta-pocket results) or
Discovery Studio 3.0 tool, we found some interesting results. located at other binding cavities rather than the residues that
The YASARA-based docking with Famoxadone was found were involved in the main binding sites (Supplementary Table
comparable to the MD simulations results, as the residues S4).
involved in the protein–ligand contact were found to be
similar, and were found to be involved/constitute the major Protein–Protein Interaction Network
binding (active sites) of the FOXG_04696. The YASARA The functional interactive network formed by the FOXG_04696
based docking score and dissociation constant obtained has protein at the highest confidence level (0.90) has been
been plotted (Supplementary Figure S16). By contrast, the shown in Figure 12A. The predicted protein was shown
Oxathiapiprolin–protein complexes, when analyzed were found to have an interaction with the fatty acid synthase subunit

Frontiers in Pharmacology | www.frontiersin.org 16 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 11 | (A) The Oxathiapiprolin–FOXG_04696 interaction results during 50-ns MD simulations. (B) Interaction of the Famoxadone with FOXG_04696 showing
the residues involved in protein–fungicide docking with different type of molecular interactions. (C) The correlation plot showing the values of correlation coefficient
R2 = 0.335 based on binding affinities (kcal/mol) and MM/GBSA (1G) binding free energy calculations showing the strong correlation between the binding free
energies 1Gbind and binding affinities for ranking/scoring of the fungicides in the docked complexes. The three later symbols with different colors have been used for
representing the ligands. (D) The scatter plot displaying docking (XPG) and binding free energy MM/GBSA (1Gbind) scores represented in the quadrant view form for
all the 19 protein–fungicide docked complexes.

beta dehydratase (FOXG_06392) and the fatty acid synthase percent inhibitions recorded on 4th day were 90.53, 74.42,
subunit alpha dehydratase (FOXG_06391). However, at high 63.04, and 44.36 at 50, 100, 150, and 200 µL concentrations,
confidence level (0.70), we found the interaction of our respectively. By contrast, the percent inhibitions recorded on
predicted protein FOXG_04696 with acetyl-CoA carboxylase 8th day post inoculation were 25.73, 19.99, 11.22, and 7.04
(FOXG_02375; interacting score 0.847). The interaction network at 50, 100, 150, and 200 µL concentrations, respectively. The
of FOXG_04696 at high confidence level has been shown in growth inhibition recorded on the 4th and 8th days at different
Figure 12B. The interactive associative protein network formed concentrations of fungicides has been shown in Figures 13A-I,II,
by various interacting partners, with their interacting score respectively. The statistical data for growth measured at different
annotation identities and accession identities values have been concentrations and on even days have been shown in bar diagram
shown (Supplementary Table S5). (Figure 13B).

In vitro Assessment of Fungicides In silico Toxicity Assessment

The Famoxadone solution used for in vitro assessment The Famoxadone was checked and evaluated for toxicity
against the FOL pathogen showed growth inhibition at assessment for its safe environmental disposition. The FAF-drugs
each and every increasing concentration of fungicides. With 4.0 tool performed the computational prediction of some
increasing concentrations of fungicides, the growth rates were ADME-Tox properties (adsorption, distribution, metabolism,
correspondingly retarded and sporulation was reduced. The excretion, and toxicity) for Famoxadone and it was found that
maximum growth inhibition was recorded on eighth day post the drug is non-carcinogenic and acceptable (Yadav et al., 2017).
inoculation. The percentage inhibitions measured in the form Furthermore, the ligand (Famoxadone) was found to follow
of radial growth were subjected to statistical analysis. The Lipinski Rule of five for drug likeness with molecular mass

Frontiers in Pharmacology | www.frontiersin.org 17 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

TABLE 2 | Comparative evaluation of protein-ligand (fungicide) docking interactions from YASARA programme and XP Glide score (docking score) values.

S. No. Fungicide YASARA score Dissociation XPG score (kcal/mol) MM/GBSA (1Gbind ) (kcal/mol)
constant (K d ) (µM)

1. Oxathiapiprolin 7.81 1.86 −1.89 ± 0.32 −75.50 ± 0.54

2. Famoxadone 7.65 2.43 −3.30 ± 0.28 −66.90 ± 0.47
3. Metiram 4.10 976.17 −6.13 ± 0.25 −55.63 ± 0.38
4. Dithane 4.10 976.17 −6.13 ± 0.22 −55.63 ± 0.42
5. Pterostilbin 6.29 24.17 −4.30 ± 0.20 −50.69 ± 0.56
6. Tolclofos-methyl 4.91 249.18 1.38 ± 0.10 −40.70 ± 0.47
7. Fluberidazole 6.63 13.64 −1.79 ± 0.33 −37.50 ± 0.52
8. Tolprocarb 6.64 13.50 −4.55 ± 0.64 −35.02 ± 0.64
9. Cymoxanil 5.70 57.90 −3.35 ± 0.24 −33.39 ± 0.76
10. Carbendazim 5.78 57.09 −2.98 ± 0.38 −32.28 ± 0.49
11. Coumarin 6.13 32.05 −1.19 ± 0.19 −14.58 ± 0.36
12. Pyraclostrobin 7.05 6.69 −7.02 ± 0.10 −14.57 ± 0.39
13. Triazoquinoline 6.43 19.25 −0.45 ± 0.30 −13.19 ± 0.33
14. Fludioxonil 5.98 40.87 −3.53 ± 0.24 −8.07 ± 0.13
15. Iprodione 5.98 41.08 −4.42 ± 0.34 −3.30 ± 0.31
16. Ethyl phosphonate 5.05 197.07 −4.78 ± 0.25 −2.14 ± 0.51
17. Prothioconazole 5.53 87.65 1.64 ± 0.11 8.94 ± 0.43
18. Benomyl 5.58 80.15 2.53 ± 0.39 29.98 ± 0.74
19. Prochloraz 5.43 103.25 5.68 ± 0.36 43.86 ± 0.80

The YASARA based dissociation constant (Kd) have been given in a separate column. The docked complexes were further rescored for binding free energy assessment
using the MM/GBSA method. The MM/GBSA based binding free energies have been arranged in the increasing order, reflects the decreasing order of stability and
steadiness of the complexes. The Glide based interaction of the protein-fungicide docking complexes and binding free energy assessment through MM/GBSA were set
up with three replication and data were analyzed by Mean (± SE) was calculated from three replicates for each of the docked complexes. We have shown the docking
score values only for the significant docked protein-ligand complexes (19 fungicides; as others were found unable to dock at the intended site through Glide XP dock and
also had positive free energies as calculated through MM/GBSA approach.

FIGURE 12 | The protein–protein interaction associative network for the FOXG_04696 through STRING server. The active interaction sources were set based on the
seven parameters including experiments, co-expression, gene fusion, co-occurrence, databases, text mining, and neighborhood. (A) The interactions analyzed at
the highest confidence level (0.90) with maximum five interacting partners from both shells of interactors. (B) Interaction at high confidence level (0.70). The color
nodes describe query proteins and the first shell of interactors, whereas white nodes are the second shell of interactors. The large node size represents
characterized proteins and smaller nodes for uncharacterized proteins.

374.000000 (<500 Da), hydrogen bond donor 1, hydrogen (0.7751), non-AMES toxic (0.5395), and non-inhibitor (0.8941),
bound acceptor 6, with Log P score values 4.699, and molar with weak hERG (the human Ether-à-go-go-Related Gene)
refractivity 103.70. The admetSAR results predicted that the inhibitor (0.9732) and with non-required carcinogenetic
selected ligand (Famoxadone) was found to be non-carcinogen (0.4799).

Frontiers in Pharmacology | www.frontiersin.org 18 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

FIGURE 13 | In vitro assessment of the fungicide on growth response of FOL pathogen. Four different concentrations were selected 50, 100, 150, and 200 µL along
with control at 4 days interval (A-I) and at 8 days interval (A-II). (B) The growth was recorded at even day’s interval (2, 4, 6, and 8) and the percentage inhibition was
calculated using statistical tools.

DISCUSSION formation of intermediates including (+) scytalone, 2,3,8-

trihydroxynaphthalene (T, H, and N). DHN may be then oxidized
Vascular wilt caused by F. oxysporum f.sp. lycopersici (FOL) and polymerized to form melanin (Bell and Wheeler, 1986;
is very destructive and widespread plant disease that causes Feng et al., 2001). The 1,3,6,8-tetrahydroxynaphthalene/1,3,8-
enormous economic losses. The wilt pathogen directly penetrates trihydroxynaphthalene reductase gene has been isolated from
roots and colonizes the vascular tissue (Inoue et al., 2002). M. grisea (Vidal-Cros et al., 1994). The melanin biosynthetic
One of the most important characteristics of Fusarium pathway was recently demonstrated in other Ascomycetous
wilt disease is the discoloration of vascular tissues, which fungi based on sequence similarity, percent identity to the
is due to the brownish-black melanoid pigment. Melanin T4HNR protein (encoded by teh gene). Engh et al. (2007)
biosynthesis, therefore, is a good target for designing the reported the DHN-based melanin pathway in the Sordaria
antifungal agents. The biosynthesis of fungal melanin is macrospora, an Ascomycetous fungal model system, which
derived from a pentaketide intermediate which cyclized into accumulates the melanin during its sexual development. It was
1,3,6,8-tetrahydroxynaphthalene. The final step of the reaction found that the T4HNR protein showed sequence similarity
is accomplished by series of reductions and dehydrations and homology with Aspergillus fumigatus (taxid: 746128)
and forms 1,8-dihydroxynaphthalene (DHN) through the (51.8% identity), Cochliobolus heterostrophus (taxid: 5016) (79.1%

Frontiers in Pharmacology | www.frontiersin.org 19 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

identity), M. grisea (taxid: 148305) (51.6% identity), and O. floccusm, the complemented M. grisea buf mutants produced
Neurospora crassa (taxid: 5141) (96.3% identity). By contrast, a black pigment like a wild-type strain, and the mutants were
the polyketide synthase (encoded by pks gene) (DHN melanin found to restore the pigment biosynthesis, which predicted that
enzyme) had sequence similarity with A. fumigatus (42.2% the existence of functional homology exists in between the fungal
identity), C. heterostrophus (46.0% identity), Colletotrichum genera for the melanin biosynthetic mechanism (Eagen et al.,
lagenarium (66.5% identity), M. grisea (69.6% identity), and 2001).
N. crassa (85.6% identity) (Engh et al., 2007). The orthologs It was reported that the polyketide synthases involved in
for the genes encoding for enzymes polyketide synthase fungal DHN melanin biosynthetic pathways belong to the
(pks), a tetrahydroxynaphthalene reductase (teh), a scytalone group of iterative type I polyketide synthases similar to fatty
dehydratase (sdh), and a trihydroxynaphthalene reductase (tir) acid synthases (Hopwood and Sherman, 1990) and the PKS
were used from other Ascomycetous fungi (mentioned above) reported for S. macrospora was predicted to contain a β-ketoacyl
to retrieve the sequences of the above genes and further synthase, two acyl carrier protein domains, thioesterase, an acetyl
for their experimental demonstration in S. macrospora (Engh transferase, and two acyl carrier protein domains. Furthermore,
et al., 2007). Interestingly, the polypeptide products obtained the comparative analysis of non-ribosomal peptide synthetases
after comparative sequence analysis had significant homology (NRPSs) and polyketide synthases (PKSs) of 12 different species
to DHN-melanin pathway enzymes of other filamentous belonging to Fusarium genera revealed the 52 NRPSs and 52
ascomycetes, and confirm the role of DHN melanin in PKSs orthology group (Hansen et al., 2015). The study revealed
S. macrospora. (Engh et al., 2007). By contrast, the homology the conservation of eight NRPSs and (NRPS2–4, 6, 10–13) and
search of the FOXG_04696 (XP_018239507.1) with S. macrospora two PKSs (PKS3 and PKS7) (Hansen et al., 2015). However,
(TER) and M. grisea (TER), the protein Blast-p results revealed existence of the DHN based melanin in the FOL is rather
the 99% sequence query coverages and 51% sequence identity controversial as it was reported that the PKS encoding gene for
with S. macrospora (XP_003345723.1), 93% query coverages and DHN melanin biosynthesis is not present in bikaverin producing
49% sequence identity with M. grisea (TER) (PDB ID 1JA9), Fusarium genera including F. verticilloides, F. oxysporum, and
and only 91% query coverages and 46% sequence identities F. fujikuroi (Kroken et al., 2003). By contrast, Amany and
with M. grisea (TIR). Based on such in silico-based comparative Ellil (2005) characterized the brown colored melanin, in the
studies, one could predict the existence of DHN melanin FOL pathogen, and also evaluated sensitivity of the FOL
pathways in other filamentous fungi (Engh et al., 2007). However, pathogen against the Tricyclazole and Chlobenthiazone (melanin
the reduction reactions in the fungal DHN melanin pathway biosynthesis inhibitor). In the last few years, several melanin
can be performed by only one hydroxynaphthalene reductase, biosynthetic inhibitors have been designed to target various
whereas other ascomycetes (M. grisea and S. macrospora) utilize phytopathogenic fungi. Furthermore, FOXG_04696 have been
two reductases the 1,3,8-THN reductase (3HNR) and the 1,3,6,8- shown to have alcohol dehydrogenase (NAD) (GO: 0004022) and
THN reductase (T4HNR). In other cases, scytalone dehydratase NADH binding activity (GO: 0070404). Corrales et al. (2011)
was considered to activate both dehydration steps, of the reported an alcohol dehydrogenase gene (SDR), adh1, has dual
scytalone and vermelone (Bell and Wheeler, 1986). Based on fermentative and oxidative functions, and is involved in the
Blast-p annotation, we have found only one hydroxynaphthalene fungal (FOL) virulence in tomato plants. In this context, the
reductases (T4HNR) in the FOL pathogen, and more identical functional relevance of the FOXG_04696 could be predicted from
in sequential homology to the S. macrospora rather than the conserved functional motif and domains, measured in terms
M. grisea. The Blast-p search revealed the sequential similarity of gene ontology, and/or shared domain–domain interaction.
and homology with our target protein (the FOXG_04696) Since, the protein structure is 3–10 times more conserved
with 49% identity and 93% (query coverages) (FOXG 04696: than its sequence (Illergård et al., 2009), and the shared
XP_018239507.1). By contrast, the Blast-p results with the protein domains might be useful for structural and functional
M. grisea 1,3,8-trihydroxynaphthalene reductases (TIR) (PDB annotation of genes or their encoded products. This could
ID: 1G0O) revealed the lesser sequence similarity with the be possibly employed to evaluate the molecular functions
FOXG_04696 (46% identity and 91% query coverages) which and biological processes of interacting proteins or domains.
reflected, the identity of our target protein as the T4HNR and The functional annotation as revealed through CATH server
were found to be more closer to the M. grisea T4HNR (1JA9). revealed that the FOXG_04696 belong to SDR family and
Moreover, the scytalone dehydratase (EC: 4.2.1.94) protein might have possible role in versicolorin reductase activity
(KEGG ID: FOXG_13320; Uniprot ID: A0A0D2YAJ4; NCBI ID: (GO: 0042469), tetrahydroxynaphthalene reductase activity
XP_018252510.1) and the mRNA (XM_018393275.1) have been (GO: 0047039), (S,S)-butanediol dehydrogenase activity (GO:
well characterized in the FOL pathogen. 0047512), and tropinone reductase activity (GO: 0050358). The
The DHN pathway for melanin biosynthesis is reported significant biological process measured in terms of gene ontology
in many other fungi including Ophiostoma floccosum. It was was melanin biosynthetic process (GO: 0042438), secondary
demonstrated that the hydroxynaphthalene reductases (HNR) metabolite (bikaverin, fumonisins, fusaric acid, and fusarins)
of the fungus O. floccosum, shared the functional homology biosynthetic processes (GO: 0044550), pigment biosynthetic
with other fungal HNR (Eagen et al., 2001). For instance, the process (GO: 0046148), sterigmatocystin biosynthetic process
HNR reductases deficient buf mutant of the rice blast fungus (GO: 0045461), butanediol metabolic process (GO: 0034077),
M. grisea when provided with the functional HNR reductases of and acetoin metabolic process (GO: 0045149). The structure

Frontiers in Pharmacology | www.frontiersin.org 20 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

of the FOXG_04696 was predicted based on comparative were good enough signifying the consistency of the model
modeling. It was reported that up to date, comparative prediction and explained that the predicted model was reliable
modeling is the most successful and accurate method as and satisfactory, as it was reported that, for a model having
evolutionarily related proteins usually share a similar structure good resolution (approximately 2.5–3.0Å), the ideal score values
(sequence identity >30%) (Errami et al., 2003; Choong for Verify-3D should be 80%, and that for the ERRAT around
et al., 2011) and structural dynamics is the cornerstone of 95% (Colovos and Yeates, 1993). Furthermore, the predicted
the protein function and its regulation (Berezovsky et al., model was measured in terms of its quality from PROSA score
2017). values. The Z score value for the predicted structure was -8.07
The functional conservation of protein homology was also (against the X-ray resolved template protein having Z score value
evaluated based on the protein interaction networks. Since, −9.63), which is within the range observed for the native set
protein–protein interaction studies are mediated by a limited set of proteins of the same size. It was reported that the Z score
of domain–domain interactions (Itzhaki et al., 2006; Reimand values for any modeled structure lying outside the range of native
et al., 2012), and protein domains represent the structural, proteins that were resolved through X-ray and NMR predict the
functional, and evolutionary unit of proteins (Vogel et al., 2004; erroneous structure (Wiederstein and Sipp, 2007). This was also
Jin et al., 2009). The STRING based results for finding the T4HNR confirmed from the ProtSAV score values as all the qualitative
interacting partner in S. macrospora revealed that at confidence parameters measured the predicted model lying in the zone of
level from high to highest the T4HNR (teh) interacted with good resolution (2.5–3.0Å).
fatty acid synthase alpha subunit reductase (XP_003349949.1), In our results, we have evaluated and compared the binding
fatty acid synthase beta subunit dehydratase (XP_003349948.1), efficacy of commercial fungicides that could be used against the
and 3-oxoacyl (acyl carrier protein) synthase (XP_003351602.1). FOL pathogen to control the Vascular wilt disease. The crystal
Moreover, the STRING based results for our characterized and structure of the T4HNR complexed with the NADP(H) and
predicted model the FOXG_04696, revealed the same interacting pyroquilon (1JA9) revealed that fungicide pyroquilon binds with
partner at highest confidence level values such as fatty acid the crucial residues forming active site of the T4HNR protein
synthase subunit alpha-reductase (FOXG_06391), fatty acid and therefore, interrupt its functional mechanism. The fungicides
synthase subunit beta hydratase (FOXG_06392) (enoyl-[acyl that interact with the residues forming active sites or interact
carrier protein] reductase (NADH) activity) fatty acid synthase with the major residues that form the catalytic center of protein
subunit beta hydratase (FOXG_15138), and fatty acid synthase might have good results, for disrupting the functional aspect of
subunit beta-dehydratase (FOXG_14342) indicating the similar proteins and therefore, would affect its possible biological roles.
functional association with the FOXG_04696 protein. The metapocket server analyzed all the possible binding sites that
might be occupied with the ligands, during the protein–fungicide
interactions. The structural alignment unravelled the conserved
In silico Characterization and Model T4HNR and replaced the key residues such as TYR178 with
Validation TYR154 , LYS182 with LYS158 , PRO208 with PRO184 , THR213 with
The functional characterization of both template 1JA9 and THR186 , ASP214 with ASP187 , MET215 with MET188 , and TYR223
the predicted model FOXG_04696) through the ScanPROSITE with TYR196 in the FOXG_04696. It was found that all the
program revealed that the input protein sequences have signature replaced residues in the FOXG_04696 were present in either
sequences, belonging to SDR family. A broad range of different major or another major (first two) cavities predicted by the
activities is catalyzed by the enzyme (SDRs) that includes metapocket server. In our previous studies, the structure of the
metabolism of organic biomolecules such as carbohydrates, functional domain of the proteins that belong to the WRKY gene
lipids, amino acids, steroids, cofactors, and aromatic compounds superfamily members has been modeled for its qualitative and
and act in redox sensing (Tang and Le, 2014). Sequence quantitative evaluation, to unravel the DNA–protein interaction
analysis revealed that the FOXG_04696 belongs to classical studies, in a stimulus-specific manner in tomato (Aamir et al.,
SDRs and has the conserved catalytic tetrad (NSYK) composed 2017, 2018). In our results, the computational screening revealed
of ASN115 SER141 TYR154 and LYS158 . The three conserved the docking site and energy score values for all the 37 fungicides,
residues including SER141 , TYR154 , and LYS158 form the to evaluate their efficacy against the FOL pathogen. The fungicide
structural motif with ASN115 through H bonding with other Famoxadone interacted with maximum energy (kcal/mol) with
residues. The superimposition of the template protein over the the key residues that constituted the prominent active site.
predicted model resulted into the structural resemblances with
the minimum RMSD (0.47Å) and the relative RMSD values
(0.025Å). The superposition results also aligned the identical Protein–Fungicide Docking and MD
residues found in between the T4HNR and the FOXG_04696 Simulations Analysis
protein. The optimized model was found to be suitable based The MD simulations of the protein–fungicide interaction
on several qualitative backgrounds including the RAMPAGE, reflected the time-dependent behavior of the biological
ProSA, ERRAT, PROCHECK (PDB Sum), and Verify-3D. The complexes. The molecular docking and virtual screening
Ramachandran plot which evaluated that the predicted models revealed the two better fungicides including the Oxathiapiprolin
were closer to the template (98% residues lying in the favored and Famoxadone. The stability of the protein–fungicide docked
regions). The ERRAT score values (92%) and Verify-3D results complexes was measured at 50-ns MD simulations. It was

Frontiers in Pharmacology | www.frontiersin.org 21 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

found that both Oxathiapiprolin and Famoxadone disclosed famoxadone), and also had lesser K d (Oxathiapiprolin followed
the steadiness of the docked complexes, with an average by Famoxadone). The MD simulations of Oxathiapiprolin
potential energy of −113166.16 and −112628.96 kcal/mol, complex had minimum interaction energy and K d values, but
respectively. However, Famoxadone had comparatively better the residues involved in interaction were non-significant, and
docking with XPG score of −3.30 kcal/mol (compared to were present beyond the binding sites (both major and minor)
Oxathiapiprolin; XPG score of −1.87 kcal/mol) along with lower of target receptor protein. The residues that were found to be
values of RMSD and RMSF for protein–ligand contact, and involved in Oxathiapiprolin binding during MD simulations
better interaction within the particular specified docking site were LEU100 , VAL103 , ILE108 , LEU112 , VAL116 , TRP146 , GLY147 ,
(within 4 Å binding site region of FOXG_04696). The lower VAL148 , PRO149 , ARG150 , HIS151 , ALA152 , LEU153 , SER155 ,
XPG score −3.30 kcal/mol (more negative), and lesser RMSD ALA156 , SER157 , and ALA160 . Moreover, Famoxadone had good
and RMSF values for Famoxadone-FOXG_04696 (compared binding affinity from all the platforms with having maximum
to Oxathiapiprolin_FOXG_04696) predicted the stability and residues from first binding site (major) including catalytic
reproducibility of the docking results, to find the crystallographic tetrad, of the FOXG_04696. The residues involved with the
relevant and accurate binding pose. Famoxadone binding during MD simulations were GLY13 , SER15 ,
We analyzed our docking results from two different molecular ARG16 , GLY17 , ILE18 , GLY19 , TYR37 , VAL38 , ASN91 , SER92 ,
docking and virtual screening platforms including the Glide GLY93 , ILE94 , GLU95 , ILE139 , SER140 , SER141 , ILE142 , SER143 ,
XP dock and an auto docked-based YASARA server. Finally, TYR154 , LYS158 , PRO184 , LYS185 , THR186 , ASP187 , MET188 ,
the complexes having good docking score, better K d values, TYR189 , ALA192 , ALA193 , and TYR196 (exclusively forming
and accurate docking poses were further refined and rescored major binding site of the receptor protein) (Supplementary
through the MD simulations, and MM/GBSA methods, to Table S4).
validate the top scored docking results. In each case, we found It has been reported that multiple orientations (multiple
that the Famoxadone docked complex with FOXG_04696 had different conformations adopted by ligands upon binding)
good docking score, with accurate docking pose, and was could be involved in binding a ligand with proteins, and
reliable and reproducible. The YASARA results evaluated the small conformational changes might have big effects on
docking calculations based on YASARA score and dissociation binding affinities (Mobley and Dill, 2009). Furthermore, these
constant (K d ). Based on YASARA results, one could predict binding events are highly affected by multifarious factors,
that fungicides having high YASARA scores and low K d such as waters, ions, or cofactors, protonation state (changed
must bind with the receptor protein FOXG_04696 in a good protonation state on ligand binding), and/or conformational
docking pose. However, MD simulations analysis for these or solvation entropies that could have unexpected involvement
complexes (top docking score) were either failed to bind with and therefore, play unpredictable roles, in deforming the
the target receptor protein or were reported to be docked in protein and ligands (Mobley and Dill, 2009). It has been
an alternative conformation (other binding sites). Moreover, demonstrated through several studies that the free energy
the fungicides having low dissociation constant (K d ) values for calculations and MD simulations were done for refining and
the docked complexes such as Thiophanate methyl (K d value docking the docked complexes, starting from the docked poses,
4.85), Trifloxystrobin (K d value 4.86), Boscalid (K d value 5.27), could be effective in increasing the accuracy of binding affinity
Pyraclostrobin (K d value 6.69), and Isopyrazam (K d value 6.84), predictions (Claussen et al., 2001; Andér et al., 2008). Numerous
despite of having good docking score and lesser K d values, did studies on molecular docking program have demonstrated
not bind in a good docking pose in the MD simulations, or that the computational screening for ranking the affinities
bound with sites in altered conformation (residues that were of ligands binding to receptor proteins may results into a
not involved in binding active sites, or other minor binding higher enrichment of active compounds than random screening
sites). By contrast, fungicides such as Carbendazim (K d value (Stahl and Rarey, 2001; Wyss et al., 2003). However, they
57.09), Cymoxanil (K d value 22.45), Dithane (K d value 976.17), may suffer from sufficient false positive and false negatives,
Famoxadone (K d value 2.43), Fluberidazole (K d value 13.64), and are not sufficiently accurate to rank the compounds
Metiram (Zineb) (K d value 976.17), Pterostilbin (K d value 24.47), according to their binding affinities (Pearlman and Charifson,
Tebuconazole (K d value 39.05), and Oxathiapiprolin (K d value 2001).
1.86) bounded with some of the core residues that constituted, In our results, we found the discrepancies in the ranking of
the major or minor binding sites of receptor protein in a ligand binding affinities from two different popular molecular
good docking pose as revealed through YASARA. The YASARA docking programs (Glide XP and YASARA scores). In the
score, K d values, and MM/GBSA free energy binding values YASARA binding energy function, the energy was calculated
for Metiram (Zineb) and Dithane were found to be similar as the difference between the sum of potential and solvation
−55.63 (±0.38) as both share similar structure, and Dithane energies of the separated compounds, and the sum of potential
is the dimer unit of Metiram. The docking conformation of and solvation energies of the complex in the YAMBER3 force
the Famoxadone and Oxathiapiprolin with the FOXG_04696 field. Thus, more positive YASARA score (difference) means
analyzed, and reported to be good from all the docking higher affinity (Jakubík et al., 2013, 2015). In this context,
servers. Moreover, the MD simulations of Famoxadone and Jakubík et al. (2015) analyzed the performance of four molecular
Oxathiapiprolin protein complexes showed better results with modeling and docking programs (Autodock and Glide for
minimum interaction energies (Oxathiapiprolin followed by docking; AutoDock binding energy function, Glide XP, Prime

Frontiers in Pharmacology | www.frontiersin.org 22 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

MMGB/SA, and YASARA binding function for pose scoring) calculations and continuum (implicit) solvation models. With
in the pose evaluation of re-docked antagonists/inverse agonists this view, the computational calculation for estimating the free
to 11 original crystal structures of the aminergic G protein- binding energies are predicted from the difference between
coupled receptors (GPCRs), and found differences in the ranking the free energy of each ligand bound to the protein and
of ligand binding affinities, from all the four molecular docking the free energies of the components of the complex, i.e.,
programs. In one study, Suenaga et al. (2012) reported that 1Gbinding = 1Gcomplex − (1Gfree receptor + 1Gfree ligand ). The
in the docking process, the top-scored docking pose does not enthalpic contributions for docked complexes are assessed
always correspond to the optimal docking structure. Thus, the through molecular mechanics. MM/GBSA 1Gbind negative value
abilities to determine the optimal docking structure among indicates stronger binding of the ligands with receptor protein.
multiple docking poses generated by the docking process, MM/GBSA (1Gbind ) can be expected to agree reasonably well
as well as to correctly rank the ligands according to their with ranking based on experimental binding affinity. The results
binding affinities, are important for successful computational obtained for binding energies calculations of the protein–ligand
screening. Furthermore, investigation of top posed protein– interactions through MM/GBSA calculation were reported to
ligand interactions revealed substantial differences from actual be highly reproducible and stable (Genheden and Ryde, 2015),
crystallographic structures. In this way, the discrepancies and independent of solvation of the receptor protein, selection
observed in top docked poses and actual crystal structures, of alternative conformation in the starting crystal structure,
or bad ranking of top poses render all current docking and uncertainty in protonation and conformation of various groups
scoring schemes completely inefficient to rank-order drug leads (if employed with care) (Genheden and Ryde, 2015). The
for efficient drug optimization (Warren et al., 2006; Whalen calculations set up by different groups and procedures are likely
et al., 2011; Suenaga et al., 2012; Ramírez and Caballero, to give similar results, in spite of the many more or less arbitrary
2016). choices made during the setup (Genheden and Ryde, 2015).
It has been reported that docking calculations performed Furthermore, MM/GBSA provides more rigorous solutions for
through different servers and tools has several limitations better prediction of reliable and accurate binding positions,
such as a wrong binding site of target receptor protein, the and to estimate the free energies of the bound molecular
choice of docking poses, high docking scores, but failed in complexes (Zhang et al., 2017). This could be attributed due
MD simulations (Chen, 2015). Furthermore, sometimes MD to the fact that MM/GBSA based scoring is physics-based
simulations results revealed docking poses, that were actually term, which contains explicit terms for hydrophobic, V DW ,
unstable, but possess high docking score (Chen, 2015). In this or solvation components. By contrast, other docking and
regard, MD simulations could be deployed for calculating the scoring based programs calculate an empirical scoring function
conformational entropic changes upon receptor–ligand binding. likewise machine based learning procedure, and with having no
This could be derived from time-dependent changes in atomic relevance with other physical parameters. The binding energy
coordinates of the protein and ligand in both bound and was calculated as the difference between the MM/GBSA energy
unbound forms (Du et al., 2016). The stability and reliability of of the complex and the sum of MM/GBSA energies of the
the docked complexes over the simulation time course provides unliganded receptor and the free ligand. It has been found
a good indication for their reliability, accuracy, and stability as that the top docking ranked poses are the lowest ranked poses
it was demonstrated that the unstable and incorrectly docked using MM/GBSA rescoring, that indicates the rescoring of few
structures during MD simulations results into an unstable top poses, if binding could not be determined through docking
trajectories, that finally lead into disruption of the complex. By programs or binding is nonspecific. In many studies it has been
contrast, the realistic complexes provide stable behavior (Yunta, well demonstrated that MM/GBSA approach is most accurate
2016). It has been shown that MD simulations are necessary and reliable for ranking (“scoring”) the efficacy/affinities of a
for some systems to identify the correct binding conformations ligand binding to the receptor proteins in the protein–ligand
(Hou et al., 2011; Sakano et al., 2016). Therefore, MD can docked complexes (Singh and Warshel, 2010; Sun et al., 2014;
additionally be used to estimate the stability of a ligand–receptor Wright et al., 2014; Genheden and Ryde, 2015; Maffucci et al.,
complex proposed by molecular docking (Alonso et al., 2006). 2018). MD simulations analysis therefore, could be employed
However, the more accurate prediction of binding affinity for accurate ranking of ligands following the post docking
can be obtained through free energy calculations, dependent program in terms of their binding affinities (Okimoto et al.,
on thermodynamically important parameter that includes the 2009; Chen, 2015). Recently, MM/GBSA based on short MD
interaction of protein and ligands in complexes, their interaction simulations has been employed for prediction of the accurate
with water and other counter ions in unbounded formed, explicit poses among the generated docking poses (Terayama et al.,
inclusion of the solvent protein dynamics/flexibility (Du et al., 2018).
2016). Overall, one important conclusion from our study revealed
that, docking studies must be harmonized with MD simulations,
Molecular Mechanics and Binding as MD simulations provide core information to complement
Energy Assessment the docking prediction, and unravelled the docking poses
The binding energy calculations for molecular complexes could that were actually unstable. The MD simulations equilibrate
be calculated from MM/GBSA methods, which calculate binding the system to achieve a stable conformation. If the initial
free energies for molecules by combining molecular mechanics structure was energetically unstable, the system appropriately

Frontiers in Pharmacology | www.frontiersin.org 23 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

changes the conformation in subsequent MD simulations The protein–ligand interaction also targeted the functional
(Sakano et al., 2016). Moreover, MD simulations consider the residues that constituted the (active sites) and in a good
natural motion of protein whereas docking usually utilizes a docking pose with least binding energy. Interestingly, the X-ray
single structure obtained by experiment. The binding energy diffracted crystal structures or NMR-derived solution structures,
predictions were highly correlated with a correlation coefficient of protein–ligand complexes, could be used for interaction
r2 = 0.335 for the selected protein–fungicide docked complexes, studies with unknown hypothetical proteins. Moreover, the
and reported Famoxadone and Oxathiapiprolin having better inhibitors discovered through hierarchical in silico screening
binding efficiency with FOXG_04696 than other fungicides. approach (pharmacophore modeling and molecular docking)
The selected fungicide (Famoxadone) was further evaluated could be employed for comparative binding studies of an
for in vitro inhibitory test against the FOL pathogen. It was experimentally derived molecular complex, with unknown
found that the selected fungicide was good enough as the hypothetical protein and novel ligands. The experimental data
mycelial growth was found to be inhibited at every increased available for protein–ligand interaction at good resolutions could
concentration at an increased time interval. The in silco toxicity help in analyzing the other relevant proteins and complexes
assessment tools further predicted the toxicity assessment of the for the better modulation of their functional activity in a more
fungicide and was found to be acceptable for environmental efficacious and reliable manner. The computational screening
disposition, and could be used safely against the FOL pathogen for getting a novel inhibitor (fungicide) followed by in vitro
for controlling the vascular wilt disease. assessment, could be useful to develop commercial formulations
It is well known that fungal SDRs are large family enzymes either alone or in combination with other better fungicides,
and play a crucial role in various metabolic processes, their or used with other integrated approaches, for the better
functional characterization in the FOL pathogen, is an interesting management of the Fusarium wilt disease.
approach. The predictive function of the desired protein could
be useful in understanding the virulence mechanism and
resistance of the FOL pathogen to target fungicides. Moreover, AUTHOR CONTRIBUTIONS
this protein could be better deployed in structure-based
drug design and catalysis. The functional relevance of the MA and VS conceived the idea and planned the experiments. MA
FOXG_04696 (T4HNR like) is not quite understood. In performed all the experiments, did the computational analysis
this context, we could predict that the hypothetical protein of results, and finally prepared and wrote the manuscript. VS
FOXG_04696 might have possible functional role in secondary assisted in the computational analysis of results. MM performed
metabolic process (3-oxoacyl-[acyl-carrier protein] reductase), the in vitro experimental work, and also helped in computational
versicolorin reductase (melanin pigment biosynthesis), or play analysis of the results. MD assisted MM in in vitro experimental
crucial role in the FOL virulence (alcohol dehydrogenase) work, and also helped MA in the computational analysis of
(based on the results of significant hits of Blast-p annotation). the results. SKK, SPK, and AU analyzed the MD simulations
The possible functional relevance of the in silco predicted analysis of the protein–fungicide interactions. MA, SKK, and MD
protein could be deduced and determined experimentally using prepared the final version of manuscript. SS, RU, and AU assisted
mutant analysis and genetic complementation studies. The in manuscript writing, data validation, and supervised the work
data from our study will drive future experimentation for throughout the study. All authors revised and approved it for
determining the predictive function of this protein in the FOL publication.
pathogen.

ACKNOWLEDGMENTS
CONCLUSION
MA is thankful to the Indian Council of Medical Research
The present research work provides an insight into the structural, (ICMR), New Delhi for research facilities in the form
functional, and dynamical aspects of fungal SDR (T4HNR like) ICMR-Junior Research Fellowship and ICMR-SRF. We
in the FOL pathogen. The computational modeling of protein 3D acknowledge the Centre for Bioinformatics, School of
structures, with high accuracy and functional characterization, Biotechnology, and Banaras Hindu University (BHU) for
revealed the core information regarding the homology and providing the Discovery Studio 3.0 tool. We finally acknowledge
conservation of SDRs among the closely related fungal taxonomic to the Head, Department of Botany, Institute of Science, Banaras
groups. The fungal SDRs play a crucial role in various Hindu University, and DST-FIST programme for necessary
metabolic processes including biosynthesis of melanin and facilities during the course of the study.
other pigments, mycotoxin biosynthesis, secondary metabolism,
fungal defense response, and fungal pathogenicity; these enzymes
could be deployed as novel targets, for the discovery of SUPPLEMENTARY MATERIAL
novel agrochemicals against the phytopathogenic fungi. We
reported the interaction of Famoxadone with FOXG_04696 The Supplementary Material for this article can be found
(T4HNR like) with best protein ligand contacts through the online at: https://www.frontiersin.org/articles/10.3389/fphar.
core residues from major binding site of receptor protein. 2018.01038/full#supplementary-material

Frontiers in Pharmacology | www.frontiersin.org 24 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

REFERENCES with acute leukemia. Mycopathologia 182, 879–885. doi: 10.1007/s11046-017-

0156-2
Aamir, M., Singh, V. K., Meena, M., Upadhyay, R. S., Gupta, V. K., and Chiranjeevi, P., Swargam, S., Pradeep, N., Kanipakam, H., Katari, S. K.,
Singh, S. (2017). Structural and functional insights into WRKY3 and WRKY4 Madhulitha, N. R., et al. (2016). Inhibitor design for VacA toxin of Helicobacter
transcription factors to unravel the WRKY–DNA (W-Box) complex interaction pylori. J. Proteomics Bioinform. 9:9. doi: 10.4172/jpb.1000409
in tomato (Solanum lycopersicum L.). A computational approach. Front. Plant Choong, Y. S., Lim, T. S., Chew, A. L., Aziah, I., and Ismail, A. (2011).
Sci. 8:819. doi: 10.3389/fpls.2017.00819 Structural and functional studies ofa 50 kda antigenic protein from Salmonella
Aamir, M., Singh, V. K., Dubey, M. K., Kashyap, S. P., Zehra, A., Upadhyay, R. S., entericaserovar Typhi. J. Mol. Graph. Model. 29, 834–842. doi: 10.1016/j.jmgm.
et al. (2018). Structural and functional dissection of differentially expressed 2011.01.008
tomato WRKY transcripts in host defense response against the vascular wilt Chubb, A. J., Fitzgerald, D. J., Nolan, K. B., and Moman, E. (2006). The productive
pathogen (Fusarium oxysporum f. sp. lycopersici). PLoS One 13:e0193922. conformation of prostaglandin G2 at the peroxidase site of prostaglandin
doi: 10.1371/journal.pone.0193922 endoperoxide H synthase: docking, molecular dynamics and site-directed
Abo Ellil, A. H., and Sharaf, E. F. (2000). “Growth, morphological alteration and mutagenesis studies. Biochemistry 45, 811–820. doi: 10.1021/bi051973k
adaptation of some plant pathogenic fungi to benlate and dicarboximide; a new Cichero, E., D’Ursi, P., Moscatelli, M., Bruno, O., Orro, A., Rotolo, C.,
look,” in Proceedings of the 1st International Conference of Biological Sciences – et al. (2013). Homology modeling, docking studies and molecular dynamic
Faculty of Science, Vol. 1 (Tanta: Tanta University), 568–579. simulationss using graphical processing unit architecture to probe the type-
Alonso, H., Bliznyuk, A. A., and Gready, J. E. (2006). Combining docking and 11 phosphodiesterase catalytic site: a computational approach for the rational
molecular dynamic simulations in drug design. Med. Res. Rev. 26, 531–568. design of selective inhibitors. Chem. Biol. Drug Des. 82, 718–731. doi: 10.1111/
doi: 10.1002/med.20067 cbdd.12193
Altschul, S. F., Madden, T. L., Schäffer, A. A., Zhang, J., Zhang, Z., Miller, W., et al. Claussen, H., Buning, C., Rarey, M., and Lengauer, T. (2001). Flex E: efficient
(1997). Gapped BLAST and PSI-BLAST: a new generation of protein database molecular docking considering protein structure variations. J. Mol. Biol. 308,
search programs. Nucleic Acids Res. 25, 3389–3402. doi: 10.1093/nar/25.17.3389 377–395. doi: 10.1006/jmbi.2001.4551
Amany, H., and Ellil, A. A. (2005). Melanin inhibitors and dicarboximide Colovos, C., and Yeates, T. O. (1993). Verification of protein structures: patterns
interconversion in some phytopathogenic fungi. Egpt. J. Phytopathol. 33, 21–32. of non-bonded atomic interactions. Protein Sci. 2, 1511–1519. doi: 10.1002/pro.
Amini, J., and Sidovich, D. F. (2010). The effects of fungicides on Fusarium 5560020916
oxysporum f. sp. lycopersici associated with Fusarium wilt of tomato. J. Plant Corrales, E. A. R., Rangel, P. R. A., Meza, C. V., Gonzalez, H. G. A., Torres, G. J. C.,
Protect. Res. 50, 172–178. doi: 10.2478/v10045-010-0029-x Roncero, M. I., et al. (2011). Fusarium oxysporum Adh1 has dual fermentative
Anand, Y. R., Begum, S., Dangmei, R., and Nath, P. S. (2013). Evaluation of and oxidative functions and is involved in fungal virulence in tomato plants.
trifloxystrobin 25%+ tebuconazole 50% (Nitro 75MG) against Exserohihum Fungal Genet. Biol. 48, 886–895. doi: 10.1016/j.fgb.2011.06.004
turcicum causing leaf blight disease of maize. J. Crop Weed 9, 198–200. D’Ursi, P., Guariento, S., Trombetti, G., Orro, A., and Cichero, E., Bruno, O., et al.
Andér, M., Luzhkov, V. B., and Aqvist, J. (2008). Ligand binding to the voltage- (2016). Further insights in the binding mode of selective inhibitors to human
gated Kv1.5 potassium channel in the open state—docking and computer PDE4D enzyme combining docking and molecular dynamics. Mol. Inform. 35,
simulations of a homology model. Biophys. J. 94, 820–831. doi: 10.1529/ 369–381. doi: 10.1002/minf.201501033
biophysj.107.112045 Das, S., Lee, D., Sillitoe, I., Dawson, N. L., Lees, J. G., and Orengo, C. A.
Asha, B. B., Nayaka, C. S., Shankar, U. A., Srinivas, C., and Niranjana, S. R. (2011). (2016). Functional classification of CATH superfamilies’: a domain-based
Biological control of F. oxysporum f. sp. lycopersici causing wilt of tomato by approach for protein function annotation. Bioinformatics 32:2889. doi: 10.1093/
Pseudomonas fluorescens. Int. J. Microbial. Res. 3:79. bioinformatics/btw473
Barhate, B. G., Musmade, N. A., and Nikhate, T. A. (2015). Management of de Castro, E., Sigrist, C. J. A., Gattiker, A, Bulliard, V., Langendijk-Genevaux,
Fusarium wilt of tomato by bioagents, fungicides and varietal resistance. Int. P. S., Gasteiger, E., et al. (2006). Scan Prosite: detection of PROSITE signature
J. Plant Prot. 8, 49–52. doi: 10.15740/HAS/IJPP/8.1/49-52 matches and ProRule-associated functional and structural residues in proteins.
Bell, A. A., and Wheeler, M. H. (1986). Biosynthesis and formation of fungal Nucleic Acids Res. 34(Web Server issue), 362–365. doi: 10.1093/nar/gkl124
melanins. Annu. Rev. Phytopathol. 24, 411–451. doi: 10.1146/annurev.py.24. de la Isla, A. L., and Macías-Sánchez, K. L. (2017). Fusarium oxysporum f.
090186.002211 sp. lycopersici: how can we control this fungus? Adv. Biotech. Microbiol..
Berezovsky, I. N., Guarnera, E., Zheng, Z., Eisenhaber, B., and Eisenhaber, F. 4:AIBM.MS.ID.555637. doi: 10.19080/AIBM.2017.04.555637
(2017). Protein function machinery: from basic structural units to modulation DeVay, J. E., Garber, R. H., and Wakeman, R. J. (1988). “Field management
of activity. Curr. Opin. Struct. Biol. 42, 67–74. doi: 10.1016/j.sbi.2016.10.021 of cotton seedling diseases in California using chemical and biological seed
Bergougnoux, V. (2014). The history of tomato: from domestication to treatments,” in Proceedings of Beltwiae Cotton Conference (Memphis, TN:
biopharming. Biotechnol. Adv. 32, 170–189. doi: 10.1016/j.biotechadv.2013. National Cotton Council of Americana), 29–35.
11.003 Du, X., Li, Y., Xia, Y.-L., Ai, S.-M., Liang, J., Sang, P., et al. (2016). Insights into
Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., et al. protein–ligand interactions: mechanisms, models, and methods. Int. J. Mol. Sci.
(2000). The protein data bank. Nucleic Acids Res. 28, 235–242. doi: 10.1093/nar/ 17:144. doi: 10.3390/ijms17020144
28.1.235 Eagen, R., Kim, S. H., Kronstad, J. W., and Breuil, C. (2001). A hydroxynaphthalene
Castrignano, T., De Meo, P. D., Cozzetto, D., Talamo, I. G., and Tramontano, A. reductase gene from a wood staining fungus Ophiostoma floccosum
(2006). The PMDB protein model database. Nucleic Acids Res. 34, D306–D309. complements the buff phenotype in Magnaporthae grisea. Mycol. Res.
doi: 10.1093/nar/gkj105 105, 461–469. doi: 10.3390/ijms17020144
Chen, D. E., Willick, D. L., Ruckel, J. B., and Floriano, W. B. (2015). Principal Eisenberg, D., Luthy, R., and Bowie, J. U. (1997). VERIFY3D: assessment of protein
component analysis of binding energies for single-point mutants of hT2R16 models with three- dimensional profiles. Methods Enzymol. 277, 396–404.
bound to an agonist correlate with experimental mutant cell response. doi: 10.1017/S0953756201003744
J. Comput. Biol. 22, 37–53. doi: 10.1089/cmb.2014.0192 Eliahu, N., Igbaria, A., Rose, M. S., Horwitz, B. A., and Lev, S. (2007). Melanin
Chen, Y. C. (2015). Beware of docking! Trends Pharmacol. Sci. 36, 78–95. biosynthesis in the maize pathogen Cochliobolus heterostrophus depends on
doi: 10.1016/j.tips.2014.12.001 two mitogen-activated protein kinases, Chk1 and Mps1, and the transcription
Cheng, F., Li, W., Zhou, Y., Shen, J., Wu, Z., Liu, G., et al. (2012). admetSAR: factor Cmr1. Eukaryot. Cell 6, 421–429. doi: 10.1016/S0076-6879(97)
a comprehensive source and free tool for assessment of chemical ADMET 77022-8
properties. J. Chem. Inf. Model. 52, 3099–3105. doi: 10.1021/ci300367a Engh, I., Nowrousian, M., and Kück, U. (2007). Regulation of melanin biosynthesis
Chiewchanvit, S., Chongkae, S., Mahanupab, P., Nosanchuk, J. D., Pornsuwan, S., via the dihydroxynaphthalene pathway is dependent on sexual development
Vanittanakom, N., et al. (2017). Melanization of Fusarium keratoplasticum in the ascomycete Sordaria macrospora. FEMS Microbiol. Lett. 275, 62–70.
(F. solani species complex) during disseminated fusariosis in a patient doi: 10.1111/j.1574-6968.2007.00867.x

Frontiers in Pharmacology | www.frontiersin.org 25 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

Errami, M., Geourjon, C., and Deléage, G. (2003). Detection of unrelated proteins Lewia, M., and Penning, T. M. (1997). Comparative anatomy of the aldo-keto
in sequences multiple alignments by using predicted secondary structures. reductase superfamily. Biochemistry 326(Pt 3), 625–636. doi: 10.1042/bj326
Bioinformatics 19, 506–512. doi: 10.1111/j.1574-6968.2007.00867.x 0625
Feng, B., Wang, X., Hauser, M., Kaufmann, S., Jentsch, S., Haase, G., Jin, J., Xie, X., Chen, C., Park, J. G., Stark, C., James, D. A., et al. (2009). Eukaryotic
et al. (2001). Molecular cloning and characterization of WdPKS1, a gene protein domains as functional units of cellular evolution. Sci. Signal. 2:ra76.
involved in dihydroxynaphthalene melanin biosynthesis and virulence in doi: 10.1126/scisignal.2000546
Wangiella (Exophiala) dermatitidis. Infect. Immun. 69, 1781–1794. doi: 10. Jones, P., Binns, D., Chang, H. Y., Fraser, M., Li, W., McAnulla, C., et al. (2014).
1093/bioinformatics/btg016 InterProScan 5: genome-scale protein function classification. Bioinformatics 30,
Filling, C., Berndt, K. D., Benach, J., Knapp, S., Prozorovski, T., Nordling, E., 1236–1240. doi: 10.1093/bioinformatics/btu031
et al. (2002). Critical residues for structure and catalysis in short-chain Jörnvall, H., Persson, M., Krook, M., Atrian, S., Gonzalez-Duarte, R., Jeffrey, J.,
dehydrogenases/reductases. J. Biol. Chem. 277, 25677–25684. doi: 10.1128/IAI. et al. (1995). Short-chain dehydrogenases/reductases (SDR). Biochemistry 34,
69.3.1781-1794.2001 6003–6013. doi: 10.1021/bi00018a001
Foster, A. J. (2018). “Identification of fungicide targets in pathogenic fungi,” in Katari, S. K., Nataranjan, P., Swargam, S., Kanipakam, H., Pasala, C.,
Physiology and Genetics, eds D. Bolton, Melvin and B. P. H. J. Thomma (Cham: and Umamaheswari, A. (2016). Inhibitor design against JNK1 through
Springer), 277–296. doi: 10.1074/jbc.M202160200 e-pharmacophore modeling docking and molecular dynamics simulations.
Frandsen, R. J. N., Rasmussen, S. A., Knudsen, P. B., Uhlig, S., Petersen, D., J. Recept. Signal Transduct. 36, 558–571. doi: 10.3109/10799893.2016.1141955
Lysøe, E., et al. (2016). Black perithecial pigmentation in Fusarium species is due Kavanagh, K. L., Jornvall, H., Persson, B., and Oppermann, U. (2008). Medium-
to the accumulation of 5-deoxybostrycoidin-based melanin. Sci. Rep. 6:26206. and short-chain dehydrogenase/reductase gene and protein families: the SDR
doi: 10.1038/srep26206 superfamily: functional and structural diversity within a family of metabolic and
Frederick, B. A., Caesar-Ton That, T. C., Wheeler, M., Sheehan, K. B, regulatory enzymes. Cell Mol. Life Sci. 65, 3895–3906. doi: 10.1007/s00018-008-
Edens, W. A, and Henson J. M. (1999) Isolation and characterization of 8588-y
Gaeumannomyces graminis melanin mutants. Mycol Res. 103, 99–110. doi: 10. Khan, K. Z., Lal, A. A., and Simon, S. (2014). Integrated strategies in the
1017/S0953756298006959 management of tomato wilt disease caused by Fusarium oxysporum f. sp.
Friesner, R. A., Murphy, R. B., Repasky, M. P., Frye, L. L., Greenwood, J. R., lycopersici. Bioscan 9, 1305–1308.
Halgren, T. A., et al. (2006). Extra precision Glide: docking and scoring Kheder, A. A., Akagi, Y., and Akamatsu, H. (2012). Functional analysis of the
incorporating a model of hydrophobic enclosure for protein- ligand complexes. melanin biosynthesis genes ALM1 and BRM2-1 in the tomato pathotype of
J. Med Chem. 49, 6177–6196. doi: 10.1021/jm051256o Alternaria alternata. J. Gen. Plant. Pathol. 78, 30–38. doi: 10.1007/s10327-011-
Genheden, S., and Ryde, U. (2015). The MM/PBSA and MM/GBSA methods 0356-4
to estimate ligand-binding affinities. Exp. Opin. Drug Discov. 10, 449–461. Kimura, N., and Tsuge, T. (1993). Gene cluster involved in melanin biosynthesis
doi: 10.1517/17460441.2015.1032936 of the filamentous fungus Alternaria alternata. J. Bacteriol. 175, 4427–4435.
Gupta, S., and Rashotte, A. M. (2014). Expression patterns and regulation of doi: 10.1128/jb.175.14.4427-4435.1993
SlCRF3 and SlCRF5 response to cytokinin and abiotic stresses in tomato Krieger, E., and Vriend, G. (2014). View—molecular graphics for all devices—
(Solanum lycopersicum). J. Plant Physiol. 171, 349–358. doi: 10.1016/j.jplph. from smartphones to workstations. Bioinformatics 30, 2981–2982. doi: 10.1093/
2013.09.003 bioinformatics/btu426
Hall, T. A. (1999). BioEdit: a user-friendly biological sequence alignment editor Kroken, S., Glass, N. L., Taylor, J. W., Yoder, O. C., and Turgeon, B. G. (2003).
and analysis program for windows 95/98/NT. Nucleic Acids Symp. Ser. 41, Phylogenomic analysis of type I polyketide synthase genes in pathogenic and
95–98. saprobic ascomycetes. Proc. Natl. Acad. Sci. U.S.A. 100, 15670–15675. doi: 10.
Hansen, F. T., Gardiner, D. M., Lysøe, E., Fuertes, P. R., Tudzynski, B., Wiemann, 1073/pnas.2532165100
P, Sondergaard, T. E., et al. (2015). An update to polyketide synthase and non- Krzywinski, M., Schein, J., Birol, I., Connors, J., Gascoyne, R., Horsman, D., et al.
ribosomal synthetase genes and nomenclature in Fusarium. Fungal Genet. Biol. (2009). Circos: an information aesthetic for comparative genomics. Genome
75, 20–29. doi: 10.1016/j.fgb.2014.12.004 Res. 19, 1639–1645. doi: 10.1101/gr.092759.109
Hopwood, D. A., and Sherman, D. H. (1990). Molecular genetics of polyketides Kubo, Y., and Furusawa, I. (1991). “Melanin biosynthesis: prerequisite for
and its comparison to fatty acid biosynthesis. Annu. Rev. Genet. 24, 37–66. successful invasion of the plant host by appressoria of Colletotrichum and
doi: 10.1146/annurev.ge.24.120190.000345 Pyricularia,” in The Fungal Spore and Disease Initiation in Plants and Animals,
Hou, T., Wang, J., Li, Y. Y., et al. (2011). Assessing the performance of eds G. T. Cole and H. C. Hoch (New York, NY: Plenum Publishing), 205–218.
the molecular mechanics/Poisson Boltzmann surface area and molecular doi: 10.1007/978-1-4899-2635-7_9
mechanics/generalized Born surface area methods, II. The accuracy of ranking Kumar, A., Sharma, A., Kaur, G., Makkar, P., and Jagdeep, K. (2017). Functional
poses generated from docking. J. Comp. Chem. 32, 866–877. doi: 10.1002/jcc. characterization of hypothetical proteins of Mycobacterium tuberculosis with
21666 possible esterase/lipase signature: a cumulative in silico and in vitro approach.
Huang, B. (2009). MetaPocket: a meta approach to improve protein ligand binding J. Biomol. Struct. Dyn. 35, 1226–1243, doi: 10.1080/07391102.2016.1174738
site prediction. OMICS 13, 325–330. doi: 10.1089/omi.2009.0045 Kurahashi, Y. (2001). Melanin biosynthesis inhibitors (MBIs) for control of rice
Illergård, K., Ardell, D. H., and Elofsson, A. (2009). Structure is three to ten times blast. Inst. Phys. Chem. Res. 12, 32–35. doi: 10.1039/b100806o
more conserved than sequence: a study of structural response in protein cores. Kwon, M., Kim, K. S., and Lee, Y. H. (2010). A short-chain
Proteins 77, 499–508. doi: 10.1002/prot.22458 dehydrogenase/reductase gene is required for infection-related development
Inoue, I., Namiki, F., and Tsuge, T. (2002). Plant colonization by the vascular wilt and pathogenicity in Magnaporthe oryzae. Plant Pathol. J. 26, 8–16.
fungus Fusarium oxysporum requires FOW1, a gene encoding a mitochondrial doi: 10.5423/PPJ.2010.26.1.008
protein. Plant Cell 14, 1869–1883. doi: 10.1105/tpc.002576 La Torre, A., Caradonia, L., Matere, F., and Battaglia, V. (2016). Using plant
Itzhaki, Z., Akiva, E., Altuvia, Y., and Margalit, H. (2006). Evolutionary essential oils to control Fusarium wilt in tomato plants. Eur. J. Plant Pathol.
conservation of domain-domain interactions. Genome Biol. 7:R125. doi: 10. 144, 487–496. doi: 10.1007/s10658-015-0789-2
1186/gb-2006-7-12-r125 Lagorce, D., Bouslama, L., Becot, J., Miteva, M. A., and Villoutreix, B. O. (2017).
Jayaram, B., Singh, T., Mukherjee, G., Mathur, A., Shekhar, S., and Shekhar, V. FAF-drugs4: free ADME-tox filtering computations for chemical biology
(2013). Sanjeevini: a freely accessible web-server for target directed lead and early stages drug discovery. Bioinformatics 33, 3658–3660. doi: 10.1093/
molecule discovery. BMC Bioinformartics 13:S7. doi: 10.1186/1471-2105-13- bioinformatics/btx491
S17-S7 Laskowski, R. A., Chistyakov, V. V., and Thornton, J. M. (2005). PDBSum more:
Jakubík, J., Randáková, A., and Doležal, V. (2013). On homology modeling of the new summaries and analysis of the known 3D structure of proteins and nucleic
M2 muscarinic acetylcholine receptor subtype. J. Comput. Aided Mol. Des. 27, acids. Nucleic Acids Res. 33, 266–268. doi: 10.1093/nar/gki001
525–538. doi: 10.1007/s10822-013-9660-8 Liang, D., Chen, Q., Guo, Y., Zhang, T., and Guo, W. (2017). Insight into
Jakubík, J., El-Fakahany, E. E., and Dolezal, V. (2015). Towards predictive docking resistance mechanisms of AZD4547 and E3810 to FGFR1 gatekeeper mutation
at aminergic G-protein coupled receptors. J. Mol Model. 21:284. doi: 10.1007/ via theoretical study. Drug Des. Dev. Ther. 11, 451–461. doi: 10.2147/DDDT.
s00894-015-2824-9 S129991

Frontiers in Pharmacology | www.frontiersin.org 26 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

Liao, D. I., Thompson, J. E., Fahnestock, S., Valent, B., and Jordan, D. B. involved in the regulation of both DOPA and DHN types of pigments
(2001). Structures of trihydroxynaphthalene reductase-fungicide complexes: in submerged culture? Microorganisms 5:22. doi: 10.3390/microorganism
implications for structure-based design and catalysis. Biochemistry 40:8696. s5020022
doi: 10.1016/S0969-2126(00)00548-7 Pandey, K. K., and Gupta, R. C. (2014). Pathogenic and cultural variability among
Lipinski, C. A. (2004). Lead- and drug-like compounds: the rule-of-five revolution. Indian isolates of Fusarium oxysporum f. sp. lycopersici causing wilt in tomato.
Drug Discov. Today Technol. 1, 337–341. doi: 10.1016/j.ddtec.2004.11.007 Ind. Phytopathol. 67, 383–387.
Lovell, S. C., Davis, I. W., Arendall, W. B., de Bakker, P. I., Word, J. M., Prisant, Patón, L. G., Marrero, M. D. R., and Llamas, D. P. (2017). In vitro and field efficacy
M. G., et al. (2003). Structure validation by Ca geometry: phi-psi and C-beta of three fungicides against Fusarium bulb rots of garlic. Eur. J. Plant Pathol. 148,
deviation. Proteins 50, 437–450. doi: 10.1002/prot.10286 321–328. doi: 10.1007/s10658-016-1091-7
Ludwig, N., Lohrer, M., Hempel, M., Mathea, S., Schliebner, I., Menzel, M., et al. Park, D. K., Son, S.-H., Kim, S., Lee, W. M., Lee, H. J., Choi, H. S., et al.
(2014). Melanin is not required for turgor generation but enhances cell-wall (2013). Selection of melon genotypes with resistance to Fusarium wilt and
rigidity in appressoria of the corn pathogen Colletotrichum graminicola. Mol. Monosporascus root rot for rootstocks. Plant Breed. Biotechnol. 1, 277–282.
Plant Microbe Interact. 27, 315–327. doi: 10.1094/MPMI-09-13-0267-R doi: 10.9787/PBB.2013.1.3.277
Lyne, P. D., Lamb, M. L., and Saeh, J. C. (2006). Accurate prediction of the relative Prava, J., Pranavathiyani, G., and Pan, A. (2018). Functional assignment for
potencies of members of a series of kinase inhibitors using molecular docking essential hypothetical proteins of Staphylococcus aureus N315. Int. J. Biol.
and MM/GBSA scoring. J. Med. Chem. 49, 4805–4808. doi: 10.1021/jm060522a Macromol. 108, 765–774. doi: 10.1016/j.ijbiomac.2017.10.169
Mace, M. E., Beckman, C. H., and Mace, M. (2012). Fungal Wilt Diseases of Plants. Prihatna, C., Barbetti, M. J., and Barker, S. J. (2018). A novel tomato fusarium wilt
New York, NY: Academic Press. tolerance gene. Front. Microbiol. 9:1226. doi: 10.3389/fmicb.2018.01226
Madhulitha, N. R., Pradeep, N., Sandeep, S., Hema, K., and Chiranjeevi, P. (2017). Pearlman, D. A., and Charifson, P. S. (2001). Are free energy calculations useful
E-pharmacophore model assisted discovery of novel antagonists of nNOS. in practice? A comparison with rapid scoring functions for the p38 MAP
Biochem. Anal. Biochem. 6:307. doi: 10.4172/2161-1009.1000307 kinase protein system. J. Med. Chem. 44, 3417–3423. doi: 10.1021/jm010
Maffucci, I., Hu, X., Fumagalli, V., and Contini, A. (2018). An efficient 0279
implementation of the Nwat-MMGBSA method to rescore docking results in Rachman, M. M., Barril, X., and Hubbard, R. E. (2018). Predicting how drug
medium-throughput virtual screenings. Front. Chem. 6:43. doi: 10.3389/fchem. molecules bind to their protein targets. Curr. Opin. Pharmacol. 42, 34–39.
2018.00043 doi: 10.1016/j.coph.2018.07.001
Marchler-Bauer, A., Derbyshire, M. K., Gonzales, N. R., Lu, S., Chitsaz, F., Geeret, Ramaiah, A. K., and Garampalli, R. K. H. (2015). In vitro antifungal activity of some
L. Y., et al. (2015). CDD: NCBI’s conserved domain database. Nucleic Acids Res. plant extracts against Fusarium oxysporum f. sp. lycopersici. Asian J. Plant Sci.
43, 222–226. doi: 10.1093/nar/gku1221 Res. 5, 22–27.
Marchler-Bauer, A., Bo, Y., Han, L., He, J., Lanczycki, C. J., Lu, S., et al. Ramírez, D., and Caballero, J. (2016). Is it reliable to use common molecular
(2017). CDD/SPARCLE: functional classification of proteins via subfamily docking methods for comparing the binding affinities of enantiomer pairs for
domain architectures. Nucleic Acids Res. 45, 200–203. doi: 10.1093/nar/gkw their protein target? Int. J. Mol. Sci. 17:525. doi: 10.3390/ijms17040525
1129 Ravooru, N., Ganji, S., Sathyanarayanan, N., and Nagendra, H. G. (2014). Insilico
Marklevitz, J., and Harris, L. K. (2016). Prediction driven functional annotation analysis of hypothetical proteins unveils putative metabolic pathways and
of hypothetical proteins in the major facilitator superfamily of S. aureus essential genes in Leishmania donovani. Front. Genet. 5:291. doi: 10.3389/fgene.
NCTC 8325. Bioinformation 12, 254–262. doi: 10.6026/9732063001 2014.00291
2254 Reimand, J., Hui, S., Jain, S., Law, B., and Bader, G. D. (2012). Domain-mediated
Matsuoka, M., Kumar, A., Muddassar, M., Matsuyama, A., Yoshida, M., and Zhang, protein interaction prediction: from genome to network. FEBS Lett. 586,
K. Y. (2017). Discovery of fungal denitrification inhibitors by targeting copper 2751–2763. doi: 10.1016/j.febslet.2012.04.027
nitrite reductase from Fusarium oxysporum. J. Chem. Inf. Model 57, 203–213. Rongai, D., Pulcini, P., Pesce, B., et al. (2017). Antifungal activity of pomegranate
doi: 10.1021/acs.jcim.6b00649 peel extract against fusarium wilt of tomato. Eur. J. Plant Pathol. 147, 229–238.
Minton, E. B. (1986). “Half a century dynamics and control of cotton disease,” doi: 10.1007/s10658-016-0994-7
in Proceedings of Beltwiae Cotton Conference (Memphis, TN: National Cotton Sakano, T., Mahamood, M. I., Yamashita, T., and Fujitani, H. (2016). Molecular
Council of America), 33–35. dynamics analysis to evaluate docking pose prediction. Biophys. Physicobiol. 13,
Mobley, D. L., and Dill, K. A. (2009). Binding of small-molecule ligands to proteins: 181–194. doi: 10.2142/biophysico.13.0_181
“what you see” is not always what you get. Structure 17, 489–498. doi: 10.1016/ Shen, M., Zhou, S., Li, Y., Pan, P., Zhang, L., and Hou, T. (2013). Discovery and
j.str.2009.02.010 optimization of triazine derivatives as ROCK1 inhibitors: molecular docking,
Mohamed, G. M., and Amer, S. M. (2014). Application of salicylic acid and some molecular dynamics simulations and free energy calculations. Mol. Biosyst. 9,
fungicides as seed treatment for controlling damping-off and root rot diseases 361–374. doi: 10.1039/c2mb25408e
of squash and cantaloupe plants under field conditions. J. Plant Prot. Path. Sigrist, C. J., Cerutti, L., de Castro, E., Langendijk-Genevaux, P. S., Bulliard, V.,
Mansoura Univ. 5, 1025–1043. Bairoch, A., et al. (2010). PROSITE, a protein domain database for
Motoyama, T., and Yamaguchi, I. (2003). “Fungicides, melanin biosynthesis functional characterization and annotation, Nucleic Acids Res. 38, D161–D166.
inhibitors,” in Encyclopedia of Agrochemicals, ed. J. R. Plimmer (Hoboken, NJ: doi: 10.1093/nar/gkp885
John Wiley & Sons, Inc.) Sillitoe, I., Lewis, T. E., Cuff, A., Das, S., Ashford, P., Dawson, N. L., et al.
Pradeep, N., Munikumar, M., Swargam, S., Hema, K., Sudheer Kumar, K., and (2015). CATH: comprehensive structural and functional annotations for
Umamaheswari, A. (2015). 197 Combination of e-pharmacophore modeling, genome sequences. Nucleic Acids Res. 43. D1, D376–D381. doi: 10.1093/nar/
multiple docking strategies and molecular dynamic simulations to discover of gku947
novel antagonists of BACE1. J. Biomol. Struct. Dyn. 33, 129–130. doi: 10.1080/ Silva, P. F. F., Novaes, E., Pereira, M., Soares, C. M. A., Borges, C. L., and Salem-
07391102.2015.1032834 Izacc, S. M. (2015). In silico characterization of hypothetical proteins from
Nirmaladevi, D., Venkataramana, M., Srivastava, R. K., Uppalapati, S. R., Gupta, Paracoccidioides lutzii. Genet. Mol. Res. 14, 17416–17425. doi: 10.4238/2015.
V. K., Yli-Mattila, T., et al. (2016). Molecular phylogeny, pathogenicity and December.21.11
toxigenicity of Fusarium oxysporum f. sp. lycopersici. Sci. Rep. 6:21367. Singh, A., Kaushik., R., Mishra, A., Shanker, A., and Jayaram, B. (2016). ProTSAV: a
doi: 10.1038/srep21367 protein tertiary structure analysis and validation server. Biochim. Biophys. Acta
Okimoto, N., Futatsugi, N., Fuji, H., Suenaga, A., Morimoto, G., Yanai, R., 1864, 11–19. doi: 10.1016/j.bbapap.2015.10.004
et al. (2009). High-performance drug discovery: computational screening by Singh, V. K., Chand, R., and Singh, B. D. (2014). In silico 17β-Hydroxysteroid
combining docking and molecular dynamics simulations. PLoS Comput. Biol. dehydrogenase fungicide for leaf spot disease (Cercospora sp). Online J.
5:e1000528. doi: 10.1371/journal.pcbi.1000528 Bioinform. 15, 198–209.
Palonen, E. K., Raina, S., Brandt, A., Meriluoto, J., Keshavarz, T., and Singha, I. M., Kakoty, Y., Unni, B. G., Kalita, M. C., Das, J., Naglot, A., et al.
Soini, J. T. (2017). Melanisation of Aspergillus terreus-is butyrolactone I (2011). Control of Fusarium wilt of tomato caused by Fusarium oxysporum f.

Frontiers in Pharmacology | www.frontiersin.org 27 October 2018 | Volume 9 | Article 1038

Aamir et al. Fungal SDRs Against Fusarium Wilt

sp. lycopersici using leaf extract of Piper betle L.: a preliminary study. World J. Whalen, K. L., Chang, K. M., and Spies, M. A. (2011). Hybrid steered molecular
Microbiol. Biotechnol. 27, 2583–2589. doi: 10.1007/s11274-011-0730-6 dynamics-docking: an efficient solution to the problem of ranking inhibitor
Singh, N., and Warshel, A. (2010). Absolute binding free energy calculations: on affinities against a flexible drug target. Mol. Inform. 30, 459–471. doi: 10.1002/
the accuracy of computational scoring of protein ligand interactions. Proteins minf.201100014
78, 1705–1723. doi: 10.1002/prot.22687 Wheeler, M. H., Tolmsoff, W. J., Bell, A. A., and Mollenhauer, H. H. (1978).
Spanu, F., Scherm, B., Camboni, I., Balmas, V., Pani, G., Oufensou, S., et al. (2018). Ultrastructural and chemical distinction of melanins formed by Verticillium
FcRav2, a gene with a ROGDI domain involved in Fusarium head blight and dahliae from (+) -scytalone, 1,8-dihydroxynaphthalene, catechol, and L-3,4-
crown rot on durum wheat caused by Fusarium culmorum. Mol. Plant Pathol. dihydroxyphenylalanin. Can. J. Microbiol. 24, 289–297. doi: 10.1139/m78-049
19, 677–688. doi: 10.1111/mpp.12551 Wiederstein, M., and Sipp, M. J. (2007). PROSA-web: interactive web
Supek, F., Bošnjak, M., Škunca, N., and Šmuc, T. (2011). REVIGO summarizes and service for the recognition of errors in three-dimensional structures of
visualizes long lists of gene ontology terms, Gibas C, ed. PLoS One 6:e21800. proteins. Nucleic Acids Res. 35(Suppl. 2), W407–W410. doi: 10.1093/nar/
doi: 10.1371/journal.pone.0021800 gkm290
Sun, H., Li, Y., Shen, M., Tian, S., Xu, L., Pan, P., et al. (2014). Assessing the Willard, L., Ranjan, A., Zhang, H., Monzavi, H., Boyko, R. F., Sykes, B. D., et al.
performance of the MM/PBSA and MM/GBSA methods. 5, improved docking (2003). VADAR: a web server for quantitative evaluation of protein structure
performance using high solute dielectric constant MM/GBSA and MM/PBSA quality. Nucleic Acids Res. 31, 3316–3319. doi: 10.1093/nar/gkg565
rescoring. Phys. Chem. Chem. Phys. 16, 22035–22045. doi: 10.1039/c4cp03179b Wright, D. W., Hall, B. A., Kenway, O. A., Jha, S., and Coveney, P. V.
Suneeta, P., Aiyanathan, K., and Nakkeeran, S. (2016). Efficacy of Bacillus spp. (2014). Computing clinically relevant binding free energies of HIV-1 protease
in the management of collar rot of Gerbera under protected cultivation. Res. inhibitors. J. Chem. Theory Comput. 10, 1228–1241. doi: 10.1021/ct4007037
Crop. 17, 745–752. Wyss, P. C., Gerber, P., Hartman, P. G., Hubschwerlen, C., Locher, H., Marty, H. P.,
Stahl, M., and Rarey, M. (2001). Detailed analysis of scoring functions for virtual et al. (2003). Novel dihydrofolate reductase inhibitors, structure-based versus
screening. J. Med. Chem. 44, 1035–1042. doi: 10.1021/jm0003992 diversity-based library design and high-throughput synthesis and screening.
Suenaga, A., Okimoto, N., Hirano, Y., and Fukui, K. (2012). An efficient J. Med. Chem. 46, 2304–2312. doi: 10.1021/jm020495y
computational method for calculating ligand binding affinities. PLoS One Xiong, D., Wang, Y., Ma, J., Klosterman, S. J., Xiao, S., and Tian, C. (2014).
7:e42846. doi: 10.1371/journal.pone.0042846 Deep mRNA sequencing reveals stage-specific transcriptome alterations
Swarupa, V., Ravishankar, K. V., and Rekha, A. (2014). Plant defense response during microsclerotia development in the smoke tree vascular wilt pathogen,
against Fusarium oxysporum and strategies to develop tolerant genotypes in Verticillium dahliae. BMC Genomics 15:324. doi: 10.1186/1471-2164-15-324
banana. Planta 239, 735–751. doi: 10.1007/s00425-013-2024-8 Yadav, S., Pandey, S. K., Singh, V. K., Goel, Y., Kumar, A., and Singh, S. M.
Szklarczyk, D., Franceschini, A., Wyder, S., Forslund, K., Heller, D., Huerta-Cepas, (2017). Molecular docking studies of 3-bromopyruvate and its derivatives to
J., et al. (2014). STRING v10: protein-protein interaction networks, integrated metabolic regulatory enzymes: implication in designing of novel anticancer
over the tree of life. Nucleic. Acids Res., 43, D447–D452. doi: 10.1093/nar/ therapeutic strategies, Maga G, ed. PLoS One 12:e0176403. doi: 10.1371/journal.
gku1003 pone.0176403
Tamura, K., Stecher, G., Peterson, D., Filipski, A., and Kumar, S. (2013). MEGA6: Yamaguchi, I., and Kubo, Y. (1992). “Target sites of melanin biosynthesis
molecular evolutionary genetics analysis version 6.0. Mol. Biol Evol. 30, inhibitors,” in Target Sites of Fungicide Action, ed. W. Koller (Boca Raton, FL:
2725–2729. doi: 10.1093/molbev/mst197 CRC Press), 101–118.
Tang, N. T. N., and Le, L. (2014). Comparative study on sequence–structure– Yeole, G., Kotkar, H. M., and Mendki, P. S. (2016). Herbal fungicide to control
function relationship of the human short-chain dehydrogenases/reductases Fusarium wilt in tomato plants. Biopestic. Int. 12, 25–35.
protein family. Evol. Bioinform. 10, 165–176. doi: 10.4137/EBO.S17807 Yu, C.-S., Cheng, C.-W., Su, W.-C., Chang, K.-C., Huang, S.-W., Hwang, J.-K.,
Terayama, K., Iwata, H., Araki, M., Okuno, Y., and Tsuda, K. (2018). Machine et al. (2014). CELLO2GO: a web server for protein subCELlular localization
learning accelerates MD-based binding pose prediction between ligands and prediction with functional gene ontology annotation. PLoS One 9:e99368.
proteins. Bioinformatics 34, 770–778. doi: 10.1093/bioinformatics/btx638 doi: 10.1371/journal.pone.0099368
Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994). Clustal-W –improving Yunta, M. J. R. (2016). Docking and ligand binding affinity: uses and pitfalls. Am.
the sensitivity of progressive multiple sequence alignment through sequence J. Model. Optim. 4, 74–114.
weighting, position-specific gap penalties and weight matrix choice. Nucleic Zhang, X., Perez-Sanchez, H., and Lightstone, F. C. (2017). A comprehensive
Acids Res. 22, 4673–4680. doi: 10.1093/nar/22.22.4673 docking and MM/GBSA rescoring study of ligand recognition upon
Umamaheswari, A., Pradhan, D., and Kumar, M. H. (2010). Identification of binding antithrombin. Curr. Top. Med. Chem. 17, 1631–1639. doi: 10.2174/
potential Leptospira phosphor-heptose isomerase inhibitors through virtual 1568026616666161117112604
high throughput screening. Genomics Proteomics Bioinformatics 8, 246–255.
doi: 10.1016/S1672-0229(10)60026-5 Conflict of Interest Statement: The authors declare that the research was
Vidal-Cros, A., Viviani, F., Labesse, G., Boccara, M., and Gaudry, M. (1994). conducted in the absence of any commercial or financial relationships that could
Polyhydroxynaphthalene reductase involved in melanin biosynthesis in be construed as a potential conflict of interest.
Magnaporthe grisea, purification, cDNA cloning and sequencing. Eur. J.
Biochem. 219, 985–992. doi: 10.1111/j.1432-1033.1994.tb18581.x Copyright © 2018 Aamir, Singh, Dubey, Meena, Kashyap, Katari, Upadhyay,
Vogel, C., Bashton, M., Kerrison, N. D., Chothia, C., and Teichmann, S. A. (2004). Umamaheswari and Singh. This is an open-access article distributed under the terms
Structure, function and evolution of multidomain proteins. Curr. Opin. Struct. of the Creative Commons Attribution License (CC BY). The use, distribution or
Biol. 14, 208–216. doi: 10.1016/j.sbi.2004.03.011 reproduction in other forums is permitted, provided the original author(s) and the
Warren, G. L., Andrews, C. W., Capelli, A.-M., Clarke, B., LaLonde, J., Lambert, copyright owner(s) are credited and that the original publication in this journal
M. H., et al. (2006). Critical assessment of docking programs and scoring is cited, in accordance with accepted academic practice. No use, distribution or
functions. J. Med. Chem. 49, 5912–5931. doi: 10.1021/jm050362n reproduction is permitted which does not comply with these terms.

Frontiers in Pharmacology | www.frontiersin.org 28 October 2018 | Volume 9 | Article 1038