Discovery of succinate dehydrogenase candidate fungicides via lead optimization for effective resistance management of Fusarium oxysporum f. sp. capsici

Abstract

Fusarium wilt of chili caused by the fungus Fusarium oxysporum f. sp. capsici (FCO) severely reduces the production of chili worldwide. There is growing evidence of resistance to commercial fungicides targeting succinate dehydrogenase (Sdh) of FCO soliciting the development of new Sdh inhibitors (SdhIs). In the current work, optimized docking and virtual screening were used to mine twelve SdhIs from the ZINC database, followed by in vitro antifungal evaluation on spore and radial mycelium development. Four new promising SdhIs exhibiting a mean mycelium inhibition rate greater than 85.6% (F = 155.8, P = 0.001, P < 0.05) were observed on ten strains of virulent and resistant FCO. Importantly, three of the discovered molecules exhibited potent spore germination inhibition (≥ 80%, P = 0.01, P < 0.05) compared to the commonly used fungicide penthiopyrad. A significant positive correlation (r* ≥ 0.67, P < 0.05) between the activities of the newly discovered SdhIs compared to penthiopyrad against all tested FCO strains indicated a broad-spectrum fungicidal activity. The current findings indicate that the four SdhI’s discovered could judiciously replace certain commercial SdhIs that some FCO displays resistance to.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13205-022-03157-8.

Introduction

Chili pepper (Capsicum annum L.) is considered one of the most important vegetable crops farmed over 19.89 million hectares with an annual output of 33.52 million tons worldwide (Patel et al. 2014). Chili pepper production stands at 2.31 tons/hectare per annum in Pakistan contributing significantly to the economy (Khan et al. 2017; Nawaz et al. 2018a). Over the years, chili pepper production area in Pakistan shrunk by 2.68% from 64,175 ha in 2007 to an estimated 62,456 ha (Khan et al. 2017). Because land infestation by soilborne pathogens is difficult to manage by crop rotation, raised beds, water management, and fungicides application, farmers have been compelled to abandon chili farmland (Nawaz et al. 2018a). Root-rot and collar-rot diseases of chili pepper induced by the oomycete—Phytophthora capsici (Nawaz et al. 2018a; Nawaz et al. 2018b), die-back, and fruit rot caused by Colletotrichum species in chili pepper (Khaleeque et Khan 1991), and Fusarium wilt of chili pepper (Skaggs et al. 2000) are widely reported in Pakistan. Among these groups of fungal diseases, Fusarium wilt of chili pepper incited by the soilborne Fusarium oxysporum (Schlect.) emend. Synd. and Hans. f. sp. capsici (FCO) is one of the most severe plant diseases worldwide.

FCO incites vascular and root wilts, vein clearing in younger leaflets, epinasty, stunting, yellowing of older leaves, and progressively colonized xylem tissues and disrupt translocation of nutrients leading to death (Agrios 2005; Siddiqui and Akhtar 2007). In Pakistan, Fusarium wilt of chili pepper causes an estimated 10–50% of the total yield losses (Siddiqui and Akhtar 2007; Irum 2007; Bashir et al. 2017). In horticultural practices, concurrent use of synthetic fungicides, grafting, and cultural practices variably control FCO and also lead to the emergence of virulent strains that are resistant to conventional fungicides (Akaram et al. 2018). Thus, there is a need to develop strategies to counter the FCO fungicide resistance.

Introducing biocontrol agents such as Trichoderma species to curb fungal diseases of chili pepper and mitigate fungicide resistance in Pakistan has been initiated (Nawaz et al. 2018b). An alternative containment measure for FCO–fungicide resistance management could be the introduction of new or structurally diverse antifungal molecules for the following reasons: (1) to delay the onset of resistance, and (2) to effectively control diseases as compounds with better binding conformation provide enhanced antifungal activity by specifically blocking other amino acid residues on the receptor site of the target protein. Carboxamide fungicides which inhibit succinate dehydrogenase (Sdh; EC 1.3.5.1) enzyme could be used as a target protein to control FCO (Thomas et al. 2014; Iftikhar et al. 2017; Xiong et al. 2016; Wei et al. 2021; Li et al. 2020). The FCO-Sdh is universal and a multi-subunit enzyme at the crossways of tricarboxylic acid cycle (TCA) and electron transport chain (ETC) that catalyzes the oxidation of succinate into fumarate. Succinate dehydrogenase inhibitors (SdhIs) disrupt the ETC by binding to ubiquinone (UQ) site and thus, effectively control fungal diseases of plants. With the prevalence of diverse virulent FCO strains and the emergence of fungicide resistance in the chili farming region of Pakistan, finding an ideal fungicide that can control most of the resistant FCO strains would provide a durable solution in cutting down yield losses. The objective of this study was: (1) to mine candidate unreported SdhIs from the ZINC database, dock, and virtually screen for interaction with F. oxysporum f. sp. capsici Sdh, and (2) to evaluate the in vitro antifungal potential of candidate SdhIs at the level of mycelium and conidia. It was found that candidate SdhIs with ZINC accession numbers ZINC03102709, ZINC17007371, ZINC32180696, and ZINC00781454 were significantly active against resistant F. oxysporum f. sp. capsici compared to commercially fungicide—penthiopyrad.

Materials and methods

Mining Sdh potential inhibitors

A pipeline was developed (Fig. 1) and a preliminary in silico screening was performed using 18 reference fungicides active ingredients viz., benodanil, benzovindiflupyr, bixafen, boscalid, fenfuram, fluopyram, flutolanil, fluxapyroxad, furametpyr, isofetamid, isopyrazam, mepronil, oxycarboxin, penflufen, penthiopyrad, sedaxane, and thifluzamide (Supplementary Fig. S1) from the Fungicide Resistance Action Committee (FRAC group: 7) to interrogate the ZINC database (http://zinc.docking.org/). These active ingredients were screened for homologs in the ZINC database at default mode (Table 1). The 2D structures of generated homologous molecules were constructed and converted to 3D structures on ChemDraw Ultra v12.0 (Li et al. 2004). The 3D structures were energetically minimized using MOE suit v2009.10 (Molecular Operating Environment.2013.08) until the RMSD gradient of 0.05 kcal.mol⁻¹.Å⁻¹ was reached (Fig. S2).

Fig. 1.

A pipeline for in silico discovery of SdhIs targeting UQ site for the management of Fusarium wilt of chili pepper

Table 1.

Twelve compounds selected as novel SdhIs with their ZINC ID, common names, molecular weight, and codes

No.	ZINC ID	Common names	Molecular weight (g mol−1)	Unique code name
1	ZINC00781454	(Z)-2-cyano-3-(4-isoindolin-2-ylphenyl)-N-(2-thenyl)acrylamide	385.492	C1
2	ZINC04498541	N-benzyl-2-cyano-3-(4-pyrrolidin-1-ylphenyl)-prop-2-enamide	331.419	C2
3	ZINC27260697	3-[(2-chlorobenzoyl)amino]-N-(3-cyano-4,5-dimethyl-2-thienyl)benzamide	409.898	C3
4	ZINC02409017	4-cyano-5-[[3-(2-ketochromen-3-yl)benzoyl]amino]-3-methyl-thiophene-2-carboxylic-acid-ethyl-ester	458.495	C4
5	ZINC00689200	2-amino-4-(3-bromophenyl)-1-(2-cyanophenyl)-5-oxo-1,4,5,6,7,8-hexahydro-3-quinolinecarbonitrile	446.328	C5
6	ZINC03102709	[4-bromo-2-[(2-hydroxy-2-phenyl-acetyl)aminoiminomethyl]phenyl]	453.292	C6
7	ZINC05176439	N-[3-[(2-hydroxy-3-methoxy-phenyl)methyleneaminocarbamoyl]phenyl]-2-methyl-benzamide	403.438	C7
8	ZINC06183077	(2Z)-3-(5-chlorothiophen-2-yl)-2-cyano-N-(1-phenylethyl)prop-2-enamide	316.813	C8
9	ZINC06512448	(2Z)-3-(4-tert-butylphenyl)-2-cyano-N-(pyridin-2-ylmethyl)prop-2-enamide	319.408	C9
10	ZINC17007371	N-(2-chloro-3-methylphenyl)-2-cyano-2-cyclopentylideneacetamide	274.751	C10
11	ZINC32180696	(Z)-N-benzyl-2-cyano-3-[4-[(2-fluorophenyl)methoxy]phenyl]prop-2-enamide	386.426	C11
12	ZINC34378628	(2Z,4E)-2-cyano-5-phenyl-N-[(1S)-1-phenylethyl]penta-2,4-dienamide	302.377	C12

Pharmacophore generation and compound selection

Pharmacophore modeling was performed to screen for the output homologous drug-like compounds from the ZINC database as previously described (Irwin et al. 2012). Pharmacophore models were generated by superimposing the 3D structures of the 18 fungicides based on the positions of annotation points (such as aromatic center, H-bond donors and acceptors, and hydrophobic centers) for the homologous fungicides using MOE suit v2009.10. To identify novel Sdh inhibitors, a set of 17,900,742 compounds were retrieved from the ZINC database drug-like category. The drug-like categories were further screened by the pharmacophore model at default mode, retrieving 50,000 compounds which were subsequently subjected to the structure-based virtual screening (SBVS) protocol.

SBVS of 50,000 compounds by auto-dock tools

A library of 50,000 screened drug-like pharmacophore compounds were docked into the ubiquinone binding site of reference Alternaria solani Sdh by AutoDock Tools (ADT) v4.0 (Morris et al. 2010), given that Sdh enzyme for fungi is universal. The docking calculations were performed on Intel-Xeon-QuadTM core processor 3.0 GHz Linux workstation. AutoDock Tools v4.0 was used with empirical free energy function and a Lamarckian genetic algorithm (Morris et al. 2009). The polar hydrogen atoms and Kollman charges were added to the protein model. Partial charges of UQ were assigned with Gasteiger charges. The active site was defined on UQ binding site by AutoGrid. The grid size was 70 Å × 70 Å × 70 Å, grid spacing of 0.375 Å, step size of 1.0 Å for translation, 50 degrees for rotation, and a maximum number of energy evaluations was set at 250,000. A total of ten conformations were generated for each compound. The docked protein–ligand complexes were evaluated based on the interaction energy of each conformation for the compounds.

Re-scoring and lead optimization of compounds

From docking analysis, the top 500 molecules (⁓1%) were re-scored in MOE suit v2009 and a consensus score based on ADT and MOE scores was selected for visual inspection of the geometrical matching of ligands with residues of amino acids at the docking site. Molecules with inconsistent docking conformations were removed after visual inspection. From the pipeline, 25 molecules were generated as potential Sdh inhibitors but 13 out of the 25 molecules were already in the market. Thus, 12 uncommercialized molecules were synthesized by MolPort^® (www.molport.com; Lacplesa iela, Rega, Lativa) and subjected to in vitro testing and bioassays in this study.

Model building of Sdh for Fusarium oxysporum and template search

The UQ site is a hydrophobic cleft composed of the amino acid residues of SdhB, SdhC, and SdhD which are highly conserved among a wide range of organisms (Sun et al. 2005; Huang et al. 2006). So, the Sdh model for F. oxysporum was constructed using different reference subunit sequences for F. oxysporum (Table S1). The SWISS-MODEL template library (SMTL version 2019–03-06, PDB release 2019–03-01) was searched with BLAST (Camacho et al. 2009), and HHBlits (Remmert et al. 2011), for evolutionary related structures matching the target sequence. The target sequence of four subunits was searched with BLAST against the primary amino acid sequence contained in the SMTL. An initial HHblits profile was built using the procedure outlined in Remmert et al. (2011), followed by 1 iteration of HHblits against NR20. The obtained profile was searched against all profiles of SMTL. A total of 32 templates were found.

Refined model building, quality estimation, and docking analysis

Models were built based on the target-template alignment using open ProMod3 (https://openstructure.org/promod3/1.3/). Coordinates that were conserved between the target and the template were copied from the template to the model. Insertions and deletions were re-modeled using a fragment library and side chains were rebuilt. Finally, the geometry of the resulting model was regularized using a force field. In cases where loop modeling with ProMod3 failed, an alternative model was built with PROMOD-II (Guex et al. 1997). To evaluate the internal reliability and consistency, the stereo-chemical properties of models based on energy minimization, refinement, a simulation program called PROCHECK v3.4.4 (Laskowski et al. 1993) was used. Furthermore, Ramachandran plot was used to assess whether the values of the dihedral Psi/Phi/Chi/Omega angels agree with the values of allowed conformation for protein backbones.

The drug discovery tool 1-Click Docking (Mcule Inc, https://mcule.com/apps/1-click-docking/) was used to dock the discovered compounds within the UQ site of F. oxysporum f. sp. capsici Sdh at the default setting. 1-Click docking used a Vina filter (Trott and Olsen 2010) and docked the ligand into a single target specifying binding sites. Gasteiger charges and hydrogens were added in the input files; non-polar hydrogens, lone-pairs, and non-standard residues were removed. Molecular graphics and analyses were performed with the UCSF Chimera package (Pettersen et al. 2004).

Succinate dehydrogenase activity of F. oxysporum f. sp. capsici strains

To confirm that all fungal strains were actively producing succinate dehydrogenase (Sdh) and that the new molecules were specifically targeting the Sdh, a colorimetric assay was performed. All F. oxysporum f. sp. capsici strains used in this study were isolated from chili roots grown within the chili hotspot of Lahore, Punjab province, Pakistan between the 2017–2019 farming season. Based on greenhouse preliminary screening, a single spore culture isolates that caused 50% wilting of 3 weeks-old chili seedlings within 7 days were referred to as virulent and used herein. All virulent FCO isolates were further pre-screening in vitro at the discriminatory dose of penthiopyrad (PEN, 3 µg/mL), and mycelium of isolates that were inhibited over 79–100% (were considered sensitive and not used). On the other hand, virulent FCO isolates whose mycelium were inhibited below 79% were considered resistant (R) namely FCO1 (74.33% resistant (R)), FCO2 (75.3% R), FCO3 (75.6% R), FCO4 (74.6% R)), FCO5 (71.6% R), FCO6 (74.6% R), FCO7 (77.6% R), FCO8 (73.3% R), FCO9 (76.3% R), and FCO10 (74.6% R) and used herein (Supplementary Table S2). Isolate FCO1 was the most virulent. The abovementioned FCO strains were grown for 5 days on Potato Dextrose Agar (PDA; Difco) in a 9 cm diameter Petri plates at 25 °C with a 16 h photoperiod.

Succinate dehydrogenase (Sdh) activity was assessed with BioVision^® colorimetric assay Kit 1/14 (BioVision Inc, Milpitas, CA, USA). Briefly, fungal mycelia (10 mg) were rapidly homogenized with 100 μL ice-cold Sdh assay buffer, kept on ice for 10 min, and centrifuged at 10,000 g for 5 min. A 15 μL supernatant of the homogenate and 5 μL (10 μg/mL) of each synthesized compound mined from the in silico pipeline (Fig. 1) were added. The volume of the reaction mixture was adjusted to 50 μL with Sdh assay buffer and transferred to a 96-well plate. For the positive control, 15 μL of Sdh positive control was taken into desired well(s) and the final volume was adjusted to 50 μL with Sdh assay buffer. 0, 4, 8, 12, 16, and 20 μL of 2 mM Dichlorophenolindophenol (DCIP) standard solution were added into a series of wells to generate 0, 8, 16, 24, 32, and 40 nM/well of DCIP standard. The volume was adjusted to 100 μL per well with Sdh assay buffer. The reaction mixture of 50 μL was prepared to contain 46 μL SDH assay buffer, 2 μL Sdh substrate mix, and 2 μL Sdh probe. The reaction mixture of 50 μL was added to each well containing the samples and positive control and mixed well. The absorbance was measured spectrophotometrically immediately at 599 nm in kinetic mode for 2 h at 25 ºC. Three replicates for each treatment were used and the experiment was conducted three times. The standard reading (ΔOD) was obtained by subtracting blank from sample reading and extrapolating on DCIP standard curve to get “B” at two-time frames (ΔT = T₀ − T₃₀) separated by 30 min.

The activity of mined compounds (AS) was measured as follows:

$$\text{AS}=\frac{\left[B\right]\times \left[\text{Dilution factor}\right]}{\left[\left(\mathrm{\Delta }T\times V\right)\right]},$$

where B = Amount of reduced DCIP from standard curve (nmol), ΔT = Reaction time (min), V = Sample volume added into the reaction well (μL), and D = Dilution factor. The activity was expressed as nanomoles of DCIP reduced in 1 min per mg protein.

In vitro conidial germination assay for mined compounds on F. oxysporum f. sp. capsici strains

The potential of the compounds to inhibit conidia germination was evaluated as previously described in Pasche et al. (2004), by comparing the conidial germination on water agar plates amended with or without the compounds. The spore suspension was prepared by flooding the plates with 5 mL water and the mycelial mat was gently rubbed to dislodge the conidia. The conidial suspension was then filtered through a nylon mesh (43 mm pore diameter) and adjusted to 1 × 10⁵ conidia/mL using a hemocytometer. The compounds and pure technical grade penthiopyrad (positive control) were used to amend the agar plates at a concentration of 3 μg/mL. The control was amended with DMSO (1% v/v) with final concentration of 0.1% (v/v). The aliquots (50 μL) of the conidial suspensions of the F. oxysporum (1 × 10⁵ conidia/mL) were spread on fungicide-amended and fungicide-free Petri dishes (9 cm) with a glass rod.

The plates were incubated at 25 ± 1 ºC under continuous light for 4 h and 50 conidia/plate were examined for the development of germ tube. The conidium was considered germinated either when the germ tube was equal (or greater) in length as conidium diameter or if multiple germ tubes were emerging from one conidium. Three replicates were used for each treatment and the experiment was repeated three times. Inhibition percentage (PI) was calculated using the formula by Leroux and Berthier (Leroux and Berthier 1988), as follows:

$$PI=\frac{100\left(C-T\right)}{C},$$

where C is the average conidia germinated in the control plate and T is the average conidia germinated in the treated plate.

In vitro mycelial growth assay against F. oxysporum f. sp. capsici strains

The food-poison technique was used (Pasche et al. 2004). Briefly, autoclaved PDA at 55 °C was amended with the compounds at 3 μg/mL and distributed in 9 cm diameter Petri dishes. The mycelial plugs (5 mm diameter) were cut from the margin of actively growing 5-day-old culture on PDA and placed upside down in the center of fungicide-amended and non-amended PDA plates. The DMSO (0.1% v/v) amended plates served as negative control and the commercial fungicide penthiopyrad served as the positive control. The plates were incubated at 25 ± 1 ºC with 16 h of photoperiod for 7 days. Each treatment was triplicated, and the experiment was repeated three times. The diameter of each colony was measured by taking two perpendicular diameters from edge-to-edge of each colony after 7 days. Percentage inhibition (PI) of mycelial growth was computed as follows.

$$PI=\frac{100\left(C-T\right)}{C},$$

where C is the mean diameter of the fungal colony (in cm) in the control plate and T is the mean diameter of the fungal colony (in cm) in the treated plate.

Statistical analysis

All data were pooled and subjected to one-way analysis at P < 0.05 in open-source R STUDIO v3.5.2. Correlation analysis based on performance analytics at P < 0.05 and principal component analysis (PCA, at Jolliffe cut-off value ≤ 0.7) were performed to determine the possibility of synergism or antagonism vis-à-viz all the fungal strains.

Results

Structural overview of the homology model

The developed pipeline (Fig. 1) successfully generated 25 potential cohort of ShdIs. A total of 12 out of the 25 molecules were not yet found in the commercial domain but reported herein (Table 1). Template and target sequence similarity was as follows SdhA (86.8% identity/96.8% similar), SdhB (80.2% identity/94.8% similar), SdhC (27.8% identity/61.1% similar), and SdhD (59.4% identity/84.2% similar). The SDH template of Sus_scrofa (PDB cone: 1ZOY) with the highest queried amino acid sequence identity was chosen as the modeling template. The homology model of Sdh for F. oxysporum (FCO) showed two hydrophilic proteins and two trans-membrane proteins. The two hydrophilic proteins included flavin adenine dinucleotide (FAD) binding protein (Fp of subunit A or FCO–SdhA) and iron–sulfur protein (FCO-SdhB). The two trans-membrane proteins included large cytochrome b (CybL or subunit C) and small cytochrome b (CybS or subunit D). The model contained a total of six trans-membrane helices (Fig. 2). In FCO–Sdh model, the soluble catalytic heterodimer was composed of subunit A (FCO–SdhA) in which FAD cofactor was added by superimposition onto the template of Sdh crystal structure of Sus_scrofa (PDB ID: 1ZOY). The subunit B (FCO–SdhB) contained three iron–sulfur clusters: [2Fe-2S], [4Fe-4S], and [3Fe-4S]. The 3D molecular structure of FCO-Sdh also contained one heme molecule. The overall 3D structure was “q” letter shaped comprising of a hydrophilic head and a hydrophobic multi-pass trans-membrane-anchor tail. There was no direct interaction between FCO-SdhA and the trans-membrane hydrophobic anchors. The contact surfaces between the four subunits were dominated by the hydrophobic interactions.

Fig. 2.

The docked orientation of C1, C2, and C10 SdhIs. The interacting residues and the compounds are presented in magenta and yellow stick colours, respectively. H-bonds are displayed in black dotted lines. A C1 (ZINC00781454), (B) C2 (ZINC04498541), (C) C10 (ZINC17007371)

Model evaluation and validation of discovered molecule docking sites

The generated model of FCO-Sdh presented the specific signature of Sdh and exhibited structural features comparable to the template (PDB code: 1ZOY). The superimposed view of the template and FCO-Sdh model are shown (Fig. 2). The comparison of FCO-Sdh and template (PDB code: 1ZOY) showed consistency in the built model and good overlap. The RMSD values between chain A, B, C, and D of FCO-Sdh model vis-à-vis the template (1ZOY) were 0.563 Å, 0.108 Å, 1.044 Å, and 1.119 Å, respectively.

The quality and stereo-chemical properties of the built FCO–Sdh model and the subunits were scrutinized by PROCHECK Ramachandran plot. Ramachandran plot analysis of subunit FCO-SdhA showed 89.7% of amino acid residues within the most favored region, whereas one residue was found in the disallowed region (Supplementary Fig. S3; Table S3). The PROCHECK results of subunit FCO-SdhB indicated that 86.9% of the residues have phi- and psi- angles falling in the most favored regions (Supplementary Fig. S3; Table S3). For subunit FCO-SdhC, Ramachandran statistics showed that 92.6% of the residues were in the most favored region, while no residue was found in the disallowed region (Supplementary Fig. S3; Table S3). In the case of subunit FCO-SdhD, only one residue was in the disallowed region with 88.2% residues in the most favored region.

Based on molecular docking performed in 1-Click Docking tool (Mcule Inc, https://mcule.com/apps/1-click-docking/) and Autodock Vina Configuration (Trott and Olson 2010), the best fit posed with the lowest binding energy from initial molecular docking (Fig. 2). The compound C1 formed H-bond with Phe83 of chain C at 3.13 Å and the binding energy was -10.75 kcal/M. The binding energy for compound C2 was − 11.02 kcal/M and it formed an H-bond with Phe83 of chain C with bond length 2.89 Å. Compound C10 had a binding energy of − 10.86 kcal/mol and H-bonded with Trp82 at 3.07 Å. The interaction analysis revealed that Phe83 and Trp82 were critical amino acid residues for ligand binding, while Leu73.C, Trp205.B, Trp204.B, Pro201.B, Tyr76.C, Trp82.C, Ser84.C, Trp88.C, and Ile91.C provided hydrophobic interactions to the ligands (Fig. 2A, B, C). Amino acid residues forming strong hydrogen bond networks were found in subunit C (Table 2).

Table 2.

Interaction details of the most active compounds and residues that initiated strong hydrogen bonding network

Sdhi	Sdh subunits	Interacting amino acids	Residues forming strong hydrogen bond network
C1	C	Phe83	Phe83
	C	Ile91
	C	Leu73
	C	Trp82
	C	Trp88
	C	Ala87
	D	Try134
C2	C	Phe83	Phe83
	C	Trp69
	C	Trp82
	C	Tyr76
	B	Trp205
	C	Trp82	Trp82
C10	C	Phe83
	C	Ser84
	B	Pro201
	C	Leu73
	C	Trp82
	C	Ile91

Discovered SdhIs differentially inhibit succinate dehydrogenase enzyme

It was found that all the virulent strains of F. oxysporum f. sp. capsici produced succinate dehydrogenase with activity ranging from 3.0 to 3.3 U/mL in the presence of DMSO (Fig. 3). This signified that all the resistant virulent FCO strains used in this study constitutively expressed Sdh for metabolism. All the discovered compounds differentially decreased the activity of FCO-Sdh (Fig. 3A). A significant decrease in Sdh activity was observed for the compound C1 (1.77 U/mL; F = 113.1, P < 0.05, P = 0.00), C2 (1.75 U/mL; F = 113.1, P < 0.05, P = 0.00), C6 (1.86 U/mL; F = 113.1, P < 0.05, P = 0.00), and C10 (1.83 U/mL; F = 113.1, P < 0.05, P = 0.00) and relative to penthiopyrad (2.57 U/mL; F = 113.1, P < 0.05). The commercial fungicide penthiopyrad showed Sdh activity ranging from 2.26 to 3.03 U/mL against all the FCO isolates. For the other tested compounds, Sdh activity varied from 1.3 to 2.96 U/mL for all the isolates. To confirm the uniqueness of the FCO strains used in this study, principal component analysis (PCA) based on their Sdh expression pattern generated two groups (Fig. 3B). Group one was composed of FCO2, FCO5, FCO8, FCO10 strains, and the principal component was 7.79%. Also, group two was composed of FCO1, FCO3, FCO4, FCO6, and FCO7 strains and the principal component was 69.4%. The diverse nature of FCO strains enriched the usefulness of the discovered compounds in the effective management of FCO strains and relatives. PCA analysis revealed that the discovered SdhIs differentially inhibited succinate dehydrogenase enzyme of FCO strains (Fig. 3C).

Fig. 3.

Specific inhibitory activity of tested compounds on Sdh activity. A Mean decrease in Sdh activity (U/mL), and asterisks indicate levels of significant differences (*P < 0.05; **P < 0.001, ***P < 0.0001). B Principal component analysis (PCA) biplot for the comparative distribution of studied F. oxysporum f. sp. capsici strains in function of their Sdh activity to the tested compounds. C A PCA spanning–tree showing the activity map of the tested compounds relative to Sdh activity in each FCO strains revealed the compounds have different bioefficacy

Discovered SdhIs effectively inhibits germ tube germination

The compounds were evaluated against the FCO strains at the spore level by in vitro conidial germination assay. The antifungal activity of the twelve tested compounds and reference control penthiopyrad (PEN) showed divergent important antifungal potential. Using R package PerformanceAnalystics, a correlation analysis was performed to determine the global performance of all the tested compounds to the various FCO strains with different degree of sensitivity (Fig. 4). Compound C1, C2, C6, C9, and C10 exhibited mean significant germ tube inhibition against all the strains of FCO. The inhibitory pattern of compounds C1 vs C2, C9 vs C10, C10 vs C12, and C9 vs C12 on all the FCO strains indicated a significant positive correlation (*r = 0.67, P < 0.05), (*r = 0.65, P < 0.05), (**r = 0.77, P < 0.05), (*r = 0.71, P < 0.05), respectively. The significance of these findings is that the discovered compounds with positive correlation could be used for formulation to generate a stronger synergistic effect to target spore germination whereas significant negative correlation could generate antagonistic effects, thus, low antifungal potential. The inhibitory pattern of compound C2 vs C6 and C3 vs C6 on all the FCO strains indicated positive correlation (r = 0.51, P< 0.05) and (r = 0.62, P < 0.05), respectively.

Fig. 4.

Pearson correlation coefficient (r) for germ tube inhibition by discovered compounds relative to commercial fungicide penthiopyrad on FCO strains. The asterisks indicate levels of significant differences (*r < 0.05; **r < 0.001, ***r < 0.0001)

Discovered SdhIs effectively inhibit FCO mycelial growth in vitro

To confirm whether the molecule’s inhibitory pattern at the mycelial level was akin to that observed at the germ tubes, a PCA analysis was performed. Two strong principal components generated were 87.9% and 3.2% (Fig. 5A). The FCO strains clustered into two groups vis-à-vis the active compounds as observed with germ tube assay (Fig. 5A). Cluster I consisted of FCO4, FCO5, FCO6, FCO8, and FCO10, and cluster II consisted of FCO9 and FCO3, and a singleton FCO2 (Fig. 5A). This clustering pattern differed from those of germ tube assay signifying that the compounds affected the physiology of the fungi differently at unique growth stages. Furthermore, it was observed from the PCA analysis that the most active compounds (C1, C2, and C10) formed cluster III and were active against all the FCO strains (Fig. 5A).

Fig. 5.

Principal component analysis (PCA) showing the distribution of germ tube inhibition of F. oxysporum f. sp. capsici strains in function of tested compounds

Mycelial growth inhibition of the twelve compounds was evaluated against the ten virulent F. oxysporum isolates on PDA medium. All the tested compounds effectively reduced the linear mycelial growth of all the isolates at 3 µg/mL relative to control (Supplementary Table S4). Differential level of mycelial growth inhibition was observed for the molecules relative to all the FCO strains. The compounds C6 and C10 were highly active against FCO mycelial growth hallmarked by 89.5% and 88.6% inhibition while compound C11 was least potent with 64.5% inhibition. The compound C1 was found the most effective exhibiting mycelial growth inhibition at 88.33% for FCO1, 87.33% for FCO2, 86.33% for FCO3, 87.33% for FCO6, 86.33% for FCO7, and 87.67% for FCO8. The compound C6 also showed strong mycelial inhibition of 89.50% for FCO4, 86.67% for FCO9, and 85.67% for FCO10 (Fig. 6). The rest of the compounds were relatively less effective in controlling all the resistant and virulent FCO strains (Supplementary Table S4).

Fig. 6.

Mycelial growth inhibition of the most active compounds on: (A) FCO1, (B) FCO2 (C) FCO2, (D) FCO3, (E) FCO4, (F) FCO5, (G) FCO6, (H) FCO7, (I) FCO8, (J) FCO9, (K) FCO10 and the asterisks indicate levels of significant differences (*P < 0.05; **P < 0.001, ***P < 0.0001)

Discussion

The Fusarium wilt of chili pepper caused by Fusarium oxysporum f. sp. capsici is a serious threat to chili farming and food safety. Based on a worldwide survey and the economic impact, F. oxysporum was voted by plant pathologists as the 5th most significant fungal pathogen (Dean et al. 2012). This has led to the deployment of transgenic resistant plants (Gaspar et al. 2014), the use of crop rotation, cultural techniques, biological control, and chemical control methods in many chili farming regions to manage Fusarium wilt. Because chemical control strategies provided both protective and curative solutions, most Pakistani farmers have opted for fungicides to enhance production and revenue, thus, there is a need to develop new fungicides. Integrated computational drug designing (CADD) has been demonstrated to serve as a privileged scaffold for drug and agrochemical discovery (Xiong et al. 2016, 2017; Chen et al. 2020). The discovered lead molecule (C1) reported herein has the antifungal potential that may be used for targeted control of F. oxysporum f. sp. capsici of diverse levels of virulence and resistance at the conidia and mycelial level of development.

In the present study, the application of integrated computational drug designing (Fig. 1) approach combined with structure-based virtual screening (SBVS), docking, and scoring was used to identify lead compounds with previously untested molecular scaffolds as succinate dehydrogenase inhibitors (SdhIs). The Sdh enzyme catalyzes the oxidation of succinate to fumarate and the reduction of ubiquinone (UQ) to ubiquinol (QH2). The competitive inhibition of Sdh by antifungal compounds prevents the reduction of UQ, interrupts the mitochondrial ETC, and shuts down the TCA cycle. Thus, this leads to a rapid decrease of cellular ATP levels (Matsson and Hederstedt 2001; Glättli et al. 2009). This mode of action enables the antifungal compounds to exert highly effective and selective control of fungal growth and development. In this study, the Sdh 3D homology model (Fig. 2) exhibited an analogous allocation of the main secondary structure and well superimposed but few loops and turns showed some structural deviations (Supplementary Fig. S4). This is not unforeseen, since the secondary structure is well conserved and has fewer amino acid insertions and deletions than the loop regions (Fig. 2). The modeled Sdh showed high sequence homologies with template subunit A (GenBank acession EXL62804.1), B (GenBank: EXL48360.1), C (GenBank: SCO81136.1), and D (GenBank: EXL44012.1). These results are in line with Burger et al. (Burger et al. 1996), and Adams et al. (Adams et al. 2001), who reported that the amino acid sequences of Sdh1 and Sdh2 have 80% homology among eukaryotes at the binding site. In the FCO model of Sdh, the subunits SdhC and SdhD showed low sequence homology with that of A. alternata (Avenot et al. 2008).

Intriguingly, we found that the UQ binding site in the FCO model of Sdh is a hydrophobic pocket that comprised of residues from SdhB, SdhC, and SdhD (Leu73.C, Trp205.B, Trp204.B, Pro201.B, Tyr76.C, Trp82.C, Ser84.C, Trp88.C, and Ile91) (Fig. 2) which corresponds to the findings in Horsefield et al. (2006). Frequent mutations have been recognized in the Sdh at different positions in subunits B, C, and D notably in field isolates of Alternaria and Botrytis species (Avenot et al. 2008; Veloukas et al. 2011), thus, reducing the efficacy of SdhIs (Scalliet et al. 2012; Sierotzki et al. 2013; FRAC 2017). Resistance against SdhIs has become a common phenomenon in many plant fungal pathogens of chili pepper in Pakistan (Khan et al. 2017; Nawaz et al. 2018b; Khaleeque and Khan 1991, Siddiqui and Akhtar 2007). Based on in silico analysis, we found that the binding modes of the discovered SdhI compounds were through the formation of hydrogen bonds with either Phe83 and/or Trp82. This signified that the residues are vital in stabilizing the ligands in the UQ site of FCO of model Sdh (Fig. 2). These results agree with those of Shimizu et al. (Shimizu 2012), who reported that rhodoquinone (RQ) site is surrounded by conserved residues (C-Ser72, C-Arg76, D-Asp106, and D-Tyr107) and involved in H-bond networks with RQ. We found from docking that the 12 selected SdhIs compounds bound to Sdh protein block the active site.

One modeling strategy to enhance disease control and limit the risk of resistance development is to integrate novel fungicides with structural diversity into a spray program (Staub 1991). Among various drug screening approaches, in silico virtual screening is the most practical tool to screen enormous chemical libraries owing to a relatively minimal effort, low cost and have the potential to discover innovative drug candidates (Schapira et al. 2000; Shoichet 2004). The activity of the discovered SdhIs on FCO was accessed in vitro by Sdh enzyme activity assay (Fig. 3), conidia germ tube inhibition (Fig. 4), and mycelial growth inhibition assays (Fig. 5). The enzymatic inhibition varied from (3.0–3.3 U/mL) among all the tested molecules (Fig. 3). This suggested that the molecules formed different numbers and intensity of H-bonds at the UQ binding pocket of FCO-Sdh translated by the formation of two principal components (7.79% and 69.74%) in the clustering analysis (Fig. 3B). These results are similar to previous studies for the interaction of carboxin with Sdh in Botrytis cinerea (Fritz et al. 1993). The positive results prompted the evaluation of the efficacy of the compound to inhibit germ tube development in vitro (Fig. 4). Importantly, strong positive correlation between C1 vs C2 (r* = 0.67), C9 vs C10 (r* = 0.65), and C10 vs C12 (r** = 0.77) are promising indications of the synergistic effects should they be use in combination against diverse FCO strains in disease management.

Using a predefined discriminatory dose of penthiopyrad (3 µg/mL) for assessing the sensitivity of FCO strains, significant and varied mycelial growth inhibition was obtained (Fig. 5, 6). It is possible, that the significant differences in mycelial growth inhibition could be due to variations in the degree of hydrophobicity (or hydrophobicity), penetration strength over the cell wall and cell membrane, SdhIs solubility in the fungal cell protoplasm, and inherent differences in the respiratory system of each fungal strain. Variations in the mycelial antifungal effects of fungicides in Fusarium species are a common phenomenon. For instance, Peters et al. (2008) showed that reference strains of Fusarium coeruleum and F. sambucinum were sensitive to fludioxonil, but all field test strains of F. coeruleum and F. sambucinum were resistant to fludioxonil hallmarked by no growth inhibition at 100 mg/L. All the tested compounds actively reduced mycelial growth of FCO at 3 µg/mL as compared to the negative control (Fig. 6). In this present study, C2 (at 3 μg/mL) was found to be the most effective inhibitor against all the FCO strains at the level of conidia, succinate dehydrogenase enzyme, and mycelia. This finding is in harmony with the ability of compound C2 to form multiple hydrophobic interactions with Trp69, Trp82, Tyr76, and Trp205 at the active site of FCO-Sdh enzyme than the other compounds (Supplementary Fig. S5). Also, the positive correlation of C1 vs C2 (r* = 0.67), C9 vs C10 (r* = 0.65), and C10 vs C12 (r** = 0.77) on germ tube germination inhibition provides wider possibilities for exploiting the synergistic potential of the discovered compounds. Thus, signifying that the discovery of all these SdhIs could be used for the effective control of a wide range of Fusarium oxysporum strains.

Conclusion

In this study, we have integrated computational drug designing (CADD) to accelerate the discovery of structurally diverse antifungal lead compounds that could mitigate the ongoing FCO resistance in Pakistan. The discovered compounds displayed better control of FCO by forming strong conformational binding and providing enhanced antifungal activity effect. Without diversity in fungicide products readily available to farmers in the future, the gains to continue in business and produce adequate food would be threatened. Importantly, compounds C1, C2, C6, and C10 showed high potential to control Fusarium wilt of chili. Furthermore, the compounds identified in the present research work can function as new promising SdhIs and can be used as a starting point for the fungicide formulation, greenhouse and field trials, and prospective integration into pesticide spray programs.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

SI and AAS conceived and designed research. SI, KN, WA, and AAK conducted experiments. AAS and LB contributed new reagents or analytical tools. SI and LB analyzed data. LB and SI wrote the manuscript. All authors read and approved the manuscript.

Declarations

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.