The fieldwork was conducted in three rooftop gardens and three nurseries located around Jahangirnagar University, Savar, Dhaka, Bangladesh. Three plants of each sampling area were observed for the appearance of disease symptoms. During July to September 2023, disease symptoms were recorded and characterized following standard phytopathological manual.

Infected leaves of guava with distinct blight symptoms were collected in sterile polyethylene bags from the rooftop gardens and nurseries. From each location, three diseased leaves with similar symptoms were collected for further investigations. Necessary photographs were taken and samples were cut into pieces at 2 mm diameter. Samples were then surface sterilized with 5% sodium hypochlorite for 3 minutes, followed by washing 3 - 4 times in sterilized distilled water. After surface drying, samples were immediately transferred into petri-dishes containing sterilized Potato Dextrose Agar (PDA) medium, then the plates were incubated at room temperature (25˚C ± 2˚C) for 3 - 5 days. After the fungal appearance, it was subsequently transferred to fresh PDA medium following fungal tip culture [23].

Morphological and molecular methods were used to identify the fungus. Visual characteristics like colony color, appearance, mycelium and reproductive structures were recorded for morphological identification. Molecular identification was done by analyzing the internal transcribed spacer (ITS) sequence although multilocus sequence analysis could be used for more precise detection. Fungal DNA isolation was isolated according to standard protocol and PCR were performed using ITS1 and ITS4 primer sets. DNA sequencing was performed in the First Base Laboratories, Sdn Bhd, Malaysia by commercial service provider. National Center for Biotechnology Information (NCBI) BLASTn search with the nucleotide sequence data were conducted for similarity indexing. The evolutionary history of the fungus was inferred by using the Neighbor-joining method and the Tamura 3-parameter model derived from the ITS1/5.8S rDNA/ITS2 sequence [11]. Evolutionary analyses were conducted in MEGA (version 12.1) software [24]. Invitro pathogenicity test was carried out following detached leaf technique to confirm Koch’s Postulates. Healthy and susceptible leaves of guava were detached from plants and subsequently were surface sterilized with cotton soaked 70% ethanol. After that leaves were placed on petri-dishes containing sterilized water soaked blotting paper to maintain moisture condition. Artificial wounds were made using sterilized needle on the leaves and 5 mm mycelial disc from actively growing fungal culture were inoculated onto the wounded leaves. Agar plug without fungus was also inoculated as control. The inoculated plates were placed into vacuumed desiccator at room temperature for 7 days. Three replications of each treatment were maintained during this experiment. Fungal pathogen was re-isolated and characterized from artificially produced diseased tissues.

To determine the optimum temperature for the fungal growth, sterilized Potato Dextrose Agar (PDA) medium was used and each petri plate was inoculated with a 2 mm diameter active growing fungal disc at the center. Inoculated petri plates were incubated at 15˚C, 20˚C, 25˚C, 30˚C, and 35˚C temperature. To evaluate the effect of H⁺-ion concentration on the fungal growth, pH of PDA medium was adjusted to five different levels viz., 5.0, 6.0, 7.0, 8.0 and 9.0 by calibrated digital pH meter using sterile 1 M HCl or 1 M NaOH solution before autoclave. The inoculated plates were incubated in a controlled environment. The effect of seven different fungal culture media (MA: Malt Agar; HPA: Honey Peptone Agar; HEA: Host Extract Agar; SGA: Sucrose Dextrose Agar; PDA: Potato Dextrose Agar; PSA: Potato Sucrose Agar; CA: Carrot Agar) was also examined to observe the growth response of the fungus. In all the cases, radial mycelial growth was recorded using millimeter scale after 7 days of incubation.

2.5.1. InVitro Control Measures

In the current study, three commercially available fungicides, namely—Kazim 80 WG (carbendazim), Copper blue 50 WG (copper oxychloride), and Amistar top 325 SC (azoxystrobin + difenoconazole) were evaluated at different concentrations (ppm) against the radial growth of the pathogenic fungus following food poison technique. Besides, two Trichoderma species (T.harzianum and T. erinaceum) were employed to assess their efficacy against the fungal isolate following the dual culture technique. Petri plates with untreated PDA medium were used as control. Data on radial growth were recorded after 7 days of post-incubation (dpi), and the following method was used to calculate the percent growth inhibition:

Here, I = Percent growth inhibition; C = Growth of fungus on control plate; T = Growth of fungus on treatment plate.

2.5.2. InSilico Drug Prediction Against the Fungus

The selection of effective phytocompounds through virtual screening and the development of biogenic compounds from selective plants against the target pathogen using insilico approaches were conducted during this study [25].

1)Targetproteinmodeling,3Dstructurerefinement,andvalidation

Translation elongation factor 1-alpha (Tef-1α) is essential for protein synthesis and usually present as a single‑copy, highly conserved gene, so silencing it strongly impairs fungal growth. For this reason, the protein was selected as a target protein of our study [26], and the FASTA sequence of the protein was retrieved from the UniProt website (https://www.uniprot.org/) (Entry: A0A8E6NX89) [27]. The 3D structure of the Tef-1 α protein was generated using the selected sequence through Ab initio modeling [28]. The GalaxyTBM server (https://galaxy.seoklab.org/cgi-bin/submit.cgi?type=TBM) was used in this purpose. The Tef-1α 3D structure was refined using Galaxy Refine server (https://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE) [29]. The best structure was chosen based on GDT-HA, RMSD, MolProbity score, clash score, weak rotamers, and Rama preference. The Ramachandran plot provides relevant information and ratios regarding the anticipated residual structure in protein models within favored, permitted, and disallowed regions [30].

2)Virtualscreeningforphytocompounds

In this study, ten phytocompounds from two common medicinal plants in Bangladesh were carefully chosen based on their antifungal activity. The 3D structures of these compounds were retrieved in SDF format from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/) and converted them into PDB format using Open Bable [31][32].

3)Moleculardocking

a)Proteinandligandpreparation

The protein and ligand structures were prepared for docking using MGL tools and saved as PDBQT format [33]. The energy minimization of both protein and ligand was done by UCSF Chimera software [34].

b)Activesiteprediction,gridboxpreparation,anddocking

Pharmacology studies show that the functional activity of a protein is terminated if ligands bind in the active site of the protein [35]. The active site in the target protein was predicted by the CASTp server (http://sts.bioe.uic.edu/castp/) [36]. Moreover, docking methods determine the conformations assumed around the binding sites of macromolecular targets and calculate the ligand-receptor’s binding free energy [37]. The centre of the grid box was placed at the coordinates X = 25.735 Å, Y = 30.988 Å and Z = 30.846 Å, and it covers most of the binding sites of the protein. The dimensions were X = 60 Å, Y = 60 Å and Z = 60 Å, with a spacing of 0.375 between each grid point. The exhaustiveness value was 8. All docking was performed using AutoDock Vina and the 2D diagram of complexes was visualized using Discovery Studio 2021 to recognize the interactions (hydrogen bonds, electrostatic interactions, and hydrophobic interactions) of the amino acids between protein and ligand [38][39].

4)Fungicideslikeness

The pharmacokinetic properties of the best promising compounds were checked using SWISS ADME (https://www.swissadme.ch/) [40]. Drug-like ligands should exhibit molecular weight ≤ 500 Da, logP ≤ 5, hydrogen bond donors ≤ 5, and hydrogen bond acceptors ≤10 [41].

5)Moleculardynamics(MD)simulation

To thoroughly assess the binding stability of the protein-ligand complexes, comprehensive MD simulations were performed utilizing Desmond software on a Linux-based operating system. The protein-ligand complex structures were hydrated using the cubic three-point transferable interaction potential (TIP3P) system construction tool. To normalize the constructed model, Na⁺ and Cl⁻ charged ions were added for the formation of the 0.15 M physiological salt. To attain optimal efficacy concerning energy usage, the integrated force field, OPLS3e, was implemented. The simulation was performed employing the isothermal isobaric ensemble (NPT) method, with a temperature set at 310 Kelvin and a pressure of 1.013 bar. The capture interval was 100 ps long and occurred inside the 100 ns time frame. Meanwhile, the trajectory’s memory stored a thousand frames.

All statistical observations were carried out on the mean value of the three replications. Analysis of variance (ANOVA) was subjected to statistical analysis using the data analysis software SPSS (version.20). Mean was separated using DMRT to compare the effects of treatment on the different fungi.

Nigrospora leaf blight disease symptoms appeared at the tip which were initially brown-colored and later turned into black (Figure 1(a)). Primarily, a small necrotic spot was formed, which gradually became large to form an overall blight symptom. As the symptom develops the surrounding tissues and the midrib become yellow and later turn into a deep brown to black necrotic lesion.

On PDA medium, the fungus was found moderate to fast-growing nature with white circular colony. Initially colony texture was floccose to cottony and become dense with age (Figure 1(b)). The conidia were globose to sub-globose, single-celled, shiny, smooth, and initially pale brown color. At maturity, the fungus produces abundantly distinct black conidia that often produce dot on the colony surface, giving a peppered appearance. Conidiophores were reduced to conidiogenous cells, which were solitary, ampulliform, or subspherical (Figure 1(c)). Based on the recorded characteristics, the fungus was primarily identified as Nigrospora sp.

Figure 1. Diseased sample and morphology of the fungus. (a) Leaf blight symptom; (b) Fungal colony on PDA medium; (c) Structure of mycelia, conidia and conidiophore (10 × 40X).

The rDNA-ITS was successfully amplified and the PCR amplicon size was ~555bp after necessary trimming from both ends. A BLASTn analysis of the ITS sequence against NCBI GenBank database revealed a high sequence similarity with closely related taxa. The isolate showed >99% identity and 100% query coverage with other Nigrosporachininsis of the database. The sequence alignment led to the creation of a Neighbor-joining tree separates the N.chinensis into a distinct clade including our taxa with a strong bootstrap support, thus confirmed the fungal identity as Nigrosporachinensis Mei Wang & L. Cai [42] (Figure 2). The ITS sequence data was deposited in the GenBank database with the accession number PP442566.1. Successful invitro pathogenicity test confirmed the pathogenic nature of N.chinensis to create the disease. The re-isolated fungus produced similar morphological traits during micro-and macro-morphological characterization.

Figure 2. Neighbor-joining tree showing the phylogenetic relationship among the selected fungal isolates including the current taxa (PP442566.1) with 100% bootstrapping.

The growth response of the target fungus was significantly influenced by the investigated parameters, including culture media, media pH and incubation temperature. Results of this study revealed that there is a significant effect of culture media on the mycelial growth of N.chinensis (p < 0.05), and the highest mycelial growth (73.33 mm) was obtained on PDA medium at 7 dpi followed by HPA, PSA, MA, CA and HEA media (Figure 3(a)). The vegetative growth of the fungus was completely ceased at the host extract media. This is maybe due to high concentration metabolic compounds. Data on the effect of pH revealed that the mycelial growth of the fungus was significantly varied by media pH (p > 0.05). There was an increasing trend of mycelial growth of N.chinensis from pH 5 to pH 7 and it attained the highest diameter of 79.22 mm at pH 7. After reaching the peak the vegetative growth started declining and the growth was very slow at both lowest and highest tested pH conditions (Figure 3(b)). Significant effect of temperature was found on the growth and development of the fungus. The utmost vegetative growth (90 mm) was recorded at 25˚C, followed by 20˚C, while no growth was found at 15˚C and 35˚C (Figure 3(c)). This might have occurred because of the strain’s temperature specificity.

Figure 3. Effect of different factors on the radial vegetative growth of the fungus. (a) Effect of culture media, (b) Effect of pH, (c) Effect of temperature and d-Inhibitory effect of different antifungal agents on the radial mycelial growth of N.chinensis (T-1: Amistar top 250 ppm, T-2: Amistar top 500 ppm, T-3: Copper blue 1500 ppm, T-4: Copper blue 2000 ppm, T-5: Kazim 500 ppm, T-6: Kazim 1000 ppm, T-7: T.harzianum and T-8: T.erinaceum). [The same letter denotes no significant difference at p < 0.05].

3.5.1. InVitro Control Measures of the Fungus

The efficacy of selective chemical fungicides and fungal antagonists was evaluated and all of them exhibited varying degrees of growth inhibition with significant differences among treatments. Amistar top at 250 ppm and 500 ppm concentration showed complete growth inhibition of the fungus and Kazim at 500 ppm and 1000 ppm showed a similar inhibitory response (Figure 3(d)). However, copper blue at 1500 ppm and 2000 ppm concentrations showed about 50% and 80% growth inhibitory effects, respectively. The fungal antagonist Trichoderma spp. demonstrated substantial inhibitory effects on N.chinensis through rapid colonization and overgrowth in the dual culture assay. Between two antagonists, T.harzianum showed >70% growth inhibition, followed by T.erinaceum (57%), which is often comparable to lower dose of chemical fungicide treatments.

3.5.2. InSilico Drug Prediction Against the Fungus

1)3DStructurePrediction,Validation,Refinementandthebindingpocketofthetargetprotein

The GalaxyTBM server was used as the best template to generate the top five models of the target protein Among the five models, model 1 was selected as recommended by the server. Then this protein structure is refined using the GalaxyWEB server. GalaxyWEB generated 5 refined models of the predicted 3D structure of the targeted protein. Among them, model 5 was selected based on an RMSD value of 0.88, MolProbity 0.80, Clash score of 0.9, and Poor rotamers score of 0.0. Validation of the protein structure through the Ramachandran plot from the SAVES server showed 94.5% residues in the favored region (Figure 4). Multiple active sites were assumed by CastP server (sequence no. TRP 35, ILE 72, ALA 108, LEU 184, TYR 13, LEU 17, MET 156, PHE 151).

Figure 4. Validation of the 3D structure of the protein. The Ramachandran plot statistics represent the most favorable regions with a percentage of 94.5%.

2)Moleculardockinganalysis

The target protein contained four active sites marked as pockets where the ligands preferred to dock (Figure 5). Autodock Vina performed docking between the protein and selective ligands to anticipate their binding interactions. Ten phytocompounds were docked with the target protein (Tef-1) and seven were shown to have a higher affinity for binding than the commercial fungicide (carbendazim) which was designated as a control in this study (Table 1).

Figure 5. Visualization of the protein. (a) The 3D structure of the selected target protein; (b) Binding sites in the target protein.

Table 1. Docking outcome of selected phytocompounds and control chemical with the target protein. The table shows the binding affinity score of phytocompounds and control chemicals with their sources.

The phytocompounds with higher binding affinity than the control (−7.1) have been selected as the potential fungicide candidate. In this study, seven phytocompounds exhibited better binding affinity score over the control.

3)Fungicidelikeness

The SwissADME online program integrates different pharmacokinetic guidelines used for analysis to define leads that will be appropriate for use as an effective antifungal compound. Among the seven phytocompounds, five compounds displayed no signs of breaking the laws governing fungicide likeness (Table 2).

Table 2. Fungicide likeness properties of the selected phytocompounds.

4)Visualizationofa2Ddiagramofthebest-hitcompounds

Figure 6. Illustration of 3D and 2D complexes representation. The white circle depicts the binding pockets and includes ligands and target molecules. (a) 3D and 2D interaction of tef-1 with lawsaritol complex. (b) 3D and 2D interaction of tef-1 with andrograpanin complex. (c) 3D and 2D interaction of tef-1 with carbendazim (control) complex.

In the molecular docked complexes, the H-bonds and non-bonded interactions indicate the stabilization of binding interactions. In this study, Lawsaritol from L.enermis and Andrograpanin from A.paniculata had the highest binding affinity with tef-1 protein (Table 1). For the protein-lawsaritol complex, the compound forms an H-bond with the amino acid in position 24 of tef-1 (Figure 6(a)). For the protein-andrograpanin complex, the compound forms an H-bond with the amino acid in position 134, 126, 127 of tef-1 (Figure 6(b)). For the protein-control complex, the compound forms an H-bond with the amino acid in position 45 of tef-1 (Figure 6(c)). Afterward, these complexes were run for a stability check through MD simulation.

5)AnalysisoftheresultoftheMDsimulation

A 100-ns MD simulation was conducted using the two top-performing compounds, lawsaritol, andrograpanin, and the control fungicide carbendazim to verify the stability of the ligands at the active site of the protein. The RMSD, RMSF, hydrogens, and several bonds were assessed. The average RMSD plots of the backbone of Tef-1 protein regarding lawsaritol, andrograpanin, and carbendazim were 2.588 Å, 1.807 Å, and 2.026 Å, respectively (Figure 7(a)). Throughout the remainder of the 100-ns simulation period, the deviation of the curve remained below 3.00 Å. The RMSD of the ligands was more stable than that of the control. The curve of the control was not stable during the 100 ns simulation; in the meantime, the other compounds became stable before 20 ns, and the average RMSD values of the lawsaritol and andrograpanin were 3.023 and 2.935 Å, respectively (Figure 7(b)). The average RMSF value of the protein regarding the ligands was almost the same as the control, 3.89, 5.65, 4.51 Å (Figure 7(c)).

Figure 7. RMSD and RMSF of the protein-ligands complex. The graph demonstrated (a) the RMSD value of the backbone of the protein regarding ligands, (b) the RMSD of the ligands, (c) the RMSF of amino acids of the protein.

The non-covalent interactions between the protein and ligands were evaluated in the 100-ns simulation period (Figure 8). Regarding lawsaritol, 4 interactions with ARG24, PHE37, ASP48, and PHE151 of the protein lasted more than 20% of the total simulation time. Another ligand, andrograpanin, interacted with 3 (TYR13, TRP35, and ASN110) amino acids of the protein for more than 35% of the simulation period. On the other hand, carbenzadim showed better interaction with THR45 of the protein, and the interaction time was 125% of the simulation. Usually, it happens when ligands interact with the same amino acid several times [43].

Figure 8. Histogram of the protein-ligand interaction. a, b, and c indicate the interaction of lawsaritol, andrograpanin, and carbendazim, respectively.