Physicochemical, Pharmacokinetic and Cytotoxicity of the Compounds Isolated from an Endophyte Fusarium oxysporum: In Vitro and In Silico Approaches

The present study was intended to characterize the secondary metabolites of the endophyte Fusarium oxysporum isolated from the plant Aglaonema hookerianum Schott. And to investigate the cytotoxic and other pharmacological properties of the isolated compounds as part of the drug discovery and development process. Different chromatographic techniques were adopted to isolate the bioactive compounds that were identified by spectroscopic techniques. The cytotoxic properties of the compounds were assessed in the Vero cell line via the trypan blue method. Moreover, physicochemical, pharmacokinetic, bioactivity and toxicity profiles of the compounds were also investigated through in silico approaches. After careful spectral analysis, the isolated compounds were identified as 3β,5α-dihydroxy-ergosta-7,22-dien-6-one (1), 3β,5α,9α-trihydroxy-ergosta-7,22-dien-6-one (2), p-hydroxybenzaldehyde (3), 3-(R)-7-butyl-6,8-dihydroxy-3-pent-11-enylisochroman-1-one (4) and beauvericin (5). An in vitro study in the Vero cell line revealed that the presence of the compounds reduced the number of cells, as well as the percentage of viable cells, in most cases. An in silico cytotoxic analysis revealed that compounds 1, 2 and 5 might be explored as cytotoxic agents. Moreover, compounds 3 and 4 were found to be highly mutagenic. The present study suggested that further thorough investigations are necessary to use these molecules as leads for the cytotoxic drug development process.


Introduction
Most endophytes are symbiotically associated with their host plants and are able to produce bioactive secondary metabolites without causing apparent damage to the plant. These secondary metabolites become an attractive source for therapeutic compounds, which are still a poorly explored field in drug discovery [1]. The literature has established the genus Fusarium as a repository of bioactive compounds, including anti-fungal agents [2], antimicrobial agents [3], fungal toxins [4,5] and immunosuppressive agents [6]. More specifically, the genus Fusarium could be a potent source for isolating anticancer compounds such as taxol isolated from Fusarium redolens [7], camptothecin and podophyllotoxins isolated from Fusarium solani [8,9] and vincristine isolated from Fusarium oxysporum [10]. Furthermore, the above examples proved that the production of microorganism-based natural drugs can minimize the vulnerability of plant species as sources of natural drugs. Fusarium oxysporum has been established as a promising source of numerous bioactive molecules, such as jasmonic acid and 9,10-dihydrojasmonic acid [11], fumonisin (C1-C4) [12], bikaverin [13], fusarinolic acid, beauvericin, cerevesterol [14], sambutoxin [5], fusarin C [15] and many others.

Morphological Identification of the Fungal Strain
The tested fungal strain was recognized as Fusarium oxysporum on the basis of the key features of colony morphology. In this isolate, violet macroconidia were produced in the central spore's mass and dark magenta or violet pigment were produced in the agar medium. Mycelia was scarce, floccose and white to violet in color ( Figure 1). The isolate also produced abundant, pale orange sparsed sporodochia. Most macroconidia were of short to medium length, slightly curved or straight, sometimes with a slight hook, relatively slender and thin-walled and usually 3-septate. The microconidia were elliptical, oval or reniform (kidney shaped) and usually non-septated. The chlamydospores were infinite singly or in pairs and quickly formed within 2 to 4 weeks for most isolates [21].

Characterization of Compound 1 as 3β,5α-Dihydroxy-ergosta-7,22-diene-6-one
In the 1 H nuclear magnetic resonance (NMR) spectrum, the existence of one one-proton multiplet at δ = 4.07 ppm and one one-proton singlet at δ = 5.67 ppm represents H-3 and H-7 of a sterol molecule. The two double-doubles at δ = 5.18 ppm and δ = 5.26 ppm could be assigned to two olefinic protons located at the side chain of steroidal compounds. The 1 H NMR data also revealed that compound 1 (Figure 2) contains six methyl signals at δ = 1.05, 0.97, 0.93, 0.84, 0.87 and 0.62 ppm. Two signals at δ = 67.5 and 71.1 in the 13 C NMR spectrum clearly specify the presence of two hydroxyl groups attached to the C-3 and C-5 positions and a signal at δ = 197. 4 showed the presence of a

Characterization of Compound 1 as 3β,5α-Dihydroxy-ergosta-7,22-diene-6-one
In the 1 H nuclear magnetic resonance (NMR) spectrum, the existence of one oneproton multiplet at δ = 4.07 ppm and one one-proton singlet at δ = 5.67 ppm represents H-3 and H-7 of a sterol molecule. The two double-doubles at δ = 5.18 ppm and δ = 5.26 ppm could be assigned to two olefinic protons located at the side chain of steroidal compounds. The 1 H NMR data also revealed that compound 1 (Figure 2) contains six methyl signals at δ = 1.05, 0.97, 0.93, 0.84, 0.87 and 0.62 ppm. Two signals at δ = 67.5 and 71.1 in the 13 C NMR spectrum clearly specify the presence of two hydroxyl groups attached to the C-3 and C-5 positions and a signal at δ = 197.4 showed the presence of a carbonyl group at the C-6 position ( Figures S3-S8). Finally, based on the above analysis, compound 1 was identified as 3β,5α-dihydroxy-ergosta-7,22-diene-6-one, a steroidal molecule containing two hydroxyl groups, two double bonds, six methyl groups and a carbonyl group with 28 carbons [22]. The structure of compound 2 ( Figure 2) was elucidated by a comparison of its spe tral data with those of compound 1. Differences were found in the C-9 substituent. A hough both the 1 H NMR and 13 C NMR spectra of compound 2 were in close corr spondence to those of compound 1, one new signal at δ = 74.7 appeared, and the signal δ = 44.7 disappeared in the 13 C NMR spectrum of compound 2 ( Figures S9-S14). The hig resolution electrospray ionization mass spectrometry (HRESIMS) spectrum with [M Na] + at m/z = 467.3142, in conjunction with the other spectral data, suggested the mole ular formula C28H44O4 (Figure S15), and the compound was recognized 3β,5α,9α-trihydroxy-ergosta-7,22-diene-6-one [23].

Characterization of Compound 3 as p-Hydroxybenzaldehyde
The 1 H NMR and 13 C NMR spectra of compound 3 ( Figure 2) showed signals for a aldehydic proton at δ = 9.76 and δ = 191.4, respectively. The proton signals at δ = 6.89 an 7.73 represent the presence of disubstituted aromatic protons, which is also supported b the presence of olefin methine carbon signals at δ = 132.4 and 115.8 and one oxygenate olefin carbon signal at δ = 163.2. A comparison of these data with the published valu [24] allowed for the characterization of this compound as p-hydroxybenzaldehyde (Fi ures S16-S19).

Characterization of Compound 4 as
The presence of one one-proton singlet at δ 6.21 in the 1 H NMR spectrum is a tributed to one aromatic proton. The presence of only one aromatic proton indicated th compound 4 (Figure 2) might carry a pentasubstituted benzene ring. The presence of on 2.3. Characterization of Compound 2 as 3β,5α,9α-Trihydroxy-ergosta-7,22-diene-6-one The structure of compound 2 ( Figure 2) was elucidated by a comparison of its spectral data with those of compound 1. Differences were found in the C-9 substituent. Although both the 1 H NMR and 13 C NMR spectra of compound 2 were in close correspondence to those of compound 1, one new signal at δ = 74.7 appeared, and the signal at δ = 44.7 disappeared in the 13 C NMR spectrum of compound 2 ( Figures S9-S14). The high resolution electrospray ionization mass spectrometry (HRESIMS) spectrum with [M + Na] + at m/z = 467.3142, in conjunction with the other spectral data, suggested the molecular formula C 28 H 44 O 4 ( Figure S15), and the compound was recognized as 3β,5α,9α-trihydroxyergosta-7,22-diene-6-one [23].

Characterization of Compound 3 as p-Hydroxybenzaldehyde
The 1 H NMR and 13 C NMR spectra of compound 3 ( Figure 2) showed signals for an aldehydic proton at δ = 9.76 and δ = 191.4, respectively. The proton signals at δ = 6.89 and 7.73 represent the presence of disubstituted aromatic protons, which is also supported by the presence of olefin methine carbon signals at δ = 132.4 and 115.8 and one oxygenated olefin carbon signal at δ = 163.2. A comparison of these data with the published values [24] allowed for the characterization of this compound as p-hydroxybenzaldehyde (Figures S16-S19).

Characterization of Compound 4 as 3-(R)-7-Butyl-6,8-dihydroxy-3-pent-11-enylisochroman-1-one
The presence of one one-proton singlet at δ 6.21 in the 1 H NMR spectrum is attributed to one aromatic proton. The presence of only one aromatic proton indicated that compound 4 ( Figure 2) might carry a pentasubstituted benzene ring. The presence of one sharp oneproton singlet at δ = 11.44 could be ascribed to one phenolic chelated hydroxyl group and the doublet signal at δ = 1.66 (J = 5.6 Hz) could be ascribed for the methyl group at C-13. Two double doublets at δ = 5.51 and 5.43 confirm the presence of a trans-olefinic bond between C-11 and C-12. The signal at δ = 170.5 in the 13 C NMR spectrum revealed the presence of a carbonyl carbon in lactone moiety and an oxygenated methine carbon at C-3 (δ = 78.5). The 13 C NMR spectrum also showed two signals at δ = 160.0 and 162.2 for the two phenolic carbons at C-6 and C-8, respectively. The 13 C and DEPT-135 NMR spectra revealed the presence of two methyls, four methines, six methylenes and six quaternary carbons in compound 4 ( Figures S20-S28). An accurate mass measurement of compound 4 obtained by HRESIMS yielded a parent mass of m/z 327.1565 in positive ionization mode, corresponding to the sodium adduct [M + Na] + with a molecular formula of C 18 H 24 O 4 ( Figure S29). On the basis of the above observations and compared with the literature data [25], compound 4 was identified as 3-(R)-7-butyl-6,8-dihydroxy-3-pent-11-enylisochroman-1-one, a dihydroisocoumarin derivative.

Characterization of Compound 5 as Beauvericin
The resonance at δ = 0.41, 0.82 and 3.04 in the 1 H NMR and at δ = 17.4, 18.3 and 32.2 in the 13 C NMR spectra, in conjunction with the DEPT-135 spectrum, could be attributed to three methyl groups in compound 5 ( Figure 2). In a similar way, δ H = 2.97 and δ C = 34.7 represented one methylene, δ H = 1.99, 4.88 and 5.58 along with δ C = 29.7, 75.6 and 57.2 represented three methines and δ H = 7.24 and δ C = 126.8-128.8 could be attributed to a benzene ring. Furthermore, the signals at δ H = 3.04 and δ C = 32.2 revealed the presence of a N-methyl amino acid moiety of beauvericin ( Figures S30-S37). The 13 C NMR spectrum confirms the presence of fifteen carbons, and an X-ray crystallography report ( Figure S39) confirms three moieties with symmetrical structures in the compound. The HRESIMS spectra of compound 5 yielded a parent mass of m/z 806.3984, corresponding to the sodium adduct [M + Na] + with a molecular formula of C 45 H 57 N 3 O 9 (calcd. mass 806.9404 [C 45 H 57 N 3 O 9 + Na] + ), accounting for 19 degrees of unsaturation ( Figure S38). By careful inspection of its 1 H and 13 C NMR spectral data along with mass and X-ray crystallography reports [26], this compound was characterized as beauvericin.

Trypan Blue Test
The cytotoxic efficacy of the isolated compounds was assessed against the Vero cell line (African Green Monkey kidney cell) after 24 h of exposure. The toxicity of the compounds was determined where a dose-dependent reduction in cell viability was observed. In case of compound 1, the percentage of viable cells was lowered significantly at the concentration of 1.0 µM, but afterwards the percentage of viable cells was augmented with the increased concentrations ( Figure 3). On the other hand, compounds 3, 4 and 5 reduced the cell viability with their increased concentrations ( Figure S40). The results were consistent for the total cell number for all the compounds except compound 1, where total cell growth was induced with the increased concentrations ( Figure 4). On the other hand, cell size was reduced with the higher concentrations of all the tested compounds except for compound 4. In case of compound 4, cell size was enlarged and leveled off when the concentration was 5.0 µM; afterwards, cell size dropped with the augmented concentrations of the compound ( Figure 5). It is necessary to conduct further specific and higher studies to explore the mechanism of anticancer efficacy of these compounds.     The results are expressed as the mean ± SEM (n = 3). The degrees of significance calculated by employing ANOVA with Post Hoc Tukey's test were * p < 0.05, ** p < 0.01 and *** p < 0.001.

In Silico Prediction of Physicochemical and Pharmacological Properties
In silico analyses are often performed to predict important information such as physicochemical, pharmacodynamic and pharmacokinetic properties of bioactive compounds in a faster manner. It has become the method of choice as an early drug discovery process to improve the efficacy and druggability properties, as well as to avoid various Here, compound 1 = 3β,5α-dihydroxy-ergosta-7,22-diene-6-one, compound 3 = p-hydroxybenzaldehyde, compound 4 = 3-(R)-7-butyl-6,8-dihydroxy-3-pent-11-enylisochroman-1-one, compound 5 = Beauvericin. Different shaded bars represent different concentrations of the test samples and the vehicle DMSO. The results are expressed as the mean ± SEM (n = 3). The degrees of significance calculated by employing ANOVA with Post Hoc Tukey's test were * p < 0.05, ** p < 0.01 and *** p < 0.001.

In Silico Prediction of Physicochemical and Pharmacological Properties
In silico analyses are often performed to predict important information such as physicochemical, pharmacodynamic and pharmacokinetic properties of bioactive compounds in a faster manner. It has become the method of choice as an early drug discovery process to improve the efficacy and druggability properties, as well as to avoid various toxicities and other side effects of drug candidates. Therefore, the probability of success in drug development has increased and overall expenses have decreased [27]. In this study, several online tools were used to assess the physicochemical, bioactivity, absorption, distribu-tion, metabolism, elimination and toxicity (ADMET) properties of the isolated compounds (Table 1). Predicted data based on SwissADME, Molinspiration, Osiris Property Explorer and pkCSM web service. Here, TPSA = topological polar surface area; N or O = NH or OH = number of NH and OH; BBB = blood brain barrier; GPCR = G-protein coupled receptor; GI = gastrointestinal.

Physicochemical Properties
Molecular descriptors such as lipophilicity (n-octanol/water partition coefficient, Log P O/W ), solubility (Log S), hydrogen bond donor and acceptor, rotational bonds, topological polar surface area (TPSA) and molar refractivity contribute to the potency, selectivity against the target and ADMET profiles of the drug candidates [27]. Therefore, this study predicted the physicochemical properties of isolated molecules using the SwissADME tool (http: //www.swissadme.ch/ by the Swiss Institute of Bioinformatics, Switzerland) (accessed on 10 November 2021) ( Table 1). As stated in Lipinski's rule of five, the Log P O/W of a compound needs to be less than 5.0 for adequate absorption through the cell membrane [28]. The consensus Log P O/W value obtained from five different predictions (iLogP, WLogP, XLogP3, Silicos-IT and MLogP) indicated that compounds 2, 3 and 4 met this criterion, whereas compounds 1 and 5 have slightly higher values than 5.0. The water solubility (Log S) of the isolated compounds was also determined using three different predictions (ESOL, Ali and Slilicos-IT), where the order of the solubility was compound 3 > 2 > 1 and 4 > 5. To determine the probability of the molecules to be active orally, drug-likeness was assessed through the Lipinski, Veber, Ghose, Muegge and Egan filters (Table 1), which were developed on the basis of several physicochemical features. It is a qualitative hypothesis to delineate the relationship between physicochemical and pharmacokinetic parameters [29]. In this study, compound 4 satisfied all the rules of five different filters except for one rule (XLogP3 > 5) of Muegge's filter, and compound 2 violated only two rules of Ghose's filter. Compounds 1 and 3 met the parameters of Lipinski, Veber and Egan's filters, but failed to satisfy all the parameters of Ghose and Muegge's filters. Compound 5 met only the rules of Veber's filter.

Bioactivity
The Molinspiration online tool (https://www.molinspiration.com/ by Molinspiration Cheminformatics, Slovak Republic) (accessed on 11 November 2021) provides an activity score of the compounds between −3.0 and 3.0 to identify their activity as G-protein coupled receptor (GPCR) ligands, nuclear receptor ligands, protease inhibitors, ion channel modulators, kinase inhibitors and enzyme inhibitors. Molecules with the highest score possess the highest possibility to be active. Here, compounds 1, 2, and 4 were found to be biologically active as nuclear receptor ligands, where compound 2 (0.91) has the highest bioactivity score, followed by compound 1 (0.75) and compound 4 (0.49). These compounds also have similar bioactivity scores to act as enzyme inhibitors (Table 1). On the contrary, compounds 3 and 5 showed bioactivity scores <0.00, which indicates their inactivity against these targeted sites.

Pharmacokinetic Properties
Pharmacokinetic features such as the absorption, distribution, metabolism and excretion of drug molecules play a major role in reaching the clinic for end users. An analysis of the isolated compounds using the SwissADME tool revealed that all compounds except for compound 5 were expected to be highly absorbable through the gastrointestinal (GI) tract. Compounds 3 and 4 showed their ability to be permeable to the blood brain barrier (BBB), whereas compounds 1, 2 and 5 were non-permeable to the BBB. In case of skin permeability, the higher the negative log Kp value, the lower the skin permeation of the molecule [29]. Therefore, the permeability of the compounds to the stratum corneum of the skin is as follows: compound 4 > 1 > 5 > 2 > 3. P-glycoprotein (P-gp), a protein of intestinal cell membrane, has the capability to pump out the absorbed drug into the gut lumen. Compounds 2 and 5 were predicted to be P-gp substrate, whereas compounds 1, 3 and 4 were non-substrate to this membranous protein. An important class of enzymes for metabolism is Cytochrome P 450 , and inhibition or induction of these enzymes by drug molecules produces undesirable metabolites, which may result many unwanted drug reactions [30]. In this study, all compounds except for compound 4 were found to be non-inhibitors of important isoenzymes such as CYP2C19, CYP1A2, CYP2C9, CYP3A4 and CYP2D6. Compound 4 was predicted to be an inhibitor of the CYP1A2 and CYP2C9 isoenzymes (Table 1).

In Silico Analysis of Toxic Properties
An analysis of the compounds by the Osiris Property Explorer revealed that compounds 1, 2 and 5 were predicted to be safe, as toxicological parameters such as mutagenicity, tumorigenicity, irritation and effect on the reproductive system were absent. Mutagenic and irritant effects were highly present in compound 3, whereas compound 4 has moderate mutagenic properties. The maximum tolerated dose (MTD) of the compounds for human use was predicted through the pkCSM online tool (http://biosig.unimelb.edu.au/pkcsm/ by Bio21 Institute, University of Melbourne, Melbourne, Australia) (accessed on 7 November 2021), where compounds having a value below 0.477 (log (mg/kg/day)) needed to be given in low doses due to their toxic effects. The MTD of isolated compounds 1, 2, 4 and 5 was found to be low, indicating their capacity to act as toxic, whereas compound 3 can be given at high doses due to its lower potency of toxicity. The activity of the compounds against human ether-a-go-go related gene (hERG) was also investigated. Potassium channels are encoded by two genes (hERG I and hERG II), and inhibition of these channels develops an acquired long QT syndrome that may lead to severe ventricular arrhythmia. An in silico prediction against hERG I revealed no inhibitory features of the isolated compounds. Compounds 1, 2 and 3 were not found as inhibitors against hERG II, whereas compounds 4 and 5 showed their probability to act as inhibitors for hERG II. Drug-induced liver toxicity is an important parameter of drug development. Using the pkCSM prediction tool, compounds 1, 2 and 5 were found as hepatotoxic, whereas compounds 3 and 4 did not show any relation to hepatotoxicity (Table 1).

In Silico Cell Line Toxicity
The in silico cytotoxic activity of the isolated compounds in different cell lines is shown in Table 2, where three cancerous cell lines with a high probability value are represented for each compound. In CLC-Pred (http://www.way2drug.com/cell-line/ developed by Institute of Biomedical Chemistry of Russian Academy of Medical Sciences, Moscow, Russia) (accessed on 15 November 2021), the activity of the compounds was calculated based on the statistics of the Multilevel Neighbourhood of Atoms (MNA) descriptors and labelled with the probability of the compound being active (Pa) or inactive (Pi), ranging from 0.000 to 1.000 [31].

Discussion
Nature is considered to be the best source to identify new bioactive compounds with potent anticancer activity. More than 60% of clinically used drugs against cancer are developed from natural sources, and they demonstrate their activity via apoptosis, autophagy, immune function regulation or cell proliferation inhibition [32]. Therefore, the focus of this study was to investigate the toxicity profiles of five compounds obtained from Fusarium oxysporum isolated from the plant Aglaonema hookerianum Schott. Moreover, the physicochemical, bioactivity and pharmacokinetic profiles of these molecules were analyzed to determine their eligibility to be the drug/lead molecules for the anticancer drug development process.
Compounds p-hydroxybenzaldehyde (3), 3-(R)-7-butyl-6,8-dihydroxy-3-pent-11enylisochroman-1-one (4) and beauvericin (5) inhibited total cells in number and viable cells in percentage in transformed Vero cells. Similar to our observation, various laboratories also found that compound 5 inhibited different transformed cell lines, such as Vero cells, human breast cancer cells BC-1, human monocytic lymphoma cells U-937, etc. [33][34][35][36][37][38]. Previous in vitro studies have reported that compound 5 has slowed the proliferation of different types of cells. This inhibition was also dependent on dose and time of incubation [39,40]. The trypan blue results suggested that compounds 3, 4 and 5 may hinder cell growth or initiate apoptosis or necrosis. It is also indicated that compounds 3, 4 and 5 induced cell damage of some sort, but it is not known whether this is apoptosis or necrosis. Apoptosis or necrosis induced by extracellular signaling pathways in in vitro cultures would be a possible explanation of the reduction of the viability of cells in culture. It is not well established from our results that there was any co-relation between the inhibition of cell proliferation, cell size and total cell number, but it is presumed that any co-relation may remain. Compound 3β,5α-dihydroxy-ergosta-7,22-diene-6-one (1) decreased the percentage of viable Vero cells at a certain concentration, but afterwards it was augmented. Compound 1 also increased the total cell number, opposite to the actions of the other compounds. This finding revealed that compound 1 may activate cell division or inhibit apoptosis or necrosis in Vero cells. This result indicated that compound 1 may protect cells, especially kidney cells, from cell damage, though further thorough investigation is required to confirm this.
The physicochemical, bioactivity, pharmacokinetic and toxic properties of the isolated compounds were evaluated through different online tools. Compound 4 met almost all the eligibility criteria of different drug-likeness filters and possessed the probability to be a nuclear receptor ligand. It showed less probability to be active against different cancer cell lines. However, it demonstrated the probability to have moderate mutagenic properties and to be an inhibitor of CYP1A2 and CYP2C9 isoenzymes and hERG II. Compound 2 satisfied the criteria of all drug-likeness features except for the Ghose filter. It has shown the highest probability (0.91) of being a nuclear receptor ligand. Moreover, it may be a P-gp substrate and a non-inhibitor of important isoenzymes. However, it showed a probability for hepatotoxicity along with activity in both cancerous and non-cancerous cell lines. Compounds 1 and 3 met the drug-likeness criteria of Lipinski, Veber and Egan's filters. Compound 1 showed its probability to be a nuclear receptor ligand, whereas compound 3 is predicted to be a blood brain barrier (BBB)-permeant molecule and demonstrated negative bioactivity scores against all target sites. Compound 1 showed the probability to be active against colon and ovarian adenocarcinoma cells and compound 4 demonstrated its activity against Oligodendroglioma (Hs 683). Moreover, compound 1 did not show any undesired effects except for hepatotoxicity, while compound 3 demonstrated a probability to be highly mutagenic and highly irritant. Compounds 1 and 2 have almost similar structures with the only exception at C9 position, where compound 2 possesses a hydroxyl moiety. The presence of this functional group made an impact on the solubility and absorption (being a P-gp substrate) profiles (Table 1) of compound 2. Compound 5 is lipophilic in nature and met only the drug-likeness features of Veber's filter. It showed a higher probability of being active against glioblastoma (SF-268), breast adenocarcinoma (MDA-MB-231) and pancreatic carcinoma (MIA PaCa-2) cell lines (Table 2), which is supported by an in vitro study conducted by Zhan et al. [35]. It was found to be an inactive molecule (bioactivity score <0.00) against the target sites available in the Molinspiration tool, which may be due to its high molecular weight. A recent in vitro study has reported it as a potent anticancer molecule against KB cells through the inhibition of acetyl-CoA acetyltransferase 1 (ACAT1) [41]. However, this compound requires structural modification to improve oral bioavailability as well as to avoid the features of hERG II inhibitor and hepatotoxicity.

Conclusions
Our present study highlighted the possibility of establishing Fusarium oxysporum as a treasury of promising secondary metabolites. Considering in vitro and in silico analyses, it can be summarized that the isolated compounds can be used as prominent lead compounds for anticancer drug development, and further detailed studies are required to design the synthetic analogs of these compounds along with the establishment and optimization of their therapeutic and pharmacokinetic activities.

Collection and Identification of the Plant Material
In August 2014, the plant A. hookerianum was collected from Pablakhali, Chittagong Hill Tracts, Bangladesh. The plant's taxonomical characterization was completed by the Bangladesh National Herbarium and a specimen bearing the no. DACB 40633 was kept for future reference ( Figure S1).

General Experimental Procedures
The silica gel for column chromatography and the silica plates for thin layer chromatography were procured from Merck, Darmstadt, Germany. A Bruker 400 MHz spectrophotometer was operated to record all the NMR spectra of the compounds using deuterated solvents (CDCl 3 and MeOD). The Vero cell lines were supplied by CLS cell lines service GmbH, 605372, Eppelheim, Germany.

Isolation and Extraction of Fungal Material
After surface sterilization, the edges of sliced leaves, roots and petioles of A. hookerianum were cut under sterile conditions and placed on a water agar media. To inhibit bacterial growth, streptomycin with a concentration of 100 mg L −1 was mixed with the media. Within 4-5 weeks, fungal growth initiated from the sliced segments of the plant that were taken off of the water agar media and placed onto a potato dextrose agar (PDA) medium [16][17][18][19][20]. Two endophytic fungi were collected from the petiole and characterized as Fusarium sp. based on their macroscopic and microscopic morphological characters. Similarly, two endophytic fungi were collected from the leaf and characterized as Colletotrichum sp. One of the Fusarium sp. was selected to isolate bioactive compounds based on its preliminary bioassay screening and was subjected to elaborated microscopical identification to confirm the species ( Figure S2). The selected fungal strain was cultivated at a large scale by using approximately 20 L of PDA media and maintaining the temperature at 28 ± 2 • C. After 21 days, the cultured media with the matured fungal materials were soaked with ethyl acetate for 7 days with intermittent shaking. After that period, the solvent was filtered sequentially through cotton plug and filter paper. The collected filtrate was concentrated to obtain the crude extract (8.0 g).

Identification of the Selected Endophytic Fungus
A fungal isolate was cultured on PDA for 7 days for morphological examination. The isolate was characterized by its macroscopic and microscopic characteristics (i.e., color, size, length and width of macroconidia and microconidia and septation) according to the manual [21].

In Vitro Cytotoxicity Analysis by Trypan Blue Assay
The investigation of cytotoxicity of the isolated compounds was conducted using a modified trypan blue exclusion method [42,43]. The Vero cell line was used to evaluate the cytotoxicity and was cultured at 37 • C following the process narrated by Khan et al., 2018 [17]. Different doses (0.1 to 20 µg/mL) of the compounds were added into T-25 cell culture containers containing around 2.5 × 10 6 cells and incubated for 24 h. After the incubation period, the treated cells were collected using 0.5% trypsin. Trypan blue (0.4% w/v) was used to make the unviable cells stained and viable cells remained unstained. An automated cell counter was used to count the number of unviable (stained) cells [44]. The percentage of unviable cells was calculated as follows: Percentage of unviable cells = [number of unviable cells/total number of cells] × 100.

In Silico Analysis of Physicochemical, Bioactivity, Pharmacokinetic and Toxicity Properties
Several online tools were used to assess the physicochemical, bioactivity and AD-MET properties of the isolated compounds. The physicochemical, pharmacokinetic and medicinal chemistry aspects of the compounds were screened through the SwissADME tool (http://www.swissadme.ch/index.php) (accessed on 10 November 2021). The Molinspiration online tool version 2018.03 (https://www.molinspiration.com/) (accessed on 11 November 2021) was used to assess the bioactivity score of the compounds against drug targets such as GPCR, kinase enzymes, ion channels, nuclear receptors and other enzymes. The toxicity risks such as the mutagenic, tumorigenic, irritant and reproductive effects of the compounds were determined using the Osiris Property Explorer, and other toxic features were analyzed via the pkCSM online tool (http://biosig.unimelb.edu.au/pkcsm/) (accessed on 7 November 2021). In each case, canonical simplified molecular input line entry system (SMILES) notations of the isolated compounds were fed into the tools to obtain the outcomes.

In Silico Cell Line Toxicity Analysis
The cytotoxic effect of the compounds was screened in both normal (non-tumor) and cancer cell lines by a web service (http://way2drug.com/Cell-line/) (accessed on 15 November 2021) known as the Cell Line Cytotoxicity Predictor (CLC-Pred), where the cytotoxicity of chemicals is predicted based on a PASS (Prediction of Activity Spectra for Substances) algorithm using a training dataset of 59,882 cytotoxic compounds obtained from the experimental data of ChEMBL (version 23) (https://www.ebi.ac.uk/chembldb/) (accessed on 15 November 2021) to generate and establish the 'structure-cytotoxicity' relationship models against 278 cancer cell lines and 27 normal cell lines of humans [31]. The chemical structure of each compound in the form of SMILES was submitted to CLC-Pred to predict the cytotoxic activity.

Statistical Analysis
A statistical analysis was completed using Prism v5.0 (GraphPad Software Inc., San Diego, California, USA). Data are stated as mean ± SEM (standard error of mean). A student's t-test or ANOVA followed by Post Hoc Tukey's test were used for analyzing the cytotoxicity data. To determine the correlation between variables, a linear regression was carried out. Statistical significance was considered when p < 0.05.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.