Metabolomics Analysis and Antioxidant Potential of Endophytic Diaporthe fraxini ED2 Grown in Different Culture Media

Endophytic fungi are a promising source of bioactive metabolites with a wide range of pharmacological activities. In the present study, MS-based metabolomics was conducted to study the metabolomes variations of endophytic Diaporthe fraxini ED2 grown in different culture media. Total phenolic content (TPC), total flavonoid content (TFC), 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging, 2,2-azinobis(3-ethylbenzothiazoline-6-sulfonic acid (ABTS), and ferric reducing antioxidant power (FRAP) assays were conducted to assess the antioxidant potential of the fungal extracts. Multivariate data analysis (MVDA) was employed in data analysis and interpretation to elucidate the complex metabolite profile. The supplemented culture medium of D. fraxini fungal extract stimulated the production of metabolites not occurring in the normal culture medium. Antioxidant activity studies revealed the potential of supplemented cultured fungal extract of D. fraxini as a source of antioxidants. The present findings highlight that fungal culture medium supplementation is an effective approach to unravelling the hidden metabolome in plant-associated fungal diversity.


Introduction
The plant kingdom houses a diverse group of beneficial endophytic species. Endophytes inhabit host plant tissues without causing any apparent disease symptoms to them. This plant-endophyte relationship is mutualistic and important in plant micro-ecosystems. The host provides habitation and nutrients for endophytes to complete their life cycles. Meanwhile, endophytes enhance the host's ability to tolerate biotic and abiotic stress conditions by producing functional metabolites [1,2]. The literature has reported that each plant accommodates at least one endophyte [3]. Among the endophytes, endophytic fungi have attracted substantial attention as a potential source of biologically active metabolites. They produce metabolites with various biological activities, such as antioxidant, antimicrobial, antiviral, antidiabetic, anticancer, insecticidal, and immunosuppressive. These natural bioactive metabolites are useful for medicinal and pharmaceutical applications [4,5].
Research in endophytic fungi associated with medicinal plants has emerged as an exciting field in discovering novel bioactive metabolites. The literature has shown that The organic phase was collected and concentrated to dryness using a rotary evaporator under reduced pressure to yield the fungal extract. The concentrated extract was stored in an amber bottle in a freezer at −80 • C and lyophilized in a freeze dryer. The obtained extract was then stored in an amber bottle in a refrigerator at −20 • C until the analysis. Six individual culture media from each medium group were used as biological replicates to prepare the fungal extracts.

Total Phenolic Content (TPC)
TPC was performed according to the Folin-Ciocalteu method as described by Bobo-Garcia et al. (2015) [18]. In a flat-bottom 96-well microplate, 20 µL of 1 mg/mL extract was mixed with 100 µL of 1:4 diluted Folin-Ciocalteu reagent (R&M Chemicals, Subang, Malaysia) and gently shaken. After 2 min, 75 µL of 100 mg/mL Na 2 CO 3 (Bendosen, Shah Alam, Malaysia) solution was added and the mixture was briefly shaken. After 2 h of incubation at room temperature, the absorbance was measured at 750 nm. The calibration curve was constructed using gallic acid (Sigma-Aldrich, Burlington, MA, USA) at concentrations ranging from 0.2 to 400 µg/mL. TPC was estimated as gallic acid equivalent (GAE) in µg per mg of extract.

Total Flavonoid Content (TFC)
TFC was measured following Horszwald & Andlauer (2011) with slight modifications [19]. The AlCl 3 solution was prepared by mixing 10% (w/v) AlCl 3 .6H 2 O (Fisher Scientific, Waltham, MA, USA), 1 M sodium acetate (Sigma-Aldrich, Burlington, MA, USA), and deionised water in a 1:1:28 (v/v/v) ratio. A total of 100 µL of 1 mg/mL extract prepared in 95% ethanol were mixed with 150 µL freshly prepared AlCl 3 solution in a 96-well plate and incubated at room temperature for 30 min. The absorbance was then measured at 415 nm against the AlCl 3 solution as a blank. The standard curve was generated using quercetin (Sigma-Aldrich, Burlington, MA, USA) at concentrations ranging from 1 to 63 µg/mL in 95% ethanol. The results obtained were expressed as µg quercetin equivalent (QE) per mg of extract.

ABTS Cation-Radical Reduction Activity Assay
The ABTS cation-radical reduction activity of the samples was determined using the method described by Seo et al. (2015) [22]. The ABTS powder (Roche Life Science, Indianapolis, IN, USA) was first dissolved in deionised water to produce a 7 mM ABTS solution. The ABTS cation-radical was generated during 16 h reaction period with 2.45 mM potassium persulfate (Sigma-Aldrich, Burlington, MA, USA) in the dark at room temperature. Before the assay, the solution was diluted with deionised water to an absorbance of 0.7 at 734 nm. The ABTS cation-radical solution (100 µL) was then added to a 96-well plate containing the 100 µL of 0.005-5000 µg/mL test sample. Trolox (Acros Organics, Geel, Belgium) was used as a positive control, with a concentration of 0.06-31 µg/mL. After 5 min incubation, the absorbance was immediately measured at 734 nm using a microplate reader. The following equation was used to calculate the extract scavenging activity: ABTS scavenging activity (%) = abs 0 − abs extract abs 0 × 100 where abs 0 = absorbance of negative control abs extract = absorbance of extract The activity-concentration curve was evaluated using a plot of percentage ABTS scavenging activity against concentration. The concentration of extract required for 50% reduction in ABTS radical (IC 50 ) was calculated using the QuestGraph TM IC 50 Calculator. The Trolox equivalent antioxidant activity (TEAA) was calculated using IC 50 value (µg/mL) obtained from the QuestGraph TM IC 50 Calculator as follows [21]: where IC 50 T = IC 50 value of Trolox IC 50 E = IC 50 value of extract

Statistical Analysis
The results were expressed as the means ± standard deviation (SD) of six independent experiments (n = 6). The statistical significance of differences between means was established by an unpaired t-test using GraphPad Prism 9. p values < 0.05 were considered to indicate statistical significance.

LC-HRMS Metabolomic Analysis
Six biological fungal extracts from each DFC and DFS were analysed using an Agilent 1290 Infinity LC system coupled to Agilent 6520 Accurate-Mass Q-TOF mass spectrometer with a dual ESI source. One mg/mL of extract was analysed using an Agilent Zorbax Eclipse column (XDB-C18, Narrow-Bore 2.1 × 150 mm, 3.5 µm) in positive mode. The gradient elution was conducted at the 0.5 mL/min flow rate using purified water (A) and acetonitrile (B) with 0.1% formic acid in each mobile phase. The gradient program started with 5% B and increased gradually to 100% B. The total analysis period for each extract was 25 min. The injection volume was 1 µL, and the column temperature was maintained at 20 • C. HRMS analysis was performed in positive ESI ionisation modes coupled with a spray voltage at 4.0 kV; nitrogen gas was used as the drying gas at 320 • C with a flow rate of 10 L/min, nebuliser pressure: 45 psig, fragmentor voltage: 125 V, and mass range from 100 to 3200 m/z at a resolving power up to 20,000 (1 s acquisition). The obtained raw MS data files were converted to mzML format using ProteoWizard software (Palo Alto, CA, USA) and MS-DIAL version 4.8 for peak discrimination, filtering, and alignment. Dereplication and metabolite identification for the positive ionisation mode dataset were carried out using the METLIN and DNP databases. The level of identification was L2-putatively identified metabolites through library matching [23]. ChemDraw Professional 20.0 (PerkinElmer, Waltham, MA, USA) software was used for chemical structure drawing.

Multivariate Data Analysis (MVDA)
MetaboAnalyst 5.0 is employed to perform MVDA in the present study. It is a web-based metabolomics data processing platform for statistical, functional, and metaanalyses [24]. A data file (.csv) containing a table with the information of sample name, sample group, peak list, and peak intensity was uploaded onto MetaboAnalyst 5.0 server (https://www.metaboanalyst.ca/, accessed on 24 April 2022). Data were subjected to log transformation and Pareto scaling. Subsequently, univariate analysis was performed followed by multivariate analysis (principal component analysis (PCA), supervised partial least squares-discriminant analysis (PLS-DA), hierarchical clustering, and K-means partitional clustering) [25].

Effects of Culture Medium Supplementation on TPC and TFC
The effects of supplementation in the culture medium of D. fraxini on the TPC and TFC are shown in Figure 1. Overall, DFS exhibited higher TPC (215.63 µg GAE/mg extract) and TFC (2.74 µg QE/mg extract) than DFC (TPC: 36.98 µg GAE/mg extract; TFC: 2.62 µg QE/mg extract). It is known that phenolic and its derivatives are the major contributors to antioxidant activity by exhibiting free radical inhibition in biological systems. They are regarded as good electron donors due to the presence of aromatic hydroxyl groups that play a key role in scavenging free radicals [26]. According to a study conducted by Verma et al. (2022), Diaporthe sp. SAUCC194 extract has been shown to give a TPC of 78.91 µg GAE/mg extract. Thus, different endophytic fungal extracts exhibited different phenolic profiles that could affect their biological activities, particularly antioxidants [27]. According to the literature, flavonoids constituted a small portion of the TPC [28]. The obtained TFC was lower than other fungal extracts from Diaporthe sp. [29]. It was reported that fungal endophytes produced bioactive secondary metabolites to inhibit pathogen attacks and self-survival in their specific niches [7]. Thus, a rosmarinic acid-supplemented culture medium may have stimulated the production of related phenolic metabolites in D. fraxini, contributing to its high TPC.
obtained TFC was lower than other fungal extracts from Diaporthe sp. [29]. It was reported that fungal endophytes produced bioactive secondary metabolites to inhibit pathogen attacks and self-survival in their specific niches [7]. Thus, a rosmarinic acid-supplemented culture medium may have stimulated the production of related phenolic metabolites in D. fraxini, contributing to its high TPC.

Effects of Culture Medium Supplementation on DPPH and ABTS Radical Scavenging and FRAP Reducing Activities
In assessing the antioxidant potential of DFC and DFS, different antioxidant assays were employed. This is due to the fact that complex chemical interactions may occur within the sample. Thus, several assays are warranted to assess different modes of action of antioxidants within a biological system [30]. DPPH, FRAP, and ABTS assays are widely used to evaluate the antioxidant capacities of samples owing to their fast and reproducible results [31]. Generally, DFS showed higher antioxidant activity than DFC in all the assays (Table 1). DPPH assay measures the reducing or scavenging ability of the sample towards the free radicals using the spectrophotometric method. The colour change occurs when there is a reduction of an oxidan. Additionally, the degree of colour change is positively correlated to the concentration of antioxidants in the sample [32]. Based on the results, DFC and DFS recorded an activity of 9.71 ± 2.64 and 332.20 ± 51.07 µg AAE/mg extract, respectively. Additionally, IC50 values were obtained and compared with the standard, ascorbic acid. IC50 is the amount of sample required for 50% inhibition of a given biological activity. Thus, a lower IC50 value exhibited by the sample indicates higher biological activity [33]. The IC50 for DFC and DFS were 250.66 and 7.11 µg/mL, respectively, as compared to the standard (IC50: 2.31 µg/mL). In a study conducted by Rai et al. (2022), extracts from Diaporthe tulliensis and Diaporthe tectonendophytica showed an IC50 of >200 µg/mL when evaluated using a DPPH assay [34]. Weak activity (IC50: >200 µg/mL) was also recorded by fungal extract of Diaporthe sp. SAUCC194 isolated from a medicinal plant, Oroxylum indicum (L.) Kurz [26]. The present findings highlight the significant DPPH radical scavenging activity displayed by DFS when cultured in a supplemented medium. In FRAP assay, DFS recorded 188.41 ± 18.67 µg AAE/mg extract as compared to DFC (53.88 ± 4.31 µg AAE/mg extract). This assay is based on the reduction of Fe 3+ -TPTZ to give Fe 2+ -TPTZ complex by antioxidants that give an intense blue colour [32]. Meanwhile, the ABTS assay is one of the widely used antioxidant assays. It is also known as the TEAA assay. The stable radical cation ABTS will lose its blue-green colour when reacting with hydrogen-donating antioxidants. It absorbs at a wavelength of 734 nm, which has advantages over the elimination of colour interference and reduction in sample turbidity [35,36]. It is worth noting that DFS (1159.44 ± 67.70 µg TE/mg extract) showed an exceptionally higher antioxidant potential than DFC (37.77 ± 6.13 µg TE/mg extract). In addition, DFS (IC50: 17.83 µg/mL) exhibited potent IC50 as compared to DFC (IC50: 560.32 µg/mL) and the standard trolox (IC50: 20.61 µg/mL) in ABTS assay. Compared to fungal

Effects of Culture Medium Supplementation on DPPH and ABTS Radical Scavenging and FRAP Reducing Activities
In assessing the antioxidant potential of DFC and DFS, different antioxidant assays were employed. This is due to the fact that complex chemical interactions may occur within the sample. Thus, several assays are warranted to assess different modes of action of antioxidants within a biological system [30]. DPPH, FRAP, and ABTS assays are widely used to evaluate the antioxidant capacities of samples owing to their fast and reproducible results [31]. Generally, DFS showed higher antioxidant activity than DFC in all the assays (Table 1). DPPH assay measures the reducing or scavenging ability of the sample towards the free radicals using the spectrophotometric method. The colour change occurs when there is a reduction of an oxidant. Additionally, the degree of colour change is positively correlated to the concentration of antioxidants in the sample [32]. Based on the results, DFC and DFS recorded an activity of 9.71 ± 2.64 and 332.20 ± 51.07 µg AAE/mg extract, respectively. Additionally, IC 50 values were obtained and compared with the standard, ascorbic acid. IC 50 is the amount of sample required for 50% inhibition of a given biological activity. Thus, a lower IC 50 value exhibited by the sample indicates higher biological activity [33]. The IC 50 for DFC and DFS were 250.66 and 7.11 µg/mL, respectively, as compared to the standard (IC 50 : 2.31 µg/mL). In a study conducted by Rai et al. (2022), extracts from Diaporthe tulliensis and Diaporthe tectonendophytica showed an IC 50 of >200 µg/mL when evaluated using a DPPH assay [34]. Weak activity (IC 50 : >200 µg/mL) was also recorded by fungal extract of Diaporthe sp. SAUCC194 isolated from a medicinal plant, Oroxylum indicum (L.) Kurz [26]. The present findings highlight the significant DPPH radical scavenging activity displayed by DFS when cultured in a supplemented medium. In FRAP assay, DFS recorded 188.41 ± 18.67 µg AAE/mg extract as compared to DFC (53.88 ± 4.31 µg AAE/mg extract). This assay is based on the reduction of Fe 3+ -TPTZ to give Fe 2+ -TPTZ complex by antioxidants that give an intense blue colour [32]. Meanwhile, the ABTS assay is one of the widely used antioxidant assays. It is also known as the TEAA assay. The stable radical cation ABTS will lose its blue-green colour when reacting with hydrogen-donating antioxidants. It absorbs at a wavelength of 734 nm, which has advantages over the elimination of colour interference and reduction in sample turbidity [35,36]. It is worth noting that DFS (1159.44 ± 67.70 µg TE/mg extract) showed an exceptionally higher antioxidant potential than DFC (37.77 ± 6.13 µg TE/mg extract). In addition, DFS (IC 50 : 17.83 µg/mL) exhibited potent IC 50 as compared to DFC (IC 50 : 560.32 µg/mL) and the standard trolox (IC 50 : 20.61 µg/mL) in ABTS assay. Compared to fungal extracts from Diaporthe sp. isolated from mangrove plants which displayed IC 50 ranging from 0.77 to 13.56 mg/mL, the present findings suggested the potential of DFS as a source of natural antioxidants [37].

LC-HRMS-Based Metabolomics Analysis
The ethyl acetate extracts of D. fraxini grown in different media were analysed in positive ion mode using LC-HRMS. An untargeted metabolomics approach was conducted to characterize the metabolites present in the 12 extracts by considering the low molecular weight ionisable molecules. Six independent biological replicates were used for each culture medium to explore the metabolome differences between the fungal extracts. A total of 3164 features were detected in the 12 fungal extracts. The dataset from LC-HRMS analysis was subjected to MetaboAnalyst 5.0 to interpret and analyse the large metabolomics data generated from the samples. In unsupervised analysis, principal component analysis (PCA) was performed to distinguish the key differences between the sample groups. Generally, it is used to reduce many dimensionalities into important essential factors, thus providing an overview of the dataset [38]. Figure 2 shows the PCA pairwise score plots between the principal components (PCs). It was depicted that five PCs explained 78.8% of the total variation. In detail, PC1 and PC2 contributed to 59.6% of the total variation while PC1, PC2, and PC3 accounted for 67.8% of the total variation. Different culture media extracts were clearly distinguished based on the PCA 2D scores plot (Figure 3). The fungal extracts DFC and DFS were distributed into two distinct areas indicating these two extracts are statistically different from each other. It was observed that the six replicates from each DFC and DFS showed the scores closely, indicating the reproducibility of the fungal culture extracts. Moreover, the PCA was depicted in loading plot to distinguish the discrimination of the metabolites (Figure 4). The PCA analysis exhibited clear discrimination of the fungal extracts DFC and DFS, suggesting their unique patterns as shown in the heatmap ( Figure 5). In clustering analysis, it was distinctive that samples were grouped into two main clusters. Hierarchical cluster analysis (HCA) demonstrated a better distance matrix of the relationship between the culture extracts. HCA was found in accordance with the PCA results and allowed a better resolution of classification of the sample replicates [39]. HCA utilises a similarity metric between pairs of samples to give a dendrogram of nested clusters [40]. Based on the dendrogram (Figure 6), the vertical axis showed the arrangement of the cluster while the horizontal axis showed the clusters' similarity. In this aspect, each sample begins with its own cluster and similar clusters will then be merged as the hierarchy shifts to the left [41]. Thus, HCA further strengthened the PCA findings that DFC and DFS were well discriminated. According to previous literature, the metabolite variations were noticed among the different fungal cultures. Fungal interactions were believed to be the determining factor in contributing to the differences of the metabolite profiles [42]. In a study conducted by Tawfike and co-workers, different culture extracts of endophytic Curvularia sp. showed changes in their chemical profiles when examined using heatmap analysis. On different culture media, the occurrence of the metabolites was well discriminated, indicating chemical diversity of the studied fungal extracts [43].         In supervised analysis, partial least squares-discriminant analysis (PLS-DA) is performed to distinguish biological samples within-group variation from between-group variation. It is a widely used supervised technique that integrates the extracted features and discriminants into one algorithm [44]. The accuracy, correlation coefficient R2 as well as the cross-validation correlation coefficient Q2 of the dataset were more than 0.8, suggesting good predictability of the model. PLS-DA 2D scores and loading plots are shown in Figures 7 and 8. In PLS-DA scores plot, 56.8% of the total variations were explained by the two PLS components. The first and second components recorded 49.6% and 7.2%, respectively. In addition to that, PLS-DA established the 15 highest values of variable importance in projection (VIP) scores, as shown in Figure 9. The most 15 important features were shown on the vertical axis in an ascending order. These metabolites were dereplicated using METLIN and DNP to give 12 metabolites ( Figure 10) and three unknowns (

S) (2), 3-acetyl-4-hydroxy-6-methyl-2H-pyran-2-one (C 8 H 8 O 4 ) (3)
and N-methyl-14-O-demethylepiporphyroxine (C 20 H 21 NO 6 ) (4), respectively. Hexamethylquercetagetin was previously isolated as a new metabolite in Citrus plant [45], while thioquinolactobactin was found as a siderophore from Pseudomonas. This metabolite possessed significant antimicrobial activity with iron-chelating property [46]. 3-Acetyl-4-hydroxy-6methyl-2H-pyran-2-one, also known as methylacetopyronone, was detected in the chloroform extract of Solandra nitida [47]. Meanwhile, N-methyl-14-O-demethylepiporphyroxine is an alkaloid previously found in Papaver somniferum. This medicinal plant contains various alkaloids with potent pharmacological and bioactive properties [48]. Vermopyrone (6), an αpyrone, was first isolated from the fungus Gliocladium vermoesenii [49]. This metabolite was also present in Cephalotaxus hainanensis for anticancer activity [50]. Metabolite 2-amino-3-(3,4-dihydroxyphenyl)propanoic acid (7) is a catechol α-amino acid. Commonly, it is known as 3,4-dihydroxyphenylalanine and is used to treat Parkinson's disease [51]. A [M+H] + peak at m/z 200.0914 suggested the presence of α-amino-5-oxo-7-oxabicyclo[4.1.0]heptane-2-propanoic acid (anticapsin) (9). This metabolite was found to act as an inhibitor of glucosamine synthetase in Staphylococcus aureus [52]. Additionally, 12-decarboxy-4 ,5dihydromuscaaurin I (11) was reported as a pigment from Amanita muscaria [53], while 5-acetyl-2-hydroxybenzaldehyde (12) was found as a new p-hydroxyacetophenone derivative from Senecro graveolens [54]. Metabolite aculeatin A (13) was isolated as a novel dioxadispiro[5.1.5.2]pentadeca-9,12-dien-11-one derivative from Amomum aculeatum. It showed a potent cytotoxic effect against KB cells and anti-protozoal activities on Plasmodium strains [55]. Interestingly, metabolite toxicol B (14) was previously isolated as a novel structure from the extract of Toxiclona toxius, which showed activity on human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) [56]. 3-O-Demethyldehydroamorphigenin (15) was previously isolated as a bioactive phenolic from the fruits of Amorpha fruticosa. Nonetheless, the metabolite exhibited weak antibacterial and cytotoxic activities [57]. The literature has reported that the fungal metabolomes depend on various experimental parameters. To date, there are various techniques to conduct metabolomics analyses to identify the fungal metabolites from the extracts [58]. Plant-associated fungal endophytes could produce bioactive metabolites that may be used as therapeutic agents against several diseases. The possibilities of discovering novel bioactive metabolites are exciting, particularly in unknown structures. Thus, searching for novel chemical skeletons from endophytic fungi is essential for the sustainable production of desirable natural products [7]. In supervised analysis, partial least squares-discriminant analysis (PLS-DA) is performed to distinguish biological samples within-group variation from between-group variation. It is a widely used supervised technique that integrates the extracted features and discriminants into one algorithm [44]. The accuracy, correlation coefficient R2 as well as the cross-validation correlation coefficient Q2 of the dataset were more than 0.8, suggesting various experimental parameters. To date, there are various techniques to conduct metabolomics analyses to identify the fungal metabolites from the extracts [58]. Plantassociated fungal endophytes could produce bioactive metabolites that may be used as therapeutic agents against several diseases. The possibilities of discovering novel bioactive metabolites are exciting, particularly in unknown structures. Thus, searching for novel chemical skeletons from endophytic fungi is essential for the sustainable production of desirable natural products [7].

Conclusions
Plant-associated fungal research has developed over the past decades along with the advancement of new technologies. In this aspect, metabolomics is tremendously contributing to the generation of comprehensive information to uncover the metabolomes of a fungal biological system. It is an indispensable technique used to study the global fungal metabolites at a particular condition and/or state. Moreover, MVDA in a metabolomics study is crucial to modeling variances, and thus, to presenting an overview of the dataset. An untargeted metabolomics approach was performed by employing an LC-HRMS to characterize the fungal extracts of D. fraxini grown in different culture media. A supplemented culture medium of D. fraxini exhibited potent antioxidant activity when evaluated using DPPH, ABTS, and FRAP assays. Metabolites discrimination was generated using unsupervised analysis for visual representation and explorative study. Additionally, the application of supervised PLS-DA analysis allowed the extraction of important metabolites features, which contributed to the discrimination of the fungal culture media. These metabolite markers are warranted for targeted metabolomics profiling for specific culture media conditions. The present study offers an important reference to producing bioactive metabolites from fungal endophytes residing in medicinal plants.