Fractionation Coupled to Molecular Networking: Towards Identification of Anthelmintic Molecules in Terminalia leiocarpa (DC.) Baill

Terminalia leiocarpa is a medicinal plant widely used in ethnoveterinary medicine to treat digestive parasitosis whose extracts were shown to be active against gastrointestinal nematodes of domestic ruminants. The objective of our study was to identify compounds responsible for this activity. Column fractionation was performed, and the activity of the fractions was assessed in vitro on Haemonchus contortus and Caenorhabditis elegans as well as their cytotoxicity on WI38 fibroblasts. Two fractions were the most active on both nematode models and less cytotoxic. LC-MS/MS analysis and manual dereplication coupled to molecular networking allowed identification of the main compounds: ellagic acid and derivatives, gallic acid, astragalin, rutin, quinic acid, and fructose. Other potentially identified compounds such as shikimic acid, 2,3-(S)-hexahydroxydiphenoyl-D-glucose or an isomer, quercetin-3-O-(6-O-galloyl)-β-D-galactopyranoside or an isomer, and a trihydroxylated triterpenoid bearing a sugar as rosamultin are reported in this plant for the first time. Evaluation of the anthelmintic activity of the available major compounds showed that ellagic and gallic acids were the most effective in inhibiting the viability of C. elegans. Their quantification in fractions 8 and 9 indicated the presence of about 8.6 and 7.1 µg/mg ellagic acid and about 9.6 and 2.0 µg/mg gallic acid respectively. These concentrations are not sufficient to justify the activity observed. Ellagic acid derivatives and other compounds that were found to be positively correlated with the anthelmintic activity of the fractions may have additive or synergistic effects when combined, but other unidentified compounds could also be implicated in the observed activity.


Introduction
Terminalia leiocarpa (DC.) Baill (Combretaceae) (previously Anogeissus leiocarpus) is a 15-18 m tall tree found in India and Africa (especially in West and Central Africa). It is a very important tree because of its high use in traditional medicine, as wood, and in energy production. Indeed, T. leiocarpa is widely used by various communities to treat numerous ailments including cough, tuberculosis, diarrhea, dysentery, helminthiasis, malaria, trypanosomiasis, hemorrhoids, skin diseases, fever, and diabetes [1][2][3][4][5]. Several pharmacological studies have concluded that the plant has antibacterial, antioxidant, anthelmintic, and anti-tuberculosis properties [4,6,7].
The anthelmintic activity of T. leiocarpa has been already evaluated and the results obtained showed that its leaf extracts were very active on ruminant digestive parasites both in vitro and in vivo. Indeed, Kabore et al. [8] showed that the aqueous extract of T. leiocarpa leaves was very active in vitro on eggs, larvae, and adult worms of H. contortus, a digestive parasite of small ruminants. The IC 50 value of the extract was estimated to be 409.5 µg/mL for the inhibition of H. contortus egg hatching. Similarly, Ndjonka et al. [6] showed that the ethanol extract of T. leiocarpa leaves exhibited strong anthelmintic activity in vitro on C. elegans. These results were later confirmed by the work of Soro et al. [9] who showed that the ethanol extract of the plant roots was very effective in vivo (in sheep) on H. contortus and Trichostrongylus colubriformis at a concentration of 80 mg/kg orally. Furthermore, a screening of the in vitro anthelmintic activity of some of the most common plants used in ethnoveterinary medicine in Benin to treat digestive parasitosis of small ruminants on H. contortus larvae migration showed that the MeOH extract of T. leiocarpa leaves was one of the most active ones [10].
Although the interesting anthelmintic activity of T. leiocarpa has already been established [6,[8][9][10], little work has been conducted to identify the molecules responsible for this activity. Ndjonka et al. [11] linked the anthelmintic activity of the plant to phenolic acids including ellagic acid, gentisic acid, and gallic acid. On the other hand, the work of Waterman et al. [2] concluded that punicalagin was partly responsible for the anthelmintic activity of the aqueous extract of the leaves of T. leiocarpa. Indeed, the authors considered that the concentration of punicalagin in the extract was too low to justify the strong anthelmintic activity of the plant. There would therefore have to be other anthelmintic compounds that act in addition or synergy with punicalagin. Furthermore, Ademola and Eloff [1] concluded that the anthelmintic activity of T. leiocarpa was due to several different compounds, but did not identify them. Thus, in view of the pharmacological importance of this plant in the treatment of digestive parasitosis in ruminants, it appears necessary to identify the main compounds responsible for its anthelmintic activity. This is particularly relevant to identify new anthelmintic molecules in the context of the development of resistance against the synthetic anthelmintics currently used [12,13].
The identification of compounds in an extract is a difficult, tedious, and sometimes time-consuming and expensive task due to the complexity of some plant matrixes. In recent years, molecular networking, an organization and dereplication LC-MS/MS based technique has been developed. This technique allows rapid identification proposals of molecules and their visualization and organization into clusters based on the similarity between their MS/MS fragmentation [14][15][16]. Molecular networking is increasingly used for the tentative identification of natural substances by comparison of the experimental data with reference MS/MS fragmentation spectra [17,18].
In the present study, we combined fractionation of the MeOH extract of T. leiocarpa leaves with HPLC-PDA-HRMS/MS analysis and the use of molecular networking to identify the major compounds responsible for its anthelmintic activity. The anthelmintic activity was evaluated individually for identified major compounds commercially available, some of which were quantified by HPLC-PDA.

Cytotoxicity and Anthelminthic Activities of Fractions
We chose the MeOH extract of leaves of T. leiocarpa in view of its high anthelmintic activity and low cytotoxicity observed previously [10]. Open column chromatography fractionation of the extract yielded nine fractions labelled 1 to 9. Cytotoxicity of the fractions was evaluated on WI38 cells with the MTT assay. Fractions 1 to 6 showed mild to moderate toxicity with IC 50 values ranging from 58.9 to 78.9 µg/mL (Table 1). On the other hand, fractions 7, 8, and 9 were considered as not cytotoxic with an IC 50 greater than 100 µg/mL [18].
The anthelmintic activity of the fractions was evaluated on infective H. contortus larvae and young adult of C. elegans. These two nematode models are often used to evaluate anthelmintic activity and to identify anthelmintic molecules in natural substances [2,6,8]. Unfortunately, after the cytotoxicity evaluation, the remaining amount of fractions 1 and 3 was not sufficient to evaluate their anthelmintic activity. Nevertheless, these fractions were the two most cytotoxic on WI38 cells, after fraction 6, and would be less interesting to promote as anthelmintic.
On the other hand, fractions 8 and 9 exhibited strong anthelmintic activity (superior to that obtained for the MeOH extract of T. leiocarpa) on H. contortus and C. elegans (Table 1). In addition, fractions 5, 6, and 7 showed moderate anthelmintic activity with inhibition of larval migration ranging from 21.0 to 40.0% at a concentration of 600 µg/mL. The anthelmintic activity of these three fractions on young adult of C. elegans was in the same range as that observed on H. contortus, with approximately 20% inhibition of viability. These results suggest that anthelmintic compounds are more concentrated in fractions 8 and 9. Furthermore, the lower anthelmintic activity observed in fractions 5, 6, and 7 suggests that these fractions also contain anthelmintic compounds with lower activity or present in lower quantities. These results corroborate the work of Ademola and Eloff [1] who concluded that the anthelmintic activity of T. leiocarpa was due to several compounds with various polarities. In general, the anthelmintic activity of the different fractions on H. contortus is similar to that observed on C. elegans, the two nematodes sharing nearly 70% similarity [19].These results support the use of C. elegans in the identification of anthelmintic compounds instead of ruminant parasitic nematodes that are difficult to obtain and maintain in the laboratory [6,11]. Furthermore, the high anthelmintic activity observed in both nematode species and different physiological stages (larva and adult) suggests that the compounds responsible for anthelmintic activity in T. leiocarpa could be multitarget. Generally, compounds/extracts with anthelmintic properties show some specificity of action on nematode stages. For example, the water extract of Daniellia oliveri leaves was more active on eggs than larvae [8]. Similarly, levamisole is very active on larvae and adult worms but ineffective on eggs [20].
As fractions 8 and 9 were the least toxic and were more active than other fractions, we considered the identification of their major compounds.

Molecular Networking and Major Compounds in the Most Active Fractions
All fractions as well as the MeOH extract, were analyzed by HPLC-PDA-HRMS/MS in negative ion mode. The mass spectrometry data of all fractions (fraction 1 to 9) processed on MZmine 2.5.3 allowed the generation of a spectral alignment with 362 features. The molecular network was built with these 362 features on GNPS and is available on the link: http://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=a9c6688a76d54309817b5a61705 4b536 (accessed on 10 November 2022).
The constructed molecular network shows the metabolites present in fractions 8 (green nodes) and 9 (red nodes) compared to those observed in fractions 1-7 (blue nodes) ( Figure 1). We mainly focused on fractions 8 and 9 because of their stronger anthelmintic activity and lower cytotoxicity. The molecular network classified the detected metabolites into several clusters. In the molecular network presented in Figure 1, we identified triterpenic derivatives ( Figure 1A Figure 1H). This is in accordance with the previously performed phytochemical analyses which showed that T. leiocarpa contains mainly triterpenes, phenolic acids including ellagic acid and its derivatives, flavonoids, fatty acids, tannins, and sugars [2,4,7,21].
The dereplication performed on GNPS was completed by manual dereplication comparing HRMS/MS data to existing literature in order to identify the major compounds in fractions 8 and 9. Table 2 presents the mass spectrometry data of the major compounds identified putatively or confirmed with reference standards, in fractions 8 and 9.
Compounds ] -. So, these two phenolic acids were identified as quinic acid (3) and gallic acid (8). Their identification was confirmed by injection with the corresponding standards. Compounds (3) and (8) were previously identified in T. leiocarpa [2,3]. Compound (4) showed a deprotonated [M-H]ion at m/z 173.0456 and one of its major MS/MS fragments was observed at m/z 119.0353. A comparison of the MS/MS data of compound (4) with the literature identified it as shikimic acid [22]. To the best of our knowledge, this is the first time the presence of compound (4) has been reported in T. leiocarpa. Compound (19), eluting later, yielded a deprotonated molecular ion [M-H]at m/z 300.9978 that fragmented in MS/MS to give two main fragments at m/z 163.0398 and m/z 169.0144 (Table 2). Its correspondence to ellagic acid was confirmed by injection of a standard. Like compounds (3) and (8), ellagic acid (19) was previously identified in T. leiocarpa extracts [4,7]. It was one of the major compounds in fractions 8 and 9 (Figures 2 and 3, respectively) and can be visualized in the molecular network ( Figure 1E). The dereplication performed on GNPS was completed by manual dereplication comparing HRMS/MS data to existing literature in order to identify the major compounds in fractions 8 and 9. Table 2 presents the mass spectrometry data of the major compounds identified putatively or confirmed with reference standards, in fractions 8 and 9.
Compounds  (8). Their identification was confirmed by injection with the corresponding standards. Compounds (3) and (8) were previously identified in T. leiocarpa [2,3]. Compound (4) showed a deprotonated [M-H] -ion at m/z 173.0456 and one of its major MS/MS fragments was observed at m/z 119.0353. A comparison of the MS/MS data of compound (4) with the literature identified it as shikimic acid [22]. To the best of our knowledge, this is the first time the presence of compound (4) has been reported in T. leiocarpa. Compound (19), eluting later, yielded a deprotonated molecular ion [M-H] -at m/z 300.9978 that fragmented in MS/MS Compounds (9), (10), (11), (12), (13), (14), (15), and (16), grouped into cluster 1C ( Figure 1) were identified as derivatives of ellagic acid (19). Indeed, these compounds showed different molecular ions in the full scan spectrum and a main MS/MS fragment at m/z 300.9987, corresponding to the ellagic acid fragment ( Table 2) , which suggests that the ellagic acid moiety was potentially linked to a deoxyhexoside, a glucuronide, a dideoxyhexoside, or a gallate unit, respectively. However, the ∆ ppm obtained for these structural proposals were too high (sometimes >200). Isolation should allow their precise identification and characterization of these compounds will likely identify new compounds in T. leiocarpa. The presence of ellagic acid derivatives has long been suspected in some extracts of T. leiocarpa, in the genus Terminalia or other Combretaceae species [4,7,[23][24][25].
A total of five flavonoids were identified in both fractions. The first eluted at 25.  (18) but had a different retention time. We therefore concluded that these were isomers, and that compound (18) could correspond to quercetin-3-O-(6-O-galloyl)-β-D-galactopyranoside or an isomer. The latter was previously identified in Guiera senegalensis (Combretaceae) [3] and Tapirira guianensis (Anacardiaceae) [26]. To the best of our knowledge, this is the first time that compound (18) has been identified in T. leiocarpa. Mass spectrometry chromatograms showed that it is also one of the major compounds in fractions 8 and 9 (Figures 2 and 3). Compound (23) ([M-H]at m/z 447.0932) showed two major MS/MS fragments at m/z 285.0404 and m/z 284.0323. It was identified as astragalin [27]. The identification was confirmed by injection of the standard. The compound was previously identified in Pteleopsis suberosa (Combretaceae) [27]. To the best of our knowledge, this is the first time this compound has been identified in T. leiocarpa where it seems to be present as the major compound of the leaves. The molecular network organized the flavonoids detected into two major groups. Cluster H consists of glycosylated galloylated flavonoids (including compounds 18 and 22) and cluster B contained O-glycosylated flavonoids (including compounds 17, 20, and 23) ( Figure 1C,D).
Another major metabolite was eluted at 3.44 min and showed a signal in the full scan spectrum at m/z 481.0605 [M-H]with a high MS/MS fragment at m/z 331.0672. These spectral data are similar to those obtained by Fernandes et al. [28] who identified the compound as 2,3-(S)-hexahydroxydiphenoyl-D-glucose, or an isomer (5), a hydrolysable tannin. This compound was previously identified in Terminalia myriocarpa and Terminalia calamansanai [29,30]. To the best of our knowledge, this is the first time that compound (5) has been identified in T. leiocarpa. Combretaceae in general and species of the genus Terminalia in particular are well known for their high content of hydrolysable tannins [29,30]. Compound (5) is one of the major compounds in fraction 8 ( Figure 3) and was visualized in the molecular network ( Figure 1G).
Compound (29) is a trihydroxylated triterpene wearing a sugar, which could correspond to rosamultin, already identified in Terminalia alata [31]. Similarly, compounds (30) and (31) were putatively identified as fatty acids in comparison with literature data [32][33][34]. Compound (2) exhibited a deprotonated molecular ion [M-H]at m/z 179.0564 and was identified as fructose after injection of the standard, while compound 1 was tentatively identified as an hexitol.

Prediction of Anthelmintic Activity of Features Detected in T. leiocarpa Fractions by Pearson Correlation
Phytochemical analysis of the fractions of T. leiocarpa showed that they contain a variety of metabolites. Thus, in order to identify those potentially responsible for anthelmintic activity, we calculated the correlation coefficient between the intensity of the different metabolites within each fraction and the anthelmintic activity of the fractions on C. elegans. To achieve this, we added the anthelmintic activity on C. elegans of the nine fractions to the spectral alignment file containing the 362 features detected. The file was then exported to RStudio to calculate the Pearson correlation coefficient between the detected features and the anthelmintic activity. The calculated correlation coefficient along with the probability (p-value) was used to generate a three-value score (−1, 0 and +1) which was then imported into Cytoscape to visualize the features/compounds. A −1 signifies a significant (p < 0.05) negative correlation between the feature and anthelmintic activity on C. elegans. In other words, these compounds would exhibit antagonistic activity. On the other hand, a +1 means a significant (p < 0.05) positive correlation between the feature concerned and the anthelmintic activity on C. elegans. Compounds represented in the molecular network by green colored nodes are significantly positively correlated to the anthelmintic activity ( Figure 4). This means that these compounds could have anthelmintic activity and could partially account for the anthelmintic activity of the fractions on C. elegans. Those not significantly (p > 0.05) correlated with the anthelmintic activity of the fractions on C. elegans are represented in yellow.

Anthelmintic Activity of Major Compounds and Their Quantification
The anthelmintic activity of the major identified and commercially available compounds of fractions 8 and 9 was evaluated in order to confirm or not the predictions in Section 2.3 and to determine if these compounds could be responsible for the anthelmintic activity of the MeOH extract of T. leiocarpa. The results of the anthelmintic activity of available standards are presented in Figure 5.
In general, the different compounds inhibited the viability of young adults of C. ele- The results showed that 43 features (11.88%) were significantly positively correlated with anthelmintic activity compared to 318 (87.85%) that were not significantly correlated, and only one feature (0.27%) was significantly and negatively correlated with anthelmintic activity. The results showed that ellagic acid (19) and its derivatives (compound 9 for example) were more abundant in fractions 8 and 9, and positively correlated with the anthelmintic activity of the fractions (Figure 4C,E). They may explain at least in part the higher activity of fractions 8 and 9 as several of them are present in higher quantities in these fractions. Ellagic acid (19) is known to possess interesting anthelmintic activity on H. contortus and C. elegans [11,36]. Like ellagic acid (19), some flavonoids identified in fractions 8 and 9 were positively correlated with anthelmintic activity on C. elegans ( Figure 4B,D). These include compounds (17) and (20). Many studies have shown that flavonoids are endowed with anthelmintic activity on various nematodes [37][38][39]. Nevertheless, all flavonoids are not correlated with anthelminthic activities, as we observed that astragalin (23) which was quite abundant in the most active fractions did not possess a significant correlation with anthelmintic activity ( Figure 4B). The same was observed for the triterpenic derivative tentatively identified as rosamultin (29) ( Figure 4A). A significant positive correlation was also observed for compound (5) (a hydrolysable tannin). Previously conducted studies show that tannins are endowed with strong anthelmintic activity [38], but this activity may depend on the type and structure of the tannins that are present. A positive correlation was also observed for fructose (2), which is a common sugar that should not have anthelminthic activity, but whose polarity may be close to active compounds.
The results of the correlation between the detected metabolites and the anthelmintic activity on C. elegans of the fractions remain indicative and should be taken with caution. Indeed, antagonistic, additive, or synergistic activity are possible when compounds are in a mixture, and their activity in these extracts is not related to their activity when tested individually. These results are nevertheless a lead towards an identification of anthelmintic molecules in T. leiocarpa. The evaluation of the anthelmintic activity of each compound and several mixtures would allow the confirmation or not of the results of the correlation.

Anthelmintic Activity of Major Compounds and Their Quantification
The anthelmintic activity of the major identified and commercially available compounds of fractions 8 and 9 was evaluated in order to confirm or not the predictions in Section 2.3 and to determine if these compounds could be responsible for the anthelmintic activity of the MeOH extract of T. leiocarpa. The results of the anthelmintic activity of available standards are presented in Figure 5.
In general, the different compounds inhibited the viability of young adults of C. elegans. The anthelmintic activity varied with the compounds and concentrations tested. Ellagic acid (19) and gallic acid (8) were the most effective with a reduction in viability approaching 70% at 500 µM (compound 19: 151.1 µg/mL and compound 8: 85.06 µg/mL). These results confirm the positive correlation between the anthelmintic activity of the fractions and these compounds. Furthermore, studies conducted previously had concluded that compounds (8) and (19) have strong anthelmintic activity on H. contortus and C. elegans [11,36]. Astragalin (23) and rutin (17) moderately inhibited the viability of C. elegans at the highest concentration tested (1000 µM) (compound 17: 610.5 µg/mL and compound 23: 448.4 µg/mL) and the inhibition rate for both compounds was around 50% ( Figure 5). The moderate anthelmintic activity of compound (23) confirms the absence of significant correlation between the anthelmintic activity of the fractions and this feature ( Figure 4D). The anthelmintic activity of compound (23) was previously evaluated on Fasciolopsis buski, a parasitic trematode of pig [40]. To our knowledge, this is the first time that the anthelmintic activity of astragalin has been evaluated on C. elegans. Rutin (17) on the other hand showed low anthelmintic activity despite its strong anthelmintic activity prediction. Previous work also showed weak anthelmintic activity of compound (17) on H. contortus [37,38]. Like the other flavonoids, 2-O-galloylhyperin (an isomer of compound 18) moderately inhibited the viability of adult C. elegans worms at a concentration of 1000 µM (616.5 µg/mL). This activity seems low in view of the significant positive correlation between compound (18) (its isomer) and the anthelmintic activity of the fractions, but as the structures are different, we cannot draw a conclusion.
Previous work also showed weak anthelmintic activity of compound (17) on H. contortus [37,38]. Like the other flavonoids, 2-O-galloylhyperin (an isomer of compound 18) moderately inhibited the viability of adult C. elegans worms at a concentration of 1000 µM (616.5 µg/mL). This activity seems low in view of the significant positive correlation between compound (18) (its isomer) and the anthelmintic activity of the fractions, but as the structures are different, we cannot draw a conclusion. Furthermore, the major available identified compounds that showed highest anthelmintic activity were quantified in the most active fractions (fractions 8 and 9). Astragalin (23) which showed moderate anthelmintic activity was also quantified in the two most active fractions, as it could serve as an analytical marker, given its high concentration. As some ellagic acid derivatives were significantly positively correlated with anthelmintic activity (Figure 4A), we also quantified the ellagic acid derivatives in ellagic acid equivalents. The results are presented in Tables 3 and 4. The concentrations of ellagic acid (19) and gallic acid (8) in fraction 8 were estimated to be 8.6 ± 0.7 and 9.7 ± 0.8 µg/mg of the fraction (Table 3). These compounds were most concentrated in fraction 8 compared to astragalin (23) which accounted for only 0.8 ± 0.1 µg/mg of this fraction. Compounds (8) and (19) were more concentrated in fraction 8 while compound (23) was more concentrated in fraction 9 ( Table 3). The total concentrations of ellagic acid derivatives (9-16) were 2.1 and 6.0 µg of ellagic acid equivalents/mg fraction respectively in fractions 8 and 9 ( Table 4). The concentration of ellagic acid derivatives plus ellagic acid in fraction 8 (10.8 µg/mg of fraction) was lower than that obtained in fraction 9 (13.2 µg/mg of fraction).  Furthermore, the major available identified compounds that showed highest anthelmintic activity were quantified in the most active fractions (fractions 8 and 9). Astragalin (23) which showed moderate anthelmintic activity was also quantified in the two most active fractions, as it could serve as an analytical marker, given its high concentration. As some ellagic acid derivatives were significantly positively correlated with anthelmintic activity (Figure 4A), we also quantified the ellagic acid derivatives in ellagic acid equivalents. The results are presented in Tables 3 and 4. The concentrations of ellagic acid (19) and gallic acid (8) in fraction 8 were estimated to be 8.6 ± 0.7 and 9.7 ± 0.8 µg/mg of the fraction (Table 3). These compounds were most concentrated in fraction 8 compared to astragalin (23) which accounted for only 0.8 ± 0.1 µg/mg of this fraction. Compounds (8) and (19) were more concentrated in fraction 8 while compound (23) was more concentrated in fraction 9 ( Table 3). The total concentrations of ellagic acid derivatives (9-16) were 2.1 and 6.0 µg of ellagic acid equivalents/mg fraction respectively in fractions 8 and 9 ( Table 4). The concentration of ellagic acid derivatives plus ellagic acid in fraction 8 (10.8 µg/mg of fraction) was lower than that obtained in fraction 9 (13.2 µg/mg of fraction).   (14) --8.0 ± 0.1 0.8 ± 0.0 Ellagic derivative (15) 0.3 ± 0.5 0.1 ± 0.1 0.4 ± 1.0 <LOQ Ellagic derivative (16) 11.1 ± 0.6 1.1 ± 0.1 1.8 ± 0.2 0.2 ± 0.0 ID codes, F8 Fraction 8, F9 Fraction 9.

Plant Collection
Fresh leaves of T. leiocarpa were collected in North Benin (N'Dali municipality). The sample was authenticated at the National Herbarium of Benin (NHB), University of Abomey-Calavi, Benin (AAC 1504/HNB). The leaves were washed with water to remove dust and other contaminants before being dried in the laboratory at 25 • C for 2 weeks. The dry leaves were ground in a 0.5 mm diameter mill. The powder obtained was stored in hermetically sealed boxes at 25 • C in the laboratory.

Extraction Procedure
The extraction procedure has been described previously [10]. Briefly, 250 mL of hexane was added to 50 g of powder and the mixture was macerated on a shaker for 12 h. After filtration, a second 250 mL portion of hexane was used for a further 12 h maceration under shaking. The same procedure was repeated for dichloromethane (DCM) and MeOH on the same powder sample. The MeOH extract was evaporated with a rotavapor, weighed, transferred to labelled boxes, and stored at +4 • C.

Open Column Chromatography (OCC) Fractionation of T. leiocarpa MeOH Extract
A series with thin layer chromatography (TLC) was performed to identify the solvent system to be used for the fractionation of the MeOH extract of T. leiocarpa. Twenty grams of extract was solubilized in MeOH and added to 50 g of silica gel (0.063-0.2 mm), mixed, and evaporated. In parallel, a silica column was mounted (350 g silica gel in DCM in a glass column: 33 × 35 cm). Extract mixed with silica gel was deposited on the top of the mounted column and the different solvent systems were prepared to elute the column (Table 5). TLC (on silica gel and using the solvent system used to elute the column as mobile phase) was performed in parallel to pool the collected fractions. The plates were sprayed with sulfuric anisaldehyde solution [41] and heated. Sub-fractions were formed by mixing the fractions showing a similar TLC profile. They were dried and stored at +4 • C until use.

Fractions Cytotoxicity
Cytotoxicity of the fractions was evaluated on WI38 cells (non-cancerous human fibroblast cell line) using MTT-assay according to a procedure described in the literature [42]. They were solubilized in DMSO at a concentration of 20 mg/mL. Then 5000 cells per well were seeded overnight in 96-well plates in 180 µL of DMEM (Dulbecco's Modified Eagle's Medium supplemented with 10% inactivated fetal calf serum and 1% penicillin/streptomycin). Solubilized extract/fractions (20 mg/mL) were diluted with DMEM to give concentrations from 0.5 to 1000 µg/mL and 20 µL of each diluted solution was added to the seeded cells in each well. Final concentrations tested ranged from 0.05 to 100 µg/mL. After 72 h of incubation, medium was replaced by 100 µL of MTT ([3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide] tetrazolium salt) solution to measure the metabolic activity of the cells, an indicator of cell viability. MTT solution was prepared by dissolving 15 mg of MTT in 50 mL (5 mL of PBS and 45 mL of DMEM). After 45 min, MTT solution was replaced with an equal volume of DMSO, and absorbance was measured with a spectrophotometer (SpectraMax M3) at 570 and 620 nm to measure formazan formed by the reduction of MTT. The assay was repeated twice in duplicate.
3.6. Anthelmintic Activities of Fractions 3.6.1. Viability of C. elegans Adult Worms Treated with Fractions and Pure Compounds Anthelmintic activity of the fractions was evaluated on young adults of the wild type (N2) strain of C. elegans. The young adults were provided by the Laboratory of Neurophysiology (Neuroscience Institute, Université Libre de Bruxelles, Brussels, Belgium). Briefly, ten young C. elegans adults were manually transferred into each well of a 48-well plate containing 250 µL of M9 buffer solution. Two hundred and fifty microliters of each sample (at a concentration of 1200 µg/mL in M9 buffer) was added to the worms. The final concentration tested for each fraction was 600 µg/mL. This dose was shown to be very discriminating in recently published work [10] and could enable us to easily distinguish the most active fractions from the less active ones. Viability of young adults of C. elegans was measured under a binocular microscope after 24 h incubation. Worms that were elongated and immobile even after shaking were considered dead or non-viable [11]. Levamisole was used as a positive control at a concentration of 25 µM. Each treatment was tested in duplicate, and the assay was repeated twice.
Anthelmintic activity of the pure compounds was also evaluated on young adults of C. elegans. The assay was conducted in the same way as for the fractions but at three concentrations: 100, 500, and 1000 µM. Each treatment was tested in triplicate and the assay was repeated twice.

Larval Migration Inhibition Assay (LAMIA)
Anthelmintic activity of the fractions was also evaluated on infested larvae (L3) of H. contortus using the larval migration inhibition assay (LAMIA). The larvae were obtained by artificially infesting sheep with a pure strain of H. contortus provided by the Laboratoire d'Ethnopharmacologie et de Santé Animale (LESA), University of Abomey-Calavi. The droppings of the infested sheep were cultured in the laboratory for ten days and the larvae were collected by the Baermann device. The collected larvae were stored at +4 • C for three months before use. The methodology used for LAMIA was described in the literature [10,43]. In summary, larvae (1000 L3s/mL) were incubated at 25 • C with the fractions at a concentration of 600 µg/mL in phosphate-buffered saline (PBS) solution. After three hours of incubation, the larvae were washed by centrifugation (67× g) with PBS solution and deposited on inserts (20 µm diameter) for migration for 3 h. The inserts were previously deposited on canonical tubes, containing PBS solution. After 3 h, the larvae contained in the inserts were discarded and those that migrated into the canonical tubes under the inserts were recovered and counted under the microscope. The rate of inhibition of larval migration was calculated according to the following formula: where A is the rate of inhibition of larval migration, T is the total number of larvae deposited on the insert, and M is the number of larvae counted in the canonical tube.

HPLC-PDA-HRMS/MS Analysis
T. leiocarpa fractions as well as MeOH extract were analyzed by HPLC-PDA-HRMS/MS to identify major compounds. HPLC-PDA (Thermo Scientific Accela LC Systems) coupled with mass spectrometry (Thermo Scientific LTQ orbitrap XL, Bremen, Germany) constituted the system used for analysis. Instruments were controlled using Thermo Scientific Xcalibur X software. HPLC separation was performed on a Luna C18 column, 250 × 4.6 mm packed with 5 µm particles. The mobile phase consisted of water + 0.1% formic acid (A) and 100% acetonitrile (B). The gradient used for elution was as follows: 0-10 min, 95% A; 10-40 min, 95-40% A; 40-45 min, 40% A; 45-50 min, 40-95% A, and 50-55 min, 95% A. Samples to be analyzed (10 mg/mL) were solubilized in MeOH and 20 µL was injected per fraction. Standards were prepared at a concentration of 500 µg/mL and 20 µL was injected. HRMS/MS analyses were performed in APCI (atmospheric pressure chemical ionization) in positive and negative modes with the following input conditions for the negative mode: capillary temperature 250 • C; APCI vaporizer temperature 400 • C; sheath gas flow rate 20 a.u.; auxiliary gas flow rate 5 a.u. and sweep gas flow rate 5 a.u. For positive mode: capillary temperature 250 • C; APCI vaporizer temperature 400 • C; sheath gas flow 25 a.u.; auxiliary gas flow 25 a.u. and sweep gas flow 5 a.u.; discharge current of 5 µA; capillary voltage of 21 V, and tube lens voltage of 75 V. Chromatograms were recorded between 200 and 600 nm.

Data Processing on MZmine
Raw mass spectrometry data of the fractions as well as that of MeOH extract of T. leiocarpa were pre-processed in the MZmine software (version 2.5.3). We only worked with the negative data as they were more sensitive and more informative in comparison with the positive mode. In summary, an ion list was generated by setting the noise level to 1.5 ×10 5 and 1, respectively for MS1 and MS2. The ion list thus created was used to construct the chromatogram with the MZmine ADAP Chromatogram builder function. The minimum number of scans in the cluster was set to 5. The group intensity threshold and the highest minimum intensity were set to 1.5 × 10 5 . Deconvolution of the constituted ion list was performed using the wavelets (ADAP) algorithm. Deconvolution was performed by setting the main parameters as follows: S/N threshold: 6, SN estimator: intensity window S/N, minimum feature height: 120.000, coefficient/area threshold: 3, peak duration: 0.00 to 0.50, and RT wavelet range: 0.00 to 0.10. Isotopes were grouped using the "Isotope grouper" function and setting the m/z tolerance to 10 ppm and the RT tolerance to 0.3 min (absolute). Lists of deisotoped ions were aligned by setting the parameters at the same level as for isotope grouping (m/z tolerance: 10 ppm and RT tolerance: 0.3 min absolute). The list of aligned ions was filtered by removing duplicate peaks (m/z tolerance: 0.02 and RT tolerance: 0.4 min absolute) and using the "Feature list rows filter" function. The aligned list was deisotoped, gap-filled, and exported as a .csv and .mgf file for submission to GNPS-FBMN (Global Natural Product Social Molecular Networking-Feature Based Molecular Networking).

Dereplication on GNPS
Exported MZmine files as well as the raw mass spectrometry data of the fractions were sent to the GNPS platform using Win SCP software (version 5.21.2). The m/z tolerance for MS1 and MS2 was set to 0.02 Da by default. Molecular networking was created on the GNPS platform (http://gnps.ucsd.edu (accessed on 10 November 2022)), version 28.2. Nodes were filtered to have a cosine score greater than 0.7 and at least 6 matched peaks. Dereplication against GNPS libraries was set to a cosine score of 0.7, with at least 6 matching peaks. The molecular network was finally processed and visualized on Cytoscape (version 3.8.2).

Quantification of Major Compounds
The quantification of the major compounds from fractions 8 and 9 was performed using an HPLC-PDA system (Accela Thermo ScientificTM, Bremen, Germany) based on the UV absorbance of the compounds. The system was controlled using ChromQuest software (version 4.2.34). Separation of compounds was performed on a Luna C18 (250 × 4.6 mm, 5 µm particles). The mobile phase consisted of water + 0.1% formic acid (A) and 100% acetonitrile (B). The column was eluted in gradient mode: 0-10 min, 5-12% B; 10-20 min, 12-18% B; 20-45 min, 18-25% B; 45-46 min, 25-5% B; and 46-55 min, 5% B. This gradient is different from that used for LC-MS/MS analysis and was intended to allow for better separation of compounds to facilitate quantification. The quantification was performed with an injection of 20 µL and a flow rate of 700 µL/min. The standard compounds to be quantified were prepared at different concentrations in MeOH (HPLC grade) varying from 150 to 25 µg/mL for compounds (8) and (19) and 50 to 5 µg/mL for compound (23). Samples of fractions 8 and 9 were solubilized in MeOH (HPLC grade) at a concentration of 10 mg/mL. The PDA wavelength was set between 220 and 360 nm and the chromatogram was integrated at 254 nm. The assay was conducted in triplicate and repeated three times. The limit of detection (LOD) and limit of quantification (LOQ) were determined from the residual standard deviation (σ) of the regression curves and slopes (S), according to the following equations: LOD = 3.3 σ/S and LOQ = 10 σ/S [18].

Statistical Analysis
Means ± standard deviation of the rate of inhibition of larval migration (H. contortus) and viability of young adult worms (C. elegans) were calculated for each fraction and control tested. The Pearson correlation coefficient between the metabolite intensities and anthelmintic activity of the fractions and the MeOH extract of T. leiocarpa was determined by the methodology developed by Nothias et al. [44]. Analysis was performed on RStudio software (Version 1.4.1717). The calculated correlation coefficient was used to identify on the molecular network, the metabolites whose intensities are significantly correlated (p < 0.05) or not, with the anthelmintic activity of the fractions on C. elegans.

Conclusions
In summary, our study identified several compounds in the most active fractions on adult C. elegans worms. Several of these compounds had already been previously identified in T. leiocarpa. These include quinic acid (3), gallic acid (8), ellagic acid (19), and rutin (17). On the other hand, shikimic acid (4), 2,3-(S)-hexahydroxydiphenoyl-D-glucose (5), or an isomer, quercetin-3-O-(6-O-galloyl)-β-D-galactopyranoside (18) or an isomer, and a glycosylated trihydroxylated triterpene, as rosamultin (29) were identified for the first time in the plant as well as several ellagic acid derivatives. The results of the biological activity prediction showed that several of these compounds are significantly positively correlated with the anthelmintic activity of the fractions on C. elegans. Evaluation of the anthelmintic activity of the major available compounds identified showed that gallic acid (8) and ellagic acid (19) were the most active. The other compounds tested moderately inhibited the viability of C. elegans. These results suggest an additive/synergistic effect of the different compounds present but may also indicate that some active substances may not have been identified by our LC-MS method. Further studies could focus on the verification of this hypothesis as well as the isolation and characterization of the other compounds positively and significantly correlated with the anthelmintic activity of the fractions. The anthelmintic activity of these compounds could be evaluated as well as their mechanism of action.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the University of Abomey-Calavi, Abomey-Calavi, Benin (approval number 2019UAC352). All efforts were made to minimize animal suffering.

Informed Consent Statement: Not applicable.
Data Availability Statement: All data generated or analyzed during this study are included in this manuscript.