Differential Metabolomic Fingerprinting of the Crude Extracts of Three Asteraceae Species with Assessment of Their In Vitro Antioxidant and Enzyme-Inhibitory Activities Supported by In Silico Investigations

: The Asteraceae is a large family, rich in ornamental, economical, and medicinally valuable plants. The current study involves the analytical and pharmacological assessment of the methanolic extracts of three less investigated Asteraceae plants, namely Echinops ritro , Centaurea deﬂexa , and Tripleurospermum decipiens, obtained by three different extraction methodologies viz. maceration (MAC), ultrasound-assisted extraction (UAE), and homogenizer-assisted extraction (HAE). LC-MS-MS analysis of E. ritro , C. deﬂexa , and T. decipiens extracts led to the identiﬁcation of ca. 29, 20, and 33 metabolites, respectively, belonging to ﬂavonoids, phenolic acids, and fatty acids/amides. Although there were signiﬁcant differences in the quantitative metabolite proﬁles in the extracts of E. ritro and T. decipiens based on the used extraction method, no signiﬁcant variation was observed in the extracts of C. deﬂexa in the three implemented extraction techniques. The antioxidant activities of the nine extracts were assessed in vitro using six different assays viz. DPPH, ABTS, CUPRAC, FRAP, PDA, and metal chelation assay (MCA). The HAE/UAE


Introduction
Oxidative stress has been linked to several complications, such as cancer, diabetes, neurodegenerative, and cardiovascular disorders [1][2][3][4]. This is attributed to the imbalance between the production of reactive oxygen/nitrogen radicals and the internal antioxidant defense system [5]. Similarly, the inhibition of some enzymes has been reported to contribute to the healing process in many diseases. For example, α-amylase and glucosidase inhibitors can be used in the management of type 2 diabetes, acetyl cholinesterase inhibitors for the management of Alzheimer symptoms, tyrosine-kinase inhibitors in cancer, and tyrosinase inhibitors for skin whitening and skin diseases [6].
Nature remains a rich source for the discovery and development of new drug entities that can be used in the management of various health conditions [7][8][9][10][11][12]. Members of family Asteraceae have been reported to display antioxidant effects due to their rich content of polyphenolics, yet many factors affected their antioxidant power; among these are the geographical location and climatic conditions, as well as the extraction procedure and the type of solvent used [13,14]. They have been reported to inhibit the activity of several enzymes, such as acetylcholinesterase, tyrosinase, and α-amylases [15,16]. Many members of Asteraceae have not yet been investigated for their chemical composition and pharmacological values. Echinops ritro L. is an Asteraceae plant widely cultivated in Europe, North Africa, and Asia. In China, it is commonly used to stimulate milk secretion in traditional medicine. The plant is rich in quinoline alkaloids, polyacetylenes, sesquiterpenes, polyphenolics, and thiophenes [17]. Few studies were reported on its medicinal value, which focused on its antifungal and antimicrobial properties [18,19]. Centaurea deflexa, belonging to the largest genus of the Asteraceae, is mainly distributed in the Mediterranean region and is rich in secondary phytoconstituents, the most characteristic of which are the sesquiterpene lactones responsible for its documented anticancer activity [20][21][22]. Meanwhile, little is known about its antioxidant activity or its polyphenolic content. Tripleurospermum species are distributed in Europe, North America, North Africa, and Asia, with many species predominant in Turkey [23]. Their essential oils were reported to inhibit acetylcholinesterase activity in a concentration-dependent manner [24]; nevertheless, nothing has so far been reported on the pharmacological value of T. decipiens.
Therefore, we herein report on the LC-MS-MS metabolic profiling of the methanol extracts of three-little investigated or uninvestigated-plants belonging to the family Asteraceae. In vitro antioxidant activities are intensely studied using seven different assays and correlated with their polyphenolic and flavonoidal content. Moreover, the enzyme-inhibitory capacity of the extracts was evaluated against α-glucosidase, α-amylase, tyrosinase, and acetyl-and butyrylcholinesterases.

Plant Materials and Preparation of Extracts
The plant samples were collected in the city of Konya in the 2021 summer season (June) and location information is given below. The plants were confirmed by one co-author (Dr. Evren Yildiztugay) in Selcuk University and one voucher specimen was deposited in Selcuk University. The plant samples (aerial parts) were dried in the shade at room temperature for approximately one week. The samples were then pulverized using a mill and they were placed in a dark environment.
In the present study, three extraction methods (maceration (MAC), homogenizerassisted (HAE) and ultrasound-assisted (UAE)) were performed using methanol. The extraction procedures are summarized below. The solid-solvent ratio was 1/20 in all extraction methods.
Maceration (MAC): The plant materials (5 g) were stirred with 100 mL of methanol at room temperature for 24 h in a shaking device.
Ultrasound-assisted extraction (UAE): The plant materials (5 g) were extracted with 100 mL of methanol in one ultrasound bath at room temperature for 30 min.
After the extraction procedures, all extracts were filtered using Whatman No.1 filter paper in Büchner flask under vacuum. The solvents were removed using rotary evaporator. All extracts were stored at 4 • C until analysis.

HPLC-ESI-MS/MS Analysis of the Methanol Extracts of Three Asteraceae Species
The extracts of Echinops ritro, Centaurea deflexa, and Tripleurospermum decipiens obtained by three extraction techniques were analyzed using high-performance liquid chromatography coupled to electrospray ionization mass spectrometry for investigating their metabolic profiles. Analysis was conducted on Shimadzu ® 8045 HPLC-ESI-MS/MS using a C 18 reversed phase column (Shimpack UPLC-2.7 µm, 2 × 150 mm). Negative and positive ion acquisition modes were implemented using a triple quadrupole mass analyzer, Shimadzu ® Corporation. Samples were dissolved in HPLC-grade methanol and filtered using a PTFE membrane (0.2 µm). MS-grade mobile phases were used as follows: A: water with 0.1% formic acid (v/v) and B: methanol with 0.1% formic acid (v/v). The elution profile was 0-2 min, 10% B (isocratic); 2-5 min, 10-30% B in A; 5-15 min, 30-70% B in A; 15-22, 70-80% B in A; 22-26, 80% B in A (isocratic); 29-30 min, 80-10% B in A; 30-35 min, 10% B in A (isocratic) with a flow rate of 0.2 mL/min. Mass detection was performed in a mass range over m/z 100-1200. The temperature of the ion source was adjusted to 200 • C, capillary voltage 3000 eV, desolvation and interface temperatures were set to 526 • C and 300 • C, respectively. Cone gas flow was 50 L/h, while the nebulizing gas flow was 3 L/min. For collision-induced dissociation (CID) MS/MS measurements were performed. The cone voltage for fragmentation was adjusted for each mass peak in a range from 10 to 40 eV. Data were processed using lab solutions software.

Enzyme Inhibitory Assays
The enzyme inhibitory assays were carried out according to previously reported methodologies [25,26]. The acetylcholinesterase (AChE) and butyrylcholinesterase (BChE) inhibition were expressed as mg galanthamine equivalents (GALAE)/g extract; tyrosinase inhibition was expressed as mg kojic acid equivalents KAE/g extract; amylase and glucosidase inhibition were expressed as mmol acarbose equivalents (ACAE)/g extract.
Docking grid and parameter files were generated using the binding coordinates of each ligand in its respective crystal structure in AutodockTools (https://autodock.scripts.edu, accessed on 3 June 2022) [32]. Autodock 4.2 s Lamarckian genetic algorithm was used to generate distinct ligand conformers and docked to the active site of each protein. Multiple ligand poses with different binding energies were returned and the ligand pose with the lowest binding energy was examined for reasonable binding pose using Biovia DS Visualizer.

Statistical Analysis
Firstly, for each species, one-way analysis of variance followed by Tukey's post hoc test were used to assess significant differences between the extracts in terms of their antioxidant and enzyme inhibitory activity (p < 0.05). In addition, the relationship between bioactive compounds and antioxidant activities as well as enzyme inhibitory activities was assessed by calculating Pearson correlation coefficient. Pearson's coefficients greater than 0.7 were considered significant. Afterwards, principal component analysis (PCA) following by clustered Image Maps were achieved to compare the biological activities of the three species samples. The statistical analysis was conducted using R software v. 4.1.2

ESI-MS-MS Fingerprinting of the Three Asteraceae Crude Extracts under Three Different Extraction Methods
Metabolic profiling of E. ritro, C. deflexa, and T. decipens extracts obtained by three extraction techniques viz. homogenizer-assisted extraction (HAE), maceration (MAC), and ultrasonic-assisted extraction (UAE) was performed using HPLC-ESI-MS/MS. A wide range of diversification was observed between the three species due to their different taxonomical backgrounds as well as the implementation of different extraction techniques. The identification of the compounds was based on their mass data, their characteristic MS 2 fragments, their UV, and by comparison with previously reported compounds in the literature. As can be noticed in Table 1, twenty compounds with different concentrations were identified in the extracts of E. ritro, depending on the extraction method. The majority of these were phenolic acids, fatty acids, and flavonoids. Comparative analysis of the percentage of the detected constituents using the three different extraction methods showed that some techniques improved the extraction of certain metabolites in higher yield compared to others. Caffeic and chlorogenic acids extracted by HAE were almost double and 1.5-times the amount of those extracted by MAC or UAE, respectively. Similarly, quercetin-O-hexoside extracted by HAE had a yield 18-times more than that obtained by UAE or MAC. Fatty acid amides were better extracted by HAE; however, they were never detected in the macerated extract. Ultrasonic-assisted extraction (UAE) was the best methodology for the extraction of dicaffeoylquinic acid, trihydroxy-octadecenoic acid, apigenin-O-hexouronide, and apigenin-p-coumaroyl-hexoside isomer, while MAC was better in the extraction of protocatechuic acid hexoside, shimobashiraside C, betulinic acid, and the amine derivative N ,N ,N -tris-p-coumaroyl spermidine. Surprisingly, naringenincoumaroyl hexoside was only detected in Echinops extract prepared by MAC and betulinic acid was not observed in the extract obtained by HAE.
The chemical composition of Centaurea deflexa is summarized in Table 2. Results revealed the presence of 29 compounds belonging to flavonoids (free or as glycosides), fatty acids and their amides, and organic and phenolic acid derivatives. In contrast to Echinops, Processes 2022, 10, 1911 5 of 22 there is no significant quantitative variation in the metabolites extracted by the three applied techniques, except for luteolin-O-hexoside, whose quantity was almost doubled in MAC if compared to HAE. Some compounds could be extracted by one method but not with the other(s). One example is the flavonoid eupatorine, which was only successfully extracted by applying MAC. Similarly, salvigenin was only extracted and in high yield (ca. 9%) by MAC. On the other hand, MAC was not successful in extracting dihydroxyoctadecadienoic acid. Additionally, hispidulin could only be extracted by HAE, as coumaroyl quinic acid and octadecadienoic acid were not detected in the homogenizer-assisted extract. Indeed, the percentage of the major compounds of HAE, MAC, and UAE extracts were, respectively, 10.03, 12.26, and 11.28% for oleamide, 10.28, 9.00, and 10.18% for caffeoyl hexoside, 8.86, 7.90, and 7.77% for arctiin, 5.72, 5.66, and 4.98% for apigenin-di-C-hexoside (vicenin-2), and 5.41, 4.70, and 4.81% for chlorogenic acid/neochlorogenic acid. Our results are in concordance with other published studies, which showed the richness of C. deflexa extracts through different phenolics compounds [20].
Tripleurospermum decipens extract showed the presence of 33 compounds with predominance in flavonoids, phenolic acids, and fatty acids ( Table 3). Different yields of the secondary metabolites were detected using the three extraction methods. Tartaric acid, cirsimaritin, and isorhamnetin-O-hexouronoide were only detected in UAE, while caffeoyl hexoside, naringenin-coumaroyl-hexoside, and syringic acid were only observed in HAE. Although isorhamnetin-O-hexouronide was not detected in extracts obtained by MAC or by HAE, its aglycone was detected by applying these techniques. MAC was the only effective method in extracting the triterpene compound butanoyl botulinic acid. MAC, likewise, improved the extraction yield of medioresinol by 1.2-fold compared to HAE and 1.8-fold compared to UAE. Fertaric acid was best extracted by MAC, showing double and triple the yields obtained by HAE and UAE, respectively. Most fatty acids in T. decipens were obtained in better yields if extracted by UAE; however, the lowest quantities were observed if HAE was implemented. To the best of our knowledge, our study is the first to report the chemical profile of T. decipens extract. However, previously published works studied its volatile compounds, reporting the presence of terpenoids [23,33].
Our results showed the impact of the extraction technique on the chemical composition of the three studied Asteraceae species. The extraction methods can influence the nature of the compounds identified as well as their percentage. In general, HAE and UAE are invasive techniques that result in cellular membrane disruption, thus, showing leakage of the metabolites and, consequently, an improvement in the phytochemical yield. Therefore, many metabolites are better extracted by HAE and UAE. In MAC, the extraction efficacy depends on the passive diffusion of the metabolites outside the plant cell, which might also be affected by the molecule size compared to the pore size between the membrane. For example, quercetin-O-hexoside yield by HAE and UAE was compared to MAC; however, HAE was better than UAE, which might indicate that the compound underwent chemical degradation after applying ultrasonic waves, hence, decreasing its yield dramatically.

In Vitro Assessment of the Antioxidant Activities in the Extracts
With the aim of highlighting the antioxidant properties in the extracts of three plants, six antioxidant tests were carried out in vitro. Firstly, the antioxidant activities in each Asteraceae species were discussed individually. As shown in Figure 1, all extracts showed significant antioxidant effects with variability related to the nature of the extracts.   [72][73][74]. The antioxidant activities of plant extracts can be attributed to their major bioactive compounds. Indeed, for E. nitro, DPPH showed positive and significant correlation with En23 and En24 (Figure 2). Similarly, a strong positive correlation was found between ABTS, CUPRAC, and FRAP and En4, En5, En7, En11, and En25. ABTS activity was also bound to En23 and En24. In addition, MCA was significantly linked to En6, En10, En14, En17, En20, and En21, whereas PBD was positively correlated with En16 and En22. Regarding C. deflexa, a positive significant correlation was observed between DPPH and Cd1, Cd3, Cd4, Cd5, Cd9, Cd12, Cd15, Cd18, Cd19, Cd21, Cd23, and Cd28. Furthermore, a significant positive Pearson coefficient was obtained between ABTS and CUPRAC, FRAP, MCA, and PBD and Cd6, Cd13, and Cd20.
Main identified substances, such as oleamide (found in E. ritro and C. deflexa extracts), showed antioxidant properties according to some reported pharmacological studies [75,76]. Moreover, chlorogenic acid, present in plant extracts, also demonstrated interesting remarkable in vitro and in vivo activities by several investigations [77,78].

Enzyme Inhibitory Effects
Inhibition of carbohydrate and glyceride-hydrolyzing enzymes is a promising therapeutic strategy in the management of type 2 diabetes mellitus (T2DM). In our case, we evaluated, in vitro, the inhibitory capacity of EOs, obtained from different phenological stages, on the catalytic activity of α-glucosidase and α-amylase. The results showed that extracts obtained from C. deflexa using MAC and UAE methods showed important inhibition of α-amylase, with inhibitory values of 0.30 ± 0.01 and 0.28 ± 0.01 mmol ACAE/g for MAC and UAE, respectively. However, the extract obtained from T. decipens using the UAE and HAE methods revealed the highest inhibitory value of α-glucosidase (0.98 ± 0.01 and 0.91 ± 0.01 mmol ACAE/g). Interestingly, all the extracts of E. nitro showed the same inhibitory effects on α-amylase and α-glucosidase. Similarly, T. decipens samples exhibited the inhibitory effects on α-amylase, while those of C. deflexa had the same inhibitory effects on α-glucosidase.
On the other hand, skin aging is a natural process related to endogenous (metabolic, cellular, and hormonal processes) and exogenous (chronic exposure to pollutants, toxic chemicals, ionizing radiation, etc.) factors that cumulatively damage skin appearance and physiology [79,80]. In our study, the evaluation of the dermatoprotective activity of plant extracts was carried out by the inhibitory effect on tyrosinase, an enzyme activating the oxidation of tyrosine, leading to melanin secretion. As can been seen in Table 5, the extract obtained by MAC and HAE methods from E. ritro showed the important inhibitory value (62.19 ± 0.38 and 62.28 ± 0.59 mg KAE/g) of tyrosinase. No significant difference was observed between the tyrosinase inhibition potentiality of the extract obtained from C. deflexa ( Figure 3). Regarding T. decipens, the extract obtained from UAE showed the highest anti-tyrosinase effect, followed by the extract obtained from MAC and HAE.

Enzyme Inhibitory Effects
Inhibition of carbohydrate and glyceride-hydrolyzing enzymes is a promising therapeutic strategy in the management of type 2 diabetes mellitus (T2DM). In our case, we evaluated, in vitro, the inhibitory capacity of EOs, obtained from different phenological stages, on the catalytic activity of α-glucosidase and α-amylase. The results showed that extracts obtained from C. deflexa using MAC and UAE methods showed important inhibition of α-amylase, with inhibitory values of 0.30 ± 0.01 and 0.28 ± 0.01 mmol ACAE/g for MAC and UAE, respectively. However, the extract obtained from T. decipens using the UAE and HAE methods revealed the highest inhibitory value of α-glucosidase (0.98 ± 0.01 and 0.91 ± 0.01 mmol ACAE/g). Interestingly, all the extracts of E. nitro showed the same inhibitory effects on α-amylase and α-glucosidase. Similarly, T. decipens samples exhibited the inhibitory effects on α-amylase, while those of C. deflexa had the same inhibitory effects on α-glucosidase.
On the other hand, skin aging is a natural process related to endogenous (metabolic, cellular, and hormonal processes) and exogenous (chronic exposure to pollutants, toxic chemicals, ionizing radiation, etc.) factors that cumulatively damage skin appearance and physiology [79,80]. In our study, the evaluation of the dermatoprotective activity of plant extracts was carried out by the inhibitory effect on tyrosinase, an enzyme activating the oxidation of tyrosine, leading to melanin secretion. As can been seen in Table 5, the extract obtained by MAC and HAE methods from E. ritro showed the important inhibitory value (62.19 ± 0.38 and 62.28 ± 0.59 mg KAE/g) of tyrosinase. No significant difference was observed between the tyrosinase inhibition potentiality of the extract obtained from C. deflexa (Figure 3). Regarding T. decipens, the extract obtained from UAE showed the highest anti-tyrosinase effect, followed by the extract obtained from MAC and HAE.    On the other hand, hyperactivities of acetylcholinesterase enzymes significantly impact memory functions and can lead, with other risk factors, to Alzheimer's disease. In this regard, natural inhibitors that can reduce or minimize catalytic activities of acetylcholinesterase lead to an increase in acetylcholine levels in the synapses and may, therefore, improve memory function in Alzheimer's patients. In our study, the HAE extract from E. ritro suppressed AChE levels at inhibitory values of 2.41 ± 0.04 mg GALAE/g. In addition, HAE and MAC extracts from E. nitro demonstrated stronger galantamine equivalent values against BChE (HAE = 0.80 ± 0.10 and MAC = 0.87 ± 0.11 mg GALAE/g). Concerning C. deflexa, UAE and HAE extracts revealed the highest anti-AChE activity (UAE = 2.25 ± 0.02; HAE = 2.27 ± 0.01 mg GALAE/g); nonetheless, all extracts exhibited the same anti-BChE activity. As for T. decipens, among the three extracts, HAE and MAC had the greatest capacity to inhibit AChE and BChE, respectively, with a value of 2.46 ± 0.01 and 1.96 ± 0.18 mg GALAE/g, respectively (Figure 3). Furthermore, the observed bioactivity may be correlated to the high levels of the numerous bioactive compounds present in the extracts. Indeed, examination of Pearson correlation coefficient in Figure 4 highlighted the presence of significant correlation between the evaluated bioactivities and the bioactive compounds. As an example, concerning E. nitro, significant correlation was found between AChE and En4, En5, En7, En11, and En25. Similarly, amylase was bound to En4, En5, En7, En11, En23, En24, and En25, while glucosidase was correlated to En16 and En22. Regarding C. deflexa, a strong positive correlation was found between AChE and Cd1, Cd3, Cd4, Cd5, Cd9, Cd12, Cd15, Cd19, Cd21, Cd23, and Cd28. BChE was positively linked to Cd1, Cd4, Cd12, Cd14, Cd18, Cd21, and Cd23. Further, anti-tyrosinase activity was significantly correlated to Cd19, Cd26, and Cd28. In addition, significant positive correlation was obtained between α-amylase inhibition and Cd2, Cd6, Cd13, Cd16, Cd20, Cd22, Cd24, Cd27, Cd29, and between α-glucosidase and Cd7, Cd8, Cd10, Cd11, Cd17, and Cd25.
Processes 2022, 10, x FOR PEER REVIEW 16 of 24 mg GALAE/g, respectively ( Figure 3). Furthermore, the observed bioactivity may be correlated to the high levels of the numerous bioactive compounds present in the extracts. Indeed, examination of Pearson correlation coefficient in Figure 4 highlighted the presence of significant correlation between the evaluated bioactivities and the bioactive compounds. As an example, concerning E. nitro, significant correlation was found between AChE and En4, En5, En7, En11, and En25. Similarly, amylase was bound to En4, En5, En7, En11, En23, En24, and En25, while glucosidase was correlated to En16 and En22. Regarding C. deflexa, a strong positive correlation was found between AChE and Cd1, Cd3, Cd4, Cd5, Cd9, Cd12, Cd15, Cd19, Cd21, Cd23, and Cd28. BChE was positively linked to Cd1, Cd4, Cd12, Cd14, Cd18, Cd21, and Cd23. Further, anti-tyrosinase activity was significantly correlated to Cd19, Cd26, and Cd28. In addition, significant positive correlation was obtained between α-amylase inhibition and Cd2, Cd6, Cd13, Cd16, Cd20, Cd22, Cd24, Cd27, Cd29, and between α-glucosidase and Cd7, Cd8, Cd10, Cd11, Cd17, and Cd25. To our knowledge, there are few published data concerning the enzymatic inhibitory effects of the studied plants, which makes our investigation the first original work. It was reported in the literature that enzyme inhibitory effects of T. decipiens were investigated over a long time by Göger, et al. [81]. However, other species in Echinops, Centaurea, and Tripleurospermum genus exhibited different biological and pharmacological activities, including antidiabetic and dermatoprotective effects, as well as neuroprotective properties [82,83]. Different studies have proved the biological effects of chlorogenic acid, including its antidiabetic effects [77,84]. To our knowledge, there are few published data concerning the enzymatic inhibitory effects of the studied plants, which makes our investigation the first original work. It was reported in the literature that enzyme inhibitory effects of T. decipiens were investigated over a long time by Göger, et al. [81]. However, other species in Echinops, Centaurea, and Tripleurospermum genus exhibited different biological and pharmacological activities, including antidiabetic and dermatoprotective effects, as well as neuroprotective properties [82,83]. Different studies have proved the biological effects of chlorogenic acid, including its antidiabetic effects [77,84].

Comparison of the Biological Activities of the Three Species Samples
The principal component analysis was conducted to compare the extracts of the three Asteraceae species in terms of their antioxidant and enzyme inhibitory activities. A summary of variability and eigenvalues obtained from PCA is provided in Figure 5. The first three components captured approximatively 82% of the variability and were the only components for which the eigenvalues were higher than 1. These three components were linked with a variable degree covariance to the following group of bioactivities: (1) DPPH, ABTS, FRAP, CUPRAC, BChE, and MCA, (2) Amylase, PBD, and tyrosinase, (3) AChE. Figure 5 depicts the distribution of samples on the three scatter plots, derived from PC1, PC2, and PC3. In the first two scatter plots, the samples of E. nitro was separated from those of C. deflexa and T. decipiens, along the first component. In addition, in the first scatter plot, the extracts obtained from C. deflexa using HAE were separated from the other samples. Similarly, in the second scatter plot, the extracts derived from T. decipens using HAE were removed from the remaining samples. These two tendencies were also observed in the third scatter plot. These observations suggested the existence of two main clusters as well as some subgroups in one of the two main groups. For better visualization of the clusters, a heatmap was produced for the result of the PCA, in consideration of the retained components. As can be seen in Figure 6, the extracts were split into two mains clusters. Cluster A comprised the extracts obtained from E. nitro using HAE and UAE. Both samples showed the highest antioxidant and anti-BChE activities. Cluster B, enclosing the remaining samples, can be divided into four sub-clusters, namely B1, B2, B3, and B5. Cluster B1 contained T. decipens-HAE and was characterized by the strongest anti-AChE activity. Cluster B4 comprised T. decipens-UAE and T. decipens-MCA extracts. Both exhibited remarkably anti-tyrosinase, anti-BChE, and anti-amylase activities, compared to other samples.

Comparison of the Biological Activities of the Three Species Samples
The principal component analysis was conducted to compare the extracts of the three Asteraceae species in terms of their antioxidant and enzyme inhibitory activities. A summary of variability and eigenvalues obtained from PCA is provided in Figure 5. The first three components captured approximatively 82% of the variability and were the only components for which the eigenvalues were higher than 1. These three components were linked with a variable degree covariance to the following group of bioactivities: (1) DPPH, ABTS, FRAP, CUPRAC, BChE, and MCA, (2) Amylase, PBD, and tyrosinase, (3) AChE. Figure 5 depicts the distribution of samples on the three scatter plots, derived from PC1, PC2, and PC3. In the first two scatter plots, the samples of E. nitro was separated from those of C. deflexa and T. decipiens, along the first component. In addition, in the first scatter plot, the extracts obtained from C. deflexa using HAE were separated from the other samples. Similarly, in the second scatter plot, the extracts derived from T. decipens using HAE were removed from the remaining samples. These two tendencies were also observed in the third scatter plot. These observations suggested the existence of two main clusters as well as some subgroups in one of the two main groups. For better visualization of the clusters, a heatmap was produced for the result of the PCA, in consideration of the retained components. As can be seen in Figure 6, the extracts were split into two mains clusters. Cluster A comprised the extracts obtained from E. nitro using HAE and UAE. Both samples showed the highest antioxidant and anti-BChE activities. Cluster B, enclosing the remaining samples, can be divided into four sub-clusters, namely B1, B2, B3, and B5. Cluster B1 contained T. decipens-HAE and was characterized by the strongest anti-AChE activity. Cluster B4 comprised T. decipens-UAE and T. decipens-MCA extracts. Both exhibited remarkably anti-tyrosinase, anti-BChE, and anti-amylase activities, compared to other samples.

Molecular Docking
Compounds accounting for ≥5% of the total bioactive compound content were further studied using molecular docking to estimate their binding strength and to predict their binding mode to each of the five enzymes (AChE, BChE, tyrosinase, amylase, and glucosidase). These compounds show the potential to bind to all enzymes, as suggested by their binding energy scores (Table 6). Furthermore, the majority of compounds displayed binding preference for AChE, BChE, and glucosidase. For instance, apigenin-di-Chexoside (Vicenin-2) and arctiin bound strongly to the AChE and, to a lesser extent, to BChE, moderately to the amylase and glucosidase, but modestly to the tyrosinase. On the other hand, chlorogenic acid and dicaffeoylquinic acid preferentially bound to glucosidase. Hence, we visualized protein-ligand interaction details for these compounds to examine the interaction patterns.
The major contributors to the interaction in all the docking complexes are H-bonds and π-π interactions formed between hydroxyl groups and aromatic rings on the ligands and the residues in the active site of the target enzymes (Figure 7). In addition, a few hydrophobic contacts and several van der Waals interactions increased the binding strength. Apigenin-di-C-hexoside (Vicenin-2) spanned the cavity of AChE by forming multiple Hbonds with polar amino acid residues at the entrance to and deep inside the tunnel ( Figure  7A). Arctiin bound strongly to BchE, mainly via π-π interactions and a couple of van der Waals interactions ( Figure 7B). Amylase formed multiple H-bonds, a couple of π-π interactions, and van der Waals interactions throughout the amylase catalytic channel ( Figure  7C). Similarly, the major interactions between glucosidase and chlorogenic acid are Hbonds formed throughout the glucosidase active site ( Figure 7D). Therefore, the biological activities displayed by these compounds are likely due to the inhibition of these enzymes.

Molecular Docking
Compounds accounting for ≥5% of the total bioactive compound content were further studied using molecular docking to estimate their binding strength and to predict their binding mode to each of the five enzymes (AChE, BChE, tyrosinase, amylase, and glucosidase). These compounds show the potential to bind to all enzymes, as suggested by their binding energy scores (Table 6). Furthermore, the majority of compounds displayed binding preference for AChE, BChE, and glucosidase. For instance, apigenin-di-C-hexoside (Vicenin-2) and arctiin bound strongly to the AChE and, to a lesser extent, to BChE, moderately to the amylase and glucosidase, but modestly to the tyrosinase. On the other hand, chlorogenic acid and dicaffeoylquinic acid preferentially bound to glucosidase. Hence, we visualized protein-ligand interaction details for these compounds to examine the interaction patterns.
The major contributors to the interaction in all the docking complexes are H-bonds and π-π interactions formed between hydroxyl groups and aromatic rings on the ligands and the residues in the active site of the target enzymes (Figure 7). In addition, a few hydrophobic contacts and several van der Waals interactions increased the binding strength. Apigenin-di-C-hexoside (Vicenin-2) spanned the cavity of AChE by forming multiple H-bonds with polar amino acid residues at the entrance to and deep inside the tunnel ( Figure 7A). Arctiin bound strongly to BchE, mainly via π-π interactions and a couple of van der Waals interactions ( Figure 7B). Amylase formed multiple H-bonds, a couple of π-π interactions, and van der Waals interactions throughout the amylase catalytic channel ( Figure 7C). Similarly, the major interactions between glucosidase and chlorogenic acid are H-bonds formed throughout the glucosidase active site ( Figure 7D). Therefore, the biological activities displayed by these compounds are likely due to the inhibition of these enzymes.

ADMET Prediction
ADMET (Absorption-Distribution-Metabolism-Excretion-Toxicity) properties in key metabolites in Asteraceae species extracts obtained by different extraction methods were predicted using Biovia DS. Further, 95 and 99% of a compound with high gastrointestinal absorption is expected to fall in ellipses colored in red and green, respectively. Moreover, 95 and 99% of a compound with blood-brain permeability is expected to be in ellipses colored in magenta and aqua, respectively (Figure 8). Palmitamide and caffeic acid were predicted to have high gastrointestinal (GI) absorption and blood-barrier penetration probability. Oleamide was predicted to have low probability of crossing the blood-brain barrier and low GI absorption due to its high polarity. Similarly, neochlorogenic acid, chlorogenic acid, dicaffeoylquinic acid, palmitamide, quinic acid, caffeoyl hexoside, apigenin-di-C-hexoside (vicenin-2), arctiin, medioresinol, and caftaric acid were found to have low GI absorption and low blood-brain barrier penetration probability. Nonetheless, all the compounds are not likely to be associated with any toxicities.

ADMET Prediction
ADMET (Absorption-Distribution-Metabolism-Excretion-Toxicity) properties in key metabolites in Asteraceae species extracts obtained by different extraction methods were predicted using Biovia DS. Further, 95 and 99% of a compound with high gastrointestinal absorption is expected to fall in ellipses colored in red and green, respectively. Moreover, 95 and 99% of a compound with blood-brain permeability is expected to be in ellipses colored in magenta and aqua, respectively (Figure 8). Palmitamide and caffeic acid were predicted to have high gastrointestinal (GI) absorption and blood-barrier penetration probability. Oleamide was predicted to have low probability of crossing the blood-brain barrier and low GI absorption due to its high polarity. Similarly, neochlorogenic acid, chlorogenic acid, dicaffeoylquinic acid, palmitamide, quinic acid, caffeoyl hexoside, apigenin-di-C-hexoside (vicenin-2), arctiin, medioresinol, and caftaric acid were found to have low GI absorption and low blood-brain barrier penetration probability. Nonetheless, all the compounds are not likely to be associated with any toxicities. Figure 8. ADMET plot of logarithm of logP (octanol-water partition coefficient) against topological polar surface area (PSA); 95 and 99% of a compound with high gastrointestinal absorption is expected to fall in ellipses colored in red and green, respectively; 95 and 99% of a compound with blood-brain permeability is expected to be in ellipses colored in magenta and aqua, respectively.

Conclusions
Species belonging to the Asteraceae are rich in secondary metabolites. LC-MS-MSguided profiling of the crude extracts of three Asteraceae plant samples, each obtained by three different extraction methods, namely HAE, MAC, and UAE, revealed the presence of a wide array of phytoconstituents. E. ritro extracts are predominately rich in oleamide, representing ca. 50% of the whole chromatogram as well as the phenolic acids dicaffeoyl Figure 8. ADMET plot of logarithm of logP (octanol-water partition coefficient) against topological polar surface area (PSA); 95 and 99% of a compound with high gastrointestinal absorption is expected to fall in ellipses colored in red and green, respectively; 95 and 99% of a compound with blood-brain permeability is expected to be in ellipses colored in magenta and aqua, respectively.

Conclusions
Species belonging to the Asteraceae are rich in secondary metabolites. LC-MS-MSguided profiling of the crude extracts of three Asteraceae plant samples, each obtained by three different extraction methods, namely HAE, MAC, and UAE, revealed the presence of a wide array of phytoconstituents. E. ritro extracts are predominately rich in oleamide, representing ca. 50% of the whole chromatogram as well as the phenolic acids dicaffeoyl quinic acid and chlorogenic acid and the fatty acid amide palmitamide. Similarly, the major peaks in T. decipiens extract are the simple phenolic acids chlorogenic, dicaffeoyl quinic, and caftaric acids, as well as the lignan compound medioresinol. On the other hand, C. deflexa showed predominance in the flavonoids, such as vicenin-2 and lignans as arctiin, but also with significant quantities of oleamide, chlorogenic acid, and caffeoyl hexoside. In addition to the chemical profiles, the extracts were tested for antioxidant and enzyme inhibitory properties. The biological activities depended on the used extraction solvents for each species and, in general, E. nitro exhibited stronger antioxidant ability as compared to other species. With regard to the enzyme inhibitory effects, all tested extracts showed inhibitory potentials. Our results could provide valuable insights to produce functional applications using the Asteraceae species and they could be considered as important sources of healthpromoting compounds. However, further studies, such as toxicity and bioavailability, need to understand the full functional pictures of the tested species.