In Vitro and In Silico Investigation of Polyacetylenes from Launaea capitata (Spreng.) Dandy as Potential COX-2, 5-LOX, and BchE Inhibitors

Diverse secondary metabolites are biosynthesized by plants via various enzymatic cascades. These have the capacity to interact with various human receptors, particularly enzymes implicated in the etiology of several diseases. The n-hexane fraction of the whole plant extract of the wild edible plant, Launaea capitata (Spreng.) Dandy was purified by column chromatography. Five polyacetylene derivatives were identified, including (3S,8E)-deca-8-en-4,6-diyne-1,3-diol (1A), (3S)-deca-4,6,8-triyne-1,3-diol (1B), (3S)-(6E,12E)-tetradecadiene-8,10-diyne-1,3-diol (2), bidensyneoside (3), and (3S)-(6E,12E)-tetradecadiene-8,10-diyne-1-ol-3-O-β-D-glucopyranoside (4). These compounds were investigated for their in vitro inhibitory activity against enzymes involved in neuroinflammatory disorders, including cyclooxygenase-2 (COX-2), 5-lipoxygenase (5-LOX), and butyrylcholinesterase (BchE) enzymes. All isolates recorded weak–moderate activities against COX-2. However, the polyacetylene glycoside (4) showed dual inhibition against BchE (IC50 14.77 ± 1.55 μM) and 5-LOX (IC50 34.59 ± 4.26 μM). Molecular docking experiments were conducted to explain these results, which showed that compound 4 exhibited greater binding affinity to 5-LOX (−8.132 kcal/mol) compared to the cocrystallized ligand (−6.218 kcal/mol). Similarly, 4 showed a good binding affinity to BchE (−7.305 kcal/mol), which was comparable to the cocrystallized ligand (−8.049 kcal/mol). Simultaneous docking was used to study the combinatorial affinity of the unresolved mixture 1A/1B to the active sites of the tested enzymes. Generally, the individual molecules showed lower docking scores against all the investigated targets compared to their combination, which was consistent with the in vitro results. This study demonstrated that the presence of a sugar moiety (in 3 and 4) resulted in dual inhibition of 5-LOX and BchE enzymes compared to their free polyacetylenes analogs. Thus, polyacetylene glycosides could be suggested as potential leads for developing new inhibitors against the enzymes involved in neuroinflammation.


Introduction
Nature remains a crucial resource for drug leads and has a considerable contribution to the field of drug discovery [1]. Many medicinal plants were reported to have beneficial effects on patients with Alzheimer's disease (AD), such as Gingko biloba, Bacopa monnieri, aliphatic carbon signals distinguished as; two methylenes at δ C 59.1 (C-1A; δ H 3.66, m, overlapped) and 41.3 (C-2A; δ H 1.86, m), and a methine carbon signal at δ C 60.3 (C-3A; δ H 4.58, t, J = 6.8 Hz). The HMBC spectrum showed correlations of H-1A/C-2A, H-3A/C-1A, and H-3A/C-2A confirming their connectivity. Furthermore, it showed HMBC correlations for the proton H-3A with the quaternary carbon signals at δ C 83.5 (C-4A), 69.7 (C-5A), and 72.4 (C-6A), and for the proton H-9A with the quaternary carbon signal at δ C 78.1 (C-7A), thereby confirming the presence of a central diyne moiety of a polyacetylene derivative. The HR-MS (positive mode) spectrum of compound 1 ( Figure S7) showed a pseudo molecular ion peak for 1A at m/z 187.0728 for [M + Na] + (Calcd. 187.0730), which agreed with the molecular formula, C 10 H 12 O 2 . By reviewing the data with those published in the literature, compound 1A was confirmed to be the known polyacetylene derivative, (8E)-deca-8-en-4,6diyne-1,3-diol isolated before from Gymnaster koraiensis (Asteraceae) [19,20]. However, it is worth noting that in their first report, Jung et al. [20] reported the absolute configuration of this compound as (3R). However, the same research group later published the acyl-CoA cholesterol acyltransferase (ACAT) inhibitory activity of the same compound, although they assigned (3S) for its absolute configuration by referring to the same spectral data in their previously published paper [19]. Thus, we compared 1A to its C 1 -O-glycoside, namely bidensyneoside A1 (3), which was isolated before from L. capitata [16]. Since it was concluded by the last research group that all compounds isolated from L. capitata had an S configuration [16], compound 1A was biosynthetically suggested as (3S,8E)-deca-8-en-4,6diyne-1,3-diol, the aglycone of bidensyneoside A1 [16,21].
The 1 H NMR, DEPT135, and HSQC spectra of 1 (Table 1 and Figures S1, S3 and S5) showed the presence of two olefinic proton signals, each was integrated for one proton at δH 5.59 (d, J = 15.8 Hz, H-8A; δC 110.5) and 6.31 (m, H-9A; δC 145.1). A methyl proton signal at δH 1.82 (dd, J = 6.9, 1.6 Hz, H-10A; δC 18.9) was found to be attached to the olefinic carbon (C-9A), as revealed from the HMBC spectrum ( Figure S6), which showed correlations for this methyl group with C-8A and C-9A. The DEPT135 spectrum ( Figure S3) showed three aliphatic carbon signals distinguished as; two methylenes at δC 59.1 (C-1A; δH 3.66, m, overlapped) and 41.3 (C-2A; δH 1.86, m), and a methine carbon signal at δC 60.3 (C-3A; δH 4.58, t, J = 6.8 Hz). The HMBC spectrum showed correlations of H-1A/C-2A, H-3A/C-1A, and H-3A/C-2A confirming their connectivity. Furthermore, it showed HMBC correlations for the proton H-3A with the quaternary carbon signals at δC 83.5 (C-4A), 69.7 (C-5A), and 72.4 (C-6A), and for the proton H-9A with the quaternary carbon signal at δC 78.1 (C-7A), thereby confirming the presence of a central diyne moiety of a polyacetylene derivative. The HR-MS (positive mode) spectrum of compound 1 ( Figure S7) showed a pseudo molecular ion peak for 1A at m/z 187.0728 for [M + Na] + (Calcd. 187.0730), which agreed with the molecular formula, C10H12O2. By reviewing the data with those published in the literature, compound 1A was confirmed to be the known polyacetylene derivative, (8E)-deca-8-en-4,6-diyne-1,3-diol isolated before from Gymnaster koraiensis (Asteraceae) [19,20]. However, it is worth noting that in their first report, Jung et al. [20] reported the absolute configuration of this compound as (3R). However, the same research group later published the acyl-CoA cholesterol acyltransferase (ACAT) inhibitory activity of the same compound, although they assigned (3S) for its absolute configuration by referring to the same spectral data in their previously published paper [19]. Thus, we compared 1A to its C1-O-glycoside, namely bidensyneoside A1 (3), which was isolated before from L. capitata [16]. Since it was concluded by the last research group that all compounds isolated from L. capitata had an S configuration [16], compound 1A was biosynthetically suggested as (3S,8E)-deca-8-en-4,6-diyne-1,3-diol, the aglycone of bidensyneoside A1 [16,21]. The second compound of the mixture (1B) was closely related to 1A with minor differences. Compound 1B showed the absence of the olefinic bond as it was found to be replaced by an additional acetylene group and appeared to be a biosynthetic product of the dehydrogenation of 1A [22]. This was evident from the 1 H NMR, DEPT135, and HSQC The second compound of the mixture (1B) was closely related to 1A with minor differences. Compound 1B showed the absence of the olefinic bond as it was found to be replaced by an additional acetylene group and appeared to be a biosynthetic product of the dehydrogenation of 1A [22]. This was evident from the 1 H NMR, DEPT135, and HSQC spectra of 1 (Table 1 and Figures S1, S3 and S5), which showed a highly shielded methyl group at δ C 3.81 (C-10B; δ H; 1.99, s). This methyl group showed HMBC correlation with two quaternary carbon signals at 64.5 (C-8B), and 78.0 (C-9B) confirming the presence of an additional acetylene group at C-8B. The HR-MS (positive mode) spectrum of compound 1 ( Figure S7) showed a pseudo molecular ion peak for 1B at m/z 167.0126 for [M-H 2 O + Na] + (Calcd. 167.0473), which agreed with a molecular formula of C 10 H 10 O 2 . Thus, compound 1B was confirmed as (3S)-deca-4,6,8-triyne-1,3-diol, which was reported before in the Asteraceae family in Artemisia capillaris [23] and Lactuca sativa [24]. It could be concluded that the C 1 -O-glucosides of both 1A and 1B were previously reported from the Asteraceae plant, Bidens parviflora WILLD. and were termed bidensyneoside A1 and bidensyneoside B, respectively [21]. However, their aglycones (1A, and its biosynthetic dehydrogenated product, 1B) were reported herein for the first time from L. capitata.

Identification of Compound 4
The spectral data of 4 ( Figures S22-S28) were closely related to that of 3, except for the presence of a C 14 polyacetylene chain and appeared to be the 3-β-O-glucosyl derivative of compound 2. This was revealed from 1 H NMR, DEPT135, and HSQC (Table 2 Figure S28) confirmed a molecular formula of C 20 H 28 O 7 based on the pseudo molecular ion peak of [M + Na] + at 403.1719 (Calcd. 403.1722). Therefore, the structure of compound 4 was confirmed as (3S)-(6E,12E)-tetradecadiene-8,10-diyne-1-ol-3-O-β-D-glucopyranoside, which was previously reported from Coreopsis tinctoria (Asteraceae) [27]. It is worth noting that compound 4 was confirmed as the C 3 -O-glucoside of 2, isolated, herein, for the first time from L. capitata. However, its C 1 -O-glucoside derivative, termed bidensyneoside E, was previously reported in the same plant [16]. These findings supported the presence of the same configuration at C-3 for the three isolated plyacetylene derivatives from L. capitata.

Docking Study
Molecular docking is a broadly used computational tool in drug discovery. It is very helpful for predicting the mode of interaction and the binding affinity of the ligands towards the investigated proteins [32,33]. Molecular docking was performed to explain the observed results of the in vitro enzyme inhibition assays and to investigate the mode of interaction of the tested molecules. The observed H-bonding of the tested compounds with the amino acid residues in the active sites of the investigated proteins are listed in Table S2. It is worth noting that compounds 1A and 1B were tested as a mixture in the in vitro enzyme assays, which is why the simultaneous docking function that permits the docking of multiple ligands to the same target was used to study their interaction with the investigated proteins. Interestingly, the use of this function has enabled us to explain the in vitro enzyme inhibition assay results. The tested 1A and 1B individual molecules showed low docking scores against all the investigated targets (Table 4). However, the tested 1A/1B combination showed a reasonable docking score (−7.861 kcal/mol) against COX-2 enzyme, which was consistent with its slightly better in vitro inhibitory activity among the tested molecules with IC 50 of 170.48 ± 20.15 µM ( Table 3). Visualization of the best docking pose showed that the alcoholic groups of the two compounds formed H-bonds with Arg-120 (Figure 2a) located in the opening of the cyclooxygenase channel and is essential for COX-2 catalysis [34,35]. Compound 1A also formed an H-bond with Glu-524 [34,35]. The cocrystallized ligand of COX-2 showed the highest docking score (−8.004 kcal/mol) and exhibited H-bonding interactions with Ser-530, and Tyr-385 amino acid residues essential for the enzyme activity (Table S2, Figure S29a) [35]. Compound 4 showed a reasonable docking score against COX-2 enzyme (−6.895 kcal/mol, Table 4) and showed the formation of H-bonding with Asp-347, Gln-350, His-351, Tyr-355, and phe-580 amino acid residues (Table S2, Figure 3b), which explain its highest in vitro inhibition activity against the COX-2 enzyme with an IC 50 of 146.38 ± 7.70 µM.
Although the 1A/1B combination showed the highest docking scores against 5-LOX and BchE, as shown by simultaneous docking (Table 4), no H-bonding interactions with key amino acids were observed during the visualization of their obtained docking poses against 5-LOX and BchE (Figures S29d and S29g, respectively). The natural substrate of 5-LOX and arachidonic acid was reported to form van der Waals contacts within the active site of 5-LOX, and no H-bonds were observed, as illustrated in Figure S29c [36]. While compound 4 showed H-bonding with the amino acid residues Val-175, Asp-176, and Ala-606 in the active site of 5-LOX (Figure 2c). Residues Val-175 and Asp-176 were reported to be a part of the V4 anchor of the 5-LOX's active site and were suggested to be involved in the binding of novel 5-LOX inhibitors [36]. These obtained docking interactions together with the high docking score of compound 4 (−8.132 kcal/mol, Table 4) explained the highest in vitro inhibitory activity of this compound against 5-LOX. Compound 3 showed H-bonding interactions with His-372, His-367, and His-550 amino acid residues ( Figure S29e) in the Fe coordination sphere of the 5-LOX active site explaining its in vitro enzyme inhibition activity [37]. Two molecules of the cocrystallized ligand were examined in the active site of BchE, as presented in Figure S29f. They were reported to form aromatic stacking with Tyr-332, although no H-bonds were reported to be observed [38]. Compound 4 showed H-bonding interactions with Trp-82, Gly-116, Gly-117, and Trp-430 amino acid residues in the active site of BchE (Figure 2d). Particularly, Gly116 and Gly117 amino acids were involved in the oxyanion hole in the active site of BchE [38]. Hence, the highest in vitro enzyme inhibition activity of compound 4 IC 50 (14.77 ± 1.55 µM) could be attributed to its interaction with these residues. Compound 3 displayed three H-bonding interactions with Tyr-128, Trp-82, and His-438 residues ( Figure S29h) in the active site of BchE [39,40]. It showed a lower docking score (−7.018 kcal/mol, Table 4) and a higher IC 50 (48.81 ± 6.34 µM) than compound 4 against the BchE enzyme.
It can be concluded that the compound 4W showed the best docking interactions against the investigated enzymes explaining its highest in vitro enzyme inhibition activity. The presence of the sugar moiety in the structures of compound 4 and compound 3 greatly enhanced their enzyme inhibitory activity in comparison to the free polyacetylene molecules, as in the case of compound 2.

Plant Material
The whole plant of Launaea capitata (Spreng.) Dandy (Figure 3) was harvested in Riyadh, Saudi Arabia, in March 2022. A voucher sample (ID #16741) was kept at the herbarium of the Pharmacognosy Department, Prince Sattam bin Abdulaziz University, Al-Kharj. The identity of the plant was confirmed by Professor Ibrahim A. Mashaly, Department of Plant Ecology, Faculty of Sciences, Mansoura University, Egypt. The plant material was shade-dried, powdered, and kept for further phytochemical processing.

Extraction and Purification
About 1200 g of the dry powder of the title plant was exhausted by maceration in cold MeOH (5 × 2000 mL). The combined methanolic extracts were evaporated at 45 • C by an R-215 rotavapor (Buchi, Switzerland). The obtained extract (325.0 g) was extracted by suspending it in water and shaking it with n-hexane (4 × 500 mL) to give a non-polar nhexane fraction (87.3 g). Chromatographic separation of the obtained n-hexane fraction was carried out on a silica gel CC., (5 cm i.d. × 50 cm L.), packed in n-hexane, eluted with EtOAc (from 0→100%, gradient), then, by EtOAc-MeOH (0→100%, gradient), and the effluent volume was 250 mL. The obtained fractions (400) were monitored by normal phase silica gel and grouped based on their components to provide ten groups (Hex-I-X). Group Hex-IV (109-144, weight; 500 mg) eluted with n-hexane-EtOAc

BchE Inhibition Assay
The in vitro butyrylcholinesterase (BchE) inhibitory activity of the isolated polyacetylenes was determined according to the published method, using horse serum butyrylcholinesterase (Equine BchE; CAS#9001-08-5) with a slight change [18,41,42]. To create a useful stock solution, the materials were dissolved in DMSO. The maximal DMSO concentration was 0.1%, and twelve sample concentrations (ranging from 500 to 0.25 g/mL) were created by two-fold serial dilution in Tris buffer (pH 8.0) [18].

COX-2 Inhibition Assay
The fluorometric approach was used to measure the in vitro cyclooxygenase (COX-2) inhibitory activity. It was carried out in accordance with the COX-2 Inhibitor Screening Kit methodology, as advised by the manufacturer (CAT # K547-100, BioVision, Milpitas, CA, USA) [18,43,44]. The sample solutions were prepared in the manner described before providing twelve concentrations (ranging from 500 to 0.25 g/mL) [18]. The positive control was nordihydroguaiaretic acid (NDGA; CAS # 500-38-9, MO, Sigma-Aldrich, USA). The concentration-response curve created by GraphPad Prism version 8.0 (San Diego, CA, USA) was used to measure the IC 50 (the concentration of the test sample that inhibited 50% of COX-2 enzyme activity).

Statistical Analysis
The IC 50 values were represented as means (± SD) from triplicates of two independent experiments. The IC 50 value was determined from the concentration-response curve processed using GraphPad Prism version 8.0 (San Diego, CA, USA), representing the concentration, which inhibited 50% of the enzyme activity.

Docking Study
Autodock vina 1.2.3 was used for performing the molecular docking study [47,48]. The crystal structures for COX-2, 5-LOX, and BchE were acquired from the RCSB protein data bank. Their PDB codes and the cocrystallized ligands are mentioned in Table S1. The structures of the proteins and ligands were prepared for the docking study with the help of Autodock tools. The docking parameters were recognized by selecting the coordinates of the binding sites using a grid box around the cocrystallized ligand with spacing of 0.375 Å. The dimensions of 30 × 30 × 30. The X, Y, and Z coordinates of the used grid box in each investigated protein are described in Table S1. The number of runs was determined by the exhaustiveness parameter and set to a value of 10, while the number of modes was set to 20. The docking process was validated by removing the cocrystallized ligand and redocking it into the active site using Autodock vina. Subsequently, the root mean square deviation (RMSD) value of the redocked ligand superimposed with the cocrystallized ligand was determined using Pymol and it was found to be lower than 2 Å. Docking poses showing minimum RMSD values were demonstrated using Pymol (version 2.5) [49].

Limitation of the Study
Although the active compounds showed inhibition of the target enzymes through endpoint inhibition reactions, determinations of the kinetic inhibition modes are essential to show their inhibition mechanisms. Furthermore, additional in vivo animal models are required to study the pharmacodynamics and pharmacokinetic properties of the active compounds.

Conclusions
In this study, as a continuation of our effort to discover the phytochemical composition of Launaea capitata (Spreng.) Dandy, the n-hexane fraction of the methanol extract of the whole plant was investigated. Four polyacetylene derivatives including three aglycones and one glycoside were isolated for the first time from this plant. in addition to a previously isolated glycoside derivative. The isolates showed reasonable 5-LOX and BchE inhibitory activities, however, they showed weak-moderate activities against COX-2. Docking experiments were conducted to clarify the mode of binding and to explain the results of the in vitro inhibitory activities. The current study suggested polyacetylene glycosides as promising leads for the dual inhibition of 5-LOX and BchE enzymes, which could be applied for the prevention and treatment of neuroinflammatory disorders, such as Alzheimer's disease.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/molecules28083526/s1, Figures S1-S28: 1H, 13 C, DEPT135, 1 H-1 H COSY, HSQC, HMBC, and HR-MS spectra of isolated compounds 1-4; Figure S29: 3D binding models of the cocrystallized ligand in the active site of COX-2; 3 in the active site of COX-2; 4 in the active site of COX-2; the cocrystallized ligand (arachidonic acid) in the active site of 5-LOX; simultaneously docked molecules 1A and 1B in the active site of 5-LOX; the cocrystallized ligand in the active site of BchE. Table S1: PDB codes of the crystal structures and grid box coordinates for the enzymes used in the docking study. Table S2