7-Acetoxyhorminone from Salvia multicaulis Vahl. as Promising Inhibitor of 3-Hydroxy-3-methylglutaryl Coenzyme A (HMG-CoA) Reductase

3-Hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase is a key enzyme involved in cholesterol biosynthesis and one of the most important targets for the treatment of hypercholesterolemia. A limited number of studies on the HMG-CoA reductase inhibitory potential of natural products are available. Thus, in the current study, we aimed to test the HMG-CoA reductase inhibitory capacity of extracts from the roots and aerial parts of Salvia multicaulis Vahl., through activity-guided isolation. Our findings revealed that the root extract prepared with dichloromethane–acetone (1:1) showed the highest inhibition (71.97 ± 0.37%) at 100 µg/mL. The extract was then initially fractionated by column chromatography and the obtained fractions were monitored by thin layer chromatography. Fractions which were similar to each other were combined and a total of 15 fractions were obtained. Further conventional chromatographic studies were carried out on the active fractions. Based on these fractions, 10 known compounds, comprising 9 terpenes and 1 steroid derivative in total, were isolated and their structures were verified by a combination of IT-TOF-MS, and 1D and 2D NMR techniques. According to the enzyme inhibition data of the identified compounds, 7-acetoxyhorminone exerted the highest inhibition (84.15 ± 0.10%, IC50 = 63.6 ± 1.21 µg/mL). The molecular docking experiments on 7-acetoxyhorminone and horminone indicated that both compounds strongly bind to the active site of the enzyme.


HMG-CoA Reductase Inhibitory Activity of the Isolated Compounds
In this study, the extracts of the root and aerial parts of the plant in different polarities were initially subjected to enzyme inhibition assays. Then, activity-guided fractionation was performed on the most active extract, i.e., dichloromethane-acetone (1:1). The obtained fractions were combined and again subjected to enzyme inhibition assays (Table 1). Since five fractions, i.e., M-3, M-6, M-7, M-8, and M-10, exhibited inhibition over 50%, it was decided that studies should be continued on them. The HMG-CoA reductase inhibitory activity of the compounds isolated from the active fractions was also determined. Atorvastatin was used as the reference drug, while horminone-7-acetoxyhorminone mixture (1:1) (6) and 7-acetoxyhorminone (7) were found to possess the highest inhibition ( Table 2). Table 2. HMG-CoA reductase inhibitory activity of the isolated compounds.

Molecular Docking Data
A molecular docking simulation has been found useful to obtain a better understanding of the inhibition mechanisms of small molecules-including how the ligands stick to the catalytic domains of receptors, which amino acids are more likely to contribute to stabi-Pharmaceuticals 2022, 15, 198 5 of 11 lizing the ligands, and the chemical bonds that would be formed between the ligands and receptor. In this section, the binding positions and the free energies of the two compounds, horminone-7-acetoxyhorminone and 7-acetoxyhorminone, inside the active site of HMG-CoA reductase, were determined using the docking simulation method. The neighboring amino acids surrounding and stabilizing the ligands of horminone-7-acetoxyhorminone (1) and 7-acetoxyhorminone (2), by forming polar and non-polar interactions, were displayed with 2D diagrams in Figure 2. The amino acids, i.e., Ser684, Lys692, Lys735, Hie752, and Asn755, have been found to play a dominant role in bonding the ligands to the receptor. Both systems suffer from the effect of less aromatic interactions, since neither of the aromatic domains in the ligands is likely to form strong pi-pi stacking or hydrophobic bonds with the amino acids in the binding domain, which can result in an increase in free energy. a Standard deviation: Values expressed are the means ± SD of three parallel measurements and the values were calculated according to a negative control, b NA: Not active, c Compounds with inhibition over 50% were bolded.

Molecular Docking Data
A molecular docking simulation has been found useful to obtain a better understanding of the inhibition mechanisms of small molecules-including how the ligands stick to the catalytic domains of receptors, which amino acids are more likely to contribute to stabilizing the ligands, and the chemical bonds that would be formed between the ligands and receptor. In this section, the binding positions and the free energies of the two compounds, horminone-7-acetoxyhorminone and 7-acetoxyhorminone, inside the active site of HMG-CoA reductase, were determined using the docking simulation method. The neighboring amino acids surrounding and stabilizing the ligands of horminone-7-acetoxyhorminone (1) and 7-acetoxyhorminone (2), by forming polar and non-polar interactions, were displayed with 2D diagrams in Figure 2. The amino acids, i.e., Ser684, Lys692, Lys735, Hie752, and Asn755, have been found to play a dominant role in bonding the ligands to the receptor. Both systems suffer from the effect of less aromatic interactions, since neither of the aromatic domains in the ligands is likely to form strong pi-pi stacking or hydrophobic bonds with the amino acids in the binding domain, which can result in an increase in free energy. More than one binding energy has been calculated for every single ligand, following the fact that ligands inside the binding cavity can fall into diverse poses with different binding energies, as shown in Figure 3. The distribution of the binding energies in the boxplots were classified as the minimum, first quartile, median, third quartile, and maximum scores. A convergence of the median scores for both compounds, falling under −5 kcal/mol, has been observed, which demonstrates the ability of the ligands to stick to the binding site of the receptor. Compound 2 has been found to have a wider interquartile range and a lower minimum score, of under −7 kcal/mol, compared to compound 1. Compound 2 has been able to pose more diverse conformers with lower binding energies during sticking to the binding site, due to more polar groups in the structure, which, in turn, form more polar bonds with the amino acids. More than one binding energy has been calculated for every single ligand, following the fact that ligands inside the binding cavity can fall into diverse poses with different binding energies, as shown in Figure 3. The distribution of the binding energies in the boxplots were classified as the minimum, first quartile, median, third quartile, and maximum scores. A convergence of the median scores for both compounds, falling under −5 kcal/mol, has been observed, which demonstrates the ability of the ligands to stick to the binding site of the receptor. Compound 2 has been found to have a wider interquartile range and a lower minimum score, of under −7 kcal/mol, compared to compound 1. Compound 2 has been able to pose more diverse conformers with lower binding energies during sticking to the binding site, due to more polar groups in the structure, which, in turn, form more polar bonds with the amino acids.

Discussion
Cardiovascular disease, hypercholesterolemia in particular, is among the leading health problems in the world. It ranks first among the causes of death in the USA [33]. The most prescribed drug group in the current treatment of hypercholesterolemia is

Discussion
Cardiovascular disease, hypercholesterolemia in particular, is among the leading health problems in the world. It ranks first among the causes of death in the USA [33]. The most prescribed drug group in the current treatment of hypercholesterolemia is statins, which act through the HMG-CoA reductase inhibitory mechanism [34][35][36]. Although statins are considered the most effective drug class against hypercholesterolemia, they have various adverse effects, such as myopathy, rhabdomyolysis, and increased liver enzyme levels. In addition, findings have pointed to the fact that they increase the incidence of type-2 diabetes [37][38][39][40][41][42][43]. According to our detailed literature research, only a few studies on the inhibitory potential of Salvia L. taxa on HMG-CoA reductase have so far been reported, which inspired us to conduct the present research. In one of the previous studies, a traditional Chinese drug consisting of S. miltiorrhiza Bunge and Carthamus tinctorius L., was found to reduce HMG-CoA reductase mRNA expression in female ApoE-/-and LDLR-/type mice [44]. In another study, it was reported that the protein fraction obtained from S. hispanica L., known as "chia", inhibited HMG-CoA reductase [45]. As mentioned, a limited number of reports are available on the HMG-CoA reductase inhibitory capacity of natural molecules and medicinal herbs. Therefore, more research on natural sources to find lead compounds is needed. In this context, we can conclude that the abietane-type diterpenes, i.e., horminone and horminone-7-acetoxyhorminone (1:1), were the leading compounds responsible for the HMG-CoA reductase inhibitory activity of S. multicaulis root extract. The HMG-CoA reductase inhibitory effects of diterpenes against the mentioned enzyme have been reported in some studies, such as Polyalthia longifolia (Sonn.) Thwaites [46]. For instance, 16α-hydroxycleroda-3,13(14)Z-diene-15,16-olide, a clerodane-type diterpene isolated from pendula, has been described as a "new class of HMG-CoA reductase inhibitory natural compounds" due to its potent inhibition [6,46,47]. It has been determined that the diterpene-derivative compound called cafestol in coffee, inhibited HMG-CoA reductase by 40% at a concentration of 20 µg/mL [48].
The early discovered statins, i.e., mevastatin and lovastatin, became model parent molecules in the discovery of new HMG-CoA reductase inhibitors. Guided by this structural similarity, simvastatin, pravastatin, cerivastatin, pitavastatin, etc., were synthesized as novel inhibitors. The hexahydronaphthalene ring systems in mevastatin and lovastatin, as well as monacolins, were reported to be similar to the HMG part of the enzyme that induces the activity through binding to the domain of the active gorge in HMG-CoA reductase [49]. A polyketide substitution is also common in natural statins [50]. In this study, 7-acetoxyhorminone ( Figure 4) was predicted to exhibit strong bonding through its decalin ring structure, while the fluorophenyl group of the atorvastatin molecule, which was used as the reference, was thought to play an active role in the binding. However, it can be predicted that the methyl groups provide the Van der Waals interaction, while the hydroxyl functional groups that are common in both structures, interact with the active site of the enzyme through hydrogen bonding. It can be said that the efficacy difference between the reference molecule and the natural molecule is due to the pi-pi interactions arising from the phenyl groups in atorvastatin, and the additional bonding energies of the groups capable of hydrogen bonding, may contribute in part to the differences in pharmacological properties.
Considering the genus Salvia, it has been reported that nearly 500 diterpene-derivative compounds have been isolated. In general, the effects of diterpenes on the inhibition of the mentioned enzyme have been reported in some studies. However, no study has yet been found in the literature on the HMG-CoA reductase inhibitory activity of abietane-type diterpenes. In our study, it was concluded that the abietane-type diterpenes, horminone, and horminone-7-acetoxyhorminone (1:1), were the leading compounds responsible for the HMG-CoA reductase inhibitory activity of S. multicaulis root extract. The results of our study may also form a basis for the use of some Salvia species in folk medicine against cardiovascular diseases, while horminone and 7-acetoxyhorminone may be considered as model molecules for designing new HMG-CoA reductase inhibitors.
interact with the active site of the enzyme through hydrogen bonding. It can be said that the efficacy difference between the reference molecule and the natural molecule is due to the pi-pi interactions arising from the phenyl groups in atorvastatin, and the additional bonding energies of the groups capable of hydrogen bonding, may contribute in part to the differences in pharmacological properties.  Considering the genus Salvia, it has been reported that nearly 500 diterpene-derivative compounds have been isolated. In general, the effects of diterpenes on the inhibition of the mentioned enzyme have been reported in some studies. However, no study has yet been found in the literature on the HMG-CoA reductase inhibitory activity of abietane-type diterpenes. In our study, it was concluded that the abietane-type diterpenes, horminone, and horminone-7-acetoxyhorminone (1:1), were the leading compounds responsible for the HMG-CoA reductase inhibitory activity of S. multicaulis root extract. The results of our study may also form a basis for the use of some Salvia species in folk medicine against cardiovascular diseases, while horminone and 7-acetoxyhorminone may be considered as model molecules for designing new HMG-CoA reductase inhibitors.

Plant Material
The sample of S. multicaulis Vahl. was collected from the vicinity of Van Province

Extraction and Fractionation
The aerial parts and roots of S. multicaulis were dried in the shade and then powdered in a grinder. Both plant parts were sequentially macerated with petroleum ether, dichloromethane-acetone (1:1), and ethanol (95%), by occasional shaking at room temperature. After each filtration, the solvent was evaporated to dryness under a vacuum. Then, the enzyme inhibition assay was performed on each extract ( Table 1). The root dichloromethane-acetone (1:1) extract was fractionated on a silica gel glass column (5 × 150

Extraction and Fractionation
The aerial parts and roots of S. multicaulis were dried in the shade and then powdered in a grinder. Both plant parts were sequentially macerated with petroleum ether, dichloromethane-acetone (1:1), and ethanol (95%), by occasional shaking at room temperature. After each filtration, the solvent was evaporated to dryness under a vacuum. Then, the enzyme inhibition assay was performed on each extract ( Table 1). The root dichloromethane-acetone (1:1) extract was fractionated on a silica gel glass column (5 × 150 and 2 × 100 cm, respectively) using petroleum ether (40-60 • C), followed by a gradient elution using dichloromethane, acetone, methanol, and water, up to 50%, where 52 fractions in total were obtained. After combining phytochemically similar fractions according to thin layer chromatography (TLC) monitoring, 15 subfractions were obtained, which were immediately subjected to HMG-CoA reductase inhibition assays (Table 1). Following UV light checking, TLC plates were visualized by spraying them with cerium (IV) sulfate dissolved in 10% sulfuric acid. Silica gel, Sephadex LH-20 columns, and preparative TLC techniques were used to isolate compounds from the active fractions, which led to isolation of compounds 1-10. The spectral data of the compounds is presented in supplementary material file.

Microtiter Assay for HMG-CoA Reductase Inhibition
The enzyme inhibition method developed by Wang et al. (2015) was applied. In the experimental procedure, 10 µL of the sample to be tested was first added to the microplate wells. Then, 10 µL of potassium phosphate buffer, prepared to contain EDTA (pH 7), dithiothrethiol ( reader (Eon Biotek, Winooski, VT, USA). Atorvastatin was used as the reference drug and dimethylsulfoxide (DMSO) as a control [51].

Molecular Docking Experiments
The crystal structure of HMG-CoA reductase, coded 2Q1L, was downloaded from the Protein Data Bank (PDB)-the chains A and B were excluded from the structure and the rest of the system was managed for use in the simulations. All the hydrogen and missing atoms of the amino acids were added in the process of protein preparation with the preparation wizard of Maestro [52,53]. The protonation state of the system was set in a biological pH and its total energy was minimized using a force field method to remove any clashes between the atoms.
The 2D structures of the ligands were sketched and then their total energies were minimized in order to search for the most stable conformers. The docking method used here allowed both the ligand and the receptor to be flexible and optimized during the simulation. The idea behind this algorithm is to apply an induced fit docking method for exploring docking poses with the lowest binding free energies inside the active site of the receptor [54][55][56].