Medicinal Chemistry Friendliness of Pigments from Monascus-Fermented Rice and the Molecular Docking Analysis of Their Anti-Hyperlipidemia Properties

: In this study, the physicochemical properties, pharmacokinetics properties, and drug-likeness of pigments from Monascus -fermented rice ( Monascus pigments, MPs) were predicted in silico using SwissADME tool. In silico prediction of physicochemical properties showed that MPs had desirable lipophilic drug-like physicochemical properties including molecular weight (236 to 543), TPSA (44.76 to 179.77), lipophilicity ( − 0.81 to 4.14), and water solubility ( − 4.94 to − 0.77). The pharmacokinetic properties of MPs (i.e., GIA, P-glycoprotein substrate, and CYP3A4 inhibitor) illustrated that most MPs had high intestinal absorption and bioavailability, but some MPs might cause pharmacokinetics-related drug–drug interactions. Following this, six main well-known MPs (monascin, ankaﬂavin, rubropunctatin, monascorubrin, rubropunctamine, monascorubramine) were selected for molecular docking with some enzyme receptors. The docking results were shown with the best molecular docking poses, and the interacting residues, number and distance of hydrogen bonds of the MPs and monacolin K (for docking with 3-hydroxy-3-methyl glutaryl coenzyme A reductase (HMG-CoA reductase)), or MPs and oleic acid (for docking with lipase). Dissociation constants showed that MPs had lower inhibitory potential for HMGR (compared with Monacolin K), and higher inhibitory potential for lipase. Individual pigments from Monascus -fermented rice, therefore, have the potential to be developed as drug candidates for controlling hyperlipidemia.


Introduction
Hyperlipidemia refers to acquired or genetic disorders resulting in high levels of cholesterol, triglycerides (TG), free fatty acids, and low-density lipoprotein cholesterol (LDL-C) in the blood circulation, along with an excessive accumulation of TG in the liver [1]. Hyperlipidemia has become a global public health problem as a risk factor for strokes or heart attacks [2,3]. Individuals with hyperlipidemia have responded well to lipid-lowering agents such as statins, nicotinic acids, and fibrates [4]. However, nowadays, many people prefer to use nutraceuticals due to interest in 'natural' avenues of disease control, and the concern of potential drug interactions, long-term safety, and the economic implications of long-term (or life-long) dependence on drugs [5,6]. Moreover, some of the medications used to treat hyperlipidemia e.g., statins have side effects such as nausea, headaches, myalgias, memory loss and cognitive function impairment, and increased risk of diabetes [7][8][9][10][11].
Monascus fermented rice has been used as a functional food worldwide for decades [12,13]. A large number of studies has shown that Monascus fermented rice is effective in hyperlipidemia 2. Methods

In Silico Evaluation of Physicochemical Properties and Drug-Likeness of Monascus Pigments
In silico evaluation of physicochemical properties and drug-likeness for the 61 Monascus pigments was evaluated using the SwissADME tool (http://www.swissadme.ch/index.php#). This tool computes physicochemical properties (molecular weight, TPSA (topological polar surface area), lipophilicity, and water solubility), and estimates the indicators of pharmacokinetics (GIA (gastrointestinal absorption), P-glycoprotein (permeability glycoprotein or multidrug resistance protein 1) substrate, and CYP3A4 inhibitor (cytochrome P450 3A4 inhibitor)) as well as drug-likeness and bioavailability of MPs.

Molecular Docking
Molecular docking was performed using the Discovery Studio software 2019 (DS 2019) and Autodock Vina. For docking onto HMG-CoA Reductase receptor, a crystal structure of human HMG-CoA Reductase (1HW9 (PDB) bound with simvastatin (an HMG-CoA Reductase inhibitor) was used. The simvastatin and water molecules were removed, and polar hydrogen atoms were added prior to docking. A binding site was searched in a box of 24 × 24 × 24 Å with centered on the coordinates x: 10.8, y: 10.8 and z: 13.6. For lipase docking, a crystal structure of porcine lipase 1ETH (PDB) was used. Unnecessary structures such as water molecules and metals were removed, and polar hydrogen atoms were added. A binding site was searched in a box of 24 × 24 × 24 Å with centered on the coordinates x: 63.7, y: 28.5, z: 120.6, and automated molecular docking was then performed using Autodock Vina. In the docking process, the ligand was flexible, while the receptor (protein) was rigid. All generated docking modes were evaluated according to affinity energy values. The dissociation constant (K i ) were calculated using affinity energy according to our previous study [33]. The DS 2019 software was utilized to view hydrogen bonds as well as hydrophilic, hydrophobic, and electrostatic interactions between residues at the HMG-CoA Reductase or lipase active sites and the pigment compounds.

Physicochemical Properties of Monascus Pigments
The physicochemical characteristics of MPs are presented in Figure 1. The mol. wt. of 61 MPs ranged from 236 to 543. Twenty-six (26) yellow MPs and 6 orange MPs had a mol. wt. range of 236-430 and 354-440, respectively, while 29 red MPs had mol. Wt. between 353 and 543 ( Figure 1A). The mol. wt. of most (56) MPs were less than 500 in accordance with the rules-of-five [34]. Only 5 red MPs had mol. wt. greater than 500.
Topological polar surface area (TPSA) is a critical descriptor in many estimation models and rules of drug-likeness or bioavailability, especially in regards to absorption [35]. TPSA ranges of 20-130 Å 2 are indicative of the compound having the ability to permeate into cells [36]. The TPSA of MPs were in the range of 44.76 to 179.77. As shown in Figure 1B The lipophilicity of MPs was calculated by using the partition coefficient between n-octanol and water (log P o/w ) [37]. To increase the prediction accuracy, consensus Log P o/w was conducted, which is the average of five predictive methods (iLOGP, XLOGP3, WLOGP, MLOGP and SILICOS-IT) [37]. The lipophilicity of MPs ranged from −0.81 to 4.14. The yellow MPs had a lipophilicity range of 0.4 to 4.14, red MPs ranged from −0.81 to 4.09, while orange MPs ranged between 2.98 and 4.04 ( Figure 1C). The lipophilicity of all MPs was below 5, implying they have good permeability across the cell membrane, while only two MPs (one red and one yellow) had lipophilicity below 1. This shows that most MPs are more lipophilic and less water-soluble [34].  The water solubility is also an important parameter, since a high Log S values correspond to good drug absorption [38]. The Log S of MPs were calculated with method ESOL, and a qualitative estimation of the solubility class is given according to the following scale: Insoluble < −10 < Poorly soluble < −6 < Moderately soluble < −4 < Soluble <−2 < Very soluble < 0 < Highly soluble [39].  Figure 1D). Therefore, the water solubility results suggested yellow MPs would have better drug absorption compared with red MPs.

Pharmacokinetics Properties
The pharmacokinetics and drug-likeness of all the 61 Monascus pigments were predicted using physicochemical and ADME properties calculated from SwissADME [37]. As shown in Figure 2, the pharmacokinetics were evaluated by GIA (gastrointestinal absorption), P-glycoprotein (permeability glycoprotein or multidrug resistance protein 1) substrate, and CYP3A4 inhibitor (cytochrome P450 3A4 inhibitor).
The prediction of GIA is based on the Brain Or IntestinaL EstimateD permeation (BOILED Egg) model, which uses lipophilicity and polarity of molecules to predict passive gastrointestinal absorption of compounds [35].
The Bioavailability Score is a semi-quantitative rule-based score relying on total charge, TPSA, and violation to the Lipinski filter, with four classes of probabilities (11%, 17%, 56% or 85%) that compounds have >10% bioavailability in rat or measurable Caco-2 permeability [37]. A score of 55% means the compound passes the rule-of-five, and a score of 17% is a fail [37]. The bioavailability score results showed that most of the MPs have moderate bioavailability (0.55-0.56), while only one yellow MPs has low bioavailability (0.11) (Figure 3).
The Bioavailability radar of 6 well-known MPs was shown in Figure 4. All six main Monascus pigments had physicochemical profiles that makes them suitable for oral administration. Moreover, P-glycoprotein is a cell membrane transport protein pumping xenobiotic (including drug molecules) out of cells [40]. This transporter can protect cells against potentially toxic substances by promoting biliary and renal elimination. It could also decrease intestinal absorption and bioavailability of drugs through limiting cytosolic accumulation [40]. Thirty-two (32) MPs including six well-known MPs were not P-glycoprotein substrates, suggesting that these MPs were likely to have high intestinal absorption and bioavailability ( Figure 2B).

Orange
The cytochrome P450 enzymes are major (accounting for 75%) enzymes involved in drug metabolism. They play a key role in drug elimination through metabolic biotransformation [41]. Members of the CYP3A subfamily, especially CYP3A4 and CYP3A5 (having similar substrate specificities) in adults, are considered to be critical enzymes for drug metabolism, synergistically processing drug molecules with P-glycoprotein to improve the protection of tissues and organisms [42]. Eighteen (18) MPs exhibited in silico inhibition of CYP3A4 (9 yellow, 8 red, and 1 orange) ( Figure 2C, Supplementary File 1). Due to the lower clearance and accumulation of the drug or its metabolites, inhibition of these MPs might cause pharmacokinetics-related drug-drug interactions leading to toxic or other unwanted adverse effects.
The Bioavailability Score is a semi-quantitative rule-based score relying on total charge, TPSA, and violation to the Lipinski filter, with four classes of probabilities (11%, 17%, 56% or 85%) that compounds have >10% bioavailability in rat or measurable Caco-2 permeability [37]. A score of 55% means the compound passes the rule-of-five, and a score of 17% is a fail [37]. The bioavailability score results showed that most of the MPs have moderate bioavailability (0.55-0.56), while only one yellow MPs has low bioavailability (0.11) (Figure 3).   The Bioavailability radar of 6 well-known MPs was shown in Figure 4. All six main Monascus pigments had physicochemical profiles that makes them suitable for oral administration. Moreover, the six known MPs had bioavailability values that were inside the desired range for enhanced bioavailability.

HMGR Receptor
The best molecular docking poses of the six main Monascus pigments and monacolin K (see structures in Figure 5) interactions with the active site of HMGR are shown in Table 2 and Figure 6. The results indicate affinity energies (kJ/mol) and dissociation constants (Ki) for monascin, ankaflavin, rubropunctatin, monascorubrin, rubropunctamine, monascorubramine, monankarin A, monankarin B and monacolin K ( Table 2). The findings suggest that, compared to monacolin K, six main MPs had lower inhibitory properties of HMGR. These results agree with a previous in vitro study by Jeun, et al. [43] who reported that the red and orange pigments exhibited lower inhibitory levels (36%, 15%) against HMGR compared with monacolin K (98%). On the other hand, in their in vitro experiments, the total cholesterol (TC) level of mouse serum was reduced 16% by orange pigments and 9% by monacolin K [43]; but the mechanism of in vitro regulation is not well-known, and therefore warrants further study.       In previous docking studies of statins and other chemicals with HMGR, the dissociation constants (K i ) are usually underestimated compared with the in vitro experimental data [44,45]. Therefore, the estimated affinity energies and dissociation constants (K i ) could only be useful at a qualitative level.
The best poses of each MPs and monacolin K were stabilized by hydrophobic interactions and hydrogen bonds with simvastatin binding pockets of HMGR ( Figure 6). Four MPs (monascin, rubropunctatin, rubropunctamine and monascorubramin) and monacolin K mainly occupied a hydrophobic pocket with polar groups via hydrogen bonds and hydrophobic interactions with Arg590, Asn755, His752, Leu853, Leu857, Lys691, and Val683. These interactions are similar to a previous study of docking statins and simvastatin to HMGR [45]. However, two other MPs, ankaflavin and monascorubrin mainly occupied simvastatin binding pockets of HMGR via different residues, including Ala654, Asn658, Met655, Met657, Met659, Ser661, and Val805 (Table 3). Those residues could be a new hydrophobic region that needs experimental validation.

Monascin Ankaflavin Rubropunctatin Monascorubrin Rubropunctamine Monascorubramine Monacolin
The interacting residues, number and distance of hydrogen bonds in the docking model of the six main MPs and Monacolin K within HMGR have been shown in Table 4. The functional importance of hydrogen bonds contributing to inhibition of enzyme activity in the docking experiment has been reported in many previous studies [33,46,47]. However, Monacolin K with only two hydrogen bonds showed higher inhibitory properties compared with MPs with more hydrogen bonds (3-6), suggesting hydrophobic interactions could be more important when docking to simvastatin binding pockets of HMGR. This result is in good agreement with the conclusions drawn by Ressaissi, et al. [45].

Lipase Receptor
The best molecular docking poses of the six main MPs and oleic acid within the active site of lipase is shown in Table 5 and Figure 7. Figure 5 shows the structures of oleic acid. Table 5 shows affinity energies and dissociation constants (K i ) of monascin, ankaflavin, rubropunctatin, monascorubrin, rubropunctamine, monascorubramine and oleic acid. The results suggest that compared to oleic acid, six MPs have higher lipase-inhibitory activities, and monascorubrin and rubropunctamine had the best binding affinity for lipase of all the MPs. These findings agree with previous research that three MPs (monascin, monasuore B, and ankaavin) exerted lipase-inhibitory effects [20].  Lipase inhibitors have been used as commercial anti-obesity drugs through their ability to prevent or control the hydrolysis of dietary fats into absorbable glycerol and free fatty acids [48]. In a previous in vivo studies, the weights of rats were reduced significantly by two yellow MPs, monascin and ankaflavin [49], but the weights of mice were not significantly changed by red and orange MPs [43]. The anti-obesity effects of Monascus pigment derivatives have also been reported in mice, and it is assumed that this anti-obesity effects were caused by inhibition of lipases [48]. In fact, Liu, et al. [50] and Chen, et al. [51] reported that an extract of Monascus fermented rice, Ankascin 568 (containing monascin and ankaflavin), significantly reduced the levels of serum total cholesterol and low-density lipoprotein cholesterol in human clinical studies. However, to date, the in vitro inhibitory effects on pure lipase by the six main MPs in this study have never been reported in the published literature. The outcomes of this study show that these six MPs have the potential to act as inhibitors of lipase.
The best poses of each MPs and oleic acid were mainly stabilized by hydrogen bonds, and hydrophobic interactions (Figure 7). In the previous study of Fang, et al. [20], the interactions of MPs with lipase were assumed to be located between two loops: the lid-domain loop (residues 237 to 261) and the β5 loop (residues 75 to 84), and these indirectly affected the active site. However, in this study, there were only a few interacting residues of MPs-lipase located in the two loops (only Ile249) ( Table 6). Three MPs (monascin, ankaflavin, monascorubrin,) shared similar interacting residues via hydrogen bonds, as well as hydrophobic and electrostatic interactions with Asn89, Asp 273, Asn272, Lys269, Phe336 (Table 6). Rubropunctamine had only two interacting residues (Lys269, Phe336), but had the highest binding affinity of the six MPs; whereas for monascorubramine, although it also had only two interacting residues (Ile249, Lys269), it showed the lowest binding affinity. This implies that Phe336 could play an important role in stabilizing MPs-lipase interactions.

Monascin Ankaflavin Rubropunctatin Monascorubrin Rubropunctamine Monascorubramine Oleic Acid
The interacting residues of hydrogen bonds of the six MPs and oleic acid within lipase are shown in Table 7. Oleic acid interacted with one hydrogen bond and showed lowest inhibitory property. Rubropunctamine however, did not have any hydrogen bond interaction and yet showed the highest inhibitory properties. This seems to suggest that hydrogen bonds are not important in lipase docking.

Conclusions
In this study, in silico evaluation of physicochemical properties, pharmacokinetics, and drug-likeness of 61 Monascus pigments (MPs) was done using the SwissADME tool. Physicochemical features of MPs showed desirable lipophilic drug-likeness, including molecular weight (236 to 543), TPSA (44.76 to 179.77), lipophilicity (−0.81 to 4.14), and water solubility (−4.94 to −0.77.). Most (55) of the MPs met all conditions of Lipinski's rule-of-five of the filter (6 MPs had one violation each).
Pharmacokinetics (gastrointestinal absorption, interaction with P-glycoprotein, inhibition of CYP3A4) and bioavailability predictions showed that most of the MPs had high absorption and bioavailability. However, 18 MPs were predicted to be inhibitors of cytochrome P450 3A4 (CYP3A4), one of the key oxidizing enzymes responsible for clearing xenobiotics from the body. An interference with the activity of CYP3A4 means these MPs may cause pharmacokinetics-related drug-drug interactions that may lead to adverse effects when taken together with other drugs. Among these 18 MPs were 8 red (mostly rubropunctamine, monascorubramine, and Monascopyridine derivatives) 1 orange (monapilol A), and 9 yellow (monankarin derivatives, monapurone derivatives, and rubropunctin) pigments (See Supplementary File 1). The evaluation of drug-likeness and bioavailability showed that most MPs have good drug-like properties, are suitable for oral administration, and had moderate bioavailability.
A docking model for HMGR and lipase active sites was developed using an automated docking approach. Compared to monacolin K, six main MPs had lower inhibitory properties of HMGR, and ankaflavin and monascorubrin had unique binding residues, perhaps showing binding at a new hydrophobic region. The results of lipase docking suggested that, compared to oleic acid, six MPs have higher lipase-inhibitory activities, and we found the hydrophobic Phe336 could play an important role in stabilizing MPs-lipase interactions. Although hydrogen bonds are usually important for interactions between ligand and proteins, in this study, they were not found to be important for docking of lipases.
The limitation of this study is that there are many factors involved in metabolism of lipids and control of hyperlipidemia, and that HMG-CoA reductase and pancreatic lipase [18,20] are not the only targets. It is understandable that MPs may also have inhibiting or activating effects on other proteins involved in lipid metabolism. In vitro and in vivo studies involving interactions of MPs and these other protein targets should therefore be studied to gain a complete picture of the effects of MPs on hyperlipidemia.
Moreover, in silico data are usually underestimated or overestimated when compared with in vitro and in vivo experimental data. This suggests that in silico findings need experimental validation from in vitro and in vivo studies to confirm the hypolipidemic effects of MPs. Another important recommendation is the need to for in vivo research to be done on the effects of individual MPs on the activities of HMG-CoA reductase and pancreatic lipase, thus helping to unravel how these MPs control hyperlipidemia.