Virtual Screening, Molecular Docking and Pharmacophore Modeling of Phytoconstituents of Flavones as Aldose Reductase Inhibitors

Flavones are an important class of naturally occurring heterocycles possessing various pharmacological activities. An in silico approach was carried out where 506 compounds containing flavone ring were utilised as ligand against the target aldose reductase enzyme. Aldose reductase is the rate-limiting enzyme in the polyol pathway, which indirectly causes diabetic complications like diabetic nephropathy and diabetic retinopathy. The flavone containing compounds retrieved from the PubChem were investigated by HTVS (high throughput virtual screening) followed by molecular docking using glide SP and XP docking module in Maestro of Schrodinger software. Among them, the best fifteen compounds were selected for further studies. The binding energy calculation was done using the Prime MM-GBSA module. PASS online prediction tools were used for predicting the antidiabetic activity of the compounds. Also, a pharmacophore model was generated for best interacted fifteen compounds by Phase, which can be used for evaluation of the characteristic features essential for this specific biological activity. The ADMET properties of the compounds were determined using the Qikprop module in the Schrodinger software.


INTRODUCTION
An alarming increase in diabetic cases significantly impacted the spike of both mortality and morbidity rates worldwide. In 2016, WHO reported an enormous increase in diabetic patients from 108 million to 422 million in the past decade [1]. Thus, it is essential to develop a better therapeutic portofolio for diabetes management. Moreover, diabetic complications are some the worst part which has to be controlled.
Aldose reductase plays a vital role in polyol pathway, as it is involved in rate limiting step [2]. The hexokinase pathway is used for the metabolism of sugar in average conditions. During hyperglycemia, excess glucose saturates the latter and activates aldose reductase leading to the conversion of glucose into sorbitols or other sugar alcohols [3]. This fact (pathways/ transformation) indirectly causes diabetic complications like diabetic nephropathy and diabetic retinopathy [4]. Hence, aldose reductase inhibitors will reduce the glucose flux and plays a crucial role in the management of diabetes which is scarcely available. Alrestatin, benurestat, epalrestat, fidarestat, imirestat, lidorestat, minalrestat, ponalrestat, ranirestat, risarestat, sorbinil, tolrestat, zenarestat, and zopolrestat were some of the aldose reductase inhibitors which were withdrawn during clinical trials due to various adverse effects [5]. Fever, nausea, diarrhea, increases in liver enzymes, skin rashes, including toxic epidermal necrolysis and Stevens-Johnson syndrome, marked thrombocytopenia, lymphadenopathy, splenomegaly, and adult respiratory distress syndrome were some of the main adverse effects [6].Tolrestat was withdrawn because of deaths from fatal hepatic necrosis [7].
Flavones are an essential class of oxygencontaining heterocyclic systems. The benzopyran ring system in the flavones occurs as secondary metabolites in the plant kingdom [8]. Generally, the term 'Flavonoids' are used, which possess C6-C3-C6 carbon framework having a phenylbenzopyran skeleton. Further divided into the flavonoids, the isoflavonoids and the neoflavonoids based on the position of phenyl ring linkage to the benzopyran moiety [9]. Flavones are derived from chalcones through a biosynthetic pathway utilising malonyl coA and pcoumaroyl CoA [10].
Molecular docking is efficient in silico drug design approach to predict the possible binding site of the target protein for a ligand [11]. It is a virtual screening technique where many compounds are reduced to a minimum subset of compounds having a high binding affinity towards the receptors [12].The docking procedure involves the computational method to search for an appropriate ligand into the binding pockets with minimised energy conformation [13]. By applying such computational tools, we have performed the anticancer action of flavonoids [14], COX inhibitory actions of synthetic compounds [15] and in continuation of the in silico studies, this study focuses on developing flavones as leads as aldose reductase inhibitors.

MATERIALS AND METHODS
The flavones were retrieved from PubChem obeying Rule of Five properties, and 506 compounds were chosen. The target protein aldose reductase was selected as 3RX2 from protein data bank for the specified study.

Ligand Preparation
The ligands were imported to LigPrep and preparation and minimization was done by using the Ligprep module [16,17].

Protein Preparationand Receptor Grid Generation
Aldose reductase preparation was realized using the wizard of the Schrodinger software for the molecular docking studies. The threedimensional structure of the target proteins was obtained from the RCBS (PDB ID:3RX2) [18]. Preprocess was done by assigning hydrogen orders. The optimization was also carried out by hydrogens assignment and water removal. Then minimization was carried out to ease the docking process. Grid generation is an essential feature in molecular docking. The active site was determined, and grid generation was done [19].

Molecular Docking
Molecular docking is used to determine the interaction between the ligand and the target protein. 339 compounds were performed for glide SP docking followed by glide XP docking. In the GLIDE ligand docking module, the corresponding generated grid of the receptor and ligprep files were inserted, anddocking was performed [17,21].The ligands which interacted with 3RX2 by XP docking method were listed in Table 1.

Pharmacophore Modeling
A pharmacophore model was generated for bestinteracted compounds by PHASE [17]. It can investigate the characteristic features essential for biological activity. The generated five featured pharmacophores in this study showed three main elements: hydrogen bond acceptors (HBAs), hydrogen bond donors (HBDs) and aromatic rings (AR). HBDs are shown in blue, HBAs in pink and ARs in orange. The three and four featured pharmacophore hypotheses were rejected due to the low value of survival score. They were unable to define the entire binding space of the selected molecules. Five featured pharmacophore hypotheses were established and subjected to rigorous scoring function analysis. Common pharmacophore hypotheses (CPH) were searched, which included at least five sites common-to-all molecules. Further, the best CPH was selected depending on the survival score until at least one hypothesis was found and scored successfully. Pharmacophorematching tolerance was set to 2 A • . The highest survival score for the common pharmacophore hypothesis gives the best alignment of the active ligands to this hypothesis. This alignment provides fitness to all of the inhibitors [22].

Physicochemical and ADMET Studies
The physicochemical and ADMET properties of the compounds were determined by using the Qikprop module in the Schrodinger software [17,[23][24]. ADME properties prediction before expensive experimental processes can eliminate unnecessary testing on compounds that will probably fail in clinical trials.

Antidiabetic Activity Prediction in PASS Online
Prediction of antidiabetic activity for the selected 15 flavones wasmade with the help of the computer program PASS (Prediction of activity spectra for substances) [25]. PASS is a computer-based program used to predict different pharmacological activities for other substances, including phytoconstituents. Prediction of this spectrum by PASS is based on structural activity relationship (SAR) analysis of the training set containing more than 205,000 compounds exhibiting more than 3750 kinds of biological activities. The predicted activity spectrum of a compound is estimated as probable activity (Pa) and probable inactivity (Pi). The compounds showing more Pa value than Pi are the only constituents considered as possible for a particular pharmacological activity.

Virtual Screening and Molecular Docking
Among the 506 selected compounds, the high throughput virtual screening produced 339 compounds against aldose reductase enzymes. The docking scores and interactions are given in Table 2 and Fig. 1(a-b), 2(a-b).

Pharmacophore Hypothesis Generation and Modeling
The results of all featured pharmacophore hypotheses are presented in Table 3. The first hypothesis, AADRR_1, is the best hypothesis in this study, characterised by the highest survival score (4.8906), which consists of two hydrogen bond acceptors (A), one hydrogen bond donors (D), and two aromatic rings (R) (Fig. 2). The distances between the sites in the common pharmacophore hypothesis AADRR_1 are given in Figs. 3(a-c) and Table 4 and 5.

ADMET and Physicochemical Properties
The selected flavonoids were screened for their physicochemical and ADMET properties and the results are portrayed in Tables 6. The predicted properties state that all the flavonoids, except naringin, showed values within the permissible limit, confirming that they might act as good oral bioavailable drugs.

PASS Predictions for Antidiabetic Activities
The anticancer activity spectra of myricetin and quercetin were obtained by online PASS version.
These predictions were interpreted and used in a flexible manner. PASS predicted probable activity (Pa) of myricetin and quercetin for different cancer targets was predicted and reported in the given Table 7.        Leachianone G 14.

) Distances of Pharmacophore Hypothesis AADRR_1
studies proved to be a better methodology for predicting biological activity. In this study, 506 flavones were undergone virtual screening for the aldose reductase inhibitory action and found that 339 flavones interacted and on further screening igands interacted well. Among them tanariflavanone D shows the highest XP docking score with aldose reductase. Thus, this study concludes that flavones possess antidiabetic action, which has to be further confirmed by vitro and in vivo studies. The resu utilised for the drug development process and aid to develop newer medicinally important moiety.

CONSENT
It is not applicable.

ETHICAL APPROVAL
It is not applicable.