In Silico Evaluation of 5‐Arylidine Glitazone Esters as Potential Antidiabetic Agents: ADMET, Molecular Docking, Dynamics, MMGBSA and DFT Studies

ABSTRACT

Diabetes Mellitus remains a severe cause of death globally. In our present study, we report an in silico study of our previously synthesized 5‐arylidine glitazone esters 3ai–ev to find an alternative treatment for T2DM. To this end, computational methods such as ADMET, Molecular docking, dynamics, MMGBSA, and DFT were employed to investigate drug‐like characteristics, safety, binding affinity, stability, free binding energy, and the electronic properties of compounds 3ai‐ev. The results showed that all compounds exhibited favorable physicochemical and pharmacokinetic properties. Molecular docking showed that all compounds are the best inhibitors of SGLT2 when compared to other enzymes. Alaninate 3biii emerged as the most potent inhibitor with a docking score of −9.322 kcal/mol, followed by butanoate 3civ, norvalinate 3eiii, valinate 3div, and glycinate 3av with docking scores of −9.322, −8.787, −8.710, and −8.135 kcal/mol. Parento algorithm trade‐off analysis between synthetic yield and binding affinity confirmed that compound 3biii has a higher synthetic yield than all compounds. In addition, molecular dynamics confirmed the stability of compounds 3ai‐ev using both RMSD and RSMF fluctuations, while MMGBSA revealed favorable free binding energy. Furthermore, DFT provided the acceptable electronic properties using the HOMO‐LUMO energy gap, which is very significant towards the development of new anti‐diabetic drugs.

This study presents an in silico evaluation of 5‐Arylidine Glitazone Esters 3ai‐ev as Potential Antidiabetic Agents utilizing ADMET, Molecular Docking, Dynamics, MMGBSA, and DFT Studies. The drug‐like characteristics, safety, binding affinity, stability, free binding energy, and electronic properties of 31 compounds were investigated. The results showed that compounds 3biii, 3civ, 3eiii, 3div, and 3av emerged as the most potent inhibitors against SGLT2.

1. Introduction

Diabetes mellitus (DM) is a chronic metabolic disorder that occurs when the blood glucose is too high due to a lack of insulin production in the pancreas [1]. Type 2 diabetes (T2DM) is more prevalent, accounting for 90 %, followed by type 1 diabetes mellitus (T1DM) (10%) and gestational diabetes (GDM), accounting for 7% of all diabetes cases [2]. According to a report by the IDF, it is estimated that in 2021 alone, around 537 million people were living with diabetes globally, with healthcare expenses estimated at around $966 billion. This number is projected to increase to 783 million by 2045 [3]. The treatment of diabetes involves using insulin injections for type 1, while type 2 uses oral anti‐diabetic medications [4]. Among these oral medications, Rosiglitazone, Voglibose, Miglitol, Sitagliptin, and Dapagliflozin (Figure 1) are the current anti‐diabetic medications approved to treat type 2 diabetes [5]. However, challenges such as adverse side effects and insulin resistance have limited the use of these medications [6, 7].

FIGURE 1.

Some known oral anti‐diabetic medications.

Given challenges with these current treatments, DM has become a disease which is getting out of control [8]. Hence, the solution for this problem is an urgent need to introduce and develop new alternative anti‐diabetic drugs which are effective and have fewer side effects. Glitazones, also known as 2,4‐thiazolidinediones, are five‐membered heterocyclic compounds that have recently gained much attention due to their wide range of biological properties, such as anti‐diabetic, anti‐cancer, and antimalarial, amongst others [9, 10]. A recent study by Gharge and co‐workers on design, synthesis, characterization and antidiabetic evaluation of 3,5‐substituted thiazolidinediones 1 and found that ethoxy glitazone was the most potent against α‐glucosidase and α‐amylase with IC₅₀ of 86.06 ± 1.1 µM and 74.97 ± 1.23 µM, respectively, as shown in Figure 2 [11]. Another study by Ibrahim et al. revealed that quinoline‐thiazolidinedione hybrid 2 was potent against PPAR‐γ Agonists and found that this hybrid reduced glucose levels by 23% after 15 days of treatment with a single oral dose [12]. While there have been numerous studies exploring glitazones as an antidiabetic agent, there is still a gap in exploring their detailed computational or in silico profile.

FIGURE 2.

Recently reported glitazones with anti‐diabetic properties.

In silico methods have recently emerged as important tools in providing high‐throughput screening and in drug design and optimization using large chemical libraries [13, 14]. More interestingly, these methods are capable of performing various complex calculations at high speed at reduced costs, thereby providing important information that guides the in vitro and animal testing through prioritization during candidate drug trials. In continuation of our previous study on the design and synthesis of glitazone derivatives with anti‐diabetic properties [15, 16, 17, 18, 19]. Our initial in vitro screening revealed that our previously tested glitazones 3 (Figure 3) did not show significant α‐glucosidase activity, having 51.49% average inhibition [15]. In the present study, our aim is to perform the computational simulation of our synthesized 5‐arylidine glitazone esters 3 to see if they have the potential to inhibit α‐amylase, PPAR‐γ, DPPIV, and SGLT2 enzymes in an attempt to find new potent anti‐diabetic drugs. Molecular docking, dynamics, MMGBSA, ADMET, and DFT simulations were conducted with a clear objective to investigate the binding affinity, stability, interactions, safety, drug‐like characteristics, physicochemical and pharmacokinetic properties, as well as their electronic properties, which are very significant in drug discovery and development of new oral Type 2 antidiabetic drugs.

FIGURE 3.

Our previously synthesized glitazone esters 3.

2. Materials and Methods

2.1. Ligand Preparation and Selection of Molecules for Study

All ethyl‐(2‐(5‐Arylidine‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev were drawn using ChemDraw and imported to the PubChem website to obtain their respective smiles. Furthermore, all the target ethyl‐(2‐(5‐Arylidine‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev were subjected to in silico computational methods as shown in Figure 4. ADMET, molecular docking, dynamics, MMGBSA, and DFT studies were employed to predict the absorption, distribution, metabolism, and excretion (ADME), toxicity, binding affinity, interactions, stability, and free energy binding, and the HOMO‐LUMO electronic properties. The detailed computational methodology is available in our supplementary materials.

FIGURE 4.

Schematic representation of ligand screening flowchart.

2.2. Chemistry

The synthesis of ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev was carried out using five synthetic protocols with yields ranging from 4%–98% as shown in Table 1, as we have previously reported [19]. All compounds were characterized using a combination of NMR, IR and HRMS spectroscopies. Their characterization spectroscopic spectra can be accessed online on https://doi.org/10.24820/ark.5550190.p011.397.

TABLE 1.

Physicochemical properties of ethyl (2‐(5‐arylidine‐2,4‐dioxothiazolidin‐3‐yl) esters 3ai–ev.

Compound	R1	R2	Yield	MW (g/mol)	Rotable bonds	NHBDs	NHBAs	MR	TPSA	LOG P	LOG S	Lipinski violation
3ai	H	Ph	50	348.37	8	1	5	92.35	118.01	1.52	−2.92	No violation
3aii		4‐MeOPh	54	378.40	9	1	6	98.84	127.31	1.53	−3.01
3aiii		4‐NO2Ph	98	393.38	9	1	7	101.17	163.90	0.84	−3.01
3aiv		4‐OMe, 3OHPh	97	394.40	9	2	7	100.87	147.54	1.14	−2.88
3av		Piperonayl	54	392.38	8	1	7	98.42	136.54	1.38	−3.06
3avi		4‐MePh	46	362.40	8	1	5	97.32	118.08	1.91	−3.24
3avii		4‐FPh	12	366.36	8	1	6	92.31	118.08	1.91	−3.09
3bi	Me	Ph	50	362.40	8	1	5	97.16	118.08	1.88	−3.25
3bii		4‐MeOPh	98	392.43	9	1	6	103.65	127.31	1.89	−3.34
3biii		4‐NO2Ph	90	407.40	9	1	7	105.98	163.90	1.19	−3.34
3biv		4‐MeO, 3OHPh	61	408.43	9	2	7	105.67	147.54	1.56	−3.22
3bv		Piperonayl	53	406.41	8	1	7	103.22	136.54	1.69	−3.40
3bvi		4‐Meph	31	376.43	8	1	5	102.13	118.08	2.13	−3.57
3bvii		4‐FPh	11	380.39	8	1	6	97.12	118.08	2.11	−3.42
3bviii		Furanyl	34	352.36	8	1	6	89.43	131.22	1.16	−2.60
3ci	Et	Ph	10	376.43	9	1	5	101.97	118.08	2.12	−3.60
3cii		4‐OMePh	37	406.45	10	1	6	108.46	127.31	2.15	−3.69
3ciii		4‐MePh	8	390.45	9	1	5	106.93	118.08	2.53	−3.91
3civ		4‐OHPh	6	392.43	9	2	6	103.99	138.31	1.74	−2.47
3cv		Furanyl	11	366.39	9	1	6	94.23	131.22	1.53	−2.95
3di	iPr	Ph	24	390.45	9	1	5	106.77	118.08	2.42	−3.96
3dii		4‐MeOPh	38	420.48	10	1	6	113.27	127.31	2.45	−4.05
3diii		4‐Meo, 3‐OHPh	12	436.48	10	2	7	115.29	147.54	2.04	−3.92
3div		4‐OHPh	4	406.45	9	2	6	108.80	138.31	1.95	−3.83
3dv		4‐FPh	11	408.44	9	1	6	106.73	118.08	2.74	−4.13
3ei	Pr	4‐MeOPh	17	420.48	11	1	6	113.27	127.31	2.50	−3.93
3eii		4‐MePh	12	404.48	10	1	5	111.74	118.08	2.72	−4.15
3eiii		4‐OHPh	26	406.45	10	2	6	108.80	138.31	1.94	−3.71
3eiv		4‐FPh	21	408.44	10	1	6	106.73	118.08	2.72	−4.01
3ev		Furanyl	65	380.42	10	1	6	99.04	131.22	1.82	−3.19
Pioglitazone	—	—	—	356.44	7	1	4	102.17	93.59	2.61	−4.31

3. Results and Discussion

3.1. ADMET

3.1.1. Analysis of Physicochemical Properties and Drug‐Likeness

Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) is one of the most widely used aspects to assess the safety and crucial requirements of any candidate drug [20]. Recent studies have proved ADMET is one of the reliable methods which provides necessary information in investigating the drug likeness characteristics [21]. Although the in silico ADMET predictions do not replace the experimental study, it is crucial for guidance and priority during experimental studies. The Lipinski rule has emerged as one of the most important rules which is used to predict physicochemical properties such as absorption or permeation, and drug likeness characteristics of any candidate drug [22]. This rule uses five criteria to evaluate a new orally active drug that should not violate more than one of the following: less than or equal to 5 hydrogen bond donors, less than or equal to 10 hydrogen bond acceptors, a molecular weight of less than 500, and a logP (AlogP) of less than 5 and TPSA less than or equal to 140. All the target ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev showed favorable drug likeness characteristics by conforming to the Lipinski rule (Table 1) when compared to the piogitazone standard. Furthermore, all derivatives showed a molecular weight of less than 500 Da, less than 5 hydrogen bond acceptors, and less than 10 hydrogen bond donors, meeting criteria for the Lipinski rule of 5. Lipophilicity (Log P) values ranged from 0.84 to 2.74, falling within an acceptable borderline for good oral and intestinal absorption of any candidate drug. Topological polar surface area (TPSA) values of ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl) acetyl)esters 3ai–ev further supported favorable absorption, with all derivatives showing a TPSA of less than 140. Furthermore, the water solubility (Log S) of all target ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev were found to be less than 6, suggesting that these compounds have good solubility properties.

3.1.2. Pharmacokinetic Predictions

Pharmacokinetic properties are crucial for efficacy, toxicity, and understanding how drugs behave in the body. The most important predictor of pharmacokinetic properties is intestinal absorption, which is a measure of the proportion of compounds that are easily absorbed through the intestines [23]. For this study, all target ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev showed intestinal absorption ranging from 60.325%–86.019%, suggesting that they will be absorbed well through the intestines, as shown in Table 2. The human colon epithelial cancer cell line (Caco‐2) of all target ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev ranged from 0.148–1.998, suggesting that compounds would be permeable. P‐glycoprotein, which is also known as ATP‐binding Cassette (ABC) transporter, is responsible for forcing out toxins and xenobiotics from the body cells. According to the results, only compound 3aiii, among the gyclinates 3ai–3avii, was predicted to be a substrate of Pgp and an inhibitor of Pgp I. With regards to alaninates 3bi–3bviii, compound 3biii emerged as a Pgp substrate, Pgp I and II inhibitors. Most butanoate 3ci‐3cv, vallinates 3di–3dev, and norvaliates 3ei–3ev were found to be Pgp substrates and Pgp inhibitors. In addition, the skin permeability of all target compounds 3ai–ev was predicted to be higher with log kp > −2.5. The bioavailability scores for all the derivatives were 0.55, indicating that these molecules are easily available to their intended biological destinations. In addition, the BOILED‐egg and physicochemical radar analysis of the most potent inhibitor 3biii is shown in Figure 5. This analysis enables accurate predictions of gastrointestinal absorption, lipophilicity, and polar surface area, amongst others. The BOILED‐egg results for compound 3biii, as represented by a pink dot, suggest that this compound could be a substrate or inhibitor of Pgp glycoproteins with the highest probability of being absorbed in the gastrointestinal tract. Furthermore, the drug likeness descriptors of compound 3biii are also shown in Figure 5 with the red‐colored zone identified as a feasible physicochemical domain to enhance oral bioavailability (LIPO, lipophilicity; SIZE, molecular weight; INSOLU, insolubility; INSATU, saturation; and FLEX, flexibility).

TABLE 2.

Pharmacokinetic properties of ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev.

Compound	Metabolism
	Caco‐2 permeability	Intestinal absorption	Skin permeability	Pgp Substrate	Pgp I inhibitor	Pgp II inhibitor	Bioavailability
3ai	0.511	66.521	−3.149	No	No	No	0.55
3aii	0.535	63.524	−2.98	No	No	No	0.55
3aiii	0.626	63.603	−2.788	Yes	Yes	No	0.55
3aiv	0.418	63.178	−2.966	No	No	No	0.55
3av	0.672	66.980	−3.028	No	No	No	0.55
3avi	0.521	66.989	−3.147	No	No	No	0.55
3avii	0.553	65.436	−3.016	No	No	No	0.55
3bi	0.517	67.346	−3.144	No	No	No	0.55
3bii	0.496	62.181	−2.917	No	No	No	0.55
3biii	0.623	64.495	−2.815	Yes	Yes	Yes	0.55
3biv	0.386	61.835	−2.944	No	No	No	0.55
3bv	0.674	59.758	−2.937	No	No	No	0.55
3bvi	0.909	62.738	−3.032	No	No	No	0.55
3bvii	0.925	58.215	−2.913	No	No	No	0.55
3bviii	1.490	92.743	−2.735	No	No	No	0.55
3ci	1.049	63.516	−3.458	Yes	No	No	0.55
3cii	1.031	59.716	−3.284	Yes	No	No	0.55
3ciii	1.049	63.985	−3.45	Yes	No	No	0.55
3civ	0.188	55.075	−3.163	Yes	No	No	0.55
3cv	0.055	62.226	−3.198	No	No	No	0.55
3di	1.117	64.125	−3.424	Yes	Yes	No	0.55
3dii	1.099	60.325	−3.287	Yes	Yes	No	0.55
3diii	1.998	81.092	−2.735	No	No	No	0.55
3div	0.171	55.684	−3.13	Yes	No	No	0.55
3dv	1.130	62.237	−3.337	Yes	Yes	No	0.55
3ei	1.038	70.967	−3.296	Yes	No	No	0.55
3eii	1.417	86.019	−2.735	No	No	No	0.55
3eiii	0.148	66.326	−3.157	Yes	No	No	0.55
3eiv	1.069	72.880	−3.362	Yes	No	No	0.55
3ev	0.79	73.477	−3.23	No	No	No	0.55
Pioglitazone	1.022	91.713	−2.616	No	Yes	No	0.55

FIGURE 5.

BOILED‐egg and physicochemical radar of the most potent alaninate 3biii inhibitor.

3.1.3. Distribution and Excretion

Volume of distribution (VDss) is one of the most important pharmacokinetic parameters that deals with the total dose of theoretical volume of drug needed to be uniformly distributed to give the same concentration as in blood plasma [24]. If the Vdss is higher than 0.281 (log Vdss 0.45), this means that the drug will be distributed in tissue rather than in plasma. In contrast, all our target ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev had VDss ranging from −0.188 to −0.985, which is lower than 0.281, suggesting that these compounds will be distributed more in plasma than in tissues. On the other hand, the blood‐brain barrier (BBB) is one of the most important parameters which protects the brain from exogenous compounds. In most cases, BBB permeability is measured by carrying out in vivo experiments in animal models as the logarithmic ratio of brain to plasma drug concentrations (log BB). BB is considered to cross the blood‐brain barrier if log BB is greater than 0.3 for any given compound. All target ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai‐ev showed log BB < −1.2, suggesting that they might be poorly distributed to the brain, as shown in Table 3. In addition, one of the equally most important processes in candidate drug examination is excretion, defined as the removal of a drug from the body through the kidneys and other routes. OCT2 and total clearance are the most important parameters which are used to assess the excretion of any drug candidates. The results showed that all target ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev do not act as OCT2 substrates, revealing their poor clearance from the body.

TABLE 3.

Distribution and excretion properties of ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev.

Compound	Distribution		Excretion
	VDss	BBB permeability	Total clearance	Renal OCT2 substrate
3ai	−0.238	−0.803	0.122	No
3aii	−0.355	−1.032	0.158	No
3aiii	−0.441	−1.301	0.027	No
3aiv	−0.523	−1.181	0.142	No
3av	−0.458	−1.209	−0.067	No
3avi	−0.216	−0.796	0.066	No
3avii	−0.358	−1.004	−0.023	No
3bi	−0.214	−0.799	0.074	No
3bii	−0.270	−1.027	0.117	No
3biii	−0.474	−1.278	−0.022	No
3biv	−0.448	−1.084	0.101	No
3bv	−0.426	−1.278	−0.113	No
3bvi	−0.188	−0.866	0.021	No
3bvii	−0.328	−1.074	−0.067	No
3bviii	−0.985	0.703	0.595	No
3ci	−0.348	−0.615	0.108	No
3cii	−0.481	−0.81	0.14	No
3ciii	−0.322	−0.591	0.052	No
3civ	−0.312	−1.192	−0.01	No
3cv	−0.476	−1.061	0.198	No
3di	−0.383	−0.48	0.094	No
3dii	−0.501	−0.675	0.125	No
3diii	0.006	0.462	−21.604	No
3div	−0.348	−1.155	−0.025	No
3dv	−0.513	−0.664	−0.055	No
3ei	−0.517	−0.764	0.166	No
3eii	−2.313	0.277	1.111	No
3eiii	−0.363	−1.226	0.016	No
3eiv	−0.527	−0.754	−0.015	No
3ev	−0.523	−1.044	0.224	No
Pioglitazone	0.02	−0.614	−0.043	Yes

3.1.4. Metabolism and Toxicity

Metabolism refers to the bioconversion of a drug in the body so that it can be absorbed and eliminated easily. In most cases, the metabolic processes that involve the drug occur in the liver, where the enzymes facilitate biological reactions and transformation. Cytochrome P450 has been identified as an important detoxification enzyme in the body, located in the liver [25]. Furthermore, this enzyme is known to be converted to its different iso forms, namely CYP2D6, CYP3A4, CYP1A2, CYP2C19, and CYP2D6. The results showed that most target ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev did not inhibit CYP2D6, CYP3A4, CYP1A2, CYP2C19, and CYP2D6 bioenzymes when compared to the standard pioglitazone, suggesting that they will be metabolized more easily, as shown in Table 4a.

TABLE 4a.

Predicted Metabolism of ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev.

Compound	Metabolism
	Substrates		Inhibitors
	CYP2D6	CYP3A4	CYP1A2	CYP2C19	CYP2C9	CYP2D6	CYP3A4
3ai	No	No	No	No	No	No	No
3aii	No	No	No	No	No	No	No
3aiii	No	Yes	No	No	No	No	No
3aiv	No	No	No	No	No	No	No
3av	No	Yes	No	No	No	No	No
3avi	No	No	No	No	No	No	No
3avii	No	No	No	No	No	No	No
3bi	No	No	No	No	No	No	No
3bii	No	Yes	No	No	No	No	No
3biii	No	Yes	No	No	No	No	No
3biv	No	No	No	No	No	No	No
3bv	No	Yes	No	No	No	No	No
3bvi	No	Yes	No	No	No	No	No
3bvii	No	Yes	No	No	No	No	No
3bviii	No	No	No	No	No	No	No
3ci	No	No	No	No	No	No	No
3cii	No	No	No	No	No	No	No
3ciii	No	No	No	No	No	No	No
3civ	No	No	No	No	No	No	No
3cv	No	No	No	No	No	No	No
3di	No	Yes	No	No	No	No	No
3dii	No	Yes	No	No	No	No	No
3diii	No	No	Yes	No	No	No	No
3div	No	No	No	No	No	No	No
3dv	No	Yes	No	No	No	No	No
3ei	No	Yes	No	No	No	No	No
3eii	No	No	Yes	No	No	No	No
3eiii	No	No	No	No	No	No	No
3eiv	No	No	No	No	No	No	No
3ev	No	No	No	No	No	No	No
Pioglitazone	No	Yes	Yes	Yes	No	No	No

Toxicity is defined simply as a state where the drug can poison the body. The parameters involved in toxicity assessments include AMES, maximum tolerated dose (MRTD), human Ether‐à‐go‐go Related Gene (HERG I and II), Acute oral toxicity (ORAT), Hepatoxicity and T. Pyroformis. AMES is used to assess mutagenic potential. A positive test usually means that a compound is mutagenic, acting as a carcinogen. The results revealed that among ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev, most derivatives were found to be non‐mutagenic as summarized in Table 4b. In addition, the MRTD results of compounds 3ai–ev ranged from −0.231 to 0.699, which is acceptable. All the target ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai‐ev did not inhibit the potassium channels encoded by HERG I and HERG II, suggesting that they may cause ventricular arrhythmia. Furthermore, all compounds displayed acceptable ORAT values and tested positive against hepatotoxicity, suggesting that they may disrupt the normal function of the liver. With regard to the toxicological endpoint, T. Pyriformis, also known as protozoan bacteria, the predicted results of compounds 3ai–ev were found to be higher than expected, ranging from 0.269 to 2.482, suggesting that they may be toxic.

TABLE 4b.

Predicted toxicity result of ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai‐ev.

Compound	Toxicity
	AMES	MRTD	HERG I	HERG II	ORAT	Hepatotoxicity	T. Pyriformis
3ai	No	0.551	No	No	2.705	Yes	0.579
3aii	No	0.601	No	No	2.715	Yes	0.411
3aiii	Yes	0.287	No	No	2.403	Yes	0.438
3aiv	No	0.599	No	No	2.672	Yes	0.307
3av	Yes	0.259	No	No	2.728	Yes	0.367
3avi	No	0.484	No	No	2.752	Yes	0.582
3avii	No	0.631	No	No	2.784	Yes	0.450
3bi	No	0.482	No	No	2.767	Yes	0.585
3bii	No	0.699	No	No	2.792	Yes	0.391
3biii	Yes	0.150	No	No	2.429	Yes	0.441
3biv	No	0.677	No	No	2.632	Yes	0.297
3bv	No	0.234	No	No	2.655	Yes	0.402
3bvi	No	0.439	No	No	2.675	Yes	0.692
3bvii	No	0.604	No	No	2.727	Yes	0.519
3bviii	Yes	0.51	No	No	2.482	No	0.285
3ci	No	0.19	No	No	2.872	Yes	0.576
3cii	No	0.139	No	No	2.95	Yes	0.448
3ciii	No	0.122	No	No	2.913	Yes	0.590
3civ	No	−0.060	No	No	2.17	Yes	0.351
3cv	No	0.610	No	No	3.247	Yes	0.209
3di	No	−0.023	No	No	2.88	Yes	0.552
3dii	No	−0.085	No	No	2.989	Yes	0.444
3diii	Yes	0.519	No	No	2.482	No	2.482
3div	No	−0.231	No	No	2.25	Yes	0.379
3dv	No	−0.060	No	No	3.011	Yes	0.469
3ei	No	0.015	No	No	3.024	Yes	0.474
3eii	Yes	0.703	No	No	2.482	No	0.285
3eiii	No	−0.132	No	No	2.246	Yes	0.389
3eiv	No	0.037	No	No	3.067	Yes	0.506
3ev	No	0.504	No	No	3.274	Yes	0.269
Pioglitazone	No	0.505	No	No	2.314	Yes	1.0078

3.2. Molecular Docking

Molecular docking was carried out to investigate the interaction between the proteins and the ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–ev in comparison with the Epigallocatechin gallate (EGCG) and Pioglitazone standards. Docking uses binding energy to predict the bioactivity of a compound, where the lowest binding energy translates to more activity. For this study, all compounds 3ai–ev were docked against α‐glucosidase, α‐amylase, PPAR‐γ, DPPIV, and SGLT2 in comparison with the experimental in vitro activity and the standards EGCG and pioglitazone, as shown in Table 5. Although it is well known that SGLT2 inhibitors are not a first‐line treatment for T2DM, with reports of poor intestinal absorption, high costs, and risk of urinary/genital infections, they were selected in this study for the unique mechanism of action in blocking glucose reabsorption in the kidneys. In addition, these drugs can be used with other diabetes medications like metformin or GLP‐1 agonists, which is a big advantage. The results showed good correlation between the experimental α‐glucosidase inhibition and binding affinity to the α‐glucosidase. Generally, all compounds did not show significant activity against α‐glucosidase as confirmed by the experimental results at 200 µM. However, all the glycinates 3ai–vii were not done (ND), whereas all the rest of the compounds displayed α‐glucosidase inhibition ranging from 31.83%–51.49%. Notably, hydroxy‐Phenyl butanoate 3civ and Furanyl norvalinate 3ev were found to be the best inhibitors of α‐glucosidase, which was consistent with docking scores of −4.685 and −4.385 kcal/mol. This suggests that the presence of oxygen plays a crucial role in α‐glucosidase activity.

TABLE 5.

Experimental and molecular docking results of ethyl (2‐(5‐arylidene‐2,4‐dioxothiazolidin‐3‐yl)acetyl)esters 3ai–e.

Compound	In vitro inhibition (%)	Docking scores (Kcal/mol)
	α‐lucosidase (200 µM)	α‐glucosidase	α‐amylase	PPAR‐γ	DPPIV	SGLT2
3ai	ND	−4.805	−4.631	−7.252	−5.399	−7.456
3aii		−5.307	−6.825	−7.559	−5.428	−7.914
3aiii		−4.907	−4.431	−6.908	−5.144	−7.714
3aiv		−5.611	−7.729	−7.744	−5.958	−7.935
3av		−5.463	−6.368	−7.614	−5.383	−8.135
3avi		−5.087	−6.093	−6.848	−5.043	−7.303
3avii		−4.378	−4.646	−7.248	−4.965	−7.757
3bi		−4.084	−6.275	−7.497	−5.243	−8.135
3bii		−4.895	−6.642	−8.171	−5.081	−8.885
3biii		−4.875	−5.551	−7.746	−4.370	−9.322
3biv		−4.537	−6.564	−7.698	−5.282	−6.866
3bv		−4.124	−6.092	−7.975	−6.030	−8.153
3bvi		−4.537	−6.471	−7.635	−4.911	−8.000
3bvii		−4.018	−6.143	−7.860	−5.682	−8.270
3bviii		−4.106	−6.350	−7.474	−5.058	−7.235
3ci	33.38 ± 5.65	−4.018	−6.143	−7.860	−5.682	−7.235
3cii	ND	−4.743	−6.082	−7.909	−5.207	−8.535
3ciii		−4.293	−6.344	−7.884	−5.454	−8.885
3civ	37.69 ± 0.39	−4.685	−7.044	−7.617	−6.036	−9.290
3cv	ND	−4.226	−6.625	−7.210	−5.278	−6.891
3di	32.66 ± 4.31	−4.399	−6.133	−7.423	−5.546	−8.361
3dii	ND	−3.901	−6.173	−7.376	−5.021	−7.608
3diii	29.66 ± 4.31	−4.793	−7.246	−7.258	−5.211	−8.654
3div	ND	−4.797	−6.570	−7.617	−5.636	−8.710
3dv		−4.710	−7.190	−7.491	−5.016	−7.712
3ei		−5.093	−6.761	−7.703	−5.283	−7.972
3eii		−4.405	−5.437	−7.722	−4.968	−8.618
3eiii	31.83 ± 2.85	−4.970	−6.697	−7.409	−5.605	−8.787
3eiv	ND	−4.986	−5.826	−7.651	−5.280	−8.117
3ev	51.49 ± 5.05	−4.375	−6.655	−7.533	−5.649	−6.682
EGCG	97.03 ± 0.41	ND	ND	ND	ND	ND
Pioglitazone	ND	−2.791	−3.578	−4.258	−5.950	−7.999

The heatmap of the docking scores for the five receptors of the thiazolidinedione ester compounds was plotted to enable a comparative visualization of their binding affinities, as shown in Figure 6. In the heatmap, light yellow to orange colours represent weaker binding energies, whereas dark blue to black colours indicate stronger binding affinities. Visual inspection of the heatmap reveals that the thiazolidinedione esters display generally weak interactions against α‐glucosidase, followed by relatively moderate activity against DPP‐IV, with average docking scores clustering of −6 kcal/mol. In contrast, these compounds exhibit substantially stronger binding energies toward SGLT2 and PPAR‐γ, indicating a preference for these targets. The compounds were clustered into five structural classes (glycinate to norvalinate) based on the R¹ substitution for simple identification. Within the SGLT2 target, glycinated compounds 3ai‐3avii showed their strongest interaction with compound 3av (−8.135 kcal/mol), while in alaninates 3bi–3bviii the most potent inhibitor was 3biii (−9.322 kcal/mol). For butanoate 3ci–3cev, compound 3civ exhibited the strongest binding energy (−9.290 kcal/mol), whereas among valinates 3di–3dev, compound 3div had the highest activity (−8.710 kcal/mol), and with norvalinates 3ei–3ev, compound 3eiii had the highest binding affinity of −8.787 kcal/mol. These findings further support that, although both theoretical and experimental studies indicate limited activity of these compounds against α‐glucosidase, they demonstrate considerable potential as inhibitors of SGLT2 and PPAR‐γ, suggesting their promise as multitarget anti‐diabetic scaffolds warranting further biological validation.

FIGURE 6.

Heatmap representation of docking scores of thiazolidinedione ester derivatives against five antidiabetic receptors.

Ligand efficiency (LE) and lipophilicity efficiency (LLE) were also calculated for the thiazolidinedione ester compounds against the SGLT2 receptor and are presented in Figure 7a,b, respectively. Ligand efficiency is an important metric as it describes the binding affinity of a compound relative to its molecular size, providing insight into how effectively each atom contributes to target interaction. Lipophilicity efficiency, on the other hand, offers valuable information on the balance between binding potency and lipophilicity, thereby helping to assess the drug‐like quality and developability of the compounds. Inspection of Figure 7a reveals a clear trend in which ligand efficiency increases as the SGLT2 docking scores become more negative, with an R² value of 0.6629, indicating a moderate to strong correlation between improved binding affinity and ligand efficiency. Similarly, Figure 6b shows that lipophilicity efficiency also increases with more favourable (more negative) docking scores, with an R² value of 0.602, suggesting that enhanced binding is accompanied by improved lipophilicity–potency balance. The calculated LLE values ranged approximately from 5.0 to 8.0, which is generally considered a favourable range for lead‐like compounds. In Figure 7c,d, Pareto front analyses were performed by plotting synthetic yield (%) against docking score and lipophilicity efficiency, respectively. A Pareto front represents the set of non‐dominated compounds for which no other molecule simultaneously outperforms them in both optimisation criteria, thereby highlighting optimal trade‐offs between potency and synthetic accessibility. In the yield versus docking score plot (Figure 7c), compounds 3biii and 3bii were identified on the Pareto front, as they combined high synthetic yields (above 90%) with strong binding affinities (docking scores below −8.5 kcal/mol), suggesting that these molecules offer an attractive balance between potency and synthetic feasibility. In the yield versus LLE Pareto analysis (Figure 7d), compounds 3div, 3cii, 3ci, 3dv and 3ev were highlighted. Notably, 3div and 3cii displayed relatively low synthetic yields but higher LLE values, indicating efficient potency–lipophilicity balance despite synthetic challenges. In contrast, 3ci and 3dv exhibited both low yields and low LLE values, reflecting less favourable overall profiles. Compound 3ev stood out by combining a comparatively high synthetic yield with one of the lowest LLE values, suggesting that its Pareto ranking was driven primarily by its synthetic accessibility rather than optimal efficiency metrics. Overall, this analysis demonstrates that the Pareto‐based unmasking was governed by trade‐offs between synthetic feasibility and efficiency‐based binding performance, enabling identification of structurally diverse candidates with complementary optimisation strengths.

FIGURE 7.

Ligand efficiency, lipophilicity efficiency and Pareto front analyses of thiazolidinedione ester derivatives against SGLT2.

The key intermolecular interactions of the best‐docked molecules from each structural class against SGLT2 were visualised and are presented in Figure 8. Compound 3av exhibited π–π stacking interactions with the aromatic residues Trp291 and Tyr290. For 3biii, a single hydrogen bond was observed with Gln457. Compound 3civ formed three hydrogen bonds with Asp454, Asn75, and His80, indicating a strong polar interaction network within the binding pocket. In the case of 3div, three hydrogen bonds were detected with Asp454, Tyr526, and Glu457, in addition to one π–π stacking interaction involving Phe96. Similarly, 3eiii established three hydrogen bonds, with its hydroxyl group forming two hydrogen bonds with Tyr526 and Asp454, and an additional hydrogen bond formed between its carbonyl group and Gln457. Furthermore, 3eiii also showed a π–π stacking interaction with Phe96. These interaction patterns suggest that residues Asp454, Gln457, Tyr526, Phe96 and Trp291 play critical roles in ligand recognition and stabilization within the SGLT2 binding cavity, highlighting their importance as key anchoring points in the active site.

FIGURE 8.

Molecular interaction profiles of the best‐docked thiazolidinedione ester derivatives from each structural class within the SGLT2 binding site.

The binding poses of all compounds were further visualized to evaluate their spatial orientation and consistency within the SGLT2 active site. These poses are presented in Figure 9 together with the native co‐crystallized ligand, which is shown in black for reference. This comparative visualization highlights the degree of overlap and alignment between the thiazolidinedione ester derivatives and the native ligand, providing insight into their binding mode similarity and structural compatibility with the active site.

FIGURE 9.

Superimposed binding poses of thiazolidinedione ester derivatives within the SGLT2 active site, shown together with the native co‐crystallized ligand (black).

3.3. Molecular Dynamics

3.3.1. RMSD

Molecular dynamics (MD) simulations were performed for the best‐docked ligand from each structural class in complex with the SGLT2 protein to provide insight into the stability of the ligand–protein interactions over time. The root mean square deviation (RMSD) profiles of the complexes are presented in Figure 10. In all simulations, the protein backbone remained stable, with RMSD values consistently below 4.0 Å, indicating overall structural stability of the receptor, with the exception of the 3civ complex, which showed a gradual increase in RMSD toward the end of the simulation period.

FIGURE 10.

RMSD profiles of SGLT2–ligand complexes obtained from molecular dynamics simulations of the best‐docked thiazolidinedione ester derivatives.

The ligands exhibited varying degrees of RMSD fluctuations throughout the simulations. However, these fluctuations remained consistently lower than the corresponding protein RMSD values, suggesting that the ligands were well accommodated within the binding pocket and maintained stable binding orientations. Among the studied compounds, 3div displayed the most stable ligand RMSD profile, showing no abrupt fluctuations and reaching a well‐defined equilibrium. This behaviour suggests a highly stable binding mode and strong persistence of key intermolecular interactions throughout the simulation, supporting the reliability of its docking‐predicted binding pose.

3.3.2. RMSF

The root mean square fluctuation (RMSF) of the protein residues was further analyzed to evaluate residue‐level flexibility and ligand‐induced effects, and the results are presented in Figure 11. Overall, the terminal regions of the protein exhibited the highest fluctuations, as expected due to their intrinsic mobility. Residues in the range 200–300 showed moderate fluctuations for most complexes; however, for compounds 3biii and 3eiii, these residues remained relatively stable, suggesting that these ligands may confer additional local stabilization to this segment of the binding site. Notably, compound 3eiii displayed pronounced fluctuations in residues 0–45, indicating potential mobility or conformational adjustments at the N‐terminal region upon ligand binding. These observations provide insights into the dynamic behavior of SGLT2 in the presence of different thiazolidinedione ester derivatives and highlight specific regions that may contribute to ligand‐induced conformational flexibility.

FIGURE 11.

Residue‐level flexibility of SGLT2 in complex with thiazolidinedione ester derivatives, assessed by RMSF analysis from molecular dynamics simulations.

3.3.3. Interaction Bonds

Interaction fraction analyses were also performed to investigate the types of intermolecular interactions occurring during the simulations and to identify the protein residues engaged with the ligands (reported in Figure 12). For compound 3av, the interactions were relatively evenly distributed across different types of bonds, but overall, fewer residues at the active site were involved, indicating limited engagement with the ligand. A similar trend was observed for 3biii, where many active site residues remained largely uninvolved; however, the nature of the dominant interactions differed between the two compounds, with 3av showing predominantly hydrogen bonding, while 3biii was dominated by hydrophobic interactions. In contrast, 3civ exhibited extensive engagement with the active site, forming multiple hydrogen bonds, hydrophobic contacts, and water‐bridge interactions with several residues, suggesting a highly stabilised binding mode. The remaining two compounds (3div and 3eiii) showed fewer interactions compared to 3civ, indicating more limited residue engagement. Notably, ionic interactions were generally absent, except for minor contributions from Asp454 in the 3av complex.

FIGURE 12.

Interaction fraction analysis of SGLT2 residues with best docked thiazolidinedione ester derivatives, showing the hydrogen bonds, hydrophobic contacts, water bridges, and ionic interactions during molecular dynamics simulations.

3.4. Molecular Mechanics, General Born Surface Area (MMGBSA)

The binding free energy (ΔG_bind) analysis of the thiazolidinedione ester derivatives complexed with SGLT2 is reported in Table  6 . Among the compounds, 3eiii exhibited the most favourable overall binding free energy of −79.14 kcal/mol, followed by 3civ with −76.97 kcal/mol and 3biii with −72.64 kcal/mol, indicating that these ligands have the strongest thermodynamic propensity for SGLT2 binding. In contrast, 3div and 3av showed relatively weaker binding energies, with −63.16 and −66.93 kcal/mol, respectively. When examining the contribution per interaction type, van der Waals (vdW) interactions were the dominant stabilising force for all compounds, ranging from −52.02 to −56.82 kcal/mol, highlighting the critical role of hydrophobic contacts within the SGLT2 binding pocket. Hydrogen bonding contributed modestly to the overall binding, with ΔG bind(Hbond) values ranging from −0.56 to −1.37 kcal/mol, indicating that while polar interactions are present, they play a secondary stabilizing role compared to van der Waals (vdW) interactions. Solvation effects, as represented by ΔG_bind(solv), were largely destabilising for all compounds, with positive contributions ranging from 19.10 to 43.34 kcal/mol. Notably, 3div displayed the largest solvation penalty of 43.34 kcal/mol, which partly explains its comparatively weaker total binding free energy despite a reasonable vdW contribution. Normalizing the binding energies per covalent atom (ΔG_bind/covalent) provides further insight into ligand efficiency at the atomic level, showing that 3eiii and 3div have higher per‐atom binding contributions (5.69 and 5.01 kcal/mol, respectively), suggesting the efficient utilization of their molecular framework in binding SGLT2.

TABLE 6.

Components of the binding free energy for thiazolidinedione ester derivatives complexed with SGLT2.

Protein	Compounds	dG Bind (kcal/mol)	dG Bind/covelent (kcal/mol)	dGBindHbond (kcal/mol)	dGBindvdW (kcal/mol)	dGBindsolv(GB) (kcal/mol)
SGLT2	3av	−66.93	2.65	−0,56	−56.82	23.98
	3biii	−72.64	4.05	−0.57	−55,20	19.10
	3civ	−76.97	4.56	−1.37	−53.98	21.48
	3div	−63.16	5.01	−1.13	−52.02	43.34
	3eiii	−79.14	5.69	−1.10	−53,58	32.08

3.5. Density Functional Theory (DFT)

DFT was performed to investigate the electronic properties of a series of compounds (3av, 3biii, 3civ, 3div, 3eiii). The total energies were evaluated to assess the relative thermodynamic stability of each compound, while the dipole moments were computed to provide insight into molecular polarity. As shown in Figure 13, compound 3biii exhibits the highest dipole moment of 7.9821 D, indicating significant charge separation within the molecule, which may influence its reactivity and interaction with polar environments. In contrast, 3div has the lowest dipole moment of 0.7563 D, suggesting a more symmetric electron distribution and potentially lower polarity‐related interactions. Regarding total energies, 3biii also shows the most negative energy at −1747.438 a.u., suggesting it is the most thermodynamically stable among the series. On the other hand, 3civ has the least negative energy of −1649.28 a.u., indicating it is comparatively less stable. The lack of direct correlation between dipole moment and total energy highlights that molecular stability is not solely determined by polarity but also by the overall electronic structure and bonding interactions. The error bars, representing an estimated 5% uncertainty, show that the calculated values are reasonable.

FIGURE 13.

Comparison of dipole moments and total energies of selected compounds evaluated by DFT.

Figure 14 presents the HOMO (highest occupied molecular orbital) and LUMO (lowest unoccupied molecular orbital) distributions of the thiazolidinedione ester derivatives. The HOMO orbitals, depicted in green and red, highlight regions of high electron density that are available for nucleophilic interactions, whereas the LUMO orbitals indicate regions susceptible to electrophilic attack. The calculated HOMO energies span from −6.01 to −8.60 eV, while the LUMO energies range from −3.43 to +0.95 eV. Consequently, the HOMO–LUMO energy gaps (ΔE) vary between 3.56 and 9.55 eV, reflecting pronounced differences in electronic stability and chemical reactivity across the series. Analysis of the orbital distributions reveals that the HOMO electron density is predominantly localized on the thiazolidinedione core and its adjacent substituents, indicating that these regions may play a central role in intermolecular interactions, such as hydrogen bonding and π–π stacking, within the SGLT2 active site. In contrast, the LUMO orbitals are generally more delocalized over the extended π‐systems, indicating potential regions for electron acceptance. Compounds with smaller HOMO–LUMO gaps, notably 3biii and 3av (≈3.6–3.7 eV), are expected to exhibit higher chemical reactivity, which may facilitate stronger binding interactions. In contrast, compound 3civ displays a markedly larger energy gap (9.55 eV) and a positive LUMO energy, indicative of a highly stabilized and weakly accepting electronic structure. Although 3civ is an organic derivative, its wide gap suggests limited π‐electron delocalization and reduced charge‐transfer capability, rendering it less chemically reactive but potentially more electronically stable than the other compounds in the series.

FIGURE 14.

Frontier molecular orbital (FMO) s of the most active thiazolidinedione ester derivatives.

The molecular electrostatic potential (MEP) surfaces in Figure 15 reveal clear charge distribution patterns across the thiazolidinedione ester scaffold. The red regions, corresponding to electron‐rich zones, are mainly localized around the carbonyl oxygen atoms of the thiazolidinedione ring and ester functionalities, indicating their strong potential as hydrogen‐bond acceptor sites. In contrast, the blue regions, which represent electron‐deficient areas, are predominantly located around hydrogen‐bearing heteroatoms and proximal alkyl or aromatic regions, highlighting potential hydrogen‐bond donor interaction sites. A comparison across the five compounds shows that substitution at the R1 position modulates both the intensity and spatial distribution of electrostatic potential, with 3civ displaying a broader separation of positive and negative regions, consistent with its larger HOMO–LUMO energy gap and increased electronic stability. Compounds 3av, 3biii, 3div, and 3eii exhibited more localized negative potential around the active heteroatoms, which may favour stronger and more directional interactions with polar amino acid residues in the SGLT2 binding pocket.

FIGURE 15.

Molecular electrostatic potential (MEP) surface maps of the thiazolidinedione ester derivatives (3av, 3biii, 3civ, 3div, and 3eii). The color scale ranges from red (regions of highest negative electrostatic potential) through yellow/green (neutral regions) to blue (regions of highest positive electrostatic potential).

4. Conclusion

In this study, a series of thiazolidinedione ester derivatives 3ai–cv were evaluated using in silico approaches to assess their anti‐diabetic potential. ADMET results revealed acceptable drug likeness characteristics. Docking results showed good correlation with experimental α‐glucosidase inhibition with predicted binding affinities. Both experimental and computational findings consistently demonstrated that these compounds did not exhibit significant inhibitory activity against α‐glucosidase enzyme. Furthermore, molecular docking revealed considerably stronger binding affinities toward SGLT2 and PPAR‐γ, indicating that these molecules possess promising multitarget anti‐diabetic potential. Pareto front analyses, based on synthetic yield versus docking score and lipophilicity efficiency, identified nondominated compounds that optimally balance binding potency and synthetic accessibility, thereby strengthening the rational lead selection process. Molecular dynamics simulations confirmed the structural stability of the most promising ligand–protein complexes, supported by stable RMSD and RMSF trajectories and favorable interaction fraction profiles dominated by hydrogen bonding, hydrophobic, and van der Waals interactions. MMGBSA supported molecular dynamics with a favorable overall binding free energy. Furthermore, DFT calculations provided insights into the electronic features of the compounds, revealing HOMO–LUMO energy gaps consistent with their chemical reactivity and stability. Collectively, these results demonstrate that while these thiazolidinedione ester derivatives 3ai–cv are weak α‐glucosidase inhibitors, as they show strong potential as SGLT2 and PPAR‐γ modulators and represent promising scaffolds for the development of next‐generation multitarget anti‐diabetic agents. (Supporting Information).

Author Contributions

Kabelo P. Mokgopa: methodology, validation, formal analysis, investigation, data curation, and writing – original draft, writing – review and editing. Mofeli B. Leoma: methodology, validation, formal analysis, investigation, data curation, and writing – original draft. Tendamudzimu Tshiwawa: formal analysis, writing, review, and editing. Ndivhuwo R. Tshiluka: conceptualization, formal analysis, resources, writing – review and editing, and funding acquisition, validation, and supervision. All authors participated in the discussions

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: cbdv71426‐sup‐0001‐SuppMat.docx

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.