Investigation of metformin analogues in targeting signaling molecules of head and neck cancer pathways by computational approaches

Abstract

The aggressive progression of head and neck cancer (HNC) and the limited availability of effective targeted therapies render it as a significant global health challenge. Several key biological pathways, including PI3K/AKT/mTOR, Wnt/β-catenin, AMPK, and glycolysis, play crucial roles in the progression of HNC. Although conventional therapies such as chemotherapy remain widely utilized, there is a pressing need for innovative interventions to address the major limitations of these approaches. In response to this critical issue, this study seeks to evaluate the efficacy of the FDA-approved drug metformin as an anti-cancer agent. To identify and assess the inhibition of these biological molecules by metformin analogs, we employed a comprehensive in-silico approach that integrated molecular docking, molecular dynamics (MD) simulations, and MM/PBSA (Molecular Mechanics/Poisson–Boltzmann Surface Area) free energy calculations. PubChem compound CID 222300 emerged as a top-scoring ligand with a high binding affinity for the mTOR protein (PDB: 4JSV). A 100 ns MD simulation was conducted to evaluate the stability and conformational dynamics of the mTOR–ligand complex. The results demonstrated favourable interaction characteristics, including a consistently decreasing radius of gyration (Rg), low root mean square fluctuation (RMSF), and stable root mean square deviation (RMSD), indicating overall structural compaction and stability. Further evidence from hydrogen bond analysis revealed that the ligand contributed to complex stability by maintaining sporadic yet regular polar interactions throughout the trajectory. These findings suggest that CID 222300 is a promising lead compound capable of interacting with the mTOR target in a stable and targeted manner, indicating its potential as a novel therapeutic agent against HNC.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-49860-x.

Introduction

Cancers affecting the head and neck (HNC) exhibit heterogeneity among individual samples, involving tumors tracing their origins to distinct anatomical regions of the body. The vast majority (~ 90%) are head and neck squamous cell carcinomas (HNSCCs) of mucosal epithelial origin. Global incidence is high (~ 1–2% of all cancers). Principal risk factors are tobacco and alcohol consumption, betel chewing, and high‐risk human papillomavirus (HPV) infection¹. HNSCCs account for the majority of HNCs, representing one of the most prevalent types of human cancer, with the condition affecting various body organs such as the oral cavity, pharynx, and larynx. These biologically diverse tumor occurrences in HNSCC are further accompanied by different clinical presentations and consequential outcomes. From this arises the need for treatment strategies which carefully take into account each unique aspect of the cancer in each patient². While challenges underscoring the efficient treatment of HNC have been highlighted and discussed repeatedly, various risk factors hold paramount importance in relation to the initiation and pathogenesis of HNC. These include alcohol abuse, tobacco use, and infections by certain oncogenic viruses, such as HPV and Epstein-Barr virus³. HPV‑positive oropharyngeal carcinomas possess different tumor biology and highly improved prognosis. Even after multimodal therapy, HNC continues to be aggressive: locoregional recurrence and distant metastasis are frequent, and 5‑year overall survival is only ~ 50–60%. The significance of various factors like body mass index (BMI), tumor site, TNM stage, sleep apnoea, gender, treatment intention, and age-related treatment modalities have also been highlighted in HNC⁴. According to global cancer statistics based on estimates from the International Agency for Research on Cancer (IARC) for 36 different types of cancers in 185 countries in 2022, significant number of new cases and deaths have been attributed to cancer of the lip and oral cavity (389,485 cases, 188,230 deaths), larynx (188,960 cases, 103,216 deaths), nasopharynx (120,416 cases, 73,476 deaths), oropharynx (106,316 cases, 52,268 deaths), and hypopharynx (86,276 cases, 40,917 deaths)⁵. Based on 37 population-based cancer registries (PBCRs) representing different regions of India, HNC was found to constitute 26% and 8% of all cancer cases in males and females, respectively. Further, the age-standardised incidence rate (ASIR) of HNC was found to be 25.9 and 8.0 per 100,000 in males and females, respectively. India reported a higher burden of HNC than the U.S.A., Brazil, Australia, Africa, and the United Kingdom, with the most prominent cancer sites being the mouth and hypopharynx⁶. Analysis of the Scottish Cancer Registry data from 2001 to 2020 revealed no significant changes with respect to the sociodemographic profile of HNC, although incidence rates of oropharynx cancer (OPC) were marked by a considerable increase of 78%.⁷.

Apart from the possible risk factors, a complex network consisting of a number of critical signalling pathways have been found to characterize HNC pathogenesis. These include the phosphoinositide 3-kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR), Janus kinase/signal transducer and activator of transcription (JAK/STAT), hepatocyte growth factor/mesenchymal–epithelial transition factor (HGF/c-MET), nuclear factor kappa B (NF‐κB), and TP53/RB pathways^8,9. Figure 1 illustrates the different biological pathways involved in the aggressive progression of HNC. Although a range of treatment approaches exist for HNSCC, survivors still face grave threats, with lifetime risks of death by cardiac and respiratory illnesses. Development of second primary tumors often associated with smoking has also been noted in the treated patients¹⁰. As a consequence, dynamic fluctuations in treatment procedures opted for HNSCC have been witnessed in global healthcare settings. Elaboration of the molecular genetic landscape that accurately maps the tumor biology of HNSCC has provided novel opportunities for therapeutic interventions, with the integration of current knowledge of tumor biology and immunobiology aimed at discovering effective cancer therapies¹¹.

Figure 1.

Altered signalling pathways in HNC demonstrating a diverse molecular landscape involved in the development of the tumor microenvironment of the cancer (Created in BioRender. Rai, N. (2026), https://BioRender.com/nq2w35b, Concept:⁹).

This computational study thus focuses on a possible therapeutic intervention of HNC using a widely available FDA-approved drug, metformin. While metformin has been extensively investigated for its potential as an anti-cancer drug candidate as discussed in the next section of this paper, research focused on the direct utilization of metformin in the inhibition of HNC is heavily unexplored yet, as observed during an exhaustive literature analysis. In addition to studies catapulting its advantageous usage in different forms of cancer, we believe that metformin can also be directly employed for mitigating HNC. This study progresses with a clearly defined objective: investigation of direct interactions of metformin with pathways responsible for HNC pathogenesis. Thus, several proteins pertaining to PD-L1, PI3K/AKT/mTOR, AMPK, Wnt/β-catenin, p53, and glycolysis have been comprehensively evaluated for their interactions with metformin, following which it might be possible to narrow down a broad search to a few proteins that can be specifically targeted by the drug as a mechanism of arresting the spread of HNC.

A number of different biological proteins have been used as receptors in this study. To maintain scientific accuracy and validity, the target proteins were selected from published studies which have used the same proteins for prior in-silico experimental analysis. The immunogenic receptor PD-L1, represented by PDB 4Z18, was used as a novel immune target in liver cancer to evaluate the therapeutic efficacy of the drug Dostarlimab¹². In a study aiming to discover kinase inhibitors of PI3Kα for the treatment of non-small cell lung cancer (NSCLC), PDB 4JPS was used as the X-ray crystal structure of the target protein against various ligands, providing significantly positive docking scores as the results¹³. Potential mTOR inhibitors were also identified using docking analysis and binding site evaluation in PDB 4JSV and 4JT6¹⁴. In the virtual screening of naphthoquinone analogs as inhibitors of the PAM pathway, PDB 3MVH and 4JT6 were utilized as the structures for human AKT1 and mTOR, respectively¹⁵. Nootkatone was shown to inhibit glucose metabolism and suppression of stemness in human breast cancer stem cells MCF-7/SC. Inhibition of cancer stem cell proliferation was shown to be AMPK activation-mediated, which was confirmed by in-silico docking with PDB 4CFE¹⁶. Various elements of the Wnt/β-catenin signalling pathway were evaluated against potential inhibitory drug candidates in order to treat colon cancer. PDB 6MXY was utilized for the protein TP53, out of the five protein structures used in the study¹⁷. PDB 1Q5K has been the protein used among different representatives of Glycogen synthase kinase 3 beta (GSK-3β) involved in molecular docking studies aimed at identifying potential inhibitors of this kinase¹⁸. In a study evaluating the possible targeting of β-catenin in the Wnt signalling pathway by 50 isoforms of Ganoderic acid, PDB 4DJS was selected in the in-silico analysis as the protein β-catenin, with the results further revealing Ganoderic acid A as the best suited isoform on the basis of the docking results¹⁹. Another study on the analysis of bioactive compounds sourced from the seeds of Cichorium intybus L. as a potential therapy against hepatocellular carcinoma used PDB 2OCJ as the target protein p53 in the molecular docking and molecular dynamics simulation studies²⁰. A study aimed at identification of potential inhibitors of hexokinase-2 (HK-2) through in-silico analysis used the PDB 2NZT as the receptor to represent human HK-2 against the set of ligands investigated in the study²¹.

Thus, considering the general well-being of the human population and the current global emphasis on sustainability under the sustainable development goals (SDGs), this field of study has cemented its importance in the mainstream of global focus. This domain is intricately aligned with several SDGs put forth by the 2030 Agenda for Sustainable Development. A total of four SDGs are associated with the need to study the current incidence of HNC and plausible treatment methodologies. The first is SDG 3, which is concerned with good health and well-being. Development of efficient therapeutic regimens based on precision medicine, along with enhanced diagnostic procedures, can be instrumental in ensuring patient longevity and disease eradication. Next, the requirement of quality education is upheld by SDG 4. Inclusion of scientific and technical skills in foundational education curriculums, such as introduction to computational tools and data repositories, can bolster real-time outcomes of education offered in participating institutions. Further, SDG 9, which encompasses industry, innovation, and infrastructure, holds direct relevance to the study of HNC. Integration of cutting-edge computational biology tools into biomedical research and drug discovery can enhance research outcomes and promote upgradation of technological capabilities. Lastly, SDG 17, which emphasizes global partnerships for achievement of the SDGs, holds prominence in reiterating the importance of large scale collaborations among developing and developed nations in a bid to achieve the aforementioned goals consistent with the progressive framework of all nations.

Metformin as an anti-cancer agent

Metformin, a cost-effective and safe-to-use biguanide derivative, is widely employed as a first-line drug for the treatment of type 2 diabetes mellitus (T2DM). Metformin inhibits hepatic gluconeogenesis in an adenosine monophosphate-activated protein kinase (AMPK)-dependent as well as AMPK-independent manner. In addition to this inhibitory effect, metformin alters the gut microbiome composition, maintains intestinal barrier integrity, and reduces systemic low-grade inflammation. Further, metformin also regulates lipid metabolism in the body²². Apart from its effective mechanisms of action as an anti-diabetic drug, metformin has found several other diverse applications, including its use as a potential anti-cancer agent. The major anti-cancer mechanisms of metformin based on the hallmarks of cancer have been identified as de-regulation of cancer metabolism, metabolic regulation of epigenetics, suppression of cancer proliferation, inhibition of invasion and metastasis, stimulation of cell death and senescence, inhibition of cancer stemness, promoting immunity and inflammation, and positive regulation of the gut microbiome²³. Metformin has been extensively explored as an effective anti-cancer drug in various literature sources based on experimental findings and subsequent validation of the therapeutic efficiency of metformin, apart from its conventional use as an oral anti-diabetic medicine^24–26.

Treatment of HNSCC cell lines with metformin resulted in AMPK activation and mTOR inhibition, along with other prominent effects like G0/G1 cell cycle arrest, thus reducing proliferation, viability, and colony formation in vitro. Activation of AMPK by metformin also results in induction of endoplasmic reticulum stress, causing degradation of epidermal growth factor receptor (EGFR) through the c-Cbl lysosome pathway which ends in apoptosis of the cancer cells²⁷. Epidemiologic and preclinical evidence has sparked interest in exploring metformin for its potential as an anti-cancer agent. A well-designed observational study found that diabetic patients prescribed metformin had decreased rates of cancer incidence and mortality²⁸. With a history of more than 60 years of usage, it has been revealed that the mitochondria and lysosome act as the organelle targets of metformin, with the latter providing potentially novel mechanistic insights into the action of metformin through the AMPK signalling pathway. Further, metformin extends its actions beyond its anti-hyperglycaemic effects on the liver, influencing the gut microbiota and directly or indirectly regulating components of the innate as well as adaptive immune response²⁹. The primary site of action of metformin is through direct inhibition of complex 1 present in the mitochondrial respiratory chain, where metformin produces a partially time-dependent, self-limiting inhibition of the respiratory chain as part of its pharmacological effects. This limits hepatic gluconeogenesis and increases uptake of glucose in peripheral tissues³⁰. A meta-analysis consisting of 14,694 patients revealed that metformin users had an improved overall survival (OS) rate (HR 0.87), with increased probability of benefit from metformin treatment linked with patients below 65 years of age³¹. However, whether metformin actually offers benefits as an effective anti-cancer drug remains disputed. This is attributed to the failure of metformin in producing favourable results. Multiple cohort studies have identified that metformin users did not present reductions in incidence of cancer. Recent critiques have argued that earlier observational studies do have some biases like the immortal time bias, and recent randomized trials conducted for treatment of other cancers have not demonstrated preventive effect³². Therefore, the degree of clinical benefit metformin offers still maintains active interest in scientific studies. In a phase study involving 20 patients with locally advanced HNSCC (LAHNSCC), metformin was administered in different doses along with cisplatin and standard radiotherapy in order to determine the adequate dose of metformin needed to treat LAHNSCC. Although the maximum tolerable dose could not be determined due to a small sample size, the combined treatment produced a 2-year overall survival (OS) and progression-free survival (PFS) of 90% and 84%, respectively³³.

Taking into account the promising results observed from a thorough literature analysis conducted prior to the initiation of the project, along with the projected efficiency and relative safety associated with metformin, it has been selected as the target drug to be investigated as a potential molecular inhibitor of different biological pathways activated in HNC, which in turn can further validate and guide the future use of metformin as a viable option for treatment of HNC.

Materials and methods

Retrieval of compounds from repositories

The keyword “metformin” when searched as text on PubChem (https://pubchem.ncbi.nlm.nih.gov/) provided one structure with absolute identity and 89 structures bearing similarity to the original structure. Since the study aims to investigate the analogs of metformin as potential HNC inhibitors, 3D conformers of the 89 analogous structures of metformin were downloaded in the structured data file (SDF) format. However, the structure retrieval process from PubChem provided only 59 downloaded structures. The analogs based on repetitive parent compounds were removed during the download process to maintain uniqueness. In order to facilitate the molecular docking procedure, all the ligands were collectively converted to the PDBQT format through the use of the Open Babel tool. Thus, the final array of metformin analogs to be used as ligands was composed of 59 structures.

Protein structure preparation

Proteins to be utilized as receptors during the molecular docking and dynamics simulations were retrieved from RCSB PDB in the .pdb format, using the PDB ID for the search process. The retrieved proteins have been listed in Table 1. Each protein used in this study has been selected after a through literature survey related to various experimental studies of computational biology involving these proteins. This ensured that each protein used as a receptor to be acted upon by the metformin analogs has already been validated in other studies for use in similar purposes. The proteins were processed prior to the start of the docking procedure using BIOVIA Discovery Studio. Each protein was opened on the workspace of the visualizer, following which hetero atoms and water molecules were removed from the protein. Further, all chains except chain A of the protein being processed were also removed. This powerful molecular modelling tool ensured that the protein structures to be used as receptors in the study were well prepared and processed for successful docking results. The post-docking visualization and analysis of receptor-ligand complexes allowed detailed views of binding pockets as well as interaction profiles. Further, 2D visualizations and surface diagrams depicting the different receptor-ligand interactions were also performed using this tool for the purpose of inclusion in the results of the project.

Table 1.

Proteins used as receptors in the molecular docking analysis and molecular dynamics simulation studies.

S. No	PDB ID	Description	Resolution (in Å)	Role in HNC	Reference(s)
1	4Z18	Crystal Structure of Human PD-L1	1.95	PD-L1 is overexpressed in HNC and promotes immune evasion by inhibiting T-cell activation	34,35
2	4JPS	Co-crystal Structures of the Lipid Kinase PI3K alpha with Pan and Isoform Selective Inhibitors	2.20	PI3Kα mutations activate the PI3K/AKT pathway, promoting HNC cell proliferation and survival	36–38
3	3MVH	Crystal structure of Akt-1-inhibitor complexes	2.01	Akt-1 is a key effector of the PI3K pathway, involved in resistance to apoptosis and metastasis in HNC	39–41
4	4JSV	mTOR kinase structure, mechanism and regulation	3.50	mTOR regulates growth and metabolism in HNC cells and is associated with poor prognosis, inhibition of mTORC1/2 reduces HNC cell growth and enhances the effect of radiotherapy	36,42,43
5	4JT6	Structure of mTORDeltaN-mLST8-PI-103 complex	3.60		44,43
6	4CFE	Structure of full length human AMPK in complex with a small molecule activator, a benzimidazole derivative	3.02	AMPK acts as a tumor suppressor in HNC by regulating cellular energy homeostasis and inhibiting mTOR	45–47
7	6MXY	Structure of 53BP1 tandem Tudor domains in complex with small molecule UNC3351	1.62	53BP1 regulates DNA damage response and may influence sensitivity to chemoradiotherapy in HNC	48–50
8	1Q5K	Crystal structure of Glycogen synthase kinase 3 in complexed with inhibitor	1.94	GSK3β modulates Wnt signaling in HNC and contributes to tumor cell migration and epithelial-mesenchymal transition	51–53
9	4DJS	Structure of beta-catenin in complex with a stapled peptide inhibitor	3.03	β-catenin is involved in aberrant Wnt signaling, contributing to HNC progression and invasion	54–56
10	2OCJ	Human p53 core domain in the absence of DNA	2.05	p53 is frequently mutated in HNC, leading to impaired DNA repair and evasion of apoptosis	57–59
11	2NZT	Crystal structure of human hexokinase II	2.45	Hexokinase II supports the Warburg effect in HNC, promoting glycolysis and resistance to therapy	60–62

Molecular docking

PyRx⁶³ was used for virtual screening, performing ligand-energy minimization and molecular docking simulations. The tool was employed to calculate binding affinities between the ligands and the target proteins, using metformin analogs as ligands and proteins of biological importance in HNC as receptors, thus enabling the identification of the most stable conformations of metformin analogs against the tested proteins, subsequently leading to advanced analytical measures for the identification of suitable inhibitors to be used in the management of HNC. The library of metformin analogs prepared above was used for the molecular docking procedure with the 11 different target proteins previously discussed for their involvement in cancer progression, either directly or indirectly. Using PyRx, each protein was subjected to docking with each of the 59 ligands, thus producing a total of 531 results for each receptor-ligand interaction profile. Prior to the commencement of the procedure, the coordinates of the space in which the ligand produces interactions with the receptor were accurately determined, based on the dimensions of the receptor protein. The coordinates used for each of the 11 proteins in this study have been listed in Table 2.

Table 2.

Center and size coordinates in three planes (x, y, z) set for each target receptor protein used for the molecular docking simulations.

S. No	Receptor (PDB ID)	Centre			Size
		x	y	z	x	y	z
1	4Z18	− 26.0211	− 4.3491	16.2986	47.0701	66.4514	66.0448
2	4JPS	− 5.3162	− 24.3168	26.9928	101.3415	77.4082	93.9454
3	3MVH	11.6328	0.2640	19.6238	50.0408	49.3586	63.4931
4	4JSV	72.7363	− 7.5154	− 55.7216	94.8536	99.7681	114.8319
5	4JT6	71.6243	− 6.9877	− 55.4335	89.2884	92.3835	112.9438
6	4CFE	490.6153	18.8583	1006.1150	70.4451	64.8206	80.6693
7	6MXY	− 10.4531	30.8415	− 15.0927	47.3869	44.4450	30.8177
8	1Q5K	21.8904	37.7636	0.3248	61.6459	71.7427	56.3778
9	4DJS	2.7983	11.0798	− 38.9032	58.8998	50.3762	120.1218
10	2OCJ	3.3802	− 0.2244	31.1710	49.5080	40.8894	48.8358
11	2NZT	− 8.2683	− 20.4435	25.7746	77.6829	128.9501	63.4862

ADME studies and toxicity assessment

SwissADME (http://www.swissadme.ch/,⁶⁴), an online predictive tool developed by the Swiss Institute of Bioinformatics, was applied in this study for the purpose of prediction of pharmacokinetic properties and drug-likeness of the metformin analogs to be used as ligands in the evaluation of receptor-ligand interactions. Assessment of the Absorption, Distribution, Metabolism, and Excretion (ADME) properties was performed using physicochemical descriptors, Lipinski’s rule of five, and bioavailability scores⁶⁵. In the study, the results obtained from SwissADME were used to identify lead compounds with acceptable oral bioavailability, gastrointestinal absorption, and drug-likeness to prioritize compounds that have the potential as effective anti-cancer drug candidates. The metformin analogs were studied for their toxicity profile with the help of ProTox 3.0 (https://tox.charite.de/protox3/,⁶⁶), a toxicity prediction web-based platform. This tool integrates machine learning algorithms and chemical similarity measures to predict a number of toxicity end-points: LD₅₀ values, hepatotoxicity, carcinogenicity, and immunotoxicity. ProTox 3.0 was used in this study to identify analogs with acceptable toxicity levels, allowing the selection of safe candidates to be further explored in HNC therapy as inhibitory molecules for target proteins. Final selection of ligands for molecular dynamics and MM/PBSA analysis was based on the achievement of crucial parametric values from ADME analysis and toxicity profiles. Using SwissADME, the molecular weight, number of hydrogen bond donors and acceptors, molar refractivity, lipophilicity, and drug-likeness was assessed and recorded. Similarly, ProTox 3.0 was used to predict the toxicity class (1–6) of the chemical compounds. Analysis on both tools was done using Simplified Molecular Input Line Entry System (SMILES) of each ligand obtained from PubChem. The SMILES were entered as input in both tools and the query was executed, returning the required parameters as the results.

Molecular dynamics simulations

GROMACS 2024.4⁶⁷ was utilized for the molecular dynamics simulation. Energy minimization was performed using GROMACS 2024.4 with 50,000 steps (nsteps = 50,000), employing the steepest descent algorithm (steep) within a dodecahedral simulation box (distance = 2.0 nm) with extended simple point charge (SPC/E) water molecules⁶⁸. The Amber ff99SB-ILDN force field⁶⁹ was used to model the protein complex, while the target molecule topologies were generated using the Antechamber module within the AMBER22 suite^70,71. Energy minimization of the system was carried out using the steepest descent algorithm for 50,000 steps to relieve any steric clashes. This was followed by two equilibration steps: a 1,000 ps simulation under the NVT ensemble (constant number of particles, volume, and temperature) to stabilize the temperature at 300 K, and a 1,000 ps NPT simulation (constant number of particles, pressure, and temperature) to equilibrate the pressure at 1 atm. Subsequently, a production MD simulation was run for 100 ns to assess the dynamic behaviour and structural stability of the complex over time.

Binding free energy calculations

To further assess the binding stability of the complex, binding free energy was computed using the GMXPBSA module with MMPBSA.py⁷². Key output files from the simulation such as the .tpr file, index file, trajectory, and topology were used as inputs to generate the corresponding AMBER-compatible topology files.

Results

Molecular docking

The set of 59 metformin analogs, subjected to energy minimization using PyRx and converted to the PDBQT format, were then used for the docking procedure with one protein at a time. After the completion of each docking, the results (531) were filtered on the basis of RMSD values. With only an RMSD value of 0 considered for selection of a particular receptor-ligand interaction, the screening process resulted in 59 interactions, corresponding to the number of ligands for each protein. The docking results have been discussed in Supplementary Tables S1(a) and S1(b). Among the 11 different biological proteins involved in this study, only 4 proteins presented satisfactory values of the binding affinities recorded through docking. The lowest binding affinity recorded was − 6.7 kcal/mol between PDB 4JSV and PubChem CID 222300.

Virtual screening of potential lead molecule(s)

With the results of all the molecular docking simulations compiled, the potential lead molecules among the entire set of analogs were selected based on the binding affinities presented by the receptor-ligand interactions. The criteria maintained for selection of any ligand for further evaluation was a binding affinity of − 6.0 kcal/mol or lower. All ligands not meeting this criteria were rejected by reason of not appropriately binding to the target proteins. Taking into consideration this criteria of screening, ligands were accepted for further virtual simulations. These ligands, along with the proteins with which they produce the acceptable binding affinities, have been listed in Table 3. Following the virtual screening of ligands, the selected molecules were further tested for biological suitability in terms of solubility and permeability. Further, the toxicity class of the compounds was also determined to ensure drugs involving the compounds mentioned in Table 3 are safe for human consumption. The values recorded from the ADME analysis and the toxicity evaluation have been listed in Tables 4 and 5.

Table 3.

Target receptor(s) interacting with each of the six ligands selected on the basis of acceptable binding affinities.

S. No	Ligand (PubChem CID)	Interacting receptor(s)
1	222300	4JPS (− 6.4), 4JSV (− 6.7)
2	12717755	4JPS (− 6.2), 3MVH (− 6.2), 2NZT (− 6.1)
3	55281088	4JSV (− 6.3)
4	86292531	4JPS (− 6.0)
5	138986770	4JPS (− 6.5), 4JSV (− 6.4)
6	171758279	4JPS (− 6.2), 3MVH (− 6.3), 2NZT (− 6.2)

Table 4.

ADME analysis and toxicity evaluation of the six selected ligands.

S. No	Ligand (PubChem CID)	Molecular weight (g/ mol)	Number of H-bond donors	Number of H-bond acceptors	Consensus Log Po/w	Molar refractivity	Drug-likeness	Toxicity class
1	222300	140.15	3	3	− 1.19	36.62	Yes	5
2	12717755	154.17	2	3	− 0.82	41.52	Yes	4
3	55281088	127.15	2	2	− 1.07	45.48	Yes	4
4	86292531	143.19	2	2	− 0.56	41.83	Yes	4
5	138986770	140.15	3	3	− 1.22	36.62	Yes	5
6	171758279	171.20	4	3	− 1.50	48.13	Yes	4

Table 5.

Toxicity assessment of screened ligands (+ denotes active, −denotes inactive).

Ligand (PubChem CID)	Hepatotoxicity	Neurotoxicity	Cardiotoxicity	Carcinogenicity	Immunotoxicity	Mutagenicity	Cytotoxicity
222300	+	+	−	+	−	−	−
12717755	+	+	−	+	−	−	−
55281088	−	+	−	−	−	−	−
86292531	−	−	−	−	−	−	−
138986770	+	+	−	+	−	−	−
171758279	−	−	−	−	−	−	−

Based on the docking results, the receptor 4JSV produced the lowest binding affinity (−6.7 kcal/mol) with the ligand 222300. Thus, all receptor-ligand interactions involving 4JSV as the receptor were visualized using 2D interactions as well as surface images. The interactions of the three ligands interacting with 4JSV, namely 222300, 55281088, and 138986770, have been illustrated in Fig. 2A–C. This study has employed a blind docking methodology during the docking procedure. When we compared our interacting residues with other studies, it was seen in one of the studies which aimed to determine the efficacy of benzoxazole derivatives as potential inhibitors of mTOR in breast cancer that the key residues in the active site prominently included Lys2187, Asp2357, Val2240, and Trp2239. Other residues included Gln2167, Glu2190, Thr2245, Arg2251, and Hid2247⁷³. It is worth noting that only one out of the four benzoxazole derivatives presented a binding affinity better than the PDB 4JSV-CID 222300 pair (− 6.7 kcal/mol). Another study involving Rhizoma polygonati extract with kinase inhibitors to determine blockade of kinase pathways used PDB 4JSV as the reference mTOR structure. Molecular docking with the compound having the best docking score, diosgenin, revealed interactions with many active site residues, including Gly1897, Asn1898, Leu1900, Glu1937, Pro1940, Glu2196, Gln2200, Thr2207, Arg2224, and Leu2207⁷⁴. Although interacting active site residues differ, the pair PDB 4JSV-CID 222300 was chosen because it presented the lowest binding affinity among all receptor-ligand pairs at an RMSD value of 0. Variation among ligands, evident upon comparison of published literature and the current study, naturally accounts for differences among interacting residues despite the same target. Thus, the PDB 4JSV-CID 222300 complex was selected for further analysis using MD simulations and MM/PBSA studies because it presented the lowest binding affinity (− 6.7 kcal/mol) while also producing appropriate functional interactions with mTOR as evident from surface and 2D interaction diagrams. While other ligands also produced affinity values close to the selected ligand, prioritizing the best-scoring complex, a recurring strategy in structure-based screening studies, facilitates focused and rigorous simulations while also bolstering the possibility of attaining the most thermodynamically favourable binding scenario.

Figure 2.

(A) (a) Surface diagram showing hydrogen donors and acceptors involved in the interactions of ligand 222300 with 4JSV, and (b) 2D interaction diagram showing the amino acid residues of 4JSV interacting with 222300. (B) (a) Surface diagram showing hydrogen donors and acceptors involved in the interactions of ligand 55281088 with 4JSV, and (b) 2D interaction diagram showing the amino acid residues of 4JSV interacting with 55281088. (C) (a) Surface diagram showing hydrogen donors and acceptors involved in the interactions of ligand 138986770 with 4JSV, and (b) 2D interaction diagram showing the amino acid residues of 4JSV interacting with 138986770.

Molecular Dynamics Simulation Analysis

The MD simulations revealed positive associations between the screened ligand (222300) and the target mTOR protein (PDB: 4JSV). The Root Mean Square Deviation (RMSD) plot (Fig. 3) reveals that the system undergoes a sharp increase in RMSD within the first 20 ns, suggesting notable conformational changes or structural instability during the initial phase of the simulation. This early rise likely reflects the system adapting and deviating from its initial conformation during equilibration. From approximately 22 ns onward, the complex exhibits more stable behaviour, with RMSD values fluctuating around 6 nm, though occasional spikes reaching up to ~ 8.5 nm are observed. These spikes may correspond to transient structural events such as local unfolding, domain rearrangements, or flexibility within loop regions. The behaviour may be caused by ligand-induced conformational changes or a spontaneous reorganization of the protein to better hold the ligand. Thus, the receptor-ligand system does not present a complete loss of stability, indicative of an overall structural integrity throughout the duration of the simulation, albeit with some flexibility. Systems experiencing rapid relaxation from initial strain as a result of steric or electrostatic clashes during the early simulation stage are characterized by this RMSD trajectory. Given the hydrophobic and electrostatic effects of PubChem CID 222300, the early deviations indicate an adjustment period during which the tertiary and quaternary structures of the receptor most likely realign to accommodate the presence of ligands. The successful establishment of a reasonably stable binding pose is highlighted by the stabilization post-22 ns, and the moderate fluctuations suggest a dynamic equilibrium as opposed to rigid docking. The observed transient spikes might be an indication of flexibility that is functionally relevant, especially in loop regions that are known to modify allosteric transitions or ligand accessibility. This behaviour is noteworthy because it is consistent with the known domain-level plasticity of mTOR, where conformational adaptability essential to its kinase regulation is made possible by partial flexibility.

Figure 3.

Result of RMSD for protein ID: 4JSV and ligand CID: 222300.

The RMSD profile indicates an initial equilibration phase during the first 20 ns, where the PDB ID 4JSV and ligand CID 222300 complex adjusts from the docked structure to a stable conformation in the solvated environment. A transient increase in RMSD around 20–23 ns likely reflects a conformational rearrangement of flexible regions within the protein. After this transition, the RMSD stabilizes around ~ 5.8–6.2 nm, with moderate fluctuations observed between 60 and 100 ns, which are typical of natural protein dynamics. Importantly, the RMSD does not show continuous drift and remains within a stable range during the majority of the simulation, suggesting that the PDB ID 4JSV and ligand CID 222300 complex maintains structural stability throughout the simulation period.

Further, the Root Mean Square Fluctuation (RMSF) plot (Fig. 4) shows that the majority of residues display fluctuations within the range of 0.2 to 0.5 nm, indicating overall structural stability. However, a few pronounced peaks are present, including an extreme outlier near atom index 18,000 with a fluctuation of approximately 2.8 nm. These elevated RMSF values typically indicate flexible or disordered regions, such as terminal residues or solvent-exposed loops, which are common in large protein–ligand complexes. Localized flexibility is further supported by the RMSF data, which enhances the global trends found in the RMSD analysis. High fluctuations at certain indices, especially around atom index 18,000, are probably due to the high mobility inherent in surface-exposed loops, linker segments, or terminal domains. These pliable residues could serve as functional hotspots for conformational gating or ligand recognition. It is well known that flexible domains in the context of mTOR experience structural changes when ATP or ligands bind to them. The ligand may be dynamically stabilized in the active site by these regions, which may promote transient hydrogen bonding or π–π stacking interactions. Furthermore, the flexibility of these zones may indicate possible allosteric control points or areas of induced fit, which are essential for effective molecular recognition and catalysis. These moderate fluctuations are thus suggestive of conformational adjustments occurring in the protein structure upon binding of the ligand exhibiting consistency with the RMSD data previously discussed.

Figure 4.

Result of RMSF for protein ID: 4JSV and ligand CID: 222300.

Furthermore, the Radius of Gyration (Rg) plot (Fig. 5) demonstrates a slight but consistent decrease in Rg values from ~ 3.55 to ~ 3.47 nm over the course of the simulation, along with minor fluctuations. This downward trend suggests a gradual compaction of the protein–ligand complex, supporting the notion of increasing structural stability over time. The structure of the protein gradually condenses, as indicated by the Rg profile, which is consistent with a thermodynamically advantageous binding event. The decrease in Rg suggests that ligand engagement causes the protein to undergo entropic ordering, which could entail stabilizing mobile loops or rearranging secondary structural motifs. Van der Waals, electrostatic, and hydrophobic contacts that form between the ligand and important receptor residues may be the cause of this compaction, which is suggestive of an energetically optimized conformation. Additionally, no significant unfolding or destabilization takes place, as indicated by the plateau in Rg after 40 ns, which supports the integrity of the folded protein domain architecture throughout the simulated timeframe. This implies that CID 222300 may be appropriate as a stabilizing ligand for modifying mTOR structural states. The relative fluctuational stability post 40,000 ps is thus indicative of a sustained compact protein–ligand structure associated with thermodynamic favourability and an overall well-folded protein structure.

Figure 5.

Radius of Gyration analysis for protein ID: 4JSV and ligand CID: 222300.

MM/PBSA analysis

The MM/PBSA studies provided comprehensive information about the receptor-ligand interactions involving hydrogen bonds. Also, the binding energy calculations provided satisfactory results with respect to the overall interaction profile of the protein and the ligand. The hydrogen bond analysis (Figs. 6 and 7) presents low occupancy values of 15% between Asp1712 and the ligand, indicating that these interactions are generally transient or weak. However, a few interactions such as those involving Gln1715 (0.78%), Asp1506 (0.76%), and Met1535 (0.52%) occur more frequently, suggesting that these residues may play a stabilizing role in maintaining the ligand–protein complex through persistent hydrogen bonding. The ligand thus maintains effective polar interactions with the target mTOR protein, with the observed results consistent with those observed and inferred from the results of RMSD, RMSF, and radius of gyration. Figure 8 further elaborates the variety of prominent hydrogen bonding interactions recorded from the simulation. Among the various residues mentioned in the graph corresponding to 62 hydrogen bonds in the complete interaction profile, Asp1712 presents the highest percent occupancy among all residues, with a value of 16.37%. Key residues like Asp1712 regularly interact with the ligand, which possibly helps to anchor the molecule within the active site, according to the observed hydrogen bond dynamics, even though the majority of polar contacts are temporary. Although semi-stable interactions are not sustained throughout the trajectory, they recur frequently enough to affect binding energetics, as evidenced by the modest occupancy exhibited by Asp1712. Crucially, interactions with residues such as Asp1506 and Gln1715 point to a dispersed hydrogen-bonding network that could limit rotational degrees of freedom and strengthen ligand orientation. Such interactions might be essential for ligand specificity and retention given the polar binding environment of mTOR. A correlated network of enthalpic and entropic stabilization mechanisms is indicated by these interactions, which also reflect the structural compaction and conformational stability deduced from the Rg and RMSD profiles.

Figure 6.

H-Bonds between protein ID: 4JSV and ligand CID: 222300.

Figure 7.

H-Bond involving residues Asp1539 and Ligand (UNL) for protein ID: 4JSV and ligand CID: 222300.

Figure 8.

Histogram representation of percentage occupancy of the hydrogen bond contacts between protein ID: 4JSV and ligand CID: 222300.

In terms of binding free energy analysis (Fig. 9), the gas-phase interaction energy (ΔGGAS) is strongly negative (− 115.44 kcal/mol), indicating that the ligand and receptor interact very favourably in the absence of solvent. Conversely, the solvation free energy (ΔGSOLV) is positive (~ + 105.52 kcal/mol), reflecting the energetic cost associated with de-solvating both the ligand and receptor upon binding. Despite this penalty, the total binding free energy (ΔTOTAL) remains moderately favourable at − 9.92 kcal/mol, supporting the conclusion that the ligand forms a stable and energetically viable complex with the protein. The thermodynamic landscape controlling complex formation is revealed by the MM/PBSA binding energy results. A high intrinsic affinity between the ligand and the mTOR receptor in a vacuum is highlighted by the noticeably negative ΔGGAS, which reflects strong Van der Waals and electrostatic contributions. Nevertheless, polar systems frequently exhibit the de-solvation penalty (ΔGSOLV), particularly when hydrogen bonds and hydrophilic surface residues interact with the bulk solvent. The net negative ΔTOTAL indicates that the solvent reorganization and entropic costs are outweighed by the advantageous enthalpic contributions from molecular contacts. According to this balance, CID 222300 may be able to occupy the mTOR binding pocket steadily in physiological settings. In lead compounds, a moderate binding energy is frequently preferred because it allows for the achievement of adequate affinity without irreversible binding, maintaining drug-likeness and reducing off-target interactions.

Figure 9.

Binding Free Energy for protein ID: 4JSV and ligand CID: 222300.

Discussion

With the aggravation of the issue of cancer resistance to conventional chemotherapeutic treatments in global healthcare settings, the need to develop, test, and deploy novel methods for advanced curative effects and minimal side effects has been gaining traction. Among the diverse cancerous tumors prevalent in humans, HNC presents a worrisome challenge to the overall well-being and quality of life of global populations, with several statistics discussed previously. Targeting biological pathways involved in the sustenance of proliferative cellular conditions, pathogenesis of the cancerous tumors, and the development of therapeutic resistance can provide access to the synthesis of highly precise, effective medication procedures. This study is concerned with determining the effectiveness of the chemical analogs of metformin, an FDA-approved first-line anti-diabetic drug, against different biological proteins representing essential cellular pathways like PAM, Wnt/β-catenin, AMPK, and glycolysis. With an integrated in-silico approach involving molecular docking, ADMET analysis, MD simulations, and MM/PBSA studies, the study produced considerably positive results with respect to RMSD, RMSF, radius of gyration, hydrogen bond interactions, and calculations of binding energy for the mTOR protein PDB 4JSV and the metformin analog (6e)-6-Imino-1-methyl-1,6-dihydro-1,3,5-triazine-2,4-diamine (PubChem CID 222300). The receptor-ligand pair produced the lowest binding affinity of − 6.7 kcal/mol. Although the molecular docking result with the metformin analog does not present highly significant binding affinity, visualization of surface and 2D interaction diagrams reveal multiple conventional hydrogen bonds, Van der Waal forces, π-π T-shaped bonds, and π-σ bonds which strengthen the receptor-ligand interactions and reinforce the occurrence of a stable complex formation without inducing unusual structural disruptions. Although direct molecular docking and molecular simulation studies involving metformin and the mTOR protein PDB 4JSV for a potential treatment of HNC are very scarce, a few studies present results pertaining to other objectives. A study investigated the therapeutic effects of metformin nuclei-based drugs modified with Tulbaghia violacea extract compounds and metal ions. The study illustrated the potential of these metformin derivatives against cancer cells and anti-viral efficacy against COVID-19. PDB 4JSV, representing the mTOR protein considered responsible for virus proliferation, was used in the docking study with standard metformin and the derivatives. While metformin demonstrated a very limited binding affinity with 4JSV (− 1.63 kcal/mol), the metformin derivatives produced significant results, with a binding affinity of − 9.56 kcal/mol reported with Cu-NBM (condensation of metformin with [4-(Diethylamino) benzaldehyde (NBM) and copper ions), thus indicating effective mTOR inhibition⁷⁵. This supports the possibility of enhanced binding affinity of metformin with the target receptor upon modification of the ligand. Another study involving a combinational treatment of Dapagliflozin and Etoricoxib for cervical cancer inhibition investigated interactions of the drugs with each of the three components of the PAM pathway, wherein mTOR (PDB ID: 4JSV) was one of the targets of the study. While dapagliflozin yielded an affinity of − 7.3 kcal/mol, etoricoxib presented an affinity of − 6.8 kcal/mol⁷⁶. The binding affinity with dapagliflozin is close to the value determined in this study (− 6.7 kcal/mol), while etoricoxib yielded a value almost identical to the results of this study.

Interpretation of RMSD, RMSF, and Rg values coherently indicate an overall stable receptor-ligand complex with greater compaction as compared to the native protein and enhanced thermodynamic favourability. Apart from inducing some minor structural flexibilities, the ligand does not stimulate any major conformational variations in the protein structure that might translate to instability and degradation. Instead, it greatly stabilizes the structure through persistent hydrogen bonding interactions and the production of an appropriately negative binding free energy. The MD simulations results and MM/PBSA analysis thus present strong correlation and parallelism. This metformin analog can thus be safely concluded as an effective ligand for mTOR targeting and subsequent inhibition of downstream biological processes involving this crucial protein.

The present study focused specifically on metformin analogs, and, therefore, the compound library was intentionally limited to 59 structurally related molecules selected based on their reported structural similarity and potential pharmacological relevance. The aim was to systematically evaluate the interaction patterns of metformin-derived compounds with the selected targets rather than to perform a large-scale virtual screening. The ligand binding site for mTOR was defined based on the ATP-binding catalytic pocket reported in the crystal structure of the mTOR—mLST8 complex⁴³. The docking grid was centred on the ATP-binding site residues of the kinase domain, which are known to accommodate ATP-competitive inhibitors. The MD simulations were analysed using RMSD, RMSF, radius of gyration, and hydrogen bond profiles. The RMSD trajectory shows an initial increase during the early phase of the simulation as the system adjusts from the docked conformation to a solvated equilibrium state, followed by stabilization after approximately 20–25 ns, indicating attainment of conformational equilibrium. The RMSF analysis reveals that most residues exhibit low fluctuations (< 0.4 nm), while slightly higher fluctuations are limited to flexible loop or terminal regions, suggesting that the core protein structure remains stable during the simulation. Furthermore, the radius of gyration remained within a narrow range (~ 3.47–3.60 nm) throughout the trajectory, indicating that the protein maintained its overall compactness without significant structural unfolding. Hydrogen bond analysis shows the formation of 0–2 hydrogen bonds between the ligand and protein during the simulation, demonstrating persistent intermolecular interactions that contribute to binding stability. Collectively, these analyses suggest that despite the observed RMSD fluctuations during the equilibration phase, the protein–ligand complex maintains structural stability and stable binding interactions throughout the simulation period.

Metformin, used in an in-silico approach to supplement the anti-cancer efficacy being investigated in the same study using cancer cell lines, presented highly significant interaction with FKBP12-mTOR (PDB: 1FAB), with the receptor-ligand pair recording the highest binding free energy (− 15.63 kcal/mol)⁷⁷. Studies involving the use of metformin or its analogs as ligands for targeted inhibition of biological proteins, particularly mTOR, through in-silico methods are rare. Though in vitro studies investigating the anti-cancer potential of metformin are numerous, in-silico approaches involving this promising drug have fallen behind in the course of scientific conduct. However, the target protein used in the study, PDB 4JSV has been used as an mTOR receptor in studies focused on different other types of cancer. Compounds from a novel synthesized series of mono- and bis(dimethylpyrazolyl)-s-triazine derivatives were evaluated for their anti-proliferative activity through the use of cancer cell lines as well molecular docking with target receptors. PDB 4JSV was used as the mTOR representative protein. All the compounds synthesized and analysed during the study produced considerable binding affinities with the protein, ranging from − 7.7 to − 8.6 kcal/mol, thus exhibiting appreciable interactions with the components of the EGFR/PI3K/AKT/mTOR signaling cascade⁷⁸. A marine-derived compound, 1,4:3,6:5,7-Tribenzal-beta-mannoheptitol was studied for its potential use in lung, breast, and colorectal cancer therapy through molecular docking analysis and was found to be effective against the protein mTOR, represented by the PDB 4JSV in the study⁷⁹. Ten new quinoline and eighteen benzofuran derivatives were screened for their possible inhibitory activity against colon cancer and triple-negative breast cancer (TNBC). For molecular docking simulations aimed at determining possible targets, PDB 4JSV was involved as the mTOR receptor chosen against the ligands being evaluated and appropriate binding affinities were recorded following the conduct of docking⁸⁰. Future work will involve expanding the chemical space and performing large-scale virtual screening to identify more potent inhibitors.

Conclusion

This computational study evaluated 59 different metformin analogs as potential inhibitors of different biological pathways activated in HNC. The evaluated pathways and receptors, namely PD-L1, PI3K/AKT/mTOR, AMPK, Wnt/β-catenin, p53, and glycolysis, consisted of a set of 11 different proteins. From the entire docking results of different receptor-ligand interactions corresponding to each protein docked with the 59 ligands, PDB 4JSV produced the lowest binding affinity (− 6.7 kcal/mol) with PubChem CID 222300. The compound, present on PubChem by the name (6e)-6-Imino-1-methyl-1,6-dihydro-1,3,5-triazine-2,4-diamine, proved sufficiently viable for future research endeavours not only due to acceptable binding affinity with the protein and ADMET analysis, but also due to the favourable MD simulation results at 100 ns using the pair PDB 4JSV-CID 222300, characterizing the RMSF, RMSD, radius of gyration, hydrogen bond analysis, and binding free energy calculations for the receptor-ligand pair. While a highly significant binding affinity value could not be obtained, literature analysis reveals comparable results when different compounds are used in molecular docking studies with the same receptor. Also, the presence of multiple molecular interactions discussed earlier effectively anchor the ligand to the receptor. Comprehensive evaluation of the obtained results after MD simulations provided the conclusion that, despite a moderate binding affinity value, CID 222300 stably and appropriately binds to PDB 4JSV without inducing any structural deformities or destabilizations in the protein structure. Although major protein structural and conformational changes occur, the overall structure does not progress to an unstable or inactive state. Instead, the protein, towards the end of the simulations, takes up an evolved, compact, thermodynamically stable conformation, with the ligand contributing to the stabilization of the molecular structure and formation of an energetically viable complex. The net negative ∆TOTAL (− 9.92 kcal/mol) provides affirmation to the aforementioned point of an energetically favourable complex. Structural changes, rearrangements, and intermittent outliers do not hold sufficient ground for the dismissal of the interactions as inadequate. Ligand engagement, influencing final binding energetics, was further proven by the transient yet significant polar interactions between the amino acid residues and the ligand, particularly Asp1712 (16.37%). All of these findings provide compelling evidence that CID 222300 does not induce any deleterious structural disruptions in the protein and contributes to a progressively stabilized structural conformation of the mTOR protein, interacting with the binding pocket of mTOR in a stable and efficient manner. The ligand actively stabilizes the receptor-ligand interface and holds great potential as an inhibitor that can alter the mTOR-driven signalling pathway, which is a key axis for cellular survival, proliferation, and metabolic reprogramming in HNC. The conduct of MD simulations at 100 ns, along with analysis using MM/PBSA, provided highly detailed insights into direct metformin-mTOR interactions not yet comprehensively evaluated in literature with respect to the use of metformin in HNC. Based on the obtained results, we propose the conduct of in vitro experimental research to confirm the appropriateness of physicochemical profiles associated with the ligand. Such confirmatory studies, combined with the already established results in this paper, can cement the position of this metformin analog as a potential therapeutic intervention which can be applied to treat HNC patients in global healthcare settings.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ADME

Absorption, Distribution, Metabolism, and Excretion

AMPK

Adenosine monophosphate-activated protein kinase

ASIR

Age-standardised incidence rate

ATP

Adenosine triphosphate

BMI

Body mass index

EGFR

Epidermal growth factor receptor

HGF/c-MET

Hepatocyte growth factor/mesenchymal–epithelial transition factor

HNC

Head and Neck Cancer

HNSCC

Head and neck squamous cell carcinoma

HPV

Human papillomavirus

IARC

International Agency for Research on Cancer

JAK/STAT

Janus kinase/signal transducer and activator of transcription

LAHNSCC

Locally advanced HNSCC

MD

Molecular dynamics

MM/PBSA

Molecular Mechanics/Generalized Born Surface Area

mTOR

Mammalian target of rapamycin

NF- κB

Nuclear factor kappa B

OCC

Oral cavity cancer

OS

Overall survival

PBCR

Population-based cancer registries

PFS

Progression-free survival

PI3K

Phosphoinositide 3-kinase

RCSB PDB

Research Collaboratory for Structural Bioinformatics Protein Data Bank

Rg

Radius of gyration

RMSD

Root mean square deviation

RMSF

Root mean square fluctuation

SDF

Structured data file

SDGs

Sustainable development goals

T2DM

Type 2 diabetes mellitus

Author contributions

P.A. and P.S. conceptualized and designed the study. S.P.K.,P.M.S, B.Y, M.T performed data acquisition, molecular docking, molecular dynamics simulations, and analysis. N.R and P.S. supervised the research, contributed to methodology development, and provided critical revisions. S.P.K, V.J.U, P.M.S did the final review of the manuscript. All authors contributed to the interpretation of results, manuscript drafting, and approved the final version for publication.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Data availability

All the data analyzed/ generated in this work is included within the manuscript and the supplementary files, Tables S1(a) and S1(b) provided along with the manuscript.

Competing interests

The authors declare that they have no conflict of interest.