In-Silico Drug Discovery of Flavonoids Targeting the Human Serotonin Transporter: A Multi-Approach Computational Study for Depression

Depression is a chronic and debilitating neuropsychiatric disorder that affects over 280 million people worldwide, making it a leading cause of disability and a major contributor to the global disease burden[1]. Characterized by persistent sadness, loss of interest, cognitive impairment, and emotional dysregulation, depression significantly impacts an individual’s social, occupational, and personal well-being[2]. The lifetime prevalence of major depressive disorder (MDD) is estimated between 10% and 20%, with a higher incidence in women. In India, one in every twenty individuals suffers from depression, a condition exacerbated by socioeconomic stressors, lifestyle changes, and increasing comorbidities like diabetes and cardiovascular diseases. Despite the availability of antidepressant treatments, 30%–50% of patients fail to achieve full remission, indicating a dire need for novel, more effective, and safer therapeutic alternatives[3, 4].

Central to the pathophysiology of depression is serotonin (5-hydroxytryptamine, 5-HT) deficiency in the brain. The human serotonin transporter (hSERT) plays a pivotal role in regulating serotonergic neurotransmission by facilitating serotonin reuptake from the synaptic cleft into presynaptic neurons, thus terminating its signaling[5]. As a critical therapeutic target, hSERT is the primary site of action for selective serotonin reuptake inhibitors (SSRIs), which function by blocking serotonin reuptake and enhancing serotonergic activity. However, SSRIs and serotonin-norepinephrine reuptake inhibitors (SNRIs) like fluoxetine, sertraline, venlafaxine, and duloxetine suffer from several limitations, including delayed onset of action, adverse side effects (such as sexual dysfunction, weight gain, sleep disturbances), treatment resistance, increased suicidal risk in younger populations, and withdrawal symptoms[6]. These drawbacks emphasize the need for novel antidepressant strategies that offer enhanced efficacy, faster onset, and reduced side effects.

Flavonoids, a diverse class of plant-derived polyphenolic compounds, have garnered significant attention for their potential antidepressant properties. Found in fruits, vegetables, and medicinal plants, flavonoids have demonstrated the ability to modulate monoaminergic neurotransmission, reduce oxidative stress, and promote neuronal survival. Certain flavonoids, such as quercetin, kaempferol, and baicalein, have shown promise in inhibiting hSERT, suggesting their potential as natural antidepressants[7]. These compounds offer several advantages over synthetic antidepressants, including their natural origin, multi-target mechanisms, lower toxicity, and fewer side effects. Additionally, flavonoids have demonstrated fast-acting antidepressant-like effects in preclinical models[8]. Despite these promising findings, the molecular interactions of flavonoids with hSERT remain largely unexplored, which underscores the need for further investigation using advanced computational drug discovery techniques.

1.1.In-Silico Approach for Flavonoid-Based hSERT Inhibition

In this study, we employ a multi-approach in-silico strategy to evaluate the potential of flavonoids as hSERT inhibitors, leveraging the power of computational drug discovery. Using the crystal structure of hSERT (PDB ID: 5I6X), our workflow integrates molecular docking, pharmacokinetic profiling, binding free energy calculations, and molecular dynamics simulations to identify promising flavonoid candidates.Molecular Docking – This technique is used to assess the binding affinities and interaction patterns of flavonoids within the hSERT active site[9]. The docking process helps predict how these compounds interact with key amino acid residues responsible for serotonin reuptake inhibition.ADMET Screening – The absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties of flavonoids are analyzed to evaluate their pharmacokinetic profile and drug-likeness, ensuring that selected compounds have favourable bioavailability and minimal toxicity.MM/GBSA Calculations – Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) calculations are performed to estimate the binding free energy of flavonoid-hSERT complexes, providing a more accurate measure of binding stability and affinity.Molecular Dynamics (MD) Simulations – To analyze the stability and dynamic behaviour of flavonoid-hSERT interactions over time, MD simulations are conducted under physiological conditions. This step ensures that the identified flavonoids form stable and sustained interactions with hSERT, which is crucial for their potential antidepressant effects[10].

All computational analyses are performed using Schrödinger 2023-1, a state-of-the-art software suite for structure-based drug discovery. By integrating these in-silico methodologies, we aim to identify novel flavonoid-based hSERT inhibitors with high binding affinity, favourable pharmacokinetics, and stable interactions. Our findings could pave the way for the development of safer, naturally derived antidepressant agents that overcome the limitations of current treatments. Ultimately, this study provides valuable insights into flavonoid-hSERT interactions, contributing to the rational design of antidepressant therapies that offer improved efficacy, reduced side effects, and enhanced patient outcomes.

MATERIALS AND METHODS:

2.1. Protein preparation:

The (hSERT) protein (PDB ID: 5I6X) was retrieved from the RCSB Protein Data Bank and prepared for molecular modelling using the Protein Preparation Wizard in Schrödinger Suite 2023-1. The preprocessing step involved removing crystallographic water molecules and adding missing side chains and hydrogen atoms to ensure structural integrity. Bond orders were assigned appropriately, and the most probable protonation states for ionizable residues were determined under physiological conditions. The structure was then subjected to energy minimization using the OPLS4 force field, where side-chain hydroxyl groups were optimized, and heavy atoms were converged to an RMSD of 0.30 Å. Restraints were applied to maintain the backbone conformation while allowing local flexibility for improved stability. The active site was identified, and a grid box was generated to define the binding pocket for subsequent docking studies[11, 12].

2.2. Ligand preparation

The phytochemical structures were processed using the LigPrep module of Schrödinger Suite 2023-1 to ensure high-quality molecular structures suitable for docking studies. The Epik module was employed to generate ionization states, tautomeric forms, and desalted structures at a physiological pH of 7.0 ± 2.0, ensuring that each compound was represented in its most relevant protonation state. This step is critical for accurate binding affinity predictions, as improper charge states can significantly affect ligand-receptor interactions. During LigPrep processing, the structures underwent stereochemical refinement, where specified chiralities were retained, and unnecessary counterions or solvent molecules were removed. Additionally, alternative stereoisomers were generated where applicable to account for potential biological relevance. The geometry optimization was performed using the OPLS4 force field within the Impact package of Schrödinger, allowing for energy minimization while preserving the correct molecular conformation. This step ensures that ligands adopt a stable and low-energy state, crucial for reliable docking and binding studies.

The energy minimization process was executed until a root mean square deviation (RMSD) of 1.8 Å was achieved, refining atomic positions to eliminate steric clashes and ensure structural stability. For molecules containing flexible rings, a single low-energy ring conformation was generated to maintain biological relevance while reducing excessive computational complexity during docking. The final optimized phytochemicals were thoroughly analyzed to confirm correct stereochemistry, protonation states, and overall molecular integrity before proceeding to docking analysis. These well-prepared ligand structures enhance the accuracy of molecular docking by ensuring that only biologically relevant conformations are used, ultimately improving predictions of ligand-receptor interactions and potential drug-like behaviour[13, 14].

2.3. Receptor grid generation

The Receptor Grid Generation Wizard in Schrödinger was used to define the docking region within the prepared protein structure. A grid box was created at the centroid of the active site with coordinates x = -32.92, y = −21.86, z = 1.68, ensuring precise ligand docking. The Van der Waals (VdW) scaling factor was set to 1.0 for the receptor, while a partial charge cutoff of 0.25 was applied. To accommodate ligand flexibility, the grid dimensions were expanded appropriately to cover the entire binding site without introducing unnecessary computational overhead. Additionally, constraints such as hydrogen bonding and metal coordination were applied as needed, and excluded volumes were defined to restrict docking to relevant regions. Electrostatic and hydrophobic site maps were also incorporated to improve docking accuracy. The final receptor grid was validated to ensure it accurately represented the binding pocket, optimizing docking efficiency for subsequent virtual screening and drug discovery studies[15, 16].

2.4. Molecular docking studies using Glide:

The phytochemicals obtained from the LigPrep module were docked using the Glide module of the Schrödinger suite 2023-1 in Extra Precision (XP) mode to achieve highly accurate binding affinity predictions. The best Glide G-scores were selected based on their ability to balance various molecular interactions, punishing steric conflicts while promoting favourable lipophilic, hydrogen bonding, and metal-ligand interactions. The docking results were analyzed using the XP Visualizer, which provided detailed insights into the molecular interactions at the receptor site. The Glide scores of the phytochemicals were compared with those of commonly used antidepressant medications, such as Fluoxetine, to evaluate their potential efficacy. Several docking factors influenced the scoring, including lipophilic perseverance, where ligands are deeply enclosed within the hydrophobic pocket, enhancing their stability. Additionally, electrostatic interactions, hydrogen bonding, and π–π stacking interactions played a crucial role in strengthening ligand-receptor binding. The binding affinity could also be enhanced by metal coordination and water-mediated hydrogen bonding in some cases. However, unfavourable factors such as rotational penalties, desolvation energy penalties, and steric clashes could negatively impact the Glide score. The presence of excessive ligand flexibility also led to higher penalties, reducing docking accuracy[11, 17].

2.5. ADMET profile:

The ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) profile of the phytochemicals was analyzed using the QikProp module in the Schrödinger suite 2023-1. This computational approach provided insights into key physicochemical and pharmacokinetic properties, which are essential for evaluating the drug-likeness and oral bioavailability of the compounds. Various descriptors were predicted, including molecular weight, hydrogen bond acceptor and donor count, partition coefficient (log P), total solvent-accessible surface area (SASA), and molar volume. The Lipinski’s Rule of Five was assessed to determine drug-like properties, and the number of violations was recorded to ensure compliance with oral drug delivery standards. Additionally, aqueous solubility (log S), van der Waals volume, and human oral absorption percentage were predicted to assess the compounds' solubility, permeability, and overall bioavailability[18, 19].

2.6. Molecular Dynamics studies:

Molecular Dynamics (MD) Simulation Protocol Using Desmond

The molecular dynamics (MD) simulation was conducted using Desmond in the Schrödinger suite 2023-1, following a structured protocol to ensure accuracy and reproducibility. The initial biomolecular structure protein was obtained in PDB format and pre-processed using Maestro 13.5 to add missing atoms or residues, assign protonation states at physiological pH (7.4), and optimize hydrogen bonding networks. For ligand-containing systems, LigPrep was used to generate parameters, validated against the OPLS4 force field. The system was solvated using the TIP3P -Transferable Intermolecular Potential and widely used three-site water model in molecular dynamics simulations, designed to accurately represent the physical and chemical properties of water model within an orthorhombic simulation box, maintaining a 10 Å buffer distance around the solute. To neutralize the system and achieve an ionic concentration of 0.15 M NaCl, Na? or Cl? ions were added using the System Builder tool in Maestro. The OPLS4 force field was applied, and non-standard residues or ligands were validated.

The Desmond Setup Wizard was used to configure the NPT ensemble, maintaining a temperature of 300 K and pressure of 1 atm for a 100 ns simulation. The system underwent three equilibration steps: energy minimization using the steepest descent algorithm to eliminate steric clashes, a short-restrained MD simulation (100 ns) to allow solvent relaxation, and a brief unrestrained MD simulation (1 ns) to equilibrate the full system. The 100 ns production run was executed on a high-performance computing (HPC) cluster, saving trajectory frames every 10 ps for further analysis[16].

Post-simulation analysis was carried out using tools from the Schrödinger suite, including root-mean-square deviation (RMSD) to assess protein backbone stability, root-mean-square fluctuation (RMSF) to evaluate residue flexibility, and hydrogen bond analysis to identify key interactions. Additionally, MM/GBSA binding free energy calculations were performed to estimate ligand binding affinities. The trajectory and interactions were visualized using Maestro, providing insights into conformational changes, ligand binding mechanisms, and solvent effects. This workflow ensures a robust and reproducible MD simulation process, enabling a detailed exploration of biomolecular systems at atomic resolution[20, 21].

2.7. Binding Free Energy Estimation Using Prime/MM-GBSA:

The binding free energy of the phytochemicals was estimated using the Prime MM-GBSA module in the Schrödinger suite 2023-1. The OPLS4 force field was applied to minimize the energy of the XP-docked ligand-receptor complex, ensuring an optimized conformation. The Generalized Born/Surface Area (GB/SA) model with VSGB 2.0 (Variable-dielectric Surface Generalized Born) solvent mode was used to estimate the solvation free energy of the molecule in the given solvent. The Prime MM-GBSA binding free energy (ΔGbind) was determined using the following equation:

ΔG_bind?=E _{complex (minimized)?} − (E _{receptor (minimized)}?+E _{ligand (minimized)?)}

This calculation provided a quantitative measure of ligand binding affinity, aiding in the identification of potential lead compounds for further experimental validation[11, 22].

RESULTS

3.1 Comprehensive in Silico ADMET Screening of Bioactive Compounds Using the QikProp Module:

The ADMET screening of the compounds was performed in silico using the QikProp module of the Schrödinger Suite 2023-1. The results of the ADMET properties for the compounds are summarized in Table 1. Based on the Insilco screening, most compounds exhibit properties within the recommended limits. All parameters were evaluated according to the recommended range limits specified in the QikProp Module User Manual 4.4. Additionally, a comprehensive list of all ADMET parameters is provided in Supplementary Data Table 1. The molecular weight of the compounds ranges from 254.242 to 610.568. The predicted octanol/water partition coefficient (QPlogPo/w) falls within -2.35 to 4.62, while the predicted IC50 value for HERG K+ channel blockade ranges from -3.428 to -5.99. The predicted brain/blood partition coefficient (QPlogBB) varies between -4.58 and 0.74. Additionally, the predicted qualitative human oral absorption is within the recommended limits. The number of violations of Lipinski’s Rule of Five for most compounds falls between 0 and 3. The compounds also exhibit excellent human oral absorption, with values ranging from 94% to 100%. Overall, the in silico ADMET screening indicates that most compounds meet the recommended criteria, with only a few parameters deviating for certain compounds.

Table 1: Detailed Prediction of ADMET Properties for Various Flavonoids Using the QikProp Module

3.2. Molecular Docking Analysis and Binding Interactions of Flavonoids with (hSERT)

In Schrödinger 2023-1 (Maestro 13.5), molecular docking was performed using Extra Precision (XP) mode, which provides a more accurate assessment of ligand-protein interactions by incorporating advanced scoring functions and penalties for non-ideal interactions. The docking results revealed a root mean square deviation (RMSD) of 0.2 for the molecules, indicating an excellent fit and high reliability of the predicted binding conformations.

Molecular docking analysis was conducted to evaluate the interaction and binding affinity of selected ligands against the human serotonin transporter (hSERT). Among the tested compounds, the results are summarized in Table 2, where all flavonoids were docked into the same binding pocket using Extra Precision (XP) mode. Globularicitrin exhibited the highest docking score (-15.357), indicating a strong binding interaction, whereas Fluoxetine, a standard drug, showed a lower docking score of (-10.306). These results suggest that Globularicitrin has a stronger binding potential at the active site. The enhanced binding affinity of Globularicitrin is primarily attributed to multiple hydrogen bonds, electrostatic interactions, and extensive hydrophobic contacts, contributing to its stability within the binding pocket. The 2D interaction analysis revealed that Globularicitrin forms strong hydrogen bonds with key residues ASP-98, GLN-332, and SER-336, along with π-π stacking interactions with PHE-335 and additional hydrophobic contacts with TYR-95, ALA-169, and LEU-434 are given in figure 1. In contrast, Fluoxetine primarily relies on hydrophobic interactions with TYR-95, LEU-99, and PHE-341 and forms only a single hydrogen bond with ASN-177, leading to a comparatively lower binding affinity and reduced stability within the active site are given in figure 2. The higher docking score of Globularicitrin suggests a more stable and deeply penetrating interaction with hSERT, whereas Fluoxetine’s weaker electrostatic interactions and limited hydrogen bonding contribute to its lower stability. These findings indicate that Globularicitrin may serve as a more effective ligand for hSERT modulation due to its superior interaction profile. However, further studies, including molecular dynamics simulations, has been validated its potential as a novel hSERT inhibitor as detailed in the following section. The 3D and 2D docking interaction images of Globularicitrin and Fluoxetine are presented in Figures 1 and 2, respectively. The docking results of other flavonoids, including their interaction profiles, are provided in the Supplementary File for further reference.

Table 2: Tabular Representation of XP Docking Scores for In-Silico Screening of Flavonoids Against Human Serotonin Transporter (PDB ID: 5I6X)

3.3. Binding Free Energy Calculation Using the Prime/MM-GBSA Approach

The stability of the docking complex was assessed using MM-GBSA free energy calculations as a post-scoring approach following molecular docking with the human serotonin transporter (hSERT), with values presented in Table 3. The dG Bind values ranged from -44.83 to 21.64 kcal/mol, dG Coulomb values varied between -54.26 and 7.1 kcal/mol, and dG hydrogen bond contributions were between -5.03 and -0.25 kcal/mol, with binding energy positively contributing to the total energy. The lowest energy poses of the scoring function confirmed the stability of the docking complex, with G-scores closely aligning with experimentally determined binding energies from X-ray crystallography. The G-score and MM-GBSA energy values within the active site further supported the stability of the docking interactions. To evaluate binding stability, MM-GBSA calculations showed Globularicitrin exhibited a ΔG Bind of -23.16 kcal/mol, whereas Fluoxetine had a more favorable ΔG Bind of -40.86 kcal/mol. Further decomposition of the MM-GBSA results indicated that Globularicitrin had a significant Coulombic energy contribution (-54.26 kcal/mol), supporting strong electrostatic interactions, whereas Fluoxetine’s Coulombic energy was much lower (-1.47 kcal/mol), implying that van der Waals and hydrophobic interactions played a more dominant role in its binding. Additionally, hydrogen bonding contributions were -5.03 kcal/mol for Globularicitrin and -0.65 kcal/mol for Fluoxetine, confirming that Globularicitrin formed stronger hydrogen bonds than Fluoxetine.

3.4. Molecular Dynamics stimulations:

Structural Stability Analysis - Protein Stability and RMSD Analysis:

The Root Mean Square Deviation (RMSD) analysis was performed to assess the stability of 5I6X after Globularicitrin binding. The RMSD of Cα atoms showed fluctuations between 1.8 and 2.9 Å, indicating a relatively stable protein structure. The stabilization of RMSD towards the later stages of the trajectory confirms that the system reached equilibrium, ensuring the reliability of further structural and interaction analyses. Minor fluctuations observed suggest localized conformational changes, but the overall protein structure remained intact, validating the stability of the ligand-protein complex as given in figure 3.

Figure 3: Demonstrates Structural Stability Analysis - Protein Stability and Root Mean Square Deviation (RMSD) Analysis

Ligand Stability and RMSD Analysis:

The ligand RMSD was analyzed to determine the stability of Globularicitrin in the 5I6X binding pocket. The initial fluctuations between 1.6 and 2.8 Å indicate that the ligand was adjusting to the binding site. However, after 18-25 ns, the ligand stabilized in the range of 2.0 to 2.4 Å, maintaining this conformation for the remainder of the simulation. This suggests that Globularicitrin remained bound within the active site without significant displacement, confirming its strong binding affinity as shown in figure 3.A stable ligand RMSD relative to the protein backbone indicates that the ligand maintained a consistent binding orientation, which is crucial for potential drug candidates.

Protein Flexibility and RMSF Analysis:

The Root Mean Square Fluctuation (RMSF) was used to assess local flexibility in the 5I6X protein. Most of the protein exhibited low RMSF values, indicating high structural stability. However, residues 120-140 showed higher fluctuations (~3.5 Å), corresponding to loop regions, which are inherently more flexible. Another fluctuation peak was observed in the 420-520 residue range, suggesting localized dynamic movements. Despite these fluctuations, the overall structure remained stable, ensuring a secure ligand-binding environment as shown in figure 4. The interaction of Globularicitrin with rigid regions of the protein further reinforces the stability of the protein-ligand complex.

Figure 4: Root-mean-square fluctuation (RMSF) of the simulated protein 5I6X in complex with Globularicitrin during a 100 ns molecular dynamics (MD) simulation.

Protein Secondary Structure Stability:

The secondary structure analysis of the protein during the simulation revealed that alpha-helices and beta-strands were largely maintained throughout the 100 ns trajectory, ensuring that the protein did not undergo major conformational changes. The percentage of secondary structure elements (SSE) remained at ~59.2%, with 58.54% helices and 0.66% strands, confirming that the protein retained its native structure after ligand binding as Visualized in figure 5.

Figure 5 Illustrated Protein secondary structure elements (SSE) like alpha-helices and beta-strands are monitored throughout the simulation.

The stable secondary structure distribution supports the idea that Globularicitrin does not destabilize the 5I6X transporter, making it a promising candidate for further studies.

Protein-Ligand Interaction Analysis:

The protein-ligand interaction analysis revealed multiple stabilizing forces, including hydrogen bonds, hydrophobic interactions, ionic interactions, and water bridges. The most persistent hydrogen bonds were observed with ASP 98 (90% occupancy), ARG 104 (94%), and TYR 95 (86%), indicating their significant role in ligand stabilization. Hydrophobic interactions were frequent with TYR 95, ALA 169, and TYR 175, reinforcing ligand positioning within the active pocket. Furthermore, water-bridged interactions were prominent with ASN 177, PHE 334, and TRP 103, suggesting that solvent molecules contribute to the ligand’s stability in the binding site as Presented in given figure 6 & 7.

Figure 6: Interaction fraction of the 5I6X protein complexed with Globularicitrin during a 100 ns molecular dynamics (MD) simulation.

Figure 7: Timeline representation of various contacts formed by Globularicitrin in complex with the 5I6X protein during a 100 ns molecular dynamics (MD) simulation.

Interaction Timeline and Stability:

The interaction timeline analysis confirmed that key residues maintained stable interactions with the ligand. SER 336 exhibited interaction for the first 20 ns, lost contact temporarily, and regained it from 60-100 ns, suggesting a transient binding pattern. In contrast, LEU 337 and GLY 435 did not participate in significant interactions. The 2D ligand interaction diagram indicated that ASP 98, ARG 104, and TYR 95 consistently engaged in hydrogen bonding, while hydrophobic and water-mediated interactions remained stable throughout the simulation. The strong and persistent interactions suggest a high binding affinity of Globularicitrin for 5I6X, reinforcing its potential as a SERT transporter inhibitor as outlined in figure 8.

Figure 8: 2D interaction diagram of Globularicitrin in complex with the 5I6X protein during the 100 ns MD simulation trajectory of the protein-ligand complex.

Ligand Conformational Analysis and Torsion Profile:

The ligand RMSF analysis provided insights into atom-level fluctuations, revealing minimal changes, which suggests that Globularicitrin maintained its binding conformation without excessive movement. The torsional analysis of the ligand confirmed that the rotatable bonds remained stable, preventing major entropic penalties and reinforcing a well-fitted docking pose. Additionally, the ligand’s molecular surface area (MolSA), solvent-accessible surface area (SASA), and polar surface area (PSA) remained consistent, indicating that the ligand did not undergo significant conformational strain as represented in figure 9.

Figure 9 represents the ligand properties, including ligand RMSD, radius of gyration (rGyr), intramolecular hydrogen bonds (intraHB), molecular surface area (MolSA), solvent-accessible surface area (SASA), and polar surface area (PSA), were analyzed. Additionally, the root-mean-square fluctuation (RMSF) of both the ligand and protein in complex with 5I6X was evaluated during the 100 ns molecular dynamics (MD) simulation.

DISCUSSION

The in-silico evaluation of flavonoid-based inhibitors for hSERT revealed Globularicitrin as a promising candidate, exhibiting superior binding affinity and stability compared to the standard SSRI, Fluoxetine. Molecular docking analysis in XP mode yielded a significantly higher docking score (-15.357) for Globularicitrin versus Fluoxetine (-10.306), suggesting enhanced ligand-receptor complementarity. ADMET profiling confirmed its favorable pharmacokinetic properties, including optimal oral bioavailability (94–100%) and adherence to Lipinski’s Rule of Five. MM-GBSA free energy calculations (-23.16 kcal/mol) indicated thermodynamic stability of the Globularicitrin-hSERT complex, supported by favorable Coulombic and hydrogen bonding contributions. Molecular dynamics simulations over a 100 ns trajectory further validated its stability, with RMSD fluctuations confined to 2.0–2.4 Å and consistent hydrogen bonding interactions with key residues ASP-98 (90%), ARG-104 (94%), and TYR-95 (86%). Secondary structure analysis demonstrated minimal conformational perturbations, reinforcing the structural integrity of hSERT upon ligand binding. Collectively, these findings position Globularicitrin as a potent natural modulator of serotonin reuptake, necessitating further biochemical and preclinical investigations to elucidate its antidepressant potential.

CONCLUSION:

This study presents a computational framework for the rational design of flavonoid-based serotonin transporter (hSERT) inhibitors as potential antidepressant agents. A multi-tiered in-silico strategy, integrating molecular docking, ADMET profiling, MM-GBSA binding free energy calculations, and molecular dynamics (MD) simulations, identified Globularicitrin as a lead candidate with superior binding affinity and stability compared to Fluoxetine. The structural and energetic analyses revealed key interaction patterns, including hydrogen bonding and hydrophobic contacts, contributing to its inhibitory potential. ADMET screening confirmed its favorable pharmacokinetic and safety profile, reinforcing its drug-likeness. The study provides mechanistic insights into flavonoid-hSERT interactions, supporting the development of naturally derived antidepressants with improved therapeutic indices. The 100 ns molecular dynamics simulation demonstrated that Globularicitrin binds stably within the 5I6X binding site, with strong hydrogen bonds, hydrophobic contacts, and water bridges supporting its binding affinity. The protein structure remained stable, showing only minor fluctuations in loop regions, while the ligand maintained its binding pose with minimal RMSD deviations. The persistent interactions of ASP 98, ARG 104, and TYR 95 suggest that Globularicitrin could effectively target the SERT transporter, making it a strong candidate for further experimental validation and optimization of flavonoid-based hSERT inhibitors in preclinical and clinical Applications.

Abbreviations

MDD – Major Depressive Disorder

hSERT – Human Serotonin Transporter

SSRI – Selective Serotonin Reuptake Inhibitor

SNRI – Serotonin-Norepinephrine Reuptake Inhibitor

ADMET – Absorption, Distribution, Metabolism, Excretion, and Toxicity

MM/GBSA – Molecular Mechanics/Generalized Born Surface Area

MD – Molecular Dynamics

HT – 5-Hydroxytryptamine (Serotonin)

PDB – Protein Data Bank

XP – Extra Precision (used in molecular docking)

ΔGbind – Binding Free Energy in Prime/MM-GBSA calculations

VSGB – Variable-dielectric Surface Generalized Born (solvent model in Prime)

OPLS4 – Optimized Potentials for Liquid Simulations (Force Field used in Prime calculations)

dG Coulomb – Electrostatic Energy Contribution in Prime/MM-GBSA

dG Hbond – Hydrogen Bond Energy Contribution in Prime/MM-GBSA

dG Bind Covalent – Covalent Energy Contribution in Prime/MM-GBSA

Prime Energy – Total Energy of the Ligand-Receptor Complex in Prime

 G-score – Glide Score (used in docking before Prime/MM-GBSA refinement)

Clinical trial number: not applicable.’

Ethics approval and consent to participate- Not applicable

Human Ethics and Consent to Participate declarations-NOT APPLICABLE.

FIGURE LEGEND:  Molecular docking results using Glide Module (XP- precision)

1. The below figure represents Ligand interaction of flavonoid Globularicitrin with Human Serotonin Transporter (5I6X) dock score (-15.357)

2. The below figure represents Ligand interaction of flavonoid kameferol 7,4 Dimethyl Ether -3-O-Beta-D-Gluco pyranoside with Human Serotonin Transporter (5I6X) dock score (-14.13)

3. The below figure represents Ligand interaction of flavonoid Isoquercetin with Human Serotonin Transporter (5I6X) dock score (-13.208)

4. The below figure represents Ligand interaction of flavonoid hesperidin with Human Serotonin Transporter (5I6X) dock score (-13.048)

5. The below figure represents Ligand interaction of flavonoid Hyperoside with Human Serotonin Transporter (5I6X) dock score (-10.76)

6. The below figure represents Ligand interaction of flavonoid Fluoxetine with Human Serotonin Transporter (5I6X) dock score (-10.306)

7. The below figure represents Ligand interaction of flavonoid Naringin with Human Serotonin Transporter (5I6X) dock score (-10.215)

8. The below figure represents Ligand interaction of flavonoid Fiestin with Human Serotonin Transporter (5I6X) dock score (-9.993)

9. The below figure represents Ligand interaction of flavonoid Scutellarin with Human Serotonin Transporter (5I6X) dock score (-9.846)

10. The below figure represents Ligand interaction of flavonoid Miquelianin with Human Serotonin Transporter (5I6X) dock score (-9.702)

Table legend: ADMET Profile

A comprehensive list of all ADMET parameters is provided according to the recommended range limits specified in the QikProp Module User Manual 4.4