Chromone–Thiophene Hybrids as Non-Peptidic Inhibitors of SARS-CoV-2 Mpro: Integrated ADME, Docking, and Molecular Dynamics Approach

Introduction

The COVID‑19 pandemic, caused by the emergent SARS‑CoV‑2 coronavirus, has exposed the vulnerability of health systems to highly mutable and rapidly spreading pathogens (Acter et al., 2020). Although effective vaccines were deployed at record speed, the continual emergence of variants with broadened tropism and partial immune escape underscores the need for direct‑acting antivirals that target essential viral functions independently of spike‑protein drift (Acter et al., 2020). Chief among these functions is the main protease (Mpro, nsp5), which cleaves the viral polyprotein pp1ab into eleven non‑structural proteins and thereby initiates genome replication and transcription (Yadav et al., 2021). Because Mpro lacks close human homologues, off‑target liabilities are minimal, making it an ideal target for rational small‑molecule design (Friman et al., 2021).

First‑generation inhibitors such as nirmatrelvir rely on a covalent warhead grafted onto a peptidomimetic scaffold (Chen et al., 2023). Although clinically validated, these compounds exhibit high polarity (TPSA > 110 Å²) and marked flexibility, leading to limited oral absorption and the frequent requirement for a pharmacokinetic booster (ritonavir) and strict therapeutic monitoring (Ahkam et al., 2020). By contrast, low‑molecular‑weight non‑peptidic inhibitors promise simplified dosing and improved tissue distribution provided they combine (a) an ADME profile compatible with oral administration, (b) sufficient selectivity to avoid human proteases and (c) adequate metabolic resilience (Bekkouch et al., 2024). Against this backdrop, the chromone scaffold offers an attractive platform: its rigid, electron‑rich bicyclic system common to many natural and synthetic antivirals minimizes entropic penalties upon binding and provides two strategic hydrogen‑bond acceptors (the C‑4 carbonyl and the ring oxygen at position 1) (Silva et al., 2018). Conjugation with a thiophene ring, renowned for π‑sulfur contacts and metabolic stability, strengthens hydrophobic interactions and exploits the cysteine‑rich environment of the Mpro catalytic cleft (Lv et al., 2022). Controlled installation of hydroxyl groups further modulates polarity without exceeding the critical polar‑surface threshold, thereby optimizing membrane permeability (Wang et al., 2025). Complementing synthetic chemistry, in silico tools have revolutionized early candidate triage: SwissADME, ProTox‑III, and AutoDock Vina rapidly assess pharmaceutical potential, predicted toxicity, and binding affinity, respectively, directing synthetic effort toward the most promising scaffolds (Bhat & Chatterjee, 2021). Molecular‑dynamics (MD) simulations add a kinetic dimension by revealing the persistence of key interactions and the plasticity of the targeted pocket crucial criteria for anticipating inhibitor robustness in a biological context (Bourassi et al., 2023). Starting from these design principles, we generated a small library of chromone‑thiophene hybrids and filtered them through successive ADME and toxicity checkpoints (Bozorgpour et al., 2023). One analogue here designated M3 emerged as the lead scaffold because it formed a dual hydrogen‑bond clamp with key residues in the S1 pocket of Mpro and established complementary π‑sulfur and π‑cation interactions with the catalytic dyad and surrounding cysteines. Explicit‑solvent molecular‑dynamics simulations confirmed that this interaction network remains stable over an extended time scale, while the protein backbone itself exhibits minimal conformational drift, suggesting that M3 enforces a “locked” active‑site geometry. These convergent in silico data nominate M3 as a synthetically accessible lead for biochemical and cellular validation and illustrate more broadly how chromone–thiophene hybridisation can be leveraged to develop next‑generation inhibitors of coronavirus proteases.

Materials and Methods

Pharmacokinetic Analysis Using Computational Tools

A compound’s pharmacokinetic fate absorption, distribution, metabolism, and excretion (ADME) is dictated by a cascade of biochemical and physiological events that determine how it moves through, transforms within, and exits the body (Loukili et al., 2024). A rigorous grasp of these parameters clarifies the pathways by which molecules cross biological membranes, circulate systemically, undergo biotransformation, and are ultimately cleared. In silico approaches have become indispensable for probing ADME behavior: they simulate membrane permeation, ligand-biomolecule interactions, and metabolic stability, yielding early predictions of efficacy and safety (Fraj et al., 2025).

In the present study, each candidate structure was sketched in ChemDraw and exported as a SMILES string, which served as the input for SwissADME calculations (Abbaoui et al., 2024). This web‑based platform delivered an integrated pharmacokinetic forecast covering permeability, metabolic liability, and toxicity indices thereby furnishing a detailed picture of the compounds’ prospective performance under physiological conditions.

Prediction of the Toxicity Analysis (Pro Tox III)

Predictive toxicity profiling was carried out with the ProTox‑III web platform, following established best‑practice protocols. Using the SMILES strings generated and verified in ChemDraw (Permatasari et al., 2024), the tool leverages expansive toxicological databases, machine‑learning algorithms, and advanced statistical models to interrogate each molecule’s structural alerts. ProTox‑III yields forecasts for multiple toxicological endpoints most notably acute LD₅₀ values and corresponding GHS hazard categories thereby illuminating liabilities that might arise in vivo (Drwal et al., 2014). This streamlined, high‑resolution analysis underpins the rational triage of candidate compounds for downstream experimental validation and expedites data‑driven decision‑making in both drug development and environmental safety contexts.

PyRx: Preparation, Configuration, and Validation of the Docking Protocol

Molecular‑docking calculations were executed in PyRx, which couples the AutoDock Vina engine to the AutoDock Tools (ADT v1.5.7) interface (Guarimata et al., 2023). The crystal structure of Mpro was pre‑processed by deleting all crystallographic waters, appending polar hydrogens, and assigning Gasteiger charges in accordance with AutoDock prescriptions; residual clean‑up and structural checks were completed in Discovery Studio, which was also used post‑docking to inspect poses and map intermolecular contacts (Bourassi et al., 2024). Each ligand was sketched and conformationally refined in ChemDraw, subjected to an energy-minimization step, and finally converted to PDBQT format via ADT to ensure correct atom types and charge states. Docking searches were confined to a grid box centered at (‑14.30, 14.40, 1.00) Å with edge lengths of 9.87 × 6.59 × 8.82 Å; the exhaustiveness parameter was fixed at 8, while all other Vina settings were left at their defaults. To verify the protocol, the co-crystallized inhibitor was re‑docked into the active site, yielding a best‑fit pose with an RMSD of 1.151 Å relative to the experimental geometry-well beneath the 2 Å benchmark-thereby attesting to the robustness and predictive reliability of the docking workflow (Ouahabi et al., 2024).

Implementation of Molecular Dynamics Simulations Using GROMACS

The protein model (P.pdb) was processed with gmx pdb2gmx using the AMBER99SB‑ILDN force field, which appended all missing hydrogens and assigned the appropriate protonation states. The ligand (LIG.pdb) was parametrised separately; its topology (.itp) and coordinate (.gro) files were generated and verified before being merged with the protein to yield the protein-ligand complex (complex.pdb) (Behera et al., 2025). This complex was centered in a cubic box, solvated with TIP3P water, and electrically neutralized by the addition of counter‑ions. After the steepest‑descent energy minimisation, the system underwent the standard NVT and NPT equilibration phases. A production molecular dynamics run of 100 ns was then executed, saving coordinates and velocities at regular intervals (Merzouki et al., 2024). This trajectory provides a detailed picture of the complex’s stability, conformational fluctuations, and intermolecular interactions under near‑physiological conditions.

Results and Discussion

Comparative Analysis of ADME Profiles and Drug-Likeness Using SwissADME Radar Charts

The analysis of physicochemical and pharmacokinetic profiles generated by SwissADME radar charts (Figure 1) highlights distinct characteristics for chromone derivatives substituted with a thiophene ring (M1–M4), along with a marked contrast to ritonavir (M5). The M1 molecule, which lacks hydroxyl substituents, exhibits moderate lipophilicity and low polarity, favoring efficient membrane permeability but slightly limiting aqueous solubility. This behavior directly stems from the lipophilic nature of the thiophene moiety combined with the chromone scaffold, which lacks polar groups capable of significantly enhancing solubility. Molecules M2 and M3, bearing hydroxyl groups at positions 3 and 5 respectively, show a notable increase in polarity, resulting in enhanced potential solubility and a slight reduction in lipophilicity compared to M1. These structural modifications, through hydroxyl substitution, promote hydrogen bonding with water molecules, thereby improving potential oral bioavailability without significantly impairing membrane permeability. Molecule M4, bearing two hydroxyl groups, displays the most balanced profile among the chromone-thiophene derivatives. Its markedly increased polarity, and consequently improved expected solubility, results from the combined presence of hydroxyl groups at positions 3 and 5, substantially enhancing interactions with aqueous biological environments while maintaining sufficient lipophilic character to support effective membrane penetration. In contrast, ritonavir (M5), with its highly complex structure composed of multiple heterocycles (thiazoles), amide and carbamate linkages, and methylated substituents, exhibits a radically different profile: high lipophilicity, large molecular size, increased molecular flexibility, and higher polarity. These features result in limited solubility and potentially problematic pharmacokinetics due to poor intestinal absorption and reduced oral bioavailability. In summary, the chromone-thiophene derivatives (M1–M4) display advantageous ADME profiles, strongly influenced by progressive hydroxylation, offering an optimal balance between solubility and lipophilicity. In contrast, ritonavir, due to its high structural complexity, presents less favorable pharmacokinetic properties, thus underscoring the therapeutic potential of substituted chromone derivatives with improved ADME characteristics (Oualdi et al., 2025).

Profils ADME des molécules étudiées évalués par SwissADME et la méthode « Boiled Egg »

Table 1. Physicochemical, pharmacokinetic, and drug-likeness properties of the studied compounds compared to Ritonavir.

The molecular weights of the investigated compounds range from 228.27 to 260.27 g/mol, significantly lower than that of ritonavir (720.94 g/mol), suggesting improved potential for oral bioavailability due to more efficient diffusion across biological membranes. Regarding hydrogen bonding, the thiophene derivatives display a moderate number of hydrogen bond acceptors (2 to 4) and donors (0 to 2), markedly lower than ritonavir (7 acceptors and 4 donors). This reduced hydrogen-bonding capacity limits nonspecific and potentially undesirable interactions with plasma proteins or metabolic enzymes, thereby favoring better absorption (Table 1). Each compound possesses only one rotatable bond, in stark contrast to the 22 found in ritonavir. This increased structural rigidity is beneficial for enhancing ligand-receptor binding specificity and selectivity while reducing the likelihood of extensive metabolism and associated toxicity. Topological Polar Surface Area (TPSA) increases with hydroxyl substitution (from 58.45 Ų for compound 1 to 98.91 Ų for compound 4), yet remains within the optimal range for good oral bioavailability, in contrast to the very high TPSA of ritonavir (202.26 Ų), which likely hinders its intestinal absorption. The calculated logP (CLogP) values for the chromone-thiophene compounds, ranging from 2.5 to 3.08, reflect a well-balanced lipophilicity conducive to gastrointestinal absorption. In contrast, ritonavir exhibits a much higher CLogP value (5.19), which compromises aqueous solubility and poses formulation and absorption challenges. Moreover, all four chromone-thiophene derivatives fully comply with both Lipinski's and Veber's rules, with zero violations, while ritonavir shows two violations of each rule—highlighting its reduced potential for oral bioavailability.Ces observations sont corroborées par un « bioavailability score » stable et élevé (0,55) pour les dérivés chromoniques substitués, comparé à une valeur basse (0,17) pour le Ritonavir.

The studied thiophene-substituted chromone derivatives exhibit favorable ADME profiles compared to ritonavir, thereby demonstrating strong potential for effective therapeutic application with optimal oral bioavailability. The Boiled-Egg diagram analysis (Figure 2) for thiophene-substituted chromone derivatives (M1–M4) highlights the influence of structural modifications on human intestinal absorption (HIA) and the ability to cross the blood-brain barrier (BBB). Compound M1, which lacks hydroxyl substitution, is distinctly located within the yellow region, indicating excellent BBB penetration capacity. This efficient permeation is attributed to its low polarity—resulting from the absence of highly polar functional groups—and moderate lipophilicity conferred by the thiophene and fused chromone rings, which facilitate passive diffusion across the lipophilic membranes of the BBB. Compounds M2 and M3, each bearing a single hydroxyl group at positions 3 and 5, respectively, are positioned at the boundary between optimal intestinal absorption and potential BBB permeability. The addition of a hydroxyl group moderately increases their polarity and topological polar surface area (TPSA), slightly reducing their ability to cross the BBB while maintaining excellent gastrointestinal absorption. These characteristics directly reflect structural modifications that enhance hydrogen-bonding potential, thereby decreasing passive membrane diffusion across the brain barrier. Compound M4, bearing two hydroxyl substituents, exhibits the highest TPSA among the thiophene derivatives studied, placing it squarely within the region associated exclusively with good intestinal absorption and negligible BBB penetration. The dual hydroxylation significantly increases its polarity and hydrogen-bonding capacity, severely limiting its ability to diffuse passively across lipophilic barriers, thus making it more suitable for peripheral therapeutic applications. In contrast, ritonavir, with its extremely high TPSA and considerable lipophilicity, would not be favorably positioned on the diagram, displaying substantially limited intestinal absorption and minimal to no BBB permeability. These limitations stem directly from its complex structure, which includes multiple highly polar functional groups and large molecular size, both of which significantly hinder passive membrane diffusion.

Structure–Toxicity Correlation of the Investigated Molecules Compared to Ritonavir

The predictive analysis of acute oral toxicity (LD₅₀) conducted using ProTox-III (Figure 3) reveals that the thiophene-substituted chromone derivatives (M1–M4) exhibit moderate toxicity levels, falling within toxicity class 4, which corresponds to potentially harmful compounds when administered orally (LD₅₀ values ranging from 600 to 1070 mg/kg). Compounds M1 and M3, characterized by the absence or presence of only one hydroxyl group, display slightly higher LD₅₀ values (1070 mg/kg), indicating relatively moderate toxicity within this class. Structurally, these results can be explained by the relatively simple architecture of the compounds, which lack highly polar or reactive substituents, thereby limiting their potential interactions with biomolecules. The absence or singular presence of hydroxyl groups reduces the likelihood of electrophilic or nucleophilic interactions that might generate toxic effects. These findings are supported by average similarity scores of 65.35% (M1) and 63.25% (M3), with a predictive accuracy of 68.07% for each, suggesting a moderate to good reliability of these predictions and reinforcing the proposed structure–toxicity rationale. In contrast, compounds M2 and M4 show lower LD₅₀ values (600 mg/kg), indicating a higher predicted acute toxicity. This relative increase in toxicity compared to M1 and M3 is directly linked to the presence of one (M2) or two hydroxyl groups (M4), which enhance chemical reactivity and polarity, increasing the potential for intense and potentially harmful interactions with biological targets. Furthermore, the greater number of hydroxyl groups may render these compounds more susceptible to metabolic transformation, potentially leading to the formation of more reactive and toxic intermediates. These results are supported by average similarity scores of 59.75% (M2) and 61.15% (M4), with predictive accuracy of 67.38% (M2) and 68.07% (M4), respectively, indicating acceptable confidence in the structural validity of the predictions. In comparison, ritonavir has an estimated LD₅₀ of 1000 mg/kg (also toxicity class 4), similar to the thiophene-based derivatives. However, ritonavir shows a much lower average similarity score (42.32%) and limited predictive accuracy (54.26%), reflecting its structural complexity, which includes thiazole, ureido, carbamate, and methyl-substituted groups. This complexity increases the likelihood of nonspecific interactions and unpredictable metabolic transformations, potentially resulting in higher toxicity under real biological conditions. The low predictive accuracy thus highlights the inherent difficulty in reliably estimating the toxicological effects of highly complex molecules. In summary, the predictive data from ProTox-III demonstrate how the presence and positioning of hydroxyl substituents, as well as overall chemical complexity, strongly influence the toxicological profiles of the studied compounds. Despite exhibiting comparable toxicity classes, the relative structural simplicity of the thiophene–chromone derivatives confers a more predictable and interpretable toxicological profile compared to ritonavir, whose structural intricacy leads to greater predictive uncertainty.

Molecular Docking

The X‑ray structure of the SARS‑CoV‑2 main protease (Mpro, PDB 9C8Q; 1.69 Å), a homodimeric cysteine protease that orchestrates polyprotein cleavage and hence viral replication. Catalysis hinges on the canonical Cys145-His41 and Glu166 dyad, whose precise orientation is stabilized only in the dimeric A₂ assembly and is framed by subsites S1′-S4 that dictate substrate specificity (Barkan et al., 2024). To gauge the aptitude of our four chromone‑thiophene candidates (M1-M4) to occlude this pocket, we performed a structure‑based docking campaign with ritonavir as a benchmark. Although developed against HIV protease, ritonavir’s broad protease‑inhibition profile makes it a stringent reference for binding efficiency. By comparing the geometry and energetics of key ligand-residue contacts to those of ritonavir, we aim to single out the scaffold(s) most capable of neutralising Mpro and thus warranting prioritisation for optimization and subsequent in‑vitro validation.

Table 2. Calculated binding energies (ΔG) and principal ligand‑residue contacts for the candidate molecules docked into SARS‑CoV‑2 main protease (M^pro).

A comparative reading of the binding energies reported in Table 2 (‑6.5 to ‑6.8 kcal mol⁻¹) reveals a nearly uniform energetic landscape for the chromone‑thiophene series, yet molecule M3 clearly distinguishes itself. It is the only candidate that forges two hydrogen bonds toward His163 (2.24 Å) and Glu166 (3.26 Å) anchoring its aromatic core deep within the S1 sub‑site while simultaneously locking the rim of the catalytic channel, a strategic position that constrains the geometry of the Cys145-His41 dyad and promotes closure of the active site. This polar engagement, absent in M1, M2, and M4, provides a distinct enthalpic contribution that, despite its modest absolute ΔG (‑6.8 kcal mol⁻¹), is expected to translate in dynamics into enhanced complex stability and reduced susceptibility to water‑mediated displacement compared with analogs lacking H‑bond locks. Coupled with the intrinsic rigidity of M3 only one rotatable bond, as highlighted in the ADME profile lowers the entropic penalty and orients the scaffold optimally toward the functional triad Glu166-His41-Cys145, mimicking the anchoring motif observed for ritonavir but with a steric footprint and lipophilicity far more compatible with drug‑likeness constraints. Collectively, these factors rationalize the selection of M3 as the lead compound; its complete interaction network with SARS‑CoV‑2 Mpro is detailed in Figure 4, which will serve as the structural basis for subsequent optimization and experimental validation cycles.

Figure 4 reveals an intricately interlaced interaction network that explains compound M3’s superiority within the active cleft of SARS‑CoV‑2 Mpro (PDB 9C8Q). The carbonyl at position 4 of the chromone scaffold forms a short conventional hydrogen bond with His163 (2.24 Å), while the side‑chain carbonyl at position 2 secures the ligand’s rear flank through a second H‑bond to Glu166 (3.26 Å); together, these polar locks immobilize the ligand and curb re‑orientation of the catalytic dyad Cys145-His41. The thiophene ring serves as a dual hydrophobic anchor: π‑sulfur and π‑alkyl contacts with Cys44 and Cys145 (5.38 Å and 5.86 Å, respectively) wrap around the enzyme’s nucleophilic sulfur atoms, and strong π‑cation interaction with His41 (3.97 Å and 4.47 Å) further seals the catalytic channel. Additional π‑alkyl links to Met49 and Cys145, complemented by an extensive van‑der‑Waals sheath involving S2/S4 residues (Phe140, Leu141, Asn142, Ser144, Met165, His164), fashion a hydrophobic cradle that excludes water and minimizes entropic penalties. Surface analyses reinforce this picture: the donor/acceptor map (magenta = donor, green = acceptor) places M3’s two carbonyl oxygens precisely on green niches defined by His163 and Glu166, rationalising the inevitability of the twin H‑bonds; the aromatic edge/face plot colors the thiophene π‑face orange against a blue edge on Cys44/Cys145, evidencing a stabilising face‑to‑edge stack; the interpolated electrostatic surface displays a red anionic basin from Glu166 that dovetails with the partially positive chromone carbonyl, while the neutral crest above accommodates the non‑polar thiophene, collectively minimising desolvation costs; and the solvent‑accessible‑surface (SAS) map shows more than two‑thirds of the ligand buried in low‑exposure blue (< 12 Å²) regions, confirming tight steric complementarity. This integrated arrangement of twin H‑bonds toward the S1 recognition site, a π‑cation latch on His41, π‑sulfur bridges to key cysteines, and surface‑level electrostatic and topological complementarity underpins the binding energy of ‑6.8 kcal mol⁻¹ reported in Table 1 and enables M3 to dwell in the pocket long enough to impair protease maturation, positioning it as a compelling lead for further optimization and biological validation.