In silico prediction, molecular docking, and dynamics analysis of steroidal alkaloids from the genus Fritillaria: implications for designing novel antiparkinsonian therapeutic strategies

Abstract

Parkinson’s disease (PD) is the second most prevalent neurodegenerative disorder, affecting nearly 0.3% of the global population. Its pathology is primarily linked to dopaminergic neuronal loss in the substantia nigra, leading to hallmark motor impairments such as tremor, rigidity, and bradykinesia. A defining molecular feature of PD is the aberrant aggregation of α-synuclein, alongside dysregulation of proteins such as MAO-B, COMT, and LRRK2, which collectively contribute to disease progression. Within the current research, these proteins were designated as docking targets to explore the enzyme-modulating activity and the therapeutic promise of steroidal alkaloid candidates from the genus Fritillaria, a taxon long recognized in traditional medicine for its neuroprotective properties. Docking analyses revealed that among 70 compounds analysed, compound 65 exhibited strong MAO-B inhibitory activity (binding energy − 11 kcal/mol), compound 5 demonstrated pronounced COMT inhibition (− 9 kcal/mol), and compound 42 emerged as a promising dual-acting agent capable of targeting both enzymes. Favorable physicochemical attributes, including optimal lipophilicity, low polar surface area, and blood-brain barrier permeability, further support their suitability. These findings identify preliminary computational leads that warrant further experimental validation for potential future development.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-44050-1.

Introduction

Parkinson’s disease (PD) is a progressive neurodegenerative disorder affecting over 6 million people globally, with approximately 60,000 new U.S. cases annually. It typically emerges after the age of 55 and was linked to nearly 18,000 U.S. deaths in 2003. The disease is mechanistically driven by dopaminergic neuronal loss in the substantia nigra, leading to motor symptoms such as tremors, bradykinesia, rigidity, and postural instability, as well as non-motor manifestations including cognitive decline, depression, and dementia. Both genetic and environmental factors contribute to PD pathogenesis, with potential mechanisms involving oxidative stress, mitochondrial dysfunction, and α-synuclein aggregation^1–3.

Existing therapeutic modalities, including pharmacotherapy ranging from carbidopa/levodopa, rehabilitation, and advanced options like deep brain stimulation for drug-resistant cases, remain largely palliative, addressing symptoms without halting neurodegeneration⁴.

Emerging therapies including stem cell-based strategies and α-synuclein-targeted immunotherapy, remain at exploratory stages^5–7. Monoamine oxidase (MAO) enzymes, particularly MAO-B, are key therapeutic targets due to their role in neurotransmitter metabolism. MAO-B inhibitors like selegiline enhance dopamine levels and may reduce α-synuclein accumulation, offering benefits in PD and Alzheimer’s diseases^8–11. Cholinesterase inhibitors are also being explored for related neurodegenerative conditions¹².

Despite advances in symptomatic management, the lack of disease-modifying therapies has intensified interest in plant-derived compounds as alternatives or adjuncts to conventional treatments. In this context, phytochemicals have emerged as compelling candidates for disease-modifying interventions. Among these, A growing body of evidence strongly support steroidal alkaloids as promising lead candidates for Parkinson’s disease drug discovery, owing to their structural diversity, ethnopharmacological precedent, and multi-target neuroprotective mechanisms¹³. In particular, steroidal alkaloids from the genus Fritillaria hold distinctive promise due to their pronounced structural diversity, well-established ethnopharmacological relevance, and capacity to engage multiple therapeutic targets^14–17. Their molecular scaffolds, spanning solanidine, jervine, verazine, veratramine, and cevanine types¹⁸, are well represented in Fritillaria species historically used in traditional Chinese and Iranian medicine. Importantly, precedent for clinical translation exists, as illustrated by the steroidal alkaloid derivative abiraterone acetate (Zytiga), an FDA-approved therapeutic¹⁹.

Computational and experimental investigations substantiate their pharmacological potential: Bhat et al. demonstrated potent MAO-B inhibition with favorable pharmacokinetic profiles of steroidal alkaloids and their structural compliance with Lipinski rules and drug-likeness²⁰, while Kumar et al.. revealed extensive multi-target interactions of Fritillaria steroidal alkaloids using network pharmacology and molecular docking¹³. Empirical studies further support their biological breadth. Steroidal alkaloids from F. cirrhosa and F. thunbergii exhibit pronounced anti-inflammatory effects²¹, and nine alkaloids from F. unibracteata, including two unusual jervine-type derivatives with an unusual cis A/B ring system, significantly reduced inflammation in a CuSO₄-induced inflammation model. Likewise, isosteroidal alkaloids from F. ussuriensis inhibited LPS-induced nitric oxide production in vitro, underscoring their modulatory impact on inflammatory pathways²².

Beyond anti-inflammatory activity, these compounds demonstrate broad neuroprotective actions relevant to PD pathophysiology, including inhibition of acetylcholinesterase and butyrylcholinesterase²³ and identification of novel AChE-inhibiting alkaloids from F. cirrhosa²⁴. More broadly, alkaloids act through diverse mechanisms, muscarinic and adenosine receptor agonism, antioxidant effects, anti-amyloid activity, MAO inhibition, suppression of α-synuclein aggregation, and modulation of dopaminergic and nicotinic receptors, as well as NMDA antagonism²⁵. Preclinical and in silico studies also implicate steroidal alkaloids in the regulation of PD-related kinases such as CDK5, LRRK2, and GSK-3β, thereby reducing oxidative stress, restoring mitochondrial function, inhibiting apoptotic cascades, and mitigating α-synuclein toxicity; computational docking consistently reveals favorable affinities for multiple kinase domains²⁶. Collectively, these findings justify systematic interrogation of Fritillaria steroidal alkaloids against PD-related targets, bridging traditional pharmacognosy with modern computational pharmacology to identify novel scaffolds for disease-modifying therapy.

Computational approaches, such as molecular docking and dynamics simulations, provide a rational, cost-effective framework for accelerating drug discovery and generating mechanistic insights that lead to the identification of novel therapeutics^27,28. Although Fritillaria alkaloids are pharmacologically diverse, their computational interrogation against PD-related targets remains underexplored. Previous studies have identified natural MAO inhibitors and alkaloids targeting prolyl oligopeptidase (POP), which is involved in α-synuclein accumulation^29,30.

α-Synuclein, a presynaptic protein whose pathological aggregation contributes to neuronal loss and mitochondrial dysfunction, is a critical therapeutic target³¹. COMT (catechol-O-methyltransferase) inhibition enhances levodopa bioavailability by reducing its peripheral metabolism, thus improving PD treatment efficacy³². MAO-B inhibitors increase dopamine levels in the brain and are used as monotherapy or adjuncts to levodopa³³. LRRK2 (leucine-rich repeat kinase 2), a protein with kinase and GTPase activities, is also implicated in PD pathogenesis, and its inhibition may correct aberrant signaling and reduce dyskinesia³⁴.

Given the absence of disease-modifying therapies for PD and the compelling mechanistic profile of Fritillaria-derived steroidal alkaloids, systematic evaluation of these molecules against key PD-associated targets is warranted. Building upon traditional pharmacognostic insights and leveraging molecular docking and dynamics simulations, the present study investigates 67 such alkaloids with respect to their interactions with α-synuclein, MAO, COMT, and LRRK2. This integrative computational approach aims to delineate binding mechanisms, infer structure-activity relationships, and identify promising scaffolds possibly leading to novel antiparkinsonian therapeutic agents.

Materials and methods

Virtual screening and molecular docking

This study investigated a library of 67 steroidal and isosteroidal alkaloids derived from plants belonging to the Liliaceae family, particularly species within the Fritillaria genus. These specialized metabolites are characterized by a C27 cholestane carbon skeleton, a structural feature that contributes to their diverse biological activities. For detailed structural and mechanistic analysis, five representative core scaffolds from this library were highlighted: Cevanine, Veratramine, Jervine, Solanidine, and Verazine (Fig. 1). These representative compounds were selected based on their prominence in the Fritillaria genus and their relevance to Parkinson’s disease-related pharmacophores. All 67 compounds were evaluated for their potential therapeutic effects in Parkinson’s disease through two primary mechanisms: inhibition of α-synuclein aggregation targeting the pathological protein aggregation characteristic of Parkinson’s like Leucine-rich repeat kinase 2 (LRRK2), and Enzyme modulation Focusing on key Parkinson’s-related enzymes such as Monoamine oxidase B (MAO-B) and Catechol-O-methyl transferase (COMT). In addition to studying the inhibition of Parkinson’s-related enzymes, α-synuclein was directly docked to the targeted sites. Selective steroidal and isosteroidal alkaloids from Fritillaria were selected based on structural features, documented PD enzymes inhibitory activity, standard features with known bioactive steroidal alkaloids, and the presence of key pharmacophores, reactive moieties, and bioactive groups relevant to the neurodegenerative disease pathogenesis. For ligand preparation, Initial 2D structures were drawn using ChemDraw and saved in .cdx format. They were converted to 3D structures (.mol2 format) using ChemBio3D Ultra. The protonation states of ionizable groupsincluding the steroidal nitrogen atoms were assigned at physiological pH (~ 7.4) using the “Protonate3D” function in ChemBio3D Ultra prior to energy minimization and format conversion. The most probable tautomeric form for each ligand was selected based on standard heuristics and energetic stability. Following protonation state and tautomer assignment, the 3D structures of all ligands underwent a geometry optimization step. This optimization was performed using the MM2 force field implemented in ChemBio3D Ultra, with convergence criteria set to a gradient norm of 0.001 kcal/(mol·Å). The optimization process minimized the potential energy of each ligand, ensuring stable starting conformations free of significant steric clashes before proceeding to partial charge assignment and format conversion. Gasteiger partial charges and rotatable bonds were added, and the structures were finally converted to PDBQT format^35,36. The 3D structures of the target enzymes MAO-B (PDBID: 2V5Z) and COMT (PDBID: 3A7E) were retrieved from the RCSB Protein Data Bank. Both structures include their essential cofactors: the FAD (flavin adenine dinucleotide) molecule in MAO-B and the catalytic Mg²⁺ ion along with its coordinating water molecule and the SAM (S-adenosyl-L-methionine) cofactor in COMT. All co-crystallized water molecules were removed, except for the structural water molecule coordinating the catalytic Mg²⁺ ion in COMT, which was retained. Missing side chains and loops in the proteins were modeled and refined using the SWISS‑MODEL server’s automated homology modeling pipeline. Protonation states of protein residues were assigned at pH 7.4 using the PROPKA algorithm integrated into the PDB2PQR suite. The prepared structures subsequently underwent a brief energy minimization (500 steps of steepest descent followed by 500 steps of conjugate gradient) using the AMBER ff14SB force field to relieve steric clashes and optimize hydrogen bonding networks. The minimized structures were then converted to PDBQT format. Structures for LRRK2 (PDBID: 8TXZ) and α-synuclein (PDBID: 7YNL) were also retrieved and prepared accordingly. The active site for docking was explicitly defined for each target protein based on the location of the native co-crystallized ligand and key catalytic residues. For MAO‑B (2V5Z), the binding site encompasses the FAD cofactor and the surrounding aromatic cage residues (Tyr398, Tyr435). For COMT (3A7E), the site includes the catalytic Mg²⁺ ion, the SAM cofactor, and the substrate-binding residues (Lys144, Asp141, Tyr200). For LRRK2 (8TXZ), the ATP-binding site was defined by the bound inhibitor in the crystal structure. For α‑synuclein (7YNL), a region implicated in early oligomerization (residues 36–55) was targeted based on prior structural studies. Docking was performed using DockFace (in-house software)³⁷. A uniform grid box of 40 × 40 × 40 Å was centered on the centroid of the native co-crystallized ligand in each target protein. This dimension was selected to be significantly larger than the volume of any ligand in our screening library, ensuring comprehensive sampling of the binding site and its surrounding pockets. The exhaustiveness was set to 100 for all docking runs to ensure adequate conformational sampling. The docking protocol was validated by performing self-docking (re-docking of the native ligand), and the result was reported with a root mean square deviation (RMSD) value < 2 Å (Fig. 2). Additionally, the clinical inhibitors selegiline (MAO‑B) and entacapone (COMT) were docked as reference compounds to benchmark the scoring function. DockFace employs a hybrid scoring function that combines empirical energy terms (van der Waals, hydrogen bonding, electrostatic, and desolvation contributions) with a knowledge-based potential derived from statistical analysis of protein-ligand complex structures in the PDB. The docking search is performed using a Lamarckian Genetic Algorithm (LGA), which explores ligand conformational flexibility, rotational freedom, and translational orientation within the defined grid box. Pose Generation and Selection: For each ligand, DockFace was configured to generate 50 independent docking poses. All poses were ranked primarily according to the DockFace scoring function (a hybrid empirical and knowledge-based score approximating binding affinity). The top-ranked pose (the conformation with the most favorable score) for each ligand was selected for subsequent binding energy analysis and interaction studies. This approach ensures that the reported binding energy and interaction profile represent the most stable predicted binding mode for each compound.

Fig. 1.

General steroidal alkaloid scaffold found in Fritillaria species, illustrating the characteristic C27 cholestane backbone and the tetracyclic A–D ring system that forms the basis of major alkaloid subclasses evaluated in this study.

Fig. 2.

Schematic representation of the crystallographic native ligand (red) and the redocked ligand (blue) for: (a) PDB 2V5Z, (b) PDB 3A7E, (c) PDB 8TXZ, and (d) PDB 7YNL.

To ensure reliability, the protocol was rigorously validated through self-docking (re-docking) of the native co-crystallized ligand into each target protein’s active site. The root-mean-square deviation (RMSD) between the docked pose and the original crystallographic pose was below 2.0 Å for all targets (MAO-B: 1.2 Å; COMT: 1.5 Å; LRRK2: 1.8 Å; α-synuclein: 1.3 Å), confirming satisfactory geometric reproducibility. Additionally, we docked two clinically approved inhibitors, selegiline (MAO-B) and entacapone (COMT) and obtained binding energies consistent with their known potencies and with results commonly reported using established software such as AutoDock Vina in the literature. This two-tier validation supports the robustness of DockFace for the comparative virtual screening performed in this study. The Discovery Studio Client 2016 was applied to visualize the interactions and binding poses of all selected compounds.

Molecular dynamics simulation

Further assessment was conducted on the interactions between the top-ranked compounds and the least active compound against human Catechol O-methyltransferase (COMT) (PDB ID: 3A7E) and human Amine oxidase [flavin-containing] B (MAOB) (PDB ID: 2V5Z) within a dynamic context. MD simulation was accomplished employing the Desmond³⁸. The output systems obtained from the docking calculations were used for the MD simulation. The ligand-protein complex was placed within an orthorhombic box, and the transferable intermolecular potential with a three-point (TIP3P) solvent model was used for the system preparation³⁹. The establishment of equilibrium inside the system was attained by introducing sodium and chloride ions at a concentration of 0.15 M, subsequent to the use of Schrödinger’s System Setup⁴⁰. The simulation was performed for 250 nanoseconds with the default relaxation protocol and NPT ensemble framework. During the simulation, the NPT ensemble preserved a constant number of atoms, alongside stable pressure and temperature. The system’s temperature was accurately set at 310.15 K (37 °C) using the Nose-Hoover method, while the pressure was sustained at 1 atm by isotropic scaling. The OPLS force field was used for the simulation of the systems³⁸.

ADMET profile

A comprehensive in silico evaluation was conducted to assess key pharmacological characteristics of the selected steroidal alkaloid ligands, their co-crystal structures, and reference FDA-approved drug compounds. The results were obtained from SwissADME and the preADMET online servers (http://preadmet.bmdrc.org/).

DFT studies

Density functional theory (DFT) analysis was conducted for compounds 5, 42, and 65 using the Gaussian 09 program with the 6–31 + G**(d, p) basis set⁴¹. This analysis covered the energy calculations of frontier highest occupied molecular orbital (HOMOs) and lowest unoccupied molecular orbital (LUMOs), Electrostatic surface potential (ESP), thermochemical parameters including total energy (E_total), enthalpy (Hº), Gibbs free energy (Gº), and Entropy (S), as well as chemical reactivity descriptors including hardness (η), softness (σ), electron affinity (A)⁴². The thermochemical parameters were determined at 298 K using the B3LYP/6–31 + G** level of theory, applying zero-point energy corrections to E, H^°, and G^°, according to the following equation. 

1

2

3

4

5

6

7

8

Here, k_B is the Boltzmann constant, and the subscripts t, r, v, and e correspond to the translational, rotational, vibrational, and electronic contributions, respectively.

The 6–31 + G(d, p) basis set was chosen as a balanced compromise between computational accuracy and feasibility for our set of steroidal alkaloids. This basis set provides a reasonable description of electron polarization and includes diffuse functions, which are important for modeling non-covalent interactions and electron density distributions relevant to chemical reactivity. However, we acknowledge that for large, flexible molecular systems, more complete basis sets (e.g., cc-pVTZ) would provide more rigorous energetic descriptions. Our primary aim in this comparative study was to obtain consistent trends in reactivity descriptors across compounds under identical theoretical conditions.

Results and discussion

Virtual screening and molecular docking studies

In this study, molecular cross-docking analyses were conducted to evaluate the binding affinities and inhibitory potential of 67 plant-derived alkaloid compounds (Table 1s) against key enzymes involved in dopamine metabolism related to Parkinson’s disease (PD), specifically monoamine oxidase B (MAO-B) and catechol-O-methyltransferase (COMT), and also, in α-synuclein protein aggregation. High-resolution protein structures for MAO-B, COMT, LRRK2, and α-synuclein were selected from the RCSB Protein Data Bank, prioritizing structures with the best available resolution and the presence of co-crystallized ligands or suitable binding site definitions. The selected structures were MAO-B (PDB: 2V5Z, 1.6 Å), COMT (PDB: 3A7E, 2.0 Å), LRRK2 (PDB: 8TXZ, 2.8 Å), and α-synuclein (PDB: 7YNL, 2.6 Å). Docking accuracy was validated through self-docking with root-mean-square deviation (RMSD) values below 2 Å, ensuring reliable geometric ligand pose prediction. Figure 2 illustrates the positioning of the native ligand after the self-docking process (blue) compared to the crystallographic ligand of the enzyme (red). Based on the alignment comparison and an RMSD below 2 Å, it can be concluded that the ligand’s placement in the enzyme’s active site is accurate and suitable for subsequent stages of the study.

Initial screening revealed that none of the tested alkaloids surpassed the docking performance of internal ligands for LRRK2 and α-synuclein, resulting in their exclusion from further consideration as likely targets (Table 2s). This suggests that these compounds likely exhibit reduced therapeutic efficacy in inhibiting or modulating these targets. This screening phase narrowed down the number of candidate enzymes for subsequent analysis, focusing on the most promising targets. In contrast, multiple compounds demonstrated more favorable binding energies compared to reference internal ligands for MAO-B and COMT, two well-established pharmacological targets in PD. To validate the docking protocol and ensure result reliability, we conducted cross-docking simulations with two clinically approved reference inhibitors: Selegiline (MAO-B inhibitor) and Entacapone (COMT inhibitor), alongside the steroidal alkaloids and native ligands. The internal ligands of MAO-B and COMT exhibited binding energies of − 9.4 kcal/mol and − 6.1 kcal/mol, respectively. Notably, Compound 65 showed superior MAO-B inhibition (− 11.0 kcal/mol) compared to the reference drug Selegiline (− 7.3 kcal/mol). For COMT, Compound 5 demonstrated the strongest binding (− 9.0 kcal/mol), outperforming Entacapone (− 6.6 kcal/mol) (Table 1).

Table 1.

Binding energies (ΔG, kcal/mol) of the selected key steroidal alkaloids and reference inhibitors against the crystal structures of MAO-B (PDB: 2V5Z), COMT (PDB: 3BWM), LRRK2 (PDB: 4YZI), and α-synuclein (PDB: 1XQ8). Full docking results for all 67 compounds are provided in Supplementary Table 2s.

Compound	MAO-B (PDB 2V5Z)	3A7E/COMT	8TXZ/LRKK2	7YNL/α-synuclein	Notes
65	− 11.0	− 7.3	− 8.4	− 5.8	Best MAO-B inhibitor
42	− 10.6	− 8.4	− 8.3	− 6.3	Dual-target activity
5	− 8.1	− 9.0	− 9.6	− 6.3	Best COMT inhibitor
36	− 7.7	− 6.5	− 7.3	− 5.7	Least active representative
Selegiline	− 7.3	–			Reference MAO-B inhibitor
Entacapone		6.6			Reference COMT inhibitor

It is important to note that docking scores are primarily predictive indicators of binding affinity; actual inhibitory potency (IC₅₀/K_i) must be confirmed through experimental in vitro assays. Detailed analysis of ligand-protein interactions within the MAO-B active site revealed several promising compounds. The interaction of the best (65 and 42), the least potent³⁶, and selegiline, as a reference drug, is shown in Fig. 3. Notably, compound 65 exhibited the most potent binding affinity (ΔGbinding = − 11 kcal/mol), facilitated by an extensive network of hydrophobic, Pi-Alkyl, Pi-Pi stacking, and hydrogen-bonding interactions with critical active site residues such as Cys172, Leu171, Tyr398, and Tyr435. Compound 65 forms a hydrogen bond with CYS172 (a key residue in the MAOB active site), π-π stacking with TYR435, and π-alkyl interactions with LEU171/ILE198, aligning with known MAOB pharmacophores as seen in rasagiline’s binding mode. Compound 42 stabilizes via hydrogen bonds with ASN116/THR478 and π-cation interaction with GLU483, mirroring the electrostatic features of safinamide. Compound 36 shows weaker binding, limited to ARG100 (H-bond) and LEU88 (π-alkyl), lacking interactions with catalytic residues like FAD or TYR 435. Selegiline primarily engages TYR435/PHE343 (π-alkyl) but lacks strong hydrogen bonds, explaining its moderate potency. Overall, Compounds 65 and 42 surpass selegiline by targeting CYS172, TYR435, and GLU483 residues critical for MAOB inhibition.

Fig. 3.

The binding interaction of compounds (65, 42, 36, and selegiline) in the active site of the MAO-B enzyme.

Similarly, for COMT, our docking results revealed significant differences in binding energetics and interaction patterns (Table 1). Compound 5 exhibited the strongest binding (− 9.0 kcal/mol), forming critical interactions with Tyr200 and Lys45 (Fig. 4). These residues are strategically positioned near the catalytic Mg²⁺ ion, suggesting Compound 5 may directly influence the enzyme’s active site. The binding superiority of Compound 5 arises from three key factors: simultaneous interactions with both Tyr200 (hydrogen bonding/π-stacking) and Lys45 (electrostatic) create a stable binding configuration, engagement of 12 residues (including Phe197 and Leu198) provides substantial van der Waals stabilization, and finally, the ligand’s structure effectively fills the hydrophobic pocket while maintaining polar contacts. Notably, Compound 5’s interaction profile differs significantly from that of entacapone, which exhibits fewer stabilizing contacts despite binding to similar residues. This suggests that the spatial orientation and additional hydrophobic interactions in Compound 5 contribute to its enhanced affinity. Compound 42 displayed intermediate binding (− 8.4 kcal/mol), primarily through Trp143 (π–π stacking) and Lys144 interactions. While these are favorable, the absence of key catalytic site interactions likely accounts for its reduced affinity compared to Compound 5.

Fig. 4.

The binding interaction of compounds (5, 42, 25, and Entacapone) in the active site of the COMT enzyme.

The MAO-B enzyme relies on a FAD cofactor to mediate oxidative deamination. Key catalytic residues include Tyr435 and Tyr398, which form the characteristic “aromatic cage” responsible for substrate orientation and stabilization of amine-containing ligands during catalysis^8,10. Cys172, positioned near the flavin ring, has been implicated in influencing the redox environment of the active site. In our study, the superior predicted binding of compound 65 is mechanistically supported by its π–π stacking interaction with Tyr435 and hydrogen bonding with Cys172. Engagement of these residues suggests a binding mode capable of interfering with substrate access to the FAD cofactor. Compound 42 also interacts with Glu483 via a π–cation interaction, a residue involved in substrate recognition, which may alter the electrostatic landscape of the catalytic pocket. In contrast, the weaker binder compound 36 and the reference inhibitor selegiline lack these key interactions, consistent with their lower predicted affinities. COMT catalyzes O-methylation of catechol substrates using Mg²⁺ as a catalytic cofactor and S-adenosyl-L-methionine (SAM) as the methyl donor. The mechanism involves deprotonation of the catechol hydroxyl group by Lys144, facilitated by Mg²⁺, followed by methyl transfer from SAM. In our docking results, compound 5 forms critical interactions with Tyr200 and Lys45. Tyr200 stabilizes the catechol substrate, whereas Lys45 is involved in coordinating the catalytic Mg²⁺ ion. The binding pose of compound 5 therefore suggests that it may directly compete with the substrate and perturb the Mg²⁺ coordination sphere. Compound 42, although interacting with Trp143 and Lys144, does not engage the Mg²⁺-coordinating network to the same extent, which may explain its comparatively lower predicted affinity relative to compound 5. The catalytic mechanism of COMT critically depends on the Mg²⁺ ion coordinated within the active site, which polarizes the catechol substrate for methylation. Although the Mg²⁺ ion is not explicitly visualized in the interaction diagram (Fig. 4), it was retained throughout all docking and MD simulations. The predicted strong binding of compound 5 is consistent with its interactions near the Mg²⁺ coordination region (Tyr200, Lys45), suggesting a potential competitive inhibition mechanism through spatial overlap with the substrate-binding and metal-coordination environment.

Collectively, the findings underscore the selective inhibitory potential of these alkaloids towards MAO-B and COMT over LRRK2 and α-synuclein, congruent with their biochemical roles in dopaminergic neurotransmission and PD pathology. The superior docking scores relative to standard inhibitors suggest these compounds merit further exploration as novel therapeutic candidates. The bulky, multi-ring steroidal frameworks enriched with hydroxyl, methyl, and nitrogenous groups appear to facilitate stable and potent interactions within the enzymes’ active sites, highlighting possible directions for future structure-based optimization. Overall, the tested compounds displayed higher binding affinities than standard inhibitors, suggesting they may serve as preliminary computational hits for further evaluation.

Molecular dynamics simulation results

A molecular dynamics simulation was run for 250 ns to evaluate the ligand-protein interactions of the most and least favorable compounds for COMT and MAOB. The compounds 5 and 25 were selected for COMT, and the compounds 65 and 36 were selected for MAOB. The RMSD values of the proteins throughout the simulations are illustrated in Fig. 5.

Fig. 5.

The RMSD values of the COMT and MAOB throughout the MD simulations for ligand-protein complexes of 5-COMT, 25-COMT, 65-MAOB, and 36-MAOB. These complexes are converged at approximately 1.5 Å, 1.5 Å, 2 Å, and 2.5 Å, respectively, during the simulation.

After the simulation, the proteins exhibited a convergence in the fluctuations of RMSD. For COMT, the RMSD was converged at about 1.5Å, demonstrating system stability. For MAOB, the RMSD values for compounds 65 and 36 were converged at about 2Å and 2.5Å, respectively. As can be seen, the protein MAOB has shown more stability when interacting with the compound 65 when compared to the compound 36. It shows the compound 65 constructs more stable interactions with MAOB residues. Although COMT is more stable at the beginning of the simulation time when interacting with the compound 5 than when interacting with the compound 25, the stability of the systems is almost equal for both compounds 5 and 25 at the end of the simulations. These findings suggest that during the interactions with the target proteins, the compounds 5 and 65 are more stable than their pair compounds.

Figure 6 shows the ligand-protein interactions between MAOB and the compounds 65 and 36 during a 250 ns. Gln206, His115, Pro104, and Thr201 are the residues with the highest interaction fraction for the compound 65 (Fig. 6A). For compound 36, the residues with high interaction fraction are His115, Pro102, and Leu164 (Fig. 6B). The interactions shown by Gln206 and His115, notably the hydrogen bond, are the most significant among the interactions seen in the simulation of the compound 65. As can be seen, the total number of created interactions between the compound 65 and MAOB is larger than those created by the compound 36, which shows the compound 65 is more effective than the compound 36 against MAOB.

Fig. 6.

The interactions constructed between the ligands (A: 65, B: 36) and MAOB during the MD simulation.

Figure 7 shows the interactions formed between compounds 5 and 25 and COMT. The interactions between the compound 5 and COMT throughout the simulation are depicted graphically in Fig. 7A. The distribution of secondary structural elements (SSE) in the 65-MAOB and 5-COMT systems is shown in Fig. 8. The provided illustration depicts the stability of beta-strands, indicated by the blue segments, and alpha-helices, denoted by the red segments, throughout the duration of the simulation. This observation corroborates the reliability of the results by affirming that the simulation effectively preserves the structural integrity of the protein.

Fig. 7.

The interactions constructed between the ligands (A: 5, B: 25) and COMT during the MD simulation.

Fig. 8.

The percent variation of SSE for MAOB (A) and COMT (B) during the MD simulation in the presence of the compounds 65 and 5, respectively. Alpha helices are represented by red regions, and beta strands are represented by blue regions.

The results of the binding free energy calculation are presented in Fig. 9. As can be seen, the lowest energies belong to C5 and C25 when interacting with COMT. The average values of free energy for C5 and C25 are about − 130 kcal/mol and − 150 kcal/mol, respectively. This value for C36 and C65 when interacting with MAOB is about − 20 kcal/mol.

Fig. 9.

The results of binding free energy calculations for the compounds C5, C25, C36, and C65 when interacting with the target proteins COMT and MAOB.

Pharmacokinetic profile

The compounds exhibit molecular weights ranging from 413 to 594 g/mol, approaching the 500 g/mol threshold of Lipinski’s Rule of Five, which may impact their permeability. Log values vary significantly from 1.37 (compound 16) to 4.68 (compound 34), reflecting distinct lipophilicity profiles. Notably, compound 34 demonstrates the highest lipophilicity, suggesting enhanced membrane permeability, while compound 12 (Log P = 2.15) shows greater hydrophilicity, potentially favoring aqueous solubility. Topological polar surface area (TPSA) ranges from 40.54 Ų (compound 5) to 143.08 Ų (compound 16), with lower TPSA values (compounds 5 and 34) correlating with improved passive diffusion across biological membranes. In contrast, the high TPSA of compound 16 (143.08 Ų) may limit its permeability despite its moderate Log P. The number of donors (HBD) varies from 1 to 6, and acceptors (HBA) range from 3 to 9. Compound 12, with the highest counts (5 HBD, 8 HBA), likely exhibits strong target binding but reduced membrane penetration due to its polar nature. Rotatable bonds (0–3) also influence bioavailability, where rigid scaffolds (e.g., compounds 5 and 34, with zero rotatable bonds) often demonstrate superior metabolic stability compared to more flexible analogs (compound 12) (Table 2).

Table 2.

Physicochemical properties of selected steroid alkaloids predicted by SwissADME.

Molecule	Molecular Weighta (g/mol)	Num. Rotatable bondsf	Num. H-bond acceptorsd	Num. H-bond donorsc	TPSAe (Å2)	Log Po/wb
5	413.64	0	3	1	40.54	4.53
12	577.79	3	8	5	122.85	2.15
16	593.79	3	9	6	143.08	1.37
34	415.65	0	3	2	43.70	4.68
35	463.65	0	6	5	104.39	2.18
42	421.57	0	4	2	60.77	4.06
65	427.31	1	3	5	58.56	2.97

The study of seven steroidal alkaloids revealed important pharmacokinetic insights and structure-activity relationships. Compounds 34 and 35 showed excellent blood-brain barrier (BBB) penetration (ratios of 7.44 and 6.98) due to moderate lipophilicity and low polar surface area, making them promising for CNS-targeted therapies. Although all compounds exhibited good predicted intestinal absorption (> 68%), permeability varied, likely influenced by passive diffusion and active transport mechanisms. All analogs were predicted as P-glycoprotein inhibitors, impacting bioavailability and brain exposure. Most compounds had high plasma protein binding (> 85%), potentially reducing free drug levels, except compound 16, with lower binding but poor BBB penetration. Compound 35 emerged as the best lead candidate, combining strong BBB penetration, high absorption, and good permeability, though its high plasma protein binding suggests a need for further optimization (Table 3).

Table 3.

Pharmacokinetic ADME properties predicted by PreADMET.

Molecule	BBBa (C.brain/C.blood)	Caco2b (nm/sec)	HIAc  %	MDCK (nm/sec)	P-gp inhibitione	PPBf %
5	5.6	37.81	95.95	0.069	Inhibitor	89.75
12	0.38	20.54	81.41	0.046	Inhibitor	74.00
16	0.11	19.51	68.66	0.049	Inhibitor	49.72
34	7.44	32.86	93.75	0.07	Inhibitor	88.88
35	6.98	35.98	98.0	0.059	Inhibitor	87.90
42	1.26	22.86	94.16	7.06	Inhibitor	73.90
65	2.45	38.90	95.76	8.9	Inhibitor	89.8

From www.preadmet.qsarhub.com, ^ain vivo blood-brain barrier penetration (BBB),^bin vitro Caco2 cell permeability (Human colorectal carcinoma),^cHuman intestinal absorption (HIA), in vitro P-glycoprotein inhibition, ^fin vitro plasma protein binding (PPB).

The ADMET properties reported here are derived from in silico predictive models (SwissADME, preADMET). These tools provide valuable preliminary estimates of pharmacokinetic behavior based on quantitative structure-activity relationship (QSAR) algorithms. However, it is crucial to acknowledge that such predictions are inherently probabilistic rather than definitive. They serve to prioritize compounds for further experimental testing (e.g., in vitro permeability assays, metabolic stability tests, and in vivo pharmacokinetic studies) and should not be regarded as conclusive evidence of a compound’s actual biological behavior.

DFT analysis

Density Functional Theory (DFT) calculations are utilized to investigate the stability and chemical behavior of the compounds. The frontier orbitals, specifically the Highest Occupied Molecular Orbital (HOMO) and the Lowest Unoccupied Molecular Orbital (LUMO), for compounds 5, 42, and 65 were determined using Gaussian 09 software with the B3LYP functional and 6–31 + G(d, p) basis set, as shown in Fig. 10. The energy gap between the HOMO and LUMO, calculated as the difference between the energies of the LUMO and the HOMO, is a crucial parameter for predicting the chemical reactivity of compounds^43,44. A smaller energy gap results in increased electronic reactivity of the compound, which can enhance its interactions with the desired targets through greater electron transfer, redox activity, or improved binding affinity. The energy gaps of compounds 5, 42, and 65 were 4.775, 4.321, and 3.973 eV, respectively. Compound 5 exhibited the highest energy gap, suggesting weaker interactions with receptors, as confirmed by the docking results. In contrast, compounds 42 and 65 showed smaller energy gaps, indicating higher electronic reactivity. This increased reactivity likely facilitates more efficient interactions with the target receptors MAO-B and COMT.

Fig. 10.

The HOMO, LUMO, and energy gap for compounds 5, 42, and 65 at the B3LYP/6–31 + G (d, p) level of theory.

The Electrostatic Surface Potential (ESP) maps of compounds 5, 42, and 65 are shown in Fig. 11. Nucleophilic regions, such as carbonyl and hydroxyl functional groups, are marked in red, while electrophilic areas, specifically nitrogen atoms, are shown in blue. This representation supports the identification and comprehension of the reactive regions of the molecules^45,46.

Fig. 11.

Electrostatic surface potential (ESP) maps of compounds 5, 42, and 65.

Thermodynamic parameters such as total energy, enthalpy, Gibbs free energy, and entropy for compounds 5, 42, and 65 were calculated using the B3LYP/6–31 + G(d, p) level of theory and are presented in Table 4. According to the thermodynamic parameters, compound 5 exhibits lower total energy and, consequently greater stability compared to compounds 42 and 65. These findings are consistent with the HOMO-LUMO analysis. The hardness (η) and softness (σ) of the compounds 5, 42, and 65 were illustrated in Table 4. The higher hardness of compound 5 indicates superior kinetic stability, suggesting that it is less likely to undergo non-specific interactions. In contrast, the greater softness of compounds 42 and 65 leads to enhanced reactivity, which may facilitates more favorable interactions of these compounds with the active sites of the target receptors.

Table 4.

Thermodynamic parameters of compounds 5, 42, and 65 calculated at the B3LYP/6–31 + G(d, p) level of theory.

Entry	Etota	Ha	Ga	Sb	ɳc	σd	Ac
5	− 1332.81	− 1332.80	− 1332.89	183.290	2.377	0.420	0.58
42	− 1299.56	− 1299.56	− 1299.23	182.36	2.175	0.45	0.46
65	− 1258.92	− 1258.92	− 1258.00	175.299	1.98	0.50	0.95

^a in Hartree/particle. ^b in cal/mol K. ^c in ev. ^d in ev⁻¹.

Discussion

The present study employed a comprehensive in silico strategy to screen steroidal alkaloids from Fritillaria species for their potential as multi-target inhibitors against key proteins implicated in PD pathogenesis. Our results highlight compounds 65 and 5 as particularly promising candidates for MAO-B and COMT inhibition, respectively, with superior binding affinities compared to clinically used reference compounds. Furthermore, compound 42 emerged as a dual-target inhibitor, underscoring the potential of steroidal alkaloid scaffolds in designing multi-target-directed ligands (MTDLs) for complex neurodegenerative disorders.

The preferential inhibition of MAO-B and COMT over LRRK2 and α-synuclein aligns with the known biochemical roles of these enzymes in dopamine catabolism. MAO-B inhibitors, such as rasagiline and safinamide, are well-established in PD therapy for their ability to elevate synaptic dopamine and possibly exert neuroprotective effects by reducing oxidative stress and α-synuclein aggregation [7, 10]. Our findings that compounds 65 and 42 interact with critical residues (Cys172, Tyr435, Glu483) mirror the binding modes of these drugs, suggesting a similar mechanistic basis. Recent studies have emphasized the importance of targeting multiple pathways in PD. For instance, dual MAO-B/COMT inhibition could provide a more sustained dopaminergic effect by simultaneously preventing dopamine degradation via both major enzymatic pathways, potentially reducing motor fluctuations associated with long-term levodopa therapy^10,33.

The failure of the screened steroidal alkaloids to surpass native ligands for LRRK2 and α-synuclein is instructive. While LRRK2 kinase inhibitors are a major focus in current PD drug discovery due to their potential disease-modifying effects³⁴, our results suggest that the steroidal alkaloid scaffold in its current form may not be optimal for targeting the ATP-binding site of LRRK2. Similarly, direct inhibition of α-synuclein aggregation remains a formidable challenge, often requiring molecules designed to disrupt protein-protein interfaces or stabilize non-toxic oligomers³¹. Future work could involve scaffold modification or fragment-based design to incorporate pharmacophores known to interact with these challenging targets.

Our ADMET predictions reveal a crucial structure-property relationship. The most promising compounds (5, 34, 35, 65) consistently exhibited favorable LogP values (< 5), low TPSA (< 70 Ų), and minimal rotatable bonds properties strongly associated with efficient blood-brain barrier (BBB) penetration, a non-negotiable requirement for PD therapeutics. This computational observation is supported by recent studies on CNS drug design, which stress the importance of molecular rigidity and controlled lipophilicity⁴⁶. However, the predicted high plasma protein binding (PPB) for most candidates, while common for lipophilic natural products, warrants caution. High PPB can limit free drug concentration at the target site, potentially necessitating structural optimization to improve unbound fraction without compromising permeability.

The DFT analyses provided a quantum-mechanical rationale for the observed binding trends. The smaller HOMO-LUMO gap and higher softness (σ) of compounds 42 and 65 correlate with their higher predicted reactivity and stronger binding affinities. These electronic properties likely facilitate more favorable interactions, such as charge transfer or stronger van der Waals contacts, within the enzyme active sites. This integrative approach, combining docking, dynamics, and quantum chemistry, strengthens the validity of our virtual screening hits.

Future perspectives

Although the current in silico results are promising, they represent only an initial step toward drug development. The next essential phase involves in vitro validation, where the top-ranked compounds (5, 42, and 65) should be tested using standard MAO-B and COMT inhibition assays and evaluated for neuroprotective activity in dopaminergic cell models such as SH-SY5Y.

Subsequently, medicinal chemistry optimization is required to refine their pharmacological profiles. Priorities include improving BBB penetration and PPB balance for compound 5, enhancing the dual-target efficacy of compound 42, and designing hybrid analogs that integrate the key pharmacophores of compounds 65 and 5.

Given the multifactorial nature of Parkinson’s disease, future studies should also explore target expansion, including AChE, adenosine A₂A receptors, and the NLRP3 inflammasome, to assess the polypharmacological potential of these steroidal alkaloids.

Finally, the most promising candidates should progress to in vivo pharmacokinetic and efficacy studies using established PD models (MPTP or 6-OHDA) to evaluate behavioral effects, biomarker modulation, and safety.

Conclusion

In this integrated in silico study, steroidal alkaloids from Fritillaria species were identified as promising computational scaffolds with potential relevance to MAO-B and COMT interactions. Molecular docking, molecular dynamics simulations, and supporting ADMET and DFT analyses consistently highlighted compounds 65 and 5 as the most favorable predicted binders for MAO-B and COMT, respectively, while compound 42 showed dual-target engagement. These results, however, represent predictive computational insights only, and do not constitute evidence of biological potency or therapeutic efficacy. The predicted physicochemical and ADMET profiles, including estimated BBB permeability, indicate that several of these steroidal alkaloids exhibit characteristics desirable for CNS-active agents; however, these findings are based solely on in silico predictions and require rigorous experimental validation. The inability of the tested alkaloids to outperform native ligands for LRRK2 and α-synuclein also provides mechanistic insight, indicating that this scaffold may be more suited to enzymatic targets than to protein-aggregation or kinase-related mechanisms. Importantly, all conclusions drawn in this work are based solely on computational modeling. Experimental validation is essential as a next step, beginning with in vitro enzyme inhibition assays and followed by structure–activity optimization guided by SAR and ADMET findings. Subsequent in vivo pharmacokinetic and pharmacodynamic studies in Parkinson’s disease models will be required to evaluate biological relevance and safety. Overall, this study provides a set of prioritized computational hits and a methodological framework to support future discovery efforts aimed at multi-target neuroprotective agents from natural products.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

M,.A.F: Conceptualization, Supervision, Project administration, Sh.H.: Running and interpretation of Docking section. S.S: Running and interpretation of DFT section. M.M: Analysis and interpretation of data. A.P.: Running and interpretation of M.D. section. H.S.: Funding acquisition and Supervision,. L.E. conceptualization, Resources, Supervision, Writing - review & editing. All authors reviewed and approved the final manuscript.

Funding

Financial assistance from the Shiraz University of Medical Sciences, grant numbers (Grant No.: 31897) is gratefully acknowledged.

Data availability

The data sets used and analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.