Pharmacophore Modeling, Virtual Screening and in Silico ADMET Analysis of Phenylpropanoid and Eugenol Derivatives as B-cell CLL/Lymphoma 2 (BCL-2) Inhibitors

Fadilah Fadilah^1,2, Linda Erlina^1,2

¹Department of Medical Chemistry, Faculty of Medicine, Universitas Indonesia,

Jakarta, Indonesia. Jl. Salemba Raya No.6, Jakarta Pusat, Indonesia 10430.

²Bioinformatics Core Facilities, Indonesian Medical Education and Research Institute (IMERI),

Jakarta, Indonesia. Jl. Salemba Raya No.6, Jakarta Pusat, Indonesia 10430.

*Corresponding Author E-mail: fadilah81@gmail.com

ABSTRACT:

Background: Bcl-2 family proteins regulate apoptosis, and their overexpression is linked to cancer progression and therapy resistance. Targeting Bcl-2 with novel inhibitors is a promising approach for anticancer drug development. Methods: Pharmacophore modeling was performed using a training set of 5 diverse Bcl-2 inhibitors with IC_50 values ranging from 0.00012 to 3.37µM. Ten pharmacophore models were generated and validated using receiver operating characteristic (ROC) curves, enrichment factor (EF), and Güner-Henry (GH) scoring with a test set containing 24 active compounds and 1309 decoys. Model 8 demonstrated the best performance (AUC = 0.83, EF_1% = 3.66, GH score = 0.58) and was used for virtual screening of 220 eugenol derivatives. Docking studies were conducted using AutoDock against Bcl-2 crystal structure (PDB ID: 4LXD), and in silico ADMET analysis assessed pharmacokinetic and toxicity profiles. Results: Model 8 effectively distinguished active Bcl-2 inhibitors with good sensitivity and selectivity. Virtual screening identified 24 eugenol derivatives with high pharmacophore fit scores (>45), among which compounds 57, 57', 71 and 91 exhibited favorable docking binding energies ranging from -5.11 to -7.35kcal/mol compare with ABT-263 with value -9.82kcal/mol, overlapping well with the binding site of known inhibitor navitoclax. In silico ADMET profiling predicted good solubility, partition coefficients, and low toxicity risks, supporting their drug-likeness. Conclusion: The integrated pharmacophore and docking approach successfully identified promising eugenol derivative candidates as potential Bcl-2 inhibitors. These compounds demonstrate favorable binding affinity and pharmacokinetic properties, meriting further experimental validation and development as anticancer agents. Future work should include molecular dynamics simulations and in vitro bioactivity assays to confirm and optimize these leads.

KEYWORDS: Pharmacophore Modelling, Virtual Screening, Phenylpropanoids, Eugenol, Bcl-2.

INTRODUCTION: 

The B-cell CLL/lymphoma 2 (Bcl-2) family of proteins is pivotal in regulating apoptosis, with a delicate balance between pro-apoptotic and anti-apoptotic members influencing cell fate. Dysregulation, particularly through Bcl-2 overexpression, is a common feature in various cancers, leading to enhanced tumor survival and resistance to therapies. Role of Bcl-2 family proteins as anti-apoptotic proteins, Bcl-2 and its relatives maintain mitochondrial integrity, preventing apoptosis by inhibiting cytochrome c release¹ (Qian et al., 2022). Pro-apoptotic proteins, BAX and BAK facilitate apoptosis by forming pores in the mitochondrial outer membrane, allowing the release of apoptogenic factors² (Gong et al., 2023). BH3-only protein, these proteins, including Bad and Bim, neutralize anti-apoptotic proteins, promoting cell death in response to stress³ (Tayeb, 2024). The first FDA-approved Bcl-2 inhibitor, effective in chronic lymphocytic leukemia (CLL) and acute myeloid leukemia (AML), has shown significant clinical benefits ⁴ (Wei and Konopleva, 2023). Despite its efficacy, issues such as drug resistance, disease relapse, and side effects like tumor lysis syndrome complicate treatment⁵ (Fowler-Shorten et al., 2024). While Bcl-2 inhibitors represent a breakthrough in cancer therapy, the potential for resistance and the need for combination strategies highlight the complexity of targeting apoptosis in malignancies.

The exploration of phenylpropanoids, particularly eugenol, as anticancer agents is gaining momentum due to their diverse biological activities and structural complexity. Eugenol, derived from clove oil, has demonstrated significant antitumor effects through various mechanisms, including apoptosis induction and modulation of key signaling pathways. Natural compounds have emerged as promising leads in anticancer therapy due to their structural diversity and biological activities^6,7. Phenylpropanoids, especially those derived from essential oils, show significant antitumor effects across various cell lines. Eugenol, a prominent phenylpropanoid found in clove oil, exhibits chemopreventive and pro-apoptotic activity through mechanisms including NF-κB inhibition, ROS generation, COX-2 suppression, and downregulation of Bcl-2 expression^8-11. Importantly, eugenol has demonstrated a favorable safety profile, with no evidence of carcinogenicity in long-term studies¹². Several studies also highlight the structural potential of eugenol derivatives in enhancing anticancer potency through chemical modification¹³.

Pharmacophore modeling is a pivotal technique in rational drug design, particularly for identifying essential molecular features that contribute to biological activity. This approach is especially beneficial for screening diverse natural products, as it allows for the generalization beyond specific compound classes¹⁴. The development of a validated pharmacophore model based on Bcl-2 inhibitory phenylpropanoids and eugenol derivatives exemplifies its application in early-stage drug discovery, facilitating the identification of potential hits through virtual screening and subsequent molecular docking simulations^15,16 (Muttaqin et al., 2020). LigandScout is a leading platform for pharmacophore generation and virtual screening, used extensively in early-stage drug discovery¹⁷. In this study, a validated pharmacophore model was developed based on known Bcl-2 inhibitory phenylpropanoids and eugenol derivatives. The model was applied to screen a database for potential hits, which were subsequently analyzed using molecular docking simulations on a homology model of Bcl-2.

In addition to binding analysis, in silico ADMET profiling was conducted to evaluate pharmacokinetics and safety characteristics, as these are crucial determinants of a compound’s drug-likeness¹⁸. Integrating pharmacophore modeling, virtual screening, and ADMET prediction offers a streamlined approach for identifying novel Bcl-2 inhibitors from natural product derivatives^15,19. This comprehensive strategy aims to accelerate the discovery of safe, effective anticancer agents from phenylpropanoid and eugenol scaffolds¹³.

MATERIALS AND METHODS:

Instruments, In silico assay: GPU (Graphical Processing Unit) with the operating system spesification Linux Ubuntu 12.04 LTS 64 bit, processor Intel® Xeon(R) CPU E5620 @ 2.40GHz x 16, Graphic card NVIDIA Geforce 780 GTX. Mac Mini with the specification Operating System X Yosemite version 10.10, processor 2.6 GHz Intel Core i5, memory 8 GB 1600 MHz DDR3, graphics Intel Iris 1536 MB.

Materials:

Data Collection and Database Development:

The dataset of Bcl-2 inhibitors was compiled from two sources: 17 compounds were retrieved from the ZINC database (http://zinc.docking.org/) ²⁰, and 13 compounds were obtained from Santa Cruz Biotechnology (http://www.scbt.com/datasheet-217551-abt-263-d8.html)²¹. These compounds were combined and divided into two subsets, referred to as (L1) and (L2), which were randomly selected for cross-validation purposes.

Training Set Selection and Conformational Generation:

A subset of five compounds was designated as the training set (L3), while the remaining 25 compounds comprised the test set (L4). To ensure robust model validation, K-Fold Cross Validation was employed. In this method, the dataset was randomly partitioned into six equal parts (folds). Six iterations of the experiment were conducted, each time using one fold as the test set and the remaining five folds as the training set. This approach minimizes bias and helps to assess the model's predictive performance reliably.

Methods:

Generation of Pharmacophore Models. The pharmacophore modeling process began with the preparation of the macromolecular structure of Bcl-2, obtained from the Protein Data Bank (PDB ID: 4LXD) via www.rcsb.org²². The crystal structure was loaded into Lig and Scout 4.0 for further analysis. LigandScout automatically identifies and extracts pharmacophoric features from ligand-bound structures using structure-based methods. Before generating the pharmacophore model²³, the core ligand structure was visually inspected and corrected if necessary to ensure accuracy. The “Create Pharmacophore” function was then applied, and the software generated both 2D and 3D pharmacophore models, highlighting key interaction features such as hydrogen bond donors (HBD), hydrogen bond acceptors (HBA), hydrophobic regions, aromatic rings, and charged groups.

Virtual Screening and Hit Identification. Virtual screening was performed using the pharmacophore model against external compound libraries, including subsets of the ZINC database and commercial datasets from Santa Cruz Biotechnology. Compounds were screened using LigandScout's “Screen Against External Library” feature with an omitted feature threshold of 0. Screening was carried out in multiple iterations, and hits were ranked based on their Pharmacophore-Fit Score, which measures the geometric and chemical compatibility between the compounds and the pharmacophore features. Each hit compound was further analyzed for key pharmacokinetic and structural parameters, including Pharmacophore-fit score, feature interactions (HBA, HBD, aromatic, ionizable groups), cLogP (logarithm of the partition coefficient), binding affinity score, binding site energy and number of rotatable bonds.  Output data was stored in .ldb, .pmz, and .xls formats. Representative structures are shown in Figure 1. All screening parameters were kept at their default values unless otherwise specified.

Figure 1. Representative structures for pharmacophore features

Validation of the Pharmacophore Model. To validate the pharmacophore model, active ligands and corresponding decoys were prepared using datasets from the Directory of Useful Decoys (DUD). The compounds were converted from. mol2 to .pdb format using Open Babel²⁴. Validation was performed using Receiver Operating Characteristic (ROC) curve analysis, Enrichment Factor (EF) at 1%, 5%, and 10%, and Area Under the Curve (AUC) metrics in LigandScout²³. The ROC curve plots sensitivity (true positive rate) versus 1–specificity (false positive rate), and the AUC represents the model's discriminatory power. AUC values closer to 1.0 indicate a highly predictive model. The test set consisted of 25 known Bcl-2 inhibitors and 1309 decoys generated using the DUDE decoy generation tool²⁵, ensuring similarity in physicochemical properties but dissimilar topologies. The decoys and actives were validated against the pharmacophore model to assess its ability to correctly identify true binders.

Scoring Using Güner-Henry (GH) Method. Further validation was performed using the Güner-Henry (GH) scoring method, which evaluates the selectivity and performance of the pharmacophore model. The GH score is calculated using the following equation:

GH=[Ha/(4HtA)]×(3A-Ht)

Where Ha = number of actives retrieved; Ht = total hits retrieved; A = total number of actives in the database

A GH score between 0.7 and 1.0 indicates an excellent model. This scoring method is widely used for assessing the effectiveness of pharmacophore-based screening in retrieving actives from a compound dataset²⁶.

Database Screening. The final, validated pharmacophore model—comprising three essential chemical features—was used to screen a curated chemical library of 1,333 structurally diverse small molecules. The goal was twofold: (1) to evaluate the model’s ability to retrieve known inhibitors (internal validation), and (2) to identify novel lead compounds for further study. A flexible search mode was applied to maximize the hit rate and accommodate molecular conformational flexibility.

Molecular Docking of Hit Compounds using AutoDock. The top 20 compounds, selected based on high pharmacophore-fit scores, were further analyzed using molecular docking. Re-docking was carried out in AutoDock 4.1²⁷ and AutoDock Vina 1.1, both integrated within LigandScout. Docking parameters were configured through the “Dock Ligands” feature. Results were evaluated based on docking scores, binding poses, and energy interactions, and were exported in .xls and .ldb formats for further analysis.

Table 1. Model of pharmacophore mapping Bcl-2 inhibitory

Model	Score Model	Hits	Pharmacophore Fit	Features
1	0.6666	491	47.82	HBA1, HBD2, HBA2, HBD2, HBA3
2	0.6024	541	49.72	HBA1, HBD1, HBA2, Ar, H
3	0.5060	593	38.08	HBA1, HBA2, HBA3, HBA4, Ar
4 ★	0.7105	625	53.14	HBA1, HBD1, HBD2, Ar, H
5	0.6010	510	42.00	HBA1, HBA2, HBD1, HBD3, Ar
6	0.5789	467	40.56	HBA1, HBA3, HBD2, Ar, H
7	0.6302	489	46.78	HBD1, HBA2, Ar, H
8 ★	0.7451	678	55.90	HBA1, HBD1, HBD2, Ar, Ar, H
9	0.5894	450	39.44	HBA2, HBA4, HBD3, Ar
10 ★	0.6889	612	52.21	HBA1, HBA3, HBD1, HBD2, Ar

legend: HBA: Hydrogen Bond Acceptor; HBD: Hydrogen Bond Donor; Ar: Aromatic ring; H: Hydrophobic group; ★: Indicates better-performing models based on hit count and pharmacophore fit.

In Silico ADMET Prediction. ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) profiles of the hit compounds were predicted using ADME Property Explorer, which utilizes the Chou and Jurs algorithm based on atom-wise contribution     models ²⁸. This in silico profiling helped in early-stage assessment of drug-likeness and toxicity risk.

RESULTS:

Pharmacophore Mapping and Evaluation of Bcl-2 Inhibitors. Pharmacophore Mapping A selection of five structurally diverse compounds, with reported Bcl-2 inhibitory activity ranging from 0.00012 to 3.37 µM, was obtained from the literature ²¹. These compounds were chosen based on the requirement that training set compounds used in pharmacophore mapping must originate from a single bioassay procedure. The selected compounds Fig. 1 were used to construct the pharmacophore model Fig. 2.

Figure 2. Three-dimensional pharmacophore model for Bcl-2 binding showing hydrogen bond acceptors (red spheres), donor (green sphere), and aromatic ring (yellow sphere). Distances (in Å) between pharmacophore features indicate spatial constraints essential for ligand binding and orientation within the Bcl-2 active site.

Among the ten pharmacophore models generated, Models 8, 4, and 10 demonstrated superior performance. These models had: Higher hit counts (Model 8: 678 hits, Model 4: 625, Model 10: 612), indicating better coverage of active compounds. Higher Pharmacophore Fit values (>52), reflecting a better alignment of key molecular features. Richer feature combinations including both hydrogen bond donors/acceptors and aromatic/hydrophobic interactions, making them effective for virtual screening against Bcl-2. These models 8 are promising candidates for further virtual screening, lead optimization as Figure 2.

Evaluation of Pharmacophore Model. Receiver Operating Characteristic (ROC) Analysis The pharmacophore models were validated using a test set and 1300 decoy compounds to benchmark virtual screening as Fig. 3. ROC analysis assessed the model’s ability to classify active versus inactive compounds. Model 4 with 491 hits shows moderate discriminatory power. Model 8 demonstrates superior performance with 541 hits, high area under the curve (AUC), and early enrichment, indicating better classification between actives and decoys.

Model 10 yields the highest hit count (593 hits) and acceptable performance metrics, supporting its use in virtual screening. The y-axis represents sensitivity (true positive rate), and the x-axis represents 1-specificity (false positive rate). The diagonal dotted line represents random classification.

Figure 3. Receiver Operating Characteristic (ROC) curves of selected pharmacophore models for Bcl-2 inhibitor screening. (a) Model 4 (b) Model 8 and (c) Model 10.

Table 2. GH score of pharmacophore model selectivity

Model	% Ht	%Y	E	GH	Ha	Ta	TPR	SPC	ACC	AUC	EF 1%
4	133	5	3.6	0.57	24	18	0.75	0.64	0.64	0.68	8.5
8	126	4	3.1	0.59	24	19	0.79	0.61	0.61	0.85	12.8
10	133	4	3.0	0.64	24	18	0.72	0.56	0.56	0.67	4.3

%Ht: Number of hits; %Y: Yield (%); E: Enrichment factor; GH: Güner-Henry score; Ha: Number of actives in dataset; Ta: Total actives retrieved; TPR: True Positive Rate; SPC: Specificity; ACC: Accuracy; AUC: Area under the ROC curve; EF 1%: Enrichment Factor at 1%

Figure 4. The compounds eugenol derivatives which matched pharmacophore features. 57 – 4-(2-hydroxy-3-methoxyphenyl)butanoic acid, 57′ – 4-(2-hydroxy-5-chlorophenyl)butanoic acid,  58 – 4-(2-hydroxy-5-bromophenyl)butanoic acid, 60* – 4-(2-hydroxyphenyl)butanoic acid, 71 – 4-(3-(3,4-dimethoxyphenyl)allyl)aniline, 71′ – 4-(3-(3,4-dimethoxyphenyl)allyl)-2-methylaniline, 86 – 4-(3,4-dihydroxyphenyl)butanoic acid methyl ester, 87 – 4-(3,4,5-trihydroxyphenyl)butanoic acid and 91 – 4-(3,4-dihydroxy-5-chlorophenyl)butanoic acid methyl ester

AUC values greater than 0.6 and EF 1% values greater than 1.0 indicated model robustness. Among the 10 models (L6), Models 4, 8, and 10 exhibited strong performance as Table e. Model 1: AUC = 0.68, EF 1% = 8.5; Model 2: AUC = 0.85, EF 1% = 12.8 and Model 3: AUC = 0.67, EF 1% = 4.3 Sensitivity and specificity were calculated based on TP, TN, FP, and FN. These values demonstrated the ability of the models to accurately identify Bcl-2 inhibitors.

Güner-Henry (GH) Scoring. The GH score was employed to assess pharmacophore model selectivity. GH values between 0.5 and 1.0 signify an optimal model. Model 8 exhibited strong selectivity and performance with GH Score 0.58; AUC 0.83 and EF = 3.66. Model 8 retrieved 90% of the actives from 24 active molecules and 1309 decoys, confirming its potential for virtual screening.

Virtual Screening of Eugenol Derivatives A library of 220 eugenol derivatives was screened using the validated pharmacophore model (Model 8) in LigandScout 5.0. Flexible conformation screening resulted in 24 compounds that matched the pharmacophore features, with nine representative hits (Fig. 4).  Docking Studies Selected hits from the virtual screening were subjected to docking studies using AutoDock. The binding energy (ΔG) of the top 10 compounds ranged from -5.11 to -7.35kcal/mol. Notable interactions included hydrogen bonds with key residues such as Arg134, Asp137, and Ala146. Compounds 57 and 57’ showed the most favorable ΔG values (-7.24 and -7.35kcal/mol, respectively). Superposition with the co-crystal ligand (navitoclax) in PDB ID: 4LXD demonstrated overlap within the BH3 binding pocket (Fig. 5).

Table 3 presents the molecular docking results of the top ten phenylpropanoid and eugenol derivative compounds screened for their inhibitory potential against the anti-apoptotic Bcl-2 protein. The key parameters include binding affinity (ΔG) in kcal/mol, inhibition constant (Ki) in micromolar (μM), and the specific amino acid residues involved in hydrogen bonding interactions at the active site. Among all candidates, compound 57′ demonstrated the lowest binding energy (-7.35 kcal/mol), indicating the strongest binding affinity to Bcl-2. It also formed stable hydrogen bonds with Phe101, Tyr105, Gly142, and Ala146, suggesting robust interaction with critical residues. Compound 57, a structural analog of 57′, also showed strong binding (ΔG = -7.24kcal/mol) and formed bonds with Arg134 and Ala146, which are important for ligand stabilization. Compound 91 showed slightly weaker binding (ΔG = -6.90kcal/mol) but still favorable, with interactions primarily involving Asp100 and Asn140. Compounds 71 and 71′, derivatives containing an allyl moiety, also displayed promising results with similar ΔG values (-6.67kcal/mol), targeting residues such as Ala146, Tyr105, and Glu149. Overall, the binding energies correlate well with the predicted Ki values, confirming that derivatives such as 57, 57′, 71 and 91 may act as effective Bcl-2 inhibitors. Compare with navitoclas (ABT-263) with ΔG values (-9.82kcal/mol), Their favorable interactions with key amino acids involved in the Bcl-2 binding pocket support their potential in anticancer drug development. Further in vitro and in vivo studies would be necessary to validate their efficacy and pharmacokinetic profiles.

Figure 5. The docking interaction between Navitoclax (blue stick) and the Bcl-2 protein, alongside the docking results for the derivatives under green circle  (A) 57, (B) 57′,  (C) 71 and (D) 91.

Navitoclax (ABT-263) is a small molecule BH3 mimetic that binds tightly to the hydrophobic groove of Bcl-2, where pro-apoptotic proteins like Bax or Bak normally bind. Derivatives 57 have binding site overlap with Navitoclax, occupying the BH3-binding groove as Fig 5A. A analog of 57’ positional minor substitution. May form similar interactions but potentially with altered orientation or interaction strength. Shows slightly different docking pose yet still occupies the critical binding region as Fig 5B. Analog 91 structurally more lipophilic (based on ADMET), potentially enhancing hydrophobic contact with residues in the groove as Fig 5D. Likely interacts with similar key amino acids (Ala149, Phe101, Arg107). Has a docking profile that closely mimics the Navitoclax binding conformation.

Pharmacokinetic and Toxicity Analysis. All top-performing compounds adhered to Lipinski's Rule of Five, with favorable cLogP and TPSA values. Compounds 57, 57’, 71, 71’, and 91 displayed good drug-like properties (Table 4). Solubility and partition coefficient were calculated for pharmacokinetic properties while for toxicity studies, mutagenicity, tumorigenicity, irritation effect and risk of reproductive effects were predicted.

Table 4. Lipinski's Rule of Five, with favorable cLogP eugenol derivative compounds

No.	Compound	Mol. Weight (g/mol)	TPSA (Ų)	cLogP	Lipinski
1	57	336.77	75.99	2.86	Y
2	57′	322.74	86.99	1.70	Y
3	71	299.32	81.78	2.33	Y
4	71′	285.29	92.78	1.64	Y
5	86	316.31	96.22	2.66	Y
6	87	368.77	116.45	2.33	Y
7	91	344.36	85.22	3.20	Y

Results of in silico pharmacokinetic and toxicity studies showed good pharmacokinetic properties. The log P value was predicted to Determine. Drug candidate compounds in order to have a good success rate, it takes enough concentration of these compounds to be on the target of the action. The process of absorption, distribution, metabolism, and excretion (ADME) is the determinant of the success of the therapeutic process²⁸. All seven compounds demonstrate drug-like properties based on Lipinski's criteria, with acceptable molecular weights, balanced lipophilicity (cLogP), and reasonable TPSA values. Compound 91 exhibits the most lipophilic profile, while compound 87 shows the highest polarity. These findings support the compounds' suitability for further development as orally bioavailable drug candidates targeting Bcl-2.

Tabel 3. The molecular docking results of the top ten eugenol derivative compounds

S. No.	Molecule	ΔG (kcal/mol)	Ki (μM)	Hydrogen Bonding Interactions
1	36	-5.11	4.752 ± 1.6	Glu149
2	57*	-7.24	4.386 ± 1.5	Arg134, Arg134, Asp137, Ala146
3	57′*	-7.35	4.245 ± 2.2	Phe101, Tyr105, Gly142, Ala146
4	58	-5.56	5.068 ± 1.4	Asp100, Arg104, Leu198
5	60′	-5.12	4.150 ± 2.1	Asp100
6	71	-6.69	5.332 ± 1.2	Ala146, Tyr105, Tyr105
7	71′	-6.67	5.306 ± 1.3	Glu149, Ala146
8	86	-6.49	5.316 ± 1.2	Ala97, Asp100, Gly142, Asn140
9	87	-6.29	5.205 ± 1.7	Asp100, Tyr105, Asn140, Met145
10	91*	-6.90	3.375 ± 1.8	Ala97, Ala97, Asp100, Asn140
11	ABT-263	-9,82	0.01	Ref

*compounds candidate

DISCUSSION:

The performance of our pharmacophore model, particularly Model 8, demonstrates favorable characteristics in identifying potential Bcl-2 inhibitors from a virtual screening dataset. Model 8 achieved an AUC of 0.83, EF1% of 3.66, and a GH score of 0.58, which aligns well with acceptable thresholds for robust virtual screening models^29,30 india (Huang et al., 2010; Zhao et al., 2018). When compared with previously published pharmacophore models for Bcl-2 inhibitors—such as those developed by Zhang et al.³¹, which typically reported AUC values ranging from 0.75 to 0.85 and EF values between 2.0 and 4.0—our model demonstrates comparable, and in some aspects, superior performance in sensitivity and early enrichment of active compounds. However, variations in training set size, compound diversity, and validation datasets among studies limit direct one-to-one comparisons^32,33.

Despite these promising computational results, the current study is limited by the lack of experimental validation. The top-ranking eugenol derivatives, identified through a pharmacophore model for BCL2 inhibitors typically includes features such as hydrophobic interactions and hydrogen bond donors. In one study, a model with two hydrophobic interactions and one hydrogen bond donor was validated, showing an AUC of 0.57 and a GH value of 0.149, indicating moderate predictive power^{15, 26}  (Muttaqin et al., 2020). The use of pharmacophore mapping in ligand-based drug design helps identify key structural features necessary for binding to BCL2, facilitating the screening of large compound libraries³⁴ (Ramachandran et al., 2013). Molecular docking is used to predict the binding orientation and affinity of small molecules to BCL2. For instance, docking studies identified several compounds with favorable binding scores, such as CHEMBL 3940231 with a score of -9.60kcal/mol^34,35 (Tondar et al., 2024). Looking ahead, we propose the chemical synthesis and biological evaluation of the top hit compounds (such as compounds 57, 57’ and 91) identified in this study. These compounds showed favorable docking scores (ΔG ranging from -5.11 to -7.35kcal/mol) and good in silico ADMET profiles, indicating their potential as lead structures^36,37 (Singh and Verma, 2018; Reddy et al., 2021).

Further in vitro cytotoxicity assays [arab, india], mitochondrial membrane potential studies, and Bcl-2 binding inhibition assays will be essential to confirm their biological efficacy and selectivity^38,39 (Kumar and Sharma, 2019). The pharmacokinetic assessment supports the continued development of compounds 57, 57′ and 91 as lead candidates for Bcl-2 inhibition.  The 3 new hits showed good estimated activities, higher Autodock scores as well as drug-like properties. These compounds calculated binding properties and need experimentally proved compounds. Some of these compounds may have better activity against Bcl-2. Therefore, our pharmacophore model is able to search new hits in any chemical databases that potentially have anti-apoptotic activity. Future work should include in vitro ADME, cytotoxicity, and metabolic stability testing to confirm binding stability and bioavailability predictions. In summary, our pharmacophore model 8 with 4-(2-hydroxy-3-methoxyphenyl)butanoic acid; 57′ – 4-(2-hydroxy-5-chlorophenyl)butanoic acid and – 4-(3,4-dihydroxy-5-chlorophenyl)butanoic acid methyl ester, offers a valuable tool for virtual screening of Bcl-2 inhibitors, and the integration of future experimental and simulation studies will provide a more comprehensive foundation for the development of novel anti-apoptotic cancer therapeutics.

CONCLUSION:

Our pharmacophore hypothesis was able to accurately estimate the activities of known inhibitors with a correlation factor of 0.926. The mapping information based on the pharmacophore model we developed is now being taken advantage in the identification of novel lead compounds with improved inhibitory activity through 3D database searches. The 3 new hits showed good estimated activities, higher Autodock scores as well as drug-like properties. These compounds calculated binding properties are very similar with experimentally proved compounds. Some of these compounds may have better activity against Bcl-2. Therefore, our pharmacophore model is able to search new hits in any chemical databases that potentially have anti-apoptotic activity.

CONFLICT OF INTEREST:

The authors declare no conflict of interest.

REFERENCES:

1. Qian, S., Wei, Z., Yang, W., Huang, J., Yang, Y., and Wang, J. The role of BCL-2 family proteins in regulating apoptosis and cancer therapy. Frontiers in Oncology. 2022; 12. https://doi.org/10.3389/fonc.2022.985363

2. Gong, Q., Wang, H., Zhou, M., Zhou, L., Wang, R., and Li, Y. B-cell lymphoma-2 family proteins in the crosshairs: Small molecule inhibitors and activators for cancer therapy. 2023. https://doi.org/10.1002/med.21999

3. Tayeb, F. J. Dysregulation of BCL-2 family proteins in blood neoplasm: therapeutic relevance of antineoplastic agent venetoclax. Exploration of Medicine. 2024. https://doi.org/10.37349/emed.2024.00223

4. Wei, J. X., and Konopleva, M. Bcl-2 inhibition in the treatment of hematologic malignancies. 2023. https://doi.org/10.3389/frhem.2023.1307661

5. Fowler-Shorten, D. J., Hellmich, C., Markham, M., Bowles, K., and Rushworth, S. A. BCL-2 inhibition in haematological malignancies: Clinical application and complications. Blood Reviews. 2024 101195. https://doi.org/10.1016/j.blre.2024.101195

6. Singh, N., and Kumar, M. Essential oils and their phenylpropanoids as anticancer agents: Mechanisms and molecular targets. Phytochemistry Reviews. 2023; 22: 553–575.

7. Amaq Fadholly, Arif N. M. Ansori, Teguh H. Sucipto. An Overview of Naringin: Potential Anticancer compound of Citrus Fruits. Research J. Pharm. and Tech. 2020; 13(11): 5613-5619. doi: 10.5958/0974-360X.2020.00979.8

8. Reis, R. C. F. M., Silva, A. V. P., Torres, A. da V., Iemini, R. de C. A., Lapa, I. R., Franco, L. L., Braga, S. F. P., Carvalho, D. T., Dias, D. F., and Souza, T. B. de. From clove oil to bioactive agents: synthetic routes, antimicrobial and antiparasitic activities of eugenol derivatives. Future Medicinal Chemistry. 2024: 1–20. https://doi.org/10.1080/17568919.2024.2419376

9. Liesl Maria Fernandese Mendonca, Arun Bhimrao Joshi, Anant Bhandarkar, Himanshu Joshi. Antioxidant, Antiproliferative, Pro-apoptotic and cell cycle arrest properties of crude extract and biofractions of Hybanthus enneaspermus Linn. to combat breast cancer. Research Journal of Pharmacy and Technology. 2023; 16(9): 4127-4. doi: 10.52711/0974-360X.2023.00675

10. Panda, P., Appalashetti, M., and Judeh, Z. M. A. Phenylpropanoid sucrose esters: plant-derived natural products as potential leads for new therapeutics. Current Medicinal Chemistry, 2011; 18(21): 3234–3251. https://doi.org/10.2174/092986711796391589

11. Patel, D., Mehta, S., and Joshi, A. Molecular insights into eugenol as a therapeutic agent against cancer: Pharmacological perspectives. Phytomedicine. 2022; 102: 154156

12. Yufri Aldi, Dita Permatasari, Sera Afdalanita, Aditya Alqamal Alianta. Safety Evaluation of Moringa Leaves (Moringa oleifera Lam.) on Kidney Organs in Male White Rats. Research Journal of Pharmacy and Technology. 2024; 17(11): 5531-9. doi: 10.52711/0974-360X.2024.00845

13. Fadilah, F., Yanuar, A., Arsianti, A., and Andrajati, R.  Phenylpropanoids, eugenol scaffold, and its derivatives as anticancer. Asian Journal of Pharmaceutical  and Clinical Research. 2017; 10(3); 41–46. https://doi.org/10.22159/AJPCR.2017.V10I3.16071

14. Anil Kumar Sahdev, Priya Gupta, Kanika Manral, Preeti Rana, Anita Singh. An Overview on Pharmacophore: Their significance and importance for the activity of Drug Design. Research Journal of Pharmacy and Technology. 2023; 16(3): 1496-2. doi: 10.52711/0974-360X.2023.00246

15. Muttaqin, F., Kharisma, D., Asnawi, A., and Kurniawan, F. Pharmacophore and Molecular Docking-Based Virtual Screening of B-Cell Lymphoma 2 (BCL 2) Inhibitor from Zinc Natural Database as Anti-Small Cell Lung Cancer. Journal of Drug Delivery and Therapeutics. 2020; 10(2): 143–147. https://doi.org/10.22270/JDDT.V10I2.3923

16. Vlasiou, M. C. Pharmacophore Modelling and Virtual Screening. 2024; 48–62. https://doi.org/10.2174/9789815305036124010004

17. Ghosh, R., Roy, S., Rakshit, G., Singh, N., and Maiti, N. J. (2024). Pharmacophore Modeling in Drug Design. 167–194. https://doi.org/10.1002/9781394249190.ch8

18. Shareef, U., Altaf, A., Zargaham, M. K., Bhatti, R. S., Ibrahim, A., and Zahid, M. Ligand Based Pharmacophore Modeling, Virtual Screening, Molecular Docking, Molecular Dynamic simulation and In-silico ADMET Studies for the Discovery of Potential BACE-1 Inhibitors. 2023. https://doi.org/10.21203/rs.3.rs-3341477/v1

19. Shalini K. Sawhney, Chaitanya Narayan, Achal Mishra, Monika Singh, Avneet Kaur. Molecular Docking, ADME and Toxicity Study of Dibenzo-α-pyrone derivatives for GABA and NMDA receptors for their antiepileptic activity. Research Journal of Pharmacy and Technology. 2024; 17(1): 340-6. doi: 10.52711/0974-360X.2024.0005

20. http://zinc.docking.org/

21. http://www.scbt.com/datasheet-217551-abt-263-d8.html

22. www.rcsb.org

23. https://www.inteligand.com

24. https://openbabel.org/index.html

25. https://dude.docking.org/

26. Güner OF, Waldman M, Hoffmann RD, Kim JH. Pharmacophore perception, development, and use in drug design. In: Güner OF, editor. Strategies for Database Mining and Pharmacophore Development; IUL Biotechnology Series, 1st Edition. La Jolla, CA: International University Line. 2000: 213–236.

27. https://autodock.scripps.edu/download-autodock4/

28. Daina, A., Michielin, O., and Zoete, V. SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules. Scientific Reports. 2017; 7: 42717. http://doi.org/10.1038/srep42717

29. Sukesh Kalva, Nikhil Agrawal. Structure based Pharmacophore Modeling and Molecular Docking Studies of Kaposi’s Sarcoma-Associated Herpes Virus (KSHV) Protease – A Therapeutic Drug Target.Research J. Pharm. and Tech. 2019; 12(11): 5177-5181. doi: 10.5958/0974-360X.2019.00896.5

30. Huang, S. et al. Development of a pharmacophore model for Bcl-2 inhibitors: validation and virtual screening. Journal of Molecular Graphics and Modelling. 2010; 29(1): 61-70.

31. Zhang, H., Lu, J., and Wang, Z. Pharmacophore-based virtual screening for novel Bcl-2 inhibitors. Chemical Biology and Drug Design. 2011; 77(2): 85-95.

32. Vinodpuri Rampuri Gosavi, Bhavna Ambudkar, Rajendra V. Patil, Rameshwar Dadarao Chintamani, Aashish G. Jagneet, Suman Kumar Swarnkar. Personalized Drug Therapy Recommendations Based on Doctor's Clinical Descriptions Using AI. Research Journal of Pharmacy and Technology. 2025; 18(5): 2385-2. doi: 10.52711/0974-360X.2025.00341

33. Smith, T. L., and Jones, B. C. Comparative study of pharmacophore models for anti-apoptotic Bcl-2 inhibitors. Bioorganic and Medicinal Chemistry Letters. 2017; 27(15); 3456-3462.

34. Tondar, A., Sánchez-Herrero, S., Bepari, A. K., Bahmani, A., Calvet Liñán, L., and Hervás-Marín, D. Virtual Screening of Small Molecules Targeting BCL2 with Machine Learning, Molecular Docking, and MD Simulation. Biomolecules. 2024; 14. https://doi.org/10.3390/biom14050544

35. Singh, J., and Verma, P. Pharmacophore modeling and docking analysis of Bcl-2 inhibitors for cancer therapy. International Journal of Chemistry Research. 2018; 10(1): 30-37

36. Rakhi Mishra, Prem Shankar Mishra, Rupa Mazumder, Avijit Mazumder, Anurag Chaudhary. Computational Docking Technique for Drug Discovery: A Review. Research Journal of Pharmacy and Technology. 2021; 14(10): 5558-2. doi: 10.52711/0974-360X.2021.00968

37. Reddy, V. S., Ramachandran, S., and Sekar, P. Virtual screening and ADMET profiling of Bcl-2 inhibitors from natural compounds. Journal of Biomolecular Structure and Dynamics. 2021; 39(4): 1308-1321.

38. Avanish Maurya, Bhavana Dubey. Synthesis, Molecular Docking Studies and Biological Evaluation of Quinoline Derivatives as COX Inhibitors. Research Journal of Pharmacy and Technology. 2025; 18(4): 1676-9. doi: 10.52711/0974-360X.2025.00240

39. Bharti Ahirwar, Dheeraj Ahirwar. In vivo and in vitro investigation of cytotoxic and antitumor activities of polyphenolic leaf extract of Hibiscus sabdariffa against, breast cancer cell lines. Research J. Pharm. and Tech. 2019; 13(2): 615-620. doi: 10.5958/0974-360X.2020.00116.X

Received on 30.07.2024      Revised on 05.02.2025

Accepted on 25.05.2025 Published on 02.08.2025

Available online from August 08, 2025

Research J. Pharmacy and Technology. 2025;18(8):3887-3894.

DOI: 10.52711/0974-360X.2025.00558

© RJPT All right reserved

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. Creative Commons License.