Bioinformatics Screening of Phenylpropanoids from Pyrostegia venusta in ER+ Breast Cancer ^†

Abstract

This study investigated the cytotoxic, antiproliferative, and molecular interaction profiles of the phenylpropanoids verbascoside and isoverbascoside from Pyrostegia venusta using in silico approaches. Computational predictions suggested differential cytotoxicity trends between tumor and non-tumor breast cell models compared with tamoxifen. QSAR analyses indicated antiproliferative potential, while docking studies revealed stable ligand–protein interactions with estrogen-related targets and PTEN. ADMET predictions suggested favorable metabolic characteristics, including limited CYP3A4 interaction. Overall, these results provide predictive insights that support further experimental investigation of these phenylpropanoids in ER⁺ breast cancer models.

1. Introduction

Breast cancer remains one of the leading causes of morbidity and mortality among women worldwide, representing a major global public health challenge [1]. In Brazil, it was estimated that by the end of the 2023–2025 triennium approximately 704,000 new cancer cases per year would be diagnosed, excluding non-melanoma skin cancer, of which around 244,000 would affect women. Female breast cancer remains the most prevalent cancer type in the South and Southeast regions [2].

Drug resistance and the adverse effects associated with conventional therapies, such as chemotherapy and hormone therapy, have driven the search for safer and more effective therapeutic strategies [3]. In this context, natural compounds derived from medicinal plants have emerged as promising sources of bioactive molecules with selective cytotoxic potential and modulatory effects on molecular pathways associated with carcinogenesis [4]. Among the various classes of pharmacologically relevant secondary metabolites, phenylpropanoids stand out due to their antioxidant, anti-inflammatory, and antiproliferative properties [5].

The species Pyrostegia venusta (commonly known as flame flower or “cipó-de-são-joão”), widely used in Brazilian traditional medicine, is rich in phenolic compounds and exhibits recognized antioxidant and antimicrobial activities [6]. The abundance of phenylpropanoids in this species, including verbascoside [7] and isoverbascoside [6], suggests a chemopreventive and antitumor potential that remains largely unexplored in breast cancer models.

The identification and in silico characterization of these compounds represent an essential initial approach to understanding their mechanisms of action and possible molecular interactions with relevant therapeutic targets [8]. Accordingly, the present study aimed to evaluate the cytotoxic and antitumor potential of verbascoside and isoverbascoside from Pyrostegia venusta against breast cancer cell lines using in silico methodologies. Based on their chemical structure and previously reported bioactive properties, it is hypothesized that verbascoside and isoverbascoside interact favorably with molecular targets involved in breast cancer progression, resulting in predicted cytotoxic and antitumor activity in breast cancer cell lines as assessed by in silico analyses.

2. Materials and Methods

2.1. Selection of the Plant and Compounds of Interest

The cytotoxic potential and molecular mechanisms of action of the phenylpropanoids verbascoside (Figure 1A) and isoverbascoside (Figure 1B) [9], isolated from Pyrostegia venusta (Figure 1C), were evaluated using in silico methods. The selection of this plant species was based on the availability of specimens cultivated in the Medicinal Garden of the State University of Minas Gerais (UEMG), Academic Unit of Passos, Passos, Brazil, which enables the continuity of experimental studies and the optimization of cultivation conditions aimed at obtaining higher concentrations of bioactive compounds.

2.2. Prediction of Cytotoxicity and Antitumor Activity in Breast Cancer

The cytotoxic potential of the compounds was evaluated using in silico methodologies. Initially, the CLC-Pred software 2.0, Moscow, Russia “https://way2drug.com/clc-pred/ (accessed on 11 July 2025)” [10] was employed to estimate cytotoxicity against the non-tumorigenic epithelial cell line MCF-10A, used as a reference model for assessing selectivity. Predictions were expressed as pIC₅₀ values, according to the software’s validation criteria. Subsequently, the antitumor potential was investigated using the BC CLC-Pred 1.0 platform, Moscow, Russia “https://way2drug.com/bc/ (accessed on 11 July 2025)” [11], which predicts activity against breast cancer cell lines, notably MCF-7 (mammary gland adenocarcinoma) and T-47D (mammary gland ductal carcinoma), both estrogen receptor–positive, expressed as pIC₅₀ and pIG₅₀. All predictions were interpreted within the applicability domain of the models, ensuring that the evaluated compounds fall within the chemical space covered by the training datasets, thereby increasing the reliability of the predictions. The evaluated compounds included the phenylpropanoids isolated from Pyrostegia venusta, verbascoside (Compound CID: 5281800) and isoverbascoside (Compound CID: 6476333), along with the reference drug tamoxifen (Compound CID: 2733526), all retrieved from the PubChem database Maryland, United States (USA) “https://pubchem.ncbi.nlm.nih.gov/ (accessed on 11 July 2025)”.

2.3. Identification of Interactions with the Estrogen Receptor and CYP3A4

The molecular interactions of the compounds with the estrogen receptor (ER) and the CYP3A4 enzyme were evaluated using the ADMETlab 2.0 platform Zhejiang, China “https://admetmesh.scbdd.com/ (accessed on 11 July 2025)” [12]. For these classification endpoints, the platform outputs qualitative probability classes rather than absolute numerical affinity values. Specifically, the prediction probabilities generated by the underlying machine-learning models are internally mapped into six categorical symbols: 0–0.1 (---), 0.1–0.3 (--), 0.3–0.5 (-), 0.5–0.7 (+), 0.7–0.9 (++), and 0.9–1.0 (+++).

These symbols therefore represent probability ranges associated with the predicted likelihood of interaction or classification outcome, and should be interpreted as qualitative indicators of confidence rather than quantitative binding scores. Consequently, the ER and CYP3A4 outputs allow comparative qualitative assessment of interaction tendencies and metabolic behavior, but do not provide direct numerical measures suitable for statistical comparison. The predictive reliability of the models is supported by their reported performance metrics, including area under the ROC curve, accuracy, Matthews correlation coefficient, specificity, and sensitivity, as provided by the ADMETlab 2.0 validation framework [12].

2.4. Molecular Docking

The three-dimensional structure of the protein was obtained from the Protein Data Bank (PDB) “https://www.rcsb.org/ (accessed on 11 July 2025)” under the accession code PDB ID: 1D5R. The binding to the tumor suppressor protein PTEN (Phosphatase and Tensin Homolog) was evaluated using in silico molecular docking methodology through the CB-Dock2 platform Chengdu, China “https://cadd.labshare.cn/cb-dock2/ (accessed on 11 July 2025)” [13]. Tamoxifen was used as the reference compound for model validation and comparison of binding affinity. The selection of PTEN [14] as a molecular target was based on its crucial role in regulating the PI3K/AKT/mTOR signaling pathway, as PTEN suppression or loss-of-function leads to constitutive activation of this pathway, promoting uncontrolled cell proliferation, enhanced survival signaling, metabolic reprogramming, and resistance to apoptosis, hallmarks frequently observed in breast cancer progression. Additionally, PTEN presents a well-characterized phosphatase domain [15] with a defined active site and available high-resolution crystallographic structures, which support its suitability for structure-based molecular docking analyses. Docking results were expressed as binding free energy (kcal.mol⁻¹).

3. Results

The phenylpropanoids (Figure 2) exhibited lower predicted cytotoxicity toward the non-tumorigenic MCF-10A cell line compared to tamoxifen, with pIC₅₀ values of 5.76 (verbascoside) and 5.66 (isoverbascoside), while the reference drug presented a pIC₅₀ of 5.10. Although these differences are numerically small, they suggest a trend toward reduced cytotoxicity in non-tumor cells. However, given the in silico nature of the predictions and the absence of associated statistical dispersion metrics, these differences should be interpreted as indicative rather than biologically definitive.

In the antitumor activity assessment (Figure 3A), the pIC₅₀ values in MCF-7 cells were similar between the phenylpropanoids (pIC₅₀ = 5.20 and 5.16), whereas tamoxifen showed higher cytotoxic potency (pIC₅₀ = 5.35). In T47D cells, the phenylpropanoids exhibited lower activity (pIC₅₀ = 4.28 and 4.17), while tamoxifen maintained a stronger cytotoxic effect (pIC₅₀ = 5.35).

Regarding antiproliferative potential (Figure 3B), the QSAR models for MCF-7 cells predicted higher pIG₅₀ values for the phenylpropanoids (5.76 and 5.85) compared to tamoxifen (pIG₅₀ = 5.48), indicating greater anticipated antiproliferative activity. In T47D cells, verbascoside reached the highest predicted value (pIG₅₀ = 6.03), surpassing both isoverbascoside (pIG₅₀ = 5.41) and tamoxifen (pIG₅₀ = 5.44). According to the platform’s criteria, values above 6 are considered highly relevant.

The in silico predictions indicated low interaction of the phenylpropanoid isoverbascoside in the NR-ER and NR-ER-LBD assays (

Table 1. Determination of integration between phenylpropanoid compounds of Pyrostegia venusta against endocrine receptors and modulation of P450 isoenzymes (CYP3A4).

Test	Verbascoside	Isoverbascoside	Tamoxifen
NR-ER	0.3–0.5	0.5–0.7	0.9–1.0
NR-ER-LBD	0.1–0.3	0.3–0.5	0–0.1
Substrate CYP3A4	0–0.1	0.7–0.9	0.9–1.0
Inhibition CYP3A4	0–0.1	0–0.1	0.7–0.9

), whereas tamoxifen demonstrated strong estrogen receptor engagement, as expected. Regarding metabolic parameters, both verbascoside and isoverbascoside were not predicted to inhibit CYP3A4, in contrast to tamoxifen, which showed a higher likelihood of acting as a substrate and a moderate inhibitor of this isoenzyme.

Molecular docking analysis (Figure 4A) indicated a higher predicted affinity for isoverbascoside, with a binding energy of −8.4 kcal.mol⁻¹, surpassing tamoxifen (−7.5 kcal.mol⁻¹) and suggesting a more stable interaction with the modeled target. The docking pose (Figure 4B) revealed a favorable binding mode within the protein’s active site, characterized by multiple non-covalent interactions. The complex was primarily stabilized by six conventional hydrogen bonds involving residues GLN149, ARG172, LYS163, ASN329, ASP324, and ARG173, highlighting the essential role of polar groups and glycosidic moieties in site recognition.

Additionally, aromatic interactions contributed significantly to anchoring the phenolic ring, including a T-shaped π–π interaction with TYR176 and a strong π–anion interaction with ASP153. Van der Waals forces involving PRO169, LEU320, and other nearby residues further supported the precise accommodation of the ligand within the hydrophobic pocket. Together, these findings confirm that the ligand binds in a stable and specific manner, engaging both polar and aromatic regions of the active site.

4. Discussion

The in silico analyses conducted in this study indicate that the phenylpropanoids verbascoside and isoverbascoside exhibit predicted trends of cytotoxic selectivity when compared with tamoxifen, particularly in the non-tumorigenic MCF-10A cell line. Although these findings are derived exclusively from computational predictions and should not be interpreted as direct evidence of an improved safety profile, they highlight the value of in silico approaches as early-stage screening tools for identifying compounds with differential activity toward tumor and non-tumor cells [16]. Such predictions provide a rational basis for prioritizing candidates for subsequent experimental validation.

In this context, the observed selectivity trends are consistent with previous reports describing the antioxidant properties and low toxicity of Pyrostegia venusta extracts [6,8]. Moreover, the higher predicted antiproliferative activity of verbascoside reinforces its relevance as a promising molecular scaffold for ER⁺ breast cancer research. Similar applications of in silico early-screening strategies have been successfully employed to identify natural compounds with selective antitumor potential, supporting the translational relevance of computational methodologies in natural product–based drug discovery [17,18].

Phenylpropanoid-rich extracts have been reported to interact with molecular pathways associated with carcinogenesis, including estrogen-related signaling and tumor suppressor–associated proteins [19]. In this context, the predicted interaction of isoverbascoside with the estrogen receptor is consistent with previous in silico and experimental reports describing the affinity of phenolic compounds toward ER⁺ tumor models. Regarding PTEN, the observed binding affinity suggests a potential structural compatibility between isoverbascoside and the protein; however, molecular docking alone does not imply functional activation, stabilization, or enhancement of PTEN signaling [14]. Any possible influence on the PI3K/AKT/mTOR pathway remains speculative and cannot be inferred without functional assays.

The CYP3A4 profiling revealed distinct metabolic behaviors. Verbascoside showed neither substrate affinity nor inhibitory effect, indicating a low likelihood of CYP3A4-mediated interactions and suggesting a favorable metabolic safety profile. Isoverbascoside displayed moderate substrate potential but no inhibitory activity, while tamoxifen, as expected, exhibited strong substrate behavior and moderate inhibition, consistent with reports of its extensive CYP3A4 metabolism [20]. These findings collectively indicate that verbascoside may present fewer pharmacokinetic interactions than tamoxifen.

The docking results indicate the formation of stable ligand–protein complexes mediated by non-covalent interactions, such as hydrogen bonds, π-interactions, and hydrophobic contacts. These interactions reflect favorable binding geometry rather than direct evidence of biological modulation, in agreement with the limitations inherent to structure-based computational approaches [8,13]. Thus, the stronger predicted affinity of isoverbascoside toward PTEN should be interpreted as a starting point for hypothesis generation rather than proof of tumor-suppressive activity.

Overall, the integrated in silico analyses performed in this study provide a coherent preliminary framework for the evaluation of verbascoside and isoverbascoside as natural phenylpropanoids of interest in ER⁺ breast cancer drug discovery and early-stage screening [12]. The predicted trends of cytotoxic selectivity, receptor interaction profiles, and docking results highlight these compounds as candidates for further investigation [21]. However, the study is limited by its exclusive reliance on computational models, which do not account for biological complexity, do not provide statistical dispersion metrics, and do not imply functional modulation of molecular targets. Consequently, the findings should be interpreted as hypothesis-generating rather than confirmatory, and experimental validation through cell-based and molecular assays is required to establish cytotoxic efficacy, receptor modulation, and any potential involvement of PTEN-related signaling pathways [14].

5. Conclusions

The results indicate that these phenylpropanoids, particularly isoverbascoside, exhibit promising in silico antitumor potential for breast cancer, demonstrating predicted cell selectivity and relevant interactions with hormone receptors and molecular biomarkers. However, these findings are preliminary and based on computational analyses, highlighting the need for further in vitro and in vivo studies to validate their antitumor efficacy and safety profiles.