In vitro study and in silico reveal Teucrium Polium L. methanolic extract as an apoptosis inducer and a potential Bcl-2 inhibitor for breast cancer Authors: Monira Alaujan, Sabine Matou-Nasri, Fatimah G. Albani, Sahar S. Alghamdi, Rasha S. Suliman, Rehab AlRoshody, Sarah Huwaizi, Aljawharah Almogren, Mikhlid H. Almutairi, Zeyad Alehaideb Published: 2026 ABSTRACT Therapy of breast cancer (BC) overexpressing the pro-survival protein Bcl-2 remains a major clinical challenge. The discovery of natural anticancer drugs with potential as Bcl-2 inhibitors has generated increasing interest. The medicinal plant Teucrium polium L. (TPL) exhibits pharmacological properties, including anticancer activities. This study aimed to assess the antiproliferative and pro-apoptotic effects of TPL extracts on BC cells and to predict the potential of TPL extract-derived metabolites as Bcl-2 inhibitors. Plant extraction was performed using several solvents, and cell viability was determined. The expression of proteins and genes related to apoptosis was evaluated using Western blot, the Proteome Profiler™ Human Apoptosis Array Kit, and qPCR, respectively. Caspase-3/7 and mitochondrial permeability transition pore opening (mPTPO) activities were visualized. GC/MS analysis, molecular docking, and in silico modeling were used for metabolite identification, prediction of their molecular interactions with Bcl-2, and their pharmacokinetic profiles. Among extracts, TPL methanolic extract (TPLME) dose-dependently reduced hormone-dependent and triple-negative BC cell viability, while sparing normal mammary epithelial cells. TPLME induced apoptosis by increasing caspase-3/7 activity, activating distinct apoptotic pathways, and modulating BAX, TP53, and BCL-2 gene expression. Differential BCL-2 expression levels in TPLME-treated BC cells were confirmed by variations in mPTPO activity. After TPLME-derived metabolite identification, molecular docking revealed interactions between key metabolites and the active site of Bcl-2. Predictive analysis revealed the safe pharmacokinetic profiles of TPLME-derived metabolites. These findings highlight the promising potential of TPLME-derived metabolites for the development of novel Bcl-2 inhibitors for BC cells, requiring further chemical development and preclinical investigations using in vivo BC models.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s44446-026-00106-x. Introduction Breast cancer (BC), the most frequently diagnosed cancer in women worldwide, remains the leading cause of cancer-related mortality, despite significant advances in early detection and treatment strategies (Sharma and Saripilli 2025). Projections indicate that if current trends persist, the global burden of BC will increase to an estimated 6 million new cases and 1.1 million deaths annually by 2050 (Freihat et al. 2025). BC is a heterogeneous disease comprising distinct molecular subtypes that differ in their genetic, molecular, histological, and clinical characteristics, which significantly influence prognosis and treatment responses (Xiong et al. 2025). The classification of BC into molecular subtypes, primarily based on the expression of hormone receptors (i.e., estrogen (ER) and progesterone (PR) receptors) and human epidermal growth factor receptor 2 (HER2), has been essential in personalizing therapy (Carvalho et al. 2025). The major intrinsic subtypes include luminal A (ER+/HER2−), luminal B (ER+/PRlow/HER2±), HER2-enriched, and triple-negative/basal-like BC (Carvalho et al. 2025). Unlike luminal A and B tumors that respond well to endocrine therapies and to a combination of hormone therapy and chemotherapy, respectively, HER2-enriched tumors are typically more aggressive but have become more manageable with HER2-targeted therapies (e.g., trastuzumab) (Mercogliano et al. 2023; Agelidis et al. 2025). Triple-negative breast cancer (TNBC), defined by the absence of ER, PR, and HER2 amplification, represents approximately 15–20% of BC cases and is associated with a poorer prognosis due to its aggressiveness and limited targeted treatment options, including chemotherapy as the mainstay of treatment (Xu et al. 2025). To overcome the side effects or drug resistance resulting from conventional cancer chemotherapy, natural products have attracted considerable interest in the search for novel anti-TNBC compounds (Li et al. 2025). In the discovery of therapeutic drugs, natural products derived from medicinal plants constitute a valuable source of anticancer agents (Newman and Cragg 2020). The exploration of these natural products as anticancer agents has gained renewed interest due to their diverse bioactivities, including their antioxidant, anti-inflammatory, antiproliferative, and pro-apoptotic effects (Sun et al. 2024). Many plant-derived compounds have shown promise due to their ability to target multiple cancer signaling pathways and anti-apoptotic proteins such as B-cell lymphoma (Bcl)−2, often with lower toxicity than synthetic drugs (Xu et al. 2023; Al Khalily et al. 2025). The discovery of Bcl-2 inhibitors, including FDA-approved Venetoclax and plant-derived compounds, represents a novel therapeutic strategy for BC patients and is generating increasing interest in research on treatments for this cancer, particularly to combat drug resistance (Xu et al. 2023). During the apoptosis process, the expression of the mitochondrial protein Bcl-2 is reduced, which allows the oligomerization of Bax and Bak, leading to the pore formation in the outer membrane of the mitochondria and subsequently to the release of the cytochrome c, a key component for the activation of the intrinsic apoptotic pathway (Vogler et al. 2025). Overexpression of Bcl-2 occurs in 85% of ER+ BC and in 41% of TNBC, developing resistance against chemotherapeutic drugs (Tosun and Kaplan 2025). Venetoclax has been reported to act synergistically with endocrine therapy in a preclinical model of ER+ BC (Tiong et al. 2025) or with poly(ADP-ribose) polymerase (PARP) inhibitors on TNBC cells (Almudimeegh et al. 2025). Clinical trials have highlighted the enhanced efficacy of conventional BC treatment (e.g., anti-estrogen therapy, anti-HER2 therapy, CDK4/6 inhibitors) after their combination with Bcl-2 inhibitors (Ploumaki et al. 2023). Natural compounds targeting Bcl-2 encompass a wide range of chemical classes, such as terpenoids, alkaloids, and polyphenols (Abood et al. 2024). Given that combined treatment with Venetoclax causes side effects, further research on these natural products holds promise for the development of novel Bcl-2 inhibitors in solid tumor treatment (Ploumaki et al. 2023; Goldsmith et al. 2020). Teucrium polium L. (T. polium), commonly known as felty germander, is a perennial herbaceous plant of the Lamiaceae family, native to the Mediterranean region and Southwest Asia (Bahramikia et al. 2022). It has a long history of use in traditional medicine to treat ailments, including diabetes, inflammation, and infections (Huseini et al. 2024). Pharmacological studies have reported a broad spectrum of biological effects attributed to T. polium, such as antioxidant, antimicrobial, anti-inflammatory, antidiabetic, and anticancer activities against colon, lung, and ER+ BC (Al-Hamwi et al. 2021; Bsharat et al. 2025). However, research exploring the effects of T. polium extracts on TNBC and their potential as Bcl-2 inhibitors remains limited. Therefore, the present study aimed to investigate the antiproliferative and apoptotic effects of different T. polium extracts on the TNBC MDA-MB-231 cell line, as well as comparative analyses with MCF-7 cells and non-tumorigenic mammary epithelial cells, in order to evaluate both the efficacy and selectivity of the anticancer effects of the T. polium extract. Chemical analysis, molecular docking, and computer-based predictions were performed to identify metabolites derived from the TPL extract and to predict their molecular interactions with Bcl-2 as potential Bcl-2 inhibitors, as well as their pharmacokinetics properties. Materials and methods Reagents All reagents were purchased from Thermo Fisher Scientific (Waltham, MA, USA), unless otherwise specified. Plant material collection and authentication The aerial parts of Teucrium polium (T. polium) L. were collected in the Sedeer area, near Jalajil City, north of Riyadh, Saudi Arabia, during the flowering season. A qualified botanist carried out identification and authentication. A reference specimen was deposited in the herbarium of King Saud University (KSU), Riyadh, under reference number KSU 24645, registered in the author’s name for documentation and future reference (Supplementary Figure S1). The sample was collected on March 15, 2023. Extraction of T. polium L. aerial parts The aerial parts of T. polium L. were carefully rinsed with distilled water to remove debris, then dried in the shade under well-ventilated ambient conditions to preserve the thermolabile bioactive compounds. The dried material was ground into a fine powder using a standard electric grinder and stored in a clean, airtight container until use. Approximately 500 mg of the powdered plant material was extracted with 10 mL of each of six different high-grade analytical solvents—polar protic solvents (i.e., water, ethanol, and methanol), polar aprotic solvents (i.e., ethyl acetate and dichloromethane), and nonpolar chlorocarbone solvent (i.e., chloroform)—in separate preparations. Ultrasound-assisted extraction was carried out using a XUB10 Digital Ultrasonic Bath (Grant Instruments, Shepreth, Cambridgeshire, UK) operating at 50 kHz for 30–40 min at maximum power. The temperature was maintained at 20 °C using a water bath to prevent the degradation of heat-sensitive compounds. After sonication, the mixtures were filtered through ashless quantitative filter paper, and the resulting filtrates were concentrated under reduced pressure using a Thermo Scientific Savant SPD2010 SpeedVac Concentrator. The dried residues were reconstituted in 1 mL of dimethyl sulfoxide (DMSO), vortexed until fully dissolved, and stored at −20 °C until further analysis. Extraction yields varied depending on the solvent used and were calculated as the ratio of the mass of the dried extract to the initial mass of the plant material according to the following equation:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Extraction\, yield\, \left(\%\right)= \frac{Weight \,of \,extraction (Final \,residue)}{Weight\,of \,plant \,material\, \left(Initial \,plant \,material\right) }\times 100$$\end{document}Extractionyield%=Weightofextraction(Finalresidue)WeightofplantmaterialInitialplantmaterial×100 Cell line culture and treatment Triple-negative MDA-MB-231 (#HTB-26), ER+ MCF-7 (#HTB-22) human BC cell lines and non-tumorigenic mammary epithelial cell line MCF-10A (#CRL-10317) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and grown in complete medium consisting of Gibco® Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 1% penicillin (100 IU/mL), and streptomycin (100 µg/mL). Cells were cultured at 37 °C in a humidified incubator with a controlled atmosphere of 5% CO₂ and 95% humidity. The culture medium was refreshed every 2–3 days, and cells were harvested using TrypLE™ Express solution then sub-cultured when they reached 70–80% confluence. For the treatment conditions, the cells were exposed to various concentrations of T. polium L. (TPL) extracts reconstituted in DMSO. The controls included an untreated control group consisting of cells maintained in complete culture medium without any added extract and cells treated with DMSO, the vehicle group. For most of the assays, the cells (5 × 105 cells) were seeded in complete medium in 6-well plates, and the next day they were exposed to TPL extract concentrations for 48 h of incubation. Cell viability assay Cells were seeded at a density of 1,000 cells per well into opaque 96-well Greiner® microplates (Greiner Bio-One, Kremsmünster, Austria) containing complete culture medium. The next day, after cell adhesion, treatments were applied as previously described, including exposure to varying concentrations of TPL extracts, alongside appropriate controls. Each condition was performed in triplicate. Cell viability was subsequently evaluated using the CellTiter-Glo® Luminescent Cell Viability Assay kit (Promega Corporation, Madison, WI) according to the manufacturer’s protocol. This assay measures intracellular adenosine triphosphate (ATP) levels, which serve as a reliable indicator of metabolically active and viable cells. The addition of the assay reagent induces luminescence by enzymatic reaction, which is proportional to the concentration of ATP produced. The emitted luminescent signal was recorded using an EnVision™ microplate reader (Perkin Elmer, Waltham, MA, USA). Visualization of morphological alterations Cells were incubated in the presence or absence of TPL methanolic extract (TPLME) tested 30 µg/mL (optimal concentration), as well as with 0.5% DMSO, used as a vehicle control. After 48 h, random cell fields were photographed using Invitrogen EVOS® cell imaging system. Apoptosis-related protein detection analysis using Western blot and protein arrays Western blotting was employed to assess the expression levels of apoptosis-related proteins, with particular focus on cleaved caspase-7. Untreated and treated cells were washed twice with ice-cold phosphate-buffered saline (PBS) and lysed using 120 μL of NP40 lysis buffer supplemented with cocktails of protease and phosphatase inhibitors. The lysates were left on ice for 30 min, then centrifuged at 14,000 × g for 40 min at 4 °C to remove cell debris. Proteins were then estimated, separated by 11% SDS–polyacrylamide gel electrophoresis, then transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica, MA, USA) using a semi-dry transfer apparatus (Trans-Blot® SD, Bio-Rad) as previously described (Matou-Nasri et al. 2022). PVDF membranes were blocked for 1 h at room temperature in Tris-buffered saline with 0.1% Tween-20 (TBS-T, pH 7.4) containing 1% bovine serum albumin. Primary antibodies, including rabbit monoclonal anti-cleaved caspase-7 (1:1000, clone D6H1, #8438, Cell Signaling Technology, Danvers, MA, USA) and mouse monoclonal anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH, 1:5000, clone 6C5, #ab8245, AbCam, Cambridge, UK), used as a loading control, were diluted in Intercept™ blocking buffer (#927–70001, LI-COR Biosciences, Lincoln, NE) and incubated with the membranes overnight at 4 °C with continuous shaking. After washing, the membranes were incubated for 1 h at room temperature with species-specific secondary antibodies labeled with infrared dyes: IRDye® 680RD goat anti-rabbit (#926–68071) and IRDye® 800RD goat anti-mouse (#926–32210) IgG secondary antibodies (LI-COR Biosciences). After the final washes, fluorescence signals were detected using the LI-COR Odyssey CLx Imaging System (LI-COR Biosciences). Band intensities were analyzed using ImageJ software (http://rsbweb.nih.gov/ij/index.html, NIH, Bethesda, MD, USA). The Proteome Profiler™ Human Apoptosis Array Kit (#ARY009, R&D Systems, Minneapolis, MN, USA) was used to simultaneously assess the expression levels of multiple apoptosis-related proteins in MCF-7 and MDA-MB-231 cell lysates after cell treatment with 30 µg/mL TPLME or 0.5% DMSO for 48 h. The assay was carried out according to the manufacturer’s protocol, enabling a global analysis of apoptosis regulatory protein expression upon experimental treatments. Protein expression was detected using the enhanced chemiluminescence (ECL) method, facilitated by the ECL Chemiluminescent Kit (Bio-Rad). The resulting chemiluminescent signals appeared as black spots against a white membrane background, reflecting the relative abundance of each protein. Signal intensities were captured using a high-resolution imaging system, and densitometric quantification was performed with ImageJ software. Live imaging of apoptotic cellular processes using confocal microscopy To assess apoptosis induction and mitochondrial membrane integrity, MDA-MB-231 and MCF-7 cells were seeded at a density of 80,000 cells per well in Nunc™ Lab-Tek™ 8-well chambered slides. After allowing the cells to adhere overnight, all treatments were administered for 48 h under standard culture conditions. The cells were treated either with TPLME at 30 µg/mL (effective concentration) or with 0.5% DMSO, used as a vehicle control. Each condition was performed in triplicate and subjected to confocal microscopy. Caspase-3 and caspase-7 activities, which are key executioner enzymes of the apoptotic cascade, were detected using the Invitrogen Image-iT® LIVE Red Caspase-3 and Caspase-7 Detection Kit (Molecular Probes Inc., Life Technologies, Carlsbad, CA, USA). Concurrently, mitochondrial outer membrane permeability, indicative of mitochondrial permeability transition pore opening (mPTPO) and hallmark of intrinsic apoptosis, was evaluated using the Image-iT® LIVE Mitochondrial Transition Pore Assay kit (Molecular Probes Inc.). Both assays were conducted according to the manufacturers’ protocols to ensure accurate and reliable detection of apoptotic events at the cellular level using the LSM 780 confocal laser scanning microscope system (Carl Zeiss Microscopy GmbH, Jena, Germany). RNA extraction and reverse transcription-quantitative polymerase chain reaction (RT-qPCR) Untreated or treated MDA-MB-231, MCF-7, and MCF-10A cells (5 × 105 cells per well, 6-well plates) were washed with PBS and harvested using TrypLE™ Express solution. The cells were collected by centrifugation at 15,000 rpm for 5 min, and the resulting pellets were stored at −20 °C until further use. Total RNA was isolated from cell pellets using the RNeasy™ Plus Mini Kit (#74106, Qiagen, Hilden, Germany) following the manufacturer’s instructions, with minor modifications adapted from Alehaideb et al. (2021). Complementary DNA (cDNA) was synthesized from 200 ng of total RNA per reaction using the High-Capacity cDNA Reverse Transcription Kit (#43–688-14, Applied Biosystems™, Waltham, MA, USA). Quantitative real-time PCR was performed using SYBR™ Green PCR Master Mix on an Applied Biosystems™ 7900 Real-Time PCR System. Each reaction included a no-template negative control to confirm the absence of contamination. The primers targeting BCL-2, BCL-2-associated X (BAX), Tumor protein p53 (TP53), and the housekeeping gene GAPDH were used at 10 pmol/µL. Primer sequences are provided in Table 1. Relative gene expression levels were determined using the comparative Ct method (2−ΔΔCt). This approach involves normalizing the target gene expression to the internal control (GAPDH) and comparing the normalized values between treated and untreated samples to quantify fold changes in gene expression. Table 1Primers used for qPCR analysis and their corresponding nucleotide sequencesGenePrimer Sequence (5' → 3')BAXForward (F): TTTGCTTCAGGGTTTCATCCReverse (R): GCAAAGTAGAAGAGGGCAACCBCL-2F: GGTGAACTGGGGGAGGATTGR: GAGACAGCCAGGAGAAATCAAACTP53F: CCAGCCAAAGAAGAAACCACR: GGTACAGTCAGAGCCAACCTCGAPDHF: CCAGGGCTGCTTTTAACTCTGGR: CCAGTGGACTCCACGACGTAC Table 1Primers used for qPCR analysis and their corresponding nucleotide sequencesGenePrimer Sequence (5' → 3')BAXForward (F): TTTGCTTCAGGGTTTCATCCReverse (R): GCAAAGTAGAAGAGGGCAACCBCL-2F: GGTGAACTGGGGGAGGATTGR: GAGACAGCCAGGAGAAATCAAACTP53F: CCAGCCAAAGAAGAAACCACR: GGTACAGTCAGAGCCAACCTCGAPDHF: CCAGGGCTGCTTTTAACTCTGGR: CCAGTGGACTCCACGACGTAC Primers used for qPCR analysis and their corresponding nucleotide sequences Forward (F): TTTGCTTCAGGGTTTCATCC Reverse (R): GCAAAGTAGAAGAGGGCAACC F: GGTGAACTGGGGGAGGATTG R: GAGACAGCCAGGAGAAATCAAAC F: CCAGCCAAAGAAGAAACCAC R: GGTACAGTCAGAGCCAACCTC F: CCAGGGCTGCTTTTAACTCTGG R: CCAGTGGACTCCACGACGTAC Statistical analysis Data are presented as mean ± standard deviation (SD) from three independent experiments. Dose–response curves were generated by plotting the logarithm of extract concentration (log₁₀ [µg/mL]) against the percentage of viable BC cells to estimate half-maximal inhibitory concentration (IC50) values. Statistical analysis was performed using a one-way ANOVA followed by Tukey’s post hoc test for multiple comparisons, while for the comparison of two groups, an unpaired t-test was used. The values of p less than 0.05 were deemed statistically significant. Chromatographic fingerprinting of TPLME TPLME was subjected to gas chromatography/mass spectrometry (GC/MS) analytical technique, allowing the separation of the main metabolites. Briefly, the extract was analyzed using a GC/MS system equipped with a TriPlus RSH SMART Autosampler, TRACE 1610 GC Oven, and TSQ 9610 Triple Quadrupole Mass Spectrometer. The separation was achieved using a TG-5MS (30m × 0.25 mm, 0.25 µm) capillary column. The temperature program was as follows: 0 to 5 min, 80 °C; 5 to 25 min, 80 to 300 °C; and 25 to 30 min, 300°C. Mass spectra were recorded with a scanning range of 50–600 m/z. Peak identification was performed by comparing the mass spectra with reference spectra from the National Institute of Standards and Technology (NIST) library software. Molecular docking analysis A molecular docking study was carried out to investigate the binding modes of metabolites identified from the TPLME against the anti-apoptotic protein Bcl-2. The three-dimensional crystal structure of Bcl-2 was retrieved from the Protein Data Bank (PDB ID: 6O0K; (Birkinshaw et al. 2019), which is co-crystallized with the selective Bcl-2 inhibitor Venetoclax. The co-crystallized ligand was used to define the active binding site and to serve as a native reference ligand for docking validation. Protein preparation was performed using the Protein Preparation Wizard in Maestro (Schrödinger Release 2025–4: Protein Preparation Workflow, Schrödinger, LLC, New York, NY, 2025), which included the assignment of bond orders, the addition of missing hydrogen atoms, the optimization of hydrogen bonding networks, and the minimization of energy to generate an optimized receptor structure suitable for docking (Sastry et al. 2013). The identified TPLME-derived metabolites were prepared using the LigPrep module (Schrödinger Release 2025–4: LigPrep, Schrödinger, LLC, New York, NY, 2025). The two-dimensional chemical structures were converted into three-dimensional conformations, and appropriate ionization states and stereochemical variations were generated to ensure a reliable ligand representation during docking. A receptor grid was generated using the Receptor Grid Generation tool in Maestro, centered on the binding site occupied by Venetoclax in the Bcl-2 crystal structure. Molecular docking was then performed using the Glide docking module (Schrödinger Release 2025–4: Glide, Schrödinger, LLC, New York, NY, 2025) to predict the preferred binding orientations and interactions of the prepared metabolites within the Bcl-2 binding pocket (Yang et al. 2021). A post-docking analysis was conducted to evaluate ligand–protein interactions, including hydrogen bonding, hydrophobic contacts, and aromatic interactions within the active site of Bcl-2. To validate the docking protocol, Venetoclax was re-docked into the prepared Bcl-2 structure, and the predicted binding pose was compared with its crystallographic conformation to confirm the reliability of the docking methodology. In silico computational analysis Activity prediction The chemical structures of the identified TPLME-derived metabolites were converted into SMILES to represent their molecular features in a simplified textual format, and their corresponding two-dimensional structures were generated using ChemDraw 12 Pro software. Subsequently, the potential anticancer (antineoplastic) activity of these metabolites was computationally evaluated using the online Prediction of Activity Spectra for Substances (PASS) platform (https://www.way2drug.com/PassOnline/index.php). This online platform estimates multiple biological activities by analyzing the relationship between chemical structure and biological function, using prediction models developed from an extensive database of experimentally validated bioactive compounds. The predicted results are expressed as probability values, including the probability of activity (Pa) and the probability of inactivity (Pi) for each biological effect (Filimonov et al. 2014). Absorption, distribution, metabolism, and excretion (ADME) prediction The pharmacokinetic properties related to ADME of the identified TPLME-derived metabolites were computationally assessed using the SwissADME web-based platform (http://www.swissadme.ch/), which is developed and maintained by the Molecular Modeling Group at the Swiss Institute of Bioinformatics (SIB). The SMILES representations of each metabolite were employed as input data to generate the corresponding ADME parameters. This online tool provides a comprehensive evaluation of pharmacokinetic behavior and medicinal chemistry features, thereby contributing to rational decision-making from the earliest stages of drug discovery and development (Daina et al. 2017). Cytochromes P450 (CYP) enzyme inhibition profile The SwissADME web platform (http://www.swissadme.ch/) was used to predict the inhibitory effects of the identified TPLME-derived metabolites on CYP enzymes. The potential interactions of these compounds were evaluated against five major CYP isoforms, namely CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4. This analysis provides valuable insight into possible metabolic interactions and drug–drug interaction risks, thus contributing to the assessment of metabolic stability and safety profiles (Daina et al. 2017). Organ toxicity and safety profile predictions The potential toxicological and safety profiles of the identified TPLME-derived metabolites were comprehensively evaluated using the ProTox-II web server (https://tox.charite.de/protox3/#). The SMILES representations of each metabolite were used as input data to estimate oral toxicity parameters, including toxicity class and median lethal dose (LD₅₀, mg/kg). In addition to organ-specific toxicities (i.e., hepatotoxicity, neurotoxicity, nephrotoxicity, respiratory toxicity, and cardiotoxicity), other toxicity endpoints were assessed, including carcinogenicity, immunotoxicity, mutagenicity, cytotoxicity, blood–brain barrier (BBB) permeability, ecotoxicity, clinical toxicity, and nutritional toxicity. The predicted results were expressed as probability percentages, indicating the likelihood of each toxicological effect. These values provide a comprehensive overview of the safety and toxicity profiles of the studied metabolites (Banerjee et al. 2018). Results Impact of various solvents on the antiproliferative effects of T. polium L. extracts against BC cells MCF-7 and MDA-MB-231 The aerial parts of T. polium L. (TPL) were subjected to extraction using six solvents, including water, ethyl acetate, ethanol, dichloromethane, chloroform, and methanol. The antiproliferative effects of extracts were examined on immortalized ER+, poorly invasive MCF-7 cells and on highly invasive, TNBC MDA-MB-231 cells. The generated sigmoid curves summarized the inhibitory effects of these extracts, tested at increasing concentrations, on the viability of MCF-7 (Fig. 1) and MDA-MB-231 (Fig. 2) cells. The IC50 values of each extract causing a 50% decrease in cell viability were determined and listed in Table 2. On MCF-7 cells (Fig. 1), the chloroform extract showed the strongest antiproliferative effect (IC50 = 170.6 ± 40.4 µg/mL), followed by the dichloromethane extract (IC50 = 211.5 ± 126.8 µg/mL). The methanolic extract demonstrated moderate activity (IC50 = 398.8 ± 160.1 µg/mL), while the water, ethanolic, and ethyl acetate extracts gave IC50 values above 2000 µg/mL, indicating weak or negligible cytotoxic effects. Conversely, using MDA-MB-231 cells (Fig. 2), the methanolic extract exhibited the strongest antiproliferative effect (73.7 ± 9.1 µg/mL), followed by dichloromethane (IC50 = 110.4 ± 24.8 µg/mL), ethyl acetate (IC50 = 182.2 ± 74.9 µg/mL), chloroform (IC50 = 487.5 ± 17.8 µg/mL) and ethanol (IC50 = 572.4 ± 5.1 µg/mL). The remaining aqueous extract displayed IC50 values greater than 2000 µg/mL. Overall, extracts obtained with nonpolar or moderately nonpolar solvents exhibited more pronounced antiproliferative effects than extract obtained with polar solvent extracts. The difference in solvent polarity influenced the types of bioactive compounds extracted. Exhibiting the strongest antiproliferative effect against TNBC MDA-MB-231 cells, the TPL methanolic extract (TPLME) was used to study in depth the molecular mechanisms underlying cell death.Fig. 1Effects of various solvent extracts of T. polium L. (TPL) on MCF-7 cell viability. MCF-7 BC cells were treated with increasing concentrations (log₁₀ scale) of TPL extracts prepared using different solvents (water, methanol, ethanol, ethyl acetate, dichloromethane, and chloroform). After 48 h of incubation, cell viability was assessed using the CellTiter-Glo® Luminescent Cell Viability Assay, and results are presented as the mean ± SD of the percentage of viable cells relative to untreated cells from three independent experiments. (*), (**), and (***) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.001), compared to untreated cells (control) corresponding to 100%Fig. 2Effect of various solvent extracts of TPL on MDA-MB-231 cell viability. MDA-MB-231 BC cells were treated with increasing concentrations (log₁₀ scale) of TPL extracts prepared using different solvents (water, methanol, ethanol, ethyl acetate, dichloromethane, and chloroform). After 48 h of incubation, cell viability was assessed using the CellTiter-Glo® Luminescent Cell Viability Assay, and results are presented as the mean ± SD of percentage of viable cells relative to untreated cells from three independent experiments. (*), (**), and (***) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.001), compared to untreated cells (control) corresponding to 100%Table 2IC50 values (µg/mL) of Teucrium polium L. extracts causing a 50% decrease in MCF-7 and MDA-MB-231 breast cancer cell viability after 48 h of incubationSolvent extraction from TPL aerial partsIC50 value (µg/mL) against MCF-7 cell viabilityIC50 value (µg/mL) against MDA-MB-231 cell viabilityWater > 2000 > 2000Ethyl Acetate > 2000182.2 ± 74.9Ethanol > 2000572.4 ± 5.1Dichloromethane211.5 ± 126.8110.4 ± 24.8Chloroform170.6 ± 40.4488.6 ± 17.7Methanol398.8 ± 160.173.7 ± 9.1 Effects of various solvent extracts of T. polium L. (TPL) on MCF-7 cell viability. MCF-7 BC cells were treated with increasing concentrations (log₁₀ scale) of TPL extracts prepared using different solvents (water, methanol, ethanol, ethyl acetate, dichloromethane, and chloroform). After 48 h of incubation, cell viability was assessed using the CellTiter-Glo® Luminescent Cell Viability Assay, and results are presented as the mean ± SD of the percentage of viable cells relative to untreated cells from three independent experiments. (*), (**), and (***) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.001), compared to untreated cells (control) corresponding to 100% Effect of various solvent extracts of TPL on MDA-MB-231 cell viability. MDA-MB-231 BC cells were treated with increasing concentrations (log₁₀ scale) of TPL extracts prepared using different solvents (water, methanol, ethanol, ethyl acetate, dichloromethane, and chloroform). After 48 h of incubation, cell viability was assessed using the CellTiter-Glo® Luminescent Cell Viability Assay, and results are presented as the mean ± SD of percentage of viable cells relative to untreated cells from three independent experiments. (*), (**), and (***) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.001), compared to untreated cells (control) corresponding to 100% Table 2IC50 values (µg/mL) of Teucrium polium L. extracts causing a 50% decrease in MCF-7 and MDA-MB-231 breast cancer cell viability after 48 h of incubationSolvent extraction from TPL aerial partsIC50 value (µg/mL) against MCF-7 cell viabilityIC50 value (µg/mL) against MDA-MB-231 cell viabilityWater > 2000 > 2000Ethyl Acetate > 2000182.2 ± 74.9Ethanol > 2000572.4 ± 5.1Dichloromethane211.5 ± 126.8110.4 ± 24.8Chloroform170.6 ± 40.4488.6 ± 17.7Methanol398.8 ± 160.173.7 ± 9.1 IC50 values (µg/mL) of Teucrium polium L. extracts causing a 50% decrease in MCF-7 and MDA-MB-231 breast cancer cell viability after 48 h of incubation Apoptotic features observed in BC cell lines exposed to TPLME Morphological hallmarks of apoptosis were noticed in BC MCF-7 and MDA-MB-231 cell lines exposed to TPLME for 48 h of incubation. Compared to healthy untreated (control) and DMSO-treated MCF-7 cells with typical cobblestone morphology, classic apoptotic features, such as cell rounding, membrane blebbing, chromatin condensation, and formation of apoptotic bodies, were observed, highlighting the induction of programmed cell death (Fig. 3). More specifically, compared to healthy, elongated untreated MDA-MB-231 cells (control) and DMSO-treated cells, TPLME-treated cells exhibited extensive morphological changes, including marked cell shrinkage, detachment, nuclear condensation, and abundant apoptotic bodies, collectively suggesting greater susceptibility of this aggressive, therapy-resistant TNBC subtype to TPLME treatment (Fig. 3). However, at the same concentration of TPLME promoting apoptotic morphological features, normal mammary epithelial cells (MCF-10A) retained their typical epithelial morphology, exhibiting intact membranes and no signs of apoptotic alterations, indicating a selective pro-apoptotic activity of TPLME against BC cells, particularly TNBC, while sparing normal mammary epithelial cells (Fig. 3).Fig. 3Morphological alterations observed in MCF-7 and MDA-MB-231 BC cells after treatment with 30 µg/mL TPLME, but not in normal mammary epithelial cells MCF-10A. Untreated (control) and DMSO-treated cells retained typical morphology. In the TPLME-treated MCF-7 and MDA-MB-231 cells, apoptotic features are indicated on the representative photomicrographs: the blue arrow denotes an example of cells undergoing cell shrinkage; the red arrow highlights membrane blebbing; the green arrow marks chromatin condensation and nuclear fragmentation. Scale bar: 400 μm Morphological alterations observed in MCF-7 and MDA-MB-231 BC cells after treatment with 30 µg/mL TPLME, but not in normal mammary epithelial cells MCF-10A. Untreated (control) and DMSO-treated cells retained typical morphology. In the TPLME-treated MCF-7 and MDA-MB-231 cells, apoptotic features are indicated on the representative photomicrographs: the blue arrow denotes an example of cells undergoing cell shrinkage; the red arrow highlights membrane blebbing; the green arrow marks chromatin condensation and nuclear fragmentation. Scale bar: 400 μm TPLME induces cleaved caspase-7 expression and increases caspase-3/−7 activity in BC cell lines A Western blot analysis was performed to evaluate TPLME-induced cell death in BC cells by measuring the expression levels of cleaved caspase-7 protein in MCF-7 and MDA-MB-231 cells exposed to increasing concentrations of TPLME (15, 30, and 60 µg/mL) for 48 h. Cleaved caspase-7 was not detected in healthy untreated and DMSO-treated cells of both cell lines (Fig. 4). Treatment with TPLME resulted in a net, dose-dependent elevation of cleaved caspase-7 expression in both BC cell types (Fig. 4). Quantitative densitometry confirmed significant increases at all concentrations compared to controls, the untreated cells and DMSO-treated cells (Fig. 4). Notably, MDA-MB-231 cells showed a stronger induction of cleaved caspase-7 than MCF-7 cells across all doses, with the highest expression observed at 30 µg/mL (Fig. 4).Fig. 4TPLME induces cleaved caspase-7 expression in human BC cells in a concentration-dependent manner. Representative Western blots showing a significant increase in cleaved caspase-7 expression observed in a dose-dependent manner, especially in MDA-MB-231 cell lysates, compared to MCF-7 cell lysates. Densitometric quantification of cleaved caspase-7 bands was performed using Image J software and normalized with respect to GAPDH, the loading control. A bar graph summarizing the quantitative analysis and the data are presented as mean ± SD from three independent experiments. (*), (**), and (***) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.001), compared to untreated cells (control) and DMSO-treated cells TPLME induces cleaved caspase-7 expression in human BC cells in a concentration-dependent manner. Representative Western blots showing a significant increase in cleaved caspase-7 expression observed in a dose-dependent manner, especially in MDA-MB-231 cell lysates, compared to MCF-7 cell lysates. Densitometric quantification of cleaved caspase-7 bands was performed using Image J software and normalized with respect to GAPDH, the loading control. A bar graph summarizing the quantitative analysis and the data are presented as mean ± SD from three independent experiments. (*), (**), and (***) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.001), compared to untreated cells (control) and DMSO-treated cells Confocal laser scanning microscopy using the Image-iT® LIVE Caspase-3 and −7 Detection Kit was performed to evaluate apoptosis in ER+ MCF-7 and TNBC MDA-MB-231 BC cells following treatment with 30 μg/mL of TPLME. In both cell lines, DMSO-treated controls exhibited predominantly blue nuclear staining with minimal red fluorescence, indicating high cell viability and low caspase-3/−7 activity, respectively (Fig. 5 and Supplementary Figure S2). In contrast, TPLME-treated MCF-7 cells displayed a significant reduction in cell number by 90% (p < 0.01) and increased caspase-3/−7 activity by 50% (p < 0.01), compared to DMSO-treated MCF-7 cells (Fig. 5 and Supplementary Figure S2). Conversely, in MDA-MB-231 cells, the addition of TPLME resulted in a substantial decrease in total cell number by 70% (p < 0.01), compared to DMSO-treated MDA-MB-231 cells, while TPLME augmented caspase-3/−7 activity by 85% (p < 0.001) (Fig. 5 and Supplementary Figure S3).Fig. 5Confocal laser scanning microscopy analysis of caspase-3/−7 activity in MCF-7 and MDA-MB-231 BC cells treated with TPLME. MCF-7 and MDA-MB-231 cells were treated with 30 μg/mL TPLME or vehicle control (DMSO) and stained using the Invitrogen Image-iT® LIVE Caspase-3 and −7 Detection Kit. Representative confocal images show nuclei stained with Hoechst 33,342 (blue), dead cells (green), and apoptotic cells with activated caspase-3/7 (red). In both BC cell lines, TPLME treatment induced a significant increase in red fluorescence compared to controls, indicating apoptosis induction. Bar graphs reporting the quantitative analysis indicates a significant decrease in total cell number per field and an increase in the percentage of caspase-3/7-positive cells in treated groups, compared to DMSO groups. Data are presented as mean ± SD from three independent experiments. (**) and (***) signify a statistically significant difference (p < 0.01 and p < 0.001), compared to DMSO-treated cells. Scale bar = 50 μm Confocal laser scanning microscopy analysis of caspase-3/−7 activity in MCF-7 and MDA-MB-231 BC cells treated with TPLME. MCF-7 and MDA-MB-231 cells were treated with 30 μg/mL TPLME or vehicle control (DMSO) and stained using the Invitrogen Image-iT® LIVE Caspase-3 and −7 Detection Kit. Representative confocal images show nuclei stained with Hoechst 33,342 (blue), dead cells (green), and apoptotic cells with activated caspase-3/7 (red). In both BC cell lines, TPLME treatment induced a significant increase in red fluorescence compared to controls, indicating apoptosis induction. Bar graphs reporting the quantitative analysis indicates a significant decrease in total cell number per field and an increase in the percentage of caspase-3/7-positive cells in treated groups, compared to DMSO groups. Data are presented as mean ± SD from three independent experiments. (**) and (***) signify a statistically significant difference (p < 0.01 and p < 0.001), compared to DMSO-treated cells. Scale bar = 50 μm Apoptosis-related protein expression profiling in MCF-7 and MDA-MB-231 cells treated with TPLME using the human proteome profiler™ array The expression profiling of apoptosis-related proteins detected in MCF-7 and MDA-MB-231 cell lysates after treatment with the optimal effective concentration of TPLME (30 μg/mL) was established using the Human Proteome Profiler™ Array, revealing a distinct modulation of the main apoptotic regulators. In MCF-7 cells, at the difference of the low level of adaptor protein Fas-associated death domain (FADD) and moderate level of the mitochondrial pro-apoptotic protein SMAC/Diablo detected in DMSO-treated cells, the addition of TPLME modulated the expression of proteins playing a dual role in apoptosis, such as catalase– a crucial antioxidant enzyme inhibiting apoptosis by scavenging free radicals and promoting apoptosis by inhibiting stress responses– and heat shock protein (HSP)60. Concurrently, a concomitant increase in the expression levels of anti-apoptotic heme-oxygenase 1 (HO-1/HMOX1) and 2 (HO-2/HMOX2) was observed in TPLME-treated MCF-7 cells (Fig. 6). In MDA-MB-231 cells, compared to the protein expression levels detected in DMSO-treated cells, TPLME surprisingly suppressed the expression of several key apoptosis-related proteins, including Bad, a pro-apoptotic member of the Bcl-2 family that promotes mitochondrial outer membrane permeabilization, and cleaved caspase-3, the effector enzyme responsible for coordinating the destruction of cellular structures (Fig. 6). The protein expression levels of death receptors (DR), including tumor necrosis factor (TNF)-related apoptosis-inducing ligand receptor (TRAIL R)2/DR5 and FADD, were significantly decreased by the addition of TPLME, compared to DMSO-treated MDA-MB-231 cells (Fig. 6). In contrast, TPLME treatment led to an increase in the expression levels of pro-caspase-3 and SMAC/Diablo compared to the DMSO effect (Fig. 6). Notably, activated forms of the tumor suppressor and pro-apoptotic phosphorylated p53 (Ser15) were increased, while phosphorylation in Ser46 and Ser392 was decreased in TPLME-treated MDA-MB-231 cells compared to DMSO-treated cells (Fig. 6). Additionally, like MCF-7 cells, TPLME markedly increased the expression of stress proteins such as the anti-apoptotic HO-1/HMOX1 and the dual apoptotic protein HSP60 in MDA-MB-231 cells, compared to the basal expression levels monitored in DMSO-treated cells (Fig. 6).Fig. 6Apoptosis protein array analysis of MCF-7 and MDA-MB-231 cells treated with TPLME. The left panel shows representative images of the apoptosis-related protein array membranes for MCF-7 (top) and MDA-MB-231 (bottom) cells treated with vehicle control (DMSO) and 30 µg/mL TPLME. Each spot corresponds to a specific apoptosis-related protein, with reference (Ref) spots indicated for normalization. Numbers correspond to individual proteins listed in the x-axis legends of the bar graph. The right panel quantifies the relative expression levels of apoptosis-associated proteins, comparing TPLME-treated cells (white bars) to DMSO-treated cells (black bars). Data are presented as mean ± SD from three independent experiments Apoptosis protein array analysis of MCF-7 and MDA-MB-231 cells treated with TPLME. The left panel shows representative images of the apoptosis-related protein array membranes for MCF-7 (top) and MDA-MB-231 (bottom) cells treated with vehicle control (DMSO) and 30 µg/mL TPLME. Each spot corresponds to a specific apoptosis-related protein, with reference (Ref) spots indicated for normalization. Numbers correspond to individual proteins listed in the x-axis legends of the bar graph. The right panel quantifies the relative expression levels of apoptosis-associated proteins, comparing TPLME-treated cells (white bars) to DMSO-treated cells (black bars). Data are presented as mean ± SD from three independent experiments Effect of TPLME on apoptosis-related gene expression (BAX, BCL-2, and TP53) in MCF-10A, MCF-7, and MDA-MB-231 breast cell lines The relative gene expression levels of BAX, BCL-2, and TP53 were quantified in mammary cell lines MCF-10A (non-tumorigenic), MCF-7 (ER+ BC), and MDA-MB-231 (TNBC) after a 48-h treatment with 30 µg/mL of TPLME. In MCF-10A cells, cell exposure to TPLME resulted in a pronounced upregulation of BCL-2 gene expression (more than tenfold greater than the control), accompanied by a moderate increase in BAX and TP53 gene expression levels, compared to baseline levels measured in untreated cells and DMSO-treated cells (Fig. 7A). In contrast, MCF-7 cells exhibited a strong downregulation of BCL-2 gene expression upon TPLME treatment, suggesting therapeutic potential as a Bcl-2 inhibitor, while BAX and TP53 levels remained largely unchanged compared to controls (Fig. 7B). In MDA-MB-231 cells, TPLME treatment significantly upregulated BAX, BCL-2, and TP53, compared to controls, the untreated cells and DMSO-treated cells (Fig. 7C).Fig. 7Quantitative PCR analysis of apoptosis-related gene expression in MCF-10A (A), MCF-7 (B), and MDA-MB-231 (C) cells following treatment with 30 µg/mL TPLME. The relative mRNA expression levels of pro-apoptotic gene BAX (black bars), anti-apoptotic gene BCL-2 (white bars), and tumor suppressor gene TP53 (gray bars) were measured and normalized to control and DMSO-treated cells. Data are presented as mean ± SD from three independent experiments. (*), (**), and (****) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.0001), compared to untreated (control) and DMSO-treated cells Quantitative PCR analysis of apoptosis-related gene expression in MCF-10A (A), MCF-7 (B), and MDA-MB-231 (C) cells following treatment with 30 µg/mL TPLME. The relative mRNA expression levels of pro-apoptotic gene BAX (black bars), anti-apoptotic gene BCL-2 (white bars), and tumor suppressor gene TP53 (gray bars) were measured and normalized to control and DMSO-treated cells. Data are presented as mean ± SD from three independent experiments. (*), (**), and (****) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.0001), compared to untreated (control) and DMSO-treated cells TPLME differentially increased mitochondrial permeability transition pore opening activity in ER+ MCF-7 (BCL-2 downregulation) and TNBC (BCL-2 upregulation) MDA-MB-231 cells At the functional level, Bcl-2 inhibition leads to an increase in the permeabilization of the outer mitochondrial membrane, causing the activation of the intrinsic apoptotic pathway after the release of cytochrome c. A confocal laser scanning microscopy analysis using the Image-iT® LIVE Mitochondrial Transition Pore Assay kit was conducted to evaluate the effect of TPLME treatment on mitochondrial membrane permeability in MCF-7 and MDA-MB-231 BC cells. For both BC cell lines, TPLME significantly decreased cell numbers (Fig. 8), with similar antiproliferative effects observed when assessing caspase-3/−7 activity (Fig. 5). The DMSO-treated control groups retained the green fluorescence of mitochondrial calcein, indicating intact mitochondrial membranes and low mitochondrial permeability transition pore opening (mPTPO) activity (Fig. 8). In contrast, MCF-7 cells treated with 30 μg/mL TPLME, causing Bcl-2 downregulation (Fig. 7B), showed virtually no detection of green fluorescence due to calcein leakage and resulted in an 80% increase in mPTPO activity, indicating a significant loss of mitochondrial membrane integrity (Fig. 8 and Supplementary Figure S4). However, with regard to MDA-MB-231 cells, TPLME, previously shown to cause Bcl-2 upregulation (Fig. 7C), slightly increased mPTPO activity by 20% (p < 0.05), compared to DMSO-treated cells (Fig. 8 and Supplementary Figure S5).Fig. 8Confocal laser scanning microscopy analysis of mitochondrial permeability transition pore opening (mPTPO) activity in MCF-7 and MDA-MB-231 BC cells following treatment with TPLME. Cells were stained using the Image-iT® LIVE Mitochondrial Transition Pore Assay kit, where green fluorescence indicates intact mitochondrial membranes and mPTPO status. Representative images show strong green fluorescence in DMSO-treated cells, revealing low mPTPO activity, whereas TPLME-treated cells exhibit reduced green fluorescence consistent with increased mPTPO activity and mitochondrial membrane permeabilization. Quantitative graphs depict the significant decrease in total cell number per field and the corresponding increase in mPTPO activity in both BC cell lines after TPLME treatment. Data are presented as mean ± SD from three independent experiments. (*), (**), and (****) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.0001), compared to DMSO-treated cells Confocal laser scanning microscopy analysis of mitochondrial permeability transition pore opening (mPTPO) activity in MCF-7 and MDA-MB-231 BC cells following treatment with TPLME. Cells were stained using the Image-iT® LIVE Mitochondrial Transition Pore Assay kit, where green fluorescence indicates intact mitochondrial membranes and mPTPO status. Representative images show strong green fluorescence in DMSO-treated cells, revealing low mPTPO activity, whereas TPLME-treated cells exhibit reduced green fluorescence consistent with increased mPTPO activity and mitochondrial membrane permeabilization. Quantitative graphs depict the significant decrease in total cell number per field and the corresponding increase in mPTPO activity in both BC cell lines after TPLME treatment. Data are presented as mean ± SD from three independent experiments. (*), (**), and (****) signify a statistically significant difference (p < 0.05, p < 0.01, and p < 0.0001), compared to DMSO-treated cells Identification of the main metabolites in TPLME solution GC/MS analysis of the TPLME allowed the preliminary identification of a group of chemically diverse compounds, mainly belonging to the class of terpenoids, steroids, glycosides, and nitrogen-containing metabolites (Fig. 9). The presence of several sesquiterpene alcohols, including τ-cadinol and α-cadinol, highlights the dominance of terpenoid constituents in the extract. Additionally, several steroid- and triterpenoid-related compounds were also tentatively identified, such as androstane derivatives, cholestane-3-ol derivatives, and andrographolide-related diterpenoids (Fig. 9). The detection of glycosidic and sulfur-containing compounds, including desulphosinigrin, indicates the presence of secondary metabolites commonly associated with antioxidant, chemopreventive, and enzyme-modulating activities. Furthermore, nitrogen-containing compounds such as 1-Hydroxy-4,4-dimethylcyclohexanecarbonitrile were also detected in TPLME (Fig. 9). The metabolic compounds tentatively identified from the TPLME are listed in Table 3.Fig. 9GC/MS chromatogram of TPLMETable 3Compounds (M1–M9) tentatively identified from TPLMEPeak NoArea Counts *minRetention time (min)Probability%Relative Area %Molecular formulaTentative IdentificationLiterature Review of the CompoundsPeak A (M1)400865.81216.13324.9418.11C15H26Oτ-Cadinol(Ho et al. 2011)Peak B (M2)313946.93616.2977.5814.18C15H26Oα-Cadinol(Ho et al. 2011)Peak C (M3)453830.42328.35018.7820.50C20H22O62-Deoxychamaedroxide(Bruno et al. 1987)Peak D (M4)124168.72228.0615.615.61C22H36O3SiAndrostane-11,17-dione, 3-[(trimethylsilyl)oxy]-, (3alpha,5beta)(Uthirasamy et al. 2021)Peak E (M5)25154.3707.01112.331.14C9H15NO1-Hydroxy-4,4-dimethylcyclohexanecarbonitrile(Kadhim et al. 2018)Peak F (M6)13680.39213.15115.950.62C10H17NO7SDesulphosinigrin(Krishnaveni 2015)Peak G (M7)31784.05128.6115.631.44C19H22O3Androsta-1,4-diene-3,6,17-trione(Collins and Krause 1988)Peak H (M8)10249.33919.08713.780.46C28H48OCholestane-3-ol-2 methylene 3ß 5a)-(Ashour et al. 2025)Peak I (M9)65804.35725.16435.282.97C20H28O414-Deoxy-11,12-dehydroandrographolide(Liu et al. 2020) GC/MS chromatogram of TPLME Table 3Compounds (M1–M9) tentatively identified from TPLMEPeak NoArea Counts *minRetention time (min)Probability%Relative Area %Molecular formulaTentative IdentificationLiterature Review of the CompoundsPeak A (M1)400865.81216.13324.9418.11C15H26Oτ-Cadinol(Ho et al. 2011)Peak B (M2)313946.93616.2977.5814.18C15H26Oα-Cadinol(Ho et al. 2011)Peak C (M3)453830.42328.35018.7820.50C20H22O62-Deoxychamaedroxide(Bruno et al. 1987)Peak D (M4)124168.72228.0615.615.61C22H36O3SiAndrostane-11,17-dione, 3-[(trimethylsilyl)oxy]-, (3alpha,5beta)(Uthirasamy et al. 2021)Peak E (M5)25154.3707.01112.331.14C9H15NO1-Hydroxy-4,4-dimethylcyclohexanecarbonitrile(Kadhim et al. 2018)Peak F (M6)13680.39213.15115.950.62C10H17NO7SDesulphosinigrin(Krishnaveni 2015)Peak G (M7)31784.05128.6115.631.44C19H22O3Androsta-1,4-diene-3,6,17-trione(Collins and Krause 1988)Peak H (M8)10249.33919.08713.780.46C28H48OCholestane-3-ol-2 methylene 3ß 5a)-(Ashour et al. 2025)Peak I (M9)65804.35725.16435.282.97C20H28O414-Deoxy-11,12-dehydroandrographolide(Liu et al. 2020) Compounds (M1–M9) tentatively identified from TPLME Binding modes of TPLME-derived metabolites against the Bcl-2 active site After observing a predominant downregulation of BCL-2 in TPLME-treated MCF-7 cells, a molecular docking study was conducted to investigate the binding mode of metabolites identified from TPLME against the anti-apoptotic protein Bcl-2 (PDB ID: 6O0K). Venetoclax was co-crystallized with Bcl-2 and used as the Bcl-2 native ligand to validate the docking protocol and provide a benchmark for comparison. Venetoclax exhibited a docking score of − 8.549 kcal/mol, confirming its strong and specific binding to the Bcl-2 active site. Compared to the native ligand, the TPLME-derived metabolites showed moderate-to-good binding affinities, with docking scores ranging from − 4.360 to − 6.139 kcal/mol as summarized in Table 4. Table 4Molecular docking scores of TPLME-derived metabolites against Bcl-2 (PDB ID: 6O0K)Compound IDMetabolite NameDocking Score (kcal/mol)M1τ-Cadinol−5.694M2α-Cadinol−5.667M32-Deoxychamaedroxide−5.821M4Androstane-11,17-dione, 3-[(trimethylsilyl)oxy]-, (3alpha,5beta)-−4.384M51-Hydroxy-4,4-dimethylcyclohexanecarbonitrile−4.764M6Desulphosinigrin−6.139M7Androsta-1,4-diene-3,6,17-trione−4.360M85.alpha.-Cholestan-3.beta.-ol, 2-methylene-−5.691M914-Deoxy-11,12-dehydroandrographolide−6.127Native ligandVenetoclax−8.549 Table 4Molecular docking scores of TPLME-derived metabolites against Bcl-2 (PDB ID: 6O0K)Compound IDMetabolite NameDocking Score (kcal/mol)M1τ-Cadinol−5.694M2α-Cadinol−5.667M32-Deoxychamaedroxide−5.821M4Androstane-11,17-dione, 3-[(trimethylsilyl)oxy]-, (3alpha,5beta)-−4.384M51-Hydroxy-4,4-dimethylcyclohexanecarbonitrile−4.764M6Desulphosinigrin−6.139M7Androsta-1,4-diene-3,6,17-trione−4.360M85.alpha.-Cholestan-3.beta.-ol, 2-methylene-−5.691M914-Deoxy-11,12-dehydroandrographolide−6.127Native ligandVenetoclax−8.549 Molecular docking scores of TPLME-derived metabolites against Bcl-2 (PDB ID: 6O0K) Among the tested metabolites, Desulphosinigrin (M6) demonstrated the highest binding affinity (− 6.139 kcal/mol), followed closely by 14-Deoxy-11,12-dehydroandrographolide (M9) (− 6.127 kcal/mol). Other metabolites, including 2-Deoxychamaedroxide (M3) (− 5.821 kcal/mol), τ-Cadinol (M1) (− 5.694 kcal/mol), 5α-Cholestan-3β-ol, 2-methylene- (M8) (− 5.691 kcal/mol), and α-Cadinol (M2) (− 5.667 kcal/mol), exhibited moderate binding affinities. In contrast, Androstane-11,17-dione, 3-[(trimethylsilyl)oxy]- (M4), 1-Hydroxy-4,4-dimethylcyclohexanecarbonitrile (M5), and Androsta-1,4-diene-3,6,17-trione (M7) showed comparatively weaker interactions with docking scores above − 4.7 kcal/mol. In addition, the comparison of the two-dimensional chemical structures of the TPLME-derived metabolites and the native ligand Venetoclax (Fig. 10) further supported the differences observed in docking scores. Venetoclax has an extended molecular structure with multiple aromatic and hydrophobic moieties that allow proper binding with the Bcl-2 binding groove. Unlike Venetoclax, the identified TPLME-derived metabolites are smaller, less complex, and structurally diverse, with limited substituent extension and fewer functional groups capable of spanning the full length of the binding pocket.Fig. 10Chemical structure comparison of the TPLME-derived metabolites and the native ligand Venetoclax, indicating differences in complexity and structural extension Chemical structure comparison of the TPLME-derived metabolites and the native ligand Venetoclax, indicating differences in complexity and structural extension The three-dimensional structural comparison explains why, despite the occupation of the binding site (Fig. 11), the metabolites were unable to fully reproduce the binding mode of Venetoclax. Their shorter molecular length restricts their ability to interact with several key regions of the Bcl-2 pocket, resulting in lower docking scores.Fig. 11Binding mode comparison of A) Venetoclax within Bcl-2, B) Overlay between TPLME-derived metabolites and Venetoclax within Bcl-2 active site, and C) TPLME-derived metabolites within Bcl-2 protein (PDB ID: 6O0K), illustrating pocket occupancy in Bcl-2 Binding mode comparison of A) Venetoclax within Bcl-2, B) Overlay between TPLME-derived metabolites and Venetoclax within Bcl-2 active site, and C) TPLME-derived metabolites within Bcl-2 protein (PDB ID: 6O0K), illustrating pocket occupancy in Bcl-2 Docking analysis and comparison of chemical structures jointly indicate that these metabolites require structure-guided optimization to improve binding efficiency. Consequently, these metabolites could be considered as promising compounds or early lead candidates for the development of novel Bcl-2 inhibitors by medicinal chemistry. Antineoplastic activity prediction by PASS analysis PASS (Prediction of activity spectra for substances) analysis of the potential antineoplastic activity of the compounds revealed a variation in the probability of activity (Pa) and probability of inactivity (Pi) values, indicating differences in their predicted biological potentials (Table 5). Among the compounds identified, M7 exhibited the highest predicted antineoplastic activity, with a Pa value of 0.539 and a very low Pi value of 0.015, suggesting a strong likelihood of anticancer potential. This was followed by M9, which showed a relatively high Pa value of 0.443 and a low Pi value of 0.025, indicating promising predicted activity. In contrast, M1 and M2 demonstrated moderate predicted antineoplastic activity, with identical Pa values of 0.319 and low Pi values of 0.022, reflecting moderate biological potential. Similarly, M4 showed moderate predicted activity (Pa = 0.286, Pi = 0.059). Compounds M3, M5, and M8 exhibited relatively low Pa values, ranging from 0.231 to 0.253, accompanied by comparatively higher Pi values, indicating a weaker predicted antineoplastic potential. Notably, M6 recorded the lowest Pa value (0.154) among all compounds analyzed, suggesting the lowest likelihood of anticancer activity. Overall, these findings indicate that M7 and M9 represent the most promising candidates for antineoplastic activity among the identified compounds. Table 5PASS prediction of the antineoplastic activity of identified TPLME-derived metabolitesCompound nameCompound IDProbability of activity (Pa)Probability of inactivity (Pi)τ-CadinolM10.3190.022α-CadinolM20.3190.0222-DeoxychamaedroxideM30.2530.070Androstane-11,17-dione,3-[(trimethylsilyl)oxy]-, (3alpha,5beta)-M40.2860.0591-Hydroxy-4,4-dimethylcyclohexanecarbonitrileM50.2380.077DesulphosinigrinM60.1540.046Androsta-1,4-diene-3,6,17-trioneM70.5390.0155.alpha-Cholestan-3.beta.-ol,2-methyleneM80.2310.08014-Deoxy-11,12-dehydroandrographolideM90.4430.025 Table 5PASS prediction of the antineoplastic activity of identified TPLME-derived metabolitesCompound nameCompound IDProbability of activity (Pa)Probability of inactivity (Pi)τ-CadinolM10.3190.022α-CadinolM20.3190.0222-DeoxychamaedroxideM30.2530.070Androstane-11,17-dione,3-[(trimethylsilyl)oxy]-, (3alpha,5beta)-M40.2860.0591-Hydroxy-4,4-dimethylcyclohexanecarbonitrileM50.2380.077DesulphosinigrinM60.1540.046Androsta-1,4-diene-3,6,17-trioneM70.5390.0155.alpha-Cholestan-3.beta.-ol,2-methyleneM80.2310.08014-Deoxy-11,12-dehydroandrographolideM90.4430.025 PASS prediction of the antineoplastic activity of identified TPLME-derived metabolites ADME property prediction by SwissADME analysis The SwissADME analysis was performed to evaluate the physicochemical properties (Fig. 12) and pharmacokinetic behavior (Table 5) of the identified TPLME-derived metabolites. The molecular weights ranged from 153.22 to 400.68 g/mol, and all values were below 500 g/mol, indicating compliance with oral bioavailability requirements (Table 6). Lipophilicity values (Log Po/w) showed that most of the compounds M1, M2, M3, M4, M5, M7, and M9 were within the acceptable pharmacological range (1—5), while compound M6 had a very low value (−2.09), suggesting a hydrophilic nature. In contrast, compound M8 recorded a very high value (6.54), reflecting high lipophilicity which can negatively affect solubility and absorption, thus explaining its violation of Lipinski’s rule. The predicted solubility values (Log S) ranged from −0.44 to −7.76, with most compounds showing moderate solubility, while compound M8 had the lowest solubility, which may limit its bioavailability. In addition, gastrointestinal absorption analysis revealed that compounds M1, M2, M3, M4, M5, M7, and M9 exhibit high intestinal absorption, whereas M6 and M8 showed low absorption. It is noteworthy that the same group of compounds demonstrated the ability to cross the blood–brain barrier (BBB), suggesting that favorable physicochemical properties, such as appropriate lipophilicity, may contribute to both enhanced intestinal permeability and BBB penetration. The evaluation according to Lipinski’s rule indicated that all compounds complied with the drug-likeness criteria without violations, except for M8, which had one violation due to its high lipophilicity (Table 6). Overall, these results confirm that most compounds possess adequate pharmacokinetic properties, which supports their potential development as promising therapeutic candidates.Fig. 12Radar charts illustrating the predicted oral bioavailability profiles of identified TPLME-derived metabolites, based on key physicochemical parameters, including molecular size (SIZE), polarity (POLAR), aqueous solubility (INSOLU), degree of unsaturation (INSATU), molecular flexibility (FLEX), and lipophilicity (LIPO), as evaluated by SwissADMETable 6Predicted ADME properties for the TPLME-derived metabolites by SwissADME analysisCompound IDMolecular weight (g/mol)Log Po/wLog SGI absorptionBBB permeantLipinskiM1222.373.67−3.26HighYesYes; 0 violationM2222.373.67−3.26HighYesYes; 0 violationM3358.391.84−3.52HighYesYes; 0 violationM4376.603.38−4.81HighYesYes; 0 violationM5153.221.07−1.79HighYesYes; 0 violationM6279.31−2.09−0.44LowNoYes; 0 violationM7298.382.44−3.13HighYesYes; 0 violationM8400.686.54−7.76LowNoYes; 1 violation: MLOGP > 4.15M9332.432.72−3.74HighYesYes; 0 violation Radar charts illustrating the predicted oral bioavailability profiles of identified TPLME-derived metabolites, based on key physicochemical parameters, including molecular size (SIZE), polarity (POLAR), aqueous solubility (INSOLU), degree of unsaturation (INSATU), molecular flexibility (FLEX), and lipophilicity (LIPO), as evaluated by SwissADME Table 6Predicted ADME properties for the TPLME-derived metabolites by SwissADME analysisCompound IDMolecular weight (g/mol)Log Po/wLog SGI absorptionBBB permeantLipinskiM1222.373.67−3.26HighYesYes; 0 violationM2222.373.67−3.26HighYesYes; 0 violationM3358.391.84−3.52HighYesYes; 0 violationM4376.603.38−4.81HighYesYes; 0 violationM5153.221.07−1.79HighYesYes; 0 violationM6279.31−2.09−0.44LowNoYes; 0 violationM7298.382.44−3.13HighYesYes; 0 violationM8400.686.54−7.76LowNoYes; 1 violation: MLOGP > 4.15M9332.432.72−3.74HighYesYes; 0 violation Predicted ADME properties for the TPLME-derived metabolites by SwissADME analysis Prediction of the CYP inhibition profile using the SwissADME website The potential interactions of compounds identified in the TPLME with CYP enzymes were evaluated by SwissADME analysis in order to predict their inhibitory effects on these key drug-metabolizing enzymes (Table 7). The results showed that none of the compounds exhibited inhibitory activity against CYP1A2. Regarding CYP2C19, only compounds M1 and M2 were predicted to act as inhibitors, while the other compounds showed no inhibitory effect. For CYP2C9, inhibitory activity was predicted only for compound M9, whereas all other identified metabolites were non-inhibitory. Furthermore, compounds M3 and M4 were predicted to inhibit CYP2D6, while no inhibitory activity was observed for the other compounds. With respect to CYP3A4, only compound M9 exhibited predicted inhibitory potential, while all other compounds showed no inhibitory activity. These findings indicate that most of the compounds possess a relatively safe metabolic profile with a low likelihood of drug–drug interactions, except a few compounds that may interfere with specific CYP enzymes, which should be considered in future pharmacokinetic and drug development studies. Table 7Prediction of the CYP450 inhibition profile of TPLME-derived metabolites using SwissADMECYP450 EnzymesCompound NumberM1M2M3M4M5M6M7M8M9CYP1A2NoNoNoNoNoNoNoNoNoCYP2C19YesYesNoNoNoNoNoNoNoCYP2C9NoNoNoNoNoNoNoNoYesCYP2D6NoNoYesYesNoNoNoNoNoCYP3A4NoNoNoNoNoNoNoNoYes Table 7Prediction of the CYP450 inhibition profile of TPLME-derived metabolites using SwissADMECYP450 EnzymesCompound NumberM1M2M3M4M5M6M7M8M9CYP1A2NoNoNoNoNoNoNoNoNoCYP2C19YesYesNoNoNoNoNoNoNoCYP2C9NoNoNoNoNoNoNoNoYesCYP2D6NoNoYesYesNoNoNoNoNoCYP3A4NoNoNoNoNoNoNoNoYes Prediction of the CYP450 inhibition profile of TPLME-derived metabolites using SwissADME Organ toxicity prediction using the ProTox-II website An in silico toxicity prediction was conducted to assess acute oral toxicity, organ-specific toxicity, and major toxicological endpoints of TPLME-derived metabolites (Table 8). Most of the identified compounds were classified in toxicity class 5, indicating relatively low acute toxicity, with M1, M2, M4, M7, and M8 showing a high safety of median lethal dose (LD)₅₀ values between 2280 and 5000 mg/kg, while M9 exhibited the lowest toxicity (class 6, LD₅₀ = 6060 mg/kg). In contrast, M5 and M6 predicted high acute toxicity (class 2), with LD₅₀ values of 10 and 15 mg/kg, respectively, while M3 showed moderate toxicity (class 3, LD₅₀ = 280 mg/kg). All compounds were predicted to be non-hepatotoxic, and most displayed inactive neurotoxicity, nephrotoxicity, and cardiotoxicity profiles, except for M8 (neurotoxicity) and M6 and M9 (nephrotoxicity and cardiotoxicity). Most of the compounds exhibited predicted respiratory toxicity. All compounds were predicted to be non-mutagenic and non-cytotoxic, and all except M7 were non-carcinogenic. Immunotoxicity was predicted for M1, M2, M3, M4, M7, M8 and M9. All compounds exhibited predicted blood–brain barrier permeability. Ecotoxicity was predicted for M1, M2, M4, M5, M7 and M8, while M3, M6 and M9 were inactive. Clinical toxicity was predicted for M6, M8, and M9, and nutritional toxicity for M1, M2, M8, and M9. Overall, these findings indicate that most compounds possess a relatively acceptable safety profile, particularly M1, M2, M4, M7, M8 and M9, while M5 and M6 require caution due to their predicted high toxicity. Table 8Predicted toxicity profile of TPLME-derived metabolites using the ProTox-II serverClassification of PredictionCompound IDM1M2M3M4M5M6M7M8M9Oral ToxicityToxicity Class553522556LD50 (mg/kg)2830283028022801015292050006060Predicted Organ toxicity %HepatotoxicityInactive 82Inactive 82Inactive 79Inactive 77Inactive 80Inactive 67Inactive 69Inactive 94Inactive 88NeurotoxicityInactive 50Inactive 50Inactive 82Inactive 73Inactive 61Inactive 78Inactive 56Active 51Inactive 93NephrotoxicityInactive 93Inactive 93Inactive 60Inactive 75Inactive 88Active 65Inactive 87Inactive 90Active 52Respiratory toxicityActive 75Active 75Active 52Active 50Inactive 71Inactive 59Active 91Active 88Active 72CardiotoxicityInactive 85Inactive 85Inactive 54Inactive 62Inactive 78Active 67Inactive 74Inactive 77Active 57Predicted Toxicological Endpoints %CarcinogenicityInactive 66Inactive 66Inactive 50Inactive 60Inactive 78Inactive 57Active 67Inactive 62Inactive 82ImmunotoxicityActive 69Active 69Active 60Active 57Inactive 99Inactive 96Active 77Active 98Active 98MutagenicityInactive 91Inactive 91Inactive 81Inactive 64Inactive 88Inactive 55Inactive 95Inactive 94Inactive 71CytotoxicityInactive 87Inactive 87Inactive 71Inactive 63Inactive79Inactive 71Inactive 72Inactive 94Inactive 60BBB-barrierActive 88Active 88Active 70Active 91Active 86Active54Active97Active 85Active 57EcotoxicityActive 60Active 60Inactive 61Active 57Active 58Inactive 62Active52Active 61Inactive 70Clinical toxicityInactive 50Inactive 50Inactive 55Inactive 65Inactive 55Active 52Inactive 57Active 59Active 67Nutritional toxicityActive 56Active 56Inactive 55Inactive 72Inactive 61Inactive 52Inactive 60Active 90Active 83 Table 8Predicted toxicity profile of TPLME-derived metabolites using the ProTox-II serverClassification of PredictionCompound IDM1M2M3M4M5M6M7M8M9Oral ToxicityToxicity Class553522556LD50 (mg/kg)2830283028022801015292050006060Predicted Organ toxicity %HepatotoxicityInactive 82Inactive 82Inactive 79Inactive 77Inactive 80Inactive 67Inactive 69Inactive 94Inactive 88NeurotoxicityInactive 50Inactive 50Inactive 82Inactive 73Inactive 61Inactive 78Inactive 56Active 51Inactive 93NephrotoxicityInactive 93Inactive 93Inactive 60Inactive 75Inactive 88Active 65Inactive 87Inactive 90Active 52Respiratory toxicityActive 75Active 75Active 52Active 50Inactive 71Inactive 59Active 91Active 88Active 72CardiotoxicityInactive 85Inactive 85Inactive 54Inactive 62Inactive 78Active 67Inactive 74Inactive 77Active 57Predicted Toxicological Endpoints %CarcinogenicityInactive 66Inactive 66Inactive 50Inactive 60Inactive 78Inactive 57Active 67Inactive 62Inactive 82ImmunotoxicityActive 69Active 69Active 60Active 57Inactive 99Inactive 96Active 77Active 98Active 98MutagenicityInactive 91Inactive 91Inactive 81Inactive 64Inactive 88Inactive 55Inactive 95Inactive 94Inactive 71CytotoxicityInactive 87Inactive 87Inactive 71Inactive 63Inactive79Inactive 71Inactive 72Inactive 94Inactive 60BBB-barrierActive 88Active 88Active 70Active 91Active 86Active54Active97Active 85Active 57EcotoxicityActive 60Active 60Inactive 61Active 57Active 58Inactive 62Active52Active 61Inactive 70Clinical toxicityInactive 50Inactive 50Inactive 55Inactive 65Inactive 55Active 52Inactive 57Active 59Active 67Nutritional toxicityActive 56Active 56Inactive 55Inactive 72Inactive 61Inactive 52Inactive 60Active 90Active 83 Predicted toxicity profile of TPLME-derived metabolites using the ProTox-II server Active 54 Active 97 Active 52 Discussion In the search for new natural apoptosis inducers and potential Bcl-2 inhibitors against BC development, our study provides compelling findings revealing that, among the different solvents used for extraction, the methanolic extract of TPL exhibited stronger antiproliferative effects against aggressive TNBC cells than poorly invasive hormone-dependent MCF-7 cells, without affecting normal mammary epithelial cells. Teucrium Polium L. methanolic extract (TPLME) exerted differential pro-apoptotic mechanisms, including moderate caspase-3/−7 activation and high mPTPO activity associated with a significant downregulation of BCL-2 gene expression in MCF-7 cells; conversely, in MDA-MB-231 cells, high caspase-3/7 and low mPTPO activities accompanied by BAX, BCL-2, and TP53 upregulation were observed upon TPLME treatment. Through chemical analysis of the TPLME and a molecular docking study, bioactive metabolites were identified and predicted molecular interactions with the active site of Bcl-2. In silico predictions revealed a high probability of antineoplastic activity and a safe pharmacokinetic profile for most metabolites. These findings suggest the promising potential of TPLME-derived metabolites, particularly M9 (14-Deoxy-11,12-dehydroandrographolide), as a safe novel natural compound and as an early lead candidate for the development of novel Bcl-2 inhibitors against BC. Evaluation of the antiproliferative properties of various extracts against ER+ BC MCF-7 cells and TNBC MDA-MB-231 cells revealed that chloroform and dichloromethane TPL extracts exhibited the most promising activity against the MCF-7 cells, while the methanolic extract was the most potent against MDA-MB-231 cell growth. In addition, TPLME was reported to enhance the cytotoxic effects of chemotherapeutic agents such as vinblastine, vincristine, and doxorubicin against various cancer cell lines, including ER+ BC MCF-7 cells (Rajabalian 2008). Previous studies reported that the chloroform extract of several plants, including Salvia species, Cryptocarya massoy bark, and Melastoma malabathricum had marked cytotoxic activity against various cancer cell lines, including MCF-7 cells, compared with methanolic or ethanolic extracts (Widiyastuti et al. 2018). In contrast, other studies have demonstrated that polar solvent fractions tend to exhibit stronger anticancer activities. In line with our findings, the methanolic extract of Artemisia Sieberi Besser exhibited greater antiproliferative activity against TNBC cells than MCF-7 cells (Albani et al. 2025). However, Achillea fragrantissima dichloromethane reduced TNBC cell growth more than polar extracts (Alshuail et al. 2022). In our current study, the difference in IC₅₀ values between MCF-7 and MDA-MB-231 cells could be attributed to variations in receptor expression and metabolic characteristics between ER+ and TNBC BC cells, which often make the latter less responsive to phytochemicals (Kooti et al. 2017). The strong antiproliferative activity observed in chloroform and dichloromethane extracts suggests that nonpolar and polar phytochemicals, such as terpenoids, alkaloids, or certain flavonoids, could contribute significantly to the antiproliferative effects against BC cells. These classes of compounds are known for their ability to disrupt cancer cell membranes, interfere with intracellular signaling, and induce apoptosis (Sun et al. 2024; Al Khalily et al. 2025). Thus, BC cell line-specific susceptibility in response to some methods of extraction suggests further purification and isolation of the active constituents to identify more potent bioactive molecules targeting TNBC, the most challenging BC subtype. TPLME triggers selective programmed cell death (i.e., apoptosis) mechanisms in BC cells while sparing non-tumorigenic mammary epithelial cells. Morphological analysis revealed that MCF-10A cells maintained normal epithelial characteristics after treatment with TPLME, indicating minimal toxicity to healthy breast tissue. This selective profile is particularly important given the need for cancer therapies that minimize damage to normal cells (Obidiroet al. 2023). Conversely, upon TPLME treatment, the MCF-7 and MDA-MB-231 BC cell lines exhibited classical apoptotic morphological changes, including cell rounding, membrane blebbing, chromatin condensation, and apoptotic body formation. These apoptotic features are consistent with previous studies (Qian et al. 2020). Here, MDA-MB-231 TNBC cells displayed more pronounced apoptotic changes than MCF-7 cells, paving the way for underlying molecular investigation and the establishment of the structure–function relationship between TPLME and apoptotic proteins. The current study highlights the role of caspase-mediated apoptosis in the pro-apoptotic effects of TPLME on BC cells. Western blot analysis revealed a dose-dependent increase in cleaved caspase-7 detected in MCF-7 and MDA-MB-231 cells after TPLME treatment, confirmed by the caspase-3/−7 activity assessed by real-time cell imaging. Notably, the TNBC MDA-MB-231 cells exhibited significantly stronger induction of caspase-7 cleavage than ER+ MCF-7 cells, reflecting caspase-7 expression dependent on estrogen therapy in ER+ BC cells (Chaudhary et al. 2016). The absence of caspase-7 activation in control cells and DMSO-treated cells indicates that this effect is specific to the bioactive compounds in the extract rather than to nonspecific stress. These current results are consistent with prior evidence highlighting that TPL extract contains phytochemicals capable of triggering caspase activation and intrinsic apoptotic pathways involving mitochondrial signaling (Rajabalian 2008). The comparatively lower caspase-7 activation in MCF-7 cells may be due to their lack of functional caspase-3, which typically amplifies apoptotic signaling, thereby attenuating the response (Jänicke et al. 1998). Apoptosis protein array analysis revealed distinct protein expression profiles in cell lysates from ER+ BC and TNBC. In MCF-7 cell lysates, TPLME mainly modulated the expression of proteins playing a dual role in apoptosis, such as catalase and hsp60, while the expression levels of HO-1 and HO-2 were significantly increased. In vitro studies have reported that many natural compounds induce apoptosis in MCF-7 cells by modulating catalase expression and activity, resulting in the generation of elevated levels of reactive oxygen species, which trigger cytochrome c release and, consequently, mitochondria-dependent apoptosis (Wei et al. 2023). Among the antioxidant properties of natural compounds, HO enzymes, particularly HO-1, are well known to be highly inducible by chemical and physical agents such as hydrogen peroxides, nitric oxide and hypoxia, in order to degrade heme and produce antioxidants for cell protection (Funes et al. 2020). Regarding TNBC cell lysates, TPLME markedly increased tumor suppressor p53 phosphorylation at Ser15 while reducing phosphorylation at Ser46 and Ser392. Ser15 phosphorylation is an early event in p53 activation, usually mediated by ATM/ATR kinases in response to DNA damage or cellular stress, leading to p53 stabilization and prevention of its degradation (Liebl and Hofmann 2019). In contrast, p53 phosphorylation at Ser46 is closely linked to transcriptional activation of pro-apoptotic genes such as p53AIP1, BAX, and PUMA, which promote mitochondrial apoptosis (Yamada and Yoshida 2019). Phosphorylation of p53 at Ser392, on the other hand, enhances the structural stability and DNA-binding affinity of p53 (Kim et al. 2004). The observed increase in Ser15 phosphorylation of p53 alongside reductions in Ser46 and Ser392 suggests that TPLME activates p53 as part of a stress-response mechanism, but may not fully commit the cell to p53-dependent transcriptional apoptosis. The differential modulation of apoptosis-related genes observed across the MCF-10A, MCF-7 and MDA-MB-231 cell lines suggests that TPLME exerts cell type-specific effects on survival and death pathways in breast cells. In normal MCF-10A cells, the pronounced upregulation of BCL-2 following exposure to TPLME indicates the activation of cytoprotective mechanisms aimed at maintaining cellular homeostasis under stress conditions. This adaptive response could counterbalance the moderate increases in BAX and TP53, thereby preventing unnecessary apoptosis in non-tumorigenic cells. In contrast, the marked downregulation of BCL-2 in MCF-7 cells, without significant alterations in BAX or TP53, implies that TPLME selectively impairs anti-apoptotic signaling in ER+ BC cells. Such a decrease in BCL-2 gene expression may enhance apoptotic susceptibility by altering the BAX/BCL-2 ratio, thereby potentiating mPTPO activity, as previously observed in MCF-7 cells by confocal microscopy. Meanwhile, in the TNBC MDA-MB-231 cell line, the concomitant upregulation of BAX, BCL-2, and TP53 suggests that TPLME activates intrinsic apoptotic cascades, with the relative increase in BAX and TP53 promoting a net pro-apoptotic outcome. These findings align with previous evidence indicating that phytochemicals in TPL extract possess cytotoxic and pro-apoptotic properties that are often more pronounced in malignant than in normal cells (Basharat et al. 2025). Collectively, current results highlight the potential of TPLME as a selective anticancer agent capable of targeting malignant cells while preserving the viability of normal breast epithelial cells. However, the pronounced upregulation of BCL-2 in MCF-10A cells upon TPLME treatment underscores the need for rigorous dose optimization to mitigate potential cytoprotective or stress-related effects in healthy tissues (Alatawi et al. 2025). In addition, TPL has been reported to contain phytoestrogen substances (e.g., phenylpropanoid glycosides), which can induce BCL-2 gene expression via estrogen-responsive elements (Simonyan and Chavushyan 2016), as observed in TPLME-treated MCF-10A and TNBC cells, whose estrogenic activity occurs despite the lack of ER (Treek et al. 2020). Further studies focusing on phytoestrogen substances and on upstream regulators and downstream effectors of these apoptotic pathways would help clarify the molecular basis of the selective action of TPLME on different BC subtypes, in particular ER+ BC showing a surprising BCL-2 downregulation in MCF-7 cells. Deeper investigations aimed at analyzing the molecular pathways involved, determining the active constituents of the extract, and evaluating its efficacy in vivo would be of great interest. An attempt was made to identify the bioactive metabolites separated from the TPLME, and 9 peaks were revealed as follows:Peak A (M1) was detected at a retention time of 16.133 min. The compound showed a proposed molecular formula of C₁₅H₂₆O with a matching probability of 24.94%. Based on the relatively high probability score and good concordance with reference spectral data, this peak was tentatively identified as τ-Cadinol. This compound has been previously reported as a characteristic sesquiterpene alcohol in phytochemical investigations of essential oils and plant extracts, supporting the reliability of the present identification (Ho et al. 2011).Peak B (M2) was detected at a retention time of 16.297 min. The proposed molecular formula was C₁₅H₂₆O, with a matching probability of 7.58%. Despite the lower probability score compared with Peak A, the spectral profile showed acceptable concordance with reported data, allowing tentative identification as α-Cadinol. This compound has been previously documented among sesquiterpene alcohols in phytochemical analyses of essential oils (Ho et al. 2011).Peak C (M3) appeared at a retention time of 28.350 min and was associated with a molecular formula of C₂₀H₂₂O₆, exhibiting a matching probability of 18.78%. The relatively strong probability value and consistency with reference spectral information supported the tentative identification of this peak as 2-Deoxychamaedroxide. This compound has been reported in earlier phytochemical studies and was cataloged in the Knapsack database, reinforcing the plausibility of this assignment (Bruno et al. 1987).Peak D (M4) was observed at a retention time of 28.061 min with a proposed molecular formula of C₂₂H₃₆O₃Si and a matching probability of 5.61%. Although the probability score was modest, the detected mass characteristics aligned with published spectral data, leading to the tentative identification of Androstane-11,17-dione, 3-[(trimethylsilyl)oxy]-, (3α,5β). Similar steroidal derivatives have already been reported in analytical studies involving silylated compounds (Uthirasamy et al. 2021).Peak E (M5) was detected at an earlier retention time of 7.011 min and showed a proposed molecular formula of C₉H₁₅NO, with a matching probability of 12.33%. Based on the probability score and concordance with available reference data, this compound was tentatively identified as 1-Hydroxy-4,4-dimethylcyclohexanecarbonitrile. This identification is consistent with compounds previously described in related chemical profiling studies (Kadhim et al. 2018).Peak F (M6) appeared at a retention time of 13.151 min and was assigned a molecular formula of C₁₀H₁₇NO₇S, with a matching probability of 15.95%. The probability value and spectral concordance supported the tentative identification of this peak as Desulphosinigrin. This compound has been previously reported in phytochemical investigations, particularly in studies focusing on sulfur-containing metabolites (Krishnaveni 2015).Peak G (M7) was detected at a retention time of 28.611 min with a proposed molecular formula of C₁₉H₂₂O₃ and a matching probability of 5.63%. Despite the relatively low probability score, the spectral features were consistent with reported data for steroidal compounds, allowing tentative identification as Androsta-1,4-diene-3,6,17-trione. Similar compounds have been described in earlier analytical and biochemical studies (Collins and Krause 1988).Peak H (M8) was observed at a retention time of 19.087 min and showed a molecular formula of C₂₈H₄₈O, with a matching probability of 13.78%. The close concordance between the detected spectral data and literature references supported tentative identification as Cholestane-3-ol-2-methylene (3β,5α). Cholestane derivatives have been widely reported in sterol profiling studies, supporting this assignment (Ashour et al. 2025).Peak I (M9) was detected at a retention time of 25.164 min and demonstrated the highest matching probability among the identified peaks at 35.28%. The proposed molecular formula was C₂₀H₂₈O₄. Based on the strong probability score and excellent concordance with reference spectral data, this compound was tentatively identified as 14-Deoxy-11,12-dehydroandrographolide, a diterpenoid compound previously described in phytochemical and pharmacological studies (Liu et al. 2020). Peak A (M1) was detected at a retention time of 16.133 min. The compound showed a proposed molecular formula of C₁₅H₂₆O with a matching probability of 24.94%. Based on the relatively high probability score and good concordance with reference spectral data, this peak was tentatively identified as τ-Cadinol. This compound has been previously reported as a characteristic sesquiterpene alcohol in phytochemical investigations of essential oils and plant extracts, supporting the reliability of the present identification (Ho et al. 2011). Peak A (M1) was detected at a retention time of 16.133 min. The compound showed a proposed molecular formula of C₁₅H₂₆O with a matching probability of 24.94%. Based on the relatively high probability score and good concordance with reference spectral data, this peak was tentatively identified as τ-Cadinol. This compound has been previously reported as a characteristic sesquiterpene alcohol in phytochemical investigations of essential oils and plant extracts, supporting the reliability of the present identification (Ho et al. 2011). Peak B (M2) was detected at a retention time of 16.297 min. The proposed molecular formula was C₁₅H₂₆O, with a matching probability of 7.58%. Despite the lower probability score compared with Peak A, the spectral profile showed acceptable concordance with reported data, allowing tentative identification as α-Cadinol. This compound has been previously documented among sesquiterpene alcohols in phytochemical analyses of essential oils (Ho et al. 2011). Peak B (M2) was detected at a retention time of 16.297 min. The proposed molecular formula was C₁₅H₂₆O, with a matching probability of 7.58%. Despite the lower probability score compared with Peak A, the spectral profile showed acceptable concordance with reported data, allowing tentative identification as α-Cadinol. This compound has been previously documented among sesquiterpene alcohols in phytochemical analyses of essential oils (Ho et al. 2011). Peak C (M3) appeared at a retention time of 28.350 min and was associated with a molecular formula of C₂₀H₂₂O₆, exhibiting a matching probability of 18.78%. The relatively strong probability value and consistency with reference spectral information supported the tentative identification of this peak as 2-Deoxychamaedroxide. This compound has been reported in earlier phytochemical studies and was cataloged in the Knapsack database, reinforcing the plausibility of this assignment (Bruno et al. 1987). Peak C (M3) appeared at a retention time of 28.350 min and was associated with a molecular formula of C₂₀H₂₂O₆, exhibiting a matching probability of 18.78%. The relatively strong probability value and consistency with reference spectral information supported the tentative identification of this peak as 2-Deoxychamaedroxide. This compound has been reported in earlier phytochemical studies and was cataloged in the Knapsack database, reinforcing the plausibility of this assignment (Bruno et al. 1987). Peak D (M4) was observed at a retention time of 28.061 min with a proposed molecular formula of C₂₂H₃₆O₃Si and a matching probability of 5.61%. Although the probability score was modest, the detected mass characteristics aligned with published spectral data, leading to the tentative identification of Androstane-11,17-dione, 3-[(trimethylsilyl)oxy]-, (3α,5β). Similar steroidal derivatives have already been reported in analytical studies involving silylated compounds (Uthirasamy et al. 2021). Peak D (M4) was observed at a retention time of 28.061 min with a proposed molecular formula of C₂₂H₃₆O₃Si and a matching probability of 5.61%. Although the probability score was modest, the detected mass characteristics aligned with published spectral data, leading to the tentative identification of Androstane-11,17-dione, 3-[(trimethylsilyl)oxy]-, (3α,5β). Similar steroidal derivatives have already been reported in analytical studies involving silylated compounds (Uthirasamy et al. 2021). Peak E (M5) was detected at an earlier retention time of 7.011 min and showed a proposed molecular formula of C₉H₁₅NO, with a matching probability of 12.33%. Based on the probability score and concordance with available reference data, this compound was tentatively identified as 1-Hydroxy-4,4-dimethylcyclohexanecarbonitrile. This identification is consistent with compounds previously described in related chemical profiling studies (Kadhim et al. 2018). Peak E (M5) was detected at an earlier retention time of 7.011 min and showed a proposed molecular formula of C₉H₁₅NO, with a matching probability of 12.33%. Based on the probability score and concordance with available reference data, this compound was tentatively identified as 1-Hydroxy-4,4-dimethylcyclohexanecarbonitrile. This identification is consistent with compounds previously described in related chemical profiling studies (Kadhim et al. 2018). Peak F (M6) appeared at a retention time of 13.151 min and was assigned a molecular formula of C₁₀H₁₇NO₇S, with a matching probability of 15.95%. The probability value and spectral concordance supported the tentative identification of this peak as Desulphosinigrin. This compound has been previously reported in phytochemical investigations, particularly in studies focusing on sulfur-containing metabolites (Krishnaveni 2015). Peak F (M6) appeared at a retention time of 13.151 min and was assigned a molecular formula of C₁₀H₁₇NO₇S, with a matching probability of 15.95%. The probability value and spectral concordance supported the tentative identification of this peak as Desulphosinigrin. This compound has been previously reported in phytochemical investigations, particularly in studies focusing on sulfur-containing metabolites (Krishnaveni 2015). Peak G (M7) was detected at a retention time of 28.611 min with a proposed molecular formula of C₁₉H₂₂O₃ and a matching probability of 5.63%. Despite the relatively low probability score, the spectral features were consistent with reported data for steroidal compounds, allowing tentative identification as Androsta-1,4-diene-3,6,17-trione. Similar compounds have been described in earlier analytical and biochemical studies (Collins and Krause 1988). Peak G (M7) was detected at a retention time of 28.611 min with a proposed molecular formula of C₁₉H₂₂O₃ and a matching probability of 5.63%. Despite the relatively low probability score, the spectral features were consistent with reported data for steroidal compounds, allowing tentative identification as Androsta-1,4-diene-3,6,17-trione. Similar compounds have been described in earlier analytical and biochemical studies (Collins and Krause 1988). Peak H (M8) was observed at a retention time of 19.087 min and showed a molecular formula of C₂₈H₄₈O, with a matching probability of 13.78%. The close concordance between the detected spectral data and literature references supported tentative identification as Cholestane-3-ol-2-methylene (3β,5α). Cholestane derivatives have been widely reported in sterol profiling studies, supporting this assignment (Ashour et al. 2025). Peak H (M8) was observed at a retention time of 19.087 min and showed a molecular formula of C₂₈H₄₈O, with a matching probability of 13.78%. The close concordance between the detected spectral data and literature references supported tentative identification as Cholestane-3-ol-2-methylene (3β,5α). Cholestane derivatives have been widely reported in sterol profiling studies, supporting this assignment (Ashour et al. 2025). Peak I (M9) was detected at a retention time of 25.164 min and demonstrated the highest matching probability among the identified peaks at 35.28%. The proposed molecular formula was C₂₀H₂₈O₄. Based on the strong probability score and excellent concordance with reference spectral data, this compound was tentatively identified as 14-Deoxy-11,12-dehydroandrographolide, a diterpenoid compound previously described in phytochemical and pharmacological studies (Liu et al. 2020). Peak I (M9) was detected at a retention time of 25.164 min and demonstrated the highest matching probability among the identified peaks at 35.28%. The proposed molecular formula was C₂₀H₂₈O₄. Based on the strong probability score and excellent concordance with reference spectral data, this compound was tentatively identified as 14-Deoxy-11,12-dehydroandrographolide, a diterpenoid compound previously described in phytochemical and pharmacological studies (Liu et al. 2020). Overall, the compounds identified across the chromatographic profile demonstrate a dominance of oxygenated mono and sesquiterpene derivatives, supported by high probability scores and consistent molecular formulas. These findings indicate a chemically rich extract, containing constituents previously associated with bioactive and functional properties, thus reinforcing the reliability of the analytical approach and the relevance of the identified metabolites. A molecular docking analysis was performed to explore the binding mode of metabolites identified from TPLME against the anti-apoptotic protein Bcl-2, using Venetoclax as the native reference ligand. The inclusion of Venetoclax provided a reliable benchmark for evaluating binding orientation and pocket engagement within the Bcl-2 binding groove. Comparison of chemical structures revealed marked differences between the metabolites and the native ligand. Venetoclax possesses an elongated and optimized structure, containing multiple aromatic and hydrophobic moieties, enabling it to efficiently span the Bcl-2 binding groove and establish extensive stabilizing interactions. In contrast, TPLME-derived metabolites exhibit comparatively shorter and less extensive molecular architectures, with a limited distribution of functional groups and reduced structural complexity. Furthermore, the docking study analysis demonstrated that the metabolites were able to occupy the Bcl-2 pocket; however, this occupancy was partial and did not extend to the entire binding site, unlike Venetoclax. This incomplete pocket coverage could explain the lower docking scores. The combined comparison of structures and modes of binding suggests that, although these natural compounds possess core scaffolds compatible with Bcl-2 recognition, their molecular structures are insufficient to fully mimic the native ligand’s interaction profile. Consequently, these metabolites should be regarded as preliminary hit structures rather than optimized inhibitors. Structure-guided modifications would be necessary to enhance binding pocket coverage and improve interaction strength with the active site of Bcl-2. At the functional level, Bcl-2 plays a crucial role primarily in neutralizing the mitochondrial proteins Bax and Bak, which then prevent pore formation in the outer mitochondrial membrane and hinder the release of cytochrome c. The downregulation of BCL-2 gene expression in TPLME-treated MCF-7 cells, corresponding also to the impact of Bcl-2 inhibitors, resulted in a significant increase in mPTPO activity in MCF-7 cells. Conversely, the upregulation of BCL-2 gene expression in TPLME-treated MDA-MB-231 cells resulted in a low mPTPO activity. mPTPO activity is a key event in the intrinsic apoptotic pathway, causing the loss of mitochondrial membrane potential, thereby inducing the intrinsic apoptotic pathway and triggering cell death (Vogler et al. 2025). The observed increase in mPTPO activity suggests that TPLME induces apoptosis by disrupting mitochondrial integrity. Overall, current findings highlight the importance of molecular length and structural extension for effective Bcl-2 inhibition and indicate that TPLME-derived metabolites may serve as promising starting points for further medicinal chemistry development aimed at designing more potent Bcl-2 inhibitors. These could be used in combination with conventional endocrine therapy for ER+ BC (Muttiah et al. 2022) and as a promising therapeutic agent for TNBC patients (Liu et al. 2025). Moreover, the selective apoptotic induction by TPLME, particularly against TNBC cells, coupled with the preservation of normal breast epithelial cell morphology, supports its further investigation as a promising phytotherapeutic candidate. Using in silico computational studies, the TPLME-derived metabolites M1 and M2 showed identical prediction profiles due to their high structural similarity as stereoisomers, which could not be distinguished by models primarily based on two-dimensional molecular descriptors; therefore, M1 and M2 were treated as a single bioactivity profile. In line with this limitation, previous studies have reported that stereochemistry is often ignored or considered only as a post-processing step in many current methods, which may limit the identification of optimal candidates (Tom et al. 2025). PASS analysis revealed variability in predicted biological activity among compounds, with M7 exhibiting the highest Pa and a very low Pi, suggesting strong potential biological relevance. This observation is supported by studies on structurally related steroidal aromatase inhibitors, including one compound bearing an androsta-diene-dione framework, which has demonstrated antiproliferative effects in ER+ BC cells, highlighting the therapeutic importance of multi-target mechanisms in ER+ BC (Amaral et al. 2025). In addition, the predicted activity of M9 is consistent with previous reports demonstrating that 14-deoxy-11,12-dehydroandrographolide derivatives exhibit antiproliferative effects in multiple cancer cell lines, including breast cancer MCF-7 cells (Kumar et al. 2024). ADME properties are key determinants of the potential of compounds to be developed as drug candidates, as they allow prediction of in vivo behavior and pharmacological suitability (Ancuceanu et al. 2025). The SwissADME analysis indicated that most compounds identified in TPLME possess favorable physicochemical and pharmacokinetic characteristics. All compounds had molecular weights below 500 g/mol, consistent with Lipinski’s rule and indicating good oral absorption potential (Wu et al. 2024). Most compounds showed Log Po/w values within the acceptable pharmacological range (1—5), reflecting an appropriate balance between lipophilicity and aqueous solubility, promoting efficient intestinal absorption (Wu et al. 2024; Klimoszek et al. 2024). In contrast, compound M6 displayed very low lipophilicity, which may limit membrane permeability, while compound M8 exhibited excessively high lipophilicity, negatively affecting its solubility and absorption and resulting in a violation of Lipinski’s rule. These findings are supported by solubility (Log S) predictions, with most compounds demonstrating moderate solubility, a desirable feature for oral drug development, while the poor solubility of compound M8 may hinder its pharmaceutical potential (Klimoszek et al. 2024). Furthermore, compounds with moderate lipophilicity were predicted to have high intestinal absorption, thus highlighting the close link between physicochemical properties and gastrointestinal absorption behavior (Menestrina et al. 2025). In addition, the ability of these compounds to cross the blood–brain barrier suggests that appropriate molecular size and moderate lipophilicity may contribute to enhanced permeability. Overall, this analysis indicates that most of the studied compounds exhibit promising drug-like properties and merit further biological evaluation, whereas compounds with unfavorable characteristics may require structural optimization or advanced delivery strategies. The focus on drug metabolism has resulted in predictive analysis targeting CYP, the primary enzymatic system responsible for phase I drug metabolism, and its inhibition or induction may alter pharmacokinetics and lead to drug-drug interactions (Iacopetta et al. 2023). The SwissADME analysis revealed that most TPLME-derived metabolites exhibit a favorable metabolic profile, with no predicted inhibition of CYP1A2. Overall, these results indicate a low potential for CYP-mediated interactions, although metabolites with predicted inhibitory activity, particularly M9, require further experimental validation. The ProTox-II platform indicates that most TPLME-derived metabolites exhibit an acceptable safety profile with respect to acute and organ-specific toxicity. Most of the compounds were classified in toxicity class 5, corresponding to low acute oral toxicity and confirming their potential use within safe therapeutic ranges (Setiani et al. 2023; Banerjee et al. 2024). Moreover, compound M9 was classified in toxicity class 6, the least toxic category, further highlighting its promise as a drug candidate. In contrast, compounds M5 and M6 were classified in toxicity class 2, indicating very high acute toxicity and requiring caution in their pharmaceutical development. Regarding organ-specific toxicity, the predicted absence of hepatotoxicity across all metabolites represents a favorable finding, particularly considering the central role of the liver in drug metabolism (Zhao et al. 2021). ProTox-II further enables toxicity assessment by integrating predictive models for multi-organ toxicity and key toxicological assessment criteria, including mutagenicity, carcinogenicity, cytotoxicity, immunotoxicity, and blood–brain barrier permeability, thus enabling a more comprehensive evaluation of target organ risks and expected effects (Setiani et al. 2023). Conclusion This study underscores the cell line-specific antiproliferative and pro-apoptotic effects of TPLME against BC cells, revealed by a pronounced cytotoxic activity against aggressive MDA-MB-231 TNBC cells compared to hormone-dependent MCF-7 ER+ BC cells, while sparing normal mammary epithelial cells. TPLME increased executioner caspase-7 cleavage and exhibited distinct apoptotic proteome profiling, including BCL-2 downregulation with high mPTPO activity in MCF-7 cells, while upregulating key pro-apoptotic BAX and TP53 genes and anti-apoptotic BCL-2 leading to moderate mPTPO activity in MDA-MB-231 cells. In silico analyses have predicted molecular interactions between TPLME-derived metabolites, particularly M9 (14-Deoxy-11,12-dehydroandrographolide), and the active site of Bcl-2, suggesting them as potential and pharmacologically safe early lead candidates for Bcl-2 inhibitor development. These findings warrant further anticancer mechanistic investigations and in vivo studies to evaluate the efficacy and safety of TPLME-derived metabolites and their potential integration into combination therapeutic strategies for drug-resistant BC overexpressing Bcl-2, particularly clinically challenging TNBC. Supplementary Information Below is the link to the electronic supplementary material. ESM 1Representative image of a specimen of Teucrium polium L. after identification and authentication. A voucher specimen was deposited in the herbarium of King Saud University, Riyadh, under voucher number KSU 24645 (PNG 991 KB)Supplementary file1 (TIF 3591 KB)ESM 2Representative photomicrographs showing separate fluorescent dyes from merged confocal images of MCF-7 cells treated with DMSO (top panel) and TPLME (bottom panel): Hoechst 33342 (blue fluorescence) staining nuclei, SytoxTM (green fluorescence) staining nucleic acids of the dead cell, and red fluorochrome staining caspase-3 and caspase-7 (PNG 1.82 MB)Supplementary file2 (TIF 20815 KB)ESM 3Representative photomicrographs showing separate fluorescent dyes from merged confocal images of MDA-MB-231 cells treated with DMSO (top panel) and TPLME (bottom panel): Hoechst 33342 (blue fluorescence) staining nuclei, SytoxTM (green fluorescence) staining nucleic acids of the dead cell, and red fluorochrome staining caspase-3 and caspase-7 (PNG 2.48 MB)Supplementary file3 (TIF 27899 KB)ESM 4Representative photomicrographs showing separate fluorescent dyes from merged confocal images of MCF-7 cells treated with DMSO (top panel) and TPLME (bottom panel): Hoechst 33342 (blue fluorescence) staining nuclei, green fluorescence representing calcein/quencher cobalt, and red fluorochrome staining the mitochondria (PNG 1.66 MB)Supplementary file4 (TIF 17297 KB)ESM 5Representative photomicrographs showing separate fluorescent dyes from merged confocal images of MDA-MB-231 cells treated with DMSO (top panel) and TPLME (bottom panel): Hoechst 33342 (blue fluorescence) staining nuclei, green fluorescence representing calcein/quencher cobalt, and red fluorochrome staining the mitochondria (PNG 2.38 MB)Supplementary file5 (TIF 26266 KB) Representative image of a specimen of Teucrium polium L. after identification and authentication. A voucher specimen was deposited in the herbarium of King Saud University, Riyadh, under voucher number KSU 24645 (PNG 991 KB) Supplementary file1 (TIF 3591 KB) Representative photomicrographs showing separate fluorescent dyes from merged confocal images of MCF-7 cells treated with DMSO (top panel) and TPLME (bottom panel): Hoechst 33342 (blue fluorescence) staining nuclei, SytoxTM (green fluorescence) staining nucleic acids of the dead cell, and red fluorochrome staining caspase-3 and caspase-7 (PNG 1.82 MB) Supplementary file2 (TIF 20815 KB) Representative photomicrographs showing separate fluorescent dyes from merged confocal images of MDA-MB-231 cells treated with DMSO (top panel) and TPLME (bottom panel): Hoechst 33342 (blue fluorescence) staining nuclei, SytoxTM (green fluorescence) staining nucleic acids of the dead cell, and red fluorochrome staining caspase-3 and caspase-7 (PNG 2.48 MB) Supplementary file3 (TIF 27899 KB) Representative photomicrographs showing separate fluorescent dyes from merged confocal images of MCF-7 cells treated with DMSO (top panel) and TPLME (bottom panel): Hoechst 33342 (blue fluorescence) staining nuclei, green fluorescence representing calcein/quencher cobalt, and red fluorochrome staining the mitochondria (PNG 1.66 MB) Supplementary file4 (TIF 17297 KB) Representative photomicrographs showing separate fluorescent dyes from merged confocal images of MDA-MB-231 cells treated with DMSO (top panel) and TPLME (bottom panel): Hoechst 33342 (blue fluorescence) staining nuclei, green fluorescence representing calcein/quencher cobalt, and red fluorochrome staining the mitochondria (PNG 2.38 MB) Supplementary file5 (TIF 26266 KB)