Introduction

Natural products have long served as a foundation for therapeutic development, with plant-derived compounds recognized for their immense pharmacological potential. To date, numerous plant species have been extensively researched for their biomedical properties, leading to the identification of many bioactive molecules, some of which have advanced to clinical trials. Lablab purpureus (hyacinth bean, LP) is one such plant with a long history of traditional use. Various parts of LP have been employed in Ayurvedic and Chinese traditional medicine to treat a wide range of ailments. Despite its rich history, the full spectrum of its bioactive potential remains underexplored in modern biomedical research. The seeds of this plant, in particular, are a source of phytochemicals, such as flavonoids, saponins, proteins, tannins, which have been linked to antioxidant, antiviral, anti-inflammatory and anticancer activities1,2. Genistein, Kievitone, Trigonelline, a mannose-specific lectin (FRIL), and other bioactive molecules from this plant have demonstrated cytotoxic effects against various cell lines through various mechanisms, with Genistein even being evaluated for its anticancer potential in clinical trials1.

Given the rise of cancer incidence globally, with lung cancer having the highest incidence and mortality followed by breast cancer in females in 2022, as well as the limitations in the existing treatment options, the identification of novel, plant-based anticancer agents has become a critical area of research3. Many plant-derived compounds target specific tumour cells while preserving healthy cells, rendering them potential candidates for treatment research. In this context, exploring underutilized plants with a history of medicinal use, like L. purpureus, whose phytochemicals have been evaluated for cytotoxicity against a variety of cancers, offers exciting potential for discovering new bioactive molecules. These could address the urgent need for innovative, plant-based anticancer therapies.

Since cancer is a highly diverse disease, enhancing the specificity and effectiveness of treatment requires focusing on the major molecular drivers of the condition. Therefore, finding new cancer therapeutic targets is equally important. while many current treatments target well-established targets, their efficacy can diminish due to resistance mechanisms and mutations in cancer cells. Differentially Expressed Genes (DEGs) are genes whose expression levels vary statistically significantly between cancerous and non-cancerous tissues or among different cancer types. They provide insights into the fundamental mechanisms of carcinogenesis, identification of important molecular pathways that propel cancer progression, and provision of prospective biomarkers for diagnosis, prognosis, or treatment. Since breast and lung cancers are some of the leading types of cancers in the world, identifying DEGs between breast and lung cancer also allows us to determine probable treatment targets that may be unique to each cancer type or shared across both.

The MCF-7 (breast) and A549 (lung) cancer cell lines serve as well-established models for studying hormone receptor-positive breast cancer and non-small cell lung cancer, respectively. Evaluating the cytotoxic effects of L. purpureus on these two cell lines allows for an assessment of its potential across diverse cancer types, potentially identifying shared molecular targets for broader therapeutic applications. This study evaluated the antioxidant, anti-inflammatory potential of L. purpureus seed extract. In addition, its preliminary anticancer activity, and flow cytometry-based detection of apoptosis and cell cycle analysis, against two cancer cell lines, MCF-7 and A549 was also ascertained. To the best of our knowledge, the analysis of DEGs between breast and lung cancer datasets have not been reported so far, and hence are a novelty of this study to ascertain common upregulated proteins as common cancer target proteins. Molecular docking, and phytochemical characterization further deepens our understanding of L. purpureus seed extracts bioactivity and also highlights their potential role in targeted cancer therapy.

Results

Extraction of L. purpureus phytochemicals provided good yield with methanol solvent

The phytochemicals of L. purpureus seeds were extracted using methanol and chloroform. The overall yield of phytochemicals gained from 60 g of seed powder in methanol and chloroform solvents was 3.75 g (6.25%) and 0.72 g (1.2%), respectively (Table 1).

Table 1 Percent yield of LP seed extract in methanol and chloroform solvents.
Full size table

Further, LPM and LPC were subjected to various qualitative biochemical evaluations to determine their phytochemical composition. Both extracts contained carbohydrates, phenols, proteins, and tannins, with LPM displaying additional phytochemicals such as glycosides and flavonoids. The list of phytochemicals present is recorded in the Table 2.

Table 2 The biochemical evaluation of phytochemicals present in LPM and LPC.
Full size table

L. purpureus methanol extract demonstrated strong concentration-dependent antioxidant and anti-inflammatory activities

The antioxidant activity of LPM and LPC was determined using: 1) DPPH free radical scavenging assay; 2) Hydroxyl radical scavenging assay; 3) FRAP assay; and 4) ABTS assay. For all assays, increasing concentrations of LPM and LPC (62.5 to 1000 μg/ml) were incubated with the respective reagents, and the resulting colour changes were quantified using a multimode plate reader at their respective wavelengths.

In the DPPH assay, both LPM and LPC exhibited DPPH radical scavenging activity. At the highest concentration of 1000 μg/ml, LPM exhibited 76.32 ± 0.75% DPPH radical scavenging activity, whereas LPC scavenged 57.27 ± 0.709% DPPH radicals. The DPPH radical scavenging activity of LPM was notably comparable to that of the positive control, which exhibited 92.66 ± 1.4% at 1000 μg/ml concentration. These findings suggested the strong potential of LPM to neutralize free radicals (Fig. 1a). The hydroxyl radical scavenging activity, an important indicator of antioxidant potential, was also concentration-dependent for LPM and LPC. At a concentration of 1000 μg/ml, LPM exhibited 72.15 ± 0.95% hydroxyl radical scavenging activity, while LPC exhibited an activity of 61.62 ± 1.32%. This potential of LPM was comparable to that of the positive control, Tocopherol, which exhibited 93.16 ± 0.87% scavenging at 1000 μg/ml (Fig. 1b).

Fig. 1
figure 1

LPM and LPC exhibited anti-oxidant and anti-inflammatory potential. Graphical representation of the (a) DPPH radical scavenging potential of LPM and LPC. (b) hydroxyl radical scavenging potential of LPM and LPC. (c) FRAP value of LPM and LPC. (d) % antiradical activity LPM and LPC. (e) % anti-inflammatory activity of LPM and LPC as observed in the protein denaturation assay (f) anti-inflammatory activity of LPM and LPC as demonstrated by inhibition of RBC lysis. n = 3, Mean ± SD, p-value of < 0.05 was significant, **p < 0.002, ***p < 0.001 and ns = not significant.

Full size image

The FRAP assay further assessed the ferric-reducing power of LPM and LPC by measuring the reduction of Fe3+ to Fe2+ ions. A ferrous sulphate standard curve was plotted (y = 0.0007x + 0.0517, R2 = 0.9865), and the FRAP values of LPM and LPC were extrapolated on the standard curve. The data revealed a concentration-dependent increase in the FRAP values of both extracts, indicating an enhanced ferric ion-reducing ability. The calculated FRAP value of LPM was 498.75 ± 1.02 µmol Fe2+ equivalents per gram of LPM and that of LPC was 321.74 ± 1.92 µmol Fe2+ equivalents per gram of LPC. This suggested the stronger electron-donating ability of LPM compared to LPC (Fig. 1c). The ABTS assay evaluated the ability of LPM and LPC to neutralize ABTS radicals. At the highest concentration, 75.14 ± 0.91% antiradical activity was demonstrated by LPM while LPC demonstrated 51.83 ± 1.02%. LPM effectively inhibited free radicals, demonstrating strong antioxidant activity comparable to the Trolox positive control, which exhibited 92.44 ± 0.91% scavenging activity at the same concentration. These results were indicative of both hydrophilic and lipophilic antioxidant activity, and demonstrate the higher inhibition of free radicals by LPM in comparison with LPC (Fig. 1d), highlighting the strong antioxidant activity of LPM.

The anti-inflammatory activities of LPM and LPC were further measured with the bovine serum albumin (BSA) denaturation and the RBC membrane stabilization assays. The extracts were tested at various concentrations, and their inhibitory effects were equated compared to the positive control, Aspirin. In the BSA denaturation assay, LPM demonstrated 70.12 ± 0.39% inhibition of protein denaturation at the highest concentration (1000 μg/ml), while the same for LPC was found to be 51.51 ± 1.65% (Fig. 1e). The anti-inflammatory effect was also ascertained by incubating ascending LPM and LPC concentrations with the RBC and measuring the quantity of haemoglobin in the medium. The powerful antioxidants were found to prevent the RBC lysis caused by membrane damage. Observations made indicated the membrane stabilizing efficacy of LPM and LPC in comparison to Aspirin. LPM and LPC were found to inhibit RBC membrane lysis by 76.27 ± 0.87% and 49.32 ± 0.89%, respectively (Fig. 1f). These results were strongly indicative of the higher anti-inflammatory potential of LPM as compared to LPC.

Total phenolic and flavonoid content

Gallic acid standard curve was constructed and the absorbance of the reaction mixture treated with test sample was plotted on the standard curve to measure the total phenolic acid content (TPC) in LPM. The standard curve was obtained (y = 0.007x + 0.0972, R2 = 0.9956) and TPC was determined to be 10.971 mg GAE/g of dry LPM extract. Similarly, a standard curve of catechin was constructed (y = 0.0013x + 0.0504, R2 = 0.998) in order to estimate the total flavonoid content. Flavonoid concentration was estimated as 3.01 mg CE/g of dry LPM extract.

L. purpureus extracts exhibited concentration-dependent cytotoxicity on MCF-7 and A549 cell lines, while exhibiting minimal toxicity towards HEK-293 cells

LPM extract, as demonstrated earlier as an antioxidant and anti-inflammatory agent, was assessed for its cytotoxicity potential using MTT assay. MCF-7 and A549 cell lines were treated with varying concentrations (12.5 to 250 µg/ml) of LPM extract for 24, 48 and 72 h incubation timepoints. The LPM extract showed time- and concentration-dependent cytotoxicity on MCF-7 and A549 cell lines. At the highest concentration (250 µg/ml), 60.01 ± 1.16% inhibition was observed at 24 h incubation, with a significant increase to 65.32 ± 0.38% inhibition at 48 h, and at 73.39 ± 1.25% inhibition at 72 h in LPM-treated MCF-7 cells. Inhibition of growth caused by LPM at higher concentrations (100 and 250 µg/ml) was comparable to that caused by 50 µg/ml concentration of positive control Cisplatin, which caused 82.21 ± 0.99% cell death upon incubation for 72 h. The calculated IC50 value of LPM decreased over time, indicating increased cytotoxicity with prolonged exposure. The IC50 values were calculated as 126 μg/ml at 24 h, and 40.91 μg/ml at 72 h (Fig. 2). Upon treatment of A549 cells for 24 h with LPM, at the highest concentration of 250 µg/ml, 65.39 ± 0.29% inhibition of cell viability was observed, increasing to 67.76 ± 1.55% at 48 h, and 77.55 ± 1.3% at 72 h. The IC50 values decreased with prolonged exposure from 96.67 µg/ml at 24 h, to 79.48 µg/ml at 48 h and 35.7 µg/ml at 72 h. The cytotoxic effects, on A549 cells, caused at higher concentrations were comparable to that of the positive control Doxorubicin (50 µg/ml), which caused 81.3 ± 0.08% cell death after 72 h of treatment (Fig. 3).

Fig. 2
figure 2

LPM exhibited cytotoxicity against MCF-7 cell lines. (a) Cytotoxicity of MCF-7 cell lines when treated with LPM for 24, 48 and 72 h. (b) IC50 of LPM calculated. n = 3, Mean ± SD, Two-way ANOVA and Tukey’s multiple comparisons test were performed to analyse data. P-value of < 0.005 was considered to be significant, **p < 0.002, ***p < 0.001 and ns = not significant. (c) Photomicrographs of untreated and treated MCF-7 cells at 24, 48 and 72 h. The arrows point towards the observed morphological changes in the cells.

Full size image
Fig. 3
figure 3

LPM displayed cytotoxicity against A549 cell lines: (a) Cytotoxicity of LPM on A549 cell lines when treated for 24, 48 and 72 h. (b) Calculated IC50 of LPM. n = 3, Mean ± SD, Two-way ANOVA and Tukey’s multiple comparisons test were performed to analyse data. p-value of < 0.005 was considered to be significant, *p < 0.033, **p < 0.002, ***p < 0.001 and ns = not significant. (c) Photomicrographs of treated and untreated A549 cells at 24 h, 48 h and 72 h. The arrows point towards the observed morphological changes in the cells.

Full size image

In addition, the cytotoxicity of LPM towards Human Embryonic Kidney (HEK)-293 normal cells was also evaluated. At the highest LPM concentration (250 μg/ml), cell viability remained high, with 87.01 ± 0.8% cell viability observed upon 24 h incubation, 84.54 ± 1% at 48 h incubation and 82.45 ± 0.65% viability observed upon 72 h of incubation in HEK-293 cells (Fig. 4). This indicated the minimal toxicity of LPM towards normal cells. Hence, LPM exhibited selective toxicity towards cancer cells.

Fig. 4
figure 4

LPM displayed minimal toxicity towards Hek-293 cell lines: (a) Cytotoxicity of LPM on HEK-293 cell lines when treated for 24, 48 and 72 h. (b) Calculated IC50 of LPM. n = 3, Mean ± SD, Two-way ANOVA and Tukey’s multiple comparisons test were performed to analyse data. p-value of < 0.005 was considered to be significant, *p < 0.033, **p < 0.002, ***p < 0.001 and ns = not significant. (c) Photomicrographs of treated and untreated HEK-293 cells.

Full size image

Flow cytometry-based detection of cell cycle and apoptosis validated the cytotoxic properties of LPM

Analysis of Cell cycle phases

To understand the effect of LPM on MCF-7 and A549 cell cycle phases, flow cytometry was performed. MCF-7 cells treated with LPM at the IC50 concentration for 72 h showed an increase in DNA content in the Sub G0/G1, S, and G2/M phases, with higher accumulation in the S phase as compared to untreated control cells (Fig. 5a and b). Similarly, in LPM-treated A549 cells, DNA content increased in the G2/M phase compared to untreated cells (Fig. 5a and c). These findings revealed that LPM induced arrest at the S phase in MCF-7 cells and at the G2/M phase in A549 cells.

Fig. 5
figure 5

LPM induced cell cycle arrest in MCF-7 and A549 cells. (a) Percentage distribution of MCF-7 and A549 cells across different cell cycle phases following treatment with LPM at IC50 concentrations (40.91 µg/ml for MCF-7 and 35.7 µg/ml for A549) for 72 h. The variation in percent of DNA content in multiple cell cycle phases in (b) MCF-7 cells and (c) A549 cells. The y-axis shows the % of cells in various phases of cell cycle.

Full size image

Apoptosis detection

To assess the inhibition of cell proliferation and induction of apoptosis, flow cytometric detection was carried out by staining the untreated and treated groups with Annexin V & Dead Cell Reagent. 72 h treated MCF-7 and A549 cells with LPM were used to carry out the analysis. The analysis demonstrated an increase in percentage of cells in early and late apoptotic stages, as well as necrotic stages, in both MCF-7 and A549 cell lines when treated with IC50 concentrations of LPM. Untreated MCF-7 control cells consisted of 92.75% live cells, 5.05% early apoptotic, 2.15% late apoptotic, and 0.05% necrotic. MCF-7 cells treated for 72h showed 16.33% of early apoptotic cells, 22.31% of late apoptotic cells, and 1.2% of necrotic cells. Untreated A549 control cells showed 92.6% live cells, 1.75% in early apoptotic stage, 5.05% in late apoptotic stage, and 0.6% cells in necrotic stage. A549 cells treated for 72h showed 3.23% of early apoptotic cells, 27.26% of late apoptotic, and 4.93% of necrotic cells. These findings indicate that LPM effectively induces apoptosis in both MCF-7 and A549 cells with similar efficacy (Fig. 6).

Fig. 6
figure 6

Apoptosis was detected in MCF-7 and A549 cells treated with LPM. Apoptosis analysis in untreated and LPM-treated cells was performed using Muse® Annexin V & Dead Cell Kit and Muse® Cell Analyzer. LPM-treated cells (at IC50 concentrations, i.e., 40.91 µg/ml for MCF-7, and 35.7 µg/ml respectively for A549 for 72 h) and untreated cells were analysed. (a) and (c) Untreated MCF-7 and A549 cells depicting majority of viable cells, respectively. (b) LPM-treated MCF-7 cells (d) LPM-treated A549 cells.

Full size image

GC–MS analysis

GC–MS analysis of the LPM extract was carried out and the eluted compounds were identified with the NIST library (2014). Upon comparing the mass spectra of the components (Supplementary Fig. 1) with the library, these compounds were tentatively characterized and identified. The list of predicted compounding to each retention time (RT) are provided in Supplementary Table 1.

Identification of DEGs and overlapping hub genes

Four microarray datasets acquired from the GEO database were examined for DEGs using the GEO2R tool. DEGs were filtered based on a LogFC and adjusted p-value to ensure statistical significance and biological relevance, and classified as noteworthy DEGs. A total of 1,020 genes were identified as common across the four datasets, as visualized in a Venn diagram highlighting their shared pattern (Fig. 7a). To further explore the biological significance of these overlapping genes, PPI was constructed with a confidence score of over 0.9 to ensure reliability. The resulting PPI network yielded 967 nodes (representing the proteins) and 2,909 edges (indicating the interactions between these proteins). The PPI network was subsequently introduced into the Cytoscape software for further analysis.

Fig. 7
figure 7

Identification and analysis of common DEGs and hub genes. (a) Venn diagram depicting the common DEGs across the four datasets filtered based on adjusted p-value of less than 0.05, and log fold change of less than 1. (b) The top five hub genes recognized with highest connectivity identified using the Cytoscape software. (c) Visualization of increased expression of the recognised hub genes in breast and lung cancers, versus normal tissues. ∗ p < 0.05. In the image, BRCA: breast cancer, LUAD: Lung adenocarcinoma; genes expressed in tumours are highlighted in orange, while their expression in normal tissues is depicted in grey. Y-axis represents the expression levels of genes.

Full size image

Key genes within the PPI network were identified using the Cytohubba plugin in Cytoscape software, which ranked genes based on their degree of connectivity. The hub genes, CDK1, CCNA2, CCNB1, CDC20, and BUB1B, were recognized (Fig. 7b). Further, Gepia analysis was conducted to examine the expression levels of the recognized hub genes between cancerous and normal tissues. The results indicated that all five identified hub genes were significantly overexpressed in both breast and lung cancer tissues compared to normal tissues, reinforcing their potential as key drivers of cancer progression and probable therapeutic targets (Fig. 7c).

Molecular docking and molecular dynamics simulations

In silico assessment of the identified possible LPM phytochemicals revealed effective molecular docking with the top five hub gene-encoded proteins identified by differential gene analysis of breast and lung cancer datasets. Detailed analysis of interactions with the cell division cycle protein 20 homolog (CDC20, PDB ID: 4N14) protein highlighted the most favourable interaction of 4-O-β-D-galactopyranosyl-β-D-Glucopyranose, forming seven hydrogen bonds with amino acids LEU449, LEU274, ARG316, TRP317, TRP234, and ASP191, achieving a glide gscore of -6.076 kcal/mol, suggesting strong binding affinity. The second-best interaction was detected with 2-(acetylamino)-2-deoxy-α-D-Galactopyranose, which formed six hydrogen bonds with the amino acids LEU449, ASN188, ASP191, ARG316, and VAL232, with a glide gscore of -5.94 kcal/mol. For the G2/mitotic-specific cyclin-B1 (CCNB1, PDB ID: 4Y72) protein, the best interaction was seen with N-Acetyl-D-glucosamine, which showed five hydrogen bonds involving SER227, ASP230, GLN184, and PHE338 amino acids with a glide gscore of -6.963 kcal/mol. 2-(acetylamino)-2-deoxy-α-D-Galactopyranose showed the best interaction with the Cyclin A2 (CCNA2, PDB ID: 3BHT) protein, glide gscore of -6.881 kcal/mol, involving two hydrogen bonds with hydrophobic residues TYR350, and TYR347, two bonds with positively charged ASN237, and SER340, and two bonds with negatively charged ASP240 and ASP343. d-Glycero-l-gluco-heptose also formed six hydrogen bonds, however, with a lower glide gscore of -5.326 kcal/mol. 2-(acetylamino)-2-deoxy-α-D-Galactopyranose showed the best interaction with the CDK1 (PDB ID: 4YC6) protein, with a glide gscore of -6.372 kcal/mol, and five hydrogen bonds connecting hydrophobic amino acids LEU83, ILE10, negatively charged ASP86, and positively charged LYS89. Furthermore, the ligand d-Glycero-l-gluco-heptose also showed the strongest interaction with BUB1B protein (PDB ID: 2WVI) with a glide gscore of -5.157 kcal/mol, forming six hydrogen bonds involving amino acids PRO90, TYR89, GLU86, and ARG130. The overall binding energy and bonding pattern, suggest the possible role of d-Glycero-l-gluco-heptose in modulating the BUB1B protein. The best-interacting ligands with each of the proteins have been displayed in Fig. 8.

Fig. 8
figure 8

2D interaction analysis of the best-interacting ligand–protein complexes. Interaction analysis of top-ranked ligands with the respective identified hub genes after molecular docking using Maestro suite revealed many hydrogen bonds. The hydrogen bonds are indicated in purple-coloured arrows.

Full size image

The best-docked complexes from the molecular docking results were selected for additional molecular dynamics simulations, each performed for 300 ns. In addition, to validate the docking protocol, prebound inhibitor molecules retrieved from the Protein Data Bank (PDB) database were subjected to docking using the same protocol. The resulting conformations were superimposed with the prebound molecule conformation and the Root Mean Square Deviation (RMSD) values were calculated to assess alignment accuracy. For the 4Y72 protein, the LZ9 prebound ligand showed an RMSD of 0.670 Å. Similarly, for the 3BHT protein, docking with the inhibitor Merlion resulted in an RMSD of 0.803 Å. The highest RMSD was seen with the 4N14 protein, when docked with Apcin inhibitor, which showed an RMSD value of 1.953 Å. All RMSD values were below the acceptable threshold of 2 Å, confirming that the ligands were well-aligned with their prebound conformations (Supplementary Fig. 2).

Cyclin A2-2-(acetylamino)-2-deoxy-α-D-Galactopyranose complex had a protein RMSD which was fairly steady from 0–75 ns trajectory at the range of 1.6–2.4 Å, and slightly increased to the range of 2.4–3.2 Å at the 100 ns trajectory, reached equilibrium again between the range 1.6–2.8 Å, reached the highest RMSD peak at 3.6 Å and attained equilibrium between the 200–300 ns trajectory between 2–2.8 Å (Fig. 9a). Analysis of CDC20 protein-4-O-β-D-galactopyranosyl-β-D-Glucopyranose complex displayed a protein RMSD range of 1–1.8 Å throughout the 300 ns trajectory. The ligand RMSD ranged between 1.5 and 2.7 Å, and was completely aligned with the RMSD of the protein. The complete alignment and the low RMSD values (less than 3 Å fluctuation) suggest the stability of this complex (Fig. 9b). Cyclin-B1-N-Acetyl-D-glucosamine complex had a stable protein RMSD in the range of 1–2 Å, and an aligned ligand RMSD, which was mostly stable in the range of 2–3 Å, with small variations (Fig. 9c). These indicate stable protein–ligand complex formations in the 3 selected complexes. The protein Root Mean Square Fluctuation (RMSF) of Cyclin A2 (Fig. 9d), CDC20 (Fig. 9e), and Cyclin-B1 (Fig. 9f) showed no fluctuations at either the N or the C terminal regions, although slight fluctuations within were observed. These indicate the stable nature of the protein structures, with small conformational changes post formation of the complex with the respective ligands.

Fig. 9
figure 9

The Protein–Ligand (P-L) RMSD and RMSF analysis of the best-interacting complexes over the 300 ns trajectory of molecular dynamics simulations. x-axis depicts the time (in ns) and y-axis represents the Protein, ligand RMSD and RMSF respectively. The analysis offered insights into the stability and adaptability of the studied protein–ligand complexes during the simulation.

Full size image

45.51% of the total secondary structural elements of the CDC20 (4N14) protein was found to exist in strands, and 0% helices, while conversely, 55.49% and 56.49% of the Cyclin A2 and Cyclin-B1 proteins were observed as helices, respectively throughout the 300 ns simulation duration. The ligand 2-(acetylamino)-2-deoxy-α-D-Galactopyranose exhibited an RMSD ranging from 0.5 to 2 Å (Fig. 10a), and a fairly stable rGyr between 3.18 and 3.3 Å. This ligand also showed a single intra-hydrogen bond, with a SASA of 0–10 Å2 over the first 200 ns trajectory, and raised to 20 Å2 till the 300 ns trajectory, and the PSA ranged between 216 to 232 Å2 (Fig. 10b). The ligand 4-O-β-D-galactopyranosyl-β-D-Glucopyranose was found to have an RMSD in the range 0.4–1 Å (Fig. 10c). The radius of gyration (rGyr), which is used to measure the inertia of the ligand, for the 4-O-β-D-galactopyranosyl-β-D-Glucopyranose ligand was observed to be 3.66–3.84 Å. This ligand was found to have one intra-hydrogen bond, with water solvent-accessible surface area (SASA) ranging between 50–125 Å2, while the polar surface area (PSA), which is made up of oxygen and nitrogen atoms, ranges in between 315 and 360Å2 (Fig. 10d). N-Acetyl-D-glucosamine RMSF was between 1 to 2 Å (Fig. 10e), with rGyr between 3.15–3.25 Å, also with 1 intra-hydrogen bond, fairly stable SASA of 0–15 Å2, which acquires a peak to 30 Å2 at 200 ns trajectory, and the PSA, although initially stable at 230 Å2, during the later stages of the simulation duration, fluctuated between 210–240 Å2 (Fig. 10f).

Fig. 10
figure 10

The ligand properties of the 3 ligands 4-O-β-D-galactopyranosyl-β-D-Glucopyranose, 2-(acetylamino)-2-deoxy-α-D-Galactopyranose, and N-Acetyl-D-glucosamine.

Full size image

Cyclin A2-2-(acetylamino)-2-deoxy-α-D-Galactopyranose complex was found to be made of interactions with residues ARG211 (54%), ASN237 (92%), ASP240 (98%), SER340 (56%) making up hydrogen bonds, and water bridges with PRO309 (36%), ILE311 (40%), SER340 (56%), ASP343 (73%), and TYR347 (44%) (Figs. 11a, 12a). CDC20-4-O-β-D-galactopyranosyl-β-D-Glucopyranose complex was found to have hydrogen bonds with residues LEU274 (87%), SER275 (87%), TRP276 (87%), ARG316 (41%), LEU449 (55%), and THR450 (46%), and to make salt-bridges with residues ASP191 (97%), TRP234 (97%), ARG316 (41%), and TRP363 (51%) for over 30% of the total simulation duration (Figs. 11b, 12b). The 4Y72-N-Acetyl-D-glucosamine complex showed multiple hydrogen bonds involving residues GLN184 (38%), ASP230 (99%, and 98%), MET335 (41%), and PHE338 (84%), and water bridges with GLN184 (31%), and ARG201 (44%), maintained for over 30% of total duration (Figs. 11c, 12c).

Fig. 11
figure 11

The P-L interactions which remained stable for over 30% of the total simulation duration (300 ns) across the studied complexes are depicted.

Full size image
Fig. 12
figure 12

Histogram of the observed key protein–ligand interactions across the studied complexes are depicted.

Full size image

Cyclin A2-2-(acetylamino)-2-deoxy-α-D-Galactopyranose complex, had an average MMGBSA dGBind of -45.95 ± 6.2 kcal/mol, -50.47 ± 5.11 kcal/mol for N14-4-O-β-D-galactopyranosyl-β-D-Glucopyranose complex, and -43.81 ± 4.34 kcal/mol for the cyclin B1-N-Acetyl-D-glucosamine complex during the 300 ns simulation (Fig. 13). These results confirm the strong binding of the ligands to the respective proteins active sites, resulting in the formation of stable complexes. Analysis of amino acid contributions revealed significant residues accountable for binding in each complex. Most noteworthy amino acid residue contributors to binding in the 3BHT-2-(acetylamino)-2-deoxy-α-D-Galactopyranose complex were ARG211 (-2.42 kcal/mol), ASN237 (-6.07 kcal/mol), ASP240 (-2.76 kcal/mol), SER340 (-1.94 kcal/mol), PRO309 (-0.63 kcal/mol), ILE311 (-3.12 kcal/mol), ASP343 (-1.82 kcal/mol), and TYR347 (-3.16 kcal/mol). Chief contributors in the 4N14-4-O-β-D-galactopyranosyl-β-D-Glucopyranose complex were LEU274 (-2.65 kcal/mol), SER275 (-6.13 kcal/mol), TRP276 (-5.05 kcal/mol), ARG316 (-9.72 kcal/mol), LEU449 (-1.48 kcal/mol), THR450 (-7.4 kcal/mol). In the 4Y72-N-Acetyl-D-glucosamine complex, key contributors were of GLN184 (-3.06 kcal/mol), ASP230 (-3.05 kcal/mol), and moderate contributors were MET335 (-2.23 kcal/mol), PHE338 (-2.05 kcal/mol). These results indicate the formation of stable complexes between the respective ligands and proteins.

Fig. 13
figure 13

The binding free energies of the studied protein–ligand complexes throughout the 300 ns trajectory.

Full size image

ADME prediction

Lipinski’s rule of five is a generally recognized guideline for evaluating the drug-likeness of molecules. As per the rule, a molecule can be considered drug-like if it adheres to the criteria of QPlogPo/w < 5, molecular weight less than 500 Da, number of donor hydrogens ≤ 5, and number of acceptor hydrogens ≤ 10. In this study, Qikprop predictions of the identified LP phytochemicals indicated several compounds adhering to this rule, while molecules such as Cyclohexanol, 1R-4cis-acetamido-5,6cis-epoxy-2trans,3cis-dimethoxy- (donor Hydrogen bond: 7.917), d-Glycero-l-gluco-heptose (acceptor hydrogen bond: 11), exceeded the number of hydrogen bonding thresholds. Notably, several molecules displayed predicted partition coefficient between octanol and water values greater than 5, thereby violating the Lipinski’s criteria. These included: n-Hexadecanoic acid, Palmitoyl chloride, Dibenz[a,h]anthracene, 1,2,3,4-tetrahydro-, Undecane, 4,5-dimethyl-, Octane, 3-ethyl-, Oleic Acid, 4-Butylbenzoic acid, 3,5-dimethylphenyl ester, 8,11,14-Eicosatrienoic acid, (Z,Z,Z)-, 7,10-Octadecadienoic acid, methyl ester, Isopropyl palmitate, Eicosanoic acid, Octadecanoic acid, 2,3-dihydroxypropyl ester, Palmitic anhydride, 7,10-Octadecadienoic acid, methyl ester, Cyclopropaneoctanoic acid, 2-[[2-[(2-ethylcyclopropyl)methyl]cyclopropyl]methyl]-, methyl ester, 2-Heptadecanol, acetate, Octadecanoic acid, 2-(dimethylamino)ethyl ester, Dimethylaminoethyl palmitate, Butyl 9,12-octadecadienoate, and n-Propyl 9,12-octadecadienoate. In addition, Hexadecanoic acid, 1-(hydroxymethyl)-1,2-ethanediyl ester, and l-(+)-Ascorbic acid 2,6-dihexadecanoate both exhibited molecular weights higher than 500 Da and high QPlogPo/w values of 9.794 and 9.653, respectively, further breaching the drug-likeness of these molecules (Supplementary Table 2). 2-(acetylamino)-2-deoxy-α-D-Galactopyranose was predicted to have high intestinal absorption, whereas molecules 4-O-β-D-Galactopyranose-β-D-Glucopyranose, and N-Acetyl-D-glucosamine showed low gastrointestinal absorption capacity (Fig. 14).

Fig. 14
figure 14

BOILED-EGG plot of the top three ligands. The molecules falling in the white part of the egg plot are predicted to have high intestinal absorption capacity, while those in the grey region indicate low gastrointestinal absorption. In the figure, Molecule 1 corresponds to α-D-Galactopyranose, 2-(acetylamino)-2-deoxy-; Molecule 2 corresponds to β-D-Glucopyranose, 4-O-β-D-galactopyranosyl-; and Molecule 3 corresponds to N-Acetyl-D-glucosamine.

Full size image

Discussion

Current chemotherapeutic options for breast and lung cancers are failing due to limited efficacy, drug resistance and toxicity4. As a result, natural products are being assessed for the presence of anticancer compounds to treat cancers. Numerous natural compounds, such as genistein, have been shown to act on breast cancer via various pathways5. Likewise, this study evaluated the anticancer activity of LP seed phytochemicals.

Previous studies have reported the presence of carbohydrates, proteins, phenols, flavonoids and tannins in LP seed extracts, akin to the results obtained in the current study6,7,8. The total phenolic content of LP has been reported to differ considerably across diverse research studies and in various genotypes, with one study reporting TPC ranging from 5.2 to 39.3 mg GAE/100 g of dry extract and another noting a lower TPC of 2.05 mM GAE/g6,9. Conversely, a higher average of 13.47 mg GAE/g was also documented10. Another study of the methanol extract of Bangladeshi white lablab beans exhibited 42.55 mg/g of quercetin equivalents of TFC11. However, variations in TPC and TFC across various studies highlight the effect of geographical conditions and genotypes. TPC of LPM was calculated to be 10.971 mg GAE/g of dry extract and TFC was 3.01 mg CE/g of dry extract, falling within the previously reported ranges.

Potent antioxidant capacity has been reported in LP extracts, as measured by DPPH, and hydroxyl radical scavenging assays, in various studies. Dry-heated seeds exhibited strongest DPPH free radical scavenging ability (IC50 of 2.5 μg/ml), followed by raw samples (4.9 μg/ml) and pressure-cooked samples (5.7 μg/ml). The raw, dry heated and pressure-cooked seed extracts further exhibited 94% to 99% hydroxyl radical scavenging effect at 1 mg/ml concentration. At 500 µg/ml, FRAP values were 1.04 µM Fe2+/mg for acetone extract, 2.36 µM Fe2+/mg for ethyl acetate extract, 4.72 µM Fe2+/mg for 70% ethanol extract, and 6.12 µM Fe2+/mg for water extract12. Similar to the formerly reported results, this study found 76.32% and 57.27% DPPH free radical scavenging potential, and 72.15% and 61.62% hydroxyl radical scavenging potential for 1 mg/ml concentration of methanol and chloroform extracts, respectively. Likewise, the ABTS radical scavenging potential of these extracts at 1 mg/ml concentration was observed to be 75.14% and 51.83% respectively. FRAP values of 498.75 and 321.74 µmol Fe2+ equivalents per gram of LPM and LPC, respectively were also detected for the same at the highest concentration of 1 mg/ml. The LPM extract exhibited superior free radical, hydroxyl radical, ABTS radical scavenging activities as well as ferric oxide reducing potential when compared to LPC. Crude lectin extracts from LP showed varying concentrations required for hemagglutination across different blood groups13. Further, raw seed samples exhibited 17.28% antihaemolytic activity against erythrocytes, while the dry-heated samples demonstrated an activity of 13.81%14. Ethanol extracts were found to exhibit moderate thrombolytic activity ranging from 37.25% to 2.40%, while the methanolic extract was found to significantly protect the lysis of erythrocyte membranes, causing 61.48% inhibition15,17 Similarly, this study identified the stronger potential of LPM in comparison to LPC, although both extracts exhibited concentration-dependent anti-inflammatory activity, as evaluated using the RBC membrane stabilization and protein denaturation assays.

Pepsin hydrolysate of LP has been reported to exhibit the lowest IC50 values of 119.60 μg/mL against A549 cancer cells and 9.8 μg/mL against MCF-7 cancer cells, indicating higher effectiveness compared to the protein isolate16. Pepsin hydrolysates were found to induce early apoptosis in A549 cells at a rate of 41.5%, compared to 26.4% for the isolate, with caspase activation confirming these apoptotic effects. In a further investigation of apoptotic effects, 85.4% of A549 and 89.6% of MCF-7 cells underwent apoptosis16. Similarly in this study, the LPM extract also exhibited the time- and dose-dependent cytotoxicity towards MCF-7 (IC50 value of 40.91 μg/ml) and A549 cell lines (IC50 value of 35.7 μg/ml), which warrants further validation to validate its mechanism of action, ascertain the active phytochemicals essential for the exhibited cytotoxicity, and assess possible therapeutic value in preclinical studies. The extract was found to cause apoptosis as ascertained by flow cytometry-based assays. Cell cycle arrest in MCF-7 cells was found to be occurring at the S phase, and in A549 cells at the G2/M phase when these cells were treated with IC50 concentrations of LPM. In comparison to untreated controls, the LPM-treated MCF-7 and A549 cells showed higher number of cells in the early and late apoptotic stages.

The common differentially expressed genes as identified by a previous study in breast, lung, and prostate cancer include Acyl-CoA synthetase short-chain family member 3, angiopoietin-1, Aldehyde oxidase 1, Baculoviral IAP repeat-containing protein 5, Caveolin 1 and 2, Coiled-Coil Domain Containing (CCDC) 69, CCDC85A, CUGBP Elav-Like Family Member 2, Complement Factor D, clusterin, and dermatopontin17. These DEGs could serve as biomarkers for advancing new diagnostic and therapeutic solutions, which is crucial given the high rates of cancer-related mortality associated with these types of cancer. This study identified Cyclin-dependent kinase 1 (CDK1), Cyclin A2 (CCNA2), Cyclin B1 (CCNB1), cell division cycle protein 20 (CDC20), and BUB1 Mitotic Checkpoint Serine/Threonine Kinase B (BUB1B) as the top five hub genes, as common DEGs in the selected 4 breast and lung cancer microarray datasets. All these genes are involved in cell division and cell cycle regulation, represent promising therapeutic targets18,19,20,21,22. Several studies have studied plant-based novel drugs against these targets. Hence, the phytochemicals identified in the seed crude extract were subjected to molecular docking against these targets.

Molecular docking analysis, performed using Maestro Schrodinger, revealed successful docking of all ligands with the proteins, with several ligand–protein complexes displaying strong hydrogen bonding interactions and low glide gscores. The best-performing complexes were subjected to molecular dynamics simulations, using Desmond Schrodinger, to explore atomic variations, which revealed stable protein–ligand complexes. Of the studied complexes, Cell division cycle protein 20 homolog protein- 4-O-β-D-galactopyranosyl-β-D-Glucopyranose complex was the most stable complex. It exhibited the lowest protein–ligand RMSD, and the RMSF analysis suggested that the ligand fits well at the CDC20 protein active site. Further, the radius of gyration indicated the compact nature of this complex, and the secondary structure analysis and snapshots generated confirm that the 4-O-β-D-galactopyranosyl-β-D-Glucopyranose ligand remained bound at the active site of the protein throughout the simulation. Both complexes [Cyclin A2 -2-(acetylamino)-2-deoxy-α-D-Galactopyranose and G2/mitotic-specific cyclin-B1-N-Acetyl-D-glucosamine], displayed fairly stable and equilibrated protein–ligand RMSD and RMSF. The secondary structural elements of both proteins were found to exist as helices. These complexes also exhibited stable bonding interactions. MMGBSA snapshots generated confirmed stable ligand binding in the respective protein active sites. The residues critical for ligand binding were identified as key contributors in the complexes studied. For the Cell division cycle protein 20 homolog-4-O-β-D-galactopyranosyl-β-D-Glucopyranose complex, the primary contributors were LEU274, SER275, TRP276, ARG316, LEU449, and THR450. In the cyclin B1-N-Acetyl-D-glucosamine complex, key contributors were GLN184 and ASP230, and moderate contributors were MET335 and PHE338. The most noteworthy contributors to binding in the Cyclin A2-2-(acetylamino)-2-deoxy-α-D-galactopyranose complex were ARG211, ASN237, ASP240, SER340, PRO309, ILE311, ASP343, and TYR347. ADME property prediction of all the ligands revealed the drug-likeness of the three ligands (4-O-β-D-galactopyranosyl-β-D-Glucopyranose, 2-(acetylamino)-2-deoxy-α-D-Galactopyranose and N-Acetyl-D-glucosamine), which obeyed all of Lipinski’s rules. However, while 4-O-β-D-Galactopyranose-β-D-Glucopyranose and N-Acetyl-D-glucosamine were found to have low gastrointestinal absorption capacity, 2-(acetylamino)-2-deoxy-α-D-Galactopyranose was predicted to be absorbable via the intestine.

This study presents intriguing new perspectives; however, several limitations must be acknowledged. Standardized extraction and characterization techniques are essential, as variations in LP phytochemical composition due to environmental and genetic factors may undermine repeatability. Although in vitro assays demonstrated cytotoxicity and induction of cell death, validation in in vivo models is essential to confirm therapeutic efficacy and assess pharmacokinetics and toxicity. Furthermore, although molecular docking provides insights into potential ligand–protein interactions, experimental validation is essential to confirm binding efficacy and elucidate the precise mechanisms of action. Future research must focus on isolating and characterizing active compounds, evaluating their efficacy in preclinical models, and exploring potential synergistic interactions with existing chemotherapeutics to enhance therapeutic strategies. These could help in the development of non-toxic, novel, anti-cancer drug development.

Conclusion

The in vitro analysis of Lablab purpureus phytochemicals using antioxidant, anti-inflammatory and cytotoxicity assays demonstrated their strong biomedical potential. LPM crude extract exhibited potent cytotoxicity against MCF-7 and A549 cells in vitro by means of apoptosis and cell cycle arrest at the S and G2/M phases, respectively. Additionally, in silico analysis of the identified phytochemicals ascertained their action against the identified differentially expressed proteins in breast and lung cancers. These in silico results must be validated with comprehensive in vitro analysis and in vivo validation to ascertain the mechanistic basis of cytotoxicity of LPM phytochemicals against MCF-7 and A549 cell lines. This could lead to the development of novel plant-based, non-toxic cancer therapeutic agents.

Materials and methods

Extraction

Lablab purpureus pods were purchased at the local market in Mysore district, Karnataka. The seeds were separated from the pods by hand, and subsequently identified and validated at the Indian Council of Agricultural Research – Indian Institute of Horticulture Research, Bangalore, India. After washing with tap water, the seeds were dried in shade for ten days at room temperature, powdered using mixer grinder, and extracted with methanol and chloroform. For the extraction, 60 g each of seed powder were processed with 300 ml of methanol, and chloroform solvents, to extract phytochemicals of varying polarities, for 4 h at 50 °C using the Soxhlet method6. Whatman No. 1 filter paper was used for the filtration, and the filtrate was subjected to rotary evaporation (water bath at 40 °C, rotational speed of 100 rpm, under reduced pressure) to concentrate the extracted phytochemicals to each solvent, and were subsequently air-dried. The respective extracts were labelled LPM (L. purpureus Methanol extract), and LPC (L. purpureus chloroform extract). The extracts were stored at 4 °C.

Biochemical evaluation

The LPM and LPC extracts were subjected to various qualitative assays to understand their possible phytochemical makeup, as follows:


Molisch’s test for carbohydrates: a few drops of Molisch’s reagent were added to extracts (0.2 ml), followed by the addition of sulphuric acid (0.2 ml) along the side of the test tube23.


Braymer’s test for tannins: 0.2 ml of extracts were incubated with 2 ml of water for 10 m in a water bath. After filtration, 10% ferric chloride solution was added to the mixture24.


Test for glycosides: one drop of 5% ferric chloride and glacial acetic acid (3 ml) were added to the extracts (2 ml), followed by the addition of concentrated sulphuric acid (0.5 ml) along the sides of the test tube25.


Test for flavonoids: 1.5 ml of the extracts were warmed, and magnesium was mixed, following which 5–6 drops of concentrated hydrochloric acid were added to this solution26.


Wagner’s test for alkaloids: A solution containing 2 g of potassium iodide and 1.27 g of iodine in distilled water was made up to 100 ml. A few drops of the resultant solution were added to the samples27.


Frothing test for saponins: extracts (1 ml) were mixed with 3 ml distilled water in test tubes. These tubes were securely stoppered, and vigorously shaken for several min and left undisturbed for 30 min to observe froth formation28.


Test for reducing sugars: 5 ml of Fehling’s solutions 1 and 2 were added to 2 ml of the extracts, and the mixture was boiled29.


Biuret test for proteins: a few drops of both 3% copper sulphate and sodium hydroxide (10%) were added to 1 ml of the extracts30.


Folin-Ciocalteu test for phenols: 0.2 ml of the extracts were mixed with 4 ml of 50% Folin-Ciocalteu reagent and a few drops of 10% solution of sodium carbonate. The mixture was incubated at room temperature for 30 m31.

Evaluation of antioxidant and anti-inflammatory effects

2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging assay

Antioxidant efficacy of LPM and LPC were evaluated using DPPH radical scavenging assay32. The antiradical activity was quantified by the reduction in absorbance at 517 nm of a coloured methanolic DPPH solution, induced by the sample. Various concentrations (62.5 μg/ml- 1000 μg/ml) of the methanol extracts were prepared from a stock solution (10 mg/10 ml). 1.6 ml of each sample concentrations were incubated with 2.4 ml of DPPH solution (0.1 mM in methanol) for 30 min at room temperature in the absence of light, following which the absorbance was measured at 517 nm. Ascorbic acid served as the positive control and all the samples were analysed in triplicate. The percentage of DPHH radical scavenging activity was calculated using the following formula:

$${\text{DPPH}}\;{\text{Radical}}\;{\text{Scavenging}}\;{\text{activity}}\left( \% \right) = \left[ {\left( {{\text{A}}0 - {\text{A}}1} \right)/{\text{A}}0} \right]*100$$

where A0 is the absorbance of the control, and A1 is the absorbance of the extracts/standard.

Hydroxyl radical scavenging assay

The hydroxyl radical scavenging activity was evaluated by comparing the amount of hydroxyl radical produced by the Fe3+/ascorbate/EDTA/H2O2 system between deoxyribose and the extract33. The reaction mixture comprising 3000 μM deoxyribose in phosphate buffer (pH 7.4), 100 μM ferric chloride, 100 μM EDTA, 100 μM ascorbic acid, 1.5 mM hydrogen peroxide, and the extracts at varying concentrations (ranging from 62.5 μg/ml- 1000 μg/ml) were incubated for 1 h at 37 °C. Equal volumes of 1% trichloroacetic acid and 2.8% thiobarbituric acid were subsequently added and incubated at 100 °C for 20 min. After cooling, the absorbance was quantified at 532 nm relative to a blank containing deoxyribose and buffer. Tocopherol served as the reference standard, and all the samples were tested in triplicate. The inhibition of deoxyribose degradation in percent was calculated using the following formula:

$${\text{Radical}}\;{\text{Scavenging}}\;{\text{activity}} \left( \% \right) = \left[ {\left( {{\text{A}}0 - {\text{A}}1} \right)/{\text{A}}0} \right]*100$$

where A0 is the absorbance of the control, and A1 is the absorbance of the extracts/standard.

ABTS radical cation assay

The free radical scavenging activity of samples was quantified using the ABTS radical cation assay34. A solution of 7 mM ABTS in water was mixed with 2.45 mM potassium persulfate in equal quantities and kept at room temperature, in the absence of light to produce the ABTS cation radicals. The solution, after 12h, was subsequently diluted with methanol to achieve the absorbance of 0.7 at 734 nm. Next, 10 µl of the sample was incubated with 200 µl of ABTS working solution for 30 min in darkness. Trolox served as the standard, while distilled water served as the blank. Absorbance was quantified at 734 nm. All the samples underwent testing in triplicate and the % scavenging activity was calculated using:

$${\text{ABTS}}\;{\text{Radical}}\;{\text{Scavenging}}\;{\text{activity}} \left( \% \right) = \left[ {\left( {{\text{A}}0 - {\text{A}}1} \right)/{\text{A}}0} \right]*100$$

where A0 is the absorbance of the control, and A1 is the absorbance of the extracts/standard.

Ferric reducing antioxidant power (FRAP) assay

FRAP assay was conducted according to the protocol established by Benzie and Strain35. A freshly prepared FRAP reagent, containing 2.5 ml of 10 mM tripyridyl-s-triazine (TPTZ) in 40 mM hydrochloric acid and heated at 50 °C for 5 min, 2.5 ml of 20 mM FeCl3 solution, and 25 ml of 300 mM acetate buffer, at pH 3.6, was utilized in the assay. 190 μl of the reagent was incubated with LPM and LPC extracts or ferrous sulphate standards at various concentrations in a darkened room for 30 min. The absorbance was recorded at 593nm and expressed as µmol Fe2+ equivalents per gram of sample.

RBC membrane stabilization assay

The anti-inflammatory activity of LPM and LPC was determined using the RBC membrane stabilization assay (2019)36. Blood (5ml) was collected from the JSS Medical College and Hospital, JSSAHER, in Mysore, and centrifuged for 5 min at 2000 rpm to separate the RBC pellet. Re-suspending the pellet after washing it with 4 ml of 0.9% NaCl solution produced a 10% RBC suspension. This suspension underwent incubation for 1 h at 37 °Cwith 500 µl of various test sample concentrations and 1 ml of 0.015 M phosphate buffer. After centrifuging the samples, the absorbance of supernatant was assessed at 560 nm, with aspirin serving as a positive control and RBC suspension alone as the lysis control. The percentage inhibition of RBC lysis was calculated using:

$$\% {\text{ inhibition}}\;{\text{of}}\;{\text{RBC}}\;{\text{lysis}} = \left[ {\left( {{\text{A}}0 - {\text{A}}1} \right)/{\text{A}}0} \right]*100$$

where A0 is the absorbance of 10% RBC suspension incubated with distilled water, and A1 is the absorbance of the 10% RBC suspension incubated with extracts/standard.

Protein denaturation assay

Using the protein denaturation assay, the anti-inflammatory activity of the extracts was assessed37. In 1.5 ml of centrifuge tubes, 1ml of Phosphate buffered saline and 50µl of bovine serum albumin were mixed with different concentrations of samples and the standard solution. The mixture was incubated for 15 min at room temperature. Denaturation was induced by keeping it at 70 °C in a hot water bath for 15 min. The absorbance was quantified at 660 nm. Aspirin served as a positive control. The inhibition of protein denaturation was measured using:

$$\% {\text{Denaturation}}\;{\text{activity}} = \left[ {\left( {{\text{A}}0 - {\text{A}}1} \right)/{\text{A}}0} \right]* 100$$

where A0 and A1 are the absorbance of control and sample, respectively.

Assessment of total phenolic and total flavonoid content (TPC and TFC)

TPC of LPM was assessed using the Folin-Ciocalteu (FC) method standardized to microtiter plate volumes38. 150 µl of 10% FC reagent was added to 30 µl of the plant sample dissolved in 95% methanol. After adding 120 µl of 6% sodium bicarbonate solution, this mixture was incubated for 1 h in the dark. The absorbance was recorded at 760 nm. Gallic acid was used as the standard and the standard curve obtained was expressed as mg gallic acid equivalent (GAE) per gram of dry extract.

TFC was quantified by using the aluminium chloride method39. Extract (25 µl) was diluted with 125 µl of water, and 7.5 µl of 5% sodium nitrite and incubated for 5 min. 15 µl of 10% aluminium chloride was incorporated into this mixture and incubated for 6 min. Subsequently, 50 µl of 1 M sodium hydroxide and 27.5 µl of water were added, and the absorbance was noted at 510 nm. Catechin served as the standard, and all samples were tested in triplicate. TFC was ascertained from the prepared curve and represented as mg Catechin equivalent (CE) per gram of dry extract.

MTT assay

The effects of LPM on the viability of MCF-7 and A549 cell lines (procured from NCCS, Pune) were evaluated using MTT assay (2021)40. A total of 1 × 104 MCF-7, A549 and HEK-293 cells per well were seeded separately in 96-well plates and incubated at 37 °C in a humidified environment with 5% CO₂. After 48 h, cells were treated with varying concentrations of LPM (12.5 – 250 µg/ml) and incubated for 24 h. The LPM-containing media was then discarded, washed with PBS, followed by addition of 200 µl of MTT reagent (Sigma-Aldrich, M-5655) before incubation for 4 h to facilitate formazan crystal formation. Subsequently, the MTT reagent was carefully removed, and 200 µl of DMSO was added to solubilize the formed formazan crystals. Absorbance was quantified at 570 nm using a microplate reader. Cisplatin and Doxorubicin served as positive controls for MCF-7 and A549 cells, respectively. The samples were tested in triplicate for 48 h and 72 h treatments as well. The percentage inhibition was determined using the formula:

$$\% {\text{Inhibition}} = \left[ {\left( {{\text{A}}0 - {\text{A}}1} \right)/{\text{A}}0} \right]*100$$

where A0 and A1 are the absorbance of control and sample, respectively.

Flow cytometry-based analysis of cell cycle and apoptosis

Cell cycle analysis

To determine the effect of LPM on MCF-7 and A549 cell cycle, analysis was performed using flow cytometry4. MCF-7 and A549 cells were cultured and treated with the IC50 concentrations of LPM (40.91 µg/ml for MCF-7 and 35.7 µg/ml for A549, calculated based on the MTT assay with treatment for 72 h) and incubated for 72 h in a humidified 5% CO2 incubator. Untreated wells were maintained as controls. Post-incubation, the cells were centrifuged at 300 rpm for 5 min. The supernatant was discarded and the cell pellet was washed with PBS and re-centrifuged at 300 rpm for 5 min, pellets were washed with PBS, and centrifuged again. The pellets were resuspended in residual PBS, and fixed with 1 ml of ice-cold 70% ethanol by storing overnight at -20 °C. 250 µl of cell cycle reagent (Muse®, MCH100106) was added to the pellets. The pellets were incubated in the dark for 30 min and analysed for cell cycle distribution using a flow cytometer (Muse® Cell Analyzer). Gating was performed using untreated cells as controls.

Apoptosis analysis

MCF-7 and A549 cell lines were cultured according to the standard protocol described earlier. Cells were treated with the IC50 concentrations of the LPM (40.91 µg/ml for MCF-7 and 35.7 µg/ml for A549), and incubated for 72 h at 37 °C in a humidified 5% CO2 incubator. Untreated cells served as controls. After incubation, the cells were centrifuged for 3 min at 1500 rpm. The pellets were retrieved to which 100 μL of the Annexin V & Dead Cell Reagent (Muse®, MCH100105) were added, mixed thoroughly, and incubated at room temperature in the dark for 20 min. The cells were then analysed using a flow cytometer (Muse® Cell Analyzer) and processed with Muse FCS 3.0 software. Gating was performed against untreated control cells41.

Gas chromatography-mass spectrometry (GC–MS) analysis of the LPM crude extract

The Perkin Elmer GC Clarus 680 system fitted with a fused silica column, packed with Elite-5MS (5% biphenyl 95% dimethylpolysiloxane, 30 m × 0.25 mm ID × 250 μm df) was used to perform the GC–MS analysis of L. purpureus seed methanol extract. Helium was used as a carrier gas to separate the components at a steady flow rate of 2 ml/min. The injector temperature was set to 220 °C for this analysis. After injecting 1 μl of LPM into the device, the oven temperature was set for a discontinuous adjustment of temperature to 50 °C for 2 min, eventually increasing to 150 °C at a rate of 15 °C/min, 150 °C for two minutes, and finally increasing to 250 °C at a rate of 30 °C/min, which was sustained for eight minutes. The mass detector parameters used were 230 °C for the ion source, 250 °C for the inlet line, 70 eV for the ionization mode electron impact, 0.2 s for the scan time, and a scan interval of 0.1 s. The component spectrums were compared to the database of standards available in the GC–MS NIST (2014) library.

Identification of Differentially expressed genes, PPI network construction and hub gene identification

The expression profiles of the GSE48984, GSE45827, GSE19804, GSE18842 were downloaded from the GEO database (https://www.ncbi.nlm.nih.gov/geo). The details of the datasets are given in the Table 3.

Table 3 Details of the breast and lung cancer datasets retrieved from the GEO database.
Full size table

The GEO2R tool (https://www.ncbi.nlm.nih.gov/geo/geo2r/) available on the NCBI GEO database was employed to evaluate each of the profiles, to determine the differentially expressed genes (DEGs) to distinguish diseased and normal/control samples. These genes were further filtered with an adjusted p-value of less than 0.05, and log fold change of less than 1. The identified DEGs were then added to the Venny 2.1 free web server (https://bioinfogp.cnb.csic.es/tools/venny/) to determine the overlapping DEGs between the series. The list of overlapping DEGs was then imported into the online software interface, the Search Tool for the Retrieval of Interacting Genes (STRING) software (https://string-db.org/) to construct a protein–protein interaction (PPI) network for the cross-disease analysis. The average node degree parameter was then determined and the PPI network topology was visualized using Cytoscape program (V 3.10.2). Hub genes were defined as those with a node degree of 10 or higher. GEPIA (http://gepia.cancer-pku.cn) was used to analyse the expression of DEGs in cancer and normal tissues41,42.

Molecular docking and dynamics simulations of phytochemical-protein interactions

Molecules identified in the crude extract by GC–MS, were subjected to molecular docking against the identified hub gene-encoded proteins, BUB1B (PDB: 2WVI, Chain A, sequence length of 164, and resolution of 1.8 Å), CDC20 (PDB: 4N14, chain A, sequence length of 314, and resolution of 2.10 Å), CCNB1 (PDB: 4Y72, chain B, sequence length of 273, resolution of 2.30 Å), CCNA2 (PDB: 3BHT, chain B, sequence length of 262, resolution of 2.00 Å), and CDK1 (PDB: 4YC6, chain A, sequence length of 297, resolution of 2.60 Å). The protein preparation panel was used to minimise protein energies. The Prime module filled in the lacking chains and loops while the pH was set to 7.5. The protein regulatory components were removed. At pH 7.5 hydrogen atoms were added and the suitable protonation state was assigned using the Epik module. Using PROPKA pH 7.5, the OPLS4 force field was applied to minimise atoms beyond 0.3 Å RMSD, water molecules at a distance of over 5 Å were also eliminated, and partial atomic charges were assigned. Additionally, the Sitemap tool allowed receptor grid generation around the pre-bound ligands or following active site generation using the Sitemap tool. The default settings were then used to create a receptor grid removing partial charge, scaling the van der Waals radius, and scaling factor 1.0. The produced grid sizes were: 4Y72: 72 Å × 72 Å × 64 Å, 4N14: 88 Å × 88 Å × 80, 2WVI: 64 Å × 64 Å × 64 Å, 3BHT: 56 Å × 56 Å × 56 Å, and 4YC6: 72 Å × 72 Å × 72 Å.

Further, the ligands were docked against the proteins with the Schrodinger glide suite (v14.1)43,44,45,46,47,48,49,50,51. In addition, the native ligands were separated, and subjected to the same for validation of the docking protocol. Protein–ligand complexes with the strongest interactions and lowest Glide gscores were further validated with molecular dynamics simulations (MDS). The protein–ligand complex was submerged in an orthorhombic simulation box of 10 × 10 × 10 Å of boundary condition, filled with a TIP3P solvent system, and the OPLS4 forcefield was applied using the system builder program. The complexes were neutralized with the addition of salt ions (sodium and Chlorine) in order to balance the overall charge of the system. The temperature (K) and pressure were held constant throughout the experiment at 300 K and 1.01325 bar, respectively, and the total duration of simulation was set for 300 ns at the recording energy interval of 12 ps simulation run.

Furthermore, thermal MM/GBSA was performed using a script to calculate the binding free energies of the analysed complexes. The complete trajectory was segmented, with snapshots obtained every 10 trajectories, utilising the OPLS4 force field and the VSGB 2.1 solvent model to calculate the binding energies of ligands. Per-residue decomposition was performed to quantitatively evaluate the energetic contributions of each amino acid, including both backbone and sidechain analyses. The analysis of binding energy decomposition was crucial for delineating the contribution of each residue to the examined ligands.

ADME prediction and boiled-egg plot

The identified phytochemicals were further subjected to ADME property prediction using the Qikprop application in Schrodinger. Additionally, blood–brain barrier permeability and gastrointestinal absorption capacity were predicted and visualized using the SwissADME free web server (http://www.swissadme.ch/)52.

Statistical analysis

The results were expressed as mean ± standard deviation, and were performed in triplicate. The GraphPad Prism statistical software (v10.1.2) was used to perform the Anova tests. A p-value of < 0.05 was considered statistically significant.