Pro-Apoptotic and Anti-EMT Activity of Wild Ginseng Adventitious Root Extract in MDA-MB-231 TNBC Cells: Association with GSK-3β/β-Catenin Signaling

Abstract

Background/Objectives: Triple-negative breast cancer (TNBC) lacks targeted therapies and has a poor prognosis. Wild ginseng (Panax ginseng) is traditionally valued for its medicinal properties, but its scarcity limits therapeutic application. Adventitious root culture technology provides a sustainable source of wild ginseng-derived bioactive compounds. This study investigated the anticancer effects of wild ginseng adventitious root extract (WGAR) on MDA-MB-231 TNBC cells and elucidated the underlying molecular mechanisms. Methods: WGAR was prepared from cultured adventitious roots of 100-year-old wild ginseng, and its chemical composition was analyzed by LC-MS/MS. Anticancer effects were evaluated using MTT assay, acridine orange/propidium iodide (AO/PI) staining, Matrigel invasion assay, Western blot analysis, and proteome profiler array. Molecular docking was performed to predict interactions between WGAR constituents and target proteins poly (ADP-ribose) polymerase (PARP)-1 and β-catenin. Results: LC-MS/MS analysis tentatively identified 17 compounds, including ginsenosides (Rg3, Rh1, Rf) and terpenoids (ursolic acid). WGAR reduced cell viability with an IC50 of 79 μg/mL at 48 h, inducing 51.2% cell death. WGAR activated the intrinsic apoptotic pathway through sequential caspase-9 and caspase-3 activation, followed by PARP cleavage, and was associated with changes in epithelial–mesenchymal transition (EMT)-related markers (reduced N-cadherin, Slug, and β-catenin) alongside decreased inhibitory Ser9 phosphorylation of GSK-3β. Proteome array analysis revealed suppression of ECM remodeling proteins (tenascin C, u-PA) and inflammatory mediators (IL-6, CXCL8). Molecular docking predicted that selected WGAR constituents, particularly terpenoid-type compounds, may potentially interact with PARP-1 and β-catenin; however, these in silico findings are hypothesis-generating and require experimental validation. Conclusions: WGAR exerts multi-target anticancer effects on TNBC cells through apoptosis induction and EMT suppression associated with modulation of GSK-3β/β-catenin signaling, suggesting its potential as a source of therapeutic agents for TNBC.

1. Introduction

Triple-negative breast cancer (TNBC) is defined by the absence of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) expression, accounting for approximately 15–20% of all breast cancer cases [1]. Compared with other breast cancer subtypes, TNBC is characterized by a worse prognosis and increased risk of metastasis to vital organs [2]. A key contributor to the aggressive behavior of TNBC is the epithelial–mesenchymal transition (EMT), a process by which epithelial cells lose cell–cell adhesion and acquire mesenchymal characteristics, thereby enhancing migratory and invasive properties [3]. TNBC cells frequently exhibit mesenchymal gene expression patterns, and hybrid epithelial/mesenchymal phenotypes have been implicated in driving metastasis in both preclinical models and patient samples [4]. At the molecular level, the Wnt/β-catenin signaling pathway is aberrantly activated in TNBC and has been associated with tumor progression, metastasis, and therapy resistance [5]. This pathway regulates EMT-related transcription factors and contributes to the maintenance of cancer stem cell properties [6]. Despite these mechanistic insights, there are currently no effective targeted therapies for TNBC, and the development of poly (ADP-ribose) polymerase (PARP) and epidermal growth factor receptor (EGFR) inhibitors has shown limited clinical benefit [2]. Furthermore, high toxicity of conventional chemotherapeutic agents and multidrug resistance remain significant obstacles to the effective management of TNBC [7]. Given the molecular heterogeneity of TNBC and the involvement of multiple interconnected signaling pathways, single-target therapeutic approaches have proven insufficient [8]. In this context, natural products have attracted attention as potential anticancer agents due to their structural diversity and ability to modulate multiple molecular targets simultaneously [9]. Among various natural products, ginseng (Panax species) and its bioactive constituents, ginsenosides, have been extensively studied for their anticancer properties, including induction of apoptosis, inhibition of metastasis, and modulation of signaling pathways in various cancer types, including breast cancer [10].

Based on growing environment and cultivation method, Panax ginseng can be classified into three categories: cultivated ginseng (4–6 years in farms), mountain-cultivated ginseng (forest-cultivated to mimic natural conditions), and wild ginseng (naturally growing for decades to over a century) [11]. Wild ginseng has traditionally been considered more effective than cultivated ginseng, requiring more than 30 years to mature compared to 5–6 years for cultivated varieties [12,13]. This pharmacological superiority is attributed to 2–6-fold higher saponin concentrations [14], unique ginsenosides such as C-Mc1 and C-O not detected in cultivated ginseng [15], and elevated contents of rare ginsenosides and amino acids [16,17]. Wild ginseng extract has been reported to exhibit more prominent anti-inflammatory and anticancer activities than cultivated ginseng [14]. However, its extreme rarity due to long growth periods and specific habitat requirements has significantly limited research and industrial applications [18], and mountain-cultivated ginseng developed as an alternative requires further comparative studies [15].

Adventitious root culture of wild ginseng has emerged as a biotechnological approach to overcome supply limitations, offering genetic stability and consistent production of secondary metabolites [19]. Adventitious root cultures have been established from wild P. ginseng specimens over 100 years old and proliferated in bioreactors for large-scale production [18,20]. These adventitious roots retain the biochemical characteristics of native wild ginseng, including genomic DNA composition and saponin content [21], and mutation breeding has achieved 1.8- to 2.3-fold increases in ginsenoside content [22]. These characteristics make adventitious root culture suitable for mass production [23]. Despite these advances, the anticancer potential of wild ginseng adventitious roots in breast cancer cells remains largely unexplored.

The anticancer properties of wild ginseng have been investigated in various cancer models, including T cell lymphoma xenografts where tumor growth and Akt/Src activation were suppressed [24], and colorectal and gastric carcinoma cells where anti-metastatic activity was demonstrated through AXIN2 inhibition [11]. Clinical case reports have documented favorable outcomes in metastatic breast cancer patients treated with wild ginseng pharmacopuncture [14,25]. In contrast, research on wild ginseng adventitious roots has focused on non-cancer applications; total saponins exhibited anti-inflammatory effects through suppression of nuclear factor kappa B (NF-κB), mitogen-activated protein kinase (MAPK), and Akt pathways in macrophages [26]. To our knowledge, no study has examined the anticancer effects of wild ginseng adventitious root extracts on breast cancer cells, particularly triple-negative breast cancer.

In the present study, we investigated the anticancer effects of wild ginseng adventitious root extract (WGAR) on MDA-MB-231 triple-negative breast cancer cells and compared its efficacy with the conventional chemotherapeutic agent cisplatin. Given that wild ginseng exhibits prominent anticancer activities [14] and that aberrant GSK-3β/β-catenin signaling drives EMT and metastasis in TNBC [5,6], we hypothesized that (1) WGAR would induce apoptosis via the intrinsic mitochondrial pathway, and (2) WGAR would suppress EMT-related markers through modulation of the GSK-3β/β-catenin axis. The bioactive compounds in WGAR were characterized using liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis. The anticancer activity of WGAR was evaluated through cell viability, migration, and invasion assays. To elucidate the underlying molecular mechanisms, we examined apoptosis-related markers (caspase-9, caspase-3, and PARP), epithelial–mesenchymal transition (EMT) markers (E-cadherin, N-cadherin, β-catenin, and Slug), and glycogen synthase kinase-3β (GSK-3β) signaling pathway by Western blot analysis. Additionally, proteome profiling was performed to identify differentially expressed oncology-related proteins, and molecular docking analysis was conducted to predict potential target proteins of the identified compounds. Our findings provide the first evidence for the anticancer potential of wild ginseng adventitious root extract against triple-negative breast cancer and offer insights into the molecular mechanisms underlying its therapeutic effects.

2. Results

2.1. Influence of Ethanol–Water Ratio on Chromatographic Profiles

The WGAR extracts prepared with six different ethanol–water ratios (0, 20, 40, 60, 80, and 100% ethanol) were analyzed by ultra-high-performance liquid chromatography (UHPLC)-Orbitrap MS within a retention period of 32 min (Figure 1). The chromatographic profiles exhibited differences in peak distribution, intensity, and retention behavior across the solvent systems, reflecting chemically diverse metabolite compositions ranging from highly polar to strongly lipophilic constituents.

The major peaks were concentrated in two distinct regions: the polar region (t_R < 4.5 min) containing small molecules with m/z values less than 400 Da, and the non-polar region (t_R 21–25.5 min) dominated by higher molecular weight compounds. Among the six extraction conditions, the 100% ethanol extract demonstrated distinctive enrichment of late-eluting peaks in the t_R 21–24 min region (Figure 1a). The three-dimensional mass profile (Figure 1b) revealed that these constituents were characterized by higher m/z values, consistent with non-polar metabolites such as fatty acid derivatives and highly substituted saponins.

Based on cytotoxicity screening results (Section 2.3), the 100% ethanol extract exhibited the most potent anticancer activity against MDA-MB-231 cells. Therefore, this extract was selected for further metabolomic analysis to identify the bioactive constituents. Extraction yields differed across ethanol–water solvent systems (

Table 1. Extraction yields of WGAR obtained using different ethanol–water solvent systems. Extraction yield (%) was calculated as (dry extract weight/raw material weight) × 100. Each extraction was performed in triplicate using ultrasonic extraction at 40 °C for 2 h (three consecutive cycles). Values are presented as mean ± SD (n = 3).

Ethanol (%, v/v)	Raw Material (g)	Extract (mg)	Yield (%)
0	1.0	273.30 ± 5.63	27.33 ± 0.56
20	1.0	255.73 ± 2.84	25.57 ± 0.28
40	1.0	256.70 ± 5.51	25.67 ± 0.55
60	1.0	200.43 ± 4.96	20.04 ± 0.50
80	1.0	47.63 ± 3.96	4.76 ± 0.40
100	1.0	203.10 ± 4.86	20.31 ± 0.49

). The yield ranged from 4.76 ± 0.40% (80% ethanol) to 27.33 ± 0.56% (20% ethanol). One hundred percent ethanol condition was selected for subsequent experiments based on downstream chemical profiling and biological activity results.

2.2. Identification of Chemical Constituents via Molecular Networking

To characterize the chemical constituents in the active extract, the 100% ethanol extract was analyzed using Feature-Based Molecular Networking (FBMN) coupled with the Global Natural Products Social (GNPS) platform. This approach enables rapid annotation of complex mass spectrometry datasets by organizing spectra into structured molecular networks, where each node represents a compound and edges denote structural relatedness based on fragmentation similarity (Figure 2).

The peaks detected in the region of interest (t_R 21–24 min) were further analyzed by molecular networking, which connects mass spectra based on the similarity of their precursor ions and MS² fragmentation patterns. A total of 17 compounds (83–87, 90–95, 98–103) were tentatively identified by comparing precursor and fragmentation ions against public MS/MS databases (

Table 2. Identification of compounds from wild ginseng adventitious root (WGAR) extracted with 100% ethanol. RT, retention time; m/z, mass-to-charge ratio; mDa, millidalton. Dagger (†) indicates compounds identified by matching with in-house library.

No.	Compound	RT (min)	Formula	Adduct	m/z	Error (mDa)	Class
82	Unknown	21.205	-	[M+H]+	708.3732	-	-
83	Pseudoginsenoside F11	21.629	C42H72O14	[M+H]+	801.5006	0.002	Terpenoids
84	Myristic acid †	21.927	C14H28O2	[M+H]+	229.1551	1.709	Fatty acids
85	Poecillastroside D †	21.988	C41H68O13	[M+K]+	807.4413	0.160	Terpenoids
86	Licoricesaponin G2	22.426	C42H62O17	[M+H]+	839.4097	3.723	Terpenoids
87	Majonoside R2 †	22.484	C41H70O14	[M+NH4]+	805.4254	0.050	Terpenoids
88	Unknown	22.528	-	[M+H]+	765.4796	-	-
89	Unknown	22.691	-	[M+H]+	851.4803	-	-
90	Soyasapogenol C	22.864	C30H48O2	[M+H]+	441.3733	0.305	Terpenoids
91	Glochidone	22.884	C30H46O	[M+H]+	423.3626	0.580	Terpenoids
92	Methyl palmitate †	22.937	C17H34O2	[M+NH4]+	288.2901	0.916	Fatty acids
93	Ginsenoside Rf †	22.943	C42H72O14	[M+H-H2O]+	783.4901	1.221	Terpenoids
94	Ginsenoside Rh1 †	23.401	C36H62O9	[M+H-2H2O]+	603.4266	0.001	Terpenoids
95	Ursolic acid	23.456	C30H48O3	[M+H-H2O]+	439.3577	2.258	Terpenoids
96	Unknown	23.467	-	[M+H]+	297.2062	-	-
97	Unknown	23.576	-	[M+H]+	407.3680	-	-
98	Austinoneol	23.628	C24H30O6	[M+H]+	415.2120	1.038	Terpenoids
99	Octadecane †	23.669	C18H38	[M+COOH]+	293.2115	0.458	Fatty acids
100	(1R,2R,4aS,8aS)-2-[(2R)-2-Hydroxybutyl]-1,3-dimethyl-1,2,4a,5,6,7,8,8a-octahydro-1-naphthalenecarboxylic acid	23.763	C17H28O3	[M+H-H2O]+	263.2008	0.275	Naphthalenes
101	Methyl Stearate	23.772	C19H38O2	[M+H-H2O]+	281.2956	0.380	Fatty acids
102	20(S)-ginsenoside Rg3 †	24.070	C42H72O13	[M+H-H2O]+	767.4958	1.831	Terpenoids
103	20(R)-ginsenoside Rg3	24.090	C42H72O13	[M+H-2H2O]+	749.4847	0.002	Terpenoids

Compound identities were tentatively assigned based on accurate mass and MS/MS spectral matching against public databases (GNPS, MS-DIAL) and were not confirmed with authentic reference standards.

, Figure 3). The identified compounds were classified into five categories: terpenoids (55%), fatty acids (18%), naphthalenes (5%), and others (23%) (Figure 3b).

Among the ginsenosides, pseudoginsenoside F11 (83) was detected at t_R 21.629 min with a protonated precursor ion at m/z 801.5006 [M+H]⁺. Its fragmentation pattern showed a characteristic dehydration ion at m/z 783.4904 [M+H−H₂O]⁺ along with an aglycone fragment at m/z 441.3733 [M+H−rutinose]⁺ (Figure S1). The peak at m/z 143.0705 arises from retro-Diels-Alder (RDA) cleavage, a characteristic feature of ocotillol-type ginsenosides. Similarly, majonoside R2 (87), detected at t_R 22.484 min, shares the same ocotillol-type aglycone. Ginsenoside Rf (93) was tentatively identified at tR 22.943 min with a dehydrated precursor ion at m/z 783.4901 [M+H−H₂O]⁺, showing characteristic protopanaxatriol (PPT)-type ginsenoside fragments at m/z 459.3836, 441.3732, 423.3627, and 405.2256. Ginsenoside Rh1 (94) at t_R 23.401 min exhibited a precursor ion at m/z 603.4266 [M+H−2H₂O]⁺, differing from ginsenoside Rf by a reduction of 163 Da corresponding to one hexose sugar. Additionally, 20(S)-ginsenoside Rg3 (102) and 20(R)-ginsenoside Rg3 (103) were detected at t_R 24.070 and 24.090 min, respectively.

Poecillastroside D (85) was detected at t_R 21.988 min with prominent ions at m/z 807.4413 [M+K]⁺, showing fragmentation consistent with loss of a Glc–Glc disaccharide unit. The triterpenoid derivatives soyasapogenol C (90) and glochidone (91) were tentatively identified at t_R 22.864 and 22.884 min, respectively, based on their characteristic triterpenoid fragmentation patterns. Ursolic acid (95) was detected at t_R 23.456 min with fragment ions at m/z 439.3577 [M+H−H₂O]⁺ and 393.2259 [M+H−2H₂O−COOH]⁺ (Figure S2).

Fatty acid derivatives including myristic acid (84), methyl palmitate (92), octadecane (99), and methyl stearate (101) were also tentatively identified based on their mass fragmentation pathways. Additional compounds austinoneol (98) and (1R,2R,4aS,8aS)-2-[(2R)-2-hydroxybutyl]-1,3-dimethyl-1,2,4a,5,6,7,8,8a-octahydro-1-naphthalenecarboxylic acid (100) were annotated by matching with public mass databases using the MS-DIAL version 5.1 program.

2.3. Cytotoxicity of WGAR on MDA-MB-231 Cells

To optimize the extraction conditions for WGAR, we first evaluated the cytotoxic effects of WGAR extracts prepared with different ethanol–water ratios (0, 20, 40, 60, 80, and 100% ethanol) on MDA-MB-231 cells using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Figure 4a). Among all extraction conditions, 100% ethanol extract exhibited the strongest cytotoxicity, with cell viability reduced to 55.7 ± 2.0% at 50 μg/mL and 6.9 ± 1.6% at 5000 μg/mL. In contrast, 80% ethanol extract showed the weakest cytotoxicity, maintaining cell viability above 90% even at 500 μg/mL. Extracts prepared with 0–60% ethanol showed intermediate effects, with significant cytotoxicity observed only at concentrations above 500 μg/mL. Based on these results, 100% ethanol extract was selected for subsequent experiments.

The cytotoxic effects of WGAR (100% ethanol extract) and cisplatin were then compared (Figure 4b). Both WGAR and cisplatin exhibited dose-dependent cytotoxicity against MDA-MB-231 cells. WGAR treatment resulted in a gradual decrease in cell viability with increasing concentrations. At 5 μg/mL, cell viability was reduced to 77.7 ± 2.6%, while 50 μg/mL treatment further decreased viability to 61.5 ± 0.4%. At 500 μg/mL, WGAR almost completely inhibited cell viability, reducing it to 4.2 ± 0.4%. Based on the dose–response curve, the half-maximal inhibitory concentration (IC50) and 20% inhibitory concentration (IC20) values of WGAR were determined to be 79 μg/mL and 3 μg/mL, respectively.

Cisplatin demonstrated stronger cytotoxicity than WGAR across all concentrations tested. Treatment with 3 μg/mL cisplatin reduced cell viability to 73.4 ± 4.0%, and 12 μg/mL treatment decreased viability to 44.4 ± 4.2%. At 400 μg/mL, cisplatin reduced cell viability to 7.7 ± 0.9%. The IC50 and IC20 values of cisplatin were 9 μg/mL and 2 μg/mL, respectively. Although WGAR exhibited approximately 9-fold lower potency than cisplatin based on IC50 values, the natural product demonstrated significant anticancer activity against this aggressive breast cancer cell line. These IC20 and IC50 concentrations were used for subsequent mechanistic studies. IC20 and IC50 values were calculated from dose–response curves and used for subsequent experiments: IC20 was used as a sub-cytotoxic exposure level to evaluate marker modulation with minimal confounding from cell death, whereas IC50 was used to compare responses under equipotent cytotoxic stress between WGAR and cisplatin.

2.4. Cell Death and Apoptosis Analysis

To validate the cytotoxic effects observed in the MTT assay, acridine orange (AO)/propidium iodide (PI) dual fluorescent staining was performed (Figure 5a).

In control cells, the majority of cells exhibited green fluorescence, indicating high viability (98.7 ± 0.1%), with minimal dead cells (1.3 ± 0.1%). Treatment with WGAR at IC20 (3 μg/mL) resulted in a noticeable increase in red fluorescent cells, with viability decreasing to 81.2 ± 3.9% and dead cells increasing to 18.8 ± 3.9% (p < 0.01). At IC50 (79 μg/mL), WGAR treatment showed a marked shift from green to red fluorescence, with approximately half of the cells being non-viable (48.8 ± 10.0% live cells, 51.2 ± 10.0% dead cells; p < 0.01).

Similarly, cisplatin treatment induced dose-dependent cell death. At IC20 (2 μg/mL), cisplatin reduced viability to 79.1 ± 0.9% (p < 0.001), and at IC50 (9 μg/mL), viability decreased to 49.8 ± 0.9% with a corresponding increase in dead cells to 50.2 ± 0.9% (p < 0.001). Notably, both WGAR and cisplatin at their respective IC50 concentrations induced approximately 50% cell death, which is consistent with the MTT assay results (Figure 5b).

To further characterize the mode of cell death, Annexin V/7-AAD flow cytometry was performed to distinguish between early apoptotic, late apoptotic, and necrotic cell populations (Figure 5c). In control cells, the majority of cells were viable (Annexin V−/7-AAD−; 91.65 ± 1.68%), with minimal early apoptotic (Annexin V+/7-AAD−; 5.55 ± 1.87%) or late apoptotic/dead cells (Annexin V+/7-AAD+; 2.28 ± 0.21%). WGAR treatment increased total apoptotic cells from 27.50 ± 10.34% at IC20 (p < 0.05) to 86.53 ± 3.02% at IC50 (p < 0.001). Cisplatin similarly induced apoptosis, with total apoptotic populations of 19.67 ± 6.80% at IC20 (p < 0.05) and 39.09 ± 13.67% at IC50 (p < 0.05). These results demonstrate that both WGAR and cisplatin induce cell death primarily through apoptosis, as evidenced by increased early and late apoptotic populations, while the necrotic population (Annexin V−/7-AAD+) remained comparatively low.

2.5. Effects of WGAR on Cell Invasion

The invasive capacity of MDA-MB-231 cells was evaluated using Matrigel-coated Transwell chambers to assess whether WGAR could inhibit cancer cell invasion through an extracellular matrix barrier (Figure 6a).

Quantitative analysis showed no significant inhibition of invasion at IC20 for either WGAR (3 μg/mL) or cisplatin (2 μg/mL), whereas invasion was markedly reduced at IC50 for both treatments (Figure 6b). At IC20 concentrations, neither WGAR (3 μg/mL) nor cisplatin (2 μg/mL) significantly affected cell invasion, showing relative invasion rates of 0.88 ± 0.19 (p = 0.39) and 0.94 ± 0.13 (p = 0.61), respectively, compared to the control. At IC50 concentrations, both treatments markedly reduced cell invasion. WGAR at IC50 (79 μg/mL) reduced the relative invasion to 0.12 ± 0.09, representing an 88% reduction compared to the control (p < 0.001). Similarly, cisplatin at IC50 (9 μg/mL) decreased relative invasion to 0.13 ± 0.04, an 87% reduction (p < 0.001).

These results show that WGAR reduces invasion at IC50; however, because invasion was not significantly inhibited at IC20, the data do not support a definitive anti-invasive effect independent of cytotoxicity. Additional validation under sub-cytotoxic or proliferation-controlled conditions is warranted. The comparable reductions in invasion observed at IC50 suggest that both agents can be associated with reduced invasive readouts under cytotoxic exposure; however, interpretation as direct suppression of invasive potential is limited by the lack of significant inhibition at IC20.

2.6. Western Blot Analysis

2.6.1. Effects on Apoptosis-Related Proteins

To investigate whether WGAR induces apoptosis in MDA-MB-231 cells, the expression of apoptosis-related proteins was examined by Western blot analysis (Figure 7a).

Treatment with WGAR and cisplatin induced caspase activation in a dose-dependent manner. The ratio of cleaved caspase-9 to total caspase-9 increased significantly following treatment. WGAR at IC50 increased this ratio to 1.35 ± 0.09 (p < 0.001), while cisplatin at IC20 and IC50 showed even greater increases to 2.08 ± 0.10 (p < 0.01) and 2.49 ± 0.12 (p < 0.001), respectively (Figure 7b). Similarly, the cleaved caspase-3 to caspase-3 ratio was elevated, with WGAR IC50 showing a ratio of 1.10 ± 0.08 (p < 0.01) and cisplatin IC50 showing 1.73 ± 0.11 (p < 0.05).

Cleaved PARP (89 kDa), a hallmark of apoptosis, was significantly increased following WGAR IC20 treatment (1.32 ± 0.15, p < 0.05), while full-length PARP (116 kDa) decreased in WGAR IC50 and cisplatin IC50 groups (p < 0.05). These results indicate that both WGAR and cisplatin induce apoptosis through the caspase cascade, although the specific activation patterns differ between the two agents.

To functionally validate caspase activation observed in Western blot analysis, a caspase activity assay was performed using flow cytometry (Figure 7c). Control cells showed minimal caspase activity, with 9.35 ± 0.67% of cells exhibiting caspase positivity. WGAR treatment induced dose-dependent caspase activation, with total caspase-positive cells increasing from 19.30 ± 6.05% at IC20 (p < 0.05) to 37.47 ± 2.42% at IC50 (p < 0.001). Cisplatin similarly activated caspases, showing 14.20 ± 3.25% and 33.35 ± 1.72% caspase-positive cells at IC20 and IC50, respectively (p < 0.001 for IC50).

2.6.2. Effects on EMT and GSK-3β Signaling Pathway

EMT is a critical process in cancer metastasis, and GSK-3β is a key regulator of EMT through its control of β-catenin stability. We examined whether WGAR and cisplatin modulate EMT and GSK-3β signaling in MDA-MB-231 cells (Figure 8a).

Total GSK-3β expression remained relatively constant across all treatment groups, indicating that neither WGAR nor cisplatin affected GSK-3β protein levels (Figure 8b). However, phosphorylated GSK-3β (Ser9) was significantly decreased by all treatments. WGAR reduced p-GSK-3β levels to 0.80 at IC20 and 0.75 at IC50 (p < 0.05), while cisplatin decreased p-GSK-3β to 0.70 at IC20 and 0.80 at IC50 (p < 0.05). Since phosphorylation at Ser9 inactivates GSK-3β, this decreased phosphorylation suggests reduced inhibitory regulation of GSK-3β by both WGAR and cisplatin.

Consistent with reduced inhibitory Ser9 phosphorylation of GSK-3β, β-catenin expression was markedly reduced in all treatment groups compared to the control (p < 0.001). WGAR at IC20 and IC50 reduced β-catenin levels to 0.46 ± 0.05 and 0.53 ± 0.06, respectively, while cisplatin at IC20 and IC50 decreased β-catenin to 0.32 ± 0.04 and 0.50 ± 0.05, respectively. This 47–68% reduction in β-catenin is consistent with enhanced GSK-3β-mediated degradation.

The suppression of GSK-3β/β-catenin signaling was accompanied by changes in EMT markers. E-cadherin, an epithelial marker, showed increased expression following cisplatin IC20 treatment (1.47 ± 0.12, p < 0.05). In contrast, the mesenchymal marker N-cadherin was significantly reduced by WGAR IC50 (0.75 ± 0.09, p < 0.01), cisplatin IC20 (0.34 ± 0.06, p < 0.05), and cisplatin IC50 (p < 0.001). The EMT-inducing transcription factor Slug was also significantly downregulated by WGAR IC50 (0.35 ± 0.07, p < 0.001), cisplatin IC20 (0.25 ± 0.05, p < 0.001), and cisplatin IC50 (0.30 ± 0.06, p < 0.001).

To validate Slug suppression at the transcriptional level, qPCR analysis was performed after 24 h treatment (Figure 8c) to capture early transcriptional responses. WGAR reduced SNAI2 (Slug) mRNA expression to 0.72 ± 0.10-fold at IC20 and 0.58 ± 0.12-fold at IC50 versus control (p < 0.05 and p < 0.01, respectively). Cisplatin showed modest effects on SNAI2 expression (IC20: 0.88 ± 0.14-fold, ns; IC50: 0.71 ± 0.11-fold, p < 0.05). Notably, the magnitude of Slug protein reduction in cisplatin-treated cells appeared greater than the corresponding decrease in SNAI2 mRNA levels.

These findings suggest that both WGAR and cisplatin reduce inhibitory Ser9 phosphorylation of GSK-3β, which is consistent with modulation of the GSK-3β/β-catenin axis and may be associated with the reduced β-catenin levels and EMT-related changes observed in this study.

2.7. Proteome Profiling and Validation

To comprehensively profile the effects of WGAR and cisplatin on oncology-related protein expression, we employed the Human XL Oncology Array, which enables simultaneous detection of 84 cancer-related proteins (Figure 9a).

At IC20 concentrations, WGAR and cisplatin exhibited distinct protein expression profiles (Figure 9b). WGAR treatment broadly suppressed proteins across multiple functional categories. Inflammatory mediators including interleukin (IL)-6 and macrophage colony-stimulating factor (M-CSF) were reduced to 0.13-fold, while C-X-C motif chemokine ligand 8 (CXCL8), C-C motif chemokine ligand 2 (CCL2), and granulocyte-macrophage colony-stimulating factor (GM-CSF) decreased to 0.38-, 0.28-, and 0.38-fold, respectively. Extracellular matrix (ECM) remodeling proteins were also suppressed, including tenascin C (0.19-fold), urokinase-type plasminogen activator (u-PA, 0.34-fold), thrombospondin-1 (0.35-fold), and plasminogen activator inhibitor-1 (PAI-1, 0.38-fold). Among growth factor signaling proteins, mesothelin showed the greatest reduction (0.06-fold), followed by galectin-3 (0.04-fold), amphiregulin (0.24-fold), and Axl receptor tyrosine kinase (0.49-fold). In contrast, cisplatin IC20 showed selective effects on inflammatory markers: CCL2 was significantly reduced (0.35-fold), while IL-6, M-CSF, and GM-CSF remained near control levels (0.95-, 1.12-, and 1.05-fold, respectively). Cisplatin effectively suppressed tenascin C (0.12-fold) and vascular endothelial growth factor (VEGF, 0.29-fold).

At IC50 concentrations, both treatments showed convergent effects on ECM remodeling and tumor-associated proteins (Figure 9c). Tenascin C was comparably reduced by WGAR (0.20-fold) and cisplatin (0.21-fold), consistent with the anti-invasive effects observed in the Matrigel invasion assay. Similarly, mesothelin was suppressed by both treatments (WGAR: 0.32-fold; cisplatin: 0.31-fold). Tumor suppressor p53 was upregulated approximately 2-fold by both WGAR and cisplatin, correlating with the caspase activation observed in apoptosis analysis. Differential responses were observed for CCL2, which was suppressed by cisplatin (0.29-fold) but remained unchanged by WGAR (0.91-fold). Cathepsin S showed a slight increase with cisplatin treatment (1.45-fold).

These results indicate that WGAR suppresses a broader spectrum of pro-tumorigenic proteins at sub-cytotoxic concentrations, whereas cisplatin shows more selective effects. At cytotoxic concentrations, both treatments converge on key targets including ECM remodeling proteins and p53 activation.

To independently validate the proteome array findings, IL-6 and M-CSF secretion were measured by ELISA (Figure 9d). WGAR treatment at IC20 significantly reduced IL-6 levels to 0.28 ± 0.09-fold (p < 0.01) and M-CSF levels to 0.22 ± 0.08-fold (p < 0.01) compared with control, which was directionally consistent with the array results. At IC50, IL-6 showed a modest decrease (0.65 ± 0.18-fold) that did not reach statistical significance, whereas M-CSF remained significantly suppressed (0.48 ± 0.12-fold, p < 0.05). In contrast, cisplatin had minimal effects on IL-6 at both IC20 (0.92 ± 0.14-fold) and IC50 (0.85 ± 0.16-fold), but significantly reduced M-CSF at IC50 (0.52 ± 0.14-fold, p < 0.05). These results indicate that WGAR suppresses inflammatory mediators at sub-cytotoxic concentrations, while M-CSF suppression at IC50 represents a convergent effect of both WGAR and cisplatin treatment.

2.8. Molecular Docking Analysis

To investigate the potential molecular targets of WGAR constituents, molecular docking studies were performed on PARP-1 (Protein Data Bank (PDB) ID: 7KK5) and β-catenin (PDB ID: 1JDH), which are key regulators of apoptosis and EMT, respectively. The docking protocol was validated by redocking the native ligands, which showed root-mean-square deviation (RMSD) values of 0.30 Å for PARP-1, confirming the reliability of the docking method.

For PARP-1, the native ligand exhibited a binding energy of −9.95 kcal/mol (

Table 3. Free binding energies of active compounds against poly (ADP-ribose) polymerase 1 (PARP-1) and β-catenin. Free binding energies were calculated by using AutoDock4.2.

No.	PARP-1	β-Catenin
83	−4.54	−2.47
84	−3.41	−4.07
85	−10.96	−5.90
86	−4.22	−1.42
87	−4.88	−2.48
90	−9.49	−6.23
91	−9.65	−7.19
92	−2.99	−2.67
93	−4.37	−0.35
94	−6.09	−4.89
95	−9.33	−6.76
98	−8.98	−6.91
99	−3.12	−3.02
100	−6.03	−5.84
101	−2.71	−1.19
102	−4.35	−1.42
103	−4.99	−3.14

). Among the 17 compounds identified from WGAR, compound 85 (poecillastroside D) showed the most favorable binding affinity (−10.96 kcal/mol), exceeding that of the native ligand (Figure 10a). Compounds 91 (glochidone, −9.65 kcal/mol), 90 (soyasapogenol C, −9.49 kcal/mol), 95 (ursolic acid, −9.33 kcal/mol), and 98 (austinoneol, −8.98 kcal/mol) also demonstrated strong binding affinities comparable to the native ligand. These compounds formed hydrogen bonds with key catalytic residues including ARG878, TYR896, and SER864, and exhibited hydrophobic interactions with TYR907 and TYR889 within the NAD⁺ binding pocket. The remaining compounds showed weaker binding affinities ranging from −2.71 to −6.09 kcal/mol (Figure S3).

For β-catenin, the native ligand showed a binding energy of −7.79 kcal/mol (Table 2). Compound 91 (glochidone) exhibited the strongest binding affinity (−7.19 kcal/mol), followed by compounds 98 (austinoneol, −6.91 kcal/mol), 95 (ursolic acid, −6.76 kcal/mol), and 90 (soyasapogenol C, −6.23 kcal/mol) (Figure 10b). These compounds interacted with key residues within the armadillo repeat domain of β-catenin. The remaining compounds exhibited weaker binding affinities (Figure S4).

Compounds 91, 95, 90, and 98 showed favorable predicted binding to PARP-1 and β-catenin in silico. Because docking is hypothesis-generating, direct binding would require biochemical confirmation. These predictions are directionally consistent with the observed PARP cleavage and β-catenin reduction, but do not constitute evidence of direct target engagement.

3. Discussion

The present study investigated the anticancer effects of WGAR on MDA-MB-231 triple-negative breast cancer cells. Chemical profiling by LC-MS/MS tentatively identified 17 compounds, which were classified into terpenoids (55%), fatty acids (18%), naphthalenes (5%), and others (Table 2, Figure 3). The major ginsenosides tentatively identified include 20(S)-ginsenoside Rg3, ginsenoside Rh1, and ginsenoside Rf, while ursolic acid was tentatively identified as a major triterpenoid constituent. WGAR induced apoptosis through caspase activation and PARP cleavage and suppressed metastatic potential by inhibiting EMT via the GSK-3β/β-catenin pathway. Molecular docking was performed as an exploratory computational analysis to propose potential interactions between WGAR constituents and PARP-1/β-catenin. These predictions are hypothesis-generating and do not provide direct biological evidence of binding.

Ginseng, particularly its ginsenoside constituents, exhibits a broad spectrum of biological activities, including anticancer effects [27,28,29]. WGAR treatment reduced the viability of MDA-MB-231 cells in a dose-dependent manner, as determined by MTT and AO/PI assays. The IC50 value of WGAR was 79 μg/mL at 48 h, and treatment at this concentration induced 51.2 ± 10.0% cell death. Western blot analysis revealed that WGAR activated the intrinsic apoptotic pathway through sequential caspase activation. Treatment with WGAR increased cleaved caspase-9 and cleaved caspase-3 levels in a dose-dependent manner, followed by proteolytic cleavage of PARP. These findings are consistent with a previous report demonstrating that ginsenoside Rg3 induces apoptosis in MDA-MB-231 cells through caspase-3 activation and PARP proteolysis, effects that were attenuated by the pan-caspase inhibitor z-Val-Ala-Asp-fluoromethyl ketone (z-VAD-fmk) [30]. Given that 20(S)-ginsenoside Rg3 was tentatively identified as a major constituent of WGAR (Table 2), this ginsenoside may contribute to the observed caspase-dependent apoptosis. Proteome array analysis further demonstrated that WGAR upregulated p53 expression in MDA-MB-231 cells. This observation is supported by previous findings that 20(S)-ginsenoside Rg3 upregulates p53 expression through the AMP-activated protein kinase (AMPK) signaling pathway, subsequently activating caspase-9 and caspase-3 in cancer cells [31]. The tumor suppressor p53 induces the apoptotic cascade by transcriptionally activating pro-apoptotic genes and modulating Bcl-2 family proteins; the balance between pro-apoptotic Bax and anti-apoptotic Bcl-2 controls mitochondrial cytochrome c release and subsequent caspase activation [31,32]. Although most studies on ginsenoside-induced apoptosis have utilized cultivated P. ginseng, the underlying molecular mechanisms are attributed to individual ginsenoside components that are also present in wild ginseng. WGAR also contains ursolic acid, a pentacyclic triterpenoid that induces apoptosis in breast cancer cells through caspase-3 and caspase-9 activation via the mitochondrial pathway [33].

Given the presence of multiple terpenoid constituents in WGAR, molecular docking was performed as an exploratory approach to identify potential molecular targets. Terpenoid constituents were predicted to dock into the PARP-1 catalytic pocket: poecillastroside D (−10.96 kcal/mol) showed the highest predicted affinity, followed by glochidone (−9.65 kcal/mol), soyasapogenol C (−9.49 kcal/mol), and ursolic acid (−9.33 kcal/mol). In silico analysis suggested that these compounds may form interactions with key catalytic residues including Gly863 and Tyr907 [34]. In contrast, ginsenosides showed comparatively weaker predicted binding (Rh1: −6.09 kcal/mol; Rg3: −4.35 kcal/mol). Importantly, PARP cleavage observed in this study reflects caspase-mediated proteolysis, which is mechanistically distinct from direct PARP-1 enzymatic inhibition predicted by docking. Therefore, these docking outcomes are hypothesis-generating and would require enzyme inhibition assays for validation. The presence of multiple bioactive compounds with similar apoptotic mechanisms suggests that these components may act synergistically to enhance the pro-apoptotic effect of WGAR. Compared with cisplatin, WGAR and cisplatin appeared to induce apoptosis through partially distinct mechanisms; while WGAR treatment predominantly induced PARP proteolysis, cisplatin showed stronger caspase activation, suggesting different apoptotic pathways between the natural extract and the conventional chemotherapeutic agent. The caspase activity assay further confirmed functional caspase activation, with WGAR IC50 showing 37.85% total caspase-positive cells compared to 9.56% in control, consistent with the Western blot data showing cleaved caspase-9 and caspase-3 induction. These results indicate that WGAR induces apoptosis in MDA-MB-231 cells through the intrinsic mitochondrial pathway, involving p53 upregulation, Bcl-2 family modulation, sequential caspase activation, and PARP proteolysis, with ginsenoside Rg3 and ursolic acid likely serving as major bioactive contributors.

While apoptosis induction reduces cancer cell viability, inhibition of metastatic potential is equally critical for preventing tumor dissemination. Beyond its pro-apoptotic effects, WGAR reduced Matrigel invasion in vitro, with a marked reduction observed at IC50. Because invasion was significantly reduced only at IC50, interpretation of a cytotoxicity-independent anti-invasive effect should be made cautiously and warrants further validation. In the Matrigel invasion assay, WGAR at IC50 (79 μg/mL) reduced the number of invaded cells to 12.1% of control (p < 0.001), which was comparable to cisplatin at IC50 (9 μg/mL; 13.2% of control, p < 0.001). This substantial reduction in invasive capacity prompted investigation of the underlying molecular mechanisms, particularly those related to EMT, intracellular signaling pathways, and extracellular matrix remodeling.

EMT is a process whereby epithelial cells lose cell–cell adhesion and acquire mesenchymal characteristics with enhanced migratory and invasive properties, representing a critical step in breast cancer metastasis [35]. Western blot analysis revealed that WGAR modulated the expression of EMT-related proteins. At IC50, WGAR significantly reduced the mesenchymal marker N-cadherin to 0.75 ± 0.09-fold (p < 0.01) and the EMT-inducing transcription factor Slug to 0.35 ± 0.07-fold (p < 0.001), while E-cadherin showed a modest increase (1.10-fold). Cisplatin also suppressed these mesenchymal markers, reducing N-cadherin to 0.34 ± 0.06-fold at IC20 (p < 0.05) and Slug to 0.25–0.30-fold (p < 0.001), with increased E-cadherin expression at IC20 (1.47 ± 0.12-fold, p < 0.05). These findings indicate that WGAR shifts the EMT balance toward an epithelial phenotype. In nasopharyngeal carcinoma cells, ginsenoside Rg3 has been shown to inhibit EMT by increasing E-cadherin expression while decreasing N-cadherin and vimentin, accompanied by reduced matrix metalloproteinase (MMP)-2 and MMP-9 activity [36]. Furthermore, in MDA-MB-231 cells specifically, ginsenoside Rh1 inhibited migration and invasion through suppression of MMP-2, MMP-9, and VEGF-A via mitochondrial reactive oxygen species (ROS)-mediated inhibition of STAT3 and NF-κB signaling [37]. Given that both 20(S)/20(R)-ginsenoside Rg3 and ginsenoside Rh1 were tentatively identified as constituents of WGAR (Table 2), these ginsenosides may contribute to the EMT-inhibitory effects observed in the present study. Ursolic acid, another constituent tentatively identified in WGAR, has also been reported to suppress EMT markers including Snail, Slug, and fibronectin in MDA-MB-231 cells [38].

Since β-catenin acts as a transcriptional co-activator for EMT-related genes including Slug and Snail [39], we examined whether the observed EMT suppression was associated with modulation of GSK-3β/β-catenin signaling. WGAR decreased phosphorylated GSK-3β at Ser9 to 0.80-fold at IC20 and 0.75-fold at IC50 (both p < 0.05), indicating reduced inhibitory phosphorylation of GSK-3β. Cisplatin showed similar effects (0.70–0.80-fold, p < 0.05). Total GSK-3β levels remained unchanged across all treatment conditions. Because phosphorylation at Ser9 inactivates GSK-3β, this dephosphorylation releases the kinase from inhibition, enabling it to phosphorylate β-catenin and target it for proteasomal degradation. Consistent with reduced inhibitory Ser9 phosphorylation of GSK-3β, β-catenin expression was reduced by WGAR to 0.46 ± 0.05-fold at IC20 and 0.53 ± 0.06-fold at IC50 (both p < 0.001). Similarly, cisplatin reduced β-catenin to 0.32–0.50-fold (p < 0.001). In hepatocellular carcinoma cells, ginsenoside Rh2 has been shown to activate GSK-3β and promote β-catenin degradation [39,40], and ginsenoside Rg3 has been reported to inhibit EMT progression via the Wnt/β-catenin pathway in glioblastoma cells [41]. Although Rh2 was not directly identified in WGAR, structurally related ginsenosides including Rg3 were detected (Table 2), suggesting that similar mechanisms may be operative. Since β-catenin nuclear translocation promotes Slug transcription, the reduction in β-catenin observed in this study provides a mechanistic link to the downregulation of Slug and the consequent preservation of E-cadherin expression. Overall, these data indicate changes in EMT-related marker expression under WGAR and cisplatin treatment. Because functional EMT assays under proliferation-arrested conditions were not performed, these findings should be interpreted at the marker level rather than as definitive evidence of EMT suppression. In addition, the invasion readout was markedly reduced only at IC50, where cytotoxicity may confound interpretation. Notably, the apparent discrepancy between the modest SNAI2 mRNA changes and the stronger Slug protein reduction observed by Western blot in cisplatin-treated cells may reflect additional post-transcriptional regulation and/or altered protein stability under cytotoxic stress. Accordingly, we describe these observations as EMT-related marker modulation rather than definitive EMT suppression.

Molecular docking was also performed as an exploratory approach to evaluate potential interactions between WGAR constituents and β-catenin. Among the terpenoid constituents, glochidone exhibited the strongest predicted binding affinity (−7.19 kcal/mol), followed by austinoneol (−6.91 kcal/mol) and ursolic acid (−6.76 kcal/mol). In contrast, ginsenosides showed weaker predicted binding to β-catenin (Rh1: −4.89 kcal/mol; Rg3: −1.42 kcal/mol). Ursolic acid has been reported to suppress cancer cell proliferation through inhibition of Wnt/β-catenin signaling in colorectal cancer cells [42]. The experimentally observed reduction in β-catenin and reduced inhibitory Ser9 phosphorylation of GSK-3β indicate pathway modulation at the protein level. In contrast, docking predictions suggesting potential interactions between terpenoid constituents and β-catenin are computational and hypothesis-generating and require validation. However, experimental validation through binding assays would be required to confirm these computational predictions. However, it should be noted that GSK-3β activation was inferred from reduced Ser9 phosphorylation without functional validation using GSK-3β inhibitors; therefore, further studies employing pharmacological inhibitors such as LiCl or CHIR99021 would be required to confirm the direct involvement of GSK-3β in the observed EMT suppression.

The EMT program endows cancer cells with enhanced migratory capacity, but successful invasion also requires degradation of the surrounding extracellular matrix. Proteome array analysis revealed that WGAR suppressed multiple ECM remodeling proteins at IC20 concentrations. Tenascin C, an ECM glycoprotein associated with tumor invasion and metastasis [43,44], was reduced to 0.19-fold by WGAR. WGAR also suppressed urokinase-type plasminogen activator (u-PA; 0.34-fold), PAI-1 (0.38-fold), and thrombospondin-1 (0.35-fold). Cisplatin showed similar suppression of tenascin C (0.12-fold) and u-PA (0.37-fold). In breast carcinomas, large splice variants of tenascin C are abundant at the invasive tumor front and correlate with enhanced tumor cell proliferation [45]. In triple-negative breast cancer, tenascin C accumulation promotes resistance to T cell-mediated cytotoxicity, and its degradation restores anti-tumor immune responses [46]. The u-PA/PAI-1 system is involved in cancer cell invasion through activation of plasmin-mediated ECM degradation; plasmin not only directly degrades ECM components but also activates MMPs to further promote matrix breakdown. Red ginseng extract has been reported to suppress cancer metastasis by inhibiting MMP-2 and MMP-9 activities [47], and ginsenoside Rg1 suppressed PMA-induced tumor cell invasion through NF-κB-dependent inhibition of MMP-9 expression in MCF-7 breast cancer cells [48]. Although MMP activities were not directly measured in the present study, the observed suppression of u-PA by WGAR may be consistent with a potential upstream modulation of the plasmin–MMP cascade. These proteome array results provide screening-level, semi-quantitative observations. While IL-6 and M-CSF were independently validated by ELISA, other ECM/invasion-related targets (e.g., tenascin C, u-PA) remain unvalidated and should be considered hypothesis-generating. The concurrent suppression of multiple ECM remodeling proteins by WGAR suggests a coordinated inhibition that may contribute to the reduced invasion observed in the Matrigel assay.

Beyond direct effects on ECM remodeling, the tumor microenvironment is shaped by inflammatory mediators that promote cancer progression. IL-6 activates STAT3 signaling to drive breast cancer progression and metastasis [49], and CXCL8 promotes cancer cell migration and invasion through autocrine and paracrine mechanisms [50]. At IC20, WGAR substantially reduced IL-6 and M-CSF to 0.13-fold, representing a >85% reduction compared to control. CXCL8, CCL2, and GM-CSF were also suppressed (0.28–0.38-fold). This broad suppression of inflammatory mediators was specific to WGAR; cisplatin IC20 selectively reduced only CCL2 (0.35-fold) while maintaining IL-6, M-CSF, and GM-CSF near control levels. This differential profile suggests that WGAR may exert anti-inflammatory effects through mechanisms distinct from conventional chemotherapy. Ginsenoside Rg3 has been shown to inhibit chemokine secretion from macrophages through suppression of MAPK and NF-κB pathways [51], and similar mechanisms may underlie the inflammatory mediator suppression observed with WGAR treatment.

In addition to inflammatory mediators, WGAR modulated tumor-associated proteins involved in growth factor signaling and cell adhesion. Galectin-3 and mesothelin, both implicated in cancer cell adhesion and invasion [52,53], were significantly reduced at IC20 (0.04-fold and 0.06-fold, respectively). Galectin-3 promotes cancer cell adhesion to the ECM and endothelium, facilitating metastatic dissemination [53], while mesothelin overexpression in triple-negative breast cancer is associated with increased invasiveness and distant metastasis [52]. WGAR also suppressed amphiregulin, progranulin, and Axl receptor tyrosine kinase (0.24–0.49-fold), which are involved in growth factor signaling and EMT regulation. Of interest, Dickkopf-1 (Dkk-1), a Wnt pathway antagonist, was also reduced (0.23-fold); since Dkk-1 can have context-dependent effects in cancer, further investigation is needed to determine the biological significance of this observation. At IC50, both WGAR and cisplatin showed convergent suppression of mesothelin (0.32-fold vs. 0.31-fold) and cathepsin D (0.55-fold vs. 0.56-fold), suggesting that these proteins may be common targets of anticancer agents regardless of their mechanism of action.

Taken together, WGAR was associated with reduced Matrigel invasion in vitro at IC50 and with concomitant changes in EMT-related protein markers. Because invasion inhibition was not significant at IC20 and IC50 invasion readouts may be confounded by cytotoxicity, these findings should be interpreted cautiously and warrant further functional validation to establish a cytotoxicity-independent anti-invasive effect. The multi-target trends observed with WGAR may be influenced by the diverse constituents tentatively annotated in wild ginseng adventitious roots, including putative ginsenosides (Rg3, Rh1, and Rf) and ursolic acid (Table 2), which have been reported in other systems to be associated with modulation of EMT-related signaling and invasion-related phenotypes. To strengthen confidence in key observations, we independently validated IL-6 and M-CSF changes using ELISA, which were directionally consistent with the array trends (Figure 9d). Notably, M-CSF suppression at IC50 was observed with both WGAR and cisplatin treatment, suggesting that this mediator may reflect a shared cytotoxic-stress–associated response under cytotoxic exposure. However, other array-identified targets relevant to ECM remodeling and invasion (e.g., u-PA and tenascin C) were not validated in the current revision and should be considered preliminary. Further confirmation using orthogonal methods will be required to define the extent to which WGAR modulates tumor microenvironment-related pathways.

4. Materials and Methods

4.1. Plant Materials

The source wild ginseng, a specimen of P. ginseng estimated to be over 100 years old, was authenticated by the Korea Sanwoncho Wild Ginseng Association (dated 5 June 2021). Adventitious roots were established from this wild ginseng through callus culture and subsequently cultivated by MIRAE FARM Agricultural Corporation Co., Ltd. (Gokseong-gun, Jeollanam-do, Republic of Korea) using a smart farm hydroponic system. The product, commercially known as “Nunkkot Sansam” (Snow Flower Wild Ginseng), was cultivated in a modified Murashige and Skoog (MS) liquid medium (Sigma-Aldrich, Cat# M0529, St. Louis, MO, USA) continuous aeration for approximately 60 days. Although the >100-year-old wild ginseng specimen was used only as source material to establish the adventitious root culture, all experiments in this study were performed using extracts prepared from the cultured adventitious roots, not from the original wild ginseng root. The culture medium contained the following macronutrients (mg/L): KNO₃ (1875), CaCl₂ (181.2), MgSO₄ (146.29), (NH₄)H₂PO₄ (225), FeNaEDTA (19.8), and inositol (100); micronutrients (mg/L): H₃BO₃ (5.0), MnSO₄·H₂O (10.0), ZnSO₄·7H₂O (1.0), KI (1.0), Na₂MoO₄·2H₂O (0.10), and CuSO₄·5H₂O (0.2); vitamins (mg/L): nicotinic acid (5.0), pyridoxine HCl (0.5), and thiamine HCl (5.0); and sucrose (30 g/L) as the carbon source.

4.2. Preparation of Extracts

The harvested WGAR were subjected to an extraction procedure designed to compare solvent-dependent extraction efficiencies. The roots were first air-dried and subsequently pulverized into a fine powder using a mechanical grinder. For each extraction condition, 1 g of powdered material was accurately weighed and extracted with 20 mL of one of six ethanol–water solvent systems (0, 20, 40, 60, 80, and 100% ethanol, v/v). Each sample–solvent mixture was placed in an ultrasonic bath and sonicated at 40 °C for 2 h to enhance solvent penetration and facilitate the release of phytochemical constituents. This extraction step was repeated three consecutive times to ensure exhaustive extraction. Following sonication, the combined extracts for each solvent system were filtered to remove insoluble residues, concentrated under reduced pressure, and subsequently dried to a constant weight. The dried extracts were then weighed to determine extraction yield. All extract powders were stored at 4 °C until use.

4.3. Chromatographic and Mass Spectrometric Conditions

LC-MS/MS analysis was performed using a Vanquish UHPLC system coupled with an Orbitrap Exploris 120 mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Chromatographic separation was achieved using an ACQUITY Premier HSS T3 column (4.6 × 100 mm, 1.8 μm; Waters, Milford, MA, USA) maintained at 40 °C with a flow rate of 0.2 mL/min and an injection volume of 4 μL. The mobile phase consisted of (A) water containing 0.1% formic acid and (B) acetonitrile containing 0.1% formic acid. The gradient elution program was as follows: 8–15% B (0–4 min), 15–32% B (4–8 min), 32–53% B (8–11 min), 53–100% B (11–24 min), held at 100% B for 3 min, followed by 100–8% B for 1 min and re-equilibration at 8% B. Prior to analysis, sample solutions were filtered through a 0.22 μm polytetrafluoroethylene (PTFE) membrane (Agilent Technologies, Santa Clara, CA, USA).

4.4. Cell Culture

The human triple-negative breast cancer cell line MDA-MB-231 was obtained from the Korean Cell Line Bank (KCLB, Seoul, Republic of Korea). Cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM; Cat# 11965092, Gibco, Carlsbad, CA, USA) containing 10% fetal bovine serum (FBS; Cat# 16000044, Gibco) and 1% penicillin-streptomycin (100 U/mL penicillin and 100 µg/mL streptomycin; Cat# 15140122, Gibco). Cultures were incubated at 37 °C in a humidified atmosphere of 5% CO₂. Cisplatin was obtained from Sigma-Aldrich (Cat# 1134357) and dissolved in dimethyl sulfoxide (DMSO) to prepare stock solutions. WGAR extract was also dissolved in DMSO and diluted with culture medium to the desired concentrations. The final DMSO concentration did not exceed 0.1% (v/v) in all experiments. Untreated cells in complete medium were used as negative controls.

4.5. Cell Viability Assay

Cell viability was determined using the MTT assay. MDA-MB-231 cells were seeded into 96-well plates at a density of 5 × 10³ cells per well and incubated overnight to allow attachment. Cells were then treated with various concentrations of WGAR extract or cisplatin for 48 h. After treatment, 20 µL of MTT solution (5 mg/mL in phosphate-buffered saline (PBS); Cat# M2128, Sigma-Aldrich) was added to each well, and plates were incubated for an additional 4 h at 37 °C. The resulting formazan crystals were solubilized in 150 µL of DMSO, and absorbance was measured at 570 nm using a microplate reader (Sunrise^®, Tecan, Männedorf, Switzerland). Cell viability was calculated as a percentage relative to untreated control cells. IC50 and IC20 values were determined from dose–response curves.

4.6. AO/PI Staining

Live and dead cells were distinguished using AO (Cat# A6014, Sigma-Aldrich) and PI (Cat# P4170, Sigma-Aldrich) dual fluorescent staining. MDA-MB-231 cells were treated with WGAR extract or cisplatin for 48 h, then stained with AO (excitation 490 nm, emission 515 nm) and PI (excitation 535 nm, emission 617 nm). AO is retained by viable cells, producing uniform green fluorescence throughout the cytoplasm. In contrast, PI is membrane-impermeable in live cells but penetrates damaged membranes in dead cells, binding to nuclear DNA and emitting red fluorescence. Fluorescence images were captured using a fluorescence microscope (Eclipse Ti2, Nikon, Tokyo, Japan), and the percentages of live and dead cells were quantified using ImageJ software (version 1.54; National Institutes of Health (NIH), Bethesda, MD, USA).

4.7. Invasion Assay

The invasive capacity of MDA-MB-231 cells was evaluated using Transwell chambers (Cat# CLS3414, Corning, Lake Placid, NY, USA) with 8 µm pore size polycarbonate membranes. The upper chamber was pre-coated with Matrigel (Cat# CLS354234, Corning) to simulate the extracellular matrix barrier. Cells were seeded in the upper chamber with serum-free medium, while medium containing 10% FBS was added to the lower chamber as a chemoattractant. Following treatment with WGAR extract or cisplatin for 24 h, non-invaded cells remaining on the upper surface were removed using a cotton swab. Cells that had invaded through the Matrigel-coated membrane were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet. The number of invaded cells was counted in five randomly selected fields under a light microscope (IX73, Olympus, Tokyo, Japan) at 100× magnification, and results were expressed as the average number of cells per field.

4.8. Western Blotting

Total protein was extracted from treated cells using Radioimmunoprecipitation assay (RIPA) lysis buffer (Cat# 89901, Thermo Fisher Scientific) supplemented with protease and phosphatase inhibitors. Protein concentrations were determined using the Bicinchoninic acid (BCA) Protein Assay Kit (Cat# 23227, Thermo Fisher Scientific). Equal amounts of protein (30 µg) were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes (Cat# IPVH00010, Millipore, Burlington, MA, USA). After blocking with 5% non-fat milk in Tris-buffered saline containing 0.1% Tween-20 (TBST) for 1 h at room temperature, membranes were incubated with primary antibodies overnight at 4 °C.

The following primary antibodies from Cell Signaling Technology (Danvers, MA, USA) were used at 1:1000 dilution: caspase-9 (Cat# 9502), cleaved caspase-9 (Cat# 7237), caspase-3 (Cat# 9662), cleaved caspase-3 (Cat# 9664), PARP (Cat# 9542), cleaved PARP (Cat# 5625), E-cadherin (Cat# 3195), N-cadherin (Cat# 13116), β-catenin (Cat# 8480), Slug (Cat# 9585), GSK-3β (Cat# 12456), phospho-GSK-3β (Ser9) (Cat# 5558), and β-actin (Cat# 4970). Following primary antibody incubation, membranes were washed with TBST and incubated with horseradish peroxidase (HRP)–conjugated anti-rabbit immunoglobulin G (IgG) secondary antibody (Cat# 7074, Cell Signaling Technology) at 1:2000 dilution for 1 h at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) detection system (Cat# 28980926, GE Healthcare, Chicago, IL, USA) and quantified by densitometric analysis using ImageJ software (version 1.54, NIH). Band intensities were normalized to β-actin as a loading control.

4.9. Proteome Profiler Array

To analyze the differential expression of multiple cancer-related proteins simultaneously, the Proteome Profiler Human XL Oncology Array Kit (Cat# ARY026, R&D Systems, Minneapolis, MN, USA) was employed according to the manufacturer’s instructions. This array enables the parallel detection of 84 cancer-related proteins on nitrocellulose membranes containing capture antibodies spotted in duplicate. Cell lysates (200 µg) from control, WGAR-treated (IC20 and IC50), and cisplatin-treated (IC20 and IC50) MDA-MB-231 cells were prepared using the provided lysis buffer supplemented with protease inhibitors. Membranes were blocked with Array Buffer for 1 h at room temperature and then incubated with diluted cell lysates overnight at 4 °C on a rocking platform. Following washing steps, membranes were incubated with biotinylated detection antibody cocktail for 1 h, followed by streptavidin-HRP for 30 min at room temperature. Protein spots were visualized using chemiluminescent detection reagents and exposed to X-ray film. Spot pixel densities were quantified using ImageJ software (version 1.54, NIH) with a transmission-mode scanner. Background signals were subtracted, and duplicate spot intensities were averaged. Relative protein expression levels were compared between treatment conditions and untreated controls.

4.10. ELISA

To independently validate the proteome array findings, IL-6 and M-CSF levels in cell culture supernatants were quantified using the Human IL-6 Quantikine ELISA Kit (Cat# D6050, R&D Systems) and Human M-CSF Quantikine ELISA Kit (Cat# DMC00B, R&D Systems), respectively, according to the manufacturer’s instructions. MDA-MB-231 cells were seeded in 6-well plates and treated with WGAR (IC20: 3 μg/mL; IC50: 79 μg/mL) or cisplatin (IC20: 2 μg/mL; IC50: 9 μg/mL) for 48 h. Cell culture supernatants were collected, centrifuged to remove debris, and stored at −80 °C until analysis. Cytokine concentrations were calculated from standard curves generated using recombinant human standards. Results are expressed as fold change relative to control and represent mean ± SD from three independent experiments.

4.11. Annexin V Apoptosis Assay

To quantitatively assess apoptosis induction, the Muse Annexin V & Dead Cell Kit (Cat# MCH100105, Luminex Corporation, Austin, TX, USA) was used. MDA-MB-231 cells were treated with WGAR or cisplatin for 48 h. Following treatment, adherent cells were detached using trypsin-EDTA and resuspended in medium containing 1% FBS at a concentration of 1 × 10⁵ to 5 × 10⁵ cells/mL.

Cell suspension (100 μL) was mixed with 100 μL of Muse Annexin V & Dead Cell Reagent and incubated for 20 min at room temperature in the dark. Samples were analyzed using the Muse Cell Analyzer (Luminex Corporation). The assay distinguishes four populations: live cells (Annexin V−/7-AAD−), early apoptotic cells (Annexin V+/7-AAD−), late apoptotic/dead cells (Annexin V+/7-AAD+), and dead cells (Annexin V−/7-AAD+). Results are expressed as percentage of total cells and represent mean ± SD from three independent experiments.

4.12. Caspase Activity Assay

To further validate apoptosis through caspase activation, the Muse MultiCaspase Kit (Cat# MCH100109, Luminex Corporation) was used. MDA-MB-231 cells were treated with WGAR or cisplatin for 48 h. Following treatment, cells were harvested and resuspended in 1× Caspase Buffer at a concentration of 1 × 10⁵ cells/mL.

The MultiCaspase Reagent was reconstituted with 50 μL DMSO and diluted 1:160 with 1× PBS to prepare the working solution. Cells (50 μL) were incubated with 5 μL of MultiCaspase working solution at 37 °C for 30 min. Subsequently, 150 μL of Caspase 7-AAD working solution (2 μL 7-AAD in 148 μL 1× Caspase Buffer) was added. Samples were mixed thoroughly and analyzed using the Muse Cell Analyzer.

The assay distinguishes four cell populations: live cells (Caspase−/7-AAD−), cells exhibiting caspase activity (Caspase+/7-AAD−), late apoptotic/dead cells (Caspase+/7-AAD+), and dead cells (Caspase−/7-AAD+). Results are expressed as percentage of total cells and represent mean ± SD from three independent experiments.

4.13. qPCR

To validate Slug expression at the transcriptional level, quantitative real-time PCR was performed. MDA-MB-231 cells were seeded in 6-well plates and treated with WGAR or cisplatin for 24 h. Total RNA was extracted using the RNeasy Mini Kit (Cat# 74104, Qiagen, Hilden, Germany) according to the manufacturer’s instructions. RNA concentration and purity were assessed using a NanoDrop spectrophotometer (Thermo Fisher Scientific), and samples with A260/A280 ratios between 1.8 and 2.0 were used for subsequent analysis.

Complementary DNA (cDNA) was synthesized from 1 μg of total RNA using the High-Capacity cDNA Reverse Transcription Kit (Cat# 4368814, Applied Biosystems, Waltham, MA, USA). qPCR was performed using SYBR Green Master Mix (Cat# 4367659, Applied Biosystems) on a QuantStudio 3 Real-Time PCR System (Applied Biosystems). The following primers were used: SNAI2 (Slug) forward 5′-CGAACTGGACACACATACAGTG-3′ and reverse 5′-CTGAGGATCTCTGGTTGTGGT-3′; GAPDH forward 5′-GAAGGTGAAGGTCGGAGTC-3′ and reverse 5′-GAAGATGGTGATGGGATTTC-3′.

The thermal cycling conditions were: 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Melting curve analysis was performed to confirm primer specificity. Relative gene expression was calculated using the 2^−ΔΔCt method with GAPDH as the internal reference gene. Results represent mean ± SD from three independent experiments, each performed in technical triplicate.

4.14. Molecular Docking

Molecular docking studies were performed to predict the binding affinities between the identified compounds from WGAR and target proteins. The three-dimensional crystal structures of PARP-1 (PDB ID: 7KK5) and β-catenin (PDB ID: 1JDH) were retrieved from the RCSB Protein Data Bank (https://www.rcsb.org, accessed on 20 December 2025). The 3D conformers of compounds (83–87, 90–95, and 98–103) tentatively identified from LC-MS/MS analysis were obtained from the PubChem and ChemSpider databases.

Protein and ligand preparation was performed using MGL Tools (The Scripps Research Institute, La Jolla, CA, USA), and docking scores were calculated using the AutoDockTools-1.5.7 algorithm. The scoring function incorporates van der Waals forces, electrostatic interactions, hydrogen bonding, and alkyl interactions for accurate prediction of ligand-binding modes and affinities. Grid box dimensions for PARP-1 were defined based on previously reported binding-site residues, including HIS862, GLY863, GLY888, TYR889, TYR896, ALA898, LYS903, SER904, and TYR907. The docking protocol was validated by redocking the native ligands into their respective proteins; the RMSD between the native ligand (3JD) and redocked ligand in PARP-1 was 0.30 Å, confirming the reliability of the protocol. Binding affinities and ligand–protein interactions were visualized using Discovery Studio Visualizer 2025 Client (BIOVIA, San Diego, CA, USA).

4.15. Statistical Analysis

All statistical analyses were conducted using GraphPad Prism version 9.5.1 (GraphPad Software, San Diego, CA, USA). Experiments were independently repeated three times (n = 3), and results are expressed as mean ± standard deviation (SD). Parametric one-way ANOVA was applied based on its robustness to minor deviations from normality in balanced designs with equal group sizes. For comparisons of multiple treatment groups against the untreated control, Dunnett’s post hoc test was used; for comparisons among all groups, Tukey’s post hoc test was applied. Statistical significance was defined as * p < 0.05, ** p < 0.01, and *** p < 0.001. Exact p-values are reported for key comparisons in the figure legends.

5. Conclusions

The present study demonstrates that WGAR induces apoptosis in MDA-MB-231 TNBC cells via the intrinsic mitochondrial pathway, as evidenced by caspase-9/caspase-3 activation and PARP cleavage. WGAR was associated with modulation of EMT-related markers alongside alterations in GSK-3β/β-catenin signaling; however, because invasion was significantly reduced only at IC50, interpretation of cytotoxicity-independent anti-invasive effects warrants further validation. WGAR was also associated with decreased levels of tumor microenvironment-related proteins (e.g., IL-6, u-PA, tenascin C); because these changes were identified by proteome array, they represent screening-level observations requiring orthogonal confirmation. Chemical profiling by LC-MS/MS tentatively identified 17 compounds, with ginsenosides (Rg3, Rh1, Rf) and terpenoids (ursolic acid) as putative major constituents. Docking analyses predicted potential interactions with PARP-1 and β-catenin; however, these in silico findings are hypothesis-generating and require biochemical validation. In conclusion, WGAR demonstrates multi-target anticancer potential against TNBC cells, and adventitious root culture technology may facilitate sustainable production for further therapeutic development.