Global gene expression profiling of clinical breast cancer has provided molecular classification of subtypes based of differential expression of genes for hormone and growth factor receptors (1). This molecular classification dictates specific conventional chemo-endocrine therapy or pathway selective small molecule inhibitor based targeted therapy. Thus, for the hormone receptor-positive/human epidermal growth factor receptor-2 (HER-2)-positive luminal B molecular subtype, treatment with selective estrogen receptor modulators, aromatase inhibitors and HER-2 inhibitors represent common options (2). By contrast, for the hormone receptor-negative/HER-2-positive HER-2 enriched subtype HER-2-targeted therapy represents a viable option (3).

Long-term treatment with conventional and/or targeted therapy is frequently associated with systemic toxicity and acquired tumor resistance that collectively compromise patient compliance and favor therapy-resistant tumor progression predominantly due to the emergence of drug-resistant cancer stem cells (4).

In preclinical investigations, human breast carcinoma-derived BT474, MDA-MB-361 and MCF-7/HER cell models for the luminal B subtype, and SKBr-3 and MDA-MB-435 cell models for the HER-2-enriched subtype (5,6), represent valuable experimental systems with which to elucidate the mechanisms of drug resistance and for the identification of efficacious lead compounds. Multiple drug resistance via acquired resistance to HER-2/EGFR-selective small molecule inhibitors in the HER-2-enriched model (3) or cross-resistance to aromatase inhibitors in the MCF-7^AROM model (7,8) represent additional approaches. Recent advances in the development of human cancer models include genetically engineered mouse models and constituent cells lines, patient-derived xenografts and constituent organoid models, and drug-resistant cancer stem cell models. Collectively, these approaches promise the clinical translation of preclinical data.

Acquired drug resistance leads to the lack of a response to conventional and/or targeted chemo-endocrine therapy, and is frequently is associated with therapy-resistant cancer stem cells that are characterized by pluripotency, tumor-initiating potential and the upregulated expression of several cell surface proteins and nuclear transcriptional factors (9).

The limitations of chemo-endocrine and/or targeted therapy emphasize an unmet need to develop clinically relevant cancer stem cell models and identify efficacious testable alternatives for stem cell targeted therapy of chemo-endocrine therapy resistant breast cancer.

The present review summarizes a comparison of recently generated unpublished data with that from previously published data on the following: i) Growth inhibitory efficacy of mechanistically distinct natural phytochemicals on a cellular model for the HER-2-enriched breast cancer subtype; ii) drug-resistant stem cell model for HER-2-enriched breast cancer; and iii) proof of concept for efficacy of natural phytochemicals as testable alternatives for drug resistant breast cancer stem cells.

These agents are selected based on their documented chemo-preventive efficacy in preclinical models of epithelial organ site cancers. To ascertain the feasibility of the experimental approach, select phytochemicals, such as cruciferous glucosinolate indole-3-carbinol (I3C), tea polyphenol epigallocatechin gallate (EGCG), soy isoflavone genistein (GEN) and rosemary terpenoids carnosic acid (CA) and carnosol (CSOL), are tested on the non-tumorigenic 184-B5 cells. Subsequently these agents are tested for their growth inhibitory efficacy on the 184-B5/HER model. Compared to the 184-B5 model, the 184-B5/HER model exhibits substantially greater growth inhibitory efficacy. At the mechanistic levels, I3C alters the cellular metabolism of estradiol to generate the anti-proliferative metabolite 2-hydroxyestrone (14). EGCG and GEN inhibit the expression of proliferative cell nuclear antigen (PCNA) and cyclin D1, and increase the expression of the cyclin-dependent kinase inhibitor (CDKI) p16^INK4 (15). CA and CSOL inhibit tyrosine kinase expression and upregulate p16^INK4 expression (15), induce cytostatic G₂/M arrest and accumulate G₂-specific cyclin B1(16). CSOL also inhibits the expression of inducible cyclo-oxygenase-2 (COX-2) via the protein kinase C/mitogen-activated protein kinase (PKC/MAPK) pathway (17).

The data on the status of hyper-proliferation in 184-B5/HER model are summarized in Table I. In comparison with the non-tumorigenic 184-B5 cells, the hyper-proliferative 184-B5/HER cells exhibit an approximately 55% reduction in population doubling times, an approximately 63% increase in saturation density, an approximately 96% reduction in the sub-G₀ (apoptotic) population and a substantial increase in anchorage-independent colony formation, the latter being an established in vitro surrogate end point for in vivo tumor development (16). In addition, the 184-B5/HER cells exhibit an approximately 178-fold increase in HER-2 expression and an approximately 231-fold increase in pHER-2 expression.

The data on the growth inhibitory efficacy of natural phytochemicals is summarized in Table II. Treatment with individual phytochemicals at their respective maximum cytostatic (IC₉₀) concentrations resulted in a substantial increase in the G₁:S+G₂/M ratio ranging from approximately 43% for CA to approximately 3-fold for GEN. This increase in the G₁:S+G₂/M ratio is indicative of cytostatic growth arrest. At the mechanistic level, phytochemical treatment resulted in the downregulated expression of p-HER-2, ranging from approximately 20% for CA to approximately 85% for GEN. In addition, these data facilitate the rank ordering of natural phytochemicals based on their efficacy for biomarker modulation. Thus, with regard to the G₁:S+G₂/M ratio, the rank order is GEN>I3C>EGCG>CA=CSOL. For pHER-2 inhibition the rank order is GEN>EGCG>I3C>CSOL>CA. Thus, the rank order suggests distinct efficacy for p-HER-2 inhibition that may be responsible for the observed effect on cell cycle progression.

The data on the pro-apoptotic effects of the phytochemicals are summarized in Table III. Treatment with the phytochemicals results in a substantial increase of the cell population in the sub-G₀ (apoptotic) phase of the cell cycle ranging from approximately 8-fold for I3C to approximately 17-fold for GEN, relative to the solvent control. At the mechanistic level, the induction of cellular apoptosis was associated with a reciprocal modulation of apoptotic-specific proteins due to the reduced expression of the anti-apoptotic BCL-2 protein, ranging from approximately 23% for I3C to approximately 67% for GEN, and the increased expression of pro-apoptotic BAX protein, ranging from approximately 23% for I3C to approximately 76% for CSOL (Table IV). Thus, with regard to the induction of apoptosis, the rank order is GEN>CSOL>EGCG>CA>I3C. The rank order for the decrease in BCL-2 expression is GEN>CSOL>CA>EGCG>I3C, while the rank order for the increase in BAX expression is CSOL>GEN>CA>EGCG>I3C. Collectively, the rank order for the efficacy for induction of cellular apoptosis indicates distinct mechanisms of action for individual phytochemical. It is also notable that the growth inhibitory effects of these phytochemicals have been reported in the HER-2-negative human breast carcinoma derived cell lines, MCF-7, MCF-10 CA 1a and MDA-MB-231 (18-21). However, the effective concentrations differ from the ones presented for the 184-B5/HER model.

In the preclinical models for HER-2-enriched breast cancer, the HER-2-specific antibody, trastuzumab, and the HER-2/EGFR dual-specific small molecule inhibitors display acquired drug resistance (22,23). Consistent with these data, the 184-B5/HER cells treated with the HER-2/EGFR small molecule inhibitor, lapatinib (LAP), provided the drug-resistant stem cell model. Long-term treatment with the maximum cytostatic (IC₉₀) concentration of LAP generated progressively growing LAP-R cells. The data summarized in Table V demonstrate an approximately 1-fold increase in the number of tumor spheroids, an approximately 3-fold increase in cluster of differentiation (CD44), a 3-fold increase in DNA binding homeobox nuclear transcription factor (NANOG) and a 2-fold increase in octamer binding transcription factor-4 (OCT-4), relative to that observed in LAP-sensitive (LAP-S) cells. The cell surface marker, CD44, and the nuclear transcription factors, NANOG and OCT-4, represent well-established markers of the stem cell population (9). Thus collectively, these data indicate that the LAP-R cells may represent putative cancer stem-like cells in the present model.

Frequent occurrence of acquired tumor resistance against pharmacological therapeutic agents (2,3,7,8) and the emergence of drug-resistant cancer stem cells (22,23) emphasizes an unmet need to identify stem cell targeting testable alternatives. Unlike pharmacological agents, natural phytochemicals, due to their low toxicity profiles, are less likely to induce acquired tumor resistance. Published studies on the rosemary terpenoids, CA and CSOL, have documented potent anti-proliferative and/or pro-apoptotic effects at cytostatic low µM concentrations on the present HER-2-enriched model (16). It is therefore of considerable relevance to examine the stem cell targeted efficacy of natural phytochemicals. The data presented in Table VI demonstrate that in response to a treatment with CSOL the LAP-R cells exhibit an approximately 87% decrease in the number of tumor spheroids, an 86% decrease in CD44 expression, a 77% decrease in NANOG and an 80% decrease in OCT-4 expression, relative to the solvent-treated controls. Collectively, these data provide a proof of concept for a mechanistic lead for the rosemary terpenoid, CSOL, as a stem cell targeting natural alternative. With regard to the cancer stem cell-targeted efficacy of natural products, it is noteworthy that several mechanistically distinct natural products have documented efficacy against drug resistant stem cell population. For example, the vitamin A derivative, all-trans retinoic acid (ATRA), inhibits gastric cancer stem cell growth via inhibiting CD44, aldehyde dehydrogenase-1 (ALDH1), Kruppel-like factor-4 (KLF-4) and sex determining region Y-box-2 (SOX-2) expression (24). Sulphoraphane present in broccoli inhibits tumor spheroid formation and ALDH-1 expression in pancreatic and prostate cancer stem cells (25), inhibits NANOG, ALDH-1, Wnt-3 and Notch expression in triple-negative breast cancer stem cells (26), and benzyl isothiocyanate present in cruciferous vegetables inhibits breast cancer stem cells via upregulation of KLF-4-p21 axis (27). The inhibition of established stem cell markers provides potential mechanistic leads for the stem cell-targeted efficacy of natural products.

In addition to natural phytochemicals, nutritional herbs may represent testable alternatives for therapy resistant breast cancer. Herbal aqueous extracts are extensively used in traditional Chinese medicine for a variety of health issues in women, including cancer (28). Non-fractionated aqueous extracts from Chinese nutritional herbs, simulating patient consumption, exhibit anti-proliferative and pro-apoptotic effects in cellular models for molecular subtypes for clinical breast cancer. In the MCF-7 model for the luminal A subtype, Epimedium grandiflorum extract inhibits cancer growth via altering estrogen metabolism and generating anti-proliferative metabolite 2-hydroxyestrone (29). In the MCF-7^AROM model for aromatase-expressing post-menopausal breast cancer, Taheebo-NFD-Marugoto (TNM), an extract from the Tabebuia avellanedae tree, exhibits anti-proliferative effects via inhibition of estrogen regulated genes and pro-apoptotic effects via increase in pro-apoptotic caspase 3/7 activity and upregulation of BAX gene. Additionally, TNM exerts superior inhibitory effects on aromatase activity than clinical aromatase inhibitors (30). In the MDA-MB-231 model for triple-negative breast cancer, Cornus officinalis extract induces anti-proliferative and pro-apoptotic effects via the inhibition of cyclin D1 and pRB, and via the upregulation of caspase-3/7 activity and BAX expression, respectively (31). Collectively, these data provide mechanistic leads for the efficacy of nutritional herbs on models for breast cancer subtypes, and thereby, provide a rationale for investigations on developed stem cell models. It is also notable that the stem cell-specific transcription factors, OCT-4, SOX-2 and NANOG, represent markers for poor survival in HER-2-positive clinical breast cancer (32).

The data reviewed herein lead to the conclusion that the cellular model for the HER-2-enriched breast cancer subtype in general and drug-resistant stem cell model in particular, provide valuable experimental approaches for identifying natural phytochemicals as testable alternatives for treatment of chemotherapy therapy resistant breast cancer.

Human tissue-derived cellular models represent valuable experimental approaches to reduce preclinical data extrapolation for their clinical translatability. However, to further enhance clinical translatability, future investigations utilizing patient derived ex vivo breast cancer organoid models (33,34) are likely to provide valuable clinically translatable data on stem cell-targeting natural products.