Selective concomitant inhibition of mTORC1 and mTORC2 activity in estrogen receptor negative breast cancer cells by BN107 and oleanolic acid

Abstract

Hormonal, targeted and chemotherapeutic strategies largely depend on the expression of their cognate receptors and are often accompanied by intolerable toxicities. Effective and less toxic therapies for estrogen receptor negative (ER−) breast cancers are urgently needed. Here, we present the potential molecular mechanisms mediating the selective pro-apoptotic effect induced by BN107 and its principle terpene, oleanolic acid (OA), on ER− breast cancer cells. A panel of breast cancer cell lines was examined and the most significant cytotoxic effect was observed in ER− breast lines. Apoptosis was the major cellular pathway mediating the cytotoxicity of BN107. We demonstrated that sensitivity to BN107 was correlated to the status of ERα. Specifically, the presence of functional ERα protected cells from BN107-induced apoptosis and absence of ERα increased the sensitivity. BN107, an extract rich in OA derivatives, caused rapid alterations in cholesterol homeostasis, presumably by depleting cholesterol in lipid rafts (LRs), which subsequently interfered with signaling mediated by LRs. We showed that BN107 or OA treatment in ER− breast cancer cells resulted in rapid and specific inhibition of LR-mediated survival signaling, namely mTORC1 and mTORC2 activities, by decreasing the levels of the mTOR/FRAP1, RAPTOR and RICTOR. Cotreatment with cholesterol abolished the proapoptotic effect and restored the disrupted mTOR activities. This is the first report demonstrating possible concomitant inhibition of both mTORC1 and mTORC2 activities by modulating the levels of protein constituents present in these signaling complexes, and thus provides a basis for future development of OA-based mTOR inhibitors.

Despite the favorable advances that treatment options have had on survival, current regimens still lead to toxic side effects and are mostly ineffective against ER− metastatic breast cancer. Currently, patients with ER−/progesterone receptor negative (PR−)/HER2 negative (Her2−) tumors still represent a therapeutic challenge for oncologists. Novel and effective therapies with minimal toxicities are urgently needed for this patient population.

The anticancer effect of the fruit of Gleditsia sinensis Lam (G. sinensis) has been attributed to the induction of cytotoxicity.1–5 BN107 is an aqueous extract of G. sinensis that has been shown to exhibit antiproliferative activity on a panel of human breast cancer lines.4G. sinensis contains several terpenoidal saponins that possess a similar base structure to oleanolic acid (OA).5 These oleanane saponins have been shown to exhibit differential cytotoxicities against tumor cells, which depend greatly on the presence and position of the oligosaccharide moieties and the monoterpene units.5 In addition, OA and its synthetic derivatives have been shown to induce strong antitumor activity in a wide variety of tumor cells in culture and in animal models.5–7

The physiological activity of terpenoidal saponins is usually associated with their ability to complex with plasma membrane cholesterol.8 It is now well established that cholesterol is important for the functions of lipid rafts (LRs) and that agents which bind to and/or extract cholesterol from the rafts alter the localization and function of the raft-associated proteins.9–11 LRs are sites where cell surface receptors and signaling molecules are concentrated and which spatially organize signal transduction at the cell surface.12 Some proteins selectively partition into the LRs, including glycosylphosphatidylinositol-anchored proteins, myristoylated or palmitoylated proteins (such as Akt, flotillin),13, 14 doubly acylated proteins (such as Src-family kinases), phospholipid bound proteins (such as annexins), and cholesterol-bound transmembrane proteins (such as caveolins).12 In addition, LRs have been shown to provide a “platform” for proper assembly of functional protein complexes: mTOR activities were shown to be dependent on the presence of the complex components on LRs.15, 16

The Akt/mammalian target of rapamycin (mTOR) pathway is the prototypic survival mechanism that is aberrantly activated in many types of cancer, including ER− breast cancer.17 This pathway is central in the transmission of the growth regulatory and survival signals originating from cell surface receptors.18, 19 Growth factors and cytokines activate Akt via phosphatidylinositol-3-kinase (PI3K), which phosphorylates phosphatidylinositol-4,5-bisphosphate [PI(4,5)P2] to generate [PI(1,4,5)P3] that binds to the PH domain of Akt and phosphoinositide-dependent protein kinase 1 (PDK1), recruiting them to the plasma membrane. Once in the membrane, Akt is phosphorylated by PDK1 at Thr 308 and by mTORC2 at Ser 473.20 When Akt is fully activated, signaling through activated Akt can be propagated to a diverse array of substrates, including mTORC1,21 a key regulator of protein translation.19

mTOR/FRAP1 is an atypical serine/threonine kinase that belongs to the PI3K-like kinase family and regulates a variety of cellular activities that are sensitive to environmental stress.18, 22, 23 In mammalian cells, two mTOR/FRAP1-containing complexes have been identified, mTORC1 and mTORC2. mTORC1 is comprised of mTOR/FRAP1, RAPTOR and mLST8, while mTORC2 contains mTOR/FRAP1, RICTOR, mLST8, sin1 and protor.23–25 mTORC1 phosphorylates p70S6K (ribosomal p70S6 kinase) and 4E-BP (eukaryotic initiation factor 4E binding protein), both regulators of protein translation.22 Activated-mTORC2 regulates the actin skeleton and phosphorylates the hydrophobic motif of SGK1 at Ser42226, 27 and Akt at Ser473,23, 28 which in conjunction with PDK1-mediated phosphorylation (Thr308) drives full activation of Akt.29, 30 It has been proposed that mTOR/FRAP1 polypeptide and other complex components of mTORC1 and mTORC2 reside on the LRs and mTOR/FRAP1 polypeptide is shared between mTORC1 and mTORC2 complexes.16 Recently, Copp et al. showed that mTOR/FRAP1 is phosphorylated differentially when associated with mTORC1 and mTORC2 complexes. Specifically, mTORC1 contains predominately phosphorylated mTOR/FRAP1 on Ser2448; whereas mTORC2 contains mTOR/FRAP1 phosphorylated predominantly on Ser2481.31

Given that aberrant mTOR activities are critical for tumor cells to grow and survive, efforts have been directed toward developing potential cancer therapies which inhibit mTOR activity.18, 30 mTOR inhibitors, rapamycin and its derivatives (rapalogs) have been developed to target mTORC1 complex. Some rapalogs have recently demonstrated significant activity in the treatment of metastatic renal cell carcinoma.32 Activities against other solid tumors, including breast, are not as impressive.18, 30 The molecular mechanisms responsible for these differences in sensitivity have not yet been clearly identified. Rapalogs blocking mTORC1 were shown to exert direct antiproliferative effects. However, the overall antitumour effects of rapalogs appear to be more complex than just inhibition of tumour growth and could be a result of their apoptotic-inducing effects in carcinoma cells along with the inhibition of tumour angiogenesis. Despite extensive cognitive research, it is difficult to appraise which of those mechanisms is predominant in patients.33 Evidence shows that mTORC1 inhibition can lead to pathway reactivation: abrogation of the negative-feedback loop which is normally initiated by the direct mTORC1 substrate p70S6k that can lead to strong PI3K-Akt reactivation (chemoresistance). Moreover, rapalogs cannot efficiently inhibit mTORC2 in certain cell types.34 Together, this would suggest that pathway activation and reactivation could be avoided by agents that lead to concomitant inhibition of mTORC1 and mTORC2.35 To this end, we report here the molecular mechanism through which BN107 and OA induce apoptosis selectively in ER− breast cancer cells by simultaneous inhibition of both mTORC1 and mTORC2 complexes.

4E-BP: eukaryotic initiation factor 4E binding protein; Akt: v-akt murine thymoma viral oncogene homolog 1; CHL: cholesterol; CAV-1: caveolin 1; ER−: estrogen receptor negative; ERα: estrogen receptor α; FOH: farnesol; GGOH: geranylgeraniol; GM-1: monosialotetrahexosylganglioside; GAPDH: glyceraldehyde-3-phosphate dehydrogenase; Her2: v-erb-b2 erythroblastic leukemia viral oncogene homolog 2; LacZ: β-galactosidase containing plasmid; LRs: lipid rafts; non LR PM: non-lipid raft plasma membrane; LDL: low-density lipoprotein; mTOR/FRAP1: mammalian target of rapamycin/ FK506 binding protein 12-rapamycin associated protein 1; mTORC1/2: mammalian target of rapamycin complex 1/2; MTP: mitochondrial transmembrane potential; MβCD: methyl-β-cyclodextrin; OA: oleanolic acid; p70S6K: ribosomal p70S6 kinase; PI3K: phosphoinositide-3-kinase; PR: progesterone receptor; Raptor: regulatory associated protein of mTOR; Rictor: rapamycin-insensitive companion of mTOR; SGK1: serum glucocorticoid-induced protein kinase 1; TsA: trichostatin A; TR: transferrin receptor

Material and Methods

Reagents and antibodies

BN107 is an aqueous preparation of the grounded fruit of Gleditsia sinensis (Sichuan Medicines and Health Products, Chengdu, China).4 All chemicals were purchased from Sigma (St Louis, MO) unless noted otherwise. The following antibodies were purchased from Cell Signal except noted : phospho-mTOR, total mTOR, RICTOR, RAPTOR, phosho-4EBP, total 4EBP, pS6 kinase, total S6 kinase, phospho-AKT, total AKT (Cruz Technology, Santa Cruz, CA), Caveolin 1 (BD Biosciences, San Jose, CA), CD44 (Epitomics, Burlingame, CA), ERα (Upstate Biotechnology, Temecula, CA) and cytochrome C (Biovision, Mountain View, CA). GM-1 was detected by using subunits of cholera toxin (CT) subunit B conjugated to HRP (Invitrogen, Carlsbad, CA).

Cell cultures and treatments

All the cell lines used were purchased from ATCC (Manassas, VA). Cells were treated with 70 μg/ml of BN107, 110 μM of OA, 500 μM of cholesterol or as indicated in the text and/or legends to Figures.

ERα transduction

MDA-MB-231 cells were transduced with adenovirus particles (ViraQuest, North Liberty, IA) expressing LacZ or ERα on day 0. Infected cells (300,000) were trypsinized and plated in the presence of 10 nM estradiol per well in 6-well plate on Day 1. Treatment of these cells began 8 hr later on Day 1 and continued for 16 hr.

ERα knockdown

Chemical knockdown—a selective estrogen receptor down-regulator ICI182,780 (1 μM) was used to induce degradation of ERα protein in MCF7 cells in RPMI1640 media containing regular fetal bovine serum for 3 days. The cells were then treated for additional 3 days in 4% charcoal-stripped serum containing phenol-red free RPMI-1640 media to deplete estradiol, followed by 16 hr of BN107 treatment. Cells were harvested and subjected to Annexin V/PI analysis. Small hairpin RNA (shRNA) stable knockdown—various shRNA constructs in the pGIPZ vector specifically targeting ERα were purchased from Open Biosystem (ThermoScientific, Huntsville, AL). The following shRNAs targeting the following sequences of ERα were used: shRNA 2-CTTATTGTCTGTAATTGAA, shRNA4-TTACAGACAATAAGGTCAC and shRNA9-GACTTACTGATAATTTACT. Lentivirus was produced as described.36 Transduced-MCF7 cells were selected in 2.5 μg/ml of puromycin for one week to allow efficient knockdown. Two populations of transduced-MCF7 cells exhibited significant ERα knockdown (shRNA 2 and shRNA 4) and 1 shRNA 9, did not produce appreciable ERα knowdown. These 3 MCF7 pooled populations along with a selected population infected with vector only lentivirus were depleted of estradiol and treated with BN107 as described earlier. Cell viability was assessed using MTT assays after 16 hr of treatment.

Cell death/apoptosis measurement

Apoptosis was quantified using flow cytometry analysis of AnnexinV/PI-stained cells (Invitrogen, Carlsbad, CA). MTP was determined using JC-1 loading (Invitrogen, Carlsbad, CA) of live cells and analyzed using flow cytometry. Caspase 3 and 9 activities were measured using specific caspase peptide inhibitors (Calbiochem, San Diego, CA) conjugated to FITC followed by measurement with fluorescence plate reader (SpectraMax, Sunnyville, CA). Cytosolic cytochrome C release was examined in cytosol versus mitochondrial fractions (Biovision kit, Biovision, Mountain View, CA) followed by immunoblotting. DNA fragmentation was analyzed following manufacturer's instruction (DNA laddering Kit, Roche, Indianapolis, IN). MTT assay was used to assess cell survival according to manufacturer's instruction (Invitrogen).

Immunofluorescence

MDA-MB-231 cells were plated in 8-chamber slides on day 0. The cells were treated with BN107 for 4 hr and fixed with methanol: acetone (1:1) at −20°C for 5 min. Cells were rinsed in phosphate-buffered-saline (PBS) and blocked in 2% bovine serum albumin (BSA) in PBS for 1 hr before applying anti-caveolin (1/1000) or anti-CD44 (1:250) in 2% BSA in PBS overnight at 4°C. The chamber slides were rinsed with PBS and incubated with appropriate Alexa 488 conjugated secondary antibody for one hour. The nuclei were stained with 1 μg/ml Hoechst 33258 for 5 min, and the slides were mounted with Fluoromount-G (Southern Biotech, Birmingham, AL) before viewing.

Cholesterol determination

Total cellular cholesterol levels were determined by lysing cells in lysis buffer (50 mM TrisHCl pH 7.4, 150 mM NaCl, 1 mM ethylenediaminetetraacetic acid (EDTA), 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 1% triton) and extracted using chloroform (3 times). Cholesterol content in the LRs was determined by analyzing fractions enriched in GM-1, which were subjected to chloroform extraction (3 times). The pooled organic phase was dried down and subjected to vacuum drying. The Amplex Red cholesterol assay kit was used to quantitate the amount of cholesterol and cholesterol ester in the samples (Invitrogen).

Lipid raft isolation

A modified procedure for density gradient centrifugation using Nycodenz was used to fractionate Triton X-100-solubilized cell lysate.33 The fractions were dialyzed against PBS to remove the gradient sugars and concentrated using Amicon Ultra 4 centrifugal filter device (Millipore, Jeffrey, New Hampshire) before protein quantitation (BCA reagent, ThermoFisher, Pittsburgh, PA). The enrichment of GM1 in fractions was tested using horseradish peroxidase-conjugated cholera toxin B subunit (Invitrogen, Carlsbad, CA) and dot blot analysis.

Results

BN107 selectively induces apoptosis in ER− breast cancer cells

We analyzed the cytotoxicity of BN107 on a panel of human breast cell lines and investigated whether there was a correlation between genotypic characteristics of the cells and their sensitivity to BN107. Table 1 shows that cells lacking ER expression were highly sensitive to BN107, while cells containing functional ER were relatively insensitive to BN107 at the same dose (70 μg/ml). Cytotoxicity from BN107 in ER− lines was apoptotic and primarily mediated by the mitochondrial pathway, as evidenced by mitochondrial transmembrane potential (MTP) dissipation, caspases 3 and 9 activation, cytochrome C release into cytosol, Annexin V binding and DNA fragmentation (Figs. 1a–1c, and Supporting Information Figs. 1a and b).

Figure 1. BN107 induces apoptosis in ER− breast cancer cells, assessed by (a) mitochondrial transmembrane potential (MTP) assessed by JC-1 staining, (b) activation of caspases 3 and 9, and (c) Western blot showing cytosolic Cytochrome C release. All cells (Hs578T, MDA-MB-231 and MCF7) were treated with BN107 (70 μg/ml) and harvested after 6 (a), 16 (b) or indicated hours (c) of treatment. Data are represented as mean ± SEM (b). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Table 1. Cells without ER are more sensitive to BN107-induced apoptosis, while Her2 status appears not correlative with BN107 sensitivity

To evaluate if functional ER plays a protective role in BN107-induced apoptosis, we expressed ERα via adenoviral transduction in MDA-MB-231 cells that are null for ER expression and highly sensitive to BN107. Figure 2a shows ERα protein expression after transduction and Supporting Information. Figure 2a shows expression of RNA of WISP2, an ER responsive gene that is indicative of functional ER status. Figure 2b shows that ERα versus control lacZ expression in MDA-MB-231 cells conferred significant protection from BN107-induced apoptosis.

Figure 2. ERα expression rescues MDA-MB-231 cells from BN107-induced apoptosis. (a) Western showing ERα expression in MDA-MB-231 cells infected with LacZ or ERα adenovirus. (b) These cells were treated with 60 μg/ml of BN107 for 18 hr in the presence of 10 nM estrogen and analyzed with Annexin/PI binding. The chart shows percentage of Annexin- PI- (live) cells (mock-treated □, BN107-treated ▪, # denotes p < 0.01 compared to BN107-treated ERα expressing cells, * denotes p > 0.05 compared to mock-treated ERα expressing cells) and (c) MDA-MB-231 cells were treated with 50 nM TsA or DMSO for 2 days. The cells were then treated with BN107 and analyzed with Annexin/PI binding as in b (** denotes p < 0.01 compared to TsA-BN107 treated). Data are represented as mean ± SEM (b and c).

We treated parental MDA-MB-231 cells with differentiating agent trichostatin A (TsA) in an attempt to reverse their mesenchymal phenotypes and induce expression of the epithelial markers. For example, MDA-MB-231 cells have been shown to re-express ERα, E-cadherin and CD24, and to downregulate CD44, caveolin and vimentin expression on prolonged treatment with low-dose TsA (50 nM), which does not cause cell death.37 We examined the levels of RNA and protein expression for these genes in MDA-MB-231 cells after treatment with TsA for 2 days and confirmed that these cells indeed express ERα and CD24 and downregulate CD44 (Supporting Information Table 1, and Supporting Information Figs. 2b and 2c). We then treated these TsA-differentiated MDA-MB-231 cells with BN107 and showed that these cells were more resistant to BN107 (Fig. 2c), consistent with our hypothesis that ERα plays a protective role against BN107-induced apoptosis.

To further examine the role of ERα in protecting breast cancer cells from BN107-induced death, we investigated the effect of knocking down ERα expression in ER+ MCF7 cells. We approached this question using 2 experimental strategies. First, we attempted to eliminate ERα protein using ICI 182,780, which has been shown to specifically degrade ERα RNA and protein.38 This approach yielded results that show elimination of ERα protein followed by removal of estradiol from the medium leads to enhanced sensitivity of MCF7 to BN107 (Figs. 3a and 3b).

Figure 3. ERα knockdown increases the sensitivity of MCF7 cells to BN107. (a) Western showing ERα expression in MCF7 cells treated with ethanol (ET), ICI (1 μM) or estradiol (E2, 10 nM). (b) MCF7 cells as in (a) were depleted with estradiol, treated with BN107 for 16 hr and analyzed with Annexin/PI binding. The chart shows % survival. (c) Western blot showing ERα expression in MCF7 pool populations transduced with various shRNA constructs targeting ERα. (d) MCF7 ERα shRNA pooled populations were depleted with estradiol, treated with BN107 for 16 hr, and cell survival was analyzed using the MTT assay. The chart shows % cell survival normalized to mock-treated cells. (* denotes p < 0.01 compared to ETOH or E2 treated cells, ** denotes p < 0.01 compared to Vector or ERα shRNA9 transduced cells).

The second approach involved stable knockdown of ERα using shRNAs specifically targeting ERα. MCF7 cells were infected with lenti-virus expressing shRNA constructs specifically targeting ERα and selected to generate transduced populations of MCF7 cells. We analyzed the ERα expression after one week of selection since ERα protein was shown to be relatively stable (data not shown and Ref. 39). To rule out off-target effect, we used 2 different hairpin sequences targeting ERα. MCF7 ERα shRNA transduced pools 2 and 4 showed specific knockdown of ERα protein (Fig. 3c). In addition, we used ERα shRNA pool 9 that did not produce appreciable ERα knockdown (Fig. 3c). We then depleted estradiol from these MCF7 populations along with a vector control transduced population, and treated them with BN107. Consistent with the result obtained from chemical depletion with ICI 182780, we showed that ERα knockdown in MCF7 cells followed by estradiol depletion increased the sensitivity of these cells to BN107-induced death (Fig. 3d).

Cholesterol depletion induced by BN107 could potentially be responsible for its proapoptotic effect

To investigate the underlying mechanism of the proapoptotic effect of BN107 in ER− breast cancer cells, we performed expression array analysis on Hs578T and MCF-7 cells that had been treated with BN107 for 4 hr (Supporting Information Tables 2 and 3). Expression profiles were analyzed using Ingenuity Pathway Analysis (IPA, version 7) to identify potential cellular pathways collectively responsible for BN107-induced death. IPA analysis between these 2 BN107-treated cell lines revealed distinct patterns of gene expression in response to BN107. Specifically, ER− Hs578T breast cancer cells responded to BN107 by upregulating genes involved in cell death, oxidative stress response, MAPK signaling and cholesterol synthesis/uptake pathways. In addition, statistically significant overlapping patterns of gene expression were observed in BN107- or OA-treated ER− MDA-MB-231 cells (p = 4.8 e −137 and p = 2.1 e −70, respectively, data not shown). Conversely, ER+ MCF-7 breast cancer cells responded by regulating a relatively small set of genes involved in growth receptor and survival signaling.

Array expression data for Hs578T cells exhibited upregulation of cholesterol synthesis and transport genes, suggesting that cholesterol or intermediates of the cholesterol synthetic pathway might play a role in the proapoptotic effect of BN107. The cholesterol synthetic pathway provides isoprenoid precursors that are important for the functions of several signaling proteins;40 therefore, we investigated whether exogenously added isoprenoid precursors, farnesol (FOH) and geranylgeraniol (GGOH),41 could rescue cells from BN107-induced death. As shown in Figure 4a, preincubation of Hs578T cells with FOH or GGOH did not protect cells from BN107-induced apoptosis, implying that isoprenoid starvation was not the underlying cause of death. Similar results were obtained with MDA-MB-231 cells cotreated with BN107 and these isoprenoid precursors (data not shown).

Figure 4. Cholesterol depletion in the LRs is potentially the mechanism mediating the pro-apoptotic effect of BN107 in ER− breast cancer cells. (a) Percent live cells in Hs578T cells pretreated with 50 μM isoprenoid precursors, FOH or GGOH, followed by treatment with BN107 for 18 hr. (b) Cholesterol content in sucrose-density fractions collected from MDA-MB-231 (control □, BN107-treated ▪) or MCF7 (control , BN107-treated ) cells treated with BN107 for 4 hr. LR, lipid rafts (Fractions 3–5), non-LR plasma membrane (Fractions 6). (c) Percent live cells in Hs578T cells treated with 70 μg/ml BN107, 0.5 mg/ml BZL101 or 500nM taxol alone (□) or with 500 μM cholesterol (▪) for 18 hr. Data are represented as mean ± SEM.

Oleanane saponins have been shown to form complexes with cholesterol and to extract cholesterol from the outer face of erythrocyte membranes.42 We examined the levels of total cellular cholesterol and measured a significant decline 4 hr after treatment with BN107 in MDA-MB-231 (Fig. 4b) and in Hs578T cells (Supporting Information Figure 3). The total levels of cholesterol did not change significantly in MCF7 cells (Fig. 4b). The decline in total cellular cholesterol might be responsible for the proapoptotic effect of BN107. Therefore, we cotreated ER− Hs578T cells with cholesterol and BN107. Figure 4c shows that addition of cholesterol completely and specifically rescued cells from BN107-induced death, while it had no effect on cell death induced by BZL10143 or taxol. BZL101 induces necrotic death that involves inhibition of glycolysis. Cotreatment with low-density lipoproteins (LDL) and BN107 or OA also completely abolished the proapoptotic effect of BN107 and OA (Supporting Information Figure 4). These observations confirmed our hypothesis that cholesterol depletion was the underlying mechanism responsible for the proapoptotic effect of BN107.

Lipid rafts are disrupted by BN107

Cholesterol is critically important for the functions of LRs, so we investigated the effect of BN107 treatment on cholesterol in LRs. Figure 4b and Supporting Information Figure 3 show that the cholesterol in the lipid raft region was depleted in BN107-treated MDA-MB-231 and Hs578T cells but not in MCF7 cells, consistent with the reduction in total level of cellular cholesterol. Next, we examined the distribution pattern of caveolin and CD44, 2 LR resident proteins,44, 45 using immunofluorescent staining. Figure 5a shows that BN107 treatment caused a rapid redistribution of these 2 LR resident proteins to intracellular, lysosomal-like structures within 4 hr. This observation was corroborated by data obtained from a biochemical subcellular fractionation approach. Specifically, after a 4-hr treatment of MDA-MB-231 cells with BN107, the level of cytosolic caveolin protein increased and the level of plasma membrane caveolin protein reciprocally decreased, while total level of caveolin protein remained unchanged (Supporting Information Figure 5).

Figure 5. LRs-resident proteins and LR-mediated mTORC1 and mTORC2 signalings are disrupted by BN107 or oleanolic acid treatment. (a) Immunofluorescence staining of caveolin 1 (CAV-1) and CD44 (green) in MDA-MB-231 cells. (b) Analysis of relative density of GM-1 in LR of Hs578T (Hs), MDA-MB-231 (MD) and MCF-7 (MC) cells. The chart was calculated based on GM-1 densities obtained from 3 independent experiments. The image to the right is representative of a typical dot blot analysis using LR fractions isolated from cells treated with BN107 (70 μg/ml) for 4 hr. Western analysis of LRs, non-LR plasma membrane and cytosolic fractions (c), and total cytosolic lysate (d). All cells were treated with 70 μg/ml BN107 (a–d) or 110 μM oleanolic acid (c) (± 500 μM cholesterol, CHL) for 4 or indicated hours (d). LR fractions were spotted directly from fractions (b), or were precipitated to load the same amount of protein (c).

Inhibition of mTORC1 and mTORC2 activities

To determine whether LR-mediated survival signaling was disrupted by BN107, we analyzed the lipid raft fractions obtained from ultracentrifugation of Triton-X100 solubilized cell lysate. The LR region was identified as fractions enriched in gangliosides GM-1, an LR marker.46 As shown in Figure 5b, the levels of GM-1 measured by dot blot analysis were significantly decreased in the LR fractions of BN107-treated Hs578T and MDA-MB-231 cells, thereby supporting the hypothesis that BN107 causes the disruption of lipid rafts. Cholesterol replenishment led to the restoration of raft structures as evidenced by the reappearance of GM1 in the LR fractions. In BN107-resistant MCF-7 cells, levels of GM-1 were unchanged.

To determine whether Akt/mTOR signaling that occurs spatially on LRs was disrupted by BN107 and OA, LR sample fractions were analyzed by Western blotting. We first examined the levels of mTOR/FRAP1, RAPTOR, RICTOR, Akt, 4E-BP, p70S6k in these collected fractions and confirmed that they were all enriched in the LR fractions (see Ref. 16 and data not shown). We then measured the levels of phospho-mTOR, total mTOR, as well as the mTORC1 and mTORC2 complex partners RAPTOR and RICTOR, respectively, in BN107- or OA-treated Hs578T and MDA-MB-231 cells. All were significantly decreased in the LR fractions of both sensitive cell lines after 4 hr of treatment (Fig. 5c), indicating that the components of the mTORC1 and mTORC2 complexes were disrupted/displaced from this region. mTORC1 activity was also inhibited as phosphorylation of its substrates, 4E-BP and p70S6k was found to be significantly inhibited (Fig. 5c).

As mTORC1 and mTORC2 complexes may share the same pool of mTOR/FRAP1 polypeptide,16 and levels of RICTOR were also decreased in the LRs of BN107- or OA-treated Hs578T and MDA-MB231 cells (Fig. 5c), we investigated whether mTORC2 activity is inhibited. Given the recent discovery that mTORC2 is one of the main kinases phosphorylating Akt at Ser473, and that Akt signaling was shown to take place on LRs,13 we analyzed the level of Ser473 phosphorylated Akt in LR fractions as a read-out for mTORC2 activity. We showed that BN107 or OA treatment strongly decreased the level of Ser473-phosphorylated Akt particularly in Hs578T cells that harbor high levels of endogenous phospho-Akt (Ser473), while it had minor effect in cells containing low (MDA-MB-231) or undetectable (MCF7) endogenous levels of phospho-Akt on LRs (Fig. 5c). The addition of exogenous cholesterol restored the signaling events disrupted by BN107 or OA (Fig. 5c). None of these decreases were observed in the resistant MCF-7 cells (right panel, Fig. 5c). The levels of transferrin receptor marking the non-LR plasma membrane region (Non LR PM) and GAPDH marking the cytosolic fractions were not affected in these cell lines treated with BN107 or OA.

As mTORC1 and mTORC2 complexes are also found in the cytosol, we examined the changes of relevant proteins in total cytosolic extracts after BN107 treatment. Consistent with the data for the lipid raft fractions, levels of cytosolic mTOR/FRAP1, phospho-mTOR, RAPTOR and RICTOR were all decreased within 1 hr of BN107 treatment, resulting in minimal mTORC1 and mTORC2 activities to phosphorylate 4E-BP, p70S6k and Ser473-Akt, respectively (Fig. 5d). The decrease in total protein levels of mTOR/FRAP1, RAPTOR and RICTOR occurred posttranscriptionally since the levels of their corresponding mRNAs were not modulated by BN107 or OA treatment in ER− Hs578T cells (Supporting Information Figure 6). Also consistent with data for LRs, cytosolic Ser-473 phosphorylated Akt was markedly decreased; while levels of total cytosolic Akt were slightly decreased at later time points. These data collectively indicated that LRs and LR-mediated mTORC1 and mTORC2 signaling were specifically disrupted by BN107 and OA in ER− breast cancer cells. These effects were presumably due to their cholesterol depleting effect on LRs, as addition of exogenous cholesterol reversed these changes (Figs. 5c, 6a and 6b). The cytosolic levels of phospho-mTOR, total mTOR/FRAP1, RICTOR and RAPTOR were increased in a time-dependent manner in the resistant MCF-7 cells treated with BN107 (Fig. 5d).

Figure 6. Cholesterol specifically restores mTOR signaling disrupted by BN107 (a) or OA (b) induced apoptosis. Western analysis of the total levels of phospho-4E-BP or phospho-p70S60 kinase and phosphor-Akt(ser 473) in BN107- or OA-treated Hs578T, MDA-MB-231 or MCF7, as a read out for mTORC1 and mTORC2 activities, respectively. All cells were treated with 70 μg/ml BN107 (±500 μM CHL) or 125 μM OA (±500 μM CHL) for 4 hr.

Discussion

Gene-expression profiling of primary tumors has identified different subtypes of breast cancers. The basal/mesenchymal-like subtype does neither express the ER or PR nor do they have amplified HER2. These tumors are less differentiated, more aggressive and do not respond to available targeted therapies. Instead they are treated with conventional chemotherapies that have limited efficacy and many side effects. Therefore, it is critically important to identify alternative therapeutic strategies for these patients.

In our study, we showed that BN107 and oleanolic acid specifically targeted mesenchymal-like, ER− breast cancer cells, while ER+ cells were not as sensitive to these treatments. Consistent with this observation, we clustered the publically available expression profiles of various breast lines47 according to their sensitivity to BN107 and found that expression of many ER down-stream targets was associated with insensitivity to BN107 (not shown). At the same time, the proapoptotic effect induced by BN107 showed no correlation with Her2 status (Table 1). We further demonstrated that when ERα status was restored in breast cancer cells lacking functional ER by forced expression with adenovirus or induced expression with TsA, the sensitivity to BN107 in these cells was significantly decreased (Fig. 2). Moreover, we have eliminated ERα protein expression in MCF7 cells using a selective estrogen receptor downregulator, ICI 182780,48 or shRNA-mediated knockdown approaches. Both experiments show that elimination of ERα protein followed by removal of estradiol from the medium leads to enhanced sensitivity to BN107 (Fig. 3). Collectively, these findings demonstrated that functional ER status played a protective role in BN107-induced apoptosis.

To elucidate the mechanism mediating the selective proapoptotic effect of BN107 on ER− breast cancer cells, we performed expression profiling analysis comparing the expression patterns in the sensitive (ER−) versus insensitive (ER+) cell lines. Using IPA analysis, we identified that genes involved in cholesterol synthesis/uptake pathway were over-represented. Similar results were obtained when ER− cells were treated with OA suggesting that cholesterol or cholesterol synthetic pathway intermediates might be involved in apoptosis induced by BN107. We demonstrated that depletion of LR cholesterol, but not isoprenoid precursors, appeared to be responsible for BN107-induced apoptosis, since pretreatment with exogenous cholesterol or cholesterol equivalents (i.e., LDL) protected cells from treatment with BN107 and OA. The integrity of lipid rafts is very dependent upon the presence of cholesterol.9 The loss of cholesterol from lipid rafts after treatment with cholesterol-sequestering agents [methyl-β-cyclodextrin (MβCD)], through increased sterol oxidation or by inhibiting its de novo synthesis leads to loss of LR-associated proteins and decreased cell survival.13, 14, 49

We further showed that BN107- or OA-induced apoptosis was based on decreased mTORC1 and mTORC2 activities in the LRs of ER− breast cancer cells. This presumably led to subsequent inhibition of Akt activity. We cannot rule out that a decrease in the level of phosphatidylinositol phosphates in LRs from cholesterol depletion might lead to reduction of Akt membrane recruitment and phosphorylation. However, the total amount of Akt proteins remained relatively unchanged in the LRs, albeit the level of total cytosolic Akt protein decreased slightly at later time points (Fig. 5d). Therefore, Akt inactivation was likely a consequence of mTORC2 inhibition, due to decrease in mTOR/FRAP1 and RICTOR components. Aberrant activation of the Akt/mTOR pathway has been shown to exist in many cancers, including ER− breast cancer,17 allowing malignant tumor cells to proliferate and evade death signaling or become resistant to various therapies. Despite numerous efforts directed at developing therapeutics that target mTOR, most clinical testing has not shown promising results against solid tumor cancers.18 The clinical data point to our incomplete understanding of the regulation of mTOR pathways, especially regarding mTORC2 activity.35 Recent data have implicated mTORC2 as the major kinase that phosphorylates Ser473 on Akt, and, along with PDK1, facilitates the full activation of Akt.28 Rapamycin showed a minimal acute inhibitory effect on mTORC2, as compared to mTORC1, and only prolonged incubation with high dose can lead to some inhibition on mTORC2.34 Rapamycin-induced mTORC1 inhibition alone can lead to PI3K/Akt pathway reactivation. Conversely, disruption of mTORC2 activity alone might also lead to increased mTORC1 activity.16 Altogether, these observations would suggest that pathway activation and reactivation could be avoided by treatments that lead to concomitant mTORC1 and mTORC2 inhibition resulting in subsequent inactivation of Akt.35 Indeed, recent studies reported the development of ATP-competitive inhibitors that specifically bind to the ATP-binding site of mTOR/FRAP1 kinase and inhibit the catalytic activity of both mTORC1 and mTORC2.50, 51 These ATP-competitive mTOR inhibitors showed significantly stronger anti-proliferative activity on cells than rapamycin.50, 51

The inhibition of mTORC1 and mTORC2 activities by BN107 or OA appeared to be based on disruption of the mTORC1 and mTORC2 complex formation on LRs in the ER− breast cancer cells. This is likely a consequence of the lower levels of complex components mTOR/FRAP1, RAPTOR and RICTOR components present in LRs of the BN107- or OA-treated cells. The disruption of the complexes likely leads to degradation of these proteins, as the roles of RAPTOR and RICTOR are mostly structural and the total cellular levels of them also showed a concomitant decrease. The decreased levels of mTOR/FRAP1, RAPTOR and RICTOR were not caused by downregulation in their steady-state mRNA levels. It is possible that BN107 or OA specifically downregulates the levels of these proteins posttranscriptionally which results in decreased mTORC1 and mTORC2 activities on LRs. Furthermore, since we also observed that the activities of 2 important protein translation regulators, p70S6k and 4E-BP, were decreased, it is possible that the translation of the mTOR components was inhibited in sensitive breast cancer cells. Interestingly, we observed a time-dependent increase in the cytosolic levels of phospho-mTOR, mTOR/FRAP1, RICTOR and RAPTOR in the ER+ MCF7 cells treated with BN107, which produced increased phosphorylation of 4E-BP and p70 S6 kinase (Fig. 5d). It is tempting to postulate that the MCF-7 cells' relative resistance to BN107 is in part a result of the increased activity of mTORC1 and mTORC2 complexes caused by increases in these protein levels. It is also possible that BN107 or OA treatment disrupts the homeostasis of phosphatidic acid on the plasma membrane, which is required for the stabilization of mTORC1 and mTORC2 complexes.52–54 To our knowledge, this is the first report demonstrating downregulation of the activities for both mTOR complexes concomitantly by treating cells with agents that decrease the levels of mTORCs components on LRs, as well as the total cytosolic level.

Experimental and epidemiological evidence suggested that cholesterol may play a promotional role in cancer development and progression.55, 56 Others have proposed that progressive increases in membrane cholesterol contribute to the expansion of rafts, which may potentiate oncogenic pathways (i.e., Akt).11, 57, 58 These findings collectively suggest that agents interfering with cholesterol homeostasis in LRs, such as BN107 and OA, represent a novel approach to disrupt tumor cell survival signaling.

A major concern connected to the potential clinical application of raft-ablating chemicals is that these agents may also nonselectively alter LRs and interfere with the function of vital organs. MβCD derivatives, however, are widely utilized as carriers for water-insoluble drugs for parenteral use,59 implying that lower doses of these compounds do not ultimately exert marked systemic toxicity. Even though the Gleditsia saponins have been shown to strip plasma membrane cholesterol from erythrocytes in vitro, antitumor doses of OA have exhibited minimal toxicity in animals.6 It must be noted that distinct types of LRs have been identified that differ in their biochemical composition, compartmentalization and functions.60 Many studies have shown that depletion of cholesterol from cells leads to the disruption of LRs and the release of raft constituents into the bulk plasma membrane;60 however, not all LRs appear to be equally sensitive to cholesterol depletion. Rajendran et al. showed that in Jurkat and U937 cells, several raft proteins, including lck, lyn and LAT, were released from rafts by treatment with MβCD, but flotillins remained in low-density, detergent-resistant domains.61 These findings suggest that there is heterogeneity in the LR population based on its dependence on or interaction with cholesterol. Consistent with this notion, Ostapkowicz et al. showed that lipid rafts undergo significant structural reorganization during transition from ER+ breast cancer cells to the more aggressive ER− genotype.37 Potentially only a specific subset or composition of LRs supports mTOR signaling that was inhibited by BN107 or OA in the ER− breast cancer cells. How ERα contributes to the protection of BN107-induced lipid raft disruption and apoptosis remains unknown and requires further investigation as well as detailed characterization of the specific interactions between BN107/OA and various LRs components that will facilitate the development of drugs selectively targeting raft components associated with Akt/mTOR signaling.

In summary, our findings demonstrate for the first time that BN107 and OA are strong inhibitors of both mTORC1 and mTORC2 activities. Unlike allosteric mTORC1 inhibitors such as rapalogs, BN107 and oleanolic acid inhibit mTOR activities through decreasing the protein components present in the complexes. This approach may negate development of chemoresistance and growth factor driven activation of Akt and potentially provide better clinical outcomes for ER− breast cancers.