Aromatase inhibitors (AI) have become the first-line endocrine treatment of choice for postmenopausal estrogen receptor–positive (ER+) breast cancer patients, but resistance remains a major challenge. Metabolic reprogramming is a hallmark of cancer and may contribute to drug resistance. Here, we investigated the link between altered breast cancer metabolism and AI resistance using AI-resistant and sensitive breast cancer cells, patient tumor samples, and AI-sensitive human xenografts. We found that long-term estrogen deprivation (LTED), a model of AI resistance, was associated with increased glycolysis dependency. Targeting the glycolysis-priming enzyme hexokinase-2 (HK2) in combination with the AI, letrozole, synergistically reduced cell viability in AI-sensitive models. Conversely, MCF7-LTED cells, which displayed a high degree of metabolic plasticity, switched to oxidative phosphorylation when glycolysis was impaired. This effect was ER dependent as breast cancer cells with undetectable levels of ER failed to exhibit metabolic plasticity. MCF7-LTED cells were also more motile than their parental counterparts and assumed amoeboid-like invasive abilities upon glycolysis inhibition with 2-deoxyglucose (2-DG). Mechanistic investigations further revealed an important role for miR-155 in metabolic reprogramming. Suppression of miR-155 resulted in sensitization of MCF7-LTED cells to metformin treatment and impairment of 2-DG–induced motility. Notably, high baseline miR-155 expression correlated with poor response to AI therapy in a cohort of ER+ breast cancers treated with neoadjuvant anastrozole. These findings suggest that miR-155 represents a biomarker potentially capable of identifying the subset of breast cancers most likely to adapt to and relapse on AI therapy. Cancer Res; 76(6); 1615–26. ©2016 AACR.

A majority of breast tumors at primary diagnosis express estrogen receptor-α (ERα, called hereafter ER) and rely on ER signaling for growth and survival. Several endocrine therapies have been developed clinically to target the ER pathway, including aromatase inhibitors (AI), which block the conversion of androgens to estrogens; selective ER modulators such as tamoxifen, which compete with estrogen for ER; and fulvestrant (ICI182,780), which induces ER degradation (1). In postmenopausal patients, AIs have become the first-line treatment choice, showing higher efficacy than tamoxifen therapy (2). However, resistance remains a problem. Several molecular mechanisms have been proposed to contribute to AI resistance including hypersensitization to estrogen (3, 4) and activation via aberrant growth factor signaling (5). These findings have been exploited clinically by combination therapy approaches in which AIs have been effectively combined either with HER2-targeting compounds (6, 7) or with inhibitors targeting downstream signaling effectors such as mTORC1 (8). However, given the adaptability of tumor cells, targeting a single growth factor or a downstream signaling hub will likely lead to compensatory upregulation. Indeed, many patients fail to benefit from these combined therapeutic approaches and there remains an urgent need for more efficient therapeutic strategies. It has been shown that targeting bioenergetic alterations sensitizes breast cancer cells to chemotherapy (9–11) and to biologic therapy, such as Herceptin (12). However, it is unclear whether cancer cell metabolic reprogramming may contribute to AI resistance and whether targeting it may sensitize cancer cells to AIs.

Most cancer cells, even in the presence of oxygen, show increased glycolysis followed by fermentation of pyruvate to lactate (Warburg effect) using only a small fraction of glucose for oxidative phosphorylation (OXPHOS). Clinically, highly glycolytic tumors can be detected using labeled glucose for PET imaging. These tumors show worse outcome (13), underscoring the therapeutic potential of inhibiting glycolysis. Several glycolytic inhibitors have been shown to hinder tumor growth, particularly in preclinical models, and are currently under clinical investigation. However, no definitive answers are available for glycolysis inhibitors used as single agents, possibly due to tumor adaptation to monotherapy (www.clinicaltrial.gov, ID: NCT00633087; ref. 14).

In this study, we have characterized the metabolic reprogramming that ER+ breast cancers undergo in response and adaptation to long-term estrogen deprivation (LTED). Crucially, we have identified the miR-155/hexokinase-2 (HK2) axis as an important regulator of tumor plasticity and demonstrate that it may have predictive utility in response and adaptation to AI for breast cancer patients.

Cell lines and reagents

Wild-type (wt) MCF-7 and ZR75-1 were obtained from ATCC and independently validated by STR profiling by DNA Diagnostic Centre (DCC). Cells were amplified, stocked, and once thawed were kept in culture for a maximum of 4 months. Cells were maintained in phenol red–free RPMI1640 supplemented with 10% FBS (Euroclone), 2 mmol/L glutamine, and 1 nmol/L 17-β estradiol (E2, Sigma). Long-term E2-deprived (LTED) MCF-7 and ZR75-1 were cultured in steroid-depleted medium in phenol red–free RPMI1640 medium containing 10% dextran charcoal-stripped (DCC) FBS (Hyclone) and 2 mmol/L glutamine (DCC medium). MCF7 cells expressing full-length human aromatase, known as MCF7-2A and MCF7-AROM1, were generated and maintained as described previously (15, 16). For functional assay, cells were cultured for three days in DCC medium.

2-Deoxyglucose and metformin were from Sigma and 3-bromopyruvic acid from Santa Cruz Biotechnology. ICI182,780 was from Tocris Bioscience and the matrix metalloproteinase (MMP) inhibitor, ilomastat, was from Chemicon International. All of the antibodies were from Santa Cruz Biotechnology, except for HKII (Cell Signaling Technology), anti-RAC1 (BD transduction laboratories) and GLUT1 (Abcam).

Cell viability

Cell survival fraction was conducted as reported previously (17).

Western blot analysis

Western blot analyses were conducted as described previously (15). Cells were lysed in RIPA buffer and 20–50 μg of total proteins were loaded on precast SDS-PAGE gels (Bio-Rad).

In vitro Boyden motility and invasion assay

Motility and invasion assays were conducted as described previously (18).

RhoA or Rac1 activity assay

RhoA or Rac1 activity assay was conducted as described previously (19).

Quantitative real-time RT-PCR and miRNA analysis

Total RNA was extracted using RNeasy Kit (Qiagen) and cDNA synthesis performed using Reverse Transcription Kit (Applied Biosystems). For miRNA analysis, miRNeasy Kit (Qiagen) and miScript II RT Kit (Qiagen) were used. qRT-PCR was conducted as described previously (15, 18, 19) using both Power SYBR green dye (Applied Biosystems) or TaqMan assays for HK2 (Hs0060686_m1). Data were normalized on β-2 microglobulin for SYBR or GAPDH (Hs02758991_g1) and ACTB (4310881E), for TaqMan. For miRNA analysis on cell lines, miScript Primer Assay Hs_miR_143_1 and Hs_miR_155_2 were used and normalized on Hs_SNORD61_1 (Qiagen). For miR-155 analysis on patient-derived RNA, we used a ΔCt technique normalizing on both SNORD61 and RNU6-2.

RNAi transfection

Cells were transfected with 20 nmol/L anti-miR-155 (Ambion) or negative controls (Ambion, AM17011) using Lipofectamine 2000 Reagent (Invitrogen) and analyses were performed 3 days after transfection.

Radioactive assays

Breast cancer cells were treated as described in the Figures for 3 days and subjected to radioactive assays as described previously (18).

In vivo experiments

Female Ncr Foxhead nude 6- to 8- week-old mice (Harlan) were kept under sterile conditions with free access to food and water. Mice were ovariectomized and then allowed to acclimatize for approximately 14 days. MCF7-AROM1 xenografts were initiated by subcutaneous inoculation of 100 μL cell suspension containing 1 × 107 cells in Matrigel (BD Biosciences) into the right flank. Growth was maintained by androstenedione support through intradermal daily injection (100 μg/day). Tumors were grown to approximately 8-mm diameter and assigned to treatment groups. Mice continued to receive androstenedione support and were randomized to receive daily doses of vehicle (10% N-methyl-pyrollidone (NMP)/90% polyethylene glycol (PEG300) or letrozole (1 mg/kg in 150 μL of 10% NMP/90% PEG300). Letrozole was administered daily and mice were sacrificed after 21 days. All animal work was carried out with UK Home Office approval.

Immunostaining

For GLUT1 IHC, antigen retrieval was carried out by microwave for 5 minutes at full power (900 W) in citrate buffer pH 6.0. Anti-GLUT1 antibody was applied at 1/3,000 dilution. The stained slides were then scanned on a whole slide scanner (Nanozoomer 2.0-HT, Hamamatsu).

Statistical analysis

Statistical analysis was conducted using GraphPad Prism Software as reported in the figure legends and Results.

The combination of letrozole and glycolysis inhibitors synergistically inhibits cancer cell growth in vitro

As described previously, MCF7-2A cells, which stably express aromatase, showed a dose-dependent decrease in proliferation in response to letrozole in the presence of a standard concentration of androstenedione (Fig. 1A; refs. 15, 20). Of note, 3-day letrozole treatment caused a dose-dependent decrease in glycolysis capacity, indicated by reduced radioactive [3H] glucose uptake (Fig. 1B). However, at this time point, no significant cell growth inhibition was observed. These results suggest that the inhibition of glycolysis by letrozole precedes cell growth inhibition, indicating that glycolysis impairment is not merely a bystander effect of the cell growth inhibition. In fact, the inhibition of glycolysis (glucose uptake) positively correlates with the inhibition of cell viability induced by letrozole (r = 0.92, P < 0.001; Fig. 1C), suggesting that combining letrozole with a glycolysis inhibitor may be more efficient than using AI alone. Indeed, the glycolysis inhibitor 2-deoxyglucose (2-DG) significantly enhanced the effect of letrozole in inhibiting MCF7-2A cell survival, but had no effect on parental MCF7 cells, which have no endogenous aromatase expression (Fig. 1D). Of note, combination index (CI) analysis for 2-DG showed that the effect was synergistic (Supplementary Table S1). Comparable results were obtained when letrozole was combined with the glycolytic inhibitor 3-bromopyruvate (Supplementary Fig. S1).

Figure 1.

Letrozole impairs glycolysis in MCF7-2A cells and targeting glycolysis in combination with letrozole synergistically inhibits cancer cell growth. A, MCF7-2A were E2 deprived for 3 days with addition of 10 nmol/L androstenedione for the last 24 hours. Cells were then subjected to letrozole treatment with the indicated concentration. B, cells were treated as in A but glucose uptake was measured after 3 days of letrozole treatment. Data represent mean ± SEM, n = 3. One-way ANOVA; Dunnett corrected; **, P < 0.01; ***, P < 0.001. C, correlation scatter plot of inhibition of glycolysis (glucose uptake) and inhibition of cell viability (r = 0. 92, P < 0.001). D, cells were treated as in A and after 24 hours received either letrozole, 2-DG (1 mg/mL), or the combination of both for further 72 hours. Parental MCF7 cells that have no endogenous aromatase expression responded to 1 nmol/L E2 but not to androstenedione (andro, 10 nmol/L), letrozole (10 nmol/L), or a combination of the two. Data are presented as the ratio of cell survival inhibition measured compared with untreated cells. Data represent mean ± SEM, n = 3. One-way ANOVA; Dunnett corrected; ***, P < 0.001; ns, not significant.

Figure 1.

Letrozole impairs glycolysis in MCF7-2A cells and targeting glycolysis in combination with letrozole synergistically inhibits cancer cell growth. A, MCF7-2A were E2 deprived for 3 days with addition of 10 nmol/L androstenedione for the last 24 hours. Cells were then subjected to letrozole treatment with the indicated concentration. B, cells were treated as in A but glucose uptake was measured after 3 days of letrozole treatment. Data represent mean ± SEM, n = 3. One-way ANOVA; Dunnett corrected; **, P < 0.01; ***, P < 0.001. C, correlation scatter plot of inhibition of glycolysis (glucose uptake) and inhibition of cell viability (r = 0. 92, P < 0.001). D, cells were treated as in A and after 24 hours received either letrozole, 2-DG (1 mg/mL), or the combination of both for further 72 hours. Parental MCF7 cells that have no endogenous aromatase expression responded to 1 nmol/L E2 but not to androstenedione (andro, 10 nmol/L), letrozole (10 nmol/L), or a combination of the two. Data are presented as the ratio of cell survival inhibition measured compared with untreated cells. Data represent mean ± SEM, n = 3. One-way ANOVA; Dunnett corrected; ***, P < 0.001; ns, not significant.

Close modal

Aerobic glycolysis is enhanced in an in vitro model of AI resistance

MCF7-LTED provides a widely accepted model of breast cancer cells that have developed resistance to AI treatment (3). To investigate the central carbon metabolism in the context of AI resistance, we assessed key molecular components of the glycolytic pathway by Western blot analysis and qRT-PCR. Compared with parental MCF7 cells, independent of estrogen (E2) stimulation, MCF7-LTED cells showed higher expression at both mRNA and protein level of HK2, an enzyme that catalyzes the priming step of glycolysis (Fig. 2A and B). Importantly, MCF7-LTED cells showed increased expression of the glucose importer GLUT1 (also known as SLC2A1) (Fig. 2C) and of the lactate exporter monocarboxylate transporter-4 (MCT4, also known as SLC16A4; Fig. 2A). This increase in glycolysis-associated components reflected enhanced glucose uptake (Fig. 2D) and was accompanied by reduced glucose respiration, assessed by [14C] CO2 release (Fig. 2E). Conversely, MCF7-LTED cells did not differ from parental MCF7 cells in lactate consumption, revealed by analyses of lactate upload (Fig. 2F) and MCT1 (also known as SLC16A1) expression (not shown), but reduced the amount of lactate respired (Fig. 2G), a general indication of OXPHOS impairment in culture conditions. No significant differences were observed in basal metabolic activity when cells were cultured in human normoglycemic 5 mmol/L glucose compared with 11.1 mmol/L glucose, which is the routinely used RPMI culture medium (Supplementary Fig. S2).

Figure 2.

MCF7-LTED cells display higher aerobic glycolytic activity. A, MCF7-LTED cells were compared with wt-MCF7 in presence or absence of 1 nmol/L E2. Total protein lysates were subjected to Western blot analysis as indicated. B–G, MCF7-LTED cells were compared with wt-MCF7 and subjected to qRT-PCR (B and C) after 3-day culture or to radioactive assays (D–G) as described in Materials and Methods. Data represent mean ± SEM, n = 3. Student t test; *, P < 0.05, **, P < 0.01; ***, P < 0.001; ns, not significant.

Figure 2.

MCF7-LTED cells display higher aerobic glycolytic activity. A, MCF7-LTED cells were compared with wt-MCF7 in presence or absence of 1 nmol/L E2. Total protein lysates were subjected to Western blot analysis as indicated. B–G, MCF7-LTED cells were compared with wt-MCF7 and subjected to qRT-PCR (B and C) after 3-day culture or to radioactive assays (D–G) as described in Materials and Methods. Data represent mean ± SEM, n = 3. Student t test; *, P < 0.05, **, P < 0.01; ***, P < 0.001; ns, not significant.

Close modal

Metabolic targeting does not affect MCF7-LTED cell growth revealing their metabolic plasticity

To investigate whether targeting MCF7-LTED glycolysis dependency could lead to decreased cell survival, parental MCF7 and their LTED derivatives were exposed to the metabolic poisons: 2-DG, which inhibits glycolysis, and metformin, which targets complex 1 of the electron transport chain, leading to inhibition of OXPHOS. Single drug treatment significantly impaired parental MCF7 cell survival, both in presence and absence of E2 (Fig. 3A). Conversely, no effects were observed, either on cell survival (Fig. 3A) or apoptosis (Supplementary Fig. S3A), when MCF7-LTED cells were treated with metformin, and only minor survival fraction changes were noted when MCF7-LTED cells were treated with 2-DG (Fig. 3A, black bars). As expected, 2-DG plus metformin dramatically impaired cell survival of both parental MCF7 and LTED cells, suggesting that the MCF7-LTED cells were capable of switching from glycolysis to OXPHOS metabolism when glycolysis was impaired. Indeed in MCF7-LTED cells, 2-DG treatment decreases HK2 expression (Fig. 3B) and glucose uptake (Fig. 3C) while inducing their ability to upload lactate (Fig. 3D), which can be diverted to OXPHOS metabolism.

Figure 3.

MCF7-LTED cells display metabolic plasticity and are insensitive to glycolysis targeting. A, MCF7-LTED cells were compared with wt-MCF7 in presence or absence of 1 nmol/L E2 and were subjected to 1 mg/mL 2-DG and 5 mmol/L metformin (Met) treatments. Data are presented as fold change of survival cell fraction compared with untreated cells. B–D, MCF7-LTED cells were treated with or without 2-DG or metformin and subjected to radioactive glucose uptake (B), radioactive lactate uptake (C), or Western blotting (D). One-way ANOVA; Dunnett corrected; *, P < 0.05; ***, P < 0.001.

Figure 3.

MCF7-LTED cells display metabolic plasticity and are insensitive to glycolysis targeting. A, MCF7-LTED cells were compared with wt-MCF7 in presence or absence of 1 nmol/L E2 and were subjected to 1 mg/mL 2-DG and 5 mmol/L metformin (Met) treatments. Data are presented as fold change of survival cell fraction compared with untreated cells. B–D, MCF7-LTED cells were treated with or without 2-DG or metformin and subjected to radioactive glucose uptake (B), radioactive lactate uptake (C), or Western blotting (D). One-way ANOVA; Dunnett corrected; *, P < 0.05; ***, P < 0.001.

Close modal

Compared with MCF7 cells, ZR75-1 cells have similar ER expression. However, in contrast to MCF7-LTED cells, ZR75-1-LTED cells have undetectable levels of ER (15). 2-DG and metformin administration reduced the survival fraction of both parental and ZR75-1-LTED cells (Supplementary Fig. S3B). Crucially, combining metformin with 2-DG had an additive effect. This suggests that ER expression is a prerequisite for the metabolic plasticity observed in MCF7-LTED cells. This prerequisite was further reinforced by administration of the ER downregulator ICI182,780. Importantly, parental MCF7 cells and their derivatives (LTED and MCF7-2A) treated with ICI182,780 showed impaired glucose uptake and enhanced OXPHOS rate (Supplementary Fig. S4).

2-DG treatment enhances amoeboid-dependent cell motility and invasion of MCF7-LTED cells

Metabolic reprogramming and motile plasticity correlates with enhanced aggressiveness of cancer cells (21). To characterize whether metabolic plasticity was paralleled by motile plasticity, MCF7-LTED cells were first subjected to a Boyden assay, in the presence or absence of a Matrigel barrier. MCF7-LTED cells showed increased migration compared with parental MCF7, both in the presence and absence of E2 (Fig. 4A and B) as well as increased invasion (Fig. 4E). E-cadherin and vimentin protein levels were unchanged, suggesting that epithelial-to-mesenchymal transition (EMT) is not directly involved in the enhanced motility (Supplementary Fig. S5A). Alternative to mesenchymal motility, cells can also adopt amoeboid motility allowing them to slide through the extracellular matrix (22). Amoeboid migration is characterized by MMP exclusion, inhibition of Rac1, and activation of RhoA (23). Accordingly, MCF7-LTED cells showed enhanced RhoA-GTP expression levels and decreased Rac1-GTP (Fig. 4C). In addition, treatment with the MMP inhibitor ilomastat did not reduce the invasive capacity of MCF7-LTED cells through Matrigel (Fig. 4D). Finally, gelatin zymography of parental and MCF7-LTED cells revealed no significant differences in MMP activity (Supplementary Fig. S5B) and fluorescein isothiocyanate (FITC)-collagen release assays showed no increase in collagen degradation (Supplementary Fig. S5C).

Figure 4.

MCF7-LTED cells display high motile and invasive abilities and 2-DG administration enhances these features. A and B, MCF7-LTED cells were compared with wt-MCF7 in presence or absence of 1 nmol/L E2 and were subjected to 48-hour migration assay with or without drug treatments, as indicated; 1 mg/mL 2-DG and 5 mmol/L metformin (Met). One-way ANOVA; Bonferroni corrected; ***, P < 0.001. C, Rac1-GTP and RhoA-GTP were assayed as described in Materials and Methods. D, MCF7-LTED cells were subjected to invasion assay in presence or absence of the MMP inhibitor ilomastat (50 μmol/L). Student t test; ns, not significant. E, cells were treated as in B and subjected to invasion assay. Data represent mean values ± SEM, n = 3. One-way ANOVA; Bonferroni corrected; **, P < 0.01; ***, P < 0.001.

Figure 4.

MCF7-LTED cells display high motile and invasive abilities and 2-DG administration enhances these features. A and B, MCF7-LTED cells were compared with wt-MCF7 in presence or absence of 1 nmol/L E2 and were subjected to 48-hour migration assay with or without drug treatments, as indicated; 1 mg/mL 2-DG and 5 mmol/L metformin (Met). One-way ANOVA; Bonferroni corrected; ***, P < 0.001. C, Rac1-GTP and RhoA-GTP were assayed as described in Materials and Methods. D, MCF7-LTED cells were subjected to invasion assay in presence or absence of the MMP inhibitor ilomastat (50 μmol/L). Student t test; ns, not significant. E, cells were treated as in B and subjected to invasion assay. Data represent mean values ± SEM, n = 3. One-way ANOVA; Bonferroni corrected; **, P < 0.01; ***, P < 0.001.

Close modal

As 2-DG and metformin monotherapies had no substantial effect on MCF7-LTED cell survival, we investigated whether these treatments impeded their migratory capacity. Surprisingly, 2-DG treatment enhanced MCF7-LTED cell motility (Fig. 4B) and invasive abilities (Fig. 4E), without affecting MMP activity (Supplementary Fig. S5B) or their collagen-degrading capacity (Supplementary Fig. S5C). These features were confirmed in a three-dimensional tumor spheroid invasion assay (24). MCF7-LTED spheroids embedded in Matrigel treated with 2-DG were more invasive than those generated from untreated MCF7-LTED cells and the invading cells were characterized by single cell dispersal, resembling typical amoeboid migration (Supplementary Fig. S6; ref. 24). Therefore, MCF7-LTED cells are characterized by amoeboid motility, in line with the absence of EMT markers expression and MMP activity.

ER-dependent miR-155 is responsible for metabolic and motile plasticity of MCF7-LTED cells

As targeting the glycolytic metabolism with 2-DG in the AI-resistant cell model led to surprising results, we investigated the molecular mechanism responsible for the metabolic and motile plasticity of MCF7-LTED cells.

Deregulation of miRNA expression has been demonstrated in many types of cancer as they act as upstream regulators of mRNA expression of genes involved in carcinogenesis (25). In particular, miR-155 has been shown to be overexpressed in breast cancers (26–28) and is modulated by estrogens (29). In addition, miR-155 controls miR-143 expression, a miRNA known to target HK2 in MCF7 cells (30). Indeed, miR-143 inhibits HK2 expression via a conserved miR-143 recognition motif located in the 3′-untranslated region (3′UTR) of HK2 mRNA (31). Therefore, we hypothesized that miR-155 and miR-143 may be deregulated in MCF7-LTED cells and responsible for the glycolytic metabolism increase noted in this model. qRT-PCR showed that E2 induced a significant increase in miR-155 expression in parental MCF7 and that MCF7-LTED cells had higher (∼3-fold) miR-155 expression when compared with their parental counterpart (Fig. 5A). This increased miR-155 expression was paralleled by a significant reduction in miR-143 (Fig. 5B), consistent with the increased HK2 expression both at mRNA and protein levels (Fig. 2A and B). E2-induced miR-155 expression was observed in MCF7-2A cells upon androstenedione administration (Supplementary Fig. S7A). Importantly, blocking androgen to estrogen conversion by letrozole administration restored basal miR-155 expression (Supplementary Fig. S7A). In addition, ER dependency was validated in both E2-treated MCF7 and MCF7-LTED cells by administration of ICI182,780 (Supplementary Fig. S7B). We therefore hypothesized that the enhanced expression of miR-155 and HK2 in MCF7-LTED cells was related to their retention of a functional ER. This was further supported by our observation that miR-155 and HK2 expression were not significantly altered in ER ZR75-1-LTED cells (Supplementary Fig. S7C)

Figure 5.

miR-155/miR-143 axis controls metabolic and motile plasticity of MCF7-LTED cells. A and B, MCF7-LTED cells were compared with wt-MCF7 in presence or absence of 1 nmol/L E2 and subjected to qRT-PCR. Data represent mean ± SEM, n = 3. One-way ANOVA; Dunnett corrected; *, P < 0.05; **, P < 0.01. C and D, MCF7-LTED cells were transfected with the indicated oligos and treated with 5 mmol/L metformin for further 72 hours before survival fraction calculation (C) or subjected to migration assays after further 48 hours of 2-DG treatment (D). Data represent mean ± SEM, n = 3. Student t test; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 5.

miR-155/miR-143 axis controls metabolic and motile plasticity of MCF7-LTED cells. A and B, MCF7-LTED cells were compared with wt-MCF7 in presence or absence of 1 nmol/L E2 and subjected to qRT-PCR. Data represent mean ± SEM, n = 3. One-way ANOVA; Dunnett corrected; *, P < 0.05; **, P < 0.01. C and D, MCF7-LTED cells were transfected with the indicated oligos and treated with 5 mmol/L metformin for further 72 hours before survival fraction calculation (C) or subjected to migration assays after further 48 hours of 2-DG treatment (D). Data represent mean ± SEM, n = 3. Student t test; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

We subsequently targeted miR-155 to impair metabolic and motile properties of MCF7-LTED cells. Anti-miR-155 caused a marked reduction of miR-155 and HK2 expression and a corresponding increase of miR-143 expression (Supplementary Fig. S8). Anti-miR-155–transfected MCF7-LTED cells showed a small change in cell survival when compared with anti-miR scramble control. However, exposure of the anti-miR-155–transfected cells to metformin led to a marked reduction in cell survival (Fig. 5C). To demonstrate that miR-155 may also be responsible for the 2-DG–induced migration of MCF7-LTED cells, we compared the anti-miR-155 transfected to the anti-miR scramble control cells. Notably, anti-miR-155 transfection reverted the motile ability of MCF7-LTED cells when exposed to 2-DG (Fig. 5D).

Glycolytic key players and miR-155 levels are decreased by letrozole administration in AI-sensitive human breast cancer xenografts

The MCF7 aromatase-transfected cell line, MCF7-AROM1, was injected subcutaneously into immunocompromised mice and formed tumors under androstenedione support, due to the conversion into estrogen. Tumor-bearing mice were then treated with or without letrozole and sacrificed after 21 days. Letrozole-treated tumors were significantly smaller than those of the vehicle-treated mice (P = 0.009, Supplementary Fig. S9). To determine the effects of letrozole treatment on the glycolytic key players, GLUT1 and HK2, and on miR-155 levels in breast cancer xenografts, MCF7-AROM1 tumors were subjected to IHC and/or protein and RNA were extracted for Western blot analysis and qRT-PCR analyses, respectively. Letrozole significantly reduced the expression levels of GLUT1, as shown by IHC (Fig. 6A) and Western blot analysis (Fig. 6B), and that of HK2, both at protein (Fig. 6B) and at mRNA levels (Fig. 6C), although the latter with borderline statistical significance. Importantly, miR-155 levels were significantly reduced by letrozole administration (Fig. 6D), reinforcing the in vitro findings linking miR-155 expression and the glycolytic phenotype.

Figure 6.

Letrozole administration decreases glycolytic key players and miR-155 levels in AI-sensitive human breast cancer xenografts. A, representative IHC GLUT1 images of vehicle and letrozole-treated human MCF7-AROM1 xenografts (day 21 of treatment) and relative quantification per field of view (FOV). Data shown are from 7 mice per group ± SEM (Student t test, P = 0.004). Scale bar, 1 mm. B, total lysates extracted from MCF7-AROM1 tumors were subjected to Western blot analysis as indicated. Quantification shown below the blots was performed using ImageJ and normalized on GAPDH. Data represent mean ± SEM (n = 3). C and D, RNA was extracted and subjected to qRT-PCR for HK2 (C) and miR-155 (D). Data represent mean ± SEM (n = 4); Student t test: HK2, P = 0.084; miR-155, P = 0.048.

Figure 6.

Letrozole administration decreases glycolytic key players and miR-155 levels in AI-sensitive human breast cancer xenografts. A, representative IHC GLUT1 images of vehicle and letrozole-treated human MCF7-AROM1 xenografts (day 21 of treatment) and relative quantification per field of view (FOV). Data shown are from 7 mice per group ± SEM (Student t test, P = 0.004). Scale bar, 1 mm. B, total lysates extracted from MCF7-AROM1 tumors were subjected to Western blot analysis as indicated. Quantification shown below the blots was performed using ImageJ and normalized on GAPDH. Data represent mean ± SEM (n = 3). C and D, RNA was extracted and subjected to qRT-PCR for HK2 (C) and miR-155 (D). Data represent mean ± SEM (n = 4); Student t test: HK2, P = 0.084; miR-155, P = 0.048.

Close modal

Glycolytic key players correlate with response to AI treatment in vivo

Next, we evaluated whether key glycolytic players, such as GLUT1 and MCT4, which were differentially expressed in MCF7-LTED versus parental control, correlated with response to AI treatment. In silico analysis was performed by retrieving publicly available gene expression data from biopsies of 52 ER+ breast cancer patients taken before and after 2 weeks of neoadjuvant letrozole treatment (32). The patients were subsequently divided into responders and nonresponders defined by a more than 50% and less than 50% reduction, respectively, in tumor volume, following a further 3 months of letrozole treatment. Pairwise comparison shows a significant decrease in GLUT1 expression after 2 weeks of letrozole treatment in the responder cohort (P = 0.008) but not in the nonresponder cohort (Fig. 7A). Conversely, MCT4 expression increased in the nonresponder cohort with a borderline statistical significance (Fig. 7B, P = 0.055), but not in the responder cohort. HK2 expression was also analyzed and although no significant differences were reported by paired t test in both responder and nonresponder patients, it was interesting to note that 57% (21/37) of the responders showed a decreased HK2 expression, whereas 67% (10/15) of the nonresponders showed an increased expression after 2 weeks of treatment with letrozole.

Figure 7.

MCT4, GLUT1, and miR-155 expression levels in the response to AIs in clinical specimens. A–D, correlation of metabolic key players GLUT1 and MCT4 expression with response to AIs. A and B, changes in 52 paired ER+ breast cancer samples pre- and post-2-week letrozole treatment (32). A, responder patients show a significant decrease in GLUT1. No significant change was observed in the nonresponder group. B, conversely, MCT4 expression is increased in the nonresponder group (Wilcoxon test). C–F, correlation of the MCT4 expression with Ki67 staining before receiving anastrozole treatment (C) or of the change of the 2-week change in GLUT1 expression with the change in Ki67 staining between pre- and posttreatment biopsies (Spearman correlation; D). Paired pre- and posttreatment gene expression profiles and Ki67 IHC staining were available for 69 patients (33). For miR-155 analysis, correlation of the pretreatment miR-155 expression with the Ki67 staining after anastrozole treatment (E) or with the change in Ki67 staining between pre- and posttreatment biopsies (Pearson correlation; F). RNA was available for 64 patients.

Figure 7.

MCT4, GLUT1, and miR-155 expression levels in the response to AIs in clinical specimens. A–D, correlation of metabolic key players GLUT1 and MCT4 expression with response to AIs. A and B, changes in 52 paired ER+ breast cancer samples pre- and post-2-week letrozole treatment (32). A, responder patients show a significant decrease in GLUT1. No significant change was observed in the nonresponder group. B, conversely, MCT4 expression is increased in the nonresponder group (Wilcoxon test). C–F, correlation of the MCT4 expression with Ki67 staining before receiving anastrozole treatment (C) or of the change of the 2-week change in GLUT1 expression with the change in Ki67 staining between pre- and posttreatment biopsies (Spearman correlation; D). Paired pre- and posttreatment gene expression profiles and Ki67 IHC staining were available for 69 patients (33). For miR-155 analysis, correlation of the pretreatment miR-155 expression with the Ki67 staining after anastrozole treatment (E) or with the change in Ki67 staining between pre- and posttreatment biopsies (Pearson correlation; F). RNA was available for 64 patients.

Close modal

These findings were independently validated in gene expression data derived from 69 paired ER+ tumor biopsies taken pre- and post- 2-week neoadjuvant treatment with the nonsteroidal AI anastrozole (33). As previously reported, lack of response to AI treatment can be monitored by a decrease in Ki67 staining (15), a parameter that has been shown to predict poor long-term disease outcome (1). Pretreatment (i.e., samples before receiving anastrozole) MCT4 expression showed a significant inverse correlation with pretreatment Ki67 levels (Fig. 7C; r = −0.28, P = 0.02), suggesting that MCT4 expression is not merely a surrogate marker of highly proliferating tumors. Furthermore, the change in GLUT1 expression in the pre- and posttreatment samples positively correlated with the proportional 2-week change in Ki67 (Fig. 7D; r = 0.26, P = 0.03). In addition, we investigated whether the glycolytic components monitored in the AI setting could also be responsible for patient stratification in breast cancers that had been treated with tamoxifen. Notably, Kaplan–Meier analysis revealed that patients characterized by higher levels of HK2 (HR = 1.72, P = 0.0015), MCT4 (HR = 1.46, P = 0.019), or GLUT1 (HR = 1.36, P = 0.062) showed poorer relapse-free survival when compared with lower expressing tumors (Supplementary Fig. S10A–S10C). Crucially, high miR-155–expressing tumors also showed a poorer prognosis (HR = 3.62, P = 0.0048; Supplementary Fig. S10D), suggesting that miR-155 and glycolysis could also have a role in tamoxifen response. These data support the concept that glycolytic metabolism plays an important role in the response and adaptation of breast cancer patients to AI treatment and potentially to other endocrine agents, such as tamoxifen.

miR-155 expression in ER+/HER2 breast cancers identifies a subset of patients that do not respond to AI anastrozole

As our in vitro data suggested miR-155 as a key component of the metabolic and motile reprogramming of AI-resistant cells, we wanted to validate whether this could be of clinical relevance. Therefore, we analyzed by qRT-PCR the expression levels of miR-155 in 64 ER+ and HER2 breast cancer patients before undergoing anastrozole treatment (33). Ki67 levels were also monitored before and after 2 weeks of anastrozole treatment and used as an indicator of therapy response (15). Notably, pretreatment miR-155 levels positively correlate with the Ki67 levels posttreatment (Fig. 7E; r = 0.3593, P = 0.0035) and with the proportional 2-week change in Ki67 (Fig. 7F; r = 0.2973, P = 0.0171). Pretreatment miR-155 levels did not show a significant correlation with pretreatment Ki67 levels (r = 0.22, P = 0.08), thus excluding a possible association between miR-155 and proliferation. These results indicate that high miR-155 levels correlate with poor response to anastrozole therapy, reinforcing the idea that miR-155 plays a role in the response and adaptation of breast cancer patients to AI treatment.

Resistance to AI therapy is an important clinical challenge as AIs have become the standard of care for many patients, particularly in the western world. Current preclinical data suggest that a fundamental role in developing resistance is played by growth factor–induced signaling pathways and clinical trials have shown encouraging results. However, cancer cells are highly plastic and eventually undergo signaling rewiring when trying to block a growth-driving signaling pathway: this could explain why targeting downstream pathway hubs (e.g., mTOR) or impacting directly on tumor cell cycle (e.g., cyclin-dependent kinases; ref. 34) gave better results than targeting single growth factor receptors. Metabolic adaptation is essential to satisfy the different energetic requirements that support a cancer cell during proliferation, exit from the primary site (dissemination), and engraftment to distal tissues. As such, we investigated primarily whether metabolic reprogramming could be responsible for breast cancer cell adaptation to LTED and whether metabolic targeting could be of any benefit in AI-sensitive and AI-resistant in vitro models.

Several recent studies have reported that OXPHOS metabolism is associated with aggressiveness of cancer cells (35) and characterizes cancer cells that have developed resistance to different chemotherapeutic agents in various cancer models (36, 37). Conversely, reprogramming toward a hyperglycolytic metabolism has been associated with resistance to biologic agents such as Herceptin and Avastin in breast cancer (12, 38). To study the metabolic reprogramming that a cell undergoes in response and resistance to AI treatment, we have used an array of different in vitro and in vivo models. As breast cancer cell lines are characterized by low or no expression of endogenous aromatase, we have used cells transfected with the human aromatase gene to study AI response. In addition, we have used ER+ cells adapted to LTED to study AI resistance, as lack of estrogen in the medium mimics the hormone withdrawal that occurs during AI treatment. However, such models have some limitations, that is, MCF7 cells overexpressing aromatase do not account for the fact that androgen to estrogen conversion predominantly occurs in the stromal cells of breast cancers, and LTED models cannot mimic the rewiring that cancer cells undergo when chronically treated with an AI. Therefore, to confirm the clinical relevance of our in vitro findings, we have used tumor biopsies derived from patients enrolled in neoadjuvant trials of AIs and a xenograft model that mimics letrozole clinical treatment. To identify the metabolic pathways associated with AI resistance, we first analyzed the expression of key glycolytic components and performed tracking radioactive assays in ER+ breast cancer cells and then assessed the expression of glycolytic-related genes MCT4 and GLUT1 in publicly available clinical datasets. Doyen and colleagues reported the increased expression of GLUT1 and MCT4 in triple-negative breast cancers. Importantly, their expression levels correlate with worse clinical outcome. In particular, MCT4 was suggested to be a new independent prognostic factor in breast cancer (39). Our data implies that AI-resistant patients are characterized by enhanced MCT4 and GLUT1 expression (i.e., hyperglycolytic phenotype) and therefore may have a potential value in predicting AI response. In addition, we observed reduced expression of the glycolytic-related components GLUT1 and HK2 in a human xenograft model that were responsive to treatment with letrozole. Taken together, our results highlight a “glycolysis dependency” of the MCF7-LTED cells and of the AI-sensitive xenograft. Furthermore, letrozole sensitivity of MCF7-2A cells was potentiated in vitro by concomitant antiglycolytic agent administration. To our surprise, however, targeting glycolytic metabolism in AI-resistant MCF7-LTED cells did not impact on cell survival. Indeed, these cells were capable of switching from glycolysis to OXPHOS metabolism; furthermore, 2-DG treatment enhanced their promigratory and proinvasive capabilities. This different response might explain the discrepancy in 2-DG clinical trials. In fact, it has been reported that luminal B breast cancers show a higher FDG-PET signal (i.e., higher glucose uptake) when compared with luminal A (40). Notably, luminal B tumors have poorer prognosis compared with luminal A with an increased risk of early relapse and resistance to endocrine therapy and chemotherapy (41). This suggests that the acquisition of a hyperglycolytic phenotype correlates with aggressive clinical features of the luminal B subtype. Consequently, 2-DG used in combination with standard therapy regimes could be beneficial if patient subsets are characterized and properly assigned. For example, in the preclinical models described here, combining glycolysis targeting with 2-DG might be beneficial in the AI-sensitive tumors but detrimental to AI-resistant ones. In particular, our data support the concept that the aggressive features of tumor cells are not necessarily linked to a specific steady-state metabolic phenotype. Conversely, such tumors are characterized by a “dynamic” phenotype that confers to cancer cells many escape routes to bypass signaling, metabolic and/or motile blockades; this plasticity allows cancer cells adaption to target treatments and is ultimately responsible for resistance and relapse.

We therefore explored the molecular components that could be responsible for the metabolic and motile reprogramming that MCF7 cells undergo under LTED conditions and after 2-DG treatment. Deregulation of miRNA expression has been demonstrated in many types of cancer and the involvement of miRNAs has been previously described in breast cancer etiology (42) and in endocrine therapy response and resistance (43). For example, miR-221/222 was able to confer tamoxifen resistance (44), while reexpression of miR-375 (45), let-7 (46), miR-342 (47), or miR-320 (48) induced tamoxifen sensitivity. As the metabolic plasticity described in the current article was exclusive to the LTED cells that retain ER expression, that is, MCF7, but not in those who lose it during LTED adaptation, for example, ZR75-1, we focused our attention on those miRNAs known to be regulated by ER signaling.

miR-155 was found to be associated with metastasis events and invasive properties of breast cancer (49) and E2 has been shown to upregulate miR-155 expression in MCF7 cells (29). Furthermore, miR-155 ectopic overexpression in MCF7 cells induced the upregulation of the key glycolytic enzyme HK2 (30). Indeed, Jiang and colleagues showed that miR-155 promotes HK2 transcription by activation of STAT3 and via targeting C/EBPβ (a transcriptional activator for miR-143); miR-155 represses miR-143, a negative regulator of HK2, thus resulting in upregulation of HK2 expression at the posttranscriptional level (30). In addition, miR-155 transfection increases the motility of lymphoma cells by impacting on RhoA activity and RhoA downstream effectors regulating stress fiber formation and actin polymerization (50).

As miR-155 and miR-143 expression levels were found to be deregulated in MCF7-LTED cells, we hypothesized that miR-155 could be responsible for the altered metabolism observed in these cells and for the motile and invasive abilities displayed. Notably, impairing ER signaling by adding letrozole to androstenedione-treated MCF7-2A cells or by ICI182,780 administration in MCF7-LTED cells, confirms ER dependency of miR-155 expression. Moreover, it suggests why ER ZR751-LTED cells do not have such characteristics. This was further confirmed in the xenograft-derived samples, where miR-155 expression was significantly decreased following ER signaling impairment, induced by in vivo letrozole administration.

Crucially, we identify miR-155 levels to be significantly associated with response to AI therapy in ER+ breast cancers and identified a subset of patients that could benefit from a combinatorial approach (i.e., high miR-155) rather than an AI monotherapy. In silico analysis of publicly available datasets also showed that high miR-155 levels associated with poor prognosis on tamoxifen. Several studies have reported that miR-155 levels could be monitored in patients' blood samples, and that this miRNA is differentially expressed in breast cancer patients when compared with healthy women (26, 51, 52). Therefore, our results highlight a potential diagnostic approach and therapeutic intervention based on miR-155 patient stratification.

Finally, targeting miR-155 with anti-miR agents impairs 2-DG–induced motility of MCF7-LTED cells and potentiated the effect of metformin treatment, indicating that miR-155 targeting could have a potential therapeutic implication in AI-resistant tumors that retain ER expression.

Here, we show that metabolic reprogramming has a fundamental role in response and adaptation to LTED and therefore is important for AI therapy response. Particularly, our in vitro, in vivo, and in silico analyses suggest that concomitant targeting of glycolysis and the ER signaling pathway may be beneficial in the AI-sensitive setting but has to be carefully considered in the AI-resistant setting, where 2-DG treatment may enhance aggressive features of cancer cells. However, miR-155 targeting may be achieved in the latter scenario and could ameliorate the current therapeutic strategies, particularly for the AI-resistant setting.

No potential conflicts of interest were disclosed.

Conception and design: A. Morandi

Development of methodology: R. Ribas, M.L. Taddei, A. Morandi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Bacci, A. Fearns, R. Ribas, M. Dowsett, C.M. Isacke, L.-A. Martin, A. Morandi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Bacci, E. Giannoni, R. Ribas, Q. Gao, M. Dowsett, L.-A. Martin, A. Morandi

Writing, review, and/or revision of the manuscript: M. Bacci, E. Giannoni, G. Pintus, M. Dowsett, L.-A. Martin, P. Chiarugi, A. Morandi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Morandi

Study supervision: P. Chiarugi, A. Morandi

The authors thank David Robertson and the ICR Histopathology for immunohistochemistry optimization, Allan Thornhill for in vivo experiments assistance, and Needs Hope and Gianluca Mattei for useful comments.

The work was funded by Fondazione Italiana Ricerca sul Cancro and Fondazione Umberto Veronesi (A. Morandi), Associazione Italiana Ricerca sul Cancro (AIRC#8797), Istituto Toscano Tumori (#0203607), Programma operativo regionale Obiettivo “Competitività regionale e occupazione” della Regione Toscana cofinanziato dal Fondo europeo di sviluppo regionale 2007–2013 grants (P. Chiarugi), and Breast Cancer Now grants (M. Dowsett, L.-A. Martin, and C.M. Isacke). The authors acknowledge NHS funding to the Royal Marsden Hospital's NIHR Biomedical Research Centre.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Forbes
JF
,
Cuzick
J
,
Buzdar
A
,
Howell
A
,
Tobias
JS
,
Baum
M
, et al
Effect of anastrozole and tamoxifen as adjuvant treatment for early-stage breast cancer: 100-month analysis of the ATAC trial
.
Lancet Oncol
2008
;
9
:
45
53
.
2.
Winer
EP
,
Hudis
C
,
Burstein
HJ
,
Wolff
AC
,
Pritchard
KI
,
Ingle
JN
, et al
American Society of Clinical Oncology technology assessment on the use of aromatase inhibitors as adjuvant therapy for postmenopausal women with hormone receptor-positive breast cancer: status report 2004
.
J Clin Oncol
2005
;
23
:
619
29
.
3.
Martin
LA
,
Farmer
I
,
Johnston
SR
,
Ali
S
,
Marshall
C
,
Dowsett
M
. 
Enhanced estrogen receptor (ER) alpha, ERBB2, and MAPK signal transduction pathways operate during the adaptation of MCF-7 cells to long term estrogen deprivation
.
J Biol Chem
2003
;
278
:
30458
68
.
4.
Martin
LA
,
Pancholi
S
,
Chan
CM
,
Farmer
I
,
Kimberley
C
,
Dowsett
M
, et al
The anti-oestrogen ICI 182,780, but not tamoxifen, inhibits the growth of MCF-7 breast cancer cells refractory to long-term oestrogen deprivation through down-regulation of oestrogen receptor and IGF signalling
.
Endocr Relat Cancer
2005
;
12
:
1017
36
.
5.
Aguilar
H
,
Solé
X
,
Bonifaci
N
,
Serra-Musach
J
,
Islam
A
,
López-Bigas
N
, et al
Biological reprogramming in acquired resistance to endocrine therapy of breast cancer
.
Oncogene
2010
;
29
:
6071
83
.
6.
Kaufman
B
,
Mackey
JR
,
Clemens
MR
,
Bapsy
PP
,
Vaid
A
,
Wardley
A
, et al
Trastuzumab plus anastrozole versus anastrozole alone for the treatment of postmenopausal women with human epidermal growth factor receptor 2-positive, hormone receptor-positive metastatic breast cancer: results from the randomized phase III TAnDEM study
.
J Clin Oncol
2009
;
27
:
5529
37
.
7.
Johnston
S
,
Pippen
J
,
Pivot
X
,
Lichinitser
M
,
Sadeghi
S
,
Dieras
V
, et al
Lapatinib combined with letrozole versus letrozole and placebo as first-line therapy for postmenopausal hormone receptor-positive metastatic breast cancer
.
J Clin Oncol
2009
;
27
:
5538
46
.
8.
Baselga
J
,
Campone
M
,
Piccart
M
,
Burris
HA
,
Rugo
HS
,
Sahmoud
T
, et al
Everolimus in postmenopausal hormone-receptor-positive advanced breast cancer
.
N Engl J Med
2012
;
366
:
520
9
.
9.
Zhao
Y
,
Butler
EB
,
Tan
M
. 
Targeting cellular metabolism to improve cancer therapeutics
.
Cell Death Dis
2013
;
4
:
e532
.
10.
Morandi
A
,
Chiarugi
P
. 
Metabolic implication of tumor:stroma crosstalk in breast cancer
.
J Mol Med
2014
;
92
:
117
26
.
11.
Tennant
DA
,
Durán
RV
,
Gottlieb
E
. 
Targeting metabolic transformation for cancer therapy
.
Nat Rev Cancer
2010
;
10
:
267
77
.
12.
Zhao
Y
,
Liu
H
,
Liu
Z
,
Ding
Y
,
Ledoux
SP
,
Wilson
GL
, et al
Overcoming trastuzumab resistance in breast cancer by targeting dysregulated glucose metabolism
.
Cancer Res
2011
;
71
:
4585
97
.
13.
Ben-Haim
S
,
Ell
P
. 
18F-FDG PET and PET/CT in the evaluation of cancer treatment response
.
J Nucl Med
2009
;
50
:
88
99
.
14.
Raez
LE
,
Papadopoulos
K
,
Ricart
AD
,
Chiorean
EG
,
Dipaola
RS
,
Stein
MN
, et al
A phase I dose-escalation trial of 2-deoxy-D-glucose alone or combined with docetaxel in patients with advanced solid tumors
.
Cancer Chemother Pharmacol
2013
;
71
:
523
30
.
15.
Morandi
A
,
Martin
LA
,
Gao
Q
,
Pancholi
S
,
Mackay
A
,
Robertson
D
, et al
GDNF-RET signaling in ER-positive breast cancers is a key determinant of response and resistance to aromatase inhibitors
.
Cancer Res
2013
;
73
:
3783
95
.
16.
Macaulay
VM
,
Nicholls
JE
,
Gledhill
J
,
Rowlands
MG
,
Dowsett
M
,
Ashworth
A
. 
Biological effects of stable overexpression of aromatase in human hormone-dependent breast cancer cells
.
Br J Cancer
1994
;
69
:
77
83
.
17.
Morandi
A
,
Barbetti
V
,
Riverso
M
,
Dello Sbarba
P
,
Rovida
E
. 
The colony-stimulating factor-1 (CSF-1) receptor sustains ERK1/2 activation and proliferation in breast cancer cell lines
.
PLoS One
2011
;
6
:
e27450
.
18.
Giannoni
E
,
Taddei
ML
,
Morandi
A
,
Comito
G
,
Calvani
M
,
Bianchini
F
, et al
Targeting stromal-induced pyruvate kinase M2 nuclear translocation impairs oxphos and prostate cancer metastatic spread
.
Oncotarget
2015
;
6
:
24061
74
.
19.
Taddei
ML
,
Giannoni
E
,
Morandi
A
,
Ippolito
L
,
Ramazzotti
M
,
Callari
M
, et al
Mesenchymal to amoeboid transition is associated with stem-like features of melanoma cells
.
Cell Commun Signal
2014
;
12
:
24
.
20.
Banerjee
S
,
Zvelebil
M
,
Furet
P
,
Mueller-Vieira
U
,
Evans
DB
,
Dowsett
M
, et al
The vascular endothelial growth factor receptor inhibitor PTK787/ZK222584 inhibits aromatase
.
Cancer Res
2009
;
69
:
4716
23
.
21.
Hanahan
D
,
Weinberg
RA
. 
Hallmarks of cancer: the next generation
.
Cell
2011
;
144
:
646
74
.
22.
Sanz-Moreno
V
,
Gadea
G
,
Ahn
J
,
Paterson
H
,
Marra
P
,
Pinner
S
, et al
Rac activation and inactivation control plasticity of tumor cell movement
.
Cell
2008
;
135
:
510
23
.
23.
Parri
M
,
Chiarugi
P
. 
Rac and Rho GTPases in cancer cell motility control
.
Cell Commun Signal
2010
;
8
:
23
.
24.
Vinci
M
,
Gowan
S
,
Boxall
F
,
Patterson
L
,
Zimmermann
M
,
Court
W
, et al
Advances in establishment and analysis of three-dimensional tumor spheroid-based functional assays for target validation and drug evaluation
.
BMC Biol
2012
;
10
:
29
.
25.
Croce
CM
. 
Causes and consequences of microRNA dysregulation in cancer
.
Nat Rev Genet
2009
;
10
:
704
14
.
26.
Liu
J
,
Mao
Q
,
Liu
Y
,
Hao
X
,
Zhang
S
,
Zhang
J
. 
Analysis of miR-205 and miR-155 expression in the blood of breast cancer patients
.
Chin J Cancer Res
2013
;
25
:
46
54
.
27.
Kong
W
,
He
L
,
Coppola
M
,
Guo
J
,
Esposito
NN
,
Coppola
D
, et al
MicroRNA-155 regulates cell survival, growth, and chemosensitivity by targeting FOXO3a in breast cancer
.
J Biol Chem
2010
;
285
:
17869
79
.
28.
Kong
W
,
He
L
,
Richards
EJ
,
Challa
S
,
Xu
CX
,
Permuth-Wey
J
, et al
Upregulation of miRNA-155 promotes tumour angiogenesis by targeting VHL and is associated with poor prognosis and triple-negative breast cancer
.
Oncogene
2014
;
33
:
679
89
.
29.
Zhang
C
,
Zhao
J
,
Deng
H
. 
17β-estradiol up-regulates miR-155 expression and reduces TP53INP1 expression in MCF-7 breast cancer cells
.
Mol Cell Biochem
2013
;
379
:
201
11
.
30.
Jiang
S
,
Zhang
LF
,
Zhang
HW
,
Hu
S
,
Lu
MH
,
Liang
S
, et al
A novel miR-155/miR-143 cascade controls glycolysis by regulating hexokinase 2 in breast cancer cells
.
EMBO J
2012
;
31
:
1985
98
.
31.
Peschiaroli
A
,
Giacobbe
A
,
Formosa
A
,
Markert
EK
,
Bongiorno-Borbone
L
,
Levine
AJ
, et al
miR-143 regulates hexokinase 2 expression in cancer cells
.
Oncogene
2013
;
32
:
797
802
.
32.
Miller
WR
,
Larionov
AA
,
Renshaw
L
,
Anderson
TJ
,
White
S
,
Murray
J
, et al
Changes in breast cancer transcriptional profiles after treatment with the aromatase inhibitor, letrozole
.
Pharmacogenet Genomics
2007
;
17
:
813
26
.
33.
Dunbier
AK
,
Ghazoui
Z
,
Anderson
H
,
Salter
J
,
Nerurkar
A
,
Osin
P
, et al
Molecular profiling of aromatase inhibitor-treated postmenopausal breast tumors identifies immune-related correlates of resistance
.
Clin Cancer Res
2013
;
19
:
2775
86
.
34.
Finn
RS
,
Crown
JP
,
Lang
I
,
Boer
K
,
Bondarenko
IM
,
Kulyk
SO
, et al
The cyclin-dependent kinase 4/6 inhibitor palbociclib in combination with letrozole versus letrozole alone as first-line treatment of oestrogen receptor-positive, HER2-negative, advanced breast cancer (PALOMA-1/TRIO-18): a randomised phase 2 study
.
Lancet Oncol
2015
;
16
:
25
35
.
35.
Viale
A
,
Pettazzoni
P
,
Lyssiotis
CA
,
Ying
H
,
Sánchez
N
,
Marchesini
M
, et al
Oncogene ablation-resistant pancreatic cancer cells depend on mitochondrial function
.
Nature
2014
;
514
:
628
32
.
36.
Vellinga
TT
,
Borovski
T
,
de Boer
VC
,
Fatrai
S
,
van Schelven
S
,
Trumpi
K
, et al
SIRT1/PGC1α-dependent increase in oxidative phosphorylation supports chemotherapy resistance of colon cancer
.
Clin Cancer Res
2015
;
21
:
2870
9
.
37.
Liu
Q
,
Harvey
CT
,
Geng
H
,
Xue
C
,
Chen
V
,
Beer
TM
, et al
Malate dehydrogenase 2 confers docetaxel resistance via regulations of JNK signaling and oxidative metabolism
.
Prostate
2013
;
73
:
1028
37
.
38.
Curtarello
M
,
Zulato
E
,
Nardo
G
,
Valtorta
S
,
Guzzo
G
,
Rossi
E
, et al
VEGF-targeted therapy stably modulates the glycolytic phenotype of tumor cells
.
Cancer Res
2015
;
75
:
120
33
.
39.
Doyen
J
,
Trastour
C
,
Ettore
F
,
Peyrottes
I
,
Toussant
N
,
Gal
J
, et al
Expression of the hypoxia-inducible monocarboxylate transporter MCT4 is increased in triple negative breast cancer and correlates independently with clinical outcome
.
Biochem Biophys Res Commun
2014
;
451
:
54
61
.
40.
Ueda
S
,
Saeki
T
,
Takeuchi
H
,
Shigekawa
T
,
Matsuura
K
,
Nakamiya
N
, et al
Genomic profiling shows increased glucose metabolism in luminal B breast cancer
.
J Breast Cancer
2013
;
16
:
342
4
.
41.
Sørlie
T
,
Perou
CM
,
Tibshirani
R
,
Aas
T
,
Geisler
S
,
Johnsen
H
, et al
Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications
.
Proc Natl Acad Sci U S A
2001
;
98
:
10869
74
.
42.
D'Amato
NC
,
Howe
EN
,
Richer
JK
. 
MicroRNA regulation of epithelial plasticity in cancer
.
Cancer Lett
2013
;
341
:
46
55
.
43.
Hayes
EL
,
Lewis-Wambi
JS
. 
Mechanisms of endocrine resistance in breast cancer: an overview of the proposed roles of noncoding RNA
.
Breast Cancer Res
2015
;
17
:
40
.
44.
Miller
TE
,
Ghoshal
K
,
Ramaswamy
B
,
Roy
S
,
Datta
J
,
Shapiro
CL
, et al
MicroRNA-221/222 confers tamoxifen resistance in breast cancer by targeting p27Kip1
.
J Biol Chem
2008
;
283
:
29897
903
.
45.
Ward
A
,
Balwierz
A
,
Zhang
JD
,
Küblbeck
M
,
Pawitan
Y
,
Hielscher
T
, et al
Re-expression of microRNA-375 reverses both tamoxifen resistance and accompanying EMT-like properties in breast cancer
.
Oncogene
2013
;
32
:
1173
82
.
46.
Zhao
Y
,
Deng
C
,
Lu
W
,
Xiao
J
,
Ma
D
,
Guo
M
, et al
let-7 microRNAs induce tamoxifen sensitivity by downregulation of estrogen receptor α signaling in breast cancer
.
Mol Med
2011
;
17
:
1233
41
.
47.
Cittelly
DM
,
Das
PM
,
Spoelstra
NS
,
Edgerton
SM
,
Richer
JK
,
Thor
AD
, et al
Downregulation of miR-342 is associated with tamoxifen resistant breast tumors
.
Mol Cancer
2010
;
9
:
317
.
48.
M
,
Ding
K
,
Zhang
G
,
Yin
M
,
Yao
G
,
Tian
H
, et al
MicroRNA-320a sensitizes tamoxifen-resistant breast cancer cells to tamoxifen by targeting ARPP-19 and ERRγ
.
Sci Rep
2015
;
5
:
8735
.
49.
Mattiske
S
,
Suetani
RJ
,
Neilsen
PM
,
Callen
DF
. 
The oncogenic role of miR-155 in breast cancer
.
Cancer Epidemiol Biomarkers Prev
2012
;
21
:
1236
43
.
50.
Zhang
Y
,
Roccaro
AM
,
Rombaoa
C
,
Flores
L
,
Obad
S
,
Fernandes
SM
, et al
LNA-mediated anti-miR-155 silencing in low-grade B-cell lymphomas
.
Blood
2012
;
120
:
1678
86
.
51.
Zhu
W
,
Qin
W
,
Atasoy
U
,
Sauter
ER
. 
Circulating microRNAs in breast cancer and healthy subjects
.
BMC Res Notes
2009
;
2
:
89
.
52.
Sochor
M
,
Basova
P
,
Pesta
M
,
Dusilkova
N
,
Bartos
J
,
Burda
P
, et al
Oncogenic microRNAs: miR-155, miR-19a, miR-181b, and miR-24 enable monitoring of early breast cancer in serum
.
BMC Cancer
2014
;
14
:
448
.

Supplementary data