Metabolic Effect of Estrogen Receptor Agonists on Breast Cancer Cells in the Presence or Absence of Carbonic Anhydrase Inhibitors

Metabolic shift is one of the major hallmarks of cancer development. Estrogen receptor (ER) activity has a profound effect on breast cancer cell growth through a number of metabolic changes driven by its effect on transcription of several enzymes, including carbonic anhydrases, Stearoyl-CoA desaturase-1, and oncogenes including HER2. Thus, estrogen receptor activators can be expected to lead to the modulation of cell metabolism in estrogen receptor positive cells. In this work we have investigated the effect of 17β-estradiol, an ER activator, and ferulic acid, a carbonic anhydrase inhibitor, as well as ER activator, in the absence and in the presence of the carbonic anhydrase inhibitor acetazolamide on the metabolism of MCF7 cells and MCF7 cells, stably transfected to express HER2 (MCF7HER2). Metabolic profiles were studied using 1D and 2D metabolomic Nuclear Magnetic Resonance (NMR) experiments, combined with the identification and quantification of metabolites, and the annotation of the results in the context of biochemical pathways. Overall changes in hydrophilic metabolites were largest following treatment of MCF7 and MC7HER2 cells with 17β-estradiol. However, the carbonic anhydrase inhibitor acetazolamide had the largest effect on the profile of lipophilic metabolites.


Introduction
Cancers, including breast cancer, remain a major challenge for healthcare. The complexity and adaptability of this disease continues to evade the development of effective and safe therapies. In the era of personalized medicine, the high throughput analysis of cancer metabolism under different challenges, i.e. metabolomics, is expected to provide significant novel information and tools for the analysis of drug resistance, which remains one of the major clinical setbacks in cancer treatment [1,2]. Breast cancer is a heterogeneous disease with different subtypes presenting distinct cellular and molecular characteristics. The presence or absence of a number of hormone receptors in breast cancer subtypes is an important indicator used for the optimization of therapeutic strategies [3]. Hormone receptors defining breast cancer subtypes are estrogen receptor alpha (ERα), progesterone receptor (PR) and the human epidermal growth factor receptor 2 HER2/neu (HER2 or ERBB2). These receptors may be present individually or in various combinations, which may provide information into the

Results and Discussion
The present study investigates metabolic differences between the ERα +/low HER2 mammary carcinoma cell line MCF7 and MCF7 cells stably transfected with HER2 (MCF7HER2) (ERα +/high HER2) (Figure 1), with the immortalized MCF10A normal mammary epithelial line used as an estrogen insensitive control [12]. Metabolism in these cells was compared in untreated controls and following incubation with 17β-ED or FA, in the presence or absence of the CA inhibitor acetazolamide. The effect of these compounds on the three cell lines was investigated through the analysis of changes in cellular metabolic profiles using NMR 1 H analysis with methods previously developed and used by our group [24][25][26]. Previously indicated relations between the tested molecules and some of the major proteins responsible for cancer progression and metabolism are schematically presented (Scheme 1) and were explored through metabolomics analysis in this work. 17β-estradiol (ED) is an estrogen receptor activator that induces cell division in ER+ breast cancer cells. This effect is clearly confirmed in both MCF7 and MCF7HER2 cell lines where incubation with ED causes significant increases in cell proliferation in both MCF7 and MCF7HER2 cells ( Figure 2). Ferulic acid treatment also increased cell proliferation in both MCF7 and MCF7HER2 cells (Figure 2), in agreement with previous studies [23,27]. Addition of the pan-CA inhibitor acetazolamide leads to a slight decrease in cell proliferation in both cell lines, in agreement with the previously observed anti-proliferative effect of CA inhibitors or CA knockdown [20,28]. In MCF10A cells, cellular proliferation was not affected by the treatments with ED, FA or acetazolamide (data not shown). The presence of CA in MCF7 cells has been previously established [15,20,29] and although the role of CA is particularly pronounced in hypoxia [20], CAs are expressed in cancer cell lines, even under normoxic conditions, where metabolism is enhanced due to the Warburg effect, with roles in pH regulation [30] and lactate transport to the extracellular medium [20]. Indeed removal of lactate from  17β-estradiol (ED) is an estrogen receptor activator that induces cell division in ER+ breast cancer cells. This effect is clearly confirmed in both MCF7 and MCF7HER2 cell lines where incubation with ED causes significant increases in cell proliferation in both MCF7 and MCF7HER2 cells ( Figure 2). Ferulic acid treatment also increased cell proliferation in both MCF7 and MCF7HER2 cells (Figure 2), in agreement with previous studies [23,27]. Addition of the pan-CA inhibitor acetazolamide leads to a slight decrease in cell proliferation in both cell lines, in agreement with the previously observed anti-proliferative effect of CA inhibitors or CA knockdown [20,28]. In MCF10A cells, cellular proliferation was not affected by the treatments with ED, FA or acetazolamide (data not shown). The presence of CA in MCF7 cells has been previously established [15,20,29] and although the role of CA is particularly pronounced in hypoxia [20], CAs are expressed in cancer cell lines, even under normoxic conditions, where metabolism is enhanced due to the Warburg effect, with roles in pH regulation [30] and lactate transport to the extracellular medium [20]. Indeed removal of lactate from Scheme 1. Schematic presentation of proposed effects of Estradiol and carbonic anhydrase on metabolism.
17β-estradiol (ED) is an estrogen receptor activator that induces cell division in ER+ breast cancer cells. This effect is clearly confirmed in both MCF7 and MCF7HER2 cell lines where incubation with ED causes significant increases in cell proliferation in both MCF7 and MCF7HER2 cells ( Figure 2). Ferulic acid treatment also increased cell proliferation in both MCF7 and MCF7HER2 cells (Figure 2), in agreement with previous studies [23,27]. Addition of the pan-CA inhibitor acetazolamide leads to a slight decrease in cell proliferation in both cell lines, in agreement with the previously observed anti-proliferative effect of CA inhibitors or CA knockdown [20,28]. In MCF10A cells, cellular proliferation was not affected by the treatments with ED, FA or acetazolamide (data not shown). The presence of CA in MCF7 cells has been previously established [15,20,29] and although the role of CA is particularly pronounced in hypoxia [20], CAs are expressed in cancer cell lines, even under normoxic conditions, where metabolism is enhanced due to the Warburg effect, with roles in pH regulation [30] and lactate transport to the extracellular medium [20]. Indeed removal of lactate from metabolically active cells is required to prevent intracellular acidification and a reduction in cell metabolism and proliferation [20]. metabolically active cells is required to prevent intracellular acidification and a reduction in cell metabolism and proliferation [20].    metabolically active cells is required to prevent intracellular acidification and a reduction in cell metabolism and proliferation [20].   Overall metabolic profile changes are best observed from the analysis of spectra. Principal component analysis (PCA) of the measured hydrophilic metabolites for the three cell lines incubated under different treatment conditions is shown in Figure 4A. In the untreated control condition, MCF10A cells had a highly distinct profile from that of the other cell lines, with no significant difference in the major variances between MCF7 and MCF7HER2 cells. A similar overall result was observed following incubations with acetazolamide and ferulic acid. Treatment with 17β-estradiol created change in the metabolic profile between all three cell types with significantly reduced difference between MCF10A and MCF7 and MCF7HER2 cells. According to PCA shown in Figure 4A, major PC1 variance between these three groups goes from over 70% in control group to under 50% in 17β-estradiol treated cells. At the same time PC2 component shows variance between MCF7 and MCF7HER2 cells only in 17β-estradiol treated cells. Analysis of treatments and controls for each cell line ( Figure 4B) show no significant effect on the overall profile in MCF10A cells for any of the tested treatments. In MCF7 and MCF7HER2 cells, 17β-estradiol once again had the most profound effect on PCA spectra when applied alone or in combination with acetazolamide resulting in significant changes from untreated controls. There was no measured effect on PCA of metabolic profiles following treatment of cells with ferulic acid or acetazolamide. Overall metabolic profile changes are best observed from the analysis of spectra. Principal component analysis (PCA) of the measured hydrophilic metabolites for the three cell lines incubated under different treatment conditions is shown in Figure 4A. In the untreated control condition, MCF10A cells had a highly distinct profile from that of the other cell lines, with no significant difference in the major variances between MCF7 and MCF7HER2 cells. A similar overall result was observed following incubations with acetazolamide and ferulic acid. Treatment with 17β-estradiol created change in the metabolic profile between all three cell types with significantly reduced difference between MCF10A and MCF7 and MCF7HER2 cells. According to PCA shown in Figure 4A, major PC1 variance between these three groups goes from over 70% in control group to under 50% in 17β-estradiol treated cells. At the same time PC2 component shows variance between MCF7 and MCF7HER2 cells only in 17β-estradiol treated cells. Analysis of treatments and controls for each cell line ( Figure 4B) show no significant effect on the overall profile in MCF10A cells for any of the tested treatments. In MCF7 and MCF7HER2 cells, 17β-estradiol once again had the most profound effect on PCA spectra when applied alone or in combination with acetazolamide resulting in significant changes from untreated controls. There was no measured effect on PCA of metabolic profiles following treatment of cells with ferulic acid or acetazolamide. Percentage separation was shown wherever there is a visible separation between sample groups. Separation of sample groups in any principal component axis is more significant if the percent score is larger. The third replicate in the acetamidazol/ferulic acid treatment of MCF7 cells was not included due to the quality of the NMR signal.
Further biological analysis of the effects of treatments can be performed using quantified metabolic data. Methods previously developed and utilized by our group [24][25][26]32] were used to quantify 38 metabolites previously listed as present in breast cancer cells using publically-available spectra, as shown in Figure 5. Further biological analysis of the effects of treatments can be performed using quantified metabolic data. Methods previously developed and utilized by our group [24][25][26]32] were used to quantify 38 metabolites previously listed as present in breast cancer cells using publically-available spectra, as shown in Figure 5. This relative comparison of the effects of different treatments on metabolite concentrations in three cell lines allows us to focus on changes caused by specific differences between cell lines.   Several metabolites have lower concentrations in both MCF7 and MCF7HER2 cells in all treatments compared to MCF10A cells. These include a number of amino acids including some branched chain amino acids (leucine, valine, glutamate, glycine, methionine, serine, alanine, creatine, proline, asparagine, tryptophan), as well as betaine, glycerol-3-phosphate (G3P), myoinositol, GSSG. At the same time methylamine, malate, lactic acid cis-aconic acid, citric acid, glucose, adenosine, glycerophosphocholine (GCP), tyrosine, glutamine, phosphocholine (PC), choline, histidine, arginine and ethanol are consistently more concentrated in cancer cell lines. In general the two breast cancer cell lines responded similarly to treatments with the most apparent difference between the effects of treatments on MCF7 relative to MCF7HER2 cells being in concentration changes of serine. In order to put the observed differences into the context of cellular metabolism, we show relative metabolite concentration changes following individual treatments in MCF7 and MCF7HER2 cells relative to the equivalent treatment effect in MCF10A cells.
Relative concentrations for observed metabolites involved in glycolysis, Krebs cycle, and choline pathways as well as related amino acids are shown for different treatments in Figure 7. Imbedded graphs show metabolite concentration changes following acetazolamide, 17β-estradiol and ferulic acid treatment in MCF10A, MCF7, and MCF7HER2 relative to untreated controls in the same cell lines. Several metabolites have lower concentrations in both MCF7 and MCF7HER2 cells in all treatments compared to MCF10A cells. These include a number of amino acids including some branched chain amino acids (leucine, valine, glutamate, glycine, methionine, serine, alanine, creatine, proline, asparagine, tryptophan), as well as betaine, glycerol-3-phosphate (G3P), myoinositol, GSSG. At the same time methylamine, malate, lactic acid cis-aconic acid, citric acid, glucose, adenosine, glycerophosphocholine (GCP), tyrosine, glutamine, phosphocholine (PC), choline, histidine, arginine and ethanol are consistently more concentrated in cancer cell lines. In general the two breast cancer cell lines responded similarly to treatments with the most apparent difference between the effects of treatments on MCF7 relative to MCF7HER2 cells being in concentration changes of serine. In order to put the observed differences into the context of cellular metabolism, we show relative metabolite concentration changes following individual treatments in MCF7 and MCF7HER2 cells relative to the equivalent treatment effect in MCF10A cells.
Relative concentrations for observed metabolites involved in glycolysis, Krebs cycle, and choline pathways as well as related amino acids are shown for different treatments in Figure 7. Imbedded graphs show metabolite concentration changes following acetazolamide, 17β-estradiol and ferulic acid treatment in MCF10A, MCF7, and MCF7HER2 relative to untreated controls in the same cell lines. In the untreated control group ( Figure 6) the glycolysis pathway including lactate production is enhanced in the cancer cell lines compared to MCF10A cells. In addition, concentrations of Krebs cycle intermediates are also increased in MCF7 cells possibly resulting from increased utilization of glutamate (with its concentration decrease) or from other amino acids (e.g., branched amino acids) also showing reduced concentrations compared to MCF10A cells. Increases in Krebs cycle further leads to enhanced citrate production as well as possibly contributing additionally to lactate production. Higher concentrations of histidine in the cancer cell lines possibly indicates increased PPP pathway as previously noted [33][34][35]. Average concentration change for lactic acid, succinic acid and histidine across different groups of samples are shown in in Supplementary Figure S3. Additionally, choline concentrations as well as concentrations of phosphocholine (PC) and glycerophosphocholine (GPC) are increased in cancer cell lines, once again in agreement with previous results.
Addition of 17β-estradiol leads to a number of concentration changes between these three cell lines with differences in behavior when compared against normal MCF10A cells, as well as when comparing MCF7 and MCF7HER2 cells. MCF7 cells respond to estradiol by increasing the consumption of glucose, possibly increasing the flux through the PPP cycle with a noted relative increase in histidine concentration. The effect of estradiol on Krebs cycle is more significant in MCF7HER2 cells with higher increases in the concentration of all Krebs cycle intermediates except citrate, possibly due to the ER activation of fatty acid synthesis. The effect of estrogen receptor activation also leads to significant increases in relative concentrations of choline as well as In the untreated control group ( Figure 6) the glycolysis pathway including lactate production is enhanced in the cancer cell lines compared to MCF10A cells. In addition, concentrations of Krebs cycle intermediates are also increased in MCF7 cells possibly resulting from increased utilization of glutamate (with its concentration decrease) or from other amino acids (e.g., branched amino acids) also showing reduced concentrations compared to MCF10A cells. Increases in Krebs cycle further leads to enhanced citrate production as well as possibly contributing additionally to lactate production. Higher concentrations of histidine in the cancer cell lines possibly indicates increased PPP pathway as previously noted [33][34][35]. Average concentration change for lactic acid, succinic acid and histidine across different groups of samples are shown in in Supplementary Figure S3. Additionally, choline concentrations as well as concentrations of phosphocholine (PC) and glycerophosphocholine (GPC) are increased in cancer cell lines, once again in agreement with previous results.
Addition of 17β-estradiol leads to a number of concentration changes between these three cell lines with differences in behavior when compared against normal MCF10A cells, as well as when comparing MCF7 and MCF7HER2 cells. MCF7 cells respond to estradiol by increasing the consumption of glucose, possibly increasing the flux through the PPP cycle with a noted relative increase in histidine concentration. The effect of estradiol on Krebs cycle is more significant in MCF7HER2 cells with higher increases in the concentration of all Krebs cycle intermediates except citrate, possibly due to the ER activation of fatty acid synthesis. The effect of estrogen receptor activation also leads to significant increases in relative concentrations of choline as well as phosphocholines and glycerophosphocholines and a decrease in betaine in MCF7 cells but not MCF7HER2 cells. Production of lactate increases in both MCF7 and MCF7HER2 cells.
Treatment with ferulic acid leads to lesser change in relative concentrations, however, major significant additional changes are the same as in 17β-estradiol treated cells. Once again lactate production is increased particularly in MCF7HER2 cells. Serine concentration is also increased in MCF7HER2 cells. Choline and phosphocholine are not significantly affected, however there is an increase in the relative concentration of glycerophosphocholine in MCF7HER2 cells. HER2 activated MCF7 cells can represent a hybrid between MCF7 and SKBR3 cells and treatment with hypothetic estrogen receptor activators can lead to activation of PLA 2 expression, altering ratio the between PC and GPC [36].
Treatment with the CA inhibitor acetazolamide does not alter relative metabolite concentrations between the three cell lines significantly for the majority of metabolites. Major changes are in lysine and citrate levels with an increase in concentration in MCF7 cells relative to the other two cell lines, suggesting an effect of CA inhibition on fatty acid biosynthesis in this cell line. Figure S3) possibly due to the reduced efflux out of the cell and this observation will need to be further investigated. Addition of acetazolamide to estradiol treatment leads to GPC concentration changes with a major decrease in MCF7 cells relative to cells treated only with estradiol as well as control cells. Choline and phosphocholine are directly related through the Kennedy pathway and their comparable relative concentrations are apparent in all treatment cases. Glycerophosphocholine can result from phosphatidylcholine in a PLA 2 -driven reaction further producing choline and glycerol-3-phosphate. Combined treatment with acetazolamide and ferulic acid leads to major concentration increases in lactate in MCF7 cells possibly indicating reduced efflux out of cells. As there is a reduction in the relative concentration of serine in MCF7 cells it can also be hypothesized that glycolysis in these cells increasingly goes to the lactic acid shunt, which is corroborated by decreased relative concentrations of several Krebs cycle metabolites. Concentrations of choline as well as PC and GPC show relative increases in MCF7 cells in this case while choline, PC and GPC concentration changes were observed in ER activated cells.

Concentrations of lactic acid increased the most in MCF7HER2 cells (Supplementary
Changes in lipid profiles are apparent from the analysis of NMR metabolic profiles of lipophilic metabolites. Figure 8 shows PCA of spectra for the three cell lines comparing control and treated samples. The largest differences are in acetazolamide-treated cancer cells. According to the trace analysis of PC1 ( Figure 8B), the most significant variances leading to the observed separation of control and acetazolamide-treated MCF7HER2 cells are in CH2 peak of lipids or cholesterols, CH2CH2COO and CHOCOR peak of lipids, and HC=CH peak of lipids and cholesterols. The observed differences suggest increases in lipid concentrations and a decrease in cholesterol concentrations following acetazolamide treatment.
To investigate this observation, we measured the expression of sterol-response element binding protein-1 (SREBP-1) as shown in Figure 9 under different conditions. SREBP-1 is a transcription factor that shows higher activity in the induction of genes involved in fatty acid synthesis than those participating in cholesterol synthesis [37,38]. As expected, the activation of ERα with both agonists, 17β-estradiol and FA, enhances the expression of SREBP-1 in both MCF7 and MCF7HER2 cell lines compared to untreated controls. The pan-CA inhibitor acetazolamide alone induces SREBP-1 expression in MCF7 cells but not in MCF7HER2 cells. Further lipidomics analysis is warranted in order to explore in more detail the effect of these treatments on lipid metabolism.

Reagents
Cell culture media (DMEM/F12, RPMI-1640 and phenol red free-media) and charcoal stripped FBS were purchased from Hyclone Fisher. Normal FBS was from Wisent and penicillin/streptomycin were from Lonza. 17beta-estradiol (17β-ED), deuterium oxide and DMSO were purchased from Sigma-Aldrich. The ferulic acid and acetazolamide were from Biovision and Cedarlane Labs, respectively.

Cell Culture and Western Blot
MCF7 and MCF10A cell lines were purchased from the American Type Culture Collection (Manassas, VA). The MCF7HER2 cells were generously provided by Dr L.C. Murphy from the University of Manitoba. MCF7 and MCF7HER2 cells were cultured in RPMI 1640 supplemented with 10% FBS. MCF10A cells were cultured in DMED/F12 supplemented with 100 ng/mL cholera toxin, 10% FBS, 100 U/mL penicillin and 10 µg/mL streptomycin. The three cell lines were maintained at 37˝C in a humidified environment with 5% CO 2 . Before treatments cells were cultured for one week in phenol red-free medium supplemented with 10% charcoal-stripped FBS to starve cells from steroid hormones as previously described [12]. The cells were then treated for 5 days with the ER agonists FA (10 µM) or 17β-ED (2 nM). Acetazolamide (100 µM) was then added to the culture medium for one more day prior to analyses. 17β-ED was dissolved in ethanol, ferulic acid and acetazolamide were both prepared in DMSO.

Cell Viability and Western Blot
5ˆ10 3 cells were seeded in 96-well plates and cultured under the conditions described above. Cell viability was assessed using Cell titer blue (Promega) following the manufacturer's instructions. Briefly, 20 µL of Cell titer blue was added to 100µl of media containing the treated cells and incubated for 1h at 37˝C. Then the microplate was subjected to analysis (544nm excitation/590nm emission) on a fluorescence microplate reader (VarioSkan, Thermo Scientific Corporation,). For Western blots, trypsinized MCF7, MCF7HER2 and MCF10A cells were washed in cold PBS and lysed in 50 mM tris-HCl pH 7.6, 150 nM NaCl, 2 mM EDTA and 1% Nonidet-P40 containing a cocktail of protease inhibitors (Roche). Proteins were quantified with the EZQ Protein Quantitation kit (Molecular Probes) and were separated on 4%-15% polyacrylamide Criterion TGX precast gel (Bio Rad). The separated proteins were transferred to polyvidine difluoride membranes and HER2 and SREBP-1 were detected using Cell Signaling (cat # 3250S) and BD Biosciences (Cat#557036) antibodies, respectively.

Hydrophilic and Lipophilic Metabolites Extraction
The hydrophilic metabolites were extracted as previously described [32]. Briefly, the treated cells were trypsinized, rinsed twice with PBS, centrifuged at 300ˆg for 1-min, and the pellet was kept on ice for 5 min. The pellets were resuspended in 1 mL 50% (v/v) acetonitrile/water, incubated on ice for 10 min, centrifuged at 16,000ˆg for 10 min at 4˝C, and the resulting supernatant was collected and evaporated under a stream of N 2 .
For the lipophilic metabolites the treated and trypsinized cells were washed in PBS and resuspended in 96 mL cold methanol/water (3.36/1, v/v) as previously described [25] . Samples were sonicated 3 times with 1-min cycles and 1-min wait periods. Cold chloroform was then added to the suspension in glass tubes and stored overnight at 4˝C. The following day cold chloroform and cold water (3.7 mL/pellet each) were added to the tubes, and samples were vortexed for 30 s. Homogenates were centrifuged at 200ˆg for 5 min at 4˝C. The lower phase containing the lipophilic metabolites was recuperated and dried down under a stream of N 2 [39].

NMR Experimentation and Preprocessing
The dried hydrophilic and lipophilic residues were immediately dissolved in 0.7 mL of deuterium oxide or deuterated chloroform, respectively, and pipetted into a 5 mm NMR tube for NMR analysis. All 1 H NMRs were performed on a Bruker Avance III 400 MHz spectrometer at 298 K. One-dimensional spectra for the hydrophilic fractions were obtained using a gradient water presaturation method with 512 scans, while the spectra of the lipophilic fractions were obtained using a regular proton experiment with 128 scans. Unless otherwise indicated, all analyses were performed on samples from 5 independent experiments performed on different days. NMR spectra were processed using Mnova 9.1.0. Spectral preprocessing for hydrophilic spectra included exponential apodization (exp 1), global phase correction, and normalization using the total spectral area. Spectral regions from 0.5 to 9.5 ppm were included in the normalization and analysis. For lipophilic spectra preprocessing included exponential apodization (exp 1), global phase correction including manual correction when needed, baseline corrected using Whittaker Smoother, and normalization using the total spectral area. Spectral regions from 0.5 to 8 ppm (excluding 7.0-7.5ppm solvent peak) were included in the normalization and analysis.
Data pre-processing, including data organization, removal of undesired areas and binning, as well as data presentation were performed with Matlab (Matworks Inc.). Minor adjustments in peak positions (alignment) between different samples were performed using in-house alignment software (GASP), as well as Icoshift [40] The qualitative analyses of the major variances in the spectra were performed by using principal component analysis (PCA), as well as fuzzy k-means cluster analysis using the MATLAB platform. Feature selection was performed with the Significance Analysis for Microarrays (SAM) method [41].

Metabolite Quantification
Peak assignment was performed using several methods developed in our group and elsewhere and was based on metabolic NMR databases [42,43] Spectra for around 40 metabolites used in quantification were obtained from the Human Metabolomics Database [42] or the Biological Magnetic Resonance Databank [43] and were further analyzed visually and compared to the obtained spectra. An automated method for quantification based on multivariable linear regression of spectra with appropriately aligned metabolite data from databases was previously described in detail [24][25][26] and was used in this study. The assumption behind this approach is that the spectrum of a mixture is the same as the combination (sum) of spectra of individual components measured under the same conditions. Relative metabolite concentrations were estimated using nonlinear curve-fitting with the multivariate least-squares approach. The linear regression result was used as the starting point, and the model was constrained to concentrations: c ě 0. The NMR spectra of the mixtures (samples) are modeled as a sum of spectra for components (metabolites) in the mixture. Metabolite concentrations have been normalized across samples (assuming equal total metabolite concentration). For visualization, all samples were scaled across metabolites. All metabolite concentrations for all experiments are available from the authors.

Conclusions
NMR metabolomics analysis was used to investigate the effects of 17β-estradiol, an ER activator and ferulic acid, a CA inhibitor as well as ER activator, in the absence and in the presence of the CA inhibitor acetazolamide on the metabolism of MCF7 and MCF7HER2 cells. Overall changes in hydrophilic metabolites were largest following treatment of MCF7 and MC7HER2 cells with 17β-estradiol. However, the CA inhibitor acetazolamide has the largest effect on the profile of lipophilic metabolites. Quantification of metabolites from NMR data provided information for analysis of the effects of treatments on individual metabolites and pathways. This analysis shows some similarities and differences in the effects of the three compounds suggesting possible effects of CA inhibition as well as ER activation on production and utilization of lactate, energy, fatty acids and choline derivatives. Further work will be needed to explore the effects of these types of compounds in hypoxic conditions and on fluxes of metabolites through specific pathways, as well as to further investigate the effects of specific inhibitors of different CA isozymes.