Investigation of discriminant metabolites in tamoxifen-resistant and choline kinase-alpha-downregulated breast cancer cells using 1H-nuclear magnetic resonance spectroscopy

Metabolites linked to changes in choline kinase-α (CK-α) expression and drug resistance, which contribute to survival and autophagy mechanisms, are attractive targets for breast cancer therapies. We previously reported that autophagy played a causative role in driving tamoxifen (TAM) resistance of breast cancer cells (BCCs) and was also promoted by CK-α knockdown, resulting in the survival of TAM-resistant BCCs. There is no comparative study yet about the metabolites resulting from BCCs with TAM-resistance and CK-α knockdown. Therefore, the aim of this study was to explore the discriminant metabolic biomarkers responsible for TAM resistance as well as CK-α expression, which might be linked with autophagy through a protective role. A total of 33 intracellular metabolites, including a range of amino acids, energy metabolism-related molecules and others from cell extracts of the parental cells (MCF-7), TAM-resistant cells (MCF-7/TAM) and CK-α knockdown cells (MCF-7/shCK-α, MCF-7/TAM/shCK-α) were analyzed by proton nuclear magnetic resonance spectroscopy (1H-NMRS). Principal component analysis (PCA) and partial least square discriminant analysis (PLS-DA) revealed the existence of differences in the intracellular metabolites to separate the 4 groups: MCF-7 cells, MCF-7/TAM cells, MCF-7-shCK-α cells, and MCF-7/TAM/shCK-α cells. The metabolites with VIP>1 contributed most to the differentiation of the cell groups, and they included fumarate, UA (unknown A), lactate, myo-inositol, glycine, phosphocholine, UE (unknown E), glutamine, formate, and AXP (AMP/ADP/ATP). Our results suggest that these altered metabolites would be promising metabolic biomarkers for a targeted therapeutic strategy in BCCs that exhibit TAM-resistance and aberrant CK-α expression, which triggers a survival and drug resistance mechanism.


Introduction
Recent advancements in high-throughput technologies, such as nuclear magnetic resonance spectroscopy (NMRS), to obtain cancer-associated metabolites have made a significant a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 were transfected with a CK-α shRNA lentiviral vector (pLenti-CK-α shRNA), a packing vector (pCMV-dR8.2) and an envelope vector (pCMV-VSVG) using lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) for lentivirus packaging. Forty-eight hours later, the virus-containing supernatant medium was collected, filtered, and concentrated by ultracentrifugation. In brief, 1x10 5 MCF-7 and MCF-7/TAM cells were seeded in a six-well plate and infected with lentivirus for 6 hours, and after replacing the culture medium, the cells were incubated for an additional 72 hours. The GFP-expressing cells were separated by a FACSCalibur flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA) equipped with a 530-nm filter (bandwidth, ± 15 nm) and a 585-nm filter (bandwidth, ± 21 nm) then analyzed using CellQuest software (BD Biosciences). The downregulation of CK-α in transduced cells was evaluated by RT-PCR and Western blot. The specific CK-α knockdown cells for MCF-7 and MCF-7/TAM were denoted as MCF-7/shCK-α and MCF-7/TAM/shCK-α, respectively.

Western blot
The cells were lysed in RIPA buffer containing a protease inhibitor cocktail (Sigma, St. Louis, MO, USA). The proteins were separated through sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membranes. The membranes were blocked with 5% skim milk in Tris-buffered saline and incubated with a primary anti-CK-α antibody (Proteintech Group, Inc., Chicago, IL, USA) and β-actin antibody (Sigma) overnight at 4˚C, followed by incubation with horseradish peroxidase-conjugated secondary antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) at room temperature for 30 minutes. The blots were developed using enhanced chemiluminescence reagents (Amersham Biosciences, Piscataway, NJ, USA). The relative intensity of the bands observed by Western blotting was analyzed using ImageJ.

Mitochondrial staining with MitoTracker CMXRos
Functional mitochondria were labeled with the mitochondria-specific red-fluorescent dye Mitotracker Red CMXRos (Molecular Probes, Eugene, OR). In brief, cells were treated with 100 nM Mitotracker Red for 10 min, fixed with 2% paraformaldehyde and counterstained with 4 0 ,6 0 -diamidino-2-phenylindole. Mitotracker Red fluorescence was detected using a fluorescence microscope (Leica, Wetzlar, Germany). A total of 1.0×10 6 cells were pelleted and suspended in 1 ml of pre-warmed complete medium with a final concentration of 100 nM Mitotracker Red for 10 min at 37˚C. Excess dye was removed with 2 washes in pre-warmed complete medium at 37˚C and the mean fluorescence intensity (MFI) value in Mitotracker Red -labeled cells was analyzed using flow cytometry (Becton Dickinson). MCF-7, MCF-7/TAM, MCF-7/shCK-α, and MCF-7/TAM/shCK-α cells were harvested, collected as cell pellets containing 3x10 7 cells per sample, and stored at -80˚C until the onset of   1 H-NMR data acquisition. In brief, frozen cell pellets were thawed with D 2 O made in PBS, mixed with 1.5 mM sodium 3-(trimethylsilyl)-propionate-2,2,3,3-d4 (TSP; Cambridge Isotope Laboratories, Inc., Andover, MA) as an internal standard, centrifuged to remove precipitates, and then placed immediately on ice until the onset of data acquisition. One-dimensional 1 H-NMR spectra were acquired on a Bruker Avance 600 system (14.1 T) equipped with 5-mm TXI cryoprobe spectrometers (Bruker BioSpin Corp. Ettlingen, Germany). Spectra were acquired using a CPMG sequence at 20˚C±1 with the following sequence parameters: flip angle = 90˚/180˚, fixed echo time = 1 ms, loop for T2 filter = 20, spectral width = 16 kHz, relaxation delay = 2 s, 32k data points, and 128 scans. NMR spectra were processed using MestRe-Nova (Mestrelab research, Santiago de, Spain) and jMRUI [20]. The time-domain data were apodized with an exponential function (1 Hz) and then Fourier transformed followed by phase-and baseline-correction. The chemical shifts of the peaks were calibrated relative to the TSP signal at 0.00 ppm. The individual metabolites were quantified by estimating the peak areas in the corresponding spectral regions of interest (Table 1), followed by normalization to the measurements for TSP.

Data analysis
The quantitative analysis of the 1 H-NMR spectra was performed on 4-6 samples of MCF-7 (n = 4), MCF-7/TAM (n = 4), MCF-7/shCK-α (n = 6) and MCF-7/TAM/shCK-α (n = 5) cells. Given the number of cell groups and 1 H-NMR-detectable metabolites and the potential correlations among the metabolites, a multivariate analysis was performed using SIMCA (v.13; Umetrics Inc., San Jose, CA). Data were first inspected by performing principal components analysis (PCA), followed by partial least-squares discriminant analysis (PLS-DA). Then, a set of 1 H-NMR measures were sorted out according to the variable influence on projection (VIP) values as a measure of the relative discriminatory potential of the individual 1 H-NMR measures [21]. Those 1 H-NMR measures with VIP>1 were considered to have contributed most to the differentiation of the cell groups, for which further statistical analyses were performed for multiple group comparisons using the Student's t-test and analysis of variance (ANOVA). A p value of <0.05 was considered to be statistically significant.

Discussion
Although it has anti-estrogen activity, TAM is widely used to treat ER-positive breast tumors as an adjuvant therapy for early-stage hormone-sensitive breast cancer or first line therapy for metastatic hormone-sensitive breast cancer, and as many as 30% of patients can be refractory to TAM and acquire resistance to TAM along with the loss of ER-α [6,[22][23][24]. CK-α is one of the targets of therapeutic strategies because it is a key enzyme in choline metabolism and might contribute to TAM resistance [11,19]. In addition, the metabolic survival-promotion function of autophagy, which is activated in drug resistant BCCs, is also one of the therapeutic targets [19]. Our recent study reported a potential connection between autophagy and CK-α in the mechanisms driving TAM resistance in ER+ BCCs. Interestingly, TAM-resistant and/or CK-α-knockdown BCCs (MCF-7/TAM and MCF-7/TAM/shCK-α) exhibited notable induction of autophagy through accumulation of LC3-II and P62 as well as the suppression of AKT, ERK and mTOR, which can contribute to the protective and survival mechanisms of MCF- 7/TAM and MCF-/TAM/shCK-α cells [19]. According to the study reported by Sanchez-Lopez et al [25], the pharmacological inhibition of CK-α by MN58b and RSM932A changes CK-α protein folding and leads to apoptosis via CHOP-mediated ER stress in cancer cells, including MCF-7, but partial genetic inhibition of CK-α by small interfering RNA (siRNA) does not induce apoptosis. The potent downregulation of endogenous CK-α protein using siRNA in breast cancer cells (MDA-MB-231, MDA-MB-468) and cervical cancer cells (HeLa) reduces proliferation, and results in significant cell death through apoptosis [12,26,27]. We rarely observed few caspase-3-stained cells, indicating that there is an apoptotic response in MCF-7/shCK-α but not MCF-7/TAM/shCK-α as well as a reduction of proliferation activity in MCF-7/TAM/shCK-α, suggesting that there is partial downregulation (approximately 30%) of the CK-α proteins in our shRNA system that is not sufficient to render apoptotic cell death but reduces proliferation activity in MCF-7/TAM/shCK-α. The partial knockdown of CK-α protein in our study may limit the reproducibility of previous studies. In addition, these discrepancies with the many previous reports would be due to distinct pharmacokinetic or target selectivity of pharmacological inhibitors as well as different knockdown efficiency of the siRNA or shRNA. When CK-α is inhibited either genetically (shRNA) or pharmacologically (CK37) in our previous study [19], shRNA and CK37 increased the autophagosomal marker LC3-II expression, but rendered differential effects on the expression level of p62, a marker of autophagic flux as shRNA, which suggest that genetic or pharmacological inhibition of CK-α can perturb a biological and metabolic system in different ways. Besides being a competitive CK inhibitor, CK37 suppresses choline uptake [28]. In general, different cellular responses can be triggered by concentration-and time-dependent pharmacokinetics of CK37. Therefore, pharmacological inhibitor should be used with caution. For this reason, the metabolic analysis of CK37-treated cells was not performed in this study. In our study, the lack of correlation between the levels of mRNA and proteins of CK-α was observed in CK-α knockdown cells. This is because protein levels are generally affected by many steps in their synthesis, stability and degradation [29]; cells can control the rates of degradation and synthesis of proteins depending on a number of different conditions, even for those proteins with similar functions. We speculate that the lack of a strong downregulation of the CK-α protein levels in CK-α knockdown cells may be associated with the steps of high stability or low degradation.
We designed the study to depict metabolic differences based on TAM resistance and CK-α expression linked with protective autophagy, which could potentially provide a direction toward targets for validation studies and the development of therapeutics in ER+ BC patients. To the best of our knowledge, this is the first study to apply 1 H-NMR to identify altered metabolites in the total lysate of TAM-resistant and/or CK-α-knockdown BCCs linked with TAM resistance as well as protective autophagy for use as predictors of the hormone and CK-α gene therapy. In the present study, we quantified a total of 33 metabolites (including 3 unknown resonances) in the MCF-7, MCF-7/shCK-α, MCF-7/TAM and MCF-7/TAM/shCK-α cells. In the subsequent multivariate analysis, a statistical model was constructed that effectively differentiated cell types according to TAM-resistance and CK-α expression. The metabolites that contributed most to differentiation were found to be fumarate, UA, lactate, myo-inositol, glycine, phosphocholine, UE, glutamine, formate, and AXP. Increased glycolysis has been linked to drug resistance through increased lactate production [30]. It was also reported very recently that lactate is critical for sustaining protective autophagy in cancer cells, including ovarian carcinoma cells, glioblastoma cells and gastric cancer cells [31,32]. In addition, elevated lactate is associated with drug resistance and stemness of BCCs, which drives recurrence, metastasis and poor clinical outcomes in BC patients [30,32,33]. Glycine provides the essential precursors for the synthesis of proteins, nucleic acids, and lipids, and it is a crucial metabolite for cancer cell proliferation and growth [34][35][36]. Glutamine participates as a nutrient in energy formation, redox homeostasis, and macromolecular synthesis and influences the suppression/ induction of autophagy through a complex mechanism [37,38]. High expression and activity of CK-α in BCCs elevates phosphocholine levels, which serves as a biomarker, thus reflecting breast cancer progression, especially in cases of drug resistance [9,10,22,39,40]. As noted in the results above, the MCF-7/TAM cells exhibiting TAM resistance and protective autophagy had high levels of lactate, glycine, glutamine, and phosphocholine, whereas CK-α knockdown reduced the levels of these metabolites, suggesting that these metabolites could be regulated by CK-α-mediated mechanisms and could serve as biomolecules and energy sources to overcome TAM insult and activate protective autophagy of damaged products. The level of myo-inositol was reported to be higher in MCF-7 and BT-474 cells (luminal A type of breast cancer cell line) than in MDA-MB-468 and MDA-MB-231 cells (triple negative breast cancer cell line), and there was also an increase in paclitaxel [41].
The CK defect causes a decreased phosphatidylcholine (PtdCho) level in the mitochondrial membrane, leading to mitochondrial dysfunction and degradation by autophagy through a process called mitophagy [27,42]. The measurement of PtdCho requires separate preparation of cells, for instance, cells dissolved in methanol-chloroform solvent, for precise quantification of PtdCho levels from 1H-NMR spectra [43,44], which was not available in our study. For this reason, we did not attempt to quantify PtdCho in our samples dissolved in D 2 O, which is one of the limitations of our study. Given its importance in the metabolism of BCCs [43,45], the measurement of PtdCho would have provided additional information in the interpretation of our data. For instance, it might have corroborated the relationships of PtdCho with mitochondrial dysfunction, especially autophagy and drug resistance in both CK-α knockdown cells and TAM-resistant cells.
Interestingly, we found that CK-α knockdown in ER+BCCs led to a decrease in Mito-Tracker CMXRos-stained mitochondria and the MIF of MitoTracker CMXRos, which indicates that there is mitochondrial dysfunction. In cancer patients, cancer cells with "healthy mitochondria" are actually more resistant to conventional therapies [46]. The treatment with TAM can induce the aggregation of mitochondria and mitochondrial-mediated apoptosis in BCCs [47], but mitochondrial function is improved in TAM-resistant BCCs [48]. In accordance with these findings, we observed that MCF-7/TAM cells exhibited higher levels in mitochondria staining and MIF of MitoTracker CMXRos compared to parent MCF-7 cells and MCF-7/shCK-α and MCF-7/TAM/shCK-α exhibited a loss of mitochondria compared to MCF-7 and MCF-7/TAM. According to a previous study by Rodríguez-González et al. [49], a strong CK-α inhibition by MN58b in cervical cancer cells induces cytochromc c release, followed by a loss of mitochondrial potential, which are associated with mitochondria-mediated apoptosis. Our study shows that a partial knockdown of CK-α by shRNA in breast cancer cells preferentially induced mitochondria-mediated autophagy. The different outcomes reported by Rodríguez-González et. al may likely be due to stronger inhibition of CK-α activity in their study. Our findings suggest that understanding mitochondrial dysfunction (mitochondriamediated autophagy or apoptosis) may be the key to unlocking new anticancer therapies and preventing the onset of drug resistance in cancer patients. Additionally, mitochondrial metabolism may be a key target for designing novel anticancer therapies.
The depletion of myo-inositol induces protective autophagy in human cells by influencing both ER and mitochondria [50,51]. Therefore, the low levels of myo-inositol in TAM-resistant cells (MCF-7/TAM and MCF-7/TAM/shCK-α) might have contributed to the induction of protective autophagy. Some cancer cells exhibit a high rate of formate release when growing in a standard culture medium, and a recent study has shown that electron transport chain dysfunction leads to a reduction in mitochondrial formate production from serine [38]. An excess amount of intracellular fumarate due to mutation or loss of fumarate hydratase is detected in the majority of primary tumors and metastases, and the involved mechanisms are potentially associated with treatment resistance [52]. Formate and fumarate also have critical roles in regulating epigenetic changes and maintaining cellular redox homeostasis [53]. Of note, formate and fumarate were remarkably decreased by CK-α knockdown and were hardly detected in the MCF-7/TAM cells. Based on the above reports, undetectable formate and fumarate in MCF-7/TAM cells might be associated with excessive activation of fumarate hydratase and dysfunction of the mitochondrial electron transport chain. These results suggest that the fumarate hydratase and mitochondrial electron transport chain could also be the target for TAMresistance and protective autophagy in ER+ BC patients.
In conclusion, the discriminant metabolites found in TAM-resistant and/or CK-α knockdown BCCs can be used as metabolic markers to predict TAM resistance and aberrant CK-α expression and may represent changes in the metabolic activity of signaling pathways underlying the protective autophagy triggered by TAM resistance and CK-α knockdown. Our findings may provide further insight into the different therapeutic responses of BCCs and additional information about current diagnostic and therapeutic assessments of TAM-resistant BCCs. Further studies on the blood samples or needle biopsies from TAM-resistant BC patients would be required to validate the potential clinical applicability of the metabolic biomarkers found in the BCCs with TAM resistance and aberrant CK-α expression.