Altered glutamine metabolism exposes EMT derived mesenchymal cells to PI3K/Akt/mTOR pathway inhibition

Background: Epithelial-to-mesenchymal transition (EMT) is a fundamental developmental process with strong implications in cancer progression. Understanding the metabolic alterations associated with EMT may open new avenues of treatment and prevention. Methods: We utilize 13 C carbon analogs of glucose and glutamine to examine difference in internal flux patterns of metabolites in central carbon and lipid metabolism following EMT in breast epithelial cell lines. Furthermore, an isotopomer spectral analysis, 13 C-metabolic flux analysis and weighted correlation network analysis are utilized for investigation of the alterations in metabolic functionality following EMT in breast. Results: There are inherent differences in metabolic profiles before and after EMT. We observed EMT-dependent re-routing of TCA-cycle flux characterized by increased mitochondrial IDH2 -mediated reductive carboxylation of glutamine to lipid biosynthesis with a concomitant lowering of glycolytic rates and glutamine-dependent glutathione (GSH) generation. Our network approach identified specific subtype of cancer drugs that are significantly associated with GSH abundance and we confirmed this in vitro . Conclusions: EMT-linked alterations in GSH synthesis modulate the sensitivity of breast epithelial cells to PI3K/Akt/mTOR inhibitors.

Metabolic reprogramming is recognized as one of the ten cancer hallmarks as proposed by Hanahan and Weinberg [2]. In contrast to rapidly dividing cancer cells, a mesenchymal phenotype faces a different set of metabolic requirements whose relation to malignant transformation has been intensely studied and associated with enhanced glycolysis, increased glutaminolysis, nucleotide metabolism and abnormal choline metabolism [3,4,5]. Quantitative understanding of the metabolic requirements of mesenchymal cells is however lacking, particularly the changes to the turnover and quantity of metabolites involved in xenobiotic clearance, i.e. the drug response of cells.
Cancer cells that undergo EMT have increased resistance to various drugs [6,7,8], which indicates that the xenobiotic clearance of the cells is altered. There are three steps involved in the metabolism of xenobiotics: 1) Modification, 2) Conjugation and 3) Excretion.
Conjugation involves the binding of special metabolites (e.g. glutathione, UDP-glucuronate, PAPS, S-adenosylmethionine) to a xenobiotic compound, which leads to the assumption that the availability of these particular metabolites within cells influences the activity of the drug. Therefore, accurate metabolic measurements of EMT may contribute to better understanding of the drug resistance of cancer cells and lead to novel therapeutic approaches aimed at eliminating metastatic cancer cells.
We have previously used both ultra-performance liquid chromatography coupled mass spectrometry (UPLC-MS) and NMR to study EMT and cancer metabolism [9,10,11]. Integrated analyses of these metabolomics data with transcriptomic and proteomic data within genome-scale metabolic models predicted metabolic differences that occur following EMT in breast epithelium [11]. These included alterations to glycolysis, the pentose phosphate pathway TCA flux and fatty acid synthesis. Although these models provided useful insights into metabolic alterations associated with EMT, they lacked accuracy in predicting internal fluxes in a quantitative manner in the compartmentalized central carbon metabolism.
In order to better understand the metabolite flux changes that accompany EMT we characterized the internal flow of metabolites in D492 breast epithelial cells and their mesenchymal variant, D492M, to provide a detailed, quantitative analysis of metabolic reprogramming following EMT in breast. We performed stable isotope tracing of 13 C labelled glucose and two separate 13 C labelled glutamine analogs. UPLC-MS and NMR were used to measure label incorporation into metabolites associated with central carbon metabolism and lipid biosynthesis. Next, we utilized these data to perform 13 C-metabolic flux analysis (13C-MFA) and isotopomer spectral analysis (ISA) to provide computational predictions of 1) absolute flux values based on measured uptake rates of different carbon sources, 2) fractions of lipogenic acetyl-CoA from each carbon source and 3) fractions of newly synthesized fatty acids. We subsequently performed shRNA lentiviral silencing of key genes to further elucidate their role in EMT metabolic re-programming.
Finally, using an integrated network analysis of the NCI-60 Human Tumor Cell Line panel and an untargeted metabolomic analysis, we investigate how the EMT-dependent re-routing of central carbon metabolism affects drug responsiveness in D492 and D492M cells.

Methods
Cell culture D492 and D492M cells were cultured in DMEM/F12-based medium H14 at 37C in 5% CO2 as previously described [12]. For the labeling experiments, the cells were fed with medium containing 100% 13 C labelled glutamine at the 1 or 5 position (Cambridge Isotope Laboratories, Inc.) or 13

Proliferation assay
Cells were seeded in quadruplicates in 48 well plates (10.000 cells/well). They were grown in a large chamber incubation system (PeCon) at 37°C in 5% CO2 and imaged for 12-72 hours using Leica DMI6000B. Images of cells were opened with Fiji [13], where the cells were counted with the help of an in-house script. The slope of the best fitting line through log-transformed cell number over time was used to represent proliferation rate.

Detection of intracellular NADP + and NADPH
NADP + and NADPH were measured using NADP/NADPH-Glo™ Assay (G9071, Promega, Madison, WI). Cells were seeded in triplicates in opaque 96 well plates (10.000 cells/well) and incubated at 37C in 5% CO2. After 24 hours, the medium was removed, cells were washed with cold PBS and then supplemented with 50µL and 50µL 1% DTAB in 0.2N NaOH solution to induce cell lysis. Next steps were according to manufacturer's protocol. The luminescence was measured 50 minutes after addition of the NADP-NADPH-Glo TM Detection reagent in SpectraMax M3 Microplate Reader from Molecular Devices (San Jose, CA, USA).

Nuclear magnetic resonance (NMR)
For NMR analysis, D492 and D492M cells were cultured in T225 flasks in supplemented DMEM/F12 until they reached approximately 70% confluency. Cells were then fed with either 1,2-13 C glucose or 1-13 C glutamine for 6 hours. Parallels without 13 C tracers were also cultured. Culture medium was collected after incubation. Methanol extracts from glucose-and glutaminelabelled cells were prepared as described previously [14]. analysis. The spectra were baseline corrected using asymmetric least squares method [16] and peak aligned using icoshift [17]. The water peak and areas in the spectra with contamination and noise only were removed. All spectra were mean normalized and mean centered. Principal component analysis (PCA) was performed using PLS toolbox v8.2.1 (Eigenvector Research). Proton decoupled 13 C spectra (Bruker: zgpg30) were acquired using a power gated decoupling sequence with a 30° pulse angle as described in Bettum et al. [18]. The spectra were collected with either 4K (for 1,2-13 Cglucose) or 16 K (for 1-13 C-glutamine) scans and 16 dummy scans.

UPLC-MS
Before UPLC-MS analysis, the organic phase was reconstituted in MTBE before a methanol solution containing 1M NaOH was added   [20]. When calculating the total contribution of carbon sources to metabolites, we used the following equation [21]: Where n is the number of C atoms in the metabolite, i represents the isotopologues and m is the relative fraction of the isotopologues.
In order to evaluate the percentage of glucose that enters the pentose phoshate pathway, we utilized a formula from Lee et al. [22]: In equation 2, m1 and m2 are the fractional abundances of M+1 and M+2 lactate isotopologues, respectively, (e.g. from Fig. 1).

Isotopomer spectral analysis
To determine the fractional contribution of glutamine and glucose to fatty acid biosynthesis, we used the convoluted isotopic spectral analysis algorithm [23]. The chosen proxy for fatty acid labeling was palmitate (16:0). The MDV vector for palmitate was corrected for natural abundance using IsoCor. The corrected MDVs for palmitate were imported into MATLAB where the analysis was carried out. 13 Cglucose, 1-13 C-glutamine and 5-13 C-glutamine were used for simultaneous fitting of our network. More specifically, the 6-hour timepoints were used. In our analysis, the following assumptions were made: 1 The cells are in the exponential growth phase, a phase where metabolic steady state is present.
3 Serine one-carbon metabolism was not included. We saw that glucose contribution to glutathione was minimal, indicating that the glycine used for GSH generation is not derived from glucose.
The flux of individual reactions was estimated relative to uptake of glutamine. The actual flux value (units of µmol/gDW/hour) was then found by multiplying by measured glutamine uptake rates of the cells from [11]. The sum of squared residuals (SSR) was minimized between experimentally determined MIDs of 7 metabolites and simulated fluxes in the network. Goodness-of-fit was determined using the chi-square test. The fit was accepted within the 95% confidence interval of the chi-square distribution.

Weighted correlation network analysis
Drug sensitivity data were gathered from the NCI-60 Human Tumor Cell Lines [25] using the rcellminer R-package [26]. We focused specifically on FDA-approved drugs and cells that were contained within the independent dataset from Ortmayr et al. [27]

Statistical analysis
A Student's t-test was employed for comparison of two treatments, and in the case of non-parametric data, a Mann-Whitney U -test was used. One-way ANOVA was used to compare data from three or more treatments. The asterisks in each figure represent the p-values (* < 0.05, ** < 0.01, *** < 0.001). Data were assumed to be normally distributed unless otherwise stated in figure legends. Statistical analysis and image generation was carried out in the R environment [28] using the ggplot2 and ggpubr packages. All data are presented as mean + standard deviation.

Results
Glycolysis rates determine the pentose phosphate shunt in the D492 EMT cell model Unlike the flux rate differences observed within glycolysis, these results indicated significant changes to the flux network topology following EMT.
Citrate is used for energy production following its oxidation in the TCA cycle or as a precursor for lipids [30]. Based on the reduced synthesis of citrate from glucose, we hypothesized that the generation of citrate in D492M could be more dependent on another carbon source the amino acid glutamine. Glutamine the second most consumed carbon source after glucose in D492 and D492M cells [11] and we found that it is essential for their proliferation (Additional file 5). The overall contribution of glutamine to the TCA cycle was analyzed by label incorporation from 13 C glutamine to citrate. By utilizing either 1-13 C or 5-13 C-labelled glutamine analogs ( Fig. 2A), we quantified both the reductive and oxidative flow of carbons from glutamine within the TCA cycle [31] (Additional files 6-7). D492 epithelial cells primarily oxidized glutamine in the TCA cycle while the D492M mesenchymal cells had significantly higher flux through reductive carboxylation to citrate ( Fig. 2B and C), confirming our hypothesis of increased reliance on glutamine for citrate synthesis following EMT. Furthermore, there was higher incorporation of glutamine into glutathione (Fig. 2D).
We next investigated the transfer of labelled carbons to fatty acids from glucose and glutamine to further determine the metabolic fate of citrate. To this end we specifically looked at the label incorporation from both 1,2-13 C-glucose and 5-13 Cglutamine into palmitic acid (C16:0). The contribution of glucose to fatty acid synthesis was relatively higher in D492, whereas the contribution of glutamine to fatty acids was higher in D492M ( Fig. 3A and B). These results are consistent with the increased contribution of glutamine to the citrate pool following EMT. Isotopomer spectral analysis (ISA) of labelled palmitic acid revealed that the pool of lipogenic acetyl-CoA derived from glucose was significantly diminished while the glutamine-derived pool is increased following EMT in D492 epithelial cells (Fig. 3C).

Metabolic re-routing following EMT affects redox metabolism in D492 cells
The synthesis of fatty acids requires both acetyl-CoA from a carbon source and reducing potential in the form of NADPH. The spatial difference in requirement of reducing potential (cytosol vs. To put the results from the 13 C tracing experiments into quantitative context, we combined our results with the cell-specific uptake rates from our previous study [11] via 13

IDH2 knockdown increases glucose-dependent fatty acid synthesis
Based on the difference in reductive carboxylation ( Fig. 2B and 3G) and recent literature [33,34], we hypothesized that isocitrate dehydrogenase (IDH) would contribute to the discrimination between the D492 and D492M flux phenotypes through glutamine consumption and redox balance. RNA sequencing and proteomic data from the cell lines clearly showed that one isoform of IDH in particular was higher in abundance in D492M cell lines compared to D492. This was the mitochondrial NADPH + -dependent isocitrate dehydrogenase IDH2 (Fig. 4A) which indicated that the increased reductive carboxylation in D492M could depend on the increased expression of this isoform. After knocking down the mRNA levels of IDH2 using shRNA lentiviral transduction, we investigated its metabolic and morphological effects on both cell lines. When the IDH2 gene was silenced (Fig. 4B), the mRNA levels from the cytosolic isoform IDH1 were increased (Fig. 4C). There were no morphological differences observed when IDH2 expression was diminished in neither D492 nor D492M cells (Fig. 4D). However, the acetyl-CoA (Fig. 4G). Taken together, these results demonstrate a fundamental role of IDH2 in the increased reductive carboxylation of glutamine for lipogenesis following EMT of D492 cells.
Alteration in redox metabotype drives sensitivity to PI3K/Akt/mTOR inhibition The metabolic phenotypes characterized by different glycolytic rates and altered carbon source preference for TCA, changes to total concentrations and synthesis rates of proline and glutathione are reminiscent of cancer stem cell metabotypes [35,36]. Due to glutathione's role in drug resistance in various cell types [37,38,39], we hypothesized that metabolic rerouting centered around alternate utilization of glutamine-derived glutamate for glutathione synthesis would contribute to the different drug sensitivities of D492 and D492M.
To identify drugs that are selectively affected by glutathione concentrations within cells, we performed an integrated network analysis of I) drug sensitivity profiles within the NCI-60 Human Tumor Cell Line database [25] and II) untargeted metabolomic analysis of NCI-60 cell lines from Ortmayr et al. [27]. The network analysis revealed that FDA-approved drugs in the NCI-60 are grouped into 6 intracorrelated drug modules (Fig. 5A). and rapamycin from the blue module (Fig. 5C), but also to paclitaxel, a taxane drug belonging to the red module. In order to establish a functional link between glutathione and the blue module drugs, we co-treated D492 and D492M cells with buthionine sulphoximine (BSO), an inhibitor of the rate-limiting enzyme glutamate-cysteine ligase (GCL) in glutathione synthesis (Fig. 5D-E), and either everolimus (Fig. 5F) or paclitaxel (Fig. 5G). We saw that the sensitivity of both D492 and D492M to everolimus could be enhanced by cotreatment with BSO, whereas these effects were not observed when the cells were co-treated with paclitaxel and BSO. Together, these data suggest that glutathione availability directly affects sensitivity to drugs that affect the PI3K/Akt/mTOR pathway.

Discussion
D492 and D492M cells represent only two of the numerous phenotypes within the spectrum of EMT [40]. Herein, we have thoroughly characterized the central carbon metabolic activity of these cell types using 13 C-labelled carbon sources along with metabolic flux analysis, isotopomer spectral analysis, and both in silico and in vitro drug-sensitivity analyses.

IDH2 plays a key role in EMT in breast
The Finally, the incorporation of glutamine carbons into fatty acids from reductive carboxylation is replaced by glucose carbons when the IDH2 levels are lowered in both cell lines (Fig. 4G).
Our findings highlight the role of IDH2 in the increased reductive carboxylation following EMT in breast. However, we cannot exclude the importance of IDH1 in this context. It is reasonable to assume that the reductive carboxylation works through IDH1 in the cytosol, and citrate is subsequently transported into the mitochondria where IDH2 takes part in its ultimate oxidation (as proposed by Jiang et al. [33]). When IDH2 levels are diminished, the activity of this pathway is inevitably halted. Nevertheless, our results demonstrate that IDH2 knockdown significantly affects the reductive carboxylation of glutamine to citrate and ultimately fatty acids which unequivocally establishes a crucial, functional role of the mitochondrial isoform in this process.

Alterations in reductive carboxylation and redox metabolism follow EMT in breast
Our results show that glutamine-derived citrate is being utilized for fatty acid synthesis in the D492 cell model, but the reliance on this pathway is enhanced following EMT. We show that there is a concurrent increase in NADPH/NADP + ratio and proline synthesis along with a decrease in glutathione synthesis (Fig. 3D-F showed that its degradation and cycling is higher in breast cancer cells grown in 3D culture than in 2D [41], which ultimately altered the NADPH/NADP + balance. Phang hypothesized that this could be due to the fact that proline is being directed away from protein synthesis and towards redox regulation [42], a pathway that proline has previously been shown to take part in [43]. More recently, Liu et al.
showed that under hypoxic conditions, proline synthesis and reductive carboxylation, both of which require NADPH, act as alternative bins for electrons so that electron transfer can occur for cellular proliferation [36]. These studies fit well with our observations, where D492M cells indeed have lower oxygen consumption rate [11] and show higher activity of aforementioned pathways. More specifically, the higher NADPH availability of D492M is reflected in increased reductive carboxylation and proline synthesis in the cells. Furthermore, D492M cells display a concomitant lowering of glutathione synthesis and its overall abundance, a phenomenon shown to occur when EMT-inducing transcription factor Snail is overexpressed in MCF7 breast cancer cells [44].

Diminished glutathione abundance potentiates sensitivity to PI3K/Akt/mTOR inhibitors
We have previously reported a reduction in oxidative phosphorylation following EMT in D492 cells. This causes a metabolic shift towards anaplerosis and upregulation of pathways receiving otherwise ETC-directed electrons (i.e. proline synthesis and reductive carboxylation) [36] and decreased flux of glutamine towards glutathione synthesis. Glutathione is the most abundant non-protein thiol in animal cells, and it plays a crucial role in the conjugation phase of xenobiotic metabolism. This leads to increased water-solubility of foreign compounds such as drugs and reduced efficacy [45,46].
Due to the clear differences we observed in glutathione synthesis and overall abundance between D492 and D492M cells ( Fig. 3 and additional file 3B), we hypothesized that this would result in altered drug sensitivity of the cells. Integrated network analysis suggested that D492M cells are more affected than D492 by drugs that specifically target mTOR, PI3K and Akt along with acetalax and A2O3, a drug known to be affected by glutathione abundance in cancer cells [37] (Fig. 5A-B). Furthermore, the lack of a significant relationship of the other metabolites shown, S-adenosylmethionine and UDP-glucuronate, which are also known to partake in the conjugation to xenobiotic compounds, suggests that these drugs are specifically affected by glutathione availability. We tested several drugs that target mTOR, PI3K and Akt and found that D492M cells were more sensitive than D492 cells (Fig. 5C). Furthermore, we showed that by manipulating the glutathione levels within the cells via BSO treatment (Fig. 5E), we could increase the sensitivity to these drugs (Fig. 5F), but the same manipulation did not affect the sensitivity to paclitaxel, a microtubule stabilizer and mitotic inhibitor (Fig. 5G). These results indicate that glutathione availability primarily affects drugs that target the PI3K/Akt/mTOR pathway. In recent years, studies have shown a direct relationship between the PI3K/Akt/mTOR signaling pathway and oxidative stress response [47,48,49]. Furthermore, the PI3K/Akt/mTOR pathway has been shown to be highly involved in the EMT process and chemoresistance of ovarian cancer cells and melanoma [50,51].    . E) Proline synthesis in D492 and D492M cells cultured with 5-13 C-glutamine for 6 hours (n=3). F) Glutathione synthesis from glutamine in D492 and D492M in cells cultured with 1-13 C-glutamine for 6 hours (n=3). G) Relative flux differences according to steady-state metabolic 13 C flux analysis using INCA. Different pathways are indicated in grey. For both D492 and D492M models, the expected 95% confidence interval for the sum-of-squared residuals (SSR) for 108 degrees-of-freedom was 81.1 -138.7. Both models had lower SSR than expected (44.5 and 100.8 for D492 and D492M, respectively). Results in A-F are presented as mean + std.