Methotrexate elicits pro-respiratory and anti-growth effects by promoting AMPK signaling.

One-carbon metabolism fuels the high demand of cancer cells for nucleotides and other building blocks needed for increased proliferation. Although inhibitors of this pathway are widely used to treat many cancers, their global impact on anabolic and catabolic processes remains unclear. Using a combination of real-time bioenergetics assays and metabolomics approaches, we investigated the global effects of methotrexate on cellular metabolism. We show that methotrexate treatment increases the intracellular concentration of the metabolite AICAR, resulting in AMPK activation. Methotrexate-induced AMPK activation leads to decreased one-carbon metabolism gene expression and cellular proliferation as well as increased global bioenergetic capacity. The anti-proliferative and pro-respiratory effects of methotrexate are AMPK-dependent, as cells with reduced AMPK activity are less affected by methotrexate treatment. Conversely, the combination of methotrexate with the AMPK activator, phenformin, potentiates its anti-proliferative activity in cancer cells. These data highlight a reciprocal effect of methotrexate on anabolic and catabolic processes and implicate AMPK activation as a metabolic determinant of methotrexate response.

needed to build nucleotides 18 . MTX also inhibits 5-aminoimidazole-carboxyamide ribonucleotide formyltransferase (ATIC) and thymidylate synthetase (TYMS), which are enzymes of the purine and pyrimidine biosynthesis pathways respectively 19 . Despite its wide use as a cancer drug, MTX can have high toxicity by targeting metabolic enzymes found in both transformed and non-transformed cells, hence reducing its therapeutic index 20,21 . In this manuscript, we uncover that MTX promotes AMPK signaling in breast cancer, which results in stimulation of mitochondrial metabolism and inhibition of cellular proliferation. This knowledge will assist in refining cancer-specific therapeutic strategies involving MTX.

Results
Methotrexate induces AMpK signaling. Given that MTX inhibits purine biosynthesis, we first quantified the levels of metabolic intermediates in this pathway upon treatment. MTX caused a strong increase (~30 fold) in endogenous AICAR levels in breast cancer cells and mouse embryonic fibroblasts (MEFs) (Fig. 1A, Supplementary Fig. 1), with little effect on the downstream metabolites IMP and AMP, suggesting a blockage in de novo purine biosynthesis at the ATIC step. AICAR is used as an exogenous compound to activate AMPK in various cell models 22 , hence we assessed whether the increase in endogenous AICAR levels upon methotrexate treatment was sufficient to promote AMPK activation. MTX treatment increased the phosphorylation of Ser 79 on acetyl-CoA carboxylase (pACC) 23 , and the phosphorylation of Thr 172 on AMPK, indicating that AMPK is activated (Fig. 1B,C). PGC-1α signaling is a known downstream effector of AMPK activation in both non-transformed and transformed cells [24][25][26] . Accordingly, MTX treatment increased the expression of PPARGC1A and its partner ESRRA in BT-474 cells, indicating that MTX upregulates the PGC-1α/ERRα axis (Fig. 1D). In addition, MTX decreases the expression of MTHFD1L (Fig. 1D), a folate cycle gene that is repressed by AMPK/PGC-1α/ ERRα signaling 26 . Collectively, these data show that MTX treatment promotes AMPK signaling.
Methotrexate promotes AMpK-dependent mitochondrial respiration. To test the biological implications of AMPK activation upon MTX treatment, we first performed respirometry experiments given that AMPK engages the PGC-1α/ERR axis, which is a central regulator of mitochondrial oxidative phosphorylation. In accordance with the role of AMPK in promoting catabolic reactions, MTX increased cellular respiration in breast cancer cells and non-transformed mammary cells, including the respiration linked to ATP synthesis (coupled respiration) and the respiration linked to proton leak (uncoupled respiration) ( Fig. 2A, Supplementary  Fig. 2A-F). We also formally quantified the impact of MTX on global cellular bioenergetics 28 . MTX treatment increased basal total cellular ATP production (J ATP total), which was largely due to an increase in oxidative phosphorylation (J ATP ox), with a small contribution from glycolysis (J ATP glyc) (Fig. 2B). MTX treatment also increased maximal total bioenergetic capacity (Fig. 2C,D) and the levels of aspartate, a metabolite linked to increased respiration in proliferating cells 27 (Fig. 2E). In addition, MTX promoted mitochondrial metabolism in non-transformed MEFs. Indeed, MEFs treated with MTX displayed increased total, uncoupled and coupled respiration at baseline, similar to cancer cells (Fig. 2F,G-I blue bars). To determine if the MTX-induced increase in oxidative metabolism was AMPK-dependent, MEF cells deficient for AMPKα1/2 were treated with MTX. AMPK-null MEF cells showed no significant increase in oxidative metabolism upon MTX treatment (Fig. 2F,G-I purple bars). Taken together, these results demonstrate that MTX promotes mitochondrial respiration in an AMPK-dependent manner.
Methotrexate exerts AMPK-dependent anti-proliferative effects. It is well established that MTX acts as a chemotherapeutic agent by inhibiting cellular proliferation 29 . Given that AMPK-mediated metabolic reprogramming has previously been shown to impact proliferation and tumour growth 30,31 , we tested if the anti-proliferative effect of MTX is dependent on AMPK. We first examined this in eμ-Myc B cell lymphoma cells using an AMPKα1/α2 hairpin construct (shAMPK) targeting AMPK. Knockdown of AMPK in cancer cells formate supplementation does not rescue the AMpK-dependent effects of MtX treatment. Genetic defects in one-carbon metabolism have been rescued by formate supplementation 32 . To determine if the AMPK-dependent effects of MTX treatment can be rescued by formate supplementation, BT-474 cells were treated with a combination of MTX and sodium formate. Formate supplementation did not rescue the MTX-mediated induction of AICAR or ZMP, which resulted in unchanged AMPK signaling activation ( Fig. 4A-D), and unaltered cellular bioenergetics (Fig. 4E,F). Furthermore, formate did not rescue the MTX-mediated decrease in cell count (Fig. 4G). Ultimately, the AMPK-dependent effects of MTX do not appear to arise due to lack of formate.
Biguanides potentiate the anti-neoplastic effect of methotrexate. Cancer cells are dependent on one-carbon metabolism to build nucleotides de novo 2 . Recently, we have shown that one-carbon metabolism is inhibited through activation of AMPK/PGC-1α/ERRα signaling, resulting in increased sensitivity to MTX 26 . In addition, the AMPK/PGC-1α/ERRα axis is activated by treatment with biguanides 24,26,[33][34][35] . To investigate whether co-treatment with biguanides can potentiate the anti-proliferative effects of MTX, cells were pre-treated with phenformin for 24 hours before treatment with MTX for 72 hours. Cells treated with both phenformin and methotrexate demonstrated lower cell counts than those treated with methotrexate or phenformin alone (Fig. 5A,B). AMPK-null MEF cells were also less sensitive to the combination of phenformin and MTX in terms of cell proliferation, viability, and cell cycle progression compared to WT cells (Fig. 5C-E), further highlighting the impact of AMPK on MTX response. These data show that activation of AMPK using phenformin can potentiate the anti-proliferative effect of MTX in cancer cells.

Discussion
In this study, we report that AMPK signaling plays a key role in mediating the metabolic and anti-proliferative effects of MTX (Fig. 5F). MTX treatment increases endogenous levels of AICAR, leading to AMPK activation. Increased AMPK signaling results in a decrease in one-carbon metabolism gene expression and cell proliferation. In addition, elevated AMPK signaling augments cellular global bioenergetic capacity, mainly by promoting oxidative phosphorylation. The anti-proliferative effects of MTX was potentiated by phenformin, a complex I inhibitor that activates AMPK, which suppresses one-carbon metabolism through PGC-1α/ERRα signaling 26 . These data highlight potential synergy between MTX and modulators of mitochondrial bioenergetics, and underscore the importance of understanding the metabolic impact of cancer drugs to improve their chemotherapeutic action.
MTX is a potent inhibitor of one-carbon metabolism and nucleotide biosynthesis. Previous work has shown that MTX treatment causes an accumulation of AICAR 36 sufficient to activate AMPK in breast cancer cells [37][38][39][40] . MTX-induced AMPK signaling was shown to promote glucose uptake and lipid oxidation 39 , as well as inhibit protein synthesis 40 . The anti-folate drug pemetrexed also induces AICAR-mediated AMPK activation and inhibits anabolic processes in colon cancer cells and lymphoblastic leukemia cells 41,42 . Here, we show that MTX does not only inhibit anabolic processes, but also promotes catabolic metabolism by stimulating mitochondrial respiration.
Our group has previously shown that the AMPK/PGC-1α/ERRα axis can inhibit one-carbon metabolism and purine biosynthesis, resulting in decreased proliferation 26 . Given that the anti-proliferative effects of MTX are linked to AMPK activity, we show in this study that the AMPK activator phenformin can improve the anti-proliferative action of MTX. This is line with reports showing that pharmacological AMPK activation with AICAR also increases the antineoplastic effect of MTX in breast, skin, and prostate cancer models 37 . However, given that AICAR has poor bioavailability 43 , combining MTX with AMPK activators that have higher bioavailability, such as biguanides, may prove to be an effective option in the clinic.
There has been increased interest in repurposing biguanides for cancer treatment. Indeed, both phenformin and metformin display AMPK-dependent and -independent anti-neoplastic effects in several cancer subtypes including breast, prostate, and colorectal cancers 44,45 . Despite these successes, clinical trials regarding the usage of biguanides in cancer have been disappointing 44 . Given that biguanides and MTX both share a common effector in AMPK, the combination of both drugs may have strong clinical implications in improving AMPK-dependent anti-neoplastic response. Indeed, AMPK has been reported to have both anti-Warburg and anti-proliferative effects 31 . However, the overall outcome of AMPK activation in cancer is still unclear as AMPK can limit anabolic processes required for proliferation, yet promote metabolic adaptation of tumours to cope under energetic stress 46 . www.nature.com/scientificreports www.nature.com/scientificreports/ Overall, our results highlight the importance of AMPK in mediating chemosensitivity. Our work will likely stimulate translational research aimed at testing whether biguanides could be used as neoadjuvant therapy to improve chemotherapeutic response to anti-folates.
Quantitative Rt-pcR. RNA was extracted and purified using the Aurum Total RNA Mini Kit (Bio-Rad) following the manufacturers' protocols. Reverse transcriptase reactions were performed using iScript cDNA Synthesis Kit (Bio-Rad). Samples were then analyzed by qRT-PCR with SYBR-green-based qRT-PCR with a MyiQ2 Real-Time Detection System (Bio-Rad). Gene-specific primers are found in Supplementary Table 1. Values were normalized to PUM1. www.nature.com/scientificreports www.nature.com/scientificreports/ Respirometry. Cellular respiration was determined using a Digital Model 10 Clark Electrode (Rank Brothers) as previously described 50 . 1 × 10 6 cells (BT-474, MCF10A, MCF7) or 1.5 × 10 6 (NMuMG, NT2196) cells were treated with methotrexate for 72 hours and used for respiration measurements. Oligomycin (2.5 μg/ mL/1 × 10 6 cells) was added to inhibit the ATP synthase, which allows for the calculation of respiration coupled to ATP synthesis (coupled respiration) and respiration linked to proton leak (uncoupled respiration). Myxothiazol (0.5 μM/1 × 10 6 cells) was added to inhibit complex III, and is used to determine the contribution of non-mitochondrial respiration, which was not observed.

Seahorse bioenergetic analysis. Oxygen Consumption Rate (OCR) and Extracellular Acidification Rate
(ECAR) were measured using the XFe24 Extracellular Flux Analyzer (Agilent Technologies, Santa Clara, CA, USA) according to the manufacturer's protocol. BT-474 cells were treated for 72 hours with methotrexate, then trypisizinzed and 50,000 cells per well were plated overnight in Seahorse cell culture plates (Agilent Technologies, Santa Clara, CA, USA). The next day, cells were washed with XF media (Seahorse Bioscience) without glucose and incubated in a CO 2 -free incubator at 37 °C for 2 hours to establish equilibrium. Basal conditions include XF media with 10 mM glucose. Oligomycin (1 μM), FCCP (1.5 μM), rotenone and antimycin A (0.5 μM each), and monensin (20 μM) were used. Oligomycin is an inhibitor of ATP synthase. FCCP uncouples the inner mitochondrial membrane, thereby allowing for maximal oxygen consumption. The combination of rotenone (complex I inhibitor) and antimycin A (complex III inhibitor) is used to maximally perturb mitochondrial respiration. Monensin is used to calculate the maximum glycolytic capacity of cells, by driving increased ATP demand by the Na + /K + -ATPase. In addition, HCl (hydrochloric acid) was injected into wells containing media with no cells (four injections of 0.25 mM each) in order to calculate the buffer capacity of the XF media. OCR, ECAR, and PPR (proton production rate) measurements were taken before and after each injection, and were used to calculate ATP production (J ATP total) and bioenergetic capacity as previously described 28 . Briefly, bioenergetic capacity is the product of the maximal ATP from glycolysis (J ATP glyc) and ATP from oxidative phosphorylation (J ATP ox). Values were normalized to cell counts. cell proliferation. Cell counts were quantified using an automated TC10 cell counter (Bio-Rad) and viability was determined using trypan blue exclusion.

Metabolomics. Nucleotide detection and analysis was performed using LC-MS/MS at the Metabolomics
Core Facility of the Goodman Cancer Research Centre. Cultured cells were treated with methotrexate for 72 hours. Cells were washed in ammonium formate three times, then quenched in cold 50% methanol (v/v) and acetonitrile. Cells were lysed following bead beating at 30 Hz for 2 minutes. Cellular extracts were partitioned into aqueous and organic layers following dichloromethane treatment and centrifugation. Aqueous supernatants were dried down using a refrigerated speed-vac. Dried samples were subsequently resuspended in 25 μl of water. A 10 μl volume of sample was injected onto an Agilent 6430 Triple Quadrupole (QQQ)-LC-MS/MS for targeted metabolite analysis of AICAR (5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside) and a 10-fold dilution was injected for nucleotides: ZMP (5-aminoimidazole-4-carboxamide ribonucleotide), IMP (inosine monophosphate), and AMP (adenosine monophosphate). The liquid chromatography separation was performed using a 1290 Infinity ultra-performance binary LC system (Agilent Technologies, Santa Clara, CA, USA). The chromatography run was conducted as follows: column temperature was maintained at 10 °C and the separation was realised by reverse phase separation using a flow rate of 0.4 mL/min with a Scherzo SM-C18 column 3 μm, 3.0 × 150 mm (Imtakt Corp, JAPAN) The gradient started at 100% mobile phase A (5 mM ammonium acetate in water) with a 5 min gradient to 100% B (200 mM ammonium acetate in 80% water/20% ACN) at a flow rate of 0.4 ml/min. This was followed by a 5 min hold time at 100% mobile phase B and a subsequent re-equilibration time (6 min) before next injection.
The mass spectrometer was equipped with an electrospray ionization (ESI) source and samples were analyzed in positive ionization mode. Multiple reaction monitoring (MRM) transitions were optimized on standards for each metabolite quantitated. Transitions for quantifier and qualifier ions were as follows: AICAR (259. . Gas temperature and flow were set at 350 °C and 10 l/min respectively, nebulizer pressure was set at 40 psi and capillary voltage was set at 3500 V. Relative concentrations were determined from external calibration curves prepared in water and compared to sample area under the curve. Note that no corrections were made for ion suppression or enhancement. Data were analyzed using MassHunter Quant (Agilent Technologies, Santa Clara, CA, USA). flow cytometry analysis of cell cycle. WT and AMPK-null MEF cells were seeded in 12-well plates and grown for 24 h, after which they were treated with control, MTX, phenformin or both for 72 h. Cells were counted and 100,000 cells were washed with cold wash buffer (PBS + 5% FBS + 0.01 M NaN 3 ), spun down, and resuspended in hypotonic buffer (0.1% sodium citrate, 0.1% Triton X-100 in water). Cells were stained with Propidium iodide (50 μg/ml) (cat #: P4170, Sigma) for 30 min at 37 °C in the dark. Samples were analyzed with a FACS Canto II (San Jose, CA). Fluorescence was detected by excitation at 488 nm and acquisition on the 585/42 PI-A channel.

Data availability
The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request.