Mitochondria preserve an autarkic one-carbon cycle to confer growth-independent cancer cell migration and metastasis

Kiweler, Nicole; Delbrouck, Catherine; Pozdeev, Vitaly I.; Neises, Laura; Soriano-Baguet, Leticia; Eiden, Kim; Xian, Feng; Benzarti, Mohaned; Haase, Lara; Koncina, Eric; Schmoetten, Maryse; Jaeger, Christian; Noman, Muhammad Zaeem; Vazquez, Alexei; Janji, Bassam; Dittmar, Gunnar; Brenner, Dirk; Letellier, Elisabeth; Meiser, Johannes

doi:10.1038/s41467-022-30363-y

Download PDF

Article
Open access
Published: 16 May 2022

Mitochondria preserve an autarkic one-carbon cycle to confer growth-independent cancer cell migration and metastasis

Nature Communications volume 13, Article number: 2699 (2022) Cite this article

9431 Accesses
30 Altmetric
Metrics details

Subjects

Abstract

Metastasis is the most common cause of death in cancer patients. Canonical drugs target mainly the proliferative capacity of cancer cells, which leaves slow-proliferating, persistent cancer cells unaffected. Metabolic determinants that contribute to growth-independent functions are still poorly understood. Here we show that antifolate treatment results in an uncoupled and autarkic mitochondrial one-carbon (1C) metabolism during cytosolic 1C metabolism impairment. Interestingly, antifolate dependent growth-arrest does not correlate with decreased migration capacity. Therefore, using methotrexate as a tool compound allows us to disentangle proliferation and migration to profile the metabolic phenotype of migrating cells. We observe that increased serine de novo synthesis (SSP) supports mitochondrial serine catabolism and inhibition of SSP using the competitive PHGDH-inhibitor BI-4916 reduces cancer cell migration. Furthermore, we show that sole inhibition of mitochondrial serine catabolism does not affect primary breast tumor growth but strongly inhibits pulmonary metastasis. We conclude that mitochondrial 1C metabolism, despite being dispensable for proliferative capacities, confers an advantage to cancer cells by supporting their motility potential.

Mitochondrial RNA modifications shape metabolic plasticity in metastasis

Article Open access 29 June 2022

Glioblastoma cell motility depends on enhanced oxidative stress coupled with mobilization of a sulfurtransferase

Article Open access 30 October 2022

Bedaquiline, an FDA-approved drug, inhibits mitochondrial ATP production and metastasis in vivo, by targeting the gamma subunit (ATP5F1C) of the ATP synthase

Article Open access 13 May 2021

Introduction

The survival rates of patients harboring primary tumors steadily increase due to targeted treatment schemes. However, in case of tumor relapse and metastatic disease, effective therapies are mostly lacking with the consequence that secondary tumors account for the majority of cancer deaths¹. Classical chemotherapeutic approaches to counteract cancer growth aim to target biomass production by interfering with the synthesis of proteins, lipids, and nucleotides and thereby directly interfere with central pathways of cancer cell metabolism^2,3. Prospectively, therapeutic approaches targeting metabolic pathways that also support the invasive and migratory properties of cancer cells would help to prevent metastatic progression of the initial disease. However, profound knowledge of specific metabolic dependencies that support growth-independent processes during the metastatic cascade, are only starting to be understood⁴.

Nucleotide synthesis is an essential requirement for cancer cell proliferation. Consequently, antifolates such as Methotrexate (MTX) and related compounds have been proven as successful chemotherapeutics for several decades. Although differences between the various antifolates exist, a shared mode of action is the inhibition of one-carbon (1C) metabolism which provides essential building blocks for thymidylate and purine synthesis. Additionally, 1C metabolism has implications in supporting methylation reactions, glutathione, heme, sphingolipid and protein synthesis⁵. 1C metabolism describes a metabolic cycle that is spread across mitochondrion and cytoplasm and it is mainly fueled by the non-essential amino acid serine. De-novo serine synthesis is catalyzed by the rate limiting enzyme phosphoglycerate dehydrogenase (PHGDH) which is amplified in different tumors^6,7,8. Hence, targeting serine synthesis, serine catabolism or serine availability by starvation or inhibition of PHDGH has been proven successful to inhibit cancer progression^{9,10,11,12,13,14,15,16,17}. To support the various anabolic programs, cytosolic 1C metabolism is of particular importance, while mitochondrial 1C metabolism is dispensable due to the reversibility of the cytosolic 1C metabolism¹⁶. In fact, depending on intracellular folate levels, cytosolic 1C metabolism can be the dominant route to provide one-carbon units for nucleotide synthesis¹⁷. In that light, reasons for conserved mitochondrial 1C metabolism are still not fully understood. Previous evidence suggests implications of mitochondrial 1C metabolism for mitochondrial translation^18,19. One additional explanation resides in the observation that mitochondrial derived formate lowers the pool of free cytosolic folates by increasing the 10-CHO-THF pool thereby protecting it from degradation. Upon MTX, degradation is facilitated as folates accumulate in the oxidized DHF form, impeding the 10-CHO-THF synthetase activity of MTHFD1²⁰. However, if mitochondrial folate-dependent 1C metabolism and formate production remains functional under conditions of cytosolic 1C metabolism impairment and whether mitochondrial 1C metabolism itself confers additional functions that contribute to tumorigenesis remains unknown.

We have previously shown that a reduction of biomass demand can result in increased formate release from cancer cells to their surrounding microenvironment²¹ and that the rate of serine catabolism to formate is significantly increased in adenomas of the small intestine and breast tumors compared to normal tissue²². Importantly, enhanced formate release rates and increased extracellular formate concentrations have been shown to increase the invasiveness of glioblastoma cells in vitro²².

Apart from our previous findings on formate overflow and its potential implications for cancer cell invasion^21,22, other reports substantiate a role of serine metabolism in the context of metastasis^{7,14,23,24,25}. However, such studies were so far focused on the relevance of serine de novo synthesis via PHGDH or extracellular serine availability for metastatic outgrowth (proliferative aspect) in the secondary tissue^{7,14,23,24,25}. In contrast, whether mitochondrial serine catabolism also has a growth-independent role in promoting cancer cell motility that potentially increases the rate of cancer cell escape from the primary tumor has not been addressed so far. Being able to identify and target metabolic programs that promote the early steps of the metastatic cascade would help to contain the tumor, a clinical phenotype with more favorable therapeutic outcome compared to advanced metastatic stages.

In the present study, we observed that antifolates, albeit being strongly cytostatic, did not decrease the cell’s motility potential. Using MTX as a tool compound to selectively inhibit cytosolic 1C metabolism, we show that mitochondrial 1C metabolism runs as an autarkic, compartmentalized pathway that is important to drive cancer cell motility. In contrast, the cytosolic part of 1C metabolism, which is essential for nucleotide synthesis during proliferation, is dispensable for the cellular motility potential. Hence, we provide mechanistic evidence that one selective advantage of mitochondrial 1C metabolism is to support cell motility. We corroborate this hypothesis in an orthotopic breast cancer model. Here, we demonstrate that ablating mitochondrial 1C metabolism effectively blocks mitochondrial serine catabolism but does not affect primary tumor growth. Yet, it strongly inhibits pulmonary metastasis formation.

Results

Inhibition of anabolic synthesis routes differentially impacts the motility potential of cancer cells

To assess the relative importance of different anabolic pathways for cancer cell migration, we employed a panel of metabolic perturbations targeting glycolysis, mitochondrial electron transport chain (ETC) as well as lipid, protein, and nucleotide synthesis (Fig. 1a). As expected, the chosen metabolic interventions inhibited proliferation and affected cell cycle progression of MDA-MB-468 breast cancer cells with no to mild toxic effects after 48 h treatment (Fig. 1b–c, S1A). Inhibition of glycolysis, ETC, lipid, and protein synthesis significantly reduced migration of MDA-MB-468 cells (Fig. 1d). Surprisingly and in contrast to all other growth-inhibiting conditions, inhibition of nucleotide synthesis with various drugs did not diminish migration of MDA-MB-468 cells (Fig. 1e). Calculation of the area under curve (AUC) allowed us to quantitatively compare wound closure over time (Fig. 1d, e) and confirmed that wound closure was significantly reduced upon treatment with rotenone (Rot), galactose (Gal), simvastatin (SIM) and sirolimus (rapamycin, SIR). However, treatment with the antifolates methotrexate (MTX) and pemetrexed (PEM), as well as with hydroxyurea (HU) and clofarabine (CLO) did not affect or did even significantly increase wound closure in scratch assays in comparison to fully proliferative, untreated control cells. While cell migration correlates with reduced cell growth for all other metabolic perturbations, nucleotide synthesis inhibition does not reduce cell migration in correlation to growth inhibition (Fig. 1f, g). Sustained migration in response to nucleotide synthesis inhibition was verified in LN229 glioblastoma and 4T1 breast cancer cells (Fig. S1B–D). Of note, using dialyzed or normal FBS did not change the result (Fig. S1E). This observation was also confirmed in trans-well migration assays using Boyden chambers (Fig. 1h, S1F, G). Furthermore, using ECM-collagen coating in trans-well assay, we found that nucleotide synthesis inhibition also had no impact on the invasive capacity of MDA-MB-468 cells (Fig. 1h), LN229 and 4T1 cells (Fig. S1F, G). To validate if MTX resistance results in a pro-migratory effect, we generated MTX-resistant MDA-MB-468 cells by long-term cultivation in 50 nM MTX for 2 months. The resulting MTX-resistant MDA-MB-468 cells proliferate in the presence of increasing concentrations of MTX (Fig. 1i) and display enhanced migratory capacity compared to the parental MDA-MB-468 cell line (Fig. 1j).

**Fig. 1: Inhibition of major anabolic synthesis routes differentially impacts the motility potential of cancer cells.**

In conclusion, the perturbation of any major anabolic route results in growth repression. However, the impact on cell migration highly depends on the targeted metabolic pathway, with nucleotide synthesis being an ineffective target to abrogate cell motility. Furthermore, prolonged nucleotide synthesis inhibition in drug-resistant cells acts as a selection pressure for more motile cancer subpopulations.

Sustained motility upon MTX is supported by a ROS-driven EMT phenotype

As MTX induced the strongest growth repression without increasing cell death (Fig. 1b, c), we used it as a tool compound to disentangle proliferation from migration to investigate the metabolic processes that support cancer cell migration. After testing multiple concentrations (Fig. S2A), we chose a concentration of 50 nM MTX for subsequent experiments. 50 nM MTX is significantly growth- and cell cycle-arresting in MDA-MB-468 cells (Fig. 1b, S1A) and has also been shown to be within the therapeutic window^26,27,28,29. As this concentration of 50 nM is significantly lower compared to earlier in vitro studies^{20,30,31,32,33}, we first set out to validate known drug effects of MTX in our system. We observed that 48 h MTX treatment of MDA-MB-468 cells resulted in altered cellular morphology with individual cells visibly increasing in cell size and exhibiting an outstretched cell shape with protruding edges (Fig. 2a). MTX-resistant MDA-MB-468 cells showed comparable morphologic alterations (Fig. S2B). While the observed cell cycle and proliferation arrest upon MTX treatment was rapidly reversible (Fig. S2C, D) upon MTX removal, [U-¹³C]serine tracing and metabolic flux analysis confirmed that 50 nM MTX did strongly inhibit nucleotide synthesis (Fig. 2b). Of note, nucleotide synthesis was partially restored in MTX-resistant MDA-MB-468 cells (Fig. S2E). Stable isotope labeling by amino acids in cell culture (SILAC) and quantitative proteomics analysis further confirmed the efficacy of the selected MTX dose as mechanistic targets of MTX such as dihydrofolate reductase (DHFR), thymidylate synthase (TYMS), and thymidine kinase (TK1) were upregulated by treatment as expected (Fig. S2F). DHFR expression was also increased in MTX resistant cells, while FolRa expression was slightly decreased (Fig. S2G). Moreover, superoxide dismutase (SOD) was one of the most significantly upregulated proteins in response to treatment (Fig. S2F), which corresponds to prior reports that depict MTX as a potent inducer of oxidative stress^30,34,35. We validated these prior findings with our low dose of MTX by measuring a moderate but significant increase of mitochondrial and cytosolic reactive oxidant species (ROS) upon acute MTX treatment and in the MTX resistant cells (Fig. S2H, I, J). In line with increased ROS levels we observed NRF2 stabilization, however, this did not result in activation of the BACH1 axis, as reported earlier for metastatic lung cancer^36,37 (Fig. S2K, I). Increased ROS levels upon MTX correspond to earlier reports that highlight the potential of oxidative stress as a driver of cancer cell transformation via epithelial-mesenchymal transition (EMT)^{38,39,40,41,42,43,44}. EMT is also a known general mechanism to escape chemotherapy induced effects.

**Fig. 2: Enhanced motility under MTX is supported by a ROS-driven EMT phenotype.**

In line with previous reports, we found that ZEB1 and ZEB2 expression as well as VIM expression was upregulated in a time-dependent manner in response to acute MTX treatment and in the MTX-resistant cell line (Fig. 2c, d, S2F, M, N). MTX-treatment also induced the expression of the collagenase MMP9 (Fig. S2O), an enzyme which is involved in the degradation of the extracellular matrix during cancer cell invasion⁴⁵. Treatment of cells with the antioxidant N-acetylcysteine (NAC) reduced ROS levels in MTX-treated MDA-MB-468 cells (Fig. S2P) and resulted in a significant reduction of MTX-induced cell migration (Fig. 2e) and EMT phenotype induction (Fig. 2f). These findings confirm that the chosen low dose of MTX is sufficient to promote EMT in a ROS dependent manner, supporting earlier reports suggesting EMT onset in context of pulmonary fibrosis using high doses of MTX^46,47. Such moderate increase in cellular ROS levels after MTX might contribute to a pro-migratory stimulus that subsequently results in the observed sustained migration.

MTX-treated cells sustain high metabolic rates and enhance de novo serine synthesis

In addition to the here reported MTX-specific effects on ROS and EMT, we wanted to take advantage of the MTX effect (sustained migration) to identify growth-independent metabolic liabilities that support cell migration.

To that end, we profiled central carbon metabolism using [U-¹³C]glutamine and [U-¹³C]glucose tracing to monitor glycolytic activity and glutamine- or glucose-derived carbon oxidation through the TCA cycle (Fig. 3a, d). Absolute consumption and release (CORE) rates of glutamine and glutamate were sustained in response to MTX treatment (Fig. 3b), while the relative TCA-cycle flux of [U-¹³C]glutamine was significantly increased in response to MTX (Fig. 3c). Absolute quantification of CORE rates of glucose and lactate also revealed sustained high glycolytic rates upon MTX treatment (Fig. 3e). This came as a surprise, as lactate release rates were previously shown to correlate with cell growth rates⁴⁸. In MDA-MB-468 cells, the maintained glycolytic rate and the associated lactate release at constant high levels (Fig. 3e) indicate a constant generation of glycolysis-derived ATP even in the presence of reduced energetic demand for anabolic reactions in growth-arresting MTX conditions. Consequently, [U-¹³C]glucose distribution within the TCA-cycle was sustained by MTX treatment in MDA-MB-468 cells (Fig. 3f) as well as in 4T1 and LN229 cells (Fig. S3A, B). It has previously been shown that MTX inhibits oxygen consumption rates (OCR) in HCT116 cells¹⁸. We could replicate this finding and observed a ~30 % reduction of OCR in HCT116 cells upon MTX (Fig, S3C), while OCR in MDA-MB-468, LN229, and 4T1 cells was sustained (Fig. 3g). This indicates cell line specific effects of MTX on OCR. Furthermore, MTX treatment significantly increased the relative flux of glucose to serine despite decreased anabolic demands for nucleotide synthesis (Fig. 3h, S3D). Protein levels of PHGDH, PSPH, and PSAT1 that catalyze the serine de novo synthesis pathway from glucose were unaltered in response to MTX treatment in multiple cell lines (Fig. S3E–H). In agreement with Diehl et al⁴⁹. who showed that changes in NAD⁺/NADH ratio can increase PHGDH activity and in consequence serine de novo synthesis rates, we observed an increased NAD⁺/NADH ratio upon MTX (Fig. 3i). As MTX is well characterized as an inhibitor of DHFR and thus 1C metabolism, it came to our surprise that MTX treatment did also increase the abundance of extracellular formate M + 1 isotopologues derived from [U-¹³C]glucose through serine and the mitochondrial 1C metabolism (Fig. 3j, S3I).

**Fig. 3: MTX-treated cells sustain high metabolic rates and enhance de novo serine synthesis.**

In summary, the metabolic profiling demonstrates that, despite decreased metabolic demand for biomass production, MTX-treated cells sustain high metabolic rates that are comparable to fully proliferating cells. Additionally, and in contrast to general assumptions, cytosolic DHFR inhibition results in increased rates of serine synthesis and downstream mitochondrial formate excretion suggesting an uncoupling of mitochondrial 1C metabolism from cytosolic 1C metabolism.

Mitochondria protect 1C metabolism-dependent serine catabolism upon cytosolic 1C pathway inhibition

Our finding of increased [U-¹³C]glucose-derived M + 1 formate excretion under MTX (Fig. 3j, S3I), allows the hypothesis that mitochondrial 1C metabolism can function in an autarkic manner when cytosolic 1C metabolism is inactivated (Fig. 4a). To corroborate this hypothesis, we employed [U-¹³C]serine tracing to monitor serine flux through 1C metabolism. We observed that cytosolic block of 1C metabolism via MTX did not reduce the exchange rates of serine, glycine and formate in MDA-MB-468 or LN229 cells (Fig. 4b, C, S4A, C). In contrast, the formate release rate was significantly increased and the increased fraction of labeled formate proved that most of the released formate is derived from [U-¹³C]serine and not from other carbon sources that could alternatively generate formate in presence of MTX. Of note, sustained serine consumption and formate overflow via 1C metabolism upon DHFR inhibition could also be confirmed in Plasmax medium⁵⁰, which is a culture medium closer to human physiology (Fig. S4D). Since cell dry mass composition is constituted by around 60% of proteins⁵¹, growth arresting conditions generally result in decreased consumption rates of proteinogenic amino acids. In fact, a general trend for decreased consumption of essential amino acids was observed (Fig. S4b). Hence, sustained serine consumption rates upon growth arrest suggest that spared serine that is otherwise used for anabolic processes such as nucleotide synthesis, is used for alternative metabolic pathways that support cell motility and survival.

Mechanistically, 1C metabolism follows a cycle in which serine can be resynthesized in the cytoplasm and mitochondrion from glycine via serine hydroxymethyltransferase 1 and 2 (SHMT1, SHMT2) (Fig. 4a). As the SHMT reaction is highly reversible, the serine labeling pattern from [U-¹³C]serine is composed of a mix of M + 1, M + 2 and M + 3 isotopologues, which represents the different recombination events with labeled and unlabeled glycine and formate^16,21,52. Consequently, M + 1 and M + 2 serine isotopologues are expected to be absent upon complete 1C metabolism inhibition. Intriguingly, cytosolic DHFR inhibition with MTX neither completely abrogated serine M + 1 and M + 2 isotopologues upon [U-¹³C]serine nor did it significantly affect the serine to glycine level ratio in MDA-MB-468 or LN229 cells (Fig. 4d, e, S4E, F), further supporting our hypothesis of a persistent mitochondrial 1C metabolism upon DHFR inhibition. To mechanistically disentangle cytosolic from mitochondrial 1C metabolism, we analyzed a panel of CRISPR knockouts in HAP1 cells in which either the cytosolic (SHMT1), the mitochondrial (MFT, SHMT2), or both compartments (FPGS, MFT + SHMT1) of 1C metabolism were abrogated⁵³. Here, cytosolic SHMT1 KO did not eliminate intermediary M + 1 and M + 2 serine isotopologues from [U-¹³C]serine comparable to MTX treatment, whereas mitochondrial or combined cytosolic and mitochondrial inhibition of 1C metabolism upon MFT KO, SHMT2 KO, FPGS KO and MFT + SHMT1 KO resulted in a complete loss of M + 1 and M + 2 serine isotopologues (Fig. 4f). In mutants harboring a full inhibition of 1C metabolism or mitochondrial 1C metabolism alone, we additionally observed an increase in the ratio of serine to glycine levels, whereas MTX treatment and SHMT1 KO did not substantially increase serine to glycine ratio (Fig. 4g).

Of note the protein abundance of mitochondrial serine catabolism enzymes was not increased upon MTX. MTHFD2 abundance even decreased upon MTX, however, this did apparently not affect the rate of serine catabolism (Fig. S4G, H).

To test if knock-out of mitochondrial serine catabolism stops formate overflow, we monitored formate release flux upon MTX in HAP1 cells deficient for SHMT1 or SHMT2. As expected, loss of SHMT2 completely abolished formate overflow, while SHMT1 KO increased formate overflow (Fig. 4h). This experiment was then validated in MDA-MB-468 cells harboring MTHFD1L KO²⁰ or shRNA mediated MTHFD2 knock-down and MTHFD2 KO (Fig. 4i, j, S4I-M). All genetic interventions resulted in a strong suppression of formate overflow during MTX treatment, further suggesting that increased formate overflow upon MTX depends on mitochondrial serine catabolism via the folate pathway (Fig. 4i, j, S4J, M).

As an alternative approach to genetic silencing we used Metformin to target complex 1 of the electron transport chain (ETC). We have previously shown that mitochondrial formate production depends on active ETC^21,22. Here, we demonstrate that targeting of complex 1 with Metformin strongly suppresses mitochondrial formate production both at baseline and in presence of MTX, further supporting our finding that formate production in presence of MTX depends on mitochondrial 1C metabolism (Fig. 4k). To proof that Metformin indeed inhibits respiration under the applied conditions, we performed Seahorse experiments to measure the expected drop in oxygen consumption rate (Fig. S4N). We also monitored a reduction in proliferation (Fig. S4O) and we performed [U-¹³C]glutamine tracing to monitor the expected strong increase in reductive carboxylation (Fig. S4P, Q), a well-known phenomenon upon ETC inhibition^54,55.

In summary, these results show that mitochondria provide a protected cellular environment that permits autarkic folate-mediated serine catabolism independent of the cytosolic folate metabolism.

Limiting serine availability decreases formate overflow and cancer cell migration

As we observed increased glucose derived serine synthesis and subsequent downstream catabolism of serine via mitochondrial 1C metabolism, we investigated if interfering with serine synthesis is effective to reduce sustained cell migration upon pharmacologic inhibition of cytosolic 1C metabolism. First, to identify an effective inhibitor of serine de novo synthesis, we tested multiple available allosteric and one competitive inhibitor of PHGDH. Indicated by the strong abrogation of labeled serine, we observed that the competitive inhibitor BI-4916 (BI)¹² had superior efficacy compared to the allosteric inhibitors WQ-2101 (WQ)¹¹, NCT-502 (NCT)⁹, and CBR-5884 (CBR)¹⁵ (Fig. 5a). Released M + 1 formate from [U-¹³C]glucose was also effectively inhibited after BI treatment alone or in combination with MTX (Fig. 5b). Additionally, routine screening for metabolic side effects revealed that treatment with the allosteric inhibitors WQ, NCT, and CBR did result in a significant reduction of mitochondrial OCR (Fig. 5c), which was not observed with the competitive inhibitor BI (Fig. 5c). WQ and NCT also negatively affected TCA cycle activity as characterized by a reduction of M + 2 isotopologue abundance of TCA cycle associated metabolites (Fig. 5d). In contrast to the other inhibitors, BI had no adverse effect on TCA cycle activity and proliferation rate and thus emerged as the preferred PHGDH inhibitor in all subsequent experiments (Fig. 5d, S5A–D). Combined treatment with the specific PHGDH inhibitor BI significantly reduced MTX-mediated cell migration (Fig. 5e). This inhibitory effect was further enhanced upon combined PHGDH inhibition and serine and glycine (S/G) starvation (Fig. 5f). While S/G starvation alone resulted only in moderate inhibitory effects on migration of MTX-treated cells, combined treatment with PHGDH inhibition fully blunted cell migration (Fig. 5f). Similar results were observed in 4T1 cells (Fig. S5F). As all MTX conditions resulted in full growth arrest (Fig. S5E), this finding indicates a role for mitochondrial 1C metabolism in cell motility that is independent of serine’s anabolic function to support biosynthetic processes.

While neither BI treatment alone nor S/G starvation alone were sufficient to prevent the induction of the observed EMT-phenotype, combined inhibition and starvation did minimize ZEB1, ZEB2, and VIM upregulation upon DHFR inhibition (Fig. 5g). This indicates that serine represents an underlying requirement for EMT onset upon DHFR inhibition. Importantly, S/G starvation did not further increase cell death after combined MTX and BI treatment (Fig. 5h). Of note, we also found that BI treatment was effective to reduce the migratory capacity of MTX-resistant, pro-migratory MDA-MB-468 cells (Fig. 5i). In summary, using MTX as a tool compound to study cell migration in absence of proliferation revealed that serine is essential to sustain full cancer cell migratory capacity (Fig. 5j).

Mitochondrial 1C metabolism supports cell motility and depends on mitochondrial THF availability

As withdrawal of serine can affect different aspects of cell physiology, we aimed to investigate more specifically the role of mitochondrial 1C metabolism in promoting cell migration. Analogue to the interventions presented in Fig. 4i–k, we used Metformin to target the ETC and genetic interventions to target MTHFD2 or MTHFD1L both in MDA-MB-468 and 4T1 cells.

Metformin treatment of MTX exposed (growth arrested) cells reduced the migration of MDA-MB-468 cells (Fig. 6A), suggesting that one function of the ETC is to support cell migration.

**Fig. 6: Mitochondrial 1C metabolism supports cell motility and depends on mitochondrial THF availability.**

Next, we investigated the role of mitochondrial serine catabolism in context of migration directly, by monitoring the migration of MTX treated MTHFD2 or MTHFD1L silenced MDA-MB-468 and 4T1 cells. Of note, neither MTHFD2 nor MTHFD1L silencing reduced cell proliferation or sensitivity to the growth-arresting effects of MTX (Fig. S6A–D). Also, cell death rates upon MTX were not affected by MTHFD2 or MTHFD1L silencing (Fig. S6E–H). Scratch assay analysis showed that MTHFD2 silencing markedly reduced MTX-mediated cell motility both in MDA-MB-468 and 4T1 cells (Fig. 6b, c). As shRNA used to silence human and murine MTHFD2 were different, we do not expect shRNA specific off-target effects. Nevertheless, we confirmed the shMTHFD2 KD results in MDA-MB-468 cells with MTHFD2 knock-out cells (Fig. S6I).

Mthfd1l silencing of MTX treated 4T1 cells also resulted in decreased migration (Fig. 6d). Somewhat surprisingly, loss of MTHFD1L in MDA-MB-468 cells had no negative effect on the migration of MTX treated cells (Fig. 6e). By following different hypotheses and various experiments, we realized that there is a striking difference in ALDH1L2 expression in 4T1 versus MDA-MB-468 cells, while MTHFD1L expression was comparable between these two cell lines (Fig. 6f, g).

This observation reminded us that in MDA-MB-468 cells, ALDH1L2 can act as an alternative outlet to oxidize formyl-THF to CO₂ thereby regenerating mitochondrial THF (in contrast to 4T1). That said, in MTHFD1L deficient cells, the ALDH1L2 status can be decisive whether THF can be regenerated or if THF will be trapped as 10-CHO-THF in the mitochondrion. Loss of free THF will potentially inhibit the autarkic mitochondrial cycle that can otherwise act in presence of MTX (Fig. 6h). In support of this hypothesis, we also measured increased formate release flux in 4T1 cells compared to MDA-MB468 (Fig. S6J), suggesting that in 4T1 cells less 10-CHO-THF is oxidized via Aldh1l2 and instead catabolized through Mthfd1l.

To test this hypothesis, we targeted ALDH1L2 in MTHFD1L KO cells and monitored migration. We found that combinatorial depletion of MTHFD1L and ALDH1L2 in MDA-MB-468 cells resulted at least in a reduction of migration upon MTX compared to single depletion of MTHFD1L alone. This was observed both for shRNA as well as for siRNA mediated knockdown of ALDH1L2 (Fig. 6i, S6K).

In conclusion, these results reveal a growth-independent role of mitochondrial 1C metabolism for cancer cell motility. Interestingly, the pool of free THF might be relevant for the functionality of the autarkic mitochondrial 1C cycle and ALDH1L2 can serve as an alternative outlet to regenerate THF when MTHFD1L is missing (summarized in Fig. 6J).

Genetic targeting of mitochondrial 1C metabolism reduces metastasis formation in vivo

According to our model, we predict that sole inhibition of mitochondrial 1C metabolism might be sufficient and effective to reduce cancer cell escape from the primary tumor and subsequent metastasis formation. To test this hypothesis, we employed the 4T1 cells with Mthfd1l KD to study the metastatic cascade in orthotopic tumors in vivo.

To evaluate first if Mthfd1l KD cells are more prone to oxidative stress, which could impact metastatic dissemination in vivo⁵⁶, we measured ROS levels in MTHFD1L silenced MDA-MB-468 and 4T1 cells at baseline and after MTX treatment. At baseline, ROS levels between MTHDF1L silenced and control cells were unchanged. MTX treatment induced ROS levels to comparable extends and 500 µM of H₂O₂ treatment did induce cell death rate comparable to Ctrl cells (Fig. S7A–C). In summary, these experiments do not indicate that loss of MTHFD1L results in increased cellular ROS levels or increased susceptibility to oxidative insults.

We then injected 4T1 cells with or without silenced Mthfd1l into the mammary fat pad of immunocompetent BALB/c mice to form orthotopic tumors (Fig. 7a). Primary tumor growth was monitored over 6 weeks. In concordance with the in vitro data, primary tumor growth was not affected and tumor weight at endpoint was not different compared to Ctrl conditions (Fig. 7b, c). Knock-down of Mthfd1l in the primary tumors was confirmed by Western Blot (Fig. 7d, S7D–F). In line with our hypothesis, the number of macroscopic lung metastases was significantly reduced in mice that were carrying tumors deficient in Mthfd1l (Fig. 7e). While all mice that were injected with Mthfd1l-expressing 4T1 cells exhibited lung metastases, over 50% of mice injected with Mthfd1l KD cells did not show macroscopic metastatic lesions. Of note, expression of EMT related genes in the primary tumor was unchanged between the two groups (Fig. S7G). H&E staining of lungs and quantification of the metastatic area per lung confirmed reduced metastasis in Mthfd1l deficient tumors (Fig. 7f, g).

**Fig. 7: Genetic targeting of mitochondrial 1C metabolism reduces metastasis formation in vivo.**

These in vivo findings further strengthen the key relevance of mitochondrial 1C metabolism as a cornerstone to support cancer cell migration independent of primary tumor growth and cancer cell proliferation rates.

Discussion

In this study, we uncover a growth-independent function of mitochondrial serine catabolism to drive cancer cell motility. Importantly, by using MTX as a tool compound to study migration independent of proliferation, we observed that upon inhibition of cytosolic 1C metabolism, mitochondria sustain an autarkic 1C metabolism that is sufficient to sustain full migratory capacity.

Such compartmentalization of 1C metabolism during MTX treatment is feasible because release of the 1C unit from THF in the mitochondrial MTHFD1L reaction is, in contrast to cytosolic thymidylate synthetase reaction, not oxidizing THF to DHF. Therefore, the mitochondrial THF pool is sustained to allow SHMT2-dependent mitochondrial serine catabolism, while cytosolic 1C metabolism enzymes are inhibited due to accumulating DHF species. Our finding that ALDH1L2 can serve as an alternative source for THF regeneration in MTHFD1L deficient cells, supports the notion that mitochondria can maintain an independent THF pool that is required for an autarkic mitochondrial 1C metabolism as long as there is THF available that can accept the hydroxymethyl group in the first (SHMT2 dependent) step of serine catabolism.

An autarkic mitochondrial 1C metabolism is further supported by a chemical modification of folate species that alters their transport activities across the mitochondrial membrane. For maximal biologic activity, folate species need to be polyglutamated by folylpolyglutamate synthase (FPGS) and the resulting polyglutamate species were shown to be only poorly transported across the mitochondrial membrane^57,58. Consequently, polyglutamated folate species can be chemically trapped within the mitochondria to sustain the mitochondrial 1C cycle. Especially, upon growth arrest when cells can retain their mitochondrial content and don’t need to amplify their content as it is required during proliferation. Additionally, there is evidence that transport activity of the mitochondrial folate transporter is limited to reduced folates⁵⁹. This indicates a limited transport of MTX and folic acid itself across the mitochondrial membrane. Taken together, such compartmentalization of 1C metabolism and autarkic function of mitochondrial 1C metabolism emerges as a selective advantage in cells upon perturbation of cytosolic 1C metabolism and indicates a function of the mitochondrial 1C metabolism pathway to support cell motility.

While our data demonstrate that mitochondrial serine catabolism is needed to allow full migratory capacity, the specific biochemical output conferring this function still needs to be determined in more detail. Mitochondrial serine catabolism can generate NADH, NADPH, ATP, glycine, CO₂, methylene- and formyl-THF and formate. The plasticity of this pathway allows tailoring the catabolism towards current cellular needs³⁸. Therefore, it can be possible that serine is differentially catabolized along the metastatic cascade in a context dependent manner. We have previously shown that exogenous formate promotes cancer cell invasion²² and various cellular stress situations such as energy stress²¹ or MTX treatment (this study) can increase the rate of formate overflow which results in increased extracellular formate concentrations. It is therefore tempting to speculate that increasing formate concentrations in the tumor microenvironment promote a more invasive phenotype of stressed cancer cells. In more general terms, it could be possible that mitochondrial serine catabolism can selectively promote cell motility in response to extrinsic or intrinsic stress stimuli such as growth inhibition or nutrient deprivation. On the other hand, MDA-MB-468 MTHFD1L KO cells are also formate overflow negative but do not show reduced migration. As ROS levels are neither increased in MDA-MB-468 nor in 4T1 MTHFD1L deficient cells, we do not think that the observed migration effects in this study are ROS dependent. Given previous work²² and the current study, there might also be differences when comparing migration with invasion. Ongoing work in our lab will help clarifying these questions in the future. To do so, it will be of particular importance to further disentangle effects arising from exogenous formate from effects depending on endogenous serine catabolism. Independent of these yet unknown details, we remain with the conclusion that mitochondrial serine catabolism contributes to cancer cell migration and metastasis independent of its known anabolic functions.

Finally, we would like to note that although we mainly use MTX as a tool compound to separate proliferative from migratory processes and to inhibit cytosolic 1C metabolism, our results have also some implications in regard to MTX as an anchor drug in chronic treatment of autoimmune diseases. Intracellular erythrocyte and mean plasma concentration of MTX were reported to be in a comparable, even slightly higher, nM range to our chosen drug concentration^60,61. Finally, we also would like to refer to a parallel and independent (back-to-back) publication that provides additional evidence that purine depletion promotes serine de novo synthesis and cell migration⁶². Interestingly, in this parallel work, the authors provide data that MTX also promotes the metastatic potential of melanoma cells in vivo. Hence, our findings on mitochondrial 1C metabolism might justify future investigations in cancer patients and patients that undergo chronic MTX therapy for an arthritic comorbidity.

Methods

Chemicals

Methotrexate, rotenone, PKUMDL-WQ-2101, galactose, hydroxyurea, NAC, metformin and fluorouracil were purchased from Sigma Aldrich. Clofarabine, pemetrexed disodium hydrate, sirolimus were purchased from Bio Connect. Simvastatin was purchased from Sanbio. CBR-5884 and BI-4916 were purchased from MedChemExpress. NCT-502 was purchased from ApeXBio.

Cell culture

All cell lines (except HAP1 cells) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) without phenol red, glucose, and glutamine (Thermo Fisher Scientific) and supplemented with 2 mM glutamine, 17 mM glucose, and 10% fetal bovine serum (FBS) at 37 °C and 5% CO₂. HAP1 cells were cultured in IMDM medium supplemented with 10% FBS. For galactose treatment, supplemented glucose was replaced by 17 mM galactose at the beginning of the experiment. For serine/glycine starvation experiments cells were kept in MEM (Thermo Fisher Scientific) supplemented with or without 400 μM glycine and serine, 2 mM glutamine, 17 mM glucose and 10% dialyzed FBS.

MDA-MB-468 and MTHFD1L CRISPR/Cas9 KO cells were obtained from Lewis Cantley lab, Weill Cornell Medical College, USA²⁰. MDA-MB-468 were authenticated by STR profiling in our lab in 2021 (originally from ATCC); LN229 cells obtained from Alexei Vazquez lab Beatson Institute for Cancer Research (authenticated by STR profiling in our lab in 2021) (originally from ATCC); 4T1 cells from Luxembourg Institute of Health, Clement Thomas lab (authenticated by Clement lab in 2018) (originally from ATCC). HCT-116 cells from University of Luxembourg, Elisabeth Letellier lab (originally from ATCC), HEK293T: NorLux Lab, Luxembourg Institute of Health (originally from ATCC). HAP1 cells were obtained from Patel’s laboratory⁵³ (originally from Horizon Discovery). MTX-resistant MDA-MB-468 cells were generated through prolonged culturing of MDA-MB-468 cells with 50 nM MTX for 2 months until cells started to regrowth under MTX. All cell lines were routinely tested for mycoplasma contamination.

Metabolic characterization

For metabolic characterization of cell lines, previously established protocols for absolute quantification of exchange fluxes and intracellular fluxes of one-carbon metabolism were applied²¹.

Stable isotope tracing and metabolite extraction

Stable isotope tracing experiments with [U-¹³C]-glucose tracer or [U-¹³C]-glutamine tracer (Cambridge Isotope Laboratories, CLM-1396) were performed in DMEM supplemented with 2 mM glutamine, 17 mM glucose, and 10 % FBS. Stable isotope tracing experiments with [U-¹³C]-serine tracer (Cambridge Isotope Laboratories) were performed in MEM (Thermo Fisher Scientific) supplemented with 400 μM glycine, 2 mM glutamine, 17 mM glucose, 400 μM serine tracer, and 10% FBS. To allow for adaptation, cells were cultivated in MEM for at least one passage prior to [U-¹³C]-serine tracer experiments. 150,000 to 200,000 cells were seeded in 12-well plates in triplicates for each experimental condition. Identical triplicate wells were seeded to allow for cell count and cell volume determination (to calculate the packed cell volume (PCV)) at the start and end of each tracing experiment. The day after seeding, growth medium was replaced by tracer medium and cells were cultured for 24 h. In parallel, 3 wells per condition were counted to assess starting PCV. After 24 h, triplicate wells were counted to assess PCV at the end of the experiment and one set of triplicates was used for subsequent metabolite extraction. Medium of these triplicates were collected and analyzed for exchange rates. To determine the basal medium composition for the subsequent calculation of exchange rates, identical medium was incubated in triplicates in empty 12 wells throughout the experiment and analyzed in parallel to the medium samples. Collected medium samples were centrifuged at 300 g for 5 min. Supernatant was collected and stored at −20°C until extraction of metabolites. Formate extraction, derivatization, and quantification as well as metabolite extraction for LC-MS analysis was performed as described in²¹. For metabolite extraction for GC-MS analysis after [U-¹³C]-glucose tracing and [U-¹³C]-glutamine tracing, cells were washed with cold 0.9% NaCl solution. 400 μl ice-cold MeOH/H₂O_MQ [(ratio, 1:1) containing the internal standards pentanedioc-d6 acid and [U-¹³C]-ribitol at a final concentration of 1 μg/ml and Tridecanoid-d25 acid at a final concentration of 5 μg/ml] was added to each well. Plates were incubated for 5 min at 4°C on a rocking shaker. Supernatant was collected, mixed with 200 μl CHCl₃, and centrifuged for 5 min at 13,000 g at 4°C. Upper polar phase was collected and stored at −20°C for subsequent MS analysis of polar metabolites.

GC-MS measurements

Analysis of formate release rates

Formate derivatization in the culture medium was performed using MCF derivatization as described in²¹. GC-MS analysis was performed using an Agilent 7890 A GC coupled to an Agilent 5975 C inert XL Mass Selective Detector (Agilent Technologies). A sample volume of 1 μl was injected into a Split/Splitless inlet, operating in split mode (20:1) at 270 °C. The gas chromatograph was equipped with a 30 m (I.D. 250 μm, film 0.25 μm) DB-5MS capillary column (Agilent J&W GC Column, 122-5532 G). Helium was used as carrier gas with a constant flow rate of 1.4 ml/min. GC oven temperature was held at 80 °C for 1 min and increased to 130 °C at 10 °C/min followed by a post run time of 4 min at 280 °C. Total run time was 15 min. Transfer line temperature was set to 280 °C. Mass selective detector (MSD) was operating under electron ionization at 70 eV. MS source was held at 230 °C and the quadrupole at 150 °C. For precise quantification, measurements were performed in selected ion monitoring mode. Target ions (m/z) and dwell times are shown in Table S1. GC-MS chromatograms were processed using Agilent MassHunter Quantitative Analysis for GC-MS, Version B.08.00. Final determination of release rates was performed as described in²¹.

Determination of mass isotopomere distribution (MID) of Intracellular TCA Cycle Metabolites following Stable Isotope Tracing

First, polar metabolites were derivatized for 90 min at 55 °C with 20 μl of methoxyamine (c = 20 mg/ml) in pyridine under continuous shaking and subsequently for 60 min at 55 °C with additional 20 μl of MTBSTFA w/ 1% TBDMCS. GC-MS analysis was performed using an Agilent 7890B GC coupled to an Agilent 5977 A Mass Selective Detector (Agilent Technologies). A sample volume of 1 μl was injected into a Split/Splitless inlet, operating in splitless mode at 270 °C. Gas chromatograph was equipped with a 30 m (I.D. 250 μm, film 0.25 μm) ZB-35MS capillary column with 5 m guard column (Phenomenex). Helium was used as carrier gas with a constant flow rate of 1.2 ml/min. GC oven temperature was held at 100°C for 2 min and increased to 300°C at 10°C/min and held for 4 min. Total run time was 26 min. Transfer line temperature was set to 280°C. Mass selective detector (MSD) was operating under electron ionization at 70 eV. MS source was held at 230°C and the quadrupole at 150 °C. For precise quantification of the MID, measurements were performed in selected ion monitoring mode. Target ions (m/z) and dwell times are shown in Table S2.

The MetaboliteDetector software package (Version 3.220180913) was used for mass spectrometric data post processing, quantification, MID calculations, correction of natural isotope abundance, and determinations of fractional carbon contributions⁶³.

Analysis of medium exchange rates

Polar metabolites of the culture medium were derivatized for 90 min at 45°C with 20 μl of methoxyamine (c = 20 mg/ml) in pyridine under continuous shaking and subsequently for 30 min at 45 °C with 20 μl of MSTFA. GC-MS analysis was performed using an Agilent 7890B GC coupled to an Agilent 5977 A Mass Selective Detector (Agilent Technologies). A sample volume of 1 μl was injected into a Split/Splitless inlet, operating in split mode (10:1) at 270°C. Gas chromatograph was equipped with a 30 m (I.D. 250 μm, film 0.25 μm) ZB-35MS capillary column with 5 m guard column (Phenomenex). Helium was used as carrier gas with a constant flow rate of 1.2 ml/min. GC oven temperature program: 90°C for 1 min, 9°C/min to 270°C, 25°C/min to 320°C and held for 7 min. Total run time was 30 min. Transfer line temperature was set to 280°C. MSD was operated under electron ionization at 70 eV. MS source was held at 230°C and the quadrupole at 150°C. Full scan mass spectra were acquired between m/z 70 and 700.

The MetaboliteDetector software package (Version 3.220180913) was used for quantification. Briefly, peak areas of all isotopologues of defined quantification ions were summed up and divided by the peak area of the internal standard for normalization. In addition, a calibration curve was prepared to calculate absolute concentrations. Absolute uptake and release rates were calculated as described in²¹

LC-MS measurements

Untargeted LC-MS analysis was carried out as previously described in²¹.

Nucleotide and NAD/NADH analysis

The following analytical conditions are based on a protocol from²¹. Metabolite analyses were performed using a Thermo Vanquish Flex Quaternary LC coupled to a Thermo Q Exactive HF mass spectrometer. Chromatography was carried out with a SeQuant ZIC-pHILIC 5 µm polymer (150 × 2.1 mm) column connected to the corresponding SeQuant ZIC-pHILIC Guard (20 × 2.1 mm) pre-column. Column temperature was maintained at 45 °C. The flow rate was set to 0.2 mL/min and the mobile phases consisted of 20 mmol/L ammonium carbonate in water, pH 9.2 (Eluent A) and Acetonitrile (Eluent B). The gradient was: 0 min, 80% B; 2 min, 80% B; 17 min, 20% B; 18 min 20% B; 19 min 80 % B; 20 min 80% B (0.4 mL/min); 24 min 80% B (0.4 mL/min); 24.5 min 80% B. The injection volume was 5 µL. All MS experiments were performed using electrospray ionization with polarity switching enabled (+ESI/-ESI). The source parameters were applied as follows: sheath gas flow rate, 25; aux gas flow rate, 15; sweep gas flow rate, 0; spray voltage, 4.5 kV (+) / 3.5 kV (–); capillary temperature, 325°C; S-lense RF level, 50; aux gas heater temperature, 50°C. The Orbitrap mass analyzer was operated at a resolving power of 30,000 in full-scan mode (scan range: m/z 75…1000; automatic gain control target: 1e6; maximum injection time: 250 ms). Data were acquired with Thermo Xcalibur software (Version 4.3.73.11) and analyzed with TraceFinder (Version 4.1). Subsequent data analysis for normalization and natural isotope subtraction were performed using in house scripts as in²¹.

Western blot

Total cell lysates from cells cultured in vitro were prepared by 30 min incubation of cell pellets on ice in cell lysis buffer (150 mM NaCl, 1 mM EDTA, 50 mM Tris-HCl, 1% NP-40). Protein lysates from tumor tissue were obtained by lysing 10 mg tumor tissue pieces in 150 μl cell lysis buffer. Homogenization of tumor tissue lysates was achieved with TissueLyser II (Quiagen) using 5 mm metal beads. Lysis efficiency was maximized by sonification. Lysis solution was centrifuged at 13,000 g for 15 min at 4°C and supernatant was collected and stored at −80°C. Protein concentration was determined by Bradford assay. 15 – 30 μg of total protein were loaded on NuPAGE 4–12% Bis-Tris gels (Life Technologies) or RunBlue 4–12% Bis-Tris gels (Westburg) using 4x NuPage LDS Sample buffer (Thermo Fisher Scientific) supplemented with 10 mM DTT (Sigma Aldrich) and blotted on nitrocellulose membrane according to standard protocols. Membranes were stained with REVERT staining solution (LI-COR) and analyzed for total protein abundance. Subsequently, membranes were blocked with Odyssey TBS blocking buffer (LI-COR) or 5% milk-powder in TBST and incubated with the indicated primary antibodies over night at 4°C. Incubation with secondary antibody occurred for 2 h at RT. Detection was performed with the Odyssey CLx Infrared Imaging System (LI-COR). ImageStudioLite Software Vers.5.2 (LI-COR) was used for image analysis. All uncropped and unprocessed scans are available in the source data file. Antibodies used for Western blot analysis in this study:

MTHFD1L (16113-1-AP, LOT: 00024322, RRID: AB_2250974) from Proteintech; Vimentin (3390, clone 5G3F10, LOT:2, RRID: AB_2216128), β-actin (3700, clone 8H10D10, LOT: 18, RRID: AB_2242334), NRF2 XP(R) (12721, clone D179C, LOT: 8, RRID: AB_2715528) and MTHFD2 (41377, D8W9U, LOT: 1, RRID: AB_2799200) from Cell Signaling Technology (CST); PHGDH (HPA021241, LOT: B115626, RRID: AB_1855299), PSAT1 (HPA042924, LOT: R40839, RRID: AB_2678223), PSPH (HPA020376, LOT: B115600, RRID: AB_1855867), SHMT1 (HPA023314, LOT: A78402, RRID: AB_1856830), and SHMT2 (HPA020549, LOT: A96534, RRID: AB_1856834) from Sigma Aldrich; LaminB (ab16048, LOT: GR32836742, RRID: AB_443298) from Abcam; IRDye 680RD Goat Anti-Mouse IgG (H + L) (926-68070, LOT: C90910-21, RRID: RRID AB_10956588) and IRDye 800CW Donkey Anti-Rabbit IgG (H + L) (926-32213, LOT: 926-32213, RRID: RRID AB_621848) from LI-COR (used in a 1:10,000 dilution). All antibodies were used in a 1:1,000 dilution unless stated otherwise.

Poly-L-Lysine coating

Poly-L-lysine (P1274) was purchased from Sigma-Aldrich and reconstituted at 500 μg/ml in H₂O_MQ. Wells were coated with PLL prior to Seahorse measurement and Scratch assay. To this end, PLL solution was diluted 1:20 in H₂O_MQ and added to the plates at least 1 h prior to seeding. Following incubation with PLL at 37°C and prior to cell seeding, plates were washed twice with H₂O_MQ and allowed to air-dry.

Seahorse measurements

The day prior to measurement, 40,000 cells were seeded on poly-L-lysine coated plates and treated the subsequent day as indicated. XF96 Extracellular Flux Analyzer (Seahorse Bioscience) was used to measure basal OCR following manufacturer’s instructions using WAVE software. OCR was normalized to the protein concentration or cell number in the wells following the protocol described in⁶⁴ using Bradford assay.

Flow cytometric analysis of cell cycle distribution and cell death

200,000 cells were seeded in 2 ml DMEM and treated the subsequent day as indicated. After incubation, medium was collected and cells were washed with PBS. PBS fraction was collected and cells were detached with trypsin. Detached cells were collected in DMEM. The combined, collected solutions were centrifuged and pellet was washed with PBS.

Cell cycle distribution

Centrifugation yielded a pellet that was resuspended in 100 μl PBS and fixed with ice-cold 80% EtOH. Fixed cells were stored at −20°C for at least 1 h and maximum 5 days prior to measurement. Cells were centrifuged, pellet was incubated for 1 h in 200 μl RNAse A in PBS (30 μg/ml) at RT. Immediately prior to measurement, 98 μl propidium iodide (PI) in PBS (50 μg/ml) was added. Flow cytometric analysis was performed using BD FACSCanto and BD FACSDiva software (version 8.0.1). Analysis was performed in FlowJo Version 10.6.2.

Cell death analysis

Pellet following centrifugation was resuspended in 50 μl AnnexinV-FITC staining solution (5% AnnexinV-FITC in AnnexinV binding buffer (10 mM HEPES pH 7.4, 140 mM NaCl, 2.5 mM CaCl₂, 0.1% BSA in _ddH₂O)) and incubated for 15 min on ice in the dark. 450 μl PI-staining solution (1.1 μg/ml PI in AnnexinV binding buffer) was added immediately prior to measurement using BD FACSCanto and BD FACSDiva software (version 8.0.1) (for wild-type cells) or NovoCyte Quanteon and software NovoExpress (version 1.5.0) (for GFP-positive transfected cells). Analysis was performed in FlowJo Version 10.6.2.

Flow cytometric analysis of ROS levels

250,000 MDA-MB-468 cells were seeded in 2 ml medium and treated the subsequent day as indicated. Following incubation, cells were detached with trypsin, centrifuged at 350 g for 5 min and washed with warm DMEM. Cells were stained in 100 μl DMEM supplemented with 1:2,000 DAPI and 1:500 DCFDA for 30 min at 37°C. Following incubation, samples were centrifuged at 350 g for 5 min and washed with PBS. Following centrifugation at 350 g for 5 min, cells were resuspended in 100 μl PBS and measured using the BD LSRFortessa^TM system and BD FACSDiva software (version 8.0.1). Data analysis was performed using FlowJo software Version 10.6.2.

Cell proliferation

Cell proliferation was determined using the IncuCyte^® Live-Cell Analysis system (Essen Bioscience). Cell proliferation was either determined as measurement of cell density (confluence) or by quantification of red nuclei using the IncuCyte^® NucLight Rapid Red Reagent (Essen Bioscience). Staining was performed according to manufacturer’s instructions. Alternatively, viable cell number was determined by trypan blue staining and automatic counting using a Countess™ Cell Counting Chamber Slide (Thermo Fisher).

Migration assay (wound closure)

Cell migration was determined using the IncuCyte® Scratch Wound Assay system for 96 well plates (Essen Bioscience). 96 well image lock plates (Essen Bioscience) were PLL coated prior to seeding of 40,000 cells per well. Wound application was performed 24 h post seeding using the 96-pin IncuCyte WoundMaker tool (Essen Bioscience) and indicated treatments were applied simultaneously. Measurement was performed in technical replicates (n = 4–8) and repeated as independent, biological experiments as stated for the respective experiment in figure legends.

Invasion and migration assay (transwell assay)

Invasion and migration was analyzed by Boyden chamber assay as described in Meiser et al., 2018²². Transwells were coated with ECM-Collagen for invasion assays and left un-coated for migration assay. Treatment and migration/invasion was carried out for 18 h to 24 h.

RNA extraction, cDNA synthesis, and qPCR analysis

Total RNA was extracted from cell culture dishes following manufacturer instructions of RNeasy Mini Kit (Qiagen, 74104) or NucleoSpin RNA Kit (Machery&Nagel, 740955). For RNA extracts from tumor tissue, 10 mg tissue were homogenized in 350 μL RNA using 5 mm metal beads in TissueLyser II (Quiagen). RNA was transcribed to cDNA with High-capacity cDNA Reverse Transcription Kit (ThermoFisher, 4368814). qPCR was performed from 20 ng cDNA per sample using Fast SYBR™ Green Master Mix (ThermoFisher, 4385612). All samples were analyzed in technical duplicates or triplicates. qPCR using Fast SYBR Green was conducted at 95°C for 20 s, 40 cycles of 95°C for 1 s and 60°C for 20 s using the QuantStudio 5 Real-Time PCR System (Applied Biosciences, ThermoFisher Scientific). Specificity was verified by melt curve analysis. Relative quantification of each mRNA was done by QuantStudio Design&Analysis v1.5.1 software (Applied Biosciences, ThermoFisher Scientific) and comparative Ct method. All expression data was normalized to two housekeeping genes (GAPDH and CycloA). All primers used in this study are listed in Table S3.

shRNA-mediated gene silencing

Knockdown with shRNA was performed in MDA-MB-468 and 4T1 cells using lentiviral particles expressing two different shRNAs targeting the gene of interest and one non-silencing control vector (pGIPZ-shSCR: TTACTCTCGCCCAAGCGAG (Code: RHS4346). For silencing of the MTHFD2 gene in MDA-MB-468 cells, pGIPZ shMTHFD2-2: ATTGCATTTCTATTGGCCT (V2LHS_90968) purchased from Horizon Discovery was used. For silencing of the ALDH1L2 gene in MDA-MB-468 cells, pGIPZ shALDH1L2: ACTGCTTTATCAACATCCG (Code: V3LHS_385780) purchased from Horizon Discovery was used. For silencing of the Mthfd1l gene in 4T1 cells, pGIPZ shMTHFD1L-2: TAGATTTCAATTTCATCTG (Code: V2LHS_96542) purchased from Horizon Discovery was used. For silencing Mthfd2 gene in 4T1 cells, pGIPZ shMthfd2-2: ATGACAACGGCTTCATTTC (Code: V2LMM_23800) purchased from Horizon Discovery was used. Plasmids used for generation of lentivirus were amplified from glycerol stocks, isolated (Machery&Nagel plasmid isolation kit), and used for transfection of HEK293T cells. Lentivirus was produced for 24 h in HEK293T cells co-transfected with the viral core packaging construct pCMVR8.74, the VSV-G envelope protein vector pMD2.G and the respective pGIPZ-shRNA-target plasmid using lipofectamine 3000 (Invitrogen #L3000008). Filtered virus supernatant was used to transduce MDA-MB-468 and 4T1 cells in DMEM supplemented with 2% FBS. After 24 h incubation at 37 °C, virus was removed and cells were cultured in DMEM supplemented with 10 % FBS for 48 h. 96 h post transduction, cells were cultured in selection media containing 2 µg/mL of puromycin (Sigma #P8833) for five passages to obtain stably transduced GFP-positive cells.

Generation of KO cells

Gene knock-out (KO) using the CRISPR-Cas9 system was performed in MDA-MB-468 using a vector expressing hCas9 and the respective guideRNA to target the gene of interest. For silencing of the MTHFD2 gene in MDA-MB-468 cells, two different gRNA sequences (gRNA#918: CGCCAACCAGGATCACACTC (Vector ID: VB190718-1104ncn); gRNA1.Ex: GCTCGCGGCAGTTCGGTAAGA (Vector ID: VB190719-1026vzh) purchased from VectorBuilder were used. A control vector was used scramble_gRNA (GTGTAGTTCGACCATTCGTG (Vector ID: VB190718-1103vja) together with both constructs. To generate a MTHFD2-KO cell line, 300.000 cells were seeded the day prior to transfection into a 6 well plate in DMEM supplemented with 10% FBS. 24 h later, the culture medium was replaced with 2 mL of DMEM-F12 (LONZA #12-719 F) supplemented with 10 % FBS, Glucose (17 mM final concentration), Glutamine (2 mM final concentration), and 1 mM Sodium Formate and the cells were transfected using Opti-MEM, Lipofectamine, P3000 and 2 µg of plasmid DNA. Post incubation for 24 h, the transfection mix was removed and cells were replaced with fresh DMEM-F12. 72 h post transfection, cells were sorted into single cells by FACS based on their GFP fluorescence. Single cells were grown in DMEM-F12 supplemented with 1% Pen-Strep (Sigma, # P4333) until the formation of single colonies. These were further expanded and characterised by means of Western Blot.

siRNA-mediated gene silencing

500.000 MDA-MB-468 cells were transfected with 40 pmol siRNA upon seeding in 6 well-plates using Lipofectamine 3000 (ThermoFisher) according to the manufacturer’s instructions. Non-silencing control siRNA (siRNA ON-TARGETplus Human non-targeting (1810-10-05) siRNA SMARTpool) and siRNA directed against human ALDH1L2 mRNA (siRNA ON-TARGETplus Human ALDH1L2 (160428; 26918-01-0005) siRNA SMARTpool) were purchased from Dharmacon.

Animal model

Animal experiments were performed according to all applicable laws and regulations, after receiving approval by the institution’s Animal Experimentation Ethics Committee at UL (AEEC) and the veterinarian service of the Ministry of Agriculture, Viniculture & Rural Development (TumorMetab LUPA 2020/01). They ensure that care and use of animals for research purposes was conducted according to the EU Directive 2010/63/EU, as well as the Grand-Ducal Regulation of January 11, 2013 on the protection of animals used for scientific purposes. These included the justification of the use of animals, their welfare and the incorporation of the principles of the 3 R’s (Replacement, Reduction and Refinement). A biostatistician reviewed all animal protocols. Mice were housed in a specific pathogen free (SPF) facility at a relative humidity of 40–70%, at 22 °C, and in 12 h dark/light cycles. Syngeneic 4T1 mammary carcinoma cells were orthotopically implanted into the left mammary fat pads (1 injection/mouse) of immune competent female Balb/c mice (8-10 month old) according to established protocols⁶⁵. Briefly, each injection contained 2000 cells in a mixture of 25 μL PBS and 25 μL matrigel. 9 mice per group were injected with either 4T1 SCR or 4T1 Mthfd1l KD cells. Primary tumor growth was monitored between day 7 and day 35 of the experiment. Weight was monitored and no weight loss was observed. Experiment was terminated after 6 weeks and lung and liver were prepared for examination of metastatic outgrowth. No metastases were found in the liver. Macroscopic lung metastases were blindly counted under a microscope and microscopic lung metastases were visualized by H&E staining.

H&E staining

H&E staining was performed on 10 μm sections of paraffin embedded lung tissue. Staining was performed according to established protocols. Briefly, selected sections were dehydrated with MeOH and stained with Gill 2 hematoxylin. Sections were neutralized with successive washes of tap water, hard water (10 g MgSO₄ and 0.7 g NaHCO₃ per L), and distilled water. Subsequently, sections were stained with Eosin-solution and dehydrated through successive washes with 80%, 95%, 100% EtOH and Xylol prior to mounting. Images for quantification were acquired using BioTek Cytation 5 Cell Imaging Multimode Reader. Area of lung and metastasis were measured in ImageJ. Representative images for publication were acquired using Olympus IX83 microscope. Collapsed lungs were excluded from quantitative analysis.

Proteomics

SILAC (Stable Isotope Labeling with Amino acids in Cell culture) strategy was used for proteomic analysis. In general, MDA-MB-468 cells were cultivated in DMEM-F12 SILAC medium supplemented either with Lys⁰ and Arg⁰ (light channel) or Lys⁸ and Arg¹⁰ (heavy channel). After 6 passages, labeling efficiency of heavy channel was confirmed using LC-MS/MS. Cells in light channel were treated with 50 nM MTX for 24 h and 72 h, while heavy channel was used as control. Cell pellets of biological duplicates were collected. Proteins were extracted in lysis buffer (50 mM ammonium bicarbonate, 6 M Urea, 2 M Thio-urea, pH 8) following a 30 min incubation at 4°C in the presence of protease inhibitors (cOmplete™ EDTA-free Protease Inhibitor Cocktail, Roche). Following centrifugation at 16,000 g for 10 min, supernatants were taken for protein quantification. Samples from light channel (50 μg protein) were mixed with control heavy channel (50 μg protein) for protein reduction (5 mM DTT, 1 h incubation at 37°C) and alkylation (10 mM IAA, 45 min in dark at room temperature). Protein digestion was performed with Lys-C (FUJIFILM Wako, 125-05061) at 1:30 ratio (enzyme/protein substances) for 4 h at 37 °C, then samples were diluted 4 times with 50 mM ammonium bicarbonate and digested overnight with 1 μg of trypsin at 37°C. Digestion was terminated through addition of formic acid (1% final concentration). Digested peptides were cleaned up with reverse phase Sep-Pak C18 1 cc Vac Cartridge (Waters, WAT054955) and eluted with 1 mL 50% ACN. Eluted peptides were dried by Speedvac (Thermo Fisher Scientific) and re-suspended in 0.1% formic acid. Peptide concentration was measured with Nanodrop. Peptides were measured by LC-MS/MS on Q-Exactive HF mass spectrometer (Thermo Fisher) connected to a Dionex Ultimate 3000 (Thermo Fisher). 500 ng of peptides were loaded onto a trap column (Acclaim PepMap 75 μm x 2 cm, C18, 3 μm) and separated on a 25 cm Acclaim pepmap RSLC column (75 μm x 25 cm, C18, 2 μm) using a 150 min gradient (2% to 90% acetonitrile) with a flow rate of 0.3 μL/min. MS data were acquired in data-dependent mode. Survey scans of peptide precursors from 375 to 1500 m/z were performed at 70,000 resolution with a 3 × 10⁶ ion count target and the top 12 abundant peaks from survey scan were selected for fragmentation. Tandem MS was performed by isolation at 1.4 m/z with the quadrupole, HCD fragmentation with a normalized collision energy of 28. The MS2 ion count target was set to 1 × 10⁵ and the max injection time was 45 ms. Only precursors with a charge state of 2–7 were sampled for MS2. The dynamic exclusion duration was set to 20 s with a 10 ppm mass tolerance around the selected precursor and its isotopes. Each sample was analyzed twice as technical replicates. All raw data was analyzed with MaxQuant (version 1.6.7.0) and searched with Andromeda against the Homo sapiens database from Uniprot. The minimal peptide length was set to 7 amino acids and the maximum of 3 missed cleavages were allowed. The search included variable modifications of methionine oxidation and N-terminal acetylation, deamidation (N and Q) and fixed modification of carbamidomethyl cysteine. The “Match between run” was checked within 1 min retention time window. Mass tolerance for peptide precursor and fragments were set as 10 ppm and 20 ppm, respectively. The FDR was set to 0.01 for peptide and protein identifications. SILAC based protein quantification (MaxQuant built-in) was used for quantitative evaluation of identified proteins. ProTIGY (https://github.com/broadinstitute/protigy), an R based tool, was used for differential analysis of MaxQuant output.

Statistics

Unpaired t-test with Welch’s correction was applied for pairwise comparison (two-sided) using GraphPad Software Vers.8. For comparison of multiple groups ordinary Brown Forsythe & Welch one-way or two-way ANOVA with Dunnett’s multiple comparison or Games-Howell’s multiple comparisons test was performed using GraphPad Software Vers.8. Resulting p-values for comparisons of interest are given as numerical values within the respective graphs. For normalization, data points of one experiment were either normalized to the untreated control or divided by the global mean of the individual experiment. We define one n as one independent biological experiment (in some cases further consisting of several wells, e.g. triplicate wells for all stable isotope tracing experiments). The technical mean of one biological experiment was considered as one n. The mean values of several independent, biological experiments (as indicated in figure legends) were plotted and used for statistical analysis as indicated.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

Source data are provided with this paper. Proteomics data is available at the proteomics identification database (PRIDE) under the id: PXD027175 (INSERT URL here, once available). The remaining metabolomics data from targeted analysis are available within the article and the supplementary information files (source data file). Source data are provided with this paper.

References

Siegel, R. L., Miller, K. D. & Jemal, A. Cancer statistics, 2019. CA Cancer J. Clin. 69, 7–34 (2019).
Article PubMed Google Scholar
Luengo, A., Gui, D. Y. & Vander, M. G. Heiden, targeting metabolism for cancer therapy. Cell Chem. Biol. 24, 1161–1180 (2017).
Article CAS PubMed PubMed Central Google Scholar
Counihan, J. L., Grossman, E. A. & Nomura, D. K. Cancer metabolism: current understanding and therapies. Chem. Rev. 118, 6893–6923 (2018).
Article CAS PubMed Google Scholar
Bergers, G. and S. M. Fendt, The metabolism of cancer cells during metastasis. Nat. Rev. Cancer 21(3),162–180 (2021).
Locasale, J. W. Serine, glycine and one-carbon units: cancer metabolism in full circle. Nat. Rev. Cancer 13, 572–583 (2013).
Article CAS PubMed PubMed Central Google Scholar
Locasale, J. W. et al. Phosphoglycerate dehydrogenase diverts glycolytic flux and contributes to oncogenesis. Nat. Genet 43, 869–874 (2011).
Article CAS PubMed PubMed Central Google Scholar
Pollari, S. et al. Enhanced serine production by bone metastatic breast cancer cells stimulates osteoclastogenesis. Breast Cancer Res Treat. 125, 421–430 (2011).
Article CAS PubMed Google Scholar
Possemato, R. et al. Functional genomics reveal that the serine synthesis pathway is essential in breast cancer. Nature 476, 346–350 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Pacold, M. E. et al. A PHGDH inhibitor reveals coordination of serine synthesis and one-carbon unit fate. Nat. Chem. Biol. 12, 452–458 (2016).
Article CAS PubMed PubMed Central Google Scholar
Tajan, M. et al. Serine synthesis pathway inhibition cooperates with dietary serine and glycine limitation for cancer therapy. Nat. Commun. 12, 366 (2021).
Article CAS PubMed PubMed Central Google Scholar
Wang, Q. et al. Rational design of selective allosteric inhibitors of PHGDH and serine synthesis with anti-tumor activity. Cell Chem. Biol. 24, 55–65 (2017).
Article CAS PubMed Google Scholar
Weinstabl, H. et al. Intracellular trapping of the selective phosphoglycerate dehydrogenase (PHGDH) inhibitor bi-4924 disrupts serine biosynthesis. J. Med Chem. 62, 7976–7997 (2019).
Article CAS PubMed Google Scholar
Maddocks, O. D. K. et al. Modulating the therapeutic response of tumours to dietary serine and glycine starvation. Nature 544, 372–376 (2017).
Article ADS CAS PubMed Google Scholar
Ngo, B. et al. Limited environmental serine and glycine confer brain metastasis sensitivity to PHGDH inhibition. Cancer Disco. 10, 1352–1373 (2020).
Article CAS Google Scholar
Mullarky, E. et al. Identification of a small molecule inhibitor of 3-phosphoglycerate dehydrogenase to target serine biosynthesis in cancers. Proc. Natl Acad. Sci. USA 113, 1778–1783 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Ducker, G. S. et al. Reversal of cytosolic one-carbon flux compensates for loss of the mitochondrial folate pathway. Cell Metab. 24, 640–641 (2016).
Article CAS PubMed Google Scholar
Lee, W. D. et al. Tumor reliance on cytosolic versus mitochondrial one-carbon flux depends on folate availability. Cell Metab. 33, 190–198.e6 (2021).
Article CAS PubMed Google Scholar
Morscher, R. J. et al. Mitochondrial translation requires folate-dependent tRNA methylation. Nature 554, 128–132 (2018).
Article ADS CAS PubMed PubMed Central Google Scholar
Minton, D. R. et al. Serine catabolism by SHMT2 is required for proper mitochondrial translation initiation and maintenance of formylmethionyl-tRNAs. Mol. Cell 69(4), 610–621.e5 (2018).
Article CAS PubMed PubMed Central Google Scholar
Zheng, Y. et al. Mitochondrial one-carbon pathway supports cytosolic folate integrity in cancer cells. Cell 175(6), 1546–1560.e17 (2018).
Article CAS PubMed PubMed Central Google Scholar
Meiser, J. et al. Serine one-carbon catabolism with formate overflow. Sci. Adv. 2(10), e1601273 (2016).
Article ADS PubMed PubMed Central CAS Google Scholar
Meiser, J. et al. Increased formate overflow is a hallmark of oxidative cancer. Nat. Commun. 9(1), 1368 (2018).
Article ADS PubMed PubMed Central CAS Google Scholar
Rossi, M., et al., Heterogeneity in PHGDH protein expression potentiates cancer cell dissemination and metastasis. https://doi.org/10.1101/2021.01.24.427949 (2021).
Samanta, D. et al. PHGDH expression is required for mitochondrial redox homeostasis, breast cancer stem cell maintenance, and lung metastasis. Cancer Res 76, 4430–4442 (2016).
Article CAS PubMed Google Scholar
Rinaldi, G. et al. In vivo evidence for serine biosynthesis-defined sensitivity of lung metastasis, but not of primary breast tumors, to mTORC1 inhibition. Mol. Cell 81, 386–397.e7 (2021).
Article CAS PubMed Google Scholar
Stoller, R. G. et al. Use of plasma pharmacokinetics to predict and prevent methotrexate toxicity. N. Engl. J. Med 297, 630–634 (1977).
Article CAS PubMed Google Scholar
Chabner, B. A. & Young, R. C. Threshold methotrexate concentration for in vivo inhibition of DNA synthesis in normal and tumorous target tissues. J. Clin. Invest 52, 1804–1811 (1973).
Article CAS PubMed PubMed Central Google Scholar
Huang, S. et al. Study on relationships of tumor status and gene polymorphism with blood concentration of MTX and toxicities in 63 pediatric mature B cell lymphoma in chinese population. Technol. Cancer Res Treat. 20, 1533033821995288 (2021).
CAS PubMed PubMed Central Google Scholar
Kawakatsu, S. et al. Population pharmacokinetic analysis of high-dose methotrexate in pediatric and adult oncology patients. Cancer Chemother. Pharm. 84, 1339–1348 (2019).
Article CAS Google Scholar
Phillips, D. C., Woollard, K. J. & Griffiths, H. R. The anti-inflammatory actions of methotrexate are critically dependent upon the production of reactive oxygen species. Br. J. Pharm. 138, 501–511 (2003).
Article CAS Google Scholar
Soula, M. et al. Metabolic determinants of cancer cell sensitivity to canonical ferroptosis inducers. Nat. Chem. Biol. 16, 1351–1360 (2020).
Article CAS PubMed PubMed Central Google Scholar
Wu, S. et al. Activation of RSK2 upregulates SOX8 to promote methotrexate resistance in gestational trophoblastic neoplasia. Lab Invest. 2021.
AlBasher, G. et al. Methotrexate-induced apoptosis in human ovarian adenocarcinoma SKOV-3 cells via ROS-mediated bax/bcl-2-cyt-c release cascading. Onco Targets Ther. 12, 21–30 (2019).
Article CAS PubMed Google Scholar
Garcia-Sanchez, A., Miranda-Diaz, A. G. & Cardona-Munoz, E. G. The role of oxidative stress in physiopathology and pharmacological treatment with pro- and antioxidant properties in chronic diseases. Oxid. Med Cell Longev. 2020, 2082145 (2020).
Article PubMed PubMed Central CAS Google Scholar
Babiak, R. M. et al. Methotrexate: pentose cycle and oxidative stress. Cell Biochem Funct. 16, 283–293 (1998).
Article CAS PubMed Google Scholar
Lignitto, L. et al. Nrf2 activation promotes lung cancer metastasis by inhibiting the degradation of Bach1. Cell 178, 316–329 e18 (2019).
Article CAS PubMed PubMed Central Google Scholar
Wiel, C. et al. BACH1 stabilization by antioxidants stimulates lung cancer metastasis. Cell 178, 330–345 e22 (2019).
Article CAS PubMed Google Scholar
Benzarti, M. et al. Metabolic potential of cancer cells in context of the metastatic cascade. Cells 9, 2035 (2020)
Nguyen, H. L. et al. Oxidative stress and prostate cancer progression are elicited by membrane-type 1 matrix metalloproteinase. Mol. Cancer Res 9, 1305–1318 (2011).
Article CAS PubMed PubMed Central Google Scholar
Rhyu, D. Y. et al. Role of reactive oxygen species in TGF-beta1-induced mitogen-activated protein kinase activation and epithelial-mesenchymal transition in renal tubular epithelial cells. J. Am. Soc. Nephrol. 16, 667–675 (2005).
Article CAS PubMed Google Scholar
Chen, D. et al. MiR-373 drives the epithelial-to-mesenchymal transition and metastasis via the miR-373-TXNIP-HIF1alpha-TWIST signaling axis in breast cancer. Oncotarget 6, 32701–32712 (2015).
Article CAS PubMed PubMed Central Google Scholar
Reczek, C. R. & Chandel, N. S. ROS-dependent signal transduction. Curr. Opin. Cell Biol. 33, 8–13 (2015).
Article CAS PubMed Google Scholar
Hunter, M. V. et al. Oxidative stress orchestrates cell polarity to promote embryonic wound healing. Dev. Cell 47, 377–387 e4 (2018).
Article CAS PubMed Google Scholar
Porporato, P. E. et al. A mitochondrial switch promotes tumor metastasis. Cell Rep. 8, 754–766 (2014).
Article CAS PubMed Google Scholar
Piperigkou, Z. et al. Key matrix remodeling enzymes: functions and targeting in cancer. Cancers (Basel), 13(6):1441 (2021)
Kawami, M. et al. Methotrexate-induced epithelial-mesenchymal transition in the alveolar epithelial cell line A549. Lung 194, 923–930 (2016).
Article CAS PubMed Google Scholar
Ohbayashi, M. et al. Involvement of epithelial-mesenchymal transition in methotrexate-induced pulmonary fibrosis. J. Toxicol. Sci. 39, 319–330 (2014).
Article CAS PubMed Google Scholar
Vazquez, A. et al. Catabolic efficiency of aerobic glycolysis: The Warburg effect revisited. BMC Syst. Biol. 4, 58 (2010).
Article PubMed PubMed Central CAS Google Scholar
Diehl, F. F. et al. Cellular redox state constrains serine synthesis and nucleotide production to impact cell proliferation. Nat. Metab. 1, 861–867 (2019).
Article CAS PubMed PubMed Central Google Scholar
Vande Voorde, J. et al. Improving the metabolic fidelity of cancer models with a physiological cell culture medium. Sci. Adv. 5, eaau7314 (2019).
Article ADS PubMed PubMed Central CAS Google Scholar
Palm, W. & Thompson, C. B. Nutrient acquisition strategies of mammalian cells. Nature 546, 234–242 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Tibbetts, A. S. & Appling, D. R. Compartmentalization of Mammalian folate-mediated one-carbon metabolism. Annu Rev. Nutr. 30, 57–81 (2010).
Article CAS PubMed Google Scholar
Burgos-Barragan, G. et al. Mammals divert endogenous genotoxic formaldehyde into one-carbon metabolism. Nature 548, 549–554 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Mullen, A. R. et al. Reductive carboxylation supports growth in tumour cells with defective mitochondria. Nature 481, 385–U171 (2012).
Article ADS CAS Google Scholar
Metallo, C. M. et al. Reductive glutamine metabolism by IDH1 mediates lipogenesis under hypoxia. Nature 481(7381), 380–384 (2011).
Article ADS PubMed PubMed Central CAS Google Scholar
Piskounova, E. et al. Oxidative stress inhibits distant metastasis by human melanoma cells. Nature 527, 186–191 (2015).
Article ADS CAS PubMed PubMed Central Google Scholar
Lawrence, S. A. et al. Mammalian mitochondrial and cytosolic folylpolyglutamate synthetase maintain the subcellular compartmentalization of folates. J. Biol. Chem. 289, 29386–29396 (2014).
Article CAS PubMed PubMed Central Google Scholar
Lowe, K. E. et al. Regulation of folate and one-carbon metabolism in mammalian cells. II. Effect of folylpoly-gamma-glutamate synthetase substrate specificity and level on folate metabolism and folylpoly-gamma-glutamate specificity of metabolic cycles of one-carbon metabolism. J. Biol. Chem. 268, 21665–21673 (1993).
Article CAS PubMed Google Scholar
Horne, D. W., Holloway, R. S. & Said, H. M. Uptake of 5-formyltetrahydrofolate in isolated rat liver mitochondria is carrier-mediated. J. Nutr. 122, 2204–2209 (1992).
Article CAS PubMed Google Scholar
den Boer, E. et al. Determinants of erythrocyte methotrexate polyglutamate levels in rheumatoid arthritis. J. Rheumatol. 41, 2167–2178 (2014).
Article CAS Google Scholar
Zhu, T. et al. Investigation of potential drug-drug interactions between peficitinib (ASP015K) and methotrexate in patients with rheumatoid arthritis. Clin. Drug Investig. 40, 827–838 (2020).
Article CAS PubMed PubMed Central Google Scholar
Soflaee, M. H. et al. Purine nucleotide depletion prompts cell migration by stimulating the serine synthesis pathway. Nat. Commun. https://doi.org/10.1038/s41467-022-30362-z (2022).
Hiller, K. et al. MetaboliteDetector: comprehensive analysis tool for targeted and nontargeted GC/MS based metabolome analysis. Anal. Chem. 81, 3429–3439 (2009).
Article CAS PubMed Google Scholar
Dranka, B. P. et al. Assessing bioenergetic function in response to oxidative stress by metabolic profiling. Free Radic. Biol. Med 51, 1621–1635 (2011).
Article CAS PubMed PubMed Central Google Scholar
Kocaturk, B. & Versteeg, H. H. Orthotopic injection of breast cancer cells into the mammary fat pad of mice to study tumor growth. J. Vis. Exp. 8(96), 51967 (2015).

Download references

Acknowledgements

We thank Fabien Rodriguez (Letellier lab) for technical support during the tissue analysis of the in vivo experiment. We are grateful to Lewis Cantley (Weill Cornell Medical College) for providing MDA-MB-468 cells with MTHFD1L KO²⁰ and to Patel’s laboratory (Cambridge, UK) for providing various HAP1 KO cells⁵³ We thank Clément Thomas (LIH, Luxembourg) for providing 4T1 cells and Saverio Tardito (Cancer Research UK Beatson Institute, Glasgow, UK) for providing Plasmax medium.

We would like to thank: the LCSB Metabolomics Platform, especially Xiangyi Dong and Floriane Vanhalle, for providing technical and analytical support; the National Cytometry Platform (Quantitative Biology Unit, LIH) and especially Thomas Cerutti for support with flow cytometric analyses; Magretta Adiamah for her diligent proofreading of the manuscript; all our collaboration partners for fruitful discussions and constructive feedback. All graphical figures were produced with BioRender.com (Agreement Number OJ23O9J9QD).

J.M. is supported by the FNR-ATTRACT program (A18/BM/11809970), J.M. and L.H. are supported by the FNR-PRIDE NEXTIMMUNE (PRIDE/11012546) and J.M. and K.E. by the FNR-PRIDE i2Tron (PRIDE19/14254520) program. N.K. is supported by the LIH Career Launchpad program (Legs Baertz) and by a DFG fellowship (KI 2508/1-1). D.B. is supported by the FNR-ATTRACT program (A14/BM/7632103), the FNR-CORE program (C18/BM/12691266 and C21/BM/15796788) and the i2Tron (PRIDE19/14254520) grant. D.B. and L.S.B. are funded by the FNR-PRIDE (PRIDE/11012546/NEXTIMMUNE) scheme. E.L. is supported by the FNR-CORE program (C16/BM/11282028 and C20/BM/14591557), by a Proof of Concept FNR grant (PoC/18/12554295), a PRIDE17/11823097 and by i2Tron (PRIDE19/14254520). G.D. and F.X. are supported by FNR-CORE (C17/BM/11642138). B.J. and M.Z.N. are supported by Luxembourg National Research Fund (C18/BM/12670304/COMBATIC) and FNRS Televie grant (7.4579.20).

Author information

Authors and Affiliations

Cancer Metabolism Group, Department of Cancer Research, Luxembourg Institute of Health, Luxembourg, Luxembourg
Nicole Kiweler, Catherine Delbrouck, Laura Neises, Kim Eiden, Mohaned Benzarti, Lara Haase & Johannes Meiser
Faculty of Science, Technology and Medicine, University of Luxembourg, 2 avenue de Université, Esch-sur-Alzette, Luxembourg
Catherine Delbrouck, Leticia Soriano-Baguet, Kim Eiden, Mohaned Benzarti, Lara Haase & Gunnar Dittmar
Faculty of Science, Technology and Medicine, Department of Life Sciences and Medicine, Molecular Disease Mechanisms Group, University of Luxembourg, Esch-sur-Alzette, Luxembourg
Vitaly I. Pozdeev, Eric Koncina, Maryse Schmoetten & Elisabeth Letellier
Experimental & Molecular Immunology, Department of Infection and Immunity, Luxembourg Institute of Health, 29 Rue Henri Koch, Esch-sur-Alzette, Luxembourg
Leticia Soriano-Baguet & Dirk Brenner
Immunology & Genetics, Luxembourg Centre for Systems Biomedicine, University of Luxembourg, 7 Avenue des Hauts Fourneaux, Esch-sur-Alzette, Luxembourg
Leticia Soriano-Baguet & Dirk Brenner
Proteomics of cellular signaling, Department of Infection and Immunity, Luxembourg Institute of Health,1a Rue Thomas Edison, Strassen, Luxembourg
Feng Xian & Gunnar Dittmar
Luxembourg Centre for Systems Biomedicine, University of Luxembourg, Esch-sur-Alzette, Luxembourg
Lara Haase & Christian Jaeger
Tumor Immunotherapy and Microenvironment (TIME) Group, Department of Cancer Research, Luxembourg Institute of Health, Luxembourg, Luxembourg
Muhammad Zaeem Noman & Bassam Janji
Institute of Cancer Sciences, University of Glasgow, Glasgow, UK
Alexei Vazquez
Odense Research Center for Anaphylaxis (ORCA), Department of Dermatology and Allergy Center, Odense University Hospital, University of Southern Denmark, Odense, Denmark
Dirk Brenner

Authors

Nicole Kiweler
View author publications
You can also search for this author in PubMed Google Scholar
Catherine Delbrouck
View author publications
You can also search for this author in PubMed Google Scholar
Vitaly I. Pozdeev
View author publications
You can also search for this author in PubMed Google Scholar
Laura Neises
View author publications
You can also search for this author in PubMed Google Scholar
Leticia Soriano-Baguet
View author publications
You can also search for this author in PubMed Google Scholar
Kim Eiden
View author publications
You can also search for this author in PubMed Google Scholar
Feng Xian
View author publications
You can also search for this author in PubMed Google Scholar
Mohaned Benzarti
View author publications
You can also search for this author in PubMed Google Scholar
Lara Haase
View author publications
You can also search for this author in PubMed Google Scholar
Eric Koncina
View author publications
You can also search for this author in PubMed Google Scholar
Maryse Schmoetten
View author publications
You can also search for this author in PubMed Google Scholar
Christian Jaeger
View author publications
You can also search for this author in PubMed Google Scholar
Muhammad Zaeem Noman
View author publications
You can also search for this author in PubMed Google Scholar
Alexei Vazquez
View author publications
You can also search for this author in PubMed Google Scholar
Bassam Janji
View author publications
You can also search for this author in PubMed Google Scholar
Gunnar Dittmar
View author publications
You can also search for this author in PubMed Google Scholar
Dirk Brenner
View author publications
You can also search for this author in PubMed Google Scholar
Elisabeth Letellier
View author publications
You can also search for this author in PubMed Google Scholar
Johannes Meiser
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

Conceptualization, J.M., N.K.; Methodology, J.M., N.K., C.J., M.Z.N., B.J., E.L., D.B., G.D., V.I.P.; Software, A.V., F.X.; Validation, N.K., J.M., L.N., C.D., K.E., L.H.; Formal Analysis, N.K., C.D., L.N., M.B., L.S.B., F.X., V.I.P., M.S., C.J., K.E., E.K., L.H.; Investigation, N.K., C.D., L.N., M.B., L.S.B., F.X., V.I.P., M.S., E.L., K.E., L.H.; Resources, C.J., A.V., J.M., E.L.; Data Curation, N.K., J.M., F.X., A.V.; Writing – Original Draft, N.K., J.M.; Writing – Review & Editing, All Authors; Visualization, N.K., J.M., L.S.B., F.X., V.I.P.; Supervision, J.M., D.B., A.V., E.L., G.D.; Project Administration, J.M., Funding Acquisition, J.M., D.B., E.L., A.V., G.D.

Corresponding author

Correspondence to Johannes Meiser.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Kiweler, N., Delbrouck, C., Pozdeev, V.I. et al. Mitochondria preserve an autarkic one-carbon cycle to confer growth-independent cancer cell migration and metastasis. Nat Commun 13, 2699 (2022). https://doi.org/10.1038/s41467-022-30363-y

Download citation

Received: 24 January 2022
Accepted: 22 April 2022
Published: 16 May 2022
DOI: https://doi.org/10.1038/s41467-022-30363-y

This article is cited by

The significant role of amino acid metabolic reprogramming in cancer
- Xiaohong Liu
- Bo Ren
- Yupei Zhao
Cell Communication and Signaling (2024)
Cycling back to folate metabolism in cancer
- Younghwan Lee
- Karen H. Vousden
- Marc Hennequart
Nature Cancer (2024)
Formate overflow drives toxic folate trapping in MTHFD1 inhibited cancer cells
- Alanna C. Green
- Petra Marttila
- Johannes Meiser
Nature Metabolism (2023)
Purine nucleotide depletion prompts cell migration by stimulating the serine synthesis pathway
- Mona Hoseini Soflaee
- Rushendhiran Kesavan
- Gerta Hoxhaj
Nature Communications (2022)