Genetic Disruption of Lactate/H⁺ Symporters (MCTs) and Their Subunit CD147/BASIGIN Sensitizes Glycolytic Tumor Cells to Phenformin

Unlike in normal differentiated cells, most rapidly growing tumors rely primarily on glycolysis as a major source of energy even in the presence of oxygen, which is referred to as “fermentative glycolysis” (1). This phenomenon, confirmed through the use of ¹⁸Fluoro-deoxyglucose PET (2), is now considered as a key metabolic hallmark of invasive cancers (3). Recent studies linked the Warburg effect to metabolic reprogramming, in which oncogenic transformation, as well as the transcriptional activity of c-Myc and the hypoxia-inducible factor1 (HIF1), induce glycolytic activity (4–6) associated to parallel inhibition of the pyruvate dehydrogenase (PDH) complex (7, 8).

Even if fermentative glycolysis generates only a few ATP molecules considering the large amounts of glucose consumed, this pathway appears to be well fitted for rapid growth of normal and cancer cells, as it provides not only energy, but also anabolic precursors (6, 9, 10). However, the high glycolytic rate associated with a production of lactic acid rapidly creates a hostile acidic and hypoxic microenvironment that induces dissemination of metastatic variants (11). Hypoxia via HIF1 controls intracellular pH (pHi) by inducing the membrane-bound carbonic anhydrases IX and XII (12–14) and a major H⁺/lactate⁻ symporter, MCT4 (15–17). Considering the key role of glycolysis in “glucose-addicted” tumors, targeting this bioenergetic and anabolic pathway represents an attractive therapeutic approach (18–22).

The regulation of pHi is a key determinant for many physiologic processes and is certainly essential in maintaining a high glycolytic rate. For this reason, we have exploited our knowledge on lactic acid transport to evaluate the impact of its inhibition as an anticancer strategy (21, 23). Transport of lactate across the plasma membrane is carried out by a family of solute carrier family 16 (SLC16), or monocarboxylate transporters (MCT) in which four (MCT1-4) catalyze bidirectional proton-coupled transport of lactate. MCT1 is induced by Myc and MCT4 by HIF1, and are upregulated in several cancers (17, 24). These two transporters require an accessory glycoprotein CD147/BASIGIN (BSG) for proper folding, stability, and trafficking to the cell surface (25, 26). BSG is a type I transmembrane glycoprotein that has been reported to be overexpressed in different metastatic tumor cells (24, 26). Thus targeting BSG/MCTs complex using a pharmacologic inhibitor or shRNA-mediated knockdown was reported to impair the in vitro and in vivo growth of pancreatic tumor cells (27), Ras-transformed fibroblasts, colon adenocarcinoma (21), and Myc-induced human malignancies (22), which suggested that blocking lactic acid export provides an efficient metabolic therapy to limit tumor cell growth (14).

In the present study, we explored further the mechanism by which blockade of lactic acid export exerts cytostasis and could be turned into a strategy to provoke tumor cell death. Because there is no pharmacologic inhibitor for MCT4, we designed genetic experiments to knockout, using the zinc finger nuclease (ZFN), either MCT4 or BSG or both in two glycolytic tumor models, the colon adenocarcinoma (LS174T), and glioblastoma (U87) human cell lines. We demonstrated that tumor cells grown in normoxia or hypoxia survived the MCT blockade by reactivating OXPHOS. This key finding prompted us to test the sensitivity of these cells to phenformin a mitochondrial complex I inhibitor. Indeed, the two tumor cell lines that survive MCT blockade with minimal growth rapidly enter into “ATP crisis” and die in vitro and in vivo upon treatment with phenformin.

Colon adenocarcinoma LS174T cells (kindly provided by Dr. Van de Wetering, Oxford University, Oxford, United Kingdom), and glioblastoma U87 cells (kindly provided by Dr. Adrian Harris, Oxford University) were grown in DMEM (Gibco) supplemented with 10% FBS, penicillin (10 U/mL), and streptomycin (10 μg/mL). Chinese hamster lung CCL39 fibroblasts obtained from the ATCC and the CCL39-derived mutants impaired either in respiration (res⁻) or in glycolysis (gly⁻), were obtained as described (28, 29) and maintained in DMEM supplemented with 7.5% FBS. Cell lines were identified by DNA fingerprint and STR loci (DSMZ). With the exception of “Seahorse” experiments, all in vitro studies were performed in parallel, under normoxia or hypoxia 1% O₂ as reported (21) to fully express glycolytic enzymes, particularly MCT4.

The different LS174T stable cell lines (1 × 10⁶ cells) suspended in 300 μL of serum-free DMEM supplemented with insulin–transferrin–selenium (Life Technologies) were injected subcutaneously into the back of 8-week-old male athymic mice (Harlan). All procedures were approved by the Institutional Animal Care and Use Committee at the University of Nice-Sophia Antipolis (Nice, France; CIEPAL-azur agreement NCE/165; see Supplementary Data for details).

The different LS174T cell lines (6 × 10⁴ cells) were seeded onto 6-well plates, three wells per cell line and per condition. We measured proliferation by detaching the cells by trypsinization, and counting them daily with a Coulter Z1 (Beckman) during 5 days. Proliferation units were calculated by dividing the cell number obtained for each day by the one obtained 24 hours after seeding.

The rate of extracellular acidification (ECAR) and the oxygen consumption rate (OCR) were measured using the Seahorse XF24 analyzer (Seahorse Bioscience). LS174T cells (9 × 10⁴) and U87 cells (3 × 10⁵) were seeded on Seahorse 24-well plates, 24 and 3 hours, respectively, before analysis. Cells were maintained in cell culture medium without glucose, pyruvate, serum, or bicarbonate, and incubated for 45 minutes in a non-CO₂ incubator at 37°C. The wells were mixed for 2 minutes and the pH and oxygen concentration measured every 22 seconds for 3 minutes. Different agents were then injected into the wells at a final concentration of 10 mmol/L of glucose (Sigma), 300 nmol/L of MCT1i, 50 μmol/L of phenformin (Sigma), and 1 μmol/L of rotenone (Sigma). After each addition, four data points of 3 minutes were taken to determine the OCR (pMoles O₂/minutes) and ECAR (mpH/minute). The protein concentration was measured to normalize the OCR and ECAR values.

Data are expressed as mean ± SD. Each experiment was performed at least three times and the number of experiments is represented by n. Statistical analysis was done with the unpaired Student t test. Differences between groups were considered statistically significant when P < 0.05.

We first knocked out the MCT4 gene in LS174T cells that express high levels of BSG, MCT1, and MCT4; the mutations altering the two alleles are depicted in Supplementary Fig. S1A. We then knocked out the BSG gene in wild-type (WT) LS174T (Supplementary Fig. S1B) or in LS174T-MCT4^−/− generating the double knocked out BSG^−/−, MCT4^−/− cell line. The MCT4^−/− cell line shows the lack of expression of MCT4 in both normoxia and hypoxia (Supplementary Fig. S1A). Ablation of MCT4 was accompanied by a slight decrease in the expression of BSG in normoxia and hypoxia (Fig. 1A), and by an increase in MCT1 expression that is more apparent in hypoxia (Fig. 1A). BSG knockout lead to the complete loss of both forms, the heavily glycosylated (around 65 kDa) and the immature low-molecular weight form of 39 kDa. BSG disruption, as expected from its chaperone role, induces decreased expression of both transporters MCT1 and MCT4 (Fig. 1A). Similar results were obtained for the human glioblastoma cell line. Knockout of the BSG gene in U87 cells induced a major decrease in the protein level of MCT1 and MCT4 in both normoxia and hypoxia (Supplementary Fig. S2A).

Next, we investigated the impact of these mutations on the V_max values of MCT1 and MCT4 by measuring initial lactic acid transport rates. The initial entry rates of total lactic acid transport in hypoxic LS174T cells was 61 ± 5% for MCT1 and 39 ± 5% for MCT4. Quantification was facilitated by the use of the specific MCT1 inhibitor (AR-C155858) referred to as MCT1i and the observation that LS174T cells do not express MCT2 and MCT3 (data not shown). Disruption of the MCT4 gene did not modify the transport activity of MCT1 (Fig. 1B). However, addition of 300 nmol/L MCT1i to the MCT4^−/− cell line totally abolished facilitated lactic acid transport, confirming the nonexpression of other MCT isoforms (Fig. 1B). Disruption of BSG severely reduced the lactic acid transport activity. Thus, MCT1 and MCT4 expression was reduced by 10- and 6-fold, respectively, in LS174T (Fig. 1B), and by 3.9- and 3.3-fold, respectively, in U87 (Supplementary Fig. S2B), confirming that in the absence of BSG, MCTs, fully transcribed (Supplementary Fig. S3), were retained and degraded in the endoplasmic reticulum as nonfunctional transporters (25).

The next step was to analyze the pHi in the LS174-mutant derivatives. The alkaline resting pHi of the parental LS174 cell line, close to 8, was essentially not affected by the complete inhibition of MCT1 (Fig. 1C). The additional knockout of MCT4 had a very modest effect on pHi. In sharp contrast, in the MCT4^−/− cells, MCT1i increased by 10-fold the intracellular H⁺ concentration; the resting pHi decreased from about 7.9 to 6.9 (Fig. 1C).

When mutants derived from LS174T cells were starved for 1 hour of glucose, they rapidly accumulated intracellular lactic acid upon glucose addition with a steady-state equilibrium that was reached within 3 to 5 hours (Fig. 2A). In agreement with the results described above, BSG^−/− and BSG^−/−, MCT4^−/− cells, compared with parental cells, accumulated a large intracellular pool of lactic acid ranging from 5 to 6 mmol/L lactate in parental cells to 20 mmol/L in the double knockout BSG^−/−, MCT4^−/− cells (Fig. 2A, left). The intracellular pool of lactic acid of the single mutant MCT4^−/− was not affected (7 mmol/L), indicating the major role played by MCT1 alone in lactic acid export. As expected, treatment of the cells with MCT1i further increased the intracellular lactate levels particularly for the MCT4^−/− cells reaching 38 to 40 mmol/L (Fig. 2A, right). This accumulation was less pronounced in the BSG^−/− cells, whereas parental cells remained less sensitive to MCT1 inhibition. With such a block in lactic acid export, retro-inhibition on lactic dehydrogenease-A (LDHA) was expected to lead to a parallel increase in pyruvic acid pool. Pyruvate concentrations were about 20-fold lower than lactate pool in parental cells but increased 4- to 6-fold in the three MCT/BSG knockout cells (Fig. 2B, right).

The intracellular glycolytic metabolite and proton accumulation mediated by MCT blockade had, not surprisingly, a strong negative impact on the rate of glycolysis. Data from the Seahorse analyzer confirmed the rapid decrease in the ECAR after BSG^−/− knockout (2-fold) or BSG^−/−,MCT4^−/− double knockout (3-fold), whereas the ECAR was similar in MCT4^−/− and parental cells (Fig. 2C). Interestingly, MCT1 inhibition did not affect the ECAR of parental cells, a result that was consistent with the relatively high expression of MCT4 in normoxic LS174T cells (Fig. 1A). However, compared with parental cells, MCT1i virtually suppressed the ECAR in the MCT4^−/− cells (8 and 16-fold decreases of ECAR in mct4^−/− and BSG^−/−,MCT4^−/− cells, respectively; Fig. 2C).

The corresponding OCR, which was low in parental cells, in contrast increased progressively with mutations affecting the intensity of the glycolytic blockade (WT < MCT4^−/− < BSG^−/− < BSG^−/−,MCT4^−/−; Fig. 2D). As expected, the OCR was increased in the disrupted cell lines upon MCT1 inhibitor. These results were confirmed in U87 cells. Indeed, both U87 BSG-null cells displayed a decreased ECAR (50%) and even an increased OCR (53%) compared with the parental U87 cells. These changes were magnified by the pharmacologic inhibition of MCT1 (Supplementary Fig. S2C and S2D).

Taken together, these findings highlight the plasticity of cancer cells and clearly demonstrate the ability of highly glycolytic tumor cells, such as LS174T and U87 cells, to rapidly shift their metabolism to OXPHOS, a mechanism of resistance to inhibition of fermentative glycolysis.

To further understand how this metabolic switch occurred and to which substrate the knockout cells relied on to increase their respiration, we first assessed the impact of glutamine deprivation on the OCR. Indeed, glutamine metabolism is well documented as a carbon and bioenergetic source for cancer cells (30).

Our results showed that the highly glycolytic parental LS174T cells displayed, in the presence of 2 mmol/L glutamine as the only exogenous carbon source (no glucose), a rather low respiratory rate (OCR = 180) that was increased when the MCT activity was partially disrupted (OCR = 380 in MCT4^−/− and 600 in BSG^−/−; Fig. 2E). Removal of glutamine from this medium sharply reduced by 3- to 4-fold the rate of respiration for the three cell lines thereby highlighting the key contribution of glutamine to respiration (Fig. 2E and Supplementary Fig. S4A and S4B). We next conducted the experiment without glutamine to analyze the contribution of glucose in OXPHOS (Fig. 2F). The rate of respiration increased in proportion to the decreased level of the MCT activity (WT < MCT4^−/− < BSG^−/−) and became maximal in the presence of MCT1i (Fig. 2F). It is interesting to note that in the presence of MCT1i, both MCT4^−/− or BSG^−/− cells respired at the same level, that is, 4-fold above the parental cells (Fig. 2F), a value that paralleled the increased intracellular pool of pyruvate in these two cell lines (5-fold increase in the first hour of MCT1i addition; Fig. 2B, right). Because this increase in OXPHOS is sensitive to rotenone (Fig. 2F) or metformin/phenformin (Supplementary Fig. S5B), mitochondrial complex I inhibitors (31–33), we propose that the accumulation of pyruvate increases, by mass action, the activity of PDH.

We next reported the impact of inhibition of glycolysis on the in vitro growth rates of the four LS174T-derived cell lines (Fig. 3A and B) and of the three U87-derived cell lines (Supplementary Fig. S2E). In normoxia, the disruption of MCT4, BSG, or both had only a modest effect on the growth rate (less than 20%, for BSG^−/−; Fig. 3A and Supplementary Fig. S2E, left). In hypoxia, the trend in general is that many tumor cells display a reduced growth rate (2-fold reduction for the parental LS174T cells). Again, it is the BSG^−/−,MCT4^−/− cells that display the most reduced growth rate in hypoxia compared with the parental cells (1.5-fold, Fig. 3A, right). In contrast with these modest effects on growth, the addition of the MCT1i, that does not affect the parental cell growth, decreased the growth of U87 BSG^−/− cells by 2-fold (Supplementary Fig. S2E), and severely reduced the growth of LS174T -MCT4^−/− cells by, respectively, 3-fold (Fig. 3A, left) and up to 6-fold in hypoxia (Fig. 3A, right).

Interestingly, this decrease in growth rate correlated with the decrease in pHi (Fig. 1C), which is known to restrict mTORC1 (34) as seen by inhibition of p70 S6 kinase phosphorylation (Fig. 3B). In other words, the MCT1i sensitized MCT4-null cells to growth arrest by inducing a pHi drop of 1 pH unit. The most remarkable finding is that ablation of lactic acid export induced only cytostasis with minimal loss of cell viability (data not shown). The bioenergetic plasticity reported above is certainly the basis of this acquired tumor cell viability under MCT blockade. We thus analyzed the in vitro growth of LS174T cells (WT or BSG^−/−) treated with MCT1i or the lipophilic biguanide metformin analog, phenformin, or both. Figure 3C shows that the parental cells, either grown in normoxia or hypoxia, are not affected by any of the treatments, MCT1i, phenformin, or both. Disruption of BSG alone, however, severely restricted clonal growth in response to phenformin in normoxia but not in hypoxia (Fig. 3C). In hypoxia, although MCT1i or phenformin alone has no effect, their combination suppressed clonal growth, highlighting a clear case of synthetic lethality (Fig. 3C). Similar results were obtained by replacing 50 μmol/L phenformin by 1 mmol/L of the hydrophilic and less permeable analog metformin.

The diverse growth phenotypes displayed in normoxia/hypoxia and in response to phenformin by the four cell lines, WT, MCT4^−/−, BSG^−/−, and BSG^−/−,MCT4^−/−, are easily explained by their energy metabolic features. In normoxia, the rates of glycolysis (ECAR), although high are further stimulated in response to phenformin for the WT and MCT4^−/− cell lines but not affected in the most MCT-disrupted cells BSG^−/− and BSG^−/−,MCT4^−/− (Supplementary Fig. S5A). The corresponding OCR showed a collapse of respiration for the four cell lines in response to phenformin (Supplementary Fig. S5B). The most dramatic phenformin-mediated growth restriction is only seen in the most restricted cell lines for MCT/glycolysis: BSG^−/− and BSG^−/−,MCT4^−/− (Supplementary Fig. S5C).

These findings received confirmation in two independent BSG-null tumor cell lines, glioblastoma U87 (Supplementary Fig. S2E) and lung adenoma carcinoma A549 (data not shown), pointing that targeting glycolysis (MCT inhibition) sensitized cells to phenformin. In hypoxic conditions, however, MCT4 induction induced a slight resistance to phenformin for BSG^−/− cells, whereas cells lacking both BSG and MCT4 remained highly sensitive (Fig. 3C and Supplementary Fig. S5C). For BSG^−/− and BSG^−/−,MCT4^−/− cells, we postulate that the elevated intracellular lactate levels played a negative feedback on the glycolysis, thus inhibiting the increased ECAR due to phenformin (Supplementary Fig. S5A and S5B). These findings were supported by other results obtained from the Ras-transformed fibroblast CCL39 cell line, for which we possess mutants impaired either in glycolysis (gly⁻; Supplementary Fig. S5D; refs. 29, 35) or in respiration (res⁻; Supplementary Fig. S5E; refs. 21, 28). As expected, only the CCL39 glycolysis–defective cells (gly⁻) were rapidly killed by phenformin treatment (Supplementary Fig. S5F).

In conclusion, phenformin/metformin severely impairs growth and viability of tumor cells only when fermentative glycolysis is compromised.

As mentioned above, phenformin treatment abolishes OXPHOS (Supplementary Fig. S5B). This resulted in decreased mitochondrial ATP and led to a compensatory increase in glycolysis with increased lactate production and extracellular acidification (Supplementary Fig. S5A). Here, we showed that 24 hours of phenformin treatment had no effect on intracellular ATP levels of LS174T WT cells in both normoxia and hypoxia (Fig. 4A and B). In the case of MCT4 knockout, phenformin only decreased the intracellular ATP by 10% showing that the compensatory glycolysis is sufficient to maintain high ATP levels. However, BSG^−/− and BSG^−/−,MCT4^−/− cells, impaired in their glycolytic flux, were not able to restore their intracellular ATP, which dropped by 60% and 80%, respectively, compared with WT cells in normoxia (Fig. 4A). MCT4 induction in hypoxia limited this decrease for BSG^−/− cells, confirming thus our findings on clonal growth.

Although MCT1 inhibition slightly decreased the intracellular ATP of BSG/MCT-disrupted cells, combination with phenformin induced an “energetic crisis” represented by a striking decrease of ATP (up to 85%; Fig. 4A and B). This ATP drop resulted in enhanced cytotoxicity and consequently rapid cell death. Indeed, 50% of either MCT4^−/− or BSG^−/− cells, and 90% of BSG^−/−,MCT4^−/− cells died in both normoxia and hypoxia after 24 hours of combined treatment; and only 8% of BSG^−/− cells survived under hypoxic conditions after 96 hours (Fig. 4C and D). These results pointed out the acute synthetic lethality achieved through the simultaneous block of lactic acid export and mitochondrial complex I.

To assess the efficacy of our combined strategy in vivo, LS174T-derived cell lines were injected subcutaneously to nude mice. After tumor appearance (>30 mm³), mice received 200 mg/kg/d phenformin in the drinking water, and/or, when specified, 100 mg/kg MCT1 inhibitor AZD3965 twice daily by oral gavage for 10 days. Interestingly, even if no relevant differences were seen in vitro, knockout of either MCT4, BSG, or both induced in untreated mice a modest (20%–25%) decrease in tumor growth (Fig. 5A–D). Comparing WT and BSG^−/− LS174T tumor growth, phenformin treatment alone had no effect (Fig. 5A and B), suggesting, as shown in Fig. 3C, an in vivo hypoxic environment. In contrast and consistent with in vitro clonal growth (Supplementary Fig. S5C), phenformin alone reduced by 2-fold tumor growth of MCT4^−/− and BSG^−/−,MCT4^−/− tumors (Fig. 5C and D). Treatment with MCT1i alone affected tumor growth of the four tumor cell lines. The reduction in the tumor growth rate of WT and BSG^−/− tumors (2-fold) was stopped with halt in treatment (arrows, Fig. 5A and B). As expected, the strongest inhibitory effect of MCT1i alone was observed in the two MCT4-null cell lines: 75% and 86% reduction in tumor volume, respectively, for MCT4^−/− and BSG^−/−,MCT4^−/− tumors (Fig. 5C and D). Finally, combination of treatment (phenformin/MCT1i) reduced further the tumor volume to 84% and 96% when measured at 20 days in the two MCT4-null cell lines (Fig. 5C and D).

Taken together, these findings support the importance of combining blockers of lactate/H⁺ transporters (MCT1 and MCT4) with inhibitors of OXPHOS (phenformin) for a novel efficient anticancer strategy.

In this study, we extended our previous work on the last step of glycolysis and further investigated via gene disruption the key role of MCTs/BSG complexes in controlling lactic acid export, bioenergetics, and tumor growth.

We showed that the growth of the two glycolytic cell lines, LS174 and U87, was not impaired by pharmacologic inhibition of MCT1 (MCT1i). The basal and hypoxia-induced expression of MCT4 was sufficient to ensure export of lactic acid and growth. Thus, it was clear from previous MCT knockdown studies that both MCT1/4 have to be suppressed for anticancer strategies, unless only MCT1 is expressed like in transformed fibroblasts (21) or neoplastic B cells (22). Interestingly BSG-knockout, as expected from Halestrap's work (26), severely reduced the MCT1/4 lactate transport activities but only minimally affected in vitro and in vivo growth of both cell lines. It is only when the full MCT activity was ablated, in MCT4-null cells treated with MCT1i that in vitro and in vivo growth was strongly reduced but not abolished (Figs. 3A and Fig. 5C). This strong inhibitory growth effect of MCT1i (300 nmol/L) raises the question of its specificity that we addressed previously with cells expressing only MCT1 or both MCT1/4 (21). We confirmed the specificity of MCT1i by showing that the residual growth rate seen in MCT1/MCT4 double knockout is not affected by MCT1i (data not shown). This tumor growth inhibition, however, is not due to cell death but due to cytostasis induced by mTORC1 inhibition in response to intracellular acidic stress (Fig. 3B), as we previously reported (34).

BSG-disruption in LS174T and U87 cells impacted strongly on the plasma membrane expression of MCT1/4 thereby reducing lactate transport and inducing a 2-fold reduction in the rate of glycolysis and almost no change in the rate of growth. This minimal impact on growth was surprising until we realized reactivation of OXPHOS in BSG-null cells. Thus, treating BSG- or MCT4-null cells with MCT1i reduced further lactate export, leading to substantial intracellular accumulation of lactate, pyruvate, and H⁺. This drastic metabolic and acidic stress inhibited further the glycolytic rate thereby mirroring a sharp increase in oxygen consumption. This metabolic plasticity, observed within minutes of addition of MCT1i, explains the cell survival by rapid maintenance of ATP levels. These findings contrasted with similar approaches that targeted MCT1 (22), MCT4 (36), or the upstream glycolytic step of LDHA (18, 20). In these three cases, cell death was reported to be associated with a decrease in ATP and an increase in oxidative stress. This was not the case for all the MCT-disrupted cell lines issued from LS174 even when associated with MCT1i treatment; ATP levels decreased no more than 20% and no reactive oxygen species (ROS) were detected (Fig. 4A and Supplementary Fig. S6A), a result apparently consistent with the antioxidant character of lactate (37). Nevertheless, this is an important issue that needs further exploration to fully understand resistance or vulnerability depending of the cancer model used.

How can we turn the cytostatic effect imposed by MCTs blockade into cell death? As ATP levels, rescued from OXPHOS, were due to activation of mitochondrial complex I (Fig. 2F and Supplementary Fig. S5B), the solution to the question came from associating phenformin with MCTs inhibition. The most remarkable point here is that only a 2-fold reduction in the glycolytic rate of BSG-null cells sensitized normoxic cells to death in response to phenformin or metformin. Note that these two biguanides, at their respective active concentrations (50 μmol/L, 1 mmol/L), had no detectable effect on growth and viability of WT cells (Fig. 3C, data not shown). Phenformin, alone or combined with MCT1i, reduced the ATP pool of BSG-null cells, respectively, by 60% and 80% in 24 hours. Loss of cell viability in either normoxia or hypoxia paralleled the rapid fall in ATP with no detectable apoptosis; this was acute necrotic cell death resulting from ATP crisis. Indeed combining phenformin with MCTs blockade induced ROS formation (Supplementary Fig. S6B), as reported by Cleveland's group (22), but cell death was not avoided by addition of N-acetyl-L-cysteine confirming cell death by “metabolic catastrophe.”

Lactic acid generated from tumors is emerging as a major metabolite that impacts on the tumor microenvironment not only as a signaling molecule, but also as an acid stressor compromising immune surveillance (14, 38, 39). High levels of tumor lactate in cervical cancers (40) as well as high expression of MCT4, a major lactate transporter in breast cancers (17), are reported to be associated with a high frequency of metastases, tumor recurrence, and low survival. High lactate levels also signal “pseudo hypoxia” by increasing HIF1α stability (41), which promotes or enhances tumor hypoxia. We confirmed that four hypoxic markers, CA9, LDHA, PDK1 and VEGF-A, were further induced 3- to 8-fold in hypoxia by the increased pool of 40 mmol/L lactate (Supplementary Fig. S3E). Lactate as a signaling metabolite was also reported to promote M2 polarization of tumor-infiltrated macrophages (42). The second edged sword of lactate is its acidic function that together with carbonic acid (43) contributes to the extracellular acidic tumor microenvironment (14). Few reports have demonstrated the devastating impact of lactic acid on several immune cell function in vitro (44) and inhibition of cytolytic NK in vivo (45). Interestingly, the anergic state established at low pH (6–6.5) of tumor-specific CD8(+) T lymphocytes could be reverted by restoring a physiologic pH (39).

Altogether, these protumoral features of lactate levels and acidic tumor microenvironment converge to raise optimism concerning an anticancer strategy that will reduce production and secretion of lactic acid. Here, we provided genetic evidence that targeting MCT1 and MCT4 reduced tumor growth and, if associated with the antidiabetic drugs phenformin/metformin (33), could, in a short tolerable therapeutic window, induce tumor cell death. Our optimism took off from our in vitro preliminary experiments on several cancer cell lines combining MCT1i and the novel specific MCT4 inhibitor in current development by AstraZeneca (46). Further experimentation with immune-competent mouse genetic cancer models and the MCT1/MCT4 inhibitors should demonstrate the benefit of this novel anticancer strategy.

No potential conflicts of interest were disclosed.

Conception and design: I. Marchiq, J. Pouyssegur

Development of methodology: I. Marchiq, R. Le Floch, M.-P. Simon, J. Pouyssegur

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Marchiq, D. Roux

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): I. Marchiq, M.-P. Simon

Writing, review, and/or revision of the manuscript: I. Marchiq, J. Pouyssegur

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): I. Marchiq, D. Roux, J. Pouyssegur

The authors thank Dr. Susan Critchlow (AstraZeneca) for providing the MCT1i, Prof. Andrew Halestrap (Bristol University) for the MCT4 vector, Prof. Laurent Counillon (Nice University) for helping with pHi experiments, Agnes Loubat for FACS analysis (CYTOMED), and Dr. Christiane Brahimi-Horn for editorial correction of the manuscript.

This work was funded by Ligue Nationale Contre le Cancer (LNCC; Equipe labellisée), Fondation ARC, INCa, ANR, the EU-FP7 “METOXIA,” SERVIER-CNRS, and Centre Lacassagne. I. Marchiq received a fellowship from LNCC.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.