Wild-type IDH2 promotes the Warburg effect and tumor growth through HIF1α in lung cancer.

Hotspot mutations of isocitrate dehydrogenase 1 and 2 (IDH1/2) have been studied in several cancers. However, the function of wild-type IDH2 in lung cancer and the mechanism of its contribution to growth of cancer cells remain unknown. Here, we explored the role and mechanism of wild-type IDH2 in promoting growth of lung cancer.
Methods: Information regarding genomic and clinical application focusing on IDH2 in cancer was examined in several databases of more than 1,000 tumor samples. IDH2 expression was assessed by immunohistochemistry in tissues from lung cancer patients. The biological functions of IDH2 were evaluated by using cell-based assays and in vivo xenograft mouse models.
Results: Here we reported that wild-type IDH2 is up-regulated and is an indicator of poor survival in lung cancer and several other cancers. Targeting IDH2 with shRNA resulted in decreased HIF1α expression, leading to the attenuation of lung cancer cell proliferation and tumor growth. Treatment of lung cancer cells with AGI-6780 (a small molecule inhibitor of IDH2), PX-478 (an inhibitor of HIF1α) or incubation with octyl-α-KG inhibited lung cancer cell proliferation.
Conclusion: IDH2 promotes the Warburg effect and lung cancer cell growth, which is mediated through HIF1α activation followed by decreased α-KG. Therefore, IDH2 could possibly serve as a novel therapeutic target for lung cancer.

Introduction

Lung cancer is the leading cause of cancer-related death worldwide 1. Non-small cell lung cancer (NSCLC) accounts for 80% to 85% of all lung cancer cases, including lung adenocarcinoma, squamous cell lung carcinoma and large cell lung carcinoma 2. The four main genetic mutations of NSCLC include K-RAS (24%), EGFR (13%), ALK (5%) and TP53 (5%) 3. EGFR inhibitors have been used as first-line treatment drugs against lung cancer 4. Unfortunately, the clinical outcome of NSCLC patients remains poor, with a 5-year overall survival (OS) rate of 18% in the USA and 10-15% in China 5; 6.

Reprogramming energy metabolism is proving to be a common survival mechanism in cancer cells and is recognized as a hallmark of cancer 7. Cancer cells take up glucose and glutamine at high rates as the two most important nutrients to support their survival and growth 8. Glucose and glutamine are utilized for biosynthesis, proliferation and NADPH production in cancer cells and both can be controlled by oncogenes such as K-RAS and C-MYC 9-14.

Interestingly, gain-of-function mutations of the isocitrate dehydrogenase 1 and 2 (IDH1/2) enzymes have been identified recently in several cancers 15. IDH1/2 catalyze the oxidative decarboxylation of isocitrate to α-ketoglutarate (α-KG) or the reverse reaction. Gain-of-function mutations in IDH1/2 have been shown to activate oncogenes and regulate the expression level of multiple genes 16. These mutations produce the “onco-metabolite” 2-hydroxyglutarate (2-HG) which is from α-KG 17; 18. 2-HG is a competitive inhibitor of α-KG-dependent dioxygenases 17; 18. Thus, IDH1/2 mutations impair the function of α-KG-dependent dioxygenases by consuming α-KG and competitively substituting α-KG with 2-HG. α-KG was shown to have antitumor effects through inhibition of angiogenesis in a murine tumor model 19. The α-KG-dependent dioxygenases use α-KG as a substrate. The dioxygenases include KDM, TET2, PHD2 and PLOD1-3, which control histone demethylation and hypoxia-inducible factor-1α (HIF1α)-dependent cellular signaling and collagen formation 15. HIF1α is broadly expressed and correlates with poor prognosis in human cancers by regulating genes involved in glycolysis, angiogenesis, cell cycle progression and other cellular pathways 20. HIF1α was shown to be a positive factor in solid tumor growth and was shown to be required for tumor growth and metastasis of NSCLC 21-23.

In addition to the well-known function of the mutant IDH2, the wild-type IDH2 protein (IDH2wt) was found to participate in glutamine metabolism and promoted cell survival. IDH2 carboxylates α-KG from glutamine to citrate in hypoxia to promote glioblastoma cell growth and increased viability 24. IDH1/2 participate in reductive carboxylation of glutamine to support redox homeostasis during anchorage-independent tumor spheroid formation 25. The protein level of wild-type IDH2 was found to be markedly up-regulated in esophageal squamous cell carcinoma (ESCC) tissues and is associated with worse overall survival and reduced progression-free survival of ESCC patients 26. IDH2wt suppresses melanoma cell growth, increases tumor-free survival in animal models and contribute to 5-hmC loss 27. High expression of 5-hydroxymethylcytosine and IDH2wt was associated with favorable prognosis after curative resection of hepatocellular carcinoma 28.

Here we reported that wild-type IDH2 is highly expressed in multiple cancers, especially in lung cancer, and this high level of IDH2 correlates with poor survival. We found that IDH2 enhances the Warburg effect and increases cell growth by promoting a lower intracellular level of α-KG. Targeting IDH2 with shRNA resulted in decreased HIF1α levels, leading to attenuated cell viability, proliferation and tumor growth. Treatment of cells with AGI-6780 (a small molecule inhibitor of IDH2), PX-478 (an inhibitor of HIF1α) or incubation with octyl-α-KG inhibited cell proliferation. Overall, these results suggest that high expression of wild-type IDH2 has an oncogenic function and might be a new therapeutic target against lung cancer.

Methods

Reagents and antibodies

CoCl2, dihydroethidium and propidium iodide were from Sigma/Aldrich. Octyl-α-ketoglutarate was from Cayman and PX-478 and AGI-6780 were from Selleck. Antibodies to detect IDH2, HIF1α, LDHA, ALDOA, PDK1, and total ubiquitination and the NADP/NADPH kit were from Abcam and the β-action antibody was from Sigma.

Immunohistochemistry analysis

Lung cancer tissues and paired normal tissues were from Xiangya Hospital and the Third Affiliated Hospital of Kunming Medical University and were validated by a pathologist from the Xiangya Hospital. The study protocol was approved by the hospital ethical review committees. Immunohistochemistry was performed as previously described 14; 29; 30.

Western blot analysis

Cells were harvested and washed with PBS twice, disrupted in IP buffer (Thermo) and centrifuged at 12,000 × g for 20 min. Protein (50 μg) from the supernatant fraction (quantified by the BCA Protein Assay Kit, Thermo) was subjected to SDS-PAGE, and transferred to a PVDF membrane (Millipore). Membranes were blocked with 5% non-fat milk for 1 h at room temperature and then incubated with the specific primary antibody, followed by the corresponding HRP-conjugated anti-mouse or anti-rabbit secondary antibody. Protein bands were visualized by the Western lightening plus-ECL kit (Pierce).

Generation of IDH2-overexpressing and IDH2-knockdown cell pools

For lentivirus production, an IDH2-overexpressing plasmid LV105-IDH2, its empty vector LV105 (Genecopoeia), shRNA plasmids, GV-248-sh-IDH2#1 and GV-248-sh-IDH2#2, and control vector GV-248 (Genechem,) were separately co-transfected with packaging (psPAX2) and envelope (pMD2.G) vectors into HEK293T cells. Lentivirus was harvested at 48 h post transfection from the supernatant fraction, and mixed with 5 mg/mL polybrene (Selleck) to increase the infection efficiency. H460 and A549 lung cancer cells were infected with the lentivirus and selected in 1 mg/mL puromycin (Selleck) for 1 week. For rescue experiments, an IDH2-overexpressing plasmid, LV105-IDH2-res (Genecopoeia), was transfected into lung cancer cells using Lipofectamine 2000 reagent (Invitrogen).

Cell culture

A549 lung cancer cells were originally purchased from American Type Culture Collection (kind gifts from Dr. Xingming Deng, Emory University Atlanta, Georgia 31). The H1299, H460, H520, H522, PC9 and MRC5 cells were obtained from Cancer Research Institute of Central South University. The H460 and A549 stable IDH2-overexpressing and IDH2-knockdown cells were maintained in RPMI 1640 with 10% fetal calf serum and penicillin/streptomycin/gentamicin antibiotics. Cells were grown in a 37 °C incubator with 5% CO2.

Cell viability and cell growth assays

Cells (5×104) were seeded in a 6-well plate and cell numbers were counted by Trypan blue exclusion using a blood counting chamber. Cells (2×103) were seeded in a 96-well plate and cell viability was estimated by MTS assay (Promega). Cells were treated with different concentrations of octyl-α-KG, PX-478, or AGI-6780 and then cell counting and cell viability estimation were performed as indicated above.

Cell death assay

Cell death was measured using propidium iodide (100 ng/mL) and a BD FACSCantoTM II flow cytometer (BD Biosciences).

Intracellular glucose uptake, lactate production, ATP assays

Cells (5×105) were seeded in a 6-well plate and culture medium was replaced after 8 h. The culture medium was collected after another 16 h. Glucose and lactate were measured using the Beckman AU680 Automatic Biochemistry Analyzer normalized by cell number. Cells (5×105) were seeded in a 6-well plate and treated with octyl-α-KG (1 mM) and PX-478 (10 uM) for 24 h and then glucose and lactate were measured as indicated above. Intracellular ATP concentration was measured by using an ATP bioluminescent somatic cell assay kit (Perkin Elmer). Cells (2×104 cells/well) were plated in a 96-well plate and harvested after 12 h. Luminescent intensity as the ATP value from each well was measured using an EnSpireTM Multilabel reader (Perkin Elmer).

ROS, α-KG, 2-HG and oxygen consumption rate assays

Intracellular levels of reactive oxygen species (ROS) were measured by flow cytometry. Cells were collected and suspended at 1×106 cells/mL in PBS and incubated with 1 μM dihydroethidium (Sigma) for 30 min before analysis by a BD FACSCantoTM II flow cytometer (BD Biosciences). The mean values were set as the ROS level. α-KG and 2-HG was measured using an alpha-ketoglutarate colorimetric fluorometric assay kit and D-2-hydroxyglutarate colorimetric assay kit (Biovision). The oxygen consumption rate (OCR) was measured using the MitoXpress® Xtra Oxygen Consumption Assay kit (Luxcel) and the slope of the fluorescence curve was set as the OCR. The OCR was also analyzed by Seahorse XF24 at baseline and after adding oligomycin, FCCP, and rotenone.

Database search

Clinical data were from publically available databases, including cBioprotal (TCGA) and Oncomine, which were used to explore the genomic changes and mRNA expression of IDH2 in cancers 32-34. Kmplot was used to compare survival profiles of cancer patients who were divided by high and low IDH2 expression 35. Survexpress was utilized to analyze the mRNA of IDH2 in differential pathological types and clinical stage of lung cancer tissues 36. The Cancer Cell Line Encyclopedia (CCLE) was employed to analysis the correlation of copy number and mRNA level.

Ubiquitination analysis

Cells incubated with CoCl2 (500 μM) for 24 h were harvested and the HIF1α antibody was used to immunoprecipitate total HIF1α. Ubiquitination of HIF1α was examined using a ubiquitination primary antibody and its corresponding secondary antibody.

Carbon-labeled isotopologues analysis

Cells were seeded and the culture medium was replaced with carbon-labeled [U-13C] glucose or [U-13C] glutamine for 24 h and then GC-MS was used for analysis.

Xenograft studies

Animal experiments were conducted according to the protocols approved by the Central South University Animal Care Committee. Nude mice (BCLB/c-nu, male, 4-6 weeks old) were subcutaneously injected with H460 stable IDH2 overexpressing, IDH2 knockdown cells or paired control cells. Tumor growth was recorded by measuring length and width of the tumors. Tumor size was calculated using the formula length×width2/2 and the tumors were harvested and weighed at the experimental endpoint. Tumor cell proliferation was determined by Ki-67 IHC staining.

RT-qPCR

Total mRNA was isolated utilizing the Trizol reagent (Invitrogen) and used for reverse transcription (Thermo Fisher Scientific). Then qRT-PCR was performed using the ABI7500 (Applied Biosystems) by mixing cDNA, primers and SYBR®Green Real-Time PCR Master Mixes (Invitrogen). The primer sequences are included in Supplementary Material 37.

Statistical analysis

Differences between lung cancer tissues and paired normal lung tissues were assessed by a paired student's t-test. Two-tailed unpaired t-tests and correlation analysis were conducted using GraphPad Prism 5. Statistical significance is reported with a p value of < 0.05.

Results

IDH2 is up-regulated in lung cancer tissues and is a negative prognostic factor for lung cancer patients

To obtain genomic and clinical outcome data regarding IDH2 expression in cancers, a search for the IDH2 gene was conducted in several databases containing more than 1,000 tumor samples. Amplification was the main genomic change occurring in the IDH2 gene and was found in 1.6% of the total samples (Figure ) 32; 33. IDH2 expression was found to be higher in lung cancer than in normal lung tissues (Table and Table ) and is the most altered of all the tricarboxylic acid cycle (TCA) genes (Table ) 34. High levels of IDH2 correlated with poor overall survival and shorter progression-free survival in lung cancer (Figure ) 35. Analysis of the correlation of IDH2 expression with clinical characteristics revealed that IDH2 is not associated with clinical stage but instead with differentiation and pathological type (Figure ). The correlation of IDH2 expression and overall survival of lung cancer patients was more obvious in adenocarcinomas in non-smoker patients (Figure ).

Additionally, we performed immunohistochemistry analysis of IDH2 in lung cancer and in paired noncancerous tissues derived from lung cancer patients (n = 33). IDH2 was mainly distributed in the cytoplasm of NSCLC cells (Figure ) and was significantly (p < 0.001) higher in adenocarcinoma and squamous cell carcinoma compared to the paired noncancerous pneumocyte tissues (Figure ).

From the TCGA and other databases, we found that the IDH2 mRNA level is positively correlated with its copy number in lung cancer cells (Figure ) and, in addition, breast cancer and pancreatic cancer carried the IDH2 amplification (Figure ). IDH2 is also an indicator of poor survival in these two cancers (Figure ). These data suggest an important role of IDH2 in multiple cancers.

Wild-type IDH2 controls intracellular α-KG concentration in lung cancer

To confirm that IDH2 is not mutated in lung cancer cells, we sequenced the genome of the NCI-H460 and A549 lung cancer cell lines and no mutations were found, which indicated that IDH2 is wild-type in these cells (Figure ). To determine the influence of wild-type IDH2 on intracellular α-KG concentration, H460 and A549 lung cancer cells stably overexpressing IDH2 (H460-IDH2 and A549-IDH2) or knockdown IDH2 (H460-shIDH2#1, #2 and A549-shIDH2 #1, #2) were generated by the lentivirus method and IDH2 expression was confirmed by Western blotting (Figure ).

The intracellular α-KG concentration was then analyzed. We found that the α-KG concentration was markedly decreased in H460-IDH2 and A549-IDH2 cells, but was increased in H460 and A549 IDH2 knockdown cells (Figure ). Additionally, we performed a xenograft experiment in which nude mice were subcutaneously injected with H460-IDH2 or H460-shIDH2 #1 cells and found that the α-KG concentration was also lower in H460-IDH2 xenograft tumor tissues, but higher in H460-shIDH2 #1 xenograft tumor tissues (Figure ). This suggests that IDH2 might consume α-KG in lung cancer cells.

In breast cancer cells, IDH2 can carboxylate α-KG from glutamine to citrate in hypoxic conditions and increase the 2-HG level 38. Hypoxia could induce production of 2-HG by LDHA and MDH1/2 39. We found that 2-HG was elevated in H460-IDH2 and A549-IDH2 cells and decreased in H460 and A549 IDH2 knockdown lung cancer cells under normal oxygen conditions (Figure ). Thus, wild-type IDH2 can decrease α-KG concentration and enhance the production of 2-HG. Reductive carboxylation of glutaminolysis can generate NADPH, which could be used to eliminate cellular ROS 25. We measured the total ROS level and the NADP/NADPH ratio by flow cytometry in IDH2-overexpressing cells. IDH2 overexpression reduced cellular ROS levels (Figure ) along with the NADP/NADPH ratio (Figure ), whereas knocking down IDH2 enhanced the total cellular ROS levels (Figure ) and increased the NADP/NADPH ratio (Figure ). These results suggest that the reverse reaction of wild-type IDH2 might consume intracellular α-KG, produce 2-HG and lower the level of cellular ROS under normal oxygen conditions in lung cancer cells.

To identify the metabolic flux that contributes to changes in α-KG, carbon-labeled glucose or glutamine was added in the culture medium and TCA cycle metabolites were measured. IDH2 overexpression decreased all carbon-labeled isotopologue TCA cycle metabolites generated from [U-13C] glucose but increased slightly those from [U-13C] glutamine (Figure ). The ratio of carbon-labeled α-KG from [U-13C] glucose was the most altered metabolite of all the tricarboxylic acid cycle metabolites (Figure ).

Wild-type IDH2 promotes the Warburg effect

Expression of IDH2 correlates with time to first progression and overall survival time, which indicates that IDH2 might influence key hallmarks of cancer, such as aerobic glycolysis, cell proliferation and death. Wild-type IDH2 regulates the concentration of α-KG, which is a substrate of PHD2 that degrades HIF1α. HIF1α plays a key role in the Warburg effect by transcriptionally regulating multiple genes, and specifically glycolytic genes, including aldolase A (ALDOA), lactate dehydrogenase A (LDHA) and pyruvate dehydrogenase kinase 1 (PDK1) 22; 37. We first analyzed the mRNA levels of IDH2 and HIF1α-targeted genes in lung cancer tissues taken from the database 36. IDH2 mRNA was positively correlated with 44 of a total of 74 HIF1α target genes analyzed in lung cancer tissues (Table ).

By examining glucose uptake and lactate production of NSCLC cells, we next determined whether IDH2 could influence the Warburg effect. We found that IDH2 overexpression promoted cellular glucose uptake, lactate production and lowered cellular ATP content (Figure ). In contrast, knockdown of IDH2 decreased cellular glucose uptake, lactate production and increased cellular ATP content (Figure ).

Enhancing cellular glycolysis and glutaminolysis might lower the amount of α-KG available for oxidative phosphorylation, thus decreasing the cellular ATP concentration and oxygen consumption rate (OCR). We found that IDH2 overexpression reduced the OCR, whereas knockdown IDH2 increased the OCR of lung cancer cells (Figure ).

We then determined the effect of IDH2 expression on HIF1α-targeted glycolytic gene expression. Lung cancer cells overexpressing IDH2 potentiated the mRNA expression of the HIF1α-targeted glycolytic genes ALDOA, LDHA and PDK1 (Figure ), whereas knockdown IDH2 had lower ALDOA, LDHA and PDK1 mRNA expression (Figure ). The protein level of ALDOA, LDHA and PDK1 was further analyzed in xenograft tissues. The expression of these three proteins was higher in H460-IDH2 xenograft tumor tissues and lower in H460-shIDH2 #1 cell xenograft tumor tissues (Figure ).

Wild-type IDH2 promotes lung cancer cell proliferation and tumor growth

We next examined the role of IDH2wt in cell proliferation and tumor growth. IDH2 overexpression enhanced cell viability and promoted proliferation. Knocking down IDH2 significantly decreased cell viability and inhibited proliferation (Figure ). IDH2 also diminished lung cancer cell death (Figure and Figure ). We also found that IDH2 overexpression enhanced colony formation (Figure and Figure ) and reduced apoptosis and TSZ-induced necroptosis in these cancer cells (Figure ).

Additionally, we performed a xenograft experiment in which nude mice were subcutaneously injected with H460-IDH2 cells or H460-shIDH2 #1 cells. Results showed that volume and weight of tumors from mice implanted with H460-IDH2 cells were higher compared with tumors from mice injected with empty vector cells (Figure and Figure ). In contrast, volume and tumor weight of tumors from mice implanted with H460-shIDH2 #1 cell were lower compared with tumors from mice injected with control cells (Figure and Figure ). These results overall demonstrated that IDH2 could promote tumor growth.

To confirm the specificity of shRNA manipulations, rescue experiments were conducted. H460 and A549 lung cancer cells stably expressing sh-IDH2 RNA were transiently transfected with an IDH2 wild type plasmid. Cell proliferation rates were significantly higher in IDH2 wild-type plasmid-transfected cells compared to controls (Figure ).

Wild-type IDH2 promotes the Warburg effect and cell growth mediated through HIF1α

We have shown that IDH2wt could promote the Warburg effect and cell growth, and α-KG might be a primary contributing molecular metabolite. One important α-KG-dependent dioxygenase is hypoxia-inducible factor prolyl hydroxylase 2 (PHD2), which can affect cellular metabolism by degrading HIF1α. HIF1α-targeted glycolytic gene expression was found positively correlated with IDH2 expression. We thus determined whether IDH2 could promote the Warburg effect and cell growth by modulating HIF1α. First, IDH2 and HIF1α expression was analyzed and found to be positively correlated in a panel of lung cancer cells and in normal MRC5 lung fibroblasts (Figure ). H460 and A549 IDH2-overexpressing or -knockdown cells were treated with CoCl2 and HIF1α expression was assessed. HIF1α protein levels were higher in H460 and A549 IDH2-overexpressing cell lines and decreased in IDH2-knockdown cell lines (Figure ). The total HIF1α ubiquitination levels were decreased in IDH2-overexpressing cell lines, which indicated that IDH2 increased HIF1α levels by lowering HIF1 α ubiquitination (Figure ).

Next, we determined whether changing the concentration of α-KG could affect the expression of the HIF1α protein. Cell-permeable octyl-α-KG and CoCl2 were added to the culture medium and at 24 h, the intracellular concentration of α-KG and the HIF1α protein level were analyzed. Increased amounts of intracellular α-KG were associated with lower cellular HIF1α protein levels (Figure ).

We further examined whether inhibiting HIF1α would influence the Warburg effect. Treatment of H460-IDH2 and A549-IDH2 cells with the HIF1α inhibitor, PX-478, reduced glucose uptake and lactate production rate (Figure ). This suggested that IDH2 promotes the Warburg effect by modulating the cellular HIF1α protein.

The effect of IDH2 on cell growth might be due to its effect on α-KG, 2-HG or NADPH. Hypoxia can induce production of 2-HG by LDHA, MDH1/2, but not wild type IDH1/2 39. The 2-HG concentration should be low in those cells with wild type IDH1/2 and not be a primary contributor to IDH2 function. Treatment with the ROS scavenger, N-acetyl-cysteine (NAC), had little effect on cell viability or number (Figure ) and could not restore the effect of the IDH2 inhibitor on cell viability (Figure ). We next examined the effect of the HIF1α inhibitor, PX-478, on cell proliferation. Incubation with PX-478 obviously diminished cell number at various time points (Figure ). These results overall suggest that the oncogenic function of wild-type IDH2 in lung cancer is mainly due to its effect on α-KG and IDH2 promotes cell growth by mediating the cellular HIF1α protein.

Targeting wild-type IDH2 with AGI-6780 and replenishing cellular α-KG inhibits NSCLC cell viability and growth

IDH2 was up-regulated in human cancers and attenuation of IDH2 inhibited lung cancer cell proliferation and tumor growth, which implicated IDH2 as a potential anticancer target. AGI-6780 is a commercially available IDH2 inhibitor that can target both the IDH2 mutant and wild type enzymes with different IC50s 40. We found AGI-6780 selectively inhibited proliferation of lung cancer cells, but did not affect normal MRC5 lung fibroblasts (Figure ). H460 and A549 cells treated with 20 μM AGI-6780 for 24 h had a 30-40% reduction in IDH activity (Figure ). AGI-6780 decreased lung cancer cell viability (Figure ) and proliferation (Figure ) in a time- and dose-dependent manner.

We found that incubation with the cell-permeable octyl-α-ketoglutarate diminished cellular glucose uptake and lactate production (Figure ). Additionally, this compound inhibited viability and proliferation of lung cancer cells (Figure ).

Discussion

IDH1 and IDH2 mutations have greatly altered our understanding of cancer biology by the discovery that they are new, unconventional oncogenes. Metabolic enzymes, including IDHs, have now gained a rightful place on the 'oncogene hit list' along with kinases and transcription factors 41. IDHs are highly mutated in glioma, cartilaginous tumors, acute leukemia and thyroid carcinomas. In a previous study, we showed that IDH2 rs11540478 is a risk factor for lung cancer 42. Here, we found that IDH2wt is up-regulated in lung cancer; and, wild-type IDH2 also performs a function similar to that of mutant IDHs. Wild-type IDH2 decreased the level of intracellular α-KG to promote the Warburg effect and cell growth, which is mediated by increasing HIF1α protein expression. HIF1α inhibited PDH by activating PDKs to decrease the flux of pyruvate to the TCA cycle. At the same time, LDHA also transcriptionally activated HIF1α and more lactate was produced from pyruvate, which contributes to the Warburg effect and cell growth.

Hypoxia can induce production of 2-HG by LDHA and MDH1/2 39. Because mutation-induced 2-HG is more than 100-fold higher than that found in normal cells, while in IDH2-overexpressed lung cancer cells is up to 2-fold than its control cells; the 2-HG may be produced by LDHA or MDH1/2. From these results, the function of IDH2wt seems to rely more on the effect of α-KG rather than 2-HG or ROS. The reduction of α-KG by IDH2wt might have an oncogenic function in lung tumorigenesis.

Interestingly, the AMPK/α-ketoglutarate axis dynamically regulates differentiation of progenitor cells 43. We noted that IDH2 expression is the highest in poorly differentiated lung cancer tissues (Figure ). We also found that IDH2 expression is the highest in large cell carcinoma (LCC). LCC is defined as undifferentiated NSCLC that lacks differentiation, which also provides clues for the relationship of IDH2 and cellular differentiation 44. Mechanistically, wild-type IDH2 might inhibit lung cancer cell differentiation through α-KG-dependent dioxygenases.

More importantly, we found that treatment of cells with AGI-6780 reduced IDH2 activity and decreased cellular viability and proliferation. Replenishing cellular α-KG also decreased cellular viability and growth in vivo. The inhibitors of IDHs have been used in clinical trials and α-ketoglutarate salts have been used as clinical nutrition and in metabolic care 45-49, which provides new clues for lung cancer treatment. However, survival data generated using public datasets have limitations, such as lack of confounding variables (such as gender and smoking status). In lung cancer, IDH1 and IDH2 are both overexpressed (Table ) and play the same role. IDH1 was also shown to promote tumor growth 50. IDH2 expression levels in gastric and kidney cancer are opposite to those of other cancers. The role and mechanism of IDH2 may be reversed in different types of cancer, which needs further study.

In summary, we reported a growth promotion mechanism of wild-type IDH2 in lung cancer and identified the role of α-KG and HIF1α in IDH2wt's oncogenic function. These results suggested that treatment of lung cancer cells with AGI-6780 (a small molecule inhibitor of IDH2), PX-478 (an inhibitor of HIF1α), and replenishing cellular α-KG might be a new strategy for the treatment of lung cancer in the future.

Supplementary figures, tables and experimental procedures.

Click here for additional data file.

Table 1

IDH2 expression profiles across multiple cancer types compared to normal tissue.

Analysis by Cancer type	Cancer vs. Normal
	Increase	Decrease
Breast	21	9
Lung	12	0
Ovarian	11	0
Gastric	0	9
Kidney	1	9

Numbers are datasets containing IDH2 mRNA expression (Oncomine).

Table 2

TCA cycle gene expression profiles in lung cancer compared to normal tissue.

TCA cycle genes	Cancer vs. Normal
	Increase	Decrease
IDH2	12	9
FH	7	0
MDH2	6	1
IDH1	5	1
OGDH	0	0

Numbers are from datasets containing TCA cycle gene mRNA expression (Oncomine).