Oncometabolite D-2-Hydroxyglutarate Impairs Α-Ketoglutarate Dehydrogenase and Contractile Function in Rodent Heart

Oncometabolite D-2-hydroxyglutarate impairs α-ketoglutarate dehydrogenase and contractile function in rodent heart

Anja Karlstaedta, Xiaotian Zhangb,c,d,e, Heidi Vitracf, Romain Harmanceyg,h,i, Hernan Vasqueza, Jing Han Wangj, Margaret A. Goodellb,c,d,e,1, and Heinrich Taegtmeyera,1,2

aDepartment of Internal Medicine, Division of Cardiology, McGovern Medical School, The University of Texas Health Science Center, Houston, TX 77030; bDepartment of Molecular and Human Genetics, Baylor College of Medicine, Houston, TX 77030; cCenter for Cell and Gene Therapy, Baylor College of Medicine, Houston, TX 77030; dStem Cells and Regenerative Medicine Center, Baylor College of Medicine, Houston, TX 77030; eDepartment of Pediatrics, Baylor College of Medicine, Houston, TX 77030; fDepartment of Biochemistry and Molecular Biology, McGovern Medical School, The University of Texas Health Science Center, Houston, TX 77030; gDepartment of Physiology and Biophysics, University of Mississippi Medical Center, Jackson, MS 39216; hMississippi Center for Obesity Research, University of Mississippi Medical Center, Jackson, MS 39216; iCardiovascular-Renal Research Center, University of Mississippi Medical Center, Jackson, MS 39216; and jDepartment of Hematology, The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou 310003, China

Edited by Gregg L. Semenza, Johns Hopkins University School of Medicine, Baltimore, MD, and approved July 19, 2016 (received for review January 29, 2016) Hematologic malignancies are frequently associated with cardiac The starting point for the present work were reports that my- pathologies. Mutations of isocitrate dehydrogenase 1 and 2 (IDH1/2) eloid malignancies are associated with cardiac pathologies, which occur in a subset of acute myeloid leukemia patients, causing are commonly considered a side effect of chemotherapy (7). Other metabolic and epigenetic derangements. We have now discovered recent reports suggest that systemically produced D2-HG by that altered metabolism in leukemic cells has a profound effect on neomorphic IDH2 is associated with cardiomyopathy, suggesting cardiac metabolism. Combining mathematical modeling and in vivo that D2-HG influences cardiac cellular responses (8, 9). However, as well as ex vivo studies, we found that increased amounts of the the extent to which D2-HG can directly affect cardiac function oncometabolite D-2-hydroxyglutarate (D2-HG), produced by IDH2 and metabolism, and which processes are involved, has remained mutant leukemic cells, cause contractile dysfunction in the heart. unknown. It is known, however, that the heart adapts to stress by This contractile dysfunction is associated with impaired oxidative remodeling, both metabolically and structurally. Similar to cancer SYSTEMS BIOLOGY decarboxylation of α-ketoglutarate, a redirection of Krebs cycle in- cells, metabolic remodeling in the heart is characterized by a shift termediates, and increased ATP citrate lyase (ACL) activity. Increased from fatty acid utilization toward glucose utilization and, ulti- availability of D2-HG also leads to altered histone methylation and mately, mitochondrial dysfunction. This remodeling involves acetylation in the heart. We propose that D2-HG promotes cardiac changes in pathways that regulate energy and redox homeostasis, α dysfunction by impairing -ketoglutarate dehydrogenase and in- growth, and autophagy, resulting in altered enzyme activities (10, duces histone modifications in an ACL-dependent manner. Collec- 11). Consequently, we proposed that D2-HG mediates metabolic tively, our results highlight the impact of cancer cell metabolism on stress in the heart, and we tested this hypothesis using a targeted function and metabolism of the heart. multiomics approach, together with the predictive and integrative value of mathematical modeling. D-2-hydroxyglutarate | IDH2 | metabolism | cardiomyopathy | flux rate analysis Results D2-HG Promotes Cardiac Remodeling. To investigate the impact etabolic dysregulation in cancer cells changes the way nu- of Idh2 mutations on cardiac remodeling, we generated mice R140Q Mtrients are consumed and macromolecules are produced to bearing hematopoietic cells with an Idh2 mutation, which meet the increased demands for cell growth. Somatic mutations in IDH1/2 isocitrate dehydrogenase 1 and 2 ( )arecommonandare Significance described in several cancer types (i.e., gliomas and acute myeloid leukemia). IDH mutations lead to increased production and We show that the oncometabolite D-2-hydroxyglutarate (D2- accumulation of the oncometabolite D-2-hydroxyglutarate (D2- HG) affects cardiac function in the isolated working heart by HG) through a neomorphic enzymatic function (1). WT IDH1/2 α α inhibiting -KGDH, a key regulatory enzyme of cellular energy catalyzes the oxidative decarboxylation of isocitrate to -keto- metabolism. Analyzing metabolic flux rates by using in vitro α + glutarate ( -KG), while reducing NADP to NADPH either in and ex vivo approaches in combination with integrative math- the cytosol and peroxisome (IDH1), or in mitochondria (IDH2). ematical modeling enabled us to identify the mechanisms by In this reaction, D2-HG is produced in small amounts but con- which D2-HG perturbs metabolic flux and induces epigenetic verted back to its structural homolog α-KG by D2-HG dehydro- modifications in the heart. The results provide knowledge about genase. Common features of tumors with IDH1/2 mutations are malignancy-related changes in enzymatic activity and post- abnormal histone and DNA methylation, connecting metabolic translational modifications in the context of cardiac remodeling. changes with epigenetic control of gene expression (2). In hematologic malignancies, IDH1/2 are often co-mutated with epigenetic Author contributions: A.K., M.A.G., and H.T. designed research; A.K., X.Z., H. Vitrac, R.H., regulatory genes encoding enzymes that are important in DNA and H. Vasquez performed research; H. Vitrac, J.H.W., and M.A.G. contributed new TET2 reagents/analytic tools; A.K., X.Z., M.A.G., and H.T. analyzed data; and A.K., X.Z., M.A.G., hydroxymethylation (i.e., tet methylcytosine dioxygenase 2, ) and H.T. wrote the paper. DNMT3A and methylation (i.e., DNA methyltransferase 3 A, )(3). The authors declare no conflict of interest. Accumulation of D2-HG contributes to leukemogenesis, likely due This article is a PNAS Direct Submission. α – to inhibition of -KG dependent dioxygenases, including histone 1M.A.G. and H.T. contributed equally to this work. lysine demethylases (KDMs) and TET2 (4). This hypothesis has 2To whom correspondence should be addressed. Email: [email protected]. been supported by recent reports linking the hypermethylation edu. phenotype in cancer cells to IDH, fumarate hydratase, and succi- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. nate dehydrogenase mutations (5, 6). 1073/pnas.1601650113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1601650113 PNAS Early Edition | 1of6 Downloaded by guest on October 1, 2021 IDH1/2 600 mimics one of the most common mutations in acute Epinephrine and A B ** myeloid leukemia (AML) patients. Wild-type (WT) C57BL/6 Increased Afterload (40%) 500 mice were lethally irradiated and reconstituted with hemato- 400 poietic stem/progenitor cells (HSPCs) transduced with a retro- D2-HG (0.5 to 1.0 mM) 300 R140Q 200 virus expressing either WT Idh2 or Idh2 generating WT L-Lactate (0.5 mM) * WT R140Q Idh2 Idh2 A no substrate Glucose (5 mM) 100 control ( ) or single-mutant ( )HSPCs(Fig. S1 ). D2-HG tissue Time 55 0 We found no difference in the survival rate between mice (min) 0 20 45 75 D2HG 00.51.0 WT R140Q content (μg/g dry wt.) with Idh2 and mice with Idh2 (Fig. S1B). We used liquid (mM) – chromatography mass spectrometry (LC-MS) to measure serum C Control D Control total 2-hydroxyglutarate (2-HG) in mice with neomorphic IDH2 D2-HG (0.5 mM) D2-HG (0.5 mM) R140Q D2-HG (1.0 mM) D2-HG (1.0 mM) (Idh2 ) and WT control HSPCs 6 mo after bone marrow 15.0 1.5 * * transplantation (BMT). The serum total 2-HG markedly in- 12.5 creased fivefold in all mice transplanted with BM cells over- ** ** 10.0 1.0 expressing mutant Idh2 compared with controls (Fig. S1C). A ** previous report indicated that overproduction of D2-HG, caused 7.5 by a broadly expressed mutant IDH2, is associated with cardiac 5.0 0.5 2.5

hypertrophy, dilatation, and failure (8). We found no cardiac mol/min/g dry wt.)

WT μ Hydraulic Power (mW) Hydraulic Power Idh2 0.0 ( 0.0 hypertrophy in transplanted mice with HSPCs 6 mo after acute prolonged rate Glucose oxidation acute prolonged BMT. In contrast, we observed substantial changes at the mo- Normal Stimulation R140Q Normal Stimulation lecular level in the hearts of mice with Idh2 HSPCs 6 mo α after BMT. In this group, levels of myosin-heavy-chain E α-KG F Pronase - 1:3000 G 1.5 (α-MHC) expression were decreased and myosin-heavy-chain β O O D2-HG (μM) 0 0 100 500 (β-MHC) expression levels were increased ([β-MHC]:[α-MHC] D–F HO OH α-KGDH ratio increased) (Fig. S1 ). We and others have previously O 1.0 observed these same isoform changes in both hypertrophied and D2-HG α-KGDH/α-Actinin 1.8 2.3 2.5 atrophied heart, suggesting cardiac remodeling (10–13). A shift O O α-KGDH/GAPDH 1.9 2.2 2.9 in gene expression of MHC isoforms occurs only in response to ATP5B 0.5 HO OH

ATP5B/α-Actinin -KGDH activity stress and is dependent on the severity and duration of the stress. OH 2.3 3.0 3.1 α ** In other words, our findings support the hypothesis that D2-HG ATP5B/GAPDH 2.5 2.9 3.6 ** α-Actinin

mediates metabolic stress, causing cardiac remodeling in mye- (nmol NADH/min/mg protein) 0 loid malignancies. GAPDH D2-HG 00.51.0 (mM) D2-HG Impairs Cardiac Energy Substrate Metabolism. To understand whether overproduction of D2-HG alone was responsible for the H IJ effects observed in the Idh2 mutant mouse model, we measured 1.5 1.5 0.8 rates of substrate metabolism in the isolated working rat heart ** ** ** 0.6 1.0 1.0 and conducted computational flux rate analysis using the Car- ** **

(5) - FCCP) 0.4

dioNet model of mammalian cardiac metabolism (14). Rat 3 0.5 NS 0.5 hearts were perfused ex vivo in the presence or absence of D2- 0.2 production rate production /min/mg protein)

HG in concentrations similar to those found in the plasma of 2 2

R140Q O

0.0 O 0 0.0 Idh2 mutant mice (0.5 mM) and AML patients (8, 15–17) 2 2

H D2-HG 00.51.0 D2-HG 00.51.0 D2-HG 00.51.0 (nmol/min/mg protein) and those reported by Latini et al. (18) to promote inhibition of Catalase activity (mM) (mM) Membrane Potential (mM) ATP synthase in cardiac muscle in vitro (range 0.05–5 mM; F0/F1 μmol H ( ATP synthase Ki = 0.47 ± 0.18 mM) (Fig. 1A). Pretreatment DiSC (Normalized levels of serum 2-HG in AML patients range from 19 to Fig. 1. D2-HG impairs cardiac energy metabolism by inhibiting α-KGDH. 96,000 ng/mL (15–17). Glucose and lactate were the only other (A) Protocol for the isolated working rat heart with (0.5 mM or 1.0 mM) or substrates provided. The stressed heart (e.g., adrenergic stimula- without D2-HG. (B) LC-MS analysis of D2-HG concentration in hearts perfused tion) shifts to increased oxidation of glucose, whereas in the fasted with (0.5 mM or 1.0 mM) or without D2-HG (n = 3). (C and D) Hydraulic power – state the heart oxidizes predominantly fatty acids for energy pro- (C) and glucose oxidation rate (D) at near-physiologic (100 cmH2O, 45 55 min) – vision (19). Importantly, we limited the substrate supply to car- and increased workload (140 cmH2O, acute stimulation 55 58 min, prolonged – α bohydrates to assess any potential de novo fatty acid synthesis. stimulation 65 75 min). (E) Chemical structures of -KG and D2-HG. (F) DARTs blotting showing D2-HG as a substrate of α-KGDH and ATP5B. Susceptibility of D2-HG was taken up by the perfused heart at a constant rate both α-KGDH and ATP5B to pronase digestion is increased in the presence of and accumulated in the tissue (Fig. 1B and Fig. S2A). The D2- – α ± μ D2-HG. (G I) Effect of D2-HG on -KGDH activity (G), H2O2 production rate (H), and HG tissue content was 110 48 g/g dry weight (0.5 mM D2-HG catalase activity (I) in mitochondria isolated from hearts perfused with or with- ± μ group) and 442 65 g/g dry weight (1.0 mM D2-HG group). out D2-HG. (J) MMP assessed by 3,3′-dipropylthiadicarbocyanine iodide (DiSC35) With high D2-HG levels, we observed a significant decrease in staining and corrected by carbonyl cyanide-4-(trifluoromethoxy)phenyl- cardiac power before and after an imposed increase in cardiac hydrazone (FCCP) of mitochondria isolated from hearts perfused with or work (Fig. 1C). However, myocardial oxygen consumption was without D2-HG. n = 3 rats per group. All data shown are mean ± SEM. the same in all groups, likely due to inefficient oxidative phos- Statistical analyses were performed with Kruskal–Wallis test, ANOVA, and ’ < < phorylation of ADP, as indicated by the decline in hydraulic Student s t test. *P 0.05; **P 0.01; NS, not significant. power and cardiac efficiency (Fig. S2 B and C). Next, we determined the effect of D2-HG on glucose oxidation and mea- 14 14 provision through inhibition of ATP synthase by D2-HG both sured CO2 production from D-[U- C]glucose (19). In perfusions with 1.0 mM D2-HG, glucose oxidation rates increased signifi- in vivo and in vitro (18, 20). To assess whether D2-HG differen- cantly at high workloads (Fig. 1D). At the same time ATP levels tially affects energy substrate metabolism in the heart, we perfused were reduced and AMP levels were increased ([ATP]:[AMP] ratio rat hearts (n = 4 animals per group) with or without D2-HG decreased) (Fig. S2D). The results are consistent with prior re- (1 mM) in presence of glucose (5 mM) and oleate (0.4 mM) (Fig. ports that an increased supply of D2-HG leads to impaired ATP S3A). Likewise, D2-HG accumulated in the tissue of perfused rat

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1601650113 Karlstaedt et al. Downloaded by guest on October 1, 2021 hearts (Fig. S3B). At normal workload we observed after 5 min 0.98) between model simulations and flux rate measurements of perfusion with D2-HG a steady decrease in cardiac power (Fig. S4A). Predicted flux rates were used to calculate the con- (Fig. S3C). Hearts perfused with D2-HG showed a decreased tribution of glycolysis, β-oxidation of fatty acids, and oxidative adrenergic response upon stimulation with epinephrine and sig- phosphorylation to ATP provision. Increased D2-HG supply nificant decline in cardiac power. The oxidation of both glucose caused a decrease in ATP synthesis, with most ATP being pro- and oleate was significantly reduced in presence of D2-HG (Fig. vided oxidatively (Fig. S4B). Glucose oxidation was the main S3D). We now found that with deficient oxidative phosphoryla- source of ATP with or without D2-HG. However, ATP provision tion at high D2-HG concentration cardiac metabolism shifted from glycolysis and β-oxidation of endogenous fatty acids was toward glycolysis, as demonstrated by higher glucose uptake and increased in simulations with D2-HG supply. D2-HG caused a increased lactate release (Fig. S3E). shift in cardiac metabolism, increasing the reliance on endogenous substrates under experimental conditions (Fig. 2). D2-HG Inhibits α-KG Dehydrogenase Activity. Next, we hypothesized We identified metabolic differences by performing pairwise that D2-HG, as a structural homolog to α-KG (Fig. 1E), can bind analysis of estimated flux rates with or without D2-HG supply (P < to α-KG dehydrogenase (α-KGDH). Using drug-affinity respon- 0.05). Metabolic reactions and their metabolic subsystems, classi- sive target stability (DARTS) (21), an unbiased molecular ap- fied in the Kyoto Encyclopedia of Genes and Genomes database proach, we found decreased protease susceptibility of α-KGDH (24), are presented (Fig. S5 A and B). Increased supply of D2-HG and ATP5B, the beta subunit of the catalytic core of the ATP is associated with distinct metabolic alterations, including in- synthase, in the presence of D2-HG in a concentration-dependent creased β-oxidation of short chain fatty acids and degradation of manner (Fig. 1F). This decreased susceptibility suggests that D2- amino acids. These reactions provided intermediates directly HG binds to and stabilizes both α-KGDH and ATP5B, affecting feeding into the Krebs cycle through succinyl-CoA, fumarate, and their activity, as evidenced by the reduced ATP levels and con- oxaloacetate. For example, mitochondrial provision of oxaloace- sistent with previous reports (18, 20). We measured the level of tate through malate dehydrogenase (MDH) in the Krebs cycle α-KGDH activity and production rate of hydrogen peroxide decreased by 0.65-fold (Fig. 2). Further, D2-HG caused increased (H2O2) fluorometrically in the ex vivo perfused rat hearts. There flux in the D2-HG dehydrogenase reaction and conversion of D2- was a distinct reduction in α-KGDH activity at both 0.5 mM and HG to α-KG. Impairment of the Krebs cycle at the α-KGDH 1.0 mM D2-HG (Fig. 1G and Fig. S3F), indicating an impaired reaction shifted the model toward increased production of citrate NADH generation in the Krebs cycle and increased reductive through the NADPH-dependent reverse function of IDH2 (Fig. 2). carboxylation. H2O2 was produced at the same rate in both control This shift, in turn, increased fluxes for the transport of citrate into SYSTEMS BIOLOGY and at 0.5 mM D2-HG perfused hearts (Fig. 1H). However, H2O2 the cytosol and conversion of citrate to acetyl-CoA and oxaloac- formation rose markedly at high extracellular D2-HG concentra- etate via the ATP citrate lyase (ACL). Our simulations indicate tions (1.0 mM), implying that D2-HG impairs mitochondrial that D2-HG–dependent inhibition of α-KGDH shifts cardiac me- membrane potential (MMP) and induces generation of reactive tabolism toward increased reductive carboxylation, which increases oxygen species (ROS). To assess the activation of ROS defense the reliance on glucose oxidation and endogenous substrates to mechanisms, we determined the catalase activity colorimetrically maintain flux through the Krebs cycle. in isolated mitochondria from ex vivo perfused rat hearts. Catalase activity was increased at both 0.5 mM and 1.0 mM D2-HG (Fig. D2-HG Promotes Metabolic and Epigenetic Alterations in the Heart. 1I), indicating that D2-HG promotes an acute rise in ROS gen- Based on the model predictions, we determined the impact of D2- eration and cellular adaptation. The production of ROS correlates HG on the cytosolic and mitochondrial redox states. We measured with the MMP. A slight decrease in the membrane potential can cause a significant decrease in ROS production (22). We found the MMP markedly decreased in D2-HG perfused rat hearts (Fig. Extracellular < 0.5 0.5 - 0.8 1J). The diminished MMP impairs the ability of perfused hearts to Glucose no change Glut 71.8 1.3 - 1.6 provide ATP by oxidative phosphorylation. Together, these data 1.41 α-KG 1.6 - 2 α Ac-CoA >2 indicate that the observed changes in -KGDH activity were in- G6P Asp duced by D2-HG rather than a rise in H2O2 and show that D2-HG 200 OAA impairs Krebs cycle function and mitochondrial electron transport. 1.41 ATP 200 1.47 200 Lactate Pyruvate Citrate Mal Computational Estimation of Metabolic Changes in the Presence of Glut 0.5 200 Isocitrate D2-HG D2-HG. To understand the effect of D2-HG and the impact of CO 200 α 14 2 200 -KGDH inhibition on metabolic reactions in the heart, we next Tryp Ac-CoA Citrate 200 performed computational simulations by flux balance analysis α-KG CO2 (23). We applied this algorithm to identify the metabolic reac- SCFAs 200 1.59 0.29 CO2 tions involved in the metabolic adaptation to high levels of OAA 100 D2-HG. To estimate flux rates, we used uptake and release rates Succ-CoA Valine of D2-HG, glucose, oxygen, and lactate, which were determined 0.65 0.65 during the isolated working rat heart perfusions. We combined Mal Succ these measurements with previously reported rates for glycogen, 0.25 0.65 Fum protein, and lipid turnover to constrain the flux bounds of associated reactions during calculations (Table S1). Furthermore, 64.6 57 we incorporated the experimentally measured decrease in α-KGDH Asp Try activity to reflect the inhibition caused by D2-HG in the com- putations. We used the metabolic model of mammalian cardiac Fig. 2. Flux rate analysis reveals dysregulation of cardiac energy substrate metabolism, CardioNet (14), and flux balance analysis to identify metabolism with increased D2-HG supply. Schematic of in silico flux rate flux distributions that would optimally fit the experimental data- analysis for glucose and D2-HG metabolism in ex vivo working heart perfusions. Colors indicate flux changes in the presence of D2-HG compared sets. The objective of the optimization problem was to maximize with control conditions. Ac-CoA, acetyl-CoA; Asp, aspartate; Fum, fumarate; ATP hydrolysis to reflect cardiac work. To validate the model Glut, glutamate; Homocys, homocysteine; Mal, malate; OAA, oxaloacetate; predictions, we compared estimated glucose oxidation rates with SAM, S-adenosylmethionine; Succ, succinate; Succ-CoA, succinyl-CoA; Tryp, 2 tracer measurements and found a high correlation (R = 0.9– tryptophane; and 5,10-Met-THF, 5,10-methenyl-tetrahydrofolate.

Karlstaedt et al. PNAS Early Edition | 3of6 Downloaded by guest on October 1, 2021 the [pyruvate]:[lactate] and the [acetoacetate]:[β-hydroxybutyrate] HATs increased when exposed to D2-HG in a concentration- ratios (25). The conversion of pyruvate to lactate and the forma- dependent manner (Fig. S6D), suggesting that the level of D2- tion of β-hydroxybutyrate to acetoacetate increased in the pres- HG affects the histone acetylation state of cardiomyocytes. We ence of D2-HG (Fig. S6 A and B). The differential expression also assessed the ability of D2-HG to induce histone acetylation pattern corresponds to a decreased cytosolic and mitochondrial in histone extractions from perfused hearts. Pan-acetylation of + [NAD ]:[NADH] ratio. We further quantified citrate, α-KG, and histone 3 increased in D2-HG–perfused hearts compared with succinate levels and determined the glucose-derived triglyceride controls, whereas histone 3 K9 acetylation (Ac-H3K9) and tri- turnover by measuring 14C labeling in extracted lipids from per- methylation (Me3-H3K9) decreased (Fig. S6E). Thus, the bal- fused hearts. Increased amounts of D2-HG significantly increased ance between histone acetylation and deacetylation shifted the incorporation of glucose-derived glycerol into lipids (Fig. S6C). toward increased histone deacetylation in isolated working rat Decreased α-KGDH activity suggested that Krebs cycle inter- heart perfusions with D2-HG. The observed effects can most mediates might not be fully oxidized in the heart. Indeed, the likely be attributed to the relatively short perfusion time with D2- levels of both citrate and succinate were elevated, whereas α-KG HG (30 min). levels were decreased, resulting in a decreased [α-KG]:[succinate] + ratio (Fig. 3 A and B). NAD -dependent decarboxylation of pyruvate Prolonged D2-HG Supply Promotes Unique Metabolic and Epigenetic to acetyl-CoA by pyruvate dehydrogenase is followed by the con- Changes. To test whether the D2-HG–dependent metabolic and densation of acetyl-CoA with oxaloacetate to form citrate, which is, epigenetic changes in the heart also exist in vivo, we injected WT in turn, transported into the cytosol (26). In the cytosol, citrate is mice daily for 32 d with PBS (control, 0.2 mL) or D2-HG converted to acetyl-CoA and oxaloacetate by ACL. In silico mod- (250 mg D2-HG/ kg body weight). We observed significant eling suggested an increased flux rate in presence of D2-HG. skeletal muscle atrophy and reduction in total body weight (Fig. Therefore, we determined ACL and MDH activity in perfused hearts 4A and Fig. S7A). The serum D2-HG level was 657 ± 164 ng/mL freeze-clamped at the end of the protocol. As expected, high levels of in controls and 3,962 ± 356 ng/mL in with D2-HG–treated mice, D2-HG (1.0 mM) increased ACL activity by 37% and MDH activity thus within the range of reported serum D2-HG levels from by 58% compared with controls (Fig. 3 C and D). These results AML patients (Fig. S7B). Next, we perfused mouse hearts from suggest an increased conversion of citrate to acetyl-CoA and oxa- those groups ex vivo to assess the metabolic consequences of loacetate in the presence of D2-HG (Fig. 3E). Previous studies prolonged exposure to D2-HG on the heart. Consistent with the have shown that changes in metabolite concentrations of Krebs perfused rat hearts, we observed increased concentrations of D2- cycle intermediates (e.g., citrate and succinate) and ACL activity HG in the hearts of mice injected with D2-HG (Fig. S7B). affect epigenetic mechanisms regulating the posttranslational Glucose was oxidized at a higher rate in hearts from D2-HG– modification of the chromatin-modifying machinery (27, 28). treated mice, suggesting that continuous D2-HG supply causes a To test whether the observed changes in ACL activity from shift of cardiac metabolism toward higher glucose utilization D2-HG–perfused hearts affected the acetylation and methyl- rates (Fig. 4B). We also found a marked reduction in α-KGDH ation of histones, we determined the activity of global histone activity in the chronic exposure model, whereas there was no acetyltransferases (HATs), and protein levels of acetylated and difference in H2O2 formation and catalase activity between methylated histone 3. Enzymatic assays performed on nuclear control and D2-HG–treated animals (Fig. S7 C–E). The results extracts from the perfused hearts revealed that the activity of suggest that glucose-derived pyruvate is decarboxylated to acetyl- CoA, promoting increased citrate synthesis. Under these conditions, ROS generation rises acutely, and with chronic exposure to D2-HG in vivo antioxidant defense mechanisms seem to be ac- A B tivated. We also asked whether prolonged exposure to D2-HG Citrate α-KG Succinate 1.75 0.08 0.75 0.10 affected the activity of ACL and MDH, and histone methylation * ** 1.50 0.60 0.08 and acetylation. Consistent with measurements from rat heart 1.25 0.06 * perfusions, treating animals for 4 wk with D2-HG leads to a 1.00 0.45 0.06 0.04 ** ** NS decreased [α-KG]:[succinate] ratio and a rise in ACL and MDH 0.75 NS 0.30 0.04 ** C D F

0.50 μmol/g dry wt.) activity (Fig. 4 and and Fig. S7 ). Furthermore, we observed 0.02 0.15 0.02 0.25 that with D2-HG both the activity of HATs (Fig. S7G) and pan- 0.00 0.00 0.00 0.00

D2-HG 00.51.0 α-KG (μmol/g dry wt.) D2HG 0 0.51.0 D2HG 00.51.0 D2HG 0 0.51.0 acetylation of histone 3 (Ac-H3) increased, whereas specifically [α-KG]:[Succinate] ratio Citrate (μmol/g dry wt.) (mM) (mM) (mM) (mM) H3K9 acetylation (Ac-H3K9) and trimethylation decreased (M3- Succinate ( H3K9) (Fig. 4E). Increased demethylation of H3K9 resulted in a – CDE decreased [Ac-H3K9]:[Me-H3K9] ratio in D2-HG treated hearts compared with controls. The results suggest an increased turnover 25 16 FADH2 ** ** rate for H3K9 deacetylation and demethylation during short 20 FAD D2-HG NS 12 periods of increased D2-HG supply, as observed in the isolated 15 NS 8 α-KGDH working rat heart perfusions. However, prolonged supply of D2- 10 α-KG Succinyl-CoA + HG redirects pyruvate toward citrate and acetyl-CoA and conse- 5 4 NAD NADH mito quently increases histone acetylation. Our data provide evidence 0 0 cyto D2HG 0 0.51.0 ACL that chronic exposure of the heart to D2-HG leads to reductive D2HG 00.51.0 Citrate Acetyl-CoA (mM) (mM) carboxylation, which contributes to both increased intracellular Oxaloacetate citrate concentrations and modifications of the cardiac epige- ATP citrate lyase activity ATP

(nmol NADH/min/mg protein) F (nmol NADH/min/mg protein) nome (Fig. 4 ). Malate dehydrogenase activity Malate dehydrogenase

Fig. 3. Perfusion with D2-HG redirects Krebs cycle intermediates. (A and B) Discussion α α Analysis of citrate, -KG, and succinate concentrations (A)and[ -KG]:[succinate] Metabolic alterations in AML caused by IDH2 mutations initiate ratio (B) in perfused hearts freeze-clamped at the end of the protocol. (C and D) the production of a small molecule, D2-HG, which is a structural Effect of D2-HG on ACL (C)andMDH(D) activity in hearts perfused with or α without D2-HG. (E) Schematic overview of ACL and α-KGDH reaction in the homolog to the Krebs cycle intermediate -KG. We show that cytosol (cyto) and mitochondria (mito). In A–D, n = 3 rats per group; data are IDH2 mutation in HSPC contributes to accelerated serum total mean ± SEM. ANOVA and Student’s t test. *P < 0.05; **P < 0.01; NS, not 2-HG production, and that this increase promotes cardiac remod- significant. eling as indicated by the shift of α-MHC to β-MHC expression.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1601650113 Karlstaedt et al. Downloaded by guest on October 1, 2021 A BCD ( µ mol/min/g dry wt.) Glucose oxidation rate Glucose oxidation

EFFig. 4. Mice injected with D2-HG exhibit metabolic and epigenetic alterations in the heart. (A)Body weight and heart weight to tibia length ratio in mice after 32 d of PBS (control, Cnt) or D2-HG injection. 14 (B) Rates of glucose oxidation (D-[U- C]glucose) in hearts from mice treated with or without D2-HG. (C and D) Measurement of ACL (C)andMDH(D)activity in heart tissue from mice treated with PBS (control, Cnt) or D2-HG. (E) Analysis of global histone 3 acetylation (Ac-H3), lysine 9 histone 3 acetylation (Ac-H3K9), lysine 9 histone 3 trimethylation (M3-H3K9), and [Ac-H3K9]:[M3-H3K9] ratio in histone extractions from heart tissue in mice treated with PBS (control, Cnt) or D2-HG. (F) Schematic summarizing the proposed concept of metabolic and epigenetic modifications in the heart. In A–E, n = 10 mice per group. In F–H, n = 8 mice per group. Data expressed as mean ± SEM. Student’s t test. *P < 0.05, **P < 0.01;

NS, not significant. SYSTEMS BIOLOGY

Thus, increased levels of D2-HG impose metabolic stress not MMP can be associated with ROS production and indicates a only in cancer cells but also in other tissues, driving adaptive dysfunction of respiratory chain components (30). We observed in changes in cellular processes. We demonstrate that cardiac dys- the isolated working rat heart that D2-HG supply decreased the function can be initiated directly by D2-HG through inhibition of MMP, which reduces the electron transfer across the mitochon- α-KGDH, which impairs oxidative phosphorylation and promotes drial membrane and decreases the ability to provide ATP by ox- compensatory epigenetic modifications in the heart. idative phosphorylation. This decrease in MMP was most likely In the isolated working rat heart, D2-HG initiates cardiac caused by several mechanisms including deficiency of oxidizable dysfunction and metabolic alterations. We focused on glucose substrates, due to impaired α-KGDH activity, and the uncoupling and lactate as energy-providing substrates because we were in- of the inner membrane, as indicated by the rise in ROS pro- terested in assessing the possible effects of increased D2-HG duction. Prolonged exposure to D2-HG in mice did not display the supply on lipid remodeling, which cannot be assessed in the increase in ROS observed in perfusions with D2-HG for 30 min, presence of fatty acids. Our results indicate that in response to and thus catalase activation and MMP decrease were sufficient to increased levels of D2-HG cardiac metabolism shifts toward in- normalize ROS production over time. Computational flux rate creased glycolysis and intermediary metabolites such as pyruvate analysis indicates that specific metabolic processes are part of a and citrate are redirected into pathways other than the Krebs response system to meet cellular energy demands to maintain cycle. These processes are driven by both decreased α-KGDH cardiac output. Our findings suggest that, in conditions of in- activity and decreased production of reducing equivalents in the creased D2-HG, citrate and acetyl-CoA formation is enhanced by form of NADH. D2-HG is known to be a potent inhibitor of F0/F1 two converging processes: (i) the decarboxylation of glucose to ATP synthase (18) and mediator of histone methylation modifica- pyruvate driving acetyl-CoA formation through ACL and (ii)re- tions by inhibiting α-KG–dependent dioxygenases (e.g., KDMC4) in ductive carboxylation of α-KG by the reverse function of mito- IDH2 mutant cells (4, 5). chondrial IDH (6). Reductive carboxylation requires an intracellular + We have now demonstrated that D2-HG also perturbs mito- reducing environment (low [NAD(P) ]:[NAD(P)H] ratio) to gen- chondrial metabolism in the heart by inhibiting α-KGDH and erate NAD(P)H for the conversion of D2-HG to α-KG. Our decreasing the MMP. Analysis of substrate binding by DARTS results suggest that prolonged exposure to D2-HG shifts cardiac demonstrated that D2-HG binds to α-KGDH, and that this redox homeostasis toward NAD(P)H oxidation. binding leads to decreased α-KGDH activity and contributes to a Dynamic regulation of histone modification is critical to the + decreased mitochondrial [NAD ]:[NADH] ratio. The regenera- maintenance of cellular processes in response to stress (31). D2- tive function of many antioxidative and ROS-scavenging enzymes HG affects histone methylation by inhibiting α-KG–dependent requires NADH or NADPH, and thus changes in the redox state dioxygenases, resulting in epigenetic changes and promoting can affect the antioxidant capacity of cells. In the isolated working tumorigenesis in cancer cells (4). Prolonged systemic exposure to rat heart, we show that increased D2-HG supply is associated with D2-HG in mice altered histone methylation and acetylation in increased mitochondrial ROS production. A corresponding in- the heart, specifically increasing pan-acetylation of histones and crease in catalase activity confirmed the cellular adaptation to H3K9 demethylation. Recent studies have shown a link between increased mitochondrial ROS. Previous studies in isolated mito- histone trimethylation status and the development of cardiac + chondria demonstrated that the mitochondrial [NAD ]:[NADH] hypertrophy, corroborating our conclusion that D2-HG acts as a ratio and ROS production are positively correlated with the MMP metabolic signal promoting metabolic and structural adaptation (29). However, ROS changes can also be transient. A drop in in the heart. ACL-dependent production of acetyl-CoA has been

Karlstaedt et al. PNAS Early Edition | 5of6 Downloaded by guest on October 1, 2021 shown to contribute to increased histone acetylation (28). Fur- Methods ther, the activity of ACL and HATs synergize to maintain global Animals. All animal experiments were approved by the Institutional Animal histone acetylation and regulate chromatin structure by changes Care and Use Committee and conducted according to the guidelines issued in the generation of acetyl-CoA. We now show that activities of by The University of Texas Health Science Center at Houston and by Baylor ACL and HATs are increased in the presence of D2-HG. Thus, College of Medicine, respectively. Animals were fed a standard laboratory it seems that D2-HG–induced alterations in cardiac substrate chow (LabDiet 5001; PMI Nutrition International). All rats were male Spra- utilization and energy metabolism are linked to changes in the gue-Dawley rats obtained from Harlan Laboratories, and all mice were with acetylation of histones by ACL activity. a C57BL/6 background, obtained from Charles River Laboratories. In conclusion, we demonstrate that cardiac dysfunction is α initiated directly by D2-HG through inhibition of α-KGDH, Metabolic Assays. -KGDH activity and H2O2 production rate were measured promoting both impairment of oxidative phosphorylation and fluorometrically as described by Starkov et al. (34). ACL activity was de- compensatory epigenetic modifications. We offer insight into the termined as described by Srere (35). Metabolite concentrations were synergy of epigenetic and metabolic modifications caused by assessed colorimetrically using established enzymatic assays as described D2-HG in hematologic malignancies, which underscores the in- previously (19). Further detailed information on reagents and antibodies, teraction of tumor metabolism and heart metabolism. The in- isolated working rat heart and Langendorff mouse perfusions, in silico teraction between metabolism and epigenetics is of particular analysis of metabolic flux rates, Western blotting, and other methods is interest in light of recent investigations showing that the use of given in Supporting Information. hypomethylating agents and histone deacetylase inhibitors may be ACKNOWLEDGMENTS. We thank Dolly A. Fernandez Palomino and Patrick beneficial in the treatment of cancer and heart failure (32, 33). Guthrie for help during the final stages of the work and Nataliya Bulayeva However, our study is limited to fully elucidate all of the molecular at the Clinical Laboratory Improvement Amendments Molecular Diagnos- mechanisms linking metabolic dysregulation with structural tics Laboratory and Arun Sreekumar at the Baylor College of Medicine remodeling in the heart in response to elevated D2-HG levels. Metabolomics Core Facility for help with the LC-MS analysis. This work was supported by the Friede Springer Herz Stiftung (A.K.), the Roderick Further studies on the mechanisms leading to metabolic and epi- MacDonald Research Fund (A.K.), NIH Grants R01-HL-61483 (to H.T.) and genetic dysregulation may offer new opportunities to treat cancer K99/R00-HL-112952 (to R.H.), and The Adrienne Helis Malvin Medicial and prevent (rather than cause) heart failure at the same time. Research Foundation (M.A.G.).

1. Ward PS, et al. (2010) The common feature of leukemia-associated IDH1 and IDH2 22. Korshunov SS, Skulachev VP, Starkov AA (1997) High protonic potential actuates a mutations is a neomorphic enzyme activity converting alpha-ketoglutarate to 2- mechanism of production of reactive oxygen species in mitochondria. FEBS Lett hydroxyglutarate. Cancer Cell 17(3):225–234. 416(1):15–18. 2. Figueroa ME, et al. (2010) Leukemic IDH1 and IDH2 mutations result in a hyper- 23. Holzhütter H-G (2004) The principle of flux minimization and its application to esti- methylation phenotype, disrupt TET2 function, and impair hematopoietic differen- mate stationary fluxes in metabolic networks. Eur J Biochem 271(14):2905–2922. tiation. Cancer Cell 18(6):553–567. 24. Ogata H, et al. (1999) KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic 3. Lawrence MS, et al. (2014) Discovery and saturation analysis of cancer genes across 21 Acids Res 27(1):29–34. tumour types. Nature 505(7484):495–501. 25. Williamson DH, Lund P, Krebs HA (1967) The redox state of free nicotinamide-adenine 4. Xu W, et al. (2011) Oncometabolite 2-hydroxyglutarate is a competitive inhibitor of dinucleotide in the cytoplasm and mitochondria of rat liver. Biochem J 103(2): – α-ketoglutarate-dependent dioxygenases. Cancer Cell 19(1):17–30. 514 527. 5. Xiao M, et al. (2012) Inhibition of α-KG-dependent histone and DNA demethylases by 26. Kauppinen RA, Hiltunen JK, Hassinen IE (1982) Compartmentation of citrate in re- fumarate and succinate that are accumulated in mutations of FH and SDH tumor lation to the regulation of glycolysis and the mitochondrial transmembrane proton suppressors. Genes Dev 26(12):1326–1338. electrochemical potential gradient in isolated perfused rat heart. Biochim Biophys – 6. Mullen AR, et al. (2014) Oxidation of alpha-ketoglutarate is required for reductive Acta 681(2):286 291. carboxylation in cancer cells with mitochondrial defects. Cell Reports 7(5):1679–1690. 27. Wang Q, et al. (2010) Acetylation of metabolic enzymes coordinates carbon source – 7. Jaiswal S, et al. (2014) Age-related clonal hematopoiesis associated with adverse utilization and metabolic flux. Science 327(5968):1004 1007. outcomes. N Engl J Med 371(26):2488–2498. 28. Wellen KE, et al. (2009) ATP-citrate lyase links cellular metabolism to histone acety- – 8. Akbay EA, et al. (2014) D-2-hydroxyglutarate produced by mutant IDH2 causes car- lation. Science 324(5930):1076 1080. 29. Gottlieb E, Armour SM, Harris MH, Thompson CB (2003) Mitochondrial membrane diomyopathy and neurodegeneration in mice. Genes Dev 28(5):479–490. potential regulates matrix configuration and cytochrome c release during apoptosis. 9. Kranendijk M, et al. (2010) IDH2 mutations in patients with D-2-hydroxyglutaric Cell Death Differ 10(6):709–717. aciduria. Science 330(6002):336. 30. Chacon E, Acosta D (1991) Mitochondrial regulation of superoxide by Ca2+:Anal- 10. Depre C, et al. (1998) Unloaded heart in vivo replicates fetal gene expression of ternate mechanism for the cardiotoxicity of doxorubicin. Toxicol Appl Pharmacol cardiac hypertrophy. Nat Med 4(11):1269–1275. 107(1):117–128. 11. Razeghi P, et al. (2001) Metabolic gene expression in fetal and failing human heart. 31. Kaneda R, et al. (2009) Genome-wide histone methylation profile for heart failure. Circulation 104(24):2923–2931. Genes Cells 14(1):69–77. 12. Harada K, Sugaya T, Murakami K, Yazaki Y, Komuro I (1999) Angiotensin II type 1A 32. Haberland M, Montgomery RL, Olson EN (2009) The many roles of histone deacety- receptor knockout mice display less left ventricular remodeling and improved survival lases in development and physiology: Implications for disease and therapy. Nat Rev after myocardial infarction. Circulation 100(20):2093–2099. Genet 10(1):32–42. 13. Miyata S, Minobe W, Bristow MR, Leinwand LA (2000) Myosin heavy chain isoform 33. McKinsey TA (2012) Therapeutic potential for HDAC inhibitors in the heart. Annu Rev expression in the failing and nonfailing human heart. Circ Res 86(4):386–390. Pharmacol Toxicol 52:303–319. 14. Karlstädt A, et al. (2012) CardioNet: A human metabolic network suited for the study 34. Starkov AA, et al. (2004) Mitochondrial alpha-ketoglutarate dehydrogenase complex of cardiomyocyte metabolism. BMC Syst Biol 6:114. generates reactive oxygen species. J Neurosci 24(36):7779–7788. 15. DiNardo CD, et al. (2013) Serum 2-hydroxyglutarate levels predict isocitrate de- 35. Srere PA (1959) The citrate cleavage enzyme. I. Distribution and purification. J Biol hydrogenase mutations and clinical outcome in acute myeloid leukemia. Blood Chem 234:2544–2547. – 121(24):4917 4924. 36. Challen GA, et al. (2011) Dnmt3a is essential for hematopoietic stem cell differenti- 16. Janin M, et al. (2014) Serum 2-hydroxyglutarate production in IDH1- and IDH2- ation. Nat Genet 44(1):23–31. mutated de novo acute myeloid leukemia: A study by the Acute Leukemia French 37. Taegtmeyer H, Hems R, Krebs HA (1980) Utilization of energy-providing substrates in – Association group. J Clin Oncol 32(4):297 305. the isolated working rat heart. Biochem J 186(3):701–711. 17. Borger DR, et al. (2014) Circulating oncometabolite 2-hydroxyglutarate is a potential 38. Wu SP, et al. (2015) Increased COUP-TFII expression in adult hearts induces mito- surrogate biomarker in patients with isocitrate dehydrogenase-mutant intrahepatic chondrial dysfunction resulting in heart failure. Nat Commun 6:8245. cholangiocarcinoma. Clin Cancer Res 20(7):1884–1890. 39. Harmancey R, Wilson CR, Wright NR, Taegtmeyer H (2010) Western diet changes 18. Latini A, et al. (2005) Mitochondrial energy metabolism is markedly impaired by D-2- cardiac acyl-CoA composition in obese rats: A potential role for hepatic lipogenesis. hydroxyglutaric acid in rat tissues. Mol Genet Metab 86(1-2):188–199. J Lipid Res 51(6):1380–1393. 19. Goodwin GW, Ahmad F, Doenst T, Taegtmeyer H (1998) Energy provision from gly- 40. Farrelly E, Amaral MC, Marshall L, Huang SG (2001) A high-throughput assay for cogen, glucose, and fatty acids on adrenergic stimulation of isolated working rat mitochondrial membrane potential in permeabilized yeast cells. Anal Biochem 293(2): hearts. Am J Physiol 274(4 Pt 2):H1239–H1247. 269–276. 20. Fu X, et al. (2015) 2-Hydroxyglutarate inhibits ATP synthase and mTOR signaling. Cell 41. Swanton EM, Saggerson ED (1997) Effects of adrenaline on triacylglycerol synthesis Metab 22(3):508–515. and turnover in ventricular myocytes from adult rats. Biochem J 328(Pt 3):913–922. 21. Lomenick B, et al. (2009) Target identification using drug affinity responsive target 42. Chua BH, Siehl DL, Morgan HE (1980) A role for leucine in regulation of protein stability (DARTS). Proc Natl Acad Sci USA 106(51):21984–21989. turnover in working rat hearts. Am J Physiol 239(6):E510–E514.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1601650113 Karlstaedt et al. Downloaded by guest on October 1, 2021