Reductive glutamine metabolism by IDH1 mediates lipogenesis under hypoxia

Abstract

Acetyl coenzyme A (AcCoA) is the central biosynthetic precursor for fatty-acid synthesis and protein acetylation. In the conventional view of mammalian cell metabolism, AcCoA is primarily generated from glucose-derived pyruvate through the citrate shuttle and ATP citrate lyase in the cytosol. However, proliferating cells that exhibit aerobic glycolysis and those exposed to hypoxia convert glucose to lactate at near-stoichiometric levels, directing glucose carbon away from the tricarboxylic acid cycle and fatty-acid synthesis. Although glutamine is consumed at levels exceeding that required for nitrogen biosynthesis, the regulation and use of glutamine metabolism in hypoxic cells is not well understood. Here we show that human cells use reductive metabolism of α-ketoglutarate to synthesize AcCoA for lipid synthesis. This isocitrate dehydrogenase-1 (IDH1)-dependent pathway is active in most cell lines under normal culture conditions, but cells grown under hypoxia rely almost exclusively on the reductive carboxylation of glutamine-derived α-ketoglutarate for de novo lipogenesis. Furthermore, renal cell lines deficient in the von Hippel-Lindau tumour suppressor protein preferentially use reductive glutamine metabolism for lipid biosynthesis even at normal oxygen levels. These results identify a critical role for oxygen in regulating carbon use to produce AcCoA and support lipid synthesis in mammalian cells.

Figure 1

Reductive carboxylation is the primary route of glutamine to lipids. a) Schematic of carbon atom (circles) transitions and tracers used to detect reductive glutamine metabolism. Isotopic label from [1-¹³C]glutamine (red) is lost during oxidative conversion to succinate (Suc) but retained on citrate (Cit), oxaloacetate (Oac), aspartate (Asp), malate (Mal), and fumarate (Fum) in the reductive pathway (green arrows). [5-¹³C]glutamine (blue) transfers label to AcCoA through reductive metabolism only. Molecular symmetry is shown for oxidative metabolism. b) Contribution of [5-¹³C]glutamine and [U-¹³C₅]glutamine to lipogenic AcCoA in cell lines. c) IDH1 levels in A549 cells expressing IDH1-specific (IDH1a and IDH1b) or control shRNAs. d) Metabolite labeling from [1-¹³C]glutamine from cells in (c). e) IDH flux estimates from ¹³C MFA model in control or IDH1-knockdown A549 cells cultured with [U-¹³C₅]glutamine. f) Cell proliferation of A549 cells expressing IDH1-shRNAs. Error bars indicate 95% confidence interval (CI) for (b, e) and s.e.m. (n=3) for (d, f). * denotes p < 0.05.

Figure 2

Hypoxia reprograms cells to rely on reductive glutamine metabolism for lipid synthesis. a,b) Labeling of palmitate extracts from A549 cells cultured under normoxia or hypoxia with [5-¹³C]glutamine (a) or [U-¹³C₆]glucose (b). Similar results were observed in myristate, oleate and stearate pools (not shown). c,d) Relative contribution of glucose oxidation or glutamine reduction to lipogenic AcCoA in A549 cells under normoxia and hypoxia (c) or A549 cells expressing control or IDH1-targeting shRNAs under hypoxia (d). e) Absolute fluxes of [U-¹³C₆]glucose and [5-¹³C]glutamine to palmitate in A549 cells. Error bars indicate 95% CI from model for c – e; * denotes p < 0.05. f) Huh7 cell proliferation after 4 days in the presence or absence of glutamine. Error bars indicate s.e.m. (n=3) for a, b, and f. ** denotes p < 0.005 comparing glutamine-free cultures. *** denotes p < 0.001 comparing normoxia and hypoxia.

Figure 3

Reductive TCA metabolism increases under hypoxia. MRC5 cells were cultured under normoxia or hypoxia for 3 days in the presence of tracer. A) Relative level of glucose oxidation as determined by M2 labeling from [U-¹³C₆]glucose (see Fig. S16A for atom transition map). M2 isotopologues were the most abundant labeled metabolites in mass spectra. B) Relative abundance of citrate and αKG. c,d) Relative contribution of reductive glutamine metabolism to TCA metabolites, determined by M1 labeling from [1-¹³C]glutamine. M1 isotopologues were the only species with significant abundance. e) Contribution of glucose oxidation and glutamine reduction to lipogenesis in A549 cells cultured with or without 5 mM DCAError bars indicate s.e.m. for (a-d; n=3) and 95% CI for (e). * denotes p < 0.01 and ** denotes p < 0.001 comparing normoxia to hypoxia. † denotes p < 0.05 comparing control to DCA in hypoxia.

Figure 4

HIF/ARNT/VHL signaling regulate carbon utilization for lipogenesis. A-C) Contribution of glucose oxidation ([U-¹³C₆]glucose) and glutamine reduction ([5-¹³C]glutamine) to lipogenesis in RCC lines (A), parental control (PRC3) and VHL+ (WT8) cells derived from 786-O line (B), or vector control (pTV) or HIF2α shRNA (pTR) cells derived from 786-O line (C). D) Western blot to determine HIF2α levels for cells in (B-C). Error bars indicate 95% CIs obtained from ISA model. * indicates p < 0.05. E) Model depicting the metabolic reprograming of mammalian cells by hypoxia or VHL loss to employ reductive glutamine metabolism for lipogenesis. HIF stabilization drives transcription of PDK1, which decreases PDH activity and subsequently intracellular citrate levels. IDH1 and ACO1 reductively generate lipogenic citrate from glutamine-derived αKG. DCA can inhibit PDKs, forcing increased glucose oxidation in hypoxic cells.