Oxidation of alpha-ketoglutarate is required for reductive carboxylation in cancer cells with mitochondrial defects

Abstract

Mammalian cells generate citrate by decarboxylating pyruvate in the mitochondria to supply the tricarboxylic acid (TCA) cycle. In contrast, hypoxia and other impairments of mitochondrial function induce an alternative pathway that produces citrate by reductively carboxylating α-ketoglutarate (AKG) via NADPH-dependent isocitrate dehydrogenase (IDH). It is unknown how cells generate reducing equivalents necessary to supply reductive carboxylation in the setting of mitochondrial impairment. Here, we identified shared metabolic features in cells using reductive carboxylation. Paradoxically, reductive carboxylation was accompanied by concomitant AKG oxidation in the TCA cycle. Inhibiting AKG oxidation decreased reducing equivalent availability and suppressed reductive carboxylation. Interrupting transfer of reducing equivalents from NADH to NADPH by nicotinamide nucleotide transhydrogenase increased NADH abundance and decreased NADPH abundance while suppressing reductive carboxylation. The data demonstrate that reductive carboxylation requires bidirectional AKG metabolism along oxidative and reductive pathways, with the oxidative pathway producing reducing equivalents used to operate IDH in reverse.

Figure 1. Metabolomic features of cells using reductive carboxylation

(A) Relative abundance of metabolites extracted from triplicate samples of 143Bwt, 143Bcytb, UOK262FH and UOK262EV cells. Peak areas of each metabolite were normalized to protein abundance. The color reflects a log₂ scale. (B) Principal component analysis for the twelve extracts used to generate the heat map in (A). (C-E) Metabolites demonstrating consistent and significant alterations in abundance in cells using reductive carboxylation. In all cases, p<0.005 for the comparison between cells lines in each pair. (F) Abundance of succinate in several models of reductive carboxylation induced by impaired ETC activity. High doses of metformin and rotenone inhibit complex I. Antimycin inhibits complex III. Hypoxia functions as an inhibitor of complex IV. These treatments were all applied to 143Bwt cells. CCL16-B2 are Chinese hamster fibroblast cells with impaired Complex I activity as a consequence of NDUFA1 mutation. CCL16-NDI1 cells were generated by stably infecting CCL16- B2 with yeast NADH quinone oxidoreductase (NDI1) which restores oxidative capacity and eliminates reductive carboxylation. Data are the average and S.D. of three independent cultures. *p<0.05; **p<0.005, Student's t-test. NS, not significant.

Figure 2. Oxidative glutamine metabolism is the primary route of succinate formation in cells using reductive carboxylation to generate citrate

(A-C) 143Bwt cells were cultured in medium containing unlabeled glucose and [U-¹³C]glutamine, and labeling patterns in succinate, fumarate and malate were followed. Isotopologues with four ¹³C atoms (m+4) are formed through oxidative metabolism (green), and isotopologues with three ¹³C atoms (m+3) are formed through reductive metabolism (red). See Fig. S2 for labeling scheme. (D-F) 143Bcytb cells were cultured in medium containing unlabeled glucose and [U-¹³C]glutamine, and labeling patterns in succinate, fumarate and malate were followed as in (A-C). Note that succinate formation is primarily oxidative, but that fumarate and malate formation is almost exclusively reductive, indicating bidirectional metabolism of glutamine.

Figure 3. Pyruvate carboxylase contributes to citrate formation in cells using reductive carboxylation

(A-B) 143B and UOK262 cells were cultured with [3,4-¹³C]glucose and unlabeled glutamine and labeling patterns in malate and citrate were analyzed. Data are the average and S.D. of three independent cultures. *p<0.05; ** p<0.005, Student's t-test. (C) Left, abundance of pyruvate carboxylase (PC) protein in 143Bcytb cells stably expressing non-targeting (shGFP) or two independent hairpins targeting PC (shPC_1, shPC_2). Cells were cultured with [3,4-¹³C]glucose and unlabeled glutamine (middle), or with unlabeled glucose and [U- ¹³C]glutamine, and labeling patterns were analyzed. Data are the average and S.D. of three independent cultures. *p<0.05; ** p<0.005, Student's t-test.

Figure 4. Elevated succinate abundance is dispensable for reductive carboxylation

(A) Abundance of the alpha subunit of succinate-CoA ligase (encoded by SUCLG1) in 143Bcytb cells transfected with a control siRNA (siLUC) or siRNA directed against SUCLG1 (siSUCLG1). (B) Relative abundance of succinate in 143Bcytb following transient knockdown of LUC or SUCLG1. Data are the average and S.D. of three independent cultures. *p<0.05, Student's t-test. (C) Fraction of citrate containing five glutamine-derived ¹³C nuclei (m+5) following transient knockdown of LUC or SUCLG1 and culture with [U-¹³C]glutamine and unlabeled glucose. Data are the average and S.D. of three independent cultures. NS, not significant.

Figure 5. AKG dehydrogenase is required for reductive carboxylation

(A) Abundance of OGDH protein in 143Bcytb cells transfected with a control siRNA (siLUC) or siRNA directed against OGDH (siOGDH). (B) Relative abundance of succinate in 143Bcytb cells following transient knockdown of siLUC or siOGDH. Data are average and S.D. of three independent cultures. **p<0.005, Student's t-test. (C) Abundance of m+5 labeled citrate in 143Bcytb cells after transient knockdown of LUC or OGDH cultured with [U-¹³C]glutamine and unlabeled glucose. Abundance of labeled citrate was calculated by multiplying the relative citrate pool size by the percent citrate m+5. Data are the average and S.D. of three independent cultures. *p< 0.05, Student's t-test. (D) Abundance of OGDH protein in 143Bwt cells transiently transfected with LUC or OGDH. (E) Ratio of citrate to AKG in 143Bwt cells following knockdown of LUC or OGDH. Data are the average and S.D. of four independent cultures. **p<0.005, Student's t-test. (F) Mass isotopomer distribution of citrate in 143Bwt cells following transient knockdown of LUC or OGDH and cultured with [U-¹³C]glutamine and unlabeled glucose. Data are the average and S.D. of three independent cultures. **p<0.005, Student's t-test. (G) Western blot demonstrating expression of fumarate hydratase lacking its mitochondrial targeting sequence and containing a V5 epitope tag (FHΔMTS), in UOK262 cells. EV, empty vector. (H) Relative fumarate abundance in UOK262 cells containing or lacking FHΔMTS. Data are the average and S.D. of three independent cultures. *p< 0.05, Student's t-test. (I) Mass isotopomer distribution of citrate in UOK262 cells containing or lacking FHΔMTS. Data are the average and S.D. of three independent cultures. NS, not significant.

Figure 6. AKG dehydrogenase and NNT contribute to NAD+/NADH ratio

(A) Two-photon fluorescence imaging of NADH in 143Bwt and 143Bcytb. Red indicates areas of increased NADH abundance. Scale bar 22μm. (B) NAD+/NADH ratio in 143Bwt and 143Bcytb cells. Data are the average and S.D. of 3-4 independent cultures. **p<0.005, Student's t-test. (C) NAD+/NADH ratio in 143Bcytb following transient transfection with siRNAs against Luciferase (siLUC) or OGDH (siOGDH). Data are the average and S.D. of 3-4 independent experiments. **p<0.005, Student's t-test. (D) NAD+/NADH ratio in 143Bcytb following transfection with siLUC) or an siRNA pool directed against niconatinomide nucleotide transhydrogenase (siNNT). Data are the average and S.D. of 3-4 biological replicates. *p< 0.05, Student's t-test. (E) Abundance of m+5 labeled citrate in 143Bcytb cells transfected with siLUC or siNNTand cultured with [U-¹³C]glutamine. Abundance of labeled citrate was calculated by multiplying the relative abundance of total citrate by the fractional contribution of the m+5 isotopomer. Data are the average and S.D. of three independent cultures. **p<0.005, Student's t-test. (F) NADP+/NADPH ratio in 143Bcytb cells following transient transfection with siLUC or siOGDH. Data are the average and S.D. of three independent cultures. *p< 0.05, Student's t-test. (G) NADP+/NADPH ratio in 143Bcytb cells following transient transfection with siLUC or siNNT. Data are the average and S.D. of three independent cultures. *p< 0.05, Student's t-test. (H) NADP+/NADPH ratio in 143Bcytb cells following transient transfection with siLUC or siIDH2. Data are the average and S.D. of three independent cultures. Inset is western blot depicting bundance of IDH2 protein in both conditions. *p< 0.05, Student's t-test. (I) Model for induction of reductive carboxylation in 143Bcytb cells. Flux through AKG pdehydrogenase generates succinate and NADH, which is dissipated by NNT to generate NADPH during reductive carboxylation.