NADH Shuttling Couples Cytosolic Reductive Carboxylation of Glutamine with Glycolysis in Cells with Mitochondrial Dysfunction

Abstract

The bioenergetics and molecular determinants of the metabolic response to mitochondrial dysfunction are incompletely understood, in part due to a lack of appropriate isogenic cellular models of primary mitochondrial defects. Here, we capitalize on a recently developed cell model with defined levels of m.8993T>G mutation heteroplasmy, mTUNE, to investigate the metabolic underpinnings of mitochondrial dysfunction. We found that impaired utilization of reduced nicotinamide adenine dinucleotide (NADH) by the mitochondrial respiratory chain leads to cytosolic reductive carboxylation of glutamine as a new mechanism for cytosol-confined NADH recycling supported by malate dehydrogenase 1 (MDH1). We also observed that increased glycolysis in cells with mitochondrial dysfunction is associated with increased cell migration in an MDH1-dependent fashion. Our results describe a novel link between glycolysis and mitochondrial dysfunction mediated by reductive carboxylation of glutamine.

Keywords: GAPDH; MDH1; NADH; cancer metabolism; cell migration; glycolysis; malate-aspartate shuttle; mitochondrial dysfunction; mitochondrial metabolism; reductive carboxylation.

Graphical abstract

Figure 1

Increasing Levels of m8993T>G Mutation Are Associated with Changes in Mitochondrial Function (A) Schematic representation of m8993T>G heteroplasmy in the mTUNE models. (B) RFLP analysis of last-cycle hot PCR products (mtDNA nt positions 8339–9334) amplified from total DNA samples of 143B cells harboring indicated levels of m.8993T>G, obtained by treatment with mtZFNs. Wild-type cells and 99% m.8993T>G cybrids were used as controls. (C) Basal extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) in mT7, mT45, and mT80 cells. Data are normalized to protein content. (D) Heatmap representation of mitochondrial respiratory complex protein expression in mT7, mT45, and mT80 cells as determined by mass spectrometry. Data represent values from three independent experiments and log₂ fold change values are color-coded as indicated. (E–G) Respiration of digitonin permeabilized mT7, mT45, and mT80 cells in the presence of (E) glutamate-malate (complex I) and succinate (complex II), (F) duroquinol (complex III), and (G) TMPD-ascorbate (complex IV). (H) Proliferation of mT7, mT45, and mT80 cells in the presence of galactose instead of glucose. Cell growth was determined by calculating the slope of respective proliferation curves. (I) Heatmap representation of significantly different intracellular metabolites (corrected ANOVA, p < 0.05) in mT7, mT45, and mT80 cells. Data represent values from three independent experiments and log₂ fold change values are color-coded as indicated. (C and E–H) Data are mean ± SEM from three independent cultures. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, one-way ANOVA.

Figure 2

Mitochondrial Function of mT7, mT45, and mT80 Cells Is Associated with Induction of Reductive Carboxylation in the Cytosol (A) Bubble representation of reactions involved in glycolysis, respiration, MAS, and cytosolic reductive carboxylation as predicted by mT7 and mT80 metabolic models. Bubble size is indicative of predicted reaction flux (μmoles/min/gDW). Blue and red bubbles indicate forward and reverse reactions. Gray arrows show the predicted direction of reactions, while gray dots represent reactions present in the depicted pathways, but with no predicted flux change. (B) Schematic representation of metabolite labeling pattern from (U)-¹³C-glutamine. Gray circles indicate ¹³carbon. (C) Proportion of total pool of metabolites originating from reductive carboxylation of U-¹³C-glutamine; aconitate m+5, citrate m+5, malate m+3, and fumarate m+3 are shown. Data are mean ± SEM from three independent cultures. ^∗∗∗p ≤ 0.001, one-way ANOVA.

Figure 3

Reductive Carboxylation Is Dependent on NAD⁺/NADH Levels (A) Total levels of NAD⁺/NADH in mT7, mT45, and mT80 cells measured using an enzymatic assay. Data are mean ± SEM from four independent experiments. (B) Levels of basal mitochondrial NAD(P)H measured by NAD(P)H autofluorescence. (C) Schematic representation of the rescue of mitochondrial NADH oxidoreductase activity by the expression of yeast-derived NADH dehydrogenase internal (NDI-1). (D) Western blot analysis of NDI-1 expression upon infection of mT80 cells with pWPI control or NDI-1 lentiviral vectors. Representative blot of two independent experiments. (E) OCR of pWPI and NDI-1 mT80 cells in basal conditions or after injection of oligomycin, FCCP, rotenone, and antimycin A. Data are mean ± SD from one representative experiment, performed with five replicates (out of two). (F) Total levels of NAD⁺/NADH in pWPI and NDI-1 mT80 cells. Data are mean ± SEM from three independent cultures. (G) Proportion of total pool of citrate, malate, and fumarate in pWPI and NDI-1 mT80 cells grown in the presence of 1-¹³C-glutamine. Data are mean ± SEM from three independent cultures. ^∗p ≤ 0.05, one-way ANOVA (A and B). ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, two-sided t test (E–G).

Figure 4

Reductive Carboxylation Supports Glycolytic Flux via NADH Channeling between MDH1 and GAPDH (A) Percentage of contribution to ATP production for the indicated enzymes, as predicted by metabolic modeling in mT80 and mT7 models. (B) Total levels of NAD⁺/NADH in shIDH1 and shMDH1 mT7 and mT80 cells, compared to shNTC controls, in basal conditions measured by enzymatic assay. (C) Schematic representation of labeling pattern originating from 4-²H₁-glucose. Deuterium atoms are represented as green filled circles. (D and E) Proportion of total pool of malate m+1 originating from 4-²H₁-glucose in mT7, mT45, and mT80 cells (D) and shMDH1 mT80 cells (E). (F) Levels of secreted lactate m+3 upon incubation of shMDH1 mT7 and mT80 cells with (U)-¹³C-glucose. Data are normalized to shNTC controls. (G) Proportion of total pool for citrate m+1 and malate m+1 originating from 1-¹³C-glutamine in mT7, mT45, and mT80 cells upon treatment with 0.5 μM of the GAPDH inhibitor heptelidic acid. (H) GAPDH IP on lysates of mT7, mT45, and mT80 cells. The interaction between GAPDH and MDH1 is shown by coIP. Immunoglobulin G (IgG) is used in negative isotype controls. Representative images from two independent experiments. (I) Immunofluorescence images of mT7, mT45, and mT80 cells stained with DAPI (blue), GAPDH (green), and MDH1 (red). (B and D–G) Data are mean ± SEM from at least three independent cultures. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, two-sided t test (B and E–G). ^∗p ≤ 0.05, one-way ANOVA (D). GAA, glutamate aspartate antiporter.

Figure 5

Aspartate Transamination Supports Flux through MDH1 and Generation of Malate (A) Schematic representation of MAS and labeling patterns originating from (U)-¹³C-aspartate. (B and C) Proportion of total pool of malate m+4 (B) and fumarate m+4 (C) in mT7, mT45, and mT80 cells grown in the presence of U-¹³C-aspartate upon treatment with vehicle control or 0.5 μM rotenone. (D) Malate m+4 levels originating from (U)-¹³C-aspartate in mT80 cells upon silencing of GOT1. Data are normalized to intracellular levels of aspartate m+4 and are mean ± SD from one independent experiment. (E and F) Cell growth of mT7, mT45, and mT80 cells grown in 25 mM galactose and supplemented with 5 mM aspartate (E) upon treatment with 2 mM of the transaminase inhibitor aminooxyacetate (F). Data are normalized on cell growth of vehicle control (E) or on cell growth in the presence of aspartate only (F). (G) Cell growth of mT80 cells grown in 25 mM galactose and supplemented with 5 mM aspartate upon silencing of GOT1. Data are normalized to the cell growth rate of vehicle control. (H and I) Total levels of NAD⁺/NADH in mT7, mT45, and mT80 cells (H) or shMDH1 mT80 cells (I) upon supplementation with 5 mM aspartate. Data are normalized on vehicle control. (J and K) Secretion of lactate of mT7, mT45, and mT80 cells (J) or shMDH1 mT80 cells (K) upon supplementation with 5 mM aspartate. Data are normalized on vehicle control. (B, C, and E–K) Data are mean ± SEM from at least three independent cultures. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, two-sided t test; n.s., not significant (B, C, E, and G–K). ^∗∗∗p ≤ 0.001, one-way ANOVA (F).

Figure 6

Mitochondrial Dysfunction Is Linked with Cell Migration (A) Enrichment p values (−log₁₀) of gene ontology (GO) biological processes involved in cell migration and cytoskeleton remodeling as obtained with measurements of protein abundance by proteomics. Red dashed line indicates false discovery rate (FDR) = 0.05. (B and D) Migration speed of mT7, mT45, and mT80 cells (B) or shMDH1 mT80 cells (D) measured by wound healing assay. (C and E) Values of J_ATP consumption due to cytoskeleton remodeling based on calculations from OCR and ECAR data upon treatment with 1 μM nocodazole in mT7, mT45, and mT80 cells (C) or mT80 shMDH1 cells (E). (F) Immunofluorescence images of mT7, mT45, and mT80 cells stained with DAPI (blue), phalloidin (green), or antibody against MDH1 (red). White arrows indicate areas of co-localization between MDH1 and actin in mT80 cells. (G) Quantification of co-localization between MDH1 and phalloidin (actin). Data were obtained from 20–30 ROIs per condition. (B–E) Data are mean ± SEM from three to four independent cultures and were normalized on mean values of each experiment. ^∗p ≤ 0.05 and ^∗∗∗p ≤ 0.001, ANOVA (B, C, and G) or Dunnett’s test (D).

Figure 7

Reductive Glutamine Carboxylation Regulates NAD Redox Balance and Supports Glycolysis in Response to Mitochondrial Dysfunction Reduced turnover of NADH by mitochondria leads to impairment of the MAS and increase of cytosolic NADH. This in turn induces reductive carboxylation of glutamine, providing carbons for NADH-coupled MDH1, thus regulating NAD redox state and enhancing GAPDH activity. Increased glycolytic turnover supports ATP production in the cytosol, and this is associated with cell migration.