SLC25A51 is a mammalian mitochondrial NAD+ transporter

Abstract

Mitochondria require nicotinamide adenine dinucleotide (NAD⁺) to carry out the fundamental processes that fuel respiration and mediate cellular energy transduction. Mitochondrial NAD⁺ transporters have been identified in yeast and plants^1,2, but their existence in mammals remains controversial^3-5. Here we demonstrate that mammalian mitochondria can take up intact NAD⁺, and identify SLC25A51 (also known as MCART1)-an essential^6,7 mitochondrial protein of previously unknown function-as a mammalian mitochondrial NAD⁺ transporter. Loss of SLC25A51 decreases mitochondrial-but not whole-cell-NAD⁺ content, impairs mitochondrial respiration, and blocks the uptake of NAD⁺ into isolated mitochondria. Conversely, overexpression of SLC25A51 or SLC25A52 (a nearly identical paralogue of SLC25A51) increases mitochondrial NAD⁺ levels and restores NAD⁺ uptake into yeast mitochondria lacking endogenous NAD⁺ transporters. Together, these findings identify SLC25A51 as a mammalian transporter capable of importing NAD⁺ into mitochondria.

Extended Data Figure 1.. SLC25A51 is a mitochondrial protein that affects cellular NAD⁺ distribution, proliferation, and metabolome profiles.

qPCR quantification of SLC25A51 mRNA expression in a, HEK293T (n=3) and b, HeLa cells (n=3) expressing shRNA targeting SLC25A51. NAD⁺ content of c, isolated mitochondria (n=3) and d, whole cell lysates (n=3) from HeLa cells with stable shRNA knockdown of SLC25A51 (KD) and non-targeting control (Ctrl). e, Western blot confirming shRNA targeting murine Slc25a51 reduces SLC25A51 protein expression in cells transfected with cDNA encoding SLC25A51-FLAG. f, Mitochondrial free NAD⁺ levels in mouse embryonic stem cells expressing shRNA against Slc25a51 and non-targeting shRNA (shFF2), as measured with the mitochondrial cpVenus NAD⁺ biosensor (n=3). g, qPCR quantification of SLC25A51 mRNA expression in HeLa cells transfected with siRNA targeting SLC25A51 (n=3). h, Western blot confirming protein expression of Flag-tagged mitochondrial carriers. Controls include stable expression of the NAD⁺ biosensor (sensor) and anti-Tubulin for loading. i, Immunofluorescent detection of SLC25A51 and SLC25A52 subcellular localization. Cells were transiently transfected with cDNA encoding Flag-HA-tagged SLC25A51 or SLC25A52 and probed with anti-Flag and the mitochondrial marker, anti-MTC02. Scale bar: 10 μM, 2 μM on inset. Inset represents zoomed view of Flag localization and mitochondria. Proliferation of j, HAP1 SLC25A51 KO (n=8), k, HEK293T SLC25A51 shRNA-knockdown (n=8) l, HeLa SLC25A51 shRNA-knockdown cells (n=8) and their respective controls. Proliferation was measured by CyQuant, a fluorescent DNA dye, at 0h and 96h after plating and expressed as fold change. qPCR quantification of SLC25A52 mRNA expression in m, HAP1 SLC25A51 KO, n, HEK293T SLC25A51 shRNA-knockdown and o, HeLa SLC25A51 shRNA-knockdown cells (n=3). p, Western blot of whole cell protein lysates from HAP1 wildtype (WT) and SLC25A51 knockout (KO) cells confirming SLC25A51 loss. Loading control is total protein measured by Revert 700 Total Protein. Heat map of top 30 q, mitochondrial and r, whole cell metabolites that differ between HAP1 wildtype and SLC25A51 KO cells (n=3). Data represented as mean ± SEM. P values were determined by unpaired, two-tailed Student’s t-test (for two groups) or one-way ANOVA with multiple comparisons analysis using Dunnett’s method (for groups of three or more). *P<0.05, and ***P<0.001 vs control or WT (exact P values are provided in the source data).

Extended Data Figure 2.. NAD⁺ and SLC25A51 affect oxidative phosphorylation.

a, Respiration of isolated mitochondria from HEK293T cells treated with either vehicle (Veh) or the NAMPT inhibitor FK866 to deplete mitochondrial NAD⁺. Mitochondria were treated with pyruvate and malate (state 2), then ADP was added to induce state 3 respiration. 1 mM NAD⁺ was added to test the ability of exogenous NAD⁺ to rescue respiration in the setting of mitochondria NAD⁺ depletion (Trace is representative of n=4 independent experiments). P values were determined by two-way ANOVA with multiple comparisons analysis using the Sidak method. b, Oxygen consumption rate (OCR) was measured in SLC25A51 shRNA knockdown (KD) and control (Ctrl) HeLa cells using a Seahorse XF96e. Basal OCR was measured prior to the addition of treatments and maximal respiration was measured after the sequential addition of oligomycin (Oligo, ATP synthase inhibitor) and FCCP (uncoupler). Rotenone (Rot) and Antimycin A (AA) were added as a control to completely block mitochondrial oxygen consumption (n=6). c, Respiration of isolated mitochondria from SLC25A51 knockdown HEK293T cells. Mitochondria were treated with pyruvate/malate (state 2), and then ADP was added to induce state 3 respiration. Oligomycin was added to block ATP synthase-mediated respiration (n=3 independent experiments). d, Mitochondria were isolated from HEK293T control, SLC25A51 shRNA knockdown cells, and controls treated with FK866 to deplete mitochondrial NAD⁺. Mitochondrial oxygen consumption rate was measured after treatment with pyruvate/malate (state 2), ADP (state 3), and 1 mM NAD⁺ (n=4 independent experiments). e, Mean volume per mitochondrial unit and f, number of distinct mitochondria per cell quantified from confocal image reconstructions of mitochondrial voxels in SLC25A51 shRNA knockdown (n=31 cells) and control (n=32 cells) HeLa cells. Data represented as mean ± SEM. P values were determined by unpaired, two-tailed Student’s t-test. *P<0.05, **P<0.01, and ***P<0.001 vs vehicle or control; ^###P<0.001 vs state 3 (exact P values are provided in the source data).

Extended Data Figure 3.. Intact NAD⁺, but not nicotinamide or nicotinamide mononucleotide contributes to the mitochondrial NAD⁺ pool.

a, Mitochondrial NAD⁺ content was measured in isolated mitochondria from a, HeLa control (Ctrl) cells, cells treated with FK866 (Ctrl+FK), and SLC25A51 shRNA-knockdown (KD) cells. NAD⁺ content of isolated mitochondria was determined before (untreated) and after a 40-min incubation with 1 mM NAD⁺ (n=3 independent experiments). b, NAD⁺ levels in HEK293T mitochondria incubated with 1 mM nicotinamide (NAM), 1 mM nicotinamide mononucleotide (NMN), or 1 mM NAD⁺ (n=3 independent experiments). c, NAD⁺ uptake in NAD⁺-depleted mitochondria isolated from HEK293T cells incubated with NAD⁺ ± 2 mM NAM or 2mM NMN (n=4 independent experiments). d, Fractional labeling of mitochondrial NAD⁺ in HAP1 cells treated with isotopically double labeled NaR (n=3 biological independent replicates). Data represented as mean ± SEM. P values were determined by unpaired, two-tailed Student’s t-test (for two groups) or one-way ANOVA with multiple comparisons analysis using Dunnett’s or Tukey’s method (for groups of three or more). *P<0.05 and ***P<0.001 vs untreated, vehicle, and wildtype M+0; ^#P<0.05 vs wildtype M+1.

Extended Data Figure 4.. Generation and validation of yeast strains for testing mitochondrial NAD⁺ transport.

a, PCR genotyping to confirm double knockout gene deletion in BY4727 S. Cerevisiae via antibiotic-resistance cassette replacement at the NDT1 and NDT2 loci. b, c, Deletion of the mitochondrial NAD⁺ carriers NDT1 and NDT2 in DKO strain phenocopied previously described growth defects on non-fermentative media (YP, 3% glycerol media), which was rescued by plasmid expression of NDT1 d, Western blot confirmed enrichment of mitochondrial markers (MTC02 and COXIV) and absence of cytoplasmic proteins (actin) or ER (SC2) in isolated mitochondria from yeast. e, RT-PCR confirmed ectopic expression from pRS415-SLC25A51 and pRS415-SLC25A52 in DKO strains.

Extended Data Figure 5.. Kinetics and selectivity of NAD⁺ transport by human SLC25A51 expressed in yeast mitochondria.

a, Co-incubation with excess unlabeled NAD⁺ (n=5 independent experiments for 1mM NAD⁺) b, supraphysiological levels of NMN (100 μM, n=4 independent experiments; 500 μM, n=5 independent experiments), or c, NADH (n=3 independent experiments) with ³H-NAD⁺ to measure uptake competition in mitochondria from DKO yeast expressing SLC25A51. d, Proportional relationship between integrated peak intensities from mass spectrometry of mitochondrial samples compared to a known meta dataset of absolute protein abundances; used to quantitate SLC25A51 abundance in yeast samples. e, Uptake measured with indicated NAD⁺ concentrations; calculated from specific activity (n=3 independent experiments, mean ± SEM). f, Lineweaver-Burk plot based on a non-linear fit with datapoints overlaid (n=3 independent experiments). P values were determined by two-way ANOVA with multiple comparisons analysis using Sidak’s method. *P<0.05 and **P<0.01 vs 100 μM cold NAD⁺.

Figure 1.. SLC25A51 and SLC25A52 expression dictates mitochondrial NAD⁺ concentration.

a, NAD⁺ content of isolated mitochondria (n=4) and b, whole cell lysates (n=3) from HEK293T cells stably depleted of SLC25A51 (shRNA1–3) and stably expressing non-targeting control shRNA (Ctrl). c, Mitochondrial free NAD⁺ levels in HeLa cells transfected with siRNA targeting SLC25A51 (siRNA1 and 2) and non-targeting siRNA (Ctrl), measured by the mitochondrially-targeted cpVenus NAD⁺ biosensor, (n=4). d, Mitochondrial free NAD⁺ levels in HeLa cells overexpressing NDT1 (yeast mitochondrial NAD⁺ transporter) (n=24), SLC25A32 (n=4), SLC25A33 (n=4), SLC25A36 (n=4), SLC25A51 (n=4), SLC25A52 (n=4), and vector control (n=24) measured by the cpVenus NAD⁺ biosensor. e, Mitochondrial free NAD⁺ levels in HEK293 cells with stable shRNA-mediated knockdown of SLC25A51 (n=3) and stable expression of non-targeting shRNA (Ctrl) (n=3). f, Mitochondrial free NAD⁺ levels in U2OS cells overexpressing SLC25A51, SLC25A52 and vector control (n=6), as measured by NAD⁺-Snifit. g, Mitochondrial (n=3) and h, whole cell NAD⁺ content (n=3) of lysates collected from CRISPR/Cas9-mediated SLC25A51 knockout (KO) and wildtype (WT) HAP1 cells. i, Mitochondrial (n=3) and j, whole cell (n=3) metabolomes of HAP1 SLC25A51 KO and WT cells measured by liquid chromatography–mass spectrometry. Significantly changed metabolites were determined by setting a false discovery rate of 1% (two-stage step-up method of Benjamini, Krieger, and Yekutieli) and are represented in a volcano plot. n represents biological independent replicates. Data represented as mean ± SEM. P values were determined by unpaired, two-tailed Student’s t-test (for two groups) or one-way ANOVA with multiple comparisons analysis using Dunnett’s method (for groups of three or more). **P<0.01, ***P<0.001 vs. control, vector, or wildtype (exact P values are provided in the source data).

Figure 2.. SLC25A51 modulates mitochondrial respiratory capacity.

Oxygen consumption rate (OCR) for a, SLC25A51 shRNA-depleted HEK293T (n=5), b, HAP1 SLC25A51 KO cells (n=6), c, HAP1 SLC25A51 KO cells rescued using adenovirus-mediated SLC25A51 expression (multiplicity of infection or MOI-4) (n=6), and d, HAP1 wildtype cells with low (MOI-2), medium (MOI-4) and high (MOI-6) overexpression of SLC25A51 (n=6) (80,000 cells per well). Basal OCR was measured prior to the addition of treatments and maximal respiration was measured after the sequential addition of oligomycin (Oligo, ATP synthase inhibitor) and FCCP (uncoupler). Rotenone (Rot) and Antimycin A (AA) were then added as a control to completely block mitochondrial oxygen consumption. e, Quantification of mitochondrial membrane potential using the cell permeant fluorescent dye tetramethylrhodamine, ethyl ester (TMRE) (n=4). f, Mitochondrial content measured by fluorescence intensity of the mitochondrial localization dye, MitoTracker Deep Red (n=4). Relative fluorescence intensities were determined by flow cytometry. g, Cumulative mitochondrial volume per cell quantified from confocal image reconstructions of mitochondrial voxels in SLC25A51 shRNA knockdown (n=31 cells) and control (n=32 cells) HeLa cells. h, Representative images of mitochondrial voxels (mitochondrial marker, anti-MTC02) reconstructed using 0.1 μm optical slices and Imaris Surface Analyses. Scale bar: top, 5 μm; bottom, 1 μm. i, Western blot of mitochondrial oxidative phosphorylation complexes in HAP1 SLC25A51 KO cells. TOM20 was blotted as a mitochondrial loading control. n represents biological independent replicates unless otherwise indicated. Data represented as mean ± SEM. P values were determined by unpaired, two-tailed Student’s t-test or two-way ANOVA with multiple comparisons analysis using Dunnett’s method (for groups of three or more). *P<0.05, **P<0.01, and ***P<0.001 vs. control, wildtype, or KO (exact P values are provided in the source data).

Figure 3.. SLC25A51 expression is required for NAD⁺ uptake in isolated mitochondria.

NAD⁺ content of isolated mitochondria measured before and after a 40-min incubation with 1 mM NAD⁺ from a, HEK293T control (Ctrl) cells, control + FK866 (Ctrl+FK) to deplete mitochondrial NAD⁺, SLC25A51 shRNA knockdown (KD) cells, and SLC25A51 KD + murine Slc25a51 cDNA (KD+A51) cells; b, HAP1 wildtype (WT) cells, WT + FK866 (WT+FK), SLC25A51 (KO) knockout cells, and KO cells transduced with adenovirus encoding SLC25A51 (KO+A51); and c, HEK293T control (Ctrl) cells, control +FK866 (Ctrl+FK), SLC25A51 (KD) cells, and SLC25A51 KD cells + cDNA encoding the yeast mitochondrial NAD⁺ transporter NDT1 (KD+NDT1). d, NAD⁺ content of mitochondria isolated from HEK293T control cells and cells overexpressing SLC25A51 (OE) before and after a 20-min incubation with 1 mM NAD⁺ (a-d; n=3 independent experiments). e, HeLa cells were transfected 3 days in advance with non-targeting siRNA (siNT) and siRNA targeting SLC25A51 (siA51), or NAMPT (siNampt) and mitochondrial free NAD⁺ levels were measured after 16-hours of nicotinamide riboside (NR) treatment (n=3). f, HEK293 cells stably expressing Arabidopsis thaliana NDT2-FLAG (AtNDT2), SLC25A32-FLAG, SLC25A51-FLAG or g, SLC25A52-FLAG were transfected with mitochondrially-targeted eGFP (mito-eGFP) or the catalytic domain of PARP1 (mitoPARP1cd). Mitochondrial PARylation levels reflect mitochondrial NAD⁺ availability. h, Mitochondrial NAD⁺ content after 7-hours of treatment with doubly isotopically labeled nicotinic acid riboside (NaR). i, Whole cell and j, mitochondrial fractional labeling patterns normalized to total ion counts in WT to reflect relative abundance of NAD⁺ after NaR treatment (n=3). n represents biological independent replicates unless otherwise indicated. Data represented as mean ± SEM. P values were determined by unpaired, two-tailed Student’s t-test. *P<0.05, **P<0.01, and ***P<0.001 vs untreated, wildtype M+0; ^###P<0.001 vs control + NAD⁺, wildtype M+1; ^{^^^}P<0.001 vs wildtype M+2 (exact P values are provided in the source data).

Figure 4.. SLC25A51 is sufficient for transport of NAD⁺ into yeast mitochondria lacking their endogenous transporters, NDT1 and NDT2.

³H-NAD⁺ uptake measured in isolated mitochondria from wildtype (n=4 independent experiments), NDT1 and NDT2 double knockout (DKO) (n=4 independent experiments), and a, DKO + overexpression of SLC25A51 (n=4 independent experiments) or b, DKO + overexpression of SLC25A52 (n=3 independent experiments) yeast; grey line indicates mean basal radioactivity in DKO samples. P values were determined by a two-way ANOVA with multiple comparisons analysis using Dunnett’s method. c, Initial rates of NAD⁺ uptake into isolated yeast mitochondria (n=4 independent experiments for WT, DKO, and DKO + SLC25A51; n=3 for DKO + SLC25A52). d, NAD⁺ content measured from isolated yeast mitochondria (n=3 biological independent replicates). P values were determined by one-way ANOVA with multiple comparisons analysis using Tukey’s method. Data represented as mean ± SEM. **P<0.01, and ***P<0.001 vs DKO, ^##P<0.01 and ^###P<0.001 vs wildtype (exact P values are provided in the source data).