Ketogenic Diet Treatment of Defects in the Mitochondrial Malate Aspartate Shuttle and Pyruvate Carrier

Abstract

The mitochondrial malate aspartate shuttle system (MAS) maintains the cytosolic NAD+/NADH redox balance, thereby sustaining cytosolic redox-dependent pathways, such as glycolysis and serine biosynthesis. Human disease has been associated with defects in four MAS-proteins (encoded by MDH1, MDH2, GOT2, SLC25A12) sharing a neurological/epileptic phenotype, as well as citrin deficiency (SLC25A13) with a complex hepatopathic-neuropsychiatric phenotype. Ketogenic diets (KD) are high-fat/low-carbohydrate diets, which decrease glycolysis thus bypassing the mentioned defects. The same holds for mitochondrial pyruvate carrier (MPC) 1 deficiency, which also presents neurological deficits. We here describe 40 (18 previously unreported) subjects with MAS-/MPC1-defects (32 neurological phenotypes, eight citrin deficiency), describe and discuss their phenotypes and genotypes (presenting 12 novel variants), and the efficacy of KD. Of 13 MAS/MPC1-individuals with a neurological phenotype treated with KD, 11 experienced benefits-mainly a striking effect against seizures. Two individuals with citrin deficiency deceased before the correct diagnosis was established, presumably due to high-carbohydrate treatment. Six citrin-deficient individuals received a carbohydrate-restricted/fat-enriched diet and showed normalisation of laboratory values/hepatopathy as well as age-adequate thriving. We conclude that patients with MAS-/MPC1-defects are amenable to dietary intervention and that early (genetic) diagnosis is key for initiation of proper treatment and can even be lifesaving.

Keywords: AGC1; Citrullinemia; aspartate glutamate carrier 1 deficiency; citrin deficiency; epilepsy; hepatopathy; mitochondrial disease; modified Atkins diet; serine; treatment.

Figure 1

Malate aspartate shuttle (MAS) and the mitochondrial pyruvate carrier (MPC). The MAS is mainly responsible for the transfer of NADH across the inner mitochondrial membrane. MPC transport the end product of glycolysis—pyruvate—to the mitochondrial matrix, where it enters TCA cycle and finally oxidative phosphorylation. Fatty acids—from ketogenic diet—enter β-oxidation and thus bypass MAS and MPC. Proteins associated with human disease are written in bold pink. Glycolysis is inked in purple, MAS in green and β-oxidation (the bypass on KD) in blue. Of note, MDH2 takes part within the MAS and the TCA cycle. * Somatic variants in OGC have been related to pheochromocytoma. Abbreviations: NAD = nicotine amid adenine dinucleotide (NAD+ oxidated form, NADH reduced form), AGC1 = aspartate glutamate carrier, AGC2 = citrin, OGC1 = oxaloglutarate carrier, MDH1/2 = malate dehydrogenase (cytosolic/mitochondrial isoenzymes), GOT1/2 = aspartate aminotransferase (cytosolic/mitochondrial isoenzymes), MPC1/2 = mitochondrial pyruvate carrier 1/2 complex. G3P = glycerol-3-phosphate, DHAP = dihydroxyacetone phosphate, TCA = tricarboxylic acid.

Figure 2

Phenotypic overlap between different types of neurological malate aspartate shuttle (MAS) and mitochondrial pyruvate carrier (MPC) defects. Common features are written in bold. AGC1 = aspartate glutamate carrier 1/2, MDH1/2 = malate dehydrogenase (cytosolic/mitochondrial isoenzymes), GOT2 = aspartate aminotransferase, MPC1 = mitochondrial pyruvate carrier 1 complex. NAA = N-acetyl-aspartate.

Figure 3

MRI course AGC1-1. Selected T2-weighted images from subject AGC1-1 at identical ana- tomical planes to allow comparison at four different time points: (A1–D1) at axial level through the upper part of the body of the lateral ventricles; (A2–D2) at axial level of the foramen of Monroe; (A3–D3) at coronal plane through the foramen of Monroe. (A1–A3) MRI at 4 months showing enlarged size of lateral ventricles and extracerebral space as well as inappropriate myelination. The MRI (B1–B3) at 7 months is clearly abnormal with evidence of cerebral volume loss and widespread white matter signal alteration, compatible with a secondary hypomyelination. The findings are subsequently progressive, with impressive volume loss and extensive white matter changes at 21 months (C1–C3). The MRI (D1–D3) at age 40 months, after 18 months on classical KD (cKD), showed clear improvement. The size of the ventricles and the extracerebral space was regressive, myelination in the cerebral white matter advancing in general, and in the optic radiation, the corpus callosum and the subcortical (gyral) areas, in particular.

Figure 4

MRS course AGC1-1. Basal ganglia voxel positions and corresponding MR spectra acquired at an age of 3 months (left) and 1 year 9 months (middle), before introduction of classical KD (cKD), and at 3 years 4 months (right), after 18 months on the cKD. MRS data were analysed with LCModel, a fully automated spectral fitting package, which models each in-vivo spectrum as the linear combination of known basis spectra from a standard set of metabolites (see Supplementary Materials for more information). For each spectrum, the raw MRS data are depicted in black and the LCModel fit is overlaid in red. The residuals between the fit and the data are plotted above each spectrum. Using the standard basis set, a peak is seen in the residuals at 3.6–3.65 ppm (black arrows), indicating the presence of a metabolite in the spectrum, which is not included in the basis set. This peak appears most prominent in the spectrum acquired at 1 year 9 months (middle row, middle column), and diminishes in size following introduction of the cKD (middle row, right column). After including glycerol in the LCModel basis set (bottom row), this residual peak decreases in size (grey arrows), and the spectral fit in the area of the spectrum around 3.6 ppm improves, suggesting that glycerol may contribute to this unknown signal. (mI: myo-inositol, Cho: Choline, Cr: Creatine, Glu + Gln: Glutamate + Glutamine, NAA: N-acetyl-aspartate).

Figure 5

Functional validation of AGC1 variants in AGC1-2/3 and AGC1-5. (A) Transport assays of wild-type and p.D540N AGC1 variant in liposomes. Wild type (WT) and the p.D540N AGC1 variant of AGC1-2 and AGC1-3 patients were overexpressed in Escherichia coli, purified and reconstituted in liposomes, as previously described [40]. The uptake rate of ¹⁴C-glutamate (white bars) or ¹⁴C-aspartate (grey bars) was measured by adding 1 mM of radiolabelled glutamate or aspartate to liposomes reconstituted with purified WT or p.D540N AGC1 and containing 20 mM of glutamate. Transport reaction was terminated after 1 min by adding 20mM of pyridoxal 5′-phosphate and 20 mM batophenathroline. The means and SDs from three independent experiments are shown. (B) Expression analysis of AGC1 in fibroblasts from unrelated healthy controls and AGC1-5 proband. 100 μg of fibroblasts cultured at 37 °C in a humidified atmosphere with 5% CO2 in high glucose DMEM (D6546 SIGMA) supplemented with 10% foetal bovine serum and glutamine 2 mM, were harvested at passage #3 and lysed for western blot analysis with an antibody against AGC1. Densitometry analysis revealed the absence of AGC1 in AGC1-5 cells in comparison with two unrelated healthy controls showing similar AGC1 expression. An antibody against GAPDH was used for protein expression normalisation. The full western blot analysis is provided in Supplementary Figure S1 (Supplementary Materials).