Altered mitochondria-associated ER membrane (MAM) function shifts mitochondrial metabolism in amyotrophic lateral sclerosis (ALS)

Abstract

Mitochondrial function is modulated by its interaction with the endoplasmic reticulum (ER). Recent research indicates that these contacts are disrupted in familial models of amyotrophic lateral sclerosis (ALS). We report here that this impairment in the crosstalk between mitochondria and the ER impedes the use of glucose-derived pyruvate as mitochondrial fuel, causing a shift to fatty acids to sustain energy production. Over time, this deficiency alters mitochondrial electron flow and the active/dormant status of complex I in spinal cord tissues, but not in the brain. These findings suggest mitochondria-associated ER membranes (MAM domains) play a crucial role in regulating cellular glucose metabolism and that MAM dysfunction may underlie the bioenergetic deficits observed in ALS.

Fig. 1. Defects in mitochondrial respiratory complexes in tissues from SOD1^G93A mice.

a Cytochrome c oxidase (COX) and succinate dehydrogenase (SDH) activities are shown by double staining in 20-µm-thick coronal sections of the cerebellum and b in cross sections of SPC tissues from SOD1^G93A and NTg (control) mice at pre-symptomatic stage P60. Note that, in control sections, the purple stain representing SDH activity is masked by the brown stain representing COX activity, whereas the purple stain is prominently visible in mutant tissues. This suggests that COX activity is reduced in mutant samples, enabling visualization of the SDH stain. The images shown are representative of 3 independent animals per group.

Fig. 2. Progressive alterations in mitochondrial respiration in SOD1^G93A mice.

Oxygen consumption rate (OCR, in pmole O₂/min/mg protein) in mitochondria isolated from brain (a, c) and spinal cord (SPC) (b, d) from SOD1^G93A mice at the indicated ages (in days) compared to age-matched NTg controls (set at 100%; dotted lines) at states 2/3/4/3U. The columns represent the mean ± SE. n = 4 biological replicates consisting of a pool of tissues from 4 animals (2 male and 2 female). Dots indicate the OCR average at the specific time point using Seahorse (see methods). e Complex II enzymatic activity in isolated mitochondria from SOD1^G93A tissues at P60 relative to the NTg control average (set at 100%; dotted line). Data represent the mean ± SD; n = 3 biologically independent samples (2 male and 1 female). f RCR (ratio of state 3: state 4) calculated from the respiratory values in Fig. 2 (a–d) in SOD1^G93A brain (top panel) and SPC (bottom panel) mice relative to the NTg control average (set at 100%; dotted lines). The columns represent the mean of the RCR ± SE; n = 4. Dots indicate the relative RCR values of SOD1^G93A relative to NTg of each group at the specific time point (set at 100%; dotted lines). g OCR in hMNs^A4V relative to WT controls (set at 100%; dotted lines) in the presence of pyruvate as a mitochondrial substrate. Data represent the mean ± SD; n = 3 (DIV2), n = 5 (DIV5) and n = 7 (DIV14). h, i NADH- and FADH₂-OCR in permeabilized hMNs^A4V relative to WT controls (set at 100%; dotted lines). Data represent the mean ± SE. n = 4. Statistical test: Student’s two-tailed t-test for normal data or the Mann-Whitney U test for non-normal data at an α = 0.05 significance level. Statistical significance is shown as: n.s. p>0.05; * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001.

Fig. 3. Defects in glycolysis and pyruvate metabolism in ALS models shift mitochondria towards fatty acid oxidation.

a Hexokinase (HK) activity in total homogenates of brain and SPC from SOD1^G93A mice at the indicated ages relative to NTg control average (set at 100%; dotted line). The columns represent the mean ± SD. Brain: n = 3 (P15), n = 3 (P30), n = 3 (P60), n = 3 (P90); SPC: n = 3 (P15), n = 3 (P30), n = 4 (P60), n = 5 (P90) biologically independent samples. b Pyruvate dehydrogenase complex (PDHC) activity in crude mitochondria fractions or total homogenates of brain and SPC from SOD1^G93A mice at the indicated ages relative to NTg control average (set at 100%; dotted line). The columns represent the mean ± SD. Brain: n = 4 (P60), n = 3 (P90); SPC: n = 3 (P60), n = 3 (P90). c Hexokinase (HK) activity in total homogenates from hMNs^A4V at the indicated days in vitro (DIV) relative to WT control average (set at 100%; dotted line). The columns represent the mean ± SD; n = 4 (DIV2), n = 3 (DIV7) n = 3 (DIV14). d Pyruvate dehydrogenase complex (PDHC) activity in total homogenates from hMNs^A4V at the indicated (DIV) relative to WT control average (set at 100%; dotted line). The columns represent the mean ± SD; n = 3 (DIV7), n = 3 (DIV14). e Lactate dehydrogenase (LDH) activity in hMNs^A4V compared to WT controls. The columns represent the mean ± SE; n = 3. f CPT1 activity in hMNs^A4V relative to WT controls (set at 100%; dotted line). The columns represent the mean ± SD; n = 3 (DIV2), n = 4 (DIV7), n = 3 (DIV14). g CPT1 activity in crude mitochondria fractions from the brain (left panel) and SPC (right panel) at the indicated ages relative to NTg control average (set at 100%; dotted line). The columns represent the mean ± SD. Brain: n = 3 (P15), n = 4 (P30), n = 4 (P60); SPC: n = 4 (P15), n = 4 (P30), n = 4 (P60). Data at 120 days in brain and SPC is representative of one experiment using 3 technical replicates from 3 mice in each group. h Lipidomics analysis in SOD1^G93A mouse tissues (P60) and hMNs^A4V (DIV14). Heat maps show fold changes in the levels of the indicated acylcarnitines (ACs), with significant fold changes (p<0.05) indicated with asterisks. This increase in C₁₄-C₁₈ ACs is more evident in brain mitochondria from SOD1^G93A mice (P60) compared to NTg controls. Conversely, these AC species were reduced in SPC mitochondria. Note that shorter ACs (C₂-C₈) were enriched in SPC and hMNs^A4V. Data was obtained from 3–5 biological replicates balanced by sex. Colored areas indicate statistical significance (p<0.05); Student’s two-tailed t-test at an α = 0.05 significance level.

Fig. 4. Defects in NADH levels and complex I activation in mitochondria from SOD1^G93A mice.

a NAD⁺:NADH ratio in hMNs^A4V at DIV14 relative to WT control average (set at 100%; dotted line) and SPC at P60 relative to NTg controls (set at 100%; dotted line). The columns represent the mean ± SD; n = 3 biologically independent samples. b Quantification of hydrogen peroxide (H₂O₂) produced during the oxidation of different substrates in mitochondria from SOD1^G93A mice relative to NTg controls (set at 100%; dotted line). The columns represent mean ± SD; brain: n = 4, SPC: n = 4. c Analysis of the D-to-A transition constant for CI, calculated and expressed as min⁻¹ (see Supplementary Fig. 3 for absolute values). The data are shown as the values in SOD1^G93A mice relative to NTg (set at 1). The columns represent the mean ± SD; brain: n = 3 (P15), n = 4 (P90); SPC: n = 4 (P15), n = 3 (P90). d Schematical representations of Complex I conformational changes during forward and reverse electron transfer (Created with BioRender released under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, OE26XW1WA2). Statistical test: Student’s two-tailed t-test for normal data or the Mann-Whitney U test for non-normal data at an α = 0.05 significance level. Statistical significance is shown as: n.s. p>0.05; * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001.

Fig. 5. Downregulation of MAM activity in familial and sporadic models of ALS.

a Scheme of MAM domains formed between the ER and mitochondria, showing the phospholipid synthesis and transfer assay (Created with BioRender released under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, WN26XW044I). b Quantification of MAM activity by the synthesis and transfer of phospholipids between the ER and mitochondria in crude mitochondria fractions from SOD1^G93A mouse SPC at the indicated ages. The columns represent the mean of PtdSer and PtdEtn content/μg crude mitochondria ± SD relative to controls (set at 100%; dotted line). n = 5 (P15), n = 3 (P30), n = 3 (P60), n = 3 (P90) biologically independent samples. c MAM activity via the phospholipid synthesis and transfer assay in hMNs^A4V (DIV14) at 2, 4, and 6 h. The columns represent the mean of the PtdEtn/PtdSer ratio of ALS vs WT controls (black bars set at 100%; dotted line). n = 3 (2 h), n = 4 (4 h), n = 4 (6 h). d MAM activity via the phospholipid synthesis and transfer assay at 6 h in iPSCs from sALS patients: Ctrl-1/sALS-1 and Ctrl-2/sALS-2 (see Supplementary Information for additional details about these samples). The columns represent the mean of the PtdEtn/PtdSer ratio of sALS vs healthy controls (black bars set at 1; dotted line); n = 3. Statistical test: Student’s two-tailed t-test at an α = 0.05 significance level. Statistical significance is shown as: n.s. p>0.05; * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001.

Fig. 6. Lipidomics analysis of mouse and human ALS tissues.

a, b Quantification of cholesterol and sphingomyelin levels in subcellular fractions from SOD1^G93A mouse brain and SPC tissues at pre-symptomatic stage P60 and in c, d postmortem frontal cortex samples from sALS and fALS patients with mutations in SOD1 (SOD1^G93A). In mouse samples, data are shown as the mean ± SD of n = 4 biologically independent samples balanced by sex. For human samples, data are shown as mean from 3–5 replicates consisting of a pool of tissues from 2 fALS cases with SOD1^G93A and 3 sALS cases (balanced by sex) normalized to control values ± SE. See Supplementary Information for additional details about these tissues. Statistical test: Student’s two-tailed t-test at an α = 0.05 significance level. Statistical significance is shown as: n.s. p>0.05; * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001.

Fig. 7. Proteomics changes in MAM fractions from NSC-34 cells expressing mutant SOD1.

a Scheme of PhotoClick-cholesterol methodology. b Proteomics analysis of the PhotoClick-cholesterol interactome in MAM fractions from NSC-34 cells expressing either SOD1^WT or SOD1^G93A. Heat maps show fold changes in the indicated proteins compared to non-transfected controls (Ctr), with statistically significant (p<0.05) fold changes indicated with asterisks. c Proteomics analysis of biotinylated proteins from cells expressing either myc-BioID2-SOD1^WT or myc-BioID2-SOD1^G93A. Left: spectral counts for the indicated proteins from MAM fractions are shown normalized to TH values for both SOD1 mutants. Right: the normalized MAM values from SOD1^G93A are normalized to the normalized MAM values from SOD1^WT. For all heat maps in this figure, the results show the mean of n = 3 biologically independent samples (colored areas indicate statistically significant fold changes; p<0.05, Student’s two-tailed t-test at an α = 0.05 significance level); red indicates increases and blue indicates decreases. Glycolytic enzymes: hexokinase-1 and -2 (Hk1, Hk2); glyceraldehyde 3-phosphate dehydrogenase (Gapdh); phosphoglycerate kinase (Pgk1); lactate dehydrogenase A (Ldha); phosphofructokinase (Pfkl, Pfkp); triose-phosphate isomerase 1 (Tpi1); pyruvate kinase isoform M (Pkm); fructose-bisphosphate aldolase A (Aldoa). Enzymes involved in metabolic adaptation: citrate synthase (Cs); phosphoenolpyruvate carboxykinase 2 (Pck2); glycerol-3-phosphate dehydrogenase (Gpd2); hydroxy-3-methylglutaryl-CoA lyase (Hmgcl); 3-ketoacyl-CoA thiolase A (Acaa1a). Proteins involved in oxidative stress: glutaredoxin-3 (Glrx3); thioredoxin (Txn); peroxiredoxins (Prdx1, 2 and 5); heat-shock protein 90 b1 (Hsp90b1). MAM-resident proteins: ER lipid raft associated 2 (Erlin2); oxysterol-binding protein like-8 (Osbpl8); acyl-CoA:cholesterol acyltransferase-1 (Acat1); acyl-CoA synthase long chain 4 (Acsl4); neutral sphingomyelinase 2 (nSMase2/Smpd3).

Fig. 8. Stimulation of MAM formation reverses mitochondrial phenotypes in ALS motor neurons.

a Quantification of OCR in hMNs^A4V using pyruvate as a mitochondrial substrate before and after SMase treatment compared to untreated WT cells (set at 100%). Data are shown as mean ± SD of n = 6 biologically independent samples. b PDHC activity in hMNs^A4V before and after SMase treatment compared to untreated WT cells (set at 100%). Data are shown as mean ± SD; n = 3. c HK activity in hMNs^A4V before and after SMase treatment compared to untreated WT cells (set at 100%). Data are shown as mean ± SE; n = 3. Statistical test: Student’s two-tailed t-test at an α = 0.05 significance level. Statistical significance is shown as: n.s. p>0.05; * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001.

Fig. 9. Model of mitochondrial metabolic dysfunction in ALS.

MAM downregulation shifts mitochondrial metabolism and precedes mitochondrial defects in SOD1-mutant models. a In healthy neurons, mitochondria associate with the ER, promoting pyruvate metabolism and NADH-driven oxygen consumption with relatively low ROS generation. Note that the cell maintains a high NAD⁺:NADH ratio to sustain the continuous production of NADH to fuel ATP production, and maintain CI working in forward mode. b In contrast, in the SPC of ALS mice, the dysfunction of MAM hinders pyruvate metabolism and induces a shift from pyruvate to other carbon sources such as FAs, promoting increases in succinate and FADH₂-driven respiration, and the transfer of electrons from CII to CoQ. Over time, this surge in FA oxidation increases the pool of reduced CoQ, triggering RET and subsequently higher levels of ROS. The radicals generated in this process differentially affect the kinetics of CI activation in SPC compared to the brain (Fig. S6). In this scenario, the SPC displays a low NAD⁺:NADH ratio (Created with BioRender released under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, HZ26XVZMOD).