Regional metabolic signatures in the Ndufs4(KO) mouse brain implicate defective glutamate/α-ketoglutarate metabolism in mitochondrial disease

Abstract

Leigh Syndrome (LS) is a mitochondrial disorder defined by progressive focal neurodegenerative lesions in specific regions of the brain. Defects in NDUFS4, a subunit of complex I of the mitochondrial electron transport chain, cause LS in humans; the Ndufs4 knockout mouse (Ndufs4(KO)) closely resembles the human disease. Here, we probed brain region-specific molecular signatures in pre-symptomatic Ndufs4(KO) to identify factors which underlie focal neurodegeneration. Metabolomics revealed that free amino acid concentrations are broadly different by region, and glucose metabolites are increased in a manner dependent on both region and genotype. We then tested the impact of the mTOR inhibitor rapamycin, which dramatically attenuates LS in Ndufs4(KO), on region specific metabolism. Our data revealed that loss of Ndufs4 drives pathogenic changes to CNS glutamine/glutamate/α-ketoglutarate metabolism which are rescued by mTOR inhibition Finally, restriction of the Ndufs4 deletion to pre-synaptic glutamatergic neurons recapitulated the whole-body knockout. Together, our findings are consistent with mTOR inhibition alleviating disease by increasing availability of α-ketoglutarate, which is both an efficient mitochondrial complex I substrate in Ndufs4(KO) and an important metabolite related to neurotransmitter metabolism in glutamatergic neurons.

Keywords: Genetics; Leigh syndrome; Metabolism; Mitochondria; Mouse; Rapamycin; Reactive oxygen species.

Figure 1 –. Regional Mitochondrial Energetic Capacity.

(A) Schematic of the mitochondrial electron transport chain (ETC). (B) Schematic of the histochemical staining method for ETC complex IV (CIV) activity using the DAB reaction. (C) Representative sagittal brain section from a 28-day old control mouse stained for ETC CIV activity. (D-H) Magnified images of cerebellum (D), olfactory bulb (E), brainstem (F), hippocampus (G), and cortex (H) from the section in (C). (I) Representative sagittal brain section from a 28-day old Ndufs4(KO) mouse stained for ETC CIV activity. (J-N) Magnified images of cerebellum (J), olfactory bulb (K), brainstem (L), hippocampus (M), and cortex (N) from the section in (I). (O) Schematic of brain sectioning for DNA damage assessment by qPCR. (P) Quantification of long-range PCR of the nuclear gene beta-globulin in DNA isolated from key brain regions in control and Ndufs4(KO) mice. All results are normalized to the mean values for control samples from the respective regions. (Q) Quantification of long-range qPCR targeting a mitochondrial DNA encoded 10kb fragment in DNA isolated from key brain regions in control and Ndufs4(KO) mice. All results are normalized to the mean values for control samples from the respective regions. (P-Q) Data presented are the means of 3 technical repeats for each of 3 biological samples per genotype. No statistically significant differences observed between control and Ndufs4(KO) samples. In this figure and all subsequent figures; Ctl RB= control resistant brain region, KO RB= knockout resistant brain region, Ctl CB = control cerebellum, KO CB = knockout cerebellum, Ctl BS = control brainstem, KO BS= knockout brainstem, Ctl OB= control olfactory bulb, KO OB= knockout olfactory bulb. Values are given as the means + standard errors of the mean (SEM).

Figure 2 –. Region Specific mTOR Activity in Pre-Lesion CNS .

Western blots and quantification of the phosphorylation of ribosomal protein S6 (rpS6), a downstream target of mTORC1, in the lesion resistant brain region (A), brainstem (B), cerebellum (C), and olfactory bulb (D) in ad libitum fed pre-disease onset (P25–28) animals. Western blots and quantification of the phosphorylation of AKT at serine 473 (S473), a downstream target of mTORC2, in the lesion resistant brain region (E), brainstem (F), cerebellum (G), and olfactory bulb (H). For histograms, N=4 in all cases. Ctl = wild type without rapamycin, KO = Ndufs4(KO) without rapamycin, Ctl Rapa = wildtype treated with rapamycin, KO Rapa = Ndufs4(KO) treated with rapamycin. Dots in histograms represents 4 biological samples (even where the adjacent blot shows only 3 of the samples). (A) p<0.0001, one-way ANOVA. (B) p=0.001, one-way ANOVA. (C) p<0.0001, one-way ANOVA. (D) p=0..0007, one-way ANOVA. (E-H) p>0.05, one-way ANOVA, groups not significantly different. (A-H) * designates p<0.05. ** p<0.005, ***p≤0.0005 by pairwise t-test.

Figure 3 –. Region-Specific Metabolomics in CNS of Control and Ndufs4(KO) Mice.

(A) Schematic of brain regions physically separated and analyzed by metabolomics. (B) Metabolites in control brain samples that show significantly different abundance between the lesion ‘resistant’ region and each of the lesion associated regions, with the same direction of change in each pairwise comparison. Relative fold change (larger squares) and nominal p-values (inset rectangles) of each comparison to the ‘resistant’ region are indicated by heat maps. (C) Metabolites in Ndufs4(KO) brain samples that show significantly different abundance between the lesion ‘resistant’ region and each of the lesion associated regions, with the same direction of change in each pairwise comparison. Relative fold change (larger squares) and nominal p-values (inset rectangles) of each comparison to the ‘resistant’ region are indicated by heat maps. (D-G) Metabolites significantly higher in the ‘resistant’ region, versus the lesion associated regions, in both the control and Ndufs4(KO) datasets (see also Supplemental Figure 2). (H-I) Metabolites significantly lower in the ‘resistant’ region, versus the lesion associated regions, in both the control and Ndufs4(KO) datasets. (J-O) Additional neurotransmitters appearing in these metabolomics datasets. (D-I) ANOVA p<0.0001 for each of these metabolites. Pairwise t-tests between each sensitive and the resistant region appear in (B-C). No significant genotype differences observed except in (G) and (M), where indicated (*p<0.05, **p<0.005 by pairwise t-test). (J) ANOVA p=0.0075, no significant differences by genotype. (M) ANOVA p=0.48, no significant differences by genotype. (N) ANOVA p=0.026, no significant differences by genotype. (O) ANOVA p=0.0119, no significant differences by genotype. For all panels (D-O) dots represent biological replicates (N>4), histograms and error bars designate median values and standard error of the mean. Ctl RB= control resistant brain region, KO RB = knockout resistant brain region, Ctl CB = control cerebellum, KO CB = knockout cerebellum, Ctl BS = control brainstem, KO BS= knockout brainstem, Ctl OB= control olfactory bulb, KO OB= knockout olfactory bulb.

Figure 4 –. Regional Metabolic Profiles of Control and Ndufs4(KO) Mice.

(A-B) 3-dimensional principal component analysis (PCA) of brain region metabolomics data with region (A) or genotype (B) indicated (see Methods). (C) Projection of first two major principal components with 95% confidence intervals of groups indicated (colored ellipses). Brain region is the major determinant of sample clustering by PCA. (D) Unsupervised clustering of brain region metabolite level data in control and Ndufs4(KO) mice. (E) Cluster 1, which includes the majority of amino acids. (C-E) Glycolysis intermediates appearing in cluster 4. ANOVA p-values: (F) glucose p<0.0001, (G) DHAP p<0.0001, (H) G-1-P/G-6-P/F-6-P/F-1-P p<0.0001. *p<0.05 and **p<0.005 by two-tailed pairwise t-test. (I) Lactate, which appears in cluster 5. ANOVA p=0.0008. (J-N) Glycolysis intermediates appearing in cluster 6. ANOVA p-values: (J) D-GA3P p<0.0001, (K) 2/3 phosphoglycerate p<0.0001, (L) F-1,6-BP/F-2,6-BP/G-1,6-BP, p<0.0001 (M) glycerol-3-P, Not significant, (N) pyruvate p<0.008. For all panels (F-N) individual points represent biological replicates (individual mice), N>4 per group; histograms and error bars designate median values and standard error of the mean. Ctl RB= control resistant brain region, KO RB = knockout resistant brain region, Ctl CB = control cerebellum, KO CB = knockout cerebellum, Ctl BS = control brainstem, KO BS= knockout brainstem, Ctl OB= control olfactory bulb, KO OB= knockout olfactory bulb.

Figure 5 –. Region-Specific Metabolic Impact of mTOR Inhibition.

(A-B) Unsupervised clustering of brain region metabolite levels in untreated and rapamycin treated control (A) and Ndufs4(KO) (B) mice. Metabolites appearing in cluster 1 in both control and Ndufs4(KO) datasets indicate a rapamycin-responsive glutamate/glutamine/α-ketoglutarate metabolite group. (C) α-ketoglutarate levels by brain region, genotype, and treatment. ANOVA *p=0.01. (D) Data-reduced α-ketoglutarate data using median values for each brain region, revealing genotype and treatment dependent, region independent, changes. ANOVA ***p<0.0001. (E) Glutamine levels by brain region, genotype, and treatment. ANOVA **p=0.005. (F) Data-reduced glutamine data using median values for each brain region, revealing genotype and treatment dependent, region independent, changes. ANOVA ***p<0.0001. (G) Glutamate levels by brain region, genotype, and treatment. ANOVA **p<0.005. (H) Data-reduced glutamate data using median values for each brain region, revealing genotype and treatment dependent, region independent, changes. ANOVA **p=0.0055. (I) GABA levels by brain region, genotype, and treatment. ANOVA ***p<0.0001. (J) Data-reduced GABA data using median values for each brain region, revealing genotype and treatment dependent, region independent, changes. Not significantly different by ANOVA. (K) Pyroglutamate levels by brain region, genotype, and treatment. ANOVA *p=0.03. (L) Data-reduced pyroglutamate data using median values for each brain region, revealing genotype and treatment dependent, region independent, changes. ANOVA **p=0.004. (M) Symbol key for panels D, F, H, J, and L; triangles, squares, circles, and diamonds represent the median values (across biological replicates) for ‘resistant brain’ (RB), cerebellum (CB), brainstem (BS), and olfactory bulb (OB), respectively. (C, E, G, I, K) Dots represent biological replicates (N>=4); columns and error bars designate median values and standard error of the mean. (D, F, H, J, L) Dots represent median values (among biological replicates) for each brain region. Columns represent median values among regions within the same genotype and treatment, and error bars represent standard error of the mean. (C-L) pairwise t-test p-values * designates p≤0.05, ** designates p≤0.005, and *** designates p≤0.0005. (N) Schematic of the glutamine/glutamate/α-ketoglutarate metabolite group with key CNS functions of each glutamate metabolite indicated. Ctl RB= control resistant brain region, KO RB= knockout resistant brain region, Ctl CB = control cerebellum, KO CB = knockout cerebellum, Ctl BS = control brainstem, KO BS= knockout brainstem, Ctl OB= control olfactory bulb, KO OB= knockout olfactory bulb. For (D,F,H,J,L) Ctl and KO represent untreated wild type and Ndufs4(KO) averages, respectively; Ctl R and KO R represent rapamycin treated wild type and Ndufs4(KO) averages, respectively.

Figure 6 –. Neuronal Effects of Ndufs4 Deficiency.

(A-C) VGlut2 specific knockout of Ndufs4 leads to shortened lifespan (A), progressive weight loss after approximately post-natal day 37, and (C) onset of neurologic symptoms by approximately P40 roughly equivalent to that observed in the whole-body Ndufs4(KO) mouse. (A) Log-rank test p=0.1. (B) No significant differences at any age. (C) Log-rank test ***p<0.0001. (D) Western blot analysis of whole-brain lysates from Ndufs4(KO) and control animals as a function of age. (E-N) Quantification of protein abundance. (E, G, I, K, M) Analysis of genotypes with all ages pooled, *p<0.05, **p<0.005 by pairwise t-test. (F, H, J, L, N) Scatter-plots of target abundance by genotype and age. GAD1 (Glutamate Decarboxylase 1); GAD2 (Glutamate Decarboxylase 2); MBP (myelin basic protein). (O-P) VGlut2-Cre promoter driven ZsGreen (Ai6 construct) expression in P60 control (O) and Ndufs4(KO) (whole-body knockout) (P) brains. Green = ZsGreen, blue = DAPI (DNA dye). Arrows indicate known lesion-associated areas of the cerebellum (right) and brainstem (left). An advanced lesion is visible in the brainstem region of the Ndusf4(KO) sample (P), as determined by the abnormal accumulation of nuclei (DAPI, blue) at this location. No overt changes to VGlut2-Cre expressing cell populations are present in the Ndufs4(KO) compared to control. Representative images, n>3 brains imaged per genotype. (Q-R) Higher magnification imaging of brainstem (Q-R) and cerebellum (S-T) in these control (Q, S) and Ndusf4(KO) (R, T) brains. In each set of panels, a-ZsGreen, b-DAPI, c-overlay.