Imaging brain glucose metabolism in vivo reveals propionate as a major anaplerotic substrate in pyruvate dehydrogenase deficiency

Abstract

A vexing problem in mitochondrial medicine is our limited capacity to evaluate the extent of brain disease in vivo. This limitation has hindered our understanding of the mechanisms that underlie the imaging phenotype in the brain of patients with mitochondrial diseases and our capacity to identify new biomarkers and therapeutic targets. Using comprehensive imaging, we analyzed the metabolic network that drives the brain structural and metabolic features of a mouse model of pyruvate dehydrogenase deficiency (PDHD). As the disease progressed in this animal, in vivo brain glucose uptake and glycolysis increased. Propionate served as a major anaplerotic substrate, predominantly metabolized by glial cells. A combination of propionate and a ketogenic diet extended lifespan, improved neuropathology, and ameliorated motor deficits in these animals. Together, intermediary metabolism is quite distinct in the PDHD brain-it plays a key role in the imaging phenotype, and it may uncover new treatments for this condition.

Keywords: brain; glucose; imaging; ketogenic diet; metabolism; propionate; pyruvate; pyruvate dehydrogenase deficiency.

Figure 1.. Structural and molecular characterization of the Pdha1^hGFAP brain.

(A) Immunofluorescence staining to visualize Cre-recombinase expression indicated by GFP in a P21 hGFAP-cre;Rosa26^GFP reporter mouse brain. Scale bar: 2 mm. (B) Cresyl violet staining on sagittal sections of control and Pdha1^hGFAP mouse brain. Scale bar: 2 mm. (C) Western blot and quantification of Pdha1 expression and total PDH activity in control and Pdha1^hGFAP cortex, hippocampus, and cerebellum (N=3/group). Western blot of Pdha1 in granule cell neurons and glial cells from P7 cerebella of Pdha1^hGFAP and control mice. Molecular weight markers are displayed in Kd. (D) Tunel staining and quantification of Tunel positive area normalized to the Dapi area in cortex, hippocampus, and cerebellum (N=3/group). (E) Immunofluorescence staining and quantification of neuronal density, as reported by NeuN, and density of reactive astrocytes, as reported by cell size of GFAP⁺ cells, in cortex (scale bar: 200 μm), hippocampus (scale bar: 200 μm), and cerebellum (scale bar: 50 μm) (N=5/group). GCL: granule cell layer. Unpaired Student t-test was used to carry out the statistical analysis hereafter unless specified otherwise. Values are expressed in mean ± S.E.M in all figures unless indicated otherwise. *: p< 0.05, **: p< 0.01, ***: p< 0.001.

Figure 2.. Brain MRI, ¹H-MRS, and ¹³C-Hyperpolarized MRI.

(A) Sagittal and coronal T₂-weighted MRI of control and Pdha1^hGFAP brain. Yellow squares depict where voxels were placed to perform ¹H-MRS analysis. Yellow and orange arrows illustrate T_2-weighted hyperintensities in the cortex and hippocampus in the Pdha1^hGFAP mouse brain, respectively. (B) PCA analysis of animal groups and individual metabolites. (C) Representative ¹H-MRS of cortex, hippocampus, and cerebellum normalized to creatinine and quantification of peak intensity areas for each metabolite (N=8/group). A Welch T-test with Benjamini-HochBerg correction for multiple comparison analysis was performed, and p-adjusted values were reported. (D) Schematic of ¹³C-hyperpolarized (HP) pyruvate metabolism in the cell and quantification of bicarbonate:lactate ratio are illustrated (Control: N=5, Pdha1^hGFAP: N=4).

Figure 3.. Gliovascular unit and expression of glucose transporters.

(A) Schematic of the gliovascular unit and immunostaining of blood vessels and astrocytes in the hippocampus using GLUT1 and GFAP markers, respectively (3 sections/animal, 3 animals/group). Scale bars: 20 μm. (B) Immunofluorescence staining of blood vessels using GLUT1 marker (Bar = 200 μm), Imaris 3D reconstruction of blood vessels (Bar = 20 μm), and quantification of average and total blood vessel diameter in cortex, hippocampus, and cerebellum (N = 5/group). (C) Quantitative Western blot analysis of GLUT1 and GLUT3 in cortex, hippocampus, and cerebellum (N = 5/group).

Figure 4.. FDG-PET/CT and ¹⁴C-2DG autoradiography analyses from control and Pdha1^hGFAP brains.

(A) Schematic showing metabolism of FDG and ¹⁴C-2DG in neurons and astrocytes. (B) Representative coronal and sagittal FDG-PET/CT brain images of control and Pdha1^hGFAP mice and brain atlas–guided PET quantification in standardized uptake value (SUV) (Control: N=8, Pdha1^hGFAP: N=6). (C) Autoradiography and cresyl violet staining of coronal sections from both groups and quantification of ¹⁴C-2DG signal intensity (N=4/group). Cereb: cerebellum, Hip: hippocampus, DCN: deep cerebellar nuclei.

Figure 5.. Analysis of [U-¹³C]-glucose metabolism in the brain.

(A) Design of the experiment. (B) Analysis of metabolites pool size based on mass spectrometry signal intensity in blood, cortex, hippocampus, and cerebellum and Pdha1^hGFAP:control ratio of signal intensities (N=7/group). (C) Heatmaps of Pdha1^hGFAP:control fold change of isotopologue enrichments of metabolites identified in blood, brain, cortex, hippocampus, and cerebellum. Scale restricted from 1 to −1 for illustration purposes. (D) Average of Pdha1^hGFAP:control fold change of total ¹³C enrichment of metabolites in the brain. Cluster dendrogram demarcating metabolite groups based on labeled/total pools. (E) Isotopologue analysis of brain metabolites in the Pdha1^hGFAP brain relative to the control brain expressed as fold changes. (F) Schematic of ¹³C enrichments of glycolytic metabolites and the first turn of the Krebs cycle metabolites in the Pdha1^hGFAP brain relative to the control expressed as fold changes. 3-PG: 3-phosphoglycerate, OAA: oxaloacetate, α-KG: α-ketoglutarate.

Figure 6.. Analysis of [U-¹³C]-propionate metabolism in the brain.

(A) Design of the [U-¹³C]propionate experiment. (B) Heatmap of Pdha1^hGFAP:control fold change of isotopologue enrichment of metabolites identified in cortex, hippocampus, and cerebellum (Control: N=7, Pdha1^hGFAP: N=6). (C) Pdha1^hGFAP : control fold change of M3 enrichment of metabolites in the brain. (D) Isotopologue analysis of brain metabolites in the Pdha1^hGFAP brain relative to the control brain expressed in fold change. (E) Representative ¹³C-NMR spectrum from control and Pdha1^hGFAP mice. Demarcated regions in each spectrum (star and parenthesis) are expanded in panel F. (F) ¹³C-NMR spectrum and isotopomer analysis of glutamate, glutamine, and aspartate C3. (G) Isotopologue fractional enrichment of CoAs and ratios of M3 enrichments between Pdha1^hGFAP and control brains (H) Schematic of the proposed compartmentalized propionate metabolism in astrocytes and neurons. Glut: glutamate, Gln: glutamine, OAA: oxaloacetate, α-KG: α-ketoglutarate, Asp: aspartate, Ala: alanine, Tau: taurine, NAA: N-acetyl-aspartate. C#: carbon labeled in position #. Sx: singlet, Dxx: doublet, Q: quartet.

Figure 7.. Analysis of dietary propionate on survival, weight gain, pathological findings, and motor dysfunction of Pdha1^hGFAP mice.

(A) Schematic of the three dietary regimens administered during gestation and the postnatal period, and the metabolic pathways targeted with each diet. (B) Survival curves of Pdha1^hGFAP mice on chow diet (N=7), ketogenic diet (KD) (N=8), KD with sodium propionate 50 mM on the drinking water (KDP) (N=8), and chow diet with 150 mM of sodium propionate (Prop) (N=7). Log-rank statistical test was used to compare the survival distributions of animals among experimental diets. (C) Analysis of weight gain during the first 30 postnatal days of Pdha1^hGFAP mice subjected to each dietary condition (Chow, N=5, KD, N=9, KDP, N=8, Prop, N=7). A pairwise t-test analysis followed by Benjamini-Hochberg p-value correction was used hereafter in this figure to compare the chow diet group with other experimental dietary groups individually. (D) Analysis of neuronal density as defined by NeuN⁺ cells in each region (normalized to the wildtype group of each experimental condition for comparison purposes), and reactive astrocytosis, denoted by GFAP⁺ cells of sizes established in Pdha1^hGFAP mice on a regular diet, within the cortex, hippocampus, and cerebellum (Chow, N=7, KD, N=5, KDP, N=4, Prop, N=3). Scale bars in cortex (100 μm), hippocampus (200 μm), and cerebellum (50 μm). (E) Analysis of lateral body sway (pixels) of both control and Pdha1^hGFAP mice while walking on an open field platform (Chow, N=9, KD, N=9, KDP, N=8, Prop, N=9). GCL: granule cell layer. NS: not significant.