Multimodal assessment of mitochondrial function in Parkinson's disease

Abstract

The heterogenous aetiology of Parkinson's disease is increasingly recognized; both mitochondrial and lysosomal dysfunction have been implicated. Powerful, clinically applicable tools are required to enable mechanistic stratification for future precision medicine approaches. The aim of this study was to characterize bioenergetic dysfunction in Parkinson's disease by applying a multimodal approach, combining standardized clinical assessment with midbrain and putaminal 31-phosphorus magnetic resonance spectroscopy (31P-MRS) and deep phenotyping of mitochondrial and lysosomal function in peripheral tissue in patients with recent-onset Parkinson's disease and control subjects. Sixty participants (35 patients with Parkinson's disease and 25 healthy controls) underwent 31P-MRS for quantification of energy-rich metabolites [ATP, inorganic phosphate (Pi) and phosphocreatine] in putamen and midbrain. In parallel, skin biopsies were obtained from all research participants to establish fibroblast cell lines for subsequent quantification of total intracellular ATP and mitochondrial membrane potential (MMP) as well as mitochondrial and lysosomal morphology, using high content live cell imaging. Lower MMP correlated with higher intracellular ATP (r = -0.55, P = 0.0016), higher mitochondrial counts (r = -0.72, P < 0.0001) and higher lysosomal counts (r = -0.62, P = 0.0002) in Parkinson's disease patient-derived fibroblasts only, consistent with impaired mitophagy and mitochondrial uncoupling. 31P-MRS-derived posterior putaminal Pi/ATP ratio variance was considerably greater in Parkinson's disease than in healthy controls (F-tests, P = 0.0036). Furthermore, elevated 31P-MRS-derived putaminal, but not midbrain Pi/ATP ratios (indicative of impaired oxidative phosphorylation) correlated with both greater mitochondrial (r = 0.37, P = 0.0319) and lysosomal counts (r = 0.48, P = 0.0044) as well as lower MMP in both short (r = -0.52, P = 0.0016) and long (r = -0.47, P = 0.0052) mitochondria in Parkinson's disease. Higher 31P-MRS midbrain phosphocreatine correlated with greater risk of rapid disease progression (r = 0.47, P = 0.0384). Our data suggest that impaired oxidative phosphorylation in the striatal dopaminergic nerve terminals exceeds mitochondrial dysfunction in the midbrain of patients with early Parkinson's disease. Our data further support the hypothesis of a prominent link between impaired mitophagy and impaired striatal energy homeostasis as a key event in early Parkinson's disease.

Keywords: 31phosphorus magnetic resonance spectroscopy; Parkinson’s disease; disease stratification; fibroblasts; mitochondria.

Figure 1

³¹P-MRS allows quantification of energy-rich metabolites in vivo. (A) Sagittal, (B) coronal and (C) axial images demonstrating spectroscopic grid (14 × 14) positioning for the midbrain voxels. Analysed voxels are highlighted in yellow and placement is performed to capture a voxel from the right and left side. Analysis focused on the mean values of the right and left voxels for each of the midbrain, posterior putamen and anterior putamen voxels. Voxel placement ensures the substantia nigra will be included within the voxel of interest. (D) Sagittal, (E) coronal and (F) axial images demonstrating spectroscopic grid (12 × 12) positioning for the putaminal voxels. Analysed voxels are highlighted in yellow and placement is performed to capture a voxel from the right and left side for both the anterior and posterior putamen. (G) An example spectrum obtained from the midbrain of a healthy control. This spectrum has been phased and apodized to aid visualization with phosphocreatine frequency shifted to 0 ppm, for quantification spectra are not apodized. (H) An example output of peak fitting following AMARES analysis. Thirteen resonances are fitted: 1 = phosphocholine; 2 = phosphoethanolamine; 3 = Pi (inorganic phosphate); 4 = glycerophopshocholine; 5 = glycerophosphoethanolamine; 6 = phosphocreatine; 7 and 8 = γ-ATP; 9 and 10 = α-ATP; 11–13 = β-ATP. Phosphocholine and phosphoethanolamine form the phosphomonoesters (PME). Glycerophosphocholine and glycerophosphoethanolamine form the phopshodiesters (PDE). PME and PDE are only quantified to facilitate normalization of each amplitude to the total phosphorus signal detected in the spectra and were not used in any statistical analysis.

Figure 2

Considerably greater variance of mitochondrial and lysosomal function in fibroblasts derived from patient with Parkinson's disease. (A) Intracellular ATP, F-test P = 0.0056; (B) MMP, F-test P < 0.0001; (C) long mitochondria MMP only, F-test P = 0.0007; (D) short mitochondria MMP, F-test P = 0.0003; (E) mitochondria count per cell, F-test P < 0.0001; (F) lysosomal count per cell, F-test P < 0.0001. Mean (purple diamond) ± standard deviation (SD) presented. Blue dashed lines denote 2 SDs from the control mean values. The parkin (PRKN^−/−) mutant patient is depicted with a green diamond, the SNCA-G51D mutant patient is depicted with a purple square. All fibroblast assays repeated in triplicate. Data presented are from experiments performed in glucose-containing media and all data from each participant are normalized to mean control values per repeat. Variances between groups were tested with the F-test of equality of variances and statistical significance is displayed on each panel. Control: n = 23, Parkinson's disease: n = 34. *P < 0.05, **P < 0.01, ***P < 0.001. MMP = mitochondrial membrane potential; PD = Parkinson’s disease.

Figure 3

Analysis of mitochondrial and lysosomal function suggests impaired mitophagy and uncoupling in Parkinson's disease patient fibroblasts. Fibroblast measures of mitochondrial and lysosome function/morphology; correlations between measures for both Parkinson's disease and controls assess using Pearson's correlation coefficient. (A) Intracellular ATP and MMP (control r = −0.044, P = 0.918; Parkinson's disease r = −0.552, P = 0.0018); (B) MMP and mitochondria per cell (control r = −0.379, P = 0.0895; Parkinson's disease r = −0.723, P < 0.0001); (C) MMP and lysosomes per cell (control r = −0.205, P = 0.432; Parkinson's disease r = −0.623, P = 0.0003); (D) mitochondria per cell and lysosomes per cell (control r = 0.43, P = 0.0603; Parkinson's disease r = 0.7, P < 0.0001). (E) Long mitochondria MMP (peach) and short mitochondria MMP (turquoise) and lysosomes per cell in patients only (Parkinson's disease long mitochondria MMP r = −0.65, P < 0.0001; Parkinson's disease short mitochondria MMP r = −0.73, P < 0.0001). (F) Intracellular ATP and mitochondria per cell (control r = 0.548, P = 0.0137; Parkinson's disease r = 0.719, P < 0.0001). All fibroblast assays repeated in triplicate. Data presented are from experiments performed in glucose-containing media and all data from each participant are normalized to mean control values per repeat. The parkin (PRKN^−/−) mutant patient is depicted with a green diamond, the SNCA-G51D mutant patient is depicted with a purple square. Control (blue) n = 23, Parkinson's disease (red) n = 34. The P-values reported are adjusted for multiple comparisons using the Benjamini-Hochberg method. *P < 0.05, **P < 0.01, ***P < 0.001. MMP = mitochondrial membrane potential; PD = Parkinson’s disease.

Figure 4

Putaminal, but not midbrain Pi/ATP ratios correlate with indices of mitophagy in Parkinson's disease patient-derived fibroblasts. All ³¹P-MRS data are expressed as composite z-scored data of all putaminal voxels. All fibroblast assay data are expressed as a composite z-score of triplicate repeats in both glucose-containing and galactose-containing media. Pearson's correlation coefficient was used for all analyses. The parkin (PRKN^−/−) mutant patient is depicted with a green diamond, the SNCA-G51D mutant patient is depicted with a purple square. (A) Putaminal Pi/ATP ratio and mitochondria per cell (control r = −0.326, P = 0.13; Parkinson's disease r = 0.369, P = 0.0319). (B) Putaminal Pi/ATP ratio and MMP in short mitochondria only (control r = −0.345, P = 0.107; Parkinson's disease r = −0.522, P = 0.0016). (C) Putaminal Pi/ATP ratio and MMP in long mitochondria only (control r = −0.249, P = 0.252; Parkinson's disease r = −0.469, P = 0.0052). (D) Putaminal Pi/ATP ratio and lysosomes per cell (control r = −0.073, P = 0.741; Parkinson's disease r = 0.476, P = 0.0044). (E) Midbrain Pi/ATP ratio and mitochondria per cell (control r = 0.023, P = 0.92; Parkinson's disease r = 0.018, P = 0.919). (F) Midbrain Pi/ATP ratio and MMP in short mitochondria only (controls r = 0.245, P = 0.271; Parkinson's disease r = 0.016, P = 0.93). (G) Midbrain Pi/ATP ratio and MMP in long mitochondria only (controls r = 0.308, P = 0.164; Parkinson's disease r = 0.124, P = 0.484). (H) Midbrain Pi/ATP ratio and lysosomes per cell (controls r = 0.183, P = 0.414; Parkinson's disease r = 0.345, P = 0.0457). Putamen data: controls n = 23, Parkinson's disease n = 34. Midbrain data: controls n = 22, Parkinson's disease n = 34. MMP = mitochondrial membrane potential; PD = Parkinson’s disease.

Figure 5

Increased variance of putaminal, but not midbrain Pi/ATP in Parkinson's disease. ³¹P-MRS in Parkinson's disease compared to controls. All ³¹P-MRS values are normalized to total phosphorus signal detected in the spectra. (A) Mean posterior putamen Pi/ATP ratio, F-test P = 0.0036; (B) mean posterior putamen Pi, F-test P = 0.0042; (C) mean posterior putamen ATP, F-test P = 0.5624. (D) Mean midbrain inorganic phosphate (Pi)/ATP ratio, F-test P = 0.1708; (E) mean midbrain Pi, F-test P = 0.3674; (F) mean midbrain ATP, F-test P = 0.0030. Mean (purple diamond) ± standard deviation (SD). Variances between groups were tested with the F-test of equality of variances and statistical significance is displayed on each panel. Blue dashed lines denote 2 SDs from the control mean values. The parkin (PRKN^−/−) mutant patient is depicted with a green diamond, the SNCA-G51D mutant patient is depicted with a purple square. Midbrain data: controls n = 24, Parkinson's disease n = 35. Putamen data: controls n = 25, Parkinson's disease n = 35. *P < 0.05, **P < 0.01. Pi = inorganic phosphate; ns = not significant.