Delineating the mechanisms of cerebellar degeneration in paediatric and adult primary mitochondrial disease

Clinical details

In a cohort of 310 adult patients with PMD (n = 22 bi-allelic POLG, n = 248 m.3243A > G, n = 34 m.8344A > G, n = 6 m.14709T > C, Supplementary Table 1), 73 patients (23.5%) had predominantly cerebellar ataxia, assessed by the NMDAS,23 at the first appointment (Table 2). In comparison to the most recent follow-up assessment, cerebellar ataxia had progressed in 38.8% of patients, and a total of 123 of 310 patients (39.7%) presented with cerebellar ataxia. Mixed sensory and cerebellar ataxia was common in the POLG disease patient group, affecting 36.4% and 81.8% of patients at the first and most recent assessment, respectively.

Table 2 Prevalence of ataxia in adult patients with PMD

The PMD post-mortem tissue cohort consisted of 13 patients with Alpers’ syndrome (Patient 1 – 13), five adult patients with bi-allelic late-POLG disease (Patient 14–18), and ten adult patients with mtDNA disease (Patient 19–28: m.3243A > G, n = 4; m.8344A > G, n = 4; m.13094T > C, n = 1; m.14709T > C, n = 1) (Table 1). The mean age of disease onset was significantly younger (P < 0.0001), and the duration of disease was significantly shorter (P < 0.001), in the Alpers’ syndrome patient group (mean onset: 5.3 years, mean duration: 4.9 years) in comparison to the late-POLG and mtDNA disease group (mean onset: 26.9 years, mean duration: 19.3 years).

In this patient cohort, 19 of 28 patients displayed clinical signs of ataxia, including 7 patients with Alpers’ syndrome and 12 adult patients with POLG and mtDNA PMD. However, ataxia may not have been assessed in young infants who had not yet acquired key motor developmental milestones. Refractory epilepsy, which included frequent episodes of status epilepticus, was a defining feature of all patients with Alpers’ syndrome. Developmental delay and/or psychomotor regression, cortical blindness, and hepatic dysfunction were also common in these younger patients. In the adult PMD cohort, two of five patients with bi-allelic POLG-related disease and eight of ten patients with mtDNA disease had epilepsy. Stroke-like episodes were confirmed through the identification of cerebral stroke-like lesions in nine patients, and subacute signal abnormalities involving the cerebellar cortex were reported in two patients (Table 1).

Cerebellar neuropathology

In line with previous neuropathological studies [8, 19], assessment of post-mortem cerebellar tissues revealed variable degrees of degeneration in the PMD tissue cohort involving a decreased abundance of Purkinje cells, thinning of the molecular cell layer, and depletion of the granular cell layer (Fig. 1A–B, Supplementary Table 6). Degeneration of Purkinje cells was typically accompanied by Bergmann gliosis and microgliosis predominantly affecting both the molecular cell layer and Purkinje cell layer. The white matter mostly appeared normal, except for regions affected by severe neurodegeneration where loss of myelin, evident by decreased Luxol fast blue staining, and vacuolation in the subcortical white matter were observed (Fig. 1B). Assessment of the deep cerebellar nuclei demonstrated an apparent moderate loss of neurons and gliosis in the dentate nucleus, although dentate tissue was only available for analysis in 15 cases.

Fig. 1figure 1

Degeneration of the cerebellar cortex in patients with mitochondrial disease. A Cresyl fast violet (CFV) stain reveals intact cerebellar cortex and preserved Purkinje cell density in control tissue (iii). However, cerebellar tissue from a patient with Alpers’ syndrome (Pt.03, iiiiv) demonstrates severe thinning of the molecular cell layer, loss of Purkinje cells and granular cell layer depletion. B Luxol fast blue, Haematoxylin and Eosin (LFB H&E) stain reveals intact myelin (blue) in control tissue (i). However, cerebellar tissue from a young adult patient with POLG-related disease (Pt.13, ii) demonstrates spongiosis in the subcortical white matter. (iii) A focal necrotic lesion characterised by total Purkinje cell loss and severe thinning of the cortical layers is presented from an adult patient with late-POLG disease (Pt.15). Preserved cortex (white box—iv) is outlined in the cerebellar folia adjacent to the lesioned cortex (yellow box—v). Scale bars = 100 μm. C Purkinje cell densities and D molecular cell layer width measurements (n = 20 per case) were analysed using a linear regression model (C) and linear mixed regression model (D) respectively. Grey bar and asterisk indicate comparisons where data from all patients with Alpers’ syndrome are pooled, and all late-onset POLG and mtDNA patients are pooled and compared to age-matched control groups. Black bar and asterisk indicate further comparisons splitting the Alpers’ syndrome patients into two groups based on age, and comparisons between the late-onset POLG group to controls, and mtDNA group to controls. *P < 0.05, **P < 0.01, ***P < 0.001

Qualitative assessment of cerebellar tissues revealed visibly more severe atrophy of the cerebellar cortex in patients with Alpers’ syndrome relative to age-matched controls (Fig. 1A iii), in comparison to adult patients with POLG-related disease and mtDNA disease relative to adult controls (Fig. 1B ii, iii). Therefore, we sought to characterise and compare the severity of neuropathological features in cerebellar tissues across the PMD patient groups, with the aim of identifying shared and distinct pathomechanisms of cerebellar degeneration.

Decreased Purkinje cell density and molecular cell layer width

Quantification of the density of Purkinje cells in control and PMD patient tissues revealed a significantly decreased abundance of Purkinje cells (P < 0.05) in the Alpers’ syndrome, late-POLG and mtDNA disease patient groups relative to age-matched controls (Fig. 1C). Interestingly, Purkinje cell loss tended to be more severe in PMD patients with epilepsy. We also quantified the surface area of individual Purkinje cells which revealed in cerebellar tissues from patients with Alpers’ syndrome and mtDNA disease, and Purkinje cells were significantly smaller compared to the size of control Purkinje cells suggestive of atrophy (P < 0.05) (Supplementary Fig. 3). Quantification of the width of the molecular cell layer revealed a significantly thinner layer in all PMD patient groups compared to controls (P < 0.01). However, the density and size of Purkinje cells and the width of the molecular cell layer did not differ significantly between different PMD patient groups (P > 0.05).

Focal lesioned cerebellar cortex is characterised by decreased parvalbumin immunoreactivity

Focal necrotic lesions in the cerebellum were identified in eight patients including four patients in the Alpers’ syndrome group, two adult patients with late-POLG disease, and two adult patients with mtDNA disease, all of whom had epilepsy (Table 3 and Supplementary Table 6). These lesions were characterised by demarcated regions of the cerebellar cortex affected by severe neurodegeneration, typically affecting all three cortical layers. In all cases, these lesions were identified by severely decreased parvalbumin-positive immunoreactivity in comparison to adjacent non-lesioned preserved cortex with intact parvalbumin staining (Fig. 2, Supplementary Figs. 4 and 5). This novel observation suggests that parvalbumin is a sensitive and reliable marker in the neuropathological identification of focal necrotic lesions in the cerebellar cortex of patients with PMD.

Table 3 Summary of cerebellar neuropathological features in PMD patient tissuesFig. 2figure 2

Pathological characterisation of focal lesions in the cerebellar cortex in POLG-related disease. Slide-scanned serial cerebellar cortical sections from a young adult patient (Pt.13) harbouring pathogenic variants in POLG (p.[Ala467Thr]/p.[Trp748Ser]) demonstrate focal lesions (example identified by dashed black box) characterised by: total Purkinje cell loss and decreased parvalbumin immunoreactivity (a), increased glial fibrillary acidic protein (GFAP; marker of reactive astrocytes) immunoreactivity (b), and increased HLA-DR (marker of activated microglia) immunoreactivity (c), in comparison to adjacent non-lesioned cortex. Increased c-Fos immunoreactivity (marker of neuronal hyperactivity) is observed in multiple cerebellar folia, labelling demarcated non-lesioned regions of the granular cell layer, molecular cell layer, and Purkinje cells (d). Blue arrow heads (d) indicate the region of cerebellar cortex which was imaged at a higher magnification. All sections were counterstained with Haematoxylin. Scale bars = 100 μm

Comparison of PMD patient tissues revealed a striking increase in GFAP and HLA-DR immunoreactivity localised to the focal lesions in tissues from the patients with Alpers’ syndrome (Fig. 2, Supplementary Fig. 4). Neighbouring cortical regions demonstrated a high abundance of HLA-DR + microglia, albeit astrocyte densities appeared normal (Supplementary Fig. 4). However, interestingly, the focal lesions in tissues from adult patients with late-POLG disease and mtDNA disease were not as clearly defined by gliosis (Supplementary Fig. 5). Furthermore, the focal lesions in tissues from patients with Alpers’ syndrome were characterised by a total loss of Purkinje cells. However, often some, albeit few, Purkinje cells were observed in focal lesioned cortex from adult patients with late-POLG disease and mtDNA disease, suggesting less severe degeneration and inflammatory pathology within these late-onset lesions.

Increased c-Fos immunoreactivity

In response to a severe degeneration of inhibitory Purkinje cells and focal necrotic lesions in the cerebellar cortex of patients with PMD, we next investigated whether c-Fos protein expression, a marker of neuronal hyperactivity [9], was increased. In cerebellar tissues from control cases, minimal c-Fos immunoreactivity was observed suggesting a low basal expression of c-Fos (Supplementary Fig. 6). However, 7 of 13 patients with Alpers’ syndrome, 3 of 5 patients with late-POLG disease, and 4 of 10 patients with mtDNA disease showed detectable levels of c-Fos, in total constituting 50% of the patient cohort (Table 3, Supplementary Fig. 6), suggesting recent neuronal hyperactivity. Increased c-Fos protein expression was often localised to Purkinje cells within non-lesioned cortex (Supplementary Fig. 6); however, some patient tissues also demonstrated c-Fos labelling in demarcated regions of the granular cell layer and molecular cell layer (Fig. 2). Interestingly, the patient with the most striking pattern of c-Fos immunoreactivity (Fig. 2), which was observed in focal regions of the superior and inferior cerebellar lobes, was Patient 13, a young adult with early onset POLG disease who died during status epilepticus. These findings suggest hyperexcitability of the cerebellar cortex in patients with PMD.

Mitochondrial OXPHOS protein deficiencies in Purkinje neurons

Since we quantified a severely decreased density of Purkinje cells in PMD patient tissues, we sought to investigate the vulnerability of this cell type to mitochondrial OXPHOS protein deficiencies to infer changes to OXPHOS function. Previously, OXPHOS protein defects have been reported in Alpers’ syndrome and late-onset PMD [4, 8]; however, comparative analyses have not been performed. Therefore, a quadruple immunofluorescence assay was used to quantitatively assess and compare the level of complex I (NDUFB8) and complex IV (COXI) subunit protein expression within Purkinje cells of PMD and control tissues (Fig. 3A).

Fig. 3figure 3

Mitochondrial oxidative phosphorylation protein deficiencies in Purkinje cells in mitochondrial disease. a Representative confocal images demonstrating quadruple immunofluorescence for parvalbumin (blue; Purkinje cell marker), NDUFB8 (red; complex I subunit), COXI (green; complex IV subunit), and porin (purple; mitochondrial mass marker) within the cerebellar cortex. Purkinje cell from Patient 11 (white arrow head) demonstrates decreased NDUFB8 and COXI immunoreactivity in comparison to adjacent Purkinje cell. b NDUFB8/Porin z-scores for individual Purkinje cells are presented for patients with Alpers’ syndrome (blue), adult-onset POLG (pink), and mtDNA disease (purple). Complex I deficiencies were significant in Alpers’ syndrome and adult-onset mitochondrial disease compared to age-matched control groups (P = 0.0179 and 0.0281, respectively, linear mixed-effects model). c COXI/Porin z-scores for Purkinje cells were significantly decreased in the Alpers’ syndrome group and did not reach significance in the adult mitochondrial disease group (P = 0.023 and 0.0918, respectively, linear mixed-effects model). (d) Porin z-scores were significantly increased in the Alpers’ syndrome group and remained unaltered in the adult mitochondrial disease group (P = 0.0122 and 0.4736, respectively, linear mixed-effects model). *P < 0.05

Quantification of the intensity of NDUFB8 and COXI proteins, normalised to porin, within Purkinje neurons revealed OXPHOS protein deficiencies involving both complexes in the Alpers’ syndrome patient group relative to matched controls (P < 0.05) (Fig. 3B and C). However, in the late-POLG and mtDNA PMD cohort, only NDUFB8 protein expression was significantly decreased (P = 0.0281). Furthermore, comparison of the three PMD groups revealed that the Alpers’ syndrome patients demonstrated significantly lower expression of COXI in comparison to the mtDNA disease group (P = 0.0406). A summary of the percentage of neurons deficient for complex I and IV and Purkinje cell loss is provided in Table 3, which highlights the increased severity of changes to Purkinje cells in Alpers’ syndrome in comparison to late-POLG and mtDNA PMD groups (Supplementary Table 7).

To assess changes to mitochondrial mass within Purkinje neurons, the intensity of porin was quantified. Analyses revealed that porin was significantly increased in Alpers’ syndrome patient Purkinje cells relative to the control, late-POLG and mtDNA groups (P < 0.05), suggestive of increased mitochondrial mass. In contrast, porin remained unaltered in the late-POLG and mtDNA group relative to controls (Fig. 3D).

Finally, we quantified the expression of the activity-dependent calcium-binding protein parvalbumin (Supplementary Fig. 7) [34]. A lower expression of parvalbumin was detected within Purkinje neurons in the late-POLG and mtDNA disease groups (P = 0.0139); however, parvalbumin expression was unaltered at the Alpers’ syndrome patient group level (P = 0.0564).

OXPHOS protein expression in the granule cell layer and dentate nucleus neurons

Since the expression of mitochondrial OXPHOS proteins has not previously been interrogated in neurons of the granule cell layer of patients with PMD, we sought to quantify OXPHOS protein expression within this predominantly excitatory cortical layer, as a comparison to inhibitory Purkinje cells (Supplementary Fig. 8A).

In Alpers’ syndrome, both NDUFB8 and COXI were significantly decreased in the granule cell layer relative to controls (P < 0.05), whereas in the late-POLG and mtDNA group, only COXI protein expression was significantly decreased (Supplementary Fig. 8B and C). Analyses between the three disease groups revealed a significantly increased NDUFB8 and COXI deficiencies in Alpers’ syndrome in comparison to the mtDNA and late-POLG disease groups (P < 0.01, Supplementary Table 7). Mitochondrial mass within the granular cell layer was significantly higher in patients with Alpers’ syndrome compared to controls (P = 0.0344), whereas mitochondrial mass remained unaltered in the late-POLG and mtDNA disease patient groups (Supplementary Fig. 8D). Interestingly, the level of OXPHOS protein deficiencies was similar between Purkinje neurons and the granule cell layer in patients with Alpers’ syndrome. However, the late-POLG and mtDNA patient groups demonstrated a more pronounced decrease in NDUFB8 and COXI expression in Purkinje neurons in comparison to the granule cell layer.

Additionally, we probed OXPHOS subunit expression in neurons of the dentate nucleus, as they are innervated by Purkinje cells and are primarily composed of excitatory neuronal populations. Intriguingly, dentate nucleus neurons did not demonstrate significantly altered NDUFB8, COXI, or porin protein levels in the PMD patient tissues relative to control tissues (P > 0.05, Supplementary Fig. 9). However, there was a significant decrease in the surface area of the dentate nucleus neurons, indicative of atrophy, in the Alpers’ syndrome and mtDNA groups (P < 0.05, Supplementary Table 7).

Altered expression of autophagy-related proteins

Since OXPHOS-deficient Purkinje cells from patients with Alpers’ syndrome demonstrated increased porin abundance, indicative of increased mitochondrial mass, we sought to investigate whether this might be associated with impairments to mitophagy, the selective autophagic process whereby dysfunctional mitochondria are degraded [33]. We therefore probed the expression of mitophagy- and autophagy-related proteins, which are involved in the formation of mitophagosomes, semi-quantitatively in the PMD and control cerebellar tissues (Table 3). The markers employed included BNIP3 (mitophagy receptor), LC3B (marker of autophagosome formation), p62 (cargo autophagy receptor), and LAMP2 (lysosomal membrane protein).

In cerebellar tissues from patients with Alpers’ syndrome, remaining Purkinje cells demonstrated a marked increased expression of both BNIP3 and LC3B proteins, compared to controls and patients with late-POLG and mtDNA disease (Fig. 4), mirroring the more extensive and severe OXPHOS deficiencies also observed in this patient group. A similar, albeit less pronounced, visibly increased expression of LC3B was observed in Purkinje cells from adult patients with late-POLG and mtDNA disease, whilst BNIP3 expression was only mildly increased in some of these cases (Table 3). However, LAMP2 protein expression appeared to be similar across all control and PMD cases (Supplementary Fig. 10).

Fig. 4figure 4

Increased expression of autophagy and mitophagy-related proteins in the cerebellar cortex in mitochondrial disease. Representative images demonstrate immunoreactivity of the mitophagy receptor BNIP3, autophagosome membrane marker LC3B, and autophagy receptor p62 in control tissue, as well as in patients with Alpers’ syndrome (Patient 9), late-POLG (Patient 17), and m.3243A > G MELAS syndrome (Patient 21). BNIP3 expression was increased in patients with Alpers’ syndrome, relative to controls, late-POLG, and mtDNA cases. LC3B protein expression was detected in a subset of Purkinje cells in control tissues (black arrowhead), in comparison to cells with undetectable levels of LC3B (green arrowhead). Most PMD patients demonstrated increased expression of LC3B relative to control cases. P62 immunoreactivity was observed primarily in corpora amylacea, some molecular layer inhibitory interneurons and vessels in adult controls and patients with PMD. However patient tissues demonstrated some p62-expressing glial cells, which appear to morphologically resemble microglia. All sections were counterstained with Haematoxylin. Scale bars = 100 μm

P62-immunoreactive corpora amylacea and inhibitory interneurons within the molecular layer of the cerebellum were detected in most adult PMD patient tissues, and to a lesser extent in adult controls (Fig. 4). Infant controls unsurprisingly did not demonstrate any p62 corpora amylacea, since these structures are typically associated with degenerative and ageing processes, albeit p62 immunoreactivity was evident in multiple young patients with Alpers’ syndrome (Table 3). In some PMD patient tissues, p62-expressing glial cells were observed in close proximity to Purkinje neurons; however, the majority of patient tissues did not demonstrate any p62 immunoreactivity within the remaining Purkinje cells. An exception to this was Patient 18, in whose tissue intranuclear p62-immunoreactive puncta were present in virtually all Purkinje neurons.

Overall, we found an increased expression of autophagy-related proteins in PMD patient cerebellar tissues. However, the most prominent changes were detected in the Alpers’ syndrome patient tissues, which potentially reflects greater OXPHOS defects and increased mitochondrial mass in this disease group.

Comments (0)

No login
gif