Neurochemical Differences in Spinocerebellar Ataxia Type 14 and 1

Spinocerebellar ataxias (SCA) are a group of autosomal-dominant disorders that present as progressive cerebellar ataxia. They have a prevalence of up to 5.6 patients out of 100.000 individuals [1] however with a suspected large number of genetically unconfirmed cases. Depending on the genotype, these disorders lead to atrophy patterns that include the cerebellum and variable regions in the brainstem, basal ganglia, and cerebral cortex. Hence, patients share a cerebellar syndrome but show different concomitant symptoms. The cerebellar involvement and especially the extra-cerebellar degeneration display SCA subtype-specific patterns [2, 3]. The most prevalent SCA subtypes, among them SCA1, are caused by trinucleotide repeat expansion mutations, and the majority presents widespread central nervous system (CNS) involvement [4]. Many among the more recently identified SCA subtypes caused by missense and deletion mutations, such as SCA14, display a cerebellar syndrome with scarce extra-cerebellar manifestations. Neuroimaging studies in SCA14 are confined to case series, and specific atrophy patterns are not well described. SCA14 is caused by missense or deletion mutations in the protein kinase C gamma gene (PRKCG), which is mainly attributed to signal transduction, cell proliferation, and synaptic transmission in Purkinje cells [4]. SCA14 patients suffer from a slowly progressive ataxia. Other symptoms like cognitive impairment, mild sensory impairment, tremor, or axial myoclonus have been reported [5]. SCA1 is caused by a CAG-repeat expansion in the gene ataxin-1 (ATXN1) on chromosome 6p23 [6]. This gene encodes for the protein ataxin-1, which is suggested to regulate gene expression [7]. Besides progressive ataxia, the phenotype includes supranuclear ophthalmoplegia, pyramidal signs, peripheral neuropathy [5], or retinal nerve degeneration [8].

This study was designed based on our previous studies in SCA14 [9] and SCA1 [10]. Compared with these previous total n-acetylaspartate (tNAA) focused studies, we optimized our spectroscopic methods, used a shorter echo time (TE), included more metabolites into the analysis, examined extended patient cohorts of both SCA1 and SCA14, and introduced two cerebellar volumes of interest (VOI) and an extra-cerebellar VOI in the brainstem. In the present study, we aimed to assess metabolic changes in SCA14 and SCA1 patients in both cerebellar and extra-cerebellar regions in order to differentiate them from a control group as well as to identify differences and similarities of these subtypes based on their metabolic profile. An extensive neurochemical profiling might be able to reveal subtype-specific and/or cross-subgroup biomarkers that could be used for facilitation of diagnostic procedures and demonstration of starting points for future therapeutics. Since SCA1 is characterized by widespread neuronal loss and brainstem atrophy [4], we hypothesized that metabolic alterations will be present in the pons. In contrast, we hypothesized that metabolic alterations in SCA14 will be limited to the vermis and cerebellar hemisphere due to the localization of Purkinje cells and expression of protein kinase C gamma. Moreover, we correlated clinical examination scores and characteristics with in vivo levels of each metabolite to check for associations with prognostic significance on the clinical course. For the assessment of brain metabolites, we considered ¹H magnetic resonance spectroscopy as a powerful tool to analyze the human brain metabolism in vivo. This technique allows for non-invasive detection of biochemical alterations and quantification of metabolites in symptomatic [11] and presymptomatic SCA patients [12]. In addition, ¹H magnetic resonance spectroscopy is able to distinguish between SCA subtypes [11], and the results of different medical centers are reproducible if identical spectroscopy settings are applied [13].

This study assessed metabolic changes in cerebellar and extra-cerebellar brain regions in SCA1 4 and SCA1 patients using 3-T single-voxel ¹H magnetic resonance spectroscopy. In-vivo metabolites were compared between the three groups of SCA14, SCA1, and HC as well as linear relationships between metabolite levels, and clinical features were evaluated for merged patient groups with adjustment for age.

In SCA14, metabolic changes were strongly restricted to the vermis, while extra-cerebellar brain regions and particularly the cerebellar hemisphere were not involved in neurochemical alterations. In the vermis of SCA14 patients, tNAA, tCho, tCr, GABA, and Glu were mainly decreased as well as Glc was increased. The tCho reduction and Glc elevation have not been described for SCA14 yet, while the reductions in tNAA, tCr, GABA, and Glu confirm previous research [9]. The tNAA reduction is characteristically for neurodegenerative diseases and corresponds to neuronal cell loss or neuronal dysfunction [18]. Low tCr may indicate suppressed energy metabolism with a consecutive accumulation of Glc and low membrane turnover as illustrated by tCho reduction. The decreased neurotransmitters GABA and Glu may reflect loss of GABAergic Purkinje cells and loss of climbing fibers, whose efferences maintain glutamatergic synapses with Purkinje cells.

SCA1 patients demonstrated similar results in vermis: There was a significant reduction in tNAA and tCho but less pronounced than in SCA14. Although results on tCho levels in SCA1 are heterogeneous, the observed tCho reduction of the present study supports previous indications for low tCho in the cerebellum [19, 20]. Alike SCA14, Glu was reduced in the vermis, but did not reach statistical significance in SCA1 likely due to significant longer disease duration of the SCA14 cohort. Furthermore, both subtypes had a Glc elevation in common, which was more pronounced in the SCA1 than in the SCA14 cohort (247% vs. 155%). Supporting these results, Glc levels were elevated in the cerebellum of mice overexpressing mutant human ataxin-1 [21]. Also, SCA1 patients revealed an increase of Glc+tau amounts in the vermis [12] and cerebellar hemisphere [12, 22]. In positron emission tomography, a reduced Glc metabolism was seen in the cerebellum and brainstem of SCA1 patients [23]. In fact, a Glc hypometabolism has also been reported for other polyglutamine diseases such as SCA2, SCA3, SCA6 [24], and Huntington’s disease [25] as well as for neurodegenerative diseases without trinucleotide repeats such as Alzheimer’s disease [26], amyotrophic lateral sclerosis [27], and Parkinson’s disease [28]. Accordingly, the Glc elevation in the vermis could represent an accumulation due to a reduced Glc metabolism. In distinction from SCA14, Asp was reduced, and tCr was not changed in the vermis of SCA1 patients. Interestingly, aspartate-glutamate homeostasis in neuronal cells is affected by the availability of glucose and ketone bodies. In the presence of ketone bodies only, neurons showed higher aspartate and lower glutamate levels compared with presence of glucose alone [29]. Thus, high levels of Glc may affect aspartate levels in the vermis. However, Asp also serves as substrate for NAA synthesis, and low Asp could simply be linked to low NAA levels [30]. Since the SCA1 cohort was characterized by distinct changes in Glc and Asp as opposed to SCA14, despite a significant shorter disease duration, these metabolites might point to metabolic pathways differing between SCA14 and SCA1.

In SCA1 patients, more widespread metabolic changes were present. The tNAA reduction was not only found in vermis of SCA1 patients but also in the cerebellar hemisphere, pons, and as a trend in the motor cortex. Furthermore, the SCA1 cohort featured an Ins elevation in the cerebellar hemisphere and pons, whereas the vermis was excluded from an increase. This result confirms previous studies [12, 22]. Ins is considered a marker for glial proliferation or hyperplasia in the sense of gliosis [18]. High Ins levels could indicate an involvement of glial cells in the pathomechanism of SCA1. In fact, the Bergmann glia of a SCA1 mouse model revealed a reduced expression of the Glu-Asp transporter (GLAST) that correlated with the loss of Purkinje neurons [31]. Moreover, Asp showed a reduction not only in the vermis of SCA1 patients but also in the pons, which did not reach statistical significance probably due to the small sample size in pons (n = 9). The pons is anatomically located in an area, where shimming in the majority of cases does not result in sufficient B₀ field homogeneity and subsequently only a small number of data sets met sufficient spectral quality. Interestingly, each metabolite of tNAA, Ins, and Asp was altered to the same extent across the affected brain areas in the metencephalon (tNAA in the vermis − 24%, cerebellar hemisphere − 26%, and pons − 25%; Ins in the cerebellar hemisphere + 51% and pons + 46%; Asp in the vermis − 62% and pons − 68%). However, no changes were identified for Glu, Glc, and tCho in SCA1 outside the vermis. In contrast to the reduction of tCr in the vermis of SCA14 patients, but in line with previous research [22], we found a tCr elevation selectively in the cerebellar hemisphere of SCA1 patients. Adanyeguh and colleagues reported about tCr increases in the vermis and pons of SCA1 without investigating the cerebellar hemisphere [32]. Since tCr levels are related to mental activity [33], the discrepancy in the distribution of tCr elevations might be attributed to recently varying activity states. TCr is considered a marker for energy metabolism and serves as buffer for adenosine triphosphate (ATP) [30]. Besides a Glc hypometabolism in the cerebellum of SCA1 patients, downregulation of certain enzymes involved in glycolysis and ATP synthesis, lower ATP levels [34], and reduced activities of the respiratory chain complexes [35] were identified in the cerebellum of SCA1 mouse models. Therefore, high tCr could be a reaction on suppressed energy metabolism in order to maintain fast energy supply in case of high energy demand. Moreover, Scyllo was reduced in the motor cortex of SCA1 patients, which should be reevaluated in future studies.

Neurochemical changes were observed in SCA14 selectively in the vermis and in SCA1 throughout the cerebellum and pons. Besides changes related to the different atrophy patterns, both subtypes differed by direct comparison in the metabolites tCr, Ins, and GSH in the cerebellum as well as Scyllo and Tau in the motor cortex. Since GSH and Tau were unchanged in both patient groups compared with the control group, these results seem questionable.

Regression analysis did not reveal any significant associations between clinical data and metabolite levels in the vermis. This could be explained by focal atrophy with distinct loss of cerebellar tissue and consequently very low metabolite levels. Of note, the provided analysis tested only for linear relationships. In the cerebellar hemisphere as well as in the pons, low tNAA levels were associated with worse cognitive power (lower DemTect scores) and, in accordance with a previous study [22], worse motor performances (higher SARA scores). Although no alterations in tCho were found outside the vermis, low tCho in the cerebellar hemisphere and/or pons were indicative for poorer clinical presentation (worse SARA, worse DemTect, lower age of onset, longer disease duration). Moreover, weak associations between high Ins level in the cerebellar hemisphere and higher age of onset, respectively, shorter disease duration were identified. Since SCA14 did not reveal alterations in Ins, these associations seem to be an effect of the SCA1 cohort. Glc showed only isolated associations with clinical parameters, which were merely present in brain areas without altered Glc levels. Further investigations on Ins, tCho, and Glc levels in SCA1 and SCA14 patients are needed to confirm these results. With regard to Deelchand and colleagues [13], a multi-center study could overcome the issue of underpowered patient cohorts and clearly enhance the quality of analysis. Future studies using ¹H magnetic resonance spectroscopy would also benefit from a two-staged data collection for evaluation of metabolite fluctuations/courses over time, a combination of spectroscopic with structural MRI measurements, and an inclusion of presymptomatic mutation carriers in addition to symptomatic patients.

The various SCA genotypes share cerebellar ataxia as a joint clinical characteristic and cerebellar atrophy of yet different degree and pattern. Beyond, SCA patients may share disease features common with other neurodegenerative diseases, especially mitochondrial disorders. Against this background, it seems possible that different SCA genotypes share pathological mechanisms and pathways. It remains to be shown, if SCAs with similar atrophy patterns share more pathophysiological pathways than SCAs with different patterns. Our study examined two SCA subtypes with different atrophy patterns. Still, we found common metabolic changes that are in line with previous results and might be attributed to dysfunctional energy metabolism in both SCA14 and SCA1. Since mitochondrial and glycolytic pathways have great impact on neuronal energy metabolism, these findings offer an eligible approach for further investigations in SCA14 and SCA1.