Subcortical alterations were also found with structural and functional imaging techniques. Voxel based morphometry (VBM) studies showed no significant differences in white or gray matter of the basal ganglia in M-D patients, but increased severities of dystonia [42] and myoclonus [59] were associated with larger putaminal volumes. Functional imaging with [¹⁸F]-fluorodeoxyglucose binding revealed genotype related alterations of subcortical metabolic activity in M-D patients and asymptomatic SGCE-positive mutation carriers [43]. Metabolic increases in pontine and thalamic brain areas were present in all SGCE mutation carriers compared to healthy controls [43].

The hypothesis of a subcortical deficit in M-D is also supported by treatment effects in M-D, particularly DBS of the GPi or the VIM [38], and also by altered activity of GPi neurons [47, 54].

In patients with dystonia, it is thought that the direct motor pathway via striatal D1 receptors is hyperactive, which might result in a reduced GPi activity, and therefore, a disinhibition of the thalamus, and an increased thalamocortical output [63]. On the other hand, it is suggested that the activity of the indirect motor pathway via striatal D2 receptors is reduced in patients with dystonia [63]. This explanatory framework might also fit for the hyperkinetic symptoms, e.g., dystonia and myoclonus, present in M-D.

In this regard, an interesting strategy to investigate the hubs of the direct and indirect motor pathways is the use of intracranial DBS electrodes, either intraoperative during DBS implantation surgery or shortly after the operation when the impulse generator is not connected yet and electrodes are externalized. A coherence analysis investigating the synchrony (i.e., correlation) between muscle discharges, recorded with electromyography (EMG), and (motor associated) neural activity, recorded by local field potentials (LFP) of the basal ganglia via DBS electrodes, or cortical activity recorded by electroencephalography (EEG) [64], is very helpful to draw further conclusions how movements are controlled or influenced by (sub-)cortical activity [65]. Increased cortico-muscular coherence (3–15 Hz) between several muscles and the GPi-LFP during rest and voluntary muscle activation was identified in M-D patients after GPi-DBS surgery and could reflect abnormal, e.g., increased GPi activity [47]. In another study, GPi recordings revealed a higher burst activity, which correlated with a higher preoperative severity of myoclonus in M-D patients in comparison to generalized dystonia patients, and thus seems to be specific for the myoclonus phenotype [54].

The hypothesis, that hypoactivation of the indirect pathway might contribute to the pathogenesis of M-D, is supported by findings of reduced striatal [¹²³I]-Iodobenzamide binding, reflecting lower dopamine D2 receptor binding [39]. After treatment with GPi-DBS for 2 years, striatal dopamine D2 receptor binding did not decrease further, as it was the case in M-D patients without GPi-DBS, suggesting a stabilization on dopamine pathways due to GPi-DBS [30].

Defective basal ganglia mediated motor inhibition, investigated with a “Go/NoGo” task, where it is required to react to a “Go” cue and to suppress a reaction to a “NoGo” cue, has been found in a group of SGCE-positive and SGCE-negative M-D patients, suggesting that these abnormalities might be related to phenotype rather than genotype [46].

The role of the basal ganglia, especially the GPi, during motor inhibition in M-D was further undermined using a “stop signal” task, which investigates reactive (stop an already started action) and proactive (action inhibition during action preparation) inhibition [55]. M-D patients with GPi-DBS showed impaired reactive inhibition and M-D patients without GPi-DBS exhibited impaired proactive inhibition [55]. The impairments in reactive inhibition correlated with the intrinsic/preoperative severity of myoclonus. This could indicate that GPi-DBS on the one hand improves proactive inhibition, but on the other hand impairs reactive inhibition. Moreover, response inhibition involves the hyper-direct and the direct pathway and potentially also the striato-nigral pathway, which is known to be modulated by GPi-DBS [55].

To sum up, results of neurophysiological, behavioral, and imaging studies, as well as treatment effects of DBS, point to a subcortical deficit in M-D. Especially a dysfunction of the basal ganglia circuits, including the GPi, might contribute to abnormal excitatory output and connectivity with other subcortical and cortical output regions.

Besides the basal ganglia and its connections, there is emerging evidence that the cerebellum might be involved in the pathogenesis of M-D [40, 41–46, 56, 58, 60]. With regards to structural brain imaging, a VBM study revealed a higher gray matter signal in the left motor cerebellum (lobules V and VI) in M-D patients compared to healthy control participants [59]. Furthermore, changes of white matter bilateral in subthalamic areas of the brain stem, connecting the cerebellum with the basal ganglia, were observed via VBM and diffusion tensor imaging (DTI) [45].

Increased metabolic activity in the parasagittal cerebellum has been found in M-D patients, but not in asymptomatic SCGE-positive mutation carriers and healthy controls [43]. Furthermore, these metabolic increases were also found in patients with posthypoxic myoclonus, which was interpreted as phenotype- (i.e., myoclonus) specific cerebellar metabolic abnormality [43].

In functional magnetic resonance imaging (fMRI) studies using different motor tasks, cerebellar regions were hyperactive in M-D patients [40, 44, 46], and asymptomatic SGCE-positive mutation carriers, who did not report symptoms by themselves, but had subtle signs of M-D in a detailed motor examination [41]. These cerebellar hyperactivations might be genotype-specific, as they also allowed separating asymptomatic SGCE mutation carriers from healthy controls [41], and SGCE-positive from SGCE-negative M-D patients [46].

Cerebellar function can also be assessed with the help of behavioral tasks, as the cerebellum contributes to non-declarative forms of learning, e.g., motor learning or classical conditioning [66]. Classical conditioning can be investigated in an experimental setting with an eyeblink conditioning task [67]. Thereby, the connection between an neutral conditioned stimulus (CS), usually a tone, and a response to be conditioned (conditioned response), a blink (e.g., triggered by an air puff or electrical stimulus) is learned and subsequently unlearned. It is thought, that eyeblink conditioning is mediated via brainstem-cerebellar connections, e.g., between pontine nuclei, the inferior olive, and the cerebellum [66]. Studies in patients with cerebellar lesions suggest a strong cerebellar involvement in the acquisition of the conditioned response [66]. Moreover, functional imaging of healthy participants showed activation in the cerebellar lobules VI, Crus I and II, VIIb, VIII, interposed nuclei, and dentate nuclei during acquisition of the conditioned response [68].

M-D patients showed decreased cerebellar motor learning, reflected by reduced acquisition of the conditioned response, and therefore, a poorer performance in the eyeblink conditioning task [60]. After consuming alcohol, myoclonus and the acquisition of the conditioned response improved in M-D patients, but decreased in healthy controls [60]. A proposed mechanism might be, that alcohol probably increases inhibitory GABAergic transmission and improves dysfunctional cerebellar disinhibition in M-D, but disrupts physiological cerebellar activity in the healthy brain [60]. This hypothesis of an “overactive”, disinhibited cerebellum might be supported by the fMRI results of cerebellar hyperactivation during motor tasks as described above [40, 41, 44, 46].

Contrary to these findings, in another study, a smaller cohort of M-D patients showed normal eyeblink conditioning acquisition/learning rates, but lower extinction rates, i.e., difficulties in unlearning the conditioned response [52]. Methodological differences (air vs. electrical stimulation, trial numbers) and differences in sample size and characteristics should be kept in mind, when interpreting and comparing results [52, 60].

Other blink reflex measurements such as the blink reflex recovery cycle, which analyzes brain stem-basal ganglia interactions, showed a greater/faster mean recovery [50]. However, in a larger group of M-D patients, the blink reflex recovery cycle was normal, suggesting normal brain stem-basal ganglia interactions [60], and a more pronounced cerebellar deficit in M-D [60].

Another technique to assess cerebellar motor learning is saccadic adaptation [56]. Adaptation in general requires learning of an artificially induced movement error [69]. In the case of saccadic adaptation, a visual target is moved after movement initiation, so a corrective eye movement, i.e., a saccade, must be learned [69]. In a backward reactive saccade adaptation task, which involves the cerebellar vermis (lobules VI and VII), M-D patients showed slower saccadic adaptation [56]. Interestingly, impairment in a saccadic adaptation task can also be observed in healthy participants following inhibitory repetitive transcranial magnetic stimulation (rTMS) of the posterior vermis [70]. This suggests reduced function of the posterior vermis in M-D patients, which is also consistent with the observed metabolic differences along the parasagittal cerebellum in M-D patients as described above [43].

To test, if potential deficits in cerebellar adaptation function might also affect limb adaptation, visuomotor or forcefield perturbation in symptomatic body parts was examined [57]. In comparison to the deficits in saccadic adaptation [56], no differences were found in other tasks, suggesting that deficits in saccadic adaptation are not easily translatable to other body regions. However, potential deficits in limb adaptation might be more subtle and need to be investigated in larger sample sizes, as the current study only included five patients [57]. Also, methodological differences, and the potential contribution of different brain regions have to be kept in mind [57].

The association between the cerebellum and motor learning deficits in M-D is also supported by results from a paternally-inherited cerebellar Purkinje cell-specific Sgce conditional knockout mouse model [71]. These mice showed motor learning deficits, whereby, paternally-inherited Sgce heterozygous (non-conditional) knockout mice showed additional myoclonus, psychiatric alterations (depression- and anxiety-like behaviors), and motor impairments [72]. This might be an indication, that an impairment of motor learning is mostly influenced by a loss of function of cerebellar SGCE, whereas defective SGCE in other brain regions might contribute to the development of other symptoms of M-D, like myoclonus [71, 72].

In conclusion, although the cerebellum (and its connections to other brain areas) is recognized as a region of particular interest in M-D, its contribution to M-D symptoms remains largely unclear. Imaging data showed that the cerebellum is a promising region to discriminate between SGCE-positive mutation carriers and healthy controls, and SGCE-positive and SGCE-negative M-D patients. In addition, behavioral tasks such as eye-blink conditioning and saccadic adaptation are an important research strategy as they are less prone to motion artefacts than imaging data and can be used in patients who are not suitable for MRI.

There is emerging evidence, that dystonias are sensorimotor disorders, as evidences, for instance, by sensory phenomena including symptom improvement by sensory tricks [73], or increased temporal discrimination thresholds, even in unaffected relatives of patients [74]. The cerebellum itself is involved in somatosensory processing, as it receives somatosensory input via the spinal cord, visual and auditory systems, and trigeminal nuclei [75, 76]. It monitors and adjusts executed movements by comparing planned movements (efference copy) and somatosensory feedback [76]. Thus, sensory deficits found in M-D might be influenced by cerebellar and basal ganglia dysfunction in M-D.

In this regard, in addition to motor learning and motor inhibition difficulties, M-D patients have also shown sensory abnormalities in the visual and tactile domain [58, 59].

Visual sensory processing seems to be impaired in M-D patients with GPi-DBS, who had higher visual temporal discrimination thresholds than M-D patients without GPi-DBS and healthy control participants [59]. Sensory accumulation, which is a computational analysis of the response times connected to the gain of visual sensory information, was lower in the whole M-D patient group compared to healthy controls in the visual temporal discrimination task, and also in a movement orientation and a movement speed discrimination task [59]. Patients with more severe myoclonus showed lower sensory accumulation in the visual temporal discrimination task, and had a thicker primary visual cortex. Because the deficits in visual sensory processing were correlated with the thickness of the primary visual cortex, a brain area elementary responsible for visual perception, the authors interpreted these abnormalities as a primary part of M-D, and not as a secondary phenomenon [59]. Moreover, the role of tactile sensory processing in M-D is underscored by another study, which showed increased tactile temporal discrimination thresholds with preserved tactile perception thresholds in these patients [58].

Furthermore, besides cerebellar hyperactivation during motor tasks (as described above), hyperactivation of the somatosensory cortex was found in M-D patients [40, 41], and further separated them from asymptomatic SGCE mutation carriers [41].

Moreover, the cerebellum is also connected with the motor cortex via dentato-thalamo-cortical pathways that are predominantly facilitatory, whereas connections between cerebellar Purkinje cells and the dentate nucleus are inhibitory [77].

Abnormal cerebellar activity could potentially influence the motor cortex, which could result in defective motor cortex and corticospinal excitability.

Transcranial magnetic stimulation (TMS), a non-invasive brain stimulation technique, is suitable to investigate cerebellar-primary motor cortex connections. The majority of studies analyzed motor cortex excitability by measuring resting or active motor thresholds [49–53], or intracortical inhibitory processes [49–53].

Measures at rest were normal in M-D across studies, e.g., resting motor threshold [52] and recruitment curve of motor evoked potentials [53]. Measures with muscular preactivation, e.g., active motor threshold, were normal [52, 53], or, in contrast, increased [51, 52]. Myoclonic symptoms, as a part of M-D, appear more frequently during action compared to rest [38]. The increased active motor thresholds, reflecting a reduced excitability of the axon membranes during muscle activation [52], might show, that the deficit in M-D is action-specific. In the context of the hypothesis of abnormal cerebellar hyperactivation, increased active motor thresholds could reflect increased Purkinje cell inhibition on the deep cerebellar nuclei, and therefore, an enhanced inhibition of the motor cortex [76], reflected by increased active motor thresholds. On the other hand, if human M-D is associated with a dysfunction, e.g., decreased activity, of cerebellar Purkinje cells, it might also reduce the inhibition of the dentate nucleus, and therefore, increase the faciliatory cerebellar-thalamo-cortical loop and its excitatory output on the motor cortex. Although this would fit with the hypothesis of cerebellar hyperactivation and the hyperkinetic symptoms of M-D, i.e., myoclonic jerks and dystonia, it does not explain increased active motor thresholds. However, contradictory findings might also be explained with methodological differences, as, e.g., the increased active motor thresholds have only been found using biphasic TMS pulses in one study [52]. Additionally, motor thresholds might not be sensitive enough to reveal potential (subtle) network deficits [78].

The majority of other TMS protocols investigating short-interval intracortical inhibition [49, 51–53] and (short-interval) intracortical facilitation [49, 52, 53] were normal in different groups of M-D patients, suggesting intact cortical GABA_Aergic inhibitory and glutaminergic excitatory networks in M-D [49, 79, 80]. In addition, GABA_Bergic inhibitory networks during action and rest [81–83], investigated with silent periods [50, 53] and long-interval intracortical inhibition [49, 50], seem to be normal in M-D patients as well.

The above described method of coherence, referring to EMG-LFP investigations [47], can also be non-invasively applied with either the combination of EMG and EEG (to analyze cortico-muscular coherence) or of two EMG channels (to analyze intermuscular coherence), if, e.g., the application of an EEG is not possible due to data contamination because of muscle artifacts [64]. The investigations of EEG-EMG and EMG-EMG coherence can give insights into potentially altered (sub-) cortical neuronal activity, as different oscillations have different neuronal generators, e.g., the olivo-cerebellar system with frequencies between 6–12 Hz, and the primary motor cortex with frequencies between 15–30 Hz and 30–60 Hz [64].

In a group of M-D patients and non-affected SGCE mutation carriers, physiological EEG-EMG coherence in the 15–30 Hz frequency band was absent during muscular contractions [48]. This altered cortical activity may be influenced by subcortical dysfunction [48]. Furthermore, phenotype-specific alterations of intramuscular coherence, i.e., a significantly increased EMG-EMG coherence (3–10 Hz), were present in M-D patients with pronounced dystonia, but not in those with mild dystonia and/or predominating myoclonus [48]. This might indicate altered subcortical activity, as increased EMG-GPi-LFP coherence was found in M-D patients as well (as previously reported), suggesting abnormal GPi activity [47]. With regards to the hypothesis of cerebellar hyperactivation in M-D, the increased coherence might also be influenced by altered olivo-cerebellar oscillations, because these occur at a similar frequency as those associated with increased coherence in M-D [64].

Overall, M-D patients show sensory dysfunction, reflected by altered visual and tactile processing on a behavioral level, and structural and functional imaging abnormalities. TMS results point towards an action-specific network deficit. Cortical GABA_A- and GABA_Bergic inhibitory networks, and excitatory glutaminergic networks seem to be unaffected in M-D. Abnormal cortico- and intermuscular coherence in M-D might be a consequence of altered subcortical activity.