The cerebellum plays an important role in sensorimotor integration, cognitive tasks, and emotional processing (Stoodley and Schmahmann, 2009), rendering it an interesting target region for non-invasive brain stimulation (NIBS) to alter its excitability and connectivity. Beside repetitive transcranial magnetic stimulation, transcranial current stimulation (tCS) paradigms are NIBS techniques, which are based on the application of a low-intensity current flow between two electrodes. The tCS technique currently most commonly used is transcranial direct current stimulation (tDCS). However, the effects of tDCS show large inter- and intra-individual variability (Batsikadze et al., 2019). Therefore, other methods including transcranial alternating current stimulation (tACS) and transcranial random noise stimulation (tRNS) might be interesting alternatives (Naro et al., 2017; Miyaguchi et al., 2020).

TACS has already been used for cerebellar stimulation with frequencies ranging from 5 Hz to 300 Hz with different results (Naro et al., 2016, 2017; Miyaguchi et al., 2019, 2020; Wessel et al., 2020; Schubert et al., 2021; Spampinato et al., 2021). One reason for using different frequencies is related to the fact that different target cell populations have different oscillatory properties. It has been shown that when using low intensities, stimulation frequencies close to the endogenous oscillation frequency induce the strongest spiking resonance at a cellular level (Reato et al., 2010). When using higher intensities, effects can also be induced at subharmonics, and the power of the endogenous frequency may be modulated (Reato et al., 2010, 2013). As Purkinje cells have an endogenous firing rate of about 50 Hz (Raman and Bean, 1999) and play an important role in coordination, 50 Hz seems to be a promising frequency to influence movements and coordination, especially when using low intensities. However, concerning 50 Hz, previous studies revealed different results. Whereas motor performance of the upper limb was reported to be improved after offline tACS (Naro et al., 2017), online tACS it did not improve or even impaired the acquisition of a motor skill when used online (Wessel et al., 2020; Giustiniani et al., 2021). Also, neurophysiological results differed between studies: While Naro et al. (2016) found a reduction of cerebellar brain inhibition (CBI) and an increase of motor evoked potential (MEP) amplitudes, other studies showed no changes in cortical excitability (Spampinato et al., 2021). Reasons for these differences are probably related to the use of different protocols including on-/offline stimulation, which is very relevant because of the known state-dependency of tACS effects (Feurra et al., 2013), stimulation duration, and different methods to investigate CBI.

In contrast to tACS, during tRNS the current alternates with randomized frequencies ranging from 0.1 to 640 Hz. In a recent study, 20 min of left primary motor cortex (M1) tRNS stimulation was more efficient in increasing motor cortex excitability compared to tDCS and 140 Hz tACS (Inukai et al., 2016), which was probably mediated by the high-frequency spectrum (> 100 Hz) (Terney et al., 2008). Due to its high efficacy when applied to M1, positive findings following the application to other brain regions including the dorsolateral prefrontal cortex (Pena et al., 2019) and its possible differential effect regarding location and frequency of the stimulation (Campana et al., 2016), tRNS might be a promising tool for cerebellar stimulation. Given the limited experience with tRNS and its mode of action, the expected effects are currently largely unclear, so its use has to be considered exploratory.

To our knowledge, due to the high variability of study protocols and modes of application used in different studies no standardized comparison has been carried out between tDCS, tACS, and tRNS in the same study population (Kumari et al., 2019; Antal et al., 2022). To fill this gap, we directly compared anodal tDCS, 50 Hz tACS, and high-frequency tRNS (hf-tRNS) with sham stimulation as to their efficacy to induce plasticity in the cerebellum.

We examined the size of MEPs because MEPs reflect corticospinal excitability, which has been shown to be increased following tACS (Naro et al., 2017). To investigate the excitability of the cerebello-thalamo-cortical pathway, we included CBI as an additional readout. To test CBI, a conditioning stimulus is applied to the cerebellum before stimulating the contralateral M1, which decreases cortical excitability reflected in reduced MEP amplitudes (Ugawa et al., 1995; Werhahn et al., 1996). This effect is thought to be mediated by the activation of Purkinje cells reducing the excitatory drive of dentato-thalamo-cortical pathways (Celnik, 2015). CBI can thus be taken as a marker of Purkinje cell excitability, which is likely modulated by tCS, e.g., following 50 Hz tACS (Naro et al., 2017). We also tested short-interval intracortical inhibition (SICI), where a subthreshold conditioning stimulus over M1 precedes a suprathreshold pulse given to M1 by 1 to 6 ms (Berardelli et al., 2008). It is a measure of GABA A-ergic mediated M1 inhibition and has been documented to be modifiable by cerebellar tDCS and a conditioning cerebellar TMS pulse, but not by 50 Hz tACS (Daskalakis et al., 2004; Ates et al., 2018; Spampinato et al., 2021). The effects of tRNS on SICI are currently unclear. Because of the important role of the cerebellum for sensorimotor processing (Dubbioso et al., 2015; Yildiz et al., 2018) we also included short-latency afferent inhibition (SAI) where a peripheral nerve is stimulated before stimulation of the contralateral M1 (Chen et al., 1999) as a readout. Finally, to capture tCS effects on motor performance, we measured arm movements using inertial measurement units (IMUs). In previous studies, kinematic measurements including IMUs or optoelectronic devices have successfully been used to investigate patients with cerebellar dysfunction including those with ataxia and dystonia (Bologna et al., 2016; Krishna et al., 2019). Thus, to capture the effects of the different tCS techniques on the excitability/activity of the cerebellum and interconnected pathways we used a comprehensive battery of behavioral and neurophysiological measures.

To the best of our knowledge, this is the first sham-controlled study investigating different cerebellar tCS techniques intra-individually. The main finding of our study is that corticospinal excitability, reflected by the MEP amplitudes, is enhanced by tACS and tRNS. This effect lasted longer after cerebellar tACS stimulation i.e., for at least an hour. We found no effect of tDCS nor effects of any intervention on CBI, SICI, SAI, or the behavioral parameters.

Previous data on cerebellar tCS are heterogeneous. In line with our findings, an increase of corticospinal excitability after cerebellar tACS (Naro et al., 2016, 2017) but no change of CBI, SAI, or SICI was described previously (Doeltgen et al., 2016; Spampinato et al., 2021). Also, in keeping with our results, no effect of isolated anodal tDCS on motor cortex excitability has been found before (Behrangrad et al., 2019). However, some studies reported effects of tACS or tDCS on CBI, SICI, and behavioral parameters but used a study design that differed from ours regarding coil type (double cone coil), electrode size (larger), and stimulation parameters (Galea et al., 2009; Naro et al., 2017; Ates et al., 2018). For tRNS, no previous data on cerebellar application exists.

The mechanisms of how tACS and tRNS influence neural circuits are not fully understood. However, it has been hypothesized that tACS modulates neuronal firing rates through locking to their frequency and phase (Battleday et al., 2014). In a recent animal study, it has been shown that Purkinje cells could be entrained to external stimulation at certain frequencies (Asan et al., 2020). Probably, this represents a resonance effect as cells were stimulated with their natural/baseline frequency, e.g., 50 Hz. In humans, it has been shown that the stimulation effect of tACS is state-dependent, i.e., that changes of oscillatory activity were present when participants performed a task (Alagapan et al., 2016). Besides, effects of cerebellar 10 Hz tACS on the learning-related alpha power and premotor-cerebellar connectivity could be shown recently (Schubert et al., 2021). This underlines the relevance of network activity regarding oscillatory brain activity. However, network responses after tACS are variable and until now difficult to predict (Reato et al., 2010).

Two different hypotheses are postulated for tRNS effects with neuronal oscillations playing an important role. The first is based on the influences of natural oscillations by a phenomenon called “stochastic resonance” (van der Groen et al., 2019), i.e., that a non-linear system can be amplified by means of noise, which leads to periodic exceeding of the threshold (Gammaitoni et al., 1998). This in turn possibly leads to a higher sensitivity to weak external input (Moss et al., 2004). Alternatively, it was hypothesized that neurons with a sufficiently long time constant could be influenced by the summation of the stimulation effect of two or more following electric stimuli. This is particularly relevant when using hf-tRNS, which avoids the development of biochemical homeostasis during the stimulation due to an inconstant influence on the ion channels and consequently membrane potentials (Fertonani et al., 2011). Regarding neurochemical mechanisms, it has been shown that the effect of tRNS can be suppressed by benzodiazepines and depends on sodium channels, but not on NMDA receptors (Chaieb et al., 2015).

Probably, cerebellar tACS and tRNS reduce the inhibitory output of the cerebellum, which leads to an increase of excitability in cerebello-thalamo-cortical pathways. This was reflected by an increase of unconditioned MEP amplitudes in the present study. In animal studies, it has been shown that the cerebello-dentato-cortical pathway can have both inhibitory and excitatory effects on the motor cortex, mediated by different interneurons (Na et al., 1997; Holdefer et al., 2000). In addition to classical CBI that is thought to be caused by activation of Purkinje cells, which in turn inhibit the excitatory effect of cerebello-thalamo-cortical pathways and therefore lead to a decrease of corticospinal excitability, it has recently been hypothesized that a stronger activating effect of parallel fibers on Purkinje cells could also lead to an increase of CBI (Celnik, 2015; Spampinato et al., 2021). Taking this into account, there are different explanations for the excitability changes in our study. The stimulation could directly interact with the firing rate of the Purkinje cells. Even if one predicts a short-term increase in their activity due to a resonance effect after 50 Hz tACS, the long-lasting after effect could be a result of a trans-synaptic long-term depression-like effect, resulting from the modulation in network activity over 20 min and subsequent biochemical changes (D’Angelo, 2011). Alternatively, 50 Hz oscillations could interfere with the firing rates of granule cells. Granule cells are the only neurons with an excitatory output to the Purkinje cells and the molecular layer interneurons (D’Angelo, 2011). The stimulation could negatively influence their activating output over parallel fibers on the Purkinje cells, as their natural frequency lies in the theta band (5–7 Hz) (Gandolfi et al., 2013). This could also lead to an increase of corticospinal excitability due to the reduction of inhibitory influence of Purkinje cells on cerebello-thalamo-cortical pathways. The possibility of influencing Purkinje cells and/or inter-connected neuronal circuits non-invasively potentially has high clinical relevance, because especially the pathophysiology of alcohol-responsive movement disorders, e.g., essential tremor and myoclonus dystonia, seems to be related to activity of these cell types (Frucht and Riboldi, 2020; Madelein van der Stouwe et al., 2020). All these models explain an effect of stimulation of the cerebellum via direct or indirect cerebello-neocortical projections. In our study, we found an increase of MEP amplitudes but no effect on intracortical excitability in the motor cortex nor an effect on sensorimotor integration. The mechanism leading to the increase in MEP amplitudes may have occurred at the cortical level or in cerebellar spinal projections. This cannot be answered unequivocally based on the available data. As spinal excitability measures including direct current motor cortex stimulation, H-reflex and F-response were not measured in our study, the site of effective stimulation cannot be determined with certainty.

When interpreting our results, some limitations need to be considered. Given the lack of previous studies testing the effects of tRNS on the measures determined in the present study, the tRNS effects we found should be considered preliminary and have to be replicated in future studies. Regarding CBI, we aimed to compare cerebellar interventions in a representative sample. Therefore, it was important not to pre-select our participants. In a pilot study in our laboratory on CBI with a double-cone coil used for cerebellar conditioning and also in other published studies using this coil (Kassavetis et al., 2011), subjects did not tolerate the procedure due to pain and discomfort. Therefore, we used another published CBI protocol using a figure-of-eight coil, and no participant had to be excluded in our study. Although other studies (Carrillo et al., 2013; Brusa et al., 2014; Koch et al., 2014; Benussi et al., 2018, 2019) with similar experimental conditions found CBI, this was not the case in the present study. In general, it is still unclear if figure-of-eight coils are capable of activating Purkinje cells in the cerebellar cortex (Hardwick et al., 2014; Fernandez et al., 2018). Therefore, we cannot make any conclusions on the effect of cerebellar interventions on CBI.

Regarding SICI, in view of time constraints only one conditioning intensity was tested, so that recruitment curves could not be determined. Using lower intensities or recruitment curves might have revealed different effects.

We could show that the sensors detected variability changes of hand movements concerning speed, direction, handedness, and time course. However, there were no intervention-specific changes of behavioral performance. This is probably explained by the fact that the task we used was too simple and not specific enough to target cerebellar function. A sequence learning task might have been more sensitive. Including such a task, however, might have caused interference effects of (cerebellar) learning and plasticity induced by tCS. Even though we tried to avoid such effects, a possible influence of the motor task we used on our readouts cannot be excluded.

Taken together, corticospinal excitability increased after cerebellar tRNS and tACS, but not following cerebellar tDCS. The effects of tACS lasted up to 1 h, i.e., longer than those of cerebellar tRNS. These findings are relevant for clinical applications of tCS.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethikkommission der Universität zu Lübeck. The patients/participants provided their written informed consent to participate in this study.

RH, AM, TB, and AW conceived the study design. TMB and RH acquired the data. TMB and HX preprocessed the data. RH and HX analyzed the data. RH, TB, MP, HX, ER, AM, TMB, and AW interpreted the data and revised the manuscript. RH drafted the manuscript. All authors approved the final version of the manuscript.