Our study shows that anodal stimulation of the right cerebellar cortex modulates static and, especially, dynamic resting-state FC between the stimulated right lobule VII and ICNs. In particular, we show for the first time that tDCS caused greater temporal variation of BOLD signal between crus 2 and SN/DAN/SMN and between lobule 7b and DAN/SMN whereas tDCS induced preferentially lesser temporal variability between crus 1 and ICNs and between crus 2 and DMN/CEN/VN. Moreover, our study also reveals that tDCS alters differentially the cerebello- cortical time variability with respect to low-to-high β values converted into z-score, whatever the time variability might be increased or decreased compared to sham. More precisely, increased time variability is accompanied by enhanced frequency of windows in high z-scores (60%), and dimished time variability is accompanied by increased frequency of windows in small and moderate z-cores. We also noticed tDCS-induced between-lobuli FC reorganization within the cerebellum.
tDCS stimulation applied over the right neocerebellum modulates all the tested ICNs. The cerebellar region stimulated by the anodal electrode mainly corresponds to the underlying hemisphere of lobule VIIab where the electric field strength is known to reach focally its maximal value [15], even if more discrete and anatomically restricted stimulation of adjoining lobules VI and VIII cannot be ruled out. Moreover, the lobule VII occupies 49, 75% of the whole cerebellar volume [16], and is massively structurally and functionally interconnected with associative cortices. More precisely, functional coherence was found between crus 1–2 and prefrontal/ parietal/ temporal/ posterior cingulate cortices (BA 8/9/10/46), between crus 1 and rostral inferior parietal cortex and ACC, between crus 2-lobule VIIb and posterior parietal cortex (BA 39 including the overlying intraparietal sulcus, precuneus) [17, 18, 19] and between lobule VIIb and DAN [4]). Cerebellum is also functionally linked to the insula [20]. Cerebello-cortical FC was bilateral with a contralateral predominance. At the network level, the lobule VII constitutes a hub being part of the CEN, DMN, SN [6] and VN [21]. It is noteworthy that the CEN, also called frontoparietal network (FPN), proved to be two-fold overrepresented within the cerebellar cortex [22].
We have found, with the static analysis, bilateral dynamic FC connections between lobule VII and DMN, CEN, SN, VN and DAN. Bilateral FC between cerebellum and cerebral cortex have been reported previously [6]. Such bilaterality may rely, despite the massive projection of the dentate nuclei to the contralateral thalamus through the superior cerebellar peduncle, on recrossed thalamic projection to ipsilateral thalamus [23], on ipsilateral collaterals arising from pontine or reticular nuclei, to transcallosal connection, or to another brain relay not accounted by our predefined regions-of-interest used in the connectivity analyses [8]. Furthermore, the parallel fibers may also participate in the changes observed, by linking cerebellar cortical sites. It is noteworthy that Park et al. [24] has demonstrated tDCS-induced enhancement of interhemispheric connectivity during the resting state.
CEN is subdivided into right and left frontoparietal networks likely subserving, besides general executive functions, more specialized functions in visuospatial or linguistic/logical domains, respectively. tDCS-induced temporal variability was increased between crus 2 and left CEN, whereas temporal variability was decreased between crus 2 and right CEN, pointing out a dual influence of crus 2 upon CEN. However, decreased temporal variability was observed between CEN and the other cerebellar sublobuli. On this vein, anodal tDCS stimulation over the right parietal cortex yielded to enhanced static FC between cerebellum (crus 1 and 2) and precuneus (DMN) and contralateral CEN [25].
We have found changes of static and dynamic FC between lobule VII and SMN/VN, although no anatomical nor direct known functional links exist between lobule VII and motor or visual areas. Two explanations can be proposed: either a “hidden” brain node or spread of electrical stimulation to nearby lobules VI and VIII for SMN. There might be a direct or indirect recruitment of the oculomotor vermis of lobules VI caudal and VII, and of lobules IV-V through intracerebellar functional connection [26]. Kelly and Strick [27] also traced in monkey connections between lobule VIIb and M1. In addition, a contribution of brainstem relays cannot be ruled out.
If the main effect of cerebellar tDCS consisted in temporally stabilizing FC between cerebellum and ICNs, we measured a greater temporal variability between crus 2 and the main nodes of SN (anterior insula, ACC and the supramarginal cortex). SN plays a major role in switching activity between DMN involved in task-negative mind wandering and CEN. Resting state high activity of SN can also be associated with greater FC between DMN and CEN [28].
Anodal tDCS can transiently alter alpha, beta and gamma brain oscillations [29, 30, 31, 32]. In rat, crus 1 processed phases and phase differences between prefrontal and hippocampal oscillations [33] suggesting a cerebellar role in timing and temporal interaeal coordination. Therefore, tDCS may modulate within- and cross-network synchronization which may concern predominantly DMN and CEN. During the resting state, using co-activation pattern analysis, Karahanoglu and van De Ville [34] showed that: (1) DMN including crus 1 had the longest dwell time, (2) DMN and SN were anticorrelated, and (3) DMN and CEN activations tended to co-occur with the same or opposite (posterior DMN) sign. Moreover, DMN, especially the posterior parietal cortex, transiently correlated in the beta band with other networks [35]. Therefore, it can be hypothesized that anodal tDCS might modulate brain oscillations involved in dynamic synchronization/desynchronization of DMN, CEN and SN, which contributed to reinforce cerebellum-DMN/CEN FC and to diminish crus 2-SN/SMN/DAN FC. In particular, the more flexible interaction between crus 2 and SN, and conversely, the reverse pattern between crus1 and SN, could reflect a dual and antagonistic cerebellar control of lobule VIIa upon the SN ability to switch between CEN and DMN activity, likely in relation to different prefronto-cerebellar afferents. Of interest, crus 2 showed predominantly increased or dual increased/decreased dynamic connectivity with ICNs (except with DMN). Crus 2 should participate in dynamic switching between ICNs or between specific nodes of ICNs all the more easily that it is functionally connected with prefrontal, parietal and cingulate cortices, and with the DMN/SN/CEN.
Dynamic FC exhibited more widespread tDCS-induced effects of lobule VII onto ICNs than static FC. We have found agreement between static and dynamic results for crus1 and lobule VIIb
which manifested preferentially increased FC. However, after tDCS, crus 2 showed decreased static FC with DMN whereas dynamic FC was weakened. Further studies are required to reconcile these discrepant results. Moreover, dynamic FC between lobule VIIb and ICNs had the highest FDR p- value (p ≤ 0.01). Lobule VIIb is involved, at least, in executive, linguistic and visual working memory [6, 7], and is in functional coherence with CEN [6], SN [6] and DAN [7]. Lobule VIIb thus seemed to constitute an important cerebellar hub controlling ICNs.
tDCS also changed static and dynamic FC between cerebellar lobuli. Such changes might modulate cooperation between ICNs and between homotopic lobuli within the cerebellum.
Our results should be replicated and complemented with a larger population, a longer duration of fMRI recording, a characterization of the thought content during the mind wandering, and neuropsychological tests performed before and after tDCS/MRI sessions to seek for tDCS- induced transient modifications, for example, of executive functions. Although a larger population is required to address generalization of our results to the general population, our data clearly showed that tDCS caused important transient FC reorganization. Finally, the variable z-score distributions between lobule 7 and ICNs observed during tDCS versus sham requires also further studies to grasp its functional and underlying neurophysiological meaning.
In conclusion, anodal tDCS over the right cerebellum causes static and mostly dynamic changes in resting state FC characterized by global reduced cerebello-cortical temporal variability with the notable exception of crus 2 whose FC with SN was enhanced. Crus 2 could be considered as a hub dually and differentially influencing intra-network nodes. The elucidation of these effects are particularly relevant given the major implication of the neocerebellum in cognitive operations [36]. Our results reinforce the notion that cerebellar circuitry is a major site for internal models, According to this leading theory, expectations and estimates of future motor or cognitive states are critical for performing motor or mental operations [37]. These internal models require updates on a constant basis. Temporal variability can be seen as one of the parameters tuned by the neocerebellum.