The basal ganglia and the cerebellum are major subcortical brain structures that control mainly motor, but also non-motor aspects of behavior initiated by the cerebral cortex (Gold et al., 1989; Middleton and Strick, 2000; Schmahmann and Caplan, 2006; Strick et al., 2009; D’Angelo and Casali, 2012; Schultz, 2016). Traditionally, the basal ganglia and the cerebellum were thought to function in parallel to each other, by having efferent projections to essentially non-overlapping regions of thalamic nuclei (Ilinsky and Kultas-Ilinsky, 1984; Deniau et al., 1992), thereby only interacting at cerebral cortex level (Percheron et al., 1996; Teune et al., 2000; Gallay et al., 2008). In the last decade, however, the use of transneuronal tracers has shown the existence of subcortical disynaptic connections between the basal ganglia and the cerebellum (Bostan et al., 2010). These connections suggest that the basal ganglia and the cerebellum, at least to some extent, might work in concert at a subcortical level.

After establishing a link from the cerebellar nuclei, by way of thalamic nuclei, to the striatum in macaque (Hoshi et al., 2005), Peter Strick’s group established a disynaptic subcortical link between the subthalamic nucleus (STN) and the cerebellum, in particular to lobule VII B and crus II of the cerebellar cortex (Bostan and Strick, 2010). Based on the differential topographically organized connections to both investigated cerebellar regions they postulated that this connection plays a role in both motor and non-motor functions of the cerebellum (Bostan and Strick, 2010; Bostan et al., 2010). Imaging studies that use diffusion magnetic resonance imaging and tractography have shown that this disynaptic connection between the STN and the cerebellar cortex may also be present in the human brain (Pelzer et al., 2013; Milardi et al., 2016). As such, these findings provide neuroanatomical evidence for the putative role of an affected interplay between basal ganglia and cerebellum in movement disorders. This has been largely derived from clinical and animal studies that showed a cerebellar component in disorders typically associated with basal ganglia dysfunction like Parkinson disease (especially in the generation of rest tremors; Benabid et al., 1991; Rascol et al., 1997; Wu and Hallett, 2005, 2013) or dystonia (Chen et al., 2014).

In rodents, a subcortical connection relating cerebellar output to basal ganglia functioning and its potential for inducing dystonia has recently been studied using wildtype and genetically modified mice (Fremont et al., 2015; Liu et al., 2015; also see Shakkottai et al., 2016). However, a reverse subcortical pathway from the basal ganglia to the cerebellum, to our knowledge, has not been demonstrated or investigated in rodents. Given the demonstration of this pathway in the primate and the surge of recent interest in the interplay between basal ganglia and cerebellum, we wanted to see if the STN—cerebellar connection could also be established in rodents and to what extent it differed with that of primates. Therefore, similar to the primate study by Bostan et al. (2010), we have studied a potential disynaptic connection between the STN and the cerebellar cortex in the rat by injecting a mixture of unmodified rabies virus (RABV: CVS 11) and cholera toxin B-subunit (CTb) at several places of the posterior cerebellar cortex (Suzuki et al., 2012). In total, nine small and confined injections were made in the posterior cerebellar cortex ranging from the vermis of lobule VII, to the paravermis and hemispheres of the paramedian lobule and crus IIb. The survival time was set to allow for disynaptic retrograde transneuronal infection of RABV, the resulting second order labeling was then related to the RABV/CTb injection site.

The results indicate that in the rat a disynaptic connection is present but more prominently reaches the vermis as compared to the hemispheres of lobule VII or crus2b.

The experiments with RABV/CTb injections were all performed on male Wistar rats (200–300 g), which have also been used previously in a related but different study (Suzuki et al., 2012). Surgical procedures adhered to the European guidelines for the care and use of laboratory animals and were approved by the ethics committee in Neurosciences at the Institut de Neurosciences de la Timone, INT-Marseille. This study was carried out in accordance with the recommendations of the same committee. The protocol was approved by the same Ethics committee in Neurosciences (n° 02167–01). Vaccinated personnel conducted all RABV handling and surgery and animal care procedures were carried out at the appropriate biosafety containment level (level 2) in the lab of PC (Marseille). In all nine cases the unmodified “French” fixed strain of the Challenge Virus Standard (CVS) 11 was used as a tracer (Ugolini, 2011; Ruigrok et al., 2016).

A total of nine injections with a mixture of RABV and CTb into lobule VII were evaluated for the present study. Three injections were confined to the vermis, four injections to the paravermal and hemispheral parts of the paramedian lobule and the remaining two injections were centered on crus 2B. The location and size of the injection was evaluated in CTb-processed sections (Figure 1A) and, as the organization of the olivocerebellar projection is particularly well known in rats (Sugihara and Shinoda, 2004; Voogd and Ruigrok, 2004; Pijpers et al., 2005), incorporated evaluation of the extent of retrograde labeling in the inferior olive (Suzuki et al., 2012). Retrograde, i.e., somatic, labeling of neurons outside of the cerebellum was carefully evaluated in both CTb- and RABV-processed sections.

For proper evaluation of the results it is vital to demonstrate unequivocally that no transneuronal labeling beyond second-order could have occurred, as this would dramatically increase potential routes of infection. The restriction of transneuronal labeling to one synaptic step was confirmed in the cerebral cortex, where in all studied cases, RABV -infected neurons were always confined to layer V and resembled pyramidal neurons (Bostan and Strick, 2010). No RABV labeling was encountered in other cortical layers (Figures 1B1,B2) or in the dorsal thalamus (not shown). As the pontine nuclei as well as the reticulotegmental nucleus of the pons both serve as major sources of mossy fibers to lobule VII and receive their major afferent input from layer V pyramidal cells of the cerebral cortex (e.g., see Ruigrok et al., 2015), this observation was in line with the disynaptic corticopontocerebellar pathway. Moreover, this indicated that all other RABV labeling was due to either first- or second-order infection.

In the diencephalon, although no RABV labeling was encountered in the dorsal thalamus and epithalamus, many infected neurons were seen in the subthalamic region, in particular in the prerubral field, fields of Forel, zona incerta and, more sparsely, within the STN. As these areas did not contain any CTb-labeled neurons, we conclude that the RABV labeling consisted of second-order infected neurons only. Scattered RABV labeling was also observed in hypothalamic areas, where, occasionally, also a CTb-labeled neuron could be observed (Dietrichs and Haines, 1989). Here, however, we will further focus on RABV labeling observed within the STN (Figures 1C1,C2). In evaluating this labeling it is also important to note that no RABV infected neurons could be observed in the external globus pallidus.

Only in one of the examined cases (case 1080), not a single RABV-infected neuron was observed within the STN despite infection of neurons in the zona incerta and prerubral field. The remaining eight cases all showed at least several and up to 84 infected STN neurons (determined in a one out of four series of plotted sections). Figure 2 shows the location and density of RABV-labeled neurons in color-coded diagrams based on 160 μm wide bins that were oriented along the long axis of the STN. The binning diagram is seen at the top of Figure 2. After color-coding the number of RABV neurons within every bin overlying the STN bin and flattening the result, a dorsomedial view of both the ipsi- (right side) and contralateral (left side) of the STN is presented. The accompanying injection site is indicated in a figurine of the rat cerebellum as seen from caudal. It is striking that the number of RABV-labeled neurons in the STN was most impressive when the injection was centered on the vermis of lobule VII (top two cases). Injections in the paramedian lobule or crus 2A resulted in only a few scattered RABV-infected neurons that were rather evenly distributed over the contra- and ipsilateral STN (Figure 3A). Figure 3B shows the summed mediolateral and rostrocaudal distribution of STN labeling of all eight cases with RABV neurons in the STN. These diagrams show that these neurons could be found throughout every aspect of the STN but mostly in the central part, which seems to correlate well with the general distribution of the number of STN neurons in this lens-shaped nucleus.

From these results it can be concluded that in the rat the vermis of lobule VII is more prominently reached by a disynaptic connection from the STN than its hemisphere. We have asked ourselves what the intermediary of this connection might be. It has been proposed that the pontine nuclei could be a likely candidate (Bostan and Strick, 2010). As, in the course of another study (Suzuki et al., 2012) we have plotted and determined the number of CTb-labeled neurons within the basal pontine nuclei and the reticulotegmental nucleus of the pons of the cases examined in the present study, we correlated the summed numbers of CTb-labeled pontine neurons with the numbers of plotted RABV neurons in the STN (Figure 4). It is obvious that both cases with vermal injections that resulted in the highest numbers of STN labeled neurons also harbored the highest number of CTb-labeled neurons in the pontine nuclei.

Yet, we think alternative pathways should also be considered as, in rodents, an STN projection to the basal pontine nuclei has not been reported in an extensive retrograde study (Mihailoff et al., 1989) and also with anterograde tracing an STN projection to the pontine nuclei is not observed in mice (e.g., see case 146986331 image 74–75 for the STN injection site and images 89–97 (source: Website: © 2015 Allen Institute for Brain Science. Allen Mouse Brain Connectivity Atlas [Internet]. Available from: http://connectivity.brain-map.org). Therefore, we also want to propose a potential alternative route. The STN mostly is a source of afferents of the entopeduncular nucleus (i.e., internal part of the globus pallidus) and the substantia nigra. In addition, connections have been described to the pedunculopontine tegmental nucleus (PPTg; Takada et al., 1988; Granata and Kitai, 1989; Kita and Kitai, 1991; Semba and Fibiger, 1992). As the PPTg is known to contain many cholinergic neurons and, among many other targets, also projects to the cerebellum where efferents terminate as thin varicose fibers in the upper granular and deep molecular layer (Jaarsma et al., 1997), we have qualitatively examined the density of these cholinergic fibers in lobule VII, paramedian lobule and crus 2b. Figure 5 illustrates that, indeed, the density of varicose ChAT-positive fibers is highest in the vermis and much reduced in the hemispheres of lobule VII. It may be noted that in the vermis of lobule VII also ChAT-positive mossy fibers can be observed. Although the origin of specifically these fibers is not known, we have shown that the lateral paragigantocellular nucleus is a likely candidate for similar ChAT-positive mossy fibers to lobule IXa/b of the rat (Jaarsma et al., 1997). Moreover, as the same study shows that the PPTg does not provide mossy fibers to the cerebellar cortex, it was concluded that the ChAT-positive varicose fibers in the cerebellar cortex, at least partly, may be derived from the PPTg. The differential distribution of these ChAT-positive varicose fibers over vermis and hemisphere of lobule VII suggests that the PPTg input may favor vermal areas and, as a consequence, may correlate with a potential STN-PPTg-cerebellar projection (Figure 6).

Recently, in both physiological and pathophysiological conditions, the function of the basal ganglia and the cerebellum systems is increasingly advocated to the subcortical interconnections between these structures (Bostan and Strick, 2010; Helmich et al., 2013; Chen et al., 2014; Louis, 2016; Caligiore et al., 2017). Anatomically, however, disynaptic connections have only been established in primates where they were demonstrated to originate from the cerebellar nuclei to the striatum and from the STN to the cerebellar cortex (for review, see Caligiore et al., 2017). In the present study, we can confirm a disynaptic link from the STN to the cerebellar cortex in rats. However, contrasting primate results we found a more prominent route from the STN to the vermis as compared to the cerebellar hemisphere (see Bostan et al., 2010). Moreover, as anatomical routes from the STN to the pontine nuclei are essentially ill-described, we propose a potential alternative pathway, with the PPTg acting as an intermediary of the disynaptic STN cerebellar pathway.

Using transneuronal and conventional tracers, we show that a principal basal ganglia nucleus, the STN, is disynaptically connected to the vermis of lobule VII of the cerebellar cortex of the rat. Additionally, we show that the PPTg might be a potential candidate as the intermediary element in this connection. We furthermore suggest that this connection has the potential to modulate cerebellar cortical functioning, especially in non-motor related behavior. It is clear that further assessment of the role of the STN-cerebellar link requires a definite establishment of the intermediary in the STN—cerebellar connection. Furthermore, the observed differences between rat and primates need to be confirmed and further investigated. This information may be vital for understanding the ultimate potential of this basal ganglia—cerebellar connection as well as of its potential role in the generation of pathophysiology of both systems.

TJHR and PC designed the experiment. SJ and TJHR analyzed the data and wrote the manuscript.

This research was supported by the Dutch Ministry of Health, Welfare and Sports (TJHR and SJ) and the Centre National de la Recherche Scientifique (CNRS) and Aix-Marseille Université through UMR 7289 (PC).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.