From back to front: A functional model for the cerebellar modulation in the establishment of conditioned preferences for cocaine‐related cues

It is now increasingly clear that the cerebellum may modulate brain functions altered in drug addiction. We previously demonstrated that cocaine‐induced conditioned preference increased activity at the dorsal posterior cerebellar vermis. Unexpectedly, a neurotoxic lesion at this region increased the probability of cocaine‐induced conditioned preference acquisition. The present research aimed at providing an explanatory model for such as facilitative effect of the cerebellar lesion. First, we addressed a tracing study in which we found a direct projection from the lateral (dentate) nucleus to the ventral tegmental area (VTA) that also receives Purkinje axons from lobule VIII in the vermis. This pathway might control the activity and plasticity of the cortico‐striatal circuitry. Then we evaluated cFos expression in different regions of the medial prefrontal cortex and striatum after a lesion in lobule VIII before conditioning. Additionally, perineuronal net (PNN) expression was assessed to explore whether the cerebellar lesion might affect synaptic stabilization mechanisms in the medial prefrontal cortex (mPFC). Damage in this region of the vermis induced general disinhibition of the mPFC and striatal subdivisions that receive dopaminergic projections, mainly from the VTA. Moreover, cerebellar impairment induced an upregulation of PNN expression in the mPFC. The major finding of this research was to provide an explanatory model for the function of the posterior cerebellar vermis on drug‐related memory. In this model, damage of the posterior vermis would release striatum‐cortical networks from the inhibitory tonic control exerted by the cerebellar cortex over VTA, thereby promoting drug effects.


INTRODUCTION
For decades, the cerebellum's role has been viewed to be restricted only to motor functions. However, numerous investigations in the last decades have described the involvement of the cerebellum in non-motor functions including language, spatial and emotional processing, reward, working memory, and executive functions (1)(2)(3)(4)(5)(6)(7)(8)(9).
Indeed, the cerebellum plays an important role in the consolidation of non-motor Pavlovian memory and in the establishment of automatic behavioral protocols (4,26).
Moreover, neuroimaging studies of drug-induced cue reactivity in drug addicts described cerebellar activation after the presentation of drug-related cues (27)(28)(29)(30)(31)(32). In a mice model of cocaine-induced conditioned preference, we showed that only those animals that developed preference for cocaine-related cues exhibited increased activity at the apical region of the cerebellar vermis (33)(34)(35). Although this effect was found throughout the cerebellar cortex, only activity in lobule VIII was significantly correlated with the level of preference towards cocaine-related cues (34). Furthermore, cocaine-induced conditioned preference also increased the expression of PNNs surrounding Golgi inhibitory interneurons located in the same region of the vermis (35), suggesting that drug-induced Pavlovian memory encouraged one of the main mechanisms for synaptic stabilization (36). On that basis, it could be expected a neurotoxic lesion localized in lobule VIII may prevent the acquisition of cocaine-induced conditioned preference. On the contrary, the cerebellar lesion dramatically raised by up to 100 the percentage of rats that acquired cocaine-induced conditioned preference (37). The same effect was observed after a reversible deactivation of the infralimbic (IL) cortex (37). Moreover, simultaneous IL-cerebellar deactivation prevented the effect of either of the separate manipulations (37). These results were in agreement with findings reporting that the IL cortex is required for the suppression of cocaine-seeking response and expression of extinction memory (38,39). Overall, our findings suggested that both the cerebellum and IL cortex might act together in regulating the establishment of drug-cue Pavlovian associations.
The present research aimed at: (1) further investigating cerebellum-infralimbic functional relationships for the acquisition of cocaine-induced conditioned preference; and (2) proposing a functional model to explain the effects of the cerebellar lesion in cocaineconditioned memory. First, we addressed a tracing study using anterograde and retrograde tracers in order to build a working neuroanatomical model to explain our prior findings.
Second, we assessed cFos expression in different regions of the medial prefrontal cortex and striatum after the neurotoxic lesion in the apical region of lobule VIII before conditioning. Finally, we explored if the cerebellar lesion might affect synaptic stabilization mechanisms through PNN expression in the medial prefrontal cortex.

Subjects
Twenty-two male Sprague-Dawley rats (Janvier, ST Berthevin Cedex, France) weighing between 175 and 200 g were individually housed under standard laboratory conditions, with controlled temperature and humidity (12-
For the anatomic study, retrograde and anterograde tracers were infused in different regions of the brain. As a retrograde tracer, we used FluoroGold with DAPI (FG) (  immunostaining and camera lucida ( Fig 1A). More detailed information can be found in (37).

Cocaine-induced preference conditioning procedure
The cocaine-induced conditioning procedure has been published previously (37). Briefly, conditioning was conducted using two equally preferred olfactory stimuli located in the walls of a black chamber (20 × 20 × 60 cm) at the opposite arms of a corridor. One of the odors acted as the conditioned stimulus (CS+) and was associated with an IP injection of cocaine hydrochloride (15 mg/kg, IP) (Alcaliber S.A., Madrid, Spain). On alternate days, rats were exposed to the other scent (CS−) placed at the opposite black chamber in the corridor and received 0.9% saline injections. During pairing sessions (15 minutes) animals remained confined in the chamber. A total of eight cocaine-paired sessions were conducted. The olfactory cues and locations in the corridor were counterbalanced between animals. Preference for the cocaine-related cue was evaluated 48 hours after the last cocaine administration in a 30-minute, drug-free test in which CS+ and CS− odors were present simultaneously but in opposite arms of the corridor. The first ten minutes were not considered in order to allow the animal to explore the location of the odors, which was the opposite to the conditioning phase. The preference score was calculated as

Immunohistochemistry and immunofluorescence
The brain tissue was frozen with liquid nitrogen, and sections were performed at 40 µm In a second step, brain samples were exposed to FICT-Streptavidin (1:50; Jackson InmunoResearch Laboratories, Inc., West Grove, PA, USA) and goat anti-rabbit Cy5

Image acquisition and analysis
Fluorescent images of the infusion locations and tracer diffusion were acquired as eight confocal stacked images using the tile-scan tool (Leica DMi8, Leica Microsystems CMS GmbH, Wetzlar, Germany) in order to obtain complete coronal sections in which ROIs were presented.
Images of immunoperoxidase cFos expression were acquired using an optic microscope We included bregma coordinates between 3.20 mm to 2.20 mm for PL and IL, and for the striatum and NAc between 1.60 mm to 0.70 mm.
Fluorescence images of cFos and PNNs were captured in the above-mentioned confocal microscope with 20x lenses and resolution 1,024 x 1,024 dpi Laser intensity (1.0%), gain (600) and offset (-4) remained constant in each analysis. Each image was formed by a stack of ten images. Image stacks were pre-processed applying a maximal projection process with Leica Application Suite LAS X (Leica Microsystems CMS GmbH, Wetzlar, Germany). Three image stacks in coronal sections for brain regions and in sagittal sections for deep cerebellar nuclei were acquired by structure and hemisphere.
FIJI free software (41) was used for all quantifications. Unmodified images were used in all analyses. cFos expression was evaluated using the cell-counter plug-in of FIJI software. Additionally, PNN expression was estimated using a densitometry assessment of WFA intensity (brightness range 0-255) in all the PNNs that were found in three sections of each ROI (35,42,43).

Experimental design and statistics
All statistical analyses were performed using GraphPad Prism 7 software (GraphPad Software Inc., La Jolla, CA, USA). One-way ANOVAs and Student t-tests for independent samples were carried out as parametric tests. Individual scores were provided for each variable. In addition, data were presented as mean ±SEM. Post hoc comparisons were performed using Tukey's HSD tests. The statistical level of significance was set at p < 0.05. When non-parametric statistics was required Kruskal-Wallis and Mann-Whitney-Wilcoxon tests were conducted and data were represented as the median.

Cocaine-induced conditioned preference
In an earlier study, we showed that an excitotoxic lesion in the apical region (dorsal) of lobule VIII before conditioning facilitated cocaine-induced conditioned preference (37).
Indeed, the whole sample of lesioned animals developed a preference for the cocainerelated cue, as compared to only one third of the sham group. In the present study, we confirmed the results with the smaller sample of rats in which immunohistochemical studies were conducted (N=18) [the sham group (n = 6), the lesioned group (QA) (n = 6), and the pseudo-conditioned group (Unp) (n = 6)]. As expected, the cerebellar lesion promoted the acquisition of cocaine-induced conditioned preference [F (2,15) = 5.34, p = 0.0178]. Post hoc analysis revealed that the QA group showed significant higher preference that the sham (p = 0.0463) and Unp (p = 0.0235) groups. Remarkably, no difference was found between the sham and Unp groups (P = 0.9350). Therefore, the neurotoxic lesion was ineffective in affecting the behavior of the Unp group, indicating that lesion effects were learning-related (Fig S1).

Anterograde and retrograde tracing
To propose a working neuroanatomical model that may explain behavioral effects of the lesion in the posterior vermis (37), we carried out an anatomical study using anterograde (BDA/red) and retrograde (FG/blue) tracers. Very recent findings pointed to a direct control of the cerebellum over the ventral tegmental area (9). Therefore, BDA was infused into the apical region of lobule VIII in the vermis (the same location as the excitotoxic lesion) and the Lat nucleus. In turn, we infused FG into the VTA and IL cortex. We searched for colocalization between BDA and FG within the VTA and Lat nucleus (n = 3) (Figs 1-2; S2-S3).
Despite the fact that the infusion point was restricted to lobule VIII in the vermis, BDAlabeled projections were found throughout the whole vermis, but also the hemispheres, reaching Crus I (Fig. 1). However, the molecular layer along the cerebellar cortex was devoid of BDA labeling. BDA is an anterograde tracer that identifies only a neural projection from their source to their point of termination. Hence, one plausible explanation for these results is that Purkinje-to-Purkinje collaterals would have spread the tracer laterally from the middle line along the cerebellar cortex. Purkinje collateral branches originated within the parasagittal plane can reach 2.0 mm towards the apex of the granule cell layer and the base of the lobule in adult rodents (44). Importantly, these findings suggested that lobule VIII in the vermis might be interconnected with distal regions in the cerebellar cortex. As expected, we observed a great number of BDAlabeled terminals within the deep cerebellar nuclei (DCN), especially within the medial nucleus, but also in the IntA; IntP and Lat nuclei (Fig. 1).
When infusing FG into the VTA, FG-labeled somata were found within the contralateral IntA, IntP, and Lat, but not in the Med (Fig. 1). We also observed some FG+ somata in mPFC and NAcSh, among other already described afferences to VTA, what replicated the results observed in other studies (45,46). Then, this cerebellar afference to VTA was confirmed by infusing BDA into the Lat nucleus (Fig. 2). Unilateral BDA infusions into the Lat nucleus reached ipsilateral cerebellar hemisphere, lobule VIII and IX of the vermis, and contralateral parabrachial pigmented nucleus of the VTA (PBP), the most caudal part of the VTA (Fig. 2). At higher magnifications it was possible to observe in the Lat close putative contacts between Purkinje BDA+ terminals from lobule VIII and FG+ cerebellar cells retrogradely labeled from VTA tracer injection (Fig 1).
Additionally, by infusing FG in the IL cortex and BDA in the Lat (n = 2), we observed close apposition between anterogradely labeled BDA fibers and retrogradely labeled FG neurons of the PBP (Fig 2; S2-S3). This finding supports the above-mentioned observations and pointed to the caudal VTA as an interface through which the cerebellum would regulate prefrontal activity and striatal function. Projections from the lateral nucleus seemed to be glutamatergic and contacted TH+ cells and TH-within the VTA  (Fig. 3). The idea of a cerebellar modulation of VTA has been grounded in a few previous findings which described indirect cerebellar-VTA pathways (47,48), as well as direct control of the cerebellum onto the VTA (9,22). Indeed, the cerebellum could reach the VTA through the reticulotegmental and pedunculopontine nuclei (47) and the mediodorsal and ventrolateral thalamus (48). More importantly, there is a direct cerebellar-VTA pathway (22), whose functional properties have been recently delineated in an elegant study (9). PNNs have been proposed as one of the mechanisms for the stabilization of drug-induced memories (36). Interestingly, previous findings demonstrate that both drug-induced conditioned preference (35) and drug self-administration (49)(50)(51) increase PNN expression in the cerebellum and different prefrontal areas, respectively. Furthermore, PNN digestion within the PL cortex (50) and the anterior lateral hypothalamus (49) prevented the expression of cocaine-induced conditioned place preference and cocaine self-administration. The present data suggest that the cerebellar lesion could facilitate the acquisition of cocaine-cue associations and strengthen drug-induced memories by increasing synaptic stabilization mechanisms in the mPFC.

A lesion of lobule VIII in the vermis increases neural activity and PNN expression in
To our mind, the major contribution of the present research is to propose an explanatory hypothetical model for the function of the posterior cerebellar vermis on drug-related memory. In this model that integrate our results with those previously published (9), damage of the posterior dorsal vermis would release striatum-cortical networks from the inhibitory tonic control exerted by the cerebellum over the VTA, thereby promoting drug conditioning ( Figure 6). The model may help to explain why patients with lesions or diseases affecting the posterior cerebellum presented difficulties in controlling their behavior and emotions (52). Moreover, our model predicts that the stimulation of the posterior cerebellar vermis would reduce drug-related effects and improve behavioral inhibition ( Figure 6). Nevertheless, one crucial question for future research is to establish which functional characteristics and connectivity patterns make the dorsal region of lobule VIII in the vermis of special relevance to regulate DCN outputs to the VTA.
In summary, the present results indicate that: (1) the posterior cerebellar cortex may exert inhibitory control over the striatum and mPFC; (2) the Lat nucleus is the main efferent pathway relaying the cerebellar cortex processing to modulate activity and plasticity in the prefrontal-striatal network; and (3) the VTA could be a candidate to mediate cerebellar modulation on activity and plasticity in prefrontal-striatal loops that in turn can regulate cocaine-related behavior.

Authors acknowledge the constructive criticism from Kamran Khodakhah and Saleem
Nicola to the present manuscript.

AUTHOR DISCLOSURE
All authors declare no conflicts of interest.

AUTHORSHIP
All authors made a notable contribution to the manuscript, and they were involved in critically revising the present version. Isis Gil-Miravet performed the stereotaxic surgeries and behavioral experiments; Isis Gil-Miravet and Francisco Olucha-Bordonau performed the tracing study; Isis Gil-Miravet, Edgar Arias de Saavedra-Sandoval, Ignacio Melchor-Eixea, and Lizbeth Vásquez-Celaya were involved in image and data analysis.
Finally, Marta Miquel designed the study, supervised the surgeries and behavioral experiments, was involved in data analysis, and drafted the manuscript. All authors approved the present version of the manuscript.

DATA ACCESSIBILITY
The database is available from the corresponding author upon request.    Data are shown as mean ± SEM and individual scores (*p < 0.05; **p < 0.01; ***p < 0.001). All images on the right panels were taken at 20x magnification. Scale bar 100 µm.

Supplementary information
Figure S1 Effect of a lesion in the apical region of lobule VIII on cocaine-induced conditioned preference. Quinolinic acid lesion were made before conditioning. Lesion Preference scores for the CS+ on the test day in the sham (n = 6), QA lesion (QA) (n = 6) and unpaired (Unp) (n = 6) groups. The QA lesion increased the proportion of rats that expressed cocaine-induced conditioned preference. Data are shown as individual preference scores overlaid mean ± SEM (*P < 0.05)

Figure S2 Infusions of BDA in the Lat nucleus (red) and FG in the IL cortex (blue).
A coronal section of the VTA. White arrows indicate an example of a synaptic contact between the axon of a projection neuron from the Lat nucleus (BDA+) and one THneuron in the VTA (FG+) expressing VGLUT. All confocal images were taken a 40x and 2.5x zoom in a. Scale bar of 10 µm. Right panels: Digital amplification of 300x.

Figure S3 Infusions of BDA in the Lat nucleus (red) and FG in the IL cortex (blue).
A coronal section of the VTA. White arrows indicate an example of a synaptic contact between the axon of a projection neuron from the Lat nucleus (BDA+) and one TH+ neuron (green) in the VTA (FG+) expressing Synapsin 1 in the presynaptic puncta (SYN/purple). All confocal images were taken a 40x and 2.5x zoom in a. Scale bar of 10 µm. Right panels: Digital amplification of 300x.

Figure S4
Effect of an excitotoxic lesion in lobule VIII on DAT expression. IL (A) PL (B). Dopaminergic terminals were identified as TH+. Data are shown as mean ± SEM of the sham (n = 6), QA (n = 6) and Unp (n = 6) groups. DAT expression did not change as an effect of the cerebellar lesion. All images were taken at 20x magnification. Scale bar 100 µm.