Response to “Fallacies of Mice Experiments”

Abstract

In a recent Editorial, De Schutter commented on our recent study on the roles of a cortico-cerebellar loop in motor planning in mice ^1,2. Two issues were raised. First, De Schutter questions the involvement of the fastigial nucleus in motor planning, rather than the dentate nucleus, given previous anatomical studies in non-human primates. Second, De Schutter suggests that our study design did not delineate different components of the behavior and the fastigial nucleus might play roles in sensory discrimination rather than motor planning. These comments are based on anatomical studies in other species and homology-based arguments and ignore key anatomical data and neurophysiological experiments from our study. Here we outline our interpretation of existing data and point out gaps in knowledge where future studies are needed.

Anatomical studies in macaques suggest a prominent connection between primate frontal cortex and the dentate nucleus via thalamus ³. In contrast, projections from the fastigial nucleus to the thalamus are thought to be relatively weak ⁴. Therefore, the dentate nucleus is often associated with cognitive functions. However, the degree of homology across species and the strength of projections assessed anatomically do not necessarily correspond to their importance functionally ^5–7.

We began our analysis in the mouse anterior lateral motor cortex (ALM), a brain region that is critically involved in planning of directional tongue movements ^8–10. ALM projects to the cerebellum via the basal pontine nucleus ^11,12, which sends widespread projections to the cerebellum ^13,14. We examined ALM-cerebellar connections using triple injections in the same brain. A retrograde tracer was injected into ALM to label the ALM-projecting thalamus; two separate anterograde tracers were injected into the fastigial and dentate nucleus to map their projections to thalamus. The anatomical data show clearly that fastigial projections selectively target the ALM-projecting thalamus (primarily VM), whereas the dentate neurons project mostly to regions of VAL that do not project to ALM (Gao et al Extended Data Fig 8).

Electrophysiological data further support the anatomical data. Inactivation of ALM abolished movement-selective preparatory activity in the fastigial nucleus. Thus ALM drives preparatory activity in the fastigial nucleus. Conversely, photoactivation of the fastigial nucleus, but not the dentate nucleus, destroyed preparatory activity in ALM. Therefore, ALM and the fastigial nucleus, form a selective cortico-cerebellar loop that is segregated from loops in which the dentate nucleus participates.

Furthermore, these findings may not be at odds with the primate studies cited by De Schutter. A recent study in non-human primates found that the cerebellar vermis, which provides input to the fastigial nucleus, receive input from the motor cortex via the pons ¹⁵. In addition, anterograde tracing studies have identified fastigial projections to the thalamus ¹⁶. But which part of the frontal cortex the fastigial output targets via thalamus remains unclear. In rodents ¹⁷ and dogs ¹⁸, the fastigial nucleus projects to parts of VM, consistent with the results in Gao et al. More quantitative comparative anatomy is needed to bridge the anatomical organizations of cortico-cerebellar connectivity in rodent and primate brains.

With regard to the fastigial function, citing fastigial roles in axial and proximal motor control, De Schutter argues that the involvement of the fastigial nucleus in the behavioral task used in Gao et al. is primarily in tactile discrimination rather than motor planning.

We do not find support in our data for this claim. To investigate brain regions involved in motor planning we used a standard delayed response task modified from primate studies ^19–24. Mice judge the location of a tactile stimulus using their whiskers and report their choice using directional licking. Importantly, an intervening delay period separates the sensory stimulus and the time when mice are instructed to respond. The intervening delay thus separates sensation and motor response in time and provides a window of time during which neural activity related to motor planning may be examined ^8,11,25. In Gao et al, fastigial involvement in the behavior was probed using temporally specific optogenetic manipulations during individual epochs of the behavior. Optogenetic manipulation experiments, including both neuronal activation and inactivation, produced the largest effects when perturbing the fastigial activity during the delay and response epochs, as compared to the sample epoch (see Gao et al. Figs 1h and 1k, Extended Data Figs 1e and 2g). Importantly, the same manipulations in the dentate nucleus did not produce any detectable effect in the delayed response task. These results are consistent with the view that the fastigial nucleus plays a role in preparing and initiating directional tongue movement.

Electrophysiological data further support this interpretation: movement-selective preparatory activity is observed in the fastigial nucleus and is strongest during the delay and response epochs. The time course of the optogenetic manipulation behavioral effect size matched the time course of neuronal selectivity (Compare Gao et al. Figs 1 and 2).

Our studies do not exclude potential involvements of the fastigial nucleus in other aspects of the behavior, including tactile discrimination. We note that some behavioral effect was induced by perturbing fastigial activity during the late sample epoch. The fastigial nucleus is known to control proximal body parts, including eye movement ^26,27 and licking ²⁸. The fastigial nucleus also receives input from Purkinje cells that respond to whisking and touch ^29–31. It is conceivable that the fastigial nucleus could play roles in integrating whisker touch into motor plans that instruct proximal body parts ^32,33. The dentate nucleus on the other hand, may be recruited during motor planning in other motor behaviors.

Citing a recent study on the involvement of the dentate nucleus in a visually-guided behavior ³⁴, De Schutter suggests that changing the modality of the sensory stimulus (e.g., from tactile to visual) would cease the involvement of the fastigial nucleus in the delayed response task, and instead cause the dentate nucleus to be involved. We find this unlikely for two reasons. First, the involvement of the dentate nucleus in the delayed response task was clearly ruled out by multiple experiments (e.g., lesion, optogenetic photoactivation). Second, important task differences between Chabrol et al. and Gao et al. should be taken into consideration. In Chabrol et al., mice ran through a virtual corridor, and based on visual stimulus mice stopped running and licked for anticipated reward. Licking and locomotion were strongly anti-correlated. Mice therefore likely planned and coordinated running and licking. In Gao et al., mice sat in a tube and they were primarily concerned with licking. The fastigial nucleus forms a cortico-cerebellar loop with ALM through VM thalamus, whereas the dentate nucleus targets somatic M1 through VAL thalamus. Rather than a single deep cerebellar nucleus performing motor planning, it is more likely that different regions of the cerebellum interact with distinct regions of the frontal cortex during different motor behaviors through parallel cortico-cerebellar loops ³⁵.

De Schutter acknowledges that the findings in Gao et al. are in line with a growing body of literature that supports the role of the cerebellum beyond online motor control ^33–47. The functions of the cerebellum in nonmotor behaviors are just beginning to be elucidated. Resolving the functional organization of cortico-cerebellar loop underlying different forms of cognitive behaviors will be an important area of future research.