Understanding how circuits controlling action in space are organised requires the development of tools that allow us to reveal these circuits and manipulate them.
Hence, the first part of our work consisted in the development of a novel method to trace and manipulate neural circuits. We designed a self-inactivating Rabies virus (SiR), which retains the ability to jump synapses,but does not cause neuronal death. During the initial viral infection, the neurons in the network are permanently labelled and genetic manipulation is made possible. The virus can modify the neurons so they constantly produce a fluorescent protein, allowing the network to be visualised. Crucially, we modified the virus by adding a tuneable switch – the SiR contains selected proteins which can be targeted for degradation. Without these proteins, the virus is unable to replicate and switches off after the initial infection, thus avoiding death of the neurons and leaving them permanently labelled but virus-free. This will be important at a basic science level to understand how the brain works, and excitingly will also offer insight into how neural networks change upon experience and learning. In addition, this new technique will pave the way to the engineering of the genetic content of selected neurons within a network, which would be extremely useful in neurodegenerative and psychiatric disorders where disease genes could be targeted, with therapeutic potential. Ref. ((Ciabatti et al., 2017; Ciabatti et al., 2020; Lee, Ciabatti et al., in press).
We then began to use these tools to dissect circuits controlling goal-oriented actions. We found that the superior colliculus, a region of the midbrain, is where three-dimensional motor space is mapped, and importantly we show how the activity of neurons in this brain region leads to controlled spatially targeted movement. As mice primarily interact with their surroundings by orienting their head, rather than moving their limbs or eyes towards targets, we developed a lightweight head-mounted inertial sensor system to record rotations of the head in three-dimensional space. This system was paired with electrophysiological recordings to determine whether neural activity was related to head movements and whether a map of motor space exists within the superior colliculus. The findings revealed that neurons in the superior colliculus of mice code for head rotations around any of the three rotational axes (the yaw, pitch and roll axes). Neurons whose activity was motor-related showed a preference for a particular angle and direction of rotation, while increased activity of these motion-tuned neurons further modulated the speed of head rotations.
These results provide the first evidence of a three-dimensional representation of motor space in the superior colliculus and give new insights into the mechanisms underlying the control of spatially-targeted movement. This finding suggests that neurons in the superior colliculus not only code for the direction of movement, but also the speed depending on the demands of the task.
By studying the neural representation of spatially-tuned actions in mice, this work also opens up a door to the study of motor-space representations at a genetic level. This will allow for a better understanding of how networks of neurons represent the relationship between space and motor behaviour. Ref. (Wilson et al., 2018)
Following up on this work, we then defined the genetic identity of neurons responsible for the selection of spatially tuned actions and discovered an unexpected anatomical organization of these neurons.
Using an arrays of different methodological approaches from in vitro preparations to freely moving mouse models, we dissected the properties, tuning and circuit organisation of neurons controlling the metric of head movements. We identified a neuronal population organised in clusters tiling the entire SC, forming a grid of discrete spatial modules. These modules are responsible for the execution of head movements whose amplitude vary topographically with the activated module. Our work reveals for the first time a motor map for head movements, begins to address the functional relevance of this SC modularity and paves the way for future experiments to investigate principles of sensorimotor integration in SC circuits. Ref. (Masullo, Mariotti et al, 2019).
Finally, we moved on to characterise the logic governing the sensory motor alignment in the SC. Sensory-guided behaviours require the transformation of sensory stimuli into goal locations in egocentric space in order to elicit an appropriate motor action. In order to understand the nature and logic of sensory-motor integration, intracellular and extracellular recordings were performed across the SC to assess both the motor and the sensory tuning of individual neurons. Whilst traditional models of sensory motor alignment have centred on the mapping between static spatial features, such as stimulus location and movement endpoints, we show that motor tuned neurons in the deep layers of the superior colliculus respond primarily to kinetic visual features, and reveal the existence of an alignment in vectorial space between the encoded sensory flow and movement vectors rather than between visual receptive fields and movement endpoints as previously hypothesised (Reuda et al., in preparation).
