Periodic Reporting for period 2 - EscapeVector (Circuit and cellular mechanisms for computing spatial vectors to shelter during escape)
Période du rapport: 2022-03-01 au 2023-08-31
The goal of this project is to understand at the cellular and neural circuit-level how the brain computes actions directed to memorised locations in space. We aim to produce a mechanistic model of how neurons integrate information from sensory and memory systems to determine the position of the body in relation to a memorised target location, and how neural circuits generate actions that move towards the target. This is important because animals, including humans, must constantly navigate through space and move body parts, such as limbs, to locations that are valuable (to obtain food, for example), and often the target location is not immediately reachable by sensory systems (e.g.: it is not visible from the current position); therefore, the movement must be made towards a previously memorised location. This process is essential for satisfying basic survival needs, and obtaining a mechanistic neural model of how actions are generated to reach memorised spatial goals will explain how the brain implements a series of computations essential to every-day life.
To achieve this goal, we study mice and investigate their instinctive escape behaviour to shelter upon imminent threat. In this paradigm, animals are presented with innately aversive sensory stimuli, such visual images that mimic a rapidly approaching predator. When mice see these stimuli, they immediately escape to a shelter, and importantly, they do so not by looking around to see where the shelter is, but by using memory and remembering the shelter location. We believe that mice continuously compute a vector to the shelter (i.e.: the escape route), which represents the goal of the actions that should be taken to reach the memorised location in space where the shelter is. We will use the shelter vector as a model for goal-directed memory-based actions and ask how the brain computes the vector and turns it into motor actions. Answering these questions will not only advance our understanding of how escape actions are computed to reach a memorised shelter location but may also reveal general principles of how spatial goal-directed actions are computed in the brain.
To achieve this goal, we study mice and investigate their instinctive escape behaviour to shelter upon imminent threat. In this paradigm, animals are presented with innately aversive sensory stimuli, such visual images that mimic a rapidly approaching predator. When mice see these stimuli, they immediately escape to a shelter, and importantly, they do so not by looking around to see where the shelter is, but by using memory and remembering the shelter location. We believe that mice continuously compute a vector to the shelter (i.e.: the escape route), which represents the goal of the actions that should be taken to reach the memorised location in space where the shelter is. We will use the shelter vector as a model for goal-directed memory-based actions and ask how the brain computes the vector and turns it into motor actions. Answering these questions will not only advance our understanding of how escape actions are computed to reach a memorised shelter location but may also reveal general principles of how spatial goal-directed actions are computed in the brain.
The main results achieved since starting the project is having identified a neural circuit node that is critical for computing the angle of the shelter vector, and determining how this information is encoded by this circuit. We started by using silicon probes to record neural activity in the brain of mice exploring an environment with a shelter and found two brain regions that continuously represent the shelter direction in body-centred coordinates; that is, they represent how much and in which direction the animal needs to turn in order to face the shelter. These areas are in the cortex and in the midbrain, specifically the retrosplenial cortex and the superior colliculus, respectively. We then used high-resolution electrical recordings and circuit tracing techniques to find that these two areas are synaptically connected. When we inactivated this synaptic connection with a chemo-genetic technique we found that the midbrain lost its representation of shelter direction and that mice now escaped in random directions and missed the shelter. Importantly, the effects of this manipulation were specific to escape: mice could still navigate perfectly fine around the environment to find food, for example, or orient accurately to a salient sensory stimulus.
From these data we built a computer model that contains all the neural elements, synaptic connections and properties that the mouse brain has. The model not only recapitulates the representation of shelter direction, but also suggests that shelter direction is computed in the cortex and passed onto the midbrain because of the specific organization and properties of this neural circuit. A possible advantage of this organization is to use advanced cortical circuits to perform complex computations and distil the result to variables that can be easily converted into actions—here, the shelter direction continuously mapped already in a body-centred reference frame. We believe that the circuit organization that we have found may have evolved to decrease the time to execute accurate actions, which in the case of escaping from imminent threat is of great survival value. The model that emerges from these results may therefore represent a generic brain strategy for using cortical output to generate fast and accurate memory-based goal-directed actions.
From these data we built a computer model that contains all the neural elements, synaptic connections and properties that the mouse brain has. The model not only recapitulates the representation of shelter direction, but also suggests that shelter direction is computed in the cortex and passed onto the midbrain because of the specific organization and properties of this neural circuit. A possible advantage of this organization is to use advanced cortical circuits to perform complex computations and distil the result to variables that can be easily converted into actions—here, the shelter direction continuously mapped already in a body-centred reference frame. We believe that the circuit organization that we have found may have evolved to decrease the time to execute accurate actions, which in the case of escaping from imminent threat is of great survival value. The model that emerges from these results may therefore represent a generic brain strategy for using cortical output to generate fast and accurate memory-based goal-directed actions.
In addition to advancing our understanding of how the mammalian brain computes actions to memorized places in space, we have made two notable technical advances beyond the state of the art. First, the combined implementation of new generation high-density silicon probe recordings with optical fibres and retrograde viral strategy to identify the molecular and anatomical identify of recorded neurons in a freely behaving mouse. Second, the development of an advanced molecular and viral circuit tracing strategy that combines anterograde and retrograde elements and allowed us to demonstrate directly the existence of a specific neural circuit motif between the cortex and the midbrain.
Until the end of the project, we expect to understand how the representation of the shelter vector changes in more complex spatial environments; identify neural circuits responsible for terminating escape, and that therefore have access to the length of the shelter vector; and how synaptic activity from these circuits is integrated to terminate escape.
Until the end of the project, we expect to understand how the representation of the shelter vector changes in more complex spatial environments; identify neural circuits responsible for terminating escape, and that therefore have access to the length of the shelter vector; and how synaptic activity from these circuits is integrated to terminate escape.