Motor cortex encodes forelimb joint movements according to contextual relevance

Abstract

Primary motor cortex (M1) contributes to the control of limb movements which have to be applied not only in isolation but also within purposeful sequences during everyday life. How the M1 neuronal network encodes motion of individual limb joints and in particular how their encoding depends on contextual features of a movement sequence remains unclear. Here we combined two recent techniques to investigate these questions, optogenetic stimulation and two-photon calcium imaging of genetically encoded calcium indicators. Using transgenic mice which express the light-excitable cation channel channelrhodopsin-2 in cortical layer 5 neurons, we first applied optogenetic stimulation through a chronic cranial window to map the motor cortex. Without having to impair the architecture of the underlying neuronal networks with electrode penetrations, we were thus able to identify equivalent motor cortex circuits across mice that are involved in the control of proximal and distal forelimb joints. In the focal area of these M1 circuits, we subsequently employed two-photon calcium imaging to record the activity in neuronal networks of layer 2/3 (L2/3) while head-fixed mice moved across regularly or irregularly spaced rungs on ladder wheels. During skilled locomotion of the animals, we also tracked the motion in proximal and distal forelimb joints using high-speed videography. By predicting kinematics of the individual forelimb joint angles from M1 L2/3 network activity we discovered that finger motion was represented under both conditions whereas encoding of shoulder motion increased for the irregular pattern. Condition-related encoding differences of individual joints correlated with condition-related differences of their grasp-to-grasp variability during the entire movement sequence. This correlation persisted when we only considered discrete grasping actions on the regular and irregular pattern that featured equivalent kinematics in both conditions. We additionally classified three salient forelimb grasp types that occurred under both conditions ('standard', 'corrective', and near-slip 'digit tip' grasps). While the representation of finger motion was particularly high during digit tip grasps, the encoding of shoulder motion on the irregular pattern originated mainly from corrective grasps. Additionally, corrective and digit tip grasps, both of which are associated with an impending fall, could be directly predicted from the activity in neuronal networks of M1 L2/3. Our results suggest that neuronal populations in M1 L2/3 encode motion of individual joints according to their contextual relevance. In a learned movement sequence, neuronal networks incorporate the required grasp-to-grasp variability of individual joints as contextual signature to strengthen the representation of joints with frequent amplitude recalibration. Moreover, the encoding of motion in a joint is upregulated when its control seems to be especially relevant during the execution of a particular grasp type. Our findings are also associated with the forelimb deficits rodents modelling motor cortex stroke or Parkinson’s disease exhibit on the rung ladder, thereby providing a novel framework to investigate the cortical pathophysiology in these motor disorders.