Methodological considerations for a chronic neural interface with the cuneate nucleus of macaques

The mechanisms underlying the emergence of learned motor skill representation in primary motor cortex (M1) are not well understood. Specifically, how motor representation in the deep output layer 5b (L5b) is shaped by motor learning remains virtually unknown. In rats undergoing motor skill training, we detect a subpopulation of task-recruited L5b neurons that not only become more movement-encoding, but their activities are also more structured and temporally aligned to motor execution with a timescale of refinement in tens-of-milliseconds. Field potentials evoked at L5b in vivo exhibit persistent long-term potentiation (LTP) that parallels motor performance. Intracortical dopamine denervation impairs motor learning, and disrupts the LTP profile as well as the emergent neurodynamical properties of task-recruited L5b neurons. Thus, dopamine-dependent recruitment of L5b neuronal ensembles via synaptic reorganization may allow the motor cortex to generate more temporally structured, movement-encoding output signal from M1 to downstream circuitry that drives increased uniformity and precision of movement during motor learning.

The capability to learn novel, complex motor skills is a remarkable ability of human and many other mammals. Although multiple extracortical brain areas, notably the cerebellum and the basal ganglia, contribute to motor learning1, it has been argued that the primary motor cortex (M1) is ideally suited not only for the execution of movement but also the acquisition and storage of motor memory2. Recent finding also highlights the essential role of the motor cortex in tutoring subcortical motor circuits in acquiring motor skills3. M1 neurons are known to comprise a functionally heterogeneous population, encoding distinct motor parameters such as direction, velocity, position and muscle activity4,5,6,7,8,9,10, while jointly they may constitute a pattern generator to encode for the large diversity of possible movements11. The motor representations of individual neurons and neuronal ensembles in M1 are not static, exhibiting adaptation to task requirements, as well as improved coding and predictability of behavioural outcomes with motor learning.

M1 circuitry exhibits interlaminar specificity. Layer 2/3 (L2/3) provides excitatory input to layer 5a (L5a) and layer 5b (L5b), while L5a also relays feedforward excitatory drive to L5b (refs 18, 19, 20, 21). Functional imaging in animals undergoing motor training demonstrated that M1 L2/3 neurons consist of functionally distinct types with different sensory, motor and decision correlates during motor task which are modulated by learning22,23. In the forelimb area of M1, after repeated lever-manipulation task L2/3 neurons exhibit more reproducible population activity in relation to motor execution24. Interestingly, L2/3 neurons retain similar predictability of motor outcome during training of lever-press task, while L5a neurons become progressively recruited for the task and a substantial proportion of neurons become more predictive of lever trajectory after motor training25. However, it remains unknown, how these learning induced reorganizations of neuronal coding in L2/3 and L5a are eventually conveyed to and reflected in the motor representation of the output layer L5b, which is beyond the reach of optical functional imaging techniques (for example, two-photon microscopy) due to its depth. Previous studies showed that M1 neurons encode movement parameters with temporal leads or lags on the order of tens to hundreds of milliseconds5,8,9,26,27; however, how these temporal dynamics in motor representation in M1 change with learning remains virtually unknown.

It has been suggested that reorganization of both the interlaminar and intrinsic horizontal connections in M1 underlies learning-induced plasticity of motor representation. Great insight has been provided by recent in vivo imaging studies of the motor cortex, which revealed rapid changes in spine dynamics on the apical dendrites of L5 neurons within the first few hours28,29,30. These findings suggest a crucial role of long lasting synapse remodelling and connectivity reorganization in the formation as well as consolidation of motor memory. Electrophysiological studies, on the other hand, have revealed that motor skill training could strengthen horizontal connections in the superficial layers 1–3 of the motor cortex, manifested as an increase in local field potentials (FPs) evoked in cortical slices studied in vitro31,32. However, up to now, physiologically induced long-term potentiation (LTP) during motor skill learning has not been demonstrated convincingly and tracked in vivo in the motor cortex.

To probe the process of motor learning and the refinement of motor representation by L5b, we performed a longitudinal study by chronic recording of single-unit activities from M1 in rats performing repetitive motor skill task up to 7 days. We elucidated the fine-scale temporal dynamics of single neuron and population activities from ensembles of L5b neurons, as well as the maintenance of training-induced synaptic plasticity in vivo. We also interrogated the process by examining the effect of depleting dopamine, the neuromodulator that has been consistently implicated in different types of cortical plasticity and learning.