Getting ready to move: transmitted information in the corticospinal pathway during preparation for movement

Abstract

Corticospinal interactions are considered to play a key role in executing voluntary movements. Nonetheless several different studies have shown directly and indirectly that these interactions take place long before movement starts, when preparation for forthcoming movements dominates. When motor-related parameters are continuously processed in several premotor cortical sites, segmental circuitry is directly exposed to this processing via descending pathways which originate from these sites in parallel to descending fibers that derive from primary motor cortex. Recent studies have highlighted the functional role of these interactions in priming downstream elements for the ensuing motor actions. Time-resolved analysis has further emphasized the dynamic properties of pre-movement preparatory activity.

Figure 1. Examples of cortical and spinal preparatory activity during a 2-dimensional isometric wrist task

a, monkeys sat in a primate chair in front of a computer screen and controlled an on-screen cursor by generating a 2D isometric torque at the wrist. Task required acquisition of a memorized target out of 8 presented targets in a circular arrangement. Trials were composed of a pre-cue period, a delay period and a “go” signal after which the monkey acquired the cued-target. During task performance neuronal activity was recorded from motor cortex and spinal cord. b-f, examples of cortical (b-c) and spinal (d-f) neurons recording during the task. Each example shows neuronal activity aligned on cue onset (left panel) a torque onset (right panel); the two panels are separated because the duration of the delay period varied. Note also that there was an overlap between the post-cue period and the pre-torque period. Trials were sorted according to target direction (target 1 is the upper-most target and numbering is clockwise). Peri-event time histograms (upper left plot in each example) quantify neuronal activity around cue onset irrespective of target direction. Examples e and f show two spinal neurons recorded simultaneously: one neuron (e) shows robust pre-movement activity and the second neuron (f) shows only torque-related activity. These examples suggest that pre-movement activity is not a mere reflection of subtle pre-movement modulations in torque level.

Figure 2. Population-based measures of pre-movement delay period activity in motor cortex and spinal cord

a, we measured the evoked multi-unit activity (MUA) around cue-onset for cortical (red) and spinal (blue) sites recorded from three monkeys during the task described in figure 1a. Single site data were first normalized using pre-cue activity to provide a normalized response (expressed in standard deviation units) and were then averaged across all sites recorded from three different monkeys. Note that about 150 ms after cue onset there was a clear increase in cortical MUA. Spinal MUA at roughly the same time shows a brief decrease followed by a monotonous increase that resembles the cortical response pattern. b, histograms of the peak-time of cue-triggered cortical and spinal responses show overlap indicating a similar time scale for responses at the two sites. Note that as spinal response appeared monotonous with no clear peak we used the peak of the response derivative. c, we first computed the cross-correlation between cortical and spinal evoked MUA potentials before and after each behavioral event. The scatter plots show the peak amplitude of the cross-correlation computed after the event vs. the peak amplitude for the cross-correlation computed before the event. Scatter plots are shown for the following events: trial onset (left), cue onset (middle) and torque onset (right). Structure of trials is shown in figure 1a. Note that although for trial onset (a cue which provided no motor-related information) the event had no effect on peak correlation magnitude, for both cue onset and torque onset CS correlations were stronger after the event compared to the pre-event correlation. d-e, for two monkeys that were trained to perform a one-dimensional wrist task which included only flexion and extension targets we analyzed the fraction of cortical (d ) and spinal (e) neurons that were directionally tuned at different time bins along the trial. For each population of neurons we computed the coding fraction for the entire set of task-related neurons (light red and blue dots) as well as for the subpopulation of neurons that were directionally tuned during the active hold period (dark red and blue dots). The gray block reflects the fact the delay period varied in time on a trial-to-trial basis. Dashed line reflects the threshold above which the fraction of neurons deviated from the expected number based on the pre-cue level and variability. The population recruitment fraction (PRF - fraction of neurons that were significantly tuned to torque direction) varied along the trial and was maximal during hold. Note that cortical PRF exhibited a sustained and robust increase while the spinal PRF at the same period expressed only a transient and smaller peak.

Figure 3. Dynamic modulation in the coordinate frame during the delay period

Monkeys were trained to perform the 2-dimensional wrist task with their hand in either a pronation or supination position. The difference between the preferred directions (ΔPD) of tuned neurons in the two hand postures was used to identify the coordinate frame in which these neurons operated as either extrinsic (ΔPD = 0°) or intrinsic (ΔPD ≈ 70°). We found that for cortical neurons (red triangles) the coordinate frame slowly rotated during the trial from a nearly extrinsic frame near cue onset to a nearly intrinsic frame during active hold. Spinal neurons at early stages of the delay period were hardly tuned and the obtained averages were noisy compared to the cortical population. The figure indicates that during the delay period a continuous processing of parametric information takes place especially for cortical neurons.

Figure 4. Illustration of possible model to account for the events that take place during preparation for movement in the motor system

In our model two pathways affect spinal circuitry during preparation for movement. The first delivers information from premotor cortex to motor cortex and subsequently via the corticospinal tract to the spinal cord. The pathway delivers task-related information and its main property is the fact that the convergent pattern of the terminal is organized to preserve the information delivered via this pathway. The second pathway connects premotor cortex with spinal cord indirectly via the reticulospinal pathway. Here, the termination pattern is not organized in a manner that preserves task-related parameters and thus a global, mostly inhibitory signal is transmitted. This pathway serves to modulate spinal circuitry by applying global inhibition and thus prevents premature release of the motor action. This organization is consistent with the fact that when stimulating the premotor cortex, the main excitatory spinal measured by muscle response; the effect is obtained indirectly via M1 (see text for details). In this scheme both excitatory and inhibitory processes that take place during preparation for movement have a supra-spinal origin. Note that the premotor cortex is represented here by multiple boxes to illustrate the fact that several premotor sites may operate in parallel during preparation for movement.