Developmental “awakening” of primary motor cortex to the sensory consequences of movement

Before primary motor cortex (M1) develops its motor functions, it functions like a somatosensory area. Here, by recording from neurons in the forelimb representation of M1 in postnatal day (P) 8-12 rats, we demonstrate a rapid shift in its sensory responses. At P8-10, M1 neurons respond overwhelmingly to feedback from sleep-related twitches of the forelimb, but the same neurons do not respond to wake-related movements. By P12, M1 neurons suddenly respond to wake movements, a transition that results from opening the sensory gate in the external cuneate nucleus. Also at P12, few M1 neurons respond to twitches, but the full complement of twitch-related feedback observed at P8 can be unmasked through local disinhibition. Finally, through P12, M1 sensory responses originate in the deep thalamorecipient layers, not primary somatosensory cortex. These findings demonstrate that M1 initially establishes a sensory framework upon which its later-emerging role in motor control is built.


INTRODUCTION
In placental mammals, primary motor cortex (M1) plays a critical role in adapting behavior 34 to an ever-changing environment (Kawai et al., 2015). Interestingly, M1 does not assume The method used to record electrophysiologically from a head-xed pup. Dots denote locations of the EMGs (forelimb, red; hindlimb, green; nuchal muscle, blue). (b) Top: Flattened cortex sectioned tangentially to the surface and stained for cytochrome oxidase (CO); primary somatosensory cortex (S1) appears darker than the surrounding tissue. Bottom: Boundaries of primary sensory areas from CO-stained tissue, illustrating S1 and primary motor cortex (M1), as well as primary auditory (A1) and visual (V1) cortex.
(c) Enlargement of gray box in (b) showing the somatotopic organization within S1 and M1. hl: hindlimb. Twitch Wake Movement † † Figure 1-supplement 2. Frequency and Kinematics of Twitches and Wake Movements Across Age (a) Mean (± SEM) number of triggered twitches (blue) and wake movements (red) across all pups over 30 min of recording at each age. Dots indicate values for each individual pup. There was no e ect of age on the number of wake movements, but there was for twitches. † signi cant di erence from two days prior (P < 0.005). (b) Mean recti ed EMGs for twitch (light-blue lines) and wake movements (light-red lines) for each pup at each age. Heavy blue and red lines represent median values. Arrowheads denote movement onset. For EMG records of twitches, insets are enlargements of the data in the gray rectangles. To determine whether individual neurons were twitch-or wake-responsive, perievent 129 histograms triggered on twitches or wake movements were constructed for each isolated 130 M1 neuron. We then fit perievent histograms to models of idealized M1 neuronal 131 responses using custom-written MATLAB code. Twitches, being discrete movements, (twitch-responsive, wake-responsive, twitch-and wake-responsive), are shown in Figure   143 2c.

144
Whereas unresponsive neurons (adjusted r 2 ≤ 0.35) did not exhibit increased firing rates 145 after forelimb twitches or wake movements at any age (  P11 and P12 (Figure 3b, blue bars). Correspondingly, the adjusted r 2 wake increased from 160 P8 to P12 (Figure 3b, red bars). The age-related reversal in responsiveness is most 161 visually apparent as a change in the percentage of responsive neurons that are twitch-162 responsive, twitch-and wake-responsive, and wake-responsive across age (Figure 3c).

163
Importantly, all responsive neurons also responded to exafferent stimulation of the  feedback from self-generated movements.

167
The decreased percentage of twitch-responsive neurons observed between P8 and P12 168 is accompanied by decreased variability in those neurons that did respond to twitches.

169
Specifically, twitch-responsive neurons exhibited increases in peak firing rate, narrower 170 response tuning, and decreased latency across these ages, and all of these measures 171 were accompanied by less variability across neurons (Figure 3-supplement 2; top; Table   172 2). Further, P11 and P12 twitch-responsive neurons responded more consistently to 173 every twitch than did twitch-responsive neurons at P8-10. Gaussian-exponential fits of  Table 2), most likely because wake movements are inherently 176 more noisy and variable than twitches.

178
Sensory inputs to M1 in adults have two distinct origins: one arriving from S1 via horizontal conclude that sensory input to M1 arrives directly from the thalamus.

189
Although anatomical evidence suggests that horizontal projections between S1 and M1 190 do not drive neural activity in the deeper layers before P12 (Anastasiades and Butt, 2012), 191 we nonetheless assessed whether M1 reafference is conveyed via S1 by recording 192 simultaneously from M1 and S1 at P8 (n = 6, 94 M1 neurons, 91 S1 neurons) and P12 (n 193 = 6, 135 M1 neurons, 107 S1 neurons; Figure 5a,b). At both ages we observed striking 194 similarities between the activity profiles of S1 and M1 neurons to the same movements 195 (Figure 5c,d). We then calculated cross-correlations of all M1-S1 pairs that were 196 responsive to forelimb movement. To differentiate cross-correlations due to a stimulus 197 (e.g., twitch) from those due to neuron-neuron interactions (e.g., horizontal projections), 198 we analyzed the data using the shift predictor (Engel et al., 1990;Perkel et al., 1967).

199
This analysis revealed that event-triggered S1 activity does not systematically precede   f Lag (s) 3 S1 spikes / M1 spikes / sec M1 S1 M1 S1 Figure 5. Dual Recordings in M1 and S1 (a) Left: CO-stained tissue of the M1 and S1 forelimb representation of a P8 rat showing the location of electrodes. Right: Location of M1-S1 dual recordings for all P8 rats. M1 is shown in blue, S1 in red.
Recording sites from stained tissue (left panels) are designated with yellow highlights in the right panels.
(d) Same as in (c) except at P12. (e) Cross-correlation (bin size = 1 ms) of all available pairs of responsive M1-S1 neurons, minus the shift predictor, for twitches (top row, blue) and wake movements (bottom row, red). These plots signi cant peaks of S1 activity at lags of 0 ms in relation to M1 activity. Gray regions denote con dence bands (P = 0.01).

228
Based on these findings, we conclude that the increase in wake responsiveness of M1   Figure 6. Developmental Change in State-Dependent ECN Activity (a) Model of ECN neuronal activity in response to twitches and wake movements at P9, as proposed previously (Tiriac and Blumberg, 2016). Neurons in the ECN convey twitch-related rea erence to downstream structures, including thalamus and, ultimately, M1. For wake movements, a motor copy inhibits the ECN neuron, preventing the conveyance of rea erence to downstream structures. (b) Proposed model of ECN neuronal activity in response to twitches and wake movements at P12. The ECN's gating of twitch-related rea erence is identical to that at P9. However, at P12, we propose that wake-related rea erence ceases to be gated in the ECN, permitting this rea erence to be conveyed to downstream structures. (c) Left: At P9, representative data depicting 20-s periods of active sleep (blue) and wake (red), showing forelimb movements, multi-unit activity (MUA), sorted unit activity from the forelimb representation of M1, and recti ed EMGs from ipsilateral forelimb and nuchal muscles. Right, top: Raster sweeps for an individual ECN neuron triggered on twitches (blue) and wake movements (red), with each row showing the unit activity surrounding a single movement. Right, bottom: Perievent histogram (bin size = 10 ms) showing mean ring rate for this neuron triggered on twitches (blue) and wake movements (red). (d) Same is in (c) except for a P12 rat.     The percentage of twitch-responsive neurons in M1 decreased suddenly and dramatically 306 by P12 (Figure 3b). At P11, of the responsive M1 neurons, 73% were twitch-responsive; 307 by P12, that percentage had decreased to just 24% (Figure 3c). This sudden decrease in 308 twitch responsiveness is attributable to local inhibition, as disinhibiting M1 using the 309 GABAA antagonist, bicuculline, restored twitch responsiveness to 71% (Figure 8c-e). That

364
In adults, the functional importance of wake-movement reafference for M1 motor control 365 is well established (Iriki et al., 1989(Iriki et al., , 1991. However, at P12, M1 neurons are not yet 366 involved in motor control. This motivates the question: What is the function of wake-367 movement reafference during this early period? We propose that early wake-movement 368 reafference shapes emerging motor control and plasticity.

369
Although adult M1 neurons receive a wide diversity of cortical and thalamic inputs, 370 plasticity is driven through the coincident activation of both horizontal projections from S1 371 and ascending projections from thalamus (Iriki et al., 1989(Iriki et al., , 1991 The traditional designation of primary motor and sensory cortices as exclusively "motor" what has been referred to as a "sensorimotor amalgam" that exhibits a mix of features of 400 S1 and M1 as defined in placental mammals (Karlen and Krubitzer, 2007;Lende, 1963a, 401 b, c). This has led to the hypothesis that S1 and M1 are derived from the same ancestral 402 brain area and that opossums have retained this ancestral state (Beck et al., 1996).

403
Even though M1 in rats is highly specialized, its shared evolutionary history with 404 somatosensory cortex suggests that it should develop similarly to other somatosensory 405 areas, including S1. This suggestion is consistent with a developmental-evolutionary 406 perspective, according to which evolution enables phenotypic transformations in cortical 407 structure through alterations in developmental processes (Krubitzer and Dooley, 2013  were also performed in the ECN (n = 6 at P8-9; n = 7 at P11-12) or the forelimb 450 representation of primary somatosensory cortex (S1; n = 6 at P8 and P12). Additional 451 laminar electrode recordings were performed at P8 and P12 (n = 2 at each age). For M1 452 recordings before and after injection of saline or bicuculline, a total of 12 male and female 453 rats were used at P12 (n = 6 per group). See Supplemental Table 1  At the end of data collection, the pup was euthanized with an overdose of 10:1 558 ketamine/xylazine (> 0.08 mg/kg) and perfused with phosphate-buffered saline (PBS), 559 followed by 4% paraformaldehyde. The brain was immediately extracted and post-fixed 560 in 4% paraformaldehyde for at least 24 h. Next, 24-48 h before the brain was sectioned,

561
it was transferred to a 20% solution of sucrose in PBS until it was no longer buoyant in 562 solution.

563
For all but 3 brains with four shank M1 and S1 recordings, the cortical hemispheres were  Cortical sections were stained for cytochrome oxidase (CO), which has been shown in 577 developing rats as young as P5 to reliably delineate primary sensory areas, including S1

656
The relationship between neural response type and firing rate was assessed as follows:

657
First, using either twitches or wake movements as triggers, perievent histograms (2000-658 ms windows; 10-ms bins) of neural activity were constructed. The average firing rate (in 659 spikes per second, sps) was calculated for each bin and plotted. Next, the twitch-triggered The adjusted r 2 of these models was then calculated and the neuron was considered 672 responsive to a given movement (twitch or wake) if it had an adjusted r 2 > 0.35. Thus, 673 every neuron was classified as one of four types: unresponsive, twitch-responsive, wake-674 responsive, or twitch-and wake-responsive depending on its fit to these models.

675
For responsive neurons, these models also provided estimates of baseline firing rate, 676 peak time, peak firing rate, half-width at half-height, and, for wake movements, half-life. The percentage of twitch-responsive, wake-responsive, and twitch-and wake-responsive 683 neurons were then calculated relative to the total number of responsive neurons.

685
For both twitches and wake movements, triggered movements did not always result in a 686 neural response. Thus, we quantified the percentage of movements (with each twitch or 687 wake movement being an individual trial) that resulted in more activity than would be 688 expected during baseline activity. First, we used the models described above to the correlated activity was induced by the stimulus (twitch or wake movement) and how 715 much was due to neuron-neuron interactions (i.e., between S1 and M1), we calculated 716 the average shift predictor for each neuron pair. The shift predictor compared the activity 717 of an M1 neuron for one movement to that of an S1 or ECN neuron in all instances of the 718 same movement class (twitch or wake movement). Because changes in the patterns of The custom MATLAB code used to fit data to models will be made available upon request.

734
The overall data that support the findings of this study are available from the 735 corresponding author upon reasonable request.  Table 1