| TYPE Original Research PUBLISHED 12 June 2023 DOI 10.3389/fresc.2023.1205456 Emergence of functionally aberrant and subsequent EDITED BY Carolee Winstein, University of Southern California, United States reduction of neuromuscular REVIEWED BY Vicki Maria Tysseling, connectivity and improved motor Northwestern University, United States Gustavo Balbinot, performance after cervical spinal University Health Network (UHN), Canada *CORRESPONDENCE cord injury in Rhesus V. Reggie Edgerton vre@ucla.edu Gregory Wai1, Sharon Zdunowski1, Hui Zhong2, Jessica L. Nielson3, RECEIVED 13 April 2023 ACCEPTED 22 May 2023 Adam R. Ferguson4, Sarah C. Strand5, Rod Moseanko5, PUBLISHED 12 June 2023 Stephanie Hawbecker5, Yvette S. Nout-Lomas6, Ephron CITATION S. Rosenzweig7, Michael S. Beattie4, Jacqueline C. Bresnahan4, Wai G, Zdunowski S, Zhong H, Nielson JL, Ferguson AR, Strand SC, Moseanko R, Mark H. Tuszynski7,8, Roland R. Roy1,9 and V. Reggie Edgerton2,10,11* Hawbecker S, Nout-Lomas YS, Rosenzweig ES, 1 Departments of Integrative Biology and Physiology, University of California, Los Angeles, Los Angeles, CA, Beattie MS, Bresnahan JC, Tuszynski MH, United States, 2Rancho Los Amigos National Rehabilitation Center, Rancho Research Institute, Downey, Roy RR and Edgerton VR (2023) Emergence of CA, United States, 3Department of Psychiatry & Behavioral Sciences and the Institute for Health functionally aberrant and subsequent reduction Informatics, University of Minnesota, Minneapolis, MN, United States, 4Brain and Spinal Injury Center, of neuromuscular connectivity and improved Department of Neurological Surgery, University of California, San Francisco, San Francisco, CA, United motor performance after cervical spinal cord States, 5California National Primate Research Center, University of California, Davis, Davis, CA, United injury in Rhesus. States, 6Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Front. Rehabil. Sci. 4:1205456. Colorado State University, Fort Collins, CO, United States, 7Veterans Administration Medical Center, La Jolla, CA, United States, 8Department of Neuroscience, University of California, San Diego, La Jolla, CA, doi: 10.3389/fresc.2023.1205456 United States, 9Brain Research Institute, University of California, Los Angeles, Los Angeles, CA, United COPYRIGHT States, 10Institut Guttmann, Hospital de Neurorehabilitacio, Universitat Autonoma de Barcelona, Badalona, © 2023 Wai, Zdunowski, Zhong, Nielson, Spain, 11Neurorestoration Center, University of Southern California, Los Angeles, CA, United States Ferguson, Strand, Moseanko, Hawbecker, Nout-Lomas, Rosenzweig, Beattie, Bresnahan, Tuszynski, Roy and Edgerton. This is an open- Introduction: The paralysis that occurs after a spinal cord injury, particularly during access article distributed under the terms of the the early stages of post-lesion recovery (∼6 weeks), appears to be attributable to Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other the inability to activate motor pools well beyond their motor threshold. In the forums is permitted, provided the original later stages of recovery, however, the inability to perform a motor task author(s) and the copyright owner(s) are effectively can be attributed to abnormal activation patterns among motor credited and that the original publication in this journal is cited, in accordance with accepted pools, resulting in poor coordination. academic practice. No use, distribution or Method: We have tested this hypothesis on four adult male Rhesus monkeys reproduction is permitted which does not (Macaca mulatta), ages 6-10 years, by recording the EMG activity levels and comply with these terms. patterns of multiple proximal and distal muscles controlling the upper limb of the Rhesus when performing three tasks requiring different levels of skill before and up to 24 weeks after a lateral hemisection at C7. During the recovery period the animals were provided routine daily care, including access to a large exercise cage (5’ × 7’ × 10’) and tested every 3-4 weeks for each of the three motor tasks. Results: At approximately 6-8 weeks the animals were able to begin to step on a treadmill, perform a spring-loaded task with the upper limb, and reaching, grasping, and eating a grape placed on a vertical stick. The predominant changes that occurred, beginning at ∼6-8 weeks of the recovery of these tasks was an elevated level of activation of most motor pools well beyond the pre- lesion level. Frontiers in Rehabilitation Sciences 01 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 Discussion: As the chronic phase progressed there was a slight reduction in the EMG burst amplitudes of some muscles and less incidence of co-contraction of agonists and antagonists, probably contributing to an improved ability to selectively activate motor pools in a more effective temporal pattern. Relative to pre-lesion, however, the EMG patterns even at the initial stages of recovery of successfully performing the different motor tasks, the level of activity of most muscle remained higher. Perhaps the most important concept that emerges from these data is the large combinations of adaptive strategies in the relative level of recruitment and the timing of the peak levels of activation of different motor pools can progressively provide different stages to regain a motor skill. KEYWORDS muscle activity patterns, motor performance, spinal plasticity, spinal cord injury, EMG Introduction stimulation parameters in uninjured individuals (4). In addition, unusually diverse combinations of muscles are activated after an Significant levels of recovery of upper limb function can occur incomplete spinal cord injury when the subjects attempt to after a cervical spinal cord injury and the severity of the deficits are voluntarily initiate a movement (9). A spatial reorganization of largely a function of the level of the cervical injury and the severity sensory-evoked potentials projecting to the sensorimotor cortex of the lesion. We have reported that after a unilateral cervical lesion also can occur within hours after a peripheral nerve injury (12) (C7) in nonhuman primates (1–3), there is extensive branching of or amputation of the distal tip of a digit (13). Similar the corticospinal axons projecting across the midline to the injured observations have been observed in response to the stimulation side. We also showed that recovery of locomotor function was of spinal networks via spinal epidural (14) or noninvasive substantial and there was a gradual change in the coordination transcranial stimulation (15) following a spinal cord injury in patterns of agonist and antagonist muscle activity. The present human subjects. paper examines further details of the post-lesion recovery of In the present study we quantified EMG patterns of proximal motor function. We reasoned that to regain some systemic-level and distal flexor and extensor muscles in the upper limb of a understanding of this reparative process every movement is nonhuman primate while performing a relatively automatic task, defined essentially by two basic physiological variables, i.e., the i.e., stepping on a treadmill, as well as during tasks requiring net level of excitation of each motor pool and the precision of finer “conscious” control to include pulling a spring-loaded the coordination among the motor pools that generate the motor handle and removing a grape on a stick before and after a task. Given the fundamental importance of these two variables, cervical lateral hemisection. Comparison of the performance of we systematically and comprehensively assessed the physiological these different motor tasks was to determine whether any re- properties that reflect these variables before and up to 24 weeks organization associated with aberrant connections would be more after a cervical lateral hemisection (C7) in the Rhesus. critical in performing a fine motor task vs. locomotion. These One might predict that the loss of function after a cervical comparisons could have clinical implications as to whether any lesion would be due largely to a loss of the ability to activate the expanding functional connectivity could be more prominent motor pools relevant for a particular motor task. This indeed within and/or among the supraspinal and spinal networks. appears to be the case during the very early post-lesion period, but during the subsequent stages of recovery several observations suggest that the reverse may be the case (4–11), i.e., an Materials and methods excessively high number of motor pools and motor units are recruited when performing a given task post-injury. This time Subjects course of events suggests the formation of functionally aberrant connections during the initial stages of recovery of movement Four adult male Rhesus monkeys (Macaca mulatta) (6–10 followed by a gradual decrease in the functional connectivity years old, mean of 8.6 years; mean weight: 12.2 kg) were studied. with post-injury time. All surgical and experimental procedures were carried out using Theoretically, the consequences of forming additional the guidelines set by the Laboratory Animal Care (US National functionally aberrant connections at any point along the Institutes of Health Publication 85–23, revised 1985) and were sensorimotor axis would be an excessive excitation of motor approved by the Institutional Animal Care and Use Committee pools and an increase in instances of co-contraction. For (IACUC) of the California National Primate Research Center example, electromagnetic stimulation of the motor cortex in (CNPRC) at the University of California, Davis. Specific human subjects after a spinal cord injury activates multiple protocols with regard to sleep, shelter, and nourishment are muscles within a few days: a result rarely observed with the same outlined in Nout et al. (1, 2). Frontiers in Rehabilitation Sciences 02 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 Task training and testing transducer attached in series with the handle and spring was used to record force and this signal was synchronized with EMG Pre-lesion, the monkeys were trained to step on a treadmill recordings from multiple muscles (see below). twice/week. They also were trained to perform a handle pull and a grape on a stick (GOAS) task with their ipsilateral (lesion side) Exercise cage activity forelimb while in a restraint chair three times/week for 30-min/ All animals were placed 4–5 times per week in a large open session. Each chair trial began with the limb in a trained resting wire-mesh cage (7 feet [H ] × 10 ft [W ] × 6 ft [D]) with a series position using a clear plexiglas partition to only allow access to of 4 ascending elevated perches on the back and side wall of the the food reward with the arm on the impaired side and to serve enclosure, and a series of food cups hanging diagonally at as a signal for the subject to begin the task. The trial ended different heights (1.5, 2.7, 3.5, 4.3, 4.7 ft) along the front of the when the handle was pulled maximally and released, the grape cage. Each session began with cage entry through a chute onto was retrieved, or there was no retrieval response after 15 s. Food the lowest perch. A large sized Kong® (a hollow rubber toy filled rewards were used for positive reinforcement. After the spinal with small food rewards) was placed on the top perch to cord hemisection (see below), the monkeys were retrained as encourage movement. Use of the impaired forelimb was soon as they were able to participate in the tasks 2–3 times/week encouraged by the need to use both hands to retrieve items from for 30-min/session. the Kong. Next the food cups on the front of the cage were loaded to encourage climbing and impaired limb use. Next, presentation of a large food item (e.g., an apple or orange), a Treadmill task task requiring both hands for support and manipulation during The monkeys were trained to walk quadrupedally on a motor- eating, was used to promote use of the impaired hand. Each driven treadmill. A plexiglass enclosure was utilized to keep the session lasted approximately 30 min and the cage was located in animal in position while allowing video recording of their the housing room. movements. Each training session consisted of 8 trials (two trials at 0.45, 0.89, 1.34, and 1.79 m/sec, 30 s/trial) with approximately one min of rest between each trial. Post-operatively, the Surgical and post-operative procedures maximum treadmill speed was adjusted according to each monkey’s level of recovery, not to exceed 1.79 m/sec. The All surgical procedures were performed under aseptic monkeys were stepped until they performed at least 10 conditions and with the monkeys sedated intramuscularly with consecutive and consistent steps at each speed with the ketamine (1 mg/kg) and then anesthetized to a surgical state with contralateral hindlimb or stepped a maximum of 45 s. isoflurane gas to effect. Body temperature, heart rate, blood gases, respiration rate, and indirect blood pressure were monitored and Grape on a stick task maintained within acceptable levels. The monkeys were required to retrieve a food item (usually a grape) attached to the top of a vertical post that was positioned EMG electrode implantation at about maximum reach length and shoulder height. The post Approximately 6–8 weeks prior to the spinal cord lesion, an also was anchored at the center of a platform containing a force EMG telemetry system (Konigsberg Instruments, Pasadena, CA, sensor with six degrees of freedom. Ten trials were performed USA) having 6 EMG leads was implanted in each monkey. A during each session. For five of these trials, a funnel (base transmitter and battery pack were implanted and secured diameter, 7.6 cm; height, 8.3 cm) was placed over the bottom of between the internal and external oblique muscles on the non- the post to promote retrieval of the grape using a pincer-grip injured side. Two wires (one lead) were tunneled subcutaneously and to prevent substitute forms of grasping. The remaining five into each of the following muscles: triceps brachii (TB), biceps trials were conducted without the funnel. A successful trial was brachii (BB), pronator teres (PT), flexor digitorum superficialis defined as getting the food item to the mouth within 15 s after (FDS), extensor digitorum communis (EDC), and flexor pollicis the food item was presented. No attempt to reach, failure to brevis (FPB) and implanted intramuscularly as described remove the item within 15 s, or removal and drop before previously (1, 2). reaching the mouth were defined as unsuccessful trials and subsequently not analyzed. Spinal cord hemisection A partial laminectomy was performed between vertebral levels Handle pull task C5 caudal and C6. An incision was made in the dura to expose the The monkeys were required to pull a round handlebar (11.4 cm spinal cord and a lateral hemisection was performed at spinal cord long × 6.4 cm wide × 0.64 cm diameter) attached to a plastic guide level C7 using a micro-blade mounted on a stereotaxic carrier as equipped with springs of differing tensions (20 N—easy; 60 N— described previously (1, 2). medium; 98 N—difficult; and 107 N—very difficult). The handle was positioned horizontally or vertically (5 trials in each position Post-operative care per spring). Progressively greater spring tensions were utilized An analgesic (oxymorphone: 0.15 mg/kg) was given thrice post-lesion as the monkey recovered from the surgery. A daily for 3 days and an antibiotic (cephazolin: 25 mg/kg) twice Frontiers in Rehabilitation Sciences 03 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 daily for 7 days. The monkeys were returned to their home cage (21), retaining factors with at least 3 loadings above. (4) PCs after recovery from surgery: the home cages were lined with meeting all three criteria were examined and named using fleece pads and porous rubber mats. Monkeys were monitored loadings above, thereby accounting for at least 20% of the every 2–4 h and were given preferred food items to encourage variance. To test the hypothesis that multivariate PC patterns them to sit up, reach, and stand. change together over time, we first performed a PCA on muscle recruitment and success metrics across all tasks and all timepoints then applied a general linear model (GLM), one-way EMG and video recordings analysis of variance (ANOVA) with Tukey’s post-hoc test for significance to assess the impact of timepoint. Two-way GLM Video and EMG were recorded during all task pre-lesion and ANOVA was performed on PC scores to test the hypothesis that approximately 6, 8, 10, 12, 16, 20, and 24 weeks post-lesion. each task significantly changed over time for PC1. All analyses Video was captured at 100 frames per second and edited using were performed in SPSS v.20 (IBM). Simi Motion software (Simi Realty Motion Systems, Unterschleissheim, Germany). EMG was digitized at 2,000 Hz (National Instruments A/D board, Austin, TX) using custom Results software. The alterations in the patterns of activation of proximal and distal extensor and flexor upper limb muscles were characterized Data processing over a period of up to 24 weeks post-injury. With the chronically implanted electrodes it was possible to make consistent Raw EMG signals were band-pass filtered (20–500 Hz) and comparisons of these patterns of activation throughout the stages rectified. Cycle durations for the handle pull task were of recovery after the lesion and to compare these patterns to determined as the time between the activation of the BB in the those prior to the lesion. The activation patterns were compared ipsilateral forelimb and the return of the spring to its original when the animal was performing a relatively automatic type of position after the pull. For the GOAS task, the cycle duration movement, i.e., treadmill locomotion, and when performing tasks was determined as the time between the onset of the reach of the that required finer motor control, i.e., pulling a spring-loaded ipsilateral forelimb (as seen on video) and the positioning of the handle and removing a grape from the end of the stick that was grape in the mouth. If the monkey was unsuccessful in retrieving instrumented with the 3-D force transducer. Several features were the grape, the end of the cycle was marked when the grape was quantified that reflected the level of activation of each motor pool dropped. Trials where the grape did not leave the stick were not and the coordination among the motor pools during each task. marked. For the treadmill task, an average of ∼10 consecutive and most consistent steps from each trial were chosen for analysis. EMG burst durations, amplitudes, and integrals were EMG activity patterns during locomotion determined as described previously (3). The peak filtered and rectified EMG amplitude for each 0.5 The average EMG patterns for each muscle pre-lesion and up msec time bin was plotted for the duration of each trial for the to 24 weeks post-lesion for Subject 1 are shown in Figure 1A. GOAS and handle pull tasks and for two consecutive step cycles For the three remaining subjects similar data are shown for pre- for the treadmill locomotion. Joint probability distributions were lesion, at an acute stage (6–8 weeks post-lesion), i.e., the first generated to compare the EMG activation levels between selected time that the animal could make a minimal attempt at muscle pairs for each trial for each task (16). We determined the performing the motor tasks, and at the most chronic stage (24 percentage of data points exceeding 15% of a 2 mv maximum on weeks post-lesion), i.e., and the final recording session at both the x and y axes, i.e., 0.3 mv. Given that the majority of the approximately 24 weeks post-lesion (Figures 1B–D). Subjects 1, data points are located less than 0.3 mv for both the x and y 2, and 3 had recovered sufficiently to resume locomotor testing axes (baseline), these data points were excluded in the calculation at 6 weeks post-lesion, whereas Subject 4 was able to resume of the percentage of co-activation (see Supplementary Figure S1). locomotor activity only after 9 weeks. At the end of the experimental period, the strategies used by the subjects to accomplish the task differed significantly (Supplementary Video S1). Subject 1 Statistics displayed only a slight limp and its gait most closely resembled that recorded pre-lesion. Subject 2 contacted the treadmill with the To extract multivariate patterns across correlated muscle dorsal rather than the palmar surface of the hand with the groups during recovery, we applied non-linear principal ipsilateral forelimb during locomotion. Its body weight also component analysis (NL-PCA) using optimal scaling method seemed to be more heavily distributed toward the hindlimbs. (17). We applied well-established rules for PCA extraction and Similar to Subject 2, Subject 3 contacted the treadmill with the PC retention (18). In brief, PCs were retained using 4 criteria: dorsal surface of the ipsilateral hand while stepping on the (1) the Kaiser rule, retaining PCs with eigenvalues >1.0 (19), (2) treadmill. Subject 3 did not bear any weight with its ipsilateral Scree plot (20), and (3) the over-determination of the factors hindlimb, i.e., the limb remained elevated during locomotion and Frontiers in Rehabilitation Sciences 04 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 FIGURE 1 EMG patterns during locomotion. (A) Activation patterns (rectified EMG for ∼10 consistent, consecutive steps) of six upper limb muscles during locomotion on a treadmill at 0.89 m/s pre-lesion and 6 to 24 weeks post-spinal cord hemisection for Subject 1. Similar data for Subjects 2−4 pre- lesion and at an acute stage (6−8 weeks post-lesion) and a chronic (final recording session at 24 weeks post-lesion) stage are shown in (B–D). The beginning of the EMG trace is synchronized with the initiation of stance and all step cycle lengths are normalized. TB, triceps brachii; BB, biceps brachii; PT, pronator teres; FDS, flexor digitorum superficialis; EDC, extensor digitorum communis; FPB, flexor pollicis brevis. resorted to a slight hopping motion on its contralateral side. Subject 4 bursting pattern at the chronic phase. In a number of muscles, also contacted the treadmill with the dorsal surface of its ipsilateral the peak amplitudes remained well above pre-lesion levels even hand during stepping, but displayed a more even interlimb weight- at the chronic stage. The mean amplitudes of the EMG bursts, bearing distribution and more normal stepping pattern. EMG burst durations, and total EMG activity per step cycle pre- While all subjects were able to step on the treadmill and adjust lesion and post-lesion (normalized to the pre-lesion condition) to a range of speeds, there were both similarities and differences in are shown in Figure 2. In many muscles these three EMG the changes in EMG burst amplitudes during the recovery period measures showed greater levels of activation post-lesion across the subjects. The data in Figures 1, 2 demonstrate a compared to pre-lesion. rather consistent and marked increase in the level of activation of The level and pattern of activation in a given muscle all motor pools early post-lesion in Subject 1 and generally in represent two of the characteristics that define the output of each of the other three subjects. In some cases, the EMG activity the muscle. A third factor in defining motor performance is increased in the acute phase and eventually declined, even to a the relative levels of activation of agonist and antagonist level less than was present before the lesion in some muscles. muscles, i.e., how well are the activation patterns coordinated. The other change that occurred in the activation patterns was the Figure 3 demonstrates a rather consistent pattern in the absence of a clear bursting pattern during a specific phase of a changes in coordination between pairs of muscles when step. This was most evident in the acute phase while many of the comparing the pre-lesion and acute and chronic phases of motor pools recovered the ability to generate a more normal recovery for each subject. The amplitudes of the EMG patterns Frontiers in Rehabilitation Sciences 05 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 FIGURE 2 EMG characteristics during locomotion. Mean EMG burst amplitudes (A), durations (B), and integrals (C) and the overall mean (black traces) for each muscle and each subject for each time point. All post-lesion values are normalized to pre-lesion values. Muscle abbreviations, same as in Figure 1. are consistently elevated post-lesion, particularly in the acute phase as noted above (Figure 3A). The joint probability distributions demonstrate a marked increase in the incidence of co-contraction in agonist and antagonist muscles during locomotion post-lesion (Figure 3B). The percent of co- contraction between each muscle pair studied in Subjects 1, 2, and 4 was higher than pre-lesion at the acute phase and then returned toward near pre-lesion levels at the chronic phase (Figure 3C). The maximum treadmill speed at which each FIGURE 3 subject could step at each time point was initially decreased EMG co-activation and locomotor performance. (A) Examples of raw but recovered to a near pre-lesion level during the following EMG activity (2 step cycles for each condition) during treadmill locomotion at 0.89 m/s for Subject 1 for the same muscles and at the weeks (Figure 3D). pre-lesion and an acute and chronic time points post-lesion. (B) Joint probability distributions (see Methods and Materials) of the EMG burst amplitudes between pairs of muscles for the same step cycles shown in (A). The y scales are 2 mv for all plots. (C) Percent co-contraction EMG activity and force patterns during the for each muscle pair at the pre-lesion stage and at an acute and chronic time point post-lesion. (D) Maximum speed of locomotion at GOAS task each time points for each subject. Each subject was able to complete the task by the end of the post-lesion period. The regularity with which they could successfully complete the task and the strategy they used to do for grasping, (b) activation of the PT, FDS, EDC, and FPB to so, however, varied from subject to subject (Supplementary adjust the hand to perform the chosen grasping strategy (Grasp), Video S2). Similar to that observed during the locomotion task, and (c) prolonged activation of the BB as the subject moved the a stereotypical EMG pattern was associated with the pre-lesion grape from the stick to its mouth (Retrieve) (Figure 4A). performance of the GOAS task. This included (a) an initial Compared to pre-lesion, the EMG burst amplitude of most activation of the TB and EDC as the subject extended the arm muscles was higher at both the acute and chronic stages post- towards the grape (Reach) and opened the hand in preparation lesion and Subjects 2, 3, and 4 showed higher levels of forces Frontiers in Rehabilitation Sciences 06 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 FIGURE 5 EMG activity during reaching, grasping and retrieving. Mean EMG integrals for each muscle and for each subject and the overall mean EMG integrals (black line) and durations (bottom row) during the reach (A), grasp (B), and retrieve (C) phases of the grape on a stick task. The mean EMG integrals are normalized to pre-lesion values at each time point. changes in levels of co-activation were generally unrelated to successful performance of the task (compare Figures 4C,D). This variation seems to reflect the unique strategies that each subject adopted to successfully retrieve the grape (see FIGURE 4 EMG characteristics and performance during the GOAS task. Data for Supplementary Video S2). Although the response varied Subject 1 while performing retrieval of a grape placed on a stick among subjects, the percent success was near a pre-lesion level instrumented for recording 3-D forces imposed while obtaining the grape presented in a similar format to that in Figure 2. The red by the end of the 24-week recovery period in three of four vertical lines on the raw EMG graphs in (A) divide the task into reach, subjects (Figure 4D). grasp, and retrieve phases. The resultant of the three force vectors The mean EMG burst integrals during the reach, grasp, and imposed on the stick is shown below the EMG traces. (B). Distribution plot of EMG amplitudes of agonist-antagonists muscles (C) Percent retrieve phases of the GOAS task at all time points are shown in co-contraction for each muscle pair at the pre-lesion stage and at an Figure 5. All four subjects showed similar changes in most acute and chronic time point post-lesion. (D) The integral of the force x time for all successful trials for the task for each subject at each muscles over the course of the experiment. During the reach time point. phase, the integral values remained relatively unchanged post- lesion (Figure 5A). All integral values throughout the recovery period were elevated in all four subjects in the TB, BB, and FDS during the retrieval phase (Figure 5C). This also was true for the imposed on the stick while retrieving the grape. Concomitant BB and FDS during the grasp phase (Figure 5B). The durations with the increase in EMG amplitudes there was a much higher of the reach and retrieval phases remained relatively close to the incidence of co-activation among most pairs of muscles pre-lesion values throughout the recovery period, whereas the studied post-lesion (Figures 4B,C). In addition, the patterns of duration of the grasp phase was elevated at most time points changes in co-activation were unique for each subject and the post-lesion in all subjects (Figures 5A–C). Frontiers in Rehabilitation Sciences 07 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 FIGURE 7 EMG activity and kinetics during the handle pull task. (A) Mean EMG integrals for each muscle and for each subject and the overall mean EMG integrals (black line) and durations (bottom row) during the handle pull task pre-lesion and between 6−24 weeks post-lesion. The mean EMG integrals are normalized to pre-lesion values at each time point. (B) Relationships between individual force values and the corresponding EMG integrals for each muscle for Subject 1 (B) and Subject 4 (C) pre-lesion and at 24 weeks post-lesion. and pulled on it (Figure 6A). As with the other tasks, there was a general increase in the EMG burst duration and amplitude in most muscles of all subjects in the acute and chronic stages post-lesion. There were high incidences of co-contraction (Figure 6B) and this FIGURE 6 EMG characteristics and performance during the handle pull task. Data persisted throughout the recovery period in Subjects 1 and 2 but for each subject while performing the handle pull task instrumented for not in Subjects 3 and 4 (Figure 6C). Although the increased recording 3-D forces imposed presented in a similar format to that in incidence of co-activation observed at the acute stage had subsided Figure 2. Traces of the amount of force generated over the duration of the trial are shown below the raw EMG traces. (B) Joint probability completely in the TB and BB in Subject 3 at the chronic stage distributions of the EMG burst amplitudes between pairs of muscles (Figure 6C), the mean maximum force (Figure 6D) in the handle for the same step cycles shown in (A). (C) Percent co-contraction for each muscle pair at the pre-lesion stage and at an acute and chronic pull task remained high across all time points. time point post-lesion. (D) The average peak force during the task for The mean EMG burst integrals were elevated in most muscles in each subject at each time point. most subjects during the handle pull task post-lesion (Figure 7A). The most common pattern of recovery of EMG patterns and the time taken to complete the task was reaching a peak within about 10 EMG activity and force patterns during the weeks post-lesion and thereafter generally plateauing up to 24 weeks. handle pull task The forces generated during the task at both the acute and chronic phases of recovery were either elevated or about the same as pre- Pre-lesion, the stereotypical pattern of EMG activation during the lesion in Subjects 1, 2, and 3 as illustrated for Subject 1 in Figure 7B. handle pull task included a) activation of the TB and EDC as the In 5/6 muscles studied in Subject 1 the EMG integral was consistently subject extended the arm and prepared to grasp the handle, and b) elevated relative to the force generated (Figure 7B). In contrast, the activation of the FDS, BB, and PT as the subject grasped the handle forces generated in Subject 4 were much smaller post-lesion Frontiers in Rehabilitation Sciences 08 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 compared to pre-lesion and there was little relationship between the calculated the correlation (PC loading) of each individual variable EMG levels and the forces generated (Figure 7C). Subject 1 was able with the PC axis. To measure the hypothesized effects of time post- to successfully complete the spring pull task with the stiffest spring at injury on the multidimensional pattern, we calculated the precise PC the pre and all chronic time points, as reflected by the large spread of score for each subject. forces in Figure 7B. Subject 4 was able to pull only the least stiff PC1 loadings revealed an inverse relationship between recovery spring throughout the post-surgical recovery period (Figure 7C). success and muscle co-contraction for each task (Figure 8A). Analysis of PC scores by time point-post injury revealed a significant global pattern of co-contraction (Figure 8B) associated Global assessment of the relationship with impairments in function at the acute post-injury time point between task performance and the levels of that were partially restored at the chronic time point. There was not co-contraction of antagonistic muscles a significant difference between the tasks in this global pattern, i.e., over time the time x task interaction was not significant (Figure 8C). To assess the global impact of the post-injury time-point on the relationship between the levels of task recovery and muscle co- Discussion contraction we applied non-linear principal component analysis (NL- PCA) separately for the spring, grape-on-a-stick, and locomotion on Motor performance after a partial spinal a treadmill tasks. NLPCA analyzes the cross correlations across all cord injury (hemisection) was limited by outcomes, combined with optimal scaling transformations, to detect poor coordination of motor pools multidimensional patterns using an alternating least squares approach and represent these patterns as synthetic axes known as the principal The present results suggest that the limitations in performance of components (PCs). To understand what the PC axis represents, we the motor behaviors studied, once beyond the acute stage of the injury, FIGURE 8 Multivariate analysis of the recovery of behavioral success and muscle co-contraction post-injury. Non-linear principal component analysis (NLPCA) was performed on time series data from muscle integrals and co-contraction distributions during the recovery of success rate for the spring, grape on a stick (GOAS) and treadmill locomotion tasks. (A) The first principal component (PC1) accounted for a substantial percentage of the variance in each dataset (39.5%, 32.4%, 50.2%, respectively), and generally represents the inverse relationships between the recovery of task success rate and muscle co- contraction and integrals. PC loadings represent the correlations for each variable onto the entire multivariate PC space (shown separately for spring, GOAS and locomotion tasks): red indicates a positive correlation, blue a negative correlation, and bolded PC loadings (>|0.4|). (B) Composite PC scores for each monkey at each time point. A significant main effect of time was found [F(2,27) = 4.396, partial η 2 = 0.246, 1 − β = 0.709, p = 0.022], with the acute time point showing a significant decrease compared to pre-injury (p < 0.05), and recovering towards the chronic phase that was not significantly different from pre-injury. (C) Two-way GLM ANOVA showed no significant time by task interaction [F(4,27) = 0.443, partial η 2 = 0.062, 1 − β = 0.137, p = 0.777]. Histograms are plotted as mean ± standard error. *p < 0.05. NS, not significant. Frontiers in Rehabilitation Sciences 09 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 were more closely linked to the abnormal levels of coordination from newly formed functional connections, particularly if there among the motor pools rather than the inability to recruit the was a proliferation of aberrant supraspinal and/or spinal motor pools. In most cases the motor pools were much more active networks that directly or indirectly project to cervical spinal post-lesion compared to pre-lesion in each of the motor tasks motor pools that are not normally functionally connected. As studied. Although some muscles continued to be activated at functional reorganization of connectivity also could explain the relatively low levels during some of the tasks, this particular factor present observations reflecting changes in the reorganization of alone did not seem to be a major limitation in task performance. the neural control for stepping, as well as for tasks requiring fine An important implication of these changes in coordination among motor control of the hand. There were no consistent differences the motor pools combined with the elevated levels of activity was in the changes of aberrant connectivity observed either during that there appears to be a significant neural reorganization of locomotion, retrieving a grape, or pulling on a handle, three functional connectivity at one or more points along the tasks differing in the degree of automaticity required to perform sensorimotor axis extending from the sensorimotor cortex to the the tasks, i.e., fine motor control of the more distal muscles vs. spinal motor pools as a result of a partial spinal cord injury (22, 23). stepping on a moving treadmill belt. These data are consistent This elevated activation level of motor pools in response to with there being a proliferation aberrant connectivity formed supraspinal and sensory input are consistent with the observation within or among supraspinal and/or spinal networks. that motor-evoked potentials induced with transcranial Other observations are consistent that a loss of precision in the simulation in human subjects after a partial cervical spinal cord temporal modulation and levels of activation of different motor injury continued to increase in amplitude when generating up to pools contributed to the formation of functionally aberrant 50% maximum voluntary contraction whereas the amplitude connectivity within the spinal networks. For example, after a plateaued at about 10% of maximum voluntary contraction for complete mid-thoracic spinal cord transection in adult rats and control subjects (7, 24). The increased muscle activation after the cats, there is a significant increase in the amount of co-activation partial spinal cord injury observed in the present study also is between flexors and extensors during stepping (16, 27–30). After consistent with the observation that although spontaneous cage weeks of step training of these animals, however, the incidence of activity was relatively low after a spinal cord injury in rats, co-contraction among agonists and antagonists can be reduced increasing the excitability level of the lumbosacral spinal cord significantly. Thus, the spinal networks controlling stepping can networks to ∼20% below motor threshold via epidural function with near-normal coordination patterns without any stimulation increased cage activity five-fold (25). These supraspinal control after a complete spinal cord transection, observations, combined with the present data, are consistent with whereas there is clearly a greater dependence on supraspinal the resting level of excitability of the sensorimotor networks control when performing the GOAS and handle pull tasks after a being relatively low after paralysis (14, 26), but once the motor spinal cord hemisection. The similarity in the degree of threshold is reached the aberrant connectivity can lead to greater functionally elevated aberrant connections during each motor excitatory input to the motor pools. task performed in the present study leaves the possibility that the abnormality of the connectivity observed could be attributed Recovery remains continuously dynamic for largely to spinal networks, particularly during locomotor tasks. months after a spinal cord injury Given the multiple examples of cortical and brainstem In the acute stages after a spinal cord injury there is a period of reorganization reflecting the formation of aberrant connectivity days to weeks in which the excitatory state of the spinal circuitry and elevated neuronal activity after an injury, however, it seems below the lesion is relatively quiescent electrically, a fundamental likely that the supraspinal networks also played some role in property of “spinal shock”. In the present study, over a period of shaping the motor recovery observed in the present study (31–33). two months there was a gradual and substantial recovery in the ability to activate groups of muscles resulting in the generation What Are the structural mechanisms associated of considerable forces. The activation levels of the muscles were with the continuously adapting functional routinely higher post-lesion than pre-lesion. What are the neural connectivity mechanisms that could underlie this elevated activation of motor What could be the mechanisms for the changing functional pools post-injury? From a fundamental physiological perspective connectivity among neurons? Commonly used terms and this must involve activation of more motor units and/or concepts that are presumed to be the basis of neural activation of motor units at a higher frequency within each burst reorganization include “collateral sprouting” of uninjured axons of activity. We suggest that the elevated EMG can most logically and dendrites and “axonal re-growth” of injured neurons, each of be explained by the greater excitability derived from more motor which could occur concomitant with a proliferation of synaptic units being activated and not more action potentials generated inputs along the sensorimotor axis at one or more sites (34). per motor unit. Other possibilities could be less anatomically based, such as rapid An unusually high level of functional connectivity among changes in the efficacy of synaptic connections. Although there interneurons that directly or indirectly project to motoneurons are multiple mechanisms that might underlie the observed could explain, at least in part, the observed increase in motor changes post-injury, we cannot attribute the changes in unit activity (8). The higher level of co-activation between functional connectivity to a specific anatomical mechanism. agonists and antagonists also would be a predictable outcome Based on the present data we propose that a common Frontiers in Rehabilitation Sciences 10 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 mechanism for elevated neuromuscular activity of the spinal within a few weeks after the initiation of epidural stimulation circuitry when performing a motor task after a spinal cord injury below the lesion in four human subjects with chronic motor is attributable to the relatively widespread development of complete paralysis, some neural connectivity might emerge functionally novel, aberrant connections as well as some that initially from functionally incompetent neurons (34). While become more functionally useful via activity-dependent actual growth of axons in a more macro-perspective to new mechanisms. For example, in subjects with a central cord targets is theoretically possible, the speed at which these changes syndrome, a cervical spinal cord injury that affects the upper occurred in these paralyzed subjects and the spread of the limbs more than the lower limbs, clear examples of the loss of activation patterns from a unilateral to a bilateral one as reported specificity in being able to voluntarily activate only the intended by Alexeeva et al. (4) makes this possibility seem less likely. muscles in response to transcranial magnetic stimulation was Similar phenomena of forming aberrant projections at the observed as early as three days post-injury (4). More specifically supraspinal level have been reported (31). when the subjects were asked to activate the left adductor pollicis We do know from studies of recovery of locomotion after a brevis, there was simultaneous EMG activity in the left extensor complete spinal cord injury that as the animal improves in its carpi radialis and right abductor hallucis muscles followed by the ability to step there is a concomitant improvement in the pattern left abductor hallucis, right quadriceps, and soleus muscles. of reciprocity of activation of agonist and antagonist muscles (16, These types of nonspecific activation are rarely observed in able- 28, 40). A decrease in the incidence of co-activation over time bodied subjects. As stated by these authors, these results occurred in the present study in most muscles over a 24-week demonstrate a “diffuse excitation of nonspecific populations of recovery period. Given that the pattern and timing of activation interneurons and motoneurons”. of multiple motor pools involved in the movement are The present data demonstrate that although initially there was determined largely by the interneurons, this gradual reduction in an extensive development of newly formed connectivity there also co-contraction seems likely to be a consequence of some type of was a continuous reshaping of a less diffuse connectivity over time. functional “pruning” of synaptic connectivity among interneurons In some cases nearly normal functionally coordinated patterns and their eventual projections to motoneurons (41, 42). returned by 24 weeks post-lesion, although to some degree the The present data provides a comprehensive ensemble of functionally aberrant connectivity persisted among most motor observations which formulates a systems level biological concept pools. We propose that aberrant functional connections which reflects a highly integrated plastic response to a severe contributed to the loss of specificity in the ability to activate the neuromuscular injury. The nature of this response has significant appropriate combination of motor pools and thus the poor implications for the rehabilitative strategies that we embrace coordination of motor pools when attempting to perform a following such injuries. First, the potential for recovery of specific task. We also have described this loss of specificity in function given the apparent level of plasticity, particularly in the chronic spinal cord injured patients. For example, when neuromuscular systems are clearly evident. Multiple examples instructed to move a specific joint there was a simultaneous supporting this general concept are evident in Figure 1. For activation of flexors and extensors across multiple joints within a example in almost all cases, the amplitude of the EMG during limb and even in both limbs (9). locomotion is significantly higher beginning around six weeks Although it is clear that some of this adaptation in circuitry post-lesion and this elevation persist in almost all muscles in could be facilitated directly to individual motor pools, it is highly each of the subjects. In addition, during this 6 to 24 week some likely that mediation via interneurons will be effective in muscles shows a gradual decline, but remains higher than the establishing more normal patterns among motor pools. This pre-lesion state. For example, the amplitude of the FDS and EDC interpretation is consistent with the multiple experiments in subject one declines between after 20 and 24 weeks. In extensively reviewed by (35) and to experiments involving, full addition, the temporal pattern of the EMG changed markedly in load-bearing stepping over a range of speeds, directions of the EDC as well as the time of the peak EMG amplitude, stepping based on body orientation to the treadmill belt, demonstrating a change in how a specific muscle can be including the reversal of the treadmill belt in animals that have a coordinated which probability of change in the mechanics of the complete mid-thoracic spinal cord transection (15, 36, 37). In movement unless there is another muscle that changes in a these latter experimental preparations, the only source of control manner that compensates for the changes in another motor pool. had to be derived from the total ensembles of proprioception Further important implications of these variations in patterns of derived from a multitude of mechanoreceptors from the EMG during locomotion is that many combinations of activation neuromuscular system of the hindquarters of the animal. patterns of the motor pools can be recruited with little change in Results from studies by Bizzi and colleagues (38) and the mechanics of the movement, i.e., there are many ways for the Lacquaniti and Ivenenko (39), as well as from our lab (28) have nervous system to generate different patterns of activation of demonstrated the dominate role that the spinal circuitry can play multiple motor pools in the uninjured as well as in the injured in controlling locomotion without any influence from state. This is a common observation when changes the supraspinal control. Thus, the key question remains: “what is the mechanics of stepping in response to different acute and chronic anatomical-biochemical substrate of these functionally aberrant, perturbations, such as simply injuring the foot or as occurs as well as useful, connections that seem to emerge so quickly?” during normal aging. The importance of this concept in Given our observations of the recovery of voluntary movements rehabilitation is that when some of these options are eliminated Frontiers in Rehabilitation Sciences 11 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 due to some pathology or trauma, there remains alternate patterns significant connectivity that leads to better coordination is that can achieve virtually the same or at least a similar outcome and derived from ensembles of proprioceptive input to the spinal thus should be taken advantage when possible. interneurons that play a major role in defining the patterns of The concept noted above is also relevant from a biochemical activation among multiple motor pools. perspective in that the performance of a given motor task reflects a functional change in network by changing the activities of selected chemoreceptors, such as serotonergic subtypes that Data availability statement respond to a specific presynaptic transmitter. The classic example, is the monosynaptic circuit in which there is not only The original contributions presented in the study are included monosynaptic activation of motor neurons but simultaneously in the article/Supplementary Material, further inquiries can be there are interneurons that inhibit antagonistic motor neurons. It directed to the corresponding author. has been demonstrated in many cases for example improve stepping via disinhibition. A dramatic demonstration of this phenomenon was to transform the physiological states of the spinal circuitry in an animal that cannot step following a Ethics statement complete transaction regained full weight bearing stepping within 30 min following administration of a pharmacological The animal study was reviewed and approved by Institutional disinhibitor such as strychnine (16, 43). The mechanism for the Animal Care and Use Committee (IACUC) of the California spinal networks to achieve this inhibitory effect normally is via National Primate Research Center (CNPRC) at the University of glycinergic or gabaergic neuro transmission. This type of California, Davis. circuitry is a key component of interneurons of central pattern generation. But they play important roles in defining the temporal patterns of activation of most motor pools. This Author contributions temporal pattern of activation of motor pools defines the pattern that generates every motor behavior (34, 43, 44, 45). In GW, data analysis and interpretation, edited manuscript; SZ, rehabilitation one of the key objectives in many cases is to data recording and analysis generation of figures and tables; HZ facilitate adaptations toward a more functional balance of edited manuscript, surgical procedures; JLN, study design, inhibition and excitation of agonists and antagonist motor pools edited manuscript, data interpretation; ARF study design, data and thus the muscle which they innervate. analysis, interpreting data edited manuscript; SCS, RM, SH, study design, presentation of interventions and daily care of Clinical implications subjects, edited manuscript, data interpretation; YSN-L, edited The present observations following a partial spinal cord injury manuscript, data interpretation; ESR, study design, data demonstrate (1) initially, a marked loss of the ability to activate the analysis, surgery, interpreting data, edited manuscript; MSB, appropriate muscles, (2) the subsequent formation of excessive and JCB, study design, data analysis, interpreting data, edited aberrant inter-neuronal connections within the spinal and/or manuscript; MHT, study design, data analysis, interpreted supraspinal circuitry, and (3) finally a gradual reduction, i.e., data, edited manuscript; RRR, surgery, data collection and “pruning” of these aberrant connections. Understanding the analysis, edited manuscript, data interpretation; VRE, study systems-level mechanisms that could generate this kind and level design, data collection and analysis, edited manuscript, data of spontaneous motor recovery after a cervical spinal cord lesion interpretation; MHT and VRE contributed equally to this could be critical in the development of interventions that can work and share senior authorship. All authors contributed to facilitate recovery of function. More specifically, the question the article and approved the submitted version. becomes “to what extent can the early post-injury quiescent phase be followed by the subsequent formation of more functional but less aberrant connections, thus, maximizing recovery by enhancing recruitment levels of motor pools, while Funding also sustaining more normal coordination among motor pools”? Previous studies of the recovery of stepping after a complete This study was supported by the National Institutes of Health mid-thoracic spinal cord transection suggest that improvements (NS42291), the Veterans Administration, and the California in the coordination of motor pools can be attributable to training Roman-Reed Funds. (37, 41, 46–48). Accordingly, we propose that activity-dependent- learning mechanisms, as occurs during training, will facilitate pruning and functionally selective guidance of supraspinal and Acknowledgments spinal network connectivity leading to more effective coordination among motor pools (49). We further suggest that a The authors thank Steve Guzman for his assistance in data major source of guidance in generating more functionally analysis and interpretation. Frontiers in Rehabilitation Sciences 12 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 Conflict of interest their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be The authors declare that the research was conducted evaluated in this article, or claim that may be made in the absence of any commercial or financial by its manufacturer, is not guaranteed or endorsed by the relationships that could be construed as a potential conflict of publisher. interest. Supplementary material Publisher’s note The Supplementary Material for this article can be found All claims expressed in this article are solely those of online at: https://www.frontiersin.org/articles/10.3389/fresc.2023. the authors and do not necessarily represent those of 1205456/full#supplementary-material References 1. Nout YS, Ferguson AR, Strand SC, Moseanko R, Hawbecker S, Zdunowski S, et al. 17. Linting M, Meulman JJ, Groenen PJ, van der Koojj AJ. Nonlinear principal Methods for functional assessment after C7 spinal cord hemisection in the rhesus components analysis:introduction and application. Psychol Method. (2007) 12 monkey. Neurorehabil Neural Repair. (2012) 26(6):556–69. doi: 10.1177/ (3):336–58. doi: 10.1037/1082-989X.12.3.336 1545968311421934 18. Ferguson AR, Irvine KA, Gensel JC, Nielson JL, Lin A, Ly J, et al. Derivation of 2. Nout YS, Rosenzweig ES, Brock JH, Strand SC, Moseanko R, Hawbecker S, et al. multivariate syndromic outcome metrics for consistent testing across multiple models Animal models of neurologic disorders: a nonhuman primate model of spinal cord of cervical spinal cord injury in rats. PLos One. (2013) 8(3):e59712. doi: 10.1371/ injury. Neurotherapeutics. (2012) 9(2):380–92. doi: 10.1007/s13311-012-0114-0 journal.pone.0059712 3. Rosenzweig ES, Courtine G, Jindrich DL, Brock JH, Ferguson AR, Strand SC, et al. 19. Kaiser HF. The application of electronic computers to fac tor analysis. Educ Extensive spontaneous plasticity of corticospinal projections after primate spinal cord Psychol Meas. (1960) 20:141–51. doi: 10.1177/001316446002000116 injury. Nat Neurosci. (2010) 13(12):1505–10. doi: 10.1038/nn.2691 20. Cattell RB. The scree test for the number of factors. Multivar Behav Res. (1966) 4. Alexeeva N, Broton JG, Suys S, Calancie B. Central cord syndrome of cervical 1:245–76. doi: 10.1207/s15327906mbr0102_10 spinal cord injury: widespread changes in muscle recruitment studied by voluntary 21. Guadagnoli E, Velicer WF. Ralation of sample size to the stability of contractions and transcranial magnetic stimulation. Exp Neurol. (1997) 148 component patterns. Psychol Bull. (1988) 103:265–75. doi: 10.1037/0033-2909.103.2. (2):399–406. doi: 10.1006/exnr.1997.6689 265 5. Bunday KL, Perez MA. Impaired crossed facilitation of the corticospinal pathway 22. Garcia-Alias G, Truong K, Shah PK, Roy RR, Edgerton VR. Plasticity of after cervical spinal cord injury. J Neurophysiol. (2012) 107(10):2901–11. doi: 10.1152/ subcortical pathways promote recovery of skilled hand function in rats after jn.00850.2011 corticospinal and rubrospinal tract injuries. Exp Neurol. (2015) 266:112–9. doi: 10. 6. Bunday KL, Oudega M, Perez MA. Aberrant crossed corticospinal facilitation in 1016/j.expneurol.2015.01.009 muscles distant from a spinal cord injury. PLoS One. (2013) 8(10):e76747. doi: 10. 23. Taccola G, Sayenko D, Gad P, Gerasimenko Y, Edgerton VR. And yet it moves: 1371/journal.pone.0076747 recovery of volitional control after spinal cord injury. Prog Neurobio. (2018) 7. Davey NJ, Smith HC, Savic G, Maskill DW, Ellaway PH, Frankel HL. Comparison 160:64–81. doi: 10.1016/j.pneurobio.2017.10.004 of input-output patterns in the corticospinal system of normal subjects and 24. Jutzeler CR, Streijger F, Aguilar J, Shortt K, Manouchehri N, Okon E, et al. incomplete spinal cord injured patients. Exp Brain Res. (1999) 127(4):382–90. Sensorimotor plasticity after spinal cord injury: a longitudinal and translational doi: 10.1007/s002210050806 study. Ann Clin Transl Neurol. (2019) 6(1):68–82. doi: 10.1002/acn3.679 8. Fenrich KK, Rose PK. Spinal interneuron axons spontaneously regenerate after 25. Gad P, Choe J, Shah P, Garcia-Alias G, Rath M, Gerasimenko Y, et al. Sub- spinal cord injury in the adult feline. J Neurosci. (2009) 29(39):12145–58. doi: 10. threshold spinal cord stimulation facilitates spontaneous motor activity in spinal 1523/JNEUROSCI.0897-09.2009 rats. J Neuroeng Rehabil. (2013) 10:108. doi: 10.1186/1743-0003-10-108 9. Maegele M, Müller S, Wernig A, Edgerton VR, Harkema SJ. Recruitment of spinal 26. Alaimo MA, Smith JL, Roy RR, Edgerton VR. EMG Activity of slow and fast motor pools during voluntary movements versus stepping after human spinal cord ankle extensors following spinal cord transection. J Appl Physiol. (1984) 56 injury. J Neurotrauma. (2002) 19(10):1217–29. doi: 10.1089/08977150260338010 (6):1608–13. doi: 10.1152/jappl.1984.56.6.1608 10. Pierrot-Deseilligny E. Transmission of the cortical command for human 27. Courtine G, Gerasimenko YP, van den Brand R, Yew A, Musienko P, Zhong H, voluntary movement through cervical propriospinal premotoneurons. Prog et al. Transformation of nonfunctional spinal circuits into functional and adaptive Neurobiol. (1996) 48(4–5):489–517. doi: 10.1016/0301-0082(96)00002-0 states after the loss of brain input. Nature Neurosci. (2009) 12(10):1333–42. doi: 10. 11. Wang Z, Dai Z, Gong G, Zhou C, He Y. Understanding structural-functional 1038/nn.2401 relationships in the human brain: a large-scale network perspective. Neuroscientist. 28. de Leon RD, Hodgson JA, Roy RR, Edgerton VR. Locomotor capacity (2015) 21(3):290–305. doi: 10.1177/1073858414537560 attributable to step training versus spontaneous recovery following spinalization in 12. Merzenich MM, Jenkins WM. Reorganization of cortical representations of adult cats. J Neurophysiol. (1998) 79(3):1329–40. doi: 10.1152/jn.1998.79.3.1329 the hand following alterations of skin inputs induced by nerve injury, skin island 29. Edgerton VR, Courtine G, Gerasimenko YP, Lavrov I, Ichiyama RM, Fong AJ, transfers, and experience. J Hand Ther. (1993) 6(2):89–104. doi: 10.1016/S0894- et al. Training locomotor networks. Brain Res Rev. (2008) 57(1):241–54. doi: 10. 1130(12)80290-0 1016/j.brainresrev.2007.09.002 13. Merzenich MM, Nelson RJ, Stryker MP, Cynader MS, Schoppmann A, Zook JM. Somatosensory cortical map changes following digit amputation in adult monkeys. 30. Edgerton VR, de Guzman CP, Gregor RJ, Roy RR, Hodgson JA, Lovely RG. J Comp Neurol. (1984) 224(4):591–605. doi: 10.1002/cne.902240408 Trainability of the spinal cord to generate hindlimb stepping patterns in adult spinalized cats. In: Shimamura M, Grillner S, Edgerton VR, editors. Neurobiological 14. Harkema S, Gerasimenko Y, Hodes J, Burdick J, Angeli C, Chen Y, et al. Effect of basis of human locomotion. Tokyo: Japan Scientific Societies Press (1991). p. 411–23. epidural stimulation of the lumbosacral spinal cord on voluntary movement, standing, and assisted stepping after motor complete paraplegia: a case study. Lancet. (2011) 377 31. Bronfeld M, Bar-Gad I. Loss of specificity in basal ganglia related movement (9781):1938–47. doi: 10.1016/S0140-6736(11)60547-3 disorders. Front Syst Neurosci. (2011) 5:38. doi: 10.3389/fnsys.2011.00038 15. Gerasimenko YP, Lu DC, Modaber M, Zdunowski S, Gad P, Sayenko DG, et al. 32. Milliken GW, Plautz EJ, Nudo RJ. Distal forelimb representations in primary Noninvasive reactivation of motor descending control after paralysis. J Neurotrauma. motor cortex are redistributed after forelimb restriction: a longitudinal study in (2015) 32(24):1968–80. doi: 10.1089/neu.2015.4008 adult squirrel monkeys. J Neurophysiol. (2013) 109(5):1268–82. doi: 10.1152/jn. 00044.2012 16. de Guzman CP, Roy RR, Hodgson JA, Edgerton VR. Coordination of motor pools controlling the ankle musculature in adult spinal cats during treadmill 33. Oudega M, Perez MA. Corticospinal reorganization after spinal cord injury. walking. Brain Res. (1991) 555(2):202–14. doi: 10.1016/0006-8993(91)90343-T J Physiol. (2012) 590(16):3647–63. doi: 10.1113/jphysiol.2012.233189 Frontiers in Rehabilitation Sciences 13 frontiersin.org Wai et al. 10.3389/fresc.2023.1205456 34. Edgerton VR, Gad P. Spinal automaticity of movement control and its role in spinal cord transection in rats. J Neurosci. (2007) 27(16):4460–71. doi: 10.1523/ recovering function after spinal injury. Expert Rev Neurother. (2022) 22(8):655–67. JNEUROSCI.2302-06.2007 doi: 10.1080/14737175.2022.2115359 43. Chen B, Li Y, Yu B, Zhang Z, Brommer B, Williams PR, et al. Reactivation of 35. Prochazka A, Mushahwar V, Yakovenko S. Activation and coordination of spinal dormant relay pathways in injured spinal cord by KCC2 manipulations. Cell. (2018) motoneuron pools after spinal cord injury. Prog Brain Res. (2002) 137:109–24. doi: 10. 174:521–36. doi: 10.1016/j.cell.2018.06.005 1016/S0079-6123(02)37011-0 44. Grillner S, Manira AE. Current principles of motor control, with special 36. Gerasimenko Y, Roy RR, Edgerton VR. Epidural stimulation: comparison of the reference to vertebrate locomotion. Pysiol Rev. (2020) 100(1):271–320. doi: 10.1152/ spinal circuits that generate and control locomotion in rats, cats and humans. Exp physrev.00015.201910.1152/physrev.00015.2019 Neurol. (2008) 209(2):417–25. doi: 10.1016/j.expneurol.2007.07.015 45. Musienko P, Courtine G, Tibbs JE, Kilimnik Savochin A, Garfinkel A, Roy 37. Rossignol S, Frigon A. Recovery of locomotion after spinal cord injury: some RR, et al. Somatosensory control of balance during locomotion in facts and mechanisms. Annu Rev Neurosci. (2011) 34:413–40. doi: 10.1146/annurev- decerebrated cat. J Neurophysiol. (2012) 107(8):2072–82. doi: 10.1152/jn.00730. neuro-061010-113746 2011 38. Bizzi E, Ajemian R. From motor planning to execution: a sensorimotor loop 46. Joseph MS, Ying Z, Zhuang Y, Zhong H, Wu A, Bhatia H, et al. Effects of diet perspective. J Neurophysiol. (2020) 124(6):1815–23. doi: 10.1152/jn.00715.2019 and/or exercise in enhancing spinal cord sensorimotor learning. PLoS One. (2012) 7 39. Lacquaniti F, Ivanenko YP, Zago M. Patterned control of human locomotion. (7):e41288. doi: 10.1371/journal.pone.0041288 J Physiol. (2012) 590(10):2189–99. doi: 10.1113/jphysiol.2011.215137 47. Shah PK, Garcia-Alias G, Choe J, Gad P, Gerasimenko Y, Tillakaratne N, et al. 40. Tillakaratne NJ, de Leon RD, Hoang TX, Roy RR, Edgerton VR, Tobin AJ. Use- Use of quadrupedal step training to re-engage spinal interneuronal networks and dependent modulation of inhibitory capacity in the feline lumbar spinal cord. improve locomotor function after spinal cord injury. Brain. (2013) 136 J Neurosci. (2002) 22(8):3130–43. doi: 10.1523/JNEUROSCI.22-08-03130.2002 (Pt11):3362–77. doi: 10.1093/brain/awt265 41. Ichiyama RM, Broman J, Roy RR, Zhong H, Edgerton VR, Havton LA. 48. Weaver J. How brain circuits adapt to changes in sensory experience. PLoS Biol. Locomotor training maintains normal inhibitory influence on both alpha- and (2014) 12(2):e1001802. doi: 10.1371/journal.pbio.1001802 gamma-motoneurons after neonatal spinal cord injury. J Neurosci. (2011) 31 49. Urban LS, Thornton MA, Ingraham Dixie KL, Dale EA, Zhong H, Phelps PE, (1):26–33. doi: 10.1523/JNEUROSCI.6433-09.2011 et al. Formation of a novel supraspinal-spinal connectome that relearns the same 42. Petruska JC, Ichiyama RM, Jindrich DL, Crown ED, Tansey KE, Roy RR, et al. motor task after complete paralysis. J Neurophysiol. (2021) 126(3):957–66. doi: 10. Changes in motoneuron properties and synaptic inputs related to step training after 1152/jn.00422.2020 Frontiers in Rehabilitation Sciences 14 frontiersin.org