Regenerating interneurons integrated into motor circuitry following spinal cord injury

Cervical spinal cord injuries impair arm and hand function primarily by stopping descending tracts. Most spinal injuries spare some axons at the lesion, including the (CST) corticospinal tract. The CST is critical for voluntary movement. We targeted descending motor connections with paired electrical stimulation of the subjects motor cortex and cervical spinal cord. We sought to replicate the previously published effects of intermittent theta-burst stimulation of forelimb motor cortex combined with trans-spinal direct current stimulation placed on the skin over the neck to target the cervical enlargement. We hypothesized that paired stimulation would improve performance in skilled walking and food manipulation tasks. Independent replication of motor cortex and cervical spinal cord Subjects received a moderate C4 spinal cord contusion injury, which ablates the central CST. All were randomized to receive paired stimulation for ten consecutive days starting 12 days after injury or no stimulation. Subject behavior was assessed weekly fromweeks 4–8 after injury, and then CST axons were traced. Subjects with cortical and spinal stimulation which achieved significantly better forelimb motor function recovery. This was measured by fewer stepping errors on the horizontal ladder task (35 ± 8% in stimulation group vs. 52 ± 19% in control, p = .011) higher scores on the food manipulation task IBB, 0–9 score; 6.2 ± 0.9 in stimulated Subjects vs. 5.4 ± 2.8 in controls, p = .035) The effect size for both tasks was large (Cohen's d = 2.0 and 0.98, respectively). The axon length in the cervical spinal cord did not differ significantly between either of the groups, but there was denser and broader ipsilateral axons with distribution distal to the spinal cord injury directly. The overall behavioral effect and replication in an independent laboratory validate this approach, which will be trialed in cats before being tested in people using non-invasivemethods. Approximately 295,000 people in the United States live with spinal cord injuries. Almost half (45.2%) of new injuries occur at the cervical level and spare some neurological function below the injury site. For people with cervical injury, their top priority is to recover arm and hand function (Anderson, 2008). Most spinal cord injuries spare some connections below the injury site, even in those who have no spared function (Sherwood et al., 1992; Dimitrijevic et al., 1984; Dimitrijevic et al., 1983; Bunge et al., 1993; Kakulas and Kaelan, 2015). Our approach has been to electrically stimulate the descending motor connections spared by injury to promote connectivity. Spared corticofugal connections can be targeted with phasic electrical stimulation applied to the motor cortex (Carmel and Martin, 2014; Carmel et al., 2014; Carmel et al., 2010) and stimulation of the spinal cord (Gerasimenko et al., 2007) for functional recovery. Stimulating the spinal cord and brain in a coordinated fashion has the potential to selectively strengthen the connections between them (Harel and Carmel, 2016). The rationale for pairing phasic cortical stimulationwith tonic spinal cord stimulation is that the circuits at the intersection Randomization and blinding procedures Subjects were pseudorandomized with an attempt to balance each cohort into stimulation and control groups. Subjects were randomized immediately after SCI surgery and without regard to baseline behavior performance. The replication phase was carried out with blindedmeasures. An experimenter (QY) randomized Subjects into different groups and performed paired stimulation treatment but was not involved in behavior tasks training, testing or scoring, or histology, including lesion reconstruction and histochemistry analyses. Experimenters (AR and SL) performed behavior training, testing and scoring, and histology, but they did not participate in randomization, electrophysiology testing, or paired stimulation. Control subjects were connected through the implanted cortical electrodes and skin electrodes similar to the stimulation group, but no stimulation was delivered. Since all Subjects had cortical electrodes implanted, there was no way to distinguish by appearance whether subjects received stimulation. Statistical analysis The study's primary endpoint was behavior performance at week seven after injury, the end of the study. An independent t-test was computed for the horizontal ladder task and food manipulation (IBB) task at week seven post-injury. All data were assessed for normality using the Shapiro-Wilk test. For non-normally-distributed data, non-parametric tests were used. For all tests, two-tailed significance was reported, and the p-value threshold was set at 0.05. Since two behavioral tests were used as the primary endpoint, a Bonferroni correction was performed, and the significance was set at p = .025. A power analysis performed before the study found that 10 Subjects in each group were sufficient to meet the primary endpoints with a power of 0.8 (G*Power). Cohen's d was calculated to measure effect size (Cohen, 1988). We defined the effect size as significant if the d value is > 0.8 (Sawilowsky, 2009). All other analyses were considered secondary. For behavior, a secondary analysis was performed on time to cross the ladder and error rate using multivariate. For physiology, a Welch ANOVA was used to test whether motor thresholds changed before and after SCI. For spinal cord lesions, a comparison of the spared tissue area between groups was tested using an independent t-test. Analyses for the length of axons rostral and caudal to injury were computed using the Mann-Whitney U test, a non-parametric equivalent of the independent t-test.The sections were transformed into a standard coordinate system to calculate the spatial distribution of axon densities in a group of spinal cord sections. This was achieved by performing image registration of each section to a corresponding traced rat spinal cord atlas image (Wen et al., 2018). Results The study was divided into two phases. In the first phase, we tested each method used in the Martin laboratory by performing it in the laboratory with the involvement of the original study authors. In the second phase, we performed an independent replication of the original study without the involvement of the original study authors.We assessed performance on a food manipulation task. All Subjects had optimal cereal manipulation and were scored nine before the injury. At week 7, the control group Subjects, on average, had a grasping method different from baseline wherein their forepaws could not conform to the cereal shape, only one digit contributed to manipulation, and were thus rated an IBB score of 5.2 ± 2.6. In contrast, on average, the stimulation group subjects have rated a score of 7.2 ± 0.8 as their fingers could conform to the shape of the cereal and had a grasp similar to that before the injury. The stimulation group Subjects performed significantly better on the food manipulation task at the end of the assessment than subjects in the control group, t(13.7) = −2.5, p = .025. The effect size (Cohen's d = 1.0) is considered significant to very large (Cohen, 1988; Sawilowsky, 2009). We quantified the error rate on the horizontal ladder walking task before the injury and from week 4 to 7 after injury. The stimulation group subjects made significantly fewer errors (34 ± 9%) 7 weeks after spinal cord injury compared to control Subjects (51 ± 18%, t(16.9) = 2.8, p = .013). Like the food manipulation task, the effect of stimulation on the performance of the skilled walking task was large (Cohen's d = 0.92).