Redundancy and multifunctionality among spinal locomotor networks.

c-Fos is used to identify system-wide neural activation with cellular resolution in vivo. However, c-Fos can only capture neural activation of one event. Targeted recombination in active populations (TRAP) allows the capture of two different c-Fos activation patterns in the same animal. So far, TRAP has only been used to examine brain circuits. This study uses TRAP to investigate spinal circuit activation during resting and stepping, giving novel insights of network activation during these events. The level of colabeled (c-Fos+ and TRAP+) neurons observed after performing two bouts of stepping suggests that there is a probabilistic-like phenomenon that can recruit many combinations of neural populations (synapses) when repetitively generating many step cycles. Between two 30-min bouts of stepping, each consisting of thousands of steps, only ∼20% of the neurons activated from the first bout of stepping were also activated by the second bout. We also show colabeling of interneurons that have been active during stepping and resting. The use of the FosTRAP methodology in the spinal cord provides a new tool to compare the engagement of different populations of spinal interneurons in vivo under different motor tasks or under different conditions.NEW & NOTEWORTHY The results are consistent with there being an extensive amount of redundancy among spinal locomotor circuits. Using the newly developed FosTRAP mouse model, only ∼20% of neurons that were active (labeled by Fos-linked tdTomato expression) during a first bout of 30-min stepping were also labeled for c-Fos during a second bout of stepping. This finding suggests variability of neural networks that enables selection of many combinations of neurons (synapses) when generating each step cycle.

Locomotor Training Increases Synaptic Structure With High NGL-2 Expression After Spinal Cord Hemisection.

BACKGROUND: We previously demonstrated that step training leads to reorganization of neuronal networks in the lumbar spinal cord of rodents after a hemisection (HX) injury and step training, including increases excitability of spinally evoked potentials in hindlimb motor neurons. METHODS: In this study, we investigated changes in RNA expression and synapse number using RNA-Seq and immunohistochemistry of the lumbar spinal cord 23 days after a mid-thoracic HX in rats with and without post-HX step training. RESULTS: Gene Ontology (GO) term clustering demonstrated that expression levels of 36 synapse-related genes were increased in trained compared with nontrained rats. Many synaptic genes were upregulated in trained rats, but Lrrc4 (coding NGL-2) was the most highly expressed in the lumbar spinal cord caudal to the HX lesion. Trained rats also had a higher number of NGL-2/synaptophysin synaptic puncta in the lumbar ventral horn. CONCLUSIONS: Our findings demonstrate clear activity-dependent regulation of synapse-related gene expression post-HX. This effect is consistent with the concept that activity-dependent phenomena can provide a mechanistic drive for epigenetic neuronal group selection in the shaping of the reorganization of synaptic networks to learn the locomotion task being trained after spinal cord injury.

Cell Counting and Segmentation of Immunohistochemical Images in the Spinal Cord: Comparing Deep Learning and Traditional Approaches.

Estimation of cell nuclei in images stained for the c-fos protein using immunohistochemistry (IHC) is infeasible in large image sets. Use of multiple human raters to increase throughput often creates variance in the data analysis. Machine learning techniques for biomedical image analysis have been explored for cell-counting in pathology, but their performance on IHC staining, especially to label activated cells in the spinal cord is unknown. In this study, we evaluate different machine learning techniques to segment and count spinal cord neurons that have been active during stepping. We present a qualitative as well as quantitative comparison of algorithmic performance versus two human raters. Quantitative ratings are presented with cell-count statistics and Dice (DSI) scores. We also show the degree of variability between multiple human raters' segmentations and observe that there is a higher degree of variability in segmentations produced by classic machine learning techniques (SVM and Random forest) as compared to the newer deep learning techniques. The work presented here, represents the first steps towards addressing the analysis time bottleneck of large image data sets generated by c-fos IHC staining techniques, a task that would be impossible to do manually.

Brain-machine interface facilitated neurorehabilitation via spinal stimulation after spinal cord injury: Recent progress and future perspectives.

Restoration of motor function is one of the highest priorities in individuals afflicted with spinal cord injury (SCI). The application of brain-machine interfaces (BMIs) to neuroprostheses provides an innovative approach to treat patients with sensorimotor impairments. A BMI decodes motor intent from cortical signals to control external devices such as a computer cursor or a robotic arm. Recent BMI systems can now use these motor intent signals to directly activate paretic muscles or to modulate the spinal cord in a way that reengage dormant neuromuscular systems below the level of injury. In this perspective, we review the progress made in the development of brain-machine-spinal-cord interfaces (BMSCIs) and highlight their potential for neurorehabilitation after SCI. The advancement and application of these neuroprostheses goes beyond improved motor control. The use of BMSCI may combine repetitive physical training along with intent-driven neuromodulation to promote neurorehabilitation by facilitating activity-dependent plasticity. Strong evidence suggests that proper timing of volitional neuromodulation facilitates long-term potentiation in the neuronal circuits that can promote permanent functional recovery in SCI subjects. However, the effectiveness of these implantable neuroprostheses must take into account the fact that there will be continuous changes in the interface between the signals of intent and the actual trigger to initiate the motor action.