Physiological and pathological motor unit phenotypes coexist within single motor pools after cervical SCI

Individuals with chronic cervical spinal cord injury (SCI) retain voluntary motor unit (MU) activation below the level of the lesion despite profound impairments in muscle relaxation, yet the MU-level mechanisms underlying this dissociation remain unclear. Here, we investigated MU recruitment and derecruitment dynamics using high-density surface EMG (HD-sEMG) and intramuscular EMG (iEMG), in functionally paralyzed muscles across open- and closed-loop tasks in four participants with chronic cervical SCI. Building on prior work that distinguished task-modulated from non-modulated MUs, we combined closed-loop MU feedback with targeted intramuscular implants to determine whether tonic units remain accessible to voluntary control and whether distinct firing phenotypes coexist within individual motor pools. Participants voluntarily recruited MUs and modulated discharge rates (45.2%), but a significant fraction (54.8%) of active MUs could not be derecruited during attempted relaxation. Quantitative analysis of 409 MUs revealed that derecruitment impairment spans a graded continuum rather than the binary distinction between modulated and non-modulated units reported previously. Three phenotypes, controllable (45.2%), modulated-tonic (42.8%), and tonic (12%), partition this spectrum based on derecruitment timing, discharge regularity, and firing persistence and coexist within the same motor pools, including within single muscle compartments sampled intramuscularly. Kaplan-Meier survival analysis confirmed graded derecruitment dynamics: controllable units ceased firing within 500 ms of rest onset, whereas most tonic units persisted throughout the observation window (log-rank{chi} 2= 164.10, P < 0.001). Tonic units displayed highly regular discharge consistent with intrinsic motoneuron excitation via persistent inward currents (PIC). Phenotypes were consistent across recording modalities and task contexts, and phenotype effects on derecruitment metrics exceeded movement-type effects by an order of magnitude. These findings identify impaired MU derecruitment as a core feature of spastic paralysis, driven by maladaptive motoneuron and spinal network properties that preserved descending drive cannot fully counteract. A proof-of-concept spike-train-level temporal filter selectively suppressed tonic firing while preserving voluntary modulation (>94% correlation retained), demonstrating a physiologically grounded strategy for improving neural interfacing after SCI.