Role of the medullary reticular formation in motor control and functional recovery following spinal cord injury

Spinal cord injury (SCI) interrupts the flow of information between the brain and the spinal cord, thus leading to a loss of sensory information and motor paralysis of the body below the lesion. Surprisingly, most SCIs are incomplete and spare supraspinal pathways, especially those located within the peripheral white matter of the spinal cord, which includes reticulospinal pathways originating from the medullary reticular formation. Whereas there is abundant literature about the motor cortex, its corticospinal pathway, and its capacity to modulate functional recovery after SCI, less is known about the medullary reticular formation and its reticulospinal pathway.

Anatomically, motor circuits of the medullary reticular formation are organized into distinct but not necessarily well-delineated nuclei (Brownstone and Chopek, 2018). Among them, the most ventrolateral nucleus corresponds to the lateral paragigantocellular nucleus, the most dorsomedial one is the gigantocellular reticular nucleus (Gi), the most rostro-ventral one is the α pars of the Gi, and the most caudoventral one is the ventral pars of the Gi (Figure 1A and B). Functionally, if activation of all neurons of the medulla fails to elicit locomotion, targeting optogenetic stimulation of glutamatergic lateral paragigantocellular nucleus neurons induces locomotion and increases speed in mice (Capelli et al., 2017). Furthermore, a subset of glutamatergic neurons expressing the transcription factor Lhx3 and Chx10 (also called V2a neurons) of the Gi are involved in several motor functions (Bretzner and Brownstone, 2013): bilateral optogenetic stimulation of V2a Gi neurons stops ongoing locomotion (Bouvier et al., 2015), while their unilateral stimulation evokes lateral movements (Cregg et al., 2020). Nevertheless, locomotor arrest is not restricted to V2a Gi neurons, as unilateral photostimulation of all glutamatergic Gi neurons can also modulate locomotor patterns in a phase-dependent manner in addition to resetting locomotor rhythm by inducing co-contraction, and slowing down and stopping locomotion (Lemieux and Bretzner, 2019). Taken together, these suggest the existence of several glutamatergic medullary subpopulations with distinct functional roles depending on their medullary location.

Figure 1.

Plasticity of the glutamatergic reticulospinal drive after spinal cord injury.

(A) Schematic illustrating the sagittal and coronal views of the medullary reticular formation in the brainstem. (B) Stimulation sites initiating, accelerating, decelerating, or stopping locomotion upon 1-second train photostimulation of glutamatergic neurons of the medulla. (C) Schematic illustrating hindpaw dragging and loss of body weight support during spontaneous motor recovery after a lateral hemisection at the low thoracic level. (D) Changes in reticulospinal efficacy upon 10 ms pulse photostimulation before and after SCI. (E) Graph analysis illustrating the number of stimulation sites (1 site per mouse) evoking potentiation, depression, absence of changes, or recovery to control levels in hindlimb motor responses over time following SCI. The thickness of the bar illustrates the proportion of change in the number of sites from control to week 1, and from week 1 to week 7. Bar graphs illustrate the total number of stimulation sites at one and seven weeks after SCI. (F) Schematic illustrating stepping ability of chronically impaired SCI mice upon 1-second train photostimulation. (G) Hindpaw dragging before and during 1-second train photostimulation. (H) Schematic illustrating locomotion on the rungs of a horizontal ladder before and after 2-week priming of the reticulospinal drive at rest. (I) Skilled locomotor score as a function of changes in reticulospinal efficacy. Reprinted and adapted with permission from Lemieux et al. (2024). Gi: Gigantocellular reticular nucleus; Giv: ventral pars of the Gi; Giα: α pars of the Gi; LPGi: lateral paragigantocellular nucleus; SCI: spinal cord injury.

After incomplete SCI, tracing studies have reported extensive anatomical reorganization of spinal projections of reticulospinal neurons above and below the lesion that seems to be associated with spontaneous motor recovery (Filli et al., 2014; Zörner et al., 2014; Asboth et al., 2018). Moreover, pharmacogenetic inhibition of reticulospinal neurons of the Gi or glutamatergic neurons of the ventral pars of the Gi prevents spontaneous motor recovery from SCI (Asboth et al., 2018; Engmann et al., 2020), thus suggesting that glutamatergic neurons of the medulla could be necessary for locomotor recovery after SCI.

Combining detailed kinematic and electromyographic recordings with optogenetic tools in freely behaving mice, we investigated the functional contribution of glutamatergic neurons of the medullary reticular formation to control and recovery of locomotion following a lateral thoracic hemisection as an SCI model (Lemieux et al., 2024). Using short-pulse photostimulation delivered throughout the step cycle to probe motor efficacy, we found that the glutamatergic reticulospinal drive evokes phase-dependent motor responses in hindlimb muscles during treadmill locomotion at steady speed (Lemieux and Bretzner, 2019; Lemieux et al., 2024). Interestingly, if glutamatergic reticulospinal neurons project presumably unilaterally in the spinal cord, short pulse photostimulations evoked strong or weak motor responses in hindlimb muscles regardless of their stimulation site across the medulla, thus suggesting a lack of functional left-right lateralization of the reticulospinal command in normal settings.

Despite this, changes in reticulospinal motor efficacy occur over time in motor responses of the ipsilesional hindlimb flexor muscle between one and seven weeks after SCI (Figure 1C–E). Although some sites evoked no change, some exhibited an initial depression followed by a return to pre-SCI levels, whereas others evoked potentiation over time (Figure 1E). Interestingly, seven weeks after SCI, the percentage of changes in reticulospinal efficacy was inversely correlated with pre-SCI responses. In other words, sites evoking the strongest motor responses prior to SCI evoked the weakest changes seven weeks after SCI, whereas sites evoking the weakest responses prior to SCI evoked the strongest changes in reticulospinal efficacy after SCI, thus suggesting an unmasking of potential silent reticulospinal pathways. Regarding the time course of responses (Figure 1E) from the multiple sites stimulated through the medulla, a bit less than half evoked no changes in motor responses one week after SCI, whereas a bit more than a third of sites evoked an initial depression at 1 week post-SCI, which either returned to pre-SCI levels or was potentiated seven weeks after SCI. Finally, a subgroup of sites already evoked a potentiation in motor responses one week after SCI, persisting up to seven weeks after SCI. Changes in such a short time frame cannot be supported by neuroanatomical plasticity, thus arguing for homeostatic plasticity (i.e., physiological mechanisms). From a functional viewpoint, changes in motor responses evoked by the glutamatergic reticulospinal drive were correlated with the locomotor score: The higher the potentiation of the reticulospinal command, the higher the locomotor score of the ipsilesional hindlimb, regardless of the stimulation site. Taken together, these results argue that changes in the glutamatergic reticulospinal drive contribute to spontaneous motor recovery of the ipsilesional hindlimb from SCI.

We next used long-train photostimulation to investigate the functional contribution of these glutamatergic neuronal populations of the medulla to initiating locomotion before and after SCI (Figure 1B). Prior to SCI, we found that sites located in the most dorsal part of the medulla evoked mainly head turning and body turning, as well as a few episodes of locomotion, whereas those located in the most ventral part and especially the most ventrolateral part of the medulla evoked primarily locomotion. Regarding locomotor initiation, changes occurred after chronic SCI: Sites located in the most ventral and ventrolateral sites of the medulla kept their capacity to initiate locomotion, whereas those located in the most dorsal part lost it, regardless of their left-right lateralization.

Concerning the capacity of these neuronal populations to modulate locomotor patterns and rhythm during ongoing locomotion, we identified four groups loosely organized as a continuum of motor functions from the dorsomedial to the ventrolateral regions of the medulla. A first group located in the most dorsal part of the medulla evoked deceleration and stops, suggesting that locomotor arrest may not be specific to V2a Gi neurons (Bouvier et al., 2015); a second and third group evoked a transient deceleration with or without a rebound acceleration in the ventral part of the medulla. The fourth group, located in the ventrolateral part of the medulla, in the vicinity of the lateral paragigantocellular nucleus, evoked acceleration, as previously shown (Capelli et al., 2017). After chronic SCI, sites associated with a transient deceleration lost their function and instead induced locomotor arrests, thus revealing changes in network dynamics. Whether these changes relate to homeostatic plasticity or anatomical reorganization will require further investigation.

Although mice usually recover locomotor functions within a few weeks after a lateral thoracic hemisection (mimicking a Brown-Sequard syndrome), a subgroup of our chronic SCI mice still exhibited strong functional deficits eight weeks after SCI (Figure 1F and G). Surprisingly, long-train photostimulation improved functional recovery of their paw placement by decreasing the frequency of incorrect paw placement and the duration of hindpaw dragging.

As a therapeutical approach, we also investigated whether the plasticity of the reticulospinal drive can improve skilled locomotor coordination on a horizontal ladder (Figure 1H and I). To test this, the motor coordination of the ipsilesional hindlimb of chronic SCI mice was assessed by crossing a horizontal ladder before and after a 2-week priming of the reticulospinal drive. The stimulation protocol consisted of delivering long trains of photostimulation at low intensity to evoke only slight motor twitches in chronic SCI mice at rest for 2 weeks. After priming the reticulospinal command, we found that medullary sites evoking an increase in the locomotor score were the same sites evoking a potentiation of reticulospinal efficacy (with short-pulse photostimulation), whereas those evoking a decrease in the locomotor score, which were located ipsilateral to the lesion site, evoked a depression of reticulospinal efficacy. This finding suggests that sites evoking reticulospinal plasticity could be modulated after chronic SCI to promote skilled locomotor coordination.

In conclusion, we have shown that glutamatergic neurons of the medullary reticular formation contribute to the control and spontaneous recovery of motor functions from SCI. Although the spinal cord, the motor cortex, and the mesencephalic locomotor region can be appealing as neuromodulatory targets, being at the junction between these supraspinal structures and the spinal cord, the medulla could be a promising target for promoting functional motor recovery. Indeed, the medulla is likely crucial in relaying important cortical and/or mesencephalic inputs to spinal locomotor circuits after SCI (Zörner et al., 2014; Asboth et al., 2017; Roussel et al., 2023). Therefore, we plan to further investigate how reticulospinal pathways of the medulla integrate and relay cortical and mesencephalic inputs pertaining to motor control and recovery of locomotion after SCI. Furthermore, we also plan to study how the descending reticulospinal drive translates into a bilateral motor command through local brainstem circuits and/or descending cervical propriospinal interneurons projecting to the lumbar spinal circuit.

Despite a lack of left-right lateralization, our results argue that neuronal populations of the medulla are not necessarily circumscribed to discrete and well-delineated nuclei, but rather constitute a functional continuum through the medulla, ranging from the dorsomedial region promoting locomotor arrest for postural adjustments to the ventrolateral region promoting locomotion (Figure 1B). Therefore, we plan to study this continuum by investigating the intrinsic and extrinsic properties of reticular and reticulospinal tract neurons, as well as their local and remote networks’ connectivity. Ultimately, with the emergence of new genetic tools and identification of new genetically identified neuronal populations from transcriptomic studies, it will be important in the future to generate a new integrative functional, neurophysiological, neuroanatomical, and transcriptomic atlas of the medulla that could improve our understanding of the medullary circuits and their networks in normal settings and neuropathological diseases and neurotraumas affecting gait and posture.

Given their deep location within the brain, neuronal subpopulations of the brainstem can be difficult to access anatomically and physiologically with standard techniques. Optogenetic tools have therefore emerged as a promising alternative to electrical neurostimulation to promote motor recovery in humans with gait and posture disorders. However, several questions must still be resolved prior to optogenetically neurostimulating deep brain structures in individuals. In comparison to the first viruses used in the early trials of gene therapies in the 1990s, we now have access to safe vectors with adeno-associated viruses to deliver therapeutic genes to the brain. Although the long-term expression of light-activated channels remains to be validated, optogenetic tools can—using adequate promoters—target specific neuronal subpopulations according to their neurotransmitters and network connectivity. The development of new optical technologies with new optogenetic tools will also enable biomedical research to photostimulate/probe the neural activity of deep brain regions. Some of these tools in the near-infrared and red wavelengths will allow us to neurostimulate deep brain structures in a non-invasive manner. As suggested by our priming experiments of the reticulospinal drive (Figure 1H and I; Lemieux et al., 2024), another promising approach could be to combine focal ultrasound neuromodulation with functional resonance magnetic imaging to induce plasticity of deep brain regions in individuals at rest during short weekly sessions at the hospital. With the development of lightweight wearable technologies, these focal ultrasound neurostimulations could be combined with rehabilitative training to promote functional recovery after SCI. Nevertheless, prior to any clinical trials, it will be very important to get a better understanding from animal studies of neural dynamics of deep brain regions in physiological and pathological conditions and to assess their potential clinical translation with bioethicists and clinicians.

The author wants to recognize the participation of M. Lemieux (Centre de Recherche du CHU de Québec, Canada) in earlier control experiments and M. Lemieux and N. Karimi (Centre de Recherche du CHU de Québec, Canada) in spinal cord injury experiments.

This work was supported by Craig H. Neilsen Foundation, Wings for Life Foundation, Canadian Institutes of Health Research, and Fonds de Recherche Québec-Santé (to FB).