Key points

Introduction

The mammalian lumbar spinal cord contains the necessary circuitry to drive a wide variety of locomotor behaviors. This spinal capacity for locomotion in the absence of supraspinal influence was well recognized by the early twentieth century. In that period, Sherrington’s reflex chain mechanism was the prevailing theory whereby spinal locomotion is generated from serial triggering of reflex actions by peripheral afferent feedback as a consequence of limb movement. Contemporary studies by Thomas Graham Brown found to the contrary that acutely spinalized cats following lumbosacral dorsal rhizotomy exhibited rhythmic stepping behavior. Brown, therefore proposed that the spinal cord is inherently capable of driving rhythmic locomotor behavior in the absence of supraspinal or sensory input. This fundamental concept of a self-contained spinal mechanism for locomotion gave rise to the modern conceptualization of the locomotor central pattern generation (CPG). There is now a growing body of evidence for a locomotor CPG mechanism in the human lumbar spinal cord that is amenable to rehabilitation following a spinal cord injury (SCI). A primary goal of SCI therapeutic development is to identify strategies that maximize the retained locomotor potential of the spinal locomotor circuits following injury.

Current therapeutic strategies aim to (1) increase baseline excitability of spinal interneuronal circuits underlying CPG function and (2) optimize sensorimotor integration processes governing locomotor activity ( Fig. 1 ). Several approaches have been explored both individually and in combination to accomplish this. The most widely studied approaches are based on repetitive task-specific training and pharmacologic neuromodulatory agents. Recently, epidural stimulation (EDS) was used in conjunction with locomotor training on a patient classified as ASIA B (American Spinal Injury Association classification). This patient regained full weight-bearing standing and was able to execute voluntary controlled movement in his lower extremity with EDS. Similar findings of voluntary-controlled locomotion was seen in rats treated with a combination of EDS, locomotor training, and pharmacologic modulatory agents after a complete thoracic spinal transection and after a double hemisection at different spinal levels. Together, these studies suggest that EDS can effectively elevate the functional state of the spinal locomotor circuitry, thereby facilitating sensory-mediated locomotion and voluntary movements via spared supraspinal fibers. In this article, the physiologic basis of EDS and future directions of EDS-based spinal neuromodulation is discussed.

This illustration provides a global perspective on the sources of neural control of posture and locomotion, which normally includes the exchange of information between the brain and the spinal cord and between the sensory receptors within the muscles, joints, and skin and the spinal cord. This article emphasizes the importance of the afferent information from the periphery as a source of control of posture or locomotor tasks, in conjunction with the spinal circuitry to which CPG is routinely attributed, when there can be no exchange of information between the brain and spinal cord (ie, following a complete SCI). Given that it is possible to generate locomotor movements with a tonic stimulation of the mesencephalic locomotor area, details of the control of posture and locomotion are not likely to be derived from a tonic signal from the brainstem. However, this tonic stimulation can be an important source of modulatory control of the spinal circuitry, as can be epidural stimulation, afferent stimulation, and pharmacologic modulation. Factors listed on the right side of the figure point out details of the physical environment that can be detected by the spinal cord and used to instruct the spinal circuitry to activate the appropriate motor pools at the appropriate time. The sensory perception of the factors listed in the box on the right side can play an important role in shaping the physiology of the spinal circuitry through repetitive activity (ie, training). Further evidence of the importance of the sensory information is demonstrated by the acute effects of unilateral deafferentation after which epidural stimulation will induce locomotor movements only on the intact side.

Introduction

The mammalian lumbar spinal cord contains the necessary circuitry to drive a wide variety of locomotor behaviors. This spinal capacity for locomotion in the absence of supraspinal influence was well recognized by the early twentieth century. In that period, Sherrington’s reflex chain mechanism was the prevailing theory whereby spinal locomotion is generated from serial triggering of reflex actions by peripheral afferent feedback as a consequence of limb movement. Contemporary studies by Thomas Graham Brown found to the contrary that acutely spinalized cats following lumbosacral dorsal rhizotomy exhibited rhythmic stepping behavior. Brown, therefore proposed that the spinal cord is inherently capable of driving rhythmic locomotor behavior in the absence of supraspinal or sensory input. This fundamental concept of a self-contained spinal mechanism for locomotion gave rise to the modern conceptualization of the locomotor central pattern generation (CPG). There is now a growing body of evidence for a locomotor CPG mechanism in the human lumbar spinal cord that is amenable to rehabilitation following a spinal cord injury (SCI). A primary goal of SCI therapeutic development is to identify strategies that maximize the retained locomotor potential of the spinal locomotor circuits following injury.

Current therapeutic strategies aim to (1) increase baseline excitability of spinal interneuronal circuits underlying CPG function and (2) optimize sensorimotor integration processes governing locomotor activity ( Fig. 1 ). Several approaches have been explored both individually and in combination to accomplish this. The most widely studied approaches are based on repetitive task-specific training and pharmacologic neuromodulatory agents. Recently, epidural stimulation (EDS) was used in conjunction with locomotor training on a patient classified as ASIA B (American Spinal Injury Association classification). This patient regained full weight-bearing standing and was able to execute voluntary controlled movement in his lower extremity with EDS. Similar findings of voluntary-controlled locomotion was seen in rats treated with a combination of EDS, locomotor training, and pharmacologic modulatory agents after a complete thoracic spinal transection and after a double hemisection at different spinal levels. Together, these studies suggest that EDS can effectively elevate the functional state of the spinal locomotor circuitry, thereby facilitating sensory-mediated locomotion and voluntary movements via spared supraspinal fibers. In this article, the physiologic basis of EDS and future directions of EDS-based spinal neuromodulation is discussed.

This illustration provides a global perspective on the sources of neural control of posture and locomotion, which normally includes the exchange of information between the brain and the spinal cord and between the sensory receptors within the muscles, joints, and skin and the spinal cord. This article emphasizes the importance of the afferent information from the periphery as a source of control of posture or locomotor tasks, in conjunction with the spinal circuitry to which CPG is routinely attributed, when there can be no exchange of information between the brain and spinal cord (ie, following a complete SCI). Given that it is possible to generate locomotor movements with a tonic stimulation of the mesencephalic locomotor area, details of the control of posture and locomotion are not likely to be derived from a tonic signal from the brainstem. However, this tonic stimulation can be an important source of modulatory control of the spinal circuitry, as can be epidural stimulation, afferent stimulation, and pharmacologic modulation. Factors listed on the right side of the figure point out details of the physical environment that can be detected by the spinal cord and used to instruct the spinal circuitry to activate the appropriate motor pools at the appropriate time. The sensory perception of the factors listed in the box on the right side can play an important role in shaping the physiology of the spinal circuitry through repetitive activity (ie, training). Further evidence of the importance of the sensory information is demonstrated by the acute effects of unilateral deafferentation after which epidural stimulation will induce locomotor movements only on the intact side.

Spinal sensorimotor control of locomotor behavior

The spinal locomotor circuit normally operates under the direction of supraspinal input and sensory cues. Supraspinal areas, including the motor cortex, red nucleus, vestibular nuclei, reticular nuclei, and the mesencephalic locomotor area (MLR), are known to modulate spinal locomotor circuits. In concert, spinal locomotor circuits are receiving a constant inflow of information from sensory receptors that gauge the environment and proprioceptive receptors that transduce both the limb configuration and movement state. One fundamental concept is that even in a normal spinal cord, rhythmic locomotor behaviors, although initiated by supraspinal input, are largely a product of spinal interneuronal circuits (ie, CPG) but shaped by sensory information (see Fig. 1 ).

The underlying sensorimotor integration processes governing spinal locomotor behavior are highly sophisticated. This sophistication is demonstrated by the wide varieties of locomotor behaviors, such as trotting, galloping, backwards and sideways walking, and obstacle avoidance, that can be executed by spinal animals with sensory cues alone. The constant influx of sensory information must be interpreted and responded to in a contextually appropriate manner. For example, studies on stumbling responses in chronic spinal cats demonstrate that electrical stimulation of cutaneous receptors of the dorsum paw surface during the flexion phase enhanced ongoing flexion, a response that is appropriate of obstacle avoidance. If this same stimulus elicits a flexion response when the paw is bearing weight, it would result in a fall. If the paw was in the stance phase and bearing weight, that same stimuli that enhanced flexion during the swing phase will instead enhance the ongoing extension. The mechanism underlying this locomotor phase-dependent selection of motor responses has been partly examined through intracellular recordings, demonstrating phasic modulation of cutaneous input postsynaptic potential in acute spinal cats, chronic spinal cats, and in decerebrate preparations. Results from these studies suggest that the transmission of cutaneous information through oligosynaptic pathways strongly influences the spinal locomotor circuitry.

Therefore, the CPG’s responses to sensory information are not mere fixed reactions. They are highly contingent on the behavioral state (ie, rest vs walking), the task being performed (eg, stand vs walking), and the phase of the task being executed (eg, extension vs flexion phase of walking). These characteristics argue that the spinal locomotor circuits operate with similarities to engineered intelligent control systems. In fact, even simple reflex responses in spinalized animals are modifiable through an instrumental learning paradigm. Furthermore, noncontingent stimuli can actually block future attempts at instrumental learning through a central sensitization mechanism. Therapeutic development for SCI should, therefore, take advantage of these features to promote functional recovery.

Rehabilitating the injured spinal cord

Two general strategies, based on leveraging the retained spinal locomotor ability following SCI, have been widely studied. These strategies are repetitive task-specific training and pharmacologic modulation. The basis of repetitive task-specific therapy is to optimize spinal locomotor performance through use-dependent plasticity. Initial studies showed that with repetitive task-based training alone, spinal cats can regain weight-bearing stepping or standing ability comparable with normal cats. Although there is a decline in locomotor performance following training cessation, repeated training leads to faster gains than with initial training, suggesting a use-dependent plasticity process. This process is likely mediated by axonal sprouting, alterations in the strength of afferent pathways, and changes in the level of expression of neurotrophic factors and receptors brought on by locomotor training. Functional improvements are largely limited to the task practiced in training. For example, stand training in spinal cats results in the recovery of weight-bearing standing ability; however, this does not translate into improvement of weight bearing during stepping. In fact, stand-trained spinal cats step more poorly, suggesting that task-specific training can negatively affect the acquisition of other behaviors. Task-specific locomotor training in the form of overground and partial bodyweight support locomotor training have been studied in human patients with SCI. Although human patients with SCI gain improvement in walking speed and maximal walking distance with training, their recovery does not match the results seen in spinal cats. Further, improvement in locomotor function in response to locomotor training is well documented in individuals that have had an incomplete spinal injury for more than 1 year, but full weight-bearing locomotion has not yet been achieved with training after a clinically complete injury.

A variety of pharmacologic agents have been investigated for the purpose of promoting or enhancing locomotor activity. These agents include drugs that target catecholamine receptors (eg, l -dopa, dopamine, and clonidine) and serotonin receptors (eg, quipazine). Some agents, such as l -dopa and clonidine, can illicit locomotor activity from a nonlocomotor state, whereas agents affecting the serotonergic system result in the enhancement of muscle activation during locomotion. These agents may partially mimic the cellular actions of supraspinal tracts on spinal locomotor circuits and can modulate different aspects of locomotor activity following spinal transection in animal studies. The differential enhancement in the features of spinal locomotion of agents acting on the serotonergic, dopaminergic, or noradrenergic pathways were found to be additive in combination therapy. A potential clinical implication is that pharmacologic treatments may be tailored to the specific locomotor deficits in patients. To date, studies on drug therapy alone in human patients with SCI found only slight improvement in locomotor performance, less than the functional gains made with locomotor training alone. It remains to be determined if further systematic evaluation of combination pharmacologic therapy alone will demonstrate greater improvements in human patients with SCI.

Electrical stimulation of spinal locomotor circuits

Spinal cord stimulation aims to activate and recruit intrinsic spinal mechanisms driving multi-joint motor actions. Two main approaches have been used to deliver electrical stimulation directly to the spinal cord for this purpose, (1) intraspinal electrodes and (2) epidural grid-electrode arrays. Microstimulation of the spinal cord gray matter via intraspinal electrodes is an attractive means to precisely stimulate a small volume of neural tissue. The function of local spinal circuits in movement generation can be closely examined using this method. This strategy was initially applied in frogs, rats, and feline models. A common finding from these studies is that intraspinal stimulation, only at distinct sites, resulted in a small set of unique force patterns. When simultaneous intraspinal stimulation was carried out at 2 separate sites, the resulting force patterns corresponded to the simple summation of force patterns generated if each site was stimulated independently. One hypothesis is that spinal circuits generate complex movement behaviors by superimposing basic movement building blocks. A theoretical approach for the neuromodulation of spinal behavior, based on this hypothesis, is to drive behavior through intraspinal stimulation at multiple sites. Only a small number of studies have shown that intraspinal stimulation can evoke locomotor activity ; however, this required multiple electrodes over multiple levels or site-specific single-electrode stimulation following the administration of clonidine. It remains to be determined if the modularity observed in spinal behavior is a result of modularity in the underlying spinal circuits or a consequence of the complex bidirectional interaction between spinal circuits, musculoskeletal system, and the environment. Nevertheless, spinal neuromodulation via intraspinal stimulation is likely challenging to translate due to the technical issues of placing and securing multiple intraspinal electrodes without significant tissue damage.

Another means to deliver electrical stimuli directly to the spinal cord is with an epidural electrode-grid array. Epidural stimulation has been used for some time for the management of intractable pain. Percutaneous and laminectomy-based placements of epidural electrodes are well established and common practice in neurosurgery. EDS was found to elicit locomotor-like activity in humans and rats with complete spinal cord injuries. Initial studies on epidural stimulation in human patients with SCI aimed to identify the optimal electrode placement site and stimulation parameters to promote locomotor activity. Stimulation was delivered via quadripolar electrodes (Medtronic, Minneapolis, MN) percutaneously placed at the L1-2 level in patients with an ASIA A SCI, more than 1 year from injury. The level of injury ranged from C5 to T8. Patients were examined in a supine position, and the responses to the stimulation of the ventral and dorsal surface were studied. The ventral spinal cord was targeted because projections from the MLR course ventrally. Because stimulation of the MLR in cats can dependably promote locomotor activity, stimulation of the ventral cord may activate areas targeted by MLR projection fibers. However, ventral cord stimulation elicited only tonic muscle contraction in all 8 patients tested, with background rhythmic hip movements in 4 of 8 patients for the entirety of stimulation. The tonic response was likely caused by activation of the anterior-horn motoneuronal pools.

In contrast, dorsal cord stimulation at L2 resulted in rhythmic, step-like electromyogram (EMG) activity associated with leg flexion and extension (nonpatterned 5–9 V, 25–50 Hz). When the electrode was located rostral or caudal from L2, only tonic or irregular rhythmic EMG activity was elicited, but there was no locomotor-like activity. With increasing amplitude of stimulus intensity at L2, EMG progressed from tonic to rhythmic activity, ultimately converting to organized rhythmic locomotor-like movements in both limbs (at 5.5 V, 30 Hz). This progression of activity was highly reproducible among the patients tested. Stimulation frequencies of 20 to 70 Hz were most effective for producing oscillatory movements of the legs. Bilateral locomotor-like activity was produced only with the electrode placed in the midline, whereas unilateral locomotor-like activity was observed when the electrode was offset from the midline. Similar findings were made with epidural stimulation in T10 spinal cats, with stimulation at the L5 level being most effective for producing step-like activity. (Nota Bene The cat has 7 lumbar segments.) These studies demonstrate that EDS is able to invoke locomotor-like activity following complete SCI, further supporting the presence of a locomotor CPG in the human spinal cord. Additionally, activity is highly dependent on the site of stimulation, frequency of stimulation, and intensity of stimulation.