Probing the Human Spinal Locomotor Circuits by Phasic Step ...klab4recovery.org › wp-content ›...

Send Orders for Reprints to [email protected]

Current Pharmaceutical Design, 2017, 23, 1-16 1 REVIEW ARTICLE

1381-6128/17 $58.00+.00 © 2017 Bentham Science Publishers

Probing the Human Spinal Locomotor Circuits by Phasic Step-Induced Feedback and by Tonic Electrical and Pharmacological Neuromodulation

Ursula S. Hofstoettera, Maria Knikoub, Pierre A. Guertinc and Karen Minassian*a,d

aCenter for Medical Physics and Biomedical Engineering, Medical University of Vienna, Vienna, Austria; bThe Graduate Center, City University of New York, NY, USA; cDepartment of Psychiatry and Neurosciences, Laval University, Québec City, Canada; dCenter for Neuroprosthetics and Brain Mind Institute, School of Life Sciences, Swiss Federal Institute of Technology (EPFL), Lausanne, Switzerland

A R T I C L E H I S T O R Y

Received: November 2, 2016 Accepted: December 6, 2016 DOI: 10.2174/1381612822666161214 144655

Abstract: The mammalian lumbar spinal cord experimentally isolated from supraspinal and afferent feedback input remains capable of expressing some basic locomotor function when appropriately stimulated. This ability has been attributed to spinal neural circuits referred to as central pattern generators (CPGs). In individuals with a severe spinal cord injury, rhythmic activity in paralyzed leg muscles can be generated by phasic proprioceptive feedback during therapist- or robotic-assisted stepping on a motorized treadmill. Here, we critically review to what extent the resulting motor output represents locomotor-like activity, and whether these motor patterns are the result of activation of CPGs, as commonly suggested in the literature. Attempts will be made to further de-lineate the pivotal roles played by mechanisms such as spinal proprioceptive reflexes and their alterations after spinal cord injury, the central excitability level, and by neurotransmitters critical for spinal locomotor activity. We will discuss the view that the muscle activity produced during assisted passive treadmill stepping is resulting from the entrainment of spinal reflex circuits by the cyclically generated proprioceptive feedback. We suggest that the activation of CPG circuits depends rather on the presence of a sustained tonic excitatory drive, as can be provided by electrical spinal cord stimulation, or by specific combinations of dopaminergic agonists, adrener-gic/dopaminergic precursors and/or 5-HT receptor agonists. Novel rehabilitation strategies using spinal cord stimulation and rhythmic-activity producing drugs during locomotor therapy will pave the way for clinically relevant advances in restoration of motor function in people with severe spinal cord injury.

Keywords: Central pattern generator, locomotion, neuromodulation, rehabilitation, spinal cord injury, spinal cord stimulation, spinal reflexes, treadmill stepping.

1. INTRODUCTION Locomotion is probably one of the most complex motor func-tions, normally involving virtually all areas of the central and pe-ripheral nervous system. Under normal conditions, full expression of locomotion requires inputs from cortical areas (visual detection, planning and decision-making), mesencephalic areas (trigger on/off signals), as well as cerebellar and vestibular structures (equilibrium and tonus) [1–3]. The final integration and interpretation of these supraspinal signals takes place in the spinal cord. Indeed, locomo-tion critically depends upon the complex interaction with several neural circuits that are located for most parts within the lumbar area of the spinal cord [4–10]. Monoaminergic and catecholaminergic terminals from descending pathways and their respective receptor targets in the spinal cord as well as spinal glutamatergic, GABAer-gic and glycinergic systems have all been shown to play modula-tory, excitatory, and inhibitory roles in spinal circuitry-driven lo-comotor activity [11,12]. Both, segmental spinal reflexes and dis-tributed CPGs (networks of spinal neurons that can produce rhyth-mic activity autonomously because of single-cell properties and specific connections between cells; [13]) are believed to control locomotor rhythm and pattern generation [8,9]. Peripheral inputs from proprioceptors (muscle spindles, Golgi tendon organs, and joint receptors) provide sensory feedback capable of enhancing, resetting or adapting each step to external factors [14–19]. A severe cervical or thoracic spinal cord injury (SCI) classified clinically as complete (American Spinal Injury Association Impairment Scale A; *Address correspondence to this author at the EPFL SV BMI UP-COURTINE, Station 19, 1015 Lausanne, Switzerland; E-mail: [email protected]

AIS A) or motor complete/sensory incomplete (AIS B) is associated with an integral loss of signaling activity and communication be-tween supraspinal structures and the lumbar spinal circuits. As ex-pected, an immediate and generally irreversible loss of the ability to generate and control locomotion occurs after such SCI. Another consequence associated with SCI is the adaptive transformation, including gene expression and receptor expression level changes, of sublesional structures in the spinal cord [20,21]. Following the iso-lation of the lumbar spinal cord from supraspinal control and the injury-related plasticity [21–24], proprioceptive afferent-input processing [25–27] and rhythmic motor output generating capabili-ties [8,28–31] have been shown to enable, if appropriately stimu-lated, the expression of basic rhythmic leg muscle activity in para-lyzed individuals. One approach used to investigate the ability of the spinal cir-cuits to produce rhythmic activity in individuals suffering of AIS A or AIS B injuries is analysis of the effects associated with assisted and body-weight supported (BWS) stepping on a treadmill. This method involves a motorized treadmill system where partial BWS is provided using a harness to a patient placed in upright position while bilateral alternating leg movements on the moving treadmill belt are manually assisted or executed by therapists, mechanical devices, or robotic exoskeletons [25,32–37]. Another method that can probe the involvement of spinal circuits in the generation of rhythmic leg motor activity in humans is epidural spinal cord stimu-lation (SCS) [28,38–41]. When applied over the lumbar and upper sacral spinal cord (located between the 11th thoracic and 1st lumbar vertebral levels), epidural stimulation has been shown to predomi-nantly activate large-to-medium diameter afferent fibers entering the spinal cord through the posterior roots (dorsal roots) [30,42–46].

2 Current Pharmaceutical Design, 2017, Vol. 23, No. 00 Hofstoetter et al.

Additional neural structures that may be recruited by epidural stimulation are longitudinally running myelinated fiber branches in the most superficial layers of the dorsal and dorsolateral columns of the spinal cord white matter [47,48]. SCS may thus excite groups of afferents that are also activated mechanically during treadmill step-ping, but in a tonic rather than phasic manner. The neural activation pattern that SCS provides when applied continuously at a constant stimulation frequency can be viewed indeed as a bilateral tonic drive to several spinal cord segments simultaneously. In contrast, the kinematic- and load-related feedback input to the spinal cord generated by assisted stepping movements on a treadmill can be viewed as phasic input to specific spinal cord segments. Hence, SCS and treadmill stepping may provide complementary ap-proaches to study some of the spinal mechanisms involved in rhythmic motor pattern generation after SCI. In addition to probing the human lumbar spinal circuits through afferent projections acti-vated mechanically or electrically, recent findings introduced the possibility of their pharmacological activation (clinicaltrials.gov/ ct2/show/NCT01484184). Pharmacological stimulation using a specific combination of orally-active molecules (buspirone + levodopa and carbidopa) generated involuntary rhythmic muscle activity and left-right alternating leg movements in paralyzed indi-viduals lying supine, thought to be mediated by CPGs (see compan-ion paper, [49]). In this article we discuss in-depth the results of studies that focus on mechanisms underlying rhythmic motor pattern generation by the human spinal cord isolated from supraspinal inputs by a severe lesion and probed by different types of stimulation. Pertinent information derived from animal experimental studies is provided on the mechanisms underlying the generation of locomotor activity and its regulation by afferent input.

2. AFFERENT STIMULATION TO PROBE SPINAL PAT-TERN GENERATING MECHANISMS: ANIMAL EXPERI-MENTAL STUDIES

2.1. Afferent Input can Modify Ongoing Fictive Locomotion Induced by Tonic Stimulation One method in classical physiology experiments to study sen-sorimotor interaction during locomotion is to centrally induce fic-tive locomotion (rhythmic activity recorded without any movement generated), and then test whether selective afferent stimulation disturbs or resets the regular locomotor rhythm [50,51]. For in-stance, in cat preparations, fictive locomotion is induced either by intravenous administration of L-DOPA (and the monoaminoxidase inhibitor nialamide) in curarized, acutely spinalized animals, or by tonic electrical stimulation of the mesencephalic locomotor region (MLR) in the decerebrate model. Under these conditions, rhythmic alternating activity is recorded from the nerves supplying flexor and extensor muscles. The spinal circuits underlying the generation of such rhythmic patterns of fictive locomotion in absence of phasic descending or peripheral inputs are referred to as a CPG [13]. A disturbance of the rhythm of fictive locomotion by specific afferent stimulation is typically considered as evidence for direct access of the stimulated fibers to the CPG. Normally, in absence of locomotor activity, stimulation of group Ib afferents from extensor muscles evokes a short-latency, nonreciprocal inhibition among homonymous and close synergistic extensor motoneurons [52,53]. The same stimulation given during locomotor activity produces instead an oligosynaptic excitation in homonymous and synergistic extensor motoneurons throughout the limb [17,50]. Specifically, brief electrical stimulation of group I afferents from knee and ankle extensors effectively resets fictive locomotion during swing (to stance) or enhances extensor activity during stance. A contribution of muscle spindle Ia afferents and Golgi tendon organ Ib afferents has been demonstrated using differ-ent types of stimulation such as muscle vibration and low-intensity

nerve stimulation [17]. These findings support the notion that ex-tensor group I afferents have excitatory connections to the extensor-related elements of the CPG [17,54–56]. Resetting of an otherwise steady fictive locomotor activity can also be produced by brief trains of stimulation of the flexor reflex afferents (FRA) that include cutaneous, joint, and groups II and III muscle afferents. In curarized spinal cats, FRA stimulation deliv-ered during the extension phase of fictive locomotion terminates the extensor activity and initiates an abrupt transition to a phase-advanced flexor burst, while the same stimulation delivered during the flexion phase results in prolongation of this phase [57]. It was earlier observed by Lundberg and colleagues that trains of FRA stimulation could induce short sequences of self-sustained rhyth-mic, alternating discharges in efferents to flexors and extensors in L-DOPA treated spinal cats [58,59]. These effects are state-dependent, occurring only under the same conditions (i.e., injection of L-DOPA and monoaminoxidase inhibitors) during which fictive locomotor activity may develop in the acute spinal cat [4]. To-gether, these results are interpreted such that at least part of the interneurons intercalated within the FRA pathways are elements of the CPG networks [58–60]. Natural stimulation of the proprioceptors from the hindlimbs can also alter the timing of induced fictive locomotion. Frigon and Gossard demonstrated that sensory input from load- and stretch-sensitive receptors in ankle extensors influence the centrally gener-ated rhythm during fictive locomotion in adult decerebrate cats [61]. Sustained slight dorsiflexion applied during episodes of spon-taneous fictive locomotion significantly decreased the burst fre-quency of fictive locomotion by increasing the extension phase durations. Andersson and Grillner investigated the effect of pro-prioceptive feedback from hip movements on the spinal circuits underlying fictive locomotion in the acute spinal cats [62]. Fictive locomotion was pharmacologically elicited by intravenous admini-stration of L-DOPA, and curarization prevented generation of ac-tual movements. Imposed passive sinusoidal hip movements en-trained the rhythmic motor pattern, even after the entire hindlimb including the skin was denervated, except for the hip joint and sur-rounding muscles. The studies in cats reviewed in this section demonstrate that proprioceptive inputs from ankle and hip act phase-specifically upon the CPG to elicit powerful effects onto several muscle groups in a coordinated and synergistic manner. Afferent projections have parallel connectivity to motoneurons through segmental reflex pathways, but also through more complex, multi-segmental circuits, including the CPG. The coordinated effects of specific afferent inputs clearly require ongoing locomotor activity to be centrally generated. Indeed, afferent input was found in those studies to modulate CPG activity, but not to cause the locomotor activity. The CPG circuits were rather activated by a tonic drive provided by electrical stimulation of the MLR or by pharmacological neuro-modulation.

2.2. Afferent Feedback Input can Produce Stepping on the Treadmill Under Specific Conditions In the experimental animal studies reviewed in section 2.1, the CPGs producing fictive locomotion were uncoupled from produc-ing any movements. However, movement-induced afferent feed-back can be utilized to assist generation or maintenance of func-tional rhythmic motor activity. Cats spinalized as adults can step involuntarily and bear weight through their hindquarters on a tread-mill after intense training [63]. The stance and swing phase durations, the joint angular excursions, and electromyographic (EMG) activity of hindlimb muscles are similar to those of non-injured cats stepping actively at comparable speeds [63]. However, even in these cases some additional excitatory inputs, such as perin-eal stimulation, tail pinching/crimping, or plantar pressure is re-quired during step training, at least at the initial stages of injury and

Human Spinal Locomotor Circuits Current Pharmaceutical Design, 2017, Vol. 23, No. 00 3

step training [64,65]. This suggests that although the adult spinal cat can perform weight-bearing hindlimb stepping on a treadmill after task-specific training [66], additional tonic excitatory inputs are required to initiate or maintain locomotor activity. Further, even after intense step training and tonic stimulation, locomotor abnor-malities persist in the adult spinal cat, involving a later onset of EMG bursts, especially that of the medial gastrocnemius muscle, reduced EMG amplitude and muscle force, changes in synchroniza-tion of flexor muscles acting at various joints which are related to foot drag at the onset of the swing phase, clonic EMG activity, and absence of equilibrium control [67–69]. Rhythmogenic capacity is also present in other mammalian models used for studies of spinal locomotor circuits, such as mice and rats. A recent study showed that spinal rats can recover invol-untary hindlimb stepping capability by treadmill training, but when pinching of the perineal area was applied throughout the training sessions [70]. Without such additional tonic stimulation, both un-trained and trained rats produced only occasional hindlimb flexion movements, likely in response to stretch when the limb was drawn backwards by the motorized treadmill belt. Other means of stimula-tion are normally used in these species to augment the central excit-ability to a level that allows lumbar circuits to respond to step-related feedback with the generation of coordinated stepping movement on a treadmill. For example, robotic devices that me-chanically modulate afferent input to the lesioned spinal cord [71], tonic epidural SCS that interacts with phase- and task-specific af-ferent feedback [72,73], and pharmacological agents such as sero-tonin and noradrenaline agonists [74,75] are used to increase loco-motor activity of these chronic spinalized animal models for study-ing restored stepping on a motorized treadmill. Thus, step-specific sensory feedback generated on a treadmill can induce hindlimb stepping in spinal cats, mice, and rats. Yet, in order to translate the afferent input into useful motor output, the lumbar circuits require additional activation by different types of tonic stimulation.

2.3. Drug Combinations can Induce CPG Activity with or with-out Significant Afferent Inputs Smaller animal models such as mice and rats have been used to demonstrate in-vivo that putative CPG-mediated locomotor-like movements using pharmacological aids in non-curarized, spinalized animals are inducible. Air-stepping with mice or rats completely suspended with a harness has been used to remove most significant load-related group I inputs. No other stimulation, such as tail pinch-ing, is applied in such studies. McEwen and colleagues have shown that neonatal rats with mid-thoracic spinal transection can express locomotor-like movements following L-DOPA (adrener-gic/dopaminergic precursor) and quipazine (5-HT2A/2C/3 agonist) administration subcutaneously [76]. Comparable effects found in adult mice were shown to critically depend upon spinal dopaminer-gic, serotonergic, and glutamatergic systems [77–80]. Effective CPG activation is best obtained using specific combinations of molecules [79,81]. Different types of agonists or precursors (for example 5-HT1A/7 receptor agonists, D1/5 receptor agonists, L-DOPA, 5-HT2A receptor agonists) can trigger some CPG activity, often mixed with non-locomotor movements, in air-stepping or treadmill conditions but only specific combinations with these molecules have been reported to elicit full weight-bearing stepping on a treadmill in untrained, non-otherwise stimulated mice follow-ing spinal transection [75,82–84]. This is somehow reminiscent of findings made earlier in isolated spinal cord preparations showing that dopamine and 5-HT with or without NMDA induce superior CPG-activating effects than either molecule administered separately [85,86]. When drawing conclusions from these studies, we should consider that the cells on which the drugs are acting and their exact mechanism of action are not completely defined, yet no neural sys-tem below a complete spinal transection other than the CPG is

known to trigger rhythmic motor activity in response to such phar-macological stimulation.

3. TREADMILL STEPPING AFTER SPINAL CORD INJURY IN HUMANS It has been suggested that “... on a moving treadmill, individu-als with spinal cord injury are enabled to perform rudimentary step-ping movements. These movements evoke an appropriate afferent input to the spinal cord leading to leg muscle activation comparable with that during walking ...” [87]. The aim of this section is to criti-cally review these assumptions.

3.1. Neural Activation Patterns Produced by Treadmill Step-ping in SCI Individuals In the generation of rhythmic leg muscle activity on a treadmill, proprioceptive feedback from the moving legs becomes one of the major sources for phasic modulation of spinal circuits after motor complete SCI [25,34,35,72]. In paralyzed individuals, each step is produced by external assistance under BWS conditions [88], either by physiotherapists [89] or robotic-driven gait orthosis systems [35]. The rhythmic, alternated loading and unloading of the legs together with the multi-joint flexion and extension movements re-sult in the mechanical activation of peripheral receptors in the hip and lower-limb muscles, tendons, and joints. This stimulation pro-duces gait-phase dependent sequences of afferent inflow to the spinal cord with complex temporal and spatial patterns (cf. [90,91]). When external assistance is withdrawn, the phasic EMG activation patterns ceases and the legs drag or collapse [92,93]. 3.1.1. Stretch-Mediated Neural Activation Patterns The role of muscle spindle afferents in the control of move-ment, and the contribution of stretch reflexes in the generation of motor output during assisted treadmill stepping in individuals with motor complete SCI have received much attention for more than two decades [25,92,94,95]. The muscle spindle is a sensory organ that detects muscle length and the velocity of muscle stretch [96]. There are large-diameter group Ia and medium-diameter group II muscle spindle fibers. The primary endings are sensitive to both the static and dynamic components of muscle stretch, while the secon-dary endings are mainly sensitive to the static component. Thus, the steady-state length of a muscle is signaled by the discharge of both group Ia and II muscle spindle afferents, while the group Ia affer-ents are highly sensitive to the velocity of stretch during the period of the actual joint movement. Further, muscle spindle activity is controlled by gamma motor innervation through corticospinal pathways that can increase the firing rate of the sensory endings [96]. Firing rates of muscle spindle afferents increase with increas-ing muscle lengths during passive stretch and active eccentric mus-cle contractions. During passive shortening of the muscle, firing pauses because of the unloading of the spindles. However, during concentric muscle contractions gamma motoneurons are activated together with alpha motoneurons and spindle afferents discharge despite the shortening of the muscle. The consequences of the disruption of descending gamma mo-toneuron activation after severe SCI are not well studied. Informa-tion gained from stroke patients [97,98] and from one study in indi-viduals with chronic motor complete SCI and spasticity [99] sug-gest that firing rates of muscle spindles recorded at rest and during passive stretch are not different from that of control subjects with intact nervous system. Nevertheless, during rhythmic muscle short-ening and lengthening movements, the absence of descending gamma modulation will inevitably affect the muscle spindle firing patterns. In individuals with intact nervous system, the gamma mo-tor system is activated in different types of voluntary concentric contractions, resulting in substantial spindle activation during mus-cle shortening [100]. In paralyzed individuals, muscle spindles will not be activated during the shortening phases of a movement and their firing patterns during imposed flexion-extension movements


may be similar to those generated by passive movements in indi-viduals with intact nervous system. Burke and colleagues studied the responses of muscle spindles of the tibialis anterior in individu-als with intact nervous system during passive and voluntary move-ments of the ankle joint [101]. Passive stretching and shortening movements of the relaxed muscle produced afferent firing only during the stretching phase. When the same movement was repro-duced actively, particularly against an external load, the muscle spindles also discharged throughout the concentric phase of the movement. The voluntary activation under external load thus de-creased the contrast between the afferent firing frequencies during the shortening and lengthening phases of the movements (Fig. 1). In addition, the firing frequency of the muscle spindle afferents was lower during passive stretch than during active eccentric contrac-tions due to the gamma modulation occurring in parallel with the muscle activity.

Fig. (1). Firing patterns of muscle spindle afferents during rhythmic movements. The mean firing rates of muscle spindle afferents in the tibialis anterior during passive and voluntary ankle movement in individuals with intact nervous system. The bars to the left represents the values obtained with passive movements of the relaxed muscle, the three pairs of bars to the right are based on voluntary movements, with the receptor-bearing muscle activated without external load or with different degrees of load. Data from different conditions were derived from stretching or shortening sequences with the same amplitude and speed of movement. Sustained firing of the afferents throughout the eccentric and concentric phases of active movement under load reduces the ratio of the firing frequencies between lengthening and shortening of the muscle. The situation in a paralyzed individual during externally imposed movements would be best represented by the passive movements described by the far left bars. Modified with permission from [101].

During active stepping, activity of group Ia and II fibers from muscle spindles is influenced by gamma drive with tonic and phasic components that can have different contributions in different mus-cles, as suggested in cats [18,20,102,103]. Thus, the exact firing patterns of the stretch sensitive receptors will be modified during passive treadmill stepping in the absence of volitional-contraction related modulation of gamma drive. Passive stretch of a muscle during imposed stepping movements on the treadmill would in-crease the firing frequencies of the muscle spindle afferents, while firing frequencies will decrease and drop to zero during the passive shortening phases of that muscle. This means that the spindle firing frequencies during alternated lengthening and shortening move-ments would become more phasic when compared to active walk-ing of an individual with intact nervous system. Spindle firing would only contribute to muscle activation during the lengthening

phases of the receptor-bearing muscle. The contribution of stretch reflexes to muscle activation may hence decelerate forward move-ments of the leg during swing or increase extensor-muscle and joint stiffness during stance of imposed stepping movements on the treadmill. Stewart and colleagues observed that when the EMG activity during assisted treadmill stepping in individuals with SCI was reduced (following the administration of clonidine), “... the research assistants reported a reduction in the resistance previ-ously felt when pulling the leg forward ...“ [92]. Physiological muscle spindle firing patterns would be further related to the exact kinematics of the externally imposed stepping-like motions on the treadmill. Some potential differences between the ankle- and knee-joint movement during assisted passive step-ping of individuals with SCI and physiological stepping of indi-viduals with intact nervous system on a treadmill shall be men-tioned here. When the patient makes contact with the ground through the heel first, a rapid passive plantar flexion movement can follow due to lack of controlled eccentric contraction of the dorsi-flexors at the early stance phase, resulting in a sudden stretch of the tibialis anterior. The occurrence of this non-physiological muscle spindle firing is specifically likely when using exoskeletons that do not permit active control of the ankle-joint movements. When the patient makes ground contact through the forefoot first, ankle ex-tensor muscle spindle afferents will fire earlier than during normal locomotion. Last, during manually-assisted stepping and minimum BWS of the patient, the knee joint at the end of the stance phase sometimes gives away, when the therapist stops supporting exten-sion while preparing to initiate the swing phase of the stepping movements, resulting in a sudden stretch of the knee extensors and thus to another phase-inappropriate muscle spindle discharge. Many spinal pathways control the synaptic effectiveness of afferent inputs and stretch reflex excitability. Specifically, sensory information is gated by presynaptic inhibition [104–107] depending on the state of the reflex and locomotor circuits [108]. When pre-synaptic inhibition is exerted at the afferent axon terminals on mo-toneurons, it can reduce the amount of transmitters that will be released by these afferents following their activation. Presynaptic inhibition is under descending control [95] and is modulated in a phase-dependent manner during stepping in humans [109,110], as well as in absence of any movement during induced fictive locomo-tion in the cat [111–114]. Presynaptic inhibition is thus regulated by descending control, sensory feedback as well as by the operation of the CPG. At the onset of selective voluntary contractions of indi-vidual muscles, presynaptic inhibition of Ia fibers to motoneurons of the contracting muscle is decreased, while presynaptic inhibition of Ia afferents on motoneurons not involved in the contraction is increased [115]. Thus, presynaptic inhibition at the onset of move-ment is organized in a way to increase selectivity of muscle activa-tion. During locomotion, presynaptic inhibition filters afferent in-puts according to the phase of the step cycle before the input can affect post-synaptic neural targets, and can thus reduce the impact of afferent inputs in phases where it could produce unwanted motor output [95]. The absence or dysfunction of any supraspinal or spinal control mechanisms of presynaptic inhibition could thus result in the generation of non-functional, non-timely appropriate muscle activity during passive movement. Indirect methods to study presynaptic inhibition in humans include the assessment of the soleus H-reflex size during specific motor tasks, and the amplitude modulation of the soleus H-reflex following conditioning stimulation of the femoral or common per-oneal nerve, or mechanical tendon vibration [116]. The H-reflex size depends on the excitability state of motoneurons, as well as on the level of presynaptic inhibition acting upon the Ia afferent termi-nals responsible for the monosynaptic depolarization of alpha mo-toneurons [116]. When the H-reflex size alone is used in studies of presynaptic inhibitory control during a motor task or movement, it is based on the assumption that at similar background EMG activity


levels, postsynaptic inhibition remains largely constant and thus changes in the H-reflex size that are not in accordance to the back-ground EMG modulation can be attributed mostly to presynaptic inhibition [116]. Based on this assumption, the task-dependent modulation of presynaptic inhibition was described in humans. During the transi-tion from seated to standing, and from standing to walking, there is a rapid reduction of the H-reflex size even when there is little change in the EMG level during the initiation phase of the move-ment [117,118]. Capaday and Stein compared soleus H-reflex am-plitudes and modulations in individuals with intact nervous system during active stepping on a treadmill and standing [14]. During stepping, the soleus H-reflex was strongly modulated in amplitude throughout the step cycle. The soleus H-reflex was small at the time of foot contact but increased rapidly to a maximum amplitude dur-ing late stance, and then decreased to a low value or was completely absent after toe-off and throughout the swing phase. This amplitude modulation of the soleus H-reflex was not always closely correlated to the background EMG modulation. Further, at similar soleus EMG levels, the soleus H-reflex was always much larger during tonic contractions while standing than during rhythmic activity while walking. Since the H-reflex amplitude was not always closely related to the EMG produced during stepping and was task-dependent (standing vs. walking), it was suggested that the modula-tions were not simply a reflection or function of motoneuronal ex-citability, but that presynaptic inhibition was a major contributor to these observations. A subsequent study of the same group found that while peak EMG activity of soleus during running was consid-erably higher than during walking on a treadmill, the maximum amplitude of the H-reflex was significantly smaller during running than during walking [14]. A tonic increase in the amount of presyn-aptic inhibition of the Ia terminals to alpha motoneurons during running was suggested as the most likely mechanism accounting for the difference in H-reflex gain in the two motor tasks. Knikou and colleagues compared the soleus H-reflex following conditioning stimulation to the antagonist nerve at an interval that the soleus H-reflex depression is associated with the presynaptic inhibitory net-work [110]. At similar background EMG levels, the conditioned soleus H-reflexes were modulated in a manner similar to the un-conditioned soleus H-reflexes, and it was suggested that presynaptic inhibition was upregulated at heel contact and downregulated at late stance and early swing phases [110]. The pronounced locomotor-task specific presynaptic inhibition suggests a downregulation of Ia monosynaptic feedback contribu-tion to muscle activation during active stepping in individuals with intact nervous systems [95]. There is a general understanding that presynaptic inhibition of Ia afferent terminals is decreased in spastic SCI patients [119–122]. Knikou and colleagues studied the modula-tion of the soleus H-reflex and of presynaptic inhibition during treadmill walking in people with motor complete and incomplete SCI [36,37,123]. The soleus H-reflexes showed considerably less modulation compared to healthy subjects during stepping, but the modulation patterns varied among patients. The most common pathological patterns observed were lack of or reduced soleus H-reflex depression during the swing phase and absent progressive facilitation of the soleus H-reflex from mid- to late-stance phases (Fig. 2). In patients with more severe spasticity, the H-reflexes elicited at the different phases of the step cycle showed little or no modula-tion. Further, presynaptic inhibition exerted on Ia afferents, as-sessed via a conditioning soleus H-reflex protocol, was absent in motor complete and incomplete SCI individuals when tested at rest, and its modulation during assisted stepping was different to that of healthy control subjects [122]. It was suggested that a combination of mechanisms had resulted in the abnormal behavior of the soleus H-reflex during stepping, including impaired function of reciprocal

!"#$%&'( )*$+#$,'!-./'0.!)1

!"#$%&'()*$

+#$,'& 23$'45'"+'6

7,267#'8

)97:$ '$ :";$<'7= '$7 >?'@2A B

C D E F G H I J K CL CC CD CE CF CG CH

L

CL

DL

EL

FL

GL&=7A>$ &92AM

!"#$%&'( )*$+#$,'-"A=*"#&

N2A'A%6@$*'"+'=?$'&=$O'>P>#$

L

CL

DL

EL

FL

GL

L

D

F

H

J

CLQ&

!

R7&&2:$ S"#%A=7*P'6":$6$A=

TA#"7<$< LUG'V6 CUCF' V6

"

8$7A'+2*2AM'*7=$'<%*2AM &=*$=>?

<%*2AM &?"*=$A2AM

Fig. (2). Step-related modulation of spinal pathways underlying the stretch reflex alters after spinal cord injury. Soleus H-reflex modulation during treadmill stepping in people with intact nervous system (controls, top) and in people with incomplete SCI (bottom). In the patients, H-reflexes elicited at the different phases of the step cycle were not physiologically modulated, partly because of altered proprioceptive-input processing after SCI. Modified with permission from [37,123].

inhibition that contributed partly to co-contractions between ankle antagonistic muscles (Fig. 3) and clonic EMG bursts in soleus [124], reduced levels of presynaptic inhibition, reduced levels of postsynaptic inhibition, and abnormal excitability of soleus moto-neurons.

!"#$%&'()*'+,-,(#,&'()+$.,".'/"(/+,0(+,")

!"#$%&'(#)(*%&'(#

+",&-.

/

0

1

2

3

4+",&5

-6

--

-.

-/

-0

-1

-2+",&-

.

/

0

1

2

3

4+",&5

-6

--

-.

-/

-0

-1

-2

789:%";9%"8,&<6&&&.6&&&06&&&26&&&46&&&-66&&&

789:%";9%"8,&<6&&&.6&&&06&&&26&&&46&&&-66&&&

Fig. (3). Coactivation of ankle antagonistic muscles during stepping in individuals with spinal cord injury. Coactivation indices (0–100%), shown as grey circles, for the left and right soleus and tibialis anterior mus-cles for each bin of a step during assisted treadmill stepping (divided into 16 equal bins; bin 1 corresponds to the onset of the stance phase and bin 9 to the swing phase initiation). Coactivation between the ankle antagonistic muscles is evident throughout all phases of the step cycle. The coactivation suggests impaired modulation of pre- and postsynaptic inhibitory mecha-nisms acting on the associated alpha motoneurons. Modified with permis-sion from [122].


In conclusion, a basic correlation of muscle spindle firing pat-terns with shortening and stretching of the receptor-bearing muscle during passive joint movements on the treadmill will remain after SCI. The exact firing patterns, however, will change in absence of supraspinal-controlled gamma modulation, probably resulting in a more phasic pattern synchronized to the muscle lengthening phases of movement. Additional differences of the muscle spindle firing patterns will result from differences between the exact kinematics of the assisted joint movements on the treadmill and voluntary stepping. Considering enhanced transmission of impulse trains at the Ia-motoneuron-synapse due to reduced homosynaptic depres-sion in SCI individuals [108,122], and reduced functional presynap-tic inhibition of Ia monosynaptic feedback normally associated with an active locomotor task [14,95,122,125,126], the effects of stretch-induced muscle spindle afferent volleys will play a major role in the generation of the motor outputs as observed during assisted tread-mill stepping. Yet, stretch reflex pathways may not contribute to a concentric muscle contraction, and altered proprioceptive-input processing following SCI can result in stretch-related muscle activ-ity in physiologically inappropriate phases of the stepping move-ments. 3.1.2. Load-Mediated Neural Activation Patterns Load-sensitive feedback provides another important contribu-tion to the muscle activity during assisted treadmill stepping in SCI individuals [25,35]. The Golgi tendon organ, a proprioceptive re-ceptor that senses changes in muscle tension and transmits this information via the Ib afferent pathway, is normally thought to provide the major load-related feedback, but primary and secondary muscle spindle afferents and cutaneous afferents also have the ca-pacity to signal load-related feedback information [127]. Load-related afferent feedback facilitates the ankle extensor EMG activity in individuals with intact nervous system at the stance phase [128,129]. Studies applying sudden perturbations of the ankle angle during treadmill stepping have been conducted to identify the contributions of the afferent modalities to the soleus EMG activity during locomotion. An abruptly imposed ankle plan-tar flexion during the stance phase of walking unloads the muscle-tendon complex of soleus and produces a transient spinal-mediated drop in the extensor EMG activity. The amount of EMG activity removed by the rapid plantar flexion is then interpreted as the pro-prioceptive contribution to the extensor activity during stance. It was shown that the soleus EMG activity and its unloading response to a rapid plantar-flexion perturbation were not influenced by an anaesthetic block of feedback from the foot and ankle [130]. Thus, cutaneous or proprioceptive afferents from intrinsic muscles of the foot likely did not contribute to the extensor EMG activity during the stance phase of stepping. The unloading effect remained unchanged after ischemic depression of the group Ia afferents from the lower leg, as well as following common peroneal nerve block [129]. These results imply that short-latency soleus stretch reflexes, or reciprocal Ia inhibition following a stretch of the tibialis anterior by the perturbation did not contribute to the unloading response in the soleus EMG. Grey and colleagues studied the effects of im-posed rapid plantar flexion perturbations on the unloading response in the soleus EMG at the late stance phase in subjects during step-ping on an inclined, level, or declined treadmill [131]. It was shown that Ib afferent feedback was modulated differently from the muscle spindle feedback during the different stepping conditions. The so-leus background activity and its unloading response were related directly to the Achilles tendon tension and inversely to the ankle displacement and angular velocity. These observations suggest that force feedback contributes predominantly to the late stance phase enhancement of the soleus EMG during locomotion. af Klint and colleagues showed that the amount of depression in soleus activity following plantar-flexion perturbation decreased with reduced limb load during treadmill stepping with transient alterations of the amount of BWS provided [132]. The pharmaceutical depression of

the secondary muscle spindle afferent pathway using the α2-adrenergic agonist tizanidine hydrochloride, however, did not affect the unloading response [132]. These series of studies suggest that Golgi tendon organ feed-back constitutes an essential part of the sensory feedback to the extensor activity, providing reinforcing force feedback during the late stance phase of locomotion via an excitatory group Ib pathway. It is not clear to what extent these observations are related to the results in experimental animals, where afferent activity from exten-sor nerves during the stance-like phases of fictive locomotion acts upon several muscles synergistically probably through the extensor part of the CPG circuits (cf. Section 2.1). Studies in humans were largely confined to the soleus muscle only. Grey and colleagues studied the unloading response also in the medial gastrocnemius, however, the response was not present in most subjects and when present, it was of very short duration compared with that of the soleus muscle [131]. The importance of limb-load related afferent feedback in the generation of EMG activity during assisted and BWS treadmill stepping in individuals with motor complete SCI has been repeat-edly reported. Dobkin and colleagues observed that the EMG activ-ity in soleus and medial gastrocnemius muscles increased with de-creasing BWS and suggested that load and joint kinematics were among the components that modulated the motor output [89]. Harkema and colleagues showed that EMG activity in soleus and medial gastrocnemius muscles were tightly correlated with the limb peak loading at the stance phase during BWS and manual assisted stepping in motor complete SCI [25]. The correlation between so-leus mean EMG amplitude and soleus muscle-tendon stretch during the lengthening EMG activity of the muscle was lower than the correlation between the soleus mean EMG amplitude and peak load, and lower than the correlation between the soleus EMG mean amplitude and muscle-tendon length changes during the entire step cycle in a person with motor complete SCI [25]. These results were interpreted such that the level of loading is a sensory feedback from the periphery that can modulate efferent activity in a manner similar to people with full supraspinal control. However, co-contraction between dorsiflexors and plantarflexors was apparent during the step cycle, and the non-physiological EMG activity of tibialis ante-rior that occurred during stance was related to the limb load as well [25]. This finding might thus suggest that limb loading was not simply triggering a critical feedback mechanism for ankle extensor activity, but enhanced the overall responsiveness of the spinal cir-cuits (including non-functional activity) to the step-induced afferent feedback. In a similar context, it was suggested that the generation of EMG activity on a treadmill in persons with motor complete SCI required load-sensitive feedback [35]. Simulated stepping-movements produced by a robotic exoskeleton during 100% BWS without ground contact generated no or low-amplitude EMG activ-ity in the paralyzed legs. This was interpreted such that pathological stretch reflexes induced by the movements contribute little to the leg muscle activation during walking. Robotic-driven treadmill stepping with 70% BWS on the other hand was reported to generate EMG activity “which in several aspects was similar to that ob-served in the healthy subject” [35]. Close inspection of the reported EMG data however suggests that the medial gastrocnemius was active at both stance and swing phases, and tibialis anterior showed only a short period of activation at the stance-to-swing transition phase, following the unloading foot-off movement when the muscle was passively stretched. Further, biceps femoris was active during swing and rectus femoris briefly during late stance, in both cases during the respective phases when the muscles were passively stretched, and both activities were absent in a healthy control group. In an attempt to modify the stretch-related feedback while continu-ing to provide alternating limb load, the knee joints of the robot were blocked during the imposed step-like movements and the re-sults were reported to be “about the same as seen with normal leg


movements” [35]. The published data however show the absence of activity in rectus femoris and tibialis anterior, and alterations in the activation of other muscles as well. A careful interpretation of these findings – also within the context of the studies discussed in Section 3.1.1 – then might be that both, muscle-stretch related as well as load related feedback input contribute to the EMG activation on the treadmill in individuals with motor complete SCI. One effect of the load sensitive feedback may be to add to the central state of excit-ability of the lumbar spinal circuits [33,133]. To what extent the resulting EMG activation pattern represents ‘locomotor’ activity is further discussed in section 3.2. 3.1.3. Step-Related Neural Activation Patterns and Sensory-Input Processing are Altered After SCI In conclusion of section 3.1, a severe SCI results in disruption of the supraspinal activation of alpha and gamma motoneurons, alternations in the neural firing patterns of muscle-stretch related proprioceptive feedback during movement, impaired modulation of incoming afferent inputs, and impaired function of the pre-motoneuronal spinal circuits. The spinal lesion transforms the effect of stretch-induced feedback to such an extent that it becomes cru-cial for the muscle activation during assisted locomotor movements as compared to its more limited contribution to muscle activation during active walking in individuals with intact nervous system. Load-sensitive feedback additionally contributes to muscle activity generated on the treadmill. The absence of the supraspinal activa-tion of neuronal circuits underlying the reciprocal organization of antagonistic muscles, and of the supraspinal task- and phase-dependent modulation of presynaptic inhibition results in altered sensory-input processing after SCI and hence may alter the inte-

grated actions of the afferent input upon the motoneurons and gen-erate pathological as well as phase-inappropriate motor activity.

3.2. Rhythmic Motor Activity Patterns During Treadmill Step-ping After SCI in Humans The human spinal cord circuits below a severe cervical or tho-racic SCI have the ability to produce motor activity during BWS and assisted stepping. To what extent this motor activity is ‘loco-motor-like’ and resembles the voluntary muscle activity observed during normal human walking and whether this motor output is due to the activation of CPGs, as suggested by some researchers [134–136], are issues that warrant an in-depth discussion. In Fig. 4, the muscle activation profiles from people with motor complete (AIS A, B) and motor incomplete (AIS C, D) SCI and healthy subjects (controls) during robotic-assisted stepping at 50% BWS are indicated [37,123]. In motor incomplete SCI, the muscle activation profiles have some similarities to those observed in healthy subjects, but the activation profiles of both ankle and knee muscles are greatly impaired in motor complete SCI. For example, the medial hamstrings activity at the swing-to-stance transition and the prolonged activity in tibialis anterior during swing associated with muscle contraction for foot clearance as produced during nor-mal walking are absent in the motor complete SCI subjects. Both activities are largely occurring in the absence of muscle lengthening in healthy individuals. A common explanation of the EMG patterns produced in indi-viduals with severe SCI on the treadmill is that the activity is due to partial or full operation of CPGs, but this cannot explain the patho-

Fig. (4). Leg EMG activation profiles during stepping in individuals with motor complete SCI (AIS A-B), motor incomplete (AIS C and AIS D) SCI, and intact spinal cord (controls). Mean average normalized EMG from the left and right leg muscles during robotic-assisted treadmill stepping at 50% BWS from one subject with AIS A, one with AIS B, four with AIS C, and eight with AIS D. Average normalized EMG for controls are from 13 healthy subjects. The EMG is normalized to the maximal homonymous EMG recorded during stepping. Grey background denotes the stance phase duration. Data adopted and modified with permission from [37,123].


logical EMG patterns that are apparent in motor complete spinal lesions. Further, the EMG patterns produced do not regularly evoke phases of ‘automatic’ or ‘spontaneous’ involuntary movement on the treadmill, such as the hip-stretch induced synergy to initiate the swing phase. Such observations are rather the exceptions. The acti-vation of spinal circuits underlying flexion and extension synergies in response to step-induced feedback input indeed seems to require a combination of some degree of spastic muscle tone (background excitability) and residual descending connectivity, as well as spe-cific movement-mediated afferent cues. Wernig and Müller re-ported that individuals with severe SCI could induce a few unaided bipedal steps after prolonged periods of treadmill training [33]. Specifically, five individuals with motor incomplete SCI with over-all low motor scores and with one leg demonstrating complete func-tional paralysis (when tested in a resting supine position) could elicit alternating flexion-extension synergies when in an upright position over a motorized treadmill belt. Initiation and maintenance of stepping was possible only when the hip joint was extended be-yond 180° after mid-stance to stretch the iliopsoas muscle, the knee joint was fully extended during the stance phase, and the leg was fully unloaded after bearing full weight. The stepping patterns in-duced in this study may suggest the activation of spinal circuits incorporating ipsilateral flexor or crossed extensor reflexes. In a study of Dobkin and colleagues, patients with incomplete SCI, yet with non-functional residual motor function, were encouraged to contribute actively to the stepping motions while manual assistance was continuously provided [89]. The voluntary facilitation not only augmented the EMG activity as generated by passive movements

alone, but also produced additional EMG bursts, like a burst in iliopsoas, medial hamstrings, and tibialis anterior muscles in the swing phases. After a few training sessions, these patients became able to self-initiate the swing phases on the treadmill when the hip was moved into extension. These observations suggested that su-praspinal inputs clearly added functional components to the EMG patterns during stepping that could not be generated by step-induced afferent feedback alone. We may consider at this point that EMG activities timed to the gait cycle are not necessarily locomotor-like, or step-like, just be-cause they are rhythmic in nature. In the following, we review in detail the characteristic activation patterns of various thigh and lower leg muscle groups with respect to the phases of the gait cycle as produced during assisted and BWS treadmill stepping in indi-viduals with motor complete SCI (Fig. 5). The timing of the activation in rectus femoris most frequently coincided with the stretch applied to the muscle in the late stance phase when the knee was passively flexed as the leg was drawn backward [35,88,92,137–141]. A similar activation during the stance-to-swing transition was described for the knee extensor vas-tus lateralis by Harkema [88], while Dobkin and colleagues [89] observed its activation from swing-to-stance transition, possibly during weight acceptance with knee flexion. Activity generated in biceps femoris and medial hamstrings were found to be timed to phases within the gait cycle when the leg was moved forward during swing phases, that is, when the biarticu-lar muscles of the posterior thigh were stretched [35,92,

Fig. (5). Timing of EMG activity in leg muscles during BWS and assisted stepping in individuals with motor complete SCI. The duration of activity is shown on a normalized time scale from heel strike to heel strike; shaded areas mark stance phases. RF, rectus femoris; VL, vastus lateralis; BF/MH, biceps femoris/medial hamstrings; TA, tibialis anterior; MG, medial gastrocnemius; SOL, soleus. Hip and knee goniometric data illustrated below RF, VL, and BF/MH derived from robotic-assisted stepping; knee and ankle data below lower-leg muscle groups derived from motor complete SCI subjects during manu-ally assisted stepping (black lines) and actively stepping non-injured controls (gray lines).


137,138,140–142]. One study showed that the hamstrings EMG activity increased by increasing the amplitude and the velocity of muscle stretch [141]. Tibialis anterior activity frequently occurred during the stance phase of the gait cycle [25,88,89,138,142], also in the form of clonus [92,141]. The onset and waning of the EMG activity typi-cally coincided with that of medial gastrocnemius [25,89,92,138]. A brief activation at the stance-to-swing transition [35,88,137] or no activation throughout the gait cycle [140] was reported as well. The main activation of medial gastrocnemius occurred during the stance phase, and often started well before the onset of loading of the leg [25,35,89,137,140]. It should be noted that the gastroc-nemius is a biarticular muscle that is also stretched by the knee extension during late swing. Activity produced in soleus consis-tently started during mid- or late-swing phases and terminated be-fore the end of stance [25,34,88,89,92]. Clonic EMG activity was frequently observed across the vari-ous muscles [25,89]. Particularly, clonic EMG patterns were de-scribed at or following foot contact, throughout the stance phase and stance-to-swing transition during stepping [25,88,92,141–144]. Clonus was also found to occur before swing phase initiation and during mid-to-end swing in the gastrocnemius muscle [142,145]. Characteristically, muscle activation in motor complete SCI in response to afferent feedback on a moving treadmill belt occurs during the lengthening phases of the respective muscles and can be related to immediate stretch. One exception is the non-functional tibialis anterior activity during stance. This type of activity may be clonus triggered by the initial foot contact (cf. section 3.1.1) or reflect impaired reciprocal inhibition.

3.3. Influence of Spinal Circuits’ Excitability and Residual Su-praspinal Input on the Generation of Rhythmic Motor Output Increasing the central state of excitability of the lumbar spinal circuits caudal to an SCI by pinching of the leg, reducing BWS, electrical SCS, or reduced antispasticity medication, facilitates the motor output during assisted treadmill stepping [33,133,141, 146,147]. Specifically, some degree of spastic muscle tone in the lower limbs is critical for the generation of rhythmic EMG activity in response to the externally induced stepping movements [33,87,92,143]. In this context, spasticity may be interpreted as an increased excitability of the spinal circuits, also reflected by an increase in their susceptibility to respond to afferent input. Inter-individual differences in the relative excitability of various spinal pathways may also be the cause of the variability in motor patterns generated during assisted treadmill stepping with relatively similar kinetics and kinematics across the population of motor complete SCI individuals [88]. Stewart and colleagues investigated the ef-fects of clonidine, a noradrenergic agonist, on spasticity and on the EMG activity generated during treadmill stepping in people with SCI, of whom six had no motor function below the lesion [92]. The stepping movements in the paraplegic patients were manually in-duced under BWS. The EMG patterns seen in the placebo sessions in most paraplegic subjects during assisted treadmill stepping were stretch-related activity in the medial hamstrings in the swing phase, and clonic activity in tibialis anterior and gastrocnemius evoked at foot contact. Similarly, when tested in a sitting position at rest, hamstrings activity was consistently produced during passive exten-sion of the knee, and clonus in the lower leg muscles could be evoked as well. During the clonidine sessions, the manually im-posed leg movements on the treadmill largely failed to generate any muscle activity. In the sitting position, the tonic stretch reflexes and clonus were abolished while on clonidine. The responsiveness of the spinal circuits may be further influ-enced by some residual descending neural connections. Even after a motor complete SCI according to standard neurological classifica-tion [148], a small amount of white matter crossing the lesion is

commonly preserved [149–151], and propriospinal connections [152] may also survive and pass neural information around the lesion site [153,154]. Conduction along such preserved systems is clearly compromised, but may provide some translesional inhibition or excitation to the lumbar spinal circuits [33,147,155,156]. In conclusion, motor complete SCI individuals require a certain excitability level of the lumbar spinal circuits on which step-related peripheral feedback can act to produce rhythmicity during assisted treadmill stepping. Such ongoing activity within the circuits can result from some degree of spinal spasticity. Most of the leg muscle activation patterns described in motor complete SCI occur during the muscle-lengthening phases of the stepping movement and can be related to immediate stretch. Sensory-driven flexion synergies across multiple joints that produce muscle contraction and actual leg movements seem to require some motor-incompleteness of the SCI.

4. SPINAL CORD STIMULATION AFTER SPINAL CORD INJURY IN HUMANS

4.1. Neural Activation Patterns Produced by Spinal Cord Stimulation The spinal locomotor network studied in reduced animal prepa-rations is normally activated by tonic pharmacological or electrical stimulation, that is, by sustained excitatory inputs with rather in-variable temporal and spatial aspects. It has been assumed that such excitation mimics an unpatterned descending drive from the brain-stem that spinal cord networks convert into rhythmic outputs [157]. With the application of SCS in motor disorders [158,159] and the specific placement of epidural electrodes over the lumbar spinal cord [160], a method that could deliver a tonic excitatory drive to spinal cord segments that are essential for locomotion was devel-oped in humans. The spinal circuits are likely activated through a combination of afferent stimulation of posterior roots (also termed dorsal roots), containing axons of the dorsal root ganglion neurons, and of posterior columns (dorsal columns), containing the ascend-ing intraspinal continuations of these axons. Afferent orthodromic action potentials in the posterior roots [29,45,46,161] and antidro-mic action potentials in the posterior columns [47,48], in turn, will result in excitation of spinal interneurons and motoneurons through collateral projections. Due to the compact arrangement of the lum-bar and upper sacral roots in the dural sac, as well as of the afferent branches in the posterior columns, each electrical pulse normally activates sensory fibers associated with several spinal cord seg-ments bilaterally [30,48]. Therefore, tonic SCS generates a series of highly synchronized multi-segmental excitatory inputs. The result-ing repeated release of neurotransmitters by the stimulated afferents may maintain a relatively continuous level of excitation of a wide range of spinal circuits, including those with pattern generating capability [162,163].

4.2. Rhythmic Motor Activity Patterns Produced by Spinal Cord Stimulation Tonic electrical stimulation of the lumbar spinal cord was shown to generate rhythmic motor outputs and movements in cats and rats, with or without pharmacological facilitation [4,73,164,165]. By applying epidural stimulation to the correspond-ing spinal cord segments in humans, locomotor-like activity can be generated in the paralyzed legs of individuals with complete SCI studied in the supine position [28,38,166,167]. The term locomotor-like in these studies denoted rhythmic EMG patterns with reciprocal activation in antagonistic lower-limb muscle groups and the genera-tion of involuntary flexion-extension movements. The finding that the lumbar spinal cord isolated from supraspinal control could con-vert a tonic excitation into coordinated rhythmic motor activity was interpreted as evidence for the existence of CPGs in humans [8,28]. Notably, SCS with appropriate parameters immediately induces rhythmic movements on the examination bed without manual assis-


tance provided to the legs. Such activity ceases after stimulation is stopped. The characteristics of the spinal pattern generating circuits were analyzed by Danner and colleagues [162]. In individuals with motor complete SCI lying supine, ongoing tonic SCS (at intensities three-fold the threshold for evoking EMG responses in the legs, and fre-quencies around 30 Hz) generated various rhythmic multi-muscle activation patterns in the lower limbs, including coactivation, mixed muscle synergies, and locomotor-like patterns (Fig. 6). Such rhyth-mic activity present in the main thigh and lower leg muscle groups were elicited in seven out of ten subjects with (motor-) complete SCI studied. Statistical decomposition of the EMG data identified three basic temporal components shared across the different EMG patterns, a flexor and an extensor basic activation pattern, and one that was characteristically associated with the activation of the bi-articular hamstrings muscle group during the rhythmic activity. The basic activation patterns were interpreted as excitatory neural drives provided by spinal burst generators, each imposing a specific spa-tiotemporally organized activation on the motoneuron pools [162]. The flexor and extensor basic patterns corresponded to the output of a CPG in reduced animal preparations [5,7]. The EMG burst activities produced by lumbar SCS are com-posed of a series of stimulus-synchronized responses, each repre-senting an afferent-induced compound muscle action potential, at short latencies [29,168]. Successively evoked responses are rhyth-mically amplitude-modulated, resulting in the burst-like EMG en-velopes. It was suggested that the rhythmic motor patterns had re-sulted from the integrated activities of repetitively activated spinal reflex circuits, as well as of the operation of multisegmental pat-tern-generating circuits, the latter being activated due to the tonic nature of the afferent input [30,41]. Experiments in rats have come to the same conclusion [169,170]. The stimulus-triggered EMG composition under SCS could be the result of synchronized excita-tory postsynaptic potentials and excitation of motoneurons and interneurons at multiple spinal segments. Because of the functional integrity of neuronal circuits and peripheral nerves below the lesion site, the contribution of move-ment mediated afferent feedback in the generation of rhythmic ac-tivity in the studies reviewed above cannot be excluded. However,

in the supine position, significant contribution from proprioceptors mediating loading and hip extension, being restricted by the exa-mination table, is unlikely. The switch between extensor- and flexor-like phases likely was controlled centrally in the absence of proprioceptive feedback cues, otherwise critical for phase transi-tions during stepping. It should though be noted that proprioceptive feedback during or after contraction and relaxation of leg muscles, and joint movements, could have added phasic activation to the spinal cord circuits. This said, SCS also generates antidromic action potentials propagating in the electrically stimulated sensory fiber toward the periphery [171,172]. The antidromic activity generated by repetitive SCS, such as at 30 Hz, may thus partially cancel the sensory feedback related to the induced rhythmic activities, and overall, reduce the influence of phasic sensory feedback on the rhythmic motor outputs. The burst frequency of the SCS-induced rhythmic output patterns must then have been largely determined by phasic changes innate in the spinal circuits. Under tonic drive, these circuits generated a wide range of physiologically relevant burst frequencies, including those corresponding to slow and fast gait [162]. The specific spinal circuits underlying the central generation of the rhythmic motor patterns remain unknown. In principle, a simple network with appropriate pattern of interconnections between neu-rons can generate rhythmic activity under ongoing tonic stimula-tion. A network consisting of two populations of flexor and exten-sor motoneurons, reciprocally coupled by Ia inhibitory interneu-rons, and Renshaw cells that provide a delayed inhibition of the Ia inhibitory interneurons, can produce oscillatory output in response to tonic input [173]. There are some observations to suggest that circuits involving Ia inhibitory interneurons and Renshaw cells are activated by SCS [47,163]. However, such inhibitory circuits alone cannot explain the burst-like patterns of rhythmic activities as pro-duced by SCS. For instance, non-modulated responses in the tibialis anterior muscle in response to low-frequency SCS (2–10 Hz) have relatively low EMG amplitudes, but the responses in the same mus-cle can attain large amplitudes in the form of rhythmic bursts of activity when SCS frequency is close to 30 Hz, without changing the intensity of the stimulation [168]. This observation suggests that during the rhythmic activity additional excitation acted upon the

Fig. (6). Rhythmic patterns of EMG activity and envelopes of the rectified EMG in response to tonic multi-segmental driving input provided by epidural SCS. Data derived from three motor complete SCI individuals lying in the supine position. Tonic stimulation was continuously applied with parameters given at the top of the figures; electrode pairing (‘0’, most rostral and ‘3’, most caudal electrode along a linear array, ‘+’ and ‘-’, polarity of the active electrodes), stimula-tion frequency and intensity. Gray background denotes flexion-like phases defined by the activity of tibialis anterior muscles. Q, quadriceps; Ham, hamstrings; TA, tibialis anterior; TS, triceps surae. Modified with permission from [162].


tibialis anterior motoneurons. Furthermore, the stimulus-synchro-nized responses constituting the burst-shaped activities during the flexion phases under SCS delay with respect to the shortest-latency reflexes [29,30,166]. This may be the result of the time needed for facilitation/disfacilitation of excitatory spinal circuits with interca-lated interneurons and their synaptic delays. Similar burst-phase related delayed responses were also identified during rhythmic activity induced by SCS in rats and were suggested to be related to the flexor reflex [174], or to be mediated through some elements of the spinal locomotor networks [169].

4.3. Tonic Excitation of the Human Lumbar Spinal Cord Acti-vates Circuits with Pattern Generating Properties In conclusion of section 4, it is obvious from the differences of the inputs provided and motor output patterns generated, that tonic SCS and treadmill stepping involved different mechanisms of rhythm generation. In the studies using SCS, the rhythmic motor output was not entrained by the excitatory inputs provided. Rhyth-micity was rather generated by spinal circuits that switched between flexion- and extension-like phases. To become and be maintained operational, the rhythm generating networks seem to require a con-tinuous excitation, as provided by tonic stimulation. While some of the rhythmic motor activity patterns produced by SCS in humans resemble those of fictive locomotion in reduced animal prepara-tions, further studies are required to clarify to what extent the cir-cuits engaged by SCS in humans correspond to CPG networks. Clearly, the leg muscle activation patterns produced by SCS are not the same as during active walking. Bipedal human locomotion in-volves more than just a simple pattern of flexion and extension. One question for future work would be to address to what extent task-specific sensory feedback could tune the rhythmic patterns evoked by SCS to become more functional.

5. PHARMACOLOGICAL STIMULATION AFTER SPINAL CORD INJURY IN HUMANS

5.1. Cellular and Intracellular Mechanisms Underlying Drug-Induced, CPG-Mediated Leg Movements As mentioned earlier, animal studies (in-vitro isolated spinal cord and in-vivo spinal transected mice) have shown that specific combinations of drugs using dopaminergic agonists, adrener-gic/dopaminergic precursors and/or 5-HT receptor agonists most effectively elicit putative CPG-mediated hindlimb locomotor movements (see section 2.3). Our understanding of the synergistic activating effects on the CPG enabled by such combinations of different molecules is incomplete. Using animals pretreated with selective receptor antagonists or knock-out models with missing specific receptor subtypes, Guertin and colleagues have obtained evidence suggestive that 5-HT1A, D1, and NMDA receptors are critically involved in pharmacological activation of CPGs [21,78–80]. Stepping induced by subcutaneous, intraperitoneal or oral ad-ministration of CPG-activating drugs in low-thoracic spinal tran-sected mice is likely centrally-mediated by lumbar CPG elements since well-coordinated movements are found. This conclusion is further supported by the similar effects that can be achieved when animals are suspended during air-stepping conditions as on a motor-ized treadmill, without requiring additional excitatory inputs, such as tail or sexual organ pinching. At the transmembrane level, it is well documented that buspirone is a 5-HT1A receptor agonist whereas L-DOPA is a noradrenaline/dopamine precursor. Carbi-dopa does not cross the blood brain barrier, it is a decarboxylase inhibitor typically used to augment by fivefold the central bioavail-ability of L-DOPA. The corresponding targets (e.g., 5-HT1A and D1 receptors) are expressed in lumbar spinal cord segments [175–178] and dopamine release (mainly from descending tracts originating supraspinally in

A11 nuclei; [179]) exists also in the spinal cord (D cells; [180]). L-DOPA is converted to dopamine and noradrenaline in catechola-minergic axon terminals and, in absence of descending dopaminer-gic terminals due to spinal transection, D cells may compensate by upregulation of aromatic L-amino acid decarboxylase (AADC; [180–183]). However, the mechanisms at an intracellular level are not well understood. For example, how L-DOPA and buspirone act together for effective CPG activation in spinal transected animals remains largely unclear. At least two possible explanations exist. Firstly, since CPG neurons belong to various classes (V0-V3, HB9, etc.) with different properties and neurotransmitters, it may be that L-DOPA and buspirone act on different classes of CPG elements that, once activated, can trigger overall activity in the entire network. Secondly, both compounds may act upon the same key CPG neu-rons (e.g., pacemaker HB9 cells) for better activation of G-coupled protein cascades involving cAMP and adenyl cyclase-induced [Ca2+]i increase leading to robust NMDA-dependent pacemaker property expression [75,78,79,184]. This is supported by results obtained in lampreys showing various converging actions upon intracellular Ca2+-dependent mechanisms including Ca(V)1.3 chan-nel-mediated robust post-inhibitory rebounds [185].

5.2. First Tests of CPG-Activating Drugs in Patients with Com-plete or Motor-Complete SCI For clinically relevant use in humans, it is imperative to identify drugs that can mimic the CPG-activating effects of comparable cocktails in animal models, but cross the blood brain barrier and are centrally active upon oral administration. According to these crite-ria, Guertin and colleagues developed SpinalonTM, a drug that is composed of L-DOPA, carbidopa and buspirone for activation of both spinal dopaminergic and serotonergic mechanisms, and dem-onstrated its efficacy in mice [82]. This combination was also cho-sen because each of the molecules were shown to be relatively safe in patients with Parkinson’s disease or anxiety, respectively. Moreover, co-administration of L-DOPA, carbidopa (or other de-carboxylase inhibitors) and 5-HT1 agonists (buspirone or compara-ble agents) were already tested in patients with Parkinson’s disease for improved levodopa-induced effects without increasing side effects [186]. A phase I/IIa study on the effects of SpinalonTM was recently completed in 45 individuals with (motor) complete SCI (see com-panion paper, [49]). The study was a randomized, placebo-con-trolled, double-blind dose escalation study. The participants re-ceived SpinalonTM in the form of over-encapsulated buspirone and Sinemet® (levodopa/carbidopa), or a placebo, an over-encapsulated cornstarch powder. The tests were performed in the supine position. Blood samples before and at regular intervals after treatment were collected for hematological and pharmacokinetic analyses. EMG activity of vastus lateralis, biceps femoris, tibialis anterior, and gastrocnemius muscles bilaterally was monitored before and at several time points after drug administration. Buspirone/levodopa/ carbidopa (10-35 mg/100-350 mg/25-85 mg) showed no sign of safety concerns. Only mild nausea was reported by 3 participants. At higher doses (50 mg/500 mg/125 mg), buspirone/levodopa/ car-bidopa was considered to have reached maximal tolerated dose, since 3 out of 4 subjects experienced related adverse events includ-ing vomiting. The SpinalonTM-treated groups displayed significant EMG ac-tivity with locomotor-like characteristics, that is, with rhythmic and bilaterally alternated EMG bursts. The onset of the rhythmic activ-ity was spontaneous and occurred within 15 and 120 minutes post-administration while the subjects were lying supine, without manual assistance provided by the examiners. The study therefore provided preliminary evidence on safety and efficacy following a single ad-ministration of a rhythmic-activity producing drug that putatively acted upon the CPG in people with SCI.


Guertin and colleagues are currently planning further tests to provide evidence on the potential of this approach in triggering stepping movements on a motorized treadmill in people with severe SCI, without or with additional means of applying excitatory input to the spinal cord, such as muscle vibration or electrical stimulation. It will be also of interest whether repeated administration of Spi-nalonTM over a longer period of time would increase the activity of the spinal CPG and the induced-motor output.

6. CONCLUSION AND FUTURE DIRECTIONS Experimental studies have demonstrated CPG capabilities of the isolated mammalian lumbar spinal circuits and the interaction of specific afferent inputs with the CPG activity. In-vivo animal ex-periments have further shown that proprioceptive feedback can produce hindlimb stepping on a motorized treadmill, but additional sources of excitatory drive provided to the spinal cord are normally required to initiate or maintain effective weight-bearing stepping. Externally assisted treadmill stepping in individuals with motor complete SCI can produce leg muscle activity that is synchronized to the phases of stepping, yet the activation patterns differ from those during normal walking. Much of the EMG activity produced can be explained by the entrainment of spinal reflex circuits by proprioceptive feedback related to cyclically appearing muscle stretch. SCI alters the rhythmic patterns of muscle spindle firing during passive stepping movements on a treadmill, and an altered input-processing by spinal cord circuits is apparent. This leads to an increased contribution of stretch-induced afferent volleys to the generation of the motor outputs in these patients. Some critical level of sustainable state of excitability of the circuits, as can result from spinal spasticity, is required to respond to the feedback inputs with the generation of motor output. Load-related proprioceptive input is probably further increasing the central excitability state of the cir-cuits. SCS can provide a tonic excitation to the human lumbar spinal cord, a type of sustained drive also used in animal experimental studies to activate spinal CPGs. The SCS-induced muscle activity occurs largely within two alternating phases of rhythmicity, thus being more similar to fictive locomotion in reduced animal prepara-tions than to the complex muscle activation patterns of active bi-pedal human walking. Preliminary evidence suggests that specific combinations of pharmacological substances that can effectively activate CPGs in animal models can also generate rhythmic activity in the paralyzed legs in individuals with a motor complete SCI studied in the supine position. It remains to be shown whether and how proprioceptive inputs generated during stepping interact with SCS and pharmaco-logically-driven CPG activity. The development of new therapeutic strategies for restoring motor function of paralyzed muscles after SCI is a continuous en-deavor from both the scientific and clinical community. While sig-nificant advances have been made in understanding better the pathological expression of muscle tone and paralysis in humans, it is apparent that rehabilitation requires the simultaneous utilization of different strategies. Locomotor training may be administered along with SCS and CPG-activating drugs to restore motor function in patients who remain wheelchair bound after standard-of-care rehabilitation. Thus, there is a great need for clinical research stud-ies that utilize neurophysiological, clinical, and pharmacokinetic biomarkers under conditions during which training of the motor system is multidimensional.

GRANTS The Craig H. Neilsen Foundation (339705) and the New York State Department of Health (DOH01-C30836GG-3450000) [M.K.]; Canadian Institutes of Health Research and Natural Sciences and Engineering Research Council of Canada [P.A.G.].

CONFLICT OF INTEREST P.A.G. is president and CEO of Nordic Life Science Pipeline Inc. U.S.H., M.K., and K.M. have no conflicts of interest to report.

ACKNOWLEDGEMENTS Declared none.

REFERENCES [1] Petersen TH, Willerslev-Olsen M, Conway BA, Nielsen JB. The

motor cortex drives the muscles during walking in human subjects. J Physiol 2012; 590: 2443-52.

[2] Pearson K, Gordon J. Locomotion. In: Kandel E, Schwartz J, Jes-sell T, Siegelbaum S, Hudspeth A, Eds. Principles of neural sci-ence. New York: McGraw-Hill 2013; pp. 812-34.

[3] Drew T, Marigold DS. Taking the next step: cortical contributions to the control of locomotion. Curr Opin Neurobiol 2015; 33: 25-33.

[4] Grillner S, Zangger P. On the central generation of locomotion in the low spinal cat. Exp Brain Res 1979; 34: 241-61.

[5] Grillner S. Control of locomotion in bipeds, tetrapods, and fish. In: Brooks V, Ed. Handbook of physiology Section 1: The nervous system, vol II Motor control. Bethesda, MD: American Physiologi-cal Society 1981; pp. 1179-236.

[6] Stuart DG, Hultborn H. Thomas Graham Brown (1882--1965), Anders Lundberg (1920-), and the neural control of stepping. Brain Res Rev 2008; 59: 74-95.

[7] Grillner S. Neuroscience. Human locomotor circuits conform. Science 2011; 334: 912-3.

[8] Guertin PA. Central pattern generator for locomotion: anatomical, physiological, and pathophysiological considerations. Front Neurol 2012; 3: 183.

[9] Guertin PA. Preclinical evidence supporting the clinical develop-ment of central pattern generator-modulating therapies for chronic spinal cord-injured patients. Front Hum Neurosci 2014; 8: 272.

[10] McLean DL, Dougherty KJ. Peeling back the layers of locomotor control in the spinal cord. Curr Opin Neurobiol 2015; 33: 63-70.

[11] Jordan LM, Liu J, Hedlund PB, Akay T, Pearson KG. Descending command systems for the initiation of locomotion in mammals. Brain Res Rev 2008; 57: 183-91.

[12] Hägglund M, Borgius L, Dougherty KJ, Kiehn O. Activation of groups of excitatory neurons in the mammalian spinal cord or hindbrain evokes locomotion. Nat Neurosci 2010; 13:246-52.

[13] Grillner S. Neurobiological bases of rhythmic motor acts in verte-brates. Science 1985; 228: 143-9.

[14] Capaday C, Stein RB. Amplitude modulation of the soleus H-reflex in the human during walking and standing. J Neurosci 1986; 6: 1308-13.

[15] Pearson KG, Ramirez JM, Jiang W. Entrainment of the locomotor rhythm by group Ib afferents from ankle extensor muscles in spinal cats. Exp Brain Res 1992; 90: 557-66.

[16] Pearson KG. Generating the walking gait: role of sensory feedback. Prog Brain Res 2004; 143: 123-9.

[17] Guertin P, Angel MJ, Perreault MC, McCrea DA. Ankle extensor group I afferents excite extensors throughout the hindlimb during fictive locomotion in the cat. J Physiol 1995; 487: 197-209.

[18] Rossignol S, Dubuc R, Gossard J-P. Dynamic sensorimotor interac-tions in locomotion. Physiol Rev 2006; 86: 89-154.

[19] Frigon A, Rossignol S. Experiments and models of sensorimotor interactions during locomotion. Biol Cybern 2006; 95: 607-27.

[20] Giroux N, Rossignol S, Reader TA. Autoradiographic study of alpha1- and alpha2-noradrenergic and serotonin1A receptors in the spinal cord of normal and chronically transected cats. J Comp Neu-rol 1999; 406: 402-14.

[21] Landry ES, Rouillard C, Levesque D, Guertin PA. Profile of im-mediate early gene expression in the lumbar spinal cord of low-thoracic paraplegic mice. Behav Neurosci 2006; 120: 1384-8.

[22] Edgerton VR, Leon RD, Harkema SJ, et al. Retraining the injured spinal cord. J Physiol 2001; 533: 15-22.

[23] Bareyre FM, Kerschensteiner M, Raineteau O, Mettenleiter TC, Weinmann O, Schwab ME. The injured spinal cord spontaneously forms a new intraspinal circuit in adult rats. Nat Neurosci 2004; 7: 269-77.

[24] Lapointe NP, Ung R-V, Guertin PA. Plasticity in sublesionally located neurons following spinal cord injury. J Neurophysiol 2007; 98: 2497-500.


[25] Harkema SJ, Hurley SL, Patel UK, Requejo PS, Dobkin BH, Edg-erton VR. Human lumbosacral spinal cord interprets loading during stepping. J Neurophysiol 1997; 77: 797-811.

[26] Knikou M, Conway BA. Modulation of soleus H-reflex following ipsilateral mechanical loading of the sole of the foot in normal and complete spinal cord injured humans. Neurosci Lett 2001; 303: 107-10.

[27] Knikou M, Rymer WZ. Hip angle induced modulation of H reflex amplitude, latency and duration in spinal cord injured humans. Clin Neurophysiol 2002; 113: 1698-708.

[28] Dimitrijevic MR, Gerasimenko Y, Pinter MM. Evidence for a spinal central pattern generator in humans. Ann N Y Acad Sci 1998; 860: 360-76.

[29] Minassian K, Jilge B, Rattay F, et al. Stepping-like movements in humans with complete spinal cord injury induced by epidural stimulation of the lumbar cord: electromyographic study of com-pound muscle action potentials. Spinal Cord 2004; 42: 401-16.

[30] Minassian K, Persy I, Rattay F, Pinter MM, Kern H, Dimitrijevic MR. Human lumbar cord circuitries can be activated by extrinsic tonic input to generate locomotor-like activity. Hum Mov Sci 2007; 26: 275-95.

[31] Guertin PA. The mammalian central pattern generator for locomo-tion. Brain Res Rev 2009; 62: 45-56.

[32] Wainberg M, Barbeau H, Gauthier S. Quantitative assessment of the effect of cyproheptadine on spastic paretic gait: a preliminary study. J Neurol 1986; 233: 311-4.

[33] Wernig A, Müller S. Laufband locomotion with body weight sup-port improved walking in persons with severe spinal cord injuries. Paraplegia 1992; 30: 229-38.

[34] Beres-Jones JA, Harkema SJ. The human spinal cord interprets velocity-dependent afferent input during stepping. Brain 2004; 127: 2232-46.

[35] Dietz V, Müller R, Colombo G. Locomotor activity in spinal man: significance of afferent input from joint and load receptors. Brain 2002; 125: 2626-34.

[36] Knikou M, Angeli CA, Ferreira CK, Harkema SJ. Soleus H-reflex modulation during body weight support treadmill walking in spinal cord intact and injured subjects. Exp Brain Res 2009; 193: 397-407.

[37] Knikou M. Functional reorganization of soleus H-reflex modula-tion during stepping after robotic-assisted step training in people with complete and incomplete spinal cord injury. Exp Brain Res 2013; 228: 279-96.

[38] Shapkova E. Spinal locomotor capability revealed by electrical stimulation of the lumbar enlargement in paraplegic patients. In: Latash M, Levin M, Eds. Progress in motor control, vol 3: effect of age, disorder, and rehabilitation. Champaign, IL: Human Kinetics 2004; pp. 253-89.

[39] Harkema S, Gerasimenko Y, Hodes J, et al. Effect of epidural stimulation of the lumbosacral spinal cord on voluntary movement, standing, and assisted stepping after motor complete paraplegia: a case study. Lancet 2011; 377: 1938-47.

[40] Minassian K, Hofstoetter US. Spinal Cord Stimulation and Aug-mentative Control Strategies for Leg Movement after Spinal Pa-ralysis in Humans. CNS Neurosci Ther 2016; 22: 262-70.

[41] Minassian K, McKay WB, Binder H, Hofstoetter US. Targeting Lumbar Spinal Neural Circuitry by Epidural Stimulation to Restore Motor Function After Spinal Cord Injury. Neurotherapeutics 2016; 13: 284-94.

[42] Coburn B. A theoretical study of epidural electrical stimulation of the spinal cord--Part II: Effects on long myelinated fibers. IEEE Trans Biomed Eng 1985; 32: 978-86.

[43] Coburn B, Sin WK. A theoretical study of epidural electrical stimu-lation of the spinal cord--Part I: Finite element analysis of stimulus fields. IEEE Trans Biomed Eng 1985; 32: 971-7.

[44] Struijk JJ, Holsheimer J, Boom HBK. Excitation of dorsal root fibers in spinal cord stimulation: a theoretical study. IEEE Trans Biomed Eng 1993; 40: 632-9.

[45] Rattay F, Minassian K, Dimitrijevic MR. Epidural electrical stimu-lation of posterior structures of the human lumbosacral cord: 2. quantitative analysis by computer modeling. Spinal Cord 2000; 38: 473-89.

[46] Sayenko DG, Angeli C, Harkema SJ, Edgerton VR, Gerasimenko YP. Neuromodulation of evoked muscle potentials induced by epidural spinal-cord stimulation in paralyzed individuals. J Neuro-physiol 2014; 111: 1088-99.

[47] Hunter JP, Ashby P. Segmental effects of epidural spinal cord stimulation in humans. J Physiol 1994; 474: 407-19.

[48] Holsheimer J. Which Neuronal Elements are Activated Directly by Spinal Cord Stimulation. Neuromodulation 2002; 5: 25-31.

[49] Radhakrishna M, Steuer I, Prince F, et al. Double-blind, placebo-cotrolled, randomized phase I/IIa study with buspi-rone/levodopa/carbidopa (SpinalonTM) in subjects with complete AIS A or motor-complete AIS B spinal cord injury. Curr Pharm Des; in press.

[50] Gossard JP, Brownstone RM, Barajon I, Hultborn H. Transmission in a locomotor-related group Ib pathway from hindlimb extensor muscles in the cat. Exp Brain Res 1994; 98: 213-28.

[51] Hultborn H, Conway BA, Gossard JP, et al. How do we approach the locomotor network in the mammalian spinal cord? Ann N Y Acad Sci 1998; 860: 70-82.

[52] Laporte Y, Lloyd DPC. Nature and significance of the reflex con-nections established by large afferent fibers of muscular origin. Am J Physiol 1952; 169: 609-21.

[53] Eccles JC, Eccles RM, Lundberg A. Synaptic actions on motoneu-rones caused by impulses in Golgi tendon organ afferents. J Physiol 1957; 138: 227-52.

[54] Conway BA, Hultborn H, Kiehn O. Proprioceptive input resets central locomotor rhythm in the spinal cat. Exp Brain Res 1987; 68: 643-56.

[55] Angel MJ, Jankowska E, McCrea DA. Candidate interneurones mediating group I disynaptic EPSPs in extensor motoneurones dur-ing fictive locomotion in the cat. J Physiol 2005; 563: 597-610.

[56] Rybak IA, Stecina K, Shevtsova NA, McCrea DA. Modelling spinal circuitry involved in locomotor pattern generation: insights from the effects of afferent stimulation. J Physiol 2006; 577: 641-58.

[57] Schomburg ED, Petersen N, Barajon I, Hultborn H. Flexor reflex afferents reset the step cycle during fictive locomotion in the cat. Exp Brain Res 1998; 122: 339-50.

[58] Jankowska E, Jukes MG, Lund S, Lundberg A. The effect of DOPA on the spinal cord. 5. Reciprocal organization of pathways transmitting excitatory action to alpha motoneurones of flexors and extensors. Acta Physiol Scand 1967; 70: 369-88.

[59] Jankowska E, Jukes MG, Lund S, Lundberg A. The effect of DOPA on the spinal cord. 6. Half-centre organization of interneu-rones transmitting effects from the flexor reflex afferents. Acta Physiol Scand 1967; 70: 389-402.

[60] Lundberg A. Multisensory control of spinal reflex pathways. Prog Brain Res 1979; 50: 11-28.

[61] Frigon A, Gossard J-P. Evidence for specialized rhythm-generating mechanisms in the adult mammalian spinal cord. J Neurosci 2010; 30: 7061-71.

[62] Andersson O, Grillner S. Peripheral control of the cat’s step cycle. II. Entrainment of the central pattern generators for locomotion by sinusoidal hip movements during “fictive locomotion”. Acta Physiol Scand 1983; 118: 229-39.

[63] Barbeau H, Rossignol S. Recovery of locomotion after chronic spinalization in the adult cat. Brain Res 1987; 412: 84-95.

[64] Lovely RG, Gregor RJ, Roy RR, Edgerton VR. Effects of training on the recovery of full-weight-bearing stepping in the adult spinal cat. Exp Neurol 1986; 92: 421-35.

[65] Bélanger M, Drew T, Rossignol S. Spinal locomotion: a compari-son of the kinematics and the electromyographic activity in the same animal before and after spinalization. Acta Biol Hung 1988; 39: 151-4.

[66] Rossignol S. Locomotion and its recovery after spinal injury. Curr Opin Neurobiol 2000; 10: 708-16.

[67] de Guzman CP, Roy RR, Hodgson JA, Edgerton VR. Coordination of motor pools controlling the ankle musculature in adult spinal cats during treadmill walking. Brain Res 1991; 555: 202-14.

[68] de Leon RD, Tamaki H, Hodgson JA, Roy RR, Edgerton VR. Hindlimb locomotor and postural training modulates glycinergic inhibition in the spinal cord of the adult spinal cat. J Neurophysiol 1999; 82: 359-69.

[69] Lovely RG, Gregor RJ, Roy RR, Edgerton VR. Weight-bearing hindlimb stepping in treadmill-exercised adult spinal cats. Brain Res 1990; 514: 206-18.

[70] Alluin O, Delivet-Mongrain H, Rossignol S. Inducing hindlimb locomotor recovery in adult rat after complete thoracic spinal cord section using repeated treadmill training with perineal stimulation only. J Neurophysiol 2015; 114: 1931-46.


[71] Timoszyk WK, De Leon RD, London N, Roy RR, Edgerton VR, Reinkensmeyer DJ. The rat lumbosacral spinal cord adapts to ro-botic loading applied during stance. J Neurophysiol 2002; 88: 3108-17.

[72] Edgerton VR, Courtine G, Gerasimenko YP, et al. Training loco-motor networks. Brain Res Rev 2008; 57: 241-54.

[73] Courtine G, Gerasimenko Y, van den Brand R, et al. Transforma-tion of nonfunctional spinal circuits into functional states after the loss of brain input. Nat Neurosci 2009; 12: 1333-42.

[74] Steuer I, Rouleau P, Guertin PA. Pharmacological approaches to chronic spinal cord injury. Curr Pharm Des 2013; 19: 4423-36.

[75] Guertin PA. Recovery of locomotor function with combinatory drug treatments designed to synergistically activate specific neu-ronal networks. Curr Med Chem 2009; 16: 1366-71.

[76] McEwen ML, Van Hartesveldt C, Stehouwer DJ. L-DOPA and quipazine elicit air-stepping in neonatal rats with spinal cord tran-sections. Behav Neurosci 1997; 111: 825-33.

[77] McCrea AE, Stehouwer DJ, Van Hartesveldt C. Dopamine D1 and D2 antagonists block L-DOPA-induced air-stepping in decerebrate neonatal rats. Brain Res Dev Brain Res 1997; 100: 130-2.

[78] Guertin PA. Synergistic activation of the central pattern generator for locomotion by l-beta-3,4-dihydroxyphenylalanine and quipaz-ine in adult paraplegic mice. Neurosci Lett 2004; 358: 71-4.

[79] Guertin PA. Role of NMDA receptor activation in serotonin ago-nist-induced air-stepping in paraplegic mice. Spinal Cord 2004; 42: 185-90.

[80] Lapointe NP, Rouleau P, Ung R-V, Guertin PA. Specific role of dopamine D1 receptors in spinal network activation and rhythmic movement induction in vertebrates. J Physiol 2009; 587: 1499-511.

[81] Lapointe NP, Guertin PA. Synergistic effects of D1/5 and 5-HT1A/7 receptor agonists on locomotor movement induction in complete spinal cord-transected mice. J Neurophysiol 2008; 100: 160-8.

[82] Guertin PA, Ung R-V, Rouleau P. Oral administration of a tri-therapy for central pattern generator activation in paraplegic mice: proof-of-concept of efficacy. Biotechnol J 2010; 5: 421-6.

[83] Guertin PA, Ung R-V, Rouleau P, Steuer I. Effects on locomotion, muscle, bone, and blood induced by a combination therapy eliciting weight-bearing stepping in nonassisted spinal cord-transected mice. Neurorehabil Neural Repair 2011; 25: 234-42.

[84] Ung R-V, Rouleau P, Guertin PA. Functional and physiological effects of treadmill training induced by buspirone, carbidopa, and L-DOPA in clenbuterol-treated paraplegic mice. Neurorehabil Neu-ral Repair 2012; 26: 385-94.

[85] Kiehn O, Kjaerulff O. Spatiotemporal characteristics of 5-HT and dopamine-induced rhythmic hindlimb activity in the in vitro neona-tal rat. J Neurophysiol 1996; 75: 1472-82.

[86] Jiang Z, Carlin KP, Brownstone RM. An in vitro functionally ma-ture mouse spinal cord preparation for the study of spinal motor networks. Brain Res 1999; 816: 493-9.

[87] Dietz V, Fouad K. Restoration of sensorimotor functions after spinal cord injury. Brain 2014; 137: 654-67.

[88] Harkema SJ. Plasticity of interneuronal networks of the function-ally isolated human spinal cord. Brain Res Rev 2008; 57: 255-64.

[89] Dobkin BH, Harkema S, Requejo P, Edgerton VR. Modulation of locomotor-like EMG activity in subjects with complete and incom-plete spinal cord injury. J Neurol Rehabil 1995; 9: 183-90.

[90] Aoyagi Y, Stein RB, Branner A, Pearson KG, Normann RA. Capa-bilities of a penetrating microelectrode array for recording single units in dorsal root ganglia of the cat. J Neurosci Methods 2003; 128: 9-20.

[91] Holinski BJ, Everaert DG, Mushahwar VK, Stein RB. Real-time control of walking using recordings from dorsal root ganglia. J Neural Eng 2013; 10: 056008.

[92] Stewart JE, Barbeau H, Gauthier S. Modulation of locomotor pat-terns and spasticity with clonidine in spinal cord injured patients. Can J Neurol Sci 1991; 18: 321-32.

[93] MacKay-Lyons M. Central pattern generation of locomotion: a review of the evidence. Phys Ther 2002; 82: 69-83.

[94] Rossignol IS, Barbeau H. New approaches to locomotor rehabilita-tion in spinal cord injury. Ann Neurol 1995; 37: 555-6.

[95] Nielsen JB. Human Spinal Motor Control. Annu Rev Neurosci 2016; 39: 81-101.

[96] Pearson K, Gordon J. Spinal Reflexes. In: Kandel E, Schwartz J, Jessell T, Siegelbaum S, Hudspeth A, Eds. Principles of neural sci-ence. New York: McGraw-Hill 2013; pp. 790-811.

[97] Hagbarth KE, Wallin G, Löfstedt L. Muscle spindle responses to stretch in normal and spastic subjects. Scand J Rehabil Med 1973; 5: 156-9.

[98] Wilson LR, Gandevia SC, Inglis JT, Gracies J, Burke D. Muscle spindle activity in the affected upper limb after a unilateral stroke. Brain 1999; 122: 2079-88.

[99] Macefield VG. Discharge rates and discharge variability of muscle spindle afferents in human chronic spinal cord injury. Clin Neuro-physiol 2013; 124: 114-9.

[100] Vallbo AB. Muscle spindle response at the onset of isometric vol-untary contractions in man. Time difference between fusimotor and skeletomotor effects. J Physiol 1971; 218: 405-31.

[101] Burke D, Hagbarth KE, Löfstedt L. Muscle spindle activity in man during shortening and lengthening contractions. J Physiol 1978; 277: 131-42.

[102] Prochazka A, Gorassini M. Models of ensemble firing of muscle spindle afferents recorded during normal locomotion in cats. J Physiol 1998; 507: 277-91.

[103] Prochazka A, Gorassini M. Ensemble firing of muscle afferents recorded during normal locomotion in cats. J Physiol 1998; 507: 293-304.

[104] Frank K, Fuortes M. Presynaptic and postsynaptic inhibition of monosynaptic reflexes. Fed Proc 1957; 16: 39-40.

[105] Eccles JC, Eccles RM, Magni F. Central inhibitory action attribut-able to presynaptic depolarization produced by muscle afferent vol-leys. J Physiol 1961; 159: 147-66.

[106] Eccles JC, Schmidt RF, Willis WD. Presynaptic inhibition of the spinal monosynaptic reflex pathway. J Physiol 1962; 161: 282-97.

[107] Rudomin P, Schmidt RF. Presynaptic inhibition in the vertebrate spinal cord revisited. Exp Brain Res 1999; 129: 1-37.

[108] Pierrot-Deseilligny E, Burke D. The Circuitry of the Human Spinal Cord. 2nd ed. Cambridge: Cambridge University Press 2012.

[109] Brooke JD, Cheng J, Misiaszek JE, Lafferty K. Amplitude modula-tion of the soleus H reflex in the human during active and passive stepping movements. J Neurophysiol 1995; 73: 102-11.

[110] Mummidisetty CK, Smith AC, Knikou M. Modulation of recipro-cal and presynaptic inhibition during robotic-assisted stepping in humans. Clin Neurophysiol 2013; 124: 557-64.

[111] Dueñas SH, Rudomin P. Excitability changes of ankle extensor group Ia and Ib fibers during fictive locomotion in the cat. Exp Brain Res 1988; 70: 15-25.

[112] Gossard JP, Cabelguen JM, Rossignol S. An intracellular study of muscle primary afferents during fictive locomotion in the cat. J Neurophysiol 1991; 65: 914-26.

[113] Gosgnach S, Quevedo J, Fedirchuk B, McCrea DA. Depression of group Ia monosynaptic EPSPs in cat hindlimb motoneurones dur-ing fictive locomotion. J Physiol 2000; 526: 639-52.

[114] Ménard A, Leblond H, Gossard J-P. Sensory integration in presyn-aptic inhibitory pathways during fictive locomotion in the cat. J Neurophysiol 2002; 88: 163-71.

[115] Hultborn H, Meunier S, Pierrot-Deseilligny E, Shindo M. Changes in presynaptic inhibition of Ia fibres at the onset of voluntary con-traction in man. J Physiol 1987; 389: 757-72.

[116] Knikou M. The H-reflex as a probe: pathways and pitfalls. J Neu-rosci Methods 2008; 171: 1-12.

[117] Katz R, Meunier S, Pierrot-Deseilligny E. Changes in presynaptic inhibition of Ia fibres in man while standing. Brain  1988; 111: 417-37.

[118] Edamura M, Yang JF, Stein RB. Factors that determine the magni-tude and time course of human H-reflexes in locomotion. J Neuro-sci 1991; 11: 420-7.

[119] Calancie B, Broton JG, Klose KJ, Traad M, Difini J, Ayyar DR. Evidence that alterations in presynaptic inhibition contribute to segmental hypo- and hyperexcitability after spinal cord injury in man. Electroencephalogr Clin Neurophysiol 1993; 89: 177-86.

[120] Faist M, Mazevet D, Dietz V, Pierrot-Deseilligny E. A quantitative assessment of presynaptic inhibition of Ia afferents in spastics. Dif-ferences in hemiplegics and paraplegics. Brain 1994; 117: 1449-55.

[121] Stein RB. Presynaptic inhibition in humans. Prog Neurobiol 1995; 47: 533-44.

[122] Knikou M, Mummidisetty CK. Locomotor training improves pre-motoneuronal control after chronic spinal cord injury. J Neuro-physiol 2014; 111: 2264-75.

[123] Knikou M, Hajela N, Mummidisetty CK, Xiao M, Smith AC. So-leus H-reflex phase-dependent modulation is preserved during


stepping within a robotic exoskeleton. Clin Neurophysiol 2011; 122: 1396-404.

[124] Smith AC, Rymer WZ, Knikou M. Locomotor training modifies soleus monosynaptic motoneuron responses in human spinal cord injury. Exp Brain Res 2015; 233: 89-103.

[125] Capaday C, Stein RB. Difference in the amplitude of the human soleus H reflex during walking and running. J Physiol 1987; 392: 513-22.

[126] Yang JF, Fung J, Edamura M, Blunt R, Stein RB, Barbeau H. H-reflex modulation during walking in spastic paretic subjects. Can J Neurol Sci 1991; 18: 443-52.

[127] Duysens J, Clarac F, Cruse H. Load-regulating mechanisms in gait and posture: comparative aspects. Physiol Rev 2000; 80: 83-133.

[128] Yang JF, Stein RB, James KB. Contribution of peripheral afferents to the activation of the soleus muscle during walking in humans. Exp Brain Res 1991; 87: 679-87.

[129] Sinkjaer T, Andersen JB, Ladouceur M, Christensen LO, Nielsen JB. Major role for sensory feedback in soleus EMG activity in the stance phase of walking in man. J Physiol 2000; 523: 817-27.

[130] Grey MJ, Mazzaro N, Nielsen JB, Sinkjær T. Ankle extensor pro-prioceptors contribute to the enhancement of the soleus EMG dur-ing the stance phase of human walking. Can J Physiol Pharmacol 2004; 82: 610-6.

[131] Grey MJ, Nielsen JB, Mazzaro N, Sinkjaer T. Positive force feed-back in human walking. J Physiol 2007; 581: 99-105.

[132] af Klint R, Mazzaro N, Nielsen JB, Sinkjaer T, Grey MJ. Load rather than length sensitive feedback contributes to soleus muscle activity during human treadmill walking. J Neurophysiol 2010; 103: 2747-56.

[133] Wernig A. No dawn yet of a new age in spinal cord rehabilitation. Brain 2015; 138: e362.

[134] Dietz V. Spinal cord pattern generators for locomotion. Clin Neu-rophysiol 2003; 114: 1379-89.

[135] Dietz V. Body weight supported gait training: from laboratory to clinical setting. Brain Res Bull 2008; 76: 459-63.

[136] Dietz V, Harkema SJ. Locomotor activity in spinal cord-injured persons. J Appl Physiol (1985) 2004; 96: 1954-60.

[137] Dietz V, Wirz M, Curt A, Colombo G. Locomotor pattern in para-plegic patients: training effects and recovery of spinal cord func-tion. Spinal Cord 1998; 36: 380-90.

[138] Dietz V, Michel J. Human bipeds use quadrupedal coordination during locomotion. Ann N Y Acad Sci 2009; 1164: 97-103.

[139] Dietz V. Improving outcome of sensorimotor functions after trau-matic spinal cord injury. F1000Res 2016; 5: 1018.

[140] Lünenburger L, Bolliger M, Czell D, Müller R, Dietz V. Modula-tion of locomotor activity in complete spinal cord injury. Exp Brain Res 2006; 174: 638-46.

[141] Minassian K, Hofstoetter US, Danner SM, et al. Spinal Rhythm Generation by Step-Induced Feedback and Transcutaneous Poste-rior Root Stimulation in Complete Spinal Cord-Injured Individuals. Neurorehabil Neural Repair 2016; 30: 233-43.

[142] Hubli M, Dietz V. The physiological basis of neurorehabilitation--locomotor training after spinal cord injury. J Neuroeng Rehabil 2013; 10: 5.

[143] Dietz V, Colombo G, Jensen L, Baumgartner L. Locomotor capac-ity of spinal cord in paraplegic patients. Ann Neurol 1995; 37: 574-82.

[144] Forrest GF, Sisto SA, Barbeau H, et al. Neuromotor and muscu-loskeletal responses to locomotor training for an individual with chronic motor complete AIS-B spinal cord injury. J Spinal Cord Med 2008; 31: 509-21.

[145] Knikou M, Mummidisetty CK. Reduced reciprocal inhibition dur-ing assisted stepping in human spinal cord injury. Exp Neurol 2011; 231: 104-12.

[146] Minassian K, Persy I, Rattay F, Dimitrijevic MR. Effect of periph-eral afferent and central afferent input to the human lumbar spinal cord isolated from brain control. Biocybern Biomed Eng 2005; 25: 11-29.

[147] Angeli CA, Edgerton VR, Gerasimenko YP, Harkema SJ. Altering spinal cord excitability enables voluntary movements after chronic complete paralysis in humans. Brain 2014; 137: 1394-409.

[148] Kirshblum SC, Burns SP, Biering-Sorensen F, et al. International standards for neurological classification of spinal cord injury (re-vised 2011). J Spinal Cord Med 2011; 34: 535-46.

[149] Kakulas BA. Pathology of spinal injuries. Cent Nerv Syst Trauma 1984; 1: 117-29.

[150] Kakulas A. The applied neurobiology of human spinal cord injury: a review. Paraplegia 1988; 26: 371-9.

[151] Bunge RP, Puckett WR, Becerra JL, Marcillo A, Quencer RM. Observations on the pathology of human spinal cord injury. A re-view and classification of 22 new cases with details from a case of chronic cord compression with extensive focal demyelination. Adv Neurol 1993; 59: 75-89.

[152] Nathan PW, Smith MC. Fasciculi proprii of the spinal cord in man. Brain 1959; 82: 610-68.

[153] Faganel J, Dimitrijevic MR. Study of propriospinal interneuron system in man. Cutaneous exteroceptive conditioning of stretch re-flexes. J Neurol Sci 1982; 56: 155-72.

[154] Pohland M, Glumm J. Propriospinal interneurons in the spotlight for anatomical and functional recovery after spinal cord injury. Neural Regen Res 2015; 10: 1737-8.

[155] Dimitrijevic MR, Dimitrijevic MM, Faganel J, Sherwood AM. Suprasegmentally induced motor unit activity in paralyzed muscles of patients with established spinal cord injury. Ann Neurol 1984; 16: 216-21.

[156] Kakulas BA, Kaelan C. The neuropathological foundations for the restorative neurology of spinal cord injury. Clin Neurol Neurosurg 2015; 129 Suppl: S1-7.

[157] Parker D. Descending interactions with spinal cord networks: a time to build? J Physiol 2009; 587: 4761.

[158] Cook AW, Weinstein SP. Chronic dorsal column stimulation in multiple sclerosis. Preliminary report. N Y State J Med 1973; 73: 2868-72.

[159] Dimitrijevic MM, Dimitrijevic MR, Illis LS, Nakajima K, Sharkey PC, Sherwood AM. Spinal cord stimulation for the control of spas-ticity in patients with chronic spinal cord injury: I. Clinical obser-vations. Cent Nerv Syst Trauma 1986; 3: 129-44.

[160] Pinter MM, Gerstenbrand F, Dimitrijevic MR. Epidural electrical stimulation of posterior structures of the human lumbosacral cord: 3. Control Of spasticity. Spinal Cord 2000; 38: 524-31.

[161] Ladenbauer J, Minassian K, Hofstoetter US, Dimitrijevic MR, Rattay F. Stimulation of the human lumbar spinal cord with im-planted and surface electrodes: a computer simulation study. IEEE Trans Neural Syst Rehabil Eng 2010; 18: 637-45.

[162] Danner SM, Hofstoetter US, Freundl B, et al. Human spinal loco-motor control is based on flexibly organized burst generators. Brain 2015; 138: 577-88.

[163] Hofstoetter US, Danner SM, Freundl B, et al. Periodic modulation of repetitively elicited monosynaptic reflexes of the human lumbo-sacral spinal cord. J Neurophysiol 2015; 114: 400-10.

[164] Iwahara T, Atsuta Y, Garcia-Rill E, Skinner RD. Spinal cord stimu-lation-induced locomotion in the adult cat. Brain Res Bull 1992; 28: 99-105.

[165] Barthélemy D, Leblond H, Rossignol S. Characteristics and mechanisms of locomotion induced by intraspinal microstimulation and dorsal root stimulation in spinal cats. J Neurophysiol 2007; 97: 1986-2000.

[166] Gerasimenko Y, Daniel O, Regnaux J, Combeaud M, Bussel B. Mechanisms of locomotor activity generation under epidural spinal cord stimulation. In: Dengler R, Kossev A, Eds. Sensorimotor con-trol. Washington, DC: IOS Press 2001; pp. 164-71.

[167] Gerasimenko YP, Makarovskii AN, Nikitin OA. Control of loco-motor activity in humans and animals in the absence of supraspinal influences. Neurosci Behav Physiol 2002; 32: 417-23.

[168] Jilge B, Minassian K, Rattay F, et al. Initiating extension of the lower limbs in subjects with complete spinal cord injury by epidural lumbar cord stimulation. Exp Brain Res 2004; 154: 308-26.

[169] Lavrov I, Dy CJ, Fong AJ, et al. Epidural stimulation induced modulation of spinal locomotor networks in adult spinal rats. J Neurosci 2008; 28: 6022-9.

[170] Wenger N, Moraud EM, Raspopovic S, et al. Closed-loop neuro-modulation of spinal sensorimotor circuits controls refined locomo-tion after complete spinal cord injury. Sci Transl Med 2014; 6: 255ra133.

[171] Yingling CD, Hosobuchi Y. Use of antidromic evoked potentials in placement of dorsal cord disc electrodes. Appl Neurophysiol 1986; 49: 36-41.

[172] Buonocore M, Bonezzi C, Barolat G. Neurophysiological evidence of antidromic activation of large myelinated fibres in lower limbs during spinal cord stimulation. Spine 2008; 33: E90-3.


[173] Miller S, Scott PD. The spinal locomotor generator. Exp Brain Res 1977; 30: 387-403.

[174] Gerasimenko YP, Lavrov IA, Courtine G, et al. Spinal cord re-flexes induced by epidural spinal cord stimulation in normal awake rats. J Neurosci Methods 2006; 157: 253-63.

[175] Commissiong JW, Gentleman S, Neff NH. Spinal cord dopaminer-gic neurons: evidence for an uncrossed nigrospinal pathway. Neu-ropharmacology 1979; 18: 565-8.

[176] Holstege JC, Van Dijken H, Buijs RM, Goedknegt H, Gosens T, Bongers CM. Distribution of dopamine immunoreactivity in the rat, cat and monkey spinal cord. J Comp Neurol 1996; 376: 631-52.

[177] Perrin FE, Gerber YN, Teigell M, et al. Anatomical study of sero-tonergic innervation and 5-HT(1A) receptor in the human spinal cord. Cell Death Dis 2011; 2: e218.

[178] Zhu H, Clemens S, Sawchuk M, Hochman S. Expression and dis-tribution of all dopamine receptor subtypes (D(1)-D(5)) in the mouse lumbar spinal cord: a real-time polymerase chain reaction and non-autoradiographic in situ hybridization study. Neuroscience 2007; 149: 885-97.

[179] Sharples SA, Koblinger K, Humphreys JM, Whelan PJ. Dopamine: a parallel pathway for the modulation of spinal locomotor net-works. Front Neural Circuits 2014; 8: 55.

[180] Jaeger CB, Teitelman G, Joh TH, Albert VR, Park DH, Reis DJ. Some neurons of the rat central nervous system contain aromatic-

L-amino-acid decarboxylase but not monoamines. Science 1983; 219: 1233-5.

[181] Li Y, Li L, Stephens MJ, et al. Synthesis, transport, and metabo-lism of serotonin formed from exogenously applied 5-HTP after spinal cord injury in rats. J Neurophysiol 2014; 111: 145-63.

[182] Wienecke J, Ren L-Q, Hultborn H, et al. Spinal cord injury enables aromatic L-amino acid decarboxylase cells to synthesize mono-amines. J Neurosci  2014; 34: 11984-2000.

[183] Ren L-Q, Wienecke J, Hultborn H, Zhang M. Production of Dopa-mine by Aromatic l-Amino Acid Decarboxylase Cells after Spinal Cord Injury. J Neurotrauma 2016; 33: 1150-60.

[184] Guertin PA, Hounsgaard J. L-type calcium channels but not N-methyl-D-aspartate receptor channels mediate rhythmic activity in-duced by cholinergic agonist in motoneurons from turtle spinal cord slices. Neurosci Lett 1999; 261: 81-4.

[185] Wang D, Grillner S, Wallén P. 5-HT and dopamine modulates CaV1.3 calcium channels involved in postinhibitory rebound in the spinal network for locomotion in lamprey. J Neurophysiol 2011; 105: 1212-24.

[186] Kleedorfer B, Lees AJ, Stern GM. Buspirone in the treatment of levodopa induced dyskinesias. J Neurol Neurosurg Psychiatry 1991; 54: 376-7.

Probing the Human Spinal Locomotor Circuits by Phasic Step ...klab4recovery.org › wp-content ›...

Documents

Transcript of Probing the Human Spinal Locomotor Circuits by Phasic Step ...klab4recovery.org › wp-content ›...