Cerebellar compartments for the processing of kinematic and kinetic information related to hindlimb stepping.

We previously showed that proprioceptive sensory input from the hindlimbs to the anterior cerebellar cortex of the cat may not be simply organized with respect to a body map, but it may also be distributed to multiple discrete functional areas extending beyond classical body map boundaries. With passive hindlimb stepping movements, cerebellar activity was shown to relate to whole limb kinematics as does the activity of dorsal spinocerebellar tract (DSCT) neurons. For DSCT activity, whole limb kinematics provides a solid functional framework within which information about limb forces, such as those generated during active stepping, may also be embedded. In this study, we investigated this idea for the spinocerebellar cortex activity by examining the activity of cerebellar cortical neurons during both passive bipedal hindlimb stepping and active stepping on a treadmill. Our results showed a functional compartmentalization of cerebellar responses to hindlimb stepping movements depending on the two types of stepping and strong relationships between neural activities and limb axis kinematics during both. In fact, responses to passive and active stepping were generally different, but in both cases their waveforms were related strongly to the limb axis kinematics. That is, the different stepping conditions modified the kinematics representation without producing different components in the response waveforms. In sum, cerebellar activity was consistent with a global kinematics framework serving as a basis upon which detailed information about limb mechanics and/or about individual limb segments might be imposed.

Plasticity and modular control of locomotor patterns in neurological disorders with motor deficits.

Human locomotor movements exhibit considerable variability and are highly complex in terms of both neural activation and biomechanical output. The building blocks with which the central nervous system constructs these motor patterns can be preserved in patients with various sensory-motor disorders. In particular, several studies highlighted a modular burst-like organization of the muscle activity. Here we review and discuss this issue with a particular emphasis on the various examples of adaptation of locomotor patterns in patients (with large fiber neuropathy, amputees, stroke and spinal cord injury). The results highlight plasticity and different solutions to reorganize muscle patterns in both peripheral and central nervous system lesions. The findings are discussed in a general context of compensatory gait mechanisms, spatiotemporal architecture and modularity of the locomotor program.

The organization of cortical activity in the anterior lobe of the cat cerebellum during hindlimb stepping.

We recorded from over 280 single cortical neurons throughout the medial anterior lobe of the cat cerebellum during passive movements of the hindlimbs resembling stepping on a moving treadmill. We used three stepping patterns, unilateral stepping of either the ipsilateral or contralateral leg and bipedal stepping in an alternating gait pattern. We found that over 60% of the neurons, mostly Purkinje cells, responded to stepping of one or both legs, and over 40% to more than one type of stepping pattern. Responsive cells were distributed throughout the five anterior lobules, and the highest concentration was found in traditional hindlimb areas in lobules 2 and 3. A comparison of response waveforms showed that they are similar for neighboring cells in many parts of the cerebellar cortex, and they tend to form local blob-like groupings. Response patterns, i.e., relationship among responses to each stepping type, tended to be similar within a local group. The groupings extend further in the parasagittal dimension (up to about a third of a lobule) than in the transverse dimension (about 1 mm), and they may form functional modules. A principal component analysis also showed that the responses were composed of a four basis waveforms (principal components) that explained about 80% of the response waveform variance that were nearly identical to those derived from dorsal spinocerebellar tract (DSCT) responses to similar stepping movements. We reconstructed the locations of the recorded neurons on a 2D map of the cerebellar cortex showing the spatial distribution of responsive cells according to their response characteristics. We propose, on the basis of these results, that the sensory input to the cerebellum from the hindlimbs is distributed to multiple zones that may each contribute to a different component of cerebellar function.

A novel approach to mechanical foot stimulation during human locomotion under body weight support.

Input from the foot plays an essential part in perceiving support surfaces and determining kinematic events in human walking. To simulate adequate tactile pressure inputs under body weight support (BWS) conditions that represent an effective form of locomotion training, we here developed a new method of phasic mechanical foot stimulation using light-weight pneumatic insoles placed inside the shoes (under the heel and metatarsus). To test the system, we asked healthy participants to walk on a treadmill with different levels of BWS. The pressure under the stimulated areas of the feet and subjective sensations were higher at high levels of BWS and when applied to the ball and toes rather than heels. Foot stimulation did not disturb significantly the normal motor pattern, and in all participants we evoked a reliable step-synchronized triggering of stimuli for each leg separately. This approach has been performed in a general framework looking for "afferent templates" of human locomotion that could be used for functional sensory stimulation. The proposed technique can be used to imitate or partially restore surrogate contact forces under body weight support conditions.

Distributed neural networks for controlling human locomotion: lessons from normal and SCI subjects.

The control of human locomotion engages various brain structures and numerous muscles. Even though the hypothetical central pattern generator (CPG) and sensory feedback can sustain the basic locomotor rhythm, the resultant motor output is highly adaptable and context-dependent. Indeed, while the temporal architecture of the locomotor output (basic EMG components) is relatively conserved across subjects and conditions, the spatial architecture (muscle activations) shows considerable non-linear changes with walking speed, level of body unloading or the direction of progression. Even so, leg kinematics are remarkably similar in all cases. Spinal cord injured (SCI) patients may learn new motor patterns with training rather than re-activate normal motor patterns, and such locomotor improvements may not transfer to untrained tasks. Redundancy in the neuromuscular system or malfunctioning of injured 'elements' may often result in motor equivalent compensatory solutions. Injured pathways can partially recover while uninjured pathways can augment or modify their activity. As a result, the reconstructed spatiotemporal maps of motor neuron activity in SCI patients might be quite different from those of healthy subjects but they nevertheless achieve nearly normal foot kinematics. Kinematics training may thus provide a more successful rehabilitation than training based on reconstructing normal muscle activation patterns. Taken together, recent data support the idea of plasticity and distributed networks for controlling human locomotion. A new generation of robotic devices takes advantage of this by providing the opportunity for patients to generate and correct limb movements rather than just adapting muscle activation to the fixed kinematic template imposed by a gait orthosis.

Spatiotemporal organization of alpha-motoneuron activity in the human spinal cord during different gaits and gait transitions.

Here we studied the spatiotemporal organization of motoneuron (MN) activity during different human gaits. We recorded the electromyographic (EMG) activity patterns in 32 ipsilateral limb and trunk muscles from normal subjects while running and walking on a treadmill (3-12 km/h). In addition, we recorded backward walking and skipping, a distinct human gait that comprises the features of both walking and running. We mapped the recorded EMG activity patterns onto the spinal cord in approximate rostrocaudal locations of the MN pools. The activation of MNs tends to occur in bursts and be segregated by spinal segment in a gait-specific manner. In particular, sacral and cervical activation timings were clearly gait-dependent. Swing-related activity constituted an appreciable fraction (> 30%) of the total MN activity of leg muscles. Locomoting at non-preferred speeds (running and walking at 5 and 9 km/h, respectively) showed clear differences relative to preferred speeds. Running at low speeds was characterized by wider sacral activation. Walking at high non-preferred speeds was accompanied by an 'atypical' locus of activation in the upper lumbar spinal cord during late stance and by a drastically increased activation of lumbosacral segments. The latter findings suggest that the optimal speed of gait transitions may be related to an optimal intensity of the total MN activity, in addition to other factors previously described. The results overall support the idea of flexibility and adaptability of spatiotemporal activity in the spinal circuitry with constraints on the temporal functional connectivity of hypothetical pulsatile burst generators.

On the origin of planar covariation of elevation angles during human locomotion.

Leg segment rotations in human walking covary, so that the three-dimensional trajectory of temporal changes in the elevation angles lies close to a plane. Recently the role of central versus biomechanical constraints on the kinematics control of human locomotion has been questioned. Here we show, based on both modeling and experimental data, that the planar law of intersegmental coordination is not a simple consequence of biomechanics. First, the full limb behavior in various locomotion modes (walking on inclined surface, staircase stepping, air-stepping, crouched walking, hopping) can be expressed as 2 degrees of freedom planar motion even though the orientation of the plane and pairwise segment angle correlations may differ substantially. Second, planar covariation is not an inevitable outcome of any locomotor movement. It can be systematically violated in some conditions (e.g., when stooping and grasping an object on the floor during walking or in toddlers at the onset of independent walking) or transferred into a simple linear relationship in others (e.g., during stepping in place). Finally, all three major limb segments contribute importantly to planar covariation and its characteristics resulting in a certain endpoint trajectory defined by the limb axis length and orientation. Recent advances in the neural control of movement support the hypothesis about central representation of kinematics components.

Cerebellar cortical activity in the cat anterior lobe during hindlimb stepping.

We recorded from over 250 single cortical neurons throughout the medial anterior lobe of the cat cerebellum during passive movements of the ipsilateral hindlimb resembling stepping on a moving treadmill. We applied three different quantitative analysis techniques to determine the extent of neuronal modulation that could be accounted for by the stepping movements. The analyses all indicated that up to half the recorded neurons in all five lobules responded to these passive hindlimb movements. We reconstructed the locations of the recorded neurons on a 2-D map of the cerebellar cortex in order to determine the spatial distribution of responsive cells. Cells that were located in the classical hindlimb projection areas of the anterior lobe (in lobules 2 and 3) were generally most responsive to the limb movement with activity patterns that generally had a linear relationship to hindlimb kinematics. Cells in lobules 4 and 5, considered as classical forelimb areas of the cerebellum, were also responsive. Although these cells tended to have noisier firing patterns, many were found to be modulated nevertheless by the hindlimb movements. We also found a clear demarcation between zones b and c, with a higher fraction of responsive cells in all lobules located in zone b.

Phase-specific sensory representations in spinocerebellar activity during stepping: evidence for a hybrid kinematic/kinetic framework.

The dorsal spinocerebellar tract (DSCT) provides a major mossy fiber input to the spinocerebellum, which plays a significant role in the control of posture and locomotion. Recent work from our laboratory has provided evidence that DSCT neurons encode a global representation of hindlimb mechanics during passive limb movements. The framework that most successfully accounts for passive DSCT behavior is kinematics-based having the coordinates of the limb axis, limb-axis length and orientation. Here we examined the responses of DSCT neurons in decerebrate cats as they walked on a moving treadmill and compared them with the responses passive step-like movements of the hindlimb produced manually. We found that DSCT responses to active locomotion were quantitatively different from the responses to kinematically similar passive limb movements on the treadmill. The differences could not be simply accounted for by the difference in limb-axis kinematics in the two conditions, nor could they be accounted for by new or different response components. Instead, differences could be attributed to an increased relative prominence of specific response components occurring during the stance phase of active stepping, which may reflect a difference in the behavior of the sensory receptors and/or of the DSCT circuitry during active stepping. We propose from these results that DSCT neurons encode two global aspects of limb mechanics that are also important in controlling locomotion at the spinal level, namely the orientation angle of the limb axis and limb loading. Although limb-axis length seemed to be an independent predictor of DSCT activity during passive limb movements, we argue that it is not independent of limb loading, which is likely to be proportional to limb length under passive conditions.

Motor patterns in human walking and running.

Despite distinct differences between walking and running, the two types of human locomotion are likely to be controlled by shared pattern-generating networks. However, the differences between their kinematics and kinetics imply that corresponding muscle activations may also be quite different. We examined the differences between walking and running by recording kinematics and electromyographic (EMG) activity in 32 ipsilateral limb and trunk muscles during human locomotion, and compared the effects of speed (3-12 km/h) and gait. We found that the timing of muscle activation was accounted for by five basic temporal activation components during running as we previously found for walking. Each component was loaded on similar sets of leg muscles in both gaits but generally on different sets of upper trunk and shoulder muscles. The major difference between walking and running was that one temporal component, occurring during stance, was shifted to an earlier phase in the step cycle during running. These muscle activation differences between gaits did not simply depend on locomotion speed as shown by recordings during each gait over the same range of speeds (5-9 km/h). The results are consistent with an organization of locomotion motor programs having two parts, one that organizes muscle activation during swing and another during stance and the transition to swing. The timing shift between walking and running reflects therefore the difference in the relative duration of the stance phase in the two gaits.

Spinal cord maps of spatiotemporal alpha-motoneuron activation in humans walking at different speeds.

Functional MRI (fMRI) imaging of motoneuron activity in the human spinal cord is still in its infancy, and it will remain difficult to apply to walking. Here we present a viable alternative for documenting the spatiotemporal maps of alpha-motorneuron (MN) activity in the human spinal cord during walking, similar to the method recently reported for the cat. We recorded EMG activity from 16 to 32 ipsilateral limb and trunk muscles in 13 healthy subjects walking on a treadmill at different speeds (1-7 km/h) and mapped the recorded patterns onto the spinal cord in approximate rostrocaudal locations of the motoneuron pools. This approach can provide information about pattern generator output during locomotion in terms of segmental control rather than in terms of individual muscle control. A striking feature we found is that nearly every spinal segment undergoes at least two cycles of activation in the step cycle, thus supporting the idea of half-center oscillators controlling MN activation at any segmental level. The resulting spatiotemporal map patterns seem highly stereotyped over the range of walking speeds studied, although there were also some systematic redistributions of MN activity with speed. Bursts of MN activity were either temporally aligned across several spinal segments or switched between different segments. For example, the center of mass of MN activity in the lumbosacral levels generally shifted from rostral to caudal positions in two cycles for each step, revealing four major activation foci: two in the upper lumbar segments and two in the sacral segments. The results are consistent with the presence of at least two and possibly more pattern generators controlling the activation of lumbosacral MNs.

Kinematic and non-kinematic signals transmitted to the cat cerebellum during passive treadmill stepping.

Previous work from this laboratory has shown that activity in the dorsal spinocerebellar tract (DSCT) relates strongly to global hindlimb kinematics variables during passive displacements of the hindlimb. A linear relationship to limb axis orientation and length variables accounts for most of the response variance for passive limb positioning and movement. Here we extend those observations to more natural movements by examining the information carried by the DSCT during passive stepping movements on a treadmill, and we compare it to information transmitted during passive robot-driven hindlimb movements. Using a principal component analysis approach, we found that a linear relationship between the responses and hindlimb kinematics was comparable across experimental conditions. We also observed systematic non-linearities in this relationship for both types of movement that could be attributed to events corresponding to the touch-down and lift-off phases of the movement. We concluded that proprioceptive information transmitted to the cerebellum by the DSCT during locomotion has at least two major components. One component is associated with limb kinematics (limb orientation) and may be more or less related to the metrics of the step (stride length, for example) or its velocity. The other component is associated with limb length and/or limb loading, and it may signal some aspect of limb stiffness.

Five basic muscle activation patterns account for muscle activity during human locomotion.

An electromyographic (EMG) activity pattern for individual muscles in the gait cycle exhibits a great deal of intersubject, intermuscle and context-dependent variability. Here we examined the issue of common underlying patterns by applying factor analysis to the set of EMG records obtained at different walking speeds and gravitational loads. To this end healthy subjects were asked to walk on a treadmill at speeds of 1, 2, 3 and 5 kmh(-1) as well as when 35-95% of the body weight was supported using a harness. We recorded from 12-16 ipsilateral leg and trunk muscles using both surface and intramuscular recording and determined the average, normalized EMG of each record for 10-15 consecutive step cycles. We identified five basic underlying factors or component waveforms that can account for about 90% of the total waveform variance across different muscles during normal gait. Furthermore, while activation patterns of individual muscles could vary dramatically with speed and gravitational load, both the limb kinematics and the basic EMG components displayed only limited changes. Thus, we found a systematic phase shift of all five factors with speed in the same direction as the shift in the onset of the swing phase. This tendency for the factors to be timed according to the lift-off event supports the idea that the origin of the gait cycle generation is the propulsion rather than heel strike event. The basic invariance of the factors with walking speed and with body weight unloading implies that a few oscillating circuits drive the active muscles to produce the locomotion kinematics. A flexible and dynamic distribution of these basic components to the muscles may result from various descending and proprioceptive signals that depend on the kinematic and kinetic demands of the movements.

Modulation of dorsal spinocerebellar responses to limb movement. I. Effect of serotonin.

Spinocerebellar neurons (DSCT) receive converging sensory information from various sensory receptors in the hindlimbs and lower trunk. Previous studies have shown that sensory processing by DSCT neurons results in a representation of global hindlimb kinematic parameters such as the length and the orientation of the limb axis. In addition to the sensory input, the DSCT circuitry also receives a descending input from the raphe nuclei in the brain stem. Recent studies have demonstrated that the raphe serotonergic terminals synapse directly on DSCT neurons and exert a differential modulatory influence on their sensory inputs. We examined the role of serotonergic modulation on the DSCT representation of hindlimb kinematic parameters by recording DSCT activity during passive hindlimb movements before and after perturbing serotonergic transmission. We used two types of perturbation: electrical stimulation of the raphe areas in the brain stem to release serotonin in the spinal cord (42 neurons) and intravenous administration of serotonergic agonists or antagonists, mostly the 5HTP2 antagonist ketanserin (30 neurons). We found that movement responses were altered in approximately 70% of the DSCT units studied with each protocol. Changes could include shifts in mean firing rate, increases or decreases in response amplitude, and changes in response waveform. We used a principal component analysis (PCA) to examine waveform components and to determine how they contributed to the response waveform changes caused by serotonin perturbation. Such changes could be explained by new or different response components that might indicate a modification in the data processing or by a different weighting of existing components that might indicate a modification of synaptic weighting. The results were consistent with the second alternative. We found that the same underlying response components could account for both control responses and those altered by serotonin perturbations. The observed changes in waveform could be entirely accounted for by a re-weighting of response components. In particular, the changes observed after raphe stimulation could be accounted for by selective changes in the weighting of the first principal component (PC) with only minor changes of the weighting of the second PC. Because these response components were shown previously to correlate with the limb axis orientation and length trajectories respectively, the finding is consistent with the idea that limb axis length and orientation information are processed separately within the spinal circuitry.

Modulation of dorsal spinocerebellar responses to limb movement. II. Effect of sensory input.

Dorsal spinocerebellar tract (DSCT) neurons receive converging sensory inputs from muscle, skin, and joint receptors and their cerebellar projection is a product of the spinal sensory processing of movement-related information. We concluded earlier that DSCT activity relates to global rather than to local parameters of hindlimb postures and movement, specifically to a kinematic representation of the limb endpoint. The waveforms of principal components (PCs) derived from an ensemble of DSCT movement responses were found to correlate with either the waveform of the limb axis length or orientation trajectories. It was not clear, however, whether these global representations resulted from neural processing or from biomechanical factors. In this study, we perturbed the limb biomechanical factors by decoupling limb geometry from endpoint position during passively applied limb trajectories patterned after a step cycle. We used two types of perturbations: mechanical constraints that limited joint rotations and electrical stimulation of hindlimb muscles. We found that about half of the 89 cells studied showed statistically different response patterns during the perturbations. We compared the PCs of the altered responses with the PCs of the control responses, and found two basic results. With the joint constraints, >85% of the total variance in both control and changed responses was accounted for by the same five PCs that were also observed in the earlier study. The differences between altered and control responses could be fully accounted for by changes in the PC weighting, suggesting a modulation of global response components rather than an explicit representation of local parameters. With the muscle stimulation, only the first and third PCs were the same for the control and altered responses. The second PC was modified, and additional PCs were also required to account for the altered responses. This suggests that the stimulus parameters were specifically represented in the responses. The changes induced by both types of perturbation affected primarily the weighting or waveform of the second PC, which relates to the limb axis length trajectory. The results are consistent with the suggestion that information about limb orientation and length may be separately modulated.

Dorsal spinocerebellar tract neurons respond to contralateral limb stepping.

Proprioceptive sensory information carried by spinocerebellar tracts provides a major input to the spinocerebellum, which has an important role in coordinating motor output for posture and locomotion. Until recently it was assumed that the information transmitted by the dorsal spinocerebellar tract (DSCT) was organized to represent single muscles or single joints in the ipsilateral hindlimb. Recent studies have shown, however, that DSCT activity represents global kinematic parameters of the hindlimb. We now present evidence that the DSCT neurons are also modulated by passive step-like movements of either hindlimb, implying they receive a bilateral sensory input. About two-thirds of 78 cells studied had significant responses to movements of the contralateral limb alone and about 70% responded differently to bipedal movements than to ipsilateral movement alone. The same basic behavior was observed in anesthetized, paralyzed cats and in unanesthetized, decerebrate cats, although decerebrate cats may have had larger responses on average. The results suggest that many DSCT cells may encode information about interlimb coordination.

A single-camera method for three-dimensional video imaging.

We describe a novel method for recovering the three-dimensional (3D) point geometry of an object from images acquired with a single-camera. Typically, multiple cameras are used to record 3D geometry. Occasionally, however, there is a need to record 3D geometry when the use of multiple cameras is either too costly or impractical. The algorithm described here uses single-camera images and requires in addition that each marker on the object be linked to at least one other marker by a known distance. The linkage distances are used to recover information about the third dimension that would otherwise be lost in single-camera two-dimensional images. The utilities of the method are its low-cost, simplicity, and ease of calibration and implementation. We were able to estimate 3D distances and positions as accurately as with a commercially available multi-camera 3D system. This method may be useful in pilot studies to determine whether 3D imaging systems are required, or, it can serve as a low-cost alternative to multi-camera systems.

Encoding of hindlimb kinematics by spinocerebellar circuitry.

Earlier work from our laboratory showed that principal component waveforms (PCs) from an ensemble of DSCT movement responses correlated with either the waveform of the limb axis length or orientation trajectories, suggesting that DSCT circuitry might elaborate an explicit representation of limb endpoint kinematics independent from limb geometry. In this study, we tested this idea by decoupling limb geometry from endpoint position with mechanical constraints that blocked the motion of the knee joint during step-like movements applied passively to the hindlimb of anesthetized cats. Only about half of the 50 cells studied showed statistically different response patterns when the limb was constrained compared to the unconstrained condition (control). However, the PC waveforms extracted from responses that showed significant changes with the knee constrained were found to be identical to those extracted from control responses. Instead, the differences between constrained and control responses could be accounted for by changes in the weighting of PCs suggesting a modulation of global response components rather than an explicit representation of local parameters.

Independent representations of limb axis length and orientation in spinocerebellar response components.

Dorsal spinocerebellar tract (DSCT) neurons transmit sensory signals to the cerebellum that encode global hindlimb parameters, such as the hindlimb end-point position and its direction of movement. Here we use a population analysis approach to examine further the characteristics of DSCT neuronal responses during continuous movements of the hind foot. We used a robot to move the hind paw of anesthetized cats through the trajectories of a step or a figure-8 footpath in a parasagittal plane. Extracellular recordings from 82 cells converted to cycle histograms provided the basis for a principal-component analysis to determine the common features of the DSCT movement responses. Five principal components (PCs) accounted for about 80% of the total variance in the waveforms across units. The first two PCs accounted for about 60% of the variance and they were highly robust across samples. We examined the relationship between the responses and limb kinematic parameters by correlating the PC waveforms with waveforms of the joint angle and limb axis trajectories using multivariate linear regression models. Each PC waveform could be at least partly explained by a linear relationship to joint-angle trajectories, but except for the first PC, they required multiple angles. However, the limb axis parameters more closely related to both the first and second PC waveforms. In fact, linear regression models with limb axis length and orientation trajectories as predictors explained 94% of the variance in both PCs, and each was related to a particular linear combination of position and velocity. The first PC correlated with the limb axis orientation and orientation velocity trajectories, whereas second PC with the length and length velocity trajectories. These combinations were found to correspond to the dynamics of muscle spindle responses. The first two PCs were also most representative of the data set since about half the DSCT responses could be at least 85% accounted for by weighted linear combinations of these two PCs. Higher-order PCs were unrelated to limb axis trajectories and accounted instead for different dynamic components of the responses. The findings imply that an explicit and independent representation of the limb axis length and orientation may be present at the lowest levels of sensory processing in the spinal cord.

Proprioception from a spinocerebellar perspective.

This review explores how proprioceptive sensory information is organized at spinal cord levels as it relates to a sense of body position and movement. The topic is considered in an historical context and develops a different framework that may be more in tune with current views of sensorimotor processing in other central nervous system structures. The dorsal spinocerebellar tract (DSCT) system is considered in detail as a model system that may be considered as an end point for the processing of proprioceptive sensory information in the spinal cord. An analysis of this system examines sensory processing at the lowest levels of synaptic connectivity with central neurons in the nervous system. The analysis leads to a framework for proprioception that involves a highly flexible network organization based in some way on whole limb kinematics. The functional organization underlying this framework originates with the biomechanical linkages in the limb that establish functional relationships among the limb segments. Afferent information from limb receptors is processed further through a distributed neural network in the spinal cord. The result is a global representation of hindlimb parameters rather than a muscle-by-muscle or joint-by-joint representation.