Multifactorial motor behavior assessment for real-time evaluation of emerging therapeutics to treat neurologic impairments

Integrating multiple assessment parameters of motor behavior is critical for understanding neural activity dynamics during motor control in both intact and dysfunctional nervous systems. Here, we described a novel approach (termed Multifactorial Behavioral Assessment (MfBA)) to integrate, in real-time, electrophysiological and biomechanical properties of rodent spinal sensorimotor network activity with behavioral aspects of motor task performance. Specifically, the MfBA simultaneously records limb kinematics, multi-directional forces and electrophysiological metrics, such as high-fidelity chronic intramuscular electromyography synchronized in time to spinal stimulation in order to characterize spinal cord functional motor evoked potentials (fMEPs). Additionally, we designed the MfBA to incorporate a body weight support system to allow bipedal and quadrupedal stepping on a treadmill and in an open field environment to assess function in rodent models of neurologic disorders that impact motor activity. This novel approach was validated using, a neurologically intact cohort, a cohort with unilateral Parkinsonian motor deficits due to midbrain lesioning, and a cohort with complete hind limb paralysis due to T8 spinal cord transection. In the SCI cohort, lumbosacral epidural electrical stimulation (EES) was applied, with and without administration of the serotonergic agonist Quipazine, to enable hind limb motor functions following paralysis. The results presented herein demonstrate the MfBA is capable of integrating multiple metrics of motor activity in order to characterize relationships between EES inputs that modulate mono- and polysynaptic outputs from spinal circuitry which in turn, can be used to elucidate underlying electrophysiologic mechanisms of motor behavior. These results also demonstrate that proposed MfBA is an effective tool to integrate biomechanical and electrophysiology metrics, synchronized to therapeutic inputs such as EES or pharmacology, during body weight supported treadmill or open field motor activities, to target a high range of variations in motor behavior as a result of neurological deficit at the different levels of CNS.


Introduction
Damage to the central nervous system (CNS), due to either acute events such as cerebral ischemia or spinal cord injury (SCI), or prolonged neurodegenerative diseases such as Parkinson Disease (PD) or Multiple Sclerosis (MS), can result in lifelong impairment of sensorimotor functions (e.g., reaching, grasping, standing, and/or walking). Once damages, the CNS undergoes a cascade of complex changes both within the brain and across spinal cord sensorimotor networks that integrate sensory signaling and motor control commands to produce coordinated neuromuscular activity. Neuromodulation has emerged over the last decades as a clinically available therapeutic to alleviate neurologic deficits, such as gait dysfunction due to PD 1-4 , tremor 5 , medically refractory central pain syndromes 6 , , dystonia 7 and epileptic seizures 8 .
Additionally, over the last few years, investigations using EES have shown it hold great potential to improve function in humans with spasticity 9,10 and restoration of motor control after spinal cord injury [11][12][13][14] .
One of the most critical concepts of spinal cord neuromodulation is a central pattern generation (CPG) that describes the organization of distinct neural networks that receive patterned sensory signaling and, in turn, produce predictable patterns of motor activity 15,16 . The anatomical overlap of sensory signals that originate in the peripheral nervous system and converge upon spinal sensorimotor networks via multi-segmental innervations to the spinal cord and the brain suggests that, multiple CPGs across central nervous system could integrate sensory signaling to allow real-time optimization of motor outputs in response to external perturbations [15][16][17][18] (Fig.   1a). It is expected that CPGs at different level of CNS are able to coordinate activity during complex motor functions, such as standing, stepping, or running. However, the distinct role of the neuronal circuitry of CPG during motor task performance, as well as the mechanism by which multiple CPGs coordinate their activity remains undefined [19][20][21][22] .
The majority of this work has focused on interpreting electromyography recordings from EESenabled muscle activity synchronized to biomechanical assessments of movements, however, a gap in knowledge remains with respect to direct modulation of spinal sensorimotor activity in relation to end organ function. This gap is largely due to a lack of affordable in vivo tools capable of modulating and recording EES-enabled sensorimotor activity synchronized to each electrical pulse that is delivered to neuronal structures, and synchronized to biomechanical assessments of motor performance, all of which are correlated to sensory input to the CNS during different motor patterns.
Several experimental models have been established to estimate interactions between the electric fields that are produced during stimulation and neural tissues in close proximity to stimulation 31,32 . Although, these studies revealed the anatomical structures most likely to be facilitated by EES, and the roles these structures play in producing both desirable and undesirable outcomes, in silico properties of computational modelling and limitations of in vitro models fail to describe key observations made during in vivo investigations, such as the degree to which different parameters of stimulation influence motor output via activation of distinct sensorimotor networks, the impact of pharmacological dosing on network activity, or which motor training-induced network changes are responsible for improvements in motor task performance over the course of EES. Currently available in vivo models of neurologic damage lack approaches to investigate spinal sensorimotor network inputs and outputs simultaneously.
Additionally, outcomes generated by available models are typically captured using isolated, nonsynchronized, assessments of motor task performance and electrophysiological metrics 24,26,28,33 .
Multiple studies have shown that motor training following SCI reorganizes neural circuitry via repetitive reinforcement of sensorimotor network activation patterns 34,35 . For example, in rodents, step training on a treadmill following SCI induces reorganization of spinal sensorimotor ensembles via selective reinforcement of locomotor networks 26 . Furthermore, step training combined with EES, and/or pharmacological neuromodulation, both of which alter spinal network excitability, enhances motor performance 23 . Positive outcomes from task-specific training in humans after incomplete SCI led to the development of several body weight support systems (BWSs) to achieve appropriate posture in animal models and humans during treadmill training [36][37][38] or over-ground locomotion 39 . These BWSs were developed to maximize the effect of task-specific motor training while assessing general aspects of motor performance, however, these systems lack the capability to integrate data recordings across multiple dynamic variables such as limb loading, kinematics, and timing of exogenous neuromodulatory application (e.g., electrical pulse frequency, intensity, pharmacological dosage, time of application). Additionally, mechanisms responsible for locomotion network plasticity in neuromotor impairments (e.g. SCI, PD, tremor, spasticity, MS), as well as the progression of plasticity over the course of motor training, have yet to be elucidated. To our knowledge, currently available assessments of motor performance in rodents are not capable of simultaneously recording and integrating critical parameters to gain knowledge on the mechanisms through which neuromodulation with motor training enables different types of motor behavior and re-organizes spinal networks.
In order to evaluate impaired neural circuitry facilitated by neurostimulation combined with motor-training, we proposed a multifactorial behavioral assessment (MfBA) and designed a new BWS assessment system, which integrates multiple behavioral and electrophysiological parameters and allows simultaneous recording of sensorimotor network inputs and outputs, synchronized to multiple metrics of motor behavior (Fig. 1 a-b). Proposed MfBA approach and system's capabilities were tested using neurologically intact rodents, parkinsonian (PD), and SCI rodent models that exhibit distinct motor dysfunctions related to the brain and spinal cord lesion. Following validation of proposed MfBA approach in healthy animals, PD, and SCI rats, we hypothesized that, during functional tasks in SCI rats, EES and pharmacology enabled modulation of mono-and polysynaptic components of spinal circuitry is reflective of changes in motor outputs and animal behavior.

Materials and methods
Multi-factorial behavioral assessment system configuration The multi-factorial assessment system consists of the following main components: (1) BWS apparatus integrated with force and torque transducers, (2) motion tracking system, (3) open field camera, (4) interface with chronically implanted electrophysiological recording electrodes, (5) interface with chronically implanted neural stimulation electrodes, (6) motorized treadmill, and (7) open field platform. Through hardware level communication between these modules the system combines behavior, locomotion and electrophysiology in either an open field or treadmill environment with the flexibility to position animals for bipedal or quadrupedal stepping. This system is capable of synchronizing input variables including EES, pharmacology, treadmill speed, variation of load, direction of locomotion, and extent of BWS. Body weight support system (BWS) The BWS system (Fig. 1b, d) provides trunk support to motor impaired rodents while imposing minimal friction forces in order to provide a neutral environment in which to study motor behavior. Rodents were secured to the BWS using a custom-made, fully adjustable fabric padded jacket, to provide BWS during training and assessment activities on a treadmill or in open field. Components of the BWS system were chosen with the intent to achieve friction forces that allow unrestricted motor behavior in the open field environment while also considering overall system cost. The total cost to build the BWS was ~$400 (Sup. table 1 for cost details). The major components of the BWS are described herein:

Aluminum frame structure
The BWS structure consists of rectangular aluminum alloy extrusion bars and corner plates (47065T503, Mcmaster-Carr, Elmhurst, Illinois) ( Fig. 1b-c). The rigid framework is easily modifiable to accommodate assessment equipment required during motor training and data recording, such as camera angle and height adjustment to during treadmill or open field motion analysis. The goal of the frame is to provide consistent placement of the equipment and be adaptable for new experiments. To ensure that open field test can be performed in the system, the frame was constructed to provide a 120cmx120cmx60cm unobstructed environment.

Two axis linear motion
To allow for planar motion with Z axis support, a linear 2 axis system was created using guiding rods and bearings. By taking weight into account, the system used 6.35 mm precision ground 6061 aluminum rods and stainless steel linear ball bearings with aluminum housings (5911K11, NB corporations, Hanover Park, Illinois). For a 300-gr rat (1 Kg of total moving load) the max deflection of the guiding rod was calculated to be 3.023mm. The static and dynamic forces required to move the 2-axis system is reduced further by removing the linear bearings seals.
Using the formula (equation 1) provided by the manufacturer (NB Corporation), the dynamic friction force of 0.0294 N/bearing was calculated after removing the seals.

F=μW+f
(1) Where, F represents dynamic friction force (N), µ is the dynamic friction coefficient, W is the applied load (N) and f is the resistance from seal (N). Theoretical friction value of the linear bearings after removing the seal was calculated and then compared with the friction value experienced by the bearing during dynamic trials (Sup . table 2). To further define the static and dynamic friction forces of the BWS system, we conducted tests using an actuator (MT1-28, Thorlabs, N, New Jersey) and a load cell (Nano17, ATI, Apex, NC) (Fig. 1e,f). The actuator utilized micro-stepping and acceleration controls to provide an applied force to acquire friction and inertial data through the load cell across X and Y axis, and 45 0 angles.

Weight manipulation
As shown in Fig. 1c,d, the rodent's weight can be manipulated by changing the Z axis position, variation of the pitch angle, and by shifting the rat's weight bilaterally. The Z axis support uses a stepper motor and ball screw to accurately change the amount of body weight support provided in small increments.

Surgical procedures
EMG wire and electrode implants for healthy and SCI rats A small skin incision was made at the midline of the skull. The muscles and fascia were retracted laterally, and the skull was thoroughly dried. A 12-pin Omnetics circular connector (Omnetics, Minneapolis, MN) and 12 Teflon-coated stainless-steel wires (AS632, Cooner Wire, CA) were attached to the skull with screws and dental cement as previously described 28,40 . Skin and fascia incisions were made to expose the bellies of the medial gastrocnemius (MG), and tibialis anterior (TA) muscles bilaterally. Using hemostats, the EMG wires were routed subcutaneously from the back incision to the appropriate locations in the hind-limb. Bipolar intramuscular EMG electrodes were inserted into the muscles as described previously 40 . The EMG wires were coiled near each implant site to provide stress relief. Electrical stimulation through the head-plug was used to visually verify the proper placement of the electrodes in each muscle.
A partial laminectomy was then performed at the L2 vertebral level (S1 spinal cord level) and one wire was affixed to the dura at the midline using 9.0 sutures as previously described 28 . A small notch made in the Teflon coating (about 0.5-1.0 mm) of the wire used for EES was placed toward the spinal cord and served as the stimulating electrode. The wire was coiled in the back region to provide stress relief. Teflon coating (about 1 cm) was stripped from the distal centimeter of one wire that was inserted subcutaneously in the back region and served as a common ground 25 .

Spinal cord transection
One week after electrode implantation surgery, a complete spinal cord transection was performed on the three SCI rats. Rats were anaesthetized with a mixture of oxygen and Isoflurane (≈1.5%). Mid-dorsal skin incision was made between T6 and L4 and the paravertebral muscles were retracted as needed. A partial laminectomy was performed at the T8 level and the dura was opened longitudinally. Lidocaine was applied locally and the spinal cord was completely transected using a micro-scissors. Completeness of the lesion was verified by two surgeons under microscope. If so, tissues were sutured by layers and animals were allowed to recover in individual cages with soft bedding. Manual bladder expression was performed four times daily during two weeks post transection.

6-OHDA injection
Three male Sprague-Dawley rats were intraperitoneally anesthetized with a mixture of ketamine (90mg/kg) and Xylazine (10mg/Kg) and received atropine methyl bromide (0.03mg/Kg, IM) and buprenorphine SR to reduce bronchial secretion and to provide 72 hours of pain relief, respectively. They were also pre-treated with Pargyline (50mg/Kg, IP) to prevent 6-OHDA catabolism by monoamine oxidase. Following induction of anesthesia, each rat was placed in a standard stereotaxic frame in which the skull was secured with a nose clamp, incisor bar, and ear bars. Body temperature was maintained constant at 37 degrees C by a heating pad. weeks post transection surgery. In control rats, data collection started 1 week post EMG wire implant surgery and in PD rats gait information has been collected 3 weeks after 6-OHDA injection. Custom MATLAB scripts were used to analyze the data.

Spinal Cord Electrical Epidural Stimulation
A single channel manually controllable isolated (A-M systems, Sequim, WA) or an eight-channel real-time programmable (STG4008, Multichannel Systems, Reutlingen) stimulators was used to deliver biphasic square wave pulses (250 µs pulse width) at 40 Hz with amplitudes ranging from 0.5 to 2 volts to the epidural electrode placed on the rat lumbosacral (S1) spinal cord.

Gait analysis
The motion tracking system (Vicon, UK) was used to record three-dimensional digital position of back and the hind limb joints (100 Hz). Six motion-sensitive Infra-Red (IR) cameras were aimed at the treadmill or the open field volumes. Another high-speed video camera synchronized with motion tracking system was positioned in a side to provide a lateral view of the motor performance. Retro-reflective markers were placed on bony landmarks at the iliac crest, greater trochanter, lateral condyle of the femur, lateral malleolus and the distal end of the fifth metatarsal on both legs of the rat to record the kinematics of the hip, knee and ankle joints.
Nexus system was used to obtain three-dimensional coordinates of the markers. Gait cycles were defined as the time interval between two successive paw contacts of one limb. Successive paw contacts were visually defined by the investigators based on video records. Cycle duration, and stance and swing durations were also manually determined from the kinematic recordings.
Each step of a sequence was resampled to average cycle duration of the steps collected during the session, thus minimizing elimination of temporal information. Limb endpoint trajectories were determined from motion of the toe marker. The computed step parameters included stanceswing phase, step frequency, stride length, maximum toe height and joint angles.

EMG and fMEP analysis
EMG activity was collected from TA and MG muscles at 4000 Hz during stepping and later highpass filtered at 0.5 Hz to remove the DC offset. Synchronization pulse sent from motion tracking system was used to identify the same time window used in gait analysis. To perform fMEP (functional Motor Evoked Potential) analysis of EMG activity recorded during stepping, EMG signals were processed based on EES inter-pulse time windows, e.g. 25ms for 40 Hz and 10ms for 100 Hz stimulation, and organized sequentially. To quantify the number of mono and poly synaptic responses during 40 Hz stimulation, the numbers of peaks were counted within 5.5 to 9.1ms and 9.1 to 25ms. In order to determine flexor (TA) and extensor (MG) coordination, EMG signals were band pass filtered (10Hz to 1000Hz), rectified, normalized and plotted.

Kinetic recordings
A transducer (Nano17, ATI, Apex, NC) was used to record 6 measurements including force and torque across the X (F x, τ x ), Y (F y , τ y ) and Z (F z , τ z ) axis. This load cell was placed directly above the rat's centerline and measures force and torque produced by the rodent during locomotion tasks. A custom LabVIEW (National Instrument, Austin, Texas) script was used to acquire force and torque data. The signals were recorded at 100 Hz sampling rate and later high-pass filtered at 0.5 Hz to remove DC offset. Beginning and end of stepping phase was determined using the video recorded simultaneously. Average force and torque were determined dividing the total amount force with duration of recordings. Additionally, force vectors for each time points were determined using, F= F x i + F y j + F z k.

Assessment of BWS system's mechanical properties
After building the system, we calculated the theoretical dynamic friction values to the direction parallel to, Y axis ( ), diagonal to X and Y axis ( 45 ), and across X axis ( ) for a 1 Kg load. The values of , 45 and were calculated to be 0.29 N, 0.12 N and 0.18 N, respectively. After building the BWS system, the experimental static and dynamic friction profile were obtained, while a constant force was applied ( Fig. 1d-  and bipedal (~60% BWS) locomotion were compared (Fig. 2a-c). Quadrupedal stepping after unilateral 6-OHDA injection in hemi-Parkinsonian rats (n=3, ~20% BWS, Fig 2d)   Comparison of gait parameters, (e) step frequency, (f) stride length, (g) maximum toe height, and (h) stance-swing cycle. (*p <0.05, **p<0.01 and ***p<0.001, one-way ANOVA, data represents mean± standard deviation, n=5 (healthy rats), n=3 (PD rats)).

(b) Simultaneous kinematic and open field behavioral assessment in neurologically intact rodents
To determine if the friction level of the BWS system is within tolerable range to allow open field locomotion in rodent, we performed open field test using healthy rats in the 80x80 cm area within the BWS system. The same rats (n=4) moved freely around the open field and also while supported by the BWS system for 10 minutes (Fig. 3a-d). The BWS system was found to significantly reduce the total distance travelled by the rats (Fig. 3b,  steps/min), that difference was not significant (Fig. 3f). Additionally, stride lengths (BWS: 4.21±2.26 cm and freely walking: 4.14±1.77 cm) were found to be similar (Fig. 3g). However, maximum toe height during BWS stepping (3.52±0.61 cm) was found to be significantly higher compared to freely walking (6.25±2.8 cm) (p<0.001) (Fig. 3h).

(c) Simultaneous kinematic and electrophysiological assessment during open field behavior
The feasibility of combining kinematic and EMG recording during open field locomotion was evaluated next (Fig. 3i). Open field trajectories were recorded continuously using a camera positioned above the open field platform, which was synchronized to limb kinematics recorded during two epochs (i and ii) (Fig. 3i). Using the open field trajectory, we quantified the maximum and average velocities of the rat during the two epochs of kinematic recordings (Fig. 3j) and found a maximum velocity (14.12 cm/s) during epoch-i, which was corroborated by a maximum step frequency (211.64± 17.71 steps/min) recorded by the kinematic system (Fig. 3l). We also quantified the peak-to-peak voltage of EMG traces recorded from the TA muscle (Fig. 3k) and found minimum peak-to-peak amplitudes occurred during epoch-i (2.91±0.3 mV). No significant differences were observed in strident length (Fig. 3m) and maximum toe height (Fig. 3n).
However, significantly higher toe fluctuation was observed during epoch-i (9.83±3.86 cm compared to epoch-ii (4.03±1.4 cm) (Fig. 3o). Variation in directional force and torque during treadmill locomotion in intact and SCI rodents Ichiyama et al. 33 and multiple following research reports using electrophysiological and behavior assessments described that, 0.3mg/Kg Quipazine (a 5HT2A/C agonist) injection can improve stepping outcome in SCI rat with or without EES. In this study we tested kinetic outcome of healthy and SCI rodent behavior with and without Quipazine. Using a six-axis force and torque sensor placed directly above the rat (Fig. 4a)  We quantified open field, kinematic, kinetic, EMG and fMEP variables (n=34 variables) across the modalities of output collected for both sub-threshold and threshold EES. As a large number of these parameters changed substantially between the conditions, in order to summarize their effect on output modalities, we performed principle component analysis (PCA). PCA identified the distribution of the steps facilitated by sub-threshold and threshold EES (Fig. 5b). The first three PCs explained 98% of the variance associated with the data. The sub-threshold and threshold steps found to be isolated in the PC space. In addition, sub-threshold EES steps were clustered together with minimum variation. In contrast, threshold EES steps were more spread across the PC space.
To demonstrate how the sub-threshold and threshold EES-enabled stepping vary across multiple modalities, we also presented direct comparison between tested modalities (Fig. 5c-g).
This allowed to demonstrate how the higher level motor outputs (e.g. open field and kinematic parameters), correlates with functional state of spinal cord circuitry (assessed with fMEP) during tested behaviors. Increasing EES intensity from sub-threshold level to threshold level resulted in significant increase in distance travelled by the rat (Fig. 5c, p<0.001), improved forward force (Fig. 5d, P<0.01), and increased toe height (Fig. 5e, p<0.001). EMG amplitude in both TA and MG muscles also increased significantly (p<0.001), although these changes were more prominent in MG (Fig. 5f). In order to evaluate the contribution of polysynaptic components of spinal cord circuitry, during sub-threshold and threshold EES, we quantified the number of polysynaptic peaks during these two conditions (Fig. 5g). Polysynaptic peak counts did not change significantly during sub-threshold and threshold EES in TA, but increased significantly in MG at threshold level, indicating predominant influence of EES at threshold level on activity in extensor muscles during stance phase (p<0.001), reflecting circuitry modulation by sensory input during step cycle. This multifactorial analysis of simultaneously collected multimodal data shows a direct correlation between alteration of the states of spinal cord locomotor circuit and subsequent effect on locomotor behavior.

Multifactorial behavioral assessment of EES and pharmacologically-enabled locomotion in rodents with complete SCI
The MfBA of spinal neural circuitry responsible for locomotion based on behavior and electrophysiological features was further extended in SCI rats (n=3) 4-5 weeks after complete T8 transection under a combination of EES and pharmacology facilitated stepping (Fig. 6).
Although, specific kinematic characteristics of these synergistic neuromodulators have been discussed, previous studies 27  Sub-threshold level (<0.6 V) of stimulation was applied in healthy rats in order to obtain fMEP without influencing motor output during bipedal stepping on the treadmill while supported by the BWS (60% BWS). Healthy rats did not show any sign of discomfort during sub-threshold stimulation and exhibited regular pattern of stepping (Fig. 6b). Following SCI, no stepping was observed during sub-threshold EES (Fig. 6c). EES at threshold level alone (Fig. 6d), at subthreshold level with Quipazine injection (Fig. 6e), and at threshold level with Quipazine injection (Fig. 6f), produced consistent stereotypical steps. The number of steps taken per minute decreased significantly from healthy rats (76±14 steps/min) to SCI (0 steps) rats when subthreshold EES was applied in both groups (Fig. 6g,  cm to 0.13±0.04 cm after SCI when sub-threshold EES was applied (Fig. 6h, p<0.001).
EMG activity recorded from TA and MG muscles of SCI rats was evaluated under the same conditions (Fig. 6i). During sub-threshold EES, SCI rats produced minimal to no coordination between TA and MG muscles. However, at threshold level EES, coordination between TA and MG muscles was evident. Quantification of peak to peak EMG amplitude in both TA and MG muscles showed, significant reduction following SCI when sub-threshold EES was applied (p<0.001) (Fig 6j).Threshold EES, sub-threshold EES with Quipazine or threshold EES with Quipazine significantly increased TA amplitude (p<0.01). Interestingly, in MG muscle, even though threshold EES restored amplitude to preinjury level (p<0.001), sub-threshold EES with Quipazine injection failed to increase amplitude. However, threshold EES with Quipazine injection was effective in increasing MG amplitude (p<0.001).  (Fig. 6l-m). In TA muscle, healthy rats with sub-threshold EES generated ~400 polysynaptic peaks/step, which decreased to ~12 peaks/step after SCI when same sub-threshold EES was applied (p<0.01, Fig. 6l). Threshold EES failed to significantly increase number of polysynaptic peaks. However, sub-threshold EES with Quipazine injection or threshold EES with Quipazine injection, significantly increased polysynaptic peaks (~1000 and ~700 peaks/step, correspondingly; p<0.001). Interestingly, subthreshold EES with Quipazine injection increased polysynaptic peaks/step even higher than healthy rats with sub-threshold EES (p<0.05). MG muscle of healthy rat during sub-threshold EES produced ~500 peaks/step, but following SCI the number of polysynaptic peaks decreased to ~10 peaks/step (p<0.001). Threshold EES increased the number of polysynaptic peaks to ~500 peaks/step (p<0.001). However, sub-threshold EES with Quipazine injection failed to increase the number of polysynaptic peaks/step to ~80 peaks/step; however, threshold EES with Quipazine injection did increased the number of polysynaptic peaks (1100 peaks/step, p<0.001), even higher than healthy rats with sub-threshold EES (p<0.05).
In TA muscle, AOC did not change after SCI with sub-threshold or threshold EES (Fig 6m). In SCI rats, Quipazine injection with threshold EES increased AOC higher than healthy rats with sub-threshold EES or SCI rats with sub-threshold or threshold EES (p<0.001). In MG muscle, AOC also did not change after SCI when sub-threshold EES was applied. Threshold EES however, significantly increased AOC (p<0.001), which decreased again when Quipazine was injected during sub-threshold EES (p<0.001). Threshold EES with Quipazine did increase MG AOC compared to sub-threshold EES with Quipazine (p<0.01), but this was still lower than threshold EES (p<0.001).
10 fMEP variables were quantified (Sup. The results of the PCA are presented in the 3D graph in Fig. 6n, by plotting the first three PCs. These  To validate designed for this study BWS system, we performed extensive mechanical tests to determine the friction forces imposed on rodents by BWS components of the system. The maximum dynamic frictional force of 0.28±0.1 N in the Y direction is within the range of propulsive force (~± 2N/Kg) produced by healthy rats during locomotion 51 . Additionally, the system's static and dynamic friction values are in line with values of Rabinowicz's sticking and sliding model, which describes the friction profiles of two lubricated sliding metal surfaces 52 .
Therefore the system we developed imposes minimal extrinsic force during motor task performance in contrasts with recently developed, robotically-assisted BWS systems that are comprised of large, expensive structural and mechanical components and contain active serial actuators, which convolute results through inherent integration of intrinsic friction forces produced by the system and forces produced by SCI rodents during locomotion 53 . Some earlier versions of rodent BWS system relies on attaching robotic arms to the rat's hindlimb in order to measure hindlimb trajectories 54,55 . However, this method is only suitable for less severe injury and may fail to adapt with the sudden change in locomotion.
During system validation, we characterized control bipedal, control quadrupedal, and PD quadrupedal treadmill stepping (Fig. 2). As expected, contralateral to lesion left hind limb gait deficits were identified in the PD rats. Specifically, PD rats exhibited reductions in step frequency, stride length, and maximum toe height, with increased stance phase duration ( Fig.   2e-h). Previous studies also reported impairments in contralateral hind limb of unilateral 6-OHDA -rats in the form of shorter steps with reduced toe clearance 56,57 , which reflects the shuffling gait in human PD patients 58 . It has been reported that, pattern-generating networks responsible for stepping in dopamine-depleted rats are still functional and can produce coordinated rhythmic gait pattern 59 . A novel assessment system capable of providing time varying sensory input while simultaneously recording motor behavior can be further deployed to study the mechanism of motor deficit in rodents with brain lesion and for evaluation of therapeutic options.
In this study several functional tests, including open field assessment of neurologically intact rodents with BWS were used to validate system performance and MfBA approach. We found that rodents explored less in the BWS compared to freely exploring (Fig. 3b), however, average velocity and duration of activity were not significantly different between freely exploring rats and rats in BWS, indicating that BWS did not impede motor activity. We also characterized open field gait parameters of healthy rats (Fig. 3e-h). During over-ground stepping increased toe height was observed during BWS supported locomotion compared to freely exploring rats (Fig.   3h). Other gait parameters (e.g. step frequency, stride length) were not statistically different between freely walking BWS supported stepping, which also supports the conclusion that the BWS system did not hinder gait. The differences in toe height during BWS versus free open field locomotion were not observed during treadmill locomotion (Fig. 2g), indicating that kinematic parameters of open field locomotion may be more sensitive to perturbations, and in turn, may prove useful as targeted gait characteristics to enhance recovery via emerging therapeutics. In contrast, treadmill training is less prone to external perturbations yet provides the opportunity to investigate repetitive gait characteristics at investigator-controlled presets.
The system's capability of recording locomotor kinetics using a force and torque transducer integrated into the BWS apparatus was demonstrated (Fig. 4, 5). Without any intervention, SCI rodents generated substantial forces in lateral directions, but only minimal forces forward and backward, which are essential in order to propel during locomotion. Following Quipazine (5-HT 2A and 5-HT 3 receptors agonist) administration, forward and reverse directional forces increased to levels that were higher than those recorded prior to SCI (Fig. 4e). Lumbosacral EES produced stepping characterized by low toe height clearances during swing phase with increased step frequency while the administration of Quipazine without EES increased toe height and reduced stepping frequency. At the same time, Quipazine and EES synergistically modulated spinal cord networks to enhance treadmill stepping performance after SCI (Fig. 6).
Previous studies have shown that EES enabled a greater number of steps compared to just Quipazine administration 33  Established in this study characteristics of mono-and polysynaptic components in flexor and extensor muscles during EES synchronized functional tasks in healthy and SCI rats indicate on significant alteration of motor outputs (e.g. open field movement, kinematic and kinetic output).
Associated with alteration of mono-and polysynaptic components recorded from related muscles during EES synchronized functional tasks, modulation of mono-and polysynaptic components in this study for the first time provided integrative approach to evaluate spinal locomotor circuitry and resultant motor behavior simultaneously. Specific changes in polysynaptic activity observed between tested groups could reflect complex changes in spinal circuitry. Thus, decreased number of polysynaptic components of fMEP after SCI was successfully compensated by administration of EES alone or with Quipazine, although EES alone was able to compensate polysynaptic activity only in MG muscle active during stance phase when circuitry related to MG receiving specific sensory facilitation from the foot, while component of circuitry related to TA could has limited sensory facilitation during swing phase and accordingly may require additional pharmacological modulation to became sensitive to EES. These findings suggest that fMEP integrated with MfBA approach could be a powerful tool in evaluation of spinal circuitry and related in vivo motor behavior.
In recent clinical studies [11][12][13] , a wide range of stimulation parameters were used to produce activity-dependent modulation of spinal sensorimotor networks, indicating a clear lack of information available to investigators during stimulation parameter optimization. Additionally, it remains unknown how various EES parameters influence human spinal circuit activity and longterm plasticity over the course of motor training. Currently available systems for motor behavior assessment, unfortunately, do not support necessary integration and real-time visualization of multiple assessments of spinal cord inputs and outputs [36][37][38][39]54 . In order to optimize EES-enabled activity-dependent modulation of human spinal network in an efficient and effective fashion, the faciliatory effects of EES on locomotor circuitry must be studied using integrative clinical assessment tools, which could be further design with MfBA approach similar to the system we have developed for rodent use.
In summary, by combining multiple assessment parameters including kinematic, kinetic, open field, and electrophysiology during therapeutically-enabled locomotion, we have established and tested a comprehensive real-time evaluation of motor behavior in healthy rodents and in rodents with neurologic deficits at the different CNS levels that significantly impact motor activity.
Additionally, we provide evidence that MfBA combined with fMEP analysis is effective tool for at dissecting therapeutically-enabled gait characteristics and at targeting of different components of locomotor circuitry related with variations in motor behavior. Novel MfBA approach and designed for this purpose BWS system provide a platform for future investigations of the interactions between CNS inputs and outputs while manipulating external perturbations, and therapeutic administration in order to better understand how the CNS coordinates and executes complex motor tasks in healthy animals and in animals with neurologic deficit at different CNS levels.  Step height 2

Supplementary
Step length 3 Toe fluctuation 4 Step duration 5 Stance phase 6 Swing phase 7 Drag phase