Edited by: Michelle Maria Rank, The University of Melbourne, Australia
Reviewed by: Jürg Streit, University of Bern, Switzerland; Marie-Pascale Cote, Drexel University, United States
This article was submitted to Cellular Neurophysiology, a section of the journal Frontiers in Cellular Neuroscience
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Electrical stimulation of the spinal cord is a promising rehabilitation intervention to restore/augment motor function after spinal cord injury (SCI). Combining sensory feedback with stimulation of remaining motor circuits has been shown to be a prerequisite for the functional improvement of SCI patients. However, little is known about the cellular mechanisms potentially underlying this functional benefit in the injured spinal cord. Here, we combine computer simulations with an isolated whole-tissue adult mouse spinal cord preparation to examine synaptic, cellular, and system potentials measured from single motoneurons and ventral roots. The stimulation protocol included separate and combined activation of the sensory inputs (evoked by dorsal root stimulation) and motor inputs (evoked by stimulation of spinal cord tissue) at different frequencies, intensities, and neuromodulatory states. Our data show that, while sensory inputs exhibit short-term depression in response to a train of stimulation, motor inputs exhibit short-term facilitation. However, the concurrent activation of both inputs elicits a stronger and steadier motor output. This effect is enhanced by the application of pharmacological neuromodulators. Furthermore, sensorimotor excitatory postsynaptic potentials (EPSPs) summate sublinearly (i.e., their combination produces an excitatory potential smaller than the sum of the excitatory potentials they would individually produce). However, ventral root compound action potentials (CoAPs) summate supralinearly generating much higher outputs. Computer simulations revealed that the contrasting summation and disproportionality in plasticity between the excitatory postsynaptic potentials (EPSPs) and CoAPs result from the motoneuronal firing threshold acting as an amplitude-selective filter. Together, these results provide the mechanistic basis for the cellular processes contributing to the generation of steady motor outputs using spinal stimulation. This data has great potential to guide the design of more effective stimulation protocols in SCI patients.
Electrical stimulation of the spinal cord is currently showing promise for motor rehabilitation after spinal cord injury (SCI). With electrical stimulation, patients showed improved trunk control, standing, stepping, and urogenital function (
To investigate the effects of electrical stimulation after SCI at the synaptic, cellular, and system levels, we used electrophysiological recordings and computer simulations to study the plasticity, integration, and neuromodulation of electrically evoked sensory and motor synaptic potentials in an isolated spinal cord preparation. Specifically, our goals were to: (1) Identify the characteristics of electrically triggered sensory and motor synaptic potentials generated in motoneurons (synaptic level), (2) understand how electrically triggered sensory and motor synaptic potentials integrate within the motoneuron (cellular level), and (3) examine the transformation of electrically triggered sensorimotor potentials into a motor output at the ventral roots (system level), which determines the muscle force.
Motor behaviors are determined by how different synaptic inputs to motoneurons are integrated to produce the firing patterns of spinal motoneurons, and by how these individual motoneuron outputs are integrated into system output (muscle force). In the spinal cord, three excitatory sources determine the firing pattern of motoneurons: Descending motor commands from supraspinal structures (referred to as ‘
As premotor interneurons fire repetitively, their synapses on motoneurons undergo typical short-term plasticity alterations (i.e., short-term facilitation or depression). Such plasticity is common to all neurons in the nervous system (
We therefore hypothesized that combined electrical stimulation of sensory and motor inputs in the transected spinal cord yields higher and more stable motor output than that generated by either input separately. To test this hypothesis, we used a whole-tissue
The fifty three male mice (B6SJL, Jackson laboratory, Bar Harbor, ME) were used in this study, 30–200 days old. The sacrocaudal spinal cord was surgically transected and isolated, maintained
The procedures for surgical isolation of the sacrocaudal spinal cord has been previously described (
Ventral roots were connected to a differential amplifier (Kinetic Software, GA) with 1000× gain and 300 Hz – 3 KHz bandwidth filter. We performed most of our recordings at the S4 segments where the responses were most stable and highest in amplitude (see below in ‘sensory and motor inputs’). During the first 30–40 min in the recording chamber, the ventral root response to dorsal root stimulation steadily increased in amplitude. Hence, the cord was allowed to recover for nearly 1 h before any recordings were started. The ventral root responses to trains of electrical stimulation were quantified as the peak-to-peak measurements of its compound action potentials (CoAPs).
Using sharp intracellular glass microelectrodes, single motoneurons were recorded in the isolated whole tissue. These glass electrodes were pulled using a micropipette puller (P97, Sutter instruments, CA), and filled with 3M potassium acetate and 100 mM KCl and had a resistance of 25–40 MΩ. The microelectrodes were advanced into the ventral horn using a micro-positioner (2660, Kopf instruments, CA). Motoneurons were identified by antidromic stimulation of the ventral root and were accepted for recording when the resting membrane potential was below −60 mV and the antidromic spike was ≥ 60 mV. The Na+ channel blocker, QX-314 (50–100 mM), was used in the internal electrode solution to inhibit action potential generation, so that the exact amplitude of the synaptic potentials can be determined. The amplitudes of synaptic potentials during a train of stimulation were measured as the voltage change at the peak of the EPSP relative to the baseline resting membrane potential before the stimulation train. Intracellular recordings were performed using an Axoclamp 2B amplifier (Molecular Devices, CA) running in bridge or discontinuous current clamp (DCC) mode and low-pass filtered at 3 kHz.
The outputs of both the intracellular and extracellular amplifiers were digitized using Power 1401-3 data acquisition interface (CED, United Kingdom) at 10–20 kHz. Data were acquired into a computer controlled by Spike2 software (version 8.06, CED) and stored for offline analysis.
The sensory inputs (S) were induced by electrical stimulation of the dorsal roots. Dorsal roots were connected through bipolar wire electrodes (
Experimental setup for recording synaptic responses to sensory and descending stimulation. The spinal cord is placed ventral side up in a perfusion chamber. Ventral and dorsal roots are mounted on bipolar wire electrodes above the solution level, and covered with petroleum jelly. The dorsal roots
The motor inputs (M) were induced by electrical stimulation of the spinal cord tissue (
The response to each synaptic input was recorded separately, and then both pathways were stimulated simultaneously to study integration. To induce a higher neuromodulatory state, the α1-adrenergic receptor agonist, methoxamine (10 μM), was added to the recording solution. After 10 min of continuous exposure, the electrical stimulation paradigm was repeated.
To measure the effective synaptic currents generated by the sensory and motor inputs in
The nACSF was formed of the following (in mM): 128 NaCl, 3 KCl, 1.5 MgSO4, 1 NaH2PO4, 2.5 CaCl2, 22 NaHCO3, and 12 glucose. The osmolarity of the solution was ∼295 mOsm, and the pH was 7.35–7.4 when aerated with carbogen (95% O2 and 5% CO2).
The mACSF contained the following (in mM): 118 NaCl, 3 KCl, 1.3 MgSO4, 5 MgCl2, 1.4 NaH2PO4, 1.5 CaCl2, 24 NaHCO3, and 25 glucose. This high Mg2+/low Ca2+ solution decreases the activity in the cord during dissection. The osmolarity of the solution was ∼310 mOsm, and the pH was 7.35–7.4 when aerated with carbogen (95% O2 and 5% CO2).
N-(2,6-Dimethylphenylcarbamoylmethyl) triethyl ammonium bromide (QX-314), a membrane-impermeable blocker of the voltage-gated Na+ channels; strychnine (STR), a blocker of the glycine receptors; picrotoxin (PTX), a blocker of the GABAA receptors; methoxamine, an agonist of the alpha-1 adrenergic receptors. All of the drugs were purchased from Sigma (St. Louis, MO, United States), and all of the chemical components of the physiological solutions were purchased from Thermo Fisher Scientific (Waltham, MA, United States).
To simulate the response of spinal motoneurons to electrically evoked sensory and motor inputs, we employed a multiscale, high-fidelity computer model of the alpha-motoneuron pool, which is described in full detail in
Each individual motoneuron cell model was driven by two sources of synapses: sensory and motor synapses. The distribution of excitatory and inhibitory sensory and motor synapses was uniform on the motoneuron dendrites and their conductances were the same across all cells in the pool. This assumption is supported by the equal synaptic input limb motoneurons receive from inhibitory Renshaw and Ia-reciprocal inhibitions and the small variability in amplitude of inputs motoneurons receive from excitatory Ia-afferents and vestibulospinal input (
Synaptic conductances of the sensory and motor inputs in the simulations.
P1 | 0.016 | 0.014 | 0.009 | 0.0022 |
P2 | 0.00976 | 0.0063 | 0.00954 | 0.0022 |
P3 | 0.0096 | 0.00588 | 0.009504 | 0.00187 |
P4 | 0.0104 | 0.00868 | 0.00954 | 0.00242 |
P5 | 0.01008 | 0.00868 | 0.009765 | 0.00308 |
To mimic our EPSP recordings in presence of QX-314 in the micropipette, we measured EPSPs in simulations when Na+ channels were blocked.
To investigate the effects of changing the driving force on synaptic integration (
Different timing parameters for the synaptic potentials of the sensory and motor inputs.
Delay from trigger | 1.97 ± 0.54 | 1.85 ± 0.32 | |
Time to peak (TPeak) | 1.69 ± 0.58 | 2.14 ± 0.59 | |
Half decay time (T1/2) | 4.3 ± 1.86 | 5.71 ± 1.78 |
The statistical analyses were performed using GraphPad Prism software (La Jolla, CA, Version 7.01). Repeated-measures one-way ANOVA was used to test changes in the amplitude of coAPs and EPSPs at different stimulation pulses to a specific pathway (i.e., sensory, motor, or sensorimotor). When the response at each pulse was compared between two different pathways, or before and after drug treatment, a repeated-measures two-way ANOVA was used. For
In this study, the whole-tissue adult sacrocaudal spinal cord was used to study the plasticity, integration, and neuromodulation of electrically evoked sensory and motor synaptic inputs generated in spinal motoneurons. The sensory input was activated by electrical stimulation of the ipsilateral dorsal roots, which generates a response that contains a prominent monosynaptic component, presumably from the Ia muscle afferents. On the other hand, the motor input was activated via surface electrical stimulation of the ventrolateral funiculus of the rostral end of the sacrocaudal preparation. This stimulation activates local interneurons in addition to the remaining axons of descending tracts, primarily the lateral vestibulospinal tract (LVST, see section “Discussion” for more details). The synaptic, cellular, and system responses to electrical stimulation were recorded as EPSPs and somatic APs measured from single motoneurons using intracellular sharp electrodes and coAPs measured from the ventral roots using extracellular wire electrodes, respectively (
Short trains of electrical stimuli (five pulses) of low intensity (1.5×T, T is threshold of each pathway) or high intensity (10×T) at physiological frequencies (25 Hz or 50 Hz) were delivered to either the dorsal roots (sensory input, S), the remaining descending axons (motor input, M), or both simultaneously (sensorimotor input, S&M). Threshold was determined separately for the S and M inputs, and was identified as the minimum stimulation intensity needed to activate that particular pathway to evoke the smallest observable response in the ventral roots.
Electrical stimulation of the dorsal roots (S input), at both intensities (1.5×T and 10×T) and frequencies (25 Hz and 50 Hz), generated coAPs in the ventral roots that became progressively smaller with successive pulses of stimulation (
Adapting sensory and motor inputs generate a steady motor output. The response to electrical stimulation (5 pulses at 25 Hz) of the dorsal roots (
When sensory synaptic responses were recorded intracellularly in single motoneurons, the cells initially fired action potentials, but later failed with subsequent pulses (
The depression seen in the response to sensory inputs could result from one or more of several factors: (1) sensory axons fail to fire at the stimulation frequencies used, (2) activation of polysynaptic inhibitory pathways in the cord affects later pulses than earlier ones, and/or (3) depression of synaptic transmission caused by depletion of vesicular neurotransmitter release at the terminal. The following experiments aimed to separate and quantify these factors.
First, to test if the observed depression is due to the failure of sensory axons in conducting APs at the stimulating frequencies, we used a recording electrode on the dorsal root (labeled ‘C’ in
Depression of the sensory inputs is partially due to activation of inhibitory pathways, not stimulation failure. The dorsal root potentials recorded in response to electrical stimulation of the roots’ distal end at 1.5×T
Second, to test the contribution of inhibitory pathways to the measured sensory depression, strychnine and picrotoxin (selective blockers of glycine and GABA) were administered to block inhibitory synaptic transmission. In presence of strychnine and picrotoxin, the sensory input showed less depression (
To activate the motor input, electrical stimulation was delivered to the ventral surface of the spinal cord below the lumbosacral enlargement next to the midline. A similar protocol of 5-pulse trains at intensities of 1.5×T or 10×T and frequencies of 25 or 50 Hz was used. Electrical stimulation evoked coAPs in the ventral roots that gradually increased in amplitude (
Summary of the responses to individual and integrated sensory and motor inputs at 1.5×T/25 Hz stimulation. Summary of responses in the ventral roots
Summary of the responses to individual and integrated sensory and motor inputs at 10×T/25Hz stimulation. Summary of responses to 10×T/25 Hz electrical stimulation of the sensory and descending pathways (The layout is similar to
In single motoneurons, motor EPSPs evoked by a train of stimulation gradually increased in amplitude, which sometimes resulted in generation of action potentials with the later pulses (
The degree of motor facilitation was more profound in the coAPs compared to EPSP synaptic potentials (in
The data above show that electrically evoked sensory and motor inputs to motoneurons have different synaptic plasticity patterns in response to a train of electrical stimuli. We then investigated the characteristics (amplitude and profile) of the motor output generated from concurrent activation of both inputs (S&M). When dorsal roots and descending fibers were stimulated simultaneously, the S&M coAPs at the ventral roots exhibited less adaptation and had higher amplitudes than those generated from either input separately (
At the synaptic level, the S&M EPSPs also exhibited a non-adapting pattern: Their amplitudes did not change at any pulse for 1.5×T and 10×T intensities [
To quantify the integration of inputs, we compared the amplitudes of coAPs resulting from simultaneous stimulation of both pathways (referred to as ‘S&M’) to the linear sum of amplitudes of separate S and M coAPs (referred to as ‘S + M’). At the system level, the amplitudes of S&M coAPs were significantly larger than those of S + M coAPs at all five pulses at 1.5×T (
To this point, our results show that: (1) The S input exhibits depression at the synaptic (EPSP data) and system (coAP data) levels; (2) the M input exhibits facilitation at the synaptic and system levels; (3) there is a disproportion in the magnitude of induced plasticity between the synaptic and system levels for both inputs, in that plasticity is more profound at the system level than at the synaptic level; (4) simultaneous activation of both inputs (S&M condition) generates a stronger and steadier (i.e., less adaptive) responses at the synaptic and system levels; and (5) there is a discrepancy in summation of S and M inputs between the synaptic and system levels, in that EPSPs summate sublinearly, whereas coAPs summate supralinearly.
Our experimental data revealed that sensorimotor inputs (S&M) always summate supralinearly at the system level (
We used the computational model of the spinal motor pool published in
Computational model showing the effect of changing the firing threshold of individual motoneurons on the summation of sensory and motor inputs. Data obtained from a computational model of a motor pool (50 cells) which is stimulated using two separate uniformly-distributed synaptic inputs, and generates an output analogous to the experimental data.
To test our hypothesis on the effect of the motoneuron firing threshold, we varied the firing threshold of individual motoneurons in the pool model to different levels while measuring the amplitudes of the resulting S&M coAPs and EPSPs. Each level was compared to the amplitudes from separate inputs at that level (
At all the tested frequencies and intensities, the summation of the sensorimotor EPSPs was sublinear. This effect could result from: (1) an increase in the effective cell conductance, and/or (2) a decrease in the driving force of the synaptic current. The following set of experiments and simulations examined these potential mechanisms.
First, to test whether synaptic activation changed the motoneuron input conductance, we followed the method of
The activation of sensory and/or motor inputs does not cause a significant change in motoneuron conductance. Voltage measurements versus injected current for a motoneuron in response to injected current alone (open circles), and injected current combined with sensory input activation (S, squares), motor input activation (M, triangles), or combined sensorimotor input (S&M, filled circles). The solid lines indicate the best linear fit to the data points. Effective synaptic current for any of the synaptic inputs is equal in magnitude, but opposite in sign to the injected current at which V (injected + synaptic) = 0. The slopes of the fit lines are not different (Linear regression,
Second, to investigate the effect of changing the driving force of the synapse, we measured different characteristics of the EPSPs as described in
Dependence of EPSP summation on driving force.
Taken together, the data show that sublinear summation of sensory and motor EPSPs at the synaptic level is caused by changes in local driving force, and not by increased input conductance of the cell.
The level of neuromodulation sets the excitability level of both presynaptic and postsynaptic neurons in the spinal network and thus affects patterns of synaptic plasticity (
Recent studies show that pharmacological neuromodulation improves the response to electrical stimulation after SCI in
The effect of methoxamine on plasticity and integration of sensorimotor inputs.
Interestingly, methoxamine also prolonged the decay of EPSPs (compare the falling phases of the EPSPs before and after methoxamine in
The current study provides comprehensive investigation of the plasticity, integration, and neuromodulation of electrically evoked sensorimotor inputs in spinal motoneurons, and the resultant motor output in absence of supraspinal inputs. Using electrophysiological recordings and computer simulations, we show that integration of electrically evoked sensory and motor inputs, despite having different plasticity patterns, help generate a stronger and steadier motor output, which is more readily achievable at higher neuromodulatory states. Our data revealed, for the first time, contrasting types of summation between the synaptic and system levels. In motoneurons, sensory and motor EPSPs undergo sublinear summation due to reduction in the driving force of the sensory and motor synaptic currents during their concurrent activation. Nonetheless, the amplitude of the generated sensorimotor EPSPs are large enough to maintain the motoneuronal membrane potential above the firing threshold; thereby increasing the number of motoneurons recruited by each stimulus in the train. This leads to two functional outcomes: The coAPs evoked by motoneurons at the ventral roots, compared to those generated by either input separately, become larger and steadier in amplitude throughout the stimulation train. This leads to a stronger and more stable spinal motor output. Accordingly, these results provide, for the first time, mechanistic explanation for the cellular processes contributing to the functional motor improvement observed in subjects with complete SCI when electrical stimulation is delivered. Additionally, these data could potentially guide and/or refine the design of more effective stimulation protocols in patients with SCI, ultimately improving the restoration of motor control and patient independence.
Repeated activation of synapses can result in either gradual increase or decrease in the resulting synaptic current in the postsynaptic cell. This phenomenon is known as use-dependent (or short-term) plasticity, which is a hallmark of synaptic transmission in the nervous system (
The motor inputs were activated by stimulation of the local motor circuits and remaining descending axons in the ventral funiculus of the sacral cord. Only a few descending tracts reach the sacral cord, including the LVST (
The synaptic potentials and effective synaptic currents generated by the motor input were generally smaller than those generated by the sensory input. This is in agreement with studies of cat motoneurons which showed that cells with high input resistance have smaller effective synaptic currents from the LVST than from Ia afferents (
Upon successive stimulation of the descending inputs, motor EPSPs showed facilitation, resulting in gradual enhancement of coAPs in the ventral roots. Interestingly, the membrane potential between EPSPs at different pulses was depolarized, and sometimes remained elevated for few seconds after the train. The facilitation of the descending response, thus, could be due to: (1) short-term synaptic facilitation (STF), caused by gradual accumulation of Ca2+ in the synaptic terminals, or (2) increased background network excitation resulting from summation of asynchronous EPSPs (
Of note, with either input, the ventral root coAPs show larger changes than the EPSPs recorded in single motoneurons (discussed below). When the two inputs are combined, two features of the resulting sensorimotor response are notable. First, opposite summation styles between the EPSPs (synaptic level) and coAPs (system level) is observed: sublinear summation of EPSPs vs. supralinear summation of coAPs. Second, the motor output of the integrated sensorimotor inputs is larger and steadier (i.e., without adaptation during stimulation) than the individual responses.
Sublinear summation of EPSPs has been reported in different neuronal types (
Despite this sublinear summation of EPSP amplitudes at the synaptic level, the increased amplitude of the integrated S&M EPSPs is large enough to maintain the motoneuron membrane potential above the firing threshold. Thus, the probability of the motoneuron firing APs by electrical stimuli becomes much higher than the sum of the probability of each separate input (i.e., supralinear summation). In other words, the motoneuronal firing threshold filters out subthreshold sublinearly summating synaptic events, and only converts probabilistic supra-threshold EPSPs into system events, effectively acting as an amplitude-selective filter separating the synaptic and system levels. This explains the mismatch in summation and plasticity between EPSPs and coAPs observed in the present study. This also explains the effectiveness of including proprioceptive sensory feedback to induce stepping and standing in patients with complete SCI via spinal cord stimulation (
In this study, we examined the effect of neuromodulation on short-term synaptic plasticity of the sensory and motor inputs to spinal motoneurons. Methoxamine, the α1-adrenergic receptor agonist, has been shown to increase the excitability of the spinal motor networks (
This study mechanistically highlights the role of sensorimotor integration in generating steady motor outputs. Our data could explain the need to combine spinal cord electrical stimulation with peripheral sensory feedback (generated during motor training) to successfully induce stepping and standing in patients with complete SCI. In addition, our results support that this effect could be boosted by pharmacological neuromodulators. The data also suggest that dorsal root stimulation might be combined with spinal cord stimulation to improve the clinical outcome in patients who failed to independently stand and step with epidural stimulation alone. Taken together, the mechanistic understanding provided by these results is expected to guide the field in designing more refined and effective electrical stimulation interventions, with the ultimate goal of maximizing restored movement, quality of life, and independence in SCI patients.
All datasets generated for this study are included in the manuscript and/or the
This study was carried out in accordance with the recommendations of Wright State University Animal Care and Use Committee. The protocol was approved by the Wright State University Animal Care and Use Committee.
AM and SE conceived the presented idea, discussed the results, and verified the analytical methods. AM planned and carried out the experimental work, and drafted the manuscript. MM and SE developed the computer models and performed the computer simulations. AM, MH, and SE edited and approved the final manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Supplementary Material for this article can be found online at: