The present study investigated the effect of trans-spinal direct current (tsDC) on the firing rate, pattern, and amplitude of spontaneous activity of the tibial nerve and on the magnitude of cortically elicited triceps surae (TS) muscle contractions. The effect of combined tsDC and repetitive cortical electrical stimulation (rCES) on the amplitude of cortically elicited TS twitches was also investigated. Stimulation was applied by two disk electrodes (0.79 cm2): one was located subcutaneously over the vertebral column (T10–L1) and was used to deliver anodal DC (a-tsDC) or cathodal DC (c-tsDC) (density range: ± 0.64 to ± 38.2 A/m2), whereas the other was located subcutaneously on the lateral aspect of the abdomen and served as a reference. While the application of a-tsDC significantly increased the spike frequency and amplitude of spontaneous discharges compared with c-tsDC, c-tsDC made the spontaneous discharges more rhythmic. Cortically elicited TS twitches were depressed during a-tsDC and potentiated after termination. Conversely, cortically elicited TS twitches were enhanced during c-tsDC and depressed after termination. While combined a-tsDC and rCES produced similar effects as a-tsDC alone, combined c-tsDC and rCES showed the greatest increase in cortically elicited TS twitches. tsDC appears to be a powerful neurostimulation tool that can differentially modulate spinal cord excitability and corticospinal transmission.
- spinal cord
- motor cortex
direct current (DC) stimulation is a noninvasive technique used to modulate the excitability of the central nervous system (57, 58). When DC stimulation is delivered transcranially, a positively or negatively charged stimulating electrode (anode or cathode, respectively) is positioned at the cortical area to be stimulated while a reference electrode is usually situated at a distance (52). Transcranial DC stimulation is used to modulate the excitability of the motor cortex (12, 14, 44, 48, 57), ameliorate the perception of pain (3, 4, 9, 65), modulate cognitive functions (33, 51, 68), and treat depression (30, 32, 33, 47, 54). The effect of DC stimulation depends on the topography of neurons relative to the applied field (26), interactions between functional neuronal circuits (7, 61), and the polarity of the electrode. For example, whereas cathodal stimulation depresses neuronal activity, anodal stimulation activates neurons (7, 8).
The spinal cord contains various populations of excitatory (34) and inhibitory interneurons (75) that mediate cortical and subcortical inputs. By acting on these interneurons as well as motoneurons and ascending and descending processes, DC stimulation at the spinal level could exert modulatory effects on cortical and subcortical inputs to the spinal cord. Although DC stimulation has been found to improve functional recovery after spinal cord injury (27–29, 60), only a few studies (22, 36) have investigated the effects of trans-spinal direct current (tsDC) on the excitability of spinal neurons, and its effects on corticospinal transmission have never been investigated.
We hypothesized that the application of tsDC on the intact spinal cord (in anesthetized animals) would 1) modulate spinal neurons excitability and activity as a function of electrical field orientation, which is determined by electrode positions and polarity; and 2) be accompanied by changes in corticospinal system outputs. This hypothesis was attractive because the corticospinal system is impaired after either spinal or cortical injury, and finding a neuromodulatory intervention is of paramount importance.
This study aimed to test whether 1) tsDC could modulate the spontaneous activity of spinal motoneurons in a polarity-dependent manner, 2) tsDC could modulate corticospinal transmission, and 3) repetitive cortical electrical stimulation (rCES) could affect spinal cord responses to tsDC. Using a one disk electrode situated subcutaneously over the vertebral column from T10 to L1 and another disk electrode at an extravertebral location (lateral abdominal aspect), the effects of anodal tsDC (a-tsDC) or cathodal tsDC (c-tsDC) were tested on the spontaneous activity and amplitude of cortically elicited triceps surae (TS) muscle twitches. The results showed differential modulatory effects of tsDC polarity on spontaneous activity. Cortically elicited TS twitches were increased during c-tsDC, depressed after termination, decreased during a-tsDC, and then potentiated after termination. In a different set of experiments, the effects of a-tsDC or c-tsDC combined with rCES were tested. Whereas a-tsDC and rCES produced similar effects as a-tsDC alone, c-tsDC and rCES showed the greatest augmentation in cortically elicited TS twitches. Thus, the present data demonstrate a unique pattern of modulation of corticospinal pathway activity by tsDC.
Experiments were carried out in accordance with National Institutes of Health guidelines for the care and use of laboratory animals. Protocols were approved by the Institutional Animal Care and Use Committee of the College of Staten Island. Adult CD-1 mice (n = 33) were used for this study. Animals were housed under a 12:12-h light-dark cycle with free access to food and water.
Animals were anesthetized using ketamine-xylazine (90/10 mg/kg ip), which has been reported to preserve corticospinal evoked potentials (13, 77). Anesthesia was kept at this level using supplemental dosages (∼5% of the original dose) as needed, and animals were kept warm throughout the procedure by a lamp.
The skin covering the two hindlimbs, thoracic and lumbar spines, and skull was removed. On one side, the TS muscle was carefully separated from the surrounding tissue, with care taken to preserve the blood supply and nerves. The tendon of each of TS muscle was threaded with a hook-shaped 0-3 surgical silk, which was then connected to force transducers. Tissue surrounding the distal part of the sciatic nerve was removed. Both the sciatic nerve and TS muscle were soaked in warm mineral oil.
A craniotomy was performed to unilaterally expose the primary motor cortex (M1; usually on the right side) of the hindlimb muscles, which is located between 0 to −1 mm from the bregma and between 0 and 1 mm from the midline (31). The dura was left intact. The exposed motor cortical area was explored with a stimulating electrode to locate the motor point from which the strongest contraction of the contralateral TS muscle was obtained with the weakest stimulus (1).
DC was induced through a gold surface electrode (0.79 cm2, Grass Technologies, West Warwick, RI) situated over the vertebral column from T10-L1. A similar reference electrode (0.79 cm2) was situated over the lateral aspect of the abdominal muscles, as shown in Fig. 1. A layer of salt-free electrode gel (Parker Laboratories, Fairfield, NJ) was applied between the electrodes and the tissue. Cortical stimulation was induced by a concentric electrode (shaft diameter: 500 μm, tip: 125 μm, FHC, Bowdoinham, ME), which was placed over the motor cortex presentational field of the TS muscle. Extracellular recordings were made from the TS branch of the sciatic nerve with pure iridium microelectrodes (shaft diameter: 180 μm, tip: 1–2 μm, resistance: 5.0 MΩ, WPI, Sarasota, FL). Tibial nerve potentials were recorded from the same location (∼3 mm from the TS muscle) in all animals. The proper location was confirmed by penetration-elicited motor nerve spikes, which were correlated with muscle twitches (2).
Muscle force recording.
The hindlimb and proximal end of the tail were rigidly fixed to the base of the apparatus. The knee was also fixed to the base to prevent any movements from being transmitted between the stimulated muscles and the body. The tendon of the TS muscle was attached to force displacement transducers (FT10, Grass Technologies), and the muscle length was adjusted to obtain the strongest twitch force (optimal length). The head was fixed in a custom-made clamping system. Animals were kept warm during the experiment with radiant heat.
Extracellular activity was passed through a standard head stage, amplified (Neuro Amp EX, AD Instruments, Colorado Springs, CO), filtered (bandpass, 100 Hz–5 kHz), digitized at 4 kHz, and stored in the computer for further processing. A power lab data-acquisition system and LabChart 7 software (AD Instruments) were used to acquire and analyze the data.
Polarization and stimulation protocols.
DC was delivered by a battery-driven constant-current stimulator (North Coast Medical, Morgan Hill, CA). A pretest of cortical stimulation consisting of 10 pulses delivered at 1 Hz (intensity: ∼5 V, pulse duration: 1 ms) was used to elicit TS muscle twitches. The intensity of a-tsDC was increased in 30-s steps (0.5, 1, 1.5, 2, 2.5, and 3 mA) over a total duration of 3 min. Thus, the maximal current density = current/area = 0.003/(3.14 × 0.0052) = 38.22 A/m2. To avoid a stimulation break effect, the current intensity was always ramped for 10 s (ramping was also done between steps). During each tsDC step, a test (identical to the pretest) was conducted; this test was repeated immediately (∼10 s) after the termination of tsDC and then again 5 and 20 min later. To avoid complications by excitability changes resulting from current applications, each a-tsDC and c-tsDC protocol was tested in a different group of animals (n = 5 animals/group).
In addition, in two different groups of animals (n = 5 animals/group), paired stimulation was delivered, consisting of rCES (∼5 V, 1 ms, 1 Hz, 180 pulses) combined with either a-tsDC (+2 mA) or c-tsDC (−2 mA). A pretest and three posttests (0, 5, and 20 min after) of cortical stimulation (∼5 V, 1 ms, 1 Hz, 10 pulses) were also performed.
To control for possible effects of conducting the testing procedure during tsDC, we performed experiments (n = 3 animals/group) in which only pre- and posttests were conducted but no tests were performed during tsDC stimulation. The procedure was performed identically to the procedure previously described, in which tsDC was increased in 30-s steps. In addition, to control for possible tsDC-independent effects of rCES used in a paired stimulation protocol, we also performed experiments (n = 2) in which rCES (180 pulses, 1 Hz) was performed alone.
To ensure that the current intensity used to stimulate the motor cortex was not spreading into deeper brain structures and the muscle and nerve responses were driven by the motor cortex, using a microinjection pump (WPI), we injected 1 μl lidocaine (40 μg/μl) over a 4-min period (49) into the motor cortex, which was previously identified. Lidocaine is a local anesthetic that blocks Na+ channels and, thus, prevents action potential initiation and conduction (35).
After mice had been exposed to a-tsDC (n = 2) or c-tsDC (n = 2), segments of the spinal cord (∼1 cm) located directly below the stimulating electrode were dissected and exposed to Hoechst stain to evaluate whether tsDC damaged spinal cord tissue. A similar spinal cord segment from an unstimulated control animal (n = 1) was also analyzed. Tissues were kept overnight (4°C) in 4% paraformaldehyde in 0.1 M PBS and then cryoprotected in 20% sucrose in PBS at 4°C for 24 h. Spinal segments were freeze mounted, cut into 30-μm sections, and placed on poly-l-lysine-coated glass slides. Sections were treated with Hoechst stain (5 μg/ml, Sigma) for 30 min and then washed with PBS four times. Sections were mounted and glass coverslipped using mounting medium. Immunofluorescence was visualized using a Leica TCS SP2 confocal microscope with 405- and 488-nm lasers.
Injection of glycine and GABA blockers.
Spinal cord segments (T13–L3) were exposed by a laminectomy in anesthetized animals (n = 2). The spinal column was clamped, and the gastrocnemius muscle and sciatic nerve were exposed. The muscle was attached to a force transducer, and recording microelectrode was situated as shown in Fig. 1. The spinal cord was injected at the level of L3–L4 with the inhibitory neurotransmitter blockers picrotoxin and strychnine (5 μM in 200 nl/2 min) using a microinjection pump (WPI).
Calculations and statistics.
Cortically elicited TS muscle twitches were calculated as the height of the twitch force relative to the baseline. The results of the pretest, tests during tsDC, and posttests were calculated as the average of 10 responses evoked at 1 Hz. Spike Histogram software (AD Instruments) was used to discriminate and analyze extracellular spontaneous motoneuronal activity. The amplitude and frequency of spontaneous activity were measured as the average activity during a 20-s recording period before and at different points during and after stimulation. One-way ANOVA, repeated-measures ANOVA, and Kruskal-Wallis one-way ANOVA on ranks were used to test differences between the various treatment conditions. Post hoc tests (Holm-Sidak method or Dunn's method) were then performed to compare cortically elicited TS twitches at baseline or during paired stimulation with those after stimulation. In addition, paired t-tests and Wilcoxon signed-rank tests were used to compare the two treatment conditions. All data are reported as group means ± SE. Statistical analyses were performed using SigmaPlot (SPSS, Chicago, IL) and LabChart software (AD Instruments) with the level of significance set at P < 0.05.
To examine the safety of our DC protocol, we used the blue fluorescent Hoechst dye, which is a sensitive determination of cell numbers. There were no alterations observed in the histochemical analysis of the spinal cord after a-tsDC or c-tsDC compared with controls, as shown in Fig. 2.
To ensure that the responses recorded from the muscle and nerve were actually cortically elicited and not due to the spread of current to deeper brain structures, we inactivated the motor cortex by injecting lidocaine (n = 2) into the motor cortex area according to Martin and Ghez (49). Cortically elicited TS muscle contraction and tibial nerve potentials were completely abolished by this procedure, indicating that the cortex is the source driving the muscle and nerve responses (data not shown).
tsDC stimulation modulates the spontaneous activity of the tibial nerve.
To characterize the effect of tsDC on the spontaneous activity of spinal neurons, firing frequency was examined before, during, and after tsDC, as shown in Fig. 3, A (a-tsDC) and B (c-tsDC). As shown in Fig. 3C, a-tsDC increased the firing frequency from a baseline of 3.3 ± 0.3 spikes/s to 8.5 ± 0.5, 66.5 ± 4.9, and 134.2 ± 6.7 spikes/s at +1, +2, and +3 mA, respectively, yielding a significant effect of condition (repeated-measures ANOVA). Immediately after the termination of a-tsDC, the spontaneous firing frequency returned to baseline levels. As shown in Fig. 3D, c-tsDC increased the firing frequency from a baseline of 2.2 ± 0.6 spikes/s to 6.5 ± 3.0, 20.1 ± 3.1, and 93.1 ± 3.8 spikes/s at −1, −2, and −3 mA, respectively, yielding a significant effect of condition (repeated-measures ANOVA). Immediately after the termination of c-tsDC, the spontaneous firing frequency returned to baseline levels and was not statistically significantly different from baseline (P > 0.05).
The a-tsDC effect on spontaneous firing frequency was significantly greater than that of c-tsDC (Kruskal-Wallis ANOVA). Post hoc tests revealed that all three a-tsDC intensity steps induced significantly higher changes in the frequency of spontaneous activity compared with the changes induced by corresponding intensities of c-tsDC (P < 0.05).
Changes in spike amplitude recorded during different intensities and polarities of tsDC were recorded across conditions (at baseline, at each intensity step, and after tsDC was terminated). Amplitude was measured as the height between negative and positive peaks of a spike. Repeated-measures ANOVA showed a significant overall effect of condition on the amplitude of activity recorded during baseline (16.8 ± 0.3 μV), which increased during a-tsDC steps (step of +1 mA mA = 16.7 ± 0.5 μV, step of +2 mA = 63.2 ± 3.0 μV, and step of +3 mA = 484.2 ± 3.5 μV) and then decreased after termination (11.9 ± 0.7 μV), as shown in Fig. 3E. Subsequent post hoc tests showed that the spike amplitude of activity recorded during intensity steps of +2 and +3 mA were significantly higher than baseline activity (P < 0.05). Repeated-measures ANOVA also showed a significant overall difference in the amplitude of activity recorded at baseline (7.0 ± 0.3 μV), during c-tsDC (step of −1 mA = 17.3 ± 1.5 μV, step of −2 mA = 80.4 ± 2.2 μV, and step of −3 mA = 123.7 ± 4.3 μV), and after termination (5.6 ± 0.29 μV), as shown in Fig. 3F. Subsequent post hoc tests showed that the amplitude of activity recorded during steps of −2 and −3 mA was significantly higher than baseline (P < 0.05). Taken together, the increase in spike amplitude and firing frequency reflects an increase in the number of recruited neurons and an increase in the firing frequency of already recruited neurons, respectively. Furthermore, the differences between amplitudes of activity recorded during a-tsDC of +2 mA and c-tsDC of −2 mA and between a-tsDC of +3 mA and c-tsDC of −3 mA were statistically significant (t tests, P < 0.001). Overall, these findings indicate that a-tsDC and c-tsDC affect spinal neuron excitability through different mechanisms.
Spinal motoneurons and excitatory spinal interneurons are under tonic inhibitory inputs from glycinergic and GABAergic spinal interneurons. Therefore, blockade of glycine and GABA receptors would disinhibit the excitatory spinal interneurons, causing them to fire spontaneously in an oscillatory manner and excite motoneurons (16, 42). To further investigate the differential effects of a-tsDC and c-tsDC on spontaneous activity, we generated autocorrelograms for activity induced by these two conditions as well as by injections of glycine and GABA receptor blockers. The results showed tonic activity with no bursting or oscillation during a-tsDC, as shown in Fig. 4A. Conversely, c-tsDC induced bursting as well as oscillatory activity, as shown in Fig. 4B. Similar to c-tsDC, glycine and GABA receptor blockers induced bursting and oscillatory activity, as shown in Fig. 4C. This similarity indicates that c-tsDC and glycine and GABA receptor blockers may share a mechanism of effect, which involves rhythmic-generating circuitry in the spinal cord. Note that oscillatory activity resulted from the injection of glycine and GABA receptor blockers and was faster compared with the oscillation of c-tsDC-induced activity.
tsDC modulated cortically elicited TS twitches.
To address whether tsDC could modulate cortically elicited TS twitches in an intensity- and polarity-dependent manner, TS twitches were elicited by stimulating the motor cortex before stimulation, at five intensity steps during tsDC, and after stimulation (at 0, 5, and 20 min). Repeated-measures ANOVA combined with post hoc tests showed that a-tsDC affects the ability of the motor cortex to elicit TS twitches (P < 0.001). Examples are shown in Fig. 5A. As shown in Fig. 5C, the baseline average of TS twitch peak force was 0.52 ± 0.04 g, which was depressed to 0.35 ± 0.02, 0.32 ± 0.01, 0.34 ± 0.02, and 0.28 ± 0.01 g at intensities of +1, +1.5, +2, and +2.5 mA, respectively. In contrast, immediately after the termination of a-tsDC, cortically elicited TS twitches were significantly increased (1.51 ± 0.12 g), and this increase persisted at 5 min (1.20 ± 0.15 g) and at 20 min (1.9 ± 0.38) after a-tsDC.
Compared with a-tsDC, the application of c-tsDC had an opposite effect on cortically elicited twitches. Repeated-measures ANOVA combined with post hoc tests showed a significant enhancement of cortically elicited TS twitches during c-tsDC and depression after c-tsDC. Examples are shown in Fig. 5B. As shown in Fig. 5D, the average baseline TS twitch peak force was 0.53 ± 0.04, which was enhanced to 1.23 ± 0.08, 1.98 ± 0.13, 2.88 ± 0.13, 4.35 ± 0.14, and 5.28 ± 0.17 g at −1, −1.5, −2, −2.5, and −3 mA, respectively. A depressive effect was seen after the termination of c-tsDC, with a peak force of 0.23 ± 0.10, 0.12 ± 0.12, and 0.12 ± 0.012 g at 0, 5, and 20 min, respectively. Taken together with the a-tsDC results, these data indicate that trans-spinal application of DC can modulate the ability of the motor cortex to elicit activity at the level of the lumbar spine. This modulation depends on the polarity and intensity of the stimulation as well as on the timing of the test relative to the stimulation.
Testing procedures did not change tsDC aftereffects.
To investigate a possible effect of conducting the testing procedure during a-tsDC or c-tsDC, we repeated these experiments (n = 3 animals/group) with only pre- and posttests but no tests during the tsDC stimulation. For a-tsDC, there was no significant difference between conditions that included or excluded testing during the a-tsDC stimulation (H = 5.3, P = 0.06, Kruskal-Wallis ANOVA). In conditions with and without testing during stimulation, a-tsDC induced an immediate increase of TS twitches (301.14 ± 49.33% vs. 366.9 ± 46.9%), which persisted after 5 min (229.59 ± 66.03% vs. 325.9 ± 170.14%) and 20 min (387.87 ± 117.13% vs. 299.6 ± 137.57%). Similarly, there was no effect of the testing procedure on the c-tsDC depressive aftereffect (H = 5.3, P > 0.05, Kruskal-Wallis ANOVA). In conditions with and without testing during stimulation, c-tsDC depressed cortically elicited TS twitches immediately (33.48 ± 6.40% vs. 17.65 ± 6.40%), after 5 min (21.24 ± 3.8% vs. 25.45 ± 2.98%), and after 20 min (23.95 ± 3.44% vs. 25.35 ± 3.0%). These results confirm that the testing procedure used in this study had no effect on the aftereffects induced by a-tsDC or c-tsDC.
Effects of a-tsDC and c-tsDC on latency of cortically elicited tibial nerve potentials.
The latency of cortically elicited tibial nerve potentials was measured before, during, and after a-tsDC and c-tsDC. Only latencies measured at a-tsDC of +2 mA and c-tsDC of −2 mA are presented because no differences were found between the latencies at these intensities and those at other intensities that caused significant increases in TS twitches. However, the mean latency was calculated based on measurements at all time points after tsDC. For a-tsDC, Kruskal-Wallis ANOVA showed a significant effect of time (baseline, during, and after stimulation), as shown in Fig. 6A. Post hoc tests revealed that the latency of cortically elicited tibial nerve potentials was significantly longer during +2-mA stimulation (21.5 ± 0.34 ms) and shorter after termination (17.92 ± 0.21 ms) relative to baseline (19.82 ± 0.17 ms). Similarly, for c-tsDC application, Kruskal-Wallis ANOVA showed a significant effect of time. Post hoc tests revealed that the latency of cortically elicited tibial nerve potentials was significantly shorter during −2-mA stimulation (17.42 ± 0.22 ms) and longer after termination (23.90 ± 1.19 ms) relative to baseline (20.33 ± 0.19 ms). Taken together, these results indicate that tsDC affects the excitability of spinal neurons in a way that changes their ability to respond to the inputs from the motor cortex. In rodents, the motor cortex excites the spinal motoneurons through two indirect, disynaptic pathways: 1) a slower pathway from the cortex to segmental interneurons and/or propriospinal neurons and 2) a faster pathway from the cortex to the medial part of the reticular formation and then to motoneurons (41). Thus, changes in latency during tsDC may reflect changes in the involvement of these pathways or may simply be due to changes in the recruitment pattern of spinal neurons.
Paired rCES and tsDC stimulation.
The motor cortex was stimulated for 3 min (180 pulses, 1 Hz, maximal intensity: ∼5.5 mA) during either a-tsDC (+2 mA) or c-tsDC (−2 mA), as shown in Fig. 7, A and B. Paired rCES and a-tsDC was associated with a significant increase in cortically elicited TS twitches after the termination of stimulation (0.80 ± 0.10 g) compared with baseline (0.39 ± 0.05 g, P < 0.001), as shown in Fig. 7C. Notably, paired rCES and c-tsDC showed a similar increase after termination (3.67 ± 0.51 g) compared with baseline (0.21 ± 0.51 g, P < 0.001), as shown in Fig. 7D. The increase after those two different stimulation paradigms persisted, with no notable changes immediately, at 5 min, and at 20 min after termination. Thus, results presented after the termination represent the average of these three time points. The effect of rCES alone was tested in a separate group of animals (n = 2), and no change was found after termination compared with baseline (t-test, P > 0.05; data not shown).
A total of four stimulation paradigms used in the current experiments affected cortically elicited TS contraction: a-tsDC, c-tsDC, a-tsDC with rCES, and c-tsDC with rCES. Kruskal-Wallis ANOVA showed a significant effect of condition (H = 66.97, P < 0.001). Multiple comparisons showed that paired c-tsDC and rCES was more effective than all other paradigms (2,287.07 ± 342.49%, all P < 0.05), especially for reversing the depressive effect seen after c-tsDC (33.66 ± 9.82%). Paired a-tsDC and rCES showed no significant difference (252.88 ± 30.79%) compared with a-tsDC alone (329.18 ± 38.79%, P > 0.05). These findings indicate that cortical activity had a strong influence on c-tsDC aftereffects; however, it had no influence on a-tsDC aftereffects.
Histological analysis demonstrated no harmful morphological effects of the tsDC parameters used in the present study. The maximal current density used was 38.22 A/m2 for a duration of 3 min, which was within the range that was considered safe in rats (45) and mice (14). In this study, spinal cord stimulation differed from cranial stimulation in three respects: 1) the distance from the electrode surface to the ventral aspect of the spinal cord was ∼7 mm, as opposed to the distance to the cranium of ∼0.3 mm; 2) bone, muscle, and fat tissues were present between the electrode and spinal cord, whereas only bone was present at the cranium; and 3) the volume of the conductor surrounding the target tissue was much larger in the spinal cord than in the brain, potentially deforming the current and reducing its density.
Both a-tsDC and c-tsDC markedly increased the frequency and amplitude of spontaneous tibial nerve activity in an intensity-dependent fashion. Interestingly, a-tsDC was more effective than c-tsDC in increasing firing frequency and recruiting units with larger amplitude. These results are in agreement with data from a-tsDC stimulation of the cerebral cortex (8), hippocampal slices (7), and cerebellum (15). The effects of c-tsDC on neuronal discharges were more complex in three respects. First, c-tsDC only caused significant changes at higher intensities (−2 and −3 mA). Second, c-tsDC did not cause firing of neurons with large spikes but was observed in some experiments to inhibit firing of large spikes (>1 mV) while increasing firing of smaller spikes. Third, as shown in Fig. 3B, c-tsDC evoked rhythmic firing. The c-tsDC-induced increase in firing rate supports previous observations by Bindman et al. (8) in which negative currents occasionally increased the firing rate. During stimulation, a-tsDC depressed cortically elicited TS twitches, whereas c-tsDC markedly potentiated twitches. From immediately after the termination of tsDC until at least 20 min later, cortically elicited TS twitches were markedly potentiated after a-tsDC and depressed after c-tsDC. Interestingly, pairing a-tsDC with rCES (1 Hz) potentiated cortically elicited TS twitches but was not different from a-tsDC alone. Conversely, pairing c-tsDC with rCES potentiated cortically elicited TS twitches and had the greatest effects of any stimulation condition.
Moreover, whereas a-tsDC increased the latency of cortically elicited tibial nerve potentials, c-tsDC decreased this latency. After a-tsDC or c-tsDC stimulation was terminated, the effect on latency was reversed. Changes in latency were observed despite a steady intensity of cortical stimulation, suggesting that factors underlying these changes are not likely to include the switch from a cortical site of activation to a deeper location (62). Instead, these factors may include 1) axonal hyperpolarization (53) by c-tsDC or 2) activation of preferential spinal circuits that mediate corticospinal transmission. In rodents, the corticospinal pathway has two indirect routes: a faster route mediated via reticulospinal neurons and a slower route mediated via segmental interneurons (41). The present findings suggest that c-tsDC may shift the pattern of excitability at the spinal cord toward the faster reticulospinal route.
The differences in the effects of a-tsDC and c-tsDC on neuronal activity suggest that the two conditions affect distinctive neuronal types through different mechanisms. The topography of spinal neurons relative to the direction of current determines the current locus and type of effect (i.e., increase or decrease in excitability) (7, 23, 26). As shown in Fig. 8, a dorsal cathodal current should depolarize neuronal compartments closer to the electrode and hyperpolarize compartments farther from the electrode. Thus, an interneuron with its dendrites and soma at the ventral aspect of the spinal cord and its axon at the dorsal aspect would have a hyperpolarized dendritic tree and soma and a depolarized axon and nerve terminal. Such a neuron would be less responsive to synaptic activation but would have a lower threshold to spontaneously fire an axonally generated action potential. A spinal neuron oriented in the opposite direction would show an opposite response to cathodal stimulation. This argument is supported by the finding that motoneuron responses to dorsolateral and medial funiculus stimulation were facilitated by depolarizing currents in the dendrites and soma but were not affected by hyperpolarizing currents (20), which have also been shown to occur in the hippocampus (7).
Thus, we predict that c-tsDC may cause motoneurons to be more responsive to synaptic activation but less inclined to generate spontaneous activity. This may also explain why cortically elicited TS twitches were potentiated during c-tsDC application. Moreover, presynaptic hyperpolarization has been shown to increase excitatory postsynaptic potentials (EPSPs) (23, 38, 39), and such hyperpolarization is expected to occur in corticospinal tract terminals and spinal interneurons between the corticospinal tract and spinal motoneurons. Thus, nerve terminal hyperpolarization and dendrite depolarization induced by c-tsDC would cause the potentiation of cortically elicited TS twitches. Moreover, cathodal stimulation has been shown to increase the excitability of axons aligned perpendicular to the direction of current (5). Therefore, in the present study, the corticospinal tract, which passes below the cathodal electrode, would be expected to increase axonal excitability and, hence, spinal output. Conversely, the dendrites and soma of motoneurons would be hyperpolarized and axons would be depolarized in response to a-tsDC stimulation. Axonal depolarization at locations that affect voltage-sensitive membrane conductances could increase the firing rate and amplitude of spontaneous activity during a-tsDC. Presynaptic depolarization has been shown to decrease presynaptic nerve action potentials and EPSPs (37, 40), which may play a role in depressing cortically elicited TS twitches during a-tsDC. In addition, hyperpolarization of the soma and dendrites could depress motoneuron responses to cortical stimulation during a-tsDC. Alternative explanations could include 1) increased numbers of refractory motor neurons due to increased spontaneous firing or 2) preferential activation of the spinal or supraspinal inhibitory pathway.
Rhythmic activity was observed during c-tsDC but not a-tsDC, indicating that c-tsDC may have a depressive effect on spinal inhibitory interneurons. Such interneurons might be inhibited because of their topography relative to the applied electrical field. c-tsDC might hyperpolarize both excitatory and inhibitory spinal interneurons. If we assume that inhibitory and excitatory spinal interneurons contain different membrane channels (e.g., fewer low-voltage-activated T-type Ca2+ channels and hyperpolarization-activated cation channels in inhibitory interneurons), then hyperpolarization would silence inhibitory interneurons, thus disinhibiting the excitatory interneurons. In contrast, in spinal rhythmogenic neurons, hyperpolarizing tsDC might activate hyperpolarization-activated, nonselective cation current (Ih) (17). In combination with T-type Ca2+ channels (71), Ih should gradually depolarize the cell membrane to reach the threshold for an action potential, which could be another mechanism (11) mediating c-tsDC-induced potentiation of cortically elicited TS twitches. A direct test of the effect of tsDC on Ih will be investigated in future studies.
Excitable membranes contain large and diverse number of voltage-gated channels (70). For example, many types of voltage-gated Na+ and Ca2+ channels are expressed throughout the neuronal membrane (soma, axon, and dendrites), where they mediate specific functions. Voltage-gated Ca2+ channels are located in every synaptic nerve terminal to regulate neurotransmitter release. Na+ and Ca2+ channels are found in dendritic processes of neurons, where they contribute electrically to augment synaptic inputs (73, 74). Specifically, the presence of TTX-sensitive slowly inactivating Na+ channels in the dendrites of neurons can enhance transmission of postsynaptic glutamate currents (64, 66). Another example is TTX-resistant Na+ channels like PN1 (21), which responds to subthreshold depolarizing currents, amplifying subthreshold inputs. Widely distributed in the nervous system, including in the lumbar spinal cord (72), are N-methyl-d-aspartate (NMDA) receptors. The NMDA receptor regulates ligand- and voltage-sensitive Ca2+ channels and could be influenced by tsDC, playing a key role in the plastic changes found in our study. Thus, as shown by the hypothetical diagram in Fig. 8, a depolarizing tsDC would strengthen the spread of synaptic input in motoneuronal dendritic processes. This could be one of the mechanisms mediating c-tsDC action.
Subthreshold depolarization or hyperpolarization affects the probability of opening of a voltage-gated channel by changing activation (m) and inactivation (h) coefficients. Therefore, the ensuing stimulus current (I) depends on the value calculated by the following equation: I = mh(Gmax)(V − Erev). In this context, Gmax is the specific maximal conductance, V is the membrane potential, and Erev is the reversal potential. According to this relationship, the voltage strongly alters the number of channels that will be available to open when the membrane potential changes by altering m and h and also the driving force on the ions (35). In conclusion, more or fewer channels will open if the membrane has been for while either hyperpolarized or depolarized, respectively. This might partly explain the difference between the depression and amplification of cortically elicited TS twitches during a-tsDC and c-tsDC, respectively.
According to the cell position relative to the tsDC-induced field and cellular morphology besides depolarization, the membrane can be also hyperpolarized. This hyperpolarization will influence cells differently depending on a specific composition and distribution of voltage-gated channels contained in neuronal membranes. For example, A-type K+ (KA) channels are activated when the cell is depolarized after a period of hyperpolarization. This hyperpolarization gradually removes the inactivation of KA channels. This repriming of KA channels will allow this channel to open when the membrane is depolarized in response to an excitatory stimulus. Subsequently, this tsDC-induced KA current cancels the stimulus current, reducing cell excitability.
Persistent subthreshold depolarization caused by tsDC application could create a new steady state of voltage-gated channels activity with an altered excitability. In addition to its direct effect on neuronal responses, this new excitability would change the ionic composition of the cytoplasm. Alterations in the intracellular concentrations of ions like Ca2+ and Na+ influence gene transcription and the protein synthetic machinery of the cell (69, 76), potentially contributing to long-term changes in cell function. Therefore, these channels and their influence on excitable cell physiology constitute a possible target for tsDC action.
In the present study, cortically elicited TS twitches were depressed after c-tsDC and potentiated after a-tsDC. DC stimulation of the brain has similar results, as anodal stimulation increases, whereas cathodal stimulation decreases, the excitability of the motor cortex in humans (56, 57, 59) and in mice (14). Anode-induced excitability appears to depend on membrane depolarization, whereas cathode-induced depression depends on membrane hyperpolarization. In addition, aftereffects of both anodal and cathodal stimulation involve the NMDA glutamate receptor (46, 55). In the spinal cord, L-type Ca+2 channels present in motoneuron dendrites mediate the facilitatory action of depolarizing currents (36, 67). In addition, we suggest that the pattern of a-tsDC-induced polarization might activate depression-mediating mechanisms, such as retrograde signaling by endocannabinoids, which selectively depress inhibitory presynaptic terminals (25). However, the exact cellular mechanisms mediating DC stimulation aftereffects are not clear (6, 10). Notably, mechanisms mediating the depressive aftereffects of cathodal DC stimulation are completely unknown. Future studies are necessary to delineate the cellular mechanisms of depression after c-tsDC termination.
In the present study, pairing rCES with c-tsDC not only prevented the depression of cortically elicited TS twitches after c-tsDC termination but remarkably increased twitches. It seems most likely that c-tsDC induces a polarizing pattern, as shown in Fig. 8, including presynaptic hyperpolarization and postsynaptic depolarization within the corticospinal pathway. This pattern, combined with rCES, would evoke long-term potentiation. Specifically, presynaptic hyperpolarization has been shown to increase the size of EPSPs (19, 23, 24, 38), which would subsequently increase neurotransmitter release and thereby cortical input. Although a low-frequency stimulation was applied to the motor cortex in the present study, the actual frequency of cortical input was probably much higher (63). In addition, postsynaptic depolarization would activate the NMDA receptor. The association between the presynaptic increase of neurotransmitter release and steady postsynaptic depolarization would trigger the induction of long-term potentiation (18). This could serve as the main mechanism for the c-tsDC-induced enhancement of cortically elicited TS twitches observed in the present study. Furthermore, reduction of inhibitory inputs to spinal circuits could also mediate the aftereffects of paired rCES and c-tsDC.
The present study describes a novel use of DC stimulation to augment spinal responses to cortical stimulation. In many neurological disorders, connectivity between the cortex and spinal cord is compromised (e.g., spinal cord injury or stroke), and stimulation protocols similar to those in the present study could strengthen spinal responses. The present data clearly show that neuronal activity is important in shaping c-tsDC aftereffects. Specifically, c-tsDC should optimize corticospinal activity during stimulation but should depress it at other times. The ability of c-tsDC to interact with cortical activity to cause different outcomes is an interesting phenomenon that could support many clinical uses of c-tsDC. Translating this to rehabilitative strategies would require either artificial cortical stimulation (when voluntarily muscle activation is impossible) or voluntary training during the application of c-tsDC. Moreover, the depressive aftereffects of c-tsDC could be used to manage spasticity resulting from many neurological disorders, as indicated by Elbasiouny and Mushahwar (26). Another important practical observation seen in our findings is that spinal neurons were recruited in an orderly fashion as a function of tsDC intensity. In other words, tsDC elevated the excitability of spinal circuits yet maintained the normal pattern of motoneuron recruitment, which is important to control the speed and force. As shown in Fig. 3, spike amplitude and frequency seemed to follow the size principle (50) in which small motoneurons were activated first with the progressive addition of large motoneurons. This distinguishes the tsDC-induced activity from drug-induced activity (43). In practice, tsDC would be suitable to increase the excitability of spinal neurons to produce a sequential activation of muscles. Therefore, tsDC has many potential uses in rehabilitative therapy applied alone or in combination with rCES.
This work was supported by New York State Department of Health Grant CO23684 and Professional Staff Congress-City University of New York Grant 60027-37-39.
No conflicts of interest, financial or otherwise, are declared by the author.
- Copyright © 2011 the American Physiological Society