Electrophysiologic Evidence for Increased Endogenous GABAergic but Not Glycinergic Inhibitory Tone in the Rat Spinal Nerve Ligation Model of Neuropathy

Male Sprague-Dawley rats (Central Biological Services, University College London) weighing 120–160 g at the time of the surgery were used. Rats were housed in groups of five in plastic cages in artificial lighting with a fixed 12-h light–dark cycle. Laboratory chow and water were available ad libitum . Guidelines for animal research by the United Kingdom Home Office and the Council of the American Physiologic Society were followed. The spinal nerve ligation (SNL) model of neuropathic pain was used. 25In brief, the animals were anesthetized with halothane (Fluothane; Zeneca, Macclesfield, United Kingdom) in nitrous oxide and oxygen (50:50). The left L5 and L6 spinal nerves were exposed by removing a small piece of the paravertebral muscle and a part of the left transverse process of the L5 lumbar vertebra. The L5 and L6 spinal nerves were then carefully isolated and tightly ligated with 6-0 silk. A sham operation was performed by exposing but not ligating the spinal nerves. After checking for hemostasis, the muscle, the adjacent fascia, and the skin were closed with sutures.

The behavioral signs of mechanical and cold allodynia were measured with a series of von Frey filaments and by application of a drop of acetone as previously described. 26In brief, the rats stood on a metal mesh platform, and the plantar surface of the paw was touched with different von Frey filaments (Semmes-Weinstein monofilaments; Stoelting, Wood Dale, IL) with a bending force of 0.2–15 g. Each filament was applied five times at 5-s intervals until the weakest filament that induced paw withdrawal on more than half the occasions it was presented was found. Cold allodynia was measured as the number of brisk foot withdrawal responses after five consecutive applications of a drop of acetone to the plantar surface of the paw. The assessment of mechanical and cold sensitivity was performed in the SNL and sham-operated rats 7 and 14 days after the operation to verify that the animals entered to the electrophysiologic study exhibited mechanical and cold allodynia.

The electrophysiology was performed 15–18 days after the surgery and on nonoperated control animals of similar size, as described previously. 27,28The rat was anesthetized with halothane (3% during induction, 1.5% during the recordings) in nitrous oxide and oxygen (66%:33%), and the trachea was cannulated. No neuromuscular blockers were used. The rat was secured in a stereotaxic frame, and a laminectomy was performed over lumbar segments L1–L3. The spine was clamped rostral and caudal to the laminectomy, and the dura covering the exposed part of the spinal cord was removed. Parylene-coated tungsten electrodes were used to make single-unit extracellular recordings of dorsal horn neurons receiving afferent input from the hind paw. The electrode was lowered into the spinal cord using a SCAT microdrive (Digitlmer, Welwyn, UK), which allowed measurement of the depth of the neuron relative to the surface of the dorsal horn. Gentle tapping of the plantar surface of the hind paw with the fingers was used as the search stimulus for finding wide-dynamic-range neurons.

Transcutaneous electrical stimulation of the receptive field was applied at three times the C-fiber threshold, and poststimulus histograms to a train of 16 stimuli at 0.5 Hz with a 2-ms pulse width were constructed (Spike 2 software, C.E.D. 1401 interface; Cambridge Electronic Design, Cambridge, UK). The resulting Aβ-, Aδ-, and C-fiber–evoked responses were separated by latency (0–20, 20–90, and 90–300 ms, respectively) and quantified. In addition, after discharge, the activity after the main C-fiber–evoked band between 300 and 800 ms was recorded and quantified. Poststimulus histograms were constructed at 10-min intervals before administration of the drug until a stable (< 10% change between tests) response was obtained, and thereafter for 40 min after each drug dose. For analyzing wind-up, the initial response of the neuron to the first stimulus in the train and the final response of the neuron after the repeated 16 stimuli in the train were measured over the C-fiber and postdischarge latencies (90–800 ms).

Bicuculline methobromide was obtained from Tocris Cookson Ltd. (Bristol, United Kingdom), and strychnine-HCl was obtained from RBI (Poole, United Kingdom). Both drugs were dissolved in 0.9% NaCl. The vials containing strychnine solutions were kept wrapped in aluminium foil to prevent exposure to light. All drugs were stored at 4°C. The doses of bicuculline and strychnine were administered on the spinal cord in volume of 50 μl using a glass Hamilton microsyringe in a cumulative manner, with a 1-h interval between each dose. The number of neurons recorded was 5–12 for each drug dose in each study group. The exact numbers are given in the figure legends.

Normally distributed continuous variables, such as the properties of the recorded neurons, were analyzed using analysis of variance followed by the Dunn post hoc  test when appropriate. Discrete variables, such as the measures of mechanical and cold sensitivity, were analyzed using the Mann–Whitney U test for paired comparisons, and the Wilcoxon signed rank and Kruskal-Wallis tests were used for comparisons between the groups, as appropriate. Significance was set at P  less than 0.05.

The SNL group rats developed mechanical and cold allodynia in the operated paw. The von Frey filament force that induced paw withdrawal in SNL rats 7 days after the operation was 7.0 ± 1.5 g (median and median absolute deviation) and 8.5 ± 3.2 g 14 days after the operation, whereas applying the strongest filament used (15 g) did not cause paw withdrawal in any of the rats in the sham-operated group at either 7 (P < 0.001) or 14 days after the operation (P < 0.001). No paw withdrawal responses to mechanical stimulation occurred in the paw contralateral to the operation in either the SNL or the sham-operated rats. In the acetone drop test 7 days after the operation, the median number of responses was 1.5 in the operated paw (range, 0–4) and 0 in the sham-operated group (P < 0.001), and 14 days after the operation was 1.0 (range, 0–4) in the operated paw and 0 in the sham-operated group (P < 0.001). No paw withdrawal responses to the acetone stimulation occurred in the paw contralateral to the operation in either the SNL or the sham-operated rats. These behavioral manifestations of nerve injury did not differ between the rats used in the bicuculline and strychnine studies (data not shown).

All neurons recorded were wide-dynamic-range type and had receptive fields located on the plantar surface of the hind paw. The baseline characteristics of the neurons recorded are described in tables 1 and 2. Neurons that showed constant spontaneous activity during the baseline recordings were not used in this experiment.

Bicuculline significantly increased the Aδ-fiber–evoked activity in a dose-related manner in all three experimental groups (fig. 1). The peak of this effect was observed 20 min after each dose. There were no significant differences between the SNL rats and either of the control groups with respect to the effect of bicuculline. The C-fiber–evoked activity was significantly increased with bicuculline in the SNL animals but not in the control groups (P = 0.008 for SNL vs.  nonoperated, and P = 0.004 for SNL vs.  sham with 50 μg bicuculline;fig. 1). Bicuculline did not significantly alter the Aβ-fiber–mediated activity or the postdischarge in any of the experimental groups (fig. 1).

The initial neuronal response was increased by bicuculline in a dose-related manner in all three study groups (fig. 2). There were no statistically significant differences among the study groups (fig. 2). The final response after 16 consecutive electrical stimuli (0.5 Hz) was significantly increased after bicuculline in the group with SNL-induced neuropathy but not in the sham-operated or nonoperated control animals (fig. 2;P = 0.05 for SNL vs.  sham and 0.05 for SNL vs.  normal). In the SNL group, 4 of the 15 neurons started to fire spontaneously after bicuculline administration, with a mean peak firing rate of 12 ± 4.7 Hz with 0.5 μg bicuculline (n = 3), 14 ± 7.0 Hz with 5 μg bicuculline (n = 3), and 17 ± 10 Hz with 50 μg bicuculline (n = 4). In the sham-operated control group, bicuculline administration caused 2 of 16 neurons to fire spontaneously, with a mean peak firing rate of 1.2 Hz with 0.5 μg bicuculline (n = 1), 14 Hz with 5 μg bicuculline (n = 2), and 18 Hz with 50 μg bicuculline (n = 2).

Because intrathecal strychnine did not produce any effect in the pilot experiments in normal animals, only SNL rats and nonoperated control animals were studied. Strychnine did not significantly alter the activity mediated by any of the fiber types (fig. 3). The initial response to the first electrical stimulus of the train and the final response after all 16 stimuli (see above) were not changed by strychnine in either group of animals. However, in a number of animals, segmental motor responses, such as spontaneous tail flicks or muscle twitches in the body occurred, although there were no changes in the activity of the dorsal horn neurons.

The mechanical allodynia that occurred in the SNL group after the nerve injury was comparable to the results reported previously. 25,26,28,29In the electrophysiologic recordings, the GABA^Aantagonist bicuculline produced statistically significant increases in the Aδ-fiber–evoked activity in SNL and sham-operated and nonoperated control rats, and increased C-fiber–mediated activity in the SNL group but not in either of the control groups. The glycine-receptor antagonist strychnine did not have any statistically significant effect on any of the parameters studied in either SNL rats or nonoperated control animals.

In an earlier electrophysiologic study in normal rats, bicuculline facilitated significantly Aδ- and C-fiber–evoked responses in a dose-dependent manner. The facilitation of the C-fiber–evoked response was statistically significant but small compared with the profound potentiation observed on the Aδ-fiber–evoked response. 30This would agree with findings reported in normal rats with spinal transection, where the benzodiazepine midazolam, which potentiates GABA^Areceptor function, caused only a weak depression of C-fiber–evoked responses but had a marked dose-dependent depressive effect on the Aδ-fiber–evoked response of the spinal cord wide-dynamic-range neurons. 31This GABA^Aeffect on Aδ-fiber–evoked responses may have a morphologic basis: GABAergic terminals contact many more Aδ-fiber terminals than nonglomerular C-fiber terminals and form no contacts at all with glomerular C-fiber terminals. 32In a recent study, the benzodiazepine agonist midazolam, acting spinally, reduced Aδ-evoked responses in a dose-dependent manner in both SNL and control animals but inhibited C-fiber–mediated activity significantly only in the SNL-injured animals. 33Overall, these results perfectly mirror the present findings in that the profile of midazolam, which enhances GABA transmission, was the exact opposite of the effects of bicuculline, which blocks the actions of GABA at the GABA^Areceptor. Thus, it can be surmised that GABAergic neurons are subject to plasticity after nerve injury. In the present study, the profound facilitation of the Aδ-fiber–evoked responses by bicuculline was unchanged in all study groups. In contrast, an enhancement of the C-fiber responses and the final response, a measure of the enhanced activity in a neuron produced by wind-up, was observed only after nerve injury. The overall evoked response after C-fiber stimulation was increased in these animals, although the initial response, a measure of the ability of the first C-fiber volley to evoke a response, was unaltered (fig. 2). Thus, it appears likely that the enhanced C-fiber responses and final responses produced by the drug after nerve injury result from postsynaptic mechanisms, possibly increased GABA^A-mediated controls acting on interneurons in polysynaptic nociceptive C-fiber pathways. Figure 4is a schematic diagram illustrating the proposed changes in the spinal GABAergic systems in neuropathy. The marked GABAergic inhibitory tone on the Aδ-fiber–evoked responses is not altered in neuropathy, whereas the less pronounced GABA control of C-fiber responses is enhanced, which may reflect changes in the polysynaptic C-fiber pathways after nerve injury.

In behavioral experiments, intrathecal administration of both bicuculline and strychnine produce allodynia-like behavior, facilitation of nocifensive flexor reflexes, and cardiovascular responses similar to those produced by nociceptive stimuli.^10–14In the present study, administration of strychnine (1.0–100 μg intrathecally) did not significantly change the responses of the wide-dynamic-range neurons in the dorsal horn of the spinal cord in the electrophysiologic recordings. This discrepancy between the previous behavioral studies and the electrophysiologic findings in the present study could be a result of several factors, including the type of stimulation used, the species, the population of cells studied, and the depth of anesthesia. However, bicuculline was effective under the same conditions.

Strychnine-induced allodynia has been shown to be selective for nonnoxious light tactile stimulation 34,35and initiated by primary afferents not normally involved in nociception, possibly Aββfibers 13,34. The fact that intrathecal strychnine does not produce hyperalgesia to mechanical, thermal, or chemical noxious stimuli further indicates the independence of nociceptive pathways and strychnine-sensitive afferent inputs to the spinal cord. 13,34In the present study, suprathreshold electrical stimulation at three times the C-fiber threshold was used to study changes in both the C- and the Aδ-fiber–mediated effects. Strychnine applied near the dorsal horn neurons in α-chloralose–anesthetized cats through a dialysis fiber causes a significantly enhanced response to hair follicle activation only. 15Similar treatment with bicuculline results in increased responses to hair deflection, low- and high-threshold tonic mechanical stimulation of the receptive field, and increased background discharge in nociceptive dorsal horn neurons. 12 

The behavioral allodynia produced by spinal strychnine is segmentally localized and can be prevented with intrathecally administered glycine, 13which would indicate a spinal effect. However, the cardiovascular responses produced by intrathecal administration of strychnine can be observed only in lightly anesthetized animals, but not in animals during surgical anesthesia. 13We required anesthesia depth at surgical levels for electrophysiologic recordings. Despite this, some animals showed segmental spontaneous and, above all, evoked motor responses to low-threshold stimuli, such as touching the animal’s tail. It is conceivable that a large part of the agitation in the behaving animals could result from blockade of the recurrent inhibitory glycinergic interneurons in the motor response arc, the Renshaw neuron rather than a selective effect on sensory neurons.

In conclusion, bicuculline produced dose-related facilitations of the Aδ-fiber–evoked activity in all study groups and significantly increased C-fiber–mediated activity in the SNL group. This suggests a modality-dependent shift in the inhibitory controls exerted by GABA after nerve injury, possibly as compensation for increased C-fiber–evoked excitability. This would indicate that the extent of hyperalgesia in neuropathy is modulated by endogenous spinal GABA neurons. There were no differences in the effect of bicuculline on low-threshold responses among the study groups, indicating that GABA^A-mediated control of low-threshold inputs is not altered by nerve injury, at least under these conditions. These findings would suggest that the purported loss of GABA controls on low threshold inputs in neuropathy, proposed to be a basis for allodynia, is unlikely. The morphologic evidence for reduced level of GABA in neuropathy may, in fact, result from increased release of the transmitter rather than a downregulation of these inhibitory controls. This, in turn, may partly be responsible for the reduced baseline responses of the spinal neurons in the SNL model of neuropathic pain. These results and our previous findings with midazolam would suggest that a reassessment of GABAergic mechanisms in neuropathic pain is needed.