The Effect of Intrathecal Gabapentin and 3-Isobutyl gamma-Aminobutyric Acid on the Hyperalgesia Observed After Thermal Injury in the Rat

Clinical work has shown that a structural analog of gamma-aminobutyric acid (GABA), gabapentin (neurontin), can significantly alter the hyperpathic state observed after nerve injury states ^[1-3]. These observations in humans were confirmed in animal models of nerve injury, in which the systemic and considerably smaller doses given spinally could dose-dependently reverse the tactile allodynia in the Chung model of neuropathy ^[4] and the thermal hyperalgesia in the Bennett model (W.H. Xiao and G. Bennett, personal communication). The functional specificity of this action seems not to be limited to altered sensory processing induced by nerve injury. Recent work has shown that gabapentin can reverse the second phase of the formalin test ^[5] and diminish the hyperalgesia induced by intrathecal injections of substance P (Partridge, Chaplan, and Yaksh; unpublished observations). This ability to alter these hyperalgesic components suggests that gabapentin may exert an effect on facilitated processing generated by the persistent small afferent discharge evoked by tissue injury. We sought to specifically address this issue by assessing the spinal actions of gabapentin on the hyperalgesia induced by thermal injury to the plantar surface of the paw.

To further define the mechanisms of action of this agent, we also examined the effects of two gabapentin analogs, S(+)-3-isobutyl GABA and R(-)-3-isobutyl GABA. The S(+) isomer displaces gabapentin binding in the nanomolar range and, like gabapentin, is a potent anticonvulsant. In contrast, the stereoisomer is without affinity or anticonvulsant activity ^[6-8]. Finally, the anticonvulsant effects of gabapentin are reversed by D-serine, but not L-serine ^[9]. D-Serine acts as an agonist at the nonstrychnine-sensitive glycine site ^[10]. This site serves to enable the N-methyl-D-aspartate receptor. We sought to determine whether this ability of D-serine to stereospecifically reverse the effects of gabapentin also occurred with respect to its antihyperalgesic effects and whether the antagonism extended to the effects of S(+)-3-isobutyl GABA.

Methods

The following investigations were carried out under a protocol approved by our institution's animal care committee. Male Holtzman Sprague-Dawley rats (300-400 g; Harlan Industries, Indianapolis, IN) were individually housed and maintained on a 12-h light/12-h dark cycle. Animals had free access to food and water at all times. Chronic intrathecal catheters were implanted under halothane anesthesia according to a modification of the method described by Yaksh and Rudy ^[11]. Briefly, under halothane anesthesia, the rat was placed in a stereotaxic head holder. The dorsal skull and back of the neck was shaved and blotted with povidone-iodine. The atlantooccipital membrane was exposed, and a polyethylene (PE-10) catheter was advanced through an incision in the membrane to a position 9 cm caudal to the cisterna at the level of the lumbar enlargement. The catheter was externalized on the top of the skull and secured with a piece of steel wire. The wound was closed with 3-0 silk sutures. Rats showing neurologic deficits postoperatively were killed immediately by CO₂ inhalation. After the implantation of intrathecal catheters, rats were housed in individual cages. Intrathecal injection studies were performed 5-7 days after surgery.

Gabapentin (1-[aminomethyl] cyclohexanacetic acid; neurontin; molecular weight [MW] 171), S(+)-3-isobutyl GABA (MW 157; Parke-Davis, Ann Arbor, MI), and D-serine and L-serine (MW 105; Sigma Chemical, St. Louis, MO) were used. Gabapentin was stored at 5[degree sign]C in an opaque container. All drugs for intrathecal injection were freshly prepared in physiologic saline so that the required dose was delivered in an appropriate injection volume (10 micro L). All doses are expressed in micrograms as the total intrathecal dose given per rat.

Previous work has shown that a mild thermal injury of the plantar surface of the paw induces clear thermal hyperalgesia (unpublished observations). To prepare this model, the rat was briefly anesthetized in an induction box with halothane anesthesia (2%). On loss of spontaneous movement and response to a toe pinch, the plantar surface of the right hind paw of the rat was placed flat on a 52.5 +/- 1[degree sign]C surface hot plate for 45 s. A 10-mg weight was placed on the dorsum of the paw to maintain a constant, equal pressure to the heel area of the paw during the thermal exposure. After the paw was removed from the surface, a significant thermal hyperalgesia could be observed by 30 min, and this was sustained for approximately 3 h. In initial studies, we demonstrated that this treatment did not produce blistering of the paw during the subsequent 24-h interval.

To measure the thermal escape latency, a thermal stimulator that permitted us to focus a radiant stimulus on the plantar surface of the hind paw though a glass surface on which the animal stood ^[12,13] was used. The rats were placed in a clear plastic cage (10 x 20 cm) that rested on an elevated floor of clear glass. A focused radiant heat source was contained in a movable holder placed beneath the glass floor. The current to the thermal source was controlled by a constant supply. To reduce the variability in plate surface caused by room temperature, the interior of the box under the animal was regulated at 30[degree sign]C with a feedback-controlled heater fan under the glass. The calibration of the thermal test system was such that the average response latency in normal untreated rats, measured before the initiation of an experimental series, was 10 +/- 1 s (mean +/- SE).

To initiate a test, the rat was placed in the box and allowed approximately 30 min to adjust to it. The under-floor heat source was then positioned so that it was focused at the heel portion of planter surface of one hind paw where it was in contact with the glass. The light was then activated, which in turn activated a timing circuit. The time interval between the application of the light and the brisk hind paw withdrawal response was measured to the nearest 0.1 s. Paw withdrawal was sensed automatically by photodetectors. In the absence of a response by 20 s, the stimulus was automatically terminated, and this time was assigned as the escape latency (cutoff time).

Drugs and doses were assigned randomly to be given over the course of these studies. Control animals were interspersed concurrently with drug-treated animals. This prevented all of the controls being run on a single group of animals at one time in the course of the investigation. Two sets of experiments were performed.

The first set of experiments involved characterization of the dose-response and time-response curves of intrathecally administered gabapentin, S(+)-3-isobutyl GABA, and R(-)-3-isobutyl GABA. The thermal injury was induced at -30 min. The drugs were then delivered at time (t) 0. Thermal escape thresholds were defined at intervals before and after the drug injection.

The second set of experiments involved examination of the effects of D- and L-serine on the antihyperalgesic effects of gabapentin and S(+)-3-isobutyl GABA. Thermal injury was induced at -30 min. D- or L-serine was injected intrathecally (100 micro g) at t-10. Gabapentin and S(+)-3-isobutyl GABA were intrathecally administered at t = 0. Thermal escape thresholds were measured at intervals before and after the drug injection.

Motor function was evaluated by the observation of two specific behaviors: 1) the placing/stepping reflex. This response was evoked by drawing the dorsum of either hind paw across the edge of the table. This act results in an upward lifting of the paw from the surface of the table (stepping) and a subsequent placement of the plantar surface of the paw on the Table topwith the toes slightly spread (placing). 2) Righting reflex. A rat placed horizontally in a supine position will display an immediate coordinated twisting of the body around its longitudinal axis to regain its normal upright posture.

The mean +/- SE of the paw withdrawal latency (PWL) were plotted. Dose-response curves were plotted against the maximal PWL. The results were analyzed by using one-way analysis of variance, followed by a multiple comparison test or by a Student's t-test in a paired series. In these cases, multiple comparisons were undertaken using Bonferroni's correction to prevent alpha build-up. To compare the potency of gabapentin and S(+)-3-isobutyl GABA, the dose required to raise the escape latency to 10 s (defined as the ED_10s) and 95% confidence interval were calculated using the computer program of Tallarida and Murray ^[14]. The analyses were carried out using (Abbacus StatView/Concepts, Inc., Berkeley, CA). Values of P < 0.05 were considered statistically significant.

Results

Control Response to Thermal Injury

After placing the right hind paw on the thermal surface (52.5[degree sign]C) for 45 s, the response latency in saline-treated rats decreased from the 10.5 +/- 0.1 s measured before injury to 6.35 +/- 0.33 s (n = 6) 30 min after injury. In comparison, the withdrawal latency of the uninjured paw was not altered (e.g., escape latency 10.9 +/- 0.4 s before injury and 12.9 +/- 0.6 s 30 min after injury). Examination of the time course of the reduced thermal escape latency (hyperalgesia) indicated that the escape latency of the injured paw had returned to preinjury values by 120-150 min after the injury (Figure 1).

Intrathecal Gabapentin/3-Isobutyl GABA and Escape Latencies

The intrathecal delivery of gabapentin (30-300 micro g) and S(+)-3-isobutyl GABA (30-300 micro g) produced a significant, dose-dependent reversal of the thermal hyperalgesia, but had no effect on the response latency of the normal hind paw, even at the largest doses (Figure 1 and Figure 2).

The stereoisomer R(-)-3-isobutyl GABA had no effect on hyperalgesia or the normal escape latency at its largest dose (300 micro g) (Figure 3).

The effects on the thermal injury-induced hyperalgesia of intrathecal gabapentin, S(+)-3-isobutyl GABA, and R(-)-3-isobutyl GABA were dose-dependent (Figure 4). The ED_10s (and confidence interval) values for gabapentin and S(+)-3-isobutyl GABA were 192 (51-720) micro g and 119 (67-210) micro g, respectively.

Intrathecal D- and L-Serine and Escape Latencies

Intrathecal D-serine(100 micro g) and L-serine (100 micro g) alone have no effect on thermal injury-induced hyperalgesia (Figure 5). On the other hand, the antihyperalgesic effects of both intrathecal gabapentin (300 micro g) and S(+)-3-isobutyl GABA (300 micro g) were significantly reversed by pretreatment with intrathecal D-serine (100 micro g), but not L-serine (100 micro g) (Figure 6 and Figure 7). All effects were observed at doses that had no significant effect on motor function.

Intrathecal Gabapentin/3-Isobutyl GABA and Motor Function

Assessment of placing/stepping reflexes or righting reflexes revealed no difference between normal and thermally injured paws. The intrathecal delivery of the largest dose of gabapentin or either isomer of 3-isobutyl GABA had no effect on the expression of these end points. At the largest doses, there was a reduction in the briskness of the responses, but they were never blocked. In addition, there seemed to be a reduction in spontaneous activity, but this was not systematically quantified and was judged not to alter the response of the rat to the environment, e.g., as when being handled or briefly restrained. Similarly, intrathecal D- or L-serine were without detectable effects on motor function, either alone or in combination with either gabapentin or S(+)-3-isobutyl GABA.

Discussion

Brief exposure of the paw to a 52.5[degree sign]C surface results in mild erythema and well defined hyperalgesia, as indicated by a reliable decrease in the paw escape latency. The contralateral paw shows no change in latency. In this model, intrathecal gabapentin and S(+)-3-isobutyl GABA had no effect on the thermal escape latency of the normal, uninjured paw, but produced a well defined reversal of the hyperalgesia otherwise observed in the injured paw.

Several points should be stressed. First, although some reduced motor function could be induced by very large doses, the magnitude of these effects were not judged to interfere with the animal's ability to respond. Moreover, although the hyperalgesic paw latency returned to normal after gabapentin, the normal paw's corresponding response latency was unaltered. This argues against the likelihood that the increased latency of the injured paw simply reflected a general motor weakness leading to increased response latencies. Second, although these studies do not exclude a supraspinal effect, the potent effect of the spinally delivered agent (10-100 micro g) compared with the systemic activity (10-300 mg/kg), as previously demonstrated (^[4,5]; Partridge, Chaplan, and Yaksh; unpublished observations), suggests that the spinal cord is a likely site of this antihyperalgesic action. This finding is consistent with the fact that gabapentin has a low lipid solubility (log [octanol/water] = -1.2) ^[15].

There are now data that suggest that gabapentin may represent a class of drugs, which exert their actions through a novel mechanism. First, gabapentin binds brain tissue with high nanomolar affinity ^[7,15]. Second, although this binding seems not to correspond with any known transmitter site, it is displaced by a structural analog S(+)-3-isobutyl GABA, but not by the stereoisomer R(-)-3-isobutyl GABA ^[8]. Third, the stereoselectivity in binding is reflected in the anticonvulsant activity and now in the spinal antihyperalgesic effects. Finally, the effects of both active drugs are reversed by intrathecal D-, but not L-serine (see below). These convergent observations jointly suggest that gabapentin and 3-isobutyl GABA may share a common, if as yet undefined, mechanism of action.

Gabapentin was synthesized to be a systemically active GABA analog and was found to have anticonvulsant actions ^[16-18]. Several mechanisms may account for the antihyperalgesic and anticonvulsant activity.

Gabapentin has, however, been shown to increase GABA synthesis ^[19] and release from the brain ^[20]. In addition, gabapentin may enhance extracellular levels of GABA released from astrocytes induced by the reversal of the GABA transporter ^[21]. On the other hand, the drug has no affinity for either GABA A or B binding sites ^[7]. Similarly, in studies on the Chung model of neuropathy, spinal delivery of either GABA A or B receptor antagonists had no effect on the antiallodynic effects of intrathecal gabapentin ^[4].

Functionally, the actions of gabapentin are limited to those models that involve a facilitated state of processing. Models of facilitated processing initiated by tissue injury (e.g., the formalin test, carrageenin-induced inflammation) or nerve injury are reversed by spinal NMDA receptor antagonists ^[22] and are associated with an increase in the spinal release of glutamate ^[23-25]. Although gabapentin may reduce NMDA-evoked currents ^[26], such effects are not seen at concentrations likely to be achieved in vivo ^[27]. Nevertheless, gabapentin could interfere with glutamate transmission by altering glutamate synthesis or release. Gabapentin has moderate inhibitory effects in vitro on branched chain amino acid transferase, an enzyme that metabolizes cytosolic amino acids to form glutamate ^[28]. With regard to receptor interaction, binding studies have shown no affinity of gabapentin for NMDA, alpha-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid, or nonstrychnine glycine sites ^[7]. Despite the lack of binding at the nonstrychnine site, it has been reported that the anticonvulsant effects of gabapentin are antagonized by D-, but not L-serine ^[9]. D-Serine is an agonist at the nonstrychnine-sensitive glycine site of the NMDA receptor ^[29]. Agonist occupancy of this site facilitates NMDA-mediated membrane currents ^[10]. Intrathecal D-serine, but not L-serine, reverses the effects of drugs that antagonize the nonstrychnine-glycine site ^[30,31]. Although intrathecal D-serine alone reportedly facilitates the thermally evoked tail flick ^[32], we have not observed this action under any condition in the present work. The lack of effect of D-serine alone suggests that, in the present model, the glycine site is fully occupied.

Gabapentin and its congeners bind with high affinity at a site that is associated with the alpha₂ delta-subunit. This subunit is found in many calcium channels. This interaction may regulate terminal excitability and alter transmitter release ^[33].

In summary, this study indicates a dose-dependent and stereospecific antihyperalgesic effect at the spinal level for gabapentin and its structural analog 3-isobutyl GABA. The parallels between these drugs, e.g., overlapping binding sites, similar profiles of functional activity, and ability to antagonize the effects of D-serine, but not L-serine, suggest that these drugs may constitute a class of drugs, gabapentinoids, that by a novel mechanism of action serves to alter spinally facilitated processing induced by tissue injury.