Abstract
Central sensitization is a condition of enhanced excitability of spinal cord neurons that contributes to the exaggerated pain sensation associated with chronic tissue or nerve injury.N-methyl-d-aspartate (NMDA) receptors are thought to play a key role in central sensitization. We have tested this hypothesis by characterizing in vitro and in vivo a novel antagonist of the NMDA receptor acting on its glycine site, GV196771A. GV196771A exhibited an elevated affinity for the NMDA glycine binding site in rat cerebral cortex membranes (pKi = 7.56). Moreover, GV196771A competitively and potently antagonized the activation of NMDA receptors produced by glycine in the presence of NMDA in primary cultures of cortical, spinal, and hippocampal neurons (pKB = 7.46, 8.04, and 7.86, respectively). In isolated baby rat spinal cords, 10 μM GV196771A depressed wind-up, an electrical correlate of central sensitization. The antihyperalgesic properties of GV196771A were studied in a model of chronic constriction injury (CCI) of the rat sciatic nerve and in the mice formalin test. In the CCI model GV196771A (3 mg/kg twice a day p.o.), administered before and then for 10 days after nerve ligature, blocked the development of thermal hyperalgesia. Moreover, GV196771A (1–10 mg/kg p.o.) reversed the hyperalgesia when tested after the establishment of the CCI-induced hyperalgesia. In the formalin test GV196771A (0.1–10 mg/kg p.o.) dose-dependently reduced the duration of the licking time of the late phase. These antihyperalgesic properties were not accompanied by development of tolerance. These observations strengthen the view that NMDA receptors play a key role in the events underlying plastic phenomena, including hyperalgesia. Moreover, antagonists of the NMDA glycine site receptor could represent a new analgesic class, effective in conditions not sensitive to classical opioids.
Glutamate is the dominant excitatory neurotransmitter of the mammalian central nervous system. Almost all types of central neurons can be excited by glutamate acting on a variety of ligand-gated ion channels or G protein-coupled (metabotropic) receptors (Hollmann and Heinemann, 1994). Glutamatergic transmission is critically involved in many important events of the central nervous system, such as synaptic plasticity in the developing and adult brain, as well as neuronal survival and death (Choi, 1994; Collingridge and Bliss, 1995). The glutamate-gated ion channel receptors have been classified into three main categories, those sensitive to the agonistN-methyl-d-aspartate (NMDA; NMDA receptors), those sensitive to the agonist kainate (kainate receptors), and those sensitive to the agonist α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA; AMPA receptors). Molecular cloning studies have revealed that NMDA receptors are heteromultimeric protein complexes, composed of at least two subunits, NMDAR1 (NR1) and NMDAR2 (NR2). Whereas NR1 exists in several variants generated by alternative splicing from a single gene, four distinct genes are responsible for the expression of NR2A, 2B, 2C, and 2D (Hollmann and Heinemann, 1994). The NMDA receptor is blocked by external physiological Mg2+ at resting membrane potential. Depolarization removes this block, allowing entry of Ca2+, which can in turn activate a variety of signal transduction pathways (Mayer et al., 1984). As a consequence of this mechanism, NMDA receptors can function as detectors of temporally coincident synaptic inputs to the same neuron. Such a cascade of events is thought to be at the basis of the long-term potentiation of synaptic inputs to CA1 pyramidal neurons in the hippocampus, an event responsible for certain forms of learning (Herron et al., 1986). However, the crucial role of NMDA-based detectors of coincident activity is not restricted to hippocampal synaptic plasticity. Indeed, evidence is present to support the crucial role of the NMDA receptors in other forms of neuronal plasticity induced by prolonged electrical activity. Peripheral tissue injury or inflammation induces a state of sensory hypersensitivity that manifests itself as allodynia (decreased pain threshold) and hyperalgesia (an increased response to noxious stimuli). This pain hypersensitivity results from both an increase in transduction sensitivity of primary afferent receptors and an increase in the excitability of spinal cord neurons (Ma and Woolf, 1995). Several lines of investigation have shown the involvement of the glutamatergic system in the development and maintenance of pain hypersensitivity. Iontophoretic applications of glutamate produce a facilitation of responses to low and high intensity of mechanical stimulation of the skin (Dougherty and Willis, 1991), whereas an increase in glutamate release has been observed in rat spinal cord after injection of an irritant agent or nerve injury (Sluka and Westlund, 1992; Malmberg and Yaksh, 1995). Moreover, it appears that the dorsal horn neuronal plasticity and hyperexcitability after tissue injury involve the effect of excitatory amino acid on NMDA receptors. In fact, NMDA antagonists have been shown to suppress formalin-induced pain behavior (Haley et al., 1990), hyperalgesia induced by chronic constriction injury (CCI) to the rat sciatic nerve (Davar et al., 1991), and peripheral inflammation (Ren et al., 1992).
The NMDA receptor is unique because opening of the channel requires the simultaneous binding of glutamate and glycine (Corsi et al., 1996). Therefore, the NMDA receptor blockade can also be obtained through the antagonism of the glycine site, which may represent a centerpiece for a novel strategy to explore in vivo the role of NMDA receptors (Dickenson and Aydar, 1991; Ren et al., 1992). Therefore, to further study the role of NMDA receptors in chronic pain, we have used a novel, potent, and selective antagonist of the glycine site, GV196771A, [E-4,6-dichloro-3-(2-oxo-1-phenyl-pyrrolidin-3-ylidenemethyl)-1H-indole-2-carboxylic acid sodium salt, Fig. 1; Giacobbe et al., 1998].
Materials and Methods
Binding Studies
Animals and Tissue Preparation.
Male Sprague-Dawley rats (200–250 g) were used. Animals were supplied by Charles River (Lecco, Italy) and were allowed food and water until used. Immediately after sacrifice, brains were removed and used for the preparation of cerebral cortex synaptic membranes to be used in radioligand binding studies.
The protocols used for the preparation of membranes for [3H]glycine binding experiments were as described by Mugnaini et al., (1998), whereas for [3H]AMPA or [3H]kainic acid experiments they were as described by Giberti et al. (1991).
For [3H]1-[1-(2-thienyl)cyclohexyl] piperidine ([3H]TCP) binding experiments, extensively washed membranes were prepared as described by Foster and Wong (1987), with some additional modifications to further reduce the presence of endogenous amino acids, as follows. Briefly, the cerebral cortex tissue was homogenized with a Polytron PT-MR 3000 (Littau, Switzerland) in 9 volumes (v/original wet weight) of 0.32 M sucrose and centrifuged at 1000g for 10 min. The supernatant was centrifuged again at 10,000g for 10 min and the upper “buffy coat” of the resulting pellet was collected and frozen in liquid N2 for approximately 30 s. After thawing at room temperature and resuspension in 20 volumes of MilliQ water (MilliQ ultra-pure water system, Millipore, France) the suspension was sonicated for 30 s (model 300 sonic dismembrator; Fisher, Milano, Italy) and centrifuged at 50,000g for 20 min. The last three steps were repeated three more times, with the addition of an incubation at 25°C for 20 min before the last centrifugation. The final pellet was resuspended in 10 volumes of MilliQ water and dialyzed (D0655 dialysis tubing; Sigma, St. Louis, MO) against 5 liters of Tris/HCl 5 mM (pH 7.7) for 72 h, with the dialysis buffer refreshed every day. Finally, membranes were centrifuged at 50,000g for 20 min and the pellet was resuspended in 3 volumes of MilliQ water, aliquoted, frozen in liquid N2, and conserved at −80°C until the day of the experiment. Only laboratory glassware previously treated at high temperature (250°C) for more than 4 h or treated with 1 M HCl and then extensively washed with fresh MilliQ water was used throughout the preparation procedure.
[3H]Glycine Binding Assay.
[3H]Glycine displacement experiments were performed as described by Mugnaini et al., (1998). GV196771A was dissolved in dimethyl sulfoxide (DMSO) at 5 mM and tested at seven concentrations in duplicate (0.1 nM to 100 μM) in five separate experiments.
Antagonism of Glycine-Induced [3H]TCP Binding Enhancement.
[3H]TCP binding experiments were performed as follows. To minimize glycine contamination all steps were performed in laboratory glassware previously treated at high temperature (250°C) for more than 4 h or treated with 1 M HCl and then extensively washed with fresh MilliQ water. Briefly, on the day of the experiment, rat cerebral cortex membranes were homogenized in 20 volumes of 5 mM Tris/HCl (pH 7.7) and centrifuged at 48,000g for 15 min. Pellets were then given 4 cycles of washing, each cycle consisting of resuspending the membranes in 20 volumes of buffer, incubating at 25°C for 20 min, and centrifuging at 48,000g for 15 min. Glycine concentration response curves (CRC) at enhancing [3H]TCP binding have been obtained in the presence of increasing concentration of GV196771A and 1 μM glutamic acid. Final pellets were resuspended in 30 volumes of buffer and [3H]TCP binding was performed in a final volume of 1000 μl containing: 180 μl of buffer, 100 μl of buffer with 1 μM glutamic acid (final concentration), 20 μl of buffer containing increasing concentrations of glycine (from 0.1 nM to 100 μM final concentration, including intermediate concentrations), 100 μl of buffer with 10X the final desired concentration of GV196771A, 100 μl of buffer with the radioligand ([3H]TCP at a final concentration of 2.5 nM), and 500 μl of final membrane suspension. The reaction was allowed to proceed for 2 h at 30°C and then stopped by filtration through glass fiber filters (GF/C, Whatman International, Maidstone, UK) on a Brandel M48R cell harvester (Brandel, Gaithersburg, MD), followed by rapid washing of the filters two times with 3 ml of ice-cold buffer. Filters were collected in polyethylene vials (Biovials; Beckman Instruments, Berkeley, CA) to which 3.5 ml of scintillation fluid (Filter Count, Packard Instruments, Downers Grove, IL) were added. The vials were capped, shaken, and counted in a Packard TRI-CARB 1900 CA liquid scintillation analyzer. CRC to glycine in the absence (control) or presence of antagonist (0.3, 1, and 3 μM GV196771A) were simultaneously obtained on the same experiment with each incubation performed in triplicate.
[3H]AMPA, [3H]-cis-4-(Phosphonomethyl)-2-piperidinecarboxylic acid (CGS-19755), and [3H]Kainic Acid Binding Assay.
[3H]AMPA and [3H]-CGS-19755 displacement experiments were performed essentially as described previously by Giberti et al. (1991)and Murphy et al. (1988). [3H]Kainic acid displacement experiments were performed as follows. Membranes for the [3H]kainic acid binding assay were incubated (60 min, 4°C) in Tris/acetate 50 mM (pH 7.10) with 2 nM radioligand. The reaction was stopped by dilution with ice-cold buffer solution and filtration as indicated for [3H]TCP binding assay.
GV196771A was dissolved in DMSO at 5 mM and tested at five concentrations in duplicate (10 nM to 100 μM); glutamic acid, at six concentrations in duplicate (1 nM to 100 μM), was the reference compound tested as a positive control.
Data Analysis for Binding Experiments.
Data of [3H]glycine and [3H]kainic acid displacement experiments were analyzed using the nonlinear curve fitting program Ligand (Munson and Rodbard, 1980) to determine the inhibition constants of displacer ligands (Ki); the requiredKD values of the radioligands were set to 177 nM and 2.19 nM, as determined previously in saturation studies with [3H]glycine and [3H]kainic acid, respectively (data not shown). Given the ability of [3H]AMPA to label two populations of binding sites in the conditions used in this study (Giberti et al., 1991) and the consequent complexity of estimation ofKi values, data of [3H]AMPA displacement experiments were analyzed using Allfit (De Lean et al., 1978) and only the concentration of displacer ligands inhibiting 50% of [3H]AMPA-specific binding (IC50) were estimated.Ki and IC50values have been expressed as pKi(−LogKi) and pIC50 (−Log IC50) ± S.E.M. In [3H]TCP binding experiments, CRCs to glycine in the presence of increasing concentrations of GV196771A were simultaneously fitted to the following equation:
Drugs and Solutions for Binding Experiments.
[3H]Glycine (NET 004, specific radioactivity 1620.6 GBq/mmol), [3H]AMPA (NET 833, specific radioactivity 1961.0 GBq/mmol), [3H]kainic acid (NET 875, specific radioactivity 2146.0 GBq/mmol), [3H]TCP (NET 886, specific radioactivity 1835.2 GBq/mmol), and [3H]-CGS-19755 (NET 988, specific radioactivity 9.25 MBq/nmol) were obtained from DuPont-NEN (Boston, MA). GV196771A was synthesized by the Medicinal Chemistry department of Glaxo Wellcome S.p.A. (Verona, Italy). Glycine, glutamate, and various salts were purchased from Sigma Chemical Company (St. Louis, MO). All salts and reagents were of the highest analytical grade available.
Electrophysiological Recordings
We have used the whole cell patch-clamp recording technique to measure macroscopic currents. The preparations used included primary cultures of rat embryonic neurons. Moreover, in parallel experiments, we have used extracellular recordings from the ventral root of isolated baby rat spinal cords to measure spinal cord wind-up.
Culture of Cortical Neurons.
Neurons were prepared as described by Tang and Aizenman (1993). Briefly, cortices of 16-day-old Sprague-Dawley embryonic rats (Charles River) were incubated in a dissociation medium [minimal essential medium (MEM) containing 2 mM glutamine, 1% penicillin/streptomycin, and 0.2% glucose] containing 0.032 mg/ml trypsin for 2 h. The tissue was then incubated for 20 min in Earle’s balanced salt solution and dissociated by means of a Pasteur pipette. Cells were suspended in plating medium (Dulbecco’s medium, supplemented with 10% calf serum, 10% nutrient mixture F-12 ham medium, 2 mM glutamine, 1% penicillin/streptomycin, and 25 mM HEPES) and plated onto glass coverslips coated with 10 mg/ml poly-l-lysine. The culture was treated on day 15 with 1.5 μg/ml cytosine-B-d-arabinofuranoside. Half changes of medium were done three times a week.
Culture of Hippocampal Neurons.
Cells were prepared as described by Goslin and Banker (1991). Briefly, hippocampi of 18-day-old Sprague-Dawley embryonic rats (Charles River) were incubated in a low-calcium saline with trypsin for 20 min. The trypsin was inactivated with MEM supplemented with 10% horse serum, 2 mM glutamine, 1% penicillin/streptomycin, and 0.6% glucose (plating medium). The tissue was mechanically dissociated and plated onto a confluent layer of cortical glial cells. The second day in culture the plating medium was substituted with feeding medium (MEM containing 2 mM glutamine, 1% penicillin/streptomycin, 0.6% glucose, 1 mg/ml bovine serum albumin, 0.1 mg/ml transferrin, 100 μM putrescine, 30 nM Na-selenite, 0.5 mg/ml insulin, 20 nM progesterone, and 1 mM Na-piruvate).
Culture of Spinal Cord Neurons.
Spinal cord neurons were prepared as described by Fitzgerald (1989). Briefly, spinal cords of 14-day-old Sprague-Dawley embryonic rats (Charles River) were incubated in a low-calcium saline with trypsin for 30 min. The enzyme was inactivated in a plating medium consisting of MEM supplemented with 10% fetal bovine serum, 10% horse serum, 0.37% Na-bicarbonate, 0.6% glucose, 2 mM glutamine, and 40 μg/ml deoxyribonuclease I. The tissue was mechanically dissociated and plated onto 10 mg/ml poly-l-lysine-coated glass coverslips. The next day the plating medium was substituted with a feeding medium composed of MEM with 0.37% Na-bicarbonate, 0.6% glucose, 2 mM glutamine, 5% horse serum, 10 μg/ml bovine serum albumin, 0.2 mg/ml transferrin, 32 μg/ml putrescine, 10 ng/ml Na-selenite, 20 ng/ml triiodothyronine, 10 ng/ml insulin, 12 ng/ml progesterone, and 40 ng/ml corticosterone. All cultures were treated 5 days after plating with a mixture of 5′-fluoro-2-deoxyuridine and uridine (20 and 50 μg/ml, respectively) to suppress overgrowth of background cells. Half changes of medium were done twice weekly.
Isolation of Rat Neonatal Spinal Cords.
Spinal cords were prepared from 4- to 10-day-old Sprague-Dawley rats (Charles River) as described by Thompson et al. (1992).
Electrical Recordings.
Whole cell patch-clamp recordings were performed on cultured neurons after 1 week. The extracellular solution contained (in mM): NaCl 140, KCl 5, CaCl2 1, HEPES 10, glucose 10, pH adjusted to 7.4. Tetradotoxin (0.1 μM) was used to block spontaneous activity. The intracellular (pipette) solution contained (in mM): CsCl 140, EGTA 11, MgCl2 4, Mg-ATP 2, HEPES 10, pH adjusted to 7.3. The recording chamber was placed on the stage of an inverted microscope and continuously superfused by gravity with the extracellular solution containing glycine and GV196771A at the specified concentrations. Test NMDA solution was applied rapidly with the “U-tube” method, positioning the ejection hole within 100 to 200 μm of the cell. The cells were voltage-clamped at −60 mV. Only cells having a resting potential more negative than −40 mV were used. Currents were digitized (100 Hz), low-pass filtered (40 Hz), and stored on-line using an IBM-compatible PC running pClamp (Axon Instruments, Foster City, CA).
Spinal wind-up was studied as follows. Closely fitting glass suction electrodes filled with Krebs’ solution were used for both stimulation and recording from the dorsal and ipsilateral ventral roots, respectively (levels L4 or L5). For stimulation, an analogic stimulator (Grass Instruments, Quincy, MA) controlled by a personal computer (PC; see below), and coupled to a standard stimulus isolation unit (SIU; Grass Instruments) was used. For recording of the ventral root potential, an Axoprobe-1A (Axon Instruments) was used. The resulting signal was filtered at 0.2 kHz, digitized at 1 kHz using a TL-1 interface (Axon Instruments), and finally stored on PC hard disk. For stimulation, simultaneous data acquisition, as well as off-line data analysis, a personal computer using custom-made software written in Axobasic (Axon Instruments) was used. Recruitment of the nociceptive afferent fibers (C-/group IV) in the dorsal root was achieved by applying rectangular electrical pulses with an amplitude of 50 V and a duration of 1 ms (Thompson et al., 1992, 1993). Short trains (20 s, 1 Hz) of this type of stimulus produced cumulative synaptic responses on the ipsilateral ventral root, a phenomenon called “wind-up” (Sivilotti et al., 1993). During the course of the experiment, the wind-up was evoked every 3 min. At least 1 h of recording in basal conditions (20 trains; aerated Krebs’ solution only) was allowed in each preparation to test whether the wind-up remained stable. Subsequently, DMSO (0.1%, the largest concentration used in wind-up experiments; see below) was added to the Krebs’ solution used for the superfusion and the wind-up recorded for another 30-min period. Only preparations with a stable level of wind-up during superfusion with both Krebs and DMSO in Krebs solutions were used. GV196771A,d(−)-2-amino-5-phosphonopentanoic acid (d-AP5), or morphine was dissolved in artificial cerebral spinal fluid from stock solution and superfused into the recording chamber at the same flow rate as control artificial cerebral spinal fluid.
Data Analysis for Electrophysiology.
Current amplitude was measured and analyzed by means of custom made analysis software written in Axobasic (Axon Instruments). All current measurements, unless otherwise stated, refer to the steady-state value of the responses (Iss) and are the average of the last 3 to 4 s of agonist application. For the generation of the agonist CRC, values were expressed as a percentage and normalized to the maximal effect. Glycine CRCs were obtained in the presence of 100 μM NMDA, whereas NMDA CRCs were obtained in the presence of 3 or 10 μM glycine (as indicated). Data were fitted to the equation to estimate the response at the steady state:
The effect of GV196771A was measured using the maximal response to NMDA and glycine. Then, data were fitted to the equation:
In the spinal cord wind-up experiments, cumulative response (wind-up) was expressed as area under the curve (AUC) calculated during the train stimulation (0–20 s). For each experiment the average of three AUC values were taken before the GV196771A superfusion (AUC-control) and at the plateau drug effect (AUC-test response).
Statistical analyses were performed to compare AUC-test response with AUC-control using a paired Student’s t test.
Drug and Solution for Electrophysiological Studies.
NMDA, morphine hydrochloride, and glycine were purchased from Sigma, whereasl-AP5 was purchased from Tocris (Langford, UK). GV196771A was stored at 10 mM either in a solution of 50% DMSO-50% distilled water (patch-clamp experiments) or 100% DMSO (wind-up experiments). The maximal concentration of DMSO present in the final solutions was 0.5% for patch-clamp experiments and 0.1% for wind-up experiments. We found that at these concentrations DMSO did not affect the measured responses.
Behavioral Studies
Effect of GV196771A on Thermal Hyperalgesia in A Rat Model of Painful Mononeuropathy: CCI Model.
Male Sprague-Dawley rats (Charles River) weighing 200 to 300 g were used. Animals were housed in groups of 2 to 3 and fed with chow pellet diet with free access to water. They were fasted overnight before the study and were allowed free access to water. Rats were anesthetized with pentobarbital sodium (50 mg/kg i.p.). The left common sciatic nerve was exposed, and proximal to the sciatic trifurcation about 10 mm of nerve was freed of adhering tissue and four ligatures (3.0 chromic gut) were tied loosely around it with about 1 mm of spacing (Bennett and Xie, 1988). Rats were tested for thermal hyperalgesia using a commercial available analgesimeter (Plantar test, Ugo Basile, Comerio, Italy) by applying a heat stimulus (50W, 8V) directed onto the plantar surface of each hind paw, and the paw withdrawal latency (s) was determined. Four latency measurements were taken for each hind paw and averaged. The results were expressed as the difference score (DS) by subtracting the latency of the control side from the latency of the ligated side. Negative DSs indicated a lower threshold on the ligated side supporting an hyperalgesic state. The animals developed thermal hyperalgesia within 14 to 21 days after surgery.
Two different protocols were followed: 1) prophylactic treatment: GV196771A (6 mg/kg) or vehicle were administered orally to animals on the day of surgery in a dose volume of 10 ml/kg. Then, postoperatively, 3 mg/kg GV196771A or vehicle was administered twice a day for 10 days. Hyperalgesia was measured on days 0, 3, 9, 14, 21, and 30; and 2) therapeutic treatment: GV196771A (1–10 mg/kg) or vehicle was administered orally to animals on day 14 or 21 after ligation and the hyperalgesia was tested 1, 4, and 8 h after treatment. Only the animals that developed thermal hyperalgesia higher than −1.5 s as DS values were used for this study. The evaluation of the drug effects was carried out by a blind operator.
Effect of GV196771A and Morphine on Pain Behavior in Mice Paw Formalin Test: Acute Administration.
Male albino CD mice (Charles River) weighing 25 to 30 g were used. Animals were housed in groups of 5 to 6 and fed with chow pellet diet with free access to water. They were fasted overnight before the study and were allowed free access to water.
Before the formalin injection, mice were placed individually into clear perspex cages that served as observation chambers. After 15 min of adaptation to the cage, 20 μl of 1% formalin was injected into the plantar surface of the left hind paw. The amount of time, in seconds, the animals spent licking the injected paw for the first 5 min (early phase, EP) and then from 20 to 60 min (late phase, LP) after formalin was used as measurement of the intensity of pain. GV196771A (0.1–10 mg/kg po), the standard opioid morphine (0.3–10 mg/kg i.p.), or vehicle (0.5% methocel for GV196771A or saline for morphine) were administered in a dose volume of 10 ml/kg 1 h or 30 min before formalin injection for GV196771A and morphine, respectively. Five to ten animals were used for each group.
Effect of Chronic Treatment with GV196771A or Morphine in Mice Paw Formalin Test.
Mice were divided randomly into seven groups (10–21 mice per group) and administered once daily for 8 days as follows: groups g1 and g3 received saline i.p.; groups g2 and g5 received methocel p.o.; group g4 received morphine 10 mg/kg i.p.; and groups g6 and g7 received GV196771A 3 and 10 mg/kg p.o., respectively (Table 1). On day 9 mice were treated with saline i.p. (g1), methocel po (g2), morphine 3 mg/kg i.p. (g3 and g4), and GV196771A 3 mg/kg (g5, g6, and g7); the protocol treatment is described in Table 1. GV196771A 3 mg/kg p.o. was administered 1 h before formalin injection whereas morphine 3 mg/kg i.p. was administered 30 min before formalin injection. The evaluation of the drugs effects was carried out by a blind operator.
Statistic Analysis for Behavioral Studies.
For all experiments the data are expressed as mean ± S.E.M.
CCI test: prophylactic treatment.
Statistical analysis was performed within each group to compare DS calculated at each time point after ligation versus basal values before surgery. In addition, the time course of DS of the vehicle-treated group was compared with the GV196771A-treated group. One-way ANOVA followed by Dunnett’s test, where p < .05 was considered significant, was used.
CCI test: therapeutic treatment.
Statistical analysis was performed to compare DS of control response (vehicle) and GV196771A-treated groups, taken at various time after administration, versus pretreatment values, using one-way ANOVA followed by Dunnett’s test where p < .05 was considered significant. Dose-response curve regression analysis was then performed to evaluate regression line parameters to calculate the ED50(dose of GV196771A in mg/kg that reduced the thermal hyperalgesia by 50%) and 95% confidence intervals.
Formalin test: acute treatment.
Statistical analysis was performed to compare control response (vehicle) with test response using one-way ANOVA followed by Dunnett’s test where p< .05 was considered significant. Dose-response curve regression analysis was then performed to evaluate regression line parameters to calculate the ED50 (dose of GV196771A or morphine, in mg/kg, that reduced the licking time by 50%) and 95% confidence intervals.
Formalin test: chronic treatment.
Statistical analysis was performed to evaluate the following comparisons: 1) g1 with g3 and g4; 2) g2 with g5, g6, and g7; and 3) g5, g6, and g7 against each other.
For all comparisons, one-way ANOVA followed by Dunnett’s step-down test was used; p < .05 was considered significant (Dunnett and Tamhane, 1991).
Drugs and Solutions for Behavioral Studies.
GV196771A was prepared as a stock solution of 1 mg/ml in 0.5% methylcellulose (methocel); further dilutions were prepared in 0.5% methocel. Morphine (Sigma) was prepared as a stock solution of 10 mg/ml in saline; further dilutions were prepared in saline.
Results
Binding Experiments
[3H]Glycine.
To characterize GV196771A as a glycine site antagonist, we studied its effects on the binding of glycine in rat cerebral cortex membranes. It was observed that GV196771A inhibited [3H]glycine binding in a concentration-dependent fashion. Moreover, GV196771A was able to completely suppress the specific binding of glycine (Fig.2A). The resulting pKi was 7.56 ± 0.09 (n = 5).
[3H]TCP.
Because TCP binds only to open NMDA receptors, labeled TCP can be used to measure the fraction of open channels independently from electrophysiological methods. In agreement with its expected behavior, very low specific [3H]TCP binding to rat cerebral cortex membranes was observed in the absence of glycine and glutamate. After addition of glutamic acid (1 μM) increasing concentrations of glycine (0.1 nM to 100 μM) progressively enhanced [3H]TCP binding until it reached a maximum. The glycine CRC had an estimated pEC50 of 7.32 (7.27–7.39; 95% C.L.). The action of GV196771A was examined on TCP binding. In the presence of increasing concentrations of GV196771A (0.3–3 μM), parallel rightward shifts of the glycine CRC could be observed (Fig. 2B) with no significant depression of the maximal response. The Shild slope factor was not significantly different from unity (m = 1.09; 0.99–1.19; 95% C.L.) and a pKB value of 7.13 (7.06–7.21; 95% C.L.) was estimated.
[3H]AMPA, [3H]-CGS-19755, and [3H]Kainic Acid.
GV196771A did not displace [3H]AMPA and [3H]-CGS-19755 binding up to a concentration of 10 μM. Therefore, a pIC50 <4 could be inferred for GV196771A for both AMPA and NMDA binding sites. In the same experiments, glutamic acid dose-dependently inhibited [3H]AMPA binding with an estimated pIC50 of 6.64, and [3H]-CGS-19755 binding with an estimated pIC50 of 7.90.
Similarly, GV196771A did not displace [3H]kainic acid binding up to 1 μM. At 100 μM, 60% inhibition of specific [3H]kainic acid binding was observed. A pKi of 4.47 was calculated for GV196771A, whereas in the same experiment a pKi of 7.24 was obtained for glutamic acid.
Electrophysiological Recordings
GV196771A Antagonizes NMDA/Glycine-Induced Currents in Embryonic Rat Neurons.
Because the NMDA receptor regulates the gating of an intrinsic ion channel, one can quantitatively study the effect of GV196771A by measuring the currents flowing through this channel. In the primary cultures of cortical, hippocampal, and spinal neurons, NMDA (1–300 μM) and glycine (10 nM-10 μM) induced currents in a concentration-dependent manner only in the presence of both agonists. Currents were characterized by the presence of two phases: a transient and a sustained phase (Fig.3A-C). NMDA- and glycine-activated currents were measured at steady state and data were transformed into CRC. The agonist pEC50 and slope values are reported in Table 2.
These NMDA- and glycine-activated currents were completely inhibited by GV196771A (0.01–10 μM) in all three preparations (Fig.4A). The estimated pIC50s and the corresponded pKBs are reported in Table 2. GV196771A appeared to be a competitive antagonist because at 1 μM it shifted the glycine CRC to the right (in the presence of 100 μM NMDA) without modifying the agonist maximum response (Fig. 4B). On the other hand, GV196771A 1 μM (n = 5–6) showed an insurmountable antagonism of NMDA CRC obtained in the presence of 10 μM glycine (Fig. 4C).
Suppression of Spinal Cord Wind-Up by GV196771A.
A significant reduction (p < .01) of the AUC “wind-up” response was observed after 60 min of superfusion of the isolated spinal cords with GV196771A at 10 μM (Fig. 5A;n = 6); the “wind-up” was depressed by 25.5 ± 2% with respect to controls. In parallel experiments, we confirmed that in the same preparation morphine (n = 2) ord-AP5 (n = 3) suppressed the “wind-up” (Fig. 5, B and C).
Antihyperalgesic Activity of GV196771A
GV196771A Reduces Thermal Hyperalgesia in CCI Model.
Prophylactic treatment. Before surgery, no significant difference between left and right thermally induced paw withdrawal latencies was detected in both vehicle- (mean ± S.E.M.: 10.56 ± 0.51 s versus 10.47 ± 0.52 s) and GV196771A- (mean ± S.E.M.: 10.60 ± 0.36 s versus 10.48 ± 0.32 s) treated groups.
After surgery, the vehicle-treated group showed a decrease in thermally induced paw withdrawal latency in the ligated paw. DSs were negative and significantly different (p < .05) from preoperative values from days 14 to 30 after surgery (mean ± S.E.M. in s: −1.33 ± 0.19; −1.44 ± 0.18, and −1.23 ± 0.49 on days 14, 21, and 30, respectively, versus 0.09 ± 0.18 on day 0; Fig. 6A). The thermally induced paw withdrawal latency of the nonligated paw remained unchanged during the whole experiment (Fig. 6B).
In GV196771A-treated animals, DSs were not significantly different from preoperative values during the whole experiment (mean ± S.E.M. in s: −0.64 ± 0.42; −0.98 ± 0.26; −0.64 ± 0.15; −0.58 ±0.66; and −0.98 ± 0.24 on days 3, 9, 14, 21, and 30, respectively, versus 0.12 ± 0.20 on day 0; Fig. 6A). Furthermore, DSs at days 14 and 21 in the GV196771A-treated group were significantly lower (p < .05) compared with vehicle-treated animals at the same days showing a reduced thermal hyperalgesia (Fig. 6A).
In GV196771A-treated animals, the thermally induced paw withdrawal latency of the nonligated paw remained unchanged during the whole experiment (Fig. 6B).
Therapeutic treatment.
GV196771A (1–10 mg/kg p.o.) produced a dose-related reversal of thermal hyperalgesia increasing the latency of the ligated paw. At 3 and 10 mg/kg the reduction of thermal hyperalgesia was significant (p < .05), whereas it was not at 1 mg/kg p.o. (Fig. 6C and Table3). By measuring the analgesic effect at 1 h after treatment, an ED50 of 2.95 (1.50–8.44) mg/kg was calculated.
GV196771A Reduces Pain Behavior in Mice Paw Formalin Test: Acute Administration.
In the vehicle-treated groups, s.c. injection of formalin induced marked spontaneous nociceptive behaviors. The values of total licking time measured during the EP and the LP were 125.5 ± 7.2 and 317.3 ± 42.8 s, respectively, in the methocel-treated group (Fig. 7A) and 148.8 ± 12.1 and 529.9 ± 46.0 s, respectively, in the saline-treated group (Fig. 7B).
GV196771A (0.1–10 mg/kg p.o.) had no effect on the EP at all doses tested (Fig. 7A, left). On the other hand, a dose-related inhibition of the LP response was observed (amount of licking in s): 187.7 ± 38.9 at 0.1 mg/kg; 205.8 ± 42.2 at 0.3 mg/kg; 154.4 ± 24.4 at 1 mg/kg; 111.8 ± 25 at 3 mg/kg; and 61.9 ±10 at 10 mg/kg (Fig. 7A, right). The effects of GV196771A were significant (p < .05) at 0.1, 1, 3, and 10 mg/kg. The calculated ED50 value was 0.6 (0.1–2.1) mg/kg p.o. At all doses tested no overt behavioral changes were observed.
Morphine (0.3–10 mg/kg i.p.) induced a dose-related inhibition of the nociceptive response in both phases. In the EP the amounts of licking (s) observed were 89.6 ± 12.7 at 0.3 mg/kg; 70.2 ± 6.4 at 1 mg/kg; 52.4 ± 2.6 at 3 mg/kg; and 17.9 ± 4.9 at 10 mg/kg (Fig. 7B, left). In the LP the amounts of licking (s) observed were 279.9 ± 32.7 at 0.3 mg/kg; 233.9 ± 16.8 at 1 mg/kg; 131.9 ± 5.0 at 3 mg/kg; and 119.0 ± 7.9 at 10 mg/kg (Fig.7B, right). Morphine effects were significant (p < .05) at all doses tested. The calculated ED50values were 0.73 (0.24–1.37) and 0.56 (0.20–1.63) mg/kg i.p. for EP and LP, respectively.
Chronic Treatment with GV196771A Did Not Induce Tolerance in Mice Paw Formalin Test.
In g2, s.c. injection of formalin at day 9 induced marked spontaneous nociceptive behaviors. The licking time values measured during the EP and the LP were 125.1 ± 11.9 and 555.4 ± 147.1 s, respectively (Fig.8A).
No reduction in the EP was observed in animals receiving GV196771A 3 mg/kg p.o. at day 9 after chronic treatment with methocel (g5; Fig. 8A, left). On the contrary, a significant reduction (p < .05) of the hyperalgesic responses in LP was observed (Fig. 8A, right). The values of licking time measured were 111.8 ± 9.6 and 280.0 ± 34.1 s for the EP and LP, respectively.
When GV196771A at 3 mg/kg p.o. was administered at day 9 after chronic treatment with GV196771A at 3 and 10 mg/kg p.o. for 8 days (g6 and g7, respectively) the antihyperalgesic activity of the compound was maintained in the LP (Fig. 8A, right). However, the licking time values were higher than that observed in g5. In particular, in g6 the licking time values measured after formalin were 117.7 ± 13.6 and 377.5 ± 41.2 s for EP and LP, respectively. In g7 the licking time values measured were 131.0 ± 8.7 and 354.8 ± 35.9 s for EP and LP, respectively. Moreover, the antihyperalgesic effects observed in the LP in g5, g6, and g7 were not significantly different from each other.
Chronic Treatment with Morphine-Induced Tolerance in Mice Paw Formalin Test.
In g1 s.c. injection of formalin at day 9 induced marked spontaneous nociceptive behaviors. The licking time values measured during the EP and the LP were 119.4 ± 20.5 and 494.6 ± 92.1 s, respectively (Fig. 8B).
Animals receiving 3 mg/kg i.p. morphine at day 9 after chronic treatment with saline (g3) showed significant attenuation (p < .05) on basal nociceptive responses in both phases. Indeed, the values of licking time measured were 73.5 ± 16.0 and 233.3 ± 50.9 s in the EP and LP, respectively (Fig.8B).
In g4, animals chronically treated with morphine 10 mg/kg i.p. did not show a reduction in both EP and LP when treated with morphine 3 mg/kg i.p. at day 9. The values of licking time measured during the EP and LP were 123.4 ± 9.2 and 449.5 ± 48.8 s, respectively (Fig. 8B).
Discussion
Among the different classes of NMDA receptor antagonists, ligands at the glycine site have recently been reported to elicit antinociception against prolonged noxious stimulation in the absence of a marked influence upon motor coordination (Millian and Seguin, 1994). The results obtained with GV196771A further support the role of NMDA receptor in the development and maintenance of chronic pain and the analgesic effect of the NMDA glycine site antagonists.
Inhibition of NMDA Receptors.
GV196771A was found to possess a high affinity (pKi value of 7.56) for the glycine site of the NMDA receptors in the rat cerebral cortex. This value was in agreement with that found in functional experiments carried out with [3H]TCP. As expected, [3H]TCP could bind to the NMDA channel only when the receptor is activated by the simultaneous presence of the two agonists, NMDA and glycine. In these experiments GV196771A showed a competitive behavior displacing in a parallel manner the [3H]TCP binding induced by glycine CRC with an apparent pKB of 7.13. On the other hand GV196771A was a very week ligand for the AMPA and kainate receptors and for the glutamate binding site of the NMDA receptor.
Electrophysiological Recordings: Patch-Clamp in Primary Cultures and Wind-Up in Baby Rat Spinal Cord.
The binding results were confirmed and extended by further functional studies using the patch-clamp technique in embryonic rat neurons taking from cortex, hippocampal, and spinal cord. In these functional preparations, the simultaneous presence of NMDA and glycine induced currents flowing through the NMDA receptor channel. GV196771A antagonized the NMDA glycine-induced currents in all three preparations, although in the spinal cord neurons the pKB value was higher (8.05) in respect to those found in cortical (7.47) and hippocampal neurons (7.86). Also in this study, GV196771A was found to be a competitive antagonist of the glycine-induced currents whereas the antagonism of GV196771A on NMDA-induced currents was not reversed by increasing the agonist concentration.
Repetitive low-frequency electrical stimulation of the dorsal roots of the spinal cord with a sufficient intensity to recruit C-/group IV fibers evokes a temporal summation of the synaptic potentials recorded in the ipsilateral ventral root. This results in a cumulative ventral root potential known as “wind-up” (Sivilotti et al., 1993), which is considered to be an electrophysiological phenomenon mechanistically similar to spinal central sensitization (Thompson et al., 1993). Our findings further confirm the involvement of the NMDA receptor in the generation of wind-up as already suggested by Boxall et al. (1996) because d-AP5 and GV196771A both depressed this phenomenon. Moreover, the activity of GV196771A was qualitatively similar to that observed with the standard opioid morphine.
Antihyperalgesic Activity.
We investigated the effects of GV196771A in a rat model of painful mononeuropathy (CCI) and in the formalin test in mice.
The results obtained in the CCI model, where the hyperalgesia to thermal stimulation was measured, suggest that the compound was active when administered both pre- and postinjury. Indeed, GV196771A given before the nerve ligature blocked the development of the hyperalgesia for a long period of time (at least 30 days). Moreover, it was able to reverse the hyperalgesia in animals that had already developed thermal hypersensitivity (14–21 days after ligation). The antinociceptive effect lasted for up to 4 and 8 h at 3 and 10 mg/kg, respectively, indicating a dose-related long-lasting activity.
NMDA receptors contribute to the generation of central sensitization (Woolf and Thompson, 1991) being located in the dorsal horns of the spinal cord where long-term changes in the processing of nociceptive information occur. This phenomenon is elicited by brief, low-frequency C-fiber conditioning stimuli lasting a few seconds, which can produce a central facilitation lasting several hundred times longer. This facilitation is apparent as an expansion in the size of the cutaneous receptive fields of dorsal horn neurons, a drop in their threshold, and an increase in responsiveness, paralleling hypersensitivity. In addition, molecular changes, like the synthesis of new receptors or higher expression of native receptors, can occur during central sensitization (Neumann et al., 1996; Harris et al., 1996). The finding that GV196771A is active also after the establishment of thermal hypersensitivity indicates that whatever the molecular mechanism that occurs in the spinal cord, it does not influence the activity of the compound. Moreover, the results support the hypothesis that NMDA receptors are involved in the development as well as in the maintenance of a hyperalgesic state because GV196771A both prevents the establishment of central sensitization and returns the spinal cord toward baseline levels of excitability when given once central sensitization is already established.
The antihypersensitivity effect of GV196771A was specific because the withdrawal latencies in the nonoperated paw were unaffected by the treatment. This is consistent with the peculiarity of the NMDA receptor blockade, which affects only a pathological pain state without interfering with the physiological activity (Woolf and Thompson, 1991), in keeping with the hypothesis that NMDA receptor enhances rather than transmits noxious information.
Formalin injected s.c. into the animal paw is a frequently used pain assay. Electrophysiological studies showed that formalin administration to a hind paw excites primary C-fibers in a biphasic manner with a similar time course as observed in behavioral studies (Dickenson and Sullivan, 1987). Indeed there is an intense display of pain response for the first 5 min after formalin injection, followed by a lull in responding, and then a re-emergence of pain response, namely LP, starting about 15 min after injection, which continues for about 45 min. Specifically, although the initial EP of pain responding appears to be a direct result of chemical activation of primary nociceptive afferent fibers by formalin (Heapy et al., 1987), there is evidence that the second phase is mediated largely by processes within the spinal cord. Consistent with the role of spinal NMDA receptors in tonic pain and hyperalgesia, the sensitization of spinal neurons held to underlie the second phase of the formalin pain is mediated by NMDA receptors (Coderre et al., 1990).
Oral administration of GV196771A induces a dose-related inhibition of the LP response, without acting on the EP supporting concept that NMDA receptors are involved only in a sustained nociceptive transmission. These results are in agreement with previous experiments using different NMDA antagonists, such as ketamine and MK-801 (Haley et al., 1990) showing the selective reduction of dorsal horn activity during the LP. Therefore, these findings suggest a dissociation between the ability of NMDA antagonists to prevent central changes without actually affecting the behavioral or spinal expression of the EP.
Tolerance Study.
The results reported here indicate that GV196771A, different from opioids, does not induce tolerance of its antihyperalgesic activity after 8 days of chronic treatment.
In the formalin test we found that morphine inhibited both the EP and LP with a similar potency, suggesting, differently from NMDA receptors, an active participation of opioid receptors in acute as well as in a sustained nociceptive transmission. Interestingly, GV196771A and morphine were equally active in the LP of the test with ED50s of about 0.6 mg/kg. Enormous differences were found between these two compounds when a protocol to test for tolerance of the antinociceptive activity was carried out. Eight days of chronic treatment with morphine 10 mg/kg produced significant tolerance, in both phases of the test, in mice treated at day 9 with morphine 3 mg/kg. On the contrary, chronic treatment with GV196771A at 3 and 10 mg/kg did not alter the analgesic action of GV196771A 3 mg/kg given at day 9. The lack of tolerance observed with GV196771A cannot be attributable to the low doses or to the time of exposure used. In fact, 3 and 10 mg/kg, administered in repeated treatments, are, respectively, 5 and 15 times greater than its ED50 value detected in the acute experiment. Differently, morphine induced tolerance at a dose 15 times greater than its ED50 value estimated in the acute treatment. As far as exposure time is concerned, these results are in agreement with the experiment carried out in a neuropathic model where 10 days of repeated treatment with GV196771A prevented the development of thermal hyperalgesia without any sign of tolerance.
The lack of tolerance observed after chronic treatment with GV196771A underlines the difference in the mechanism involved in the induction of analgesia between NMDA antagonists and opioids, although it has recently been suggested that the contribution of opioid tolerance may derive from an involvement of NMDA receptors through a protein kinase C activation (Smart and Lambert, 1996). This hypothesis is supported by the evidence that the NMDA channel blocker MK-801 prevents the development of morphine tolerance in animal models (Mao et al., 1994).
Recent evidence obtained in our laboratories with GV196771A in rat cortical neurons indicated that GV196771A up to 2 g/kg p.o. did not induce vacuolization differently from the effect of the channel blocker MK-801 1 mg/kg s.c. (data not shown). Moreover, in the passive avoidance test GV196771A (30 mg/kg p.o.) did not induce learning and memory deficit (data not shown).
In conclusion, the results obtained with GV196771A suggest that this compound is a potent and selective NMDA glycine site antagonist. Moreover, the in vivo data indicate that GV196771A may offer an innovative and safe way for the management of chronic pain and the lack of tolerance of its antihyperalgesic activity could offer a possible advantage for the clinical treatment of chronic pain in comparison with opioids.
Footnotes
-
Send reprint requests to: Dr. Mauro Quartaroli, Glaxo Wellcome S.p.A., Medicines Research Centre, Department of Pharmacology, Via Fleming 4, 37135 Verona, Italy. E-mail:mq7886{at}glaxowellcome.co.uk
-
1 Glycine-induced [3H]TCP binding, equal to total [3H]TCP binding minus basal binding.
- Abbreviations:
- NMDA
- N-methyl-d-aspartate
- AMPA
- α-amino-3-hydroxy-5-methylisoxazole-4-propionate
- TCP
- 1-[1-(2-thienyl)cyclohexyl] piperidine
- DMSO
- dimethyl sulfoxide
- MEM
- minimal essential medium
- d-AP5
- d(−)-2-amino-5-phosphonopentanoic acid
- AUC
- area under the curve
- CCI
- chronic constriction injury
- DS
- difference score
- EP
- early phase
- LP
- late phase
- CRC
- concentration response curve
- CGS-19755
- cis-4-(phosphonomethyl)-2-piperidinecarboxylic acid
- Received November 16, 1998.
- Accepted March 20, 1999.
- The American Society for Pharmacology and Experimental Therapeutics