Glycine Transporter Inhibitors as a Potential Therapeutic Strategy for Chronic Pain with Memory Impairment

All of the experimental protocols used here were approved by the Animal Care and Use Committee of Nagoya City University (Nagoya, Japan) and were conducted in accordance with the guidelines of the National Institutes of Health (Tokyo, Japan) and the Japanese Pharmacological Society (Tokyo, Japan).

The surgical procedure was based on that described by Seltzer et al.  21In brief, 4-week-old, male ddY-strain mice were anesthetized by intraperitoneal administration of pentobarbital sodium (60 mg/kg). One third to one half of the dorsal aspect of the right sciatic nerve was ligated just distal to its branch to the posterior biceps and semitendinosus muscles. Thermal and mechanical hypersensitivity was assessed 7 days after ligation.

Three- to 4-week-old male ddY mice were injected with streptozotocin (200 mg/kg, intraperitoneal) dissolved in saline after a 14- to 17-h fast. All animals were housed in solid floored cages with a deep layer of paper chips that was changed daily. A 12:12 h light–dark cycle was used, and the animals were allowed free access to sufficient food and water. The blood glucose level was measured 1 day before and 2 weeks after streptozotocin injection by measuring the glucose concentration in a blood sample obtained by tail prick using an Accu-Chek blood glucose monitoring system (Roche Diagnostics, Indianapolis, IN). The mice were defined as diabetic when their blood glucose concentration exceeded 350 mg/dl. Mechanical hypersensitivity was assessed 2 weeks after streptozotocin administration.

Thermal hypersensitivity was assessed by the plantar test (Ugo Basile, Comerio, Italy) following a modification of the method of Hargreaves et al.  22Mice were placed in a clear plastic chamber with a glass floor and were allowed to acclimate to their environment before testing. During this time, the mice initially demonstrated exploratory behavior, but subsequently stopped exploring and stood quietly with occasional bouts of grooming. A mobile radiant heat source, which was located under the glass floor, was focused onto the plantar surface of the right hind paw, and paw withdrawal latencies (PWLs) were recorded. The intensity of radiant heat was adjusted to give a 7- to 8-s withdrawal latency in naive mice. A cutoff latency of 15 s was imposed to avoid tissue damage. PWLs were measured in duplicate for the right hind paw of each animal, and the mean of the two values was used for analysis.

Mice were placed in individual transparent Perspex cubicles with a wire mesh bottom, and a series of calibrated von Frey filaments (Semmes-Weinstein monofilaments; Stoelting, Wood Dale, IL) was used to determine the 50% likelihood of a paw withdrawal response (50% threshold) by the up–down method of Dixon.23Eight von Frey filaments, with approximately equal logarithmic incremental bending forces, were chosen (von Frey numbers 2.36, 2.44, 2.83, 3.22, 3.61, 3.84, 4.08, and 4.17; equivalent to 0.02, 0.03, 0.07, 0.17, 0.41, 0.69, 1.20, and 1.48 g force, respectively). Testing was initiated with the 0.17-g hair, and each hair was applied perpendicularly to the plantar surface of the right hind paw (partial nerve ligation model), with sufficient force to bend the filament, for 3–4 s. Lifting of the paw indicated a positive response and prompted the use of the next weaker (i.e. , lighter) filament. Absence of a paw withdrawal response prompted the use of the next stronger (i.e. , heavier) filament. This paradigm was continued until four measurements had been obtained after an initial change in behavior, or until four consecutive positive scores (score of 0.01 g) or five negative scores (score of 1.5 g) had been obtained. The resulting scores were used to calculate the 50% threshold.24In diabetic mice, testing was made on both the right and left hind paws of each animal, and the mean of the two 50% threshold values was used for analysis.

In the study presented here, mice that exhibited a PWL of less than 5 s in the plantar test and a 50% threshold of 0.1 g in the von Frey test 7 days after partial ligation of the sciatic nerve were considered to be developing thermal and mechanical hypersensitivity. In diabetic mice, a 50% threshold of 0.2 g in the von Frey test 14 days after streptozotocin administration was considered to indicate development of mechanical hypersensitivity.

In the formalin test, 3- to 4-week-old male ddY mice were used. Mice were placed individually in a clear plastic chamber and were allowed to acclimate to their environment before testing. A mirror was situated behind the chamber to allow an unobstructed view of the animals’ paws. Ten minutes after administration of either drugs or vehicle, mice were gently restrained, and 20 μl formaldehyde, 1% (2.7% formalin in distilled water), was subcutaneously injected into the plantar surface of the left hind paw using a 27-gauge needle. After formalin injection, each mouse was placed back in the testing chamber, and the incidence of formalin-induced nociceptive behavior characterized by licking/biting of the affected paw was measured for 50 min. Formalin-induced nociceptive behavior is biphasic, with the first and second phases recorded during 0–10 min and 10–40 min after injection of formalin, respectively. Time spent performing the licking/biting behavior in each 5-min block was recorded continuously during the first and second phases.

The degree of antinociception was determined by the plantar test (see Assessment of Thermal Hypersensitivity) and the paw pressure test in normal (nonligated) mice.

In the plantar test, a higher intensity of radiant heat than that used in ligated animals was applied. Otherwise, at the weaker intensity used in ligated animals (including before ligation), mice occasionally exhibited PWL values above 10 s, which is close to the cutoff latency of 15 s and prevented us from making a proper evaluation of analgesic effects on acute nociception.

After the plantar test, mice were subjected to the paw pressure test (Pressure Analgesy-Meter; Muromachi Kikai, Tokyo, Japan) to assess their threshold for acute mechanical nociception. In brief, while the experimenter gently held the body of each mouse, the right hind paw was exposed to increasing mechanical pressure. The pressure level was increased at a rate of 10 mmHg/s, and the pressure (mmHg) required to elicit a response was determined for each mouse, that pressure being defined as the nociceptive threshold. Paw pressure measurements were made in duplicate, and the mean of the two values was used for calculations. The cutoff pressure was 200 mmHg.

Mice either developing mechanical hypersensitivity after partial nerve ligation or subjected to a sham operation were killed by cervical dislocation during ether anesthesia. The brain was then removed and both hippocampi were quickly dissected out in ice-cold low-sodium artificial cerebrospinal fluid (pH 7.4 after bubbling with 95% O²and 5% CO²) containing 215.5 mm sucrose, 3.5 mm KCl, 1.24 mm KH²PO⁴, 26 mm NaHCO³, 10 mm D-glucose, 2.4 mm CaCl², and 1.3 mm MgSO⁴. Transverse hippocampal slices 400 μm thick were prepared using a vibratome (DSK-2000; Dosaka, Kyoto, Japan) and then maintained for at least 60 min at 32°C. The slices were then transferred to a recording chamber and superfused with standard artificial cerebrospinal fluid (at 28°–29°C, pH 7.4 after bubbling with 95% O²and 5% CO²) containing 124 mm NaCl, 2 mm KCl, 1.24 mm KH²PO⁴, 26 mm NaHCO³, 10 mm D-glucose, 2.4 mm CaCl², and 1.3 mm MgSO⁴. Each slice was completely submerged in the medium and superfused continuously at a rate of 2.0–2.5 ml/min.

The field excitatory postsynaptic potentials (fEPSPs) were recorded from the striatum radiatum of the CA1 area with glass electrodes (1–5 MΩ) filled with 2 m NaCl after electrical stimulation of the Schaffer-collateral/commissural fibers with square wave pulses (0.1-ms duration at a baseline frequency of 0.033 Hz) via  a bipolar enamel-coated stainless steel electrode. Test stimuli were applied to produce 30% of the maximum fEPSPs. After stable baseline fEPSPs had been recorded for approximately 60 min, long-term potentiation (LTP) was induced by high-frequency stimulation with a train of 100 shocks delivered at 100 Hz. The fEPSPs were acquired and their slopes were measured using the Measurement and Analysis System for Extracellular Potentials (Furusawa Lab Appliance, Kawagoe, Japan). The values of the measured slopes were normalized to the average of baseline responses obtained during 60 min before induction of LTP.

Glycine and sarcosine were purchased from Nacalai (Kyoto, Japan) and Sigma Chemical (St. Louis, MO), respectively, and dissolved in 0.9% physiologic saline. NFPS was obtained from Tocris Cookson (Bristol, United Kingdom; in vivo  experiments with the permission of NPS Pharmaceuticals Inc., Parsippany, NJ) and dissolved in 20% 2-hydroxypropyl-β-cyclodextrine. In the behavioral studies, glycine, sarcosine, and NFPS were injected intrathecally. These drugs or their vehicle were administered in a volume of 5 μl via  a disposable 27-gauge needle, which was inserted into the subarachnoid space through the intervertebral foramen between L5 and L6 in accordance with the method described by Hylden and Wilcox.25In the electrophysiologic study, NFPS was dissolved in 20% 2-hydroxypropyl-β-cyclodextrine and then administered by bath application at 1,000-fold dilution with standard artificial cerebrospinal fluid.

All data are expressed as mean ± SEM. The effects of drugs (glycine, sarcosine, and NFPS) on the nociceptive threshold in both the plantar and von Frey tests were evaluated with respect to time; the time of administration of drugs was defined as time 0. In the formalin test, glycine, sarcosine, or NFPS was administered 10 min before formalin injection, and the time of formalin injection was defined as time 0. The statistical analysis was made using Java Applets & Servlets for Biostatistics software (programmed by Hideki Ono, Ph.D., Professor, Nagoya City University, Nagoya, Japan). Two-tailed multiple t  test with Bonferroni correction after one-way analysis of variance was used.26Differences with P < 0.05 were considered significant.

We first injected glycine intrathecally to examine whether the increase in glycine actually reduces thermal and/or mechanical hypersensitivity developing after partial ligation of the sciatic nerve. As figure 1Ashows, intrathecal glycine (0.3 and 1 μg) significantly increased the PWL in the plantar test and the 50% threshold in the von Frey test in a dose-dependent manner. Similar significant reduction of thermal and mechanical hypersensitivity was obtained after intrathecal injection of the GlyT1 inhibitors sarcosine (10 and 30 μg; fig. 1B) and NFPS (0.03 and 0.1 μg; fig. 1C). Maximal analgesic effects, which at the doses used were not increased beyond the averaged PWL or the 50% threshold values measured before ligation, were generally observed within 30 min after injection of these drugs, and the effects mostly returned to the predrug level within 120 min.

Sixty mice developing diabetic mechanical hypersensitivity 2 weeks after streptozotocin administration were used. Their mean blood glucose concentrations and 50% thresholds before and 2 weeks after streptozotocin administration were 138.4 ± 3.4 mg/dl and 0.503 ± 0.021 g, and 552.9 ± 8.2 mg/dl and 0.060 ± 0.004 g, respectively. Intrathecal injection of glycine (0.3, 1, and 3 μg), sarcosine (10 and 30 μg), and NFPS (0.03 and 0.1 μg) reduced mechanical hypersensitivity dose dependently and significantly, and the effect of NFPS was relatively long-lasting (fig. 2). Again, these drugs at the doses used did not generally produce analgesic effects beyond the nociceptive withdrawal levels obtained before ligation.

Glycine (0.3 and 1 μg), sarcosine (10 and 30 μg), and NFPS (0.03 and 0.1 μg), which were respectively intrathecally administered 10 min before formalin injection, reduced the second phase of formalin-induced nociceptive behavior. By contrast, none of these drugs had any significant influence on the first phase (fig. 3). Generally, mice pretreated with 20% 2-hydroxypropyl-β-cyclodextrine (vehicle for NFPS) exhibited less nociceptive behavior after formalin injection compared with those pretreated with saline (vehicle for glycine and sarcosine). Nonetheless, NFPS revealed prominent inhibition of the second phase of formalin-induced nociceptive behavior (fig. 3C). Although the first phase of formalin-induced nociceptive behavior (10–20 min after drug administration) was not influenced, these drugs generated obvious analgesic effects (both in the Seltzer and streptozotocin-induced diabetic neuropathy models) 15 min after drug administration, suggesting that the analgesic effects were not elicited secondarily via  impairment of motor activity.

The experiment on acute nociception was in addition made to know whether blockage of GlyT1 by NFPS at the effective doses in chronic pain states generates analgesic effects on acute nociception. When assessing the effects on acute nociception in the plantar test, a higher intensity of radiant heat than that used in ligated animals was applied. Otherwise, at the weaker intensity used in ligated animals (including before ligation), mice occasionally exhibited PWL values above 10 s, which is close to the cutoff latency of 15 s and prevented us from making a proper evaluation of analgesic effects on acute nociception. In nonligated naive mice, intrathecally injected NFPS (0.03 and 0.1 μg) significantly increased the PWL in the plantar test and the nociceptive threshold in the paw pressure test (fig. 4). The antinociceptive effects elicited at 0.1 μg seemed to be equivalent to those of morphine hydrochloride at 10 μmol/kg, subcutaneous (approximately 3 mg/kg) assessed in the plantar test (see our previous data, fig. 1of Sakaue et al.  27).

We have recently demonstrated that LTP in the hippocampal CA1 region is impaired after peripheral nerve injury.17In agreement with this, the fEPSP slope for hippocampal slices prepared from mice after peripheral nerve injury was significantly less potentiated compared with that in mice given a sham operation (fig. 5; 165.0 ± 5.6% in Seltzer vs.  207.1 ± 11.5% in sham control [n = 8 each, P < 0.05], 60 min after high-frequency stimulation in the presence of vehicle for NFPS). By contrast, in the presence of NFPS (25 nm), which at this concentration alone did not significantly influence LTP in slices prepared from sham-treated mice, the fEPSP slope was equally potentiated in slices after nerve injury and a sham operation (fig. 5; 211.3 ± 10.0% in Seltzer vs.  223.2 ± 9.8% in sham control [n = 8 each], 60 min after high-frequency stimulation). Therefore, NFPS improved the maintenance of LTP, which was impaired after peripheral nerve injury.

A net predominance of excitatory over inhibitory influences on spinal neuronal networks related to pain signaling is likely to be one crucial factor that underlies chronic pain conditions. Hence, normalization of excitatory and inhibitory balance in the spinal dorsal horn by strengthening inhibition may have therapeutic consequences for neuropathic and/or inflammatory pain. Accordingly, the current study clearly demonstrated that administration of glycine or blockade of endogenous glycine reuptake by GlyT1, and therefore an increase of extracellular glycine in the spinal cord, reduced thermal and mechanical hypersensitivity developing after partial ligation of the sciatic nerve, mechanical hypersensitivity in diabetic neuropathy, and formalin-induced persistent nociceptive behavior. More importantly, blockade of glycine uptake via  GlyT1 also normalized the weakened LTP in hippocampal CA1 hippocampal synaptic transmission, suggesting that an increase of endogenous glycine not only removes chronic pain at the spinal level but also supraspinally ameliorates the decrease in synaptic efficacy presumably related to cognitive disturbance which is often described in patients with chronic pain.18,19 

Dysfunction of inhibitory synaptic transmission in the spinal dorsal horn, and therefore a net increase in spinal excitation, is considered to be a crucial mechanism involved in the pathophysiology of chronic pain states.3However, it is currently unclear whether impairment of GABAergic and/or glycinergic inhibition is of presynaptic or postsynaptic origin. Peripheral nerve injury reduces presynaptic releasable GABA pool and induces functional loss of GABAergic inhibitory synaptic transmission in the superficial dorsal horn.7Moreover, peripheral inflammation also leads to a reduction of glycine release from glycinergic nerve terminals in the spinal superficial layer.8By contrast, Polgár et al.  28,29revealed no evidence for a reduction of GABAergic or glycinergic neurons after peripheral nerve injury. While prostaglandin E², one of the important mediators of pain and inflammation, reduces glycinergic transmission by a postsynaptic mechanism involving the glycine receptor α³subunit in the spinal superficial dorsal horn,30,31this mechanism does not seem to contribute to pain after peripheral nerve injury or the formalin test.32Therefore, various presynaptic and/or postsynaptic mechanisms may contribute to spinal disinhibition in neuropathic and inflammatory pain conditions. Our study focusing on glycinergic inhibition has demonstrated that it is possible to normalize excitatory and inhibitory balance by using existing functional inhibitory mechanisms including presynaptic and postsynaptic components. As presented here, successful pain relief by blockade of glycine uptake both in neuropathic pain models including peripheral nerve injury and diabetic neuropathy and in the formalin test implies that GlyT1 blockers have potential for the treatment of chronic pain. In addition, pathophysiologically, it seems that glycine is functionally released presumably from spinal glycinergic interneurons even in these chronic pain conditions. Hence, the use of available extracellular glycine may provide analgesic effects. It was noteworthy that NFPS generated only analgesic effects below the basal nociceptive withdrawal levels obtained before ligation of the sciatic nerve or streptozotocin administration at doses that produced antinociceptive effects in naive animals. This may reflect reduced functional glycinergic inhibition after peripheral nerve injury or in diabetic neuropathy.

It has been generally accepted that there are many similarities in the synaptic plasticity underlying memory and central sensitization.1,2Hence, it should be borne in mind that drugs influencing central sensitization potentially impair memory function. In fact, the benzodiazepine diazepam impairs memory function in human and animals33,34and blocks LTP in the rat hippocampus35and LTP of C fiber–mediated field potentials in the rat spinal dorsal horn,36a synaptic model that has been proposed to describe the hypersensitivity underlying increased efficacy of nociceptive transmission.37However, as we demonstrated here, GlyT1 blockade had different influences on synaptic plasticity in the forebrain and spinal cord. In addition to the blockade of glycine reuptake near inhibitory glycinergic terminals in the spinal cord, glial GlyT1 regulates glycine levels close to NMDA receptors, where glycine acts as a coagonist of glutamate.11Hence, the analgesic effect in chronic pain models and the antinociception in naive animals may reflect a net neuronal inhibition at the spinal dorsal horn as a result of overcoming any excitatory effects via  glycine-mediated activation of NMDA receptors. After blockade of GlyT1, glycine may reach strychnine-sensitive glycine receptors that are also expressed functionally in the hippocampus.38–40Nevertheless, excitatory effects of glycine on synaptic plasticity were observed at the hippocampus after peripheral nerve injury. We have recently demonstrated that the LTP of the Schaffer-collateral synapses in the hippocampal CA1 region elicited by high-frequency electrical stimulation in slices prepared from neuropathic mice was maintained at a significantly lower level than that in sham-treated mice.17Such impairment of LTP after peripheral nerve injury was never observed when the LTP was induced in the presence of the GlyT1 blocker NFPS. Although not significant, the LTP in the sham-treated groups was also enhanced by NFPS. These results are consistent with studies demonstrating facilitatory effects of GlyT1 blockers on LTP in the CA1 region41and dentate gyrus42of the hippocampus. Chattipakorn and McMahon et al.  43have suggested that increases in glycine concentration can elicit neuronal inhibition particularly when excitability is high. This not only supports the analgesic effects produced by increasing the spinal glycine level but also implies decreased neuronal activities of the hippocampus after peripheral nerve injury. The latter may be plausibly ascribed to reduced activities of NMDA receptors that can be restored by blockade of glycine uptake. Further studies are ongoing in our laboratory to clarify whether peripheral nerve injury alters the intrinsic activities of hippocampal NMDA receptors or glycine uptake via  GlyT1.

Because GlyT2 inhibitors are not commercially available, we were unable to examine their effects on chronic pain. Because GlyT2 exhibits exclusively neuronal expression and is located in inhibitory glycinergic nerve terminals,44its blockade may increase glycinergic inhibition and therefore potentially lead to pain relief. However, a recent study using GlyT2-deficient mice has revealed that GlyT2 has an essential role in the recycling of glycine at glycinergic nerve terminals for vesicular release.12Hence, long-term use of GlyT2 blockers may possibly lead to diminished glycinergic inhibition with a consequent deterioration of chronic pain conditions, in contrast to the case of GlyT1-deficient mice that develop hyperglycinergic symptoms.45 

In conclusion, the current study has demonstrated that either exogenously applied glycine or endogenously increased extracellular glycine via  blockade of GlyT1 can lead to sufficient pain relief in neuropathic and inflammatory pain. In particular, GlyT1 blockers seem to elicit both supraspinal excitatory and spinal inhibitory actions of glycine in chronic pain conditions, suggesting that this could be a novel therapeutic approach for patients with chronic pain.