Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The dicarbamate compound FBM (2-phenyl-1,3-propanediol dicarbamate) has been shown to have anticonvulsant activity in several animal seizure models including maximal electroshock-, pentylenetetrazol- and picrotoxin-induced seizures in rodents (Swinyard et al., 1986; Coffin et al., 1994), and focal seizures induced by aluminum hydroxide injection into pre- and postcentral gyri in rhesus monkeys (Perhach et al., 1986). Moreover, FBM is efficacious in the treatment of human seizures including partial complex seizures and Lennox-Gastaut syndrome (Bourgeois et al., 1993; Burdette et al., 1992; Faught et al., 1993; Leppik et al., 1991; Sachdeo et al., 1992; The Felbamate Study Group in Lennox-Gastaut Syndrome, 1993; Theodore et al., 1991). Although the primary mechanism of action has not been firmly established, the neuropharmacological profile of this compound appears to be distinct from more classic anticonvulsant medications (McCabe et al., 1993; Porter, 1989; Rho et al., 1994; Sofia et al., 1991; White et al., 1992).

FBM appears to act as a functional antagonist of the NMDA receptor-ionophore complex, a multi-subunit heterooligomer (Kutsuwada et al., 1992) with multiple, allosterically coupled recognition sites for glutamate, glycine, polyamines, ions and use-dependent channel blockers (for review see McBain and Mayer, 1994). Furthermore, converging lines of evidence have shown that FBM may produce this effect through an interaction with the strychnine-insensitive glycine recognition site of the NMDA receptor. First, FBM reduces the increase in intracellular [Ca⁺⁺]_i stimulated by NMDA and glycine (Taylor et al., 1995; White et al., 1995). Second, the anticonvulsant effects of FBM in mice are reversed by glycine (Coffin et al., 1994; De Sarro et al., 1994) and D-serine, a glycine site agonist (De Sarro et al., 1994; White et al., 1995). Third, FBM acts as a neuroprotectant (Wasterlain et al., 1992; Wallis et al., 1992) in a glycine reversible fashion (Wallis and Panizzon, 1993). Fourth, FBM is capable of inhibiting the binding of DCKA, a high affinity glycine site antagonist, in membranes from both rat (McCabe et al., 1993) and human brain (Wamsley et al., 1994). Because occupancy of both the glycine and glutamate sites are necessary for NMDA-channel opening, and agonist activation of these sites causes neuronal depolarization (Monaghan et al., 1989), inhibition of glycine action at strychnine-insensitive glycine receptors would be expected to result in a decreased frequency of channel opening and a reduction of neuronal activation (Wong and Kemp, 1991). These observations strongly support the hypothesis that the anticonvulsant activity of FBM involves an action at the strychnine-insensitive glycine recognition site of the NMDA receptor.

Because our previous studies (McCabe et al., 1993) have demonstrated that FBM inhibits the binding of the competitive antagonist [³H]5,7-DCKA, we assessed the ability of FBM to modulate the binding of an agonist, [³H]glycine, to rat forebrain membranes and human brain sections. In addition, since various ions are capable of influencing the binding of glycine-site ligands, we assessed the effects of several ions on the modulation of [³H]glycine binding by FBM. We now report that, in contrast to its ability to inhibit [³H]5,7-DCKA binding, pharmacologically relevant concentrations of FBM increase [³H]glycine binding to both rat membrane homogenates and human cortical sections, and furthermore that this increase in binding is sensitive to ions which modulate the NMDA receptor.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Rat tissue preparation. Membranes were prepared from forebrains of male Sprague-Dawley rats (Taconic Farms, Germantown, NY) weighing 150 to 200 g. Animals were housed under a 12-hr light/dark cycle (lights on 0700) with access to food and water ad libitum. Animals were anesthetized with CO₂ and killed by decapitation. Tissues were assayed according to previously described methods (McCabe et al., 1993). Brains were immediately removed and forebrains rapidly dissected, weighed and placed in 10 volumes (original wet weight:volume) of 5 mM HEPES/4.5 mM Tris buffer (HTS; pH 7.8 at room temperature) containing 0.32 M sucrose. Tissue preparation was performed at 4°C. Tissues were homogenized by 8 to 10 passes of a motor-driven Teflon pestle in a glass tube, diluted to 50 volumes in HTS-sucrose, and centrifuged at 1,000 × g for 10 min. The pellet (P1) was discarded and supernatant centrifuged at 20,000 × g for 20 min. The pellet (P2) was resuspended in HTS using a Brinkmann Polytron (setting 6, 5 sec) and centrifuged at 8000 × g for 20 min. Subsequently, the supernatant and outer buffy coat (remaining pellet core discarded) was centrifuged at 20,000 × g for 20 min. The resultant pellet (P2/P3) was resuspended in HTS containing 1 mM EDTA and centrifuged at 20,000 × g for 20 min. The P2/P3 pellet was resuspended in HTS and the "washing" procedure repeated two more times. The P2/P3 pellet was then resuspended in 5 volumes of HTS, frozen on dry-ice and stored at -80°C for at least 72 hr before binding assay.

[³H]Glycine binding to rat brain membrane homogenates. Radioligand binding assays using [³H]glycine were performed as described (Baron et al., 1991) with minor modifications. On the day of the assay, the tissue was washed twice by resuspension (50 mM HEPES-KOH buffer, pH 7.4 at 4°C) and centrifugation at 20,000 × g for 20 min. Assays were performed at 4°C using quadruplicate samples in a total volume/tube of 0.5 ml consisting of: 250 µl membrane suspension (200 µg protein/assay tube) in 50 mM HEPES-KOH buffer (pH 7.4 at 4°C), 50 µl [³H]glycine (20 nM; 48.4 Ci/mmol; Du Pont/NEN, Boston, MA) solution, 50 µl drug or buffer and 150 µl buffer. FBM (10 mM stock) was dissolved 75% dimethylsulfoxide, and serial dilutions in buffer were prepared for assay. Pilot experiments revealed that this concentration of dimethylsulfoxide did not interfere with the assay (data not shown). Glycine or D-serine (100 µM) were used to define nonspecific binding. The ability of FBM to modulate [³H]glycine (20 nM) binding was also studied in rat lung tissue (prepared as described for rat forebrain) or heat-inactivated rat forebrain tissue (heated to 70°C for 60 min). Binding reactions were terminated after 60 min by centrifugation (20,000 × g for 20 min at 4°C). The supernatants were aspirated and pellets rinsed with 2 × 1-ml aliquots of ice-cold buffer. The pellets were solubilized in Solvable (25 µl; Packard Instrument Co., Meriden, CT). Scintillation cocktail (4 ml; F989, Du Pont/NEN) was added, and radioactivity measured in a Beckman LS 5801 liquid scintillation counter. Where appropriate, radioligand binding data were analyzed via iterative curve fitting software (GraphPad Prism Version 1.03, San Diego, CA).

[³H]Glycine binding to slices of human brain. Dissected blocks of human brain were obtained from the brain bank at the University of Auckland. These tissues were kept frozen at 70°C. For binding studies, blocks of cerebral cortex (middle temporal gyrus) from two female individuals (ages 62 and 66 yr with postmortem times of 14 and 15.5 hr) were thawed, pooled and lightly homogenized. These cortical tissues were placed in small cylindrical centrifuge tubes and refrozen. The frozen cylinders of tissue were cut into sections (18 µm in thickness) in a cryostat and thaw-mounted onto microscope slides. Binding of [³H]glycine to these sections was accomplished as previously described (McDonald et al., 1990).

2

9

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Effects of FBM on [³H]glycine binding to rat brain membranes. FBM (10-1000 µM) produced an increase in the specific binding of [³H]glycine (20 nM), with an EC₅₀ value of 485 µM and Hill slope of 1.8 (fig. 1). Maximum enhancement of [³H]glycine binding by FBM (1000 µM) was 211% of control. Under the described conditions, baseline specific binding of [³H]glycine was 155 ± 12.5 fmol/assay (100%). The specific binding of [³H]glycine was 60 to 80% of total binding. D-serine, (16-8000 nM), an agonist at the strychnine-insensitive glycine receptor, inhibited the binding of [³H]glycine (20 nM) with an IC₅₀ value of 370 nM. Incubation of rat brain membranes for 1 hr at 70°C eliminated specific [³H]glycine binding. FBM (650 µM) did not increase nonspecific binding in heat treated tissue. Specific binding of [³H]glycine was not observed in membranes prepared from rat lung, and FBM did not increase nonspecific [³H]glycine binding in this tissue.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Stimulation of [3H]glycine (20 nM) binding to rat forebrain membranes by FBM. [3H]Glycine binding was assayed in the presence of increasing concentrations of FBM (10-1000: µM; ). Assays were performed as described in "Materials and Methods." Data are expressed as the percent of specific binding, were fit using a sigmoidal dose-response (variable slope) equation and represent mean ± S.E.M. (intra-assay variability) of quadruplicate samples. Each assay tube contained 200 µg protein. In this experiment EC50 was 485 µM, Emax was 211% and Hill slope was 1.8 (r2 = 0.91). Comparable values were obtained from two separate determinations.

Effects of FBM on [³H]glycine binding to human brain slices. Approximately 85% specific binding to strychnine-insensitive glycine receptors in sections of human cortex was obtained with [³H]glycine using the conditions outlined in "Materials and Methods." This binding was enhanced by FBM in a dose-dependent fashion with an EC₅₀ value of 142 µM and Hill slope of 1.6, with an E_max measured at 157% of control in the presence of a 1 mM concentration of the drug (fig. 2). Inclusion of higher concentrations of FBM in the incubation media was impractical due to limited solubility of the drug. The increase in [³H]glycine binding was unique to FBM because the other anticonvulsant drugs tested (carbamazepine, phenytoin, valproic acid and phenobarbital; 300 µM) did not increase [³H]glycine binding (data not shown).

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Stimulation of [3H]glycine (50 nM) binding to human brain slices by FBM. [3H]Glycine binding was assayed in the presence of increasing concentrations of FBM (0.1-1000 µM; ). Assays were performed as described in "Materials and Methods." Higher concentrations of FBM were not be used due to solubility limitations. Data are expressed as the percent of specific binding, were fit using a sigmoidal dose-response (variable slope) equation and represent mean ± S.E.M. of three experiments with triplicate samples. The EC50 was 142 µM, Emax was 157% and Hill slope was 1.6 (r2 = 0.99).

Effects of ions on FBM-enhanced [³H]glycine binding to human brain slices. Inclusion of various ions in the incubation media modified FBM-induced increases in [³H]glycine binding (figs. 3 and 4). Reduced binding of [³H]glycine in the presence of 300 µM FBM (which increased binding to 120% of control in the absence of ions) was found when sodium chloride (81%), potassium hydroxide (71%) or calcium nitrate (58%) were included at a 10 mM concentration (fig. 3). Calcium chloride produced only a modest reduction in FBM-stimulated [³H]glycine binding (89%). The presence of 10 mM zinc chloride had no effect on [³H]glycine binding, but was able to prevent the increase in binding elicited by 600 µM FBM (150% of control; fig. 4). However, zinc acetate caused an increase in [³H]glycine in the absence of drug and still decreased the binding in the presence of FBM. Slight, but consistent differences in the ability of FBM to induce increases in [³H]glycine binding in the presence of magnesium ions were also noted (fig. 3). When FBM was present at 300 to 600 µM, a 10 mM concentration of magnesium sulfate enhanced the binding (120% of FBM alone), whereas magnesium chloride did not share this effect.

View larger version (46K): [in this window] [in a new window]   Fig. 3.   Stimulation of [3H]glycine (50 nM) binding by FBM in human postmortem brain: modulation by ions. [3H]Glycine binding to human brain slices was assayed in the presence of FBM (300 µM) and 10 mM of MgSO4, MgCl2, CaCl2, NaCl, KOH or Ca(NO3)2. The dashed line represents maximal binding enhancement by FBM in the absence of ions. Data are expressed as the percent of specific binding, and represent mean ± S.E.M. of three experiments with six replicate samples.

View larger version (46K): [in this window] [in a new window]   Fig. 4.   Stimulation of [3H]glycine (50 nM) binding by FBM in human postmortem brain: modulation by Zn++. [3H]Glycine binding to human brain slices was assayed in the absence or presence of FBM (600 µM) and in the absence or presence or either of ZnCl2 or Zn(C2H3O2)2 [abbreviated ZnAc], (each 10 mM). Effect of zinc on FBM enhanced [3H]glycine binding. FBM (600 µM) increased [3H]glycine binding to 150% of control. The dashed line represents basal [3H]glycine binding in the absence of FBM or ions. Data are expressed as the percent of specific binding, and represent mean ± S.E.M. of three experiments with six replicate samples.

Autoradiographic analysis of [³H]glycine binding to human brain slices. Autoradiographic localization of [³H]glycine binding showed the ubiquitous presence of these sites in most of the human brain regions examined. In the cortex (middle temporal gyrus), the binding was concentrated in the superficial layers (I-III), but was present to a lesser extent in all layers (fig. 5; table 1). Very high levels of binding were found in the molecular layer of the dentate gyrus, stratum pyramidale of the CA1 and CA2 regions of the hippocampus (fig. 6; table 1). Somewhat lower levels of binding were found in the stratum oriens and radiatum of CA1 and CA2, and in the hilus of the area dentata (table 1). Intermediate to low binding was also noted throughout the CA3 region, the subiculum, and extending out into the parahippocampal gyrus (fig. 6; table 1).

View larger version (137K): [in this window] [in a new window]   Fig. 5.   Autoradiographic localization of [3H]glycine binding in the human temporal cortex. Strychnine-insensitive glycine receptors appeared with the greatest density in the superficial layers (2) of the cortex lining the sulcus (A). The deep layers (d) also had a substantial density of these receptors which decreased incrementally as the deepest layer was approached (VI). Addition of FBM (600 µM) caused an increase in all layers (B), but was especially apparent in the superficial laminae of this area of cortex. No change in binding was noticed after the addition of carbamazepine (C). Glycine was used in inhibit the specific binding associated with [3H]glycine and the images reflect areas of nonspecific binding (D). Bar = 1 mm.

View larger version (139K): [in this window] [in a new window]   Fig. 6.   Autoradiographic localization of [3H]glycine binding in the human hippocampal formation: Effects of anticonvulsant drugs. A, Labeling of strychnine-insensitive glycine receptors with [3H]glycine is shown in the human hippocampus (h), dentate gyrus (d) and surrounding regions including the subiculum (s) and parahippocampal gyrus (p). The highest density of receptors is indicated by the red color appearing in the CA1 and CA2 regions of the hippocampus, and in the molecular layer of the dentate gyrus. B, Most of the binding is increased in the presence of FBM (600 µM), but is slightly decreased when C, carbamazepine (100 µM); D, phenytoin (100 µM); E, valproic acid (100 µM) or F, phenobarbital (100 µM), are included in the incubation medium. Bar = 1 mm.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

FBM is a pharmacologically unique anticonvulsant drug (White et al., 1992). Converging lines of evidence indicate that FBM exerts its anticonvulsant effects through an action at the strychnine-insensitive glycine site of the NMDA receptor (Coffin et al., 1994; De Sarro et al., 1994; McCabe et al., 1993; Taylor et al., 1995; Wallis and Panizzon, 1993; Wamsley et al., 1994; Wasterlain et al., 1992; White et al., 1995). Previously we have demonstrated that FBM, at pharmacologically relevant concentrations, is able to inhibit the binding of [³H]5,7-DCKA, an antagonist at the strychnine-insensitive glycine recognition site (McCabe et al., 1993; Wamsley et al., 1994). In our study we examined the binding of [³H]glycine, an agonist, and likely the endogenous ligand at this site. Surprisingly, we found that FBM, at concentrations similar to those inhibiting [³H]5,7-DCKA binding, was able to enhance specific, D-serine sensitive [³H]glycine binding (figs. 2 and 3) in a concentration-dependent manner. The ability of D-serine to inhibit [³H]glycine (20 nM) binding [IC₅₀ = 370 nM; in agreement with previously reported data (Baron et al., 1991)] to rat forebrain membranes suggests that we are measuring [³H]glycine binding to the glycine recognition site of the NMDA receptor complex. Specific binding of [³H]glycine was not seen in either rat lung tissue or heat-inactivated rat brain tissue, and FBM did not increase nonspecific binding in these tissues, indicating that the FBM-induced increase in [³H]glycine binding represents an increase in specific binding. In addition to an action on rat brain membranes, FBM was able to increase [³H]glycine binding in human brain slices (figs. 6, 9, 10 and 11; table 2). These results suggest that the ability of FBM to produce a consistent, concentration-dependent increase in [³H]glycine binding is not species specific, and of potential clinical relevance in man.

Our results are in part consistent with the results of Subramaniam et al. (1995) who demonstrated that felbamate may inhibit the NMDA receptor through an allosteric mechanism. However, whereas our previous studies (McCabe et al., 1993; Wamsley et al., 1994) demonstrated the ability of felbamate to inhibit [³H]5,7-DCKA binding, Subramaniam et al. (1995) did not see inhibition of [³H]5,7-DCKA binding by felbamate. Furthermore, Subramaniam et al. (1995) reported that felbamate acts as a competitive antagonist of [³H]MK-801 binding, although previous reports were unable to show inhibition of agonist-enhanced [³H]MK-801 binding (McCabe et al., 1993; White et al., 1992). The reason for the lack of agreement in these conflicting sets of data are unclear. Interestingly, Rho et al. (1994) have shown that at low glycine concentrations (<0.3 µM) 1 mM FBM potentiated NMDA responses in cultured rat hippocampal neurons, which is consistent with the results of our study. It is clear from these recent studies that FBM is producing a functional antagonism of the NMDA receptor. Moreover, because FBM inhibits [³H]5,7-DCKA binding although enhancing [³H]glycine binding, it is unlikely that this antagonism is the result of competitive antagonism of the glycine site, rather, FBM likely influences the glycine site through an allosteric mechanism.

The ability of several other structurally diverse anticonvulsant compounds to modulate [³H]glycine binding was then assessed. Previously, we have demonstrated that other anticonvulsant compounds including carbamazepine, phenobarbital, phenytoin and valproic acid were unable to inhibit [³H]5,7-DCKA binding to rat brain membranes (McCabe et al., 1993). Similarly, in our study these same anticonvulsant drugs were unable to increase [³H]glycine binding in slices of human brain (table 1). The finding that phenobarbital significantly reduced [³H]glycine binding in many regions of the human brain (table 1) was unexpected. This property of phenobarbital was not noticed in the tissue slices analyzed by scintillation counting even though this involved a higher concentration of the drug. In both the present study and our previous study (McCabe et al., 1993), the amount of anticonvulsant drug to include in similar incubations was arbitrarily chosen at a level eight times that achievable in brain and plasma. Thus, a 600 µM concentration of FBM is reasonable and achievable, whereas a 100 to 300 µM concentration of the other agents is probably too high to be clinically relevant. It is interesting to note that valproic acid showed some propensity to interact with glycine receptors labeled with [³H]5,7-DCKA (Wamsley et al., 1994), but did not show any such trend with [³H]glycine. These data further support the hypothesis that FBM has a mechanism of action that is distinct from other clinically used anticonvulsant drugs (White et al., 1992).

Strychnine-insensitive glycine receptors exist in the forebrain as part of the NMDA receptor/ionophore complex (Monaghan et al., 1989; Wong and Kemp, 1991). Stimulation of the glycine receptor sites is thought to enhance channel opening and hence NMDA-mediated stimulatory effects. In fact, some strychnine-insensitive glycine receptor activity appears to be requisite in this channel opening process (Kleckner and Dingledine, 1988). Inhibition of glycine receptors would therefore be expected to reduce channel opening and thus reduce NMDA-mediated seizure-genic and potential neuronal damaging effects (Carter, 1992). Clinical studies have shown FBM to be efficacious as an anticonvulsant drug (Bourgeois et al., 1993; Burdette et al., 1992; Faught et al., 1993; Leppik et al., 1991; Sachdeo et al., 1992; The Felbamate Study Group in Lennox-Gastaut Syndrome, 1993; Theodore et al., 1991), and preclinical studies have indicated that the strychnine-insensitive glycine receptors might be involved in mediating the anticonvulsant and neuroprotectant effects of FBM (McCabe et al., 1993). An antagonist action of FBM at these glycine receptors would explain its anticonvulsant and neuroprotectant effects. Although the ability of FBM to inhibit [³H]5,7-DCKA binding is consistent with this hypothesis, the ability of FBM to stimulate the binding of [³H]glycine at similar concentrations [IC₅₀ against [³H]5,7-DCKA binding: 374 µM (McCabe et al., 1993); EC₅₀ for enhancement of [³H]glycine binding: 485 µM (our study)] is surprising. Nevertheless, the ability of a compound that interacts with the NMDA receptor to produce opposite effects on the binding of agonists and antagonists is not without precedent. For example, competitive NMDA antagonists will inhibit [³H]5,7-DCKA binding, although agonists such as NMDA will enhance [³H]5,7-DCKA binding (Baron et al., 1991). The polyamine spermine increases the affinity of NMDA receptor antagonists although decreasing the affinity of NMDA receptor agonists (Pullan and Powel, 1991).

Several possible explanations may account for the seemingly paradoxical ability of FBM to enhance [³H]glycine binding in vitro although behaving as a glycine antagonist in vivo (and inhibiting the binding of [³H]5,7-DCKA in vitro). All of the strychnine-insensitive glycine receptors in the human forebrain may not be involved to the same extent in mediating the actions of FBM. For instance, FBM may be a partial agonist or may be differentially affecting subtypes of glycine receptors. That different subtypes of glycine receptors exist is suggested by studies which demonstrate discrepancies between the binding of [³H]glycine and [³H]5,7-DCKA. This is most apparent in the cerebral cortex where [³H]5,7-DCKA shows a uniform labeling of all layers of the cortex except a narrow band of intense labeling bordered by an equally narrow band of light binding in the intermediate layers of the middle temporal gyrus (Wamsley et al., 1993). In contrast, [³H]glycine labels the outer layers of cortex much more than the deeper layers and does not show this banded appearance in the intermediate zone. Different conformations (high and low affinity) of glycine receptors might also exist in the cortex. The antagonist would recognize both conformations with equal high affinity whereas the agonist would preferentially bind to the high affinity sites. These high affinity receptors would be expected to be concentrated in the superficial layers of the cortex, whereas the low affinity sites would predominate in the deeper laminae. Previous studies have also indicated discrepancies exist between the distribution of sites recognized by glutamate and glycine, leading these investigators to speculate on the possible existence of subtypes of glycine receptors (McDonald et al., 1990; O'Shea et al., 1991). Another possibility is that glycine receptor subtypes exist with overlapping but nonidentical binding sites for FBM, glycine and glycine antagonists such as 5,7-DCKA. This could explain the discrepancies in binding which exist between [³H]5,7-DCKA and [³H]glycine and indicate how FBM can act as an anticonvulsant agent and inhibit [³H]5,7-DCKA binding at the same time as increasing [³H]glycine binding. That FBM interacts in a complex manner with NMDA receptors to increase [³H]glycine binding is further suggested by the finding that the hill coefficient (~1.8 in rat forebrain membranes, ~1.6 in human cortex sections) is greater than unity. The final significance of the ability of FBM to increase [³H]glycine binding will require further investigation.

The finding that FBM both increases [³H]glycine binding and decreases [³H]5,7-DCKA binding prompted us to examine the potential interaction of FBM with other sites known to modulate through allosteric means the binding of glycine site ligands. Polyamines produce a differential effect on the potency of glycine agonists and antagonists (Sacaan and Johnson, 1989). However, in a previous report (McCabe et al., 1993) we demonstrated that FBM did not influence the ability of spermine to enhance [³H]MK-801 binding, suggesting that an action at the polyamine site was unlikely. Because divalent cations can modulate the binding of [³H]glycine (Marvizón and Skolnick, 1988) and the glycine site partial agonist [³H]1-aminocyclopropanecarboxylic acid (Popik et al., 1995), as well as [³H]MK-801 (Reynolds and Miller, 1988; Wong et al., 1988), we assessed the interaction of FBM with various ions that might modulate [³H]glycine binding.

Zinc chloride had no effect on [³H]glycine binding, whereas zinc acetate caused a modest increase in binding to human brain slices. Both compounds attenuated the FBM-induced increase in [³H]glycine binding in human brain slices. Previous studies have indicated zinc inhibits glycine binding and selectively blocks NMDA receptor-mediated responses (Wong and Kemp, 1991). This led to a proposed extracellular binding site for zinc which is independent of other ion sites. It is clear from the present data that the presence of zinc interferes with the ability of FBM to increase [³H]glycine binding. Whether zinc interferes directly with the binding of felbamate or indirectly by opposing a conformational change induced by felbamate remains to be elucidated.

Similar to the effect of FBM, magnesium ions have been shown to increase the specific binding of [³H]glycine (Marvisón and Skolnick, 1988) as well as [³H]1-aminocyclopropanecarboxylic acid (Popik et al., 1995). Magnesium also has been shown to block NMDA receptor associated channels in a voltage-dependent fashion. A distinct site for magnesium has been proposed within the channel where the gating mechanism is manifest (Wong and Kemp, 1991). In our study, a high concentration of magnesium ions did not attenuate the ability of FBM to stimulate [³H]glycine to human brain slices. Rather, magnesium sulfate slightly increased the binding of [³H]glycine in the presence of FBM although magnesium chloride had no effect. Because FBM and magnesium produce similar effects on [³H]glycine binding, and are not additive in our study, it is possible that felbamate and magnesium may be acting through a common mechanism, and perhaps a common site, to increase [³H]glycine binding. Further experiments will be required to resolve this issue.

The alteration of the ability of FBM to increase [³H]glycine binding in the presence of NaCl in human cortex slices might suggest an uptake mechanism is involved. Sequestration of [³H]glycine intracellularly might be viewed as increased binding to a transporter complex. Several experiments were performed to address this possibility. Preloading cells with glycine in the presence of FBM prevented the subsequent increase in [³H]glycine elicited by FBM alone. This blockade could be duplicated with D-serine but could not be accomplished with equimolar concentrations of D, L-serine. D-Serine is thought to share binding sites with glycine (Monaghan et al., 1989; Wong and Kemp, 1991) but is not thought to bind to glycine uptake sites. Preloading the cells with glycine might saturate the uptake process such that further uptake of [³H]glycine is reduced. This does not seem to be a feasible explanation due to the ability of D-serine to block the FBM-induced increase in [³H]glycine binding that would apparently require initial binding and recognition of the uptake site. Other data that would apparently rule out this possibility was obtained with taurine. Taurine recognizes and binds to the glycine transporter complex, but is not taken up by the cells (McDonald et al., 1990). Taurine did not disturb [³H]glycine binding alone or in the presence of FBM. Thus, the increase in [³H]glycine binding induced by FBM occurs in the presence of an agent which is thought to block the glycine uptake process. Taken together, these data imply that the effect of felbamate is not a result of increased binding to the glycine reuptake site.

Autoradiographic localization of [³H]glycine binding indicated a prevalence of these receptor sites in portions of the human cerebral cortex and hippocampal formation. All of the binding in these areas was increased to varying degrees by the presence of FBM. Least affected was the binding in the molecular layer and hilus region of the dentate gyrus. The greatest change in [³H]glycine binding occurred in the CA1 region of the hippocampus and in the cortex. These same areas showed some discrepancies in the distribution of [³H]glycine binding with that of the glycine receptor antagonist [³H]5,7-DCKA used in previous studies (Wamsley et al., 1993).

In conclusion, FBM is capable of inhibiting binding of the antagonist [³H]5,7-DCKA (McCabe et al., 1993), although concomitantly enhancing [³H]glycine binding (our study), two ligands that compete for the same site. The most parsimonious explanation for this phenomenon is that FBM is interacting with a distinct site allosterically linked to the strychnine-insensitive glycine site of the NMDA receptor. FBM may have varying efficacy at this site depending on the subunit stoichiometry of the particular NMDA receptor. If true, then there may exist subtypes of NMDA receptors with varying sensitivity to FBM. Action at only a subset of NMDA receptors may further account for the lack of side effects typically seen after administration of competitive NMDA antagonists or use dependent channel blockers like MK-801. Finally, these data further support the hypothesis that the effectiveness of FBM as an anticonvulsant and neuroprotectant is most likely due to an interaction with, and functional antagonism of, the NMDA-receptor complex.