Abstract
The potencies of various N-methyl-d-aspartate (NMDA) receptor channel blockers were determined at recombinant NMDA receptors containing differing combinations of NR1 and NR2 subunits expressed inXenopus laevis oocytes. When the NR1 subunit was varied (NR1e/NR2A or NR1b/NR2A), none of the 9 channel blockers tested displayed a statistically different affinity. In contrast, altering NR2 composition changed the affinities of several channel blockers. Three of 10 compounds displayed significantly higher affinities for NR1b/NR2C receptors than NR1b/NR2A receptors, and three of five compounds had higher affinity at NR1b/NR2C than NR1b/NR2B receptors. Both MK-801 and N-[1-(2-thienyl)cyclohyxyl]piperidine displayed identical affinities at all receptor subunit combinations tested. However, these two compounds displayed significantly slower rates of blockade and unblockade at NR1b/NR2C than at NR1b/NR2A receptors, perhaps reflecting the shorter mean open times of NR1/NR2C receptors. NR1b/NR2B and NR1b/NR2A were distinguished by one of five compounds tested. Taken together, these results indicate that NR2 subunits impart differing pharmacological profiles to NMDA receptors; thus, it may be possible to develop NMDA receptor channel blocker antagonists of greater subtype selectivity.
Pharmacological agents that can selectively block subpopulations of NMDA receptors may have important therapeutic applications. Excessive NMDA receptor activation is thought to be critical to a wide variety of pathological conditions, such as seizure activity and neuronal cell death after ischemia, hypoglycemia, HIV infection and head trauma (Lipton and Rosenberg, 1994; Meldrum and Garthwaite, 1990). However, NMDA receptors also play important roles in various normal functions of the nervous system, such as long-term potentiation and long-term depression (Bear and Malenka, 1994), experience-dependent formation of synaptic connections in development (Artola and Singer, 1994), neuronal migration (Komuro and Rakic, 1993), pain modulation (Dickenson, 1990) and oscillatory depolarization patterns for locomotion (Headley and Grillner, 1990; Hochman et al., 1994). Thus, it is not surprising that NMDA receptor blockers have been associated with various adverse side effects, especially psychotomimetic effects and motor impairment (Carter, 1994; Rogawski, 1993; Willetts et al., 1990). If NMDA receptor blockers can selectively block different subpopulations, it may be possible to obtain therapeutic actions with minimal adverse side effects. Indeed, some recently characterized compounds display improved therapeutic profiles that could be due to improved subtype selectivity (Palmeret al., 1995; Rogawski, 1993; Subramaniam, et al., 1996).
The number of NMDA receptor subtypes is unknown; however, there are at least several subtypes and potentially many more. Two families of NMDA receptor subunits have been cloned: eight alternatively spliced forms of a single gene coding for NR1 subunits (Hollmann et al., 1993; Sugihara et al., 1992) and four different NR2 subunits (NR2A–NR2D) (Ikeda et al., 1992; Ishii et al., 1993; Kutsuwada et al., 1992; Meguro et al., 1992; Monyer et al., 1992, 1994). Native NMDA receptors are thought to be heteromers composed of NR1 and NR2 subunits because their coexpression is necessary for full activity (Meguro et al., 1992; Monyer et al., 1992). Recently, an additional subunit (χ-1 or NMDAR-L) has been cloned that may act to modulate NMDA receptor activity (Ciabarra et al., 1995; Sucher et al., 1995). Thus, if each NMDA receptor complex contains four or five subunits, then there are a large number of potential subunit combinations that may represent distinct NMDA receptor subtypes.
NMDA receptor channel blockers have been developed as potential therapeutic agents; however, relatively little is known regarding their relative potencies at different combinations of NMDA receptor subunits. NMDA receptors in rat brain display consistent regional variations in their pharmacological properties for channel blocker antagonists (Beaton et al., 1992; Ebert et al., 1991; Porter and Greenamyre, 1995), thus suggesting that NMDA receptors of differing subunit composition may have differing channel blocker pharmacological profiles. To date, the principal observation of NMDA receptor channel blocker subtype selectivity comes from studies of recombinant NMDA receptors expressed in X. laevis oocytes that report varied sensitivities to MK-801 among NR1/NR2 receptors composed of differing NR1 subunits (Rodriguez-Paz et al., 1995) or differing NR2 subunits (Yamakura et al., 1993). The purpose of the present study was to determine whether NMDA receptors that differ in NR1 and NR2 subunit composition display differing channel blocker pharmacologies. As was found for the pharmacology of the glutamate recognition site (Buller et al., 1994; Laurie and Seeburg, 1994), the NR2 subunit appears to confer distinct, pharmacological properties to the NMDA receptor complex.
Methods
In vitro transcription.
cDNAs encoding NMDAR1 subunits NR1b and NR1e (Sugihara et al., 1992) were the generous gift of Dr. Shigetada Nakanishi. NR1b is the splice variant containing each of the three alternatively spliced cassettes (also termed NR1111; Durand et al., 1993), and NR1e is the splice variant lacking all three cassettes (also termed NR1000). cDNAs, including NR2A, NR2B and NR2C (Monyeret al., 1992) (generously provided by Dr. Peter Seeburg) were linearized by digestion with NotI (NR1b and NR1e),EcoRI (NR2A and NR2C) or EcoRV (NR2B) and transcribed in vitro using T7 (NR1s) or T3 (NR2s) mMessage mMachine kits (Ambion, Austin, TX).
Translation in X. laevis oocytes.
Oocytes were removed from mature X. laevis (Xenopus One, Ann Arbor, MI, or NASCO, Fort Atkinson, WI) and were dissociated in 2 mg/ml collagenase (Type IA, Sigma Chemical, St. Louis, MO) in Ca++-free oocyte Ringers-2 (82.5 mM NaCl, 2.5 mM KCl, 1 mM MgCl2, 5 mM HEPES, pH 7.6), and the remaining follicle layer was removed manually. Stage V and VI oocytes were maintained in ND-96 (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, 5 mM HEPES, pH 7.6) supplemented with 2.5 mM sodium pyruvate and 50 μg/ml gentamicin. NMDA receptor RNAs were dissolved in sterile distilled H2O. NR1 RNA was mixed 1:3 with either NR2A, NR2B or NR2C RNA. Fifty nanoliters of the final RNA mixture (15–30 ng total) was microinjected into the oocyte cytoplasm. Oocytes were incubated in complete ND-96 solution at 17°C for 1–4 days before the electrophysiological assay. Animal handling procedures conformed to the Declaration of Helsinki and the National Institutes of Health “Guide for the Care and Use of Laboratory Animals.”
Electrophysiology.
Electrophysiological responses were measured using a standard two-microelectrode voltage-clamp (model OC-725A or OC-725B Oocyte Clamps, Warner Instruments, Hamden, CT) at a holding potential of −60 mV. Electrodes were filled with 3 M KCl and had resistances of 0.6–3.0 MΩ. The extracellular buffer contained 116 mM NaCl, 2 mM KCl, 2.0 mM BaCl2 and 5 mM HEPES, pH 7.4. Glutamate at 100 μM (or 3 or 10 μM in a few cases with NR2A-containing receptors) and 10 μM glycine were perfused until a stable plateau response was obtained (cell was eliminated if the response would not plateau). Channel blocker plus agonists were added until a steady state blockade was obtained and then agonist alone was added until a full agonist response recovered (to ensure complete antagonist removal from the channel). Agonists were then removed until base-line returned. To obtain plateau agonist responses, it was necessary to use NR1/NR2A receptors within 24 hr after RNA injection; responses of >90–100 nA did not plateau presumably due to barium-induced chloride currents (Leonard and Kelso, 1990). NR1/NR2C receptors were used for 1–3 days before responses were too large to plateau (which occurred at higher conductance levels: 200–300 nA). The ability of NR1/NR2C receptors to have larger currents without Ba++-induced Cl− currents is consistent with the reduced fraction divalent cation conductance of NR1/NR2C receptors (Burnashev et al., 1995). NR1/NR2B receptor responses were smaller (50–100 nA) and consistently would plateau.
Current responses to drug application were recorded on a strip-chart with or without digital capture using an ITC-16 computer interface (Instrutech, Great Neck, NY) and a MacIntosh computer with AxoData software (Axon Instruments, Foster City, CA). Dose-response curves for antagonist blockade of heteromeric NMDA receptor complexes expressed in oocytes were fit (Prism, GraphPAD Software, San Diego, CA) according to the equations: I = Imax/[1 + (IC50/A)n] and I = Imax/[1 + (IC50/A)], where I is the current response, Imax is the current response evoked at zero antagonist concentration, A is the antagonist concentration,n is the Hill coefficient and IC50 is the concentration of antagonist inhibiting 50% of the response. AxoGraph software (Axon Instruments) was used to determine time constants (τ) of channel blockade and unblockade using the equation: fractional current response at time t = 1 − exp(−t/τ), where t is time (Hille, 1984). The on-rate and off-rate constants were determined by standard methods: 1/τoff =k −1; k 1 is derived from the equation 1/τon = k on (observed) = k −1 + [drug] * k 1 (Svensson et al., 1994;Weiland and Molinoff, 1981) (see fig. 4).
MK-801 was obtained from Research Biochemicals (Natick, MA). ARL-designated compounds were kindly provided by Astra Pharmaceuticals (Rochester, NY), and the remaining compounds and chemicals were obtained from Sigma Chemical (St. Louis, MO). Chemical structures of compounds tested were (+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine hydrogen maleate [(+)-MK-801], 1,2-diphenyl-2-proplamine monohydrochloride (ARL 12495),cis-2,3,4,4a,9,9a-hexahydro-1H-fluoren-4a-amine monohydrochloride (ARL 15609), (+)-α-methyl-α-phenyl-2-pyridineethanamine dihydrochloride (ARL 15853), (−)-α-methyl-α-phenyl-2-pyridineethanamine (ARL 15859), (−)-α-phenyl-2-pyridineethanamine (ARL 15895), 2-amino-N-(1-methyl-1,2-diphenylethyl)acetamide monohydrochloride (remacemide), S-(+)-α-phenyl-2-pyridineethanamine dihydrochloride (ARL 15896) and N-[1-(2-thienyl)cyclohyxyl]piperidine (TCP).
Results
From 1 to several days after RNA injection into the oocytes, 100 μM glutamate plus 10 μM glycine evoked currents that were fully blocked by NMDA receptor antagonists. Uninjected oocytes did not give glutamate/glycine-evoked responses, and cells injected with NR1 alone yielded, at most, responses of 8 nA. Antagonist potencies were determined as illustrated in figure 1, with basal and maximal responses determined both before and after each antagonist application.
To determine the influence of different NR1 subunits on channel blocker pharmacology, we evaluated antagonist potency at NMDA receptors composed of NR2A subunits coexpressed with either NR1b or NR1e subunits. These are the two most diverse NR1 subunits, containing either all three alternatively spliced cassettes (NR1b) or none of the three cassettes (NR1e). None of the corresponding IC50values for nine antagonists differed between NR1e/NR2A and NR1b/NR2A receptors (table 1). In three cases (ARL 15609, ARL 15896 and remacemide), there was a nonsignificant trend for NR1b-containing receptors to display a lower affinity for the antagonist.
When the NR2 subunit was varied and the NR1 subunit was held constant (NR1b), several statistically significant pharmacological differences were found (table 1). The potencies of 10 channel blockers at inhibition of NR1b/NR2A and NR1b/NR2C receptor responses are shown in figure 2. Overall, there was a general trend for channel blockers to have a higher affinity at the NR1b/NR2C receptor, and in three cases (ARL 15896, ARL 15859 and dextromethorphan) the difference in affinity was statistically significant. Three compounds (ARL 15895, ARL 15609 and ARL 15853) distinguished between NR1b/NR2C and NR1b/NR2B receptors, and in each case the NR2C-containing receptor displayed higher channel blocker affinity. Receptors composed of NR1b/NR2A and NR1b/NR2B displayed very similar pharmacological properties, being weakly distinguished by only one of five compounds (ARL 15895). Unexpectedly, NR1b/NR2A receptors did not appear to be fully blocked by high concentrations of MK-801 (fig. 2A). Dose-response analysis by letting the percentage of maximal inhibition be a variable indicated maximal inhibition to be 76.4 ± 8.0% (mean ± S.E.M.). However, this value was not statistically different from 100% maximal inhibition. With this analysis, the corresponding MK-801 IC50 values were slightly altered from those presented in table 1. MK-801 affinity/percentage of maximal inibition values were 7.3 ± 0.6 nM/85.7% (NR1e/NR2A), 4.9 ± 0.8 nM/76.4% (NR1b/NR2A), 11.4 ± 1.9 nM/97.6% (NR1b/NR2B) and 6.9 ± 0.1 nM/83.3% (NR1b/NR2C). For MK-801 at NR1b/NR2A receptors (assuming 100% maximal inhibition), the corresponding Hill coefficient was 0.47 ± 0.10, whereas for the other drug/receptor combinations, the average Hill coefficient was 0.92, with a range of 0.68–1.21. In control experiments, we found that MK-801 does not increase (or decrease) the holding current in X. laevis oocytes (data not shown); thus, it is unlikely that full blockade is masked by an MK-801-induced increase in the holding current.
MK-801, as well as TCP, displayed very similar affinities at all of the heteromeric receptors tested. These compounds did, however, display differing rates of blockade and unblockade at receptors with differing NR2 subunits. TCP was significantly slower to develop steady state block and unblock at the NR1b/NR2C-containing receptor compared with the NR1b/NR2A receptor (fig. 3).
Kinetic analysis of TCP blockade and reversal of blockade shows that NR1b/NR2C receptors were associated with slower TCP on-rate (k 1) and off-rate (k -1) constants (fig. 4). The average TCP dissociation constant (± S.E.M.) derived from thek -1/k 1 ratio from individual experiments was 110 ± 21 nM for NR1b/NR2A and 119 ± 28 nM for NR1b/NR2C. The apparent on-rates were concentration dependent, whereas the off-rates generally were concentration independent. However, at the highest concentration, the off-rate was markedly reduced for the NR1b/NR2A receptor, perhaps suggesting that some of the compound is retained by the cell. NR1b/NR2C receptors also displayed slower rates of block and unblock by MK-801 than NR1b/NR2A receptors. For example, the half-time for channel block by 10 nM MK-801 at NR1b/NR2A receptors was ∼30 sec, whereas at NR1b/NR2C receptors, half-time for channel block was 150 sec. Due to the length of time to achieve steady state and the need to use lower membrane holding potentials to more quickly reverse MK-801 block at NR1b/NR2C receptors, kinetic rates were not quantified. It was not possible to evaluate the kinetic properties of blockade by any of the other drugs; their rate of blockade could not be resolved from the time it takes to change the drug solution around the oocyte. Drug exchange time was several-fold faster than the rate of blockade by TCP at NR1b/NR2A receptors.
Discussion
The major finding of this study was that alteration of the NR2 subunit composition in NR1/NR2 heteromers corresponded to small, but significant, changes in the antagonist pharmacology. In contrast, NMDA receptors that differ in their NR1 composition did not differ in their antagonist pharmacological profiles. These results suggest that differing NMDA receptor subtypes differ in their channel blocker pharmacological properties. The compounds tested in this study have been shown to be open channel blockers (Mealing et al., in press; Rogawski, 1993; Subramaniam et al., 1996)2 and were specifically chosen because of their varied therapeutic indices.,3 Some of the blockade action may also occur via nonchannel sites, as recently described for remacemide. However, the other ARL compounds do not have a glycine moiety, which is necessary for the nonchannel action of remacemide (Subramaniam et al., 1996).
The channel blocker binding site on NR1 homomeric NMDA receptors is thought to be in the apparent pore lining M2 and M3 regions (Ferrer-Montiel et al., 1995). Because this region is distant from the NR1 alternative splice sites (Sugihara et al., 1992), this might explain why the NR1 splice variants displayed identical channel blocker pharmacologies. NR2 subunits do, however, have varied residues in the region of the channel blocker binding site (Ishii et al., 1993; Kutsuwada et al., 1992; Monyer et al., 1992); thus, these residues could be contributing to the differential pharmacology of the channel blockers.
Because differing NR2 subunits correspond to small, but significant, variations in channel blocker potency, it is possible that differences in the therapeutic index of these drugs may be partially due to differing NR2 selectivity. Thus, a therapeutically useful compound might selectively block those NMDA receptors responsible for epileptic activity or stroke damage while weakly blocking those NMDA receptors whose blockade causes psychotomimetic or motor-impairment effects. Along these lines, it has been proposed that channel blockers with a higher affinity for the cerebellar NMDA receptors (NR2C-containing) may correlate with a higher therapeutic index (Porter and Greenamyre, 1995). In the present study, this correlation partially extends to NR1/NR2C NMDA receptors compared with NR1/NR2A and NR1/NR2B receptors. Thus, the drugs ARL 15896, dextromethorphan and ARL 15895 have a better therapeutic index than TCP, MK-801 or ARL 15609 (Greene et al., 1996; Rogawski, 1993)4; specifically, the former drugs have a higher potency at NR1/NR2C receptors than at NR1/NR2A receptors. The general correspondence between well-tolerated compounds and those with higher affinity at NR1/NR2C receptors does not extend, however, to remacemide; this compound is well tolerated but does not have a higher affinity for NR1/NR2C receptors. However, this disparity could potentially be explained by the potent des-glycinated metabolite of remacemide (ARL 12495) that appears to have some NR1/NR2C selectivity, although this was not statistically significant. Alternatively, improved therapeutic potential of some NMDA receptor channel blockers has also been attributed to their having a low affinity for NMDA receptors in general. As discussed elsewhere (Rogawski, 1993), low affinity channel blockers have faster on- and off-rate kinetics and thus may interfer less-with-normal synaptic transmission. Overall, given the compounds tested in this study, the present results are consistent with the latter suggestion that the higher affinity antagonists display a lower therapeutic index.
The subunit-specific variations in channel blocker pharmacology reported here are largely, but not entirely, comparable to the pharmacological variations reported for [3H]MK-801 binding to native NMDA receptors in differing brain regions. Relative to TCP and other channel blockers, dextromethorphan (Beaton et al., 1992; Ebert et al., 1991) and ARL 15853 (Porter and Greenamyre, 1995) display a significantly higher relative affinity in the cerebellum (lower ratio of cerebellar IC50 to forebrain IC50). Because the cerebellum contains NR1, NR2A and NR2C subunits and forebrain contains NR1, NR2A and NR2B subunits, these findings are consistent with NR2C-containing NMDA receptors having a relatively higher affinity for dextromethorphan and ARL 15853.
The major disparity between this study and the radioligand binding literature is the relative potency of MK-801. Several studies have shown that cerebellar NMDA receptors display 2–25-fold lower affinities for [3H]MK-801 (Beaton et al., 1992; Bresink et al., 1995; Ebert et al., 1991;Porter and Greenamyre, 1995; Quarum et al., 1990; Reynolds and Palmer, 1991; Yoneda and Ogita, 1991), whereas in the present study, MK-801 was equipotent between all receptor subunit combinations. Likewise, radioligand binding studies of recombinant receptors revealed a 40–50-fold lower affinity for [3H]MK-801 at NR1a/NR2C than at NR1a/NR2A receptor complexes (Chazot et al., 1994;Laurie and Seeburg, 1994). TCP is also sometimes reported to have a lower affinity at cerebellar NMDA receptors (Ebert et al., 1991; Quarum et al., 1990), whereas other studies indicate similar affinities between forebrain and cerebellum (Beaton et al., 1992; Porter and Greenamyre, 1995). In one physiological study, the structurally related compound PCP appeared to display higher affinity for cerebellar than for forebrain NMDA receptors (Yi et al., 1988). The apparent difference between radioligand binding studies and the present physiological study of recombinant receptors could be due to the differing resting membrane potentials in the respective studies (0 mV vs. −60 mV) or to the use of nonequilibrium conditions in some radioligand binding studies, since equilibrium conditions are frequently not established for each receptor subtype being examined.
Although TCP and MK-801 displayed identical affinities at the different NMDA receptors, they displayed slower association and dissociation rates at NR1/NR2C receptors than the others tested. These findings are consistent with the results of Yamakura et al. (1993), who demonstrated that when agonists and MK-801 were coapplied in pulses onto recombinant NMDA receptors, NR1/NR2A and NR1/NR2B receptor responses were more quickly blocked by MK-801 than were NR1/NR2C receptor responses. For NR1 homomeric receptors, Rodriguez-Paz et al. (1995) reported a faster blockade rate for NMDA receptors containing NR1 subunits that have the first alternatively spliced cassette (termed insert 1, Sugihara et al., 1992), which may account for observations of lowered MK-801 affinity at NR1 homomers missing insert 1 (Hollmann et al., 1993; Rodriguez-Pazet al., 1995) if the drug application times were not sufficient to achieve steady state.
The mechanism responsible for the slower blockade kinetics of TCP and MK-801 at NR1/NR2C receptors is unknown. It is possible that drug ability to enter and leave the channel is more restricted for the NR1/NR2C receptor. This would be consistent with single-channel studies that show that NR2C-containing receptors display briefer channel open times (Stern et al., 1992). Thus, all channel blockers may display slower kinetics at the NR1/NR2C receptor. We were not able to evaluate this in the present study because only two compounds (TCP and MK-801) displayed kinetics that were sufficiently slow to resolve from the kinetics of drug application.
Overall, these studies indicate that the NR2 subunit contributes to the heterogeneity of channel blocker specificity for NMDA receptors. Thus, the heterogeneous distribution of NR2 subunits can account for the pharmacological heterogeneity of channel blockers for native NMDA receptors in differing brain regions (Beaton et al., 1992;Ebert et al., 1991). In a parallel manner, NR2 subunits appear to be predominantly responsible for generating pharmacological heterogeneity at the glutamate recognition site on the NMDA receptor complex (Buller et al., 1994, 1997; Laurie and Seeburg, 1994).
Acknowledgments
The authors are grateful to Astra Arcus for kindly providing several of the compounds used in this study and to Dr. Nakanishi and Dr. Seeburg for providing NMDA receptor cDNAs.
Footnotes
-
Send reprint requests to: Dr. Daniel T. Monaghan, Department of Pharmacology, University of Nebraska Medical Center, 600 S. 42nd Street, Omaha, NE 68198-6260. E-mail:dtmonagh{at}unmc.edu
-
↵1 This work was supported by United States Army Medical Research contract DAMD17–94-C-4050 and National Institutes of Health grant NS28966.
-
↵2 Eric Harris, Astra, personal communication.
-
↵3 Gene Palmer, Astra, personal communication.
-
↵4 Gene Palmer, Astra, unpublished observations.
- Abbreviations:
- NMDA
- N-methyl-d-aspartate
- HEPES
- 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid
- Received June 7, 1996.
- Accepted October 21, 1996.
- The American Society for Pharmacology and Experimental Therapeutics