Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Oral levodopa replacement therapy remains the single most effective medication for the symptomatic relief of Parkinson's disease (PD). However, predictable and unpredictable fluctuations in clinical status and abnormal involuntary movements (dyskinesias) eventually develop in most patients following chronic levodopa treatment (Marsden, 1990). Recent pharmacological evidence suggests that levodopa-related motor response complications in experimental parkinsonism in rats are due, at least in part, to hyperfunctioning of certain central glutamatergic pathways (Engber et al., 1994; Oh et al., 1997, 1998). In support of this, a variety of N-methyl-D-aspartate (NMDA) receptor antagonists have been shown to be effective in animal models of PD. For example, in 1-methyl-4-phenyl-tetrahydropyridine (MPTP)-lesioned parkinsonian monkeys treated with levodopa, a competitive NMDA receptor glutamate-site antagonist (LY 235959) significantly attenuated dyskinesias while sparing the motor benefit derived from levodopa (Papa and Chase, 1996). The therapeutic index of this novel agent was far superior than that of the potent NMDA channel blocker MK-801 (Crossman et al., 1989; Close et al., 1990; Rupniak et al., 1992; Domino and Sheng, 1993) and appreciably better than the weak channel blocker amantadine (Blanchet et al., 1998).

Mammalian NMDA receptors are ligand-gated ion channels composed of di- or triheterooligomeric assemblies of NR1 subunits and NR2 subunits (Moriyoshi et al., 1991; Hollmann and Heinemann, 1994). Individual NR1 isoforms and NR2 subunits have distinct anatomical and developmental patterns of expression (Sheng et al., 1994). In addition, different subunit combinations (or "subtypes") have distinct biophysical and pharmacological characteristics (Williams, 1993; Hollmann and Heinemann, 1994; Priestly et al., 1995; Woodward et al., 1995). Overall, the molecular biology studies give strong support to the idea that different NMDA subtypes mediate different aspects of brain function, and that the subtype-selectivity profile of NMDA antagonists will affect their therapeutic potential and side effect profile.

The key pharmacological properties of NMDA receptor antagonists important for the treatment of PD remain poorly understood. Having an optimum pattern of subtype selectivity is only the beginning. Other variables are certainly important, for example: 1) site/mechanism of antagonist action (e.g., channel blockers, competitive antagonists, allosteric inhibitors etc.), 2) affinity/binding kinetics, and 3) efficacy of inhibition (i.e., full versus partial). Comparison of activity in vivo with mechanism and subtype selectivity in vitro should yield important insights into structure-function relationships in NMDA responses as well as improved therapies for PD. In view of the antidyskinetic effect reported earlier with the competitive glutamate-site antagonist LY235959 (Papa and Chase, 1996), we evaluated another competitive glutamate-site antagonist (R)-4-oxo-5-phosphononorvaline (MDL 100,453) and a novel allosteric NMDA antagonist 4-hydroxy-N-[2-(4-hydroxyphenoxy)ethyl]-4-(4-methylbenzyl)piperidine (Co 101244/PD 174494) for antiparkinsonian and antidyskinetic efficacy in monkeys. We also assayed these compounds, together with LY 235959, for inhibitor potency and subtype selectivity at four cloned binary NMDA subtypes expressed in Xenopus oocytes (NR1A or NR1E in combination with NR2A, NR2B, NR2C, or NR2D). The results indicate that different NMDA receptor antagonists can have opposite effects on motor function and that subtype selectivity may play a role in determining these differences.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Animal Study Design. Six cynomolgus (Macaca fascicularis) monkeys of both sexes (2 females and 4 males) weighing 3.25 to 7.4 kg were studied under an approved protocol that met the ethical standards of the institutional Animal Care and Use Committee. Subjects were housed individually, fed with a standard biscuit diet twice a day supplemented with fruits, and had free access to water. They were kept under stable room conditions and maintained under a 12-h light/dark cycle. All monkeys were exposed to MPTP hydrochloride (Research Biochemicals Intl., Natick, MA) administered s.c. at a weekly dose of 0.5 to 1 mg/kg until sustained parkinsonian features with action tremor appeared. The average cumulative MPTP dose was 4.4 mg/kg (range, 2.1 to 9.75 mg/kg). All animals were left drug-free for 6 to 8 weeks and scored on a regular basis using the Laval University Disability Scale for MPTP Monkeys (Gomez-Mancilla et al., 1993), where the normal state extends from 0 to 2 points and maximal disability is 10 points. Once a mild (n = 5) to moderate (n = 1) parkinsonian syndrome had stabilized (baseline disability scores between 4-6 points), monkeys began chronic treatment with 0.5 to 2 tablets levodopa/carbidopa (Sinemet; 100 mg/25 mg tablets; Roane-Barker Inc., Greenville, SC) once daily hidden in food, with the dose depending only on individual intake and spillage. All monkeys developed dyskinesias of a predominantly choreic nature within 1 month, which were subsequently reproduced consistently and predictably following each oral dose of levodopa/carbidopa. At that point, they were put on a maintenance oral dose of levodopa/carbidopa administered 3 times a week to maintain priming events and avoid the confounding effect of variable levodopa withdrawal during testing with glutamate antagonists. On testing days, the morning meal and oral dose of levodopa/carbidopa were withheld.

Electrophysiology. cDNA clones encoding the rat NMDA receptor subunits NR1A, NR2A, NR2B, NR2C, and NR2D were provided by Dr. P. H. Seeburg (Heidelberg University, Heidelberg, Germany; Monyer et al., 1992). The NR1E cDNA was provided by Dr. S. Nakanishi (Kyoto University, Kyoto, Japan). Preparation and maintenance of Xenopus laevis oocytes and microinjection with cRNA were all as previously described (Woodward et al., 1995; Whittemore et al., 1997). Membrane current responses were recorded at room temperature (18-22°C) with a conventional two electrode voltage clamp (Dagan TEV-200, Minneapolis, MN) in a nominally Ca²⁺-free Ringer in which Ca²⁺ was replaced by equimolar Ba²⁺ (Whittemore et al., 1997). Drugs were applied by bath perfusion (5-10 ml/min) in a conventional flow-through chamber (volume ~0.2 ml). Under these conditions, provided levels of expression were moderate, coapplication of glutamate and glycine elicited a monophasic NMDA response (Whittemore et al., 1997).

(1)

(2)

(3)

Drugs. Co 101244/PD 174494 (CoCensys, Inc., Irvine, CA/Parke-Davis, Ann Arbor, MI) was synthesized as described elsewhere (Zhou et al., 1998). MDL 100,453 and LY 235959 were kindly provided by Hoechst Marion Roussel Inc. (Cincinnati, OH) and Eli Lilly (Indianapolis, IN). Other chemicals were purchased from Sigma Chemical Co. (St. Louis, MO), or as noted in the text. For electrophysiological recordings, the test compounds were initially dissolved in dimethyl sulfoxide (DMSO) and diluted in a series of DMSO stocks over the range 0.003 to 30 mM. Ringer solutions (0.001-100 µM) were made by 300- to 3000-fold dilution of stocks, just before application to the oocyte (final %DMSO = 0.03-0.3%).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Antiparkinsonian Efficacy. Two different, well tolerated doses of each NMDA antagonist were selected from a wide range of doses tested alone up to toxic levels. Doses of MDL 100,453 (50 and 300 mg/kg s.c.) alone dose dependently increased PAM counts, and two animals showed an apparent antiparkinsonian motor response on the rating scale as well with the high dose (Table 1). In contrast, doses of Co 101244 (0.1 and 1 mg/kg s.c.) alone produced no behavioral effect in five animals, and only one animal displayed motor effects with both doses (Table 1).

Antidyskinetic Efficacy. The near-threshold dose of levodopa did not produce dyskinesias by itself. However, all animals exhibited dyskinesias when that dose was combined with MDL 100,453 (300 mg/kg): the mean DSI with this combination (17 ± 5) approached that seen with the high dose of levodopa alone (Fig. 1). By comparison, only one of three animals showing evidence for potentiation when the low dose of levodopa was combined with Co 101244 (1 mg/kg) also displayed dyskinesias.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Mean DSI following administration of two different doses of levodopa/benserazide alone or in combination with an NMDA glutamate-site antagonist. The index was derived from the formula: (sum of all dyskinesia scores/duration of antiparkinsonian effect determined by direct observation) × 100; *p < .05 versus respective Dopa dose + vehicle.

Inhibition of Cloned NMDA Receptors Expressed in Oocytes. The previously tested NMDA receptor glutamate-site antagonist LY 235959 inhibited agonist-evoked currents with IC₅₀ values between 0.25 and 0.6 µM, dependent on the subunit combination (Fig. 2 and Table 2). The calculated K_b values demonstrate that the apparent affinity of LY 235959 for different NMDA receptor subunit combinations varies by only 3-fold across the five subunit combinations tested. The competitive, glutamate-site antagonist MDL 100,453 inhibited currents with IC₅₀ values ranging between 3.6 and 36 µM depending on the subunit combination (Fig. 2 and Table 2). Converting IC₅₀ values for individual subunit combinations into K_b values demonstrated that, in terms of apparent affinity, MDL 100,453 had 5- to 10-fold selectivity for the NR1A/2A versus other subunit combinations (Fig. 2 and Table 2). In contrast, the allosteric site antagonist Co 101244 inhibited agonist-evoked currents at NR1A/2B and NR1E/2B with IC₅₀ values of 0.024 and 0.048 µM, respectively, but was essentially inactive at NR1A/2A, NR1A/2C, and NR1A/2D (Fig. 2 and Table 2). Thus, Co 101244 showed ~10,000-fold selectivity for the NR1A/2B subunit combination (Zhou et al., 1998).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Concentration-inhibition curves for MDL 100,453, Co 101244/PD 174494, and LY 235959 at cloned NMDA receptors expressed in oocytes. Data are plotted as mean ± S.E., expressed as a fraction of control responses to coapplication of the agonists glycine and glutamate. For the different subunits, the concentrations of agonists were: 10 mM glycine + 100 µM glutamate for NR1A/2A; 1 µM glycine + 100 µM glutamate for NR1A/2B, NR1E/2B, NR1A/2C, and NR1A/2D. Experiments were performed in zero-Ca2+ Ringers, with Ba2+ in place of Ca2+.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The two NMDA antagonists assayed in the present experiments, MDL 100,453 and Co 101244, had distinct, and to a large extent opposite behavioral effects in the MPTP-treated monkeys. The two compounds differ in terms of their mechanism of antagonism: MDL 100,453 is a competitive antagonist, whereas Co 101244 is an allosteric inhibitor. The electrophysiological measurements showed that the two compounds also have profoundly different subtype-selectivity profiles for inhibition of binary NMDA receptor subunit combinations expressed in oocytes. Most strikingly, MDL 100,453 showed ~5-fold greater potency for NR1A/NR2A (K_b = 0.6 µM) versus NR1A/2B (K_b = 3.2 µM), whereas Co 101244 potently inhibited the NR1A/2B (IC₅₀ = 0.0026 µM) and was essentially inactive at other NR2 subtype combinations (IC₅₀ > 200 µM). Combining the NR2B subunit with NR1E, an NR1 isoform that contains the C-terminal insert cassette (Zukin and Bennett, 1995), did not affect the potency of either compound. This is consistent with previous studies indicating that the potency of competitive and allosteric antagonists shows dependence on the NR2 subunit and not the NR1 isoform (Lynch and Gallagher, 1996; Gallagher et al., 1996; Fischer et al., 1997).

Given the distinct localization of NMDA receptor subtypes in mammalian brain (Monyer et al., 1994; Sheng et al., 1994; Zukin and Bennett, 1995), it is not surprising that MDL 100,453 and Co 101244 produced widely different behavioral effects. Less easy to explain are the behavioral differences seen between MDL 100,453 and the previously tested competitive antagonist LY 235959 (Papa and Chase, 1996). The present experiments indicate that LY 235959 is essentially nonselective across the four binary NMDA subunit combinations, with potencies varying only ~2-fold (K_b = 0.04-0.09 µM). Again, incorporation of NR1E did not affect potency, at least for NR2B-containing receptors. Thus, LY 235959 has a subtype profile and mechanism of inhibition closer to MDL 100,453 than Co 101244 and yet, in terms of behaviors, it more closely resembles Co 101244. Reasons for the apparent disparity between subtype profile and antiparkinsonian and antidyskinetic activity are uncertain. It is possible that the ~5-fold selectivity for NR2A seen with MDL 100,453 is sufficient in itself to have behavioral consequences. However, western blots generally suggest that diheteromeric NR1/2A receptors are not common in adult mammalian brain (Luo et al., 1997). Another possibility is that LY 235959 and MDL 100,453 have differential potencies for triheteromeric NR1/2A/2B receptors, which are considered to be a major brain subtype (Luo et al., 1997). Unfortunately, assaying drugs on triheteromeric receptors in oocytes is problematic because it is difficult to distinguish between populations of true triheteromeric subunit combinations and mixtures of tri- and diheteromeric combinations, all of which form functional receptors (Stocca and Vicini, 1998). When extrapolating from work with the clones to complex behavioral models, it is also important to appreciate the numerous other factors that could come into play in vivo. These include differences in binding kinetics between antagonists, competition with and modulation by endogenous agents, and interactions at other NMDA subtypes or at other classes of receptor/channel. Such differences are certainly worth investigating to compare the antiparkinsonian efficacy of nonselective NMDA antagonists in monotherapy, like amantadine and MDL 100,453, to NR2B-selective NMDA antagonists like CP-101,606 (Steece-Collier et al., 1995) and ifenprodil (Nash et al., 1997) that have shown some efficacy in preliminary studies conducted in primates. The foregoing results and others (Papa and Chase, 1996) on the efficacy of NMDA antagonists cast doubt on the superiority of NR2B-selective antagonists to improve mobility in monotherapy. Comparative studies between NR2A- and NR2B-selective NMDA antagonists would be worth pursuing.

The two antagonists also differed in their interaction with exogenous levodopa. MDL 100,453 consistently potentiated levodopa responses, as reported previously with other NMDA blockers (Löschmann et al., 1991; Wüllner et al., 1992; Greenamyre et al., 1994): the highest dose turned a near-threshold low dose of levodopa into an antiparkinsonian and dyskinetic dose, and increased the dyskinetic scores following a high dose of levodopa. In contrast, Co 101244 less consistently potentiated the motor response to levodopa and produced a good antidyskinetic effect similar to the competitive NMDA glutamate-site antagonist LY 235959 (Papa and Chase, 1996). Just how certain NMDA antagonists potentiate levodopa-related benefit whereas others attenuate dyskinesia remains unexplained, but because both Co 101244 and LY 235959 displayed higher affinity for the NR1A/NR2B combination than MDL 100,453, this in itself may well be a feature for attenuation of dyskinesias.

The differences in affinity and selectivity found between the two antagonists tested for subtypes of NMDA receptors are likely to be linked to different mechanisms of action and circuits in the basal ganglia. Although NMDA receptors are widely expressed in the brain (Monyer et al., 1992; Buller et al., 1994), previous work indicates that intrastriatal administration of the NMDA channel blocker MK-801 reverses response alterations recorded in 6-hydroxydopamine-lesioned rats following chronic levodopa treatment (Papa et al., 1995). Moreover, a similar effect following the intrastriatal injection of a protein kinase A inhibitor (Oh et al., 1997) and a tyrosine kinase inhibitor (Oh et al., 1998) has been reported in the same model, the latter drug also attenuating the enhanced tyrosine phosphorylation of striatal NMDA NR2A and NR2B subunits observed following chronic levodopa treatment (Oh et al., 1998). This clearly suggests a striatal site of action. Nonetheless, pharmacological blockade of extrastriatal glutamate pathways may also affect basal ganglia outflow and in doing so promote antiakinetic and prodyskinetic effects, as seen following local perfusion of kynurenic acid (broad spectrum glutamate antagonist) in the internal pallidal segment (GPi) in normal monkeys (Robertson et al., 1989). Thus, subthalamo-pallidal (GPi) mechanisms could conceivably play a greater role in the case of MDL 100,453 compared with Co 101244 to enhance dyskinesia. In contrast, Co 101244 could antagonize glutamate influences at the level of the external pallidal segment (GPe), where local application of kynurenic acid did not provoke dyskinesia (Robertson et al., 1989). Interestingly, injections of kynurenic acid in the ventral part of the GPe in awake monkeys provoked contralateral leg dystonia (Robertson et al., 1989) and high doses of the NMDA antagonist MK-801 (Rupniak et al., 1992) and LY 235959 (Papa and Chase, 1996) have increased dystonia severity. In one PD patient off levodopa, some doses of the noncompetitive NMDA receptor channel blocker dextrorphan apparently worsened foot dystonia, whereas it attenuated peak-dose chorea in combination with levodopa (Blanchet et al., 1996). Thus, direct and/or indirect modulation of neuronal activity in subregions of the GPe may partly explain the antidyskinetic activity of certain NMDA blockers.

The foregoing results provide insights for the development of NMDA antagonists with optimal pharmacological properties for treating the disabling and prevalent condition of levodopa-induced dyskinesia. Antagonists showing selectivity for receptors containing the NR2B subunit may be particularly efficacious in this respect. Whether the antidyskinetic effects of NMDA antagonists will be maintained with chronic treatment and whether these drugs can slow the onset of dyskinesias following early combination with levodopa remain important questions well worth pursuing.