Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Since the first topochemical demonstration of "free" (chelatable) Zn²⁺ in sharply delineated regions of the mammalian hippocampal formation (Maske, 1955), it has been firmly established that zinc is contained in synaptic vesicles of several neuronal pathways. Zinc has anticonvulsant and neuroprotective properties, but a better understanding of the molecular mechanisms of action of zinc appears to be necessary to take selective advantage from this knowledge. Thus, the beneficial effects of zinc in a number of seizure models are contrasted by its possible involvement in processes leading to neurodegeneration (for a review, see Choi and Koh, 1998). Notwithstanding the possible detrimental role of zinc in cerebral ischemia (Koh et al., 1996), it is in a well established animal model of cerebral ischemia in which zinc proved to be beneficial and neuroprotective (Matsushita et al., 1996).

At micromolar concentrations, which might be attained in the synaptic cleft during neuronal activity, Zn²⁺ has pronounced effects on ligand- and voltage-gated ion channels (Harrison and Gibbons, 1994). Particular attention has been devoted to the inhibitory effect of Zn²⁺ at the N-methyl-D-aspartate (NMDA) receptor complex mediated by an allosteric regulatory site near the external face of the membrane (Peters et al., 1987; Westbrook and Mayer, 1987). Specific binding of the open NMDA channel blocker [³H]MK-801 to rat neuronal membranes (Wong et al., 1988; Yoneda and Ogita, 1989) is inhibited noncompetitively by micromolar concentrations of Zn²⁺ (Greenberg and Marks, 1988; Reynolds and Miller, 1988). Increasing the concentrations of the coagonists glutamate and glycine has only marginal effects on the inhibition of [³H]MK-801 binding by Zn²⁺ (Reynolds and Miller, 1988), in agreement with the observation that the electrical responses of mouse cultured hippocampal neurons can be blocked by Zn²⁺ independently of the concentrations of NMDA and glycine used to stimulate the cells (Mayer et al., 1989). On the other hand, addition of the polyamine spermidine, which is supposed to increase the opening frequency of the NMDA channel via a separate polyamine-sensitive mechanism (Ransom and Stec, 1988; Williams et al., 1990; Rock and Macdonald, 1991; Benveniste and Mayer, 1993), greatly reduces the inhibitory effect of Zn²⁺ on [³H]MK-801 binding (Enomoto et al., 1992; Reynolds, 1992). The hypothesis that Zn²⁺ interacts as negative modulator with the same site at the NMDA receptor complex as the positive modulators spermine and spermidine was, however, rejected: the IC₅₀ value of Zn²⁺ was not increased to the extent predicted for competitive interaction by increasing the concentration of the agonist spermidine.

Here, we reinvestigate the possibility that Zn²⁺ inhibits the NMDA receptor complex via a polyamine-sensitive regulatory site. We compared Zn²⁺ as an inhibitor of the NMDA receptor complex with three other compounds exhibiting polyamine-sensitive inhibition of the NMDA receptor complex: 1,12-dodecanediamine (N-12-N) (Berger et al., 1992), pentamidine (Reynolds and Aizenman, 1992), and ifenprodil (Carter et al., 1990).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Membrane Preparation. Triton-treated membranes were prepared from the hippocampal cornu ammonis 1 and dentate gyrus part (CA1/DG part, the region with the highest density of NMDA receptors) of adult male Wistar rats and stored at 80°C as described previously (Berger et al., 1992). For some experiments, membranes were prepared from the CA3 part of the hippocampus, the piriform cortex and the amygdala, dissected on a cold plate from the unfrozen brain as described previously (Berger et al., 1986), and from the parietal cortex, striatum, bulbus olfactorius, gyrus cinguli, and superior colliculi (also dissected from the unfrozen brain). No EDTA was included into the homogenization medium because in experiments performed in parallel, similar results were obtained with EDTA-treated and untreated membranes.

[³H]MK-801 Binding Assay. Binding assays were performed in polypropylene vials (duplicates or triplicates) in 1.0 ml of 50 mM Tris acetate, pH 7.0, at 24°C. [³H]MK-801 (5 nM, 23.9 Ci/mM; New England Nuclear Research Products, Boston, MA) was incubated for 2 h with glutamic acid (1 µM), glycine (1 µM), and various concentrations of spermine (1, 3, 10, 30, 100, and 300 µM; Serva, Heidelberg, Germany). Incubation times beyond 2 h did not result in any further increase in binding (this will occur only with buffer concentrations lower than 50 mM; unpublished observation). For nonspecific binding, glutamic acid and glycine were replaced by their respective antagonists, D-2-amino-5-phosphonovaleric acid (10 µM) and 5,7-dichlorokynurenic acid (1 µM; both from Tocris Cookson, Northpoint, UK). The incubation was started by adding membranes corresponding to approximately 1 mg fresh tissue and stopped by the addition of 3 ml (room temperature) 20 mM Tris-acetate, pH 7.0, and rapid filtration through Whatman (Hassel, Munich, Germany) GF/C filters presoaked for 1 h in polyethyleneimine (0.3% in H₂O), using a 48-place Brandel (Gaithersburg, MD) harvester. Filters were washed 3 times with 4 ml buffer (room temperature) and transferred into counting vials. After addition of 2.5 ml scintillation standard cocktail (Rotiscint 11; Roth, Karlsruhe, Germany), vials were warmed to 40°C, agitated for 1 h, and counted in a -scintillation counter. Zinc acetate dihydrate (99.999%) and N-12-N were obtained from Aldrich- Chemie (Steinheim, Germany), pentamidine from Sigma Chemical Co. (St. Louis, MO). Ifenprodil was obtained from Tocris and also as a gift from Synthélabo Recherche (Bagneux, France).

Separation of High- and Low-Affinity Components of Inhibition. Inhibition of [³H]MK-801 binding by Zn²⁺ consisted of two components, with IC₅₀ values sufficiently different from each other (3-7 µM and 2.0 mM, see below), to allow nearly complete resolution of the two components. In several experiments, the high-affinity component was masked by 100 µM; in some others, it was masked by 300 µM Zn²⁺. It can be calculated that under these conditions, only 0.9% to 2.7% of the high-affinity component remained unmasked (0.2-0.6% in presence of 300 µM Zn²⁺; n_H = 1.35), whereas 95% of the low-affinity component was unaffected (87% in presence of 300 µM Zn²⁺; n_H = 1.00). Also, inhibition of [³H]MK-801 binding by ifenprodil consisted of two components, with IC₅₀ values sufficiently different from each other (143-231 nM and 362 µM, see below). In several experiments, the high-affinity component was masked by 10 µM; in some others, it was masked by 30 µM ifenprodil. As can be calculated, under these conditions, only 1.7% to 2.7% of the high-affinity component remained unmasked (0.6-1.0% in the presence of 30 µM ifenprodil; n_H = 0.95), whereas 97% of the low-affinity component was unaffected (92% in the presence of 30 µM ifenprodil; n_H = 1.00).

Data Analysis. Monophasic and biphasic inhibition curves were subjected to computerized curve fitting (Johnson and Faunt, 1992). For the evaluation of biphasic inhibition curves, the Hill coefficient (n_H) for the low-affinity component was fixed to 1.0. In special situations (e.g., in the presence of 30 µM spermine), n_H of the high-affinity component also had to be fixed to 1.0 to avoid inconsistent results. ANOVA was applied to identify significant differences in the components of specific [³H]MK-801 binding in several brain regions (F test; posthoc Newman-Keuls test). Computerized curve fitting was used for the determination of EC₅₀ values of spermine stimulation of [³H]MK-801 binding, allowing for n_H  1.0 (usually 1.1 < n_H < 1.5). For Schild plot analysis, IC₅₀/K_i  1 was plotted double logarithmically against [spermine]/EC₅₀ (K_i = IC₅₀ value in the absence of added spermine). To study the influence of ifenprodil on the IC₅₀ value of Zn²⁺, and vice versa, the influence of Zn²⁺ on the IC₅₀ value of ifenprodil, the IC₅₀ values were determined with and without the influencing agent within the same incubation and filtration procedure. The results of four separate experiments were evaluated by Student's paired t test (3 df).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Biphasic Inhibition of [³H]MK-801 Binding by Zn²⁺. Low concentrations of Zn²⁺ displaced 65 ± 5% of specifically bound [³H]MK-801 from membranes prepared from the CA1/DG part of the rat hippocampus (Fig. 1A, Table 1); the remainder was inhibited by millimolar Zn²⁺. Increasing the concentrations of glutamic acid and glycine from 1 to 10 µM did not change the IC₅₀ value of Zn²⁺ (high-affinity component: 11.9, 6.6 µM with 1 µM; 11.8, 6.5 µM with 10 µM glutamic acid and glycine, n = 2). Spermine shifted the high-affinity component to higher IC₅₀ values but not the low-affinity component (Fig. 1A and Table 1). In the absence and in the presence of spermine, the inhibition curves were steep (see n_H > 1 in Table 1). Spermine (10 µM) (i.e., 2.93 times its EC₅₀ for stimulation of [³H]MK-801 binding in these experiments) shifted the high-affinity IC₅₀ value by a factor of 5.0 ± 3.1 (range, 2.4-10.4) (i.e., by a factor compatible with competitive interaction between Zn²⁺ and spermine). In Fig. 2, this relationship (in the form of a Schild plot analysis) is compared with results obtained with two compounds inhibiting the NMDA receptor complex via a polyamine-sensitive mechanism: N-12-N (Berger et al., 1992) and pentamidine (Reynolds and Aizenman, 1992). Only the results obtained with Zn²⁺ scatter around a correlation line with unity slope. Linear correlation analysis resulted in the following slopes (±S.D.): 1.05 ± 0.11 (for Zn²⁺), 0.89 ± 0.03 (for N-12-N), and 0.62 ± 0.04 (for pentamidine), which are significantly different from each other (p < .001, ANOVA). All IC₅₀, EC₅₀, and K_i values have been obtained by computer analysis of several independent experiments. In the case of Zn²⁺, computer analysis had to operate on a greater number of parameters than in the case of N-12-N and pentamidine due to the existence of a high- and a low-affinity component; this might explain the relatively high extent of scattering in the Zn²⁺ data.

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Inhibition of specific [3H]MK-801 binding to membranes prepared from the CA1/DG part of rat hippocampus (1 µM glutamic acid and glycine) by Zn2+ (A) and by ifenprodil (B). Mean values of normalized data were pooled from six independent experiments; 100% is 40.8 ± 2.6 fmol/mg tissue in A and 38.7 ± 4.8 fmol/mg tissue in B (mean ± S.D.). Influence is shown of increasing concentrations of spermine (1, 3, 10, and 30 µM, ).

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Spermine-stimulated [3H]MK-801 binding showing Schild plot analysis of inhibition by Zn2+ (, high-affinity component only), by 1,12-dodecanediamine (N-12-N, ), and by pentamidine (crosses). The relationship indicated by the dotted diagonal is predicted for simple competitive interaction between the inhibitor and spermine.

Biphasic Inhibition of [³H]MK-801 Binding by Ifenprodil. Ifenprodil displaced only 20.7 ± 7.0% of specifically bound [³H]MK-801 (CA1/DG membranes) with high affinity (Fig. 1B, Table 1); the remainder was inhibited by high micromolar concentrations. The addition of spermine shifted the high-affinity component to higher IC₅₀ values but not the low-affinity component (Fig. 1B, Table 1). The Hill coefficients n_H did not deviate significantly from unity, neither without nor with 10 µM spermine (Table 1). The shift of the high-affinity IC₅₀ value by spermine was compatible with competitive interaction: 10 µM spermine (i.e., 3.62 times its EC₅₀ value for stimulation of [³H]MK-801 binding in these experiments) shifted the IC₅₀ value by a factor 6.2 ± 2.3 (range, 4.1-11.2). Schild plot analysis of the dependence of the IC₅₀ value on a more extended range of spermine concentrations yielded data with an even higher degree of scattering than observed with zinc (not shown). Obviously, it is more difficult to obtain accurate data on a relatively small high-affinity component (as in the case of inhibition by ifenprodil) than on a high-affinity component representing the main effect of the inhibitor (as in the case of zinc).

Additivity of Inhibition by Zn²⁺ and by Ifenprodil. Figure 3 illustrates that in the CA1/DG part of the hippocampus, inhibitions of [³H]MK-801 by Zn²⁺ and by ifenprodil were additive. Low concentrations of ifenprodil (up to 10 µM) inhibited the same fraction, in the absence (Fig. 3A, shaded) and in the presence (Fig. 3B, shaded area) of 100 µM Zn²⁺. Similarly, the fraction inhibited by low Zn²⁺ concentrations (up to 100 µM) did not change after the addition of 10 µM ifenprodil (arrows in Fig. 3).

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Additivity of the inhibition of [3H]MK-801 binding by Zn2+ and by ifenprodil in membranes from hippocampal CA1/DG part. Data are representative for two independent experiments performed in triplicate (fmol/mg tissue); bars indicate S.D. Displacement curves by Zn2+ yielded the same high-affinity component under control conditions (arrow in B) as in the presence of 10 µM ifenprodil (arrow in A). In separate experiments, displacement curves obtained with various concentrations of ifenprodil revealed the same high-affinity component under control conditions (shaded in A) as in the presence of 100 µM Zn2+ (shaded in B).

Inhibition of [³H]MK-801 Binding by Zn²⁺ and by Ifenprodil in Other Brain Regions. In all brain regions analyzed, inhibition of [³H]MK-801 binding by Zn²⁺ and by ifenprodil was biphasic. The IC₅₀ value for the high-affinity components did not vary between the regions by more than a factor of 2 (for neither inhibition by Zn²⁺ nor inhibition by ifenprodil; Table 2). Also, the corresponding n_H values were similar in all regions. However, the regions differed from each other in the extent to which [³H]MK-801 binding was sensitive to low concentrations of either Zn²⁺ or ifenprodil. For example, from piriform cortex membranes, only 40.1% of specifically bound [³H]MK-801 was displaced by 100 µM Zn²⁺ (Table 3, column B), but 74.1% was displaced from gyrus cinguli membranes. Also, the sensitivity to 10 µM ifenprodil varied among the regions, from 20.6% (CA3) to 39.7% (bulbus olfactorius; Table 3, column E).

Apparent Deviation from Additivity in Many Brain Regions. In contrast to the results obtained with membranes prepared from the CA1/DG part of the hippocampus, inhibition by Zn²⁺ and by ifenprodil was apparently nonadditive in several other brain regions. For example, in the piriform cortex, 40.1% of specific [³H]MK-801 binding was sensitive to 100 µM Zn²⁺ [i.e., 11.9 fmol of 29.7 fmol specifically bound (mean value of three experiments, Table 3, column B; Fig. 4B, arrow)]. However, in the presence of 10 µM ifenprodil, Zn²⁺ displaced with high affinity 16.9 fmol specifically bound [³H]MK-801 (56.8%, Table 3, column F; Fig. 4A, arrow); this is significantly more than 11.9 fmol (P < .001). Micromolar concentrations of Zn²⁺ had a similar effect on the ifenprodil sensitivity of [³H]MK-801 binding to piriform cortex membranes. Without Zn²⁺, only 20.7% (i.e., 6.15 fmol) could be inhibited by 10 µM ifenprodil (Table 3, column E; Fig. 4A, shaded), but in the presence of Zn²⁺, this fraction amounted to 35.9% (column C; i.e., 10.7 fmol; Fig. 4B, shaded), significantly more than 6.15 fmol (P < 001). Thus, in the piriform cortex, ifenprodil increased the fraction of bound [³H]MK-801 sensitive to low Zn²⁺, and Zn²⁺ increased the fraction of bound [³H]MK-801 sensitive to low ifenprodil. In membranes prepared from the amygdala (no significance) and from the hippocampal CA3 part (weak significance), a tendency into the same direction could be observed (Table 3), but membranes prepared from several other regions demonstrated opposite relationships. In the gyrus cinguli, one of the most extreme examples (one of three experiments is illustrated in Fig. 4, C and D), the fraction of [³H]MK-801 binding sensitive to low Zn²⁺ was significantly reduced by ifenprodil, from 74.1% (i.e., 13.9 fmol, arrow in Fig. 4D) to 58.6% (i.e., 11.0 fmol, P < .001; Table 3; Fig. 4C, arrow), and the fraction sensitive to low ifenprodil was reduced by Zn²⁺ from 35.3% (i.e., 6.64 fmol; shaded in Fig. 4A) to 18.0% (i.e., 3.38 fmol, P < .001, Table 3; Fig. 4D, shaded).

View larger version (29K): [in this window] [in a new window]   Fig. 4.   Inhibition of [3H]MK-801 binding by Zn2+ (A and C) and by ifenprodil (B and D). Absolute values (fmol [3H]MK-801 totally bound/mg tissue) are shown; the bottom x-axis is at the level of the nonspecific binding, and the top x-axis at the control level, representative of three independent experiments (mean results are given in Table 3). Positive interaction between the two inhibitors in rat piriform cortex membranes (A and B). In the presence of 10 µM ifenprodil, a greater amount of radioligand is displaced by Zn2+ with high affinity (arrow in A) than in the absence of ifenprodil (arrow in B); vice versa, in the presence of 100 µM Zn2+, a greater amount of radioligand is displaced by ifenprodil with high affinity (shaded in B) than in the absence of Zn2+ (shaded in A). No interaction could be observed in membranes prepared from the CA1/DG part of the hippocampus (see Fig. 3), whereas negative interaction was seen in cingulate cortex membranes (C and D).

Mutual Elimination of Spermine Sensitivity. In four experiments, the sensitivity of the inhibition of [³H]MK-801 binding by Zn²⁺ to spermine (i.e., the factor, by which the IC₅₀ value of the high-affinity component was increased by the addition of 10 µM spermine) was determined simultaneously in the absence and in the presence of 30 µM ifenprodil. In the absence of ifenprodil, the addition of 10 µM spermine resulted in a 4-fold shift of the high-affinity IC₅₀ value of Zn²⁺ (in agreement with data given in Table 1). Ifenprodil (30 µM) not only reduced the stimulatory effect of spermine but also almost eliminated the spermine-induced shift in the IC₅₀ value of Zn²⁺ (to 1.49-fold; Table 4 and Fig. 5A). In four other experiments, the sensitivity of the inhibition of [³H]MK-801 binding by ifenprodil to spermine was determined simultaneously in the absence and in the presence of 300 µM Zn²⁺. In the absence of Zn²⁺, the addition of 10 µM spermine resulted in a 5-fold shift of the high-affinity IC₅₀ value (again in agreement with data given in Table 1). This shift was eliminated (to 0.96-fold; Table 4 and Fig. 5B) by 300 µM Zn²⁺.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Inhibition of [3H]MK-801 binding (membranes from hippocampal CA1/DG part) by Zn2+ in the presence of 30 µM ifenprodil (A) and by ifenprodil in the presence of 300 µM Zn2+ (B). Results are representative for four independent experiments. Note that the addition of 10 µM spermine did not result in a shift of the inhibition curves (in contrast to Fig. 1).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The main results of this study are that 1) the dose-response curves for the inhibition of [³H]MK-801 binding by Zn²⁺ and ifenprodil consisted of a low-affinity and a high-affinity component, respectively; 2) the respective high-affinity components were roughly additive and were shifted to the right by the addition of spermine; and 3) the spermine reversal of Zn²⁺ inhibition was prevented by ifenprodil and, vice versa, the spermine reversal of ifenprodil inhibition was prevented by Zn²⁺.

Possible Complex Formation between Zn²⁺ and Spermine. The interpretation of results obtained with metal ions like Zn²⁺ and Cu²⁺ is complicated by the propensity of these ions to form tight complexes with several organic molecules, including glutamic acid, glycine, and spermine (Smith and Martell, 1975; Prince, 1987). The free concentration of Cu²⁺, which forms stronger complexes than Zn²⁺, is sensitive to the presence of amino acids (Vlachová et al., 1996). The effects of Zn²⁺ at the NMDA receptor complex, however, seem to be largely independent of the concentration of amino acids (Westbrook and Mayer, 1987; Mayer et al., 1989; Vlachová et al., 1996; and our own observations). Furthermore, 30 µM ifenprodil abolished the influence of spermine on the inhibitory potency of Zn²⁺, making it unlikely that our results may be explained simply by spermine forming an inactive complex with Zn²⁺ (although the formation of this complex cannot be excluded, see 933 ff in Prince, 1987, and 101 ff in Smith and Martell, 1975).

Biphasic Inhibition. Several reports have described monophasic inhibition of [³H]MK-801 binding by Zn²⁺ (Greenberg and Marks, 1988; Reynolds and Miller, 1988; Reynolds, 1992), in contrast to our results. A reason for the discrepancy may be that under the conditions of low ionic strength and slightly alkaline pH (as used in these studies), the two components of Zn²⁺ inhibition are practically indistinguishable (M. L. Berger and P. Rebernik, unpublished observation). Electrophysiological experiments leave no doubt that the inhibition of the NMDA receptor complex by Zn²⁺ involves at least two separate mechanisms: low micromolar (Christine and Choi, 1990; Legendre and Westbrook, 1990), or even lower (Chen et al., 1997; Paoletti et al., 1997), concentrations of Zn²⁺ act at the outer surface of the membrane; a second mechanism mediates direct inhibition of the channel at higher concentrations. Thus, our detection of two components of inhibition by Zn²⁺ also with biochemical techniques is not unexpected.

Additivity of Independent Components? In membranes prepared from the CA1/DG part of the rat hippocampus, the inhibition of [³H]MK-801 binding by Zn²⁺ and by ifenprodil was additive (i.e., each of the two substances inhibited its own fraction of bound [³H]MK-801, apparently independent of the presence of the other substance). In both cases, the inhibition was reversed by the addition of spermine to the extent predicted for competitive interaction; therefore, we adopted the hypothesis that both substances inhibited the NMDA receptor complex via independent polyamine regulatory sites, with the first site sensitive to stimulation by polyamines like spermine and spermidine and at the same time sensitive to inhibition by Zn²⁺, and the second site also sensitive to stimulation by polyamines and not sensitive to inhibition by Zn²⁺ but sensitive to inhibition by ifenprodil. However, more detailed investigations revealed that this additivity was limited to certain brain regions such as the hippocampal CA1/dentate gyrus and the parietal cortex. In membranes prepared from several other brain regions, Zn²⁺ and ifenprodil mutually influenced the extent to which they inhibited the NMDA receptor complex with high affinity; for example, specific [³H]MK-801 binding to membranes prepared from the gyrus cinguli could be reduced (see Table 3) by 74.1% by 100 µM Zn²⁺ and by 35.3% by 10 µM ifenprodil. Nevertheless, 6.0% to 7.7% proved insensitive to either. These fractions sum up to 100% if mutual influences are taken into consideration: in the gyrus cinguli, Zn²⁺ inhibited only 58.6% of specifically bound [³H]MK-801 under the influence of ifenprodil (instead of 74.1% in its absence). In contrast, pronounced positive interaction was seen in the piriform cortex. Taking together the results from nine brain regions, negative interaction appears to correlate with relatively pronounced high-affinity components of inhibition by zinc and by ifenprodil, whereas small high-affinity components for both inhibitors seem to favor positive interaction. This regional variability does not correlate with the regional distribution of any of the known NMDA receptor subunits. An exception might be the striatum and the olfactory bulb, where a higher level of NR2B expression has been found than in many other brain regions (Portera-Cailliau et al., 1996; Wenzel et al., 1997) and where a relatively high fraction of [³H]MK-801 binding was sensitive to inhibition by ifenprodil.

Competitive or Allosteric Interaction? Reversal of Zn²⁺ inhibition of [³H]MK-801 binding by the polyamine spermidine was first demonstrated by Reynolds (1992), who, while considering a direct competitive interaction between Zn²⁺ and spermidine unlikely, did not take into account components of high- and low-affinity Zn²⁺ inhibition. Direct competitive interaction between stimulatory polyamines and inhibitory ifenprodil has also been questioned (Reynolds and Miller, 1989; Ogita et al., 1992), but here, too, the investigations did not distinguish between high- and low-affinity components in the inhibitory action of ifenprodil. Our data characterizing the interaction between zinc (high-affinity component) and spermine, although compatible with a competitive mechanism, exhibited a high degree of scattering (Fig. 2) and represent no direct proof for such a mechanism; analogous data for ifenprodil (not shown) exhibited an even higher degree of variability. One reason for the difficulty in obtaining accurate data on the reversibility of the high-affinity inhibitions produced by zinc and by ifenprodil over a more extended concentration range could be the direct channel blockade at spermine concentrations slightly above stimulating concentrations (Rock and Macdonald, 1991; Benveniste and Mayer, 1993; and our own observations). On the other hand, our observation that spermine reversibility of Zn²⁺ inhibition was lost in the presence of ifenprodil (and vice versa, that spermine reversibility of ifenprodil inhibition was lost in the presence of Zn²⁺) strongly argues against a competitive and in favor of an allosteric mechanism of action. This hypothesis is in agreement with molecular biology studies, indicating that polyamine stimulation is primarily mediated by NR1 receptor subunits (Williams et al., 1995; Kashiwagi et al., 1996), whereas high-affinity inhibition by zinc depends on the presence of the NR2A (Chen et al., 1997; Paoletti et al., 1997) and high-affinity inhibition by ifenprodil on the presence of the NR2B subunit (Williams, 1993; Gallagher et al., 1996). Furthermore, it has been demonstrated with site-directed mutagenesis that other amino acid residues of the NR2B subunit are involved in polyamine stimulation than those responsible for inhibition by ifenprodil (Gallagher et al., 1996).

Comparison with Other Polyamine-Sensitive Inhibitors. Other compounds postulated to inhibit the NMDA receptor complex via a polyamine regulatory site, such as N-12-N and pentamidine, exhibit monophasic inhibition of [³H]MK-801 binding; the concentrations of polyamines needed to overcome the inhibitions are higher than those needed to overcome high-affinity inhibition by zinc or by ifenprodil. In contrast to Zn²⁺ and ifenprodil, these compounds have two positive charges separated from each other by some distance essential for their potency (Romano et al., 1992). It may be speculated that these compounds interact with the NMDA receptor complex via two separate sites, with the first site corresponding to the interaction of Zn²⁺ and the second corresponding to the interaction of ifenprodil with the NMDA receptor complex. In this case, the zinc site could only be occupied with concomitant occupation of the ifenprodil site, and vice versa, the ifenprodil site could only be occupied with concomitant occupation of the zinc site. Because occupation of one of these sites compromises the spermine reversibility of inhibition via the other site (as shown in this report; see above), concomitant occupation of both sites by "bidentate" compounds would provide an explanation for the reduced spermine sensitivity of this type of inhibition in comparison to inhibition by Zn²⁺ or by ifenprodil alone (with only one of the two sites inhibited).