Introduction

Summary Introduction Procedures Results Discussion References

A number of modulators and noncompetitive antagonists, including spermine and protons, affect the activation of NMDA receptors (1, 2), and there are complex interactions between some of these modulators (Fig. 1A). Spermine has several macroscopic effects on NMDA receptors, including stimulation (Fig. 1A, 1 and 2) and voltage-dependent block (Fig. 1A, 3). Protons inhibit NMDA receptors (Fig. 1A, 5), with a tonic inhibition of 50% at physiological pH (3-5). The glycine-independent form of spermine stimulation (Fig. 1A, 1), seen with saturating concentrations of glycine, may involve relief of proton inhibition (Fig. 1A, 6), processes that are also influenced by the alternatively spliced insert encoded by exon 5 in the NR1 subunit (Fig. 1A, 9). Voltage-dependent inhibition by spermine is due to an open-channel block, and spermine can also permeate NMDA channels (Fig. 1A, 11). Spermine is a weak NMDA channel blocker, being active at high micromolar concentrations, but some polyamine derivatives, such as N¹-DnsSpm, are potent channel blockers, being active at nanomolar to micromolar concentrations (1, 6). N¹-DnsSpm is a useful new tool with which to study the structure and properties of glutamate receptor channels (6).

View larger version (51K): [in this window] [in a new window]   Fig. 1.   Modulation and mutagenesis of NMDA receptors. A, Diagram of NMDA receptor showing effects of spermine, N1-DnsSpm, pH, and ifenprodil. Numbers, various macroscopic effects and interactions of these modulators. B, Relative positions of M1-M4 segments in NR1 and NR2 subunits. NR1 and NR2 subunits have different-sized amino- and carboxyl-terminal domains. sp, signal peptide. Inset, proposed topology of glutamate receptor subunits, with an extracellular amino-terminal domain, an intracellular carboxyl-terminal domain, three membrane-spanning domains (M1, M3, and M4), and a reentrant loop (M2) that contributes to the ion-permeation pathway (10, 42, 43). Below schematic, amino acid sequences of NR1A (residues 561-650), NR2A (residues 556-648), and NR2B (residues 557-649) between the beginning of M1 and the end of M3. Vertical lines, amino acids are numbered at intervals of 10 residues. Horizontal lines above each sequence, positions of M1-M3. Gaps, introduced at three positions in NR1A to maximize homology with NR2A and NR2B. Dots above amino acid, positions at which mutations were studied; the particular mutations (W-to-L, D-to-N, E-to-Q, and so on) are shown below each amino acid. Residues are numbered from the initiator methionine in each subunit (20, 27).

NMDA receptors are hetero-oligomers composed of combinations of NR1 and NR2 subunits in as-yet undefined combinations and stoichiometries (7-9). The topology of NMDA receptor subunits may be similar to that proposed for -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and kainate receptor subunits and kainate binding proteins, with an extracellular amino-terminal domain, three membrane-spanning domains (M1, M3, and M4), and a reentrant loop (M2) that contributes to the ion channel pore (10, 11) (Fig. 1B). Mutations at asparagine residues, such as NR1A(N616Q), in the M2 regions of NR1 and NR2 subunits influence Ca²⁺ permeability and block by Mg²⁺, and these residues may form the narrowest constriction of the channel pore (Fig. 1A, 10) (12-17).

In this study, we initially set out to determine the role of residues in and around the M1-M3 regions of NMDA receptor subunits in voltage-dependent block by spermine and N¹-DnsSpm. The focus of the experiments was on mutations at aspartate, glutamate, tryptophan, and tyrosine residues because these amino acids are involved in binding of spermidine to PotD, a bacterial polyamine binding protein (18, 19). Because spermine has multiple effects on NMDA receptors, we also examined stimulation by spermine and found, surprisingly, that some mutations in the M2 region reduced spermine stimulation. These mutations also reduced inhibition by protons, an effect that may be largely responsible for the loss of spermine stimulation seen at pH 7.5. A number of mutations in the M1-M3 regions of NR1 and NR2B were found to affect block by spermine and N¹-DnsSpm. Residues near the extracellular side of the M1 and M3 regions, as well as residues in M2, form part of a binding site for N¹-DnsSpm, suggesting that at least part of the M1 and M3 segments contributes to the pore or vestibule of the ion channel.

	Experimental Procedures

Summary Introduction Procedures Results Discussion References

cDNA clones and site-directed mutagenesis. The wild-type NR1A and NR1B clones (20, 21) and the NR1A mutants E594Q, E595Q, E596Q, E597Q, E598Q, D599N, T602G, S604A, S605A, A606S, W608L, F609L, S610A, N616Q, and E621Q (15) were gifts from Dr. S. Nakanishi (Institute for Immunology, Kyoto University Faculty of Medicine, Kyoto, Japan). The NR1A(N616R) and NR1A(S617N) mutants (14) were gifts from Dr. R. J. Dingledine (Department of Pharmacology, Emory University School of Medicine, Atlanta, GA). The wild-type NR2A and NR2B clones (22) were gifts from Dr. P. H. Seeburg (Center for Molecular Biology, University of Heidelberg, Germany).

Expression in oocytes and voltage-clamp recording. The preparation of cRNAs and the preparation, injection, and maintenance of oocytes were carried out as previously described (28-30). Oocytes were injected with NR1 plus NR2 cRNAs in a ratio of 1:5 (0.2-4 ng of NR1 plus 1-20 ng of NR2). Macroscopic currents were recorded with a two-electrode voltage-clamp using a GeneClamp 500 amplifier (Axon Instruments, Foster City, CA) or an OC-725 amplifier (Warner Instruments, Hamden, CT) as previously described (28, 30). Oocytes were continuously superfused with a saline solution consisting of 96 mM NaCl, 2 mM KCl, 1.8 mM BaCl₂, and 10 mM HEPES, pH 7.5, and in most experiments, oocytes were injected with K⁺-1,2-bis(2-aminophenoxy)ethane-N,N,N,N-tetraacetic acid (100 nl of 40 mM, pH 7.0-7.4) on the day of recording (28). To study the pH sensitivity of NMDA receptors, glutamate was applied in buffer at a particular pH with a 20-30-sec wash at the same pH before and after application of glutamate. I-V relationships were measured by using linear voltage ramps from 150 to +30 mV over 6 sec as previously described (6).

Data analysis. Data analysis and curve fitting were carried out using Axograph (Axon Instruments) or SigmaPlot (Jandel Scientific, San Rafael, CA) on Macintosh computers. To obtain IC₅₀ and Hill slope (n_H) values of antagonists, concentration-inhibition curves were fit to eq. 1:

(1)

where I_glu is the response to glutamate and I_glu+ant is the response to glutamate measured in the presence of the antagonist. Concentration-response curves for agonists were fit to eq. 2 to obtain n_H and EC₅₀ values of agonists:

(2)

where I_glu is the agonist-induced current, and I_max is the maximum response. For analysis of the voltage dependence of block by N¹-DnsSpm, data were analyzed using the model of Woodhull (31) by fitting the data to eq. 3:

(3)

where I_glu is the control response to glutamate, I_glu+DS is the response to glutamate measured in the presence of N¹-DnsSpm,  is the fractional recovery from block at depolarized potentials, K_d(0) is the equilibrium dissociation constant of N¹-DnsSpm at a transmembrane potential of 0 mV, z is the charge of N¹-DnsSpm,  is the fraction of the membrane electric field sensed by the blocker at its binding site within that field, F is the Faraday constant, R is the gas constant, and T is the absolute temperature. The  function was included in eq. 3 because in some cells the glutamate response showed a small run-down or run-up over time, and the fractional recovery from block at depolarized potentials was often slightly different from 1.0. The inclusion of the  variable improves the fitting procedure, but block by N¹-DnsSpm (0.1-1 µM) does not show a voltage-independent component. For example, the values of  were 1.01 ± 0.02 at NR1A/NR2B receptors (21 oocytes), 1.00 ± 0.01 at NR1A(N616Q)/NR2B receptors (8 oocytes), and 1.02 ± 0.01 at NR1A/NR2B(W559L) receptors (21 oocytes). In most experiments, only data negative to the reversal potential were used to fit eq. 3.

Materials. Glutamate and glycine were purchased from Sigma Chemical (St. Louis, MO). Spermine tetrahydrochloride was purchased from Calbiochem (San Diego, CA). N¹-DnsSpm was provided by Drs. J. Renault and N. Seiler (University of Rennes, France). Ifenprodil was from Synthélabo Recherche (Bagneux, France). Other reagents were from Sigma Chemical or Fisher Scientific (Pittsburgh, PA).

	Results

Summary Introduction Procedures Results Discussion References

Spermine stimulation and pH sensitivity. Spermine has four macroscopic effects on NMDA receptors (Fig. 1A, 1-4) that can be studied in relative isolation by the use of particular subunit combinations and by manipulating the concentrations of glutamate and glycine and the holding potential (1, 32, 33). The glycine-independent form of spermine stimulation (Fig. 1A, 1) was studied at NR1A/NR2B receptors activated by saturating concentrations of glutamate and glycine in oocytes voltage-clamped at 20 mV (Figs. 2A and 3A). A number of mutations in the M1-M3 regions of NR1A, including W563L, S610A, and W636L, produced a small increase in spermine stimulation. However, a surprising finding was that mutations at W608, W611, and N616 in M2 and at Y647 in M3 of NR1A reduced or abolished spermine stimulation (Figs. 2A and 3A). In the case of N616, any of three mutations (N-to-Q, N-to-R, or N-to-G) reduced spermine stimulation. Mutations at the equivalent positions in the M2 and M3 regions of NR2B (W607L, W610L, N615Q, Y646L) had no effect or increased spermine stimulation (Fig. 3, A and B). Thus, the decrease in spermine stimulation seen with mutations in M2 and M3 is largely selective for the NR1A subunit.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Screening mutant NMDA receptors. A, The effects of spermine (100 µM) on responses to glutamate (glu; 10 µM, with 10 µM glycine) were studied on oocytes expressing NR1A/NR2B receptors with wild-type or mutant NR1A subunits and voltage-clamped at 20 mV. B, The effects of spermine (10 µM) were studied on oocytes expressing NR1A/NR2A receptors with wild-type or mutant NR1A subunits and voltage-clamped at 100 mV. During recovery from block by spermine, the oocytes were briefly depolarized to +40 mV to speed recovery. C, The effects of N1-DnsSpm (1 µM) were determined in the same oocytes as shown in B, but oocytes were voltage-clamped at 70 mV. Horizontal calibration bars, 20 sec in A, 40 sec in B and C.

View larger version (63K): [in this window] [in a new window]   Fig. 3.   Screening mutant NMDA receptors. A, The effects of 100 µM spermine were determined at NR1A/NR2B receptors containing wild-type and mutant subunits in oocytes voltage-clamped at 20 mV. The effects of 10 µM spermine (B) and 1 µM N1-DnsSpm (C) were determined at NR1A/NR2A receptors containing wild-type and mutant subunits in oocytes voltage-clamped at 100 mV (spermine) or 70 mV (N1-DnsSpm). For experiments with NR1A mutants, the mutants were coexpressed with wild-type NR2B (A) or wild-type NR2A (B and C). Similarly, for experiments with NR2B mutants (A) and NR2A mutants (B and C), the mutants were coexpressed with wild-type NR1A. In all paradigms, NMDA receptors were activated by glutamate (10 µM; with 10 µM glycine), and currents measured in the presence of spermine and N1-DnsSpm are expressed as a percentage of the control response to glutamate in each oocyte. Solid horizontal line, control response to glutamate (100%). Dashed horizontal line, mean effect of spermine or N1-DnsSpm at wild-type NR1A/NR2 receptors. Values are mean ± standard error from 4-14 oocytes for each mutant and from 20-66 oocytes for wild-type receptors, which were measured in all batches of oocytes. , p < 0.05, *, p < 0.01 compared with wild-type NR1A/NR2 receptors (one-way analysis of variance with post hoc Dunnett's test).

View this table: [in this window] [in a new window]   TABLE 1 Sensitivity of mutant NMDA receptors to glutamate and glycine EC50 values for glutamate and glycine were determined from concentration-response curves at wild-type and mutant NMDA receptors in oocytes voltage-clamped at 70 mV. Concentration-response curves for glutamate were carried out in the presence of 10 µM glycine, and those for glycine were carried out in the presence of 10 µM glutamate. The maximum response to glutamate and glycine (Imax) in the same experiments is also shown.

View larger version (56K): [in this window] [in a new window]   Fig. 4.   Proton inhibition of mutant NMDA receptors. A, Inward currents induced by glutamate (10 µM; with 10 µM glycine) at pH 8.5, 7.5, 7.0, and 6.5 in oocytes expressing NR1A/NR2B and NR1A(N616Q)/NR2B receptors and voltage-clamped at 70 mV. B, Inhibition by protons was determined using the experimental protocols shown in A. Values are mean ± standard error from seven to nine oocytes for each subunit combination. Responses to glutamate at a given pH (Iglu) are expressed as a fraction of the response to glutamate at pH 8.5 [Iglu(8.5)]. C, IC50 values for proton inhibition were determined by curve fitting of data from pH inhibition curves, similar to those shown in B, for various NR1/NR2 subunit combinations. For these analyses, the raw data (glutamate current versus pH) were fit to the equation Iglu = Imax/[1 + (pH/IC50)n], in which Iglu is the glutamate current at a given pH, Imax is the maximum glutamate-induced current, and n is the slope factor. Dashed line, mean IC50 value at the respective wild-type receptors (hatched bars). Values are mean ± standard error from 4-12 oocytes for each mutant and from 41 (NR1A/NR2B), 4 (NR1B/NR2B), and 21 (NR1A/NR2A) oocytes for wild-type receptors. *, p < 0.01 compared with respective wild-type receptor (one-way analysis of variance with post hoc Dunnett's test, or unpaired t test).

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Interactions of spermine and protons. A, Responses to glutamate (glu, 10 µM; with 10 µM glycine) were measured in the absence and presence of 100 µM spermine at different extracellular pH in oocytes voltage-clamped at 20 mV. Data are from the same oocyte for each subunit combination. Right, traces normalized to the glutamate response immediately before application of spermine at each pH. B, Effects of spermine were measured at different extracellular pH using the protocols shown in A at NR1A/NR2B receptors containing wild-type and mutant NR1A subunits. Responses measured in the presence of spermine are expressed as a percentage of the control response to glutamate at each pH. Values are mean ± standard error from four to six oocytes for each subunit combination.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Block of NMDA channels by extracellular Mg2+ is not pH sensitive. A, I-V curves were constructed by voltage ramps (100 to +40 mV over 4 sec) in an oocyte expressing NR1A/NR2B receptors activated by glutamate (glu, 10 µM; with 10 µM glycine) in the absence and presence of 100 µM Mg2+ at pH 7.0, 7.5, and 8.5. Leak currents have been subtracted. (Note the different scales on the ordinate at each pH.) B, Glutamate currents measured in the presence of Mg2+ are expressed as a fraction of the control current at each pH. Data around the reversal potentials have been masked for presentation.

Block by spermine and N¹-DnsSpm. We initially screened block by spermine and N¹-DnsSpm at NR1A mutants coexpressed with NR2A (Fig. 2, B and C, and Fig. 3, B and C). For these experiments, NR1A mutants were coexpressed with NR2A to avoid the complication of spermine stimulation, which occurs at NR1A/NR2B but not NR1A/NR2A receptors. Those analyses showed that mutations W563L, D581N, N616Q, N616R, and E621Q in NR1A reduced block by spermine and/or N¹-DnsSpm (Fig. 3, B and C). In contrast, mutations N616G and Y647L in NR1A increased block by N¹-DnsSpm.

View larger version (21K): [in this window] [in a new window]   Fig. 7.   Voltage-dependence of block by N1-DnsSpm. A, I-V relationships for responses to glutamate (10 µM; with 10 µM glycine) in the absence (control) and presence of 1 µM N1-DnsSpm were constructed by using voltage ramps (150 to +30 mV over 6 sec; inset) with oocytes expressing NR1A/NR2B and NR1A(W563L)/NR2B receptors. Leak currents have been subtracted. B, Currents in the presence of N1-DnsSpm are expressed as a fraction of the glutamate-induced current (Iglu), over the range 16 to 110 mV. Values of Kd(0) and z were derived by fitting the data to text eq. 3. Smooth lines, fits.

View this table: [in this window] [in a new window]   TABLE 2 Voltage-dependent block by N1-DnsSpm at mutant NMDA receptors I-V curves were constructed by using voltage ramps (150 to +30 mV over 6 sec) in the absence and presence of N1-DnsSpm. Values of Kd(0) and z were determined by fitting data from voltage ramps to text eq. 3 as illustrated in Fig. 7B.

View larger version (30K): [in this window] [in a new window]   Fig. 8.   N1-DnsSpm permeates NR1A/NR2B(W607L) and NR1A/NR2A(W606L) channels. A, I-V relationships for responses to glutamate (10 µM; with 10 µM glycine) in the absence (control) and presence of 1 µM or 0.1 µM N1-DnsSpm were constructed by using voltage ramps (150 to +30 mV over 6 sec; inset) on oocytes expressing wild-type and mutant receptors. Leak currents have been subtracted. B, Currents in the presence of N1-DnsSpm are expressed as a fraction of the glutamate-induced current (Iglu) at 50, 100, and 145 mV at each receptor type. Values are mean ± standard error from 7-15 oocytes for each subunit combination. Note the incomplete block and very shallow slope conductance with NR1A/NR2B(W607L) in A and the partial recovery from block at 145 mV with NR1A/NR2B(W607L), NR1A(N616G)/NR2B, and NR1A/NR2A(W606L) in B.

	Discussion

Summary Introduction Procedures Results Discussion References

The major effects of the mutations that we studied are summarized in Fig. 9. Mutations at some residues in M2, in particular at the critical asparagine residues, including NR1A(N616) and NR2B(N615), have previously been shown to affect permeation and block by inorganic divalent cations (13-17). Here, we extended characterization of the pore-forming M2 region by determining the effects of mutations in M2 on modulation and block by polyamines and protons. In addition, residues in the M1 and M3 regions that affect sensitivity to channel blockers and modulators have been identified.

View larger version (27K): [in this window] [in a new window]   Fig. 9.   Summary of the effects of mutations in M1-M3. A, The effects of mutations at residues in NR1A and, in some cases, at the corresponding position in NR2B are summarized. Upward arrows, an increase. Downward arrows, a decrease. Dashes, no effect of the mutation on a particular experimental paradigm. Thickness of the arrows, relative magnitude of the effects. In the case of N1-DnsSpm, the effects of mutations on the values of Kd(0) and z are included to facilitate comparisons of the mechanisms that may influence block. Thus, a decrease in block by N1-DnsSpm at receptors containing NR1A(W563L) is due entirely to an increase in Kd(0), whereas a decrease in block at receptors containing NR1A(N616Q) is due to a decrease in voltage-dependence (z). nd, not determined. *, Mutation NR2B(W607L) greatly increased the apparent permeation of N1-DnsSpm, and block at NR1A/NR2B(W607L) channels was very shallow. The equivalent mutation in NR2A, NR2A(W606L), also increased permeation of N1-DnsSpm, and the mutation had no significant effect on the values of Kd(0) and z (Table 2). B, Putative transmembrane topology of NR1 and NR2B subunits and the relative positions of the key residues described in this study (). , Positions of residues E342 and D669 in NR1, which influence spermine stimulation and proton inhibition (32, 33). The binding sites for glutamate (Glu) and glycine (Gly) are formed by regions in the distal part of the amino-terminal domain and the M3-M4 loop (37, 38). Mutations at residues such as NR1A(N616) and NR1A(Y647) may alter spermine stimulation, proton inhibition, and sensitivity to glutamate and glycine by disrupting channel gating and/or by long-range allosteric effects that alter the binding and coupling of these agents. Because of their effects on the affinity [i.e., Kd(0)] of block by N1-DnsSpm, residues NR1A(W563), NR2B(N616), and NR2B(Y646) are proposed to form part of a binding site for N1-DnsSpm and to contribute directly to the pore or vestibule of the ion channel. Other residues, including NR2B(W559), NR1A(N616), and NR1A(Y647), may influence or interfere with binding of N1-DnsSpm. Residue W607 in NR2B may contribute to a narrow constriction in the pore and/or form part of the binding site for N1-DnsSpm because a NR2B(W607L) mutation greatly increases permeation of N1-DnsSpm.

Block by spermine is voltage dependent and presumably involves a spermine binding site within the channel pore. In contrast, spermine stimulation and proton inhibition are not voltage dependent, and these effects are unlikely to involve direct interactions of spermine and protons with residues in the pore-forming regions of the channel. Thus, a surprising finding was that mutations at W608, W611, and N616 in M2 and at Y647 in M3 of NR1A altered spermine stimulation and proton inhibition. In the case of mutations at W608, N616, and Y647 in NR1A, a decrease in spermine stimulation was correlated with a decrease in proton inhibition. These effects are similar to those of mutations at residues E342 and D669, which are located in the extracellular domains of NR1A (32, 33) (Fig. 9B). The change in sensitivity to spermine may be secondary to a change in pH sensitivity if spermine functions to relieve proton inhibition (mechanism 6, Fig. 1A) (5). Alternatively, the mutations may disrupt a common determinant that couples the spermine and proton binding sites to channel gating. In this case, the effects of spermine and protons would not be functionally interdependent, and spermine stimulation would not involve a relief of proton inhibition (i.e., mechanism 6, Fig. 1A, would not exist). Rather, the separate and opposing effects of spermine and protons may be mediated through a structural motif that involves residues W608, N616, and Y647 in NR1A or at least through a motif that is disrupted by mutations at these residues.

With the NR1A(N616) mutants, the largest effect on pH sensitivity was seen with an N-to-Q mutation, whereas N-to-R or N-to-G mutations at this position produced smaller changes in pH sensitivity. This profile is different from the effects of N616 mutations on Mg²⁺ block, in which substitution of the positively charged R residue produces the largest reduction in Mg²⁺ block, and on cation permeability, in which the N-to-G mutation produces the largest increase in pore size (12-15). Proton inhibition was reduced in receptors containing an N-to-Q mutation at N637 in the NR1B variant of NR1, similar to the reduction seen in receptors with an N-to-Q mutation at the equivalent position, N616, in NR1A. The NR1B variant contains the exon 5 insert, which itself reduces proton inhibition, possibly by interacting with a proton sensor on the receptor. These observations, together with the lack of voltage-dependence of proton inhibition and the lack of interaction of protons and Mg²⁺, suggest that NR1A(N616) does not contribute directly to a proton sensor. Mutations at N616, and at other residues in M2 and M3, may indirectly alter sensitivity to spermine and pH through subtle disruptions of the subunit protein structure or disruptions to gating processes. Changes in gating produced by mutations at residues in the M1-M3 region could also account for the increase in sensitivity to glutamate and glycine seen with these mutants. The glutamate and glycine binding sites are thought to be formed by residues in the extracellular amino-terminal and M3-M4 loop regions (37, 38) (Fig. 9B), and residues in M1-M3 are unlikely to directly affect agonist binding. Indeed, the NR1A(N616Q) mutation was recently shown to affect the coupling of cation permeation to channel gating (39), which may also account for some of the effects of this mutant seen in the current study.

The effects of mutations in M2 and M3 on pH sensitivity were selective for the NR1 subunit. Mutations at W608, N616, and Y647 in NR1A reduced pH sensitivity, whereas mutations at the equivalent or adjacent positions in NR2B did not (Fig. 9). Thus, mutations in NR1 may affect a proton sensor and a spermine binding site located on the NR1 subunit. It is notable that mutations at NR1A(N616) produce receptors with a complex pharmacological phenotype. Thus, in addition to effects on block and permeation of divalent cations, mutations at NR1A(N616) produce changes in pH sensitivity and agonist potency, which should be kept in mind when evaluating the macroscopic effects of mutations at this position.

Block by spermine and N¹-DnsSpm was influenced by mutations in M1-M3. Using the Woodhull model of voltage-dependent channel block (31), we determined the effects of the mutations on the K_d(0) and z of N¹-DnsSpm block. In this analysis, z is the valence of N¹-DnsSpm, which is +3 at pH 7.5 (6), and  is the fraction of the membrane electrical field sensed by the blocker. As previously shown (6), block by N¹-DnsSpm was strongly voltage dependent with a z of 2.6-2.7 at wild-type channels. This suggests that the binding site is very deep within the channel, although there are a number of potential limitations to use of this model to analyze block by N¹-DnsSpm. For example, it is not known if only one molecule of N¹-DnsSpm enters the channel at one time or if the entire polyamine tail on the molecule enters the membrane field (6). Nevertheless, the data derived from voltage ramps were well-fit by eq. 3, and these analyses provide some information about amino acid residues that seem to form part of a polyamine binding site and residues that may interfere with block by polyamines.

The proposed topology of the M1-M3 regions of NR1 and NR2 subunits (40-43) and the proposed positions of residues in the M1-M3 regions are shown in Fig. 9B. Thus, M2 is proposed to be a hair-pin loop with the critical asparagine residues that form the narrowest part of the pore (N616 in NR1 and NR2B) near the top of the loop and NR1A(E621), immediately after M2, at the cytoplasmic side of the pore. Residues W563 and Y647 in NR1A are close to the extracellular ends of M1 and M3, respectively, as are the corresponding residues, W559 and Y646, in NR2B. Residues at which mutations selectively increase the K_d(0) value (i.e., decrease the affinity) of N¹-DnsSpm may contribute directly to a binding site for N¹-DnsSpm; these include W563 and E621 in NR1A, and N616 and Y646 in NR2B. Interestingly, the corresponding residues in the other subunit have different effects. For example, mutation NR1A(W563L) increased the K_d(0) value, whereas mutation NR2B(W559L) did not. A Y-to-L mutation at NR1A(Y647L), equivalent to NR2B(Y646L), actually increased N¹-DnsSpm block by increasing the affinity of binding (Fig. 9). This mutation, however, did not affect block by spermine. Thus, residue NR1A(Y647) may normally interfere with binding of the head group of N¹-DnsSpm at wild-type channels, possibly through steric hindrance because the Y-to-L mutation reduces the size of the side chain at this position. Residues N616 in NR1A and N615 and W607 in NR2B may also normally interfere with binding of N¹-DnsSpm because mutations at those positions alter the z and/or apparent permeation of N¹-DnsSpm. Some residues in the pore-forming region that are involved in binding channel blockers may also play a role in gating processes because mutations at W563, N616, and Y647 in NR1 increased sensitivity to glutamate and glycine (Fig. 9A).

We previously found that an N-to-G mutation at NR1A(N616G) or NR2A(N615G) increased the potency of N¹-DnsSpm and that N¹-DnsSpm could easily permeate the channel of receptors containing these mutants. This is presumably because the N-to-G mutation increases pore size, allowing passage of the bulky naphthalene head group of N¹-DnsSpm (6). These results are consistent with the finding, based on the permeability of organic cations, that the asparagine residues seem to form the narrowest part of the channel (12). An effect of the second (N616) rather than the first (N615) asparagine residue in the M2 region of NR2B on block by N¹-DnsSpm (Fig. 9A) is consistent with a staggered position of the M2 regions in NR1 and NR2 subunits, with the second asparagine in NR2 having a function similar to that of N616 in NR1A (12) (Fig. 9B). The W-to-L mutation at NR2B(W607L) produced channels with a very shallow and incomplete block by N¹-DnsSpm, suggesting that N¹-DnsSpm can easily permeate these channels. Thus, NR2B(W607) may also form part of the narrowest constriction of the channel, and the W-to-L mutation may increase the size of the pore, allowing passage of N¹-DnsSpm. Importantly, a W-to-L mutation at the equivalent position (W608) in NR1A did not affect block by N¹-DnsSpm, whereas a mutation at the equivalent position in NR2A, NR2A(W606L), did increase permeation of N¹-DnsSpm. This suggests that only in NR2 subunits is the tryptophan residue at this position critical for permeation of N¹-DnsSpm. The NR2B(W607L) mutation is presumably sufficient to make the pore large enough to allow passage of the bulky-head group of N¹-DnsSpm.

In a model of the structure of M2 based on cysteine-accessibility mutagenesis, the tryptophan residues in the NR1 and NR2C subunits at positions equivalent to NR2B(W607) were proposed to face the lumen of the channel (40); this is consistent with a direct interaction of NR2B(W607) with N¹-DnsSpm in the channel pore. However, in the model proposed by Kuner et al. (40), the narrow constriction of the channel is formed by the critical asparagine residues, and the residues at positions equivalent to NR2B(W607) lie some distance below the narrow constriction. Indeed, the tryptophan residue in NR2C was accessible to cytoplasmic but not to extracellular thiol reagents after cysteine mutagenesis (40). Thus, mutations at the tryptophan residue (W607 in NR2B, W606 in NR2A) would not be expected to alter permeation of blockers, such as N¹-DnsSpm, entering the channel from the extracellular side.

Although it is difficult to reconcile our data with the data of Kuner et al. (12, 40), there are a number of possibilities that could account for the effects of the NR2B(W607L) mutant. First, residue NR2B(W607) may contribute to the narrow constriction of the channel together with the asparagines in NR1 and NR2 subunits. This would require that W607 lies in the same plane as the asparagines rather than below them, as in the model by Kuner et al. In this case, all or part of the M2 loop may exist parallel to the plane of the membrane rather than perpendicular to it, as was proposed for the structure of the loop region in a cyclic nucleotide-gated channel (44). Second, the effects of W607L may reflect changes in the binding of N¹-DnsSpm rather than changes in pore size. In this case, the relief from block at extreme negative potentials would reflect some process other than permeation of N¹-DnsSpm. This seems unlikely because N-to-G mutations such as NR1A(N616G), which increase pore size (12), also increase permeation of N¹-DnsSpm as determined by the relief of block at extreme negative potentials (6). Furthermore, the NR2A(W606L) mutation, which increased the permeation of N¹-DnsSpm, did not affect the values of either K_d(0) or z for block by N¹-DnsSpm. This suggests that the binding and unbinding of N¹-DnsSpm within NR1A/NR2A(W606L) channels remain intact and that the NR2A(W606L) mutation alters the ability of N¹-DnsSpm to pass through the channel. Finally, we cannot exclude the possibility that the NR2B(W607L) and NR2A(W606L) mutations disrupt the asparagine residues further along the M2 loop or cause a general disruption of the secondary structure or positioning of the M2 loop, which could affect block and permeation of N¹-DnsSpm. However, this seems unlikely given the lack of effect of a corresponding mutation in the NR1 subunit, NR1A(W608L).