Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

N-Methyl-D-aspartate (NMDA) receptors are ligandgated ion channels that have structural similarities with the -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and kainate receptors (Dingledine et al., 1999). NMDA receptors have several characteristics that distinguish them from the other glutamate receptors. These include the requirement for two different agonists (glutamate and glycine) to activate the receptor, a high permeability of the channel to Ca²⁺, and a voltage-dependent block of the channel by Mg²⁺ (Dingledine et al., 1999). NMDA channels are also blocked by a large number of structurally dissimilar organic blockers including ketamine, MK-801, memantine, and various spider toxins and polyamine derivatives (Huettner and Bean, 1988; Collingridge and Lester, 1989; Chen et al., 1992; Igarashi et al., 1997).

With the cloning of cDNAs encoding subunits of glutamate receptors, it has been possible to begin to delineate the structural features that may contribute to various binding sites and/or functional domains of these receptors. The glutamate binding site is formed by two domains (S1 and S2) in the amino terminus and M3-M4 loop of the NR2 subunit, whereas homologous domains in NR1 form the glycine binding site (Kuryatov et al., 1994; Laube et al., 1997; Anson et al., 1998). The amino-terminal domain preceding S1 seems to be a regulatory domain that may contain binding sites for modulators and antagonists such as spermine, protons, ifenprodil, and Zn²⁺ (Masuko et al., 1999; Low et al., 2000; Paoletti et al., 2000). The M2 loop region is a critical determinant of divalent cation permeability and Mg²⁺ block. In particular, asparagine residues in this region form part of a Mg²⁺ binding site and contribute to the selectivity filter of the channel (Dingledine et al., 1999). These asparagine residues, which are in positions analogous to the Q/R sites that control divalent cation permeability of AMPA and kainate channels, have also been found to influence block by organic channel blockers such as MK-801 (Dingledine et al., 1999). There is little information about how other residues in the membrane-spanning and pore-forming regions of NMDA receptors affect block by compounds such as MK-801, memantine, and other organic channel blockers.

In this article, we have studied mutations in the M2 pore-forming loop and in the M1, M3, and M4 membrane-spanning regions and in some adjacent regions. We determined the effects of these mutations on block by three structurally distinct channel blockers: MK-801, memantine, and N¹-N⁴-N⁸-tribenzyl-spermidine (TB-3-4). MK-801 is a prototypical, high-affinity NMDA channel blocker with a very slow onset and recovery of block (Wong et al., 1986; Huettner and Bean, 1988), whereas memantine is a low-affinity blocker with much faster rates of block and unblock (Chen et al., 1992; Blanpied et al., 1997). Both molecules have quite rigid structures. TB-3-4 is a potent and selective NMDA blocker and is a much larger and more flexible molecule than either MK-801 or memantine (Igarashi et al., 1997). Our results show that block by these three compounds is affected not only by residues in the M2 loop but also by residues toward the extracellular ends of M1, M3, and M4. Notably, mutations in these regions have differential effects on the different blockers. Surprisingly, many mutants that influence block by TB-3-4 also affect MK-801 but have no effect on block by memantine.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

NMDA Clones and Site-Directed Mutagenesis. The NR1 clone used in these studies is the NR1A variant (Moriyoshi et al., 1991) which lacks the 21-amino acid insert encoded by exon-5. This clone, and some of the NR1 mutants in the M2 and M1-M2 linker region (Sakurada et al., 1993) were gifts from Dr. S. Nakanishi (Institute for Immunology, Kyoto University Faculty of Medicine, Kyoto, Japan). The rat and mouse NR2B clones (Kutsuwada et al., 1992; Monyer et al., 1992) were gifts from Drs. P. H. Seeburg (Center for Molecular Biology, University of Heidelberg, Germany) and M. Mishina (University of Tokyo, Tokyo, Japan).

Numbering of Residues. In NR1 and NR2 subunits, amino acids are numbered from the initiator methionine as in the original paper reporting the sequence of NR1 (Moriyoshi et al., 1991). This differs from the numbering system used in some other laboratories, in which residues are numbered from the start of the mature peptide (Kuner et al., 1996; Beck et al., 1999). In the case of NR1, there is an 18 amino acid signal peptide so, for example, residue NR1(N616) described in this study corresponds to residue NR1(N598) using the alternative numbering scheme (Kuner et al., 1996).

Expression in Oocytes and Voltage-Clamp Recording. The preparation of capped cRNAs and the preparation, injection, and maintenance of oocytes were carried out as described previously (Williams et al., 1993). Oocytes were injected with NR1 plus NR2 cRNAs in a ratio of 1:5 (0.1-4 ng of NR1 plus 0.5-20 ng of NR2). Macroscopic currents were recorded with a two-electrode voltage-clamp using a GeneClamp 500 amplifier (Axon Instruments, Union City, CA) as described previously (Williams, 1993). Electrodes were filled with 3 M KCl and had resistances of 0.4 to 4 M. Oocytes were continuously superfused with a saline solution (100 mM NaCl, 2 mM KCl, 1.8 mM BaCl₂, and 10 mM HEPES, pH 7.5) that contained BaCl₂ rather than CaCl₂ to minimize Ca²⁺-activated Cl currents, and in most experiments oocytes were injected with K⁺-BAPTA (100 nl of 40 mM, pH 7.0-7.4) on the day of recording.

(1)

(2)

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Screening Mutations. We initially made a series of individual point mutations in and around the membrane-spanning and pore-forming loop regions of the NR1 subunit. The mutants were constructed in most cases to alter the functional side chain, for example by neutralizing the charge (E-to-Q, D-to-N), removing an aromatic ring (W-to-L, Y-to-L) or removing a hydroxyl group (S-to-A, T-to-A). The regions that were studied are shown schematically in Fig. 1A. We screened each mutation by measuring block using a single concentration of TB-3-4, memantine, and MK-801 at NR1/NR2B receptors (Fig. 1). In these experiments, we also studied the effects of mutations on block by 100 µM Mg²⁺.

View larger version (52K): [in this window] [in a new window]   Fig. 1.   Screening NR1 mutants. A, schematic of the NR1 subunit, which contains three membrane-spanning domains (M1, M3, and M4), a re-entrant loop (M2), and a large extracellular loop between M3 and M4. The stippled boxes indicate the regions in which mutations were studied. B, examples of experiments designed to screen for effects of mutations on inhibition by channel blockers. The traces are inward currents induced by glutamate (glu, 10 µM; with 10 µM glycine) in the absence or presence of blockers in oocytes voltage-clamped at 70 mV. In these examples, the effects of 0.3 µM TB-3-4 (top row) and 1 µM memantine (bottom row) were determined at NR1/NR2B receptors containing wild-type NR1, NR1(N616Q), and NR1(T807C) subunits. All horizontal scale bars are 20 s, and all vertical scale bars are 50 nA. C, the effects of TB-3-4 (0.3 µM), memantine (1 µM), MK-801 (30 nM), and Mg2+ (100 µM) were determined at NR1/NR2B receptors containing wild-type and mutant subunits using protocols similar to those shown in B. Currents measured in the presence of blockers are expressed as a percentage of control currents.

The M2 Loop/Pore-Forming Region. The M2 loop of NR1 contains an asparagine (N616) that has previously been shown to influence divalent cation permeability and block by Mg²⁺, MK-801, and other channel blockers (Dingledine et al., 1999). An N-to-Q mutation at this position reduced block by TB-3-4, memantine, and MK-801. The NR2B subunit contains two asparagines (N615 and N616) at positions equivalent to N616 and S617 in NR1, and we studied mutations at both N residues in NR2B. In the M2 region of NR1, mutations at two tryptophan residues (W608 and W611) had small effects on block by memantine (Fig. 1C). We have previously found that mutations at NR1(W608) had little or no effect on Mg²⁺ block, whereas mutations at NR2B(W607), a position equivalent to NR1(W608), had dramatic effects on Mg²⁺ block (Williams et al., 1998). In light of this, we studied the effects of NR2B(W607) mutations on block by TB-3-4, memantine, and MK-801. Thus, the studies of the M2 region are focused on NR1(N616), NR2B(N615), NR2B(N616), and NR2B(W607).

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Effects of mutations in the M2 region. Concentration-inhibition curves for TB-3-4, memantine, and MK-801 at wild-type and mutant NR1/NR2B receptors were constructed using four or more oocytes for each concentration of blocker. Data for wild-type receptors () are shown only in the top, and the fitted curves for wild-type receptors are shown by broken lines. Data are currents measured in the presence of blocker expressed as a percentage of the control current measured in the absence of the blocker. The structures of the three blockers are shown at the top, and the relative positions of the mutations are shown schematically at the left side.

View this table: [in this window] [in a new window]   TABLE 1 Effects of NR1 and NR2B mutants in the M2 loop region on block by TB-3-4, memantine, and MK-801 Values of the IC50 were determined from concentration-inhibition curves, and values of the Kd(0) and z were determined using voltage ramps analyzed with the Woodhull model of voltage-dependent block.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Effects of mutations in the M2 regions of NR1 and NR2B. Concentration-inhibition curves for MK-801 at wild-type and mutant NR1/NR2B receptors were constructed using four or more oocytes for each concentration of MK-801. The data for wild-type receptors are not shown, but the fitted curves for wild-type receptors are shown by broken lines. Data are currents measured in the presence of MK-801 expressed as a percentage of the control current measured in the absence of MK-801.

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Voltage-dependent block by TB-3-4. Current-voltage (I-V) relationships were measured by voltage ramps in the absence (control) and presence of TB-3-4 at receptors containing wild-type (A), L655A (B), and W607N(C) NR1 mutants. Bottom, currents measured in the presence of TB-3-4 (Iglu+TB-3-4) are expressed as a fraction of the control current (Iglu). Data around the reversal potential have been masked. The solid lines for NR1/NR2B and NR1(L655A)/NR2B are fits to Woodhull model (eq. 2). Note the incomplete block at NR1/NR2B(W607N) receptors and the apparent reversal of the slope at very negative membrane potentials.

The M1, M3, and M4 Domains. The effects of mutations in these regions are listed in Table 2 and are summarized (together with the effects on mutations in the M2 region) in Fig. 5. Mutations that produced a 3-fold change in IC₅₀ (shaded areas on Fig. 5) were considered to be significant. In the pre-M1 and M1 regions of NR1, an F558L mutation had a modest effect on block by TB-3-4 and MK-801 without affecting memantine (Table 2, Fig. 5), whereas mutations at W563 had larger effects on block by TB-3-4 and MK-801, again with no effect on block by memantine (Fig. 6B, top, and Table 2).

View this table: [in this window] [in a new window]   TABLE 2 Effects of NR1 and NR2B mutants in and around the M1, M3, and M4 regions on block by TB-3-4, memantine, and MK-801 Values of the IC50 were determined from concentration-inhibition curves, and values of the Kd(0) and z were determined using voltage ramps analyzed with the Woodhull model of voltage-dependent block.

View larger version (73K): [in this window] [in a new window]   Fig. 5.   Summary of the effects of mutants on the potencies of channel blockers. The IC50 values for channel blockers at each mutant are expressed relative to wild-type receptors. Data are from concentration-inhibition curves and are re-plotted from Tables 1 and 2. The shaded area represents a 3-fold change in potency. Different mutations at the same position are grouped by thin vertical bars on the left. Thick vertical bars group residues within the various regions (M1, M2, M3, etc.).

View larger version (35K): [in this window] [in a new window]   Fig. 6.   Effects of mutations in and adjacent to the M1, M3, and M4 regions of NR1 and NR2B. A, schematic showing the positions of the key residues described in this figure and in Tables 1 and 2 and Fig. 5. B, concentration-inhibition curves for TB-3-4, memantine, and MK-801 at wild-type and mutant NR1/NR2B receptors were constructed using four or more oocytes for each concentration of blocker. For clarity, the data for wild-type receptors are not shown, but the fitted curves for wild-type receptors are shown by broken lines. Data are currents measured in the presence of blocker expressed as a percentage of the control current measured in the absence of the blocker. The relative positions of the mutations are shown schematically at the left side.

View larger version (35K): [in this window] [in a new window]   Fig. 7.   Effects of different mutations at individual residues. A number of different mutations (e.g., W-to-L, W-to-A, W-to-Y, and W-to-F) were studied at residues W563 (A), N650 (B), and T807 (C) in the NR1 subunit. Concentration-inhibition curves for TB-3-4 and MK-801 at wild-type and mutant NR1/NR2B receptors were constructed using four or more oocytes for each concentration of blocker. For clarity, the data for wild-type receptors are not shown, but the fitted curves for wild-type receptors are shown by broken lines. Data are currents measured in the presence of blocker expressed as a percentage of the control current measured in the absence of the blocker.

Mutations That Generate Constitutively Open Channels. In initial experiments in which we screened a large number of mutations in the NR1 subunit, we found that oocytes expressing NR1/NR2B receptors with a T-to-A or T-to-S mutation at T648 (located in M3) had very large holding currents when voltage-clamped at 70 mV. An example is shown in Fig. 8A. Application of glutamate and glycine to these mutants did induce inward currents, but the currents were often irreversible on wash-out of the agonists (data not shown), and we did not attempt to study the effects of TB-3-4, memantine, and MK-801 at the T648 mutants. We reasoned that the large holding currents required to voltage-clamp cells expressing NR1(T648) mutants may be due to the expression of constitutively open channels in receptors containing these mutants. If this were the case, replacing extracellular Na⁺ (the main charge carrier) with the large impermeant cation N-methyl-D-glucamine (NMDG) should reduce the currents. We also hypothesized that extracellular Mg²⁺, applied in the absence of agonists, would block these constitutively open channels, provided that the mutation does not disrupt the Mg²⁺ binding site. To test this, we studied the effects of substituting NMDG for Na⁺ in the extracellular solution, and we also studied the effects of extracellular Mg²⁺. Substitution of Na⁺ by NMDG or the addition of 100 µM Mg²⁺ greatly reduced the holding currents in cells expressing receptors containing NR1(T648) mutants but had little effect on wild-type receptors (Fig. 8, A and C). In light of this, we also examined the effects of NMDG and Mg²⁺ at many of the other key mutations that we had characterized and at some adjacent and nearby residues (Fig. 8C). Only mutations at Q556, P557, T648, and L657 in NR1 produced holding currents that were larger than wild-type and were sensitive to NMDG and Mg²⁺ (Fig. 8, B and C). An NR2B(N649D) mutation had a similar effect. In all cases the increased holding current, and consequent reduction by NMDG or Mg²⁺, were modest compared with the effects seen with NR1(T648) mutants.

View larger version (26K): [in this window] [in a new window]   Fig. 8.   Residues that may influence gating of NMDA receptors. A, holding currents were recorded in oocytes expressing wild type NR1/NR2B and NR1(T648S)/NR2B receptors and voltage-clamped at 70 mV. Buffer containing 100 mM NMDG (substituted for extracellular Na+) or 100 µM Mg2+ was applied during the times indicated by horizontal bars. Note that NMDG substitution and application of Mg2+ was carried out in the absence of agonists. B, schematic of the NR1 subunit showing positions of residues at which NMDG and Mg2+ reduced the holding current. C, holding current (in oocytes voltage-clamped at 70 mV), and effects of NMDG and Mg2+ on the holding current in NR1/NR2B receptors containing wild-type and mutant subunits. The effects of NMDG and Mg2+ were determined using protocols similar to those shown in A, and the data represent the difference in current measured in the absence and presence of NMDG or Mg2+.

Effects of Mutations on Permeability of Ba²⁺. Mutations at the N residues in the M2 loop and at NR2B(W607) have previously been shown to affect permeability of divalent cations including Ca²⁺ and Ba²⁺ (Williams et al., 1998; Dingledine et al., 1999). We carried out experiments to determine whether the key residues identified in this study affect permeability of Ba²⁺. Reversal potentials were measured in extracellular solutions that contained Na⁺ or Ba²⁺ as the main charge carrier. The shift in reversal potential between Na⁺ and Ba²⁺ is an index of the permeability of the channel to Ba²⁺. We used Ba²⁺ rather than Ca²⁺ in these studies to minimize Ca²⁺-activated Cl conductances (Leonard and Kelso, 1990). At wild-type NR1/NR2B receptors, the reversal potentials were 1.6 ± 0.3 mV (Na⁺) and +21.1 ± 0.6 mV (Ba²⁺). As shown in Fig. 9, mutations at NR1(N616) and NR2B(W607) in the M2 loop had characteristic effects on Ba²⁺ permeability (Dingledine et al., 1999). For example, mutations of NR1(N616) to Q, G, or W markedly reduced Ba²⁺ permeability and an N-to-R mutation had an even larger effect. Mutations of NR2B(W607) to L, N, or A, but not to Y or F, reduced Ba²⁺ permeability, highlighting the importance of an aromatic ring (as found in W, Y, and F) at this position. An NR2B(W559L) mutation had a small effect on Ba²⁺ permeability but two other mutations at this position had no effect. All of the other key mutants that were studied had no effect on Ba²⁺ permeability (Fig. 9). This highlights the specificity of these mutations and indicates that these residues are determinants of block by TB-3-4 and MK-801 but not of Ba²⁺ permeability.

View larger version (38K): [in this window] [in a new window]   Fig. 9.   Ba2+ permeability at NMDA channels. The shifts in reversal potential seen when currents were measured in Ba2+ saline versus Na+ saline were determined at wild-type and mutant NR1/NR2B receptors. Different mutations at the same position are grouped by thin vertical bars on the left. Thick vertical bars group residues within the various regions (M1, M2, M3, etc.). Data for some mutants in the M2 region are from Williams et al. (1998).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we have identified residues in several regions of NMDA receptor subunits that differentially affect block by MK-801, memantine, and TB-3-4. A model that summarizes this work is shown in Fig. 10. The model is based, in part, on previous studies in which solvent-accessible residues were probed with MTS reagents after cysteine substitution (Kuner et al., 1996; Beck et al., 1999). The M2 segment of NR1 was proposed to contain an -helix terminating at N616 followed by an extended structure or random coil (Kuner et al., 1996). The critical asparagines in the M2 loop of NR1 and NR2B form the binding sites for Mg²⁺ and contribute to the narrow constriction of the pore (Wollmuth et al., 1996; Wollmuth et al., 1998a,b).

View larger version (41K): [in this window] [in a new window]   Fig. 10.   Modeling residues that affect blockers in the pore and vestibule region of the NMDA receptor. The M1-M2-M3 region is depicted as a helix-pore-loop-helix similar to the structure reported for the KcsA potassium channel (Doyle et al., 1998). The M2 region contains a helix followed by a random coil structure, with the critical asparagines (N616 in NR1 and N615 in NR2B) at the tip of the helix. Only the top half of each M4 helix is shown, and the position of these helices (for which there is no equivalent in KcsA) relative to M1-M2-M3 is uncertain. Red circles indicate residues at which mutations affect block by MK-801, TB-3-4, and memantine. Green circles indicate residues at which mutations affect block by MK-801 and TB-3-4 but not by memantine. Dark blue circles are residues at which mutations generate constitutively open channels. Circles outlined in light blue indicate residues that, when mutated to cysteine, were reactive with MTS reagents and are presumed to be solvent-exposed within the lumen of the channel (Kuner et al., 1996; Beck et al., 1999). Blue asterisks indicate residues at which solvent accessibility has not been determined. Residue N649 in NR2B is assumed to be solvent accessible because the equivalent residue (N650) in NR1 is solvent accessible. Although only two subunits are shown, the receptor is most probably a tetramer or a pentamer containing two NR1 and two NR2 subunits. MK-801 is shown, drawn to scale, inside the channel, with the amino group positioned near NR1(N616) and NR2B(N615). The structures of memantine and TB-3-4 are shown adjacent to the model for comparison with MK-801. In these structures, carbon atoms are gray, hydrogen atoms are light blue, and nitrogen atoms are dark blue.

The model is also related to the known structure of the KcsA potassium channel (Doyle et al., 1998). There is some amino acid homology between the M2 loops of glutamate receptor subunits and the pore helix of KcsA, although in the case of NMDA receptor subunits, the homology is very limited (Fig. 11). It is possible that the M1 through M3 region of NMDA receptors has a structure similar to KcsA; M1 corresponds to the outer helix and M3 to the inner helix of KcsA. Indeed, secondary structure prediction of these regions of NR1 and NR2B suggests the presence of  helices in M1, M2, and M3 (Fig. 11). KcsA channel subunits have a pore-forming loop containing an  helix followed by a random coil; this structure may be similar to the M2 loop region of glutamate receptor subunits (Panchenko et al., 2001). Notably, the secondary structure predicted for NR1 has an M2 helix that ends just before N616, consistent with the model proposed by Kuner et al. (1996). A shorter helix is predicted in the M2 region of NR2B (Fig. 11). In this context, it is notable that there is very little amino acid identity between the M2 regions of NR1 and NR2B. It is thus conceivable that the pore loops in NR1 and NR2B are somewhat different in terms of their structure, length, or positioning within the pore. Compared with KcsA, NMDA receptor subunits contain another putative helix, M4, so the overall packing and arrangement of the membrane spanning helices in glutamate receptor subunits may be different from those in KcsA.

View larger version (28K): [in this window] [in a new window]   Fig. 11.   Sequences of NR1 and NR2B in the M1 through M3 regions aligned with KcsA. Colons indicate similar residues in all three sequences; asterisk indicate identical residues. Sequence alignment was performed using CLUSTAL W. The segments originally designated M1, M2, and M3 are shown by lines above the sequences. Secondary structure prediction was carried out using SOPMA (http://pbil.ibcp.fr/) and is indicated above the amino acid sequences: h, -helix (shown in red); e, extended strand; t,  turn; c, random coil. Residues in KcsA that form the outer, inner, and pore helices are shown in red.

Mutations at the critical asparagines in M2 (N616 in NR1, N615 and N616 in NR2B) had marked effects on block by MK-801 and memantine and modest effects on block by TB-3-4. It is likely that these residues interact with the amino group of MK-801 and memantine and, possibly, with one of the amino groups of TB-3-4. The effects of mutations at NR1(N616) and NR2B(N615) were additive, suggesting that they make separate contributions to a binding site for the blocker. Some N-site mutations not only reduced the affinity but also increased the apparent permeation of TB-3-4 and memantine, similar to effects on block by N¹-dansyl-spermine (Chao et al., 1997). Mutations such as NR1(N616G) presumably increase the size of the narrow constriction, allowing the cationic blockers to actually permeate the channel when the driving force is sufficiently large. The values of z for TB-3-4 and memantine at wild-type receptors were 1.38 and 0.89, respectively. Assuming that only one molecule of each blocker binds within the transmembrane electrical field, this would correspond to  values of 0.46 for TB-3-4 and 0.89 for memantine. TB-3-4 is a long and highly flexible molecule. If it bound in the channel in an extended conformation, then the positive charges might be distributed over some distance and it is difficult to strictly interpret the  values as a depth of field for this molecule. In the case of memantine, it is possible that two molecules of memantine bind simultaneously within the channel (Sobolevsky and Koshelev, 1998). This would yield an average  value of 0.45 for memantine, consistent with the tip of the helix being located about halfway across the membrane electrical field (Fig. 10).

We have shown previously that mutations at NR2B(W607) alter block and permeation of extracellular Mg²⁺ (Williams et al., 1998). The effects of the NR2B(W607) mutants seen in the present study were reminiscent of their effects on block and permeation of Mg²⁺, with an aromatic residue at this position being important for block by the organic blockers. Mutations at the equivalent residue in NR1, W608, have little or no effect. We suggested previously that NR2B(W607) may contribute directly to the narrow constriction and Mg²⁺ binding site (Williams et al., 1998). In this case, it could also contribute to the binding site for organic channel blockers such as MK-801, TB-3-4, and memantine (Fig. 10). Another possibility is that NR2B(W607) forms part of a structural backbone that is important for control of the pore structure. There is a conserved tryptophan in an equivalent position in KcsA (W67), the side chain of which is not exposed to the lumen of the channel but forms part of a cuff of aromatic residues that constrain the opening of the channel pore (Doyle et al., 1998).

In addition to M2, the other regions in which we identified residues that influence channel block were the pre-M1, M1, M3, post-M3, and pre-M4 regions. In the model (Fig. 10), the channel is shown as a structure with a pore formed by the helix and random coil in M2 and an outer vestibule formed by the M3 and post-M3 segments. The pre-M1 and M4 regions may also contribute to the vestibule (Beck et al., 1999). The helices are proposed to span a membrane distance of 34 Å with the dimension of the narrow constriction being 5.5 Å (Villarroel et al., 1995; Wollmuth et al., 1996). Thus, in this model, it is possible to show the relative positions of residues along each helix, but the relative positioning of residues between helices is uncertain and the relative positions of residues that lie in random coil structures (e.g., in the pre-M1 and post-M3 regions) is unknown. In addition to residues that affect sensitivity to blockers, we identified several mutations that generate constitutively open NMDA channels. The largest effects were seen with mutations at T648 in M3, but mutations at Q556 and P557 in M1 and L657 in M3 also had effects (Fig. 10, dark blue circles). These mutants presumably affect gating of the channel. Residue T648 is near residue A653, which is equivalent to a residue in the 2 channel (an "orphan" channel related by sequence homology to glutamate receptors) at which an A-to-T mutation generates constitutively open channels responsible for the Lurcher mouse phenotype (Zuo et al., 1997). A-to-T mutations at NR1(A653) or NR2B(A652) do not generate constitutively open channels (K. Williams, unpublished observations), but the effects of the T648 mutants suggest that the M3 region in NMDA receptors is, like the equivalent region in 2, important for channel gating.

Many, but not all, of the residues identified in the current study that influence channel blockers were previously shown to be exposed in the channel lumen (Kuner et al., 1996; Beck et al., 1999). The side chains of residues that are exposed to the lumen could interact directly with channel blockers, forming part of their binding sites. It is possible that residues N650, Y647, and possibly A653 in NR1, and Y646 and N649 in NR2B interact directly with MK-801 and TB-3-4 (Fig. 10). At least some of these residues are solvent exposed and positioned at a level such that MK-801 and TB-3-4 could interact with their side chains and simultaneously interact with the N residues in M2. In particular, NR1(N650) may bind to an amino group of TB-3-4 because an N-to-A mutation reduces the affinity for TB-3-4, whereas an N-to-D mutation increases affinity. The roles of residues at the top of M1 and M3 and in the pre-M1 and post-M3 regions are more difficult to interpret. Mutations at residues NR1(W563) and NR2B(W559) had pronounced effects on block by MK-801 and TB-3-4, but these residues were reported to not be solvent exposed (Beck et al., 1999). We found that an aromatic residue (W, Y, or F) at these positions was important for block by MK-801 and TB-3-4. The aromatic residue may be important for stabilizing the channel structure and/or stabilizing a binding site within the channel. Also, it seems unlikely that MK-801 (drawn to scale on Fig. 10) could interact directly with residues at the top of the M1 and M3 helices while simultaneously interacting with the M2 asparagines, although a direct interaction of TB-3-4 with these residues is conceivable.

Clearly, there are many residues that affect block by MK-801 but do not affect block by memantine (Fig. 10, green circles). Memantine and MK-801 have a similar overall size and both are influenced by the critical asparagines in M2. It is unlikely that differences in the effects of the M1, M3, and M4 mutants can be explained on the basis of the sizes of MK-801 versus memantine. However, MK-801 contains benzyl rings and is more likely to make hydrophobic interactions with aromatic residues in the pore. It is possible that some residues that influence MK-801 (but not memantine) do so by altering access of the blocker to its binding site or by allosteric disruptions of the binding site. It is also possible that some mutants that influence two or more antagonists do so indirectly or allosterically rather than by directly altering side chains (or the positioning of backbone peptide groups) within the binding site. However, the mutations at these positions do not simply have general nonspecific effects on the pore structure, evidenced by their lack of effect on block by memantine and Mg²⁺ and their lack of effect on Ba²⁺ permeability.