JBC Avanti Polar Lipids

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published In Press as doi:10.1074/jbc.M610077200 on March 2, 2007

J. Biol. Chem., Vol. 282, Issue 17, 12773-12784, April 27, 2007
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
282/17/12773    most recent
M610077200v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ahmed, A. H.
Right arrow Articles by Oswald, R. E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ahmed, A. H.
Right arrow Articles by Oswald, R. E.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Dynamics of the S1S2 Glutamate Binding Domain of GluR2 Measured Using 19F NMR Spectroscopy*

Ahmed H. Ahmed{ddagger}, Adrienne P. Loh{ddagger}§, David E. Jane, and Robert E. Oswald{ddagger}1

From the {ddagger}Department of Molecular Medicine, Cornell University, Ithaca, New York 14853, §Department of Chemistry, University of Wisconsin, La Crosse, Wisconsin 54601, and Department of Pharmacology, MRC Centre for Synaptic Plasticity, School of Medical Sciences, University of Bristol, Bristol BS8 1TD, United Kingdom

Received for publication, October 27, 2006 , and in revised form, January 16, 2007.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
Ionotropic glutamate receptors mediate the majority of vertebrate excitatory synaptic transmission. Although the structure of the GluR2 binding domain (S1S2) is well known (agonist binding site between two lobes), little is known about the time scales of conformational transitions or the relationship between dynamics and function. 19F NMR (19F-labeled tryptophan) spectroscopy was used to monitor motions in the S1S2 domain bound to ligands with varying efficacy and in the apo state. One tryptophan (Trp-671) undergoes chemical exchange in some but not all agonists, consistent with µs-ms motion. The dynamics can be correlated to ligand affinity, and a likely source of the motion is a peptide bond capable of transiently forming hydrogen bonds across the lobe interface. Another tryptophan (Trp-767) appears to monitor motions of the relative positions of the lobes and suggests that the relative orientation in the apo- and antagonist-bound forms can exchange between at least two conformations on the ms time scale.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
Ionotropic glutamate receptors (GluRs)2 mediate the majority of excitatory synaptic transmission in the central nervous system of higher vertebrates (1) and play important roles in the formation of synaptic plasticity underlying higher order processes such as learning and memory as well as in neuronal development (2). In addition, ionotropic GluRs have been implicated in various neurodegenerative disorders such as Parkinson and Alzheimer diseases, Huntington chorea, and neurologic disorders including epilepsy and ischemic brain damage. Antagonists of glutamate receptors have been shown to limit tumor growth in a variety of human tumors and to inhibit tumor cell migration (3). In recent years many advances in characterizing the relationship between ionotropic GluR structure and function have been made. Ionotropic GluRs are membrane-bound receptor ion channels composed of multiple subunits arranged as a rosette, forming a central ion channel in which each subunit contributes to pore formation. Individual subunits are categorized by pharmacological properties, sequence, functionality, and biological roles into those that are sensitive 1) to the synthetic agonist N-methyl-D-aspartic acid (NR1, NR2A-D, NR3A-B), 2) to the synthetic agonist {alpha}-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA; GluR1–4), and 3) to the naturally occurring neurotoxin kainate (GluR5–7, KA1,2).

The structural analysis of glutamate receptors was dramatically advanced by the finding that the extracellular agonist binding domain (S1S2 domain) could be expressed in isolation as a soluble protein (4) and by the subsequent solution of the crystal structure bound to kainate (5). As predicted by homology models (6), the S1S2 domain was found to be a bilobed structure with the agonist binding pocket located between the two lobes. The first glimpse of the structural basis of channel activation was the finding by Armstrong and Gouaux (7) that the apo state exhibited a considerably larger angle between the two lobes than the full agonist-bound state, and at least for kainate relative to full agonists, the degree of lobe closure correlates with agonist efficacy. Furthermore, the dimer interface, and in particular the stability of the interface, seems to control desensitization. That is, the dissociation of the dimer interface leads to desensitization (8, 9), suggesting that the receptor works as a dimer of dimers, with Lobe 2 moving relative to Lobe 1 to promote channel activation (9). Although the simplistic idea would be that the degree of lobe closure dictates the single channel conductance, numerous studies have shown that multiple conductance levels are present with both partial and full agonists (10, 11) and that the conductance levels are independent of the efficacy of the agonist (12). However, at saturating concentrations, high efficacy agonists preferentially exhibit the higher conductance levels, whereas partial agonists preferentially populate the lower conductance levels. This has led to the suggestion that other dynamic processes either within the S1S2 domain or downstream of the agonist binding domain are involved in controlling the channel gate (13).

Although the availability of numerous crystal structures has provided many clues to the details of the coupling between binding and channel opening, little is known about the time scales of conformational transitions. This is particularly important since ligand binding and/or the control of channel function may be related to dynamic processes within S1S2. 19F NMR spectroscopy was employed as a monitor of conformational transitions in the GluR2 S1S2 domain in the presence of a variety of ligands with varying efficacy. We find that two of the four tryptophans in S1S2 (labeled with 19F) sense motions that may be related to conformational transitions associated with binding and lobe closure. Estimates of the time scales of these motions suggest that contacts across the interlobe interface can exchange on the order of µs-ms for the glutamate-, bromowillardiine-, and iodowillardiine-bound forms and that chemical exchange, likely related to dynamics of lobe closure, can occur on the time scale of ms for the apo form and the antagonist-bound form.


   EXPERIMENTAL PROCEDURES
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
RESULTS
 DISCUSSION
 REFERENCES
 
Materials—5-Fluorotryptophan was obtained from Sigma and Fisher. Glutamate was obtained from Sigma. Quisqualate, aniracetam, AMPA, and CNQX were obtained from Tocris. The willardiine compounds were synthesized as described previously (41). The GluR2 S1S2J construct was obtained from Eric Gouaux (Vollum Institute; Ref. 7).

Protein Preparation and Purification—The S1S2 protein was prepared as described by Furukawa and Gouaux (42) with modifications to incorporate fluorotryptophan as described by Salopek-Sondi and Luck (43). Briefly, wild type and mutant pET-b22(+) plasmids were transformed in Escherichia coli strain Origami B (DE3) cells and were grown at 37 °C to A600 of 0.8 in LB medium supplemented with the antibiotic. Cells were harvested, washed with M9 medium, and resuspended in M9 medium supplemented with 2 mM MgSO4, 0.2% NH4Cl, 0.1 mM CaCl2, 0.04% glucose, 1% casamino acids, 0.1% thiamine, 1% glycerol, 1 g of 19F-labeled Trp, and the antibiotics. The cultures were incubated at 37 °C to exhaust the residual Trp in the medium for 1 h (or reach A600 between 0.9 and 1.0). The cultures were cooled to 20 °C, and isopropyl-beta-D-thiogalactoside was added to a final concentration of 0.5 mM. Cultures were allowed to grow at 20 °C for 20 h. The cells were then pelleted, and the S1S2 protein was purified using a nickel nitrilotriacetic acid column followed by a HT-SP-Sepharose column (Amersham Biosciences) connected to a fast protein liquid chromatography system. Glutamate (10 mM) was maintained in all buffers throughout purification. For NMR experiments, the protein was concentrated to 0.2–0.5 mM using a Centricon 10 centrifugal filter (Millipore, Bedford, MA). Ligands were exchanged by multiple steps of dilution and centrifugation in the Centricon filter until no peaks characteristic of the glutamate-bound form were observed in the NMR spectrum. Glutamate and kainate were used at 10 mM, and all other ligands were used at 2 mM.

To produce mutants in which the tryptophans were replaced by phenylalanine leucine or valine, two primers for each mutant were custom-synthesized by Integrated DNA Technologies (Coralville, IA). For Trp-463 and Trp-671, site-directed mutagenesis employed the QuikChange mutagenesis kit (Stratagene, La Jolla, CA). The site-directed mutagenesis for residues 766 and 767 was carried out according to Ho et al. (44). The mutant sequences were confirmed by DNA sequencing at the BRC Biotechnology Resource Center at Cornell University (Ithaca, NY).

NMR Spectroscopy—All measurements were made on Varian Inova 500 and 600 spectrometers (470 and 564 MHz for 19F). 19F measurements were made using a broadband probe (at 470 MHz) or a triple resonance gradient probe (at 564 MHz) with the proton channel down-tuned to fluorine. Unless otherwise indicated, spectra were collected at 25 °C. Spectra were processed using NMRPipe Version 1.6 (45). The first five points were back-predicted using linear prediction to remove the background signal from the probe. The data were apodized with a mixed exponential-Gauss window function and zerofilled to double the original number of data points before Fourier transform. For quantitation, the peaks were fit with Gaussian functions. 19F T1 and T2 measurements were obtained by acquiring one-dimensional spectra obtained with standard Varian pulse sequences. The T2 measurements were made with a standard CPMG echo sequence (46), and the delay was set to 0.2, 0.5, 1, or 2 ms. If chemical exchange was present and the exchange rate is in the time scale of the delays, then one would expect the T2 to vary with CPMG delay (16). Although the protein concentration was well below the dimerization dissociation constant (6 mM, 9), the monomeric form of the protein was verified in the NMR tube using diffusion measurements which employed pulsed field gradients (21). Incorporation of 19F into each of the four sites was equal to within 10% as measured by integration of the peaks in the AMPA-bound spectrum.

Analysis of Exchange Data—The data from the CPMG experiments for Trp-671 were analyzed using the general expression for the transverse relaxation rate for two-site exchange, which is applicable to all chemical exchange time scales (47–49).

Formula 1(Eq. 1)
where

Formula 1(Eq. 2)

Formula 1(Eq. 3)

Formula 1(Eq. 4)

Formula 1(Eq. 5)
and R02A and R02B refer to the exchange-free transverse relaxation rates of the two states (A and B), pA and pB are the populations of the two states, and kex is the exchange rate for a two-site model. pA and pB were assumed to be 0.5, although varying the ratio had little effect on the rates. R02A and R02B were assumed to be identical and were tested for a range of values approximating 1/T2 for Trp-766, Trp-767, and Trp-460, which did not appear to undergo chemical exchange. The fitting procedure was developed in the laboratory and used a Simplex algorithm to optimize the fit to the data.

For the CNQX-bound form, the Trp-671 resonance showed clear decoalescence at lower temperatures The time constant for exchange ({tau}) at the decoalescence temperature can be estimated from the frequency splitting ({nu}A - {nu}B) at the stopped-exchange limit using Equation 6.

Formula 6(Eq. 6)
A more detailed analysis of the line shape as a function of temperature allows for extraction of rate constants for exchange using a slightly simplified version of the McConnell equation (25),

Formula 6(Eq. 7)

Formula 6(Eq. 8)

Formula 6(Eq. 9)

Formula 6(Eq. 10)
where K is a constant describing maximum amplitude, {tau} is the average time in seconds between isomerizations (= 1/k), {nu}A and {nu}B are the frequencies (in Hz) of the two resonances in the absence of exchange (the "stopped exchange limit"), and T2 is the spin-spin relaxation time. To maintain the Lorentzian line shape, the data were apodized with a matched exponential before Fourier transform. The line-shape analysis was performed using IGORPro (Wavemetrics, Portland, OR). The peak for Trp-671 was fit to a Lorentzian line shape and subtracted from the spectra before fitting Trp-767. The splitting between peaks was similar between 3 and 7 °C, so that the data for 3 °C was assumed to be stopped exchange (in the stopped exchange limit, {tau}->{infty}). To determine the values of {nu}A and {nu}B, the 3 °C spectrum was fit by fixing {tau} at values of 1, 0.1, and 0.03 s while allowing the other parameters to vary. Similar results for {nu}A and {nu}B were obtained from all three cases, with {nu}A - {nu}B = 98 Hz. For the higher temperature data, {nu}A and {nu}B were held fixed, whereas {tau} and the remaining parameters were allowed to vary. An additional parameter ({Delta}{nu}) was used in the higher temperature fits to account for chemical shift changes as a function of temperature. The Arrhenius plot was prepared from the resulting best-fit values of {tau} using k = 1/{tau}. Errors in the y axis (ln k) are given by

Formula 6(Eq. 11)
where {sigma}{tau} is the error in {tau} from the fit. Values of T2 from the fits were consistent with those measured directly (Table 1) and were weakly temperature-dependent.


View this table:
[in this window]
[in a new window]

  		TABLE 1
Relaxation parameters (ms) All values are given in ms. Except where otherwise noted, all measurements were made at 25 °C. The values in parentheses after T2 refer to the CPMG delay (see "Experimental Procedures"). Trp-767 for the CNQX-bound form showed decoalescence and could not be fit by a single Gaussian.

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
DISCUSSION
 REFERENCES
 
Assignments—There are four Trp residues in wild type S1S2 (Fig. 1); one in Lobe 1 (Trp-460), one in Lobe 2 (Trp-671), and two in helix K (Trp-766, Trp-767) near the C terminus. Wild type S1S2 was prepared with 5-fluoro-Trp (19F-labeled Trp) amino acids substituted for natural Trp. Assignment of the 19F-labeled resonances was achieved using a combination of mutations and experiments using line-broadening agents. Four mutant constructs, each with a single 19F-labeled Trp -> Phe mutation, were initially prepared. Only two mutants (W460F and W671F) produced well folded protein. Spectra of these mutants in the glutamate-bound form are shown in Fig. 2A, where the missing 19F signal relative to wild type corresponds to the mutated Trp residue. The W766F and W767F point mutations in helix K failed to produce well folded protein. However, W766V did produce a stable, well folded protein, but chemical shift changes precluded its use in assignments. No mutations of Trp-767 produced useable protein. Therefore, to assign the two 19F-labeled resonances from helix K, the differential exposure of the fluorine atoms to water was exploited. The fluorine atom of Trp-766 is solvent-exposed, whereas the fluorine atom of Trp-767 is completely buried. Thus, the addition of a paramagnetic substance such as Gd-DTPA should result in preferential broadening of the Trp-766 resonance over the Trp-767 resonance. Spectra of wild type 19F-labeled are shown in Fig. 2B as a function of Gd-DTPA concentration. Only one resonance broadens significantly; this is assigned to Trp-766. The remaining unassigned resonance is then assigned to Trp-767. The Trp-671 resonance also broadens somewhat more than those of Trp-460 and Trp-767, as its fluorine is slightly more solvent-exposed. A similar procedure was used to assign resonances in the presence of other glutamatergic ligands.

Full Agonists—19F resonances are sensitive reporters of protein structure and dynamics (14). In particular, the 19F chemical shift is very sensitive to changes in the electronic environment due to nearby hydrogen bonds, electrostatic fields, and van der Waals contacts (15), whereas the line-shape and relaxation parameters are good reporters of subtle changes in local dynamics. The fluorine spectra at 470 MHz of 19F-labeled S1S2 bound to AMPA, quisqualate, and glutamate are shown in Fig. 3, A–C. The substitution of quisqualate or glutamate for AMPA (Fig. 3, A versus B and C) results in a significant up-field change (~0.7 ppm) in chemical shift for Trp-767 relative to the other resonances. This is most likely because of the orientation of the isoxazole group of AMPA, which occupies subsite G, leading to the rearrangement of side chains (Tyr-405 and Met-708) in proximity to Trp-767 (7).

The line widths for Trp-460, Trp-766, and Trp-767 are similar regardless of the bound agonist, but that of Trp-671 is broadened only in S1S2 glutamate (Fig. 3, C and E). This broadening relative to the AMPA- and quisqualate-bound forms is consistent with the measured T2 values (Table 1), which are ~40% lower for the glutamate-bound than for the quisqualate- and AMPA-bound forms. The line-shape and T2 differences between the glutamate-bound and the quisqualate- and AMPA-bound forms suggest a change in dynamics of either the tryptophan itself or the environment of the tryptophan. To explore this further, the T2 was measured using a CPMG sequence with delays ranging from 0.2 to 2 ms. If exchange occurs on this time scale, then the measurement of T2 should vary with the delay (16). As shown in Table 1 and Fig. 3F, no systematic variation in T2 was observed at 25 °C. An alternative explanation is that the line broadening is because of processes on the ns-ps time scale. Given the large contribution of chemical shift anisotropy (CSA) to 19F relaxation, line broadening could arise from variations in the CSA on this faster time scale. If, however, the line-broadening arises from chemical exchange, variation in T2 as a function of CPMG delay could possibly be seen if the exchange rate slows sufficiently at a lower temperature. For that reason, the relaxation experiments were repeated at 15 and 5 °C. As shown in Table 1 and Fig. 3F, the T2 values decrease as expected for all four resonances at the lower temperatures. However, the value of T2 for Trp-671 at both 15 and 5 °C decreases as a function of CPMG delay. This behavior is consistent with chemical exchange at or near Trp-671 rather than with processes occurring on the ns-ps time scale. Although calculation of precise kinetic parameters is difficult from these data (due largely to the relatively rapid relaxation of the 19F nucleus in this protein), the time scale of motion can be estimated by assuming a two-site exchange model between two putative conformers. Using the Equation 1 ("Experimental Procedures") for two-site exchange, which is valid in all time scales, and assuming equal populations (fits were only marginally sensitive to changing the relative populations), the time constant for the exchange would be ~0.02 ms at 25 °C, increasing to 0.1 ms at 15 °C and 1.8 ms at 5 °C. To the extent that these assumptions are valid, this would suggest a very high activation energy for exchange (on the order of 40 kcal/mol). In any event, the variable dependence of T2 on CPMG delay is consistent with the presence of chemical exchange on the order of 10s of µs to ms, depending upon the temperature.


Figure 1
View larger version (41K):
[in this window]
[in a new window]

  		FIGURE 1.
A, structure of the GluR2 S1S2 domain bound to glutamate (PDB code 1ftj 7). Tryptophans are shown in blue using a space-filling format. The helices are labeled as described by Armstrong et al. (5), and the figure was created using PyMol (50). B, backbone representation of the peptide flip region showing the three copies of the glutamate-bound S1S2 in the asymmetric unit. Yellow is the unflipped protomer A, blue is the intermediate protomer B, and green is the flipped protomer C. C, the two additional hydrogen bonds formed when the peptide bond is flipped is illustrated. A water molecule mediates the hydrogen bond between Tyr-450 and Asp-651.

 
Given that the line broadening in the glutamate-bound form is very likely because of chemical exchange, the next question would be the origin of the exchange process. One possible candidate is a flip of the peptide backbone described originally by Armstrong and Gouaux (Ref. 7; see Fig. 1, B and C). Relative to the apo form, all crystal forms of AMPA bound S1S2, and four of the five protomers of quisqualate-bound S1S2 show the peptide bond between Asp-651 and Ser-652 flipped by 180°. This conformational change makes it possible to form two additional hydrogen bonds across the interface between the two lobes (Fig. 1C). On the other hand, of the three copies in the asymmetric unit of the glutamate-bound form, one is flipped, one is unflipped, and one is intermediate (Fig. 1B). This suggests that a dynamic equilibrium exists at least in the glutamate-bound form. In a study of the backbone NH dynamics of S1S2 bound to quisqualate, AMPA, and glutamate, Valentine and Palmer (17) found that exchange on the ms time scale occurred near the flip region and that exchange was less evident in the AMPA- and quisqualate-bound forms than in the glutamate-bound form. Given these findings, our initial hypothesis is that the broadening of the Trp-671 line shape is a result of the peptide flip and that the flip occurs on a time scale of 0.02 to 2 ms in the glutamate-bound form.


Figure 2
View larger version (32K):
[in this window]
[in a new window]

  		FIGURE 2.
The strategy for assigning the 19F resonances. A, Trp-460 (top spectrum) and Trp-671 (middle spectrum) were mutated to phenylalanine. Only minor changes in the spectra were observed so that the peak that is lost can be assigned to the mutated residue. B, gadolinium is paramagnetic and increases relaxation in resonances for which it comes in close proximity. Gd:DTPA selectively broadens the resonance at –47.6 ppm. Because the crystal structure of the glutamate-bound form of S1S2 (PDB code 1ftj (7)) indicates the Trp-766 is the tryptophan most exposed to solvent, this resonance was assigned to Trp-766. The remaining resonance at –48.2 ppm was assigned to Trp-767.

 
Hydrogen bonding across the interface between the two lobes has been postulated to affect the dissociation rate of the ligand (7, 18). If this is true, then agents that slow deactivation (a process reflecting channel closure and/or dissociation of the ligand from the active state) may affect the dynamics of the peptide flip, which should in turn affect the line width of the Trp-671 signal in the presence of glutamate. Aniracetam has been shown to slow deactivation (as well as having a small effect on desensitization) in the flop form of GluR2 (19) and has similar effects on GluR1 (20). As shown in Fig. 3D, the line width of Trp-671 narrows in the presence of 2 mM aniracetam, consistent with a measured increase in the T2 value (Table 1). No changes were observed in any of the other fluorine resonances. These observations are consistent with decreased dynamics near Trp-671 and presumably the peptide flip region. Although aniracetam binds at the dimer interface and can marginally promote the dimerization of the GluR2 S1S2 domain (19), the effect here is presumably on the monomer as the T2 values of the fluorine resonances were not decreased, and gradient diffusion measurements (21) indicated no change in the diffusion rate of the protein (data not shown). To explore the dynamics of Trp-671 further, a series of partial agonists was investigated that is expected to differ in the degree and perhaps kinetics of the peptide flip conformational transition.

Partial Agonists—Partial agonists are characterized by lower integrated currents at saturating concentrations, suggesting a lower efficacy. Armstrong and Gouaux (7) showed that S1S2 bound to the partial agonist, kainate, exhibits a smaller degree of lobe closure than when bound to full agonists. Likewise, the structures of the willardiine derivatives (partial agonists) seem to exhibit a smaller degree of lobe closure (particularly with iodowillardiine) than full agonists (Refs. 12 and 22; Table 2). The 19F spectra for S1S2 bound to four partial agonists (kainate, fluorowillardiine, bromowillardiine, iodowillardiine) are shown in Fig. 4, A–D (willardiine was also tested, but peak overlap of Trp-671, Trp-766, and Trp-767 precluded further analysis.) The peak for Trp-767 shows the largest changes in chemical shift across the series, moving up-field with increasing electronegativity and size of the halogen on the willardiine derivative (Fig. 4, B–D). This is not surprising since the Trp-767 fluorine is ~10.5 Å from the willardiine halogen, whereas the other three fluorines are ~15 Å from the willardiine halogen. In addition to differences in chemical shift, Trp-671 is broadened in the presence of some (bromowillardiine and iodowillardiine) but not all of the partial agonists. Decreasing the temperature further broadens Trp-671 in the case of bromowillardiine- and iodowillardiine-bound S1S2 (Fig. 4E). In the presence of both partial agonists, the T2 value for Trp-671 is also decreased relative to the other three resonances (Table 1 and Fig. 4F), and the T2 value decreased with increasing CPMG delays. As noted above for glutamate, this is indicative of chemical exchange rather than dynamic processes on a faster time scale. Using the approach and assumptions described above for glutamate, this would suggest an exchange rate in the bromowillardiine-bound form of 0.05–0.08 ms and in the iodowillardiine-bound S1S2 of 0.08–0.12 ms. On the other hand, in the presence of kainate and fluorowillardiine, Trp-671 does not appear to be broadened by exchange (Fig. 4, A, B, and E), although in the case of fluorowillardiine, peak overlap precludes a detailed analysis. In the presence of kainate, T2 for Trp-671 is similar to that of the other fluorine resonances at all CPMG delays (Fig. 4F). Unlike the case with glutamate, no effect of aniracetam was observed in the line widths or T2 values for iodowillardiine (Table 1). Thus, partial agonists appear to vary in the dynamics of the peptide flip region, with the bromowillardiine- and iodowillardiine-bound forms clearly exhibiting motions in the µs-ms time scale and little or no motion detected for the kainate- and fluorowillardiine-bound forms.


View this table:
[in this window]
[in a new window]

  		TABLE 2
Crystal structures, binding affinity, and chemical exchange Lobe opening was measured relative to protomer C in the glutamate-bound crystal structure of S1S2 (PDB code 1ftj) using residues Ile-500 —Lys-506, Pro-632—Asp-728, and the Gly-Thr linker to superimpose Lobe 1 in each structure (12). Opening is given in degrees with the errors as S.D. Only one protomer was present in the cases where no S.D. is shown. KD values were taken from Glu, AMPA, quisqualate, kainate, DNQX (27) and fluoro-, bromo-, and iodowillardiine (12). PDB codes: glutamate (1ftj), AMPA (1ftm), quisqualate (zinc, 1mm7; no zinc, 1mm6), kainate (1fwo), fluorowillardiine (1mqi), bromowillardiine (zinc, 1my3; no zinc, 1mqh), iodowillardiine (zinc, 1my4; no zinc, 1mqg), DNQX (1ftl), apo (1fto). The peptide flip is shown as a question mark for fluorowillardiine because the one crystal structure is unflipped, but HD exchange studies3 suggest that the hydrogen bond, suggestive of the flipped conformation, is present in solution. Although the apo form is given as unflipped, as suggested by the two crystal structures, the dynamic nature of this form would suggest that an equilibrium between the flipped and unflipped forms cannot be ruled out.

 
Antagonist-bound and Apo Form—In contrast to full and partial agonists, the apo and antagonist forms exhibit a much larger degree of separation between the two lobes (7). Using the glutamate-bound structure as a reference (PDB code 1ftj [PDB] , C protomer) and determining the degree to which the lobes change orientation using DynDom (23, 24), the apo form (PDB code 1fto [PDB] ) is 19.4 ± 2.1° more open, and the DNQX-bound form (PDB code 1ftl [PDB] ) is more open by 16.3 ± 1.1° (Table 2). Fig. 5A shows the 19F spectra with CNQX in the binding site. In contrast to the agonist structures, Trp-767 is broadened relative to the other three resonances. Upon cooling, the broadened resonance is clearly resolved into two peaks, indicating the presence of two-site exchange. The time constant for exchange at each temperature can be determined by fitting the data to the McConnell equation (Ref. 25; see "Experimental Procedures"). Assuming stopped exchange at 3 °C, the time constant ranges from ~10 ms at 10 °C to 0.8 ms at 35 °C, with an activation energy of ~16 kcal/mol (Fig. 5C). These results agree with the estimate for the time constant for exchange (2.3 ms) that can be obtained at a decoalescence temperature of 25 °C by assuming a stopped exchange peak separation of 100 Hz (measured at 3 °C). Trp-767 faces directly into the cleft between the two lobes and interacts with Tyr-711 on helix I. Although this region of the cleft exhibits a smaller degree of translation relative to some other portions of Lobe 2 when the lobes open, helix I (and Tyr-711) does rotate relative to the position of Trp-767. This may suggest that the chemical exchange in Trp-767 is because of an opening and closing of the cleft on the ms time scale. However, the time scale and not the extent of the motion can be discerned from these measurements.


Figure 3
View larger version (34K):
[in this window]
[in a new window]

  		FIGURE 3.
19F NMR spectra of full agonists as a function of temperature. A, AMPA, B, quisqualate. C, glutamate. The spectrum of glutamate- and aniracetam-bound 19F-labeled is shown in D. Aniracetam decreases the line width of Trp-671, consistent with an increase in the T2 for this resonance (Table 1). E, line-widths of all four resonances for S1S2 bound to AMPA, quisqualate (quis), and glutamate. The line width was determined as the width at half-height of a Gaussian fit to the spectra. The line width for Trp-671 in S1S2 bound to glutamate is clearly broader than Trp-671 in the AMPA- and quisqualate-bound forms and broader than the other three resonances in all cases. F, dispersion curves showing the measured T2 as a function of the CPMG delay ({tau}cp) in S1S2 bound to glutamate. The data were fit to the general equation for the transverse relaxation rate given under "Experimental Procedures" (Equation 1).

 
In the apo form, the resonances of Trp-671 and Trp-767 are considerably more broadened than the corresponding resonances of the agonist- or antagonist-bound forms (Fig. 5B). In both cases, cooling the sample leads to further line broadening. In the case of Trp-767, the broad peak at 25 °C resolves into at least two peaks at lower temperatures. As in the case of CNQX, this is consistent with the opening and closing of the cleft between lobes on the ms time scale. The dynamics of Trp-671 are more difficult to interpret, as the crystal structure (7) has provided no evidence for the peptide flip in the apo form (although it may be present in solution and not in any of the crystal forms). The protein is also relatively unstable in the apo form relative to the ligand-bound forms, such that careful measurements of the state of the peptide flip in solution may be difficult or impossible.


Figure 4
View larger version (38K):
[in this window]
[in a new window]

  		FIGURE 4.
19F NMR spectra of partial agonists as a function of temperature. A, kainate. B, fluorowillardiine. C, bromowillardiine. D, iodowillardiine. E, line widths of all four resonances for S1S2 bound to kainate (kai), bromowillardiine, and iodowillardiine. Fluorowillardiine was not shown because of peak overlap. The line widths for Trp-671 in S1S2 bound to bromowillardiine and iodowillardiine are clearly broader than Trp-671 in the kainate-bound forms and broader than the other three resonances in all cases. F, dispersion curves showing the measured T2 as a function of the CPMG delay ({tau}cp) in S1S2 bound to kainate, bromowillardiine, and iodowillardiine. The data were fit as described in the legend to Fig. 3.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
REFERENCES
 
19F NMR studies of GluR2 S1S2 labeled with fluorotryptophan demonstrate differences in dynamics for different positions in the protein and in the presence of different ligands. In particular, the 19F studies may reflect the dynamics of a peptide flip in Lobe 2 (Trp-671) and the global motions of the two lobes (Trp-767). The challenge in all structural and dynamic studies is to relate the results to the function of the protein. The problem with using isolated domains is lack of a definitive functional assay in the same preparation. One approach is to use correlations between structure or dynamic processes in the S1S2 domain measured in the presence of different ligands with function induced by those ligands in the intact, membrane-bound receptor. Drawing on insights from crystal structures, this approach has shed considerable light on the mechanisms of activation (7, 12) and desensitization (9) of glutamate receptors. Nevertheless, the details of the coupling between ligand binding and channel activation remain elusive.

Coupling between Binding and Channel Opening—Even before the availability of the crystal structures of the S1S2 domain, hypotheses based on homology modeling proposed that the coupling between ligand binding and channel opening resulted from the closure of two lobes upon ligand binding, which would in turn produce the movement that leads to channel opening (26). The crystal structures confirmed that ligand binding leads to lobe closure (5, 7) and added the additional important observation that the degree of lobe closure is dependent upon the ligand. That is, in some cases partial agonists lead to partial lobe closure, and full agonists lead to full lobe closure (7). The simple notion of strong coupling between lobe closure and channel opening might suggest that partial agonists would have lower single channel conductances than full agonists, but this is complicated somewhat by the fact that an intact receptor has four binding domains (10). The agonist concentration dependence of the single channels suggests that binding to two subunits is required for a low conductance channel opening, and the occupation of three and four subunits leads to two higher conductance levels (10, 12). Full and partial agonists share the same three conductance levels, although partial agonists are less likely to populate the highest conductance level (12), suggesting that the degree of lobe closure does not directly determine the conductance level but, rather, influences which conductance levels are populated. That is, additional processes must be involved either within the S1S2 domain or outside of the domain (e.g. within the channel domain or perhaps the linkers between the binding and channel domains). Furthermore, the degree of lobe closure is not always a fixed quantity in partial agonism. For example, AMPA is a partial agonist for the L650T mutation of S1S2, but only one of five unique structures (three in the presence of zinc and two in the absence) is partially open, with the other four fully closed as observed for most full agonists (27). Also, Jayaraman and co-workers (35) have shown using fluorescence resonance energy transfer measurements that the L650T mutant of the AMPA-bound form of GluR2 S1S2 is more closed than would be expected from the efficacy of AMPA. In the case of the willardiines, no correlation exists between lobe closure and efficacy for structures determined in the presence of Zn (all are slightly more open than the structures in the presence of glutamate), but iodowillardiine does show a more open conformation in the crystals in the absence of Zn (Table 2; Refs. 12, 22, and 28) and has the lowest efficacy of the series.


Figure 5
View larger version (32K):
[in this window]
[in a new window]

  		FIGURE 5.
19F NMR spectra of GluR2 S1S2 bound to CNQX (A) and in the apo form (B) as a function of temperature. Trp-767 exhibits significant broadening and decoalescence into two peaks in both spectra. C, a Arrhenius plot of the rate constant for chemical exchange for the CNQX-bound form determined as described under "Experimental Procedures." The time constants for exchange were 0.83 ± 0.05 ms at 35 °C, 1.37 ± 0.05 ms at 30 °C, 2.11 ± 0.05 ms at 25 °C, 4.58 ± 0.3 ms at 20 °C, and 3.83 ± 0.16 ms at 15 °C. The fits at 10 °C were less well defined but centered at ~10 ms. D, an example of the fit of the data to the McConnell equation (20 °C). The data were first fit to a Lorentzian centered at the Trp-671 resonance, and the resulting fit was subtracted from the spectrum (Trp-671 subtracted). The Trp-671-subtracted spectrum was fit to the McConnell equation as described under "Experimental Procedures." The fits to Trp-671 and Trp-767 are shown as dashed lines. The peaks are somewhat broader in spectrum in D versus the corresponding spectrum in A because of the apodization, which was a mixed-Gauss window in A and a matched exponential in D.

 
Assuming that the S1S2 domain does control channel gating, several possible mechanisms can be envisioned. One possibility is that a strong coupling between lobe closure and channel gating does exist but that multiple degrees of lobe closure are present in the protein under physiological conditions and that only the fully closed lobe would lead to channel gating. This would suggest that lobe closure is not fixed in solution but is a dynamic equilibrium between two or more conformations. This notion is supported by the multiple degrees of lobe closure seen in the L650T mutant (27) and in the iodowillardiine crystals (12, 22). Another possibility would be that the coupling between lobe closure and gating is weak and that although lobe closure places the system in an enabled state, other dynamic processes control channel gating. The partial lobe closing observed with partial agonists would be less likely to trigger a gating event than the full closure of full agonists (13). Gating could be a result of dynamic processes within S1S2 or in other parts of the protein. For example, backbone NMR dynamics measurements (17, 29) have found that Lobe 2 is considerably more dynamic than Lobe 1, and this could be associated with channel gating. One specific possibility concerns motions near the linker region (i.e. the region that would attach to the channel domain). Some disorder is seen in crystal structures in this region (e.g. PDB code 1mqg), and NMR studies of backbone dynamics suggest µs-ms motions (17, 29).

The 19F NMR studies described here were motivated by the fact that tryptophans in S1S2 were located in positions of potential interest and that 19F is a particularly sensitive probe of changes in structure and dynamics. Trp-460 is in Lobe 1 near but not in the agonist binding site. Trp-671 is situated in a hydrophobic pocket in Lobe 2, the end of which composes the peptide flip region. Finally, Trp-766 and Trp-767 are near the C terminus. Trp-766 is oriented toward the solvent and perhaps would be less sensitive to changes in conformation. Trp-767, on the other hand, interacts with Tyr-711 in helix I, which can be seen as the pivot for lobe closure. The changes in helix I as a function of agonist are small relative to the remainder of Lobe 1, so that only substantial changes in lobe closure or dynamics are likely to change the Trp-767 resonance. The question was then whether the 19F NMR spectra would reveal protein motions (inferred from chemical exchange associated with line broadening) and if those motions could be correlated to functional processes. The ability to observe chemical exchange is dictated by the time scale of the chemical exchange and by the difference in chemical shift between two or more environments, but observable time scales would tend to be on the order of µs-ms. The Trp-460 and Trp-766 resonances showed small chemical shift changes with different agonists, but the line widths were largely invariant, suggesting little or no ms motion relative to the remainder of the protein. In contrast, the Trp-767 and Trp-671 resonances did exhibit ligand-specific changes that suggest motions which may be associated with functional processes. These are discussed in the sections below.

Trp-671 and Ligand Binding Affinity—Line broadening arising from chemical exchange is a result of movement between two or more distinct chemical environments. This can be because of the movement of the probe (in this case, the 19F atom on Trp-671) or to fluctuations of the local environment. As shown in Fig. 1A, Trp-671 is situated in a hydrophobic pocket that includes the peptide flip region (Asp-651 and Ser-652). NMR studies of backbone dynamics of the glutamate-bound form have provided evidence for chemical exchange for some residues in this pocket (17, 29). In agreement with the results presented here, quisqualate- and AMPA-bound forms exhibit considerably less motion than the glutamate-bound form in those residues (17). In the crystal structures of GluR2 S1S2 (7), the peptide bond is flipped in all copies of the AMPA-bound protein and in four of the five copies of the quisqualate-bound form (30). As shown in Fig. 1B, the peptide bond is flipped in one glutamate-bound structure, unflipped in another, and intermediate in a third (7). These findings suggest that the Asp-651–Ser-652 peptide bond is in dynamic equilibrium that depends upon the ligand in the binding site.

The crystallographic studies can only show what conformations are possible; the relative populations of the states cannot be determined because of the small sample size (e.g. three structures for glutamate), the possibility of distortion in the crystal, or the possibility that one conformation can crystallize better than another. Also, an estimate of the time scale of exchange between conformers cannot be extracted from the crystal structures. Nevertheless, the multiple peptide flip conformations seen in crystal structures with full agonists can be correlated with chemical exchange observed by NMR, in that both the backbone NH chemical exchange data and the line broadening in 19F-labeled 671 are more pronounced in the glutamate-bound form than in quisqualate- or AMPA-bound forms.

Similar comparisons can be made for S1S2 bound to partial agonists or antagonist. In the case of the willardiine derivatives, the crystal structures and 19F NMR data can be directly compared for only two of the willardiine-bound forms (bromo- and iodowillardiine), as only one structure is available for fluorowillardiine (unflipped conformation for Asp-651/Ser-652). Similar to the case of the glutamate-bound form, both bromowillardiine- and iodowillardiine-bound forms have flipped and unflipped conformations present in the crystal structures (Table 2; Refs. 12 and 22). In addition, both the bromowillardiine- and iodowillardiine-bound forms show line broadening, with exchange rates in the 0.05–0.12 ms range at 25 °C (Table 1 and Fig. 4F). Although it is difficult to draw conclusions from a single crystal structure for the fluorowillardiine form, preliminary hydrogen-deuterium exchange NMR data with fluorowillardiine bound to 15N-labeled S1S2 has detected what is presumably the water-mediated hydrogen bond between Asp-651 and Tyr-450 that is stable for greater than 5 min,3 suggesting that the flipped or an intermediate conformer predominates in solution. Furthermore, to the extent that it can be analyzed because of peak overlap, Trp-671 shows much less broadening in the fluorowillardiine-bound protein (Fig. 4B), probably because of a relatively stable flipped conformation in the solution state. Neither kainate (Fig. 4A) nor CNQX (Fig. 5A) shows line broadening or a decrease T2 for Trp-671, and none of the crystal structures for kainate (5, 7) or DNQX (an analog of CNQX 7) shows a flipped conformation. Thus, an apparent correlation exists between the presence of multiple peptide-flip conformations and dynamics observed by NMR.

Comparing these data with binding affinity, the KD values for quisqualate and AMPA are 3–8-fold higher than that for glutamate (27). The binding affinity for fluorowillardiine is 10-fold higher than for bromo- and iodowillardiine, and kainate and DNQX show the lowest affinities (Table 2). Although other factors such as the electrostatic and hydrophobic interactions in the binding pocket are also essential contributors to binding affinity, a possible interpretation would be that binding affinity correlates with the state of the peptide flip such that those ligands that stabilized the flipped conformation (with the associated hydrogen bonds across the lobe interface) have the highest affinity, those that show a more balanced equilibrium between the two forms have an intermediate affinity, and those that favor largely the unflipped conformation have the lowest affinity (Table 2).

The stability of the interface between the two lobes has been proposed previously as one determinant of ligand affinity (7, 18) and possibly efficacy (31). In addition to the peptide flip region described above, an interlobe hydrogen bond between Glu-402 and Thr-686 also seems to affect the stability of the interface (18), as mutations of Thr-686 increase deactivation. Weston et al. (32) engineered two mutations (S652D and A455E), which allowed for additional electrostatic interactions across the interface and produced a significant slowing of deactivation. Thus, the release of ligand from its binding site may be controlled in part by the stability of the contacts across the lobe interface. The Trp-671 19F NMR data in wild type appear to report on one part of the interface; the additional hydrogen bonds formed when the Asp-651–Ser-652 peptide bond is flipped. The time scale of the exchange between the flipped and unflipped forms based on the NMR data are on the time scale of 0.02 to 1.2 ms for the glutamate-bound form, depending upon the temperature. The question then becomes whether this relates to ligand dissociation. Aniracetam decreases the rate of receptor deactivation (19, 20), which is presumably a result of an effect on ligand dissociation and/or channel closing. In the case of glutamate, aniracetam decreases the line width and increases T2, suggesting that chemical exchange is diminished or changed to a different time scale. This supports the notion that the dynamic processes affecting the relaxation of Trp-671 are associated with ligand dissociation. Although in the crystal structure (PDB code 1al5) aniracetam binds at the dimer interface (19), the effects observed here were presumably on the monomer (or perhaps a transient dimer), consistent with the fact that aniracetam only weakly promotes dimerization (19) and that the effects on deactivation are observed at lower concentrations than the effects on desensitization (19, 33). As suggested by Lawrence et al. (33), aniracetam may bind to more than one site, one of which could be the site at the dimer interface and the second of which may exist even in the monomer. When tested, however, on the iodowillardiine-bound form, no effect of aniracetam was observed. Likewise, aniracetam did not affect deactivation of kainate-induced currents (33). At least in the case of the full agonist, glutamate, aniracetam may stabilize the interface between the two lobes, perhaps decreasing the exchange between the flipped and unflipped Asp-651/Ser-652 backbone.

Using intrinsic tryptophan fluorescence at 5 °C, Abele et al. (34) suggest that glutamate dissociation from the GluR4 S1S2 protein is relatively slow (7.6 s-1), with AMPA dissociation occurring at an even slower rate (0.06 s-1). However, in intact GluR2, the deactivation after removal of glutamate is much faster at room temperature (time constant of 1.3 ms, Ref. 18). AMPA and quisqualate both show deactivation kinetics considerably slower than glutamate (18). The slower rate measured by Abele et al. (34) may in part be a result of the difference in temperature and a difference in measured parameters between the fluorescence and whole cell experiments (deactivation is not strictly a measure of ligand dissociation) as well as the slower binding kinetics for the GluR4 versus GluR2 S1S2 domain (see the supplement to Ref. 35). One possibility is that the lobe interface is composed of more than one contact. The formation and dissociation of the hydrogen bonds of the peptide flip are just one part of the story. Other positions, such as the hydrogen bond between Glu-402 and Thr-686 also contribute to the stability of the interface. Thus, a likely explanation is that the peptide flip may occur on the µs-ms time scale, but the lobe opening would represent a less probable event since more than one set of contacts would have to be broken at the same time.

Trp-767 and Potential Global Motions of the Two Lobes—The side chain of Trp-767 is positioned near helix I which acts as a pivot with changes in lobe closure, showing very little change in the crystal structures with different ligands (7). Thus, it is not surprising that no clear evidence for chemical exchange is observed for any of the agonists or partial agonists (line widths are the same as those for Trp-460 and Trp-766 (Figs. 3E and 4E) as are the T2 values (Table 1 and Figs. 3F and 4F)). However, Trp-767 does show significant decoalescence in the presence of CNQX and in the apo form. Although these results do not address any dynamic changes in lobe closure for full or partial agonists, the CNQX and apo forms show at least two-site exchange on the ms time scale, suggesting a motion between two distinct conformations. The activation barrier to exchange for the CNQX motion was found to be 16 ± 2 kcal/mol, which is similar to the barrier to peptide bond rotation observed in N,N-dimethylacetamide (36). Thus, the binding of agonist presumably stabilizes at least the range of the motion of the protein. This is consistent with the isothermal titration calorimetric studies of Madden et al. (37) that suggest the binding of agonist is driven by enthalpy rather than by a favorable change in entropy. In the case of antagonist or in the apo form, the lobes would have a much greater degree of possible motion, likely exchanging between at least two conformers. It is important to note that these measurements can identify the time scale of the motion, but the exact nature of the conformers is less clear. For example, these two conformers could be the states observed in the apo and DNQX crystal structures (7) as well as a more closed state or even the hyperextended state recently described by Kasper et al. (38). Also possible is a twist rather than a lobe opening or closing (39) or a more local motion.

Summary—The 19F NMR studies presented here are consistent with a range of techniques that have provided evidence that the functional motions of GluR2 are more complex than a simple rigid body movement of one lobe relative to the other in the binding domain. Hydrogen bonds across the interface between the lobes contribute to the binding affinity. Also, the apo and antagonist-bound forms are flexible, possibly exhibiting multiple degrees of lobe closure that exchange in the ms time scale. The binding of agonist then stabilizes the binding domain in a closed lobe form (at least on time scales of ms and slower), presumably allowing other processes within the protein to lead to channel opening. Although arising from a completely different structure, a conformational wave similar to that proposed by Auerbach (40), could subsequently produce changes in the channel domain that allow the passage of ions.


   FOOTNOTES
 
* This work was supported by National Institutes of Health Grant R01-GM068935 and National Science Foundation Grant IBN-0323874. Travel funds were provided by the Abraham and Henrietta Brettschneider Oxford and Cornell Exchange Fund. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Back

1 To whom correspondence should be addressed: Dept. of Molecular Medicine, Cornell University, Ithaca, NY 14853. Tel.: 607-253-3877; Fax: 607-253-3659; E-mail: reo1{at}cornell.edu.

2 The abbreviations used are: GluR, glutamate receptor; AMPA, {alpha}-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid; CNQX, 6-cyano-7-nitroquinoxaline-2,3-dione; CPMG, Carr, Purcell, Meiboom, Gill (an NMR pulse sequence element); DNQX, 6,7-dinitroquinoxaline-2,3-dione; Gd-DTPA, diethylenetriaminepentaacetic acid, gadolinium (III); S1S2, extracellular ligand binding domain of GluR2; T1, longitudinal relaxation time; T2, transverse relaxation time. Back

3 M. Fenwick and R. E. Oswald, unpublished results. Back


   ACKNOWLEDGMENTS
 
We thank Prof. Linda Luck (SUNY Plattsburgh, NY) for suggesting the use of fluorine NMR and for providing insightful and valuable advice throughout all aspects of this study including the preparation of protein and interpretation of results. We also thank Ivan Keresztes and Anthony Condo of the Cornell Chemistry and Chemical Biology NMR facility for expert assistance and Prof. Eric Gouaux (Vollum Institute) for the GluR2 S1S2J construct. Prof. Linda Nicholson (Cornell), Prof. Linda Nowak (Cornell), Dr. Michael Fenwick (Cornell), Dr. Chris Ptak (Cornell), Kathy Partin (Colorado State University), and Vasanthi Jayaraman (University of Texas, Houston) provided useful discussions and important advice.



   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Dingledine, R., Borges, K., Bowie, D., and Traynelis, S. (1999) Pharmacol. Rev. 51, 7-61[Abstract/Free Full Text]
  2. Asztely, F., and Gustafsson, B. (1996) Mol. Neurobiol. 12, 1-11[Medline] [Order article via Infotrieve]
  3. Rzeski, W., Turski, L., and Ikonomidou, C. (2001) Proc. Natl. Acad. Sci. U. S. A. 98, 6372-6377[Abstract/Free Full Text]
  4. Arvola, M., and Keinänen, K. (1996) J. Biol. Chem. 271, 15527-15532[Abstract/Free Full Text]
  5. Armstrong, N., Sun, Y., Chen, G. Q., and Gouaux, E. (1998) Nature 395, 913-917[CrossRef][Medline] [Order article via Infotrieve]
  6. Sutcliffe, M. J., Smeeton, A. H., Wo, Z. G., and Oswald, R. E. (1998) Method Enzymol. 293, 589-620[Medline] [Order article via Infotrieve]
  7. Armstrong, N., and Gouaux, E. (2000) Neuron 28, 165-181[CrossRef][Medline] [Order article via Infotrieve]
  8. Armstrong, N., Jasti, J., Beich-Frandsen, M., and Gouaux, E. (2006) Cell 127, 85-97[CrossRef][Medline] [Order article via Infotrieve]
  9. Sun, Y., Olson, R., Horning, M., Armstrong, N., Mayer, M., and Gouaux, E. (2002) Nature 417, 245-253[CrossRef][Medline] [Order article via Infotrieve]
  10. Rosenmund, C., Stern-Bach, Y., and Stevens, C. F. (1998) Science 280, 1596-1599[Abstract/Free Full Text]
  11. Swanson, G. T., Kamboj, S. K., and Cull-Candy, S. G. (1997) J. Neurosci. 17, 58-69[Abstract/Free Full Text]
  12. Jin, R., Banke, T. G., Mayer, M. L., Traynelis, S. F., and Gouaux, E. (2003) Nat. Neurosci. 6, 803-810[CrossRef][Medline] [Order article via Infotrieve]
  13. Oswald, R. E. (2004) Adv. Protein Chem. 68, 313-349[Medline] [Order article via Infotrieve]
  14. Luck, L. A., and Falke, J. J. (1991) Biochemistry 30, 4248-4256[CrossRef][Medline] [Order article via Infotrieve]
  15. Zemsky, J., Rusinova, E., Nemerson, Y., Luck, L. A., and Ross, J. B. (1999) Proteins 37, 709-716[CrossRef][Medline] [Order article via Infotrieve]
  16. Luz, Z., and Meiboom, S. (1963) J. Chem. Phys. 39, 366-370[Medline] [Order article via Infotrieve]
  17. Valentine, E. R., and Palmer, A. G., III (2005) Biochemistry 44, 3410-3417[CrossRef][Medline] [Order article via Infotrieve]
  18. Robert, A., Armstrong, N., Gouaux, J. E., and Howe, J. R. (2005) J. Neurosci. 25, 3752-3762[Abstract/Free Full Text]
  19. Jin, R., Clark, S., Weeks, A. M., Dudman, J. T., Gouaux, E., and Partin, K. M. (2005) J. Neurosci. 25, 9027-9036[Abstract/Free Full Text]
  20. Partin, K. M., Fleck, M. W., and Mayer, M. L. (1996) J. Neurosci. 16, 6634-6647[Abstract/Free Full Text]
  21. Altieri, A. S., Hinton, D. P., and Byrd, R. A. (1995) J. Am. Chem. Soc. 117, 7566-7567[CrossRef]
  22. Jin, R., and Gouaux, E. (2003) Biochemistry 42, 5201-5213[CrossRef][Medline] [Order article via Infotrieve]
  23. Hayward, S., and Berendsen, H. J. (1998) Proteins 30, 144-154[CrossRef][Medline] [Order article via Infotrieve]
  24. Hayward, S., and Lee, R. A. (2002) J. Mol. Graph. Model. 21, 181-183[CrossRef][Medline] [Order article via Infotrieve]
  25. McConnell, H. M. (1958) J. Chem. Phys. 28, 430-431[CrossRef]
  26. Wo, Z. G., and Oswald, R. E. (1995) Trends Neurosciences 18, 161-168[CrossRef][Medline] [Order article via Infotrieve]
  27. Armstrong, N., Mayer, M., and Gouaux, E. (2003) Proc. Natl. Acad. Sci. U. S. A. 100, 5736-5741[Abstract/Free Full Text]
  28. Inanobe, A., Furukawa, H., and Gouaux, E. (2005) Neuron 47, 71-84[CrossRef][Medline] [Order article via Infotrieve]
  29. McFeeters, R. L., and Oswald, R. E. (2002) Biochemistry 41, 10472-10481[CrossRef][Medline] [Order article via Infotrieve]
  30. Jin, R., Horning, M., Mayer, M. L., and Gouaux, E. (2002) Biochemistry 41, 15635-15643[CrossRef][Medline] [Order article via Infotrieve]
  31. Mayer, M. L. (2006) Nature 440, 456-462[CrossRef][Medline] [Order article via Infotrieve]
  32. Weston, M. C., Gertler, C., Mayer, M. L., and Rosenmund, C. (2006) J. Neurosci. 26, 7650-7658[Abstract/Free Full Text]
  33. Lawrence, J. J., Brenowitz, S., and Trussell, L. O. (2003) Mol. Pharmacol. 64, 269-278[Abstract/Free Full Text]
  34. Abele, R., Svergun, D., Keinanen, K., Koch, M. H., and Madden, D. R. (1999) Biochemistry 38, 10949-10957[CrossRef][Medline] [Order article via Infotrieve]
  35. Ramanoudjame, G., Du, M., Mankiewicz, K. A., and Jayaraman, V. (2006) Proc. Natl. Acad. Sci. U. S. A. 103, 10473-10478[Abstract/Free Full Text]
  36. Reeves, L. W., Shaddick, R. C., and Shaw, K. N. (1971) Can. J. Chem. 49, 3683-3691
  37. Madden, D. R., Abele, R., Andersson, A., and Keinanen, K. (2000) Eur. J. Biochem. 267, 4281-4289[Medline] [Order article via Infotrieve]
  38. Kasper, C., Pickering, D. S., Mirza, O., Olsen, L., Kristensen, A. S., Greenwood, J. R., Liljefors, T., Schousboe, A., Watjen, F., Gajhede, M., Sigurskjold, B. W., and Kastrup, J. S. (2006) J. Mol. Biol. 357, 1184-1201[CrossRef][Medline] [Order article via Infotrieve]
  39. Holm, M. M., Lunn, M. L., Traynelis, S. F., Kastrup, J. S., and Egebjerg, J. (2005) Proc. Natl. Acad. Sci. U. S. A. 102, 12053-12058[Abstract/Free Full Text]
  40. Auerbach, A. (2005) Proc. Natl. Acad. Sci. U. S. A. 102, 1408-1412[Abstract/Free Full Text]
  41. Jane, D. E., Hoo, K., Kamboj, R., Deverill, M., Bleakman, D., and Mandelzys, A. (1997) J. Med. Chem. 40, 3645-3650