HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 276, Issue 41, 37821-37826, October 12, 2001

This Article Abstract Full Text (PDF) All Versions of this Article: 276/41/37821    most recent M106171200v1 Submit a Letter to Editor Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Madden, D. R. Articles by Jayaraman, V. Search for Related Content PubMed PubMed Citation Articles by Madden, D. R. Articles by Jayaraman, V. Social Bookmarking                      What's this?

Stereochemistry of Quinoxaline Antagonist Binding to a Glutamate Receptor Investigated by Fourier Transform Infrared Spectroscopy^*

Dean R. Madden, Shalita Thiran^§, Herbert Zimmermann, Jonathan Romm^§, and Vasanthi Jayaraman^§¶

From the ^§ Department of Chemistry, Marquette University, Milwaukee, Wisconsin 53233 and  Max Planck Institute for Medical Research, 69120 Heidelberg, Germany

Received for publication, July 3, 2001, and in revised form, July 29, 2001

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

The stereochemistry of the interactions between quinoxaline antagonists and the ligand-binding domain of the glutamate receptor 4 (GluR4) have been investigated by probing their vibrational modes using Fourier transform infrared spectroscopy. In solution, the electron-withdrawing nitro groups of both compounds establish a resonance equilibrium that appears to stabilize the keto form of one of the cyclic amide carbonyl bonds. Changes in the 6,7-dinitro-2,3-dihydroxyquinoxaline vibrational spectra on binding to the glutamate receptor, interpreted within the framework of a published crystal structure, illuminate the stereochemistry of the interaction and suggest that the binding site imposes a more polarized electronic bonding configuration on this antagonist. Similar spectral changes are observed for 6-cyano-7-dinitro-2,3-dihydroxyquinoxaline, confirming that its interactions with the binding site are highly similar to those of 6,7-dinitro-2,3-dihydroxyquinoxaline and leading to a model of the 6-cyano-7-dinitro-2,3-dihydroxyquinoxaline-S1S2 complex, for which no crystal structure is available. Conformational changes within the GluR ligand binding domain were also monitored. Compared with the previously reported spectral changes seen on binding of the agonist glutamate, only a relatively small change is detected on antagonist binding. This correlation between the functional effects of different classes of ligand and the magnitude of the spectroscopic changes they induce suggests that the spectral data reflect physiologically relevant conformational processes.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Glutamate receptor (GluR)¹ ion channels are the predominant mediators of excitatory synaptic signals in the central nervous system. Glutamate binding triggers the channels to open, permitting cations to flow down their electrochemical gradients, depolarizing the postsynaptic membrane and thereby stimulating the receiving cell. According to agonist affinity profiles, the GluR can be subdivided into three subfamilies: -amino-5-methyl-3-hydroxy-4-isoxazole propionate (AMPA), N-methyl-D-aspartate, and kainate receptors (1, 2). Recent large scale expression of the extracellular ligand binding domains of two AMPA receptor subunits, namely GluR2 and GluR4, has paved the way for a number of structural and spectroscopic investigations. This ligand binding domain, known as S1S2, retains ligand binding parameters quite similar to those of the intact receptor and hence serves as an excellent model to study the ligand-protein interactions of the complete molecule (3-6).

Crystal structures have been determined for the GluR2S1S2 protein in the apo form and in complex with kainate, glutamate, 6,7-dinitro-2,3-dihydroxyquinoxaline (DNQX), and AMPA (7). These structures indicate that the ligand binding domain is a bilobate structure and that the degree of cleft closure between the two lobes is related to the type of ligand bound to the protein (7). Full agonists such as glutamate and -amino-5-methyl-3-hydroxy-4-isoxazole propionate induce a 20° closure of the two domains relative to the apo form, whereas the partial agonist kainate induces only a 12° closure of the two domains (7). On the other hand, the antagonist DNQX-bound form does not exhibit any large conformational changes relative to the apo form (7). These results suggest that the degree of domain closure most likely controls the extent of channel activation, thus providing the first structural insight into the activation mechanism of the glutamate receptors.

The interactions of the ligands kainate and glutamate with GluR4-S1S2 have also been investigated using vibrational spectroscopy (8). The ability to assign vibrational modes to individual functional groups using site-specific isotopic labeling, coupled with the high inherent sensitivity of these vibrations to the electronic environments of the groups, permitted a detailed investigation of the interactions between each of the carboxylate moieties of the ligands with the protein. Furthermore, the amide I vibrations of the protein backbone, which are sensitive to the secondary structure of the protein, indicated larger and different changes in the protein on binding glutamate relative to kainate, consistent with the crystallographic data (7). The spectroscopic results extended the structural information by providing a more detailed picture of the strength of the important interactions between the ligand and protein and within the protein.

In this study, we performed similar investigations of the interactions of the antagonists DNQX and 6-cyano-7-nitro-2,3-dihydroxyquinoxaline (CNQX) with GluR4-S1S2 using vibrational spectroscopy. Because of the presence of electron-withdrawing nitro and cyano substituents, the cyclic amide groups of the free quinoxaline GluR antagonists exist in both delocalized (enol) and localized (keto) bonding configurations; interactions with the GluR binding site specifically stabilize one of these configurations. The DNQX vibrational frequency shifts also provide valuable information on the stereochemistry of the binding interactions and can be used to interpret the published crystal structure of the S1S2-DNQX complex (7). This complex in turn serves as a reference for modeling the interactions in the CNQX complex, which has not been structurally characterized but which we have analyzed by Fourier transform infrared spectroscopy.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Synthesis of Isotopes of DNQX and CNQX-- Fig. 1 shows the structures of DNQX, CNQX, and the isotopomers used in this study. The isotopomers of DNQX were synthesized using the following reaction scheme. Equimolar quantities of o-phenylenediamine and either oxalic acid or 1,2-¹³C-oxalic acid (Campro Scientific) were condensed in 4 N hydrochloric acid under reflux to 2,3-dihydroxyquinoxaline or 2,3-¹³C₂-2,3-dihydroxyquinoxaline, respectively. The recrystallized (ethanol) products were nitrated in concentrated H₂SO₄ using KNO₃ at temperatures increasing from 0 °C to room temperature. The mononitrated products were isolated and purified and subjected to a second nitration step under identical conditions. 2,3-Dihydroxyquinoxaline was ¹⁵N-labeled by two-step nitration with K¹⁵NO₃. The labeled and unlabeled 6,7-dinitro-2,3-dihydroxyquinoxalines were stirred in ethanol (absolute, twice), filtered, and dried. 7-¹⁵NO₂-CNQX was prepared by nitration of 6-cyano-2,3-dihydroxyquinoxaline in 98% H₂SO₄ with a large excess of K¹⁵NO₃ at 0 °C.

¹H NMR spectra (500 MHz, dimethylsulfoxide-d6) exhibited the expected resonances for the isotopomers of DNQX (S, 2H, 7.718 ppm; and S, (OH)₂, 12.448 ppm) and CNQX (S, 1H, 7.558 ppm; D, 1H, 8.038-8.043 ppm; S, OH, 12.432 ppm; and S, OH, 12.472 ppm). The chemical purity of the isotopomers was >99%. Mass spectra revealed the expected peak at m/z 254 for both 2,3-¹³C₂-DNQX and 6,7-(¹⁵NO₂)₂-DNQX, with 99 and 95% isotopic purity, respectively, and the expected peak at m/z 233 for 7-¹⁵NO₂-CNQX, with 95% isotopic purity.

Protein Preparation and Characterization-- The GluR4-S1S2 protein was expressed, purified, and characterized as described (9). In brief, S1S2 was expressed as a secreted construct in the baculovirus system. After clarification and concentration of the cell-culture supernatant, it was purified to homogeneity by immunoaffinity and ion exchange chromatography. The protein was concentrated and dialyzed to yield a final concentration of 0.25-0.5 mM in 25 mM phosphate buffer, pH 7.4, containing 250 mM NaCl and 0.02% NaN₃. Protein-bound spectra were obtained in the presence of saturating concentrations of DNQX and CNQX. Because water has a large infrared absorption band at ~1600 cm¹, D₂O was used as the solvent to obtain spectra in the 1450-1800 cm¹ region. However, for studying the CN stretching vibration that occurs at ~2200 cm¹, water was used as the solvent.

FTIR Difference Spectroscopy-- The FTIR spectra were measured on a Nicolet Magna 870 infrared spectrophotometer using an FTIR cell with CaF₂ windows and a 50-µm spacer. Spectra were collected at 4 cm¹ spectral resolution. A modified variable-length sample holder (Aldrich), which allowed liquid from a constant temperature bath to be circulated around the holder, ensured that the protein solutions were kept at a constant temperature of 15 °C. All spectra reported are difference spectra, and the subtraction was performed using the band at 2045 cm¹ arising from the sodium azide present in the buffer as an internal standard. Furthermore, for generating the difference spectra between the ligand-bound and apo forms of the protein, the peaks that arise from excess unbound ligand (DNQX, CNQX, and isotopomers thereof) were subtracted using the spectrum of free ligand.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Vibrational Modes of DNQX in Solution-- The FTIR spectrum of DNQX at pH 7.4 is shown in Fig. 1, trace A. Comparison of the spectrum to spectra of DNQX isotopomers labeled at the carbonyl (¹³CO-DNQX) or nitro (¹⁵NO₂-DNQX) positions (Fig. 1, traces B and C, respectively) reveals that the amide, nitro, and aromatic ring moieties of DNQX are the main infrared active chromophores in the 1450-1800 cm¹ region. The assignments are described in detail below and are summarized in Table I.

View larger version (33K): [in this window] [in a new window]   Fig. 1.   FTIR difference spectra versus D2O for DNQX (A), DNQX isotopomers labeled at the nitro (B; 15NO2-DNQX) and carbonyl (C; 13CO-DNQX) positions, and CNQX (D). E, FTIR difference spectrum between CNQX and an isotopomer labeled at the nitro position (CNQX-15NO2CNQX). Corresponding chemical structures of the antagonists and isotopomers are shown to the right of each trace. a.u., absorbance units; asym, asymmetric vibrations.

¹³C labeling at the two carbonyl positions of DNQX (¹³CO-DNQX) downshifts all the bands in the infrared spectra of DNQX in the 1450-1800 cm¹ region (Fig. 1, traces A and B) except for a band at 1542 cm¹, which corresponds to the asymmetric vibration of the nitro groups, as identified by ¹⁵N labeling (see below). The DNQX bands at 1690 and 1662 cm¹ (Fig. 1, trace A) are downshifted to 1647 and 1624 cm¹, respectively, in ¹³CO-DNQX (Fig. 1, trace B), indicating that they arise from amide I-type CO stretching vibrations of the DNQX cyclic amide groups (10). The presence of two amide I vibrations suggests that the carbonyl groups in free DNQX at pH 7.4 can adopt multiple bonding configurations, as illustrated in Fig. 2. The high and low frequency vibrations would thus correspond, respectively, to strongly (e.g. Fig. 2, keto group in Structure I) or weakly (e.g. Fig. 2, partial double bonds in Structures I and II) localized carbonyl double bonds. In the absence of electron-withdrawing substituents, the carbonyl moieties would be expected to be delocalized at neutral pH, but in DNQX the keto form is apparently stabilized by resonance structures (e.g. Fig. 2, Structure I) made possible by the presence of the electron-withdrawing nitro groups.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   FTIR difference spectra between DNQX and D2O at pH 7.4 (A), pH 10 (B), and pH 10 (C) divided by a factor of 5. Shown to the right are the DNQX solution bonding configurations proposed to account for amide I vibrational modes observed at pH 7.4 (Structures I and II) and at pH 10 (Structures III and IV). a.u., absorbance units; asym, asymmetric vibrations.

This interpretation is supported by the differential sensitivities of the 1690- and 1662-cm¹ bands to pH (Fig. 2). Increasing the pH from 7.4 to 10 should titrate the amide protons, as illustrated in Fig. 2, Structures III and IV, leaving the carbonyl groups with little or no double bond character. As predicted, at pH 10, no band is observed at 1690 cm¹, whereas a much weaker signal is obtained at 1662 cm¹ (Fig. 2, trace B). An accompanying pH-induced shift is observed in the frequency and intensity of the amide II band found at 1547 cm¹ in the pH 7.4 spectrum (Fig. 2, traces A and C), which has been identified on the basis of its frequency and ~34-cm¹ downshift on ¹³CO labeling (Fig. 1, trace B). The amide II shifts observed at pH 10 can be attributed to changes in the composition of the band from a hybrid CN stretching and NH bending mode (Fig. 2, Structures I and II), to a predominantly CN stretching vibration (Fig. 2, Structures III and IV, and Ref. 10).

The band at 1614 cm¹ in the spectrum of DNQX at pH 7.4 (Fig. 1, trace A) can be assigned to an aromatic ring-breathing mode on the basis of its frequency and its relatively small shift on ¹³CO labeling (Fig. 1, trace B), which presumably reflects coupling of the benzene and adjacent cyclic amide rings.

¹⁵N labeling of the DNQX nitro groups at pH 7.4 downshifts a shoulder on the band at ~1547 cm¹ (Fig. 1, trace A) to 1510 cm¹ (Fig. 1, trace C). The frequency of the shoulder peak in the unlabeled molecule is 1542 cm¹, as revealed by a shift of the overlapping amide II peak from 1547 to 1513 cm¹ on ¹³CO labeling (Fig. 2, trace B). On the basis of its frequency and ~30-cm¹ downshift on ¹⁵NO₂ labeling, the 1542-cm¹ band can be assigned to the asymmetric stretching vibrations of the DNQX nitro groups.

The resonance structure invoked above to explain the presence of two DNQX amide I vibrational modes (Fig. 2, Structure I) also implies the existence of two distinct bonding configurations of the nitro groups. However, only a single nitro vibrational mode is detected. One explanation is that there is little difference between the asymmetric vibrational frequencies of the two NO₂ bonding configurations and that the band observed at 1542 cm¹ represents a hybrid of two unresolved component frequencies. Consistent with this interpretation, an increase in pH from 7.4 to 10 causes the band to shift, but only by 5 cm¹ (data not shown), presumably reflecting selective stabilization of the uncharged form of the nitro moiety (Fig. 2, Structures III and IV).

Vibrational Frequencies of CNQX in Solution-- The structures of DNQX and CNQX are very similar, except that a single nitro group in DNQX is replaced by a cyano group in CNQX (Fig. 1). As a result, their spectra are also comparable (Fig. 1, traces A and D), and the assignments made by isotopic labeling of DNQX can be transferred to the CNQX spectrum. The small upshifts in the amide I, amide II, and aromatic ring vibrations of CNQX relative to DNQX (Fig. 1 and Table I) can be attributed to the fact that the CNQX cyano group is less electron-withdrawing than the corresponding DNQX nitro moiety.

Like DNQX, free CNQX in solution exhibits dual amide I vibrational modes at pH 7.4 (Fig. 1, trace D), presumably also reflecting the existence of alternative bonding configurations (Fig. 2, Structures I and II). In the case of CNQX, Structure I in Fig. 2 is resolved into two distinct structures by the nonequivalence of the 6 and 7 substituents (Fig. 3, Structures I and II). The amide I equilibrium should thus be coupled to the stretching vibration of the CNQX cyano group; the FTIR spectrum exhibits the expected two bands in the region of the CN stretching vibration (Fig. 3, trace A).

View larger version (15K): [in this window] [in a new window]   Fig. 3.   FTIR spectra of CNQX in H2O (A) and CNQX bound to GluR4 S1S2 protein (B). Shown above the spectra are the alternative CNQX bonding configurations proposed to account for the dual vibrational frequencies observed in A. These two configurations correspond to the single DNQX Structure I in Fig. 2 and reflect the molecular asymmetry of CNQX. a.u., absorbance units.

The frequency of the asymmetric stretching vibration(s) of the single NO₂ group of CNQX is revealed by the difference spectrum between CNQX and ¹⁵NO₂-labeled CNQX (Fig. 1, trace E). The positive band at 1536 cm¹ reflects asymmetric vibrations of the unlabeled NO₂ group. As for DNQX, this mode is likely to represent a hybrid of the component frequencies associated with the two different nitro group configurations shown in Figs. 2 and 3.

Vibrational Modes of DNQX in Complex with S1S2-- The FTIR difference spectrum between DNQX-bound protein and apo protein (Fig. 4, trace A) reflects changes in the vibrational modes of both DNQX and the protein. Difference spectra have also been obtained between complexes of protein with unlabeled DNQX and with DNQX isotopomers (Fig. 4, traces B and C). Isotopic labeling is essential not only to assign bands to specific DNQX functional group vibrations but also to detect bands that are masked in the unlabeled difference spectrum (e.g. Fig. 4, 1500-1550 cm¹) because of overlap between the protein and DNQX difference features. The assignments of the individual bound vibrational modes are described below and summarized in Table I.

View larger version (31K): [in this window] [in a new window]   Fig. 4.   FTIR difference spectra between DNQX-bound and unligated GluR4 S1S2 protein (A), DNQX-bound and 13CO-DNQX-bound GluR4 S1S2 protein (B), DNQX-bound and 15NO2-DNQX-bound GluR4 S1S2 protein (C), CNQX-bound and unligated form of GluR4 S1S2 protein (D), and CNQX-bound and 15NO2-CNQX-bound form of GluR4 S1S2 protein (E). All spectra were measured in D2O. a.u., absorbance units.

In the difference spectrum between DNQX bound to protein and ¹³CO-DNQX bound to protein (Fig. 4, trace B), a positive band at 1689 cm¹ is downshifted by ¹³CO labeling to yield a negative band at 1647 cm¹, indicating that it corresponds to an amide I vibration of bound DNQX. A second positive band at 1540 cm¹ is downshifted by isotopic labeling to 1506 cm¹, suggesting that it corresponds to an amide II vibration. Two small difference features are observed in the difference spectrum between DNQX-bound S1S2 and apo-S1S2 at 1619 and 1597 cm¹ (Fig. 4, trace A). On the basis of their frequency, these have been tentatively assigned to benzene ring vibrations of bound DNQX.

The difference spectrum between DNQX bound to protein and ¹⁵NO₂-DNQX bound to protein (Fig. 4, trace C) has two positive overlapping bands at 1545 and 1527 cm¹ which are downshifted by ¹⁵NO₂ labeling to 1514 and 1504 cm¹, respectively, suggesting that they can be assigned to the two NO₂ groups of the bound DNQX.

As shown in Table I, S1S2 binding clearly affects the vibrational spectra of the DNQX nitro and cyclic amide moieties. Although the difference between the two bonding configurations of the DNQX nitro groups is too small to be resolved in solution, their distinct interactions with the S1S2 binding site upshift one asymmetric stretching vibration slightly and downshift the other significantly. At the carbonyl groups, the interaction with S1S2 simultaneously stabilizes the high frequency (keto) bonding configuration and apparently eliminates the low frequency (enol) configuration observed in free DNQX. Interpreted within the framework of crystallographic data on the S1S2-DNQX complex (7), these spectroscopic data establish the electronic bonding configuration of the bound antagonist and provide the basis for a model of the S1S2-CNQX interaction, which has not yet been structurally characterized.

Environment of DNQX Nitro Groups in S1S2-- Comparing the dual frequencies of the nitro groups of bound DNQX (1545 and 1527 cm¹) with the corresponding single hybrid frequency (1542 cm¹) of the two nitro groups in free DNQX (Table I), it is evident that one of the bands is downshifted by ~15 cm¹. This indicates that one of the nitro groups has become more electron-withdrawing, has formed a hydrogen-bonding interaction with the protein, or both. The other band is slightly upshifted by ~3 cm¹ relative to the hybrid frequency, which would be consistent with a more negative electrostatic environment for the other nitro group in the binding pocket.

In the crystal structure of the DNQX-bound GluR2 ligand binding domain (7), two different complexes are observed in the asymmetric unit. In one complex, a sulfate ion is located in the binding pocket together with the DNQX molecule, causing small but significant rearrangements of both ligand and protein side-chain moieties relative to the second, sulfate-free complex. In the sulfate-bound complex, each DNQX nitro group makes a single weak hydrogen bond to S1S2 (3.4 Å to Tyr-733 and 3.5 Å to Thr-687, respectively, using GluR4 sequence numbering) and another hydrogen bond to a water molecule (2.7 and 3.2 Å, respectively), and each is 3.4-3.7 Å from the closest protein carboxylate moieties (Glu-403 and Glu-706). Approximately equivalent shifts would therefore be expected for the two groups relative to their (overlapping) frequencies in solution, in contrast to the unequal effects seen in the vibrational spectra (Table I).

In the sulfate-free complex (Fig. 5), the DNQX nitro group that is distal to the S1S2 hinge makes a single 3.0-Å hydrogen bond to Thr-687 in S1S2 and is distant (4.5 Å) from both protein carboxylates, so that a downshift of the observed magnitude would be expected for this group. The hinge-proximal nitro group makes hydrogen bonds to Tyr-733 (2.7 Å) and to a water molecule (2.6 Å) but is located 3.2 Å from the negatively charged carboxylate of Glu-403 and 4.3 Å from the Glu-706 carboxylate. The competing influences of these interactions would be consistent with the small upshift we observe for the other nitro group. Our data therefore suggest that the sulfate-free crystal structure more accurately reflects the DNQX complex formed in solution. Furthermore, the negative electrostatic environment (Glu-403 and Glu-706) of the hinge-proximal nitro group causes the hinge-distal nitro group to be more electron-withdrawing, thus favoring the specific DNQX bonding structure shown in Fig. 5.

View larger version (10K): [in this window] [in a new window]   Fig. 5.   Environment of DNQX in one of the subunits of the crystal structure of DNQX in complex with the GluR2 ligand binding domain (7).

Environment of the DNQX Cyclic Amide Carbonyl Groups in S1S2-- The negative charge on the hinge-distal nitro group imposes a keto configuration on the opposite, hinge-proximal carbonyl group (Fig. 5), consistent with the high frequency amide I band observed in the bound DNQX and CNQX spectra. The localized double bond at the hinge-proximal carbonyl group is also stabilized by the hydrogen-bonding interaction of Pro-479 with the hinge-proximal amide proton (Fig. 5). In parallel, the absence of an enol vibrational mode suggests that a negatively charged single-bonded configuration is induced at the hinge-distal carbonyl group by electrostatic interactions with the nearby positively charged Arg-486 guanidino moiety (2.8 Å). The structure stabilized by the sum of these interactions would therefore exhibit a single, high frequency amide I vibrational mode, as observed.

Vibrational Modes of CNQX in Complex with S1S2-- S1S2 binding has broadly similar effects on the distribution of CNQX vibrational frequencies as it does for DNQX. In the FTIR difference spectrum between CNQX-S1S2 and apo-S1S2 (Fig. 4, trace D), the positive bands at 1693 and 1623 cm¹ can be assigned in analogy to the corresponding bands at 1689 and 1619 cm¹ in the bound DNQX spectrum. The four-wave number upshift in these vibrations in bound CNQX versus DNQX (Fig. 4) mirrors the similar upshift in the corresponding frequencies of free CNQX versus DNQX caused by the weaker electron-withdrawing character of the CNQX cyano group (Table I). It thus appears that S1S2 binding stabilizes the same carbonyl bonding configuration in CNQX as it does in DNQX, with one single bond and one strongly localized double bond.

In the difference spectrum between CNQX bound to protein and ¹⁵NO₂-CNQX bound to protein (Fig. 4, trace E), there is one positive band at 1523 cm¹, which is downshifted to 1500 cm¹ by isotopic labeling of the NO₂ moiety. The band at 1523 cm¹ can therefore be assigned to the asymmetric vibrational frequency of the single NO₂ group of bound unlabeled CNQX. The cyano stretching vibrational mode of the bound CNQX (Fig. 3, trace B), on the other hand, exhibits a single peak that is upshifted by 4 cm¹ from the highest frequency band of free CNQX.

Structural Interpretation of CNQX-S1S2 Interactions-- The similar effects of S1S2 binding on the vibrational frequencies of DNQX and CNQX suggest a highly conserved stereochemical interaction of both antagonists with the binding pocket. Within this framework, the behavior of the vibrational frequencies of the CNQX cyano and nitro moieties permits us to identify their likely positions in the S1S2 ligand binding pocket. The 13-cm¹ downshift in the asymmetric vibration of the single CNQX nitro group is comparable with the 15-cm¹ downshift assigned to the hinge-distal nitro group of DNQX. As a result, the CNQX nitro group would be expected to carry a negative charge and to occupy the hinge-distal nitro binding site (Fig. 5). The resulting bonding configuration of S1S2-bound CNQX is also reflected in the behavior of the cyano stretching vibration, for which only a single peak is found, in contrast to the dual peaks observed in the free form (Fig. 3). The high frequency of this peak corresponds to the less conjugated form of the cyano group (Fig. 3, Structure I), consistent with the negative charge being located on the nitro group in the S1S2-bound state (Fig. 5). Furthermore, assuming that the nitro group occupies the hinge-distal position as postulated, the cyano group would have to be placed in the negative electrostatic environment of the hinge-proximal binding site, which would further localize the electrons and in addition decrease the conjugation between the CN group and the CNQX aromatic ring. Consistent with this prediction, the frequency of the cyano stretching vibration in the difference spectrum of bound CNQX, (2247 cm¹) is four wave numbers higher than the higher of the two frequency bands observed in free CNQX (2243 cm¹).

Protein Secondary Structural Changes-- Although most of the bands in the difference spectra of DNQX-S1S2 and CNQX-S1S2 versus apo S1S2 (Fig. 4, traces A and D) can thus be assigned to vibrational modes of the bound antagonist, this is not the case for the features at 1645 and 1644 cm¹, respectively. The frequency of these bands appears not to be shifted by isotopic labeling of the antagonists, nor is the CNQX frequency upshifted relative to the DNQX frequency, as seen for most antagonist vibrational frequencies. The frequencies and lack of agonist sensitivity of these bands suggest that they reflect the amide I vibrational mode of -helices in the protein (10, 11), and positive bands in the difference spectra thus indicate a modest increase in the -helical content of the S1S2 domain on binding of the quinoxaline antagonists. The increase appears to be similar for DNQX and CNQX, suggesting that both compounds induce similar, small conformational changes in the binding domain.

Compared with the modest effect of the antagonists DNQX and CNQX, much more extensive changes in the vibrational spectra of GluR4 S1S2 have been described on binding of full and partial agonists (8). The full agonist glutamate induces a significant increase in -sheet content and simultaneous moderate increases in both -helical and turn contents (8). Similar changes in -sheet and turn contents and a small increase in -helical content are observed on binding of the partial agonist kainate (8).² Thus, there is a correlation between the magnitude of the infrared spectral changes and the extent of channel activation induced by GluR ligands. Furthermore, as outlined in the Introduction, the extent of spectral changes mirrors the crystallographically observed differences in the extent of domain closure on the binding of antagonists, partial agonists, and full agonists (7). It is therefore likely that infrared spectral changes reflect physiologically relevant conformational changes in the binding domain.

Conclusions-- Infrared spectroscopy can be used to refine and extend the information provided by crystal structures. Here, the availability of isotopically labeled quinoxaline antagonists has revealed their electronic configurations and charge distributions as stabilized either in solution or by their interactions with the glutamate receptor binding site. This information is not directly accessible to crystallographic analysis alone but rather to a combination of structural and spectroscopic techniques. In addition, FTIR difference spectra confirm the similarity of the binding interactions of both CNQX and DNQX with S1S2 and permit us to model the previously uncharacterized CNQX interaction on the basis of the crystal structure of the DNQX-S1S2 complex. Finally, the vibrational spectra detect subtle changes in protein secondary structure content that appear to distinguish clearly the conformational effects of different functional classes of ligands, including antagonists, partial agonists, and full agonists. The ability to monitor such conformational changes spectroscopically may ultimately permit a time-resolved analysis of the conformational changes induced by agonist binding.

	ACKNOWLEDGEMENTS

We thank U. Reygers for excellent technical assistance and gratefully acknowledge Dr. P. Jacobsen (Novo Nordisc) for advice and for a sample of 6-cyano-2,3-dihydroxyquinoxaline. V. J. thanks Dr. W. A. Donaldson and Dr. R. Rathore for helpful discussions. D. R. M. thanks Dr. K. Keinänen (University of Helsinki) for collaborative support in establishing S1S2 expression and purification.

	FOOTNOTES

^* This work was supported by NSF, National Institutes of Health Grants NSF-9982759 and NSF-0096635 (to V. J.) and by a grant from The Max-Planck Society (to D. R. M.). D. R. M. and S. T. contributed equally to this work.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^¶ To whom correspondence should be addressed: Dept. of Chemistry, Marquette University, Milwaukee, WI 53233. Tel.: 414-288-7859; Fax: 414-288-7066; E-mail: vasanthi.jayaraman@marquette.edu.

Published, JBC Papers in Press, July 31, 2001, DOI 10.1074/jbc.M106171200

² S. Thiran, R. Madden, and V. Jayaraman, unpublished data.

	ABBREVIATIONS

The abbreviations used are: GluR, glutamate receptor; FTIR, Fourier transform infrared spectroscopy; CNQX, 6-cyano-7-dinitro-2,3- dihydroxyquinoxaline; DNQX, 6,7-dinitro-2,3-dihydroxyquinoxaline; AMPA, -amino-5-methyl-3-hydroxy-4-isoxazole propionate.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				M. Du, A. Rambhadran, and V. Jayaraman Luminescence Resonance Energy Transfer Investigation of Conformational Changes in the Ligand Binding Domain of a Kainate Receptor J. Biol. Chem., October 3, 2008; 283(40): 27074 - 27078. [Abstract] [Full Text] [PDF]

				D. R. Madden, N. Armstrong, D. Svergun, J. Perez, and P. Vachette Solution X-ray Scattering Evidence for Agonist- and Antagonist-induced Modulation of Cleft Closure in a Glutamate Receptor Ligand-binding Domain J. Biol. Chem., June 24, 2005; 280(25): 23637 - 23642. [Abstract] [Full Text] [PDF]

				M. Du, S. A. Reid, and V. Jayaraman Conformational Changes in the Ligand-binding Domain of a Functional Ionotropic Glutamate Receptor J. Biol. Chem., March 11, 2005; 280(10): 8633 - 8636. [Abstract] [Full Text] [PDF]

				Q. Cheng and V. Jayaraman Chemistry and Conformation of the Ligand-binding Domain of GluR2 Subtype of Glutamate Receptors J. Biol. Chem., June 18, 2004; 279(25): 26346 - 26350. [Abstract] [Full Text] [PDF]

				R. L. McFEETERS and R. E. OSWALD Emerging structural explanations of ionotropic glutamate receptor function FASEB J, March 1, 2004; 18(3): 428 - 438. [Abstract] [Full Text] [PDF]

				D. Deming, Q. Cheng, and V. Jayaraman Is the Isolated Ligand Binding Domain a Good Model of the Domain in the Native Receptor? J. Biol. Chem., May 9, 2003; 278(20): 17589 - 17592. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) All Versions of this Article: 276/41/37821    most recent M106171200v1 Submit a Letter to Editor Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Madden, D. R. Articles by Jayaraman, V. Search for Related Content PubMed PubMed Citation Articles by Madden, D. R. Articles by Jayaraman, V. Social Bookmarking                      What's this?