Search

We have expressed, purified, and characterized glutamate receptor ion channels (GluR) assembled as homomers of the subunit GluRB. For the first time, single-milligram quantities of biochemically homogeneous GluR have been obtained. The protein exhibits the expected pharmacological profile and a high specific activity for ligand binding. Density-gradient centrifugation reveals a uniform oligomeric assembly and a molecular mass suggesting that the channel is a tetramer. On the basis of electron microscopic images, the receptor appears to form an elongated structure that is visualized in several orientations. The molecular dimensions of the molecule are approximately 11 x 14 x 17 nm, and solvent-accessible features can be seen; these may contribute to formation of the ion-conducting pathway of the channel. The channel dimensions are consistent with an overall 2-fold symmetric assembly, suggesting that the tetrameric receptor may be a dimer of dimers.

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.

An AMPA-specific ionotropic glutamate receptor binding domain was overexpressed using the baculovirus system and purified by immunoaffinity and metal-affinity chromatography. Purified protein was enzymatically deglycosylated. Both glycosylated and deglycosylated proteins crystallized under the same conditions and in the same space group (P2). In both cases, it was observed that the use of MPD as a cryoprotectant induced a significant reduction in the unit-cell volume compared with glycerol or sucrose. For crystals of deglycosylated protein, cryoprotection with MPD also yielded a dramatic improvement in resolution.

Fourier transform infrared spectroscopy was used to investigate ligand-protein interactions in the ligand-binding domain of the GluR4 glutamate receptor subunit. Glutamate binding induces more extensive secondary structural changes in the ligand-binding domain than does kainate binding. Glutamate also alters the hydrogen bonding strength of the single free cysteine side chain in the domain, while kainate does not. On the other hand, the interaction of a binding site arginine residue with kainate appears to be stronger than that with glutamate. These results identify chemical and structural differences that may explain the different functional characteristics of the two agonists acting on ionotropic glutamate receptors. In doing so, they complement and extend recent crystallographic structures of the ligand-binding domain.

Agonist binding to glutamate receptor ion channels occurs within an extracellular domain (S1S2) that retains ligand affinity when expressed separately. S1S2 is homologous to periplasmic binding proteins, and it has been proposed that a Venus flytrap-style cleft closure triggers opening of glutamate receptor ion channels. Here we compare the kinetics of S1S2-agonist binding to those of the periplasmic binding proteins and show that the reaction involves an initial rapid association, followed by slower conformational changes that stabilize the complex: "docking" followed by "locking." The motion detected here reflects the mechanism by which the energy of glutamate binding is converted into protein conformational changes within S1S2 alone. In the intact channel, these load-free conformational changes are harnessed and possibly modified as the agonist binding reaction is used to drive channel opening and subsequent desensitization. Using mutagenesis, key residues in each step were identified, and their roles were interpreted in light of a published S1S2 crystal structure. In contrast to the Venus flytrap proposal, which focuses on motion between the two lobes as the readout for agonist binding, we argue that smaller, localized conformational rearrangements allow agonists to bridge the cleft, consistent with published hydrodynamic measurements.

The ionotropic glutamate receptors (GluR) are the primary mediators of excitatory synaptic transmission in the brain. GluR agonist binding has been localized to an extracellular domain whose core is homologous to the bacterial periplasmic binding proteins (PBP). We have established routine, baculovirus-mediated expression of a complete ligand-binding domain construct at the 10-L scale, yielding 10-40 milligrams of purified protein. This construct contains peptides that lie outside the PBP-homologous core and that connect the domain core to the transmembrane domains of the channel and to the N-terminal 'X'-domain. These linker peptides have been implicated in modulating channel physiology. Such extended constructs have proven difficult to express in bacteria, but the protein described here is stable and monomeric. Isothermal titration calorimetry reveals that glutamate binding to the domain involves a substantial heat capacity change and that at physiological temperatures, the reaction is both entropically and enthalpically favorable.

Solution scattering studies were performed on a ligand-binding domain (S1S2) of a glutamate receptor ion channel (GluR) in order to study GluR-binding and signal-transduction mechanisms. The core of the ligand-binding domain is homologous to prokaryotic periplasmic binding proteins (PBP), whose binding mechanism involves a dramatic cleft closure: the "Venus flytrap". Several models of GluR function have proposed that a similar cleft closure is induced by agonist binding. We have directly tested this putative functional homology by measuring the radius of gyration of S1S2 in the presence and absence of saturating concentrations of agonists. In contrast to the PBP, S1S2 shows no reduction in radius of gyration upon agonist binding, excluding a comparably large conformational change. Furthermore, we determined an ab initio molecular envelope for our S1S2 construct, which also contains the peptides that connect the PBP homology core to the three transmembrane domains and to an N-terminal domain. By fitting an atomic model of the ligand-binding domain core to the envelope of our extended construct, we were able to establish the likely position of these connecting peptides. Their positions relative to one another and to the expected sites of an agonist-induced conformational change suggest that ion channel gating and desensitization may involve more subtle and complex mechanisms than have been assumed based on the structural homology to the PBP.

We prepared and crystallized five complexes of the human histocompatibility molecule HLA-A2 with peptides derived from human immunodeficiency virus type 1, human T lymphotropic virus type 1, influenza A virus and hepatitis B virus proteins. Each HLA-A2 complex was refolded in vitro from insoluble proteins produced in bacteria; to crystallize, two of the complexes required seeding with microcrystals of another complex. Maintained at -160 degrees C, single co-crystals of each of the five peptide-HLA-A2 complexes yielded complete X-ray diffraction data sets to a resolution of approximately 2.5 A. After a sufficient number of diffraction peaks were acquired during data collection, the direct analysis of integrated intensities established the point group of the co-crystal, thus allowing an efficient data collection strategy to be designed. The subsequent examination of systematic absences revealed that the five peptide-HLA-A2 co-crystals formed in space groups P1, P2(1), or P2(1)2(1)2(1). Molecular replacement structure solutions yielded unambiguous protein electron density maps, thus confirming the space group determinations. The system of obtaining HLA-A2 co-crystal structures described here is applicable to other crystallographic problems where structures of several related molecules from uncharacterized single crystals are required.

Complexes of five peptides (from HIV-1, influenza A virus, HTLV-1, and hepatitis B virus proteins) bound to the human class I MHC molecule HLA-A2 have been studied by X-ray crystallography. While the peptide termini and their second and C-terminal anchor side chains are bound similarly in all five cases, the main chain and side chain conformations of each peptide are strikingly different in the center of the binding site, and these differences are accessible to direct TCR recognition. Each of the central peptide residues is seen to point up for some bound peptides, but down or sideways for others. Thus, although fixed at its ends, the structure of an MHC-bound peptide appears to be a highly complex function of its entire sequence, potentially sensitive to even small sequence differences. In contrast, MHC structural variation is relatively limited. These results offer a structural framework for understanding the role of nonanchor peptide side chains in both peptide-MHC binding affinity and TCR recognition.

Cell surface complexes of class I MHC molecules and bound peptide antigens serve as specific recognition elements controlling the cytotoxic immune response. The 2.1 A structure of the human class I MHC molecule HLA-B27 provides a detailed composite image of a co-crystallized collection of HLA-B27-bound peptides, indicating that they share a common main-chain structure and length. It also permits direct visualization of the conservation of arginine as an "anchor" side chain at the second peptide position, which is bound in a potentially HLA-B27-specific pocket and may therefore have a role in the association of HLA-B27 with several diseases. Tight peptide binding to class I MHC molecules appears to result from the extensive contacts found at the ends of the cleft between peptide main-chain atoms and conserved MHC side chains, which also involve the peptide in stabilizing the three-dimensional fold of HLA-B27. The concentration of binding interactions at the peptide termini permits extensive sequence (and probably some length) variability in the center of the peptide, where it is exposed for T cell recognition.

The class I major histocompatibility (MHC) antigen HLA-B27 was purified by immunoaffinity chromatography from the homozygous human B lymphoblastoid cell line LG-2. Detergent-soluble HLA-B27 was cleaved with the protease papain to remove the hydrophobic transmembrane region and the cytoplasmic tail. Crystals of the resulting water-soluble extracellular fragments were obtained in hanging drops by the vapor-diffusion method. The crystals are triclinic, space group P1, with unit cell dimensions a = 45.9 A, b = 71.0 A, c = 83.7 A, alpha = 79.4 degrees, beta = 88.5 degrees, gamma = 89.9 degrees, and diffract beyond 2.5 A resolution.