Ionotropic glutamate receptors (GluRs) are molecular pores that facilitate the passage of ions across cell membranes. The mechanism mediates excitatory signal transmission in the mammalian nervous system at the majority of synapses. Given the receptors’ essential role in normal brain function and development and increasing evidence that dysfunction of GluR activity underlies multiple diseases of the central nervous system as well as damage during stroke, we have directed much effort toward analyzing GluR function at the molecular level. A major goal of our work is to obtain structural data that will allow the development of subtype-selective antagonists and allosteric modulators. In humans, the ionotropic glutamate receptors are encoded by seven gene families named after the ligands that were first used to identify the major subtypes on a functional basis: AMPA, kainate, and NMDA. In addition, a growing family of prokaryotic GluRs is emerging from the sequencing of microbial genomes, of which GluR0 from the photosynthetic bacterium syncheocystis PCC 6803 was the first to be identified. Much work remains to be done, including studies on other subtypes and domains that were excluded from the constructs used in our initial studies.

Expression of Kainate Receptor Ligand-Binding Cores
Mayer
The recent crystallization of the ligand-binding cores of an AMPA receptor subunit and GluR0 has revealed, for the first time, the molecular mechanisms underlying the binding of agonists and antagonists. By studying the structural and functional changes produced by mutations that alter the gating of ion channel activity of the GluR2 AMPA receptor subunit, we have obtained evidence that defines a major structural role for an interface between the ligand-binding cores of individual subunits. Our studies show that, for ion channel gating to proceed, the subunits must assemble as dimers. Drugs or mutations that increase dimer stability attenuate desensitization and facilitate gating while mutations that reduce dimer stability enhance desensitization. Our studies reveal, for the first time, a molecular mechanism for desensitization of a ligand-gated ion channel.

Ongoing work in the laboratory has focused on the establishment of procedures for the expression, biophysical characterization, and crystallization of the ligand-binding cores of kainate subtype GluRs. Our goal is to characterize members of both the high- (KA1 / KA2) and low- (GluR5-GluR7) affinity kainate receptor gene families. Based on the design of the previously characterized GluR2 and GluR0 ligand-binding cores, we assembled related constructs for GluR5, GluR6, KA1, and KA2 for expression in E. coli. At present, procedures have been established for the production of highly purified GluR6 at the 20 mg scale and of GluR5 at the 5 to 10 mg scale. Both proteins were expressed in the cytoplasm of E. coli with N-terminal His tags and purified by NTA affinity chromatography followed by proteolytic removal of the His tag and final purification by ion exchange chromatography. The proteins are monodisperse when concentrated to 20 mg/ml and have the expected molecular weight when analyzed by electrospray mass spectrometry. We have grown crystals of a GluR6 complex with L-glutamate and characterized them by x-ray diffraction. The lattice is orthorhombic with unit cell dimensions a = 57.8 A, b= 90.8 A, c= 104.8 A. We have tentatively identified the space group as P21212, and diffraction to < 2.2 A is obtained on a rotating anode x-ray source. With the crystals not sufficiently well ordered for a high-resolution structure determination, crystallization conditions are undergoing optimization to address this deficiency. Additional agonists for which we intend to solve GluR6 complexes include kainate, quisqualate, and 4methly-glutamate, a high-affinity kainate receptor agonist. Of particular interest will be measurements of the extent of domain closure produced by kainate compared with data obtained for GluR2. Kainate acts as a partial agonist for GluR2, producing approximately 10° less domain closure than glutamate or AMPA. Coupled with this analysis, we plan to design non-desensitizing kainate receptors and use them to quantify the action of partial agonists, as we have done for AMPA receptors.

Until now, we have not succeeded in obtaining diffraction-quality crystals for GluR5 and thus will continue optimization of crystallization conditions. GluR5 is of special interest because recent studies show that GluR5 antagonists have anticonvulsant activity. A large number of GluR5-selective ligands that do not bind to GluR6 have been developed, including ATPA, 5-iodowillardiine, and the (S)-enantiomer of E4(2,2dimethylpropylidene) glutamic acid, but the structural basis for this selectivity is not well understood. Attempts to express soluble KA1 and KA2 have not met with success, and so far sparse matrix folding screens have not yielded sufficient protein from solubilized inclusion bodies for structural work. We will examine different expression systems for these proteins.

Structural Basis for Glutamate Receptor Desensitization
Horning, Mayer; in collaboration with Gouaux
Receptor desensitization is a common process that occurs in molecules as diverse as ion channels and G protein–coupled receptors.

Figure 15

Mechanistic scheme for GluR desensitization based on crystal structures of the ligand binding core.

While the role(s) of desensitization may differ among different pathways, the concept that desensitization participates in shaping the amplitude, duration, and frequency of signals at synaptic circuits is gaining acceptance. Despite the fact that desensitization is a ubiquitous phenomenon in receptor signal transduction cascades, the molecular mechanisms underlying desensitization have defied analysis. By performing structural and functional studies on the AMPA-selective GluR2 receptor that localize crucial elements of the receptor involved in desensitization, we have obtained evidence for quantitative relationships between the extent of receptor desensitization and the strength of intradimer subunit interface interactions. Our goal was to test whether the interactions that are seen in the dimer interface in the crystal structures are present in the intact receptors and to determine the extent to which the strength of the dimer interface, as measured by the dimer dissociation constant (Kd) using analytical ultracentrifugation experiments, is correlated with the degree of receptor desensitization. To this end, we made a number of single, double, and triple mutants and examined their behaviors by using the patch clamp technique and rapid solution exchange. Our analyses revealed a remarkable correlation between the extent of equilibrium desensitization and the Kd of dimer dissociation for mutations at the binding site for the allosteric modulator cyclothiazide and at the position 483 mutation from leucine to tyrosine, which blocks desensitization. Our experiments suggest a model in which the dimer interface acts as a supporting point so that the conformational strain caused by ligand binding–evoked domain closure can be transferred to the gate, which we assume is located at the extracellular surface of the membrane, thus opening the channel. In the event of desensitization, the ligand-binding cleft also closes and traps agonist, but the dimer interface is disrupted, thus releasing the domain closure energy and decoupling it from the channel gate. For the wild-type receptor, the energy barrier for desensitization is higher than for activation; therefore, the receptor activates faster than it desensitizes. The desensitized receptor is more stable than the activated receptor, however, and thus, during a prolonged incubation with agonist, the large share of the receptors becomes desensitized.