Ionotropic glutamate-like receptor delta2 binds D-serine and glycine

Abstract

The orphan glutamate-like receptor GluRdelta2 is predominantly expressed in Purkinje cells of the central nervous system. The classification of GluRdelta2 to the ionotropic glutamate receptor family is based on sequence similarities, because GluRdelta2 does not form functional homomeric glutamate-gated ion channels in transfected cells. Studies in GluRdelta2(-/-) knockout mice as well as in mice with naturally occurring mutations in the GluRdelta2 gene have demonstrated an essential role of GluRdelta2 in cerebellar long-term depression, motor learning, motor coordination, and synaptogenesis. However, the lack of a known agonist has hampered investigations on the function of GluRdelta2. In this study, the ligand-binding core of GluRdelta2 (GluRdelta2-S1S2) was found to bind neutral amino acids such as D-serine and glycine, as demonstrated by isothermal titration calorimetry. Direct evidence for binding of D-serine and structural rearrangements in the binding cleft of GluRdelta2-S1S2 is provided by x-ray structures of GluRdelta2-S1S2 in its apo form and in complex with D-serine. Functionally, D-serine and glycine were shown to inactivate spontaneous ion-channel conductance in GluRdelta2 containing the lurcher mutation (EC(50) values, 182 and 507 microM, respectively). These data demonstrate that the GluRdelta2 ligand-binding core is capable of binding ligands and that cleft closure of the ligand-binding core can induce conformational changes that alter ion permeation.

Fig. 1.

The structure of the ligand-binding core of GluRδ2. (A) Domain organization of the GluRδ2 receptor subunit. The architecture is similar to other glutamate receptors (AMPA, kainate, and NMDA receptors), with an extracellular N terminus, three transmembrane segments (M1, M2, and M3), a reentrant membrane loop (P), and an intracellular C terminus. The extracellular regions harbor the N-terminal domain (NTD) and the ligand-binding core (D1 and D2). The red dot shows the approximate position of the lurcher mutation (A654T). The boundaries of the GluRδ2–S1S2 construct are indicated by scissors, and the dotted line represents the Gly–Thr linker. (B) Representation of the twofold symmetric dimer of GluRδ2–S1S2 apo (in yellow). The structure of GluRδ2–S1S2 in complex with d-serine has been superimposed onto GluRδ2–S1S2 apo and is shown in red. d-serine is displayed as green spheres.

Fig. 2.

ITC studies on GluRδ2–S1S2. Raw data (Upper) and isotherms (Lower) of d-serine, l-serine, and glycine are presented. The filled squares represent titrations of GluRδ2–S1S2 with ligands, and the open squares represent titrations with ligands in buffer only. The graphs show that heat is absorbed after each injection of the ligand and that the heat signal diminishes as the protein becomes saturated with the ligand. The K_d values of d-serine and glycine binding correspond to fitted 1/K_d values of 1.11 ± 0.04 (10³·M⁻¹) and 0.36 ± 0.03 (10³·M⁻¹), respectively. N.b., no binding. Molar ratio is the ratio of ligand relative to GluRδ2–S1S2.

Fig. 3.

The structure of the ligand-binding core of GluRδ2 in complex with d-serine. (A) The d-serine-binding site of GluRδ2–S1S2. The F_o – F_c electron-density map of d-serine before introduction of this molecule in the refinements is shown. d-serine and potential hydrogen-bonding residues of GluRδ2 are represented as sticks, and dashed lines indicate hydrogen bonds. (B) Contour of the ligand-binding cavity of GluRδ2–S1S2 in complex with d-serine (shown in gray). d-serine and GluRδ2 residues within 3.5 Å are represented as sticks. No water molecules were located within the hydrogen-bonding distance of d-serine.

Fig. 4.

Electrophysiology on WT GluRδ2 and the lurcher mutant GluRδ2^Lc receptor. (A) Mean concentration–response curves for d-serine (■) and glycine (□) at GluRδ2^Lc receptors expressed in Xenopus oocytes. (Inset) The responses (60 s) are normalized to the maximal current response (ΔI_max) of the indicated ligand. (Calibration bar: 100 nA.) The EC₅₀ values of d-serine and glycine are 182 ± 11 μM (n = 9) and 507 ± 40 μM (n = 7), respectively. The maximal current response elicited by glycine was 0.95 ± 0.02 (n = 7) relative to the maximal current response elicited by d-serine in the same recording. Error bars (SEM) are small and are contained within the data points. (B) Inhibition of GluRδ2^Lc currents in oocytes by application of 1 mM various amino acids, including d-serine and glycine. Values are normalized to the inhibition produced by application of 100 μM NASP and are the means ± SEM from 3 to 16 oocytes. (C) Representative traces from electrophysiological recordings showing the current response to fast application of 10 mM d-serine to outside-out membrane patches from HEK-293 cells transfected with GluRδ2 (Upper), GluRδ2^Lc (Lower Left), and GluR1 (Lower Right). (Calibration bars: Upper Left, 10 pA, 10 s; Upper Right, 10 pA, 5 ms; Lower Left, 10 pA, 10 s; Lower Right, 50 pA, 5 ms.) d-serine did not elicit a current response from WT GluRδ2. For comparison, Lower Right depicts the current–response from the rapidly desensitizing AMPA receptor GluR1 to fast application of 10 mM l-glutamate shown on the same timescale as WT GluRδ2. The dashed line indicates the average baseline current, as measured by complete NASP inhibition of the spontaneously activated GluRδ2^Lc current. Upon application of d-serine to GluRδ2^Lc, the current is inhibited to a level corresponding to ≈81% of the NASP inhibition. (D) (Lower) The effect of 1 mM d-serine (■) or 50 μM NASP (●) on the constitutive whole-cell current (○) at different membrane potentials in HEK-293 cells expressing GluRδ2^Lc. (Upper) Current–voltage relationship was generated by ramping the holding voltage from +45 mV to −50 mV (50 ms) with 10 μM spermine present in the recording pipette, and normalization was performed corresponding to the maximal observed current. (Calibration bar: 100 pA, 50 ms.)