Exploring the GluR2 ligand-binding core in complex with the bicyclical AMPA analogue (S)-4-AHCP Bettina B. Nielsen 1 , Darryl S. Pickering 2 , Jeremy R. Greenwood 1 , Lotte Brehm 1 , Michael Gajhede 1 , Arne Schousboe 2 and Jette S. Kastrup 1 1 Biostructural Research, Department of Medicinal Chemistry, Danish University of Pharmaceutical Sciences, Copenhagen, Denmark 2 Department of Pharmacology, Danish University of Pharmaceutical Sciences, Copenhagen, Denmark The main excitatory amino acid in the central nervous system (S)-glutamate, exerts its actions by binding to three different classes of ionotropic glutamate receptors (iGluRs) and three classes of metabotropic receptors, which all have important functions in neuronal signal- ling (for a review, see [1]). With the glutamatergic sys- tem implicated in a variety of brain disorders such as schizophrenia and Alzheimer’s disease, these receptors are potential targets for pharmacotherapy [2,3]. The iGluRs form ligand-gated ion channels and have been classified according to their agonist selectivity as 2-amino-3-(3-hydroxy-5-methyl-4-isoxazolyl)pr opionic acid (AMPA), kainic acid (KA) and N-methyl-d-aspar- tic acid (NMDA) receptors [3]. Four subunits, assem- bled as a pair of dimers, constitute the receptor ion-channel complex [4–11]. Among iGluR subunits, homomeric and heteromeric receptors constructed from cloned GluR1–4 are most sensitive to activation by AMPA, and thus native AMPA receptors are identified with these genes [12–14]. The subunits exist in two different alternatively spliced isoforms, flip (i) and flop (o), which have different desensitization properties. In addition, two RNA-edited isoforms of GluR2 are found in which a crucial amino-acid residue in the channel pore region is either glutamine or arginine (the Q ⁄ R site) and this affects channel properties, e.g. rectification and ion-selectivity of GluR2-containing heteromeric channels [2]. Keywords (S)-4-AHCP; bicyclical AMPA analogue; ionotropic glutamate receptor; ligand-binding core; X-ray crystallography Correspondence J. S. Kastrup, Biostructural Research, Department of Medicinal Chemistry, Danish University of Pharmaceutical Sciences, Universitetsparken 2, DK-2100 Copenhagen, Denmark Fax: +45 3530 6040 Tel: +45 3530 6486 E-mail: jsk@dfuni.dk (Received 28 September 2004, revised 18 January 2005, accepted 25 January 2005) doi:10.1111/j.1742-4658.2005.04583.x The X-ray structure of the ionotropic GluR2 ligand-binding core (GluR2- S1S2J) in complex with the bicyclical AMPA analogue (S)-2-amino-3-(3-hyd- roxy-7,8-dihydro-6H-cyclohepta[d]-4-isoxazolyl)propionic acid [(S)-4-AHCP] has been determined, as well as the binding pharmacology of this construct and of the full-length GluR2 receptor. (S)-4-AHCP binds with a glutamate- like binding mode and the ligand adopts two different conformations. The K i of (S)-4-AHCP at GluR2-S1S2J was determined to be 185 ± 29 nm and at full-length GluR2(R) o it was 175 ± 8 nm.(S)-4-AHCP appears to elicit par- tial agonism at GluR2 by inducing an intermediate degree of domain closure (17°). Also, functionally (S)-4-AHCP has an efficacy of 0.38 at GluR2(Q) i , relative to (S)-glutamate. The proximity of bound (S)-4-AHCP to domain D2 prevents full D1–D2 domain closure, which is limited by steric repulsion, especially between Leu704 and the ligand. Abbreviations 4-AHCP, 2-amino-3-(3-hydroxy-7,8-dihydro-6H-cyclohepta[d]-4-isoxazolyl)propionic acid; AMPA, 2-amino-3-(3-hydroxy-5-methyl-4- isoxazolyl)propionic acid; Br-HIBO, 2-amino-3-(4-bromo-3-hydroxy-5-isoxazolyl)propionic acid; EC 50 , concentration of drug producing 50% of the maximal response; GluR2-S1S2J, soluble construct of the ionotropic GluR2 ligand-binding core; iGluR, ionotropic glutamate receptor; KA, kainic acid; 2-Me-Tet-AMPA, 2-amino-3-[3-hydroxy-5-(2-methyl-2H-5-tetrazolyl)-4-isoxazolyl]propionic acid; n H , Hill coefficient; NMDA, N-methyl- D-aspartic acid. FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS 1639 Overexpression and purification of a GluR2 construct (GluR2-S1S2J) containing the extracellular segments S1 and S2 linked by a small peptide has provided a soluble form of the ligand-binding core of the GluR2 receptor, belonging to the AMPA class of iGluRs [15]. Binding of agonists to this construct creates a pharmacological profile comparable to that seen for the full-length recep- tor [16,17]. A number of structures of GluR2-S1S2J in complex with different agonists (e.g. [15,18–22]) and antagonists [15,23] have provided evidence that compet- itive ligands bind in a cleft between two domains, D1 and D2. Domain movement occurs upon ligand bind- ing, resulting in closure of the binding cleft. The extent of domain closure is correlated with activation and desensitization of the receptor (for a review, see [24]). The AMPA receptor agonist, 2-amino-3-(3-hydroxy- 7,8-dihydro-6H-cyclohepta[d] -4-isoxazolyl)propionic acid (4-AHCP) was originally synthesized as a con- formationally restricted analogue of AMPA [25]. As is often the case, the agonist activity resides with one enantiomer (S)-4-AHCP. The agonist shows activity at AMPA receptors, but it is 35–115 times more potent at GluR5 homomers, classified as low affinity KA receptors [26]. (S)-4-AHCP distinguishes itself from the majority of iGluR agonists by its bicyclical structure, and the extra carbon atom between the a-amino acid moiety and the distal isoxazole 3-hydroxy anion (Fig. 1). The seven-membered ring offers opportunities for design and stereospecific derivatization in direc- tions not easily accessible from other scaffolds. Accu- rate knowledge of the binding mode of (S)-4-AHCP may assist in the structure-based design of ligands that are targeted to iGluR specific subtypes, as well as highlighting regions that could be occupied by small hydrophobic substituents to increase affinity. Here, we present the first structure of GluR2-S1S2J in complex with the bicyclical AMPA analogue (S)-4-AHCP, as well as the pharmacology of (S)-4-AHCP binding to GluR2-S1S2J and full-length GluR2 receptors. Results and Discussion Pharmacology of (S)-4-AHCP The affinity of (S)-4-AHCP for GluR2-S1S2J was deter- mined to be (mean ± SEM): K i ¼ 185 ± 29 nm; Hill coefficient (n H ) ¼ 1.03 ± 0.04 (n ¼ 4) and for full- length GluR2(R) o :K i ¼ 175 ± 8 nm;n H ¼ 0.95 ± 0.02 (n ¼ 3) (Fig. 2A). No statistically significant difference between the K i values was observed using the t-test (P ¼ 0.80). The affinity for the construct GluR2-S1S2J is identical to the affinity for the full-length receptor, implying that the binding of (S)-4-AHCP observed in the crystal structure represents the binding mode at the full-length receptor. Analysis of concentration–response curves for (S)-4-AHCP activation of GluR2(Q) i exp- ressed in Xenopus laevis oocytes gave: concentration of drug producing 50% of the maximal response (EC 50 ) ¼ 17.5 ± 1.2 lm,n H ¼ 0.95 ± 0.04 (n ¼ 7) and an effic- acy of 0.381 ± 0.046 (n ¼ 7), relative to (S)-glutamate (Fig. 2B). Similar EC 50 values have been reported for (S)-4-AHCP at the other AMPA receptors: GluR1 o (4.5 lm), GluR3 o (7.2 lm) and GluR4 o (15 lm) [26]. Interactions of (S)-4-AHCP with GluR2 The GluR2-S1S2J:(S)-4-AHCP complex crystallizes with one molecule in the asymmetric unit of the crystal and the structure has been determined at 1.75 A ˚ reso- lution (Table 1). (S)-4-AHCP was modelled in two dif- ferent conformations in the ligand-binding site; the major variation is two conformationally enantiomeric puckering modes of the seven-membered ring of the ligand but the orientation of the ligand in the two con- formations is also slightly different (Fig. 3). However, the positions of the atoms of the 3-isoxazolol moiety only differ between 0.2 and 0.4 A ˚ , which may be within the experimental error. The interactions of the a-amino acid moiety of both conformations of the lig- and with binding site residues are tabulated in Table 2. Multiple ligand conformations, implying some retent- ion of conformational entropy upon binding, have not previously been observed at GluR2. The barrier to ring inversion in (S)-4-AHCP is lower than the barrier to Fig. 1. Chemical structures of the neutral forms of the GluR2 agon- ists (S)-glutamate, (S)-AMPA, (S)-2-Me-Tet-AMPA and (S)-4-AHCP (including atom numbering as in pdb-file). GluR2 ligand-binding core in complex with (S)-4-AHCP B. B. Nielsen et al. 1640 FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS binding and receptor activation, hence, ring inversion occurs on a much more rapid timescale than binding and activation. Significant flexibility in the protein is not seen in response to the two conformations and both are substantially populated, which suggests that they are similar in internal energy, in interaction energy and in binding energy. We expect that the lig- and can rearrange while bound, at least at the tem- perature at which affinity is measured. Thus, the observed affinity will be a Boltzmann average of the two states in rapid equilibrium. The interactions of the a-amino acid moiety of the ligand with binding site residues (Table 2) are con- served compared to other GluR2-agonist complexes [15,18–21]. Besides the residues forming ion pairs or hydrogen bonds with the ligand, a number of addi- tional residues are involved in hydrophobic or van der Waals’ interactions (Table 2). Practically all the same residues have been shown to be in van der Waals’ con- tact in other agonist complexes (e.g. [20]). The ligand is ‘squeezed in’ between Tyr450 and Leu650 on the one side and Glu705 and Tyr732 on the opposite side, forming a hydrophobic sandwich (Fig. 3B and C). The Tyr450 side chain interacts extensively with the a-amino acid group, as well as with C4, C5, C6 and C7 of the seven-membered ring. Leu650 interacts with the isoxazole, C4, C5, C6 and C9. The residues Glu402 and Thr686, which form an interdomain hydrogen-bond lock upon agonist binding, also inter- act with atoms of the seven-membered ring. Leu704 forms contacts to C3, N1 and O1 of the isoxazole ring and C8 of the seven-membered ring, while the residue Fig. 2. Pharmacology of (S)-AHCP at GluR2. (A) Binding affinities of (S)-AHCP at the soluble GluR2-S1S2J construct and at the full- length GluR2(R) o receptor. One representative [ 3 H]AMPA radiolig- and binding experiment is shown for each (mean ± SD of tripli- cates). Experiments were replicated a total of three or four times. (B) Potency of (S)-AHCP activation of GluR2(Q) i expressed in X. laevis oocytes. Shown are means ± SD of concentration– response data pooled from seven oocytes, normalized to each maximal steady-state response. (Inset) Current traces showing the relative efficacy (here, 0.361) of (S )-4-AHCP vs. (S)-glutamate. Cyclothiazide (100 l M CTZ; white box) was preapplied for 60 s before application of 1 m M (S)-glutamate (G; black box) or 500 lM (S)-AHCP (A, black box). Note that a perfusion delay of 4–5 s occurs in this system. Scale bars, 200 nA, 10 s. Table 1. Crystal data, data collection and refinement statistics for GluR2-S1S2J:(S )-4-AHCP. Crystal data Space group P2 1 2 1 2 Unit cell parameters (A ˚ )a¼ 94.4, b ¼ 59.5, c ¼ 47.8 No. of molecules per a.u. 1 Data collection Resolution range (A ˚ ) a 20.1–1.75 (1.78–1.75) No. of unique reflections 27472 Average redundancy 3.5 Completeness (%) 99.1 (96.3) R sym (%) 3.7 (22.7) I ⁄ r(I) 26.7 (4.1) Refinements Total number of atoms Non-hydrogen 2393 Protein 2039 Ligand 17 Water 309 Sulfate and glycerol 28 R-values R work (%) 16.9 (24.8) R free, 5% (%) 21.4 (31.8) Rms deviations Bond lengths (A ˚ ) 0.017 Bond angles (°) 1.6 Residues in allowed regions of Ramachandran plot (%) b 99.6 Mean B-values (A ˚ 2 ) Protein atoms 18.7 Ligand atoms 14.9 Water 31.5 Sulfate and glycerol 52.5 a Values in parentheses correspond to the outermost resolution bin. b The Ramachandran plot was calculated according to Kleywegt and Jones [49]. B. B. Nielsen et al. GluR2 ligand-binding core in complex with (S)-4-AHCP FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS 1641 Met708 follows the contour of the seven-membered ring, interacting hydrophobically with C6, C7 and C8. (S)-4-AHCP may be classified as a barely exposed lig- and, as only a small part of the bicyclical ring system can be seen from the exterior of the protein. The conformation of the peptide bond Asp651– Ser652, which has been shown to be flipped 180° in the complexes with most full agonists [15,18,19], is similar to the nonflipped conformation observed in, for example, the apo and (S)-2-amino-3-(3-hydroxy- 5-methyl-4-isoxazolyl)propionic acid [(S)-Br-HIBO] complex. The backbone oxygen atom of Ser652 is connected to the 3-hydroxy anion (O2) of (S)-4-AHCP through the water molecule W1. Comparison with other isoxazole-based agonists The binding modes of GluR2 agonists thus far charac- terized can be broadly divided into two classes, the glutamate mode and the AMPA mode. In the glutam- ate binding mode, the agonist approaches D2 more closely and interacts directly with the hydrogen-bond donor atoms of Ser654 and Thr655. In the AMPA binding mode, the interaction with D2 is mediated via a water molecule (here denoted W4) [15,18,19]. To characterize the binding mode of (S)-4-AHCP, the structure was compared to those of GluR2-S1S2J in complex with (S)-glutamate (S)-AMPA and (S)-2- amino-3-[3-hydroxy-5-(2-methyl-2 H -5 -tetrazolyl)-4-is- oxazolyl]propionic acid [(S)-2-Me-Tet-AMPA]. The binding mode of (S)-4-AHCP falls between those of (S)-2-Me-Tet-AMPA and (S)-AMPA with regard to the position of the isoxazole ring (Fig. 4A). As both (S)-4-AHCP and (S)-2-Me-Tet-AMPA bind with the distal anionic moiety interacting directly with the back- bone N atom of Thr655, the binding mode resembles most closely that of (S)-glutamate. The 3-hydroxy anion of (S)-4-AHCP is also hydrogen bonded to W1, as seen for (S)-2-Me-Tet-AMPA and (S)-glutamate, but no water molecule is observed corresponding to W4 in the (S)-AMPA complex. The glutamate binding mode of (S)-4-AHCP would be expected from the size of the 5-substituent of the isoxazole ring (the fused seven-membered ring) since the limited hydrophobic Fig. 3. Binding of (S)-4-AHCP to GluR2-S1- S2J (shown in stereo). (A) 2F o -F c electron density map of (S)-4-AHCP contoured at 1 r. The electron density was generated with program ARP ⁄ wARP and before the ligand was introduced into the model. Two conformations were modelled with confor- mation 1 of the ligand shown in magenta and conformation 2 in cyan. Additional F o -F c electron density (green and contoured at 3 r) was apparent after modelling only con- formation 1. (B) Selected residues of the ligand-binding site and their interactions with the ligand are shown. Dashed lines indicate hydrogen bonds ⁄ ionic interactions (< 3.3 A ˚ ). The red spheres represent water molecules, the nitrogen atoms are coloured blue, the oxygen atoms red and the sulphur atoms yellow. (C) Same as in A, but rotated )90° about a vertical axis. GluR2 ligand-binding core in complex with (S)-4-AHCP B. B. Nielsen et al. 1642 FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS space available in the pocket (formed mostly by D1) forces larger ligands closer to D2. The seven-membered ring of (S)-4-AHCP fills out some of the same space as the 2-methyl-tetrazole ring of (S)-2-Me-Tet-AMPA, but does not protrude as dee- ply into the pocket. However, unlike the tetrazole ring, the seven-membered ring is not planar and it occupies additional space towards the residues Glu402 from D1 and Thr686 from D2, constituting the lock between the two domains of the agonist-bound GluR2 ligand- binding core. Atoms C5, C6 and C7 in particular approach the lock, but without major disturbance. This interdomain interaction is still intact in the sense that a hydrogen bond is formed between the two resi- dues, however, the distance is somewhat longer (3.1 A ˚ ) than in the other three complexes ( 2.7 A ˚ ). The residues Leu650 and Met708 that are involved in hydrophobic interactions with the ligands show con- formational variability in the structures under discus- sion. The side chain of Leu650 is flipped  180° in (S)-4-AHCP (v 1 ¼ )87°; v 2 ¼ )167°) relative to the (S)-glutamate (v 1 ¼ 179°; v 2 ¼ 65°; mol C) (S)-2-Me- Tet-AMPA (v 1 ¼ 172°; v 2 ¼ 68°; mol B) and (S)- AMPA (v 1 ¼ 179°; v 2 ¼ 67°; mol C) complexes and adopts a conformation more like the one seen in the GluR2-S1S2J:kainate complex (v 1 ¼ )94°; v 2 ¼ 166°). Leu650 apparently adjusts its conformation for opti- mal hydrophobic contact with (S)-4-AHCP. The side chain of Met708 lends flexibility to the binding pocket; its conformation adjusting to fit various ligands [15,18–20]. In the (S)-4-AHCP complex, the tail of this side chain skirts the binding pocket to avoid clashing with the ligand. The Ce atom of Met708 points back into favourable van der Waals’ contact with the seven- membered ring of (S)-4-AHCP. (S)-4-AHCP is a partial agonist at GluR2 The D1–D2 domain closure in the GluR2-S1S2J:(S)-4- AHCP structure is 16.9° (relative to the structure of apo GluR2-S1S2J), which is less than the domain closure for full agonists ( 21°). The apparent explanation for this is twofold. Firstly (S)-4-AHCP has an extra carbon atom between the a-amino acid and the distal anionic moieties compared with other 3-hydroxy isoxazole analogues. Although conformational restriction shor- tens the distance between the a-amino acid group and the distal 3-hydroxy anion, the isoxazole is nonetheless pushed deeper into D2 than in the other three [(S)-glu- tamate (S)-AMPA and (S)-2-Me-Tet-AMPA] structures when adopting the conformation required for recogni- tion of a-amino acids. This favours intermediate domain closure. Secondly, driving the domains closer to each other would result in steric clashes between the bicyclical ring system of (S)-4-AHCP and the backbone atoms of Leu704. Also, the side chains of Tyr450, Thr655, Glu705 and Met708 would need to rearrange. In the present structure, the Ca atom of Leu704 is displaced 0.9 A ˚ (Fig. 4B) compared to its position in GluR2-S1S2J:(S)-glutamate. One consequence of the position of the isoxazole ring close to D2 is the hydro- gen bond formed between the isoxazole oxygen atom and the backbone nitrogen atom of Glu705 (Fig. 3C). This is different from other known structures of GluR2 in complex with isoxazole-based agonists. The GluR2-S1S2J:(S)-4-AHCP complex forms a dimer in the crystal as observed in all other agonist complexes reported. The dimer is generated by apply- ing crystallographic symmetry to the monomer that is observed in the asymmetric unit of the crystal. The dis- tance between the GT-linker (replacing the M1 and M2 transmembrane regions) of both protomers has been shown to be linearly related to the degree of domain closure [15,18,21]. In this structure, the Table 2. Residues in GluR2-S1S2J involved in ionic interactions and hydrogen bonds (< 3.3 A ˚ ) with (S )-4-AHCP. Residues and water molecules within 5 A ˚ from any ligand atom: Glu402,Tyr450, Pro478, Leu479, Thr480, Arg485, Leu650, Ser652, Gly653, Ser654, Thr655, Lys656, Thr686, Leu703, Leu704, Glu705, Met708, Tyr732, W1-W3 and W6-W10. Conformation 1 (A ˚ ) Conformation 2 (A ˚ ) Carboxylate oxygen O3 a Thr480 N 2.8 2.9 Arg485 Ng1 2.8 2.7 Carboxylate oxygen O4 Ser654 N 2.7 2.9 Ser654 Oc 3.2 (3.5) Arg485 Ng2 2.8 2.8 Ammonium group N2 Pro478 O 2.9 2.7 Thr480 Oc1 2.9 2.8 Glu705 Oe1 3.1 3.3 Glu705 Oe2 2.6 2.8 Isoxazole oxygen O1 Glu705 N 2.8 3.1 W3 b, c (3.7) 3.3 Isoxazole nitrogen N1 Thr655 Oc1 2.6 2.8 W2 d 3.0 2.9 3-hydroxy anion O2 Thr655 N 3.3 3.3 W1 e 2.6 2.9 a For atom numbering, see Fig. 1. b Numbering of water molecules as per Kasper et al. [19]. c W3 is further hydrogen bonded to the side chains of Thr686 and Tyr702. d W2 also interacts with the backbone of residues Leu650 and Leu703. e W1 is further connec- ted to the backbone of residues Ser652, Thr655 and Lys656. B. B. Nielsen et al. GluR2 ligand-binding core in complex with (S)-4-AHCP FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS 1643 distance is 34.1 A ˚ (between Ile633-Ile633 Ca-atoms). It has been suggested previously that the movement of D2 to close over the ligand causes conformational strain that is transferred to the ion channel, leading to pore opening [27]. (S)-4-AHCP has an efficacy of 0.38 at GluR2(Q) i relative to (S)-glutamate. The efficacy combined with the observed domain closure (16.9°) and D2–D2 linker separation imply that (S)-4-AHCP acts as a partial agonist at the AMPA-type receptor GluR2. Based on comparisons with structural and pharmacological studies on a range of other agonists [15,18,21], partial agonism was indeed expected from the observed degree of GluR2-S1S2J domain closure and D2–D2 linker separation. (S)-4-AHCP and receptor subtype selectivity (S)-4-AHCP displays selectivity for homomers of the low affinity kainate receptor subunit GluR5 over the AMPA receptors [26]. The most important differences in the ligand-binding site between GluR2 and GluR5 are the respective substitutions of Leu650 and Met708 in GluR2 to the smaller residues Val685 and Ser741 in GluR5 (numbering as in TrEMBL entry Q86SU9). In particular, mutagenesis studies have shown that the latter residue is responsible for the selectivity displayed by another GluR5 subtype selective agonist (S)- 2-amino-(5-tert-butyl-3-hydroxy-4-isoxa zolyl)propionic acid [28]. Recently, Armstrong et al. [29] reported that the mutation of Leu650 to Thr in GluR2 yields a receptor that responds more potently and efficaciously to the partial agonist kainate and less to the full agon- ist AMPA compared to unmodified GluR2. Also, the nonconserved residue at position 702 in GluR2 has been identified as the major contributor to the selectiv- ity of (S)-Br-HIBO for GluR1 (Tyr698) over GluR3 (Phe706) [30]. In GluR5, this residue is Leu735 and it may thus play a role in receptor selectivity. Thr686 A B Fig. 4. Comparison of the GluR2-S1S2J: (S)-4-AHCP complex with other agonist complexes. (A) Superposition of the struc- tures of GluR2-S1S2J in complex with: (S)-4-AHCP (grey) (S)-glutamate (yellow; pdb code 1FTJ, mol C) (S)-AMPA (cyan; pdb code 1FTM, mol C) and (S)-2-Me-Tet-AMPA (magenta; pdb code 1M5B, mol B), shown in stereo. Selected residues of the ligand- binding site are shown. Superimposition of the Ca atoms of D1 (residues 393–496 and 730–773) of the three structures on the (S)-4-AHCP complex resulted in rmsd of 0.36, 0.33 and 0.80 A ˚ , respectively. The spheres represent water molecules, the nitrogen atoms are coloured blue, the oxygen atoms red and the sulfur atoms yellow. (B) Ca-trace of the structures of GluR2-S1S2J:(S)-4-AHCP (grey) and of GluR2-S1S2 J:(S)-glutamate (yellow), super- imposed by Ca atoms of D1. (S )-4-AHCP (conformation 1) and the side chain of Leu704 are shown in ball-and-stick. GluR2 ligand-binding core in complex with (S)-4-AHCP B. B. Nielsen et al. 1644 FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS forms an interdomain interaction with Glu402, and is in close contact with atoms of the seven-membered ring system; this residue is replaced by the smaller Ser721 in GluR5. Homology modelling of the ligand-binding core of GluR5 and docking of (S)-4-AHCP to this model has shown that the binding site is larger than in GluR2; thus allowing the accommodation of larger and more bulky ligands [26]. A corresponding analysis based on the structure of GluR2-S1S2J in complex with (S)-4- AHCP supports this finding. Using single-channel recordings, domain closure has elegantly been shown to correlate with the open probability of discrete subconductance states of the channel and also with receptor desensitization [21]. Taken together, the smaller residues in GluR5 would probably allow for increased domain closure compared to that of GluR2, as well as a more complementary van der Waals’ environment, and this may explain the functional selectivity of (S)-4-AHCP towards GluR5. A fuller understanding of GluR5 selectivity awaits the publica- tion of the structure of a GluR5 construct in complex with an agonist. Experimental procedures Materials Chemicals were purchased from Sigma-Aldrich (Vallensbæk Strand, Denmark) unless otherwise specified. The synthesis of (S)-4-AHCP is as described by Brehm et al. [26]. Restric- tion and other molecular biological enzymes were obtained from New England BioLabs (Beverley, MA, USA). Protein expression and purification The GluR2-S1S2J construct described by Armstrong and Gouaux [15] was expressed, refolded and purified essentially as previously reported [17,31]. The rat AMPA receptor clone GluR2(Q) i within the vector pGEMHE [32] was used for preparation of high-expression cRNA transcripts. cDNA were grown in XL1 Blue bacteria (Stratagene, La Jolla, CA, USA) and prepared using column purification (Qiagen, Hilden, Germany). cRNA was synthes- ized from this cDNA using the mMessage mMachine T7 mRNA-capping transcription kit (Ambion Inc., Austin, TX, USA). Cell culture Sf9 insect cells were maintained in BaculoGold Max-XP serum-free medium (BD Biosciences, FranklinLakes, NJ, USA) according to standard manufacturers protocols. Receptor binding assay (S)-4-AHCP binding affinity at the GluR2-S1S2J soluble construct and at full-length GluR2(R) o was determined by a radioligand binding assay. Purified construct (0.1 lg protein) or Sf9 insect cell membranes (0.2–0.4 mg protein) expressing GluR2(R) o [33] were incubated with 2–4 nm (RS)- [5-methyl- 3 H]-AMPA (43.5 CiÆmmol )1 ; Perkin Elmer, Well- esley, MA, USA) in the presence of 1 nm)0.10 mm (S)-4- AHCP for 1–2 h on ice in assay buffer (50 mm Tris ⁄ HCl, 100 mm KSCN, 2.5 mm CaCl 2 , pH 7.2 at 4 °C; containing 10% glycerol for GluR2-S1S2J). Samples were filtered onto Millipore 0.22-lm GSWP nitrocellulose filters (for GluR2- S1S2J) or Whatman GF ⁄ B filters [for GluR2(R) o ]. Filters were washed twice with cold assay buffer and radioactivity was determined by scintillation counting. Data were analysed using grafit v3.00 (Erithacus Software Ltd, Horley, UK) and fit as previously described [34] to determine Hill coeffi- cient and K i . The K d values of [ 3 H]AMPA at GluR2-S1S2J (12.8 nm) and GluR2(R) o (16.8 nm) were determined previ- ously [18,33]. Electrophysiology All frog experimental procedures are approved by the Experimental Animal Committee, The Danish Ministry of Justice, Copenhagen, Denmark (2004/561-876-C10). Mature female X. laevis (African Reptile Park, Tokai, South Africa) were anesthetized using 0.1% ethyl 3-aminobenzoate, meth- anesulfonic acid salt (tricaine methanesulfonic acid salt) by transdermal administration and ovaries were surgically removed. The ovarian tissue was dissected and treated with 1mgÆmL )1 collagenase in nominally Ca 2+ -free Barth’s medium for 2 h at room temperature. On the second day, oocytes were injected with 50 nL ( 1 lgÆlL )1 ) cRNA and incubated in Barth’s medium (88 mm NaCl, 1 mm KCl, 0.33 mm Ca(NO 3 ) 2 , 0.41 mm CaCl 2 , 0.82 mm MgSO 4 , 2.4 mm NaHCO 3 ,10mm Hepes, pH 7.4) with 0.1 mgÆmL )1 gentamicin and 1% penicillin–streptomycin (Life Technol- ogies) at 17 °C. Oocytes were typically used for recordings from 3 to 10 days postinjection and were voltage-clamped with the use of a two-electrode voltage clamp (GeneClamp 500B, Axon Instruments, Union City, CA, USA) with both microelectrodes filled with 3 m KCl. Recordings were made while the oocytes were continuously superfused with nomin- ally Ca 2+ -free frog Ringer’s solution (115 mm NaCl, 2 mm KCl, 1.8 mm BaCl 2 ,5mm Hepes, pH 7.0). Drugs were dis- solved in Ca 2+ -free frog Ringer’s solution and added by bath application. Recordings were made at room tempera- ture at holding potentials in the range of )80 to )20 mV. For efficacy measurements, (S)-4-AHCP was applied at a saturating concentration (500 lm) in the presence of 100 lm cyclothiazide in order to block receptor desensitiza- tion (cyclothiazide EC 50 : GluR2(Q) i ¼ 7.6 lm [35]). Control B. B. Nielsen et al. GluR2 ligand-binding core in complex with (S)-4-AHCP FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS 1645 stimulations with 1 mm (S)-glutamate plus 100 lm cyclothi- azide were performed immediately prior to, and after (S)-4- AHCP application, with a washout period of 5–10 min between drug applications. The two control (S)-glutamate stimulations were each no more than 1–9% different from the mean value. Cyclothiazide (100 lm) was preapplied alone for 1 min before each agonist application. The (S)-4- AHCP maximum response was then expressed as a fraction of the mean value of the two test (S)-glutamate stimula- tions. Data analysis of pharmacology Student’s t-test was used for comparison of K i values using sigmastat for Windows v3.0 (SPSS Inc., Chicago, IL, USA). Values are given as mean ± SEM and were consid- ered statistically significantly different if P < 0.05. Concen- tration–response curves for agonists were analysed using grafit v3.00 to determine the EC 50 and Hill value (n H ), using Eqn (1), where I is the measured current and I max is the maximal steady-state current. I ¼ I max =ð1 þ 10 ðlog½EC 50 Þ=10 ðlog½AgonistÞ Þ n H ð1Þ Co-crystallization of GluR2-S1S2J with (S)-4-AHCP The GluR2-S1S2J protein was dialysed extensively in the buffer used for crystallization (10 mm Hepes pH 7.0, 20 mm NaCl, 1 mm EDTA) and concentrated to 6 mgÆmL )1 . The GluR2-S1S2J was mixed with (S)-4-AHCP at a ratio of 1 : 49. Crystals were obtained at 6 °C by the hanging drop vapour diffusion method using a reservoir solution contain- ing 0.2 m lithium sulfate, 0.1 m phosphate–citrate buffer pH 4.5 and 20% PEG 3350. Crystals were transferred through a cryo-protectant solution consisting of 4.7 mm ligand and 12% glycerol in reservoir solution prior to flash-cooling. X-ray data collection The X-ray diffraction data were collected from one crystal at 100 K and at a wavelength of 0.811 A ˚ using a MAR CCD detector at beamline X11 (DESY, Hamburg, Ger- many). The crystal diffracted to 1.75 A ˚ . The HKL package (Denzo and Scalepack) [36] was used for autoindexing and data processing, for statistics see Table 1. Structure determination and refinement The structure was solved using molecular replacement with amore [37] implemented in the ccp4i package [38]. The protein atoms of the structure of GluR2-S1S2J complexed with (S)-Br-HIBO ([18]; pdb code 1M5C), was used as a search model. Only one solution to both the rotation- and translation function was obtained. The program arp ⁄ warp [39] was used for tracing the majority of the structure. Refinements alternating with manual model building were performed using the programs refmac5 [40] and o [41], respectively. The electron density corresponding to (S)-4-AHCP was well defined and allowed unambiguous positioning of two different conformations of the ligand (refined with equal occupancy; see Fig. 3A). Initially (S)-4-AHCP was mod- elled as a single conformation; however, additional F o –F c difference electron density was present, indicating conform- ationally enantiomeric puckering of the seven-membered ring of the ligand. The atoms of the seven-membered ring were built and refined in two conformations but additional density still appeared. Therefore, two conformations (inclu- ding all ligand atoms) were modelled and all atoms refined with half occupancy. This resulted in the disappearance of the additional difference electron density. A monomer library description of the ligand for REF- MAC5 has been created. (S)-4-AHCP was built as the tri- ion and submitted to conformational analysis using the MMFFs force field with GB-SA treatment of solvation in macromodel 8.1 [42]. Three distinct low energy conform- ers were chosen to represent the repertoire of the ring and truncated systems were built (conformers of 4-ethyl-7,8- dihydro-6H-cyclohepta[d]isoxazol-3-ol anion). These were minimized using Density Functional Theory [B3LYP ⁄ 6– 311 + G(d,p)] in gaussian¢03 [43] to give highly accurate co-ordinates for the ring system. The amino-acid group was rebuilt from the ethyl side chain and the resulting glycine moiety was re-minimized with MMFFs ⁄ GB-SA with the other atoms frozen in the positions determined by quantum chemistry. The coordinates of one of the low energy con- formations of (S)-4-AHCP were used for the library des- cription. Water molecules as well as two sulfate ions and three glycerol molecules were included as refinement progressed. The refined structure comprises (using the numbering of full-length membrane bound receptor without signal pep- tide, Swiss-Prot entry P19491) residues 392–506, the GT linker and residues 632–774, as well as two additional N-terminal residues. For refinement statistics (Table 1). The coordinates of the GluR2-S1S2J structure in complex with (S)-4-AHCP have been deposited in the RCSB Protein Data Bank with accession code 1WVJ. Structure analysis and figure preparation The hingefind script [44] implemented in the program vmd [45] was used to calculate the ligand-induced domain closure relative to the apoGluR2-S1S2J structure (pdb code 1FTO, mol A). The CCP4 program contacts was used in the analysis of protein–ligand interactions. The programs molscript [46], raster3d [47] and bobscript [48] were used in the preparation of figures. GluR2 ligand-binding core in complex with (S)-4-AHCP B. B. Nielsen et al. 1646 FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS Acknowledgements L. B. Sørensen is kindly acknowledged for technical assistance. This work was supported by: DANSYNC (Danish Centre for Synchrotron Based Research); the Danish Medical Research Council; the Novo Nordisk Foundation; the Lundbeck Foundation; the computing resources of the Australian Centre for Advanced Com- puting and Communications as well as the Danish Center for Scientific Computing; and the European Community – Access to Research Infrastructure Action of the Improving Human Potential Programme to the EMBL Hamburg Outstation, contract number HPRI-CT-1999-00017. References 1 Riedel G, Platt B & Micheau J (2003) Glutamate recep- tor function in learning and memory. Behav Brain Res 140, 1–47. 2 Dingledine R, Borges K, Bowie D & Traynelis SF (1999) The glutamate receptor ion channels. Pharmacol Rev 51, 7–61. 3 Bra ¨ uner-Osborne H, Egebjerg J, Nielsen EØ, Madsen U & Krogsgaard-Larsen P (2000) Ligands for glutamate receptors: Design and therapeutic prospects. J Med Chem 43, 2609–2645. 4 Rosenmund C, Stern-Bach Y & Stevens CF (1998) The tetrameric structure of a glutamate receptor channel. Science 280, 1596–1599. 5 Mano I & Teichberg VI (1998) A tetrameric subunit stoichiometry for a glutamate receptor-channel complex. Neuroreport 9, 327–331. 6 Sun Y, Olson R, Horning M, Armstrong N, Mayer M & Gouaux E (2002) Mechanism of glutamate receptor desensitization. Nature 417, 245–253. 7 Tichelaar W, Safferling M, Keinanen K, Stark H & Madden DR (2004) The three-dimensional structure of an ionotropic glutamate receptor reveals a dimer-of- dimers assembly. J Mol Biol 344, 435–442. 8 Bowie D & Lange GD (2002) Functional stoichiometry of glutamate receptor desensitization. J Neurosci 22, 3392–3403. 9 Robert A, Irizarry SA, Hughes TE & Howe JR (2001) Subunit interactions and AMPA receptor desensitiza- tion. J Neurosci 21, 5574–5586. 10 Smith C, Lu-Yang Wang L-W & Howe JR (2000) Het- erogeneous conductance levels of native AMPA recep- tors. J Neurosci 20, 2073–2085. 11 Ayalon G & Stern-Bach Y (2001) Functional assembly of AMPA and kainate receptors is mediated by several dis- crete protein–protein interactions. Neuron 31, 103–113. 12 Nishimura S, Iizuka M, Wakamori M, Akiba I, Imoto K & Barsoumian EL (2000) Stable expression of human homomeric and heteromeric AMPA receptor subunits in HEK293 cells. Receptors Channels 7, 139–150. 13 Fleck MW, Bahring R, Patneau DK & Mayer ML (1996) AMPA receptor heterogeneity in rat hippocampal neurons revealed by differential sensitivity to cyclothia- zide. J Neurophysiol 75, 2322–2333. 14 Mansour M, Nagarajan N, Nehring RB, Clements JD & Rosenmund C (2001) Heteromeric AMPA receptors assemble with a preferred subunit stoichiometry and spatial arrangement. Neuron 32, 841–853. 15 Armstrong N & Gouaux E (2000) Mechanisms for acti- vation and antagonism of an AMPA-sensitive glutamate receptor: Crystal structures of the GluR2 ligand binding core. Neuron 28, 165–181. 16 Kuusinen A, Arvola M & Keina ¨ nen K (1995) Molecular dissection of the agonist binding site of an AMPA receptor. EMBO J 14, 6327–6332. 17 Chen GQ & Gouaux E (1997) Overexpression of a glu- tamate receptor (GluR2) ligand binding domain in Escherichia coli: Application of a novel protein folding screen. Proc Natl Acad Sci USA 94, 13431–13436. 18 Hogner A, Kastrup JS, Jin R, Liljefors T, Mayer ML, Egebjerg J, Larsen IK & Gouaux E (2002) Structural basis for AMPA receptor activation and ligand selectiv- ity: Crystal structures of five agonist complexes with the GluR2 ligand-binding core. J Mol Biol 322, 93–109. 19 Kasper C, Lunn M-L, Liljefors T, Gouaux E, Egebjerg J & Kastrup JS (2002) GluR2 ligand-binding core com- plexes: Importance of the isoxazolol moiety and 5-sub- stituent for the binding mode of AMPA-type agonists. FEBS Lett 531, 173–178. 20 Lunn M-L, Hogner A, Stensbøl TB, Gouaux E, Egebj- erg J & Kastrup JS (2003) Three-dimensional structure of the ligand-binding core of GluR2 in complex with the agonist (S) -ATPA: Implications for receptor subu- nit selectivity. J Med Chem 46, 872–875. 21 Jin R, Banke TG, Mayer ML, Traynelis SF & Gouaux E (2003) Structural basis for partial agonist action at iono- tropic glutamate receptors. Nat Neurosci 6, 803–810. 22 Jin R, Horning M, Mayer ML & Gouaux E (2002) Mechanism of activation and selectivity in a ligand- gated ion channel S: structural and functional studies of GluR2 and quisqualate. Biochemistry 41, 15635–15643. 23 Hogner A, Greenwood JR, Liljefors T, Lunn M-L, Egebjerg J, Larsen IK, Gouaux E & Kastrup JS (2003) Competitive antagonism of AMPA receptors by ligands of different classes: Crystal structure of ATPO bound to the GluR2 ligand-binding core, in comparison with DNQX. J Med Chem 46, 214–221. 24 Mayer ML & Armstrong N (2004) Structure and function of glutamate receptor ion channels. Annu Rev Physiol 66, 161–181. 25 Krogsgaard-Larsen P, Nielsen EO & Curtis DR (1984) Ibotenic acid analogues. Synthesis and biological and B. B. Nielsen et al. GluR2 ligand-binding core in complex with (S)-4-AHCP FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS 1647 in vitro activity of conformationally restricted agonists at central excitatory amino acid receptors. J Med Chem 27, 585–591. 26 Brehm L, Greenwood JR, Hansen KB, Nielsen B, Egebjerg J, Stensbøl TB, Bra ¨ uner-Osborne H, Sløk FA, Kronborg TT & Krogsgaard-Larsen P (2003) (S) -2- Amino-3-(3-hydroxy-7,8-dihydro-6H-cyclohepta[d]isoxa- zol-4-yl) propionic acid, a potent and selective agonist at the GluR5 subtype of ionotropic glutamate receptors. Synthesis, modeling, and molecular pharmacology. J Med Chem 46, 1350–1358. 27 Abele R, Keina ¨ nen K & Madden DR (2000) Agonist- induced isomerization in a glutamate receptor ligand- binding domain. A kinetic and mutagenetic analysis. J Biol Chem 275, 21355–21363. 28 Nielsen MM, Liljefors T, Krogsgaard-Larsen P & Egeb- jerg J (2003) The selective activation of the glutamate receptor GluR5 by ATPA is controlled by serine 741. Mol Pharmacol 63, 19–25. 29 Armstrong N, Mayer M & Gouaux E (2003) Tuning activation of the AMPA-sensitive GluR2 ion channel by genetic adjustment of agonist-induced conformational changes. Proc Natl Acad Sci USA 100, 5736–5741. 30 Banke TG, Greenwood JR, Christensen JK, Liljefors T, Traynelis SF, Schousboe A & Pickering DS (2001) Iden- tification of amino acid residues in GluR1 responsible for ligand binding and desensitization. J Neurosci 21, 3052–3062. 31 Chen GQ, Sun Y, Jin R & Gouaux E (1998) Probing the ligand binding domain of the GluR2 receptor by proteolysis and deletion mutagenesis defines domain boundaries and yields a crystallizable construct. Protein Sci 7, 2623–2630. 32 Liman ER, Tytgat J & Hess P (1992) Subunit stoichio- metry of a mammalian K+ channel determined by con- struction of multimeric cDNAs. Neuron 9, 861–871. 33 Coquelle T, Christensen JK, Banke TG, Madsen U, Schousboe A & Pickering DS (2000) Agonist discrimi- nation between AMPA receptor subtypes. Neuroreport 11, 2643–2648. 34 Nielsen B, Banke TG, Schousboe A & Pickering DS (1998) Pharmacological properties of homomeric and heteromeric GluR1o and GluR3o receptors. Eur J Phar- macol 360, 227–238. 35 Quirk JC & Nisenbaum ES (2003) Multiple mole- cular determinants for allosteric modulation of alter- natively spliced AMPA receptors. J Neurosci 23, 10953– 10962. 36 Otwinowski Z & Minor W (1997) Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol 276, 307–326. 37 Navaza J (1994) AMoRe: An atomated package for molecular replacement. Acta Crystallogr A50, 157– 163. 38 Collaborative Computational Project Number 4 (1994) The CCP4 suite: Programs for protein crystallography. Acta Crystallogr D50, 760–763. 39 Perrakis A, Morris R & Lamzin VS (1999) Automated protein model building combined with iterative structure refinement. Nat Struct Biol 6, 458–463. 40 Murshudov GN, Vagin AA & Dodson EJ (1997) Refinement of macromolecular structures by the maxi- mum-likelihood method. Acta Crystallogr D53, 240– 255. 41 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991) Improved methods for building protein models in elec- tron density maps and the location of errors in these models. Acta Crystallogr A47, 110–119. 42 Schro ¨ dinger Inc. (2003) Macromodel. 8.1. Schro ¨ dinger Inc, 1500 S. W. First Avenue, Suite 1180 Portland, OR 97201, USA. 43 Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Montgomery JA Jr, Vreven T, Kudin KN, Burant JC et al. (2003) Gaussian 03, Revision B.02. Gaussian, Inc, Pittsburgh, PA. 44 Wriggers W & Schulten K (1997) Protein domain move- ments: Detection of rigid domains and visualization of hinges in comparisons of atomic coordinates. Proteins 29, 1–14. 45 Humphrey W, Dalke A & Schulten K (1997) VMD – Visual Molecular Dynamics. J Mol Graphics 14, 33–38. 46 Kraulis PJ (1991) MOLSCRIPT A: a program to pro- duce both detailed and schematic plots of protein struc- tures. J Appl Crystallogr 24, 946–950. 47 Merritt EA & Murphy MEP (1994) Raster3d, Version 2.0. A Program for Photorealistic Molecular Graphics. Acta Crystallogr D50, 869–873. 48 Esnouf RM (1999) Further Additions to Molscript, Version 1.4. including reading and contouring of electron-density maps. Acta Crystallogr. D55, 938–940. 49 Kleywegt GJ & Jones TA (1996) Phi ⁄ psi-chology: Ramachandran revisited. Structure 4, 1395–1400. GluR2 ligand-binding core in complex with (S)-4-AHCP B. B. Nielsen et al. 1648 FEBS Journal 272 (2005) 1639–1648 ª 2005 FEBS . Exploring the GluR2 ligand-binding core in complex with the bicyclical AMPA analogue (S)-4-AHCP Bettina B. Nielsen 1 , Darryl S. Pickering 2 , Jeremy R. Greenwood 1 ,. 2005) doi:10.1111/j.1742-4658.2005.04583.x The X-ray structure of the ionotropic GluR2 ligand-binding core (GluR2- S1S2J) in complex with the bicyclical AMPA analogue (S)-2-amino-3-(3-hyd- roxy-7,8-dihydro-6H-cyclohepta[d]-4-isoxazolyl)propionic. substituents to increase affinity. Here, we present the first structure of GluR2- S1S2J in complex with the bicyclical AMPA analogue (S)-4-AHCP, as well as the pharmacology of (S)-4-AHCP binding to GluR2- S1S2J