1 Full Length Paper Modelling the interaction of catecholamines with the α1A Adrenoceptor - towards a ligand-induced receptor structure Gemma K. Kinsella1, Isabel Rozas1,2, Graeme W. Watson1,* 1Department of Chemistry, 2 Centre for Synthesis and Chemical Biology, University of Dublin, Trinity College, Dublin 2, Ireland. *To whom correspondence should be addressed. Phone: +353 1 608 1357, Fax: +353 1 671 2826, E-mail: watsong@tcd.ie Abstract: Adrenoceptors are members of the important G protein coupled receptor family for which the detailed mechanism of activation remains unclear. In this study, we have combined docking and molecular dynamics simulations to model the ligand induced effect on an homology derived human α1A adrenoceptor. Analysis of agonist / α1A adrenoceptor complex interactions focused on the role of the charged amine group, the aromatic ring, the N-methyl group of adrenaline, the beta hydroxyl group and the catechol meta and para hydroxyl groups of the catecholamines. The most critical interactions for the binding of the agonists are consistent with many earlier reports and our study suggests new residues possibly involved in the agonist-binding site, namely Thr-174 and Cys-176. We further conclude that a number of changes occur when moving from agonist binding to G protein activation, including a movement of TM-V away from 2 TM-III and a change in the interactions of Asp-123 of the conserved DRY motif. This may cause Arg-124 to move out of the TM helical bundle and change the orientation of residues in IC-II and IC-III, allowing for increased affinity of coupling to the G- protein. Keywords: α1A-adrenoceptor; Agonists; Molecular docking; Molecular Dynamics; Receptor Activation. Introduction The α1 adrenoceptors (α1-AR), which belong to Class A of the super family of G protein coupled receptors (GPCRs), are of particular therapeutic interest due to their important roles in the control of blood pressure and in the contraction and growth of smooth muscle. More specifically the α1A adrenoceptor (α1A-AR) located in large abundance in the prostate is thought to be influential in the condition Benign Prostatic Hyperplasia (BPH), a considerable health problem for aging men [1]. Class A GPCRs consist of a heptahelical bundle (TM I-VII), which transverses the cell membrane. They share key structural features including a disulfide bond between TM-III and the extracellular (EC) region, a tripeptide Glu/Asp-Arg-Tyr (E/DRY) motif located at the intracellular end of TM-III and a common Asn-Pro-X- X-Tyr (NPxxY) motif in TM-VII [2]. To advance our understanding of the molecular structure of the α1A-AR we recently built a homology model based on the 1l9h [3] crystal structure of bovine rhodopsin in the inactive state [4]. 3 Mutagenesis studies suggest specific binding interactions of the endogenous catecholamines, adrenaline (AD) and noradrenaline (ND), with the α1A-AR [5,6,7,8]. These include an interaction between the amino group of the catecholamines and Asp- 106 in TM-III; and with Ser-188 and Ser-192 of TM-V [9,10]. Once GPCRs receive an agonist stimulus at their extracellular side they transfer the signal across the cell membrane and initiate a variety of intracellular biochemical events. However, an activation mechanism has only been postulated for a few Class A receptors including the α2A–AR and β2-AR [11,12]. Agonist binding may induce specific rearrangements in Pro kink regions or rigid body motion of TM-V and TM-VI [13]. Experimental studies also support a movement of TM-VI relative to TM-III as a consequence of agonist induced receptor activation. Obtaining structural information about agonist/α1A-AR complexes has been hampered by the lack of a GPCR crystal structure in the activated state [14]. The direct use of the Rhodopsin crystal structure as a template for modelling activation remains questionable. The crystal structure is a snapshot of Rhodopsin in its inactive state; while upon agonist binding an active form will be induced. To address this issue Bissantz et al. [15] attempted to produce an activated form of other GPCRs such as dopamine D3, the β2-AR and the δ opioid receptors, by manually rotating TM-VI anticlockwise by 30o around its helical axis. Chambers et al. [16], developed an active form of the 5-HT2A receptor through homology modelling using a theoretically active template of rhodopsin obtained by isomerisation of the 11-cis-retinal chromophore followed by constrained MD. However, Carmine et al. [17], suggested that the agonist-binding pocket of the β2-AR is not rigid but is dynamically formed as the ligand builds an increasing number of contacts with the receptor. Therefore, as the 4 receptor is “en route” to the active form, it could also change the configuration of the binding cavity as the key agonist binding interactions and the active receptor conformation develop in parallel. Through molecular modelling studies we aim to facilitate the integration of the available experimental observations and biophysical data into a scheme to examine receptor structure and function. Hence, in this paper, we present a computational strategy to investigate the ligand-induced conformation of the α1A–AR. Computational methods Our computational approach to model the α1A–AR conformation induced by the interaction with a ligand involves: 1) Docking of agonists (AD and ND) to examine their interactions with our previously developed homology-derived model of the α1A-AR [4]. 2) Molecular Dynamics (MD) simulations were performed in a H2O/CHCl3/H2O membrane mimic to refine the agonist/α1A-AR complexes. We explore the structural changes during the dynamics and monitor the resultant conformational changes to the receptor. 3) Redocking of the agonists into the ‘agonist-induced’ receptor form developed in step 2. 5 4) Further MD simulations of the agonist/α1A-AR complexes to determine agonist binding modes and the agonist induced receptor structures. A repeated cycling of steps 3 and 4 can be used to further refine the binding site. Molecular Docking Docking studies between the catecholamines and our α1A-AR model [4] were performed using the programs Dock 4.0 [18] and FlexiDock [19]. Both agonists were protonated and charges were taken from the electrostatic potential obtained during Density Functional Theory (DFT) structural optimisations using the B3LYP hybrid functional and the 6-31G* basis set in Gaussian 98 [20]. The binding site was defined as all residues within 10 Å of Asp-106 in TM-III and Ser-188 and Ser-192 in TM-V, as suggested in mutagenesis studies [6,7,8]. The agonists were divided into rigid fragments and conformations were built into the receptors binding pocket (incremental construction). The docking modes were further optimised using a genetic algorithm to determine the optimum geometry. The parameters used in this work were the default parameters of Dock 4.0 and FlexiDock. Molecular Dynamics Simulations All MD simulations were performed with Amber 7.0 [21] using a time step of 1 fs, and the Amber force field. The simulation cell was heated gradually to 300 K over 5 ps with equilibration performed using backbone restraints for 5 ps at each of 6 15, 10 and 5 kcal mol-1 followed by 65 ps without restraints. Periodic boundary conditions were applied in all three dimensions with the Particle Mesh Ewald (PME) method being used to treat the long-range electrostatic interactions. Non-bonded interactions were calculated for 1-4 interactions and higher using a cutoff radius of 9 Å. Further analysis was performed with the Carnal module of Amber 7.0 and the Gromacs (v3.1.4) tools [22]. Root mean square deviations (RMSDs) were calculated for the backbone heavy atoms after fitting the Cα carbons to the starting homology structure as a reference. Amber was also used to optimise the docked and final dynamical structures using steepest descent (250 steps) and conjugate gradient (750 steps) methods. Following a similar procedure to our previous study, the dynamical behaviour of the agonist/receptor complexes in a membrane mimic (H2O/CHCl3/H2O) were examined [4]. Such an environment allows the simulations to sample more conformational space than a phospholipid model, while being more appropriate than a gas phase environment [4,23]. Counter-ions (Cl-) were added to ensure a charge neutral cell, by replacing solvent molecules at sites of high electrostatic potential. For AD and ND the atom types and parameters were obtained from the General Amber Force Field (GAFF) [24]. Two angular terms, ca-c3-oh (67.1 kcal.mol-1, 110.7o) and ca-c3-h1 (47.3 kcal.mol-1, 109.1o), were added to GAFF by analogy to similar parameters already present. In addition, two improper dihedral terms, (ca-ca-ca-ha, ca-ca-ca-oh) were added to enforce the planarity of the aromatic rings (1.1 kcal.mol-1, 180.0o). These parameters were tested by comparison of the geometry and vibrational frequencies of experiment and DFT calculation [25]. The 7 parameters were further tested by performing MD simulations (150 ps) in water and the structures were found to be dynamically stable. Results and Discussion Analysis of Catecholamine/α1A-AR Complexes Both agonists were docked into our homology model of the α1A-AR. For the adrenaline complex, interactions were formed between the protonated nitrogen and Asp-106 (3.23 Å) and between the meta hydroxyl and Ser-188 (2.16 Å). For the noradrenaline complex, the protonated nitrogen to Asp-106 distance is long at 6.04 Å, while the para hydroxyl is 2.93 Å from Ser-188. These initial complexes provide starting structures for further structural refinement through MD simulations (1 ns). While little structural change occurred over the optimisation step, a marked RMSD rise was observed for both simulations over the heating step of the equilibration runs, as the receptor structures moves from a ‘rhodopsin-like’ conformation (see Figure 1). Over both simulations small structural changes were observed for the short TM-VIII, while the largest deviations were, for the AD complex, TM-VI at 3.30 Å and for the ND complex, TM-IV at 3.04 Å. The time dependent RMSD data of the seven TM helices were next examined for further structural changes over the MD production runs (see trendlines in Figure 1). For the adrenaline complex, TM-I underwent a decrease in RMSD over 100 - 300 ps of the production run; TM-VII underwent a structural change between 300 – 400 ps 8 while TM-IV changed between 600 – 800 ps (see Figure 1a). A HB interaction forms between Ser-188 and the meta hydroxyl after 200 ps, which corresponds to a slight movement of TM-V at this time. For the noradrenaline complex, most of the structural changes occur in the region of 200 – 600 ps including marked rises in RMSD for TM-IV and TM-V (Figure 1b). Hence, the RMSD trendlines over the production runs for both simulations indicate that a structural rearrangement of the receptor has occurred due to the presence of an agonist, which has stabilised by 800 ps of simulation. Structural changes are observed for TM-V and TM-VI in both simulations in agreement with experimental postulates of such helical movements [12]. Throughout the 1 ns simulations, interactions between the positively charged amine of the agonists and the negatively charged side chain carboxyl of Asp-106 in TM-III, anchors the catecholamines in the binding pocket. The second extracellular loop (EC-II), moves further into the TM framework allowing for the participation of EC-II residues (Thr-174, Cys-176, Gln-177 and Ile-178) in the binding site as observed by Pedretti et al. [26]. The final properties of the complexes were averaged over the last 200 ps of simulation and optimised. Both the receptor structures of these complexes undergo a loss of inter-helical interactions and hydrogen bonds (HB) relative to the uncomplexed structure [4]. However, the AD complex experienced a larger movement of TM-VI, which is thought to be important for activation coupled with a lengthening of the HB between Asp-123 and Arg-124 of the DRY motif. The AD 9 complexed receptor structure was therefore chosen for redocking of adrenaline and crossdocking of noradrenaline. Further Docking of Catecholamines utilising the agonist induced α1A-AR For the redocked adrenaline and crossdocked noradrenaline complexes, the HB distance from the protonated nitrogen to the carboxylic group of Asp-106 was 2.43 Å and 1.42 Å, respectively. For the adrenaline complex, the protonated nitrogen also interacts with Thr-174 at a distance of 3.07 Å. For the noradrenaline complex, the beta hydroxyl is orientated towards Asp-106 at a distance of 3.46 Å, while no other close hydroxyl interactions were determined in either docked conformation. Finally, the N-methyl group of adrenaline is close to Cys-176 (3.68 Å) and Tyr-316 (2.55 Å). Unlike the previously docked complexes, no interactions were formed with the Ser residues of TM-V and novel interactions have formed with EC-II residues. To optimise the agonist binding modes, further MD simulations (2 ns) were performed on both complexes. Further Agonist Induced Changes to the α1A-AR Once again the optimisation step causes almost no change in the position of the atoms as illustrated by the small RMSDs (Figure 2), while the largest rise in RMSD is over the heating period. Over the production run a maximum Cα RMSD of 2.59 Å was observed for the adrenaline complex and a higher value of 3.59 Å for the noradrenaline complex. The helical RMSD deviations are smaller than for the original 10 MD simulations (see Figure 1), as the structural rearrangement from a ‘rhodopsin- like’ conformation occurred over the original MD runs. Specifically, for the AD/α1A-AR complex simulation, there is a gentle structural drift for TM-I, which has stabilised by 1.5 ns. For TM-VI, there is a structural change at 1.6 ns, which again appears to stabilise quickly. For the ND/α1A- AR complex simulation, there were no structural changes of significance with largely stable helical RMSDs throughout the simulation. The overall analysis indicates that there are no structural changes of significance, which signifies that most changes induced by the ligand have already occurred. The HB interactions have stabilized by ~1 ns and the following HB analysis was performed over the second ns of the simulations. For the AD complex, the protonated nitrogen interacts frequently with Asp-106 (66.7 %) and Cys-176 (34.3 %) of EC-II. Initially the meta hydroxyl interacts with Ser-188 but it also forms an interaction with Ser-192 by this time period. The N-methyl group forms occasional interactions with Trp-102 (22.9 %) and Thr-174 (10.9 %). For the ND complex, similar interactions occur for the protonated nitrogen with Asp-106 (50.5 %) and Cys- 176 (35.9 %) with a lengthening in the region of 1.6 ns. The nearby beta hydroxyl also forms a strong interaction with Thr-174 (93.2 %). Final Binding Mode Analysis of the catecholamine/α1A-AR Complexes The final properties of the two complexed structures, averaged over the last 200 ps of simulation, were optimised and examined. Both receptor structures proved 11 to be stereochemically satisfactory, with 316 residues (99.3%) in allowed regions with a further two loop residues in disallowed regions for the AD complex; while for the ND complex, four residues were in disallowed regions, three of which were loop residues. Different HB interactions were observed for the two catecholamine agonists as they induced different structural variations to the binding site. The final adrenaline binding mode (Figure 3a) had both, the beta hydroxyl (1.61 Å) and the protonated nitrogen (1.78 Å) orientated towards the carboxylate group of Asp-106, while the N- methyl group is close to Thr-174 (2.62 Å). Two Ser interactions were also formed in the final bound mode where the meta hydroxyl is close to Ser-188 (1.80 Å) and the para hydroxyl is 1.94 Å from Ser-192. For the noradrenaline complex (see Figure 3b) the main interaction formed was between the protonated nitrogen and Asp-106 at 1.80 Å and the beta hydroxyl with Asp-106 at 1.9 Å. It appears that Noradrenaline has twisted in the active site allowing the para hydroxyl to be orientated away from the Ser residues in the final binding mode. No other close hydroxyl interactions were determined. The RMSD of the active site residues of the adrenaline complex (with respect to the original homology model), shows that Asp-106 (1.26 Å) reorientates for improved interactions over the course of the simulation. Also Thr-174 of EC-II experiences an RMSD of 1.07 Å as the loop engages in interactions in the binding site. Notably, the RMSD analysis indicated a considerably smaller effect of the agonist on Ser-188 (0.08 Å) and Ser-192 (0.16 Å). For the final noradrenaline complex, larger agonist effects were exerted on Asp-106 (RMSD of 1.31 Å) and Thr- 12 174 (1.25 Å), than for the adrenaline complex. Again there is only a slight change induced on Ser-188 (0.39 Å) and Ser-192 (0.18 Å). In both simulations there are three aromatic residues (Trp-285, Phe-288 and Phe-289) in TM-VI clustered around the binding site (see Figure 4). Interactions of agonists with residues in this aromatic cluster are hypothesised to further induce or stabilize an altered configuration of the side chains within this cluster that may promote receptor activation, consistent with the findings of Visiers et al. [27]. Comparison Between Inactive and Active α1A-AR Models So far, comparisions have been made with respect to our homology model with which we obtained the initial binding modes through docking. This has allowed us to examine the degree of rearrangement during the dynamics and hence determine when structural change has occurred. However, determination of the differences between the inactive and active forms of the receptor requires comparisions with an inactive form which has been equilibrated under the same conditions [4]. To quantify the overall structural effect caused by the agonists the RMSDs were examined. The Cα RMSD differences, are similar at 5.68 Å for the AD complex and at 5.25 Å for the ND complex (Table 1). In terms of the helical regions, TM-I has the highest RMSD difference for the adrenaline complex to the uncomplexed at 4.53 Å, while for the noradrenaline complex it is TM-VII at 4.20 Å. This indicates that an overall structural change has occurred due to the presence of the agonists. 13 For the uncomplexed receptor, the binding site has interactions formed between Ser-192 and both Asp-106 and Ser-188, which do not occur for the activated receptors. Hence, the binding site is closed in the inactive receptor form and can not accommodate an agonist, highlighting the need for our protocol to develop an ‘agonist-induced’ receptor structure. Role of Helical Movement in Receptor Activation We observe upon agonist binding a movement of TM-V away from TM-III on the intracellular side, with a subsequent repositioning of the Ser residues of TM-V (see Figure 5).28,29 In addition, flexibility about a Pro-kink results in the movement of the cytoplasmic end of TM-VI away from TM-III, which is consistent with studies of the β2-AR [30,31]. Such movement indicates that conformational switches in TM α- helices can be generated via Pro containing motifs that form flexible molecular hinges as postulated by Sansom et al. [32]. Our α1A-AR results are consistent with a movement of TM-VI away from TM-III, which was identified for Rhodopsin [33,34] and in the β2-AR [35,36,37], indicating similar activation mechanisms. Role of the conserved motifs DRY and NPxxY in Receptor Activation In the uncomplexed α1A-AR structure, Asp-123 of the DRY motif donates to both neighbouring Arg-124 and Tyr-125. Arg-124, a residue entirely conserved in all Rhodopsin-like GPCRs, also interacts with residues in TM-I, TM-III and IC-III. However, for the agonist complexed receptors, novel interactions are formed with Asp-123 donating to Ile-119 and in the noradrenaline induced structure, interactions 14 between Arg-124 and TM-I and IC-III are lost, with interactions only occurring with other TM-III residues. We, therefore, hypothesise that agonists promote conformational changes by releasing Arg-124 of the DRY motif, which is then free to reposition and possibly facilitate G protein binding. In the uncomplexed form the Asn-322 of the NPxxY (Asn-322, Pro-323, Ile- 324, Ile-325, Tyr-326) motif thought to be important in receptor activation [38,39], forms various interactions with Asp-72 of TM-II, Ser-113 of TM-III and Ser-319 and Cys-320 of TM-VII. We again observe changes in these interactions upon agonist complexation. For the AD complex, Asn-322 only interacts with its neighbouring Ile- 321, while the ND complex loses interactions with TM-III and gains interactions with TM-VI. For the ND complex, Asn-322 interacts with Asp-72 of TM-II, Val-277 of TM-VI and Asn-318 of TM-VII. Hence, in our analysis, a Ser-113/Asn-322 interaction is formed in the uncomplexed receptor and broken in the complexed forms, possibly freeing Asn-322 for other interactions such as with other TM-VII residues. This is consistent with the postulate of Bruysters et al. [40], which suggests that a Ser residue of TM-III acts as a molecular switch in the activation of the Histamine H1 receptor. Such a movement may change the orientation of residues in IC-II and IC-III that allows for increased affinity of coupling to the G-protein. Conclusions Despite a large number of experimental studies on GPCRs, knowledge of how agonist binding to receptors results in G protein activation remains unclear. We have developed a computational procedure, involving docking to generate a number of 15 plausible agonist/α1A-AR complexes and structural refinement through MD simulations using the Amber force field. In this way, ligand induced receptor structures were produced, which were subsequently used for redocking and further MD simulations. Analysis of agonist interactions with the ligand induced receptor structures are in agreement with mutagenesis experiments and suggest novel residues involved in the agonist-binding site, namely Cys-176 and Thr-174. The role of the charged amine group, the aromatic ring, the N-methyl group (for adrenaline), the beta hydroxyl group and the catechol meta and para hydroxyl groups have been examined and the most critical interactions for the binding of the agonists is consistent with many earlier reports. Furthermore, we envisage a number of steps when moving from agonist binding to G protein activation. These include a separation of TM-V away from TM- III, and increased flexibility about a Pro kink, which facilitates a movement of TM- VI. Upon examining the conserved residues in α1A-AR, activation may be a consequence of a change in the interactions of Asp-123 of the DRY motif that causes Arg-124 to move out of the TM helical bundle and change the orientation of residues in IC-II and IC-III, allowing for increased affinity of coupling to the G-protein. GPCR models such as ours can be used to generate hypotheses regarding the binding and signaling functions of these receptors for further experimental testing. However, we await the production of an experimental crystal structure of an adrenoceptor in the inactive or active state for a thorough comparison with experiment. 16 Acknowledgements: This research was supported by an HEA PRTLI grant (Cycle III), an Enterprise Ireland postgraduate research scholarship (G.K.) and a Trinity College Dublin postgraduate scholarship (G.K.). We also thank Dr. Peter Oliver at RAL for access and assistance with Hrothgar, a 16 node Beowulf cluster. 17 Figures & Tables: Figure 1: Optimisation and scaling steps followed by moving average trendline (period of 50ps) over the production run of the RMSD values of the (a) adrenaline complex and the (b) noradrenaline complex simulations. 18 Figure 2: Optimisation and scaling steps followed by moving average trendline (period of 50ps) over the production run of the RMSD over the 2 ns simulations for the adrenaline complex (a) and noradrenaline complex (b) simulations. 19 Figure 3: Final binding mode of Adrenaline with Trp-102, Asp-106, Thr-175, Ser- 188 and Ser-192 (upper) and Noradrenaline with Asp-106, Thr-175, Gln-177, Ser-188 and Ser-192 (lower). 20 Figure 4: Positioning of aromatic residues in TM-III (in mauve) and TM-VI (in orange) of the final noradrenaline complex. 21 Figure 5: The original uncomplexed α1A–AR (in grey), with TM-III (in mauve) and TM-V (in blue) for the final adrenaline complex (on left) and final noradrenaline complex (on right). 22 Table 1: Comparison of uncomplexed receptor structure and final agonist complexed receptor structures via RMSD (in Å). AD complex ND complex TM-I 4.42 3.28 TM-II 2.79 2.93 TM-III 2.64 2.58 TM-IV 2.55 2.76 TM-V 2.19 2.55 TM-VI 4.09 3.70 TM-VII 3.97 3.99 TM-VIII 0.86 0.89 C α (all) 5.68 5.25 23 References: 1 Beduschi, M.C., Beduschi, R. and Oesterling, J.E. Urology, 51 (1998) 861. 2 Ballesteros, J. and Palczewski, K., Curr. Opin. Drug Disc. Devel., 4 (2001) 561. 3 Okada, T. and Palczewski, K., Curr. Opin. Struct. Biol., 11 (2001) 420. 4 Kinsella, G.K., Rozas, I. and Watson, G.W., Biochem. Biophys. Res. Commun., 324 (2004) 916. 5 Scheer, A., Fanelli, P., Costa, T., De Bennedetti, P.G. and Cotecchia, S., EMBO J., 15 (1996) 3566. 6 Hamaguchi, N., True, T., Saussy, D. L. and Jeffs, P. W., Biochemistry, 35 (1996) 14312. 7 Zhao, M., Hwa, J. and Perez, D. M., Mol. Pharmacol., 50 (1996) 1118. 8 Waugh, D., Gaivin, R., Zuscik, M., Gonzalez-Cabrera, P., Ross, S., Yun, J. and Perez, D., J. Biol. Chem., 27 (2001) 25366. 9 Cavalli, A., Fanelli, F., Taddei, C., de Benedetti, P. G. and Cotechia, S., FEBS Lett., 399 (1996) 9. 10 Piascik, M. T. and Perez, D. M., J. Pharmacol. Exp. Ther., 298 (2001) 2. 11 Kobilka, B., Mol. Pharmacol., 65 (2004) 1060. 12 Gether, U. and Kobilka, B.K., J. Biol. Chem., 273 29 (1998) 17979. 13 Javitch, J.A., Fu, D., Liapakis, G. and Chen, J., J. Biol. Chem., 272 (1997) 18546. 14 Bea, I., Jaime, C. and Kollman, P., Theor. Chem. Acta., 108 (2002) 286. 15 Bissantz, C., Bernard, P., Hibert, M. and Rognan, D., Proteins, 50 (2003) 5. 16 Chambers, J.J., Nichols, D.E., J. Comput. Aided Mol. Des., 16 (2002) 511. 17 Carmine, R., Molinari, P., Sbraccia, M., Ambrosio, C. and Costa, T., Mol. Pharmacol., 66 (2004) 2. 24 18 Kuntz, I.D., Science, 257 (1992) 1078. 19 Sybyl6.9, Molecular Modelling System, Tripos Associates: St. Louis, MO, USA. 20 Frisch, M.J., Trucks, G.W., Schlegel, H.B., Scuseria, G.E., Robb, M.A., Cheeseman, J.R., Zakrzewski, V.G., Montgomery Jr., J.A., Stratmann, R.E., Burant, J.C., Dapprich, S., Millam, J.M., Daniels, A.D., Kudin, K.N., Strain, M.C., Farkas, O., Tomasi, J., Barone, V., Cossi, M., Cammi, R., Mennucci, B., Pomelli, C., Adamo, C., Clifford, S., Ochterski, J., Petersson, G.A., Ayala, P.Y., Cui, Q., Morokuma, K., Salvador, P., Dannenberg, J.J., Malick, D.K., Rabuck, A.D., Raghavachari, K., Foresman, J.B., Cioslowski, J., Ortiz, J.V., Baboul, A.G., Stefanov, B.B., Liu, G., Liashenko, A., Piskorz, P., Komaromi, I., Gomperts, R., Martin, R.L., Fox, D.J., Keith, T., Al-Laham, M.A., Peng, C.Y., Nanayakkara, A., Challacombe, M., Gill, P.M.W., Johnson, B., Chen, W., Wong, M.W., Andres, J.L., Gonzalez, C., Head-Gordon, M., Replogle, E.S. and Pople, J.A. Gaussian 98 (Rev A.11), Gaussian, Inc., Pittsburgh PA, 2001. 21 Case, D.A., Pearlman, D.A., Caldwell, J.W., Cheatham III. T.E., Wang, J., Ross, W.S., Simmerling, C.L., Darden, T.A., Merz, K.M., Stanton, R.V., Cheng, A.L., Vincent, J.J., Crowley, M., Tsui, V., Gohlke, H., Radmer, R.J., Duan, Y., Pitera, J., Massova, I., Seibel, G.L., Singh, U.C., Weiner, P.K. and Kollman, P.A., Amber 7, University of California, San Francisco, 2002. 22 Berendsen, H.J.C., Postma, J.P.M., van Gunsteren, W.F., DiNola, A. and Haak, J.R., J. Chem. Phys., 81 (1984) 3684. 23 Wymore, T., Wong, T.C., 45th Annual Meeting of the Biophysical Society. Boston, MA, 2001. 24 Wang, J., Wolf, R.M., James W. Caldwell, J. W., Kollman, P.A., Case, D.A. J. Comp. Chem. 25 (2004) 1157. 25 25 Aldrich Library of FT-IR Spectra, 1 (1) 1296A. 26 Pedretti, A., Silva, M.E., Villa,L. and Vistoli, G., Biochem. Biophys. Res. Commun., 319 (2000) 493. 27 Visiers, I., Ballesteros, J.A. and Weinstein, H., Methods Enzymol., 343 (2002) 329. 28 Ambrosio, C., Molinari, P., Cotecchia, S. and Costa, T., Mol. Pharmacol., 57 (2000) 198. 29 Liapakis, G., Ballesteros, J. A., Papachristou, S., Chan, W. C., Chen, X. and Javitch, J. A. J. Biol. Chem., 275 (2000) 37779. 30 Shi, L., Liapakis, G., Xu, R., Guarnieri, F., Ballesteros, J.A. and Javitch, J.A., J. Biol. Chem., 277 (2002) 40989. 31 Shi, L. and Javitch, J.A., Annu. Rev. Pharmacol. Toxicol., 42 (2002) 437. 32 Sansom, M.S. and Weinstein, H., Trends Pharmacol. Sci., 21 (2000) 445. 33 Dunham, T.D. and Farrens, D.L., J. Biol. Chem., 274 (1999) 1683. 34 Farrens, D.L., Altenbach, C., Yang, K., Hubbell, W.L. and Khorana, H.G., Science, 274 (1996) 768. 35 Sheikh, S.P., Vilardarga, J.P., Baranski, T.J., Lichtarge O., Iiri, T., Meng, E.C., Nissenson, R.A. and Bourne, H.R., J. Biol. Chem., 274 (1999) 17033. 36 Jensen, A.D., Guarnieri, F., Rasmussen, S.G., Asmar, F., Ballesteros, J.A. and Gether, U., J. Biol. Chem., 276 (2001) 9279. 37 Gether, U., Asmar, F., Meinild, A.K. and Rasmussen, G.F., Pharmaco. Toxicol., 91 (2002) 304. 38 Zhou, W., Flanagan, C., Ballesteros, J.A., Konvocka, K., Davidson, J.S., Weinstein, H., Millar, R.P. and Sealfon, S.C., Mol. Pharmacol., 45 (1994) 165. 26 39 Sealfon, S.C., Chi, L., Ebersole, B.J., Rodic, V., Zhang, D., Ballesteros, J.A. and Weinstein, H., J. Biol. Chem., 270 (1995) 16683. 40 Jongejan, A., Bruysters, M., Pardo, L. and Leurs, R., Oral Presentation, XVIIth Symposium of Medicinal Chemistry, 2004.