Lipid-facing correlated mutations and dimerization in G-protein coupled receptors.

A correlated mutation analysis has been performed on the aligned protein sequences of a number of class A G-protein coupled receptor families, including the chemokine, neurokinin, opioid, somatostatin, thyrotrophin and the whole biogenic amine family. Many of the correlated mutations are observed flanking or neighbouring conserved residues. The correlated residues have been plotted onto the transmembrane portion of the rhodopsin crystal structure. The structure shows that a significant proportion of the correlated mutations are located on the external (lipid-facing) region of the helices. The occurrence of these highly correlated patterns of change amongst the external residues suggest that they are sites for protein-protein interactions. In particular, it is suggested that the correlated residues may be involved in either large conformational changes, the formation of heterodimers or homodimers (which may be domain swapped) or oligomers required for activation or internalization. The results are discussed in the light of the subtype-specific heterodimerization observed for the chemokine, opioid and somatostatin receptors.

The agonist SR 146131 and the antagonist SR 27897 occupy different sites on the human CCK(1) receptor.

1-[2-(4-(2-Chlorophenyl)thiazol-2-yl) aminocarbonyl indoyl] acetic acid (SR 27897) is an effective CCK(1) receptor antagonist, while the structurally related molecule 2-[4-(4-chloro-2, 5-dimethoxyphenyl)-5-(2-cyclohexyl-ethyl)-thiazol-2-ylcarbamoyl ]-5, 7-dimethyl-indol-1-yl-1-acetic acid (SR 146131) is a highly potent and specific agonist for the same receptor. To discover how the two molecules interact with the human cholecystokinin (CCK) CCK(1) receptor, we have carried out binding and activity studies with 33-point mutated receptors. Only six mutants showed altered [3H]SR 27897 binding properties, Lys(115), Lys(187), Phe(198), Trp(209), Leu(214) and Asn(333). In contrast, numerous mutations throughout the receptor either reduced SR 146131 agonist potency, Phe(97), Gly(122), Phe(198), Trp(209), Ile(229), Asn(333), Arg(336) and Leu(356) or increased it, Tyr(48), Cys(94), Asn(98), Leu(217) and Ser(359). Only mutations of Phe(198), Trp(209) and Asn(333) affected both SR 27897 and SR 146131 binding or activity. The collated information was used to construct molecular models of SR 27897 and SR 146131 bound to the human CCK(1) receptor. The clear difference in the binding sites of SR 27897 and SR 146131 offers a molecular explanation for their contrasting pharmacological characteristics.

Dimerization and domain swapping in G-protein-coupled receptors: a computational study.

In recent years there has been an increasing number of reports describing G protein-coupled receptor (GPCR) dimerization and heterodimerization. However, the evidence on the nature of the dimers and their role in GPCR activation is inconclusive. Consequently, we present here a review of our computational studies on G protein-coupled receptor dimerization and domain swapping. The studies described include molecular dynamics simulations on receptor monomers and dimers in the absence of ligand, in the presence of an agonist, and in the presence of an antagonist (or more precisely an inverse agonist). Two distinct sequence-based approaches to studying protein interfaces are also described, namely correlated mutation analysis and evolutionary trace analysis. All three approaches concur in supporting the proposal that the dimerization interface includes transmembrane helices 5 and 6. These studies cannot distinguish between domain swapped dimers and contact dimers as the models used were restricted to the helical part of the receptor. However, it is proposed that for the purpose of signalling, the domain swapped dimer and the corresponding contact dimer are equivalent. The evolutionary trace analysis suggests that every GPCR family and subfamily (for which sufficient sequence data is available) has the potential to dimerize through this common functional site on helices 5 and 6. The evolutionary trace results on the G protein are briefly described and these are consistent with GPCR dimerization. In addition to the functional site on helices 5 and 6, the evolutionary trace analysis identified a second functional site on helices 2 and 3. Possible roles for this site are suggested, including oligomerization.

Mutational analysis and molecular modelling of the antagonist SR 144528 binding site on the human cannabinoid CB(2) receptor.

We have investigated the binding site of the subtype specific antagonist SR 144528, (N-[(1S)-endo-1,3,3-trimethyl bicyclo [2.2. 1]heptan-2-yl]-5-(4-chloro-3-methylphenyl)-1-(4-methoxybenzyl)- pyrazo le-3-carboxamide) on the human cannabinoid CB(2) receptor based on functional studies with mutated receptors. Two serine residues in the fourth transmembrane region, Ser(161) and Ser(165), were singly mutated to the cognate cannabinoid CB(1) receptor residue, alanine, and each gave receptors with wild-type properties for the cannabinoid agonists CP 55,940 (1R,3R,4R)-3-[2-hydroxy-4-(1, 1-dimethylheptyl)phenyl]-4-(3-hydroxypropyl)cyclohexan-1-ol) and WIN 55212-2 (R)-(+)[2, 3-dihydro-5-methyl-3-[(4-morpholinyl)methyl]pyrrolo[1,2,3-de]-1, 4-benzoxazin-6-yl](1-naphthalenyl) methanone, which SR 144528 completely failed to antagonise. Molecular modelling studies show that SR 144528 interacts with residues in transmembrane domains 3, 4, and 5 of the cannabinoid CB(2) receptor through a combination of hydrogen bonds and aromatic and hydrophobic interactions. In addition, the replacement by serine of a nearby cannabinoid CB(2) receptor-specific residue, Cys(175) resulted in wild-type receptor properties with CP 55,940, loss of SR 144528 binding and eight-fold reduced binding and activity of WIN 55212-2, a result compatible with a recently-proposed binding site model for WIN 55212-2.

Essential role of extracellular charged residues of the human CCK(1) receptor for interactions with SR 146131, SR 27897 and CCK-8S.

We hypothesized that charge-charge interactions may be important for the binding of the human cholecystokinin type 1 (CCK(1)) receptor-specific non-peptide full agonist SR 146131, (2-[4-(4-chloro-2, 5-dimethoxyphenyl)-5-(2-cyclohexyl-ethyl)-thiazol-2-ylcarbamoyl ]-5, 7-dimethyl-indol-1-yl-1-acetic acid), the competitive antagonist SR 27897, (1-[2-(4-(2-chlorophenyl)thiazol-2-yl) aminocarbonyl indoyl] acetic acid) and the natural octapeptide CCK-8S to the CCK(1) receptor. Alanine replacement studies of positively charged residues in the extracellular domains of the receptor showed that only the R336A mutation affected SR 146131 potency of mutated receptors transiently expressed in monkey kidney epithelial COS-7 cells. Two residues, Lys(115) and Lys(187), were implicated in SR 27897 binding. Only the replacement of Lys(115), Arg(197) and Arg(336) significantly affected CCK-8S binding or activity. These results clearly indicated the importance of certain charged residues, but not others, in SR 146131, SR 27897 and CCK-8S binding. Furthermore, although these molecules probably occupy different binding sites on the CCK(1) receptor, we show that a small non-peptide agonist, SR 146131, can stimulate the dual signaling pathways mediated by the CCK(1) receptor.

Contrasting roles of leu(356) in the human CCK(1) receptor for antagonist SR 27897 and agonist SR 146131 binding.

A new highly specific, potent non-peptide agonist for the cholecystokinin subtype 1 receptor (CCK(1)), SR 146131 (2-[4-(4-chloro-2, 5-dimethoxyphenyl)-5-(2-cyclohexyl-ethyl)-thiazol-2-ylcarbamoyl ]-5, 7-dimethyl-indol-1-yl-1-acetic acid) was recently described [Bignon, E., Bachy, A., Boigegrain, R., Brodin, R., Cottineau, M., Gully, D., Herbert, J.-M., Keane, P., Labie, C., Molimard, J.-C., Olliero, D., Oury-Donat, F., Petereau, C., Prabonneaud, V., Rockstroh, M.-P., Schaeffer, P., Servant, O.Thurneyssen, O., Soubrié, P., Pascal, M., Maffrand, J.-P., Le Fur, G., 1999. SR 146131: a new, potent, orally active and selective non-peptide cholecystokinin subtype I receptor agonist: I. In vitro studies. J. Pharmacol. Exp. Ther. 289, 742-751]. From binding and activity assays with chimeric constructs of human CCK(1) and the cholecystokinin subtype 2 receptor (CCK(2)) and receptors carrying point mutations, we show that Leu(356), situated in transmembrane domain seven in the CCK(1) receptor, is a putative contact point for SR 146131. In contrast, Leu(356) is probably not in contact with the CCK(1) receptor specific antagonist SR 27897 (1-[2-(4-(2-chlorophenyl)thiazol-2-yl)aminocarbonyl indoyl]acetic acid), a compound structurally related to SR 146131, since its replacement by alanine, histidine or asparagine gave receptors having wild-type CCK(1) receptor SR 27897 binding affinity. Previous mutational analysis of His(381), the cognate position in the rat CCK(2) receptor, had implicated it as being involved in subtype specificity for SR 27897, results which we confirm with corresponding mutations in the human CCK(2) receptor. Moreover, binding and activity assays with the natural CCK receptor agonist, CCK-8S, show that CCK-8S is more susceptible to the mutations in that position in the CCK(1) receptor than in the CCK(2) receptor. The results suggest different binding modes for SR 27897, SR 146131 and CCK-8S in each CCK receptor subtype.

Domain swapping in G-protein coupled receptor dimers.

Computer simulations were performed on models of the beta2-adrenergic receptor dimer, including 5,6-domain swapped dimers which have been proposed as the active, high affinity form (here the dimer interface lies between helices 5 and 6). The calculations suggest that the domain swapped dimer is a high energy structure in both the apo dimer and in the presence of propranolol. In the presence of agonist the energy of the domain swapped dimer is significantly lowered. Analysis of the dimer structure suggests that the agonist-induced conformational change optimizes the helix-helix interactions at the 5-6 interface. An antagonist on the other hand has little effect on these interactions. These observations are consistent with the hypothesis that the agonist functions by shifting the equilibrium in favour of the domain swapped dimer. Indirect support for the domain swapping hypothesis was obtained from the correlated mutations amongst the external residues of the known beta2-adrenergic receptors. These occur mainly at the 5-6 interface at precisely the locations predicted by the simulations; site-directed mutagenesis data in support of a functional role for these lipid-facing correlated residues is presented. The article includes a review of the experimental evidence for G-protein coupled receptor dimerization. Many other aspects of G-protein coupled receptor activation are discussed in terms of this domain swapping hypothesis

A new approach to docking in the beta 2-adrenergic receptor that exploits the domain structure of G-protein-coupled receptors.

A novel technique for docking ligands to the beta 2-adrenergic receptor is described which exploits the domain structure of this class of receptors. The ligands (norepinephrine, an agonist; pindolol, a partial agonist; and propranolol, an antagonist) were docked into the receptor using the key conserved aspartate on helix 3 (D113) as an initial guide to the placement of the amino group and GRID maps (Goodford, P. J. J. Med. Chem, 1985, 28, 849) to identify the likely binding regions of the hydrophobic (and hydroxyl) moieties on the A domain (comprising of helices 1-5). The essence of the new approach involved pulling the B domain, which includes helices 6 and 7, away from the other domain by 5-7 A. During the subsequent minimization and molecular dynamics, the receptor ligand complex reformed to yield structures which were very well supported by site-directed mutagenesis data. In particular, the model predicted a number of important interactions between the antagonist and key residues on helix 7 (notably Leu311 and Asn312) which have not been described in many previous computer simulation studies. The justification for this new approach is discussed in terms of (a) phase space sampling and (b) mimicking the natural domain dynamics which may include domain swapping and dimerization to form a 5,6-domain-swapped dimer. The observed structural changes in the receptor when pindolol, the partial agonist, was docked were midway between those observed for propranolol and norepinephrine. These structural changes, particularly the changes in helix-helix interactions at the dimer interface, support the idea that the receptors have a very dynamic structure and may shed some light on the activation process. The receptor model used in these studies is well supported by experiment, including site-directed mutagenesis (helices 1-7), zinc binding studies (helices 2, 3, 5, and 6), the substituted cysteine accessibility method (helices 3, 5, and 7), and site-directed spin-labeling studies (helices 3-6).

A molecular dynamics approach to receptor mapping: application to the 5HT3 and beta 2-adrenergic receptors.

A molecular dynamics-based approach to receptor mapping is proposed, based on the method of Rizzi (Rizzi, J. P.; et al. J. Med. Chem. 1990, 33, 2721). In Rizzi's method, the interaction energy between a series of drug molecules and probe atoms (which mimic functional groups on the receptor, such as hydrogen bond donors) was calculated. These interactions were calculated on a three-dimensional grid within a molecular mechanics parameters, were placed at these minima. The distances between the dummy atom sites were monitored during molecular dynamics simulations and plotted as distance distribution functions. Important distances within the receptor became apparent, as drugs with a common mode of binding share similar peaks in the distance distribution functions. In the case of specific 5HT3 ligands, the important donor--acceptor distance within the receptor has a range of ca. 7.9--8.9 A. In the case of specific beta 2-adrenergic ligands, the important donor--acceptor distances within the receptor lie between ca. 7--9 A and between 8 and 10 A. These distances distribution functions were used to assess three different models of the beta 2-adrenergic G-protein-coupled receptor. The comparison of the distance distribution functions for the simulation with the actual donor--acceptor distances in the receptor models suggested that two of the three receptor models were much more consistent with the receptor-mapping studies. These receptor-mapping studies gave support for the use of rhodopsin, rather than the bacteriorhodopsin template, for modeling G-protein-coupled receptors but also sounded a warning that agreement with binding data from site-directed mutagenesis experiments does not necessarily validate a receptor model.