Molecular Architecture of G Protein-Coupled Receptors

Abstract

This review of the current literature on mutations in G protein-coupled receptors (GPCRs) of the rhodopsin-related family intends to draw inferences from amino acid sequences for single receptors and multiple sequence alignments with regard to the molecular architecture of this class of receptors. For this purpose a comprehensive list of mutations within the transmembrane helical regions (TMs; over 390 mutations from 38 different receptor subtypes) and their effects on function was compiled, and an alignment of known GPCR sequences (over 150 separate sequences) was made. Regions most prominently involved in ligand binding are located in TMs 3, 5, 6, and 7. Position 3.32 in TM3 is occupied by a D in all biogenic amine receptors (sequence conservation) but may be occupied by uncharged residues in other receptors while its role in ligand binding is analogous (function conservation). TMs 5, 6, and 7 display considerable sequence conservation throughout the majority of GPCRs investigated, but not necessarily at those positions involved in ligand binding. However, considerable function conservation is observed for positions 5.42 (frequently hydrophilic), 5.46 (small amino acids required for agonist binding to “small ligand” receptors), 6.52 and 7.39 (high variability), and 7.43 (frequently aromatic). A general conclusion of this review is that there is overwhelming conservation of structure-function correlates among GPCRs. Thus, it is now possible to cross-correlate the results of mutagenesis studies between GPCRs of different subfamilies, and to use those results to predict the function of specific residues in new GPCR sequences.

INTRODUCTION

The aim of this study was to analyze the effects of amino acid substitutions in G protein-coupled receptors (GPCRs) on ligand binding. In an effort to identify functions of particular sites among various receptors, we have focused on mutations within the helical transmembrane regions, since sequence homology is clearly discernible in those regions. The result is a comprehensive, but probably not exhaustive, list of these substitutions. We have tried to correlate substitutions of specific amino acids with the effect(s) exerted on ligand binding, to discriminate between agonist and antagonist binding, coupling to G proteins, and receptor activation. The literature is abundant with reports of structure-activity/affinity relationships (SAR) based on modelling ligands; however, very few (if any) SAR studies are based on the insights obtained from modelling the receptor-ligand interaction at the molecular level of the receptor. Other aspects of mutagenesis techniques applied to GPCRs have been reviewed extensively [Houslay, 1992; Baldwin, 1994; Coughlin, 1994; Donnelly and Findlay, 1994; Schwartz, 1994]. Structure and function of one subfamily of GPCRs, i.e., the biogenic amine receptors, have been analyzed in more detail [Ostrowski et al., 1992; Strader et al., 1994; Burstein et al., 1995].

From Isolation to Cloning of G Protein-Coupled Receptors

Isolation, purification, sequencing

The first biophysical data on GPCRs were obtained in the early seventies, when Lefkowitz et al. [1972] isolated the cardiac β-adrenoceptor. It was a very laborious process that required large amounts of receptor protein and perseverance. It took 14 years before the next large step in the elucidation of the structure of this GPCR was completed. In 1986, Dixon et al. [1986] reported the isolation and cloning of a mammalian β-adrenoceptor. In the years since, hundreds of sequences for GPCRs have been identified, and their function attributed to a specific neurotransmitter/humoral regulator system. The receptor isolation techniques developed over the years, however, have not been totally abandoned. They have been demonstrated to be useful in, for instance, determining the binding site for a specific antagonist in muscarinic acetylcholine receptors [Kurtenbach et al., 1990].

The polymerase chain reaction

Considerable progress in the identification of GPCR sequences was made with the advent of a novel molecular biology technique currently known as the polymerase chain reaction (PCR). Its immediate predecessor was described as early as 1985 [Saiki et al.], but the method gained wide popularity [Vosberg, 1989; White et al., 1989] only after the introduction of a thermostable DNA polymerase in the procedure [Saiki et al., 1988]. The method was successfully applied to the cloning of proteins in general [Bonner et al., 1987; Intres and Crabb, 1992], and to GPCRs more specifically [Libert et al., 1989; Parmentier et al., 1993]. The PCR method has since seen the development of a large number of specialized and/or more sophisticated applications [Bockstahler, 1994; Edwards and Gibbs, 1994; Rashtchian, 1995].

Cloning methodology

Cloning procedures for specific receptors have been described extensively [Allard et al., 1987; Arai et al., 1990; Arakawa et al., 1990; Barberis et al., 1993; Birdsall, 1991; Bunzow et al., 1988; Cone et al., 1993; Corjay et al., 1991; D’Angelo et al., 1994; Gantz et al., 1991; Gershengorn, 1993; Inagami et al., 1992; Ishihara et al., 1992; Jüppner et al., 1991; Kieffer et al., 1992; Kimura et al., 1993; Klein et al., 1988; Kubo et al., 1986; Lubbert et al., 1987; Lustig et al., 1993; Marsh and Herskowitz, 1988; Masu et al., 1991; McEachern et al., 1991; Miyajima et al., 1987; Morel et al., 1993; Murphy et al., 1992; Ruat et al., 1991; Salvatore et al., 1993; Straub et al., 1990; Sugimoto et al., 1992; Wank et al., 1992; Webb et al., 1993; Yamashita et al., 1991; Ye et al., 1991; Yokota et al., 1989], and an overview of such procedures was published by Parmentier et al. [1993]. Explicit procedures for cloning and expression of the serotonin 5HT_1A receptor [Albert, 1992], or site-directed mutagenesis in the serotonin 5HT₂ receptor [Shih et al., 1992] were described, and could serve as a reference for receptors yet to be cloned.

Targeted searching

The first GPCR to be cloned was rhodopsin [Nathans and Hogness, 1983]. Rhodopsin is a special case within the superfamily of GPCRs; the receptor couples only to the specific G-protein homologue transducin, and its “endogenous agonist” is a steric and electronic transition of the covalently linked retinal chromophore by photoinduction rather than a circulating/topically released chemical substance. Its sequence and apparent functional homology to other GPCRs, however, qualify it as a GPCR par excellence. It has proven to be a well-behaved model and test system for analyzing process kinetics and thermodynamics, and has served especially well in mutagenesis studies [Kaushal and Khorana, 1994, and references therein]. After the initial cloning of the β-adrenoceptor [Dixon et al., 1986], many other members of the GPCR superfamily have been identified through targeted cloning techniques (Table 1). In Table 1, receptors were classified according to their “ligand family” which is not necessarily the best option. For instance, even though adenosine (4 subtypes) and adenosine 5′-triphosphate (ATP; at least 8 subtypes) receptors are all considered members of the purinergic receptor subfamily, there seems to be minimal crossover in ligand, agonist, or antagonist, affinity between either receptors for adenosine (P₁ or receptors for ATP (P₂) [Fredholm et al., 1994; Dalziel and Westfall, 1994; Fredholm, 1995]. A better way to classify receptor subfamilies is probably through amino acid sequence homology and genealogy, as was proposed by Kolakowski [1994]. Another case in point is the interleukin 8 (IL-8) receptor. Initially classified by its supposed ligand, it was deemed the receptor for formyl-methionyl-leucyl-phenylalanyl (fMLP) peptide [Thomas et al., 1990]; it currently is more properly defined as an IL-8 receptor [Beckmann et al., 1991].

TABLE 1.

Cloned G Protein-Coupled Receptors

Class	Refcode	Name (if not human: species)
Peptide/cytokine	P30968	Gonadotropin-releasing hormone: GRH
	P34981	Thyrotropin-releasing hormone: TRH
	P25116	Thrombin: THR
	P23945	Follicle stimulating hormone: FSH
	P16473	Thyrotropin: TSH
	P22888	Luteinizing hormone: LH/CG
	Q01718	Adrenocorticotropic hormone: ACTH/MC2
	Q01726	Melanocyte Stimulating hormone: MSH/MC1
	B46647	Melanocortin MC3
	P32245	Melanocortin MC4
	P33032	melanocortin MC5
	U14108	Melatonin
	P30559	Oxytocin
	P37288	Vasopressin V1A
	A55089	Vasopressin V1B
	P30518	Vasopressin V2
	P32238	Cholecystokinin CCK-A
	P32239	Cholecystokinin CCK-B
	P25929	Neuropeptide Y: NPY1
	P25103	Neurokinin NK1
	P21452	Neurokinin NK2
	P29371	Neurokinin NK3
	P25101	Endothelin ET-A
	P24530	Endothelin ET-B
	P28336	Neuromedin B: NMB
	P30872	Somatostatin SSTR1
	P30874	Somatostatin SSTR2
	P32745	Somatostatin SSTR3
	P31391	Somatostatin SSTR4
	P35346	Somatostatin SSTR5
	P25024	Interleukin IL-8A
	P25025	Interleukin IL-8B
	P30556	Angiotensin AT1a
	P29089	Angiotensin AT1b (Rattus norvegicus)
	JC2435	Angiotensin AT2
	U12512	Bradykinin BK1
	P30411	Bradykinin BK2
	P21462	Formyl-Met-Leu-Phe peptide: fMLP
	JC2194	Vaso-intestinal peptide VIP-IVR8
	S38397	Vaso-intestinal peptide VIP-PACAP
	P32241	Vaso-intestinal peptide VIP1
	JC2195	Vaso-intestinal peptide VIP-IVR5
	P35000	Vaso-intestinal peptide VIP2 (Rattus norvegicus)
	JC2463	Vaso-intestinal peptide VIP-T1
	L04308	Parathyroid hormone
	L23332	Corticotropin releasing hormone
	U10037	Glucagon
	U20178	Secretin
lipid mediator	P34995	Prostaglandin E2 PGE-EP1
	P35408	Prostaglandin E2 PGE-EP2
	U19487	Prostaglandin E2 PGE-EP3
	L24470	Prostaglandin F2α:PGF
	P21731	Thromboxane A2: TXA2
	P25105	Platelet-activating factor: PAF
biogenic amine	P08172	Muscarinic acetylcholine M2
	P08173	Muscarinic acetylcholine M4
	P11229	Muscarinic acetylcholine M1
	P20309	Muscarinic acetylcholine M3
	P08912	Muscarinic acetylcholine M5
	P35367	Histamine H1
	P25021	Histamine H2
	P08908	Serotonin 5HT1A
	P28222	Serotonin 5HT1B
	P28221	Serotonin 5HT1D
	P28566	Serotonin 5HT1E
	P30939	Serotonin 5HT1F
	P28223	Serotonin 5HT2A
	X77307	Serotonin 5HT2B
	P28335	Serotonin 5HT2C
	X81411	Serotonin 5HT5A
	P35365	Serotonin 5HT5B (Rattus norvegicus)
	P31388	Serotonin 5HT6 (Rattus norvegicus)
	P34969	Serotonin 5HT7
	P25100	Adrenergic α1A
	P35368	Adrenergic α1B
	P35348	Adrenergic α1C
	P08913	Adrenergic α2A
	P18089	Adrenergic α2B
	P18825	Adrenergic α2C1
	P35369	Adrenergic α2C2
	P08588	Adrenergic β1
	P07550	Adrenergic β2
	P13945	Adrenergic β3
	P21728	Dopamine D1A
	P14416	Dopamine D2
	P35462	Dopamine D3
	P21917	Dopamine D4
	P21918	Dopamine D5
nucleoside	P30542	Adenosine A1
	P29274	Adenosine A2a
	P29275	Adenosine A2b
	P33765	Adenosine A3
nucleotide	Z49205	P2 purinergic P2Y/P2Y1
	P41231	P2 purinergic P2U/P2Y2
	U41070	P2 purinergic P2Y/P2Y5
	D63665	P2 purinergic P2Y/P2Y6
cyclic AMP	P13773	Cyclic AMP-1 (Dictyostelium discoideum)
	P34907	Cyclic AMP-2 (Dictyostelium discoideum)
	P35352	Cyclic AMP-3 (Dictyostelium discoideum)
octopamine	P22270	Octopamine (Drosophila melanogaster)
opsin	P03999	Blue opsin
	P04001	Green opsin
	P04000	Red opsin
	P08100	Rhodopsin
	M92039	Violet opsin (Gallus gallus)
	L03781	Opsin (Limulus polyphemus)
opioid	P41143	δ opioid
	P41145	κ opioid
	P35372	μ opioid
sensory
(olfactory)	145774	OLF1 (Ictalurus punctatis)
	H45774	OLF3 (Ictalurus punctatis)
	E45774	OLF8 (Ictalurus punctatis)
	D45774	OLF32A (Ictalurus punctatis)
	C45774	OLF32B (Ictalurus punctatis)
	B45774	OLF32C (Ictalurus punctatis)
	A45774	OLF32D (Ictalurus punctatis)
	F45774	OLF47 (Ictalurus punctatis)
	G45774	OLF202 (Ictalurus punctatis)
	P37067	OLFCOR1 (Gallus gallus)
	P37068	OLFCOR2 (Gallus gallus)
	P37069	OLFCOR3 (Gallus gallus)
	P37070	OLFCOR4 (Gallus gallus)
	P37071	OLFCOR5 (Gallus gallus)
	P37072	OLFCOR6 (Gallus gallus)
	P30955	OLF (Canis familiaris)
	P34982	OLF07E
	P30953	OLF07I
	P30954	OLF07J
	P23275	OLFOR3 (Mus musculus)
	S47014	OLF (Rattus norvegicus)
	P23265	OLFF3 (Rattus norvegicus)
	P23266	OLFF5 (Rattus norvegicus)
	P23267	OLFF6 (Rattus norvegicus)
	P23268	OLFF12 (Rattus norvegicus)
	P23269	OLF13 (Rattus norvegicus)
	P23270	OLF17 (Rattus norvegicus)
	P23271	OLF18 (Rattus norvegicus)
	P23272	OLF19 (Rattus norvegicus)
	P23273	OLF114 (Rattus norvegicus)
	P23274	OLF115 (Rattus norvegicus)
	X80391	OLFOR17–40
(gustatory)	D12820	GUST27 (Rattus norvegicus)
metabotropic	P23385	MG1 (Rattus norvegicus)
glutamate	JC2132	MG5A
	JC2131	MG5B
	P31424	MG5 (Rattus norvegicus)
	P31423	MG4 (Rattus norvegicus)
	P35400	MG7 (Rattus norvegicus)
	P35349	MG6 (Rattus norvegicus)
	P31421	MG2 (Rattus norvegicus)
	P31422	MG3 (Rattus norvegicus)
calcium	U20769	Ca2+ -sensitive GPCR
calcitonin	P30988	calcitonin
pheromone	Q00619	MAM2 (Schizosaccharomyces pombe)
	P06842	STE2 (Saccharomyces cerevisiae)
	P31302	PRA1 (Ustilago maydis)
	P31303	PRA2 (Ustilago maydis)
	P31397	MAP3 (Schizosaccharomyces pombe)
	P06783	STE3 (Saccharomyces cerevisiae)
orphan	P41146	OPRX
receptors	Z28332	AERG (Anthopleura elegantissima)
	P35412	GP01 (Mus musculus)
	P30951	R334 (Rattus norvegicus)
	P35413	GP21 (Mus musculus)
	P30731	GCRC (Mus musculus)
	P30098	TXKR
	P31392	G10D (Rattus norvegicus)
	M64749	RDC1
	P32302	BLR1
	L06797	CL5
	P30991	LCR1
	P32248	EBI1
	P35411	RBS1 (Rattus norvegicus)
	P32249	EBI2
	P32250	GCRT (Gallus gallus)
	P35414	APJ
	P23749	RTA (Rattus norvegicus)
	S77867	UHR (Rattus norvegicus)
	P25089	FMRL1
	P25090	FMRL2
	U14910	RPE
	QQBED3	HHRF1
	QQBED2	HHRF2
	QQBET9	HHRF3
	U03882	MCP-1A
	U03905	MCP-1B
	P35350	PPR1 (Bos taurus)
	QQBEQ4	ALB (saimiriine herpes virus 1)
	U17473	Calcitonin-like

Orphan receptors

It was not long until the first GPCRs were cloned where the endogenous ligand was unknown. These clones were dubbed “orphan receptors” (Table 1) [see Mills and Duggan, 1994, and references therein]. The serotonin 5HT_1A receptor started out as an orphan receptor [Kobilka et al., 1987], but its function was rapidly elucidated [Fargin et al., 1988]. The same was the case for the somatostatin receptors [Meyerhof et al., 1991; Li et al., 1992]. However, of the four orphan receptors originally identified by Libert et al. [1989], one sequence (RDC4) was assigned to the serotonin 5HT_1D receptor [Maenhaut et al., 1991], and two others were assigned to the adenosine A₁ (RDC7) [Libert et al., 1991], and adenosine A_2a (RDC8) [Maenhaut et al., 1990] receptors, respectively. Still unresolved is the functionality of the clone RDC1. Other orphan receptors are listed in Table 1.

Detection of GPCRs

Use of receptor-antibodies

GPCRs can be detected in various ways. Probably the oldest way is the registration of a functional effect certain agents exert in vitro/ex vivo/in vivo on tissues, organs, or whole organisms. The results of such experiments can be expressed in consistent and predictive structure-activity relationships. The translation of the signal to the final effect, however, requires any number of transduction steps that may be modulated/moderated by other factors. Not the least of these steps is the receptor-to-G protein coupling itself, and the concomitant coupling to effector systems, such as adenylyl cyclase and phospholipases. A more direct way of detecting GPCRs is through the use of radioligand binding assays. Data derived from such assays have been used extensively to define structure-affinity relationships to aid the development of newer ligands. In cases where radioligand binding assays were either not available, or not useable, employing (auto)antibodies raised against GPCR sequences was demonstrated to be helpful. For example, immunological detection proved to be instrumental in the isolation and identification of the β-adrenoceptor [Venter and Fraser, 1983]. More recently, this method of tracking GPCRs has been shown to be effective in determining (splice variant) subtypes of D₅ dopamine receptors [Bergson et al., 1995], monitoring the expression of chimerical D₂ dopamine [Fishburn et al., 1994], and NKl/NK2 neurokinin receptors [Vigna et al., 1994], and identifying an A_p4A dinucleotide receptor [Walker and Hilderman, 1993].

Use of epitope-tags

Another advantage of applying the PCK technique to the field of GPCR research, is that primer sequences are readily extended at either the 3′ or the 5′ end. Translated from the nucleotide sequence to the protein, this means that proteins can be supplied with a “flag” or “tag” sequence at either the N- or C-terminal side. Such a tag can be used to monitor expression levels of the tagged protein by means of radio-immunoassays (RIAs) or enzyme-linked immunosorbent assays (ELISAs) directed against the tag sequence. This method has successfully been applied to quantitate cell surface expression levels of transfected platelet-activating factor (PAF) receptors [Kunz et al., 1992] and dopamine D₂ receptors [Sanderson and Strange, 1995], to monitor the internalization of β₂-adrenoceptors [von Zastrow and Kobilka, 1994], and to establish the expression levels of mutant adenosine A_2a receptors that were pharmacologically undetectable [Kim et al., 1995].

Peptide and cytokine receptors

A very arbitrary, but useful, distinction can be made between two classes of GPCRs that is based solely on the size of the endogenous ligand. Peptide and cytokine receptors are activated by (macro)molecules that, due to their size, interact with the receptor protein over a much larger surface area than the “small ligand” receptors. The interactions vary from relatively small polypeptides, such as angiotensin II, that interact partly with the helical bundle and partly with the extracellular extensions [Yamano et al., 1995], through thrombin receptors, where the N-terminal part of the receptor sequence is a precursor for the agonist [Vu et al., 1991; Vouret-Craviari et al., 1995], to receptors such as the LH/CG receptor, where the agonist is thought to interact entirely with the N-terminal part of the receptor sequence [Kakar et al., 1992]. The binding of large ligands leads to participation of a larger portion of the helical bundle in the process of binding agonists, but not usually (non-peptide) antagonists [Beinborn et al., 1993; Bihoreau et al., 1993; Bhogal et al., 1994; Breu el al., 1995; Yamano et al., 1995]. The advantage of working with such receptors in understanding ligand receptor interactions is that a larger part of the receptor is “covered.” A disadvantage is that interactions between agonists and their receptors tend to be much harder to assign to a specific region of the receptor. In the case of the thrombin receptor, studying agonist-receptor interactions is even more difficult compared to other peptide receptors, because of the required enzymatic conversion of the N-terminus in receptor activation [Vouret-Craviari et al., 1995]. Enzymatic activation of GPCRs is, however, not limited to peptide receptors: a yet uncloned AP₄A dinucleotide receptor requires activation by a serine protease [Walker and Hilderman, 1993].

Small ligand receptors

The analysis of mutagenesis data for small ligand receptors can yield valuable information concerning the relative height of the ligand binding site in the helical bundle, and the mode of binding of agonists and antagonists to the receptor [e.g., Fong et al., 1993b; Javitch et al., 1995; Kim et al., 1995]. A recent review by Ballesteros and Weinstein [1995] indicates that the binding sites for the class of smaller ligands in GPCRs are allocated to a rather narrow region. A special case, in which more data are available than for any other receptor subgroup, is the biogenic amine receptors. Several models have been generated, and all indicate similar binding domains around the upper third to upper half of the helical transmembrane bundle towards the extracellular surface [Hibert et al., 1991; Trumpp-Kallmeyer et al., 1992]. For both agonists and antagonists for small ligand receptors, and for antagonists for peptide/cytokine receptors there is a remarkable coincidence of the ligand binding site with the position of the retinal chromophore in either bacteriorhodopsin or rhodopsin [Henderson et al., 1990; IJzerman et al., 1992; Baldwin, 1993]. The possibilities of using the smallest possible ligand, a one-atom ligand, for probing the receptor domains was demonstrated for the “zinc-receptor” that was engineered from a neurokinin receptor [Elling et al., 1995]. Data on the naturally occurring one-atom ligand receptor, the extracellular Ca²⁺-sensitive GPCR, are sparse [Garrett et al., 1995].

Agonists and antagonists

Even in such cases where ligand size is not the limiting factor for activity/affinity analysis, the existence of multiple “affinity states” for ligands, especially if not exclusively agonists, makes the deduction of cause-effect relationships very difficult. A ternary complex model has been proposed for the binding of ligands to GPCRs to explain these multiple “affinity states.” In this model the ternary complex of agonist-receptor-G protein is energetically favoured over the binary agonist-receptor complex. Agonists therefore experience two “affinity states” (affinity being the inverse of the equilibrium dissociation constant) [De Lean et al., 1980; Lefkowitz et al., 1993; Samama et al., 1993]. The effects of binary and ternary complex formation on ligand binding characteristics are tentatively mediated by conformational changes of the receptor architecture. The allosteric interactions between agonists and guanine nucleotides influencing such conformational changes have been described for, e.g., the serotonin 5HT_1A and the adenosine A₁ receptor [Mahle et al., 1992]. Recently, random saturation mutagenesis was used to demonstrate similar effects in the muscarinic m1 and m2 receptors [Page et al., 1995]. If the number of intermediary states in rhodopsin [Zvyaga et al., 1993; Fahmy et al., 1995] is any measure for the number of possible states in other G protein-coupled receptors, the interpretation of mutagenesis results will become even more complex.

Aligning and Interpreting G Protein-Coupled Receptor Sequences

Structure prediction

It is now generally accepted that GPCRs consist of a single polypeptide featuring seven α-helical transmembrane domains (TMs), an extracellular N-terminus, and an intracellular C-terminus (Fig. 1). GPCRs are therefore sometimes referred to as “7TM receptors.” The prediction of GPCR structure from its nucleotide or amino acid sequence is, however, less than straightforward [von Heijne and Manoil, 1990]. Most sequence analysis software packages, such as GCG (Sequence Analysis Software Package of the Genetics Computer Group; University of Wisconsin, Madison, WI; in our laboratory version 7.3.1-UNIX of this program was run on an SGI Challenge XL; Silicon Graphics Inc., Mountain View, CA, MIPS R4400 CPU), are tuned towards structure prediction for (soluble) globular proteins. Because GPCRs are not only membrane bound proteins, but also have to traverse the lipid membrane, their structure and properties are inherently different from the datasets used for structure predictions. The α-helical segments of the GPCRs that actually span the lipid bilayer, designated transmembrane domains (TMs) or transmembrane helical domain (TMHs), are more lipophilic than the solvent (water) exposed loops connecting them. Residues with a higher hydrophobic (or lipophilic) index tend to form β-strands/sheets in globular proteins, and GCG will invariably assign a β-strand structure to these TMs, whereas they are presumed to be α-helical in GPCRs [Schertler et al., 1993; Baldwin, 1994; Ballesteros and Weinstein, 1995]. A better indicator for locating TMs, although it does not assign any structure, is the Kyte-Doolittle (KD) hydrophobicity parameter [Kyte and Doolittle, 1982]. A caveat in such predictions is that GPCRs are thought to be composed of multiple TMs within a single contiguous sequence that are grouped together in a seven-membered helical bundle. Transmembrane domains in GPCRs therefore, often display an amphipatic character (lipid exposed on one surface, and protein/solvent exposed on the other surface), which greatly reduces the feasibility of predicting TMs by analysis of the Kyte-Doolittle hydrophobicity indices alone. This is especially pertinent when the TMs are part of the ligand binding domain and contain one or more charged residues [Oliveira et al., 1993]. In the case of the biogenic amine receptors, the conserved aspartate residue in TM3 (D3.32; sec below) is sufficient to cause a “dip” in the KD profile. The KD profile of the P_2U purinergic receptor, expressing multiple basic residues in the ligand binding domain, is greatly offset by the presence of these residues and makes accurate assignment of TMs virtually impossible [Erb et al., 1995]. More sophisticated methods using, e.g., Emini surface probability analysis [Emini et al., 1985], or amino acid preferences for specific locations near the ends of α-helices [Richardson and Richardson, 1988; White and Jacobs, 1990; Zhang and Weinstein, 1994; van Rhee et al., 1995], have been used to better define the TMs. Several such methods, including the use of α-helical periodicity, were reviewed recently by Ballesteros and Weinstein [1995].

Figure 1.

Schematic presentation of the general topology of GPCRs. NT = N-terminus; CT = C-terminus; Ix = intracellular loop x; Ex = extracellular loop x; α indicates a supposed α-helical fragment of I3; β indicates a supposed β-pleated sheet substructure in I2; S-S indicates a possible cystine bond between TM3 and TM5; stacked circles indicate tentative ribosylation sites. Glycosylation has been shown to occur at either or both NT and E2; S-CO denotes a possible acylation site in CT. The numbering of the TMs is shown in the inset in the upper right corner.

Automated methods

Various algorithms, with a varying degree of accuracy, have been implemented in automated (computer) analysis and prediction methods. Automated methods for sequence retrieval and pattern searching are available from many sources on the internet [e.g., URL: http://www.nih.gov/ or http://www.ncbi.nlm.nih.gov/], as are possibilities to obtain secondary structure predictions [URL: http://www.embl-heidelberg.de/], or even complete models of receptors [URL:http://expasy.hcuge.ch/].

Alignment strategies

To compare whole GPCR sequences (retrieved from either the SwissProtein or Genbank databases maintained at the National Center for Biotechnology Information; NCBI, Bethesda, Maryland), or the effects of residue substitutions between two or more GPCRs, with each other, it is useful to construct a sequence alignment. Sequence analysis packages like GCG will generate such an alignment based on residue identity (conserved residues: the arginine near the C-terminus of TM3 is “always” conserved as an arginine) or homology parameters (e.g., “L may substitute for M, but not for K.”). To optimize the alignment, gaps are introduced in the sequence to maximize the number of identical residue pairs, and pairs with the highest degree of homology. Since these gaps in the sequence would ultimately translate into “gaps” in the structure, which is not tolerable in the structurally explicit α-helical segments, the introduction of gaps in the sequence may be penalized by 2 parameters. The first one used by GCG is the “GapWeight” (“where can a gap be introduced?”), and the second one is the “GapLength Weight” (“how many gaps, if any, can be introduced sequentially?”). Setting GapWeight to 5, and GapLengthWeight to 0.2, is usually sufficient to align closely related sequences to a satisfactory level. For more divergent sequences, however, the sequence alignment needs to be optimized by other means [Probst et al., 1992; Cserzö et al., 1994]. For the family of rhodopsin-related GPCRs (family A), the use of “conservation patterns” in optimization of alignments has been proven useful [Oliveira et al., 1993; Ballesteros and Weinstein, 1995]. The alignment of the TMs of seven divergent GPCRs is presented in Table 2, along with the “conservation patterns” used in its construction. (A more comprehensive alignment will be made available electronically through URL: http://mgddk1.niddk.nih.gov:8000/GPCR.html).

TABLE 2.

Alignment of Transmembrane Domains of Selected GPCRs

TM1
β2	P07550	33
rho	P08100	37
NK2	P21452	33
LH/CG	P22888	359
A2a	P29274	6
V2	P30518	37
P2U	P41231	33
TM2
β2	P07550	69
rho	P08100	73
NK2	P21452	69
LH/CG	P22888	395
A2a	P29274	52
V2	P30518	75
P2U	P41231	69
TM3
β2	P07550	117
rho	P08100	111
NK2	P21452	117
LH/CG	P22888	440
A2a	P29274	78
V2	P30518	113
P2U	P41231	107
TM4
β2	P07550	146
rho	P08100	149
NK2	P21452	144
LH/CG	P22888	479
A2a	P29274	117
V2	P30518	152
P2U	P41231	146
TM5
β2	P07550	197
rho	P08100	201
NK2	P21452	195
LH/CG	P22888	524
A2a	P28274	175
V2	P30518	203
P2U	P41231	195
TM6
β2	P07550	273
rho	P08100	252
NK2	P21452	250
LH/CG	P22888	569
A2a	P29274	233
V2	P30518	271
P2U	P41231	244
TM7
β2	P07550	302
rho	P08100	282
NK2	P21452	283
LH/CG	P22888	599
A2a	P29274	264
V2	P30518	301
P2U	P41231	281

^*A short receptor name, the accession number (SwissProtein), and the first residue number for the sequence are indicated at each line. The sequence identifiers (extended notation) are indicated underneath each column. Interconversion of notation schemes can be achieved by applying the following: (1.50) = 117H = 130O = I:18B; (2.50) 209H = 224O = II:14B; (3.50) = 326H = 340O = III:25; (4.50) = 406H = 420O = IV:11B; (5.50) = 511H = 520O = V:14B; (6.50) = 615H = 620O = VI:18B; (7.50) = 722H = 730O = VII:18B; where H = notation as proposed by Hibert et al [1991]; O = notation as proposed by Oliveira et al [1993]; and B = notation as proposed by Baldwin [1993].

Sequence identifiers

The use of sequence identifiers, a term coined by Ballesteros and Weinstein [1995], in sequence alignments could aid in the understanding of similarities in, and differences between, GPCR sequences. Although Hibert et al. [1991] and Oliveira et al. [1993] do not explicitly claim a need for an integrative method of numbering corresponding residues in different receptor sequences, they apparently feel compelled to offer such a scheme. Baldwin [1993], on the other hand, clearly states that “an integrated numbering scheme for positions in the structure is necessary.” Each of the aforementioned authors offered their own solution to this problem. Hibert et al. [1991] use a system in which the start of the helix is 1 plus 100 times the assigned number of the helix. Each residue is numbered sequentially from thereon. Example: P215 in human rhodopsin becomes P511. This scheme is very simple to apply, eminently suitable for building models of receptors, and facilitating comparisons between receptor models within the same subset. Oliveira et al. [1993] recognize that helical TMs can be characterized by a conservation pattern, and propose that: “In each helix the number of the most conserved residue is a multiple of 10 while the start of the helix is at the same time as close as possible to 100 times the number of the helix.” Example: P215 in human rhodopsin becomes P520. This system has as an advantage over the method used by Hibert et al. [1991], because, not only are helices easily identified, but the alignment of sequences with a lower homology is also taken into consideration. Baldwin’s approach differs significantly from these two, and encompasses a system where: Each helix is numbered in roman numerals with the number assigned to it. Each residue is then numbered sequentially from the (estimated) onset of the helix. “The numbering scheme allows for each helix to include 26 residues.” [Baldwin, 1993] Example: P215 in human rhodopsin becomes V:14P. This method greatly resembles the scheme proposed by Oliveira et al. [1993], but is not very suitable for the purpose of building molecular models, since these have to adhere to a format consistent with the Arabic numbering used in all Protein Data Bank (PDB) files, or program-dependent internal coordinate files derived thereof. The one drawback to all three proposals is that the numbering is dependent on the assignment of the apparent/estimated start of the TMs. The aforementioned authors disagree on the assignment of this particular point, considering the range of identifiers assigned to P215 in human rhodopsin. The start of TM5 is thought to originate either 11 [Hibert et al., 1991], 13 [Oliveira et al., 1993], or 14 [Baldwin, 1993] residues upstream, i.e., towards the N-terminal side of the TM, of this conserved proline residue. Ballesteros and Weinstein [1995] use a novel approach with a dual numbering scheme: “Every amino acid identifier starts with the transmembrane helix number, e.g., 4 for TMH4, and is followed by the position relative to a reference residue among the most conserved amino acid in that TMH. That reference residue is arbitrarily assigned the number 50.” Example: P215 in human rhodopsin becomes P_5.50(215). In effect, the numbering scheme becomes independent of the perceived (absolute) start of the TM, and is therefore more generally applicable. This is advantageous, not only for GPCR molecular modellers, but perhaps even more so for molecular biologists, who have no mathematical tools to establish such criteria. Yet another proposition was put forward, where the residue numbering of bacteriorhodopsin is used throughout a model [Teeter et al., 1994]. Given the low similarity between GPCR sequences and the bacteriorhodopsin sequence (15% homology at best, with virtually no identity), and the existing disagreement on the alignment of GPCRs and bacteriorhodopsin, the method is not very well suited for comparing larger numbers of GPCR sequences. In this paper we use a slightly modified version of the Ballesteros and Weinstein [1995] proposal that is in accordance with the numbering used in protein sequence databases and constitutes common practice in molecular biology. Although we adhere to the reasoning of Ballesteros and Weinstein [1995], we propose that the residue maintains its original sequence residue number, and is supplemented with a new extension containing the residue identifier. Example: P215 in human rhodopsin becomes P215(5.50). This will circumvent the foreseeable problems arising from drastically renumbering all residues (P511 [Hibert et al., 1991], P520 [Oliveira et al., 1993], V:14P [Baldwin, 1993], and P_5.50 [Ballesteros and Weinstein, 1995] all point to the proline residue at sequence position 215 of human rhodopsin). Moreover, it keeps the numbering schemes backward- and forward-compatible, and adds functionality to the notation rather than obscuring it. An example of this “extended notation,” used throughout this paper, is presented in Table 2. To aid in the interconversion of residue designators, we have added notations for the most conserved residues in each helix as footnotes in the table.

Genetic Heterogeneity

Receptor subtypes

Considerable heterogeneity exists in the expression levels of GPCRs in various tissues [Peralta et al., 1987]. The adenosine receptors were initially divided in A₁ and A₂ receptors based on their pharmacological profile, but only later was it confirmed that the A₂ receptors could be subdivided into A_2a and A_2b receptor subtypes based on their respective sequences [Rollins et al., 1994].

Allelic variations

Another form of genetic heterogeneity is allelic variation. This form of heterogeneity is largely neglected in depositing and retrieving amino acid or nucleotide sequences in databases. Allelic variation is responsible for phenotypic differences within a single species, and for the species differences themselves. Natural variability by allelic variation accounts for, e.g., the pigmentation phenotypes regulated by the MSH receptor within one species [Robbins et al., 1993]. Species differences resulting from (minor) genetic variability have been documented for the human and hamster β₂ adrenoceptors [Caron et al., 1988], the rat and human serotonin 5HT_1B receptors [Parker et al., 1993], the human and rat neurokinin NK1 receptors [Jensen et al., 1994], adenosine A₃ receptors [Linden, 1994], mammalian adenosine A₁ receptors [Meng et al., 1994; Tucker et al., 1994], and primate dopamine D₄ receptors [Livak et al., 1995]. Moreover, allelic variation can result in an abundance of pathological conditions [Raymond, 1994]. Mutations in the vasopressin V2 receptor resulting in decreased sensitivity to vasopressin, with nephrogenic diabetes insipidus as the somatic characteristic, have been well documented [Pan et al., 1992; Bichet et al., 1993; Rosenthal et al., 1993; Tsukaguchi et al., 1993; Birnbaumer et al., 1994; Faa et al., 1994; Knoers et al., 1994; Rosenthal et al., 1994; Wildin et al., 1994]. Constitutive activation of rhodopsin in some allelic mutants leads to a decrease in the signal-to-noise ratio from light-sensitive rods in the retina. Under conditions of low light intensity this results in a condition known as retinitis pigmentosa, hence “night-blindness” [Robinson et al., 1992; Zvyaga et al., 1993; Kaushal and Khorana, 1994; Li et al., 1995; Macke et al., 1995; Millán et al., 1995; Richards et al., 1995]. Cases of naturally occurring mutations leading to constitutively active GPCRs have been described for the TSH receptor in hyperthyroidism [Parma et al., 1993; Paschke et al., 1994a; Ohno et al., 1995], and the LH/CG receptor in male precocious puberty [Shenker et al., 1993; Yano et al., 1995]. Nocturnal asthma [Turki et al., 1995], dominantly inherited bleeding disorder [Hirata et al., 1994], and an induced eating disorder in mice [Tecott et al., 1995] can be traced to hyporesponsiveness of the β-adrenoceptor, the TXA₂, and the serotonin 5HT_2C receptor, respectively.

Alternative splicing

The third form of genetic heterogeneity is alternative splicing, although strictly speaking the heterogeneity is not expressed at the genomic, but at the protein level. Alternative splicing has been demonstrated for dopamine D₂ receptors [Dal’ Toso et al., 1989; Monsma et al., 1989; Guiramand et al., 1995], where the D_2L splice variant contains a 29 amino acid insert in the third intracellular loop compared to the D_2S splice variant, resulting in differential coupling to adenylyl cyclase. Divergence caused by alternative splicing, was also demonstrated for the C-terminus of the human endothelial TXA₂ receptor [Raychowdhury et al., 1994]. In the rat, alternative splicing of the vasopressin V2 receptor accounted for the existence of two protein products: the active V_2L receptor, or complete sequence, comprising 85% of the receptor population, and the inactive V_2S receptor, consisting of TMs 1 through 6 only, comprising 15% of the receptor population [Firsov et al., 1994]. Exon-shuffling in the human adenosine A₁ receptor, without resulting protein heterogeneity, was demonstrated to regulate the level of expression of these receptors [Ren and Stiles, 1994; Deckert et al., 1995].

Structural Segmentation

GPCRs can be divided into 15 discreet structural segments (Fig. 1), although the assignment of specific residues at the interface between two such segments to either segment is highly contentious. Also, due to low sequence homology in the non-helical segments, even at low stringency, it becomes very difficult to assign residue identifiers to particular residues outside the more conserved regions.

The structural information on the intracellular and extracellular regions of GPCRs is very limited. A high flexibility in these regions is implied by electron diffraction studies of the GPCR rhodopsin [Schertler et al., 1993], and the proton pump protein bacteriorhodopsin [Henderson et al., 1990]. The motility in these regions is too high to contribute sufficiently to the electron density of the protein to allow structure elucidation.

As indicated in Figure 1, several structural features have been posited for GPCRs, such as a disulfide bond between cysteines located near the N-termini of TM3 and TM5, respectively, a β-pleated sheet in I2, an α-helical fragment in I3, glycosylation sites in NT and E2, and an acylation site in the C-terminal tail [van Galen et al., 1992], that will not be discussed within the limits of this review.

N-terminus (NT)

As was to be expected, the involvement of the NT in agonist binding has been demonstrated for GPCRs of the peptide/cytokine class [La Rosa et al., 1992; Baumgartner et al., 1994; Hjorth et al., 1994; Paschke et al., 1994b; Chini et al., 1995].

First intracellular loop (I1)

Very few data have been published on the significance of the first intracellular loop in either G protein-coupling or receptor activation, but this segment was indicated to be, at least, involved in the coupling of the human TXA₂ receptor to its effector [Hirata et al., 1994].

First extracellular loop (E1)

Not surprisingly, the first extracellular loop, like the N-terminus, was shown to be important in agonist binding for GPCRs of the cytokine/peptide class [Haraguchi et al., 1994b; Hjorth et al., 1994; Chini et al., 1995].

Second intracellular loop (I2)

The second intracellular loop has been implicated in the coupling of rat m3 acetylcholine receptors to phospholipase C [Blüml et al., 1994a]. If some degree of structural homology among all GPCRs, whether they are members of the rhodopsin-like subfamily or not, may be inferred, recent experiments with metabotropic glutamate receptors regarding G protein-coupling suggest that the second intracellular loop is a major determinant in deciding G protein-coupling specificity [Pin et al., 1995].

Second extracellular loop (E2)

Even though in the cartoon representation of Figure 1 there is considerable distance between the N-terminus and the second extracellular loop, in a 3D representation (model) of GPCRs these segments are in proximity. It is therefore only logical that this segment is also involved in binding of agonists to GPCRs of the peptide/cytokine class [Hjorth et al., 1994; Kosugi and Mori, 1994]. However, there is mounting evidence that this particular segment is also involved in the binding of agonists to small ligand GPCRs, such as adenosine receptors [Olah et al., 1994b].

Third intracellular loop (I3)

There is ample evidence for the involvement of the third intracellular loop in the coupling of GPCRs to G proteins. Variations in the sequence can result in several different functional effects. The best documented effect is probably the effect of mutations on coupling selectivity [Strader et al., 1987b; Kosagi et al., 1992; Blount and Krause, 1993; McAllister et al., 1993; Samama et al., 1993; Shapiro et al., 1993; Tsukaguchi et al., 1993], but coupling efficiency is also affected by these variations [Monsma et al., 1989; Guiramand et al., 1995]. Modulation of coupling specificity and efficiency is ultimately reflected in the levels of homologous [Moro et al., 1993; Samana et al., 1993], and heterologous regulation observed [Lee and Fraser, 1993]. Direct physical evidence for the interaction of the third intracellular loop and a G protein (homologue) was provided by cross-linking experiments between bovine rhodopsin and the α-subunit of transducin through the use of cysteine mutants [Resek et al., 1994; Ridge et al., 1995].

Third extracellular loop (E3)

The third extracellular loop is relatively short when compared to the second extracellular loop. It varies in length between at most 20 (e.g., prostaglandin EP₃ receptor), and possibly as few as 5 amino acid residues (e.g., adenosine A₁ receptor). The peptide backbone is subject to space/size constraints: individual helices may not be too close to each other to avoid van der Waals repulsion, but must be close enough to satisfy the “contiguous chain” criterion. It is therefore likely that this loop serves a general structural function, rather than direct involvement in ligand binding. The effects observed for mutations in rat and human AT₁ receptors [Hjorth et al., 1994] could reflect either possibility, and more data are needed before any definitive conclusions can be drawn.

C-terminus (CT)

As is the case for the third intracellular loop, there is ample evidence for the involvement of the C-terminus in various aspects of GPCR-to-G protein coupling [Blount and Krause, 1993; Goujon et al., 1994; Haraguchi et al., 1994a; Sasakawa et al., 1994]. In addition, there is evidence that specific residues are palmitoylated [Karnik et al., 1993], a feature involved in anchoring protein sequences to the lipid membrane bilayer, or phosphorylated [Lattion et al., 1994; Wang et al., 1995], which is thought to be important in “post-activation processing.”

Transmembrane domains

This area comprises the most extensive and rigorous part of mutational analyses performed to date. Because of the rigidity (both in requirements, and in physical coordinates) imposed by the organization of the TMs in a helical bundle, point mutations in the TMs may affect changes in affinity of agonists, antagonists, or activity exclusively, or affect more than one parameter concurrently. To facilitate comparison of residue positions between multiple GPCR sequences, we constructed alignments of over 150 sequences, and annotated it with the “extended notation.” Table 2 represents a selection of 7 sequences out of the complete alignment. A comprehensive, although most likely incomplete, list was compiled of point mutations reported in the literature (Table 3). (It is the intent of the authors to make this list electronically available at the URL above, and to update it at regular intervals after publication of this paper. The cooperation of all authors, present and future, in maintaining the database would be greatly appreciated). This list will be discussed below, in detail. Figure 3 is an inventory of the number of mutants reported in the literature before August 1995, and their effect(s), sorted by sequence identifier. This effectively yielded a geographical map of already investigated, and yet to discover areas in the architecture of GPCRs. The map could then be used to draw inferences for sequences where mutations at a specific identifier have not yet been performed [e.g., van Rhee et al., 1995], or to pinpoint areas of interest for new mutagenesis studies [e.g., Kim et al., 1995].

TABLE 3.

Genetically Engineered and Naturally Occurring Point Mutations in GPCRs

Receptor	Species	Mutation (identifier)	Effect on functiona	Reference
m5/m2	Human	T37A(1.39)	AG & AN & act. restored	Liu et al., 1995
V2	Human	L44F(1.39)	decr. act.	Knoers et al., 1994
NK3	Rat	L81V(1.41)	n.e.	Wu et al., 1994
NK3	Rat	F88L(1.48)	n.e.	Wu et al., 1994
rho	Bovine	T58R(1.53)	decr. act.	Min et al., 1993
V2	Human	L62P(1.57)	decr. act.	Knoers et al., 1994
TXA2	Human	R60L(2.36)	decr. act. (PI)	Hirata et al., 1994
β2	Human	C77S(2.48)	n.e	Fraser et al., 1989b
β2	Hamster	C77V(2.48)	AG & AN n.a.	Dixon et al., 1987
β2	Hamster	C77V(2.48)	AG & AN n.a.	Dohlman et al., 1990
m1	Rat	C69S(2.48)	decr. AG; AN & act. n.a.	Savarese et al., 1992
5HT1A	Rat	D82N(2.50)	decr. AG; AN & act. n.a.	Ho et al., 1992
5HT1A	Human	D82N(2.50)	decr. AG; AN & act. n.a.	Ho et al., 1992
5HT2	Rat	D120N(2.50)	decr. AG & AN; decr. act.	Wang et al., 1993a
α2	Porcine	D79N(2.50)	decr. act.; AG & AN n.a.; Na	Horstman et al., 1990
β2	Human	D79N(2.50)	decr. act.; AG & AN n.a.	Wang et al., 1991
α2a	Murine	D79N(2.50)	incr. AG; decr. act. (K); act. n.a. (AC)	Suprenant et al., 1992
AT1a	Rat	D74E(2.50)	AG & AN n.a.; decr. act.	Bihoreau et al., 1993
AT1a	Rat	D74N(2.50)	AG n.a.; AN aff.; decr. act.	Bihoreau et al., 1993
β2	Hamster	D79A(2.50)	decr. AG; AN n.a.	Strader et al., 1987a
β2	Hamster	D79A(2.50)	decr. AG; AN & act. n.a.	Strader et al., 1988
β2	Hamster	D79E(2.50)	decr. AG; AN & act. n.a.	Strader et al., 1989
β2	Human	D79N(2.50)	decr. AG; AN n.a.; decr. act.	Wang et al., 1989
β2	Human	D79N(2.50)	decr. AG; AN & act. n.a.	Chung et al., 1988
D2	Rat	D80A,E(2.50)	decr. AG & act.; Na; AN n.a.	Neve et al., 1991
ET-A	Human	D126A(2.50)	selec. AG; decr. act. (PLC)	Rose et al., 1995
ET-B	Human	D147A(2.50)	AG & AN n.a.; decr. act. (PI)	Rose et al., 1995
GRH	Human	N87D(2.50)	decr. AG; decr. act. (PI)	Davidson et al., 1994b
LH/CG	Rat	D383N(2.50)	decr. AG & act.	Ji and Ji, 1991
LH/CG	Rat	D383N(2.50)	decr. AG & act.; Na	Quintana et al., 1993
m1	Rat	D71N(2.50)	inc. AG; AN n.a.; decr. eff.	Fraser et al., 1989b
NK1	Human	E78A(2.50)	n.e.	Rosenkilde et al., 1994
rho	Bovine	D83A(2.50)	n.e.	Zhukovsky and Oprian, 1989
SST2	Murine	D89N(2.50)	AG & act. n.a.	Kong et al., 1993
TRH	Murine	D71A(2.50)	decr. AG & AN & act.	Perlman et al., 1992
V2	Human	D85N(2.50)	decr. act.	Knoers et al., 1994
ET-A	Human	Y129A(2.53)	sub. selec. AG	Krystek et al., 1994
ET-B	Human	H150A,Y(2.53)	selec. AG & AN	Rose et al., 1995
NK3	Rat	V121M(2.53)	incr. AN; spec. selec.	Wu et al., 1994
V2	Human	V88H(2.53)	decr. act.	Knoers et al., 1994
V2	Human	V88M(2.53)	n.e.	Bichet et al., 1993
NK1	Human	M81A(2.54)	n.e.	Fong et al., 1994b
5HT2	Human	F125L(2.55)	n.e.	Choudhary et al., 1993
ET-A	Human	D133A(2.57)	decr. AN; AG & act. n.a.	Rose et al., 1995
ET-B	Human	D154A(2.57)	n.e.	Rose et al., 1995
m3	Rat	S120A(2.57)	decr. AN; AG & act. n.a.	Wess et al., 1991
NK1	Human	N85A(2.57)	decr. AN; AG n.a.	Rosenkilde et al., 1994
MSH	Murine	E92K(2.60)	const. act.	Robbins et al., 1993
m1	Human	Y82F(2.61)	AG n.a.	Drubbisch et al., 1992
NK1	Human	N89A(2.61)	n.e.	Rosenkilde et al., 1994
NK1	Human	Y92A(2.64)	decr. AN; AG n.a.	Rosenkilde et al., 1994
NK3	Rat	G133A(2.65)	incr. AN; spec. selec.	Wu et al., 1994
NK1	Human	N96A(2.68)	n.e.	Rosenkilde et al., 1994
V2	Human	P108M,V(2.73)	n.e.	Pan et al., 1992
AT1a	Rat	C101G(3.25)	decr. AG	Yamano et al., 1992
β2	Hamster	C106V(3.25)	n.e.	Dixon et al., 1987
rho	Bovine	C110A(3.25)	decr. stab.	Davidson et al., 1994a
β2	Hamster	E107A(3.26)	n.e.	Strader et al., 1987a
D2	Human	D108C(3.26)	AN n.a.; SH +	Javitch et al., 1995
m1	Rat	D99N(3.26)	decr. AG & AN; act. n.a.	Fraser et al., 1989b
P2U	Human	K107I(3.26)	n.e.	Erb et al., 1995
V2	Human	R113W(3.26)	decr. AG; act. n.a.; imp. proc.?	Birnbaumer et al., 1994
D2	Human	I109C(3.27)	AN n.a.; SH+	Javitch et al., 1995
D2	Human	F110C(3.28)	decr. AN; SH +	Javitch et al., 1995
NK2	Human	Q109H(3.28)	decr. AG; AN n.a.	Bhogal et al., 1994
rho	Bovine	E113A,Q(3.28)	destab. Schiff base	Lin et al., 1992
rho	Bovine	E113Q(3.28)	const. act.	Cohen et al., 1992
rho	Bovine	E113Q(3.28)	const. act.	Robinson et al., 1992
D2	Human	V111C(3.29)	AN n.a.; SH ++	Javitch et al., 1995
NK1	Human	N109A(3.29)	n.e.	Fong et al., 1994b
P2U	Human	K110L(3.29)	n.e.	Erb et al., 1995
TRH	Murine	T102V(3.29)	n.e.	Perlman et al., 1994b
D2	Human	T112C(3.30)	AN n.a ; SH−	Javitch et al., 1995
TRH	Murine	Y103F(3.30)	n.e.	Perlman et al., 1994b
D2	Human	C113C(3.31)	decr. AN; SH−	Javitch et al., 1995
5HT1A	Human	D116N(3.32)	decr. AG; AN & act. n.a.	Ho et al., 1992
5HT2	Rat	D155N(3.32)	decr. AG & AN	Wang et al., 1993a
α2	Human	D113N(3.32)	decr. act. & AN	Wang et al., 1991
β2	Hamster	D113E(3.32)	decr. AG & AN; act. n.a.	Strader et al., 1988
β2	Hamster	D113E(3.32)	decr. AG & AN; AN as part. AG	Strader et al., 1989
β2	Hamster	D113N(3.32)	decr. AG & AN	Strader et al., 1987a
β2	Hamster	D113Q(3.32)	decr. AG & AN; act. n.a.	Strader et al., 1988
β2	Hamster	D113S(3.32)	decr. AN aff.; AG mod.	Strader et al., 1991
D2	Human	D114C(3.32)	decr. AN; SH ++	Javitch et al., 1995
D2	Rat	D114G,N(3.32)	decr. AG & AN	Mansour et al., 1992
H1	Human	D107A(3.32)	decr. AN & act.	Ohta et al., 1994
H1	Human	D107N(3.32)	decr. AG & AN & act.	Ohta et al., 1994
H1	Human	D107E(3.32)	decr. AN & act.	Ohta et al., 1994
H2	Canine	D98N(3.32)	decr. AG & AN & act.	Gantz et al., 1992
m1	Rat	D105(3.32)	binds aff. reag.	Kurtenbach et al., 1990
m1	Rat	D105E(3.32)	decr. AG; AN & act. n.a.; uncoupling?	Page et al., 1995
m1	Rat	D105N(3.32)	decr. AG & AN	Fraser et al., 1989b
m1	Rat	D105N(3.32)	decr. AG & AN & act.	Page et al., 1995
m2	Rat	D103E(3.32)	decr. AG; AN & act. n.a.; uncoupling?	Page et al., 1995
NK1	Human	P112A(3.32)	n.e.	Fong et al., 1994b
rho	Bovine	A117F(3.32)	Blue shift	Ridge et al., 1992
TRH	Murine	Q105A,E,L,V(3.32)	decr. AG; act. n.a.; perturb. BD	Perlman et al., 1994b
D2	Human	V115C(3.33)	decr. AN; SH ++	Javitch et al., 1995
ET-B	Rat	K181D(3.33)	decr. AG; act. n.a.	Zhu et al., 1992
m3	Rat	Y148F(3.33)	decr. AG; AN & act. n.a.	Wess et al., 1991
m3	Rat	Y148F(3.33)	decr. AG	Wess et al., 1992b
TRH	Murine	Y106F(3.33)	decr. AG; act. n.a.	Perlman et al., 1994a
D2	Human	M116C(3.34)	AN n.a.; SH−	Javitch et al., 1995
D2	Human	M116L(3.34)	AG & AN n.a.	Mansour et al., 1992
AT1	Human	N111A(3.35)	incr. AG; AG selec.; decr. AN	Perlman et al., 1995b
β2	Hamster	C116V(3.35)	AG & AN n.a.	Dixon et al., 1987
D2	Human	M117C(3.35)	AN n.a.; SH−	Javitch et al., 1995
D2	Human	M117C,G(3.35)	AG & AN n.a.	Mansour et al., 1992
A2a	Human	T88A,R,S(3.36)	decr. AG; AN n.a.	Jiang et al., 1995
D2	Human	C118(3.36)	AN n.a.; SH + +	Javitch et al., 1995
NK1	Rat	L116V(3.36)	spec. selec. AN	Jensen et al., 1994
NK1	Human	V116L(3.36)	reversed selec. AN	Fong et al., 1992b
NK2	Human	M117L(3.36)	decr. AN; AG n.a.	Bhogal et al., 1994
A2a	Human	Q89A(3.37)	inc. AG & AN; selec. AN	Jiang et al., 1995
A2a	Human	Q89H,R(3.37)	decr. AN; AG n.a.; perturb. BD	Jiang et al., 1995
A2a	Human	Q89L,N,S(3.37)	incr. AG; AN n.a.; perturb. BD	Jiang et al., 1995
D2	Human	T119C(3.37)	decr. AN; SH−	Javitch et al., 1995
rho	Bovine	E122A,Q(3.37)	blue shift	Ridge et al., 1992
rho	Bovine	E122D(3.37)	n.e.	Ridge et al., 1992
rho	Bovine	E122Q(3.37)	n.e.	Lin et al., 1992
rho	Bovine	E122Q(3.37)	blue shift	Zvyaga et al., 1993
TRH	Murine	N110A,S(3.37)	decr. AG	Perlman et al., 1994b
A2a	Human	S90A(3.38)	incr. AG; AN n.a.	Jiang et al., 1995
D2	Human	A120C(3.38)	AN n.a.; SH−	Javitch et al., 1995
A2a	Human	S91A(3.39)	n.e.	Jiang et al., 1995
β2	Hamster	S120A(3.39)	decr. expr.	Dixon et al., 1987
β2	Hamster	S120A(3.39)	imp. proc.	Strader et al., 1989a
D2	Human	S121C(3.39)	decr. AN; SH +	Javitch et al., 1995
D2	Human	I122C(3.40)	AN n.a.; SH +	Javitch et al., 1995
D2	Human	L123C(3.41)	AN n.a.; SH−	Javitch et al., 1995
rho	Bovine	W126A,L(3.41)	blue shift	Ridge et al., 1992
rho	Bovine	W126F(3.41)	n.e.	Ridge et al., 1992
V2	Human	Y128S(3.41)	decr. act.	Pan et al., 1992
V2	Human	Y128S(3.41)	decr. act.	Bichet et al., 1993
V2	Human	Y128S(3.41)	decr. act.	Faa et al., 1994
D2	Human	N124C(3.42)	AN n.a.; SH−	Javitch et al., 1995
TRH	Murine	Y106F(3.42)	decr. AG	Perlman et al., 1994a
D2	Human	L125C(3.43)	decr. AN; SH +	Javitch et al., 1995
β2	Hamster	C125V(3.44)	AG & AN n.a.	Dohlman et al., 1990
D2	Human	C126(3.44)	AN n.a.; SH−	Javitch et al., 1995
D2	Human	A127C(3.45)	AN n.a.; SH−	Javitch et al., 1995
V2	Human	A132D(3.45)	decr. act.	Bichet et al., 1993
D2	Human	I128C(3.46)	AN n.a.; SH−	Javitch et al., 1995
D2	Human	S129C(3.47)	dec. AN; SH +	Javitch et al., 1995
D2	Human	I130C(3.48)	AN n.a.; SH−	Javitch et al., 1995
5HT2	Rat	D172N(3.49)	decr. AG & AN	Wang et al., 1993a
α2	Porcine	D130N(3.49)	AG & AN n.a.	Horstman et al., 1990
α2a	Human	D130N(3.49)	AN n.a.; decr. AG & act.	Wang et al., 1991
β2	Human	D130N(3.49)	AN n.a.; incr. AG; decr. act.	Fraser et al., 1988
LH/CG	Rat	E441D,Q(3.49)	AG & AN & act. n.a.; imp.proc.	Wang et al., 1993b
m1	Rat	D122N(3.49)	act. & AN n.a.; incr. AC	Fraser et al., 1989b
rho	Bovine	E134Q(3.49)	decr. act.	Franke et al., 1992
rho	Bovine	E134Q(3.49)	n.e.	Lin et al., 1992
rho	Bovine	R135L,W(3.50)	decr. act.	Zvyaga et al., 1993
V2	Human	R137H(3.50)	decr. act.	Bichet et al., 1993
V2	Human	R137H(3.50)	AG n.a.; decr. act.	Rosenthal et al., 1993
V2	Human	R143P(3.56)	decr. act.	Tsukaguchi et al., 1993
α2a	Human	I150V(4.42)	n.e.	Wang et al., 1991
H1	Human	S155A(4.47)	n.e.	Moguilevsky et al., 1995
m3	Rat	W192F(4.50)	decr. AG & AN; act. n.a.	Wess et al., 1993
β2	Hamster	S161A(4.53)	n.e.	Strader et al., 1989a
iodop	Chicken	S177A(4.53)	red shift (int. w. ionone ring)	Lin et al., 1994
rho	Bovine	A164S(4.53)	n.e.	Chan et al., 1992
V2	Human	S167L(4.53)	decr. act.	Knoers et al., 1994
NK2	Human	L160F(4.54)	n.e.	Bhogal et al., 1994
β2	Human	T164I(4.56)	decr. AG & eff.	Turki et al., 1995
β2	Hamster	S165A(4.57)	imp. proc.	Strader et al., 1989a
m3	Rat	P201A(4.59)	decr. AG & AN; act. n.a.	Wess et al., 1993
NK1	Human	P164A(4.59)	n.e.	Fong et al., 1994b
NK1	Human	Q165A(4.60)	decr. AG & AN	Fong et al., 1994b
NK1	Human	Q165N,S(4.60)	AN selec.; decr. AG	Fong et al., 1994b
NK2	Human	C167G(4.61)	n.e.	Bhogal et al., 1994
AT1a	Rat	R167A(4.63)	decr. AG & AN	Yamano et al., 1995
NK1	Human	Y168A(4.63)	n.e.	Fong et al., 1994b
AT1a	Rat	H166A(4.64)	decr. AG & AN	Yamano et al., 1995
NK1	Human	S169A(4.64)	AN selec.	Fong et al., 1994b
V2	Human	R181C(4.67)	decr. act.	Bichet et al., 1993
AT1a	Rat	E173A(4.71)	n.e.	Yamano et al., 1995
V2	Human	C185C(4.71)	decr. act.	Bichet et al., 1993
rho	Bovine	C187A(4.76)	decr. stab.; imp. proc.	Davidson et al., 1994a
rho	Human	C187S(4.76)	decr. act.; imp. proc.?	Karnik et al., 1988
rho	Human	C187Y(4.76)	decr. act.; imp. proc.?	Richards et al., 1995
AT1a	Rat	H183A(5.26)	n.e.	Yamano et al., 1995
AT1a	Rat	E185A(5.28)	n.e.	Yamano et al., 1995
NK2	Human	G190N,K(5.31)	n.e.	Bhogal et al., 1994
NK2	Human	G191K(5.32)	n.e.	Bhogal et al., 1994
NK1	Human	E193L(5.35)	decr. AN	Gether et al., 1994
NK2	Human	L194E,T(5.35)	n.e.	Bhogal et al., 1994
V2	Human	R202C(5.35)	decr. act.	Bichet et al., 1993
NK1	Human	K194L(5.36)	decr. AN	Gether et al., 1994
V2	Human	Y205C(5.38)	decr. act.	Bichet et al., 1993
m3	Rat	T231A(5.39)	decr. AG; AN & act. n.a.	Wess et al., 1991
m3	Rat	T231A(5.39)	decr. AG	Wess et al., 1992b
NK1	Human	H197A,Y,S,K(5.39)	decr. AN; AG n.a.	Fong et al., 1993a
NK1	Human	H197F,Q(5.39)	n.e.	Fong et al., 1993a
NK2	Human	F198A,L(5.39)	decr. AG & AN	Bhogal et al., 1994
A2a	Human	F180A(5.41)	n.e.	Kim et al., 1995
5HT1A	Human	S198A(5.42)	decr. AG; AN & act. n.a.	Ho et al., 1992
5HT1A	Rat	S198A(5.42)	decr. AG; AN & act. n.a.	Ho et al., 1992
A2a	Human	N181S(5.42)	AG selec.; AN n.a.	Kim et al., 1995
α2a	Human	S200A(5.42)	decr. AG; AN & act. n.a.	Wang et al., 1991
AT1a	Rat	K199A(5.42)	decr. AG; AN n.a.	Yamano et al., 1995
AT1a	Rat	K199Q(5.42)	decr. AG; AG selec.	Yamano et al., 1992
β2	Hamster	S203A(5.42)	imp. proc.	Strader et al., 1989a
D1	Human	S198A(5.42)	decr. AG & AN; act. n.a.; imp. proc.?	Pollock et al., 1992
D2	Rat	S194A(5.42)	AG & AN n.a.	Mansour et al., 1992
H1	guinea pig	T203A(5.42)	n.e.	Leurs et al., 1994
H1	Human	T194A(5.42)	decr. AG; incr. AN; act. n.a.	Ohta et al., 1994
H1	Human	T194A(5.42)	decr. AG; incr. AN; act. n.a.	Moguilevsky et al., 1995
H2	Canine	D186A(5.42)	decr. AN & act.; AG n.a.	Gantz et al., 1992
H2	Canine	D186N(5.42)	decr. AN & act.; AG n.a.	Gantz et al., 1992
m3	Rat	T234A(5.42)	decr. AG & act.; AN n.a.	Wess et al., 1991
NMB	Rat	I216S(5.42)	decr. AG & act.	Fathi et al., 1993
5HT1A	Human	T199A(5.43)	decr. AG; act. n.a.	Ho et al., 1992
A2a	Human	F182A(5.43)	decr. AG & AN & act.	Kim et al., 1995
A2a	Human	F182W(5.43)	decr. AG; AN n.a.	Kim et al., 1995
A2a	Human	F182Y(5.43)	decr. AG; AN n.a. (selec.?)	Kim et al., 1995
α2b	Murine	S182C(5.43)	incr. AN	Link et al., 1992
α2-C10	Human	C201S(5.43)	spec. selec	Link et al., 1992
β2	Hamster	S204A(5.43)	decr. AG; AN n.a.	Strader et al., 1989b
D1	Human	S199A(5.43)	decr. AG & AN; act. n.a.	Pollock et al., 1992
m3	Rat	T234A(5.43)	decr. AG; act. n.a.	Wess et al., 1992b
NK2	Human	I202V(5.43)	decr. AG; AN n.a.	Bhogal et al., 1994
5HT2A	Human	S242A(5.46)	spec. selec. AG; incr. AN	Kao et al., 1992
α2a	Human	S204A(5.46)	decr. AG; AN & act. n.a.	Wang et al., 1991
β2	Human	S207A(5.46)	decr. AG; AN n.a.	Strader et al., 1989b
D1	Human	S202A(5.46)	AG selec.; AN & act. n.a.	Pollock et al., 1992
D2	Rat	S197A(5.46)	AG & AN selec.	Mansour et al., 1992
H1	Human	N198A(5.46)	decr. AG; AN & act. n.a.	Ohta et al., 1994
H1	guinea pig	N207A(5.46)	decr. AG & AN; AG selec.; act.n.a.	Leurs et al., 1994
H1	Human	N198A(5.46)	decr. AG & act.; AN n.a.	Moguilevsky et al., 1995
H2	Canine	T190A(5.46)	decr. AN; act. n.a.	Gantz et al., 1992
H2	Canine	T190C(5.46)	decr. AN; act. n.a.	Gantz et al., 1992
rho	Bovine	H211C,F(5.46)	n.e.; conf. change?	Weitz and Nathans, 1992
rho	Bovine	H211F(5.46)	n.e.	Cohen et al., 1992
m3	Rat	P242A(5.50)	decr. AG; AN & act. n.a.	Wess et al., 1993
LH/CG	Human	I542L(5.56)	decr. AG; Const.act. (AC)	Laue et al., 1995
AT1a	Rat	Y215F(5.58)	AN n.a.; decr. AG & act. (PI)	Hunyady et al., 1995
β2	Hamster	Y219L(5.58)	AG & AN n.a.	Strader et al., 1989a
m3	Rat	R252H(5.60)	n.e.	Blüml et al., 1994a
m3	Rat	I253A(5.61)	n.e.	Blüml et al., 1994a
m3	Rat	Y254A,F,W(5.62)	decr. act. (PI)	Blüml et al., 1994d
m3	Rat	Y254S(5.62)	AG & AN n.a.; decr. act.	Blüml et al., 1994a
rho	Bovine	C240(5.75)	crosslinking to transducin (α)	Resek et al., 1994
LH/CG	Human	D564G(6.30)	const.act. (AC)	Laue et al., 1995
α1b	Hamster	A293X(6.34)	incr. AG; const. act.	Kjelsberg et al., 1992
α2a	Human	T348C,E,K(6.34)	incr. AG.; const. act.; AN n.a.	Ren et al., 1993
TSH	Rat	A623E,K(6.34)	incr. AG; decr. act.	Kosugi et al., 1992
TSH	Human	A623I(6.34)	const. act.	Parma et al., 1993
TSH	Human	A623V(6.34)	const.act (AC); act. n.a. (PI)	Paschke et al., 1994
LH/CG	Human	M571I(6.37)	const.act. (AC)	Laue et al., 1995
AT1a	Human	A572V(6.38)	const. act. (AC); incr. AG	Yano et al., 1995
TSH	Human	F631C(6.42)	AG & act n.a. (PI); const. act. (AC)	Kosugi et al., 1994
LH/CG	Human	T577I(6.43)	const.act. (AC)	Laue et al., 1995
TSH	Human	T632I(6.43)	AG & act n.a. (PI); const. act. (AC)	Kosugi et al., 1994
TSH	Human	T632I(6.43)	const. act. (AC); act.n.a. (PI)	Paschke et al., 1994
V2	Human	ΔV279(6.43)	decr. act.	Tsukaguchi et al., 1993
V2	Human	ΔV279(6.43)	decr. act.	Faa et al., 1994
LH/CG	Human	D578G(6.44)	AG n.a.; const. act. (AC)	Shenker et al., 1993
LH/CG	Human	D578G(6.44)	const.act. (AC)	Laue et al., 1995
LH/CG	Rat	D556N(6.44)	AG & act. n.a.	Ji and Ji, 1991
LH/CG	Human	D578Y(6.44)	high const.act. (AC) [phenotypic]	Laue et al., 1995
rho	Bovine	F261Y(6.44)	red shift	Chan et al., 1992
rho	Fish	Y261F(6.44)	blue shift	Yokoyama et al., 1995
TSH	Human	D633E(6.44)	AG & act n.a. (PI); const. act. (AC)	Kosugi et al., 1994
TSH	Human	D633Y(6.44)	AG & act. n.a. (PI); const. act. (AC)	Kosugi et al., 1994
SST5	Rat	G258F(6.46)	n.e.	Ozenberger and Hadcock, 1995
β2	Hamster	C285S(6.47)	decr. act.; AG & AN n.a.	Fraser et al., 1989a
β2	Hamster	C285V(6.47)	AG & AN n.a.	Dohlman et al., 1990
LH/CG	Human	C581R(6.47)	decr. AG; const.act. (AC)	Laue et al., 1995
m3	Rat	T502A(6.47)	n.e.	Wess et al., 1991
AT1a	Rat	W253A(6.48)	decr. AG; AN n.a.	Yamano et al., 1995
m3	Rat	W503F(6.48)	decr. AG & AN: act. n.a.	Wess et al., 1993
rho	Bovine	W265A,F,Y(6.48)	blue shift	Ridge et al., 1992
TRH	Murine	W279A(6.48)	n.e.	Perlman et al., 1995a
m3	Rat	P505A(6.50)	n.e.	Wess et al., 1993
V2	Human	P286L(6.50)	decr. act.	Faa et al., 1994
V2	Human	P286R(6.50)	decr. act.	Pan et al., 1992
v2	Human	P286R(6.50)	decr. act.	Bichet et al., 1993
5HT2	Human	F339L(6.51)	selec. AN; AG n.a.	Choudhary et al., 1993
AT1a	Rat	H256A(6.51)	n.e.	Yamano et al., 1995
CCK-B	Canine	L355V(6.51)	selec. AN; act. n.a.	Beinborn et al., 1993
CCK-B	Human	V319I(6.51)	selec. AN; AG n.a.	Beinborn et al., 1993
CCK-B	Human	V319L(6.51)	rev. selec. AN; AG n.a.	Beinborn et al., 1993
m3	Rat	Y506F(6.51)	decr. AG; selec. AN; act. n.a.	Wess et al., 1991
m3	Rat	Y506F(6.51)	decr. AG; act. n.a.	Wess et al., 1992b
NK2	Human	Y266F(6.51)	n.e.	Bhogal et al., 1994
rho	Bovine	Y268F(6.51)	n.e.	Ridge et al., 1992
5HT2	Human	F340L(6.52)	selec. AN; decr. AG & act.	Choudhary et al., 1993
A1	Bovine	H251L(6.52)	decr. AN; AG n.a.	Olah et al., 1992
A2a	Human	H250A(6.52)	decr. AG & AN & act.	Kim et al., 1995
A2a	Human	H250F(6.52)	selec. AG; AN n.a.	Kim et al., 1995
A2a	Human	H250Y(6.52)	selec. AG; AN n.a.	Kim et al., 1995
iodop	Chicken	T282A(6.52)	red shift (int. w. ionone ring)	Lin et al., 1994
m3	Rat	N507A(6.52)	decr. AN & AG; act. n.a.	Blüml et al., 1994b
m3	Rat	N507D(6.52)	decr. AN; AG selec.	Blüml et al., 1994b
m3	Rat	N507S(6.52)	const. act.	Blüml et al., 1994b
NK1	Human	H265A(6.52)	decr. AN; AG & act. n.a.	Zoffmann et al., 1993
NK1	Human	H265A,F(6.52)	AN selec.; spec. selec.	Fong et al., 1994a
NK1	Human	H265F,Q(6.52)	n.e.	Zoffmann et al., 1993
NK1	Human	H265Q,S,Y(6.52)	AN selec.; spec. selec.	Fong et al., 1994a
rho	Bovine	A269T(6.52)	red shift	Chan et al., 1992
SST5	Rat	F265Y(6.52)	AG& sub. select.; act. n.a.	Ozenberger and Hadcock, 1995
TRH	Murine	R283H,K(6.52)	decr. AG; act. n.a.	Perlman et al., 1995a
TRH	Murine	R283A,E,L,S(6.52)	decr. AG & act.	Perlman et al., 1995a
AT1a	Rat	F259A(6.54)	decr. AG; AN n.a.	Yamano et al., 1995
A2a	Human	N253A(6.55)	decr. AG & AN & act.	Kim et al., 1995
A2a	Human	N253Q(6.55)	decr. AG & AN	Kim et al., 1995
A2a	Human	N253S(6.55)	decr. AG & AN	Kim et al., 1995
AT1a	Rat	T260A(6.55)	n.e.	Yamano et al., 1995
P2U	Human	R265L(6.55)	decr. AG	Erb et al., 1995
iodop	Chicken	Y274(6.57)	Stabiliz. Schiff base	Lin et al., 1994
AT1a	Rat	D263A(6.58)	decr. AG & AN	Yamano et al., 1995
NK1	Human	P271G(6.58)	n.e.	Gether et al., 1994
NK2	Human	G273P(6.58)	decr. AG & AN	Bhogal et al., 1994
NK2	Human	G273T(6.58)	decr. AG; AN n.a.	Bhogal et al., 1994
A2a	Human	F257A(6.59)	decr. AG & act.	Kim et al., 1995
NK1	Human	Y272T,A(6.59)	decr. AN; selec.	Gether et al., 1994
NK2	Human	S274T,Y(6.59)	n.e.	Bhogal et al., 1994
AT1a	Rat	C274G(7.25)	decr. AG	Yamano et al., 1992
AT1a	Rat	D278A(7.29)	decr. AG; AG selec.; AN n.a.	Hjorth et al., 1994
AT1	Human	D278A(7.29)	decr. AG; AG selec.; AN n.a.	Perlman et al., 1995b
AT1a	Rat	D281A(7.32)	decr. AG; AG selec; incr. AN	Hjorth et al., 1994
AT1	Human	D281A(7.32)	decr. AG; AG selec.; incr. AN	Perlman et al., 1995b
A2a	Human	Y271A(7.36)	decr. AG & act.	Kim et al., 1995
m2/m5	Human	T423A(7.36)	imp. proc.	Liu et al., 1995
m2/m5	Human	T423,E,H,N(7.36)	AG & AN & act. restored	Liu et al., 1995
P2U	Human	K289I(7.36)	reversed selec. AG	Erb et al., 1995
TXA2	Human	L291F(7.36)	decr. AG & AN; imp. proc.	Funk et al., 1993
NK1	Human	I290S(7.38)	reversed selec. AN	Fong et al., 1992a
NK1	Human	I290S(7.38)	decr. AG; spec. selec.	Sachais et al., 1993
NK1	Rat	S290I(7.38)	spec. selec. AN	Jensen et al., 1994
NK2	Human	L292I(7.38)	AG selec.; decr. AN	Bhogal et al., 1994
NK2	Human	L292S(7.38)	AG selec.; AN n.a.	Bhogal et al., 1994
5HT1A	Human	N385V(7.39)	decr. AN; AG n.a.	Guan et al., 1992
5HT1B	Human	T355N(7.39)	incr. aff.; spec. selec.	Oksenberg et al., 1992
5HT1B	Human	T355N(7.39)	decr. AG & AN; spec. selec.	Parker et al., 1993
A2a	Human	I274A(7.39)	decr. AG & AN; act. n.a.	Kim et al., 1995
α2-C10	Human	F412N(7.39)	AN & AG sub. selec.	Suryanarayana et al., 1991
β2	Human	N312A(7.39)	dec. AN; act n.a.; AN sel.	Suryanarayana et al., 1993
β2	Human	N312F(7.39)	imp. proc.	Suryanarayana et al., 1991
β2	Human	N312F(7.39)	imp. proc.	Suryanarayana et al., 1993
β2	Human	N312Q,T(7.39)	decr. AG & AN; act. n.a.	Suryanarayana et al., 1993
m3	Rat	Y529F(7.39)	decr. AG; selec. AN; act. n.a.	Wess et al., 1991
m3	Rat	Y529F(7.39)	decr. AG	Wess et al., 1992b
P2U	Human	R292L(7.39)	decr. AG	Erb et al., 1995
rho	Bovine	A292D(7.39)	n.e.	Ridge et al., 1992
rho	Human	A292E(7.39)	const. act.	Dryja et al., 1993
TRH	Murine	R306A,E,L(7.39)	decr. AG; act. n.a.	Perlman et al., 1995a
TRH	Murine	R306K(7.39)	n.e.	Perlman et al., 1995a
β2	Turkey	W330(7.40)	photochem. lab.	Wong et al., 1988
m3	Rat	W530F(7.40)	n.e.	Wess et al., 1993
EP3	Murine	R309E,V(7.40)	decr. AG	Huang and Tai, 1995
EP3	Human	R309K(7.40)	incr. AG	Huang and Tai, 1995
TXA2	Human	R295Q(7.40)	decr. AG & AN	Funk et al., 1993
A1	Human	T277S,A(7.42)	decr. AG; AN n.a.; spec. sel.	Townsend-Nicholsen and Schofield, 1994
A2a	Human	S277A(7.42)	decr. AG; AN & act. n.a.	Kim et al., 1995
A2a	Human	S277C(7.42)	decr. AG; AN n.a.	Jiang et al., 1995
A2a	Human	S277N,T(7.42)	n.e.	Kim et al., 1995
m1	Rat	C407S(7.42)	decr. AG; act. n.a.	Savarese et al., 1992
A1	Bovine	H278L(7.43)	decr. AG & AN	Olah et al., 1992
A2a	Human	H278A(7.43)	decr. AG & AN; act. n.a.	Kim et al., 1995
A2a	Human	H278F,Q,Y(7.43)	decr. AG & AN	Kim et al., 1995
β2	Hamster	Y316S(7.43)	imp. proc.?	Strader et al., 1989a
m3	Rat	Y533F(7.43)	decr. AG & AN; act. n.a.	Wess et al., 1991
m3	Rat	Y533F(7.43)	decr. AG	Wess et al., 1992b
rho	Bovine	K296A,E,G(7.43)	const. act.	Robinson et al., 1992
rho	Bovine	K296A,E,G,H(7.43)	const. act.	Cohen et al., 1992
rho	Human	K296E(7.43)	const. act.	Li et al., 1995
TXA2	Human	w299L(7.44)	decr. AN; AG n.a.	Funk et al., 1993
TXA2	Human	W299R(7.44)	decr. AG & AN	Funk et al., 1993
AT1	Human	N294A(7.45)	decr. AN; AG n.a.	Perlman et al., 1995b
β2	Hamster	N318K(7.45)	decr. AG; AN n.a.	Strader et al., 1987a
A2a	Human	S281A(7.46)	decr. AG & AN & act.	Kim et al., 1995
A2a	Human	5281N(7.46)	inc. AG; decr. AN; selec. AN	Jiang et al., 1995
A2a	Human	S281T(7.46)	selec. AG & AN	Kim et al., 1995
β2	Hamster	S319A(7.46)	decr. AG; AN n.a.	Strader et al., 1989a
m3	Rat	T537A(7.47)	n.e.	Wess et al., 1991
GRH	Human	D318N(7.49)	n.e.	Davidson et al., 1994b
β2	Hamster	P323S(7.50)	imp. proc.	Strader et al., 1987a
m3	Rat	P540A(7.50)	decr. act.; AG & AN n.a.	Maggio et al., 1993
m3	Rat	P540A(7.50)	incr. AG; AN n.a.; decr. act.	Wess et al., 1993
GRH	Human	S320A(7.51)	n.e.	Davidson et al., 1994b
m1	Rat	C417S(7.52)	n.e.	Savarese et al., 1992
β2	Human	C327R(7.54)	decr. AG & act.	O’Dowd et al., 1988
β2	Human	C327S(7.54)	AG & AN n.a.	Fraser et al., 1989a
α2	Porcine	D432N(7.58)	AG & AN n.a.	Horstman et al., 1990
rho	Bovine	C322S(7.69)	n.e.	Karnik et al., 1993
rho	Bovine	C323S(7.70)	n.e.	Karnik et al., 1993

^adecr., decreased; incr., increased; n.a., not affected, ≤3-fold difference with wild type receptor; n.e., no effect or ≤3-fold difference with wild type receptor; imp. proc., impaired processing, either translational/posttranslational/transport; AG;, agonist (affinity); AN, antagonist (affinity); act., activity per se; aff., any measure of affinity (K_d/K_i/EC₅₀/IC₅₀); eff., efficacy; perturb. BD, size-dependent perturbation of binding domain; (rev.) selec., (reversed) ligand selectivity; spec. selec., species selectivity; sub. selec., subtype selectivity; const. act., constitutive activity; (K), potassium channel activity; (AC), adenylyl cyclase activity; (PI), phosphatidyl inositol turnover activity; (PLC), phospholipase C activity; SH, probed with sulfhydryl reagent: − = does not react, + = reacts, + + = reacts strongly; Na, affects sodium modulatory capacity; part. AG, partial agonist(s)/agonism; conf. change, involved in conformational change(s); photochem. lab., photochemically labeled; stab, (photochemical) stability. Notes 1 “decr./incr.” denotes a >3-fold difference with the wild type receptor, regardless whether a smaller difference is deemed significant by statistical analysis. A 3-fold difference in affinity corresponds to 0.5 log unit on the concentration scale. A 3-fold decrease in affinity corresponds to a loss of approximately 1 kcal/mol in binding energy, which is less than the loss of 1 hydrogen bond (typically between 3 and 6 kcal/mol). 2 Affinity is defined by the equilibrium constants derived from any ligand-receptor interaction. In the absence of such values as K_d, (equilibrium dissociation constant), K_i (equilibrium inhibitory displacement constant) or IC₅₀ (concentration of inhibitory ligand that displaces 50% of the marker ligand) from binding studies, EC₅₀ or IC₅₀ (concentration of ligand that displays a semi-maximal excitatory or inhibitory effect, respectively) values derived from functional assays may be used. Although such functional parameters are frequently obtained under non-equilibrium conditions, for simplicity they will be treated equivalent to equilibrium constants for the purpose of this paper. 3 The term ‘activity’ in this paper is used solely for the pupose of identifying effects on the activity of the receptor-effector system, and does not reflect leftward or rightward shifts of dose-response curves. Neither does it reflect ‘intrinsic activity’, or other parameters involved in the actions of partial agonists. Consequently, the failure of a system to respond to stimulation by virtue of a loss of affinity, should not be considered a loss of activity. However, it is virtually impossible to distinguish between lack of response due to severe loss of affinity (the system will not respond at supramaximal stimulation), and that due to an impaired response system (the system will not respond to any stimulation whatsoever).

Figure 3.

Mutation distribution in the transmembrane regions of GPCRs from Table 3. Open bars (□) represent the number of ineffective mutants reported; shaded bars (▧) represent the number of reports where agonist affinity (AG) is affected; closed bars (■) represent the number of reports where antagonist affinity (AN) is affected.

Chimeric Receptors

Before focusing entirely on single-point mutations, it should be pointed out that the use of chimeric receptors has contributed considerably to the understanding of receptor structure and function.

Constructs for “homo-chimeric” receptors (mixed constructs of receptors belonging to the same subfamily) have been reported for adrenergic [Kobilka et al., 1988; Liggett et al., 1993; Samama et al., 1993], muscarinic [Kubo et al., 1988; Wess et al., 1992a; Ellis et al., 1993; Pittel and Wess, 1994; Liu et al., 1995], dopaminergic [MacKenzie et al., 1993; McAllister et al., 1993; Fishburn et al., 1994; Kozell et al., 1994; Robinson et al., 1994], serotoninergic [Choudhary et al., 1992], neurokinin [Blount and Krause, 1993; Gether et al., 1993a, b; Vigna et al., 1994], adenosine [Olah et al., 1994a, b], and cholecystokinin receptors [Mantamadiotis and Baldwin, 1994]. One such chimeric construct offers the first evidence in favour of the clockwise arrangement of TMs in the helical bundle (when viewed from the intracellular surface), as opposed to a possible anti-clockwise orientation [Liu et al., 1995]. A clockwise arrangement had been deducted earlier, based on physicochemical parameters, but experimental evidence had been lacking so far [Baldwin, 1993; Baldwin, 1994]. Liu et al. [1995] constructed hybrids containing m5 sequence in TM1, and m2 sequence in TM7. This construct is incompatible with ligand binding and receptor function. However, single point mutations into either TM1 (m5:T37(1.39) → m2:A30(1.39)) or TM7 (m2:T423(7.36) → m5:H478(7.36)) were able to rescue this hybrid. It was concluded that the sites on TM1 (1.39) and TM7 (7.36) are located at the interhelical interface, and therefore the helical bundle must be organized in a clockwise fashion (as viewed from the intracellular membrane surface).

“Hetero-chimeric” constructs have been proven useful in determining ligand specificity for a TRH/β₂ chimeric receptor [Kosugi and Mori, 1994], and in investigating G protein selectivity for a chimeric receptor containing a β₁ sequence insertion in either an m1 or an m2 receptor [Wong and Ross, 1994].

Insertion/Deletion Mutagenesis

When aligning amino acid sequences, particular care has to be taken to avoid random introduction of gaps. Gaps in one sequence signify the absence of structure in that sequence, when compared to another sequence not containing that particular gap. By the same reasoning, insertion and deletion mutagenesis can be applied to intentionally introduce structural aberration in GPCRs. Consequently, because the structure of the receptor is targeted, effects observed resulting from this approach are not easily attributed to specific interactions. Genetically engineered [Blount and Krause, 1993; Lee and Fraser, 1993; Maggio et al., 1993; Shapiro et al., 1993; Blüml et al., 1994a, c; Lattion et al., 1994; Sasakawa et al., 1994; van Koppen et al., 1994], or naturally occurring [Monden et al., 1992; Haraguchi et al., 1994a; Wang et al., 1995] insertion/deletion mutagenesis products, most notably affect the activity of these receptors.

Site-Directed Mutagenesis (Point Mutations)

Recently, we published a systematic investigation of single amino acid residue replacements within the TMs of the human adenosine A_2a receptor, and correlated the findings with ligand SAR data accumulated over the years [Kim et al., 1995; Jiang et al., 1995]. Many similar studies for other receptor subtypes (predominantly those of the biogenic amine subfamily) exist [e.g., Strader et al., 1994]. Another approach is systematic probing of the receptor surface by measuring the substituted-cysteine accessibility [Javitch et al., 1995]. The ultimate probe, would be the use of a ligand existing of a single atom. The feasibility of this approach was demonstrated by converting a NK1 receptor into a receptor for zinc ions [Elling et al., 1995].

Before venturing into the analysis of the mutations described in Table 3, one should consider which amino acids are replaced, and even more so with what other residue it is replaced. This leads, inevitably, to a different effect for homologous amino acid substitutions, than for techniques using substitutions with non-homologous residues as used in, e.g., the “alanine-scanning” procedure. The “alanine-scanning” procedure targets specific residues in the sequence, and replaces them with an alanine regardless of the nature of the wild type residue. This contrasts with the “fine-tuning” approach of replacing, e.g., an aspartate residue with either an asparagine or a glutamate, which has been applied extensively in characterizing the conserved aspartate in TM2 (Table 3: identifier 2.50). In Table 3 (and Fig. 3), correlating mutation position with mutation effect, no distinction between those two approaches was (or can be) made. Furthermore, no distinction is made between receptor (sub)types or ligands in the summation of Figure 3. Reports marked “ineffective” or “no effect” should therefore not necessarily be considered contradictory to other reports.

Moreover, the incidence (the frequency at which a particular amino acid residue appears in the sequence) of amino acids varies, not only between the 20 types of naturally occurring amino acids, but also between the segments of GPCRs. To illustrate this principle we calculated the incidence of each amino acid in either the whole sequence or only the TMs for 40 GPCRs chosen randomly from the alignment of over 100 GPCRs. To illustrate this principle we calculated the incidence of each amino acid in either the whole sequence or only the TMs for 40 GPCRs chosen randomly from the alignment of over 150 GCPRs (Fig. 2). The incidence of an amino acid in the TMs was deemed significantly different from the incidence of the same amino acid in the complete sequence (= reference value) if its value deviated by more than 10% from the reference value. If all amino acids were distributed randomly, the incidence for each amino acid would be 5%. The most abundant amino acid (more than twice the expected value if randomly distributed), in either the whole sequence or the TMs was L, with an incidence of 11.4%, and 14.6%, respectively. The lowest incidence in TMs was recorded for Q (1.0%), H (1.1%), and E (1.3%), whereas the lowest incidence in the whole sequence was observed for H (1.8%), and W (1.8%). The presence of residues that ordinarily occur with a low incidence in a particular sequence cannot be regarded as a random occurrence and begs for an explanation.

Figure 2.

Amino acid distribution throughout the GPCRs used in Table 3.

Perhaps more significant than the incidence of particular residues in the whole sequence is the incidence of the residues in the TMs. More abundant in TMs than in the whole sequence are A (8.7 vs. 7.6%), F (7.9 vs. 5.4%), I (10.3 vs. 6.6%), L (14.6 vs. 11.4%), M (3.5 vs. 2.6%), V (10.8 vs. 7.3%), W (2.3 vs. 1.8%), and Y (4.9 vs. 3.7%). This is to be expected considering the increase in the (KD) hydrophobicity profile, necessary to span the length of the lipid bilayer. Less abundant in TMs than in the whole sequence are: D (1.9 vs. 3.2%), E (1.3 vs. 3.8%), G (3.8 vs. 5.6%), H (1.1 vs. 1.8%), K (2.4 vs. 4.4%), P (3.5 vs. 5.3%), Q (1.0 vs. 2.9%), R (3.1 vs. 5.1%), and S (6.2 vs. 7.8%). Although for some residues (D, E, K, Q, R, and S) this decreased incidence could equally well be attributed to the requirements of the (KD) hydrophobicity profile, the profound structural implications of G and P residues interfere with the conditions of α-helical periodicity of the TMs [von Heijne, 1991; Sankararamakrishnan and Vishveshwara, 1992]. The low incidence of H in GPCR sequences as a whole, and its even lower incidence in the TMs therefore, signify an important role for it in the function of these receptors. Indeed, the two histidines present in the human adenosine A_2a receptor have been shown to be crucial for its function [Kim et al., 1995]. About equally distributed between TMs and whole sequences are: C (3.0 vs. 3.2%), N (4.0 vs. 4.3%), and T (5.8 vs. 6.1%). Apart from the apparent lack of correlation between position and special functionality represented by the cysteine, i.e., the ability to form disulfide bonds with adjacent cysteines [Ridge et al., 1995], no conclusions can be drawn from this particular observation.

Transmembrane domain 1 (TM1)

Fewer data are available for this TM. In accordance with its presumed structural role, rather than involvement in ligand binding [Baldwin, 1993, 1994], there is one report indicating the presence of position 1.39 at the interhelical contact surface with TM7 [Liu et al., 1995]. Other reports indicate that, mainly, the activity of the system is affected, further supporting a structural role [Min et al., 1993; Knoers et al., 1994; Wu et al., 1994].

An incidence of 21% for P at position 1.36 indicates that this position frequently functions as a helix initiator, and this coincides with an increased incidence for residues more prevalent in TMs following this position. The helical surface containing both positions 1.36, and 1.39, continues with helical periodicity along positions 1.43, 1.46, 1.50, 1.51, 1.55, and 1.58. This includes the most conserved residue (N at 1.50) in this TM. It also suggests that the S found at 1.43 in m1, m2, m4, and m5 receptors could play an important role in receptor structure. The most abundant residue at position 1.46 is G (38%), allowing for flexibility in the helix, rather than breaking the tertiary structure, whereas T (26%) and S (16%) are very likely to be involved in hydrogen bonding networks maintaining intrahelical or interhelical structure. G (10%) and S (15%) also occur at position 1.51. Position 1.55 shows an increase in W (11%) and Y (9%) residues, well above the average incidence rates of 2.3% and 4.9%, respectively. The first basic residues occur at position 1.58, indicating the protrusion of the TM into the negatively charged surface of the intracellular membrane environment, and ensuing helical periodicity seems absent until the initiation of TM2.

Transmembrane domain 2 (TM2)

It comes as no surprise that the most heavily targeted residue in this TM is the conserved aspartate at position 2.50 (Table 3). Most abundant are the reports on agonist affinity and activity of its mutants. Even the most conservative substitution, i.e., D → N or D → E, profoundly affect these parameters. It was postulated by Horstman et al [1990], that this residue in the α₂-adrenoceptor acted as an accessory site modulated by sodium ions. This hypothesis was corroborated by similar experiments on the D₂ dopamine [Neve et al., 1991], and the LH/CC: receptor [Ji and Ji, 1991; Quintana et al., 1993]. Although most reports indicate a loss of affinity for agonists, at least 2 reports to the contrary, i.e., an increase in agonist affinity, exist [Suprenant et al., 1992; Fraser et al., 1989b]. Antagonist binding to this residue has been implicated in only three cases: D(2.50)N [Wang et al., 1993a; Bihoreau et al., 1993] and D(2.50)A [Perlman et al., 1992], whereas 13 studies that explicitly investigated antagonist binding to mutant receptors found no difference between wild type and mutant behaviour. The majority of the reports suggest that the activity of the mutant receptor-ligand complexes is attenuated. It is therefore more likely that this particular position is involved in signal transduction, in casu signal propagation, than directly involved in agonist binding.

The periodicity implied by Figure 3B suggests that more residues on TM2, especially those towards the extracellular surface that are facing the interior of the helical bundle, are involved in signal transduction if not in ligand binding. Most receptor models indicate that TM2 is too far removed from other residues deemed essential, to be involved in direct ligand contact [Hibert et al., 1991; Hoflack et al., 1994; Trumpp-Kallmeyer et al., 1992; Zhou et al., 1994; Ballesteros and Weinstein, 1995]. This is substantiated by the effects observed at positions 2.53, 2.57, 2.60, and 2.64, provided the GPCR under investigation contains a “non-standard” amino acid at that position. The cumulative incidence of A, V, I, L, and M at position 2.53 is 69% vs. a total cumulative incidence of 47.7% over all positions in all TMs. “Non-standard” residues at this position occur in the GRH, green and red opsin (E), SST5, P_2Y, P_2U, and ET-A (Y), and ET-B (H) receptors. For position 2.57 the cumulative incidence is 71%, with “nonstandard residues” present in, e.g., the ET-A, and ET-B (D), all muscarinic (S), and all neurokinin (N) receptors. On the other hand, the cumulative incidence for positions 2.60, and 2.64 is 42%, and 33%, respectively, whereas the incidence for the relatively rare W (2.3% on all TMs) increases to 18% at position 2.60, and the equally rare Y (4.9% on all TMs) increases to 26%. More significant, probably, is the occurrence of P residues at positions 2.58 (28% of GPCRs), and 2.59 (43% of GPCRs) because of the structural implications of this particular residue. This P is not only not absolutely conserved (it is lacking in, e.g., all muscarinic receptors), but also not “position conserved.” Whereas in some receptors the helical angle between the conserved aspartate and the proline is only 80° (2.58), in other receptors it is 180° (2.59).

Transmembrane domain 3 (TM3)

The most heavily targeted residue in TM3 is the D that is conserved among all biogenic amine GPCRs at position 3.32 (Table 3). Aspartate (D) residues also occur at the equivalent position in opioid and somatostatin receptors, and the octopamine receptor, whereas a lysine (K) may be found in the GRH and IL-8 receptors, a glutamine (Q) in the TRH, vasopressin, endothelin, and neuromedin-B receptors, and a proline (P) in the neurokinin receptors. Site-directed mutagenesis of the conserved D in biogenic nitrogen receptors, usually led to decreased affinity of agonists and/or antagonists, although decreased activity also has been reported (Table 3). This loss of affinity is probably attributable to the binding requirements imposed by the quaternary amine present in all biogenic amine agonists, and in most antagonists. Replacing the P in the NK1 receptor with an A, apparently was without effect [Fong et al., 1994b]. Targeted mutagenesis of the Q in the TRH receptor led to a mainly size-dependent decrease in agonist affinity, but activity was not affected [Perlman et al., 1994b]. Even if this particular position is occupied by residues not generally involved in specific interactions, such as A in rhodopsin [Ridge et al., 1992], and V in the A_2a receptor [unpublished data], substitution lead to significant changes in receptor functionality.

Following the rules of strict helical periodicity, the next position up on the same helical face (i.e., towards the N-terminal end of the TM) as 3.32, is position 3.28 (or 3.29). The involvement of 3.28 in agonist binding (NK2) [Bhogal et al., 1994], antagonist binding (D₂) [Javitch et al., 1995], or coordination of the Schiff-base in rhodopsin has been demonstrated [Lin et al., 1992], whereas a (3.29)C mutant showed reactivity towards a thiol reagent (meaning it is solvent accessible), but no effect on ligand binding was observed [Javitch et al., 1995].

If the low-resolution structure of rhodopsin [Schertler et al., 1993] represents a general template for all (rhodopsin-related) GPCRs, then the fraction of the solvent- and protein-exposed surface of TM3 exceeds 180° due to the central position of this TM in the helical bundle. In this respect, this structure differs from the high resolution structure determined for bacteriorhodopsin [Henderson et al., 1990]. Support for a more central position of TM3 than suggested by the bacteriorhodopsin model, is provided by mutants at position 3.32 (see above), 3.33, 3.34, 3.35, and 3.36. Charged residues occur at position 3.33 in the endothelin receptors (K), IL-8 receptors (E), and formyl peptide receptors (D). Cysteine-replacement mutagenesis of 3.33 revealed involvement in agonist binding, and reactivity towards a thiol reagent, indicating that at least a 100° fraction of the helix is exposed [Javitch et al., 1995]. A similar mutation at position 3.34 was not efficacious [Mansour et al., 1992]. A (3.34)C mutant showed no reactivity towards a thiol reagent [Javitch et al., 1995], and mutations at 3.35 showed a similar profile [Dixon et al., 1987; Mansour et al., 1992; Javitch et al., 1995; Perlman et al., 1995b]. Residues at 3.36 were involved in ligand binding in several GPCR subfamilies, including among others, the neurokinin receptors [Fong et al., 1992b; Bhogal et al., 1994; Jensen et al., 1994], and were solvent accessible. Mutations at 3.37 also affected ligand binding, but was apparently not solvent accessible, and this site is therefore, likely located at a helix-helix interface [Javitch et al., 1995; Jiang et al., 1995]. Remarkable at position 3.37, is the presence of a glutamate (E) residue in rhodopsin [Lin et al., 1992; Ridge et al., 1992; Zvyaga et al., 1993] and the LH/CG receptor only, whereas the most prevalent residue is a threonine (T; 30% of GPCRs).

The first residue following this series, adhering to the helical periodicity requirements, is 3.39. This residue is solvent accessible [Javitch et al., 1995], and apparently influences the post-translational processing of the receptor protein [Dixon et al., 1987; Strader et al., 1989a]. In 85% of GPCRs this position is occupied by a either a serine (S) or a threonine (T), in 10% by a glycine (G), and in 4% by a proline (P) residue. Most subsequent residues seem to affect predominantly the activity of the ligand-receptor-effector system.

The conserved ‘DRY’ pattern near the C-terminus of the helix, seems mainly involved in receptor activity or G protein coupling (Table 3, Fig. 3). The D occurs in 83% of GPCRs, and is substituted with the highly homologous E in another 14% of GPCRs, totaling an acidic residue in 97% of GPCRs. In contrast, a H is found in the P_2Y and P_2U receptors, and an N in the PAF receptor. It is unclear whether this reflects a differential G protein-coupling mechanism/efficiency, and the inconsistencies reported for mutants at this position leave this issue unresolved (Table 3, Fig. 3). The R at position 3.50, on the other hand, is fully conserved, and mutants reported in certain pathological conditions invariably lead to a decrease in activity [Bichet et al., 1993; Rosenthal et al., 1993; Zvyaga et al., 1993]. Although a vast majority of GPCRs contain a Y at position 3.51 (84%), some degree of variability is allowed considering the occurrence of F (5%), C (4%), W (3%), H (2%), A (l%), and S (1%) at this position.

Transmembrane domain 4 (TM4)

The incidence of basic residues near the N-terminus (up till and including 4.43) indicates the proximity of phospholipids in the inner layer of the membrane [Richardson and Richardson, 1988]. The most conserved residue in this TM is W (96% of GPCRs) at 4.50, and only some prostaglandin receptors deviate from this rule. Mutation of W(4.50) to the somewhat similar F in the m3 receptor led to a decrease in both agonist and antagonist affinity, but the activity of the system was not affected [Wess et al., 1993]. Mutations at positions 4.53 [Strader et al., 1989a; Chan et al., 1992; Knoers et al., 1994; Lin et al., 1994] and 4.54 [Bhogal et al., 1994] are inconclusive, whereas mutations at positions 4.59 [Wess et al., 1993; Fong et al., 1994b] and 4.60 [Fong et al., 1994b] seem to affect both agonist and antagonist affinity, but not activity. Activity, or more accurately post-translational processing, is affected in a β₂ S(4.57)A mutant [Strader et al., 1989a], and considering the high incidence (57% S, and 27% A) of certain residues it is very likely involved in helix structure (intrahelical contacts) or packing of the helical bundle (interhelical contacts). Positions 4.59 and 4.60 have a high incidence of P residues (66% and 34%, respectively), indicating that this position is important for the formation of the ligand binding domain, although no obvious differentiation between GPCR subfamilies (either by ligand size, or ligand class) seems to exist. Reports of substitution effects on residues closer to the C-terminus of TM4 exist (Table 3, Fig. 3), but seem to be restricted to GPCRs of the peptide/cytokine class, which is consistent with prior observations and considerations.

Transmembrane domain 5 (TM5)

This is the first TM in which a proline residue is both present throughout the vast majority (84%) of GPCRs investigated (notable exceptions are the LH/CG receptor, all lipid mediator, and all melanocortin receptors), and its position in the sequence is conserved (without introducing gaps in the helical region). This reflects a probable role of its position in overall GPCR structure, rather than individual ligand binding domains. Since only one mutant has been reported, P(5.50)A [Wess et al., 1993], which affected only agonist affinity, but not antagonist affinity or receptor activity, such a structural role can neither be confirmed nor rejected.

Position 5.35 is the most N-terminal residue in TM5 for which efficacious mutants have been reported [Bichet et al., 1993; Gether et al., 1994]. This is consistent with the initiation of a helical periodicity pattern proceeding along the face of the helix composed of residues located at positions 5.38/5.39, 5.42/5.43, 5.46/5.47, 5.50, 5.53/554, 5.58, and 5.62/5.63, Positions 5.38 and 5.41 have a relatively high incidence of aromatic residues (10% F and 30% Y, and 13% F and 17% Y, respectively), whereas position 5.42 marks an increase in hydrophilic residues (27% S, 15% T, 3% N, 3% K, 2% D, 2% H, 1% Q, and 1% E). Substitutions mostly affect agonist affinity, with few reported effects on affinity of antagonists (Table 3, 5.42). There are also indications that point mutations at this position (5.42) may impair post-translational processing of the receptor protein [Strader et al., 1989a; Pollock et al., 1992]. The same phenomena have been observed for position 5.43 (Table 3), although the amino acid distribution is considerably different with 23% F, 17% S, 14% T, 11% L, and 10% A. This may reflect a different ligand binding environment between positions 5.42 and 5.43, where position 5.42 defines a mainly hydrophilic interaction surface, and position 5.43 defines either a hydrophobic (44%) or a hydrophilic (33%) surface. Since none of the amino acids is conserved in more than 50% of GPCRs, substantial heterogeneity exists at this position and may therefore reflect an essential role in ligand recognition and ligand specificity.

In accordance with the helical periodicity, position 5.46 has been proven essential for most agonist binding [Strader et al., 1989a; Wang et al., 1991; Kao et al., 1992; Mansour et al., 1992; Pollock et al., 1992; Leurs et al., 1994; Ohta et al., 1994; Moguilevsky et al., 1995], and less frequently for antagonist binding [Gantz et al., 1992; Kao et al., 1992; Mansour et al., 1992; Leurs et al., 1994]. This position is predominantly occupied by small amino acid residues (21% G, 20% S, and 19% A), but larger residues occur in several GPCRs (e.g., H in rhodopsin, Y in endothelin, and W in adenosine A₁ and A₃ receptors). There is preliminary evidence that the point mutation H(5.46)C, F in rhodopsin results in a conformational change [Weitz and Nathans, 1992], but this experiment was not confirmed by others [Cohen et al., 1992]. However, the high incidence of G at this position (21%) is consistent with a structural role of this position in the formation of the ligand binding site.

Few mutants have been reported C-terminal to this position, but there is some evidence that positions as far down as 5.58 may affect agonist binding [Hunyady et al., 1995]. Furthermore, residues ranging from 5.56 (not on the same helical face as positions mentioned earlier; this is reflected in the incidence of residues V 24%, L 17%, F 16%, and I 15%) to 5.62 (23% Y, 15% F, and 14% I) influence receptor activity [Laue et al., 1995].

Although not strictly belonging to TM5, residue C240 in rhodopsin, which could be designated C(5.75) (not counting gaps), was shown to be involved in coupling to transducin by direct cross-linking experiments [Resek et al., 1994].

Transmembrane domain 6 (TM6)

The hydrophilic helical periodicity seems to evolve from position 6.52, such that the absolutely conserved P(6.50) is located at the opposite surface of 6.52, facing the lipid membrane environment. The hydrophilic surface propagates from 6.52 towards the C-terminus into the extracellular environment, but is hardly detectable in the N-terminal sequence near the intracellular surface. The water-lipid interface at the N-terminus is marked by an increase in basic residues at position 6.35 (30% K, and 24% R). The low incidence of residues located between positions 6.35 and 6.47 that are capable of hydrogen bond formation, suggests that the N-terminal part of TM6 is not involved in direct ligand contact or ligand specificity. Mutations in this part of TM6, generally, lead to constitutively active receptors [Kjelsberg et al., 1992; Parma et al., 1993; Ren et al., 1993; Shenker et al., 1993; Kosugi et al., 1994; Paschke et al., 1994a; Laue et al., 1995; Yano et al., 1995], although impaired responses have been reported in several cases [Fraser et al., 1989a; Kosugi et al., 1992; Tsukaguchi et al., 1993; Faa et al., 1994; Laue et al., 1995]. The deletion mutation ΔV(6.43) reported in cases of diabetes insipidus effectively causes a rotational perturbation by about 100° within the helix. Whether this results in a changed orientation of only one face of the helix or completely destroys the helical packing is not yet clear. Although phenotyping for this condition has been performed [Tsukaguchi et al., 1993; Faa et al., 1994], no detailed data are available regarding ligand binding, and it is thought that the occurrence of diabetes insipidus results from decreased activity of the receptor, instead of constitutive activation. This further supports a more general role of the N-terminal part of TM6 in GPCR structure and activation, rather than strictly ligand binding.

Since residue 6.35 is presumably located at the water-lipid interface, it is very likely that residue 6.34, immediately preceding it, and other residues in its vicinity are involved in G protein coupling. Site-directed mutagenesis of residues at position 6.34 indicates that this supposition is viable, considering that mutations in this region invariably result in constitutive activity of the receptor-effector system [Kjelsberg et al., 1992; Kosugi et al., 1992; Parma et al., 1993; Ren et al., 1993; Paschke et al., 1994a]. In some cases, an increase in agonist affinity was also observed [Kjelsberg et al., 1992; Kosugi et al., 1992; Ren et al., 1993].

The residue at position 6.44 is most frequently F (91% of GPCRs), but this residue is replaced with a D in the LH/CG and the TSH receptors. This position must therefore be pointing into the central cavity of the helical bundle, or be involved in a specific interhelical contact. D(6.44)E, G, Y mutants were all constitutively active, but agonist affinity was apparently not affected [Shenker et al., 1993; Kosugi et al., 1994; Laue et al., 1995]. The homologous substitution D(6.44)N was well tolerated, and behaved like the wild type receptor [Ji and Ji, 1991]. The reciprocal substitutions F(6.44)Y (mammalian to fish) [Chan et al., 1992] and Y(6.44)F (fish to mammalian) [Yokoyama et al., 1995] in rhodopsin resulted in changed absorption spectra, which suggests the proximity of the retinal chromophore, and a position within the central cavity.

Cysteines have the unique ability to form dimers by establishing a covalent disulfide bond. The presence of cysteine residues, therefore, may indicate additional structural constraints within a protein. Although the incidence of C at position 6.47 is 80%, the presence of 6% T, 6% S, and residues other than these, indicate that this unique ability of C is subordinate to its role at this position. In site-directed mutagenesis studies with amino acids of similar size no effects on agonist or antagonist binding were observed, but the C(6.47)R mutant in the LH/CC, receptor caused constitutive activity of the system [Laue et al., 1995]. It would be interesting to measure the basal activity of GPCRs that contain larger residues at this position. Unfortunately, the only GPCRs reported to possess a large side chain substituent (F), are orphan receptors (EBI1, RBS1, and RTA) [Birkenbach et al., 1993; Harrison et al., 1994; Ross et al., 1990] and olfactory receptors. The 5HT_2A, 5HT_2B, and 5HT_2C receptors contain the larger than cysteine, but also sulfur-containing, methionine, and apparently do not display disparate behaviour. Moreover, since 92% of all GPCRs exhibit a residue at position 6.47 with hydrogen bonding capacity, it is very likely that this particular position takes over the peptide-backbone hydrogen-bonding network disrupted by P(6.50), which is supposedly facing the lipid phase of the membrane bilayer. A survey of the function of C, S, and T residues in protein crystal structures deposited in the PDB reveals that these amino acids are excellently suitable for reinforcing the peptide backbone [Ballesteros and Weinstein, 1995].

Adding to stabilization of the backbone, following destabilization by P(6.50), are most likely the residues located at positions 6.48 and 6.51. The occurrence of aromatic residues at position 6.48 (89% W, 5% F, and 2% Y), and the presence of aromatic residues at position 6.51 (49% F, and 28% Y), could be providing rigidity through non-covalent aromatic stacking (often, mistakenly, referred to as “π-π stacking”). Such a system would allow for a receptor-activation mechanism that requires conformational changes [Ballesteros and Weinstein, 1995]. Mutational analysis of either of these positions is inconclusive, bearing in mind the wide variety of effects observed [Wess et al., 1991, 1992b, 1993; Ridge et al., 1992; Beinborn et al., 1993; Choudhary et al., 1993; Yamano et al., 1995] or even the absence of effects [Ridge et al., 1992; Bhogal et al., 1994; Perlman, 1995a; Yamano et al., 1995]. Such observations, however, are consistent with a subtle balance between stabilizing and destabilizing effects.

Residues at position 6.52 are apparently involved in ligand recognition, more specifically ligand selectivity, considering the incidence of amino acids on the one hand; and the effects observed upon introduction of mutations on the other hand (Table 3, Fig. 3). This position is predominantly occupied by aromatic residues (38% F, and 22% H), although a wide selection of other residues is allowed (12% N, 8% Y, 4% T, 3% Q, 2% S, and 1% R). At the same position some GPCRs carry A, L, or V residues, but this seems restricted to GPCRs that have endogenous ligands consisting of (repeated) isoprene units, such as the opsins, rhodopsin, and the prostaglandin and thromboxane receptors. Site-directed mutagenesis of this position results in a remarkable number of reports citing effects on agonist or antagonist selectivity [Choudhary et al., 1993; Blüml et al., 1994b; Fong et al., 1994a; Kim et al., 1995; Ozenberger and Hadcock, 1995], instead of straightforward decreased affinities [Olah et al., 1992; Zoffmann et al., 1993; Perlman et al., 1995a].

One helical turn separated from position 6.52, position 6.55 offers an exceptional variety of amino acids (17 out of 20 of the the naturally occurring), with an incidence that differs extensively from the average distribution for TMs. In order of decreasing incidence, they are 18% N, 14% V, 10% L, 6% A, 6% F, 6% Y, 5% Q, 5% S, 5% E, 4% T, 4% M, 3% H, 3% R, 2% K, and 1% each of C, G, and I. Absent are P, D, and W. This offers the possibility to differentiate between ligands; an effect that was observed by mutational analysis of the N(6.55) residue in the adenosine A_2a receptor [Kim et al., 1995]. Such effects are even more pronounced in the case of the R(6.55) residue in the P_2U receptor [Erb et al., 1995], but could, seemingly, not be demonstrated for T(6.55) in the AT_1a receptor [Yamano et al., 1995].

The incidence of rare amino acids at position 6.58 is even more unusual: 11% N (vs. 4% in all TMs), 10% D (vs. 1.9% in all TMs), 6% R (vs. 3.1% in all TMs), 4% K (vs. 2.4% in all TMs), 3% Q (vs. 1.0% in all TMs), and 2% E (vs. 1.3% in all TMs). Site-directed mutagenesis of the AT_1a and NK2 receptors indicates that this position distinctly affects agonist and antagonist affinities [Bhogal et al., 1994; Yamano et al., 1995].

Residues occupying positions 6.59 through 6.62 are located within the last, distinguishable, full turn of the α-helix of TM6. Site-directed mutagenesis in this region has resulted in decreased affinity of agonists and decreased activity (A_2a) [Kim et al., 1995], or decreased affinity for antagonists (NK1) [Gether et al., 1994], but is not entirely consistent (NK2) [Bhogal et al., 1994].

Transmembrane domain 7 (TM7)

The helical periodicity for TM7 is initiated at position 7.29; propagates along residues 7.32, 7.35/7.36, 7.39/7.40, 7.43, 7.45/7.46, 7.49/7.50, 7.53; and terminates at position 7.57. The first helical turn is relatively rich in residues that are charged under physiological conditions, i.e., D, E, K, and R, indicating that this fragment may form an extension of the TM into the extracellular medium beyond the lipid membrane. The presence of P(7.31) in, e.g., muscarinic receptors, and P(7.32) in, e.g., the opsins, is consistent with initiation of an α-helix in this region of the sequence. The importance of this first helical turn in ligand binding, was extensively documented for the AT₁ receptor [Yamano et al., 1992; Hjorth et al., 1994; Perlman et al., 1995b].

The incidence of aromatic residues at position 7.35 is fairly high (23% F, 18% Y, 6% W), possibly signifying the entry point of the helix into the lipid environment of the membrane. The orientation of position 7.36, with regard to the arrangement of the TM in the helical bundle, is not entirely clear. Two reports suggest that this residue may be involved in interhelical contacts [Liu et al., 1995; Kim et al., 1995]. One report is explicit in this, citing a mutation that is incompatible with interhelical contacts with TM1, and several “rescue mutations” [Liu et al., 1995], whereas the other report merely mentions improper receptor processing [Kim et al., 1995]. Alternately, two reports suggest involvement of this position in ligand binding [Funk et al., 1993; Erb et al., 1995]. This position is therefore critical in more than one way, and may reflect minor differences in receptor structure between various classes of GPCRs. The high incidence of acidic (12% vs. 3.2% on all TMs) and basic (8% vs. 5.5% on all TMs) amino acids favours an orientation where this position is exposed to the central cavity of the helical bundle, but amino acids capable of hydrogen bond formation (11% S, 9% N, 7% Q, 7% H, 3% T, and 3% Y) could be accommodated in both a solvent-exposed orientation, and in an interhelical contact region. It is, therefore, likely that the size of the side chain plays a role in the decision process. Smaller residues, such as occur in the muscarinic receptor, would be important for interhelical contacts, while on the contrary larger residues extend into the central cavity.

The presence of lipophilic residues in 91% of all GPCRs rightfully marks the orientation of position 7.37 towards the lipid environment. The next position, 7.38, differing by 100° from 7.37, and 200° from 7.36, is probably involved in interhelical contacts with TM6. This is consistent with the presence of the structurally important T (22%), F (19%), G (8%), and P (4%) residues at this position. However, site-directed mutagenesis of the NK1 and NK2 receptors suggests that this position is important in determining species and subtype selectivity of agonists and antagonists [Fong et al., 1992a; Sachais et al., 1993; Bhogal et al., 1994; Jensen et al., 1994]. Although it is quite possible that changes in affinity result from changes in overall receptor structure, it can not be excluded that position 7.38 is solvent-exposed based on the current dataset [Fong et al., 1992a; Sachais et al., 1993; Bhogal et al., 1994; Jensen et al., 1994].

There is ample evidence that position 7.39 is exposed to the central cavity of the helical bundle. The presence of D, E, H, N, Q, and R residues renders it unlikely that this position is not solvent-exposed. Mutational analysis of several GPCRs from divergent subfamilies recorded profound effects on agonist and antagonist affinity (Table 3, Fig. 3). In addition, improper receptor processing and constitutive activity were observed. Especially noteworthy is the observation that an A(7.39)D mutation in bovine rhodopsin was tolerated and behaved like the wild type rhodopsin [Ridge et al., 1992], but that the A(7.39)E mutation in human rhodopsin lead to constitutive activity [Dryja et al., 1993]. The E residue, which differs from the D residue by only one methylene group in the side chain, is apparently sufficient to shift the receptor from the inactive to the (pre-) active state. In the TRH receptor the, slightly smaller, homologous mutation R(7.39)K was tolerated, but substitution with other residues greatly decreased the affinity for an agonist ligand [Perlman et al., 1995a].

Some controversy exists on the orientation of position 7.40. Despite the presence of a large majority (81%) of lipophilic residues (A, F, I, L, M, V, and W, with W in 40% of all cases) this position is also occupied by R residues in the arachidonate-derived lipid mediator receptors. Following the rules of helical periodicity, this ligand should be exposed to the lipid membrane environment, which is consistent with the high incidence of lipophilic residues, but clearly R residues can not easily be accommodated by such an environment. Moreover, it was demonstrated that it is possible to photochemically label the W(7.40) residue present in the β₂ adrenoceptor [Wong et al., 1988]. However, there is one interpretation that might afford an explanation for this apparent paradox. When the Orwellian notion that “all animals are equal, but some animals are more equal than others” [Orwell, 1945] is applied to GPCRs, it follows that the extent and positioning of TM7 may vary amongst GPCRs, regardless of homology and conservation pattern. Whereas some GPCRs may have a longer TM7 segment initiated at or around position 7.31, other receptors may consist of a smaller TM7 segment that is located with position 7.40 at about 1 helical turn from the solvent-lipid interface. Considering the length of the side chain of R, a “depth” of 1 helical turn into the membrane is sufficient for solvent-exposure of the guanidinium group. Modification of the side chain would thus affect ligand binding through optimization of the positioning of the TM with regard to the solvent-lipid interface. The R(7.40)K mutant, with a somewhat shorter but isoelectric side chain, exposes the ligand binding domain more (bringing it closer to the surface), resulting in increased affinity [Huang and Tai, 1995], than mutations abolishing the positive charge of the side chain. The R(7.40)E, V mutants in the PGE₂-EP3 receptor and the R(7.40)Q mutant TXA₂ receptor, indeed, exhibit a lower binding affinity [Funk et al., 1993; Huang and Tai, 1995]. Since this position is proximal to the solvent-lipid interface, it would also account for the photoaffinity labeling in the β₂-adrenoceptor [Wong et al., 1988], and the lack of effect observed in the m3 muscarinic receptor [Wess et al., 1993].

The location of positions 7.42 and 7.43 facing the central cavity of the helical bundle is consistent with the effects observed in site-directed mutagenesis studies. Position 7.43 is occupied by a K in the (rhod)opsins that serves as the anchoring point of the retinal chromophore by means of Schiff base formation. Destabilization of the rhodopsin protein by mutating K(7.43) to either A, E, G, or H results in constitutive activity of the receptor [Cohen et al., 1992; Robinson et al., 1992; Li et al., 1995]. Experimental data from targeted mutagenesis of the H conserved at this position among all adenosine receptors [Olah et al., 1992; Kim et al., 1995], or the Y conserved (48% of all GPCRs) among most biogenic amine receptors [Strader et al., 1989a; Wess et al., 1991; Wess et al., 1992b], demonstrate that this position is also a prominent anchoring site for ligands that are not covalently bound to the protein.

W(7.44) occurs only in some lipid mediator receptors, and according to the requirements of helical periodicity, is located at the same side of the helix as R(7.40) in those receptors. The interaction between a positively charged residue and aromatic residues is thought to be essential for ligand binding to the biogenic amine receptors [e.g., Hibert et al., 1991], and may play a role in these receptors as well. Replacing W(7.44) with an aliphatic residue such as L, or a charged residue such as R, proved detrimental for ligand affinity [Funk et al., 1993].

Residue 7.45 (76% N, 10% S, and 8% H) is located at the same face of the helix as position 7.38, but two helical turns closer to the intracellular surface. It thus exhibits a similar profile as position 7.38. The two site-directed mutagenesis studies available, suggest that this position distinguishes between agonist and antagonist binding, and a direct role in ligand binding is therefore plausible [Strader et al., 1987a; Perlman 1995b]. The abundance of hydrophilic residues at position 7.46 (94% of all GPCRs) clearly locates this position in the central cavity of the helical bundle. Considering that 70% of all GPCRs contain a S at this position, 11% a C, and another 7% a T, it is very likely that this position assists in maintaining the backbone structure of the TM. Furthermore, it is located 1 helical turn above the conserved P at position 7.50, and could take over the hydrogen bonding network disturbed by the presence of this proline. Mutation studies of this position reveal intricate effects on agonist and antagonist binding, that might be explained either by direct interaction with the ligand or by more subtle regulation of the shape of the ligand binding site through intrahelical contacts [Strader et al., 1989a; Jiang et al., 1995; Kim et al., 1995].

N(7.49) in the NPXXY conservation motif has been considered one of the better conserved residues in TM7. Indeed, 82% of all GPCRs, including all biogenic amine receptors, maintain an N at this position, but another 17% of all GPCRs express an D at the equivalent position. Furthermore, at least 1 GPCR has a K(7.49) residue, although in this latter case it concerns the uncharacterized orphan receptor RTA [Ross et al., 1990]. It has been hypothesized that in the case of the GRH receptor this constitutes a reciprocal mutation with N(2.50) [Zhou et al., 1994], but all other GPCRs express an D(2.50). The position is apparently solvent-exposed, but its function is not clear. Moreover, the one mutation reported D(7.49)N in the GRH receptor was without effect [Davidson et al., 1994b].

Position 7.50 is characterized by an incidence of 100% P. It therefore serves a function in receptor structure, that not only affects ligand binding to the receptor, but also modulates receptor activity. The effect on receptor structure of a P(7.50)S mutation in the β₂-adrenoceptor is so drastic that the receptor cannot attain proper folding [Strader et al., 1987a].

Beyond the conserved P(7.50) very few experimental data are available. One study employing site-directed mutagenesis of C(7.54) in the β₂-adrenoceptor indicates the involvement, either directly or indirectly, of this residue in agonist binding and receptor activity [O’Dowd et al., 1988], but this finding is disputed by others [Fraser et al., 1989a].

Receptor Models

Many GPCR models [e.g., IJzerman et al., 1994; Teeter et al., 1994; Prusis et al., 1995; ter Laak et al., 1995] have been published founded on the work initiated by Hibert et al. [1991], and based on a bacteriorhodopsin template [Henderson et al., 1990]. More recently, the rhodopsin template [Schertler et al., 1993] has been used for the same purpose. The use of such models has greatly aided in the understanding of receptor structure and the interpretation of mutation analyses [Donnelly and Findlay, 1994; Perlman et al., 1994b; Zhou et al., 1994; Kim et al., 1995]. The construction and application of such models was recently discussed by Ballesteros and Weinstein [1995]. We also took it upon ourselves to devise a model for the recently cloned P_2Y purinergic receptor, and are now preparing to investigate the implications of this model, and the survey presented above, by means of site-directed mutagenesis [van Rhee et al., 1995].