Proc. Nati. Acad. Sci. USA
Vol. 88, pp. 8111-8115, September 1991 Neurobiology
Molecular dynamics of dopamine at the D2 receptor (guanine nucleotide-binding regulatory protein-coupled receptors/three-dimensional structure/electrostatic potentials! molecular modeling)
SVEIN G. DAHL*, 0YVIND EDVARDSEN, AND INGEBRIGT SYLTE Department of Pharmacology, Institute of Medical Biology, University of Troms0, N-9001 Troms0, Norway
Communicated by Jean-Marie Lehn, May 20, 1991 (received for review January 28, 1991)
Cloning and sequencing of neurotransmitter receptors has demonstrated that they may be divided into various "superfamilies." The dopamine D2 receptor belongs to the superfamily of receptors transferring signals into cells through guanine nucleotide-binding regulatory (G) proteins. Antagonism of postsynaptic D2 receptors has long been regarded as the primary mechanism of action of antipsychotic drugs, and it has recently been proposed that also dopamine D1 (1), D3 (2), and D4 (3) receptors and serotonin 5-HT2 receptors (4) may be involved in antipsychotic drug action. Knowledge of the three-dimensional structure of such receptors would significantly add to our understanding of their molecular mechanisms and be useful in the search for drugs that affect these receptors. Amino acid sequences of more than 10,000 proteins are known, but only some 450 three-dimensional protein structures have been reported. No detailed three-dimensional crystal structure of a neurotransmitter receptor molecule is yet available, to our knowledge. In the absence of such data, we have constructed a model of the D2 receptor and used it to simulate the molecular dynamics of dopamine-receptor interactions. Based on structural similarities within the superfamily of G-protein-coupled neuroreceptors, the model was constructed from the rat D2 receptor sequence (5) and refined by molecular dynamics simulations and molecular mechanics energy minimization. The modeling was based on the following five hypotheses:
(i) Membrane-spanning domains of G-protein-coupled neurotransmitter receptors are a-helices. Hydropathy indices along the peptide chains show that G-protein-coupled receptors contain seven hydrophobic domains, usually assumed to correspond to seven transmembrane stretches of residues (6). Electron diffraction (7) and electron microscopy (8) experiments have demonstrated that bacteriorhodopsin has seven a-helical segments spanning the cell membrane. All available biological and chemical data indicate also that the membrane-spanning domains of visual rhodopsin form a-helices (9, 10). G-protein-coupled neurotransmitter receptors share some amino acid homology with visual rhodopsin, particularly in the seven hydrophobic domains (11) that most likely are a-helices. (ii) Each transmembrane helix contains 27 amino acids. An a-helix with 27 amino acids has a total length of =40 A and may traverse the cell membrane in an approximately perpendicular orientation. The 11 membrane-spanning a-helices in the crystal structure of the photosynthetic reaction center of Rhodopseudomonas viridis contain on average 27 amino acids (12). (iii) Membrane-spanning segments of G-protein-coupled neurotransmitter receptors have similar localizations in aligned protein sequences. Hydropathy indices provide only approximate locations of transmembrane segments in the peptide chain of a protein, and for one and the same receptor such locations often differ by several amino acids among different reports. Since G-protein-coupled neurotransmitter receptors show 50-75% sequence homology in the putative transmembrane domains, we assumed that the segments of all receptors corresponding to each a-helix would be placed in matching positions by sequence alignment, as indicated in Fig. 1. Presumably, this enabled more precise prediction of the positions of the transmembrane helices in the peptide chains, based on average hydropathy indices of 14 neuroreceptors from the same superfamily. (iv) The most polar surface areas of the transmembrane helices form a central core. The seven a-helical segments in bacteriorhodopsin are roughly perpendicular to the cell membrane, slightly inclined to one another at various angles up to 200, and closely packed to form an oval ring structure (7, 8). Site-directed mutagenesis experiments have suggested that ligands bind to the putative transmembrane regions of 132adrenergic (12, 28, 29) and muscarinic ml receptors (30), and it has been postulated that f-adrenergic receptors form a rhodopsin-like core containing a ligand binding site (28). The similarities in their primary structures indicate that this also may be the case for other G-protein-coupled neuroreceptors. (v) G-protein-coupled neurotransmitter receptors have a common ligand-binding site. Site-directed mutagenesis experiments have suggested that agonists and antagonists interact with Asp-113 and that agonists but not antagonists
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Abbreviation: G protein, guanine nucleotide-binding regulatory protein. *To whom reprint requests should be addressed.
A three-dimensional model of the dopamine ABSTRACT D2 receptor, assumed to be a target of antipsychotic drug action, was constructed from its amino acid sequence. The model was based on structural similarities within the superfamily of guanine nucleotide-binding regulatory (G) proteincoupled neuroreceptors and has seven a-helical transmembrane segments that form a central core with a putative ligand-binding site. The space between two residues postulated to be involved in agonist binding, Asp-80 and Asn-390, perfectly accommodated an anti-dopamine molecule. Molecular electrostatic potentials were mainly negative on the synaptic side of the receptor model and around aspartate residues lining the central core and positive in the cytoplasmic domains. The docking of dopamine into a postulated binding site was examined by molecular dynamics simulation. The protonated amino group became oriented toward negatively charged aspartate residues in helix 2 and helix 3, whereas the dopamine molecule fluctuated rapidly between different anti and gauche conformations during the simulation. The receptor model suggests that protonated ligands are attracted to the binding site by electrostatic forces and that protonated agonists may induce conformational changes in the receptor, leading to G-protein activation, by increasing the electrostatic potentials near Asp80.
8111
8112 r)i 02 03 51) a 5HT1c 5HT2 a2a a2h
32 mI
12
14
i'- 'Vi-
Neurobiology: Dahl et al.
Proc. Natl. Acad. Sci. USA 88 (1991)
(9) 0GTGLVVERDFSVI kILTACFLSLLII.STLLGNTLVCAAVIRFRHLRSKVTKNFFVISLAVSDLIVAVI.VMPWKAVAEIAC;F.WPFGSF.GCNIWVAFDIMCSTASILNLCVISVDRYWA I'-;. ; f20) PFNGSECKADRPPHYNYYAMILLTLLIF IIVFQIVLVCMAVSREKALT TTNYLIVSLAVADLLVATLVMPWVVYLEVVGE VKFSRI HCDI FVTl.DVMMCTASI LNLCAIS I DRYTAVA TTNYLrVVSLAVADLl.VATLVMPWVVYLFEVTGGVINFSRICCOVFVTTLDVMMCTASI I .NLC'AISIDRYTAVV' (2.8) ...... SDVTVSYQVITSLLLCTLIFCAVL(:"ACVAAIALERSLQN.VANYLIGSLAVTDLMVSVI VLPMAALYQVl,N.KIITLGQVTCDLFIALDVLCCTSS1LFLCAIALDRYWAII .';4f' (45) LF ..... QFPDGVONWPALSIVVIIIMTIGGNILVIMAVSMEKKLHN.AT YFLMSLAIADMLVGLLVMPLSLLAILYDYY"PLPRYLCPYWISLDVLFSTASIMHLCAISLDRYVAI 15;) (60) LPPTCLSILHLQEKNWSAI1LTTWVIILTIAQ#ILVIMAVSLEKKLQN.ATtYFLMS1JIADM 1lIGFI,VMPVSMLTILYGYRNPLPSKLCAIWIYLlVLFSTASIMH ZAISLDRYVA T' (37) P ...... QLDVTRAISVGLVLCAFI4LFAIVNILVILSVACNRHLRT.PTNYFIVNLAIADLLLS'TVLPFSATLEVLGY.WVLGRIFCDIWAAVDVLCCTASILSLCAISIDRY1 .V\z- ..I.,
(1i) GAENSTrGVNRARPHAYYALSYCALILAI IFGNGLVCAAVLRERALT
(19) (37) (44 f26) (I1) (12) (18)
t:CGARAT PYSLQVTLTLVCLACLLMLLTV-FVLVIIAYFTSRAXA.PQNLFLVSLASADILVATLVIPFSLArEVMGY.IYF:KTWC-EIYI.A DVLFCTSSIVHLCAISLDRYWSI ,I SWGPPRGQYSAGAVAGLAAVVGFLIVFTV'VGNVLVVIAVYlTSRALRA. PQNLFLVSLASADI LVATLVMPFSl.ANELMAY. WYFGQVWCGVYLALDVILFCTSSIVHLCAISIDRYWSVrr iPASESPEPLSQQWTAGMGLLMALIVLLIVAGNVLVIVAIAKTPR[WT. LTNLFIMSLASADLVMGLLVVPFGAT IVVWG. RWEYGSFFCELWTSVDVLCVTASIETLCVIADRYI.AI A :. QRDEVWVVGMGIVMSLIVLAIVFGNVLVITAIAKFERLT VTNYFITSLACADLVMGIlAVVPFGAAHi LM KMWTFGNFWCEFWTSI DVLCVTASI ETLCVIAVDRYFA IT -: NITVLAPGKGPWQVAFIGITTCLLSLATVTONLLVLISFKVNTELKT.VNNYFLLSLACADLIIGTFSMNLYTTYI1LMGH. ALGTLACDLWLALDYVASNASVMNLLLISEDRYFSV 1;, ALTSPP ... YKT2FEWFIVLVAGSLSLVTIIGNILYMVSIKVNRHLT.VNNYFLFSLACADLIIGVFSMNLYTLYTYIGY.NPLGPVVCDI.WLALDYVVSNASVMNLLIISFDRYF-C V i. VTSSSHNRYETVEMVFIATVTGSLSLVTVVGNI IVMISIKVNRQLCT KY PLGAVVCDLWIALDYVVS NASVMNLI. I I S FDRYF VirF W. VNNYFLFSLACADI.T1 GAFSMNLYTVY K1.
r-------
IV --
.-...
DI1 P FRY. . ERKMT PK .AA FXL ISVAWT LSVL IS F IPVQ LSWHKAK PTS PS Ic)GNATS LA E T TD NCDS SLS RTYA ISSSsv I SFY IPVAI M IVTY T R IY RIAQKQ I EkI AA L E AAVwH A K N C- O 1 D2 PMLY. .NTRYSSKRRVTVMIAIVWSVLSFTISCPI,.LFGLNNTDQNE ....C.. IIANPAFVVYSSIVSFYVPFIVTI.LVYIKIYTVLRK ...... RRKPVNTKRSSRAFRH 03 PVHYQHiGTGOSSCRRVALMITAVWVLAFAVSCPL. LFGFNTTGDPS .............. ICSISNPDFVIYSSVVSFYVPFGVTVIVYARIYIVLRQ.....RQRKRI L1'RONS,0QC:S 5HTIa PIDY.VN. . .KRTPRPRALISLTVLIGFLISIP.PMLGWRTPEDRSDPD ........... ACTISKDHGYTIYSTFGAFYIPLLLMLVLYGRIFRAARFRIRKTVKKVFKT A.
5i1rl1c
514'2 cl a2a
a2b
p1
P2
Ml M2 M4
P... IEHSRIF'NSRTKAIMKIAIVVAISIGVSVPIPVIGLRD.E.. .S.. KVFVNNTTCVL...-.NDPNFVLIGSFVAFFIPLTIMVITYFLTIYVLRRQTLMLLR .............. P ... IHHSRFNSRTKAFLKIIAVWTISVGISMPIPVFGI.Q ...... DDSKVFKEGS.CI 1L.....ADDNFVLIGSFVAFFIPLTIMVITYFLTIKSLOKEATLCVSDLSTRAKlASTS)E I). SI.QYPT... LVTRRKAILALLSVJVLSTVISIG. PLLGWKEPAPNDDKE ............ CGVTEEPFYALFSSLGSCSYIPLAVILVMYCRVYIVAKRTT . AIEY ..NLKRTPRRIKAI. IITCWVISAVISFPPLISIEKKGGGGGPOP ........ AEPRCEINDQKWYVISSCIGSFFAPCLIMILVYVRIYQIAKRRTRVPPSRHRPDAVAAIiPPC;(;IT AVEY. .NLK.RTPRRVKATIVAVWLISAVISFPPLVSLYRQPDG ............ AAYPQCGLNDETWYILSSCIGSFFAPCLIMGLVYARIYRVAKRRTRTLISEKRAPVGPI.,)(;,ASPPT I' PFRYE . QS. LLTRARARGLVCTVWAISAIVSFLPI LMHWW. RAESDEARRCY . .. NDPKCCDFVTNRAYAIASSVVSFYVPLCIMAFVyLRVFREAQK . PFKY .. QS. LLTKNKARVII LMVWIVSGLTSF. LPIQMHWYRATHQEAI NCY .... ANETCCDFFTNQAYAIASSIVSFYVPLVIMVFVYSRVFQEAKR ..................... PLSY .. RAKRTPRRAALM. IGLAWLVSFVLWAPAILFWQYI.VGERTVI . ....... AGQCYIQFlSQPI ITFGTAMAAFYLPVTVMCTLYWRIYRETENRARE. AALQGSE'1 i'(-;K PLTY. . PVKRTTKMAGMM. IAAAWVLSFI LWAPAILFWQFIVGVRTVEDGE ........ CYIQFFiSNAAVTFGTAIAAFYLPVI IMTVLYWHISRASKSRIKKDKKEPVANODPVSPSLV PLTY ..PARRTTKMAGLM. IAAAWVLSFVLWAPAILFWQFVVGKRTVPDNII ........ CFIQF..SNPAVTFGTAIAAFYLPVVIMTVLYIHISlASRSRVHKHRPECIPKEKKAKTI .AF'I. ,VI
D1 .... SQPESSFKMSFK....... RITKVLKTLSVIMGVFVCCWLPFFILNCILPFCGSGETQPFCIDSNTFDVFVWFGWANSSLNPI IY. AF'NADFRKAFSTLWC;YPIII -A 02 F.. 'EIQTMPNGKTRTSLK. TMSRRKL. SQQKBKKATQMLAIVLGVFI ICWLPFFITHILNIHCDCN ...... I PPVLYSAFTWLGYVNSAVNPI IYTTFNI EFRKAFMK ' HC D3 ... LEVRKLSNGRLSTSRIGR.PLQPR. GVPLRBKKATQKVVIVLGAFIVCWLPFFLTHVLNTHCQA.....CEYVSPELYRATTWLGYVNSALNPVIYTTFNVEFRKAFI.K I.SC SHTl a LAR .BRKTVKTLGI IMGTFILCWLPFFIVALVLPFCESSC ... HMPTLLGAI INNLGYSNSLLNPVIYAYFNKDFQNAFKK II KCNFPCRU INN .................... KKASKVLGIVFFVFLIMWCPFF1TNILSVLCGKACNO .. KLMEKLLNVFVWIGYVCSGI W LVYTLFNKIYRRAFSKYI.pRCDY KP K 5HT2 . SISNSQKACKVLGIVFFLFVVMVCPFFITNIMAVICKESCNE. CLi 1 68 NVIGAI1LNVFVWIGYLSSAVIPl.VYTLFNKTYRSAFSRY I cal FSR ............................. EKKAKTLGIVVGMFILCWLPFFIALPLGSIFSTL.....KPPDAVFKVVFVLGYFNSCLMPI IYPCSSKEFKRAFMRI LGCOCHS;R 144 a2a RGRRGSGRRLQGRGRSASGLPRRRAGAGGQNLIKRFTFVLAVVIGVyVIVCNFPFFFTYTLTAVGCS. VPRTLFKFFFWFGYCNSSLNPVIYTIFNHDFRRAFKKIL -.. CPON .4; a2b ASSRSVEFFLSRRRRARSSVCRRK, .. VAQARXKRFTFVLAVVMGV VLCWFPFFFIYSLYGICREAC.Q.. VPGPLFKFFFWIGYCNSSLNPVIYTVFNQDFRPSFKHI LF-NHNNRCRF-; ',2 LAN .............. GRAGKRRPSRLVALRBQKALKTLGIIMGVFTLCVLPFFI.ANVVKAFHREL ...... VPDRLFVFFNWLGYANSAFNPIIY.CRSPIDFRKAFG.I CI 2 P-AP 32 ....................... RRSSKFCLKEHKALKTLGIIMGTFTLCWLPFFIVNIVHVIQDNL ...... IRKEVYILLNWIGYVNSGFNPLIY.CRSPDFRIAFOELL..2 Ml KGQKPRGKEQLAKRKTFSLVK ... ........ HKKAARTLSAI LLAFIVTWTPYNIMVLVSTFCKD. ....VPETLWELGYWLCYVNSTINPMCYALCNKAFRDTFRLLLLL R ..... . .'4 M2 SGQNGDEKQNIVARKIVKMTKQPAKKKPPPSRIEKKVTRTI LAILLAFI ITVAPYNVMVLINTFCAP . C ..... IPNTVWTIGYWLCYINSTINPACYALCNATFKKTFKHLLMCiYKNIOC; M4 TPAGMRPAANVARKFASIARNOVRKKRQMAARERKVTRTIFAI LLAFI LTWTPYNVMVlVNTFCQS . C...TPDTVWSIGYWLCYVNST TNPACYALCNATFKKTFNHII,LPQ. C )YRN' T .....................
OCOYKE.NR
FIG. 1. Aligned amino acid sequences of G-protein-coupled neurotransmitter receptors. Dl, human dopamine D1 receptor (13-15); D2 and D3, rat dopamine D2 (5) and D3 (2) receptors, respectively; 5-HTla, human 5-HT1a receptor (16, 17); 5-HTlc, rat 5-HT1, receptor (18); 5-HT2, rat 5-HT2 receptor (19); alb, hamster alb-adrenergic receptor (20); a2a, human au-adrenergic receptor (21); a2b, human a2b-adrenergic receptor (22); P1, human fi-adrenergic receptor (23); ,2, human f32-adrenergic receptor (24); Ml, porcine muscarinic ml receptor (25); M2, porcine muscarinic m2 receptor (26); M4, human muscarinic m4 receptor (27).
interact with Asp-79, Asp-130, and Asn-318 (29, 31) and Ser-204 and Ser-207 (32) in the (82 receptor. The corresponding aspartic residues also seem to be involved in ligand binding to muscarinic ml receptors (30). Asp-79, Asp-113, and Asp-130 in the ,2 receptor are conserved in all sequences shown in Fig. 1 and in the dopamine D4 (4) and D5 (33) receptors, which suggests that these residues are involved in ligand binding and signal transduction in all G-proteincoupled neuroreceptors. Asn-318 in the P2 receptor is conserved in the dopamine receptors, muscarinic acetylcholine receptors, and adrenergic receptors. Ser-204 and Ser-207 in helix 5 of the p2 receptor are conserved in the dopamine receptors. It is possible that the corresponding residues (Asn-390, Ser-194, and Ser-197) also are involved in agonist binding to D2 receptors. The D2 sequence, like all the other sequences shown in Fig. 1, contains prolines in transmembrane segments IV-VII. Proline often appears at the end of a-helices in proteins and is known as a "helix breaker." However, the M subunit of the photosynthetic reaction center of R. viridis has a proline near the middle of the third transmembrane helix (12), which demonstrates that a proline does not necessarily represent a breaking point of a transmembrane a-helix.
METHODS
Sequence AlignMent The sequence of the rat D2 receptor (5) and the sequences of 13 other G-protein-coupled neurotransmitter receptors were aligned by the NeedlemanWunsch method (34), with the Gap program of the University of Wisconsin Genetics Computer Group Sequence Analysis Software Package (35). With a gap weight of 3.0, and a length weight of 0.1, hydropathy indices were calculated for each sequence by two methods (36, 37), and average indices were
calculated from the aligned sequences. The 14 sequences were selected to get a balanced contribution from various receptor families to the average indices. Receptor Modeling. Structural refinement by molecular mechanics energy minimization and molecular dynamics simulations were done on a Cray X/MP-28 supercomputer, using the AMBER all-atom force field (38). The cut-off distance for nonbonded interactions was 8 A. A distance-dependent dielectric function was used to mimic the dielectric damping effect of water on intramolecular interactions. Energy refinement was done by 500 cycles of steepest-descent minimization followed by 2000 cycles of full-conjugate gradient minimization. Molecular dynamics simulations were performed at 310 K with a step length of 0.001 psec, after an initial equilibrium dynamics phase starting from 0.1 K. Molecular electrostatic potentials 1.4 A outside the water-accessible surface were calculated with a distance-dependent dielectric function and a 10.0-A nonbonded cut-off radius. Initial models of each transmembrane a-helix, including the side chains, were constructed with the MIDAS computer graphics programs (39) from the rat D2 sequence (5). Each helix was refined by energy minimization and its wateraccessible surface (40) and molecular electrostatic potentials (39) were calculated. The helices were assembled on the computer graphics system such that the most polar region of each helix surface was oriented toward a central core. The helices were closely packed in an antiparallel arrangement, clockwise as viewed from outside the cell membrane, such that the water-accessible surface contained no holes between helices after structure refinement. The seven-helix structure, without connecting loops, was refined by energy minimization. The terminal segments and loops between helices were added, based on a method (41) that gives s50l correct predictions of secondary structures
Neurobiology: Dahl et al.
Proc. Natl. Acad. Sci. USA 88 (1991)
8113
in proteins (42). Although other methods may give more correct secondary structure predictions (43-45), we did not have any adequate method for modeling the 131-amino acid cytoplasmic loop between helix 5 and helix 6. Therefore, only 7 amino acids of this loop, taken from the segments connected to helix 5 and helix 6, were included in the model and linked to each other. Thus the segment of the third cytoplasmic loop that differs between the two D2 receptor subtypes (5, 46-48) was not included in the model. The receptor model was refined by molecular mechanics energy minimization. A dopamine molecule was then placed in the putative binding site, where it fit precisely into the space between Asp-80 and Asn-390. The receptor-ligand complex was further refined by energy minimization, which produced only negligible changes in the receptor structure. Because the initial modeling of the extracellular and cytoplasmic domains had been based on fairly inaccurate secondary structure predictions, these parts of the model were further refined by 2000 steps of molecular dynamics simulation at temperatures increasing from 0.1 K to 300 K, followed by energy minimization, while the transmembrane helices were kept fixed. This produced minor changes in the loops between helices and more noticeable conformational changes in the N- and C-terminal segments. Finally, the wateraccessible surface and molecular electrostatic potentials were calculated for the whole receptor model.
RESULTS Prediction of Transmembrane Domains. Average hydropathy indices of the 14 aligned receptor sequences are shown in Fig. 2. Plotting of areas under "windows" of 27 amino acids under the average hydropathy index curves as a function of amino acid number produced peaks that predicted the starting point of each transmembrane helix. Hydropathy indices calculated by the method of Kyte and Doolittle (36) gave well-defined peaks for helices 1-6, whereas indices calculated by the method of Hopp and Woods (37) defined the starting points of helices 1, 4, 5, 6, and 7. Three-Dimensional Receptor Structure. The D2 receptor model is shown in Fig. 3. After energy refinement, the transmembrane a-helices remained approximately antiparallel, were slightly bent, and were somewhat closer together at the cytoplasmic end than at the synaptic end. Two of the residues postulated to be involved in agonist binding, Asp-80 in helix 2 and Asn-390 in helix 7, were placed adjacent to each C
> 30 c6x
600
O_ D 20 C._ I'r 10
400-
C
Cu5 (D
2
1 0. x _
WE
D a
I
n0
~hIZ~ihiiB I1*ww 11lr a X.WXo
n
0
200
400
zuu
200
30
400
600
600
D
10- \JM\AI'~~~~~~~~~~~~\~ 1-1
,
15 kcal/mol.
other at a distance that closely accommodated the energy minimized anti conformation of dopamine. Asp-114 in helix 3 and Ser-194 and Ser-197 in helix 5 were placed nearer to the synaptic membrane surface than Asp-80 and Asn-390. Presumably, modeling of the other dopamine receptors by the same procedure would place the corresponding residues in similar geometrical positions. Electrostatic Potentials. The molecular electrostatic potentials were based on the charges of all surrounding atoms within the defined cut-off radius and depended on molecular conformation and on atomic charges. The potentials were mainly negative in the putative extracellular domains and in parts of the central core of the receptor model and positive in the postulated cytoplasmic domains (Fig. 3). The lowest electrostatic potentials in the central core were -40 kcal/mol (1 cal = 4.184 J) near Asp-80 in helix 2 and -37 kcal/mol near Asp-114 in helix 3. Strongly positive electrostatic potentials, up to 38 kcal/mol, were also found in the area between helix 4 and helix 5, extending from a level one helical turn below Asp-80 and down into the second cytoplasmic loop. Molecular Conformations and Dynamics of Dopamine. Internal movements in molecules occur on a fsec time scale. The molecular dynamics of protonated dopamine approaching the postulated receptor binding site was examined in an 80-psec simulation where the receptor was kept in a fixed position. The coordinates and energies were saved at 0.5psec intervals. The simulation required 200 min of centralprocessing-unit time on the Cray computer. The electrostatic forces were not sufficient to attract the dopamine molecule to the postulated binding site during the simulation. Weak potentials (1.0 kcal), directing the neuro-
8114
Neurobiology: Dahl et al.
transmitter toward the area between Asp-80 and Asn-390, therefore, were applied during the initial 65 psec of the simulation. Constrained atomic distances were as follows: Dopamine nitrogen to Asp-80 carboxyl oxygen, 0-25 psec at 25 A, 25-55 psec at 10 A, and 55-65 psec at 4.5 A; dopamine para-oxygen to Asn-390 carbonyl oxygen, 40-55 psec at 10 A and 55-65 psec at 4.5 A. The side chain of the dopamine molecule moved several times from one side of the phenyl ring plane to the other side and fluctuated rapidly between various anti and gauche conformations throughout the simulation. Fig. 4 shows a part of the receptor model and some of the dopamine structures that were observed during the simulation. As the neurotransmitter moved down the central core, its positively charged amino group became oriented toward negatively charged aspartate residues (Asp-80 and Asp-114) in helices 2 and 3. The amino group of dopamine remained close to Asp-80 during the final 15 psec of the simulation, after removal of the directing force. Fig. 4 Lower shows a dopamine molecule in a postulated binding site between Asp-80 in helix 2 and Asn-390 in helix 7.
DISCUSSION The negative molecular electrostatic potentials around the N-terminal segment and the loops between helices 2 and 3, helices 4 and 5, and helices 6 and 7 support the hypothesis (49) that these represent synaptic domains of the receptor. The largely negative electrostatic potentials at the synaptic side
FIG. 4. Docking of dopamine into a postulated D2 receptor binding site. (Upper) Selected structures from an 80-psec molecular dynamics simulation of dopamine approaching the binding site. A part of helices 2, 3, 6, and 7 are shown with their connecting loops. (Lower) Energy-minimized anti conformation of dopamine between Asp-80 (blue) and Asn-390 (green). Water-accessible molecular surfaces (dotted) are color coded according to electrostatic potentials as in Fig. 3.
Proc. Nati. Acad. Sci. USA 88 (1991)
indicate that protonated ligands are attracted to the receptor by electrostatic forces. f32-Adrenergic receptors are coupled to stimulatory G (Gj) proteins in 11- to 15-residue segments located near the cytoplasmic surface of the cell membrane, at each end of the third intracellular loop, and at the N-terminal part of the cytoplasmic tail (6). The cytoplasmic domains in the D2 receptor model with strong positive electrostatic potentials (Fig. 3) closely corresponded to the parts of the /32 receptor that are coupled to a G, protein. This suggests that positive electrostatic potentials around certain cytoplasmic domains of the D2 receptor may be involved in G-protein interaction. Our calculations also support, by a completely different approach, the "positive-inside" rule for membrane proteins, which has been suggested from site-directed mutagenesis experiments (50). It is interesting to note that the electrostatic potentials were strongly positive between helix 4 and helix 5, only 10 A from Asp-80, which was surrounded by an equally strong negative electrostatic field. This indicates that agonist-induced changes in the electrostatic field may play a role in signal transduction in the D2 receptor, as suggested for G-proteincoupled receptors in general (51). Such a process could be initiated by binding of a protonated agonist to Asp-80 to neutralize its negative electrostatic field. Energy-minimized anti-dopamine fit perfectly into the space between Asp-80 and Asn-390 (Fig. 4), in accordance with the hypothesis that this is the active conformation (52). However, our molecular dynamics simulations indicate that, rather than obtaining a complete lock-and-key fit to the active site while the neurotransmitter stays in this conformation, the interaction starts with an electrostatic attraction between the protonated amino group on the ligand and negatively charged aspartate residues at the binding site, while the neurotransmitter moves between various conformations (Fig. 4). Such a "zipper" mechanism whereby a flexible ligand binds to a macromolecule in several successive steps has been postulated from thermodynamic considerations (53). Still, the prevailing concept of ligand-receptor interactions has been a lock-and-key mechanism requiring a specific conformation of the ligand. The present molecular dynamics simulation supports the "zipper" type of mechanism for dopaminereceptor interactions. Fig. 5 shows the antipsychotic drug cis-(Z)-chlorprothixene with its protonated dimethylamino group in close contact with Asp-114 in helix 3. The localization of Asp-114 closer to the synaptic membrane surface than Asp-80 and Asn-390 suggests that neuroleptics act by preventing dopamine access to its endogenous binding site and offers a simple steric explanation of how Asp-80 and Asn-390 may interact with
FIG. 5. Molecular structure of cis-(Z)-chlorprothixene in a postulated D2 receptor binding site, viewed from the synaptic side. Helix 1 is at the bottom of the figure.
Neurobiology: Dahl et al. agonists but not with antagonists. The dynamics of the ligand-receptor system would still enable competitive binding of agonists and antagonists. The calculations presented here were intended to present an approximate overall model of the D2 receptor, which might provide further insight into its mechanisms. The model does not rule out other possible arrangements of the various parts of the receptor. It is possible that the seven helices may be more tilted relative to each other, that the central core is slightly narrower, and that the seven helices are arranged in an anti-clockwise order viewed from outside the cell, as suggested for bacteriorhodopsin (8), and it is quite likely that the number of residues in each helix may differ from 27. In any case, our calculations clearly indicate that, to understand the molecular mechanisms, neurotransmitterreceptor interactions should be regarded as dynamic processes and that electrostatic mechanisms may be important for ligand binding and signal transduction in the receptor. The receptor model may be used to design protein-engineering experiments to test its validity and provide further insight into receptor mechanisms. It would be interesting, for instance, to examine the relevance of the positively charged area between helix 4 and helix 5 for transducer mechanisms, by site-directed mutagenesis experiments. The model also provides a tool to simulate the molecular dynamics of ligand-receptor interactions. Although probably inaccurate in many details, we feel that the receptor model presented here may provide some insight into the mechanisms of G-protein-coupled neurotransmitter receptors. We thank Dr. T. Johansen for guidance in using the University of Wisconsin Genetics Computer Group programs. S.G.D. thanks Universitd Rend Descartes and the Institut National de la Santd et de la Recherche Mddicale, France, for providing working facilities during preparation of the manuscript. This work was supported by the Norwegian Research Council for Science and the Humanities and the Troms Fylkeskommune. 1. Altar, C. A., Boyar, W. C., Walsley, A. M., Liebman, J. M., Wood, P. L. & Gerhardt, S. G. (1988) Naunyn-Schmiedeberg's Arch. Pharmacol. 338, 162-168. 2. Sokoloff, P., Giros, B., Martres, M.-P., Bouthenet, M.-L. & Schwartz, J.-C. (1990) Nature (London) 347, 146-151. 3. Van Tol, H. H. M., Bunzow, J. R., Guan, H.-C., Sunahara, R. K., Seeman, P., Niznik, H. B. & Civelli, 0. (1991) Nature (London) 350, 610-614. 4. Meltzer, H. Y., Matsubara, S. & Lee, J.-C. (1989) J. Pharmacol. Erp. Therp. 251, 238-246. 5. Bunzow, J. R., Van Tol, H. H. M., Grandy, D. K., Albert, P., Salon, J., Christie, M., Machida, C. A., Neve, K. A. & Civelli, 0. (1988) Nature (London) 336, 783-787. 6. Hausdorff, W. P., Hnantowich, M., O'Dowd, B. F., Caron, M. G. & Lefkowitz, R. J. (1990) J. Biol. Chem. 265, 1388-1393. 7. Henderson, R. & Unwin, P. N. T. (1975) Nature (London) 257, 28-32. 8. Henderson, R., Baldwin, J. M., Ceska, T. A., Zemlin, F., Beckmann, E. & Downing, K. H. (1990) J. Mol. Biol. 213, 899-929. 9. Findlay, J. B. C. & Pappin, D. J. C. (1986) Biochem. J. 238, 625642. 10. Henderson, R. & Schertler, G. F. X. (1990) Philos. Trans. R. Soc. London Ser. B 326, 379-389. 11. Nathans, J. & Hogness, D. S. (1983) Cell 34, 807-814. 12. Deisenhofer, J., Epp, O., Miki, K., Huber, R. & Michel, H. (1985) Nature (London) 318, 618-624. 13. Dearry, A., Gingrich, J. A., Falardeau, P., Fremeau, R. T., Bates, M. D. & Caron, M. G. (1990) Nature (London) 347, 72-76. 14. Sunahara, R. K., Niznik, H. B., Weiner, D. M., Stormann, T. M., Brann, M. R., Kennedy, J. L., Gelernter, J. E., Rozmahel, R., Yang, Y., Israel, Y., Seeman, P. & O'Dowd, B. F. (1990) Nature (London) 347, 80-83. 15. Zhou, Q.-Y., Grandy, D., Thambi, L., Kushner, J. A., Van Tol, H. H. M., Cone, R., Pribnow, D., Salon, J., Bunzow, J. R. & Civelli, 0. (1990) Nature (London) 347, 76-80. 16. Kobilka, B. K., Frielle, T., Collins, S., Yang-Feng, T., Kobilka, T. S., Francke, U., Lefkowitz, R. J. & Caron, M. G. (1987) Nature (London) 329, 75-79.
Proc. Natl. Acad. Sci. USA 88 (1991)
8115
17. Fargin, A., Raymond, J. R., Lohse, M. J., Kobilka, B. K., Caron, M. G. & Lefkowitz, R. J. (1988) Nature (London) 335, 358-360. 18. Julius, D., MacDermott, A. B., Axel, R. & Jessell, T. M. (1988) Science 241, 558-564. 19. Julius, D., Huang, K. N., Livelli, T. J., Axel, R. & Jessell, T. (1990) Proc. Natl. Acad. Sci. USA 87, 928-932. 20. Cotecchia, S., Schwinn, D. A., Randall, R. R., Lefkowitz, R. J., Caron, M. G. & Kobilka, B. K. (1988) Proc. Natl. Acad. Sci. USA 85, 7159-7163. 21. Kobilka, B. K., Matsui, H., Kobilka, T. S., Yang-Feng, T. L., Francke, U., Caron, M. G., Lefkowitz, R. J. & Regan, J. W. (1987) Science 238, 650-656. 22. Regan, J. W., Kobilka, T. S., Yang-Feng, T. L., Caron, M. G., Lefkowitz, R. J. & Kobilka, B. K. (1988) Proc. Natl. Acad. Sci. USA 85, 6301-6305. 23. Frielle, T., Collins, S., Daniel, K. W., Caron, M. G., Lefkowitz, R. J. & Kobilka, B. K. (1987) Proc. Natl. Acad. Sci. USA 84, 7920-7924. 24. Kobilka, B. K., Dixon, R. A. F., Frielle, T., Dohlman, H. G., Bolanowski, M. A., Sigal, I. S., Yang-Feng, T. L., Francke, U., Caron, M. C. & Lefkowitz, R. J. (1987) Proc. Natl. Acad. Sci. USA 84, 46-50. 25. Kubo, T., Fukuda, K., Mikami, A., Maeda, A., Takahashi, H., Mishina, M., Haga, T., Haga, K., Ichiyama, A., Kangawa, K., Kojima, M., Matsuo, H., Hirose, T. & Numa, S. (1986) Nature (London) 323, 411-416. 26. Peralta, E. G., Winslow, J. W., Peterson, G. L., Smith, D. H., Ashkenazi, A., Ramachandran, J., Schimerlik, M. I. & Capon, D. J. (1987) Science 236, 600-605. 27. Bonner, T. I., Buckley, N. J., Young, A. C. & Brann, M. R. (1987) Science 237, 527-532. 28. Dixon, R. A. F., Sigal, I. S., Rands, E., Register, R. B., Candelore, M. R., Blake, A. D. & Strader, C. D. (1987) Nature (London) 326, 73-77. 29. Strader, C. D., Sigal, I. S., Register, R. B., Candelore, M. R., Rands, E. & Dixon, R. A. F. (1987) Proc. Nati. Acad. Sci. USA 84, 4384-4388. 30. Fraser, C. M., Wang, C.-D., Robinson, D. A., Gocayne, J. D. & Venter, J. C. (1989) Mol. Pharmacol. 36, 840-847. 31. Strader, C. D., Sigal, I. S., Candelore, M. R., Rands, E., Hill, W. S. & Dixon, R. A. F. (1988) J. Biol. Chem. 263, 10267-10271. 32. Strader, C., Candelore, M. R., Hill, W. S., Sigal, I. S. & Dixon, R. A. F. (1989) J. Biol. Chem. 264, 13572-13578. 33. Sunahara, R. K., Guan, H.-C., O'Dowd, B. F., Seeman, P., Laurier, L. G., Ng, G., George, S. R., Torchia, J., Van Tol, H. H. M. & Niznik, H. B. (1991) Nature (London) 350, 614-619. 34. Needleman, S. B. & Wunsch, C. D. (1970) J. Mol. Biol. 48, 443453. 35. Devereux, J., Haeberli, P. & Smithies, 0. (1984) Nucleic Acids Res. 12, 387-395. 36. Kyte, J. & Doolittle, R. F. (1982) J. Mol. Biol. 157, 105-132. 37. Hopp, T. P. & Woods, K. R. (1981) Proc. Natl. Acad. Sci. USA 78, 3824-3828. 38. Weiner, S. J., Kollman, P. A., Nguyen, D. T. & Case, D. A. (Y186) J. Comp. Chem. 7, 230-252. 39. Ferrin, T. E., Huang, C. C., Jarvis, L. E. & Langridge, R. (1988)J. Mol. Graphics 6, 13-27. 40. Connolly, M. L. (1983) Science 221, 709-713. 41. Chou, P. Y. & Fasman, G. D. (1978) Annu. Rev. Biochem. 47, 251-276. 42. Kabsch, W. & Sander, C. (1983) FEBS Lett. 155, 179-182. 43. Benchetrit, T., Bissery, V., Mornon, J. P., Devault, A., Crine, P. & Roques, B. P. (1988) Biochemistry 27, 592-596. 44. Quian, N. & Sejnowski, T. J. (1988) J. Mol. Biol. 202, 865-884. 45. Bohr, H., Bohr, J., Brunak, S., Cotterill, R. M., Lautrup, B., Noerskov, L., Olsen, 0. H. & Petersen, S. B. (1988) FEBS Lett. 241, 223-228. 46. Eidne, K. A., Taylor, P. L., Zabavnik, J., Saunders, P. T. K. & Inglis, J. D. (1989) Nature (London) 342, 865. 47. Giros, B., Sokoloff, P., Martres, M.-P., Riou, J.-F., Emorine, L. J. & Schwartz, J.-C. (1989) Nature (London) 342, 923-926. 48. Monsma, F. J., McVittie, L. D., Gerfen, C. F., Mahan, L. C. & Sibley, D. R. (1989) Nature (London) 342, 926-929. 49. Wang, H., Lipfert, L., Malbon, C. C. & Bahouth, S. (1989) J. Biol. Chem. 264, 14424-14431. 50. Nilsson, I. & von Heijne, G. (1990) Cell 62, 1135-1141. 51. Brown, G. C. (1990) FEBS Lett. 260, 1-5. 52. Tedesco, J. L., Seeman, P. 5. & McDermed, J. D. (1979) Mol. Pharmacol. 16, 369-381. 53. Burgen, A. S. V., Roberts, G. C. K. & Feeney, J. (1975) Nature (London) 253, 753-755.
Vol. 88, pp. 8111-8115, September 1991 Neurobiology
Molecular dynamics of dopamine at the D2 receptor (guanine nucleotide-binding regulatory protein-coupled receptors/three-dimensional structure/electrostatic potentials! molecular modeling)
SVEIN G. DAHL*, 0YVIND EDVARDSEN, AND INGEBRIGT SYLTE Department of Pharmacology, Institute of Medical Biology, University of Troms0, N-9001 Troms0, Norway
Communicated by Jean-Marie Lehn, May 20, 1991 (received for review January 28, 1991)
Cloning and sequencing of neurotransmitter receptors has demonstrated that they may be divided into various "superfamilies." The dopamine D2 receptor belongs to the superfamily of receptors transferring signals into cells through guanine nucleotide-binding regulatory (G) proteins. Antagonism of postsynaptic D2 receptors has long been regarded as the primary mechanism of action of antipsychotic drugs, and it has recently been proposed that also dopamine D1 (1), D3 (2), and D4 (3) receptors and serotonin 5-HT2 receptors (4) may be involved in antipsychotic drug action. Knowledge of the three-dimensional structure of such receptors would significantly add to our understanding of their molecular mechanisms and be useful in the search for drugs that affect these receptors. Amino acid sequences of more than 10,000 proteins are known, but only some 450 three-dimensional protein structures have been reported. No detailed three-dimensional crystal structure of a neurotransmitter receptor molecule is yet available, to our knowledge. In the absence of such data, we have constructed a model of the D2 receptor and used it to simulate the molecular dynamics of dopamine-receptor interactions. Based on structural similarities within the superfamily of G-protein-coupled neuroreceptors, the model was constructed from the rat D2 receptor sequence (5) and refined by molecular dynamics simulations and molecular mechanics energy minimization. The modeling was based on the following five hypotheses:
(i) Membrane-spanning domains of G-protein-coupled neurotransmitter receptors are a-helices. Hydropathy indices along the peptide chains show that G-protein-coupled receptors contain seven hydrophobic domains, usually assumed to correspond to seven transmembrane stretches of residues (6). Electron diffraction (7) and electron microscopy (8) experiments have demonstrated that bacteriorhodopsin has seven a-helical segments spanning the cell membrane. All available biological and chemical data indicate also that the membrane-spanning domains of visual rhodopsin form a-helices (9, 10). G-protein-coupled neurotransmitter receptors share some amino acid homology with visual rhodopsin, particularly in the seven hydrophobic domains (11) that most likely are a-helices. (ii) Each transmembrane helix contains 27 amino acids. An a-helix with 27 amino acids has a total length of =40 A and may traverse the cell membrane in an approximately perpendicular orientation. The 11 membrane-spanning a-helices in the crystal structure of the photosynthetic reaction center of Rhodopseudomonas viridis contain on average 27 amino acids (12). (iii) Membrane-spanning segments of G-protein-coupled neurotransmitter receptors have similar localizations in aligned protein sequences. Hydropathy indices provide only approximate locations of transmembrane segments in the peptide chain of a protein, and for one and the same receptor such locations often differ by several amino acids among different reports. Since G-protein-coupled neurotransmitter receptors show 50-75% sequence homology in the putative transmembrane domains, we assumed that the segments of all receptors corresponding to each a-helix would be placed in matching positions by sequence alignment, as indicated in Fig. 1. Presumably, this enabled more precise prediction of the positions of the transmembrane helices in the peptide chains, based on average hydropathy indices of 14 neuroreceptors from the same superfamily. (iv) The most polar surface areas of the transmembrane helices form a central core. The seven a-helical segments in bacteriorhodopsin are roughly perpendicular to the cell membrane, slightly inclined to one another at various angles up to 200, and closely packed to form an oval ring structure (7, 8). Site-directed mutagenesis experiments have suggested that ligands bind to the putative transmembrane regions of 132adrenergic (12, 28, 29) and muscarinic ml receptors (30), and it has been postulated that f-adrenergic receptors form a rhodopsin-like core containing a ligand binding site (28). The similarities in their primary structures indicate that this also may be the case for other G-protein-coupled neuroreceptors. (v) G-protein-coupled neurotransmitter receptors have a common ligand-binding site. Site-directed mutagenesis experiments have suggested that agonists and antagonists interact with Asp-113 and that agonists but not antagonists
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Abbreviation: G protein, guanine nucleotide-binding regulatory protein. *To whom reprint requests should be addressed.
A three-dimensional model of the dopamine ABSTRACT D2 receptor, assumed to be a target of antipsychotic drug action, was constructed from its amino acid sequence. The model was based on structural similarities within the superfamily of guanine nucleotide-binding regulatory (G) proteincoupled neuroreceptors and has seven a-helical transmembrane segments that form a central core with a putative ligand-binding site. The space between two residues postulated to be involved in agonist binding, Asp-80 and Asn-390, perfectly accommodated an anti-dopamine molecule. Molecular electrostatic potentials were mainly negative on the synaptic side of the receptor model and around aspartate residues lining the central core and positive in the cytoplasmic domains. The docking of dopamine into a postulated binding site was examined by molecular dynamics simulation. The protonated amino group became oriented toward negatively charged aspartate residues in helix 2 and helix 3, whereas the dopamine molecule fluctuated rapidly between different anti and gauche conformations during the simulation. The receptor model suggests that protonated ligands are attracted to the binding site by electrostatic forces and that protonated agonists may induce conformational changes in the receptor, leading to G-protein activation, by increasing the electrostatic potentials near Asp80.
8111
8112 r)i 02 03 51) a 5HT1c 5HT2 a2a a2h
32 mI
12
14
i'- 'Vi-
Neurobiology: Dahl et al.
Proc. Natl. Acad. Sci. USA 88 (1991)
(9) 0GTGLVVERDFSVI kILTACFLSLLII.STLLGNTLVCAAVIRFRHLRSKVTKNFFVISLAVSDLIVAVI.VMPWKAVAEIAC;F.WPFGSF.GCNIWVAFDIMCSTASILNLCVISVDRYWA I'-;. ; f20) PFNGSECKADRPPHYNYYAMILLTLLIF IIVFQIVLVCMAVSREKALT TTNYLIVSLAVADLLVATLVMPWVVYLEVVGE VKFSRI HCDI FVTl.DVMMCTASI LNLCAIS I DRYTAVA TTNYLrVVSLAVADLl.VATLVMPWVVYLFEVTGGVINFSRICCOVFVTTLDVMMCTASI I .NLC'AISIDRYTAVV' (2.8) ...... SDVTVSYQVITSLLLCTLIFCAVL(:"ACVAAIALERSLQN.VANYLIGSLAVTDLMVSVI VLPMAALYQVl,N.KIITLGQVTCDLFIALDVLCCTSS1LFLCAIALDRYWAII .';4f' (45) LF ..... QFPDGVONWPALSIVVIIIMTIGGNILVIMAVSMEKKLHN.AT YFLMSLAIADMLVGLLVMPLSLLAILYDYY"PLPRYLCPYWISLDVLFSTASIMHLCAISLDRYVAI 15;) (60) LPPTCLSILHLQEKNWSAI1LTTWVIILTIAQ#ILVIMAVSLEKKLQN.ATtYFLMS1JIADM 1lIGFI,VMPVSMLTILYGYRNPLPSKLCAIWIYLlVLFSTASIMH ZAISLDRYVA T' (37) P ...... QLDVTRAISVGLVLCAFI4LFAIVNILVILSVACNRHLRT.PTNYFIVNLAIADLLLS'TVLPFSATLEVLGY.WVLGRIFCDIWAAVDVLCCTASILSLCAISIDRY1 .V\z- ..I.,
(1i) GAENSTrGVNRARPHAYYALSYCALILAI IFGNGLVCAAVLRERALT
(19) (37) (44 f26) (I1) (12) (18)
t:CGARAT PYSLQVTLTLVCLACLLMLLTV-FVLVIIAYFTSRAXA.PQNLFLVSLASADILVATLVIPFSLArEVMGY.IYF:KTWC-EIYI.A DVLFCTSSIVHLCAISLDRYWSI ,I SWGPPRGQYSAGAVAGLAAVVGFLIVFTV'VGNVLVVIAVYlTSRALRA. PQNLFLVSLASADI LVATLVMPFSl.ANELMAY. WYFGQVWCGVYLALDVILFCTSSIVHLCAISIDRYWSVrr iPASESPEPLSQQWTAGMGLLMALIVLLIVAGNVLVIVAIAKTPR[WT. LTNLFIMSLASADLVMGLLVVPFGAT IVVWG. RWEYGSFFCELWTSVDVLCVTASIETLCVIADRYI.AI A :. QRDEVWVVGMGIVMSLIVLAIVFGNVLVITAIAKFERLT VTNYFITSLACADLVMGIlAVVPFGAAHi LM KMWTFGNFWCEFWTSI DVLCVTASI ETLCVIAVDRYFA IT -: NITVLAPGKGPWQVAFIGITTCLLSLATVTONLLVLISFKVNTELKT.VNNYFLLSLACADLIIGTFSMNLYTTYI1LMGH. ALGTLACDLWLALDYVASNASVMNLLLISEDRYFSV 1;, ALTSPP ... YKT2FEWFIVLVAGSLSLVTIIGNILYMVSIKVNRHLT.VNNYFLFSLACADLIIGVFSMNLYTLYTYIGY.NPLGPVVCDI.WLALDYVVSNASVMNLLIISFDRYF-C V i. VTSSSHNRYETVEMVFIATVTGSLSLVTVVGNI IVMISIKVNRQLCT KY PLGAVVCDLWIALDYVVS NASVMNLI. I I S FDRYF VirF W. VNNYFLFSLACADI.T1 GAFSMNLYTVY K1.
r-------
IV --
.-...
DI1 P FRY. . ERKMT PK .AA FXL ISVAWT LSVL IS F IPVQ LSWHKAK PTS PS Ic)GNATS LA E T TD NCDS SLS RTYA ISSSsv I SFY IPVAI M IVTY T R IY RIAQKQ I EkI AA L E AAVwH A K N C- O 1 D2 PMLY. .NTRYSSKRRVTVMIAIVWSVLSFTISCPI,.LFGLNNTDQNE ....C.. IIANPAFVVYSSIVSFYVPFIVTI.LVYIKIYTVLRK ...... RRKPVNTKRSSRAFRH 03 PVHYQHiGTGOSSCRRVALMITAVWVLAFAVSCPL. LFGFNTTGDPS .............. ICSISNPDFVIYSSVVSFYVPFGVTVIVYARIYIVLRQ.....RQRKRI L1'RONS,0QC:S 5HTIa PIDY.VN. . .KRTPRPRALISLTVLIGFLISIP.PMLGWRTPEDRSDPD ........... ACTISKDHGYTIYSTFGAFYIPLLLMLVLYGRIFRAARFRIRKTVKKVFKT A.
5i1rl1c
514'2 cl a2a
a2b
p1
P2
Ml M2 M4
P... IEHSRIF'NSRTKAIMKIAIVVAISIGVSVPIPVIGLRD.E.. .S.. KVFVNNTTCVL...-.NDPNFVLIGSFVAFFIPLTIMVITYFLTIYVLRRQTLMLLR .............. P ... IHHSRFNSRTKAFLKIIAVWTISVGISMPIPVFGI.Q ...... DDSKVFKEGS.CI 1L.....ADDNFVLIGSFVAFFIPLTIMVITYFLTIKSLOKEATLCVSDLSTRAKlASTS)E I). SI.QYPT... LVTRRKAILALLSVJVLSTVISIG. PLLGWKEPAPNDDKE ............ CGVTEEPFYALFSSLGSCSYIPLAVILVMYCRVYIVAKRTT . AIEY ..NLKRTPRRIKAI. IITCWVISAVISFPPLISIEKKGGGGGPOP ........ AEPRCEINDQKWYVISSCIGSFFAPCLIMILVYVRIYQIAKRRTRVPPSRHRPDAVAAIiPPC;(;IT AVEY. .NLK.RTPRRVKATIVAVWLISAVISFPPLVSLYRQPDG ............ AAYPQCGLNDETWYILSSCIGSFFAPCLIMGLVYARIYRVAKRRTRTLISEKRAPVGPI.,)(;,ASPPT I' PFRYE . QS. LLTRARARGLVCTVWAISAIVSFLPI LMHWW. RAESDEARRCY . .. NDPKCCDFVTNRAYAIASSVVSFYVPLCIMAFVyLRVFREAQK . PFKY .. QS. LLTKNKARVII LMVWIVSGLTSF. LPIQMHWYRATHQEAI NCY .... ANETCCDFFTNQAYAIASSIVSFYVPLVIMVFVYSRVFQEAKR ..................... PLSY .. RAKRTPRRAALM. IGLAWLVSFVLWAPAILFWQYI.VGERTVI . ....... AGQCYIQFlSQPI ITFGTAMAAFYLPVTVMCTLYWRIYRETENRARE. AALQGSE'1 i'(-;K PLTY. . PVKRTTKMAGMM. IAAAWVLSFI LWAPAILFWQFIVGVRTVEDGE ........ CYIQFFiSNAAVTFGTAIAAFYLPVI IMTVLYWHISRASKSRIKKDKKEPVANODPVSPSLV PLTY ..PARRTTKMAGLM. IAAAWVLSFVLWAPAILFWQFVVGKRTVPDNII ........ CFIQF..SNPAVTFGTAIAAFYLPVVIMTVLYIHISlASRSRVHKHRPECIPKEKKAKTI .AF'I. ,VI
D1 .... SQPESSFKMSFK....... RITKVLKTLSVIMGVFVCCWLPFFILNCILPFCGSGETQPFCIDSNTFDVFVWFGWANSSLNPI IY. AF'NADFRKAFSTLWC;YPIII -A 02 F.. 'EIQTMPNGKTRTSLK. TMSRRKL. SQQKBKKATQMLAIVLGVFI ICWLPFFITHILNIHCDCN ...... I PPVLYSAFTWLGYVNSAVNPI IYTTFNI EFRKAFMK ' HC D3 ... LEVRKLSNGRLSTSRIGR.PLQPR. GVPLRBKKATQKVVIVLGAFIVCWLPFFLTHVLNTHCQA.....CEYVSPELYRATTWLGYVNSALNPVIYTTFNVEFRKAFI.K I.SC SHTl a LAR .BRKTVKTLGI IMGTFILCWLPFFIVALVLPFCESSC ... HMPTLLGAI INNLGYSNSLLNPVIYAYFNKDFQNAFKK II KCNFPCRU INN .................... KKASKVLGIVFFVFLIMWCPFF1TNILSVLCGKACNO .. KLMEKLLNVFVWIGYVCSGI W LVYTLFNKIYRRAFSKYI.pRCDY KP K 5HT2 . SISNSQKACKVLGIVFFLFVVMVCPFFITNIMAVICKESCNE. CLi 1 68 NVIGAI1LNVFVWIGYLSSAVIPl.VYTLFNKTYRSAFSRY I cal FSR ............................. EKKAKTLGIVVGMFILCWLPFFIALPLGSIFSTL.....KPPDAVFKVVFVLGYFNSCLMPI IYPCSSKEFKRAFMRI LGCOCHS;R 144 a2a RGRRGSGRRLQGRGRSASGLPRRRAGAGGQNLIKRFTFVLAVVIGVyVIVCNFPFFFTYTLTAVGCS. VPRTLFKFFFWFGYCNSSLNPVIYTIFNHDFRRAFKKIL -.. CPON .4; a2b ASSRSVEFFLSRRRRARSSVCRRK, .. VAQARXKRFTFVLAVVMGV VLCWFPFFFIYSLYGICREAC.Q.. VPGPLFKFFFWIGYCNSSLNPVIYTVFNQDFRPSFKHI LF-NHNNRCRF-; ',2 LAN .............. GRAGKRRPSRLVALRBQKALKTLGIIMGVFTLCVLPFFI.ANVVKAFHREL ...... VPDRLFVFFNWLGYANSAFNPIIY.CRSPIDFRKAFG.I CI 2 P-AP 32 ....................... RRSSKFCLKEHKALKTLGIIMGTFTLCWLPFFIVNIVHVIQDNL ...... IRKEVYILLNWIGYVNSGFNPLIY.CRSPDFRIAFOELL..2 Ml KGQKPRGKEQLAKRKTFSLVK ... ........ HKKAARTLSAI LLAFIVTWTPYNIMVLVSTFCKD. ....VPETLWELGYWLCYVNSTINPMCYALCNKAFRDTFRLLLLL R ..... . .'4 M2 SGQNGDEKQNIVARKIVKMTKQPAKKKPPPSRIEKKVTRTI LAILLAFI ITVAPYNVMVLINTFCAP . C ..... IPNTVWTIGYWLCYINSTINPACYALCNATFKKTFKHLLMCiYKNIOC; M4 TPAGMRPAANVARKFASIARNOVRKKRQMAARERKVTRTIFAI LLAFI LTWTPYNVMVlVNTFCQS . C...TPDTVWSIGYWLCYVNST TNPACYALCNATFKKTFNHII,LPQ. C )YRN' T .....................
OCOYKE.NR
FIG. 1. Aligned amino acid sequences of G-protein-coupled neurotransmitter receptors. Dl, human dopamine D1 receptor (13-15); D2 and D3, rat dopamine D2 (5) and D3 (2) receptors, respectively; 5-HTla, human 5-HT1a receptor (16, 17); 5-HTlc, rat 5-HT1, receptor (18); 5-HT2, rat 5-HT2 receptor (19); alb, hamster alb-adrenergic receptor (20); a2a, human au-adrenergic receptor (21); a2b, human a2b-adrenergic receptor (22); P1, human fi-adrenergic receptor (23); ,2, human f32-adrenergic receptor (24); Ml, porcine muscarinic ml receptor (25); M2, porcine muscarinic m2 receptor (26); M4, human muscarinic m4 receptor (27).
interact with Asp-79, Asp-130, and Asn-318 (29, 31) and Ser-204 and Ser-207 (32) in the (82 receptor. The corresponding aspartic residues also seem to be involved in ligand binding to muscarinic ml receptors (30). Asp-79, Asp-113, and Asp-130 in the ,2 receptor are conserved in all sequences shown in Fig. 1 and in the dopamine D4 (4) and D5 (33) receptors, which suggests that these residues are involved in ligand binding and signal transduction in all G-proteincoupled neuroreceptors. Asn-318 in the P2 receptor is conserved in the dopamine receptors, muscarinic acetylcholine receptors, and adrenergic receptors. Ser-204 and Ser-207 in helix 5 of the p2 receptor are conserved in the dopamine receptors. It is possible that the corresponding residues (Asn-390, Ser-194, and Ser-197) also are involved in agonist binding to D2 receptors. The D2 sequence, like all the other sequences shown in Fig. 1, contains prolines in transmembrane segments IV-VII. Proline often appears at the end of a-helices in proteins and is known as a "helix breaker." However, the M subunit of the photosynthetic reaction center of R. viridis has a proline near the middle of the third transmembrane helix (12), which demonstrates that a proline does not necessarily represent a breaking point of a transmembrane a-helix.
METHODS
Sequence AlignMent The sequence of the rat D2 receptor (5) and the sequences of 13 other G-protein-coupled neurotransmitter receptors were aligned by the NeedlemanWunsch method (34), with the Gap program of the University of Wisconsin Genetics Computer Group Sequence Analysis Software Package (35). With a gap weight of 3.0, and a length weight of 0.1, hydropathy indices were calculated for each sequence by two methods (36, 37), and average indices were
calculated from the aligned sequences. The 14 sequences were selected to get a balanced contribution from various receptor families to the average indices. Receptor Modeling. Structural refinement by molecular mechanics energy minimization and molecular dynamics simulations were done on a Cray X/MP-28 supercomputer, using the AMBER all-atom force field (38). The cut-off distance for nonbonded interactions was 8 A. A distance-dependent dielectric function was used to mimic the dielectric damping effect of water on intramolecular interactions. Energy refinement was done by 500 cycles of steepest-descent minimization followed by 2000 cycles of full-conjugate gradient minimization. Molecular dynamics simulations were performed at 310 K with a step length of 0.001 psec, after an initial equilibrium dynamics phase starting from 0.1 K. Molecular electrostatic potentials 1.4 A outside the water-accessible surface were calculated with a distance-dependent dielectric function and a 10.0-A nonbonded cut-off radius. Initial models of each transmembrane a-helix, including the side chains, were constructed with the MIDAS computer graphics programs (39) from the rat D2 sequence (5). Each helix was refined by energy minimization and its wateraccessible surface (40) and molecular electrostatic potentials (39) were calculated. The helices were assembled on the computer graphics system such that the most polar region of each helix surface was oriented toward a central core. The helices were closely packed in an antiparallel arrangement, clockwise as viewed from outside the cell membrane, such that the water-accessible surface contained no holes between helices after structure refinement. The seven-helix structure, without connecting loops, was refined by energy minimization. The terminal segments and loops between helices were added, based on a method (41) that gives s50l correct predictions of secondary structures
Neurobiology: Dahl et al.
Proc. Natl. Acad. Sci. USA 88 (1991)
8113
in proteins (42). Although other methods may give more correct secondary structure predictions (43-45), we did not have any adequate method for modeling the 131-amino acid cytoplasmic loop between helix 5 and helix 6. Therefore, only 7 amino acids of this loop, taken from the segments connected to helix 5 and helix 6, were included in the model and linked to each other. Thus the segment of the third cytoplasmic loop that differs between the two D2 receptor subtypes (5, 46-48) was not included in the model. The receptor model was refined by molecular mechanics energy minimization. A dopamine molecule was then placed in the putative binding site, where it fit precisely into the space between Asp-80 and Asn-390. The receptor-ligand complex was further refined by energy minimization, which produced only negligible changes in the receptor structure. Because the initial modeling of the extracellular and cytoplasmic domains had been based on fairly inaccurate secondary structure predictions, these parts of the model were further refined by 2000 steps of molecular dynamics simulation at temperatures increasing from 0.1 K to 300 K, followed by energy minimization, while the transmembrane helices were kept fixed. This produced minor changes in the loops between helices and more noticeable conformational changes in the N- and C-terminal segments. Finally, the wateraccessible surface and molecular electrostatic potentials were calculated for the whole receptor model.
RESULTS Prediction of Transmembrane Domains. Average hydropathy indices of the 14 aligned receptor sequences are shown in Fig. 2. Plotting of areas under "windows" of 27 amino acids under the average hydropathy index curves as a function of amino acid number produced peaks that predicted the starting point of each transmembrane helix. Hydropathy indices calculated by the method of Kyte and Doolittle (36) gave well-defined peaks for helices 1-6, whereas indices calculated by the method of Hopp and Woods (37) defined the starting points of helices 1, 4, 5, 6, and 7. Three-Dimensional Receptor Structure. The D2 receptor model is shown in Fig. 3. After energy refinement, the transmembrane a-helices remained approximately antiparallel, were slightly bent, and were somewhat closer together at the cytoplasmic end than at the synaptic end. Two of the residues postulated to be involved in agonist binding, Asp-80 in helix 2 and Asn-390 in helix 7, were placed adjacent to each C
> 30 c6x
600
O_ D 20 C._ I'r 10
400-
C
Cu5 (D
2
1 0. x _
WE
D a
I
n0
~hIZ~ihiiB I1*ww 11lr a X.WXo
n
0
200
400
zuu
200
30
400
600
600
D
10- \JM\AI'~~~~~~~~~~~~\~ 1-1
,
15 kcal/mol.
other at a distance that closely accommodated the energy minimized anti conformation of dopamine. Asp-114 in helix 3 and Ser-194 and Ser-197 in helix 5 were placed nearer to the synaptic membrane surface than Asp-80 and Asn-390. Presumably, modeling of the other dopamine receptors by the same procedure would place the corresponding residues in similar geometrical positions. Electrostatic Potentials. The molecular electrostatic potentials were based on the charges of all surrounding atoms within the defined cut-off radius and depended on molecular conformation and on atomic charges. The potentials were mainly negative in the putative extracellular domains and in parts of the central core of the receptor model and positive in the postulated cytoplasmic domains (Fig. 3). The lowest electrostatic potentials in the central core were -40 kcal/mol (1 cal = 4.184 J) near Asp-80 in helix 2 and -37 kcal/mol near Asp-114 in helix 3. Strongly positive electrostatic potentials, up to 38 kcal/mol, were also found in the area between helix 4 and helix 5, extending from a level one helical turn below Asp-80 and down into the second cytoplasmic loop. Molecular Conformations and Dynamics of Dopamine. Internal movements in molecules occur on a fsec time scale. The molecular dynamics of protonated dopamine approaching the postulated receptor binding site was examined in an 80-psec simulation where the receptor was kept in a fixed position. The coordinates and energies were saved at 0.5psec intervals. The simulation required 200 min of centralprocessing-unit time on the Cray computer. The electrostatic forces were not sufficient to attract the dopamine molecule to the postulated binding site during the simulation. Weak potentials (1.0 kcal), directing the neuro-
8114
Neurobiology: Dahl et al.
transmitter toward the area between Asp-80 and Asn-390, therefore, were applied during the initial 65 psec of the simulation. Constrained atomic distances were as follows: Dopamine nitrogen to Asp-80 carboxyl oxygen, 0-25 psec at 25 A, 25-55 psec at 10 A, and 55-65 psec at 4.5 A; dopamine para-oxygen to Asn-390 carbonyl oxygen, 40-55 psec at 10 A and 55-65 psec at 4.5 A. The side chain of the dopamine molecule moved several times from one side of the phenyl ring plane to the other side and fluctuated rapidly between various anti and gauche conformations throughout the simulation. Fig. 4 shows a part of the receptor model and some of the dopamine structures that were observed during the simulation. As the neurotransmitter moved down the central core, its positively charged amino group became oriented toward negatively charged aspartate residues (Asp-80 and Asp-114) in helices 2 and 3. The amino group of dopamine remained close to Asp-80 during the final 15 psec of the simulation, after removal of the directing force. Fig. 4 Lower shows a dopamine molecule in a postulated binding site between Asp-80 in helix 2 and Asn-390 in helix 7.
DISCUSSION The negative molecular electrostatic potentials around the N-terminal segment and the loops between helices 2 and 3, helices 4 and 5, and helices 6 and 7 support the hypothesis (49) that these represent synaptic domains of the receptor. The largely negative electrostatic potentials at the synaptic side
FIG. 4. Docking of dopamine into a postulated D2 receptor binding site. (Upper) Selected structures from an 80-psec molecular dynamics simulation of dopamine approaching the binding site. A part of helices 2, 3, 6, and 7 are shown with their connecting loops. (Lower) Energy-minimized anti conformation of dopamine between Asp-80 (blue) and Asn-390 (green). Water-accessible molecular surfaces (dotted) are color coded according to electrostatic potentials as in Fig. 3.
Proc. Nati. Acad. Sci. USA 88 (1991)
indicate that protonated ligands are attracted to the receptor by electrostatic forces. f32-Adrenergic receptors are coupled to stimulatory G (Gj) proteins in 11- to 15-residue segments located near the cytoplasmic surface of the cell membrane, at each end of the third intracellular loop, and at the N-terminal part of the cytoplasmic tail (6). The cytoplasmic domains in the D2 receptor model with strong positive electrostatic potentials (Fig. 3) closely corresponded to the parts of the /32 receptor that are coupled to a G, protein. This suggests that positive electrostatic potentials around certain cytoplasmic domains of the D2 receptor may be involved in G-protein interaction. Our calculations also support, by a completely different approach, the "positive-inside" rule for membrane proteins, which has been suggested from site-directed mutagenesis experiments (50). It is interesting to note that the electrostatic potentials were strongly positive between helix 4 and helix 5, only 10 A from Asp-80, which was surrounded by an equally strong negative electrostatic field. This indicates that agonist-induced changes in the electrostatic field may play a role in signal transduction in the D2 receptor, as suggested for G-proteincoupled receptors in general (51). Such a process could be initiated by binding of a protonated agonist to Asp-80 to neutralize its negative electrostatic field. Energy-minimized anti-dopamine fit perfectly into the space between Asp-80 and Asn-390 (Fig. 4), in accordance with the hypothesis that this is the active conformation (52). However, our molecular dynamics simulations indicate that, rather than obtaining a complete lock-and-key fit to the active site while the neurotransmitter stays in this conformation, the interaction starts with an electrostatic attraction between the protonated amino group on the ligand and negatively charged aspartate residues at the binding site, while the neurotransmitter moves between various conformations (Fig. 4). Such a "zipper" mechanism whereby a flexible ligand binds to a macromolecule in several successive steps has been postulated from thermodynamic considerations (53). Still, the prevailing concept of ligand-receptor interactions has been a lock-and-key mechanism requiring a specific conformation of the ligand. The present molecular dynamics simulation supports the "zipper" type of mechanism for dopaminereceptor interactions. Fig. 5 shows the antipsychotic drug cis-(Z)-chlorprothixene with its protonated dimethylamino group in close contact with Asp-114 in helix 3. The localization of Asp-114 closer to the synaptic membrane surface than Asp-80 and Asn-390 suggests that neuroleptics act by preventing dopamine access to its endogenous binding site and offers a simple steric explanation of how Asp-80 and Asn-390 may interact with
FIG. 5. Molecular structure of cis-(Z)-chlorprothixene in a postulated D2 receptor binding site, viewed from the synaptic side. Helix 1 is at the bottom of the figure.
Neurobiology: Dahl et al. agonists but not with antagonists. The dynamics of the ligand-receptor system would still enable competitive binding of agonists and antagonists. The calculations presented here were intended to present an approximate overall model of the D2 receptor, which might provide further insight into its mechanisms. The model does not rule out other possible arrangements of the various parts of the receptor. It is possible that the seven helices may be more tilted relative to each other, that the central core is slightly narrower, and that the seven helices are arranged in an anti-clockwise order viewed from outside the cell, as suggested for bacteriorhodopsin (8), and it is quite likely that the number of residues in each helix may differ from 27. In any case, our calculations clearly indicate that, to understand the molecular mechanisms, neurotransmitterreceptor interactions should be regarded as dynamic processes and that electrostatic mechanisms may be important for ligand binding and signal transduction in the receptor. The receptor model may be used to design protein-engineering experiments to test its validity and provide further insight into receptor mechanisms. It would be interesting, for instance, to examine the relevance of the positively charged area between helix 4 and helix 5 for transducer mechanisms, by site-directed mutagenesis experiments. The model also provides a tool to simulate the molecular dynamics of ligand-receptor interactions. Although probably inaccurate in many details, we feel that the receptor model presented here may provide some insight into the mechanisms of G-protein-coupled neurotransmitter receptors. We thank Dr. T. Johansen for guidance in using the University of Wisconsin Genetics Computer Group programs. S.G.D. thanks Universitd Rend Descartes and the Institut National de la Santd et de la Recherche Mddicale, France, for providing working facilities during preparation of the manuscript. This work was supported by the Norwegian Research Council for Science and the Humanities and the Troms Fylkeskommune. 1. Altar, C. A., Boyar, W. C., Walsley, A. M., Liebman, J. M., Wood, P. L. & Gerhardt, S. G. (1988) Naunyn-Schmiedeberg's Arch. Pharmacol. 338, 162-168. 2. Sokoloff, P., Giros, B., Martres, M.-P., Bouthenet, M.-L. & Schwartz, J.-C. (1990) Nature (London) 347, 146-151. 3. Van Tol, H. H. M., Bunzow, J. R., Guan, H.-C., Sunahara, R. K., Seeman, P., Niznik, H. B. & Civelli, 0. (1991) Nature (London) 350, 610-614. 4. Meltzer, H. Y., Matsubara, S. & Lee, J.-C. (1989) J. Pharmacol. Erp. Therp. 251, 238-246. 5. Bunzow, J. R., Van Tol, H. H. M., Grandy, D. K., Albert, P., Salon, J., Christie, M., Machida, C. A., Neve, K. A. & Civelli, 0. (1988) Nature (London) 336, 783-787. 6. Hausdorff, W. P., Hnantowich, M., O'Dowd, B. F., Caron, M. G. & Lefkowitz, R. J. (1990) J. Biol. Chem. 265, 1388-1393. 7. Henderson, R. & Unwin, P. N. T. (1975) Nature (London) 257, 28-32. 8. Henderson, R., Baldwin, J. M., Ceska, T. A., Zemlin, F., Beckmann, E. & Downing, K. H. (1990) J. Mol. Biol. 213, 899-929. 9. Findlay, J. B. C. & Pappin, D. J. C. (1986) Biochem. J. 238, 625642. 10. Henderson, R. & Schertler, G. F. X. (1990) Philos. Trans. R. Soc. London Ser. B 326, 379-389. 11. Nathans, J. & Hogness, D. S. (1983) Cell 34, 807-814. 12. Deisenhofer, J., Epp, O., Miki, K., Huber, R. & Michel, H. (1985) Nature (London) 318, 618-624. 13. Dearry, A., Gingrich, J. A., Falardeau, P., Fremeau, R. T., Bates, M. D. & Caron, M. G. (1990) Nature (London) 347, 72-76. 14. Sunahara, R. K., Niznik, H. B., Weiner, D. M., Stormann, T. M., Brann, M. R., Kennedy, J. L., Gelernter, J. E., Rozmahel, R., Yang, Y., Israel, Y., Seeman, P. & O'Dowd, B. F. (1990) Nature (London) 347, 80-83. 15. Zhou, Q.-Y., Grandy, D., Thambi, L., Kushner, J. A., Van Tol, H. H. M., Cone, R., Pribnow, D., Salon, J., Bunzow, J. R. & Civelli, 0. (1990) Nature (London) 347, 76-80. 16. Kobilka, B. K., Frielle, T., Collins, S., Yang-Feng, T., Kobilka, T. S., Francke, U., Lefkowitz, R. J. & Caron, M. G. (1987) Nature (London) 329, 75-79.
Proc. Natl. Acad. Sci. USA 88 (1991)
8115
17. Fargin, A., Raymond, J. R., Lohse, M. J., Kobilka, B. K., Caron, M. G. & Lefkowitz, R. J. (1988) Nature (London) 335, 358-360. 18. Julius, D., MacDermott, A. B., Axel, R. & Jessell, T. M. (1988) Science 241, 558-564. 19. Julius, D., Huang, K. N., Livelli, T. J., Axel, R. & Jessell, T. (1990) Proc. Natl. Acad. Sci. USA 87, 928-932. 20. Cotecchia, S., Schwinn, D. A., Randall, R. R., Lefkowitz, R. J., Caron, M. G. & Kobilka, B. K. (1988) Proc. Natl. Acad. Sci. USA 85, 7159-7163. 21. Kobilka, B. K., Matsui, H., Kobilka, T. S., Yang-Feng, T. L., Francke, U., Caron, M. G., Lefkowitz, R. J. & Regan, J. W. (1987) Science 238, 650-656. 22. Regan, J. W., Kobilka, T. S., Yang-Feng, T. L., Caron, M. G., Lefkowitz, R. J. & Kobilka, B. K. (1988) Proc. Natl. Acad. Sci. USA 85, 6301-6305. 23. Frielle, T., Collins, S., Daniel, K. W., Caron, M. G., Lefkowitz, R. J. & Kobilka, B. K. (1987) Proc. Natl. Acad. Sci. USA 84, 7920-7924. 24. Kobilka, B. K., Dixon, R. A. F., Frielle, T., Dohlman, H. G., Bolanowski, M. A., Sigal, I. S., Yang-Feng, T. L., Francke, U., Caron, M. C. & Lefkowitz, R. J. (1987) Proc. Natl. Acad. Sci. USA 84, 46-50. 25. Kubo, T., Fukuda, K., Mikami, A., Maeda, A., Takahashi, H., Mishina, M., Haga, T., Haga, K., Ichiyama, A., Kangawa, K., Kojima, M., Matsuo, H., Hirose, T. & Numa, S. (1986) Nature (London) 323, 411-416. 26. Peralta, E. G., Winslow, J. W., Peterson, G. L., Smith, D. H., Ashkenazi, A., Ramachandran, J., Schimerlik, M. I. & Capon, D. J. (1987) Science 236, 600-605. 27. Bonner, T. I., Buckley, N. J., Young, A. C. & Brann, M. R. (1987) Science 237, 527-532. 28. Dixon, R. A. F., Sigal, I. S., Rands, E., Register, R. B., Candelore, M. R., Blake, A. D. & Strader, C. D. (1987) Nature (London) 326, 73-77. 29. Strader, C. D., Sigal, I. S., Register, R. B., Candelore, M. R., Rands, E. & Dixon, R. A. F. (1987) Proc. Nati. Acad. Sci. USA 84, 4384-4388. 30. Fraser, C. M., Wang, C.-D., Robinson, D. A., Gocayne, J. D. & Venter, J. C. (1989) Mol. Pharmacol. 36, 840-847. 31. Strader, C. D., Sigal, I. S., Candelore, M. R., Rands, E., Hill, W. S. & Dixon, R. A. F. (1988) J. Biol. Chem. 263, 10267-10271. 32. Strader, C., Candelore, M. R., Hill, W. S., Sigal, I. S. & Dixon, R. A. F. (1989) J. Biol. Chem. 264, 13572-13578. 33. Sunahara, R. K., Guan, H.-C., O'Dowd, B. F., Seeman, P., Laurier, L. G., Ng, G., George, S. R., Torchia, J., Van Tol, H. H. M. & Niznik, H. B. (1991) Nature (London) 350, 614-619. 34. Needleman, S. B. & Wunsch, C. D. (1970) J. Mol. Biol. 48, 443453. 35. Devereux, J., Haeberli, P. & Smithies, 0. (1984) Nucleic Acids Res. 12, 387-395. 36. Kyte, J. & Doolittle, R. F. (1982) J. Mol. Biol. 157, 105-132. 37. Hopp, T. P. & Woods, K. R. (1981) Proc. Natl. Acad. Sci. USA 78, 3824-3828. 38. Weiner, S. J., Kollman, P. A., Nguyen, D. T. & Case, D. A. (Y186) J. Comp. Chem. 7, 230-252. 39. Ferrin, T. E., Huang, C. C., Jarvis, L. E. & Langridge, R. (1988)J. Mol. Graphics 6, 13-27. 40. Connolly, M. L. (1983) Science 221, 709-713. 41. Chou, P. Y. & Fasman, G. D. (1978) Annu. Rev. Biochem. 47, 251-276. 42. Kabsch, W. & Sander, C. (1983) FEBS Lett. 155, 179-182. 43. Benchetrit, T., Bissery, V., Mornon, J. P., Devault, A., Crine, P. & Roques, B. P. (1988) Biochemistry 27, 592-596. 44. Quian, N. & Sejnowski, T. J. (1988) J. Mol. Biol. 202, 865-884. 45. Bohr, H., Bohr, J., Brunak, S., Cotterill, R. M., Lautrup, B., Noerskov, L., Olsen, 0. H. & Petersen, S. B. (1988) FEBS Lett. 241, 223-228. 46. Eidne, K. A., Taylor, P. L., Zabavnik, J., Saunders, P. T. K. & Inglis, J. D. (1989) Nature (London) 342, 865. 47. Giros, B., Sokoloff, P., Martres, M.-P., Riou, J.-F., Emorine, L. J. & Schwartz, J.-C. (1989) Nature (London) 342, 923-926. 48. Monsma, F. J., McVittie, L. D., Gerfen, C. F., Mahan, L. C. & Sibley, D. R. (1989) Nature (London) 342, 926-929. 49. Wang, H., Lipfert, L., Malbon, C. C. & Bahouth, S. (1989) J. Biol. Chem. 264, 14424-14431. 50. Nilsson, I. & von Heijne, G. (1990) Cell 62, 1135-1141. 51. Brown, G. C. (1990) FEBS Lett. 260, 1-5. 52. Tedesco, J. L., Seeman, P. 5. & McDermed, J. D. (1979) Mol. Pharmacol. 16, 369-381. 53. Burgen, A. S. V., Roberts, G. C. K. & Feeney, J. (1975) Nature (London) 253, 753-755.
