Structural determinants of odorant recognition by the human olfactory receptors OR1A1 and OR1A2

Journal of

www.elsevier.com/locate/yjsbi

Journal of Structural Biology 159 (2007) 400–412

StructuralBiology

Structural determinants of odorant recognition by the humanolfactory receptors OR1A1 and OR1A2

Kristin Schmiedeberg a, Elena Shirokova a, Hans-Peter Weber b, Boris Schilling b,Wolfgang Meyerhof a, Dietmar Krautwurst a,*

a German Institute of Human Nutrition, Potsdam-Rehbruecke, Department of Molecular Genetics,

Arthur-Scheunert-Allee 114-116, 14558 Nuthetal, Germanyb Givaudan Schweiz AG, Fragrance Research – Bioscience, Ueberlandstrasse 138, 8600 Dubendorf, Switzerland

Received 19 January 2007; received in revised form 20 April 2007; accepted 23 April 2007Available online 25 May 2007

Abstract

An interaction of odorants with olfactory receptors is thought to be the initial step in odorant detection. However, ligands have beenreported for only 6 out of 380 human olfactory receptors, with their structural determinants of odorant recognition just beginning toemerge. Guided by the notion that amino acid positions that interact with specific odorants would be conserved in orthologs, but variablein paralogs, and based on the prediction of a set of 22 of such amino acid positions, we have combined site-directed mutagenesis, rho-dopsin-based homology modelling, and functional expression in HeLa/Olf cells of receptors OR1A1 and OR1A2. We found that (i) theirodorant profiles are centred around citronellic terpenoid structures, (ii) two evolutionary conserved amino acid residues in transmem-brane domain 3 are necessary for the responsiveness of OR1A1 and the mouse ortholog Olfr43 to (S)-(�)-citronellol, (iii) changes atthese two positions are sufficient to account for the differential (S)-(�)-citronellol responsiveness of the paralogs OR1A1 andOR1A2, and (iv) the interaction sites for (S)-(�)-citronellal and (S)-(�)-citronellol differ in both human receptors. Our results show thatthe orientation of odorants within a homology modelling-derived binding pocket of olfactory receptor orthologs is defined by evolution-ary conserved amino acid positions.� 2007 Published by Elsevier Inc.

Keywords: Olfactory receptors; Human; Orthologs; Mutation; Docking model; Odorant

1. Introduction

Olfactory receptors in humans are encoded by about380 genes (Malnic et al., 2004; Niimura and Nei, 2003;Zozulya et al., 2001). This olfactory receptor (OR) reper-toire, is a bio-molecular interface between the chemicaloutside world and the sensory neurons of the olfactoryepithelium (OE), and enables humans to detect, discrim-inate, categorize, and qualitatively and quantitativelyevaluate a multitude of chemically diverse odorants.For example, about 8000 volatiles have been identifiedjust in food (Grosch, 2001). While each OR may recog-

1047-8477/$ - see front matter � 2007 Published by Elsevier Inc.

doi:10.1016/j.jsb.2007.04.013

* Corresponding author. Fax: +49 0 33200 88384.E-mail address: [email protected] (D. Krautwurst).

nize several odorants (Malnic et al., 1999), their ligandspecificity is, nevertheless, defined by efficacy-rankingodorant profiles (Katada et al., 2005; Shirokova et al.,2005).

If the structure of an OR determines its function,which then are the amino acid positions within OR thatodorants interact with, and how does that translate intodifferences in their odorant profiles? Over the last decade,several groups have employed computational methods,based on a structure of rhodopsin (Palczewski et al.,2000), on the few OR with known ligands, to predict sin-gle amino acids that may be involved in odorant interac-tion (Afshar et al., 1998; Araneda et al., 2004; Florianoet al., 2000; Hall et al., 2004; Singer, 2000; Singer et al.,1996; Vaidehi et al., 2002); for review see Lai et al.

mailto:[email protected]

K. Schmiedeberg et al. / Journal of Structural Biology 159 (2007) 400–412 401

(2005), Man et al. (2004), Olender et al. (2004). Ratherthan single amino acid positions, one study proposed86 amino acid motives within the superfamily of ORthat, in a combinatorial way, may enable differentialodorant detection (Liu et al., 2003). In contrast, in acomprehensive study, based on ortholog/paralog compar-isons among human and mouse OR. Man et al. (2004)predicted a set of 22 evolutionary conserved and putativeodorant-interacting amino acid positions. Recently, 8 ofthese 22 amino acid positions were, however, identifiedto overlap with the mapped interaction sites for two sin-gle odorants within two mouse OR (Abaffy et al., 2006;Katada et al., 2005). A first mutational study to correlatesingle amino acid differences within OR-I7 orthologsfrom mouse and rat with differences in their preferencetowards an activation by heptanal or octanal wasdescribed earlier (Krautwurst et al., 1998). Two hundredhuman OR display >80% amino acid identity withmouse OR (Zhang and Firestein, 2002), and, thus, arepotentially encoded by orthologous OR genes. For exam-ple, OR1A1 is the human ortholog to mouse Olfr43(Glusman et al., 2000; Lapidot et al., 2001), for whichwe have recently described an odorant response pattern(Shirokova et al., 2005).

Here, we have de-orphaned OR1A1 and its paralogOR1A2 in HeLa/Olf cells. By rhodopsin-based homologymodelling, site-directed mutagenesis, and functionalexpression of wild-type and mutant OR, we identified evo-lutionary conserved amino acid residues to be necessaryand sufficient for the specific responsiveness of OR1A1and OR1A2 to their odorants.

2. Methods

2.1. Molecular cloning

We amplified the full-length coding regions of OR1A1(NM_014565), and OR1A2 (NM_012352) from human(HeLa) genomic DNA by polymerase chain reaction(PCR) with Pfu (Promega), or PfuUltra (Stratagene) usinggene-specific primers (OR1A1: CAGGCAATTGATGAGGGAAAATAACCAGTCCTCTAC; CACTAGCGGCCGCTTACGAGGAGATTCTCTTGTTGAAGAG; OR1A2:GTCAGAATTCATGAAGAAAGAAAATCAATCCTTTAACCTG; GTCTAGCGGCCGCCTATGAGGAGATTCTCTTGCTG). Amplicons were subcloned EcoRI/NotIor MfeI/NotI into the expression vector pi2-dk(rt39),which provides the first 39 amino acids of the bovine rho-dopsin (rho-tag(39)) as a N-terminal tag for all full-lengthOR. Amino acid mutations in OR were achieved by thePCR-based QuickChange method (Stratagene). The identi-ties of all subcloned wild-type (wt) and mutated OR codingregion amplicons were verified by sequencing (MWG,Ebersberg, and UKEHH, Hamburg). Cell culture andtransient DNA transfection were performed as reportedpreviously (Shirokova et al., 2005).

2.2. Reverse transcriptase (RT)-PCR

Total RNA was prepared from a surgical biopsy ofhuman main olfactory epithelium, taken from the dorsalregion of the superior nasal concha, using Trizol (Invitro-gen). Total RNA was treated with DNaseI (Invitrogen),and 50 ng served as template for each reverse transcrip-tase-PCR (One Step RT-PCR, Quiagen), using gene-spe-cific primers for OR1A1 (CATTGTCCTAGCCATTTGCTCTGATG; CTTGAGCACGCCCTTGGTGGAAG)and OR1A2 (CATCTTGGCCATCTGTGCTGACATTC;CTTTGAATAGACTCTTGGTAGATGG).

2.3. Ca2+-FLIPR assay

FLIPR assays (Molecular Devices) and data analyseswere performed as described previously (Shirokova et al.,2005). In short, cells were loaded with 4 lM FLUO-4/AM and 0.04% Pluronic F-127 (both Molecular Probes)in Hepes-buffered saline (HBS) with 20 mM 4-(2-Hydroxy-ethyl)piperazine-1-ethanesulfonic acid (Hepes) and 2.5 mMprobenecid. After loading, cells were washed twice withHBS by an automated plate washer (Denley Cellwash,Labsystems) and transferred to the FLIPR. EC50 valuesand curves were derived from fitting the functionf(x) = (a�d)/(1+(x/C)nH) + d to the data by nonlinearregression with a, minimum; d, maximum; C, EC50, andnH, Hill coefficient.

2.4. [35S]GTPcS binding assay

[35S]GTPcS binding assays were performed as describedpreviously (Shirokova et al., 2005) with modifications thatinclude a G-protein immunoprecipitation step (Milligan,2003). HeLa/Olf cells were grown in a 10-cm dish andtransfected with 6 lg of receptor DNA, or empty vector,using PolyFect (Quiagen). For the binding reaction, cellmembrane protein (�50 lg) was incubated for 15 min at37 �C in a binding buffer (10 mM Hepes, pH 7.4, 3 mMMgCl2, and 50 mM NaCl) that included 0.05 nM[35S]GTPcS and 3 lM GDP in a total volume of 100 ll.Basal condition was determined in the absence of agonist.Parallel assays containing unlabeled GTPcS (10 lM) wereused to define non-specific binding. The reactions werestopped by addition of ice-cold binding buffer and centri-fuged at 16,000g at 4 �C for 30 min to pellet the protein.After centrifugation, the pelleted protein was resuspendedin solubilization buffer (10 mM Hepes, pH 7.4, 50 mMNaCl, 5 mM ethylenediaminetetraacetic acid [EDTA],and 0.5% Triton X) that included 1 lg anti-Gas/aolf (rab-bit, Calbiochem), and 20 ll protein A-conjugated agarosebeads, and rotated overnight at 4 �C. The beads werewashed three times with solubilization buffer, and thebound radioactivity was measured in a liquid scintillationcounter. Under these conditions, non-specific binding wastypically <10% of the total binding. The non-specific bind-ing was substracted, and the basal value was set at 100%.

402 K. Schmiedeberg et al. / Journal of Structural Biology 159 (2007) 400–412

Each data point within an experiment was determined fromduplicate samples.

2.5. Alignment of OR1A1 and OR1A2 to bovine rhodopsin

An alignment of the sequences of OR1A1 and OR1A2to the sequence of the G protein-coupled receptor (GPCR)bovine rhodopsin is the prime requisite for homology mod-elling. Since the sequence identity of OR1A1, and OR1A2,to bovine rhodopsin is as low as 14.1%, and 16.9%, respec-tively, a simple direct alignment procedure is not applica-ble. However, OR1A1 and OR1A2 show 50% amino acididentity, and about 75% functionally conserved homologyto OR1E1, one typical receptor out of the multiple ORsequence alignment with bovine rhodopsin by Man et al.(2004), which we used as a benchmark.

In this section, as in the figures showing the computedOR models with the docked ligand, the numbering ofamino acids in OR1A1 and OR1A2 follows the numberingof bovine rhodopsin in the alignment. The alignment forTM2-7 is given in Fig. 4.

2.6. Generation of 3D models for OR1A1 and OR1A2

The 3D homology models of the two OR were based onthe X-ray structure of bovine rhodopsin (PDB-code 1F88)using the modelling software MOLOC (Gerber, 1998; Krat-ochwil et al., 2005). Several rounds of visual inspection andmanual correction of side chains with sterically bad contacts,followed by computational energy minimisations, lead to thefinal model of the two OR. During the energy minimisations,the Ca positions in the TM helices, and the Ca of C179 (C187in the alignment) in EC2 (which makes the universally con-served S–S bridge to C97 [C109 in the alignment] in TM3),were kept fixed at positions as in rhodopsin (1F88). In thefinal energy minimisation all positional constraints were

Fig. 1. Odorant receptor genes OR1A1 and OR1A2 are expressed in thehuman olfactory epithelium. (a) Schematic display of the synthenic regionon mouse chromosome (MC) 11 and human chromosome (HC) 17containing OR genes Olfr43, Olfr403, OR1A1, and OR1A2. Relationshipsand amino acid identities of the olfactory receptor gene products are givenby dashed lines and numbers, respectively. Drawn to scale, withmodifications after (Lapidot et al., 2001). (b) Detection of OR1A1 andOR1A2 mRNA from a human biopsy. 1A1: (OR1A1, 574 bp), 1A2,(OR1A2, 572 bp), single tube RT-PCR products using gene-specificprimers and human total RNA. RT, omitting reverse transcriptase.Marker (M) sizes are given in base pairs.

removed, which resulted in a final model with only very smalldeviations of the Ca-positions of the TMs and of C187 fromthe positions in 1F88. The extracellular and intracellularloops, which have many insertions and deletions compared

Fig. 2. Odorant specificity patterns of OR1A1 and OR1A2. HeLa/Olfcells transfected with DNA for rho-tag(39)-OR1A1 (a) rho-tag(39)-OR1A2 (b) or mock transfected (c) and tested with 30 lM of 94 differentodorants in FLIPR experiments. All odorants and their coordinates in the96-well format are listed in Supplementary Fig. 2. , overlappingresponses in OR1A1 and OR1A2: (S)-(�)-limonene (A8), octanol (B4),(S)-(�)- and (R)-(+)-citronellal (C9, C10), E,E-2,4-decadienal (D4),geraniol (D6), Z-4-decenal (D9), Z-7-decanal (D10), nerolidol (E1), E-4-decanal (E4), nerol (F1), 1-octen-3-one (F5), helional (G3), nonanal (G4),decanal (G5), (�)-carveol mix.(G7), heptanal (G8), octanal (H6). D,differential responses to (S)-(�)-citronellol (C6), a-ionone (D7), a-ionone(E2), and trimethylamine (F8). Note: Calcium signals in (a and b) with anonset not later then 2 min after odorant application are consideredresponses. Signals in A7, E7, G6, and G9 also appear in mock-transfectedcells, and do therefore not depend on the expression of OR. Time scale,3 min each coordinate.


to 1F88, were left completely free from the beginning (withexception of C187 in ECL2) and were refined into stericallyreasonable but hypothetical positions. Since they are not indirect contact with the putative ligand binding pocket(LBP) of olfactory agents, their positions are considered oflow relevance in this context.

2.7. Docking of (S)-(�)-citronellal and (S)-(�)-citronellol

into the LBP of OR1A1 and OR1A2

The docking procedure of MOLOC was used to explorepossible binding modes. Odorant docking was completely

Table 1EC50-ranking odorant profiles of OR1A1 and OR1A2

Odorant Structure EC50-values (lM)

OR1A1 OR1A2

(S)-(�)-citronellal 2.2 ± 0.4 2.4 ± 0.7

Helional 2.6 ± 0.3 3.4 ± 1.1

Hexanal a a

Heptanal 3.2 ± 0.2 1.7 ± 0.9

Octanal 3.9 ± 1.9 4.3 ± 1.3

Nonanal 55.5 ± 12.2 127.4 ± 30.7

Hydroxy-citronellal 5.61 ± 0.6 3.9 ± 0.4

Citral 16.1 ± 3.1 14.7 ± 1.9

4-Decenal 25.1 ± 0.3 16.5 ± 3.8

2,4-Decadienal a a

(�)-Carveol(mix. of isomers)

16.1 ± 1.9 16.7 ± 5.2

geraniol >0.1* >0.1*

Octanol 38.2 ± 13 60.4 ± 15

(S)-(�)-citronellol 92.6 ± 10.2 a

(R)-(+)-citronellol 82.5 ± 6.3 81.1 ± 6.2

Citronellic acid a a

Citralva a a

Data are given as single concentrations or EC50 values as means ± SD,from 2 to 5 independent experiments. a no effect up to 300 lM.

* Effects starting from 0.03 lM, however a fit-function-derived EC50 wasnot calculated due to the lack of saturation, and OR-independent effects atconcentrations higher than 10–30 lM.

free, and not instructed by retinal coordinates in 1F88.The docking procedure started by assigning a random—but stereo-chemically reasonable—conformation to a givenligand, then positioning the centre of mass of the ligand ina random orientation onto a defined point in the cavity,and checking for unacceptable atom overlap betweenligand and receptor. An optimization of the flexible ligandin the fixed cavity was then computed, and the goodness-of-fit was retained. A large number of orientations weretried out this way, and the best orientations were retainedand stored. The procedure was repeated for many differentstarting conformations of the ligand, so that a representa-tive—but not an exhaustive—part of the degrees-of-free-dom-space is sampled. The 20 top-scoring dockings foreach ligand were finally energy optimized to convergencein the fixed receptor, OR1A1 and OR1A2, and rankedaccording to the ‘‘enthalpic binding energy’’ (DEinter),

Fig. 3. Similar and differential responses of OR orthologs and paralogs to(S)-(�)-citronellal and (S)-(�)-citronellol. Concentration-response rela-tions of (S)-(�)-citronellal (filled circles) and (S)-(�)-citronellol (opencircles) applied to HeLa/Olf cells transfected with DNA coding for (a)rho-tag(39)-Olfr43, (b) rho-tag(39)-OR1A1, and (c) rho-tag(39)-OR1A2.Data are means (SD < 10%) from a representative experiment performedin quadruplicates. Insets show the original Ca2+-fluorescence time coursesfrom FLIPR experiments. Scale: Vertical scale, 0.2 DF/F; horizontal scale,1 min. Similar results were obtained for each receptor in at least twoindependent experiments.


defined as the sum of van der Waal’s (vdW) and hydrogenbond (HB) interaction energies between ligand and recep-tor. This energy is corrected by a ‘‘ligand stress energy’’which is the difference between the enthalpic conforma-tional energy of the ligand bound in the receptor and theenergy of the ligand relaxed in vacuum, yielding the cor-rected enthalpic binding energy, DEinter, corr.

3. Results

3.1. Orthologous genes Olfr43 and OR1A1 share a function

The closest homologs to mouse OR genes Olfr43 andOlfr403 are the human OR genes, OR1A1 and its paralogOR1A2 (Fig. 1a) (Lapidot et al., 2001). Both genes arelocated in a well characterized syntenic region on humanchromosome 17(p13.3), with OR1A1 as the consideredortholog of Olfr43 (Lapidot et al., 2001). We identifiedtranscripts of both genes, OR1A1 and OR1A2, by RT-PCR experiments with RNA from human OE biopsy mate-rial (Fig. 1b). We recently identified mouse OR Olfr43 as areceptor for citronellic terpenoid structures, such as (S)-(�)-citronellal, and related odorants, shaping its EC50-

Fig. 4. A generalized OR binding pocket superimposed on the rhodopsin-aligncircles) with differences to OR1A2 marked in grey. Putative odorant interactionOR1A1 and OR1A2 at the green positions are marked red with the amino acTM1-7, putative transmembrane domains. (b) Sequence alignment of rhodointeraction sites from (a) are marked in green, and differences at these positionsboxed positions in (a) and (b) were subjected to mutation analysis. Numbers areOR amino acid sequence).

ranking odorant profile (Shirokova et al., 2005). Since trueorthologous genes are expected to share a function, weinvestigated whether the human receptors OR1A1 andOR1A2 have odorant profiles similar to that of Olfr43.We used HeLa/Olf cells (Shirokova et al., 2005) for thefunctional expression of OR. These cells express the humanG protein aolf, adenylyl cyclase type-III, and the cyclicnucleotide-gated, Ca2+-permeable, homomeric CNGA2channel. In HeLa/Olf cells transfected with DNA codingfor an OR, a stimulation with its odorant will result inOR-dependent [35S]GTPcS binding, cAMP production,and Ca2+ influx (Shirokova et al., 2005). When extendedby the N-terminal part of rhodopsin (rho-tag(39)) (Shirok-ova et al., 2005), the plasma membrane expression of bothreceptors, OR1A1 and OR1A2 could be confirmed byimmunocytochemistry (Supplementary Fig. 1). First, wecompared their odorant specificity, by screening rho-tag(39)-OR1A1 and rho-tag(39)-OR1A2 against 94 odor-ants (Supplementary Fig. 2) at 30 lM, using FLIPRCa2+-imaging (Fig. 2). At this concentration, fourodorants, muscone, farnesol, farnesene, and bourgeonalshowed OR-independent effects in mock-transfected cells(Fig. 2c). An overlapping set of 18 odorants activated both

ed OR1A1 and OR1A2 sequences. (a) A snake diagram of OR1A1 (whitesites proposed by Man et al. (2004) are given in green. Differences between

id sequence numbering. Letters show conserved positions among GPCRs.psin (Rho), Olfr43, OR1A1, and OR1A2 for TM2-7. Putative odorantbetween OR1A1 and OR1A2 are boxed in red. Purple and red circled andaccording to the rhodopsin alignment (numbers in parentheses refer to the

Table 2EC50 values of (S)-(�)-citronellol for OR1A1, OR1A2, and mutantreceptors

Receptors,mutations

Position in therhodopsinalignment

Trans-membraneregion

EC50 value(lM)

OR1A1 wt 92.6 ± 10.2 (5)OR1A2 wt a (2)OR1A1T99M 112 3 98.7 ± 6.4 (2)OR1A1G108A 121 3 a (2)OR1A1N109K 122 3 a (2)OR1A1G108A,N109K 121, 122 3 a (2)OR1A1T110A 123 3 68.7 ± 13.1 (2)OR1A1I114T 127 3 84.6 ± 18.3 (2)OR1A2A108G 121 3 122 ± 16.5 (3)OR1A2K109N 122 3 94 ± 15.6 (3)OR1A2A108G,K109N 121, 122 3 99.4 ± 7.9 (2)OR1A1A154T,N155S 166, 167 4 103 ± 14 (3)OR1A2T154A,S155N 166, 167 4 a (2)OR1A1I205V 211 5 110 ± 13.6 (3)OR1A2V205I 211 5 a (3)OR1A1V254T 264 6 85 ± 15.8 (3)OR1A2T254V 264 6 a (3)OR1A1T277V 295 7 101.5 ± 3.5 (2)OR1A2V277T 295 7 a (2)

Data are given as means ± SD with numbers of experiments inparentheses.

a No effect up to 300 lM. wt, wild-type receptor, no mutations.


rho-tag(39)-OR1A1- and rho-tag(39)-OR1A2-transfectedHeLa/Olf cells (Fig. 2a and b), but not mock-transfectedcells (Fig. 2c). Rho-tag(39)-OR1A1 responded selectivelyto (S)-(�)-citronellol (Fig. 2a, coordinate C6), whereasrho-tag(39)-OR1A2 selectively responded to a- and b-ionone, as well as to trimethylamine (Fig. 2a, coordinatesD7, E2, F8). Comparing the odorant response patterns ofboth orthologs OR1A1 (Fig. 2a) and Olfr43 (Shirokovaet al., 2005) at 30 lM, we found an overlapping set of 9activating odorants, including (S)-(�)-citronellal, (R)-(+)-citronellal, geraniol, Z-4-decenal, Z-7-decenal, E-4-decenal,nerol, helional, and octanal.

We further tested citronellic odorants and aliphaticswith varying carbon chain lengths or functional groups(Table 1). Citronellic terpenoid and aliphatic aldehydes ofa carbon chain length C7–C8 emerged as best ligands.Alcohols of the same length were about 10 times less effi-cient, with the exception of geraniol, which was an agonistat sub-micromolar concentrations, but did not saturate dueto OR-independent signals at concentrations higher than10–30 lM (Table 1). The odorants tested activated rho-tag(39)-OR1A1- or rho-tag(39)-OR1A2 with similar effica-cies, with the exception of (S)-(�)-citronellol, which selec-tively activated only rho-tag(39)-OR1A1 (Table 1;Fig. 2a, C6). (S)-(�)-citronellal activated rho-tag(39)-OR1A1, -OR1A2, and -Olfr43 (2.1 ± 0.2 lM; n = 2) withsimilar efficacies around 2 lM (Fig. 3, Table 1). In FLIPRCa2+-imaging experiments (S)-(�)-citronellol activatedboth orthologs Olfr43 and OR1A1 with EC50 values of113.4 ± 6.4 lM (n = 2), and 92.6 ± 10.2 lM (n = 5),respectively (Fig. 3a and b), but did not activate the humanparalog OR1A2 up to 300 lM (Fig. 3c; Table 1). In con-trast, the stereo-enantiomer (R)-(+)-citronellol activatedall three OR, rho-tag(39)-OR1A1, -OR1A2, and -Olfr43,with EC50 values of 82.5 ± 6.3 lM, 81.1 ± 6.2 lM, and58.7 ± 1.8 lM, respectively (n = 2; Supplementary Table1, Supplementary Fig. 4). Citronellic acid did not elicitresponses up to 300 lM in HeLa/Olf cells that were trans-fected with either of the rho-tag(39)-OR (Fig. 2a and b,B12; Table 1).

3.2. Single amino acids determine the differential (S)-(�)-

citronellol responsiveness of olfactory receptor paralogs

OR1A1 and OR1A2

We tested the hypothesis that single evolutionary con-served amino acids confer similar responsiveness towards(S)-(�)-citronellol in the orthologs OR1A1 and Olfr43.Changes in one or few of these positions in the highlyhomologous OR1A2 may then be responsible for its lackof responsiveness for (S)-(�)-citronellol. To narrow theputative (S)-(�)-citronellol-interacting amino acids downto a manageable number for site-directed mutagenesis, wemapped out a two-step strategy: First, from a predictedset of 22 evolutionary conserved and putative odorant-interacting amino acid positions (Man et al., 2004)(Fig. 4a, green positions), we superimposed 20, excluding

2 positions in the second extracellular loop, onto the rho-dopsin-aligned amino acid transmembrane sequences ofOR1A1, OR1A2, and Olfr43 (Fig. 4b, green positions).Second, of these 20 amino acid positions, we then identifiedfive amino acids which differ between the paralogs OR1A1and OR1A2, but are identical, all but one, between theorthologs OR1A1 and Olfr43 (Fig. 4b, positions boxed inred). We exchanged these five amino acids in OR1A1,one at a time, to the respective amino acids in OR1A2,and vice versa (Table 2). Two mutations at amino acidpositions 108 and 109 emerged (corresponding to positions121, 122 in the alignment), that, when combined, conferreda complete loss of function in OR1A1, and a gain of func-tion in OR1A2, towards (S)-(�)-citronellol, in both,[35S]GTPcS binding and FLIPR Ca2+-imaging assays(Fig. 5, Table 2). Despite the loss of (S)-(�)-citronellolfunction in the double mutant rho-tag(39)-OR1A1G108A,

N109K, it still responded to (S)-(�)-citronellal with anEC50 of 4.5 lM, compared to 2.2 lM of OR1A1wt(Fig. 4b). Plasma membrane expression of the mutantOR did not differ from that of the wild-type receptors (Sup-plementary Fig. 1). Changing the other three of the fivepredicted odorant-interacting amino acid positions inTM5, TM6, and TM7, from OR1A1 to OR1A2, and viceversa, did not alter the responsiveness towards (S)-(�)-cit-ronellol for either mutant OR (Table 2). We exchangedthree additional amino acids at positions 99, 110, and114 in TM3 of OR1A1 that differ from the respectiveamino acids in OR1A2 (Fig. 4b, boxed in purple), butobserved no loss of function when testing the mutant recep-tors in the FLIPR assay (Table 2). We further exchanged

a

b

c

Fig. 5. Amino acids at positions 108, 109 are responsible for the loss of (S)-(�)-citronellol function of OR1A2. (a and b), [35S]GTPcS binding inmembrane preparations of HeLa/Olf cells transfected with DNA coding for rho-tag(39)-OR1A1 and mutants (left panels), or rho-tag(39)-OR1A2 andmutants (right panels), and induced with (S)-(�)-citronellal (a) or (S)-(�)-citronellol (b). [35S]GTPcS binding was determined before (open bars), and afterimmunprecipitation with an anti-Gas/aolf antibody (filled bars). Basal [35S]GTPcS binding was set to 100% (dotted line). (c) Concentration–responserelations of (S)-(�)-citronellal (circles), and (S)-(�)-citronellol (triangles) on HeLa/Olf cells transfected with DNA coding for rho-tag(39)-OR1A1, or rho-tag(39)-OR1A2 (filled symbols), and the respective double-mutants at positions 108, 109 (open symbols). Data are given as means ± SD from 2 to 3independent experiments. EC50 values are given in the text and Table 2.


from OR1A1 to OR1A2, and vice versa, two differingamino acids at positions 154 and 155 in TM4 (Fig. 4b,boxed in purple), which had been postulated previouslyto interact with odorants (Pilpel and Lancet, 1999). Theseamino acids, when mutated either singularly, together, ortogether with amino acids 108, 109, however, did not alterthe responsiveness of mutant rho-tag(39)-OR1A1 or rho-tag(39)-OR1A2 towards (S)-(�)-citronellol (Table 2, Sup-plementary Table 1).

3.3. Modelling the ligand binding pocket (LBP) of OR1A1

and OR1A2

We assumed that the space occupied by the cis-retinoidin bovine rhodopsin, which is covalently bound as Schiff

base to L296 (Khorana, 1992; Wald, 1968), is representa-tive for the LBP of odorants in OR. In our model, the cor-responding cavity in OR1A1, or OR1A2, is ligned by 17TM-amino acids, of which 11 make hydrophobic interac-tions with the odorants (Table 3). The lipophilic part ofboth ligands, (S)-(�)-citronellol and (S)-(�)-citronellal,occupy practically identical positions in both receptors,OR1A1 and OR1A2. In the set of amino acids lining thispart of the LBP, there are only two sterically minor differ-ences between OR1A1 and OR1A2: T277V and G108A(see Table 3). T277V has practically no energetic conse-quences for the interaction with the ligands. When testedin HeLa/Olf cells, mutant rho-tag(39)-OR1A1T277V stillresponded to (S)-(�)-citronellol with a similar EC50 asthe wild-type OR1A1. The second difference, G108A, has

Table 3Amino acids lining the odorant binding pocket of OR1A1 and OR1A2

Receptordomains

Docked ligands Position inprimary sequence

Position in therhod-opsin alignment

Position predicted byMan et al. (2004)(S)-(�)-citronellal (S)-(�)-citronellol

OR1A1 OR1A2 OR1A1 OR1A2

TM2 F F F F 73 86 +TM3 M M M M 104 117 +

I I I I 105 118 +G A G A 108 b 121 +N K N a K 109 b 122 +S a S S a S 112 b 125 +

TM4 N S a N S a 155 b 167 –TM5 L L L L 201 207 –

I V I a V 205 b 211 +F F F F 206 212 +

TM6 Y a Y Y Y 251 261 +M M M M 255 265 –Y Y Y Y 258 268 –

TM7 V V V V 274 292 +T V T V 277 b 295 +A A A A 278 278 –T T T T 280 280 –

ECL2 D D D D 180 188 –I I I I 181 189 +

Amino acid positions were derived from the rhodopsin homology model.a Amino acids that are involved in H-bonding to the oxo-function in (S)-(�)-citronellal and (S)-(�)-citronellol. Other amino acids listed make

hydrophobic interactions with the odorants.b Amino acid positions that were altered by mutations. Amino acid differences are in bold. TM, transmembrane domain, ECL, extracellular loop.


a steric effect for the OR–ligand interaction: in contrast toG108 in OR1A1, the methyl group of A108 in OR1A2pushes towards the middle of the C8 chain in both, (S)-(�)-citronellol or (S)-(�)-citronellal, with the effect ofchanging direction and position of the hydroxyl, or ofthe aldehyde moiety, respectively (Figs. 6 and 7). The lipo-philic (vdW) ligand–receptor interaction energy is littleaffected by differences at position 108 in OR1A1, orOR1A2. In line with that, point mutations at this positionlead to only a partial loss of function, or gain of function,for (S)-(�)-citronellol in [35S]GTPcS binding assays withmutant rho-tag(39)-OR1A1G108A, or rho-tag(39)-OR1A2A108G, respectively (Fig. 5a). Similarly, in FLIPRCa2+-imaging, we observed only a partial gain of functionwhen testing (S)-(�)-citronellol on rho-tag(39)-OR1A2A108G. For mutant rho-tag(39)-OR1A1G108A, how-ever, we observed a complete loss of function in FLIPRCa2+-imaging.

With respect to the ligand–receptor interaction energymore significant differences, however, exist in the LBP sur-rounding the hydrophilic part of the ligands—the hydroxylgroup in (S)-(�)-citronellol, and the aldehyde group in (S)-(�)-citronellal, i.e. N109 or K109, and N155 or S155, inOR1A1 or OR1A2, respectively. These differences lead todifferent hydrogen bond patterns for both odorants (S)-(�)-citronellol, and (S)-(�)-citronellal, in both, OR1A1and OR1A2, and hence variable contributions to the bind-ing energy (Supplementary Table 2). In particular, thehydroxyl of (S)-(�)-citronellol can make three hydrogenbonds with N109, S112, and I205, in OR1A1, but only

one hydrogen bond to S155 in OR1A2. All proposed spe-cific residue contacts were via side-chain atoms, with oneexception: a hydrogen bond is proposed between thehydroxyl of (S)-(�)-citronellol and the C@O group of theI205 backbone (Fig. 6c). And while the aldehyde oxygenof (S)-(�)-citronellal supposedly forms hydrogen bondsto S112 and Y251 in OR1A1, our model suggests onlyone hydrogen bond to S155 in OR1A2. Changing aminoacid position 109, we observed only a partial gain of func-tion or loss of function for (S)-(�)-citronellol on mutantsrho-tag(39)-OR1A2K109N, or rho-tag(39)-OR1A1N109K,respectively, in [35S]GTPcS binding assays (Fig. 5). InFLIPR Ca2+-imaging, however, we observed a completegain of function or loss of function for (S)-(�)-citronellolon mutants rho-tag(39)-OR1A2K109N, or rho-tag(39)-OR1A1N109K, respectively (Table 2). The overall differ-ences in hydrogen bonding of (S)-(�)-citronellol toOR1A1, or to OR1A2, resulted in distinctly differentDEinter,corr, i.e. of �22.6, and of �15.2 kcal/mol, respec-tively (see Supplementary Table 2). Accordingly, ourexperimental results showed, that (S)-(�)-citronellol isresponsive in rho-tag(39)-OR1A1, and the double mutantrho-tag(39)-OR1A2A108G,K109N, but non-responsive inrho-tag(39)-OR1A2, and the double mutant rho-tag(39)-OR1A1G108A,N109K, in both assays (Figs. 2, 3 and 5, Tables1 and 2).

In contrast to (S)-(�)-citronellol, the hydrogen bondingof (S)-(�)-citronellal is supposedly practically equal in bothOR1A1 and OR1A2, with DEinter,corr of –19.9 kcal/mol,and of �19.2 kcal/mol, respectively (see Supplementary

a

b

c

Fig. 6. 3D models of (S)-(�)-citronellol docked into the transmembrane regions of OR1A1 and OR1A2. (a and b) Stereo pictures of docking modelsshowing the hydrogen bond interactions (dashed lines) of (S)-(�)-citronellol with OR1A1 (a) and OR1A2 (b). a-Traces of the TM-helices are shown (TM1red, TM2 orange, TM3 yellow, TM4 green, TM5 lime, TM6 blue, TM7 purple), and the amino acids interacting with the hydroxyl function of (S)-(�)-citronellol (left-right stereo). Numbering is according to the rhodopsin alignment. The stereo pictures of the docking models for (S)-(�)-citronellal inOR1A1 and OR1A2 are given in Supplementary Fig. 3. (c) Schematic presentation of (S)-(�)-citronellol docked into the transmembrane helices TM3-TM6 of OR1A1 (left panel) and OR1A2 (right panel). Numbers refer to the OR amino acid sequence. Hydrogen bonds are indicated as dashed lines, vdWrepulsion and steric hindrance as arrows.


a

b

Fig. 7. Differential interaction of (S)-(�)-citronellal with single amino acid residues in OR1A1 or OR1A2. Schematic presentations (left panels) of (S)-(�)-citronellal and the proposed main interacting amino acid residues in the homology models of OR1A1 (a) and OR1A2 (b). Hydrogen bonds are indicated asdashed lines, vdW repulsion and steric hindrance as arrows. (a and b, right panels) Concentration-response relations of (S)-(�)-citronellal on rho-tag(39)-OR1A1wt, and rho-tag(39)-OR1A2wt (circles), and the respective mutant receptors rho-tag(39)-OR1A1S112A (triangles), and rho-tag(39)-OR1A2S155A

(squares). Data were calculated as means ± SD from 2 to 3 independent FLIPR experiments, each performed in triplicates. EC50 values in (a) are 2.7 lMand 2.5 lM, for rho-tag(39)-OR1A1, and rho-tag(39)-OR1A1S112A, respectively. Dashed lines, no fitting function due to the lack of saturation.


Table 2). In line with this, (S)-(�)-citronellal activated rho-tag(39)-OR1A1, rho-tag(39)-OR1A2, and the doublemutant rho-tag(39)-OR1A1G108A,N109K with similar EC50

values (Supplementary Fig. 2a,b Figs. 3 and 5, Tables 1and 2). Altogether, our model proposes 17 amino acidpositions lining the binding pockets of OR1A1 andOR1A2, of which four are involved in hydrogen-bondingto the hydroxyl- or aldehyde moieties of (S)-(�)-citronelloland (S)-(�)-citronellal, whereas thirteen make hydropho-bic interactions with the odorants (Table 3).

3.4. The ligand binding pocket model predicts differential

binding of (S)-(�)-citronellol and (S)-(�)-citronellal

within OR1A1 and OR1A2

The model of (S)-(�)-citronellol docked into OR1A1 orOR1A2 (Fig. 6a) interestingly predicts different favoriteinteraction partners for the hydroxyl group in each recep-tor. In the case of OR1A1, the hydroxyl moiety of (S)-(�)-citronellol is predicted to favorably interact withN109, and with lower energy to S112 within TM3, and toI205 in TM5 (Fig. 6). An involvement of position N109in the interaction with (S)-(�)-citronellol has been demon-strated (Fig. 5, Table 2).

The models of (S)-(�)-citronellal docked into OR1A1 orOR1A2 again predict different favorite interaction partners

for the aldehyde group, in each receptor. In the case ofOR1A1, the aldehyde function of (S)-(�)-citronellal is pre-dicted to favorably interact with S112 within TM3 (Fig. 7a,Supplementary Fig. 3a). We have mutated position S112into an alanine, and observed in rho-tag(39)-OR1A1S112A-transfected cells a 60% decrease in amplitudein response to (S)-(�)-citronellal, as compared to the wild-type receptor (Fig. 7a). In contrast to OR1A1, thepredicted favorite interaction partner in OR1A2 for thealdehyde group of (S)-(�)-citronellal appears to be S155in TM4 (Fig. 7b, Supplementary Fig. 3b). Changing thisposition into an alanine resulted in a marked decrease inthe responsiveness of rho-tag(39)-OR1A2S155A to (S)-(�)-citronellal (Fig. 7b). The plasma membrane expression ofthe mutant OR did not differ from that of the wild-typereceptors (Supplementary Fig. 1).

4. Discussion

The functional identification of odorant ligands (de-orphaning) for an OR precedes the analysis of its struc-ture–function relation. Several lines of evidence suggestthat in humans OR1A1 and OR1A2 are indeed involvedin the specific detection of certain citronellic terpenoidodorants. First, we detected transcripts of OR1A1 andOR1A2 in biopsies from human OE. Transcript levels in


the OE can differ between OR by up to 300-fold (Younget al., 2003). The difference in transcript detection forOR1A1 and OR1A2, if excluding any bias by RT-PCRor the location of the biopsy, may be due to both unequalnumbers of expressing cells and unequal transcript levelsper expressing cell. Second, both orthologs mouse Olfr43and human OR1A1 share an evolutionary conserved func-tion, and are activated by the same citronellic terpenoidswith similar efficacies. Third and finally, both paralogsOR1A1 and OR1A2 showed a considerable overlap in theirEC50-ranking odorant profiles. The structural differences inthe carbon chain length and functional groups of theirligands appeared to be the primary determinants of effi-cacy, however, we observed the stereo-enantiomer-selectivedetection of (S)-(�)-citronellol by OR1A1.

Differential hydrogen-bonding of (S)-(�)-citronellol toOR1A1, and to OR1A2, results in distinctly different interac-tion energies. Surprisingly we found that, indeed, there is adifference in the corrected interaction energies of (S)-(�)-cit-ronellol, which is ineffective in OR1A2, but effective inOR1A1, of about 7 kcal/mol, supporting the observed ORresponses. The interaction energies of (S)-(�)-citronellal,however, which elicits a response in both, OR1A1 andOR1A2, and with similar efficacies, are almost equal for bothhuman OR. Together, the calculated interaction energiesand our experimental results on (S)-(�)-citronellol beingresponsive in OR1A1 but non-responsive in OR1A2, are inagreement with the ‘‘binding threshold hypothesis’’ for activa-tion of OR: To activate an OR, the odorant must bind to theOR with a binding energy above some threshold value(Hummel et al., 2005). If we assume a binding thresholdsomewhere between �22.6 and �15.2 kcal/mol, both, (S)-(�)-citronellol, and (S)-(�)-citronellal, would comply withthese criteria. In summary we have established a 3D modelof OR1A1 and OR1A2, with (S)-(�)-citronellol and (S)-(�)-citronellal docked, which explains the effective responseof (S)-(�)-citronellol in OR1A1, the non-effective responsein OR1A2, and the effective response of (S)-(�)-citronellalin both, OR1A1 and OR1A2.

Our model proposes 17 amino acids within TM2-7 to bethe structural determinants of odorant binding in OR1A1and OR1A2. At 12 positions our model’s prediction over-laps with a previously reported generalized binding pocketof mouse-human OR orthologs (Man et al., 2004). Of thesetwelve amino acid positions we validated two positions inTM3 of OR1A1: N109 and S112 to interact with (S)-(�)-citronellol, and S112 to interact with (S)-(�)-citronellal.Here, our experiments showed that changing the proposedfavorite interaction site for a ligand always resulted in aloss of function.

The observed gain of function and loss of function forthe mutated single positions 108, 109, and 112 in OR1A1,and the observed loss of function for mutated position155 in OR1A2, appear, however, to be incomplete.[35S]GTPcS binding, in contrast to some FLIPR Ca2+-imaging results, revealed that single amino acid mutationsat positions 108 or 109 lead to only a partial loss of function

or gain of function, when testing (S)-(�)-citronellol onOR1A1 or OR1A2, respectively. This is in line with the pre-diction of our docking model: e.g. for OR1A1 and (S)-(�)-citronellol, the vdW interaction energy is not predicted to bestrongly affected by a mutation at position 108. Experimen-tal differences observed with single amino acid mutants atpositions 108 or 109 are likely due to a higher sensitivityof the [35S]GTPcS binding assay, as compared with theFLIPR assay. Both assays, however, confirmed that thedouble mutations at positions 108 and 109 in both,OR1A1 and OR1A2, are necessary and sufficient to confera strikingly different response behavior to a specific odor-ant, (S)-(�)-citronellol, from one receptor to another.

Of the six odorant-interacting amino acid positions inTM4-7 newly proposed by our model, we validated S112in TM3 of OR1A1, and S155 in TM4 of the paralogOR1A2 to interact with (S)-(�)-citronellal. Mutations atpositions 112 or 155 resulted in a pronounced, albeit incom-plete loss of function, suggesting that other amino acid posi-tions are critically involved, or became alternativelyinvolved in an interaction with (S)-(�)-citronellal in bothmutant OR. Interestingly, (S)-(�)-citronellal activated bothwild-type OR1A1 and OR1A2 with identical efficacies. Thissuggests that with certain degrees of freedom, and within ageneralized binding pocket, certain odorants may interactwith different OR, but with similar efficacies and/or poten-cies, and hence may provide the molecular basis for the phe-nomenon of combinatorial odorant coding at the level ofOR. In a recent study, based on structural homology mod-elling of the binding region of the olfactory receptor I7, Laiet al. (2005) identified a transit—dissociative—pathway ofodorants within this receptor’s binding region, and keyamino acid residues that facilitated this transit. An interpre-tation of this modelling study may suggest that the interac-tion of odorants with certain amino acid side chains withinan OR may thus be variable, rather dynamic than static,and may occur through a series of conformational interme-diates with distinct functional properties. Thus, the value ofOR-rhodopsin homology modelling by itself may be lim-ited, since it represents an inactive form of the receptorwhich is kept static during odorant docking simulations.However, a model may very well explain experimentalresults, or render predictions that need to be experimentallyfalsified or verified.

Only two other studies have recently combined homologymodelling, mutational analysis, and functional expression toidentify odorant-interacting amino acid positions, each in amouse OR (Abaffy et al., 2006; Katada et al., 2005). Themodel by Abaffy et al., proposed four amino acid positionsthat differ in Olfr544 (alias MOR42-3; mORS6), a mouseOR responding to dicarboxylic acids (Malnic et al., 1999;Shirokova et al., 2005), and its paralog Olfr545, to determinetheir different response profiles to odorants of different car-bon chain lengths. Mutating position V113 in Olfr544resulted in some significant changes in the odorant responseprofile. Significant effects in the other three proposed posi-tions were only observed in double mutations with V113.


All four positions overlap with some of the amino acids pro-posed by Man et al., to constitute a generalized odorantbinding pocket in OR (Man et al., 2004). Interestingly,V113 in Olfr544 corresponds to position 109 in OR1A1and OR1A2, which in our study determined the responsive-ness to (S)-(�)-citronellol. In another study by Katada et al.(2005), three lines of evidence led to the identification of 4amino acid positions determining the odorant specificity ofthe mouse eugenol receptor (Olfr73, alias mOR-EG;MOR174-9): (i) they were proposed as odorant-interactingpartners by docking eugenol into the homology-modelledOlfr73, (ii) they could be verified by mutational analysisand functional expression, and (iii) they overlap with theodorant binding pocket predicted by Man et al. ( 2004).Interestingly, S113 as one of the verified amino acid posi-tions in Olfr73 corresponds to position S112 in our study,which significantly determined the responsiveness of rho-tag(39)-OR1A1 to (S)-(�)-citronellal.

In summary, our experiments suggest different dockingsites for the agonists (S)-(�)-citronellal, and (S)-(�)-citro-nellol within both human OR. We show that the orientationof odorants within a homology-modelled odorant bindingpocket of OR is defined, with certain degrees of freedom,by evolutionary conserved amino acid positions. Experi-mentally derived odorant efficacies of wild-type and mutantOR will lead to a refinement of their odorant binding pocketmodel, which in turn may serve as a prediction of odorant-receptor interaction. A full validation of our molecularmodel may, nevertheless, depend on solving the crystalstructure of OR1A1 and OR1A2. However, the combina-tion of rhodopsin homology modelling, mutational analysisand functional expression of OR provides a manageabletool to elucidate the structure-activity relations of odorants,as well as structure-function relations of OR.

Acknowledgments

We thank P. Hargrave for B6-30 anti-rhodopsin anti-body, and J.-D. Raguse for human olfactory epitheliumbiopsy material. We are grateful to S. Bandholtz and T.Rathjen for technical assistance. This work was supportedby a grant from the Deutsche Forschungsgemeinschaft(Kr-1548 to D.K.). E.S. was supported by a fellowshipfrom the German Federal Ministry of Education and Re-search (BMBF).

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at doi:10.1016/j.jsb.2007.04.013.

References

Abaffy, T., Malhotra, A., Luetje, C.W., 2006. The molecular basis forligand specificity in a mouse olfactory receptor: A network offunctionally important residues. J. Biol. Chem. 17, 17.

Afshar, M., Hubbard, R.E., Demaille, J., 1998. Towards structural modelsof molecular recognition in olfactory receptors. Biochimie 80, 129–135.

Araneda, R.C., Peterlin, Z., Zhang, X., Chesler, A., Firestein, S., 2004. Apharmacological profile of the aldehyde receptor repertoire in ratolfactory epithelium. J. Physiol. 555, 743–756, Epub 2004 Jan 14.

Floriano, W.B., Vaidehi, N., Goddard, W.A.R., Singer, M.S., Shepherd,G.M., 2000. Molecular mechanisms underlying differential odorresponses of a mouse olfactory receptor. Proc. Natl. Acad. Sci. USA97, 10712–10716.

Gerber, P.R., 1998. Charge distribution from a simple molecular orbitaltype calculation and non-bonding interaction terms in the force fieldMAB. J. Comput. Aided Mol. Des. 12, 37–51.

Glusman, G., Sosinsky, A., Ben-Asher, E., Avidan, N., Sonkin, D., Bahar,A., Rosenthal, A., Clifton, S., Roe, B., Ferraz, C., Demaille, J., Lancet,D., 2000. Sequence, structure, and evolution of a complete humanolfactory receptor gene cluster. Genomics 63, 227–245.

Grosch, W., 2001. Evaluation of the key odorants of foods by dilutionexperiments, aroma models and omission. Chem Senses 26, 533–545.

Hall, S.E., Floriano, W.B., Vaidehi, N., Goddard 3rd, W.A., 2004.Predicted 3-d structures for mouse i7 and rat i7 olfactory receptors andcomparison of predicted odor recognition profiles with experiment.Chem Senses 29, 595–616.

Hummel, P., Vaidehi, N., Floriano, W.B., Hall, S.E., Goddard 3rd, W.A.,2005. Test of the binding threshold hypothesis for olfactory receptors:explanation of the differential binding of ketones to the mouse andhuman orthologs of olfactory receptor 912-93. Protein Sci. 14, 703–710.

Katada, S., Hirokawa, T., Oka, Y., Suwa, M., Touhara, K., 2005.Structural basis for a broad but selective ligand spectrum of a mouseolfactory receptor: mapping the odorant-binding site. J. Neurosci. 25,1806–1815.

Khorana, H.G., 1992. Rhodopsin, photoreceptor of the rod cell. Anemerging pattern for structure and function. J. Biol. Chem. 267, 1–4.

Kratochwil, N.A., Malherbe, P., Lindemann, L., Ebeling, M., Hoener,M.C., Muhlemann, A., Porter, R.H., Stahl, M., Gerber, P.R., 2005.An automated system for the analysis of G protein-coupled receptortransmembrane binding pockets: alignment, receptor-based pharma-cophores, and their application. J. Chem. Inf. Model 45, 1324–1336.

Krautwurst, D., Yau, K.W., Reed, R.R., 1998. Identification of ligandsfor olfactory receptors by functional expression of a receptor library.Cell 95, 917–926.

Lai, P.C., Singer, M.S., Crasto, C.J., 2005. Structural activation pathwaysfrom dynamic olfactory receptor-odorant interactions. Chem. Senses30, 781–792.

Lapidot, M., Pilpel, Y., Gilad, Y., Falcovitz, A., Sharon, D., Haaf, T.,Lancet, D., 2001. Mouse-human orthology relationships in an olfac-tory receptor gene cluster. Genomics 71, 296–306.

Liu, A.H., Zhang, X., Stolovitzky, G.A., Califano, A., Firestein, S.J.,2003. Motif-based construction of a functional map for mammalianolfactory receptors. Genomics 81, 443–456.

Malnic, B., Godfrey, P.A., Buck, L.B., 2004. The human olfactoryreceptor gene family. Proc. Natl. Acad. Sci. USA 101, 2584–2589.

Malnic, B., Hirono, J., Sato, T., Buck, L.B., 1999. Combinatorial receptorcodes for odors. Cell 96, 713–723.

Man, O., Gilad, Y., Lancet, D., 2004. Prediction of the odorant bindingsite of olfactory receptor proteins by human-mouse comparisons.Protein Sci. 13, 240–254.

Milligan, G., 2003. Principles: extending the utility of Trends. Pharmacol.Sci. 24, 87–90.

Niimura, Y., Nei, M., 2003. Evolution of olfactory receptor genes in thehuman genome. Proc. Natl. Acad. Sci. USA 100, 12235–12240.

Olender, T., Feldmesser, E., Atarot, T., Eisenstein, M., Lancet, D., 2004.The olfactory receptor universe - from whole genome analysis tostructure and evolution. Genet Mol. Res. 3, 545–553.

Palczewski, K., Kumasaka, T., Hori, T., Behnke, C.A., Motoshima, H.,Fox, B.A., Le Trong, I., Teller, D.C., Okada, T., Stenkamp, R.E.,Yamamoto, M., Miyano, M., 2000. Crystal structure of rhodopsin: AG protein-coupled receptor. Science 289, 739–745.

http://dx.doi.org/10.1016/j.jsb.2007.04.013

http://dx.doi.org/10.1016/j.jsb.2007.04.013


Pilpel, Y., Lancet, D., 1999. The variable and conserved interfaces ofmodeled olfactory receptor proteins. Protein Sci. 8, 969–977.

Shirokova, E., Schmiedeberg, K., Bedner, P., Niessen, H., Willecke, K.,Raguse, J.D., Meyerhof, W., Krautwurst, D., 2005. Identification ofspecific ligands for orphan olfactory receptors. G protein-dependentagonism and antagonism of odorants. J. Biol. Chem. 280, 11807–11815, Epub 2004 Dec 14.

Singer, M.S., 2000. Analysis of the molecular basis for octanal interactionsin the expressed rat 17 olfactory receptor. Chem. Senses 25, 155–165.

Singer, M.S., Weisinger-Lewin, Y., Lancet, D., Shepherd, G.M., 1996.Positive selection moments identify potential functional residues inhuman olfactory receptors. Receptors and Channels 4, 141–147.

Vaidehi, N., Floriano, W.B., Trabanino, R., Hall, S.E., Freddolino,P., Choi, E.J., Zamanakos, G., Goddard 3rd, W.A., 2002.

Prediction of structure and function of G protein-coupled recep-tors. Proc. Natl. Acad. Sci. USA 99, 12622–12627, Epub 2002 Sep26.

Wald, G., 1968. The molecular basis of visual excitation. Nature 219, 800–807.

Young, J.M., Shykind, B.M., Lane, R.P., Tonnes-Priddy, L., Ross, J.A.,Walker, M., Williams, E.M., Trask, B.J., 2003. Odorant receptorexpressed sequence tags demonstrate olfactory expression of over 400genes, extensive alternate splicing and unequal expression levels.Genome Biol. 4, R71, Epub 2003 Oct 7.

Zhang, X., Firestein, S., 2002. The olfactory receptor gene superfamily ofthe mouse. Nat. Neurosci. 5, 124–133.

Zozulya, S., Echeverri, F., Nguyen, T., 2001. The human olfactoryreceptor repertoire. Genome Biol. 2, RESEARCH0018.

Structural determinants of odorant recognition by the human olfactory receptors OR1A1 and OR1A2

Documents

Transcript of Structural determinants of odorant recognition by the human olfactory receptors OR1A1 and OR1A2