Functional Variability in the Human Odorant Receptor Repertoire

Home , OR10G4, OR10G7, OR11A1, OR1A2, OR2B11, OR2J3

© 2013 Nature America, Inc. All rights reserved. Connecticut, Connecticut, USA (S.S.L.). should Correspondence be addressed to J.D.M. ( Tong University School of Medicine, Shanghai, China (H.Z.) and Department of Ecology and Evolutionary Biology, Osborn Memorial Labs, Yale University, New Haven, USA Pennsylvania, (Y.R.L.), Department of Pathophysiology, Key Laboratory of Cell and Differentiation Apoptosis of National Ministry of Education, Shanghai Jiao University Medical Center, Research Drive, Durham North Carolina, USA. of and Neurogenetics Behavior, The Rockefeller University, New York, New York, USA. Drive, Durham North Carolina, USA. 1 perception. olfactory human on in odorant receptors, and measure changes mutations the of effect functional these missense of consequences functional and prevalence the determine receptors, odorant orphan several for ligands identify (SNPs) polymorphisms in of the function odorant receptors. Here we receptors olfaction. of models viable building for and levels at all system the olfactory for understanding is essential receptors odorant to ligands matching field; the in bottleneck critical a is data of lack receptors odorant human intact ~400 the of 22 only for published been have ligands receptors; odorant human considering to ligands has seen limited success, and the picture is even worse when receptors odorant mammalian Matching pairs. receptor-odorant for odorant. an of large perception a the on have influence can receptor single a of response the odorant, given a of recognition to lead types receptor multiple of responses which in code a uses combinatorial system olfactory the that although suggest isovaleric acid agonists respective the for thresholds detection elevated in function of loss cases: four in perception predicts receptor odorant an in variation odorants of perception the in variation considerable exhibit been shown to exhibit high genetic variability have that genes receptor odorant ~800 contains genome human The and ethyl vanillin. valence for the high-affinity genotype explain over 15% of the observed variation in perceived intensity and over 10% of the observed variation in perceived receptor alleles. To show that these that polymorphisms altered receptor function. We identified agonists for 18 odorant receptors and found that 63% of the odorant receptors we examined had in olfactory perception. We used a heterologous assay to determine how often genetic in polymorphisms odorant receptors alter Humans have ~400 intact odorant receptors, but each individual has a unique set of genetic variations that lead to variation Abigail Lin Andrew H Moberly Joel D Mainland human odorant receptor repertoire The missense of smell: functional variability in the nature nature Received 6 August; accepted 5 November; published online 8 December 2013; Monell Monell Chemical Senses Center, USA. Pennsylvania, Philadelphia, Using a high-throughput system for functional testing of odorant odorant of testing functional for system high-throughput a Using Understanding the role of a single receptor requires data functional 6 , , cis NEUR -3-hexen-1-ol 1 8 , we can now study the role of missense single-nucleotide single-nucleotide of missense role the now study , we can 2 OSCI & Hiroaki Matsunami EN OR11H7P 1 – C 3 7 E , , , , Andreas Keller 1

β , 3 -ionone advance online publication online advance , , Kaylin A Adipietro , , in in vitro in in vitro 3 OR2J3 Department Department of University Neuroscience, of Pennsylvania School of Medicine, USA. Pennsylvania, Philadelphia, 8 and androstenone function. On average, two individuals have functional differences at over 30% of their odorant agonist guaiacol but do not explain phenotype variation for the lower-affinity agonists vanillin , , in in vitro OR5A1 2 4 , 1– 5 , , Yun R Li results are relevant to olfactory perception, we verified that variations in 3 and and . . In addition, humans 2 OR7D4 , , Wen Ling L Liu 9 . These . results These 2 2 6 Department Department of Molecular Genetics and Microbiology, Duke University Medical Center, Research , ,

6 8– leads to to leads , , Ting Zhou 1 4 7 6 , 5 . This This . Present Present addresses: School of Medicine, University of Philadelphia, Pennsylvania, , and and , [email protected] doi:10.1038/nn.359 F sums-of-squares (extra agonist one least at to response significant a showed receptors odorant 27 receptors. odorant 160 of agonist tive pairs odorant-receptor receptors. odorant unique 160 These representing clones 190 included comparisons). multiple for rected ( control no-odor the than activation higher significantly in resulted odorant the of concentration one least at to exposure pairs, of For in the odorant 425 triplicate. odorant-receptor a control no-odor as well as 1 against receptor odorant each we tested which in screen a secondary We selected 1,572 odorant-receptor pairs from this primary screen for primary screen, we stimulated at an odorant concentration of 100 activity receptor measure to assay used a monophosphatecyclic adenosine luciferase (cAMP)-mediated testing psychophysical previous in used been have that ants screen. the in alleles mous nonsynony multiple by represented were receptors odorant Some Project. Genomes 1000 the in present alleles 912,912 receptor odorant intact their of (47%) 428,793 and genes, 418 receptor the odorant of intact (94%) 394 represent clones These screen. heterologous high-throughput a for genes receptor odorant human 511 of library a cloned we receptors, odorant of variety a for agonists Toidentify High-throughput screening of human odorant receptors RESULTS 5 test against vector control, control, vector against test Department Department of Neurobiology and Duke Institute for Brain Sciences, Duke 2 We then constructed dose-response curves for at least one puta one least at for curves dose-response constructed Wethen odor 73 of panel a with library receptor odorant the We screened 2 , , Hanyi Zhuang , Casey Trimmer , Casey 8 ). 2 , 6 1 , Senmiao , Zhan Senmiao , Lindsey L , SnyderLindsey µ M, M, 10 P < 0.05 divided by the number of of number the by divided 0.05 < 2 0 ( µ Supplementary Fig. 1 Fig. Supplementary M M and 100 2 t , , Somin S Lee -test, -test, 1 µ , t r a M concentrations M concentrations

P OR10G4 < 0.05, uncor 0.05, < 4 Laboratory Laboratory C I ). In the the In ). 9 , 1 e l 2

9 , and and 6 µ , M. M. s - - - -

© 2013 Nature America, Inc. All rights reserved. distances were calculated using the unweighted pair group method with with method group pair unweighted the using calculated were distances similarity receptor quantify to used were scales property acid amino Grantham’s families. gene receptor odorant 13 the of 9 represent and red in highlighted are receptors agonists with receptors odorant 27 odorant than 2 Figure polymorphisms significantly a of exhibit not number did agonists different identified we receptors odorant which The for 1%. over frequency allele an with variants 1% In across the haplotypes. 2,184 contrast, 18 odorant receptors had only one with over allele an frequency allele data. Project Genomes 1000 the in receptors odorant 418 across 1% than greater frequency allele an with alleles five of median a found We repertoire? receptor odorant human the in considerable variation of amount already the to adding variants, different functionally have frames reading open intact with receptors odorant the of many can account for also a of in portion the odor variance perception receptors odorant of residues acid amino nonconserved alters that variation missense a segregating residues, acid amino conserved perception. olfactory in receptor odorant odorant receptor-agonist pairs can be used to probe the role of a single Combined with psychophysics data for a genotyped population, these eight to agonists known with pseudogenes segregating of number ( forms disrupted and segregate intact that between receptors odorant seven for agonists identified We receptors odorant in variation Genetic receptors. odorant of variety wide a across interactions ligand-receptor examining for receptors odorant of families gene 13 the of 9 throughout spread are method this by fied to 40 (refs. total the bringing agonists, known with receptors odorant human of doubles published number total the nearly This agonists. new identified we receptors odorant agonist one least at to shown respond to been previously have that nine including tested), receptors t r a (UPGMA). mean arithmetic

In addition to segregating pseudogenes and missense variation in in variation missense and pseudogenes segregating to addition In Normalized luciferase value 0 1 0 1 0 1 0 1 0 1 0 1 –12 –12 –12 –12 –12 –12 C I

OR5P3:COUM OR10G4:VAN OR8B3:+CAR OR1A1:+CAR OR2J2:EVAN OR51L1:APA Unrooted tree based on similarity of amino acid properties. properties. acid amino of similarity on based tree Unrooted e l s –2 –2 –2 –2 –2 –2 2 1 0 1 0 1 0 1 0 1 0 1 0 1 –12 –12 –12 –12 –12 –12 ( OR8D1:DMHDM OR56A4:UNDEC Fig. Fig. OR5K1:EUGME OR2J3:CINMA OR10G7:EUG OR1C1:LIN 2 ), suggesting that our assay is useful useful is assay our that suggesting ), 9 , –2 –2 –2 –2 –2 –2 F 16 6 , 1 Table 0 1 0 1 0 1 0 1 0 1 , 7 –12 –12 –12 –12 –12 8– ( [Odorant] (logM) OR2W1:C3HEX Fig. Fig. 1 OR8K3:+MEN OR10J5:LYR OR6P1:ANIS OR2A25:GA OR51A2 7 ). The receptors identi ). receptors The 1 ), bringing the total total the bringing ), 1 ). For the other 18 18 other the For ). had 19 different 3 0 –2 –2 –2 –2 –2 , and and , 1 0 1 0 1 0 1 0 1 0 –12 –12 –12 –12 –12 Vector controlandodorant Receptor andodorant OR2B11:COUM 7– OR10A6:3PPP OR7C1:ANDI OR11A1:2EF OR4E2:AA 9 . . How

6 - .

–2 –2 –2 –2 –2 high surface expression did not reliably confer additional sensitivity. sensitivity. additional confer not reliably did expression surface high agonists, to known responses receptor eliminated expression surface assay. a of lack complete cell- Although in functional the variants the (Spearman efficacy potency relative either with (Spearman correlate not did variants of set each among expression (FACS). sorting cell cell-surface Relative activated fluorescence- by followed tag an Rho with N-terminal the against immunostaining antibody live-cell using alleles encoded cloned variants 51 these receptor by odorant of expression cell-surface the We assessed expression. cell-surface altering by is function receptor 0 1 0 1 0 1 0 1 0 1 –12 –12 –12 –12 –12 OR10G3:VAN OR2C1:OTHI OR4Q3:EUG OR51E1:IVA OR7D4:AND advance online publication online advance ρ = 0.04, 0.04, = –2 –2 –2 –2 –2 ρ = 0.13, 0.13, = P = 0.82, 0.82, = Project data. Project which One through mechanism Genomes 1000 the in present alleles 34,944 receptors’ the of (77%) 27,118 of average an cloned which represented 51 cessfully alleles, Genomes 1000 the Project data. For 16 in odorant receptors we suc seen alleles coding protein- of majority the representing of goal the with DNA genomic pooled from alleles cloned and agonist known one least at with receptors odorant targeted we odorants, by activation receptor odorant affects sequence Mann-Whitney U two-sided sets, both for 5 = alleles (median agonists identified without TableSupplementary 1 replicates. Odor are abbreviations defined in indicated odorants. Error bars, s.e.m. over three odorant receptor or an empty vector to the with either a plasmid encoding the indicated for 27 receptors. Responses of cells transfected encoded by the most common functional allele Figure 1 genetic polymorphisms could influence influence could polymorphisms genetic test, test, To test how variability in amino acid acid amino in variability how test To P Z

= 0.45, 0.45, = Supplementary Fig. 2a Fig. Supplementary = 0.77, 0.77, = Dose-response curves Dose-response of the receptor P Supplementary Fig. Supplementary = 0.44). = nature nature . NEUR ) or relative relative or ) OSCI

2b EN ) of of ) C E -

© 2013 Nature America, Inc. All rights reserved. luciferase value normalized to the value for the receptor encoded by the the by encoded receptor the for value the to normalized value luciferase by ( slope. unit line, Black the by encoded receptor the of responses the 100 at odorants representative 55 the to alleles ( replicates. (see the by encoded receptor the from elicited they the along ordered were Odorants variants. t (two-tailed control no-odor the above receptors variant the of any activate 100 at odorants representative 55 against tested OR2B11 of variants five for curves tuning ordered 3 Figure potency variants, the of 5.5% for and by pseudogenes, encoded were and 6.8% 7.9% were indistinguishable of hypofunctional. the variants data revealed that Project 11% of the variantsGenomes were hyperfunctional, 1000 68% were the from receptor odorant same the of ants vari other all to variant one Comparing value). (maximum lower a efficacy and EC50) concentration; effective maximal (half potency lower a both had pair the in variant one if hypofunctional or hyper- sums-of-squares extra an using variants the compared and curve sigmoid a to data the We fit odorants. different using variants between differences to correlated highly were odorant one using variants between differences the that suggested odorants of set broad a using from results our as agonist, as well as 12 odorants additional odorant 55 receptors. all We against typically used only tested a single receptors odorant 15 the included curve dose-response a ( mM 10 to nM 10 from constructing by concentrations of range a changes. subtle more tifying for iden not ideal therefore was library and this space, odorant span ( tuning We found no case of a genetic change that resulted in a change in odor to one agonist, that variant tended to be to hypersensitive all agonists. Supplementary Fig. 4 but ( consistent remained variants of varied, responses relative the response of magnitudes absolute the odorants, Across space physicochemical the span to tively quantita chosen odorants 55 against FACSanalysis alle the in the used of les 46 by encoded variants receptor odorant screened We variation genetic of consequences Functional expression. cell-surface in was due to variation defects functional whether olution to determine confer In responses. to summary,functional FACS did not provide sufficient enough res was expression cell-surface of amount small A conserved domain in the protein, the conserved amino acid that varies is underlined. in the 1000 Genomes Project The frequency of disrupted alleles for the corresponding odorant receptor genes as found OR10G7 OR4Q3 OR2C1 OR10A6 OR8K3 OR4E2 OR2B11 name Odorant receptor Table 1 nature nature replicates. three over s.e.m. bars, Error allele. reference -test, -test, We then examined how the variant responses compared across across compared responses variant the how examined then We OR2B11 Supplementary Fig. 3 Fig. Supplementary α = 0.05/55), that odorant’s response was set to zero across all all across zero to set was response odorant’s that = 0.05/55),

Supplementary Fig. 4 Fig. Supplementary NEUR Seven Seven segregating pseudogenes with agonists Functional testing of odorant receptor variants. ( variants. receptor odorant of testing Functional alleles for three indicated odorants. odorants. indicated three for alleles b ) Responses of the receptors encoded by the four variant variant four the by encoded receptors the of ) Responses OSCI allele percentage(%) Pseudogene EN 1.40 1.50 c ). ). In other words, if a variant was hypersensitive C 22 24 30 43 ) Dose-response curves for the receptors encoded encoded receptors the for curves ) Dose-response 2 E 4 2 Fig. 3 Fig. for odor names). Error bars, s.e.m. over three three over s.e.m. bars, Error names). odor for is listed. In cases where the variant allele alters a highly advance online publication online advance F test. We classified a pair of variants as as variants of pair a classified We test. µ

c M. If a given odorant did not significantly significantly not did odorant a given If M. ), but we chose our odorant library to to library odorant our chose we but ), and and 191 191 amino acid protein 159 amino acid protein 272 amino acid protein PMLN MAYD M 8-amino-acid protein AYDRY domain Supplementary Fig. 5 Fig. Supplementary R x P Result axis according to the response response the to according axis Y Y domain LIY LIY domain 1 OR2B11 7 ( OR2B11 Supplementary Fig. 3 Fig. Supplementary µ Y M, plotted against against plotted M, axis represents the the represents axis reference allele. allele. reference reference allele allele reference a ) ) Sensitivity- Eugenol Eugenol Octanethiol 3-phenyl propyl (+)-menthol Amyl acetate Cinnamaldehyde Fig. 3a Fig. propionate ). Here we ). Agonist

, b and and ). ). - - - - - comparisons, including an untested allele, as functionally different, different, functionally as allele, untested an including comparisons, alleles; 2 × response dose in tested receptors odorant (27 differences functional possible of 54 out response dose in differences functional 16 of average an saw we identical, functionally as alleles, untested involving those including comparisons, pairwise all classified When we conservatively alleles). × 2 subjects × 1,092 receptors had odorant (27 alleles Weof 58,968 the (79%) type. allele their to according assigned we population, Project regression phenotyping; analysis, (polymorphism tool PolyPhen the from analysis, analysis, regression tolerant; regression from score, (GERP) r profiling rate evolutionary (genomic conservation evolutionary with not did our correlate assay in function altered residue a that found odds The were protein. the they throughout function; change that acids amino the to pattern in shown are function measured in vitro altered that polymorphisms had examined we receptors odorant the ( hyperfunctional or hypo- clearly as them not classify could we so concordantly, change not did efficacy and of 54 possible functional differences. These results were consistent consistent were results These differences. functional possible 54 of out response dose in differences functional 22 of average an saw we = −0.04, a Normalized luciferase value Genomes 1000 the across differences functional quantify To 0.2 0.4 0.2 0.4 0.2 0.4 0.2 0.4 0.2 0.4 0 0 0 0 0 OR2B11 V198M,T293I,D300G function. Residues that are polymorphic across variants with with variants across are that polymorphic Residues function. OR2B11 V198M,G223D OR2B11 I130S,V198M r 10 10 10 10 10 P = −0.05, −0.05, = OR2B11 V198M = 0.83), predictions from the SIFT tool (sorting intolerant 20 20 20 20 20 Odorants OR2B11 30 30 30 30 30 P 40 40 = 0.81). = 40 40 40 50 50 50 50 50 Fig. 5 Fig. b c a

Normalized luciferase value vitro in ). When we classified all pairwise pairwise all classified we When ).

Variant response 4 Figure r 100 200 100 200 200 400 0.5 = 0.05, 0.05, = 0 0 0 0 than referenceallele OR2B11 V198Mresponse [Spearmint oil](logdilutionv/v) results to each participant participant each to results –6 in vitro in More sensitive 0 b –8 Fig. 4 Fig. –8 P . There was no obvious obvious no was There . Vector control OR2B11 I130S,V198 OR2B11 V198,G223 OR2B11 OR2B11 V198M,T293I,D300G OR2B11 V198M [Coumarin] (logM) [Quinoline] (logM) = 0.80) or predictions predictions or 0.80) = than referenceallele results for 46,561 46,561 for results –6 –6 t r a a Less sensitive –4 ). 63% (17/27) of of (17/27) 63% ). –4 –4 0.5 C I –2 D M –2 –2 e l s

© 2013 Nature America, Inc. All rights reserved. our current dose-response data. Amino acid positions conserved in at least 90% of the receptors are labeled with their single-letter amino acid code. acid amino single-letter their with labeled are receptors the of 90% least at in conserved positions acid Amino data. dose-response current our given function alter they that odds the to according colored are residues gray.remaining in The shown are allele reference their by encoded receptor the and alleles minor by encoded receptors the of any between vary not did that residues acid Amino receptor. the of function the alter SNPs where residues ( variations. missense to relative variability the of portion a small for account pseudogenes that Note 4 Figure GBR, British individuals from England and Scotland; IBS, Iberian populations in Spain; JPT, Japanese in Tokyo; LWK, Luhya in Webuya, Kenya; Kenya; Tokyo; in Webuya, JPT,in LWK, Luhya Spain; Japanese in populations Iberian IBS, Scotland; and England from individuals British GBR, Finnish; FIN, Colombia; Medellin, in Colombian CLM, Chinese; Han Southern CHS, Beijing; in Chinese Han CHB, collection; CEPH the from ancestry European Western and Northern with residents Utah CEU, USA; Southwest in ancestry ASW, African collection. sample of location the to point arrows from groups ethnic displays legend The lines. grid black by separated and axes the on labeled are populations Participant histograms. the in indicated values color with participants, between differences of number the represent squares the of colors The minus differences functional 5 Figure Asian of were participants both which in Pairs alleles. receptor ant two odor words, their of (16/54) other 30% over at In functionally differed participants. individuals related 500 the excluded we if t r a Nigeria. Ibadan, in Yoruba YRI, Italy; in Tuscanians TSI, Rican; Puerto PUR, California; Angeles, Los in ancestry Mexican MXL,

C I Summary of functional variation. ( variation. functional of Summary Functional differences between participants. ( participants. between differences Functional e l 11% s 5.5% 7.9% z -scored nucleotide differences ( differences nucleotide -scored 68% 6.8% a ) Functional differences among 27 odorant receptors of 1,092 participants in the 1000 Genomes Project. Project. Genomes 1000 the in participants 1,092 of receptors odorant 27 among differences ) Functional Unclassified Hyperfunctional Indistinguishable Hypofunctional Pseudogene a – c ) The number of functional differences ( differences functional of number ) The c ) among 27 odorant receptors of 1,092 participants from the 1000 Genomes Project. Project. Genomes 1000 the from participants 1,092 of receptors odorant 27 ) among b - G (CHS) and Japanese in Tokyo (JPT)) were more functionally functionally more ancestry were Asian had participant (JPT)) neither which in Tokyo pairs than in similar Japanese Chinese and Han Southern (CHS) (CHB), Beijing in Chinese (Han ancestry N Y F G L P M Y P L L N advance online publication online advance M Q I D R A C L P NH L a 2 Y b ), nucleotide differences ( differences nucleotide ), ) Snake plot of a typical odorant receptor showing showing receptor odorant a typical of plot ) Snake L C

Y S a I C – c at the place of geographic origin; origin; geographic of place the at S H (log ofoddsratio–1to0) Low likelihoodtoalterfunction (log ofoddsratio0to1) Moderate likelihoodtoalterfunction (log ofoddsratio>1) High likelihoodtoalterfunction K A C T L

P nature nature Y P b N ), and and ), L

z -scored -scored NEUR COOH OSCI

EN C E

© 2013 Nature America, Inc. All rights reserved. relative to other groups. other to relative differences functional into translate not did diversity this of much descent, African of individuals among Mann-Whitney variability genetic greater is two-sided 0.36; = U non-African median −0.83; = African (median ancestry was of African participant neither which in pairs than similar functionally more were (ASW, YRI) LWK and ancestry African of were participants both which in pairs account, ( diversity P Mann-Whitney two-sided 15; = 16; = non-African median African (median ancestry African of was participant neither in which pairs than different were more functionally (YRI)), Nigeria States (ASW), Luhya in Webuya, Kenya (LWK), and Yoruba in Ibadan, African of were Mann- two-sided 17; = Whitney non-Asian median 13; = Asian (median nature nature d < 0.0001) a c test, test, Percentage of Percentage of Normalized luciferase value Vector control OR10G4 variant4(4.2%)T62I,Y101C,M134V,G146S,P181S,V195E,R235G,K295Q OR10G4 variant3(36.4%)A9V,M134V,V195E,R235G,K295Q OR10G4 variant2(21.8%)L24P,V195E OR10G4 variant1(28.4%)Referencesequence variance explained by –50 100 150 –50 100 150 –50 100 150 2 variance explained by 50 50 50 OR10G4 alleles (r ) 2 0 0 0 OR10G4 alleles (r ) z 10 10 15 = 149, = 149, NEUR –10 0 5 0 5 –10 –10 U [Ethyl vanillin](logM) i. 5 Fig. test, Nonyl aldehyde Eugenol acetate 2 2 [Guaiacol] (logM) , in line with those populations having a greater genetic genetic a having greater populations those , with in line

2−Decenal Bourgeonal [Vanillin] (logM) –8 –8 –8 OSCI Linalool 2−Butanone

Sandalwood Eugenol P z

Isobutyric acid

Pyrazine < 0.0001; = 127, b ancestry (African Ancestry in Southwest United United Southwest in Ancestry (African ancestry

Isobutyraldehyde Orange –6 –6 –6 Pentadecalactone Jasmine into diversity genetic taking when However, ). EN Isobutyric acid (+)−Menthol 2−Methoxy−4−methylphenol Eugenol methyl ether C –4 –4 2−Butanone Cinnamon –4

E Cedarwood Cedarwood P (−)−Menthol Cineole Fir Citronella Pairs < in participants 0.0001). both which –2 advance online publication online advance –2 –2

1−Butanol Butyric acid 5 Fig.

Diphenyl ether Hexyl butyrate b Methyl salicylate Fenchone Ethyl vanillin Pyrazine Lime Isovaleric acid 2 0 Perceived intensityrank intense Cinnamon Octyl aldehyde c Less | Isobornyl acetate Citral there although that, shows This ). Butyl acetate Pentadecalactone Butyric acid Terpinyl acetate | 4 0 | Octyl aldehyde Butyl acetate | Anise |

Propylene glycol | | |

Cineole Terpineol | | (+)−Menthol Octyl acetate | | Isoeugenol Nutmeg 6 0 Cis−3−hexen−1−ol Caproic acid

Eugenol methyl ether Propylene glycol intens Androstenone Isobutyraldehyde More

Fir 0 0 Jasmine Odorants Odorants Intensity Spearmint Valence Spearmint e ** Vanillin 4−Methylvaleric acid * * Banana Phenyl acetaldehyde Bourgeonal Heptaldehyde Orange 1−butanol Citral Geranyl acetate Paraffin oil Vanillin U test, test,

Ethyl vanillin Perceived valencerank Undecanal pleasant Isoeugenol Eugenol acetate Less 20 Ambrette Galaxolide Nutmeg 2−decenal

Diallyl sulfide Ambrette | Phenyl acetaldehyde Sandalwood z = 29, 29, = 40 |

Terpinyl acetate (−)−Menthol | Ethylene brassylate Banana | Geranyl acetate R−limonene R−limonene Paraffin oil 60 pleasant 2−ethylfenchol Androstadienone | | | | | | | | Caproic acid Ethylene brassylate More Heptyl acetate Linalool * * Lime prod encode that 2–4 alleles variant and ALTYMGPVRK, encodes >4% in the participant population: the reference allele (allele 1), which ( phenotypes perceptual each of effect the examined We then vanillin. ethyl and vanillin guaiacol, for valence and intensity perceived their OR10G4 data Project Genomes 1000 the in common were les receptor of OR10G4 the agonists of three perception for their tested been had that subjects DNA of genomic had we because testing additional for by encoded receptor odorant an selected We perception. vitro in in vitro the in widespread humanare population and thatchanges these genetic changes resultgenetic in widespread that shown far so have We variation genetic of consequences Perceptual

Terpineol * R−carvone Heptyl acetate Diacetyl Nonyl aldehyde There were four Eugenol Isobornyl acetate O Fenchone 2−Methoxy−4−methylphenol O

2−Ethylfenchol R−carvone Ethyl vanillin H H functional changes. We next determined whether the observed C Diphenyl ether C functional changes lead to the predicted consequences in in consequences predicted the to lead changes functional Citronella Guaiacol Vanillin 1 Hexyl butyrate Undecanal rated had who participants 391 the of 308 for sequences 4−Methylvaleric acid Cis−3−hexen−1−ol 9 Methanethiol Androstenone nonfunctional and functional because and Decyl aldehyde

Isovaleric acid O OC Anise Diallyl sulfide OC OH OH Galaxolide Methanethiol OH H H

Heptaldehyde Androstadienone CH 3 Decyl aldehyde Diacetyl 3 *** *** Octyl acetate Methyl salicylate 3 Guaiacol Guaiacol OR10G4 P correction, FDR *** Results. in outlined model regression linear multiple the using analyzed by ( intensity ( model). a constant from different significantly overall were that models regression for shown ** * 4%. than lower a frequency with occurred alleles unlisted All participants. 308 the in frequency allele the is listed frequency The sequence. reference hg19 the to relative are changes acid amino The rank. median the denotes violin the inside line black The rank. a given assigning participants of number the to proportional is violin each of width The allele. paternal the for once and maternal the for once twice: represented is participant of variants receptor by agonists three for rank valence and ( allele. reference the of response baseline the to normalized are values or lee wt gaao uig hetero a using guaiacol with alleles four these by encoded variants receptor odorant the We respectively. challenged acids, amino (ALICVSSEGQ) eight or (VLTYVGPEGQ) reference allele by two (APTYMGPERK), five the by encoded one the from differ that ucts Error bars, s.e.m. of three replicates. replicates. three of s.e.m. bars, Error population. participant the in 4% than greater by encoded ( OR10G4 6 Figure to the three odorants than the reference by odorant 3 encoded allele had a much lower affinity F squares of sum extra −7.7, = 2) allele for log(EC50 −7.4, = 1) allele for (log(EC50 small was effect the but 1, allele by encoded receptor odorant reference the than guaiacol to sensitive more was 2 by allele encoded tor data to a sigmoidal curve. The odorant recep resulting the fit and assay luciferase logous Fig. Fig. c a > 0.05 after FDR correction. FDR after > 0.05 3,42 , receptors of curves response ) Concentration P d OR10G4 alleles with minor allele frequency (MAF) < 0.01 in the regression model, model, regression the in < 0.01 ) Percentage of perceptual variance ( variance perceptual of ) Percentage = 6.38, 6 ).

on perceived intensity and valence. valence. and intensity perceived on Effects of genetic variation in variation genetic of Effects c ) and valence ( valence ) and allele types. Each odor was was odor Each types. allele OR10G4 P < The odorant receptor 0.002). F test. For all other odorants, odorants, other all For test. alleles with a frequency a frequency with alleles b OR10G4 ) Perceived intensity intensity ) Perceived d in vitro OR10G4 vitro in ) ranking explained explained ) ranking t r a P 2 OR10G4 < 0.001 after after < 0.001 2 . We obtained We . obtained OR10G4 P allele on the the on allele < 0.05, < 0.05, F C I Y test (only (only test -axis -axis OR10G4 . Each . Each r 2 e l F ) ) in alle

test, test,

- - -

© 2013 Nature America, Inc. All rights reserved. ing the functional consequences of segregating polymorphisms odorant receptors. Pairing odorants and odorant receptors, and verify first the investigation is of the rolefunctional of genetic variation knowledge, in a large our set of to and, agonist known a with receptors odorant human of number the doubles nearly This odorants. of tion to percep human is relevant segregation this that and demonstrated population human the in segregate that alleles different functionally have that agonists known with receptors odorant 27 identified We DISCUSSION ( type intensity and valence were significantly correlated with guaiacol only 68 compounds, Of the solvents. as two as well receptor of the agonists to be are that not known of odors 63 ception per and valence intensity for their tested psychophysically were also uncorrected model, constant the to compared valence vanillin ethyl model, constant to the pared stant model, uncorrected model, constant the to compared intensity (vanillin by predicted were vanillin ethyl and vanillin of 2 and ( 1 alleles of possession with associated significantly not were β more for pleasant copy each 3 of and alleles 4 ( ranks 3.7 and 3.3 guaiacol rated Participants odors. tested other the to relative 29th guaiacol of valence of the rank none would alleles with major the subjects that estimated model The correction). FDR model, constant the to compared 0.103, ( guaiacol of valence perceived in variance ( 2 allele of possession with associated significantly not were ratings ( P possessed they 4 and 3 alleles of copy each for intense ( possessed 1 they allele reference the of copy each for intense more ranks 2.1 guaiacol rated intensity of guaiacol 24th relative to the other tested odors. Participants the rank would none of with major alleles the that subjects estimated 15.0, r analysis, (regression guaiacol of intensity perceived in variance the rank. rating odor the against population, >4% in for the participant with frequency the four alleles in vitro predicted allele type participants’ significantly perception of the three affinity ( in decrease the cause to interact residues multiple that ing suggest 4, and 3 alleles by encoded receptors odorant the in ment impair functional the for accounted SNP single no that found and background reference a in SNPs the of each with receptors odorant control, vector squares against of sum (extra only vector with transfected cells and 4 allele with transfected cells between response in difference no was of squares sum of squares test against reference, = −5.5, (log(EC50) responses in significant resulted but still receptor t r a regarding questions inaccessible previously address to us allowed β β 6 Fig. Supplementary 2

< 0.02 and = 3.71, 3.71, = = 0.165, adjusted adjusted 0.165, = = −0.69, −0.69, = 1.01, = In contrast to guaiacol, neither perceived intensity nor valence valence nor intensity perceived neither guaiacol, to contrast In to intensity, In addition whether tested we analysis, regression multiple Using P Fig. 6c Fig. < 0.001 after false discovery rate (FDR) correction). The model agonists. We regressed the predictors, allele counts (0, 1 (0, counts or 2) allele We predictors, agonists. the regressed C I P P F = 0.32). = P < 0.03), respectively, but participants’ valence ratings ratings valence participants’ but respectively, 0.03), < test against vector control, control, vector against test e l β F , P = 0.52; 0.52; = P d = −4.34, 4,303 = 0.44; ethyl vanillin intensity compared to the con the to compared intensity vanillin ethyl 0.44; = = 0.74). As additional controls, the 308 participants participants 308 the controls, additional As 0.74). = ). s = 0.95, uncorrected r 2 β = 0.154, compared to constant model, model, constant to compared 0.154, = = 1.88, 1.88, = P ). F β < respectively. 0.005), Participants’ intensity 3,42 = 2.10, = 2.10, OR10G4 = 2.2, 2.2, = F OR10G4 P 4,303 = 0.08), respectively. 0.08), = P < 0.04), and 2.4 and 4.3 ranks less less ranks 4.3 and 2.4 and < 0.04), = 0.84, uncorrected uncorrected = 0.84, allele type predicted 10.3% of the 10.3% predicted type allele P = 0.11; 0.11; = F allele type predicted 15.4% of 15.4% predicted type allele P 3,42 F = 0.44; valence vanillin com F 3,42 = 459, 4,303 r = 149, = 149, 2 Fig. 6 Fig. = 0.115, adjusted adjusted 0.115, = = 9.85, 9.85, = β = 3.33, P OR10G4 < sum extra 0.001; a P ). We generated Wegenerated ). < 0.001). There There < 0.001). P OR10G4 F P < 0.001 after after 0.001 < P F 4,303 < and 0.002 = 0.50; and = 0.50; 4,303 allele type type allele β = −2.39, −2.39, = OR10G4 OR10G4 = 0.95, 0.95, = F = 0.50, = 0.50, in 4,303 F allele

vitro test test r 2 = = ------

effects of noncoding polymorphisms on expression. That said, there there said, That expression. on polymorphisms noncoding of effects the for testing assay of an efficient lack of the because genes receptor odors. of perception her or his affects that receptors olfactory of set personalized highly receptor one than more ing encod genes containing genome the of regions to linked been have genes receptor odorant single a to linked been previously have phenotypes perceptual tory olfac four here, demonstrated we association OR10G4-guaiacol the to addition in cases; several in individuals across perception odor in to variability Variationleads level variable. at is highly peripheral the the 800 alleles. This suggests that odor detection at the peripheral level ant receptors, we expect each pair of individuals to differ at 237–326 of data × If 2 per subject). we to alleles extrapolate the ~400 odor intact dose-response with genes receptor odorant (27 alleles 54 possible a perception. olfactory of model a in step first crucial a is code neural the affects receptor odorant an of alteration functional the how Understanding character. or intensity threshold, odor for codes array receptor odorant the how know currently not olfactory do we study, of as field alters future rich a be to promises receptor This perception. odorant individual an of activation how the influence of a single odorant receptor on the percept. Another is is Another percept. the on receptor odorant single a of influence the of guaiacol than in the perception of androstenone, therefore reducing possibility is that more odorant receptors have a role in the perception The reason for genotype. loweris value unclear.this explanatory One perception of androstenone intensity explained by variation in ception of guaiacol intensity, which is lower than the 39% of variation in investigation. further require will processing of stages later at for compensated is variation this of much how determining rules exact the but variation, perceptual to variation functional all not that suggests This ant. affinity to the ligand found the odor to be less lower intense and with more pleas receptor a encoded that allele the with ticipants for case the was As other. the than potent less much is androstenone to respond alleles major the androstenone and between association the to similar is this Indeed, receptor. the in variation by functional altered be not will receptor a activate of that only weakly odorants and is valence that intensity the ethyl interpretation one more possible or data are needed, Although vanillin. vanillin either than agonist potent more a is guaiacol that to lead can altered gene which transcription regions, noncoding in variation considerable is for guaiacol. The dose-response curves shown in in shown curves dose-response The guaiacol. for the alleles. tested the in by linkage confounded be may alterations SNP-specific receptors, odorant of majority the for SNPs of combination possible every generate not did we because not SNP; from a particular resulting changes was detect to carefully designed study our that however, Note, alogrithms. computational popular two by predicted are or evolution neutral from deviate tor, in repertoire. receptor the potential extent of variation in each individual’s expressed odorant spread in human odorant receptors Similarly, we did is not which variation, wide copy-number examine

We focused only on SNPs in the coding regions of the odorant odorant the of regions coding the in SNPs on only focused We of 16–22 in differences average, on had, individuals of pair Each Variation in Although Although we found that OR10G4 has at least three alter that SNPs suggesting evidence any find not did study Our vitro OR10G4 function are restricted to a particular domain of the recep the of domain particular a to restricted are function advance online publication online advance allele type only predicted perceived intensity and valence OR10G4 9 . In that case, the receptors encoded by both of of both by encoded receptors the case, that In . genotype explains 15.4% of the variance in per 6– 2 9 6 and five additional olfactory phenotypes phenotypes olfactory additional five and and even changes in sensory perception 23– 2 5 . Each individual, therefore, has a has therefore, individual, Each . 28 , 2 9 . . Thus, our data underestimate nature nature in vitro in OR7D4 Figure 6 Figure , but one variant variant one but , in vitro in in vitro NEUR OR7D4 OSCI genotype genotype will lead lead will agonists, a in vitro in OR7D4 reveal reveal , par , EN C 2 7 E ------.

© 2013 Nature America, Inc. All rights reserved. 4. 3. 2. 1. r at online available is information permissions and Reprints The authors declare no competing interests.financial analysis and wrote the paper with help from all authors. H.M. supervised the project. collected the psychophysical data and provided DNA samples. J.D.M. carried out the S.Z., W.L.L.L., T.Z., Y.R.L., H.Z., S.S.L., A.L. and K.A.A. performed research. A.K. J.D.M. and H.M. conceived and designed the project. J.D.M., C.T., A.H.M., L.L.S., Columbia Centre for Macular Research) for 4D2 anti-rhodopsin antibody. and DNA samples by A.K. in her laboratory, and R. Molday (University of British equipment, L.B. Vosshall for the ofcollection supervising psychophysical data Duke Cancer Institute Flow Cytometry Core. We thank D. Advancing Translational The Sciences. FACS analysis was performed using the the and Clinical Translational AwardScience program at the National Center for of Collection psychophysical data was supported by grant # UL1 TR000043 from supported by the Defense Advanced Research Project Agency RealNose Project. Institutes of Health Core NIDCD Grant A P30 DC011735. portion of the work was Analysis Core, which are supported, in part, by funding from the US National Monell Receptor Signaling Chemosensory Core and Genotyping and DNA/RNA fellowship toF32 DC008932 J.D.M. A portion of the work was performed using the andT32 DC000014 DC013339, a National Award Research Service postdoctoral This work was supported by R01 R03 R01 R01 DC011373, DC012095, DC005782, o Note: Any Supplementary Information and Source Data files are available in the the in v available are references associated any and Methods M perception. olfactory on receptor odorant single a of effects behavior, data these provide a solid platform from which to probe the gating polymorphisms segre of consequences functional the measuring receptors, odorant to ligands assigning By identity. odorant for code combinatorial the of nature for redundant the and space receptors, search odorant and large odorants the both of because part in unknown, currently may on have the effects two phenotypes. differential background, genetic and culture as such variables, confounding that nature nature COM AUTH Acknowledgmen e n e p

l ethods r The role of a single odorant receptor in olfactory perception is is perception olfactory in receptor odorant single a of role The i r n Ayabe-Kanamura, S. Ayabe-Kanamura, Olender, T. genomic from olfaction: Human T. Olender, & D. Lancet, Y., Hasin-Brumshtein, people. different for noses Y.Different Gilad, & D. Lancet, O., Man, I., Menashe, German cross-cultural study. cross-cultural German 13 diversity. phenotypic to variation Genet. Nat. s i n i e o t P

, 414 (2012). 414 , s v O n E / e i r n TI

R R s o d NEUR i o f CON e NG

n x t .

h o h et al.

FI f e t 34

m t

TRIBUTI OSCI h p N l , 143–144 (2003). 143–144 , e . a Personal receptor repertoires: olfaction as a model.

A p p a NC e p r EN e ts et al. et . r IA . C ON in vitro L E Differences in perception of everyday odors: a Japanese- a odors: everyday of perception in Differences I

N S Chem. Senses Chem. advance online publication online advance T E TrendsGenet. R and linking E STS

, 31–38 (1998). 31–38 , 25 in vitr , 178–184 (2009). 178–184 ,

o function o to function human Marchuk for sharing h t t p : / / w w BMC Genomics w .

n a t u o r e n . c l i o n m e - /

6. 17. 16. 15. 14. 13. 12. 11. 10. 9. 8. 7. 5. 30. 29. 28. 27. 26. 25. 24. 23. 22. 21. 20. 19. 18.

Menashe, I. Saito, H., Chi, Q., Zhuang, H., Matsunami, H. & Mainland, J.D. Odor coding by a by coding Odor J.D. Mainland, & H. Matsunami, H., Zhuang, Q., Chi, H., Saito, Schmiedeberg, K. heat specific A H. Hatt, & J. Barbour,W., Zhang, A., Mashukova, E.M., Neuhaus, ligand receptive extended an of Characterization H. Vogel, & H. V.,Jacquier,Pick, V. Matarazzo, Sanz, G., Schlegel, C., Pernollet, J.C. & Briand, L. Comparison of odorant specificity M. Spehr, C.H. Wetzel, variation Genetic H. Matsunami, & L.B. Vosshall, Q., Chi, H., Zhuang, A., Keller, Jaeger,S.R. J.F. McRae, odors. primary of concept the and anosmia Specific J.E. Amoore, rnhm R Aio cd ifrne oml t hl epan rti evolution. protein explain help to formula difference acid Amino R. Grantham, of variation number copy and drift Genomic M. Nei, Y.& Kawahara, M., Nozawa, S.M. Waszak, Allelic D. Drayna, & A. Manichaikul, J.P., Slack, C.T., Simons, A.A., Fushan, X. Zhang, Eriksson, N. N. Eriksson, J.F. McRae, from variation genome human of map A Consortium. Project Genomes 1000 The S. Hayden, measuring and expression cell-surface Evaluating H. Matsunami, & H. Zhuang, olfactory An Vosshall,L.B. & A.N.Gilbert, I.A., Gomez, M.,Hempstead, Keller,A., Saito, H., Kubota, M., Roberts, R.W., Chi, Q. & Matsunami, H. RTP family members olfactory receptors OR1A1 and OR1A2. and OR1A1 receptors olfactory Senses Chem. proteins. receptor olfactory mammalian of expression the enhances protein shock compounds. related structurally comprising OR17–40 receptor olfactory human the of repertoire (2005). system. cell insect Sf9 baculovirus the in expressed antagonism. for evidence and classes phylogenetic different from receptors olfactory human two of chemotaxis. sperm oocytes. and cells 293 kidney embryonic human in expressed functionally perception. odour alters receptor (2007). odorant human a in selection. food and 37 with the ability to detect the “grassy” smelling odor, cis-3-hexen-1-ol. Biol. (1977). 267–281 Science (2007). humans. in genes receptor sensory Biol. Comput. PLoS diversity.content gene receptor olfactory extensive reveals data sequencing genome sucrose. to polymorphism within the TAS1R3 promoter is associated with human taste sensitivity microarray. DNA novel a using family preference. traits. common for associations genome. human the in odors food-related to sequencing. population-scale subgenomes. (2008). 1402–1413 cells. heterologous in receptors odorant mammalian of activation metropolitandiversepopulation. a demographyof (2004). receptors. odorant mammalian of expression functional induce repertoire. receptor Mammalian , 585–593 (2012). 585–593 ,

5 , e284 (2007). e284 ,

185 J. Neurosci. J. et al. et t al. et et al. et Curr. Biol. Curr. arXiv et al. et al. et al. et t al. et , 862–864 (1974). 862–864 , t al. et J. Neurochem. J. Genome Res. Genome

et al. et t al. et t al. et t al. et 31 Characterizing the expression of the human olfactory receptor gene receptor olfactory human the of expression the Characterizing Chem. Senses Chem. Identification of a testicular odorant receptor mediating human mediating receptor odorant atesticular of Identification Ecological adaptation determines functional mammalian olfactory mammalian functional determines adaptation Ecological 1209.2096 (2012). 1209.2096 Genetic elucidation of human hyperosmia to isovaleric acid. A genetic variant near olfactory receptor genes influences cilantro A Mendelian trait for olfactory sensitivity affects odor experience odor affects sensitivity olfactory for trait Mendelian A , 445–452 (2006). 445–452 , et al. Identification of regions associated with variation insensitivity with variation associated regions of Identification eei vrain n h ooat eetr RJ i associated is OR2J3 receptor odorant the in variation Genetic Systematic inference of copy-number genotypes from personal from genotypes copy-number of inference Systematic Science ucinl hrceiain f w hmn latr receptors olfactory human two of characterization Functional

Curr. Biol. Curr. e-ae, atcpn-rvn tde yed oe genetic novel yield studies participant-driven Web-based, Specificity and sensitivity of a human olfactory receptor olfactory ahuman of sensitivity and Specificity 6

, e1000988 (2010). e1000988 , 19 Structural determinants of odorant recognition by the human , 7426–7433 (1999). 7426–7433 , , 1288–1293 (2009). 1288–1293 ,

20 299 97 Nature

, 1–9 (2010). 1–9 , , 537–544 (2006). 537–544 , 23 30 PLoS Genet. PLoS , 2054–2058 (2003). 2054–2058 , Sci. Signal. Sci. , 1601–1605 (2013). 1601–1605 , , 69–80 (2005). 69–80 , Proc. Natl. Acad. Sci. USA Sci. Acad. Natl. Proc.

467 Genome Biol. Genome J. Struct. Biol. Struct. J. , 1061–1073 (2010). 1061–1073 ,

2 Curr. Biol. Curr. 6 , ra9 (2009). ra9 , , e1000993 (2010). e1000993 , BMC Neurosci.BMC

8 Chem. Senses Chem. , R86 (2007). R86 ,

159 23 Nature , 1596–1600 (2013). 1596–1600 , t r a , 400–412 (2007). 400–412 ,

104 Cell

Chem. Senses Chem. , 20421–20426 ,

449 a. Protoc. Nat. 119 eou laevis Xenopus

, 122 (2012).122 , Chem. Senses 30 C I 468–472 , , 195–207 , , 679–691 , e l PLoS

s 3 2 , ,

© 2013 Nature America, Inc. All rights reserved. from ref. from Procedures for olfactory psychophysics. with a 3100 or 3730 Genetic Analyzer (ABI Biosystems). wereucts S-400 then(GEpurifiedHealthcare) using Sephacryl and sequenced TGAGAAATGAGTCAA-3 (5 genomic DNAs were amplifiedhuman sequencing, Forwith kit. HotStarDNA blood PAXgene TaqQiagen the (Qiagen)with prepared with primers upstream Venousfromcollectedparticipantswas ml) (8.5 andblood genomic DNA was consent to participate and were financially compensated for their informedgave time subjects All UniversityandBoard.Rockefeller Institutional Review effort. Human odorant receptor genotyping. public database). the in participants 1,092 the of each for alleles (two alleles phased 2,184 full-length into file vcf the translate to used was script Matlab custom-written A Variant calls were obtained from the ( public repository data in vcf lic format using tabix pub Project Genomes 1000 the of release phased May2011 the from derived 1000 analyzed with Microsoft Excel, GraphPad Prism 4 and Matlab (MathWorks).luciferaseactivity to control for transfection efficiency in a given well. Data were Optima plate reader (BMG). All luminescence values were divided by the Four hours after odor stimulation, luminescence was measured using a Polarstar appropriate concentration of odor diluted from the 1 M stocks24 h afterin transfection,CD293 (Gibco). transfection medium was removed and replaced with the of plasmids encoding odorant receptors. 1 driven by a cyclic AMP response element, 2.5 ng/well ofwith M3 5 (ref.ng/well of RTP1S plasmid measure receptor reponses as previously L described collection and analysis were not randomized. analysis. Cell-surface expression was quantified as PE fluorescence thatintensity. were EGFP-negative Data and/or 7AAD-positive were removed from subsequent fluorescent cells was conducted using a BD FACSCanto (BD Biosciences). Cells cell-impermeant DNA-binding agent that selectively stains dead cells. Sorting of dilution of 7-aminoactinomycin D (7AAD; 1 mg/ml; Calbiochem), a fluorescent, and resuspended in PBS containing 2% FBS and 15 mM NaN FBS and 15 mM NaN mouse antibody (Jackson Immunologicals) diluted 1:100 in PBS containing 2% lowed by secondary incubation with Phycoerythrin (PE)-conjugatedNaNmM 15containing and(vol/vol)PBS FBS2% in werewashed Cells donkey anti- R. Molday) diluted 1:50 in PBS containing 2% FBS and 15 mM NaN carried out at 4 °C using mouse monoclonal anti-rhodopsinand 4D2 detached(ref. from the dishes using Cellstripper (Cellgro).trol for transfectionPrimary incubationefficiency. 24 wash after transfection, cells were washedtransporting with proteinPBS 1 short (RTPIS) plasmid and 20 ng of EGFP plasmid1,200 to con dish was transfected with plasmids using 4 dishes (Falcon) and grown overnight at 37 °C and(Invitrogen) and5% 6 CO medium (Sigma) containing 10% FBS (Sigma), 500 ( cloneone than more had we receptorsodorantwhere 17 the forclones tested sorting. cell Fluorescence-activated (3100 Genetic Analyzer, Applied Biosystems). sequencing by verified receptorswerecloned the of sequences The tag). (Rho rhodopsin human of residues 20 first the encoding sequence the containing using Phusion polymerase and subcloned into pCI expression vectors (Promega)International participantsthe fromgenomicHapmapDNA20 of Consortium C ONLINE METHODS nature nature financiallywere andparticipate to consentinformed gave subjects All Board. Supplementary Fig. 5 Fig. Supplementary uciferaseassay. loning. ′ -ACCTGGTTGATGCAGTTTCC-3

G g h-agd drn rcpo pamd 30 g ua receptor- human ng 300 plasmid, receptor odorant Rho-tagged ng enomes Project data. Project enomes Odorant receptor open reading frames were amplified from the the from amplified were frames reading open receptor Odorant NEUR 1 ftp://ftp-tra 9 and approved by the Rockefeller UniversityInstitutionalRockefellerReview the approvedby and OSCI The Dual-Glo Luciferase Assay System (Promega) was used to µ g/ml amphotericin B (Sigma). Cells were seeded in 35-mm 3 for 30 min, covered with aluminum foil. Cells were washed ce.ncbi.nih.gov/1000 EN ). Hana3A cells were maintained in minimal essential minimal in maintained were cells Hana3A ). C ′ E ) of the OR10G4 open reading frame. The PCR prod Allele frequency in the human population was was population human the in frequency Allele 3 2 , 5 ng/well of pRL-SV40, 10 ng/well of luciferase ′ We conducted FACS analysis on all all on analysis FACS conducted We All DNA samples were approved by the ) and downstream (5

M odorant stocks were diluted in DMSO. All psychophysical data were obtained µ genomes/ftp/release/ l Lipofectamine 2000 (Invitrogen), 2 0 . Hana3A cells were transfected µ g/ml penicillin-streptomycin 2 . The following day, each ′ -AAACCTATTGA 3 containing 1:500 3 3 ) and 5 ng/well 20110521 3 3 for 30 1 ; gift from Renilla 3

, fol , min. / ) 2 3 2 4 - - - - . .

primary screen; some receptors were very promiscuous, so we tested only the topthe no-odor condition. We selected the top 5% of odorant-receptorboth applied at 10 pairs from this odorantsat 100 library with 73 odorants used in previous psychophysical testing Statistical analysis. were not randomized. phenotypinghumantheofsubjects complete. was Datacollection andanalysis sequencingconductedgenotype,allwasanalysis blindtoafterand astionwas guaiacolwas presented at a 1/1,000,000 dilution in paraffin oil. Our data collec Ethyl vanillin and vanillin were presented at a 1/200 dilution in propylene glycol; OR10G4 areguaiacol, and vanillin vanillin, odorants,ethyl these of in jars. For a detailed description of the psychophysical methods, see ref. 1 as “extremely unpleasant” and 7 as “extremely pleasant.” “extremelyStimuli were weak”presented and 7 as “extremely strong.” The valence scale was labeledvalence with of 66 odorants on a 7-point scale. The intensitybrief,participantsIn intensity66. rated thetoscale and19 range of ayears, waswith 35 labeled with 1 as Caucasian,were29 Asian andwere77 African-American. Themedian agewas under 18 were excluded from this study. Of the 308 subjects (138 male), therapy,133 were radiation to head and neck, or alcoholism. injury,headPregnant assuch smellcancerofreduced conditionsense causeda has that women and children tion, seasonal allergy, prior endoscopic surgery on the nose,respiratoryinfec preexistingupper illness, nasal medical of history a fragrances,or odors to gies compensated for their time and effort. Exclusion criteria for subjects were: aller in concentration from 10 nM to 10 mM. All numerical results are reported as reportedareresults numerical All mM. 10 to nM concentration10 in from Data collection and analysis were not randomized. of the alleles, and for than 85% of the 2,184 alleles. For of the 2,184 alleles. For 13 odorant receptors, the cloned alleles represented more 1% in the 1,092 participants and the cloned alleles represented a large percentage where we had more than one allele cloned with an allele frequency greater than responses from the clones of each odorant receptor. Wesame parentused 15 plateodorant of receptorscells. We performed an analysis of variancemeasure was collected from separate(ANOVA) wells but each well contains cellson from the the ant was screened against each receptor variant at 100 closest to the centroid of the cluster to maximize structural diversity. Each odor receptor.cluster,odoranttheagonistpresent suchthe wasin no If selected we cluster among odorants that are previously shown to activate at least oneclustering. odorantFor each cluster, we selected the odorant closest to the centroidusing clusters of55 intoodorantsthe 2,715 the divided Weodorants.then variance in mammalian odorant receptor responses space, cal we generated 20 physicochemical descriptors that predict 62% of the the vector control, which was transfected with an empty vector (extra sums-of-squares firmed that the odorant activated the receptor significantly morethe than fittedthe control,log(EC50) was less than 1 log unitconfidence and intervalsthe of extrathe top sums-of-squaresand bottom parameters didtest not overlap,con the s.d.fit the ofdata to a sigmoidal curve. We counted an odorant as an agonist if the 95% parent plate of cells, and a vector-only control was included for eachsamecontained welltheseparatefromodorant.collectedfrom eachcellswells,but was We Each odorant receptor–odorant dose was tested in triplicate, where each in measurethe secondary screen ( 10 mM for the odor-receptor pairs that were significantly different from baseline constructeddose-response usingcurves concentrations ranging fromtonM 10 tested. In addition, the tests were uncorrected for multiple comparisons. Wecedure, thenthe data distribution was assumed to be normal, but this was withnot the three formallywells stimulated with medium alone. As this was a screening pro was assessed by two-sided each well contains cells from the same parent plate of cells. Statistical performedsignificance in triplicate, where each measure was collected and100from separate wells, but each odorant receptor was tested against a no-odor controlten ligandsas forwell a givenas receptor.1 We then performed a secondary screen in which Dose-response curves. Dose-response Screening 55 odorants. . We examined the ratings of the higher of two tested concentrations.testedtwo of higher the ratingsofWe theexamined . µ M ofagonistsM identified primarytheinscreen. Eachcomparison was µ M(except for androstenone and androstadienone, which were µ P

Screening procedure M) and ranked odorant-receptor pairs by their activity above < 0.05). Data collection and analysis were not randomized. OR10G4 We tested odorant receptors with odorants ranging odorants with receptors odorant tested We To choose 55 odorants that quantitatively span chemi t -test, t -test comparing the three wells stimulated with odor , the cloned alleles represented 29.5% of the alleles. P < 0.05, uncorrected for multiple comparisons). OR2B11 . We stimulated the entire odorant receptor , the cloned alleles represented 37.5% 1 µ 7 for 2,715 commonly used M in triplicate where each doi: in vitro in 10.1038/nn.3598 9 . We applied the F agoniststo test against µ M, 10 k 9 -means . Three µ M ------

© 2013 Nature America, Inc. All rights reserved. mouse and human odorant receptors. For each SNP, we calculated the ratio ratio the SNP,calculated we each For receptors. odorant human and intact mouse 1,425 of alignment multiple-sequence a to variants receptor odorant collection and analysis were not randomized. Data participants. of pair per differences functional possible 48 of maximum a receptors,odorantwithacross summed were values These difference.tional accordingaboveremainingthecriteria. Each countedtowas pairone func as indistinguishablewere that alleles of pairsany removed andreceptor odorant a lower efficacy). showed discordant efficacy changes more(i.e.,wasone allelesensitive buthad top-bottom). A pair of alleles was designated as ‘unclassified’ if the potency and had both a higher EC50 (lower efficacy) and a lower potency (dynamic range or A pair of alleles was classified as hyper- or hypofunctional if one allele in the pair both alleles better than two separate models using the extra sums-of-squares test. P sums-of-squares (extra control vector-only that the odorant activated the receptor significantly more log thanEC50 was less than 1 log wasunit and the extra sums-of-squaresthe testcase confirmed for the intervals of the top and bottom parameters did not overlap, the s.d. of the fitted eter logistic model. We counted an odorant as an agonist if the 95% confidence from the same parent cells plate contains of well cells. We each but fit wells,the resultingseparate datafrom with a collected three-param was measure each mean doi: < 0.05). Odds thatOdds a SNP alters function. single a from alleles four the took we individuals, of pairTo compareeach Forpaireachofalleles,we determined onewhether datathe fitfrommodel 10.1038/nn.3598 ± s.e.m. and represent data from a minimum of three replicates, wherereplicates, three of minimum a from representdata and s.e.m. We of the sequences the nucleotide aligned F test against the vector control, vector the against test - -

35. 34. 33. 32. 31. population. Data collection and analysis were not randomized. were allele counts (0, 1 or 2) for the four alleles with MAF > 4% in the eral participantolfactory acuity and usage for the rating scale across subjects. The predictors We ranked each subject’s ratings of the odorants to control forthis analysis,differences we in performed gen a Monte-Carlo simulation using the data from ref. ception of the three whether the number of scores. GERP, PolyPhen and SIFT synonymous amino acid. We a used SNPnexus with allele an in occurred change functional a nonsynonymous that odds a the to with acid amino allele an in occurred variant) functional most the common to relative above, defined (as change functional a that odds the of Multiple linear regression model.

Chelala, C., Khan, A. & Lemoine, N.R. SNPnexus: a web database for functional for database web a SNPnexus: N.R. Lemoine, & A. Khan, C., Chelala, files. TAB-delimited generic from features sequence of retrieval fast Tabix: H. Li, acetylcholine muscarinic m3 the of state Activation H. Matsunami, & Y.R. Li, Zhuang, H. & Matsunami, H. Synergism of accessory factors in functional expression outer rod in rhodopsin of role the against Evidence R.S. Molday, & D.W. Laird, Bioinformatics annotation of newly discovered and public domain single nucleotide polymorphisms. Bioinformatics signaling. (2011). receptor odorant mammalian modulates receptor receptors. odorant mammalian of (1988). emn bnig o P cells. RPE to binding segment

25 27 in vitro , 655–661 (2009). 655–661 , (2011). 718–719 , OR10G4 agonists. To determine the minimum sample size for alleles significantly predicted participants’ per Multiple regression analysis was used to test J. Biol. Chem. Biol. J. net Ohhlo. i. Sci. Vis. Ophthalmol. Invest. 3 5 (Ensembl 63 build) to generate

282 nature nature , 15284–15293 (2007). 15284–15293 , NEUR c. Signal. Sci.

29 OSCI 419–428 ,

4 EN ra1 , C 9 E - - .