Odorant Receptors: A Molecular Basis for Odor Recognition PDF

Cell, Vol. 65, 175-187, April5, 1991,Copyright© 1991 by Cell Press A Novel Multigene Family May Encode Odorant Receptors: A Molecular Basis for Odor Recognition Linda Buck* and Richard Axel*l" the sense of smell may involve a large number of distinct *Departmen...

Cell, Vol. 65, 175-187, April5, 1991,Copyright© 1991 by Cell Press A Novel Multigene Family May Encode Odorant Receptors: A Molecular Basis for Odor Recognition Linda Buck* and Richard Axel*l" the sense of smell may involve a large number of distinct *Department of Biochemistry and Molecular Biophysics receptors each capable of associating with one or a small tHoward Hughes Medical Institute number of odorants. In either case, the brain must distin- College of Physicians and Surgeons guish which receptors or which neurons have been acti- Columbia University vated to allow the discrimination between different odorant New York, New York 10032 stimuli. Insight into the mechanisms underlying olfactory perception is likely to depend upon the isolation of the odorant receptors and the characterization of their diver- Summary sity, specificity, and patterns of expression. The primary events in odor detection occur in a special- The mammalian olfactory system can recognize and ized olfactory neuroepithelium located in the posterior re- discriminate a large number of different odorant mole- cesses of the nasal cavity. Three cell types dominate this cules. The detection of chemically distinct odorants epithelium (Figure 1A): the olfactory sensory neuron, the presumably results from the association of odorous sustentacular or supporting cell, and the basal cell, which Ugands with specific receptors on olfactory sensory is a stem cell that generates olfactory neurons throughout neurons. To address the problem of olfactory percep- life (Moulton and Beidler, 1967; Graziadei and Monti Grazi- tion at a molecular level, we have cloned and charac- adei, 1979). The olfactory sensory neuron is bipolar: a terized 18 different members of an extremely large dendritic process extends to the mucosal surface, where multigene family that encodes seven transmembrane it gives rise to a number of specialized cilia that provide an domain proteins whose expression is restricted to the extensive, receptive surface for the interaction of odors olfactory epithelium. The members of this novel gene with the cell. The olfactory neuron also gives rise to an family are likely to encode a diverse family of odorant axon that projects to the olfactory bulb of the brain, the receptors. first relay in the olfactory system. The axons of the olfac- tory bulb neurons, in turn, project to subcortical and corti- Introduction cal regions where higher-level processing of olfactory in- formation allows the discrimination of odors by the brain. In vertebrate sensory systems, peripheral neurons re- The initial events in odor discrimination are thought to spond to environmental stimuli and transmit these signals involve the association of odors with specific receptors on to higher sensory centers in the brain where they are pro- the cilia of olfactory neurons. Selective removal of the cilia cessed to allow the discrimination of complex sensory in- results in the loss of olfactory responses (Bronshtein and formation. The delineation of the peripheral mechanisms Minor, 1977). Moreover, in fish, whose olfactory system by which environmental stimuli are transduced into neural senses amino acids as odors, the specific binding of amino information can provide insight into the logic underlying acids to isolated cilia has been demonstrated (Rhein and sensory processing. Our understanding of color vision, Cagan, 1980, 1983). The cilia are also the site of olfactory for example, emerged only after the observation that the signal transduction. Exposure of isolated cilia from rat ol- discrimination of hue results from the blending of informa- factory epithelium to numerous odorants leads to the rapid tion from only three classes of photoreceptors (Rushton, stimulation of adenylyl cyclase and elevations in cyclic 1955, 1965; Wald et al., 1955; Nathans et al., 1986). The AMP (an elevation in inositol trisphosphate in response to basic logic underlying olfactory sensory perception, how- one odorant has also been observed) (Pace et al., 1985; ever, has remained elusive. Mammals possess an olfac- Sklar et al., 1986; Breer et al., 1990; Boekhoff et al., 1990). tory system of enormous discriminatory power (for reviews The activation of adenylyl cyclase is dependent on the see Lancet, 1986; Reed, 1990). Humans, for example, are presence of GTP and is therefore likely to be mediated thought to be capable of distinguishing among thousands by receptor-coupled GTP-binding proteins (G proteins) of distinct odors. The specificity of odor recognition is em- (Jones and Reed, 1989). Elevations in cyclic AMP, in turn, phasized by the observation that subtle alterations in the are thought to elicit depolarization of olfactory neurons by molecular structure of an odorant can lead to profound direct activation of a cyclic nucleotide-gated, cation-per- changes in perceived odor. meable channel (Nakamura and Gold, 1987; Dhallan et How are the diversity and specificity of olfactory percep- al., 1990). This channel is opened upon binding of cyclic tion accomplished? The detection of chemically distinct nucleotides to its cytoplasmic domain, and can therefore odorants presumably results from the association of odor- transduce changes in intracellular levelsof cyclic AM P into ous ligands with specific receptors on olfactory neurons, alterations in the membrane potential. which reside in a specialized epithelium in the nose. Since These observations suggest a pathway for olfactory sig- these receptors have not been identified, it has been diffi- nal transduction (Figure 1B) in which the binding of odors cult to determine how odor discrimination might be to specific surface receptors activates specific G proteins. achieved. It is possible that olfaction, by analogy with color The G proteins then initiate a cascade of intracellular sig- vision, involves only a few odor receptors, each capable of naling events leading to the generation of an action poten- interaction with multiple odorant molecules. Alternatively, tial that is propagated along the olfactory sensory axon Cell 176 A Figure 1. The Olfactory Neuroepithelium and Lumen a Pathway for Olfactory Signal Transduction (A) The olfactory neuroepithelium. The initial events in odor perception occur in the nasal Olfactorycilia < t Mucus cavity in a specialized neuroepithelium that is diagrammed here. Odors are believed to inter- act with specific receptors on the cilia of olfac- tory sensory neurons. The signals generated Jpportingcell by these initial binding events are propogated by olfactory neuron axons to the olfactory bulb. (13)A pathway of olfactory signal transduction. In this scheme the binding of an odorant mole- cule to an odor-specific transmembrane recep- tor leads to the interaction of the receptor with Olfactorysense a GTP-binding protein (G,(olf)). This interaction neuron in turn leads to the release of the GTP-coupled ~z subunit of the G protein, which then stimu- lates adenylyl cyclase to produce elevated lev- els of cAMP. The increase in cAMP opens cy- clic nucleotide-gated cation channels, thus causing an alteration in membrane potential. asalcell..... t membrane To olfactorybulb B Odorant Receptor Cyclic nucleotide 0 Adenylate gated channel cyclase ~-, Extracellular face / /.... L$~,',f_~_j_.~I Cytosolic face ~ / ~~/ Gs( ) ~ ~- GTPTP~/~ cJcAMP Na+~,~)cAMP ATP ~ to the brain, A number of neurotransmitter and hormone Results receptors that transduce intracellular signals by activation of specific G proteins have been identified. Gene cloning Experimental Strategy has demonstrated that each of these receptors is a mem- The experimental design we employed to isolate genes ber of a large superfamily of surface receptors that tra- encoding odorant receptors was based on three assump- verse the membrane seven times (for reviews see O'Dowd tions: First, the odorant receptors are likely to belong to et al., 1989b; Strader etal., 1989). The pathway of olfactory the superfamily of receptor proteins that transduce intra- signal transduction (Figure 1B) predicts that the odorant cellular signals by coupling to GTP-binding proteins. Sec- receptors might also be members of this superfamily of ond, the large number of structurally distinct odorous mol- receptor proteins. The detection of odors in the periphery ecules suggests that the odorant receptors themselves is therefore likely to involve signaling mechanisms shared should exhibit significant diversity and are therefore likely by other hormone or neurotransmitter systems, but the to be encoded by a multigene family. Third, expression of vast discriminatory power of the olfactory system will re- the odorant receptors should be restricted to the olfactory quire higher-order neural processing to permit the percep- epithelium. tion of individual odors. To address the problem of olfac- To identify molecules in the olfactory epithelium that tory perception at a molecular level, we have cloned and resemble members of the seven transmembrane domain characterized 18 different members of an extremely large superfamily, homologs of this gene superfamily were am- multigene family that encodes seven transmembrane do- plified from olfactory epithelium RNA using the polymer- main proteins whose expression is restricted to the olfac- ase chain reaction (PCR). We then asked whether any of tory epithelium. The members of this novel gene family are the PCR products we obtained consisted of a mixture of likely to encode the individual odorant receptors. DNA sequences, consistent with the amplification of mem- Candidate Odorant Receptors 177 Figure 2. A PCR Amplification Product Con- A M 1 2 34 5 6 M7 8 9 1011M M 1213 14 1515 M 17 18 19 20 21 22 M taining Multiple Species of DNA cDNA prepared from olfactory epithelium RNA was subjected to PCR amplification with a se- ries of different primer oligonucleotidas; the DNA products of appropriate size were iso- lated, further amplified by PCR, and size frac- tionated on agarose gels (A) (for details see text). Each of these semipurified PCR products was digested with the restriction enzyme Hinfl B M 1 2 3 4 5 6 M 7 8 9 1011M M 1213 14 1516 M 17 18 19 20 21 22 M and analyzed by agarose gel electrophoresis (B). Lanes M contain size markers of 23.1,9.4, 6.6, 4.4, 2.3, 2.0, 1.35, 1.08, 0.87, 0.60, 0.31, 0.28, 0.27, 0.23, 0.19, 0.12, and 0.07 kb. Twenty-two of the 64 PCR products that were isolated and digested with Hinfl are shown here. Digestion of one of these, PCR 13, yielded a large number of fragments whose sizes summed to a value much greater than that of the undigested PCR 13 DNA, indicating that PCR 13 might contain multiple species of DNA that are representatives of a multigene family. bers of a multigene family. We reasoned that restriction We next asked whether any of these discrete PCR prod- digestion of an individual PCR product consisting of a sin- ucts consisted of multiple DNA sequences reflecting the gle species of DNA would generate a set of DNA fragments amplification of a large family of genes. The isolated PCR whose molecular weights sum to the molecular weight of products were digested with Haelll or Hinfl, which recog- the original PCR product. On the other hand, if an individ- nize four base restriction sites and cut DNA at frequent ual PCR product consisted of several different DNA se- intervals. In most instances, digestion of the PCR product quences (consistent with the amplification of members of with Hinfl generated a set of fragments whose molecular a mult!gene family), restriction endonuclease digestion weights sum to the size of the original DNA (Figure 2B). should generate a large array of fragments whose molecu- These PCR bands are therefore each likely to contain a lar weights sum to far greater than the molecular weight single DNA species. In some cases, however, restriction of the original PCR product. In this manner, we identified digestion yielded a series of fragments whose molecular a multigene family that encodes a large family of proteins weights sum to a value greater than that of the original belonging to the seven transmembrane domain receptor PCR product. The most dramatic example is shown in superfamily and whose expression is restricted to the ol- Figure 2B, where the PCR 13 DNA (710 bp) is cleaved by factory epithelium. Hinfl to yield a very large number of restriction fragments whose sizes sum to a value 5- to 10-fold greater than that Cloning the Gene Family of the original PCR product. These observations indicated In initial experiments we designed a series of degenerate that PCR product 13 consists of a number of different oligonucleotides that could anneal to conserved regions species of DNA, each of which could be amplified with the of members of the superfamily of G protein-coupled seven same pair of primer oligonucleotides. In addition, when transmembrane domain receptor genes. Five degenerate PCR experiments similar to those described were per- oligonucleotides (A1-A5; see Experimental Procedures) formed using cDNA library DNAs as templates, a 710 bp matching sequences within transmembrane domain 2 and PCR product was obtained with the PCR 13 primer pair six degenerate oligonucleotides (B1-B6) matching trans- (A4/B6) with DNA from olfactory cDNA libraries, but not membrane domain 7 were used in all combinations in PCR from a glioma cDNA library. Moreover, digestion of this reactions to amplify homologous sequences in cDNA pre- 710 bp PCR product also revealed the presence of multiple pared from rat olfactory epithelium RNA. The amplifica- DNA species. In other cases (see PCR product 20, for tion products of each PCR reaction were then analyzed example), digestion yielded a series of restriction frag- by agarose gel electrophoresis. Multiple bands were ob- ments whose molecular weights also sum to a size greater served with each of the primer combinations. The PCR than the starting material. Further analysis, however, re- products within the size range expected for this family of vealed that the original PCR product consisted of multiple receptors (600 to 1300 bp) were subsequently picked and bands of similar but different sizes. amplified further with the appropriate primer pair in order To determine whether the multiple DNA species present to isolate individual PCR bands. Sixty-four PCR bands in PCR 13 encode members of a family of seven trans- isolated in this fashion revealed only one or a small number membrane domain proteins, PCR 13 DNA was cloned into of bands upon agarose gel electrophoresis. Representa- the plasmid vector Bluescript, and five individual clones tives of these isolated PCR products are shown in Fig- were subjected to DNA sequence analysis. Each of the ure 2A. five clones exhibited a different DNA sequence, but each Cell 178 The Protein Sequences of Numerous, Olfactory-Specific Members of the Seven Transmembrane Domain Superfamily Numerous clones were obtained upon screening cDNA 0 libraries constructed from olfactory epithelium or olfactory neuron RNA at high stringency (see Experimental Proce- 5.0 - 1 dures). Partial DNA sequences were obtained from 36 2.0 - clones; 18 of these cDNA clones are different, but all of them encode proteins that exhibit shared sequence motifs indicating that they are members of the family identified in PCR 13 DNA. A complete nucleotide sequence was determined for the coding regions of ten of the most diver- Figure 3. Northern Blot Analysis with a Mixture of 20 Probes gent clones (Figure 4). The deduced protein sequences of One microgram of poly(A) ÷ RNA isolated from rat olfactory epithelium, brain, or spleen was size fractionated in formaldehyde-agarose, blot- these cDNAs defines a new multigene family that shares ted onto a nylon membrane, and hybridized with a ~P-labeled mixture sequence and structural properties with the seven trans- of segments of 20 cDNA clones. The DNA segments were obtained by membrane domain superfamily of neurotransmitter and PCR using primers homologous to transmembrane domains 2 and 7. hormone receptors. This novel family, however, exhibits features different from any other member of the receptor superfamily thus far identified. Each of the ten sequences contains seven hydrophobic encoded a protein that displayed conserved features of stretches (19-26 amino acids) that represent potential the superfamily of seven transmembrane domain receptor transmembrane domains. These domains constitute the proteins. In addition, the proteins encoded by all five clones regions of maximal sequence similarity to other members shared distinctive sequence motifs not found in other su- of the seven transmembrane domain superfamily (see leg- perfamily members, indicating they were all members of end to Figure 4). On the basis of structural homologies a new family of receptors. with rhodopsin and the ~-adrenergic receptors (O'Dowd et To obtain full-length cDNA clones, cDNA libraries pre- al., 1989b), it is likely that the N-termini of the olfactory pared from olfactory epithelium RNA or from RNA of an proteins are located on the extracellular side of the plasma enriched population of olfactory sensory neurons were membrane and the C-termini in the cytoplasm. In this screened. The probe used in these initial screens was scheme, three extracellular loops alternate with three in- a mixture of PCR 13 DNA as well as DNA obtained by tracellular loops to link the seven transmembrane domains amplification of rat genomic DNA or DNA from two olfac- (see Figure 5). Analysis of the sequences in Figure 4 tory cDNA libraries with the same primers used to generate demonstrates that the olfactory proteins, like other mem- PCR 13 (A4 and B6 primers). Hybridizing plaques were bers of the receptor superfamily, display no evidence of subjected to PCR amplification with the A4/B6 primer set, an N-terminal signal sequence. As in several other super- and only those giving a PCR product of the appropriate family members, a potential N-linked glycosylation site is size (approximately 710 bp) were purified. The frequency present in all ten proteins within the short N-terminal extra- of such positive clones in the enriched olfactory neuron cellular segment. Other structural features conserved with cDNA library was approximately 5 times greater than the previously identified members of the superfamily include frequency in the olfactory epithelium cDNA library. The cysteine residues at fixed positions within the first and increased frequency of positive clones observed in the second extracellular loops, which are thought to form a olfactory neuron library is comparable to the enrichment disulfide bond. Finally, many of the olfactory proteins re- in olfactory neurons generally obtained in the purification veal a conserved cysteine within the C-terminal domain procedure. that may serve as a palmitoylation site anchoring this do- The original pair of primers used to amplify PCR 13 DNA main to the membrane (O'Dowd et al., 1989a). These fea- was then used to amplify coding segments of 20 cDNA tures, taken together with several stl0rt, conserved se- clones. A mixture of these PCR products was labeled and quence motifs (see legend to Figure 4), clearly define this used as probe for further cDNA library screens. This mixed new family as a member of the superfamily of genes en- probe was also used in a Northern blot (Figure 3) to deter- coding the seven transmembrane domain receptors. mine whether the expression of the gene family is re- There are, however, important differences between the stricted to the olfactory epithelium. The mixed probe de- olfactory protein family and the other seven transmem- tects two diffuse bands centered at 2 and 5 kb in RNA from brane domain proteins described previously, and these olfactory epithelium; no hybridization can be detected in differences may be relevant to a proposed function of brain or spleen. (Later experiments, which examined a these proteins in odor recognition. Structure-function ex- larger number of tissue RNAs with a more restricted probe, periments involving in vitro mutagenesis suggest that ad- will be shown below.) Taken together, these data indicate renergic ligands interact with this class of receptor mole- that we have identified a novel multigene family encoding cule by binding within the plane of the membrane (Kobilka seven transmembrane domain proteins that are expressed et al., 1988; Strader et al., 1989). Not surprisingly, small in olfactory epithelium, and could be expressed predomi- receptor families that bind the same class of ligands, such nantly or exclusively in olfactory neurons. as the adrenergic and muscarinic acetylcholine receptor  Candidate Odorant Receptors 179 II F3 82 F5 82 F6 MAW 85 F12 M 83 13 80 17 B3 18 80 I9 82 114 82 115 82 Ill IV F~.... 163 F5.... 163 F6.... 166 F12.... 16~ 13.... 161 17 AGFI 168 18.... 161 19.... 163 114.... 163 115.... 163 VI F3 LAEPF::~:THLE Z P H Y ~ EPN QV I Q;LT~:$::~AFL~D~V I Y FT LV LL 248 F5 248 F6 251 F12 249 I3 246 I7 253 I8 246 I9 24B I14 248 I15 2~B Vll F3 SL:FYCTGL6 EEViRSPPS LLHFFLV LCHLPC FI FCY 333 F5 CLFY:~I V i A ,MRFPSKo 313 F6 LIWY:~ST|F :GK 311 F12 SLF~S ~ GL6 6NCKVHHWTG 317 I3 SLF::Y~¥11G SMKITL 310 17 IIFyAASIF LAODOEANTNKGSKIG 327 SKK LPW 312 9 s KKO!PSFL 314 I14 TEFY~TIF(~ TK K:%:::SL 312 Ii 5 S:Le:V:~] i:6 KKKI:TFCL 314 Figure 4. The Protein Sequences Encoded by Ten Divergent cDNA Clones Ten divergent cDNA clones were subjected to DNA sequence analyses, and the protein sequence encoded by each was determined. Amino acid residues conserved in 60% or more of the proteins are shaded. The presence of seven hydrophobic domains (I-VII), as well as short conserved motifs shared with other members of the superfamily, demonstrates that these proteins belong to the seven transmembrane domain protein superfamily. Motifs conserved among members of the superfamily and the family of olfactory proteins include the GN in TM1 (transmembrane domain 1), the central W of TM4, the Y near the C-terminal end of TM5, and the NP in TM7. In addition, the DRY motif C-terminal to TM3 is common to many members of the G protein-coupled superfamily. However, all of the proteins shown here share sequence motifs not found in other members of this superfamily and are clearly members of a novel family of proteins. The nucleotide sequences from which these protein sequences were derived have been deposited in GenBank. families, exhibit maximum sequence conservation (often erate intracellular signals. In vitro mutagenesis experi- over 80%) within the transmembrane domains. In con- ments indicate that one site of association between trast, the family of receptors we have identified shows strik- receptor and G protein resides within the third cytoplasmic ing divergence within the third, fourth, and fifth transmem- loop (Kobilka et al., 1988; Hamm et al., 1988). The se- brane domains (Figure 4). The variability in the three quence of this cytoplasmic loop in 18 different clones we central transmembrane domains is highlighted schemati- have characterized is shown in Figure 6A. This loop, which cally in Figure 5. The divergence in potential ligand- is often quite long and of variable length in the receptor binding domains is consistent with the idea that the family superfamily, is relatively short (only 17 amino acids) and of molecules we have cloned is capable of associating with of fixed length in the 18 clones examined. Interestingly, 11 a large number of odorants of diverse molecular struc- of the 18 different clones exhibit the sequence motif tures. (K/R)IVSSI (or a close relative) at the N-terminus of this Receptors that belong to the superfamily of seven trans- loop. Two of the cDNA clones reveal a different motif, membrane domain proteins interact with G proteins to gen- HIT(C/W)AV, at this site. If this short loop is a site of contact Cell 180 Figure 5. Positions of Greatest Variability in the Olfactory Protein Family In this diagram the protein encoded by cDNA clone 115 is shown traversing the plasma mem- brane seven times, with its N-terminus located extracellularly and its Coterminus intracellu- lady. The vertical cylinders delineate the seven putative c~helices spanning the membrane. Po- sitions at which 60% or more of the 10 clones shown in Figure 4 share the same residue as 115 are shown as white balls. More variable residues are shown as black balls. The high degree of variability encountered in transmem- brane domains III, IV, and V is evident in this schematic. with G proteins, it is possible that the conserved motifs other G protein-coupled receptors, these residues may may reflect sites of interaction with different G proteins represent sites of phosphorylation for specific receptor ki- that activate different intracellular signaling systems in re- nases involved in desensitization (Bouvier et al., 1988). sponse to odors. In addition, the putative receptors we have cloned reveal several conserved serine or threonine S u b f a m i l i e s w i t h i n t h e M u l t i g e n e Family residues within the third cytoplasmic loop. By analogy with Figure 6A displays the sequences of the fifth transmem- Figure 6. The Presence of Subfamilies in a Di- vergent Multigene Family A F2 RVNE VV FIVVSLFLVL:P:FAL[IMSYV R:I:V:$:$i L KVP$ SQGI YK F3 F L~D ~V~Y F T LV L LATVI~LAG ~ F Y ~ F K~~ CA ! SSVH~K Partial nucleotide sequences and deduced F5 H L ~ ~ M ~ L T E G A V V M V T P F V C L! I HXTC A V l R V S~ P R6 G wE protein sequences were obtained for f 8 differ- F6 QVVE ~VS FG I AFCV I LG$CG IT L V ~ A Yt I TT|I K I P S A R G R H R ent cDNA clones. Transmembrene domain V F7 HVN~ILV~FVMGGIILVI~FVLIIV$!VVR ~!V~iLKVPSAR~IRK along with the flanking loop sequences, includ- F8 FPsH ~TMHLVPVILAAISLSG~LY~i~FK ~ Y ~ > | R S M S S V Q G K Y K ing the entire cytoplasmic loop between trans- FPSH ~IMNLVPVMLAAI S~SGWLY~F K S I S T V O~ K Y membrane domains V and VI, is shown here F12 FPSH LIMN LVPVMLAAI S~SGI L Y ~ F K VKBKYK for each protein. Amino acid residues found in F13 F23 F L~ D V z MY FA LV L L^ VV~ L L~ ,Y S~S K VQGKYK 60% or more of the clones in a given position F AEGRRE are shaded (A). This region of the olfactory pro- F2q HEI~ M I | LVLAAFNLI S$ L L V V L V ~ L teins (particularly transmembrane domain V) 13 YINE L M ! F I M S T L L I I I ~ @ F L ~ V M ~ A RiI LKVPSTQ6ICK appears to be highly variable (see Figure 4). I7 STA( ~ T D F V L A I F I L L G ~ L S V T G A ~ M AiT RI P~AA~RH~ These proteins, however, can be grouped into 18 YV~ ~M,HIMGVIIIVI~VL~VI~A Ki ! ~ L K V P ~ T O S XHK subfamilies (B-D) in which the individual sub- 19 HD~ ~A~FILGGPIVV LL|IV~YA R~i~|~FKVP~SQSIHK family members share considerable homology 111 HLNE LM LTEGAVVMVTPFVCILI~I H~TWAVLRVS~PR~GW~ in this divergent region of the protein. 112 FPSH EIMNLVPVMLGAISLSG~LY~F K~=~VR$I$~V0~KHK 114 YVNE LMIYILGGLIIIIPFLLIVM$~V R!FF$:ILKFP$IQDIYK 115 HVNE LVIFVMGGLVIVIPFVLIIVSYA RVVASrLKVPSVRGIHK B F12 F13 F~ F8 F~ 112 F F23 F F3 F C. F7 HVNE LVIFVM6GXILVXPFVLXIVSYV RIV$SI~Y~$ARG IRK 115 HVNE V IF VMGfiLV | V x PPv L! I v $ ~ R Vv A$~ v RG~H 13 Y zNE L. X F z. S T L L i Z i ~ F ~ ! V. S V A ~T~G~C~ 18 YVNE LMIH!MGVIiiV!PFV[IVISyA KiISS ~T~S~H~ 19 HONE LA!FI IVVL~ELL!IVSVA 114 YV,E L. ~Y t t. L ZI ~ LL.SV V R ~FF S ~ F ~ , ~ D ~Y D F5 111 CandidateOdorant Receptors 181 1 2 3 4 5 6 7 1-7 A S A B A B A A B A e A B A S "i 23.1 - :4 l g.4- 1 w 6.6- " W~ t W~ w 4.3- iw O.... i : l 2.3- 2.0-.- i , g 1.4 - 1.0- 0.9- Figure 7. SouthernBlot Analyseswith Non-Cross-HybridizingFragmentsof DivergentcDNAs Five microgramsof rat liver DNA was digestedwith EcoRI(A) or Hindlll (B), electrophoresedin 0.75% agarose,blottedonto a nylon membrane, and hybridizedwith the ~P-labeledprobesindicated.The probes usedwere PCR-generatedfragmentsof: 1, clone F9 (identicalto F12 in Figure 4); 2, F5; 3, F6; 4, 13;5, 17;6, 114;or 7, 115.The lane labeled"1-7" was hybridizedto a mixtureof the sevenprobes.The probesusedshowedeither no cross-hybridizationor only trace cross-hybridizationwith one another.The size markerson the left correspondto the four blots on the left (1- 4), whereasthe marker positionsnoted on the right correspondto the four blots on the right (5-7 and "1-7"). brane domain and the adjacent cytoplasmic loop encoded cDNA clones (Figure 4). In initial experiments, these DNAs by 18 of the cDNA clones we have analyzed. As a group, were labeled and hybridized to each other to define condi- the 18 sequences exhibit considerable divergence within tions under which minimal cross-hybridization would be this region. The multigene family, however, can be divided observed among the individual clones. At 70°C the seven into subfamilies such that the members of a given subfam- DNAs showed no cross-hybridization, or cross-hybridized ily share significant sequence conservation. Analysis of only very slightly. The trace levels of cross-hybridization the sequences in Figure 6A defines at least three subfamil- observed are not likely to be apparent upon genomic ies of related sequences (Figures 6B-6D). Subfamily B, Southern blot analysis, where the amounts of DNA are far for example, consists of six closely related sequences in lower than in the test cross. which pairs of sequences can differ from one another at Probes derived from these seven DNAs were annealed only four of 44 positions (91% identity) (see F12 and F13). under stringent conditions, either individually or as a The sequences encoded by clones F5 and I11 (subfamily group, to Southern blots of rat liver DNA digested with D), which differ at only one residue, differ from F12 and the restriction endonuclease EcoRI or Hindlll (Figure 7). F13 (subfamily B) at 34-36 of the 44 positions within this Examination of the Southern blots reveals that all but one region and clearly define a separate subfamily. It is possi- of the DNAs detects a relatively large, distinctive array of ble that the divergent subfamilies encode receptors that bands in genomic DNA. Clone 115 (probe 7), for example, bind odorants of widely differing molecular structures. detects about 17 bands with each restriction endonucle- Members of the individual subfamilies could therefore rec- ase, whereas clone F9 (probe 1) detects only about 5-7 ognize more subtle differences between molecules that bands with each enzyme. A single band is obtained with belong to the same structural class of odorant molecules. clone 17(probe 5). PCR experiments using nested primers (TM2/TM7 primers followed by primers to internal se- The Size of the Multigene Family quences) and genomic DNA as template indicate that the We have performed genomic Southern blotting experi- coding regions of the members of this multigene family, ments and have screened genomic libraries to obtain an like those of many members of the G protein-coupled su- estimate of the sizes of the multigene family and the mem- perfamily, may not be interrupted by introns. This observa- ber subfamilies encoding the putative odor receptors. tion, together with the fact that most of the probes encom- DNAs extending from the 3' end of transmembrane do- pass only 400 nucleotides, suggests that each band main 3 to the middle of transmembrane domain 6 were observed in these experiments is likely to represent a dif- synthesized by PCR from DNA of seven of the divergent ferent gene. These data suggest that the individual probes Cell 182 we have chosen are representatives of subfamilies that O range in size from a single member to as many as 17 ~ ~ z members. We detect a total of about 70 individual bands < = < = o < -- ; in this analysis, which could represent the presence of at least 70 different genes. Although the DNA probes used in these blots did not cross-hybridize appreciably with each other, it is possible that a given gene might hybridize to 5.0 - more than one probe, resulting in an overestimate of gene 2.0- J number. However, it is probable that the total number of bands reflects only a minimal estimate of gene number since it is unlikely that we have isolated representative cDNAs from all of the potential subfamilies and the hybrid- izations were performed under conditions of very high stringency. Figure 8. Northern Blot Analysis with a Mix of Seven Divergent Clones A more accurate estimate of the size of the olfactory- One microgram of poly(A)÷ RNA from each of the tissues shown was specific gene family was obtained by screening rat geno- size fractionated, blotted onto a nylon membrane, and hybridized with a ~P-labeled mixture of segments of seven divergent cDNA clones mic libraries. The mix of the seven divergent probes used (see legend to Figure 7). Examination of 28S and 18S rRNAs and in Southern blots, or the mix of 20 different probes used control hybridization with a mouse actin probe confirmed the integrity in our initial Northern blots (see Figure 3), was used as of the RNAs used. hybridization probe under high (65°C) or lowered (55°C) stringency conditions in these experiments. Nested PCR (see above) was used to verify that the clones giving a An estimate of the level of expression of this family can positive signal under low-stringency annealing conditions be obtained from screens of cDNA libraries. The frequency were indeed members of this gene family. We estimate of positive clones in cDNA libraries made from olfactory from these studies that there are between 100 and 200 epithelium RNA suggests that the abundance of the RNAs positive clones per haploid genome. The estimate of the in the epithelium is about one in 20,000. The frequency of size of the family we obtain from screens of genomic librar- positive clones is approximately 5-fold higher in a cDNA ies again represents a lower limit. Given the size of the library prepared from RNA from purified olfactory neurons multigene family, we might anticipate that many of these (in which approximately 75% of the cells are olfactory neu- genes are linked such that a given genomic clone may rons). The increased frequency of positive clones obtained contain multiple genes. Thus the data from Southern blot- in the olfactory neuron cDNA library is comparable to the ting and screens of genomic libraries indicate that the enrichment we obtain upon purification of olfactory neu- multigene family we have identified consists of 100 to sev- rons. These observations suggest that this multigene fam- eral hundred member genes that can be divided into multi- ily is expressed largely, if not solely, in olfactory neurons ple subfamilies. and may not be expressed in other cell types within the It should be noted that the cDNA probes we have iso- epithelium. If each olfactory neuron contains 10s mRNA lated may not be representative of the full complement of molecules, from the frequency of positive clones we pre- subfamilies within the larger family of olfactory proteins. dict that each neuron contains only 25-30 transcripts de- The isolation of cDNAs, for example, relies heavily on PCR rived from this gene family. Since the family of olfactory with primers from transmembrane domains 2 and 7 and proteins consists of a minimum of 100 genes, a given olfac- biases our clones for homology within these regions. Thus, tory neuron could maximally express only a proportion of estimates of gene number as well as subsequent esti- the many different family members. These values thus mates of RNA abundance should be considered as mini- suggest that olfactory neurons will exhibit significant diver- mal. We anticipate that further characterization of cDNA sity at the level of expression of these olfactory proteins. clones will identify representatives of new subfamilies and that this family of putative odorant receptors may be ex- Discussion tremely large. The mammalian olfactory system can recognize and dis- Expression of the Members of criminate a large number of odorous molecules. Percep- This Multigene Family tion in this system, as in other sensory systems, initially We have performed additional Northern blot analyses to involves the recognition of external stimuli by primary sen- demonstrate that expression of the members of this gene sory neurons. This sensory information is then transmitted family is restricted to the olfactory epithelium (Figure 8). to the brain, where it is decoded to permit the discrimina- Northern blot analysis with a mixed probe consisting of the tion of different odors. Elucidation of the logic underlying seven divergent cDNAs used above reveals two diffuse olfactory perception is likely to require the identification of bands about 5 and 2 kb in length in olfactory epithelium the specific odorant receptors, the analysis of the extent RNA. This pattern is the same as that seen previously with of receptor diversity and receptor specificity, as well as an the mix of 20 DNAs. No annealing is observed to RNA from understanding of the pattern of receptor expression in the the brain or retina or other, nonneural tissues, including olfactory epithelium. lung, liver, spleen, and kidney. What are the expected properties of odorant receptors, CandidateOdorantReceptors 183 and what is the evidence that the multigene family we have groups including camphoraceous, musky, pepperminty, identified encodes these receptors? First, the odorant re- ethereal, pungent, and putrid. In such a model, each group ceptors are thought to transduce intracellular signals by would contain odorants with common molecular configura- interacting with G proteins, which activate second mes- tions that bind to common receptors and share similar odor senger systems (Pace et al., 1985; Sklar et al., 1986; Jones qualities. and Reed, 1989; Breer et al., 1990; Boekhoff et al., 1990). We have provided a minimum estimate of the size of Although we have not demonstrated that the olfactory pro- the repertoire of the putative odorant receptors in the rat. teins we have identified can indeed trigger G protein-cou- Screens of genomic libraries with mixed probes consisting pled responses, these proteins are clearly members of the of divergent family members detect approximately 100 to family of G protein-coupled receptors, which traverse the 200 positive clones per genome. The present estimate of membrane seven times (O'Dowd et ai., 1989b). Second, at least 100 genes provides only a lower limit since it is the odorant receptors should be expressed specifically likely that our probes do not detect all of the possible sub- in the tissue in which odorants are recognized. We have families. Moreover, it is probable that many of these genes shown that the family of olfactory proteins we have cloned are linked such that a given genomic clone may contain is expressed in the olfactory epithelium. Hybridizing RNA multiple genes. We therefore expect that the actual size is not detected in brain or retina, nor in a host of nonneural of the gene family may be considerably higher and that tissues. Moreover, expression of this gene family in the this family of putative odorant receptors could constitute epithelium may be restricted to olfactory neurons. Third, one of the largest gene families in the genome. the family of odorant receptors must be capable of inter- The characterization of a large multigene family encod- acting with extremely diverse molecular structures. The ing putative odorant receptors suggests that the olfactory genes we have cloned are members of an extremely large system utilizes a far greater number of receptors than the multigene family that exhibits variability in regions thought visual system. Color vision, for example, allows the dis- to be important in ligand binding. The possibility that each crimination of several hundred hues but is accomplished member of this large family of seven transmembrane pro- by only three different photoreceptors (Rushton, 1955, teins is capable of interacting with only one or a small 1965; Wald et al., 1955; Nathans et al., 1986). The photore- number of odorants provides a plausible mechanism to ceptors each have different, but overlapping, absorption accommodate the diversityof odor perception. Finally, the spectra that cover the entire spectrum of visible wave- odorant receptors must bind odorants and transduce an lengths. Discrimination of color results from comparative intracellular signal leading to the activation of second mes- processing of the information from these three classes of senger systems. This criterion, at present, is difficult to photoreceptors in the brain. Whereas three photorecep- satis~ eXperimentally because the diversity of odorants tors can absorb light across the entire visible spectrum, and the large number of individual receptors make it diffi- our data suggest that a small number of odorant receptors cult to identify the appropriate ligand for a particular recep- cannot recognize and discriminate the full spectrum of tor. Nonetheless, the properties of the gene family we have distinct molecular structures perceived by the mammalian identified suggest that this family is likely to encode a large olfactory system. Rather, olfactory perception probably number of distinct odorant receptors. employs an extremely large number of receptors each ca- pable of recognizing a small number of odorous ligands. How Large Is the Multigene Family? How many structurally distinct odors can an organism de- Diversity within the Gene Family and the Specificity tect, and how large is the receptor gene family? The size of Odor Recognition of the receptor repertoire is likely to reflect the range of The olfactory proteins we have identified are clearly mem- detectable odors and the degree of structural specificity bers of the superfamily of receptors that traverse the mem- exhibited by the individual receptors. It is difficult to assess brane seven times. Analysis of the proteins encoded by the discriminatory capacity of the olfactory system accu- the 18 distinct cDNAs we have cloned reveals structural rately, but it has been estimated that humans can identify features that may render this family particularly well suited over 10,000 structurally distinct odorous ligands. How- for the detection of a diverse array of structurally distinct ever, this does not necessarily imply that humans possess odorants. I~xperiments with other members of this class of an equally large repertoire of odorant receptors. For exam- receptors suggest that ligand binds to its receptor within ple, binding studies in lower vertebrates suggest that the plane of the membrane such that the ligand contacts structurally related odorants may activate the same recep- many, if not all, of the transmembrane helices (Strader tor molecules. In fish that smell amino acids, the binding et al., 1989; Kobilka et al., 1988). The family of olfactory of alanine to isolated cilia can be competed by other small proteins can be divided into several different subfamilies polar residues (threonine and serine) but not by the basic that exhibit significant sequence divergence within the amino acids lysine and arginine (Rhein and Cagan, 1983). transmembrane domains. Nonconservative changes are These data suggest that individual receptors are capable commonly observed within blocks of residues in trans- of associating with several structurally related ligands, al- membrane regions 3, 4, and 5 (Figures 4-6); these blocks beit with different affinities. Stereochemical models of ol- could reflect the sites of direct contact with odorous li- factory recognition in mammals (Amoore, 1982) (based gands. Some members, for example, have acidic residues largely on psychophysical rather than biophysical data) in transmembrane domain 3, which in other families are have suggested the existence of several primary odor thought to be essential for binding aminergic ligands Cell 184 (Strader et al., 1987), while other members maintain hy- organization of framework and hypervariable domains drophobic residues at these positions. This divergence within the families of immunoglobulin and T cell receptor within transmembrane domains may reflect the fact that variable region sequences (for reviews see Tonegawa, the members of the family of odorant receptors must asso- 1983; Hood et al., 1985). This analogy goes beyond struc- ciate with odorants of widely different molecular struc- tural organization and may extend to the function of these tures. gene families: each family consists of a large number These observations suggest a model in which each of of genes that have diversified over evolutionary time to the individual subfamilies encodes receptors that bind dis- accommodate the binding of a highly diverse array of li- tinct structural classes of odorant. Within a given subfam- gands. The evolutionary mechanisms responsible for the ily, however, the sequence differences are far less dra- diversification and maintenance of these large gene fami- matic and are often restricted to a small number of lies may also be similar. It has been suggested that gene residues. Thus, the members of a subfamily may recog- conversion has played a major role in the evolution of nize more subtle variations among odor molecules of a immunoglobulin and T cell receptor variable domains (Bal- given structural class. At a practical level, individual sub- timore, 1981; Egel, 1981; Flanagan et al., 1984). Analysis families may recognize grossly different structures such of the sequences of the putative olfactory receptors re- that one subfamily may associate, for example, with the veals at least one instance where a motif from a variable aromatic compound benzene and its derivatives, whereas region of one subfamily is found embedded in the other- a second subfamily may recognize odorous, short-chain wise divergent sequence of a second subfamily, sug- aliphatic molecules. Subtle variations in the structure of gesting that conversion has occurred. Such a mixing of the receptors within, for example, the hypothetical ben- motifs from one subfamily to another over evolutionary zene subfamily could facilitate the recognition and discrim- time would provide additional combinatorial possibilities ination of various substituted derivatives such as toluene, leading to the generation of diversity. xylene, or phenol. It should be noted that such a model, It should be noted, however, that the combinatorial join- unlike previous stereochemical models, does not neces- ing of gene segments by DNA rearrangement during de- sarily predict that molecules with similar structures will velopment, which is characteristic of immunoglobulin loci have similar odors. The activation of distinct receptors with (Tonegawa, 1983), is not a feature of the putative odor similar structures could elicit different odors, since per- receptor gene family. We have observed no evidence for ceived odor will depend upon higher-order processing of DNA rearrangement to generate the diversity of genes we primary sensory information. have cloned. We have sequenced the entire coding region along with parts of the 5' and 3' untranslated regions of Evolution of the Gene Family and the Generation ten different cDNA clones. The sequences of the coding of Diversity regions are all different; we have not obtained any evi- Preliminary evidence from PCR analyses suggests that dence for constant regions that would suggest DNA re- members of this family of olfactory proteins are conserved arrangements of the sort seen in the immune system. in lower vertebrates as well as invertebrates. This gene These observations indicate that the diverse olfactory pro- family presumably expanded over evolutionary time, pro- teins are coded by a large number of distinct gene se- viding mammals with the ability to recognize an increasing quences. diversity of odorants. Examination of the sequences of Although it is unlikely from our data that DNA re- the family members cloned from mammals provides some arrangement is responsible for the generation of diversity insight into the evolution of this multigene family. Although among the putative odorant receptors, it remains possible we have not yet characterized the chromosomal loci en- that DNA rearrangements may be involved in the regula- coding these genes, it is likely that at least some member tion of expression of this gene family. If each olfactory genes will be tandemly arranged in a large cluster as is neuron expresses only one or a small number of genes, observed with other large multigene families. A tandem then a transcriptional control mechanism must be opera- array of this sort provides a template for recombination tive to choose which of the more than 100 genes within the events, including unequal crossing over and gene conver- family will be expressed in a given neuron. Gene conver- sion, that can lead to expansion and further diversification sion from one of multiple silent loci into a single active of the sort apparent among the family members we have locus, as observed for the trypanosome variable surface cloned (for review see Maeda and Smithies, 1986). glycoproteins (Van der Ploeg, 1991), provides one attrac- The multigene family encoding the olfactory proteins is tive model. The gene conversion event could be stochas- large: all of the member genes clearly have a common tic, such that a given neuron could randomly express any ancestral origin but have undergone considerable diver- one of several hundred receptor genes, or regulated (per- gence such that individual genes encode proteins that haps by positional information), such that a given neuron share 40%-80% amino acid identity. Subfamilies are ap- could express only one or a small number of predeter- parent, with groups of genes sharing greater homology mined receptor types. Alternatively, it is possible that posi- among themselves than with members of other subfamil- tional information in the olfactory epithelium controls the ies. Examination of the sequences of even the most diver- expression of the family of olfactory receptors by more gent subfamilies reveals a pattern in which blocks of con- classical mechanisms that do not involve DNA rearrange- served residues are interspersed with variable regions. ment. Whatever mechanisms will regulate the expression This segmental homology is conceptually similar to the of receptor genes within this large multigene family, these Candidate Odorant Receptors 185 mechanisms must accommodate the requirement that ol- different odorants. The existence of specific odorant re- factory neurons are regenerated every 30-60 days (Grazi- ceptors, randomly distributed through the olfactory epithe- adei and Monti Graziadei, 1979), and therefore the expres- lium, which converge on a common target within the olfac- sion of the entire repertoire of receptors must be tory bulb, would raise additional questions about the accomplished many times during the life of an organism. recognition mechanisms used to guide these distinct axo- nal subsets to their central targets. Receptor Diversity and the Central Processing of Other sensory systems also spatially segregate afferent Olfactory Information input from primary sensory neurons. The spatial segrega- Our results suggest the existence of a large family of dis- tion of information employed by the visual and somatosen- tinct odorant receptors. Individual members of this recep- sory systems, for example, is used to define the location tor family are likely to be expressed by only a small set of of the stimulus within the external environment as well as the total number of olfactory neurons. The primary sensory to indicate the quality of the stimulus. In contrast, olfactory neurons within the olfactory epithelium will therefore ex- processing does not extract spatial features of the odorant hibit significant diversity at the level of receptor expres- stimulus. Relieved of the necessity to encode information sion. The question then emerges as to whether neurons about the spatial localization of the sensory stimulus, the expressing the same receptors are localized in the olfac- olfactory system of mammals may use the spatial segrega- tory epithelium. Does the olfactory system employ a topo- tion of sensory input solely to encode the identity of the graphic map to discriminate among the numerous odor- stimulus itself. The molecular identification of the genes ants? The spatial organization of distinct classes of likely to encode a large family of olfactory receptors should olfactory sensory neurons, as defined by receptor expres- provide initial insights into the underlying logic of olfactory sion, can now be determined by using the procedures of in processing in the mammalian nervous system. situ hybridization and immunohistochemistry with probes specific for the individual receptor subtypes. This informa- Experimental Procedures tion should help to distinguish between different models that have been proposed to explain the coding of diverse PCR odorant stimuli (for review see Shepherd, 1985). RNA was preparedfrom the olfactoryepitheliaof Sprague-Dawleyrats accordingto Chirgwinet al. (1979) or using RNAzol B (Cinna/Biotecx) In one model, sensory neurons that express a given and then treated with DNase I (0.1 U per rig of RNA) (Promega).To receptor and respond to a given odorant may be localized obtain cDNA, this RNA was incubatedat 0.1 p.g/~lwith 5 pM random within defined positions within the olfactory epithelium. hexamers(Pharmacia), 1 mM each of dATP, dCTP, dGTP, and TTP, This topographic arrangement would also be reflected in and 2 U/p,IRNaseinhibitor (Promega)in 10 mM Tris-HCI (pH 8,3), 50 mM KCI, 2.5 mM MgCI2,and 0,001% gelatin for 10 rain at 22°C, and the projection of olfactory sensory axons into discrete re- then for a further 45 rain at 37°C following the addition of 20 U/rd gions (glomeruli) within the olfactory bulb. In this scheme, Moloney murine leukemia virus reverse transcriptase (BRL). After the central coding to permit the discrimination of discrete heating at 95°C for 3 min, cDNA prepared from 0.2 pg of RNA was odorants would depend, in part, on the spatial segregation used in each of a series of PCR experimentscontaining 10 mM Tris- of different receptor populations. Attempts to discern the HCI (pH 8.3), 50 mM KCI, 1.5 mM MgCI2,0.001% gelatin,200 pM each of dATP, dCTP, dGTP, and TTP, 2.5 U of Taq polymerase(Perkin topographic localization of specific receptors at the level ElmerCetus), and 2 pM of each PCR primer. PCR amplificationswere of the olfactory epithelium has led to conflicting results. In performedaccordingto the following schedule: 96°C for 45 s, 55°C some studies, electrophysiological recordings have re- for 4 rain (or 45°C for 2 rain), and 72°C for 3 rain with 6 s extension vealed differences in olfactory responses to distinct odor- per cycle for 48 cycles. The primers used for PCR were a series of degenerate oligonucleotidesmade according to the amino acid se- ants in different regions of the olfactory epithelium quences found in transmembranedomains 2 and 7 of a variety of (Mackay-Sim et al., 1982; Thommesen and Doving, 1977). differentmembersof the seventransmembranedomainproteinsuper- However, these experiments have been difficult to inter- family(forexample,see O'Dowdet al., 1989b).The regionsusedcorre- pret since the differences in response across the epithe- spondto aminoacids 60-70 and 286-295 of clone 115(Figure4). Each lium are often small and are not observed in all studies (for of five different5' primerswas used in PCR reactionswith each of six different 3' primers. The 5' primers had the following sequences: example, see Sicard, 1985). A second model argues that sensory neurons express- AI: AA(TIC)T(GIA)(GIC)ATI(CIA)TI(GIC)TIAA(TIC)(CIT)TIGCIGTIG- ing distinct odorant receptors are randomly distributed in CIGA; the epithelium but that neurons responsive to a given odor- A2: AA(TIC)TA(TIC)TT(TIC)(CIA)TI(GIA)TIAA(TIC)CTIGCI(TIC)TIG- CIGA; ant project to restricted regions within the olfactory bulb. A3: AA (T/C)(T/C)(T/A)II-F(T/C)(NC) TIATI(T/A)CICTIGCIT(G/C)IG- In this instance, the discrimination of odors would be a CIGA; consequence of the position of second-order neurons in A4: (CIA)GrI-FI(CIT)TIATGTG(TIC)AA(CIT)CTI(TIA)(GIC)(CiT)TF(TI the olfactory bulb but would be independent of the site of C)GCIGA; origin of the afferent signals within the epithelium. Map- A5: ACIGTITA(TIC)ATIACICA(TIC)(CIT)TI(AIT)(CIG)IATIGCIGA. ping of the topographic projections of olfactory neurons The 3' primers were: has been performed by extracellular recordings from dif- ferent regions of the bulb (Thommesen, 1978; Doving et BI: CTGI(CIT)(GIT)(GIA)'I-FCATIA(AIT)I(NC)(CIA)(NG)TAIA(TIC)IA- al., 1980) and by 2-deoxyglucose autoradiography to map (T/C)IGG(G/A)TT; B2: (G/T)(NG) T (C/G)(G/A)TTIAG(NG) CA (NG) CA (NG)TAIATIA- regional activity after exposure to different odorants (Stew- TIGG(G/A)']-F; art et al., 1979). These studies suggest that spatially local- B3: TCIAT(GIA)I-F(NG)AAIGTIGT(NG)TAIATIATIGG(GIA)I-F; ized groups of bulbar neurons preferentially respond to B4: GC(ClT)I-FIGT(NG)AAIATIGC(NG)TAIAG(GIA)AAIGG(GIA)I-F; Cell 186 B5: AA(NG)TCIGG(GIA)(CIG)(TIA)ICGI(CIG)A(NG)TAIAT(CIG)AI. Lillian Eoyang for expert technical assistance. We are also grateful to IGG(G/A)TF; Andrew Chess, Jane Dodd, Tom Jessell, Eric Kandel, and John Ngai B6: (G/C) (A/T) I (G/C) (Afl') ICCIAC (NG) AA (NG) AA (NG) TAIAT (N for helpful discussions and for critically reading the manuscript, and G)AAIGG(G/A)I-I. to George Gaitanaris for helpful discussions during the course of this work. We also wish to thank Phyllis Kisloff and Miriam Gutierrez for An aliquot of each PCR reaction was analyzed by agarose gel elec- expert assistance in the preparation of this manuscript. trophoresis, and bands of interest were amplified further by performing This work was supported by the Howard Hughes Medical Institute PCR reactions on pipet tip (,~,1 pl) plugs of the agarose gels containing and by grants from the Frederick R. Adler Education Fund and the those DNAs. Aliquots of these semipurified PCR products were di- National Institutes of Health (PO1-CA 23767) (to R. A.). gested with the restriction enzyme Haelll or Hinfl, and the digestion The costs of publication of this article were defrayed in part by products were compared with the undigested DNAs on agarose gels. the payment of page charges. This article must therefore be hereby marked =advertisement" in accordance with 18 USC Section 1734 Isolation and Analysis of cDNA Clones solely to indicate this fact. cDNA libraries were prepared according to standard procedures (Ma- niatis et al., 1982; Sambrook st al., 1989) in the cloning vector kZAP Received March 14, 1991; revised March 19, 1991. II (Stratagene) using poly(A)÷ RNA prepared from Sprague-Dawley rat epithelia (see above) or from an enriched population of olfactory References neurons that had been obtained by a "panning" procedure (L. B. and R. A., unpublished data) using an antibody against the H blood group Amoore, J. E. (1982). Odor theory and odor classification. In Fragrance antigen (Chembiomed) found on a large percentage of rat olfactory Chemistry (New York: Academic Press, Inc.), pp. 27-73. neurons. In initial library screens, 8.5 x 10S independent clones from the olfactory neuron library and 1.8 x 10e clones from the olfactory Baltimore, D. (1981). Gene conversion: some implications for immuno- globulin genes. Cell 24, 592-594. epithelium library were screened (Maniatis et al., 1982) with a ~P- labeled probe (Prime-it, Stratagene) consisting of a pool of gel-isolated Boekhoff, I., Tareilus, E., Strotmann, J., and Brser, H. (1990). Rapid PCR products obtained using primers A4 and B6 (see above) in PCR activation of alternative second messenger pathways in olfactory cilia reactions, using as template olfactory epithelium cDNA, rat liver DNA, from rats by different odorants. EMBO J. 9, 2453-2458. or DNA prepared from the two cDNA libraries. In later library screens, Bonnet, T. I., Buckley, N. J., Young, A. C., and Brann, M. R. (1987). a mixture of PCR products obtained from 20 cDNA clones with the A4 Identification of a family of muscarinic acetylcholine receptor genes. and B6 primers was used as probe ("PI" probe). In initial screens, Science 237, 527-532. phage clones were analyzed by PCR using primers A4 and B6, and Bouvier, M. W., Hausdorff, A., De Blasi, A., O'Dowd, B. F., Kobilka, those that showed the appropriate size species were purified. In later B. K., Caron, M. G., and Lefkowitz, R. J. (1988). Removal of phosphory- screens all positive clones were purified, but only those that could be lation sites from the 13-adrenergic receptor delays the onset of agonist- amplified with the B6 primer and a primer specific for vector sequence promoted desensitization. Nature 333, 370-373. were analyzed further. To obtain plasmids from the isolated phage Breer, H., Boekhoff, I., and Tareilus, E. (1990). Rapid kinetics of second clones, phagemid rescue was performed according to the instructions messenger formation in olfactory transduction. Nature 345, 65-68. of the manufacturer of X7_.AP II (Stratagene). DNA sequence analysis was performed on plasmid DNAs using the Sequenase system (United Bronshtein, A. A., and Minor, A. V. (1977). Regeneration of olfactory States Biochemical Corp.), initially with the A4 and B6 primers and later flagella and restoration of the electroolfactogram following application with oligonucleotide primers made according to sequences already of Triton X-100 to the olfactory mucosa of frogs. Tsitologiia 19, 33-39. obtained. Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J., and Rutter, W. J. (1979). Isolation of biologically active ribonucleic acid from sources Northern and Southern Blot Analyses enriched in ribonuclease. Biochemistry 18, 5294-5299. For Northern blots, poly(A)÷ RNAs from various tissues were prepared Dhallan, R. S., Yau, K.-W., Schrader, K. A., and Reed, R. R. (1990). as described above or purchased from Clontech. One microgram of Primary structure and functional expression of a cyclic nucleotide- each RNA was size fractionated on formaldehyde-agarose gels and activated channel from olfactory neurons. Nature 347, 184-187. blotted onto nylon membranes (Maniatis et al., 1982; Sambrook et al., Doving, K. B., Selset, R., and Thommesen, G. (1980). Olfactory sensi- 1989). For Southern blots, genomic DNA prepared from Sprague- tivity to bile acids in salmonid fishes. Acta Physiol. Scand. 108, 123- Dawley rat liver was digested with the restriction enzyme EcoRI or 131. Hindlll, size fractionated on agarose gels, and blotted onto nylon mem- Egel, R. (1981). Intergenic conversion and reiterated genes. Nature branes (Maniatis et al., 1982; Sambrook et al., 1989). The membranes 290, 191-192. were dried at 8O°C and then prehybridized in 0.5 M sodium phosphate buffer (pH 7.3) containing 1% bovine serum albumin and 4% SDS. Flanagan, J. G., Lefranc, M.-P., and Rabbitts, T. H. (1984). Mecha- Hybridization was carried out in the same buffer at 65°C-70°C for 14- nisms of divergence and convergence of the human immunoglobulin 20 hr with DNAs labeled with 32p. For the first Northern blot shown, the (~1 and a2 constant region gene sequences. Cell 36, 681-688. "PI" probe (see above under cDNA clone isolation) was used. For the Graziadei, P. P. C., and Monti Graziadei, G. A. (1979). Neurogenesis second Northern blot shown, a mix of PCR fragments from seven and neuron regeneration in the olfactory system of"mammals. I. Mor- divergent cDNA clones was used (see Southern blot below). For South- phological aspects of differentiation and structural organization of the ern blots, the region indicated in clone 115 by amino acids 118 through olfactory sensory neurons. J. Neurocytol. 8, 1-18. 251 was amplified from a series of divergent cDNA clones using PCR. Harem, H. E., Deretic, D., Arendt, A., Hargrove, P. A., Koenig, B., and The primers used for these reactions had the following sequences: Hofmann, K. P. (1988). Site of G protein binding to rhodopsin mapping with synthetic peptides to the (z subunit. Science 241,832-835. P 1 : ATGGCITA(r/C)GA(T/C)(C/A)GITA(T/C)GTIGC; Hood, L., Kronenberg, M., and Hunkapiller, T. (1985). T cell antigen P4: AAIA(GIA)I(GIC)(NT)IACIA(FIC)I(GICXAfT)IA(GIA)(NG)TGI(GIC) - receptors and the immunoglobulin supergene family. Cell 40, 225- (NT)~(C/G)C. 229. These DNAs (or a DNA encompassing transmembrane domains 2 Jones, D. T., and Reed, R. R. (1989). G=f: an olfactory neuron-specific through 7 for clone F6) were labeled and tested for cross-hybridization G-protein involved in odorant signal transduction. Science 244, 790- at 70°C. Those DNAs that did not show appreciable cross- 795. hybridization were hybridized individually, or as a pool, to Southern Kobilka, B. K., Kobilka, T. S., Daniel, K., Regan, J. W., Caron, blots at 70°C. M. G., and Lefkowitz, R. J. (1988). Chimeric (l=-132-adrenergic recep- tors: delineation of domains involved in effector coupling and ligand Acknowledgments binding specificity. Science 240, 1310-1316. Lancet, D. (1986). Vertebrate olfactory reception. Annu. Rev. Neurosci. We would like to thank Ore Pearlstein, Bruce Brender, Tom Livelli, and 9, 329-355. Candidate Odorant Receptors 187 Mackay-Sim, A., Shaman, P., and Moulton, D. G. (1982). Topographic somes: genetic recombination and transcriptional control of VSG coding of olfactory quality: odorant-specific patterns of epithelial re- genes. In Gene Rearrangement (New York: Oxford University Press)i sponsivity in the salamander. J. Neurophysiol. 48, 584-596. pp. 51-98. Maeda, N., and Smithies, O. (1986). The evolution of multigene fami- Wald, G., Brown, P. K., and Smith, P. H. (1955). Iodopsin. J. Gen. lies: human haptoglobin genes. Annu. Rev. Genet. 20, 81-108. Physiol. 38, 623-681. Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982). Molecular Clon- ing: A Laboratory Manual (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press). Moulton, D. G., and Beidler, L. M. (1967). Structure and function in the peripheral olfactory system. Physiol. Rev. 47, 1-52. Nakamura, T., and Gold, G. (1987). A cyclic nucleotide-gated conduc- tance in olfactory receptor cilia. Nature 325, 442-444. Nathans, J., Thomas, D., and Hogness, D. S. (1986). Molecular genet- ics of human color vision: the genes encoding blue, green, and red pigments. Science 232, 193-202. O'Dowd, B. F., Hnatowich, M., Caron, M. G., Lefkowitz, R. J., and Bouvier, M. (1989a). Palmitoylation of the human ~2-adrenergic recep- tor. Mutation of CYS TM in the carboxyl tail leads to an uncoupled non- palymitoylated form of the receptor. J. Biol. Chem. 264, 7564-7569. O'Dowd, B. F., Lefkowitz, R. J., and Caron, M. G. (1989b). Structure of the adrenergic and related receptors. Annu. Rev. Neurosci. 12, 67- 83. Pace, U., Hanski, E., Salomon, Y., and Lancet, D. (1985). Odorant- sensitive adenylate cyclase may mediate olfactory reception. Nature 316, 255-268. Reed, R. R. (1990). How does the nose know? Cell 60, 1-2. Rhein, L. D., and Cagan, R. H. (1980). Biochemical studies of olfaction: isolation, characterization, and odorant binding activity of cilia from rainbow trout olfactory rosettes. Proc. Natl. Acad. Sci. USA 77, 4412- 4416. Rhein, L. D., and Cagan, R. H. (1983). Biochemical studies of olfaction: binding specificity of odorants to a cilia preparation from rainbow trout olfactory rosettes. J. Neurochem. 41,569-577. Rushton, W. A. H. (1955). Foveal photopigments in normal and colour- blind. J. Physiol. 129, 41-42. Rushton, W. A. H. (1965). A foveal pigment in the deuteranope. J. Physiol. 176, 24-37. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989). Molecular Clon- ing: A Laboratory Manual, Second Edition (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press). Shepherd, G. M. (1985). Are there labeled lines in the olfactory path- way? In Taste, Olfaction and the Central Nervous System (New York: The Rockefeller University Press), pp. 307-321. Sicard, G. (1985). Olfactory discrimination of structurally related mole- cules: receptor cell responses to camphoraceous odorants. Brain Res. 326, 203-212. Sklar, P. B., Anholt, R, R. H., and Snyder, S. H. (1986). The odorant- sensitive adenylate cyclase of olfactory receptor cells: differential stim- ulation by distinct classes of odorants. J. Biol. Chem. 251, 15538- 15543. Stewart, W, B., Kauer, J. S., and Shepherd, G. M. (1979). Functional organization of the rat olfactory bulb analysed by the 2-deoxyglucose method. J. Comp. Neurol. 185, 715-734. Strader, C. D., Sigal, I. S., Register, R. B., Candelore, M. R., Rands, E., and Dixon, R. A. F. (1987). Identification of residues required for ligand binding to the 13-adrenergic receptor. Proc. Natl. Acad. Sci. USA 84, 4384-4388. Strader, C. D., Sigal, I. S., and Dixon, R. A. F. (1989). Structural basis of ~-adrenergic receptor function. FASEB J, 3, 1825-1832. Thommesen, G. (1978). The spatial distribution of odour induced po- tentials in the olfactory bulb of char and trout (Salmonidae). Acta Phys- iol. Scand. 102, 205-217. Thommesen, G., and Doving, K. B. (1977). Spatial distribution of the EOG in the rat. A variation with odour stimulation. Acta Physiol. Scand. 99, 270-280. Tonegawa, S. (1983). Somatic generation of antibody diversity. Nature 302, 575-581. Van der Ploeg, L. H. T. (1991). Antigenic variation in African trypano-