letter

58 nature genetics • volume 28 • may 2001

Fig. 1 Candidate genes in the region of the Sac locus.There is synteny between human 1p36.33 and mouse4pter chromosomal regions near the mouse Sac locus.Shaded circles indicate the approximate location of thepredicted start codons for each gene; arrows indicate thefull span of each gene including both introns and exons;arrowheads indicate the approximate location of eachpolyadenylation signal. Genes indicated by their fullnames were predicted by Genscan and named accordingto their closest homolog. TAS1R3 and DVL1 were experi-mentally identified and verified. The mouse markerD18346 is closely linked to the Sac locus and lies withinthe predicted pseudouridine synthase-like gene. Theregion displayed corresponds to ∼ 45,000 bp; the bottomscale marker indicates kilobases (K).

Tas1r3, encoding a new candidate taste receptor, is allelicto the sweet responsiveness locus Sac

Marianna Max2*, Y. Gopi Shanker1,2*, Liquan Huang1,2, Minqing Rong1,2, Zhan Liu1,2, Fabien Campagne2,3,Harel Weinstein2,3, Sami Damak2 & Robert F. Margolskee1,2

*These authors contributed equally to this work.

1Howard Hughes Medical Institute, 2Department of Physiology and Biophysics and 3Institute for Computational Biomedicine, Mount Sinai School ofMedicine of New York University, New York, New York, USA. Correspondence should be addressed to R.F.M. (e-mail: Bob@inka.mssm.edu).

The ability to taste the sweetness of carbohydrate-rich food-stuffs has a critical role in the nutritional status of humans.Although several components of bitter transduction pathwayshave been identified1–6, the receptors and other sweet trans-duction elements remain unknown. The Sac locus in mouse,mapped to the distal end of chromosome 4 (refs. 7–9), is themajor determinant of differences between sweet-sensitive and-insensitive strains of mice in their responsiveness to saccharin,sucrose and other sweeteners10–13. To identify the human Saclocus, we searched for candidate genes within a region ofapproximately one million base pairs of the sequenced humangenome syntenous to the region of Sac in mouse. From thissearch, we identified a likely candidate: T1R3, a previouslyunknown G protein-coupled receptor (GPCR) and the only GPCRin this region. Mouse Tas1r3 (encoding T1r3) maps to within20,000 bp of the marker closest to Sac (ref. 9) and, like humanTAS1R3, is expressed selectively in taste receptor cells. By com-paring the sequence of Tas1r3 from several independentlyderived strains of mice, we identified a specific polymorphismthat assorts between taster and non-taster strains. Accordingto models of its structure, T1r3 from non-tasters is predicted tohave an extra amino-terminal glycosylation site that, if used,would interfere with dimerization.

To identify Sac, we generated a contiguous map of the region ofthe human genome syntenous to the region of Sac in mouse. Wedetermined that D18346, the marker closest to Sac (ref. 9), mapswithin a novel mouse gene with homology to the pseudouridinesynthase gene. We used the TblastN program to identify thehomolog of this gene in finished and unfinished human genomicsequences from BAC and PAC clones known to map to humanchromosome 1pter–1p36.33 (syntenic to mouse 4pter). By

repeated Blast searches of the human genomic sequence (HTGS)database (NCBI) with portions of the sequence from this andoverlapping BAC and PAC clones, we were able to form a con-tiguous map (‘contig’) of six overlapping BAC or PAC clonesspanning approximately one million base pairs of humangenomic DNA sequence.

Using the Genscan gene prediction program, we identified 23predicted genes within this contig (Fig. 1), including pseudouridinesynthase-like, cleavage and polyadenylation-like, and glycolipidtransfer-like; a few genes within this region had been previouslyidentified and/or experimentally verified by others (for example,disheveled-1, DVL1). We then searched the Celera Genomicsmouse database to identify the mouse homologs of the genes withinthis region and pieced together the mouse contig (Fig. 1).

In our screen of the one million base pairs of genomic DNAsequence in the Sac region, only one predicted GPCR gene wasfound. The gene, which we named TAS1R3 (for taste receptorfamily 1, member 3), was of special interest because the predictedprotein it encodes (T1R3) is most similar to T1R1 and T1R2, twoorphan GPCRs expressed in taste cells14, and because it isexpressed specifically in taste cells. Human TAS1R3 is locatedabout 20 kb from the human homolog of the pseudouridine syn-thase-like gene (Fig. 1).

The intron and exon structure of the coding portion ofTAS1R3 was predicted by Genscan to span 4 kb and contain 7exons (Web Fig. A). Based on the nucleotide sequence of thegenomic DNA and cDNAs derived from a human taste cDNAlibrary, TAS1R3 is predicted to encode a protein of 843 aminoacids with 7 transmembrane helices and an extracellular domainof 558 amino acids. The corresponding mouse Tas1r3 genomicsequence was assembled from the Celera Genomics mouse

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m© 2001 Nature Publishing Group http://genetics.nature.com

letter

nature genetics • volume 28 • may 2001 59

genomic fragment database. The coding portion of Tas1r3 fromC57BL/6J spans 4 kb and contains 6 exons; the predicted proteinis 858 amino acids with an extracellular domain of 573 aminoacids. Tas1r3 is located within 20 kb of D18346 and the mousepseudouridine synthase-like gene. The close proximity of Tas1r3to D18346 is consistent with it being Sac, given the low probabil-ity of crossovers between the marker and the Sac locus in F2crosses and congenic mice9.

Human and mouse T1R3 are members of the family 3 subtypeof GPCRs, all of which contain large extracellular domains.Other family 3 subtype GPCRs include metabotropic glutamatereceptors (mGluR), extracellular calcium sensing receptors(ECaSR), candidate pheromone receptors expressed in thevomeronasal organ (V2R), and two taste receptors (T1R1 andT1R2) of unknown ligand specificity. T1R3 is most closely relatedto T1R1 and T1R2, sharing approximately 30% amino acidsequence identity with each (T1R1 and T1R2 are ∼ 40% identicalto each other). At the amino acid level, T1R3 is approximately20% identical to mGluRs and approximately 23% identical to

ECaSRs. The large N-terminal domain of family 3 GPCRs hasbeen implicated in ligand binding and dimerization15.

Northern-blot analysis identified a 7.2-kb Tas1r3 mRNA intaste tissue that was not expressed in control lingual tissuedevoid of taste buds (non-taste) or in any of the several othertissues examined (Fig. 2a). A somewhat larger (∼ 7.8 kb)mRNA species was expressed at moderate levels in testis and atvery low levels in brain. A smaller (∼ 6.7 kb) mRNA species wasexpressed at very low levels in thymus. As another measure ofTas1r3 mRNA expression, we examined the expressed sequencetag (EST) database for strong matches to Tas1r3 and the otherpredicted genes in the Sac region. Although Dvl1, glycolipidtransfer-like, cleavage and polyadenylation-like, andpseudouridine synthase-like genes each had numerous highlysignificant matches to ESTs from several different tissues,Tas1r3 showed only a single strong match to an EST from colon(Web Fig. B). The EST and northern-blot results demonstratethat the expression of Tas1r3 is highly restricted, consistentwith it being a taste receptor.

Fig. 2 Expression of Tas1r3 mRNA in mouse tissues and mouse taste cells. a, Autora-diogram of a northern blot hybridized with Tas1r3 cDNA. Each lane contained totalRNA isolated from the indicated mouse tissues (Taste, circumvallate and foliatepapillae-enriched lingual tissue; Non-Taste, lingual tissue devoid of taste buds; OlfEpi, olfactory epithelium; Small Int, small intestine; Ske Mus, skeletal muscle). A 7.2-kb transcript was detected only in the taste tissue, and a slightly larger transcriptwas detected in testis. The same blot was stripped and reprobed with a β-actincDNA (bottom) and re-exposed. The size of the RNA marker (in kilobases) is indi-cated in the right margin. b–i, Photomicrographs of frozen sections of mouse tastepapillae hybridized with 33P-labeled antisense RNA probes for Tas1r3 and Gnat3.Bright-field images of circumvallate (b), foliate (c) and fungiform (d) papillaehybridized to the antisense Tas1r3 probe demonstrate taste bud-specific expressionof Tas1r3. Control bright-field images of circumvallate (f), foliate (g) and fungiformpapillae (h) hybridized to the sense Tas1r3 probe showed no nonspecific binding.The level of expression and broad distribution of Tas1r3 expression in taste buds wascomparable to that of Gnat3 (e). Foliate papilla hybridized to the sense Gnat3probe (i) showed no nonspecific binding. j, Profiling the pattern of expression ofTas1r1, Tas1r2, Tas1r3, Gnat3, Gng13 and Plcb2 in taste tissue and taste cells. Left,Southern-blot hybridization to RT–PCR products from mouse taste tissue (T) andcontrol non-taste lingual tissue (N). Note that Tas1r1 (T1r1), Tas1r2 (T1r2), Tas1r3(T1r3), Gnat3 (α-gustducin), Gng13 (Gγ13) and Plcb2 (PLCβ2) were all expressed intaste tissue, but not in non-taste tissue. Right, Southern-blot hybridization toRT–PCR products from 24 individually amplified taste receptor cells from a trans-genic mouse expressing GFP from the gustducin promoter3: 19 cells were GFP-posi-tive (+) and 5 cells were GFP-negative (–). Expression of α-gustducin, Gγ13 and PLCβ2was fully coincident. Expression of Tas1r3 overlapped partially with that of Tas1r2,Gnt3, Gng13 and Plcb2. G3PDH served as a positive control to demonstrate success-ful amplification of products.

a b c d e

f g h i

j

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m© 2001 Nature Publishing Group http://genetics.nature.com

letter

60 nature genetics • volume 28 • may 2001

In situ hybridization demonstrated that Tas1r3 was selectivelyexpressed in taste receptor cells, but absent from the surroundinglingual epithelium, muscle or connective tissue (Fig. 2b–d). Senseprobe controls showed no non-specific hybridization to lingualtissue (Fig. 2f–h). The RNA hybridization signal for Tas1r3 waseven stronger than that for Gnat3, encoding α-gustducin (Fig.2e), indicating that Tas1r3 mRNA is very highly expressed in tastereceptor cells. This is in contrast to results with Tas1r1 and Tas1r2mRNAs, which are expressed at lower levels than is Gnat3 mRNA(ref. 14). Moreover, Tas1r3 is highly expressed in taste buds fromfungiform, foliate and circumvallate papillae, whereas Tas1r1 andTas1r2 mRNAs each show different, regionally restricted, pat-terns of expression14.

Expression profiling3 demonstrated that Tas1r3, like Tas1r1,Tas1r2, Gnat3, Gng13 (encoding the G protein γ13 subunit) andPlcb2 (encoding phospholipase Cβ2), was expressed in taste bud-containing tissue, but not in non-gustatory lingual epithelia (Fig.2j, left). Profiling the pattern of expression of these genes in indi-vidual taste cells demonstrated, as we had previously shown3,that all of the 19 cells expressing Gnat3 also expressed Gnb3(encoding the G protein β3 subunit) and Gng13. Moreover, all 19of these cells also expressed Plcb2 (Fig. 2j, right). Of these 19 cells,12 also expressed Tas1r3, 10 of these cells expressed Tas1r2, and 6cells were positive for both Tas1r2 and Tas1r3. Thus we concludethat expression of Tas1r3 and Gnat3, Gnb3, Gng13 and Plcb2,although not fully coincident, overlaps to a great extent. Theoverlapping expression of Tas1r2, Tas1r3 and Gnat3 contrastswith previous in situ hybridization results with taste receptor cellsof the vallate papillae, in which only approximately 15% ofGnat3-positive cells were positive for Tas1r2 (ref. 14); this differ-ence may be due largely to differences in sensitivity of the twotechniques. Co-expression of T1r3 with α-gustducin, Gβ3, Gγ13and PLCβ2 is significant because these proteins have been impli-cated in sweet and/or bitter taste transduction2,3,16,17.

Immunocytochemistry with an anti-T1R3 antibody demon-strated that about one-fifth of taste receptor cells in human circum-vallate (Fig. 3a,b) and fungiform (Fig. 3e,h) papillae were positivefor T1R3. We blocked T1R3 immunoreactivity by pre-incubationof the T1R3 antibody with the cognate peptide (Fig. 3c). Longitudi-nal sections of the T1R3-positive taste cells displayed an elongatedbipolar morphology typical of so-called ‘light’ cells (many of whichare α-gustducin–positive), with the immunoreactivity most promi-nent at or near the taste pore (Fig. 3a,b,e,h). Labeling adjacent sec-tions with antibodies directed against T1R3 and PLCβ2 showedmore cells positive for PLCβ2 than for T1R3 (Fig. 3b,d). Doublelabeling for T1R3 and PLCβ2 (Fig. 3e–g), or for T1R3 and α-gust-ducin (Fig. 3h–j) showed many, but not all, cells to be positive forboth (more cells were positive for PLCβ2 or α-gustducin than forT1R3), consistent with the results from expression profiling.TAS1R3 mRNA and protein are selectively expressed in a subset ofα-gustducin/PLCβ2-positive taste receptor cells, as would beexpected for a taste receptor.

We examined the nucleotide and inferred amino acid sequenceof T1r3 from taster and non-taster strains of mice for allelic dif-ferences that might explain their phenotypic differences (Fig. 4a).All four non-taster strains (DBA/2, 129/Svev, BALB/c andC3H/HeJ) examined shared the same allele with identicalnucleotide sequence, despite the fact that their most recent com-mon ancestors date back to the early 1900s or earlier (Fig. 4b).The four taster strains fell into two allelic classes (C57BL/6J ver-sus SWR, FVB/N and ST/bJ). We found 15 positions to be poly-morphic at the nucleotide level; 7 of these positions predictcoding differences (the other nucleotide differences were inintrons or were ‘silent’ codon changes). There were only two cod-ing differences that distinguished all tasters from all non-tasters:T55A and I60T (single letter amino acid code, with the non-taster amino acid listed last). The I60T change is intriguing, as itis predicted to introduce a novel N-linked glycosylation site inthe N-terminal domain of T1r3.

To gain insight into the structural determinants of T1r3 thatmight influence its function, we aligned the N-terminaldomain of T1r3 with those of four other members of the type 3subset of GPCRs (Fig. 5) and modeled it based on the recentlysolved structure of the N-terminal domain of the relatedmGluR1 receptor15 (Fig. 6). According to this model, T1r3 ispredicted to dimerize (Fig. 5). The I60T taster to non-taster

Fig. 3 Co-localization of T1R3, PLCβ2 and α-gustducin in human taste receptorcells. a,b, Longitudinal sections from human circumvallate papillae labeledwith rabbit antisera directed against a carboxy-terminal peptide of T1R3, alongwith a Cy3-conjugated anti-rabbit secondary antibody. c, T1R3 immunoreactiv-ity in longitudinal sections from human circumvallate papillae was blocked bypre-incubation of the anti-T1R3 antibody with the cognate peptide. d, A longi-tudinal section adjacent to that in (b) was labeled with an antibody directedagainst PLCβ2, along with a fluorescein-conjugated anti-rabbit secondary anti-body. The dotted lines mark the borders of three taste buds present in the twoadjacent sections. Longitudinal sections of human fungiform papillae doubleimmunostained for T1R3 (e) and PLCβ2 (f). The overlay of the two images isshown in (g). Longitudinal sections of human fungiform papillae doubleimmunostained for T1R3 (h) and α-gustducin (i). The overlay of the two imagesis shown in (j). Magnification: a–d, ×200; e–j, ×400.

a b

c d

e f g

h i j

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m© 2001 Nature Publishing Group http://genetics.nature.com

letter

nature genetics • volume 28 • may 2001 61

Fig. 5 The amino acid sequenceof T1r3 is aligned with that oftwo rat taste receptors (rT1R1and rT1R2), the mouse extracel-lular calcium sensing (mECaSR)and the metabotropic gluta-mate type 1 (mGluR1) receptors.Regions of identity among allfive receptors are indicated bywhite letters on black; regionswhere one or more of thesereceptors share identity withT1r3 are indicated by black let-ters on gray. Boxes with dashedlines indicate regions predictedto be involved in dimerization(based on the solved structurefor the N-terminal domain ofmGluR1); filled circles indicatepredicted ligand-bindingresidues based on mGluR1; bluelines linking cysteine residuesindicate predicted intramolecu-lar disulfide bridges based onmGluR1. Amino acid sequencesnoted above the alignmentindicate polymorphisms thatare found in all strains of non-taster mice. The predicted N-linked glycosylation siteconserved in all five receptors isindicated by a black squiggle;the predicted N-linked glycosy-lation site specific to T1r3 innon-taster strains of mice isindicated by the red squiggle.

substitution is predicted to introduce a novel N-linked glycosy-lation site at N58, 27 amino acids upstream from a conservedN-linked glycosylation site present in all 5 aligned receptors.The model predicts that N-linked glycosylation at N58 wouldhave a profound effect on the ability of T1r3 to dimerize (Fig.6c): the addition of even a short carbohydrate group at N58 ispredicted to disrupt at least one of the contact surfacesrequired for stability of the dimer. Therefore, if T1r3, likemGluR1, adopts a dimeric form, then the predicted N-linkedglycosyl group at N58 would be expected to preclude T1r3from forming homodimers or heterodimers.

Our identification of TAS1R3 as Sac is based on the following.TAS1R3 is the only gene encoding a GPCR present in a region ofhuman genomic DNA centered on the human ortholog of themouse gene containing the D18346 marker most tightly linked toSac. Expression of TAS1R3 is narrowly restricted to a subset of tastereceptor cells, and overlaps with that of known and proposed ele-ments of sweet transduction pathways. T1R3 has a large extracellu-lar domain that would be expected to be sensitive to proteases (aknown property of the sweet receptor18,19). Also, a polymorphismin T1r3 (I60T) was identified that differentiated all taster strains ofmice from all non-taster strains. Hence, not only is Tas1r3 identified

Fig. 4 Tas1r3 allelic differences. a, Tas1r3sequence differences between eightinbred mouse strains. All non-tasterstrains showed identical sequences andwere grouped in one row. In the bottomrow, the amino acid immediately afterthe position number is always from thenon-tasters, whereas the amino acidimmediately before the position num-ber is from whichever tasters differed atthat position from the non-tasters. Thetwo columns in bold represent positionswhere all tasters differed from all non-tasters and where the differences innucleotide sequence result in aminoacid substitutions. Nucleotide differ-ences that do not alter the encodedamino acid are indicated (s, silent).Nucleotide differences within intronsare indicated (i, intron). b, Genealogy ofthe inbred strains of mice analyzed in(a). The year in which the strains weredeveloped is indicated between brack-ets following the strain name. The labo-ratories in which these mice wereestablished are indicated.

a

b

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m© 2001 Nature Publishing Group http://genetics.nature.com

letter

62 nature genetics • volume 28 • may 2001

as Sac, but based on the model of T1r3 and this polymorphicchange it is also likely to be a sweet-responsive (that is, sweet-lig-anded) taste receptor. Moreover, T1r3-related orphan taste recep-tors (T1r1 and T1r2) are also likely to be responsive to sweetcompounds or amino acids preferred by mice (for example, gluta-mate, alanine or glycine). Proof that T1r3 is Sac will require thedemonstration that transgenic expression in non-taster mice ofT1r3 from a taster strain manifests the taster phenotype. Proof ofour model of the N-terminal domain of T1R3 will require crystal-lization and determination of the structure of the expressed N-ter-minal domain from taster and non-taster. Such information wouldbe particularly useful for the identification of novel sweeteners ofintense potency. It seems fitting that the presence or absence of asugar chain on a sweet taste receptor should determine sensitivityand preference for sweetness in life.

Note: supplementary information is available on the NatureGenetics web site (http://genetics.nature.com/supplementary_info/).

MethodsGene searching. Iterative BlastN screens of the mouse EST database identi-fied the D18346 marker-containing gene. A TBlastN search of the high-throughput genomic sequence (HTGS) database identified the humanhomolog-containing BAC clone (AL139287). Exons and genes in this clonewere predicted by Genscan (http://genes.mit.edu/GENSCAN.html; ref.20). BlastN/TBlastN searches of the NR and EST databases identified genesin this and related clones. BlastN/TBlastN searches of the HTGS, using theidentified genes, identified overlapping clones. These sequences were usedin BlastN searches of the HTGS to build a contig: the predicted genes andexons were used to order over one million bases of genomic sequence cen-

Fig. 6 The predicted three-dimen-sional structure of the N-terminaldomain of T1r3. The model shows ahomodimer of T1r3. a, The view fromthe ‘top’ of the dimer looking downfrom the extracellular space towardthe membrane. b, The T1r3 dimerviewed from the side. In this view, thetransmembrane region (not shown)would attach to the bottom of thedimer. c, The T1r3 dimer is viewedfrom the side as in (b), except the twodimers have been spread apart (indi-cated by the double-headed arrow)to reveal the contact surface. A space-filling representation (colored red) ofthree glycosyl moieties (N-acetyl-galactose-N-acetyl-galactose-Man-nose) has been added at the novelpredicted site of glycosylation of non-taster T1r3. Note that the addition ofeven three sugar moieties at this siteis sterically incompatible with dimer-ization. Regions of T1r3 correspond-ing to those of mGluR1 involved indimerization are shown by space-fill-ing amino acids. The four differentsegments that form the predicteddimerization surface are color-codedin the same way as are the dashedboxes in Fig. 5. The portions of thetwo molecules outside of the dimer-ization region are represented by abackbone tracing. The two polymor-phic amino acid residues of T1r3 thatdiffer in taster versus non-tasterstrains of mice are within the pre-dicted dimerization interface nearestthe N terminus (colored light blue).The additional N-glycosylation site ataa 58 unique to the non-taster formof T1r3 is indicated in each panel bythe straight arrows.

a b

c

tered on the D18346 marker-containing gene. TAS1R3 was identified inclones AL139287 and AC026283. A gene discovery ‘pipeline’ to identifygenes encoding seven transmembrane-helix proteins with restricted tissuedistribution (F.C. and H.W., unpublished data) was used to search allhuman genomic clones from 1pter to 1p33: TAS1R1, TAS1R2 and TAS1R3were so identified. A TBlastN screen of the Celera mouse genomic databaseusing TAS1R3 identified the mouse homolog, Tas1r3. Multiple mouse frag-ments were used to build the Tas1r3 genomic sequence (CSN001,http://www.celera.com/publicationlibrary), then Genscan was used to pre-dict the exons. Tas1r3 was amplified by PCR from genomic DNA fromtaster and non-taster strains of mice and sequenced.

Northern-blot hybridization. Isolation of mouse RNAs, electrophoresis,transfer, fixation, hybridization, washing and exposure were asdescribed3. The 32P-labeled Tas1r3 probe was generated by random non-amer priming of a 1.34-kb cDNA fragment of mouse T1r3 correspond-ing to the 5´ end coding sequence using Exo(-) Klenow polymerase inthe presence of (α-32P)-dCTP.

In situ hybridization. We used 33P-labeled RNA probes (Tas1r3 (2.6 kb)and Gnat3 (1 kb)) for in situ hybridization of frozen sections (10 µm) ofmouse lingual tissue. Hybridization and washing were as described2. Slideswere coated with Kodak NTB-2 nuclear track emulsion, exposed at 4 °C for3 weeks, and then developed and fixed.

Gene expression profiling. RT–PCR amplification of individual tastereceptor cells, single circumvallate papilla and non-gustatory tissue, as wellas gel electrophoresis of the RT–PCR products, transfer to a nylon mem-brane, hybridization and exposure, were as described3. Southern-blothybridization was carried out with cDNA probes from the 3´ ends of mouseTas1r1, Tas1r2, Tas1r3, Gnat3, Gng13, Plcb2 and G3PDH.

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m© 2001 Nature Publishing Group http://genetics.nature.com

letter

nature genetics • volume 28 • may 2001 63

Immunocytochemistry. Polyclonal antisera against a hemocyanin-conju-gated T1R3 peptide (aa 829–843) were raised in rabbits. The PLCβ2 anti-body was obtained from Santa Cruz Biotechnology. The anti-α-gustducinantibody was as described21. Frozen sections (10 µm) of human lingual tis-sue (previously fixed in 4% paraformaldehyde and cryoprotected in 20%sucrose) were blocked in 3% BSA, 0.3% Triton X-100, 2% goat serum and0.1% sodium azide in PBS for 1 h at RT and then incubated for 8 h at 4 °Cwith antibody against PLCβ2, α-gustducin or antiserum against T1R3(1:800). The secondary antibodies were Cy3-conjugated goat-anti-rabbit Igfor T1R3 and fluorescein-conjugated goat-anti-rabbit Ig for PLCβ2 or α-gustducin. PLCβ2 and T1R3 immunoreactivities were blocked by preincu-bation of the antisera with the corresponding synthetic peptides at 10 µMand 20 µM, respectively. Pre-immune serum did not show any immunore-activity. Some sections were double-immunostained with antisera againstT1R3 and PLCβ2 or against T1R3 and α-gustducin as described (M.M.Bakre et al., manuscript submitted). Briefly, sections were incubatedsequentially with T1R3 antiserum, anti-rabbit-Ig-Cy3 conjugate, normalanti-rabbit-Ig, anti-PLCβ2 or anti-α-gustducin antibody and finally withanti-rabbit-Ig-FITC conjugate with intermittent washes between each step.Control sections incubated with all of the above except the anti-PLCβ2 oranti-α-gustducin antibody did not show any fluorescence in the greenchannel.

Identification of sequence polymorphisms in Tas1r3. Based on thesequence of Tas1r3, oligonucleotide primers were designed to amplify DNAfrom regions containing ORFs. Total RNA isolated from taste papillae ortail genomic DNA isolated from one taster (C57BL/6J) and one non-taster(129/Svev) mouse strain each were used as templates to amplify T1r3cDNA and genomic DNA using RT–PCR and PCR, respectively. PCRproducts were sequenced completely in an ABI 310 automated sequencer.Based on the sequence obtained, four sets of oligonucleotide primers wereused to amplify the Tas1r3 regions where polymorphisms were foundbetween the two strains of mice. Genomic DNAs from mouse strainsDBA/2, BALB/c, C3H/HeJ, SWR, ST/bJ and FVB/N were used as templates.The amplicons were purified and directly sequenced. The genealogical treeof these strains of mice was based on Hogan et al.22 and The Jackson Labo-ratory web site (http://www.jax.org).

Modeling the structure of T1r3. The N-terminal domains of T1r3 andmGluR1 were aligned using the ClustalW program23. The alignment wasmanually edited to generate an optimal alignment based on structural andfunctional considerations. This alignment was found to be in generalagreement with an alignment generated automatically by 3D-PSSM (ref.24). Atomic coordinates of the mGluR1 N-terminal domain crystal struc-ture15 were obtained from the protein database and were used along withthe alignment as the source of spatial restraints for modeling. The structur-al model of T1r3 was generated using the program MODELLER (ref. 25).The original images were created using the programs Insight II and WeblabViewer (Molecular Simulations) and then imported into Photoshop wherethe open view was created and labels added.

Accession numbers. TAS1R3, GenBank AL139287 and AC026283; Tas1r3,Celera CSN001, GenBank AF368024 and AF368025.

AcknowledgmentsWe thank S. Zou for DNA sequencing; D. Perl and the staff of the MSSMPathology Department for human taste tissue; and I. Visiers for help withmodeling the structure of T1r3. This research was supported by grants fromthe NIH: DC03155 (R.F.M.), MH57241 (M.M.) and DC00310 (L.H.).

Received 15 February; accepted 28 March 2001.

1. McLaughlin, S.K., McKinnon, P.J. & Margolskee, R.F. Gustducin is a taste-cellspecific G protein closely related to the transducins. Nature 357, 563–569 (1992).

2. Wong, G.T., Gannon, K.S. & Margolskee, R.F. Transduction of bitter and sweettaste by gustducin. Nature 381, 796–800 (1996).

3. Huang, L. et al. Gγ13 colocalizes with gustducin in taste receptor cells andmediates IP3 responses to bitter denatonium. Nature Neurosci. 2, 1055–1062(1999).

4. Adler, E. et al. A novel family of mammalian taste receptors. Cell 10, 693–702(2000).

5. Chandrashekar, J. et al. T2Rs function as bitter taste receptors. Cell 100, 703–711(2000).

6. Matsunami, H., Montmayeur, J.-P. & Buck, L.B. A family of candidate tastereceptors in human and mouse. Nature 404, 601–604 (2000).

7. Bachmanov, A.A. et al. Sucrose consumption in mice: major influence of twogenetic loci affecting peripheral sensory responses. Mamm. Genome 8, 545–548(1997).

8. Blizzard, D.A., Kotlus, B. & Frank, M.E. Quantitative trait loci associated withshort-term intake of sucrose, saccharin and quinine solutions in laboratory mice.Chem. Senses 24, 373–385 (1999).

9. Li, X. et al. High-resolution genetic mapping of the saccharin preference locus(Sac) and the putative sweet taste receptor (T1R1) gene (Gpr70) to mouse distalchromosome 4. Mamm. Genome 12, 13–16 (2001).

10. Fuller, J.L. Single locus control of saccharin preference in mice. J. Hered. 65, 33–36(1974).

11. Lush, I.E. The genetics of tasting in mice VI. Saccharin, acesulfame, dulcin andsucrose. Genet. Res. 53, 95–99 (1989).

12. Capeless, C.G. & Whitney, G. The genetic basis of preference for sweet substancesamong inbred strains of mice: preference ratio phenotypes and the alleles of theSac and dpa loci. Chem. Senses 20, 291–298 (1995).

13. Lush, I.E., Hornigold, N., King, P. & Stoye, J.P. The genetics of tasting in mice VII.Glycine revisited, and the chromosomal location of Sac and Soa. Genet. Res. 66,167–174 (1995).

14. Hoon, M.A. et al. Putative mammalian taste receptors: a class of taste-specificGPCRs with distinct topographic selectivity. Cell 96, 541–551 (1999).

15. Kunishima, N. et al. Structural basis of glutamate recognition by a dimericmetabotropic glutamate receptor. Nature 407, 971–977 (2000).

16. Rossler, P., Kroner, C., Freitag, J., Noe, J. & Breer, H. Identification of aphospholipase C β subtype in rat taste cells. Eur. J. Cell Biol. 77, 253–261 (1998).

17. Rossler, P. et al. Protein βγ complexes in circumvallate taste cells involved in bittertransduction. Chem. Senses 25, 413–421 (2000).

18. Hiji, T. Selective elimination of taste responses to sugars by proteolytic enzymes.Nature 256, 427–429 (1975).

19. Nakashima, K. & Ninomiya, Y. Increase in inositol 1,4,5-triphosphate levels of thefungiform papilla in response to saccharin and bitter substances in mice. CellPhysiol. Biochem. 8, 224–230 (1998).

20. Burge, C. & Karlin, S. Prediction of complete gene structures in human genomicDNA. J. Mol. Biol. 268, 78–94 (1997).

21. Ruiz-Avila, L. et al. Coupling of bitter receptor to phosphodiesterase throughtransducin in taste receptor cells. Nature 376, 80–85 (1995).

22. Hogan, B., Beddington, R., Costantini, F. & Lacy, E. Manipulating the MouseEmbryo: A Laboratory Manual (Cold Spring Harbor Laboratory, Cold SpringHarbor, 1994).

23. Thompson, J.D., Higgins, D.G. & Gibson, T.J. CLUSTAL W: improving the sensitivityof progressive multiple sequence alignment through sequence weighting,position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22,4673–4680 (1994).

24. Kelley, L.A., MacCallum, R.M. & Sternberg, M.J.E. Enhanced genome annotationusing structural profiles in the program 3D-PSSM. J. Mol. Biol. 299, 501–522(2000).

25. Sali, A. & Blundell, T.L. Comparative protein modeling by satisfaction of spatialrestraints. J. Mol. Biol. 234, 779–815 (1993).

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m© 2001 Nature Publishing Group http://genetics.nature.com