Cats Lack a Sweet Taste Receptor

The ability to taste the substance phenylthiocarbamide (PTC) has been widely used for genetic and anthropological studies, but genetic studies have produced conflicting results and demonstrated complex inheritance for this trait. We have identified a small region on chromosome 7q that shows strong linkage disequilibrium between single-nucleotide polymorphism (SNP) markers and PTC taste sensitivity in unrelated subjects. This region contains a single gene that encodes a member of the TAS2R bitter taste receptor family. We identified three coding SNPs giving rise to five haplotypes in this gene worldwide. These haplotypes completely explain the bimodal distribution of PTC taste sensitivity, thus accounting for the inheritance of the classically defined taste insensitivity and for 55 to 85% of the variance in PTC sensitivity. Distinct phenotypes were associated with specific haplotypes, which demonstrates that this gene has a direct influence on PTC taste sensitivity and that sequence variants at different sites interact with each other within the encoded gene product. Show more

Introduction Olfactory receptor (OR) genes provide the basis for the sense of smell (Buck and Axel 1991) and, with more than 1,000 genes, comprise the largest gene superfamily in mammalian genomes (Glusman et al. 2001; Zozulya et al. 2001; Young and Trask 2002; Zhang and Firestein 2002; Olender et al. 2003). OR genes are organized in clusters (Trask et al. 1998; Young and Trask 2002) and in humans are found on every chromosome save the Y and 20 (Glusman et al. 2001; Zozulya et al. 2001). On the basis of sequence similarity, they are classified into two major classes and 17 families (Glusman et al. 2001). All OR genes have an approximately 1 kb coding region that is uninterrupted by introns (Ben-Arie et al. 1994; Gilad et al. 2000). Interestingly, approximately 60% of human OR genes carry one or more coding region disruptions and are therefore considered pseudogenes (Rouquier et al. 1998; Glusman et al. 2001; Zozulya et al. 2001). In nonhuman apes, the fraction of OR pseudogenes is only approximately 30% (Gilad et al. 2003). However, both humans and other apes have a significantly higher fraction of OR pseudogenes than do the mouse or the dog (approximately 20%) (Young et al. 2002; Zhang and Firestein 2002; Olender et al. 2003). Thus, there has been a decrease in the size of the intact OR repertoire in apes relative to other mammals, with a further deterioration in humans (Rouquier et al. 2000; Gilad et al. 2003). Although the causes are unclear, it seems reasonable to speculate that the high fraction of OR pseudogenes in apes reflects a decreased reliance on the sense of smell in species for whom auditory and visual cues may be more important (e.g., Dominy and Lucas 2001). We were therefore interested in investigating whether the high fraction of OR pseudogenes is characteristic of primates as a whole and, if not, to pinpoint when the proportion of OR pseudogenes increased. To this end, we randomly selected subsets of 100 OR genes in 19 primate species, including a human, four apes, six Old World monkeys (OWMs), seven New World monkeys (NWMs) and one prosimian. We find that a decrease in the size of the intact olfactory repertoire occurred independently in two evolutionary lineages: in the ancestor of OWMs and apes, and in the New World howler monkey. Results and Discussion Owing to the high levels of DNA sequence divergence among the primate species in our sample, orthologous OR genes could not be amplified by primers designed based on human sequences (Gilad et al. 2003). Instead, we used two sets of degenerate primer pairs, constructed to amplify OR genes from all of the species studied (see Materials and Methods). We then cloned the PCR products and determined the sequences of clones until we had identified 100 distinct OR genes from each species. A danger of this approach is that degenerate primers may bind preferentially to certain OR genes, thereby resulting in a biased representation of the OR repertoire. To safeguard against this, we tested the degenerate primers on human and mouse, for which the entire OR gene repertoire is known, by using them to amplify 100 OR genes from the two species. The sample thus obtained faithfully represented the composition of the full OR gene repertoire in human and mouse with respect to the 17 OR gene families (Figure 1). Moreover, the sample estimates of the fractions of pseudogenes were accurate (see Materials and Methods; Figure 2). This pilot study demonstrates that the degenerate primers yield an unbiased representation of the OR gene repertoire, as measured by the family composition and pseudogene content of the human and mouse samples. Since the primers performed well both in human and a distantly related species, the mouse, there was no reason to assume that they would not do so in nonhuman primate species. We therefore proceeded to sequence 100 genes from 18 nonhuman primates using these primer pairs. Since the genome sequence is not available for these species, we were not able to compare the familial composition of our samples of OR genes to that of the full OR repertoires. However, with the exception of OR families 3, 11, 12, and 55 (whose absence in a sample of 100 genes is not unlikely, as they represent less than 1.8% of human OR genes), we identified OR genes from all families in all species (Table 1). Moreover, the representation of the three largest OR gene families in the sample varied across species, again suggesting that there is no strong bias towards the amplification of specific families. We then tabulated the proportion of OR pseudogenes in each species (Figure 2). Consistent with previous results based on direct sequencing of full-length OR orthologs (Gilad et al. 2003), we found that the proportion of OR pseudogene in the great apes and rhesus macaque is approximately 30% (Figure 2). Together, these findings confirm the validity of this degenerate primer approach. We further found that the proportion of OR pseudogenes in OWMs (29.3% ± 2.4%) is very similar to that of nonhuman apes (33.0% ± 0.8%), but notably higher than that of NWMs (18.4% ± 5.6%). One NWM species, the howler monkey, was a conspicuous exception, with an elevated proportion of OR pseudogenes, similar to that of OWMs and apes (31.0%) (Figure 2) and significantly higher than any other NWM (one-tailed p < 0.02 for the difference between the howler monkey and the NWM with the second highest proportion of pseudogenes, the Wooly monkey, as assessed by a Fisher's exact test [FET]). Thus, it appears that a deterioration of the olfactory repertoire occurred in all apes and OWMs as well as, independently, in the howler monkey lineage. Strikingly, a second phenotype is shared only by the howler monkey, OWMs, and apes: full (or “routine”) trichromatic color vision. In primates, trichromatic color vision is accomplished by three opsin genes whose products are pigments sensitive to short, medium, or long wavelength ranges of visible light (Nathans et al. 1986). In OWMs and apes, the short-wavelength opsin gene is found on an autosome, while two distinct X-linked loci for medium and long wavelengths underlie full trichromatic color vision (and so are present in both males and females). In contrast, most NWM species carry an autosomal gene and only one X-linked gene, where different alleles encode for photopigment opsins that respond to medium or long wavelengths. Heterozygous females can therefore possess trichromatic vision, but males are dichromatic (Jacobs 1996; Boissinot et al. 1998; Hunt et al. 1998). The sole exception among NWMs is the howler monkey (Jacobs et al. 1996; Jacobs and Deegan 2001; Surridge et al. 2003), which has a duplication of the opsin genes on the X chromosome (Goodman et al. 1998; Jacobs and Deegan 2001) (Figure 3). Thus, full trichromatic vision arose twice in primates, once in the common ancestor of OWMs and apes and once in the howler monkey lineage. While OWMs, apes, and the howler monkey carry 32.5% ± 6.3% OR pseudogenes in their OR gene repertoire, species without full trichromatic vision have 16.7% ± 1.0%, significantly fewer (p < 10−4, or, excluding humans from the full trichromatic group, p < 10−3, as assessed by a Mann–Whitney U test). This p value is only indicative since the species lineages are not all independent. However, if significance is instead assessed by a FET for all pairwise comparisons of species with full trichromatic color vision and without, the difference is again striking: 94 out of 96 comparisons are significant at the 5% level. Thus, the evolution of full trichromatic vision coincided with an increase in the fraction of OR pseudogenes, indicative of a deterioration of the sense of smell. Apes and OWMs acquired trichromatic color vision approximately 23 million years ago (Yokoyama and Yokoyama 1989), while the duplication of the opsin genes in the howler monkey occurred approximately 7–16 million years ago (Jacobs 1996; Cortes-Ortiz et al. 2003). In spite of this difference in timing, the proportion of OR pseudogenes in species from both lineages is very similar. We estimated the rate of fixation of neutral gene disruptions for OR genes to be approximately 0.12 per gene per million years (Y. Gilad, S. Pääbo, and G. Glusman, unpublished data). This estimate implies that both apes, OWMs and the howler monkey could have a much higher proportion of OR pseudogenes than observed (data not shown), indicating that the process of functional OR gene loss has decreased or stopped in these species. A plausible explanation for the similar proportion of OR pseudogenes in the different lineages is that while full trichromatic vision relaxed the need for a sensitive sense of smell, it did not render olfaction unnecessary. Accordingly, while some OR genes can accumulate coding region disruptions, others are still evolving under evolutionary constraint. This model predicts that the possession of full trichromatic color vision alone allows for the loss of some but not all OR genes. A natural next step would then be to identify which OR genes or families were lost after the acquisition of full trichromatic vision. The answer to this question awaits sequence from a large number of orthologous OR genes. In this respect, it is interesting to note that the TRP2 gene, a major component of the vomeronasal pheromone transduction pathway, was found to be intact in several NWM species, but is a pseudogene in OWMs and apes (Liman and Innan 2003; Zhang and Webb 2003). The authors raised the possibility of a connection between the acquisition of full trichromatic color vision and decreased pheromone perception, based on the difference between OWMs and apes on the one hand and NWMs on the other (Liman and Innan 2003; Zhang and Webb 2003). However, since many traits can potentially be mapped to the lineage that leads to OWMs and apes, the connection between full trichromatic vision and pheromone perception was tenuous. Furthermore, Liman and Innan (2003) did not find a coding region disruption in four exons of TRP2 in the howler monkey. An intact TRP2 gene in the howler monkey would be inconsistent with the hypothesis that the enhancement of color vision replaced pheromone signaling in primates. In contrast, in the present study, we find that the deterioration of the olfactory repertoire occurred concomitant with the evolution of full trichromatic vision in two separate primate lineages. Thus, although at this point we are unable to demonstrate that the decline in the sense of smell is a direct result of the evolution of color vision, our results strongly suggest an exchange in the importance of these two senses in primate evolution. Future studies of the sensory cues involved in detection and selection of food (e.g., Smith et al. 2003), or the choice of a mate, may test this association directly. Materials and Methods Design and test of degenerate primers. OR genes have a coding region that is approximately 1 kb long and contains no introns. In order to test the performance of degenerate primers, we sequenced 30 genes amplified with each primer pair in human and mouse and compared the composition of the different OR families in the sample to that of the full OR gene repertoire of these two species (Glusman et al. 2001; Zhang and Firestein 2002). We also compared the sample estimates of the proportion of pseudogenes to the proportion in the entire OR repertoire of human and mouse. Since the degenerate primers amplify only 670 bp of the approximately 1 kb coding region of the OR gene, a subset of the coding region disruptions will fall in segments of OR genes not amplified by our primers. As a result, the true fraction of OR genes carrying coding region disruptions will be underestimated by our approach. We therefore determined the proportion of OR genes with at least one disruption within the corresponding 670 bp in the entire human and mouse OR gene repertoires (47.7% and 16.3% in humans and mouse, respectively). We first tested an existing set of primers, used by Rouquier et al. (2000), but found significant deviations from the family composition of the full OR repertoire in both species. As an illustration, among the 60 OR genes obtained in humans, 36.6% were of the subfamily 7E (all pseudogenes), significantly more than expected given the true proportion of the 7E subfamily in the full human OR gene repertoire (12.4%, p = 2 × 10−6, assessed by FET). As a consequence of these biases, estimates of the proportion of pseudogenes in human and mouse obtained with these primers (Rouquier et al. 2000) differ significantly from the true value (p < 0.01, assessed by FET). We proceeded by designing new pairs of degenerate primers for the OR gene family by using the program HYDEN (Fuchs et al. 2002; Linhart and Shamir 2002). The first primer pair, PC1 (PC1–5′: CTSCAYSARCCCATGTWYHWYTTBCT, PC1–3′: GTYYTSAYDCHRTARAYRAYRGGGTT), was designed based on class 1 human OR sequences only. The second primer pair, PC2 (PC2–5′: YTNCAYWCHCCHATGTAYTTYTTBCT, PC2–3′: TTYCTNARGSTRTAGATNANDGGRTT), was designed based on solely class 2 human OR sequences, excluding all genes that belong to subfamily 7E. Both primer pairs were designed to amplify a 670-bp product that approximately covers the region from transmembrane domains 2–7 of the OR protein. As a first step, we used each primer pair to amplify and sequence (see below) 30 genes from human genomic DNA. We found that PC1 primer pairs amplify OR class 1 and OR class 2 genes in roughly equal proportions. PC2 primer pairs amplified only OR class 2 genes, including members of the 7E OR subfamily. Based on the OR family composition that we observed for the 60 genes, we estimated that if we constructed a sample containing 25% of genes amplified with PC1 and 75% of genes amplified with PC2, we would obtain an unbiased representation of the familial composition of the human OR gene repertoire. This approach was validated by amplifying and examining 100 genes collected in the same way from human as well as from mouse. PCR and DNA sequencing Each primer pair was used to amplify a set of eight reactions in each species using a temperature-gradient PCR. The use of several annealing temperatures for each species yielded a greater diversity of amplified OR genes. PCR was performed in a total volume of 25 μl, containing 0.2 μM of each deoxynucleotide (Promega, Madison, Wisconsin, United States), 50 pmol of each primer, 1.5 mM MgCl2, 50 mM KCl, 10 mM Tris (pH 8.3), 2 U of Taq DNA polymerase, and 50 ng of genomic DNA. Conditions for the PCR amplification from all species were as follows: 35 cycles of denaturation at 94°C, annealing at a gradient temperature of 48°C to 60°C, and extension at 72°C, each step for 1 min. The first step of denaturation and the last step of extension were 3 min each. The PCR products were separated and visualized in a 1% agarose gel. From each amplification set (a given primer pair in a given species), all successful products were mixed and subjected to cloning using a TA cloning kit (Boehringer, Mannheim, Germany). Cloning was followed by a touchdown PCR using the vector primers for amplifications from isolated bacterial colonies. Products were purified using the High Pure PCR Product Purification Kit (Boehringer). Sequencing reactions were performed in both directions on PCR products, using the vector primers and the dye-terminator cycle sequencing kit (Perkin Elmer, Wellesley, Massachusetts, United States) on an ABI 3700 automated sequencer (Perkin Elmer). Sequence analysis After base calling with the ABI Analysis Software (version 3.0), the data were edited and assembled using the Sequencher program, version 4.0 (GeneCodes Corporation, Ann Arbor, Michigan, United States). Assembly of the clones was done using a similarity cutoff of 98%. This cutoff ensures that Taq-generated mutations that may have been sequenced in individual clones are not counted as independent genes. Clones that were collapsed to the same contig by the assembly process were counted as one gene. Once 25 and 75 genes (independent contigs) were identified from PC1 and PC2 primer pairs, respectively, a majority consensus was generated for each gene. In order to confirm that only OR genes were amplified from all the species, we used the consensus sequences of all genes from all species as queries in a BLAT search against the human genome sequence (http://genome.ucsc.edu/). In every case, the best hit was a human OR gene. This analysis was also used to insure that none of the genes were an artifact of (“jumping”) PCR fusion. Finally, each consensus sequence was searched for an uninterrupted open reading frame (ORF) in all six possible frames. If an uninterrupted ORF was found, the gene was annotated as intact. If no ORF was identified, the gene was annotated as a pseudogene. This approach probably results in an underestimate of the proportion of pseudogenes, as not all OR genes with an intact coding region are functional. Mutations in promoter or control regions of OR genes may lead to reduced or no expression. Similarly, radical missense mutations in highly conserved positions of the OR protein may result in dysfunction (Menashe et al. 2003). Although it is known that there are several highly conserved positions among OR genes, it is not always straightforward to ascertain which, if any, of these positions is necessary to retain function. We therefore chose the most straightforward definition of a pseudogene: a gene without a full ORF. Supporting Information Accession Numbers Sequences for all OR genes from all primate species were deposited to GenBank (http://www.ncbi.nlm.nih.gov/Genbank/) as accession numbers AY448037–AY449380 and AY454789–AY455274. Show more

Introduction The domestic cat (Felis silvestris catus), of the family Felidae in the order Carnivora, is an obligate carnivore. Its sense of taste is distinguished by a lack of attraction to, or indifference toward, compounds that taste sweet to humans, such as sweet carbohydrates (sugars) and high-intensity sweeteners [1–3]. This behavior toward sweet stimuli is in marked contrast to the avidity for sweets shown by most omnivores and herbivores and even some other carnivores such as the dog [4]. The indifference that cats display toward sweet-tasting compounds contrasts with their otherwise normal taste behavior toward stimuli of other taste modalities. For example, they show preference for selected amino acids [5] and generally avoid stimuli that to humans taste either bitter or very sour [1,5]. Congruent with these behavioral responses to taste stimuli, recordings from cat taste nerve fibers, and from units of the geniculate ganglion innervating taste cells, demonstrate responses to salty, sour, and bitter stimuli as well as to amino acids and nucleotides, but do not show neural responses to sucrose and several other sugars [5–12]. The sense of taste in the cat, in general, is therefore similar to that of other mammals, with the exception of an inability to taste sweet stimuli. The molecular basis for this sweet blindness in cats is not known. Because the taste blindness appears specific to this single modality, we postulated that the defect in the cat (and likely in other obligate carnivores of Felidae) lies at the receptor step subtending the sweet-taste modality. The possible defects at the molecular level that might cause this sweet blindness could range from a single to a few amino acid substitutions, such as is found between sweet “taster” and “nontaster” strains of mice [13,14], to more radical mechanisms, such as an unexpressed pseudogene. To distinguish among these possibilities, we identified the DNA sequence and examined the structures of the two known genes, Tas1r2 and Tas1r3, that in other mammals encode the sweet-taste receptor heteromer, T1R2/T1R3. We compared these with the sequence and structure of the same genes in dog, human, mouse, and rat—all species that display a functional sweet-taste modality. We also sought to detect the expression of the two cat genes at both the RNA and protein levels. Our results lead us to conclude that Tas1r3 is expressed in cat taste buds and very likely is functional, whereas cat Tas1r2 is an unexpressed pseudogene. The immediate repercussion of this unexpressed gene is that the heteromer normally acting as a sweet-taste receptor in most other mammals likely does not form in the cat. Results We identified DNA sequences of Tas1r3 and Tas1r2 from the domestic cat by screening a feline BAC library and using a PCR strategy on cat genomic DNA with degenerate primers. The feline sequences were compared with those of other species, and gene structures were determined. The expression of these two receptors was then evaluated by in situ hybridization and immunohistochemistry. Molecular Cloning of Cat Tas1r3 and Tas1r2: Sequence and Gene Structure By sequencing positive BAC clones retrieved from a feline genomic BAC library (Felis silvestris catus; BACPAC Resources, Oakland, California, United States), we obtained more than 3 kb of genomic sequences containing the open reading frame for cat Tas1r3, and approximately 10 kb of genomic sequences containing the open reading frame for cat Tas1r2. Because exons 1 and 2 of Tas1r2 were not found in the positive BAC clones, we employed a PCR strategy using degenerate primers to amplify these regions from cat genomic DNA (Novagen, San Diego, California, United States) (See Materials and Methods). We aligned the cDNA sequences and the deduced amino acid sequences from cat Tas1r3 and Tas1r2 with their dog, human, mouse, and rat orthologs (Figure 1). (We obtained the sequences of domestic dog genes, Tas1r3 and Tas1r2, by screening a dog genomic library using the same overgo probes and methods as for the feline genomic BAC library and by taking advantage of the limited data available at that time from the public dog genome database at http://www.ncbi.nlm.nih.gov/genome/guide/dog/). Figure 1 Alignment of Deduced Amino Acid Sequences of T1R3 and T1R2 from Five Species This figure shows the alignment of the deduced sequences of the taste receptor proteins, T1R3 and T1R2, from domestic cat, domestic dog, human, mouse, and rat. Amino acids that are identical among species are shaded in black; conservative amino acid substitutions are shaded in gray. The cat T1R3 sequence shows high similarity with that of human and rodents, with especially high similarity with that of dog. The predicted cat T1R2 sequence is truncated at amino acid 355 due to a premature stop codon at bp 57–59 in exon 4, which results from a 247-bp deletion in exon 3. The underlined amino acids from 316 to 355 of the cat T1R2 result from the frame shift brought by the 247-bp deletion in exon 3. Note that the deduced amino acid sequence of dog T1R2 predicts an apparently normal protein showing high similarity with that of rat, mouse, and human. Table 1 presents the percent similarity of the Tas1r3 and Tas1r2 genes at both the cDNA and the protein levels between all possible pairs of five species: cat, dog, human, mouse, and rat. The cat Tas1r3 gene shows high similarity at the cDNA level with that of dog (87%), human (79%), rat (75%), and mouse (74%) (Table 1). The cat Tas1r3 gene predicts a protein of 865 amino acids (Figure 1) showing 85% similarity with deduced protein of dog, and 73%, 72%, and 72% with that of human, mouse, and rat, respectively (Table 1). Initially we predicted the exon–intron boundaries of cat Tas1r3 by comparison with the known boundaries of human TAS1R3. To confirm these exon–intron boundaries for cat Tas1r3, we performed both RT-PCR on cDNA from cat taste bud–containing circumvallate and fungiform papillae, and PCR on cat genomic DNA using intron-spanning primers. By comparing the cDNA sequence with the genomic sequence, we confirmed the boundaries predicted from human TAS1R3 (Figure 2A). Both the cat Tas1r3 and human TAS1R3 genes are composed of six similarly sized exons and five introns (Figure 2A). Table 1 Similarity of Sweet Receptors between Species T1R2 and T1R3 are the protein names, and Tas1r2 and Tas1r3 are the corresponding gene names. Columns for T1R2 and T1R3 show percent similarity between predicted amino acid sequences; columns for Tas1r2 and Tas1r3 show percent similarity between cDNA sequences. Figure 2 Gene Structures of Cat Tas1r3, Human TAS1R3, and Cat Tas1r2, Human TAS1R2 The exons are shown in black (size in bp of each exon is in parentheses). Boundaries of gene sequences used to produce probes for in situ hybridization studies (Figure 3) are shown by the horizontal lines labeled “P1” and “P2” under the sketch of the cat Tas1r3 and cat Tas1r2. Boundaries of sequence used to generate peptide antigens for immunohistochemical studies (Figure 4) are shown by the horizontal lines labeled “A” under the sketch of the cat Tas1r3 and cat Tas1r2. The locations referred to in the vertical explanation text above the asterisks and the spade symbol indicate the position in bp within each exon. Intron sizes shown in the figure are not proportionally scaled on both (A) and (B) because of the large size of Tas1r2 introns. Under each human exon is the percent similarity between each human exon and its cat counterpart at the nucleotide level (Figure 2B). The exons for cat Tas1r2 refer to parts corresponding to human exons. The spade symbol (♠) indicates the position of microdeletion in exon 3 of cat Tas1r2. Asterisks (*) indicate the stop codon positions in exon 4 and 6 of cat Tas1r2. Note that nucleotide numbers of the exon 3 in human TAS1R2 and cat Tas1r2 are not identical. Figure 3 RNA Expression of Cat Tas1r2 and Tas1r3 from Circumvallate Papillae Digoxigenin-labeled sense and antisense cRNA probes corresponding to exons 3 and 6 of cat Tas1r2 and Tas1r3 were synthesized using DIG RNA labeling kit (Roche Applied Science, Indianapolis, Indiana, United States) (See Figure 2 for the locations of in situ probes, and Table 3 for identity of primers.) Hybridizations were carried as described [39]. Panel (A) shows result of antisense probes for Tas1r3, whereas panel (B) shows the result of the sense probes for Tas1r3. Panel (C) shows results of the antisense probes for Tas1r2 whereas panel (D) shows results of the sense probes. Scale bar, shown only in panel (A), = 60 μm for (A), (B), (C), and (D). Figure 4 Protein Expression of Cat T1R2 and T1R3 Cat T1R3 expression is detected in taste buds of circumvallate papilla (CV) (A) and a fungiform papilla (Fun) just anterior to the intermolar eminence (B) by labeling with anti-mouse T1R3 antibody. Cat T1R2 expression is not detectable in either circumvallate (C) or fungiform (D) using an anti-cat T1R2 antibody. Control studies demonstrated that the anti-cat T1R2 antibody labeled a subset of taste bud cells in rat circumvallate (data not shown). Scale bar, shown only in panel (A) and (B), = 60 μm for (A) and = 45 μm for (B). Scale for panel (C) is the same as that of panel (A); scale for panel (D) is the same as that of panel (B). We identified the exon–intron boundaries of cat Tas1r2 by comparison with known human boundaries (Figure 2B). Examining the sequence of cat Tas1r2, we discovered a microdeletion of 247 base pairs (bp) within exon 3. This deletion is responsible for a frame shift that results in a premature stop codon at bp 57–59 of exon 4 (Figure 2B). Assuming, for the moment, that a protein is translated from cat Tas1r2, then, because of the deletion and premature stop codon, the gene sequence predicts a peptide of 355 amino acids, the first 315 of which show high similarity with their rat, mouse, human, and dog counterparts (see Figure 1). Because of the frame shift introduced by the 247-bp deletion, the remaining deduced 40 amino acids show no similarity with their rat, mouse, human, or dog counterparts (underlined sequence of cat T1R2; Figure 1). The predicted similarity of this hypothetical 355–amino acid protein was compared with its truncated counterparts from dog, human, mouse, and rat. It ranges from 55% to 69% (Table 1). In contrast, the percent similarity of the full-length T1R2 protein within pairs of other species is between 91% (mouse–rat) and 69% (mouse–human). By aligning cat Tas1r2 DNA sequences of exons 4, 5, and 6 with their human counterparts, we found four additional stop codons: one in exon 4 due to a deletion at bp 123, and three in exon 6 due to a substitution at bp 95 and a deletion at bp 247 (Figure 2B). The multiple stop codons indicate that the cat Tas1r2 is a pseudogene. In an attempt to confirm the cat Tas1r2 exon–intron boundaries, we performed RT-PCR on cDNA from cat circumvallate and fungiform taste papillae. Despite using numerous (> 70) primers corresponding to deduced message from the Tas1r2 gene, we were unable to detect it. RNA and Protein Expression Having detected message from cat Tas1r3, but not from cat Tas1r2, by RT-PCR, we used the more tissue-specific approaches of in situ hybridization and immunohistochemistry to refine the search for cat Tas1r2 gene expression, using the cat Tas1r3 gene for comparison. Probes for in situ hybridization were constructed from the gene sequences corresponding to the lines marked “P” in Figures 2A and 2B. (See Materials and Methods for details.) Figure 3 shows that message from Tas1r3 is expressed in taste buds of cat circumvallate papillae whereas Tas1r2 expression is not detectable by in situ hybridization. Antisense probes for Tas1r3 result in positive labeling (Figure 3A); the arrows indicate three of the many labeled taste buds. The control sense probes show no labeling (Figure 3B). In contrast, antisense probes for cat Tas1r2 show no detectable labeling (Figure 3C) as is the case for the sense control (Figure 3D). To detect the presence of taste receptor proteins from Tas1r2 and from Tas1r3, we exposed 10-μm sections of cat circumvallate and fungiform papillae to polyclonal antibodies developed against deduced amino acid peptide antigens marked by the line labeled “A” in Figure 2A and 2B. T1R3-like immunoreactivity was present in the taste buds of every circumvallate (10) and fungiform (4) papilla used in this study (Figure 4A and 4B) whereas immunoreactivity to T1R2 was not detected in these same tissues (Figure 4C and 4D). (Each circumvallate papilla of the cat contains approximately 400 taste buds, whereas the large fungiform papillae used in this study, located in the area of the eminence, contain from 1 to about 15 taste buds each.) The antibody to cat T1R2 did, however, label a subset of taste buds in rat circumvallate papillae (results not shown). Confirmation of Tas1r2 Sequence in Six Individual Cats, Tiger, and Cheetah Because the feline BAC genomic library was constructed from a single individual cat, we confirmed the sequence of Tas1r2 in six additional, unrelated, healthy adult domestic cats. Genomic DNA was obtained by cheek swabs from five of the six cats and through a blood sample from the remaining cat, amplified by PCR using primers that flanked the deletion and stop codons of the known cat Tas1r2, and sequenced. In addition, to test whether other species of Felidae display similar sequence anomalies in their Tas1r2 gene, we performed PCR on genomic DNA of one tiger (Therion International, Saratoga Springs, New York, United States) and one cheetah (a gift from the San Diego Zoo). We found that Tas1r2 in all six cats, the tiger, and the cheetah show the identical 247-bp deletion in exon 3, and all have stop codons at the same positions in exon 4 (Table 2). In exon 6, we found evidence for two alleles at position 93–95 in domestic cat, wherein two cats show the stop codon, TGA (homozygotes TGA/TGA); one cat shows TGR (heterozygote TGA/TGG); and three of the domestic cats, the one tiger, and the single cheetah show TGG (homozygotes TGG/TGG) (Table 2). The second exon 6 stop codon is also common to all three species (TGA for domestic cat, TAG for tiger and cheetah). Although the third stop codon of exon 6 at bp 697–699 was found in all six domestic cats, the corresponding region in tiger and cheetah could not be amplified by PCR. Table 2 Tas1r2 Stop Codons in Species of Felidae Stop codons are shown in bold. aLocation (bp) refers to the position within each exon (Figure 2B). bTwo cats are homozygotes TGA/TGA, one cat is heterozygote TGA/TGG, and three cats are homozygotes TGG/TGG. UN, unknown, region could not be amplified by PCR. Collectively, these data indicate that cat Tas1r3 is an expressed and likely functional receptor, whereas cat Tas1r2 is an unexpressed pseudogene. Discussion The taste receptors for sweetness and for umami (an amino acid–taste modality) are members of the T1R family of taste receptors [15,16,17]. These are Class C, family 3, G protein–coupled receptors (GPCR). The three known members of the T1R family are T1R1, T1R2, and T1R3 (for review, see [18]). In rodents and primates the primary sweet-taste receptor is composed of a dimer of two closely related GPCRs, T1R2 and T1R3 [14,15,16,17]. For this study, we made the working assumption that the Felidae T1R family shows specificity similar to that known from rodents and primates. Because the umami receptor is composed of the heteromer, T1R1/T1R3, and because cats can taste amino acids, it would appear likely that both of these proteins should be functional. The sweet-taste receptor is composed of the heteromer T1R2/T1R3. Because the cat cannot taste sweet stimuli, the most likely assumption is that the cat T1R2 is non-functional. Molecular Features of Cat Tas1r3 By comparison with other known T1R3 proteins and other proteins of Class C, family 3, the sequence and gene structure of cat Tas1r3 predict a functional receptor of 865 amino acids (see Figure 1). Cat Tas1r3 is assumed to be located on cat Chromosome C1, syntenic with human 1p36, where human TAS1R3 is located [19,20]. As with other Tas1r3 genes, the cat Tas1r3 is composed of six exons, each approximately the same size as those of human (see Figure 2A). The sequence of cat Tas1r3 predicts a seven-transmembrane GPCR with extended N-terminal domain (first transmembrane region spanning amino acids 572–595), features common to other T1R3 receptors. Important Class C, family 3, structural motifs can also be located in cat T1R3 including the xPKxY motif at amino acids 814–818, and the FHSCCY motif at amino acids 517–522. Additionally, although most members of Class C, family 3, GPCRs show a highly conserved arginine residue at the extreme 3′ end of transmembrane segment 3, an exception is found with human, mouse, and rat T1R3, which substitute glutamic acid (E) for arginine (R) [21]. This substitution is also found in cat T1R3 at amino acid 660 (see Figure 1; the deduced dog T1R3 substitutes glutamine (Q) for arginine at the end of TM3). Available evidence indicates that the products of cat Tas1r3 are expressed in taste buds. RT-PCR readily detected the message from Tas1r3 in lingual taste bud–containing tissues (results not shown). In situ hybridization studies confirmed the presence of message and localized it to taste buds (see Figure 3A). Polyclonal antibodies developed against T1R3 labeled taste buds in both cat circumvallate (Figure 4A) and fungiform papillae (Figure 4B). While only a few cells showed evidence of T1R3-like immunoreactivity, nearly every taste bud was labeled by both in situ hybridization and immunohistochemistry. These commonalities in gene structure and sequence, together with evidence that the cat Tas1r3 gene is expressed, are consistent with the assumption that cat Tas1r3 codes for a functional receptor. Molecular Features of Cat Tas1r2 Cat Tas1r2, on the other hand, while retaining structure similar with that of the human TAS1R2 gene (see Figure 2B), is an unexpressed pseudogene. The likely important molecular event that resulted in cat Tas1r2 becoming a pseudogene is the 247-bp deletion in exon 3. This deletion results in a frame shift that brings about a premature stop codon in exon 4 (Figure 2B). An additional stop codon can be found in exon 4, with three more in exon 6 (Figure 2B). This apparent accumulation of mutations suggests that there is no pressure from natural selection on the cat Tas1r2 gene. To determine if this gene is expressed, we performed studies to detect message from cat Tas1r2 by RT-PCR of taste bud–containing lingual papillae and by in situ hybridization. For RT-PCR, numerous (>70) primers were constructed based on sequences from exons 1–6. For in situ hybridization, probes were designed from exon 3 and from exon 6 (Figure 2B; Table 3). Both techniques failed to detect message from cat Tas1r2 (see Figure 3C and 3D). Consistent with these attempts to detect message from cat Tas1r2, immunohistochemistry using an antibody developed from a deduced amino acid sequence spanning exons 2 and 3 revealed no labeling of taste buds in circumvallate or fungiform papillae (Figure 4C and 4D). These results suggest that the cat Tas1r2 pseudogene is not transcribed, or if it is transcribed, it rapidly degrades, perhaps through a nonsense-mediated mRNA decay pathway [22]. Table 3 Primers for In Situ Probes Tm, melting (annealing) temperature. Tas1r2 in Felidae The generality of the pseudogene nature of cat Tas1r2 was confirmed by sequencing the deletion and stop codon areas from six individual healthy adult cats. All showed the deletion and similar stop codons with some polymorphism (see Table 2). To assess the generality of the pseudogene nature of Tas1r2 in Felidae, we sequenced the stop codon areas and the area including the exon 3 microdeletion from genomic DNA of tiger and cheetah. These too displayed microdeletion and stop codons at the same location as the domestic cat. These observations, suggesting that in at least three species of Felidae Tas1r2 is not expressed, are consistent with behavioral evidence showing that, not only domestic cats, but also tigers and cheetahs do not prefer sweetened water over plain water [1]. According to morphological and molecular evidence, the available phylogeny of the order Carnivora consists of two groups, the Feliformia (cats, mongooses, civets, and hyenas) and the Caniformia (wolves, bears, raccoons, mustelids, and pinnipeds) [23,24]. It is difficult to determine when the alteration of Tas1r2 occurred and whether it preceded or followed the cat ancestor's change in diet to exclude plants. Clearly, because dogs have a human-like T1R2 structure (see Figure 1) and an avidity for sweet carbohydrates [25], the changes in the cat Tas1r2 must have occurred after the divergence of the Feliformia and the Caniformia. Genes Affecting Taste Behavior Taste receptors are shaped by and reflect a species' food choices. The genes encoding taste receptors often show a good deal of variation both among species and among individuals. These variations, both subtle and obvious, can have a variety of effects on taste sensitivity and preference behavior. A textbook example of this effect is the individual variation seen in sensitivity to the bitter compound, phenylthiocarbamide (PTC). A gene of the human TAS2R family of bitter taste receptors, TAS2R38, associated with this individual variation, shows three coding single-nucleotide polymorphisms giving rise to five haplotypes worldwide, accounting for the 55% to 85% of the variance in PTC sensitivity [26]. Further, in Drosophila, the behavioral and electrophysiological responses to trehalose are diminished in two mutants that carry deletions in the trehalose recognition gene, Gr5a [27]. In the mouse, variation in preference for sweet-tasting stimuli maps to the gene for T1R3, located within the Sac locus [28,29]. This gene is allelic in mice, and several reports identify a missense mutation (I60T) as being the most likely mutation accounting for the phenotypic differences [13,14,16,30–33]. However, the same alleles are not involved in strain-dependent sweet-taste preference in rats [34]. In addition to the modulation of behavior that can be caused by point mutations, more profound behavioral changes can result from the abolishment of gene function through, for example, the generation of pseudogenes. An example of this effect in mammalian chemoreception lies within the large repertoire of olfactory receptor genes. More than 60% of the human olfactory receptor genes are pseudogenes [35], whereas, only 20% are classified as such in mouse [35,36]. Strikingly, the accumulation of these olfactory pseudogenes in primates reportedly occurred concomitant with the acquisition of trichromatic color vision, perhaps reflecting the overarching behavioral changes that such an acquisition engendered [37]. Similar generation of bitter-taste receptor pseudogenes, accompanied by a large number of coding region single-nucleotide polymorphism, can account for the broad diversity displayed by the bitter-taste receptor family. This diversity may possibly play an important role in both species-specific and individually manifested taste preference [38]. In the extreme case, where a species fails to respond to stimuli representative of an entire modality, such as the cat with sweet taste, the development of a unique food preference behavior, based on the remaining taste receptors, might be anticipated. Because, with the exception of the sweetness modality, the taste system of the cat is organized much like that of most other mammals, discovering the molecular basis for the cats' lack of response to sweet-tasting compounds gives us a window on the development of strict carnivorous behavior in Felidae. Conclusion It is known that Felidae do not detect sweetness of carbohydrates yet can taste amino acids. Our results indicate that the gene encoding one member of the sweet-taste receptor heteromer is an unexpressed pseudogene. Given this observation, we suggest that the most parsimonious explanation for the inability of Felidae to respond to sweeteners is the lack of a functional T1R2 protein. Materials and Methods Animal tissue. We obtained cat taste tissue from healthy young-adult animals euthanized for reasons unrelated to this study. Animals were cared for under protocols 033400 and 057600 approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania to Dr. Mark Haskins of the School of Veterinary Medicine, University of Pennsylvania. Preparation of overgo probes. Overgo probes are comprised of two 22mers with a complementary eight-base overlap. They can be designed by a computer program (http://genome.wustl.edu/tools/?overgo=1) and are readily synthesized. To identify cat Tas1r2 and Tas1r3, overgo probes were designed by aligning conserved coding regions of Tas1r2 and Tas1r3 sequences from different species. The single-stranded overhangs (14 bases) were filled in with 33P-labeled dATP and dCTP, and the overgo probes were used in hybridization procedures with the BAC libraries. Screening a feline genomic BAC library. Tas1r2 and Tas1r3 overgo probes were radioactively labeled by the random hexa-nucleotide method, and hybridization and washing of membranes were as described [29]. We identified 47 positive BAC clones for cat Tas1r2 and cat Tas1r3, and sequenced all of the positive BAC ends. By aligning BAC ends sequences with human syntenic regions (human TAS1R2 and TAS1R3 are located on chromosome 1p36), we picked BAC clones positive for cat Tas1r2 and Tas1r3 for shotgun library preparation. Production of shotgun libraries for BACs containing cat Tas1r2 and Tas1r3. We prepared BAC DNAs from positive clones by using a Qiagen Large Construct Kit (Valencia, California, United States). The BAC DNAs were digested using Sau3A I and the digested BAC DNA fragments subcloned into pGEM-3Z (Promega) vector. After transformants were arrayed to a nylon membrane, two separate hybridizations were performed by using pooled Tas1r2 and Tas1r3 overgo probes. By sequencing positive clones from the shotgun libraries and by using a chromosome walking strategy, we obtained the full coding region of the cat Tas1r3 and exon 3 to exon 6 of cat Tas1r2. Identification of exon 1 and exon 2 of the cat Tas1r2 by PCR strategy. Because exon 1 and exon 2 of the cat Tas1r2 were not present in the positive BACs selected above, we designed degenerate primers based on Tas1r2 alignments from different species (human, rodents, and dog) and performed PCR using cat genomic DNA as a template. The PCR products were sequenced. The feline BAC library was then re-screened using PCR products as probes, and new positive BAC clones were retrieved. Using a chromosome walking strategy, we obtained the complete sequence of exon 1 and exon 2 of cat Tas1r2 from these newly retrieved BAC clones. RT-PCR. To examine the RNA expression and to determine the intron–exon boundaries of the cat Tas1r2 and Tas1r3 genes, we extracted total RNA using TRIZOl Reagent (Life Technologies Inc., Rockville, Maryland, United States) from cat taste bud–containing tissues, followed by reverse transcription (Superscript reverse transcriptase, Life Technologies). The cDNA samples were amplified using AmpliTaq DNA Polymerase with GeneAmp (Perkin Elmer Corporation, Branchburg, New Jersey, United States) and intron-spanning primers selected to distinguish genomic and cDNA. Single bands of expected sizes were excised from the gel, purified, and sequenced. In situ hybridization. The probes corresponding to exons 3 and 6 of cat Tas1r2 and Tas1r3 were amplified by PCR using the primers described in Table 3. Digoxigenin-labeled cRNA probes were synthesized using a DIG RNA labeling kit (Roche). Taste bud–containing vallate tongue tissue was obtained as above. Fresh frozen sections (14 μm/section) of cat circumvallate papillae were attached to clean SuperFrost/Plus slides and prepared for in situ hybridization [39]. High-stringency hybridizations were carried out at 70 °C overnight in 50% formamide, 5X SSC, 5X Denhardt's, 250 μg/ml yeast RNA, and 500 μg/ml sperm DNA using the mixed cRNA probes. Sections were washed at 72 °C with 0.2X SSC three times. Signals were detected using alkaline phosphatase–conjugated antibodies to digoxigenin and standard chromogenic substrates and observed with a Nikon SA Microphot Microscope. Control hybridizations were performed with sense probes. Immunohistochemistry. Polyclonal anti-cat T1R2 rabbit antisera directed against an N-terminal peptide of cat T1R2 (exons 2 and 3; see Figure 1) were generated by Zymed Laboratories, Inc. (South San Francisco, California, United States). Generation of antisera directed against N-terminal peptide of mouse T1R3 has been described previously [33]. Lingual tissue blocks containing cat circumvallate and fungiform papillae were fixed in 4% paraformaldehyde for 2–6 h, then processed [40]. The antibodies were incubated with the sections (10 μm/section) for 60 h at 4 °C. After washing, the sections were incubated with the secondary antibody (Cy3-conjugated goat anti-rabbit IgG; The Jackson Laboratory, Bar Harbor, Maine, United States) and observed with a Leica TCS SP2 Spectral Confocal Microscope (Leica Microsystems Inc., Mannheim, Germany). Single-channel fluorescence images (average projection of 20–25, 0.3-μm optical sections) were processed with Adobe Photoshop software and overlaid on their respective difference interference contrast images. Examination of stop codons in six individual cats and other species within Feliformia. To confirm that Tas1r2 is a pseudogene in other cats, we obtained genomic DNA from cheek swabs or blood of six unrelated healthy adult cats. We sequenced the areas around the microdeletion and the stop codons by PCR using primers that flanked these areas of interest. In addition, to test whether other species of Felidae have a functional Tas1r2 gene, we performed PCR on genomic DNA of one tiger (Therion International, Saratoga Springs, New York, United States) and one cheetah (a gift from the San Diego Zoo) using the same primers above. All the PCR products are purified and sequenced. Supporting Information Accession Numbers The GenBank (http://www.ncbi.nlm.nih.gov/Genbank/) accession numbers for the genomes discussed in this paper are cat Tas1r3 (AY819786), cat Tas1r2 (AY819787), dog Tas1r2 (AY916758), dog Tas1r3 (AY916759), human TAS1R3 (BK000152), and human TAS1R2 (NM_152232). Show more