Genetic and Molecular Basis of Individual Differences in Human Umami Taste Perception

Umami taste (corresponds to savory in English) is elicited by L-glutamate, typically as its Na salt (monosodium glutamate: MSG), and is one of five basic taste qualities that plays a key role in intake of amino acids. A particular property of umami is the synergistic potentiation of glutamate by purine nucleotide monophosphates (IMP, GMP). A heterodimer of a G protein coupled receptor, TAS1R1 and TAS1R3, is proposed to function as its receptor. However, little is known about genetic variation of TAS1R1 and TAS1R3 and its potential links with individual differences in umami sensitivity. Here we investigated the association between recognition thresholds for umami substances and genetic variations in human TAS1R1 and TAS1R3, and the functions of TAS1R1/TAS1R3 variants using a heterologous expression system. Our study demonstrated that the TAS1R1-372T creates a more sensitive umami receptor than -372A, while TAS1R3-757C creates a less sensitive one than -757R for MSG and MSG plus IMP, and showed a strong correlation between the recognition thresholds and in vitro dose - response relationships. These results in human studies support the propositions that a TAS1R1/TAS1R3 heterodimer acts as an umami receptor, and that genetic variation in this heterodimer directly affects umami taste sensitivity.


Introduction
Taste is the sensory system devoted to the evaluation of the quality of potential foods and to the determination of whether they should be ingested or rejected. The sense of taste is widely believed to be composed of a small number of primary qualities, in particular sweet, sour, bitter, salty and umami (a Japanese word that roughly translates into delicious and corresponds in many ways to savory in English). Of these five taste qualities or modalities, umami is elicited by L-glutamate, typically as its Na salt (monosodium glutamate: MSG), some amino acids and purine nucleotides (such as IMP and GMP). A salient feature of umami taste in rodents and humans is their impressive potentiation by purine nucleotides [1,2]. Many protein (amino acid) rich foods, including meat, milk and seafood taste delicious (umami) to humans, and are attractive to rodents and other animals, suggesting that umami perception plays a particularly key role in ingestion of amino acids (especially L-glutamate) which act as biosynthetic precursors of various molecules, metabolic fuels and neurotransmitters.
Here we provide additional evidence favoring TAS1R1/ TAS1R3 as an umami receptor and to elucidate underlying molecular mechanisms for individual variation in umami sensitivity, by investigating potential associations between the taste recognition thresholds for umami substances and genetic variations of TAS1R1 and TAS1R3, and functionality of TAS1R1/ TAS1R3 variants using heterologous expression system.

Taste recognition thresholds for umami compounds
Distributions of MSG, IMP and MSG plus 0.5 mM IMP (M+I) taste recognition thresholds are presented in Figure 1. The means (6S.E.) recognition thresholds for MSG, IMP and M+I were 7.6660.43, 2.1560.12 and 0.2060.016, respectively. No significant differences in the mean thresholds between males and females were found (MSG: male vs. female = 8.2760.60 vs. 6.9160.62, p = 0.12; IMP: 2.1760.17 vs. 2.1160.17, p = 0.80; M+I: 0.2060.02 vs. 0.1960.02, p = 0.81, t-test). Therefore, data for both genders were combined. A normal distribution was found for both MSG and IMP, whereas a bi-modally like distribution with the node between 0.048 and 0.098 mM, was observed for M+I.
Association between three SNPs of human TAS1Rs and taste recognition thresholds for umami compounds Based on statistical criteria we established for taste recognition threshold measurements for umami compounds [higher or lower than the mean6half a standard deviation for MSG and IMP, the node between 0.048 and 0.098 mM for M+I], samples examined both thresholds and a genotype for amino acid position 12 and 372 in TAS1R1 or 757 in TAS1R3 were divided into two groups [MSG: low taste recognition threshold group (0.2,3.13 mM, n = 86,91) vs. high (12.5,50 mM, n = 63,67), IMP: low (0.02,0.625 mM, n = 57,62) vs. high (5.0,40 mM, n = 43,45), M+I: low (0.003,0.048 mM, n = 80,85) vs. high (0.098,12.5 mM, n = 151,160)] and analyzed statistically ( Table 2). Significant associations were observed between the genotype at 372 in TAS1R1 and recognition thresholds for MSG (x 2 = 14.6, p = 0.0007) and M+I (x 2 = 6.19, p = 0.04), and between allele frequency at 757 in TAS1R3 and recognition threshold for IMP (x 2 = 6.19, p = 0.01). Regarding the genotype and allele frequency at 12 in TAS1R1, no significant differences between low and high taste recognition threshold groups were observed for MSG, IMP or M+I (x 2 = 0.0005,0.12, p = 0.83,0.98).
Distribution of taste recognition thresholds for umami substances in groups defined by human TAS1R1/TAS1R3 haplotypes Next, to analyze effects of amino acid substitutions (TAS1R1-A372T, TAS1R3-R757C) associated with differences in umami taste recognition thresholds, subjects were divided into three groups according to their TAS1R1-372/TAS1R3-757 haplotypes [(1) AR homozygotes (372A/757R, n = 84), (2) TR homozygotes (372T/ 757R, n = 37) and (3)  was higher than those of TR and AR. These results suggest that the TAS1R1-372T allele may be a more sensitive allele than the -372A allele for MSG and M+I, but not for IMP. The TAS1R3-757C allele may be the less sensitive allele than the -757R allele for MSG, IMP and M+I. For the other basic taste stimuli (sucrose, NaCl, HCl and PTC), no such differences in distributions were observed among TR, AR and TAS1R3-757C [see supporting information (SI) Figure S1].

Functional expression of human TAS1R1 and TAS1R3
In order to assess the effects of two amino acid mutations at the molecular level, we used a functional expression analysis using human embryonic kidney 293 (HEK) cells as has been used in the characterization of bitter, sweet and umami taste receptors [11,12,15,17]. Major type, TAS1R1-372A and TAS1R3-757R, were transiently transfected in HEK cells along with G a16-i3. We then monitored activation (Ca 2+ mobilization) by a Ca 2+ indicator dye, using various tastants ( Figure S2A). MSG (5,100 mM) elicited transient intracellular calcium increases in the transfected cells in a concentration dependent manner. Synergism between MSG and IMP is a hallmark of umami taste [1,2]. IMP (1, 10 mM) alone did not activate TAS1R1/TAS1R3 (data not shown), but 0.5 mM IMP potentiated the TAS1R1/TAS1R3 response to 1 and 20 mM MSG ( Figure S2A and B). The transfected HEK cells did not respond to 50 mM sucrose, 50 mM NaCl ( Figure S2A), 50 mM glucose, 1 mM saccharin and 0.3 mM SC45647 (data not shown). Mock-transfected HEK cells (without the three genes, G a16-i3 alone, G a16-i3/TAS1R1 without TAS1R3, G a16-i3/TAS1R3 without TAS1R1) did not respond to MSG and M+I ( Figure S2B and C). Lactisole is a sweet and umami taste inhibitor [15,18]. The inhibition effect of lactisole was clearly observed in the transfected cells ( Figure S5).

Functional characterization of amino acid substitutions, A372T in Human TAS1R1 and R757C in TAS1R3
To further examine the significant association between two amino acid substitutions and taste recognition threshold for umami compounds, we constructed mutant receptors encoding TAS1R1-372A, -372T, TAS1R3-757R and -757C. These constructs were tested for their responses to MSG and M+I. The concentrationresponse functions of cells transfected with different human TAS1R1/TAS1R3 variants are shown in Figure 3. We found that upon stimulation with MSG, cells containing TAS1R1/TAS1R3 variants responded with a robust transient increase of the intracellular-calcium levels. Synergisms between MSG and 0.5 mM IMP were observed in all cells with TAS1R1/TAS1R3 variants. To test whether there is functional difference in agonist sensitivity, we calculated the half maximal effective concentrations (EC 50 )  We also examined the effect of Q12H amino acid substitution of TAS1R1, which was expected to serve as a negative control, because we found no association between the genotypes (or allele frequencies) and taste recognition thresholds. The Q12H amino acid mutation of TAS1R1 did not affect MSG and M+I responses ( Figure S3).

Discussion
Distributions of MSG, IMP and M+I thresholds showed that individual subjects exhibited very different levels of sensitivity. Hence, large variations in recognition threshold sensitivity to umami substances were demonstrated. The distribution of MSG and M+I (Figure 1) may be made up of combinations of the varying underlying distributions of the TAS1R1-372 and TAS1R3-757 haplotypes, because we found only 2 SNPs (associated with differences in taste recognition thresholds) with high minor allele frequencies in human TAS1R1 and TAS1R3. Distributions of the 3 groups divided according to TAS1R1-A372T and TAS1R3-R757C haplotypes showed 3 types of curves with about 2,3-fold different mean values in MSG and M+I recognition thresholds (Figure 2). The mean values were TR,AR,TAS1R3-757C. Interestingly, the distribution of the mean values for IMP alone (TR = AR,TAS1R3-757C) was different compared with MSG (M+I). These results suggest that amino acid position 372 in TAS1R1 may modulate the binding domain for MSG but not IMP, and that the binding region for MSG may differ from that for IMP. These possibilities may be supported by previous studies using TAS1R2/TAS1R3 sweet receptor chimeras and mutants, which showed that there are at least three potential binding sites in this heterodimeric receptor [15,[18][19][20], and by a recent study using TAS1R1/TAS1R3 mutants [21]. Amino acid position 757 in TAS1R3 may be involved with broader functions (see below) because of effects on all umami compounds.
The relationship between threshold and suprathreshold taste perception is not clear [22][23][24]. In this study, therefore, we statistically examined whether the amino acid substitutions affect the receptor function at both low and middle concentrations in addition to comparing EC 50 values. The functional analysis using HEK293 cells revealed that the human TAS1R1-372T is more sensitive than -372A, while TAS1R3-757C is less sensitive than -757R at both low and middle concentrations of MSG (5 and 20 mM) and M+I (0.3 and 5 mM). The amino acid substitution, TAS1R1-Q12H, did not affect on MSG and M+I responses. Thus, a strong correlation between the taste recognition thresholds and in vitro dose -response relationships was observed. The association between TAS1R3-R757C and the recognition threshold for IMP alone remains unclear, because transfected HEK cells showed no responses to IMP alone [11,12,15]. The apparent umami taste attributed to IMP alone in vivo may actually results from the interaction of IMP with a small concentration of glutamate present in saliva [25].
TAS1R1 and TAS1R3 are GPCR and belong to family C [4][5][6][7][8][9][10]. The family C GPCR has a large extracellular region that recognizes a specific ligand molecule [26,27]. A conserved domain search in the NCBI database indicated that 372 in TAS1R1 is present in the predicted ligand binding domain, suggesting that TAS1R1-A372T amino acid substitution may lead to a confor- Ca 2+ responses to MSG and M+I were smaller in the cells transfected with TAS1R3-757C than in those with -757R, even with more sensitive TAS1R1-372T. The protein expression rates were not different between 757R and 757C (50,55%, Figure S4). These results suggest that the TAS1R3-R757C amino acid substitution may affect general functions of TAS1R1/TAS1R3 such as cell surface expression, dimerization of TAS1R1 and TAS1R3 or the dynamic equilibrium between the active and resting conformations modulated by the presence/absence of ligand as shown in mGluR1 [26,27].
Population diversities of SNPs in TAS1R1 and TAS1R3 have been reported by Kim et al. [3]. Minor allele frequencies of the SNP at 372 in TAS1R1 vary among eight populations; 10% in Cameroonian, 0% in Amerindian (native Americans), 25% in North European, 35% in Japanese, 5% in Russian, 35% in Hungarian, 40% in Chinese and 6% in Pakistani, whereas those at 757 in TAS1R3 showed no obvious difference among populations. These results suggest that there may be differences in umami sensitivity related with TAS1R1-A372T among populations in the world.
We showed that the TAS1R1/TAS1R3 genetic variation affects umami sensitivity. However, this is probably not the only factor involved. For example, three splicing variants of TAS1R1 were reported in humans (GenBank accession: NM_177539.1, NM_177540.1 and NM_177541.1). Splice variants of metabotropic glutamate receptors, mGluR1 and mGluR4, have also been proposed as taste receptors for glutamate [28,29]. It is possible that differences in expression of three TAS1R1 variants or genetic variations of mGluR1 and mGluR4 may cause differences in umami sensitivity.
Our study demonstrated that the TAS1R1-372T creates a more sensitive umami receptor than -372A, while TAS1R3-757C creates a less sensitive one than -757R for MSG and MSG plus IMP, and showed a strong correlation between the recognition thresholds and in vitro dose -response relationships. These results in human studies support the propositions that a TAS1R1/TAS1R3 heterodimer acts as an umami receptor, and that genetic variation in this heterodimer directly affects umami taste sensitivity.

Subjects
A total of 254 healthy, non-obese, non-diabetic Japanese volunteers [male/female: 141/113, age: 20-36, body mass index (BMI); 16.5-24.8] participated in this study. The purpose of the study as well as the methods and procedure were explained to the participants and the patients in this manuscript have given written informed consent. All protocols were approved by the local ethics committee at Kyushu University. Subjects were asked to refrain from consuming alcohol the day before testing, caffeine on the day of testing, and food during the experimentation period. The inclusion criteria were satisfactory state of oral hygiene, nonsmoking, regular work, sleep, and meal schedules.

Taste recognition threshold measurement
As MSG contains sodium, great care was taken to compare individual sensitivities to MSG relative to NaCl. Recognition thresholds were, therefore, measured for five qualities of taste using different concentrations of MSG (0.2,400 mM), inosine 59monophosphate (IMP, 0.02,40 mM), MSG in the presence of 0.5 mM IMP (M+I, 0.003,12.5 mM), sucrose (0.2,400 mM), NaCl (0.2,400 mM), HCl (0.02,40 mM) and PTC (0.001,2.0 mM). The differences in concentrations in the present study were in 0.3 log steps, which mean step-wise increases in concentration by a factor of 2.0. All solutions were made up in distilled water and used at room temperature. The testing procedure was the staircase-method modified from Pasquet et al. [30]. Each subject evaluated all 7 tastants. The seven series of solutions were presented one after another within a single section. Within each series, the solutions with each 2.0 ml were applied to the whole tongue of each subject with a pipet in order of ascending concentrations. The subjects were asked not to swallow test solutions and rinse their mouth between two solutions with distilled water. The subject was requested to correctly name the taste quality in each series. Once the taste of two successive concentrations was recognized successfully, the subject was given the previous unrecognized concentration (first reversal). The points at which concentration sequence changed from a decrease to an increase or vice versa were designated ''reversals''. The procedure was terminated after five reversals, and the threshold was calculated as the mean of the concentration values of the last four reversals.
Genotyping DNA samples were extracted from buccal cells by a genomic DNA purification kit (Gentra systems) after measurement of taste recognition thresholds. The entire coding regions of human TAS1R1 and TAS1R3 sequences were amplified with specific PCR primers (Table S1). Amplified PCR products were sequenced with 10 pmoles forward and reverse primers. Sequence analysis and contig assembly were performed with AssemblyLIGN (v.1.09, Oxford Molecular Group). Sequenced numbers of each region were at least fifty. Resulting data has been deposited in GenBank.

Preparation of chimeras and point mutations
Intronless human TAS1Rs and Gal6 expression constructs were generated in the pEF-DEST51 Gateway vector (Invitrogen) by genomic DNA-based methods. To subclone each gene into the vector, a Kozak cassette was introduced at the 59 end before the start codon. Gal6 chimeras were generated by PCR with mutagenic primers. The carboxyl-terminal tail of Gal6, fiveresidue (EINLL) was replaced by its counterpart from Gai3 (ECGLY; Ga16-i3), Gagust (DCGLF; Ga16-gust) [12], and Ga16-gust44 was generated as described [31]. The integrity of all DNA constructs was confirmed by DNA sequencing. Point mutations in TAS1Rs were made using the same overlapping PCR strategy with mutagenic primers. In preliminary studies, the Ga16-i3 elicited larger and more reliable responses than did Ga16, Ga16-gust or Ga16-gust44 (data not shown), therefore, Ga16-i3 was used in all experiments.

Functional expression
HEK293 cells were cultured at 37uC under a humidified atmosphere containing 5% CO 2 in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. To obtain reproducible Ca 2+ responses, cells were split every 2 days before cells become confluent. Cells were discarded after 2 months of passages and new cells were prepared from frozen-stock. For calcium imaging experiments, cells were seeded onto a 35 mm recording chamber or poly-D-lysine coated 96-well black-wall, clear-bottom recording plates. After 24 hrs at 37uC, confluent cells (60,70%) were washed in OptiMEM medium supplemented with GlutaMAX-I (invitrogen) and plasmid DNAs were transiently cotransfected into HEK293 cells using Lipofectamine2000 regent (invitrogen) (2.0 ml per 1.0 mg DNA). TAS1Rs (or their mutants) and Gal6-i3 were transfected using 1.5 and 1.0 mg of plasmid for 35 mm recording chambers, 0.09 and 0.06 mg of plasmid per well for 96-well plates, respectively. The transfection efficiencies were estimated by visualizing DsRed fluorescence (data not shown) or by immunocytochemistry of TAS1R3, and were typically 50,55% ( Figure S4). Immuno staining and Ca 2+ imaging assays were performed 36,40 hrs after transfection.

Ca 2+ imaging
Two Ca 2+ imaging systems were used. A and NaCl (50 mM). Isoproterenol (ISO, 10 mM) was used as positive control, which stimulates endogenous b-adrenergic receptors, providing that the Ga16-dependent signal transduction cascade was functional [12].
Another imaging system, FlexStation II (Molecular Devices) was used for TAS1R mutant analysis. Transfected cells in 96-well plates were washed once with HBSS and then loaded with 50 ml of FLIPR Calcium 4 (Molecular Devices) in HBSS with 20 mM HEPES (pH 7.4) for 1 hr at 37uC as described in the Kit manual. The dyeloaded cells were subjected to FlexStation II. Fluorescence changes (excitation at 485 nm, emission at 525 nm, and cutoff at 515 nm) were monitored at 2 sec intervals; 50 ml of HBSS supplemented with 26tastants were added at 30 sec, and scanning was continued for an additional 120 sec. Fluorescence responses from 20,30 sec after tastant addition were measured, corrected for background fluorescence measured before compound addition, normalized to the response to 10 mM ISO. Stimuli were tested at concentrations that do not elicit calcium responses from mock transfected Ga16 cells. Compounds were MSG (0.1, 0.3, 1, 3, 5, 10, 20, 30, 50 mM), MSG (0.1, 0.3, 1, 3, 5, 10, 20, 30, 50 mM) in the presence of 0.5 mM IMP. Lactisole (5 mM) was used as inhibitor of human TAS1R1/TAS1R3 [15,18]. Immunocytochemistry HEK293 cells transfected with TAS1Rs (or their mutants) and Ga16-i3 were rinsed with phosphate-buffered saline (PBS) and then fixed for 5 min with 4% paraformaldehyde. After three washes with PBS, cells were treated in blocking solution containing 0.5% blocking reagent (Roche) and 0.1% tween20, and then incubated with anti-TAS1R3 antibody (sc-22458, 1:50 dilution, Santa Cruz) for overnight at 4uC. After three washes with PBS, alkaline phosphatase-conjugated anti-goat secondary antibody (Jackson ImmunoResearch) was added for 1 hr at room temperature. After three washes with PBS, cells were immersed in alkaline phosphatase buffer consisting of 100 mM Tris/HCl (pH 9.5), 100 mM NaCl, and 50 mM MgCl 2 for 5 min and then the signals were developed using nitroblue-tetrazolium chloride and 5-bromo-4-chloro-3-indolylphosphate as chromogenic substrates. The reaction was stopped by rinsing in TE buffer. The signal specificities of TAS1R3 were tested by using mocktransfected HEK cells as a negative control.

Data analysis
Student's t-test was used for detection of differences in the mean recognition thresholds between males and females. The x 2 test was used for checking Hardy-Weinberg Equilibrium (HWE) for the genotype frequencies in samples to control genotyping errors, for detection of differences in genotype and allele frequencies between high and low taste recognition threshold groups. Yates'x 2 test was used when the number was less than 10. The mean values (the center6S.E. of gaussian curve on X-axis) in the distribution of recognition thresholds defined by TAS1R1/TAS1R3 haplotypes (TAS1R1-372A/TAS1R3-757R homozygotes, TAS1R1-372T/ TAS1R3-757R homozygotes, homo-and heterozygotes containing TAS1R3-757C) were calculated using gauss fit function of Origin 5.0 (OriginLab Corp). For Ca 2+ imaging analysis, data were normalized to response to 10 mM ISO, and mean values (6S.E.) were calculated from 10,18 independent analyses. EC 50 values were calculated from individual cumulative concentration-response data using sigmoid fit functions of Origin 5.0. One factor ANOVAs were conducted for detection of differences in the distributions of recognition thresholds for umami substances among 3 groups defined by TAS1R1/TAS1R3 haplotypes, and in the responses to 5, 20 and 50 mM MSG or 0.3, 5 and 50 mM M+I among TAS1R1/TAS1R3 variants in Ca 2+ imaging analysis. When the ANOVA was significant, post-hoc Student's t-test or Fisher's PLSD test was performed to compare individual means at each umami compound or concentration. Significance level was 0.05.

Supporting Information
Table S1 Nucleotide sequences for the primers used in sequencing and genotyping. Lactisole has been reported as a sweet and umami inhibitor [15,18]. The inhibition effect of lactisole (5 mM) was clearly observed in the transfected cells. Responses have been normalized to those of isoproterenol (10 mM). The values are means (6S.E.) (n = 3). Found at: doi:10.1371/journal.pone.0006717.s006 (0.24 MB EPS)