Biophysical and functional characterization of the human TAS1R2 sweet taste receptor overexpressed in a HEK293S inducible cell line

Sweet taste perception is mediated by a heterodimeric receptor formed by the assembly of the TAS1R2 and TAS1R3 subunits. TAS1R2 and TAS1R3 are class C G-protein-coupled receptors whose members share a common topology, including a large extracellular N-terminal domain (NTD) linked to a seven transmembrane domain (TMD) by a cysteine-rich domain. TAS1R2-NTD contains the primary binding site for sweet compounds, including natural sugars and high-potency sweeteners, whereas the TAS1R2-TMD has been shown to bind a limited number of sweet tasting compounds. To understand the molecular mechanisms governing receptor–ligand interactions, we overexpressed the human TAS1R2 (hTAS1R2) in a stable tetracycline-inducible HEK293S cell line and purified the detergent-solubilized receptor. Circular dichroism spectroscopic studies revealed that hTAS1R2 was properly folded with evidence of secondary structures. Using size exclusion chromatography coupled to light scattering, we found that the hTAS1R2 subunit is a dimer. Ligand binding properties were quantified by intrinsic tryptophan fluorescence. Due to technical limitations, natural sugars have not been tested. However, we showed that hTAS1R2 is capable of binding high potency sweeteners with Kd values that are in agreement with physiological detection. This study offers a new experimental strategy to identify new sweeteners or taste modulators that act on the hTAS1R2 and is a prerequisite for structural query and biophysical studies.


Results
Expression of hTAS1R2 in the HEK293S GnTI-inducible cell line. After transfection with the pcDNA5/TO-FLAG-hTAS1R2 plasmid and antibiotic selection, thirty-three HEK293S-GnTI-clones were subjected to induction in media supplemented with tetracycline or a combination of tetracycline and NaBu. To detect hTAS1R2 protein, we used the highly sensitive and specific anti-FLAG M2 tag antibody. Dot blot analysis of the cell lysates revealed that tetracycline alone does not allow an induction of the selected clones. In contrast, for two clones, the combination of tetracycline and NaBu allowed the induction of high amounts of FLAG-tagged hTAS1R2. The highest expression was observed with clone 7 (Fig. 1, Supplementary Fig. S1). Consequently, this clone was selected for subsequent experiments. Cell-based immunocytochemistry experiments confirmed expression of hTAS1R2 using a combination of tetracycline and NaBu (Fig. 2). The subcellular localization of the hTAS1R2 protein was investigated using confocal microscopy. This analysis revealed that most of the expressed protein is localized intracellularly with a minor portion present at the cell surface.
Detergent screening and purification of hTAS1R2. To purify the expressed hTAS1R2, we first investigated which detergent was able to solubilize hTAS1R2. Consequently, we performed a small detergent screen that included the main detergents that have been successfully used to solubilize several other GPCRs [40][41][42][43][44] . These detergents include zwitterionic fos-choline 14 (FC14) and three non-ionic detergents, dodecyl maltoside (DDM), octyl glucoside (OG) and lauryl maltose neopentyl glycol (LMNG). The hTAS1R2 solubilized with the different detergents was purified using FLAG immunoaffinity and analysed using gel filtration (Fig. 3). We found that the purified hTAS1R2 eluted at approximately 13.7 mL, corresponding to a molecular weight of 107 kDa, as calculated from the gel filtration calibration curve ( Supplementary Fig. S2). The chromatogram revealed an increase in absorbance between 13.5 and 15 mL for LMNG, DDM and, to a lesser extent, for FC14. Functional evaluation of these purified fractions using intrinsic fluorescence revealed that only hTAS1R2 extracted by LMNG and FC14 was able to bind sucralose, whereas hTAS1R2 extracted with DDM and OG led to a nonfunctional receptor. Since LMNG allowed us to obtain higher amount of functional receptor, this detergent was selected for all subsequent analyses.
The LMNG-solubilized hTAS1R2 receptor was purified using a two-step purification process. Coomassie blue SDS-PAGE analysis combined with western blot analysis of the protein purified by anti-FLAG M2 affinity chromatography showed four main bands migrating between 70 and 100 kDa and two other, less intense, migrating bands at 167 and 208 kDa (Fig. 4A, Supplementary Fig. S3A). Western blot analysis confirmed the presence of FLAG-tagged hTAS1R2 protein migrating around the expected molecular weight (115 kDa) (Fig. 4B, Supplementary Fig. S3B).
To further purify hTAS1R2 and remove the FLAG peptide used for protein elution, the immunoaffinitypurified FLAG-tagged hTAS1R2 protein was subjected to gel filtration analysis. Six peaks were observed (Fig. 5A). Peaks 1 and 6 corresponded to aggregates and the FLAG peptide, respectively. SDS-PAGE and western blot analysis using anti-FLAG M2 antibody showed that peaks 4 and 5 mainly contained monomeric hTAS1R2 that migrated at 100 kDa. However, three other bands migrating at 92, 76 and 71 kDa were detected by Coomassie blue staining. Peaks 2 and 3 showed an intense band migrating at approximately 75 kDa, as observed by Figure 1. Immunoblot analysis of HEK293S-GnTI-clones stably transfected with pcDNA5/TO-FLAG-hTAS1R2. Each clone was tested for induction with tetracycline alone (1 µg/mL) or in combination with 5 mM NaBu. Three microliters of solubilized protein from cell lysates (3 µg) were loaded onto PVDF membranes, analysed using the dot blot technique and probed with monoclonal FLAG antibody. The results were quantified using Image Lab (Bio-Rad) and normalized to 100% relative intensity. T: tetracycline induction; TB: tetracycline and NaBu induction. Dot blot image is presented in Supplementary Fig. S1.

Figure 2.
Immunocytochemistry of hTAS1R2-inducible HEK293S cells treated with tetracycline and NaBu. Cells from clone 7 were treated with 1 µg/mL tetracycline and 5 mM NaBu for 48 h. The level of induced hTAS1R2 protein (shown in green) was detected using a primary anti-FLAG M2 antibody and a fluorescently labelled secondary antibody (Alexa Fluor 488). The cell surface (shown in red) was detected by biotinconjugated concanavalin A and streptavidin-conjugated Alexa Fluor 568. The overlay images indicate the localization of the receptor at the cell surface (shown in yellow). (A) The cells were analysed using an epifluorescence inverted microscope (Eclipse TiE, Nikon, Champigny sur Marne, France) equipped with an ×20 objective lens and a LucaR EMCCD camera (Andor Technology, Belfast, UK). (B) The cells were analysed using a two photon confocal microscope (Nikon A1-MP) equipped with an ×60 objective lens (DImaCell platform, University of Burgundy, Dijon, France). www.nature.com/scientificreports/ SDS-PAGE but not by western blotting (Fig. 5B,C). The total yield of the hTAS1R2 protein from peak 4 (fractions 14-16) resulting from sixty T225 flasks was approximately 135 µg (2.2 µg/flask, i.e., 0.08-0.1 µg/10 6 cells in terms of cell productivity).
Secondary structure and oligomerization analysis of the purified hTAS1R2. Circular dichroism (CD) spectroscopy was used to confirm the structural integrity of hTAS1R2 purified by gel filtration (peak 4). The far-UV spectrum of purified hTAS1R2 displayed a positive peak centred at 193 nm and two negative peaks at 208 and 222 nm (Fig. 6), which clearly showed the helical character of the protein. Deconvolution of the CD Expression of hTAS1R2 was induced with tetracycline (1 µg/mL) and NaBu (5 mM www.nature.com/scientificreports/ www.nature.com/scientificreports/ spectrum revealed that hTAS1R2 was approximately 66% α-helix and 18% β-sheet. This composition is consistent with the secondary structure content of other crystallized class C GPCRs, such as mGluR (PDB 6N52) 45 and GABAb (PDB 4MS4) 46 for which the protein circular dichroism data bank (PCDDB) calculated around 37% α-helix for both proteins, and 12% and 21% β-sheet, respectively 47 .
To confirm the oligomerization state of FLAG-tagged hTAS1R2, fractions 14 to 16 corresponding to peak 4 were pooled and analysed using an online size exclusion chromatography (SEC) coupled to MALS, RI and UV detectors. SEC-MALS allows determination of the direct molecular mass of protein detergent complexes and does not require calibration curves 48 . Nevertheless, because the protein was obtained in LMNG detergent at a concentration in buffer above the critical micelle concentration, the quality of the chromatographic data was validated using two known molecular markers, bovine serum albumin (BSA) and β-amylase ( Supplementary  Fig. S4). Thus, the detergent complexes containing a BSA monomer or β-amylase were correctly resolved, with calculated molecular weights of 66 kDa and 200 kDa, respectively (theoretical molecular weights of 66 kDa and 200 kDa, respectively). The UV analysis at 280 nm-LS (90° angle)-RI overlay of the purified hTAS1R2 revealed the oligomeric states and the presence of the receptor in heterogeneous forms (Fig. 7). A predominant dimeric form with a measured mass of 204 kDa was detected after injection of concentrated fractions resulting from gel filtration. The SEC-MALS analysis of the linear and cumulative distribution of the molar mass confirmed that the dimers represented 80% of the total amount, while 20% was still present in monomeric form with an average molecular weight of 100 kDa. The theoretical mass of FLAG-tagged hTAS1R2 is 96.2 kDa; thus, hTAS1R2 appeared mainly as a dimer associated with detergents.  . Oligomeric state analysis of purified hTAS1R2. After size exclusion chromatography (SEC) and separation with a Superdex 200 3.2/300 column (GE Healthcare), the molecular mass was determined from the Raleigh ratio, measured by static light scattering and the refractive index. The calculated molecular mass (bold black curve), refraction index (blue curve), light scattering (red curve) and UV at 280 nm (green curve) are shown. A main oligomeric form with a measured mass of 204 kDa was detected, indicating the presence of a dimeric hTAS1R2 form associated with detergents. www.nature.com/scientificreports/ Ligand binding properties of the purified hTAS1R2. To characterize the interactions of purified hTAS1R2 with its ligands, we determined the dose-response relationship of its intrinsic tryptophan fluorescence upon titration with sweeteners previously demonstrated to bind the hTAS1R2 subunit 3,10-12,15,16,26,28 . We first tested the ability of neotame, sucralose, and acesulfame-K to bind to hTAS1R2. These compounds have been shown to bind hTAS1R2-VFT 10,12,15,16 . We found that the addition of neotame, sucralose, and acesulfame-K increased the fluorescence intensity of hTAS1R2. We observed that neotame was the highest affinity ligand, exhibiting a K d value of 2.78 ± 0.69 µM (Fig. 8B) whereas sucralose and acesulfame-K bound hTAS1R2 with lower affinities (K d values of 29 ± 8 µM and 164 ± 53 µM, respectively) in agreement with their lower potencies ( Fig. 8A,C,D). We also tested the sweetener perillartine shown to activate the monomeric hTAS1R2 receptor and bind its TMD [26][27][28] . We found that perillartine interacts with hTAS1R2 with lower affinity leading to a K d Figure 8. Binding activity of the purified hTAS1R2 using intrinsic tryptophan fluorescence. Intrinsic fluorescence was measured using a Cary Eclipse spectrofluorimeter. Dose-response relationship of hTAS1R2 fluorescence (λex = 280 nm, λem = 340 nm) was observed following sweetener addition. The data were fitted with a standard nonlinear regression method using SigmaPlot software. Data represent mean ± sem from at least four independent experiments. The reported K d values are the average of triplicate measurements from at least three independently purified protein samples. www.nature.com/scientificreports/ value of 373 ± 110 µM ( Fig. 8A,C,D). As a negative control, we tested the sweetener cyclamate, which has been shown to bind hTAS1R3-TMD 21,23 . As expected, cyclamate addition had no effect on the intrinsic fluorescence of hTAS1R2 (Fig. 8E). Altogether, these data demonstrated that purified hTAS1R2 protein is functional and able to specifically bind sweet tasting molecules with micromolar affinities.
To confirm the ligand binding data obtained with the purified hTAS1R2, we performed cellular assays to determine the functional activity of the heterodimeric sweet taste receptor 49,50 . First, the transient transfection rate was evaluated by immunostaining and showed that around 45% of cells expressed hTAS1R2 and hTAS1R3 proteins ( Supplementary Fig. S5). Then, HEK293T-Gα16gut44 cells were transiently co-transfected with hTAS1R2-FLAG, hTAS1R3-FLAG and pGP-CMV-GCaMP6S (fluorescent calcium indicator) and stimulated with sweeteners. The lowest EC 50 value (Fig. 9B) was measured for neotame (0.90 ± 0.09 µM), whereas cyclamate had the highest Figure 9. Human TAS1R2/TAS1R3 dose-response curves with different sweeteners. HEK293T-Gα16gust44 cells were transiently transfected with pGP-CMV-GCaMP6S (fluorescent calcium biosensor) combined with pcDNA6-hTAS1R2-FLAG and pcDNA4-hTAS1R3-FLAG (red line), or with pcDNA6-hTAS1R2-FLAG alone (green line) or pcDNA4-hTAS1R3-FLAG alone (blue line) or empty expression vector alone (mock cell) for the control (white circles, solid black line). Sweet taste stimuli were automatically applied to the transfected cells, and fluorescence changes were monitored using a FlexStation 3. The logarithmically scaled x-axis indicates the sweetener concentration in µM, and the y-axis shows the variation in fluorescence upon agonist application. Sucralose (A), neotame (B), acesulfame-K (C), perillartine (D), and cyclamate-Na (E). Data represent mean ± sem of eighteen wells from three independent experiments.  . 9A,C). These values are in accordance with those previously reported 12,26,51,52 and are in agreement with sweetener potencies (Table 1). We also tested the sweetener perillartine, which has been demonstrated to bind to hTAS1R2-TMD 26,28 . In addition to activate the hTAS1R2/hTAS1R3 heterodimer (2.54 ± 0.48 µM), perillartine is capable of activating hTAS1R2 in the absence of hTAS1R3 (61 ± 13 µM) 26,28 . We found that the perillartine-induced dose-response was strongly shifted towards a higher concentration range for the hTAS1R2 subunit expressed alone, with a slight increase in signal amplitude (by approximately 1.3-fold), compared to the responses of the heterodimeric sweet taste receptor (Fig. 9D). Our data confirm that hTAS1R2 alone, probably acting as a homodimer, is functional.

Discussion
In this study, the codon-optimized hTAS1R2 gene was overexpressed in the order of few hundreds of micrograms using the tetracycline-inducible HEK293S GnTI-cell line. Insertion of an N-terminal FLAG epitope tag allowed purification and detection of the recombinant membrane protein. We demonstrated the ability of the detergent LMNG to efficiently extract and solubilize hTAS1R2, maintaining its functional activity. LMNG is an emerging detergent that has been highlighted for its remarkable ability to enhance structural stability while protecting protein activity [53][54][55][56] . Recently, LMNG has allowed the successful crystallization of several delicate membrane proteins, such as the class A GPCR rhodopsin bound to arrestin 57 , and the class B GPCR calcitonin receptor coupled to its heterotrimeric Gs protein complex 58 . Associated with cholesterol hemisuccinate (CHS), LMNG has been used recently to determine the structure of the full-length mGluR5 by cryo-EM 45 . In our conditions, the addition of CHS during the extraction and solubilization step increased the amount of extracted protein but unfortunately led to a loss of functionality. SEC-MALS was used to determine the molecular mass of the purified FLAG-tagged hTAS1R2 protein. This analysis revealed that recombinant hTAS1R2 is mainly present in its dimeric form. Previous studies on mGluR2 demonstrated that class C GPCR dimerization is required to induce agonist activation and G-protein coupling 17,59 . It has been shown that hTAS1R3 surface expression requires hTAS1R2 co-expression in a specific membrane trafficking system different from that of mice 60 . More recently, the structural architecture of the heterodimeric sweet taste receptor was explored and it was revealed that the TAS1Rs C-terminus of the CRD needs to be properly folded for TAS1R3 dimerization and co-trafficking, but not for surface expression of TAS1R2 61 . In this study, the cell surface expression of most FLAG-tagged TAS1R2 mutant libraries was very low, with most protein remaining inside the cell, in accordance with the low expression level we observed. The authors demonstrated that inhibition of surface expression of TAS1R2 is associated with an altered sequence at the C-terminus in the transmembrane domain or cytosolic tail of TAS1R2. The authors highlighted conserved surfaces on the ECD and TMD for dimerization with TAS1R3 61 . However, it is unclear whether the TAS1R2 homodimer structure is physiological or could represent an alternative conformational state, such as the mGluR2 homodimer or heterodimer (mGluR2-mGluR4) 62 . Nevertheless, a cellular assay showed that TAS1R2 transiently expressed alone in absence of TAS1R3 subunit is able to be activated by perillartine, a sweet tasting compound that interacts with TAS1R2-TMD 26,28,63 .
High concentrations of sugars (i.e. fructose, glucose, sucrose) may modify the buffer polarity affecting tryptophan environment leading to unspecific fluorescent changes. For this reason, we characterized the functional activity of the purified hTAS1R2 by measuring its binding affinity with high potency sweeteners that have been previously shown to activate the heterodimeric sweet taste receptors. For this purpose, we monitored the changes in the intrinsic tryptophan fluorescence of hTAS1R2 as it contains 15 tryptophan residues, 13 of which are localized in the NTD. Except for cyclamate, which is known to bind to TAS1R3-TMD, addition of all of the tested sweeteners led to an increase in fluorescence of the full-length hTAS1R2, which was saturable. Using this technique, we successfully measured K d values, which were in the micromolar range. The measured K d values for sucralose and acesulfame-K are also in accordance with recently published K d values measured by the intrinsic fluorescence of hTAS1R2-NTD (40 µM and 100 µM, respectively) 30 . Interestingly, our data revealed a K d value of hTAS1R2 for neotame (2.78 µM), is 18-fold lower compared to the K d value measured with hTAS1R2-NTD (50 µM). We can speculate that the presence of the TMD may increase the affinity of hTAS1R2 for this sweetener. Table 1. Biochemical, biological and physiological features of sweet tasting compounds. K d values were determined by intrinsic tryptophan fluorescence. EC 50 values were calculated from the ligand dose-response relationship in HEK293T-Gα16gust44 cells transiently transfected with plasmid encoding pGP-CMV-GCaMP6S (fluorescent calcium biosensor), hTAS1R2-FLAG and/or hTAS1R3-FLAG. The relative sweetness of each sweetener is described on a molar basis. n.r. = no response; n.b. no binding. www.nature.com/scientificreports/ Unfortunately, the weak expression of hTAS1R2 on the cell surface does not allow us to obtain sufficient functional activity to determine EC 50 values using HEK293S-GnTI-cells co-transfected with the plasmid coding for Gα16gust44 and hTAS1R3. To overcome this difficulty, we used HEK293T cells stably transfected with Gα16gust44 and transiently transfected with a plasmid coding for hTAS1R2-FLAG and hTAS1R3-FLAG. We measured a strong expression of each subunit by immunocytochemistry (Supplementary Fig. S5). For functional assay, cells were also transiently transfected with the plasmid pGP-CMV-GCaMP6S to allow production of ultra-sensitive protein calcium sensor 64,65 . Calcium imaging assays led to determination of EC 50 values in accordance with previously published data 3,12,15,16,26,28,51 . It is interesting to note that even if sucralose, neotame and acesulfame-K were able to bind the VFT of hTAS1R2 subunit 12,15,66 they were unable to produce biological and functional response of the receptor in absence of hTAS1R3. Sucralose, which interacts with the VFT of hTAS1R3 subunit 15 is also unable to produce cellular response by itself in absence of hTAS1R2 subunit. This is the same for cyclamate which binds hTAS1R3-TMD and is unable to stimulate transfected cells expressing hTAS1R3 alone. As previously reported, perillartine, which binds hTAS1R2-TMD is able to stimulate cells expressing hTAS1R2 alone and is more effective when both sweet taste receptor subunits are expressed. This difference in response between binding at the receptor level and functional response of the sweet taste receptor at the cellular level could be explained by the mechanism of inter-subunit or intra-subunit rearrangement and the rigidity of the CRD, that lead to conformational changes after ligand binding and finally interaction with G proteins. These arguments are supported by many recent structural studies including one on the Medaka fish taste receptor T1r2-T1r3 14,32 and other class C GPCR, like CaSR, mGlu and GABA B receptors 45,62,[67][68][69][70] , which demonstrated that the reorientation of the VFT domain could lead to intra-subunit movement between VTF domain and TMD revealing multiple allosteric interactions and cooperativity between VFT domain, CRD and TMD. This rearrangement could explain why intrinsic tryptophan fluorescence can be measured for perillartine binding in TMD even if only two tryptophan were present in this part of the receptor. On the other hand, it could also suggest that binding sites for perillartine involved TMD1 in addition to TMD3, TMD5 and TMD7 that have been demonstrated with the hTAS1R2-TMD-inhibitor amiloride 28 . At the moment, this is not clear which TMD between TAS1R2 and TAS1R3 could be responsible for coupling to G protein activation. Studies performed on the mGlu2-4 heterodimers show that mGlu4-TMD lead to protein G activation even if mGlu2 homodimer can also do it 62 . Our results on hTAS1R2 homodimer with perillartine suggests that G protein could be activate preferentially by hTAS1R2 subunit, but it has been shown that hTAS1R3 homodimer could also be activated by calcium 71 . As suggested before 72 , we propose that perillartine induces conformation changes in the hTAS1R2-TMD, which in turns leads to inter-subunit rearrangement between the two TMD sufficient to activate hTAS1R2 homodimer and hTAS1R2/hTAS1R3 heterodimers. Inversely, for sucralose, neotame, and acesulfame-K the rearrangement of the VFT domain, or that of cyclamate on the hTAS1R3-TMD, are not sufficient to activate hTAS1R2 or hTAS1R3 homodimers because of difference in energy barrier or because hTAS1R2 play a key role in the activation process. Strikingly, the K d values determined for the purified hTAS1R2 receptor were in agreement with the EC 50 values measured using a cellular assay. Furthermore, these data are in agreement with the relative potencies of sweet tasting compounds (Table 1). Interestingly, we reported for the first time a K d value of 373 µM for perillartine, which is able to bind to the TMD of hTAS1R2 and is in accordance with the 61 µM evaluated using a cellular assay on HEK293T-Gα16gust44 transiently transfected with hTAS1R2 alone.
In conclusion, despite its low expression level and weak targeting to the cell surface, we successfully purified functional full-length hTAS1R2 receptor, allowing the performance of biophysical studies and measurement of its affinity for sweet tasting compounds or sweet tasting modulators. The main advantage of the stable expression is the reduction of the transfection costs, including plasmid preparation and transfection reagent, which can be limiting for protein production on a large scale. This approach paves the way also to generate nanobodies for the subsequent analysis of the functional and structural properties of hTAS1R2.

Materials and methods
The method used for construction of the tetracycline inducible HEK293S stable cell line expressing hTAS1R2, and the following steps of extraction, purification and biophysical characterization were carried out as previously described for the human olfactory receptor hOR1A1 36  www.nature.com/scientificreports/ (GCC ACC ATGG) immediately before the start codon and by the addition of the FLAG epitope tag (DYK-DDDDK) to the N-terminus of the hTAS1R2 gene after the starting codon. The synthetic cDNA encoding hTAS1R2 was subcloned into the NdeI and EcoRI restriction sites of the pcDNA5/TO-inducible expression vector (Invitrogen). The resulting expression vector pcDNA5/TO-hTAS1R2 encodes a fusion protein comprising an N-terminal FLAG-tag followed by hTAS1R2. The plasmid was amplified in E. coli DH5α cells and purified with the PureYield Plasmid Midiprep System (Promega, Charbonnières-les-Bains, France). The pcDNA5/TO-Flag-hTAS1R2 plasmid was transfected into the human inducible N-acetylglucosaminyltransferase I-negative HEK293S cell line (HEK293S GnTI − ) using Fugene HD (Promega, Madison, Wisconsin, USA) 39 . The HEK293S GnTI − cells were grown in DMEM/F12 supplemented with 10% foetal bovine serum, nonessential amino acids (0.1 mM), 2 mM GlutaMAX (Gibco, Life Technologies), HEPES (15 mM), penicillin (100 units/mL), streptomycin (100 µg/mL) and blasticidin (5 µg/mL) at 37 °C in a humidified atmosphere containing 6.7% CO 2 . The expression and selection of the clones were carried out as previously described 36 using 125 µg/mL hygromycin. Clones were expanded and screened for the inducible expression of FLAG-tagged hTAS1R2 using media supplemented with 1 µg/mL tetracycline and 5 mM NaBu for 48 h. The expression level of FLAG-hTAS1R2 was estimated by dot blot using mouse anti-FLAG primary antibody (dilution 1/2000). The clone showing the highest level of hTAS1R2 expression under the induction conditions while maintaining undetectable expression without induction was selected and expanded into large-scale culture for use in all subsequent experiments.
Cell extract preparation. The cell extraction was performed as previously described 36 . Briefly, the hTAS1R2-HEK293S GnTI − cells were grown in T225 flasks for 5 days at 37 °C until they reached 80% confluence. The cells were then incubated in medium containing tetracycline (1 µg/mL) and NaBu (5 mM). After 48 h, the adherent cells and cells in suspension were harvested into ice-cold medium, pelleted by centrifugation at 800g for 15 min at 4 °C and washed with PBS containing a protease inhibitor cocktail (Sigma-Aldrich). Cells pellets from 20 flasks were then pooled and centrifuged again. The pooled pellet was flash frozen in liquid nitrogen and stored at − 80 °C until purification.
On the day of purification, the cell pellet was thawed on wet ice. The FLAG-tagged hTAS1R2 protein was solubilized by resuspending the cell pellet in ice-cold PBS buffer containing 2% w/v LMNG and a protease inhibitor cocktail (2 mL per T175 flask). The cell homogenate was sonicated for 1 min using a Vibracell 750 sonicator (Sonics & Materials, Newtown, USA) equipped with a 2 mm tip and set to 30% maximum power. The homogenate was further disrupted by high-speed shaking with a tissue lyser (TissueLyser, Qiagen, Hilden, Germany) for 3 min after carbon beads were introduced into each microtube. Finally, the homogenate was incubated for 2 h at 4 °C under strong agitation using a Vortex Genie II mixer (Scientific Industries, Bohemia, USA) and then centrifuged at 105,000g for 1 h at 4 °C. The resulting supernatant was immediately submitted to immunoaffinity purification. hTAS1R2 purification by immunoaffinity chromatography. To purify the FLAG-tagged hTAS1R2 protein from the cell extract, we followed the protocol already described 36 using the EZview Red anti-FLAG ® M2 affinity gel, in which the monoclonal antibody ANTI-FLAG M2 is covalently attached to cross-linked agarose beads. The cell homogenate was mixed with the EZview Red anti-FLAG M2 beads (binding capacity: 0.6 mg/ mL) and rotated for 2 h at 4 °C to capture the FLAG-tagged hTAS1R2. The beads were then transferred into Pierce spin columns and collected by centrifugation at 1500g for 1 min. Then, the beads were washed 5 times with cold PBS containing 0.1% LMNG. After the last wash, the FLAG-tagged hTAS1R2 bound to the beads was eluted by competitive elution with 5 column volumes of a PBS-0.1% LMNG solution containing 100 mg/mL FLAG peptide. The eluate was loaded on SDS-PAGE, stained by Coomassie blue and analysed by western blot.
Size exclusion chromatography (SEC). The FLAG-tagged hTAS1R2 samples that had been eluted from the ANTI-FLAG M2 beads were pooled and concentrated to 0.3-0.5 mg/mL using a 30-kDa MWCO filter column (Vivaspin, Sartorius, Aubagne, France). Then the concentrated FLAG-tagged hTAS1R2 was purified by gel filtration as previously reported 36 . The samples were then loaded for gel filtration chromatography (Superdex 200 Increase 10/300GL column) on an Äkta Pure fast protein liquid chromatography system (GE Healthcare, Velizy-Villacoublay, France). The column was equilibrated with 2 column volumes of wash buffer (PBS, 0.1% LMNG, pH 7.3) before the immunopurified FLAG-tagged hTAS1R2 sample was applied. After loading, the column was rinsed with wash buffer at 0.5 mL/min, and the column flow through was monitored by UV absorbance at 280 nm. The molecular masses of the FLAG-tagged hTAS1R2-detergent complexes were estimated by calibrating the column with a gel filtration standard mixture (Sigma-Aldrich). The following standard proteins were used: thyroglobulin (669 kDa), β-amylase (200 kDa), alcohol dehydrogenase (150 kDa), monomeric BSA (66 kDa), carbonic anhydrase (29 kDa), myoglobin (17 kDa) and lysozyme (14.3 kDa). The protein fractions (0.5 mL) were collected using an automated fraction collector. The collected fractions were deposited on SDS-PAGE, stained by Coomassie blue and subjected to immunoblotting analysis.

SEC coupled with multi-angle light scattering (SEC-MALS) analysis.
The oligomeric state of the purified FLAG-tagged hTAS1R2 protein was determined using an SEC column coupled to a MALS detector. SEC was performed using a Superdex 200 Increase 3.2/300 column (GE Healthcare) and eluted with PBS containing 0.1% LMNG (pH 7.3) at 0.1 mL/min. The protein was detected with a triple-angle light scattering detector (Mini-DAWN TREOS, Wyatt Technology) connected to a UV detector (UV 100 SpectraSeries, Thermo Separation Products, Waltham, MA, USA) operating at a wavelength of 280 nm and a differential refractometer (RiD-10A, Shimadzu, Kyoto, Japan). A 100 µL volume of each sample was injected onto the column. The molecular weights of the protein detergent complexes were determined with ASTRA VI software (Wyatt Technology Santa Barbara, CA, USA). A specific refractive index increment (dn/dc value), which was estimated at 0.185 mL/g, www.nature.com/scientificreports/ was used for mass calculation. The suitability of the system was assessed by analysing the standard proteins BSA (66 kDa) and β-amylase (200 kDa).
hTAS1R2 secondary structure analysis using circular dichroism spectroscopy. The circular dichroism (CD) spectra of the FLAG-tagged hTAS1R2 samples were recorded at 20 °C using a J-815 Jasco spectropolarimeter (Jasco, Tokyo, Japan) equipped with a Peltier temperature control. The CD spectra were corrected for the buffer and ligand contributions and converted to mean residue ellipticity in deg cm 2 dmol −1 . The spectra recorded between 178 and 260 nm were averaged over 5 scans accumulated at 0.5 nm intervals with a 50 nm/ min scan speed and 5 s of response time. Spectra were smoothed using the Savitzky-Golay convolution filter with a span of 15. The secondary structure proportions were estimated using the deconvolution CDSSTR algorithm available on the DichroWeb website (http:// dichr oweb. cryst. bbk. ac. uk/ html/ home. shtml) 73 .
Intrinsic tryptophan fluorescence measurements. Intrinsic   Immunocytochemistry. For immunocytochemical staining analyses, we performed the protocol as reported earlier 36 . The stable hTAS1R2-HEK293S GnTI − clones were seeded on 4-well culture slides (Corning Inc., Corning, NY, USA) precoated with BD Cell-Tak adhesive (Corning). When the cells reached ~ 90% confluence, they were treated with 1 µg/mL tetracycline and 5 mM NaBu for 48 h. Then, the cells were rinsed twice with Hank's HEPES-balanced salt solution and permeabilized for 5 min in cold acetone-methanol (1:1).
To visualize the FLAG-tagged hTAS1R2 expression level, the cells were blocked with 5% goat serum in PBS for 30 min at 25 °C and incubated for 1 h at 25 °C with a 1/500 dilution of the mouse anti-FLAG M2 primary antibody in an antibody diluent (Dako Les Ulis, France). The cells were then rinsed twice with PBS for 5 min and incubated with a 1/400 dilution of the Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody (Life Technologies) in the Dako antibody diluent for 1 h at 25 °C to visualize FLAG-tagged hTAS1R2. The cells were analysed using an epi-fluorescence inverted microscope (Eclipse TiE, Nikon, Champigny sur Marne, France) equipped with an 20× objective lens and a LucaR EMCCD camera (Andor Technology, Belfast, UK). To study the subcellular localization of FLAG-tagged hTAS1R2, the receptor was analysed as previously described.
In addition, to visualize the plasma membrane, the cells were cooled on ice for 30 min and then incubated with 2 mg/mL biotin-concanavalin A for 1 h on ice before permeabilization. The plasma membrane was then labelled with Alexa Fluor 568 streptavidin conjugate (dilution 1/500; Life Technologies). After washing the cells with PBS, the chambers were detached from the slide and the coverslips were placed with mounting medium (Dako). The cells were analysed using a two-photon confocal microscope (Nikon A1-MP) equipped with an 60× objective lens (DImaCell platform, University of Burgundy, Dijon, France).
Heterologous expression and calcium assay. We used a calcium imaging assay to establish doseresponse curves for the sweet taste receptor hTAS1R2/hTAS1R3 and determine EC 50 values for the sweet stimuli previously tested in spectrofluorimetric experiments. We cloned the cDNAs coding hTAS1R2 and hTAS1R3 subunits into pcDNA6 and pcDNA4 expression plasmids, respectively, and we added FLAG tag to C-terminus of each construct to measure protein expression level by immunocytochemistry as described previously. HEK293T cells stably transfected with Gα16gust44 were seeded at a density of 0. 35  www.nature.com/scientificreports/ walled, clear bottom microtiter plates (Falcon) in high-glucose DMEM supplemented with 2 mM GlutaMAX, 10% dialyzed foetal bovine serum, penicillin/streptomycin and G418 (400 µg/mL) at 37 °C and 6.3% CO 2 , in a humidified atmosphere. After 48 h, the cells were transiently transfected with each TAS1R subunit (60 ng/ well for each plasmid) and plasmid pGP-CMV-GCaMP6S (Addgene #40753, 50 ng per well) coding for a green fluorescent protein-based calcium indicator, using Fugene HD (0.4 µL/well). After a further 24 h incubation, the cells were washed twice with buffer C1 and then stimulated with sweet tasting compounds. After excitation at 488 nm, calcium responses were recorded at 510 nm using a Molecular Devices FlexStation 3 system. Acquisition was made simultaneously from 8 wells corresponding to the range of taste solutions. We averaged the responses of 18 wells receiving the same stimulus one three independent experiments. We subtracted the mean calcium traces of mock-transfected cells stimulated with the same concentration of stimulus. Plots of the fluorescence variation amplitude versus concentration were fitted by four-parameter logistic nonlinear regression allowing us to calculate the EC 50 values of activation.