Regulator of G-protein Signaling-21 (RGS21) Is an Inhibitor of Bitter Gustatory Signaling Found in Lingual and Airway Epithelia*

Background: RGS21 is expressed in tastant-responsive lingual epithelium, but with unknown function. Results: RGS21 accelerated intrinsic GTPase activity of multiple Gα subunits; RGS21 over- and underexpression in epithelial cells modulated bitterant responsiveness. Conclusion: RGS21 is a negative regulator of bitterant signal transduction. Significance: RGS21 represents a nonreceptor regulatory component of gustatory signaling that alters sensitivity of bitterant responsiveness in an endogenous, cellular context. The gustatory system detects tastants and transmits signals to the brain regarding ingested substances and nutrients. Although tastant receptors and taste signaling pathways have been identified, little is known about their regulation. Because bitter, sweet, and umami taste receptors are G protein-coupled receptors (GPCRs), we hypothesized that regulators of G protein signaling (RGS) proteins may be involved. The recent cloning of RGS21 from taste bud cells has implicated this protein in the regulation of taste signaling; however, the exact role of RGS21 has not been precisely defined. Here, we sought to determine the role of RGS21 in tastant responsiveness. Biochemical analyses confirmed in silico predictions that RGS21 acts as a GTPase-accelerating protein (GAP) for multiple G protein α subunits, including adenylyl cyclase-inhibitory (Gαi) subunits and those thought to be involved in tastant signal transduction. Using a combination of in situ hybridization, RT-PCR, immunohistochemistry, and immunofluorescence, we demonstrate that RGS21 is not only endogenously expressed in mouse taste buds but also in lung airway epithelial cells, which have previously been shown to express components of the taste signaling cascade. Furthermore, as shown by reverse transcription-PCR, the immortalized human airway cell line 16HBE was found to express transcripts for tastant receptors, RGS21, and downstream taste signaling components. Over- and underexpression of RGS21 in 16HBE cells confirmed that RGS21 acts to oppose bitter tastant signaling to cAMP and calcium second messenger changes. Our data collectively suggests that RGS21 modulates bitter taste signal transduction.

hydrolysis by G␣ subunits and thereby determine the duration of GPCR signaling (15). For example, in the mammalian visual system controlled by the GPCR rhodopsin, the time constant at which rod photoreceptor cells recover from flash response increases from 0.2 s (wild-type) to 9 s in RGS9-deficient mice (16), clearly highlighting the importance of RGS proteins to proper signal transduction timing (17).
The T1R and T2R families of GPCRs compose the mammalian taste receptors that detect the taste modalities of umami, sweet, and bitter (18). The T2R family is responsible for transduction of bitter taste (19,20). In contrast to the T2R receptors, which are functional when expressed individually, T1Rs only function when expressed as heterodimers (21,22). L-Amino acids (umami compounds) signal through the T1R1-T1R3 heterodimer (22). Similarly, sweet compounds bind to the T1R2-T1R3 heterodimer to initiate taste signaling (21).
Perception of bitterness by the gustatory system helps to identify toxic or spoiled foods that should not be ingested; however, many nonharmful compounds also are perceived as bitter, including some artificial sweeteners and many medicines (23)(24)(25)(26)(27)(28)(29)(30). Tastant receptors have also been identified outside the gustatory system, including bitter receptors in ciliated epithelial airway and enteroendocrine cells (31)(32)(33)(34), but their function in these tissues is still being determined. Understanding the physiological regulation of these signaling pathways should lead to the development of more palatable therapeutics and artificial sweeteners, as well as better glucose modulators.
The critical components of taste transduction downstream of the T1R and T2R receptors are slowly being elucidated. Similar to the visual system, which has two unique G␣ subunits specific for rods and cones, lingual taste cells also have a unique G␣ called gustducin-␣ (encoded by GNAT3), which is highly related by sequence to rod transducin (G␣ t-rod , encoded by GNAT1) (35). Gustducin has a critical role in the transduction of bitter and sweet taste (36); however, the appetitive effects of umami compounds are only lost when animals are deficient in both gustducin-␣ and G␣ t-rod (37), suggesting some degree of functional redundancy. Although gustducin and transducin are widely accepted to be the critical G␣ subunits in mediating taste in lingual tissue (36,37), the role of other G␣ subunits remains unknown. The ability of gustducin-␣/transducin double knockout mice to still produce recordable action potentials on chorda tympani recordings upon exposure to bitter tastants (37), and the high levels of G␣ i2 expressed in lingual taste cells (38), suggests that G␣ i2 may also have a role in taste transduction (37,39).
Although knock-out mice have shown the importance of gustducin and transducin to gustation signal transduction, downstream second messenger systems have also been implicated. Tastant-dependent heterotrimer activation is seen to rapidly change cyclic nucleotide levels (including cAMP) in a phosphodiesterase-dependent manner, but a physiological role of these changes remains unsupported by direct evidence (40). Tastant receptor activation also results in a PLC-␤2-mediated increase in intracellular calcium and subsequent gating of the taste-transduction channel TRPM5. Both PLC-␤2-null and TRPM5-null mouse strains have reduced or a complete loss of bitter, sweet, and umami signaling (41,42).
Although multiple intracellular modulators are known to regulate G protein activity and GPCR-initiated signaling (43,44), modulators specific to tastant signal transduction are only now being identified (e.g. Refs. 45 and 46). Two independent groups identified a novel member of the RGS protein family, RGS21, that may be involved in tastant signal transduction based on its expression in lingual taste-responsive epithelium (47,48). Using in situ hybridization and RT-PCR, RGS21 expression was identified only in a subset of taste bud cells in one study (47), yet found in every tissue screened in the other (48). Neither report established the specific function of RGS21 in tastant signal transduction (47,48). Thus, our present study aimed to clarify the discordant reports of RGS21 expression and define its role in tastant signaling, thereby furthering our collective understanding of the physiological modulation of taste perception.

Chemicals and Assay Materials
Unless otherwise noted, all chemicals were the highest grade available from Sigma or Fisher Scientific (Pittsburgh, PA).

Reverse Transcription-Polymerase Chain Reaction (RT-PCR)
RNA Isolation-Cells from lung epithelial cell lines were stored in RNase-free PBS at Ϫ20°C for up to 16 h before use. RNA isolations were performed in an RNA purification tray, using the ABI Prism 6700 automated nucleic acid work station (PE Biosystems) and following the manufacturer's protocol. Total RNA samples were stored at Ϫ20°C.
Primers and Probes-The nucleotide sequences of the PCR primers and their fluorogenic probes for the target genes were designed by using the computer program Primer Express (PE Biosystems). Each fluorescent probe had a reporter dye (FAM for the target RNA and TET for the 18 S RNA control) covalently attached at its 5Ј end and a quencher dye (TAMRA) attached at its 3Ј end. Before use, the probes were purified by a PolyPak II cartridge (Glen Research, Sterling, VA) following the manufacturer's instructions.
RT-PCR Amplifications-Real-time RT-PCR amplifications were performed in a 96-well plate in an ABI Prism 7700 sequence detector (PE Biosystems) in a total volume of 30 l, which included 10 l of RNA sample from the ABI Prism 6700 plus 20 l of a reaction mixture made with minor modifications of the manufacturer's instructions. Each RT-PCR amplification was performed in triplicate: 30 min at 48°C for the RT reaction, then 10 min at 94°C, followed by a total of 40 temperature cycles (15 s at 94°C and 1 min at 60°C). During the amplification, the fluorescence of FAM (or TET), TAMRA, and ROX (a passive reference dye) was measured by the 7700 Sequence Detector in each well of the 96-well plate. The cycle numbers at which the fluorescence of the PCR starting samples crossed a set threshold (C t ) were calculated by using the sequence detector software incorporated in the ABI Prism 7700 Sequence Detector System.
Relative Quantification of Gene Expression-Amplification of the 18S RNA gene was used as an endogenous control. For each experimental sample, the relative amounts of the target and endogenous control were determined from the C t . We assume that 18S RNA is present in all tested samples of cell RNA at a constant proportion and therefore normalized the amount of total RNA in our test samples by comparing the C t of their 18S RNA with that of our test genes. For each gene, the data were then expressed as a percentage of the cell line with the highest normalized C t (49).

In Situ Hybridization
cDNA spans from the mouse Rgs21 open reading frame (fulllength), PLC␤2 (nucleotides 227-740), and gustducin-␣ (nucleotides 83-610) were separately cloned into pBluescript II SK (Stratagene) using the KpnI and NotI digestion sites. Digoxigenin-labeled riboprobes were generated using T7 RNA polymerase and a digoxigenin-RNA labeling mix (Roche Applied Science). Hybridization was performed by the University of North Carolina (UNC) Neuroscience Center In Situ Hybridization Core according to standard protocols (UNC-Chapel Hill).

Generation of Rgs21::RFP BAC Transgenic Mice
A 232-kilobase bacterial artificial chromosome (BAC) from mouse chromosome 1 (RP23-126D12: nucleotides 146,254,848 to 146,486,740) that encodes TagRFP (50) driven by the Rgs21 promoter was engineered by BAC recombineering and purified by pulsed field electrophoresis by the UNC Neuroscience Center BAC Engineering Core Facility. Pronuclear injections were performed by the UNC Animal Models Core on C57Bl/ 6XDBA2 hybrid embryos, which were subsequently implanted into pseudo-pregnant recipient females. Resultant pups were genotyped via tail biopsies at 21 days to confirm gene expression, and Rgs21::RFP BAC-transgenic founder mice were established. Rgs21::RFP BAC transgenic mice were bred with wildtype C57Bl/6J mice and the resulting F1 pups were utilized in this study. All animal protocols received prior approval by the Institutional Animal Care and Use Committee (IACUC) of UNC-Chapel Hill and all animals were cared for in an AAALAC-accredited vivarium according to NIH standards.

Immunofluorescence
Mouse tongues were removed post-humane euthanasia and fixed in 4% paraformaldehyde at 4°C overnight, followed by an overnight incubation in 30% sucrose at 4°C. Circumvallate papillae were then dissected from these fixed tongues, embedded, and frozen in Tissue-Tek OCT compound (Fisher Scientific), and coronally sectioned (12 m thick) at Ϫ22°C. Circumvallate papillae sections were mounted on FisherBrand SuperfrostPlus slides and permeabilized with 0.5% Triton X-100 in phosphate-buffered saline (PBS) for 1 h at room temperature. These slides were incubated in blocking buffer (0.5% Triton X-100 and 1% bovine serum albumin in PBS) for an additional hour at room temperature and then probed overnight at 4°C with rabbit anti-gustducin-␣ polyclonal primary antibody (sc-395; Santa Cruz) diluted in blocking buffer. After 4 washes in PBS, slides were incubated again for 30 min in blocking buffer and then probed for 1 h at room temperature with Alexa Fluor-488 goat anti-rabbit IgG (heavy and light chains) secondary antibody (Invitrogen) diluted in blocking buffer. Slides were then washed 4 times in PBS. Vectashield mounting medium with DAPI (Vector Labs) was applied to mount glass coverslips onto the slides. Immunofluorescence was detected with an IX70 epifluorescence microscope (Olympus) and mounted digital camera (Hamamatsu). Images were digitally processed and pseudocolored using Adobe Illustrator CS4 software.

Cloning
The RGS21 open reading frame (ORF) was cloned by PCR from cDNA isolated from rat taste bud cell (a gift from Dr. Lars von Buchholtz, NIH/NIDCR) using the sense primer, 5Ј-CCAGTGAAATGCTGTTTCTAC-3Ј and antisense primer, 5Ј-CAGGAAAGGCAGCCATC-3Ј with an annealing temperature of 52°C and an extension time of 22 s using Phusion thermostable DNA polymerase (New England Biolabs, Ipswich, MA). Following amplification, a 453-bp band was resolved using agarose electrophoresis and isolated by Qiagen Gel Extraction (Qiagen). A second round of PCR, using primers to extend the RGS21 ORF sequence, was used to subclone the isolated fragment into a pET-based (Novagen) ligation-independent cloning vector to make a tobacco etch virus protease-cleavable His 6 fusion protein for expression in Escherichia coli, as well as into a pcDNA3.1-based (Invitrogen) ligation-independent cloning vector to make a hemagglutinin (HA)-epitope fusion protein for expression in mammalian cells, both as previously described (51,52). QuikChange site-directed mutagenesis (Stratagene) was used to mutate Arg-126 of the RGS21 ORF to a glutamic acid codon (R126E) with sense primer, 5Ј-gtctcatggccaaggattccttccctgagtttctaaagtcagaaatttataagaaa-3Ј and antisense primer, 5Ј-tttcttataaatttctgactttagaaactcagggaaggaatccttggccatgagac-3Ј.

Protein Expression and Purification
The pET-based vector encoding the His 6 -rRGS21(wild type) or -RGS21(R126E) ORF was transformed into E. coli BL21(DE3) cells (Novagen) and grown at 37°C in Luria broth until a culture density of A 600 ϭ 0.75 was reached. Protein expression was then induced by addition of 0.75 mM isopropyl ␤-D-thiogalactopyranoside. After culture for 14 -16 h at 20°C, cells were pelleted by centrifugation and frozen at Ϫ80°C. Prior to purification, bacterial cell pellets were resuspended in N1 buffer (50 mM Tris, pH 7.5, 300 mM NaCl, 30 mM imidazole, 5% (w/v) glycerol) for lysis using high-pressure homogenization with an Emulsiflex (Avestin, Ottawa, Canada). Cellular lysates were clarified by centrifugation for 45 min at 100,000 ϫ g at 4°C before the supernatant was applied to a nickel-nitrilotriacetic acid resin FPLC column (FF HisTrap; GE Healthcare). The column was washed with 7 column volumes of N1 lacking imidazole, then 3 column volumes of N1 containing 30 mM imidazole, 5 column volumes of N1 containing 300 mM imidazole, and 5 column volumes of N1 containing 700 mM imidazole. Soluble RGS21 protein was found in the 300 mM imidazole-containing fractions. The column flow-through was pooled and resolved using a calibrated 150-ml size exclusion column (Sephacryl S200, GE Healthcare) with S200 buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5% (w/v) glycerol). Protein was concentrated to ϳ1 mM as determined by A 280 nm measurements upon denaturation in guanidine hydrochloride. Circular dichroism spectra were separately collected using 0.1 mg/ml of RGS21 and RGS21(R126E) proteins, each diluted in 10 mM potassium phosphate salt, pH 7.5, at 25°C using a PiStar-180 spectrophotometer (Applied Photophysics, UK) within the University of North Carolina Macromolecular Interactions Facility (Dr. Ash Tripathy, director). Spectra were collected from 185 to 260 nm in 0.5-nm steps with a bandwidth of 2 nm. His 6 -G␣ oA was purified using a similar chromatographic methods as previously described (53). His 6 -G␣ i1,i2,i3 and His 6 -G␣ i1 (G183S) subunits were purified exactly as previously described for the production of G␣ i1 (54). The His 6 -G␣ t/i1 chimeric G␣ subunit ("Chi6") was purified as previously described (55) and G␣-transducin was purified from bovine rod outer segments as previously described (56).

Cell Culture and Co-immunoprecipitation
COS-7 cells and the human bronchial airway epithelial cell line 16HBE were separately cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Denville Scientific, Metuchen, NJ) and penicillin-streptomycin (Mediatech, Inc., Manassas, VA) and maintained at 37°C in a humidified atmosphere containing 5% CO 2 . COS7 cells were plated in 6-well dishes and transfected with 1.5 g of DNA (when at 70% confluence) with Lipofectamine TM 2000 (Invitrogen) according to the manufacturer's instructions. Cells were lysed in lysis buffer containing 20 mM Tris (pH 7.5), 100 mM NaCl, 100 M GDP, 5 mM MgCl 2 , 1 mM EGTA, and 1% Triton X-100. To mimic the GTP hydrolysis transition state of the heterotrimeric G-protein ␣ subunit, lysis buffer was supplemented with 20 mM NaF and 30 M AlCl 3 . Cell monolayers were scraped in lysis buffer and incubated at 4°C for 45 min with sonication before insoluble components were separated by centrifugation at 14,000 ϫ g at 4°C for 10 min. Recombinant purified His 6 -RGS21 protein (10 g) was then added to the clarified cell lysate and rocked at 4°C for 1 h before addition of NTA-agarose and continued incubation overnight at 4°C with rocking. NTA-agarose was then washed four times with lysis buffer and bound proteins were resolved using SDS-PAGE electrophoresis, transferred to nitrocellulose, and detected by chemiluminescence using standard immunoblotting techniques.

Surface Plasmon Resonance (SPR) Assay
Optical detection of protein-protein interactions using SPR was performed using the Biacore 3000 (GE Healthcare) of University of North Carolina Center for Structure Biology exactly as previously described (57,58).

Single-turnover GTPase Assay
The intrinsic and RGS protein-enhanced GTP hydrolysis rates of G␣ subunits were assessed by monitoring the production of 32 P i (inorganic phosphate) during a single round of [␥-32 P]GTP hydrolysis, as previously described (59). In brief, purified G␣ subunits (100 nM) were separately incubated for 10 min at 20°C with 1 ϫ 10 6 cpm of [␥-32 P]GTP (PerkinElmer Life Science; specific activity of 6500 dpm/Ci) in reaction buffer lacking Mg 2ϩ (50 mM Tris, pH 7.5, 0.05% (v/v) polyoxyethylene 10 laurel ether (C12E10), 1 mM DTT, 10 mM EDTA, 100 mM NaCl, and 5 g/ml of BSA). The reaction was then chilled on ice for 5 min prior to initiation of the reaction by the addition of RGS protein, 10 mM MgCl 2 , and 100 M GTP␥S (final concentration). At periodic intervals, 100-l aliquots were quenched in 900 l of charcoal slurry (i.e. 5% (w/v) activated charcoal in 50 mM H 3 PO 4 , pH 3.0) and centrifuged at 4°C for 10 min at 3000 ϫ g. Subsequently, 600-l aliquots of the supernatant were counted via liquid scintillation to quantify 32 P i production. Protocol modifications for quantifying GTPase activity from transducin were made as previously described (60).

Ex Vivo Tracheal Epithelial Cultures
Mouse tracheal epithelia (MTE) were obtained from Rgs21::RFP transgenic mice and aged-matched, wild-type C57Bl/6J mice by enzymatic digestion as previously described (61) under a protocol approved by the UNC IACUC. MTE were maintained at an air-liquid interface in a modified bronchial epithelial growth medium with 5% CO 2 at 37°C and used 2-5 weeks after seeding on 12-mm T-clear inserts (Corning-Costar). To label the airway surface liquid, PBS containing 10-kDa FITC/dextran (0.2 mg/ml; Sigma) was added to MTE mucosal surfaces and then live MTE cultures were imaged by XZ confocal microscopy as previously described (62). Briefly, a 63X glycerol immersion objective was used; the airway surface liquid was imaged using a 488 nm argon laser and TagRFP protein was imaged using a 561 nm diode-pumped solid-state laser.

Stable shRNA-mediated Knockdown of Endogenous RGS21 Expression
Stable 16HBE cell lines with reduced RGS21 expression were generated via lentiviral infection. pLKO.1 plasmids encoding human RGS21-directed shRNA (Oligo IDs TRCN0000036859, TRCN0000036861, and TRCN0000036863; generated by The RNAi Consortium and purchased from Open Biosystems as catalogue number RHS3979 -9604267, RHS3979 -98492449, and RHS3979 -9604271) were prepared from bacterial stocks via maxiprep (Qiagen; Valencia, CA) and packaged into lentiviral stocks by the UNC Lineberger Comprehensive Cancer Center Lenti-shRNA Core Facility (Dr. Tal Kafri, director). A control empty lentiviral vector (Open Biosystems catalogue number RHS4080) was also packaged to establish the negative control cell line. These viruses were used to infect separate 16HBE cell cultures seeded onto 100-mm dishes at 50% confluence. Stably transfected cell lines were selected with puromycin (Cellgro, Manassas, VA) and maintained in standard medium supplemented with puromycin for several weeks prior to use in second messenger assays.

siRNA Knockdown in 16HBE Cells
All siRNA used in this study were chemically synthesized by Dharmacon RNAi Technologies and purchased as an siG-ENOME set consisting of four individual siRNA oligonucleotide duplexes directed toward human RGS21. 16HBE cells seeded in 6-well plates were co-transfected with a pool of all four RGS21-directed siRNA oligonucleotide duplexes (100 nM) with Lipofectamine 2000 (Invitrogen) using the manufacturer's protocol. A control siRNA (siGlo Lamin A/C) was separately transfected to visualize transfection efficiency and act as a negative control. After 24 h, cells were seeded onto poly-D-lysine-coated 96-well plates. Approximately 18 -24 h after seeding, fluorescence imaging plate reader (FLIPR) assays for bitterantinduced calcium flux were performed as described above. Knockdown of the RGS21 transcript was confirmed 48 h posttransfection via qRT-PCR.

FLIPR Calcium Flux Assays
Calcium flux assays were performed using the FLIPR TETRA (Molecular Devices, Sunnyvale, CA) of the NIMH Psychoactive Drug Screening Program located at UNC (PDSP; Dr. Bryan Roth, director) as previously described (63). 16HBE cells were trypsinized, counted, and seeded onto clear-bottom 96-well plates (Greiner Bio-One, Monroe, NC) pre-coated with poly-Dlysine, at a density of 7.5 ϫ 10 4 cells per well. After a 24-h incubation, media was removed and replaced with a no-wash dye (FLIPR Calcium Assay Kit; Molecular Device Corp., Sunnyvale, CA). After a 1-h incubation at 37°C, during which the cells were allowed to take up the dye, fluorescence responses of cells were measured with the FLIPR TETRA device upon the addition of variable concentrations of tastant, or vehicle, in the presence of assay buffer (20 mM HEPES, pH 7.4, 1ϫ Hanks' balanced salt solution (Invitrogen), and 2.5 mM probenecid). After 500 s of data acquisition, a subsequent addition of 5 M thapsigargin or 3 M KCl was injected into each well, and fluorescence was measured again. Net peak responses to tastants were normalized to net peak responses to thapsigargin/ KCl. Responses were compared with that of wild-type control 16HBE cells. Statistical and graphical analyses were performed using Prism version 5.0b (GraphPad Software, La Jolla, CA).

cAMP Accumulation GloSensor Assay
16HBE cell cultures were transiently co-transfected with the GloSensor cAMP-biosensor cDNA (Promega) and pcDNA3.1-based expression plasmids as previously described (64). Twenty-four hours post-transfection, cells were re-plated on poly-D-lysine-treated, clear-bottom, white 384-well plates at a density of 15,000 cells/well. Forty-eight hours post-transfection, culture medium was aspirated and cells were washed once with assay medium (DMEM (without FBS or phenol), 15 mM HEPES, pH 7.4) before being incubated for 2 h with 20 l/well of equilibration medium (assay medium with 4% GloSensor TM substrate (Promega)). After 2 h, 10 l of 3ϫ final concentration of denatonium benzoate (diluted in 3 mM forskolin-containing assay medium) was added to each well and allowed to incubate for 10 min before GloSensor emission was read on a MicroBeta Plate Counter (PerkinElmer). Before plotting, luminescence counts were normalized to 100% maximal response for each condition to account for variability in GloSensor expression, transfection efficiency, and the exact number of cells per well.

RESULTS AND DISCUSSION
RGS21 Is Expressed Outside of the Lingual Gustatory System-Given the discrepancy in expression patterns presented by the first two reports on RGS21 (47,48), we profiled the expression of Rgs21 transcripts by quantitative reverse transcription-PCR (qRT-PCR) on RNA isolated from various human as well as mouse tissues. In human tissues, RGS21 was found highly expressed in circumvallate papillae and expressed at a lower level in lung tissue (Fig. 1A). This same pattern was also observed in mouse tissue, with the addition that Rgs21 was also detected in RNA isolated from whole, homogenized mouse tongue (Fig. 1C). These results are consistent with the expression pattern of rat Rgs21 originally described by von Buchholtz et al. (47). To identify more specifically the cellular distribution of mouse Rgs21 expression within the lingual tissue, in situ hybridization was performed on sections of the mouse tongue. Using this approach, mouse Rgs21 was detected in taste bud cells in the circumvallate papillae, with an expression pattern similar to those of gustducin-␣ and PLC␤2 transcripts (Fig. 1,  E-G).
Given recent evidence suggesting that gustation signaling components are expressed in enteroendocrine cells of the digestive tract (31,32), we also analyzed RGS21 transcript levels in different regions of the human and mouse gastrointestinal tract. Although RGS21 transcripts were identified in various regions of the human gastrointestinal tract (Fig. 1B), Rgs21 expression was not detectable in the mouse gastrointestinal tract (Fig. 1D). Although this may represent a difference in expression between the two organisms, it may also be the result of differential probe sensitivity; however, our results do suggest that RGS21 expression is associated with tastant signaling components known to be expressed in gustatory cells of the human digestive tract (31,32).
To explore further the expression pattern of the mouse Rgs21 gene locus, a bacterial artificial chromosome (BAC) transgenic mouse strain was established that expresses TagRFP (50) driven by the Rgs21 promoter. Consistent with the qRT-PCR data of Fig. 1C and in situ hybridization data of Fig. 1G, TagRFP protein was expressed in the taste buds of the circumvallate papillae (Fig. 2). In combining epifluorescence with immunofluorescence, TagRFP expression was found to be co-localized with gustducin ␣-subunit expression within taste bud cells (Fig. 2,  A-D). To confirm the qRT-PCR data indicating RGS21 expression in lung tissue, we cultured MTE cells from Rgs21::RFP transgenic mice at an air-liquid interface for 3 weeks; XZ confocal scanning revealed that TagRFP fluorescence was clearly detectable in MTE cells of transgenic mouse cultures, but undetectable in control mice (Fig. 2, I-K).
RGS21 Is Predicted to be a Highly Active, and Promiscuous, G␣-directed GAP-RGS21 belongs to the R4 family (9) of RGS proteins that are generally composed of little more than a central, ϳ130 amino acid RGS domain; indeed, at 152 amino acids long, RGS21 is mainly an RGS domain with short N-and C-terminal extensions. Some of us (65) recently described applying electrostatic calculations of interaction energies across multiple, high-resolution G␣/RGS protein structures (e.g. Refs. 66 and 67) to predict RGS domain residues essential for high-level GAP function and G␣ substrate selectivity. Applying this energy-based, residue-level analysis to RGS21 (Fig. 3A) revealed that this RGS protein has the full complement of "significant and conserved" residues critical for the GAP activity of such highly active RGS proteins as RGS4 and RGS16. RGS21 also has eight "functional modulatory" residues ( Fig. 3A), which is at the high end of the range (six to eight) observed in high activity RGS domains (65). Importantly, all of the RGS21 residues in modu-latory positions that are different from those in high activity RGS domains (four positions; Fig. 3A) were predicted not to interfere with canonical G␣ recognition. Therefore, RGS21 was predicted to be a highly active and promiscuous G␣-directed GAP, similar to the high activity RGS domains that share the aforementioned structural properties (e.g. RGS4 and RGS16; Ref. 65).
The primary G␣ subunits implicated in gustation are gustducin-␣ and transducin-␣ (36 -38). Based on the RGS9/transducin-␣ structure and its comparison to other RGS domain-G␣ complexes (66,76), the predicted RGS domain contact residues within gustducin-␣ are 95% identical and 100% similar to those of transducin-␣ (Fig. 3C). Given this degree of similarity, we expect that RGS21 acts equivalently on both transducin and gustducin-␣ subunits (the latter being far more difficult to obtain in purified form than transducin). real-time, quantitative RT-PCR using fluorogenic probe detection was performed on total RNA from indicated human organs. Variable quantities of RNA were normalized by comparison to 18S ribosomal RNA and RGS21 expression in the circumvallate (c.v.) papillae was set to "100% relative expression" using the 2 Ϫ(⌬⌬Ct) method (49). C and D, total RNA was extracted from the indicated mouse organs prior to quantitative RT-PCR analysis. Variation in total amounts of RNA isolated from each organ was normalized by comparison to 18S ribosomal RNA and RGS21 expression in the whole tongue preparation was set to "100% relative expression" using the 2 Ϫ(⌬⌬Ct) method (49). E--G, coronal sections of mouse c.v. papillae were hybridized with digoxigenin riboprobes for PLC␤2 (E), gustducin-␣ (aka GNAT3) (F), and Rgs21 (G). Following hybridization and washes, probes were detected using an alkaline phosphatase-conjugated antidigoxigenin antibody and 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium chromogenic development. Taste buds within stratified squamous epithelium along the left side of the depicted central c.v. papilla are outlined in yellow; examples of probe staining are indicated with arrowheads. DECEMBER 7, 2012 • VOLUME 287 • NUMBER 50

JOURNAL OF BIOLOGICAL CHEMISTRY 41711
SPR spectroscopy was used to determine whether the RGS21/G␣ interaction is direct and occurs in a manner consistent with the canonical RGS domain/G␣ interaction (66,67). An SPR biosensor surface was generated with either immobilized biotin-G␣ i1 or His 6 -G␣ q fusion protein, and RGS21 was subsequently injected over each surface, either in buffer containing aluminum tetrafluoride (to create the transition state form) or GDP alone. RGS21 bound selectively to G␣ i1 (GDP/AlF 4 Ϫ ) and G␣ q (GDP/AlF 4 Ϫ ) with dissociation constants (K d values) of 64 nM (95% confidence interval (CI) of 40 -87 nM) and 20 nM (95% CI 11-29 nM), respectively (Fig. 4, A and B). These affinities for transition state G␣ subunits are consistent with earlier reports from other R4-family RGS proteins obtained using SPR (77). No binding was observed to either G␣ i1 or G␣ q surfaces when in their GDP-bound ground state. A charge reversal of the highly conserved Arg-126 (highlighted in Fig. 3A) to glutamic acid (R126E), previously shown to disrupt other RGS domain/G␣ interactions (53,64), was observed to disrupt the RGS21 inter-action with G␣ i1 (GDP/AlF 4 Ϫ ) and G␣ q (GDP/AlF 4 Ϫ ) surfaces. Circular dichroism studies were performed to verify that the global-fold of the mutant protein was preserved (Fig. 4D).

RGS21 Is a Potent GAP for Multiple G␣ Subunits in Vitro-
To assess directly if RGS21 accelerates the GTPase activity of GTP-bound G␣ subunits, RGS domain-mediated GAP activity FIGURE 2. The Rgs21 gene promoter is active in taste bud cells co-incident with gustducin-␣ expression, as well as in tracheal epithelium. A-G, sections of mouse circumvallate papillae from Rgs21::RFP transgenic (Tg) and wild-type (Wt) mice were stained with anti-gustducin-␣ antibody and then imaged by epifluorescence to detect the expression pattern of gustducin-␣ (green; panels A and B) and TagRFP protein expressed from the Rgs21 promoter (red; panels C and D). Images were merged, with co-localization of gustducin-␣ and RFP signals pseudocolored yellow (panels E and F) and additionally with phase-contrast and DNA staining (DAPI) colored gray and blue, respectively (panels G and H). I-L, explanted tracheal epithelium from Rgs21::RFP transgenic and wild-type mice was cultured ex vivo in an air-liquid interface for 3 weeks and then imaged for TagRFP protein expression by confocal microscopy (panels I-J) and for cellular morphology by bright-field microscopy (panels K-L). Air-liquid interface was highlighted by the addition of FITC-dextran to the airway surface liquid. Images presented are representative of n ϭ 6 Rgs21::RFP transgenic and n ϭ 6 wild-type mice were analyzed.
To ensure that these observed hydrolysis rate increases were not due to contaminating GTPase(s) in the RGS21 protein preparation, single turnover assays were repeated with a RGSinsensitive G␣ i1 point mutant, G␣ i1 (G183S) (79). The intrinsic hydrolysis rate of G␣ i1 (G183S) was not accelerated upon the addition of 250 nM of either wild-type RGS21 or RGS21(R126E) proteins (Fig. 5B).  (65). RGS21 residue arginine 126 is highlighted with an open-star symbol; this residue is known to be a highly conserved, surface-exposed constituent of the G␣-binding "A-site" of RGS domains. B, COS7 cells were transiently transfected with expression vectors for the indicated G␣ subunits and, 48 h later, lysed in buffer containing GDP, Mg 2ϩ , and AlF 4 Ϫ (ϩ) or GDP only (Ϫ). Clarified cell lysates were incubated overnight at 4°C with recombinant His 6 -RGS21 protein and nickel-nitrilotriacetic acid-agarose beads (PD, pulldown), prior to detection of co-precipitated, HA-tagged G␣ subunit by SDS-PAGE, electrotransfer to nitrocellulose, and immunoblotting (IB) with anti-HA antibody and chemiluminescence detection. C, gustducin and transducin G␣ subunits have nearly identical interfaces for RGS domains. A protein sequence alignment of human gustducin-␣ (GNAT3, GenBank accession number NP_001095856.1), transducin-␣ (GNAT1, GenBank NP_653082.1), and G␣ i1 (GNAi1, GenBank NP_002060.4) was generated using ClustalW. Contacts between the Ras-like domain of transducin-␣ (Chain A) and the RGS9 RGS domain (Chain B) from the RGS9/ transducin-␣ structure (PDB code 1FQK; Ref. 76) were identified using contact map analysis (83) and are denoted with a C above the alignment. Sw, switch region. DECEMBER 7, 2012 • VOLUME 287 • NUMBER 50

RGS21 Overexpression Blunts Bitter Tastant Responsiveness of a Human Lung Airway Epithelial Cell
Line-To determine whether RGS21 is capable of modulating gustatory signaling in an integrated, whole cell context, we first used RT-PCR to iden-tify a model cell line expressing components of the gustatory signaling cascade as well as endogenous RGS21. The transformed human bronchial epithelial cell line 16HBE (80) was found to express mRNA transcripts for RGS21, multiple tastant

. RGS21 interacts with transition-state forms of G␣ q and G␣ i/o subunits with nanomolar affinities.
A, biotin-G␣ i1 and alkaline-denatured biotin-G␣ i1 were immobilized on separate streptavidin biosensor surfaces for SPR binding analyses. Indicated concentrations of recombinant RGS21 protein were injected over both surfaces at 20 l/min for 600 s (injection start at t ϭ 0) with a subsequent 300-s dissociation time. Nonspecific binding to the alkaline-denatured biotin-G␣ i1 surface was subtracted using BiaEvaluation software. B, sensorgrams derived from experiments as illustrated in panel A were used for equilibrium saturation binding analyses to determine the affinity of wild-type RGS21 or RGS21(R126E) proteins for G␣ i1 or G␣ q in both their transition state (GDP/AlF 4 Ϫ ) and ground state forms. G␣ q biosensor sensorgrams were obtained by immobilizing His 6 -G␣ q using capture coupling. Dissociation constants (K D values) for the G␣ i1 (GDP/AlF 4 Ϫ )/RGS21 and G␣ q (GDP/AlF 4 Ϫ )/RGS21 interactions were determined to be 64 (95% CI 40 -87 nM) and 20 nM (95% CI 11-29 nM), respectively. No specific binding was observed for the loss-of-function RGS21(R126E) protein toward G␣ i1 or G␣ q nor for wild-type RGS21 toward GDP-bound G␣ i1 or G␣ q . C, GST-RGS21 fusion protein or GST alone was immobilized on an anti-GST antibody biosensor surface. 200 l of the indicated concentrations of various G␣ subunits (in their transition state form) were injected over the biosensor surface. Nonspecific binding toward GST alone was subtracted using BiaEvaluation. The binding at t ϭ 590 s was plotted and used for equilibrium binding analyses. The affinities of G␣ i1 , G␣ i2 , G␣ i3 , G␣ o , and transducin-␣ for immobilized RGS21 were determined to be 114 (95% CI 65-163 nM), 63 (30 -95 nM), 61 (25-97 nM), 24 (18 -32 nM), and 190 nM (134 -247 nM), respectively. Open symbols denote no binding was detected when the various G␣ subunits were in their ground state, GDP-loaded form. D, wild-type RGS21 and RGS21(R126E) proteins were observed to have similar overall secondary structure by circular dichroism analysis. Circular dichroism spectra were separately collected using 0.1 mg/ml of RGS21 and RGS21(R126E) proteins. Spectra were collected from 185 to 260 nm in 0.5-nm steps with a bandwidth of 2 nm. Inset, RGS21 proteins used in these biochemical analyses were equivalently purified from E. coli fermentation, as highlighted by Coomassie Blue staining after being resolved by SDS-PAGE. FIGURE 5. RGS21 acts as a GAP for G i/o subfamily G␣ subunits. Intrinsic GTPase activity of the indicated G␣ subunits, as well as RGS protein acceleration of this activity, was determined using [␥-32 P]GTP single turnover assays. In the absence of Mg 2ϩ , 100 nM G␣ i1 (A), RGS-insensitive G␣ i1 (G183S) (B), G␣ i2 (C), G␣ i3 (D), and G␣ oA (E) were preincubated with [␥-32 P]GTP at 30°C for 10 min. The reaction was then initiated by the addition of MgCl 2 (10 mM), the indicated concentrations of recombinant RGS protein, and cold GTP␥S (400 M). At the indicated time points, the production of inorganic phosphate was quantified by activated charcoal filtration and subsequent liquid scintillation. F, assays of transducin GTPase activity were conducted on a preformed complex between the G-protein heterotrimer (transducin-␣␤␥; 2 M) and the photoreceptor rhodopsin (20 M) and were initiated by addition of [␥-32 P]GTP supplemented with GTP (250 nM). Nonlinear regression was used to fit the data to a single exponential function; observed rates are presented in Table 1.
Bitter tastant signaling is known to rapidly reduce cellular cyclic nucleotide levels (40). To test the hypothesis that RGS21 can serve as a negative regulator of bitter tastant signal transduction, the bitter receptor agonist denatonium benzoate ("denatonium-B") was applied to 16HBE cells and the resultant cAMP decreases were monitored in the context of artificially heightened cAMP levels (i.e. forskolin-mediated activation of endogenous adenylyl cyclases). In a concentration-dependent fashion (Fig. 6B), denato-nium-B application reduced forskolin-mediated cAMP production, as predicted (40); overexpression of wild-type RGS21, but not the GAP-dead RGS21(R126E) mutant, significantly blunted the bitter tastant-induced reduction in cAMP production. To ensure that RGS21 overexpression did not have unanticipated, global consequences on GPCR-mediated signaling, isoproterenol-activated production of cAMP (via ␤-adrenergic receptor activation) was also monitored in 16HBE cells overexpressing wild-type RGS21, RGS21(R126E), or vector alone. No significant alteration in the potency (EC 50 values) of isoproterenol upon RGS21 overexpression was observed (Fig. 6D). FIGURE 6. Overexpression of wild-type RGS21, but not a loss-of-function, point mutant, leads to reduction of bitter tastant signaling. A, real-time qRT-PCR using fluorogenic probe detection was performed on total RNA from the human bronchial epithelium cell line 16HBE to detect the indicated gene transcripts; data were normalized to levels of 18S rRNA detection. B, 16HBE cell cultures were transiently co-transfected with the Promega GloSensor cAMPbiosensor cDNA and expression plasmids containing HA epitope-tagged open reading frames for wild-type (wt) rat RGS21, or Arg-126 to Glu point mutated (Arg 3 Glu) rat RGS21, or empty pcDNA3.1 vector, as indicated. Inhibition of forskolin (FSK) stimulated cAMP production by treatment with the indicated concentrations of denatonium-B was determined 24 h post-transfection by detection of GloSensor-dependent luminescence. Inset, Western blot of whole cell lysates from transfected 16HBE cultures with anti-HA-epitope antibody and chemiluminescence detection, demonstrating equivalent expression of wild-type and point mutant RGS21 proteins. Asterisks denote statistically significant differences from signal obtained upon empty vector expression: *, p Ͻ 0.05; **, p Ͻ 0.01 (one-way analysis of variance with Bonferroni's post-test). C, 16HBE cell cultures were transiently transfected with expression plasmids containing wild-type (wt) RGS21, or point mutated (Arg 3 Glu) RGS21, or empty vector, as indicated. Net peak intracellular calcium production, as elicited by treatment with the indicated concentrations of denatonium-B was determined using calcium indicator dye and fluorometric imaging plate reader (FLIPR) detection (relative fluorescence units, RFU), 48 h post-transfection. D, isoproterenol-stimulated production of cAMP is not perturbed by overexpression of RGS21 or RGS21(R126E) in 16HBE cells. Cells were transfected exactly as described in panel B. Isoproterenol-induced cAMP production was detected by GloSensor-dependent luminescence. EC 50 values for isoproterenol were found to be 2.9 (95% CI 2.3-3.7 nM), 3.8 (3.1-4.7 nM), and 3.8 nM (3.1-4.7 nM) for 16HBE cells overexpressing wild-type RGS21, RGS21(R126E), and empty vector, respectively.

TABLE 1 Concentration dependence of RGS21-mediated acceleration of G␣ GTP hydrolysis rates as observed in single turnover assays
Rates were determined using nonlinear regression of data obtained at 4°C (Fig. 5) DECEMBER 7, 2012 • VOLUME 287 • NUMBER 50

RGS21 Is a Promiscuous G␣ GAP and Inhibits Bitter Gustation
In tastant receptor signal transduction, the released G␤␥ heterodimer is thought to play a key role in intracellular signaling via PLC-␤2 activation (18,42), which results in production of IP 3 and DAG second messengers and a subsequent increase in intracellular calcium (81). To determine whether RGS21 is also able to inhibit tastant-mediated activation of Ca 2ϩ release, we monitored intracellular calcium in 16HBE cells upon denatonium-B treatment. A decrease in the intracellular Ca 2ϩ flux caused by treatment with denatonium-B was observed in RGS21-overexpressing 16HBE cells, as compared with cells overexpressing the GAP-dead RGS21(R126E) mutant or transfected with empty vector only (Fig. 6C).
Knockdown of Endogenous RGS21 Expression Enhances Bitter Tastant Responsiveness-Although both cAMP accumulation and Ca 2ϩ mobilization data suggest that RGS21 can act as a negative regulator of bitter tastant signaling upon its ectopic overexpression, we also wanted to test whether endogenous RGS21 expression serves to negatively regulate denatonium-B signaling. Three stable 16HBE cell lines were generated with decreased RGS21 expression using lentiviral shRNA-mediated knockdown (Fig. 7A). No difference was seen in levels of forskolin-stimulated cAMP production in the absence of denatonium-B; however, upon denatonium-B treatment, significant decreases in cAMP generation were observed in all three cell lines with reduced RGS21 expression versus the control lentivirus-treated cell line (Fig. 7B). To confirm that stable RGS21 knockdown had not disrupted GPCR-mediated cAMP signaling in a nonspecific fashion, isoproterenol responses were also tested in these cell lines. No significant different in potency or maximal efficacy of isoproterenol was observed (Fig. 7D). Intracellular Ca 2ϩ flux responses were also measured in the stable shRNA-mediated knockdown cell lines upon denatonium-B addition. At both agonist concentrations tested, decreased RGS21 expression resulted in an increase in denatonium-Binduced intracellular Ca 2ϩ accumulation (Fig. 7C). Increased calcium flux upon denatonium-B treatment was also observed upon siRNA-mediated knockdown of RGS21 expression (Fig.  7E).
Conclusions-RGS21 is capable of interacting with, and serving as a GAP for, multiple G protein ␣ subunits, including those thought to be involved in gustatory signal transduction. RGS21 is endogenously expressed in human and mouse lingual and airway epithelia and human intestinal tissues. The observed tissue expression pattern of RGS21 lies between those reported in two previous studies (47,48): whereas RGS21 is clearly expressed in lingual taste bud cells (47), it is also expressed in other tissues with gustatory epithelia. Interestingly, all of these tissues additionally express tastant receptors and other, down-FIGURE 7. RNA interference-mediated knockdown of endogenous RGS21 potentiates bitter tastant signaling. A, quantitative RT-PCR demonstrating shRNA-mediated knockdown of RGS21 transcript levels in stable, lentivirus-transfected variants of the human bronchial epithelial cell line 16HBE. Real-time, quantitative PCR using fluorogenic probe detection was performed for the RGS21 transcript, normalized to 18S rRNA signal, and expressed as the percentage of RGS21 expression observed in control 16HBE cells stably transfected with empty lentivirus. B, 16HBE cells were stably transfected with either empty lentivirus (Empty) or one of three, RGS21-directed shRNA-expressing lentiviruses (#59, #61, #63), as indicated. Inhibition of forskolin-stimulated cAMP production by treatment with the indicated concentrations of denatonium-B were determined 24 h post-transfection by detection of GloSensor-dependent luminescence. Asterisks denote statistically significant differences from empty vector, p Ͻ 0.01 (one-way analysis of variance with Dunnett's multiple comparison test). C, net peak intracellular calcium production by 16HBE cells stably expressing the indicated lentiviruses, as elicited by treatment with the indicated concentrations of denatonium-B, was determined using calcium indicator dye and FLIPR detection (relative fluorescence units, RFU). D, isoproterenol-stimulated cAMP production is not perturbed by shRNA-mediated RGS21 knockdown. Cells were treated exactly as described in panel B. Isoproterenol-induced cAMP production was detected by GloSensor-dependent luminescence. EC 50 values for isoproterenol were 4.4 (95% CI 3.6 -5.5 nM), 3.7 2.9 -4.8 nM), 5.2 (4.4 -6.1 nM), and 3.2 nM (2.4 -4.1 nM) for 16HBE cells expressing empty vector control, shRNA#59, shRNA#61, or shRNA#63, respectively. E, parental 16HBE cells were transfected with nonspecific (Control) or RGS21-directed short-interfering RNA oligonucleotide duplexes (siRNA), prior to measurement of denatonium-B-stimulated calcium flux as in panel C. *, p Ͻ 0.03 (unpaired Student's t test). stream components of the gustatory signaling cascade (31)(32)(33)(34), suggesting a role for RGS21 in negatively regulating tastant signal transduction in multiple cell types.
In identifying a human airway epithelial cell line (16HBE) that co-expresses RGS21, bitter tastant receptors, and downstream tastant signaling proteins, we have established that RGS21 indeed serves to blunt intracellular signaling by at least one tastant agonist, namely the bitterant denatonium. As lingual epithelium-derived taste bud cells are notoriously difficult to maintain in culture (82), our results suggest that established, immortalized lung epithelial cells may present a useful alternative system in which to study the action of RGS21 and other modulatory molecules on gustatory signal transduction using endogenous machinery (e.g. avoiding receptor and/or effector overexpression); however, it is possible that the signaling cascade in airway cells is different from that in taste bud cells because of differing subcellular environments and/or downstream effectors. In the taste bud, it is likely that RGS21 and/or other RGS proteins help to dampen signaling so that the gustatory system is not overwhelmed when molecules are abundant on the lingual epithelium. In the airway, detection of bitter tastants is hypothesized to help with mucus clearance of irritants by increasing cilia beat frequency (33,34); thus, the role of RGS21 in lung epithelium may be to regulate mucus clearance by cillated airway cells in response to bitter irritants.