A G Protein-biased Designer G Protein-coupled Receptor Useful for Studying the Physiological Relevance of Gq/11-dependent Signaling Pathways*♦

Designer receptors exclusively activated by a designer drug (DREADDs) are clozapine-N-oxide-sensitive designer G protein-coupled receptors (GPCRs) that have emerged as powerful novel chemogenetic tools to study the physiological relevance of GPCR signaling pathways in specific cell types or tissues. Like endogenous GPCRs, clozapine-N-oxide-activated DREADDs do not only activate heterotrimeric G proteins but can also trigger β-arrestin-dependent (G protein-independent) signaling. To dissect the relative physiological relevance of G protein-mediated versus β-arrestin-mediated signaling in different cell types or physiological processes, the availability of G protein- and β-arrestin-biased DREADDs would be highly desirable. In this study, we report the development of a mutationally modified version of a non-biased DREADD derived from the M3 muscarinic receptor that can activate Gq/11 with high efficacy but lacks the ability to interact with β-arrestins. We also demonstrate that this novel DREADD is active in vivo and that cell type-selective expression of this new designer receptor can provide novel insights into the physiological roles of G protein (Gq/11)-dependent versus β-arrestin-dependent signaling in hepatocytes. Thus, this novel Gq/11-biased DREADD represents a powerful new tool to study the physiological relevance of Gq/11-dependent signaling in distinct tissues and cell types, in the absence of β-arrestin-mediated cellular effects. Such studies should guide the development of novel classes of functionally biased ligands that show high efficacy in various pathophysiological conditions but display a reduced incidence of side effects.

Designer receptors exclusively activated by a designer drug (DREADDs) are clozapine-N-oxide-sensitive designer G protein-coupled receptors (GPCRs) that have emerged as powerful novel chemogenetic tools to study the physiological relevance of GPCR signaling pathways in specific cell types or tissues. Like endogenous GPCRs, clozapine-N-oxide-activated DREADDs do not only activate heterotrimeric G proteins but can also trigger ␤-arrestin-dependent (G protein-independent) signaling. To dissect the relative physiological relevance of G protein-mediated versus ␤-arrestin-mediated signaling in different cell types or physiological processes, the availability of G proteinand ␤-arrestin-biased DREADDs would be highly desirable. In this study, we report the development of a mutationally modified version of a non-biased DREADD derived from the M 3 muscarinic receptor that can activate G q/11 with high efficacy but lacks the ability to interact with ␤-arrestins. We also demonstrate that this novel DREADD is active in vivo and that cell type-selective expression of this new designer receptor can provide novel insights into the physiological roles of G protein (G q/11 )-dependent versus ␤-arrestin-dependent signaling in hepatocytes. Thus, this novel G q/11 -biased DREADD represents a powerful new tool to study the physiological relevance of G q/11 -dependent signaling in distinct tissues and cell types, in the absence of ␤-arrestin-mediated cellular effects. Such studies should guide the development of novel classes of functionally biased ligands that show high efficacy in various pathophysiological conditions but display a reduced incidence of side effects.
During the past few years, designer receptors exclusively activated by a designer drug (DREADDs) 2 have emerged as powerful novel chemogenetic tools to study the physiological relevance of signaling pathways activated by different functional classes of G protein-coupled receptors (GPCRs) (1)(2)(3). Structurally, DREADDs represent mutant muscarinic receptors that can be activated by clozapine-N-oxide (CNO), an otherwise pharmacologically inert agent, with high potency and efficacy (4,5). Importantly, these new designer receptors cannot be activated by acetylcholine, the endogenous muscarinic receptor agonist (4,5).
Recently, DREADDs with different G protein-coupling properties have been expressed in a cell type-specific fashion in mice and other experimental animals, for example by employing classical mouse transgenic approaches or by using viral gene transfer strategies (1)(2)(3). CNO treatment of the DREADD-expressing animals then leads to the selective stimulation a distinct GPCR signaling pathways only in DREADD-expressing cells. This approach makes it possible to assess the in vivo consequences of activating distinct GPCR signaling pathways in specific cell types. Clearly, such studies are difficult to perform with native GPCRs, which, with virtually no exception, are expressed in multiple tissues and cell types (6).
Agonist binding to GPCRs causes rapid phosphorylation of the activated receptors by GPCR kinases (7,8). This process promotes the recruitment of members of the arrestin protein family (␤-arrestin-1 and -2) to the activated receptors, disrupting receptor/G protein coupling, and promoting GPCR internalization by targeting the receptors to clathrin-coated pits (7,8). However, during the past 10 -15 years, many studies have demonstrated that ␤-arrestins can also act as signaling molecules in their own right (9 -13). This observation is not only of theoretical interest but also of potential clinical relevance (9 -11, 14 -17).
Consistent with the GPCR-like properties of DREADDs, two recent studies have demonstrated that the M 3 muscarinic receptor (M3R)-based G q DREADD (M3Dq) does not only activate G proteins of the G q family but can also interact with ␤-arrestin-1 and -2 and trigger ␤-arrestin-dependent downstream signaling (18,19). It is likely that other DREADDs are also endowed with the ability to couple to both G protein-and ␤-arrestin-dependent signaling pathways.
These new findings suggest that the physiological outcome of activating a specific GPCR (or DREADD) in a particular tissue or cell type represents an integrated response caused by the activation of both G protein-and ␤-arrestin-dependent signaling pathways. To gain insight into the physiological relevance of these two distinct GPCR signaling branches, it is important to assess the relative contribution of G protein-and ␤-arrestin-dependent signaling to a particular tissue response. To shed light on this issue, we recently generated an M3R-based, ␤-arrestinbiased DREADD (18) (M3D-arr in Fig. 1). In this study, we describe the generation of a novel G q/11 -biased, M3R-derived DREADD that lacks the ability to interact with ␤-arrestins (M3D-Gq in Fig. 1). We also demonstrate that this novel DREADD is active in vivo and that cell type-selective expression of the new designer receptor can lead to novel insights into the physiological roles of G protein-dependent versus ␤-arrestindependent signaling (in hepatocytes). This novel G q/11 -biased DREADD represents a powerful new tool to study the physiological relevance of G q/11 -dependent signal cascades in distinct tissues and cell types, in the absence of ␤-arrestin-mediated cellular effects.
These new findings suggest that the physiological outcome of activating a specific GPCR or DREADD in a particular tissue or cell type represents an integrated response caused by the activation of both G protein-and ␤-arrestin-dependent signaling pathways. To gain insight into the physiological relevance of these two distinct GPCR signaling branches, it is important to assess the relative contribution of G protein-and ␤-arrestin-dependent signaling to specific tissue response(s). Importantly, such information could be exploited for developing novel classes of clinically useful drugs, including G protein or ␤-arrestin-biased agonists (9 -11, 14 -16).

Experimental Procedures
Drugs-Acetylcholine chloride, carbamylcholine chloride (carbachol), and atropine sulfate were obtained from Sigma. Generation of a New Designer GPCR (DREADD)-To generate an M3R-based DREADD that is able to couple to G q -type G proteins but lacks the ability to recruit ␤-arrestins, we introduced the following mutations into the M3D DREADD described by Guettier et al. (5). We removed the central portion of the third intracellular (i3) loop (Ala-274 -Lys-469) and introduced four point mutations (C542A, C546S, C560S, and C562S) by using the QuikChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA), according to the manufacturer's instructions. Amino acid numbers refer to the rat M3R sequence. For the sake of simplicity, we refer to this newly generated DREADD simply as "M3D-Gq." The M3D-Gq coding sequence was inserted into the pcDNA3.1(Ϫ) vector (Invitrogen) using standard cloning techniques. The entire M3D-Gq coding sequence was sequenced to exclude the presence of any unwanted mutations. All DREADDs used in this study carried an N-terminal HA epitope tag (5,18) and were cloned into the pcDNA3.1(Ϫ) vector.
Transient Expression of DREADDs in Cultured Cells-DREADD/receptor constructs were transiently expressed in COS-7 or HEK293 cells using Lipofectamine 2000 according to the manufacturer's protocol (Thermo Fisher Scientific; for more details, see Ref. 20). Transfected COS-7 cells were incubated with 1 M atropine for the last 24 h of culture, to obtain higher receptor densities. About 48 h after transfections, cells were harvested and used for radioligand binding or functional assays.
Radioligand Binding Studies-Radioligand binding studies were performed with membrane preparations obtained from transfected COS-7 cells as described previously (20). In brief, binding reactions containing ϳ10 g of membrane protein per tube were carried out for 2 h at 22°C in 0.5 ml of binding buffer containing 25 mM sodium phosphate and 5 mM MgCl 2 (pH 7.4). In saturation binding assays, we employed six different [ 3 H]NMS concentrations (1.5-50 nM). In competition binding assays, we used a fixed concentration of [ 3 H]NMS (5 nM) in the presence of 10 different concentrations of CNO or acetylcholine. Nonspecific binding was defined as binding observed in the presence of 10 M atropine. Binding reactions were terminated by rapid filtration over GF/C glass fiber filters (Brandel, Gaithersburg, MD), followed by three washes (ϳ4 ml per wash) with ice-cold distilled water. The amount of radioactivity that remained bound to the filters was determined by liquid scintillation spectrometry. Binding data were analyzed by using the nonlinear curve-fitting program Prism 6.04 (GraphPad Software, San Diego).
Calcium Mobilization Assay-COS-7, WT HEK293, or G␣ q/11 -deficient HEK293 cells transiently expressing DREADDs with different coupling properties were resuspended in DMEM supplemented with 10% fetal bovine serum and grown in collagen-coated 96-well plates (black-walled, clear bottom; Thermo Scientific). Cells were plated at a density of 2.5 ϫ 10 4 cells per well. About 48 h after transfection, CNO-induced changes in intracellular calcium levels ([Ca 2ϩ ] i ) were measured via FLIPR (Molecular Devices, Sunnyvale, CA), as described previously (21). After a washing step (200 l of Hanks' balanced salt solution containing 20 mM HEPES (pH 7.4) per well), 30 l of loading dye (FLIPR calcium 3 assay kit containing the proprietary calcium 3 fluorophore; Molecular Devices), supplemented with 2.5 mM probenecid to increase dye retention, was added to each well, followed by a 45-min incubation at 37°C. Cells were then incubated with increasing concentrations of CNO, and changes in cell fluorescence were measured at 22°C via FLIPR (FLIPR TETRA; excitation wavelength, 470 -495 nm; emission wavelength, 515-575 nm). For each CNO dose, we measured increases in intracellular calcium levels as peak fluorescence activity minus basal fluorescence activity prior to the addition of CNO. Cells expressing the wild-type (WT) M3R were pro-cessed in the same fashion as DREADD-expressing cells, except that carbachol and acetylcholine were used as agonists. Agonist E max and EC 50 values were obtained by analyzing concentration-response curves via GraphPad Prism 6.04 (GraphPad Software, San Diego). All assays were carried out in quadruplicate.
Using the same experimental setup, we also carried out FLIPR assays with primary hepatocytes isolated from mice injected with recombinant adeno-associated viruses (AAVs; see below) coding for the various DREADDs used in this study. Primary hepatocytes were plated at a density of 1.2 ϫ 10 4 cells per well in 96-well plates. All experimental conditions remained the same as described in the previous paragraph, except that hepatocytes were incubated at 37°C for 75 min after the addition of the loading dye (100 l).
Western Blotting Studies-Immunoblotting experiments were carried out using published protocols (20,21). In brief, samples containing 20 g of solubilized membrane protein were incubated with Laemmli loading buffer (30 min at 37°C) under reducing conditions, loaded onto 10% Tris-glycine polyacrylamide gels (Invitrogen), and run at 135 V in the presence of 0.1% SDS. Immunoblots were probed with a rabbit monoclonal anti-HA antibody (catalog no. C29F4, Cell Signaling, Boston, MA) or a mouse monoclonal anti-FLAG antibody (catalog no. A00187, GenScript, Piscataway Township, NJ). Immunoreactive proteins were visualized by using SuperSignal West Pico Chemiluminescent Substrate (Pierce).
Co-immunoprecipitation-Experiments were carried out as described (21). COS-7 cells were co-transfected with DREADDs containing an N-terminal HA tag and a ␤-arrestin-1 construct carrying a C-terminal FLAG tag (ID 14687, Addgene, Cambridge, MA). About 48 h after transfection, cells were treated with 10 M CNO for 10 min at 37°C and then solubilized in lysis buffer containing 1% digitonin, 25 mM sodium phosphate, 5 mM MgCl 2 (pH 7.4), and complete protease and PhosSTOP phosphatase inhibitor cocktails (Roche Applied Science). Solubilized membrane proteins (ϳ600 g per tube) were incubated for 1 h at 4°C with an anti-FLAG monoclonal antibody (catalog no. A00187, GenScript, Piscataway, NJ). After this step, 30 l of protein A/G-agarose (Santa Cruz Biotechnology, Dallas, TX) were added, followed by an additional incubation period of 1 h at 4°C. Bound immunoreactive proteins were eluted, and immunoblotting studies were performed using the rabbit monoclonal anti-HA antibody (see previous paragraph), as described in detail previously (21).
Live-cell Imaging of DREADD Internalization Using Fluorescence Microscopy-HEK293 cells were plated onto sterile 8-well -slides (Ibidi GmbH, Martinsried/Munich, Germany), grown to 70% confluence, and then transfected using Lipofectamine 2000 with plasmids encoding DREADDs carrying a C-terminal EYFP tag (0.9 g DNA per construct). The DREADD-EYFP constructs were generated using standard cloning techniques (vector, pcDNA3). Live-cell imaging was performed 24 h after transfection using a Zeiss Axio Observer microscope with an Apo Tome Imaging System equipped with a heating insert P-Lab-Tec S1 unit. Fluorescence images were taken from living cells at 37°C on a heated microscope stage using AxioVision Rel. 4.6 software. Cells were starved for 30 min in serum-free Opti-MEM medium (Thermo Fisher Scientific), and cell nuclei were stained with Hoechst 33342 (5 g/l) for 10 min. To study DREADD internalization, cells were incubated with carbachol or CNO (100 M each) for 45 min. Images were acquired before and after agonist treatment.
Generation of DREADD-encoding Recombinant AAVs-The DREADD-coding sequences containing an N-terminal HA tag were subcloned into the AAV-TBG-MCS plasmid (a kind gift from Dr. Morris Birnbaum, University of Pennsylvania, Philadelphia) (25). Escherichia coli SURE cells (Agilent Technologies, Santa Clara, CA), which are deficient in E. coli genes involved in the rearrangement and deletion of DNA, were used as host cells to reduce the likelihood of recombination events due to the presence of the ITR sequences in the AAV-TBG-MCS plasmid. This strategy yielded the following three recombinant viral constructs: AAV-TBG-M3D, AAV-TBG-M3D-arr, and AAV-TBG-M3D-Gq. AAV viral particles were generated by the University of Pennsylvania Vector Core. The AAV-TBG-eGFP control virus (catalog no. AV-8-PV0146) was obtained through the same source.
Mouse Maintenance and Diet-C57BL/6NTac (Taconic, Germantown, NY) were housed in a specific pathogen-free barrier facility maintained on a 12-h light, 12-h dark cycle (light period from 6:00 a.m. to 6:00 p.m.). Mice were fed ad libitum with standard mouse chow (4% (w/w) fat content; Zeigler, Gardners, PA). All animal studies were approved by the Animal Care and Use Committee of the NIDDK, National Institutes of Health (Bethesda, MD).
Selective Expression of DREADDs in Mouse Liver/Hepatocytes-AAVs (AAV-TBG-M3D, AAV-TBG-M3D-arr, AAV-TBG-M3D-Gq, and AAV-TBG-eGFP) were injected into the tail vein of WT C57BL/6NTac mice (8-week-old males) by using an insulin syringe (0.5-ml; needle, 28-gauge, 1 ⁄ 2-inch long, 0.36 ϫ 13 mm) loaded with 100 l of diluted AAV solution (10 11 AAV genomic copies per 100 l). Each mouse received 10 11 genomic copies regardless of body weight. Two to 3 weeks after AAV injections, the liver-specific expression of the various DREADDs was determined via Western blotting. The number of DREADD-binding sites in the liver was determined in [ 3 H]NMS binding studies using liver membrane homogenates. [ 3 H]NMS binding studies were carried out in the same fashion as described above for transfected COS-7 cells.
DREADD-mediated Increases in Blood Glucose Levels-Two weeks after AAV treatment (see previous paragraph), mice were injected with CNO (10 mg/kg intraperitoneally). Blood samples were taken from the mouse tail just prior to injection (time "0") and 15, 30, 60, and 90 min post-injection. Blood glucose levels were measured with an automated blood glucometer (Contour, Bayer, Whippany, NJ).
DREADD-mediated Increases in Hepatic Glucose Production Studied with Primary Hepatocytes-Two weeks after AAV treatment of WT C57BL/6NTac mice, primary hepatocytes were harvested as described earlier (26). Hepatocytes were first cultured in 12-well plates (phenol red-free DMEM containing 10% FBS and 1 g/liter glucose) for ϳ4 h at 37°C. The medium was then replaced with glucose-and phenol red-free DMEM supplemented with gluconeogenic substrates (20 mM sodium lactate and 2 mM sodium pyruvate), followed by a 16-h incubation at 37°C. Subsequently, hepatocytes were incubated in the same medium in the presence or absence of glucagon (100 nM; positive control) or CNO (10 M) for 4 h at 37°C. After this time, the culture medium was collected for the measurement of glucose levels.

Knockdown of G␣ q/11 Expression in Primary Mouse
Hepatocytes-M3D-Gq-expressing hepatocytes were isolated from AAV-injected mice and plated in 6-well plates (5 ϫ 10 5 cells per well) in DMEM containing 10% FBS. To knock down G␣ q/11 expression, hepatocytes were transfected with G␣ q and G␣ 11 siRNAs (On Target plus siRNA SMARTpools, GE Dharmacon, Lafayette, CO) or scrambled control siRNA using Lipofectamine RNAiMAX transfection reagent (Thermo Fisher Scientific). About 24 h later, hepatocytes were processed for glucose output measurements as described in the previous paragraph. At the end of the experiment, cells were harvested for the isolation of total RNA. G␣ q and G␣ 11 transcript levels were determined by quantitative RT-PCR. The following primer pairs were used: G␣ q , 5Ј-TTGGTTCGAGAGGTTGA-TGTG-3Ј and 5Ј-CGTCGTCTGTCGTAGCACTC-3Ј; G␣ 11 , 5Ј-CAACGCGGAGATCGAGAAACA-3Ј and 5Ј-GCCTGC-ATGGCGGTAAAGAT-3Ј.
Receptor Internalization Studies with M3D-arr-Expressing Hepatocytes-To confirm that the M3D-arr receptor remained functional in vivo, we carried out internalization assays using primary hepatocytes expressing the M3D-arr construct. Hepatocytes growing in collagen-coated 6-well plates were incubated with CNO (10 M) for 60 min at 37°C in DMEM containing 300 mM HEPES (incubation volume: 2 ml). After this step, cells were incubated for 2 h at 4°C in the same buffer with 10 nM [ 3 H]NMS, a membrane-impermeable radioligand, as described in detail previously (27). Nonspecific binding was determined in the presence of 10 M atropine. Assays were carried out in duplicate.
Statistics-Data are expressed as means Ϯ S.E. for the indicated number of observations. For comparisons between two groups, the unpaired Student's t test (two-tailed) was used. A p value of Ͻ0.05 was considered statistically significant.

Results
Generation of a G q/11 -Biased DREADD-Using the non-biased M3Dq DREADD (4, 5) as a template, we recently generated an arrestin-biased version of this DREADD (18). For sim- ) can couple to both G q -type G proteins and recruit ␤-arrestins (4,5,18,19). The M3D-arr DREADD (center panel) represents a ␤-arrestin-biased version of M3D that contains the R165L point mutation (18). The right panel depicts the G q/11 -biased DREADD (M3D-Gq) that we developed and characterized in this study. When activated by CNO, the M3D-Gq designer receptor efficiently activates G q/11 but fails to recruit ␤-arrestins (see text for details). The M3D-Gq construct lacks the central portion of the third intracellular (i3) loop (Ala-274 -Lys-469) and contains four additional point mutations: C542A, C546S, C560S, and C562S. Amino acid numbers refer to the rat M3R sequence. M3R, M 3 muscarinic receptor.
plicity, we will refer to the non-biased DREADD as "M3D" and to the arrestin-biased DREADD as "M3D-arr," respectively ( Fig. 1). To generate an M3D-derived DREADD that retains the ability to activate G proteins of the G q family but is unable to recruit ␤-arrestins, we screened a series of mutationally modified M3D constructs in transfected COS-7 cells.
The agonist-activated M3R is phosphorylated on multiple serine residues within the third intracellular (i3) loop (28,29), and simultaneous replacement of most of these residues with alanine causes a pronounced reduction in both agonist-dependent M3R phosphorylation and ␤-arrestin recruitment (30). However, deletion of most of the i3 loop of the M3R (deletion of amino acids Ala-274 -Lys-469; rat M3R sequence) does not interfere with the ability of the WT M3R to activate G q -type G proteins (31,32). On the basis of these observations, we introduced the Ala-274 -Lys-469 deletion into the M3D construct. However, pilot BRET experiments showed that this structural modification was not sufficient to completely prevent M3Dmediated ␤-arrestin recruitment, in agreement with data presented by Kong et al. (30).
Various lines of evidence (33)(34)(35)(36), including recent high resolution structural information of GPCR-arrestin complexes (37,38), indicate that residues within the C-terminal portion of transmembrane helix 7 and the adjacent helix 8 play important roles in mediating the conformational changes required for productive GPCR/arrestin interactions. Prompted by these findings, we introduced a series of point mutations into the cytoplasmic end of transmembrane helix 7 and helix 8 of the M3D construct lacking Ala-274 -Lys-469. This approach eventually yielded a mutationally modified version of M3D that retained the ability to activate G q -type G proteins in an M3Dlike fashion, but it was no longer able to recruit ␤-arrestins. For the sake of simplicity, we refer to this G q/11 -biased DREADD as M3D-Gq (Fig. 1). Besides the Ala-274 -Lys-469 deletion, this new DREADD contains four additional point mutations, two in the C-terminal portion of transmembrane helix 7 (C542A and C546S) and two at the end of helix 8 (C560S and C562S) (Fig. 1). Below we will describe the pharmacological and functional properties of this novel DREADD. In addition, we demonstrate that the M3D-Gq DREADD represents a highly useful tool for distinguishing between G protein-mediated versus ␤-arrestinmediated signaling pathways in vivo.
Ligand Binding Properties of M3D-Gq-To determine the ligand properties of M3D-Gq, we performed [ 3 H]NMS saturation and competition binding studies, using membranes prepared from transfected COS-7 cells. WT M3R-containing membranes were included for control purposes in all experiments. [ 3 H]NMS saturation binding studies showed that M3D-Gq was expressed at similarly high levels as the M3D parent DREADD and that M3D-Gq, similar to M3D, was able to bind CNO with high affinity (K i ϳ1 M; Table 1). In contrast, both DREADDs failed to bind acetylcholine, the physiological muscarinic receptor agonist (highest concentration tested: 0.3 mM; Table 1).
M3D-Gq Retains the Ability to Couple to G q/11 -To investigate whether M3D-Gq retained the ability to activate G proteins of the G q family, we initially studied CNO-induced increases in intracellular calcium levels [Ca 2ϩ ] i in DREADDexpressing COS-7 cells. FLIPR studies showed that CNO treatment of M3D-Gq-expressing cells caused pronounced concentration-dependent increases in [Ca 2ϩ ] i , in a fashion similar to that observed with cells expressing the non-biased M3D DREADD (Table 2 and Fig. 2A), indicating that M3D-Gq retains the ability to efficiently couple to G q/11 . In contrast, exposure of M3D-Gq-or M3D-expressing cells to acetylcholine or carbachol did not lead to significant increases in [Ca 2ϩ ] i ( Table 2).
We obtained very similar results when we carried out FLIPR studies with transfected WT HEK293 cells (Fig. 2B). Strikingly, the M3D-Gq-and M3D-mediated calcium responses were completely abolished in HEK293 cells lacking both G␣ q and G␣ 11 (G␣ q/11 KO HEK293 cells (39)). Co-expression of G␣ q with M3D-Gq in G␣ q/11 KO HEK293 cells rescued the CNO/ M3D-Gq-mediated increase in [Ca 2ϩ ] i (Fig. 2C). These data clearly indicate that the CNO/M3D-Gq-triggered elevation in [Ca 2ϩ ] i is mediated by G q -type G proteins.
DREADD/Arrestin Interactions Studied Via BRET-We next used BRET technology to examine the ability of M3D-Gq to  recruit ␤-arrestins in a CNO-dependent fashion. Following established protocols (22)(23)(24), we co-transfected COS-7 cells with DREADD constructs (M3D or M3D-Gq) containing a C-terminal Renilla luciferase 8 sequence (M3D-Luc or M3D-Gq-Luc, respectively; BRET donor) and ␤-arrestin-1 or -2 constructs carrying an N-terminal Venus tag (V-␤-arrestin-1 or V-␤-arrestin-2, respectively; BRET acceptor). Initially, we transfected cells with a fixed amount of DREADD-Luc construct and varied the amount of co-transfected V-␤-arrestin plasmids. The following day, we measured luminescence at 460 and 535 nm after incubation of cells with coelenterazine-h and 20 M CNO (or vehicle for unstimulated cells). In the case of M3D-Luc-expressing cells, this approach led to saturable net BRET curves in the presence of increasing amounts of V-␤-arrestin-1 or -2 (Fig. 3, A and B). In contrast, no significant increases in net BRET were observed with M3D-Gq-Luc-expressing cells under the same experimental conditions (Fig. 3, A and B).
In another set of BRET experiments, we co-transfected COS-7 cells with fixed amounts of DREADD-Luc and V-␤arrestin-1 or -2 DNA. The next day, we carried out BRET mea-  Table 2). The data shown in B and C represent means Ϯ S.E. from three independent experiments, each performed at least in triplicate. G␣ q, human G␣ q inserted into the pcDNA3.1 expression plasmid. Veh, vehicle. surements in the presence of increasing concentrations of CNO as described in the previous paragraph. Consistent with published data (18,19), CNO treatment of cells co-expressing M3D-Luc and V-␤-arrestin-1 or -2 led to concentration-dependent net BRET signals, indicative of CNO-dependent ␤-arrestin recruitment (Fig. 3, C and D). In contrast, no significant increases in net BRET were observed with M3D-Gq-Luc-expressing cells under the same experimental conditions (Fig. 3, C  and D). This latter observation strongly suggests that the M3D-Gq DREADD lacks the ability to recruit ␤-arrestins in an agonist (CNO)-dependent fashion. On the contrary, using a similar experimental strategy, we recently showed the M3Darr-Luc construct can interact with both V-␤-arrestin-1 and -2 (18).
Detection of DREADD/Arrestin Interactions Via Co-immunoprecipitation-To confirm the inability of M3D-Gq to recruit ␤-arrestins with a different experimental approach, we carried out co-immunoprecipitation studies. Lysates from COS-7 cells co-transfected with HA-tagged versions of the different DREADD constructs and FLAG-tagged ␤-arrestin-1 were treated with an anti-FLAG monoclonal antibody to immunoprecipitate ␤-arrestin-1, followed by immunoblotting studies using an anti-HA antibody to detect co-immunoprecipitated DREADDs.
The anti-HA antibody detected a specific ϳ120-kDa immunoreactive species when cells had been co-transfected with HA-M3D or HA-M3D-arr and the ␤-arrestin-1 construct (Fig.  4A, right panel). The intensity of this band was greatly increased after exposure of cells to CNO (10 M), indicating that complex formation between M3D or M3D-arr and ␤-arrestin-1 was strongly enhanced by DREADD activation. In striking contrast, no specific immunoreactive band was observed after co-transfection of cells with HA-M3D-Gq and the ␤-arrestin-1 construct (Fig. 4A, right panel) or with cell lysates from cells expressing M3D, M3D-Gq, or ␤-arrestin-1 alone (Fig. 4B, right  panel).
Live-cell Imaging of DREADD Internalization-One of the classical roles of ␤-arrestins is to mediate the internalization of activated GPCRs (8,40). To investigate DREADD internalization, we transiently expressed M3D, M3D-Gq, and M3D-arr in HEK293 cells. For comparison, we also studied cells transfected with the WT M3R. All receptor constructs used in these studies carried a C-terminal EYFP tag.
We then used fluorescence microscopy to measure agonistdependent receptor/DREADD internalization in live cells. Prior to agonist simulation, nearly all receptors/DREADDs were expressed on the cell surface (Fig. 5). Following agonist treatment of WT M3R-, M3D-, and M3D-arr-expressing cells, a significant receptor population was internalized (Fig. 5). However, this effect was not observed after agonist (CNO) stimulation of M3D-Gq-expressing cells (Fig. 5). These findings are in good agreement with the notion that the M3D-Gq DREADD is unable to recruit ␤-arrestins.
Expression of DREADDs in Hepatocytes in Vivo Using Recombinant AAVs-To study whether the newly generated M3D-Gq DREADD was also active in vivo, we used a viral strategy to selectively express M3D-Gq in hepatocytes of WT C57BL/ 6NTac mice. Specifically, we generated a recombinant AAV coding for M3D-Gq under the control of the hepatocyte-specific TBG promoter (AAV-TBG-M3D-Gq). Using the same strategy, we also generated AAVs coding for M3D, M3D-arr, and eGFP (for control purposes) (note that all DREADDs contained an N-terminal HA tag). All AAVs (AAV-TBG-M3D, AAV-TBG-M3D-arr, AAV-TBG-M3D-Gq, and AAV-TBG-eGFP) were injected into the tail vein of WT C57BL/6NTac mice.
Two to 3 weeks after AAV injections, we carried out Western blotting experiments to confirm that all DREADDs were selectively expressed in the liver. Fig. 6 clearly shows that all DREADDs gave a strong immunoreactive signal in the liver but not in any of the other tissues investigated. We also carried out [ 3 H]NMS binding assays using membranes prepared from DREADD-expressing primary mouse hepatocytes. These studies showed all DREADDs, including M3D-Gq, were expressed at high levels in mouse hepatocytes ranging from ϳ0.7 to 1 pmol/mg protein (Table 3). A G q/11 -biased Designer GPCR APRIL 8, 2016 • VOLUME 291 • NUMBER 15

CNO Stimulation of M3D-Gq-expressing Primary Mouse
Hepatocytes Leads to Robust Calcium Responses-To examine whether M3D-Gq retained its ability to activate G q -type G proteins in hepatocytes, we isolated primary hepatocytes from mice injected with the AAV-TBG-M3D-Gq virus. In parallel, we also prepared hepatocytes from mice treated with the AAV-TBG-M3D and AAV-TBG-M3D-arr viruses. We then used FLIPR to monitor CNO-mediated increases in [Ca 2ϩ ] i in DREADD-expressing hepatocytes. CNO treatment of M3D-Gqexpressing hepatocytes led to pronounced elevations in [Ca 2ϩ ] i , in the same fashion as observed with M3D-expressing hepatocytes (Table 3; Fig. 7). In contrast, CNO had no significant effect on [Ca 2ϩ ] i in M3D-arr-expressing hepatocytes (Fig. 7). These data indicated that M3D-Gq, similar to M3D, can efficiently couple to G q/11 in endogenous cells (hepatocytes).
Activation of DREADDs Expressed in Hepatocytes In Vivo-It is well known that endogenous GPCRs can regulate hepatic glucose fluxes, resulting in changes in blood glucose levels (41). For example, treatment of mice with glucagon leads to pronounced hyperglycemia, caused by the activation of G s -coupled glucagon receptors that are abundantly expressed by hepatocytes and strongly stimulate hepatic glucose production (42,43). We recently demonstrated that CNO treatment of transgenic mice expressing the M3D DREADD selectively in hepatocytes leads to robust increases in blood glucose levels, due to enhanced hepatic gluconeogenesis and glycogen breakdown (44). To examine whether this effect is mediated exclusively by the activation of hepatocyte G q/11 or whether ␤-arrestin-dependent signaling pathways also contribute to this response, we carried out the following in vivo study. Mice that had been injected with the different DREADD AAVs (AAV-TBG-M3D-Gq, AAV-TBG-M3D, or AAV-TBG-M3D-arr) received a sin-gle injection of CNO (10 mg/kg, intraperitoneal), followed by the monitoring of blood glucose levels over the next 90 min. Strikingly, CNO treatment of mice that had been injected with the M3D-Gq virus led to robust increases in blood glucose levels (Fig. 8). The magnitude and time course of this response was very similar to that observed with mice that had been treated with the M3D virus (Fig. 8). In striking contrast, CNO treatment of mice that had been injected with the M3D-arr virus did not cause significant elevations in blood glucose levels, as compared with mice that had received the control eGFP virus (Fig.  8). Taken together, these data strongly suggest that ␤-arrestindependent signaling pathways do not play a significant role in M3D-induced hyperglycemia.
CNO Treatment of M3D-Gq-expressing Primary Hepatocytes Stimulates Glucose Release-The most likely explanation for the hyperglycemic effects observed after CNO treatment of mice injected with the M3D-Gq virus is that activation of hepatic M3D-Gq receptors promotes hepatic glucose production (HGP). To test this hypothesis in a more direct fashion, we studied HGP using primary hepatocytes prepared from mice that had received the M3D-Gq virus. CNO (10 M) or glucagon (control; 100 nM) treatment of M3D-Gq hepatocytes led to significant increases in HGP (Fig. 9). The magnitude of these effects was similar to that observed with M3D hepatocytes (Fig.  9). The M3D-Gq-mediated increase in HGP could be completely blocked in the presence of the muscarinic antagonist, NMS (100 M; n ϭ 3). CNO treatment of M3D-arr hepatocytes had no significant effect on HGP. These data strongly support the notion that the hyperglycemic effects observed after CNO treatment of mice injected with the M3D-Gq virus are caused by a direct effect on hepatocytes.

Effect of G␣ q/11 Knockdown on M3D-Gq-mediated Hepatic
Glucose Output-To demonstrate the involvement of G q -type G proteins in M3D-Gq-mediated hepatic glucose release, we treated primary mouse hepatocytes expressing the M3D-Gq construct with either scrambled control siRNA or G␣ q/11 siRNA. As expected, we found that the CNO-induced increase in glucose production in M3D-Gq-expressing hepatocytes was significantly reduced after knockdown of G␣ q/11 expression (Fig. 10A). G␣ q/11 siRNA treatment did not completely abolish the CNO/M3D-Gq-mediated glucose response, most likely due to the residual G␣ q expression (Fig. 10B).

M3D-arr Construct Undergoes CNO-induced Internalization in Mouse
Hepatocytes-To confirm that the M3D-arr construct remained functional when expressed in mouse hepatocytes, we incubated M3D-arr expressing hepatocytes for 1 h with or without CNO (10 M). Subsequently, we determined cell surface M3D-arr expression levels by incubation of intact cells with [ 3 H]NMS (10 nM), a membrane-impermeable muscarinic radioligand. These experiments confirmed that the M3D-arr    Table 3. The resulting changes in blood glucose levels were monitored over 90 min. CNO-mediated activation of the M3D and M3D-Gq designer receptors led to robust increases in blood glucose levels. In contrast, CNO treatment of M3D-arr-expressing mice, similar to eGFP control mice, had little or no effect on blood glucose levels. Data are given as means Ϯ S.E. (n ϭ 7-14 mice per group).

Discussion
GPCRs are named after their ability to interact with and activate heterotrimeric G proteins. Ligand-activated GPCRs also bind to members of the arrestin family (␤-arrestin-1 and -2), which disrupt receptor/G protein coupling and promote GPCR internalization via well known processes (7,40). Accumulating evidence indicates that ␤-arrestin-1 and -2, following their recruitment to activated GPCRs, can also act as signaling molecules in their own right. It has become clear during the past 10 -15 years that numerous signaling pathways and cellular processes can be activated or inhibited by ␤-arrestin-1 and/or -2 in a G protein-independent fashion (9 -12, 15, 40). The overall signaling outcome caused by activation of a specific GPCR in a particular cell type is considered a combination of both G protein-and ␤-arrestin (G protein-independent)-mediated signaling events (9 -13). In this study, we focused on the development of a novel, G proteinbiased DREADD as a tool to deconvolute this signaling complexity. As discussed below, a better understanding of the relative contributions of these two branches of GPCR signaling to specific GPCR-mediated physiological responses is of considerable potential clinical importance (9 -11, 14 -17).
Here, we describe the generation and pharmacological properties of a novel G q/11 -biased DREADD (referred to as M3D-Gq) that shows signal properties opposite to those of M3D-arr, a ␤-arrestin-biased DREADD (18). CNO treatment of M3D-Gq leads to the efficient activation of G q -type G proteins but does not promote ␤-arrestin recruitment (Figs. [2][3][4][5]. To demonstrate that the M3D-Gq designer receptor was also functional in vivo, we selectively expressed M3D-Gq in hepatocytes of WT mice (Fig. 6). We found that CNO treatment of the M3D-Gqexpressing mice resulted in pronounced hyperglycemic responses (Fig. 8), most likely due to an increase in HGP, as suggested by studies with isolated hepatocytes (Fig. 9). In contrast, CNO treatment of mice selectively expressing the M3D-arr DREADD in hepatocytes had no significant effect on blood glucose levels (Fig. 8), and addition of CNO to M3D-arr-expressing hepatocytes had no effect on HGP (Fig. 9). These data indicate that the ability of G q -coupled receptors to promote HGP is exclusively dependent on G q/11 -mediated signaling and does not require ␤-arrestin-activated signaling pathways. Hepatocytes express several endogenous G q -coupled receptors, including ␣ 1 -adrenergic and V 1 vasopressin receptors (41). As proposed previously (41), the ability of this class of GPCRs to stimulate glucose release from the liver is most likely due to G q/11 -mediated increases in [Ca 2ϩ ] i , which in turn lead to the activation of enzymes that enhance HGP. Interestingly, we recently demonstrated that blockade of hepatic V 1 receptors is able to significantly ameliorate the deficits in glucose homeostasis observed in a mouse model of obesity (ob/ob mice (44)). Agents that can block G q/11 -mediated signaling in FIGURE 9. DREADD-mediated increases in HGP studied with isolated hepatocytes. Primary hepatocytes were isolated from WT mice (8-week-old males) that had been injected i.v. with recombinant AAV-TBG-DREADD viruses coding for the indicated DREADDs or with a control virus (AAV-TBG-eGFP) encoding eGFP. Hepatocytes were first maintained in glucose-free medium for 16 h and then incubated with CNO (10 M) or glucagon (100 nM; positive control) for 4 h at 37°C. After this time, glucose concentrations were determined in the culture medium. Basal glucose release did not differ significantly among eGFP-and DREADD-expressing hepatocytes (data range: 0.03-0.10 g of glucose/g of protein). Data are given as means Ϯ S.E. of three independent experiments.*, p Ͻ 0.05; **, p Ͻ 0.01, as compared with nontreated cells. FIGURE 10. CNO-induced glucose output in M3D-Gq-expressing hepatocytes is greatly reduced after G␣ q/11 knockdown. Primary mouse hepatocytes expressing the M3D-Gq construct were treated with either scrambled control siRNA or G␣ q/11 siRNA. A, glucose release studies. To stimulate glucose production, hepatocytes were incubated with CNO (10 M) or glucagon (100 nM; positive control) for 4 h at 37°C. After this step, glucose concentrations were determined in the culture medium. B, effective knockdown of G␣ q/11 expression after treatment of M3D-Gq-expressing mouse hepatocytes with G␣ q/11 siRNA. G␣ q/11 transcript levels were determined by quantitative RT-PCR using standard techniques. Data are given as means Ϯ S.E. of three independent experiments.**, p Ͻ 0.01, as compared with the control group.
hepatocytes may therefore prove beneficial clinically for reducing HGP and hyperglycemia in type 2 diabetes, a disease that is characterized by unphysiologically elevated HGP (45,46).
It is well known that ␤-arrestin-mediated signaling regulates long term processes involving various signaling cascades that lead to changes in gene expression (9 -13). Thus, we cannot exclude the possibility that ␤-arrestins modulate hepatic glucose fluxes under different experimental conditions where HGP is monitored for days or weeks.
A previous study has shown that acute intraperitoneal treatment of mice with CNO does not lead to detectable levels of clozapine in the plasma (5). For this reason, it is unlikely that the findings described in this study were affected by back-transformation of CNO to clozapine. However, it should be noted that the metabolic transformation of CNO after chronic administration has not been studied systematically in mice.
Although the viral delivery approach that we employed in this study resulted in relatively high DREADD expression levels, the observed functional responses were mediated by endogenous G proteins and other signaling proteins. Also, the measured DREADD expression levels are not out of the physiological range because several GPCRs, including different muscarinic receptor subtypes (47), are expressed at comparable levels in other tissues.
The generation and analysis of G protein-and ␤-arrestin-biased DREADDs is highly relevant to recent advances in the development of G protein-and ␤-arrestin-biased agonists (11,14,15,17,48). Various lines of evidence suggest that the potential clinical use of functionally biased agonists may be associated with increased therapeutic efficacy and reduced side effects, at least in certain clinical conditions (11,14,17,48). For example, several studies suggest that G protein-biased -opioid receptor agonists may be able to relieve pain without causing the characteristic side effects (tolerance, addiction, etc.) associated with the use of morphine and other non-biased -receptor agonists (11). Similarly, G proteinbiased -opioid receptor agonists may prove clinically useful as novel analgesic agents endowed with fewer side effects such as sedation, motor incoordination, and anhedonia (49).
The outcome of physiological studies using G protein-and ␤-arrestin-biased DREADDs should prove particularly useful in guiding the search for novel biased ligands with therapeutic potential. The great advantage of DREADD technology is that it allows the activation of distinct signaling pathways in a particular cell type in vivo. As a consequence, such DREADD studies offer direct insights into the cellular basis of an observed physiological or behavioral response. In contrast, the situation is considerably more complex in the case of G protein-and ␤-arrestin-biased ligands, which after their systemic administration usually act on target receptors expressed in multiple tissues or cell types. Thus, DREADD technology clearly complements ongoing efforts in the "biased GPCR ligand" field, offering the opportunity of identifying novel targets for the development of G protein-and ␤-arrestin-biased ligands.