Chemogenetic approaches to identify metabolically important GPCR signaling pathways: therapeutic implications

Abstract

DREADDs (Designer Receptors Exclusively Activated by a Designer Drug) are designer G protein-coupled receptors (GPCRs) that are widely used in the neuroscience field to modulate neuronal activity. In this review, we will focus on DREADD studies carried out with genetically engineered mice aimed at elucidating signaling pathways important for maintaining proper glucose and energy homeostasis. The availability of muscarinic receptor-based DREADDs endowed with selectivity for one of the four major classes of heterotrimeric G proteins (G_s, G_i, G_q, and G₁₂) has been instrumental in dissecting the physiological and pathophysiological roles of distinct G protein signaling pathways in metabolically important cell types. The novel insights gained from this work should inform the development of novel classes of drugs useful for the treatment of several metabolic disorders including type 2 diabetes and obesity.

G protein-coupled receptors (GPCRs) are cell surface receptors that are activated by extracellular ligands such as neurotransmitters, hormones, paracrine factors, and many other stimuli (Pierce et al., 2002). In humans, the GPCR superfamily consists of more than 800 distinct members (Sriram and Insel, 2018). Importantly, GPCRs are the target of a very large number of therapeutic agents (Sriram and Insel, 2018). Each individual GPCR interacts with one or more of the four major classes of heterotrimeric G proteins (G_s, G_i, G_q, and G₁₂) which are named after the structural and functional properties of their α subunits (Wettschureck and Offermanns, 2005). The α subunit of heterotrimeric G proteins is tightly bound to a βγ complex in the inactive state of the protein. Receptor-activated free α subunits stimulate or inhibit distinct intracellular signaling pathways, but free βγ complexes can also modulate the activity of certain ion channels and effector proteins (Wettschureck and Offermanns, 2005).

Gene expression studies have shown that virtually each body cell expresses dozens of GPCRs that can modulate cellular functions by activating or inhibiting distinct intracellular signaling cascades (Regard et al., 2008). With almost no exception, each specific GPCR is expressed in multiple tissues and cell types (Regard et al., 2008). Moreover, highly selective ligands (agonists or antagonists) are not available for a very large number of GPCRs. For these reasons, identifying the in vivo metabolic roles of a particular GPCR expressed by a specific cell type represents a very difficult task.

General features of DREADDs (Designer Receptors Exclusively Activated by a Designer Drug)

In a pioneering study, Armbruster et al. (Armbruster et al., 2007) demonstrated that introduction of a pair of point mutations into the transmembrane core of different muscarinic acetylcholine (ACh) receptor subtype (M₁-M₄) rendered the resulting mutant receptors insensitive to activation by ACh or other endogenous ligands. However, by taking advantage of a yeast generic screen, the Roth lab identified a synthetic compound called clozapine-N-oxide (CNO) (Fig. 1) that was able to activate these mutant receptors with high potency and efficacy. Importantly, when used in the proper concentration/dose range, CNO is pharmacologically inert. For these reasons, Armbruster et al. (Armbruster et al., 2007) referred to these mutant muscarinic receptors as DREADDs (Designer Receptors Exclusively Activated by a Designer Drug). Two of these DREADDs, GqD and GiD (Fig. 2), are now widely used in the neuroscience field to activate or silence distinct sets of neurons, respectively (Rogan and Roth, 2011; Roth, 2016). However, it should be noted that treatment of mice with high doses of CNO (e.g. ≥10 mg/kg i.p) may cause off-target effects (Martinez et al., 2019).

FIGURE 1.

Chemical structures of CNO, clozapine, C21, and DCZ. All four agents can activate muscarinic receptor-based DREADDs with high potency and efficacy. However, clozapine acts on many other high-affinity targets in the CNS. Several studies have shown that CNO can be metabolized to clozapine. C21 and DCZ are CNO-derived, novel DREADD agonists that are not converted to clozapine in vivo (Chen et al., 2015; Nagai et al., 2020; Thompson et al., 2018). See text for details.

FIGURE 2.

Coupling properties of muscarinic receptor-based DREADDs. The structural features of the various DREADDs are described in the main body of the text. The four DREADDs shown here harbor the same two point mutations in the transmembrane core (Y->C^3.33 and A->G^5.46; Ballesteros-Weinstein GPCR nomenclature) (Armbruster et al., 2007; Guettier et al., 2009; Inoue et al., 2019). While acetylcholine, the endogenous muscarinic receptor agonist, shows little or no activity at these DREADDs, they can be activated by CNO with high potency and efficacy. The different G proteins are named after the α-subunit contained within the G protein heterotrimer. Activated Gα subunits regulate the activity of distinct intracellular signaling pathways, but free βγ complexes, which are released after G protein activation, can also modulate cellular functions. Gα proteins are grouped into four major subfamilies (G_s: α_s, α_olf; G_i: α_i1, α_i2, α_i3, α_o, etc.; G_q: α_q, α₁₁; G₁₂: α₁₂, α₁₃). Abbreviations: Epac, exchange protein activated by cAMP; GIRK, G-protein-regulated inward-rectifier potassium channel; LARG, Leukemia-Associated RhoGEF; PLC, phospholipase C; PKA, protein kinase A; PKC, protein kinase C; RhoGEF, Rho guanine nucleotide exchange factor; ROCK; Rho-associated coiled-coil-containing protein kinase; VDCC, voltage-dependent Ca^2+-channel.

The general concept underlying the development of DREADD technology dates back to earlier studies reporting the generation of so-called RASSLs (Receptors Activated Solely by Synthetic Ligands) (Conklin et al., 2008; Coward et al., 1998). However, the in vivo use of these first-generation designer GPCRs is complicated by the fact that RASSLs can still be activated by endogenous ligands and show strong constitutive activity when expressed in mutant mice (Conklin et al., 2008; Coward et al., 1998).

The M₃ muscarinic receptor-derived DREADD, known as hM3Dq or simply GqD, is coupled to G_q-type G proteins (Armbruster et al., 2007; Wess et al., 2007) (Fig. 2). As a result, CNO binding to GqD leads to the activation of different isoforms of phospholipase Cβ, resulting in the breakdown of phosphatidylinositol and the generation of the second messengers, inositol 1,4,5- trisphosphate (IP₃) and diacylglycerol (DAG). IP₃ releases Ca²⁺ from intracellular stores, leading to elevated intracellular Ca²⁺ levels which regulate the activity of many intracellular signaling pathways. DAG activates different isoforms of PKC which also modulate the function of many important signaling proteins including ERK (Fig. 2). The GqD-mediated increase in intracellular Ca²⁺ levels is considered a key step that causes neuronal excitation (Rogan and Roth, 2011; Roth, 2016). It remains to be elucidated whether ligand activation of GqD leads to the activation of all G proteins of the G_q family (α_q, α₁₁, α₁₄, and α_15/16).

The second DREADD that is now extensively used in neuroscience research is derived from the M₄ muscarinic receptor which is selectively linked to G proteins of the G_i family (Armbruster et al., 2007; Wess et al., 2007). CNO-induced activation of this DREADD, referred to as hM4Di or simply as GiD (Fig. 2), leads to the inhibition of adenylyl cyclase and the opening of G protein-coupled inwardly-rectifying potassium channels (GIRK) channels (Armbruster et al., 2007; Wess et al., 2007). This latter effect is thought to be responsible for the ability of activated GiD receptors to induce neuronal silencing. It needs to be explored whether CNO-mediated activation of GiD causes the activation of all members of the G_i/o family (α_i1, α_i2, α_i3, α_o, α_z, α_t, and α_gust).

Subsequently, Guettier et al. (Guettier et al., 2009) reported the generation and pharmacologic properties of a G_s-coupled DREADD (GsD) (Fig. 2). The GsD construct was created by replacing the second and third intracellular loops of GqD with the corresponding β₁-adrenoceptor sequences (Guettier et al., 2009). As expected, CNO treatment of cells expressing the GsD designer receptor resulted in increased cAMP levels (Guettier et al., 2009). In a very recent study, Inoue et al. (Inoue et al., 2019) described the development of the first G₁₂-linked DREADDs (G12D) (Fig. 2). Structurally, these new designer receptors have a GqD backbone in which the third intracellular loops are derived from GPR132 or GPR183. Somewhat surprisingly, these G12 DREADDs do not activate G₁₃ (Inoue et al., 2019). For this reason, these new designer receptors are particularly useful to study the consequences of G₁₂ signaling in vivo.

The generation of mutant mice expressing a particular DREADD in a specific cell type has made it possible to study the in vivo consequences of activating distinct G protein signaling pathways in distinct cell types in vivo (Wess, 2016; Wess et al., 2013). As discussed above, such studies are difficult to perform by using classical pharmacological tools, since the expression of endogenous GPCRs is usually not restricted to a specific cell type (Regard et al., 2008). In this review, we will discuss DREADD studies that have shed light onto metabolic roles of activating specific classes of G proteins in distinct cell types known to be critical for glucose and energy homeostasis. The new insights gained from this research should stimulate the development of novel classes of drugs useful for the treatment of type 2 diabetes (T2D) and related metabolic disorders.

Novel tools useful for in vivo applications of DREADD technology

Novel DREADD ligands derived from CNO

Since the development of the first muscarinic receptor-based DREADDs more than a decade ago, CNO (Fig. 1) has become the agonist of choice for activating this class of designer GPCRs. Guettier et al. (Guettier et al., 2009) showed that acute treatment of mice with CNO (1 mg/kg i.p.) did not lead to a significant conversion of CNO to clozapine (Fig. 1) during a 2 hr observation period (clozapine is a potent psychoactive drug). However, several studies have shown that CNO can be back-transformed to clozapine in many species including humans and monkeys under different experimental conditions (Chang et al., 1998; Gomez et al., 2017; Jann et al., 1994; Manvich et al., 2018; Raper et al., 2017). Interestingly, clozapine is ~10-fold more potent than CNO in activating muscarinic receptor-based DREADDs (Armbruster et al., 2007). However, clozapine, which is in clinical use as an antipsychotic drug, affects many functions of the CNS by binding with high affinity to numerous cellular targets (Yadav et al., 2011). For this reason, DREADD studies involving the use of CNO should ideally always be conducted with four groups of experimental animals (DREADD animals treated with CNO or saline; wild-type (WT)/control animals treated with CNO or saline). A review of past DREADD studies indicates that one or more of these control groups were not consistently included in the experimental design. The use of all four control groups is of particular importance when high doses of CNO are administered. It should also be noted that CNO metabolism after acute vs. chronic CNO administration has not been studied systematically so far.

To prevent possible effects mediated by the conversion of CNO to clozapine, recent efforts (Chen et al., 2015; Thompson et al., 2018) led to the development of a CNO derivative, referred to as compound 21 (C21) (Fig. 1). This agent activates muscarinic receptor-based DREADDs with high efficacy and potency, shows very good bioavailability, and is not metabolized to clozapine. For this reason, C21 can be employed as an alternative to CNO for studies where metabolic conversion of CNO to clozapine is a major cause of concern.

Following systemic administration, the uptake of CNO into the brain has been shown to be relatively slow (Nagai et al., 2020). Interestingly, Nagai et al. (Nagai et al., 2020) recently reported that deschloroclozapine (DCZ) (Fig. 1), a clozapine derivative, represents an extremely potent DREADD agonist that quickly enters the brain and shows high biostability. Studies with mice and monkeys demonstrated that DCZ rapidly (within 10 min after injection) activates centrally expressed DREADDs (GqD or GiD), without detectable off-target effects (Nagai et al., 2020). These data suggest that DCZ represents a highly useful, novel agonist at muscarinic receptor-based DREADDs that is selective, stable, and has a fast onset of action. We recently demonstrated that very low doses of DCZ (e.g. 10 μg/kg i.p.) are also able to efficiently activate DREADDs expressed in peripheral tissues or cell types in mutant mice (J. Wess et al., unpublished data).

Fig. 1 shows that C21 and DCZ are structurally almost identical with clozapine, except for the lack of a chloro atom (C21 and DCZ) and a methyl moiety (C21). These modifications prevent back-metabolism to clozapine and greatly reduce the affinity of clozapine for its many CNS target proteins/receptors, while maintaining high brain penetrance (Chen et al., 2015; Nagai et al., 2020; Thompson et al., 2018).

Non-muscarinic receptor-based DREADDs

During the past decade, additional DREADDs containing different GPCR backbones have also been developed. Several years ago, Vardy et al. (Vardy et al., 2015) described the generation and pharmacological characterization of a κ-opioid-derived DREADD (KORD) that is selectively activated by salvinorin B, a pharmacologically inert compound. However, salvinorin B is able to bind to the κ-opioid receptor with modest affinity (>100 nM) (Vardy et al., 2015). For this reason, salvinorin B should be used at low doses that can activate the KORD designer receptor but do not cause any undesired side effects via interaction with WT k-opioid receptors. Since this new DREADD is activated by a ligand other than CNO, it allows the regulation of neuronal activity and/or behavior in a multiplexed and bidirectional fashion. For example, co-expression of the GqD and KORD designer receptors made it possible to activate (via CNO/GqD) and inhibit (via salvinorin B/KORD) neuronal activity in a sequential fashion (Vardy et al., 2015).

Moreover, Hudson et al. (Hudson et al., 2012) succeeded in generating a DREADD variant of the free fatty acid receptor 2 (FFA2). The WT FFA2 is activated by short chain fatty acids (SCFAs) and couples to both G_q- and G_i-type proteins (Bolognini et al., 2016). The FFA2 DREADD, which shows a similar coupling profile, does no longer respond effectively to SCFAs but can be selectively activated by several other ligands, including sorbic acid (Hudson et al., 2012). Recently, Bolognini et al. (Bolognini et al., 2019) used a FFA2-DREADD knock-in mouse strain to study the physiological roles of FFA2 in the gut and white adipose tissue. Analysis of this novel mutant mouse strain demonstrated that DREADD knock-in mice can serve as highly useful tools to elucidate the physiological functions and therapeutic potential of GPCRs for which highly selective ligands are not available (Bolognini et al., 2019).

Mouse models that allow the expression of DREADDs in a conditional fashion

The recent development of a new series of DREADD mutant mice has made it possible to express various DREADDs (GqD, GiD, and GsD) in specific cell types or tissues in a Cre-dependent fashion. Zhu et al. (Zhu et al., 2016) first generated mutant mouse strains harboring the GqD or GiD sequences under the control of a strong ubiquitous promoter, which is separated from the DREADD sequences by a stop signal that is flanked by loxP sites. Using an analogous approach, Akhmedov et al. (Akhmedov et al., 2017a) developed a mutant mouse strain in which the expression of GsD requires the presence of Cre recombinase. In all these new mouse strains, the DREADD construct (preceded by the promoter and floxed stop sequences) was inserted into the ROSA26 locus. When these mice are crossed with specific Cre driver lines, Cre recombinase removes the stop signal, leading to DREADD expression in Cre-expressing cells. This approach allows the conditional expression of specific DREADDs in a cell type-specific fashion.

DREADDs that display signaling bias

As is the case with most GPCRs, is likely that ligand-activated DREADDs do not only interact with heterotrimeric G proteins but can also recruit β-arrestins (βarr1 and βarr2). This has been clearly demonstrated for the CNO-activated GqD designer receptor (Alvarez-Curto et al., 2011; Hu et al., 2016; Nakajima and Wess, 2012). The two β-arrestins are known to play a key role in receptor desensitization and internalization (Pierce and Lefkowitz, 2001). However, many studies suggest that βarr1 and βarr2 can also act as signaling proteins in their own right (Beaulieu et al., 2009; DeFea, 2011; Gurevich and Gurevich, 2014; Luttrell and Gesty-Palmer, 2010; Rajagopal et al., 2010; Shukla et al., 2011). For this reason, the in vivo consequences of activating a particular DREADD in a specific cell type or tissue may not only depend on activated G proteins but may also involve β-arrestin-dependent signaling. To assess the relative importance of these two signaling branches, mutant mice that selectively lack specific G protein α-subunits or β-arrestins in DREADD-expressing cells would prove highly valuable tools. Accumulating evidence indicates that the use of so-called ‘biased’ GPCR agonists which initiate signaling preferentially through either G proteins or β-arrestins may prove advantageous in a number of pathophysiological conditions (Gurevich and Gurevich, 2019; Kenakin and Christopoulos, 2013; Luttrell et al., 2015; Rajagopal et al., 2010; Shukla et al., 2011; Smith et al., 2018; Wootten et al., 2018). This concept is based on findings that the therapeutic effect of a particular GPCR agonist may arise from activating one signaling pathway, whereas the other signaling branch may mediate undesirable side effects.

The development of two mutationally modified versions of GqD, M3D-Gq and M3D-arr, which display biased coupling properties (Hu et al., 2016; Nakajima and Wess, 2012), may prove helpful to explore to which extent β-arrestins contribute to DREADD-mediated responses in vivo. The ligand-activated M3D-Gq designer receptor promotes signaling via G_q-type G proteins but does cause the recruitment of β-arrestins (Hu et al., 2016). On the other hand, the M3D-arr DREADD shows the opposite activity profile (Hu et al., 2016; Nakajima and Wess, 2012).

Caveats

Chronic CNO treatment of mutant mice expressing GqD in pancreatic β-cells (Jain et al., 2013) or GsD and GiD in adipocytes (Wang et al., 2019; Wang et al., 2020) did not lead to a waning of DREADD-mediated effects over time. However, additional studies are required to confirm whether this is also true for other cell types and other physiological/behavioral responses. It should also be noted DREADD-based experimental strategies mimic drug effects on target GPCRs, but that other experimental approaches (e.g., optogenetics) need to be applied when a more precise temporal control of cellular signaling is required. At present, little is known about the potential impact of DREADD overexpression on the expression of endogenous receptors or signaling molecules. Surprisingly, expression of GiD in sensory C-fibers of mice disrupted endogenous inhibitory GPCR expression and second-messenger coupling, most likely due to CNO-independent signaling of the GiD in dorsal root ganglion (DRG) neurons (Saloman et al., 2016). For this reason, experimental data obtained with systems where DREADDs display constitutive activity need to be interpreted with particular caution.

Metabolically important signaling pathways identified by using DREADD technology

In the following, we will review studies that employed DREADD technology to elucidate the metabolic roles of distinct GPCR signaling pathways operative in metabolically important cell types in regulating whole-body glucose and energy homeostasis. The focus will be on work carried out with three muscarinic receptor-based DREADDs (GqD, GsD, and GiD). The novel information resulting from this research should be of considerable translational usefulness for designing new pharmacological strategies for the treatment of major metabolic disorders, including T2D and obesity.

Pancreatic β- and α-cells

Pancreatic islets are composed of multiple cell types, including β- and α-cells which synthesize and release the hormones insulin and glucagon, respectively. These hormones play essential roles in maintaining euglycemia and modulate many other important metabolic functions. T2D is characterized by impaired β-cell function, leading to insufficient insulin release to overcome peripheral insulin resistance (Hudish et al., 2019; Roden and Shulman, 2019). For this reason, drugs capable of improving impaired β-cell function have great clinical potential for the management of T2D (Hudish et al., 2019; Roden and Shulman, 2019). Different experimental approaches, including RNA-seq studies, have demonstrated that pancreatic β-cells express numerous GPCRs that are linked to different G protein signaling pathways (Ahren, 2009; Amisten et al., 2013; Riddy et al., 2018; Xin et al., 2016).

β-cell-specific GsD mice

To examine the metabolic importance of receptor-mediated activation of G_s in β-cell function and whole-body glucose homeostasis, Guettier et al. (Guettier et al., 2009) generated and analyzed mice that selectively expressed the GsD construct in pancreatic β-cells (β-GsD mice) (Fig. 3a).

FIGURE 3.

Summary of in vivo ‘DREADD studies’ investigating signaling pathways regulating glucose and energy homeostasis. This cartoon summarizes data obtained with mutant mice expressing DREADDs in pancreatic β- and α-cells (a), adipocytes (b), hepatocytes (c), skeletal muscle cells (d), and AgRP (e) and POMC neurons (f) of the arcuate nucleus of the hypothalamus. For simplicity, only some key signaling factors/pathways and metabolic phenotypes are listed. See text for details.

Acute CNO treatment of β-GsD mice led to a pronounced increase in insulin secretion, improved glucose tolerance, and several other beneficial metabolic effects. In β-GsD mice, the GsD receptor also showed some degree of constitutive activity (Guettier et al., 2009). Interestingly, expression of the GsD receptor in mouse hepatocytes in vivo also resulted in a significant degree of CNO-independent signaling (Akhmedov et al., 2017b). However, when expressed in other cell types, including adipocytes (Wang et al., 2019) and AgRP neurons (Nakajima et al., 2016), the GsD receptor did not exhibit constitutive activity. It is likely that actual DREADD expression levels, together with cell type-specific factors, are responsible for these seemingly discrepant findings.

The glucagon-like peptide-1(GLP-1) receptor is a G_s-coupled receptor (like GsD) that is endogenously expressed at relatively high levels by pancreatic β-cells (Müller et al., 2019). As a result, GLP-1 receptor agonists, such as exenatide or semaglutide, have emerged as a very promising class of novel antidiabetic agents (Chia and Egan, 2020; Drucker, 2018). The same is true for inhibitors of dipeptidyl peptidase 4 (DPP4 inhibitors or gliptins; e.g. sitagliptin) which interfere with the degradation of GLP-1, glucose-dependent insulinotropic polypeptide (GIP), and several other bioactive peptides (Drucker and Nauck, 2006; Müller et al., 2019).

While stimulation of β-cell G_s signaling causes multiple beneficial metabolic effects, mice selectively lacking Gα_s in pancreatic β-cells (β-Gs-KO mice) showed profound metabolic deficits (Xie et al., 2007). β-Gs-KO mice displayed severe hypoinsulinemia, hyperglycemia, and glucose intolerance. Moreover, the lack of β-cell Gα_s led to greatly reduced islet insulin content, associated with a marked reduction in β-cell mass (Xie et al., 2007). These findings further highlight the importance of β-cell G_s signaling in β-cell function and whole-body glucose homeostasis.

β-cell-specific GqD mice

The development of a mutant mouse strain selectively expressing the GqD construct in pancreatic β-cells (β-GqD mice) made it possible to study the metabolic outcome of selectively stimulating G_q signaling in β-cells in vivo (Guettier et al., 2009; Jain et al., 2013) (Fig. 3a). Acute CNO treatment of β-GqD mice strongly promoted insulin secretion and led to a dramatic improvement in glucose tolerance (Guettier et al., 2009). Chronic CNO treatment of β-GqD mice caused a significant increase in pancreatic insulin content and β-cell mass, associated with the up-regulation of many genes critical for β-cell function and proliferation (Jain et al., 2013). In addition, chronically enhanced β-cell G_q signaling greatly improved glucose homeostasis in mouse models of diabetes, most likely due to G_q-mediated activation of insulin receptor substrate 2 (IRS2) signaling (Jain et al., 2013). In agreement with these findings, the G_q-coupled free fatty acid receptor 1 (FFAR1), which is expressed at relatively high levels by pancreatic β-cells (Tomita et al., 2014), is considered a promising target for the development of novel antidiabetic drugs (Li et al., 2020).

Interestingly, a recent study (Makhmutova et al., 2020) analyzing mice expressing GqD selectively in β-cells demonstrated that β-cells can communicate with vagal sensory neurons, most likely using serotonin as a signaling molecule. Since serotonin is co-released with insulin from β-cells, it may act as a reporter of the secretory state of β-cells via vagal afferent nerves.

As expected, mutant mice that lacked both Gα_q and Gα₁₁ selectively in pancreatic β-cells (β-Gq/11-KO mice) displayed impaired glucose tolerance and insulin secretion (Sassmann et al., 2010). Somewhat surprisingly, the ability of glucose to stimulate insulin release was also reduced in these mutant mice. The authors demonstrated that this deficit was caused by the loss of β cell-autonomous potentiation of insulin secretion through factors co-secreted with insulin (Sassmann et al., 2010). Such factors include uridine diphosphate which acts on G_q-coupled P2Y₆ receptors and extracellular calcium that stimulates β-cell calcium-sensing receptors. These observations further highlight the metabolic importance of G_q-dependent signaling pathways operative in pancreatic β-cells.

α-cell-specific GiD mice

Pancreatic α-cells store and release glucagon which functions to maintain euglycemia and to counteract the actions of insulin on the liver and other tissues (Ahren, 2015; Quesada et al., 2008; Unger and Cherrington, 2012). It is well documented that α-cell dysfunction contributes to deficits in glucose homeostasis in diabetic individuals (Ahren, 2015; Lee et al., 2016; Quesada et al., 2008; Wendt and Eliasson, 2020). Like pancreatic β-cells, α-cells express many GPCRs that differ in their G protein coupling properties (Ahren, 2009; Amisten et al., 2013; Riddy et al., 2018; Xin et al., 2016).

To investigate the in vivo metabolic consequences of simulating G_i signaling in α-cells, Zhu et al. (Zhu et al., 2019) analyzed mutant mice that express the GiD construct selectively in α-cells (⍺-GiD mice) (Fig. 3a). Interestingly, acute CNO treatment of ⍺-GiD mice led to profound reductions in both plasma glucagon and insulin levels. CNO-treated ⍺-GiD mice also showed impairments in glucose-stimulated insulin secretion (GSIS) and glucose tolerance, supporting the concept that glucagon released from pancreatic α-cells can promote insulin secretion (Zhu et al., 2019).

In agreement with two recent studies (Capozzi et al., 2019a; Svendsen et al., 2018), Zhu et al. (Zhu et al., 2019) also showed that intraislet glucagon primarily acts on β-cell GLP-1 receptors to stimulate insulin secretion in a paracrine fashion. Since glucagon elevates blood glucose levels by stimulating hepatic glucose production (HGP) under fasting conditions (Ali and Drucker, 2009), the ability of intraislet glucagon to promote insulin release in the presence of glucose suggests that glucagon has opposing physiological functions depending on the nutritional state of the organism (Capozzi et al., 2019b). In the light of these new findings, the development of therapeutic approaches aimed at suppressing glucagon signaling to treat hyperglycemia needs to be reassessed.

L-cells

L-cell-specific GqD mice

L-cells are enteroendocrine cells that are primarily found in the ileum and large intestine (colon). Following food intake, L-cells secrete the hormones GLP-1 and peptide YY (PYY), as well as several other bioactive peptides (Gribble and Reimann, 2016; Spreckley and Murphy, 2015). GLP-1 exerts multiple biological actions, including stimulation of insulin release from pancreatic β-cells, delayed gastric emptying, and reduced appetite. PYY also causes anorectic effects and reduces gastrointestinal motility. L-cell function is regulated by the activity of a considerable number of GPCRs (Gribble and Reimann, 2016).

Recently, Lewis et al. (Lewis et al., 2020) generated a mutant mouse line that expressed the GqD designer receptor in an inducible fashion selectively in mouse distal colonic L cells. Interestingly, CNO treatment of these mutant mice led to a significant increase in plasma GLP-1 and PYY levels, improved glucose tolerance, and decreased food intake. These data suggest that agonists that can act on G_q-coupled receptors expressed by colonic L-cells may prove useful to improve impaired glucose homeostasis under pathophysiological conditions. Such receptors may include FFAR1 and the AT_1a angiotensin II and V_1b vasopressin receptor subtypes (Billing et al., 2018).

Adipocytes

Adipocytes play a key role in regulating glucose and energy homeostasis under physiological and pathophysiological conditions including T2D (Guilherme et al., 2008). In obesity, adipose tissue is infiltrated by macrophages, resulting in the secretion of inflammatory adipokines and other factors that result in impaired glucose homeostasis (Gonzalez-Muniesa et al., 2017; Guh et al., 2009; Kusminski et al., 2016; Saltiel and Olefsky, 2017). It is well documented that adipocytes express many GPCRs endowed with different coupling properties (Amisten et al., 2015; Regard et al., 2008). While some of these receptors have been studied in considerable detail (e.g. β-adrenergic receptors), the metabolic functions of most of these GPCRs have not been studied systematically.

Adipocyte-specific GsD mice

To explore the in vivo metabolic consequences of selectively activating G_s signaling in adipocytes, two groups (Caron et al., 2019; Wang et al., 2019) recently studied mutant mice that expressed the GsD construct selectively in adipocytes (adipo-GsD mice) (Fig. 3b). Both studies demonstrated that acute CNO injection of lean or obese adipo-GsD mice resulted in increased lipolysis, elevated plasma free fatty acid (FFA) and plasma insulin levels, decreased blood glucose levels, and improved glucose tolerance (Caron et al., 2019; Wang et al., 2019). The beneficial effects on glucose homeostasis were most likely caused by increased insulin release caused by elevated plasma FFA levels. Consistent with this notion, it has been shown that FFAs can promote insulin secretion by activating β-cell FFAR1 receptors (Alquier et al., 2009; Hauke et al., 2018; Schnell et al., 2007). While Caron el al. (Caron et al., 2019) only reported the outcome of acute CNO administration studies, Wang et al. (Wang et al., 2019) also examined the metabolic consequences of chronic CNO exposure. Interestingly, chronic CNO treatment of adipo-GsD mice maintained on a high-fat diet (HFD) significantly slowed the development of adiposity, resulting in multiple metabolic benefits including improved glucose tolerance, increased whole-body energy expenditure, and reduced food intake (Wang et al., 2019). The reduced adiposity displayed by CNO-treated adipo-GsD mice maintained on an obesogenic diet is most likely due to a combination of enhanced energy expenditure and reduced appetite. Chemogenetic activation of G_s signaling also enhanced the thermogenic activity of both brown and white fat, a response predicted to increase energy expenditure (Wang et al., 2019). At present, it remains unclear which signaling pathways and molecules play a role in reducing food intake in CNO-treated adipo-GsD mice.

In agreement with the metabolic phenotypes displayed by CNO-treated adipo-GsD mice, mutant mice lacking Gα_s selectively in adipocytes (adipo-Gs-KO mice) showed impaired brown adipose tissue (BAT) function and reduced lipolysis (Li et al., 2016). However, adipo-Gs-KO mice displayed unchanged body weight and energy expenditure, but improved insulin sensitivity and glucose metabolism, possibly due to reduced plasma FFA levels. One possible explanation for the observation that both adipo-Gs-KO mice and CNO-treated adipo-GsD mice showed beneficial metabolic phenotypes may be due to compensatory changes in adipo-Gs-KO mice that lack Gα_s throughout development.

The metabolic effects displayed by CNO-treated adipo-GsD mice are similar to those observed after treatment of mice or rats with β3-adrenergic receptor (β3-AR)-selective agonists such CL316,243 (Collins and Surwit, 2001; Ghorbani and Himms-Hagen, 1997; Himms-Hagen et al., 1994; Susulic et al., 1995; Xiao et al., 2015). These effects have been shown to be mediated by activation of G_s-coupled adipocyte β3-ARs (Gavrilova et al., 2000; Grujic et al., 1997) which are abundantly expressed in mouse adipocytes (Collins et al., 1999).

In contrast, β3-ARs are expressed at relatively low levels in human adipocytes where the β1- and β2-ARs are the predominant β-AR subtypes (Robidoux et al., 2004). Despite this altered β-AR expression pattern, acute treatment of healthy male subjects with mirabegron (200 mg orally), a selective agonist at human β3-ARs (Igawa and Michel, 2013), enhanced basal metabolic rate (Cypess et al., 2015). More recently, O’Mara et al. (O’Mara et al., 2020) reported that chronic mirabegron treatment (100 mg per day for 8 weeks) of healthy female subjects stimulated BAT metabolic activity and resting energy expenditure, increased insulin sensitivity, and improved glucose tolerance, suggesting that selective β3-AR agonists might prove useful as novel antidiabetic drugs.. However, two recent studies indicated that the mirabegron-induced metabolic improvements may involve mechanisms that are independent of β3-AR-mediated BAT activation. Finlin et al. (Finlin et al., 2020) reported that treatment of obese, insulin-resistant individuals with mirabegron (50 mg per day for 12 weeks) causes improved glucose homeostasis and that this effect is most likely due to enhanced skeletal muscle oxidative capacity and improved β-cell function (BAT activity was unchanged). These beneficial metabolic effects appear to be secondary to mirabegron’s ability to decrease adipose inflammation and remodeling (Finlin et al., 2020). Another recent study demonstrated that oral administration of mirabegron stimulated BAT thermogenesis only at the maximal allowable dose (200 mg) which also led to the activation of β1- and β2-ARs (Blondin et al., 2020). The authors reported that this high dose of mirabegron causes cardiovascular side effects and stimulates lipolysis in white adipocytes and that mirabegron-induced lipolysis and thermogenesis in BAT are most likely mediated by activation of β2-ARs (Blondin et al., 2020). Cardiovascular side effects following a high dose of mirabegron have also been observed in at least two other studies (Cypess et al., 2015; Malik et al., 2012). Clearly, additional studies are needed to reconcile the seemingly discrepant findings obtained with mirabegron in human clinical trials. Differences in the mirabegron doses used, length of drug application, and the metabolic status of the individuals included in the various studies are all likely factors that may contribute to the different outcomes of these studies.

Adipocyte-specific GiD mice

Adipocytes also express many G_i-coupled receptors (Regard et al., 2008; Wang et al., 2020). In order to explore the in vivo metabolic outcome of selectively activating G_i signaling in adipocytes, Wang et al. (Wang et al., 2020) recently developed and analyzed mice that expressed the GiD designer receptor selectively in adipocytes (adipo-GiD mice) (Fig. 3b). CNO-injected adipo-GiD mice displayed robust reductions in plasma FFA, glycerol, and triglyceride levels, indicating that G_i-mediated signaling inhibits lipolysis in adipocytes. Moreover, independent of the diet that the mice consumed (regular chow or HFD), acute or chronic CNO treatment of adipo-GiD mice led to improved glucose tolerance and enhanced insulin sensitivity, most likely secondary to reduced plasma FFA levels (Wang et al., 2020).

In contrast to the observations by Wang et al. (Wang et al., 2020), a related study (Caron et al., 2019) reported that CNO treatment did not affect FFA release from adipose explants derived from adipo-GiD mice. One possible explanation for these discrepant findings is that chemogenetic activation of G_i signaling is more efficient in vivo where adipose tissue function is regulated by many other factors and neuronal pathways.

On the other hand, mice that lacked functional G_i-type proteins in adipocytes (adipo-Gi-KO mice) showed significant deficits in glucose tolerance and insulin sensitivity (as compared to control littermates) when maintained on a HFD. More detailed studies revelated that increased plasma FFA levels caused by enhanced lipolysis in adipocytes greatly reduced the responsiveness of peripheral tissues to insulin (Wang et al., 2020). In agreement with this finding, previous studies have shown that increased plasma FFA levels can trigger insulin resistance and inflammation, explaining why obesity is closely linked to insulin resistance (Boden, 2008). Taken together, these data convincingly demonstrate that adipocyte G_i signaling plays a central role in regulating lipolysis and whole-body glucose homeostasis.

Adipocytes express several endogenous G_i-coupled receptors, including the CB₁ cannabinoid receptor, hydroxycarboxylic acid receptors-1 and −2 (HCA_1/GPR81 and HCA₂ /GPR109A), and various chemokine and orphan receptors (Regard et al., 2008; Wang et al., 2020). Based on the studies discussed above, these receptors represent potential targets for reducing plasma FFA levels to improve glucose homeostasis in T2D.

Adipocyte-specific expression of GqD

A recent study (Klepac et al., 2016) reported that CNO treatment of GqD-expressing murine brown and white adipocytes inhibited adipocyte differentiation. The authors also found that endothelin acting on G_q-coupled ET_A endothelin receptors functions as an autocrine activator of G_q signaling in brown adipocytes. These data suggest that drugs able to inhibit G_q signaling in brown or beige fat may prove useful to enhance energy expenditure for therapeutic purposes.

Hepatocytes

The liver plays a central role in many key metabolic functions, including the regulation of glucose homeostasis. Hepatic glucose fluxes are regulated by insulin, glucagon, and many other hormones and neurotransmitters. In T2D, HGP is increased, particularly in the fasting state (Lin and Accili, 2011; Postic et al., 2004; Unger and Cherrington, 2012). Hepatocytes express dozens of GPCRs linked to distinct G protein signaling pathways (Exton, 1987; Regard et al., 2008). Among these receptors, the metabolic functions of the hepatic glucagon receptor (GCGR) have been studied in great detail. Glucagon binding to the GCGR leads to the activation of G_s, which eventually causes increased cAMP levels and a series of subsequent events that enhance HGP via glycogen breakdown (glycogenolysis) and gluconeogenesis. Thus, the major role of glucagon is to raise blood glucose levels under hypoglycemic conditions (D’Alessio, 2011; Estall and Drucker, 2006; Jiang and Zhang, 2003).

Hepatocyte-specific GsD mice

Akhmedov et al. (Akhmedov et al., 2017a) succeeded in generating a mutant mouse line (Hep-GsD mice) that expressed the GsD construct selectively in hepatocytes (Fig. 3c). As expected, CNO treatment of Hep-GsD mice triggered enhanced hepatic cAMP signaling, glycogenolysis, and changes in gene expression similar to these observed after stimulation of hepatic GCGRs (Akhmedov et al., 2017a). These CNO-induced changes in hepatocyte metabolism resulted in increased blood glucose levels under both fed and fasting conditions, suggesting that inhibition of hepatic G_s signaling might prove useful to lower pathologically elevated blood glucose levels.

In agreement with this notion, Chen et al. (Chen et al., 2005) demonstrated that liver-specific Gα_s KO mice (Hep-Gs-KO mice) displayed reduced blood glucose levels and increased glucose tolerance. These phenotypes were opposite to those observed with Hep-GsD mice (Akhmedov et al., 2017a), strongly suggesting that the metabolic outcomes of activating hepatic GsD receptors were indeed mediated by G_s. However, Hep-Gs-KO mice showed greatly increased glucagon and GLP-1 levels and pancreatic α-cell hyperplasia, complicating the interpretation of the metabolic phenotypes displayed by these mutant mice.

Hepatocyte-specific GiD mice

Hepatocytes also express several G_i-linked GPCRs such as certain α₂-adrenergic receptor subtypes (Exton, 1985) and the CB₁ cannabinoid receptor (Liu et al., 2012; Osei-Hyiaman et al., 2008). To study the metabolic outcomes of stimulating G_i-dependent signaling pathways in hepatocytes, Rossi et al. (Rossi et al., 2018) developed a mutant mouse line that selectively expressed the GiD construct in hepatocytes (Hep-GiD mice) (Fig. 3c). Somewhat surprisingly, CNO treatment of GiD-expressing hepatocytes led to an increase in glucose output (Rossi et al., 2018). In agreement with these in vitro data, CNO-injected Hep-GiD mice displayed elevated blood glucose levels, impaired glucose tolerance, and increased rates of gluconeogenesis and glycogenolysis in vivo (Rossi et al., 2018). On the other hand, mutant mice that lacked functional G_i proteins in hepatocytes (Hep-Gi-KO mice) showed in vivo metabolic phenotypes that were opposite to those caused by CNO treatment of Hep-GiD mice. Moreover, Hep-Gi-KO mice displayed significantly improved glucose homeostasis when maintained on an obesogenic diet (Rossi et al., 2018). Additional studies led to the identification of a signaling pathway that links the stimulation of hepatocyte G_i signaling to the production of reactive oxygen species (ROS), eventually leading to elevated HGP and impaired glucose homeostasis in a JNK-dependent fashion (Rossi et al., 2018).

Rossi et al. (Rossi et al., 2018) also showed that injection of WT mice with anandamide, a cannabinoid receptor agonist, led to impaired glucose tolerance and that this detrimental metabolic effect was mediated by hepatic G_i proteins (Rossi et al., 2018). This observation is in agreement with earlier work demonstrating that signaling initiated by hepatocyte CB₁ receptors causes severe metabolic impairments including reduced glucose tolerance (Liu et al., 2012; Osei-Hyiaman et al., 2008).

Taken together, these data strongly suggest that drugs able to inhibit hepatic G_i signaling may prove useful to restore euglycemia in T2D.

Hepatocyte-specific GqD mice

Hepatocytes also express GPCRs that are coupled to G proteins of the G_q family. The best known examples are the AT_1a angiotensin II receptor, different α₁-adrenergic receptor subtypes, and the V_1a and V_1b vasopressin receptors (Exton et al., 1981; Li et al., 2013; Lin and Accili, 2011; Regard et al., 2008).

Li et al. (Li et al., 2013) developed a mouse strain that expressed the GqD construct selectively in hepatocytes (Hep-GqD mice) (Fig. 3c). Acute CNO treatment of Hep-GqD mice, but not of WT control mice, markedly increased blood glucose levels, due to increases in the rates of both gluconeogenesis and glycogenolysis (Li et al., 2013). Previous studies suggest that G_q-mediated increases in cytoplasmic Ca²⁺ levels play a central role in G_q-dependent increases in HGP (Exton, 1987). G_q-mediated changes in gene expression may contribute to the stimulation of gluconeogenesis observed with CNO-injected Hep-GqD mice (Li et al., 2013). Studies with a mouse strain expressing a G_q-biased DREADD (M3D-Gq) in hepatocytes indicated that the increase in HGP caused by GqD activation does not require the recruitment of β-arrestins (Hu et al., 2016).

In summary, these data support the concept that agents that can block the activity of G_q-linked GPCRs endogenously expressed by hepatocytes, including, for example, members of the V₁ vasopressin receptor family (Exton, 1987; Li et al., 2013), may prove beneficial to improve glucose homeostasis for therapeutic purposes.

Skeletal muscle

In the postprandial state, skeletal muscle (SKM) is the major site of glucose uptake, and SKM insulin resistance is considered to be the key event in the development of hyperglycemia in T2D (DeFronzo and Tripathy, 2009). Like other cell types, SKM cells express dozens of GPCRs that differ in their G protein coupling preference (Jean-Baptiste et al., 2005; Regard et al., 2008). To date, the potential roles of these receptors in regulating SKM function and whole-body glucose homeostasis have not been investigated systematically.

SKM-specific GqD mice

To study the metabolic effects of stimulating receptor-mediated G_q signaling in SKM, Bone et al. (Bone et al., 2019) analyzed mutant mice that expressed the GqD designer receptor in SKM cells only (SKM-GqD mice) (Fig. 3d). Following CNO administration, lean and obese SKM-GqD mice showed improved glucose homeostasis, most likely due to increased glucose uptake into SKM. On the other hand, obese mutant mice lacking Gα_q/11 selectively in SKM displayed even more severe deficits in glucose homeostasis (Bone et al., 2019). Additional studies indicated that G_q-mediated activation of SKM AMPK triggers enhanced GLUT4 translocation to the plasma membrane, thus promoting glucose uptake in an insulin-independent fashion (Bone et al., 2019). These data should be of considerable clinical relevance since SKM glucose uptake is impaired in T2D due to peripheral insulin resistance.

Central AgRP and POMC neurons

The use of DREADD technology and optogenetic techniques has led to major novel insights into the neuronal circuits that control energy and glucose homeostasis in the CNS (Krashes, 2017; Sternson and Eiselt, 2017). Specifically, these studies have shown that various neuronal subpopulations within the hypothalamus are of particular importance in regulating key metabolic functions (Krashes, 2017; Sternson and Eiselt, 2017). The arcuate nucleus of the hypothalamus (ARC) contains two neuronal subpopulations that play central roles in the control of food intake and energy homeostasis (Morton et al., 2014). One of these neuronal subpopulations, so-called AgRP neurons, contains agouti-related protein (AgRP), neuropeptide Y (NPY), and GABA. After activation of these neurons, AgRP is released and inhibits the activity of central melanocortin receptors (predominantly melanocortin-4 receptors [MC4Rs]) to promote food intake and increase energy expenditure (Cone, 2006; Morton et al., 2014). The second major set of ARC neurons is referred to as pro-opiomelanocortin (POMC) neurons. Following their activation, these neurons release α-melanocyte stimulating hormone (α-MSH) which suppresses feeding behavior by binding to and activating central melanocortin receptors (primarily MC4Rs) that are expressed in different areas of the brain (Cone, 2006; Morton et al., 2014).

During the past decade, chemogenetic and optogenetic studies have identified several other neuronal subpopulations that modulate energy homeostasis and other important metabolic processes. Many of these studies have been reviewed recently (Atasoy and Sternson, 2018; Burnett and Krashes, 2016).

AgRP neurons

Early studies have shown that diphtheria toxin-mediated ablation of ARC AgRP neurons causes rapid cessation of feeding in mice (Gropp et al., 2005; Luquet et al., 2005), indicating that this neuronal subpopulation plays a key role play in regulating food intake. The development of optogenetic and chemogenetic strategies has made it possible to selectively modify the activity of AgRP neurons under different experimental conditions. Such studies have led to a wealth of novel information about how the activity of AgRP neurons is regulated under physiological and pathophysiological conditions and which metabolic processes are modulated by these neurons.

In the following, we will briefly summarize studies that used DREADD technology to study the function and regulation of AgRP neurons. Like all other neurons, AgRP neurons express a number of functionally diverse GPCRs (Cowley et al., 2003; Ren et al., 2012). Metabolic studies with mice expressing the GqD, GiD, and GsD designer receptors selectively in AgRP neurons have yielded novel information about the ability of different functional classes of GPCRs to regulate the activity of these neurons and whole body energy homeostasis in general (Fig. 3e).

AgRP neuron-specific GqD and GiD mice

DREADDs have emerged as powerful novel tools for neuroscience research, primarily due to the fact that CNO treatment of GqD-expressing neurons produces rapid depolarization and increased neuronal firing rate (Rogan and Roth, 2011; Roth, 2016). On the other hand, exposure of GiD-expressing neurons to CNO results in neuronal hyperpolarization and decreased firing rate (Rogan and Roth, 2011; Roth, 2016). As expected, CNO treatment of mutant mice expressing the GqD designer receptor selectively in AgRP neurons resulted in neuronal excitation (Krashes et al., 2011). The resulting release of AgRP, NPY, and GABA triggered a rapid increase in food intake and reduced energy expenditure (Krashes et al., 2011). The same laboratory (Krashes et al., 2013) also showed that the release of either NPY or GABA alone was sufficient to promote acute feeding in CNO-treated AgRP-GqD mice. In contrast, AgRP release enhanced food intake over a delayed but prolonged period of time (Krashes et al., 2013). CNO treatment of AgRP-GiD mice led to reduced food intake at the beginning of the dark cycle when mice are feeding most actively (Krashes et al., 2011).

Chemogenetic activation (via GqD) or inhibition (via GiD) of AgRP neurons has no significant effect on blood glucose levels and glucose tolerance (Üner et al., 2019). However, GqD-mediated stimulation of AgRP neurons in diabetic ob/ob mice attenuates the ability of leptin to inhibit feeding but does not interfere with the ability of leptin to lower blood glucose levels (Üner et al., 2019). These observations indicate that changes in the activity of AgRP neurons regulate food intake, but are insufficient to modulate glucose homeostasis in normal and diabetic mice.

Although acute chemogenetic activation of AgRP neurons promotes both food intake and weight gain (Ewbank et al., 2020; Krashes et al., 2011; Krashes et al., 2013), chronic, GqD-mediated stimulation of AgRP neurons causes a return of food intake to baseline levels within one week (Ewbank et al., 2020). Moreover, after chronic chemogenetic activation of AgRP neurons for two months, body weight also returns to baseline levels, most likely due to the activation of compensatory neuronal pathways that counteract the metabolic effects of chronic stimulation of AgRP neurons (Ewbank et al., 2020).

A recent study (Bunner et al., 2020) showed that a single bout of acute exercise increases the activity of ARC AgRP neurons, resulting in increased food intake. This orexigenic effect was abolished after CNO treatment of mice expressing the inhibitory GiD receptor in AgRP neurons (Bunner et al., 2020), indicating that the exercise-induced orexigenic effect is dependent on the activation of AgRP neurons.

The activity of mouse AgRP neurons can also affect other central functions and behaviors including anxiety. For example, a recent study (Li et al., 2019) demonstrated that GqD-mediated activation of AgRP neurons reduces anxiety in fed mice, while GiD-mediated inhibition of AgRP neurons blocks fasting-induced anxiolytic activity. These data indicate that hypothalamic AgRP neurons can also modulate complex behaviors such anxiety.

ARC AgRP neurons also regulate the activity of the sympathetic nervous system and cardiovascular function (Cassaglia et al., 2011; Steculorum et al., 2016). For example, a recent study (Jiang et al., 2020) showed that acute, GqD-mediated activation of AgRP neurons results in a decrease in renal sympathetic nerve activity in conscious mice. Somewhat surprisingly, chronic, GqD-mediated activation of AgRP neurons leads to an increase in blood pressure specifically during the light phase when mice are mostly inactive (Jiang et al., 2020). These findings suggest that the activity state of AgRP neurons regulates sympathetic outflow and can cause major effects on cardiovascular function.

Several of the studies summarized above raise caveats regarding the potential development of drugs aimed at inhibiting the activity of AgRP neurons for therapeutic purposes. Major obstacles are that AgRP neurons do not only regulate food intake but also modulate other complex behaviors (e.g., anxiety) and cardiovascular functions and that the metabolic benefits of this strategy may disappear over time (Ewbank et al., 2020).

AgRP neuron-specific GsD mice

To explore the potential regulation of AgRP neurons by receptor-dependent G_s signaling, Nakajima et al. (Nakajima et al., 2016) generated mice that selectively expressed the GsD construct in AgRP neurons (AgRP-GsD mice) (Fig. 3e). CNO-injected AgRP-GsD mice display a long-lasting increase in food intake that is accompanied by a significant increase in body weight, an effect that is mediated almost exclusively by AgRP release (Nakajima et al., 2016). Additional mechanistic studies suggested that increasing G_s signaling in AgRP neurons promotes the expression of the Klf4 transcription factor in AgRP neurons. Klf4 stimulates the expression of AgRP, ultimately leading to an increase in AgRP release and enhanced appetite. Interestingly, stimulation of G_s-linked PAC₁ receptors endogenously expressed by AgRP neurons can mimic the orexigenic effects displayed by CNO-treated AgRP-GsD mice (Nakajima et al., 2016). These observations raise the possibility that drugs able to interfere with G_s signaling in AgRP neurons may prove useful to reduce appetite for therapeutic purposes.

POMC neuron-specific GqD and GiD mice

Like AgRP neurons, POMC neurons play an important role in modulating food intake and energy homeostasis (Morton et al., 2014). Activation and inhibition of POMC neurons by using either DREADD technology or optogenetic techniques reduces and increases food intake, respectively (Aponte et al., 2011; Atasoy et al., 2012). However, these effects occur with considerable delay suggesting that POMC neurons are primarily involved in the long-term regulation of feeding.

In the adult rodent brain, most POMC neurons are located in the ARC (Padilla et al., 2012). However, POMC neurons are also present in the nucleus tractus solitarius (NTS) of the medulla and several other brain regions (Padilla et al., 2012). Interestingly, GqD-mediated activation of NTS POMC neurons results in an immediate reduction in food intake (Zhan et al., 2013). In contrast, only chronic, but not acute, chemogenetic stimulation of ARC POMC neurons suppresses food intake (Zhan et al., 2013), indicating that POMC neurons in the ARC and NTS have different functional roles. While NTS POMC neurons appear to respond to acute satiety signals, ARC POMC neurons seem to be involved in long-term regulation of feeding behavior and energy homeostasis.

The activity of POMC neurons is regulated by two major circulating hormones, leptin and insulin. Leptin depolarizes and activates POMC neurons, leading to the suppression of food intake (Morton et al., 2014). By contrast, insulin inhibits POMC neurons, thus limiting α-MSH release (Morton et al., 2014). However, under certain experimental conditions, POMC neurons can also be activated by insulin (Qiu et al., 2018; Qiu et al., 2014). A recent study demonstrated that the proportion of POMC neurons that is activated by insulin depends on the activity of the phosphatase TCPTP, which is increased after fasting but reduced in the fed state (Dodd et al., 2018). These authors also reported that prolonged GiD-mediated inhibition of ARC POMC neurons leads to increased HGP and impaired whole-body insulin sensitivity. On the other hand, chronic, GqD-mediated activation of ARC POMC neurons lowers HGP and improves hepatic glucose metabolism (Dodd et al., 2018). Taken together, these observations suggest that changes in the proportion of POMC neurons that are activated or inhibited by insulin may directly affect peripheral glucose metabolism. Importantly, this study highlights the fact that changes in the activity of POMC neurons can affect the function of metabolically important peripheral cell types such as hepatocytes of the liver.

In agreement with this concept, a recent study (Brandt et al., 2018) demonstrated that chemogenetic activation of ARC POMC neurons stimulates the activity of sympathetic nerves supplying the liver, thus causing changes in liver metabolism (e.g., activation of mTOR signaling) that prime the liver for incoming nutrients. The authors also reported that sensory food perception activates ARC POMC neurons, resulting in similar hepatic effects which appear to be dependent on the activation of MC4Rs (Brandt et al., 2018). These observations indicate that POMC neurons play a central role in a circuit linking sensory food anticipation to changes in hepatocyte function that prime the liver for incoming nutrients.

A related study showed that GiD-mediated inhibition of ARC POMC neurons promotes food intake and lowers blood glucose levels in normoglycemic mice (Uner et al., 2019). Additional experimentation revealed that the hypoglycemic effect caused by inhibition of POMC neurons is independent of changes in food intake (Uner et al., 2019). Although the neuronal circuits underlying this hypoglycemic response remain to be elucidated, these findings suggest that POMC neurons represent a potential target for novel drugs aimed at treating hyperglycemia.

The activity of POMC neurons can also affect cardiovascular parameters including blood pressure. A recent study reported that chronic GqD-mediated activation of mouse ARC POMC neurons leads to a significant reduction in blood pressure (Jiang et al., 2020). The molecular and cellular mechanisms underlying this effect remain to be explored in future studies.

Conclusions and future directions

The studies summarized in this short review clearly indicate that DREADD-based approaches are highly useful to assess the in vivo metabolic consequences of activating specific GPCR/G protein signaling cascades in metabolically distinct cell types. The novel information derived from these studies should stimulate efforts to develop novel classes of drugs useful for the treatment of several metabolic disorders including T2D and obesity (Table 1). Endogenous GPCRs need be identified that can be targeted to mimic the beneficial effects observed with the DREADD mutant mice. Additional studies are required to ensure that the metabolically important signaling pathways identified by the use of DREADD technology in mice are also operative in human tissues. Finally, it remains to be explored whether receptor (DREADD)-mediated recruitment of β-arrestins contributes to the physiological effects caused by DREADD activation in different cell types. Such information is important for the design of novel drugs endowed with pharmacological profiles that ensure maximum efficacy and a favorable side effect profile.

Table 1.

Potential therapeutic strategies to improve glucose hemostasis and/or reduce obesity, as suggested by the use of DREADD technology in mice

Cell type	DREADD	Major phenotypes after DREADD activation	Drugs predicted to improve glucose hemostasis and/or reduce obesity	Refs.
β-cells	GqD	Increase in insulin secretion, β-cell proliferation, and β-cell mass	GPCR agonists that stimulate β-cell Gq	(Guettier et al., 2009) (Jain et al., 2013)
	GsD	Enhanced insulin release and increased β-cell mass	GPCR agonists that activate β-cell Gs	(Guettier et al., 2009)
α-cells	GiD	Reduced glucagon secretion Deficits in glucose tolerance and GSIS	GPCR agonists that enhance α-cell activity	(Zhu et al., 2019)
L-cells (colon)	GqD	Improved glucose tolerance Reduced food intake	GPCR agonists that stimulate L-cell Gq	(Lewis et al., 2020)
Adipocytes	GsD	Increased lipolysis and energy expenditure Reduced food intake and improved glucose homeostasis	GPCR agonists that stimulate adipocyte Gs	(Wang et al., 2019) (Caron et al., 2019)
	GiD	Impaired lipolysis Improved insulin and glucose tolerance	GPCR agonists that activate adipocyte Gi	(Wang et al., 2020)
	GqD	Inhibition of brown adipocyte differentiation	GPCR antagonists that block Gq signaling in brown/beige fat	(Klepac et al., 2016)
Hepatocytes	GsD	Enhanced glycogenolysis and hyperglycemia	GPCR antagonists that block hepatocyte Gs signaling	(Akhmedov et al., 2017a)
	GiD	Elevated HGP and hyperglycemia Impaired glucose tolerance	GPCR antagonists that interfere with hepatocyte Gi signaling	(Rossi et al., 2018)
	GqD	Increased HGP and hyperglycemia	GPCR antagonists that inhibit hepatocyte Gq signaling	(Li et al., 2013)
Skeletal muscle cells	GqD	Enhanced glucose uptake into sk. musc. Improved glucose homeostasis	GPCR agonists that stimulate skeletal muscle Gq	(Bone et al., 2019)
AgRP neurons	GqD	Enhanced food intake Weight gain	GPCR antagonists suppressing Gq signaling in AgRP neurons	(Krashes et al., 2011)
	GiD	Reduced food intake	GPCR agonists enhancing Gi signaling in AgRP neurons	(Krashes et al., 2011)
	GsD	Enhanced food intake Increase in body weight	GPCR antagonists inhibiting Gs signaling in AgRP neurons	(Nakajima et al., 2016)
POMC neurons	GqD	Reduced food intake	GPCR agonists promotin Gq signaling in POMC neurons	(Zhan et al., 2013) (Atasoy et al., 2012)
	GiD	Increased food intake	GPCR antagonists suppressing Gi signaling in POMC neurons	(Krashes et al., 2011)
	GiD	Increased food intake	GPCR agonists stimulating Gi signaling in POMC neurons	(Uner et al., 2019)

Abbreviations: HGP, hepatic glucose production; GSIS, glucose-stimulated insulin secretion.

List of abbreviations

ACh

acetylcholine

AgRP

agouti-related protein

CNO

clozapine-N-oxide

DCZ

deschloroclozapine

DREADD

Designer Receptor Exclusively Activated by a Designer Drug

FFA

free fatty acid

FFAR

free fatty acid receptor 1

GCGR

glucagon receptor

GIRK channel

G-protein-regulated inward-rectifier potassium channel

GLP-1

glucagon-like peptide-1

GPCR

G protein-coupled receptor

GSIS

glucose-stimulated insulin secretion

HFD

high-fat diet

HGP

hepatic glucose production

IP₃

inositol trisphosphate

IRS2

insulin receptor substrate 2

MC4 receptor

melanocortin 4 receptor

NPY

agouti-related peptide

NPY

neuropeptide Y

POMC

pro-opiomelanocortin

PTX

pertussis toxin

PYY

peptide YY

RASSL

Receptor Activated Solely by Synthetic Ligands

ROS

reactive oxygen species

SKM

skeletal muscle

T2D

type 2 diabetes