G Protein–Coupled Receptors Targeting Insulin Resistance, Obesity, and Type 2 Diabetes Mellitus

A. Adrenoceptors

Activation of the sympathetic nervous system, with consequent release of the catecholamines, adrenaline from the adrenals, and noradrenaline from sympathetic nerve endings, is one of the most efficient ways of increasing blood glucose levels. The actions of the catecholamines are mediated by a group of nine GPCRs; these adrenoceptors are divided into three subgroups, α₁, α₂, and β, based on their sequence similarities and dominant signaling pathways. The α₁-adrenoceptor subgroup couples primarily to Gα_q/11 to activate phospholipase C, causing hydrolysis of phosphatidylinositol 4,5-bisphosphate to diacylglycerol, which activates protein kinase C (PKC), and inositol 1,4,5-trisphosphate, which releases Ca²⁺, causing a variety of effects in many tissues (Piascik and Perez, 2001; Wier and Morgan, 2003); α₂-adrenoceptors are Gα_i/o-coupled receptors that inhibit adenylyl cyclase to reduce cAMP production, activate K⁺ channels, and inhibit voltage-gated Ca²⁺ channels (Limbird, 1988); and β-adrenoceptors are Gα_s-coupled receptors that activate adenylyl cyclase to increase intracellular levels of cAMP. Each of the adrenoceptor subgroups has representatives that have a wide variety of effects on glucose metabolism; this is influenced by the receptor populations and signaling components expressed in particular cell types.

B. CB₁R

Cannabinoid receptors are highly expressed in the CNS and on immune and inflammatory cells. There are two cannabinoid receptors, CB₁R and CB₂R, which are activated by the bioactive lipid molecules anandamide and 2-arachidonoylglycerol. Activation of CB₁R in the CNS causes an increased food intake that is further enhanced by the mesolimbic system. However, activation of CB₁R in metabolic tissues, including both white and brown adipocytes, causes an increase in fatty acid synthesis and reduction in lipolysis (Gruden et al., 2016); in Zucker diabetic fatty rats, overstimulation of CB₁R expressed on macrophages caused a marked increase in the activation of the inflammasome, apoptosis, and loss of β-cell function (Rohrbach et al., 2012). More recently, increased activation and signaling of the CB₁R in podocytes have been implicated in the development of type 2 diabetic nephropathy (Jourdan et al., 2014). Multiple studies have shown an improvement in metabolic phenotype when treated with CB₁R antagonists, including SR141716 (Rimonabant; Sanofi-Aventis, Paris, France) (Lafontan et al., 2007; Scheen, 2007; Nam et al., 2012). However, development of suitable therapeutics has proved challenging due to on-target neuropsychiatric effects. To overcome this, antagonists, such as JD5037, have been developed that do not penetrate the brain and improve pancreatic β-cell function and decrease cell loss in animal models (Chorvat, 2013; Jourdan et al., 2014); however, no human studies with this compound have yet been initiated. For a comprehensive review on the endocannabinoid system, see Gruden et al. (2016).

C. EP₃R

Prostaglandin E₂ is derived from arachidonic acid and is the endogenous ligand for the EP₃R. In the BTBR mouse strain, containing the leptin^ob/ob mutation to cause obesity and diabetes, the EP₃R negatively regulates insulin secretion, and upregulation of both the ligand and receptor inhibits GLP-1R signaling (Neuman and Kimple, 2013). In a recent study by Neuman et al. (2017), enrichment of pancreatic islets with eicosapentaenoic acid, the precursor to prostaglandin E₃, caused a marked decrease in arachidonic acid, improved glucose-stimulated insulin secretion, a reduction in IL-1β production, and an improvement in glucose tolerance and β-cell function. Thus, blockade of EP₃R in pancreatic islets may be a viable therapy and recently new EP₃R antagonists have been disclosed (Abdel-Magid, 2015). However, data from Vanderbilt University (Nashville, TN) researchers have shown conflicting data. Ep3r^−/− mice, generated on the C57BL/6 background and fed a high-fat diet, displayed an increase in macrophage infiltration, increased ectopic lipid accumulation in skeletal muscle and liver, and hepatic steatohepatitis, culminating in increased obesity, lipolysis, and increased insulin resistance (Ceddia et al., 2016). These discrepancies highlight the need for further investigation into EP₃R as a metabolic target and emphasize the context dependency of data between different animal models.

D. Free Fatty Acid Receptors: FFAR1, FFAR2, and FFAR3 (GPR40, GPR43, and GPR41)

The benefits of dietary fiber have been recognized for decades, although the identification of the targets involved in these effects has only recently been established (Thorburn et al., 2014; Alvarez-Curto and Milligan, 2016). FFA receptor (FFAR)1, FFAR2, and FFAR3 bind short-chain fatty acids (SCFAs), including acetate, propionate, and butyrate, produced from the fermentation of undigested carbohydrates and dietary fibers by colonic bacteria (Bindels et al., 2013; McKenzie et al., 2015), and have subsequently been termed metabolite-sensing GPCRs. SCFAs produce many beneficial effects, including maintenance of immune and gut homeostasis, regulation of T cell activation, and control of various immune pathways, specifically via the inhibition of NF-κB signaling (Thorburn et al., 2014). Each receptor is widely expressed in multiple metabolically important tissues, including pancreatic β-cells, immune cells, adipocytes, and enteroendocrine cells (Table 1); for comprehensive reviews, see Blad et al. (2012), Husted et al. (2017), and Milligan et al. (2017).

E. Glucagon-Like Peptide 1 Receptor

GLP-1 is a key incretin, promoting glucose-dependent insulin secretion in response to meal ingestion. It has broad physiologic effects, including preservation of β-cell mass, and inhibition of glucagon secretion, gastric emptying, and food intake, in addition to its incretin role (Baggio and Drucker, 2014; Graaf et al., 2016). GLP-1 is also cardio- and neuroprotective (Salcedo et al., 2012; Graaf et al., 2016; Wiberg et al., 2016). Endogenous GLP-1 is rapidly degraded by endopeptidases, most prominently dipeptidyl peptidase-IV (DPP-IV), and has a very short plasma half-life, making it unsuitable as a therapeutic. As such, targeting of the GLP-1R has focused on peptide mimetics resistant to enzymatic degradation, and with prolonged plasma half-life (Hui et al., 2002). These peptide agonist mimetics lower HbA1c in T2DM, acting both directly on the pancreatic β-cell and via vagal afferents (Trujillo et al., 2015; Krieger et al., 2016). In addition, GLP-1R agonists cause increased β-cell proliferation, regeneration, and neogenesis, in both animal models and in vitro cultures of human islets (Tian et al., 2011), and decreased apoptosis (Vilsbøll, 2009), collectively contributing to preservation of β-cell mass. However, these effects may be specific to mouse models. There are now multiple GLP-1R agonists approved for T2DM treatment, including exenatide, albiglutide, and liraglutide (reviewed in detail; Meier 2012; Graaf et al., 2016).

Importantly, long-term treatment with GLP-1R agonists promotes satiety and weight loss in both diabetic and obese patients (van Bloemendaal et al., 2014). GLP-1R is expressed on the vagus nerve, linking the gut to the CNS, and abundantly in the hypothalamus, paraventricular nucleus, dorsomedial nucleus, and arcuate nucleus (Baggio and Drucker, 2014). The weight loss is most likely driven by combination of GLP-1 action to inhibit gastric emptying, as well as direct actions in the CNS; direct activation of propiomelanocortin expressing arcuate nucleus neurons by liraglutide induced weight loss, which was blocked by the GLP-1R antagonist exendin (9–39) (Secher et al., 2014; Sisley et al., 2014). Nonetheless, KO of GLP-1R in the CNS using nestin-Cre Glp1r^fl/fl mice, or vagus nerve using Phox2b-Cre Glp1r^fl/fl mice, did not affect food intake or body weight of animals fed normal chow or a high-fat diet, suggesting that activation of receptors at either site alone may have limited benefit. Beneficial weight loss is well-documented for treatment with exenatide and liraglutide (Raun et al., 2007; Buse et al., 2009), the earliest drugs to market; liraglutide has also been approved independently for treatment of obesity in the United States (Iepsen et al., 2015). These initial drugs are metabolically stable peptide mimetics of GLP-1 that require daily injections, and much recent work has focused on development of drugs with extended half-lives, including sustained release formulation (taspoglutide), albumin fusion (albiglutide), and linkage to modified Fc of immunoglobulin (dulaglutide), for once-weekly injection (Smith et al., 2016). Overall, GLP-1 mimetics have an excellent safety profile; the most prevalent side effects include injection-site reactions, nausea, diarrhea, or vomiting, and these effects can be dose limiting in patients. The latter events are thought to be primarily related to effects on the gastrointestinal tract; however, these tend to abate with time (Buse et al., 2009). Of note, taspoglutide was withdrawn from phase III clinical trials due to high risk for nausea (Zaccardi et al., 2016). All approved mimetics improve HbA1c levels, although the longer-acting formulations appear to have reduced efficacy for control of body weight, with limited effect on gastric emptying (Dungan et al., 2014; Pratley et al., 2014). A new generation mimetic with reversible albumin binding, semaglutide, formulated for once-weekly dosing, has shown promising effects on HbA1c and body weight control (Blundell et al., 2017), whereas implantable technologies for osmotic slow release (Intarcia) are currently in phase III clinical trials for delivery of exenatide over a 6-month period. Evidence is now emerging that liraglutide (Velez et al., 2015) and semaglutide may also have beneficial effects with respect to cardiovascular function and mortality (Marso et al., 2016), increasing the potential benefit of this drug class for treatment of T2DM patients at risk for cardiovascular disease.

There is increasing interest in benefits of dual-specificity peptide ligands that have activity at multiple glucagon-peptide family receptors, combining activity at GLP-1 and glucagon receptors (Day et al., 2009), or GLP-1 and GIP receptors (De and DiMarchi, 2010); these peptide agonists have shown promise in preclinical studies for obesity treatment with greater weight loss than seen with GLP-1 agonism alone (Day et al., 2009; De and DiMarchi, 2010). This dual targeting strategy is being pursued by companies, including Zealand Pharma and Transition Therapeutics. Moreover, combination of GLP-1 agonists with other anorectic peptides such as amylin is reported to provide synergistic effects on weight loss (Roth et al., 2012).

The GLP-1 receptor has been refractory to small-molecule drug development, and this may in part be explained by the extended pharmacophore recently revealed by structural studies (Jazayeri et al., 2017; Song et al., 2017; Zhang et al., 2017), which would create difficulties for development of mimetics acting via the orthosteric peptide binding site. A number of small-molecule compounds have been disclosed (Willard et al., 2012), but none have yet been developed for therapeutic use. Many of these act as allosteric ligands, and indeed, at least one class acts on the intracellular face of the receptor (Nolte et al., 2014). As an alternative, a range of modified peptides of ∼11 amino acids has been developed, some of which retain high affinity and potency (Huang et al., 2015a; Jazayeri et al., 2017), and are resistant to degradation (Jazayeri et al., 2017). These peptides have both overlapping and unique interactions with GLP-1 at targeting GLP-1 receptors (Jazayeri et al., 2017; Zhang et al., 2017). It is believed that such peptides may be better suited for oral formulation. Nonetheless, semaglutide has been trialled in oral formulation, and, despite limited bioavailability, can still achieve therapeutically viable concentrations (Jensen et al., 2017), with high enough dose.

The GLP-1R is a secretin-like class B GPCR (Donnelly, 2012) and is pleiotropically coupled to multiple G proteins, including Gα_i, Gα_o, and Gα_q/11, as well as other effector/regulatory proteins, most notably the arrestins (Koole et al., 2013; Graaf et al., 2016), and this has very broad consequences for intracellular signaling. Virtually all GLP-1 ligands that have been assessed appropriately exhibit biased agonism relative to the principal endogenous GLP-1 peptide [GLP-1(7-36)(NH₂)] (Koole et al., 2010; Graaf et al., 2016; Wootten et al., 2017). This is evident for both natural ligands of the receptor (e.g., oxyntomodulin) and GLP-1 mimetics (e.g., exendin-4), with divergent effects of each of these peptides in pancreatic β-cell–like INS-1 insulinoma cells (Wootten et al., 2016). This is likely to impact on treatment outcomes with different GLP-1 mimetic drugs. Indeed, although principally considered in the context of altered pharmacokinetics, there are diverse outcomes observed with currently approved therapeutics for body weight, gastric emptying, and cardiovascular improvement, as well as the extent of unwanted side effects (Graaf et al., 2016) that may also arise from biased agonism of these peptides. Of particular note, a recently discovered analog of exendin-4, termed exendin-P5, which had a different signaling bias to exendin-4, displayed a distinct impact in preclinical models of diabetes and obesity (Zhang et al., 2015). Both peptides reduced plasma glucose; however, in contrast to exendin-4, exendin-P5 was only a weak insulinotrope and caused altered adiposity. Thus, altered signaling bias could provide novel therapeutic potential for this target. To date, we have only a poor understanding of which signaling pathways interplay for beneficial effects in both the pancreas and extrapancreatic tissues that control appetite, gastric emptying, and cardio- and/or neuroprotection. Nonetheless, this promises to be a fruitful area of research for development of new-generation GLP-1R drugs.

F. G Protein–Coupled Estrogen Receptor

G protein–coupled estrogen receptor (GPER), the nongenomic receptor for estrogen, previously denoted GPR30, is widely expressed in the heart, intestines, reproductive tissues, immune cells, and metabolic tissues, including adipose, skeletal muscle, and liver (Sharma et al., 2017). Furthermore, GPER expression has been reported on pancreatic β-cells, with expression levels notably higher in islets from both female mice and humans compared with those derived from male mice (Balhuizen et al., 2010; Kumar et al., 2011). The estrogen receptor agonist, 17β-estradiol, activates GPER with nanomolar affinity (Wang et al., 2016), and the genomic estrogen receptor modulators tamoxifen and raloxifene also have moderate GPER activity (Sharma et al., 2017). Recently, a selective GPER agonist (G-1) was identified that displays no activity at the estrogen receptors ERα or ERβ (Bologa et al., 2006), and selective GPER antagonists have also been discovered (G15 and G36) (Dennis et al., 2009, 2011). Activation of GPER by 17β-estradiol and G-1 revealed coupling to both Gα_s and Gα_i proteins, yielding a net increase in both cAMP and intracellular Ca²⁺ (Martensson et al., 2009). More recently, data suggest that GPER can signal from the plasma membrane and also intracellular compartments, including the Golgi and ER (Otto et al., 2008; Cheng et al., 2011).

GPER KO mouse studies have yielded a range of reported metabolic outcomes. In one study, female Gper1^−/− mice fed a high-fat diet displayed reduced body weight and were protected from obesity-induced insulin resistance; however, no significant effects were observed in male animals (Wang et al., 2016). Contrary to these findings, Sharma et al., (2013) demonstrated that both female and male Gper1^−/− mice had increased body weight compared with wild-type controls and developed insulin resistance. More subtly, Davis et al., (2014) showed that this increased body weight in females was delayed by 6 weeks in comparison with males. Female Gper1^−/− mice aged 6 months also displayed impaired glucose tolerance, increased blood pressure, and reduced bone growth (Mårtensson et al., 2009). These disparities most likely result from different methods used to generate the transgenic animals and variations in the length of feeding, but overall they indicate that deletion of GPER results in a similar phenotype obtained from ERα KO models, suggesting that activation of GPER by estrogen may be necessary to maintain glucose homeostasis (Sharma et al., 2017). Although GPER remains an interesting drug target, further studies are required to understand the mechanism of GPER activation in metabolic tissues. Furthermore, confirmation of the effects observed in KO mouse studies in humans, both male and female, will be necessary. For a comprehensive review of GPER, see Sharma et al. (2017).

G. GPR55

GPR55 is activated by endocannabinoids and was initially classified as the third cannabinoid receptor (Ross, 2009). Subsequently, it has also been shown to be activated by lipids, including lysophospholipid, and L-α-lysophosphatidylinositol (Henstridge et al., 2010). Unlike many of the GPCRs involved in insulin resistance or pancreatic β-cell dysfunction that predominantly couple to Gα_s or Gα_i, GPR55 couples primarily to Gα_q/11 and Gα_12/13 (Ross, 2009; Simcocks et al., 2014). The glucose-lowering and insulinotropic effects of a range of GPR55 agonists in vivo have been recently described, demonstrating the therapeutic potential for this target (McKillop et al., 2013). However, many of these ligands lack specificity, thus limiting conclusions from these studies. Recently, studies show that Gpr55^−/− mice develop increased insulin resistance, adiposity, and fat mass (Meadows et al., 2015) together with a significant reduction in physical activity, even though muscle function was unaffected. These data therefore support the developmental potential of GPR55 selective agonists. For a more comprehensive review of the role of GPR55 in metabolism, see Liu et al. (2015a).

H. GPR82

GPR82 belongs to the P2Y₁₂-like family, which also contains the ADP receptor P2Y₁₃R, the UDP-glucose receptor P2Y₁₄R, and the orphan receptors GPR87 and GPR171, all of which couple to Gα_i (Cattaneo, 2015). Little else is known about the receptor, but, interestingly, Gpr82^−/− mice display a lean phenotype, have decreased serum triglyceride levels, increased insulin sensitivity, and glucose tolerance when challenged with a Western diet (32.8% sugar/21.2% raw fat) (Engel et al., 2011). Unfortunately, no further information is available regarding the physiologic role of the receptor. The lack of a synthetic or endogenous ligand has limited further studies into the receptor signaling.

I. GPR119

GPR119 is highly expressed on pancreatic β-cells and enteroendocrine cells (Hansen et al., 2012). Multiple endogenous ligands for the receptor have been identified with varying potencies, including oleoylethanolamide, N-oleyldopamine, olvanil, and 1-oleoyl-lysophosphatidylcholine (18:1-lysoPC) (Ahrén, 2009; Hansen et al., 2012; Al-Barazanji et al., 2015). The steroid glycoside, gordonoside F, from the popular herbal weight loss supplement Hoodia gordonii, activates GPR119 (Zhang et al., 2014). This plant has been used medicinally for thousands of years by Xhomani Bushmen as an anorexant (Zhang et al., 2014) without known side effects, suggesting gordonoside F may represent a novel class of compounds for further investigation. Like the FFARs, GPR119 causes the release of GLP-1 and GIP, upon activation from L and K cells, respectively (Chu et al., 2008; Lauffer et al., 2009). GPR119 couples to Gα_s to increase glucose-stimulated insulin release from pancreatic β-cells via the formation of cAMP (Chu et al., 2008). GPR119 controls enteral glucose tolerance by three main complementary mechanisms: 1) direct activation of insulin secretion in islets, 2) promotion of secretion of incretins such as GLP-1 and GIP to potentiate glucose-stimulated insulin secretion (Panaro et al., 2017), and 3) delay of gastric emptying via, at least in part, the enhancement of PYY action (Hansotia et al., 2007; Flock et al., 2011). The glucoregulatory properties of GPR119 led to the development of multiple synthetic agonists for the treatment of T2DM, including GSK1292263 (GSK), MBX-2982 (CymaBay Therapeutics), and PSN821 (Astellas/Prosidion) (Hansen et al., 2012; Nunez et al., 2014; Liu et al., 2015c; Mittermayer et al., 2015; Ritter et al., 2016). However, both GSK1292263 and MBX-2982 were discontinued due to lack of efficacy in phase II trials, in which neither compound improved glycemia handling. Clinical failure of these compounds has cast doubt on the utility of targeting GPR119 for the treatment of T2DM (Kang, 2013), but may in part be due to suboptimal compounds with low potencies or lack of sufficient exposure and activity to cause release of insulin and the incretins (Ritter et al., 2016), as most of the molecules assessed in clinical trials failed to trigger robust increases in gut hormones (Ritter et al., 2016). Nonetheless, GPR119 agonists may be effective when used in combination with current T2DM treatments. GSK2041706 in combination with metformin in diet-induced obese mice showed a synergistic effect to reduce body weight (Al-Barazanji et al., 2015). In contrast, another GPR119 agonist, GSK1292263, in combination with metformin and sitagliptin in patients with T2DM yielded no additional improvement in glycemic control (Nunez et al., 2014). Although disappointing, this latter study was performed for only 14 days, and longer-term studies may be needed to better address these key questions. The lack of complete translation of the rodent pharmacology in type 2 patients with diabetes may also result from species-specific biology of the receptor. Of note, GPR119 has been found by single-cell RNA Seq to be highly expressed in human glucagon-producing α-cells (10-fold higher levels than β-cells), where its function remains largely unknown (Blodgett et al., 2015). Interestingly, an unexpected role of GPR119 in the liver has been uncovered with some potential consequences for the treatment of fatty liver diseases (Yang et al., 2016). Whereas expression of GPR119 in hepatocytes has been a subject of debate (Odori et al., 2013), Yang et al. (2016) showed that hepatocytes do express GPR119 at both mRNA and protein levels in both mice and humans. Moreover, pharmacological activation of GPR119 led to a suppression of de novo lipogenesis via activation of AMP kinase and subsequent inhibition of the LXR-SREBP1c transcriptional cascade, ultimately resulting in a significant reduction of hepatic steatosis. These effects were abolished in Gpr119^−/− KO mice (Yang et al., 2016). Further studies are required to determine whether GPR119 agonists might be useful for the treatment of nonalcoholic fatty liver diseases.

J. GPR142

Numerous publications describe synthetic phenylalanine derivatives as agonists acting at GPR142 (Du et al., 2012; Lizarzaburu et al., 2012; Toda et al., 2013; Yu et al., 2013; Guo et al., 2016). All of these compounds were developed based on the understanding that the receptor is activated by tryptophan, an essential amino acid that is the biologic precursor to serotonin (Wurtman and Anton-Tay, 1969), and niacin (Ikeda et al., 1965). GPR142 is highly expressed on pancreatic β-cells and signals via Gα_q (Wang et al., 2016a). Activation of the receptor enhances glucose-stimulated insulin release from pancreatic β-cells, but it is not known whether GPR142 has any further role beyond glycemic control. This limited effect may prevent GPR142 from being considered a suitable (or commercially viable) target for drug discovery. However, recent data suggest that serum tryptophan levels are increased in T2DM patients with a higher degree of insulin resistance (Chen et al., 2016), suggesting that inhibition of the receptor could prove beneficial. Furthermore, Gpr142^−/− mice display reduced levels of sensitivity to a collagen antibody-induced arthritis model of inflammation (Murakoshi et al., 2016). A similar effect was observed with a GPR142 antagonist (CLP-3094) (Murakoshi et al., 2016). Identification of better tool compounds will enable greater target validation and clarification of the pathophysiological processes controlled by GPR142.

K. GPRC5B

GPRC5B is a class C GPCR that was first cloned as a retinoic acid–inducible orphan receptor (Robbins et al., 2000); its ligand, if any, remains unknown. GPRC5B is widely expressed across different tissues at the level of mRNA, with highest levels in the CNS and lowest in liver and bone marrow (Regard et al., 2008). Although relatively little is known about the function of receptor, evidence has emerged that GPRC5B plays roles in different tissues relevant to metabolic function. GPRC5B expressed in pancreatic islets negatively regulates insulin secretion, potentially by reducing cell viability, as shown by cytokine challenge in murine insulin-secreting MIN6 cells (Soni et al., 2013). Control of Gprc5b expression is thought to skew macrophage populations to the M2, anti-inflammatory phenotype (Hohenhaus et al., 2013), and a copy number variant very close to Gprc5b has been shown to be associated with body mass index (Speliotes et al., 2010). Following this latter report, Gprc5b deficiency was found to protect mice from diet-induced obesity and insulin resistance and was associated with abolition of GPRC5B-dependent inflammation in white adipose tissue (Kim et al., 2012). GPRC5B interacts with and regulates Fyn kinase, which in turn controls proinflammatory IKK-ε–NF-κB signaling. These data converge to suggest that an antagonist of GPRC5B could be effective to reduce pathophysiology and symptomology of metabolic syndrome at a number of levels. However, to date, no known small-molecule ligands of the receptor have been discovered.

L. GPRC6A

GPRC6A is a class C GPCR that has been cloned from human, mouse, rat, goldfish, and bonobo chimpanzee. The receptor mediates responses to basic L-amino acids, such as arginine, ornithine, and lysine, which bind in the N-terminal (Venus flytrap) domain of the receptor (Wellendorph and Bräuner-Osborne, 2009) in a manner analogous to the binding of L-glutamate to metabotropic glutamate receptors.

In line with its amino acid pharmacology, GPRC6A was originally thought of as a nutrient-sensing receptor. Recent studies suggest that GPRC6A is involved in an array of metabolic, endocrine, and inflammatory processes (Pi et al., 2008, 2011, 2012; Wellendorph et al., 2009; Rossol et al., 2012; Clemmensen et al., 2013a,b; Smajilovic et al., 2013), although the evidence is equivocal. One Gprc6a^−/− mouse strain displays manifestations of metabolic syndrome, with increased fasted plasma glucose levels and impaired insulin sensitivity (Pi et al., 2008), and a polymorphism in the Gprc6a gene that is associated with insulin resistance in normal weight and obese subjects has been identified (Di Nisio et al., 2017). However, a different KO mouse strain has a much less pronounced phenotype, with no evidence of disrupted glucose handling or insulin sensitivity, and impairments in glucose metabolism only observed when the mice were fed a high-fat diet (Smajilovic et al., 2013).

Perhaps the biggest area of contention with respect to GPRC6A and metabolic disorders has been the enigmatic role of the bone-derived peptide, osteocalcin (OCN). The decarboxylated form of this osteoblast-derived protein exerts a number of beneficial metabolic and endocrine effects, including control of energy expenditure regulated by the pancreas (Lee et al., 2007); Ocn^−/− mice display reduced pancreatic cell proliferation, impaired glucose handling, and insulin resistance. Furthermore, systemic administration of OCN increases circulating insulin levels in wild-type but not Gprc6a^−/− mice (Pi et al., 2011). OCN has also been shown to increase pancreatic β-cell proliferation, insulin release from pancreatic islets, insulin sensitivity, energy expenditure, and testosterone synthesis in the gonads in mice (Lee et al., 2007; Oury et al., 2011). Furthermore, OCN-mediated insulin and GLP-1 release have also been reported in immortalized pancreatic β-cell lines, effects that are blocked by siRNA targeting Gprc6a (Pi et al., 2011; Mizokami et al., 2013). However, a recent study suggests that OCN does not increase GLP-1 release from enteroendocrine cells or insulin release from islets or immortalized β-cells (Rueda et al., 2016), adding further intrigue to the biology.

The positive effects of OCN, combined with the lack of OCN efficacy in Gprc6a^−/− mice, both in vivo and ex vivo, have led to suggestions that decarboxylated OCN acts directly via GPRC6A to exert its metabolic and endocrine effects. Despite this evidence, verification of a direct action of decarboxylated OCN on GPRC6A has proven somewhat elusive. Several reports have shown that OCN stimulates cAMP accumulation or extracellular signal-regulated kinase 1/2 phosphorylation in HEK293 cells recombinantly expressing GPRC6A (although the species tested is not always stated) (Pi et al., 2011, 2012; Pi and Quarles, 2012; Oury et al., 2013), suggesting that the signaling of the receptor may not be limited to Gα_q-coupled pathways originally reported for L-amino acids. Furthermore, recent reports have used computational approaches to predict the binding locus for OCN on GPRC6A (Pi et al., 2016) and suggested that OCN shares its binding site with testosterone (De Toni et al., 2016). However, two further laboratories have evaluated the pharmacology of mouse GPRC6A and failed to replicate either the agonist activity of OCN or GPRC6A-mediated cAMP signaling (Jacobsen et al., 2013; Rueda et al., 2016), leading both groups to conclude that the receptor is primarily a Gα_q-coupled, basic L-amino acid receptor.

Given the complexities of the pharmacology of GPRC6A, it is unclear whether targeting the receptor therapeutically could replicate the effects of OCN; small-molecule chemistry targeting the receptor has been restricted to a limited number of low-potency negative allosteric modulators, which target the transmembrane domain of GPRC6A (Gloriam et al., 2011; Johansson et al., 2015). A further challenge is the potential differences between orthologs of the receptor; a recent study explained the challenges of expressing and interrogating the pharmacology of the human receptor by virtue of an insertion/deletion variation in the third intracellular loop that leads to intracellular retention (Jørgensen et al., 2017). It remains unclear whether the pharmacology of the human receptor differs markedly from its murine ortholog—or indeed, what role the human receptor might have in vivo if not expressed at the cell surface.

M. Hydroxycarboxylic Acid Receptor 2 (GPR109A)

The receptor for nicotinic acid, or niacin, was identified in 2003 and subsequently shown to be highly expressed in adipocytes and immune cells (Wanders and Judd, 2011). Activation of hydroxycarboxylic acid receptor 2 (HCA₂) by niacin in adipocytes causes inhibition of lipolysis, but also via the epidermal Langerhans cells the harmless, yet unwanted side effect of flushing (Wanders and Judd, 2011). In addition to binding niacin, the receptor also binds the SCFA butyrate, suggesting that HCA₂ is involved in gut homeostasis, similar to the FFARs described above. KO mouse studies revealed a marked increase in adiponectin concentrations (Plaisance et al., 2009). Interestingly, niacin activation of HCA₂ stimulates sodium-glucose cotransporter 1 and GLUT2 jejunal glucose uptake, indicating potential additional benefits that would improve glycemic control (Wong et al., 2015).

The pursuit of HCA₂ agonists for the treatment of T2DM has recently increased, with a focus on identifying compounds without the flushing side effect. GSK256073, in a trial of 39 diabetic patients, caused a significant decrease in serum nonesterified fatty acids and glucose concentrations when dosed three times over 2 days (Dobbins et al., 2013). These improvements in glucose tolerance were attributed to an increase in insulin sensitivity with an associated decrease in insulin concentrations. Encouragingly, no flushing or adverse effects on the gastrointestinal tract were observed (Sprecher et al., 2015). However, recent data from a randomized 12-week trial revealed that GSK256073 did not meet its primary end point in reducing HbA1c compared with placebo (Dobbins et al., 2015). Lack of efficacy could be due to an inability of the compound to produce sustained suppression of plasma nonesterified fatty acid concentrations, thought to be required for the effects on insulin sensitivity.

However, a recent publication has questioned the validity of targeting HCA₂ (Fraterrigo et al., 2012). Niaspan, an extended release form of niacin, produced insulin resistance in skeletal muscle, most probably due to increased levels of adiponectin. These opposing effects suggest that further investigation of HCA₂ as a target is required. Its close homolog, HCA₃ (GPR109B), differs by only 15 amino acids and results from a gene duplication only present in humans (Ahmed et al., 2009). This receptor does not bind niacin with high affinity, but, like HCA₂, couples to Gα_i. 3-OH-octanoic acid is an endogenous ligand for HCA₃ (Ahmed et al., 2009), and activation of the receptor by this ligand controls adipocyte lipolysis by a negative feedback loop involving counteraction of prolipolytic stimuli in situations of increased fatty acid oxidation. Targeting HCA₃ may therefore be an alternative approach to HCA₂, especially if the HCA₂-mediated side effects can be avoided.

N. Melatonin Receptors (MT₁R and MT₂R)

Melatonin (MT) is primarily secreted by the pineal gland during the dark phase, but can also be secreted to a lesser extent from the innate immune system, the gastrointestinal tract, and the retina. The effects of MT are mediated by two homologous but distinct GPCRs, namely MT₁R and MT₂R, encoded by two distinct genes, MTNR1A and MTNR1B, respectively. MTs are Gα_i-coupled and expressed in a cell-specific fashion in mice and in humans (see Jockers et al., 2016 for recent review). MTs are involved in numerous physiologic and neuroendocrine functions. Interest in MT in the pathogenesis of T2DM derived initially from three independent genome-wide association studies that led to the identification of several frequent polymorphisms located near the MTNR1B gene that were associated with increased fasting plasma glucose, a reduction of early insulin response to glucose, and ultimately an increased risk of developing T2DM (Bouatia-Naji et al., 2009; Lyssenko et al., 2009). This genetic association is very robust and was subsequently replicated by several groups in other populations (Ronn et al., 2009; Renström et al., 2015). Surprisingly, the influence of MT on insulin secretion is still debated, with conflicting results generated in vitro using rodent and human islets (Lyssenko et al., 2009; Costes et al., 2015). Furthermore, MT receptors are expressed at low levels in both α- and β-cells. In a recent study, Tuomi et al. (2016) demonstrated elegantly that the common rs10830963 variant located in the MTNR1B intron is an expression quantitative trait locus conferring increased expression (two- to fourfold) of the receptor in islets, as measured by RNA Seq in 204 Scandinavian donors, whereas MTNR1A expression is not altered by either genotypes (GG versus CC).

MT inhibits glucose-stimulated insulin secretion in INS-1 cells overexpressing MTNR1B/MT₂R by, at least in part, reducing glucose-stimulated cAMP production. Pertussis toxin treatment completely abolished this effect (Tuomi et al., 2016). The authors also characterized in depth the phenotype of Mt2^−/− KO mice. These mice display increased insulin release in response to an intravenous glucose tolerance test and impaired hepatic insulin sensitivity, as determined by hyperinsulinemic-euglycemic clamp. This exacerbated insulin response was documented both ex vivo and in vivo. Increased β-cell mass and increased cAMP levels in response to glucose stimulation are the main reasons for this altered insulin response. Tuomi et al. (2016) showed that MT treatment at bedtime for 3 months in nondiabetic individuals resulted in a significant reduction in insulin secretion in GG allele carriers (Tuomi et al., 2016). It is noteworthy that sleep duration and quality were similar in both groups in response to MT treatment. Another group reported a negative effect of acute MT treatment on glucose clearance during a 2-hour oral glucose tolerance test in a genotype-specific fashion in healthy young female athletes (Garaulet et al., 2015). Taken together, these studies suggest that increased MT signaling is a risk factor for T2DM. However, analysis of rare loss-of-function MTNR1B variants suggested that reduced MT signaling increases T2DM risk (Prokopenko et al., 2009; Bonnefond et al., 2012). Furthermore, reduced MT secretion is associated with an increased risk for diabetes in the general population (McMullan et al., 2013). Although it appears difficult to assign a definitive role of MT on insulin secretion from all these human and mouse genetic studies, the role of MT and its receptors in the CNS on energy and glucose homeostasis should be further investigated because sleep restriction and circadian misalignment are recognized as major risk factors for the development of cardio-metabolic diseases (Scheer et al., 2009). Indeed, Lane et al. (2016) recently reported that the MTNR1B risk allele influences dynamics of MT secretion with consequences on the sleep/wake cycle. Nevertheless, the species-specific receptor distribution and downstream biology complicate translation of the preclinical pharmacology into the clinical space for cardio-metabolic diseases.

A. C-C Motif Chemokine Receptor 2

CCR2 belongs to the C-C motif chemokine receptor (CCR) and C-X-C motif chemokine receptor superfamily. These receptors are involved in many inflammatory responses and are expressed on multiple inflammatory cell types, including monocytes, macrophages, T-cells, and B-cells. However, these receptors are also widely expressed in the periphery, with some expression also located in the CNS. Like many chemokine receptors, CCR2 is activated by multiple agonists, including CCL7 (MCP-3), CCL8 (MCP-2), CCL11 (eotoxin), and CCL13 (MCP-4), which all bind with similar low nanomolar affinities. However, the principal agonist for CCR2 is CCL2 (MCP-1), which binds with sub-nM affinity. Many of these agonists also bind to multiple receptors, for example, CCL2 binds with low affinity (>1000-fold compared with CCR2) to CCR4. This multireceptor, multiligand behavior suggests that a large degree of redundancy must occur across the biologic functions (Boring et al., 1997). It has been shown that the different endogenous ligands for CCR2 are able to bind and stabilize different confirmations of the receptor, enabling the activation of both G protein–dependent and independent pathways (Berchiche et al., 2011). In addition, multiple binding sites have been identified using radiolabeled antagonists (Zweemer et al., 2013). Recently, the crystal structure of inactive CCR2 was solved using a T4L fusion construct (Zheng et al., 2016) in a ternary complex with both an orthosteric (BMS-681) and an allosteric [CCR2-RA-(R)] antagonist, indicating that the receptor is able to accommodate multiple ligands at any one time. The next challenge is to identify whether the receptor could accommodate multiple endogenous agonists simultaneously, thus giving an insight into redundancy among chemokine receptors/ligands. However, this may prove too challenging, as to date over 90% of all GPCR structures have been solved in the inactive confirmation, with either an antagonist and/or an allosteric modulator bound. Of note, the double-antagonist CCR2 crystal structure is the most inactive, therefore most stable, structure solved (Zheng et al., 2016). Therefore, a double-agonist bound structure may yet be outside the current technical capabilities.

During infection, activation of the innate immune response occurs, and binding of CCL2 to CCR2 causes recruitment of Ly6C^hi monocytes. The levels of expression of CCL2 rapidly increase within inflamed tissues, and its release is further stimulated by the release of proinflammatory mediators (Shi and Pamer, 2011), although the mechanism involved is largely unknown. It has been suggested that CCL2 produces a chemokine gradient toward the site of inflammation, enabling the recruitment of circulating monocytes (Shi and Pamer, 2011). In addition, the levels of CCR2 expression are also modulated during tissue inflammation, by granulocyte-macrophage colony-stimulating factor (Croxford et al., 2015). Therefore, the increased levels of both receptor and ligand help resolve the site of infection.

It is unsurprising, based on the expression profile of CCR2, that this receptor has been implicated in the pathophysiology of T2DM. In Ccr2^−/− mice fed a high-fat diet, a reduction in food intake, macrophage recruitment, inflammatory markers in adipose tissue, and an improvement in glucose homeostasis and insulin sensitivity have been reported (Weisberg et al., 2006). A similar phenotype has also been described with siRNA knockdown of CCR2 (Kim et al., 2016). In global Ccr2^−/− and hematopoietic (bone marrow) Ccr2^−/− mice fed a high-fat diet for 12 weeks, flow cytometry revealed differences in the F4/80⁺ myeloid cell population compared with their wild-type counterparts. In the adipose tissue from all cohorts, CD11b^hiF4/80⁺ cells were detected; however, in both KOs, a population of CD11b^loF4/80^lo was seen. It was only when this population of cells was absent from adipose tissue, following 20 weeks of the high-fat diet, that improvements in adipose tissue inflammation were detected (Gutierrez et al., 2011). Loss of CCR2 resulted in the accumulation of CD11b^loF4/80^lo cells in adipose tissue, whereas in wild-type mice transient expression of these cells was identified. No explanation was given as to why these cells disappeared with increased feeding. Further data reveal that migration of monocytes from bone marrow is dependent on both CCR2 and CCL2, whereas migration of monocytes from the circulation into inflamed tissues appears to be dependent on CCR2, CCL2, and CCL7 (Tsou et al., 2007). To further explore monocyte migration, Kawano et al. (2016) generated colonic macrophage-specific Ccr2^−/− (M-Ccr2KO) and intestinal epithelial cell–specific tamoxifen-inducible Ccl2^−/− (Vil-Ccl2KO) mice. Although both strains gained significant body weight when fed a high-fat diet, only the Ccr2^−/− displayed improved glucose tolerance and insulin sensitivity. Interestingly, in a whole-body Ccl2^−/−, when fed a similar diet, an increase in adiposity and metabolic dysfunction was observed, suggesting that global CCL2 is required for the protection of sensitive metabolic tissues (Cranford et al., 2016).

As there is clear redundancy in the immune system and receptor/ligand polygamy, targeting chemokine receptors therapeutically, especially with small molecules, may appear extremely challenging. However, recent data have shown some promise with CCR2 antagonists for the treatment of T2DM and its associated complications, nonalcoholic fatty liver disease/NASH. In mice fed a high-fat diet, and in db/db mice, CCX140-B (ChemoCentryx) fully blocked the recruitment of inflammatory macrophages into adipose tissue, resulting in an improvement in glucose homeostasis, insulin sensitivity, hepatic glycogen, and triglyceride content (Sullivan et al., 2013a). Similar effects were observed in human CCR2 knock-in mice, with no increase in circulating CCL2 levels in plasma, or increased blood monocyte numbers (Sullivan et al., 2013b). Encouragingly, treatment of 110 patients with T2DM and nephropathy in a phase II trial with CCX140-B resulted in renoprotection and improved glucose handling, although there was no overall improvement in insulin resistance (de Zeeuw et al., 2015). As no improvement in insulin resistance was seen, the compound is now being further investigated in models of chronic kidney disease. Similarly, use of the Spiegelmer, NOX-E36 (Noxxon; emapticap pegol), which neutralizes CCL2, has shown promise in a phase IIa exploratory study in patients with T2DM and albuminuria (Menne et al., 2017) and in mouse models of NASH and chronic hepatic injury (Baeck et al., 2012). Finally, use of a dual CCR2/CCR5 antagonist Cenicriviroc (Allergan) has shown potent anti-inflammatory and antifibrotic effects in animal models of NASH and fibrosis (Lefebvre et al., 2016). A phase IIb combination study with the farnesoid X receptor agonist LJN 452 (Novartis) is planned to assess safety, efficacy, and tolerability. Results of these studies may further enforce the rationale for targeting CCR2 for the treatment of T2DM and chronic liver diseases.

B. FFAR4 (GPR120)

FFAR4 is a metabolite-sensing FFA GPCR activated by unsaturated long-chain FFAs (Hirasawa et al., 2005), leading to the release of GLP-1 and an increase in circulating insulin. Omega-3 polyunsaturated fatty acids display the highest potency for FFAR4 in vitro (Im, 2016), but other endogenous ligands also activate FFAR4, including docosahexaenoic acid and eicosapentaenoic acid (Oh et al., 2010; Kim et al., 2015). FFAR4 exerts its biologic effects by both G protein–dependent and independent pathways. Coupling to Gα_q causes release of intracellular Ca² to promote the release of cholecystokinin and GLP-1. In addition, activation of extracellular signal-regulated kinase phosphorylation and phosphoinositide 3-kinase leads to AKT phosphorylation and GLUT4 translocation, leading to the regulation of adipogenesis and adipogenic differentiation (Li et al., 2015a). Moreover, FFAR4 is expressed specifically on mouse pancreatic δ-islet cells, but not on β-islet cells, and activation of FFAR4 in these cells regulates the secretion of somatostatin (Stone et al., 2014).

FFAR4 may also inhibit inflammation promoted by other mechanisms. For example, activation of Toll-like receptor 4 and TNF receptor by lipopolysaccharide (LPS) and TNF-α, respectively, causes TGF-β–activated kinase 1 binding protein and TGF-β–activated kinase 1 to colocalize and activate the IKK, JNK, and NF-κB pathways, leading to inflammation (Oh et al., 2010; Talukdar et al., 2011; Li et al., 2015a; Liu et al., 2015b; Ulven and Christiansen, 2015). However, agonist activation of FFAR4 causes recruitment of β-arrestin 2, leading to the sequestration of TGF-β–activated kinase 1 binding protein away from TGF-β–activated kinase 1 and an anti-inflammatory effect (Oh et al., 2010), suggesting that signaling bias may exist for this receptor, and this may provide a means of inhibiting inflammation.

FFAR4 is highly expressed in adipose tissue, proinflammatory macrophages, and eosinophils (Konno et al., 2015), suggesting that it may provide a link between metabolic diseases and inflammation (Oh et al., 2010). In the monocyte RAW 264.7 cell line and mouse intraperitoneal macrophages, FFAR4-mediated anti-inflammatory effects, including reduction of proinflammatory M1, increased anti-inflammatory M2 macrophage levels and reduced cytokine levels that were abrogated when FFAR4 was knocked down using siRNA. Similar effects were observed in Ffar4^−/− mice on a high-fat diet with or without omega-3 polyunsaturated fatty acid supplementation. The KO phenotype indicated that a FFAR4 agonist may alleviate insulin resistance. GW9508, which has FFAR1 and FFAR4 agonist activity (Briscoe et al., 2006), inhibited LPS-induced phosphorylation of JNK, IKKβ, Iκβ degradation, cytokine secretion, and inflammatory gene expression in RAW 264.7 cells. A FFAR4-selective agonist Cpd A (Oh et al., 2014) reduced chronic inflammation in obese mice, but also improved insulin sensitivity, glucose tolerance, and decreased hyperinsulinemia. Other FFAR4 agonists have since been developed and display similar attributes (Li et al., 2015a; Liu et al., 2015b; Ulven and Christiansen, 2015). However, FFAR4 desensitizes rapidly upon chronic activation, which has hampered the progression of FFAR4 agonists in the clinic (Alvarez-Curto et al., 2016). Furthermore, increased FFAR4 expression and signaling both in vitro and in vivo are associated with an increase in colorectal carcinoma tumor growth, promotion of angiogenesis, increased motility, and induced epithelial–mesenchymal transition of human colorectal carcinoma cells (Wu et al., 2013). Further studies are therefore required to delineate the effects of FFAR4 agonist activity on tumor progression before committing to human studies.

C. GPR21

GPR21 is an orphan receptor that couples to Gα_q (Bresnick et al., 2003), but as no ligand, synthetic or endogenous, has yet been identified, further conclusions about signaling are difficult to make. It has been suggested that the endogenous ligand is a FFA (Kakarala and Jamil, 2014), and/or could be found within serum as serum-starved GPR21-expressing cells display reduced AKT phosphorylation (Leonard et al., 2016). However, no strong experimental evidence has emerged to confirm these findings.

Two independent studies using Gpr21^−/− mice indicate that GPR21 may be involved in obesity-induced insulin resistance (Gardner et al., 2012; Osborn et al., 2012). In the first study, Gpr21^−/− mice showed a robust improvement in glucose tolerance and systemic insulin sensitivity after a high-fat diet, culminating in a modest lean phenotype that was suggested to be due to a reduction in tissue inflammation caused by a decrease in the migratory ability of macrophages into adipose tissue (Osborn et al., 2012). To examine these effects in a human model, lentiviral short hairpin RNA knockdown of GPR21 in a human monocyte-like cell line (U-937) was performed and showed similar inhibition of migration. In the second study, there was a clear reduction in proinflammatory mediators in Gpr21^−/− mice, along with an improvement in glucose tolerance and systemic insulin sensitivity after a high-fat diet (Gardner et al., 2012). However, subsequently, the same group generated an alternative KO mouse model using transcription activator-like effector nucleases (TALENS) technology (Wang et al., 2016b). This was necessary because the Gpr21 gene resides within the intron of the RAB GTPase-activating protein 1 (Rabgap1) gene, and the authors showed that this gene was disrupted in the original Gpr21^−/− mice. Although there was no effect on the expression of Rabgap1 in the TALENS Gpr21^−/− KO, no improvement in glycemic control was reported. Why no beneficial phenotype was observed is unclear, and no head-to-head comparison between the two KO strains was made, nor was a high-fat diet regimen tested in the TALENS KO, making interpretation of these observations challenging. However, although GPR21 appears to be a promising target, the orphan status, and the lack of any suitable tool molecule, inhibits further investigation.

D. GPR35

GPR35 is an orphan GPCR that is highly expressed in the pancreas, immune cells, and to a lesser extent adipose tissue, liver, and skeletal muscle (MacKenzie et al., 2011). Recently, it has been reported that GPR35 can be activated by the chemokine CXCL17 (Maravillas-Montero et al., 2015). Based on this finding, it was proposed that the receptor be renamed C-X-C motif chemokine receptor 8 (Maravillas-Montero et al., 2015). Multiple other endogenous ligands have been associated with GPR35, including kynurenic acid and 2-oleoyl lysophosphatidic acid (Shore and Reggio, 2015). In addition, a vast array of synthetic agonists and antagonists has been developed for this receptor (for a summary, please see MacKenzie et al., 2011). To further add complexities to the pharmacology of GPR35, the concept of ortholog selectivity and ligand bias has been demonstrated for this receptor (Milligan, 2011). Although the expression profile of GPR35 would support its potential as a suitable target for obesity-induced T2DM, little experimental data exist to support this theory. It has been shown that CXCL17 causes the migration of macrophages from the lungs of wild-type mice, but not from Cxcl17^−/− mice, indicating a role in inflammation (Maravillas-Montero et al., 2015). However, no further investigation into the potential metabolic effect of GPR35 was performed. Further studies are required to elucidate the potential of GPR35, such as a high-fat diet study in Gpr35^−/− animals. Unfortunately, due to the complex pharmacology and multiple endogenous ligands known, interest in this receptor is perhaps limited.

E. GPR84

Medium-chain fatty acids (MCFAs), including capric, undecanoic, and lauric acid, bind and activate GPR84, resulting in Ca²⁺ mobilization and inhibition of cAMP (Wang et al., 2006). These responses are pertussis toxin sensitive, indicating that GPR84 couples to Gα_i/o. GPR84 is predominantly present on immune cells and more particularly in monocytes, macrophages, and neutrophils (Wang et al., 2006). In the periphery, GPR84 is mainly expressed in bone marrow, spleen, lung, and peripheral blood leukocytes. In the CNS, GPR84 expression is restricted to microglia (Bouchard et al., 2007). Interestingly, MCFAs caused significant upregulation of the LPS-induced proinflammatory cytokine IL-12 p40 (IL12B), which has a pivotal role in promoting cell-mediated immunity by inducing and maintaining Th1 cell responses and inhibiting Th2 response. In agreement with these results, Gpr84^−/− mice had increased levels of Th2 cytokines (Wang et al., 2006). Furthermore, under proinflammatory conditions, Gpr84 expression in macrophages is increased (Talukdar et al., 2011) and regulates TNF-α secretion from LPS-stimulated macrophages (Müller et al., 2017). Moreover, Nagasaki et al. (2012) demonstrated, using a 3T3-L1 (adipocyte) and RAW264.7 (macrophage) coculture, a clear upregulation in Gpr84 expression in the presence of TNF-α, with a concomitant downregulation of adiponectin, and concluded that the release of TNF-α from infiltrating macrophages caused the increase in Gpr84 expression.

To add further weight to this hypothesis, Suzuki et al. (2013) demonstrated, using MCFAs and a surrogate agonist, 6-n-octylaminouracil, that GPR84 agonists caused chemotaxis of human leukocytes and macrophages. Likewise, LPS stimulation increased the release of proinflammatory cytokines, IL-8 and TNF-α, in these human leukocytes and macrophages, respectively. Finally, Gpr84 was shown to be upregulated under inflammatory conditions in both the microglia and astrocytes (Madeddu et al., 2015). In line with these observations, Gpr84 gene deficiency significantly reduces microgliosis (Audoy-Remus et al., 2015). Because several groups have established a role for glia in energy balance regulation and obesity pathogenesis (see Douglass et al., 2017 for review), it is tempting to speculate that GPR84 may have a metabolic role under inflammatory conditions. Although all these studies suggest a link between GPR84 and chronic low-grade inflammation, its role in energy and glucose homeostasis remains unclear. Surprisingly, recent studies performed by Du Toit et al. (2017) indicate that Gpr84 gene deficiency does not significantly affect glucose or energy homeostasis in response to a low-chain fatty acid (LCFA)- or MCFA-enriched diet. Interestingly, the authors found that Gpr84^−/− KO mice displayed significantly increased liver triglyceride levels in response to the low-chain fatty acid diet, although infiltration of immune cells in the liver as well as expression markers of inflammation under these conditions was unfortunately not documented. Whether the exacerbated steatosis observed in Gpr84^−/− KO animals is due to cross-talk between hepatocytes and immune cells remains to be established. Further studies using KO mice and bone marrow transplantations are required to determine whether there is a major role for GPR84 in metabolic disease.

F. Leukotriene BLT₁ Receptor

Leukotriene B₄ (LTB₄) is a proinflammatory cytokine produced from arachidonic acid (Tager and Luster, 2003) that binds with high and low affinity to the leukotriene B4R1 (BLT₁) and B4R2 (BLT₂) receptors, respectively. In addition, other lipoxygenase and thromboxane synthase products, including 12-hydroxyeicosatetraenoic acid and 12-hydroxyheptadecatrienoic acid, bind and activate the BLT₂ receptor. In a recent study, Blt₁^−/− mice placed on a high-fat diet had reduced levels of CD11b⁺ monocytes in adipose tissue compared with the wild-type controls, a clear reduction in M1 proinflammatory adipose tissue macrophages, and a reduction in the expression of proinflammatory chemokine genes, including IL6 and Ccl2 (Spite et al., 2011). Deletion of the BLT₁ receptor reduced inflammation in adipose tissue and liver, resulting in protection from hepatic steatosis and insulin resistance. This suggested a novel role for the BLT₁ receptor in monocyte chemotaxis, culminating in high levels of chronic inflammation and obesity.

It has also been shown that LTB₄ causes insulin resistance in obese mice associated with enhanced macrophage chemotaxis, stimulation of proinflammatory pathways, reduced insulin-stimulated glucose uptake into myocytes, and impaired insulin-mediated suppression of hepatic glucose secretion in mouse hepatocytes (Li et al., 2015b). All of these effects were reversed or severely blunted in the Blt₁^−/− mouse. These studies suggest that BLT₁ receptor inhibition may be a useful way of preventing obesity-induced T2DM. However, to date, all studies have been performed using only Blt₁^−/− mice. As LTB₄ also binds and activates the BLT₂ receptor, further studies are required to understand whether a compensatory mechanism is present, using the Blt₂^−/−, or a double Blt₁^−/−/Blt₂^−/− KO. Finally, whereas the BLT₁ receptor appears to couple exclusively to Gα_i proteins, the BLT₂ receptor couples to both Gα_i and Gα_q proteins (Yokomizo et al., 2001). Therefore, perhaps specifically inhibiting the activation of the Gα_i pathway via the BLT₁ receptor may be sufficient to obtain the desired anti-inflammatory phenotype.

G. SUCNR1 (GPR91)

SUCNR1 is expressed on immature dendritic cells and macrophages. The receptor is activated by succinate, which is generated from citrate via the tricarboxylic acid cycle. Citrate is involved in the production of prostaglandin and fatty acid synthesis. Under homeostasis, M2 macrophages break down succinate by succinate dehydrogenase (SDH). However, under levels of high inflammation, SDH is inhibited by itaconate, an antimicrobial metabolite (Nemeth et al., 2016), and inhibition of SDH causes accumulation of succinate within the macrophage. This accumulation adds to the inflammatory phenotype by increasing LPS-induced hypoxia-inducible factor 1α equilibrium, resulting in an increase in IL-1β expression (Van den Bossche et al., 2017). Activation of SUCNR1 causes maturation and migration of these immune cells into metabolic tissues, further increasing the production of IL-1β (Rubic et al., 2008). Moreover, activation of SUCNR1 by succinate causes a SUCNR1-dependent recycling of the agonist fueling the inflammation in a paracrine mechanism, a concept termed immunometabolism (Littlewood-Evans et al., 2016). In differentiated bone marrow macrophages obtained from wild-type and Sucnr1^−/− mice, macrophage migration and IL-1β production were abrogated in the KO animals (Littlewood-Evans et al., 2016). However, Sucnr1^−/− mice fed a high-fat diet for 16 weeks exhibited improved insulin resistance (Carmone et al., 2015; van Diepen et al., 2017), whereas a similar study only showed an initial improvement, with the animals becoming progressively hyperglycemic (McCreath et al., 2015). Finally, in another study, Sucnr1^−/− mice developed dry age-related macular degeneration (Favret et al., 2013). Due to these conflicting data sets, interest in targeting SUCNR1 for metabolic disorders may be limited.