Pharmacological Advances in Incretin-Based Polyagonism: What We Know and What We Don’t

Abstract

The prevalence of obesity continues to rise in both adolescents and adults, in parallel obesity is strongly associated with the increased incidence of type 2 diabetes, heart failure, certain types of cancer, and all-cause mortality. In relation to obesity, many pharmacological approaches of the past have tried and failed to combat the rising obesity epidemic, particularly due to insufficient efficacy or unacceptable side effects. However, while the history of antiobesity medication is plagued by failures and disappointments, we have witnessed over the last 10 years substantial progress, particularly in regard to biochemically optimized agonists at the receptor for glucagon-like peptide-1 (GLP-1R) and unimolecular coagonists at the receptors for GLP-1 and the glucose-dependent insulinotropic polypeptide (GIP). Although the GIP receptor:GLP-1R coagonists are being heralded as premier pharmacological tools for the treatment of obesity and diabetes, uncertainty remains as to why these drugs testify superiority over best-in-class GLP-1R monoagonists. Particularly with regard to GIP, there remains great uncertainty if and how GIP acts on systems metabolism and if the GIP system should be activated or inhibited to improve metabolic outcome in adjunct to GLP-1R agonism. In this review, we summarize recent advances in GLP-1- and GIP-based pharmacology and discuss recent findings and open questions related to how the GIP system affects systemic energy and glucose metabolism.

Identification of GIP and GLP-1 as Incretin Hormones

Starting with the observation in 1906 that intestinal mucosal extracts decrease glucosuria in diabetic patients (1), the intestine has long been recognized to control glucose metabolism (2, 3). A seminal discovery was the demonstration in 1963–1964 that the glucose-induced rise in plasma insulin is much greater when glucose is given orally relative to its infusion into the general circulation (4–6), an effect henceforth known as the incretin effect. In 1969–1970, Brown et al. (7, 8) purified a substance from a crude mucosal cholecystokinin (CCK) preparation that showed inhibitory action on gut motility and gastric acid secretion. Based on its ability to inhibit gastric acid secretion, the substance was named gastric inhibitory polypeptide (GIP) (9). In 1973, GIP was then shown to accelerate glucose-induced insulin secretion in healthy humans (10). Since the insulinotropic action of the peptide, rather than gastric mobility inhibition, prevailed at physiological doses (9–14), the peptide was renamed to glucose-dependent insulinotropic polypeptide (15). In the early 1980s, Habener et al. (16–18) then identified a glucagon-like sequence in the anglerfish preproglucagon cDNA, soon followed by the identification of two glucagon-like peptides within the preproglucagon sequence of rats (19, 20), hamsters (21), and humans (22). The novel peptides showed ∼50% sequence homology to glucagon and were succinctly named glucagon-like peptide-1 and -2 (GLP-1 and GLP-2) (21). Studies by Mojsov et al. (23) and Holst et al. (24) subsequently showed that proglucagon processing results in different forms of GLP-1 in the intestine and the pancreas and that two NH₂-terminally truncated intestinal-originating forms, GLP-1(7–37) and GLP-1(7-36NH₂), stimulate insulin secretion in the isolated perfused pancreas of pigs (24) and rats (25). Soon after the demonstration by Drucker et al. (26) that GLP-1 autonomously acts on the pancreatic β-cells to potentiate glucose-induced insulin secretion, Bloom et al. (27) then confirmed in humans that GLP-1 is a physiological incretin hormone. GLP-1 and GIP jointly account for the vast majority of the incretin effect, as verified by a largely blunted incretin effect after adjunct antagonization (28) or concomitant deletion (29) of the GLP-1 and GIP receptors (GIPRs) in mice.

Notably, while the incretin effect accounts for 50–70% of the insulin secretory response to oral glucose in healthy humans (30–34), this effect is largely blunted in individuals with type 2 diabetes (T2D) (30–33), an observation that is primarily attributed to a diminished insulinotropic action of GIP (35–41). Consequential to this, although the insulinotropic action of GIP is restored upon normalization of glycemia (42), the pharmacological interest in GIP has ever since been overshadowed by the pharmacology of GLP-1, which over the next decades emerged as a pleiotropic hormone with therapeutic value far beyond its initially described role as an insulin secretagogue. The most prominent extrapancreatic effect of GLP-1 is its ability to act on brain feeding centers to decrease body weight via inhibition of food intake, an effect first described in rats (43, 44) and subsequently confirmed in mice, birds, nonhuman primates, and humans (45). GLP-1 receptor (GLP-1R) agonists act in the brain to mediate homeostatic and hedonic feeding, with particular importance of the hypothalamic arcuate nucleus, paraventricular nucleus, hindbrain area postrema/nucleus of the solitary tract, parabrachial nucleus, and the hippocampus (46, 47). Consistent with the key role of central nervous system (CNS) GLP-1R signaling in the control of energy metabolism, GLP-1R agonists fail to affect food intake after either antagonization (48) or genetic deletion (49) of central GLP-1R in rodents. Agonism at GLP-1R in pre- and postsynaptic neurons within these aforementioned brain regions modulates the activity of the intracellular energy sensor AMP-activated kinase (AMPK), promotes enhanced neural depolarization and firing rate, and may stimulate the trafficking of alternative receptor profiles to the plasma membrane for outcome-specific optimization (50–53). These diverse mechanisms, although unknown at the time, facilitate the CNS-mediated satiety effects of exogenously administered GLP-1 in mice and obese individuals (44, 54–56).

The Use of GLP-1R Agonists for the Treatment of Obesity and Diabetes

Endogenous GLP-1 has a short circulating half-life of ∼1–2 min, which is primarily owed to proteolytic inactivation by dipeptidyl peptidase 4 (DPP4) and rapid renal elimination (57, 58). The short half-life severely limits the therapeutic potential of native GLP-1 to improve glucose metabolism, and even more so its antiobesity indication. However, in 1992, Eng et al. (59) discovered a potent GLP-1 paralog, exendin-4, in the saliva of Heloderma suspectum that exhibited improved circulating stability and proteolytic protection due to a unique alanine to glycine substitution at the 2nd N-terminal position of the peptide, resulting in a half-life of ∼2 h (60). The improved half-life of exendin-4 led to the pharmaceutical development of an injectable application, which demonstrated antidiabetic and certain satiety-inducing effects in patients with T2D (61–64). A variety of biochemically optimized GLP-1R agonists were subsequently developed, using alternative half-life extension strategies that include increasing the size of the peptide through either linkage of two GLP-1 molecules (dulaglutide and albiglutide), or via fatty acid acylation of the glp-1 backbone (liraglutide and semaglutide). Such fatty acid acylation allows the peptide to reversibly bind to circulating albumin, with the consequence of delayed renal excretion and the unique ability of albumin to escape endosomal degradation (65). In the case of liraglutide, C16 fatty acid acylation of the GLP-1 backbone allowed for once-daily (QD) dosing and for a greater focus on achieving antiobesogenic endpoints (66–69). Liraglutide was registered in 2014 for the treatment of obesity in adults and in 2020 for obesity in adolescents aged 12–17 years. When given over 56 weeks at the indicated dose of 3 mg QD in obese patients without diabetes, weight loss attributed to liraglutide is still in the single-digit range, but with 33% and 14% of patients losing >10% and >15% body weight, respectively (70). Semaglutide, an “advanced” form of liraglutide, differs from liraglutide in that it contains a nonnatural amino acid [aminoisobutyric acid (AIB)] at position 2 to protect from DPP4 recognition and a C18 fatty diacid to further enhance albumin binding (65, 71). These modifications result in a circulating half-life of ∼160 h and hence allow for once-weekly (QW) dosing. Semaglutide (2.4 mg) was approved by the FDA for the management of obesity in 2021 and undoubtedly has performed successfully on antidiabetic and antiobesogenic endpoints (72, 73). In phase III trials, semaglutide (2.4 mg QW treatment) over 68 weeks decreased body weight in obese patients without diabetes by −14.9%, relative to −2.4% in placebo controls (73). The continued maintenance of body weight reduction is reliant on continued treatment, as treatment discontinuation results in a rapid rebound of body weight (74). Yet impressively, the SELECT phase III trial demonstrates continued semaglutide treatment (2.4 mg) to maintain an approximate −10% reduction in body weight even over the course of up to 221 weeks, indicating the lack of tachyphylaxis and sustained weight loss during long-term treatment (75). This long-term reduction in body weight during continued semaglutide treatment is associated with decreased occurrence of fatal cardiovascular (CV) events in nondiabetic obese patients with preexisting CV disease (75). Although Semaglutide sets the stage for a new era in antiobesity medication, its ability to decrease body weight in obese patients with concurrent diabetes is still limited, with placebo-corrected weight loss, after 68 weeks of treatment, still in the single-digit range (76). In summary, biochemically engineered long-acting GLP-1R agonists testify as a premier tool for the treatment of obesity and diabetes, but the weight loss efficacy of GLP-1R monoagonism remains limited in patients with concurrent obesity and diabetes.

Is GIPR Signaling a Valuable Pharmacological Target for the Treatment of Obesity and Diabetes?

Overshadowed by the pharmacological success of biochemically optimized GLP-1R agonists, GIPR agonism has traditionally been granted very little, to no, pharmacological value for the treatment of obesity and diabetes (45, 77). This is not only attributed to the dampened insulinotropic action of the hormone under conditions of T2D (35–41) but also to the observation that GIPR-deficient mice are protected from diet-induced obesity (78–81). In line with this observation are reports showing that GIP, particularly under low to absent levels of insulin, promotes lipogenesis and adipocyte lipid deposition via increased action/secretion of lipoprotein lipase (82–84). GIP further promotes insulin-induced glucose uptake (85–87) and conversion of glucose into lipids (85) by stimulating adipocyte GLUT4 translocation (87) and by increasing adipocyte insulin receptor affinity (85–87). In line with these data indicating that GIP has an energy-conserving nature, genetic, or pharmacological inhibition of GIPR signaling decreases intramuscular lipid accumulation in aged mice (78), and genome-wide association studies have identified common genetic loss-of-function variants in the GIP receptor (GIPR) to be associated with both higher and lower body mass indexes (88–90). Consistent with this is the observation that certain GIPR antagonists decrease body weight and food intake in diet-induced obese (DIO) mice and nonhuman primates (91), particularly when given in adjunct to GLP-1R agonism (91, 92). However, while these data argue that the GIP receptor should be inhibited rather than activated for the treatment of obesity, in the absence of insulin (or under baseline insulin levels) GIP stimulates lipolysis in isolated rat adipocytes (93) and differentiated 3T3L1 adipocytes (94) and in humans with T1D (95). This lipolytic action of GIP is mediated via its ability to promote cAMP production and can be antagonized by the addition of insulin, which inhibits cAMP production, or upon direct inhibition of adenylate cyclase (94). In line with this notion, GIPR is a Gα_s-coupled G protein-coupled receptor and hence not only stimulates cAMP production in the pancreatic β-cells (96, 97) but also in insulin-deprived isolated rat adipocytes (93) and differentiated 3T3L1 adipocytes (94). Consistent with the ability of GIP at the Gα_s-coupled GIPR to promote cAMP-driven lipolysis under low insulin conditions, adipocyte-specific DREADD-mediated Gα_s activation similarly reduces body fat and enhances adipocyte lipolysis (98). A single administration of a long-acting fatty acid acylated (acyl) GIP further induces fatty acid oxidation in DIO mice (99), and mice with overexpression of GIP are lean and show decreased fat mass when chronically exposed to a high-fat diet (HFD) (100). In summary, GIP differentially acts on the adipose tissue by stimulating lipolysis under baseline conditions but by accelerating the antilipolytic effect of insulin under conditions of hyperinsulinemia (FIGURE 1).

FIGURE 1.

Schematic of different activation statuses of Gα_s-coupled G protein-coupled receptors (GPCR) in white adipose tissue, and the subsequent metabolic effects A: agonism of the Gα_s pathway, as mediated by a white adipocyte-specific Gα_s-coupled DREADD receptor, leads to enhanced cAMP/PKA mediated lipolysis and decreased lipogenesis in DIO mice. B: agonism of the Gα_s pathway, as mediated by white adipocyte-specific glucose-dependent insulinotropic polypeptide receptor (GIPR), has resulted in contradictory results, with reports of enhanced lipolysis and decreased lipogenesis akin to the Gα_s-coupled DREADD receptor, and opposing reports of enhanced lipogenic and decreased lipolytic effects; additionally, there has been conflicting evidence whether GIPR agonism in the white adipocyte increases intracellular cAMP. C: antagonism of the GIPR, which prevents ligand binding and subsequent Gα_s pathways activation, has suprisingly been linked to increases in lipolysis and decreases in lipogenesis. AC, adenylate cyclase.

Notably, data related to GIPR signaling in the adipose tissue are not undisputed, given that mice with targeted deletion of Gipr in the adipose tissue do not recapitulate the obesity-protecting phenotype seen in mice with global Gipr deficiency (101–103). Reasonable evidence further indicates that Gipr in the heterogenous adipose tissue is predominantly expressed in nonadipocyte cell types, including mesothelial cells and pericytes (102). The latter is implicated in the regulation of tissue vascularization (104, 105), and as such GIP promotes adipocyte lipid deposition, at least in part, by increasing adipose tissue blood flow (106, 107). However, these nonadipocyte effects of GIP on adipose tissue function cannot explain the various reports that testify to GIP’s direct cell-autonomous effect on adipocyte lipid metabolism, as has been shown in 3T3L1 adipocytes (82–84, 87), as well as in adipocytes isolated from humans (83, 108) and rats (85–87, 109). Nonetheless, in contrast to mice with adipocyte-specific deletion of Gipr (101, 102), mice with targeted deletion of Gipr in the brain (99), and specifically in GABAergic neurons (110), show decreased body weight and fat mass when fed with an HFD. However, even these neuron-specific Gipr-deficient mice recapitulate only a mere fraction of the obesity-protecting phenotype seen in mice with global Gipr deficiency (78–81). However, the source of additional protection against DIO remains unknown, as tissue-specific knockdown of Gipr in brown adipose tissue, pancreatic β-cells, and adipose-localized immune myeloid cells all fail to recapitulate the normalization of body weight seen in the global knockout (KO) mice (FIGURE 2) (111–114). In summary, while a series of studies show GIP promotes lipid storage under conditions of hyperinsulinemia, GIPR agonism accelerates cAMP-mediated lipolysis under hypo- or normo-insulinemic conditions. Additionally, a series of genetic studies show that protection of diet-induced obesity in Gipr-deficient mice is partially mediated via central mechanisms but is unrelated to GIPR presence in adipose tissue.

FIGURE 2.

Schematic on the metabolic phenotype of conditional and global glucose-dependent insulinotropic polypeptide receptor (Gipr) knockout (KO) mice Simplified chart describing the role of tissue-specific GIPR knockouts on the degree of protection against body weight gain under chow and high-fat diet or its requirement for mediating GIPR-based pharmacological-induced weight loss. CNS, central nervous system; DIO, diet-induced obese; GLP-1R, glucagon-like peptide-1 receptor; RT/TN, room temperature/thermoneutral; VGAT, vesicular GABA amino acid transporter; WAT, white adipose tissue.

Apart from GIP effects on the adipose tissue, which seem to vary depending on the presence of insulin, long-acting GIPR agonists act in the brain to decrease body weight via inhibition of food intake (99). Chemogenetic activation of GIPR neurons in either the hypothalamus (115, 116) or the hindbrain (116) decreases food intake, and central administration of acyl-GIP into the hypothalamic third ventricle dose-dependently decreases body weight and food intake in DIO wild-type mice but not in mice with Nestin Cre-mediated neuronal loss of Gipr (99). In the hypothalamus and the hindbrain, Gipr is colocalized with vesicular GABA amino acid transporter (Vgat), a marker indicative of inhibitory GABAergic neurons, and deletion of Gipr specifically in Vgat-expressing neurons is sufficient to fully block the ability of acyl-GIP to decrease body weight and food intake in DIO mice (110).

It warrants clarification as to why both activation and inhibition of the GIP receptor decrease body weight and fat mass in experimental animals. A popular hypothesis is that GIPR agonism desensitizes the GIP receptor and hence leads to functional GIPR antagonism (117). Chronic GIPR agonism has indeed been shown to result in reduced GIPR sensitivity in DIO mice, and in isolated adipocytes (103), but similar results have been shown for GLP-1 in rat insulinoma INS-1 cells (118) and for both incretins in hamster β-cell HIT-T15 cells (119). In addition, even chronic treatment of DIO mice with acyl-GIP does not decrease the expression of Gipr in the hypothalamus or the adipose tissue (99). While there is currently no evidence indicating that GIPR agonism decreases body weight through functional GIPR antagonism, it seems possible that GIPR agonists and antagonists affect body weight through different mechanisms and target tissues, with long-acting GIPR agonists acting on brain satiety centers to affect food intake, while GIPR antagonists may compete with endogenous GIP in the periphery to inhibit the lipogenic action of endogenous GIP in the adipose tissue (120). Another possibility is that while GIPR agonism acts centrally to decrease food intake via activation of GABAergic neurons (110), GIPR antagonism may inhibit GABAergic input into the anorectic GLP-1R positive glutamatergic neurons, thereby reducing food intake via unrestrained glutamatergic effect analogous to GLP-1R agonism. Consistent with this is the observation that GIPR antagonists primarily decrease food intake when given together with GLP-1 (91, 92), while GIPR agonists also decrease food intake and body weight in mice deficient for GLP-1R (99, 121). While these hypotheses warrant experimental verification, GIPR agonism clearly depends on GIPR signaling in the CNS to decrease body weight and food intake (99, 110), and while the greatest body weight-lowering effects of GIPR antagonists are observed using antibody-based GIPR antagonists (91), antibodies have generally a very limited ability to reach the brain (122). Nonetheless, when administered directly into the brain, antibody-based GIPR antagonists also decrease body weight in DIO mice, an effect paralleled by improved leptin sensitivity (123). The obesity-protecting phenotype seen in mice with global GIPR deficiency however cannot be solely explained by improved leptin sensitivity, as deletion of Gipr in leptin-deficient ob/ob mice reduces body weight relative to ob/ob controls (81).

Another open question is why pharmacological GIPR agonism has not been as clearly delineated relative to GLP-1R agonism. An important realization is that the GIP system is evolutionary less conserved than the GLP-1 system, and human GIP is only a weak and partial agonist at the mouse GIP receptor (124). The importance of this has recently been demonstrated for the GIPR:GLP-1R coagonist tirzepatide, which is based on the human GIP sequence and stimulates islet insulin secretion in mice predominantly via the GLP-1 receptor but in human islets via the GIP receptor (125). In light of this notion, and given that neither human GIP nor tirzepatide is capable of fully activating the mouse GIP receptor (124), it is hence not overtly surprising that tirzepatide decreases body weight in mice exclusively via the GLP-1 receptor (126). Consequentially, as tirzepatide shows only weak agonism at the mouse GIP receptor, these data do not allow for the conclusion that tirzepatide also decreases body weight in humans exclusively via the GLP-1 receptor. Consistent with this, the GIPR:GLP-1R coagonist MAR709 is far more potent at the mouse GIP receptor relative to tirzepatide, and treatment of DIO mice with MAR709 leads to greater weight loss and further inhibition of food intake relative to mice treated with a pharmacokinetically matched GLP-1R agonist (99, 110). In line with intact GIPR signaling in MAR709, its superiority to a matched GLP-1R agonist vanishes in mice with neuronal loss of Gipr (99) or when Gipr is specifically deleted in Vgat-expressing GABAergic neurons (110). In summary, GIPR agonists that are based on the human GIP sequence (including tirzepatide) are insufficiently potent at the mouse GIP receptor and are hence incapable of studying the mode-of-action of GIP-based drugs in rodents. Nonetheless, if sufficiently active at the mouse receptor, GIPR agonism decreases body weight via inhibition of food intake (99, 110) and can be a vital constituent in unimolecular formulations with GLP-1R agonism (99, 110, 127, 128). Consistent with this, central coadministration of low-dose GIP and GLP-1 synergistically reduces food intake (127), and this effect correlates with synergistic increases in neural activation of pro-opiomelanocortin in the hypothalamic arcuate nucleus (129).

Fatty acid acylation of GLP-1 and GIP has further proven to be an appreciable biochemical tool to enhance CNS-driven satiety effects of both GLP-1 and GIP. Whether such fatty acid acylation increases metabolic outcome simply by the extended half-life, or maybe also due to increased brain penetrance, and exposure of deeper brain structures implicated in energy metabolism control, remains to be determined (130, 131). The observation that neither liraglutide (132) nor semaglutide (133) can cross the blood-brain barrier, however, suggests that these molecules rather act on the circumventricular organs to inhibit food intake, and recent evidence indicates that this is also the case for GIP (110, 134). Nonetheless, the body weight-lowering effects of acyl-GIP are preserved in GLP-1R KO mice (99, 121) but vanish in mice with deletion of Gipr in either CNS neurons (99) or specifically in GABAergic neurons (110), hence clearly demonstrating the involvement of central GIPR signaling in mediating these effects. In summary, there is ample evidence indicating that GIP as a single or combination therapy decreases food intake via central mechanisms, that these effects are GLP-1R independent, and that these effects are likely enhanced via acylation.

The pharmacological value of GIP is not restricted to its insulinotropic and body weight-lowering effects. GLP-1R monoagonism is frequently associated with gastrointestinal (GI) adverse effects, predominantly nausea and vomiting (135, 136). Adverse GI effects are predominantly observed in the early treatment period and can be minimized by careful gradual dose escalation (137, 138). The increased rate of GI adverse events, despite plateauing metabolic benefits, caps the maximal tolerable dose of semaglutide at 2.4 mg (73). This registered dose of semaglutide is chosen based on projected pharmacokinetic modeling of a daily 0.4-mg dose, which exhibited greater than double the GI events leading to trial discontinuation than the next lower dosage of 0.3 mg/day (139). It appears that GIPR agonism in adjunct to GLP-1R agonism mitigates the GI adverse event profile of GLP-1R pharmacology to ultimately increase both patient compliance and maximal practical dosage. Coadministration of both GLP-1 and GIP eliminates the occurrence of GLP-1-induced emetic episodes in musk shrews (140), demonstrating GIP’s capacity for complimentary antinausea effects. In line with the antiemetic effect of GIP, the maximal dosing of QW tirzepatide is as high as 15 mg, with an adverse event and discontinuation profile comparable to QW semaglutide at 2.4 mg (76, 141). However, given that tirzepatide is a 5:1 unbalanced GIPR:GLP-1R coagonist that favors the GIP receptor, it warrants clarification how much of the greater tolerability originates from the demonstrated lower potency of tirzepatide at the GLP-1 receptor (142–144) and/or from the antiemetic effect of GIPR agonism (140). Also, the mechanism behind the antiemetic effects of GIPR agonism warrants further clarification. Recent evidence indicates that the anorectic and aversive effects of GLP-1R agonism are mediated via hindbrain CCK neurons, and although GIP does not directly target these neurons, GIPR agonism reduces conditioned taste avoidance through diminished GLP-1-induced activation of these neurons (134). The area postrema/nucleus of the solitary tract is hence suggested to mediate satiety and satiety-related emesis, and single-cell transcriptomic profiling of the dorsal vagal complex has revealed that a significant portion of GABAergic neurons in this region express Gipr (145, 146). It may be coincidental that also the body weight-lowering effect of GIPR agonism is mediated by Gipr expressing GABAergic neurons (110), and these neurons may hence not only contribute to the body weight-lowering effect of the GIPR:GLP-1R coagonist tirzepatide but also explain its greater tolerability, and hence the allowance of higher doses, relative to semaglutide. In summary, central GIPR agonism not only decreases body weight via centrally mediated inhibition of food intake, it further decreases the emetic effect of GLP-1R agonism, an appreciable observation that may accelerate tolerability and higher dosing of GLP-1R agonism.

Performance of Tirzepatide versus Semaglutide on Body Weight in Humans

Ironically, while no GIPR antagonist has yet been approved for the treatment of obesity, GIPR agonism has proven remarkably efficacious in accelerating the metabolic outcome of GLP-1R agonism, either as cotherapy (127) or in unimolecular formulation (127, 128). The first unimolecular GIPR:GLP-1R coagonist was developed in 2013 by DiMarchi et al. (127). The 1st generation of these molecules (NNC0090-2746/MAR709) outperformed exendin-4 and liraglutide in body weight reduction when given at equimolar doses in obese rodents (127). The GIPR:GLP-1R coagonist tirzepatide has been derived from the GIP backbone and favors GIPR over GLP-1R by approximately fivefold in binding affinity, and ∼13-fold in a cAMP reporter assay (128). In vitro models have demonstrated tirzepatide to activate both the human GLP-1R and GIPR differentially within subsequent steps of the signal cascade to minimally recruit β-arrestin to the GLP-1R, to possess differential GLP-1R internalization and endosomal trafficking dynamics, and to ultimately potently reduce body weight in DIO mouse models (FIGURE 3) (128, 142, 143). The beneficial metabolic effects of tirzepatide have been verified in a series of human studies (147–150) and have resulted in the approval of tirzepatide for the treatment of T2D in 2022 and for the management of obesity in 2023. Very recently, the metabolic efficacies of tirzepatide and semaglutide were compared in real-world settings in patients with obesity (151). The study included data from over 41,000 individuals, from which 32,030 received semaglutide and 9,193 tirzepatide. After 1 year of treatment, weight reduction ≥5% was achieved in 81.8% and 64.6% of patients receiving either tirzepatide or semaglutide, while weight loss ≥10% was achieved in 62.1% and 38.0%, and ≥15% in 42.3% and 19.3% of patients (151). The superiority of tirzepatide over semaglutide was reflected by additional weight loss of −2.3% after 3 months of treatment, −4.3% after 6 months, and −7.2% after 12 months, with no changes between tirzepatide and semaglutide on the occurrence of GI adverse effects at any time point (151). Collectively, these data demonstrate that tirzepatide outperforms semaglutide in body weight endpoints without compromising its tolerability. In line with these data, the SURMOUNT-1 trial demonstrates −11.9%, −16.4%, and −17.8% placebo-corrected body weight loss at the used doses of 5, 10, and 15 mg of tirzepatide in nondiabetic obese patients over the course of 72 weeks of treatment (152). The lowest dose (5 mg) hence achieves roughly the same degree of body weight reduction as semaglutide 2.4 mg (−12.4%) (152). Interestingly, when considering the 10-mg and 15-mg groups pooled, the averaged placebo-corrected body weight loss for tirzepatide is −17.1% with 6.6% patient discontinuation due to adverse events out of a total 1,266 total patients. This is compared to semaglutide, in which out of 1,306 total patients, a placebo-corrected −12.4% reduction in body weight was achieved with 7.0% patient discontinuation (73, 152). Together, this suggests an approximate 4.5-5% enhancement in body weight loss with 10/15 mg tirzepatide relative to 2.4 mg semaglutide, with approximately the same degree of adverse events that lead to treatment discontinuation. These findings suggest that the chemical structure of tirzepatide, whether it be the GIP-dependent engagement of complementary satiety signals, GIP’s potential for antiemesis-mediated escalation of GLP-1 component dosage, or intrinsic GLP-1R biased agonism, or all of them, leads to substantial improvements in body weight-lowering efficacy in nondiabetic obese individuals. Similarly, the SELECT phase III clinical trial with semaglutide demonstrates an approximate −10% reduction in body weight maintained over a maximum of 221 weeks, and the SUPRASS 4 phase III trial reports tirzepatide’s maintenance of body weight loss at approximately −26% through 88 weeks (75, 153). Importantly, in obese patients with concurrent T2D, 72 weeks of 10-mg and 15-mg tirzepatide treatment resulted in −9.6% and −11.6% reduction in body weight, which is almost double that of the −6.2% by semaglutide in 68 weeks (76, 141). Similarly, in the SURPASS trials which evaluated the antidiabetic capacity of tirzepatide in patients with T2D, secondary end point changes from baseline body weight by 15 mg tirzepatide were as follows: −11.1% (SURPASS 1: 40 weeks), −13.1% (SURPASS 2: 40 weeks), −13.7% (SURPASS 3: 52 weeks), −12.8% (SURPASS 4: 52 weeks), and −9.2% (SURPASS 5: 40 weeks). Together, the SURPASS 1–5 and SURMOUNT 2 clinical trials suggest a double-digit average of induced body weight loss in T2D individuals, despite slight differences in trial durations (147–150, 154). As mentioned, it is unknown why tirzepatide performs better on weight loss within the context of T2D; however, despite the aspects of GIP’s antiemetic allowance of GLP-1 component dose escalation, and the potential for intrinsic GLP-1R biased agonism, it is considerable that the plateau in maximum HbA1c reduction by tirzepatide occurs within 24 weeks, while reductions in body weight do not plateau until week 60 (141). Hence, according to the potential for GIPR resensitization during recovery toward normoglycemic conditions, these additional weight loss benefits as time continues may be attributed to the synergistic action between GLP-1R and GIPR coagonism on reductions in food intake and body weight (42, 127–129). In summary, tirzepatide outperforms semaglutide to achieve greater weight loss, but it warrants clarification in human studies if and to what extend GIPR agonism contributes to better weight loss efficacy and enhanced tolerability.

FIGURE 3.

Tirzepatide as a “super” glucagon-like peptide-1 receptor (GLP-1R) agonist? A–D: simplified schematic describing the alternative intracellular signaling and trafficking dynamics of the GLP-1R at the single receptor level or the global cellular level elicited by GLP-1R monoagonists such as GLP-1-(7-36NH₂) and semaglutide (A and B) or tirzepatide and exendin4-Phe1 (EX4-Phe1) (C and D). GLP-1R signaling and trafficking dynamics following ligand binding and receptor activation by mono-agonists such as GLP-1-(7-36)amide and semaglutide. A1: GLP-1R activation on the level of direct G-protein interaction is maximal, which includes maximal G protein-coupled receptor (GPCR)-mediated GDP to GTP exchange on the Gα_s subunit, and maximal recruitment of GDP-bound Gα_s to the GLP-1R for continued signaling. A2a: the “activated” GTP-bound Gα_s is subsequently recruited to adenylate cyclase where it stimulates the amplified production of cAMP. A2b: Simultaneously, β-arrestin is recruited to the GLP-1R to facilitate a braking mechanism on continued GDP-bound Gα_s recruitment to the ligand-bound GPCR to prevent further signaling. A3: internalization carries the GLP-1R away from the plasma membrane into the intracellular space, where it continues to transiently signal but also is redirected into desensitizing endolysosomal pathways. B: global schematic view at the cellular level of total GLP-1R dynamics resulting from the sum of unique single receptor signaling and trafficking dynamics elicited by GLP-1R agonists such as GLP-1-(7-36)amide and semaglutide. C: GLP-1R signaling and trafficking dynamics following ligand binding and receptor activation by “biased” GLP-1R agonists such as tirzepatide and EX4-Phe1. C1: GLP-1R activation on the level of direct G-protein interaction is minimal. C2a-C2c: maximal cAMP signaling efficacy (C2a) is achieved through minimal β-arrestin recruitment brake on signaling (C2b) and greater GLP-1R retention at the plasma membrane leading to less endolysosomal colocalization and higher receptor exposure to further extracellular ligands (C2c). D: global schematic view at the cellular level of total GLP-1R dynamics resulting from the sum of unique “biased” single receptor signaling and trafficking dynamics elicited by tirzepatide and exendin4-Phe1.

GLP-1R Agonism/GIPR Antagonism

Notably, genetic or pharmacological modulation of the GIP system, whether through loss/gain-of-function or agonism/antagonism, results in different degrees of weight loss and protection from obesity (100, 155). In particular, antimurine GIPR antibodies (muGIPR-Ab) and peptide-based GIPR antagonists, while not strong stimulators of body weight loss in DIO mice, hint at some degree of induced weight loss or protection against the onset of obesity (91, 92). Nonetheless, it is clear that antibody- or peptide-based GIPR antagonist coadministration with GLP-1R monoagonists synergistically reduces body weight (91, 92). In DIO mice, liraglutide coadministration with a muGIPR-Ab (BWΔ: −27%) was superior to liraglutide alone at 80 nmol/kg (−15%) (91) and semaglutide coadministered with a peptide-based GIPR antagonist (−28%) outperformed semaglutide alone at 3 nmol/kg (−12%) (92). As an antihuman GIPR antibody (hGIPR-Ab) was found to reduce body weight both as a single therapy and as a combinatorial therapy with dulaglutide in nonhuman primates (91), a DPP4-protected GLP-1 was conjugated with a GIPR antibody (GIPR-Ab/GLP-1) and developed into the molecule AMG 133, exhibiting a half-life between 5 and 9 days depending on the species and route of administration (156). In DIO mice, mGIPR-Ab/GLP-1 reduced body weight by approximately −30%, approximately double that of GLP-1 conjugated to a nonspecific antibody (156). Similar findings were obtained in obese nonhuman primates, in which 2.5 mg/kg of hGIPR-Ab/GLP-1 induced approximately −10% body weight loss after 43 days of treatment (156). Interestingly, at the GLP-1R, both the conjugated muGIPR-Ab/GLP-1 and hGIPR-Ab/GLP-1 are prominently less potent than unconjugated GLP-1 for cAMP production, a profile similar to that of tirzepatide (128, 156). However, when both GLP-1R and GIPR are coexpressed, the cAMP potency of the antibody-GLP-1 conjugate is surprisingly greater than that of the unconjugated GLP-1, in which the improvement is suggested to occur via enhanced ligand proximity via GLP-1R/GIPR colocalization and altered endosomal signaling (156). Indeed, it is suggested that GLP-1R and GIPR interactions at the plasma membrane upon GLP-1R agonism allow for a unique signaling profile, which is demonstrated as a reduction in both Gα_q and β-arrestin recruitment to the GLP-1R, while Gα_s activity remains fully intact (157). Although clinical data has not yet been published, a press release from Amgen has indicated AMG 133 to achieve up to −14.5% body weight reduction after 12 weeks in a phase I clinical trial and has additionally indicated the start of a phase II clinical trial in early 2023 (158). However, questions remain as to how antagonism of the GIPR will influence the adverse event profile of GLP-1R agonism, particularly considering the potential for GIPR agonism’s involvement in antiemetic effect. In summary, GIPR agonism and antagonism both have therapeutic potential in adjunct to GLP-1R agonism, but it warrants clarification whether GIPR antagonism promotes its additional weight loss efficacy via GLP-1R-dependent or -independent mechanisms.

Is Tirzepatide a “Super” GLP-1?

A consideration of the enhanced efficacy of tirzepatide over semaglutide is that tirzepatide, for all intents and purposes, may be an incidental super GLP-1R agonist due to a signaling bias incurred by amino acid modifications implemented into the peptide backbone originally meant to confer dual agonism. Tirzepatide does activate the human GIPR, but this attribute is suggested to be inconsequential to the in vivo metabolic benefits relative to its biased attributes at the GLP-1R (142–144). Interestingly, at the GLP-1R, tirzepatide recruits minimal β-arrestin despite a concurrent cAMP E_max equivalent to that of semaglutide and GLP-1 (142, 143). Additionally, minimal GLP-1R internalization from the plasma membrane into the intracellular space, and the subsequent reduction in lysosomal targeting, suggest tirzepatide’s influence on GLP-1R trafficking to be protective against canonical G protein-coupled receptor (GPCR) desensitization (142, 143). Together these attributes indicate a capacity for tirzepatide to maximally stimulate a cAMP response, recruit minimal signaling inhibition, and simultaneously retain more GLP-1R at the plasma membrane (FIGURE 3). In relation, a Phe1-substituted exendin-4 (EX4-Phe1) peptide demonstrates super efficacy for both antiglycemic and antiobesogenic measures (159–161). Similar to tirzepatide at the GLP-1R, EX4-Phe1 exhibits maximal cAMP efficacy and minimal β-arrestin recruitment, receptor internalization, and receptor desensitization (159, 160). EX4-Phe1-induced insulin secretion within the rat β-cell model INS-1 832/3 outperforms exendin-4, liraglutide, dulaglutide, and semaglutide, suggesting a superior capacity to elicit an insulinotropic response (159). In vivo administration of the unacylated EX4-Phe1 evidences superior glucoregulatory control relative to exendin-4 during glucose tolerance tests (GTT) performed 4 hours and 8 hours after peptide administration (159). Similarly, in a GTT performed 72 hours posttreatment, an acylated version of Ex4-Phe1 outperforms a PK-matched alternative analog of exendin-4 characterized as favoring β-arrestin recruitment and GLP-1R internalization, suggesting enhanced long-term glucoregulatory efficacy of EX4-Phe1 to be likely due to sustained receptor sensitization and a lack of receptor internalization (160). Importantly, dose escalation of 10 to 20 nmol/kg EX4-Phe1 over the course of 15 days demonstrates superior body weight-lowering effects relative to the aforementioned exendin-4 analog that favors β-arrestin recruitment and GLP-1R internalization, henceforth suggesting minimal β-arrestin recruitment and GLP-1R internalization as a successful strategy for enhancing GLP-1R-centric body weight-lowering effects (160). This strategy has further been refined as the GLP-1R monoagonist SRB107 and has hinted at promising pre-clinical results (162). Importantly, with respect to tirzepatide’s enhanced efficacy relative to semaglutide, EX4-Phe1 cotreatment with a DPP4-protected GIP analog in DIO mice demonstrates superior enhancements in body weight reduction relative to EX4-Phe1 alone, suggesting GIPR coagonism to remain a potentiator of metabolic improvements even within the context of super GLP-1R agonism (163). In summary, it warrants clarification if and to what extent the favorable GLP-1R trafficking and β-arrestin recruitment profile of tirzepatide contributes to its observed greater weight loss efficacy, and how GIPR co-agonism likely contributes to a vital biochemical liaison with biased GLP-1R agonism.

Chronic GIPR Agonism as a Driver of Functional Antagonism?

A hypothesis to explain the enhanced efficacy attributed to the GIPR:GLP-1R coagonist tirzepatide is the suggestion that the onset of chronic GIPR agonism desensitizes the GIPR as to no longer elicit intracellular responsiveness. Through this functional antagonistic effect, tirzepatide’s long-term mechanism is suggested to be analogous to coadministration of a GLP-1R monoagonist and a GIPR antagonist. GPCR desensitization over a longer course of action (ie. repeated drug exposure) is generally mediated by receptor internalization, ubiquitination, lysosomal targeting, and ultimately receptor degradation (164). Interestingly, there has been disagreement as to the extent of tirzepatide-induced GIPR internalization (142, 143). Nonetheless, native GIP(1–42) demonstrates an ability to induce GIPR receptor internalization, while related variants including a Gly2-subtituted GIP and an AIB2-subtituted “acyl GIP,” do not (142, 165). Relative to the GLP-1R however, the ligand-bound GIPR internalizes less, internalizes slower, is completely absent from Rab5+ early endosomes, exhibits reduced endosomal signaling and enhanced receptor recycling, and is minimally targeted to lysosomes for degradation (166). In short, it seems the GIPR is capable of undergoing some ligand-induced internalization; however, this effect does not result in substantial lysosomal targeting, and the degree of internalization seems to be especially sensitive to ligand modifications. These reflexive properties of GIPR trafficking bring into question canonical GPCR degradation as a mediating pathway to the occurrence of GIPR functional antagonism. Nonetheless, it is important to not discount differential GIPR dynamics inherent within alternative tissue types and to consider the role GIPR variants may have on these trafficking processes (167, 168).

Alternatively, GIPR desensitization has been suggested to occur via persistent β-cell membrane depolarization, an attribute representative of β-cells in the diabetic state (169). In this model, persistent membrane depolarization desensitizes the Gα_s-cAMP pathway, while alternatively preserving the efficacy of the Gα_q-IP₃ pathway (169). Indeed, the GLP-1R is both Gα_s and Gα_q coupled while the GIPR is only Gα_s-coupled, giving potential reason to GIPR’s observed inefficacy, and a potential Gαq-mediated explanation for GLP-1R’s continuing glucoregulatory efficacy, in the human diabetic state. Regarding an alternative pathway toward agonism-induced GIPR functional antagonism, it is speculatively possible that chronic concurrent agonism of both the GLP-1R and GIPR Gα_s subunits in β-cells desensitizes Gα_s action and thus further GIPR agonism. Yet, this analogous situation of agonism-induced functional antagonism due to Gα_s desensitization has yet to be fully explored.

Conclusions and Future Directions

With the discovery, clinical implementation, and/or commercialization of semaglutide, tirzepatide, and AMG 133, the ground has been broken in implementing the next era of antidiabetic and antiobesogenic pharmacology. Despite major clinical success, further refinement in both understanding and approach will be needed to open pathways for future investigation, not only toward higher understanding of the associated biological dynamics, but also in the hope of optimizing toward more advanced next-generation therapeutics. Advancements in GLP-1R biology have yielded fruitful therapeutic value that has only now escalated toward revolutionizing the treatment of diabetes, obesity, and its comorbidities. Along the way, proteolytic protection and half-life extension have been key to capturing the antiobesogenic efficacy of the GLP-1 peptide. In the way these modifications are critical to the success of semaglutide, we are similarly now acknowledging the profound potential of polypharmacology in maximizing the therapeutic utility of GLP-1-centric approaches. While semaglutide has set the stage for clinical GLP-1R monoagonist approaches against diabetes and obesity, tirzepatide and AMG 133 represent novel options that pertain to GIPR-mediated complimentary enhancements of GLP-1R efficacy. Dual or triple modes of action in therapy, while allowing maximal benefit, exponentially increase the difficulty in understanding the mechanisms underpinning their enhanced efficacy. For tirzepatide, while the mechanistic resolution of GIPR biology is coming of age, the pharmacological utility of GIPR is still under debate and leaves to question, what mediates tirzepatide’s superior effectiveness: synergies of GLP-1R/GIPR coagonism, synergies of GLP-1R/GIPR functional antagonism, or biased GLP-1R agonism? Similarly to AMG 133, how will antagonism of a canonical Gα_s-coupled system involved in multiple systemic processes that include satiety, emesis, and adipocyte lipid metabolism counterintuitively amplify a therapeutic GLP-1 receptor response? There are many advancements and questions listed and not listed here that will gratuitously fill research advancement for the coming years.