How GLP-1 Receptor Signals May Cause Mouth Problems Linked to Semaglutide

GLP-1 is a peptide hormone derived from the breakdown of proglucagon, a larger precursor molecule. It is produced in three main locations: enteroendocrine L-cells of the intestinal mucosa, pancreatic α-cells, and certain neurons within the nucleus of the solitary tract in the brainstem. After food intake, intestinal L-cells release GLP-1 into circulation in response to nutrient and neuroendocrine signals. The hormone is produced as different molecular variants, including GLP-1(8-36) amide and GLP-1(8-37). Regardless of the form, its primary role is as an incretin hormone—a regulator of postprandial (after meal) blood glucose levels. GLP-1 stimulates insulin secretion in a glucose-dependent manner, suppresses glucagon release, slows gastric emptying, and promotes satiety by acting on appetite centers in the brain. However, its activity in the bloodstream is short-lived. Natural GLP-1 is rapidly degraded within one to two minutes by the enzyme dipeptidyl peptidase-4 (DPP-4), which sharply limits its therapeutic usefulness.

The GLP-1 receptor is a GPCR that responds specifically to GLP-1. While it is highly expressed in pancreatic β-cells, its distribution extends far beyond the pancreas. GLP-1R can be found in the heart, kidneys, lungs, liver, and central nervous system and gastrointestinal tract as well as in salivary glands [7]. Based on the Human Protein Atlas (HPA) page for GLP1R in salivary gland, RNA expression derives from three transcriptomic datasets: HPA bulk RNA-seq, GTEx RNA-seq, and FANTOM5 CAGE. The HPA dataset includes three normal salivary-gland tissue samples (ages 21–68; nTPM 1.3–3.2), while the GTEx minor salivary-gland panel includes 181 donors (mean 0.2 nTPM, range 0.0–1.3). Both HPA and GTEx report expression in normalized transcripts per million (nTPM), whereas FANTOM5 provides scaled CAGE tags per million (sample counts not specified on the GLP1R gene page). HPA assigns GLP1R to RNA expression cluster 8 with a cluster confidence of 1.0; however, it is worth noting that normal-tissue protein annotation is pending, meaning RNA data do not confirm protein presence or function. All dataset values and sample-level information are available via the GLP1R salivary-gland RNA page: https://www.proteinatlas.org/ENSG00000112164-GLP1R/tissue/Salivary+gland#rnaseq↗ (Accessed on 15 November 2025). Thus, GLP1R RNA is expressed at low to moderate levels in salivary glands with cellular distribution mainly in glandular and ductal cells, with variable expression across donors. Expression is biologically relevant but not abundant—consistent with the idea that GLP1R plays a modulatory rather than dominant role in salivary gland function (Table 1).

Because natural GLP-1 is unstable, GLP-1RAs result from intricate structural modifications to GLP-1, enabling them to not only replicate the pharmacological functions of GLP-1 but also impede its hydrolysis by DPP-4, thereby extending the drug's half-life [8,9]. Structural modifications allow these drugs to remain active longer in circulation, producing a stronger and more sustained biological effect [10]. Currently available GLP-1RAs vary in their molecular structure and dosing schedule. Some require daily injections, while others are long-acting and administered only once weekly. Newer formulations aim to improve patient convenience further, including orally available tablets. These innovations improve adherence and broaden therapeutic use. Thus, different farmacocinetic profiles associate with different adverse effects [1,11]. The report frequency of gastrointestinal disorders related to semaglutide was significantly higher than that of non-semaglutide in the overall database with the ROR of 4.21. The 45 semaglutide-related gastrointestinal AEs also showed statistically significant signal strengths as compared to non-semaglutide-associated gastrointestinal AEs, with values of signals ranging from a ROR025 of 1.01 (hypoaesthesia oral) to 42.03 (eructation). The association between semaglutide and gastrointestinal disorders remained when stratified by age, body weight, sex and reporter type. Among the 45 gastrointestinal AEs, 17 new and unexpected AEs, including dry mouth, hypophagia, bowel movement irregularity, etc., were detected in our pharmacovigilance study, which were not reported in the drug label (Table 2). The exact effects of semaglutide on these AEs and the mechanisms of this potential association were not completely explored, requiring further clinical investigation [1].

Based on analysis of The Food and Drug Administration's (FDA) Adverse Effect Reporting System (FAERS), adverse effects of GLP-1 RA-related medications including semaglutide, exenatide, liraglutide, dulaglutide, and tirzepatide were analyzed represent 47 cumulative years of observation across all drugs included in the analysis [11]. Semaglutide was also the only drug to display significant ROR for throat tightness (1.69) and throat irritation. Dysgeusia had significant ROR and PRR in exenatide (ROR 4.02:PRR 4.00), liraglutide (ROR 3.88; PRR 3.86), and semaglutide (ROR 3.67; PRR 3.66). Dry mouth had significant ROR signals in exenatide (1.26), liraglutide (1.80), and semaglutide (3.21) with only semaglutide having a significant PRR (3.19). However, spontaneous reporting systems are subject to substantial limitations, including under-reporting, confounding by indication (e.g., diabetes-related xerostomia), channeling bias toward older or more comorbid patients, and polypharmacy involving xerostomia-inducing drugs such as antidepressants and anticholinergics. Therefore, the observed ROR/PRR elevations indicate signal strength rather than incidence or causal relation. Recent case series has shown 3 cases for patients who were using semaglutide and developed severe secondary hyposalivation [12]. One interesting observation in this report is the onset of hyposalivation in which it was noted 4 weeks after starting semaglutide. The known mechanism of action and body water loss properties for semaglutide may explain the risk of hyposalivation in patients on semaglutide. However, water intake was within the range of normal and considering the severity and duration of reported symptoms, dehydration was less likely to be the underlying cause. In addition, no history of radiation therapy to the head and neck area, was reported by either patient. Another differential diagnosis considered was systemic disease-related hyposalivation. Although dehydration, prior head-and-neck irradiation, and systemic disease were considered and judged unlikely, the sample size is too small to infer generalizable risk or mechanism.

Glucagon-like peptide-1 (GLP-1) exerts its physiological effects by binding to the GLP-1 receptor (GLP-1R), a class B1 G-protein-coupled receptor (GPCR). [13,14] abundant in pancreas, but also present in salivary glands. In pancreas, upon activation, GLP-1R couples to heterotrimeric G-proteins, primarily stimulating adenylyl cyclase to increase intracellular cyclic AMP (cAMP). Elevated cAMP activates protein kinase A (PKA), which enhances glucose-dependent insulin gene expression and secretion while suppressing glucagon release [15,16]. In parallel, cAMP activates the exchange protein directly activated by cAMP (EPAC), which regulates vesicle priming and exocytosis, thereby fine-tuning insulin release [17,18,19]. At the transcriptional level, cAMP activates the cAMP response element-binding protein (CREB), which regulates insulin gene expression [20,21,22]. Beyond cAMP-mediated mechanisms, GLP-1 signaling also engages the phosphoinositide 3-kinase (PI3K)/protein kinase B (Akt) pathway, which is critical for preserving the survival, proliferation, and function of pancreatic β-cells [23,24,25,26]. Following GLP-1R activation, the βγ subunits of GPCRs directly activate class I PI3Ks, especially the α and β isoforms [27]. These enzymes consist of regulatory and catalytic subunits, which may interact directly with the receptor or indirectly via adaptor proteins such as insulin receptor substrate (IRS) [28,29,30]. PI3K activation leads to the phosphorylation of phosphatidylinositol-4,5-bisphosphate (PIP2), generating phosphatidylinositol-3,4,5-trisphosphate (PIP3) [31]. PIP3 serves as a docking site for Akt (protein kinase B), enabling its recruitment and phosphorylation [32]. The PI3K/Akt pathway remains essential in this context, as it ensures β-cell survival, proliferation, and effective glucose regulation [33,34]. Their activity is fine-tuned by GPCR kinases (GRKs), which phosphorylate the receptor and enable β-arrestin binding. β-arrestins were first recognized for desensitizing GPCRs by uncoupling them from G-proteins and promoting receptor internalization via clathrin-coated pits, but are now known to act as signaling scaffolds themselves, initiating distinct pathways like ERK, JNK, or NF-κB activation. This dual role underpins the concept of biased agonism, where different ligands selectively favor either G-protein-mediated or β-arrestin-mediated signaling, allowing for fine control over therapeutic outcomes and side-effect profiles. Arrestins act as scaffolds to promote recruitment and activation of mitogen-activated protein kinases (MAPKs) that are linked to protection against apoptosis and pancreatic beta-cell survival and growth [35]. GLP-1R-mediated recruitment of arrestins also leads to transactivation of epidermal growth factor receptor, activation of phosphatidylinositol 3-kinase and downstream nuclear translocation of protein kinase Cζ to promote beta cell proliferation, insulin gene expression, and insulin synthesis and secretion [27,36].

Salivary glands, however, rely on a similar set of GPCR-driven cascades to regulate secretion [37]:Parasympathetic M3 muscarinic receptors (Gq-coupled) activate phospholipase C → IP3-mediated Ca²⁺ release → fluid (aqueous) secretion.Sympathetic β-adrenergic receptors (Gs-coupled) activate adenylyl cyclase → increase cAMP → PKA-dependent pathways that regulate protein-rich exocytotic secretion (e.g., amylase, mucins).

Thus, in salivary acinar cells, cAMP/PKA/EPAC signaling plays a role analogous to pancreatic insulin release, but here it fine-tunes exocytosis of salivary proteins and enzymes rather than insulin granules. Thus, since transcriptomic atlases (HPA, GTEx, FANTOM5) suggest that GLP-1R is expressed in salivary glands, when activated, GLP-1R would engage the same cAMP–PKA–EPAC and PI3K–Akt modules, potentially influencing [38,39,40]:Vesicle priming and exocytosis of salivary proteins (via cAMP signalling).Cellular protection and survival of acinar and ductal cells (via EGF).CREB-dependent transcription, possibly altering the synthesis of secretory proteins or ion transporters that regulate saliva composition.

Biased agonism describes the ability of different ligands to stabilize distinct receptor conformations, preferentially activating some pathways (e.g., G protein–mediated cAMP signaling) while avoiding others (e.g., β-arrestin recruitment), which allows for fine-tuning pharmacological outcomes [41,42]. Unlike most GPCRs that internalize through a canonical GRK–β-arrestin–clathrin mechanism [43], GLP-1R displays atypical trafficking properties. GLP-1R internalizes via an arrestin-independent, GRK-dependent, clathrin- or caveolae-mediated pathway [44]. After endocytosis, GLP-1R continues to signal from early endosomes, producing sustained cAMP signaling by internalized GLP-1Rs, but without increasing insulin release [45]. Acute β-arrestin-1 engagement can facilitate insulin release, but prolonged or excessive recruitment may reduce insulinotropic efficiency by altering receptor recycling and signaling balance [46]. Experimental evidence from β-arrestin2 knockout models shows reduced GLP-1R trafficking toward lysosomes and increased routing toward the trans-Golgi network, thereby altering receptor recycling and degradation dynamics [47].

Different GLP-1 receptor agonists (GLP-1RAs) show distinct patterns of receptor handling. For example, semaglutide and native GLP-1 promote relatively conventional recycling and internalization, whereas biased agonists like tirzepatide favor sustained receptor retention at the plasma membrane with reduced cytosolic trafficking [48]. These divergent patterns of β-arrestin utilization and endosomal signaling may underlie differences in drug-specific clinical profiles, including glucose control, durability of weight loss, and gastrointestinal tolerability.

The study of Dawed and colleagues investigated whether genetic variants could predict response to GLP-1 receptor agonists in 4571 adults with type 2 diabetes across observational studies and RCTs, measuring HbA1c change after six months [49]. The findings underscore that GLP-1 response is a polygenic trait influenced by both common and rare loci, and that integrating pharmacogenomics with functional analyses may advance precision medicine in type 2 diabetes. Namely, they found that the common GLP1R variant Gly168Ser (rs6923761) was associated with poorer response, while rare variants in ARRB1 (notably Thr370Met) enhanced response by reducing GLP1R internalisation and boosting cAMP and insulin secretion, with stronger effects in Hispanic and American Indian/Alaska Native populations. A combined genetic score showed up to a 30% difference in HbA1c reduction be-tween low- and high-response carriers, highlighting the clinical relevance of these variants, though effects were specific to GLP-1 receptor agonists and not other drug classes.

Semaglutide, is a drug with 94% homology with human GLP-1. Its chemical structure consists of 31 amino acids with two amino acid substitutions (Aib8 and Arg34) at position 8 reduces susceptibility to degradation by dipeptidyl peptidase-4 (DPP-4). Studies such as SUSTAIN, PIONEER, and STEP highlight its superiority compared to other GLP-1 receptor agonists and anti-obesity drugs in [50]. Moreover, the modifications improve the specific high-affinity binding to albumin, which slows down the degradation of semaglutide in plasma and results in decreased renal clearance and in prolongation of the half-life of semaglutide to ~1 week, making it appropriate for once-weekly administration [51]. As capable of β-arrestin recruitment, semaglutide can trigger receptor desensitization and internalization, which may limit the duration of signaling in β-cells [52]. Thus, semaglutide shows reduced β-arrestin recruitment and receptor internalization compared to endogenous GLP-1; however, current evidence does not conclusively classify semaglutide as a functionally or clinically biased GLP-1R agonist [52]. In comparative studies, complete suppression of β-arrestin recruitment produced slower internalization, reduced downregulation, and longer-lasting cAMP signaling than semaglutide [52], indicating that semaglutide retains functionally meaningful arrestin engagement, even if not at the highest level among GLP-1R agonists. Thus, semaglutide still supports measurable GLP-1R internalization under physiological albumin conditions, leading to receptor loss from the surface which could limit the duration of its glucose-lowering effects compared with "fully biased" ligands. In line with this, the complete abolition of β-arrestin signaling prolonged anti-hyperglycemic effects (superior glucose tolerance after 72 h compared to semaglutide) and enhanced chronic weight loss in mice [52].

The study of Sonoda et al. [46] through a combination of coimmunoprecipitation, RNAi knockdown, immunoblotting, cAMP assays, and pharmacological assays demonstrated that β-Arrestin-1 knockdown broadly attenuated GLP-1 signaling, causing decreased ERK and CREB activation and IRS-2 expression as well as reduced cAMP levels and impaired insulin secretion. However, β-arrestin-1 knockdown did not affect GLP-1 R surface expression and ligand-induced GLP-1 R internalization/desensitization. Thus, β-arrestin-1 interacts directly with GLP-1R and amplifies GLP-1 signaling and may influence β-cell performance and growth.

On the other hand, study of Bitsi et al. in vivo showed that β-arrestin 2 normally promotes receptor degradation and limits sustained signaling [47]. Namely, glycemic responses to the pharmacological GLP-1R agonist exendin-4 in adult β cell–specific β-arrestin 2 knockout (KO) mice showed that when β-arrestin 2 is absent, acute responses are impaired, but prolonged responses are enhanced through reduced receptor ubiquitination, avoidance of lysosomal degradation, and redirection of active receptors toward the trans-Golgi network, where intracellular signaling is sustained [47]. Of note, the role as loss of β-arrestin-1 does not affect GLP-1R internalization on the other hand, enhancing β-arrestin-2 action by the overexpression of G protein receptor kinase 5 (GRK5) increases GLP-1R endocytosis [53]. Nevertheless, these data establish that biased agonism at the GLP-1R is a key determinant of functional outcomes. Agonists that preferentially engage G protein signaling while avoiding β-arrestin recruitment (such as exendin-phe1) produce stronger, longer-lasting insulin release. Conversely, agonists that drive β-arrestin-2 engagement promote receptor endocytosis and desensitization, reducing prolonged secretory capacity. cAMP responses to exendin-4, with an intermediate pattern, were potentiated in arrestin-null cells, showing both increased potency and a trend toward higher efficacy over time compared to wild-type. This indicates that β-arrestins act primarily as inhibitory on GLP-1R cAMP signaling during chronic agonist exposure, rather than being essential for receptor internalization [53].

Semaglutide, similarly, is characterized as "not clinically confirmed as biased" GLP1R agonist [52], which preserve acute signaling support while also enabling desensitization pathways [47], which may contribute to clinical observations such as tachyphylaxis of gastrointestinal side effects. From a therapeutic perspective, semaglutide's may thus provide sufficient β-arrestin 1 engagement in responses; on the other, it remains vulnerable to β-arrestin 2–driven desensitization over time, which could limit its signaling durability in β cells under chronic exposure. This mechanistic framework may help explain why semaglutide shows robust efficacy for glycemic control and weight loss, but also why some adverse effects and tissue-adaptive responses emerge with continued administration (Figure 1).

β-arrestins are versatile regulators of GPCR behavior, contributing both to signal termination and to alternative signaling and adaptive receptor responses.

Semaglutide contains a large hydrophobic fatty diacid side chain that anchors the peptide to human serum albumin [66]. Albumin is the most abundant plasma protein, circulating at concentrations of 35–50 g/L with a half-life of approximately 19 days. It serves as a natural carrier for fatty acids, hormones, and drugs through multiple binding sites and undergoes neonatal Fc receptor-mediated recycling, which protects it from renal clearance and lysosomal degradation. Drugs engineered to bind albumin therefore benefit from prolonged systemic exposure. For semaglutide, this strategy results in a pharmacokinetic half-life of approximately seven days, enabling once-weekly dosing [6]. Due to albumin binding, there is a continuous reservoir of semaglutide in plasma, which leads to persistent exposure of GLP-1 receptors in pancreas as well as in extra-pancreatic targets (hypothalamus, liver, salivary glands). This prolonged exposure ensures continuous activation of GLP-1R, improving fasting and postprandial glucose control compared to short-acting agonists (like exenatide) [67]; however, in exocrine tissues such as salivary glands, continuous receptor engagement may impair physiological oscillations of signaling. Salivary secretion depends on finely tuned receptor activity, and sustained activation may paradoxically reduce glandular responsiveness, contributing to the salivary hypofunction. In contrast, shorter-acting GLP-1RAs such as exenatide are rapidly cleared and do not maintain chronic receptor occupation. Liraglutide, although also albumin-associated via fatty acid acylation, displays weaker binding and a shorter half-life (~13 h) than semaglutide [6]. Thus, semaglutide's unique depth of albumin association could explain its stronger tendency to induce tissue-level adaptations, including those leading to dry mouth. Importantly, sustained receptor occupation due to albumin binding may magnify the functional consequences of even modest β-arrestin recruitment (Figure 2). Thus, due to its strong albumin binding and prolonged systemic exposure, semaglutide may maintain longer receptor occupancy, which could theoretically influence receptor adaptation dynamics in tissues expressing GLP-1R, including salivary glands; however, this remains hypothesis-generating as no direct salivary pharmacodynamic measurements have been performed to date.

Real-world pharmacovigilance studies suggest that semaglutide is disproportionately associated with gastrointestinal adverse events compared to other GLP-1Ras [10,11]. Noteworthy, due to the multifactorial nature of xerostomia, any association between semaglutide and reduced salivary function must be interpreted with caution until controlled studies accounting for metabolic, autoimmune, neurological, pharmacologic, and age-related confounders are available. Nevertheless, while nausea, vomiting, and diarrhea dominate as symptoms in the reports, oral adverse events such as hypoaesthesia and xerostomia are increasingly recognized especially with semaglutide [1,10,11]. The median time-to-onset of gastrointestinal events is approximately four weeks [12], consistent with the early phase of receptor adaptation. Given that salivary glands express GLP-1R at low but detectable levels, chronic overstimulation could plausibly extend these adverse patterns to oral secretory dysfunction.

Comparative analyses suggest that liraglutide, exenatide, and dulaglutide display different balances of albumin binding and GLP1 receptor signaling bias, and thus may pose lower xerostomia risk. For example, exenatide, with rapid clearance, minimizes prolonged receptor overstimulation, while liraglutide's weaker albumin binding and shorter half-life limit its potential for sustained β-arrestin–driven desensitization. This highlights semaglutide's unique pharmacological niche: superior efficacy accompanied by variable side effects.

Salivary secretion often requires rhythmic, pulsatile activation of secretory apparatus, for example, bursts of saliva release in response to food stimuli. The key event in activation of fluid secretion is an increase in [Ca²⁺]i triggered by inositol 1,4,5-trisphosphate (IP3)-induced release of Ca²⁺ from ER via the IP3 receptor (IP3R) and cells finely tune the generation and amplification of [Ca²⁺]i signals for regulation of cell function [68]. These pulsatile Ca²⁺ signals depend on the ability of receptors to repeatedly activate and reset, and allow acini to engage in electrolyte transport, fluid secretion, and protein content in a balanced manner [69].

Chronic exposure to GLP-1 receptor agonists (GLP-1RAs) such as semaglutide, due to albumin binding, and recruitment of β-arrestins may disturb this rhythm. Sustained stimulation stabilize β-arrestin binding, "locking" the receptor in an internalized state [70], which promotes receptor ubiquitination, and receptor down-regulation as shown with muscarinic receptor agonists [71]. The outcome thus, may be the long-term desensitization and reduced responsiveness of salivary acinar cells. On the other side, sustained GLP1R/cAMP signaling could dampen or alter the normal Ca²⁺ oscillation pattern via Ca²⁺ -cAMP cross-talk [72]. Thus, either sustained cAMP signaling or receptor desensitization could alter salivary secretion.

Namely, salivary secretion in acinar cells is regulated by the interplay of cAMP and calcium signaling pathways. Within this framework, the PKA pathway is central: cAMP activation triggers PKA, which promotes protein secretion, facilitates fluid secretion by elevating intracellular calcium levels, and activates cAMP response element–binding protein (CREB), a key transcription factor coordinating these downstream signaling events [73]. However, while short bursts of cAMP may transiently potentiate Ca²⁺ release and secretion [72], sustained cAMP elevation could suppress oscillatory Ca²⁺-dependent Cl⁻ currents, partly by: redistributing Cl⁻ efflux to non-luminal pathways, reducing the amplitude of oscillatory pulses critical for fluid secretion and, possibly dampening Na⁺-K⁺-2Cl⁻ cotransporter activity [74]. If semaglutide favors sustained Gs/cAMP signaling, cAMP responses may disrupt cross-talk between cAMP and Ca²⁺ pathways as it has been shown that cAMP-elevating agents (e.g., forskolin + IBMX) could suppress Ca²⁺-activated oscillatory Cl⁻ secretion in salivary acinar cells by reducing the amplitude/duration of Ca²⁺-driven pulses [74]. Since Cl⁻ secretion induces fluid output, suppression of oscillatory currents may result in reduced salivary flow. Furthermore, receptor internalization via β-arrestins introduces "pauses" that allow signaling to reset. If semaglutide reduces β-arrestin recruitment, receptors remain at the membrane, sustaining Gs/cAMP signaling and affecting Ca²⁺ oscillations. Over time, this alters the salivary secretion, manifesting clinically as reduced fluid output and xerostomia.

The risk of salivary dysfunction may vary across GLP-1RAs. Currently approved GLP-1 agonists, including liraglutide, semaglutide, dulaglutide and lixisenatide, as well as exendin-4 have been examined for bias between cAMP and β-arrestin-2 recruitment [75]. The results showed that all were full agonists for both pathways except exendin-4, which showed a subtle reduction in efficacy for β-arrestin-2, while only liraglutide showed statistically significant bias in favour of β-arrestin-2 recruitment over cAMP [75]. Thus, Liraglutide, which recruits β-arrestins more strongly than semaglutide, may allow for more receptor recycling [76], in terms of salivary secretion, and preservation of oscillatory Ca²⁺-driven secretion. As a result, patients on liraglutide might experience fewer xerostomia-related symptoms compared with those on prolonged semaglutide therapy. While the mechanistic framework presented here aligns with established GLP-1R signaling and trafficking paradigms, the proposed Ca²⁺–cAMP disruption and β-arrestin–mediated adaptation in salivary acinar cells should be considered hypothesis-generating, as direct salivary-tissue evidence under semaglutide exposure is currently lacking. Findings from pancreatic β-cell and other GLP-1R–expressing models may not fully recapitulate exocrine gland physiology, and future human or experimental salivary-gland studies are needed to validate these mechanistic possibilities.