Dysregulated glucagon signaling contributes to hypoglycemia after vertical sleeve gastrectomy in mice Authors: Tyler M. Cook, Maigen Bethea, Silvania da Silva Teixeira, Carissa Birznieks, Andrew McClennan, Jasmine Hendrix, Richard KP Benninger, Mary-Elizabeth Patti, Darleen A. Sandoval Published: 2026 ABSTRACT ObjectivesAfter bariatric surgery, many individuals experience debilitating bouts of hypoglycemia, termed post-bariatric hypoglycemia (PBH). Our mouse model of vertical sleeve gastrectomy (VSG) mimics key aspects of PBH in humans. As glucagon is a key element of the counterregulatory hormonal response to hypoglycemia, the objective of this manuscript is to understand if glucagon responses and sensitivity are altered after VSG in mice.MethodsMice underwent sham or VSG surgery. We measured glucagon responses to insulin-induced hypoglycemia and mixed meal tolerance tests. We then measured glucose and insulin responses to different doses of exogenous glucagon in fed and fasted states. Lastly, we performed in vitro experiments to examine glucagon responses to glucose and amino acids.ResultsPostprandial glucagon levels were elevated in mice after VSG, but glucose responses to low doses of exogenous glucagon were blunted in the fasted and postprandial state. Administration of low dose glucagon in the fed state led to hypoglycemia in some VSG mice in parallel with stimulation of insulin release. This effect was partially prevented by pretreatment with the GLP-1 receptor antagonist exendin9-39. In islets isolated from VSG mice, glucagon release in response to high glucose was suppressed, but glucagon responses to alanine and high glucose were not different compared to sham controls.ConclusionsAltogether, our data suggest both glucagon responses, and sensitivity to exogenous glucagon are altered by VSG. The glucagon dysregulation was not observed in isolated islets suggesting gut signaling may be critical in driving glucagon responses after VSG. Introduction To combat the rising prevalence of obesity and associated metabolic syndrome, bariatric surgery has been the gold standard therapy for weight loss [1]. While anti-obesity medications, such as glucagon-like peptide (GLP-1) receptor agonists, are becoming more popular, metabolic bariatric surgery remains the most effective, durable and cost-effective anti-obesity intervention [2,3]. Vertical sleeve gastrectomy (VSG), which entails the removal of over 80% of the stomach along the greater curvature, is the most commonly performed procedure, followed by Roux-en-y gastric bypass (RYGB). In addition to weight loss, both VSG and RYGB improve glucose homeostasis such that many individuals with type 2 diabetes can reduce or eliminate the need for medications for glycemic management [[2], [3], [4]]. Despite these metabolic benefits of bariatric surgery, many individuals experience complications such as vitamin deficiencies [5], dumping syndrome [6], and hypoglycemia [7,8]. Hypoglycemia after bariatric surgery, termed post-bariatric hypoglycemia (PBH), is increasingly recognized as a common and often-serious complication that impacts up to one third of patients [7]. PBH manifests typically years after surgery. Episodes of hypoglycemia most commonly occur following a meal; after an initial rapid rise in glucose (greater than in non-surgical patients), glucose rapidly declines to hypoglycemic levels [7,8]. This phenomenon also occurs in rodent models of bariatric surgery [9,10]. Specifically, our previous work demonstrated that rats and mice exhibit key features of PBH, including increased glucose variability, increased postprandial GLP-1 and insulin, and postprandial hypoglycemia in a subset of mice [10]. Because the gastrointestinal tract anatomy is so dramatically altered by bariatric surgery, food digestion, nutrient entry, and gut hormone release are dynamically changed as well [[11], [12], [13]]. Several human and rodent studies show elevated glucagon levels after a meal, a time when they are suppressed in individuals without diabetes [9,14]. Additionally, other studies find reduced glucagon levels in response to hypoglycemia in patients with PBH [[15], [16], [17]]. Thus, further mechanistic studies of glucagon signaling after bariatric surgery are warranted. While the role of glucagon as a counterregulatory hormone has long been known, more recent work has shown that it can have a glucose-lowering role in fed conditions. Capozzi et al. demonstrated that when β-cells have increased activity in response to high glucose levels, glucagon signals through the GLP-1 receptor to enhance glucose-stimulated insulin release and lower glycemia [18]. These experiments were performed in lean animals with healthy islet signaling, and similar findings have also been demonstrated in healthy humans [19]. However, more work is needed to understand how obesity and associated metabolic syndrome disrupt these dynamics of glucagon signaling. Lastly, therapeutic administration of pharmacological doses of glucagon can mitigate acute hypoglycemia in individuals with PBH [20,21]. Thus, we sought to better understand how the dynamics of endogenous glucagon release and signaling might be altered by VSG, and whether dysregulated glucagon signaling contributes to PBH. To investigate how obesity and bariatric surgery impact the interaction between glucagon signaling and glucose homeostasis, we utilized mouse models of obesity and weight loss via VSG. We found that glucagon secretion and signaling are altered by diet-induced obesity, and further disrupted by VSG, such that postprandial glucagon contributes to hypoglycemia via increased insulin secretion and concomitant increases in insulin sensitivity. Our data offers insights into mechanisms that contribute to PBH in mice. Methods Animals All animal experiments were approved by the University of Colorado Anschutz Medical Campus Institutional Animal Care and Use Committee (IACUC). Eight-week-old C57Bl/6J male mice (Jackson Laboratories; Strain #000664) were fed 60% high fat diet (HFD; Research Diet #D12492) or normal chow (Teklad #2920) for 16-weeks. Mice were housed in a temperature and light-controlled (14/10-hour light/dark cycle) facility with ad libitum (ad lib) access to food and water, unless otherwise stated for a given experiment. VSG surgery C57Bl/6J mice were single-housed and fed HFD starting at 8 weeks of age for 8 weeks to induce weight gain before undergoing VSG surgery similar to our previous work [[22], [23], [24]]. The day before surgery, solid food was removed, and mice were given Diet Gel Boost (Cat #72-04-5022). On the day of surgery, mice were anesthetized via isoflurane inhalation on a warm heating pad and were given buprenorphine SR and carprofen for pain management. For VSG surgery, a laparotomy was performed, abdominal cavity flushed, and approximately 80% of the stomach was clamped via two LigaClips. The portion outside the clips was resected and the clips were then stabilized with suture. For sham surgery, laparotomy and abdominal cavity flushing were performed similarly, without ligating the stomach. The abdominal cavity was then sutured closed, and mice recovered on a heating pad. For three days post-operation, mice were given fresh diet gel, carprofen for pain, and gentamycin to prevent infection. Three days after surgery, HFD was resumed, and diet gel was removed on the fourth day post-operation. EchoMRI (Echo Medical Systems) was performed prior to and at 4-weeks postoperatively to assess changes in body composition. General experimental setup for assessing glucose and hormone levels Singly housed mice were moved to a quiet procedure space and allowed to acclimate for 2 h before testing. Mice were studied under ad lib “fed” conditions (food removed in the morning 2 h before testing) or after an overnight fast of 16 h. Blood was sampled via tail nicks, and blood glucose was measured with Accu-Check glucometers (Roche Pharmaceuticals). In some experiments, additional blood was collected from tail bleeds in EDTA tubes, and plasma was separated via centrifugation and stored at −80 °C. for hormonal assessments (see below). Assessing plasma glucagon responses Response to Insulin: Fed mice were injected with 0.75 U/kg of insulin (Humulin R U-100, Lilly), and blood glucose was assessed at 0, 15, 30, 45, 60, and 120-min. Blood samples were collected for subsequent plasma glucagon measurement at 0 and 45 min. Estimated time spent in the hypoglycemic range was calculated by adding consecutive time intervals below 50 mg/dL. If mice became excessively lethargic after insulin and were deemed unable to complete the 120-min study, they were rescued with dextrose and removed from the study. Some mice completed the study and were then rescued with glucose immediately after the 120-min blood sampling, and these mice remained in the study. Mixed meal tolerance test: Morning fasted mice received an oral gavage of 200 μL of Ensure (Abbot Nutrition Cat. #57231) with 30 mg dextrose added as this caloric load reliably increases gut peptide responses in mice [23]. Blood glucose was measured at 0, 15, 30, 45, 60, and 120-min. Plasma glucagon was assessed at 0 and 15-min after gavage. Assessing responses to glucagon Responses to Glucagon: Glucagon (Millipore Cat#G1774) was reconstituted in PBS with 0.05% acetic acid the morning of the experiment. Fed or overnight fasted (as described above) mice were injected intraperitoneally with 1 mg/kg (high dose) or 20 μg/kg glucagon (low dose). These doses and experimental conditions have been used previously to stimulate insulin and/or glucose in lean mice [18]. Exendin 9-39 Experiments: Low dose glucagon experiments were repeated as above in the fed state 15-min after mice were injected with 50 μg Exendin9-39 (Bachem Cat. #4017799) or vehicle (saline). Calculations: Incremental Area Under the Curve (iAUC) was calculated via the trapezoid rule subtracting the baseline measurement from each subsequent time point. For Ex9 experiments, the baseline was calculated from the 0-min time point after Ex9 pretreatment, to assess the relative change in glucose after glucagon injection with and without Ex9 pretreatment. Plasma hormone analysis ELISAs: Plasma insulin and glucagon were measured using Crystal Chem ELISA (Cat #90080) and Mercodia Kits (Cat #10-1281-01), respectively, following manufacturer instructions. Assessing hepatic glucose metabolism Glycogen Content: Liver glycogen content was measured in ad lib fed mice (after 2-hour morning fast) or 2-hours after a mixed meal tolerance test similar to previous [23,25]. Livers were harvested and snap-frozen in liquid nitrogen, homogenized and 10–20 mg of liver was separated for glycogen extraction using a calorimetric assay (Abcam ab169558) according to manufacturer’s instructions. Pyruvate Tolerance Test: The ability to the liver to utilize substrates for gluconeogenesis was assessed using a pyruvate tolerance test [26]. After a 6-hour fast, mice were injected with 2 g/kg of sodium pyruvate (Millipore Sigma) in saline, and blood glucose was measured at 0, 30, 60, 120, and 180-min. Islet experiments Fed animals were anesthetized with 80 mg/kg ketamine (Zetamine, NDC 13985-584-10; Vet One/MWI) and 20 mg/kg xylazine (Rompun, NDC 17033-099-05, Dechra Veterinary Products) and then euthanized by exsanguination with cervical dislocation as a secondary method. Collagenase (Clzyme RI, VitaCyte) was injected into the pancreas through the pancreatic duct, and the pancreas was isolated immediately following infusion. Isolated pancreata were incubated at 37 °C, and islets were hand-picked. Islets were incubated overnight in 1640 RPMI medium (Gibco) with 10% FBS, 100 U/mL penicillin and 100 μg/mL streptomycin (Thermo Fisher) at 37 °C with 5% CO2. This is similar to our previous work [27,28]. After overnight incubation, replicates of 10 islets per tube (insulin, glucagon assays) or 40 islets per tube (somatostatin assays) were incubated in Krebs–Ringer buffer (128.8 mM NaCl, 5 mM KCl, 1.2 mM KH2PO4, 2.5 mM CaCl2, 1.2 mM MgSO4, 1 mM HEPES, 0.1% BSA, pH 7.4) with 2 mM glucose for 1 h at 37 °C. Islets were then incubated in 500 μL Krebs buffer for 1 h in treatment media. Supernatant was collected for analysis of insulin, glucagon, and somatostatin. To assess content, islets were lysed by 2% Triton-X-100 in deionized water, followed by a freeze–thaw cycle. Insulin, glucagon, and somatostatin concentrations were measured using mouse ultrasensitive ELISA kits (Crystal Chem, ALPCO, Phoenix Pharmaceuticals, respectively) according to manufacturer’s instructions. Islets were harvested from two independent cohorts at 12 or 15 weeks post-surgery, and data from the two cohorts were not pooled and were analyzed separately. Statistical analysis GraphPad Prism (version 10.4.0) was used for statistical analysis and graphs. The statistical test and sample size are indicated in figure legends for all experiments. Briefly, depending on the experimental design, normally distributed data were analyzed with a two-way ANOVA for repeated measures or T-tests. The Šídák’s post-hoc analysis was used for experiments with more than three time points (glucose curves) or the Fisher’s pairwise LSD was utilized as a post-hoc analysis with experiments involving two time points to determine where significant differences lie. Significance was accepted at p < 0.05. Data are reflected as mean +/− standard error of the mean (SEM). Results Glucagon levels are dysregulated after VSG Our previous work demonstrated that mice had key features of PBH after VSG, including increased glucose variability and postprandial hypoglycemia [10]. As we have previously shown [29], VSG caused significant and sustained weight loss (Supp 1A), driven by a reduction in food intake (Supp 1B). After weight loss stabilized, VSG mice did not compensate by increasing food intake (Supp 1B), in contrast to the increased food intake which often occurs after weight loss from caloric restriction [[30], [31], [32]]. The lost weight after VSG was primarily fat mass without significant changes in fat free mass (Supp 1C-D). By seven weeks post-surgery, VSG resulted in a pronounced increase in insulin sensitivity, as assessed via an insulin tolerance test (ITT; Figure 1A). During the ITT, VSG mice had significantly lower nadir glucose levels (Figure 1B), and 28% of VSG mice (7/25) had blood glucose levels below 50 mg/dL. In contrast, hypoglycemia did not occur in any sham mice (Figure 1B). We estimated the duration of hypoglycemia during the ITT, by adding consecutive time points when glucose was measured below 50 mg/dL, and VSG mice had significantly more time spent in hypoglycemia (Figure 1C). In fact, some VSG mice had to be rescued with glucose during or after the ITT. Despite persistently lower glucose level during the ITT, VSG mice had lower glucagon levels at 45 min (Figure 1D), indicating blunted glucagon response to hypoglycemia. Additionally, VSG mice had improved glucose tolerance (Figure 1E) but increased glucagon levels during a mixed meal tolerance test (Figure 1F). These data indicate that after VSG, glucagon responses are distinctly impacted depending on the glycemic state, with lower levels during hypoglycemia and higher levels postprandially.Figure 1Sham and VSG mice received insulin (0.75 U/kg) via IP injection after morning fast, and blood glucose was measured (A). Blood glucose nadir was significantly lower (B), and estimated time hypoglycemic was significantly higher in VSG vs. Sham mice (C). Glucagon levels at 45′ post insulin injection were significantly lower in VSG vs. Sham mice (D). VSG vs. Sham mice received an oral gavage of 200uL ensure +30 mg dextrose and blood glucose was measured (E). Glucagon levels were significantly higher 15-min after gavage (F). Purple dotted line indicates threshold for hypoglycemia at 50 mg/dL, and purple values (B) indicate number of mice that dropped 50 mg/dL. Two-way repeated measures ANOVA (A, D, E, F) with Šídák’s post-hoc (A, E) and Fisher’s LSD (D, F); unpaired t-test (B–C). #p < 0.01 main effect of surgery; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.00001; n = 6–18 mice/group. Sham and VSG mice received insulin (0.75 U/kg) via IP injection after morning fast, and blood glucose was measured (A). Blood glucose nadir was significantly lower (B), and estimated time hypoglycemic was significantly higher in VSG vs. Sham mice (C). Glucagon levels at 45′ post insulin injection were significantly lower in VSG vs. Sham mice (D). VSG vs. Sham mice received an oral gavage of 200uL ensure +30 mg dextrose and blood glucose was measured (E). Glucagon levels were significantly higher 15-min after gavage (F). Purple dotted line indicates threshold for hypoglycemia at 50 mg/dL, and purple values (B) indicate number of mice that dropped 50 mg/dL. Two-way repeated measures ANOVA (A, D, E, F) with Šídák’s post-hoc (A, E) and Fisher’s LSD (D, F); unpaired t-test (B–C). #p < 0.01 main effect of surgery; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.00001; n = 6–18 mice/group. Exogenous glucagon raises glucose and insulin after VSG Given that VSG mice had lower glucagon levels during the ITT, yet increased postprandial glucagon, we next examined how VSG might alter glucose and insulin responses to glucagon under different doses and nutritional states. It has been previously shown that high doses of glucagon stimulate insulin secretion when nutrients are present, but that same dose of glucagon in the fasted state failed to increase insulin and increased glucose levels [18]. Thus, the hypo-vs. hyperglycemic effect of glucagon is dependent upon nutrient status. However, these experiments were performed on lean mice [18], and diet-induced obesity causes a hyperinsulinemic and prediabetic state in which insulin secretion is increased in both fasting and prandial conditions. Thus, we first sought to understand if obesity alters the response to exogenous glucagon, and then we examined the impact of weight loss via bariatric surgery. Initially, we used the dose of glucagon (1 mg/kg) in fed mice that had been previously shown to stimulate insulin secretion in lean mice [18]. Compared to normal chow (NC)-fed mice, obese mice that had been fed a 60% HFD for 16 weeks had similar glucose responses after 1 mg/kg of glucagon in the fed state (Figure 2A–B), and both groups had significant increases in insulin levels at 15-min (Figure 2C).Figure 2Normal chow (NC) vs high fat diet (HFD)-fed mice (A-C) and sham vs VSG (D-F) animals received 1 mg/kg glucagon (Gcg) IP injection after morning fast and blood glucose was measured (A, D), iAUC was calculated (B, E), and insulin was measured (C, F). Two-way repeated measures ANOVA (A, C, D, F), Šídák’s post-hoc (A, D) Fisher’s pairwise LSD (C, F); Unpaired T-test (B, E); ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; n = 8–11 mice/group. Normal chow (NC) vs high fat diet (HFD)-fed mice (A-C) and sham vs VSG (D-F) animals received 1 mg/kg glucagon (Gcg) IP injection after morning fast and blood glucose was measured (A, D), iAUC was calculated (B, E), and insulin was measured (C, F). Two-way repeated measures ANOVA (A, C, D, F), Šídák’s post-hoc (A, D) Fisher’s pairwise LSD (C, F); Unpaired T-test (B, E); ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; n = 8–11 mice/group. When we compared VSG-operated animals and sham surgery controls, VSG mice also had similar glucose levels over time (Figure 2D) and similar glucose iAUC (Figure 2E), and both groups had increased insulin levels 15-min post injection (Figure 2F). Thus, the 1 mg/kg dose of glucagon increased glucose and insulin levels in lean and obese mice and after surgical weight loss. Hepatic effect of low dose glucagon is blunted after VSG We next sought to distinguish between hepatic vs. pancreatic responses to glucagon by utilizing a lower dose of glucagon and comparing responses in the fed vs. fasted states. Based on previous experiments in lean mice, administering a low dose of glucagon in the fasted state should primarily stimulate hepatic glucose production [18], while minimally enhancing glucose-stimulated insulin release from β-cells. Accordingly, we injected a lower dose of glucagon (20 μg/kg) after an overnight fast. Under these conditions, glucagon increased glucose in both NC and HFD-fed mice (Figure 3A). The incremental glucose response to glucagon (iAUC) for HFD-fed mice was lower than NC-fed mice (Figure 3B), likely because of the much greater basal and glucagon-stimulated insulin levels in the obese mice (Figure 3C) which likely contributed to more rapid return to baseline glucose levels.Figure 3Male mice received 20 μg/kg glucagon via IP injection after overnight fasting (A-G). Blood glucose levels were measured (A, D), incremental area under the curve (iAUC; B, E) and nadir glucose levels (F) calculated, and insulin levels at baseline and 15 min post glucagon were measured (C-G). Two-Way repeated measures ANOVA (A, C, D, G), Šídák’s post-hoc (A, D) Fisher’s pairwise LSD (C, G); Unpaired T-Test (B, E, F); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; n = 7–12 mice/group. Dotted lines (D) indicate average baseline glucose level for sham (grey) vs. VSG mice (green). Male mice received 20 μg/kg glucagon via IP injection after overnight fasting (A-G). Blood glucose levels were measured (A, D), incremental area under the curve (iAUC; B, E) and nadir glucose levels (F) calculated, and insulin levels at baseline and 15 min post glucagon were measured (C-G). Two-Way repeated measures ANOVA (A, C, D, G), Šídák’s post-hoc (A, D) Fisher’s pairwise LSD (C, G); Unpaired T-Test (B, E, F); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; n = 7–12 mice/group. Dotted lines (D) indicate average baseline glucose level for sham (grey) vs. VSG mice (green). We next assessed how VSG alters hepatic responses to this low dose of glucagon. While sham animals only exhibited modest rises in glycemia throughout the 2-hour test, glycemic responses in VSG mice were even lower in magnitude (Figure 3D–E). VSG animals had a brief spike in glucose after glucagon, but for many mice, glycemia quickly dropped below baseline levels leading to lower (Figure 3D) iAUC (Figure 3E). The glucose nadir between groups was similar (Figure 3F). Given these modest rises in glycemia for sham mice and drops in blood glucose for VSG mice, we measured plasma insulin levels and found that both groups had significant elevations 15-min after glucagon (Figure 3G). Thus, the hyperinsulinemic effect of glucagon persists even in the fasted state in obese mice and after VSG-induced weight loss. To determine whether VSG animals would be able to mount a sufficient increase in hepatic glucose production (HGP) based on substrate availability, we performed a pyruvate tolerance test where the increase in glucose in response to pyruvate is a marker of the degree of HGP. We found that VSG mice had significantly lower glucose responses to pyruvate compared to sham surgery mice confirming reduced capacity to stimulate HGP (Supp 2A). We also assessed postprandial hepatic glycogen levels after mixed meal gavage or in an ad lib fed state in sham and VSG. After a mixed meal gavage, VSG animals had lower hepatic glycogen levels (Supp 2B), and VSG animals also trended toward lower glycogen levels in the ad libitum fed state (Supp 2C). These data demonstrate that limited capacity for HGP likely contributes in part to PBH manifestation. Low dose glucagon can induce hypoglycemia after VSG Given that glucagon levels were elevated in VSG animals in the fed state and after a mixed meal gavage (Figure 1F), we next asked whether the low dose of glucagon might differentially regulate glycemia in the fed setting. Again, we started by assessing how diet-induced obesity impacted glucose and insulin responses. In the fed state, 20 μg/kg glucagon mobilized glucose similarly between NC and HFD-fed mice (Figure 4A–B), but obese mice had significantly higher insulin levels 15-min after glucagon injection (Figure 4C).Figure 4Male mice received 20 μg/kg glucagon via IP injection after feeding ad libitum (A-G). Blood glucose levels were measured (A, D), incremental area under the curve (iAUC; B, E) and nadir glucose levels (F) calculated, and insulin levels at baseline and 15 min post glucagon were measured (C, G). Islets derived from sham or VSG mice were cultured and hormones were measured after low glucose (2 mM dextrose) or high glucose (10–11 mM dextrose) treatments (H-J). Insulin secretion was measured (H), and fold-change in hormone secretion from low to high glucose were calculated (I). Insulin secretion was measured after treatment with 10 nM glucagon in high glucose media (J). Two-Way ANOVA, with repeated measures and Šídák’s post-hoc (A, D), or Fisher’s pairwise LSD (C, G, H, I, K); Unpaired T-test (B, E, F, I); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; n = 5–10 mice/group. Dotted lines (D) indicate average baseline glucose level for sham (grey) vs. VSG mice (green), and purple values (F) indicate the number of mice with a blood glucose <50 mg/dL. For islet experiments, n = 2–5 mice/group; 2–3 replicates/mouse. Male mice received 20 μg/kg glucagon via IP injection after feeding ad libitum (A-G). Blood glucose levels were measured (A, D), incremental area under the curve (iAUC; B, E) and nadir glucose levels (F) calculated, and insulin levels at baseline and 15 min post glucagon were measured (C, G). Islets derived from sham or VSG mice were cultured and hormones were measured after low glucose (2 mM dextrose) or high glucose (10–11 mM dextrose) treatments (H-J). Insulin secretion was measured (H), and fold-change in hormone secretion from low to high glucose were calculated (I). Insulin secretion was measured after treatment with 10 nM glucagon in high glucose media (J). Two-Way ANOVA, with repeated measures and Šídák’s post-hoc (A, D), or Fisher’s pairwise LSD (C, G, H, I, K); Unpaired T-test (B, E, F, I); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; n = 5–10 mice/group. Dotted lines (D) indicate average baseline glucose level for sham (grey) vs. VSG mice (green), and purple values (F) indicate the number of mice with a blood glucose <50 mg/dL. For islet experiments, n = 2–5 mice/group; 2–3 replicates/mouse. Next, we performed similar experiments in sham vs. VSG mice. In ad lib fed mice, glucagon briefly increased glycemia, but had a net glucose-lowering effect in VSG mice (Figure 4D), as indicated by a significantly lower average iAUC (Figure 4E). Glucose nadir after glucagon was also significantly lower for fed VSG animals, and glycemia dropped below 50 mg/dL for 2/11 mice (Figure 4F). Glucagon increased insulin levels in both groups, however insulin levels of VSG mice were lower 15-min after glucagon compared to sham mice (Figure 4G). The incidence of hypoglycemia, despite lower or similar insulin levels, is likely be explained by increased insulin sensitivity after VSG (Figure 1A), and slightly lower hepatic glycogen stores (Supp 2C). As an important note, after the 20 μg/kg dose of glucagon, plasma levels of glucagon reached approximately 80–100 pmol/L in both sham and VSG mice (Supp 2D), similar to the levels measured after insulin-induced hypoglycemia (Figure 1D). Therefore, we interpreted the low dose experiments as physiological responses to endogenous levels of glucagon, while the high dose experiments indicated the impact of supraphysiological glucagon levels. Next, we sought to determine if β-cell sensitivity to glucagon is altered after bariatric surgery. To do so, we cultured islets from sham or VSG mice and tested responses under high and low glucose conditions. Importantly, there were no differences in insulin content between sham and VSG islets (Figure 4H). High glucose conditions similarly increased insulin secretion in both sham and VSG islets (Figure 4I–J). Glucagon (10 nM) stimulated greater insulin secretion under high glucose conditions (10 mM) compared to low glucose (2 mM), and VSG islets had relatively lower stimulation of insulin (Figure 4K). Similar to our in vivo findings (Figure 4G), VSG islets had lower glucagon-stimulated insulin levels at 10 mM dextrose compared to sham islets (Figure 4K). These data demonstrate that changes in islet function do not explain the hypoglycemic responses to glucagon after VSG. Overall, our data shows that after VSG, the hypoglycemic effect of glucagon is predominant. In the face of lowered hepatic glycogen, lowered ability to convert gluconeogenic precursors to glucose and thus lowered capacity for hepatic glucose production with increased insulin sensitivity, the textbook counterregulatory hormone glucagon causes hypoglycemia, rather than reversing or preventing it. Exendin 9-39 prevents hypoglycemic effect of glucagon after VSG Given that it has been previously shown that glucagon stimulates insulin via activation of the GLP-1 receptor (GLP-1R) on β-cells [18], we sought to determine if exendin 9-39 (Ex9) could prevent the glucose lowering effect of glucagon in VSG mice. In a crossover experiment, mice were given vehicle or 50 μg Ex9 via IP injection [14,23,33]-min prior to a 20 μg/kg glucagon injection. Pretreatment with Ex9 before glucagon dramatically increased glucose levels in both sham and VSG animals, normalizing VSG glucose to sham vehicle levels (Figure 5A–B, Supp 3A). Although glucagon still caused a net glucose lowering effect for most VSG animals (Figure 5B), Ex9 pretreatment completely prevented hypoglycemia (Figure 5C), significantly raised the nadir glucose level compared to vehicle pretreatment (Figure 5C), and attenuated the insulin-stimulating effect of glucagon (Figure 5D, Supp 3B). Intriguingly, Ex9 pretreatment alone (prior to exogenous glucagon administration) more potently raised endogenous glucagon levels in VSG animals compared to sham controls (Supp 3C). Ex9 pretreatment did not increase glucose responses in VSG animals to the levels of shams, potentially due to the insulin resistance of the sham animals or due to additional deficits in hepatic glucagon signaling and/or reduced capacity of hepatic glucose production in VSG animals. Overall, this data supports the notion that glucagon causes hypoglycemia after VSG by stimulating insulin release via GLP-1R, and blocking GLP-1R signaling partially restores the hyperglycemic effect of glucagon.Figure 5Ad libitum fed male mice received 20 μg/kg glucagon via IP injection, after pre-treatment of vehicle or 50 μg Ex9-39 (Ex9). Blood glucose levels (A), incremental area under the curve (iAUC; B) calculated from 0′ time point, glucose nadir (minimum; C), and change (Δ) in insulin levels between 0 and 15 min (D), in sham vs VSG male mice. Number of animals with nadir below 50 mg/dL (indicated by purple dotted line) in parentheses (purple text, C). Two-Way repeated measures ANOVA; Fisher’s pairwise LSD; #p < 0.05 main effect of ex9; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; n = 7–8 mice/group. Ad libitum fed male mice received 20 μg/kg glucagon via IP injection, after pre-treatment of vehicle or 50 μg Ex9-39 (Ex9). Blood glucose levels (A), incremental area under the curve (iAUC; B) calculated from 0′ time point, glucose nadir (minimum; C), and change (Δ) in insulin levels between 0 and 15 min (D), in sham vs VSG male mice. Number of animals with nadir below 50 mg/dL (indicated by purple dotted line) in parentheses (purple text, C). Two-Way repeated measures ANOVA; Fisher’s pairwise LSD; #p < 0.05 main effect of ex9; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; n = 7–8 mice/group. Hyperglycemia suppresses glucagon secretion after VSG To determine whether the postprandial hyperglucagonemia after VSG is a surgery induced alteration in islet function, we performed another series of in vitro experiments in isolated islets. Islets were cultured from sham or VSG mice and treated with low (2 mM) or high (10–11 mM) glucose plus the amino acid alanine (Ala). Glucagon levels were similarly elevated in low glucose conditions in both sham and VSG islets, and high glucose exposure lowered glucagon secretion from VSG islets but failed to do so for sham controls (Figure 6A–B). We previously saw increased glucagon in vivo after a mixed meal tolerance test (Figure 1F) which consists of glucose, amino acids and fat. It is known that amino acids most strongly stimulate glucagon secretion from the α-cell. We found that glucagon responses to Ala were similar between sham and VSG α-cells (Figure 6C). Additionally, we did not detect differences in somatostatin (SST) between the glucose conditions or surgery groups (Supp 4A-B). Overall, glucagon secretion in response to glucose or amino acids was not higher in VSG islets, as we had predicted. Altogether, these data failed to demonstrate intrinsic alterations in pancreatic islet function after VSG that could explain either the hyperglucagonemia or hyperinsulinemic response to glucagon. Rather, there is likely a more complicated disruption in the interorgan communication from the gut to the pancreas contributing to hyperglucagonemia and PBH.Figure 6Islets derived from sham or VSG mice were cultured and glucagon was measured after low glucose (2 mM dextrose) and high glucose (11 mM dextrose) treatments (A–B). Fold-change in hormone secretion from low to high glucose were calculated (B). Glucagon secretion in high glucose media (10 mM dextrose) was measured after treatment with 1 mM and 10 mM alanine (C). Two-way ANOVA; Fisher’s pairwise LSD; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; n = 2–5 mice/group; 2–3 replicates/mouse. Islets derived from sham or VSG mice were cultured and glucagon was measured after low glucose (2 mM dextrose) and high glucose (11 mM dextrose) treatments (A–B). Fold-change in hormone secretion from low to high glucose were calculated (B). Glucagon secretion in high glucose media (10 mM dextrose) was measured after treatment with 1 mM and 10 mM alanine (C). Two-way ANOVA; Fisher’s pairwise LSD; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; n = 2–5 mice/group; 2–3 replicates/mouse. Discussion Here, we used a mouse model of bariatric surgery to advance our understanding of mechanisms for PBH. We found that after VSG, postprandial plasma levels of glucagon are elevated and glucagon response to hypoglycemia is reduced. We also saw enhanced glucagon-stimulated insulin secretion in HFD-fed mice, which when combined with the increased insulin sensitivity after VSG, promotes postprandial hypoglycemia. The hypoglycemic effect of glucagon in VSG mice was blocked with the GLP-1R antagonist, exendin 9-39, but this did not increase glucose levels to that observed in the sham animals. This suggests an important role for reduced glycogen content after VSG, as well as potential deficits in either hepatic glucagon signaling, gluconeogenesis, or additional impairments in hepatic function. In support of this, we saw reduced glucose responses to exogenous pyruvate and lower hepatic glycogen levels in VSG mice. Additionally, increased insulin sensitivity in VSG mice likely contributed to the reduced glucose responses to glucagon, even after Ex9 pretreatment. The effects on glucagon secretion and ability of glucagon to increase insulin secretion were not due to intrinsic changes in islet function but are likely a result of multiple in vivo responses. Altogether these results in our mouse model of VSG suggest that the physiological response to nutrients after VSG sets up a scenario for increased risk of hypoglycemia. Lastly, glucagon-stimulated insulin secretion may contribute to the initiation of hypoglycemia, while additional deficits in hepatic glucagon signaling, glucose production, and counterregulation contribute to hypoglycemia persistence and severity. Our findings build upon the work of Capozzi et al. [18] and Gray et al. [19], demonstrating the hyperinsulinemic potential of glucagon in the fed state, and we extend these findings demonstrating alterations in the context of diet-induced obesity (and impaired glucose tolerance) and surgical weight loss. Our data demonstrate that glucagon signaling is altered after diet-induced obesity, enhancing insulin secretion. In the fed state, a low dose of glucagon stimulated insulin to a larger degree in obese mice, compared to NC-fed lean mice. However, lean mice still had an increase in insulin levels in these experimental conditions, which was not the case in the studies performed by Capozzi et al. [18]. The timing of sampling may explain this discrepancy, as we measured insulin at 15-min to understand hormonal determinants of the glucose curves over 2 hours, whereas the prior work examined insulin at 10-min. Our HFD-fed C57Bl/6J mice are hyperglycemic, hyperinsulinemic, and less sensitive to insulin, but not completely insulin intolerant [34,35]. Accordingly, after low doses of glucagon, HFD-fed mice had higher insulin levels and relatively greater reductions in glucose in the fasted state (Figure 3B), but similar drops in the fed state (Figure 4B). Previous work has demonstrated that diet-induced obesity blunts hepatic glucagon signaling in mice [34] and humans [36], and our data suggests that glucagon-stimulated insulin secretion is also dysregulated. Our islet experiments suggest that there is not an intrinsic shift in islet function leading to hyper-secretion of glucagon or insulin with VSG. In fact, we found that VSG normalized hyperglycemic suppression of glucagon secretion (Figure 6A), indicating the restoration of normal α-cell glucose sensing, which is often disrupted in obesity and type 2 diabetes [37]. Additionally, glucose-stimulated insulin levels were lower in VSG animals, potentially indicating a slight reversal of the hyperinsulinemia which often occurs in prediabetes [38,39]. Somatostatin and delta-cell function is critical in regulating both α- and β-cell function [40,41], and diet-induced obesity diminishes somatostatin-secretion, which contributes to hyperglucagonemia [42]. However, somatostatin responses were similar between sham and VSG mice suggesting that this effect was not restored by VSG (Supp 4A-B). Conversely, in isolated islets, glucagon responses to high glucose were suppressed and insulin response were significantly lower compared to sham mice. Additionally, a mixture of glucose and amino acids did not more potently stimulate glucagon secretion from VSG islets. This differs from the in vivo experiments in which VSG mice had higher postprandial levels of glucagon. Thus, instead of an intrinsic shift in pancreatic secretory function, we suspect that altered signaling from the gut to the pancreas after VSG further increases hyperglucagonemia and hyperinsulinemia. After bariatric surgery, fasting amino acid levels are generally lower, but rapid gastric emptying after a meal results in exaggerated spikes of many amino acids [43], which may serve to enhance glucagon release. Interestingly, however, patients with PBH have reduced levels of multiple amino acids compared to patients without hypoglycemia after RYGB [43]. Furthermore, elevated postprandial glucose and gut hormones such as GLP-1 should inhibit glucagon release, and collectively contribute to enhanced insulin release [12,44,45]. Thus, the source of elevated postprandial glucagon is perplexing. Some studies suggest that bariatric surgery could potentially induce meal-stimulated glucagon production from intestinal L-cells [9]. While it is possible that the increases in postprandial glucagon are coming from the intestine rather than the pancreas, only a small amount of immunoreactive glucagon was observed in the jejunal section of patients after RYGB [9], and it is difficult to make the same prediction for VSG surgery, where no intestinal rearrangement occurs. Additionally, we have previously shown that pancreatic preproglucagon is necessary for glucose improvements after VSG, while intestinal preproglucagon is dispensable, and the latter mice had no detectable glucagon levels [23]. Thus, we suspect that other postprandial gut signals stimulate excessive pancreatic secretion of glucagon after bariatric surgery. A key finding we report here is that insulin causes prolonged hypoglycemia in VSG mice, where some mice even need to be rescued with glucose after insulin tolerance testing. The reduced glucagon response to hypoglycemia likely contributed to this persistent hypoglycemia. Clinical studies also demonstrate diminished secretion of glucagon, as well as of other counterregulatory hormones, in response to experimental hypoglycemia after bariatric surgery [14,16,46]. Additionally, it is known that repeated bouts of hypoglycemia further blunt counterregulatory responses to subsequent episodes of hypoglycemia in rodents and humans [[47], [48], [49]]. This, along with altered glucagon signaling after bariatric surgery may further blunt counterregulatory responses and endanger individuals [48,50]. Future studies should clarify the role of repeated exposure to hypoglycemia after bariatric surgery to determine whether other counterregulatory hormone responses during hypoglycemic clamps are even more impaired in individuals with PBH. Why counterregulation fails in the first place after bariatric surgery is unclear, but one hypothesis is that glucose sensing in the brain is altered. The brain relies heavily on glucose for fuel, so dedicated neurons within the hindbrain and hypothalamus detect drops in blood glucose and then trigger feeding responses, adrenal medullary secretion, glucagon secretion and more to supply more glucose to the brain [51,52]. One culprit within the brain which might contribute to diminished glucagon release during hypoglycemia after bariatric surgery is fibroblast growth factor 15 (FGF15, FGF19 is the human ortholog) expressing neurons. FGF15 levels are increased in mice after bariatric surgery [25] and FGF19 levels are higher in patients with PBH [53]. FGF15 neurons in the dorsomedial hypothalamus are glucose-sensitive, and dynamically regulate glucagon secretion [54], as well as sympathetic output to the liver controlling hepatic glucose production [55]. Intestinally-derived FGF15 is dispensable for normal glucose homeostasis [56], and is protective against muscle loss after VSG in mice [25], but the role of FGF15/19 in the brain after bariatric surgery will require further investigation. Identifying dysregulated glucose-sensing brain regions after bariatric surgery might identify new treatment targets for PBH. Several lines of evidence point to glucagon as a potential culprit in the manifestation of PBH, as well as being a therapeutic target. In human and rodent studies, postprandial glucagon levels are elevated after bariatric surgery [9,14], and glucagon levels are lower during hypoglycemia [16,17]. Glucagon or glucagon analogues have shown therapeutic promise for combating acute hypoglycemia in patients with PBH [21], and we found that supraphysiological doses of glucagon safely mobilized glucose for VSG mice, despite also stimulating insulin secretion (Figure 2D–F). Interestingly, glucagon receptor agonists have also been shown to enhance hepatic insulin sensitivity [57], so a similar effect may occur in response to hyperglucagonemia after VSG. High doses or stronger glucagon receptor agonists may be necessary to overcome the decreased capacity for hepatic glucose production, blunted counterregulation, and increased insulin sensitivity, which we show in the mouse VSG model. GLP-1R antagonists, such as avexitide (exendin 9-39) may be effective for reducing hypoglycemia in individuals with PBH, and Phase III trials are in progress [58,59]. This effect could be mediated not only by reducing GLP-1 stimulated insulin secretion but also in part by shifting glucagon signaling from the β-cell to the liver, thus promoting hepatic glucose production. However, if individuals with PBH indeed have diminished capacity for hepatic glucose production, dual molecules combining GLP-1R antagonists and glucagon agonists might be more effective in preventing hypoglycemia. Conclusions In summary, we found that elevations in postprandial glucagon contribute to postprandial hypoglycemia in mice after VSG. We speculate that increased insulin sensitivity, impaired counterregulation, and reduced capacity for hepatic glucose production, collectively contribute to the hypoglycemic effect of glucagon in the postprandial state. Hyper-secretion of glucagon is not likely caused by intrinsic alterations in the islet but is more likely due to altered nutrient signaling in the gut and changes along the gut-liver-pancreas axis, which warrants further investigation. Further investigation of the involvement of the gut–brain axis in the blunting of counterregulation after bariatric surgery is also warranted. Therapeutic strategies to bias the hepatic signaling of glucagon may be effective in preventing PBH. CRediT authorship contribution statement Tyler M. Cook: Writing – review & editing, Writing – original draft, Visualization, Supervision, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Maigen Bethea: Writing – review & editing, Writing – original draft, Investigation. Silvania da Silva Teixeira: Writing – review & editing, Writing – original draft, Resources, Investigation. Carissa Birznieks: Writing – review & editing, Writing – original draft, Investigation. Andrew McClennan: Writing – review & editing, Writing – original draft, Investigation. Jasmine Hendrix: Writing – review & editing, Investigation. Richard K.P. Benninger: Writing – review & editing, Supervision. Mary-Elizabeth Patti: Writing – review & editing, Writing – original draft, Supervision. Darleen A. Sandoval: Writing – review & editing, Writing – original draft, Supervision, Resources, Methodology, Funding acquisition, Conceptualization. Funding This work was funded or supported by grants to Dr. Cook (5T32DK007658, F32DK139710, L70DK144920, P30DK048250), Dr. Benninger (R01DK102950, R01DK106412), Dr. Patti (R01DK137518, P30DK036836), and Dr. Sandoval (K26DK138368, R01DK121995). CU-AMC Diabetes Research Core (P30DK116073). Dr. Patti is a site investigator for Amylyx Pharmaceuticals Inc. and Recordati Rare Diseases, for which her institution receives funding to conduct studies that are unrelated to the present work. Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Mary Elizabeth Patti reports a relationship with Amylyx Pharmaceuticals Inc that includes: funding grants. Mary-Elizabeth Patti reports a relationship with Recordati Rare Diseases that includes: funding grants. Mary Elizabeth Patti reports a relationship with Fractyl Health, Inc. that includes: consulting or advisory. Mary Elizabeth Patti reports a relationship with MBX Biosciences, Inc. that includes: consulting or advisory. Mary Elizabeth Patti reports a relationship with Spruce Biosciences, Inc. that includes: consulting or advisory. Mary Elizabeth Patti reports a relationship with Premier Physician Network that includes: consulting or advisory. Mary Elizabeth Patti reports a relationship with Cello Health Consulting that includes: consulting or advisory. Mary Elizabeth Patti reports a relationship with Alpha Sight that includes: consulting or advisory. Mary Elizabeth Patti reports a relationship with Boxer that includes: consulting or advisory. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.