Integration of metabolic flux with hepatic glucagon signaling and gene expression profiles in the conscious dog

Abstract

Glucagon rapidly and profoundly stimulates hepatic glucose production (HGP), but for reasons that are unclear, this effect normally wanes after a few hours, despite sustained plasma glucagon levels. This study characterized the time course of glucagon-mediated molecular events and their relevance to metabolic flux in the livers of conscious dogs. Glucagon was either infused into the hepato-portal vein at a sixfold basal rate in the presence of somatostatin and basal insulin, or it was maintained at a basal level in control studies. In one control group, glucose remained at basal, whereas in the other, glucose was infused to match the hyperglycemia that occurred in the hyperglucagonemic group. Elevated glucagon caused a rapid (30 min) and largely sustained increase in hepatic cAMP over 4 h, a continued elevation in glucose-6-phosphate (G6P), and activation and deactivation of glycogen phosphorylase and synthase activities, respectively. Net hepatic glycogenolysis increased rapidly, peaking at 15 min due to activation of the cAMP/PKA pathway, then slowly returned to baseline over the next 3 h in line with allosteric inhibition by glucose and G6P. Glucagon’s stimulatory effect on HGP was sustained relative to the hyperglycemic control group due to continued PKA activation. Hepatic gluconeogenic flux did not increase due to the lack of glucagon’s effect on substrate supply to the liver. Global gene expression profiling highlighted glucagon-regulated activation of genes involved in cellular respiration, metabolic processes, and signaling, as well as downregulation of genes involved in extracellular matrix assembly and development.

NEW & NOTEWORTHY Glucagon rapidly stimulates hepatic glucose production, but these effects are transient. This study links the molecular and metabolic flux changes that occur in the liver over time in response to a rise in glucagon, demonstrating the strength of the dog as a translational model to couple findings in small animals and humans. In addition, this study clarifies why the rapid effects of glucagon on liver glycogen metabolism are not sustained.

INTRODUCTION

Stimulation of hepatic glucose production (HGP) by glucagon is critical to the regulation of glucose homeostasis during fasting (1), exercise (2), trauma and infection (3), and hypoglycemia (4–6). The physiological effects of acute hyperglucagonemia in healthy animals and humans have been well-characterized. Glucagon stimulates glycogenolysis and causes a rapid and profound increase in HGP; an effect that characteristically wanes with time. Glucagon also plays a role in the pathogenesis of diabetes, increasing fasting glucose production and impairing hepatic glucose uptake (7–10).

Early studies, primarily utilizing isolated liver slices and cell preparations, characterized glucagon’s effects on cAMP and enzymes in the glycogenolytic, gluconeogenic, and glycolytic pathways. More recent rodent studies have expanded our understanding of the molecular effects of hepatic glucagon signaling, describing the regulation of genes and proteins involved in glucose, lipid, and protein metabolism. Given the difficulty in carrying out transhepatic metabolic flux studies in rodents, however, it has remained unclear how glucagon-mediated effects on cellular signaling and hepatic gene expression correlate temporally with changes in metabolic flux. Moreover, the mechanisms underlying the waning of glucagon’s stimulatory effect on HGP are unclear.

Given these uncertainties, and the fact that glucagon is a potential target for the treatment of diabetes (8, 11, 12), an examination of glucagon’s molecular regulation of metabolic flux in a large animal model is needed. Our primary aim was to characterize the time course of relevant glucagon-mediated cellular events during acute physiological hyperglucagonemia and to correlate those changes with hepatic transcriptomic signatures and alterations in metabolic flux in the conscious dog.

METHODS AND MATERIALS

Animal Care and Surgical Procedures

Dogs of either sex were housed and fed as described previously (13, 14) and studied after an 18 h fast. The surgical facility met the standards published by the American Association for the Accreditation of Laboratory Animal Care, and the protocols were approved by the Vanderbilt University Medical Center Animal Care Committee. Two weeks before experimentation, all dogs underwent a laparotomy to implant sampling catheters into the femoral artery, the hepatic portal vein, and the hepatic vein, as well as portal vein infusion catheters in the splenic and jejunal veins, and ultrasonic flow probes (Transonic Systems, Ithaca, NY) around the hepatic artery and the portal vein, as described previously (13, 14). All dogs were healthy (leukocyte count < 18,000/mm³, hematocrit > 35%, good appetite, and normal stools).

Experimental Design

Each study consisted of an equilibration (−150 to −30 min), a basal (−30 to 0 min), and an experimental period (30 min or 4 h) (Fig. 1). At −150 min, [3-³H]glucose (priming dose of 35 µCi followed by 0.35 µCi/min) was administered intravenously (iv), and somatostatin was infused (0.8 µg/kg/min; Bachem, Torrance, CA) to inhibit the endocrine pancreas. Glucagon (0.57 ng/kg/min; Lilly, Indianapolis, IN) and insulin (as required to maintain euglycemia; Lilly) were infused intraportally at basal rates. The basal insulin infusion rate established in the equilibration period was maintained throughout the experimental period in each animal. Experiments were terminated at 4 h in the basal glucagon/basal glucose control group (CTR; n = 6 animals) and at either 30 min or 4 h in both the basal glucagon/hyperglycemic (GLU; n = 8) and hyperglucagonemic/hyperglycemic (GGN + GLU; n = 8) groups. In the CTR and GLU groups, basal intraportal glucagon infusion was maintained throughout the experimental period, whereas in the GGN + GLU group, the rate was increased sixfold (to 3.42 ng/kg/min). In the GLU group, intravenous glucose was infused to match the arterial plasma glucose level observed in the GGN + GLU group. Since CTR animals remained at a basal steady state, the 4-h liver biopsy was assumed to provide reasonable control data for both the 30-min and 4-h time points. Immediately after obtaining the final blood sample, each animal was anesthetized with pentobarbital, and a laparotomy was performed. The hormone and glucose infusions were continued while liver sections were rapidly freeze-clamped in situ and subsequently stored at −80°C, as described previously.

Figure 1.

Study design. Glucose tracer, somatostatin, and basal intraportal insulin and glucagon infusions were started at −150 min. After a basal sampling period (−30 to 0 min), glucagon was either infused at a basal replacement rate (CTR group; n = 6 animals) or it was increased sixfold (GGN + GLU; n = 8) during the experimental period (0–240 min). In another group (GLU; n = 8), glucagon was maintained at basal, whereas glucose was infused to match the hyperglycemia that occurred in response to glucagon in the GGN + GLU group. Studies were terminated at either 30 or 240 min, and liver biopsies were taken for molecular analyses. RNA-Seq was performed using samples taken at 240 min. CTR, control (basal glucagon/basal glucose); GLU, basal glucagon/hyperglycemic; GGN + GLU, hyperglucagonemic/hyperglycemic; RNA-Seq, RNA sequencing.

Metabolite Analysis

Hematocrit levels, glucose, insulin, and nonesterified free fatty acid (NEFA) levels in plasma, and alanine, lactate, and glycerol concentrations in blood were determined using standard procedures as previously described (13). Plasma glucagon (GL-32K, Millipore-Sigma, Burlington, MA) was measured by radioimmunoassay. The Vanderbilt University Medical Center Analytical Services Core previously found that 54 ± 3% of what was measured by the GL-32K assay under baseline fasting conditions was nonspecific cross-reacting material rather than glucagon. In their analyses, glucagon was measured in plasma collected from five dogs before (37 ± 3 pg/mL; n = 75 replicates over time) and during somatostatin infusion (20 ± 1 pg/mL; i.e., the nonspecific contribution; n = 75). Complete suppression of endogenous glucagon secretion by somatostatin was verified by a total loss of the portal vein to arterial glucagon gradient (which was 11 ± 2 pg/mL before somatostatin infusion compared with 0 ± 1 pg/mL during it). Therefore, true glucagon levels were estimated by subtracting half of each dog’s basal glucagon level (i.e., the nonspecific component of the measurement) from measured values.

Liver Tissue Analyses

Representative pieces of the three largest liver lobes were combined for analysis, and SDS-PAGE and Western blotting were performed using standard methods (13). Fructose-2,6-bisphosphate (F2,6P₂) concentration, glucose-6-phosphate (G6P) concentration, liver glycogen levels, and enzyme activities of glycogen synthase (GS) and glycogen phosphorylase (GP) were assessed using established methods (15–19). cAMP was measured using the DetectX Cyclic AMP Chemiluminescent Direct Immunoassay kit (Arbor Assays, Ann Arbor, MI).

Bulk RNA Library Preparation and Sequencing

Bulk RNA sequencing (RNA-Seq) was performed as described previously (20). Briefly, nucleic acid extraction and purification were performed on nine canine liver biopsies collected under steady-state conditions at the end of the 4-h experimental period (n = 3 animals each from CTR, GLU, and GGN + GLU groups). RNA concentration and integrity (RIN > 7 for all samples) were assessed via Qubit fluorometric quantitation and tapestation bioanalyzer (Agilent), respectively. Libraries were prepared using the Takara SMART-Seq v4 Ultra-low input RNA kit and analyzed on a NovaSeq platform (Illumina) using paired-end (PE100/150) reads with 40 M total reads per sample. STAR aligner (v2.7.3a) was used for read alignment. Raw read counts were assessed using HTSeq (v0.11.2) and normalized using DESeq2. Aligned reads were used for estimating the expression of the genes using cufflinks (v2.2.1). Differentially expressed genes (DEGs) were determined using DESeq2 (R Bioconductor package, v. 3.11) on the basis of fold change (cutoff ≥ ±1.0) and adjusted P value (<0.1) as determined by the Benjamini–Hochberg false discovery rate method (21). To delineate glucagon-regulated pathways and biological processes, significantly enriched groups of coordinately regulated genes were identified by Gene Set Enrichment Analysis (GSEA) using Hallmark genesets (22) and gene ontology (GO) terms from the Molecular Signatures Database (23).

Calculations

Net hepatic glucose balance was assessed using the arteriovenous difference technique as described previously (13, 14, 24). Gluconeogenic flux was determined by taking the sum of net hepatic uptake rates of gluconeogenic precursors and dividing by two to account for three carbon precursor incorporation into one six-carbon glucose molecule (13, 14, 24). Glycolytic flux was estimated by taking the sum of net hepatic output rates (when present) of those three carbon substrates (converted to glucose equivalents) and hepatic glucose oxidation (assumed to be 0.2 ± 0.1 mg/kg/min as previously validated) (25, 26). Glucose turnover, used to determine glucose production, was determined using the GLUTRAN circulatory model of Mari et al (27).

Statistical Analysis

Statistical comparisons were carried out using one-way ANOVA (Prism10) or two-way repeated-measures ANOVA (group × time) (SigmaStat) with appropriate post hoc tests when significant F ratios were obtained. Significance was determined as P value < 0.05.

RESULTS

Hormone Levels

Insulin was maintained at a basal level in all groups (Fig. 2, A and B). Arterial and hepatic portal vein glucagon levels were clamped at basal values in the CTR and GLU groups during the experimental period, or they were increased to approximately 100 and 250 pg/mL, respectively, in the GGN + GLU group (Fig. 2, C and D). Consequently, there was about a sixfold rise in glucagon in both vessels, in keeping with the increase in portal vein glucagon infusion rate.

Figure 2.

Hormone levels. Arterial plasma insulin (A), hepatic portal vein plasma insulin (B), arterial plasma glucagon (C), and hepatic portal vein plasma glucagon concentrations (D) in 18-h fasted conscious dogs during the basal (−30 to 0 min) and experimental (0–240 min) periods. Data are represented as means ± SE, in the CTR (n = 6 animals from −30 to 240 min), GLU (n = 8 from −30 to 30 min; n = 4 from 30 to 240 min), and GGN + GLU (n = 8 from −30 to 30 min; n = 4 from 30 to 240 min) groups. Arterial and portal vein plasma glucagon levels were greater (P < 0.05) in the GGN + GLU group compared with the GLU group between 15 and 240 min. CTR, control (basal glucagon/basal glucose); GLU, basal glucagon/hyperglycemic; GGN + GLU, hyperglucagonemic/hyperglycemic.

Glucose, Glycogenolytic, and Gluconeogenic Flux Rates

Euglycemia was maintained in the CTR group, whereas hyperglycemia rapidly developed due to the effects of glucagon in the GGN + GLU group (Fig. 3A). This glycemic rise was matched in the GLU group using intravenous glucose infusion (Fig. 3B). During the basal period, the rates of net hepatic glucose balance (NHGB; Fig. 3C) and hepatic glucose production (Fig. 3D) were similar between groups. Hyperglucagonemia markedly stimulated both NHGB (fivefold) and HGP (fourfold) within 15 min (Fig. 3, C and D), after which glucose production gradually waned and returned to baseline by the fourth hour of the clamp. When glucagon-stimulated hyperglycemia was matched by glucose infusion in the GLU group, there was a rapid decline in NHGB that resulted in a switch to slight net hepatic glucose uptake by the end of the experiment (Fig. 3C). Consistent with this observation, hyperglycemia almost completely suppressed HGP (Fig. 3D). Therefore, although glucagon’s effect at 4 h was reduced by 70% from its peak, it continued to effectively stimulate glucose production relative to that in the GLU group.

Figure 3.

Glucose metabolism. Arterial plasma glucose levels (A), glucose infusion rates (B), net hepatic glucose balance (C), glucose production (D), net hepatic glycogenolytic flux (E), and hepatic gluconeogenic flux (F) in 18-h fasted conscious dogs during the basal (−30 to 0 min) and experimental (0–240 min) periods. Data are represented as means ± SE, in the CTR (n = 6 animals from −30 to 240 min), GLU (n = 8 from −30 to 30 min; n = 4 from 30 to 240 min), and GGN + GLU (n = 8 from −30 to 30 min; n = 4 from 30 to 240 min) groups. Arterial plasma glucose levels differed (P < 0.05) in the GGN + GLU and GLU groups compared with the CTR group between 15 and 240 min. The glucose infusion rate differed (P < 0.05) in the GLU group compared with the other groups between 15 and 240 min. Net hepatic glucose balance, glucose production, and net hepatic glycogenolysis were greater (P < 0.05) in the GGN + GLU group compared with the GLU group between 15 and 240 min. CTR, control (basal glucagon/basal glucose); GLU, basal glucagon/hyperglycemic; GGN + GLU, hyperglucagonemic/hyperglycemic.

The rapid effects of glucagon (stimulatory) and glucose (suppressive) on liver glucose flux were both attributable to changes in net hepatic glycogenolytic flux (Fig. 3E). At 15 min, it peaked at 10.4 ± 1.4 mg/kg/min in the GGN + GLU group, then fell to 8.3 ± 0.6 mg/kg/min by 30 min. Thus, the effect of glucagon had already begun to wane. In the GLU group, net hepatic glycogenolysis was 0.2 ± 0.5 and −0.5 ± 0.7 mg/kg/min at the two time points, respectively. By contrast, gluconeogenic flux was not significantly altered by either treatment over time (Fig. 3F). The fact that HGP was reduced by hyperglycemia in the GLU group, whereas hepatic gluconeogenic flux to G6P was not, indicates that the gluconeogenic carbon was being used to form liver glycogen rather than plasma glucose. This is in line with the switch toward net hepatic glycogen synthesis seen in the GLU group (Fig. 3E).

Lactate, Fat, and Amino Acid Metabolism

Arterial blood lactate levels were not altered over time in the CTR group, but they tended to increase in the GLU and GGN + GLU groups (Fig. 4A). Net hepatic lactate output, one indicator of hepatic glycolysis, remained at a basal rate throughout in CTR animals, but increased after a delay of 30 min in GLU, and then remained elevated (Fig. 4B). Glucagon, on the other hand, caused a rapid (within 15 min) stimulation of net hepatic lactate output, followed by a steady decline such that by the end of the study, lactate output was below baseline. Thus, glucagon caused an initial surge in glycolytic flux, but after 1 h it limited the stimulatory effect of hyperglycemia on this pathway. Arterial levels and net hepatic balances of NEFA and glycerol did not differ between groups, nor did they change over time, suggesting a lack of effect of glucagon or glucose on lipolysis or NEFA reesterification under these conditions (Supplemental Fig. S1). In the CTR and GLU groups, there were no significant changes in arterial blood levels, net hepatic uptake, or net hepatic fractional extraction of alanine (a representative gluconeogenic amino acid) (Supplemental Fig. S2). On the other hand, hyperglucagonemia increased hepatic alanine fractional extraction and net hepatic alanine uptake which caused a decrease in blood alanine levels.

Figure 4.

Lactate metabolism. Arterial blood lactate (A) and net hepatic lactate output (B) in 18-h fasted conscious dogs during the basal (−30 to 0 min) and experimental (0–240 min) periods. Data are represented as means ± SE, in the CTR (n = 6 animals from −30 to 240 min), GLU (n = 8 from −30 to 30 min; n = 4 from 30 to 240 min), and GGN + GLU (n = 8 from −30 to 30 min; n = 4 from 30 to 240 min) groups. Net hepatic lactate output was less (P < 0.05) in the GGN + GLU group compared with the GLU group between 180 and 240 min, and it was greater in the GLU group than the CTR group throughout the study. CTR, control (basal glucagon/basal glucose); GLU, basal glucagon/hyperglycemic; GGN + GLU, hyperglucagonemic/hyperglycemic.

Cellular Effects of a Physiological Increase in Glucagon

Insulin signaling.

In the presence of basal insulin, elevations in glucagon or glucose had little effect on Akt phosphorylation. Glucagon abrogated glucose-stimulated increases in GSK-3β and FOXO1 phosphorylation by 4 h (Fig. 5, A–C).

Figure 5.

Phosphorylation state of hepatic insulin signaling (A–C) and AMPK signaling proteins (D–F). Hepatic Akt (p/t) (A), GSK3β (p/t) (B), FOXO1 (p/t) (C), AMPK (Thr172 p/t) (D), AMPK (Ser485 p/t) (E), and ACC (p/t) (F). See Supplemental Table S1 for representative Western blots. Data are represented as means ± SE in the Control (n = 4 or 5 animals), Glucose (n = 4 or 5), and Glucagon (n = 4 or 5) groups for each time point. ^aP < 0.05 GLU or GGN + GLU vs. CTR, ^bP < 0.05 GLU vs. GGN + GLU within the same time point (30 min or 4 h), ^cP < 0.05 GLU vs. GLU or GGN + GLU vs. GGN + GLU between time points (30 min vs. 4 h). CTR, control (basal glucagon/basal glucose); GLU, basal glucagon/hyperglycemic; GGN + GLU, hyperglucagonemic/hyperglycemic.

AMPK signaling.

Neither hyperglycemia nor hyperglucagonemia had consistent effects on the phosphorylation of AMPK at Thr172 or Ser485, or on ACC phosphorylation (an index of active AMPK) (Fig. 5, D–F).

Regulators of glycogen metabolism.

Although the hepatic cAMP level was not altered by hyperglycemia, it was markedly elevated in the GGN + GLU group at 30 min (sevenfold) and remained elevated at 4 h, albeit at a somewhat lower level (fourfold basal; Fig. 6A). cAMP ultimately determines the extent of phosphorylation of glycogen phosphorylase (GP) and glycogen synthase (GS), thus, as expected, GP activity was markedly elevated at both 30 min and 4 h in the GGN + GLU group (Fig. 6B). Although GP activity waned modestly over time (∼25% lower at 4 h than 30 min), this also tended to occur in the GLU group (Fig. 6B), suggesting that the small decrease in activity may have been glucose-induced and that glucagon’s effect on phosphorylase activity was for the most part sustained. GS activity was not altered by hyperglycemia but decreased progressively in response to hyperglucagonemia (Fig. 6C). The ratio of GP to GS activities (an index of the net change in glycogen enzyme activity) was 12 ± 2 in the CTR group. In the GLU group, this ratio was 13 ± 5 at 30 min and 8 ± 1 at 4 h. The ratio was 40 ± 2 and 52 ± 9, respectively, in the GGN + GLU group. Thus, in response to glucagon, the activity of the glycogen metabolizing enzymes shifted toward breakdown, and there is no evidence that a change in the phosphorylation of these enzymes explains why glucose production waned over time. It should be noted, however, that liver GS is hyperphosphorylated, and a small decrease in measured (in vitro) hepatic GS activity would not necessarily have a major effect on actual GS flux (the assay measures phosphorylation-dependent rather than allosterically regulated GS activity).

Figure 6.

Hepatic cAMP and regulators of hepatic glycogen metabolism and glycolysis. Hepatic cAMP (A), glycogen phosphorylase (GP) activity ratio (B), glycogen synthase (GS) activity ratio (C), G6P (D), F2,6P₂ (E), and glycogen (F). Data are represented as means ± SE in the CTR (n = 3 or 4 animals), GLU (n = 3 or 4), and GGN + GLU (n = 4) groups for each time point. ^aP < 0.05 GLU or GGN + GLU vs. CTR, ^bP < 0.05 GLU vs. GGN + GLU within the same time point (30 min or 4 h), ^cP < 0.05 GLU vs. GLU or GGN + GLU vs. GGN + GLU between time points (30 min vs. 4 h). CTR, control (basal glucagon/basal glucose); GLU, basal glucagon/hyperglycemic; GGN + GLU, hyperglucagonemic/hyperglycemic.

Hepatic G6P, F2,6P₂, and glycogen.

Although hyperglycemia tended to reduce hepatic G6P content, hyperglucagonemia elicited a rapid (within 30 min), sustained (up to 4 h), and marked increase in G6P (Fig. 6D). Hepatic GS is primarily regulated by the allosteric effects of G6P, thus it is likely that GS flux was enhanced in the GGN + GLU group (despite decreased in vitro GS activity; Fig. 6C), at least in some hepatocytes (28), although this was clearly offset by a greater increase in GP flux during the early response to glucagon (Fig. 3E). Later there was a shift in net GS/GP balance, which was less favorable to net degradation, likely due to prolonged and sustained elevations of G6P and glucose. Hepatic levels of F2,6P₂, an allosteric activator of glycolysis and inhibitor of gluconeogenic flux, were markedly (threefold) increased by hyperglycemia at 4 h, whereas glucagon limited this glucose-induced rise by 60% (Fig. 6E). Consistent with the cessation of net hepatic glycogen breakdown in the GLU group, glycogen levels increased progressively over time relative to the other groups (Fig. 6F). As expected, given the high rate of net hepatic glycogenolysis in the GGN + GLU group, glucagon reduced the liver glycogen content (by more than 50% compared with the GLU group at 4 h).

Glucagon-regulated hepatic transcriptional signatures identified by bulk RNA-Seq.

To determine the hepatic transcriptional profiles uniquely associated with a physiological increase in glucagon (GGN + GLU) compared with matched hyperglycemia (GLU) and baseline controls (CTR), we performed bulk RNA-Seq of liver biopsies obtained from the GGN + GLU, GLU, and CTR groups at the end of the 4-h experimental periods. Comparative analysis of gene expression profiles in the GGN + GLU versus GLU group revealed a total of 559 differentially expressed protein-coding genes (DEGs) of which 318 were upregulated by hyperglucagonemia and 241 were downregulated (Fig. 7A). As anticipated, genes associated with gluconeogenesis (G6PC and PPARGC1A), amino acid transport and metabolism (SLC1A2, SLC7A2, OAT, and TAT), and cAMP signaling (ADCY1 and PDE4B) were among the most highly upregulated genes in the GGN + GLU group. Interestingly, the gene encoding follistatin (i.e., FST), glucagon- and FOXO1-regulated hepatokine that promotes insulin resistance, white adipose tissue lipolysis, and hyperglycemia in rodents (29, 30), was likewise one of the most significantly induced genes by hyperglucagonemia. On the other hand, genes associated with cellular responses to carbohydrate stimuli (PRKCE, KCNB1, EGR1, and FGF21) as well as the matrisome (i.e., extracellular matrix [ECM] related, including THBS1, MMP11, COL3A1, COL6A1, and INHBA) were among the most significantly downregulated genes in the GGN + GLU group. Of note, the waning of net hepatic glycogenolytic flux in the GGN + GLU group correlated with marked downregulation of MGAM2, a gene that encodes maltase-glucoamylase 2, an enzyme predicted to be involved in glycogen breakdown through its α-1,4-glucosidase activity (Fig. 7A).

Figure 7.

Glucagon-regulated hepatic transcriptional profiles identified by bulk RNA-Seq. A: volcano plot illustrating the log₂ fold change (FC; x-axis) and adjusted (adj.) P values (y-axis) of all differentially expressed genes (DEGs) in GLU vs. GGN + GLU at the end of the 4-h experimental period. A total of 559 protein-coding genes showed significant differential expression (i.e., −log₁₀ adj. P value ≥ 1.0 and log₂FC≥±1.0, as denoted by gray lines) in GLU vs. GGN+GLU: 318 upregulated (red circles) and 241 downregulated (green circles). B: the top 10 significantly (FDR q value < 0.1) enriched Hallmark gene sets in GLU vs. GGN + GLU as determined by Gene Set Enrichment Analysis (GSEA), with the x-axis displaying normalized enrichment scores (NES) for positively (red bars; upregulated genes in the GGN + GLU group are enriched in these gene sets) and negatively (green bars; downregulated genes in the GGN + GLU group are enriched in these gene sets). Heat map representation of the top 20 leading-edge transcripts underlying the (C) positive enrichment signals for oxidative phosphorylation (left) and MYC targets (right) or the (D) negative enrichment signals for myogenesis (left) and epithelial mesenchymal transition (right) as identified in B. Each row represents a different glucagon-regulated gene, whereas each column represents the mean log₂ normalized expression value for these genes in the CTR, GLU, and GGN + GLU groups (n = 3 animals each). Expression values are represented as colors and range from green (lowest expression) to red (highest expression). CTR, control (basal glucagon/basal glucose); GLU, basal glucagon/hyperglycemic; GGN + GLU, hyperglucagonemic/hyperglycemic; RNA-Seq, RNA sequencing.

Gene set enrichment analysis.

To identify pathways and biological processes significantly enriched in GGN + GLU versus GLU groups, we performed gene set enrichment analysis (GSEA) using Hallmark gene sets and GO terms from the MSigDB (22, 23). A total of 37 (out of 50) Hallmark gene sets were significantly enriched (FDR < 0.1) in the GGN + GLU group, with 24 positively and 13 negatively enriched (Fig. 7B). The most positively enriched glucagon-regulated pathway highlighted widespread induction of genes involved in hepatic mitochondrial oxidative phosphorylation (OXPHOS) and contained significant enrichment of related functional GO terms such as mitochondrial respiratory chain complex assembly, ATP synthesis coupled electron transport, NADH dehydrogenase complex assembly, mitochondrial electron transport chain NADH to ubiquinone, mitochondrial translation, and ribosome biogenesis. Leading-edge genes underlying the glucagon-regulated OXPHOS signal included those encoding mitochondrial respiratory chain complex proteins and ATP synthesis/transport proteins, such as CYCS, COX11, NDUFB5, NDUFS4, NDUFA8, NDUFB3, ATP5F1C, ATP6V0E1, ATP5PB, ATP5PF, and SLC25A5 (Fig. 7C). Genes annotated to the MYC Targets gene set were also significantly enriched among the upregulated genes with hyperglucagonemia, suggestive of heightened MYC transcription factor activation in the GGN + GLU versus GLU group (Fig. 7, B and C). Indeed, elevated cAMP-PKA signaling was shown to augment MYC protein expression in vitro (31), and MYC directly regulates genes that promote mitochondrial biogenesis and oxidative metabolism (32). In support of this, GO terms associated with leading edge genes underlying positive enrichment of the MYC Targets gene set included ribonucleoprotein complex biogenesis, translational initiation, peptide biosynthetic processes, spliceosomal complex assembly, and RNA binding. These findings suggest that MYC may be a molecular effector of glucagon-regulated mitochondrial OXPHOS gene expression in the liver. Protein secretion, mTORC1 signaling, and fatty acid metabolism were among the other positively enriched pathways in the GGN + GLU versus GLU group (Fig. 7B and Supplemental Fig. S3).

By contrast, genes annotated to pathways more broadly associated with regulation of ECM organization and interactions, including myogenesis, epithelial-mesenchymal transition (EMT), and apical junction, were significantly enriched among the downregulated genes with hyperglucagonemia (e.g., COL3A1, COL4A2, COL6A3, COL1A2, THBS1, MMP2, TNC, FMOD, INHBA) (Fig. 7, B and D). Likewise, GO terms related to collagen fibril organization, external encapsulating structure organization, ECM assembly, semaphoring plexin signaling pathway, and aorta development and morphogenesis were overrepresented among the downregulated genes with hyperglucagonemia. Pathological elevation in the expression of leading-edge genes underlying the signals in myogenesis and EMT, for example, has been shown to occur in response to liver injury and/or obesity and precede the development of hepatic fibrosis and insulin resistance (33, 34). In addition, T2D in humans is associated with hepatic and extrahepatic ECM remodeling (33, 34). Our data suggest that in healthy dogs, acute (i.e., 4 h) hyperglucagonemia limits hepatic ECM proliferation and/or remodeling via transcriptional downregulation of genes involved in these processes. Whether these changes impact the acute stimulatory effect of glucagon on HGP and/or amino acid metabolism remains to be determined.

DISCUSSION

A physiological sixfold rise in glucagon rapidly stimulated glycogen breakdown, HGP, and caused hyperglycemia. Glucagon’s effect on net hepatic glycogenolysis was transient, however. It peaked within 15 min, then returned to basal by 3 h (although glucagon continued to stimulate glucose production since matching hyperglycemia suppressed HGP below basal). The stimulation and waning of glucose production in response to glucagon were entirely the result of alterations in glycogen flux, with no detectable modification of hepatic gluconeogenic flux to G6P. The purpose of this study was to identify cellular and molecular correlates for the liver’s metabolic response to glucagon.

Although glucagon can stimulate HGP in a glucose-independent manner (35–37), hyperglycemia has its own independent inhibitory effects on HGP. Thus, we included a control group in which hyperglycemia was matched to isolate the unique metabolic and molecular effects of glucagon. Hyperglycemia per se, in the presence of basal glucagon and insulin, completely inhibited net hepatic glucose balance within 15 min, without initially affecting HGP. This indicates that there was an increase in HGU and glucose cycling. This was likely due to glucose-induced translocation of glucokinase to the cytoplasm (38–41), which rapidly enhances HGU. Net hepatic glycogenolysis was suppressed to near zero at this time, there was no change in glycolytic flux, and gluconeogenic flux to G6P persisted, indicating that gluconeogenesis was almost completely responsible for hepatic glucose release. By 30 min (Fig. 8A), the liver had switched to low rates of net glucose uptake and net glycogen synthesis, and liver glycogen began to accumulate (+34%). HGP was still not significantly reduced (–20%) and we did not detect any significant impact of hyperglycemia on markers of insulin signaling or gluconeogenesis. Although changes in GS and GP activities in response to hyperglycemia were not apparent in our assay at this time (Fig. 6, B and C), it should be noted that the assay reflects only the covalent (phosphorylated) state of the enzymes, and therefore does not necessarily account for the impact of allosteric effectors on their overall activities (17). Indeed, in vivo changes in the activities of these enzymes appear to have gone undetected in the in vitro assay given the clear change in net glycogenolysis. After 30 min of hyperglycemia, the plasma glucose level had doubled, but G6P content was reduced to ∼50% basal. Since G6P and glucose both allosterically activate GS and inhibit GP (42–47), the effect of the rise in glucose was dominant over the fall in G6P. The decrease in G6P, despite an increase in HGU, suggests strong glucose-induced activation of GS, which “pulled” carbon into glycogen. At 60 min, hyperglycemia had suppressed HGP by 50%, and glycolytic flux began to increase (as indicated by the increase in net hepatic lactate output). By 4 h (Fig. 8B), hyperglycemia had induced a threefold increase in F2,6P₂, which promotes glycolytic flux by increasing phosphofructokinase 1 (PFK-1) activity (48, 49). Although hyperglycemia did not reduce hepatic gluconeogenic flux (conversion of noncarbohydrate precursors into G6P), gluconeogenesis per se (glucose derived from non-carbohydrate precursors that are released into the blood) was reduced since HGP was nearly completely suppressed by the end of the study. This suggests that gluconeogenic carbon was redirected from release as glucose to other fates (i.e., glycolysis and glycogen synthesis). In support of this, both liver lactate output and glycogen levels were elevated at 4 h.

Figure 8.

Model for the temporal effects of hyperglucagonemia and/or hyperglycemia on hepatic glucose flux and signaling. A: in the GLU group, after 30 min of hyperglycemia (compared with the CTR group), the allosteric effects of glucose on glycogen synthase (GS) and glycogen phosphorylase (GP) caused an increase in net hepatic glycogen synthesis. This “pulled” carbon into glycogen, lowering glucose 6-phosphate (G6P) and increasing hepatic glucose uptake (HGU). In addition, hyperglycemia induces the translocation of glucokinase to the cytoplasm (data not shown), which would also rapidly enhance HGU. B: after 4 h of hyperglycemia in the GLU group, glucose continued to allosterically activate GS and inhibit GP. As a result, net hepatic glycogen synthesis and glucose uptake remained elevated, and the glycogen level further increased. In addition, glucose increased fructose-2,6-bisphosphate (F2,6P₂) levels threefold, which would stimulate phosphofuctokinase-1 (PFK-1) and glycolytic flux. Diversion of carbon into glycolysis caused a reduction in hepatic glucose production (HGP). C: in the GGN + GLU group the flux data demonstrate that net hepatic glycogenolysis was at its peak 15 min after the elevation of plasma glucagon (see Fig. 3E). Precipitating this rapid change, glucagon-stimulated cAMP production would have activated the protein kinase A (PKA) pathway, and this would have phosphorylated GP (stimulatory) and GS (inhibitory). During that early event, hepatic glucose and G6P levels would not yet have increased, and thus PKA’s initial effects on glycogenolysis were unopposed. But, after 30 min of hyperglucagonemia, glucose and G6P levels had now doubled in response to the increase in glycogenolysis. Although PKA-mediated stimulation of glycogenolysis was still marked at this time, glucose and G6P were now beginning to allosterically oppose PKA’s effects, and thus glycogenolysis started to fall (Fig. 3E). In this setting, we purport that the rapid and robust stimulatory effect of glucagon on cAMP/PKA-mediated signaling events preceded allosteric regulation of GS and GP by glucose and G6P. Glucose-6-phosphatase is thought to be activated by G6P, whereas glucagon inhibits the stimulatory effects of hyperglycemia on glucokinase, enhancing HGP and suppressing HGU. Not pictured are the effects of glucagon on calcium-dependent stimulation of HGP. Increased flux into the G6P pool also caused glycolytic flux to increase. D: after 4 h of hyperglucagonemia and hyperglycemia in the GGN + GLU group, the cAMP/PKA pathway remained activated, but the effects on GS and GP were now completely offset by the combined allosteric effects of glucose and G6P. As a result, net hepatic glycogenolysis returned to baseline. Suppression of net hepatic glycogenolysis below basal (as occurred in the hyperglycemic control group), however, was prevented by continued PKA pressure on GP and GS. Likewise, glucagon limited the stimulatory effect of hyperglycemia on F2,6P₂, most likely due to offsetting inhibition of PFK-2 by PKA. PKA would have also inhibited pyruvate kinase (PK), and as a result, glycolytic flux returned to baseline. Thus, our data suggest that elevations in glucose and G6P overcame the effects of PKA on glycogen metabolism but not glycolytic flux. Consequently, net hepatic glucose balance (NHGB; i.e. HGP–HGU) returned to basal. Hyperglycemia was unable to suppress NHGB below basal due to residual stimulation by glucagon. The purported dominant signaling molecules for each panel are shown in bold font. CTR, control (basal glucagon/basal glucose); GLU, basal glucagon/hyperglycemic; GGN + GLU, hyperglucagonemic/hyperglycemic.

The liver’s response to glucagon, on the other hand, was rapid (within 15 min), and led to a profound increase in net hepatic glycogenolysis that was dominant over hyperglycemia’s reverse effects (Fig. 8C). Glycogen breakdown provided substrate for increased glucose production and glycolytic flux, which were both tightly correlated with net hepatic glycogenolysis throughout the study. The glycogenolytic response was most likely caused by increased cAMP-induced activation of protein kinase-A (PKA) and its covalent effects on GP and GS activities. When glucagon initiated this event, hepatic G6P and glucose levels were baseline, and PKA’s effects were unopposed. However, the effects of glucagon on glucose production characteristically waned with time. This was due to reduced stimulation of net hepatic glycogenolysis rather than an increase in glycolytic or decrease in gluconeogenic flux.

Several reasons for glucagon’s evanescence have been suggested, including opposition by insulin, depletion of hepatic glycogen, a decline in the amount of hepatic cAMP due to adenylate cyclase inhibition or phosphodiesterase activation, desensitization of liver glucagon receptors, and CNS-mediated suppression of HGP by glucagon (50). A transient increase in cAMP has been considered a leading explanation since glucagon enhances net splanchnic plasma cAMP release with a spike-decline pattern (51, 52). On the other hand, studies performed primarily in isolated rat hepatocytes have led to the notion that cAMP cannot mediate the physiological effects of glucagon since pharmacological levels (>350 pg/mL) of the hormone were required for adenylate cyclase activation in those studies (53). The present study refutes this concept since cAMP was markedly elevated by the physiological rise in glucagon. Of note, we found that although cAMP decreased from its 30 min peak value, its levels remained elevated fourfold even after 4 h. Desensitization of the glucagon receptor or loss of downstream glucagon/cAMP/PKA signaling provide unsatisfactory explanations since the covalent effects of glucagon on GS and GP activities were largely sustained over time (there was no decline in the GP/GS ratio; Fig. 6, B and C). Furthermore, there was a sustained increase in hepatic alanine extraction over 4 h. Depletion of liver glycogen is also an unlikely cause of the loss of glucagon action since glucagon had only reduced glycogen content by ∼20% by the end of the study. Glucagon also increases HGP via increased calcium signaling, but a potential drop-off in the activity of this pathway was not assessed in this study. This did not occur in primary mouse hepatocytes, however. Instead, glucagon-stimulated calcium signaling increased over time from baseline to 15 then 45 min (54). Finally, recent rodent studies suggested that negative feedback by hypothalamic glucagon signaling can inhibit glucose production. When glucagon was administered directly into the brain in those studies, HGP was suppressed, but when activation of the hypothalamic glucagon receptor or PKA was prevented, the effect was blocked (55, 56). On the other hand, we found in dogs that although physiological engagement of the brain by glucagon altered hepatic gluconeogenic/glycogenolytic carbon flux, it was not responsible for the transient fall in HGP following the stimulation of HGP during a rise in glucagon (57).

In fact, elevations in glucose and G6P were most likely responsible for the waning of glucagon’s effect (Fig. 8D). Unlike hyperglycemia alone, cAMP-mediated activation of GP caused hepatic G6P to accumulate. At 30 min, plasma glucose and liver G6P levels had by then doubled, and although net hepatic glycogenolysis was still robust, a reduction in glycogenolysis was already becoming apparent (down 20% from 15 to 30 min). Glucose and G6P synergistically inhibit glycogenolysis via allosteric inhibition of the active form of phosphorylase (42–45). In addition, they have been shown to enhance GS activity (44–47). Indeed, over time, net hepatic glycogenolysis returned to baseline. Thus, it appears that glucose and G6P exerted time-dependent effects that eventually partially overrode the covalent phosphorylation of GP and GS. Glucose and G6P have no effect on hepatic amino acid uptake, however, so it is understandable that there was no time-dependent loss of glucagon’s effect on that process. Indeed, hepatic alanine fractional extraction increased steadily throughout the study.

An initial consequence of a rise in G6P would have been an increase in glucose-6-phosphatase activity, which would allow glucagon-stimulated HGP to increase (58). On the other hand, hyperglycemic activation of glucokinase (38–41) would be offsetting. Thus, although there was an initial increase in HGP (Fig. 8C), net hepatic glucose balance returned to baseline over time (Fig. 8D).

Glycolysis (i.e., hepatic lactate release) increased maximally during the first hour of hyperglucagonemia (Fig. 8C). This occurred without an increase in F2,6P₂, indicating that the “push” from glycogenolysis was sufficient to drive flux through the glycolytic pathway. This suggests that substrate mass was a more important regulatory factor than enzyme activity at this stage of the response. PKA opposes glucose-induced F2,6P₂ production (48, 59), and correspondingly, after an hour glycolytic flux began to fall. By 4 h, the increase in F2,6P₂ that occurred in response to hyperglycemia was diminished. In addition, PKA would have inhibited pyruvate kinase, a rate-controlling glycolytic enzyme (48). Thus, by the end of the study, glycolysis had returned to baseline, despite hyperglycemia (Fig. 8D).

It should be noted that glucagon’s effects on these pathways were not entirely lost by the end of the study—glucagon retained its ability to oppose hyperglycemic effects on glycogen, glucose, and glycolytic pathways. Thus, the data demonstrate the liver’s remarkable ability to restore homeostasis in the face of hyperglucagonemia and accompanying hyperglycemia.

Neither hyperglycemia nor hyperglucagonemia modified acute (<4 h) hepatic gluconeogenic flux. This process is highly dependent on substrate supply and NEFA availability (oxidized to support gluconeogenic flux) (60), but the glucagon receptor is not expressed in muscle, is minimally present in adipose tissue, and glucagon is not known to increase gluconeogenic precursor flux to the liver (61–63). Indeed, in the present study, elevated glucagon did not bring about net hepatic lactate uptake at any point, nor did it augment the delivery of glycerol, alanine, or NEFA to the liver. Although glucagon stimulated hepatic amino acid transport by enhancing amino acid fractional extraction, the effect was offset by a decrease in circulating amino acid levels. Thus, net hepatic amino acid uptake was maintained at a near-basal rate.

Several studies have indicated that elevations in glucagon during settings such as fasting, exercise, and hypoglycemia can enhance AMPK Thr172 phosphorylation and activity (64–67). Conversely, in the presence of hyperglycemia, hyperglucagonemia was only able to increase P-Thr172 AMPK at 30 min in the present study, indicating that some of glucagon’s effects are context-dependent. It has been suggested that glucagon’s effects on AMPK are mediated by cellular ATP depletion caused by increased gluconeogenesis (68) but gluconeogenesis was not altered, which may explain why glucagon had no effect on AMPK. It is worth noting that hyperglycemia (regardless of the glucagon concentration) increased AMPK Ser485 phosphorylation. We previously observed that this site was phosphorylated during hyperinsulinemia and was correlated with decreased Thr172 phosphorylation (67). Thus, phosphorylation at Ser485 may modify glucagon’s ability to phosphorylate AMPK’s Thr172 site.

Glucagon increases both during protein consumption and in response to stressors such as hypoglycemia and injury, conditions during which other hormones also change (e.g., insulin, catecholamines, cortisol, etc.) (69). Like glucagon, catecholamines and cortisol activate pathways that promote HGP, although not necessarily through the same mechanisms, whereas insulin inhibits HGP. The present study investigated the impact of a rise in glucagon alone to avoid the confounding effects that would be caused by changes in multiple hormones. For example, insulin reduces glucagon-stimulated increases in hepatic cAMP, and it also inhibits glycogenolysis via cAMP-independent mechanisms (70). It should be noted that hyperglycemia is not required for the waning of glucagon’s effect on HGP. In humans, although the effects of glucagon were more persistent when changes in plasma glucose were prevented by hyperinsulinemia, glucagon-stimulated HGP was still transient at basal glucose levels (71). Likewise, glucagon-stimulated HGP was transient during insulin-induced hypoglycemia in adrenalectomized dogs (no rise in epinephrine) (72). On the other hand, when glucagon and catecholamines both increased during hypoglycemia, the stimulation of HGP was sustained, presumably because epinephrine’s effect took over as glucagon’s effect waned (67). In addition, although insulin and glucagon are generally thought to antagonize each other’s effects, there is evidence that under certain circumstances they may synergize in some respects (73). Thus, in combination, these hormones can generate quite different responses compared with when they are studied in isolation. Further studies are needed to characterize the complex hormone interactions that regulate HGP.

Although recent studies have characterized the impact of acute and chronic glucagon receptor agonism and/or antagonism on liver gene expression in rodents (74–76), our study is unique in that we examined the acute transcriptional impact of hyperglucagonemia in a dog model under conditions in which the ensuing effects on blood glucose and insulin were matched across groups. This experimental design permitted discrimination of the effects of hyperglucagonemia and hyperglycemia per se (i.e., GGN + GLU group) from those of basal glucagon and hyperglycemia (i.e., GLU group) on liver gene expression. Interestingly, the most prominent hepatic transcriptomic signature in the GGN + GLU group was the broad activation of genes encoding proteins involved in hepatic mitochondrial oxidative phosphorylation (OXPHOS). Stimulation of respiration in rat liver mitochondria isolated from glucagon-pretreated animals was first described by Yamazaki and colleagues nearly 50 years ago (77, 78). More recently, Perry et al. (79) showed that glucagon stimulated hepatic mitochondrial oxidation in mice in both acute and chronic contexts in a manner requiring the mitochondrial-associated inositol triphosphate receptor 1 (79). Our findings support and extend these earlier observations by demonstrating that acute physiological hyperglucagonemia and concomitant hyperglycemia, when compared with matched hyperglycemic controls, uniquely activate an OXPHOS gene expression program that presumably supports the heightened energy demands placed on the liver to drive ATP-consuming processes like gluconeogenesis and ureagenesis. The coupling of glucagon action with induction of genes that promote mitochondrial energy production seemingly conflicts with the findings of Berglund et al. (64) who showed that a physiological increase in glucagon promotes hepatic energy depletion in mice, as evidenced by marked reduction in the liver ATP/AMP ratio and activation of AMPK. However, Berglund and colleagues (64) used a unique hyperglucagonemic-euglycemic clamp protocol to isolate hepatic glucagon action in the absence of hyperglycemia or hyperinsulinemia, whereas we examined hepatic glucagon action in a setting of basal insulin and hyperglycemia. Although we did not measure hepatic adenine nucleotide levels in our study, our gene expression findings suggest that the glycemic level differentially influenced the impact of glucagon on hepatic energy state. Alternatively, OXPHOS gene activation after 4 h of hyperglucagonemia may reflect a compensatory response to a prior fall in the ATP/AMP ratio, though such a fall was not reflected in the phosphorylation state of AMPK at 30 min or 4 h. Future studies are needed to clarify the temporal impact of hepatic glucagon action on mitochondrial oxidation and hepatic energy state under carefully controlled physiological conditions. Such studies will be of great interest as chronic glucagon infusion or glucagon receptor agonism in diet-induced obese rodents was shown to reverse hepatic steatosis via stimulation of hepatic mitochondrial fatty acid oxidation (79, 80).

Unexpectedly, genes annotated to pathways involved in skeletal muscle development (i.e., myogenesis) and wound healing, fibrosis, and metastasis (i.e., EMT) were highly enriched among the downregulated genes in the GGN + GLU group. Since myogenesis, EMT, and angiogenesis (another pathway significantly downregulated in the GGN + GLU group) are categorized together as development-related processes (22), our data suggest that glucagon signaling may reinforce mature hepatocyte identity by reducing the expression of genes involved in developmental pathways. Furthermore, functional GO terms associated with these pathways highlighted several biological processes involved in ECM assembly and organization, collagen biosynthesis and degradation, and external encapsulating structure organization. The ECM comprises a heterogeneous network of proteoglycans, polysaccharides, and proteins that provide a structural scaffold for cells and tissues. They transduce signals through ECM-receptor interactions and undergo dynamic changes in their abundance and composition to maintain tissue homeostasis in response to injury (33, 34). Although links between obesity-induced ECM remodeling and impaired insulin action in the liver, skeletal muscle, and adipose tissue have been established (33, 34), few, if any, studies have examined the impact of ECM remodeling on glucagon action or vice versa. Here, we show that hyperglucagonemia leads to downregulation of numerous matrisome genes encoding various collagens and ECM-related proteins. How glucagon signaling may regulate these processes and whether they manifest as alterations in the abundance and/or composition of hepatic ECM remains to be determined.

There is clinical interest in the development of glucose-dependent glucagon-insulin chimeras that limit postprandial hyperglycemia while also reducing hypoglycemic risk (81). To optimize the relative activation of glucagon versus insulin signaling in response to such a molecule, efforts are being made to modify the structure of glucagon to change its affinity for its receptor. Considering glucagon’s transient effects on liver glucose metabolism, future studies are needed to resolve the temporal impact of combined elevations in both glucagon and insulin on the regulation of hepatic and extrahepatic glucose flux in vivo. It would also be interesting to compare some of the glucagon molecules modified for increased stability in this context (e.g., dasiglucagon).

In summary, glucagon acutely stimulates a rapid and profound increase in HGP. This is due exclusively to increased net hepatic glycogenolysis, which is associated with increased cAMP formation and coordinated activation of GP and inhibition of GS. The waning effect of glucagon on hepatic glucose metabolism occurred even though cAMP remained substantially elevated. Hyperglucagonemia caused hyperglycemia and a substantial increase in hepatic G6P, and the allosteric effects of glucose and G6P on GS, GP, and glucokinase are likely to have mediated the evanescence of glucagon action. Hyperglucagonemia also induced gene expression programs associated with upregulation of hepatic mitochondrial oxidation and metabolism, and downregulated genes involved in ECM organization and developmental processes. These results emphasize glucagon’s potent effects on glucose metabolism while revealing novel insights into hepatic glucagon signaling that are relevant to the evaluation of glucagon antagonists as therapeutic options in diabetes treatment.

DATA AVAILABILITY

Data will be made available upon reasonable request.

SUPPLEMENTAL DATA

Supplemental Table S1 (representative Western blots for Fig. 5) and Supplemental Figs. S1–S3: https://doi.org/10.17632/rjhncsmp58.1.

GRANTS

This research was supported by the National Institutes of Health Grant DK18243 and the Diabetes Research and Training Center Grant SP-60-AM20593. Hormone analysis was performed by Vanderbilt’s Hormone Assay and Analytical Services Core, and surgical and experimental expertise was provided by Vanderbilt’s Large Animal Core, both of which are supported by the Vanderbilt University Medical Center Diabetes Research and Training Center grant DK20593. C.J.R. was supported by the American Diabetes Association Mentor-based Fellowship. A.D.C. was supported by the Vanderbilt Jacquelyn A. Turner and Dr. Dorothy J. Turner Chair in Diabetes Research.

DISCLOSURES

A.D.C has a financial interest in Abvance Therapeutics and is a consultant to Novo Nordisk. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

C.J.R., A.D.C., and D.S.E. conceived and designed research; C.J.R., G.K., B.F., and D.S.E. performed experiments; K.C.C., C.J.R., M.S., J.J.W., J.I.-D., E.P.D., P.J.R., and D.S.E. analyzed data; K.C.C., C.J.R., A.D.C., and D.S.E. interpreted results of experiments; K.C.C., M.S., A.D.C., and D.S.E. prepared figures; K.C.C., C.J.R., A.D.C., and D.S.E. drafted manuscript; K.C.C., C.J.R., J.J.W., G.K., J.I.-D., A.D.C., and D.S.E. edited and revised manuscript; K.C.C., C.J.R., M.S., J.J.W., G.K., J.I.-D., B.F., E.P.D., A.D.C., and D.S.E. approved final version of manuscript.