Postprandial Aminogenic Insulin and Glucagon Secretion Can Stimulate Glucose Flux in Humans

Insulin and glucagon are master regulators of glucose homeostasis (1). Dysfunction in the secretion and action of both hormones drives hyperglycemia in type 1 and 2 diabetes (1). Therefore, understanding how insulin and glucagon interact and are integrated at the systems level is of fundamental significance. Classically, insulin and glucagon are described as exerting opposing actions on their target tissues, although their secretion is also considered to be inversely regulated (1). Hence, any increase in insulin secretion is normally accompanied by glucagon suppression, thereby ensuring a tightly regulated feedback loop that maintains glucose homeostasis (1).

Although the reciprocal regulation of insulin and glucagon secretion is apparent after carbohydrate ingestion (1), there are in fact circumstances when this relationship does not hold. Indeed, dietary protein ingestion alone concurrently stimulates both insulin and glucagon secretion (2,3). Under these conditions, insulin and glucagon simultaneously exert agonistic and antagonistic actions on metabolic homeostasis. In relation to amino acid metabolism, through its inhibitory actions on muscle proteolysis, insulin stimulates muscle amino acid uptake (4), whereas glucagon stimulates hepatic amino acid uptake, catabolism, and ureagenesis (5). Thus, through independent actions on the muscle (insulin) and liver (glucagon), these hormones synergistically promote plasma amino acid clearance and postprandial amino acid disposal. However, insulin and glucagon exert opposing actions on hepatic glucose metabolism (1), whereas insulin also promotes tissue glucose uptake. Notably, despite this complex endocrine interaction, and the fact that amino acids are substrates for gluconeogenesis, protein ingestion does not affect blood glucose levels (3). Although the maintenance of euglycemia following protein ingestion is a well-known phenomenon (2,3), the mechanisms involved remain enigmatic.

Since insulin suppresses endogenous glucose production (EGP) and stimulates glucose disposal (Rd), protein-induced hyperinsulinemia would be expected to lower blood glucose levels, yet this does not occur. It therefore seems that protein-induced hyperglucagonemia antagonizes the glucose-lowering actions of insulin to prevent hypoglycemia (6). Interestingly, in both healthy people and those with type 2 diabetes, glucose fluxes (EGP and Rd) remain unaltered following protein ingestion (7–9). Conversely, in individuals with type 1 diabetes, the lack of endogenous insulin secretion yet the persistence of protein-induced glucagon secretion stimulate EGP, causing blood glucose to increase (7). Thus, in people with an intact endocrine pancreas, it appears that protein-induced postprandial hyperglucagonemia precisely counteracts the suppressive actions of insulin on EGP while Rd is not stimulated, thereby maintaining euglycemia.

Although the above-mentioned studies in healthy individuals (7–9) failed to demonstrate changes in glucose flux following protein ingestion, this does not preclude the potential for such effects to occur. It is possible that the magnitude and duration of the hormonal response following protein ingestion (meat or eggs) in the previous studies (7–9) was insufficient to surpass the threshold to alter glucose flux. Evidently, not all dietary protein sources are equipotent at stimulating insulin and glucagon secretion, with the speed of digestion and the rate of systemic amino acid entry being key determinants of the hormonal response (10). For example, slow-absorbing egg proteins exert negligible responses (9,11) and meat proteins exert modest responses (2, 8), whereas dairy proteins (11), particularly the whey fraction (12–14), robustly and rapidly stimulate insulin and glucagon secretion. With this in mind, we devised a physiologic paradigm using fast-absorbing whey protein to examine whether the concurrent induction of sustained hyperinsulinemia and hyperglucagonemia has the capacity to alter glucose fluxes in healthy humans during the postprandial euglycemic state. From a glucoregulatory perspective, this may reveal unique insight into the metabolic “tug-of-war” between insulin and glucagon action. This is particularly relevant since glucagon receptor antagonism and, ironically, agonism (coagonist therapy) are emerging as therapeutic strategies for diabetes and obesity.

Eight healthy individuals (four females/four males; age 30 ± 2 years; weight 66 ± 4 kg; BMI 22 ± 1 kg/m²) participated in this study, which was approved by the Deakin University Human Research Ethics Committee. The purpose, nature, and potential risks were explained to participants, and informed consent was obtained.

Participants completed two trials in random order, separated by 1–2 weeks, where whey protein, or water was ingested. A standardized diet (8,465 kJ; 52% carbohydrate, 18% protein, and 30% fat) was consumed on the day before each trial. Trials commenced following a 10- to 12-h overnight fast, consuming only water from 2130 h the evening prior. Strenuous exercise was avoided for 48 h prior. Upon arrival at the laboratory at 0700 h, height and weight were recorded. A 22-gauge cannula was inserted into a dorsal hand vein in a retrograde fashion for arterialized blood sampling using a heated box (50–55°C). A second cannula was placed in the contralateral forearm vein for [6,6-²H]glucose infusion (Cambridge Isotope Laboratories, Inc., Tewksbury, MA). A primed (33 μmol/kg infused over 5 min) continuous infusion of [6,6-²H]glucose (0.33 μmol/kg/min) commenced at 0730 h. After the primed infusion, [6,6-²H]glucose was infused at a constant rate for 120 min (−120 to 0 min) and continued throughout the experimental period (0–240 min).

At 0930 h (0 min), participants commenced either the protein-feeding or water control trials. During the protein trial, 25 g of whey protein isolate (True Protein, Brookvale, New South Wales, Australia) (French vanilla flavor; <1 g carbohydrate per serving) dissolved in 200 mL of water was ingested at 0, 30, and 60 min (total 75 g of protein ingested). During the control trial, participants ingested 200 mL of water at 0, 30, and 60 min. Arterialized blood was collected in EDTA-containing vacutainers at 7.5, 15, 22.5, 30, 37.5, 45, 52.5, 60, 67.5, 75, 82.5, 90, 105, 120, 135, 150, 180, 210, and 240 min. For glucagon, samples were collected in BD P800 vacutainers.

Plasma glucose, glycerol, amino acids, urea, and β-hydroxybutyrate were measured via gas chromatography-mass spectrometry, with absolute concentrations calculated from linear regression of serially diluted external unlabeled standards using the isotope dilution technique whereby all samples were spiked with stable isotope-labeled internal standards (see Supplementary Data for more detail). Plasma [6,6-²H]glucose enrichment and total glucose concentration were determined using the glucose methyloxime pentaproprionate derivative via positive chemical ionization gas chromatography-mass spectrometry with raw data corrected for natural isotopic background abundance skew (see Supplementary Data for more detail).

Commercially available ELISA kits were used to determine plasma insulin (ALPCO, Salem, NH), glucagon (Mercodia, Uppsala, Sweden), and C-peptide (Millipore, Billerica, MA). Plasma free fatty acids (FFAs; Wako Chemicals, Richmond, VA) and triglycerides (Roche Diagnostics, Basel, Switzerland) were determined via spectrophotometric assay.

Insulin secretion rates were calculated by C-peptide deconvolution, and glucose fluxes were modeled using Steele non–steady-state equations (see Supplementary Data for more detail). All data are reported as mean ± SEM. Two-way repeated-measures ANOVA and paired sample t tests were used where appropriate. A Bonferroni multiple-comparisons test was used for post hoc analysis. Significance was accepted at P < 0.05.

Protein ingestion had no effect on plasma glucose (Fig. 1A), yet there was a rapid and sustained insulin response (Fig. 1B), with C-peptide and insulin secretion rates following a similar pattern (Fig. 1C and D). Protein ingestion also caused marked hyperglucagonemia (Fig. 1E). Although the kinetics and magnitude of the insulin and glucagon responses were similar (approximately sixfold basal) over the first 60 min after protein ingestion (Fig. 1F), the insulin concentration gradually declined thereafter, while glucagon concentration continued to rise, peaking at approximately eightfold basal at 120 min (Fig. 1F). In relative terms, the glucagon response was larger than that of insulin, whereas the C-peptide response was substantially smaller than those of both insulin and glucagon (Fig. 1F). Although plasma triglycerides were unaffected by protein ingestion (Fig. 1G), FFAs and glycerol (Fig. 1H and I) were markedly suppressed.

Plasma essential (Fig. 2A–H) and nonessential (Fig. 3A–J) amino acid levels rapidly increased following protein ingestion with the branched-chain amino acids, particularly leucine (approximately sixfold basal) and isoleucine (approximately ninefold basal) showing the greatest response (Fig. 2A and B). Additionally, plasma urea concentrations progressively increased following protein ingestion (Fig. 3K). β-Hydroxybutyrate levels increased gradually during both the water and protein trials, with no significant differences between either condition (Fig. 3L). Following protein ingestion, there was a progressive increase in glucose flux, with EGP and Rd both increasing by ∼25% (Fig. 4A–C). As a result, on average ∼11 g of additional glucose entered the circulation during the 4-h protein ingestion trial (Fig. 4D).

Previous single-dose whey protein ingestion studies have shown robust, yet transient euglycemic insulin and glucagon responses (12–14). Here, we used a multidose protein-feeding strategy so as to potentiate the duration of this endocrine response to maximize the opportunity for insulin and glucagon to alter metabolism. As intended, this feeding paradigm caused a rapid and sustained euglycemic insulin response. Interestingly, the postprandial hyperinsulinemia was mediated by increased insulin secretion and reduced insulin clearance, as evident by the lower C-peptide response relative to that of insulin. Although a rise in glucagon was expected, the magnitude was unprecedented (approximately eightfold basal). To the best of our knowledge, this is the largest physiological, not pathological (e.g., glucagonoma), increase in glucagon that we have observed in the literature. Although these hormonal responses did not impact glycemia, plasma FFAs and glycerol were suppressed, which is consistent with adipose tissue being highly insulin sensitive (15). This supports the view that the antilipolytic actions of insulin on adipose are dominant, and, within physiological concentrations, glucagon has little, if any lipolytic effect (16).

As expected, protein ingestion had no effect on plasma glucose despite causing marked hyperinsulinemia and hyperglucagonemia. So how did blood glucose concentrations remain constant? Protein feeding proportionally increased both Rd and EGP, with glucose turnover increasing by ∼25%, which explains the euglycemia. Although the source of the extra EGP is unknown, we speculate that the combined hyperaminoacidemia and hyperglucagonemia stimulated gluconeogenic flux, thus providing an energy-costly pathway for hepatic and perhaps kidney amino acid catabolism (5). To our knowledge, this is the first time pure protein ingestion has been shown to robustly stimulate Rd in healthy humans. In light of the physiologic euglycemic-hyperinsulinemic-hyperglucagonemic state (i.e., absence of hyperglycemia) (17), the additional glucose produced was not likely taken up by the liver, but rather by muscle (skeletal and cardiac) and/or adipose tissue.

Previous studies have shown that whey protein ingestion during an insulin clamp can modestly impair insulin-stimulated glucose uptake (18). It is therefore possible that the hyperaminoacidemia in our study prevented an even greater increase in insulin-stimulated Rd from occurring. However, we do not believe that this is pathological insulin resistance, but rather a physiological compensation for acutely altered substrate availability. Regardless, our feeding protocol induced a sufficient degree of hyperinsulinemia to stimulate peripheral tissue glucose uptake, even in the face of marked hyperglucagonemia. This supports the concept that physiological concentrations of glucagon do not directly oppose skeletal or cardiac muscle glucose metabolism as these tissues lack glucagon receptors (19). Since previous studies have failed to stimulate EGP in humans in the presence of selectively high glucagon and gluconeogenic substrate concentrations, an effect attributed to hepatic autoregulation (20), we believe that the increase in EGP here is secondary to insulin-stimulated Rd. We therefore propose that the increase in Rd induced a proportional increase in EGP to maintain euglycemia in the absence of exogenous glucose. Accordingly, protein-induced stimulation of glucose flux appears to be dependent on the magnitude and duration of the hyperinsulinemia, likely explaining why previous protein-feeding studies that reported modest insulin responses did not observe changes in glucose flux (7–9). This is consistent with the notion that stimulation of peripheral (muscle) Rd is a naturally insulin-resistant process requiring substantial increments in systemic insulin (15). In regard to the multidose whey protein approach used here, it is important to note that it is not representative of typical daily protein intake derived from whole food over multiple meals, but nevertheless elicits a physiological postprandial response that was intended to push the metabolic and endocrine systems in order to reveal fundamental control mechanisms that govern metabolic homeostasis and provide teleological insight.

It is interesting to comment on the direct (hepatocyte) versus indirect (extrahepatic) effects of insulin on EGP. It has been suggested that insulin trumps glucagon at any concentration to inhibit EGP (21), and that the suppression of FFAs through the antilipolytic effects of insulin is a key determinant of this response (the “single gateway hypothesis”) (22). Our data do not support this. Here, EGP increased despite marked FFA suppression, thereby dissociating the control of postprandial EGP from adipose lipolysis. Furthermore, we show that postprandial hyperglucagonemia can completely override the effects of hyperinsulinemia on EGP, suggesting that the “tug-of-war” on hepatic glucose metabolism between insulin and glucagon is hepatocyte specific (direct).

Finally, how are glucose fluxes precisely coordinated to maintain euglycemia under these conditions and should the glucagon response be viewed as glucose counter-regulatory? These are not trivial issues to decipher but, perhaps, are most appropriately discussed from an evolutionary perspective. Through evolution of the endocrine system, complex life-forms have integrated metabolic control across multiple effector organs (i.e., liver, kidneys, adipose, and skeletal muscle). This multiorgan control has resulted in “hard-wired” metabolic flexibility, permitting survival across widely varying environmental conditions such as famine, feasting, and seasonal changes to macronutrient availability. Perhaps our view of insulin and glucagon biology has been heavily shaped by the context of our modern mixed macronutrient, yet dominantly high carbohydrate diets. Through multiple daily meals, dietary glucose from processed sugars and starch results in frequent glucose spikes, requiring a dominant insulin and weakened glucagon response, leading to the inhibition of EGP and the stimulation of Rd. As glycemia returns toward baseline, the body must ensure that glucose fluxes are not overcommitted toward glucose reduction, thus requiring counter-regulatory glucagon secretion and hence appropriate EGP recovery to prevent hypoglycemia.

However, during the Paleolithic period in which humans evolved (∼2.6 million to 10,000 years ago) (23), many, perhaps even the majority of our hunter-gatherer ancestors would have infrequently or possibly never experienced substantial blood glucose spikes. Whether the diet was periodically or persistently absent in carbohydrates, such as those diets in cold-climate hunting populations (e.g., Inuit), or whether the diet contained modest quantities of unprocessed high-fiber plant-derived food, the diets would have been by and large classed as “low glycemic” and relatively high in protein (23). Accordingly, concurrent euglycemic protein-induced insulin and glucagon secretion may have been the “prototypical” postprandial response, as would be expected to occur in carnivores (24), where the hyperaminoacidemia following meat-based meals likely results in substantial insulin and glucagon secretion. The insulin response would concomitantly inhibit adipose lipolysis and muscle proteolysis, while stimulating muscle amino acid and glucose uptake. Ultimately, this would favor adipose tissue lipid deposition and muscle protein and glycogen synthesis. Synchronously, given the predominant distribution of glucagon receptors at the liver and kidney (19), the combined postprandial glucagon response and hyperaminoacidemia would stimulate hepatic amino acid uptake, deamination, oxidation, gluconeogenesis, and ureagenesis while enhancing urea excretion at the kidney (25). This would stimulate plasma amino acid disposal, ammonia detoxification, and hepatic glycogen synthesis likely via the indirect (gluconeogenic) pathway while supporting elevated EGP to maintain euglycemia. Given the absolute requirement of protein, but not carbohydrate, for mammalian survival, it could be envisioned that such endocrine and metabolic responses were evolutionarily selected for in humans. Accordingly, we believe that glucagon is a bona fide postprandial hormone, which evolved to synergistically support insulin. Consequently, attempts to therapeutically inhibit or activate the glucagon receptor need to carefully consider not only glucose, but also amino acid and nitrogen homeostasis.

Funding. T.A. is supported by an Australian Government Research Training Program Scholarship. C.R.B. (grant FT160100017) and G.M.K. (grant DE180100859) are supported by Australian Research Council fellowships. The project was funded through the Diabetes Australia Research Program general grant scheme.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. T.A., C.R.B., and G.M.K. contributed to all aspects of the study including conceptualization and study design, data collection and analysis, and manuscript preparation. C.R.B. and G.M.K. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Data Availability. Any raw data sets generated during the current study are available from the corresponding author on reasonable request, with all reagent and analytical detail available in the online Supplementary Data.