The hepatic glucose-mobilizing effect of glucagon is not mediated by cyclic AMP most of the time

Since its discovery by the Nobel laureate Earl Sutherland and his colleagues more than a half century ago (1), cyclic adenosine monophosphate (cAMP) has been widely regarded as the predominant, if not the exclusive, intracellular signal mediating the hepatic glucose-mobilizing effects of glucagon. But substantial evidence persuasively contradicts that view, supporting instead an alternative interpretation: the glucagon receptor type 2 (GR2)/adenylate cyclase (AC)/cAMP/protein kinase A (PKA) pathway is a backup signal. It is recruited—in response to marked elevations in plasma glucagon concentrations—during times of extreme metabolic stress, notably in the early neonate (2) and the strenuously exercising adult (3), boosting glucagon-induced hepatic glucose production to meet the elevated systemic glucose demand. Whether and to what extent the cAMP-dependent pathway is involved in two other stressful conditions, starvation (fasting for more than 24 h) and diabetes (type 1 or 2), is much less clear (4–8).

At least three lines of evidence are consistent with the assertion that the GR2/AC/cAMP/PKA pathway does not mediate the hepatic glucose-mobilizing actions of glucagon most of the time, i.e., in the fed or short-term fasting (usually overnight) unstressed mammal while at rest or during normal physical activity. The first is the wide gap between hepatic portal plasma glucagon concentrations in vivo and the consensus threshold concentration required to activate hepatic AC and increase tissue cAMP levels in hepatocyte or liver preparations ex vivo. As Sutherland and coworkers stated in 1971 (9): “It is important to consider whether the concentrations of glucagon … promoting cyclic AMP accumulation in the isolated liver, lie within the normal range in portal venous blood.” A wealth of evidence gathered since then confirms that they do not. According to 25 reports, published between 1977 and 2017, the collective mean hepatic portal plasma concentration, as assessed by radioimmunoassay (RIA), is 39.4 pM, with a statistical upper limit of 56.6 pM (Table 1). If the assay method had been the more specific and sensitive enzyme-linked immunosorbent assay (ELISA) technique (35), the estimate of the mean hepatic portal concentration likely would have been half that (36, 37), around 19 pM. In contrast, according to 15 other reports (see the legend for Fig. 1) published between 1971 and 1995, including the original publication of Sutherland and coworkers (9), the threshold concentration for the activation of AC or increasing tissue cAMP levels in perfused rat livers, hepatocytes, or hepatocyte membranes is at or close to 100 pM (Fig. 1A). Thus, the ratio of the mean concentration in hepatic portal plasma in vivo to the consensus threshold concentration required to activate AC ex vivo is somewhere between 0.2 and 0.4, depending on the assay technique. Sutherland and coworkers (9) pointed to a similar concentration discrepancy to rule out a physiological role for epinephrine in regulating hepatic glucose metabolism in vivo. It remains a mystery why the same logic has not been applied more frequently to the question of whether the AC/cAMP pathway mediates the hepatic glucoregulatory effects of glucagon, especially considering that the more physiologically relevant alternative signal was discovered by Houslay and coworkers only 15 years later, in 1986 (51).

Table 1.

Hepatic portal plasma glucagon concentrations in humans, dogs, rats, and miniature pigs

Species	[Glucagon], pM	Nutritional State	References
Human	34	Fasted overnight	10
	38	Fasted overnight	11
	52	Fasted overnight	12
	53	Fasted overnight	13
Dog	11	Starved 42 h	14
	16	Fasted 16–18 h	15
	18	Fasted 16 h	16
	20	Fasted 18 h	17
	32	Fasted overnight	18
	34	Fasted 18–24 h	19
	56	Fasted overnight	20
	57	Fasted 16 h	21
	67	Fasted 12–18 h	22
	84	“Fasting”	23
	43	Fed	24
Rat	21	Fasted overnight	25
	23	Starved 27–30 h	26
	25	Fed	27
	32	Fasted 16 h	28
	34	Starved 48 h	29
	39	Fasted 20 h	30
	54	Fed high CH2O	31
	55	Fed 6 h	32
Pig	49	Fasted 24 h	33
	37	Fed	34

Mean ± SE: 39.4 ± 3.5. All values for hepatic portal glucagon concentrations were determined by radioimmunoassay (RIA), most utilizing the “Unger 30 K antibody” (16). All of the human subjects and experimental animals were nondiabetic, and were either conscious at rest or under anesthesia. The subjects in Ref. 11 were cirrhotic. The subjects in Ref. 13 were assumed to be fasted overnight. The animals in Ref. 34 were assumed to be fed because the nutritional state was not specified. Where applicable, listed values are those of untreated or unmanipulated control groups. Statistical values are as follows: n = 25; M = 39.36; SD = 17.62; SE = SD/5 = 3.52; 99.9999% confidence interval = 3.52 × 4.892 = 17.24; 99.9999% confidence interval range = 22.1–56.6 pM.

Figure 1.

Comparisons of glucagon concentration-effect curves for hepatic glucose output (A and B) with those for activation of the adenylate cyclase (AC)/cyclic adenosine monophosphate (cAMP) pathway (A) and the phospholipase C (PLC)/inositol-3,4,5-triphosphate (IP3) pathway (B). The shaded vertical bar depicts the full range of the hepatic portal plasma glucagon concentrations listed in Table 1: Mean (dotted line), 39.4 (10^1.60) pM; Lower limit, 11 (10^1.04) pM; Maximum 84 (10^1.92) pM. The glucose output curve is adapted from Ref. 38, representing the effect of glucagon on the isolated perfused rat liver. The composite AC/cAMP curve in A is the average of 15 individual curves (4, 6, 9, 39–50) generated in rat hepatocytes, hepatocyte membranes, or isolated perfused livers. The inositol-phosphate curve in B is from Refs. 51 (closed triangles) and 49 (open triangles).

That is the second line of evidence. It supports the hypothesis that the glucagon receptor type 1 (GR1)/phospholipase C (PLC)/inositol-3,4,5-triphosphate (IP3)/Ca²⁺/CaM pathway is the predominant or exclusive signal for glucagon in vivo most of the time. In their seminal report, Houslay and coworkers showed that: 1) Between 10 and 300 pM, glucagon concentration-dependently stimulated the production of inositol phosphates from membrane phospholipids (Fig. 1B) and 2) A derivative of glucagon, 1-N-α-trinitrophenylhistidine,12-homoarginine glucagon (TH-glucagon), also increased the production of inositol phosphates, to a similar extent over a similar concentration range, without affecting cellular levels of cAMP. Between 11 and 84 pM (full range of hepatic portal concentration estimates in Table 1), glucagon increases glucose output in perfused livers from 30% up to 98% of the maximum produced by 1,000 pM (Fig. 1, A and B). Over that same concentration range, glucagon activates GR1 receptors coupled to PLC/IP3 from 10% to 20% of the ultimate maximum produced by 300 pM (Fig. 1B), but has a negligible effect on GR2 receptors coupled to AC (Fig. 1A). Downstream targets of the PLC/IP3/CaM pathway are similar or identical to those of the AC/cAMP/PKA pathway, but are often influenced by subtly different mechanisms. For example, levels of intracellular calcium (Ca²⁺_i) can be increased either by binding of PLC-generated IP3 to IP3 receptors coupled to reticular calcium channels at physiological concentrations, or by PKA-mediated phosphorylation of reticular IP3 receptors and plasmalemmal calcium channels in response to supraphysiological concentrations (52). The elevations of Ca²⁺_i produced by activation of either pathway then activate glycogen phosphorylase kinase, promoting glycogenolysis. Both CaMK and PKA can phosphorylate the same downstream glucose-metabolizing enzymes, alter their activities in the same direction, or both. Common target enzymes include fructose-1,6-bisphosphatase, fructose-2,6-bisphosphatase, 6-phosphofructo-2-kinase, pyruvate kinase, glycogen phosphorylase, and glycogen synthase (53–58). Transcriptional effects common to both pathways include phosphorylation of CREB and Fox01 by either calmodulin-dependent protein kinase (CaMK) or PKA, enhancing the expressions of the gluconeogenic enzymes glucose-6-phosphatase and phosphoenolpyruvate carboxykinase (55, 57, 59–61). The important point here is that the threshold concentrations of glucagon required to produce these effects are in the low physiological range, between 1 and 50 (10^1.70) pM, not sufficient to activate AC (Fig. 1A). At those concentrations, only the PLC/IP3/CaM-dependent signal mediates the responses (Fig. 1B). Thus, by activating either the PLC/IP3/CaM pathway alone at physiological concentrations or the PLC/IP3/CaM and AC/cAMP/PKA pathways simultaneously at supraphysiological concentrations, glucagon progressively promotes glycogenolysis, gluconeogenesis, and glucose output.

Third, a key study all but rules out the possibility, proposed early on (62), that at physiological plasma concentrations, sufficient to increase activities and expressions of glycogenolytic and gluconeogenic enzymes, glucagon activates a latent AC/cAMP signal that is not detected by conventional assay methods. A downstream component of the PLC/IP3 pathway, calmodulin-dependent protein kinase kinase beta (CaMKKβ) (aka CaMKKII), phosphorylates the alpha subunit of the “metabolic switch,” adenosine monophosphate-activated protein kinase (AMPK), at Thr172. PKA phosphorylates the enzyme at a different site, Ser485 (63, 64). Infusion of glucagon into anesthetized dogs increased the mean hepatic sinusoidal plasma glucagon concentration from 14 (10^1.15) pM to a peak of 57 (10^1.76) pM, approaching the upper end of the physiological hepatic portal concentration range (Table 1 and Fig. 1). At that concentration, sufficient to enhance glucose output by ∼85% of the maximum according to the glucagon-glucose output curve in Fig. 1, A or B, glucagon increased the phosphorylation of AMPKα at Thr172, the CaMKKβ site, but not at Ser485, the PKA target (65). Seemingly, the most plausible explanation is that, between 0 and at least 57 pM, within the hepatic portal concentration range found in unstressed fed or fasting, nondiabetic adult mammals, glucagon activates the PLC/IP3/calmodulin (CaM) pathway exclusively to stimulate glucose mobilization, without activating the AC/cAMP/PKA pathway above constitutive levels.

Too often, the PLC/IP3/CaM pathway has been acknowledged only in passing or even ignored altogether, when in fact it deserves a place of prominence in future investigations (66). The administration of high pharmacological concentrations ex vivo, commonly at the extreme concentration of 100,000 (10⁵) pM, yields information that is difficult to interpret because of the complex cross talk between the two largely redundant pathways. Given the increasing appreciation of glucagon’s central role in the pathophysiology of diabetes, unequivocally establishing its true mechanism of action in health and disease is now more urgent than ever.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author.

AUTHOR CONTRIBUTIONS

R.L.R. drafted manuscript; edited and revised manuscript; approved final version of manuscript.