The role of CCK8 in the inhibition of glucose production

Abstract

A recent study suggested that a rise of cholecystokinin (CCK8) in the duodenum may bring about an inhibition of hepatic glucose production. The authors made use of the pancreatic clamp technique to characterize a gut-brain-liver signal generated by CCK8 that reduces glucose output by the liver. The pancreatic clamp conditions used created a situation in which the liver was markedly deficient in both insulin and glucagon. Although the data presented indicated that CCK8 can reduce glucose production, the authors do not establish a role for this inhibition in the reduction of glucose output seen in response to feeding. It must be remembered that in response to a meal the insulin level in the hepatic sinusoids rises markedly, as does the insulin level to which the brain is exposed. It therefore seems likely that either or both of these effects will drive the suppression of glucose production rather than any effect of CCK8. The importance of the CCK8 effect needs to be determined in the presence of elevated arterial and sinusoidal insulin before any conclusion can be drawn about its relevance.

It has been known for some time that nutrient entry into the intestine triggers the release of peptides which are involved in the regulation of gut function (1), pancreatic function (2), and appetite control (3-5). One of those peptides, CCK8, is released from I cells that line the mucosa of the duodenum in response to lipid content (6). The peptide is thought to interact with the CCK-A receptors (7; 8) on vagal afferent nerves in the lamina propria of the mucosa (8; 9) and thereby inhibit food intake (7).

A paper by Cheung et al. (10) published recently in Cell Metabolism proposed a novel role for CCK8 in glucose regulation. The authors suggested that CCK8 plays a role in meal induced suppression of hepatic glucose production. The validity of this concept depends on two factors: (i) establishment of the circuitry required for the response; and (ii) proof that, under normal physiologic circumstances, the mechanism is functional. While the data presented clearly accomplish the former, they do not establish the latter.

The authors used a conscious rat model to examine the role of duodenally infused CCK8 (35 pmol/kg/min) on the turnover of glucose during a pancreatic clamp (see Figure 1). The clamp began at 90 min and consisted of somatostatin infusion to inhibit hormone secretion by the endocrine pancreas, along with insulin replacement via a peripheral vein at a rate designed to clamp the arterial plasma insulin level at its pre-existing value. Indeed the peripheral plasma insulin level (ng/ml) was 0.85±0.14 prior to the clamp and 0.73±0.09 and 0.63±0.13 during the clamp in the saline and CCK8 groups, respectively. Euglycemia was maintained by glucose infusion, but the rate required was ∼3 mg/kg/min greater in the presence of CCK8. This difference was attributed to a greater suppression of tracer determined glucose appearance (Ra) in the CCK8 group (∼Δ4.2 mg/kg/min) which represented a 30% fall in glucose output by the liver. At the same time there was a somewhat offsetting ∼1.2 mg/kg/min CCK8-induced decrease in tracer determined glucose disappearance (Rd). Blockade of the CCK-A receptor by intraduodenal infusion of the inhibitor MK329 eliminated the effect of CCK8 on both Ra and Rd. Likewise, OLETF rats, which have a genetic absence of the CCK-A receptor, showed a deficient response to CCK8 when compared to the response seen in genetically similar rats (LETO) with a functional CCK-A receptor.

Figure 1.

The role of neural signalling in the response of the liver to CCK8 was examined in three ways. The effect of duodenal CCK8 infusion was examined in the presence of intraduodenal tetracaine infusion (blockade of neurotransmission of gut vagal afferent fibers), MK801 infusion into the NTS (blockade of the NMDA channel), or hepatic vagotomy (blockade of parasympathetic signalling in the liver). In each case, the intervention inhibited the effect of CCK8 on Ra suppression, leading to the conclusion that a neural loop involving the vagus nerve is involved in the inhibition of glucose production by the liver. This raises an interesting question, however, since the effect of CCK8 on glucose production was seen relatively quickly (30-50 min), whereas in earlier studies the effects of neural input on glucose production were slow to manifest. Pocai et al. determined that a similar neural circuit, one that involved insulin action in the hypothalamus and subsequent alteration in vagal input to the liver, inhibited glucose production entirely through an effect on gluconeogenesis that was associated with reduced gluconeogenic gene expression (11). Another study showed that hypothalamic insulin action led to the phosphorylation of hepatic STAT3, and that STAT3 was required for the inhibition of hepatic gluconeogenic gene transcription and the suppression of glucose production (12). That liver STAT3 phosphorylation was only observed after three hours of refeeding (12) is consistent with the observation (11) that the insulin-brain-liver axis requires three hours to bring about a decrease in plasma glucose level. Thus, the rapid time course of CCK8-mediated inhibition of glucose production observed by Cheung et al. is in contrast to the centrally mediated effects of insulin, despite both being brought about by the vagus nerves, and is not likely explained by a suppression of gene transcription, which would take several hours to cause a functional change in protein levels. Therefore, the time course of the hepatic response to CCK8 is surprising, since regardless of the signal activating the neural circuit, it might be expected that a change in hepatic vagal input would produce a similar response at the liver and that it would occur by the same mechanism (i.e. suppression of gluconeogenesis).

The authors next showed that infusion of lipid into the duodenum, in the presence of a pancreatic clamp, brought about the same effect on glucose turnover as did CCK8 infusion. As before with CCK8, this effect was inhibited by concurrent infusion of the CCK-A receptor inhibitor MK329. They then went on to show that high fat feeding for three days rendered the animals unresponsive to the effect of intraduodenal CKK8 infusion on glucose turnover, suggesting that high fat feeding caused CKK8 resistance. Finally, the authors attempted to assess the impact of CCK-A receptor inhibition on the response of fasted rats (40h) to the acute consumption of regular chow ad libitum. In response to refeeding, there was a slightly greater rise (∼20 mg/dl) in plasma glucose over the initial 20 min of the feeding period in the presence of the CCK-A receptor blocker or CCK-A receptor deficiency (Table 1).

Table 1.

Plasma glucose (mg/dl) in 40h fasted rats fed chow ad libitum.

GROUP	Time (min)
	-10	0	10	20
Saline	107±4	111±5	125±7	143±7
MK29	104±2	112±3	125±5	161 ±7*
LETO	----	114±3	135±5	150±5
OLETF	----	111±3	153±13	169±8*

^*p<0.01 versus control (SAL or LETO) animals. Values are shown as mean ± SEM and are derived from graphical representation by Cheung et al. (10).

Taken together, the data of Cheung et al. (10) indicate that the circuitry exists to allow an intraduodenal lipid load to participate in the meal associated reduction in hepatic glucose output. They do not, however, establish the physiologic relevance of the pathway. The reasons for this are several-fold. The experimental conditions brought about by the pancreatic clamp, while maintaining arterial insulin levels, undoubtedly resulted in marked hypoinsulinemia at the liver. Somatostatin inhibits endogenous insulin secretion, thereby eliminating the 3-fold elevation of insulin normally seen in the portal vein vs the artery (13). In addition, the tissues of the gut destroy ∼20% of the insulin in the blood reaching them (14). Therefore, compared to the basal period, the insulin level within the liver was reduced by ∼75% during peripheral insulin replacement (Table 2). Liver glucose production is extremely sensitive to small changes in circulating insulin (15) and the portal vein insulin level is a primary regulator of hepatic glucose production (16; 17). Given the above, one might have expected glucose production to increase during the clamp. However, that did not occur and euglycemia was maintained. The explanation for this probably lies in the fact that glucagon was not replaced during the clamp, even though its secretion was also inhibited by somatostatin. Thus, in the study of Cheung et al., the liver was receiving neural input at a time when the direct insulin signal was markedly deficient and the stimulatory drive of glucagon was absent. Reduced direct hepatic insulin signaling may have created an artificial circumstance where the role of CCK in regulating glucose production was amplified.

Table 2.

Plasma insulin levels (ng/ml) during basal and clamp conditions. Arterial data were directly measured by Cheung et al and presented in supplementary figure (10).

	Basal	Clamp
		SAL	CCK8
Artery	0.85	0.73	0.63
Portal Vein*	2.55	0.58	0.50
Hepatic Sinusoids*	2.21	0.61	0.53

^*Portal vein and hepatic sinusoid insulin values were estimated assuming that the portal vein insulin level was 3-fold the arterial level pre-clamp and 20% lower post-clamp, and that 80% of hepatic blood flow was portal in origin

The physiological relevance of the clamp is further weakened by the fact that when a meal is consumed insulin secretion normally increases markedly and the rise in portal vein insulin would itself shut off glucose production (18). In addition, the insulin level within the brain would be elevated following feeding and if brain insulin drives an inhibitory signal to the liver, one would expect this to increase during a meal. It thus seems unlikely that the small effect of CCK8 on vagal input would have a significant impact on the inhibition of hepatic glucose production seen in response to a meal.

The only data presented by the authors that speak to the point of physiologic relevance are those shown in Table 1. Genetic elimination or blockade of the CCK-A receptor during refeeding of 40h fasted rats was associated with a slightly greater (20 mg/dl) glucose excursion at 20 min, with the inference that CCK8-mediated suppression of Ra was impaired. There are several problems with this conclusion. First, the observation period for the plasma glucose profile was unusually brief. A longer time course may have more convincingly and reliably demonstrated the extent of the CCK8-mediated effect. Second, CCK receptors exist in the endocrine pancreas (19) and physiologic increases in CCK have been shown to stimulate insulin secretion (20-22). Therefore, inhibition of CCK8-stimulated insulin secretion during refeeding could have decreased plasma insulin and increased glycemia. Finally, the differences in glycemia may be explained by the ability of CCK8 to slow gastric emptying (20; 23), as opposed to the ability to suppress Ra.

In summary, the novel experiments of Cheung et al. clearly establish the potential of gut lipids to lead to a decrease in glucose production by the liver through a CCK8-dependent mechanism. Unfortunately, the effect was observed at a time when the liver was hypoinsulinemic and hypoglucagonemic, raising questions as to the physiologic relevance of the studies. It remains to be established whether a small CCK8-driven signal can have a meaningful impact on hepatic glucose production in the context of a physiologic rise in insulin during a meal.