Clocking Up GLP-1: Considering Intestinal Rhythms in the Incretin Effect

Pancreatic β-cells, like many other cells, contain autonomous circadian oscillators, the disruption of which leads to hypoinsulinemia and hyperglycemia in mice (1,2). Rather than slavishly following the master clock located in the suprachiasmatic nuclei, which synchronize internal oscillations with the external light-dark cycle, some peripheral clocks, including the one in the endocrine pancreas, can be reset by alternative clues, such as time-restricted food availability. The way that this food “entrainment” works has remained unclear. GLP-1, an incretin hormone that boosts glucose-dependent insulin secretion, is a candidate messenger that could potentially coordinate peripheral oscillators with the timing of food arrival in the gut. In this issue of Diabetes, Gil-Lozano et al. (3) investigated the influence of the day-night cycle on the GLP-1 axis.

To circumvent interference from previous meals, Gil-Lozano et al. fasted rats for 4 h before applying oral glucose tolerance tests (OGTTs) at intervals throughout the day. They observed clear 24-h rhythmicity of plasma GLP-1 concentrations that were in phase with insulin levels. Interestingly, the strongest responses were observed toward the end of the lights-on period—a time when nocturnal rats allowed to consume food freely do not usually eat much. The circadian rhythm of plasma glucose responses had a higher frequency, possibly reflecting the 12-h period reported for hepatic metabolic cycles (4).

To test if the observed diurnal rhythm of GLP-1 responses was food entrainable, rats were accustomed to feed only during the dark or the light phase. Dark-fed rats, like their free-feeding counterparts, had a more pronounced GLP-1 response at the end of the light phase, consistent with the normal nocturnal feeding pattern of ad libitum–fed rats. Importantly, however, light-fed animals showed a clear reversal of the responsiveness, with greater GLP-1 responses appearing at the end of the dark phase and a parallel reversal of the insulin profile. Somewhat surprisingly, increased GLP-1 and insulin responses did not translate into lower glucose concentrations, but additional factors such as the rates of intestinal glucose absorption and hepatic gluconeogenesis also could influence plasma glucose profiles.

Indeed, glucose absorption from the intestine is dependent on sodium-glucose transporter 1 (SGLT1), whose expression and function is known to show a circadian rhythm that normally peaks at the end of the light phase in rats but shifted ∼8 h when food access was restricted to the lights-on period (5). The peak expression of SGLT1 at the end of the light phase in ad libitum– and dark-fed rats (5) corresponds well with the more rapid rise of plasma glucose at this time seen by Gil-Lozano et al. As SGLT1 underlies glucose-sensing in GLP-1 secreting L cells (6), its rhythmicity also may contribute to the correspondingly enhanced GLP-1 responses.

A counterintuitive relationship between plasma glucose and the GLP-1 profile also was observed in dark-fed adapted animals deprived of external cues by their exposure to constant light. These animals were relatively glucose intolerant at “nighttime,” compared with dark-fed animals experiencing normal light-dark cycles. These experiments are, however, difficult to interpret, as small mammals can exhibit different individual behavioral responses to constant light. Some become arrhythmic or retain some rhythmicity with lengthened periods, whereas others show a phenomenon known as splitting, in which two periods of increased activity with drifting times of onset occur during each 24-h period (7,8). It should also be noted that restricting food access to just 4 or 6 h daily in the middle of the lights-on phase has previously been reported to result in an anticipatory transient elevation of plasma GLP-1 ∼1 h before the reintroduction of food (9,10). No attempts were made to link these observations to a circadian rhythm, but the results suggest that food restriction has effects beyond resetting the intestinal clock and the responsiveness of enteroendocrine L cells to glucose.

Gil-Lozano et al. investigated the intracellular mechanisms underlying circadian responsiveness in L cells, using the GLUTag cell line. Serum starvation of these model L cells combined with a brief stimulation with the adenylate cyclase activator forskolin resulted in synchronized oscillations of the clock genes Bmal1, Per2, and Rev-Erbα over the following 48 h. These genes normally maintain a 24-h periodicity through opposing negative and positive feedback loops (11,12), as illustrated in Fig. 1. In GLUTag cells, Bmal1 expression peaked 4 h after synchronization, while Rev-Erbα and Per2 were expressed at higher levels 16 h after the forskolin shock. Secretory responses to the muscarinic agonist bethanechol fluctuated approximately in phase with Bmal1, and responses to a number of other stimuli were similarly enhanced at 4 h, but largely absent 16 h, after synchronization. Expression of proglucagon, the precursor of GLP-1, was enhanced within the first 24 h after forskolin treatment, in line with previous reports (13), but fell during the subsequent day without an obvious circadian rhythm. Regulation of the proglucagon gene therefore would be unlikely to underlie the rhythmicity of GLP-1 secretory responses.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

L cells are stimulated by nutrient arrival in the intestinal lumen to secrete GLP-1. The responsiveness of L cells has a diurnal rhythm kept by the canonical mammalian clock. This consists of transcriptional loops, in which the transcriptional activator complex BMAL1/CLOCK promotes expression of the period (e.g., Per2) and cryptochrome (Cry) genes and the orphan nuclear receptor genes Rev-Erbα and Rorα. Once translated, the PER/CRY protein heterodimer inhibits BMAL1/CLOCK activity, while REV-ERBα and RORα inhibit and promote Bmal1 transcription, respectively. The phase of this clock, usually synchronized with the light-dark cycle through signals from a similar molecular clock in the suprachiasmatic nucleus, is itself reset by food availability and affects secretory responses to a range of stimuli, at least in part by Tef and Prl1 activity.

Interestingly, knockout of Bmal1 in pancreatic β-cells has been shown to interfere with insulin secretion distal to glucose-induced membrane depolarization and elevation of cytosolic Ca²⁺, possibly affecting vesicular trafficking and/or fusion (1). More recently, clock genes have been linked to NAD⁺ and sirtuin activity oscillations in hepatocytes (14), and it has been speculated that this might link the cellular clock to later stages of hormone secretion (15). Gil-Lozano et al. concentrated on two other genes, namely the transcription factor thyrotroph embryonic factor (Tef) and protein tyrosine phosphatase 4a1 (also known as phospatase of regenerating liver-1 [PRL-1]), which were differentially expressed 4 and 16 h after synchronization in GLUTag cells. They further showed that knockdown of these two candidates at the time of their respective peak expression levels affected responses to a number of secretagogues. The possible mechanisms linking these genes to GLP-1 secretion need to be investigated in future experiments to fully consolidate this aspect of the authors’ work.

In summary, Gil-Lozano et al. (3) have established that a diurnal rhythm underlies GLP-1 secretion. As the central and peripheral GLP-1 axes exert independent effects on food intake (16), it will be interesting to establish whether brain stem GLP-1–secreting neurons also exhibit a diurnal activity pattern. How the current findings relate to the reported association of metabolic disease and shift work (17) and the potential implications for the use of once-daily or once-weekly GLP-1 mimetics (18) for the treatment of diabetes will be fascinating areas for future research.

• © 2014 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered.