Angptl8 helps reset the liver’s daily clock in mice based on food timing

Mammalian circadian clocks precisely control the rhythms of behavior and physiology, and can be reset by various environmental signals. While the light–dark (LD) cycle resets the master clock, timed food intake is a potent synchronizer of peripheral clocks¹. In mammals, the circadian clock system comprises ubiquitously expressed cellular interconnecting transcriptional–translational loops of clock genes/proteins². In the core loop, the transcriptional activators of circadian locomotor output cycles kaput (CLOCK) and brain and muscle ARNT-like 1 (BMAL1) form heterodimers that activate the expression of repressors, encoded by three Period (Per 1–3) and two Cryptochrome (Cry 1/2) genes. Towards the end of the day, PER and CRY protein complexes translocate into the nucleus and inhibit CLOCK/BMAL1-mediated transcription. Due to their instability, the protein and mRNA abundance of these repressors rapidly decreases below the threshold required for autorepression, clearing the way for a new cycle^3–5. In addition, the orphan nuclear receptors of the retinoic acid receptor-related orphan receptor (ROR) and REV-ERB families are also implicated in the control of circadian clock function⁶.

Epidemiological investigations have indicated that prolonged light exposure or rotating shift work is closely correlated to an elevated risk for developing various diseases, including cancer, cardiovascular disease, immune deficiency, and metabolic disorders⁷. These observations suggest that altered sleep/wake schedules may disrupt endogenous circadian clock systems through the mistimed resetting of certain time givers, the so called Zeitgebers. The external LD cycle is the most dominant Zeitgeber to reset the central circadian pacemaker located in the hypothalamic suprachiasmatic nucleus (SCN). The SCN receives light information from the retina and generates central clock signals upon light stimulation to drive slave oscillators distributed in various peripheral tissues through behavioral and neuroendocrine transmitters^1,8. Unsurprisingly, the SCN itself is only responsive to photic timing signals⁸. However, feeding rhythms and nutritional signals play a key role in the regulation of peripheral clock function. Time-restricted feeding (e.g. food access is limited within the normal rest phase of an organism) uncouples peripheral clocks from the SCN for adaptation to the new timing of food availability⁹. This uncoupling might importantly contribute importantly to the pathogenesis of metabolic diseases found in people with abnormal life routine styles¹⁰. Animal studies also demonstrated that food intake during the normal rest phase promotes obesity in rodents¹¹. However, the mechanisms of food-dependent peripheral clock resetting has not been thoroughly elucidated to date. Specifically, as the largest metabolic organ, the liver sensitively responds to the food signals and secrets hepatokines, leading to the robust regulation of metabolic and clock processes^12–14. For example, fibroblast growth factor (FGF) 21 is a liver-secreted, insulin-responsive factor that regulates glucose and lipid metabolism¹⁵. The overexpression of FGF21 in mice increases systemic glucocorticoid levels, suppresses physical activity and alters circadian behavior. However, mice lacking the gene encoding β-Klotho (Klb) in the SCN are refractory to these effects, suggesting that the diverse physiologic and pharmacologic actions of FGF21 are still mediated by the nervous system^16–18. In contrast, the entrainment of peripheral clocks is independent on the SCN, and complex foods show much stronger effects on the phase resetting compared to carbohydrate intake, indicating that other factors must be involved¹⁹. These remaining questions prompt us to identify hepatokines that mediate the food-driven resetting of the liver clock, independent of the master clock.

To answer these questions, we cluster high-throughput RNA sequencing results from liver samples collected in mice subjected to an overnight fasting or constant darkness (DD) and find that the hepatokine Angptl8 shows daily fluctuation in both the liver and serum of mice. Angptl8 (also known as betatrophin, Lipasin, and C19orf80) belongs to the angiopoietin-like protein family, which includes 8 proteins structurally similar to the angiopoietins and are involved in the regulation of lipid metabolism, inflammation, cancer cell invasion, and hematopoietic stem activity^20–23. While Angptl8 is highly enriched in the liver and adipocytes of mice, this protein is exclusively expressed in the liver of humans^24,25. Angptl8 has been shown to play critical roles in the development of various metabolic diseases. For instance, the serum levels of Angptl8 protein are increased in type 1/2 diabetic patients, and are also positively correlated with total cholesterol (TC), LDL cholesterol and apolipoprotein B levels in morbidly obese subjects and individuals with type 2 diabetes^26,27. Upon administration of a high-fat diet, Angptl8 mRNA is induced in the liver and brown and white adipose tissues of mice. Consistently, Angptl8 overexpression by adenoviruses in mice increases serum triglyceride (TG) levels²⁴. In addition to the metabolic functions of Angptl8, a recent study reports that this factor is rhythmically expressed in the mouse liver under the control of glucocorticoid signaling and LXRα pathway, raising the possibility that Angptl8 is involved in the regulation of diurnal oscillation of metabolic processes²⁸.

In the present study, we find that intraperitoneal (i.p.) injection of recombinant Angptl8 in mice alters diurnal rhythms of locomotor activity, as well as hepatic clock and metabolic genes. In addition, Angptl8-shock induces the rhythmic expression of clock genes in mouse Hepa1c1c-7 hepatoma cells. Mechanistically, the resetting function of Angptl8 is dependent on the signal relay of the membrane receptor PirB, phosphorylation of kinases and transcriptional factors, and consequently transient activation of the central clock gene Per1. On the other hand, as the liver is the major site for Angptl8 expression and secretion, we construct mice with liver-specific Angptl8 knockdown by using adeno-associated virus serotype 8 (AAV8) system harboring an Angptl8 shRNA. We find that liver-specific Angptl8 knockdown partially attenuates food-entrained resetting of liver clock in mice. Similar results are observed when circulating Angptl8 is quenched by Angptl8-neutralizd antibody. Thus, Angptl8 is a key regulator of the liver clock in response to food.

Food intake is an important Zeitgeber for the entrainment of circadian clocks in peripheral tissues, while the SCN clock is largely controlled by the LD cycle⁴³. The sensitivity of peripheral clocks to food signals instead of light may help mammals more reasonably adapt to the environment, because the availability of food in nature does not always coincide with other external factors^44,45. However, the uncoupling of peripheral clocks from the central clock when eating occurs during the resting phase leads to internal circadian desynchrony, which is considered a risk factor to induce metabolic disorders¹⁰. Among peripheral clock systems, the liver clock is one of the most vital clock systems that can be reset by numerous food-related metabolites, such as glucose, insulin, and oxyntomodulin^39,41,46. However, the food entrainment of the liver clock does not fully depend on these factors, and the global regulation of food-induced liver clock resetting has not been investigated. In the present study, we identified the secreting hepatokine Angptl8 as a potential direct link between food intake and hepatic clock resetting and metabolic gene transcription. Angptl8 was rhythmically expressed in the liver and induced by refeeding after overnight fasting. The injection of Angptl8 purified protein in mice at ZT2 advanced the oscillation phases of major clock and metabolic genes in the liver, while the blockade of Angptl8 in refed mice by either AAV-mediated shRNA delivery or a neutralized antibody caused the changes of clock and metabolic gene oscillation toward that occuring in the fasted cohort. Mechanically, Angptl8 bound to PirB receptor at the cell membrane and acutely induced Per1 and Per2 expression at the transcriptional level. More importantly, the inhibition of Angptl8 signaling blocked the food-induced entrainment of the liver clock (Fig. 4f). Collectively, these findings add important information on molecular mechanism underlying food entrainment of the liver clock.

Angptl8 is a versatile secreted protein and plays important roles in diverse physiological processes. It is critical to select an appropriate animal model to elucidate tissue-specific functions of Angptl8. Angptl8 KO mice have been generated for more than 5 years⁴⁷, and potentially become a candidate to establish the tool platform in our present study. However, it does not escape from our notice that Angptl8 KO mice manifest various metabolic phenotypes, such as the significant decrease in the fat mass at the age of 12 weeks⁴⁷. The changes in adipose tissues, a critical endocrine organ, will lead to the alteration of circulating adipokine levels, which has a potential impact on the homeostasis of liver clock (e.g. adiponectin induces phase advance of the liver clock in KK/Ta mice)⁴⁸. In addition, the physiological roles of Angptl8 in the central nervous system (CNS) are still unknown. In our study, although injection of Angptl8 in LD cycles indeed shortened the circadian period, its effects on the period were relatively minor and may be secondary due to other metabolites which were produced in response to Angptl8 injection and were able to penetrate blood-brain barrier. Therefore, we could not exclude the possibility that the whole-body knockout of Angptl8 may influence the CNS and/or cause compensatory changes in mice, and indirectly regulates the liver clock. Therefore, although Angptl8 KO mice may serve as a feasible animal model in the present study, it is not a clean setting and possibly mask the direct regulation of liver clock by Angptl8. To overcome this problem and to authentically identify the liver-specific role of Angptl8 in resetting the peripheral clock, we constructed mice with liver-specific Angptl8 knockdown by using the AAV8 system harboring an Angptl8 shRNA, as well as quenched the circulating Angptl8 by using an Angptl8-neutrlized antibody in mice.

As acute responsive genes, Per1/2 play an important role in the entrainment of circadian clock systems. The transient induction of Per1/2 is involved in resetting the circadian clock in various tissues, including the SCN^49,50. Per1/2 expression was also upregulated in fibroblasts and hepatoma cells treated with DEX and serum shock^32,33. In addition, Per1/2 were critical in mediating the resetting of the SCN clock induced by the nonphotic Zeitgeber, melatonin⁵¹. Consistent with these findings, both Per1/2 were induced after Angptl8 treatment, while Per1 induction was more robustly induced in vivo, suggesting that additional mechanisms may finely and differently regulate Per1 and Per2. Similarly, refeeding also sensitively induces hepatic Per1/2 expression, which has been confirmed as an inalienable part of the food-evoked clock resetting mechanism. Furthermore, either AAV-mediated liver-specific Angptl8 knockdown or blockade of Angptl8 in circulation by using a neutralized antibody in refed mice after overnight fasting suppressed the upregulation of hepatic Per1/2 expression. Thus, we conclude that food intake stimulates Angptl8 secretion and increases circulating levels of Angptl8, which resets the liver clock via the induction of the core clock genes Per1/2, and Per1 functions more dominantly than Per2.

Given that Angptl8 is a secreted that is unable to penetrate cell membrane and regulate clock machinery, which is primarily localized in nuclei, the next important question is: how does Angptl8 deliver resetting signals through the cell membrane? Previous studies have demonstrated that the human LILRB2 and its mouse ortholog PirB are receptors for several angiopoietin-like proteins^36,52. Specifically, Angptl8 reverses established cardiomyopathy by PirB receptor activation on the cell membrane of resident adult cardiac progenitor cells³⁷. Therefore, we hypothesized that there is also a similar interaction between Angptl8 and PirB in hepatocytes. Indeed, although Angptl8 did not influence the expression of PirB in mouse Hepa1c1c-7 cells, Co-IP assays confirmed that these two factors physically bind in vivo and colocalize at the cell membrane. More importantly, the knockdown of PirB impaired Angptl8-induced Per1 upregulation and dampened clock gene oscillation. These data clearly suggest that PirB, as a membrane receptor, at least partially, if not totally, mediates the resetting activity of Angptl8 in the liver clock. Notably, two membrane proteins, in addition to PirB, can also bind to Angptl8. One protein, Marco, is expressed in macrophages found in the pancreas and regulates lipid metabolism. The other protein, RTN4R, is a neuronal receptor⁵³. Whether these two receptors are also involved in the resetting function of Angptl8 remains unknown.

Since Angptl8 increases Per1 expression and the signal transduction is dependent on the membrane receptor PirB, it is necessary to elucidate the molecular mechanisms by which the Angptl8-PirB axis regulates Per1 gene expression in nucleus. Previous studies have indicated that Angptl8 increases the insulin sensitivity in HepG2 cells through triggering the phosphorylation of AKT and GSK-3β proteins⁵⁴. On the other hand, PirB, acting as a receptor, responds to the external inflammatory signals or bacterial stimulation and regulates the phosphorylation of downstream MAPK and NF-κB factors through recruitment of SH2 containing phosphatases (e.g. SHP-1/2)^55–58. Notably, these factors are intensively involved in the regulation of Per1 expression via various transcription factors including c-Fos, FosB and c-Jun^33,59. These findings strongly suggest that various kinases (MAPK and AKT), as well as transcriptional factors (e.g. NF-κB), may mediate the regulation of Per1 expression by the Angptl8-PirB axis. Therefore, in the present study, we examined the changes of these kinases and transcriptional factors in response to Angptl8. Our results revealed that Angptl8-induced the phosphorylation of P38 MAPK, NF-κB, and AKT mediated its action on Per1 expression. More importantly, the knockdown of PirB impaired Angptl8-induced such phosphorylation. Co-IP analysis indicated that the Angptl8 decreased the auto-phosphorylation of PirB at tyrosine sites and reducing the recruitments of SHP1/2. Taken together, our experiments demonstrated that Angptl8 induced the phosphorylation of P38 MAPK, NF-κB, and AKT through decreasing the auto-phosphorylation of PirB at tyrosine sites and reducing the recruitments of SHP1/2, and finally activated Per1 expression.

Angptl8 is considered a novel but atypical member in the Angptl family because of its unique structure and crucial effects on lipid metabolism^24,60. Additionally, Angptl8 participates in the pathogenesis of metabolic disorders, such as non-alcoholic fatty liver disease and diabetes^61,62. The present study also revealed an unexpected role for Angptl8 in the resetting of the liver clock. Taken together, Angptl8 may serve as a critical integrator in coordinating the circadian clock and energy metabolism in the liver. At the metabolic level, recent studies have implied an essential role of Angptl8 in the regulation of TG homeostasis and VLDL secretion during the fasting-refeeding transition. Mechanistically, refeeding signals trigger the liver X receptor α (LXRα) and RXR to form a heterodimer, and further activating the transcription of Angptl8 through binding to the LXR response elements on its proximal promoter. In contrast, fasting signals induce the accumulation of glucocorticoids in the circulating system. This accumulation activates and recruits glucocorticoid receptor (GR) dimmers to the nGRE located on the Angptl8 promoter region, thus inhibiting Angptl8 transcription²⁸. Coincided with these findings, our present study indicated that the refeeding robustly increased Angptl8 mRNA expression in 10 min (Supplementary Fig. 8), suggesting a rapid turnover of Angptl8 mRNA in response to the fasting and refeeding transition. Indeed, the half-life of Angptl8 mRNA is only 15.7 min and such an instability is important for its oscillation and function²⁸.

Compared with other known integrators of clock and metabolism (such as PGC-1α and PPAR family members), we propose that the administration of Angptl8 has the following advantages: (1) Angptl8 has no effect on the SCN master clock. In vivo injection of purified Angptl8 protein did not change the oscillation of core clock genes in the SCN, suggesting that Angptl8 functions exclusively in the periphery, leaving SCN untouched. (2) We carefully titrated the doses of Angptl8 in the in vivo injection and found that a low dose (1 μg/kg) of Angptl8 showed a modest effect on systemic metabolism but sufficiently regulated the liver clock. In contrast, previous studies on the roles of metabolic regulation of Angptl8 preferred using a adenovirus-mediated gain-of-function strategy, which led to an artificially high level of Angptl8²⁸. Only in such cases can Angptl8 efficiently function in metabolic regulation. The difference in the window of manipulating Angptl8 levels when targeting clock and metabolism will enable flexibility in the modulation of these two pathways in a separate or combined manner. (3) As a secreted protein, Angptl8 is easily blocked by using a neutralized antibody. Thus, the modulation of Angptl8 is more convenient than that of nuclear factors.

Last but not least, our data indicated that Angptl8 is predominately secreted by the liver, and regulates the liver clock in turn, suggesting it is possible that this circulating hepatokine may transmit time information among different liver cells, such as hepatocytes, hepatic satellite cells and Kupffer cells. However, the physiological roles of Angptl8 in HSC and Kupffer cells currently remain unknown and is interesting to be investigated in the future study. On the other hand, it should be noted that Angptl8 is a secreted protein and can be delivered to various peripheral tissues through circulation. This factor also responds robustly to the external stimuli, in particular nutritional signals. Therefore, we believe that evaluation of the physiological functions of Angptl8 at the organ/tissue level (when specifically considering that the resetting of peripheral clock is largely dependent on the hormone cues balanced by the organ crosstalk) gets higher priority than that at the intra-tissue level.

In conclusion, we provided the evidence that Angptl8 is induced in circulation and in the liver in response to food intake. Functionally, this protein mediates the resetting of the liver clock by food through the PirB/Per1 axis. These interesting findings suggest that Angptl8 is a hepatokine in the orchestration of the liver clock, in addition to its diverse effects on metabolic regulation. Although other factors, such as insulin and glucocorticoids, also have similar actions, targeting these proteins may cause severe side-effects and clinical risks^63–65. The present study provides a safer and more convenient drug target to treat metabolic disorders arising from circadian clock disruption from the view of chronotherapy.