STATE-OF-THE-ART REVIEW IN ENDOCRINOLOGY
Unraveling the mysteries of hepatic insulin signaling: deconvoluting the nuclear targets of insulin
2023 Volume 70 Issue 9 Pages 851-866
Details
2023 Volume 70 Issue 9 Pages 851-866
Over 100 years have passed since insulin was first administered to a diabetic patient. Since then great strides have been made in diabetes research. It has determined where insulin is secreted from, which organs it acts on, how it is transferred into the cell and is delivered to the nucleus, how it orchestrates the expression pattern of the genes, and how it works with each organ to maintain systemic metabolism. Any breakdown in this system leads to diabetes. Thanks to the numerous researchers who have dedicated their lives to cure diabetes, we now know that there are three major organs where insulin acts to maintain glucose/lipid metabolism: the liver, muscles, and fat. The failure of insulin action on these organs, such as insulin resistance, result in hyperglycemia and/or dyslipidemia. The primary trigger of this condition and its association among these tissues still remain to be uncovered. Among the major organs, the liver finely tunes the glucose/lipid metabolism to maintain metabolic flexibility, and plays a crucial role in glucose/lipid abnormality due to insulin resistance. Insulin resistance disrupts this tuning, and selective insulin resistance arises. The glucose metabolism loses its sensitivity to insulin, while the lipid metabolism maintains it. The clarification of its mechanism is warranted to reverse the metabolic abnormalities due to insulin resistance. This review will provide a brief historical review for the progress of the pathophysiology of diabetes since the discovery of insulin, followed by a review of the current research clarifying our understanding of selective insulin resistance.
More than 100 years have passed since insulin was first identified and administered to a patient. Until this great discovery, for nearly 3,500 years, from the first description of diabetes in human history in 1552 B.C., type 1 diabetes was a lethal disease threatening our lives.
The year 1922 was the historical moment that transformed diabetes from an acute fatal disease to a medically manageable chronic one. This year prompted more researchers to devote themselves to elucidating the pathogenesis of diabetes to save more patients. Subsequently, to address various new questions that arose from the clinical treatment of patients, researchers have taken approaches that involve structural analysis of insulin, molecular biology, physiology, and genetics. All their research activities have contributed to the progress of science itself. In this review, we will provide a historical overview of how the pathophysiology of diabetes has been elucidated since the discovery of insulin, and review the current findings on insulin action, particularly focusing on selective insulin resistance. Finally, we will discuss the future direction of this research field.
Before the discovery of insulin, type 1 diabetes was a disease that led to death within a few years from coma caused by ketoacidosis, and starvation therapy was the only available treatment. It was not until 1889 that the cause of this devastating and deadly disease, diabetes mellitus, was identified in the pancreas [1]. In 1916, in London, the English physiologist, Dr. Edward Albert Schafer (Sir Edward Albert Sharpey-Schafer), found that the pancreatic islets secreted a factor that regulated blood glucose and named it insuline, from the Latin “insula” for “island,” after the islets of Langerhans, from which spelling the final “e” was later dropped [2]. In 1909, the Belgian physiologist, Jean De Meyer, made a similar report [3]. In May 1921, Frederick Banting, then a surgeon, and Charles Best, then a medical student, were working at Prof. John J. R. Macleod’s lab at the University of Toronto, when they identified insulin. Banting ligated the canine pancreatic duct, atrophied and degenerated the exocrine cells, and removed the degenerated pancreas to obtain the pancreatic extract. When they administered it to a diabetic dog, they confirmed that hyperglycemia was normalized, and named the extract “isletin.” In December, they succeeded in extracting isletin using alcohol extraction. Isletin was later named insulin by Dr. Macleod and has been used to date. The following year, in January 1922, insulin was administered to a 14-year-old boy with type 1 diabetes for the first time in human history. This was the moment that changed the fate of countless patients with type 1 diabetes who had been dying helplessly.
The development of injectable insulin products made it possible to deliver insulin to a large number of patients and the prognosis of diabetic patients dramatically improved. Subsequently, clinical progress accelerated diabetes research with new insights from patients living with diabetes. It was found that diabetic patients were a mixture of two types: those who required insulin and those in whom it was less effective [4-6]. Younger patients were found to be more hypoglycemic with smaller doses of insulin than older or obese patients [7, 8]. In the 1950s, the development of a technique for measuring insulin [9-11] made it possible to distinguish between patients with insulin deficiency and those with insulin excess, namely, insulin resistance. In the 1970s, the glucose clamp technique was established, to accurately measure the rate at which insulin causes each tissue to dispose of blood glucose [12-15]. Studies clearly demonstrated that the glucose metabolism was strictly regulated by a feedback loop between insulin secretion from pancreatic β-cells and insulin action on peripheral tissues. In other words, decreased insulin sensitivity is strongly associated with increased secretion of insulin. In the 1990s, by using the glucose clamp technique, the natural history of diabetes was investigated to trace the two pathological states of impaired insulin action and deficient insulin secretion from the beginning [16-19]. The results showed that insulin sensitivity was decreased for more than 10 years before the diagnosis of type 2 diabetes. At this initial point, increased pancreatic β-cell function and increased insulin secretion occurred, which compensated for the deficient insulin action, and blood glucose levels were maintained within the normal range. However, pancreatic β-cell failure due to oxidative stress, endoplasmic reticulum stress (ER stress), lipotoxicity, or dedifferentiation leads to chronic hyperglycemia. Pancreatic β-cell failure has been observed to occur approximately 3 years before the diagnosis of type 2 diabetes. In addition, when diabetes was diagnosed, pancreatic β-cell mass decreased to more than half of non-diabetes individuals [20-24]. In addition, there are ethnic differences in insulin secretion capacity, with Asians developing diabetes before insulin resistance progresses to the same level as people in Western countries [23]. In daily practice, since insulin is not measured in normoglycemic patients, in many cases the diagnosis of insulin resistance is not made until the diabetic state has been reached, delaying the diagnosis in most cases.
Under physiological conditions, insulin acts on peripheral tissues to tightly maintain blood glucose levels within a certain range. Pancreatic β-cells cease insulin secretion during fasting and rapidly initiate it in response to feeding. The major target organs of insulin to maintain blood glucose levels are the liver, fat, and muscles (Fig. 1). During the first few hours after fasting, insulin secretion is suppressed, and glycogenolysis (glycogen breakdown) occurs in the liver in combination with glucagon secreted from the pancreas. Subsequently, glucose-producing substrates (mostly amino acids derived from the muscles, predominantly glycerol, and partially fatty acids from the adipocytes) are delivered to the liver for gluconeogenesis (de novo glucose synthesis). When glucose-producing substrates are depleted by continued fasting, lipolysis occurs in the adipocytes, and fatty acids are oxidized in the liver (fatty acid oxidation: FAO). Acetyl-CoA, produced from FAO, is used for ketogenesis (ketone body synthesis from acetyl-CoA). Ketone bodies are then supplied to the whole body as the major energy source [25, 26]. In contrast, when feeding stimulates insulin secretion, the opposite reaction proceeds. In the liver, glycogenolysis and gluconeogenesis are suppressed and glucose uptake is increased. In the adipocytes, lipolysis is suppressed and glucose uptake is increased together with that in the muscle. The glucose taken into the liver is used for energy production by the glycolytic system and the citric acid cycle. The by-products of peripheral glycolysis, such as lactate, amino acid, and glycerol, are mainly used as a major source for lipogenesis (de novo lipogenesis). If in excess, the by-products are used as substrates for glucose production and stored as glycogen. Glucose taken up by the adipocytes are mainly used for the synthesis of glycerol 3-phosphate and triglyceride (TG), its long-term energy storage form.
Metabolic regulation for nutritional homeostasis during fasted and fed conditions
Schema of nutritional interplay among the three major insulin target organs (liver, muscle, and fat). Upper and lower panels show fasted and fed conditions, respectively.
[Fasted] Prolonged fasting enhances hepatic gluconeogenesis and glycogenolysis, increasing glucose output. Lactate and amino acids from muscle and glycerol processed through lipolysis in adipocytes are utilized for hepatic gluconeogenesis. The G6pc catalytic subunit, encoded by G6PC gene, is a crucial enzyme functioning in gluconeogenesis and glycogenolysis. If fasting continues, ketogenesis through fatty acid oxidation (FAO) occurs. Ketone bodies can be processed as fuel for the heart, muscles, kidney, and brain. Free fatty acids (FFAs) are processed through lipolysis, and some amino acids are provided for ketone body synthesis.
[Fed] Nutritional intake provokes insulin secretion from pancreatic β-cells. Insulin inhibits G6PC expression and increases GCK expression, suppressing gluconeogenesis and glycogenolysis and decreasing glucose output. In the process of glycolysis, glucose, lactate, amino acids, and glycerol from peripheral tissues are provided as the primary source for triglyceride synthesis and incorporated into VLDL to deliver lipids to the peripheral tissues. In parallel, insulin promotes glucose uptake in muscle and adipocytes. Glucose is stored as glycogen in muscle. Adipocytes use glucose to generate glycerol 3-phosphate for triglyceride synthesis, a long-term energy storage form. Insulin promotes glycogen synthesis in the liver and muscle, glucose uptake in muscle and adipose tissue, and de novo lipogenesis in the liver and adipose tissue, and suppresses gluconeogenesis in the liver. FAO: fatty acid oxidation, FFA: free fatty acid, TG: triglyceride, VLDL: very low-density lipoprotein, TCA cycle: tricarboxylic acid (TCA) cycle. Created with BioRender.com.
The system is sophisticated enough to maintain blood glucose levels in a narrow range during the dynamic nutritional changes in daily eating behavior. However, when insulin fails to act properly in the three target organs (i.e., liver, fat, and muscles), insulin resistance occurs, and when blood glucose levels elevate chronically, this leads to the development of diabetes mellitus. The concept of insulin resistance was conceived with the finding that the dose of insulin required to correct hyperglycemia in patients with type 2 diabetes is clearly different for each patient [27]. Insulin resistance is diagnosed when the amount of insulin to maintain normoglycemia is greater than expected. The degree of insulin resistance in each organ can be accurately assessed by using labeled glucose (radioisotope of 3-3H-glucose). The rate of glucose production in the liver and the rate of glucose disposal throughout the body (mainly skeletal muscle) are used [28]. In addition, if tissue biopsies are taken, insulin resistance indices of skeletal muscle and adipose tissue, which reflects lipolysis, can also be evaluated separately [28, 29]. However, due to the complexity of the procedure, a more convenient method of assessing insulin resistance in each tissue was warranted. One of the findings indicated that the rise in blood glucose during the first 30 minutes after oral glucose tolerance test (OGTT) is mainly due to suppression of hepatic glucose production, which does not change significantly after 60 minutes [28]. Since more than 80% of the endogenous glucose production is of hepatic origin [30], it is likely that the first 30 minutes in OGTT mainly reflects hepatic glucose production, while the decrease in blood glucose at 60 minutes is due to the predominance of glucose uptake in peripheral tissues, mainly in the muscles. The idea has been devised to distinguish between those cases with insulin resistance dominant in the liver from those with insulin resistance dominant in the muscles using OGTT [28, 31, 32]. The finding was extended to determine individualized diet strategy to improve insulin resistance [33]. A series of research findings support the existence of different pathogenesis of insulin resistance in different tissues and demonstrate the diversity of the pathogenesis of insulin resistance.
Insulin resistance leads to hyperinsulinemia, and hyperinsulinemia leads to insulin resistance [34, 35]. Which is the primary event? Hyperinsulinemia as a consequence of insulin resistance, or vice versa [36-39]? To consider this question, it seems good to refer to the insulin resistance in obese patients, as it is well recognized that obesity, hyperinsulinemia, and insulin resistance are at least mutually aggravating factors. In this section, we will look at the pathological consequence of insulin resistance in obesity.
Obesity is defined as abnormal and/or excessive fat accumulation which poses a risk to health. The accumulation of fat is considered to contribute to insulin resistance because of its positive relationship to it [40]. Since fat accumulation includes visceral and subcutaneous fat, and subcutaneous fat accumulates more readily than visceral fat, it is likely that subcutaneous fat is protective against insulin resistance and associated cardio-metabolic disorder, at least in the early stages [41]. In contrast, visceral fat accumulation increases the risk of cardio-metabolic disorder and increases mortality from cardiovascular disease. This is true even in non-obese individuals with BMI <25 [42-46]. Patients with lipodystrophy, who have reduced or absent fat accumulation in adipose tissue, have ectopic fat (i.e., fat deposits in non-adipose tissue) but demonstrate strong insulin resistance [47, 48]. In contrast to those in whom loss of adipose tissue exacerbates insulin resistance, there is a certain population of obese patients who do not have cardio-metabolic disorder, recognized as Metabolically Healthy Obesity (MHO). MHOs show a low insulin resistance with dominant fat distribution in subcutaneous tissue compared to visceral tissue. Epidemiological evidence suggests that MHOs are associated with less cardiovascular disease and a better prognosis compared to Metabolically Unhealthy Obesity (MUO) [49-51]. Recent studies have further shown that a limited capacity to retain TGs by adipocytes leads to ectopic fat accumulation and lipotoxicity in the liver and muscles, resulting in insulin resistance [52]. These facts indicate that visceral fat deposition leads to insulin resistance, but that obesity does not necessarily lead to insulin resistance in a uniform manner and may work as an insulin resistance protective response at some stage in the development of the pathophysiology of the disease.
Lipid synthesis and lipolysis are mainly regulated by the interplay between the liver and adipocytes. In the liver, insulin signaling activates a pathway that synthesizes fatty acids from glucose and amino acids (de novo lipogenesis), ultimately packing TGs into very low-density lipoproteins (VLDL) for uptake into the peripheral tissues [53]. Hyperinsulinemia enhances this pathway and exacerbates fat deposition in the liver. Chronic hyperinsulinemia then attenuates hepatic and adipose insulin signaling. In the liver, the inhibitory effect on gluconeogenesis is diminished. In adipose tissue, the inhibitory effect on lipolysis is weakened, and free fatty acids (FFAs) are supplied to the liver as a result of lipolysis, where they are broken down into glycerol and used as a substrate for gluconeogenesis. Thus, insulin resistance directly or indirectly increases hepatic glucose production [54-56]. Insulin signaling particularly attenuates its inhibitory effect on gluconeogenesis, but not lipogenesis, which is rather enhanced by hyper-insulinemia, known as selective insulin resistance in glucose metabolism (see section 9 [53, 57]). Increased hepatic lipogenesis deteriorates hepatic insulin resistance itself, and liver-derived TGs supplied to peripheral tissues result in peripheral fat deposition, exacerbating systemic insulin resistance. Increased insulin resistance further promotes insulin secretion from pancreatic β-cells. The failure to secrete sufficient insulin to overcome insulin sensitivity leads to chronic hyperglycemia, type 2 diabetes mellitus. In the scenario described here, the primary trigger is assumed to be hyperinsulinemia. What stimulates insulin secretion from pancreatic β-cells at this time could be a mild glucose intolerance that does not lead to a diagnosis of diabetes, or high nutritional intake, especially FFAs, which stimulates insulin secretion from the pancreatic β-cells [58-61]. In reality, hyperinsulinemia and insulin resistance have a mutually aggravating relationship, and it is difficult to examine them separately. It is fair to consider the possibility that they are not mutually exclusive, and may act in parallel [62].
To understand the pathogenesis of insulin resistance, we propose that the three questions be addressed: 1) Is the glucose metabolism in each tissue insulin-dependent? 2) In what order does insulin resistance occur in the liver, muscle, and fat? 3) Does insulin resistance in one tissue affect insulin resistance in another? The answers may help to build up therapeutic strategies to prevent the primary trigger and the development of insulin resistance.
Insulin resistance leads to chronic hyperglycemia and triggers the onset of diabetes. To what extent can insulin sensitivity explain glucose metabolism in insulin target organs? Research results using the glucose clamp technique have confirmed that in obese patients, regardless of insulin resistance status, hepatic gluconeogenesis can be suppressed to non-obese levels simply by increasing insulin doses [63]. In contrast, as insulin resistance progresses, insulin-independent effects predominate in the glucose uptake of peripheral tissues of muscle and adipocytes. In obese patients, a plateau is reached when glucose uptake is half that of non-obese patients, even with increased insulin doses. The mechanisms leading to insulin resistance differ between the liver, muscle, and adipocytes (Fig. 2).
Differential impact of insulin signaling on glucose handling among liver vs. muscle and adipose tissue
In the different responsiveness to insulin, the black curve shows the response to insulin in healthy individuals and the red curve that in obese individuals with insulin resistance. The curves presented are based on human clamp study [63]. In the liver, glucose production is suppressed in response to insulin stimulation. The obese patients showed a right shift in response, indicating reduced sensitivity to insulin. A high enough dose can suppress hepatic glucose production in obese patients to the same levels as healthy individuals. Therefore, hepatic insulin resistance can be overcome by high-dose insulin. In contrast, glucose uptake in muscle and adipose tissue in obese patients cannot reach the maximal biological response seen in healthy individuals. Therefore, they showed low responsiveness to insulin as well as reduced sensitivity. The results indicated that glucose uptake in obese patients could not be reversed by high insulin due to post-receptor defects.
Studies in mice have revealed the order in which insulin resistance occurs in liver, muscle, and adipocytes. A physiological animal model of insulin resistance is the high-fat diet mouse model [64-71]. One study has evaluated chronological changes in glucose production and glucose uptake in liver, muscle, and adipose tissue after feeding mice on a high-fat diet evaluated by glucose clamp technique [69]. Glucose intolerance was confirmed three days after feeding high-fat diet, and did not change until 12 weeks. At 1 week after feeding, significant insulin resistance was observed in liver and fat, but the liver has the crucial role to maintain systemic insulin resistance. The muscles showed insulin resistance approximately 3 weeks after feeding, and inflammation was seen in the adipocytes at 16 weeks, but this did not affect systemic insulin resistance. Therefore, insulin resistance in the liver and fat precedes that in the muscle. Since a mouse model in which muscle and fat predominantly lack insulin receptor function does not lead to a diabetic state [72], hepatic insulin resistance may play the major role in effecting glucose metabolic abnormality when systemic insulin resistance develops (see detailed discussion in section 5). To date, however, no studies have clarified whether this process is true in humans. For example, human glucose clamp studies have suggested that the majority of glucose uptake is due to muscle, with adipocytes accounting for only about 10% of the effect [28, 73, 74].
Does insulin resistance in one tissue affect the others? The results of studies using tissue-specific insulin receptor (IR) knockout (KO) mice provided crucial insights that address this question (Fig. 3a–c).
Propagation of IR or GLUT4 defective in one tissue to systemic insulin resistance
The metabolic impact of IR or GLUT 4 defect between the tissue involved (rounded red rectangle) and its neighbors (rounded blue rectangle). (a) IR knockout (KO) in liver tissue induced hepatic insulin resistance, resulting in fasting hyperglycemia and decreased glucose uptake in muscle. (b) IR KO in muscle showed glucose uptake decrease but no change in plasma glucose and insulin levels, that is, no insulin resistance. Instead, adipocyte expansion and increased glucose uptake occurred in adipose tissue. (c) IR KO in adipose tissue demonstrated anti-diabetic phenotype, regardless of decreased glucose uptake and increased lipolysis by insulin stimulation. Note that the simultaneous knockdown of IR in muscle and adipose tissues did not induce diabetes. (d) GLUT4 KO in muscle provoked decreased glucose uptake, resulting in insulin resistance, glucose intolerance, and hyperglycemia. This effect spread to the neighbors, as seen in increased hepatic glucose production and reduced glucose uptake in adipose tissue. (e) In contrast to IR KO, GLUT4 KO in adipose tissue provoked insulin resistance and glucose intolerance with reduced glucose uptake. As in muscle GLUT 4 KO, ADIPO-GLUT4 KO leads to increased hepatic glucose production and decreased glucose uptake in adipose tissue. Created with BioRender.com.
Insulin suppresses hepatic glucose production via IRs. In muscle or adipose tissue, insulin activates the glucose transporter type 4 (GLUT4) via IRs to uptake glucose [38]. Insulin resistance results in dysregulation of hepatic glucose production and glucose uptake, leading to the development of type 2 diabetes. Systemic IR KO results in death from ketoacidosis 4 to 5 days after birth. In heterozygous mutation, diabetic risk of the mice whose IR protein levels were less than half of their littermate control was 10% [75]. Hepatocyte specific IR KO (LIRKO) mice showed strong insulin resistance, fasting hyperglycemia, and enlargement of pancreatic β-cells [76]. In the controls, insulin suppressed hepatic glucose production by 80%, but not in LIRKO mice. Furthermore, muscle glucose uptake was reduced and insulin resistance was exacerbated [77]. In muscle specific IR KO (MIRKO) mice, glucose uptake was significantly inhibited, but glucose tolerance was normal at least until 40 weeks of age, and insulin resistance was not observed [78]. However, there was a 3-fold increase in adipocyte glucose uptake concomitant with the expansion of adipocytes. As a result, fat mass, serum TGs, and serum FFAs were increased, in a state also known as metabolic syndrome. Adipocytes are responsible for only 10% of the tissue glucose uptake throughout the body [79]. However, when IR was knocked out in adipocytes (FIRKO), basal glucose uptake was normal, but insulin failed to promote glucose uptake by 90% and failed to suppress lipolysis [80]. As expected, FIRKO mice showed normal glucose tolerance, normal serum TG levels, and weight loss were observed. Furthermore, FIRKO mice were protected against obesity, diabetes, and fatty liver even under drug-induced overeating, and prolonged life-span was observed. It has also been confirmed that substantial attenuation of IR simultaneously in muscle and fat did not lead to a diabetic state [72].
A series of findings led us to the following conclusions: 1) impaired insulin action in one location can spill over into the other tissues, exacerbating systemic insulin resistance; 2) insulin maintains glucose homeostasis largely via suppression of hepatic glucose production rather than an increase in glucose uptake in muscle or adipose tissues; and 3) adipocytes act as a buffer against insulin resistance.
In muscle and adipocytes, glucose uptake by insulin is determined by membrane translocation of GLUT4 protein [81]. Whether glucose uptake in muscle and adipocytes is insulin-IR signaling dependent can be inferred by using tissue-specific GLUT4 KO (Fig. 3d, e). Systemic GLUT4 KO induced growth retardation and reduced fat mass and cardiac muscle mass [82]. Although life expectancy was shorter, diabetes was not observed. Interestingly, heterozygous mutations led to a diabetic state. Muscle GLUT4 KO mice showed significant reduction in insulin stimulating glucose uptake and unlike MIRKO mice, insulin resistance and glucose intolerance were confirmed [83]. They also showed decreased glucose uptake in adipocytes and impaired hepatic glucose production. The abnormal changes were reversible with normalization of blood glucose levels with phloridzin, which is a sodium-glucose cotransporter inhibitor. The finding indicated that insulin-independent pathways of glucose uptake in muscle were important in the development of impaired glucose tolerance, and that hyperglycemia and hyperinsulinemia may lead to hepatic- and adipo-insulin resistance. One example reflecting this result is that exercise-induced glucose uptake contributed to maintain glucose levels even in the presence of insulin resistance. In addition to IRs, the IGF-1/IGF-1 receptor pathway was also known to be an important signal contributing to insulin resistance in muscle [84, 85]. GLUT4 KO in adipocytes decreased glucose uptake in adipocytes. Unlike FIRKO mice, insulin resistance and glucose intolerance were observed and insulin-induced glucose uptake in muscle was reduced, resulting in 50% reduction of total body glucose uptake. The inhibitory effect of insulin on hepatic glucose production was also diminished, and systemic insulin resistance progressed. In muscle and fat, an IR-independent pathway of glucose uptake may play a major role in the development of insulin resistance and diabetes mellitus. In these tissues, the activation of the insulin-independent pathway may alleviate the insulin resistance associated with the impairment of the insulin-IR axis. In the liver, on the other hand, the insulin-IR axis is important for glucose handling. Since insulin resistance in each tissue mutually aggravates one another, it is possible to break the vicious cycle by normalizing hepatic insulin signaling.
So how is insulin signaling in the liver transmitted to the nucleus? Insulin phosphorylates insulin receptor substrate (IRS) proteins via the insulin receptor. The IRS activates two main types of signals: the phosphatidylinositol 3-kinase (PI3K)-Akt/protein kinase B (PKB) pathway and the Ras-mitogen-activated protein kinase (MAPK) pathway. The former is involved in metabolic regulation, while the latter cooperates with the PI3K pathway to regulate cell proliferation and differentiation [86]. One of the reasons insulin signaling is a difficult target for drug discovery is that there are many shared properties between its metabolic and cell growth pathways.
Since excessive insulin-signaling lead to obesity, insulin resistance, and further hyperinsulinemia, insulin targets at more granular level need to be explored to ensure safe and effective, long-term therapies for modulating insulin signaling. To this end, it is necessary to clarify how insulin signaling is integrated and transmitted to the nucleus within a cell, and how to deconvolute its function.
The first milestone to address the question of how Akt signals into the nucleus was laid in 1997 [87]. The study using Caenorhabditis elegans revealed an interaction between DAF-2 and DAF-16, which are a human homologue of insulin and insulin-like growth factor receptors and forkhead transcription factor (FoxO). Mutations reducing DAF-2 activity doubled the life-span and DAF-16 was required for this effect. In 1999, four studies demonstrating the relationship between Akt and FoxO1 followed [88-91]. This was the year FoxO1 was verified as a central mediator of insulin, shifting the focus of insulin signaling research from the cell surface to nuclear signaling. In 2007, liver-specific FoxO1 knockout (LFKO) mice were reported to normalize the hyperinsulinemic hyperglycemia induced by IR KO [92]. Subsequently, from different laboratories at different times, it was reported that glucose intolerance caused by knockout of IRS1/2 [93, 94] and Akt1/2 [95] in hepatocytes was normalized by performing hepatic FoxO1 ablation. A series of research results established the finding that hepatic FoxO1 plays a crucial role in the abnormal glucose metabolism under insulin resistance. Importantly, these findings gave us an attractive idea that, in the liver, insulin may handle metabolic function via simultaneous regulation of gene transcription in the liver, where enzymatic activity was previously thought to play a dominant role. This was truly a significant paradigm shift.
Since hepatic FoxO1 plays an important role in the hepatic glucose metabolism, three possibilities can be proposed to elucidate the global picture of the regulatory mechanism of hepatic glucose production: 1) regulation via nutrients partly derived from other tissues (e.g., muscle-derived amino acids and lipolysis-induced FFA); 2) FoxO1-dependent pathway; and 3) FoxO1-independent pathway.
When insulin secretion is decreased during fasting, Akt is dephosphorylated and the FoxO1 is subsequently dephosphorylated and transferred to the nucleus. Then, a set of glucose/lipid metabolism-related genes are activated or repressed, activating glucose-6-phosphatase (G6PC) and phosphoenolpyruvate carboxykinase 1 (PCK1), which encode key rate-limiting enzymes in glucose production [89, 96-98]. Conversely, phosphorylation by Akt during feeding leads to nuclear exclusion of FoxO1 and glucose production is suppressed. FoxO1 activity thus involves acetylation as well as phosphorylation [97, 99, 100]. Under cellular stress, FoxO1 is deacetylated by Sirt1 and translocated into the nucleus and activated. This action exceeds the nuclear translocation by Akt and enhances gluconeogenesis [101]. It has been shown that the glucagon signal increases G6PC expression via cAMP and dephosphorylation of Class IIa HDACs (HDAC 4, 5, and 7). CaMKII, which is activated by cAMP, inhibits FoxO1 phosphorylation [102]. HDAC4/5 recruits HDAC3 and deacetylates FoxO1 [103]. Both pathways result in increased glucose production. In parallel with the progress in our understanding of the FoxO1-dependent pathway for hepatic glucose production, the FoxO1-independent pathway has also been gradually elucidated. A comprehensive identification of transcription factors (TFs) that bind to the promoter region of G6PC and PCK1 was performed by using mass spectrometry [104]. Among the TFs that respond to insulin and bind to G6PC and PCK1 in common, FoxK1 was identified, which was confirmed to suppress gluconeogenesis. Furthermore, the other TF, Tox4, which binds to PCK1, was found to suppress gluconeogenesis and ameliorate hyperglycemia in obese mice. This was synergistic with FoxO1 and independent of IR-mediated effects [105]. The identification of a FoxO1-independent pathway may lead to the identification of new drug targets in the hepatic glucose production pathway.
In the liver of patients with diabetes, lipid synthesis (especially atherogenic lipoproteins, such as VLDL, TGs, and small dense LDL), as well as glucose production, is increased. In addition to glucose production (reviewed in section 7), FoxO1 has also been involved in VLDL synthesis. It has been shown that FoxO1 leads to increased VLDL synthesis via increased Apo C-III activity and MTP expression [106]. However, the data have been inconsistent with the effect of FoxO1 on hepatic TG accumulation. In a mouse model in which FoxO1 was constitutively active with the mutation in the phosphorylation site by Akt, one study showed increased hepatic TG accumulation and decreased gene expression for lipid synthesis, such as sterol regulatory binding protein-1c (SREBP1c), while another study showed no change in hepatic TG accumulation [107, 108]. In a mouse model in which FoxO1 activity was constitutively active due to the mutation in the deacetylation site, hepatic glucose production was increased and the respiratory quotient in the liver was decreased, suggesting that lipids were being used as the main energy source. Consistent with this result, plasma TG levels were decreased, indicating therefore that the deacetylation of FoxO1 can be seen as a system for adaptation to starvation [109]. FoxO1 promotes lipid utilization in the liver during fasting and may contribute to lipid synthesis when inactivated by feeding.
Although hepatic FoxO1 plays a major role, the isozymes of FoxO3 and –4 are redundant and have been found to act synergistically [110]. Simultaneous KO of FoxO1 and –3 resulted in fatty liver and hypertriglycemia [111]. In addition, in mice with hepatic FoxO1, –3, and –4 KO, the scavenger receptor class B member 1 (SCARB1) and hepatic lipase (LIPC) expression levels were decreased. While the former is involved in the clearance of HDL-cholesterol, and the latter encodes a hepatic lipase, the decreased expression of selective uptake of HDL-cholesterol ester, results in increase in HDL-cholesterol [112]. In addition, FoxOs was identified as a regulator of the ApoM/S1P pathway, which confers beneficial effects of HDL on the cardiovascular system [113]. FoxOs regulate the synthesis of VLDL, an atherogenic lipid, and the peripheral delivery of cholesterol.
In response to intermittent changes in nutritional status, i.e., fasting and refeeding, insulin sharply switches between gluconeogenesis and lipid synthesis in hepatocytes to maintain metabolic state. However, obese type 2 diabetic patients with insulin resistance show hyperglycemia with increased gluconeogenesis due to impaired insulin action, while insulin sensitivity to lipid synthesis is maintained and fat accumulation in the liver is enhanced. That is, selective insulin resistance to gluconeogenesis is here being recognized [53, 57, 114]. Therefore, to cure metabolic disorders due to impaired insulin action requires us to reverse the pathophysiology of selective insulin resistance in the liver.
Although it has not been determined whether this metabolic switch between glucose vs. lipid is regulated at a single-node or multi-nodes, FoxO1 seems to play a role in it downstream of insulin signaling. FoxO1 increased glucose production by upregulating G6PC expression and suppressed glucose utilization by downregulating glucose kinase (GCK) expression [115, 116]. Feeding deactivates insulin to increase GCK expression and promotes the utilization of glucose, resulting in increased lipid synthesis. FoxO1 acts as an activator for G6PC and represses its expression together with its co-repressor SIN3A for GCK [116]. Based on this finding, drug screening was conducted to identify the small molecule which selectively inhibits the function of FoxO1 in the glucose production pathway. The full inhibitor suppressed G6PC and increased GCK, while the selective inhibitor suppressed G6PC and had no effect on GCK. This finding indicated that individual biological responses may individually modulate the critical nodes of insulin signaling and reverse selective insulin resistance.
FoxO1 has an inhibitory effect of insulin on gluconeogenesis [92-95], while FoxO1 is dispensable for insulin-induced lipid synthesis [117]. Based on these findings, a mechanistic insight of selective insulin resistance was illustrated by differential sensitivity of insulin signaling to FoxO1 vs. mTOR-SREBp1c, which is responsible for glucose production and lipid synthesis [57]. Its actual pathophysiology still remains to be uncovered, however, since it is difficult to reproduce in vivo a selective insulin resistant state, that is, a state in which the FoxO1 inactivation pathway is suppressed by insulin but the mTOR pathway is intact [118].
To date, there are some findings that may provide clues to the mechanism of selective insulin resistance. In the liver, IRS1 functions during feeding and IRS2 functions during fasting, affect the regulation of G6PC, PCK1, SREBp1c, and GCK expression, respectively [94]. And only liver-specific IRS2 KO mice were found to have fatty liver. Furthermore, there was an imbalance in the expression distribution of IRS1 and IRS2 between perivascular and periportal areas, and lipogenesis was enhanced in the perivascular areas where IRS1 was maintained. Thus, it was inferred that selective insulin resistance is determined by the distribution of IRS1 and IRS2 expression [119, 120]. However, it remains unclear how distinctively IRS1 and IRS2 are regulated by insulin, and how gene expression is coordinated downstream of IRS1 and IRS2.
In the absence of Akt signaling, activation of mTORC1 alone does not result in enhanced lipid synthesis. However, when FoxO1 was additionally ablated, increased lipogenesis occurred. Furthermore, hepatic FoxO1 KO normalized suppression of lipolysis by insulin in adipocytes and suppressed hepatic glucose production by reducing the influx of adipocyte-derived FFAs into liver [121]. FoxO1 is involved in the regulation of lipogenesis together with mTORC1 downstream of insulin signaling, while FoxO1 suppressed glucose production through direct and indirect insulin action. Based on these results, it is possible that focusing on the target of FoxO1 may shed light on the pathogenesis of selective insulin resistance.
Cistrome analysis (e.g., ChIP-seq) is a useful technique to delve into FoxO1 targets in the nucleus, to get a global, genome-wide view of FoxO1 regulation of glucose and lipid metabolism. Fasting causes several hormonal changes, suppressing insulin secretion and instead promoting pancreas-derived glucagon and adrenal-derived glucocorticoids [122-126]. Insulin suppression leads to activate FoxO1, glucagon phosphorylates CREB [127], and glucocorticoids enables fasting responses via binding to GR [128]. Suppression of insulin signaling in adipocytes promotes lipolysis and elevates plasma FFAs, which act as endogenous ligands for peroxisome proliferator-activated receptor alpha (PPARα) [129], resulting in increased ketone synthesis. Thus, during fasting, multi-hormonal waves are delivered to hepatocytes at once and are orchestrated by the signals to be delivered into the nucleus by their corresponding TFs, such as FoxO1, GR, CREB, and PPARα, to regulate gene expressions [130]. TFs also work cooperatively with each other, to promote the other TFs’ access to the chromatin and propagate TF expression. For example, GR facilitates CREB access to chromatin and increases PPARα expression [131, 132]. Therefore, fasting-inducible TFs cooperate with each other to produce common physiological responses, just as they do at the hormone level.
The first cistrome analysis of FoxO1 revealed only 400 binding regions [133, 134], which is extremely small compared to its wide variety of functions. The cistrome data of the other fast-induced TFs showed many more of them (e.g., CREB 7,000 sites [135, 136], GR 11,000 sites [137, 138], PPARα 18,000 sites [139, 140]). Considering the low expression of hepatic FoxO1 and the lack of antibodies for FoxO1 with sufficient sensitivity for chromatin immuno-precipitation (ChIP), the technical problem had to be carefully examined. Also, the use of prolonged non-physiological fasting conditions made the interpretation of the data difficult in terms of physiological function [134, 141]. We addressed the technical issue by generating a FoxO1 reporter knockin mouse (FoxO1-Venus) and using GFP antibodies for ChIP experiments [142]. By using physiological fasting and refeeding conditions, we were able to show how hepatic FoxO1 behaved in a cell in response to nutritional changes in vivo. FoxO1 ChIP-seq and immunohistochemistry demonstrated 14,000 binding sites during fasting. The data showed that FoxO1 sensitively migrated out of the nucleus upon feeding, and that more than 60% of the binding sites were detached from chromatin. By integrating cistrome and transcriptome analysis of FoxO1, we reached the following conclusions [143] (Fig. 4): (1) The transcriptional logic of FoxO1 recasts the bifurcating model of insulin signaling to an integrative model to interpret the lipid vs. glucose metabolism [57]. We showed that glucose metabolic genes are governed by intergenic and promoter/TSS enhancers, while lipid genes are governed by a bipartite intron logic that includes fasting-dependent intron enhancers and fasting-independent enhancerless introns. (2) Active enhancers of glucose metabolic genes showed transcriptional resiliency, likely through shared PPARα/FoxO1 regulatory elements. (3) Insulin resistance and hyperglycemia result in the spreading of FoxO1 binding to enhancers, resulting in quantitative and qualitative abnormalities of active FoxO1 marks. These results introduced a novel insight into the distinctive regulation taking place in glucose and lipid metabolism and its relevance to the development of insulin resistance.
Distinctive transcriptional logic of glucose vs. lipid regulation in the pathogenesis of insulin resistance
In the transcriptional logic for metabolic regulation by FoxO1, insulin directly phosphorylates FoxO1 to inhibit its activity and indirectly inhibits PPARα by reducing fatty acid supply through suppressing lipolysis in adipose tissue. In normal conditions, FoxO1 is cleared upon refeeding from resilient enhancers, enriched in glucose metabolism genes, but not in introns, and enriched in lipid metabolism genes. With the onset of insulin resistance, compensatory hyperinsulinemia can still clear FoxO1 from resilient enhancers, but not from introns, increasing serum lipoprotein and triglyceride levels. As insulin resistance progresses, compensation by PPARα and spreading of FoxO1 binding to additional sites increases the expression of glucose metabolic genes, leading to fasting hyperglycemia with dyslipidemia. A portion of the illustration is reproduced from a prior study [143]. TG; triglyceride, FFA; free fatty acid.
Inactivation of hepatic FoxO1 increased the sensitivity of direct insulin action to hepatic glucose production after prolonged fasting, as well as the sensitivity of indirect action via lipolysis suppression in adipocytes [121]. PPARα may play a role to reinforce the indirect effect, as adipocyte-derived FFAs work as endogenous ligands of PPARα. Since common binding sites were observed between FoxO1 and PPARα particularly in glucose metabolism genes, the common targets may provide a clue to the pathogenesis of selective insulin resistance. The following questions remain to be solved: 1) What is the functional interplay between FoxO1 and PPARα for glucose-lipid metabolism? 2) Are FoxO1 and PPARα cooperatively or individually bound to the same sites? 3) Do they affect each other’s transcriptional activity? The answers to these questions will bring us one step closer toward the development of a selective insulin sensitizer.
It has been more than 100 years since insulin was identified and first used to save the lives of diabetic patients. The efforts of numerous researchers have greatly advanced our understanding of the pathogenesis of insulin resistance. However, there are still limited therapeutic approaches to cure diabetes. To resolve this deadlock, it is warranted to develop drugs targeting the root of the pathogenesis, or to determine the primary trigger of insulin resistance. While insulin resistance in the liver, muscle, or fat propagates itself among the target organs (e.g., liver, muscle, and fat), as hepatic insulin resistance has a greater impact on hyperglycemia, the liver is a particularly good candidate to reverse metabolic abnormality due to systemic insulin resistance. But among drugs directly targeting hepatic glucose production, only metformin is available, and even that is incapable of achieving a cure [144, 145]. The liver is an organ that precisely switches between glucose and lipid metabolism in accordance with nutritional changes. The discovery of FoxO1 ushered in the potential to normalize the glucose metabolic abnormalities caused by hepatic insulin resistance by improving hepatic glucose production through direct or indirect insulin action on hepatic FoxO1. Genome-wide analysis of FoxO1 revealed a transcriptional logic to explain the distinctive regulation between glucose and lipid metabolism under FoxO1 nodes, and its cooperation with PPARα. The results have made great strides in understanding the pathogenesis of selective insulin resistance. The elucidation of the pathophysiology of selective insulin resistance will lead to further research in three major directions: 1) identification of individuals at high risk of developing insulin resistance; 2) establishment of strategies to prevent the onset of insulin resistance; and 3) development of a drug discovery platform to reverse insulin resistance and cure the disease. We hope that this will lead to realizing a new level of personalized medicine that can cure lifestyle-related diseases.
We wish to express our deep thanks to the members of the Accili lab for their technical support.
T. K. designed and wrote the manuscript. D. A. supervised the manuscript.
This work was supported in part by JSPS KAKENHI Grant No. JP22K16422 to T. K.
The authors declare no competing interests.
