Introduction
Although its definition and aetiology remain controversial, the metabolic syndrome is considered to be the combination of type 2 diabetes, obesity, hyperlipidaemia and hypertension [
1,
2]. Among these disorders, obesity is the most fundamental factor for diagnosis because increased fat cells and macrophages secrete inflammatory cytokines and induce insulin resistance, which is a key feature of the metabolic syndrome [
3]. Obesity is caused by an imbalance between energy intake and energy consumption, which is also termed a failure of energy homeostasis. The most important organ for regulating energy homeostasis is the hypothalamus. The hypothalamus contains two distinct neurone types: one secretes agouti-related protein (AGRP), which increases food intake and reduces energy expenditure, and the other secretes proopiomelanocortin, which decreases food intake and increases energy expenditure [
4,
5].
Recent genetic studies have revealed that insulin signalling in the hypothalamus is implicated in the regulation of energy homeostasis. Dysregulation of insulin receptor substrate 2 (IRS2) in the hypothalamus causes obesity and diabetes [
6,
7]. FoxO1, an isoform of forkhead box-containing protein of the O subfamily, is a downstream effector of insulin signalling and mediates various metabolic functions of insulin [
8]. We previously reported that hypothalamic FoxO1 regulates
Agrp transcription and plays an important role in the regulation of food intake and energy expenditure [
9,
10]. FoxO1 directly binds to the
Agrp promoter, leading to the recruitment of coactivator CBP/P300 and activating
Agrp transcription [
9]. Mice that produced constitutively active FoxO1 specifically in the hypothalamus developed obesity and insulin resistance due to increased food intake and reduced energy expenditure [
10].
Activating transcription factor 3 (ATF3) is a member of the ATF/CREB family of transcription factors.
Atf3 gene expression is induced by a variety of extracellular signals, including cytokines, chemokines, growth factors, hormones, hypoxia, DNA damage, endoplasmic reticulum stress and nutrient deprivation. Although
Atf3 has long been recognised as a stress-response gene, recent cumulative evidence has suggested that ATF3 is involved in more general adaptive responses, such as environmental, emotional and nutritional alterations [
11‐
13]. ATF3 is found in various cell types, and its disorders have been suggested to associate with inflammatory diseases, immune diseases and cancer [
14]. We also previously reported that ATF3 is highly expressed in the vascular endothelial cells of human atherosclerotic lesions and is centrally involved in endothelial cell death, which might be associated with atherogenesis [
15,
16].
To date, a number of studies have been conducted to elucidate the physiological roles of ATF3 in the pancreas. However, there are critical discrepancies in those studies, and the role of ATF3 in the pancreas is still controversial [
17‐
21]. With respect to the role of ATF3 in the brain, it has been reported that ATF3 is expressed in cortical and hippocampal neurones and protects against neuronal cell death, which reduces brain damage after cerebral ischaemia [
22]. ATF3 in the nucleus accumbens is reportedly critical for the regulation of emotional behaviour [
13]. In contrast, although genome-wide profiling of fetal hypothalamic neurones has revealed that
Atf3 is highly expressed in the murine hypothalamus [
23], its physiological role in the hypothalamus remains unknown. To elucidate these important points, we analysed knockout mice in which the ATF3 level was reduced in the pancreas and hypothalamus.
In the present study, we make two important suggestions. First, the physiological function of ATF3 in the pancreas appears to be less significant than previously predicted. Second, ATF3 has a novel, important function in the hypothalamus: hypothalamic ATF3 regulates glucose and energy metabolism by activating Agrp transcription.
Discussion
Regarding double-tissue-specific knockout mice, gene ablation in one tissue may affect the phenotype resulting from gene ablation in another tissue. However, considering that metabolic syndrome is a combined disorder involving both the hypothalamus and the pancreas, the analyses of PHT-
Atf3-KO mice should provide more practical insights into the relationship between ATF3 and the metabolic syndrome. Indeed, hypothalamus- and pancreas-specific IRS2 knockout mice have been used to determine the significance of combined IRS2 dysfunction in both the hypothalamus and the pancreas for the development of obesity and diabetes [
6,
7].
There are several contradictions in the published literature regarding the physiological role of ATF3 in the pancreas, and the critical issue of whether pancreatic ATF3 is beneficial or detrimental to the regulation of glucose homeostasis still remain unsolved [
17]. The observations in this study—normal plasma glucagon and insulin levels (Fig.
3d–g), unchanged glucose-responsive glucagon and insulin secretion (Fig.
3j, k), and normal alpha and beta cell mass (ESM Fig.
2b–d) in PHT-
Atf3-KO mice—suggested that the physiological role of ATF3 in the pancreas may be less significant than previously predicted, although the lower knockout efficiency of ATF3 in the pancreas (Fig.
1b) should also be considered.
Despite the unchanged islet cell morphology and function, we observed a significant decrease in blood glucose level and better glucose tolerance in PHT-
Atf3-KO mice compared with wild-type mice (Fig.
3a–c). Therefore, we next focused on the lean phenotype and higher insulin sensitivity in these mice (Fig.
4a–e). We observed that PHT-
Atf3-KO mice exhibited reduced food intake, which was associated with reduced
Agrp expression in the hypothalamus (Fig.
4g, h,
6a). Although
Agrp knockout mice were originally reported to have normal food intake, body weight and body composition [
33], these mice are known to show reduced body weight and adiposity after 6 months of age [
34]. Furthermore, postembryonic ablation of AGRP neurones leads to a lean phenotype due to decreased food intake and increased
Ucp1 levels in BAT [
35]. We also observed that PHT-
Atf3-KO mice have increased energy expenditure, which is associated with increased
Ppargc1a and
Ucp1 expression and increased sympathetic tone in adipose tissues (Fig.
5a–i). We further elucidated the molecular mechanism underlying these results: hypothalamic ATF3 regulates
Agrp transcription by interacting with FoxO1. To our knowledge, all of the findings presented here regarding the role of ATF3 in the hypothalamus are novel. The limitation of this study is the lack of a pair-feeding experiment, so it is unclear to what extent the phenotype of PHT-
Atf3-KO mice occurred secondary to reduced food intake. However, because the oxygen consumption without normalisation by body weight was still significantly increased in these mice (ESM Fig.
4g, h), higher energy expenditure was not simply due to the reduced food intake or reduced body weight. Another important finding in this study was the induction of
Atf3 expression by low glucose or fasting in a hypothalamic cell line or mouse hypothalamus (Fig.
6b–d). Because ATF3 is an adaptive response gene [
14], hypothalamic ATF3 level is probably regulated by nutrient status. Additional study will be needed to clarify the mechanism that determines how changes in nutrient status regulate
Atf3 expression in the hypothalamus.
Considering that ATF3 works as a transcriptional repressor rather than an activator in most cell types [
36] and that FoxO1 binds to the
Pomc promoter and inhibits its promoter activity [
9], it can be predicted that ATF3 suppresses
Pomc gene transcription, even though the level of
Pomc mRNA was unaltered in PHT-
Atf3-KO mice compared with control mice (Fig.
6a). To test this hypothesis, we performed luciferase assays using the
Pomc promoter, including the FoxO1 binding site [
9]. However, in contrast to the
Agrp promoter, ATF3 did not affect
Pomc promoter activity (data not shown). Therefore, hypothalamic ATF3 regulates food intake and energy expenditure by upregulating
Agrp transcription rather than by downregulating
Pomc transcription.
In the present study, we demonstrated that a reduction of ATF3 in the hypothalamus leads to a lean phenotype, higher insulin sensitivity and better glucose tolerance. Hypothalamic ATF3 may be a useful target for the development of a new strategy to treat or prevent obesity-based metabolic syndrome.
Acknowledgements
We thank C. Osawa for excellent technical assistance and the members of the Kitamura laboratory for discussions of the data.