Knockout of inositol-requiring enzyme 1α in pro-opiomelanocortin neurons decreases fat mass via increasing energy expenditure

Although numerous functions of inositol-requiring enzyme 1α (IRE1α) have been identified, a role of IRE1α in pro-opiomelanocortin (POMC) neurons in the arcuate nucleus of the hypothalamus is largely unknown. Here, we showed that mice lacking IRE1α specifically in POMC neurons (PIKO) are lean and resistant to high-fat diet-induced obesity and obesity-related insulin resistance, liver steatosis and leptin resistance. Furthermore, PIKO mice had higher energy expenditure, probably due to increased thermogenesis in brown adipose tissue. Additionally, α-melanocyte-stimulating hormone production was increased in the hypothalamus of PIKO mice. These results demonstrate that IRE1α in POMC neurons plays a critical role in the regulation of obesity and obesity-related metabolic disorders. Our results also suggest that IRE1α is not only an endoplasmic reticulum stress sensor, but also a new potential therapeutic target for obesity and obesity-related metabolic diseases.


Introduction
Energy balance relies on a tightly regulated homeostatic system matching food intake with energy expenditure [1]. The central nervous system (CNS), especially the hypothalamus, plays a key role in the regulation of energy homeostasis and insulin action [2]. The environmental and internal signals are integrated in the CNS to determine the different metabolic characteristics and phenotypes [3]. Because obesity and type 2 diabetes (T2D) represent an increasing health risk worldwide [4], exploring the nature of this dysfunction is high priority. Epidemiological studies have shown a relationship between dietary fat intake and obesity; therefore, high-fat diet (HFD)-induced obesity has been used commonly for the study of the mechanisms underlying dietary-induced obesity and insulin resistance [5].
In the hypothalamus, neurons expressing pro-opiomelanocortin (POMC), defined as part of the melanocortin system, play an important role in regulating food intake and energy expenditure [6], which is considered a promising target for the treatment of obesity [7].
Inositol-requiring enzyme 1a (IRE1a), as one member of three families of unique endoplasmic reticulum (ER) transmembrane proteins, is evolutionarily conserved in all eukaryotes from unicellular organisms to mammals [8] and widely expressed in brain and other tissues [9]. It contains a kinase domain and an RNase domain in the cytosolic region and is activated during ER adaptive response to stress [8,[10][11][12]. The RNase activity induces splicing of X-box binding protein (XBP) 1 mRNA, which is translated into a functional transcription factor for ER stress control [13]. IRE1a is also activated through autophosphorylation to initiate a key signalling of the mammalian unfolded protein response (UPR) pathways [12,14]. It has been shown that IRE1a is involved in the regulation of many processes, including ER stress, autophagy and adipocyte differentiation [15][16][17]. However, a role of IRE1a in the hypothalamus has not been previously described.
In our current study, we demonstrated that mice lacking IRE1a specifically in POMC neurons are resistant to obesity and obesity-related metabolic disorders, the effects of which were mediated by increased energy expenditure. Interestingly, increased a-melanocyte-stimulating hormone (a-MSH) production was observed in the hypothalamus of POMC neuron-specific IRE1a knockout (PIKO) mice.

Animals and diets
C57BL/6 J wild-type (WT) mice were obtained from the Shanghai Laboratory Animals Co. Ltd (Shanghai, China). To generate PIKO mice, IRE1a floxed mice [18] were crossed with POMC-Cre mice [19]; all studies were performed in male PIKO mice and their littermate control mice. To visualize POMC-Cre positive neurons by immunofluorescence (IF) staining, POMC-Cre mice and PIKO mice were intercrossed with tdTomato reporter/Ai9 mice (The Jackson Laboratory) [20]. All animals were under C57BL/6 J background and housed in laboratory cages at 25 + 28C with a humidity of 35 + 5% under a 12 h dark/light cycle, with free access to water and standard chow diet. For HFD experiments, male pups (n ¼ 8-11) at ages of six weeks were fed a standard chow diet (NCD) or HFD with 60% kcal from fat (Research Diets, NJ, USA) for 16 weeks, then metabolic parameters, leptin sensitivity and glucose metabolism were measured in different cohorts of mice at the ages of 22-23 weeks old.

Metabolic parameter measurements
Indirect calorimetry was measured in a comprehensive laboratory animal monitoring system (CLAMS; Columbus Instruments, OH, USA), as previously described [21]. Rectal temperature of mice was measured at 14.00 (basal metabolic state) using a rectal probe attached to a digital thermometer (Physitemp, NJ, USA). Mice body composition was measured using NMR imaging system (Bruker, Rheinstetten, Germany).

Leptin sensitivity assay
Mice were injected intraperitoneally (i.p.) with either phosphatebuffered saline (PBS) or leptin (3 mg kg 21 , R&D Systems, MN, USA) at 09.00 following fasting for 24 h as previously described [22]. Food intake and body weight were measured at 1 and 4 h post-injection of leptin or PBS.

Histological analysis of tissues
Paraformaldehyde-fixed, paraffin-embedded sections of white adipose tissue (WAT), brown adipose tissue (BAT) and liver were stained with haematoxylin-eosin (H&E) for histology. Frozen sections of liver were stained with Oil Red O.

Liver and serum triglyceride, cholesterol and free fatty acids measurements
Hepatic lipids were extracted by chloroform methanol (2 : 1) as previously described [25]. Hepatic and serum triglyceride (TG), cholesterol (TC) and free fatty acids (FFAs) were measured with a TG kit (SSUF-C, Shanghai, China), TC kit (SSUF-C) and FFA kit (Wako Pure Chemical Industries, Osaka, Japan), respectively, according to the manufacturer's instructions.

RNA isolation and relative quantitative RT -PCR
The RNA isolation and RT-PCR were performed as previously described [21]. The sequences of primers used in this study are available upon request.

Statistical analysis
All values are presented as means + s.e.m. Differences between groups were analysed by either the Student t-test or one-way ANOVA followed by the Student -Newman-Keuls (SNK) test; differences in which p , 0.05 were considered statistically significant.

PIKO mice are lean
To investigate the metabolic function of hypothalamic IRE1a, we generated PIKO mice. These mice were then intercrossed with Ai9 (tdTomato reporter) mice for examining the deletion efficiency of IRE1a in POMC neurons. IF staining showed that IRE1a was colocalized with POMC-expressing neurons in the arcuate nucleus (ARC) of control mice, but this colocalization was largely absent in PIKO mice ( figure 1a,b). Furthermore, the expression of IRE1a and its downstream effectors including spliced Xbp1 and ERdj4 [18] was also decreased in the ARC of PIKO mice (figure 1c,d). Anatomical evaluation of the POMC neurons throughout the ARC of PIKO mice and control mice revealed no changes in neuronal population distribution and size (electronic supplementary material, figure S1), suggesting that IRE1a deficiency did not alter POMC neuron distribution and survival [27]. Furthermore, because POMC promoter also drives cre recombinase expression in the pituitary [27], we examined whether the function of the pituitary-adrenal axis and the basic phenotype of the pituitary were affected by IRE1a knockout. Serum contents of hormones secreted from the pituitary, including corticosterone and growth hormone [27][28][29], were comparable between PIKO mice and control mice (electronic supplementary material, figure      the reduced adiposity in PIKO mice, the adipocyte volume in WAT was significantly smaller, and phosphorylation of proteins related to lipolysis including hormone-sensitive lipase (HSL) and substrates of protein kinase A (PKA) [30] were significantly increased in WAT of these mice (figure 2f-h).

PIKO mice have higher energy expenditure
Because energy homeostasis is maintained by a balance between food intake and energy expenditure [1], we investigated the possible reasons for the decreased fat mass in PIKO mice from these two aspects. Food intake was not altered, but energy expenditure was increased as, measured by higher oxygen consumption, and the respiratory exchange ratio (RER; V CO2 =V O2 ) was decreased in PIKO mice (figure 2i-k). Although physical activity was not affected, body temperature was much higher in PIKO mice ( figure 2l,m). Furthermore, the adipocyte volume in BAT was decreased ( figure 2n,o), Ucp1 and Pgc1a expression and UCP1 protein were increased in BAT of PIKO mice (figure 2p,q), and the serum NE level was increased in PIKO mice (figure 2r).

PIKO mice are resistant to high-fat diet-induced obesity
We then analysed the metabolic parameters in PIKO and control mice maintained on a HFD for 16 weeks. The gross morphology, body weight, fat mass component and abdominal fat mass were lower, and body lean mass component was higher in PIKO mice (figure 3a-e). WAT cell volume was decreased and levels of p-HSL and p-PKA substrates were increased in PIKO mice (figure 3f-h). In addition, PIKO mice had higher oxygen consumption and body temperature, and lower RER, with no effect on food intake and physical activity compared with control mice under HFD (figure 3i-m). BAT cell volume was decreased, Ucp1 and Pgc1a expression and UCP1 protein were increased in BAT of PIKO mice (figure 3n-q), and the serum NE level was increased in PIKO mice under HFD (figure 3r).

Insulin sensitivity is improved in PIKO mice
Because a change in body fat component is associated with changes in insulin sensitivity [1], we investigated whether insulin sensitivity was altered in PIKO mice under standard chow diet or HFD. Although fed blood glucose and fed or fasting insulin were not changed, fasting blood glucose and HOMA-IR index were significantly decreased in PIKO mice maintained on a standard chow diet (figure 4a-e). Consistently, glucose tolerance and clearance were also improved significantly in PIKO mice maintained on a standard chow diet as demonstrated by GTTs and ITTs, respectively (figure 4f,g). The HFD-induced insulin resistance in control mice as shown by the significantly increased fed and fasting blood glucose and serum insulin and HOMA-IR index, and attenuated glucose tolerance and insulin sensitivity was, however, largely prevented in PIKO mice (figure 4a-g).

Leptin sensitivity is improved in PIKO mice
Reduced body fat component and increased energy expenditure imply improved leptin sensitivity in mice [31], and serum leptin levels were decreased in PIKO mice under either a standard chow diet or HFD (figure 5a), suggesting that leptin sensitivity was improved in PIKO mice. To test this possibility, we measured food intake and body weight 1 and 4 h following injection i.p. with 3 mg kg 21 leptin or PBS [22] in control and PIKO mice maintained on a standard chow diet. Leptin (1 or 4 h) injection significantly decreased rsob.royalsocietypublishing.org Open Biol. 6: 160131 food intake and body weight in control and PIKO mice; however, the extent was much more significant in PIKO mice ( figure 5b,c). Furthermore, leptin-induced phosphorylation of signal transducer and activator of transcription 3 ( p-STAT3), an indicator for leptin signalling [22], was stronger in the POMC neurons of PIKO mice compared with control mice (figure 5d,e). Similar effects were observed in PIKO and control mice maintained on a HFD (figure 5b-e).

Hepatic lipid accumulation is reduced in PIKO mice
Because an alteration in fat mass is always associated with a change in lipid accumulation in liver [32], we therefore investigated whether this occurred in PIKO mice. On a standard chow diet, PIKO mice had lower hepatic lipid accumulation as demonstrated by the decreased lipid droplets shown by Oil Red O and H&E staining, reduced hepatic TG and FFAs levels, and reduced serum TG and TC levels ( figure 6b-d). Liver weight and hepatic TC and serum FFA levels, however, were not altered in PIKO mice ( figure 6a,c,d). HFD induced significant liver steatosis in control mice, which was largely prevented in PIKO mice (figure 6a-d).

Alpha-melanocyte-stimulating hormone production is increased in the hypothalamus of PIKO mice
To explore the mechanisms underlying the lean phenotype of PIKO mice, we measured the protein levels of a-MSH, a key component vital for the regulation of obesity and energy metabolism [27], in the hypothalamus of PIKO and control mice under either a standard chow diet or HFD. Hypothalamic a-MSH levels were increased, and the staining of a-MSH in the paraventricular nucleus (PVN) of the hypothalamus was much stronger in PIKO mice compared with control mice maintained on a standard chow diet (figure 7a-c). HFD decreased hypothalamic a-MSH levels and weakened its staining in the PVN of control mice; however, these effects were largely blocked in PIKO mice ( figure 7a-c). a-MSH originates from the processing of POMC by enzymes including prohormone convertase 1 (Pc1/3), prohormone convertase 2 (Pc2), carboxypeptidase E (Cpe), a-amidating monooxygenase (Pam) and prolylcarboxypeptidase (Prcp) [27]. Most of these genes were not changed; Pc2 and Cpe expressions were increased in the hypothalamus of PIKO mice under both diets (figure 7d,e). In addition, the expression of feeding neuropeptides Pomc, Agouti-related peptide (Agrp), neuropeptide Y (Npy) and cocaine and amphetamine-related transcript (Cart) [27] showed no significant differences in the hypothalamus of PIKO mice and that of control mice (figure 7d,e and electronic supplementary material, figure S3).

Discussion
Previous works have shown that IRE1a plays an important role in ER stress [17], autophagy [15] and insulin secretion [33]. It is also indicated that IRE1a controls lipogenic gene rsob.royalsocietypublishing.org Open Biol. 6: 160131 expression and adipocyte differentiation [16,34]. The embryonic lethality in IRE1a deletion mice [35], however, results in limiting the study of loss of gene function. Recent studies by using the Cre/loxP system identified the hepatocytespecific role of IRE1a in fatty acid b-oxidation, ketogenesis and liver regeneration [18,36]. However, the role of IRE1a in the regulation of energy homeostasis is poorly understood.
Here, we showed that knockout of IRE1a in POMC neurons decreases fat mass and protects mice from HFD-induced obesity and obesity-related metabolic disorders, the effects of which are caused by increased energy expenditure. Furthermore, the decreased fat mass is associated with increased a-MSH production in the hypothalamus of PIKO mice. We speculated that the phenotypes observed in PIKO mice were mainly caused by knockout of IRE1a in POMC neurons. POMC promoter, however, also drives cre recombinase expression in the pituitary [27,29]. There was no significant defect in the pituitary-adrenal axis observed in PIKO mice, suggesting that the pituitary-adrenal axis is unlikely to be involved in the regulation of the lean phenotype in PIKO mice.
Food intake, however, was not altered in PIKO mice compared with control mice. Consistent with our results, it is shown that pharmacological or genetic blockade of melanocortin receptors in the CNS had no effect on food intake [37,38]. As a result, the increased energy expenditure should certainly make a great contribution to the decreased fat mass in PIKO mice. We further demonstrated that the increased energy expenditure in PIKO mice was caused by induced thermogenesis in BAT, rather than a change in physical activity.
BAT oxidizes fat to produce heat via increased expression of UCPs. Deletion of UCP1 induces obesity and upregulation of UCP1 increases thermogenesis and energy expenditure in mice [39]. Lipolysis in WAT is regulated by lipases, including HSL [40]. BAT and WAT have receptors for NE that are activated by hypothalamic signals via stimulation of the sympathetic nervous system (SNS) [39][40][41]. The expression of UCP1 in BAT and phosphorylation of HSL in WAT, as well as serum NE levels, were increased in PIKO mice, suggesting that deletion of IRE1a in POMC neurons activated the SNS and increased thermogenesis and lipolysis in PIKO mice.
It has been shown that altered body fat mass is associated with other metabolic changes, including insulin sensitivity and leptin sensitivity [42]. Insulin binds to insulin receptors expressed in different tissues to maintain normal blood glucose levels [43] and leptin binds to leptin receptors highly expressed in POMC neurons to regulate energy homeostasis and lipid metabolism [44]. Therefore, insulin and leptin are critical for the maintenance of adequate glucose and lipid metabolism [27,45,46]. Here, increased responses to insulin and leptin were observed in PIKO mice, as demonstrated by the improved glucose clearance and insulin sensitivity, and increased leptin effects and p-STAT3 staining in POMC neurons of these mice. The improved insulin sensitivity and leptin sensitivity may contribute to the lean phenotypes in PIKO mice. Lipid accumulation in liver is also associated with changes in body fat component [32]. Consistent with this, we observed that PIKO mice have reduced hepatic TG accumulation and these mice were resistant to HFD-induced liver rsob.royalsocietypublishing.org Open Biol. 6: 160131 steatosis. We hypothesized that these associated metabolic changes in PIKO mice were most likely caused by the decreased fat mass in PIKO mice. It is also, however, shown hypothalamic signals have direct effect on insulin sensitivity and lipid metabolism [24,47,48]; the possibility that IRE1a in POMC neurons might also have direct effect on these metabolic changes should not be excluded. POMC is a prohormone that undergoes posttranslational proteolysis by several convertases to generate a-MSH [27,49]. In this study, we found that a-MSH levels were increased in the hypothalamus of PIKO mice, suggesting that the effects of IRE1a in POMC neurons on fat mass might be mediated by increased a-MSH production. Furthermore, the expression of Pomc and other feeding neuropeptides were not changed, whereas POMC processing enzymes including Pc2 and Cpe were increased in the hypothalamus of PIKO mice, suggesting that knockout of IRE1a might increase a-MSH levels by inducing expression of POMC processing enzymes. These possibilities, however, required further investigation.
IRE1a has a kinase domain and an RNase domain in the cytosolic region, and the RNase activity induces the splicing of XBP1 [13]. Consistently, the expression of XBP1s was reduced in the ARC of PIKO mice. In contrast to the lean phenotype and the resistance to HFD-induced obesity in PIKO mice, inhibition of XBP1 in the hypothalamus results in an obese phenotype in mice [50], and activation of XBP1 in POMC neurons protects mice against HFD-induced obesity [42]. We speculated that the possible reasons for the different phenotypes observed in our study and those of the other work [42,50] could be as follows: XBP1 splicing represents one of the downstream pathways from the activation of the RNase domain of IRE1a, whereas other signalling pathways might also be activated or inhibited by the knockout of IRE1a, which may contribute to the lean phenotype in PIKO mice. For example, altered expression of hypothalamic c-Jun N-terminal kinases (JNK), as a downstream target of the kinase domain of IRE1a [17], also regulates obesity and its related diseases [51,52]. Interestingly, the expression of ER stress markers, including phosphorylated eukaryotic translation initiation factor 2a (p-eIF2a), BiP and activating transcription factor 4 (ATF4) [27] was decreased in the hypothalamus of PIKO mice under both standard chow diet and HFD (electronic supplementary material, figure S4), suggesting another possibility in explaining the effects of IRE1a deficiency in POMC neurons. The possible involvement of these signalling pathways in IRE1a regulation of energy balance requires to be studied in the future.
Taken together, we provided the first link between energy homeostasis, lipid metabolism and POMC neuronal IRE1a function. Furthermore, the effects of POMC IRE1a knockout were mediated via increased thermogenesis, possibly caused by the increased a-MSH production in the hypothalamus. This previously unidentified role of hypothalamic IRE1a also indicates a potential novel drug target in treating obesity and obesity-related metabolic disorders.