CREBH Improves Diet-Induced Obesity, Insulin Resistance, and Metabolic Disturbances by FGF21-Dependent and FGF21-Independent Mechanisms.

Mice overexpressing the nuclear form of CREBH mainly in the liver (CREBH-Tg) showed suppression of high-fat high-sucrose (HFHS) diet-induced obesity accompanied by an increase in plasma fibroblast growth factor 21 (FGF21) levels. CREBH overexpression induced browning in inguinal white adipose tissue (WAT) and whole-body energy expenditure, which was canceled in Fgf21-/- mice. Deficiency of FGF21 in CREBH-Tg mice mostly canceled the improvement of obesity, but the suppression of inflammation of epidermal WAT, amelioration of insulin resistance, and improvement of glucose metabolism still sustained. Kisspeptin 1 (Kiss1) was identified as a novel hormone target for CREBH to explain these FGF21-independent effects of CREBH. Knockdown of Kiss1 in HFHS-fed CREBH-Tg Fgf21-/- mice showed partially canceled improvement of glucose metabolism. Taken together, we propose that hepatic CREBH pleiotropically improves diet-induced obesity-mediated dysfunctions in peripheral tissues by improving systemic energy metabolism in FGF21-dependent and FGF21-independent mechanisms.

Introduction

cAMP responsive element-binding protein 3-like 3 (CREB3L3, also named CREBH) is a basic liver-specific leucine zipper transcription factor belonging to the CREB/ATF family (Omori et al., 2001). CREBH predominantly resides in the endoplasmic reticulum (ER). Under ER stress, CREBH is transferred to the Golgi apparatus where site 1 and site 2 proteases cleave its amino-terminal portion with its subsequent transfer to the nucleus to induce its target genes (Zhang et al., 2006). CrebH expression in the liver is induced in a fasted state rather than the fed state, resulting in accumulation of the nuclear form of CREBH. Previously, our group reported that transgenic mice overexpressing the nuclear form of CREBH in the liver (CREBH-Tg) resist diet-induced obesity (DIO) by remarkably increasing hepatic gene expression and plasma levels of fibroblast growth factor 21 (FGF21) (Nakagawa et al., 2014). CREBH-Tg mice also present improvements in hyperlipidemia, hyperglycemia, and insulin resistance in DIO mice (Nakagawa et al., 2014). Conversely, CREBH knockout (KO) (CrebH) mice exhibit hypertriglyceridemia due to inefficient catalysis of triglyceride (TG) clearance by lipoprotein lipase (LPL) in blood. Hypertriglyceridemia results from the reduction of the expression of the LPL coactivators, Apolipoprotein c2 (Apoc2), Apoa4, and Apoa5, and the increase of the LPL inhibitor Apoc3 in the liver (Lee et al., 2011, Zhang et al., 2012). When fed a methionine-choline-deficient diet or ketogenic diet (KD), the deficiency of CREBH results in hepatic steatosis (Nakagawa et al., 2016a, Nakagawa et al., 2016b, Park et al., 2016).

FGF21 is a unique member of the FGF family. It has hormone-like actions and is a key mediator of starvation by activating fatty acid oxidation and ketogenesis in the liver and lipolysis in white adipose tissue (WAT) (Badman et al., 2007, Inagaki et al., 2007). FGF21 is mainly synthesized in the liver, and adipose tissues are also important organs for both the production and activity of FGF21. In adipocytes, FGF21 enhances insulin-independent glucose uptake, promotes mitochondrial oxidation, and induces secretion of adiponectin, which in turn mediates the glucose-lowering and insulin-sensitizing effects of FGF21 (Chau et al., 2010, Lin et al., 2013). In response to the cold challenge, FGF21 is induced in adipose tissues, but not in the liver (Fisher et al., 2012, Hondares et al., 2011). Deficiency of FGF21 in mice impairs cold-induced browning of inguinal WAT (iWAT), whereas recombinant FGF21 administration increases browning of iWAT and total energy expenditure in mice (Fisher et al., 2012). Recently, it was reported that plasma FGF21 levels positively correlate with the weight of iWAT in insulin-sensitive obese individuals (Li et al., 2018). Fgf21 mice show less iWAT mass and are more insulin resistant when fed a high-fat diet. Plasma FGF21 upregulates adiponectin in iWAT, accompanied by an increase of M2 macrophage polarization (Li et al., 2018). The upregulation of plasma FGF21 levels in obesity is thought to serve as a defense mechanism to protect against systemic insulin resistance (Li et al., 2018). FGF21 is proposed to have therapeutic effects on obesity-related metabolic disturbances, such as insulin resistance, diabetes, and hypertriglyceridemia in ob/ob mice, DIO mice, and diabetic monkeys (Kharitonenkov et al., 2007, Xu et al., 2009). Similar to insulin resistance, FGF21 resistance is induced in obese mice such as DIO and ob/ob mice, thereby increasing plasma FGF21 levels (Fisher et al., 2010). This contributed to a decrease of βKlotho in WAT, but not in the liver of obese mice (Fisher et al., 2010). Fgf21 expression is governed by peroxisome proliferator-activated receptor (PPAR) α and CREBH, which is induced by fasting or KD feeding (Badman et al., 2007, Inagaki et al., 2007, Nakagawa et al., 2014, Nakagawa et al., 2016b). Both CREBH and PPARα are activated in an autoloop manner in response to starvation, and thus synergistically activate Fgf21 expression (Kim et al., 2014, Nakagawa et al., 2014).

Previously, we proposed that CREBH-Tg mice ameliorate the pathology of DIO mice by inducing hepatic expression and plasma levels of FGF21. In this study, by using Fgf21 mice, we identified the FGF21-dependent or FGF21-independent effects on CREBH-mediated improvement of obesity.

Results

FGF21 Has a Crucial Role in the Suppression of Diet-Induced Obesity in CREBH-Tg Mice

In CREBH-Tg mice, body weight (BW) gain, hyperglycemia, hyperinsulinemia, and hypertriglyceridemia are clearly suppressed relative to high-fat high-sucrose (HFHS) DIO (Nakagawa et al., 2014). There is a possibility that these phenotypes of CREBH-Tg mice are a result of hepatic induction of Fgf21, a CREBH target gene (Nakagawa et al., 2014). To determine this notion, we generated CREBH-Tg Fgf21 mice by crossing CREBH-Tg mice with Fgf21 mice and then fed them with HFHS diet for 12 weeks. Consistent with the previous report (Nakagawa et al., 2014), BW gain was suppressed in CREBH-Tg mice compared with control wild-type (WT) mice following HFHS diet (Figure 1A). In contrast, BW of CREBH-Tg Fgf21 mice was reversed to the levels of WT mice, although Fgf21 mice did not show a significant increase in BW compared with WT mice (Figure 1A). There were no differences in liver weight among WT, CREBH-Tg, and CREBH-Tg Fgf21 mice, whereas Fgf21 mice had a higher liver weight than WT mice (Figure 1B). In accordance with BW changes, CREBH-Tg mice exhibited significantly lower epidermal WAT (eWAT) weight than WT mice but CREBH-Tg Fgf21mice did not (Figure 1B). Thus, suppressive effect of CREBH overexpression on DIO was completely canceled out in CREBH-Tg Fgf21 mice, indicating that fat loss effect of CREBH on a quantitative basis was mostly mediated through FGF21. There were no differences in food intake among all the groups (Figure 1C).

Figure 1

CREBH-Tg Mice Show Improvement of Diet-Induced Obesity

Six-week-old male WT, Tg, Fgf21, and Tg Fgf21 mice were fed a high-fat high-sucrose (HFHS) diet for 12 weeks.

(A) Body weight changes at indicated ages (n = 14–26 per group). aStatistical significance between WT and Tg mice (ap < 0.05, aap < 0.01, aaap < 0.001). bStatistical significance between WT and Fgf21 mice (bp < 0.05). cStatistical significance between Tg and Fgf21 mice (cp < 0.05, ccp < 0.01, cccp < 0.001). dStatistical significance between Tg and Tg Fgf21 mice (dp < 0.05, ddp < 0.01, dddp < 0.001). eStatistical significance between Fgf21 and Tg Fgf21 mice (ep < 0.05).

(B) Liver and epidermal WAT (eWAT) weights (n = 8–13).

(C) Food intake (n = 4–7).

(D) Plasma levels of glucose and insulin (n = 8–22).

(E) Plasma levels of triglyceride (TG), total cholesterol (TC), and nonesterified fatty acids (NEFA) (n = 9–14).

(F) Plasma levels of FGF21 (n = 11–12).

(G) Representative pictures of liver sections stained with H&E.

(H) Liver TG and TC (n = 9–14).

(I) mRNA expressions of CREBH target genes, lipid, and cholesterol synthetic genes in livers were determined by real-time PCR (n = 7–12). Results are represented as means ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

CREBH Overexpression Showed FGF21-Dependent Suppression of HFHS Diet-Induced Lipid Accumulation in the Liver and FGF21-Independent Amelioration of Insulin Resistance

CREBH overexpression in WT mice was associated with a significant reduction in plasma glucose, insulin, and TG levels, whereas there were no differences in plasma total cholesterol (TC) and nonesterified fatty acid (NEFA) levels compared with WT mice (Figures 1D and 1E). Meanwhile, Fgf21 mice showed no changes in plasma glucose or NEFA levels compared with WT mice, but had a tendency to increase plasma insulin levels, indicating insulin resistance (Figures 1D and 1E). Then, plasma glucose levels of CREBH-Tg Fgf21 mice did not significantly change compared with those of CREBH-Tg mice or Fgf21 mice, but CREBH-Tg Fgf21 mice sustained significantly reduced plasma insulin levels (similar to CREBH-Tg mice and significantly lower than that of Fgf21 mice) (Figure 1D). The data indicated that unlike obesity, amelioration of insulin resistance by CREBH overexpression is at least partially FGF21 independent. CREBH-Tg Fgf21 mice showed no marked further changes in TG, TC, and NEFA levels compared with their respective controls (Figure 1E). Consistent with a previous report (Nakagawa et al., 2014), plasma FGF21 levels in CREBH-Tg mice were apparently increased compared with WT mice (Figure 1F). Morphological analysis with hematoxylin and eosin staining in the liver revealed that CREBH-Tg mice clearly had fewer and smaller lipid droplets compared with WT mice (Figure 1G). Fgf21 mice exhibited exacerbation of hepatosteatosis with larger lipid droplets than WT mice. Additional absence of FGF21 in CREBH-Tg canceled amelioration of hepatosteatosis, but CREBH-Tg Fgf21 mice had still smaller lipid droplets than Fgf21 mice (Figure 1G). Consistently, the quantitative analysis revealed that liver TG was significantly reduced in CREBH-Tg mice and significantly increased in Fgf21 mice compared with WT mice (Figure 1H). Liver TG content of CREBH-Tg Fgf21 mice had a trend to decrease compared with Fgf21 mice (Figure 1H). Liver TC content significantly decreased in Fgf21 mice compared with WT mice, but no further reduction in CREBH-Tg Fgf21 mice was observed (Figure 1H).

CREBH Increased Fgf21 and Apoa4 Expression in the Liver of DIO Mice

Consistent with the elevated plasma FGF21 levels (Figure 1F), CREBH overexpression significantly increased hepatic Fgf21 expression compared with control WT mice and was undetectable in Fgf21 background mice (Figure 1I). CREBH overexpression increased target gene expression of Apoa4 and Apoc2, activators for LPL (Figure 1I), which contributes to a reduction in plasma TG levels. CREBH-Tg mice had a trend to reduce Srebp-1c expression and subsequently reduce its target gene expression of fatty acid synthase (Fasn) and stearoyl-coenzyme A desaturase 1 (Scd1) and elongation of very-long-chain fatty acid (FEN1/Elo2, SUR4/Elo3, yeast)-like 6 (Elovl6) in the liver compared with WT mice (Figure 1I). For cholesterol synthesis genes including Srebf2, 3-hydroxy-3-methylglutaryl-CoA synthase 1 (Hmgcs1) and 3-hydroxy-3-methylglutaryl-CoA reductase (Hmgcr), CREBH overexpression had no effects in both control WT and Fgf21 mice (Figure 1I). FGF21 reciprocally changed liver TG and TC contents, whereas the effects of CREBH on hepatic lipogenic and cholesterogenic gene expression were marginal.

CREBH Induces Browning in iWAT of DIO Mice

Browning is the appearance and activation of beige adipocytes within iWAT, accompanied by induction of UCP1 (Veniant et al., 2015). As CREBH-Tg mice showed suppression of DIO-induced adiposity, we determined the changes in gene expression related to browning and lipolysis in WATs, including eWAT, iWAT, and brown adipose tissue (BAT) of each DIO mice group. Although the previous report indicated that FGF21 activates lipolysis in WATs (Hotta et al., 2009), lipolysis-related genes such as lipase and hormone-sensitive (lipe) and patatin-like phospholipase domain-containing 2 (Pnpla2) in WATs and BATs were not apparently changed by CREBH overexpression despite high plasma FGF21 levels (Figures 2A–2C). Meanwhile, gene profile in eWAT among four groups highlighted browning markers, uncoupling protein 1 (Ucp1), Elovl3, and cell death-inducing DNA fragmentation factor alpha-like effector A (Cidea). Ucp1 and Elovl3 expression was significantly increased in Fgf21 mice, which were suppressed in CREBH-Tg Fgf21 mice (Figure 2A). In contrast, CREBH overexpression in both control WT and Fgf21 mice significantly increased Cidea expression in eWAT, which implicates FGF21-independent CREBH-specific effect on browning (Figure 2A). In iWAT, CREBH-Tg mice showed a significant increase in several browning-induced genes such as peroxisome proliferative-activated receptor, gamma, coactivator 1 alpha (Ppargc1a), and Cidea, and the subsequent increase in Ucp1 compared with WT mice (Figure 2B), indicating that CREBH-Tg mice might induce browning and thermogenesis in iWAT. These changes were partially canceled in CREBH-Tg Fgf21 mice, but the increase in Ucp1 and Cidea were still partially maintained (Figure 2B), suggesting that CREBH-induced browning of iWAT is both FGF21 dependent and FGF21 independent presumably depending on details in browning functions. Elovl3, a mitochondrial function marker, was markedly increased in CREBH-Tg mice, which was completely canceled in CREBH-Tg Fgf21 mice (Figure 2B). These results indicate that CREBH induces browning genes in iWAT by FGF21-dependent and FGF21-independent mechanisms. However, PR domain containing 16 (Prdm16), a master regulator for brown adipocyte differentiation, was not apparently changed in CREBH-Tg mice, indicating that CREBH-induced browning is independent of mediating the expression of PRDM16. Regarding lipolysis, Pnpla2, not lipe, was slightly increased in iWAT of CREBH-Tg mice, which was canceled in CREBH-Tg Fgf21 mice (Figure 2B). In BAT, Ucp1 and browning genes remained unchanged in WT and CREBH-Tg mice (Figure 2C). However, Elovl3 expression was significantly induced in BAT of CREBH-Tg mice (Figure 2C), activating mitochondrial function and energy expenditure. This change was blunted in Fgf21 background mice groups (Figure 2C), indicating that FGF21 governs Elovl3 expression in BAT.

Figure 2

CREBH-Tg Mice Induce Browning in Inguinal WAT (iWAT) via FGF21

Six-week-old male WT, Tg, Fgf21, and Tg Fgf21 mice were fed an HFHS diet for 2 weeks.

(A–C) mRNA expressions of beige adipocyte marker genes in epidermal WAT (eWAT) (A), iWAT (B), and brown adipose tissue (BAT) (C) were determined by real-time PCR (n = 4–8).

(D) UCP1 protein expression in eWAT, iWAT, and BAT from WT and Tg mice; α-tubulin was used as a loading control (n = 3).

(E) UCP1 protein expression in iWAT from WT, Tg, Fgf21, and Tg Fgf21 mice; α-tubulin was used as a loading control (n = 5).

(F) UCP1 immunostaining of iWAT from WT and Tg mice.

(G) Oxygen consumption VO2 (NmL/h/kg) of WT, Tg, Fgf21, and Tg Fgf21 mice measured by the ARCO2000 System (Arcosystem) (n = 5–9).

(H) The areas under the curve (AUCs) from the light, dark, and total periods in (F) (n = 5–9).

(I) Rectal core temperature of WT, Tg, Fgf21, and Tg Fgf21 mice (n = 8–14). Results are represented as means ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

We investigated UCP1 regulation at the protein level. Consistent with gene expression, UCP1 protein levels were markedly upregulated in the iWAT of CREBH-Tg mice compared with WT mice, whereas robust UCP1 protein contents in BAT were not changed between WT and CREBH-Tg mice (Figure 2D). In eWAT, UCP1 proteins were not detected (Figure 2D). FGF21 is known to induce UCP1 protein in iWAT (Fisher et al., 2012); thus, we evaluated it in CREBH-Tg Fgf21 mice. The induction of UCP1 in iWAT of CREBH-Tg mice was canceled in CREBH-Tg Fgf21 mice (Figure 2E). The induction of UCP1 in iWAT of CREBH-Tg mice was confirmed by immunocytochemistry using UCP1 antibodies (Figure 2F). These findings support the presumption that the induction of UCP1 in CREBH-Tg mice is mediated by FGF21.

CREBH Induces Whole-Body O2 Consumption Depending on FGF21

As it is predicted that thermogenesis is activated in CREBH-Tg mice, we determined the energy expenditure. The energy expenditure of 6-week-old mice fed an HFHS diet for 2 weeks was determined. Consistent with the induction of browning in iWAT, CREBH-Tg mice had higher O2 consumption than control WT mice throughout the day (Figures 2G and 2H). CREBH-Tg mice tended to have a higher body temperature than control WT mice (Figure 2I). These changes by CREBH overexpression were canceled in Fgf21 background mice (Figures 2G–2I). These results indicate that CREBH resists DIO by enhancing basal metabolism and especially thermogenesis, which is dependent on FGF21.

CREBH Overexpression Improves Both Glucose Response and Insulin Response in WT and Even Fgf21 Mice

Improvement of diet-induced insulin resistance by CREBH overexpression (Figure 1D) was further investigated by performing glucose tolerance test (GTT) and insulin tolerance test (ITT) in WT, CREBH-Tg, Fgf21, and CREBH-Tg Fgf21 mice fed an HFHS diet. In GTT, plasma glucose levels of Fgf21 mice were higher than those of other groups, but other groups showed no differences in plasma glucose levels during GTT (Figure 3A). These changes were confirmed by calculating the area under the curve (Figure 3A). The data indicate that CREBH overexpression did not change glucose tolerance following an HFHS diet but ameliorated impairment by FGF21 absence. Intriguingly, plasma insulin levels of CREBH-Tg Fgf21 mice were lower than those of WT mice but were higher than those of CREBH-Tg mice, indicating marked improvement in insulin resistance by overexpression of CREBH even in the absence of Fgf21 (Figure 3B). Homeostasis model assessment of insulin resistance also showed that CREBH overexpression improves insulin resistance in WT and even Fgf21 mice (Figure 3C). ITT demonstrated that the slightly slurred insulin response curve of WT mice on HFHS diet was markedly reduced in CREBH-Tg mice so that almost no increment in insulin response was necessary to maintain similar glucose excursion of WT mice (Figure 3D), indicating a profound improvement of insulin sensitivity. Similarly, CREBH-Tg Fgf21 mice had lower plasma glucose levels than that of control WT and even Fgf21 mice (Figure 3D). In contrast to dependence on FGF21 in amelioration of DIO, CREBH overexpression can also promote insulin response by both FGF21-dependent and FGF21-independent mechanisms with the latter more marked.

Figure 3

CREBH-Tg Mice Prevent HFHS-Induced Insulin Resistance Independent of FGF21

(A) Plasma glucose levels during glucose tolerance test (GTT) in WT, Tg, Fgf21, and Tg Fgf21 mice fed 12 weeks of HFHS diet from age 6 weeks (left) and AUC (right) (n = 7–24). aStatistical significance between WT and Fgf21 mice (ap < 0.05). bStatistical significance between Tg and Fgf21 mice (bp < 0.05, bbp < 0.01).

(B) Plasma insulin levels during GTT (n = 7–18). aStatistical significance between WT and Tg mice (aap < 0.01, aaap < 0.001). bStatistical significance between WT and Fgf21 mice (bp < 0.05). cStatistical significance between Tg and Fgf21 mice (ccp < 0.01, cccp < 0.001). dStatistical significance between Tg and Tg Fgf21 mice (dp < 0.05). eStatistical significance between Fgf21 and Tg Fgf21 mice (ep < 0.05, eep < 0.01).

(C) The homeostatic model assessment index of insulin resistance (HOMA-R) in WT, Tg, Fgf21, and Tg Fgf21 mice fed HFHS diet for 12 weeks from age 6 weeks (n = 7–18). ∗p < 0.05, ∗∗p < 0.01.

(D) Blood glucose levels during insulin tolerance test (ITT) in WT, Tg, Fgf21, and Tg Fgf21 mice fed normal chow diet for 12 weeks from age 6 weeks (n = 9–25). aStatistical significance between WT and Tg mice (ap < 0.05, aaap < 0.001). bStatistical significance between WT and Tg Fgf21 mice (bbp < 0.01, bbbp < 0.001). cStatistical significance between Tg and Fgf21 mice (cp < 0.05, ccp < 0.01). dStatistical significance between Fgf21 and Tg Fgf21 mice (dp < 0.05, ddp < 0.01). Results are represented as means ± SEM.

CREBH-Tg Mice Showed Suppression of HFHS Diet-Induced Macrophage Infiltration in eWAT

Histological analysis of eWAT using hematoxylin and eosin staining indicated that Fgf21 mice had cell size of eWAT similar to WT mice; CREBH overexpression WT and Fgf21 mice reduced this (Figure 4A). Interestingly, immunohistochemistry using antibody against F4/80 in eWAT showed that Fgf21 mice induced infiltrated macrophages compared with WT mice and CREBH overexpression reduced them compared with WT and even Fgf21 mice (Figure 4A). As FGF21 is reported to suppress inflammation in eWAT (Wang et al., 2018), deficiency of FGF21 increased expression of genes such as CD68 antigen (Cd68), tumor necrosis factor (Tnf), chemokine (C-C motif) ligand 2 (Ccl2), and interleukin-6 (Il6) compared with WT mice (Figure 4B). These changes were consistent with an increase of macrophage infiltration in eWAT of Fgf21 mice shown in Figure 4A. CREBH overexpression in both WT and Fgf21 mice reduced the macrophage marker Cd68 and inflammatory genes such as Ccl2 and Tnf compared with WT and Fgf21 mice, respectively (Figure 4B). Furthermore, CREBH overexpression significantly decreased Il6 expression in Fgf21 mice, and there were no changes in WT mice (Figure 4B). Therefore, the data support the fact that anti-inflammatory effects of CREBH might be FGF21 independent. BAT adipocytes of WT and Fgf21 mice exhibited larger and unilocular lipid droplets like a WAT adipocyte (Figure 4C). CREBH overexpression suppressed these changes in both WT and Fgf21 mice (Figure 4C). These findings suggest that the suppressive effects of CREBH on lipid accumulation in BAT are FGF21 independent. It has been known that inflammation in eWAT and BAT is associated with insulin resistance, and these observations could be a part of the FGF21-independent insulin sensitization by CREBH.

Figure 4

CREBH-Tg Mice Suppress the Inflammation in eWAT and Lipid Accumulation in BAT Independent of FGF21

(A and C) Representative pictures of (A) epidermal WAT (eWAT) sections stained with H&E (upper) and F4/80 immunostaining (bottom) and (C) brown adipose tissue (BAT) sections stained with H&E from WT, Tg, Fgf21, and Tg Fgf21 mice fed an HFHS diet for 12 weeks from age 6 weeks.

(B) Six-week-old male WT, Tg, Fgf21, and Tg Fgf21 mice were fed a normal chow or HFHS diet for 12 weeks. mRNA expressions of inflammatory genes in eWAT were determined by real-time PCR (n = 7–8). Results are represented as means ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

CREBH Ameliorates Glucose Metabolism in ob/ob Mice, a Type 2 Diabetic Mouse Model, and FGF21-Resistant Mice

Another obese, but genetic, mouse model, ob/ob mice, showed FGF21 resistance with an expected and marked increase in plasma FGF21 levels (Fisher et al., 2010, Hale et al., 2012). If ob/ob mice overexpressing CREBH improve the pathology, it would suggest that CREBH has other target genes to ameliorate metabolic disorder except for FGF21. CREBH-Tg mice were crossed with ob/ob mice. There were no apparent differences in BW changes between ob/ob and ob/ob; CREBH-Tg mice (data not shown). As expected, ob/ob mice had higher hepatic Fgf21 expression and plasma FGF21 levels compared with ob/+ control mice (Figures 5A and 5B). ob/ob; CREBH-Tg mice had a trend to increase hepatic gene expression and plasma levels of FGF21 compared with ob/ob mice (Figures 5A and 5B). ob/ob; CREBH-Tg mice showed a significant reduction in plasma glucose and insulin levels compared with ob/ob mice (Figures 5C and 5D). Although ob/ob mice have FGF21 resistance, CREBH-induced FGF21 might contribute to an improvement in glucose metabolism in ob/ob; CREBH-Tg mice. In GTT, CREBH overexpression in ob/ob mice markedly reduced plasma glucose and insulin levels compared with ob/ob mice (Figures 5E and 5F), indicating that glucose intolerance was improved in ob/ob; CREBH-Tg mice. In ITT, plasma glucose levels of ob/ob; CREBH-Tg mice were lower than in ob/ob mice, indicating that CREBH overexpression can improve insulin resistance (Figure 5G). Collectively, an improvement in glucose metabolism in CREBH overexpression in ob/ob mice despite FGF21 resistance was another piece of evidence that CREBH ameliorates insulin resistance regardless of FGF21.

Figure 5

CREBH Ameliorates Glucose Metabolism in ob/ob Mice

(A) mRNA expressions of Fgf21 of 10-week-old ob/+, Tg ob/+, ob/ob, and Tg ob/ob mice were determined by real-time PCR (n = 5–7).

(B–D) Plasma levels of FGF21 (n = 4–6) (B), glucose (n = 5–7) (C), and insulin (n = 3–6) (D).

(E and F) Plasma glucose levels (E) and insulin levels (F) during glucose tolerance test (GTT) in ob/+, Tg ob/+, ob/ob, and Tg ob/ob mice (n = 4–5).

(G) Blood glucose levels during insulin tolerance test (ITT) in ob/+, Tg ob/+, ob/ob, and Tg ob/ob mice (n = 7–8). Results are represented as means ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Kiss1 as a Novel CREBH Target Gene in the Liver

To identify novel CREBH target genes, except for FGF21, which improves insulin sensitivity, we analyzed a combination of two different microarray datasets. One was the ratio of HFHS diet-fed CREBH-Tg to WT mice liver samples (x axis), and the other was the ratio of fed state to the fasted state (y axis) based upon the observation that CREBH is activated physiologically at fasting (Figure 6A) (Danno et al., 2010). We identified Kisspeptin (Kiss1), a neuropeptide secreted by the liver and brain (Gottsch et al., 2004, Song et al., 2014), as a candidate CREBH gene target (Figure 6A). Kiss1 gene expression was significantly increased in the liver of CREBH-Tg mice compared with WT mice (Figure 6B), which did not change even with deficiency of FGF21 (Figure 6B). Conversely, CREBH KO mice showed a significant reduction of Kiss1 expression in the liver compared with WT mice in the fasting condition (Figure 6C). To determine whether CREBH directly regulates Kiss1 expression, promoter analysis including luciferase assay, electrophoretic mobility shift assay (EMSA), and chromatin immunoprecipitation (ChIP) assay was performed. CREBH can bind to the cyclic AMP response element (CRE) sequence (Nakagawa et al., 2016b, Omori et al., 2001). A previous report revealed that mouse Kiss1 promoter has two CRE sites (Song et al., 2014). In the luciferase analysis, CREBH activated Kiss1 promoter activity but blunted it using the Kiss1 luciferase vectors deleting two putative CRE sites (Figure 6D). Consistent with these results, CREBH bound to both the putative CREBH-binding sites using EMSA (Figure 6E). Next, we performed the ChIP assay using AML12 cells infected with adenoviral CREBH. Surely, CREBH bound to Kiss1 promoter (Figure 6F). Taken together, these results confirmed that Kiss1 is a target for CREBH.

Figure 6

Kiss1 Is a New Target for CREBH

(A) Microarray analysis from two independent datasets from liver samples. Relative gene expression of Tg mice fed HFHS diet for 12 weeks from age 6 weeks divided by WT mice fed HFHS diet for 12 weeks from age 6 weeks (log2, x axis), and WT mice fasted for 24 h divided by ad lib (log2, y axis).

(B) mRNA expressions of Kiss1 gene in the liver from WT, Tg, Fgf21, and Tg Fgf21 mice fed normal chow or HFHS diet for 12 weeks from age 6 weeks were determined by real-time PCR (n = 7–8).

(C) mRNA expressions of Kiss1 gene in 8-week-old 24-h-fasted liver from WT and CrebH mice were determined by real-time PCR (n = 4–9).

(D) Mouse Kiss1 promoter activity in AML 12.2 cells, as estimated by luciferase reporter assay. AML 12.2 cells were transfected with a pGL4.10-Kiss1 reporter with 300-bp deletion series from 1000 bp and the indicated expression plasmid (n = 8).

(E) EMSA analysis of the CREBH-binding sites in the Kiss1 gene promoter. The in vitro-translated CREBH protein and 32P-labeled oligonucleotides for the CREBH-binding site 1 and site 2 were used in EMSAs.

(F) Chromatin immunoprecipitation (ChIP) assays were performed with anti-hemagglutinin (HA) antibodies from Adv GFP- or Adv HA CrebH-treated AML 12.2 cells (n = 5). Results are represented as means ± SEM. ∗p < 0.05, ∗∗∗p < 0.001.

CREBH-Induced Kiss1 Contributes to Insulin Sensitivity

To determine the role of Kiss1 induced by CREBH, 7-week HFHS-diet-fed Fgf21 and CREBH-Tg Fgf21 mice were infected with Kiss1 knockdown adenovirus (Adv Kiss1i). At 4 weeks after adenovirus infection, we confirmed that Kiss1 expression in the liver of CREBH-Tg Fgf21 mice was reduced to the same level as that in Fgf21 mice infected with Adv LacZi (Figure 7A). GTT was performed at 3 weeks after adenovirus infection. There were no differences in plasma glucose levels among all mouse genotype groups (Figure 7B). Plasma insulin levels during GTT were lower in CREBH-Tg Fgf21 mice compared with Fgf21 mice irrespective of adenovirus treatment (Figure 7C). At 3 weeks after adenovirus infection, ITT was performed. CREBH-Tg Fgf21 Adv LacZi mice had lower plasma glucose levels than Fgf21 Adv LacZi mice. Administration of Adv-Kiss1i treatment to CREBH-Tg Fgf21 (CREBH-Tg Fgf21 Adv Kiss1i) mice gave rise to consistently higher glucose levels than CREBH-Tg Fgf21 Adv LacZi mice, canceling the blood glucose-lowering effect of CREBH overexpression (Figure 7D). These findings indicate that Kiss1 contributes to CREBH-induced insulin sensitivity. CREBH-Tg mice had lower plasma glucose levels than WT mice even if infected with Adv Kiss1i (Figure 7E). In the presence of FGF21, Adv Kiss1i did not suppress CREBH-induced insulin sensitivity. Taken together, FGF21 and Kiss1 might cooperate in CREBH-mediated improvement of glucose metabolism.

Figure 7

Adenoviral Kiss1 Knockdown in the Liver of Tg Fgf21KO Mice Fed an HFHS Diet Attenuates Insulin Sensitivity

(A) Fgf21 and Tg Fgf21 mice fed an HFHS diet for 7 weeks from age 6 weeks and then infected with Adv LacZi or Adv Kiss1i. At 4 weeks after treatment, mRNA expressions of Kiss1 gene in the liver from these mice were determined by real-time PCR (n = 5–7). ∗∗∗p < 0.001.

(B and C) (B) Plasma glucose levels and (C) plasma insulin levels during glucose tolerance test (GTT) in Fgf21 and Tg Fgf21 mice 3 weeks after treatment with Adv LacZi or Adv Kiss1i (n = 5–7).

(D) Blood glucose levels during insulin tolerance test (ITT) in Fgf21 and Tg Fgf21 mice 3 weeks after treatment with Adv LacZi or Adv Kiss1i (n = 5–7). ∗Statistical significance between Fgf21 Adv LacZi and Tg Fgf21 Adv LacZi mice (∗p < 0.05, ∗∗∗p < 0.001). #Statistical significance between Tg Fgf21 Adv LacZi and Tg Fgf21 Adv Kiss1i mice (#p < 0.05). Results are represented as means ± SEM.

(E) Blood glucose levels during ITT in WT and Tg mice 3 weeks after treatment with Adv LacZi or Adv Kiss1i (n = 6–7). aStatistical significance between WT Adv LacZi and Tg Adv LacZi mice (ap < 0.05, aap < 0.01, aaap < 0.001). bStatistical significance between WT Adv LacZi and Tg Adv Kiss1i mice (bbp < 0.01, bbbp < 0.001). cStatistical significance between WT Adv Kiss1i and Tg Adv LacZi mice (ccp < 0.01). dStatistical significance between WT Adv Kiss1i and Tg Adv Kiss1i mice (ddp < 0.01, dddp < 0.001). Results are represented as means ± SEM.

Discussion

In this study, we analyzed the improvement of DIO phenotypes by CREBH overexpression described in our previous report (Nakagawa et al., 2014), using CREBH-Tg, Fgf21 and CREBH-Tg Fgf21 mice. Anti-obesity effect of CREBH was mediated by the activation of FGF21. Deficiency of FGF21 in CREBH-Tg mice canceled out CREBH-induced obesity phenotypes, confirming that FGF21 is crucial in CREBH-mediated amelioration of DIO. However, CREBH-Tg Fgf21 mice still sustained some improved phenotypes such as the suppression of inflammation in eWAT and amelioration of insulin resistance. These phenotypes are FGF21 independent. Notably, we identified that Kiss1 is a potential novel target of CREBH to improve insulin sensitivity. Collectively, we revealed that CREBH ameliorates DIO phenotypes in FGF21-dependent and FGF21-independent mechanisms.

There are the following pieces of previous literature to support the mechanism by which FGF21 mediates the phenotypes of CREBH-Tg mice. The phenotypes of high-fat diet-induced obesity were improved in FGF21-Tg mice (Owen et al., 2014). These phenotypes are reported to be canceled with the deficiency of βKlotho, a component of the FGF21 receptor complex, in both the hypothalamus and dorsal-vagal complex of FGF21-Tg mice (Owen et al., 2014). Secreted FGF21 can activate the sympathetic nervous system and activate lipolysis in WATs and thermogenesis by the conversion of WAT adipocytes to beige adipocytes. FGF21 directly activates PGC-1α and in turn increases Ucp1 expression in adipocytes, leading to an increase in thermogenesis in these cells (Fisher et al., 2012). Consistent with these reports, CREBH-Tg mice revealed a marked increase of UCP1 mRNA and protein in the iWAT, subsequently increasing energy expenditure compared with WT mice. These changes were canceled in Fgf21 mice, thereby confirming that FGF21 mainly contributes to the thermogenic and anti-obesity phenotypes in CREBH-Tg mice.

It should be noticed that obesity mouse models, including ob/ob, db/db, and DIO mice, showed FGF21 resistance because of down-regulation of βKlotho (Hale et al., 2012). Even though CREBH-Tg mice have chronic higher plasma FGF21 levels, CREBH-Tg overexpression kept glucose and insulin levels lower in DIO, ob/ob, and dbdb mice (data not shown). In the current study, plasma glucose- and insulin-lowering effects of CREBH were observed even in DIO-induced CREBH-Tg Fgf21 and ob/ob; CREBH-Tg mice. Taken together, we hypothesized FGF21-independent pathway in improving insulin sensitivity by CREBH to identify Kiss1.

On feeding with high-fat diet, mice overexpressing Cidea in WATs develop obesity, but enhance insulin sensitivity, suggesting that these mice develop metabolic healthy obesity (Abreu-Vieira et al., 2015). Likewise, CREBH overexpression increased Cidea expression in eWAT and iWAT of both WT and Fgf21 mice. Although BWs of CREBH-Tg Fgf21 mice were similar to those of Fgf21 mice, CREBH-Tg Fgf21 mice kept better insulin response than Fgf21 mice. Thus, Cidea induction in adipose tissues might be a contributing factor to insulin sensitivity in CREBH-Tg mice.

CREBH overexpression was associated with the suppression of macrophage infiltration in eWAT of both WT and Fgf21 background mice. Deficiency of FGF21 induced genes related to macrophages and inflammatory cytokines, but CREBH overexpression reversed these genes to the levels of WT mice. FGF21 activates Nrf2/HO-1 and suppresses nuclear factor (NF)-κB in macrophages, thereby inducing anti-inflammatory effects by enhancing Nrf2-mediated anti-oxidant capacity and suppressing NF-κB signaling pathway (Yu et al., 2016). The effects of CREBH overexpression on macrophage infiltration are independent of FGF21. Inflammation of macrophages in eWAT is known to affect insulin sensitivity, one of the major contributors to obesity-mediated metabolic disturbances. Consistently, Fgf21 mice showed both high glucose and insulin levels during GTT, indicating that Fgf21 mice will likely progress to severe impairment of glucose response.

We identified Kiss1 as a novel CREBH target gene responsible for better insulin sensitivity and glucose metabolism even in the absence of FGF21. Besides the major well-known role as a hypothalamus neuropeptide sex hormone, there have been some conflicting reports on the functions of Kiss1 in insulin action: whereas it is reported to inhibit or activate insulin secretion from the islets (Hauge-Evans et al., 2006, Song et al., 2014), Kiss1 receptor KO mice exhibit adiposity, hyperleptinemia, and reduced energy expenditure (Tolson et al., 2016), thereby leading to glucose intolerance and obesity. These reports suggest that Kisspeptin signaling is important in peripheral tissues such as WAT and BAT, activating energy expenditure through thermogenesis. Exogenous Kiss1 administration to monkeys increases plasma adiponectin levels (Wahab et al., 2011). Adiponectin cooperates with FGF21 to improve glucose metabolism and insulin sensitivity in mice (Lin et al., 2013). These support a possibility that the effects of CREBH on adipose tissues partially contribute to Kiss1. Consistently. the knockdown of Kiss1 in CREBH-Tg Fgf21 mice partially canceled CREBH-mediated glucose-lowering effects in ITT, proposing that Kiss1 might be a CREBH-mediated insulin sensitizer. However, we need to further investigate the roles of Kiss1 on energy metabolism.

In this study, we revealed that CREBH increases hepatic expression and plasma levels of FGF21, leading to the activation of browning and subsequent thermogenesis in iWAT. CREBH suppresses inflammation of eWAT independent of FGF21. Moreover, CREBH induces its novel target Kiss1 expression and then activates insulin sensitivity. Finally, we propose that hepatic CREBH improves DIO-mediated dysfunctions in peripheral tissues by improving systemic energy metabolism in FGF21-dependent and FGF21-independent mechanism. These findings indicate that CREBH has pleiotropic effects on obesity and diabetes. Further research to identify further unknown factors in FGF21-independent pathway should be the next project.

Limitations of the Study

CREBH has pleiotropic effects on obesity and diabetes in FGF21-dependent and FGF21-independent mechanisms. We evaluated CREBH-mediated mechanisms using CREBH-Tg Fgf21 mice. In this study, we identified that Kiss1 is a CREBH-mediated FGF21-independent insulin sensitizer. However, there are still unknown CREBH-mediated FGF21-independent mechanisms. Further research to identify further unknown factors in FGF21-independent pathway should be the next project.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.