SCF^JFK is functionally linked to obesity and metabolic syndrome

Abstract

Dysregulation of lipid metabolism could lead to the development of metabolic disorders. We report here that the F‐box protein JFK promotes excessive lipid accumulation in adipose tissue and contributes to the development of metabolic syndrome. JFK transgenic mice develop spontaneous obesity, accompanied by dyslipidemia, hyperglycemia, and insulin resistance, phenotypes that are further exacerbated under high‐fat diets. In contrast, Jfk knockout mice are lean and resistant to diet‐induced metabolic malfunctions. Liver‐specific reconstitution of JFK expression in Jfk knockout mice leads to hepatic lipid accumulation resembling human hepatic steatosis and nonalcoholic fatty liver disease. We show that JFK interacts with and destabilizes ING5 through assembly of the SCF complex. Integrative transcriptomic and genomic analysis reveals that the SCF^JFK‐ING5 axis interferes with AMPK activity and fatty acid β‐oxidation, leading to the suppression of hepatic lipid catabolism. Significantly, JFK is upregulated and AMPKα1 is down‐regulated in liver tissues from NAFLD patients. These results reveal that SCF^JFK is a bona fide E3 ligase for ING5 and link the SCF^JFK‐ING5 axis to the development of obesity and metabolic syndrome.

JFK destabilizes ING5 through assembly of the SCF complex. SCF^JFK‐ING5 axis interferes with AMPK activity and fatty acid β‐oxidation to regulate hepatic lipid metabolism, supporting a functional link of JFK to the development of obesity and NAFLD.

Introduction

Metabolic syndrome is a cluster of metabolic disorders including obesity, hyperlipidemia, hyperglycemia, and insulin resistance that are attributable to a constellation of interrelated metabolic risk factors (Grundy, 2004). These disorders can develop into a set of serious pathological states including nonalcoholic fatty liver disease (NAFLD), type 2 diabetes (T2D), and cardiovascular diseases (Lonardo et al, 2015). The liver functions in maintaining lipid homeostasis by regulating the uptake/synthesis and lipolysis/β‐oxidation under physiological conditions, and dysregulation of lipid metabolism can cause excessive lipid accumulation that promotes development of hepatic steatosis and NAFLD (Younossi et al, 2018; Hodson & Gunn, 2019). Chronic liver diseases are strongly associated with systemic metabolic dysfunctions including obesity, dyslipidemia, insulin resistance, and even cardiovascular diseases, and 25% of the adult population worldwide are currently estimated to suffer from hepatic steatosis and/or NAFLD (Cotter & Rinella, 2020). However, despite a continued effort in understanding the pathogenesis of metabolic syndrome and NAFLD, it is not yet clear how imbalances in fatty acid metabolism may contribute to these pathological states.

Adenosine 5′‐monophosphate (AMP)‐activated protein kinase (AMPK) has been demonstrated as a major regulator of lipid homeostasis; dysregulation of AMPK leads to the alteration of lipid metabolism in the liver, skeletal muscle, adipose tissues, and the hypothalamus, which contributes to development of metabolic disorders (Zhang et al, 2009). Biochemically, we know that AMPK is activated by liver kinase B1 (LKB1) (Woods et al, 2003) and calmodulin‐dependent kinase kinase‐β (CaMKKβ) (Hurley et al, 2005), specifically via phosphorylation of Thr‐172 on the α‐subunit of AMPK. Activated AMPK contributes to fatty acid catabolism by catalyzing the phosphorylation‐mediated inactivation of acetyl‐CoA carboxylase (ACC1 and ACC2), thereby reducing hepatic lipogenesis (Viollet et al, 2009) and increasing fatty acid β‐oxidation (Vishwanath, 2016; Herzig & Shaw, 2018). Recent decades have seen a huge increase in our understanding about the regulatory impacts of targeted proteolysis via the ubiquitin‐proteasome pathway a huge number of cellular processes (Hershko & Ciechanover, 1998). Briefly, protein ubiquitination can be catalyzed by a cascade of enzymes, including ubiquitin‐activating enzymes (E1), ubiquitin‐conjugating enzymes (E2s), and ubiquitin ligases (E3s) (Hershko & Ciechanover, 1998; Pickart, 2004). The best‐characterized mammalian multi‐subunit RING‐finger type of E3 ligases is the SCF (SKP1‐CUL1‐F‐box) complex consisting of SKP1, CUL1, RBX1, and a flexible F‐box protein (FBP) acting as a receptor for substrate recognition and specification (Skaar et al, 2014). To date, 69 FBPs have been identified in humans. In previous studies, we reported that JFK is the only Kelch domain‐containing FBP in humans, and we showed that SCF^JFK promotes breast carcinogenesis through targeting p53 and ING4 for degradation (Sun et al, 2009; Sun et al, 2011; Yan et al, 2015). Nevertheless, the biological function of JFK in animal models is currently unknown.

The ING proteins were initially identified as putative type‐II tumor suppressors, and the five ING proteins in humans and mice have subsequently been implicated in regulation of diverse cellular processes by variously functioning as cell cycle regulators, phospholipid effectors, histone mark sensors, and as core components of several chromatin‐modifying complexes (Coles & Jones, 2009; Tallen & Riabowol, 2014). For instance, ING5 was shown to read H3K4me3 (Chi et al, 2010) and to recruit HBO1 and MOZ/MORF complexes, thereby activating target genes through acetylating H3 or/and H4 tails (Coles & Jones, 2009). Although dysregulation of ING5 has been documented in multiple human diseases (Coles & Jones, 2009), our understanding of specific pathogenic impacts from ING5 remains very limited.

In the current study, using transgenic and knockout mouse models, we show that JFK promotes spontaneous obesity and contributes to the development of metabolic syndrome and NAFLD. We found that SCF^JFK acts as an E3 ligase to destabilize ING5. We demonstrated that the SCF^JFK‐ING5 axis intersects with AMPK activity and fatty acid β‐oxidation to regulate lipid metabolism, and link our mechanistic insights about the functional significance of the SCF^JFK‐ING5 axis to clinical observations in human NAFLD patients.

Results

JFK transgenic mice develop spontaneous obesity and are prone to diet‐induced metabolic disorders

To investigate the biological function of JFK in animal models, we first generated JFK transgenic (JFK ^TG) mice by nuclear transplantation. Briefly, open reading frame (ORF) of human JFK with a FLAG tag was cloned into mammalian expression vector pRP(Exp)‐EF1A. After digestion, linearized DNA was microinjected into the pronuclei of fertilized zygotes derived from C57BL/6 female mice, and the embryos were transplanted onto pseudo‐pregnant mice to breed JFK ^TG founders. The expression of mRNA and protein of JFK in JFK ^TG mice were verified by real‐time reverse transcriptase PCR (qPCR) and Western blotting (Figs 1A and EV1A), respectively.

Figure 1. JFK transgenic mice develop spontaneous obesity and are prone to diet‐induced metabolic disorders.

1. Total RNAs or proteins from tail‐tip of wild‐type and JFK ^TG mice were extracted and analyzed for Jfk expression by qPCR or Western blotting with the indicated antibodies, respectively. Error bars represent mean ± SEM (n = 6, ***P < 0.001, paired two‐tailed Student's t‐test).
2. Measurement of the body weight of wild‐type and JFK ^TG mice under ND in the indicated time of age. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
3. Representative images of wild‐type and JFK ^TG mice fed on ND. Arrows indicate abdominal SAT under MRI analysis. EWAT and PRAT were dissected, and the sections from EWAT were stained with H&E. Scale bar, 100 μm.
4. 17‐week‐old wild‐type or JFK ^TG mice fed on ND were fasted for 6–12 h prior to intraperitoneal injection of glucose or insulin. The blood was collected from the tail vein at different time points for plasma glucose measurement. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
5. The body weight of male wild‐type or JFK ^TG mice fed on HFD. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
6. Representative images of wild‐type or JFK ^TG mice fed on HFD. Arrows indicate abdominal SAT under MRI analysis. EWAT and PRAT were dissected, and the sections from EWAT were stained with H&E. Scale bar, 100 μm.
7. GTT or ITT in 17‐week‐old wild‐type or JFK ^TG mice fed on HFD. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
8. H&E and Oil Red O staining of liver sections from wild‐type or JFK ^TG mice fed on ND or HFD. Scale bar, 100 μm.

Figure EV1. JFK transgenic mice display increased fat mass and decreased lean mass.

1. Total RNAs or proteins from liver tissues of wild‐type and JFK ^TG mice were extracted and analyzed for Jfk expression by qPCR or Western blotting, respectively. Error bars represent mean ± SEM (n = 6, ***P < 0.001, paired two‐tailed Student's t‐test).
2. Measurement of fat and lean mass of wild‐type or JFK ^TG mice. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).

JFK^TG mice were born alive following Mendelian inheritance and appeared to be grossly normal. Strikingly, however, JFK ^TG mice displayed significantly higher body weight, increased fat mass, and decreased lean mass compared with wild‐type littermates under normal diet (ND), irrespective of the gender (Figs 1B and EV1B), although their daily food intake and water drink were comparable (Table 1), implying that the discrepancy in body weight between these two groups was not due to hyperphagia. Magnetic resonance imaging (MRI) revealed that the gain of body weight in JFK ^TG mice was accompanied by expansions of subcutaneous adipose tissue (SAT) and increased fat mass in visceral adipose tissue (VAT), including epididymal white adipose tissue (EWAT) and perirenal adipose tissue (PRAT; Fig 1C). Histological analysis with hematoxylin–eosin (H&E) staining showed that the adipocytes from EWAT of JFK ^TG mice were enlarged and contained bigger lipid droplets (Fig 1C and Table 1). Notably, ND‐fed JFK ^TG mice displayed increased levels of plasma triglyceride (TG), low‐density lipoprotein (LDL), and non‐esterified fatty acid (NEFA), while exhibited reduced levels of plasma high‐density lipoprotein (HDL) and lipoprotein lipase (LPL), compared to wild‐type animals (Table 1), suggesting that JFK might also be functionally associated with a systemic dysfunction of metabolism. In support of this, glucose tolerance test (GTT) and insulin tolerance test (ITT) in ND‐fed JFK ^TG mice, in which the mice were fasted and administered with an intraperitoneal injection of glucose or insulin, respectively, followed by blood collection from tail veins at different time points for plasma glucose measurement showed that JFK ^TG mice had a significant increase in basal fasting glucose level and compromised glucose tolerance, as well as insulin resistance, at all time points tested (Fig 1D), which may be due to the increases in body weight and fat mass. Moreover, JFK ^TG mice also exhibited a significant increase in plasma insulin and a decrease in adiponectin (Table 1). Together, these observations suggest a functional link of JFK to metabolic syndrome.

Table 1.

Physiology and biochemistry of wild‐type (WT) and JFK ^TG mice under normal diet (ND) or high‐fat diet (HFD).

	ND WT (n = 6)	ND JFKTG (n = 6)	P value	HFD WT (n = 6)	HFD JFKTG (n = 6)	P value
Body weight (g)	29.45 ± 1.19	33.17 ± 1.76	0.005	38.53 ± 1.74	49.70 ± 1.52	< 0.001
Food intake (g/day)	2.05 ± 0.12	2.16 ± 0.13	0.176	2.82 ± 0.08	2.79 ± 0.10	0.559
Water drink (ml/day)	1.94 ± 0.12	2.12 ± 0.15	0.059	2.52 ± 0.09	2.55 ± 0.08	0.633
EWAT (g)	0.83 ± 0.08	1.08 ± 0.08	0.011	1.55 ± 0.11	2.23 ± 0.16	0.048
PRAT (g)	0.09 ± 0.02	0.19 ± 0.24	< 0.001	0.47 ± 0.06	0.91 ± 0.14	< 0.001
Liver (g)	1.47 ± 0.05	1.81 ± 0.08	< 0.001	2.03 ± 0.17	2.37 ± 0.07	0.007
Total cholesterol (mmol/l)	3.38 ± 0.17	3.63 ± 0.17	0.130	4.09 ± 0.17	5.82 ± 0.49	< 0.001
Cholesterol (mmol/l)	1.93 ± 0.10	2.17 ± 0.09	0.018	2.65 ± 0.13	4.70 ± 0.43	< 0.001
TG (mmol/l)	0.21 ± 0.03	0.37 ± 0.05	< 0.001	0.51 ± 0.09	1.17 ± 0.11	< 0.001
HDL cholesterol (mmol/l)	1.83 ± 0.06	1.67 ± 0.09	0.015	1.19 ± 0.16	0.66 ± 0.82	0.002
LDL cholesterol (mmol/l)	0.69 ± 0.07	0.85 ± 0.08	0.001	1.15 ± 0.09	1.54 ± 0.09	0.003
NEFA (μmol/l)	128.24 ± 5.44	139.43 ± 5.54	0.008	246.11 ± 23.82	404.25 ± 35.06	< 0.001
Insulin (ng/ml)	0.94 ± 0.10	1.15 ± 0.13	0.005	1.80 ± 0.16	3.27 ± 0.38	< 0.001
Lipoprotein lipase (pg/ml)	671.63 ± 18.31	644.25 ± 22.36	0.020	450.88 ± 18.75	256.80 ± 38.79	< 0.001
Adiponectin (μg/l)	72.80 ± 2.58	63.98 ± 3.58	0.003	47.69 ± 3.69	27.76 ± 5.92	0.002

Bold values denote statistical significance at the P < 0.05 level.

To further explore the role of JFK in animal physiology especially in metabolism, we modeled diet‐induced metabolic behaviors in rodents by feeding 5‐week‐old male (equivalent to 10‐year‐old children) wild‐type and JFK ^TG mice with a high‐fat diet (HFD) containing 45% fat for 12 weeks to eliminate estrogen‐mediated protection from metabolic disorders in females (Stubbins et al, 2012). We found that JFK ^TG mice rapidly gained significantly higher weight under HFD (Fig 1E), despite of comparable daily food intake and water drink to wild‐type mice (Table 1). Meanwhile, JFK ^TG mice had significantly increased SATs as well as fat mass in VATs, including EWAT and PRAT, compared to wild‐type animals (Fig 1F and Table 1). Consistently, H&E staining of EWAT showed a significant enlargement in adipocyte size in JFK ^TG mice (Fig 1F). Together, these observations indicate that JFK promotes lipid accumulation in SATs and VATs in mice under normal diets and exacerbates the situation in mice fed on HFD.

Further analysis showed that the exacerbated obesity in HFD‐fed JFK ^TG mice was accompanied by increased plasma levels of TG, LDL, and NEFA and decreased plasma levels of HDL and LPL, compared to wild‐type animals (Table 1). Moreover, GTT and ITT in HFD‐fed JFK ^TG mice showed that JFK ^TG mice had a significant increase in basal fasting glucose level and compromised glucose tolerance, as well as insulin resistance (Fig 1G). JFK ^TG mice also exhibited a significantly increased level of plasma insulin and decreased level of adiponectin (Table 1). These observations reflect, at least to some extent, the phenotypical manifestations of human metabolic syndrome characterized by obesity, dyslipidemia, hyperglycemia, and insulin resistance (Heindel et al, 2017).

The liver is the central hub of the body's metabolic activity (Rui, 2014). To further support the function of JFK in metabolic processes and its contribution to the development of metabolic syndrome, we thus next investigated the effect of JFK on liver function in JFK ^TG mice. Anatomical analysis showed that JFK ^TG mice had an increase in the gross weight of the liver under ND and HFD (Table 1). Histological analysis by H&E and Oil Red O staining showed that the liver of JFK ^TG mice had widespread lipid‐enriched hepatocytes and vacuoles, which were exacerbated by HFD feeding, resembling the pathological changes in hepatic steatosis (Fig 1H). Collectively, the above results support a notion that JFK promotes excessive lipid accumulation and contributes to the development of metabolic syndrome.

Jfk knockout mice are resistant to diet‐induced obesity and metabolic disorders

To gain further supports of the role of JFK in lipid metabolism, we also generated Jfk knockout (Jfk ^KO) mice by gene targeting in mouse embryonic stem cells (mESCs) using the Cre/LoxP recombination system. Briefly, homozygous floxed Jfk line (Jfk ^flox/flox) containing LoxP sequences flanking exon 5 of Jfk was produced and crossed with Cre line to disrupt Jfk through Cre‐mediated recombination to breed heterozygous and subsequently homozygous Jfk knockout (Jfk ^KO) mice. Jfk expression in Jfk ^KO mice was verified by qPCR and Western blotting (Fig 2A).

Figure 2. Jfk^KO mice are resistant to diet‐induced obesity and hepatic steatosis.

1. Total RNAs or proteins from tail‐tip of wild‐type or Jfk ^KO mice were extracted and analyzed for Jfk expression by qPCR or Western blotting with the indicated antibodies, respectively. Error bars represent mean ± SEM (n = 6, ***P < 0.001, paired two‐tailed Student's t‐test).
2. The body weight of wild‐type or Jfk ^KO mice under ND was measured at the indicated time of age. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
3. Representative images of wild‐type or Jfk ^KO mice fed on ND. Arrows indicate abdominal SAT under MRI. EWAT and PRAT were dissected, and the sections from EWAT were stained with H&E. Scale bar, 100 μm.
4. The body weight of male wild‐type or Jfk ^KO mice fed on HFD. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
5. Representative images of wild‐type or Jfk ^KO mice fed on HFD. Arrows indicate abdominal SAT under MRI. EWAT and PRAT were dissected, and the sections from EWAT were stained with H&E. Scale bar, 100 μm.
6. GTT or ITT in 17‐week‐old wild‐type or Jfk ^KO mice fed on HFD. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
7. H&E and Oil Red O staining of liver sections from wild‐type or Jfk ^KO mice fed on HFD. Scale bar, 100 μm.
8. Wild‐type or Jfk ^KO mice were fed with HFD for 11 weeks prior to injection of adenovirally delivered vector or JFK via tail vein. Seven days after injection, total proteins prepared from liver tissues were subjected to Western blotting analysis for the expression of Jfk or measurement for hepatic TG and NEFA content. Error bars represent mean ± SEM (n = 6, *P < 0.05, one‐way ANOVA with Tukey's HSD test). H&E and Oil Red O staining of liver sections are shown. Scale bar, 100 μm.
9. Wild‐type, Jfk ^KO, or Jfk ^KO mice with liver‐specific reconstitution of JFK were housed with free access to water but no food during indirect calorimetry. The white or gray horizontal bars depict the light or dark period in a 12:12‐h light–dark cycle, respectively. EE or RER was calculated and plotted at 1‐h intervals and averaged. Error bars represent mean ± SEM (n = 6, *P < 0.05, one‐way ANOVA with Tukey's HSD test).

Jfk^KO mice gave birth to pups with a normal Mendelian ratio and appeared to be grossly normal, and their daily food intake and water drink were roughly the same as wild‐type littermates (Table 2). However, in contrast to JFK ^TG mice, Jfk ^KO animals had lean bodies under ND, regardless of the gender (Fig 2B), exhibited decreased abdominal SATs, as evidenced by MRI, and displayed reduced fat mass and a smaller adipocyte size in EWATs (Fig 2C and Table 2), compared to wild‐type littermates. GTT and ITT in Jfk ^KO mice showed that Jfk ^KO mice had no significant increase in basal fasting glucose level, glucose tolerance, or insulin resistance (Fig EV2A). When fed on HFD, Jfk ^KO mice showed resistant to diet‐induced obesity (Fig 2D) and were protected against the excessive fat accumulation in SATs and VATs including EWAT and PRAT (Fig 2E and Table 2). Histological analysis showed that the adipocytes from EWAT of Jfk ^KO mice fed on HFD were even smaller in size than that in control animals (Fig 2E). Importantly, the resistance to obesity in HFD‐fed Jfk ^KO mice was accompanied by decreases in plasma levels of TG, LDL, and NEFA and increases in plasma levels of HDL and LPL, although there were no significant differences except in TG between Jfk ^KO mice and wild‐type littermates under ND (Table 2). In addition, GTT and ITT showed that Jfk ^KO mice fed on HFD had a marked decrease in basal fasting glucose level and increase in glucose tolerance and insulin sensitivity at all time points tested (Fig 2F). Moreover, HFD‐fed Jfk ^KO animals had a decreased plasma level of insulin and an increased plasma level of adiponectin (Table 2), although these plasma biochemical values in ND‐fed Jfk ^KO mice seemed to be normal (Table 2). Furthermore, anatomical analysis revealed that the gross weight of the liver decreased in Jfk ^KO mice under HFD compared with wild‐type animals (Table 2), and H&E and Oil Red O staining revealed that hepatic steatosis‐like widespread lipid‐enriched hepatocytes and vacuoles seen in wild‐type littermates were relieved in the liver from HFD‐fed Jfk ^KO (Fig 2G).

Table 2.

Physiology and biochemistry of wild‐type (WT) and Jfk ^KO mice under normal diet (ND) or high‐fat diet (HFD).

	ND WT (n = 6)	ND JfkKO (n = 6)	P value	HFD WT (n = 6)	HFD JfkKO (n = 6)	P value
Body weight (g)	29.5 ± 1.19	27.45 ± 0.94	0.018	38.2 ± 1.80	30.27 ± 1.39	< 0.001
Food intake (g/day)	2.10 ± 0.13	2.06 ± 0.13	0.577	2.75 ± 0.07	2.70 ± 0.07	0.386
Water drink (ml/day)	1.93 ± 0.11	1.96 ± 0.06	0.341	2.50 ± 0.14	2.53 ± 0.08	0.553
EWAT (g)	0.80 ± 0.06	0.71 ± 0.04	0.025	1.66 ± 0.07	1.21 ± 0.16	0.024
PRAT (g)	0.09 ± 0.03	0.06 ± 0.02	0.125	0.45 ± 0.06	0.29 ± 0.03	< 0.001
Liver (g)	1.48 ± 0.11	1.46 ± 0.09	0.181	2.05 ± 0.11	1.87 ± 0.10	0.045
Total cholesterol (mmol/l)	3.31 ± 0.11	3.33 ± 0.08	0.755	4.08 ± 0.30	3.56 ± 0.13	0.008
Cholesterol (mmol/l)	1.96 ± 0.09	1.92 ± 0.07	0.509	2.57 ± 0.12	2.18 ± 0.10	0.004
TG (mmol/l)	0.20 ± 0.02	0.19 ± 0.03	0.363	0.55 ± 0.05	0.32 ± 0.06	0.001
HDL cholesterol (mmol/l)	1.80 ± 0.07	1.85 ± 0.04	0.176	1.10 ± 0.09	1.35 ± 0.08	0.003
LDL cholesterol (mmol/l)	0.71 ± 0.07	0.69 ± 0.04	0.627	1.10 ± 0.11	0.92 ± 0.08	0.005
NEFA (μmol/l)	130.62 ± 3.24	125.18 ± 3.75	0.052	249.33 ± 14.54	218.33 ± 12.99	0.001
Insulin (ng/ml)	0.90 ± 0.10	0.94 ± 0.09	0.475	1.78 ± 0.07	1.55 ± 0.11	0.001
Lipoprotein lipase (pg/ml)	675.04 ± 11.53	678.73 ± 17.05	0.666	456.87 ± 17.26	490.60 ± 24.36	0.001
Adiponectin (μg/l)	71.95 ± 2.19	72.51 ± 3.73	0.663	48.12 ± 1.72	51.88 ± 2.73	0.008

Bold values denote statistical significance at the P < 0.05 level.

Figure EV2. Liver‐specific reconstitution of JFK offsets the increases in glucose tolerance, insulin sensitivity and energy expenditure, as well as decrease in respiratory exchange ratio in Jfk ^KO mice.

1. GTT or ITT in 17‐week‐old wild‐type or Jfk ^KO mice fed on ND. Error bars represent mean ± SEM (n = 6).
2. Representative infrared thermal images to reflect temperature of body surface in wild‐type or Jfk ^KO mice fed on HFD were shown (left). Rectal temperature of the mice at different times after cold exposure was recorded (right). Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
3. Immunohistochemical staining of Ucp1 in inguinal white adipose tissues from wild‐type or Jfk ^KO mice fed on HFD. Scale bar, 100 μm.
4. Total proteins from inguinal white adipose tissues of wild‐type or Jfk ^KO mice fed on HFD were extracted and analyzed for Ucp1 expression by Western blotting.
5. The body weight of wild‐type, Jfk ^KO, or Jfk ^KO mice with liver‐specific reconstitution of JFK mice fed on HFD. Error bars represent mean ± SEM (n = 6, *P < 0.05, one‐way ANOVA with Tukey's HSD test).
6. Measurement of fat mass of wild‐type, Jfk ^KO, or Jfk ^KO mice with liver‐specific reconstitution of JFK fed on HFD. Error bars represent mean ± SEM (n = 6, *P < 0.05, one‐way ANOVA with Tukey's HSD test).
7. GTT or ITT in 17‐week‐old wild‐type, Jfk ^KO, or Jfk ^KO mice with liver‐specific reconstitution of JFK fed on HFD. Asterisk (*) represents significant comparison to wild‐type mice; octothorpe (#) represents significant comparison to Jfk ^KO mice with liver‐specific reconstitution of JFK. Error bars represent mean ± SEM (n = 6, *P < 0.05, ^# P < 0.05, one‐way ANOVA with Tukey's HSD test).
8. Wild‐type, Jfk ^KO, or Jfk ^KO mice with liver‐specific reconstitution of JFK were housed with free access to water and food, and were subjected to indirect calorimetry. The white or gray horizontal bars depict the light or dark period in a 12:12‐h light–dark cycle, respectively. EE or RER was calculated and plotted at 1‐h intervals and averaged. Error bars represent mean ± SEM (n = 6, *P < 0.05, one‐way ANOVA with Tukey's HSD test).
9. Locomotor activity and food intake in wild‐type, Jfk ^KO, or Jfk ^KO mice with liver‐specific reconstitution of JFK fed ad libitum. Error bars represent mean ± SEM (n = 6).
10. Locomotor activity in wild‐type, Jfk ^KO, or Jfk ^KO mice with liver‐specific reconstitution of JFK challenged with 24 h fasting. Error bars represent mean ± SEM (n = 6).

Considering the obesity resistance in Jfk ^KO mice, we also performed brown adipose tissue (BAT) thermogenesis analysis of wild‐type and Jfk ^KO mice fed on HFD. The results showed that the extent of cold exposure‐induced BAT thermogenesis was significantly higher in Jfk ^KO mice than wild‐type mice (Fig EV2B). Further immunohistochemical staining and Western blotting analysis revealed that the expression level of uncoupling protein 1 (Ucp1) was elevated in inguinal white adipose tissues from Jfk ^KO mice (Fig EV2C and D), suggesting that the increases in BAT thermogenesis and WAT browning could be, at least partially, contributing to the observed decreased in body weight of the Jfk ^KO mice. These results corroborate the observations in JFK ^TG mice, supporting a functional link of JFK to the development of obesity and metabolic disorders.

As neither the JFK ^TG nor Jfk ^KO mice exhibited altered water intake or food intake, we did not focus on the effects of JFK on the nervous system; rather, we re‐introduced JFK expression specifically in the liver via adeno‐associated virus 8 (AAV8)‐mediated gene transfer to further investigate the biological function of hepatic JFK. Although liver‐specific reconstitution of JFK did not restore the decreased body weight or fat mass phenotype in Jfk ^KO mice (Fig EV2E and F), adenovirus‐mediated JFK injection did offset the increase in glucose tolerance and insulin sensitivity (Fig EV2G), and also decreased hepatic TG, and NEFA in Jfk ^KO mice (Fig 2H). Analysis by H&E and Oil Red O staining revealed that livers of Jfk ^KO mice with JFK reconstitution developed NAFLD‐like phenotypes under HFD (Fig 2H).

We also studied the metabolic consequences of Jfk knockout in the liver by monitoring energy expenditure (EE), the respiratory exchange ratio (RER), and locomotor activity, in experiments based on indirect calorimetry of mice fed ad libitum or challenged with 24‐h fasting (Jeukendrup & Wallis, 2005). Jfk ^KO mice exhibited increased EE and decreased RER, whereas liver‐specific reconstitution of JFK led to a reduction in EE and elevation of RER in the Jfk ^KO background (Figs 2I and EV2H); note that no significant differences in locomotor activity or food intake were detected (Fig EV2I and J). It is known that the metabolic switch from glucose to fatty acids by the liver occurs around 12 h after cessation of food intake–the point of negative energy balance at which liver glycogen stores are depleted and fatty acids are mobilized (Berg et al, 2002). We found that Jfk ^KO mice reached maximal rates of lipid oxidation after 4 h of fasting; similar rates were not achieved by until 18 h in wild‐type mice and Jfk ^KO mice with JFK reconstitution (Fig 2I). These results suggest that Jfk ^KO mice are capable of relatively greater fat utilization during the first 12 h of fasting; moreover, these results indicate that liver‐specific reconstitution of JFK in Jfk ^KO mice supports a fuel switch from lipids to carbohydrates. Collectively, these results support a role for JFK in lipid catabolism and indicate that JFK can apparently promote development of metabolic syndrome.

JFK is physically associated with ING5 in hepatocytes

To gain mechanistic insights into JFK‐promoted lipid catabolism and the development metabolic syndrome and NAFLD, we employed affinity purification‐coupled mass spectrometry to investigate the JFK interactome in the liver. In these experiments, hepatocarcinoma cell line HepG2 cells were stably transfected with vector or FLAG‐JFK. Whole cell extracts were prepared and subjected to affinity purification using anti‐FLAG M2 affinity gel. After extensive washing, the bound proteins were eluted with excess FLAG peptides, resolved on SDS–PAGE, and visualized by silver staining. The protein bands on the gel were recovered and analyzed by mass spectrometry. We found that JFK co‐purified with SKP1 and CUL1, two integral components of the SCF complex (Fig 3A), consistently with our previous observations (Sun et al, 2009; Sun et al, 2011; Yan et al, 2015) and confirming the association of JFK with the SCF complex in hepatocytes. Interestingly, additional proteins including ING5 were also identified in the JFK‐containing protein complex (Fig 3A), suggesting that JFK is associated with ING5 in vivo.

Figure 3. JFK is physically associated with ING5 in hepatocytes.

1. HepG2 cells were stably transfected with vector or FLAG‐JFK. Cellular extracts were immunopurified with anti‐FLAG immunoaffinity resin and eluted with FLAG peptides. The eluates were resolved on SDS–PAGE and silver‐stained. The proteins bands were retrieved and analyzed by mass spectrometry.
2. Cellular extracts from HepG2 cells were immunoprecipitated with antibodies against ING5 followed by immunoblotting with the antibodies against the indicated proteins.
3. Extracts from liver tissues of wild‐type mice were immunoprecipitated with antibodies against Ing5 followed by immunoblotting with the antibodies against the indicated proteins.
4. GST pull‐down assays with bacterially expressed GST‐ING5 and in vitro transcribed/translated JFK, CUL1, SKP1, or RBX1.
5. Schematic diagrams of wild‐type JFK and JFK mutants. Cellular extracts from HepG2 cells co‐transfected with Myc‐ING5 and FLAG‐JFK or JFK mutants were immunoprecipitated with anti‐FLAG followed by immunoblotting with anti‐Myc.
6. GST pull‐down assays with bacterially expressed GST‐JFK or JFK mutants and in vitro transcribed/translated ING5.

To verify that ING5 is associated with JFK in vivo in the context of the SCF complex, total proteins from HepG2 cells or mouse liver tissues were extracted and co‐immunoprecipitation experiments were performed. Immunoprecipitation (IP) with antibodies against ING5 and immunoblotting (IB) with antibodies against JFK, CUL1, SKP1, or RBX1 showed that ING5 was indeed efficiently co‐immunoprecipitated with JFK, as well as with the integral components of the SCF complex CUL1, SKP1, and RBX1 (Fig 3B and C). Further analysis by glutathione S‐transferase (GST) pull‐down assays with bacterially expressed GST‐ING5 and in vitro transcribed/translated JFK, CUL1, SKP1, or RBX1 showed that ING5 interacted with JFK, but not CUL1, SKP1, and RBX1 (Fig 3D), in agreement with the receptor and substrate relationship seen in the function of the SCF complex (Cardozo & Pagano, 2004). Moreover, co‐immunoprecipitation in HepG2 cells that were co‐transfected with FLAG‐tagged wild‐type JFK, F‐box‐deleted JFK (JFKΔF‐box), or Kelch domain‐deleted JFK (JFKΔKelch) together with Myc‐tagged ING5 (Myc‐ING5) demonstrated that the Kelch domain of JFK is responsible for its interaction with ING5 (Fig 3E). Furthermore, GST pull‐down assays with bacterially expressed GST‐JFK, GST‐JFKΔF‐box, or GST‐JFKΔKelch and in vitro transcribed/translated ING5 showed that while both wild‐type JFK and JFKΔF‐box were capable of interacting with ING5, JFKΔKelch was not (Fig 3F). These results are consistent with the feature of the SCF ubiquitin ligases in which only the F‐box protein directly interacts with the substrate through its variable protein–protein interaction domain (Cardozo & Pagano, 2004). Together, these results indicate that JFK physically interacts with ING5 via assembly of an SCF complex.

SCF^JFK targets ING5 for degradation

The physical interaction of ING5 with JFK in the context of the SCF complex prompted us to investigate whether ING5 is also destabilized by JFK via the SCF‐dependent mechanism. To this end, HepG2 cells were transfected with the FLAG‐tagged wild‐type or deletion mutants of JFK. Western blotting analysis of cellular lysates revealed that the steady‐state level of ING5 markedly decreased when wild‐type JFK was overexpressed (Fig 4A), whereas overexpression of either JFKΔF‐box or JFKΔKelch did not result in evident changes in the ING5 level (Fig 4A). Meanwhile, there were no detectable changes in ING5 protein steady‐state level when cells were transfected with other F‐box proteins (FBP) such as β‐TRCP, the FBP for CDC25A (Jin et al, 2003), FBW7, the FBP for cyclin E (Koepp et al, 2001), or SKP2, the FBP for p21 (Yu et al, 1998) (Fig 4B), suggesting a specific effect of JFK on ING5 decay.

Figure 4. SCF^JFK targets ING5 for degradation.

1. Western blotting analysis of cellular extracts from HepG2 cells transfected with FLAG‐JFK or JFK mutants.
2. Western blotting analysis of cellular extracts from HepG2 cells transfected with the indicated plasmids.
3. HepG2 cells treated with control siRNAs or siRNAs against CDC20, CUL1, or SKP1 were transfected with vector or FLAG‐JFK. Cellular extracts were prepared for Western blotting analysis. Total RNAs were extracted and analyzed for ING5 expression by qPCR. Error bars represent mean ± SD for triplicate experiments.
4. HepG2 cells were transfected with FLAG‐JFK and treated with MG132 prior to protein extraction for Western blotting analysis.
5. HepG2 cells were transfected with FLAG‐JFK and treated with 50 μg/ml CHX for the indicated times in the presence or absence of MG132 for Western blotting analysis. Quantitation was done by densitometry and expressed as signals of ING5/β‐actin.
6. Cellular extracts from HCT116 p53^+/+ and HCT116 p53^−/− cells transfected with empty vector or FLAG‐JFK were prepared for Western blotting analysis.
7. HepG2 cells were co‐transfected with FLAG‐tagged JFK, JFKΔF‐box, or JFKΔKelch, together with HA‐Ub or UbK48R, and treated with 5 μM MG132 for 12 h prior to in vivo ubiquitination assays with the indicated antibodies.
8. Bacterially expressed GST‐JFK or JFK mutants were incubated with in vitro transcribed/translated ING5 for in vitro ubiquitination assays.

To support this notion, HepG2 cells were treated with specific siRNAs against CUL1 or SKP1 to knock down the expression of the corresponding proteins. Western blotting analysis showed that, in either CUL1‐ or SKP1‐depleted cells, overexpression of JFK was no longer associated with an increased turnover of ING5, whereas in HepG2 cells depleted of CDC20, the activator subunit of the mitotic APC (anaphase‐promoting complex) (Peters, 2006), JFK‐promoted ING5 destruction was not affected (Fig 4C), supporting the argument that JFK specifically promotes ING5 degradation through an SCF‐dependent pathway. The decreased ING5 protein expression under JFK overexpression was not due to transcriptional regulation of ING5, as qPCR measurements indicated that JFK overexpression did not affect ING5 mRNA level (Fig 4C). Moreover, Western blotting analysis showed that the JFK dose‐dependent decay of ING5 in HepG2 cells could be effectively blocked by MG132 (Fig 4D), an inhibitor of proteasome (Lee & Goldberg, 1996). Furthermore, cycloheximide (CHX) chase assays in HepG2 cells that were transfected with FLAG‐JFK revealed that JFK overexpression was associated with a decrease in the half‐life of ING5, an effect that only occurred in the absence of MG132 (Fig 4E). Together, these results indicate that JFK promotes ING5 degradation in an SCF‐dependent proteasome pathway.

It has been reported that ING5 also interacts with p53 and promotes the transactivation of p53 (Shiseki et al, 2003). In light of our previous reports that JFK is also physically associated with p53 and promotes p53 degradation through an SCF‐dependent proteasome pathway (Sun et al, 2009; Sun et al, 2011), we asked the question whether JFK‐mediated ING5 degradation is in some way connected to the function of p53. To test this, HCT116 p53^+/+ and HCT116 p53^−/− cells were transfected with FLAG‐JFK. Western blotting analysis of the cellular lysates revealed that JFK overexpression resulted in a decrease in the level of ING5 protein, regardless of the p53 status (Fig 4F), arguing against the possibility that p53 is functionally linked to JFK‐mediated ING5 degradation.

Given the interaction of JFK with ING5 and the SCF‐dependent ING5 degradation, we next investigated whether JFK‐mediated ING5 ubiquitination. To this end, HepG2 cells were co‐transfected with JFK, JFKΔF‐box, or JFKΔKelch, together with ubiquitin (Ub) or ubiquitin mutant UbK48R which is defective in ubiquitin chain elongation (Rodrigo‐Brenni et al, 2010). The results showed that wild‐type JFK, but not JFKΔF‐box or JFKΔKelch, elevated ING5 polyubiquitination in vivo (Fig 4G). However, co‐transfection of UbK48R, not wild‐type ubiquitin, blocked JFK‐promoted ING5 polyubiquitination (Fig 4G). Consistently, in vitro ubiquitination assays with bacterially expressed GST‐tagged JFK, JFKΔF‐box, or JFKΔKelch and in vitro transcribed/translated ING5 showed that wild‐type JFK, but not JFKΔF‐box or JFKΔKelch, promotes ING5 polyubiquitination (Fig 4H). Collectively, these observations indicate that JFK targets ING5 for polyubiquitination and degradation and that SCF^JFK is a bond fide E3 ubiquitin ligase for ING5.

Genome‐wide identification of transcriptional targets for SCF^JFK‐ING5 axis

Given that ING5 reads H3K4me3 on chromatin to activate target gene transcription (Chi et al, 2010; Lalonde et al, 2013), we thus first performed chromatin immunoprecipitation‐coupled massive parallel DNA sequencing (ChIP‐seq) to analyze the genome‐wide transcriptional profile of ING5 and investigate the direct transcription targets of the SCF^JFK‐ING5 axis in hepatocytes. In these experiments, ChIP was performed first in HepG2 cells with antibodies against ING5. Following ChIP, ING5‐associated DNAs were amplified using non‐biased conditions, labeled, and sequenced with HiSeq 2500. The raw sequencing image data were examined by the Illumina analysis pipeline, aligned to the unmasked human reference genome (GRCh37, hg19) using Bowtie 2, and further analyzed by MACS (Model‐based Analysis for ChIP‐Seq) with the cutoff of P value < 1e‐5 and FDR < 0.001. After filtering through normal IgG, a total of 7,882 ING5‐specific binding peaks were called, with the majority of the peaks located at promoter sequences (88.9%) and the rest at intronic (5.2%) and distal intergenic (3.5%) regions (Fig 5A). The genomic landscape of ING5 binding peaks at promoter sequences was analyzed by Deeptools (Fig 5B).

Figure 5. Genome‐wide identification of transcriptional targets for SCF^JFK‐ING5 axis.

1. Genomic distribution of ING5 determined by ChIP‐seq analysis. ChIP was performed in HepG2 cells with antibodies against ING5 followed by deep sequencing. The raw data were analyzed by MACS with the cutoff of P value < 1e‐5 and FDR < 0.001.
2. Binding density of ING5 across its specific binding sites visualized by Deeptools. The heatmap shows the normalized ChIP‐seq tag counts ordered by signal strength.
3. Venn diagram from cross‐analysis of down‐regulated genes obtained from RNA‐seq analysis in JFK‐overexpressing and ING5‐depleted HepG2 cells with genes targeted by ING5 from ChIP‐seq. GO enrichment analysis of 140 potential target genes using DAVID is shown.
4. ChIP‐seq track at the representative promoter of target genes. The red rectangles indicate the peak regions/binding sites of ING5 on the target genes promoter.

To further explore the functional significance of SCF^JFK‐mediated ING5 degradation, we next performed genome‐wide analysis of the transcriptome in hepatocytes that is regulated by the SCF^JFK‐ING5 axis using RNA‐based deep sequencing (RNA‐seq). To this end, HepG2 cells were transfected with vector or FLAG‐JFK or treated with control siRNA or ING5 siRNA. Total mRNAs were extracted for cDNA synthesis, library construction, and sequencing using HiSeq 2500. The quality of raw data was examined by FastQC, and sequencing adapter and low‐quality reads, i.e., those with more than five “N” bases and mean Phred quality score < 15, were removed through fastp. Spliced Transcripts Alignment to a Reference (STAR) software was then utilized to align clean reads to the unmasked human reference genome (GRCh37, hg19). Raw counts of the reads mapped to genes were selected to analyze differentially expressed genes using DESeq2 Bioconductor package with FDR < 0.001 as the threshold. We identified 235 genes that were down‐regulated when JFK was overexpressed and 759 genes that were down‐regulated upon ING5 knockdown in HepG2 cells (Fig 5C), as we reasoned that these genes most likely represent the transcriptional readout of the SCF^JFK‐ING5 axis in which SCF^JFK mediates the degradation of ING5.

The down‐regulated genes obtained from RNA‐seq experiments in JFK‐overexpressing and ING5‐depleted HepG2 cells were then cross‐analyzed with the potential target genes corresponding to those peaks located at promoter sequences (≤ 1 kb) from ChIP‐seq for overlapped targets. This generated a total of 140 genes that were considered to be the targets of the SCF^JFK‐ING5 pathway (Fig 5C). These 140 genes were then assigned to various cellular biological processes using DAVID (the Database for Annotation, Visualization and integrated Discovery, http://david.ncifcrf.gov/). Strikingly, lipid metabolism, fatty acid metabolism, and fatty acid β‐oxidation represented by ACOX1, HADH, ACAD11, and AMPKα1 (McFadden & Corl, 2009; He et al, 2011; Heslegrave & Hussain, 2013; Song et al, 2018) were affected by JFK overexpression or ING5 knockdown (Fig 5C). The ChIP‐seq peak data for representative target genes ACOX1, HADH, ACAD11, and AMPKα1 are shown (Fig 5D). Together, these observations suggest that the SCF^JFK‐ING5 pathway intersects with AMPK activity and fatty acid β‐oxidation.

The SCF^JFK‐ING5 axis intersects with AMPK activity and fatty acid β‐oxidation to regulate hepatic lipid metabolism

To further support our observations, quantitative ChIP (qChIP) analysis was performed in HepG2 cells using specific antibodies against ING5 on the promoters of ACOX1, HADH, ACAD11, and AMPKα1. The results indicated that ING5 was highly enriched on these promoters, validating the ChIP‐seq experiments (Fig 6A). qChIP analysis also showed that the level of both H3K14ac and H4ac markedly increased at the promoters of ACOX1, HADH, ACAD11, and AMPKα1 in HepG2 cells upon overexpressing ING5 or knocking down JFK (Fig 6A), consistent with the reports that ING5 acetylates histone H3K14 and H4 through the assembly of the HBO1/JADE/ING5 or MOZ/MORF/BRPF/ING5 histone acetyltransferase complex (Ullah et al, 2008; Feng et al, 2016). In addition, knockdown of JFK in HepG2 cells resulted in an increased mRNA expression of ACOX1, HADH, ACAD11, and AMPKα1, effects that were offset by simultaneous knockdown of ING5, whereas overexpression of JFK led to a decreased expression of these target genes, which could be rescued by simultaneous overexpression of ING5 (Fig 6B). Moreover, depletion of JFK led to an elevation of not only the protein levels of ACOX1, HADH, ACAD11, and AMPKα1, but also the activity of AMPKα1, evidenced by increased p‐AMPKα and p‐ACC, effects that were abolished by co‐depletion of ING5 (Fig 6C), whereas overexpression of JFK was associated with a reduced level of AMPKα1, ACOX1, p‐AMPKα, and p‐ACC, which were, at least partially, rescued by simultaneous overexpression of ING5 (Fig 6C).

Figure 6. The SCF^JFK‐ING5 axis intersects with AMPK activity and fatty acid β‐oxidation to regulate hepatic lipid metabolism.

1. HepG2 cells were transfected with control or ING5, or treated with siRNAs against JFK. Soluble chromatin was prepared, and qChIP analysis was performed on the selected promoters using the indicated antibodies. Histone H3 was detected as an internal control. Error bars represent mean ± SD for triplicate experiments (*P < 0.05, paired two‐tailed Student's t‐test).
2. HepG2 cells were treated with control siRNAs or siRNAs against JFK or/and ING5, or transfected with vector or JFK or/and ING5. Total RNAs were extracted, and qPCR analysis was performed. Error bars represent mean ± SD for triplicate experiments (*P < 0.05, paired two‐tailed Student's t‐test).
3. HepG2 cells were treated with control siRNAs or siRNAs against JFK or/and ING5, or transfected with vector, JFK, or/and ING5. Cellular extracts were prepared for Western blotting analysis with the indicated antibodies.
4. HepG2 cells were transfected with vector or JFK, AMPKα1, or/and ING5, or treated with control siRNAs or siRNAs against JFK, AMPKα1, or/and ING5. Cellular extracts were prepared for AMPK activity assays by ELISA. Error bars represent mean ± SD for triplicate experiments (*P < 0.05, paired two‐tailed Student's t‐test).
5. HepG2 cells were transfected with vector, JFK, or/and ING5, or treated with control siRNAs or siRNAs against JFK, AMPKα1, or/and ING5. Mitochondria were isolated, and the rate of fatty acid β‐oxidation was measured by spectrophotometry of ferricyanide‐trapped reduced products. Error bars represent mean ± SD for triplicate experiments (*P < 0.05, paired two‐tailed Student's t‐test).
6. HepG2 cells were transfected with vector, JFK, or/and ING5, or treated with control siRNAs or siRNAs against JFK, AMPKα1, or/and ING5. Culture media was collected, and the concentration of 3‐hydroxybutyric acid and acetoacetic acid was measured by ketone body assay kit. Error bars represent mean ± SD for triplicate experiments (*P < 0.05, paired two‐tailed Student's t‐test).

Given the regulation of AMPKα1 expression and the phosphorylation/activation of AMPKα1 by the SCF^JFK‐ING5 axis, we next investigated whether the SCF^JFK‐ING5 axis impacts on the AMPK signaling pathway. To this end, HepG2 cells were transfected with JFK or/and ING5 or AMPKα1, or treated with siRNAs against JFK or/and ING5 or AMPKα1, and the activation of the AMPK signaling was measured using direct AMPK kinase assay kit. We found that the enzymatic activity of AMPK decreased upon either JFK overexpression or ING5 knockdown and increased when either JFK was depleted or ING5 was overexpressed (Fig 6D). However, the decreased AMPK activity associated with JFK overexpression was rescued when either ING5 or AMPKα1 was also overexpressed, although to a different extent (Fig 6D). Likewise, the increased AMPK activity associated with JFK depletion was offset when either ING5 or AMPKα1 was also depleted (Fig 6D). These observations are consistent with our working model that JFK targets ING5 for degradation to regulate AMPKα1 expression and influence AMPK activity.

In light of the regulation of the expression of the fatty acid β‐oxidation‐related genes ACOX1, HADH, and ACAD11 and of the phosphorylation of ACC by the SCF^JFK‐ING5 axis, we thus tested whether the SCF^JFK‐ING5 axis influences fatty acid β‐oxidation. For this purpose, the rate of fatty acid β‐oxidation was measured in mitochondria isolated from HepG2 cells transfected with JFK or/and ING5, or treated with siRNAs against ING5, AMPKα1, or/and JFK, by spectrophotometry of ferricyanide‐trapped reduced products generated by acyl‐CoA dehydrogenases in β‐oxidation (Osmundsen, 1981; Meng et al, 2018). We found that the rate of fatty acid β‐oxidation significantly decreased in either JFK‐overexpressing or ING5‐deficient cells and markedly increased in either JFK‐deficient or ING5‐overexpressing cells (Fig 6E). However, the decreased fatty acid β‐oxidation rate upon JFK overexpression was rescued when ING5 was also overexpressed, and the increased fatty acid β‐oxidation rate upon JFK depletion was offset when either ING5 or AMPKα1 was simultaneously depleted (Fig 6E). These observations indicate that the SCF^JFK‐ING5 axis inhibits fatty acid β‐oxidation.

To further support this notion, we next measured, using a ketone body assay kit, the concentration of 3‐hydroxybutyric acid (OHB) and acetoacetic acid (AcAc), two main ketone bodies produced from fatty acid oxidation (Jung et al, 2018), in culture media of HepG2 cells that were transfected with JFK or/and ING5, or treated with siRNAs against ING5, AMPKα1, or/and JFK. We found that JFK overexpression resulted in a significant decrease in the concentration of OHB and AcAc, an effect that was rescued when ING5 was also overexpressed. On the other hand, JFK depletion led to a significant increase in the concentration of OHB and AcAc, an effect that was offset when either ING5 or AMPKα1 was simultaneously depleted (Fig 6F). These results indicate that the SCF^JFK‐ING5 axis inhibits ketogenesis, consistent with its role in fatty acid β‐oxidation. Collectively, these results support a theme in which SCF^JFK, through destabilizing ING5 and inhibiting AMPK activity and fatty acid β‐oxidation, promotes lipid accumulation and the development of metabolic syndrome and NAFLD.

Pathophysiological significance of the SCF^JFK‐ING5 axis

To extend our observations to pathophysiologically relevant contexts, we first profiled the expression pattern of Jfk and Ing5 in mouse tissues. Western blotting analysis in multiple tissues collected from wild‐type mice showed that Jfk was expressed in various tissues, with the highest expression in adipose tissue and the lowest expression in lung (Fig 7A). Remarkably, the expression of Jfk overall exhibited a reverse trend relative to the expression of Ing5 (Fig 7A). Importantly, analysis by Western blotting in mouse embryonic fibroblasts (MEFs) as well as livers from Jfk ^KO mice showed the expression of Ing5 was elevated, compared to that in these cells/tissues from wild‐type animals (Fig 7B), whereas the levels of Ing5 mRNA were comparable in these cells/tissues between Jfk ^KO mice and wild‐type animals, as measured by qPCR (Fig 7C). In addition, the level of Acox1, Hadh, Acad11, Ampkα1, p‐Ampkα, and p‐Acc was also elevated in MEFs and livers from Jfk ^KO mice, which were offset by simultaneous reconstitution of JFK, whereas the level was reduced in livers from JFK ^TG mice (Figs 7B and C, and EV3A and B). Significantly, Ampk activity and the rate of fatty acid β‐oxidation were also elevated in MEFs and livers from Jfk ^KO mice (Fig 7D).

Figure 7. Pathophysiological significance of the SCF^JFK‐ING5 axis.

• A
  Western blotting analysis of mouse tissues for the expression of Jfk and Ing5.
• B, C
  Total proteins or RNAs were prepared in MEFs or liver tissues from wild‐type or Jfk ^KO mice for Western blotting analysis (B) or qPCR measurement (C), respectively. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
• D
  The measurement of AMPK activity and the rate of fatty acid β‐oxidation of wild‐type or Jfk ^KO mice. Cellular extracts prepared from MEFs or liver tissues were subjected to AMPK activity assays by ELISA. Mitochondria were isolated from MEFs or liver tissues, and the rate of fatty acid β‐oxidation was quantified by spectrophotometric detection for ferricyanide‐trapped reduced products. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).
• E, F
  Wild‐type or Jfk ^KO mice were fed with HFD for 11 weeks prior to injection of adenovirally delivered vector control, shIng5, or shAmpkα1 via tail vein. Seven days after injection, cellular extracts prepared from liver tissues were subjected to Western blotting analysis or measurement for hepatic TG and NEFA content (E). Asterisk (*) represents significant comparison to wild‐type mice, octothorpe (#) represents significant comparison to Jfk ^KO mice. Error bars represent mean ± SEM (n = 6, *P < 0.05, ^# P < 0.05, one‐way ANOVA with Tukey's HSD test). H&E and Oil Red O staining of liver sections are shown (F). Scale bar, 100 µm.
• G
  Analysis of public dataset (GSE135251) for JFK and AMPKα1 mRNA expression in liver tissues from healthy controls (n = 10) and NAFLD patients (n = 206). Whiskers indicate upper or lower quartile, and lines inside the box represent median (*P < 0.05, welch two‐sample t‐test).
• H
  The proposed model of the SCF^JFK‐ING5 axis in the development of obesity and metabolic syndrome.

Figure EV3. JFK suppresses the expression of fatty acid β‐oxidation‐related genes in livers.

• A, B
  Total proteins or RNAs were prepared in liver tissues from wild‐type, JFK ^TG, Jfk ^KO, or Jfk ^KO mice with liver‐specific reconstitution of JFK fed on HFD for Western blotting analysis (A) or qPCR measurement (B), respectively. Error bars represent mean ± SEM (n = 6, *P < 0.05, paired two‐tailed Student's t‐test).

To substantiate the inference that SCF^JFK can destabilize ING5 to inhibit AMPK activity and fatty acid β‐oxidation (thus promoting both lipid accumulation and development of NAFLD‐like symptoms) in hepatocytes, we injected (tail vein) adeno‐associated AAV8‐shng5, shIng4, shAmpkα1, or vector control into Jfk ^KO mice fed with HFD. ING4, a paralog of ING5, was reported to be degraded by SCF^JFK to suppress breast carcinogenesis (Yan et al, 2015). Liver‐specific depletion of either Ing5 or Ampkα1, but not Ing4, partially but significantly rescued the decrease in hepatic TG and NEFA in Jfk ^KO mice (Figs 7E and EV4A). Analysis by H&E and Oil Red O staining revealed that Jfk ^KO mice with liver‐specific depletion of either Ing5 or Ampkα1, but not Ing4, developed NAFLD‐like phenotypes in livers under HFD (Figs 7F and EV4B), further supporting a role for the SCF^JFK‐ING5 axis in regulating hepatic lipid metabolism.

Figure EV4. Depletion of Ing5 rescues the NAFLD‐resistant phenotypes in Jfk ^KO mice.

• A, B
  Wild‐type or Jfk ^KO mice were fed with HFD for 11 weeks prior to injection of adenovirally delivered vector control, shIng5, or shIng4 via tail vein. Seven days after injection, cellular extracts prepared from liver tissues were subjected to Western blotting analysis or measurement for hepatic TG and NEFA content (A). Asterisk (*) represents significant comparison to wild‐type mice, and octothorpe (#) represents significant comparison to Jfk ^KO mice (*P < 0.05, ^# P < 0.05, one‐way ANOVA with Tukey's HSD test). Error bars represent mean ± SEM (n = 6). H&E and Oil Red O staining of liver sections are shown (B). Scale bar, 100 μm.

To extend our observations to a clinicopathologically relevant context, we analyzed the expression level of JFK in human NAFLD datasets from http://www.ncbi.nlm.nih.gov/geo (GSE135251) (Govaere et al, 2020). Strikingly, we found that JFK is upregulated whereas AMPKα1 is down‐regulated in liver tissues from NAFLD patients as compared to healthy controls (Fig 7G). These results from analysis of human NAFLD patients are in accord with our working model wherein SCF^JFK destabilizes ING5 to inhibit AMPK activity and thus fatty acid β‐oxidation (Fig 7H), supporting that JFK targeting of ING5 for degradation promotes lipid accumulation.

Discussion

Metabolic syndrome defines a series of metabolic disorders including obesity, hyperlipidemia, hyperglycemia, and insulin resistance and represents a complex, emerging epidemic in which the metabolic homeostasis of several organs, including liver, heart, pancreas, and adipose tissue, was inflicted (Grundy, 2004; Menikdiwela et al, 2020). Despite the progress in understanding the complicated interplay between adipose tissue malfunction, insulin action, the metabolic syndrome traits, and co‐morbidities, the etiology and molecular mechanism underlying the pathogenesis of obesity and metabolic syndrome are still under intensive investigations, although it is a safe assumption that in a general sense, the interaction between environmental factors and genetic susceptibility holds the key. In the current study, using transgenic and knockout mouse models, we report that JFK, an F‐box protein, promotes spontaneous obesity and the development of hepatic steatosis, phenotypes that are further exacerbated under high‐fat diets. We showed that liver‐specific reconstitution of JFK in Jfk knockout mice accelerates hepatic lipid accumulation under high‐fat diets resembling human NAFLD. We also found that JFK is overexpressed in NAFLD patients, further supporting the role of JFK in the development of obesity and metabolic syndrome.

Targeted proteolysis of transcription regulators including chromatin modifiers via the ubiquitin‐proteasome system represents an important regulatory mechanism governing myriad cellular activities (Hershko & Ciechanover, 1998). We previously reported that SCF^JFK promotes breast carcinogenesis through targeting p53 and ING4 for degradation (Skaar et al, 2014). We report in the current study that JFK interacts with ING5 and mediates ING5 for ubiquitination and degradation through an SCF‐dependent mechanism, and SCF^JFK‐ING5 axis intersects with the AMPK activity and fatty acid β‐oxidation pathway. However, how JFK specifies its cellular target is currently unknown, and whether and how p53, ING4, and ING5 are functionally connected/coordinated in the context of SCF^JFK need further investigations. Nevertheless, multiple cellular targets have also been reported for other SCF‐type E3 ligases and it is logical that a particular E3 ligase targets different substrates in response to different cellular environments to influence distinct cellular activities. We do know now that dysregulated bioenergetic processes are intimately associated with carcinogenesis (Teoh & Lunt, 2018). It will be interesting to investigate whether JFK‐promoted breast carcinogenesis is in anyway linked to lipid homeostasis.

AMPK is believed to be a “fuel gauge” to monitor the systemic and cellular energy status and a coordinator to partition fatty acids between biosynthetic and oxidative processes by inhibiting de novo lipogenesis and encouraging fatty acid oxidation, respectively (Viollet et al, 2006). Our finding showed that SCF^JFK inhibits AMPKα1 expression and influence AMPK activity, and Ampkα1 depletion with the tail vein injection partially but significantly restored the decrease in hepatic TG and NEFA in Jfk ^KO mice. Consistent with this, we found that several fatty acid β‐oxidation‐related genes, including ACOX1, HADH, and ACAD11, are directly regulated by the SCF^JFK‐ING5 axis. Interestingly, a recent study reported that hepatic AMPK activation can alleviate obesity‐associated NAFLD through accelerating fatty acid oxidation and by inhibiting de novo lipogenesis (Garcia et al, 2019). Given that AMPK functions in metabolic homeostasis in different organs (Daval et al, 2006; Fu et al, 2013; Rabinovitch et al, 2017; Thomson, 2018), our study does not exclude the possibility that SCF^JFK could also impact de novo lipogenesis in adipose tissue or myogenesis in myoblasts through inhibiting AMPK activity. Indeed, it should be noted that JFK promotes spontaneous obesity and is associated with decreased adiponectin and lean mass in JFK ^TG mice. Although neither JFK ^TG nor Jfk ^KO mice exhibited altered food intake or locomotor activity, our observation that Jfk ^KO mice exhibited increased energy expenditure, BAT thermogenesis and WAT browning can partly explain the phenotype of obesity resistance in Jfk ^KO mice. In addition, most of calorimetry experiments were performed at room temperature (18–22°C), at which the mice are under chronic thermal stress and must increase their metabolism and food intake by about 50% (Lodhi & Semenkovich, 2009), so calorimetry experiments in thermoneutrality (30°C for mice) will be necessary to minimize energy expenditure required to maintain body temperature and to verify the energy expenditure in Jfk ^KO mice at steady state. It also will be interesting to pursue whether the SCF^JFK‐ING5‐AMPK axis is engaged on WAT browning and/or adipocyte differentiation in a similar pattern of metabolic regulation.

ING5 has been investigated as an inhibitor of growth and type‐II tumor suppressor. At the molecular level, ING5 acts to read H3K4me3 (Chi et al, 2010) and recruit HBO1 and MOZ/MORF complexes, leading to acetylation of H3/H4 and transcription activation of target genes (Ullah et al, 2008; Lalonde et al, 2013). Biologically, ING5 has been implicated in a diverse biological processes including DNA replication/repair, cell cycle, apoptosis, and angiogenesis (Aguissa‐Toure et al, 2011; Tallen & Riabowol, 2014). We propose that ING5 also acts as an important regulator of energy homeostasis via its transcriptionally regulatory function and through its influence on the AMPK activity and fatty acid β‐oxidation. Given that acetyl‐CoA is the product of fatty acid β‐oxidation and the substrate of histone acetylation (Pietrocola et al, 2015), it is reasonable to postulate that ING5, per its role as an epigenetic reader and through a long‐term transcriptional control of its target genes involving the AMPK pathway and fatty acid β‐oxidation, contributes to the reprogramming of lipid catabolism and to the sensing of the fluctuation of the cellular acetyl‐CoA level. It is even intriguing to speculate the existence of a positive feedback regulatory loop between ING5 and fatty acid β‐oxidation, in which ING5 reads H3K4me3 to encourage the consumption of acetyl‐CoA through acetylating H3/H4, rendering transcription activation of AMPKα1, ACOX1, HADH, and ACAD11, which, in turn, facilitate the production of acetyl‐CoA. If this feedback loop does exist and function in cells, the targeted proteolysis of ING5 by SCF^JFK thus braking this loop becomes more significant in that it not only adjusts lipid metabolism, but also connects lipid metabolism to transcriptional control.

It remains to be investigated the cellular environment that fosters and the upstream event and factor(s) that stimulates the activation of the SCF^JFK‐ING5 pathway. It will be interesting to delineate the scope and the variety of the cellular activities influenced by the SCF^JFK‐ING5 pathway, especially considering ING5's reading of H3K4me3, an epigenetic modification that marks all active promoters. Nevertheless, we describe in the current study that JFK promotes spontaneous obesity and contributes to the development of metabolic syndrome and NAFLD. We show that SCF^JFK is a bona fide E3 ligase for ING5. We demonstrate that SCF^JFK‐ING5 axis intersects with the AMPK activity and fatty acid β‐oxidation to regulate lipid accumulation. Our study adds to the complexity of the hierarchical regulatory network of energy homeostasis, supporting the pursuit of JFK as a potential target for therapeutic intervention of metabolic disorders.

Materials and Methods

Plasmids and reagents

The cDNAs for wild‐type or truncated JFK were amplified by PCR and ligated into pcDNA3.1 vector that contains three copies of FLAG or pGEX‐4T‐3 vector. All clones were confirmed by DNA sequencing. pcDNA3‐Myc3‐FBW7, pcDNA3‐Myc3‐SKP2, pcDNA3‐Myc3‐β‐TRCP, pcDNA3‐RBX1, and pcDNA3‐SKP1 were from Dr. Yue Xiong (University of North Carolina, USA). pcDNA3.1‐Myc‐ING5 was from Dr. Jacques Cote (Laval University, Canada). The antibodies used were αFLAG, αTubulin, αβ‐actin from Sigma; αING5 from Abcam; αHA, αMyc, αGAPDH, and αmulti‐ubiquitin from MBL; αSKP1 and αRBX1 from Zymed Laboratories; αING5, αHADH, αAMPKα1 from ProteinTech; αphospho‐AMPKα (Thr172), αphospho‐ACC, αACC from Cell Signaling Technology; αACOX1, αACAD11 from Santa Cruz. Polyclonal antibodies against JFK were raised against the C‐terminal epitope of the JFK protein (CYPKTNALYFVRAKR) in rabbits.

JFK transgenic mice

JFK transgenic mice were generated by Cyagen Biosciences on the C57BL/6 background. ORF of human JFK with a FLAG tag was cloned into the mammalian expression vector pRP (Exp)‐EF1A. After digestion, linearized DNA was used for microinjection into the pronuclei of fertilized oocytes from hormonally superovulated C57BL/6 female mice under a microscope. The injected fertilized eggs were transplanted into the oviducts of pseudo‐pregnant mice. The genomic DNA was extracted from mouse tail tips for molecular genotyping. The primers specific for the transgene (forward: 5′‐GCTTTTGGAGTACGTCGTCT‐3′; reverse: 5′‐GGCTCCTCATCTTGATCCAT‐3′) were used to amplify a 334 bp fragment, and the primers for internal control (forward: 5′‐CAACCACTTACAAGAGACCCGTA‐3′; reverse: 5′‐GAGCCCTTAGAAATAACGTTCACC‐3′) were used to amplify a 632 bp fragment.

Generation and maintenance of Jfk‐deficient mice

The targeting construct containing exon 5 of Jfk flanked by LoxP sites was microinjected into mESCs, and the targeted mESCs carrying the Jfk ^neo‐flox allele were transplanted into blastocysts for generation of chimeras. The constitutive knockout allele was obtained by crossing the Jfk ^flox/flox mice with EIIa‐cre mice for Cre‐mediated recombination and LoxP‐mediated removal of the neomycin selection marker, and then were intercrossed to generate homozygous Jfk knockout mice (van Dijk et al, 2011; Li et al, 2017). Genotyping analyses were performed by PCR with genomic DNA extracted from tail tips, and the primers used were 5′‐AGTGCTCTTAACTGCCTAGTATTTTT‐3′ (forward) and 5′‐GGGCGTGTCTTAATGTCCTAAAACCA3′ (reverse). Amplification of the wild‐type and the loxP‐containing deletion region resulted in PCR products of 1,476 and 650 bp, respectively. Animal handling and procedures were approved by the Institutional Animal Care and Use Committee at Peking University.

Liver‐specific reconstitution of JFK in mice

5‐week‐old male wild‐type or Jfk ^KO mice were fed with HFD for 12 weeks. In the last week, 1 × 10¹¹ plaque‐forming unit of adeno‐associated virus 8 (AAV8)‐Vector or AAV8‐JFK was injected into wild‐type or Jfk ^KO mice via tail vein (van Dijk et al, 2011; Chen et al, 2018). Seven days after virus injection, animals were used for subsequent experiments.

Body weight study and body composition analysis

5‐week‐old male JFK ^TG, Jfk ^KO, or wild‐type littermates were randomly assigned to ND (10% fat, MD12031; Medicience Ltd, China) or HFD (45% fat, MD12032; Medicience Ltd, China), and were weighed biweekly. Body composition of mice was measured by time‐domain nuclear magnetic resonance (TD‐NMR) with Bruker's minispec LF50 body composition analyzer (Bruker, Germany).

Glucose tolerance test and insulin tolerance test

For GTT, all mice were fasted for 12 h and injected intraperitoneally with 2 g/kg glucose. For ITT, all mice were fasted for 6 h and injected intraperitoneally with 0.75 U/kg insulin. Glucose level was measured from tail bleeds with a glucose meter at 0, 15, 30, 60, 90, and 120 min after glucose or insulin injection.

Histological analysis and Oil Red O staining

Tissues were fixed in 10% neutral‐buffered formalin at 4°C overnight. After paraffin embedding and sectioning (5 µm), tissues were stained with H&E (haemotoxylin and eosin). For Oil Red O staining of lipid droplets, the liver cryosections of 5 μm were fixed with 10% formalin and then rinsed with PBS buffer. After air dry, the slides were placed in 100% propylene and stained with 0.5% Oil Red O solution in propylene glycol for 10 min. The slides were transferred to 85% propylene glycol solution for 1 min, rinsed in distilled water for two changes, and processed for hematoxylin counter staining.

Indirect calorimetry

The metabolic cage was continuously connected to an open‐circuit, and indirect calorimetry system was controlled by a computer running data acquisition and analysis program under a constant environmental temperature (22°C). The animals were individually placed in metabolic chambers and acclimated for 24 h before data collection. Subsequently, the mice were fed ad libitum or challenged with fasting at 10:00 AM and were monitored for 24 h. Locomotor activity and animal location were monitored with infrared sensor pairs arranged in strips for horizontal and vertical activity, detecting every ambulatory movement. Energy expenditure (EE) was calculated using the Comprehensive Laboratory Animal Monitoring System. The RER (an approximation of respiratory quotient) is calculated by the ratio VCO₂/VO₂ at a 1‐h interval and averaged during fasting. The values of RER provide an approximation of carbohydrate and lipid oxidation to generate energy, ranging from 1.0 to approximately 0.7, respectively.

Cell culture and transfection

Cells were maintained according to the ATCC's recommendation. Transfections of expression plasmids were carried out using polyethylenimine (PEI, Polysciences). siRNA oligonucleotides (GeneChem Inc.) were transfected into cells using RNAiMAX (Invitrogen) according to the manufacturer's instructions. The sequences of siRNAs were as follows: JFK siRNA, 5′‐GGUGUAGCCCAUCAGUGUU‐3′; CDC20 siRNA, 5′‐GGGAAUAUAUAUCCUCUGU‐3′; CUL1 siRNA, 5′‐GUUCAUAGCAGCCAGCCU‐3′; SKP1 siRNA, 5′‐GCAAGUCAAUUGUAUAGCAG‐3′; ING5 siRNA, 5′‐GCGCUUUGAAGCAGAUCUG‐3′; AMPKα1 siRNA, 5′‐GGAUCCAUCAUAUAGUUCA‐3′; and control siRNA, 5′‐UUCUCCGAACGUGUCACGU‐3′.

Immunopurification, silver staining, and mass spectrometry

HepG2 cells stably transfected with FLAG‐JFK were treated with 5 μM MG132 (Sigma) for 12 h before harvest. Cellular lysates were prepared by incubating the cells in cold lysis buffer (50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 0.3% NP‐40, and 2 mM EDTA) containing protease inhibitor cocktail (Roche) and phosphatase inhibitor (Applygen). Anti‐FLAG immunoaffinity resin (Sigma) was prepared according to the manufacturer's protocol. Cell lysates were applied to the immunoaffinity resin to enable adsorption of the protein complex. After binding, the resin was washed with lysis buffer six times. FLAG peptide (Sigma) was added to the resin to elute the FLAG‐JFK‐associated protein complex. The eluates were collected and resolved on NuPAGE 4–12% Bis‐Tris gel (Invitrogen), silver‐stained (Pierce), and subjected to LC‐MS/MS sequencing.

Co‐immunoprecipitation and Western Blotting

HepG2 cellular lysates were prepared by incubating the cells in lysis buffer (50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 0.5% NP‐40, and 2 mM EDTA) containing protease inhibitor cocktail and phosphatase inhibitor for 20 min at 4°C, followed by centrifugation at 14,000 g for 15 min at 4°C. The protein concentration of the lysates was determined using the BCA protein assay kit (Pierce) according to the manufacturer's protocol. For immunoprecipitation, 1 mg of protein was incubated with 2 μg antibodies or normal IgG overnight at 4°C with constant rotation, followed by incubation with 60 μl of 50% protein A or G agarose beads for additional 2 h. Beads were washed six times using the lysis buffer and boiled in 2 × SDS–PAGE loading buffer for 10 min. For Western blotting, the samples were resolved on SDS–PAGE gels and transferred onto nitrocellulose membranes. The membranes were incubated with appropriate primary antibodies overnight at 4°C followed by incubation with a secondary antibody (1:5,000) at room temperature for 1 h. Immunoreactive bands were visualized using Western blotting Luminol reagent (Santa Cruz) according to the manufacturer's recommendation. The dilutions of primary antibodies were as follows: 1:500 with αACOX1 or αACAD11; 1:1,000 with αphospho‐AMPKα (Thr172), αAMPKα1, αphospho‐ACC, αACC, αHA, αHADH, αMyc, αmulti‐ubiquitin, αRBX1, or αSKP1; 1:2,000 with αβ‐actin, αGAPDH, αING5, or αTubulin; 1:8,000 with αFLAG.

GST Pull‐down assay

GST fusion constructs were expressed in BL21 E. coli bacteria, and crude bacterial lysates were prepared by sonication in lysis buffer (50 mM Tris–HCl, pH 7.4, 1.5 mM EDTA, 1 mM dithiothreitol, 10% (v/v) glycerol, 0.4 M NaCl, and 1 mM PMSF). GST fusion protein was immobilized on 50% glutathione‐Sepharose 4B slurry beads (Amersham Biosciences). After incubation for 2 h with rotation at 4°C, beads were washed three times with GST pull‐down binding buffer (10 mM HEPES, pH 7.6, 3 mM MgCl₂, 100 mM KCl, 5 mM EDTA, 5% glycerol, and 0.5% CA630). Meanwhile, in vitro transcription and translation experiments were done with rabbit reticulocyte lysate (TNT systems, Promega) according to the manufacturer's recommendation. 40 μl of transcribed/translated protein was incubated with 2 μg GST fusion protein/beads in 500 μl GST pull‐down binding buffer for 2 h with rotation at 4°C. The beads were then washed six times with binding buffer. The bound proteins were eluted by boiling in 30 μl of 2 × sample loading buffer and resolved on SDS–PAGE gel.

Ubiquitination assay

For in vivo ubiquitination assays, HepG2 cells were transfected for 36 h and then treated with 5 μM MG132 (Sigma) for 12 h before harvest. The cells were lysed in 2% SDS buffer (10 mM Tris–HCl, pH 8.0, and 150 mM NaCl) containing protease inhibitor cocktail and boiled for 10 min followed by sonication for 2 min. Lysates were diluted 1:10 in dilution buffer (10 mM Tris–HCl, pH 8.0, 150 mM NaCl, 2 mM EDTA, and 1% Triton X‐100), incubated for 60 min with rotation at 4°C, and centrifuged at 12,000 g for 15 min. 2 mg of cellular extracts was incubated with anti‐ING5 antibodies overnight at 4°C, followed by incubation with 60 μl of 50% protein A agarose beads for additional 2 h. The beads were washed five times with washing buffer (10 mM Tris–HCl, pH 8.0, 1 M NaCl, 1 mM EDTA, and 1% NP‐40), boiled in SDS sample buffer, and subjected to Western blotting analysis.

The ubiquitin conjugation kit (Boston Biochem) was utilized for in vitro ubiquitination assay. Briefly, 2 μg of GST fusion proteins (GST‐JFK, GST‐JFKΔKelch, or GST‐JFKΔF‐box) was immobilized on 50 μl of 50% glutathione‐Sepharose 4B slurry beads (Amersham Biosciences) in 0.5 ml of GST pull‐down binding buffer (10 mM HEPES, pH 7.6, 3 mM MgCl₂, 100 mM KCl, 5 mM EDTA, 5% glycerol, and 0.5% CA630) and was incubated with 40 μl of in vitro transcribed/translated ING5 for 2 h with rotation at 4°C. The beads were washed three times with binding buffer and then were added to 8 μg of fraction A (E1 and E2 enzymes), 8 μg of fraction B (E3 and deubiquitinating enzymes), 26 μg of ubiquitin, 4 μM ubiquitin aldehyde, and 2.5 μl of 10 × energy solution in a 25 μl volume. After incubation at 37°C for 30 min, the beads were washed six times with binding buffer, boiled in 2 × sample loading buffer, and subjected to SDS–PAGE followed by immunoblotting.

RNA sequencing and real‐time reverse transcription PCR

Total mRNAs were isolated with TRIzol reagents (Invitrogen) for cDNA synthesis, library construction, and sequencing using HiSeq 2500 (Beijing Genomics Institute). The quality of raw data was examined by FastQC, and sequencing adapter and low‐quality reads, i.e., those with more than five “N” bases and mean Phred quality score < 15, were removed through fastp. Spliced Transcripts Alignment to a Reference (STAR) software was then utilized to align clean reads to the unmasked human reference genome (GRCh37, hg19). Raw counts of the reads mapped to genes were selected to analyze differentially expressed genes using DESeq2 Bioconductor package with FDR < 0.001 as the threshold.

Real‐time reverse transcriptase PCR was performed using the ABI PRISM 7500 system (Applied Biosystems) that measures real‐time SYBR green fluorescence and then calculated by means of the comparative Ct method ($2^{‐\mathrm{\Delta }\mathrm{\Delta }{\mathrm{C}}_{\mathrm{t}}}$) with the expression of GAPDH as an internal control. The primers used were as follows: ACOX1: 5′‐GGAACTCACCTTCGAGGCTTG‐3′ (forward) and 5′‐TTCCCCTTAGTGATGAGCTGG‐3′ (reverse); HADH: 5′‐ACCCTGAGCACCATAGCGA‐3′ (forward) and 5′‐CAGCGAATCGGTCTTGTCTGG‐3′ (reverse); ACAD11: 5′‐TGCTACTGGCGAGTCCGAT‐3′ (forward) and 5′‐TGTGCTTTAGGAAGAAGTGAACC‐3′ (reverse); AMPKα1: 5′‐TTGAAACCTGAAAATGTCCTGCT‐3′ (forward), and 5′‐GGTGAGCCACAACTTGTTCTT‐3′ (reverse); GAPDH: 5′‐GAAGGTGAAGGTCGGAGTC‐3′ (forward) and 5′‐GAAGATGGTGATGGGATTTC‐3′ (reverse); Ing5: 5′‐AAGATCCAGAGCGCCTACAG‐3′ (forward) and 5′‐GCATCAAGTCTTCGGATGTGTT‐3′ (reverse); Acox1: 5′‐TAACTTCCTCACTCGAAGCCA‐3′ (forward) and 5′‐AGTTCCATGACCCATCTCTGTC‐3′ (reverse); Hadh: 5′‐TCAAGCATGTGACCGTCATCG‐3′ (forward) and 5′‐TGGATTTTGCCAGGATGTCTTC‐3′ (reverse); Acad11: 5′‐AGATGCTTCAGTTATCGGAACG‐3′ (forward) and 5′‐ATGTAGCCATGCCAGGGTTTC‐3′ (reverse); Ampkα1: 5′‐GTCAAAGCCGACCCAATGATA‐3′ (forward) and 5′‐CGTACACGCAAATAATAGGGGTT‐3′ (reverse); Gapdh: 5′‐AGGTCGGTGTGAACGGATTTG‐3′ (forward) and 5′‐TGTAGACCATGTAGTTGAGGTCA‐3′ (reverse).

ChIP‐seq and qChIP

HepG2 cells were cross‐linked with 1% formaldehyde for 10 min at room temperature and quenched by the addition of glycine to a final concentration of 125 mM for 5 min. The fixed cells were resuspended in lysis buffer (1% SDS, 5 mM EDTA, and 50 mM Tris–HCl, pH 8.1) containing protease inhibitors, then subjected to 30 cycles (30 s on and off) of sonication (Bioruptor, Diagenode) to generate chromatin fragments of ~ 300 bp in length. Lysates were diluted in buffer (1% Triton X‐100, 2 mM EDTA, 150 mM NaCl, and 20 mM Tris–HCl, pH 8.1) containing protease inhibitors. For immunoprecipitation, the diluted chromatin was incubated with normal IgG (control) or ING5 antibodies overnight at 4°C with constant rotation, followed by incubation with 50 μl of 50% (v/v) protein A/G Sepharose beads for an additional 2 h. Beads were successively washed with the following buffers: TSE I (0.1% SDS, 1% Triton X‐100, 2 mM EDTA, 150 mM NaCl, and 20 mM Tris–HCl, pH 8.0); TSE II (0.1% SDS, 1% Triton X‐100, 2 mM EDTA, 500 mM NaCl, and 20 mM Tris–HCl, pH 8.0); TSE III (0.25 M LiCl, 1% Nonidet P‐40, 1% sodium deoxycholate, 1 mM EDTA, and 10 mM Tris–HCl, pH 8.0). The pulled‐down chromatin complex eluted by TE (1 mM EDTA and 10 mM Tris–HCl, pH 8.0) and input were de‐cross‐linked at 55°C for 12 h in elution buffer (1% SDS and 0.1 M NaHCO₃) (Li et al, 2016). The DNA was purified with the QIAquick PCR Purification Kit (QIAGEN). ING5‐associated DNAs were amplified using non‐biased conditions, labeled, and sequenced with HiSeq 2500 (CapitalBio Corporation, Beijing). The raw sequencing image data were examined by the Illumina analysis pipeline, aligned to the unmasked human reference genome (GRCh37, hg19) using Bowtie 2, and further analyzed by MACS (Model‐based Analysis for ChIP‐Seq) with the cutoff of P value < 1e‐5 and FDR < 0.001. Enriched binding peaks were generated after filtering through normal IgG (control). Genomic distribution of ING5 binding sites was analyzed by Deeptools. Biological process ontology analysis was conducted based on the Database for Annotation, Visualization and Integrated Discovery (DAVID, https://david.ncifcrf.gov/).

qChIPs were performed using Power SYBR Green PCR Master Mix and an ABI PRISM 7500 system (Applied Biosystems, Foster City, CA). The primers used were as follows: ACOX1: 5′‐CTCTGCCAGTAAAGACGCCA‐3′ (forward) and 5′‐GCTTTGCGGAATTGCTAGGG‐3′ (reverse); HADH: 5′‐ACGTCCAACGTCAAGATGGG‐3′ (forward) and 5′‐ATGTGCACCGTTGGTTGATG‐3′ (reverse); ACAD11: 5′‐CCAGGGCCAAAGGAAGGAAA‐3′ (forward) and 5′‐CCAGGGCCAAAGGAAGGAAA‐3′ (reverse); AMPKα1: 5′‐GGCTCTCGTAACAAGCCACA‐3′ (forward) and 5′‐TCCTGGAACGATTAGGCGGA‐3′ (reverse).

ELISA for AMPK activity

AMPK activity was measured with AMPK Kinase Assay Kit (CycLex Co. Ltd.) according to the manufacturer's instructions. In brief, cells or tissues were lysed in ice‐cold lysis buffer (20 mM Tris–HCl, pH 7.5, 250 mM NaCl, 10% glycerol, 0.5% NP‐40, 1mM EDTA, 1 mM EGTA, 5 mM NaF, 2 mM Na₃VO₄, 2 mM β‐glycerophosphate, 1 mM DTT, 0.2 mM PSMF, 1 µg/ml pepstatin, and 0.5 µg/ml leupeptin). Then samples were serially diluted in kinase buffer and incubated in a pre‐coated plate with a substrate peptide at 30°C for 30 min. The phosphorylated substrate by AMPK was recognized by an anti‐phospho‐serine specific monoclonal antibody, which bound with horseradish peroxidase‐conjugated anti‐mouse IgG as a reporter molecular. The AMPK activity was quantified using spectrophotometry to measure absorbance at 450 nm.

Measurement of fatty acid β‐oxidation rate

Mitochondria were isolated from HepG2 cells, liver tissues, or MEFs using the Mitochondrial Extraction Kit (Solarbio) according to the manufacturer's recommendation. The fatty acid β‐oxidation rate was determined by recording the reduction in ferricyanide with time using Fatty Acid β‐Oxidation Assay Kit (Genmed Scientifics Inc.), in which ferricyanide is able to trap reducing equivalents generated by the acyl‐CoA dehydrogenases of β‐oxidation. Mitochondrial solution or negative reagent was added to the mixture for sample detection or blank control, followed by immediate absorbance measurement at 420 and 470 nm using dual‐wavelength spectrophotometry.

Ketone body measurement

3‐hydroxybutyric acid (OHB) and acetoacetic acid (AcAc) were measured using the Ketone Body Assay Kit (Sigma‐Aldrich) as the manufacturer's instructions. The assay was based on 3‐hydroxybutyrate dehydrogenase catalyzed reactions, in which the change in NADH absorbance measured at 340 nm is directly related to the concentration of OHB and AcAc. In brief, the culture media collected from HepG2 cells was incubated with the reaction buffer at room temperature for 5–15 min, followed by spectrophotometric detection. Total ketone body concentration is calculated as the concentration of OHB plus AcAc.

Statistical analysis

Results are reported as mean ± SD for triplicate in vitro experiments or mean ± SEM in vivo experiments unless otherwise noted. SPSS V.19.0 was used for statistical analysis. Comparisons between two groups for in vivo experiments and comparisons for in vitro experiments were performed using paired two‐tailed Student's t‐test. Comparisons among three or more groups for in vivo experiments were performed using one‐way ANOVA followed by Tukey's honestly significant difference (HSD) post hoc test. Statistical significance of differences is indicated as *P < 0.05.

Author contributions

LH, RY, and LS designed research; LH and RY conducted experiments; LH, RY, ZY, and YZ performed animal experiments and analyzed data; XinL, JY, XuL, XiL, LX, YW, JW, XW, LS, XY, JL, and YS provided technical assistance; LH, RY, YS, and LS wrote the paper.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Review Process File

EMBO reports (2021) 22: e52036.

Data availability

RNA‐seq and ChIP‐seq data have been deposited in the Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo) with accession code numbers GSE166144 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE166144) and GSE160192 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE160192). All other remaining data are available within the article or available from the authors upon request.