The novel intracellular protein CREG inhibits hepatic steatosis, obesity, and insulin resistance

ANIMALS AND TREATMENTS

Male C57BL/6J mice were reared in a standard environment with a strictly controlled room temperature (22 °C ± 2 °C) and a 12:12-hour light:dark cycle. Ten-week-old mice were fed a high-fat diet (HFD) or on a standard normal chow (NC) diet for the indicated durations. The HFD was a special mouse diet containing 20% carbohydrate, 20% protein, and 60% fat, whereas the NC diet contained 70% carbohydrate, 20% protein, and 10% fat. All of the feed was purchased from Research Diets, Inc. The animals were provided humane care based on the Guide for the Care and Use of Laboratory Animals (National Institutes of Health Publication 86-23, revised 1985). The animal experiments were approved by the Animal Care and Use Committee of the Wuhan University.

Conditional CREG knockout in the liver was performed using the CRISPR/Cas9 system.16 Up- and downstream guide RNAs (gRNA1 and gRNA2) targeting the CREG sequences were designed first. Exon 2 flanked by 2 loxP sites and 2 homology arms in a circular donor vector were designed to repair the DSB by homologous recombination. Next, pups were generated after one-cell-stage embryo injection. After floxed exon 2 amplification, screening was performed to identify mice with floxed exon2 on the same allele. Genomic DNA was used for in vitro Cre-mediated recombination. Finally, CREG conditional knockout (CKO) mice were obtained by mating flox/flox mice with Cre mice.

To generate liver-specific CREG transgenic (CREG-TG) mice,17 a construct containing full-length CREG cDNA was inserted after the CAG-loxp-CAT-loxp cassette, and this was followed by microinjection into fertilized C57BL/6J embryos. The resulting mice were crossed with Albumin-Cre mice to obtain hepatic CREG-TG mice.

HUMAN LIVER SAMPLES

This study enrolled 28 liver donors to examine CREG and JNK1 spatial and overall expression levels. The normal group included normal control livers (n = 11). The inclusion criterion for the normal group was patients who had received liver resection due to liver hemangiomas or hepatic cysts. Exclusion criteria included patients who required intensive care treatment or had lethal trauma during liver biopsy. The NAFLD group comprised NAFLD patients (n = 17). The inclusion criteria of the NAFLD group were patients who had been diagnosed with NAFLD, and the exclusion criteria included those who drank 140 or 70 g/week of alcohol (for men or women, respectively) at the time of the study or within the previous 6 months. All of the human sample experiments were conducted in accordance with the Declaration of Helsinki and were approved by the review board of the Wuhan University.

WESTERN BLOTTING ASSAYS AND IMMUNOPRECIPITATION

Cells or tissues were placed in 100 μL of ice-cold RIPA buffer (20 mM Tris [pH 7.6], 0.5% NP-40, 250 mM NaCl, 3 mM EDTA, 3 mM EGTA, 2 mM dithiothreitol, 0.5 mM PMSF, 20 mM b-glycerol phosphate, 1 mM sodium vanadate, and 1 mg/mL leupeptin) and were incubated at 4 °C for 30 minutes. Next, lysates in buffer were centrifuged at 12,000 rpm for 15 minutes at 4 °C, and polyacrylamide gel electrophoresis with 7.5%-15% sodium dodecyl sulfate was performed. The proteins were transferred to PVDF membranes and incubated with the relevant primary antibody. Immunostaining with antibodies was performed with a BIO-RAD system. For immunoprecipitation, the cells or tissues were collected in RIPA buffer. The resulting lysates were precipitated with the relevant antibody and protein G-sepharose beads by incubation at 4 °C overnight. The immune complex kinase assay for apoptosis signal-regulating kinase 1 (ASK1) was performed as described previously.18

The following antibodies were used. CREG (ab68341, 1:1000 dilution) was obtained from ABCAM (Cambridge, MA, USA). PEPCK (6924, 1:1000 dilution), AKT (4691, 1:1000 dilution), p-GSK3β (9322, 1:1000 dilution), GSK3β (9315, 1:1000 dilution), IRS1 (2832, 1:1000 dilution), p-JNK1/2 (4668, 1:1000 dilution), T-JNK1/2 (9252, 1:1000 dilution), FOXO1 (2880, 1:1000 dilution), p-AKT^ser473 (9271, 1:1000 dilution), p-FOXO1 (9461, 1:1000 dilution), p-AMPKα (5256, 1:1000 dilution), AMPK (5831, 1:1000 dilution), p-ACC (11818, 1:1000 dilution), ACC (3676, 1:1000 dilution), T-ERK1/2 (9102, 1:1000 dilution), p-ERK1/2 (4370, 1:1000 dilution), T-MEK1/2 (8727, 1:1000 dilution), p-MEK1/2 (9154, 1:1000 dilution), T-T38 (8690, 1:1000 dilution), P-P38 (4511, 1:1000 dilution), T-JNK1 (3708, 1:1000 dilution), T-JNK2 (9258, 1:1000 dilution), and β-actin (3700, 1:1000 dilution) were obtained from Cell signaling Tec (Shanghai, China). p-IRS1^Tyr608 (09-432, 1:1000 dilution), p-mTOR (ABS79, 1:1000 dilution), mTOR (04-385, 1:1000 dilution), p-p70s6k (MABS82, 1:1000 dilution), and p70s6k (06-926, 1:1000 dilution) were obtained from Merck Millipore (Beijing, China). FOXO1 (TA323072, 1:500 dilution) was obtained from ORIGENE (Beijing, China). G6Pase (ARP44223-P050, 1:1000 dilution) was obtained from Aviva Systems Biology (Beijing, China). p-JNK1 (AP3644a, 1:500 dilution) was obtained from Agbent (San Diego, CA, USA). p-JNK2 (DYC2236E, 1:500 dilution) was obtained from R&D Systems (Shanghai, China). The BCA protein assay kit was obtained from Pierce company.

METABOLIC ASSAYS

Blood glucose levels were detected with a glucometer (OneTouch Ultra Easy). Serum insulin levels were analyzed with an enzyme-linked immunosorbent assay kit from Millipore. Homeostasis model assessment of insulin resistance (HOMA-IR) was calculated according to Wallace et al.19 An intraperitoneal glucose tolerance test (IPGTT) was performed at 1 g/kg glucose (Sigma-Aldrich, St. Louis, MO) intraperitoneal injection, and an intraperitoneal insulin tolerance test (IPITT) was performed at 0.75 U/kg insulin (Novolin R; Novo Nordisk, Bagsvaerd, Denmark). For both IPGTT and IPITT, the blood glucose level was tested 0, 15, 30, 60, and 120 minutes after glucose/insulin injection. Commercial kits were used to measure triglyceride (TG), total cholesterol, and nonesterified fatty acid levels in the liver (290-63701 for TG assay, 294-65801 for total cholesterol assay, and 294-63601 for nonesterified fatty acid assay; Wako, Tokyo, Japan). All protocols were performed according to the manufacturer's instructions.

HISTOLOGICAL ANALYSES

Paraffin liver sections were prepared for hematoxylin and eosin (H&E) staining, whereas frozen liver sections were stained with Oil Red O. Both methods were used to observe the distribution of lipid accumulation and lipid droplets in the liver. In periodic acid–Schiff (PAS) staining, the liver sections were incubated in 0.5% periodic acid solution for 5 minutes at room temperature after deparaffinization and hydration. After incubation with Schiff's reagent for 15 minutes, H&E staining was performed.

BIOCHEMICAL ASSAY

Liver function tests for alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP) were measured on the basis of the ADVIA 2400 chemistry system (Siemens).

HEPATIC LIPID ACCUMULATION ASSAYS

BODIPY-C16 fluorescence staining was performed. Primary hepatic cells were infected with AdCREG or AdshCREG and the corresponding controls. Next, the cells were treated with PA or vehicle control for 24 hours. Primary hepatic cells in 6-well plates were starved for 3 hours and then incubated with 100 nM BODIPY-C16 for 5 minutes, washed with PBS, and fixed with 4% paraformaldehyde (pH 7.4) for 10 minutes. Images were obtained with a fluorescence microscope (Olympus) and were analyzed with DP2-BSW software.

CELL LINES AND PRIMARY HEPATOCYTE ISOLATION AND CULTURE

Primary hepatocytes were isolated according to a previous protocol.20 Next, these primary hepatocytes were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin in a 5% CO₂/water-saturated incubator. To mimic a fatty liver model in vitro, PA (0.25 mM; Sigma-Aldrich) was added to the culture medium for 24 hours. For Oil Red O staining of primary hepatocytes, the cells were fixed in 4% paraformaldehyde for 10 minutes and then stained in 0.3% Oil Red O working solution for 30 minutes. All of the cell culture reagents and other reagents were purchased from Sigma. Recombinant Ads encoding CREG, shCREG (a short hairpin RNA targeting CREG), GFP, or short hairpin RNA (as a nonspecific control) were constructed as described previously.21

REAL-TIME POLYMERASE CHAIN REACTION ASSAYS

Total messenger RNA (mRNA) was extracted from liver tissue or cultured cells with TRIzol Reagent (15596-026; Invitrogen) and was reverse transcribed into cDNA with a PrimeScript 1st Strand cDNA Synthesis Kit (TAKARA, China) according to the manufacturer's protocol. SYBR Green (SYBR Premix Ex Taq [Tli RNaseH Plus], ROX plus) was applied to quantify polymerase chain reaction amplification. The mRNA expression levels of target genes were normalized to those of GAPDH.

STATISTICAL ANALYSES

The data in this study are presented as the mean ± standard deviation. Two-tailed Student t tests were performed to compare two groups, and differences among more than two groups were evaluated using one-way analysis of variance followed by Tukey's post hoc tests. In contrast, in the absence of a normal distribution, comparison among groups were performed using nonparametric Mann-Whitney tests (comparison of two groups) or Kruskal–Wallis (more than two groups) tests as appropriate. The cutoff for statistically significant differences was a P value of 0.05. SPSS version 22.0 (SPSS Inc, Chicago, IL) was used for analysis. The animal cohort sizes were determined on the basis of similar previous studies.21 Randomization and blinding were used whenever possible.

CREG IS DECREASED IN FATTY LIVERS

First, the expression levels of CREG mRNA and protein were measured in fatty livers in a mouse NAFLD model. After being fed an HFD for 12 weeks, the mRNA and protein expression levels of CREG mice were clearly decreased compared with those of the NC-treated control mice (HFD 0.43 ± 0.04 versus NC 1.00 ± 0.21; P < 0.05 versus NC, Fig. 1A; HFD 2.68 ± 1.04 versus NC 5.88 ± 0.92; P < 0.05 versus NC, Fig. 1B). To directly investigate whether CREG levels changed in liver hepatocytes exposed to a dynamic change, primary hepatocytes were exposed to concentration gradients of PA (0.1 mM, 0.2 mM, 0.4 mM; Fig. 1C), and CREG expression was significantly decreased in PA-treated primary hepatocytes in a dose-dependent manner compared with the CREG expression in the control group. Considering the multiple pathogenic signs of NAFLD, a classical feed/fasted model was established, and CREG expression was measured. In addition to changes to PEPCK and G6Pase, lipid metabolism-related proteins, CREG expression levels were significantly decreased after 24 hours of fasting (P < 0.05 versus feed; data shown in Fig. 1D). Furthermore, an examination of CREG protein levels in human samples was performed by DAB staining (Dako REAL EnVision Detection System). The obtained images showed that CREG was mainly distributed in the cytoplasm of hepatocytes and was dramatically reduced in the livers of NAFLD patients compared with normal group subjects (Fig. 1E). Next, western blotting confirmed decreased CREG levels in the NAFLD group (NAFLD 1.00 ± 0.25 versus normal 2.77 ± 0.34, P < 0.05; Fig. 1F). In short, CREG mRNA and protein expression decreased as fat accumulation increased, thus suggesting a potential role of CREG in metabolism disorder.

HEPATIC CREG DELETION AGGRAVATES HFD-INDUCED OBESITY AND INSULIN RESISTANCE.

After confirming the dramatic down-regulation of CREG in fatty livers, we sought to assess the role of CREG in NAFLD and associated complications. Therefore, CREG-CKO mice were generated (Fig. 2A), and HFD treatment was administered. Western blotting was used to validate successful liver-specific CREG-KO (Fig. 2B). CREG-KO in the liver led to significantly higher body weight, liver weight/body weight and liver weight in mice administered an HFD for 12 weeks compared with the wild-type (WT) controls (Fig. 2C,D,G). In the following glucometabolism studies, CREG-CKO mice exhibited higher fasting blood glucose and fasting serum insulin levels (Fig. 2E,F,H,I), thus leading to a higher value in the homeostasis model of assessment for insulin resistance index (HOMA-IR) (Fig. 2J). Moreover, glucose tolerance tests and insulin tolerance tests demonstrated that CREG-CKO mice developed more severe glucose tolerance and insulin sensitivity (Fig. 2K,L). To further study the influence of CREG deletion on insulin resistance, insulin signaling (IRS1/Akt, GSK3β, FOXO1) was analyzed using western blotting. The phosphorylation levels of all of these tested signaling components, for which higher levels indicate higher glucometabolism ability, were markedly lower in the KO-HFD group than in the controls (Fig. 2M). PAS showed that the glycogen content in the liver was impaired by CREG deletion (Fig. 2N). In short, this loss-of-function study suggests that CREG deletion specifically in the liver aggravates glucometabolism disorder.

HEPATIC CREG OVEREXPRESSION REVERSES NAFLD PATHOLOGIES INDUCED BY HFD

After the hepatic CREG deletion study, hepatocyte-specific overexpression of CREG was evaluated to further determine the effects of CREG on hepatic steatosis and metabolic disorders (Fig. 3A). Western blotting confirmed hepatic-specific CREG overexpression (Fig. 3B), whereas nontransgenic (NTG) mice served as controls after being fed an HFD for 12 weeks. Although no significant effects on body weight were observed, CREG overexpression significantly decreased liver weight/body weight ratio (Fig. 3C,D,G). TG reversed the high blood glucose and high serum insulin levels induced by HFD feeding (Fig. 3E,F,H,I). Moreover, insulin resistance was also improved by TG in the HFD condition as assessed by HOMA-IR, glucose tolerance tests, insulin tolerance tests, and insulin signaling analysis (Fig. 3J-M). PAS results showed a higher glycogen level in the TG-HFD group than that in the littermate controls (Fig. 3N). This gain-of-function study suggested that CREG overexpression in the liver improves glucometabolism disorders.

CREG IMPROVES HFD-INDUCED HEPATIC STEATOSIS AND LIPID METABOLISM PERTURBATION

Next, we evaluated the effect of CREG on perturbation of lipid metabolism, a specific NAFLD feature with close connections to obesity, glucometabolic disorder, and insulin resistance. Liver ultrasound demonstrated that HFD administration led to greater lipid accumulation than that observed in the NC group (NC group thickness, 2.83 ± 0.25 mm; n = 3). When the ultrasonic probe was at the same position and detection angle, the liver was thicker in the KO-HFD group than in the WT-HFD group (thickness, 4.97 ± 0.25 mm versus 3.53 ± 0.35 mm; P < 0.05; n = 3) (Fig. 4A). Increased lipid accumulation in the liver demonstrated by H&E and Oil Red O staining accompanied increased nonesterified fatty acid, total cholesterol, and TG levels in the KO-HFD group (Fig. 4B-E). Furthermore, liver function as reflected by ALT, AST, and ALP levels was increased by KO after HFD treatment (Fig. 4F). According to the above glucometabolism studies, CREG overexpression reversed these changes (Fig. 4G-J). After the in vivo studies above, in vitro lipid accumulation in primary hepatic cells transfected with AdCREG to overexpress CREG or transfected with AdshCREG to down-regulate its expression was evaluated using Oil Red O staining (top two lines) and BODIPY-C16 fluorescence staining (bottom two lines) (Fig. 4L). Oil red O staining showed that lipid accumulation was decreased in the Ad-CREG group but was increased in the Ad-shCREG group compared with the corresponding controls. Furthermore, we observed that for BODIPY-C16 fluorescence staining, cells without lipid accumulation in the control group exhibited signet-ring changes with a clear cell outline, whereas PA triggered robust lipid accumulation. With AdCREG transfection, some cells recovered the signet-ring changes, but nearly all of the cells showed robust lipid accumulation with a crude contour and green fluorescence throughout the cytoplasm in the PA AdshCREG group (Fig. 4K).

Next, the lipid metabolism-related AMPK, ACC, mTOR, and p70S6K signaling pathways were analyzed, and AMPK and ACC phosphorylation were positively correlated with CREG protein expression, whereas mTOR and p70S6K were negatively correlated (Fig. 4L). In addition, the mRNA levels of lipid metabolism- and inflammation-related genes were analyzed by using real-time polymerase chain reaction. After chronic HFD administration, CREG-TG mouse livers showed decreased expression of mRNAs associated with fatty acid synthesis (HMGCR, SREBP-1C, FAS, ACCa), fatty acid uptake (CD36, FATP1, FABP1, PPAR-γ), glycogenesis (PEPCK, G6PC), and inflammatory markers (IL-1b, IL-6, TNF-α, MCP-1) and increased levels of mRNAs associated with fatty acid oxidation/anti-inflammation (ABCG1, CYP7A1, PPAR-α, CPT-1a, ACOX-1, UCP2, IL-10, PDK4), whereas CREG-KO mouse livers exhibited expression changes that were opposite from those observed in the CREG-TG mice (Fig. 4M).

Collectively, CREG deletion increased lipid accumulation in liver, whereas CREG overexpression improved this pathology.

CREG INHIBITS INSULIN RESISTANCE AND HEPATIC STEATOSIS IN OB/OB MICE.

To study the effect of CREG on genetic NAFLD, we used genetically obese ob/ob mice. CREG protein expression decreased with increased feeding time in the livers of ob/ob mice (Fig. 5A). Lipid accumulation in the liver demonstrated by H&E and Oil Red O staining also decreased in the ob/ob-AdCREG group (Fig. 5B). However, the body weight and liver weight/body weight ratio did not change with CREG overexpression in ob/ob mice (Fig. 5C,D), although CREG overexpression largely improved hepatic steatosis and metabolic disorder (Fig. 5E-I) and insulin resistance (Fig. 5J,K). Thus, CREG overexpression alleviated hepatic steatosis and metabolic disorder in genetic NAFLD.

JNK SIGNALING INACTIVATION IS REQUIRED FOR THE EFFECT OF CREG ON NAFLD

To explore the underlying CREG mechanism, the mitogen-activated protein kinase (MAPK) family, which plays a critical role in hepatic steatosis and metabolic disorder, was studied. The phosphorylation levels of MEK, ERK, JNK, and P38 were analyzed by western blotting in vivo (Fig. 6A) and in vitro (Fig. 6B). In the tested proteins (Fig. 6A,B), only JNK phosphorylation changed in by CREG manipulation in vivo and in vitro. In the CREG-KO HFD and AdshCREG PA groups, JNK phosphorylation increased, whereas JNK phosphorylation decreased in the CREG TG HFD and AdCREG-mediated PA groups compared with their respective controls. Therefore, we hypothesized a potential role of JNK1/2 in CREG-mediated hepatic steatosis and metabolic disorder, and further experiments were performed to confirm this idea. First, the JNK1/2 inhibitor SP600125 was used to block JNK1/2 phosphorylation (Fig. 6C). We found that SP600125 successfully reversed the adverse effect of CREG-KO on body weight gain, liver weight/body weight, blood glucose, serum insulin, and IPGTT and IPITT response (Fig. 6D-I). The H&E and Oil Red O staining also showed that SP600125 abolished the robust lipid accumulation in the CREG-CKO group (Fig. 6J). Importantly, immunohistochemistry results from human samples showed that the expression was negatively correlated with P-JNK (Fig. 6K). These data indicated that CREG mediates hepatic steatosis and metabolic disorder by way of the JNK signaling pathway.

JNK1, NOT JNK2, MEDIATES CREG-REGULATED HEPATIC STEATOSIS AND METABOLIC DISORDER

Given that two major isoforms of JNK, JNK1 and JNK2, are expressed in hepatocytes22 with both overlapping and diverse functions, we next investigated the precise contribution of these two major isoforms to CREG-mediated hepatic steatosis and metabolic disorders. To clarify the requirement of JNK1 and JNK2 for physiological CREG function after HFD administration, CREG-JNK1-double gene knockout (DKO) and CREG-JNK2-DKO mice were generated (Fig. 7A). H&E and Oil Red O staining showed that lipid accumulation was decreased in the CREG-JNK1-DKO group but not the CREG-JNK2-DKO group (Fig. 7B). To further confirm the reversal of this effect in CREG-JNK1-DKO mice, the body weight gain, liver weight/body weight, blood glucose, and HOMA-IR were analyzed. The adverse effect of CREG deletion was reversed by knocking out JNK1 (Fig. 7C-F). However, JNK2 deletion did not reverse the effect of CREG knock out (Fig. 7G-J). Therefore, our findings clearly demonstrate that the role of CREG in NAFLD pathologies is largely dependent on inactivation of JNK1 but not JNK2.

CREG PHYSICALLY INTERACTS WITH ASK1 IN THE JNK SIGNALING PATHWAY

We next explored the link between CREG and JNK1. ASK1 is an important upstream of JNK1 signaling pathway via activating MKK4/7. In the present session, ASK1 were studied to determine whether CREG regulates JNK1 via the ASK1-MKK4/7-JNK1 cascade. Therefore, the physical interactions of CREG with components in this cascade were examined. Cells were transiently cotransfected or separately transfected with FLAG-tagged MAPK kinase (ASK1, MKK4, MKK7, JNK1), FLAG-tagged CREG, and HA-tagged ASK1. Notably, CREG was strongly bound to ASK1, but not to other tested factors in the JNK signaling cascade (Fig. 8A-C). As expected, CREG inhibited ASK1 phosphorylation in a dose-dependent manner (Fig. 8D). The inhibitory effect of CREG on ASK1 activity was further confirmed in PA-stimulated primary hepatocytes, as evidenced by the dramatically increased ASK1 phosphorylation in the KO group but the decreased phosphorylated ASK1 after CREG overexpression compared with the corresponding controls (Fig. 8E). Collectively, these data demonstrate that CREG alleviates NAFLD pathologies by directly binding to ASK1 and inhibiting ASK1-MKK4/7-JNK1 signaling.