Over-nutrition induces low-grade inflammation that dampens insulin sensitivity, but the underlying molecular mediators are not fully understood. Comparative gene identification-58 (CGI-58) is an intracellular lipolytic activator. In the present study, we show that in mouse visceral fat-derived macrophages or human peripheral blood monocytes, CGI-58 negatively and interleukin (IL)-1β positively correlate with obesity. Saturated non-esterified fatty acid (NEFA) suppresses CGI-58 expression in macrophages and this suppression activates FOXO1 (forkhead box-containing protein O subfamily-1) through inhibition of FOXO1 phosphorylation. Activated FOXO1 binds to an insulin-responsive element in IL-1β promoter region to potentiate IL-1β transcription. Gain- and loss-of-function studies demonstrate that NEFA-induced CGI-58 suppression activates FOXO1 to augment IL-1β transcription by dampening insulin signalling through induction of SOCS3 (suppressor of cytokine signalling 3) expression. CGI-58 deficiency-induced SOCS3 expression is NLRP3 (nucleotide-binding oligomerization domain-like receptor family, pyrin domain containing 3) inflammasome-dependent. Our data thus identified a vicious cycle (IL-1β–SOCS3–FOXO1–IL-1β) that amplifies IL-1β secretion and is initiated by CGI-58 deficiency-induced activation of the NLRP3 inflammasome in macrophages. We further show that blocking this cycle with a FOXO1 inhibitor, an antioxidant that inhibits FOXO1 or IL-1 receptor antagonist alleviates chronic inflammation and insulin resistance in high-fat diet (HFD)-fed mice. Collectively, our data suggest that obesity-associated factors such as NEFA and lipopolysaccharide (LPS) probably adopt this vicious cycle to promote inflammation and insulin resistance.
We have previously shown that macrophage-specific deletion of CGI-58 in mice aggravates HFD-induced chronic inflammation and insulin resistance by activating macrophage NLRP3 inflammasome to enhance IL-1β secretion . However, CGI-58 deficiency in macrophages also augments IL-1β mRNA expression and the underlying molecular mechanism was unclear.
In the present paper, we demonstrated that macrophage CGI-58 deficiency triggers a positive feedback loop (IL-1β–SOCS3–FOXO1–IL-1β) that augments NEFA-induced chronic inflammation and insulin resistance. Furthermore, we verified this pathway in the obese mouse model and showed correlation of this pathway with the relevant pathophysiological changes in human subjects. Blocking this loop with a FOXO1-specific inhibitor, an antioxidant NAC or IL-1 receptor antagonist reduces inflammation and improves insulin sensitivity in HFD-fed mice.
Understanding the precise mechanism linking CGI-58 deficiency to IL-1β transcription holds promise of developing novel clinical strategies for the management of obesity-related diseases such as Type 2 diabetes.
Obesity is often associated with insulin resistance , the pathophysiological basis of Type 2 diabetes . It has been known that over-nutrition-induced obesity can dampen insulin sensitivity by inducing a low-grade chronic inflammatory state characterized by overproduction of pro-inflammatory cytokines such as interleukin (IL)-1β, tumour necrosis factor (TNF)-α or IL-6 [3–5]. Although multiple factors may initiate this chronic inflammation, increased infiltration and pro-inflammatory activation of macrophages in adipose tissue are believed to be a major source of many pro-inflammatory cytokines [1,4,6]. Genetic disruption of pro-inflammatory pathways such as IκB kinase (IKK)α , c-Jun N-terminal kinase (JNK)  and toll-like receptor (TLR)-4  reduces production of pro-inflammatory cytokines in macrophages and protects mice from high-fat diet (HFD)-induced inflammation and insulin resistance.
Emerging evidence supports a role of macrophage lipid overload in promoting inflammation and insulin resistance . It was reported that circulating levels of saturated non-esterified fatty acids (NEFAs) are elevated in obesity and increased NEFAs can stimulate pro-inflammatory cytokine production in macrophages via the TLR-4 pathway . Overload of free cholesterol in macrophages also activates pro-inflammatory programmes . Additionally over-nutrition-induced impairment of insulin signalling is known to promote pro-inflammatory cytokine production by activating forkhead box-containing protein O subfamily-1 (FOXO1) [11–13]. FOXO1 is a transcription factor downstream of insulin signalling. After dephosphorylation, FOXO1 translocates from cytoplasm to nucleus to regulate gene transcription . FOXO1 was shown to directly transactivate IL-1β transcription in insulin-resistant macrophages . Specific ablation of macrophage FOXO1 protects mice from HFD-induced inflammation and insulin resistance . IL-1β, a potent inducer of insulin resistance , was transcriptionally regulated by many inflammatory factors . Active IL-1β is secreted after proteolytic cleavage of pro-IL-1β by caspase-1 following activation of inflammasomes [19,20]. It was shown that NEFA induces transcription of IL-1β mRNA and NLRP3 inflammasome-mediated cleavage of pro-IL-1β protein .
Comparative gene identification-58 (CGI-58) is an intracellular lipolytic activator whose deficiency causes lipid overload in most cells including macrophages [21,22]. Our previous studies showed that CGI-58 deficiency in macrophages suppresses NLRP3 (nucleotide-binding oligomerization domain-like receptor family, pyrin domain containing 3) inflammasome-dependent IL-1β secretion in response to over-nutrition . We also found that macrophage CGI-58 deficiency increases IL-1β mRNA levels , but the molecular mechanisms linking CGI-58 deficiency to activation of IL-1β transcription was not known. In the present study, we tested the hypothesis that macrophage CGI-58 deficiency-induced insulin resistance in macrophages may activate FOXO1-dependent IL-1β transcription. Consistently, we identified a positive feedback loop, the IL-1β–SOCS3 (suppressor of cytokine signalling 3)–FOXO1–IL-1β cycle that is downstream of the CGI-58 deficiency-inflammasome pathway in macrophages. Stimulation of this loop by over-nutrition-induced suppression of CGI-58 in macrophages may be implicated in obesity-related chronic inflammation and insulin resistance.
MATERIALS AND METHODS
All mice were housed in a pathogen-free facility at the Third Military Medical University (TMMU), according to the Institutional Animal Care and Use Committee guidelines. Four-week-old mice were fed a chow diet (CD) prepared by TMMU or a HFD (D12492, fat content 60% by calorie, Research Diets, Inc.). N-Acetyl-L-cysteine (NAC) (Sigma) was administered at 10 mM in drinking water ad libitum for 3 weeks starting at 13 weeks of HFD feeding. Anakinra (Kineret) was administrated subcutaneously at 1 mg/kg/day for 2 weeks starting at 14 weeks of HFD feeding. AS1842856 (#344355, Calbiochem) was gavaged at 100 mg/kg every 12 h for one week starting at 15 weeks of HFD feeding. Clodronate liposome (FormuMax) was administered intraperitoneally (i.p.) bi-weekly at 110 mg/kg for 3 weeks starting at 13 weeks of HFD feeding.
Primary macrophages and Raw264.7 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS at 37°C in a humidified 5% CO2 atmosphere.
Preparation of saturated non-esterified fatty acid
Stearic acid (Sigma) was dissolved in ethanol at a high concentration and then diluted with 5% fatty-acid-free BSA (Sigma) to obtain a 200 mM stock solution for cell treatments.
Isolation of peritoneal macrophages
Each mouse was injected intraperitoneally with 2 ml of 3% thioglycollate (Sigma) on day 1 and killed with CO2 or isoflurane on day 3. After intraperitoneal injection of 5 ml of DMEM containing 10% FBS and penicillin/streptomycin, the peritoneal cells were collected to cell culture dishes. Two hours after culture, the floating cells were removed by washing the cells with PBS. The attached cells were considered as peritoneal macrophages (PMs) and subjected to the experiments.
Isolation of white adipose tissue-derived macrophages
Isolation of adipose tissue derived macrophages (ATMs; F4/80+ cells) from stromal vascular fraction (SVF) cells was performed by magnetic immunoaffinity isolation with anti-F4/80 antibodies conjugated to magnetic beads (MACS; Miltenyi Biotec). Cells were isolated using positive selection columns (MACS) prior to preparation of whole-cell lysates for mRNA analysis by real-time PCR.
Isolation of primary hepatocytes
Mice were anesthetized by subcutaneous injection of 200 mg ketamine/kg and 16 mg xylazine/kg. The liver was perfused via the inferior vena cava with a buffer (pH 7.5) containing 5.4 mmol/l KCl, 137 mmol/l NaCl, 0.5 mmol/l NaH2PO4, 0.42 mmol/l Na2HPO4, 0.5 mmol/l EGTA, 10 mmol/l HEPES, 4.2 mmol/l NaHCO3 and 5 mmol/l glucose. The collagenase buffer was then perfused. The collagenase buffer (pH 7.5) contained 5.4 mmol/l KCl, 137 mmol/l NaCl, 5 mmol/l CaCl2, 0.42 mmol/l Na2HPO4, 0.5 mmol/l NaH2PO4, 10 mmol/l HEPES, 0.15 g/l collagenase B (Roche), 0.05 g/l trypsin inhibitor, 4.2 mmol/l NaHCO3 and 0.016 mmol/l Phenol Red. The collagenase-perfused liver was then dissected, suspended in Hanks’ solution and filtered through cheesecloth and a 100-μm nylon membrane. Hepatocytes were subjected to centrifugation (42 g, 2 min at 4°C) and resuspended in Hanks’ solution; and this was repeated four times. The hepatocytes were then purified using density gradient centrifugation (45% Percoll solution, 42 g for 10 min at 4°C). Cell viability, measured by Trypan Blue exclusion, was more than 90%.
Glucose tolerance test and insulin tolerance test
For glucose tolerance test (GTT), the mice were fasted for 10 h during the dark cycle. The tail vein was cut and the baseline blood glucose levels were measured using a glucometer and strips (Roche). The mice were then injected intraperitoneally with glucose at 1.5 g/kg body weight (BW) and the blood glucose levels were monitored at 1, 15, 30, 60 and 120 min post glucose injection. For insulin tolerance test (ITT), the mice (1 week after GTT) were fasted for 4 h during the light cycle and the baseline blood glucose levels were measured via the tail vein. The recombinant human insulin (Sigma) was then injected intraperitoneally into the mice at 1.5 units/kg BW. The blood glucose levels were measured at 0, 15, 30, 60, 120 min after insulin administration.
The concentrations of mouse inflammatory cytokines from cultured macrophage supernatant or mouse plasma samples were measured with TNF-α, IL-1β, IL-6 and IL-10 ELISA kits from R&D system according to the manufacture's protocols. For the measurements of macrophage supernatant, the final concentration was normalized to the amount of total DNA of the cultured cells.
Western blot analysis
Tissue and cell proteins were extracted with RIPA (radioimmunoprecipitation assay) lysis buffer (Beyotime) and quantified with BCA kit (Beyotime). Immunoblotting was performed as previously described . The primary antibodies included: anti-CGI-58 (Abnova), anti-phospho-AKT [also called PKB (protein kinase B) (Thr308); Cell Signaling], anti-AKT (Cell Signaling), anti-GAPDH (Cell Signaling), anti-SOCS3 (Cell Signaling), anti-IRS2 (insulin receptor substrate 2; Cells Signaling), anti-FOXO1 (Santa Cruz) and anti-phospho-FOXO1 (Ser256) (Cell Signaling). The filter was incubated with Enhanced Chemiluminescence Substrate (PerkinElmer) for 1 min and the signals were captured with a Bio-Rad ChemiDoc MP System.
Gain-of-function and loss-of-function studies
For stable CGI-58 knockdown in Raw264.7 murine macrophages, vector-based CGI-58 shRNA (Thermo Scientific) lentivirus or PLKO control lentivirus were transfected and the stably transfected macrophages were selected with puromycin . For transient silencing of SOCS3 or FOXO1, the well-documented siRNAs specific for murine SOCS3  or FOXO1  and the scramble control siRNA were synthesized and transfected into the primary macrophages for more than 24 h.
For overexpression of CGI-58 or IRS2 in primary macrophages, the pcDNA3.1 mammalian expressing vector was used. Murine CGI-58 was subcloned from a plasmid construct containing the cDNA of murine CGI-58 (Origene). Murine IRS2 was cloned from the genomic (NM_001081212.1) DNA of Raw264.7 cells.
For overexpression of FOXO1, the adenovirus was used to express the constitutively active FOXO1, the three conserved Akt/PKB sites (Thr24, Ser256 and Ser319) of which were specifically mutated into alanine.
DNA constructs, cell transfection, site-directed reporter gene mutagenesis and reporter assays
Detailed procedures were described previously . The DNA fragments of the five mouse IL-1β promoter fusion reporter constructs shown in Figure 2A were generated from Raw264.7 cell genomic DNA by PCR amplification using KOD-Plus Kit (Toyobo). The transfection was performed by the protocol of Lipofectamine-2000 (Invitrogen). The potential FOXO1-binding site (CCAAAACAATTT) in the IL-1β promoter region was mutated as CCAGAGCGATTT with a MutanBEST Kit (Takara Bio). Luciferase activity of the cell lysate was measured according to the manufacture's instructions (Promega), and normalized by the total protein concentration in each assay.
ChIP assays were used to measure NEFA–CGI-58–SOCS3–IRS2 pathway-mediated binding between FOXO1 protein and IL-1β promoter DNA. Briefly, cultured macrophages were cross-linked with 1% formaldehyde, followed by sonication. The supernatant with equal amounts of protein were immunoprecipitated with 1 μg of mouse FOXO1 antibody or rabbit IgG as control using the ChIP Kit (Millipore Corp.) according to the manufacture's protocol. The immunoprecipitates were analysed by PCR for detection of the co-immunoprecipitated DNA containing the functional FOXO1-binding site. The primers are 5′-cttggtctccccagatctta-3′ (forward) and 5′-tgttcatgagcacagtccat-3′ (reverse). The length of the expected PCR product was 100 bp.
Isolation of peripheral blood monocytes
Human peripheral blood monocytes were isolated from healthy donors by Ficoll–Hypaque gradient centrifugation (Pharmacia Biotech), washed three times in PBS and resuspended in complete RPMI (Roswell Park Memorial Institute medium) 1640 medium containing 10% FBS, 100 units/ml penicillin, 100 mg/ml streptomycin sulfate and 2.5 mg/ml amphotericin B. The cells were cultured for 2 h at 37°C in a humidified incubator supplied with 5% CO2. The adherent monocytes were collected for measuring mRNA levels.
All data are expressed as mean ± S.E.M. and were analysed using either one-way ANOVA or two-tailed unpaired Student's t test. For each parameter of all data presented, *P<0.05, **P<0.01, ***P<0.001 and values not sharing a common small letter differ significantly (P<0.05).
Macrophage CGI-58 negatively and IL-1β positively correlated with HFD-induced weight gain
To study the relationship between macrophage CGI-58 or IL-1β expression levels and HFD-induced metabolic disorders, we fed C57BL/6 mice a HFD for 16 weeks. As expected, HFD compared with CD induced glucose intolerance and insulin resistance (Supplementary Figure S1A; Figure 1B). Meanwhile, HFD also induced an increase in IL-1β mRNA and a decrease in CGI-58 mRNA in visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT; Figure 1A). We randomly chose the male C57BL/6 mice with BW after HFD feeding and found that the fasting plasma glucose levels were positively correlated with BW. We extracted the visceral fat-derived macrophages (ATM) from these animals and measured ATM mRNA levels for CGI-58 and IL-1β. Interestingly, although IL-1β mRNA levels correlated positively with BW (Figure 1C), the opposite was found for CGI-58 mRNA levels (Figure 1D). These results imply a negative correlation between CGI-58 and IL-1β during development of metabolic disorders induced by over-nutrition.
In ATMs, CGI-58 expression negatively and IL-1β expression positively correlate with weight gain
NEFA-induced suppression of CGI-58 expression promotes FOXO1-dependent IL-1β transcription in macrophages
To explore the potential molecular link between CGI-58 expression and IL-1β transcription, we generated a series of reporter constructs harbouring different mouse IL-1β promoter regions (Figure 2A). Reporter assays revealed that -1000/-750 region was responsible for CGI-58 deficiency-induced IL-1β promoter activity (Figure 2B). Several potential transcription factor-binding sites in that region were predicted (http://www.cbrc.jp/research/db/TFSEARCH.html). By performing site-directed mutagenesis (Figure 2C), we identified a potential insulin responsive element (IRE) (CCAAACAATTT) that was responsible for CGI-58 knockdown-induced promoter activity of IL-1β (Figure 2D). Interestingly, this element was previously reported as a FOXO1-binding site in macrophages .
NEFA-induced suppression of CGI-58 expression promotes FOXO1-dependent IL-1β transcription in macrophages
To determine the relationship between CGI-58 and FOXO1 in macrophages associated with obesity, we treated primary murine PMs with saturated NEFA to mimic over-nutrition. Stearic acid (18:0) substantially reduced CGI-58 protein expression, which was associated with attenuated FOXO1 phosphorylation (indicative of functional activation of FOXO1; Figure 2E). Overexpression of CGI-58 (Supplementary Figure S2A) in NEFA-treated macrophages largely rescued FOXO1 phosphorylation (Figure 2E). These results suggest that stearic acid activates FOXO1 through suppression of CGI-58 expression in macrophages. It should be pointed out that manipulation of CGI-58 expression levels had no effects on macrophage viability (Supplementary Figure S2B), although it affected FOXO1 expression and FOXO1 was known to regulate cell apoptosis.
To investigate whether FOXO1 activation was responsible for augmented IL-1β transcription induced by CGI-58 deficiency, we overexpressed CGI-58 or silenced FOXO1 (Supplementary Figure S2C) in NEFA-treated macrophages. Stearic acid treatment decreased CGI-58 and increased IL-1β mRNA levels (Figure 2F). Overexpression of CGI-58 or silencing of FOXO1 largely abolished NEFA-induced IL-1β mRNA expression (Figure 2F). Additionally, adenovirus-mediated overexpression of constitutively active FOXO1 in NEFA-treated macrophages significantly induced IL-1β mRNA expression, which cannot be blocked by CGI-58 overexpression (Supplementary Figures S2E and S2F). These results indicate that FOXO1 is a downstream effector of the NEFA–CGI-58 axis for inducing IL-1β mRNA transcription. Furthermore, AS1842856, a specific inhibitor of FOXO1, blocked the activity of FOXO1 (Supplementary Figure S2D) and inhibited NEFA-induced IL-1β mRNA expression in primary macrophages (Figure 2G). We also found that NAC, an antioxidant, inhibited FOXO1 activity by inducing FOXO1 phosphorylation (Supplementary Figure S2D) and blocked NEFA-induced IL-1β mRNA expression (Figure 2G). Similar results were obtained when IL-1β promoter activity was examined under the same conditions (Figure 2H).
To determine how NEFA-induced CGI-58 suppression regulates FOXO1 binding to IL-1β promoter DNA, we performed ChIP assays. NEFA-induced FOXO1-binding to the aforementioned IRE in IL-1β promoter DNA and CGI-58 overexpression or FOXO1 silencing attenuated this binding (Figure 2I). Another forkhead transcription factor FOXO3a did not bind to this IRE (result not shown), suggesting the specificity of FOXO1 binding. Under in vivo conditions, macrophage-specific CGI-58 gene knockout aggravated HFD-induced glucose intolerance (Supplementary Figure S2G) and insulin resistance (Supplementary Figure S2H) and this effect was abolished by the treatment with the antioxidant NAC (Supplementary Figures S2G and S2H) that also inhibited FOXO1 activation (Supplementary Figure S2D). These results collectively demonstrate that NEFA induces IL-1β transcription by activating FOXO1 through suppression of CGI-58 expression.
NEFA-induced CGI-58 deficiency in macrophages activates FOXO1 in a SOCS3-dependent manner
To define how CGI-58 regulates FOXO1 activity, we screened gene expression changes in CGI-58-knockdown macrophages with Agilent mRNA Expression Profile Microarray (Figure 3A) and analysed the significantly altered biological processes and signal pathways (Figure 3B). Notably, gene transcription and inflammatory response were significantly altered (Figure 3B). CGI-58 deficiency negatively regulated insulin receptor signalling and positively regulated IL-1β production and secretion (Figure 3B). FOXO1 is a canonical target of insulin signalling and it is negatively regulated by insulin signalling-dependent phosphorylation . The profile of genes related to insulin signalling was changed (Figure 3A). These findings led us to speculate that FOXO1 may link insulin signalling to IL-1β production. Consistently, mRNA levels of SOCS3, protein tyrosine phosphatase non-receptor type 12 (PTPN12) and PTPN14 were increased and mRNAs for CGI-58 and IRS2 were decreased in CGI-58-knockdown macrophages (Figure 3C). Furthermore, CGI-58 knockdown increased SOCS3 and decreased IRS2 protein expression levels, which was associated with attenuated phosphorylation of AKT and FOXO1 (Figures 3D and 3E). Since SOCS and PTPN families are general suppressors of insulin signalling [26–28], our findings imply that CGI-58 deficiency may dampen insulin signalling through modulation of SOCS3, PTPN12, PTPN14 and/or IRS protein expression. Since silencing of PTPN12 (Supplementary Figure S3A) or PTPN14 (Supplementary Figure S3B) did not exert any effects on CGI-58 deficiency-induced IL-1β mRNA expression in macrophages in response to NEFA stimulation (Supplementary Figure S3C), we focused on SOCS3. NEFA treatment decreased CGI-58 expression, dampened insulin signalling (reduced IRS2 and AKT phosphorylation) and deceased FOXO1 phosphorylation, and, interestingly, siRNA-mediated silencing of SOCS3 (Supplementary Figure S3D) abolished NEFA effects on insulin signalling and FOXO1 activation (Figures 4A and 4B).
NEFA-induced suppression of CGI-58 expression dampens insulin signalling in macrophages
NEFA induces IL-1β transcription via SOCS3–IRS2 pathway in macrophages
NEFA induces IL-1β transcription via the SOCS3–IRS2 pathway in macrophages
To investigate whether NEFA also induces IL-1β transcription in a SOCS3-dependent manner, we transfected the NEFA-treated macrophages with a SOCS3-specific siRNA (Supplementary Figure S3E). To determine if IRS2 is involved in the pathway, we also overexpressed IRS2 in these cells (Supplementary Figure S3E). As expected, NEFA induced IL-1β mRNA expression and intriguingly this induction was prevented by SOCS3 silencing or IRS2 overexpression (Figure 4C). Additionally, SOCS3 silencing or IRS2 overexpression inhibited IL-1β promoter activity (Figure 4D) as well as the binding activity between FOXO1 and IL-1β promoter (Figure 4E).
CGI-58–inflammasome pathway-dependent IL-1β secretion regulates SOCS3 expression
Given that SOCS3 expression is tightly regulated by inflammatory cytokines  and that macrophage CGI-58 deficiency induces IL-1β secretion via Nlrp3 inflammasome pathway , we hypothesized that macrophage CGI-58 deficiency-induced IL-1β secretion may increase SOCS3 expression. Indeed, inactivation of the inflammasome pathway by treatment with an antioxidant (GSH; Figure 5A) or nlrp3 gene silencing (Figure 5B) largely decreased CGI-58 deficiency-induced SOCS3 mRNA expression. This finding was confirmed at the protein level (Supplementary Figure 4SA). Our previous study showed that CGI-58 deficiency potentiated ROS production . In the present paper, we show that NEFA stimulated ROS production partly through suppression of CGI-58 expression (Supplementary Figure 4SB). Furthermore, siRNA-mediated knockdown of IL-1β gene expression (Figure 5C) or Anakinra-mediated blockade of IL-1β receptor signal also abolished CGI-58 deficiency-induced SOCS3 mRNA or protein increases (Figures 5D and 5E). Recombinant IL-1β protein was a potent inducer of SOCS3 protein expression (Figure 5F). In primary PMs, Anakinra treatment also prevented NEFA-induced SOCS3 expression (Figure 5G). These results indicate that CGI-58 deficiency increases SOCS3 expression by activating the Nlrp3 inflammasome pathway. Collectively our results demonstrate that NEFA-induced suppression of CGI-58 activates a positive feedback pro-inflammatory loop (IL-1β–SOCS3–IRS2–FOXO1–IL-1β) in macrophages.
CGI-58 deficiency induces SOCS3 expression via activation of inflammasome-mediated IL-1β secretion
NEFA induces insulin resistance via CGI-58 deficiency-induced activation of the IL-1β–SOCS3–FOXO1–IL-1β loop in macrophages
To explore the role of the IL-1β–SOCS3–FOXO1–IL-1β loop in NEFA-induced insulin resistance, we treated fat tissues or primary hepatocytes isolated from wild-type C57BL/6 mice with different conditional media. NEFA-induced secretion of IL-1β was largely diminished by overexpression of CGI-58, knockdown of SOCS3 or FOXO1 or pharmacological inhibition of FOXO1 activity by AS1842856 or NAC in cultured macrophages (Figure 6A). The supernatant of NEFA-treated macrophages dampened insulin-stimulated AKT phosphorylation in fat tissues and this effect was largely abolished by CGI-58 overexpression, FOXO1 silencing in macrophages or treatment with a FOXO1 inhibitor AS1842856 or NAC (Figures 6B and 6C). Similar results were obtained in macrophage conditional medium-treated primary hepatocytes (Figures 6D and 6E).
NEFA induces insulin resistance via a CGI-58–FOXO1–IL-1β axis in macrophages
Inactivation of FOXO1 improves HFD-induced insulin resistance in mice through inhibition of CGI-58 deficiency-stimulated IL-1β production
In a previous study, we documented that macrophage CGI-58 deficiency aggravates HFD-induced insulin resistance in mice and that NAC treatment prevents this effect . In the present study, we show that NAC, a well-known antioxidant, inhibited FOXO1 activity (Supplementary Figure S2D). These results imply that macrophage FOXO1 may be involved in macrophage CGI-58 deficiency-aggravated insulin resistance in vivo. To further verify the physiological function of macrophage CGI-58–FOXO1–IL-1β axis in obesity, we fed mice with a regular CD or HFD for 16 weeks and then treated them with AS1842856, NAC and IL-1 receptor antagonist (Anakinra) at the time points indicated in Figure 7A. The pharmacological treatments had no effects on weight gain in mice on HFD (Supplementary Figure S5A). As expected, HFD relative to CD induced glucose intolerance and insulin resistance (Figures 7B and 7C). The FOXO1 inhibitor AS1842856 attenuated and NAC or Anakinra abolished the HFD-induced glucose intolerance and insulin resistance (Figures 7B and 7C). To determine whether these drugs acted through macrophages, we employed clodronate liposomes (CLO) to delete macrophages in vivo. One week after administration of CLO, HFD-stimulated macrophage infiltration into visceral fat was largely diminished (Supplementary Figure S5B). HFD-induced mRNA increases in macrophage marker F4/80 in visceral fat (Supplementary Figure S5C) and liver (Supplementary Figure S5D) were also decreased. GTTs and ITTs revealed that the beneficial effect of AS1842856 and NAC (but not Anakinra) disappeared in HFD-fed mice with macrophage deletion (Figures 7D and 7E), indicating that AS1842856 and NAC acted through macrophage FOXO1 to improve insulin sensitivity at least to some extent, whereas Anakinra targeted more tissues.
Inhibiting FOXO1 function improves HFD-induced insulin resistance in mice by inhibiting CGI-58 deficiency-induced IL-1β expression in macrophages
Additionally, we examined inflammatory cytokine production in plasma and found that the levels of pro-inflammatory cytokines IL-1β (Figure 7F), IL-6 (Figure 7G) and TNF-α (Figure 7H) were increased, whereas the anti-inflammatory cytokine IL-10 was decreased (Figure 7I) in HFD-fed mice. AS1842856 or NAC treatment largely suppressed the HFD-induced changes in plasma cytokine levels (Figures 7F–7I). To identify the origins of cytokines, we isolated ATMs and performed quantitative real-time PCR to measure gene expression levels. The mRNA expression profile of inflammatory cytokines in ATMs was identical with what seen in plasma (Figure 7J), suggesting that ATMs may be an important source of plasma cytokines. In addition, HFD induced a reduction in CGI-58 and IRS2 mRNA levels and an increase in SOCS3 mRNA in ATMs without altering FOXO1 mRNA levels (Supplementary Figure S5E). Additional FOXO1 inhibitors suppressed HFD-induced SOCS3 and pro-inflammatory cytokine expression and increased IRS2 mRNA levels in ATMs (Supplementary Figure S5E). These results are consistent with a potential role of the macrophage CGI-58–FOXO1–IL-1β axis in obesity, though definitive answers could only be established by using sophisticated, genetically altered animal models.
Peripheral blood monocyte CGI-58 negatively and IL-1β positively correlate with the body mass index in humans
The aforementioned results demonstrate an important role of the CGI-58–SOCS3–IRS2–FOXO1 pathway in NEFA-induced IL-1β production. To correlate this mechanism to physiological conditions, we isolated peripheral blood monocytes from age-matched male human donors and analysed the correlation between the mRNA level of CGI-58, SOCS3, IRS2 or IL-1β and the body mass index (BMI). We found that CGI-58 (Supplementary Figure S6A) and IRS2 (Supplementary Figure S6B) mRNA expression levels in peripheral blood monocytes were negatively correlated with the BMI in human donors, whereas the mRNA levels of SOCS3 (Supplementary Figure S6C) and IL-1β (Supplementary Figure S6D) as well as the plasma glucose levels (Supplementary Figure S6E) were positively correlated with the human BMI.
Our previous studies showed that macrophage CGI-58 is an endogenous suppressor of the NLRP3 inflammasome that proteolytically cleaves pro-IL-1β to produce mature IL-1β . We also observed that IL-1β mRNA levels were up-regulated in CGI-58-deficient macrophages, but how CGI-58 deficiency promotes IL-1β transcription was not known. In the present paper, we showed for the first time that macrophage CGI-58 deficiency increased IL-1β transcription via promoting direct binding of FOXO1 to IRE in the IL-1β promoter. Our results thus reveal a dual role of CGI-58 in suppressing IL-1β secretion in macrophages.
Administration of AS1842856, a newly identified specific inhibitor of FOXO1 transcription activity , attenuates HFD-induced chronic inflammation and insulin resistance. In the present study, we found that CGI-58 expression levels in monocytes or macrophages were suppressed in obesity at least partly via a NEFA-dependent mechanism. Together with our previous finding that macrophage-specific deletion of CGI-58 aggravates HFD-induced chronic inflammation and insulin resistance , we propose that the NEFA–CGI-58–FOXO1 axis may be implicated in HFD-induced insulin resistance. In the present study, AS1842856 treatment attenuated HFD-induced production of IL-1β and other pro-inflammatory cytokines in plasma and ATMs, suggesting that it can mimic the anti-inflammatory and anti-insulin resistant functions of Anakinra, an IL-1 receptor antagonist . These results suggest that FOXO1 may be a promising therapeutic target for obesity-related disorders, such as insulin resistance and Type 2 diabetes. Notably, NAC relative to the FOXO1-specific inhibitor is more potent in protecting HFD-fed mice from insulin resistance. Our data suggest that this difference may be due to, at least in part, NAC's dual actions on inflammation suppression: (1) NAC inhibits FOXO1 activity to suppress IL-1β transcription; and (2) NAC inactivates the NLRP3 inflammasome to reduce IL-1β protein secretion.
CGI-58 was shown to function as a co-activator of adipose triacylglycerol (triglyceride) lipase (ATGL) to promote intracellular hydrolysis . However, genetic evidence supports that CGI-58 must have ATGL-independent functions [32,33]. ATGL-deficient macrophages are prone to an M2 (anti-inflammatory) phenotype , whereas CGI-58-deficient macrophages display an M1 (pro-inflammatory) phenotype due to activation of the NLRP3 inflammasome and concomitant augmentation of IL-1β secretion . In the present study, we further showed that CGI-58 deficiency promotes FOXO1-dependent IL-1β transcription. By performing a gene expression microarray, we found that macrophage CGI-58 deficiency increased SOCS3 and decreased IRS2 expression, leading to dampened insulin signalling. IRS2 is a well-documented component of insulin signalling pathway  and SOCS3 was identified as an upstream inhibitor . SOCS3 was reported to induce insulin resistance through ubiquitin-mediated degradation of IRS2 in insulin-responsive cells  and this provided a potential mechanism linking SOCS3 overexpresion to IRS2 reduction in CGI-58 deficient macrophages. With gain-of-function or loss-of-function studies, we further verified the role of the CGI-58–SOCS3–IRS2–FOXO1 pathway in regulating IL-1β transcription ex vivo. Additionally, we found that this mechanism correlates well with obesity in mice and humans. Despite these interesting findings, the molecular mechanisms underlying obesity-associated suppression of CGI-58 expression in macrophages remain to be elucidated. In our previous study, we showed that LPS (lipopolysaccharide) inhibits CGI-58 expression in macrophages . In the future, it would be interesting to test whether TLR signalling is implicated in regulation of CGI-58 expression.
It should be pointed out that antisense oligonucleotide (ASO)-induced knockdown of CGI-58 in multiple tissues of mice improves instead of inhibits insulin sensitivity [36,37]. Although this result appears to be contrary to our previous and current studies with macrophage-specific CGI-58 knockout mice [21,38,39], one should keep in mind that ASOs knockdown gene expression in adipose tissues, liver, macrophages and, to a lesser extent, some other tissues. As we all know, each cell type has its distinct metabolic programme and function and each plays different roles in integrated pathophysiology. In those two ASO studies, what the authors reported was the net outcome of CGI-58 deficiency in multiple tissues/cell types. Actually, it is expected that fat-specific knockout of CGI-58 will improve insulin sensitivity and glucose tolerance because CGI-58-deficient adipose tissue may have a defect in mobilizing NEFA as energy and as a result insulin-sensitive tissues tend to use more glucose. In addition, CGI-58 ASO-treated mice stay lean on HFD for unknown reasons, which contribute at least in part to improved insulin sensitivity in these ASO-treated animals. Furthermore, ASOs, probably any other chemical compounds, may have off-target effects on cell metabolism. In our current study, we genetically deleted CGI-58 expression specifically in macrophages. The macrophage-specific CGI-58 knockout mice relative to controls were more insulin-resistant when challenged with a HFD . Studies from our laboratory and, more recently, from others clearly demonstrated that CGI-58-deficient macrophages are pro-inflammatory [21,38,39]. It has been well established that pro-inflammatory activation of macrophage cells promotes insulin resistance [4,6,8,9]. Our data are thus consistent with this mainstream view of the positive correlation between pro-inflammatory macrophages and insulin resistance.
In summary, our data suggest that over-nutrition may lead to inhibition of CGI-58 expression in monocytes/macrophages through NEFA and perhaps LPS. Suppression of CGI-58 in these cells activates the NLRP3 inflammasome to promote IL-1β secretion. Secreted IL-1β then induces SOCS3 expression, resulting in IRS2 reduction and FOXO1 activation. Activated FOXO1 protein binds to IL-1β promoter to increase IL-1β transcription and pro-IL-1β protein expression. Meanwhile, CGI-58 deficiency stimulates the NLRP3 inflammasome pathway to promote cleavage of pro-IL-1β protein and hypersecretion of mature IL-1β, which further increases SOCS3 expression. Thus, we identified a novel positive feedback loop (IL-1β–SOCS3–FOXO1–IL-1β) downstream of the CGI-58 deficiency-induced activation of the NLRP3 inflammasome (Figure 8). Targeting this loop probably improves obesity-related chronic inflammation and insulin resistance.
A proposed model for macrophage CGI-58-mediated protection against obesity-related insulin resistance
adipose triacylglycerol lipase
adipose tissue derived macrophage
body mass index
comparative gene identification-58
Dulbecco's modified Eagle's medium
forkhead box-containing protein O subfamily-1
glucose tolerance test
insulin responsive element
insulin receptor substrate
insulin tolerance test
non-esterified fatty acid
nucleotide-binding oligomerization domain-like receptor family, pyrin domain containing 3
protein kinase B
protein tyrosine phosphatase non-receptor
subcutaneous adipose tissue
suppressor of cytokine signalling 3
Third Military Medical University
tumour necrosis factor
Hongming Miao, Juanjuan Ou and Xuan Zhang conducted the experiments, analysed the data and wrote the manuscript. Yujuan Chen, Bingzhong Xue, Hang Shi and Lixia Gan contributed to discussion. Hongming Miao and Liqing Yu designed experiments, discussed the data and revised the paper. Houjie Liang is the guarantor of the present work, had full access to all the data and takes full responsibility for the integrity of data.
This work was supported in part by the National Natural Science Foundation of China [grant numbers 81302136 (to H.M.), 81370063 (to J.O.) and 30973430 and 81272364 (to H.L.)]; the General Financial Grant from the China Postdoctoral Science Foundation [grant number 2013M542437 (to H.M.)]; and the National Institute of Diabetes and Digestive and Kidney Diseases, U.S.A [grant number R01DK085176 (to L.Y.)].
These authors contribute equally to this work.