CCN1 expression in hepatocytes contributes to macrophage infiltration in nonalcoholic fatty liver disease in mice.

Our objective was to investigate the potential roles of CCN1 in the inflammation and macrophage infiltration of nonalcoholic fatty liver disease (NAFLD). The regulation of hepatic CCN1 expression was investigated in vitro with murine primary hepatocytes treated with free fatty acids or lipopolysaccharide (LPS) and in vivo with high-fat (HF) diet-fed mice or ob/ob mice. CCN1 protein and a liver-specific CCN1 expression plasmid were administered to mice fed a normal diet (ND) or HF diet. Myeloid-derived macrophages and RAW264.7 cells were also treated with CCN1 in vitro to determine the chemotactic effects of CCN1 on macrophages. LPS treatment significantly increased hepatic CCN1 expression in HF diet-fed mice and ob/ob mice. LPS and FFAs induced CCN1 expression in primary murine hepatocytes in vitro through the TLR4/MyD88/AP-1 pathway. CCN1 protein and overexpression of CCN1 in the liver induced more severe hepatic inflammation and macrophage infiltrates in HF mice than in ND mice. CCN1 recruited macrophages through activation of the Mek/Erk signaling pathway in myeloid-derived macrophages and RAW264.7 cells in vitro. Endotoxin and FFA-induced CCN1 expression in hepatocytes is involved in the hepatic proinflammatory response and macrophage infiltration in murine NAFLD.


Statistical analysis
Data are expressed as the mean ± SD. The group means were compared using ANOVA. All statistical analysis was performed using the SPSS statistical software version 16.0 (SPSS Inc., Chicago, IL). P values < 0.05 were considered statistically signifi cant.

LPS increases hepatic CCN1 expression in murine NAFLD
Our previous studies demonstrated that mice fed HF diets became obese and developed hepatic steatosis ( 18,19 ). In the present study, HF mice also developed signifi cant hepatic steatosis characterized by the presence of macrovesicular and microvesicular lipid droplets. Immunohistochemical analysis showed that LPS administration signifi cantly increased hepatic CCN1 expression in HF mice compared with HF mice without LPS administration and LPS-treated ND mice ( Fig. 1A , upper panel). To confi rm that hepatic CCN1 was increased in different steatosis models, ob/ob mice were treated with or without LPS. Similarly, hepatic CCN1 expression was increased in ob/ob mice compared with WT mice. LPS treatment further increased hepatic CCN1 expression in ob/ob mice ( Fig. 1A , lower panel).
FFAs and LPS in hepatocytes resulted in macrophage infi ltration and infl ammation in the liver in hepatic steatosis.

Animal and treatments
Male C57BL/6 mice (6-8 weeks of age) were purchased from the Shanghai SLAC Laboratory Animal Co. Ltd (Shanghai, China) and housed under pathogen-free conditions in the animal facility of the Shanghai Jiao Tong University School of Medicine. Mice were fed either a high-fat (HF) diet (SLAC Laboratory Animal Co. Ltd) providing 59% of calories from fat, 25% from carbohydrates, and 16% from protein for eight weeks or an isocaloric normal diet (ND) containing less fat (12% fat, 59% total carbohydrate. and 29% protein). Male C57BL/10ScN toll-like receptor 4 (TLR4) knockout mice, male C57BL/6 myeloid differentiation factor 88 (MyD88) knockout mice, their control littermates, and male C57BL/6 ob/ob mice (6-8 weeks of age) were obtained from the Model Animal Research Center of Nanjing University (Nanjing, China). All mice were maintained in a temperature-and lightcontrolled facility with ad libitum access to food and water.
Escherichia coli LPS (50 µg/mouse, Sigma, St. Louis, MO) was injected intraperitoneally, and mice were euthanized 6 h later. The CCN1 protein displays a remarkable degree of evolutionary conservation, with 92.8% identity between mouse and human CCN1 ( 11 ). Recombinant human CCN1 (5 g/mouse, R and D Systems, Minneapolis, MN) was administered intravenously through the tail vein, and mice were euthanized 24 h later. The neutralizing anti-integrin ␣ M antibody (0.1 mg/mouse, BD Pharmingen, San Diego, CA) or Z-VAD-FMK (200 g/mouse, Sigma), which inhibits induction of apoptosis, was injected through the tail vein 2 h before administration of CCN1. Control animals were injected with vehicle only.
The plasmid encoding mouse CCN1 cDNA or control vector (50 g/mouse) was dissolved in PBS with a volume of 2 ml or 3 ml and then delivered to ND and HF mice by hydrodynamic tail vein injection, respectively ( 12 ). All animal experiments were app roved by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University School of Medicine and were conducted in accordance with the National Research Council Guide for Care and Use of Laboratory Animals.

Liver-specifi c CNN1 expression plasmid construction
The construction of the liver-specifi c CCN1 expression plasmid was carried out as previously described ( 12 ), using the Enh1mTTR (ET) promoter generated by fusing a synthetic hepatocytespecifi c enhancer to the murine transthyretin promoter ( 13 ). The CCN1 cDNA inserted into the vector was expressed under the control of ET promoter. The control plasmid contained the GFP gene under control of the same promoter.

Cell culture
Primary murine hepatocytes were isolated from mice by in situ collagenase liver perfusion ( 14 ) and cultured in DMEM supplemented with 10% fetal bovine serum (FBS), 100 mg/ml streptomycin, and 100 U/ml penicillin (Invitrogen, Carlsbad, CA).
Primary murine macrophages were obtained as previously described with minor modifi cations ( 15 ) and cultured in DMEM supplemented with 10% FBS, 20 ng/ml macrophage colony-stimulating factor (M-CSF; R and D Systems), 100 mg/ml streptomycin, and 100 U/ml penicillin. The macrophage RAW264.7 cell line was purchased from the Shanghai Institute of Cell Biology (Chinese Academy of Sciences, Shanghai, China) and cultured with DMEM plus 10% FBS. in the liver is hepatocytes ( Fig. 1D ). Therefore, we chose hepatocytes as target cells for our next experiments to explore the regulation of CCN1 expression.

LPS and FFAs induce CCN1 expression in primary murine hepatocytes through the TLR4/MyD88/AP-1 signaling pathway
LPS at the concentrations of 1 µg/ml and 2 µg/ml induced CCN1 expression by approximately 3-fold in primary murine hepatocytes in vitro ( Fig. 2A , upper panel). In addition, CCN1 expression was detectable at Western blot analysis showed that both HF and LPS treatment increased hepatic expression of CCN1 in ND mice. However, LPS-treated HF mice had significantly higher hepatic CCN1 expression compared with the LPS-treated ND mice ( Fig. 1B ). The same phenomenon was also observed in ob/ob mice ( Fig. 1C ). By immunohistochemical staining analysis ( Fig. 1A ), we found that CCN1 was mainly expressed in hepatocytes. Similarly, Western blot analysis showed that the hepatocytes had signifi cantly higher CCN1 expression than the nonparenchymal cells, suggesting that the major source of CCN1 protein
To illustrate the downstream signaling pathway of TLR4/MyD88 involved in this process, inhibitors for the AP-1 (SP600125) and NF-B (IMD0354) pathways were used. SP600125 and IMD0354 signifi cantly inhibited the phosphorylations of c-Jun (component of AP-1) and IKB ␣ , respectively (supplementary Fig. I-A). However, only SP600125 inhibited the expression of CCN1 in primary hepatocytes stimulated with LPS or FFAs ( Fig. 3C ). These results suggest that LPS and FFAs induce CCN1 expression in hepatocytes through activation of the TLR4/MyD88/AP-1 signaling pathway.
12 h and reached a maximum value at 48 h after LPS treatment at 2 µg/ml ( Fig. 2A , lower panel). Primary murine hepatocytes were incubated with a mixture of long chain fatty acids (OP 2:1) to establish a cellular steatosis model ( 20 ). Similarly, FFAs (OP 2:1) also induced CCN1 expression in a concentration-dependent manner, peaking at 0.5 mmol/l ( Fig. 2B , upper panel). At 0.5 mmol/l, FFAs were shown to induce the maximum CCN1 expression level at 24 h ( Fig. 2B , lower  panel). These results demonstrated that LPS and FFAs both induced CCN1 expression of hepatocytes in a timeand concentration-dependent manner.
In vivo, LPS administration nearly tripled the hepatic CCN1 expression in WT mice; however, no signifi cant difference in CCN1 expression was detected between treatment with LPS and treatment without LPS in infl ammation in HF mice ( Fig. 4A ). CCN1 treatment increased the numbers of hepatic infl ammatory foci in both ND and HF mice compared with ND and HF mice without treatment, respectively. Furthermore, CCN1 induced

CCN1 induces severe hepatic infl ammation and liver injury in murine NAFLD
CCN1 protein induced mild to moderate hepatic infl ammation in ND mice, whereas it induced severe hepatic  HF mice, and Z-VAD-FMK, a caspase inhibitor, restrained this process ( Fig. 6A ). However, Z-VAD-FMK did not signifi cantly attenuate the hepatic infl ammation induced by CCN1 in either ND or HF mice ( Fig. 6B ). In addition, CCN1-induced macrophage infi ltration was not significantly reduced by Z-VAD-FMK ( Fig. 6C ), indicating that the role of CCN1 on hepatic infl ammation and macrophage infi ltration is independent of cellular apoptosis potentially induced by CCN1.

CCN1 induces macrophage chemotaxis through the integrin ␣ M /Mek1/2/Erk1/2 signaling pathway in vitro
CCN1 in different dosages increased the phosphorylation of Mek1/2 and Erk1/2 in the RAW264.7 cell line ( Fig. 7A ). CCN1 recruited primary murine myeloid-derived macrophages in a concentration-dependent manner, ranging from 250 to 1,000 ng/ml both at 90 min and 24 h. Pretreatment with the neutralizing antibody against integrin ␣ M or an inhibitor for Mek1/2 (U0126) signifi cantly reduced the level of macrophage migration in response to CCN1 ( Fig. 7B ). Similar results were also found in the murine macrophage RAW264.7 cell line ( Fig. 7C ). Taken together, these results suggest that CCN1 recruits macrophages through activation of the integrin ␣ M /Mek1/2/ Erk1/2 signaling pathway.

DISCUSSION
In this study, we demonstrated for the fi rst time that FFAs and LPS induced CCN1 expression in hepatocytes through the TLR4/MyD88/AP-1 signaling pathway in vitro and in vivo. CCN1 could induce severe hepatic infl ammation and apoptosis in mice with NAFLD, and CCN1 was shown to induce macrophage chemotaxis through the integrin ␣ M /Mek1/2/Erk1/2 signaling pathway. Therefore, CCN1 may play roles in the infl ammatory response and macrophage infi ltration in NAFLD.
Accumulating evidence shows that innate immunity is actively involved in insulin resistance and plays a critical role in the pathogenesis of NASH ( 23 ). TLR4 links innate immunity and fatty acid-induced insulin resistance ( 24,25 ). TLR4 is a high-affi nity receptor for LPS, a component of the cellular wall of gram-negative bacteria, and is considered a pivotal exogenous danger-signaling molecule in the pathogenesis of NAFLD ( 26 ). LPS has been shown to induce production of TNF-␣ in macrophages, thereby triggering TNF-␣ -induced hepatocyte apoptosis in a murine NASH model ( 27 ). In addition, TLR4 plays an important role in mediating proinfl ammatory effects of saturated FFAs. Saturated FFAs may function as the endogenous substances that contribute to TLR4 activation in the setting of obesity ( 28 ). FFAs more hepatic infl ammatory foci in HF mice compared with ND mice treated with CCN1 ( Fig. 4B ). Similarly, CCN1 treatment increased serum ALT and AST levels in both ND and HF mice ( Fig. 4C, D ). Consistent with the histological changes, CCN1-treated HF mice had higher ALT levels than CCN1-treated ND mice ( Fig. 4C ), suggesting that the HF mice were more susceptible to CCN1-induced hepatic infl ammation and liver injury.
As CCN1 induces macrophage adhesion and activation through integrin ␣ M ␤ 2 ( 8,21 ), a neutralizing antibody against integrin ␣ M was used to block the binding between CCN1 and macrophages. Neutralization of integrin ␣ M attenuated CCN1-induced hepatic infl ammation in ND and HF mice, indicated by less hepatic infl ammatory infi ltration in histological examinations and lower serum ALT and AST levels compared with ND and HF mice treated with CCN1 only ( Fig. 4A-D ). These results suggested that CCN1 induced monocyte infi ltration through binding to integrin ␣ M on the surface of monocytes. CCN1 also induced moderate hepatic infl ammation in ob/ob mice but not to the extent seen in HF mice (supplementary Fig. I-B). Delivery of a liver-specifi c CCN1 plasmid, confi rmed to successfully overexpress CCN1 (supplementary Fig. I-C), induced more severe hepatic infl ammation in HF mice than in ND mice ( Fig. 4E ).

CCN1 induces macrophage infi ltration through binding to integrin ␣ M in vivo
Next, we analyzed macrophage infi ltration in CCN1 protein-induced hepatic infl ammatory foci in both ND and HF mice by using F4/80 immunohistochemistry. F4/80 positive cells were frequently detected ( Fig. 5A ), suggesting that macrophages are the major cellular type infi ltrated in CCN1induced liver injury. More importantly, CCN1 induced more macrophage infi ltrates in HF mice compared with ND mice treated with CCN1 ( Fig. 5B ). Accordingly, neutralization of integrin ␣ M decreased CCN1-induced hepatic macrophage infi ltrates in ND and HF mice ( Fig. 5A, B ). Flow cytometric analysis showed that the majority of F4/80 positive cells in the liver expressed integrin ␣ M ( Fig. 5C ). These results suggest that integrin ␣ M may mediate macrophage infi ltration in CCN1-induced hepatic infl ammation. Consistent with the results of CCN1 protein administration, the forced expression of CCN1 also induced numerous macrophage infi ltrations in both ND and HF mice ( Fig. 5D ).

CCN1-induced hepatic infl ammation and macrophage infi ltration are not affected by apoptosis inhibitor in vivo
It was reported that CCN1 could facilitate apoptosis mediated by Fas and TNF-␣ in vivo and vitro ( 10,22 ). Therefore, we observed whether cellular apoptosis played a critical role in hepatic infl ammation and macrophage infi ltration induced by CCN1. Indeed, CCN1 induced cellular apoptosis in the livers not only of ND mice but also of mice that were either untreated or treated with CCN1 protein with or without neutralizing antibody against integrin ␣ M . (D) Mean (±SD) serum AST levels of ND and HF mice that were either untreated or treated with CCN1 protein with or without neutralizing antibody against integrin ␣ M . (E) Representative H and E staining of liver sections (magnifi cation 200×) from ND and HF mice with forced CCN1 expression in the liver. (n = 5 in each group, * P < 0.05, ** P < 0.01). may elicit hepatotoxicity and stimulate progression from simple steatosis to NASH via several mechanisms beyond direct cytotoxicity ( 29 ). Palmitic acid can bind to TLR4, leading to activation of NF-B and upregulate its target genes, such as TNF-␣ and IL-6, in macrophages and adipocytes. TLR4-deficient mice develop obesity when fed a HF diet; however, they are still partially protected against development of insulin resistance, possibly due to reduced infl ammatory gene expression in liver and fat tissues ( 24 ). FFAs also induce production of proinfl ammatory cytokines TNF-␣ ( 30 ) and IL-8 ( 31 ) from hepatocytes, potentially contributing to hepatic inflammation and consequent liver injury. Most recently, Csak and colleagues showed that the saturated FFA palmitic acid upregulates the infl ammasome, which cleaves pro-interleukin-1 ␤ (pro-IL-1 ␤ ) into secreted IL-1 ␤ , and induces sensitization to LPS for IL-1 ␤ release in hepatocytes ( 32 ). Our study showed that LPS and FFAs induced production of CCN1 in hepatocytes, which may be another important infl ammatory factor in the liver. Therefore, our study and other studies suggest that hepatocytes are not only passively injured targets but that they also are actively involved in orchestrating responses to insults in NASH. macrophages (Kupffer cells) positioned at the "frontline" is an essential element in the pathogenesis of NAFLD ( 25 ). It was shown that CCN1 can activate a proinfl ammatory genetic program in macrophages, such as monocyte chemotactic protein 1 (MCP-1) and macrophage infl ammatory protein 1 ␣ (MIP-1 ␣ ) ( 8 ). In our study, we showed Macrophages are another important component of innate immunity in NAFLD. In adipose tissue, macrophages are a major source of proinfl ammatory cytokines, which can function in a paracrine and potentially an endocrine fashion to cause decreased insulin sensitivity. The accumulating evidence indicates that activation of hepatic  that CCN1 could recruit macrophages through binding to integrin ␣ M on the cellular surface of macrophages, which at least partially explained the massive macrophage infi ltrates in murine liver after CCN1 administration. Therefore, production of CCN1 by hepatocytes after treatment with LPS and FFAs may activate and recruit macrophages into the liver. These activated hepatic macrophages then release various chemokines, such as MCP-1, which in turn recruit additional macrophages, setting up a feed-forward process that further increases the number of macrophages in the liver and propagates the chronic infl ammatory state ( 28 ). In other words, the proinfl ammatory process is initiated in NAFLD by CCN1mediated induction of macrophage infi ltration into the liver through direct chemotaxis and secondary recruitment by chemokines.
In summary, we propose for the fi rst time that endotoxins and FFAs induce expression of the matricellular signaling molecule CCN1 in hepatocytes through activation of the TLR4/MyD88/AP-1 signaling pathway in mice. CCN1 then drives the recruitment of macrophages into the steatotic liver, inducing the infl ammatory process and macrophage infi ltration. The potential role of CCN1 in the process of transition from simple steatosis to NASH in humans will be worth exploring in future studies. Thus, CCN1 may become a potential therapeutic target to prevent disease progression to advanced stages in NAFLD.