STAT3-mediated attenuation of CCl4-induced mouse liver fibrosisby the protein kinase inhibitor sorafenib

Yan-Ru Deng a,1, Hong-Di Ma a,1, Koichi Tsuneyama b, Wei Yang a, Yin-Hu Wang a,Fang-Ting Lu a, Cheng-Hai Liu c, Ping Liu c,**, Xiao-Song He d, Anna Mae Diehl e,M. Eric Gershwin d, Zhe-Xiong Lian a,*

a Liver Immunology Laboratory, Institute of Immunology and School of Life Sciences, University of Science and Technology of China, Hefei 230027, ChinabDepartment of Diagnostic Pathology, Graduate School of Medicine and Pharmaceutical Science for Research, University of Toyama, Toyama 930-0194,JapancKey Laboratory of Liver and Kidney Diseases (Ministry of Education), Institute of Liver Diseases, Shuguang Hospital, Shanghai University of TraditionalChinese Medicine, 528 Zhangheng Road, Shanghai 201203, ChinadDivision of Rheumatology, Allergy and Clinical Immunology, University of California at Davis School of Medicine, Davis, CA 95616, USAeDuke University Medical Center, Division of Gastroenterology, Durham, NC, USA

a r t i c l e i n f o

Article history:Received 12 July 2013Accepted 18 July 2013

Keywords:SorafenibLiver fibrosisSTAT3Kupffer cellIL-6

a b s t r a c t

There have been major advances in defining the immunological events associated with fibrosis in variouschronic liver diseases. We have taken advantage of this data to focus on the mechanisms of action of aunique multi-kinase inhibitor, coined sorafenib, on CCl4-induced murine liver fibrosis, including theeffects of this agent in models of both acute and chronic CCl4-mediated pathology. Importantly, sorafenibsignificantly attenuated chronic liver injury and fibrosis, including reduction in liver inflammation andhistopathology as well as decreased expression of liver fibrosis-related genes, including a-smooth muscleactin, collagen, matrix metalloproteinases and the tissue inhibitor of metalloproteinase-1. Furthermore,sorafenib treatment resulted in translocation of cytoplasmic STAT3 to the nucleus in its active form.Based on this observation, we used hepatocyte-specific STAT3 knockout (STAT3Hep!/!) mice to demon-strate that hepatic STAT3 was critical for sorafenib-mediated protection against liver fibrosis, and that theupregulation of STAT3 phosphorylation was dependent on Kupffer cell-derived IL-6. In conclusion, thesedata reflect the clinical potential of the multi-kinase inhibitor sorafenib for the prevention of fibrosis aswell as the treatment of established liver fibrosis and illustrate the immunological mechanisms thatunderlie the protective effects of sorafenib.

! 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Liver fibrosis is a significant health problem resulting fromresponse of the liver to injury [1,2]. Following injury, hepatocytes,biliary cells, Kupffer cells and T cells respond to create an inflam-matory milieu [3], which stimulates hepatic stellate cells (HSCs) totransform from a quiescent vitamin A-storing cell lineage to anactivated myofibroblast (a-SMA-expressing) and thereby acquiresfibrogenic properties producing extracellular matrix proteins(ECM) [2,4e8]. Other cell types also contribute to this pathologicalprocess and complex interplay amongst hepatic cell lineages occursduring fibrogenesis [3,9]. Hepatocytes are the major targets in liverfibrosis [10]. Damaged hepatocytes release ROS aswell as fibrogenicmediators and induce the recruitment of white blood cells by in-flammatory cells. Apoptosis of damaged hepatocytes also stimulatethe fibrogenic actions of liver myofibroblasts [11]. Thus, drugs

Abbreviations: HSC, hepatic stellate cell; ECM, extracellular matrix; a-SMA, a-smooth muscle actin; SORA, sorafenib; PDGF, platelet-derived growth factor; VEGF,vascular endothelial growth factor; CCl4, carbon tetrachloride; STAT3, signal trans-ducer and activator of transcription 3; wt, wild type; pSTAT3, phosphorylated STAT3;p-c-Raf, phosphorylated c-Raf; pMEK, phosphorylated MEK1/2; pErk, phosphory-lated p44/42 MAPK(Erk1/2); ALT, serum alanine aminotransferase; H&E, hematox-ylin-eosin; EMT, epithelialemesenchymal transition; HCC, hepatocellularcarcinoma; HGF, hepatocyte growth factor; IL-6, interleukin-6; JAK, Janus tyrosinekinases; MMP, matrix metalloproteinase; NASH, non-alcoholic steatohepatitis; TGF-b, transforming growth factor-b; Timp, tissue inhibitor of metalloproteinase.* Corresponding author. Tel./fax: þ86 551 63600317.** Corresponding author.

E-mail addresses: adbxdyr@mail.ustc.edu.cn (Y.-R. Deng), mahongdi@mail.ustc.edu.cn (H.-D. Ma), ktsune@med.u-toyama.ac.jp (K. Tsuneyama),ywei0702@mail.ustc.edu.cn (W. Yang), wangyh8@mail.ustc.edu.cn (Y.-H. Wang),fangting@mail.ustc.edu.cn (F.-T. Lu), chenghai_liu@yahoo.com.cn (C.-H. Liu),liuliver@vip.sina.com (P. Liu), xiaosong@stanford.edu (X.-S. He), annamae.diehl@duke.edu (A.M. Diehl), megershwin@ucdavis.edu (M.E. Gershwin), zxlian1@ustc.edu.cn, zxlian@ucdavis.edu (Z.-X. Lian).

1 Both authors contributed equally to the manuscript.

Contents lists available at ScienceDirect

Journal of Autoimmunity

journal homepage: www.elsevier .com/locate/ jaut imm

0896-8411/$ e see front matter ! 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.jaut.2013.07.008

Journal of Autoimmunity 46 (2013) 25e34

which protect hepatocytes from injury are potentially treatmentoptions for human liver fibrosis.

There are several pathways that can potentially be modulated toalter fibrosis. Amongst potential inhibitors is an oral multi-kinaseinhibitor well known for its influence on tumor signaling andvasculature, coined Sorafenib (BAY43-9006, Nexavar) [12e14].Signaling pathways affected by sorafenib include Raf, platelet-derived growth factor (PDGF), c-kit and vascular endothelialgrowth factor (VEGF). Sorafenib blocks receptor tyrosine kinasesignaling (VEGFR, PDGFR, c-Kit and RET) and inhibits downstreamRaf serine/threonine kinase activity to prevent tumor growth byantiangiogenic, antiproliferative and/or pro-apoptotic effects [15].

Signal transducer and activator of transcription 3 (STAT3) is atranscription factor that regulates cell growth and survival bymodulating the expression of many target genes [16]. Sorafenib hasthe potential to suppress STAT3 phosphorylation in humanneoplasia [17e19]. However, due to the hepatoprotective and pro-liferative functions of STAT3, STAT3 in hepatocytes plays a protec-tive role in preventing liver fibrosis [20e23].

Recent data has shown a beneficial effect of sorafenib in pa-tients with hepatocellular carcinoma (HCC) [12e14]. Other studiessuggest that sorafenib attenuates the hypertensive syndrome inpartial portal vein ligated rats [24,25]. There are also data thatnote that sorafenib has antifibrotic activity, including in vitro ef-fects of sorafenib on inhibition of activation, growth and collagenaccumulation of HSC [26,27]. Moreover, sorafenib influencesparacrine signaling between HSCs and LECs and thereby regulatesmatrix and vascular changes associated with bile duct ligation-induced liver injury in rats [28]. In hepatocytes, in vitro experi-ments demonstrated that sorafenib suppresses TGF-b1-inducedEMT [29]. In this study, we focused on the effect of sorafenib oncross talk between hepatocytes and Kupffer cells during liverfibrosis. We used CCl4 induced murine liver fibrosis model andthree independent acute liver injury models to illustrate thatsorafenib attenuates liver fibrosis and injury through up-regulation STAT3 phosphorylation in hepatocytes which depen-dent on Kupffer cell-derived IL-6.

2. Material and methods

2.1. Mice

C57BL/6 (B6) mice were purchased from the Experimental An-imal Center, Chinese Academy of Science (Shanghai, China).Hepatocyte-specific signal transducer and activator of transcription3 (STAT3) knockout (STAT3Hep!/!) mice (AlbCreþ/! STAT3flox/flox)were provided by Dr. Xin-Yuan Fu. B6.129S6-Il6tm1Kopf (IL-6!/!)mice were purchased from JAX (Bar Harbor, Maine, USA). All micewere housed in a specific pathogen-free and controlled environ-ment. All animal experiments conformed to the guidelines outlinedin the Guide for the Care and Use of Laboratory Animals by theLaboratory Animal Center, School of Life Sciences, University ofScience and Technology of China.

2.2. Experimental protocol

Three experimental protocols were followed. In all casesmale mice, in groups of 6e14, unless otherwise noted, werestudied and each experiment was repeated two to three times.Firstly, to address the influence of sorafenib on mice withCCl4-induced hepatocyte damage, both wild type (wt) andSTAT3Hep!/! mice were injected with CCl4 three times per weekfor 4 weeks, starting at 8 weeks of age. Twelve hours after thefirst CCl4 injection, daily treatment with sorafenib was startedand mice were followed as noted in Figs. 1 and 2 (wt mice) and 3

(STAT3Hep!/!) for liver function and fibrosis related geneexpression. Immunohistochemistry, immunofluorescence andimmunoblotting for STAT3 and pSTAT3 were also analyzed(Fig. 2). Secondly, to focus on the effect of sorafenib on estab-lished liver fibrosis, Fig. 4A, wt C57BL/6 mice were injected at 8weeks of age with CCl4. Two weeks later daily sorafenib treat-ment was initiated with continued CCl4 exposure (Fig. 4). At 14weeks of age, or 6 weeks following the initial exposure to CCl4,mice were sacrificed for histological assessment of liver fibrosis,immunoblot analysis of pSTAT3 and evaluation of fibrosis relatedgenes (Fig. 4). Finally, for more detailed kinetics of the influencesof sorafenib (Figs. 5 and 6), 8 weeks old wt C57BL/6, IL-6knockout C57BL/6 (Fig. 5A) and Kupffer cell depleted C57BL/6mice (Fig. 6C) were treated once with sorafenib. Twelve hourslater they were injected with one dose of CCl4. At 0, 6, 12, or 24 hpost CCl4 injection, mice were analyzed for levels of pSTAT3, IL-6and IL-22 (Figs. 5 and 6). In all groups, mice were injectedintraperitoneally (i.p.) with 2 ml/kg body weight of CCl4 dissolvedin olive oil (SigmaeAldrich, St. Louis, MO) (CCl4:olive oil ¼ 1:9)three times per week for 4 or 6 weeks, except for those receivinga single injection at the same dose. Sorafenib (Nexavar, BAY 43-9006, Germany) was dissolved in 5% carboxymethylcellulose so-dium in water and given intragastrically (i.g.) with 4 mg/kg bodyweight as noted in the protocol at the specific times. 5%carboxymethylcellulose sodium in water was given as the controlvehicle.

2.3. RNA preparation, reverse transcription and quantitative real-time PCR

Total RNA was extracted from liver of wt and STAT3Hep!/!

mice using Trizol (Invitrogen, Carlsbad, CA). M-MLV Transcrip-tase (Invitrogen) was used for reverse transcription. Quantitativereal-time PCR (Table 1) was performed using Premix Ex Taq(Takara, Japan). The expression levels of target genes (includingfibrosis-related genes and cytokine genes) were normalized tothe housekeeping gene Gapdh and the results were calculated byDDCt [30]. Known standards were included throughout allassays.

2.4. Analysis of serum transaminase activity

Serum alanine aminotransferase (ALT) activity was quantitated,including use of standardization [31].

2.5. Immunoblotting

Liver tissues were homogenized in extraction buffer (20 mMTriseHCl PH7.5, 150 mM NaCl, 1 mM Na3VO4, 1 mM EDTA, 1 mMEGTA, 1 mM phenylmethylsulfonylfluoride, 50 mM NaF, 1%Nonidet-P40) and centrifugated at 15,294 $g for 20 min to collectlysate. Protein concentrationwas quantified by a BCA protein assay.80 mg of total protein was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and then transferred toImmunobilon-P Transfer Membranes (Millipore, Billerica, Massa-chusetts), which were blocked with 5% nonfat milk dissolve inwashing buffer (20 mM TriseHCl, 150 mM NaCl, 0.1% (w/v) Tween-20), probed with appropriate primary antibodies against STAT3,pSTAT3 (Tyr705) (D3A7), p-c-Raf (Ser259), pMEK1/2 (Ser217/221)(41G9), p-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (D13.14.4E),GAPDH (1C4) or b-actin (13E5), and then detected by HRP-labeledanti-rabbit or HRP-labeled anti-mouse secondary antibodies (allfrom Cell Signaling Technology, Beverly, MA). The signal intensitywas analyzed by Image J software.

Y.-R. Deng et al. / Journal of Autoimmunity 46 (2013) 25e3426

2.6. Histological and immunohistological analysis of liver sections

Liver tissues were fixedwith 10% formalin, embedded in paraffinand cut into 4 mm sections for staining with hematoxylin-eosin(H&E). For quantitative assessment of fibrosis, sections werestained with Sirius Red for collagen content [32]. To define theactivation of hepatic stellate cells, levels of a-SMAwere assessed bystaining with rabbit anti-a-SMA antibody (clone 1A4; Dako,Glostrup, Denmark) and visualized with HRP labeled anti-rabbitantibody (ChemMate" EnVision" Detection Kit, Dako) [33].Quantification of liver fibrosis, inflammation and injury was

performed by a “blinded” liver pathologist, using the followingscales. 1) A 0-4 scale for fibrosis: 0, no fibrosis; 1, minimal portalfibrosis; 2, portal fibrosis with septa formation; 3, localizedbridging fibrosis; and 4, extensive bridging fibrosis. 2) A 0-3 scalefor inflammation: 0, no inflammation; 1, mild inflammation; 2,moderate inflammation; and 3, severe inflammation. 3) A 0-3 scalefor degeneration (based on intralobular degeneration): 0, nodegeneration; 1, mild degeneration (acidophilic bodies, ballooningdegeneration and/or scattered foci of hepatocellular necrosis in 1/3of lobules or nodules); 2, moderate degeneration (involvement of 1/3-2/3 of lobules or nodules); and 3, severe degeneration

Fig. 1. Sorafenib treatment attenuates CCl4-induced chronic liver injury and fibrosis. (A) Mice were injected with CCl4, 3 times per week, starting at 8 weeks of age. Twelve hourslater, daily treatment with sorafenib or control vehicle was begun. Four weeks later, mice were sacrificed 3 days after the last CCl4 injection. (B) Serum ALT levels in sorafenib treatedand untreated mice. (C) H&E stained liver from sorafenib treated or untreated mice 72 h after the last CCl4 injection. (D) Inflammation and degeneration scores from treated (n ¼ 14)and untreated (n ¼ 13) mice. (E) Quantitative PCR analysis of hepatic mRNA levels of fibrosis markers in sorafenib treated or untreated (n ¼ 12). Values were normalized to thehousekeeping gene GAPDH. (F) Sirius red staining (upper panels) and immunohistochemical staining (lower panels) for a-SMA for detection of activated hepatic stellate cells (HSCs).(G) Fibrosis scores of sorafenib treated (n ¼ 14) and untreated (n ¼ 14) groups. Fibrosis scores are based on the percentage of liver area positively stained by Sirius Red and anti-a-SMA. SORA, sorafenib; *P < 0.05; **P < 0.01; ***P < 0.001.

Y.-R. Deng et al. / Journal of Autoimmunity 46 (2013) 25e34 27

(involvement of %2/3 of lobules or nodules). Immunostaining forSTAT3 and pSTAT3 was performed using anti-STAT3 and anti-pSTAT3 (Tyr705) antibodies (Cell Signaling Technology) and visu-alized by ChemMate" EnVision" Detection Kit (Dako). Forimmunofluorescence, the frozen liver tissue was immersed inTissue-Teks OCT compound and 6 mm sections were stained withanti-STAT3 and incubated with Alexa Fluor 594-conjugated goatanti-rabbit IgG (Molecular Probes, Eugene, OR).

2.7. Kupffer cell depletion

Two days prior to sorafenib (or vehicle) treatment, C57BL/6 wtmice were treated with clodronate liposomes via tail vein injectionto deplete Kupffer cells [34]. Using a dose of 100 ul/per 10 g of body

weight diluted in saline [35]. Clodronate liposomes were kindlyprovided by Dr. N. van Rooijen (Vrije Universiteit, Amsterdam, TheNetherlands). Control groups were administered with the samedose of liposome-encapsulated PBS.

2.8. Isolation of murine Kupffer cells

Kupffer cells were isolated by liver perfusion [36,37]. Briefly,mice were anesthetized with sodium pentobarbital. Sequentially,the liver was perfused at 37 &C using a 0.5 mM EGTA solution for5 min at a flow rate 5 ml/min and followed by the original solutionand 0.075% collagenase IV (SigmaeAldrich), then centrifuged at50 g for 5 min to discard hepatocytes. The supernatant was trans-ferred into a new tube and centrifuged at 500 g for 10 min at 4 &C.The collected pellet was resuspended and separated by Percollgradient (20% and 50%) with centrifuging at 800 g for 15 min. Thelayer between the 20%e50% gradient interface was collected. Thecollected cells were plated in non-collagen coated culture plate(60 mm culture dish or 6 well plate) for 20 min. The cells werewashed to remove debris and unattached cells. The Kupffer cellswere then collected for the studies described above.

2.9. Statistical analysis

All data are presented as mean ' standard deviation (SD). Stu-dent’s t test was used for comparisons. All experiments werereplicated two or three times, with similar results. P < 0.05 wasconsidered statistically significant.

Table 1Real-time quantitative PCR primers used in this study.

Genes Forward (50e30) Reverse (50e30)

Acta2 AAGAGCATCCGACACTGCTGAC AGCACAGCCTGAATAGCCACATACCol1a1 CAGGGTATTGCTGGACAACGTG GGACCTTGTTTGCCAGGTTCAMmp2 GATAACCTGGATGCCGTCGTG CTTCACGCTCTTGAGACTTTGGTTCTimp1 TGAGCCCTGCTCAGCAAAGA GAGGACCTGATCCGTCCACAAIL-6 CCACTTCACAAGTCGGAGGCTTA CCAGTTTGGTAGCATCCATCATTTCIL-22 TTCCAGCAGCCATACATCGTC CTTCCAGGGTGAAGTTGAGCAGapdh AAATGGTGAAGGTCGGTGTGAAC CAACAATCTCCACTTTGCCACTGb-actin TTGCTGACAGGATGCAGAAG ACATCTGCTGGAAGGTGGAC

All the sequences of target genes were got from NCBI GenBank. The primes weredesigned and synthesized by Takara Biotechnology (Dalian) Corporation.

Fig. 2. Sorafenib treatment results in the nuclear translocation of STAT3 and enhanced phosphorylation of STAT3 in fibrotic liver. Liver sections and total liver lysates were collectedafter injection of CCl4 for 4 weeks with or without concurrent sorafenib (SORA) treatment. (A) STAT3 in total liver lysates. (B) Immunohistochemical staining for STAT3. (C)Immunofluorescent analysis of STAT3. (D) Phosphorylated STAT3 (pSTAT3) in total liver lysates. (E) Relative levels of pSTAT3, normalized to b-actin, in sorafenib treated and un-treated (n ¼ 14) mice.

Y.-R. Deng et al. / Journal of Autoimmunity 46 (2013) 25e3428

3. Results

3.1. Sorafenib attenuates CCl4-induced chronic liver injury and liverfibrosis

We first studied CCl4 induced chronic liver injury by injectingmice with CCl4 for 4 weeks and treating daily with sorafenib orcontrol vehicle (Fig. 1A). As shown in Fig. 1B, ALT levels weresignificantly lower in sorafenib-treated mice compared to controlsat 48 and 72 h (P < 0.001). Liver sections revealed less vacuolatedcells (Fig. 1C) and significantly improved portal inflammation andhepatocellular degeneration in the sorafenib-treated micecompared to controls (P < 0.001) (Fig. 1D). The levels of bothcollagen and a-SMA were significantly reduced in treated micecompared to controls (P < 0.001) (Fig. 1F and G). By quantitativePCR, the expression of Acta2, Col1a1, Mmp2, and Timp1 was signif-icantly decreased in sorafenib-treated mice (Acta2: P < 0.001;Col1a1: P < 0.001; Mmp2: P ¼ 0.005; Timp1: P < 0.001) (Fig. 1E).

3.2. Increased phosphorylation of STAT3 in hepatocytes of sorafenibtreated mice

We next examined the levels of STAT3 protein in sorafenibtreated or untreated C57BL/6 mice with or without injected CCl4.The total level of STAT3 protein was not significantly different be-tween sorafenib treated and untreated mice with or without CCl4injection (Fig. 2A). Immunohistochemical (Fig. 2B) and immuno-fluorescence (Fig. 2C) analysis further revealed that in micereceiving CCl4 injection, STAT3 was primarily located in cytoplasm.Importantly, however, in mice receiving sorafenib, most hepatic

STAT3 was located in the nucleus. To evaluate the effects of sor-afenib on activation of STAT3, we examined the phosphorylation ofSTAT3 at Tyr705. As shown in Fig. 2D and E, sorafenib treatmentmarkedly up-regulated the phosphorylation of STAT3 in the liver ofmice injected with CCl4 (P < 0.001).

3.3. The protective effect of sorafenib was abolished in STAT3Hep!/!

mice

We next induced liver fibrosis with CCl4 in hepatocyte-specificSTAT3 knockout (STAT3Hep!/!) mice and controls and treatedwith sorafenib. Without sorafenib treatment the fibrosis scoreswere higher in STAT3Hep!/! mice compared with control mice(P¼ 0.027) (Fig. 3A and B). While sorafenib protected C57BL/6 micefrom liver pathology, no significant difference was detected in theseverity of liver injury and fibrosis between treated and untreatedSTAT3Hep!/! mice (Fig. 3A and B). Twenty four hours after the lastCCl4 injection, the ALT was higher in STAT3Hep!/! than wild typemice, either treated (P ¼ 0.008) or untreated (P ¼ 0.004) withsorafenib. Although sorafenib treatment significantly inhibited ALTat 48 and 72 h in the wt B6 mice (48 h: P ¼ 0.001; 72 h: P < 0.001);this effect was not observed in STAT3Hep!/! mice (Fig. 3C).

After induction of liver fibrosis with CCl4, the mRNA levels of 3out of 4 fibrosis related genes testedwere significantly higher in theSTAT3Hep!/! mice than wild type mice (Acta2: P < 0.001, Col1a1:P ¼ 0.011 and Timp1: P ¼ 0.016) (Fig. 3D). Of note, while sorafenibtreatment significantly reduced the expression of these genes inwtB6 mice (Acta2: P < 0.001; Col1a1: P < 0.001; Mmp2: P < 0.001;Timp1: P < 0.001), such effects were not seen in STAT3Hep!/! mice(Fig. 3D).

Fig. 3. Protective effect of sorafenib against liver fibrosis was dependent on STAT3 in hepatocytes. STAT3Hep!/! mice and wt C57BL/6 (B6) mice injected with CCl4 for 4 weeks withor without concurrent sorafenib treatment. (A) H&E staining (upper panels) and Sirius red staining (lower panels). (B) Fibrosis scores from STAT3Hep!/! (sorafenib treated, n ¼ 6;untreated, n ¼ 6) or wt B6 mice (sorafenib treated, n ¼ 8; untreated, n ¼ 7) mice. (C) ALT levels after the last CCl4 injection. (D) Quantitative PCR analysis of hepatic mRNA levels offibrosis markers in STAT3Hep!/! (sorafenib treated, n ¼ 6; untreated, n ¼ 6) or wt B6 (sorafenib treated, n ¼ 6; untreated, n ¼ 6) mice. Values were normalized to the housekeepinggene GAPDH. SORA, sorafenib; ST3, STAT3; *P < 0.05; **P < 0.01; ***P < 0.001.

Y.-R. Deng et al. / Journal of Autoimmunity 46 (2013) 25e34 29

3.4. Effect of sorafenib treatment on established liver fibrosis

To investigate established disease, C57BL/6 wt mice wereinjected with CCl4 for 6 weeks to induce liver fibrosis. A 4-week

sorafenib treatment was started at week 3 of CCl4 injectionwhen liver fibrosis had been established (Fig. 4A). As shown inFig. 4, both the liver injury as well as fibrosis were significantlyreduced (P < 0.001) (Fig. 4B and C), while the expression of

Fig. 4. Effect of sorafenib treatment on established chronic liver injury and liver fibrosis induced by CCl4. (A) Wt C57BL/6 mice were injected with CCl4, 3 times per week. Twoweekslater daily sorafenib or control vehicle treatment was initiated while CCl4 injections were continued. At 14 weeks of age, or 6 weeks following the initial exposure to CCl4, mice weresacrificed 72 h after the last CCl4 injection. (B) H&E staining (upper panels) and Sirius red staining (lower panels). (C) Liver fibrosis scores of mice treated with CCl4 for 2 weeks andfor 6 weeks with or without 4-week sorafenib administration. (D) Relative levels of mRNA for fibrosis related genes. (E) Analysis of pSTAT3. (F) Relative levels of pSTAT3 in sorafenibtreated and untreated mice, normalized to b-actin. *P < 0.05; **P < 0.01; ***P < 0.001.

Y.-R. Deng et al. / Journal of Autoimmunity 46 (2013) 25e3430

fibrosis-related genes was significantly decreased (Acta2,P < 0.001; Col1a1, P < 0.001; Mmp2, P ¼ 0.04; Timp1, P ¼ 0.02)(Fig. 4D) in the sorafenib treated mice compared to untreatedcontrols. We further examined the phosphorylation of STAT3 inthe livers of sorafenib treated and untreated mice. Phosphoryla-tion of STAT3 was significantly up-regulated (P ¼ 0.002) (Fig. 4Eand F) in treated mice compared to untreated controls. These re-sults indicate that sorafenib treatment started at week 3 in micewith established liver fibrosis suppressed further development ofCCl4-induced fibrosis.

3.5. The kinetics of STAT3 activation in CCl4-induced acute liverinjury

To address the immune mechanism of sorafenib-mediatedprotection, we examined the phosphorylation of hepatic STAT3using an acute liver injury model in which the mice were injectedwith a single dose of CCl4 with or without pre-treatment with asingle dose of sorafenib administered at 12 h prior to the CCl4injection. Mice were sacrificed at serial time points for analysis(Fig. 5A). As shown in Fig. 5B, the phosphorylation of STAT3 wassuppressed at 6 h (P < 0.001), but up-regulated at 12 and 24 h(12 h, P ¼ 0.001; 24 h, P < 0.001) after CCl4 injection (Fig. 5B andC). Next, we investigated the effects of sorafenib on the upstreamkinases MEK and Erk of the STAT3 signaling pathway. As shown inFig. 5D, sorafenib pretreatment inhibited MEK and Erk phos-phorylation at 6 h and 12 h, respectively (MEK, P < 0.001; Erk,P ¼ 0.04).

3.6. Sorafenib-mediated up-regulation of STAT3 phosphorylation isdependent on Kupffer-derived IL-6

We examined the levels of IL-6 and IL-22 mRNA in sorafenibtreated or untreated mice in our acute liver injury model. Asshown in Fig. 5E, the expression of both IL-6 and IL-22 genes wassuppressed by sorafenib at 6 h (IL-6, P ¼ 0.045; IL-22, P ¼ 0.01)after CCl4 injection. However, at 24 h after CCl4 injection bothcytokines were upregulated (IL-6, P ¼ 0.03; IL-22, P ¼ 0.01).Notably, the increase of STAT3 phosphorylation was associatedwith elevated hepatic IL-6 mRNA expression (Fig. 5F). To furtherassess the role of IL-6 in pSTAT3 up-regulation, this experimentwas repeated in IL-6!/! mice. While sorafenib pre-treatmentresulted in elevated STAT3 phosphorylation at 24 h after CCl4injection in the wt B6 mice (P ¼ 0.004), such effect was notobserved in IL-6!/! mice (Fig. 6A and B). To identify the source ofsorafenib-induced IL-6, we employed the same acute liver injurymodel in normal and macrophage-depleted C57BL/6 mice(Fig. 6C). After depletion of macrophages, the up-regulation ofSTAT3 phosphorylation disappeared 24 h after CCl4 injection inmice pretreated with sorafenib (P ¼ 0.002) (Fig. 6D and E).Moreover, IL-6 mRNA expression was significantly down-regulated in Kupffer cell-depleted livers of sorafenib treatedmice at 24 h after CCl4 injection (P < 0.001) (Fig. 6F), which iscoincidental with suppression of STAT3 phosphorylation. In orderto confirm the source of sorafenib-induced IL-6, we isolatedKupffer cells from sorafenib treated or untreated mice 24 h afterCCl4 injection. Results from Q-PCR demonstrate that Kupffer cellsfrom sorafenib treated mice expressed higher level of IL-6compared with untreated controls (P ¼ 0.0217) (Fig. 6G). Sor-afenib elevated IL-6 expression in Kupffer cells and up-regulatedSTAT3 phosphorylation in hepatocytes. Therefore, sorafenib at-tenuates liver fibrosis by modulating the cross talk betweenKupffer cells and hepatocytes.

Fig. 5. Kinetics of STAT3 signaling pathway activation in CCl4 induced acute liverinjury. (A) Wt C57BL/6 mice were treated once with sorafenib or control vehicle at 8weeks of age. Twelve hours later, mice were injected i.p. with one dose of CCl4. Micewere sacrificed at 0, 6, 12, or 24 h after CCl4 injection. (B) Immunoblot analysis forpSTAT3 in sorafenib treated and untreated mice. (C) Relative levels of pSTAT3 in treatedand untreated mice (n ¼ 6), normalized to b-actin. (D) Immunoblot analysis for p-c-Raf, p-MEK, p-Erk in treated and untreated mice at serial time points after CCl4 in-jection. (E) Quantitative PCR analysis of hepatic mRNA levels of IL-6 and IL-22 (sor-afenib treated, n ¼ 6; untreated, n ¼ 6) at 6 h and 24 h after CCl4. Values werenormalized to GAPDH. (F) The relative mRNA levels for IL-6 and IL-22 genes(normalized to GAPDH) in the livers of sorafenib treated mice. *P < 0.05; **P < 0.01;***P < 0.001.

Y.-R. Deng et al. / Journal of Autoimmunity 46 (2013) 25e34 31

Fig. 6. The up-regulation of STAT3 phosphorylation induced by sorafenib is dependent on Kupffer-derived IL-6. (A) Immunoblot analysis of pSTAT3 in livers of wt B6 and IL-6!/!

mice pretreated or untreated with sorafenib at 24 h following CCl4 injection. (B) The relative levels of pSTAT3 in wt B6 and IL-6!/! mice treated or untreated with sorafenib (n ¼ 3 or4 in each group). (C) Strategy used for analyzing effects of sorafenib treatment on macrophage-depleted wild type C57BL/6 mice. Wt C57BL/6 mice were treated with clodronateliposomes to deplete macrophages at 8 weeks of age. Macrophage-depleted mice were then treated once with sorafenib or control vehicle at 48 h following the depletion. Twelvehours after the administration of sorafenib, they were injected with one dose of CCl4. Mice were sacrificed at 24 h after CCl4 injection. (D) Immunoblot analysis for pSTAT3 in livers ofcontrols and macrophage-depleted C57BL/6 mice pretreated or untreated with sorafenib 24 h following CCl4 injection. (E) The relative levels of pSTAT3 in normal controls (n ¼ 5)and macrophage-depleted (n ¼ 7) mice. (F) The relative mRNA levels of IL-6 gene (normalized to b-actin) in liver. (G) Wt C57BL/6 mice were treated once with sorafenib or controlvehicle. Twelve hours after the administration of sorafenib, they were injected with one dose of CCl4. 24 h after CCl4 injection Kupffer cells were isolated and IL-6 gene expressionwas detected at mRNA level by Q-PCR (normalized to b-actin). *P < 0.05; **P < 0.01; ***P < 0.001.

Y.-R. Deng et al. / Journal of Autoimmunity 46 (2013) 25e3432

4. Discussion

Liver fibrosis results from chronic damage to the liver inconjunction with the accumulation of ECM proteins, which is acharacteristic of chronic liver diseases [1,3]. Although experimentaldata has suggested targets that prevent fibrosis in rodents [8], theefficacy of most treatments has not been recapitulated in humans[3]. Sorafenib, the first oral multikinase inhibitor that blocks mul-tiple signaling pathways, was initially approved as an oral agent fortreating advanced renal cell carcinoma [38]. Recently, sorafenib hasbeen approved for the treatment of HCC [14,39e42]. Several studiesreport that sorafenib attenuates the hypertensive syndrome inpartial portal vein ligated rats [24,25]. There are also data thatsuggest that sorafenib has antifibrotic activities. In vitro sorafenibcan inhibit activation, growth and collagen accumulation of HSC[26,27]. Moreover, sorafenib influences paracrine signaling be-tween HSCs and LECs and thereby regulates matrix and vascularchanges associated with bile duct ligation-induced liver injury inrats [28]. In hepatocytes, in vitro experiments demonstrated thatsorafenib suppressed TGF-b1-induced EMT [29].

A complex interplay amongst hepatic cell types takes placeduring hepatic fibrogenesis [9]. Damaged hepatocytes release ROSand fibrogenic mediators and induce an inflammatory recruitment.Apoptosis of damaged hepatocytes stimulates the fibrogenic ac-tions of liver myofibroblasts [11]. Inflammatory cells, either lym-phocytes or polymorphonuclear cells, activate HSCs to secretecollagen [43]. Activated HSCs secrete inflammatory chemokines,express cell adhesion molecules, and modulate the activation oflymphocytes [44]. Therefore, a vicious circle in which fibrogenicand target cells stimulate each other is likely to occur. Our datasuggested that sorafenib can protect hepatocytes during thisfibrogenesis and break the circle to stop the fibrotic process. Wedemonstrate that sorafenib treatment significantly attenuatedCCl4-induced chronic liver injury and liver fibrosis, associated withand dependent on the up-regulated activation of hepatic STAT3.Furthermore, we demonstrate that sorafenib mediated pStat3 up-regulation is dependent on Kupffer cell derived IL-6.

Hepatic STAT3 appears to have a protective role against liverfibrosis [21e23]. In non-stimulated hepatocytes, STAT3 is inactiveand locates in the cytoplasm. The binding of STAT3-related cyto-kines to their receptors activates the receptor-associated Janustyrosine kinases (JAK), which phosphorylates STAT3. The resultantpSTAT3 dimers translocate to the nuclei and function as tran-scriptional factors for downstream genes including bcl-2, bcl-xL,mcl-1, flip, ref-1, cyclinD1, and c-myc that play important roles inpromoting hepatocyte survival and liver regeneration, as well asameliorating steatosis [20e23,45]. In our current study we usedSTAT3Hep!/! mice to demonstrate that the protection effect ofsorafenib against fibrosis is dependent on the upregulation ofpSTAT3 (Fig. 3).

Previous studies reported that sorafenib inhibited STAT3 phos-phorylation in a variety of tumors including medulloblastoma [18],cholangiocarcinoma [46] and HCC [17]. Moreover, in TGF-b1-mediated mouse hepatocyte EMT, sorafenib also inhibits TGF-b1-induced STAT3 phosphorylation [29]. These in vitro data areconsistent with data that sorafenib can inhibit STAT3 phosphory-lation in liver 6 h after CCl4 induction. However, at 12 and 24 h afterCCl4 injection, sorafenib treatment enhanced pSTAT3 levels(Fig. 5B).

To explain the effects of sorafenib on activation of STAT3, weexamined the kinetics of hepatic expression of the STAT3-relatedcytokines IL-6, IL-22, IL-10 and HGF (data not shown). We notethat only the expression of IL-6 correlates with phosphorylation ofSTAT3 (Fig. 5F). We further report that sorafenib-induced phos-phorylation of STAT3 vanishes in IL-6!/! mice (Fig. 6A). These data

indicate an important role of IL-6 in sorafenib-mediated up-regu-lation of pSTAT3, as well as protection against CCl4-induced fibrosis.Of note, since IL-6 is also a target gene of STAT3, the activation ofSTAT3 could in turn promote the transcription of IL-6. This positivefeedback is likely to play a role in the STAT3-mediated therapeuticeffect of sorafenib in our model. Since Kupffer cells are the majorsource of IL-6 [20], we investigated the effect of sorafenib onKupffer cells. Fig. 6G demonstrates that sorafenib can increase IL-6gene expression of isolated Kupffer cells from CCl4 -stimulatedmicewith or without sorafenib treatment. A recent study in Hepatologyreports that compared to untreated controls sorafenib induces IL-6(7.5-fold) transcription in cultured MV after LPS stimulation [47].Based on these data we believed that as a multi-kinase inhibitorsorafenib could inhibit STAT3 phosphorylation at 6 h after CCl4injection (Fig. 5B), while the later activation of STAT3 at 12 and 24 h(Fig. 5B) was due to increased IL-6 expression by Kupffer cells(Figs. 5F and 6).

Although in the model herein, we demonstrate an anti-fibroticactivity of sorafenib even in established cirrhosis, such was notthe case in the TAA fibrosis model in rats [27]. Future questions,with these comments in mind, need to be addressed, including whydecreased hepatocyte death results in less liver fibrosis. In addition,sorafenib is known to inhibit PDGF; PDGF activates hepatic stellatecells to myofibroblasts. Hence, the effects of sorafenib on stellatecell activation should be studied, including the related issue of whythe IL-6/STAT3 axis is increased by sorafenib. Finally, we suggestthat further studies should focus on the influence of sorafenib onother specific lineages to further define the mechanisms of atten-uation of liver injury and provide data on other pathways for po-tential therapeutic intervention.

5. Conclusions

We found that sorafenib significantly attenuates CCl4-inducedliver fibrosis and injury, with activated STAT3 playing a critical rolein this therapeutic effect. And further study has proved that Kupffercell-derived cytokine IL-6 is required for the sorafenib-mediatedphosphorylation of STAT3. These results suggest that sorafenib isa potential therapeutic agent for liver fibrosis.

Conflict of interest

The authors have declared that no conflict of interest exists.

Acknowledgments

Financial support provided by the National Basic Research Pro-gram of China (973 Program-2010CB945300, 2013CB944900), theNational Natural Science Foundation of China (30972738, 31021061,81130058), and the Hundred Talents Program of the ChineseAcademy of Sciences.

References

[1] Friedman SL. Liver fibrosisefrom bench to bedside. J Hepatol 2003;38:S38e53.[2] Shigeki TA, Parsons CJ, Rippe RA. Mechanisms of liver fibrosis. Clin Chim Acta

2006;364:33e60.[3] Bataller R, Brenner DA. Liver fibrosis. J Clin Invest 2005;115:209e18.[4] Mann DA, Smart DE. Transcriptional regulation of hepatic stellate cell acti-

vation. Gut 2002;50:891e6.[5] Liu CH, Gaca MDA, Swenson ES, Vellucci VF, Reiss M, Wells RG. Smads 2 and 3

are differentially activated by transforming growth factor-beta (TGF-beta) inquiescent and activated hepatic stellate cellseconstitutive nuclear localizationof Smads in activated cells is TGF-beta-independent. J Biol Chem 2003;278:11721e8.

[6] Parsons CJ, Takashima M, Rippe RA. Molecular mechanisms of hepatic fibro-genesis. J Gastroenterol Hepatol 2007;22:S79e84.

Y.-R. Deng et al. / Journal of Autoimmunity 46 (2013) 25e34 33

[7] Elsharkawy AM, Oakley F, Mann DA. The role and regulation of hepatic stellatecell apoptosis in reversal of liver fibrosis. Apoptosis 2005;10:927e39.

[8] Bataller R, Brenner DA. Hepatic stellate cells as a target for the treatment ofliver fibrosis. Semin Liver Dis 2001;21:437e51.

[9] Kmiec Z. Cooperation of liver cells in health and disease. Adv Anat EmbryolCell Biol 2001;161(IIIeXIII):1e151.

[10] Higuchi H, Gores GJ. Mechanisms of liver injury: an overview. Curr Mol Med2003;3:483e90.

[11] Canbay A, Friedman S, Gores GJ. Apoptosis: the nexus of liver injury andfibrosis. Hepatology 2004;39:273e8.

[12] Villanueva A, Llovet JM. Targeted therapies for hepatocellular carcinoma.Gastroenterology 2011;140:1410e26.

[13] Iavarone M, Cabibbo G, Piscaglia F, Zavaglia C, Grieco A, Villa E, et al. Field-practice study of sorafenib therapy for hepatocellular carcinoma: a prospec-tive multicenter study in Italy. Hepatology 2011;54:2055e63.

[14] Llovet JM, Ricci S, Mazzaferro V, Hilgard P, Gane E, Blanc JF, et al. Sorafenib inadvanced hepatocellular carcinoma. N Engl J Med 2008;359:378. 2508.

[15] Wilhelm S, Carter C, Lynch M, Lowinger T, Dumas J, Smith RA, et al. Discoveryand development of sorafenib: a multikinase inhibitor for treating cancer. NatRev Drug Discov 2006;5:835e44.

[16] Yu H, Pardoll D, Jove R. STATs in cancer inflammation and immunity: a leadingrole for STAT3. Nat Rev Cancer 2009;9:798e809.

[17] Chen KF, Tai WT, Liu TH, Huang HP, Lin YC, Shiau CW, et al. Sorafenib over-comes TRAIL resistance of hepatocellular carcinoma cells through the inhi-bition of STAT3. Clin Cancer Res 2010;16:5189e99.

[18] Yang F, Van Meter TE, Buettner R, Hedvat M, Liang W, Kowolik CM, et al.Sorafenib inhibits signal transducer and activator of transcription 3 signalingassociated with growth arrest and apoptosis of medulloblastomas. Mol CancerTher 2008;7:3519e26.

[19] Chai H, Luo AZ, Weerasinghe P, Brown RE. Sorafenib downregulates ERK/Aktand STAT3 survival pathways and induces apoptosis in a human neuroblas-toma cell line. Int J Clin Exp Pathol 2010;3:408e15.

[20] Wang H, Lafdil F, Kong X, Gao B. Signal transducer and activator of tran-scription 3 in liver diseases: a novel therapeutic target. Int J Biol Sci 2011;7:536e50.

[21] Plum W, Tschaharganeh DF, Kroy DC, Corsten E, Erschfeld S, Dierssen U, et al.Lack of glycoprotein 130/signal transducer and activator of transcription 3-mediated signaling in hepatocytes enhances chronic liver injury and fibrosisprogression in a model of sclerosing cholangitis. Am J Pathol 2010;176:2236e46.

[22] Mair M, Zollner G, Schneller D, Musteanu M, Fickert P, Gumhold J, et al. Signaltransducer and activator of transcription 3 protects from liver injury andfibrosis in a mouse model of sclerosing cholangitis. Gastroenterology2010;138:2499e508.

[23] Horiguchi N, Lafdil F, Miller AM, Park O, Wang H, Rajesh M, et al. Dissociationbetween liver inflammation and hepatocellular damage induced by carbontetrachloride in myeloid cell-specific signal transducer and activator of tran-scription 3 gene knockout mice. Hepatology 2010;51:1724e34.

[24] Reiberger T, Angermayr B, Schwabl P, Rohr-Udilova N, Mitterhauser M,Gangl A, et al. Sorafenib attenuates the portal hypertensive syndrome inpartial portal vein ligated rats. J Hepatol 2009;51:865e73.

[25] Mejias M, Garcia-Pras E, Tiani C, Miquel R, Bosch J, Fernandez M. Beneficialeffects of sorafenib on splanchnic, intrahepatic, and portocollateral circula-tions in portal hypertensive and cirrhotic rats. Hepatology 2009;49:1245e56.

[26] Wang Y, Gao J, Zhang D, Zhang J, Ma J, Jiang H. New insights into the anti-fibrotic effects of sorafenib on hepatic stellate cells and liver fibrosis. J Hepatol2010;53:132e44.

[27] Hong F, Chou H, Fiel MI, Friedman SL. Antifibrotic activity of sorafenib inexperimental hepatic fibrosis: refinement of inhibitory targets, dosing, andwindow of efficacy in vivo. Dig Dis Sci 2013;58:257e64.

[28] Thabut D, Routray C, Lomberk G, Shergill U, Glaser K, Huebert R, et al.Complementary vascular and matrix regulatory pathways underlie thebeneficial mechanism of action of sorafenib in liver fibrosis. Hepatology2011;54:573e85.

[29] Chen YL, Lv J, Ye XL, Sun MY, Xu Q, Liu CH, et al. Sorafenib inhibits trans-forming growth factor beta1-mediated epithelial-mesenchymal transitionand apoptosis in mouse hepatocytes. Hepatology 2011;53:1708e18.

[30] Zhou R, Wei H, Sun R, Tian Z. Recognition of double-stranded RNA by TLR3induces severe small intestinal injury in mice. J Immunol 2007;178:4548e56.

[31] Schumann G, Bonora R, Ceriotti F, Ferard G, Ferrero CA, Franck PF, et al. IFCCprimary reference procedures for the measurement of catalytic activity con-centrations of enzymes at 37 degrees C. International Federation of ClinicalChemistry and Laboratory Medicine. Part 4. Reference procedure for themeasurement of catalytic concentration of alanine aminotransferase. ClinChem Lab Med 2002;40:718e24.

[32] Safadi R, Ohta M, Alvarez CE, Fiel MI, Bansal M, Mehal WZ, et al. Immunestimulation of hepatic fibrogenesis by CD8 cells and attenuation by transgenicinterleukin-10 from hepatocytes. Gastroenterology 2004;127:870e82.

[33] Jiao J, Sastre D, Fiel MI, Lee UE, Ghiassi-Nejad Z, Ginhoux F, et al. Dendritic cellregulation of carbon tetrachloride-induced murine liver fibrosis regression.Hepatology 2012;55:244e55.

[34] Van Rooijen N, Sanders A. Kupffer cell depletion by liposome-delivered drugs:comparative activity of intracellular clodronate, propamidine, and ethyl-enediaminetetraacetic acid. Hepatology 1996;23:1239e43.

[35] Luo YP, Zhou H, Krueger J, Kaplan C, Liao D, Markowitz D, et al. The role ofproto-oncogene Fra-1 in remodeling the tumor microenvironment in supportof breast tumor cell invasion and progression. Oncogene 2010;29:662e73.

[36] Horiguchi N, Wang L, Mukhopadhyay P, Park O, Jeong WI, Lafdil F, et al. Celltype-dependent pro- and anti-inflammatory role of signal transducer andactivator of transcription 3 in alcoholic liver injury. Gastroenterology2008;134:1148e58.

[37] Vrochides D, Papanikolaou V, Pertoft H, Antoniades AA, Heldin P. Biosynthesisand degradation of hyaluronan by nonparenchymal liver cells during liverregeneration. Hepatology 1996;23:1650e5.

[38] Escudier B, Eisen T, Stadler WM, Szczylik C, Oudard S, Siebels M, et al. Sor-afenib in advanced clear-cell renal-cell carcinoma. N Engl J Med 2007;356:125e34.

[39] Zhu AX, Clark JW. Commentary: sorafenib use in patients with advancedhepatocellular carcinoma and underlying child-pugh b cirrhosis-evidence andcontroversy. Oncologist 2009;14:67e9.

[40] Pinter M, Sieghart W, Graziadei I, Vogel W, Maieron A, Konigsberg R, et al.Sorafenib in unresectable hepatocellular carcinoma from mild to advancedstage liver cirrhosis. Oncologist 2009;14:70e6.

[41] Yau T, Chan P, Ng KK, Chok SH, Cheung TT, Fan ST, et al. Phase 2 open-labelstudy of single-agent sorafenib in treating advanced hepatocellular carci-noma in a hepatitis B-endemic Asian population: presence of lung metastasispredicts poor response. Cancer 2009;115:428e36.

[42] Cheng AL, Kang YK, Chen Z, Tsao CJ, Qin S, Kim JS, et al. Efficacy and safety ofsorafenib in patients in the Asia-Pacific region with advanced hepatocellularcarcinoma: a phase III randomised, double-blind, placebo-controlled trial.Lancet Oncol 2009;10:25e34.

[43] Casini A, Ceni E, Salzano R, Biondi P, Parola M, Galli A, et al. Neutrophil-derived superoxide anion induces lipid peroxidation and stimulates collagensynthesis in human hepatic stellate cells: role of nitric oxide. Hepatology1997;25:361e7.

[44] Vinas O, Bataller R, Sancho-Bru P, Gines P, Berenguer C, Enrich C, et al. Humanhepatic stellate cells show features of antigen-presenting cells and stimulatelymphocyte proliferation. Hepatology 2003;38:919e29.

[45] Kishimoto T. IL-6: from its discovery to clinical applications. Int Immunol2010;22:347e52.

[46] Blechacz BR, Smoot RL, Bronk SF, Werneburg NW, Sirica AE, Gores GJ. Sor-afenib inhibits signal transducer and activator of transcription-3 signaling incholangiocarcinoma cells by activating the phosphatase shatterproof 2. Hep-atology 2009;50:1861e70.

[47] Sprinzl MF, Reisinger F, Puschnik A, Ringelhan M, Ackermann K, Hartmann D,et al. Sorafenib perpetuates cellular anticancer effector functions by modu-lating the crosstalk between macrophages and natural killer cells. Hepatology2013;57:2358e68.

Y.-R. Deng et al. / Journal of Autoimmunity 46 (2013) 25e3434