Protease-Activated Receptor 2 Promotes ExperimentalLiver Fibrosis in Mice and Activates Human Hepatic

Stellate CellsVirginia Knight,1,2 Jorge Tchongue,1,2 Dinushka Lourensz,1,2 Peter Tipping,1 and William Sievert1,2

Protease-activated receptor (PAR) 2 is a G-protein–coupled receptor that is activated afterproteolytic cleavage by serine proteases, including mast cell tryptase and activated coagula-tion factors. PAR-2 activation augments inflammatory and profibrotic pathways throughthe induction of genes encoding proinflammatory cytokines and extracellular matrix pro-teins. Thus, PAR-2 represents an important interface linking coagulation and inflamma-tion. PAR-2 is widely expressed in cells of the gastrointestinal tract, including hepaticstellate cells (HSCs), endothelial cells, and hepatic macrophages; however, its role in liverfibrosis has not been previously examined. We studied the development of CCl4-inducedliver fibrosis in PAR-2 knockout mice, and showed that PAR-2 deficiency reduced the pro-gression of liver fibrosis, hepatic collagen gene expression, and hydroxyproline content.Reduced fibrosis was associated with decreased transforming growth factor beta (TGFb)gene and protein expression and decreased matrix metalloproteinase 2 and tissue inhibitorof matrix metalloproteinase 1 gene expression. In addition, PAR-2 stimulated activation,proliferation, collagen production, and TGFb protein production by human stellate cells,indicating that hepatic PAR-2 activation increases profibrogenic cytokines and collagenproduction both in vivo and in vitro. Conclusion: Our findings demonstrate the capacityof PAR-2 activation to augment TGFb production and promote hepatic fibrosis in miceand to induce a profibrogenic phenotype in human HSCs. PAR-2 antagonists have recentlybeen developed and may represent a novel therapeutic approach in preventing fibrosis inpatients with chronic liver disease. (HEPATOLOGY 2012;55:879-887)

Hepatic fibrosis occurs in response to acute andchronic liver injury from a variety of sourcesand may progress to end-stage liver disease

with the development of portal hypertension, hepato-cellular carcinoma, and liver failure. A substantialbody of evidence has identified the hepatic stellate cell(HSC) as the principal source of collagen producedduring hepatic fibrogenesis,1 and thus there is consid-erable interest in factors that regulate HSC activationand collagen expression.Protease-activated receptors (PARs) are a unique

group of G-protein–coupled receptors activated byproteolytic cleavage of their extracellular N terminaldomain to reveal a ‘‘tethered’’ ligand that binds withthe second extracellular loop of the receptor to initiatesignaling. PAR-1 was initially identified in the searchfor the cellular thrombin receptor, and, to date, fourPARs have been identified. Thrombin activates PAR-1,3, and 4, and factor Xa activates PAR-1 and 2. PAR-2is also activated by trypsin, mast cell tryptase, and thetissue factor/factor VIIa and factor Xa complex.2 Thereis a strong linkage between inflammation, coagulation,and fibrosis,3 and a prothrombotic state appears toaccelerate liver fibrogenesis.4 One proposed mechanism

Abbreviations: CD, cluster of differentiation; cDNA, complementary DNA;ECM, extracellular matrix; ELISA, enzyme-linked immunosorbent assay; GI,gastrointestinal; HSC, hepatic stellate cell; IgG, immunoglobulin G; KO,knockout; MMP, matrix metalloproteinase; mRNA, messenger RNA; PAR,protease activated receptor; PDGF, platelet-derived growth factor; RT-PCR,reverse-transcriptase polymerase chain reaction; aSMA, alpha smooth muscleactin; TGFb, transforming growth factor b; TIMP, tissue inhibitor of matrixmetalloproteinase; WT, wild type.From the 1Center for Inflammatory Diseases, Monash University and

2Gastroenterology and Hepatology Unit, Monash Medical Center, Melbourne,Victoria, Australia.Received March 4, 2011; accepted October 11, 2011.This work was supported by grants from the National Health and Medical

Research Council of Australia.Address reprint requests to: William Sievert, M.D., Gastroenterology and

Hepatology Unit, Monash Medical Center, 246 Clayton Road, Melbourne,Victoria 3168, Australia fax: 613 9594 6250. E-mail: William.Sievert@monash.eduCopyrightVC 2011 by the American Association for the Study of Liver Diseases.View this article online at wileyonlinelibrary.com.DOI 10.1002/hep.24784Potential conflict of interest: Nothing to report.

879

for this linkage is signaling by coagulation factorsthrough their cellular receptors, PARs, to activatestellate cells.4,5

PAR-2 is widely expressed in the gastrointestinal(GI) tract on epithelial cells and smooth muscle cells.6

It has been shown to have important, multifacetedroles in the regulation of GI physiology and in inflam-matory processes, including pancreatitis, gastritis, andcolitis. In the healthy liver, PAR-2 is expressed onhepatocytes, Kupffer cells, bile duct epithelial cells,and endothelial cells of large vessels. Rat HSC expressPAR-2 under normal conditions and its expression ismarkedly increased in liver fibrosis.5 Mast cells areprominently recruited during hepatic fibrosis7 andhave the potential to provide a potent source of mastcell tryptase, which can activate PAR-2 receptors.PAR-2 activation augments inflammatory cell recruit-ment and profibrotic pathways through the inductionof genes encoding proinflammatory cytokines and pro-teins of the extracellular matrix (ECM). PAR-2activation has been shown to promote pulmonary8 andrenal9 fibrosis with increased expression in progressiveliver injury,10 but the contribution of PAR-2 to liverfibrosis has not been reported.We hypothesized that PAR-2 activation promotes

hepatic fibrosis in mice and induces HSC proliferationand collagen synthesis. In this study, we show thatdeletion of PAR-2 diminishes CCl4-induced hepaticfibrosis and that PAR-2 agonists promote HSC proli-feration and collagen production.

Materials and Methods

Animals. PAR-2�/� (PAR-2 knockout; KO) mice,derived on a mixed 129/SvJ and C57BL/6 back-ground, were obtained from Dr. Shaun Coughlin(University of California, San Francisco, CA) andback-crossed 10 generations onto a C57BL/6 back-ground. Their genotype was confirmed by reverse-tran-scriptase polymerase chain reaction (RT-PCR). Micewere allowed food and water ad libitum and werehoused at a constant temperature in a 12-hour lightand dark cycle. Experimental protocols were approvedby the Monash University Animal Ethics Committee,and mice received humane care as specified under theAustralian Code of Practice for the Care and Use ofAnimals for Scientific Purposes.CCl4-Induced Hepatic Fibrosis. Liver fibrosis was

induced in male mice by twice-weekly intraperitonealinjections of 1 lL/g body weight of CCl4 mixed witholive oil (1:10), starting between 8 and 10 weeks of

age and continuing for 5-8 weeks. Six groups of micewere studied: Two groups received CCl4 for 5 weeks(PAR-2�/�, n ¼ 6; wild-type [WT] C57BL/6, n ¼9), and two groups received CCl4 for 8 weeks (PAR-2�/�, n ¼ 8; WT, n ¼ 10). Two control groups ofWT C57BL/6 mice (n ¼ 8 each) received olive oilalone for 5 and 8 weeks. Mice were killed 72 hours af-ter the last dose of CCl4, and blood and tissue werecollected for analysis.Fibrosis Assessment. Liver tissue was fixed in 2%

paraformaldehyde for histological examination. Four-micron-thick sections from paraffin-embedded liver tis-sue were deparaffinized and stained with picrosiriusred (Sirius red F3BA 0.1% [w/v] in saturated picricacid) for 90 minutes, washed in acetic acid and water(5:1,000), dehydrated in ethanol, and mounted inneutral DPX. Fifteen consecutive nonoverlapping fieldswere acquired for each mouse liver, the image wasdigitized, and fibrosis area was analyzed by ScionImage for Windows (vAlpha 4.0.3.2; Scion Corpora-tion, Frederick, MD).Determination of Hepatic Hydroxyproline Con-

tent. Hepatic hydroxyproline content was quantifiedusing liver tissue frozen in liquid nitrogen, as previ-ously described, with minor modification.11 Briefly,liver samples were weighed and hydrolyzed in 2.5 mLof 6 N of HCl at 110�C for 18 hours in Teflon-coatedtubes. The hydrolysate was centrifuged at 3,000 rpmfor 10 minutes; the pH of the resulting supernatantwas adjusted to 7.4, and absorbance was measured at558 nm. Total hydroxyproline content was measuredagainst a standard curve prepared with trans-4-hydroxy-L-proline (Sigma-Aldrich, St. Louis, MO)preparations in the range of 0.156-5.0 lg/mL andexpressed per milligram of wet tissue weight.RNA Purification, Reverse Transcription, and

Real-Time Quantitative PCR. Mouse liver RNA waspurified from snap-frozen tissue using the QiagenRNeasy mini kit, according to the manufacturer’sinstructions (Qiagen Pty Ltd., Hilden, Germany).RNA from cultured cell lines was isolated using TRI-zol (Invitrogen, Carlsbad, CA), as previouslydescribed.12 RNA concentration was measured with aNanodrop ND-100 spectrophotometer (Thermo Sci-entific, Waltham, MA), and complementary DNA(cDNA) was generated using the High-CapacitycDNA Reverse Transcription Kit (Applied Biosystems,Foster City, CA), as per the manufacturer’s instruc-tions. Real-time PCR analysis was performed (FastStartSybr Green; Roche, Mannheim, Germany) using aRotor Gene 3000 light cycler (Qiagen Pty Ltd., Syd-ney, Australia), and the specific target messenger RNA

880 KNIGHT ET AL. HEPATOLOGY, March 2012

(mRNA) of interest was quantified as a ratio relativeto 18S RNA content of the sample. The followingmouse primers were used: MMP-2 forward: ACCCAG ATG TGG CCA ACT AC, reverse: TCA TTTTAA GGC CCG AGC AA; TIMP-1 forward: ACGAGA CCA CCT TAT ACC AGC CG, reverse: GCGGTT CTG GGA CTT GTG GGC (from Dr. ScottFreidman, Mt. Sinai School of Medicine, New York,NY); 18S forward: GTA ACC CGT TGA ACC CCATTC, reverse: GCC TCA CTA AAC CAT CCA ATCG (from Dr. Eric Morand, Monash University, Mel-bourne, Victoria, Australia); TGFb forward: TGCCCT CTA CAA CCA ACA CA, reverse: GTT GGACAA CTG CTC CAC CT (Primer 3 software); PAR-1forward: CTC CTC AAG GAG CAG ACC CAC;reverse: AGA CCG TGG AAA CGA TCA AC (Primer3 software); and PAR-2 and 18S primers from AppliedBiosystems TaqMan probe (Mm00433160_m1,Hs03003631_g1) using TaqMan Gene ExpressionMaster Mix (Applied Biosystems).Immunohistochemistry. Paraformaldehyde-fixed 4-

micron-thick liver tissue sections were stained with pri-mary antibody for alpha smooth muscle actin (aSMA)(monoclonal mouse antimouse a-SMA; Sigma-Aldrich), F4/80 (rat antimouse, 1:200; a gift of Dr.Richard Kitching, Monash University, Clayton, Victo-ria, Australia) and cluster of differentiation (CD)68(rat antimouse CD68, FA11, 1:100; a gift of Dr. G.Koch, Cambridge, UK). The following secondary anti-bodies were used: aSMA biotinylated rabbit antimouseimmunoglobulin G (IgG)2a antibody (1:300; Invitro-gen, Carlsbad, CA), F4/80 and CD68 polyclonalrabbit antirat IgG (1:150; Dako, Carpinteria, CA). Inbrief, sections were dewaxed, rehydrated, and thenblocked with 0.6% hydrogen peroxide and CAS pro-tein blocking solution (Invitrogen). Primary antibodyincubations for 30 minutes at room temperature(aSMA) and overnight at 4�C (F4/80, CD68) werefollowed by the application of secondary antibody.Staining was amplified using an avidin-biotin complexkit (Vector Laboratories, Burlingame, CA) and wasdetected with diaminobenzidine (Dako). Slides werecounterstained with Harris hematoxylin. For quantita-tion of immunoreactivity, 15 consecutive nonoverlap-ping fields at 250� magnification (a-SMA, F4/80,and CD68) were scored using a graticule eyepiece in ablinded fashion. Negative controls consisted of amouse IgG1 isotype control antibody (Dako,Glostrup, Denmark) and water substituting for theprimary antibody.Hepatic TGFb1 Content. Extracts were prepared

from snap-frozen liver by homogenization in lysis

buffer (Tris-HCl 50 mM, NaCl 150 mM, ethylenedia-minetetraacetic acid 1 mM, 1% Triton X-100, 0.5%Tween-20, and 0.1% sodium dodecyl sulphate), con-taining a protease-inhibitor cocktail (Roche), followedby centrifugation at 14,000�g for 15 minutes at 4�C.Supernatants were collected and activated with aceticacid/urea before analysis. Transforming growth factorb (TGFb1) content of liver protein extracts weremeasured using a mouse TGFb1 enzyme-linkedimmunosorbent assay (ELISA) kit (R&D Systems,Inc., Minneapolis, MN). Plates were read using theBio-Rad (Hercules, CA) microplate reader at 450 nm(with a 540-nm reference filter), and TGFb1 concen-trations were calculated from the standard curve by theplate-reader software.PAR-1- and PAR-2-Stimulated HSC Collagen and

TGFb1 Production In Vitro. Immortalized humanHSCs (LX-2 cells; a gift from Dr. Scott Friedman)were seeded for 3 days into six-well plates at a densityof 1 � 105 cells per well in M199 medium (Gibco,Grand Island, NY) with 5% fetal calf serum. Mediawere changed at day 3, and human PAR-1 agonisthexapeptide SFLLRN-NH2 (Sigma-Aldrich) and/orhuman PAR-2 agonist hexapeptide SLIGKV (Sigma-Aldrich) were added at varying concentrations. Ascrambled hexapeptide (Auspep, Melbourne, Victoria,Australia) was used as a control. A further dose of ei-ther agonist or scrambled peptide was added at 24 and48 hours, and culture medium and cells were har-vested after 72 hours of peptide exposure. The colla-gen content of the cell-culture supernatant was meas-ured using the Sircol Sirius red dye colorimetric assay(Biocolor, Newtown Abbey, Northern Ireland), as pre-viously described,11 and TGFb1 content was measuredby ELISA.HSC Proliferation in Response to PAR Activa-

tion. LX-2 cells were seeded onto 96-well plates at adensity of 1 � 104 per well in 5% FCS/M199 mediaand cultured overnight. The PAR-2 agonist peptide,SLIGKV, was added at concentrations from 0 to 100lM at 24 and 48 hours. Human platelet-derivedgrowth factor (PDGF)-BB (R&D Systems, Minneapo-lis, MN) was used as a positive control at a concentra-tion of 25 ng/mL. Proliferation of activated HSCs wasassessed using a colorimetric bromodeoxyuridineELISA (Roche), according to the manufacturer’sinstructions.Statistical Analysis. Data are expressed as mean 6

standard error of the mean. Statistical significance wasdetermined by one-way analysis of variance with theNewman-Keuls post-test for multiple comparisons orthe Student’s t test for comparisons between two

HEPATOLOGY, Vol. 55, No. 3, 2012 KNIGHT ET AL. 881

groups, as appropriate, using GraphPad Prism 5.03 forWindows (GraphPad Software, Inc., La Jolla, CA).

Results

PAR-2 Deficiency Prevents Progression of Histo-logical Hepatic Fibrosis. WT mice developed signifi-cant hepatic collagen deposition in response to CCl4administration (Fig. 1A). No fibrosis was observed inWTmice given olive oil alone (data not shown). Quanti-tative analysis of histological fibrosis by computer-assisted morphometry in CCl4-treated WTmice showedmarked fibrosis at 5 weeks (1.97% 6 0.16% liver area),which progressed with continued CCl4 exposure over 8weeks (3.39% 6 0.26%) (Fig. 1C). In PAR-2 KO mice,CCl4 administration induced similar fibrosis to that ofWTmice at 5 weeks (2.07% 6 0.26%). However, therewas no progression of liver fibrosis with continued CCl4exposure between 5 and 8 weeks in the PAR-2 KO mice(2.09% 6 0.28%). At 8 weeks, there was significantlyless hepatic fibrosis in the PAR-2 KO, compared to WT,mice (P ¼ 0.004) (Fig. 1B,C).Hepatic Procollagen mRNA and Hydroxyproline

Content Are Reduced in PAR-2 KO Mice Exposed toCCl4. Histological assessment of fibrosis correlated

closely with other indices of hepatic collagen contentin mice given CCl4. At 8 weeks, PAR-2 KO miceshowed significantly less induction of procollagenmRNA (1.8- 6 0.23-fold above untreated mice), com-pared with WT mice (5.9- 6 1.08-fold; P ¼ 0.002)(Fig. 1D). After 5 weeks of CCl4 administration, simi-lar increases in hepatic hydroxyproline were observedin WT and PAR-2 KO mice (0.45 6 0.02 lg/mgand 0.43 6 0.009 lg/mg, respectively) (Fig. 1E).However, after 8 weeks, whereas hepatic hydroxypro-line content increased significantly in WT mice,there was no increase in PAR-2 KO mice,compared to levels at 5 weeks. PAR-2 KO mice (0.426 0.026) had significantly less hepatic hydroxyproline,compared to WTmice (0.63 6 0.03) at 8 weeks (P <0.002).PAR-2 Deficiency Is Associated With Reduced

Stellate Cell Activation. aSMA is a marker of HSCactivation and myofibroblast differentiation. In WTmice, hepatic fibrosis induced by the administration ofCCl4 was accompanied by a progressive increase inaSMA expression at 8 weeks, compared to untreatedmice. In PAR-2 KO mice receiving CCl4, induction ofaSMA was similar to WT mice treated with CCl4 at 5weeks (Fig. 2A), but did not increase further, resulting

Fig. 1. Hepatic collagen deposition in WT mice (A) and PAR-2 KO mice (B) administered CCl4 for 8 weeks (Sirius red staining,400�).Computer-assisted morphometry of Sirius-red–stained liver sections shows no difference in hepatic fibrosis area between WT and KO miceat 5 weeks; however, in contrast to WT mice, there was no progression in fibrosis area in PAR-2 KO mice by 8 weeks (C). Procollagen mRNAexpression (D) and hepatic collagen content (E) were significantly lower in PAR-2 KO mice, compared to WT mice, at 8 weeks.

882 KNIGHT ET AL. HEPATOLOGY, March 2012

in significantly less aSMA expression, compared toWTmice at 8 weeks (P ¼ 0.014).PAR-2 Deficiency Reduces Hepatic TGFb Expres-

sion and Decreases Matrix Metalloproteinase/TissueInhibitor of Metalloproteinase mRNA. CCl4-inducedhepatic fibrosis was associated with up-regulation ofTGFb mRNA (3.44- 6 0.72-fold greater than con-trol) and protein (9.2 6 0.9 pg/mg liver, control 6.96 0.19 pg/mg) in WTmice at 8 weeks. In PAR-2 KOmice, TGFb mRNA up-regulation was significantlyreduced (1.38- 6 0.23-fold of control; P ¼ 0.016,

compared to WT) (Fig. 2B), as was TGFb protein,which was similar to control levels (Fig. 2C).Matrix metalloproteinases (MMPs) and their specific

tissue inhibitors, tissue inhibitors of metalloproteinase(TIMPs), regulate ECM composition and their expres-sion is altered in response to liver injury. In WT micetreated with CCl4 for 8 weeks, both MMP-2 andTIMP-1 mRNA increased, consistent with activeECM remodeling during the development of hepaticfibrosis (Fig. 3A,B). Both MMP-2 and TIMP-1mRNA expression were significantly reduced in PAR-2KO mice, compared to WT mice, suggesting thatECM remodeling is reduced in association with thearrest in progression of fibrosis between 5 and 8 weeksin PAR-2 KO mice.PAR-1 mRNA Is Up-regulated at Week 5, but Not

at Week 8, in PAR-2 KO Mice. The temporal patternof PAR-1 mRNA expression was examined to investi-gate the potential mechanisms for the lack of earlyprotection against hepatic fibrosis observed in PAR-2KO mice. In PAR-2 knockout mice at 5 weeks, PAR-1mRNA expression was significantly up-regulated, com-pared to CCl4-treated WT mice or untreated controls(Fig. 4A). However, at 8 weeks, PAR-1 expression inthe PAR-2 KO mice was not significantly different

Fig. 2. After CCl4 administration, the number of cells expressingaSMA was significantly lower in PAR-2 KO mice at 8 weeks, comparedto WT mice (A). TGFb mRNA expression was significantly lower in PAR-2 KO mice, compared to WT controls, at 8 weeks (B). TGFb proteinlevels were lower, remaining at control levels, in PAR-2 KO mice, com-pared to WT mice (C).

Fig. 3. Expression of MMP-2 mRNA (A) and TIMP-1 mRNA (B) wassignificantly lower in PAR-2 KO mice, compared to WT mice, at week 8of CCl4 administration.

HEPATOLOGY, Vol. 55, No. 3, 2012 KNIGHT ET AL. 883

from WT controls (Fig. 4B). Thus, up-regulation ofPAR-1 mRNA may compensate for lack of PAR-2 inthe early stages of CCl4-induced fibrogenesis, but thiscompensatory mechanism is not maintained as fibrosisprogresses, resulting in significantly less fibrosis inPAR-2 KOs at 8 weeks.Decrease in Activated Hepatic Macrophages in

Week 8 PAR-2 KO Mice. We also examined the natureof the inflammatory infiltrate at weeks 5 and 8 to investi-gate the difference in hepatic fibrosis between PAR-2KO mice and WTmice observed at week 8. Significantlyfewer F4/80þ macrophages were observed at both 5 and8 weeks in PAR-2 KO mice, compared to CCl4-treatedWT mice (Fig. 5A). In addition, at week 8, there weresignificantly fewer CD68þ macrophages in PAR-2 KOmice, compared to CCl4-treated WT mice, which is adifference that was not observed at week 5 (Fig. 5B).These observations are consistent with a role for PAR-2in the recruitment, and later activation of, macrophagesin CCl4-induced hepatic fibrosis.PAR-2 Activation Stimulates HSC Prolifera-

tion. To study the effect of PAR-2 activation directly

and specifically in HSCs, we used an immortalizedhuman stellate cell line (LX-2), which has been previ-ously well characterised. Subconfluent cultures of LX-2cells were stimulated with a specific PAR-2 agonist pep-tide (SLIGKV) for 48 hours or a scrambled hexapeptidecontrol. The PAR-2 agonist peptide stimulated dose-de-pendent proliferation of LX-2 cells (Fig. 6A). At themaximum dose of 100 lM, the PAR-2 agonist peptidecaused proliferation equivalent to PDGF (25 ng/mL),the most potent inducer of HSC proliferation.PAR-2 Activation Increases HSC Collagen Produc-

tion. HSCs spontaneously produce collagen duringculture on plastic tissue-culture plates. PAR-2 agonistpeptide (100 lM) stimulated a significant increase incollagen production by LX-2 cells, whereas the controlhexapeptide failed to stimulate collagen production(Fig. 6B). Similarly, PAR-1 agonist peptide (100 lM)stimulated a significant increase in collagen

Fig. 4. PAR-1 mRNA expression was significantly greater in PAR-2knockout mice, compared to untreated controls and CCl4-treated WTmice, at 5 weeks (A). However, at 8 weeks, PAR-1 expression wassimilar in the PAR-2 KO mice, compared to WT controls (B).

Fig. 5. There were significantly fewer F4/80þ macrophages at both5 and 8 weeks in PAR-2 KO mice, compared to CCl4-treated WT mice(A). At week 8, but not week 5, there were significantly fewer CD68þ

macrophages in PAR-2 KO mice, compared to CCl4-treated WT mice(B).

884 KNIGHT ET AL. HEPATOLOGY, March 2012

production. The combination of PAR-1 and PAR-2agonist peptide significantly increased collagen pro-duction, compared to control peptide and untreatedcontrols, but not more than the individual agonistsalone.

PAR-2 Activation Stimulates TGFb Production. TGFbis spontaneously produced by HSCs in culture. PAR-2agonist peptide (at 3 different doses) caused a signifi-cant increase in TGFb production by LX-2 cells, com-pared to the control peptide and untreated controls(Fig. 6C). The threshold for the stimulation of TGFbproduction (25 lM) was lower than that for stimula-tion of collagen production. As expected, TGFb pro-duction also increased after stimulation with PAR-1agonist peptide. The combination of PAR-1 and PAR-2 agonist peptides caused a significant increase inTGFb production by LX-2 cells, compared to controlpeptide and untreated controls. Again, the effect ofcombined agonist peptides on TGFb production wasnot significantly greater than the effect of the individ-ual agonists, suggesting a maximal response at theselected doses.

Discussion

We observed that PAR-2 deficiency in experimentalliver fibrosis leads to a reduction in hepatic collagencontent and histological fibrosis accompanied bydecreased HSC activation, as demonstrated by thereduced expression of aSMA. These findings were par-alleled by a decrease in gene and protein expression ofthe principal profibrogenic cytokine, TGFb, andaltered MMP and TIMP gene expression. We con-firmed a specific effect on HSC in vitro by showingthat PAR-2 activation stimulated proliferation, collagenproduction, and TGFb protein production. These datasuggest that PAR-2 activation promotes hepatic fibrosisby inducing a profibrogenic phenotype in HSCs.PAR-1 has been studied in animal models of hepatic

necroinflammation and fatty liver disease10 and inhuman and murine lung injury.13 PAR-1-deficientmice appear to be protected from CCl4-induced liverfibrosis.14 Thus, there is compelling evidence thatthrombin/Xa-induced PAR-1 signaling plays an impor-tant role in tissue fibrogenesis.4,5 Interest in the role ofPAR-2 in hepatic fibrosis has developed based on evi-dence that PAR-2 activation is associated with inflam-matory and fibrogenic events in the kidney and pan-creas9,15 and its expression is increased in models oflung injury,8,16 suggesting an important role for PAR-2 in mediating tissue repair. Cellular mechanismsunderlying this role have been proposed by Borensz-tajn et al., who showed that Factor Xa signaling viaPAR-2 induced fibroblast proliferation, migration, anddifferentiation into myofibroblasts.17

The role of PAR-2 in hepatic inflammation and fi-brosis has been examined, to date, only in HSC

Fig. 6. Stimulation of human LX-2 cells with a PAR-2 agonist pep-tide significantly increased cell proliferation to levels equivalent toPDGF (**values compared to 0 lM) (A). HSC collagen production wassignificantly increased by both PAR-2 and PAR-1 agonist peptide (100lM) and the combination of the two; there was no effect from ascrambled peptide used as control (*values compared to untreatedcontrols) (B). Both PAR-2 and PAR-1 agonist peptides significantlyincreased TGFb protein production by HSC; there was no effect from ascrambled peptide used as control (all values compared to untreatedcontrol). The combination of PAR-1 and PAR-2 peptides significantlyincreased TGFb production by HSC, compared to control peptide anduntreated controls, but not to PAR-1 or PAR-2 alone (C).

HEPATOLOGY, Vol. 55, No. 3, 2012 KNIGHT ET AL. 885

derived from experimental animals. Gaca et al. demon-strated PAR-2 expression in rat HSC, and showed thatPAR-2 agonists induced HSC proliferation and colla-gen production.18 Fiorucci et al. similarly showed thatPAR-2 agonist stimulation of rat HSCs resulted inproliferation and activation.10 To our knowledge, thecurrent study is the first to explore the role of thePAR-2 receptor in liver fibrosis in vivo in PAR-2knockout mice and in vitro in human HSCs. The useof the KO model is a particular strength of the studythat allows us to ascribe a profibrogenic role to PAR-2unequivocally, because antagonist studies can be trou-bled by a lack of molecular specificity. These findingssignificantly expand the evidence linking PAR-2 liga-tion with hepatic fibrogenesis that occurs most likelythrough a direct effect on HSC proliferation and colla-gen production.We confirmed the role of PAR-2 in HSC activation

through studies using the human HSC line, LX-2,which expresses PAR-2. We observed a significant doseresponse to a specific PAR-2 agonist that achieved aproliferative response comparable to PDGF, the mostpotent cytokine in regard to stimulating HSC prolifer-ation. In keeping with the overall effect that we saw inthe PAR-2 KO mice, we showed that PAR-2 ligationof human HSCs led to increased TGFb and collagenproduction. PAR-2 appears to have equivalent effectsto PAR-1 in regard to the HSC expression of TGFband collagen. We did not demonstrate an additiveeffect on these responses with the combination of ago-nists, suggesting maximal stimulation of the commondownstream effector pathways of these two receptorsunder the agonist doses and conditions of in vitrostimulation.Interestingly, we observed that the protective effect

of PAR-2 gene deletion was apparent during moreadvanced stages of fibrosis, in this case at 8 weeks ofCCl4 exposure, rather than at 5 weeks. This raises thequestion of the nature of the factor(s) leading to PAR-2 activation during continued hepatic injury. PAR-2activation can be stimulated by trypsin and mast celltryptase as well as coagulation proteases, such as factorVIIa, Xa, and tissue factor. Mast cells are recruited tothe liver during fibrogenesis and their numbers canincrease by up to 9-fold in the cirrhotic liver.7 Tryptaseaccounts for approximately 25% of mast cell protein,and its levels progressively increase with liver injury.19

Thus, we postulated that PAR-2 activation in theinjured liver might occur through tryptase generation,given the interval between injury and mast cell accu-mulation. However, we did not observe any differencein histological staining or gene expression of mast cell

chymase, a marker for mast cells, between mice treatedfor 5 or 8 weeks (data not shown). We then investi-gated PAR-1 expression in the PAR-2 KO mice andfound significant up-regulation of PAR-1 mRNA inPAR-2 KO mice at 5 weeks, which was not observedin WT mice exposed to CCl4 or the vehicle control.Interestingly, at 8 weeks, PAR-1 up-regulation was notevident in the KO mice. Thus, there appears to becompensatory PAR-1 signaling early in fibrogenesis inthe PAR-2 KO mice that is lost as fibrosis progresses,which may account for the difference in hepatic fibro-sis observed at 8 weeks that was not evident at 5weeks.Macrophages play an important role in hepatic

fibrogenesis,20 and therefore, we also examined theextent of macrophage infiltration at 5 and 8 weeks.We found that the number of F4/80þ macrophages inPAR-2 KO mice was lower than that in WT mice atboth 5 and 8 weeks; however, the number of activatedmacrophages (CD68þ cells) was significantly lower inthe KO mice, compared to WT controls, at 8 weeks.A recent study has shown that PAR-2 and Toll-like re-ceptor 4, which is highly expressed on Kupffer cellsand forms a component of the lipopolysaccharide re-ceptor, cooperate to enhance the release of proinflam-matory cytokines.21 Fewer activated macrophagesobserved at 8 weeks in the PAR-2 KO mice maytherefore lead to alterations in the inflammatory he-patic microenvironment that could contribute to thedecrease in hepatic fibrosis observed in PAR-2 defi-ciency. Thus, multiple lines of evidence suggest that itis likely that as inflammatory liver disease progresses,increasing expression of PAR-210,18 and its ligands,such as factor Xa,17 potentiate HSC activation andcollagen deposition.Changes in the expression of MMPs and their spe-

cific tissue inhibitors, TIMPs, are complex and varyover time with liver injury. MMP-2 is an autocrineproliferation and migration factor for HSC,22 whoseexpression can be induced by TGFb and is typicallyincreased after liver injury. TIMP-1, which inhibitsMMP activity, is produced by activated HSCs and itsexpression is also up-regulated with liver injury andleads to the net accumulation of ECM.23 The decreasein gene expression of MMP-2 and TIMP-1 in PAR-2-deficient mice reflects a relatively static phenotypeassociated with the failure of either fibrosis progressionor regression that we observed in these animals.Recently, peptide-mimetic compounds have been for-

mulated that bind to PAR-2 and inhibit intracellularresponses, including nuclear factor kappa light-chainenhancer of activated B cell activation and interleukin-8

886 KNIGHT ET AL. HEPATOLOGY, March 2012

expression as well as PAR-2-induced tissue responses,such as vascular (rat aorta) relaxation.24 These newlydeveloped, specific PAR-2 antagonists may represent anovel therapeutic approach in preventing fibrosis inpatients with chronic liver disease and support the needfor further research into these unique receptors.In conclusion, we have demonstrated that deletion

of the PAR-2 gene in mice chronically exposed toCCl4 leads to a significant reduction in hepatic fibro-sis. The mechanism of this effect is likely to bethrough a reduction in HSC proliferation and collagenproduction. These novel findings suggest that PAR-2may be an important therapeutic target for the treat-ment of human hepatic fibrosis.

References1. Friedman S. Mechanisms of hepatic fibrosis. Gastroenterology 2008;

134:1655-1669.2. Riewald M, Ruf W. Orchestration of coagulation protease signaling by

tissue factor. Trends Cardiovasc Med 2002;12:149-154.3. Tacke F, Luedde T, Trautwein C. Inflammatory pathways in liver injury

and liver homeostasis. Clin Rev Allergy Immunol 2009;36:4-12.4. Anstee Q, Wright M, Goldin R, Thursz M. Parenchymal extinction:

coagulation and hepatic fibrogenesis. Clin Liver Dis 2009;13:117-126.5. Borensztajn K, von der Thusen JH, Peppelenbosch MP, Spek CA. The

coagulation factor Xa/protease activated receptor-2 axis in the progres-sion of liver fibrosis: a multifaceted paradigm. J Cell Mol Med 2010;14:143-153.

6. D’Andrea M, Derian C, Leturcq D, Baker S, Brunmark A, Ling P,et al. Characterization of protease-activated receptor 2 immunoreactiv-ity in normal human tissue. J Histochem Cytochem 1998;46:157-164.

7. Farrell DJ, Hines JE, Walls AF, Kelly PJ, Bennett MK, Burt AD. Intra-hepatic mast cells in chronic liver diseases. HEPATOLOGY 1995;22:1175-1181.

8. Borensztajn K, Bresser P, van der Loos C, Bot I, van den Blink B, den BakkerMA, et al. Protease-activated receptor-2 induces myofibroblast differentiationand tissue factor up-regulation during bleomycin-Induced lung injury. Poten-tial role in pulmonary fibrosis. Am J Pathol 2010;177:2753-2764.

9. Xiong J, Zhu Z, Liu J, Wang Y, Li Z. Role of protease activated recep-tor 2 expression in renal interstitial fibrosis model in mice. J HuazhongUniv Sci Technolog Med Sci 2005;25:523-526.

10. Fiorucci S, Antonelli E, Distrutti E, Severino B, Fiorentina R, BaldoniM, et al. PAR1 antagonism protects against experimental liver fibrosis.

Role of proteinase receptors in stellate cell activation. HEPATOLOGY

2004;39:365-375.11. Patella S, Phillips DJ, Tchongue J, deKretser DM, Sievert W. Follista-

tin attenuates early liver fibrosis: effects on hepatic stellate cell activa-tion and hepatocyte apoptosis. Am J Physiol Gastrointest Liver Physiol2006;290:G137-G144.

12. Moussa L, Apostolopoulos J, Davenport P, Tchongue J, Tipping PG.Protease-activated receptor-2 augments experimental crescentic glomer-ulonephritis. Am J Pathol 2007;171:800-808.

13. Scotton C, Krupiczojc, Konigshoff M, Mercer P, Lee YC, Kaminski N,et al. Increased local expression of coagulation factor X contributes tothe fibrotic response in human and murine lung injury. J Clin Invest2009;119:2550-2563.

14. Rullier A, Duplantier J, Costet P, Cubel G, Haurie V, Petibois C, et al. Prote-ase-activated receptor 1 knockout reduces experimentally induced liver fibro-sis. Am J Physiol Gastrointest Liver Physiol 2009;294:G226-G235.

15. Masamune A, Kikuta K, Satoh M, Suzuki N, Shimosegawa T. Proteaseactivated receptor 2-mediated proliferation and collagen production ofrat pancreatic stellate cells. J Pharmacol Exp Ther 2005;312:651-658.

16. Cederqvist K, Haglund C, Heikkila P, Hollenberg MD, Karikoski R,Andersson S. High expression of pulmonary proteinase-activated recep-tor 2 in acute and chronic lung injury in preterm infants. Pediatr Res2005;57:831-836.

17. Borensztajn K, Stiekema J, Nijmeijer S, Reitsma P, Peppelenbosch M, Spek A.Factor Xa stimulates proinflammatory and profibrotic responses in fibroblastsvia protease activated receptor 2 activation. Am J Pathol 2008;172:309-320.

18. Gaca, MD, Zhou X, Benyon RC. Regulation of hepatic stellate cellproliferation and collagen synthesis by proteinase-activated receptors. JHepatol 2002;36:362-369.

19. Schwartz LB, Irani AM, Roller K, Castells MC, Schechter NM. Quan-titation of histamine, tryptase, and chymase in dispersed human T andTC mast cells. J Immunol 1987;138:2611-2615.

20. Marra F, Aleffi S, Galastri S, Provenzano A. Monocytes in liver fibrosis.Semin Immunopathol 2009;31:345-358.

21. Rallabhandi P, Nhu QM, Toshchakov VY, Piao W, Medvedev AE, Hol-lenberg MD, et al. Analysis of proteinase-activated receptor 2 andTLR4 signal transduction, a novel paradigm for receptor cooperativity.J Biol Chem 2008;283:24314-24325.

22. Ikeda K, Wakahara T, Wang YQ, Kadoya H, Kawada N, Kaneda K. Invitro migratory potential of rat quiescent hepatic stellate cells and itsaugmentation by cell activation. HEPATOLOGY 1999;29:1760-1767.

23. Hemmann S, Graf J, Roderfeld M, Roeb E. Expression of MMPs andTIMPs in liver fibrosis—a systematic review with special emphasis onanti-fibrotic strategies. J Hepatol 2007;46:955-975.

24. Kanke T, Kabeya M, Kubo S, Kondo S, Yasuoka K, Tagashira J, et al.Novel antagonists for proteinase-activated receptor 2: inhibition of cel-lular and vascular responses in vitro and in vivo. Br J Pharmacol 2009;158:361-371.

HEPATOLOGY, Vol. 55, No. 3, 2012 KNIGHT ET AL. 887