Toxicology in Vitro 35 (2016) 77–85

Contents lists available at ScienceDirect

Toxicology in Vitro

j ourna l homepage: www.e lsev ie r .com/ locate / tox inv i t

Human precision-cut liver slices as a model to test antifibrotic drugs inthe early onset of liver fibrosis

Inge M. Westra a,b, Henricus A.M. Mutsaers b, Theerut Luangmonkong b, Mackenzie Hadi a,Dorenda Oosterhuis b, Koert P. de Jong c, Geny M.M. Groothuis a, Peter Olinga b,⁎a Division of Pharmacokinetics, Toxicology and Targeting, Department of Pharmacy, University of Groningen, The Netherlandsb Division of Pharmaceutical Technology and Biopharmacy, Department of Pharmacy, University of Groningen, The Netherlandsc Department of Hepato-Pancreato-Biliary Surgery and Liver Transplantation, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands

⁎ Corresponding author at: Division of PharmaceuticalDepartment of Pharmacy, University of Groningen, AntGroningen, The Netherlands.

E-mail address: p.olinga@rug.nl (P. Olinga).

http://dx.doi.org/10.1016/j.tiv.2016.05.0120887-2333/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 2 February 2016Received in revised form 13 May 2016Accepted 24 May 2016Available online 26 May 2016

Liver fibrosis is the progressive accumulation of connective tissue ultimately resulting in loss of organ function.Currently, no effective antifibrotics are available due to a lack of reliable human models. Here we investigatedthe fibrotic process in human precision-cut liver slices (PCLS) and studied the efficacy of multiple putativeantifibrotic compounds.Our results demonstrated that human PCLS remained viable for 48 h and the early onset of fibrosis was observedduring culture, as demonstrated by an increased gene expression of Heat Shock Protein 47 (HSP47) and Pro-Col-lagen 1A1 (PCOL1A1) as well as increased collagen 1 protein levels. SB203580, a specific inhibitor of p38 mito-gen-activated protein kinase (MAPK) showed a marked decrease in HSP47 and PCOL1A1 gene expression,whereas specific inhibitors of Smad 3 and Rac-1 showed no or only minor effects. Regarding the studiedantifibrotics, gene levels of HSP47 and PCOL1A1 could be down-regulated with sunitinib and valproic acid,while PCOL1A1 expression was reduced following treatment with rosmarinic acid, tetrandrine and pirfenidone.These results are in contrastwith prior data obtained in rat PCLS, indicating that antifibrotic drug efficacy is clear-ly species-specific.Thus, human PCLS is a promising model for liver fibrosis. Moreover, MAPK signaling plays an important role inthe onset of fibrosis in this model and transforming growth factor beta pathway inhibitors appear to be more ef-fective than platelet-derived growth factor pathway inhibitors in halting fibrogenesis in PCLS.

© 2016 Elsevier Ltd. All rights reserved.

Keywords:Human precision-cut liver slicesLiver fibrosisAntifibroticTGF-β

1. Introduction

Fibrosis is an integral part of the pathophysiological mechanism of adiverse range of chronic diseases, such as Crohn's disease, chronic kid-ney disease and viral hepatitis. The fibrotic process is characterized byaugmented production and excessive deposition of extracellular matrixproteins resulting in scar formation and the progressive loss of organfunction. Liver cirrhosis, the end stage of liver fibrosis, is possibly themost clinically relevant form of tissue fibrosis in the world due to thehigh prevalence of viral hepatitis (Zeisberg and Kalluri, 2013). Conse-quently, liver fibrosis is widely studied and there is extensive knowl-edge regarding the process of fibrogenesis (Pellicoro et al., 2014). Andthroughout the years a plethora of potential therapeutic targets havebeen described (Schuppan and Kim, 2013). Nevertheless, an effective

Technology and Biopharmacy,onius Deusinglaan 1, 9713 AV

therapy for liver fibrosis remains elusive, and transplantation remainsthe sole successful treatment modality. Moreover, antifibrotic drug dis-covery is hampered by the lack of reliable and reproducible (human) invitro models. Recently, precision-cut tissue slices have been used as amodel for incipient and established fibrosis (Westra et al., 2013;Stribos et al., 2015). Of note, this model replicates most of the multicel-lular characteristics of organs and the different cells are retained in theiroriginal environment. Previously, rat and murine precision-cut liverslices (PCLS) have been successfully used to test the antifibrotic efficacyof several putative antifibrotic drugs (van de Bovenkamp et al., 2006a;Westra et al., 2014a, 2014b; Iswandana et al., 2016). Here we reportthe use of human PCLS to test antifibrotic compounds.

There are severalwell-known common signaling pathways involvedin the fibrotic process in all organs, including the archetypicaltransforming growth factor beta (TGF-β) and platelet-derived growthfactor (PDGF) pathways as well as the p38 mitogen-activated proteinkinase (MAPK) pathway (Zeisberg and Kalluri, 2013; Bonner, 2004;Parsons et al., 2007). Even though these pathways are generally in-volved in fibrogenesis, there remain tissue-, species- and strain-specific

Table 2Experimental treatment.

Inhibitor Compound Concentration(μM)

Cmax (μM), reference

PDGF Imatinib 1 –10 5, (Leveque and Maloisel, 2005)Sorafenib 0.5 – 2 9, (Strumberg et al., 2007)Sunitinib 0.5 – 5 14, (Minkin et al., 2008)

a

78 I.M. Westra et al. / Toxicology in Vitro 35 (2016) 77–85

differences (Zeisberg and Kalluri, 2013; Liu et al., 2013; Inoue et al.,2015). In general, TGF-β signaling is associated with an increased depo-sition of collagen, whereas PDGF is a potent mitogen affecting cells ofmesenchymal origin, such as myofibroblasts (Bonner, 2004). Bothgrowth factors activate a myriad of transcription factors by binding totheir respective receptors (e.g. type 1 TGF-β receptor or receptor tyro-sine kinase; (Liu, 2011)). TGF-β acts mainly via Smad signaling, whichcan be targeted by the specific inhibitor of Smad3 (Sis3), known to in-hibit TGF-β-induced Smad3 phosphorylation (Jinnin et al., 2006).PDGF stimulates cell proliferation via a host of downstream intracellularsignaling cascades involving for instance glycogen synthase kinase 3β orRho GTPases (e.g. Rac1). The latter can be specifically inhibited byNSC23766 (Gao et al., 2004). MAPK is involved in the regulation of col-lagen 1A1 gene expression and mRNA stability, and can be activated byboth PDGF and TGF-β (Parsons et al., 2007). SB203580 is a known inhib-itor of MAPK (Cuenda et al., 1995). Thus, next to studying the efficacy ofantifibrotic compounds, the current study was also designed to identifythe pathways underlying the (early) fibrotic response in healthy humanPCLS using these pathway specific inhibitors.

2. Methods

2.1. Ethics statement

This study was approved by the Medical Ethical Committee of theUniversityMedical CentreGroningen (UMCG), according toDutch legis-lation and the Code of Conduct for dealing responsibly with human tis-sue in the context of health research (www.federa.org), refraining theneed of written consent for ‘further use’ of coded-anonymous humantissue. The procedures were carried out in accordance with the experi-mental protocols approved by the Medical Ethical Committee of theUMCG.

2.2. Chemicals

All chemicals were obtained from Sigma Aldrich (Zwijndrecht, TheNetherlands) unless stated otherwise. Stock solutions were preparedin either milli-Q or dimethyl sulfoxide (DMSO) and stored at −20 °C.During experiments, stocks were diluted in culture medium with afinal solvent concentration of ≤1%.

2.3. Human liver tissue

Healthy human liver tissue was obtained either from patients fol-lowing partial hepatectomy due to metastatic colorectal cancer (PH-livers) or from donors, remaining as surgical surplus after reduced-size liver transplantation (TX), as described previously (Elferink et al.,2011). Clinical characteristics of study subjects who provided liver tis-sue are listed in Table 1. Of note, donor variability has limited impacton the fibrotic response in ourmodel (van de Bovenkamp et al., 2006b).

2.4. Precision-cut liver slice preparation and experimental treatment

All liver tissuewas perfusedwith cold University ofWisconsin (UW)organ preservation solution (DuPont Critical Care, Waukegab, IL, USA)at the time of collection and stored in ice-cold UW solution until use

Table 1Characteristics of liver donors.

Number 27

PH/TX 10/17Female/Male 20/7Age (years) 51 (10–82)

Values are shown as mean (range).PH: tissue obtained from partial hepatectomy.TX: tissue obtained from transplantation liver.

(de Graaf et al., 2010). Liver slices were prepared in ice-cold Krebs–Henseleit buffer supplemented with 25 mM D-glucose (Merck, Darm-stadt, Germany), 25 mMNaHCO3 (Merck), 10 mMHEPES (MP Biomed-icals, Aurora, OH, USA) and saturated with carbogen (95% O2, 5% CO2)using a Krumdieck tissue slicer as previously described (de Graaf et al.,2010). In addition, slices were kept on ice-cold UW solution before cul-ture, duringwhich time viability andmetabolic activity wasmaintainedas described before (Olinga et al., 1998). PCLS—diameter: 5 mm, thick-ness: 250 μm—were incubated individually in 1.3ml of Williams' Medi-um E (with L-glutamine, Invitrogen, Paisly, Scotland) supplementedwith 25 mM glucose and 50 μg/ml gentamycin (Invitrogen) at 37 °Cunder continuous supply of 95%O2, 5% CO2 in 12-well plateswhile gent-ly shaken. After 1 h of preincubation the slices were transferred to newplates with fresh medium and subsequently incubated for 24 or 48 h inthe presence or absence of antifibrotic compounds. Medium wasrefreshed every 24h. PCLSwere treatedwith antifibrotics demonstratedto be effective in previous studies utilizing animal models, primaryhuman cells and/or cell lines (Westra et al., 2014a; Jinnin et al., 2006;Tsukada et al., 2005; Xu et al., 2009) i.e. imatinib (Novartis, Basel, Swit-zerland), sorafenib (LC laboratories, Woburn, USA), sunitinib (LC labo-ratories), perindopril, valproic acid, rosmarinic acid, tetrandrine,pirfenidone and the specific MAPK inhibitor SB203580 (Bioconnect,Huissen, The Netherlands), the Smad 3 inhibitor Sis3 (Bioconnect) andthe Rac1 inhibitor NSC23766 (Tocris Bioscience, Bristol, UK). For thetested concentrations see Table 2 and to illustrate clinically relevantlevels, themaximumserumconcentration (Cmax) of compounds testedin humans is also provided. Furthermore, the PCLS were incubatedwiththe growth factors PDGF-BB (10 and 50 ng/ml; Recombinant HumanPDGF-BB, Peprotech, Bioconnect) and TGF-β1 (1 – 5 ng/ml; hTGF-β1,Roche Applied Science, Mannheim, Germany). Nonspecific binding ofTGF-β1 was prevented by preincubating the culture plates with 10%BSA in milli-Q for 20 min, whereafter the solution was removed andplates were dried at room temperature. All experiments were per-formed in triplicate (technical replicates) using liver tissue from 3 to 5different subjects.

2.5. Histomorphological examination

Integrity of the slices was assessed by immunohistochemistry aspreviously described (de Graaf et al., 2000). In short, PCLS were fixatedwith 70% ethanol at 4 °C for 24 h, and subsequently rehydrated in suc-cessive baths of xylene and graded alcohols using a Shandon 2LE Proces-sor. Afterwards, slices were vertically embedded in paraffin, sectioned(4 μm) and stained with hematoxylin and eosin. On microscopical ex-amination, viability was determined by estimating the percentage of vi-able cells in the cross-section, taking nuclear shape/staining andcytoplasmatic staining into account.

TGF-β Perindopril 10 – 100 0.04 , (Devissaguet et al., 1990)Valproic acid 100 –1000 1000, (Chavez-Blanco et al.,

2005)Rosmarinicacid

120 – 270 –

Tetrandrine 1 – 10 –Pirfenidone 500 – 2500 85, (Rubino et al., 2009)

Smad 3 Sis3 0.3 – 3 –MAPK SB203580 5 – 10 –Rac1 NSC23766 5 – 50 –

a After oral administration of 8mg perindopril. Needs to bemetabolized in vivo into theactive form perindoprilat (Devissaguet et al., 1990).

79I.M. Westra et al. / Toxicology in Vitro 35 (2016) 77–85

2.6. ATP determination

The viability of the slices was determined by measuring ATP levels,as reported previously (Hadi et al., 2013). In short, following treatment,PCLS were transferred to 1 ml sonication solution, containing 70% etha-nol and 2mMEDTA, snap frozen in liquid nitrogen and stored at−80 °Cuntil use. Subsequently, samples were thawed on ice, homogenized for45 s using aMini-BeadBeater-8 (Biospec, Bartlesville, OK, USA) and cen-trifuged for 2min at 16,000 ×g. ATP levels weremeasured in the super-natant with the ATP bioluminescence kit (Roche diagnostics,Mannheim, Germany), and corrected for the total protein content ofthe sample estimated via the Lowry assay (BIO-rad RC DC ProteinAssay; Bio Rad, Veenendaal, The Netherlands; (Lowry et al., 1951)).

2.7. Gene expression of fibrosis markers

To assess gene expression, total RNAwas isolated, from pooled (n=3) and snap frozen slices, using the RNeasy Mini kit (Qiagen, Venlo, theNetherlands) according to the manufacturer's recommendations. Theamount of isolated RNA was measured with the ND-1000 spectropho-tometer (Fisher Scientific, Landsmeer, The Netherlands), and 2 μg RNAwas reversed transcribed using the Reverse Transcription System(Promega, Leiden, The Netherlands). The RT-PCR reaction was per-formed in the Eppendorf mastercycler gradient at 25 °C for 10 min,45 °C for 60 min and 95 °C for 5 min. The mRNA expression levels ofheat shock protein 47 (HSP47), alpha smooth muscle actin (αSMA)and pro-collagen 1A1 (PCOL1A1) were detected using specific primer-probe sets (Table 3) and the qPCR master mix plus (Eurogentec, Maas-tricht, The Netherlands). The PCR reaction was performed using a7900HT Real Time PCR (Applied Biosystems, Bleiswijk, The Nether-lands) with 1 cycle of 10 min at 95 °C and 45 cycles of 15 s at 95 °Cand 1 min at 60 °C. Gene expression levels of PDGF-BB and TGF-β1were determined using the SYBRGreenmastermix (GC Biotech, Alphenaan de Rijn, The Netherlands) and specific primers (Table 3). The PCRreaction was performed using the same PCR system (see above) with1 cycle of 10 min at 95 °C and 45 cycles of 15 s at 95 °C and 25 s at60 °C followed by a dissociation curve (95 °C, 15 s: 60 °C, 15 s: 95 °C,15 s). GAPDHwas used as reference gene and relative expression levelswere calculated as fold change using the 2−ΔΔCT method.

2.8. Western blot

Collagen 1 protein expression was determined by Western blot.Stored PCLS (see above) were lysed for 1 h on ice in 250 μl RIPA buffercontaining 1 Protease inhibitor cocktail tablet (Boehringer Ingelheim,Alkmaar, The Netherlands), in 10 ml 50 mM Tris/HCl pH 7.5, 150 mMNaCl, 1% Igepal CA-630, 0.5% sodiumdeoxycholaat and 0.1% SDS. The tis-sue was homogenized on ice by a Potter homogenizer and centrifugedfor 1 h at 4 °C at 16,000 ×g. Protein concentrations were determinedin the supernatant using a Biorad DC protein assay according to themanufacturer's protocol. Lysateswere diluted 4 fold in SDS sample buff-er (50 mM TrisHCl pH 6.8, 2% SDS, 10% glycerol, 5% β-mercaptoethanol,0.05% bromophenol blue) and boiled for 2 min. Next, 100 μg of proteinwas separated via SDS/PAGE using 7.5% gels and blotted onto an activat-ed polyvinylidene difluoride membrane (Biorad). Afterwards, the

Table 3Primer sequences.

Gene Forward (5′ – 3′)

HSP47 GCCCACCGTGGTGCCGCAαSMA AGGGGGTGATGGGTGGGAAPCOL1A1 CAATCACCTGCGTACAGAACGCCGAPDH ACCAGGGCTGCTTTTAACTCTPDGF-β CTGGCATGCAAGTGTGAGACTGF-β1 GCAGCACGTGGAGCTGTAGAPDH CGCTGGTGCTGAGTATGTCG

membrane was blocked using Tris-buffered saline supplemented with5% Blocking Grade Powder (Biorad) and 0.1% Tween-20 during 1 h. Sub-sequently, membranes were incubated with rabbit-α-collagen-1(1:1000, Rockland Immunochemicals, Gilbertsville, PA, USA) ormouse-α-GAPDH (1:5000, DAKO, Heverlee, Belgium). Binding of theantibody was determined using Horseradish Peroxidase and appropri-ate secondary and tertiary antibodies (DAKO). Protein levels were visu-alized with an imaging system using Western Lightning Plus-ECL(Perkin Elmer, Groningen, The Netherlands). Collagen protein expres-sion was corrected for GAPDH levels and is displayed as relative valuescompared to the control.

2.9. Statistics

Three to five livers were used for each experiment, using triplicateslices from each liver. Results are expressed asmean± S.E.M. Statisticalanalysiswasperformedvia nonparametric ANOVA followed by aDunn'smultiple comparisons test or using an unpaired two-tailed Student's t-test. A p-value of b0.05 was considered significant.

3. Results

3.1. Viability and expression profile of fibrosis markers in PCLS duringculturing

As shown in Fig. 1A, ATP levels of the untreated PCLS increase duringthe 1 h preincubation as compared to the ATP values directly followingslicing (0 h). Moreover, the levels remain stable for at least 24 h with anaverage ATP content of 7.3 ± 0.6 pmol/μg protein. After 48 h of cultur-ing, there is a small 28% decrease of the ATP levels in untreated slicescompared to 24 h. Since there is a clear correlation between ATP contentof the slices and viability as determined by morphological scoring (FigS1, r = 0.78, p = 0.021), ATP levels can be used as a reliable viabilitymarker. These results indicate that the viability of healthy PCLS couldbe sustained for 48 h. Next, we studied the gene expression of severalfibrosis markers (e.g. HSP47, αSMA and PCOL1A1) during culturing.Fig. 1B demonstrates that the expression of HSP47 increases at all stud-ied timepoints in healthy PCLS, PCOL1A1 levelswere only elevated after48 h and αSMA expression was lower at all time points as compared tothe gene level at 0 h. Moreover, collagen 1 protein expression time-de-pendently increased 3 fold (24 h) and 10 fold (48 h) as compared to 0 h(Fig. 1C), in line with the observed PCOL1A1 gene levels. QuantitativePCR further revealed that after 48 h of culture, PCOL1A1 expression inPCLS is similar to the level observed in slices prepared from cirrhotic tis-sue (Fig. 1D). These findings illustrate the spontaneous induction offibrogenesis in healthy PCLS, with HSP47 and PCOL1A1 serving asclear markers for this process. Therefore, these two genes were usedas markers in the remainder of the study when evaluating the potencyof antifibrotics.

3.2. Pathways associated with the fibrogenic response in PCLS

Next, we investigatedwhichpathwayunderlies thefibrotic responsein human PCLS. After 48 h of culturing, expression levels of both PDGFand TGF-β were significantly up-regulated compared to control (0 h;

Reverse (5′ – 3′) Probe (5′ – 3′)

GCCAGGGCCGCCTCCAGGAG CTCCCTCCTGCTTCTCAGCGATGATGCCATGTTCTATCGG GGGTGACGAAGCACAGAGCACGGCAGGGCTCGGGTTTC CAGGTACCATGACCGAGACGTGGGTGCCATGGAATTTGCC TGCCATCAATGACCCCTTCACGAATGGTCACCCGAGTTTCAGCCGGTTGCTGAGGTACTGTGGTCATGAGCCCTTCC

Fig. 1. Viability of PCLS and expression offibrosismarkers.Human PCLSwere cultured for 48 h. (A) The viability of the sliceswas assessed by determining the ATP content (n=23). (B)After incubation, PCLSwere collected and totalmRNAwas isolated. Afterwards, cDNAwas synthesized andHSP47 (black bars; n=21),αSMA (white bars; n=7) and PCOL1A1 (red bars;n = 21) gene expression was studied using qPCR. GAPDH was used as reference gene and relative expression levels were calculated using the 2−ΔΔCT method. (C) Collagen 1 proteinexpression was studied by Western blotting. Proteins were separated via SDS/PAGE and blotted onto polyvinylidene difluoride membranes. Collagen was detected at 130 kD andGAPDH at 37 kD. Band intensities were corrected for GAPDH (n = 3). (D) PCOL1A1 gene expression was studied in PCLS prepared from healthy and cirrhotic liver tissue using qPCR.GAPDH was used as housekeeping gene and relative expression levels were calculated using the 2−ΔΔCT method (n = 3). Results are presented as mean ± SEM of minimally threeexperiments performed in triplicate. * indicates p b 0.05 compared to 0 h. (For interpretation of the references to color in this figure legend, the reader is referred to the web version ofthis article.)

80 I.M. Westra et al. / Toxicology in Vitro 35 (2016) 77–85

Fig. 2A), 6.2-fold and 2.7-fold, respectively. As shown in Fig. 2B and C,addition of either growth factor could not significantly augment the fi-brotic response in PCLS. Nevertheless, exposure of PCLS to a mixtureof PDGF and TGF-β resulted in a significant induction of both HSP47and PCOL1A1 gene expression as compared to control by 1.7-fold and2.4-fold, respectively (Fig. 2D). These results suggest that both path-ways might contribute to the development of fibrosis in PCLS. This no-tion is corroborated by the observed reduction in HSP47 and PCOL1A1gene expression following treatment with the MAPK inhibitorSB203580 (Fig. 2E). In contrast, NSC23766 solely decreased HSP47 ex-pression 1.3-fold, while Sis3 had no effect.

3.3. Impact of antifibrotic drugs on the gene expression of fibrosis markers

Since both the PDGF and TGF-β pathway appear to inducefibrogenesis in human PCLS, we investigated the capacity of multiplepotential antifibrotic compounds, associated with either pathway, tomitigate the observed increase in expression of the fibrosis markers.The inhibitors mainly targeting the PDGF pathway were not toxic forPCLS at the tested concentrations, except for the highest concentrationof imatinib (10 μM; Fig. 3A). Quantitative PCR revealed that sunitinib,in a concentration range of 0.5 μM to 5 μM, reduced the gene expressionof HSP47 and PCOL1A1withmore than 27%, and up to 65%, compared tothe control slices incubated for 48 hwithout inhibitor, whereas, imatin-ib and sorafenib did not influence the gene expression of both markers(Fig. 3B).

Regarding the TGF-β-inhibitors, toxicity was only observed forpirfenidone at the highest concentrations tested (2.5 mM) illustrated

by a reduction in ATP levels of more than 30% (Fig. 4A). As shown inFig. 4B, 0.5 mM valproic acid decreased the expression level of bothHSP47 and PCOL1A1 with 64% and 43%, respectively, while rosmarinicacid, tetrandrine and pirfenidone reduced the gene expression ofPCOL1A1 but not that of HSP47. In contrast, perindopril did not affectgene expression levels of any of the fibrosis markers. Thus, both classesof antifibrotic compounds can influence fibrogenesis in PCLS, yet theTGF-β-inhibitors seem to be more effective than the PDGF-inhibitors.

4. Discussion

Here, we describe the onset and molecular mechanism of fibrosis inhealthy human PCLS. In addition, we evaluated the antifibrotic efficacyof a broad range of different compounds using this unique model. Theexpected therapeutic targets of the tested compounds and the obtainedresults are summarized in Fig. 5.

In this study, we report the early onset of fibrosis in healthy humanPCLS during culture, as illustrated by a clear increase in the gene expres-sion of both HSP47 and PCOL1A1. However, the fibrotic process inhuman PCLS was not associated with an increased expression of thewell-known fibrosis and myofibroblast marker αSMA (Hinz et al.,2012). The observed gene expression profile is in agreementwith a pre-vious study from our group using human PCLS (van de Bovenkamp etal., 2008). As suggested previously, the decrease of αSMA expressionmight indicate a loss of myofibroblasts in human PCLS, since fibulin-2expression also decreased while gene levels of the hepatic stellate cellmarker, synaptophysin, remained stable during culture (van deBovenkamp et al., 2008). Although PCLS represent a good and

Fig. 2. Pathways involved in fibrogenesis in human PCLS. Human PCLS were cultured for 48 h in the presence or absence of PDGF, TGF-β or several inhibitors. Afterwards, geneexpression was studied using qPCR. GAPDH was used as reference gene and relative expression levels were calculated using the 2−ΔΔCT method. (A) Gene expression of TGF-β (blackbars) or PDGF (white bars) in human PCLS during culturing (n = 5). Impact of (B) PDGF (n = 4), (C) TGF-β (n = 4) or (D) PDGF + TGF-β (n = 5) on the gene expression of HSP47(black bars), αSMA (white bars) and PCOL1A1 (red bars) in PCLS. (E) Impact of several inhibitors on the gene expression of HSP47 (black bars) and PCOL1A1 (red bars) in PCLS (n =4). Results are presented as mean ± SEM of minimally three experiments performed in triplicate. * indicates p b 0.05, # indicates p = 0.059. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)

81I.M. Westra et al. / Toxicology in Vitro 35 (2016) 77–85

multicellular model for fibrotic disease, the specific role ofmyofibroblasts in this model requires further study.

Asmentioned before, there are clear species differences in the fibrot-ic process as well as the effectiveness of antifibrotic compounds. For in-stance, using rat PCLS, we previously demonstrated that the early onsetof fibrogenesis in this rodent model was accompanied by an increasedgene expression of HSP47, PCOL1A1 and αSMA (Westra et al., 2014a).Moreover, contrary to the current study, expression of these markerscould be augmented in rat PCLS by treatment with either PDGF or

TGF-β (Westra et al., 2014a). With regard to antifibrotic compounds,we now report that TGF-β pathway inhibitors are more effective indampening fibrosis in human PCLS as compared to PDGF-inhibitors.This is in disparity to rat PCLS, where the early onset of fibrosis couldbe more effectively countered with PDGF pathway inhibitors (Westraet al., 2014a). These resultsmight explainwhy several drugs, with prov-en antifibrotic effects in animal studies, lack efficacy in man. To illus-trate, imatinib, a well-known PDGF receptor tyrosine kinase inhibitor(Buchdunger et al., 2000), was demonstrated to be a potent antifibrotic

Fig. 3. Impact of PDGF-inhibitors on human PCLS.Human PCLS were treatedwith different PDGF-inhibitors for 48 h. (A) The viability of the slices was assessed by determining the ATPcontent (n = 4). (B) After treatment, PCLS were collected and total mRNA was isolated. Afterwards, cDNA was synthesized and HSP47 (black bars) and PCOL1A1 (red bars) geneexpression was studied using qPCR (n = 5). GAPDH was used as reference gene and relative expression levels were calculated using the 2−ΔΔCT method. Results are presented asmean ± SEM of minimally three experiments performed in triplicate. * indicates p b 0.05, # indicates p = 0.074, $ indicates p = 0.056. (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)

82 I.M. Westra et al. / Toxicology in Vitro 35 (2016) 77–85

compound in a broad array of rodent fibrosis models affecting severalorgans such as the lung, kidney and liver (Westra et al., 2014a, 2014b;Daniels et al., 2004;Wang et al., 2005; Yoshiji et al., 2005). Moreover, i-matinib has been shown to induce apoptosis of hepatic stellate cells invitro (Kuo et al., 2012). Yet, in the current study, imatinibwas ineffectiveas antifibrotic compound in human PCLS. This finding is in agreementwith a randomized double-blind clinical trial showing that the com-pound was ineffective as antifibrotic drug in humans, since the com-pound did not affect survival or lung function in patients withidiopathic pulmonary fibrosis (Daniels et al., 2010). In addition, it iswell known that imatinib can induce hepatotoxicity (Mindikoglu etal., 2007; Lyseng-Williamson and Jarvis, 2001), similar to our observa-tions in human PCLS. Taken together, it is clear that there are species-specific effects of potential antifibrotic compounds, further illuminatingthe pressing need for reliable and reproducible human in vitro fibrosismodels for, amongst others, drug development.

Our results further demonstrate that pirfenidone, valproic acid androsmarinic acid, amongst others, had a clear impact on the gene expres-sion offibrosismarkers in human PCLS. These results are in linewith ourprevious observations in rodent PCLS (Westra et al., 2014a; Iswandanaet al., 2016). Pirfenidone is a well-established antifibrotic compoundshown to reduce TGF-β expression in a wide variety of animal modelsof fibrosis (Schaefer et al., 2011). Also, using human LX-2 cells, it wasshown that pirfenidone diminished αSMA and COL1 gene expression(Zhao et al., 2009). Furthermore, Armendariz-Borunda et al. (2006) re-ported that twelve months of pirfenidone treatment caused a reductionin both the gene- and protein expression of COL1A, TGF-β and TIMP-1 inliver biopsies from patients with advanced liver fibrosis (Armendariz-Borunda et al., 2006). Moreover, treatment with pirfenidone was alsodemonstrated to be beneficial in patients with diabetic nephropathyand idiopathic pulmonary fibrosis (Sharma et al., 2011; King et al.,2014). Yet, in our hands, toxicity was observed with the highest testedconcentration of pirfenidone, although this did not result in a reductionof HSP47 gene expression. Therefore, the true antifibrotic potential ofpirfenidone requires further scrutiny; nevertheless, the compoundshows great potential as a therapeutic agent against fibrotic diseasesin humans, partially supported by our results in human PCLS.

Valproic acid markedly decreased gene expression of fibrosismarkers in human PCLS, although this effect might be due to hepato-toxicity, since a non-significant decrease of ATP content was ob-served at concentrations ≥0.5 mM. Nonetheless, at the lowestconcentration we also observed a clear reduction in PCOL1A1 genelevels. This suggests that valproic acid might be useful as antifibroticdrug. This notion is supported by the study of Watanabe et al. (2011)showing that valproic acid diminished TFG-β-induced collagen 1A1gene and protein expression, as well as the number of αSMA fibersin LI90 cells (Watanabe et al., 2011). These results warrant furtherinvestigation into the use of HDAC inhibitors for the treatment of(liver) fibrosis.

Rosmarinic acid is a caffeic acid ester present in a myriad of plants(Petersen et al., 2009). Several studies report antifibrotic effects of thiscompound (Li et al., 2010; Zhang et al., 2011; Domitrovic et al., 2013).Yet, to the best of our knowledge, our study is the first to report theantifibrotic properties of rosmarinic acid using a human fibrosismodel. Previously, we demonstrated that rosmarinic acid decreasedthe expression of fibrosis markers in fibrotic rat PCLS as well as healthymurine PCLS (Westra et al., 2014a, 2014b; Iswandana et al., 2016).Therefore we postulate that antifibrotic effects of rosmarinic acid inhuman liver fibrosis may be expected.

Next to testing the efficacy of antifibrotics in human PCLS we alsostrove to unravel the molecular pathways underlying fibrogenesis inthe model. Our results indicated that the MAPK pathway plays a vitalrole during the onset of the fibrotic process in human PCLS, in agree-ment with our observations in rat PCLS (unpublished data). This notionis further supported by the study from Varela-Rey et al. (2002) demon-strating that the specific p38 MAPK inhibitor SB203580 abated the in-duction of COL1A1 gene expression by TGF-β in hepatic stellate cells(Varela-Rey et al., 2002). In addition, our findings suggested that induc-tion of fibrosis in human PCLS by TGF-β goes via theMAPK pathway andis independent of Smad3. The Smad-independent activity of both path-ways in regulating collagen gene expression has been reported beforeby Tsukada et al. (2005). The trigger for MAPK activation in humanPCLS remains unclear. It is known that this pathway can be induced byreactive oxygen species (ROS) (Adachi et al., 2005), and even though

Fig. 4. Impact of TGF-β-inhibitors on human PCLS. Human PCLS were treated with different TGF-β-inhibitors for 48 h. (A) Viability of the slices was assessed by determining the ATPcontent (n = 4). (B) After treatment, PCLS were collected and total mRNA was isolated. Afterwards, cDNA was synthesized and HSP47 (black bars) and PCOL1A1 (red bars) geneexpression was studied using qPCR (n = 4). GAPDH was used as reference gene and relative expression levels were calculated using the 2−ΔΔCT method. Results are presented asmean ± SEM of minimally three experiments performed in triplicate. * indicates p b 0.05, # indicates p = 0.063, $ indicates p = 0.055, & indicates p = 0.075. (For interpretation ofthe references to color in this figure legend, the reader is referred to the web version of this article.)

83I.M. Westra et al. / Toxicology in Vitro 35 (2016) 77–85

PCLS are cultured in an atmosphere with high oxygen levels prior liter-ature demonstrated that there is no evidence for the presence of ROSin themodel (Schaffert et al., 2010; Klassen et al., 2008; Vickers et al.,2004). Moreover, previous work from our group revealed that accu-mulation of waste products or bile salts is not involved in the ob-served fibrotic response in PCLS (van de Bovenkamp et al., 2008).Thus, the causative agent(s) for fibrogenesis by culture activationin human PCLS remain to be elucidated. Of note, culture conditionshave a profound impact on PCLS and should be tailored to fit thestudy type (e.g. drug metabolism or toxicity) (Starokozhko et al.,2015).

The current study only delineated the pathways underlying the earlyonset of fibrosis in human PCLS as well as the efficacy of antifibroticcompounds during this stage of the disease process. To gain more in-sight in the complete pathophysiological mechanism of liver fibrosisstudies are currently underway using PCLS prepared from human

cirrhotic livers. These studies will hopefully grant us a better under-standing of fibrotic disease and provide new leads for therapeutictargets.

In conclusion, in this study we demonstrate that TGF-β pathway in-hibitors effectively hamper incipient fibrosis in human PCLS. Moreover,it became apparent that theMAPK signaling cascade plays a vital role inthe fibrotic process. Our results provide additional insight into themechanism of liver fibrosis and establish human PCLS as a good modelfor this disease.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.tiv.2016.05.012.

Transparency document

The Transparency document associated with this article can befound, in online version.

Fig. 5. Overview of drug targets and efficacy of the tested compounds in PCLS. Schematic overview of the expected drug targets. Compounds that show antifibrotic effects in ourmodelare emphasized in bold and underline. Compounds for which the molecular targets are not fully elucidated are presented in italic.

84 I.M. Westra et al. / Toxicology in Vitro 35 (2016) 77–85

Acknowledgements

The authors would like to thank all the transplantation coordinatorsand surgeons at the UMCG for providing human liver samples. We alsothank Marina de Jager, Miriam Langelaar-Makkinje, Marjolijn Merema,Sylvia Blomsma and Patricia Robbe for their help with preparinghuman PCLS, and Inge de Graaf for histomorphological scoring. Thiswork was supported by ZonMw (Program: Dierproeven begrensd II,project 114000098).

References

Adachi, T., Togashi, H., Suzuki, A., Kasai, S., Ito, J., Sugahara, K., et al., 2005. NAD(P)H oxi-dase plays a crucial role in PDGF-induced proliferation of hepatic stellate cells.Hepatology 41 (6), 1272–1281.

Armendariz-Borunda, J., Islas-Carbajal, M.C., Meza-Garcia, E., Rincon, A.R., Lucano, S.,Sandoval, A.S., et al., 2006. A pilot study in patients with established advanced liverfibrosis using pirfenidone. Gut 55 (11), 1663–1665.

Bonner, J.C., 2004. Regulation of PDGF and its receptors in fibrotic diseases. CytokineGrowth Factor Rev. 15 (4), 255–273.

Buchdunger, E., Cioffi, C.L., Law, N., Stover, D., Ohno-Jones, S., Druker, B.J., et al., 2000. Abl pro-tein-tyrosine kinase inhibitor STI571 inhibits in vitro signal transduction mediated by c-kit and platelet-derived growth factor receptors. J. Pharmacol. Exp. Ther. 295 (1),139–145.

Chavez-Blanco, A., Segura-Pacheco, B., Perez-Cardenas, E., Taja-Chayeb, L., Cetina, L.,Candelaria, M., et al., 2005. Histone acetylation and histone deacetylase activity ofmagnesium valproate in tumor and peripheral blood of patients with cervical cancer.A phase I study. Mol. Cancer 4 (1), 22.

Cuenda, A., Rouse, J., Doza, Y.N., Meier, R., Cohen, P., Gallagher, T.F., et al., 1995. Sb-203580is a specific inhibitor of a map kinase homologwhich is stimulated by cellular stressesand interleukin-1. FEBS Lett. 364 (2), 229–233.

Daniels, C.E., Wilkes, M.C., Edens, M., Kottom, T.J., Murphy, S.J., Limper, A.H., et al., 2004.Imatinib mesylate inhibits the profibrogenic activity of TGF-beta and preventsbleomycin-mediated lung fibrosis. J. Clin. Invest. 114 (9), 1308–1316.

Daniels, C.E., Lasky, J.A., Limper, A.H., Mieras, K., Gabor, E., Schroeder, D.R., et al., 2010. I-matinib treatment for idiopathic pulmonary fibrosis: randomized placebo-controlledtrial results. Am. J. Respir. Crit. Care Med. 181 (6), 604–610.

de Graaf, I.A., van der Voort, D., Brits, J.H., Koster, H.J., 2000. Increased post-thaw viabilityand phase I and II biotransformation activity in cryopreserved rat liver slices after im-provement of a fast-freezing method. Drug Metab. Dispos. 28 (9), 1100–1106.

de Graaf, I.A., Olinga, P., de Jager, M.H., Merema, M.T., de Kanter, R., van de Kerkhof, E.G., etal., 2010. Preparation and incubation of precision-cut liver and intestinal slices for ap-plication in drug metabolism and toxicity studies. Nat. Protoc. 5 (9), 1540–1551.

Devissaguet, J.P., Ammoury, N., Devissaguet, M., Perret, L., 1990. Pharmacokinetics ofperindopril and its metabolites in healthy volunteers. Fundam. Clin. Pharmacol. 4 (2),175–189.

Domitrovic, R., Skoda, M., Vasiljev Marchesi, V., Cvijanovic, O., Pernjak Pugel, E., Stefan,M.B., 2013. Rosmarinic acid ameliorates acute liver damage and fibrogenesis in car-bon tetrachloride-intoxicated mice. Food Chem. Toxicol. 51, 370–378.

Elferink, M.G., Olinga, P., van Leeuwen, E.M., Bauerschmidt, S., Polman, J., Schoonen, W.G.,et al., 2011. Gene expression analysis of precision-cut human liver slices indicates sta-ble expression of ADME-Tox related genes. Toxicol. Appl. Pharmacol. 253 (1), 57–69.

Gao, Y., Dickerson, J.B., Guo, F., Zheng, J., Zheng, Y., 2004. Rational design and characteri-zation of a Rac GTPase-specific small molecule inhibitor. Proc. Natl. Acad. Sci. U. S.A. 101 (20), 7618–7623.

Hadi, M., Westra, I.M., Starokozhko, V., Dragovic, S., Merema, M.T., Groothuis, G.M., 2013.Human precision-cut liver slices as an ex vivo model to study idiosyncratic drug-in-duced liver injury. Chem. Res. Toxicol. 26 (5), 710–720.

Hinz, B., Phan, S.H., Thannickal, V.J., Prunotto, M., Desmouliere, A., Varga, J., et al., 2012. Re-cent developments inmyofibroblast biology: paradigms for connective tissue remod-eling. Am. J. Pathol. 180 (4), 1340–1355.

Inoue, T., Umezawa, A., Takenaka, T., Suzuki, H., Okada, H., 2015. The contribution of epi-thelial–mesenchymal transition to renal fibrosis differs among kidney diseasemodels. Kidney Int. 87 (1), 233–238.

Iswandana, R., Pham, B.T., van Haaften, W.T., Luangmonkong, T., Oosterhuis, D., Mutsaers,H.A., et al., 2016. Organ- and species-specific biological activity of rosmarinic acid.Toxicol. in Vitro 32, 261–268.

85I.M. Westra et al. / Toxicology in Vitro 35 (2016) 77–85

Jinnin, M., Ihn, H., Tamaki, K., 2006. Characterization of SIS3, a novel specific inhibitor ofSmad3, and its effect on transforming growth factor-beta1-induced extracellular ma-trix expression. Mol. Pharmacol. 69 (2), 597–607.

King Jr., T.E., Bradford, W.Z., Castro-Bernardini, S., Fagan, E.A., Glaspole, I., Glassberg, M.K.,et al., 2014. A phase 3 trial of pirfenidone in patients with idiopathic pulmonary fibro-sis. N. Engl. J. Med. 370 (22), 2083–2092.

Klassen, L.W., Thiele, G.M., Duryee, M.J., Schaffert, C.S., DeVeney, A.L., Hunter, C.D., et al.,2008. An in vitro method of alcoholic liver injury using precision-cut liver slicesfrom rats. Biochem. Pharmacol. 76 (3), 426–436.

Kuo, W.L., Yu, M.C., Lee, J.F., Tsai, C.N., Chen, T.C., Chen, M.F., 2012. Imatinib mesylate im-proves liver regeneration and attenuates liver fibrogenesis in CCL4-treated mice.J. Gastrointest. Surg. 16 (2), 361–369.

Leveque, D., Maloisel, F., 2005. Clinical pharmacokinetics of imatinib mesylate. In vivo(Athens, Greece) 19 (1), 77–84.

Li, G.S., Jiang, W.L., Tian, J.W., Qu, G.W., Zhu, H.B., Fu, F.H., 2010. In vitro and in vivoantifibrotic effects of rosmarinic acid on experimental liver fibrosis. Phytomedicine17 (3-4), 282–288.

Liu, Y., 2011. Cellular and molecular mechanisms of renal fibrosis. Nat. Rev. Nephrol. 7(12), 684–696.

Liu, Y., Meyer, C., Xu, C., Weng, H., Hellerbrand, C., ten Dijke, P., et al., 2013. Animal modelsof chronic liver diseases. Am. J. Physiol. Gastrointest. Liver Physiol. 304 (5),G449–G468.

Lowry, O.H., Rosebrough, N.J., Farr, A.L., Randall, R.J., 1951. Protein measurement with theFolin phenol reagent. J. Biol. Chem. 193 (1), 265–275.

Lyseng-Williamson, K., Jarvis, B., 2001. Imatinib. Drugs. 61 (12), 1765–1774 (discussion75-6).

Mindikoglu, A.L., Regev, A., Bejarano, P.A., Martinez, E.J., Jeffers, L.J., Schiff, E.R., 2007. Ima-tinib mesylate (gleevec) hepatotoxicity. Dig. Dis. Sci. 52 (2), 598–601.

Minkin, P., Zhao, M., Chen, Z., Ouwerkerk, J., Gelderblom, H., Baker, S.D., 2008. Quantifica-tion of sunitinib in human plasma by high-performance liquid chromatography-tan-demmass spectrometry. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 874 (1-2),84–88.

Olinga, P., Merema, M.T., Hof, I.H., De Jager, M.H., De Jong, K.P., Slooff, M.J., et al., 1998. Ef-fect of cold and warm ischaemia on drug metabolism in isolated hepatocytes andslices from human and monkey liver. Xenobiotica 28 (4), 349–360.

Parsons, C.J., Takashima, M., Rippe, R.A., 2007. Molecular mechanisms of hepaticfibrogenesis. J. Gastroenterol. Hepatol. 22 (Suppl. 1), S79–S84.

Pellicoro, A., Ramachandran, P., Iredale, J.P., Fallowfield, J.A., 2014. Liver fibrosis and re-pair: immune regulation of wound healing in a solid organ. Nat. Rev. Immunol. 14(3), 181–194.

Petersen, M., Abdullah, Y., Benner, J., Eberle, D., Gehlen, K., Hucherig, S., et al., 2009. Evo-lution of rosmarinic acid biosynthesis. Phytochemistry 70 (15-16), 1663–1679.

Rubino, C.M., Bhavnani, S.M., Ambrose, P.G., Forrest, A., Loutit, J.S., 2009. Effect of food andantacids on the pharmacokinetics of pirfenidone in older healthy adults. Pulm.Pharmacol. Ther. 22 (4), 279–285.

Schaefer, C.J., Ruhrmund, D.W., Pan, L., Seiwert, S.D., Kossen, K., 2011. Antifibrotic activi-ties of pirfenidone in animal models. Eur. Respir. Rev. 20 (120), 85–97.

Schaffert, C.S., Duryee, M.J., Bennett, R.G., DeVeney, A.L., Tuma, D.J., Olinga, P., et al., 2010.Exposure of precision-cut rat liver slices to ethanol accelerates fibrogenesis. Am.J. Physiol. Gastrointest. Liver Physiol. 299 (3), G661–G668.

Schuppan, D., Kim, Y.O., 2013. Evolving therapies for liver fibrosis. J. Clin. Invest. 123 (5),1887–1901.

Sharma, K., Ix, J.H., Mathew, A.V., Cho, M., Pflueger, A., Dunn, S.R., et al., 2011. Pirfenidonefor diabetic nephropathy. J. Am. Soc. Nephrol. 22 (6), 1144–1151.

Starokozhko, V., Abza, G.B., Maessen, H.C., Merema, M.T., Kuper, F., Groothuis, G.M., 2015.Viability, function and morphological integrity of precision-cut liver slices duringprolonged incubation: effects of culture medium. Toxicol. in Vitro 30 (1 Pt B),288–299.

Stribos, E.G., Luangmonkong, T., Leliveld, A.M., de Jong, I.J., van Son,W.J., JL, H., et al., 2015.Precision-cut human kidney slices as a model to elucidate the process of renal fibro-sis. Transl. Res.

Strumberg, D., Clark, J.W., Awada, A., Moore, M.J., Richly, H., Hendlisz, A., et al., 2007. Safe-ty, pharmacokinetics, and preliminary antitumor activity of sorafenib: a review offour phase I trials in patients with advanced refractory solid tumors. Oncologist 12(4), 426–437.

Tsukada, S., Westwick, J.K., Ikejima, K., Sato, N., Rippe, R.A., 2005. SMAD and p38 MAPKsignaling pathways independently regulate alpha1(I) collagen gene expression inunstimulated and transforming growth factor-beta-stimulated hepatic stellate cells.J. Biol. Chem. 280 (11), 10055–10064.

van de Bovenkamp, M., Groothuis, G.M., Meijer, D.K., Olinga, P., 2006a. Precision-cut fi-brotic rat liver slices as a new model to test the effects of anti-fibrotic drugs invitro. J. Hepatol. 45 (5), 696–703.

van de Bovenkamp, M., Groothuis, G.M., Meijer, D.K., Slooff, M.J., Olinga, P., 2006b. Humanliver slices as an in vitro model to study toxicity-induced hepatic stellate cell activa-tion in a multicellular milieu. Chem. Biol. Interact. 162 (1), 62–69.

van de Bovenkamp, M., Groothuis, G.M., Meijer, D.K., Olinga, P., 2008. Liver slices as amodel to study fibrogenesis and test the effects of anti-fibrotic drugs on fibrogeniccells in human liver. Toxicol. in Vitro 22 (3), 771–778.

Varela-Rey, M., Montiel-Duarte, C., Oses-Prieto, J.A., Lopez-Zabalza, M.J., Jaffrezou, J.P.,Rojkind, M., et al., 2002. p38 MAPK mediates the regulation of alpha1(I) procollagenmRNA levels by TNF-alpha and TGF-beta in a cell line of rat hepatic stellate cells(1).FEBS Lett. 528 (1-3), 133–138.

Vickers, A.E., Saulnier, M., Cruz, E., Merema, M.T., Rose, K., Bentley, P., et al., 2004. Organslice viability extended for pathway characterization: an in vitro model to investigatefibrosis. Toxicol. Sci. Off. J. Soc. Toxicol. 82 (2), 534–544.

Wang, S., Wilkes, M.C., Leof, E.B., Hirschberg, R., 2005. Imatinib mesylate blocks a non-Smad TGF-beta pathway and reduces renal fibrogenesis in vivo. FASEB J. 19 (1), 1–11.

Watanabe, T., Tajima, H., Hironori, H., Nakagawara, H., Ohnishi, I., Takamura, H., et al.,2011. Sodium valproate blocks the transforming growth factor (TGF)-beta1 autocrineloop and attenuates the TGF-beta1-induced collagen synthesis in a human hepaticstellate cell line. Int. J. Mol. Med. 28 (6), 919–925.

Westra, I.M., Pham, B.T., Groothuis, G.M., Olinga, P., 2013. Evaluation of fibrosis in preci-sion-cut tissue slices. Xenobiotica 43 (1), 98–112.

Westra, I.M., Oosterhuis, D., Groothuis, G.M., Olinga, P., 2014a. Precision-cut liver slices asa model for the early onset of liver fibrosis to test antifibrotic drugs. Toxicol. Appl.Pharmacol. 274 (2), 328–338.

Westra, I.M., Oosterhuis, D., Groothuis, G.M., Olinga, P., 2014b. The effect of antifibroticdrugs in rat precision-cut fibrotic liver slices. PLoS ONE 9 (4), e95462.

Xu, S.W., Liu, S., Eastwood, M., Sonnylal, S., Denton, C.P., Abraham, D.J., et al., 2009. Rac in-hibition reverses the phenotype of fibrotic fibroblasts. PLoS ONE 4 (10), e7438.

Yoshiji, H., Noguchi, R., Kuriyama, S., Ikenaka, Y., Yoshii, J., Yanase, K., et al., 2005. Imatinibmesylate (STI-571) attenuates liver fibrosis development in rats. Am. J. Physiol.Gastrointest. Liver Physiol. 288 (5), G907–G913.

Zeisberg, M., Kalluri, R., 2013. Cellular mechanisms of tissue fibrosis. 1. Common andorgan-specific mechanisms associated with tissue fibrosis. Am. J. Phys. Cell Phys.304 (3), C216–C225.

Zhang, J.J., Wang, Y.L., Feng, X.B., Song, X.D., Liu, W.B., 2011. Rosmarinic acid inhibits pro-liferation and induces apoptosis of hepatic stellate cells. Biol. Pharm. Bull. 34 (3),343–348.

Zhao, X.Y., Zeng, X., Li, X.M., Wang, T.L., Wang, B.E., 2009. Pirfenidone inhibits carbon tet-rachloride- and albumin complex-induced liver fibrosis in rodents by preventing ac-tivation of hepatic stellate cells. Clin. Exp. Pharmacol. Physiol. 36 (10), 963–968.