BASIC STUDIES Fibrogenic cell phenotype modi��cations during remodelling of normal and pathological human liver in cultured slices Christelle Guyot1,2 , Sebastien �� Lepreux1,3 , Chantal Combe1, Vincent Sarrazy4, Fabrice Billet4, Charles Balabaud1, Paulette Bioulac-Sage3 and Alexis Desmouliere1,4` 1 INSERM U889 and Universite�� Victor Segalen Bordeaux 2, Bordeaux, France 2 Division of Clinical Pharmacology and Toxicology, University Hospital, Zurich, Switzerland 3 CHU Bordeaux, Hopital �� Pellegrin, Service d���Anatomie Pathologique, Bordeaux, France 4 Departement �� de Physiologie, and EA 3842 (Homeostasie �� et Pathologies Cellulaires), Facult �� e de Pharmacie, Universite�� de Limoges, Limoges, France Keywords CD 34 ��� cirrhosis ��� fibrosis ��� keratin 19 ��� liver slice ��� platelet-derived growth factor receptor-b ��� Thy-1 ��� a-smooth muscle actin Correspondence Dr Alexis Desmouliere, ` Department of Physiology, Faculty of Pharmacy, University of Limoges, 2 rue du Dr Marcland, 87025 Limoges cedex, France Tel: 133 555 435 873 Fax: 133 555 435 801 e-mail: alexis.desmouliere@unilim.fr Received 6 March 2010 Accepted 9 August 2010 DOI:10.1111/j.1478-3231.2010.02342.x Abstract Background: The debate concerning the potential remodelling and/or reversi- bility of cirrhotic lesions and biliary fibrosis is still open. Aims/Methods: In this work, we have used the precision-cut liver slice (PCLS) model, which maintains cell���cell and cell���matrix interactions to study, by immunohisto- chemistry, the behaviour of the different fibrogenic cells, i.e. hepatic stellate cells (HSC) and portal fibroblasts, in cultured (for 1 week) PCLS derived from normal and fibrotic human livers. Results: In normal liver, before and after culture, a-smooth muscle (SM) actin was present only in the vessel walls. Platelet-derived growth factor (PDGF) receptor-b was expressed before and after culture by portal fibroblasts, and appeared after culture in HSC. Before culture, CD 34 was not expressed in parenchyma, but appeared after culture in sinusoidal endothelial cells. In cirrhotic lesions, before culture, a-SM actin, PDGF receptor-b and Thy-1 were expressed in septa after culture, a-SM actin expression disappeared but the expression of the PDGF receptor-b and Thy-1 was maintained. In cholestatic liver specimens, a-SM actin, PDGF receptor-b and Thy-1 expression, which was present before culture in enlarged portal areas, disappeared after culture, and apoptosis was detected. In the parench- yma of both cirrhotic and cholestatic livers, the expression of the PDGF receptor-b and of CD 34, which was not observed before culture, was present in HSC and sinusoidal endothelial cells, respectively, after culture. Conclu- sions: These results indicate that during remodelling of pathological tissues in cultured liver slices, the myofibroblastic cells derived from HSC or from portal fibroblasts show different behaviours, suggesting different mechanisms of activation/deactivation. Recent knowledge derived from numerous pertinent studies investigating the mechanisms involved in liver fibrosis supports the notion that liver fibrosis and possibly cirrhosis are reversible (1). Even if substantial functional and structural improvements may occur in the cirrhotic liver (2), complete regression of all anatomic features of fibrosis seems highly unlikely. Moreover, regression of cholestatic liver fibrosis was observed in patients who underwent surgical decompression of an obstructed biliary system (3). Experimental data and observations of human liver fibrosis and cirrhosis may eventually resolve the debate concerning the potential reversibility of cirrhosis (4). However, to study the mechanisms involved in potential remodelling of liver fibrosis and cirrhosis in certain specific situations, the cells involved must be clearly identified. It is now well accepted that both hepatic stellate cells (HSC) and portal fibroblasts can develop a myofibroblastic phenotype, i.e. acquire a-smooth muscle (SM) actin expression and participate in extracellular matrix deposition leading to fibrosis (5). In normal liver, typical myofibroblasts are not present. In pathological situations, myofibroblastic cells expressing a-SM actin are derived either from HSC (e.g. in alcoholic cirrhosis where myofibroblasts are present in the parenchyma and in the septa) or from portal fibroblasts (e.g. in cholestatic fibrosis where myo- fibroblasts are present in enlarged portal areas). Thus, in the liver, fibrogenic cells mainly consist of HSC and portal fibroblasts, even if, in some cases, second layer Contributed equally to this work. Liver International (2010) c 2010 John Wiley & Sons A/S 1529 Liver International ISSN 1478-3223

cells located around centrilobular veins may be involved in parenchymal fibrosis (5). In addition to in vivo experiments, numerous in vitro models have been devel- oped to decrease the complexity of the experiments and to reduce animal use. Cell culture is widely used and very useful however, compared with in vivo studies, some disadvantages such as loss of differentiated functions in cultured cells are encountered. The precision-cut liver slice (PCLS) model preserves the normal lobular archi- tecture and allows the maintenance of cell���cell interac- tions within their original extracellular matrix. The PCLS culture model has been applied recently to the study of HSC activation (6, 7) it has also been shown that endothelial cells (8) and Kupffer cells (9) remain viable in PCLS culture. We have used this model to study the effects of bile acids on biliary epithelial cell proliferation and portal fibroblast activation (10), and we applied this model recently to study HSC and portal fibroblasts behaviour in rat PCLS derived from fibrotic tissue (11). Finally, this model represents a useful model in which the effects of various antifibrotic drugs can be studied (12). In this work, we have used PCLS derived from normal and pathological human livers to study, by immunohis- tochemistry, the behaviour of the different liver fibro- genic cells during the tissue modification that develops in culture. We observed that after culture of pathological tissues, myofibroblasts derived from HSC or from portal fibroblasts show different patterns of evolution, mimick- ing in part the processes that may develop during tissue remodelling in vivo. Materials and methods Human liver tissues Large liver specimens were obtained from seven patients, corresponding to subnormal liver dissected from an area distant from a cholangiocarcinoma (one case), endocrine carcinoma metastasis (one case) and segmental Caroli���s disease (one case), and to fibrotic liver including alco- holic cirrhosis (two cases) or secondary biliary fibrosis surrounding gallbladder adenocarcinoma and cholangio- carcinoma (two cases) (Table 1). Tissue was washed with pre-oxygenated ice-cold Hanks��� balanced salt solution (Invitrogen, Cergy Pontoise, France), supplemented with 5 mM glucose and 50 mg/ml gentamycin and then pro- cessed for PCLS preparation. The procedures were in accordance with the European Guidelines for the use of human tissues. Precision-cut liver slice preparation, culture and processing Precision-cut liver slice were prepared as described pre- viously (10). Briefly, tissue cores were performed using a motor-driven coring tool (8 mm diameter). Consecutive PCLS (250 mm thick) were obtained using a Krumdieck tissue slicer (Alabama Corporation, Munford, AL, USA). PCLS were placed on a stainless-steel insert (Alabama Corporation) in an incubation vial containing culture medium. Vials were set on a roller platform and gently agitated in a humidified incubator (Sanyo, Osaka, Japan) at 37 1C, with 5% CO2 and 40% O2. Slices were firstly incubated in Williams��� E medium (Invitrogen) supple- mented with 0.35 mM insulin (Sigma, Saint-Quentin Fallavier, France), 0.1 mM dexamethasone (Sigma) and 5% fetal calf serum (Dutscher, Brumath, France). After 2 h, this medium was replaced by Williams��� E medium supplemented with insulin, dexamethasone and 1% fetal calf serum. PCLS were cultured for 1 week. After culture, PCLS were routinely formalin-fixed and paraffin-embedded. Sections (4 mm thick) were stained with haematoxylin���eosin for routine histology, with Sirius red to allow visualization of fibrosis, or processed for immunohistochemistry. Antibodies and immunohistochemistry, apoptosis detection For immunohistochemistry, the following primary anti- bodies were used: mouse monoclonal antibodies against a-SM actin (IgG2a, clone 1A4, DakoCytomation, Table 1. Human normal and pathological specimens Patient Sex/age Aetiology Histology of the non-tumoral liver 1 F/71 Left hepatectomy for intrahepatic cholangiocarcinoma Histologically subnormal liver [mild steatosis (o 10%)] 2 M/52 Segmental hepatectomy for endocrine carcinoma metastasis Histologically subnormal liver [mild steatosis (o 10%)] 3 M/72 Segmental hepatectomy for segmental Caroli���s disease Histologically subnormal liver [mild steatosis (o 10%)] 4 M/57 Transplantation for a decompensated alcoholic cirrhosis associated with hepatocellular carcinomas Mixed micro/macronodular cirrhosis with moderate inflammation 5 F/46 Transplantation for a decompensated alcoholic cirrhosis Micronodular cirrhosis with mild inflammation 6 F/63 Right hepatectomy for a gallbladder adenocarcinoma Chronic obstructive cholestasis with secondary biliary septal fibrosis. 7 M/78 Left hepatectomy for intrahepatic cholangiocarcinoma Chronic obstructive cholestasis with secondary biliary septal fibrosis. Liver International (2010) 1530 c 2010 John Wiley & Sons A/S Human liver in cultured slices Guyot et al.