Gene expression profiles of hepatic cell-type specific marker genes in progression of liver fibrosis

Abstract

AIM: To determine the gene expression profile data for the whole liver during development of dimethylni-trosamine (DMN)-induced hepatic fibrosis.

METHODS: Marker genes were identified for different types of hepatic cells, including hepatic stellate cells (HSCs), Kupffer cells (including other inflammatory cells), and hepatocytes, using independent temporal DNA microarray data obtained from isolated hepatic cells.

RESULTS: The cell-type analysis of gene expression gave several key results and led to formation of three hypotheses: (1) changes in the expression of HSC-specific marker genes during fibrosis were similar to gene expression data in in vitro cultured HSCs, suggesting a major role of the self-activating characteristics of HSCs in formation of fibrosis; (2) expression of mast cell-specific marker genes reached a peak during liver fibrosis, suggesting a possible role of mast cells in formation of fibrosis; and (3) abnormal expression of hepatocyte-specific marker genes was found across several metabolic pathways during fibrosis, including sulfur-containing amino acid metabolism, fatty acid metabolism, and drug metabolism, suggesting a mechanistic relationship between these abnormalities and symptoms of liver fibrosis.

CONCLUSION: Analysis of marker genes for specific hepatic cell types can identify the key aspects of fibrogenesis. Sequential activation of inflammatory cells and the self-supporting properties of HSCs play an important role in development of fibrosis.

Key Words: Liver fibrosis, Gene expression, Microarray, Dimethylnitrosamine, Marker genes, Hepatic stellate cell, Kupffer cell, Hepatocytes, Metabolic pathway

INTRODUCTION

The pathological relationship between chronic inflammation and formation of fibrosis has been established in various organs, including the liver, kidney, lung and pancreas. Although liver fibrosis has been studied extensively, the underlying mechanisms remain unclear and drugs to prevent and treat fibrosis are only partially effective. DNA microarray technology offers an approach to this kind of complex problems, and microarray analyses of the whole liver have been reported for liver fibrosis[1,2]. However, these analyses did not address the behavior of individual hepatic cell-types and the interactions of hepatic cells during fibrosis. Therefore, in the current study we identified hepatic cell-specific marker genes that could be used to understand the in vivo behavior of each type of hepatic cells during fibrogenesis.

About 70%-80% of hepatic cells are parenchymal hepatocytes, while the non-parenchymal cells are mainly composed of Kupffer cells, hepatic stellate cells (HSC) and sinusoidal endothelial cells (SECs)[3,4]. Kupffer cells are the resident monocytes in liver, and act in phagocytosis of foreign substances such as microorganisms, as well as management of inflammatory processes. Kupffer cells and infiltrated monocytes and lymphocytes are considered to trigger inflammation in the early phase of hepatitis and then maintain chronic inflammation. HSCs control hepatic and cardiovascular contraction, and produce extracellular matrix (ECM) components and cytokines for repair of organs. HSCs are also believed to have a central role in hepatic fibrosis formation. Hepatocytes fulfill the main functions of the liver, including regulation of nutrition, production of major serum proteins, and elimination of unnecessary materials to maintain homeostasis of the whole body[4].

In the current study, we first examined gene expression profiles of hepatic cells that were isolated at different time points during liver fibrosis. Marker genes specific for different types of hepatic cells were obtained by comparing these profiles and using information from previous studies. DNA microarray data for the whole liver were then interpreted during liver fibrogenesis using the hepatic cell-type specific marker genes. Our results suggest that new pathological properties and intracellular events are associated with each cell type during liver fibrogenesis, and provide candidates for diagnostic markers of liver fibrosis.

MATERIALS AND METHODS

Animals and experimental protocols

Male Sprague-Dawley rats (Charles River Japan, Yokohama, Japan) weighing 160-190 g were housed with unrestricted access to food (CRF-1, Oriental Yeast, Tokyo, Japan) and water in air-conditioned animal quarters with a 12 h light/dark cycle (light between 07:00 and 19:00 h). Hepatic fibrosis was induced by intraperitoneal injection of 0.5% dimethylnitrosamine (DMN; Wako Pure Chemical Industries, Osaka, Japan) at 2 mL/kg of body weight for three consecutive days each week for four weeks. Blood samples were drawn from the inferior vena cava on d 0, 4, 7, 14, 21 and 28, respectively. Liver specimens obtained on these days were dissected and immediately frozen in liquid nitrogen. Fibrosis was confirmed by haematoxylin and eosin (HE) staining of the liver tissue. The hydroxyproline content of liver specimens was determined as previously described by Horie et al[5] . Plasma alanine aminotransferase (ATL) and aspartate aminotransferase (AST) were measured using commercial kits (Fuji Film, Tokyo, Japan), and hyaluronic acid levels were determined using a commercial ELISA kit (Fujirebio, Tokyo, Japan). The animal facilities and protocol were reviewed and approved by the Institutional Animal Care and Use Committee of Ajinomoto Co., Inc.

Preparation of HSCs, inflammatory cells (including Kupffer cells) and hepatocytes

HSCs were isolated from rat liver using the pronase-collagenase digestion method as previously reported[6]. Kupffer cell fraction was prepared with an elutriator, using essentially the same method as previously described[6], and hepatocytes were isolated as previously described[7].

Selection of marker genes for hepatic cells

Hepatic cells (HSCs, Kupffer cell fraction and hepatocytes) were isolated from the liver on d 0, 4, 7, 14, 21 and 28 during liver fibrogenesis. The isolated hepatic cells at each time point were subjected to DNA microarray analysis. If the maximum or minimum expression of a gene in hepatocytes was ten times higher or lower than that of the same gene in HSCs and the Kupffer cell fraction, the up- or down-regulated gene was defined as a hepatocyte-specific marker gene. If the maximum or minimum expression of a gene in HSCs was ten times higher or lower than that of the same gene in hepatocytes and three times higher or lower than the same gene in the Kupffer cell fraction, this gene was defined as a HSC-specific marker gene. Similarly, if the maximum or minimum expression of a gene in the Kupffer cell fraction was ten times higher or lower than that of the same gene in hepatocytes and three times higher or lower than the same gene in HSCs, the gene was defined as an inflammatory cell-specific marker gene. An explanation of the use of the term ‘inflammatory cell’, rather than ‘Kupffer cell’, was given in the Results section. The ratios used to determine cellular specificity were based on previous reports[8,9]: the number of hepatocytes was about ten times higher than that of the inflammatory cell fraction or HSCs, while the number of Kupffer cells was similar to that of HSCs. Use of a high ratio improved the definition of cellular specificity, but also decreased the number of marker genes. However, we found that these ratios were the most appropriate for identification of a set of genes for analysis of liver fibrogenesis.

Microarray analysis

RNAs both from frozen liver tissues and from each isolated cell type were prepared using Isogen reagent (Nippon Gene), and the quality and quantity of each RNA sample were assessed using an Agilent Bioanalyzer 2100 (Agilent Technologies, Inc.). RNA samples were reverse-transcribed with a poly (dT) oligonucleotide attached to a T7 promoter and copied into dsDNA (Invitrogen). In vitro RNA transcription was performed to incorporate biotin-labeled ribonucleotides into the cRNA transcripts using a RNA transcript labeling kit (Enzo Biochem). Some of the RNA (15 g) was utilized for hybridization to a rat genome U34A array (Affymetrix) and a quality assay using test 3 array probe chips was performed according to the manufacturer’s protocol. After hybridization and subsequent washing using the Affymetrix Fluidics Station 400, fluorescence signals amplified with streptavidin phycoerythrin were measured using the Affymetrix scanner, and the results were analyzed using the MicroArray suite software. In the whole liver analysis, two rats and two arrays were used at each time point (d 4, 7, 14, 21 and 28). In the cell-type-specific analysis, one rat which was selected by ALT/AST score and one array were used at each time point (d 4, 7, 14, 21 and 28).

Statistical analysis

Clustering of K-means was performed using the TIGR MeV (MultiExperiment Viewer)[10]. Gene expression profiles in the chronic phase were clustered into 10 patterns using K-means analysis. Clustered genes with a tendency to temporally decrease (clusters 1, 3 and 10) or increase (clusters 7 and 9) were selected as gene markers that had a strong relationship with fibrogenesis.

RESULTS

Time course of gene expression profiles for the whole liver during fibrogenesis

Administration of DMN for three days induced an inflammatory reaction in liver cells simulating the active phase in hepatitis, and the subsequent lack of administration of DMN for four days was used to simulate the remission phase in hepatitis. Repetition of this cycle led to fibrosis in three or four weeks, as shown in Figure 1. AST (GOT) and ALT (GPT) increased on the days of DMN administration and decreased on the days during which DMN was not administered (Figure 1), but both AST and the hyaluronic acid content in serum, a marker of fibrosis[11], gradually increased on days without DMN administration. The inflammatory reaction in periods without DMN administration was weaker than that in periods with DMN-administration, but gradually increased in intensity and response. Based on the behavior shown in Figure 1, gene expression profiles on d 4 (just after a period of DMN administration) were defined as representative of the acute phase response, and those on d 7, 14, 21 and 28 (just before a period of DMN administration) were defined as representative of the chronic phase response.

Figure 1 Histological and biochemical analyses of fibrogenesis. A-C: Histological staining (HE staining) of control liver sections (A) and sections obtained on d 4 (B) and d 21 (C), respectively, after DMN administration; D: Biochemical analysis of fibrogenesis showing plasma AST levels (IU/L, solid line), plasma hyaluronic acid levels (ng/mL, dashed line), and liver hydroxyproline levels (ng/mL, dotted line). The x-axis shows the days of fibrogenesis, and each value on the graph is shown as the mean ± SE, n = 5.

Marker genes indicating fibrotic activity of HSCs

HSCs were isolated from liver at each time point over the time course of development of DMN-induced fibrosis. Marker genes expressed mainly in HSCs were selected from a DNA microarray analysis, as described in the Materials and Methods, and the selected HSC-specific marker genes are listed in Figure 2. In addition, HSCs isolated from normal rats were cultured in vitro for 7 d and DNA microarray analysis of these cells was performed on d 0, 4, and 7, respectively. Marker genes identified in isolated HSCs in the DMN-induced fibrosis model showed the same behavior in the in vitro culture, supporting the HSC specificity of the selected marker genes. The behavior of HSC-specific marker genes in vivo during fibrogenesis was analyzed using DNA microarray data for the whole liver at each time point during development of DMN-induced fibrosis. These data could indicate the actual behavior of the marker genes in vivo, since the data from isolated HSCs might contain some bias due to isolation stimuli. The behavior of HSC-specific marker genes is shown in Figure 2, and supplemental background data are provided in Figure 3. The genes were separated into 2 groups as shown in Figure 2. Group 1 contained marker genes that were linearly up-regulated during fibrogenesis and in the acute inflammation phase, whereas group 2 contained marker genes that were linearly up-regulated during fibrogenesis but not in the acute inflammation phase. Both groups could be further separated into 2 subgroups. Group 1-1 included marker genes that were linearly up-regulated during fibrogenesis and remained in an up-regulated state on d 28, and group 1-2 included marker genes that were also linearly up-regulated during fibrogenesis but then decreased in expression on d 28. Group 2 was similarly separated into groups 2-1 and 2-2. In summary, two groups of HSC-specific marker genes were identified, one in which the genes responded to inflammatory stimuli and the other in which the genes did not respond to such stimuli. Furthermore, HSC-specific marker genes were found that could identify biological changes in HSCs in the late phase of fibrosis.

Figure 2 Marker genes for hepatic stellate cells (HSCs). Expression profiles of the whole liver and in vitro cultured isolated HSCs were obtained using a rat genome U34A array (Affymetrix). Gene markers for HSCs are listed. For analysis of HSCs in the whole liver, expression intensities are given for d 0, and expression intensity data for d 4, 7, 14, 21 and 28 are shown as ratios to the d 0 expression data. The classification of group1 (2) corresponds to the presence (absence) of up-regulated peak in the acute inflammation phase. For analysis of in vitro cultured HSCs, expression levels for d 0 are also shown, and data for d 4 and 7 are similarly shown as ratios to the expression level on d 0. Italicized values indicate an “absent” call by the Affymetrix software. Bold text indicates the highest ratio in the chronic phase (d 7, 14, 21 and 28).

Figure 3 Marker genes in hepatic stellate cells (HSC). Expression profiles of the whole liver, of isolated HSCs during fibrogenesis, and of in vitro cultured isolated HSCs were obtained using a rat Genome U34A Array (Affymetrix). Marker genes for HSCs are listed. Expression intensities are shown for d 0, and expression intensity data for d 4, 7, 14, 21 and 28 are displayed as ratios to the d 0 expression levels for the analysis of whole liver and isolated HSCs. For in vitro cultured HSCs, the d 0 data are similarly shown as expression intensities, and data on d 4 and 7 are shown as ratios to the expression levels on d 0. Italics indicate an “absent” call by the Affymetrix software, and bold text indicates the highest ratio in the chronic phase (d 7, 14, 21 and 28).

Marker genes indicating inflammatory activity in immune cell populations

The Kupffer cell fraction was separated from other hepatic cells at each time point during the course of fibrosis development, and marker genes expressed mainly in the Kupffer cell fraction were selected from the DNA microarray analysis. These marker genes are shown in Figure 4, and supplemental background data are provided in Figure 5. Marker genes in the Kupffer cell fraction indicated the presence of other hematopoietic cells, such as mast cells, lymphocytes, erythrocytes and their progenitors in this fraction, as shown in Figures 4 and 5. Since the source cell population for these marker genes could not be confirmed, a particular hematopoietic cell was postulated to be the source of each marker gene based on previous reports as shown in supplemental Table 1. These data indicated that the method used for isolation of Kupffer cells was not appropriate in fibrotic liver, although it was effective for isolation of normal liver cells. However, despite this drawback, the marker genes could be used to study the behavior of inflammatory cells during fibrogenesis, and these genes were therefore defined as inflammatory cell-specific marker genes. The behavior of the inflammatory cell-specific marker genes in vivo during fibrogenesis was analyzed using DNA microarray data for the whole liver at each time point during DMN-induced fibrogenesis of marker genes, as summarized in Figures 4 and 5.

Table 1 References used in the definition of types of hematopoietic cells in the Kupffer cell fraction.

No.	Reference
1	Tordjmann T, Soulie A, Guettier C, Schmidt M, Berthou C, Beaugrand M, Sasportes M. Perforin and granzyme B lytic protein expression during chronic viral and autoimmune hepatitis. Liver 1998; 18: 391-397
2	Lautenschlager I. Characteristics of the strongly Ia-positive cells in rat liver. Scand J Immunol 1984; 20: 333-338
3	Boisclair J, Dore M, Beauchamp G, Chouinard L, Girard C. Characterization of the inflammatory infiltrate in canine chronic hepatitis. Vet Pathol 2001; 38: 628-635
4	Ahlenstiel G, Woitas RP, Rockstroh J, Spengler U. CC-chemokine receptor 5 (CCR5) in hepatitis C--at the crossroads of the antiviral immune response? J Antimicrob Chemother 2004; 53: 895-898
5	Eis V, Luckow B, Vielhauer V, Siveke JT, Linde Y, Segerer S, De Lema GP, Cohen CD, Kretzler M, Mack M, Horuk R, Murphy PM, Gao JL, Hudkins KL, Alpers CE, Grone HJ, Schlondorff D, Anders HJ. Chemokine receptor CCR1 but not CCR5 mediates leukocyte recruitment and subsequent renal fibrosis after unilateral ureteral obstruction. J Am Soc Nephrol 2004; 15: 337-347
6	McCabe MJ Jr, Dias JA, Lawrence DA. Lead influences translational or posttranslational regulation of Ia expression and increases invariant chain expression in mouse B cells. J Biochem Toxicol 1991; 6: 269-276
7	Wilson KM, Labeta MO, Pawelec G, Fernandez N. Cell-surface expression of human histocompatibility leucocyte antigen (HLA) class II-associated invariant chain (CD74) does not always correlate with cell-surface expression of HLA class II molecules. Immunology 1993; 79: 331-335
8	Kiley SC, Parker PJ. Differential localization of protein kinase C isozymes in U937 cells: evidence for distinct isozyme functions during monocyte differentiation. J Cell Sci 1995; 108 (Pt 3): 1003-1016
9	Allen JN, Liao Z, Moore SA, Wewers MD. Changes in mononuclear phagocyte microtubules after endotoxin stimulation. II. Changes in microtubule composition. Am J Respir Cell Mol Biol 1997; 16: 127-132
10	Devergne O, Peuchmaur M, Humbert M, Navratil E, Leger-Ravet MB, Crevon MC, Petit MA, Galanaud P, Emilie D. In vivo expression of IL-1 beta and IL-6 genes during viral infections in human. Eur Cytokine Netw 1991; 2: 183-194
11	Thornton AJ, Ham J, Kunkel SL. Kupffer cell-derived cytokines induce the synthesis of a leukocyte chemotactic peptide, interleukin-8, in human hepatoma and primary hepatocyte cultures. Hepatology 1991; 14: 1112-1122
12	Manthey CL, Kossmann T, Allen JB, Corcoran ML, Brandes ME, Wahl SM. Role of Kupffer cells in developing streptococcal cell wall granulomas. Streptococcal cell wall induction of inflammatory cytokines and mediators. Am J Pathol 1992; 140: 1205-1214
13	Zhu XL, Zellweger R, Zhu XH, Ayala A, Chaudry IH. Cytokine gene expression in splenic macrophages and Kupffer cells following haemorrhage. Cytokine 1995; 7: 8-14
14	Loike JD, Kaback E, Silverstein SC, Steinberg TH. Lactate transport in macrophages. J Immunol 1993; 150: 1951-1958
15	Daberkow RL, White BR, Cederberg RA, Griffin JB, Zempleni J. Monocarboxylate transporter 1 mediates biotin uptake in human peripheral blood mononuclear cells. J Nutr 2003; 133: 2703-2706
16	Merezhinskaya N, Ogunwuyi SA, Mullick FG, Fishbein WN. Presence and localization of three lactic acid transporters (MCT1, -2, and -4) in separated human granulocytes, lymphocytes, and monocytes. J Histochem Cytochem 2004; 52: 1483-1493
17	Marra F, Valente AJ, Pinzani M, Abboud HE. Cultured human liver fat-storing cells produce monocyte chemotactic protein-1. Regulation by proinflammatory cytokines. J Clin Invest 1993; 92: 1674-1680
18	Guc D, Gulati P, Lemercier C, Lappin D, Birnie GD, Whaley K. Expression of the components and regulatory proteins of the alternative complement pathway and the membrane attack complex in normal and diseased synovium. Rheumatol Int 1993; 13: 139-146
19	Schwaeble W, Huemer HP, Most J, Dierich MP, Strobel M, Claus C, Reid KB, Ziegler-Heitbrock HW. Expression of properdin in human monocytes. Eur J Biochem 1994; 219: 759-764
20	Yamanouchi S, Kuwahara K, Sakata A, Ezaki T, Matsuoka S, Miyazaki J, Hirose S, Tamura T, Nariuchi H, Sakaguchi N. A T cell activation antigen, Ly6C, induced on CD4+ Th1 cells mediates an inhibitory signal for secretion of IL-2 and proliferation in peripheral immune responses. Eur J Immunol 1998; 28: 696-707
21	Manoussaka MS, Smith RJ, Conlin V, Toomey JA, Brooks CG. Fetal mouse NK cell clones are deficient in Ly49 expression, share a common broad lytic specificity, and undergo continuous and extensive diversification in vitro. J Immunol 1998; 160: 2197-2206
22	Wrammert J, Kallberg E, Agace WW, Leanderson T. Ly6C expression differentiates plasma cells from other B cell subsets in mice. Eur J Immunol 2002; 32: 97-103
23	Katz P, Whalen G, Kehrl JH. Differential expression of a novel protein kinase in human B lymphocytes. Preferential localization in the germinal center. J Biol Chem 1994; 269: 16802-16809
24	Valent P, Ashman LK, Hinterberger W, Eckersberger F, Majdic O, Lechner K, Bettelheim P. Mast cell typing: demonstration of a distinct hematopoietic cell type and evidence for immunophenotypic relationship to mononuclear phagocytes. Blood 1989; 73: 1778-1785
25	Titos E, Claria J, Planaguma A, Lopez-Parra M, Villamor N, Parrizas M, Carrio A, Miquel R, Jimenez W, Arroyo V, Rivera F, Rodes J. Inhibition of 5-lipoxygenase induces cell growth arrest and apoptosis in rat Kupffer cells: implications for liver fibrosis. FASEB J 2003; 17: 1745-1747
26	Titos E, Claria J, Planaguma A, Lopez-Parra M, Gonzalez-Periz A, Gaya J, Miquel R, Arroyo V, Rodes J. Inhibition of 5-lipoxygenase-activating protein abrogates experimental liver injury: role of Kupffer cells. J Leukoc Biol 2005; 78: 871-878
27	Sawano A, Iwai S, Sakurai Y, Ito M, Shitara K, Nakahata T, Shibuya M. Flt-1, vascular endothelial growth factor receptor 1, is a novel cell surface marker for the lineage of monocyte-macrophages in humans. Blood 2001; 97: 785-791
28	Coppola S, Narciso L, Feccia T, Bonci D, Calabro L, Morsilli O, Gabbianelli M, De Maria R, Testa U, Peschle C. Enforced expression of KDR receptor promotes proliferation, survival and megakaryocytic differentiation of TF1 progenitor cell line. Cell Death Differ 2006; 13: 61-74
29	Fernandez Pujol B, Lucibello FC, Zuzarte M, Lutjens P, Muller R, Havemann K. Dendritic cells derived from peripheral monocytes express endothelial markers and in the presence of angiogenic growth factors differentiate into endothelial-like cells. Eur J Cell Biol 2001; 80: 99-110
30	Bowling WM, Flye MW, Qiu YY, Callery MP. Inhibition of phosphatidylinositol-3'-kinase prevents induction of endotoxin tolerance in vitro. J Surg Res 1996; 63: 287-292
31	Capodici C, Hanft S, Feoktistov M, Pillinger MH. Phosphatidylinositol 3-kinase mediates chemoattractant-stimulated, CD11b/CD18-dependent cell-cell adhesion of human neutrophils: evidence for an ERK-independent pathway. J Immunol 1998; 160: 1901-1909
32	Bracke M, Nijhuis E, Lammers JW, Coffer PJ, Koenderman L. A critical role for PI 3-kinase in cytokine-induced Fcalpha-receptor activation. Blood 2000; 95: 2037-2043
33	Zweifel M, Breu K, Matozan K, Renner E, Welle M, Schaffner T, Clavien PA. Restoration of hepatic mast cells and expression of a different mast cell protease phenotype in regenerating rat liver after 70%-hepatectomy. Immunol Cell Biol 2005; 83: 587-595
34	Pemberton AD, Brown JK, Wright SH, Knight PA, Miller HR. The proteome of mouse mucosal mast cell homologues: the role of transforming growth factor beta1. Proteomics 2006; 6: 623-631
35	Schwartz LB. Analysis of MC(T) and MC(TC) mast cells in tissue. Methods Mol Biol 2006; 315: 53-62
36	Henningsson F, Yamamoto K, Saftig P, Reinheckel T, Peters C, Knight SD, Pejler G. A role for cathepsin E in the processing of mast-cell carboxypeptidase A. J Cell Sci 2005; 118: 2035-2042
37	Chen ZQ, He SH. Cloning and expression of human colon mast cell carboxypeptidase. World J Gastroenterol 2004; 10: 342-347
38	Lagasse E, Clerc RG. Cloning and expression of two human genes encoding calcium-binding proteins that are regulated during myeloid differentiation. Mol Cell Biol 1988; 8: 2402-2410
39	Bhardwaj RS, Zotz C, Zwadlo-Klarwasser G, Roth J, Goebeler M, Mahnke K, Falk M, Meinardus-Hager G, Sorg C. The calcium-binding proteins MRP8 and MRP14 form a membrane-associated heterodimer in a subset of monocytes/macrophages present in acute but absent in chronic inflammatory lesions. Eur J Immunol 1992; 22: 1891-1897
40	Ryckman C, Vandal K, Rouleau P, Talbot M, Tessier PA. Proinflammatory activities of S100: proteins S100A8, S100A9, and S100A8/A9 induce neutrophil chemotaxis and adhesion. J Immunol 2003; 170: 3233-3242
41	Yoshimoto T, Suzuki H, Yamamoto S, Takai T, Yokoyama C, Tanabe T. Cloning and expression of arachidonate 12-lipoxygenase cDNA from porcine leukocytes. Adv Prostaglandin Thromboxane Leukot Res 1991; 21A: 29-32
42	Gu JL, Natarajan R, Ben-Ezra J, Valente G, Scott S, Yoshimoto T, Yamamoto S, Rossi JJ, Nadler JL. Evidence that a leukocyte type of 12-lipoxygenase is expressed and regulated by angiotensin II in human adrenal glomerulosa cells. Endocrinology 1994; 134: 70-77
43	Nakamura M, Ueda N, Kishimoto K, Yoshimoto T, Yamamoto S, Ishimura K. Immunocytochemical localization of platelet-type arachidonate 12-lipoxygenase in mouse blood cells. J Histochem Cytochem 1995; 43: 237-244
44	Kaminski WE, Jendraschak E, Baumann K, Kiefl R, Fischer S, Marcus AJ, Broekman MJ, von Schacky C. Human mononuclear cells express 12-LX: coordinated mRNA regulation with 5-LX and FLAP genes. Blood 1996; 87: 331-340
45	Takenaga K, Nakamura Y, Sakiyama S. Expression of a calcium binding protein pEL98 (mts1) during differentiation of human promyelocytic leukemia HL-60 cells. Biochem Biophys Res Commun 1994; 202: 94-101
46	Yancopoulos GD, Oltz EM, Rathbun G, Berman JE, Smith RK, Lansford RD, Rothman P, Okada A, Lee G, Morrow M. Isolation of coordinately regulated genes that are expressed in discrete stages of B-cell development. Proc Natl Acad Sci USA 1990; 87: 5759-5763
47	Quelo I, Jurdic P. Differential regulation of the carbonic anhydrase II gene expression by hormonal nuclear receptors in monocytic cells: identification of the retinoic acid response element. Biochem Biophys Res Commun 2000; 271: 481-491
48	Baron WF, Pan CQ, Spencer SA, Ryan AM, Lazarus RA, Baker KP. Cloning and characterization of an actin-resistant DNase I-like endonuclease secreted by macrophages. Gene 1998; 215: 291-301
49	Kitabatake Y, Kawamura S, Yamashita M, Okuyama K, Takayanagi M, Ohno I. The expression of mRNA for calcitonin gene-related peptide receptors in a mucosal type mast cell line, RBL-2H3. Biol Pharm Bull 2004; 27: 896-898
50	Makino Y, Nakamura H, Ikeda E, Ohnuma K, Yamauchi K, Yabe Y, Poellinger L, Okada Y, Morimoto C, Tanaka H. Hypoxia-inducible factor regulates survival of antigen receptor-driven T cells. J Immunol 2003; 171: 6534-6540
51	Frank M. MAL, a proteolipid in glycosphingolipid enriched domains: functional implications in myelin and beyond. Prog Neurobiol 2000; 60: 531-544

The reference numbers correspond to the numbers in Figure 5 in the supplementary data.

Figure 4 Marker genes for hematopoietic cells in the Kupffer cell fraction. Expression profi les of the whole liver and the isolated Kupffer cell fraction during fi brogenesis were obtained using a rat genome U34A array (Affymetrix). Marker genes for hematopoietic cells in the Kupffer cell fraction are listed. Expression intensities are given for d 0, and expression intensity data for d 4, 7,14, 21 and 28 are shown as ratios to the d 0 expression data for analysis of hematopoietic cells in whole liver. The classifi cation of genes into groups 1-4 and corresponds to the maximum change of expression at the different time points on d 7, 14, 21, 28 and 4, respectively. Italicized values indicate an “absent” call by the Affymetrix software. The bold values indicate the highest or lowest ratio in the chronic phase (d 7, 14, 21 and 28). Cell types in the infl ammatory cell fraction are as follows: M: Monocytes and their progenitors; L: Lymphocytes and their progenitors; E: Erythrocytes and their progenitors; P: Platelets and their progenitors; Leu: Other kinds of leukocytes and their progenitors.

Figure 5 Marker genes in hematopoietic cells in the Kupffer cell fraction. Expression profiles of the whole liver and the isolated Kupffer cell fraction during fibrogenesis were obtained using a rat Genome U34A Array (Affymetrix). Marker genes of hematopoietic cells in the Kupffer cell fraction are listed. Expression intensities are shown for d 0, and the expression intensity data for d 4, 7, 14, 21 and 28 are displayed as ratios to the d 0 expression levels. Italics indicate an “absent” call by the Affymetrix software. Bold values indicate the highest or lowest ratio in the chronic phase (d 7, 14, 21 and 28). The cell types in the infl ammatory cell fraction are as follows: M: Monocytes and their progenitors; L: Lymphocytes and their progenitors; E: Erythrocytes and their progenitors; P: Platelets and their progenitors; Leu: Other kinds of leukocytes and their progenitors.

The acute phase response was followed by an immunological response, based on the increase in expression of Kupffer cell (or macrophage) markers such as Lyz, Gzmb, and Il1b, as well as surface markers of T cells, such as T-cell receptor, Il2rb, Cd8, Cd76, and Cd45. The up-regulated expression of these genes seemed to indicate activation of Kupffer cells and T lymphocytes, as well as activation of the interaction between these cells, around d 7. Temporary up-regulation of mast cell markers such as chemokines and mast cell proteases indicated the invasion and/or activation of mast cells around d 14, which is of interest since mast cells are known not only to cause acute inflammation, but also to have a role in the induction of chronic inflammation[12-14]. However, whether activation of mast cells is essential for liver fibrosis is unknown. Peak expression of B cell markers such as immunoglobulin occurred on d 21 or d 28, and therefore activation or invasion of B cells seemed to reach its peak in the late phase. Overall, these results showed that inflammatory cell-specific marker genes could be used to monitor the transition of active inflammatory cell populations in fibrosis, and this sequential activation or invasion of inflammatory cells might be related to the stage of fibrotic progression.

Marker genes indicating damage to hepatocytes

Hepatocytes were separated from other hepatic cells at each time point during the course of DMN-induced fibrogenesis, and marker genes expressed mainly in hepatocytes were selected by DNA microarray analysis. The behavior of hepatocyte-specific marker genes in vivo during fibrogenesis was analyzed using DNA microarray data for the whole liver at each time point during fibrosis development. Marker genes were categorized based on their functions, as shown in Figure 6. Many abnormally expressed genes were identified and temporal analysis revealed groups of genes showing consistent variation in expression. Gene expression profiles in the chronic phase (d 7, 14, 21 and 28) were clustered into 10 patterns using K-means analysis. Clustered genes with a tendency to temporally decrease (clusters 1, 3 and 10) or increase (clusters 7 and 9) were selected, and then genes were further selected based on a strong correlation coefficient (≥ 0.7 or ≤ -0.7) with fibrosis stage (that is, d 7, 14, 21 and 28). The time courses of expression ratios are shown in Figure 7 as supplemental data. The 41 down-regulated and 19 up-regulated genes that were finally selected are shown in Table 2. These genes appeared to have a strong relationship with progression of fibrosis, and might also share common regulatory expression mechanisms.

Table 2 Marker genes for hepatocytes that showed a strong relationship with fibrogenesis.

Functional category	Cluster No.	Direction of change	Annotation	Symbol	Common
Amino acid metabolism	7	Increase	Glutamate oxaloacetate transaminase 1	Got1	J04171
	7	Increase	Glutamic-pyruvate transaminase (alanine aminotransferase)	Gpt	D10354
	9	Increase	Glutathione-S-transferase, alpha type2	Gsta2	AA945082
	3	Decrease	Cytosolic cysteine dioxygenase 1	Cdo1	AA942685
	10	Decrease	Cysteine-sulfinate decarboxylase	Csad	M64755
Cholesterol synthesis	3	Decrease	Nuclear receptor subfamily 0, group B, member 2	Nr0b2	D86745
Steroid hormone synthesis	7	Increase	Hydroxysteroid (17-beta) dehydrogenase 9	Hsd17b9	U89280
	7	Increase	(20-alpha-hydroxysteroid dehydrogenase)	---	AA866264
	1	Decrease	Hydroxysteroid (17-beta) dehydrogenase 2	Hsd17b2	X91234
	3	Decrease	Sulfotransferase, estrogen preferring	Ste	S76489
Lipid biosynthesis, metabolism, fatty acid & lipid transport	9	Increase	Carboxylesterase 2 (intestine, liver) (drug metabolism)	Ces2	AB010635
	1	Decease	2-hydroxyphytanoyl-CoA lyase(peroxisomal) alpha-oxidation	Hpcl2	AA893239
	3	Decease	Carboxylesterase 3	Ces3	L46791
	3	Decease	Hydroxyacid oxidase 2 (long chain)(peroxisomal)( alpha-oxidation)	Hao2	AI232087
	3	Decease	Fatty acid binding protein 7, brain (cytosolic)	Fabp7	U02096
	10	Decease	Alpha-methylacyl-CoA racemase Peroxisomal)	Amacr	U89905
Retinoid synthesis & metabolism	3	Decease	Retinol dehydrogenase type II (RODH II)	RoDHII	U33500
	10	Decease	Retinol dehydrogenase type III	Rdh3	U18762
Dlucolysis & gluconeogenesis	7	Increase	Glucokinase	Gck	X53588
	3	Decrease	Phosphoenolpyruvate carboxykinase 1(PEPCK1) (cytosolic)	Pck1	K03243
	3	Decrease	Pyruvate carboxylase	Pc	U32314
	3	Decrease	Solute carrier family 37 (glycerol-6-phosphate transporter), member 4	Slc37a4	AF080468
	3	Decrease	Ribonuclease, RNase A family 4	Rnase4	AF041066
	3	Decrease	Ectonucleotide pyrophosphatase/phosphodiesterase 2(lysophospholipaseD)	Enpp2	D28560
Drug-metabolism	7	Increase	Cytochrome P450, family 3, subfamily a, polypeptide 13	Cyp3a13	U46118
	9	Increase	Aldehyde dehydrogenase family 1, member A1	Aldh1a1	AF001898
	1	Decrease	Cytochrome P450, family 1, subfamily a, polypeptide 2	Cyp1a2	E01184
	1	Decrease	Cytochrome P450, subfamily IIC (mephenytoin 4-hydroxylase)	Cyp2c	X79081
	1	Decrease	Flavin containing monooxygenase 1	Fmo1	M84719
	3	Decrease	Cytochrome P450, family 17, subfamily a, polypeptide 1	Cyp17a1	M21208
	3	Decrease	Cytochrome P450, family 3, subfamily a, polypeptide 11	Cyp3a11	M13646
	3	Decrease	Alcohol dehydrogenase 4 (class II), pi polypeptide	Adh4	X90710
	3	Decrease	3-hydroxybutyrate dehydrogenase (heart, mitochondrial)	Bdh	AA817846
	3	Decrease	Camello-like 1 N-acetyltransferase 8, NAT8)	Cml1	AA892382
	3	Decrease	(hydroxysteroid sulfotransferase )	---	M31363
	3	Decrease	(hydroxysteroid sulfotransferase subunit)	---	AA818122
Growth factors & their receptors	7	Increase	Lipopolysaccharide binding protein	Lbp	L32132
	9	Decrease	Insulin-like growth factor binding protein 2	Igfbp2	J04486
	3	Decrease	Activin beta E	Inhbe	AF089825
	10	Decrease	Growth hormone receptor	Ghr	S49003
	10	Decrease	Insulin-like growth factor binding protein 1	Igfbp1	M58634
	10	Decrease	Deiodinase, iodothyronine, type I	Dio1	X57999
Hepatic secretory proteins	7	Increase	Alpha-fetoprotein	Afp	X02361
	1	Decrease	Alpha-2u globulin PGCL4 /// alpha-2u globulin PGCL2 /// alpha-2u-globulin (L type)/// alpha-2u globulin PGCL1 /// alpha-2u globulin PGCL3	Obp3 /// LOC298109 /// LOC298116 /// LOC259246 /// LOC259244	M27434
	1	Decrease	Alpha-2u globulin PGCL4 (Ppp2r2a protein phosphatase 2 (formerly 2A), regulatory subunit B (PR 52), alpha isoform)	Obp3	J00738
	10	Decrease	Metallothionein	Mt1a	AI102562
Protease & protease inhibitor	9	Increase	Lipocalin 2	Lcn2	AA946503
	1	Decrease	Elastase 1, pancreatic	Ela1	L00117
	10	Decrease	Esterase 2 (liver carboxylesterase)	Es2	M20629
	3	Decrease	Serine protease inhibitor	Spin2a	D00752
Cell surface proteins & structural proteins	7	Increase	Keratin complex 2, basic, gene 8 (cytokeratin-8)	Krt2-8	S76054
	7	Increase	Similar to cytokeratin(keratin complex 1, acidic, gene 18)	(Krt1-18)	AI072634
Signal transduction	3	Decrease	Regucalcin	Rgn	D31662
Transporters	7	Increase	solute carrier family 17 (sodium phosphate), member 1	Slc17a1	U28504
	7	Increase	Aquaporin 7	Aqp7	AB000507
	9	Increase	ATP-binding cassette, sub-family B (MDR/TAP), member 1	Abcb1	M81855
	3	Decrease	Solute carrier family 10 (sodium/bile acid cotransporter family), member 1	Slc10a1	M77479
Protein related vitamins	10	Decrease	L-gulono-gamma-lactone oxidase	Gulo	D14564
Others	7	Increase	Rabphilin 3A-like (without C2 domains)	Rph3al	AF022774
	1	Decrease	Carbonic anhydrase 3	Ca3	AF037072

Gene expression profiles in the chronic phase (d 7, 14, 21 and 28) were clustered into 10 patterns using K-means analysis. Clustered genes that showed a tendency to temporally decrease (clusters 1, 3 and 10) or increase (clusters 7 and 9) were selected, as shown in Table 1 in the supplemental data. Among these genes, those showing a strong relation with fibrogenesis were further selected based on a t-test statistical analysis of the rate of change in expression intensity and the number of days of fibrosis. Hepatocyte-specific gene markers showing a strong relation with fibrogenesis are listed.

Figure 6 Marker genes for hepatocytes. Expression profiles of whole liver or isolated hepatocytes during fibrogenesis were obtained using a rat Genome U34A Array (Affymetrix). Marker genes for hepatocytes, which are the main contributors to the expression profile of the whole liver, are listed. Expression intensities are given for d 0, and expression intensity data for d 4, 7, 14, 21 and 28 are displayed as ratios to the d 0 expression data. Italics indicate an “absent” call by the Affymetrix software. Genes with the highest or lowest ratio in the chronic phase are shown in bold text.

Figure 7 Cluster of genes that increased or decreased in expression with progression of fibrogenesis. Gene expression profiles in the chronic phase (d 7, 14, 21 and 28) were clustered into 10 patterns using K-means analysis. Clustered genes with a tendency to temporally decrease (clusters 1, 3 and 10) or increase (clusters 7 and 9) were selected as gene markers that had a strong relationship with fibrogenesis.

Down-regulation of the expression of Cdo1 and Csad showed a strong relationship with fibrotic stage, while expression of Gsta2 was simultaneously up-regulated. Down-regulation of other metabolic enzymes in sulfur-containing metabolic pathways is also shown in Figure 6. These data suggested a broad range of abnormalities in sulfur-containing amino acid metabolic pathways in fibrosis. Regarding other genes, Hao2 and Hpcl2 had a role in fatty acid oxidation in peroxisome, while Amacr was associated with beta-oxidation of pristanoyl-CoA and C27-bile acyl-CoAs. The down-regulation of these genes and other metabolic enzymes related to fatty acid oxidation (Figure 6) suggested abnormalities in the fatty acid oxidation process in fibrosis. The expression of Amacr might be related to that of the nuclear receptor subfamily 0, group B, member 2 (Nr0b2), and changes in NrOb2 expression may affect one of the key molecules in cholesterol biosynthesis. Up-regulation of Gk and down-regulation of Pepck1, PC and Slc37a4 suggested abnormalities in gluconeogenesis, glycogen storage and glycolysis, while down-regulation of Ste and Hsd17b2 suggested an abnormality of estrogen metabolism in the liver. Decreased expression of many Cyp drug metabolism enzymes was also found in progression of fibrosis, and abnormalities in hormonal signaling were suggested by the down-regulation of Inhbe, Ghr and Dio1. These results showed that identification of hepatocyte-specific marker genes could allow analysis of functional changes in fibrosis, and all the identified abnormalities might have major effects on hepatic function.

DISCUSSION

Marker genes for HSCs

Markers of HSCs such as Acta 2, Cryab, Spp1, Prnp, and Pai-1 were strongly up-regulated on d 4, and then quickly decreased in expression following a gradual up-regulation. On the other hand, other HSC markers such as Gpc3, Lox, and Mgp did not show marked up-regulation on d 4, but their expression level increased linearly during fibrogenesis. Therefore, HSCs may be associated with events in two different phases, such as repair in the acute phase response and fibrosis in the chronic phase response. These two groups of HSC marker genes may also be differentially regulated. Marker genes in one group may be expressed both in undifferentiated and in differentiated HSCs, while marker genes in the other group may be expressed mainly in differentiated HSCs. The functional changes associated with differentiation of HSCs during fibrosis are not clear. Schnabel et al[15] have temporally divided the activation process of HSCs into an initiation phase and a perpetuation phase, and HSCs in the initiation phase may play a role in wound healing[16,17] and are then eliminated by apoptosis[18,19], although some HSCs in the initiation phase differentiate into cells in the perpetuation phase. The two kinds of marker genes found in our work may be associated with the two phases.

In the present study, gene expression in HSCs in vivo was mostly similar to that found in vitro, suggesting that HSCs can be activated and produce ECM with few factors contributed by other cell types. Since HSCs also produce auto-stimulating factors such as TGF-beta, chemokines, PDGF, and IGF-1, our results strongly suggest that HSCs have self-supporting properties and few exogenous factors are required for their activation and differentiation. Even if inflammation induced by virus activation stimulates long-term or weak hepatitis, HSCs may be able to autonomously activate and promote fibrosis, and this property of HSCs may be central to promotion of fibrogenesis. Sancho-Bru et al[20] have recently reported that a culture model of HSCs could not exactly match the activated phenotype found in DNA microarray analysis of isolated HSCs from cirrhotic human livers, because in culture marker genes for HSCs in the perpetuation phase are predominantly expressed, relative to those in the initiation phase. However, isolation of HSCs may also alter the expression of some genes, especially the expression of genes associated with inflammation, as shown in Figures 2 and 4. Therefore, compared to studies of isolated HSCs, our approach reveals the actual behavior of HSCs in vivo during fibrosis.

Gpc3 has been recently proposed as a serum and histochemical marker for hepatocellular carcinoma[21,22], since it is only weakly expressed in hepatocytes in normal and cirrhotic livers. However, in our study, Gpc3 was expressed in isolated HSCs during fibrogenesis, but weakly in cultured HSCs, suggesting that HSCs require extracellular factors for expression of Gpc3 during fibrogenesis, or that Gpc3-expressing cells with abnormal characteristics may contaminate the HSC fraction. The cell type showing expression of Gpc3 requires further study.

Marker genes for inflammatory cells

In the present study, gene expression in the inflammatory cell-fraction indicated the presence of several kinds of hematopoietic cells in this fraction, leading to some uncertainty in the data from the rat fibrosis model. However, the behavior of mast cells is of note, since temporary up-regulation of mast cell markers such as chemokines and mast cell proteases indicated invasion and/or activation of mast cells around d 14. Invasion of a marked number of mast cells could not be detected with HE staining, and therefore the number of invading mast cells must be small. Mast cells not only cause acute inflammation, but also have a role in induction of chronic inflammation[12,14], and involvement of mast cells in hepatic fibrosis has been reported[12,23-25]. RANTES, a chemokine that is produced by T cells and stimulates mast cells[26,27], showed its peak expression on d 14, as shown in Figure 4. RANTES has been suggested to be a mediator of progression from acute to chronic inflammation in colitis[28], and further studies are of importance to determine whether RANTES activation of mast cells is essential for liver fibrosis.

Marker genes for hepatocytes

DNA microarray analysis of whole liver in experimental animal models of hepatic fibrosis has been reported[1,2], and characteristic behavior of hepatocyte-specific marker genes over the time course of fibrogenesis was found in this study. DNA microarray data for the whole liver are similar to those for hepatocytes, since 70% of hepatic cells are hepatocytes[8,9], and data on d 28 are generally similar to those in previous reports. However, our temporal data indicate progressive abnormal gene expression in fibrogenesis, including in the early phase of fibrogenesis. Furthermore, since our data did not contain genes expressed in other hepatic cells, the hepatocyte-specific gene set allowed examination of the molecular network in hepatocytes.

Clustered abnormalities were found in genes associated with metabolism of sulfur-containing amino acids in this study. Similar abnormalities in metabolism of sulfur-containing amino acids have been reported[29,30], and an increase in methionine concentration in blood has been found in cirrhosis[31,32], which may be related to changes in gene expression in sulfur-containing amino acid metabolism. Furthermore, S-adenosylmethionine has an important role in methylation, including the methylation of DNA. Since abnormalities likely induce tumorigenesis due to DNA instability, it is of note that long-term suppression of S-adenosylmethionine synthetase increases the risk of tumorigenesis[33]. Glycine methyltransferase also has been implicated in DNA instability[34,35], and long-term suppression of this enzyme also has an associated risk of tumorigenesis during liver fibrosis. A metabolite of methionine, homocysteine, is also suspected as a risk factor for cardiovascular disease[36,37], and an increase in the concentration of homocysteine has been found in cirrhosis[38]. Abnormal homocysteine metabolism may also have an important role in the pathogenesis of liver failure, including fatty liver, activation of HSCs (i.e., enhancement of fibrosis), cardiovascular disease, and HCC. Finally, biosynthesis of taurine from cysteine may be suppressed, possibly leading to painful muscle cramps, which are a complication of cirrhosis caused by taurine deficiency[39]. Therefore, in summary, it is apparent that abnormalities in sulfur-containing amino acid metabolism can result in development of serious diseases.

Down-regulation of enzymes related to beta-oxidation in TAA-induced experimental fibrosis has been reported[29]. Studies of the Aox knockout mouse[40] suggest that metabolites of fatty acids in beta oxidation accumulate in the liver and stimulate PPAR-alpha, resulting in peroxisome proliferation, and tumorigenesis is a potential risk in long-term administration of PPAR-alpha agonists and in the Aox-1 knockout mouse[40,41]. Radical accumulation by blocking metabolic enzymes associated with beta-oxidation has been suggested as one explanation of tumorigenesis in Aox knockout mice, and furthermore, microarray analysis of Aox-deficient mice can show up-regulation of Lcn2, a marker of carcinogenesis[42]. Interestingly markers such as Lcn2 and Cd36 were up-regulated in hepatocytes during fibrogenesis and in the inflammatory cell fraction in the late phase in our model. Although tumorigenesis due to PPAR alpha agonist is thought not to occur in humans[41,43], abnormalities in lipid metabolism during liver fibrosis may have a role in steatofibrosis and/or enhancement of fibrosis and HCC.

Up-regulation of Gck and down-regulation of Pck1 in hepatocytes suggest a decrease in gluconeogenesis, and suppression of gluconeogenesis has been reported in cirrhosis[44]. A deficiency of Scl37a4 enzyme activity in humans causes glycogen storage disease type 1 (GSD-1) genetic disorders[45] and suppression of this molecule is associated with hepatic steatosis[46]. Abnormalities in metabolism and synthesis of sex hormone have also been found in fibrotic liver, and an increased ratio of estrogen to testosterone in serum induces feminization, which is a complication in cirrhotic males[47]. Abnormality of sex-hormone metabolism is also related to liver malignancies[48,49].

Down-regulation of Ghr is also found in hepatocytes, and long-term suppression of growth signals may greatly influence fundamental hepatic vitality and produce abnormal hepatic regeneration. Insensitivity to growth hormones in cirrhosis has been reported[50,51], and administration of growth hormone protects against experimental liver fibrosis[52]. Rgn, a regulator of calcium signaling, may have an important role in regulation of proliferation and apoptosis of hepatocytes, as well as in formation of HCC[53,54], and down-regulation of regucalcin in our model suggested a risk of HCC development. An association of Afp, Ste, Mt1a, Lcn2, Abcb1 Cml1 (Nat8) and Ca3 with hepatocellular carcinoma (HCC) formation has also been reported. Accumulation of estrogen in the liver is suspected to promote HCC[55,56], and NAT8 polymorphism may be related to HCC[57,58]. Down-regulation of metallothionein[59-61] and carbonic anhydrase[62,63], up-regulation of Mdr/Tap[64] and Lcn[42,65] occur in HCC, and proliferation of hepatoma cells is suppressed by over-expression of Rgn[53,66]. The abnormal expression of all these genes in our model is similar to that in HCC, suggesting the importance of understanding whether such changes in gene expression reflect a tumorigenic environment or even promote tumorigenesis.

Differential regulations of genes involved in key events of liver fibrosis

In addition to the expression profiles of cell type specific marker genes discussed above, we here describe how genes involved in key events of liver fibrogenesis are differentially regulated. Gene expression profiles of the whole liver for different functional categories in liver fibrogenesis, such as ECM synthesis/degradation, inflammation and oxidative stress, are shown in Figure 8. Since both synthesis and degradation of ECM occur simultaneously, both genes are put together. Figure 8 shows that most genes in each category have a common and mutually correlated expression pattern, showing different regulations for different categories. Most genes of ECM synthesis/degradation as shown in Figure 8A have a peak of up-regulation on d 4 and a following gradual up-regulation along with the progression of fibrosis, and interestingly these genes are classified into group 1 of the HSC -specific genes as shown in Figure 2. Another type of genes of ECM synthesis/degradation as shown in Figure 8B have no peak on d 4 and only a gradual up-regulation along with the progression of fibrosis, and are classified into group 2 of the HSC-specific genes as shown in Figure 2. Many genes are also involved in the inflammatory category, and here only some of them are plotted as representative in Figure 8C. Most genes of inflammation have a peak on d 4 or 7 commonly, and are classified in group 1 of the Kupffer cell fraction specfic genes as shown in Figure 4. Most genes in the catregory of oxidative stress in Figure 8D have a minimum expression on d 4, in contrast to the genes of inflammation having a peak on d 4 or 7, and are involved in hepatocyte specific genes as shown in Figure 6.

Figure 8 Differential regulations of genes involved in key events of liver fibrosis. Gene expression profiles of different events in liver fibrogenesis such as ECM synthesis/degradation, inflammation and oxidative stress, are shown. The x-axis showing the days of fibrogenesis (d 0, 4, 7 14, 21, 28) and the y-axis the normalized log ratio(scaled in terms of mean and SD, and the log base 2) of the whole liver gene expression. A and B: genes of ECM synthesis/degradation classified into group 1 (2) of HSCs as shown in Figure 2; C: genes of inflammation involved in the Kupffer cell fraction as shown in Figure 4; D: genes of oxidative stress involved in the hepatocytes as shown in Figure 6.

Gene expression profiles viewed from both cell types and functional categories have made more clear image on how the temporal expression pattern are closely associated in terms of both the cell specificity and functions in liver fibrosis, and are regulated differently in different categories but in a mutually correlated manner within the same category.

Pathological overview of the behavior of HSCs, inflammatory cells, and hepatocytes in fibrogenesis

Our results from gene-expression profiling using hepatic cell-specific marker genes support the hypothesis shown in Figure 9. Comparisons of gene expression in HSCs in vivo and in vitro strongly suggest that HSCs have self-supporting properties and that few exogenous molecules are required to activate and differentiate HSCs. Hepatocytes have been shown to suffer from serious stress during fibrogenesis, and signals from suppressed hepatocytes, such as radicals, prophlogistic substances and toxic metabolites due to abnormal metabolism, are able to stimulate Kupffer cells and HSCs, leading to subsequent production of HSC-stimulating-factors such as TNF-α and IL-1 by Kupffer cells. Therefore, HSCs, Kupffer cells and other inflammatory cells produce factors such as TGF-beta that suppress hepatocyte vitality and result in hepatocyte injury. Sequential activation of inflammatory cells such as lymphocytes and mast cells may be essential in this process, and even if this stimulatory circuit is small in scale during early remission of fibrogenesis, it can be maintained with appropriate stimulation even at long intervals, with small-scale inflammation induced by C-type hepatitis virus propagation. The self-supporting characteristics of HSCs may have a central role in maintenance of this circuit.

Figure 9 Circuit model of hepatic cells in fibrogenesis. Gene expression profiles show that HSCs have self-activating properties, and that widespread damage to hepatocytes occurred in development of fibrosis, suggesting a self-activating circuit model of fibrogenesis. After an initial stimulatory trigger caused by events such as virus infection, hepatic cells are able to stimulate each other, and the self-activating properties of HSCs maintain this cycle over the long term. Details are given in the text.

ACKNOWLEDGMENTS

We thank Ms. Sonoko Ishizaki of Ajinomoto Co., for preparation of Kupffer cells. Additionally, we thank Mr. Kazuhisa Takemoto and Ms. Kunie Shimode for experimental assistance and manuscript preparation, respectively.