Interstrain differences in chronic hepatitis and tumor development in a murine model of inflammation-mediated hepatocarcinogenesis

Authors


  • Potential conflict of interest: Nothing to report.

  • This study was supported by the Kamea Scientific Foundation of the Israeli Government (to Tamara Potikha and Daniel Goldenberg), by the Chief Scientist of the Israeli Ministry of Health (through grant 4930 to Daniel Goldenberg), by the German Research Foundation through SFB841 projects C2 (to Gabriele Sass and Gisa Tiegs) and C3 (to Eithan Galun), and by the Argentinean Agency for Science and Technology and Sales Foundation for Cancer (to Gabriel A. Rabinovich).

Abstract

Chronic inflammation is strongly associated with an increased risk for hepatocellular carcinoma (HCC) development. The multidrug resistance 2 (Mdr2)–knockout (KO) mouse (adenosine triphosphate–binding cassette b4−/−), a model of inflammation-mediated HCC, develops chronic cholestatic hepatitis at an early age and HCC at an adult age. To delineate factors contributing to hepatocarcinogenesis, we compared the severity of early chronic hepatitis and late HCC development in two Mdr2-KO strains: Friend virus B-type/N (FVB) and C57 black 6 (B6). We demonstrated that hepatocarcinogenesis was significantly less efficient in the Mdr2-KO/B6 mice versus the Mdr2-KO/FVB mice; this difference was more prominent in males. Chronic hepatitis in the Mdr2-KO/B6 males was more severe at 1 month of age but was less severe at 3 months of age in comparison with age-matched Mdr2-KO/FVB males. A comparative genome-scale gene expression analysis of male livers of both strains at 3 months of age revealed both common and strain-specific aberrantly expressed genes, including genes associated with the regulation of inflammation, the response to oxidative stress, and lipid metabolism. One of these regulators, galectin-1 (Gal-1), possesses both anti-inflammatory and protumorigenic activities. To study its regulatory role in the liver, we transferred the Gal-1–KO mutation (lectin galactoside-binding soluble 1−/−) from the B6 strain to the FVB strain, and we demonstrated that endogenous Gal-1 protected the liver against concanavalin A–induced hepatitis with the B6 genetic background but not the FVB genetic background. Conclusion: Decreased chronic hepatitis in Mdr2-KO/B6 mice at the age of 3 months correlated with a significant retardation of liver tumor development in this strain versus the Mdr2-KO/FVB strain. We found candidate factors that may determine strain-specific differences in the course of chronic hepatitis and HCC development in the Mdr2-KO model, including inefficient anti-inflammatory activity of the endogenous lectin Gal-1 in the FVB strain. (HEPATOLOGY 2013 )

Hepatocellular carcinoma (HCC) is one of the most prevalent and fatal neoplasms worldwide, and therapeutic options are limited. Chronic inflammation precedes the majority of HCC cases. The underlying mechanisms by which inflammation causes HCC development are not well understood. We investigated the role of inflammation in HCC development with the multidrug resistance 2 (Mdr2)–knockout (KO) mouse as a model. These mice lack the Mdr2 [adenosine triphosphate–binding cassette b4 (Abcb4)] P-glycoprotein responsible for phosphatidylcholine transport across the canalicular membrane; they develop chronic cholestatic hepatitis at an early age and HCC at a later age.1 Importantly, by analyzing gene expression profiles of liver tumors, we demonstrated that Mdr2-KO mice of the Friend virus B-type/N (FVB) genetic background (Mdr2-KO/FVB) share many deregulated pathways and differentially expressed genes with human HCC data sets.2 Previously, a critical role of nuclear factor kappa B (NF-κB) signaling for liver tumor development in inflammation-associated HCC was shown.3 We revealed that during the early precancerous stage, different protective mechanisms, including multiple anti-inflammatory and anti-oxidant genes, were induced in the livers of Mdr2-KO/FVB mice.4 A direct connection between chronic hepatitis at an early stage and HCC development at later stages of liver disease was demonstrated through the treatment of young Mdr2-KO/FVB mice with anti-inflammatory and anti-oxidant agents, which reduced both early hepatitis and the incidence of large tumors in the livers of aged animals.5

To identify the role of individual candidate regulatory genes in HCC development against the background of chronic inflammation, we used a strategy of combining Mdr2-KO with other mutations. We transferred the Mdr2-KO mutation from the FVB strain to the C57 black 6 (B6) strain. Genetic backgrounds of inbred murine strains may have a profound effect on manifestations of different types of liver injury and on HCC development.6-8 In this study, we investigated the phenotypic differences between well-characterized Mdr2-KO/FVB mice and newly generated Mdr2-KO/B6 mice both at early stages of chronic hepatitis and at late stages of HCC development. Using comparative gene expression analysis of both strains at an early stage of chronic hepatitis, we assessed the candidate regulatory genes that may link chronic inflammation to HCC development.

Abbreviations

Abcb4, adenosine triphosphate–binding cassette b4; Aif1, allograft inflammatory factor 1; ALP, alkaline phosphatase; ALT, alanine aminotransferase; Angptl4, angiopoietin-like 4; B6, C57 black 6; BrdU, bromodeoxyuridine; ConA, concanavalin A; CoxIV, cytochrome c oxidase subunit IV; Cxcl2, chemokine (C-X-C motif) ligand 2; Defb1, defensin beta 1; FVB, Friend virus B-type/N; Gal-1, galectin-1; GCRMA, GeneChip robust multiarray analysis; GO, Gene Ontology; HCC, hepatocellular carcinoma; H&E, hematoxylin and eosin; Hexim1, hexamethylene bisacetamide inducible 1; Hprt, hypoxanthine-guanine phosphoribosyltransferase; Ifn, interferon; Il, interleukin; Inhbe, inhibin beta E; KO, knockout; LBI, liver-to-body weight index; Lgals1, lectin galactoside-binding soluble 1; Lpin, lipin; Mat, methionine adenosyltransferase; Mbd1, methyl-CpG binding domain protein 1; Mdr2, multidrug resistance 2; mRNA, messenger RNA; Mrps21, mitochondrial ribosomal protein S21; NF-κB, nuclear factor kappa B; Ntrk2, neurotrophic tyrosine kinase receptor type 2; Por, P450 (cytochrome) oxidoreductase; Ptgds, prostaglandin D2 synthase; Rgs5, regulator of G-protein signaling 5; RT-PCR, reverse-transcription polymerase chain reaction; Slpi, secretory leukocyte peptidase inhibitor; Spink3, serine peptidase inhibitor Kazal type 1; Tff3, trefoil factor 3; Tifa, TRAF-interacting protein with forkhead-associated domain; Tnf, tumor necrosis factor; WT, wild type.

Materials and Methods

Full details are available in the Supporting Information.

Mice.

All animal experiments were performed according to national regulations and guidelines of the Institutional Animal Welfare Committee (National Institutes of Health approval number OPRR-A01-5011). FVB.129P2-Abcb4tm1Bor (Mdr2-KO/FVB) and wild-type (WT) FVB mice were purchased from the Jackson Laboratory (Bar Harbor, ME). B6 mice were obtained from Harlan Farm (Hebrew University, Jerusalem, Israel). Galectin-1 (Gal-1)–KO/B6 mice were generated previously by Francoise Poirier.9 The Mdr2-KO/B6 and Gal-1–KO/FVB mice were produced at our institute as described in the supporting information.

Concanavalin A (ConA) Treatment.

ConA was dissolved in pyrogen-free saline. Mice were injected intravenously with 0.15 mL of either ConA (12 mg/kg of body weight) or phosphate-buffered saline. Blood samples were collected either 6 and 24 hours after the ConA injection (from live animals) or 8 or 48 hours after the ConA injection (upon sacrifice), as described in the supporting information.

Gene Expression Profiling.

Total RNA was isolated from frozen liver tissues with the TRIzol reagent (Invitrogen, Carlsbad, CA) as described by the manufacturer. Gene expression profiling was performed with Mouse Genome Array 430A (Affymetrix, Santa Clara, CA). Data were analyzed by the GeneChip robust multiarray analysis (GCRMA) preprocessing algorithm with Partek software (ProSoftware, Partek, St. Louis, MO).

Immunoblotting and Immunohistochemistry.

Protein detection was performed via immunoblotting and immunohistochemical staining of formalin-fixed, paraffin-embedded liver tissue sections as previously reported2 with the antibodies described in the supporting information.

Results

Slower HCC Development in Mdr2-KO/B6 Mice Versus an FVB Genetic Background.

With the long-term goal of identifying genetic and environmental factors influencing inflammation-induced liver carcinogenesis, we first examined differences in HCC development in Mdr2-KO mice of two different genetic backgrounds. HCC development was significantly retarded in the Mdr2-KO/B6 mice versus the Mdr2-KO/FVB mice. This retardation was more prominent in males: the tumor incidence, size, and load in the Mdr2-KO/B6 males at the age of 18 months were similar to those in the Mdr2-KO/FVB males at the age of 12 months (Fig. 1). In females, the tumor incidence and load in the Mdr2-KO/B6 strain at the age of 16 months were similar to those in the Mdr2-KO/FVB strain at the age of 12 months (Fig. 1). Thus, in the Mdr2-KO/B6 mice, HCC developed approximately 6 months later in males and approximately 4 months later in females in comparison with the Mdr2-KO/FVB mice.

Figure 1.

Tumor load and size in the livers of Mdr2-KO mice with FVB and B6 genetic backgrounds at different ages. A dot diagram shows the numbers and diameters of spontaneous liver tumors in aged males (filled symbols) and females (open symbols) of both Mdr2-KO strains. The numbers of animals in the different age groups were as follows: 18 B6 males at 16 months, 20 B6 males at 18 months, 16 FVB males at 12 months, 8 FVB males at 14 months, 16 FVB males at 16 months, 24 B6 females at 16 months, 13 FVB females at 12 months, 13 FVB females at 14 months, and 14 FVB females at 16 months. Only nodules with a diameter of at least 3 mm are shown. At the age of 16 months, in the Mdr2-KO/B6 males, most nodules had a diameter less than 3 mm, and the maximum diameter of the nodules among the tested males was less than 5 mm.

Different Course of Early Chronic Hepatitis in Mdr2-KO/B6 Mice Versus Mdr2-KO/FVB Mice.

To understand the mechanisms underlying the significant differences in HCC development in Mdr2-KO mutants with B6 and FVB genetic backgrounds, we followed the dynamics of chronic hepatitis in males of both strains at early stages of liver disease (1, 2, and 3 months of age; Supporting Fig. 1). At 1 month of age, the main histological parameters of chronic hepatitis and cholangitis, bile duct proliferation and portal inflammation, progressed more rapidly in the Mdr2-KO/B6 liver versus the Mdr2-KO/FVB liver. However, in the Mdr2-KO/FVB males, these parameters increased with age, whereas in the Mdr2-KO/B6 males, they peaked at 2 months and had decreased at 3 months of age. Livers of the 3-month-old Mdr2-KO/FVB males had characteristic fibrosis with early septal formation, whereas there was no septal formation in the livers of the Mdr2-KO/B6 strain at all early ages tested (Supporting Fig. 1A).

The alanine aminotransferase (ALT) and alkaline phosphatase (ALP) serum levels in the Mdr2-KO mice of both strains were increased in comparison with controls at all ages tested (Fig. 2A,B), and this was indicative of chronic hepatitis and cholangitis. However, these parameters were more profoundly increased in the Mdr2-KO/FVB mice versus the Mdr2-KO/B6 mice, especially at 2 and 3 months of age (Fig. 2A,B). Remarkably, lower levels of serum cholesterol, a well-known result of the Mdr2-KO mutation,10 were observed only in the Mdr2-KO/FVB mice and not in the Mdr2-KO/B6 mice (Fig. 2C). As for the control Mdr2+/− mice, serum cholesterol was significantly higher in the FVB strain versus the B6 strain at all ages tested (Fig. 2C), and this was in agreement with known differences in cholesterol and triglyceride levels between WT FVB and B6 strains.11

Figure 2.

Age-associated physiological changes in Mdr2-KO mice: (A) ALT activity, (B) ALP activity, and (C) cholesterol level. Liver enzyme activities and cholesterol levels in the serum of Mdr2+/− and Mdr2-KO mice of the FVB and B6 strains are shown in the left and middle columns, respectively. The right column shows comparisons of the FVB and B6 strains [(A,B) Mdr2-KO mutants of both strains and (C) Mdr2+/− controls of both strains]. Mdr2+/−, Mdr2-KO/FVB, and Mdr2-KO/B6 are indicated by open bars, light gray bars, and dark gray bars in all graphs except the right graph of part C, in which Mdr2+/−/FVB and Mdr2+/−/B6 are indicated by open bars and striped bars, respectively. Importantly, in the control mice of both genetic backgrounds, the ALT and ALP serum levels were similar. There were four to eight males per group. The results are shown as means and standard errors of the mean. *P < 0.05 and **P < 0.005.

Different Infiltration of Inflammatory Cells in Mdr2-KO Livers of the B6 and FVB Strains.

To understand the contribution of immune cells in this model, we followed the dynamics of the infiltration of monocytes/macrophages, neutrophils, and T cells into male Mdr2-KO livers of both strains at early stages of chronic liver disease (Fig. 3A-C and Supporting Fig. 1B). The frequency of all these cells in livers from the Mdr2-KO strains was remarkably higher than the frequency in corresponding control livers (not shown). Remarkably, between 2 and 3 months of age, the frequency of monocytes/macrophages was increased in the Mdr2-KO/FVB livers but decreased in the Mdr2-KO/B6 livers (Fig. 3A). The number of T cells in the portal tracts of the Mdr2-KO/B6 mice did not change significantly with age, whereas it significantly increased from the first to second month and was still high in the third month in the Mdr2-KO/FVB livers (Fig. 3B). The number of neutrophils in the portal tracts of the Mdr2-KO/B6 mice was not altered with age, whereas it increased from the second to the third month in the Mdr2-KO/FVB livers (Fig. 3C).

Figure 3.

Kinetics of immune cell recruitment and control of hepatocyte proliferation in Mdr2-KO mice. The results are shown as mean and standard errors of the mean for Mdr2-KO/FVB (light gray bars), Mdr2-KO/B6 (dark gray bars), and respective Mdr2+/− strains (open bars). Microscopic quantification of (A) infiltrated monocytes/macrophages (F4/80+ cells per ×40 power field), (B) T cells (CD3+ cells per ×20 power field), and (C) neutrophils in liver zone 1 (periportal, per ×20 power field) of Mdr2-KO/FVB and Mdr2-KO/B6 mice. The total numbers of F4/80+ cells in the liver parenchyma were similar in the controls of both strains (not shown). Characteristic images for corresponding immunostains are presented in Supporting Fig. 1. LBI and histomorphometrical analysis of hepatocyte proliferation markers in Mdr2-KO and control mice: (D) LBI for Mdr2+/− and Mdr2-KO mice of the FVB and B6 strains, (E) quantification of the BrdU uptake into hepatocytes of control and Mdr2−/− livers of FVB and B6 backgrounds (per ×20 power field), and (F) average number of cyclin D1+ hepatocytes (per ×40 power field). Corresponding pictures of immunohistochemistry stains are presented in Supporting Figs. 2 and 3. The counting was performed mostly in zone II (middle zone) for all tested liver sections. There were four to eight males per group at each time point. *P < 0.05 and **P < 0.005.

Control of Hepatocyte Proliferation and Liver Mass in Mdr2-KO Mice of Both Strains.

We previously demonstrated an increased liver-to-body weight index (LBI) and increased cyclin D1 levels in the hepatocyte nuclei of Mdr2-KO/FVB mice at the age of 3 months.2, 4 Here we found that in the Mdr2-KO/FVB strain, a constant LBI was already established at 2 months of age, whereas in the Mdr2-KO/B6 strain, LBI sharply decreased between 2 and 3 months of age (Fig. 3D).

Bromodeoxyuridine (BrdU) incorporation into hepatocytes of both strains dropped significantly between 1 and 2 months of age for the Mdr2-KO mice and the control Mdr2+/− mice (Fig. 3E and Supporting Fig. 2). However, in the Mdr2-KO/FVB mice, BrdU incorporation was similarly increased in a significant manner in comparison with controls at 2 and 3 months of age, whereas in the Mdr2-KO/B6 mice, it was barely increased at 2 months, and it was not further elevated at 3 months of age (Fig. 3E).

Nuclear levels of cyclin D1 in hepatocytes decreased gradually with age in Mdr2-KO and control Mdr2+/− livers of both strains (Fig. 3F and Supporting Fig. 3), and they were particularly higher in the Mdr2-KO livers. In the Mdr2-KO/B6 mice, nuclear localization of cyclin D1 in hepatocytes dropped significantly between 2 and 3 months of age (Fig. 3F).

There was low and equivalent hepatocyte apoptosis in mutant and control livers at all ages tested (Supporting Fig. 4). Thus, at 2 months of age, hepatocyte proliferation and LBI were similarly increased in both Mdr2-KO mutants, whereas at 3 months of age, these parameters were significantly reduced in the Mdr2-KO/B6 strain.

Genome-Scale Gene Expression Profiling Analysis of Mdr2-KO/B6 Mice.

To dissect the mechanisms responsible for the attenuation of chronic liver disease in the Mdr2-KO/B6 mice at the age of 3 months, we performed a genome-scale gene expression profiling analysis of liver tissues from the Mdr2-KO and control Mdr2+/− B6 males with Affymetrix microarrays. We compared genes that were aberrantly expressed in Mdr2-KO livers from the B6 strain (our current study) and the FVB strain (our previous study4) with the same GCRMA algorithm. This comparative analysis revealed a striking difference between up-regulated and down-regulated genes in the Mdr2-KO mutants of both strains: there were approximately 3.8-fold more up-regulated genes in the FVB strain versus the B6 strain, whereas for down-regulated genes, there was only about a 20% difference between the strains (Fig. 4A,B). Remarkably, there were also 14 genes that were up-regulated in the FVB strain but down-regulated in the B6 strain (there were no genes with the reversed type of differential expression). This group included, among others, the regulatory genes CCAAT/enhancer binding protein beta (Cebpb), methyl-CpG binding domain protein 1 (Mbd1), and P450 (cytochrome) oxidoreductase (Por) and genes involved in lipid metabolism such as angiopoietin-like 4 (Angptl4), lipin 1 (Lpin1), and Lpin2 (Supporting Table 3). The pathway analysis of differentially expressed genes revealed that the following Gene Ontology (GO) terms were specific to the Mdr2-KO/FVB strain: regulation of immune response, regulation of cell proliferation, regulation of cell cycle, lipid biosynthesis process, and response to oxidative stress (Table 1). The most interesting and statistically significant genes that were differentially expressed in Mdr2-KO livers from both murine strains are shown in Supporting Tables 2 and 3.

Figure 4.

Genes differentially expressed in the livers of two Mdr2-KO strains. Venn diagrams present the numbers of common and unique differentially expressed genes in the livers of Mdr2-KO strains as revealed by Affymetrix microarrays: (A) up-regulated and (B) down-regulated genes in the livers of Mdr2-KO/FVB (white circles) and Mdr2-KO/B6 mice (gray circles; ≥1.9-fold change versus age-matched and sex-matched Mdr2+/− controls, P ≤ 0.04). Approximately 50% of the genes up-regulated in the B6 strain were also up-regulated in the FVB strain, whereas only 17% of the genes down-regulated in the B6 strain were also down-regulated in the FVB strain. For the FVB strain, the proportion of genes that were differentially expressed in a similar fashion in the B6 strain was approximately 13% for both up-regulated and down-regulated genes. Gene expression was tested by semiquantitative RT-PCR: (C) validation of the differential expression of selected genes from the Affymetrix data and (D) expression of selected cytokine genes that were not detected by the Affymetrix microarrays. Total RNA was extracted from liver tissues of 3-month-old Mdr2+/− and Mdr2-KO male mice of the B6 or FVB strain.

Table 1. Distribution of the Genes Differentially Expressed in Mdr2-KO/FVB and Mdr2-KO/B6 Livers Versus Corresponding Controls Into Different Most Significant GO Categories According to Biological Functions
Term for FVB TotalCountP Value
GO:0048518: positive regulation of biological processes* ↑↓975.74E−05
GO:0006952: defense response*397.10E−05
GO:0045595: regulation of cell differentiation*363.55E−05
GO:0006629: lipid metabolic process* ↑↓506.45E−04
GO:0042127: regulation of cell proliferation ↑417.16E−04
GO:0050776: regulation of immune response*190.0018
GO:0045619: regulation of lymphocyte differentiation ↑80.0128
GO:0051726: regulation of cell cycle ↑180.0128
GO:0008610: lipid biosynthetic process ↑210.0007
GO:0006979: response to oxidative stress ↑90.0362
Term for B6 TotalCountP Value
  • DAVID software (level 3) was used. Nine hundred five genes for the Mdr2-KO/FVB strain and 323 genes for the Mdr2-KO/B6 strain with a threshold for differential expression of at least 1.8-fold (the mean of Mdr2-KO mice with respect to the mean of control Mdr2+/− mice) were subjected to GO analysis. The arrows directed upward and downward designate the up-regulated and down-regulated genes, respectively, that composed the terms.

  • *

    The GO term is common for Mdr2-KO/FVB and Mdr2-KO/B6 differentially expressed genes.

GO:0048518: positive regulation of biological processes*310.002
GO:0006952: defense response*175.97E−05
GO:0045595: regulation of cell differentiation*120.005
GO:0006629: lipid metabolic process*170.007
GO:0050778: positive regulation of immune response*60.021

The expression of the selected genes in the livers of Mdr2-KO/B6 males at the age of 3 months was validated by reverse-transcription polymerase chain reaction (RT-PCR; Fig. 4C). We confirmed the up-regulation of the defensin beta 1 (Defb1), inhibin beta E (Inhbe), Jun, lectin galactoside-binding soluble 1 (Lgals1), neurotrophic tyrosine kinase receptor type 2 (Ntrk2), regulator of G-protein signaling 5 (Rgs5), and serine peptidase inhibitor Kazal type 1 (Spink3) transcripts and the down-regulation of the Lpin1 transcript. Remarkably, the allograft inflammatory factor 1 (Aif1), Cd36, and prostaglandin D2 synthase (Ptgds) genes, which were highly up-regulated in the livers of Mdr2-KO/FVB mice,4 were not differentially expressed in the Mdr2-KO/B6 livers (Fig. 4C). We also tested the expression of the selected immune regulators, which play important roles in shaping the inflammatory response, but they were not detected by the Affymetrix microarrays. We found that for both strains, the expression of the genes tumor necrosis factor α (Tnfa), interleukin-2 (Il-2), and Il-10 was significantly up-regulated in the livers of 3-month-old Mdr2-KO mice (Fig. 4D).

Validation of the Differential Expression of Selected Genes in Mdr2-KO Mice at the Protein Level.

We aimed to validate at the protein level the differential expression of 2 regulatory genes among those 14 genes that were reversely differentially expressed in two Mdr2-KO strains: Lpin1 and Mbd1. The Lpin1 gene encodes Lipin-1, one of the key regulators of lipid metabolism.12 The relevance of alterations in lipid metabolism for the pathogenesis and progression of chronic liver disease in Mdr2-KO mice was recently demonstrated.13 The expression patterns of Lipin-1 in Mdr2-KO livers versus control Mdr2+/− livers at the age of 3 months were confirmed at the protein level and showed up-regulation in FVB mice but down-regulation in the B6 strain (Fig. 5A).

Figure 5.

Detection of differentially expressed proteins by immunohistochemistry. (A) Lipin-1 expression in Mdr2+/− and Mdr2-KO livers from B6 mice (left panels) and FVB mice (right panels) that were 3 months old. Lipin-1 was detected mainly in the cytoplasm of hepatocytes with some nuclear hepatocyte staining (magnification ×100). (B) Mbd1 expression in Mdr2+/− and Mdr2-KO livers from B6 mice (left panels) and FVB mice (right panels) that were 3 or 2 months old (magnification ×200). (C) Counts of Mbd1-positive and Mbd1-negative hepatocytes in B6 mice (left panels) and FVB mice (right panels). Hepatocytes were counted (at least 250 cells per liver) as unstained (negative), weakly stained (weak), or strongly stained (strong); there were three males per group. The significant difference in the Mbd1 levels revealed by immunostaining between the B6 and FVB strains could not be attributed to the strain specificity of Mbd1 sequences because of the identity of amino acids in the antibody-specific Mbd1 epitopes in both strains. Mdr2-KO/FVB, Mdr2-KO/B6, and respective Mdr2+/− strains are indicated by light gray bars, dark gray bars, and open bars, respectively. (D) Increased Gal-1 expression in liver tissues from Mdr2-KO strains. Immunohistochemistry staining is shown for Gal-1 in Mdr2-KO and control Mdr2+/− liver sections from the B6 strain (left panels) and the FVB strain (right panels) at 1 and 3 months of age. Significantly more Gal-1–positive cells were detected in Mdr2-KO livers versus Mdr2+/− congenic livers from both strains, particularly at 3 months of age (magnification ×100).

Mbd1, the gene encoding Mbd1, was significantly down-regulated in the Mdr2-KO/B6 liver, whereas it was significantly up-regulated in the Mdr2-KO/FVB liver (Supporting Table 3). However, protein expression demonstrated a negative correlation with transcript expression in the Mdr2-KO/B6 mice. The Mbd1 protein was localized mainly in the hepatocyte nuclei and demonstrated stronger staining in hepatocytes from B6 Mdr2-KO mice versus Mdr2+/− mice; at the age of 2 months, the difference between mutant and control livers was even larger (left panels, Fig. 5B,C). Mbd1 staining was much weaker in the FVB strain versus the B6 strain, whereas the patterns of expression were similar and showed prominent localization in the hepatocyte nuclei (right panels, Fig. 5B,C). When we tested the expression levels of these proteins, Lipin-1 and Mbd1, in liver tumors from old Mdr2-KO mice of both strains, we found variable expression patterns for both proteins. They evidenced higher expression in some tumors or nodules and lower expression in others in comparison with the noncancerous surrounding tissue (Supporting Fig. 5).

Decreased Expression of Methionine Adenosyltransferase 1a (Mat1a) Protein in Mdr2-KO/FVB Mice.

Although genome-scale gene expression profiling did not reveal aberrant expression of the Mat1a transcript in the livers of Mdr2-KO mice of both strains, we compared Mat1a protein levels in total liver extracts from mutant and control mice at the age of 3 months. We found a significant decrease in Mat1a at the protein level in livers of Mdr2-KO/FVB mice but not in livers of Mdr2-KO/B6 mice (Fig. 6A,B). Both WT (Fig. 6A) and Mdr2+/− males (not shown) of the FVB and B6 strains were used as controls, and they showed similar expression of Mat1a.

Figure 6.

Detection of differentially expressed proteins by immunoblotting. (A,B) The level of the Mat1a protein was decreased in Mdr2-KO livers of the FVB strain (but not the B6 strain). (A) Representative images of Mat1a immunoblotting of liver tissue extracts from 3-month-old Mdr2-KO (right) and Mdr2-WT (left) males of FVB and B6 genetic backgrounds. There were six mice per group for Mdr2-KO and three mice per group for Mdr2-WT. (B) Graphic presentation of the results shown in panel A and normalized to the housekeeping CoxIV values. (C) Expression of the Gal-1 protein in the livers of 1- and 3-month-old Mdr2-KO/FVB and Mdr2-KO/B6 mice. CoxIV was used as a loading control. Two typical replicates of each group are shown. (D) Densitometric analysis of Gal-1 immunoblots based on triplicates and normalized to the CoxIV values (means and standard errors of the mean). **P < 0.005.

Increased Expression of Gal-1 in Livers From Both Mdr2-KO Strains.

Because of the paradoxical anti-inflammatory and protumorigenic effects of Gal-1 and its significant up-regulation in Mdr2-KO/FVB mice, we further validated the functional relevance of this endogenous lectin in both strains. We observed up-regulation of the Lgals1 transcript encoding the Gal-1 protein in the livers of 3-month-old Mdr2-KO/B6 mice by RT-PCR (Fig. 4C). We confirmed the increased expression of the Gal-1 protein in both mutant Mdr2-KO livers (Figs. 5D and 6C,D) starting from the first month of age. The Gal-1 protein was localized mainly in the cytoplasm of nonhepatocyte cells, including immune cells and cholangiocytes, and rarely in the cytoplasm of hepatocytes. Remarkably, in the bile ducts, Gal-1 was detected only in proliferating cholangiocytes (ductular reaction). Expression of the Gal-1 protein increased between 1 and 3 months of age in both Mdr2-KO strains. When the strains were compared, the level of Gal-1 was higher in Mdr2-KO/B6 livers at the age of 1 month, but the levels were similar in the two strains at the age of 3 months (Fig. 6D).

Endogenous Gal-1 Selectively Protects B6 Mice From ConA-Induced Hepatitis.

Previous studies showed that the administration of exogenous recombinant Gal-1 protected mice against autoimmune hepatitis induced by ConA.14 Here we demonstrate that B6 Gal-1–KO mice had an increased sensitivity to ConA-induced hepatitis (Fig. 7A,B), and this highlights the relevance of endogenous Gal-1. Reflecting hepatocyte injury, ALT blood levels were significantly higher in the B6 Gal-1–KO mice versus the control WT mice at every time tested (Fig. 7A); this was also true for liver tissue necrosis 48 hours after ConA injection (Fig. 7B). To test the role of the host genetic background in this process, we transferred the Gal-1–KO mutation into the FVB strain and challenged the Gal-1–KO/FVB mice with ConA. Surprisingly, the extent of injury in the Gal-1–KO/FVB livers following the ConA challenge was similar to that in FVB WT controls (Fig. 7C). This result was also confirmed with the use of a 2-fold higher ConA dose (not shown). We found that the Gal-1 transcript was up-regulated 8 hours after the ConA injection in both WT strains (Fig. 7D). These results demonstrate that endogenous Gal-1 selectively protects against ConA-induced liver injury in the B6 strain but not in the FVB strain.

Figure 7.

Enhanced liver injury after ConA-induced hepatitis in Gal-1–KO/B6 mice versus Gal-1–KO/FVB mice. WT control and Gal-1–KO mice of (A,B) B6 and (C) FVB genetic backgrounds were injected with ConA. (A,C) Serum ALT levels from these mice are shown at various time points after ConA delivery. There were three to four animals in each group; one of two independent experiments for each strain is shown. (B) H&E slides are shown for representative mouse livers obtained 48 hours after the injection of ConA into B6 WT and Gal-1–KO mice (magnification ×200). Black arrows indicate massive necrosis observed in the Gal-1–KO liver. (D) Gal-1 transcript expression was tested by semiquantitative RT-PCR in the livers of WT/B6 mice (upper panels) and WT/FVB mice (bottom panels) 8 hours after the injection of either ConA or saline. The Hprt transcript level was used as an internal control. (E) The expression of the indicated hepatic genes was detected by semiquantitative RT-PCR 8 hours after the injection of ConA into WT/B6 and Gal-1–KO/B6 mice. (F) The gene expression results shown in panel E were quantified with Scion software and normalized to Hprt expression (means and standard errors of the mean). *P < 0.05 and **P < 0.005 for the ConA-treated Gal-1–KO strain versus the WT strain.

To uncover the molecular mechanisms for the increased sensitivity of Gal-1–KO/B6 mutants to ConA-induced hepatitis, we tested the expression of selected genes 8 hours after ConA injection (Fig. 7E,F). The most significant difference between the experimental groups was the increased expression of the proinflammatory cytokines Tnfa, Il-2, and chemokine (C-X-C motif) ligand 2 (Cxcl2) and the anti-inflammatory secretory leukocyte peptidase inhibitor (Slpi) in Gal-1–KO/B6 livers (Fig. 7F).

Discussion

The Mdr2-KO mouse model of inflammation-induced HCC mimics human disease in terms of both prolonged chronic hepatitis preceding tumor development1 and aberrant gene expression in tumors.2 The phenotypic manifestations of the Mdr2-KO mutation are strain-dependent. Initially, the mutation was introduced into the 129/OlaHsd strain,15 and this resulted in a highly enlarged (up to 8-fold) nodular liver already at the age of 6 months.1 The Mdr2-KO/FVB mice have a mildly increased liver/body index (approximately 1.6-fold in males) that does not change significantly between 3 and 12 months of age.4 Now, we transferred the Mdr2-KO mutation into the B6 genetic background and demonstrated significantly retarded HCC development and inhibition of chronic hepatitis between 2 and 3 months of age in Mdr2-KO/B6 males.

Inflammatory Response in the FVB and B6 Mdr2-KO Strains.

Multiple genes involved in the control of immune/inflammatory responses were up-regulated mainly in the Mdr2-KO/FVB strain, and this was in agreement with the higher infiltration of immune cells. One of these genes, Lgals1, encodes Gal-1, an endogenous lectin that is widely expressed in epithelial and immune cells and acts both extracellularly and intracellularly by modulating innate and adaptive immunity.16, 17 In addition, Gal-1 is a key mediator of the immunosuppressive activity of regulatory T cells.18 Gal-1 overexpression in many types of tumors and/or surrounding tissues promotes tumor progression through multiple mechanisms: the inhibition of antitumor immunity,16, 19 the promotion of Ras activation,20 the stimulation of tumor angiogenesis,21, 22 and the attenuation of NF-κB activation.23 Gal-1 is overexpressed in human HCC24, 25 and in the Mdr2-KO liver. Recently, it was shown that Gal-1 promotes HCC cell adhesion through phosphoinositide 3-kinase and/or extracellular signal-regulated kinase 1/2 signaling pathways.26 Remarkably, when HCC patients were classified into low- and high-survival groups on the basis of global gene expression profiling of their liver samples,27 high Gal-1 expression was associated with low survival. Thus, in inflammation-mediated hepatocarcinogenesis, Gal-1 may act as a protective anti-inflammatory agent during early stages of chronic liver pathology but as a protumorigenic agent during the late stages of the disease.

On the basis of our finding that endogenous Gal-1 protects the liver against ConA-induced hepatitis in the B6 strain but not in the FVB strain (Fig. 7), we hypothesize that the increased inflammation (and aberrant regulation of multiple immune/inflammatory genes) in the Mdr2-KO/FVB strain can be explained in part by the impaired anti-inflammatory activity of Gal-1. This hypothesis is supported by the up-regulation of other anti-inflammatory genes (e.g., Slpi) specifically in the Mdr2-KO/FVB strain (Supporting Table 2), more robust Gal-1 induction in the Mdr2-KO/B6 liver versus the Mdr2-KO/FVB liver at the age of 1 month (Fig. 6C,D), and the induction of proinflammatory Tnfa, Il2, and Cxcl2 as well as anti-inflammatory Slpi after ConA challenge in Gal-1–KO/B6 mice (Fig. 7E,F).

Several genes that were differentially expressed only in the Mdr2-KO/B6 strain are involved in the regulation of NF-κB signaling: trefoil factor 3 (Tff3) and TRAF-interacting protein with forkhead-associated domain (Tifa), which were up-regulated, and hexamethylene bisacetamide inducible 1 (Hexim1), which was down-regulated (Supporting Table 2). Tff3 may transiently activate NF-κB,28 and Tifa also activates NF-κB,29 whereas Hexim1 inhibits NF-κB–mediated transcription.30 These examples indicate differential regulation of the NF-κB signaling pathway between the two Mdr2-KO strains.

Decreased Expression of Genes Regulating Lipid Metabolism and Reactive Oxygen Species in Mdr2-KO/B6 Mice.

Among genes up-regulated in the Mdr2-KO/FVB mice but down-regulated in the Mdr2-KO/B6 mice (Supporting Table 2), Angptl4, Lpin1, and Lpin2 encode regulators of lipid metabolism, and Por encodes cytochrome P450 reductase. In line with reduced chronic inflammation in Mdr2-KO/B6 mice, many genes involved in response to oxidative stress were up-regulated in the Mdr2-KO/FVB strain but not in the Mdr2-KO/B6 strain. This effect of the Mdr2-KO/B6 liver could be caused in part by down-regulation of the Por and Lipin-1 proteins. Por may be a source of endogenous oxidative DNA damage and genetic instability,31 and Lipin-1 promotes fatty acid oxidation.12 Importantly, the gene Aif1, encoding Aif1, was up-regulated in the Mdr2-KO/FVB liver4 but not in the Mdr2-KO/B6 liver (Fig. 4C). Aif1 is a crucial mediator in the inflammatory response, which may enhance NF-κB transcriptional activity and facilitate tumor growth,32 and thus serves as a central node in coexpression networks connecting obesity-associated genes in the liver and other tissues.33

Liver/Body Index and Control of Hepatocyte Proliferation.

Reduced liver inflammation in the Mdr2-KO/B6 strain at the age of 3 months correlated with more stringent control of the liver size. Remarkably, we revealed a significant down-regulation of the Mat1a protein specifically in the livers of Mdr2-KO/FVB mice (Fig. 6A,B). The Mat enzyme catalyzes the synthesis of S-adenosyl methionine, a universal donor of the methyl group for all methylation reactions in the cell. It is encoded by the genes Mat1a and Mat2a/2b; patients with liver cirrhosis have reduced activity of this enzyme.34 Mat1a is highly expressed in the adult liver and keeps hepatocytes in a quiescent state. In human liver cancer, Mat1a expression is reduced, whereas Mat2a is increased; this switch facilitates cancer cell growth.35 We demonstrated previously that most hepatocytes of the Mdr2-KO/FVB strain undergo cell cycle arrest between the ages 3 and 9 months, and this is characterized by a high level of cyclin D1 in hepatocyte nuclei.4, 36 Our current findings demonstrate that selectively in the Mdr2-KO/FVB mice, this stage of hepatocyte cycle arrest is characterized by a decreased level of the Mat1a protein and up-regulation of the Mat2b transcript (Supporting Table 2).

Despite the inverse differential expression of the Mbd1 transcript in the two Mdr2-KO strains (Supporting Table 2), protein expression in both strains was increased in the mutant liver versus the control liver (Fig. 5). The Mbd1 protein binds methylated DNA and functions mainly as a transcriptional repressor; its higher level in the Mdr2-KO/B6 strain versus the Mdr2-KO/FVB strain could be one of the factors responsible for a significantly lower number of up-regulated genes in the former mutant.

Differences Between the WT B6 and FVB Strains.

Published data on the genotypic and phenotypic differences between the B6 and FVB strains are summarized in Supporting Table 4. There are several known differences between these strains that could be responsible for the different courses of chronic hepatitis and HCC development in the two Mdr2-KO mutants. The most prominent among them are (1) a deficiency of complement C5 protein in the FVB strain, (2) mutations in mitochondrial DNA in the FVB strain, and (3) a Tnfaip3 (A20) polymorphism that is responsible for the less effective feedback suppression of Tnf-α–induced NF-κB activation in the B6 strain (see the references in Supporting Table 4).

In conclusion, we have demonstrated that the B6 murine strain has a remarkable resistance to both chronic hepatitis and HCC development caused by the Mdr2-KO mutation. By a comparative analysis of liver gene expression in the two Mdr2-KO strains, we determined a set of regulatory genes that could be responsible for affecting the severity of chronic hepatitis in these strains at an early age. The most prominent was the differential expression of multiple regulators of the NF-κB pathway, which is critical for manifestations of the Mdr2-KO phenotypes.3 Our finding of the inability of endogenous Gal-1 to protect the liver against ConA-induced hepatitis specifically in the FVB strain helps us to understand the greater induction of multiple pro- and anti-inflammatory genes in the Mdr2-KO/FVB mutant versus the Mdr2-KO/B6 mutant. Our results indicate a deep interconnection between genes that control inflammation, oxidative stress, and lipid metabolism and help to reveal key regulators that determine HCC development during early stages of chronic hepatitis.

Acknowledgements

The authors thank Mery Clausen for her assistance with the manuscript preparation.

Ancillary