Mice with hepatocyte-specific FXR deficiency are resistant to spontaneous but susceptible to cholic acid-induced hepatocarcinogenesis

Animals and treatments.

FXR hepatocyte-specific knockout mice (FXR^hep−/−) have been described previously (9). In brief, mice with “floxed” FXR (FXR^flox\flox) were crossed with FXR^flox\flox mice carrying albumin promoter-driven Cre (Cre^+/−) to generate FXR^flox\flox\Cre^−/− (wild-type mice, herein referred to as WT) and FXR^flox\flox\Cre^+/− (hepatocyte-specific FXR KO mice, herein referred to as FXR^hep−/−) mice. Both WT and FXR^hep−/− were on a C57BL/6J genetic background and were maintained in pathogen-free animal facilities in the Laboratory of Animal Research under a standard 12-h light-dark cycle with food and water provided ad libitum. All the protocols and procedures were approved by the Institutional Animal Care and Use Committee at the University of Kansas Medical Center.

Male WT and FXR^hep−/− mice (n = 4–15) were treated with a single intraperitoneal injection of DEN at 90 mg/kg or PBS vehicle at 4 wk of age, followed by feeding with 0.2% (wt/wt) CA-containing diet or regular chow diet for 30 wk. Liver tissues were harvested to determine mRNA and protein levels.

Serum biochemistry.

Serum alanine aminotransferase (ALT) and alkaline phosphatase (ALP) activities along with triglyceride, total cholesterol, and total bilirubin levels were determined by use of commercially available kits (Pointe Scientific, Canton, MI). Serum total bile acids were measured by using a bile acid assay kit according to the manufacturer's instructions (Bio-Quant, San Diego, CA).

Liver histology.

When livers were collected, pieces of liver were fixed in 10% neutral buffered formalin solution and embedded in paraffin, and 5-μm-thick sections were obtained for hematoxylin and eosin staining. The pathological analysis and diagnosis was provided by a board-certified pathologist.

RNA isolation and Q-PCR.

Total RNA was isolated from frozen livers by TRIzol reagent. The mRNA expression levels of genes were quantified by quantitative real-time PCR (Q-PCR) using SYBR green chemistry and normalized to β-actin mRNA levels. A list of genes and primers are in Supplemental Table S1 (Supplemental Material for this article is available online at the Journal website).

Western blot analysis.

Western blot analysis was performed as previously described on total, nuclear, and cytoplasmic proteins (9). Antibodies are described in Supplemental Table S2.

Organic extraction and UPLC/MS profiling of liver bile acids.

Three-month-old male mice (n = 5) were used to determine the bile acid profile. Total bile acids were extracted from serum, liver, gall bladder, and intestine contents by a method described before (31, 33). Bile acid extracts were analyzed by using a Thermo Finnigan Ultra Performance Liquid Chromatography (UPLC) system (Thermo Fisher Scientific, Waltham, MA) coupled with a Thermo Finnigan LTQ XL Ion Trap Mass Spectrometer (MS; ITMS; Thermo Fisher Scientific, Waltham, MA). An Electrospray (ESI)/ITMS was operated in multiple MS/MS and SRM (Selective Reaction Monitoring) modes for simultaneous determination of bile acid profile (31). All the 23 bile acids are cholic acid (CA), chenodeoxycholic acid (CDCA), deoxycholic acid (DCA), lithocholic acid (LCA), ursodeoxycholic acid (UDCA), taurocholic acid (TCA), taurochenodeoxycholic acid (TCDCA), taurodeoxycholic acid (TDCA), taurolithocholic acid (TLCA), tauroursodeoxycholic acid (TUDCA), glycocholic acid (GCA), glycochenodeoxycholic acid (GCDCA), glycodeoxycholic acid (GDCA), glycolithocholic acid (GLCA), glycoursodeoxycholic acid (GUDCA), β-muricholic acid (β-MCA), α-muricholic acid (α-MCA), ω-muricholic acid (ω-MCA), tauro-β-muricholic acid (T-β-MCA), tauro-α-muricholic acid (T-α-MCA), tauro-ω-muricholic acid (T-ω-MCA), hyodeoxycholic acid (HDCA), and taurohyodeoxycholic acid (THDCA). The proportion of each individual bile acid in the total liver bile acid or total bile acid pool was calculated.

Statistical analysis.

Results were presented as means ± SD and were analyzed by one-way ANOVA followed by the Student-Newman-Keuls test. A difference with a P value of < 0.05 was considered statistically significant.

Tumor prevalence.

As expected, WT mice did not develop any grossly identified liver tumors with the treatments of vehicle, CA alone, or DEN alone (Table 1). With combined DEN/CA treatment, 44% of WT mice developed grossly identified liver tumors. In contrast to whole body FXR^−/− mice that develop spontaneous liver tumors at 12–15 mo of age (7, 28), the FXR^hep−/− mice did not develop liver tumors spontaneously even when observed for up to 24 mo of age (data not shown), nor did they develop any tumors on DEN alone. However, with CA alone or DEN/CA treatment, 100% of the FXR^hep−/− mice developed liver tumors.

Liver histology.

As shown in Table 2 and Fig. 1, A and B, WT and FXR^hep−/− mice did not show any degree of dysplasia with vehicle treatment. But one of four FXR^hep−/− mice showed focal portal inflammation, which we did not see in any of the WT mice. After CA treatment, 29% (2 of 7) of WT mice presented with high-grade dysplasia, but they did not show any signs of focal portal inflammation. In contrast, the CA treatment resulted in 87% dysplasia in FXR^hep−/− mice with 27% low-grade and 60% high-grade dysplasia and 27% focal portal inflammation. With DEN treatment only, none of the WT mice developed dysplasia or portal inflammation, whereas 29% of FXR^hep−/− mice developed low-grade dysplasia and 57% showed focal portal inflammation. For DEN/CA treatment, all WT mice presented with high-grade dysplasia, none with low-grade dysplasia, and 33% with focal portal inflammation; and for FXR^hep−/− mice, 57% showed high-grade dysplasia, 43% low-grade dysplasia, and 14% focal portal inflammation.

The body weight was increased in WT mice with CA feeding (Fig. 1C) but decreased in FXR^hep−/− KO mice. DEN treatment prevented CA-associated weight gain in WT mice but resulted in more weight loss in FXR^hep−/− mice. CA reduced the liver-to body weight (LW/BW) ratio in WT mice but increased that in FXR^hep−/− mice. DEN slightly reduced the LW/BW ratio in not only WT and but also FXR^hep−/− mice. DEN/CA cotreatment did not change the LW/BW ratio in WT mice but markedly increased that in the FXR^hep−/− mice.

Serum lipid biochemistry and markers of liver injury.

Since bile acid metabolism is closely related to lipid homeostasis, we measured triglyceride levels in these mice. The FXR^hep−/− mice had higher serum triglyceride levels than the WT mice. Treatment with CA reduced lipid levels in WT mice but further increased those in FXR^hep−/− mice, indicating that hepatic FXR is mainly responsible for the lipid-lowering effect following FXR activation. DEN/CA cotreatment reduced triglyceride levels in both WT and FXR^hep−/− mice (Fig. 2). With vehicle or DEN treatment, both WT and FXR^hep−/− mice had normal ALT levels, indicating lack of obvious liver injury. The CA treatment increased ALT levels fourfold in FXR^hep−/− mice, but not in the WT mice, but the combined DEN/CA treatment increased ALT levels in FXR^hep−/− as well as WT (fourfold). ALP levels were increased markedly in FXR^hep−/− mice with CA or DEN/CA treatment and moderately in WT mice with DEN/CA treatment, indicating degree of bile duct epithelial cell injury. Bile acids levels were increased in WT mice under CA or DEN/CA treatment, but the levels were within the normal range. However, FXR^hep−/− mice showed much higher bile acid levels with CA treatment (100–150 μM), suggesting that hepatic FXR is required for the defense against bile acid toxicity. The increases in ALP and bile acids shared the same trend in FXR^hep−/− mice, indicating that increases in bile acid levels are associated with bile duct epithelial cell injury. The increased liver damage in FXR^hep−/− mice was also reflected by a marked increase in bilirubin levels (Fig. 2).

Expression of genes involved in bile acid homeostasis in the liver. Figure 3 shows that basal mRNA expression of Fxr, Shp, and Bsep in the liver was lower in FXR^hep−/− mice than that in WT mice, and CA treatment increased the mRNA levels of these genes in WT mice, but not in FXR^hep−/− mice, suggesting that Fxr activation is required to maintain and induce the expression of Shp and Bsep. Moreover, CA treatment further reduced Bsep expression in the FXR^hep−/− mice, likely reflecting a suppression of Bsep expression via injury/inflammation. In agreement with our previous publication (9), Cyp7a1 mRNA levels were moderately higher in FXR^hep−/− mice than those in WT mice. Treatment with CA markedly suppressed Cyp7a1 mRNA levels, but the degree of suppression was smaller in FXR^hep−/− mice than in WT mice.

Cell proliferation and death.

Figure 4 shows the protein levels of genes critical for cell proliferation and cell apoptosis, including cyclin D1, cdk4, cyclin E1, cdk2, PCNA, PARP1, and caspase-3. The basal levels of cyclin D1 were lower in WT mice and almost undetectable in FXR^hep−/− mice, consistent with our previous finding that cyclin D1 is a FXR target gene in the liver (1, 22). The cyclin D1 levels were slightly higher in WT mice after CA treatment, but markedly higher in the FXR^hep−/− mice, suggesting that cyclin D1 is induced via-FXR independent mechanism(s). Cyclin D1 levels were slightly reduced in WT mice by DEN treatment, but markedly induced in both strains by the combined DEN/CA treatment. These data suggest that cyclin D1 induction may serve as a marker for tumor development in this study. Interestingly, the trend of change for cyclin E1 was opposite of that of cyclin D1. Furthermore, despite reduced basal cyclin D1 in FXR^hep−/− mice, the levels of its partner, cdk4, were higher (perhaps to compensate for the loss of cyclin D1). The levels of cdk4 were further induced by CA and DEN/CA in WT and FXR^hep−/− mice. Cyclin E1's partner, cdk2, was elevated strongly in FXR^hep−/− mice and moderately in WT mice with DEN/CA treatment, but its expression was very low in other groups. PCNA, a marker for DNA replication, was markedly induced by DEN/CA in both strains. Cleaved caspase-3 was present only in FXR^hep−/− mice after DEN/CA treatment, indicating that hepatic FXR may prevent apoptosis. The levels of cell cycle inhibitors were also determined (Fig. 5). The level of p15 showed no change with CA or DEN treatment, whereas p21 and p27 levels followed a similar trend of changes in FXR^hep−/− mice. In detail, WT mice had reduced expression of p21 following CA treatment with and without DEN, but p21 level was almost undetectable by treatment with CA regardless of DEN treatment in FXR^hep−/− mice. These data suggest that one of the mechanisms by which FXR deficiency and bile acids cause liver tumor formation is by reducing cell cycle inhibitors. The basal levels of p53 were undetectable in the WT mouse livers but were induced by CA or DEN and also synergistically induced by DEN/CA cotreatment. Basal p53 expression was very strong in FXR^hep−/− mice compared with that of WT mice, but treatment with CA or DEN reduced its expression. These changes in FXR^hep−/− mice in response to CA or DEN treatment were opposite to that of WT mice, suggesting that FXR is critical in regulating p53 expression in the liver. Levels of Gadd45α expression followed the same trend as p53 with the treatment of vehicle, CA, and DEN in both WT and FXR^hep−/− mice. But under DEN/CA treatment, despite increased p53 levels, the levels of Gadd45a were very low. Survivin levels were slightly higher in FXR^hep−/− mice than those in WT mice, but treatment with DEN/CA markedly diminished its levels.

Activation of MAPK pathway.

Increased bile acids are known to activate MAPK signaling pathways, especially via activation of JNK and ERK (5, 19). However, in our study, we found that the total JNK1/2 levels were not changed in these groups, but the p-JNK1/2 levels were markedly reduced by CA and/or DEN treatments (Fig. 6). Even though the total bile acid levels were not increased by hepatocyte-specific deletion of FXR, the FXR^hep−/− mice had increased total and phosphorylated c-Jun, which was, surprisingly reduced by CA and/or DEN treatment (Fig. 6). CA treatment increased p-Erk1 in FXR^hep−/− mice and DEN/CA increased p-Erk1 in both strains. The results suggest that long-term bile acid treatment may desensitize MAPK activation to some degree, and levels of T- or p-c-Jun were negatively, whereas the levels of p-Erk1 were positively, correlated with liver tumor formation in this study.

Activation of Stat3 pathway.

Stat3 activation has been demonstrated to promote cell cycle progression and tumorigenesis by increasing the gene transcription of genes involved in prosurvival and proinflammatory pathways (3, 6, 30). Our previous study also indicates that constitutive activation of Stat3 may be involved in tumorigenesis in liver of FXR whole body knockout mice (14). Here we further examined activation of the Stat3 pathway in FXR^hep−/− mice (Fig. 7). The FXR^hep−/− mice appeared to have higher Stat3 activation, suggested by increased GP130, JAK1, JAK2, and p- and total-Stat3, and decreased Socs3. This result is consistent with our previous report that there was a FXR/Socs3/Stat3 regulatory loop by which FXR positively induces Socs3 to suppress Stat3 activation (14). In the FXR whole body KO mice, p-Stat3 was increased at both Y705 and S727; however, in the present study, both CA and DEN treatment resulted in downregulation of total and p-Stat3 (Y705) protein levels, whereas p-Stat3 (S727) protein levels were increased. This differential phosphorylation of Stat3 protein in FXR whole body KO and FXR^hep−/− mice suggests a more complicated regulation of Stat3 activation by FXR deficiency and increased bile acid concentration.

Altered liver bile acids profiles in FXR knockout mice.

The gathered data shown in this study indicate that hepatic FXR serves a cytoprotective role during the initiation stage of carcinogenesis. As the master regulator of bile acid homeostasis, FXR regulates the synthesis and transport of bile acids. Moreover, the alteration in bile acid pool and composition could play a role in the tumorigenesis, especially tumor promotion. The total bile acid pool size has also been published by us previously, showing that bile acid pool size was only slightly increased in FXR^hep−/− mice comparing with WT mice (9). The liver bile acid composition and bile acid pool were measured in the present study by the UPLC/MS method (Fig. 8). Levels of four bile acids (GDCA, LCA, GUDCA, and T-ω-MCA) were below the UPLC-MS detection limits among 23 bile acids measured. For bile acids that have relatively higher concentration in the liver (Fig. 8A, left) and compared with WT mice, the relative percentage of β-MCA and ω-MCA in the liver was increased significantly in FXR KO and FXR^hep−/− mice. Among the bile acids that have relatively lower concentration in the liver, CDCA, α-MCA, and HDCA were also significantly higher in FXR^hep−/− mice (Fig. 8A, right). In the total bile acid pool, the individual bile acid composition was similar to that in the liver (Fig. 8B). Overall, the levels of ω-MCA, DCA, CDCA, and α-MCA were significantly increased in FXR^hep−/− mice, whereas HDCA and β-MCA also showed a slight increase in these mice compared with WT mice.

Supporting Data Table 1

Supporting Data Table 2