Fibroblast Growth Factor 15–Dependent and Bile Acid–Independent Promotion of Liver Regeneration in Mice

ANIMALS AND TREATMENTS

Tissue-specific, inducible Fgf15 Tg mice were generated on the C57BL/6J background (Supporting Information). Whole-body Fgf15 knockout (KO) mice were described previously.22, 23 Male mice (n = 5-12/group) were used in this study, with age-matched C57BL/6J mice as wild-type (WT) controls. Mice were maintained in a pathogen-free animal facility at 22°C under the 12-hour light/dark cycle. The animal studies were approved by the Rutgers University Institutional Animal Care and Use Committee and comply with all relevant federal guidelines and institutional policies. Fgf15 Tg and WT mice were administered by oral gavage with water (10 mL/kg body weight [BW]) containing 2 mg/mL doxycycline (Dox, Sigma, D9891) daily for 5 days to stop the Fgf15 transgene expression. Two days after the treatment with Dox, 2/3 PHx was performed as described previously,22 with details in the Supporting Information. Adeno-associated virus (AAV)–mediated overexpression of Fgf15 was achieved in Fgf15 KO and WT mice through a single tail-vein injection of 1 × 10¹³ plaque-forming units/kg of AAV or AAV-Fgf15. Mice were kept in quarantine for 7 days before undergoing 2/3 PHx. Tissues were collected 48 hours after surgery. For cholic acid (CA) feeding, WT and Fgf15 Tg mice were fed chow diet containing 0% or 0.1% CA for 7 days. Recombinant Fgf15 protein (rFgf15) produced in E. coli and purified as described previously24, 25 was injected into mice through tail vein at a dosage of 50 μg/kg. Tissues were collected 3 hours after injection.

SERUM BIOCHEMICAL PARAMETER ANALYSIS

Serum activities/levels of alanine transaminase (ALT), aspartate transaminase (AST), Alkaline phosphatase (ALP), total triglycerides (TGs), total cholesterol (TC), and total bilirubin were determined by kits from Pointe Scientific (Canton, MI) and the total BA assay kit was from Diazyme Laboratories (Poway, CA).

DETERMINATION OF BA POOL SIZE

The total BA pool size was determined and expressed as micromoles of BAs per 100 g BW (μmol/100 g BW). The BA concentration in the liver was expressed as μmol/g liver weight (LW), as described previously13 and with details in the Supporting Information.

HISTOLOGICAL AND IMMUNOHISTOCHEMICAL ANALYSIS

Hematoxylin and eosin (H&E) staining and immunohistochemistry staining of Fgf15 and bromodeoxyuridine (BrdU) are described in detail in the Supporting Information.

ISOLATION OF RNA, REAL-TIME QUANTITATIVE PCR, AND WESTERN BLOT ANALYSIS

Samples were prepared and analyzed according to previous publications.13, 22 Gene expression at mRNA levels was normalized to the housekeeping gene, β-Actin, with primers listed in the Supporting Table S1. β-actin or α-tubulin protein levels were used as the loading control for western blot analysis.

STATISTICAL ANALYSIS

All experimental data are expressed as the mean ± SD. Statistical differences among multiple groups were tested using two-way analysis of variance followed by the Student-Newman-Keuls test. A value of P < 0.05 was considered statistically significant.

GENERATION AND CHARACTERIZATION OF Fgf15 TG MICE

To determine the direct effect of Fgf15 on liver regeneration in vivo, a line of Fgf15 Tg mice were generated by using the Tet-Off expression system to create a tissue-specific expression of Fgf15 that can be turned off by Dox treatment (Fig. 1A). Specifically, two lines of mice were cross-bred to make the Fgf15 Tg mice, with one expressing tetracycline transactivator fusion protein (tTA) driven by the promoter of fatty acid binding protein (FABP) 1 gene for tissue-specific overexpression of tTA protein in both the liver and intestine,26, 27 and with the other line carrying a tTA-responsive element (TRE) consisting of a repeated tetracycline operator linked to the target gene of interest, herein Fgf15 (described in detail in the Supporting Information). Therefore, Fgf15 Tg mice have Fgf15 overexpression in both enterocytes and hepatocytes where the FABP1 promoter functions. Oral administration of Dox by gavage can efficiently prevent the binding of tTA protein to the TRE sequence, which prevents Fgf15 transgene overexpression. The time course of changes in Fgf15 mRNA levels following Dox treatment is shown in Supporting Fig. S1. Fgf15 Tg mice developed and reproduced normally. When the Tg mice reached 2 to 3 months of age, the serum activities/levels of ALT, AST, ALP, TGs, TC, BAs, and total bilirubin were measured. Despite elevated serum BA levels in Tg mice compared with WT mice, these levels were still within the normal range. All other serum parameters were similar between Tg and WT mice (Fig. 1B). Fgf15 Tg mice had lower BW and LW/BW ratio compared with WT mice (Fig. 1C). No significant difference in liver histology/pathology was observed between WT and Tg mice (Fig. 1D). In WT mice, Fgf15 protein was expressed in differentiated epithelial cells of the intestine and not detected in the livers, but in Fgf15 Tg mice, strong Fgf15 protein expression was detected in both periportal hepatocytes and intestinal epithelial cells (Fig. 1E).

Levels of Fgf15 mRNA in Tg mice were 300-fold to 600-fold higher than those in WT mice (Fig. 1F). Treatment with Dox for 5 days terminated the Fgf15 overexpression in both liver and intestine, and interestingly, 5 days after Dox withdrawal, the expression of Fgf15 transgene was only recovered in the intestine but not the liver (+/−Dox, Fig. 1F), suggesting that this strategy could be used as an intestine-specific Fgf15 overexpression model. With overexpression either in both the liver and intestine (vehicle [Veh], Fig. 1F) or only in intestine (+/−Dox, Fig. 1F), Fgf15 Tg mice showed markedly reduced mRNA levels of Cyp7a1, Cyp8b1 and Cyp7b1, but not Cyp27a1 (Fig. 1F). The total BA pool size in Tg mice was 20% to 30% that in WT mice (Fig. 1G). Specifically, Fgf15 overexpression did not change BA levels in the liver, but reduced BA concentration in the intestines (Fig. 1G), where most of the BAs are stored in the body. To specifically determine the role of Fgf15 overexpression in the intestine, the Tg mice treated with Dox for 5 days were used for all of the subsequent experiments.

OVEREXPRESSION OF Fgf15 REGULATES GENES INVOLVED IN BA HOMEOSTASIS AND CELL PROLIFERATION WITH OR WITHOUT BA SUPPLEMENT

To restore their BA pool size, Fgf15 Tg mice were fed a 0.1% CA-containing diet (Fig. 2). With CA feeding, Tg mice showed a similar increase in liver BA levels compared with WT mice, but markedly increased BA levels in the intestine, and as a result, the total BA pool size increased in Tg mice (Fig. 2A). Hepatic Fgf15 mRNA levels were induced with CA feeding, with a significantly higher induction in Tg than in WT mice. Suppression of the gene expression of Cyp7a1, Cyp8b1, Cyp7b1, and Cyp27a1 was achieved with CA feeding, and further suppression of Cyp7a1 and Cyp8b1 was noted in Fgf15 Tg mice (Fig. 2B). Without CA feeding, Fgf15 Tg mice had lower mRNA levels of Shp compared with WT mice. The CA feeding reduced mRNA levels of Ntcp, but increased mRNA levels of Shp, Bsep, and Ostβ in WT mice (with the latter two not significant). However, CA feeding did not alter Ntcp and Shp mRNA levels in Fgf15 Tg mice, and markedly increased those of Bsep and Ostβ (Fig. 2B). In the ileum, the higher Fgf15 expression in Tg mice was associated with lower expression of Ibabp, which was highly induced by CA in both WT and Tg mice. Even though on chow diet the Ostβ expression was reduced in Fgf15 Tg mice compared with WT, CA feeding significantly increased Ostβ expression only in Fgf15 Tg mice (Fig. 2C).

A previous study showed that BAs stimulate hepatocyte proliferation.16 Therefore, to determine to what extent Fgf15 overexpression enhances cell cycle progression directly without liver surgery, levels of proteins involved in cell cycle progression were determined in Tg mice without (basal level) and with CA feeding. Without CA feeding, although the BA pool size was very low in the Tg mice, the basal mRNA levels of Ccnd1 (Cyclin D1), Pcna (proliferating cell nuclear antigeph [PCNA]), and Cdkn1a (cyclin-dependent kinase inhibitor 1A) (p21), but not Ccne1 (Cyclin E1), were significantly higher than those in WT mice (Fig. 2D). With CA feeding, WT mice did not show an increase in the mRNA levels of Ccnd1, Ccne1, and Pcna except for Cdkn1a, which were highly induced. In Tg mice, CA further induced the expression of all these mediators (Fig. 2D). The protein expression of cell-signaling mediators related to cell proliferation and Fgfr4 was further determined. JNK and ERK signaling pathways were activated in the Tg mice, as evidenced by increased p-JNK and p-ERK levels, as well as their downstream targets including c-Jun and c-Fos, but not c-myelocytomatosis (c-Myc). CA increased the protein levels of Cyclin D1 and PCNA but reduced those of Cyclin E1 and c-Myc, in WT and Tg mice (Fig. 2E). Overall, these results suggest that Fgf15 may directly promote cell proliferation in addition to modulating BA levels at basal levels.

OVEREXPRESSION OF Fgf15 PROMOTES LIVER REGENERATION FOLLOWING 2/3 PHx

The induction of pro-proliferative genes in the liver by Fgf15 transgene prompted us to determine the effects of Fgf15 overexpression on liver regeneration at low BA levels. PHx was performed in Tg mice with overexpression of Fgf15 in both liver and intestine (to mimic human expression pattern)-Veh/PHx, and in Tg mice with Fgf15 overexpression only in the intestine by treatment with Dox for 5 days followed by Dox-free for 2 days (mouse expression pattern)-Dox/PHx (Fig. 3A). Twenty-four hours following PHx, the total BA pool size in Tg mice remained lower (Fig. 3B). Ccnd1 and Ccne1 gene expression started to increase 24 hours post PHx,22 and the current study showed that overexpression of Fgf15 in both liver and intestine induced Ccnd1, Ccne1, and Cdkn1a following PHx, whereas overexpression of Fgf15 only in the intestine induced Ccnd1 and Ccne1 to a lesser extent, and significantly increased levels of Cdkn1a (Fig. 3C), suggesting that Fgf15 in both liver and intestine contributes to cell proliferation after PHx. The ERK, but not JNK, pathway showed higher activation following PHx in Tg mice (Fig. 3D). The protein levels of Cyclin D1, Cyclin E1, Cdk4, PCNA, and c-Myc, but not c-Jun, were markedly increased in Fgf15 Tg mice (Fig. 3D).

Thirty-six hours following PHx, overexpression of Fgf15 in the liver remained to be suppressed by Dox, but it was fully recovered in the intestine (Fig. 4A). At this time point, Ccnd1 gene expression was still much higher in mice with Fgf15 overexpression in both the liver and intestine (Veh/PHx), but not in Tg mice with overexpression of Fgf15 in intestine only (Dox/PHx). Ccne1 expression tended to be lower in Fgf15 Tg than WT mice, which may reflect a compensation. DNA proliferation, measured by BrdU incorporation in hepatocytes during liver regeneration, was higher in Tg mice with Fgf15 overexpression in both liver and intestine (Veh/PHx) compared with WT (Fig. 4B). Interestingly, Dox treatment markedly increased BrdU incorporation in WT mice with PHx (Dox/PHx), and to a similar degree when Tg mice was under the same treatment (Fig. 4B). At 36 hours after PHx, the BA pool size remained lower in Tg mice, but the hepatic BA levels were similar between WT and Tg mice, and it is noted that Tg mice tended to have higher BA levels in the liver (Fig. 4C). Fgf15 overexpression, regardless in both liver and intestine (Veh/PHx) or in intestine only (Dox/PHx), led to higher expression of cell cycle–related proteins, including Cyclin D1, Cdk4, and c-Myc (Fig. 4D). Intestine-overexpression of Fgf15 reduced c-Jun protein levels at this time point after PHx (Fig. 4D).

In another independent gain-of-function model, Fgf15 was overexpressed in the liver by AAV-mediated transduction in Fgf15 KO mice. Transduction by AAV-Fgf15 led to reduced BW, and decreased mortality following PHx (from 80% to 30%) (Fig. 5A). This liver-specific AAV-mediated Fgf15 overexpression reduced mRNA levels of hepatic Shp and intestinal Ibabp genes induced by BAs (Fig. 5B). The Fgf15 KO mice normally had higher BA levels (Fig. 5C), and following PHx, they presented with 8-fold to 10-fold higher hepatic BA concentration than WT mice (Fig. 5C). AAV-Fgf15 overexpression inhibited BA synthesis, resulting in a significant reduction in liver BA concentration, with the total BA pool size returning to levels in WT mice (Fig. 5C, AAV-Fgf15/PHx). Furthermore, the protein levels of Cyclin D1, Cyclin E1, and PCNA were induced in the AAV-Fgf15 mice (Fig. 5D).

Fgf15 ENHANCES HEPATOCYTE PRIMING IN LIVER REGENERATION FOLLWOING 2/3 PHx

The effects of Fgf15 on the priming of liver regeneration were examined at 30 minutes and 60 minutes following PHx in WT and Fgf15 Tg mice, by determining the mRNA and protein levels of genes involved in the initial stage of cell proliferation (Fig. 6). At 30 minutes, PHx increased the mRNA levels of c-Jun and c-Fos by 5-fold and 100-fold, respectively, in WT mice, and 13-fold and 250-fold, respectively, in Tg mice (Fig. 6A). This marked induction returned to baseline levels at 60 minutes (Fig. 6A). A marked induction of c-Myc and Igfbp1 was also observed, which reached their maximum expression at 60 minutes, and with significant difference in c-Myc expression between WT and Fgf15 Tg mice (Fig. 6A).

Activation of signal transducer and activator of transcription 3 (Stat3) with PHx, measured by increased p-Stat3 (S727) and p-Stat3 (Y705), is well-known as one of the major priming events during liver regeneration. The basal levels of p-Stat3 (S727) and p-Stat3 (Y705) appeared to be higher in Fgf15 Tg mice (0 minutes) (Fig. 6B). Both WT and Fgf15 Tg mice demonstrated increased activation of Stat3 at 30 minutes and 60 minutes after PHx, but Tg mice had a higher induction of p-Stat3 (S727) at 30 minutes (Fig. 6B). NF-κB signaling pathway is activated by IKK-mediated phosphorylation of the inhibitory subunit IκBα at Ser32 and Ser36, which leads to proteasomal degradation of inhibitor IκB, thus releasing NF-κB subunits and activating gene transcription in the nucleus.28 The phosphorylation and activation of NF-κB (p65) at Ser536 and p-IκBα at Ser32 in Fgf15 Tg mice was significantly higher than in WT mice at 30 minutes after PHx, and returned to basal levels in both strains at 60 minutes after PHx. There were no significant changes in protein level for NF-κB (p65) and IκBα. The basal levels of c-Jun, c-Fos, and c-Myc tended to be higher in Tg mice (0 minutes) but were not further induced by PHx in both strains (Supporting Fig. S2).

To further determine the signaling cascades activated by Fgf15, Fgf15 KO mice were injected with the recombinant Fgf15 protein (rFgf15) that was developed in our lab. The rFgf15 was functional, as it significantly suppressed Cyp7a1 gene expression (Fig. 6C). Despite having little effect on the gene expression of c-Jun and c-Fos (Supporting Fig. S3), injection of rFgf15 increased p-JNK, p-ERK and p-Stat3, indicating that Fgf15 could activate both mitogen-activated protein kinase (MAPK) and Stat3 signaling cascades in vivo (Fig. 6D).

Fgf15 MAY REGULATE THE TERMINATION OF LIVER REGENERATION AFTER 2/3 PHx

In mice, the liver regenerates and grows back to its original mass within 14 days after 2/3 PHx. Hepatocytes reenter the cell cycle and replicate after incision and return to quiescence state after liver recovers its mass.2 However, the mechanisms for the termination of regeneration are complex and not fully understood. Extensive studies have elucidated that the hippo pathway plays a critical role in organ size determination and is involved in the termination stage of liver regeneration. Activation of the hippo pathway leads to phosphorylation of Yes-associated protein (Yap) and transcriptional co-activator with PDZ binding motif (Taz), which promotes Yap/Taz subsequent cytoplasmic sequestration and inactivation.29, 30

The effects of Fgf15 on liver regeneration termination is unknown. To study this effect, activation of the hippo pathway was assessed in Fgf15 Tg mice at day 7 and day 14 after PHx (Fig. 7). Consistent with the previous result, Fgf15 Tg mice showed lower LW/BW ratio compared with WT mice (Fig. 7A, Sham). Livers in WT mice were still regenerating at day 7 after PHx and recovered their original weight 14 days after, whereas livers in Fgf15 Tg mice stopped growing at day 7 following PHx and kept a similar weight at 14 days after PHx (Fig. 7A). Consistent with the LW changes, Cyclin D1 level was still much higher in WT mice 7 days following PHx, but similar in Fgf15 Tg mouse livers between Sham and PHx, indicating that hepatocyte replication remained active in WT but not in Tg mice (Fig. 7B). There were no significant differences for p-Yap, Yap, and Taz in WT mice between Sham and PHx, but in Fgf15 Tg mice, a slight decrease in p-Yap and Taz levels following PHx, indicating an activation of the hippo pathway in Fgf15 Tg mice 7 days after PHx (Fig. 7B). At day 14 after PHx, hepatocytes in both WT and Fgf15 Tg mice returned to a more quiescence state, revealed by similar Cyclin D1 levels between the Sham and PHx groups (Fig. 7C). Compared with the WT/Sham group, the protein levels of p-Yap, Yap, and Taz were much higher in Fgf15 Tg mice (Fig. 7C). PHx decreased p-Yap (inactive form) but increased Yap (active form) and Taz expression in WT. However, compared with Sham, PHx significantly decreased both p-Yap and Yap in Fgf15 Tg mice (Fig. 7C). These data indicate that after PHx, hepatocytes from WT and Fgf15 Tg mice may have entered termination stage differently and Fgf15 may have promoted the hippo pathway activity after hepatectomy in mice.

To determine the effects of Fgf15 on changes in biliary mass after PHx, biliary mass in WT and Fgf15 Tg mice was measured at 7 days and 14 days after PHx (Supporting Fig. S4). However, there was no significant difference in the biliary mass between WT and Fgf15 Tg mice.