Beta-catenin signaling in murine liver zonation and regeneration: A Wnt-Wnt situation!


  • Potential conflict of interest: Dr. Williams received grants from Genentech.

  • Funded by NIH grants 1R01DK62277, 1R01DK100287 and Endowed Chair for Experimental Pathology to SPSM. This study was also in part funded by 5R01AR053293 and 5R21RR024887 to BOW.


Liver-specific β-catenin knockout (β-Catenin-LKO) mice have revealed an essential role of β-catenin in metabolic zonation where it regulates pericentral gene expression and in initiating liver regeneration (LR) after partial hepatectomy (PH), by regulating expression of Cyclin-D1. However, what regulates β-catenin activity in these events remains an enigma. Here we investigate to what extent β-catenin activation is Wnt-signaling-dependent and the potential cell source of Wnts. We studied liver-specific Lrp5/6 KO (Lrp-LKO) mice where Wnt-signaling was abolished in hepatocytes while the β-catenin gene remained intact. Intriguingly, like β-catenin-LKO mice, Lrp-LKO exhibited a defect in metabolic zonation observed as a lack of glutamine synthetase (GS), Cyp1a2, and Cyp2e1. Lrp-LKO also displayed a significant delay in initiation of LR due to the absence of β-catenin-TCF4 association and lack of Cyclin-D1. To address the source of Wnt proteins in liver, we investigated conditional Wntless (Wls) KO mice, which lacked the ability to secrete Wnts from either liver epithelial cells (Wls-LKO), or macrophages including Kupffer cells (Wls-MKO), or endothelial cells (Wls-EKO). While Wls-EKO was embryonic lethal precluding further analysis in adult hepatic homeostasis and growth, Wls-LKO and Wls-MKO were viable but did not show any defect in hepatic zonation. Wls-LKO showed normal initiation of LR; however, Wls-MKO showed a significant but temporal deficit in LR that was associated with decreased β-catenin-TCF4 association and diminished Cyclin-D1 expression. Conclusion: Wnt-signaling is the major upstream effector of β-catenin activity in pericentral hepatocytes and during LR. Hepatocytes, cholangiocytes, or macrophages are not the source of Wnts in regulating hepatic zonation. However, Kupffer cells are a major contributing source of Wnt secretion necessary for β-catenin activation during LR. (Hepatology 2014;60:964–976)


serum alanine aminotransferase


aspartate aminotransferase




liver-specific β-catenin knockout


littermate controls


cytochrome P450 enzymes


epidermal growth factor


Evenness Interrupted


glutamine synthetase


hepatocyte growth factor




liver regeneration


LDL-related protein


liver-specific LRP5/6 double KO


liver weight to body weight ratio


proliferating cell nuclear antigen


polymerase chain reaction


partial hepatectomy


protein kinase A


T cell factor


western blot




endothelial cell-specific Wls KO


liver-specific Wls KO


macrophage-specific Wls KO

Beta-catenin is a transcriptional coactivator that plays a critical role in liver biology. In liver homeostasis, hepatocytes exhibit molecular heterogeneity based on their location within the hepatic lobule, which is known as hepatic zonation. β-Catenin is one of the key molecules regulating the zonation pattern. Pericentral hepatocytes express cytoplasmic and nuclear β-catenin in addition to membranous localization. Here, β-catenin regulates the expression of genes such as glutamine synthetase (GS) and certain cytochrome P450 enzymes (CYPs), such as Cyp1a2 and Cyp2e1.1 In addition, β-catenin signaling is also essential for the initiation of liver regeneration (LR).[2] Liver has the unique capacity to regenerate following partial hepatectomy (PH). During LR a series of cell signaling pathways and cascades are triggered that are tightly regulated. β-Catenin signaling is one such pathway that is activated very early after PH.[3] Indeed, we and others have previously shown that liver-specific β-catenin knockout (β-catenin-LKO) that lack this protein in both hepatocytes and cholangiocytes show defective pericentral gene expression and a 24-hour delay in entry of hepatocytes to S-phase after PH that peaks at 72 hours instead of 40 hours.[4-6] However, what the upstream regulator of β-catenin is in these events remains unknown.

Numerous signaling cascades can activate β-catenin. Canonical Wnt-signaling is the major pathway that induces β-catenin activation.[7] Wnt is an extracellular glycoprotein secreted by various types of cells. Binding of Wnt to its cell surface receptor Frizzled and coreceptor low-density lipoprotein related protein-5 (Lrp5) or Lrp6 stabilizes β-catenin protein that in turn translocates to the nucleus to bind to T-cell factor (TCF) family of transcription factors to activate gene expression of tissue- and context-specific targets encoding for GS, Cyp1a2, Cyp2e1, and Cyclin-D1. β-Catenin can also be activated in a Wnt-independent manner. Some examples include β-catenin activation by hepatocyte growth factor (HGF) through phosphorylation at tyrosine-654 (Y654),[8] epidermal growth factor (EGF) by phosphorylation at Y654,[9, 10] Flt3 also by phosphorylation at Y654,[11] and protein kinase A (PKA) by phosphorylation at serine-552 (S552) and S675.12,13

In the current study, in order to address to what extent β-catenin signaling in liver is Wnt-signaling-dependent, we generated liver-specific Lrp5/6 double knockout or Lrp-LKO mice where upstream Wnt-signaling is disrupted while β-catenin expression is intact; therefore, β-catenin can still be activated by pathways other than Wnts. We found that Lrp-LKO mice phenocopied β-catenin-LKO in defective hepatic zonation and delay in LR, which suggests that β-catenin is primarily regulated by Wnt proteins in liver homeostasis and during LR. To address the cellular source of Wnt-proteins in liver, we used Wntless-floxed mice. Wntless (Wls) encodes a multipass transmembrane protein that is specific and necessary for Wnt transport from Golgi to the membrane for secretion.[14] Cell-specific Wls deletion has displayed important roles in the development of retina, brain, and bone.[15-17] We generated Wls conditional KO for liver epithelial cells (Wls-LKO), endothelial cells (Wls-EKO), and macrophages including Kupffer cells (Wls-MKO). We show that Wls-EKO was lethal during development, while heterozygous showed no decrease in Wls protein. Wls-LKO and Wls-MKO mice were viable but lacked any defect in hepatic zonation. Wls-LKO mice exhibit normal initiation of LR. However, a temporal defect in LR initiation was evident in Wls-MKO. Thus, we have identified an important cellular circuitry regulating β-catenin activation in liver at baseline and after PH.

Materials and Methods

Mice and Breeding

All animal experiments and procedures were performed under the strict guidelines of the National Institutes of Health and after approval by the Institutional Animal Use and Care Committees at the University of Pittsburgh and the Van Andel Research Institute. Homozygous Lrp5/6 double-floxed mice were reported recently.[18] To conditionally delete Lrp5 and Lrp6 from hepatocytes and cholangiocytes, homozygous Lrp5/6 double-floxed mice (Lrp5flox/floxLrp6flox/flox) were bred to Cre transgenic mice driven by an albumin promoter (Albumin-Cre) (Jackson Laboratories, Bar Harbor, ME).[19] The offspring carrying homozygous Lrp5 floxed alleles, a floxed Lrp6 allele, and an albumin-Cre allele (Lrp5flox/floxLrp6flox/wtAlb-Cre+/) were then bred to homozygous Lrp5/6 double-floxed mice (Lrp5flox/floxLrp6flox/flox). The mice with genotype Lrp5flox/floxLrp6flox/floxAlb-Cre+/ represent liver-specific Lrp5/6 KO or Lrp-LKO mice. Other genotypes Lrp5flox/wt; Lrp6flox/wt;Alb-Cre/ and Lrp5flox/flox; Lrp6flox/lwt;Alb-Cre/ and Lrp5flox/wt; Lrp6flox/flox;Alb-Cre/ and Lrp5flox/flox; Lrp6flox/flox; Alb-Cre/ were used as controls (Con). No phenotype was observed in Con. Genotyping was performed by polymerase chain reaction (PCR) analysis using genomic DNA isolated from a tail clipping, as available in the online Supporting Material.

In order to generate liver-specific Wls KO, homozygous Wls floxed mice (Wlsflox/flox) were bred with Albumin-Cre+/ mice (Jackson Laboratories).[19] The offspring carrying floxed Wls allele and Albumin-Cre (Wlsflox/wt; Alb-Cre+/) were bred to homozygous Wls floxed mice (Wlsflox/flox). Mice with genotype Wlsflox/flox; Alb-Cre+/ represent Wls-LKO mice. Mice with genotypes Wlsflox/flox; Alb-Cre/ and Wlsflox/wt; Alb-Cre/ were used as Con.

To generate macrophage-specific Wls KO mice, Wlsflox/flox mice were bred with Lyz2-Cre+/ (also called LyzM-Cre) mice (Jackson Laboratories)[20] using a similar strategy as described above. Wlsflox/flox; Lyz2-Cre+/ represent Wls-MKO and Wlsflox/flox; Lyz2-Cre/ and Wlsflox/wt; Lyz2-Cre/ as Con.

To generate endothelial cell-specific Wls knockout mice, Wlsflox/flox mice were bred to Tie2-Cre+/ mice (Jackson Laboratories)[21] in the same manner as above to obtain Wlsflox/flox; Tie2-Cre+/ or Wls-EKO mice. Since, no viable pups for Wls-EKO were obtained, mice with genotype Wlsflox/wt; Lyz2-Cre+/ or heterozygous Wls-EKO were used for experiments and mice with phenotypes Wlsflox/flox; Lyz2-Cre/ and Wlsflox/wt; Lyz2-Cre/ were used as Con.

Partial Hepatectomy

Twelve-week-old male Con or KO (Lrp-LKO, Wls-LKO, Wls-MKO, or heterozygous Wls-EKO) were subjected to partial hepatectomy (PH).[22] Equal numbers of KO and Con mice were killed by cervical dislocation after isoflurane anesthesia at different timepoints: 4 hours (n = 3), 24 houses (n = 3), 40 hours (n ≥ 3), 3 days (n ≥ 3), 4 days (n = 3), and 5 days (n = 3) after PH. The regenerating livers were harvested and used for protein extraction and paraffin embedding as described elsewhere.[23]

Separation and Staining of Macrophages

Nonparenchymal cells from mouse liver were isolated by 2-step collagenase perfusion.[24] Macrophage positive selection from nonparenchymal cells was performed using the QuadroMACS column separation kit (Miltenyi Biotech, Cambridge, MA). Antimouse F4/80 antibody (Biolegend, San Diego, CA) and specific microbeads were used according to the manufacture's instruction. After the column separation of macrophages, cytospin was performed. Cells were centrifuged at 500 rpm for 5 minutes on glass slides followed by fixation with 4% paraformaldehyde for 10 minutes. Rat antimouse F4/80 antibody (AbD Serotec, Raleigh, NC) was used for immunofluorescent staining as described elsewhere.[25]

Additional methods are available in the online Supporting Material.


Generation and Characterization of Conditional Lrp5/6 Null Mice

After strategic breeding, liver-specific Lrp5/6 null mice (Lrp-LKO) were generated. Genotyping was performed by PCR (Fig. 1A), using primer P1 and P3 described previously.[18] Livers from Lrp-LKO exhibited a significant decrease in Lrp5 and Lrp6 protein compared to control mice (Con) (Fig. 1A). Eight-month-old Lrp-LKO and Con mice had comparable and normal levels of serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), and albumin (data not shown). While total bilirubin was within the normal range, average in controls was 0.26 mg/dL (n = 5) as compared to 0.5 mg/dL (n = 5) in Lrp-LKO (P = 0.009). Lrp-LKO also displayed smaller livers and the liver weight to body weight (LW/BW) ratio was significantly lower in Lrp-LKO compared to Con (Fig. 1B). To address the discrepancy in size, we assessed hepatocytes in S-phase by proliferating cell nuclear antigen (PCNA) immunohistochemistry (IHC). Lrp-LKO livers had significantly less PCNA-positive cells at baseline (Fig. 1C,D). Concomitantly, lower Cyclin-D1 expression was also evident in KO by western blots (WB) (Fig. 1E). Intriguingly, all the above observations were similar to β-catenin-LKO mice.[6]

Figure 1.

Successful deletion of Lrp5 and Lrp6 in liver leads to alterations in downstream β-catenin signaling (*P < 0.05, **P < 0.01). Representative PCR (upper panel) shows genotype of mice for identifying Lrp-LKO with genotype Lrp5flox/flox; Lrp6flox/flox ; Cre+/− (lane 1). Mice with other genotypes (lanes 2-5) were used as Con. Representative WB (lower panel) shows a dramatic decrease in Lrp5 and Lrp6 in the tissue lysates from the Lrp-LKO livers. Decreased LW/BW ratio in 8-month-old Lrp-LKO. This difference was around 20% in both males and females and is significant. IHC for PCNA shows an occasional positive hepatocyte (arrow) in a representative section from Con liver at baseline, while none were detected in Lrp-LKO. Quantification of PCNA staining shows significantly fewer positive cells in Lrp-LKO. Representative WB shows no change in total levels of β-catenin, γ-catenin, and E-cadherin in Lrp-LKO livers as compared to Con. However, several downstream targets of β-catenin were down-regulated in Lrp-LKO livers. Prolonged exposure of WB film enabled us to observe baseline Cyclin-D1 in Con, which was notably reduced in Lrp-LKO livers. IHC for β-catenin shows predominantly membranous localization in both Lrp-LKO and Con livers. However, in Con, cytosolic localization of β-catenin was seen in hepatocytes around central vein (CV) and not portal triads (PT). Lrp-LKO show absence of cytosolic β-catenin around CV with a few hepatocytes around PT showing some cytoplasmic localization (left panel Con and KO 100×; right panel Con and KO 200×).

β-Catenin at Adherens Junctions in the Lrp-LKO Livers at Baseline

Next, we assessed β-catenin levels and levels of other key components at adherens junctions in the Lrp-LKO livers. WB shows Lrp-LKO and Con to have comparable levels of β-catenin, γ-catenin, and E-cadherin (Fig. 1E).

Disruption of Metabolic Zonation in Conditional Lrp5/6 Null Mice

Next, we assessed β-catenin localization by IHC in Con and Lrp-LKO livers. Con livers show predominantly membranous β-catenin except in pericentral hepatocytes, where there was enhanced cytosolic labeling as well (Fig. 1F). However, in Lrp-LKO, cytosolic β-catenin localization in pericentral hepatocytes was notably lacking with, occasional periportal hepatocyte showing some cytoplasmic staining (Fig. 1F). Lrp-LKO livers also showed notably low protein expression of multiple downstream targets of β-catenin, such as Regucalcin, GS, Cyp1a2, and Cyp2e1 (Fig. 1E). IHC also showed loss of pericentral expression of GS, Cyp2e1, and Cyp1a2 in Lrp-LKO livers (Fig. 2A).

Figure 2.

Loss of Lrp5/6 in liver compromises pericentral zonation (*P < 0.05). IHC shows normal pericentral expression of GS, Cyp1a2, and Cyp2e1 in Con livers, while absence of these proteins is observed in littermate Lrp-LKO livers. When injected with a lethal dose of APAP (600 mg/kg body weight), Lrp-LKO showed significantly lower ALT and AST levels than Con after 24 hours.

The functional loss of Cyp1a2 and Cyp2e1 was assessed with an acetaminophen (APAP) toxicity study. Cyp1a2 and Cyp2e1 metabolize APAP to a toxic metabolite, N-acetyl-p-benzo-quinone imine (NAPQI) to cause hepatocyte necrosis through glutathione depletion.[26] β-Catenin-LKO mice are protected from APAP overdose due to absent Cyp2e1 and Cyp1a2.5,27 Lrp-LKO also showed significantly lower serum AST and ALT as compared to Con after APAP (600 mg/kg) administration (Fig. 2B). Therefore, despite normal β-catenin expression in Lrp-LKO, ablation of Lrp5/6 disrupts Wnt signaling to impair β-catenin activity and zonation.

Retarded Liver Regeneration in Lrp-LKO Mice After PH

Next, Lrp-LKO and Con were subjected to PH. As expected, Con showed abundant hepatocytes in S-phase at 40 and 72 hours after PH with a decline at later timepoints (Fig. 3A,B). Intriguingly, Lrp-LKO livers exhibited significantly fewer PCNA-positive hepatocytes at both 40 and 72 hours, although more PCNA-positive hepatocytes were observed at 72 than 40 hours (Fig. 3A,B). In addition, numbers of mitotic figures as assessed by analysis of hematoxylin and eosin (H&E) images from Lrp-LKO and Con showed a notably lower mitosis in hepatocytes in the former at 72 hours (Fig. 3C; Supporting Fig. 1A). β-Catenin target Cyclin-D1 that regulates G1-S phase transition during LR[28] was also lower in Lrp-LKO at 40 hours with progressive increase at 72 hours, although it was still lower than Con (Fig. 3D).

Figure 3.

Abolishing Wnt/β-catenin signaling through Lrp5/6 ablation in liver impairs LR after PH. IHC for PCNA identifies several hepatocytes in S-phase at 40 and 72 hours after PH, while only a few trailing hepatocytes were PCNA-positive at 96 and 120 hours. In contrast, Lrp-LKO showed dramatically fewer PCNA-positive hepatocytes at 40 hours with an increase at 72 hours. Like Con, only a few PCNA-positive hepatocytes were detected at later times in Lrp-LKO (200×). PCNA quantification shows a gradual increase in positive hepatocytes in Lrp-LKO from 40 to 72 hours, although these are lower than Con at both times (**P < 0.01). Quantification of mitotic figures shows significantly lower numbers in Lrp-LKO as compared to Con at 72 hours after PH (**P < 0.01). WB shows low protein expression of Cyclin-D1 after PH in Lrp-LKO compared to Con at both 40 and 72 hours although its levels increase in Lrp-LKO at 72 hours. Representative WB shows comparable levels of S675-β-catenin, S552-β-catenin, Y654-β-catenin, and total β-catenin in Con and Lrp-LKO livers after PH. Representative WB shows β-catenin and TCF4 association at 4 hours after PH in Con and not Lrp-LKO livers by immunoprecipitation studies. Immunoprecipitation studies were performed by pull down of either β-catenin or TCF4. Respective input controls are included in analysis as well.

We wanted to address if there was compensatory activation of β-catenin by alternate signals such as HGF, EGF, or PKA, in the absence of Wnt. However, no differences in levels of Y654-, S552-, or S675-β-catenin were evident between Lrp-LKO and Con at 40 or 72 hours after PH (Fig. 3E).

Since β-catenin binding to TCF4 precedes Cyclin-D1 expression after PH,[6, 24] we assessed their association by immunoprecipitation at 4 hours after PH. TCF4-β-catenin complex was observed in Con but absent in Lrp-LKO (Fig. 3F).

Taken together, these data suggest that Lrp-LKO phenocopy β-catenin-LKO after PH, thereby indicating that only Wnt signaling is responsible for β-catenin activation during LR after PH.

Generation and Baseline Characterization of Liver-Specific Wls KO

To address the cell source of Wnt proteins directing β-catenin activity for zonation and during LR, we first generated hepatocyte and cholangiocyte (or liver-specific) Wls KO (Wls-LKO) by interbreeding Albumin-Cre and Wls-floxed mice. PCR confirmed the concomitant presence of floxed allele and cre-recombinase in Wls-LKO (Fig. 4A), using primer P2 and P4 described previously.[15] WB from Wls-LKO liver lysates show less Wls relative to Con (Fig. 4B). However, appreciable Wls persisted in Wls-LKO, suggesting that other nonparenchymal cells also express Wls, which would not be affected by Albumin-cre. Indeed, cell fractionation into parenchymal and nonparenchymal cells by percoll gradient after collagenase perfusion showed notably higher Wls expression in the nonparenchymal cell compartment (data not shown).

Figure 4.

Successful deletion of Wls in Wls-LKO mice. Representative PCR shows genotype of mice for identifying the Wlsflox/flox; Albumin-Cre+/− (Wls-LKO) (lane 3). Mice with genotype Wlsflox/wt; Albumin-Cre−/− and Wlsflox/flox; Albumin-Cre−/− (lanes 2 and 4) were used as Con. WB shows a modest decrease in Wls in liver of Wls-LKO compared to Con. A higher LW/BW ratio in 8-month-old Wls-LKO as compared to Con, although the difference was not significant. IHC shows occasional PCNA-positive hepatocyte in Con and Wls-LKO liver (200×). Quantification of PCNA staining (200×) shows a comparable number of positive hepatocytes between Con and Wls-LKO. Representative WB shows comparable expression of β-catenin, relevant junctional components, and β-catenin downstream targets in Wls-LKO and Con livers. Prolonged exposure of WB film enabled us to observe baseline Cyclin-D1 in Wls-LKO and Con, which was comparable in the two groups.

Eight-month-old Wls-LKO mice had comparable and normal levels of serum AST, ALT, bilirubin, and albumin levels to Con (data not shown). Wls-LKO showed marginally bigger livers than Con (Fig. 4C), although the difference was not statistically significant. Baseline proliferation examined by IHC for PCNA did not reveal any differences (Fig. 4D,E).

Normal Adherens Junctions and Hepatic Zonation in Wls-LKO Null Mice

Next, we investigated changes in β-catenin, its downstream signaling and junctional proteins in Wls-LKO mice. β-Catenin, γ-catenin and E-cadherin were unchanged in Wls-LKO (Fig. 4F). The expression of the downstream targets of β-catenin signaling was comparable between Wls-LKO and Con (Fig. 4F). IHC for pericentral expression of β-catenin targets showed no differences in staining for GS, Cyp1a2, and Cyp2e1 between Wls-LKO and Con (Fig. 5A). Therefore, blocking Wnt secretion from hepatocytes or cholangiocytes does not affect regulation of zonation by β-catenin signaling in hepatocytes.

Figure 5.

Ablation of Wls in liver epithelial cells has no effect on hepatic zonation or initiation of LR. IHC for GS, Cyp1a2, and Cyp2e1 in Con and Wls-LKO shows normal pericentral localization in both groups (100×). PCNA IHC shows comparable numbers of PCNA-positive hepatocytes in Con and Wls-LKO livers at 40 hours after PH (200×). Quantification of PCNA staining confirms similar numbers of PCNA-positive hepatocytes in Con and Wls-LKO at 40 hours after PH. Representative WB shows comparable Cyclin-D1 expression at 40 hours after PH in both Con and Wls-LKO livers.

Normal Initiation of LR in Wls-LKO Null Mice

The role of hepatocyte- and cholangiocyte-derived Wnt proteins in LR was examined next by subjecting the Wls-LKO and Con to PH. Comparably higher numbers of PCNA-positive hepatocytes were observed in Wls-LKO and Con mice at 40 hours post-PH (Fig. 5B). Quantification of PCNA staining showed no significant differences (Fig. 5C). Likewise, Wls-LKO had comparable Cyclin-D1 levels to Con at 40 hours (Fig. 5D). Analysis at 14 days after PH when liver is known to have regenerated to its pre-hepatectomy mass also revealed no difference in PCNA-positivity between the Wls-LKO and Con group (Fig. 5B,C). Therefore, lack of any differences in regeneration kinetics at early and late times after PH indicate that β-catenin activation during LR is not dependent on hepatocyte- or cholangiocyte-derived Wnts.

Generation of Endothelial Cell-Specific Wls Null Mice

Deletion of Wls using Tie2-cre led to no viable pups, suggesting embryonic lethality. Heterozygous mice that lacked an allele of Wls in endothelial cells showed no decrease in Wls protein and these livers did not show any decrease in pericentral targets (data not shown). When subjected to PH, no defect in LR was observed in heterozygous Wls-EKO (data not shown).

Generation and Baseline Characterization of Macrophage-Specific Wls Null Mice

We next bred the Wls-floxed mice to macrophage-specific Lyz2-Cre mice.[29] PCR was used to identify the Wls-MKO and Con (Fig. 6A). Decreased level of Wls protein was evident in WB from Wls-MKO livers (Fig. 6A). To verify successful loss of Wls, Kupffer cells were isolated by QuadroMACS column separation as discussed in the Materials and Methods and stained for marker F4/80 to demonstrate a notable enrichment (Fig. 6B). Quantitative reverse-transcription (RT)-PCR performed on enriched Kupffer cell-fraction showed a significant decrease in Wls expression in Wls-MKO as compared to Con (Fig. 6B).

Figure 6.

Impact of deletion of Wls in Kupffer cells on hepatic homeostasis. Representative PCR (left) showing genotype of Wls-MKO(Wlsflox/flox; Lyz2-Cre+/−) (lane 3) and Con with genotype Wlsflox/wt; Lyz2-Cre−/− and Wlsflox/flox; Lyz2-Cre−/− (lanes 2 and 4). Representative WB shows a notable decrease in Wls in Wls-MKO liver lysate. Enrichment for Kupffer cells after collagenase perfusion and column separation of nonparenchymal cells was verified by immunocytochemistry on cytospin for F4/80 (red) showing positive staining in the F4/80-positive cell fraction and not in the F4/80-negative fraction (left). Wls mRNA expression in F4/80+ cells isolated from Con and Wls-MKO, shows significantly lower expression in the latter (**P < 0.001). Bar graph depicts marginally lower LW/BW ratio in Wls-MKO, although the difference was not statistically significant. IHC (200×) shows an occasional PCNA-positive hepatocyte in Con at baseline (arrow), while no positive cell is observed in a representative view in Wls-MKO. Quantification of PCNA staining showing less positive cells in Wls-Mac KO than Con at baseline, although the differences were insignificant. WB shows no alteration in proteins levels of β-catenin, junctional proteins, and pericentral targets. However, Cyclin-D1 was notably lower in Wls-MKO livers.

Eight-month-old Wls-MKO and Con mice had comparable and normal serum levels of AST, ALT, bilirubin, and albumin (data not shown). Wls-MKO mice (n = 5) had marginally smaller livers than Con (n = 3); however, the difference did not reach statistical significance (P = 0.079) (Fig. 6C). IHC for PCNA identified fewer positive cells in Wls-MKO, although the differences were not statistically significant (Fig. 6D, 6E). Intriguingly, a notable decrease in baseline Cyclin-D1 protein expression was evident in Wls-MKO (Fig. 6F). Hence, we conclude that macrophages may secrete Wnts and contribute to baseline hepatocyte turnover to govern liver size.

Normal Adherens Junctions and Hepatic Zonation in Wls-MKO Mice

Next, the effect of abolishing Wnt secretion from macrophages on β-catenin levels and other related junctional proteins was assessed. WB showed no notable changes in the levels of β-catenin, γ-catenin, or E-cadherin in Wls-MKO when compared to Con livers (Fig. 6F). WB also shows comparable levels of Regucalcin, GS, Cyp1a2, and Cyp2e1 in Wls-MKO and Con livers (Fig. 6F). More important, comparable GS, Cyp1a2, and Cyp2e1 localization confined to centrilobular area was evident in Wls-MKO and Con livers (Fig. 7). Thus, blocking Wnt secretion from macrophages does not have any effect on β-catenin at adherens junctions or in metabolic zonation.

Figure 7.

No defect in pericentral zonation in Wls-MKO. IHC shows normal pericentral localization of GS, Cyp1a2, and Cyp2e1 in Con and Wls-Mac KO livers.

Suboptimal LR in Wls-MKO Mice

To address the role of macrophage-derived Wnts in LR, Wls-MKO and Con were subjected to PH. At 40 hours after PH, a notable decrease in PCNA-positive hepatocytes was observed in Wls-MKO; however, at 72 hours no difference was seen (Fig. 8A). Upon quantification, this deficit at 40 hours was found to be around 33% and significant (Fig. 8B). In addition, numbers of mitotic figures as assessed by analysis of H&E images from Wls-MKO and Con showed significantly lower mitosis in hepatocytes in the former at 40 hours (Fig. 8C; Supporting Fig. 1B). WB shows a concomitant decrease in Cyclin-D1 at 40 hours in Wls-Mac KO (Fig. 8D). Coprecipitation studies showed a notable decrease in β-catenin-TCF4 complex in Wls-MKO livers at 4 hours after PH (Fig. 8E). This indicates that β-catenin activation in hepatocytes after PH is at least in part regulated by Wnts derived from Kupffer cells.

Figure 8.

Dampened LR in Wls-MKO after PH. IHC (200×) showing many PCNA-positive hepatocytes in Con livers at 40 hours and 72 hours; however, several PCNA-negative hepatocytes were observed in Wls-MKO at 40 hours. Quantification of PCNA staining showing a 33% and significant decease in the number of hepatocytes in S-phase in Wls-MKO at 40 hours (**P < 0.01). Quantification of mitotic figures shows significantly lower numbers in Lrp-LKO as compared at Con at 72 hours after PH (*P < 0.05). Representative WB shows lower Cyclin-D1 levels in Wls-MKO at 40 hours after PH. A representative immunoprecipitation study shows decreased TCF4-β-catenin complex formation at 4 hours after PH in Wls-Mac KO when compared to Con. Immunoprecipitation studies were performed by pull down of either β-catenin or TCF4. Respective input controls are included in analysis as well.


β-Catenin, the critical downstream effector of Wnt signaling, acts as a coactivator for the TCF family of transcription factors to regulate expression of several target genes in a tissue-specific manner. Deletion of β-catenin in hepatocytes has yielded two major phenotypes in the liver.[4-6] The first observation at baseline is a defect in metabolic zonation as β-Catenin-LKO mice lack several downstream targets in pericentral hepatocytes such as genes encoding for GS (ammonia metabolism), Cyp2e1, Cyp1a2 (xenobiotic metabolism), and others involved in glycolytic pathway and the tricarboxylic acid (TCA) cycle.[30] The second phenotype in β-catenin-LKO mice was a defect in LR after PH. The role of β-catenin signaling in LR is now well established in rodents, zebrafish, and patients.[2, 31] Activation of β-catenin is relatively early, with its nuclear translocation evident in minutes to hours after PH, where it complexes with TCF4 to regulate the expression of Ccnd1. Cyclin-D1 is critical for driving G1 to S phase cell cycle progression and marks the point of no return for cell proliferation.[32] In fact, the absence of β-catenin in hepatocytes in β-Catenin-LKO led to a significant lag in the initiation of LR, with hepatocyte S-phase peaking at 72 hours instead of 40 hours due to decreased Ccnd1.4,6

Endogenous or exogenous activation of β-catenin has been shown to improve LR in animal models as well as in patients.[24, 27, 33] However, for translational exploitation of β-catenin signaling in regenerative medicine, it will be critical to understand the cellular and molecular circuitry upstream of β-catenin activation in hepatocytes at baseline and during LR. This is of special relevance since β-catenin activation has been also shown to occur downstream of non-Wnt mechanisms such as those mediated by HGF, EGF, Flt3, and PKA, which have been independently shown to be playing a significant role in LR after PH.[34]

In the current study using genetic mouse models we address the key regulators of β-catenin in hepatic zonation and LR. We used genetic KO of Lrp5 and Lrp6 to disrupt Wnt signaling in hepatocytes and cholangiocytes using Albumin-Cre, which was also used by us to delete β-catenin in the same two cell types previously.[6, 35] Deleting both Lrp5 and Lrp6 has been shown to prevent any compensation and hence assures complete abrogation of Wnt signaling.[18] This breach in Wnt signaling in hepatocytes prevented pericentral β-catenin activation and led to loss of GS, Cyp1a2, and Cyp2e1, which phenocopied β-catenin-LKO. This finding demonstrates that the pericentral zonation function of β-catenin is fully regulated by Wnt signaling. A more surprising result came from the analysis of LR in this model. Despite β-catenin activation by many non-Wnt dependent mechanisms and despite the physiological activation of such pathways as HGF, EGF, and PKA after PH, there was a considerable deficit in LR in the Lrp-LKO similar to β-catenin-LKO mice, as seen by decreased PCNA at 40 and 72 hours and reduced mitosis at 72 hours.[4, 36] Intriguingly, while β-catenin-LKO showed a comparable number of hepatocytes in S-phase to Con at 72 hours, the Lrp-LKO continued to show fewer hepatocytes in S-phase at this time as compared to their littermate controls. However, the numbers of hepatocytes in S-phase were notably greater at 72 hours than the 40-hour timepoint, suggesting redundant mechanisms that compensate, which was what was observed in β-catenin-LKO as well. An increase in Cyclin-D1 at 72 hours in regenerating Lrp-LKO livers over baseline and 40 hours suggests sufficient LR to restore the smaller baseline hepatic mass in Lrp-LKO. No alternate mechanism of β-catenin activation in regenerating Lrp-LKO livers was observed at any time. Thus, Wnt signaling strictly regulates β-catenin during the process of LR.

To determine the cellular source of Wnts at baseline and after PH is challenging, as several hepatic cells express multiple Wnts.[37] An exciting development to address the role of global Wnt signaling has been to interfere with either their biological activity or prevent their secretion from a cell. Deletion of the gene encoding for porcupine, which is responsible for glycosylation and acylation of Wnt proteins, has been used to disrupt Wnt function.[38] Another relevant protein, Wntless, has been shown to be indispensable for Wnt secretion.[14] In fact, conditional deletion of Wls using Wnt1-cre phenocopies Wnt1-null abnormalities in brain.[15, 39] Using Wls-floxed mice, we blocked Wnt secretion from hepatocytes and cholangiocytes, Kupffer cells, and endothelial cells. Our results show no alteration in either pericentral gene expression, or in the initiation of LR when Wnt-secretion was ablated from hepatocytes and cholangiocytes. Tie2-Cre mediated Wls deletion led to embryonic lethality, which was not surprising because of the many important roles of Wnt/β-catenin signaling in vasculogenesis and angiogenesis.[40, 41] While embryonic lethality precluded us from addressing the role of hepatic sinusoidal endothelial cells and of endothelial cells lining the hepatic vessel walls, viable heterozygous Wls-EKO mutants did not show any notable decrease in Wls. Simultaneously, no defects in hepatic zonation or LR were observed. Current studies are ongoing to generate inducible-conditional KO to address the endothelial-hepatocyte paracrine interactions in the liver. It should be noted that recent studies have divulged a role of platelet-sinusoidal endothelial cell-hepatocyte circuitry to activate β-catenin by Wnt2b during LR.[42]

A more consequential observation came from abrogating Wnt secretion from Kupffer cells. While no defect in metabolic zonation was observed, a 33% decrease in S-phase hepatocytes and hepatocyte mitosis was evident in Wls-MKO at 40 hours after PH. This coincided with a notable decrease in β-catenin-TCF4 complex and Cyclin-D1 expression at 40 hours. These observations suggest that Kupffer cells are an important source of Wnt proteins that activate hepatocyte β-catenin in a paracrine manner. Indeed, the role of macrophages in initiation of LR is unquestionable.[43]

Thus, our current study demonstrates the regulation of β-catenin in two fundamental processes inherent to the liver: hepatic zonation and regeneration. In both cases, β-catenin is under the control of Wnt signaling only. Our studies rule out the role of hepatocytes, cholangiocytes, and Kupffer cells as the source of Wnt proteins that regulate basal pericentral β-catenin activation in liver. On the other hand, it became apparent that Wnt secretion from the nonparenchymal cell compartment, especially Kupffer cells and endothelial cells,[42] is required for timely β-catenin activation in hepatocytes to initiate LR after PH. Clearly, the significance of nonparenchymal cell contribution to LR is well accepted.[44] We hope to determine the role of hepatic stellate cells and further explore the role of sinusoidal and perivenous endothelial cells in β-catenin activation in pericentral hepatocytes in regulating hepatic zonation and during LR after PH.


We thank Ammar Saladhar and Angie Lake for technical assistance.