Potential conflict of interest: Nothing to report.
This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology in Japan; the Ministry of Health, Labor and Welfare in Japan; the Japan Society for the Promotion of Science, the Japan Health Sciences Foundation; the National Institute of Biomedical Innovation; and the Foundation for Advancement of International Science.
The molecular mechanisms regulating differentiation of fetal hepatic stem/progenitor cells, called hepatoblasts, which play pivotal roles in liver development, remain obscure. Wnt signaling pathways regulate the development and differentiation of stem cells in various organs. Although a β-catenin–independent noncanonical Wnt pathway is essential for cell adhesion and polarity, the physiological functions of noncanonical Wnt pathways in liver development are unknown. Here we describe a functional role for Wnt5a, a noncanonical Wnt ligand, in the differentiation of mouse hepatoblasts. Wnt5a was expressed in mesenchymal cells and other cells of wild-type (WT) midgestational fetal liver. We analyzed fetal liver phenotypes in Wnt5a-deficient mice using a combination of histological and molecular techniques. Expression levels of Sox9 and the number of hepatocyte nuclear factor (HNF)1β+HNF4α− biliary precursor cells were significantly higher in Wnt5a-deficient liver relative to WT liver. In Wnt5a-deficient fetal liver, in vivo formation of primitive bile ductal structures was significantly enhanced relative to WT littermates. We also investigated the function of Wnt5a protein and downstream signaling molecules using a three-dimensional culture system that included primary hepatoblasts or a hepatic progenitor cell line. In vitro differentiation assays showed that Wnt5a retarded the formation of bile duct–like structures in hepatoblasts, leading instead to hepatic maturation of such cells. Whereas Wnt5a signaling increased steady-state levels of phosphorylated calcium/calmodulin-dependent protein kinase II (CaMKII) in fetal liver, inhibition of CaMKII activity resulted in the formation of significantly more and larger-sized bile duct–like structures in vitro compared with those in vehicle-supplemented controls. Conclusion: Wnt5a-mediated signaling in fetal hepatic stem/progenitor cells suppresses biliary differentiation. These findings also suggest that activation of CaMKII by Wnt5a signaling suppresses biliary differentiation. (HEPATOLOGY 2013;)
Hepatic stem cells are multipotent stem cells located within ductal plates in fetal and neonatal livers, and canals of Hering in pediatric and adult livers.1 The extrahepatic stem cell niches are peribiliary glands within the bile ducts in humans.2 Hepatic stem/progenitor cells, called hepatoblasts in the fetal liver, proliferate actively and give rise to hepatocytes and cholangiocytes.3, 4 Lineage commitment of such cells can be traced by several cell surface markers, including NCAM, ICAM-1, and EpCAM in humans.1, 5 While our group6 and others7 demonstrated roles for transcription factors regulating the biliary differentiation of hepatic stem/progenitor cells, the molecular mechanisms behind these events have yet to be fully elucidated.
The Wnt family secreted ligands and the corresponding Frizzled family cell surface receptors play a crucial role in the differentiation, proliferation, and self-renewal of stem cells in various organs.8 Wnt signaling pathways involve interactions between a complex set of molecular cognates that includes 19 different Wnt ligands and 10 Frizzled (Fzd) receptors in humans and mice (reviewed at http://www.stanford. edu/group/nusselab/cgi-bin/wnt/). Upon binding to Fzd receptors on the surface of a target cell, Wnt proteins activate one of two classes of downstream pathways distinguishable by their dependency on β-catenin. Examples of canonical β-catenin–dependent pathways include β-catenin–dependent activation of T cell factor by either Wnt1 or Wnt3.8 In contrast, Wnt4 and Wnt5a activate noncanonical β-catenin–independent pathways that include downstream molecules such as calcium/calmodulin-dependent protein kinase II (CaMKII), Rho-kinase, Rac1, calcineurin, and protein kinase C (PKC).9
In liver development, β-catenin is known to regulate the maturation, expansion, and survival of hepatoblasts, and its deletion results in increased apoptosis of hepatoblasts in midgestational fetal livers.10 While the function of noncanonical Wnt signaling in liver development is currently unknown, β-catenin–independent Wnt pathways have been shown to function predominantly as regulators of cell polarity and mobility in other organs.9 In systemic Wnt5a-deficient (knockout [KO]) mice, the size of caudal structures, lung morphogenesis, and intestinal elongation are also abnormal.11–13
Recent reports demonstrate that Wnt5a regulates hematopoietic, mesenchymal, and neural stem cell functions.14-16 Wnt5a has been shown to increase the repopulation of short- and long-term hematopoietic stem cells by maintaining these cells in a quiescent G0 state.14 Wnt5a maintains mesenchymal stem cells and promotes osteoblastogenesis in preference to adipogenesis in bone marrow,15 and also improves the differentiation and functional integration of stem cell–derived dopamine neurons.16 In healthy adult mouse liver, Wnt5a is expressed in mature hepatocytes and cholangiocytes.17 Nonetheless, the physiological functions of Wnt5a and the signaling cascades that it initiates during liver development and in hepatic stem/progenitor cells are unknown.
In this study, we investigated the function of Wnt5a and its downstream targets in the development of murine fetal hepatic stem/progenitor cells. Analysis of Wnt5a KO mice demonstrated that loss of Wnt5a abnormally promotes the formation of bile ductal structures in fetal liver in vivo. Wnt5a supplementation not only retarded the formation of bile duct–like structures, but also promoted hepatic maturation of hepatic stem/progenitor cells in vitro. CaMKII activity, which showed Wnt5a dependence in fetal liver, suppressed the formation of bile duct–like structures. These data indicate that Wnt5a-mediated CaMKII signaling plays an essential role in the differentiation of murine fetal hepatic stem/progenitor cells.
Systemic Wnt5a KO mice in C57BL/6 background were originally generated by Yamaguchi et al.11 Wnt5a KO mice and wild-type (WT) littermates were produced by crossbreeding Wnt5a heterozygous mice. All animals were treated based on the guidelines of the Institute of Medical Science, University of Tokyo, and those of Tokyo Medical and Dental University.
In Vitro Bile Duct–Like Differentiation Assay of Primary Hepatoblasts.
Bile duct–like differentiation assays were performed as described6 with some modifications. Fetal hepatic cells of embryonic day (E) 14.5 liver were dissociated with collagenase,4 and Dlk+ cells were isolated from the resulting population using a magnetic cell sorter (Miltenyi Biotec, Bergisch Gladbach, Germany) and then cultured in collagen gel (Nitta Gelatin, Osaka, Japan). After 30 minutes of incubation at 37°C on basal layer collagen, 1 or 2 × 104 cells were suspended in 1 mL Dulbecco's modified Eagle's medium (DMEM)/F12 mixed with 1 mL collagen gel solution and plated onto basal layer collagen in six-well culture dishes. Plated cells were cultured for 7 days with an additional 2 mL DMEM supplemented with 10% fetal calf serum (FCS) (Sigma, St. Louis, MO), 1× insulin/transferrin/selenium, 20 ng/mL epidermal growth factor (EGF, PeproTech, Rocky Hill, NJ), 20 ng/mL hepatocyte growth factor (HGF, PeproTech), and 25 ng/mL tumor necrosis factor α (PeproTech).
In Vitro Bile Duct–Like Differentiation Assay of Hepatic Progenitor Cell Line.
The HPPL liver progenitor cell line has been reported to exhibit characteristics of differentiated cholangiocytes in three-dimensional culture.18, 19 As in the previous report, we maintained HPPL cells in DMEM/F12 containing 10% FCS, 1× insulin/transferrin/selenium, 10 mM nicotinamide, 10−7 M Dex, and 5 ng/mL HGF and EGF and suspended cells in a mixture of type I collagen and Engelbreth-Holm-Swarm (EHS) sarcoma gel (Becton Dickinson, Bedford, MA) at a density of 4 × 104 cells/mL. Cell suspension was added to each cell culture insert (Millipore, Billerica, MA) and after incubation at 37°C for 2 hours, 500 μL of DMEM/F12 with growth factors was added above and below the insert, and the cells were cultured for 7 days. To test the effects of inhibitors of CaMKII, Rho-kinase, Rac1, calcineurin, and PKC on HPPL differentiation, KN93, KN92, KN62, Y-27632, NSC23766, cyclosporin A, and Go6976 (Supporting Information) were added individually to the culture medium when each three-dimensional culture was initiated. Independent analyses were performed in triplicate, and five fields were randomly selected for counting the cysts that indicate bile duct–like differentiation of cells.
In Vitro Hepatic Maturation Assay of Primary Hepatoblasts.
To induce hepatic differentiation, primary hepatoblasts from WT E14.5 mice were cultured as described.6 Briefly, 2.5 × 105 magnetic cell sorter–isolated Dlk+ cells were cultured in DMEM supplemented with 10% FCS, 2 mM L-glutamine, 1× nonessential amino acid, 100 U/mL penicillin, 100 μg/mL streptomycin, and 10−7 M Dex in each well of a six-well gelatin-coated dish. After 5 days, the resulting cells were supplemented with medium containing 20% EHS gel for an additional 2 days prior to analysis.
Details regarding materials, cell isolation, hematoxylin and eosin staining, reverse-transcriptase polymerase chain reaction (RT-PCR) analysis, immunostaining, immunoblot analysis, Wnt5a-blocking experiments, microarray analysis, and statistical analysis are described in the Supporting Information.
Expressions of Wnt5a and Frizzled Receptors During Liver Development.
We first analyzed Wnt5a expression during liver development using quantitative RT-PCR. Wnt5a expression was detected in fetal and neonatal livers of WT mice and showed a gradual increase during liver development (Fig. 1A). To investigate Wnt5a expression in midgestational fetal liver, we purified the fractions of hepatoblasts, mesenchymal cells, mesothelial cells, endothelial cells, and hematopoietic cells from E14.5 liver using FACS. Quantitative RT-PCR analysis indicated that Wnt5a was expressed in hepatoblasts, mesenchymal cells, mesothelial cells, endothelial cells, and hematopoietic cells. The expression level of Wnt5a was significantly higher in mesenchymal cells than in hepatoblasts and other types of cells in midgestational fetal liver (Fig. 1B). Frizzled is a family of cell surface receptors for Wnt ligands. Adult hepatocytes from 12-week-old mice served as the control. RT-PCR analysis of E14.5 hepatoblasts resulted in the detection of 9 of 10 Fzd receptors (all except Fzd9), whereas E14.5 hematopoietic cells expressed 9 of 10 Fzd receptors (all except Fzd2) (Fig. 1C and Supporting Fig. 1).
Loss of Wnt5a Promotes the Formation of Bile Duct in Fetal Liver.
Because one of the reported phenotypes of systemic Wnt5a KO mice was postpartum death,11 we investigated the function of Wnt5a in liver development using mid- to late gestational fetuses. We determined that although average liver weight in Wnt5a KO E18.5 fetal mice was significantly lower than in WT littermates, the average liver/body weight ratio in KO mice was not significantly different from the ratio in WT mice (Supporting Fig. 2).
Histological analysis of E18.5 livers showed that the number of luminal spaces around the portal vein, which we interpret to be primitive bile ducts, was greater in Wnt5a KO mice than in WT mice (Fig. 2A). To further investigate these changes in bile duct development, expression of Sox9 (a representative transcriptional factor expressed in biliary precursor cells)20 was analyzed. Expression levels of Sox9 were significantly higher in Wnt5a KO E16.5 fetal livers relative to WT livers (Fig. 2B). The Notch pathway plays an essential role in the morphogenesis of bile duct structures.21 Expression levels of Notch1, Notch2, and Jagged1 were significantly higher in Wnt5a KO E16.5 fetal livers relative to WT livers (Supporting Fig. 3A). Numbers of Hes1+ cells in E18.5 livers were significantly greater in Wnt5a KO mice than in WT mice (Supporting Fig. 3B). Expression levels of Cyclin D1 and c-Myc (target transcripts of canonical β-catenin–dependent Wnt pathway) in Wnt5a KO livers were equal to those in WT livers (Supporting Fig. 4A). We tried to assess the protein level of Sox9; however, immunostaining analysis of Sox9 did not work well, probably due to technical problems (data not shown).
During normal liver development, hepatoblasts located around the portal vein develop as hepatocyte nuclear factor (HNF) 1β+HNF4α− biliary precursor cells.22 In normal E16.5 fetal livers, monolayer rings of biliary precursor cells, termed ductal plates, can be detected.23 WT E18.5 fetal livers contained primitive ductal structures (PDSs) consisting of multiple HNF1β+ cytokeratin (CK)19+-cell lumina (Fig. 2C).
Immunohistological analysis revealed that numbers of HNF1β+HNF4α− biliary precursor cells in E16.5 livers (Fig. 2D) and in PDSs formed by these cells in E18.5 livers (Fig. 2E) were significantly higher in Wnt5a KO mice relative to WT mice. Double staining of CK19 and entactin (a component of basement membrane) confirmed that the number of PDSs formed by CK19+ cells was also significantly higher in E18.5 Wnt5a KO liver relative to WT liver (Fig. 2F). These results demonstrate clearly that loss of Wnt5a excessively promotes the formation of bile ducts in fetal liver.
Expression Analysis of Fetal Livers in Wnt5a KO Mice.
Expression of genes coincident with hepatic maturation was also analyzed in Wnt5a KO fetal livers using quantitative RT-PCR. In E16.5 fetal livers, albumin (ALB) and HNF4α messenger RNA (mRNA) levels were nearly equal between WT and Wnt5a KO mice. Similarly, we observed no significant differences between WT and Wnt5a KO E18.5 fetal livers with regard to copy numbers of tyrosine aminotransferase, carbamoyl phosphate synthetase 1 (CPS1), glucose 6-phosphatase (G6Pase), or HNF4α mRNAs (Supporting Figs. 5A and B). These data suggest that the maturation of hepatoblasts to hepatocytes is not impaired in Wnt5a KO mice.
Proliferation of fetal liver cells in Wnt5a KO mice was analyzed by immunoblot and immunostaining. Immunoblot analysis revealed that proliferating cell nuclear antigen (PCNA) production in Wnt5a KO livers was almost equal to that in WT livers (Supporting Fig. 4B). Numbers of CK19+PCNA+ cells in E18.5 were almost equal to those in WT livers (Supporting Fig. 4C). Changes in gene expression in Wnt5a KO livers were analyzed using complementary DNA microarray analysis (Supporting Fig. 5C and Supporting Table 5). Cluster analysis revealed that several molecules associated with amino acid metabolism and cell migration were up-regulated or down-regulated in Wnt5a KO fetal livers compared with those in WT livers.
Wnt5a Retards Formation of Bile Duct–Like Structures from Primary Hepatoblasts.
In collagen gel-embedding culture, mouse primary hepatoblasts differentiate into bile duct–like branching structures, coincident with the expression of biliary cell-specific genes such as CK19 (Fig. 3A, left panel).6 To investigate the effects of Wnt5a on differentiation of hepatoblasts into biliary cells in vitro, we cultured primary hepatoblasts derived from E14.5 WT fetal livers and assessed the formation of bile duct–like branching structures.
We observed that cells in cultures derived from E14.5 WT fetal liver formed approximately 10 colonies (consisting of >100 cells in large branching structures) per 1 × 104 cells (Fig. 3A, right panel); colonies with medium (50-99 cells) or small (10-49 cells) branching structures also were noted. In cultures supplemented with Wnt5a, there were significant decreases in the average number colonies with large- and medium-sized branching structures relative to vehicle-only controls.
Wnt5a Suppresses Cyst Formation Derived from HPPL in Three-Dimensional Culture.
To assess the potential of hepatic stem/progenitor cells for bile duct–like luminal formation, we used an HPPL three-dimensional culture system.19 HPPL is established from mouse E14 Dlk+ hepatoblasts and differentiates into hepatic and cholangiocytic lineages.18 In this system, HPPL cells form cysts that exhibit characteristics of differentiated cholangiocytes producing CK19, E-cadherin, and other characteristic markers. We categorized HPPL-derived colonies into one of three classes: colonies without clear lumina, small cysts (50-100 μm diameter with clear lumina), and large cysts (>100 μm diameter with clear lumina). As described,19 immuno-cytostaining of cultured cells showed that colonies without clear lumina produced both the hepatic marker ALB and the biliary marker CK19, suggesting incomplete terminal differentiation. Cells in the luminal walls of small and large cysts, in contrast, produced CK19 but not ALB, indicating their differentiation to a cholangiocyte lineage (Supporting Fig. 6). Vehicle-only controls or cultures treated with Wnt3a did not show a significant difference in overall number of cysts. In contrast, cultures supplemented with Wnt5a displayed significantly fewer cysts, due both to an absence of large cysts and a significantly reduced number of small cysts (Fig. 3B). Wnt5a is expressed in HPPL cells (Supporting Fig. 7A). We verified the specificity of effect of Wnt5a by blocking experiments. Cultures supplemented with anti-Wnt5a antibody (Ab) resulted in a significant increase in numbers of HPPL-derived cysts relative to control Ab, and blocked the effects of Wnt5a supplementation (Fig. 3C). Numbers of HPPL-derived cysts were higher in cultures supplemented with Wnt5a-specific inhibitor relative to vehicle-only controls (Supporting Fig. 7B).
Immunoblot analysis indicated that CK19 production in HPPL-derived colonies were significantly down-regulated in cultured cells supplemented with Wnt5a relative to vehicle-supplemented controls, whereas the levels of ALB, α-fetoprotein (AFP), and PCNA did not change (Fig. 3D). Expression analysis of HPPL-derived colonies revealed that HNF1β, Notch1, and multidrug resistance–associated protein 3 (MRP3, a key primary active transporter in biliary cells) were significantly down-regulated in cultured cells supplemented with Wnt5a relative to vehicle-supplemented controls (Fig. 3E). HNF1β and Sox9 were significantly up-regulated in cultured cells supplemented with anti-Wnt5a Ab relative to control Ab (Supporting Fig. 7C), whereas the levels of hepatocytic markers did not change (Supporting Fig. 7D). Consistent with our in vivo results, these data indicate that Wnt5a suppresses bile duct–like cyst formation of fetal hepatic progenitor cells in vitro.
Wnt5a Induces the Expression of Hepatic Maturation Markers in Primary Hepatoblasts In Vitro.
We evaluated the potential of primary hepatoblasts for hepatic maturation using an in vitro hepatic differentiation assay.24 Phase-contrast microscopy after addition of EHS gel identified several morphological changes within cells, including formation of highly condensed cytosol, and clear, round nuclei typical to mature hepatocytes (Fig. 4A, middle panel). Because similar gross morphological changes were also seen in cells cultured in the presence of Wnt5a (right panel), we used quantitative RT-PCR to measure the effect of Wnt5a on expression of hepatic maturation marker genes in stem/progenitor cells. Expression of tyrosine aminotransferase and CPS1 in cultured cells increased significantly with supplemental Wnt5a (Fig. 4B), whereas changes in tryptophan-2,3-dioxygenase, G6Pase, and HNF4α mRNA levels were not significantly different. These results indicate that Wnt5a contributes, in part, to primary hepatoblast maturation. Taken together, our in vitro data demonstrate that Wnt5a retards biliary differentiation and promotes hepatic differentiation of hepatoblasts.
Inhibition of CaMKII Activity Promotes the Formation of Bile Duct–Like Cysts Derived from HPPL.
While Wnt5a is known to stimulate several signaling cascades, including CaMKII, Rho-kinase, Rac1, calcineurin, and PKC, the specific cascade triggered by Wnt5a in hepatic stem/progenitor cells is unknown. To address this question, we analyzed the effects of specific inhibitors of these candidate molecules in HPPL-derived cysts, where Wnt5a is expressed (Supporting Fig. 7A). Relative to controls, inhibitors specific to CaMKII (KN93 and KN62) resulted in a significant increase in numbers of both small and large bile duct–like cysts derived from HPPL (Figs. 5A and B). In contrast, other inhibitors, including Y-27632 (Rho-kinase inhibitor), NSC23766 (Rac1 inhibitor), cyclosporin A (calcineurin inhibitor), and Go6976 (PKC inhibitor), had no effect on the number or size of HPPL-derived cysts (Fig. 5C). We examined the expression of biliary markers in HPPL-derived cysts treated with CaMKII inhibitor (KN62). Expression of MRP3, a key primary active transporter in biliary cells, in HPPL-derived cysts increased significantly with supplemental CaMKII inhibitor (Fig. 5D). There were no significant differences in mRNA levels of ALB, HNF4α, and β-catenin–related molecules between HPPL-derived cysts treated with CaMKII inhibitor and those treated with vehicle (Supporting Fig. 8). The protein level of AFP in HPPL-derived cysts treated with CaMKII inhibitor was lower than that in vehicle-supplemented controls, whereas the levels of CK19 and PCNA did not change (Fig. 5E). These data indicate that CaMKII activity suppresses the formation of HPPL-derived cysts, whereas activities of other Wnt5a-mediated candidates did not influence the efficacy of cyst formation.
Phosphorylation of CaMKII in Primary Hepatoblasts.
To investigate the activation state of CaMKII in fetal and neonatal WT livers, we used immunoblots of liver homogenates derived from E14.5, E16.5, and E18.5 and postnatal day (P) 1, P7, and P14 mice to measure CaMKII phosphorylation levels. Phosphorylation at threonine-286, specifically, has been reported to maintain CaMKII in an active state.25 Phosphorylation of PKC, a kinase that did not affect cyst formation in HPPL cells, was also examined. Whereas we detected both phosphorylated CaMKII (p-CaMKII) and PKC (p-PKC) in each fetal and neonatal liver homogenate, levels of phosphorylated CaMKII increased gradually over time (Supporting Fig. 9A, top panel), similar to the pattern of Wnt5a expression during liver development (Fig. 1A). In contrast, developmental changes in the steady-state levels and phosphorylation of PKC in these samples (Supporting Fig. 9A, lower panels) did not correspond to Wnt5a expression patterns.
Using immunostaining of FACS-purified primary hepatoblasts with anti p-CaMKII Ab, we detected p-CaMKII in >90% of FACS-purified primary hepatoblasts (Supporting Fig. 9B, upper panels); p-PKC was also detected with anti p-PKC antibodies in these cells (Supporting Fig. 9B, lower panels). These data demonstrate that both CaMKII and PKC are in an active state in primary hepatoblasts.
Wnt5a Regulates the Phosphorylation of CaMKII in Fetal Liver.
To verify whether CaMKII activation is controlled by Wnt5a, levels of p-CaMKII in HPPL grown in the absence or presence of Wnt5a were examined. Immunoblot analysis revealed that Wnt5a stimulation increased the level of phosphorylated CaMKII, with p-CaMKII levels peaking 3 hours after Wnt5a supplementation and then decreasing to baseline levels after 12 hours (Fig. 6A and Supporting Fig. 10A). Similar to a previous report,15 total CaMKII protein levels in HPPL also increased after CaMKII activation. Ratios of p-CaMKII/CaMKII also increased, peaking 3 hours after Wnt5a supplementation (Supporting Fig. 10B). In contrast, Wnt5a had no effect on p-PKC and p-Rac1 levels in HPPL (Supporting Figs. 10C and D) nor on nuclear translocation of NFAT (representative downstream molecule of calcineurin; data not shown).
We also tested the combined effect of Wnt5a plus a CaMKII inhibitor (KN62) on cyst formation in HPPL-derived cells. The number and size of cysts in HPPL-derived cells decreased with Wnt5a alone, and increased with CaMKII inhibitor alone. When used in combination (HPPL treated with both CaMKII inhibitor plus Wnt5a), the number and size of cysts was similar to CaMKII inhibitor alone, and significantly higher than cells treated with Wnt5a alone (Figs. 6B and C).
We also used immunoblots to compare p-CaMKII levels in WT and Wnt5a KO fetal liver homogenates. Levels of p-CaMKII were significantly lower in Wnt5a KO relative to WT fetal livers (Fig. 6D); quantification using densitometry revealed that p-CaMKII levels in Wnt5a KO livers were also significantly lower than those in littermate WT livers (Supporting Fig. 10E), indicating that Wnt5a mediates an increase in CaMKII phosphorylation in fetal liver.
This study provides the first evidence of a physiological role for Wnt5a in liver development, in that Wnt5a was observed to suppress the formation of bile ducts derived from hepatoblasts. Our data showed increased expression of Sox9, Notch1, Notch2, and Jagged1 in Wnt5a KO livers (Fig. 2B and Supporting Fig. 3A), as well as abnormally increased formation of primitive ductal structures (Figs. 2E and F). In Wnt5a KO livers, the numbers of HNF1β+HNF4α− biliary precursor cells and primitive ductal structures were increased around the portal vein only (zone 1), whereas such cells were not observed in zone 2 or 3 (Figs. 2D-F). At E14.5, HNF1β+HNF4α− biliary precursor cells were not detected in Wnt5a KO livers similar to WT livers (Supporting Fig. 11A). These results suggested that lineage commitment of hepatoblasts into biliary cells is determined by the microenvironment around the portal vein, depending on the presence or absence of Wnt5a protein. The lungs and intestine of systemic Wnt5a KO mice were abnormal, while tissue structures of the pancreas and kidneys were almost normal (Supporting Fig. 12). Immunostaining analysis showed that p75NTR+ cells were detected in E18.5 Wnt5a KO livers, similar to WT livers (Supporting Fig. 11B). These results implied that development of mesenchymal cells in E18.5 Wnt5a KO livers is not impaired compared with that in littermate WT livers. Wnt5a expression was significantly higher in mesenchymal cells than in hepatoblasts or other types of cells in midgestational WT fetal liver (Fig. 1B). Thus, the microenvironment around the portal vein, which consists of mesenchymal cells, other types of cells, and extracellular matrices, regulates appropriate cell fate decision of hepatoblasts, whereas loss of Wnt5a in such developmental niche leads to abnormally increased formation of primitive ductal structures (Fig. 7). Further investigation of this hypothesis will require conditional deletion of Wnt5a-downstream molecules in hepatoblasts at late gestational fetal stages.
Maturation of hepatoblasts to a hepatocyte lineage is regulated by several factors, including oncostatin M, HGF, and extracellular matrices.24 Our data showed that hepatic maturation of primary hepatic stem/progenitor cells was promoted in cultures supplemented with Wnt5a (Figs. 4A and B). On the other hand, no significant changes in hepatocyte marker expression were detected in Wnt5a KO relative to WT livers. It may be that there is functional redundancy among different Wnt family ligands in vivo, since several noncanonical-signaling Wnt ligands (Wnt4, Wnt5a, and Wnt11) are expressed in normal fetal liver.26 In support of the hypothesis that other noncanonical Wnt ligands may compensate for Wnt5a, Supporting Fig. 13A shows that Wnt4 expression levels in liver increase significantly in Wnt5a KO versus WT littermates. These data strongly support our hypothesis that the effect of Wnt5a on hepatic maturation is compensated by other noncanonical Wnt ligands, such as Wnt4.
CaMKII, a serine/threonine protein kinase present in essentially every tissue, regulates important functions including modulation of ion channel activity, cellular transport, and cell morphology in neural tissues.27 A Wnt5a-CaMKII pathway has been reported to induce osteoblastogenesis by attenuating adipogenesis in mesenchymal bone marrow stem cells.15 Our results show that in liver, inhibition of CaMKII activity promoted bile duct–like cyst formation (Figs. 5A and B), and that phosphorylation of CaMKII is dependent on Wnt5a stimulation (Fig. 6). Although these results provide strong support for our hypothesis that Wnt5a stimulates CaMKII in hepatoblasts, we have not identified which molecules function downstream of CaMKII.
CaMKII has been reported to activate the transforming growth factor β–activated kinase 1 (TAK1)-Nemo-like kinase (NLK) pathway, and that resulting phosphorylation of T cell factor inhibits β-catenin–dependent transcription.28 On the other hand, CaMKII-TAK1-NLK signaling induces bone marrow mesenchymal stem cells to undergo osteoblastogenesis depending on specific downstream signaling cascades.15 Our expression analysis showed that expression levels of Cyclin D1 and c-Myc (the direct target molecules of β-catenin activation) did not change in Wnt5a KO mice in vivo (Supporting Fig. 4) nor in HPPL-derived cysts treated with CaMKII inhibitor in vitro (Supporting Fig. 8), compared with the respective control samples. Preliminary data (not shown) demonstrated that the levels of TAK1 mRNA and protein during development did not correlate with those of Wnt5a and p-CaMKII in whole liver lysates. Moreover, Wnt5a stimulation did not increase the level of activated β-catenin in HPPL (Supporting Figs. 13B and C). These results suggest that the Wnt5a-CaMKII pathway does not activate β-catenin in hepatoblasts. On the other hand, Wnt5a stimulation increased the level of stabilized p53 (phosphorylated at Ser15) in HPPL (Supporting Figs. 13B and D), suggesting that stabilization of p53 is associated with Wnt5a-CaMKII signaling. Further study will be needed to clarify this issue.
Recent studies have shown pathological roles for Wnt5a in various organs; addition of recombinant Wnt5a significantly reduced the migratory capacity of colorectal cancer cell line.29 Whereas increased Wnt5a expression correlates with advanced stages of gastric cancer with poor prognosis,30 there is no definitive data about Wnt5a in the progression of hepatocellular carcinomas. In this study, we reveal one function of Wnt5a in fetal liver in the suppression the biliary differentiation of hepatic stem/progenitor cells. To clarify the pathological role of Wnt5a in liver disease, inducible systemic Wnt5a KO mice or liver-specific CaMKII KO mice would be needed in future studies. Any future evidence demonstrating a role for Wnt5a in adult hepatic stem/progenitor cells and cancer stem cells may lead to studies of Wnt5a signaling as a therapeutic target against abnormal bile ductal formation in the liver or cholangiocellular carcinoma.
We thank Prof. A. Miyajima (the University of Tokyo) for the gift of anti-CK19 antibody. We thank K. Itoh, K. Okada, and A. Yanagida (Institute of Medical Science, the University of Tokyo) for excellent technical assistance.