With roles in processes such as stem cell maintenance and renewal, cell survival, proliferation, embryo patterning, organogenesis, differentiation, cell migration, and polarity, β-catenin regulates numerous events in liver development, homeostasis, metabolism, regeneration, and carcinogenesis (Fig. 1). The β-catenin protein is comprised of unstructured N- and C-terminal tails flanking a core of 12 Armadillo domains. Each Armadillo domain is a 42–amino acid module comprised of three alpha-helical structures (Huber et al.,1996), with the overall structure assuming a right-handed superhelix with a positively charged groove spiraling along the helix (Graham et al.,2000). It is highly conserved and appears throughout the metazoan lineage (Schneider et al.,2003), playing a dual role in cells, as a participant in adherens junctions in linking the transmembrane protein E-cadherin to the actin cytoskeleton (McCrea and Gumbiner,1991; McCrea et al.,1991; Aberle et al.,1994; Rimm et al.,1995), and also as a signaling effector through binding and derepression of TCF family transcription factors (Behrens et al.,1996; Huber et al.,1996). Regulation of β-catenin occurs posttranslationally; it is constitutively expressed in the cytoplasm, but its activity is regulated through phosphorylation events that modulate its affinity for E-cadherin (Piedra et al.,2001) and alpha-catenin (Aberle et al.,1996; Ozawa and Kemler,1998; Piedra et al.,2003) as well as its degradation by the proteasome (Aberle et al.,1997). Signals from secreted WNT proteins transduced through transmembrane Frizzled (FZD) receptors (Bhanot et al.,1996; Wang et al.,1997) are capable of stabilizing β-catenin by inactivation of its degradation complex (Cook et al.,1996; Papkoff et al.,1996; Aberle et al.,1997; Nakamura et al.,1998; Sakanaka et al.,1999; Swiatek et al.,2004). In the absence of these signals, cytoplasmic β-catenin is phosphorylated on threonine 41 and serines 33, 37, and 45 by a protein complex comprised of glycogen synthase kinase 3 beta (GSK3β) (Rubinfeld et al.,1996; Yost et al.,1996), Adenomatous Polyposis Coli (APC) (Munemitsu et al.,1995), AXIN (Ikeda et al.,1998; Nakamura et al.,1998), and casein kinase I (CKI) (Liu et al.,2002; Swiatek et al.,2004) (Fig. 2A). Phosphorylation of these residues allows for recognition and ubiquitination by an E3 ubiquitin ligase complex including the F-box protein beta-transducin repeat containing protein βTrCP (Hart et al.,1999), and ultimate degradation by the 26S proteasome (Aberle et al.,1997). Emerging evidence indicates that activation and nuclear translocation of β-catenin can be effected by a growing number of factors other than WNT pathway proteins, including several tyrosine kinases (Roura et al.,1999; Piedra et al.,2001; Monga et al.,2002; Sekhon et al.,2004; Lilien and Balsamo,2005; Apte et al.,2006; Zeng et al.,2006; Berg et al.,2007; Rhee et al.,2007) (Fig. 2B).
β-catenin in Development
The Wnt/β-catenin pathway is critical throughout embryonic development beginning prior to implantation. In mouse and other amniotes, the fertilized zygote undergoes a series of cell divisions followed by cavitation to form a hollow ball called a blastocyst. The blastocyst implants in the maternal uterine endometrium, then flattens and elongates into a cylinder comprised of an outer cup-shaped epithelial layer called the epiblast, as well as the visceral endoderm and extra-embryonic endoderm. During gastrulation in mammals and birds, a population of epiblast cells migrates to the midline of the blastula and ingresses to form the primitive streak, ultimately giving rise to the mesoderm and definitive endoderm. Patterning of the primitive streak by WNT and TGFβ signaling results in an anterior region with potential to form both the anterior mesoderm from which cardiac tissue arises as well as the definitive endoderm from which the hepatic endoderm is generated.
Active β-catenin is detectable immediately prior to gastrulation at E5.5 in the extraembryonic visceral endoderm and in a narrow region of cells in the epiblast at E6 that go on to form the primitive streak. Thereafter, beta-catenin signaling occurs both in the primitive steak and the node (Mohamed et al.,2004). Attempts to identify the Wnts responsible for activating beta-catenin have revealed expression of Wnt1 and Sfrp1 localized to the inner cell mass of the blastocyst, Wnt3a, -6, -7b, and -10b expressed throughout the blastocyst, and Wnt9a expression in the cells around the blastocoels cavity. Shortly before gastrulation, Wnt2b expression occurs in the prestreak embryo on the posterior end of the presumptive primitive streak, and its expression expands as the streak forms. In addition, WNT3 plays a critical role in driving gastrulation, as is evident by the inability of Wnt3 mutants to form a primitive streak, maintain or produce mesoderm (Liu et al.,1999; Barrow et al.,2007). Also at this time, WNT signaling antagonists Sfrp1, Sfrp5, and Dkk1 are expressed in partially overlapping regions in the anterior visceral endoderm through midstreak stage (Kemp et al.,2005). The expression of WNTs in the posterior visceral endoderm and WNT antagonists in the anterior visceral endoderm establishes an anterior-posterior gradient of WNT activity (Fig. 3A). The WNT/β-catenin activity also promotes expression of the TGFβ family member Nodal, whose expression radiates outward from the posterior pre-gastrula epiblast, and is antagonized by proteases and Cerberus and Lefty in the anterior visceral endoderm (AVE) to create its own gradient (Perea-Gomez et al.,2002). Through regulation of the Wnt, FGF, and BMP pathways, Nodal signaling promotes the initiation gastrulation in the posterior epiblast (Haramoto et al.,2004; Onuma et al.,2005). Together these gradients promote the anterograde chemotaxis of visceral endoderm cells and thus play roles in formation of the primitive streak and establishment of the anterior-posterior axis (Conlon et al.,1994; Brennan et al.,2001; Perea-Gomez et al.,2002; Lu and Robertson,2004). After gastrulation, Nodal promotes endoderm- and mesoderm-specific gene expression programs in a dose-dependent manner, with higher levels specifying endoderm and lower levels specifying mesoderm (Vincent et al.,2003). The definitive endoderm arises from the source of Wnt-mediated Nodal expression at the anterior end of the primitive streak, and here Nodal activity promotes expression of endodermal transcription factors including GATA4-6, Eomesodermin, Mix-like proteins, and Foxa1-3, as well as Sox17 (Ang et al.,1993; Arceci et al.,1993; Laverriere et al.,1994; Suzuki et al.,1996; Gao et al.,1998; Germain et al.,2000; Aoki et al.,2002; Brennan et al.,2002; Ben-Haim et al.,2006; Hagos et al.,2007; Howard et al.,2007; Arnold et al.,2008).
Sox17, an HMG box transcription factor critical to endoderm development, partners with β-catenin to transcribe targets including Hnf1β, Foxa1, and Foxa2 in posterior endoderm. Inhibition of Sox17 activity by expression of antisense oligos (Clements et al.,2003) or dominant-negative Sox17 engrailed protein (Hudson et al.,1997) interferes with endoderm formation, and embryos with a targeted deletion of Sox17 are embryonic lethal (Kanai-Azuma et al.,2002). Furthermore, nuclear beta-catenin is detectable in early endoderm cells, and depletion of beta-catenin results in repression of Sox17 targets (Sinner et al.,2004), revealing the critical role of the interplay of these two factors in proper endoderm formation.
Embryos homozygous for a null allele of the beta-catenin gene Ctnnb1 have been generated, and exhibit disrupted anterior-posterior axis formation manifesting as a mislocation in cerberus-like and Lim1-expressing prospective anterior visceral endoderm tissue, a lack of mesoderm and head formation, as well as an absence of expression of posterior mesoderm markers Brachyury and goosecoid, and expression of anterior markers of Hex, Hesx1, Otx2, and Engrailed1 (Huelsken et al.,2000). In addition to the malformations caused by deletion of Wnt3 (Liu et al.,1999; Barrow et al.,2007), the role of beta-catenin in embryonic patterning is further supported by studies in which β-catenin activity is disrupted through deletion of WNT co-receptors Lrp5 and Lrp6 (Kelly et al.,2004), which leads to disrupted node formation, axis formation, and establishment of endoderm. Furthermore, DKK1-mediated blockade of endogenous Wnt activity inhibits mesodermal differentiation (Lindsley et al.,2006) and constitutively active beta-catenin promotes premature mesodermal differentiation of epiblast cells (Kemler et al.,2004). Recent in vitro models using embryonic stem cells also support this role, as treatment with WNT3A or low levels of activin result in a population of cells resembling posterior primitive streak cells in gene expression pattern and developmental potential, whereas treatment with high levels of activin induces a cell population resembling anterior primitive streak (Gadue et al.,2006).
β-catenin in Endoderm Patterning
Morphogenesis in the post-gastrulation embryo results in a transformation of the endoderm into a tube-like structure surrounded by mesoderm. Within this tube, precise temporal and spatial gradients of WNT signaling (McLin et al.,2007), in conjunction with BMP ligands from the septum transversum mesenchyme (Roberts et al.,1995; Rossi et al.,2001; Tiso et al.,2002; Zhang et al.,2004; Shin et al.,2007), FGF ligands (FGF1 and FGF2) from cardiac mesoderm (Jung et al.,1999; Zhang et al.,2004; Shin et al.,2007), and mesenchyme-derived retinoic acid (Stafford et al.,2004; Martin et al.,2005; Wang et al.,2006; Bayha et al.,2009), and FGF4 (Dessimoz et al.,2006), direct the patterning along the anterior-posterior axis necessary for hepatic competence in the ventral foregut region. FGF4 and WNT secreted by posterior mesoderm promote dose-dependent expression of Pdx1 and CdxB, which specify hindgut and midgut, respectively (Ohlsson et al.,1993; Heller et al.,1998; Wells and Melton,2000; Ehrman and Yutzey,2001; Kumar et al.,2003; Dessimoz et al.,2006), and suppress expression of the hepatic specification factor Hhex (Dessimoz et al.,2006). The most anterior region of the gut tube receives the lowest level of FGF4 and WNT, as well as actively inhibiting WNT signaling by the secretion of secreted Frizzled-related proteins (sFRPs) (Pilcher and Krieg,2002; Finley et al.,2003), thereby inducing hepatic competence in the foregut region by allowing Hhex expression (Dessimoz et al.,2006; McLin et al.,2007; Li et al.,2008). BMP signaling also promotes Hhex expression (Zhang et al.,2002).
Accumulating evidence supporting the role of repression of Wnt/beta-catenin signaling in gut tube patterning, the first step towards hepatic competence of the gut endoderm, is compelling (Fig. 3B). While forced expression of Wnt8 or a stabilized beta-catenin construct in presumptive foregut cells of Xenopus gastrulas leads to an absence of foregut markers and foregut organ buds, antagonism of Wnt signaling by ectopic expression of Dkk1 or overexpression of GSK3β in the posterior lateral endoderm leads to an expansion of expression of foregut markers and enlarged hepatic and pancreatic buds. Inducible expression of the beta-catenin transactivation domain revealed opposing effects on liver development depending on the timing of the induction, and enabled a mapping of the time window during which repression is required. Induction of beta-catenin activity between stage 11 (midgastrula) and stage 20 (6–7 somites) (gut patterning/hepatic competence, corresponding to approximately E7–E8.5 in mouse) leads to absence of foregut markers; induction between stages 25 and 30 (formation of liver diverticulum, approximately equivalent to E9–E9.5 in mouse) leads to normal foregut development; and induction between stages 30 and 42 (liver bud expansion, approximately E9.5–E13.5 in mouse) leads to enlarged liver buds in some embryos, underscoring the precision in both timing and dose of Wnt signal necessary for proper hepatic competence and specification (Zorn and Mason,2001; McLin et al.,2007). During gut patterning/hepatic competence, beta-catenin appears to repress Hhex expression by interaction with Vent2, a transcription factor effector of BMP signaling to prevent expression of liver-specific targets in regions with Wnt signaling active (McLin et al.,2007). The inhibition of Wnt signaling appears to be achieved by expression of sFrp5 in the foregut epithelium, which inhibits the activity of Wnt11 concomitantly expressed in nearby endodermal cells. Depletion of sFrp5 results in both disruption of hepatic specification, as well as loss of adhesion and polarity in cells of the ventral foregut, underscoring its role in modulating both canonical Wnt signaling and non-canonical Wnt/PCP signaling. In addition, ectopic sFrp5 expression in ventral posterior endoderm expands the foregut and represses the hindgut region, indicating that sFrp5 expression may be sufficient for specification of foregut (Li et al.,2008). While these observations were made in Xenopus, studies in zebrafish also show similar findings. High β-catenin activity in APC homozygous and heterozygous null embryos in zebrafish led to failure of anterior endodermal organization observed as decreased numbers of endodermal progenitors that eventually leads to decreased hepatic and pancreatic progenitors (Goessling et al.,2008). This is consistent with the requirement of inhibition of β-catenin activity for proper endodermal organization, which is the first step towards hepatic specification.
Wnt/β-catenin Signaling in Hepatic Specification
Following foregut endoderm specification, activation of Wnt/beta-catenin signaling appears to be critical in shifting the fate of these endodermal progenitors to liver-specific fates, especially in zebrafish and Xenopus. In Xenopus, the Wnt ligand Wnt2b is expressed by the mesoderm flanking the anterior gut endoderm between stage 20 and stage 32, and expression of the secreted Frizzled-related gene sFRP5 can be detected overlapping with Hex-expressing regions between stage 25 and persisting through tailbud stage 37/38 (Pilcher and Krieg,2002), at which time the liver bud is visible. In zebrafish, expression of a novel Wnt isoform, Wnt2bb, is seen by in situ hybridization at the onset of endoderm patterning at 18 h.p.f (corresponding to E7.5 in mouse development) and persists through 52 h.p.f (postnatal day 8/9 in rodents). Mutation of Wnt2bb, or inhibition of its expression with an antisense morpholino oligonucleotide, results in a severe delay in liver development evident by 24 h.p.f., a time at which the liver bud is undergoing expansion (Wallace and Pack,2003). Injection of wild-type cells into the lateral plate mesoderm, but not the endoderm, rescues this phenotype, revealing that the Wnt2bb signals driving this stage of liver development are produced by the lateral plate mesoderm (LPM) (Ober et al.,2006). Additionally, use of a heat-shock-inducible dominant-negative Tcf3 construct revealed that inhibition of β-catenin activity between 16 h.p.f and 21 h.p.f resulted in an absence or substantial decrease in hepatic tissue, while fish heat-shocked at 25 h.p.f. formed livers of variable sizes (Ober et al.,2006). Animals heterozygous for wild-type APC developed a hepatic enlargement correctable by injection with a morpholino antisense oligonucleotide targeting β-catenin or inhibition of β-catenin activity with dominant-negative TCF. APC+/− embryos show defects in endoderm patterning, exhibiting an enlargement of liver buds at the expense of pancreatic buds, consistent with an expansion of the hepatic foregut region into the pancreatic domain. Cell transplantation experiments verified that the role of β-catenin in liver formation is cell autonomous. Temporal modulation of Wnt/β-catenin activity through heat-shock-inducible expression of Wnt8 confirmed that while foregut specification of endoderm requires inhibition of Wnt signaling, this pathway becomes critical for hepatic specification (Goessling et al.,2008).
Interestingly, a mutation in the retinoic acid biosynthesis enzyme retinaldehyde dehydrogenase type2 (RALDH2) results in a strikingly similar delay in zebrafish liver bud formation as the Wnt2bb mutant, which can be partially rescued by treatment with exogenous all-trans-retinoic acid or injection of Raldh2 mRNA. Absence of Wnt2bb expression in LPM was also observed in these animals, suggesting that retinoic acid may be responsible for induction of Wnt2bb expression (Negishi et al.,2010).
Breeding Foxa3-Cre and floxed β-catenin transgenic mice results in successful deletion of beta-catenin evident at E9.5 in the HNF4α-expressing hepatic cells (Tan et al.,2008). While hepatic specification stage of foregut endoderm was not specifically assessed in this model, primitive liver bud observed at E9.5 was comparable in the control and beta-catenin-null livers and differences became apparent only later in development as discussed in the following sections. Thus, the role of beta-catenin in hepatic specification in mice has yet to be definitively addressed.
General Overview of Liver Bud Expansion and Differentiation: The Process of Hepatic Morphogenesis
Once the foregut undergoes hepatic specification, the cells in this region undergo a morphological transition from gut-like columnar cell morphology to pseudostratified epithelia that causes thickening of the epithelium into a diverticulum (Bort et al.,2006). The diverticulum is surrounded by septum transversum mesenchyme (STM), which is lined with a basal lamina and endothelial precursors. At E9.5, migration of the cells, now bipotential hepatoblasts, occurs through matrix metalloproteinase digestion of the laminin-rich basement membrane and delamination into the STM (Margagliotti et al.,2008). After migration of hepatoblasts into the STM, liver bud expansion is mediated, in part, by an activation of WNT signaling (Monga et al.,2003; Hussain et al.,2004; Micsenyi et al.,2004; Suksaweang et al.,2004; Tan et al.,2008). This phase consists of expansion of the liver bud through rapid proliferation of the resident cells.
Between E9.5 and E13.5, hepatic architecture begins to be established, with sinusoids (Enzan et al.,1997) and bile canaliculi appearing and partially maturing (Luzzatto,1981). Between E9.5–E10, the liver bud separates into the cranial lobe, from which the extrahepatic and intrahepatic bile ducts arise, and the caudal lobe, from which the gall bladder develops. The gall bladder and common bile duct join between E10.5 and E11.5, and the gallbladder subsequently elongates. Hepatic cords and sinusoids form around E10, and by E12.5, sinusoids decrease and the parenchymal compartment becomes predominant. Stellate cells also appear at this time, and interlobular spaces appear between liver lobes. Between E11.5 and E12.5, the left umbilical vein becomes the ductus venosus and the right vitelline vein becomes the portal vein (Crawford et al.,2010). Certain organelles, including lysosomes, Golgi, and rough endoplasmic reticulum, also become apparent during this period (Medlock and Haar,1983; Vassy et al.,1988). Hematopoietic cells colonize the expanding liver bud at E12 and recede at approximately E16 (Sasaki and Matsumura,1986; Crawford et al.,2010).
At approximately E13.5 in mouse development, bipotential hepatoblasts begin differentiating into biliary epithelial cells (BECs) or hepatocytes, based on proximity to portal veins. Prior to differentiation, hepatoblasts express markers of both hepatocytes and BECs, such as Albumin, α-fetoprotein, and CK19. BECs differentiate from hepatoblasts around portal veins under the influence of TGFβ and Notch signaling, first producing a monolayer, and then a bilayer of cuboidal, CK19-positive cells (Kodama et al.,2004; Clotman et al.,2005; Loomes et al.,2007; Lozier et al.,2008). At around E17.5, ductal plate remodeling is observed, in which focal dilations emerge at points in the bilayer, become surrounded by portal mesenchyme, and undergo tubulogenesis into intrahepatic bile ducts (Lemaigre,2003).
Hepatoblasts not adjacent to portal veins instead differentiate into hepatocytes and arrange into cords lined by sinudoidal epithelial cells and bile canaliculi. Once hepatoblasts are specified into hepatocytes and undergo further expansion, they begin acquiring the functions of a mature hepatocyte, a process referred to as hepatocyte maturation. This process is gradual, and eventually mature hepatocytes appear around E17 stage as highly polarized epithelial cells with abundant glycogen accumulation. This process of progressive hepatocyte differentiation or maturation appears to be prompted by oncostatin M secreted by hematopoietic cells (Kamiya et al.,1999; Matsui et al.,2002), as well as glucocorticoid hormones (Strick-Marchand and Weiss,2002; Michalopoulos et al.,2003) and HGF (Michalopoulos et al.,2003) (Fig. 3D). Oncostatin M, possibly regulated by TNFα (Kamiya and Gonzalez,2004), appears to elicit metabolic maturation of hepatocytes by activation of JAK/STAT signaling through its receptor gp130 (Kamiya et al.,1999) and promotes adhesion and polarity through upregulation of the tight junction protein claudin-2 (Imamura et al.,2007) and the adherens junction protein E-cadherin (Battle et al.,2006). Glucocorticoids may act to suppress growth and induce expression of critical hepatocyte transcription factors HNF4α (Li et al.,2000) and C/EBPα (Tomizawa et al.,1998; Michalopoulos et al.,2003; Yamasaki et al.,2006). Many other factors have demonstrated roles in liver cell differentiation, including transcription factors HNF1α, HNF1β, HNF6/the Onecut (OC) factors, HNF4α, c/ebpα, Jagged/Notch, TGFβ, TNFα, Foxa1–3, Hhex, GATA4/6, nuclear hormone receptors, and extracellular matrix interactions (Rastegar et al.,2000; Clotman et al.,2002,2005; Coffinier et al.,2002; Jacquemin et al.,2003; Parviz et al.,2003; Friedman et al.,2004; Kamiya and Gonzalez,2004; Kodama et al.,2004; Odom et al.,2004; Zhao et al.,2005; Cheng et al.,2006; Kyrmizi et al.,2006; Yamasaki et al.,2006; Hunter et al.,2007; Margagliotti et al.,2007; Watt et al.,2007; Gkretsi et al.,2008). Recent work has also revealed a role for microRNA expression in liver development, with miR-30a family microRNAs critical for biliary morphogenesis (Hand et al.,2009), and miR-495 and miR-218 responsible for regulating expression of HNF-6 and OC-2 (Simion et al.,2010).
Role of β-Catenin Activity in Hepatic Morphogenesis
The role of beta-catenin during the process of hepatic morphogenesis is quite pleiotropic and highly temporal. β-Catenin is initially required for expansion of hepatoblasts during early stages of hepatic morphogenesis and is later important for proper specification of hepatoblasts to BECs as well as hepatocyte maturation (Tan et al.,2008).
Compelling evidence exists for the role of beta-catenin during liver bud expansion in mice. In developing livers between days E10.5 to E18.5, β-catenin protein levels peak at E10–E12 and then gradually decline. Nuclear/cytoplasmic β-catenin staining also peaks at E10–E12 and decreases gradually, and correlates with percentage of PCNA-positive cells, consistent with a role for β-catenin in driving hepatoblast proliferation (Micsenyi et al.,2004). Also, infection of chick liver primordium at E3 with a viral construct expressing an N- and C-terminally truncated, constitutively active β-catenin leads to hepatomegaly at E15, as well as to a high nuclear-to-cytoplasmic ratio and hepatoblast-like morphology in targeted cells, disrupted architectural organization, loosened cell-cell contacts, decreased glycogen storage, and cytoplasmic rather than membranous localization of the E-cadherin homolog L-CAM, highlighting the function of β-catenin in expansion of liver progenitors (Suksaweang et al.,2004). Conversely, expression of the WNT antagonist Dkk-1 or a dominant-negative LEF1 to inhibit β-catenin activity produced undersized livers with decreased cell proliferation, increased apoptosis, disrupted cord structures and sinusoids, and reduced L-CAM expression and glycogen storage capacity (Suksaweang et al.,2004). In zebrafish (Goessling et al.,2008) as well as Xenopus (McLin et al.,2007), Wnt signaling also promotes the growth and differentiation of liver progenitors after liver specification.
Embryonic day-10 livers cultured ex vivo in the presence of serum-free, WNT3A-conditioned medium show biliary hyperplasia and impaired hepatocyte differentiation. Culturing in WNT3A-conditioned serum-free medium plus sFRP-1 resulted in a decreased proliferation and survival as well as morphological defects. Growth in serum-containing medium rescued liver phenotypes, as did addition of HGF to serum-free WNT3A conditioned medium, suggesting that while β-catenin is sufficient for BEC differentiation, both MET and β-catenin activity are required for hepatocyte differentiation (Fig. 3E) (Hussain et al.,2004).
A similar culture system in which embryonic day-10 livers were cultured in the presence of antisense phosphorodiamidate morpholino oligomers (PMO) targeting β-catenin for 72 hr resulted in a substantial decrease in organ mass owing to a pronounced decrease in proliferation and significant increase in apoptosis, implicating β-catenin in both hepatoblast expansion and survival. In addition, PMO-treated livers showed similar expression of albumin as well as staining with Hep-Par, a marker of hepatocyte lineage (the urea cycle enzyme carbamoyl phosphate synthetase 1) (Butler et al.,2008), relative to control-treated and untreated livers, though interestingly, the Hep-Par+ albumin+ cells also exhibited c-kit staining, suggesting a lack of differentiation. PMO-treated livers also exhibited a lack of CK19-positive cells, and a marked increase in c-kit-positive cells (marker of hepatoblasts) around ducts (Monga et al.,2003).
Furthermore, APC null mouse embryos exhibit premature beta-catenin activation leading to differentiation of hepatoblasts to BECs, which, when transplanted into adult mice, lead to formation of fully differentiated ducts, thus lending in vivo support to the temporal role of beta-catenin in BEC specification of hepatoblasts.
Finally, more compelling evidence that absence of β-catenin activity leads to failure of hepatoblast expansion and defects in both biliary specification and hepatocyte maturation comes from the generation of a mouse line with a hepatoblast-specific (FoxA3-Cre driven) β-catenin deletion (Tan et al.,2008). Embryos possessing this deletion die between E16–E17 with undersized livers due to decreased hepatic cell proliferation and increased apoptosis likely resulting from impaired regulation of oxidative stress. Intriguingly, these animals possess an apparent defect in hepatoblast differentiation, with parenchymal cells exhibiting the high nuclear to cytoplasmic ratio and unpolarized morphology characteristic of uncommitted E13/14 stage hepatoblasts. Furthermore, knockout livers show a deficit both in BECs and in expression of the transcription factors C/ebpα and HNF4α characteristic of hepatocyte differentiation. Absence of hepatic beta-catenin also leads to decreased expression of α-fetoprotein, albumin (Alb), cyclin-D1, the adherens junction protein E-cadherin, the tight junction protein ZO-2 (Tjp2), as well as β-catenin targets glutamine synthetase (Glul), regucalcin (Rgn), Egfr, cytochrome p450 oxidases Cyp2e1 and Cyp1a2, Glutathione-S-transferase Gsta3, Gsto1, and Gstm1, ornithine aminotransferase (Oat), and Leukocyte cell-derived chemotaxin 2 (Lect2), as well as numerous other proteins detected in mature hepatocytes. The resident immature cells also show elevated levels of 4-hydroxynonenal and malondialdehyde indicative of increased oxidative stress, possibly owing to the decreased expression of glutathione-S-transferases.
Differentiation of cultured hepatic progenitors from E14.5 embryos is impaired in the presence of sFRP3, a member of the family of secreted frizzled related proteins that generally act as soluble antagonists of Wnt signaling (Bi et al.,2009). While sFRPs sFRP1, sFRP2, sFRP4, and SFRP3/FZDB are all capable of binding WNT3A (Wawrzak et al.,2007), sFRP3 appears to be distinct in its function, however, as it does not inhibit Wnt3a signaling (Galli et al.,2006; Wawrzak et al.,2007), but may instead inhibit EGF signaling (ScardigLi et al.,2008).
Together, the above findings point to a highly temporal and rather pleiotropic role for β-catenin in hepatic morphogenesis (see Fig. 3 for summary). β-catenin plays a key role in the expansion of the hepatic bud directly through regulation of cell proliferation and indirectly through modulation of cell survival via regulation of oxidative stress. In addition, it plays an important and temporal role in the commitment of hepatoblasts to biliary epithelial cells, and eventually also directs the process of progressive differentiation of the hepatocytes, possibly through regulation of independent sets of targets of the Wnt pathway or through regulation of cell adhesion and/or polarity.
Wnt and Frizzled Expression in Liver Development: Major Upstream Effectors of β-catenin Signaling
While many WNT ligands and their cellular sources have yet to be identified in mouse during development, analysis of expression of Wnt and Fzd genes in whole liver has been analyzed across several stages of liver development, and in individual cell types in the adult liver. During development, many Wnt genes are expressed in the liver including Wnt1, Wnt2b, Wnt3, Wnt3a, Wnt4, Wnt9b, Wnt10a, and Wnt10b (Bi et al.,2009). Fzds 1–10 are also expressed, with Fzd 2, 3, 5, 6, 7, 8 most prominent prenatally (Bi et al.,2009). SFrp2, 3, and 5 mRNA are high at E12.5 and E14.5 and then decline with the onset of differentiation, but while expression of sFrp1 and 4 also decreases as differentiation occurs, their mRNA expression level remains significant throughout pre- and post-natal development (Bi et al.,2009), perhaps indicative of a role in postnatal liver zonation. Differentiation of cultured hepatic progenitors from E14.5 embryos is impaired in the presence of sFRP3, a member of the family of secreted frizzled-related proteins that generally act as soluble antagonists of Wnt signaling (Bi et al.,2009). While sFRPs sFRP1, sFRP2, sFRP4, and SFRP3/FZDB are all capable of binding WNT3A (Wawrzak et al.,2007), sFRP3 appears to be distinct in its function, however, as it does not inhibit Wnt3a signaling (Galli et al.,2006; Wawrzak et al.,2007), but appears instead to inhibit Wnt5a (Qian et al.,2007; Liu et al.,2008) as well as EGF signaling (ScardigLi et al.,2008).
In adult mouse liver, Wnt4 is highly expressed by sinusoidal endothelial cells (SECs), stellate cells, and biliary epithelial cells (BECs), and Wnt9b is expressed by both SECs and BECs (Zeng et al.,2007). Adult hepatocytes express Fzd2, 4, 7, and 8, as well as producing Wnt5b (Zeng et al.,2007). Immunohistochemistry to detect WNTs expressed around the developing chick liver detected WNT3A and WNT8B in the localized liver growth zones, as well as diffuse peripheral and central WNT5A and WNT11 staining (Suksaweang et al.,2004).
Evidence for the role of sinusoidal endothelial cells in secreting WNT signals necessary for liver development emerged when the generation of a Vegfr2-null mouse deficient of endothelial cells was found to exhibit failure of both hepatoblast migration from the liver bud and hepatocellular cord formation (Matsumoto et al.,2001). In the developing chick, the liver buds form around, and eventually envelop, a vein running through the septum transversum (the ductus venosus) that secretes a chemoattractant called Neurturin (Tatsumi et al.,2007). Thereafter, hepatoblasts move radially within the septum transversum mesenchyme to arrange into hepatocellular cords lined by sinusoidal endothelial cells, which are now known to mediate this process through secretion of WNT9A (Matsumoto et al.,2008). While these findings illustrate the contributions of various Wnt proteins in liver development, a better assessment of their expression at the protein level and their effect on beta-catenin activity, as well as an exhaustive characterization of all the Wnt and Fzd players and their contributions to liver development, remain to be done.
Non-Canonical Wnt Pathways and Non-Wnt Mediated β-Catenin Activity
Of note, WNT activation of FZDs can activate either canonical, β-catenin mediated signaling or β-catenin independent, non-canonical WNT signaling pathways (WNT/Ca2+ and WNT/planar cell polarity), depending on cell context (Mikels and Nusse,2006). WNT/PCP pathway activity promotes elongation of dorsal meso-endoderm during gastrulation in conjunction with FGF signals, and regulates gut elongation in Xenopus (Li et al.,2008) and gut tube fusion in zebrafish via wnt4a, silberblick/wnt11, and wnt11-related (Matsui et al.,2005). In addition, the foregut-secreted sFrp5 regulating β-catenin activity during gut patterning also inhibits Wnt11-mediated PCP activity in order to allow proper foregut morphogenesis (Li et al.,2008). Recently, activation of WNT/Ca2+ pathway by WNT5 binding to Ryk was discovered to regulate directional cell migration during gastrulation. Overall, the role of non-canonical Wnt signaling in developing liver has yet to be fully investigated.
Abundant evidence also suggests β-catenin can be activated by signals other than WNTs, such as growth factors secreted by mesenchyme cells during liver development. One such growth factor is fibroblast growth factor, or FGF. The FGF family consists of 22 FGF ligands and seven tyrosine kinase FGF receptors that activate the Ras-Raf-mitogen activated protein kinase pathway (Ornitz and Itoh,2001). Cardiac mesodermal expression of FGFs regulates endoderm commitment into hepatic or pancreatic lineages by activating expression of liver-specific genes and promoting proliferation of hepatoblasts (Jung et al.,1999). FGF receptors FGFR1 and FGFR4 are expressed by endoderm, and FGF1 and FGF2 were shown to activate expression of early liver genes in isolated mouse foregut endoderm. The hepatic commitment by FGFs is dose-dependent, as higher concentrations activate lung-specific genes, and selective inhibition of FGFR1 and FGFR4 reveals that inhibition of FGFR1 more strongly interferes with liver specification, as measured by levels of Nkx.2 (Serls et al.,2005). FGF8, also expressed by cardiac mesoderm, promotes outgrowth of nascent hepatic progenitors revealing a role for different FGF in different phases of liver development (Jung et al.,1999).
One FGF family ligand, FGF10, has been shown to regulate the proliferation of progenitor cells in other endoderm-derived tissues including lung (Nyeng et al.,2008), pancreas (Norgaard et al.,2003), stomach (Nyeng et al.,2007), and intestine (Sala et al.,2006). At mouse embryonic day 9, Fgf10 mRNA expression is detected in the septum transversum mesenchyme (STM) adjacent to the liver bud, and its expression persists until E14.5, shifting localization to the liver surface along the bile duct, then scattered within the liver at E12.5, and finally becoming limited to extrahepatic biliary system and portal vein at E14.5 (Kelly et al.,2001). Furthermore, recent work has revealed that FGF signals from embryonic stellate cells promote beta-catenin activation and liver bud expansion between E9.5 and E13.5. Fgf10 expression correlates with peak beta-catenin activity in hepatoblasts (Micsenyi et al.,2004), which express Fgfr2b, and Fgf10−/− and Fgfr2b−/− animals exhibit impaired liver bud expansion. Finally, ex vivo cultured E12.5 TOPGAL+/+ mouse livers show striking increases in β-galactosidase activity upon treatment with FGF10, consistent with nuclear translocation and canonical beta-catenin activity (Berg et al.,2007). Fgf10 and Fgfr2b knockout mice show decreased liver size and increased apoptosis relative to wild-type embryos and a decrease in proliferating hepatoblasts positive for phospho-Histone H3, albumin, and cytokeratin at E12.5, in many ways similar to the phenotypes described in liver models with β-catenin activity inhibited (Tan et al.,2008). Consistent with a role for FGF activity in stem cell renewal in developing liver, FGFR activation is detected in mouse liver at E10, E11, and E12, when the liver bud is comprised mainly of proliferating, undifferentiated hepatoblasts, but is absent thereafter, when significant differentiation occurs (Sekhon et al.,2004). Treatment of ex vivo cultured whole E10 mouse livers with FGF1, 4, or 8 enriches the organs for cells positive for c-kit, alpha-feto protein, and CK19, all markers of progenitor cells, and prevents formation of ductal structures, indicating an inhibition of differentiation. Additionally, increases in proliferation by Ki67 staining are seen in the presence of all three FGFs and significant decreases in apoptosis by TUNEL stain are also observed in the presence of FGF8. Finally, a marked increase in cytoplasmic/nuclear β-catenin staining was also observed with FGF treatment, along with an increase in membranous β-catenin (Sekhon et al.,2004). Of note, the most potent effects were seen with FGF8, the FGF family member identified as promoting expansion of liver progenitors in a previous study (Jung et al.,1999).
Another growth factor capable of WNT-independent activation of β-catenin is hepatocyte growth factor (HGF) (Monga et al.,2002), the ligand for the receptor tyrosine kinase c-MET (Naldini et al.,1991a,b). Immunofluorescently labeled MET and β-catenin co-localize at the membrane in normal adult rat liver, and co-immunoprecipitation analysis reveals that β-catenin-MET complexes exist at the membrane of hepatocytes independently of β-catenin-E-cadherin complexes. Dose-dependent nuclear translocation of β-catenin is observed in primary hepatocyte cultures treated with HGF in the absence of WNT, along with a decrease in MET-β-catenin association by co-immunoprecipitation, effects not seen with a dominant-negative MET with deleted tyrosine kinase domain (Monga et al.,2002). Mutation of various β-catenin tyrosine residues pinpoint Y654 and Y670 as MET phosphorylation targets critical for inducing nuclear translocation. Non-phosphorylatable Y654F/Y670F mutants were shown to be deficient in TCF binding and cell cycle entry (Zeng et al.,2006). The previously described study revealed that E10 ex vivo cultured livers exhibited biliary hyperplasia in serum-free, WNT3A-containing medium, but underwent normal growth and morphogenesis in the presence of WNT3A and HGF (Hussain et al.,2004). This suggests a significant role of the HGF/Met and Wnt/β-catenin signaling and their crosstalk in hepatic development. Interestingly, the respective knockouts of Met and β-catenin have a notable similarity in hepatic phenotypes (Schmidt et al.,1995; Tan et al.,2008).
Further investigation of the relationship between MET activation and β-catenin activity comes from a study in which naked plasmid DNA expressing HGF under the control of the CMV promoter was hydrodynamically injected into tail veins of mice. These mice developed hepatomegaly characterized by decreased Met-β-catenin association and increased β-catenin nuclear translocation relative to empty vector-injected mice. Most tellingly, reproducing this experiment in β-catenin null mice revealed no hepatomegaly, but rather persistence of the decreased liver weight-to-body weight ratio that characterizes these animals, confirming that HGF-induced stimulation of liver growth depends on β-catenin activity (Apte et al.,2006). HGF has been shown to promote increased nuclear translocation and transcriptional activation by beta-catenin concomitant with decreased GSK3beta activity, suggesting HGF/Met signaling may also regulate beta-catenin stabilization (Papkoff and Aikawa,1998).
Other proteins can alter β-catenin activity by regulating its affinity for its binding partners. In addition to targeting by c-Met, Y654 can be phosphorylated by both EGFR and SRC (Roura et al.,1999). Phosphorylation of residue Y142 by FYN, FER, or MET decreases binding affinity for alpha-catenin (Brembeck et al.,2004), and there is also evidence of phosphorylation of Y489 by ABL, which may regulate cadherin affinity and nuclear translocation of β-catenin (Rhee et al.,2007). Though not investigated to date, it is possible FGF also activates beta-catenin via phosphorylation of Y654 and/or Y142, promoting its dissociation from adherens junctions and allowing it to enter the nuclear/cytoplasmic pool to effect gene transcription. HGF activates MET during liver bud expansion in the mouse embryo, and it is possible that alone, or in conjunction with one of the many WNTs expressed during this time, it is responsible for activation of β-catenin activity. In addition to kinases, phosphatases also regulate β-catenin phosphorylation and likewise can affect its localization and activity (Lilien and Balsamo,2005).
Postnatal β-Catenin Activity in Liver
Postnatal liver growth and development is also dependent on β-catenin activity. Extensive cell proliferation occurs in the liver after birth, in conjunction with a substantial increase in β-catenin protein and nuclear translocation. Increased Ctnnb1 gene expression occurs at P15, and a cyclic pattern of GSK3β activation and inactivation occurs during postnatal development, with GSK3β inactivation occurring at P10 and P20, and activation occurring at P5 and P20. Beta-Catenin colocalizes with E-cadherin during postnatal development, but association with MET does not occur until P20. In mice with an albumin-Cre-driven, hepatocyte-specific deletion of β-catenin occurring by postnatal day 15, liver weight/body weight ratios are significantly decreased relative to wildtypes between day 15 and day 20, and this decrease persists throughout the lives of the animals (Apte et al.,2007). Postnatally, beta-catenin is localized mostly at the hepatocyte membrane with some additional cytoplasmic staining in centrizonal hepatocytes, though nuclear beta-catenin localization by immunohistochemistry may be underappreciated due to the harsh antigen retrieval methods required to detect it in mouse liver, and immunoblot analysis of nuclear extracts may be a more reliable means of such detection. Virtually all other cells contained in an adult liver such as BECs, endothelial cells, and stellate cells express beta-catenin, although the role of Wnt signaling in such cells is only beginning to be investigated (Zeng et al.,2007).
In the adult liver, hepatocytes are not equivalent, with position along the portocentrovenular axis within a liver lobule dictating expression of metabolic genes involved in drug metabolism, carbohydrate metabolism, ammonia detoxification, and bile production and secretion. WNT/β-catenin has now been identified to be playing a key role in this phenomenon. Complementary localization patterns exist for active β-catenin, expressed perivenously, and its negative regulator APC, expressed periportally. Induction of liver-specific deletion of Apc or inhibition of WNT signaling by virally mediated overexpression of Dkk1 reveals that β-catenin activity controls zonal expression of metabolism genes (Benhamouche et al.,2006). Shortly after Apc deletion, livers show extension of several perivenously localized enzymes, including the β-catenin targets glutamine synthetase (Glul) and the glutamine transporter Glt1 (Benhamouche et al.,2006). Likewise, expression of periportal enzymes involved in ammonia and urea metabolism, Glutaminase2 (Gls2), Arginase1 (Arg1), Carbamoyl phosphate synthetase I (Cps I), and Phosphoenolpyruvate carboxykinase 1 (Pck1) were suppressed by Apc deletion, and additional defects were observed in nitrogen metabolism. Moreover, mice expressing activated mutant forms of β-catenin, either transgene expression of ΔN131-β-catenin or adenovirally infected S37A-β-catenin, showed similar metabolism gene expression perturbations to Apc null mice. Infection of wild-type mice with an adenovirus encoding the WNT pathway inhibitor Dickkopf-1 (Dkk1) produces a massive conversion of perivenous hepatocytes to periportal hepatocytes (Benhamouche et al.,2006). Furthermore, deletion of β-catenin, but not c-myc, results in a loss of GS expression and an extension of expression of the periportally-expressed carbamoylphosphate synthetase I (CPS I) around the central vein, implicating β-catenin in c-myc-independent suppression of expression of periportal metabolic genes (Burke and Tosh,2006). Finally, while spontaneous differentiation of liver stem cells leads to their expression of periportal enzymes, stabilization of β-catenin in these cells by treatment with the GSK3β inhibitor 6-bromoindirubin-3′-oxime (BIO) switches their gene expression profile to that of a perivenous hepatocyte.
ChIP analysis of perivenously expressed GS and Cyp1a1 and periportally expressed Gls2 and H19 revealed that the level of binding of TCF family member LEF1 to HNF4a binding sites on promoters regulates their transcription. The ChIP results suggest a model in which β-catenin-induced LEF1 activation regulates LEF1 inhibition of HNF4α repressive activity, with LEF1 binding to and displacing repressive HNF4α at its consensus sites on promoters in perivenous hepatocytes with active WNT/β-catenin activity. For periportal genes Gls2 and H19, during transcriptional repression seen in the context of active WNT/β-catenin pathway activity, HNF4α is displaced, concomitant with the recruitment of LEF1 only to its own consensus sites (Colletti et al.,2009).
Very recently, β-catenin was also shown to regulate expression of both phase I drug-metabolizing Cytochrome P 450 (Cyp) enzymes and phase II glutathione-s-transferases in perivenous hepatocytes (Loeppen et al.,2005; Tan et al.,2008; Giera et al.,2010).
Role of β-catenin in Adult Hepatic Progenitors or Oval Cell Biology
Wnt/beta-catenin signaling is a key player in the process of liver regeneration across various species. Utilization of models such as surgical resection of the liver or toxicant-induced hepatic injury reveals that beta-catenin activation appears to be necessary for optimum regeneration of the remnant liver. Additionally, Wnt/beta-catenin signaling has also been uncovered in the progenitor cell-mediated regeneration of the liver that occurs only in response to regenerative stimulation in a context that prevents hepatocyte proliferation. This has been observed in response to chronic or acute liver damage, both experimentally and clinically (Hu et al.,2007; Apte et al.,2008; Yang et al.,2008). Several signaling pathways have now been shown to play an important role in emergence, expansion, and differentiation of these transiently amplifying progenitor cells, also referred to as oval cells (Erker and Grompe,2007).
In a model of oval cell activation induced by treatment with 2-acetylaminofluorine followed by 2/3 partial hepatectomy (pHx) in rats, increased presence of active nuclear and cytoplasmic β-catenin in oval cells between 5 and 10 days post-pHx correlated with increases in oval cell proliferation (Apte et al.,2008; Yang et al.,2008). Increased WNT1 and FZD2 expression appeared at these times as well, coinciding with decreased WNT pathway inhibitors WIF1 and GSK3β (Apte et al.,2008). More recently, a role for Wnt1 was also identified in the differentiation of the oval cells to hepatocytes, such that shRNA-mediated Wnt1 suppression led to not only decreased oval cell proliferation but also affected their differentiation to mature hepatocytes (Williams et al.,2010). Thus, as in development, the role of Wnt signaling in oval cell proliferation and differentiation may be highly temporal and context-dependent.
Oval cell activation in mice is also observed in animals fed a diet containing 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC). This treatment leads to biliary and hepatocyte injury followed by atypical ductular proliferation and oval cell response (Preisegger et al.,1999; Thompson et al.,2010). Wnt signaling in oval cells after DDC injury has also been investigated through the use of hepatocyte-specific beta-catenin conditional null mice (Tan et al.,2006) (Fig. 4). Treatment of β-catenin null mice with DDC resulted in a decreased number of cells positive for the oval cell marker A6 relative to treatment of wild-type animals (Apte et al.,2008). In another study, quantitative reverse transcription polymerase chain reaction and in situ hybridization identified upregulation of several Wnts in DDC-treated animals around portal triads and areas of atypical ductular response, with most pronounced increases in Wnt 7a, 7b, 9b 10a, and 11 mRNA levels. DDC-treatment of the TOPGAL transgenic β-catenin reporter mouse strain confirmed the increase in β-catenin/TCF activity in oval cells/atypical ductular cells. Moreover, increased active β-catenin, and β-catenin/TCF reporter activity was observed in oval cells isolated from livers of DDC-fed mice in response to purified WNT3A (Hu et al.,2007). Finally, hydrodynamic tail vein injection of naked plasmid DNA expressing constitutively active (S37Y) β-catenin elicited a significantly higher number of A6-positive cells in response to DDC treatment (Yang et al.,2008).
Interestingly, cells staining with an OV6 oval cell antibody and exhibiting nuclear/cytoplasmic β-catenin localization were detected in sections of cirrhotic livers and hepatocellular carcinomas from human patients, suggesting β-catenin drives oval cell expansion in human liver pathologies as well. Additionally, OV6+ cells isolated from human liver cancer cell lines show enrichment for the epithelial stem cell marker CD133+ as well as several other progenitor/stem cell markers, as well as exhibiting increased tumorigenicity in nude mice, and resistance to chemotherapy drugs. Treatment with a stabilizer of β-catenin produces enrichment of these cultures for OV6+ cells, highlighting a potential role for β-catenin in oval cell expansion in liver tumors (Yang et al.,2008). These observations are significant since a subset of hepatocellular cancers may have cancer stem cell origin (Sell and Leffert,2008) and Wnt/beta-catenin signaling, which is independently implicated in this form of tumor-type due to mutations in beta-catenin gene, may be contributing to hepatocarcinogenesis in more than one way (Monga,2009). More recently, strong activation of Wnt/beta-catenin signaling was also detected in proliferating hepatic progenitors evident in acute necrotizing hepatitis (Spee et al.,2010).