Wnt-β-catenin in hepatobiliary homeostasis, injury, and repair

Abstract

Wnt-β-catenin signaling has emerged as an important regulatory pathway in liver, playing key roles in zonation and mediating contextual hepatobiliary repair after injuries. In this review, we will address the major advances in understanding the role of Wnt signaling in hepatic zonation, regeneration, and cholestasis-induced injury. We will also touch upon some important unanswered questions, and discuss the relevance of modulating the pathway to provide therapies for complex liver pathologies that remain a continued unmet clinical need.

Introduction

The liver is vital to survival, performing many functions that can be broadly categorized into metabolism, synthesis, detoxification, and homeostasis. As a result, liver is bestowed with unique properties including metabolic zonation which allows it to deliver eclectic functions in an efficient manner. Hepatocytes and cholangiocytes, the two hepatic epithelial (‘hepithelial’) cells of the liver, which are derivatives of a common progenitor during development, play a major role in most of these functions. To ensure the continued survival of this critical organ, and since these functions require cells to maintain their highest levels of differentiation, the hepithelial cells, especially hepatocytes, have long lifespan and overall, liver shows very slow turnover. Additionally, there are no active stem cells in the liver. Therefore, the liver possesses a distinctive regenerative capacity, and responds to diverse acute and chronic hepatobiliary insults in many effective ways.

Over the past 2 decades, many labs have demonstrated an essential role of Wnt/β-catenin signaling in liver physiology and pathobiology. Its normal activity is indispensable for hepatic development, biliary morphogenesis, postnatal hepatic growth, hepatocyte maturation, and hepatic zonation (Reviewed in (1–8)). Additionally, β-catenin signaling promotes the development and progression of liver diseases such as diet-induced steatohepatitis, fibrosis, and cancer (Reviewed in (2, 5, 7–11)). One of its best-known functions, however, is regulating the regenerative process after surgical resection or toxicant-induced injury (Reviewed in (2, 12–14)). Significant progress has been made in the last few years in elucidating the cell-molecule circuitry involved in this process. This review will discuss these findings and also draw parallels between acute and chronic injury that may provide insight into the role of Wnt/β-catenin signaling in chronic liver injury, particularly in cholangiopathies.

Wnt-β-catenin signaling

In the absence of Wnt proteins, β-catenin is degraded by an active destruction complex consisting of scaffold protein Axin, adenomatous polyposis coli (APC), glycogen synthase kinase 3β (GSK3β), and casein kinase 1α (CK1α) (Reviewed in (14)). β-Catenin is phosphorylated serially by CK1α and GSK3β at serine-45 (S45), S33, S37 and threonine 41 (T41). Phosphorylated β-catenin is recognized by β-transducin repeat-containing protein (βTRCP), an E3 ubiquitin ligase, leading to its proteasomal degradation. The Wnt ligands (19 members) require glycosylation and palmitoylation by porcupine for activity, and Wntless (Wls) for secretion from a cell. When Wnts bind to one of the Frizzled receptors (10 members) and 1 of the 2 redundant co-receptors - the low-density lipoprotein receptor-related protein 5 or 6 (LRP5/6) - the degradation complex is recruited to the cell membrane, disrupting β-catenin destruction. This promotes β-catenin stabilization and nuclear translocation where it binds and activates T cell factor/lymphoid enhancer factor (TCF/LEF) transcription factors to induce tissue-specific target genes.

β-catenin is not only an essential component of the Wnt pathway but also functions as part of adherens junctions, bridging E-cadherin to α-catenin and F-actin to assist in cell-cell adhesion (2, 5). As such it plays an important role in maintaining the blood-bile barrier (15). Loss of β-catenin from both hepatocytes and cholangiocytes using floxed-β-catenin and Alb-cre mice did not cause a defect in adherens junctions (16). However, β-catenin KO did show aberrant biliary canaliculi and sluggish bile flow, resulting in signs of intrahepatic cholestasis that worsened with age and was exacerbated following methione-supplemented choline-deficient diet (17, 18). Adherens junctions were maintained in β-catenin KO by virtue of increased γ-catenin, normally a desmosomal protein, which complexed with E-cadherin in the absence of β-catenin in KO (19, 20).

Metabolic zonation and β-Catenin signaling

Because of its strategic location, liver performs over 500 functions that are pertinent to overall health and homeostasis. Following digestion, nutrients are absorbed into blood and enter the liver through the portal vein, which also brings in toxins and microbial products to be processed by the liver. Hepatocytes sense and regulate the components of the portal blood as it enters the hepatic sinusoids where it mixes with arterial blood brought in by the hepatic arterial system. As the blood flows downstream through the sinusoids, there is free exchange of metabolites, proteins, and lipids between the blood and the hepatocytes, the functioning units of the liver. The hepatocytes by virtue of their location within a hepatic lobule will synthesize glucose, store glucose in the form of glycogen, utilize lipids for β-oxidation or synthesize lipids, and maintain carbohydrate and fat homeostasis. Another important hepatic function is detoxification of ammonia, the by-product of protein metabolism, which must be prevented from exiting the central vein. Hepatocytes must also participate in synthesis of vital proteins such as those involved in coagulation cascade, transport and other functions, as well as detoxification of xenobiotics. Additionally, hepatocytes have to synthesize, secrete, and uptake bile acids for proper composition, flow, and secretion of bile through biliary canaliculi and into the intrahepatic and extrahepatic biliary tree. To perform such eclectic functions efficiently and effectively, liver cells use an admix of ‘divide and conquer’ or ‘division of labor’ and ‘conveyor belt’ approaches, that can altogether be referred to as zonation (21, 22).

The concept of divide and conquer and division of labor can be found in many metabolic functions of the liver where the expression of the genes encoding proteins regulating specific subgroups with one metabolic function is compartmentalized and restricted to hepatocytes located within specific zones of the liver lobule (Reviewed in (23)). For example, within fat metabolism, the β-oxidation related genes are expressed in zone-1 or periportal hepatocytes, whereas the lipogenesis genes are present in zone-3 or pericentral cells. Likewise, in glucose homeostasis, gluconeogenesis genes are restricted to zone-1 and glycolysis genes are found in zone-3 hepatocytes. Most ammonia detoxification genes involved in urea metabolism are present in hepatocytes in zone-1 while glutamine synthesis machinery is evident in zone-3 hepatocytes to prevent any metabolized ammonia from escaping via the central vein to the hepatic vein and in turn into systemic circulation.

In addition to spatial zonation, metabolic function is regulated by temporal zonation as well. In mammals, the circadian clock is a time-measuring device that integrates both internal and environmental stimuli, and in turn relays cues to the body’s organs (24). In the liver, this causes fluctuations in hepatic functions depending on the time of day (25). A number of key physiological functions are subject to daily oscillations, namely xenobiotic and bile acid metabolism (26, 27). Genes regulating fatty acid, glucose, and amino acid metabolism are also expressed rhythmically, and are coupled to nutrient influx and food intake (24, 25). An unexpected meal or starvation can cause deregulation of timing signals and clock-regulated gene transcripts in the liver, and in turn food metabolites can influence feeding patterns (25). Liver is also subject to changes in internal body temperature induced by feeding or environmental temperature changes, which can shift hepatic gene expression (28).

Spatiotemporal compartmentalization also determines the extent and location of damage during initiation of liver injury. With continued insult, eventually the whole liver lobule is affected, and the entire liver undergoes molecular and cellular decompensation. For example, in adult patients with non-alcoholic fatty liver disease, steatosis first occurs in zone-3 of the liver (29). Intriguingly, in pediatric cases, steatosis begins in the periportal zone (30). However, as the insult progresses, steatosis and then steatohepatitis becomes panzonal. These differences, especially during early stages of the disease may be due to initial perturbations that have been previously ignored and which if identified can help in devising specific preventive strategies (31). Likewise, due to zonated expression of several xenobiotic metabolism related genes such as cytochrome P450s including 2E1, alcohol-induced liver injury is mostly pericentral as well (32). Mice exposed to allyl alcohol, a model of toxicity caused by food flavoring agents and synthesis of industrial chemicals, develop periportal injury due to the presence of alcohol dehydrogenase which drives accumulation of acrolein in this region (33).

Efforts to identify the molecular and genetic underpinnings of hepatic zonation have demonstrated that Wnt-β-catenin signaling is an important regulator of hepatocyte gene expression in zone-3 (34, 35). Additionally, the cell-molecule circuitry of Wnt signaling in zone-3 is now known. However, there are still important questions that remain to be addressed.

The first evidence that β-catenin regulates the expression of liver-specific genes was reported two decades ago when glutamine synthetase (GS), ornithine aminotransferase (OAT) and the glutamate transporter GLT-1 were shown to be regulated by β-catenin even though the study did not address zonation of these genes per se (36). The follow-up study utilizing liver-specific APC null mice established the graded presence of basal APC with highest levels in zone-1 and lowest in zone-3 (37). Loss of APC led to activation of β-catenin throughout the lobule, and with it occurred expanded expression of zone-3 genes to other zones. Gene expression analysis on isolated pericentral and periportal hepatocytes verified various differentially expressed genes and active β-catenin in zone-3 (38). This study along with two other studies at similar times also showed conditional loss of β-catenin from hepithelial cells resulted in absence of the same zone-3 genes and others including cyp2e1 and cyp1a2 (16, 39).

Whether the β-catenin activity in zone-3 is being controlled by Wnts or by other factors also known to regulate β-catenin like insulin/IGF-1, HGF, EGF and others was addressed through generation of liver-specific Wnt co-receptor mice. Deletion of LRP5 and LRP6 from hepatocytes led to loss of zone-3 gene expression similar to β-catenin loss from hepatocytes, indicating Wnt control of this phenomenon (40). In order to further understand the molecular underpinnings of active β-catenin in zone-3, subsequent studies have addressed the cell source of Wnts, which of the 19 Wnts, if any, are specific in controlling hepatic zone-3 gene expression, and if any of the Wnt receptors (Frizzled 1–10) are zonated.

Using in situ hybridization, Wang et al showed Wnt2 and Wnt9b to be present in endothelial cells located in the pericentral region of the hepatic lobule (41). They found Wnt2 to be evident in both endothelial cells lining the central vein and liver sinusoidal endothelial cells (LSEC) in zone-3, whereas Wnt9b seemed to be restricted to the former cell type. Deletion of Wntless, a protein essential for Wnt secretion, from endothelial cells resulted in loss of zone-3 genes. Wntless expression has been eliminated from hepatic endothelial cells using 3 different cre lines: Lyve1-cre (which knocks out floxed alleles in all endothelial cells of the liver), Stab2-cre (endothelial subtype-specific elimination of floxed alleles), and VEcad-Cre^ER (an inducible endothelial cell cre line). In all cases, β-catenin zone-3 activity and target gene expression was impaired (42–44). These studies unequivocally demonstrated that endothelial cells lining the central vein and/or the sinusoidal channels in zone-3 are the source of Wnts that regulate β-catenin activity in adjacent hepatocytes (Figure 1).

Figure 1: The role of Wnt-β-catenin in metabolic zonation.

Wnt2 and Wnt9b are expressed in the pericentral region of the hepatic lobule from liver sinusoidal endothelial cells and the endothelial cells lining the central vein. As a consequence, β-catenin is constitutively active in zone-3, where it regulates expression of genes involved in glucose metabolism, xenobiotic metabolism, ammonia metabolism, and lipid metabolism, among others. Many pathological processes in the liver are zonated, which has important biological and therapeutic implications. Left, autoimmune hepatitis is predominantly periportal, while in non-alcoholic fatty liver disease macrovesicular steatosis occurs in zone-3.

Although previous experiments showed that both Wnt2 and Wnt9b were expressed in zone-3 endothelial cells, their role in establishing pericentral zonation had yet to be confirmed. A recent study generated double Wnt2-Wnt9-floxed mice which upon breeding to Lyve1-cre successfully eliminated only these 2 Wnt genes from hepatic endothelia (45). This led to loss of hepatocyte zone-3 gene expression phenocopying disruption of Wnt-β-catenin signaling shown in endothelial cell specific-Wntless knockouts, hepatocyte-specific knockouts of Wnt co-receptors, or hepithelial cell-specific knockout of β-catenin. It is notable that while Wnt2 is expressed in zone-3/2 LSEC and in some central vein endothelial cells (CVEC), Wnt9b is almost solely expressed in CVEC. The mechanism restricting the expression of these Wnts to specific endothelial cells is still unknown. But because both of these Wnts are in relative hypoxic regions, an obvious mechanistic explanation is the changing oxygen gradient in the hepatic lobule. pO₂ in the periportal region is around 60–65 mm Hg and in the pericentral region is 30–35 mm Hg (46). Although the relative hypoxia in zone-3 may be the driver of constitutive Wnt expression in endothelial cells, this needs to be directly investigated.

The role of Frizzled receptors in maintaining hepatic zonation is another area under active investigation. RNAscope and molecular mapping showed that of the Frizzled (Fzd) receptors, Fzd 1, 4, and 6 have the highest expression in the liver lobule. However, with the exception of Fzd6, which was preferentially expressed in zone-1, the expression of most Fzds was not zonated (manuscript in communication). Future work will focus on directly addressing the role of each Frizzled in liver through knockout studies.

R-spondin (RSPO) ligands and their leucine-rich repeat-containing G protein-coupled receptors (LGR) potentiate Wnt/β-catenin signaling. Binding of RSPO to LGR4/5 induces the clearance of the cell-surface transmembrane ubiquitin ligases zinc and ring finger 3 (ZNRF3) and ring finger 43 (RNF43), which are negative regulators of Wnt signaling, thus promoting Wnt receptor endocytosis and turnover (47, 48). LRG4 and LRG5 are expressed in pericentral hepatocytes with high Wnt activity, and the loss of these two receptors abrogates Wnt signaling and results in loss of metabolic liver zonation (49). Further, RSPO blockade disrupts zonation while recombinant RSPO1 expanded Wnt signaling into periportal hepatocytes, demonstrating that the RSPO-LGR4/5-ZNRF3/RNF43 module is essential in regulating hepatic Wnt signaling and metabolic zonation (49, 50).

Perturbation of the Wnt-β-catenin circuit not only causes impairment of zone-3 gene expression but also leads to expansion of genes normally restricted to zone-1 (42, 45). In fact, single cell spatial transcriptomics performed on mice where endothelial cell-Wnt2-9b or the hepatocyte-LRP5/6 or β-catenin axis was perturbed resulted in appearance of zone-1 genes in zone-3 hepatocytes (45). This periportalization of livers in the absence of Wnt-β-catenin paracrine signal implies that the zone-3 hepatocyte identity is a result of simultaneous activation and repression of target genes. It is worth noting that as more single cell studies are performed on various hepatic cell populations, it is becoming clear that all cells within liver show specific gene expression based on their location within the lobule, including macrophages and endothelial cells (51, 52). Whether this zone-specific gene expression is due to a neighborhood-wide signaling dominance or is due to more intricate cell-cell and cell-molecule interactions needs to be addressed in the future.

β-Catenin signaling and hepatobiliary repair

Another unique characteristic of the liver is its ability to regenerate especially after acute loss of hepatic mass due to surgical resection or toxicant-induced injury. In fact, liver tries to maintain an ‘appropriate-for-body weight’ mass which may represent both its functional weight and reserve capacity, although this has not been directly tested. Suffice it to say that removal of up to two-thirds of its mass by a surgical procedure called partial hepatectomy triggers the remnant tissue to undergo cell division and some hypertrophy to eventually restore the mass appropriate for the body weight of the organism. The normal homeostatic level of liver weight to body weight ratio has been referred to as the hepatostat (53, 54).

Likewise, injury to hepatocytes by drugs like acetaminophen leads to necrotic death of hepatocytes in zone-3 due to formation of toxic intermediate N-acetyl-p-benzoquinoneimine (NAPQI) by pericentrally expressed cyp2e1 and cyp1a2 (55). This induces remnant healthy cells in zone-2 and even zone-1 to under proliferation that pushes these dividing cells into the necrotic zone once it has been “mopped” by tissue macrophages (56). Intriguingly, zonation is re-established as architecture is restored and hepatocytes and pericentral endothelial cells come into each other’s proximity as seen by the reappearance of GS following acetaminophen injury and repair (57). Thus another function of zonation is the compartmentalization of injuries to xenobiotics or hypoxia, which allows survival of a subset of hepatocytes, which then fuel repair of the tissue through replication after receiving appropriate signals.

Liver regeneration

The Wnt-β-catenin pathway has been shown to play an important role in directing and coordinating the repair process following both hepatic resection or toxicant-induced injury. Over two decades ago, β-catenin nuclear translocation was first reported following partial hepatectomy in rats (58). Rats have a rapid regenerative response to resection and hepatocytes show peak proliferation at 24 hours after surgery. β-Catenin nuclear translocation was evident within minutes of surgery and persisted for up to 24 hours, after which it normalized to the hepatocyte membrane. In mice, where kinetics of regeneration are slower with peak hepatocyte proliferation at 48 hours, nuclear translocation of β-catenin was evident at 1–3 hours along with formation of β-catenin-TCF4 complex; normalization was similarly delayed and occurred after 48 hours (40, 59). Knockdown of β-catenin by antisense in rats or genetic elimination of β-catenin from hepatocytes in mice led to a delay in hepatocyte proliferation and in turn regeneration (16, 60, 61). This coincided with decrease in several cyclins involved in cell cycle progression, especially cyclin-D1, which is a known target gene regulated by the β-catenin-TCF complex. In fact, Ccnd1 and cyclin-D1 upregulation, which was evident as early as 6–12 hours following hepatectomy in mice, initiates G1 to S phase transition in hepatocytes which in turn drives their replication.

Interestingly, although loss of β-catenin in hepatocytes resulted in diminished G1 to S phase transition and proliferation at 48 hours, this delay was eventually compensated by another mechanism allowing repair (16, 60). Recently, this compensatory mechanism was identified as the insulin-insulin receptor-mTORC1-cyclin-D1 axis, which is activated in the β-catenin-deficient livers and allowed regeneration after hepatectomy (62). It was interesting to note that chronic versus acute elimination of β-catenin from hepatocytes by Alb-cre or AAV8-TBG-cre, respectively, manifested differently in terms of regeneration kinetics following hepatectomy (62). While mice survived in both scenarios, regeneration remained suboptimal and mild proliferation persisted over many days after acute β-catenin deletion, whereas chronic deletion of β-catenin merely resulted in a delayed peak of proliferation. These observations suggest that chronic elimination of a gene may allow better signaling adaptation over time, albeit often through similar pathways as mice with acute deletion of the gene.

Since β-catenin can be controlled by various growth factors in addition to Wnt ligands, and because many growth factors and their receptor tyrosine kinases are well-known drivers of regeneration including HGF, EGF, TGFα, Insulin and others, it was relevant to address the upstream effector of β-catenin activation following hepatectomy. Deletion of the dual Wnt co-receptors LRP5–6 knockout mice from hepatocytes resulted in a notable delay in regeneration that phenocopied liver-specific β-catenin knockouts, indicating that β-catenin activation in normal mice is under the influence of Wnt ligands and not other factors (40). Similarly, deletion of LGR4/5 receptors also impaired hepatocyte proliferation after hepatectomy and downregulated expression of Wnt target genes implicated in cell cycle regulation, demonstrating that RSPO-LGR4/5-ZNRF3/RNF43 signaling is required for a proper Wnt/β-catenin mediated regenerative response (49).

To next address the cell source of Wnts leading to hepatocyte β-catenin activation, mice with genetic knockout of Wntless from hepatocytes, cholangiocytes, macrophages and endothelial cells were subjected to hepatectomy. Only endothelial cell specific loss of Wntless shared the same kinetics of delayed liver regeneration that were observed in liver-specific knockouts of either LRP5–6 or β-catenin (44). Isolated endothelial cells from regenerating livers at 12 hours post hepatectomy showed a multi-fold increase in expression of Wnt2 and Wnt9b, the same Wnt genes that are basally expressed in zone-3 endothelial cells and responsible for controlling β-catenin activation and its target gene expression in pericentral hepatocytes (44). Conditional deletion of the two Wnt genes from endothelial cells led to a dramatic decrease in cyclin-D1 levels and proliferation of hepatocytes at 24 hours and 48 hours after hepatectomy, thus unequivocally demonstrating the importance of these two Wnts in both zonation and hepatic regeneration (45) (Figure 2).

Figure 2: The role of Wnt-β-catenin in liver regeneration.

After two-thirds partial hepatectomy, the remnant liver regenerates due to hepatocyte entry into the cell cycle and subsequent proliferation. The Wnt-β-catenin pathway plays an important role in directing and coordinating this process. Wnt 2 and Wnt9b, which are secreted by endothelial cells in the central vein and sinusoids, act on neighboring hepatocytes to induce nuclear translocation of β-catenin. In combination with TCF/LEF, β-catenin then induces expression of cyclin-D1, which activates the cell cycle machinery and drives replication.

It is intriguing to note that cyclin-D1, which is expressed midzonally in normal livers and is dependent on β-catenin, is induced in periportal and midzonal regions following hepatectomy and eventually becomes pan-zonal. This ‘wave’ of cyclin-D1 is followed 24 hours later by the entry of hepatocytes in S-phase following closely by cell division in these zones (44, 45). At baseline in a resting liver, mRNA detection techniques like Molecular Cartography, in situ hybridization, or RNAscope show localization of Wnt2 and Wnt9b only in zone-3 endothelial cells (41, 45). However, it is possible that Wnt2 and Wnt9b proteins pre-exist in zone-1 and zone-2 endothelial cells at baseline and are released only upon appropriate signaling following hepatectomy. Another possibility is that these two Wnts are being constitutively released, but due to high APC protein that exists at baseline in these two zones, β-catenin is degraded, canceling out the effects of these Wnt proteins (37). APC protein levels have been shown to fall after hepatectomy in rats, which may facilitate the signaling of Wnt2 and Wnt9b in zones 1 and 2 (58). Other pathways may also contribute to increased Wnt signaling in these zones. For example, VEGFR2-AKT-dependent upregulation of transcription factor Id1 was shown to mediate expression and release of angiocrine factors Wnt2 and HGF in liver sinusoidal endothelial cells following hepatectomy (63). A fourth possibility is that the transient ectopic expression of Wnt2 and Wnt9b very early after hepatectomy in zone-1 and zone-2 endothelial cells could be due to changes in biophysical processes like mechanosensing or alterations in oxygen gradient, although this hypothesis would need further experimental support.

The Wnt-β-catenin pathway not only plays a role in regeneration after hepatectomy but after other forms of acute injuries as well. Acetaminophen overdose leads to pericentral necrosis followed by proliferation of surviving hepatocytes in other zones. Wnt-β-catenin activation is an important player in this process. In the absence of hepatocyte β-catenin, there was a delay in regeneration after acetaminophen overdose. A higher dose of acetaminophen led to reduced liver regeneration and higher mortality as well as decreased β-catenin activity, as shown in an incremental dose model of acetaminophen (64). The source and identity of Wnts in acetaminophen-induced regeneration remain elusive, although macrophages have been shown to secrete signals that promote regeneration near areas of necrosis (65). Carbon tetrachloride similarly causes pericentral injury to liver lobules. β-Catenin activation under the control of endothelial cell Wnts was shown to play a role in liver repair following such insult, although the exact identity of those Wnts remains unknown (66).

Cholestatic liver injury

Primary sclerosing cholangitis (PSC) is the prototypical cholestatic liver disease and is characterized by inflammation and fibrosis of the intra- and/or extra-hepatic bile ducts, resulting in chronic disease. The underlying pathophysiology of PSC is poorly understood, as PSC is a heterogeneous disease and includes patients with variable pathogenetic backgrounds (67). Some of the etiologies proposed include genetic predisposition, ischemic damage to bile ducts, humoral and cellular immunologic abnormalities, and entry of bacteria into portal circulation due to leaky gut. The “toxic bile” concept has also been posited as a major contributor in the pathogenesis of PSC, where regurgitation of bile through leaky cholangiocyte junctions can induce hepatic injury and production of inflammatory mediators, leading to cholangitis and fibrosis (68).

Compensatory hepatocyte proliferation to replace damaged parenchymal cells may offer some protection against chronic cholestatic liver injury. Mice lacking the Mdr2 gene (Mdr2 KO) develop sclerosing cholangitis which phenotypically resembles PSC in patients. This is caused by bile duct leakage due to the absence of phosopholipid secretion in bile (69–71). Chronic damage to hepatocytes in the Mdr2 KO model results in a state of continuous regeneration which helps to maintain normal liver function despite persistent injury (72). Another common experimental model of cholestasis is bile duct ligation (BDL), a permanent surgical obstruction that results in jaundice, ductular proliferation, inflammation, and fibrosis. In response to accumulation of bile acids in liver, mice develop extensive tissue injury characterized by areas of necrosis, called bile infarcts. BDL initiates a regenerative response that occurs approximately 48 hours after the peak of injury, similar to that seen after partial hepatectomy (73). Furthermore, the number of proliferating hepatocytes correlates with the extent of injury, suggesting that the extent of regeneration may be adapted to compensate for amount of tissue lost to necrosis (74).

The role of β-catenin in cholestasis was first explored in the context of the BDL model. Conditional knockout mice that lack β-catenin in liver exhibited modestly higher basal hepatic bile acids than controls, which is most likely a consequence of sluggish bile flow due to canalicular abnormalities and loss of tight junction proteins (18). However, genetic deletion or exogenous inhibition of β-catenin after BDL significantly reduced liver injury, fibrosis, bile infarcts, and ductular proliferation, secondary to decreased total hepatic bile acids. This led to the identification of a novel association of β-catenin and the bile acid sensor, the farnesoid X receptor (FXR). This interaction regulates bile acid synthesis and transport through direct inhibition of FXR; in the absence of β-catenin, bile acid synthesis is suppressed and hepatobiliary injury improved after BDL (75, 76).

To extend these findings to other preclinical models, β-catenin was exogenously deleted from Mdr2 KO, using DsiRNA conjugated to either a lipid nanoparticle or N-acetyl-galactosamine. Unexpectedly, Mdr2 KO/β-catenin knockdown (KO/KD) mice showed exacerbated hepatobiliary injury, ductular reaction, and fibrosis (77). Although dysregulation of bile acid transporters, nuclear receptors, and bile acid detoxification/synthesis enzymes likely contributed to the phenotype, the KO/KD mice also had an impaired regenerative response. The inability to repair the ongoing injury caused by loss of Mdr2 resulted in decompensation that accelerated disease progression (77). Thus, in chronic cholestatic injury, inducing β-catenin signaling may help to increase regeneration, maintain hepatocyte polarity, preserve bile flow, or all three. Notably, β-catenin activation was increased in Mdr2 KO compared to wild-type controls, supporting the hypothesis that β-catenin is performing some compensatory functions in the absence of Mdr2.

Chronic cholestasis can also be caused by junctional defects. One such example is progressive familial intrahepatic cholestasis (PFIC), a heterogeneous group of autosomal recessive disorders that are linked by the inability to appropriately form and excrete bile from hepatocytes. The first described PFIC disorders were linked to dysfunctional proteins within the hepatocyte canalicular membrane (78). More recently, other proteins beyond the canaliculi have been identified whose mutations result in PFIC and neonatal cholestasis, including dysfunction in tight junction integrity (TJP2, USP53) (79, 80).

As mentioned previously, γ-catenin compensates for loss of β-catenin in hepatic epithelial cells to maintain the integrity of adherens junctions (19, 20). Deletion of both catenins from hepatocytes and cholangiocytes led to a PFIC-like phenotype characterized by significant morbidity and mortality, lack of hepatocyte polarity, and cholestasis consisting of increased serum and hepatic bile acids and derangements in serum alkaline phosphatase and bilirubin (81, 82). Deletion of β- and γ-catenin from hepatocytes alone resulted in an overall milder phenotype. Although para-hepatocellular leaks were still present indicating loss of cell-cell junctions, the mice did not exhibit mortality and showed mild ductular reaction and fibrosis.

β-Catenin also regulates the expression of tight junction components including claudin-2 (18). Interruption of Wnt-β-catenin signaling in hepatocytes in any of the genetic knockout mice leads to absence of claudin-2. Claudin 2 global knockout mice show decreased bile flow in the liver which was attributable to the function of this pore-forming claudin in regulating paracellular ion and water flow, which are essential for maintenance of normal bile composition and movement (83).

Hepatocyte-to-biliary reprogramming

It has been known for decades that hepatocytes can acquire a cholangiocyte-like phenotype during cholestatic liver injury (84, 85). This conversion of one type of fully differentiated epithelial cell to another is commonly known as transdifferentiation, although in most cases it might be more correctly termed hepatocyte reprogramming since in the absence of continuous injury hepatocyte-derived cholangiocytes can revert back to their original phenotype (86). Nonetheless, extensive injury to the biliary epithelium, inadequate compensatory cholangiocyte proliferation, and/or cholangiocyte senescence can induce hepatocytes to change their identity in order to replace the injured cells. Cellular plasticity during cholestasis has been demonstrated in both rodent models (86–90) and in human tissue (90–94). Depending on the disease or injury modality the contribution of hepatocytes to biliary repair can be quite extensive (95).

Hepatocytes that are undergoing reprogramming turn on a set of early biliary genes, including SRY-related HMG box transcription factor 9 (Sox9), Yap, and Notch (96, 97) (90, 94, 98–103). These pathways are linked through a complex spatiotemporal interplay of signals to regulate the acquisition of a hepatocyte-like phenotype. In addition to these pathways, Wnt signaling has also emerged as a significant contributor to hepatocyte reprogramming. In PSC, expression of both hepatocyte markers and Wnt ligands are increased in reactive ductules during the later stages of the disease (104). Other studies utilizing in vitro organoid culture systems derived from isolated hepatocytes exposed to a defined culture medium have shown that β-catenin and its downstream targets are upregulated under conditions that promote hepatocyte reprogramming to cholangiocytes, implicating a role for β-catenin in this process (94, 105). Moreover, we have previously shown that expression of Wnt7a, which is increased in the cholangiocyte population during cholestatic liver injury, can induce the expression of biliary markers in cultured hepatocytes in a β-catenin-dependent manner (106). Underscoring the relevance of this finding, transgenic mice expressing a stable form of β-catenin due to mutation at serine-45 in the liver showed increased expression of cholangiocyte markers when subjected to 3,5‐diethoxycarbonyl‐1,4‐dihydrocollidine (DDC) diet, a model for biliary injury and chronic cholestasis. Importantly, this finding was coincident with a decrease in biliary injury, as assessed by serum alkaline phosphatase and bilirubin levels (107). Conversely, mice lacking Wnt secretion from both hepatocytes and cholangiocytes had high mortality and fewer hepatocytes expressing biliary markers in response to DDC (106). Together, these results indicate that β-catenin catalyzes the transdifferentiation of hepatocytes into cholangiocytes, and that induction of a biliary phenotype through activation of β-catenin may alleviate some of the complications of cholangiopathies associated with bile stasis.

There is also evidence that canonical Wnt signaling is not required for hepatocyte-to-cholangiocyte reprogramming. Although RSPO is required for formation of cholangiocyte organoids in vitro, mice with deletion of LGR4/5 from both hepatocytes and cholangiocytes have a similar ductular reaction to littermate controls after DDC, with similar numbers of Sox9-positive hepatocytes as well (108). Additionally, Wnt/β-catenin reporter mice show that Axin2-positive periportal hepatocytes did not express Sox9 during recovery from DDC, indicating that Wnt/β-catenin signaling does not promote hepatocyte-to-cholangiocyte dedifferentiation (108). Instead, LGR4/5-dependent Wnt/β-catenin signaling promotes hepatocyte-mediated regeneration during recovery. This does not preclude the possibility that non-canonical Wnt signaling, rather than Wnt/β-catenin signaling, may drive hepatocyte reprogramming, or that β-catenin may be activated through non-canonical pathways such as receptor tyrosine kinases. Further dissection of this process will be needed to provide insight into the contribution of these pathway components.

In addition to playing a role in hepatocyte plasticity, Wnts can also facilitate repair during cholestasis by inducing cholangiocyte proliferation and/or biliary specification (109, 110). Several groups have shown that Wnts come from the bile ducts during cholestasis (106, 108, 111–113), and that one of these Wnts, Wnt7b, induces cholangiocyte proliferation in an autocrine manner (106, 114). Although this was initially thought to be a reparative process, a subsequent study showed that overexpressing Wnt7b in cholangiocytes increased both cell proliferation and expression of proinflammatory markers. Furthermore, deletion of Wnt7b from both cholangiocytes and hepatocytes improved biliary injury and decreased inflammation by promoting hepatocyte-to-cholangiocyte reprogramming, which compensated for the inability of cholangiocytes to proliferate due to the lack of Wnt7b (115). Although Wnt ligands play an important role in biliary regeneration, several groups have shown that this proliferation is β-catenin independent, thus indicating that canonical Wnt signaling is dispensable for expansion of the biliary compartment during cholestasis (50, 108, 112, 116–118).

Another potential source of Wnts during cholestasis are the inflammatory cells surrounding the ducts. In PSC, ductular proliferation is associated with sustained expression of Wnt ligands in inflammatory cells (109). Wntless was deleted from myeloid cells in order to determine the role of these Wnts. In the DDC model, loss of Wnt secretion from macrophages led to increased inflammation and injury (119). In other models of chronic liver disease, deletion of Wnts from myeloid cells exacerbated liver fibrosis and enhanced ductular reaction, indictating that macrophage-derived Wnts have a reparative or anti-fibrotic function (120). Conversely, a recent study showed that non-canonical Wnts arising from macrophages promote biliary scar deposition by activating the Wnt-PCP pathway in cholangiocytes. Importantly, therapeutic inhibition of the Wnt signaling pathway during cholestasis reduces biliary injury without affecting cholangiocyte number, suggesting that it is possible to uncouple regeneration from fibrosis in biliary disease (116). Future studies exploring the role of Wnt/β-catenin signaling in cholestasis should address the cell-molecule circuitry and the mechanisms that orchestrate this process, as has been done for liver regeneration (14) (Figure 3).

Figure 3: The role of Wnt-β-catenin in cholangiopathies.

Wnts are secreted by cholangiocytes during cholestatic liver injury. These Wnts can act on neighboring cholangiocytes in an autocrine manner and induce proliferation. Wnts can also induce hepatocyte-to-biliary reprogramming through nuclear translocation of β-catenin and expression of Sox9. Wnts can stimulate hepatocyte regeneration through activation of β-catenin and transcription of cyclin-D1. As part of adherens junctions, β-catenin maintains the blood-bile barrier and prevents paracellular leakage of bile. Finally, absence of β-catenin permits FXR activation, which regulates bile acid synthesis and transport. Figure was prepared using Biorender.com.

β-Catenin modulation for therapies

The Wnt-β-catenin pathway is required for the process of normal and optimal liver regeneration. Disruption in signaling as shown by studies in various genetic knockout models delays the regeneration process. Interestingly the reverse is true as well. As the examples below illustrate, overactivation or premature activation of the pathway can accelerate the process of regeneration following both surgical resection as well as drug-induced liver injury, especially acetaminophen overdose (49, 64, 121–123).

In an APC heterozygous mutant, where increased β-catenin activity is evident due to loss of this important component of the β-catenin degradome, accelerated regeneration was evident after hepatectomy (121). Another transgenic mouse expressing a stable form of β-catenin (S45D mutant) also showed a significant shift of regeneration kinetics to the left as compared to wild-type controls (122). It is important to note that these transgenic mice do not show constitutively active β-catenin in hepatocytes due to adaptation, but upon stimulation, these mice exhibit higher β-catenin activation. This study also demonstrated that ectopic but transient overexpression of Wnt1 in hepatocytes by naked DNA hydrodynamic tail vein injections promoted nuclear cyclin-D1 expression in hepatocytes, and increased the number of PCNA-positive hepatocytes in S-phase as early as 30 hours after hepatectomy (122). Finally, RSPO1 accelerated liver regeneration after hepatectomy, allowing RSPO1-treated mice to regain liver mass more rapidly than control mice (49).

Liver regeneration after acetaminophen overdose plays a critical role in determining the outcome of injury (124). When the S45D mutant mice were exposed to an overdose of acetaminophen, they showed a notable increase in cyclin-D1 and hepatocyte proliferation compared to wild-type counterparts (64). Finally, livers from patients with acetaminophen toxicity requiring transplants showed lack of nuclear β-catenin and reduced PCNA levels, whereas hepatocytes in the livers that spontaneously regenerated had higher nuclear β-catenin along with higher PCNA (123).

All of these studies support the hypothesis that transient β-catenin activation may be useful in specific indications where promoting liver regeneration could benefit patients by reducing morbidity and cutting down hospital. Therapeutic acceleration of liver regrowth through enhancement of β-catenin activity may overcome small for size syndrome and enable more living-donor transplantation and large resections (125). β-catenin activation may also be used to induce regeneration prematurely in an ex vivo perfusion device in order to jump-start regeneration following transplant of the graft (13). Additionally, there may be value in promoting β-catenin responsibly in a subsets of drug-induced liver injury where hepatocyte injury is the major manifestation and regenerative therapies to promote remnant hepatocyte proliferation may be beneficial (13, 14).

However, it will be important to fine-tune Wnt signaling so that metabolic zonation is not impaired. There are two important studies that highlight this consideration. In the case of RSPO1 treatment, metabolic zonation was almost entirely lost as pericentral metabolic genes were expressed throughout the liver and periportal gene expression was diminished (126). In this scenario, because therapeutic Wnt activation may reprogram zone-1 into zone-3 hepatocytes, intermittent or short-term dosing schemes may be required. In another study that used FL6.13, the tailored tetravalent antibody that engages the FZD-LRP6 receptor in the absence of natural Wnt ligands, there was a notable increase in zone-3 gene expression and induced hepatocyte proliferation, but there was no concomitant decrease in the expression of periportal genes (127). Overall, therapeutic activation of the Wnt pathway must not only induce proliferation but also maintain zonation, a delicate balance that may be achieved by intermittent dosing.

The preclinical models that have been tested in the context of β-catenin modulation include uninjured mice (to determine the impact on Wnt signaling and hepatocyte proliferation), rescue of zonation and proliferation in genetic knockout mouse models with interrupted Wnt signaling in the liver, and acute liver injury, including the partial hepatectomy model (which may mimic growth post-living donor transplant in recipients and donors as well as patient hepatectomy) and sublethal acetaminophen overdose model. While extended partial hepatectomy (90% hepatectomy) and lethal acetaminophen overdose models are theoretically important to show benefit of regenerative therapy, these are technically challenging and show very high animal-to-animal variability, thus precluding their use in these applications.

Developing tools to promote Wnt-β-catenin signaling has been an area of ongoing research in many laboratories. Recombinant proteins and conditioned media from Wnt-transfected cells have demonstrated the need for appropriate post-translational modifications essential for biological activity and provided important proof-of-concept data, although use of these treatments have been limited to in vitro studies. Repurposing existing growth factors or small molecules that may crosstalk with the β-catenin pathway is also an interesting strategy. One such hormone is triiodothyronine (T3), which induces cyclin-D1 expression in the liver in a β-catenin-dependent manner through interaction with thyroid hormone receptor β (TRβ) (128, 129). T3 causes significant side effects due to the ubiquitous expression of T3 receptors throughout the body. Since TRβ is mostly expressed in hepatocytes, the selective agonist GC-1 (Sobetirome) was thus assessed for its ability to activate proliferation. Indeed, cyclin-D1 expression was increased and hepatocyte proliferation enhanced in mice with intact β-catenin, while β-catenin knockout mice did not respond to treatment (130). The mechanism of β-catenin activation by GC-1 was through phorphorylation of serine-675 by protein kinase A, and by promotion of Wnt secretion from hepatic endothelial cells. Both T3 and GC-1 were shown to increase both baseline hepatocyte proliferation and accelerate post-hepatectomy proliferation, underscoring the usefulness of these agents for regenerative therapies (129, 130).

A more direct activation of Wnt-β-catenin signaling was achieved using artificial recombinant protein fusions as Wnt surrogates. These first-generation Wnt surrogates were bispecific molecules containing individual protein domains that bound to Frizzled and LRP5/6 receptors, and activate Wnt signaling without the presence of endogenous Wnt proteins (131). However, although these agonists showed in vivo activity, they needed adenoviral-based delivery methods. A more improved version was developed through generation of antibody-based Wnt agonists (132). This strategy used bivalent agonists comprised of one Frizzled and one LRP6 binding domain. Stronger activation was achieved with various formats of multivalent combinations of Frizzled and LRP6 binders. Using this strategy, a phase-I safety clinical trial was initiated for two agents: a Fzd5-targeted Wnt-mimetic bi-specific antibody used to treat moderate to severe ulcerative colitis; and a novel hepatocyte-specific R-spondin mimetic bispecific fusion protein targeting ASGR1 in severe alcoholic hepatitis. Unfortunately, both trials have been temporarily halted as mild to moderate transaminitis was observed in healthy subjects in dose escalation studies.

A new class of tetravalent Wnt agonists have proven to be more efficacious than the bivalent designs. Modular and engineerable Frizzled-LRP agonists (FLAgs) are tetravalent synthetic antibodies that enable selective and robust activation of any FZD receptor, resulting in induction of Wnt signaling both in vitro (intestinal organoids) and in vivo (mice) (133). FL6.13 is a tetravalent antibody that contains two pan-FZD paratopes at the N-termini of the Fc (F^P+P) and one paratope each for the WNT1 and WNT3A binding sites on the LRP6 receptor (L6¹⁺³). Upon inhibition of Wnt secretion in mice gut, administration of FL6.13 rescued the intestinal stem cells in the crypt. In a subsequent study, FL6.13 also successfully rescued metabolic zonation in both endothelial cell specific Wntless and endothelial cell specific Wnt2 and Wnt9b double knockout mice (45). However, in the absence of hepatocyte-LRP5–6, FL6.13 was unable to rescue the expression of zone-3 genes. Administration of this antibody to control mice led to increased cyclin-D1 and hepatocyte proliferation as well as accelerated liver regeneration after hepatectomy. In fact, FL6.13 was also able to rescue liver regeneration in endothelial cell specific Wnt2-9b double knockouts.

To determine the clinical applicability of this treatment, FL6.13 was administered 32 hours after a sublethal dose of acetaminophen. Activation of Wnt signaling led to a significant decrease in serum ALT and necrotic areas which also coincided with increased numbers of cyclin-D1- and Ki-67-positive hepatocytes, indicating increased liver regeneration (45). This is an important proof-of-concept because there is no effective therapy for acetaminophen overdose if treatment is not given before 24 hours. Thus, FL6.13 could be a potential new regenerative therapy in the liver post-acute liver injury. However, the timing of administration has to be carefully optimized, since Wnt is a well-known regulator of several cytochrome P450s like cyp2e1 and cyp1a2, which are involved in xenobiotic metabolism of many drugs. Their untimely or prolonged activation by Wnt agonist could have unfavorable consequences by inadvertently inducing generation of more toxic acetaminophen metabolites. Indeed, administration of FL6.13 at 12 hours post acetaminophen overdose promoted injury rather than repair. Nonetheless, this important observation now offers a new option of intervening with a regenerative therapy when N-Acetyl cysteine, the only approved therapy for acetaminophen toxicity that is most effective when given within the first 8–24 hours of ingestion, is no longer able to benefit the patient (134, 135).

Treatment for many types of cholangiopathies is limited to liver transplantation and palliative care. Because bile stasis is a hallmark of cholangiopathies, reestablishing bile flow may help reduce injury or slow the progression of cholestatic diseases. To that end, generation of de novo healthy ducts through induction of hepatocyte reprogramming may be of significance in promoting repair in diseases such as PSC (136). Indeed, a recent publication has demonstrated that transplanted mouse hepatocytes can build a biliary system in vivo by permanently transdifferentiating into mature cholangiocytes that form functional bile ducts (103).

Because of its role in promoting expression of biliary markers in hepatocytes, activation of β-catenin by clinically feasible means may be highly significant. The small molecule Wnt agonist 2-amino-4-[3,4-(methylenedioxy) benzylamino]-6-(3-methoxyphenyl) pyrimidine (AMBMP) attenuates injury and improve survival in rodent models of liver injury such as ischemia/reperfusion, alcoholic liver disease, and small-for-size transplantation (137–139). Although it has not yet been tested in a model of cholestasis, AMBMP may prove useful in reprogramming hepatocytes to increase the number of de novo bile ducts and alleviate the phenotype. Other modalities, such as the tetravalent antibody FL6.13, should also be tested in preclinical models to determine if β-catenin is a druggable target in cholangiopathies.

Chronic liver injury also creates a continuous regenerative stimulus that could be alleviated by activation of β-catenin. Thyroid hormones and their mimetics, such as GC-1, stimulate hepatocyte proliferation after partial hepatectomy in a β-catenin-dependent manner. GC-1 has been shown to impact lipid metabolism and bile acid metabolism by promoting bile acid synthesis from cholesterol and regulating cholesterol secretion in bile, which would provide an additional benefit in cholestatic disease (140–142). When administered to Mdr2 KO mice, GC-1 transiently decreased biliary injury due to a lack of bile acid export by the hepatocytes. However, GC-1 does not induce hepatocyte proliferation (143). This is likely due to decreased expression of both TRβ isoforms in KO, which commonly occurs during chronic liver disease due to loss of terminal hepatocyte differentiation. Further studies are needed to investigate the efficacy of thyromimetics in cholestatic models like Mdr2 KO after forced hepatocyte re-expression of TRβ.

Conclusions

As technologies like single cell sequencing, single cell spatial transcriptomics, and multi-omics continue to evolve, along with growth in systems and computational algorithms to better interpret these complex and high throughput data, the depth of our understanding of liver health and disease has greatly improved. We are able to apply spatial biology in the context of both liver physiology and pathology at the resolution of an individual cell, which has yielded answers to many important questions and has also allowed us to look at the unanswered questions from the lens of single cells and spatial biology. We are able to understand and apply the concept of metabolic zonation in the context of all hepatic pathologies like never before, which provides unique opportunities to understand the cellular and molecular basis of these diseases. The heterogeneity that exists in the same cell type across a liver lobule gives us an appreciation of division of labor and cellular conveyor belts, characteristics that make liver a highly efficient and effective functional organ. At the same time, these discoveries divulge the diversity in the origin of pathologies that may have been previously categorized as a single disease. Such heterogeneity in molecular and cellular underpinnings of specific liver pathologies can now be appreciated in the areas of non-alcoholic steatohepatitis, hepatocellular carcinoma, and cholangiopathies. Interpreting this information will get harder before it gets better, but will be necessary for translational and therapeutic exploitation. Nowhere is this more evident than in the field of hepatic Wnt-β-catenin signaling, as can be appreciated throughout this review. As a result, modulation of β-catenin for treatment becomes highly contextual, as similar pathologies may respond in an opposing manner depending on etiology, environment, or a host of other factors.

We believe that as we gain further insights into various hepatic pathologies and better understand cell-cell communication, cell neighborhoods within a tissue, and spatiotemporal molecular cues, we will obtain a unique understanding of liver health and disease. This knowledge can then be harnessed to devise highly precise therapies using a new class of treatments which could be delivered to special subsets of cells and at specific times to have the most efficacy and least off-target effects. By building upon past knowledge and future technologies, we believe that rational design of therapeutics that target the Wnt-β-catenin pathway will have a profound impact on the future of hepatology.