**Targeting Wnt Signaling for Gastrointestinal Cancer Therapy: Present and Evolving Views**

**Moon Jong Kim 1,**†**, Yuanjian Huang 1,**† **and Jae-Il Park 1,2,3,\***


Received: 27 October 2020; Accepted: 1 December 2020; Published: 4 December 2020

**Simple Summary:** Therapeutic targeting of Wnt has long been suggested for gastrointestinal (GI) cancer treatment because deregulation of Wnt signaling is associated with GI cancers. However, therapeutic targeting of Wnt is still challenging because of the pleiotropic roles of Wnt signaling in the human body. Thus, targeting strategies of Wnt signaling are continuously evolving. The current flows of targeting Wnt signaling for cancer treatment are focused on increasing the specificity of drugs and combinatory treatment with other cancer drugs that minimize side effects and increase efficacy. Additionally, increased knowledge about the β-catenin paradox has expanded the cases that can be treated with Wnt targeting therapy, not strictly considering Wnt upstream and downstream mutations. Here, we discuss these evolving views of targeting Wnt signaling and describe examples of current clinical trials.

**Abstract:** Wnt signaling governs tissue development, homeostasis, and regeneration. However, aberrant activation of Wnt promotes tumorigenesis. Despite the ongoing efforts to manipulate Wnt signaling, therapeutic targeting of Wnt signaling remains challenging. In this review, we provide an overview of current clinical trials to target Wnt signaling, with a major focus on gastrointestinal cancers. In addition, we discuss the caveats and alternative strategies for therapeutically targeting Wnt signaling for cancer treatment.

**Keywords:** Wnt signaling; β-catenin; cancer; gastrointestinal cancers; therapeutic targeting of Wnt signaling; β-catenin paradox; molecular targeting

#### **1. Introduction**

Evolutionarily conserved Wnt signaling was initially identified in *Drosophila* (Wingless) and the mammalian system (Int-1) [1,2]. Wnt signaling has been extensively studied, revealing its pivotal roles in orchestrating embryonic development, tissue homeostasis, and regeneration [3–5]. Notably, the deregulation of Wnt signaling is associated with many human diseases, including cancers [6]. Therefore, the manipulation of Wnt signaling has gained attention as a means of disease treatment and prevention [7,8].

Although it has been confirmed in in vitro and in vivo cancer studies that targeting Wnt signaling has drastic tumor-suppressing effects, no targeted drugs have been successively advanced to clinical applications to date [7–9]. This is mainly because Wnt signaling plays essential roles in maintaining a broad range of physiological events [3–5]. Therefore, blocking Wnt signaling has detrimental impacts on tissue homeostasis and regeneration. In this review, we discuss current views on therapeutically targeting Wnt signaling and describe related clinical trials in gastrointestinal (GI) cancer.

#### **2. Wnt Signaling**

Wnt signaling is an autocrine and paracrine signal-transducing module that is activated by lipid-modified WNT ligands and their receptors [10,11]. In humans, 19 WNT ligands and 18 receptors and coreceptors have been identified [10,12]. The Wnt ligand–receptor interaction activates a downstream cascade in a β-catenin-dependent or -independent manner [13] (Figure 1).

**Figure 1.** General view of canonical and non-canonical Wnt signaling. The switch of the canonical Wnt/β-catenin signaling pathway depends on the subcellular location of β-catenin. The stability of β-catenin is controlled by the destruction complex, consisting of AXIN, APC, CK1, and GSK3. In the absence of WNT ligands, cytoplasmic β-catenin is first phosphorylated by CK1 at Ser45 residue, followed by GSK3 phosphorylation at the Thr41, Ser37, and Ser33 residues. Next, the phosphorylated motif of β-catenin acts as a docking site for βTrCP, which induces the final ubiquitin-mediated degradation of β-catenin (Wnt off). When WNT ligands bind to Frizzled receptors (FZDs) and low density lipoprotein receptor-related protein co-receptor 5/6 (LRP 5/6), the destruction complex is recruited to the plasma membrane, triggering the translocation of β-catenin into the nucleus and activating its downstream target genes via binding directly to the TCF/LEF transcription factor family (Wnt on). Wnt/PCP signaling involves the triggering of a cascade that contains small GTPases RHOA (transforming protein RhoA) and Ras-related C3 botulinum toxin substrate 1 (RAC1), activating Rho-associated protein kinases (ROCKs) and JUN N-terminal kinases, respectively. Wnt/Ca2<sup>+</sup> signaling involves the activation of phospholipase C, which in turn triggers the release of Ca2<sup>+</sup> from intracellular stores and the activation of effectors such as calcium- or calmodulin-dependent protein kinase II, protein kinase C, and calcineurin (CaN). Next, CaN activates the nuclear factor of activated T cells, activating the transcription of downstream target genes.

β-catenin is an Armadillo repeat protein that is mainly associated with E-cadherin at the inner plasma membrane. The β-catenin level is tightly regulated by the protein destruction complex, which is composed of the axis inhibitor (AXIN1), adenomatous polyposis coli (APC), casein kinase 1 (CK1), glycogen synthase kinase 3 (GSK3), and β-transducin repeat-containing protein (βTrCP) and induces β-catenin degradation through phosphorylation-mediated ubiquitination [11,14–17]. In β-catenin-dependent Wnt signaling (canonical Wnt signaling), the destruction complex is sequestered

upon WNT ligand stimulation and disrupted by the formation of the WNT-receptor-disheveled (DVL) complex [18], resulting in the stabilization and nuclear translocation of β-catenin [19]. Next, nuclear β-catenin interacts with the TCF/LEF transcription factor family (TCF7, LEF1, TCF7L1, and TCF7L2), which recruits coactivators to transactivate downstream target genes [20–23]. β-catenin-independent Wnt signaling (also referred to as non-canonical Wnt signaling) activates downstream modules through the planar cell polarity (Wnt/PCP) pathway or Wnt/Ca2<sup>+</sup> signaling pathway [10] (Figure 1).

In the Wnt/PCP pathway, the binding of WNT-FZDs triggers a cascade involving small GTPases RHOA (transforming protein RhoA) and RAC1 (Ras-related C3 botulinum toxin substrate 1), which in turn activates ROCKs (Rho-associated protein kinases) and JUN-N-terminal kinases, respectively [10,24,25]. It mainly regulates cell polarity, cell motility, and morphogenetic movements [10,24,25]. In the Wnt/Ca2<sup>+</sup> signaling pathway, the binding of WNT-FZDs activates phospholipase C (PLC), which in turn triggers the release of Ca2<sup>+</sup> from intracellular stores and the activation of effectors such as calcium- and calmodulin-dependent protein kinase II (CAMKII), protein kinase C (PKC), and calcineurin (CaN) [10,26]. CaN activates the nuclear factor of activated T cells, which regulates the transcription of the genes that control cell fate and cell migration [10,26]. Although both β-catenin-dependent and -independent Wnt signaling are involved in tumorigenesis, β-catenin-dependent Wnt signaling is relatively well defined in various cancer models. In line with this, current pharmacological trials targeting Wnt signaling have mainly focused on β-catenin-dependent Wnt signaling.

#### **3. Wnt Signaling Alteration in GI Cancers**

Hyperactivation of Wnt signaling is frequently observed in GI cancers, including colorectal cancer (CRC), hepatocellular carcinoma, gastric cancer, and pancreatic cancer. Approximately 90% of CRC demonstrates Wnt signaling-related gene alterations [27]. More than 70% of the genetic alterations in CRC are *APC* mutations [27,28]. Unlike CRC, *APC* mutations are rare in hepatocellular carcinoma. Hepatocellular carcinoma mainly displays*CTNNB1* mutations (20–35%) [29], *AXIN1* mutations (8–15%) [30], and Frizzled-7 (*FZD7*) overexpression (90%) [31]. In addition to mutations in the negative feedback regulator of the FZD receptor, the E3 ubiquitin-protein ligases *ZNRF3* and *RNF43* and their ligands, R-spondins (RSPOs), are frequently observed in pancreatic and gastric cancers [32,33].

#### **4. Therapeutically Targeting Wnt Signaling in GI Cancer**

Targeting Wnt signaling for cancer treatment normalizes the hyperactivated Wnt signaling that promotes cancer progression. For this purpose, many targeting strategies have been evaluated, including the inhibition of Wnt ligands and receptors or coreceptors, restoration of the destructive complex, and inhibition of β-catenin/β-catenin-dependent transcriptional machinery. Although these approaches have not been studied in phase III clinical trials or used clinically, dozens of Wnt-targeting agents are currently being evaluated in phase II clinical trials (Table 1). These important phase II clinical trials include LGK974, genistein, Foxy-5, DKN-01, niclosamide, PRI-724, and chloroquine/hydroxychloroquine.

In the next section, we provide an overview of the known and potential agents that target Wnt signaling, especially for GI cancers; we also describe their mechanisms of action and related clinical trials (Table 2). All potential agents that inhibit Wnt signaling are listed in Table 3. In addition, the molecular targets of representative Wnt inhibitors on WNT signaling are illustrated in Figure 2.


**Table 1.** Agents inhibiting Wnt signaling for GI cancers in phase II clinical trials.


**Table 1.** *Cont.*

**Figure 2.** Wnt targeting agents for the Wnt/β-catenin signaling pathway. Wnt targeting agents for GI cancers mainly focus on the inhibition of the key molecules in Wnt/β-catenin signaling, such as inhibiting WNT ligands (ipafricept, LGK794), inhibiting Wnt receptors/coreceptors (vantictumab, rosmantuzumab), stabilizing the destruction complex (AZ1366, hydroxychloroquine), and inhibiting β-catenin-dependent transcriptional machinery (MSAB, PRI-724).






**Table 2.** *Cont.*









#### **5. Targeting WNT Ligands**

#### *5.1. Inhibiting WNT Ligands*

Ipafricept (OMP-54F28) is a recombinant receptor that is comprised of the cysteine-rich domain of FZD8 fused to the human IgG1 Fc domain; it inhibits Wnt signaling by neutralizing WNT ligands [54]. Three trials evaluated ipafricept and its combination therapies (Table 2). A phase I trial evaluated the best dosage of ipafricept and revealed grade 1–2 adverse events (AEs), including dysgeusia, decreased appetite, fatigue, and muscle spasms [36]. Another phase I trial evaluated ipafricept combined with nab-paclitaxel and gemcitabine in metastatic pancreatic cancer and revealed grade ≥ 3 AEs, including increased aspartate aminotransferase, nausea, maculopapular rash, vomiting, and decreased white blood cells [37].

Secreted frizzled-related proteins (SFRPs) bound directly to WNTs via the cysteine-rich domain, preventing the WNT–FZD interaction [55–57]. SFRPs also form dimers with FZDs via the respective cysteine-rich domain to activate or inhibitWNT3A/β-catenin signaling, depending on their concentration [58]. In the nucleus, SFRPs act as biphasic modulators of β-catenin-mediated transcription, which promotes TCF7L2 recruitment and transactivation of cancer stem cell-related genes by binding to the β-catenin's C-terminus; however, they suppress transcriptional activities by binding to the N-terminus [59]. The phase II trial evaluated genistein, an *SFRP2* silencer inhibitor, in combination with FOLFOX and bevacizumab in metastatic CRC; the study revealed mild AEs, including headaches, nausea, and hot flashes (Table 2) [35]. In addition, Wnt inhibitory factor 1 directly binds to WNTs through the Wnt inhibitory factor domain and prevents WNTs from transducing Wnt signaling [60]. Cerberus also binds to and inhibits WNT8, inhibiting Wnt signaling [61]. However, no agents mimicking Wnt inhibitory factor 1 and Cerberus have been identified.

#### *5.2. Targeting Lipid Modification of WNT Ligands*

The palmitoylation of WNT ligands by the protein-serine O-palmitoleoyltransferase porcupine in the endoplasmic reticulum [62] is essential for the maturation and extracellular secretion of WNT ligands. The palmitoylated WNT ligands bind to Wntless homolog in the Golgi and are ferried to the plasma membrane via secretory exosomes [63]. Porcupine inhibitors (CGX1321, ETC-159, and LGK974 [WNT794]), which suppress Wnt signaling by blocking the secretion of WNT ligands, are currently being evaluated in clinical trials (Table 2). A phase I trial evaluated the best dosage of ETC-159 and revealed well-tolerated AEs, including vomiting, anorexia and fatigue, dysgeusia, and constipation [34]. The lipid modification of WNTs can be enzymatically removed by the palmitoleoyl-protein carboxylesterase NOTUM, thereby inhibiting Wnt signaling [64]. The NOTUM inhibitor, ABC99, is effective in the treatment of benefiting osteopenia and osteoporosis by enhancing Wnt signaling (Table 3) [64,65]. However, no agents have been identified that mimic NOTUM to inhibit GI cancers. Alternatively, metalloprotease TIKI1 (Trabd2a) acts as a protease to cleave eight amino acid residues of WNTs, resulting in oxidized WNT oligomers with minimized receptor binding capability in frogs [66,67]. However, no agents have been identified that mimic the impact of TRABD on Wnt signaling in humans.

#### **6. Targeting Wnt Receptors and Co-Receptors**

#### *6.1. Antibodies against FZDs*

Vantictumab (OMP-18R5) is a monoclonal antibody that binds to FZD 1, 2, 5, 7, and 8 and inhibits Wnt signal transduction [54]. A phase I trial evaluating the best dosage of vantictumab combined with nab-paclitaxel and gemcitabine in metastatic pancreatic cancer was terminated because of the increased risk of bone fracture [39]. Moreover, FZD5 has been identified as a dominant FZD receptor in RNF43-mutant pancreatic cancer cells and may be a therapeutic index [68]. However, no agents targeting FZD5 have been introduced.

#### *6.2. Mimetic Agents Binding to FZDs*

Initially, WNT5A was classified as a non-canonical Wnt family member. It activates Wnt/Ca2<sup>+</sup> signaling by stimulating intracellular Ca2<sup>+</sup> flux in zebrafish and frogs [69–72]. In 2006, Mikels et al. found that WNT5A also activates canonical Wnt signaling via FZD4 and LRP5 [73]. Intriguingly, WNT5A additionally inhibits WNT3A-induced canonical Wnt signaling via FZD2 and tyrosine-protein kinase transmembrane receptor ROR2 [73,74]. Therefore, the function of WNT5A is considered not limited to the field of Wnt signaling and is more dependent on the context of receptors. Foxy-5, a WNT5A peptide mimic, reduces the metastatic capacity of invasive breast cancer via epithelial discoidin domain-containing receptor 1 (DDR1), which decreases the motility and the invasive potential of breast epithelial cells [75–77]. However, whether these mechanisms are also true in GI cancers remains unknown. Foxy-5 is being evaluated in phase I-II clinical trials of metastatic CRC, but no results have been published [38] (Table 2).

#### *6.3. Inhibiting LRP5*/*6*

Given that dickkopf-related protein 1 (DKK1) inhibits Wnt signaling through its direct binding to LRP5/6 [78,79], DKK1 was initially considered a tumor suppressor in the β-catenin-dependent context. Conversely, several studies have shown that DKK1 promotes tumor cell proliferation, metastasis, and angiogenesis, which might be mediated by β-catenin-independent signaling [80–86]. One available explanation is that DKK1 interacts with both glypican4 (GPC4) and the LRP/KREMEN complex to induce the endocytosis of LRP5/6, transforming the biochemical properties of FZDs and their cytoplasmic components from the Wnt/β-catenin pathway to the Wnt/PCP signaling axis [87,88]. This mechanism activating β-catenin-independent signaling and inhibiting β-catenin-dependent signaling was validated in zebrafish and frogs [87,88].

On the basis of the tumorigenic role of DKK1, DKN-01, a DKK1 monoclonal antibody, was developed for cancer therapy. Four trials evaluating DKN-01 and its combination therapies are ongoing (Table 2). A phase I trial assessing DKN-01 combined with paclitaxel in advanced esophageal and gastroesophageal junction cancer revealed that 35% of patients experienced a partial response [40,89]. Another phase I trial of the best dosage of DKN-01 combined with gemcitabine and cisplatin in advanced biliary cancer revealed that 33.3% of patients experienced a partial response [41]. Sclerostin domain-containing protein 1 can activate or inhibit Wnt signaling by mimicking WNT ligands or by competing with WNT8 for binding to LRP6, respectively [90,91]. However, no agents simulating sclerostin domain-containing protein 1 have been identified.

#### *6.4. Accelerating the Degradation of FZD*/*LRP Receptors*

Secreted RSPOs (RSPO1-3) and their receptors, RNF43/ZNRF3, are required to potentiate Wnt signaling in various development and tissue homeostasis contexts [92–94]. In addition, leucine-rich repeat-containing G-protein-coupled receptors (LGRs, LGR4-6) are required for the interaction between RSPOs and their receptors [92]. Without RSPOs and LGRs, RNF43/ZNRF3 induces the internalization and degradation of FZD receptors and negatively regulates Wnt signaling [92,95,96].

A phase I trial evaluated the best dosage of rosmantuzumab (OMP-131R10), a monoclonal antibody against RSPO3, for metastatic CRC; no results have been published (Table 2). BNC101, a monoclonal antibody against LGR5, demonstrated antitumor activity in multiple CRC patient-derived xenografts, but the clinical trial was terminated (Table 2) [97]. Niclosamide, a teniacide in the anthelmintic family, promotes FZD1 endocytosis, inhibiting WNT3A/β-catenin signaling in CRC and osteosarcoma and inducing LRP6 degradation in prostate and breast cancer [98–100]. The NIKOLO trial and NCT02687009 have been evaluating niclosamide in CRC (Table 2). The NIKOLO trial has revealed no drug-related AEs [43].

#### **7. Targeting the Destruction Complex**

#### *7.1. Inhibiting the DVL–FZD Interaction*

In the presence of WNT ligands, DVLs bind to the cytoplasmic domain of FZDs via the PDZ (PSD95, DLG1, and ZO1) domain, which provides a platform for the interaction between the LRP's tail and AXIN to recruit the destruction complex onto the cytoplasmic membrane [101,102]. This process inhibits destruction complex-mediated β-catenin protein degradation [93]. Several inhibitors (compound 3289-8625, FJ9, NSC668036, and peptide Pen-N3) that directly inhibit DVL binding with FZDs are currently being evaluated in preclinical studies (Table 3) [103–106].

#### *7.2. Stabilizing AXIN*

Tankyrase is a member of the poly ADP-ribose polymerase superfamily of proteins which mediates the PARsylation and proteasomal degradation of AXIN [107,108]. Tankyrase inhibitors (AZ1366, G007-LK, G244-LM, IWR-1, JW55, and XAV939) that stabilize AXIN and activate the destruction complex are being evaluated in preclinical studies (Table 3) [109–113]. The E3 ubiquitin-protein ligase SIAH, a potent activator of Wnt signaling, promotes the ubiquitination and proteasomal degradation of AXIN by interacting with a VxP motif in the GSK3-binding domain of AXIN [114]. Ubiquitin carboxyl-terminal hydrolase 7 (USP7), a potent negative regulator of Wnt/β-catenin signaling, promotes the deubiquitination and stabilization of AXIN by interacting with AXIN through its TRAF domain [115]. However, no agents that inhibit SIAH or mimic USP7 have been identified.

#### *7.3. Stabilizing APC*

Transmembrane protein 9 (TMEM9) binds to and facilitates the assembly of vacuolar-type H+-ATPase (v-ATPase), resulting in enhanced vesicular acidification and trafficking for subsequent lysosomal degradation of APC and hyperactivation of Wnt/β-catenin signaling [116]. Conversely, pharmacological targeting of v-ATPase using bafilomycin, concanamycin, hydroxychloroquine, or KM91104 inhibits Wnt/β-catenin signaling and suppresses intestinal tumorigenesis (Table 3) [116]. Twenty trials are currently evaluating v-ATPase inhibitors (Table 2). A phase II trial assessing hydroxychloroquine combined with gemcitabine in unresectable pancreatic cancer revealed no dose-limiting AEs [46]. Another phase II trial revealed an increased overall response rate (38.2 vs. 21.1%; *P* = 0.047) but no survival benefits (hazard ratio, 1.14; 95% CI, 0.76–1.69; *P* = 0.53) when adding hydroxychloroquine to combination therapy with nab-paclitaxel and gemcitabine for advanced pancreatic cancer [50].

#### *7.4. Activating CK1 and GSK3*

CK1 and GSK3 sequentially phosphorylate β-catenin to induce the ubiquitination and proteasomal degradation of β-catenin [16]. Therefore, CK1 and GSK3 activators likely reduce the level of β-catenin that translocates into the nucleus, consequently inactivating Wnt signaling. pyrvinium, a CK1 activator that binds to the C-terminal regulatory domain of its isoform CK1A1, has been introduced, but it has not been evaluated in clinical trials (Table 3) [117]. In addition, no GSK3 activators have been introduced.

#### **8. Targeting** β**-Catenin and** β**-Catenin-Dependent Transcriptional Machinery**

#### *8.1. Promoting* β*-Catenin Degradation*

Methyl 3-[[(4-methylphenyl)sulfonyl]amino] benzoate (MSAB) [12] binds to the Armadillo repeat domain of β-catenin and promotes its degradation [118]. NRX-252114, a protein–protein interaction enhancer, enhances the interaction between β-catenin and its cognate E3 ligase, potentiating the ubiquitination-mediated degradation of β-catenin [119]. No clinical trials have evaluated MSAB and NRX-252114.

#### *8.2. Inhibiting the* β*-Catenin–TCF*/*LEF Complex*

With its increased fold change, nuclear β-catenin replaces the transducin-like enhancer protein corepressor with coactivators by forming the β-catenin–TCF/LEF complex [93,120]. This complex transactivates Wnt target genes through its sequence-specific DNA binding and context-dependent interaction [121]. β-catenin-TCF/LEF complex inhibitors (BC21, iCRT3, and PKF115-584) were introduced in preclinical studies (Table 3) [122–124].

#### *8.3. Manipulating TCF*/*LEF Phosphatases*

TRAF2 and NCK-interacting protein kinase (TNIK) phosphorylates the serine 169 residue of TCF7L1 and the serine 154 residue of TCF7L2, acting as an activating kinase of the β-catenin-TCF/LEF transcriptional complex [125–127]. TNIK inhibitors (KY-05009 and NCB-0846) are being evaluated in preclinical studies [126,128] (Table 3). Serine/threonine-protein kinase NLK phosphorylates the threonine 155 and serine 166 residues of LEF1 and the threonine 178, 189 residues of TCF7L2, triggering their dissociation from DNA and inhibiting Wnt target gene transactivation [129,130]. Homeodomain-interacting protein kinase 2 (HIPK2) phosphorylates LEF1, TCF7L1, and TCF7L2 to dissociate them from DNA, which positively or negatively modulates Wnt/β-catenin signaling [131,132]. However, no agents targeting NLK and HIPK2 have been identified.

#### *8.4. Inhibiting Coactivators*

CREB-binding protein (CREBBP), histone acetyltransferase EP300, pygopus homolog (PYGO), and B-cell CLL/lymphoma 9 protein (BCL9) are coactivators that interact with the β-catenin–TCF/LEF complex [10]. PRI-724 competes with β-catenin to bind with CREBBP, suppressing the transcriptional activation of β-catenin target genes [133]. Three trials have been evaluating PRI-724, two of which were terminated or withdrawn because of low enrollment or a drug supply issue (Table 2). A phase I trial evaluating the best dosage of PRI-724 revealed grade 2 AEs, including diarrhea, bilirubin elevation, hypophosphatemia, nausea, fatigue, anorexia, thrombocytopenia, and alkaline phosphatase elevation [52]. Another phase I trial evaluating the best dosage of PRI-724 combined with gemcitabine as second-line therapy for advanced pancreatic cancer revealed grade ≥ 3 AEs, including abdominal pain, neutropenia, anemia, fatigue, and alkaline phosphatase elevation [53]. The inhibitors of EP300, PYGO, and BCL9 (IQ-1, pyrvinium, and carnosic acid, respectively) have been evaluated in preclinical studies (Table 3) [117,134,135]. In addition, SM08502, a CDC-like kinase (CLK) inhibitor that blocks the phosphorylation of serine/arginine-rich splicing factors and consequently disrupts spliceosome activity, has been shown to inhibit Wnt signaling in preclinical models [136–138]. A phase I trial evaluating SM08502 for advanced GI cancers is ongoing (Table 2).

#### **9. Caveats in Targeting Wnt Signaling**

#### *9.1. Targeting Core Components of Wnt Signaling*

The major caveat in Wnt targeting strategies is their detrimental side effects on normal cells in which Wnt signaling plays pivotal roles in tissue homeostasis and regeneration [3–5]. For example, intestinal stem cells replenish the intestinal epithelium every 3 to 4 days; this is tightly regulated by constitutively active Wnt signaling in the crypt bottom [139,140]. Inhibiting Wnt signaling disrupts intestinal homeostasis and induces the severe loss of the crypt-villi structure. Similarly, upon Wnt blockade, tissue homeostasis disruption also takes place in hair follicles, the stomach, and the hematopoietic system, where Wnt signaling is indispensable for the maintenance of stem cells and their niches [141–143]. Indeed, the treatment of the FZD inhibitor (vanctumab) and antagonist (ipafricept) leads to side effects, including tiredness, diarrhea, vomiting, constipation, bone metabolism disorders, and abdominal pain [36,54]. Wnt signaling is also required for tissue homeostasis and regeneration in the lungs, liver, skin, and pancreas [3–5]. Therefore, Wnt signaling targeting strategies need to be

meticulously designed and evaluated on the basis of their specificity and efficacy, which is discussed in the next section.

#### *9.2. Targeting Upstream vs. Downstream*

Targeting the downstream effectors of Wnt signaling, e.g., β-catenin and TCF/LEF, might maximize Wnt signaling inhibition on the basis of signaling convergence into downstream gene regulation. However, targeting downstream Wnt signaling might also generate severe side effects by disrupting Wnt signaling in normal tissues. Conversely, targeting the upstream molecules of Wnt signaling, e.g., ligands and receptors, was initially considered ineffectual in cancer cells carrying mutations in Wnt signaling downstream (i.e., *APC* and β-catenin/*CTNNB1*) [93]. Intriguingly, accumulating evidence suggests that targeting Wnt signaling upstream is also effective independent of Wnt signaling downstream mutations. This evolving concept, the "β-catenin paradox", is discussed below.

#### **10. Evolving Views in Targeting Wnt Signaling**

#### *10.1. Cancer- and Tissue-Specific Wnt Signaling Targeting*

Targeting cancer type- or tissue-specific Wnt signaling components or modulators may overcome the side effects of Wnt signaling blockade on normal tissues. For instance, specifically targeting the constitutively active form of β-catenin mutants may be ideal. A recent study found that small-molecule enhancers of mutant β-catenin and its E3 ligase (β-TrCP) interaction potentiate the ubiquitination-mediated degradation of mutant β-catenin [119], suggesting one possible approach to targeting the mutant form of β-catenin.

There are also several promising preclinical and clinical studies evaluating antibodies against RSPOs and LGRs, Wnt signaling amplifiers [42]. Since RSPOs and LGRs are differently expressed in different tissues and cancers [144,145], targeting them might diminish normal tissue damage. LGR5 has been suggested as a cancer stem cell marker [146,147], and targeting LGR5+ cells with anti-LGR5 antibody–drug conjugates suppressed tumor growth and metastasis in a preclinical model [145,148]. Anti-LGR5 therapy and anti-RSPO3 (rosmantuzumab) are currently being evaluated in phase I trials for the treatment of metastatic CRC (NCT02726334 and NCT02005315) (Table 2). RSPO3-LGR4-maintained Wnt signaling is essential for the stemness of acute myeloid leukemia, and the clinical-grade anti-RSPO3 antibody eradicated leukemia stem cells [149], which might be effective in GI cancer. The results of these studies indicate that blockage of cancer- or tissue-specific Wnt signaling components or regulators are viable options for GI cancer treatment.

#### *10.2. E*ffi*cacy and Combination Therapy*

An alternative method of overcoming limitations in Wnt signaling targeting strategies is to identify a safe dose that is highly effective but does not disrupt normal physiologic processes. A specific dose of LGK794 had lower severity of side effects with effective pharmacologic outcomes in a phase I clinical trial [7]. It is also noteworthy that different tissues showed different levels of Wnt signaling threshold in vivo [150], supporting the theory that localizing treatment is an alternative strategy to avoid toxicity and side effects.

In general, combination therapy is considered to result in more AEs. However, it does not always induce more AEs than does monotherapy. The incidences and degrees of AEs depend on various factors, such as the doses of single drugs, the timing of administration, the period of treatment, the supportive treatment, and the heterogeneity of the patients themselves. Thus, certain drug dose combinations may be more effective, with fewer AEs. Furthermore, monotherapy targeting one pathway does not guarantee complete anticancer activity because of multiple crosstalks and compensations by other signaling pathways. Although its efficacy may be counterbalanced by correspondingly increased toxicity, combination therapy that simultaneously targets several pathways might be more efficient. In addition, combination therapy is the most common approach to achieving survival benefits in clinical practice, and most promising phase III Wnt targeting trials use combination therapy.

ICG-001 and PRI-724 inhibit Wnt target gene expression by antagonizing CBP, a β-catenin coactivator [133,151]. PRI-724 was effective in a phase I clinical trial of PDAC when used in combination with gemcitabine (NCT01764477). Other cases include the combination of anti-FZD antibody with chemotherapy. Vantictumab (OMP-18R5) resulted in promising outcomes in the preclinical setting [152,153] and is currently being evaluated in phase I clinical trials for multiple cancers in combination with paclitaxel [154]. Ipafricept (OMP-54F28/FZD8-Fc) is being evaluated in a phase I clinical trial to treat advanced pancreatic cancer in combination with nab-paclitaxel and gemcitabine [36]. Although antibodies against pan-Wnts or pan-FZD were not tissue-specific, their combination in advanced solid tumors had promising effects [36,154]. In addition, as a neoadjuvant therapy, Foxy-5 is currently being evaluated in phase II trials for colon cancer, as described above (NCT03883802).

#### *10.3.* β*-Catenin Paradox*

The β-catenin paradox was introduced on the basis of heterogeneous Wnt signaling activity in CRC cells, carrying homogenous genetic alterations in *APC* or β-catenin/*CTNNB1* [155]. This observation was followed by discoveries of several Wnt signaling regulators and multiple crosstalks of Wnt/β-catenin signaling with MAPK and PI3K pathways [156–165]. Additionally, accumulating evidence suggests that the blockade of Wnt signaling upstream molecules suppresses tumor growth despite the presence of oncogenic mutations in Wnt signaling components [96,108,116,152,166,167], demonstrating the existence of additional regulatory modules in Wnt signaling, independent of genetic alterations. Additionally, truncated mutant APC remains partially functional to induce β-catenin protein degradation [116,167]. Moreover, the blockade of WNTs/RSPOs inhibits the growth of tumor cells that harbor *APC* mutations [96,116]. In line with this, Tankyrase inhibitor-stabilized AXIN protein suppresses the proliferation of CRC cells that carry constitutively active mutations in β-catenin or *APC* [108,110]. A recent gastric cancer mouse model study also revealed that vantictumab, the pan-FZD inhibitor, inhibits gastric adenoma growth independently of *APC* mutations [152]. Therefore, molecular targeting of the upstream molecules of APC and β-catenin might be promising in Wnt/β-catenin signaling-associated cancer.

#### *10.4. Generalization of Wnt Targeting Therapy*

Aberrant Wnt signaling is crucial for the potential clonal source of tumor cells and is considered an environmental and metastatic niche for tumor progression. Indeed, LGR5+ colon cancer cells are required for the formation of metastatic colonization in the liver [146]. A study using patient-derived pancreatic organoids revealed differing Wnt-niche dependency among organoids [168]. Furthermore, in a recent study of lung cancers that barely harbor Wnt mutations, Wnt signaling was shown to be required for lung cancer progression as a niche factor in a mouse lung adenocarcinoma model [169]. In that context, Wnt targeting by porcupine inhibitor, WNT794 (LGK794), revealed the suppression of lung tumor progression [169]. These results suggest that Wnt targeting therapy can be generalized to various types of non-Wnt-mutated cancers in which Wnt signaling has tumor-promoting or metastatic roles.

#### **11. New Candidates for Targeting Wnt Signaling in GI Cancers**

Several cancer-specific Wnt signaling regulators were identified in GI cancers. Amplification of USP21 deubiquitinase promotes pancreatic cancer cell growth and stemness via Wnt/β-catenin signaling [170]. RNF6, a CRC-upregulated E3 ligase, promotes CRC cell growth through the degradation of Tele3, a transcriptional repressor of the β-catenin/TCF4 complex [171]. Another deubiquitinase USP7 serves as a tumor-specific Wnt activator in *APC*-mutated CRC by promoting β-catenin deubiquitination [172]. Transcriptional coactivators of β-catenin, BCL9 and BCL9l, redundantly demonstrated CRC-specific upregulation, and their loss suppressed intestinal tumorigenesis in a mouse model [173]. BCL9 and BCL9l inhibitors were recently developed [135,174,175]. Targeting BCL9 and BCL9l has been suggested as a therapeutic approach to CRC-specific treatment. FZD5 mainly expressed in RNF43 mutated tumor cells was proposed as a molecular target for pancreatic cancer treatment [68]. Given that gut-specific knockout of *FZD5* is feasible in the mouse models [176,177], it is likely that targeting of FZD5 can be used in RNF43 mutated intestinal or gastric tumors. In addition, CRC-upregulated PAF/KIAA0101 hyperactivates Wnt/β-catenin signaling and accelerates tumorigenesis in vitro and in vivo [178,179]. As an amplifier of Wnt signaling, TMEM9 hyperactivates β-catenin via APC degradation to promote intestinal and hepatic tumorigenesis [116,166]. Of note, germline deletion of *Tmem9* or *Paf* did not display any discernible phenotypes, suggests that blockade of cancer-related Wnt signaling activators or amplifiers minimizes side effects in Wnt signaling targeting approaches.

Additionally, recent technological advances in organoids made it feasible to perform high-throughput chemical screening (clinical drugs or drug library) and genetic screening (gene knock-out or knock-down) of tumor organoids [180–182]. Moreover, patient-derived organoids become valuable resources to identify most effective drug(s) for precision medicine including pharmacogenomics [183–185]. Therefore, with the emergence of such new technology, it is anticipated that novel tumor-specific and druggable vulnerabilities related to Wnt signaling hyperactivation will be identified.

#### **12. Conclusions**

To date, many studies have reported the marked impact of molecular targeting of Wnt signaling on tumor suppression in preclinical settings. Despite the ongoing clinical trials, it is still imperative to overcome recurring pitfalls—catastrophic adverse effects on tissue homeostasis and regeneration. Like the sword of Damocles, targeting Wnt signaling poses a high risk but has significant potential in cancer therapy. With evolving concepts in Wnt signaling deregulation and manipulation, new and improved approaches, including molecular targeting of upstream signaling modules or cancer-specific regulators and combination therapy, are expected to open a new window of opportunity in the treatment of Wnt signaling-associated cancer.

**Author Contributions:** M.J.K., Y.H., and J.-I.P. wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by grants to the Cancer Prevention and Research Institute of Texas (RP140563 and RP200315 to J.-I.P.), the National Institutes of Health (2R01 CA193297 to J.-I.P.), the Department of Defense Peer Reviewed Cancer Research Program (CA140572 to J.-I.P.), an Institutional Research Grant (MD Anderson to J.-I.P.), a Specialized Program of Research Excellence (SPORE) grant in endometrial cancer (P50 CA83639), and an ROSI Seed Award (00057597 to M.J.K.).

**Acknowledgments:** We apologize for all of the studies we were not able to cite because of space limitations.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **p300 Serine 89: A Critical Signaling Integrator and Its Effects on Intestinal Homeostasis and Repair**

**Keane K. Y. Lai 1,2,†, Xiaohui Hu 1,†, Keisuke Chosa 1,†, Cu Nguyen 1, David P. Lin 1, Keith K. Lai 3, Nobuo Kato 4, Yusuke Higuchi 1, Sarah K. Highlander 5, Elizabeth Melendez 1, Yoshihiro Eriguchi 6, Patrick T. Fueger 2,7, Andre J. Ouellette 6,8, Nyam-Osor Chimge 1, Masaya Ono <sup>9</sup> and Michael Kahn 1,2,8,10,\***


**Simple Summary:** Given their high degree of identity and even greater similarity at the amino acid level, Kat3 coactivators, CBP (Kat3A) and p300 (Kat3B), have long been considered redundant. We describe the generation of novel p300 S89A knock-in mice carrying a single site directed amino acid mutation in p300, changing the highly evolutionarily conserved serine 89 to alanine, thus enhancing Wnt/CBP/catenin signaling (at the expense of Wnt/p300/catenin signaling). p300 S89A knock-in mice exhibit multiple organ system, immunologic and metabolic differences, compared with their wild type counterparts. In particular, these p300 S89A knock-in mice are highly sensitive to intestinal injury resulting in colitis which is known to significantly predispose to colorectal cancer. Our results highlight the critical role of this region in p300 as a signaling nexus and provide further evidence that p300 and CBP are non-redundant, playing definite and distinctive roles in development and disease.

**Abstract:** Differential usage of Kat3 coactivators, CBP and p300, by β-catenin is a fundamental regulatory mechanism in stem cell maintenance and initiation of differentiation and repair. Based upon our earlier pharmacologic studies, p300 serine 89 (S89) is critical for controlling differential coactivator usage by β-catenin via post-translational phosphorylation in stem/progenitor populations, and appears to be a target for a number of kinase cascades. To further investigate mechanisms of signal integration effected by this domain, we generated p300 S89A knock-in mice. We show that S89A mice are extremely sensitive to intestinal insult resulting in colitis, which is known to significantly increase the risk of developing colorectal cancer. We demonstrate cell intrinsic differences, and microbiome compositional differences and differential immune responses, in intestine of S89A versus wild type mice. Genomic and proteomic analyses reveal pathway differences, including lipid metabolism, oxidative stress response, mitochondrial function and oxidative phosphorylation. The diverse effects

**Citation:** Lai, K.K.Y.; Hu, X.; Chosa, K.; Nguyen, C.; Lin, D.P.; Lai, K.K.; Kato, N.; Higuchi, Y.; Highlander, S.K.; Melendez, E.; et al. p300 Serine 89: A Critical Signaling Integrator and Its Effects on Intestinal Homeostasis and Repair. *Cancers* **2021**, *13*, 1288. https://doi.org/ 10.3390/cancers13061288

Academic Editor: Daniel Louvard

Received: 3 February 2021 Accepted: 10 March 2021 Published: 14 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

on fundamental processes including epithelial differentiation, metabolism, immune response and microbiome colonization, all brought about by a single amino acid modification S89A, highlights the critical role of this region in p300 as a signaling nexus and the rationale for conservation of this residue and surrounding region for hundreds of million years of vertebrate evolution.

**Keywords:** CBP; p300; IBD; colitis; colorectal cancer; Wnt

#### **1. Introduction**

The vertebrate radiation, which was initiated approximately 450 million years ago, ushered in a major lifestyle change with a significant increase in adult lifespan [1,2] and with it, a requirement for high-fidelity, long-term homeostasis [2]. This change necessitated that somatic stem cells (SSC), in their respective niches, remain quiescent, in contrast to their differentiated daughter cells, which rapidly proliferate, in order to safeguard the integrity of the SSC's genetic material [1,3]. The gene duplication of the Kat3 coactivator family, which led to the evolution of Kat3A/CREBBP (cAMP response element binding protein (CREB)-binding protein) (CBP) and its closely related paralog Kat3B/E1A-binding protein, 300 kDa (p300), apparently occurred just prior to radiation of the vertebrate lineage [1]. The two Kat3 coactivators encode massive proteins of ~300 kDa over 33 and 31 exons, respectively [1]. CBP and p300 have maintained an extraordinarily high degree of identity—as high as 93%—and an even higher degree of similarity, particularly over an extensive central core region, which encompasses the CH1, KIX, bromodomain, CH2 and CH3 domains (Figure 1A) [1,2,4,5]. They both interact with a myriad of proteins, given their key roles in orchestrating transcription [1]. Due to their high degree of identity and even greater similarity at the amino acid level, CBP and p300 have long been considered redundant. However, mounting evidence clearly demonstrates that they are non-redundant, playing definite and distinctive roles in development and disease [1,6–10]. β-catenin, a key transcriptional component in Wnt signaling, must recruit CBP or p300 in addition to other components of the core transcriptional complex to initiate functional Wnt transcription [1,11,12]. The extreme N-terminal regions of CBP and p300, containing the lowest homology with approximately 66% identity, have been the focus of our interest. β-catenin and specific, direct small molecule CBP/catenin antagonists (ICG-001/PRI-724) [1,9,13,14] and direct small molecule p300/catenin antagonists (YH249/250) [15], competitively bind within this extreme N-terminal region (Figure 1A) [1]. This highly unstructured region of the Kat3 coactivators serves as a nexus for integrating the interactions of varied signal transduction pathways (e.g., nuclear receptor family, RAR/RXR, vitamin D, and Interferon STAT1/2) with the Wnt/catenin signaling cascade [1,9,16–19]. We originally identified p300 serine 89 as a critical residue (Figure 1B) controlling differential coactivator usage by β-catenin via post-translational phosphorylation in mouse embryonic stem cells [20]. p300 serine 89 appears to be a target for a number of kinase cascades including PKC [21,22], AMPK [23], and SIK2 [24], associated with an array of biological effects, including activation and inhibition of transcription [23,25], inhibition of histone acetyltransferase function [22], regulation of insulin/glucagon signaling [24], and differentiation of mES cells [20] and adult progenitor cells [26].

To further investigate the role that p300 serine 89 (S89) plays in vivo, we have generated p300 S89A knock-in mice. p300 S89A knock-in mice, albeit born at sub-Mendelian ratios, appear to be relatively normal. Nevertheless, these mice exhibit multiple organ system, immunologic and metabolic differences, compared with their wild type counterparts. We now initially report on the generation of these mice and their high sensitivity to intestinal injury, which is apparently related to a complex interplay between aberrant epithelial differentiation, gut immunity and changes in their intestinal microbiota and metabolites and which results in colitis, a significant risk factor predisposing to colorectal cancer [27,28].

**Figure 1.** Differential usage of homologous Kat3 coactivators CBP and p300 by β-catenin. (**A**) Schematic representation displaying identity between CBP and p300. CBP and p300 have molecular weights of approximately 300 kDa and are encoded over 33 and 31 exons and consist of 2441 and 2414 amino acids (a.a.), respectively. β-catenin competes with direct small molecule CBP/catenin antagonists (PRI-724/ICG-001) for binding to CBP's (but not p300- s) distal N-terminus, the least conserved region within these two Kat3 coactivators. CBP, cAMP response element binding protein (CREB)-binding protein; p300, E1A-binding protein, 300 kDa; RID, receptor-interacting domain; CH, cysteine/histidine region; KIX, kinase-inducible domain interacting domain; BD, bromodomain; SID, steroid receptor co-activator-1 interaction domain; QP, glutamine- and proline-rich domain. (**B**) Sequence alignment of the distal N-terminal regions of CBP and p300, showing conserved sites for binding of β-catenin (DELI motif). Note: p300 S89 is a critical residue controlling differential Kat3 coactivator usage by β-catenin. (Human CBP and p300 sequences are depicted.)

#### **2. Materials and Methods**

#### *2.1. Mice*

Animal studies were approved by the University of Southern California Institutional Animal Care and Use Committee (IACUC) as per protocol #11023. The S89A knock-in point mutation in exon 2 of the mouse p300 gene, via site-specific mutagenesis, was generated using the flip-excision (FLEx) switch construct. This mutation removes the highly conserved phosphorylation site at S89. The construct included five segments: the 5 homology arm, a point-mutated exon 2 in inverted orientation, a PGK-Neo selection cassette, the wild type exon 2, and the 3 homology arm. The design of the construct, cloning of the targeting vector, electroporation into mouse ES cells, screening of the 129 ES cells, injection into blastocysts, and screening of the chimeric mice were performed by Ozgene Pty Ltd. (Australia). In principle, transcription from the mutant exon 2 should have been activated via Cre recombinase in two steps. First, Cre recombinase would invert the mutated fragment flanked by the *loxP* site to correct the orientation to activate, and then excise the wild type fragment flanked by lox2272 to inactivate it. However, we found that mice homozygous for the knock-in construct displayed early embryonic stage lethality, similar to that of p300 knockout mice [29], suggesting that the wild type p300

protein was not being produced properly from the wild type fragment in vivo for unknown reasons. This malfunction of the wild type fragment caused us to revise our original plan of conditional mutagenesis, resulting in our decision to generate p300 S89A germ line mice. Mice were backcrossed onto C57BL/6 background (The Jackson Laboratory, Bar Harbor, ME, USA) for at least 10 generations before used for experiments. Hematology testing on mouse blood samples was performed on the Hemavet (Drew Scientific, Miami Lakes, FL, USA), and clinical chemistry testing was performed by Antech Diagnostics.

#### *2.2. Isolation of Crypt Cells from Ileum*

Crypt cells from ileum were isolated based on a previously described protocol [30] with minor modification. Ileum was dissected out, and lumen of the intestine was flushed with ice-cold PBS. Intestine was opened longitudinally and placed in tube containing icecold PBS. Tube was inverted 10–15 times, and then PBS removed and replaced with fresh ice-cold PBS. Washing with fresh ice-cold PBS was repeated until the supernatant no longer contained any visible debris. Intestine was cut into 5 mm pieces and placed into ice-cold 5 mM EDTA-PBS. Fragments of intestine were vigorously triturated by pipetting up and down 15 times, and then allowed to settle by gravity for 30 s. Supernatant was aspirated, and then 5 mM EDTA-PBS was added to the intestinal fragments and re-suspended intestinal fragments were placed at 4 ◦C on a benchtop roller for 10 min, after which supernatant was aspirated, and then intestinal fragments kept. 5 mM EDTA-PBS was added to the intestinal fragments and then placed at 4 ◦C on a benchtop roller for 30 min. Supernatant was aspirated and then ice-cold PBS was added to wash the crypts and then supernatant was aspirated. Ice-cold PBS was added, and the intestinal fragments were vigorously triturated 10 times. Supernatant fractions were collected and then mixed 1:1 with solution of basal media containing DNase I. (Final concentration of mixture: ~15 U/mL DNase I.) Mixture was first filtered through a 100 μm filter into a BSA (1%) coated conical tube, and then filtered through a 70 μm filter into a BSA (1%) coated tube, after which the filtrate was spun at 300× *g* in a tabletop centrifuge for 5 min. Supernatant was aspirated, and the cell pellet was re-suspended in basal media containing 5% FBS and then centrifuged at 100× *g* for 5 min, after which supernatant was removed and samples were frozen at −80 ◦C until further analysis.

#### *2.3. Co-Immunoprecipitation*

150–200 mg of intestinal crypt cells were re-suspended in CERI buffer (NE-PER, ThermoFisher, cat. #: 78833, Waltham, MA, USA) containing 5 mM DTT and 1X protease inhibitor cocktail, using a dounce homogenizer. After re-suspension in CERI buffer, the procedure for nuclear extraction was performed based on manufacturer protocol. Protein concentration of nuclear extract was performed using the Protein Assay Dye Reagent (Bio-Rad, cat. #: 500-0006, Hercules, CA, USA). 100–500 μg of nuclear protein was diluted in CoIP buffer (25 mM Tris-HCl, pH 8.0, 1% NP40, 5% glycerol, 150 mM NaCl, 1 mM EDTA, 5 mM DTT, 1X protease inhibitor cocktail (Calbiochem, cat. #: 539137, Burlington, MA, USA)) to a final volume of 1000 μL. 2 μg of CBP (Aviva Biosystems, cat. #: ARP43609\_P050, San Diego, CA, USA), p300 (Aviva Biosystems, cat. #: OAAF01891-100UG, San Diego, CA, USA), 14-3-3-*ε* (Abcam, cat. #: ab43057, Cambridge, MA, USA), or normal IgG (Aviva Biosystems, cat. #: OAEF01185-1MG, San Diego, CA, USA) antibody was added and mixture incubated overnight at 4 ◦C on a tube shaker/rotator. 20 μL of Dynabeads Protein A (ThermoFisher, cat. #: 10001D, Waltham, MA, USA) was added and mixture incubated for 1 h at 4 ◦C on tube shaker/rotator. The magnetic beads were washed three times with 500 μL of CoIP Buffer each time, using the magnetic stand. 20 μL of 2× Laemmli buffer was added and mixture vortexed. Beads were boiled in 2× Laemmli buffer for 10 min. Supernatant containing proteins were separated from the magnetic beads, using the magnetic stand. Protein samples were subjected to electrophoresis on a 4–20% Teo-Tricine SDS-PAGE gel (VWR, cat. #: 71003-072, Radnor, PA, USA). Proteins were transferred onto PVDF membrane. Proteins of interest on the PVDF membrane were detected by

incubating with β-catenin antibody (Santa Cruz Biotechnology, cat. #: sc-7199, Dallas, TX, USA) or p300 antibody (Aviva Biosystems, cat. #: OAAF01891-100UG, San Diego, CA, USA) as the primary antibody, and subsequent incubation with CleanBlot (ThermoFisher, cat. #: 21232, Waltham, MA, USA) as the secondary antibody, followed by application of chemiluminescent reagent ECL Plus (GE Healthcare, cat. #: RPN2132, Chicago, IL, USA) and imaging on the ChemiDoc Imaging System (Bio-Rad), after which relative protein concentration was determined by densitometry.

#### *2.4. Western Blotting*

Flash-frozen mouse ileum was resuspended in RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP-40, 0.5% deoxycholate, 5 mM EDTA, 0.1% SDS), containing protease inhibitors (Roche, cat. #: 11836170001, South San Francisco, CA, USA). Tissue was homogenized and then centrifuged at 12,000× *g* for 15 min, after which supernatants were collected. Protein concentration was determined using the Protein Assay Dye Reagent (Bio-Rad, cat. #: 500-0006, Hercules, CA, USA). 40 μg protein mixed with 4× Laemmli sample buffer were incubated at 37 ◦C for 15 min and subjected to SDS-PAGE. After overnight transfer onto PVDF membrane (Bio-Rad, cat. #: 1620177, Hercules, CA, USA), membrane was incubated overnight with primary antibody at 4 ◦C. Membrane was washed and then incubated with secondary antibody for 1 h, followed by application of chemiluminescent reagent ECL (GE Healthcare, cat. #: RPN2232, Chicago, IL, USA) and imaging on the ChemiDoc Imaging System (Bio-Rad), after which relative protein concentration was determined by densitometry. The primary antibodies used were DUOX2 (Santa Cruz, cat. #: sc-398681, Dallas, TX, USA), GAPDH (Santa Cruz, cat. #: sc-32233, Dallas, TX, USA). The secondary antibody was mouse IgG kappa binding protein conjugated to horseradish peroxidase (Santa Cruz, cat. #: sc-516102, Dallas, TX, USA).

#### *2.5. RT-qPCR and PCR*

Total mRNA was extracted by TRIzol reagent (Invitrogen Carlsbad, CA, USA) according to the manufacturer's protocol. cDNA was synthesized using qScript cDNA Synthesis Kit (Quantabio, Beverly, MA, USA). cDNA template was used for qPCR with SYBR Green detection method utilizing the CFX Connect Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). The PCR primer sequences used for mouse cells were as follows: Duox2 (F: 5- -CATTGCCACCTACCAGAACATTG-3- , R: 5- -AGATGCTGGGGTCCATG AAAG-3- ); Duoxa2 (F: 5- -TAACATTACACTCCGAGGAACAC-3- , R: 5- -AGTCCCTTCTC CAAGGCATG-3- ). Housekeeping gene PCR primer sequences for mouse Gusb were (F: 5- -TATGGAGCAGACGCAATCCC-3- , R: 5- -TTCGTCATGAAGTCGGCGAA-3- ). Relative expression levels for genes of interest were calculated using the 2−ddCt method. PCR primer sequences used for OTU217 were as follows: F: 5- -TACCGCATAAGCCTGCTGTG-3- , R: 5- -ATCGTTGTCTTGGTAGGCCG-3- . PCR products were resolved by agarose gel electrophoresis.

#### *2.6. 16S rRNA Gene Sequencing of Microbiota*

Stool samples from mice were collected and genomic DNA was extracted using the QIAAmp Fast Stool Mini Kit (Qiagen, Redwood City, CA, USA) as per manufacturer instructions. DNA samples were sent to Research and Testing Laboratory (currently RTL Genomics, Lubbock, TX, USA) for processing and analysis based on a previously described protocol [31,32]. 16S ribosomal RNA variable region was amplified and subjected to sequencing on an Illumina MiSeq as previously described [31]. Reads were processed and classified into operational taxonomic units (OTUs) as previously described [32]. Bacterial diversity between groups of mice were compared using the Shannon index (for alpha diversity) and the Bray–Curtis and Jaccard indices (for beta diversity).

#### *2.7. Intestinal Organoids*

Organoids were derived from mouse intestines and cultured based on a previously described protocol [33]. Briefly, mouse small intestine was isolated and opened longitudinally. After washing with ice-cold PBS, villi were scraped off using a coverslip. Intestine was then cut into fragments from which crypts were isolated. Crypts were resuspended with Matrigel, and cell suspension plated and cultured to form organoids.

#### *2.8. Genomic Analysis*

RNA-seq was performed on mouse intestine and intestinal organoids at the USC Core Lab, based on a previously described protocol [34]. Briefly, the Ovation RNA-Seq System V2 and Ovation Ultralow Library System V2 (NuGEN Technologies, Inc., San Carlos, CA, USA) were used for amplification of total RNA and library preparation. RNA-seq was performed on an Illumina HiSeq 2000 (Illumina, San Diego, CA, USA) using pairedend sequencing. Sequence data were analyzed using Partek Flow software (Partek Inc., Chesterfield, MO, USA). Differentially expressed gene lists were created, and differences were considered significant if false discovery rate (FDR) adjusted *p*-value, i.e., "q-value" < 0.05 or multimodel *p*-value < 0.05. Pathway analysis was performed using Ingenuity Pathway Analysis (IPA) (Qiagen).

#### *2.9. Proteomic Analysis*

Samples were prepared and subjected to proteomic analysis by 2DICAL as previously described [9]. Methanol solutions of whole cell extracts were dried and processed for trypsinization. After trypsinization, the obtained peptides were resolved and then quantified. Peptide solution was desalted, dried, and re-dissolved. The obtained peptide solution was subjected to nanoLC-Ultra 2D (AB SCIEX) coupled with to a TripleTOF5600 (AB SCIEX) mass spectrometer. The subjected peptides were directly injected onto a C18, non-end capping, ULTRON HF-ODS(N) (0.1mm I.D., 700 mm length, Shinwa Chemical Industries Ltd., Kyoto, Japan) and then separated by a binary gradient. The masses of the eluted peptides were determined using the TripleTOF5600. MS peaks were detected and quantified using 2DICAL. 2DICAL was developed as a shotgun proteomics analysis system. It analyzes the data of mass to charge ratio (m/z), retention time (RT) and peak intensity generated by liquid chromatography and mass spectrometer (LC/MS), and each sample as elemental data; it deploys various 2-dimensional images with different combinations of axes using these four elements. From the m/z–RT image, peaks derived from the same peptide in the direction of the acquiring time are integrated. By adding algorithms to ensure reproducibility of m/z and RT, the same peak can be compared precisely across different samples, and a statistical comparison of identical peaks in different samples leads to the discovery of specific differentially expressed peptide peaks. Specific peaks are designated by their m/z and RT coordinates, and further analysis is based on these identifiers. The peptide search engine used in 2DICAL is MASCOT software (version 2.5.1; Matrix Science) using the Swiss-Prot mouse database (SwissProt\_2016\_01.fasta). The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium via the jPOST [35] partner repository with the data set identifier PXD021750. Statistical Analyses was performed with the open-source statistical language R (version 3.3.0). The 2DICAL intensity data were converted to protein value averaging the intensity data of peptides derived from the protein.

#### *2.10. Ingenuity Pathway Analysis of Proteomic Data*

A total of 2836 proteins were detected in both wild type and p300 S89A mice. The proteomic data were quantile normalized and subjected to differential expression analysis. 93 proteins showed significant changes (fold change (FC) ≤ −1.2 or FC ≥ 1.2, and *p* < 0.05) and were subjected to Ingenuity Pathway Analysis (IPA) (Qiagen, Redwood City, CA, USA) to identify top canonical pathways associated with S89A mutation in mice.

#### *2.11. DSS-induced Colitis Mouse Model*

Mice were treated starting on day 1 for up to 5 days with low (2%) dose of dextran sodium sulfate (DSS) administered via the drinking water, based on a previously described protocol [36], after which treatment was discontinued for up to 7 days, and the effect of DSS on % original body weight, histology by hematoxylin and eosin (H&E) and Alcian blue staining, and colon length was assessed.

#### *2.12. Data Analysis*

Numerical data were expressed as the means ± standard deviation (s.d.) and Student's t-test was performed, unless otherwise noted. *p*-values < 0.05 were considered significant.

#### **3. Results**

#### *3.1. Generation of p300 S89A Point Mutant Knock-in Mice*

We previously utilized in vitro CRISPR/Cas9 editing of a highly conserved insertion in the N-terminus of p300 (aa 61–70) to demonstrate its importance in regulating Wnt/β-catenin/nuclear receptor interactions [9]. To further explore this critical signaling nexus within the N-terminal domain of the Kat3 coactivator p300, we generated an S89A knock-in point mutation in exon 2 of the mouse p300 gene via site-specific mutagenesis. This mutation removes the highly evolutionarily conserved phosphorylation site at S89 and consequently modulates the interaction of multiple proteins with the N-terminus of p300. The mutant fragment was cloned into a targeting vector for the murine p300 gene. The resultant flip-excision (FLEx) switch construct (Figure S1) was used to generate p300 S89A germ line mice. After crossing the mice with a CMV-Cre mouse line, we confirmed successful introduction of the point mutation and removal of the wild type fragment at both the genomic DNA and messenger RNA levels by PCR and DNA sequencing. The homozygous p300 S89A mice, although born at sub-Mendelian ratios (approximately 50% less than anticipated), did not demonstrate any obvious significant abnormalities (Figure S2 and Table S1) and were fertile, albeit both male and female S89A mice exhibited slightly decreased body weights (<10%) (Figure S2). The mice were subsequently backcrossed with C57BL/6 mice minimally for 10 generations before being used for further experiments.

#### *3.2. p300 S89A Mice and Differential β-Catenin Kat3 Coactivator Usage*

We previously reported that phosphorylation at S89 of p300 enhanced the association of β-catenin with p300 and mutation of serine 89 to alanine abrogated this phosphorylation dependent increase in vitro [20]. To confirm that this observation was also true in S89A knock-in mice, we performed a co-immunoprecipitation assay using tissue from intestinal crypts in which the Wnt signaling pathway is highly activated. As anticipated based on our previous in vitro studies, the association of β-catenin with p300 was significantly reduced in S89A mice compared with wild type (WT) mice (Figure 2).

**Figure 2.** p300 S89A mice show differential usage of Kat3 coactivators CBP and p300 by β-catenin. Co-immunoprecipitation of β-catenin with CBP (**A**) or p300 (**B**) in S89A and WT mouse intestinal crypt cells. Control (IgG) antibody and anti-CBP antibody or anti-p300 antibody were used for immunoprecipitation followed by immunoblotting for β-catenin. Numerical values above protein bands indicate densitometric quantitation of β-catenin associated with CBP or p300. Bar graphs show densitometric quantitation normalized to respective control. Data in bar graphs (mean ± s.e.m.) representative of three independent experiments are shown. S89A, p300 S89A; WT, wild type; F, female; M, male; βcat, β-catenin; IP, immunoprecipitation. Whole immunoblots corresponding to immunoblot data are included in Figure S3.

#### *3.3. p300 S89A Mice Are Extremely Sensitive to Intestinal Insult*

Although p300 S89A mice did not exhibit obvious homeostatic defects under normal feeding and housing conditions, we decided to evaluate their response to insult. Inflammatory bowel disease (IBD), including Crohn's Disease and ulcerative colitis [36], may arise from infections caused by viruses or bacteria, damage due to ischemia, or disorders of autoimmunity in genetically predisposed individuals. One popular model of colitis utilizes dextran sodium sulfate (DSS) in the drinking water, which damages the intestinal epithelium and a vigorous inflammatory reaction within the intestine generally of several days duration [36]. One variation of this model involves repeated cycles of acute insult with subsequent repair via iterative cycles of DSS administration with intervening periods of recovery, thereby simulating chronic IBD [36]. We chose a relatively low (2%) dose of DSS and proceeded to administer it to seven-week-old female mice, both S89A and wild type C57BL/6 (WT), in their drinking water. After only one round of DSS, effects on control WT mice were minor, with only slight body weight reduction at day eight/nine (~2–3 days after withdrawal of DSS), whereas in sharp contrast, there was a dramatic (~20%) body weight reduction in the mutant mice and two S89A mice died at day 12 (Figure 3A, left panel). A similar trend was observed with male mice (Figure 3A, right panel).

**Figure 3.** p300 S89A mice are extremely sensitive to intestinal insult. S89A and WT mice were treated starting on day 1 for ~5 days with low (2%) dose of dextran sodium sulfate (DSS) (or vehicle (**B**) after which treatment was discontinued for ~7 days and the effect of DSS on % original body weight (**A**), histology (**B**,**C**), and colon length (**D**) was assessed. Data are mean ± s.d. (*n* = 3–9 per group). \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001. (**B**) For both WT and S89A mice without DSS treatment: H&E staining shows normal colonic mucosa. The crypt architecture is preserved and there is no evidence of acute, neutrophil-mediated epithelial injury or histologic features suggestive of ongoing, chronic mucosal injury. Alcian blue staining highlights intact goblet cells. (**C**) With DSS treatment: For WT mice: H&E staining shows normal colonic mucosa. The crypt architecture is preserved and there is no evidence of acute, neutrophil-mediated epithelial injury or histologic features suggestive of ongoing, chronic mucosal injury. Alcian blue staining highlights intact goblet cells. For S89A mice: H&E staining shows colonic mucosa with lamina propria replacement by granulation tissue and fibrinopurulent exudate, consistent with ulcer. The few remaining crypts exhibit architectural distortion, indicative of chronic mucosal injury. Alcian blue staining highlights mucin loss and decreased goblet cells, consistent with epithelial injury.

Prior to DSS treatment, the colonic epithelium of S89A mice was normal, essentially the same as WT mice. Prior to treatment, in both S89A and WT mice, crypt architecture was intact and there was no evidence of neutrophil mediated epithelial injury or histological features suggestive of ongoing chronic mucosal injury. Alcian blue staining demonstrated the presence of intact goblet cells (Figure 3B). Histological examination of control and mutant mouse colons at day 12 after DSS treatment (five days treatment, seven days off) showed that S89A mice were nearly devoid of normal colonic epithelium. Hematoxylin and eosin staining showed colonic mucosa with lamina propria replacement by granulated tissue and fibrinopurulent exudate, consistent with ulceration. The few remaining crypts exhibited architectural distortion, indicative of chronic mucosal injury. Alcian blue staining highlighted mucin loss and decreased goblet cells consistent with epithelial injury (Figure 3C), whereas the tissue from WT mice was normal. The colon length of S89A female mice compared to WT controls, one week after 5 days of DSS administration, was somewhat shorter (5.3 versus 6.1 cm) (Figure 3D).

Perturbations of host–microbiota homeostasis induced by the host genetics and/or environmental factors can fuel inflammation at mucosal surfaces [37,38]. S89A mice were initially housed separately from control mice. We therefore decided to examine whether cohousing and thereby intermixing of the microbiota of S89A and WT mice would affect sensitivity to DSS induced colitis. Although separately housed S89A mice were dramatically more sensitive to 2% DSS treatment (Figure 4A), S89A mice were protected against the effects of the 2% DSS treatment when co-housed with WT mice (Figure 4A, left panel and Figure 4B). Interestingly, if mice subsequently were separated again for four weeks after four weeks of co-housing, both WT and S89A mice had intermediate sensitivity to DSS treatment (Figure 4C). 16S rRNA gene sequencing of stool samples from the separately housed WT and S89A mice demonstrated large taxonomic differences in the microbiota as depicted in a barplot of the top ~20 taxa in each sample (Figure 4D, upper panel). The Shannon (alpha) diversity of the two groups was not significantly different, indicating that the communities had about the same evenness and similar numbers of organisms and similar distribution. On the other hand, the Bray–Curtis and Jaccard (beta) diversity (Figures S5 and S6) was different between the two groups, indicating that the two groups were significantly different in their composition (that is members of the communities). Interestingly, at least one particular bacterial species, OTU-217, which we identified as Kineothrix alysoides and was present in S89A mice yet absent or at very low levels in WT mice housed separately, was transferred effectively during cohousing, and may contribute to enhancing the sensitivity of WT mice to DSS induced colitis (Figure 4D, lower panel).

**Figure 4.** Co-housing of p300 S89A mice with wild type mice modulates severity of intestinal injury potentially via intermixing of microbiota. S89A and WT mice were treated on day 1 for ~4–5 days with low (2%) dose of dextran sodium sulfate (DSS) after which treatment was discontinued for ~7 days (**A**, **C**) or ~2 days (**B**), and the effect of DSS on % original body weight was assessed. A single

S89A mouse (dashed red line arrow) which was co-housed with WT mice, from ~4 weeks prior to the start of DSS treatment, was protected from intestinal injury similar to WT mice (**A**, left panel: individual mouse data), while the other S89A mice (not co-housed with WT mice) remained extremely sensitive to intestinal injury (**A**, right panel: aggregate mouse data according to genotype). S89A mice co-housed with WT mice, from ~4 weeks prior to the start of DSS treatment, were protected from intestinal injury (**B**). S89A and wild type mice co-housed ~4 weeks and then separated for ~4 weeks prior to the start of DSS treatment demonstrated an intermediate sensitivity to DSS treatment (**C**). Data are mean ± s.d. (*n* = 4–5 per group) unless otherwise indicated. \* *p* < 0.05. (D) 16S rRNA gene sequencing of stool samples from separately housed WT and S89A mice demonstrated large taxonomic differences in the microbiota (**D**, top). (*n* = 3–4 per group.) Bacterial species, OTU-217, was detected by PCR in most of the S89A mice and largely absent in the separately housed WT mice, whereas OTU-217 was detected in most of the co-housed WT and S89A mice (**D**, bottom). (*n* = 12–20 per group.) OTU, operational taxonomic unit.

#### *3.4. Genomic Analysis*

To further explore the impact of the knock-in point mutation on gene expression, we performed RNA-seq on tissue from the intestines from separately housed untreated p300 S89A and WT mice. To examine epithelium-specific gene expression, intestinal organoids were also grown from both p300 S89A and WT mice and analyzed by RNA-seq. Interestingly, between the organoids and whole ileum RNA-seq (Tables S2 and S3, respectively), there was limited overlap within the statistically significantly regulated genes (400 genes, 2.18% in organoids and 785, 2.86% in ileum q < 0.05), with only nine genes (*DUOX2*, *ERO1l*, *GSR* and *MPTX2* up-regulated and *BCMO1*, *PMP22*, *PRELP*, *SLC13A1* and *SST* downregulated) (Figure 5A) being common to both. Ingenuity Pathway Analysis (IPA) showed that in the organoids, the top affected network functions were: (1) Embryonic Development, Organismal Development and Function; (2) Cell-To-Cell Signaling and Interaction; (3) Cellular Assembly and Organization, Cell-To-Cell Signaling and Interaction; and (4) Lipid Metabolism, Molecular Transport, Small Molecule Biochemistry. The NRF2-mediated Oxidative Stress Response pathway was also strongly affected. DUOX2 and DUOXA2 members of the NADPH oxidase family, serve as the first line of defense against enteric pathogens by producing microbicidal reactive oxygen species and are the predominant H2O2-producing system in human colorectal mucosa [39,40].

Duox2 expression was significantly increased in both ileum (2.6-fold *<sup>p</sup>* = 5.00 × <sup>10</sup><sup>−</sup>5) and intestinal organoids (1.5-fold *<sup>p</sup>* = 7.99 × <sup>10</sup>−6). We further confirmed increased expression of Duox2 and Duoxa2 at both the message (Figure 5B) and increased expression of Duox2 (~3-fold) at the protein level (Figure 5C) in S89A mice. The transcription factor *GATA4*, which has been demonstrated to play a crucial role in patterning the intestinal epithelium and acts as a critical determinant of enterocyte identity in the jejunum [41] was up-regulated 3.6-fold (*<sup>p</sup>* = 1.08 × <sup>10</sup>−32) in intestinal organoids. Among the significantly down-regulated genes in the ileum was *REG3A* (~50% *<sup>p</sup>* = 5.00 × <sup>10</sup>−5), an antibacterial C-type lectin, which is constitutively generated in the intestine and displays anti-Grampositive bactericidal activity [42]. There was also a significant almost 50% reduction in the expression of the interferon-induced transmembrane protein 3, *IFITM3* (*p* = 0.047) in intestinal organoids. Differential expression of *IFITM3* has been found in endoscopic biopsies from Crohn's Disease patients [43]. In addition, the sulfate transporter *SLC13A1*, an FXR transcriptional target was down-regulated 2.5-fold (*<sup>p</sup>* = 4.67 × <sup>10</sup><sup>−</sup>8) in organoids with an approximately 50% reduction in ileum (*<sup>p</sup>* = 5.00 × <sup>10</sup><sup>−</sup>5), consistent with the importance of p300 Ser89 phosphorylation on p300/nuclear receptor interactions [44,45]. Additionally, members of the HOXB cluster, which is critical for specification of the digestive tract [46], including *HOXB3* and *HOXB5-9* were all down-regulated more than 2-fold in intestinal organoids (*<sup>p</sup>* = 7.97 × <sup>10</sup>−<sup>7</sup> to 4.49 × <sup>10</sup>−5). Among the top canonical pathways affected in the IPA analysis of the ileum RNA-seq were, interferon signaling, LPS/IL1 mediated inhibition of RXR function, estrogen biosynthesis and fatty acid and xenobiotic metabolism. 1.4-fold increases in Stat1 (*<sup>p</sup>* = 5.00 × <sup>10</sup><sup>−</sup>5) and Irf8 (*<sup>p</sup>* = 5.00 × <sup>10</sup><sup>−</sup>5), 2-fold (*<sup>p</sup>* = 5.00 × <sup>10</sup><sup>−</sup>5) increases in granzyme b (*Gzmb*) and immunity-related GTPase family M member 1 (*Irgm1*) and a 1.6-fold (*<sup>p</sup>* = 5.00 × <sup>10</sup>−5) up-regulation of the Interferon Inducible Protein 47 gene (*IFI47*) were observed in S89A mice consistent with the known IFN/STAT1 pathway dysregulation in IBD [47]. Serum Amyloid A1 (*SAA1*), which demonstrates bactericidal action in vitro, may provide a feedback protective mechanism in S89A mice and was increased 5.3-fold (*<sup>p</sup>* = 5.00 × <sup>10</sup><sup>−</sup>5) [48].

**Figure 5.** Genomic analysis of p300 S89A mice intestinal tissue. (**A**) RNA-seq analysis identifies 9 genes significantly differentially expressed in intestinal tissue and intestinal organoids derived from S89A mice versus those derived from WT mice. up, up-regulated; down, down-regulated. (*n* = 3 per group.) (**B**) RT-qPCR analysis of Duox2 and Duoxa2 mRNA levels in intestine of S89A and WT mice. (**C**) Immunoblot analysis of Duox2 protein levels in intestinal tissue of S89A and WT mice. Numerical values above protein bands indicate densitometric quantitation. Data in graphs are mean ± s.d. (*n* = 5–6 per group). \* *p* < 0.05. Whole immunoblots corresponding to immunoblot data are included in Figure S4.

#### *3.5. Proteomic Analysis*

We performed a targeted proteomic analysis of intestinal proteins associated with CBP and/or p300 in wild type versus p300 S89A mice. Interestingly, the aryl hydrocarbon receptor nuclear translocator-like protein 2 (Bmal2) demonstrated increased association with CBP versus p300 in both wild type and p300 S89A mice under both fed or fasted conditions, and although the ratio of CBP to p300 binding did not change significantly in the male mice, fed female p300 S89A mice showed a somewhat decreased Bmal2/CBP versus Bmal2/p300 interaction (Table S4). *BMAL2*, similar to its paralog *BMAL1*, forms a dimer with *CLOCK*, to activate E-box-dependent transcription thereby playing an active role in circadian-regulated transcription [49]. Bmal1 regulates Bmal2, therefore Bmal1 deletion by itself effects combined Bmal1/Bmal2 deletion [50]. Clock/Bmal1-mediated transcription is associated with rhythmic recruitment of Clock to p300 by Bmal1 [51] and differential avidity and timing of binding to CBP versus p300, may affect circadian regulation in S89A mice. Differential association of the N6 methyl adenosine 70 kDa subunit Mettl3 with enhanced p300 association in the female p300 S89A mice was also demonstrated. These results are interesting given the recent report of circadian clock regulation of lipid metabolism and in particular PPARα-mediated transcription being modulated by m6A mRNA methylation [52] and the role of p300 Ser89 in PPAR transcriptional regulation [23]. The effect of the p300 S89A mutation on circadian regulation was not investigated in this study, however given the effect of this mutation on nuclear receptor signaling in both the intestine and in the liver (to be reported separately), and the crosstalk between nuclear

receptors and core circadian transcriptional regulators [53], this area will be the focus of future investigations.

We next undertook global proteomic analysis of intestinal tissue from wild type and p300 S89A mice (Table S5), which revealed that 93 of the 2836 proteins detected in both WT and S89 mice proteins were significantly differentially expressed (with fold change ≤ −1.2 or fold change ≥ 1.2 and *p* < 0.05) (Table S6). IPA analysis of the 93 proteins showed that Mitochondrial Dysfunction, Oxidative Phosphorylation and Virus Entry via Endocytic Pathway were the top canonical pathways associated with S89A mutation (Figure 6). Interestingly, we have identified similar effects on mitochondrial dysfunction and oxidative phosphorylation in other organ systems in S89A mice (e.g., brain, liver, adipose tissue, to be reported separately) as well as cell-based model systems that affect differential Kat3 coactivator usage (i.e., P19 p300 N-terminally edited cells Ono et al. [9]). Metabolic dysfunction appears to be a fundamental feature associated with aberrant differential Kat3/β-catenin coactivator usage (Kahn lab manuscript in preparation). Further, the protein expression level of the bile acid transporter protein FABP6, the 2nd most downregulated protein found in S89A mice, was approximately 10% of that in their wild type counterparts. This result is consistent with FABP6, which is required for efficient absorption and transport of bile acids in the distal intestine, being a PPAR target gene [54] that is repressed by GATA4 in the small intestine. *FABP6* message was also significantly decreased in S89A mice (*p* = 0.00035). Bile acids and their FXR nuclear receptors play important roles in inflammatory response and intestinal barrier function and are involved in IBD pathophysiology [55]. Calreticulin (CALR), which appears to play a role in leukocyte infiltration in mouse models of colitis via its interaction with alpha integrins, was downregulated in S89A mice (~30%) [56]. Calreticulin, also is secreted by macrophages and binds to target cells marking them for removal by programmed cell phagocytosis [57] and believed to function as an "eat me" signal. Viable cells also can expose calreticulin on their surfaces, apparently protected from engulfment via concurrently expressed so called "don't eat me" signals, e.g., CD200 and CD47 [57]. Interestingly, the expression of the OX-2 membrane glycoprotein (CD200) is increased 1.5-fold in S89A intestines. The role of these differentially expressed proteins in innate immunity and the intestinal phenotype displayed in S89A mice will require further investigation.


**Figure 6.** Bioinformatic analysis of proteins differentially expressed in p300 S89A mice intestinal tissue. IPA bioinformatic analysis of the 93 genes identified by proteomic analysis to be significantly differentially expressed in intestinal tissues of S89A versus WT mice (with fold change ≤ −1.2 or fold change ≥ 1.2 and *p* < 0.05) revealed the Mitochondrial dysfunction and Virus entry via endocytic pathway as top canonical pathways associated with S89A mutation (top). Differentially expressed proteins comprising the Mitochondrial dysfunction and Virus entry via endocytic pathways and associated fold change (FC) in intestinal tissues of S89A versus WT mice (bottom). (*n* = 3–4 per group).

#### *3.6. p300 S89A Is a Part of a 14-3-3 Binding Motif*

The 14-3-3 protein family of scaffolding chaperones regulates diverse intracellular signaling pathways [58]. We observed that the p300 sequence LLRSGSSP (aa 84–91) is a member of the consensus 14-3-3 binding site sequence (LX(R/K)SX(pS/pT)XP) [59]. It is unique to p300 and not conserved in CBP. We therefore anticipated that mutation of serine 89 to alanine would disrupt the binding of 14-3-3 proteins to p300. Immunoprecipitation of 14-3-3 epsilon (14-3-3ε) was performed using protein from intestines of WT and S89A mice and subsequently immunoblotted with an antibody specific for p300. As shown (Figure 7) a substantial decrease in the association of 14-3-3ε with p300 was demonstrated in intestinal tissue from S89A mutant mice. To confirm these findings, we carried out the reverse experiment, i.e., anti-p300 antibody was used for immunoprecipitation followed by immunoblotting for 14-3-3ε. Again, we found a substantial decrease in the association of 14-3-3ε with p300 in S89A mutant mouse intestinal tissue (Figure S7). Further studies are needed to address the importance of this interaction, however differential subcellular localization of p300 regulated by its interaction with 14-3-3 proteins could potentially affect its role as a nuclear transcriptional coactivator.

**Figure 7.** Association of 14-3-3ε with p300 is decreased in intestinal tissue from p300 S89A mice. Coimmunoprecipitation of p300 with 14-3-3ε in S89A and WT mouse intestinal crypt cells. Control (IgG) antibody and anti-14-3-3ε antibody were used for immunoprecipitation followed by immunoblot for p300. Bar graphs show densitometric quantitation of p300 associated with 14-3-3ε versus control IgG in intestinal tissues of S89A versus WT mice.

#### **4. Discussion**

Differential Kat3 coactivator usage by β-catenin is a fundamental regulatory mechanism in stem cell maintenance and the initiation of differentiation and repair [1,60]. Stem cells in their respective niches receive a myriad of information including oxygen and nutrient levels, circadian input, adhesion molecules, cell–cell contacts, growth factors, etc., to decide to maintain quiescence or to enter the cell cycle and undergo either symmetric or asymmetric division [1,61]. The extreme N-terminal 111 amino acids of CBP and p300, decidedly the most divergent regions of the two Kat3 coactivators [2], contain binding domains for β-catenin, nuclear receptors [9] and Stat1, an interferon-dependent transcription factor [62], as well as approximately 20 serine/threonine residues [1,2,63,64]. Post-translational modifications of these serine/threonine residues (by phosphorylation or dephosphorylation) [21–24] and combinatorial interactions, both antagonistic and synergistic [9], of multiple transcription factor families (i.e., β-catenin/TCF/LEF, β-catenin/FOXO, nuclear receptors, e.g., RAR, VDR, PPAR, etc., Stat1, 2 as well as others), provide a unique mechanism to integrate a diverse array of signal inputs. We previously demonstrated that a highly evolutionarily conserved 27bp/9aa insertion in the N-terminus of p300, which is not present in CBP, between the conserved β-catenin-binding region (DELI-sequence) and the nuclear receptor binding sequence (LXXLL), determines if the interaction will be potentially synergistic or purely antagonistic between the Wnt/β-catenin and nuclear receptor signaling cascades [2,9].

To further investigate mechanisms of signal integration effected by this domain of the Kat3 family, we generated p300 S89A knock-in mice. Based upon our earlier pharmacologic studies, p300 S89 is critical for controlling differential coactivator usage by β-catenin via post-translational phosphorylation in stem/progenitor populations [1,20,26] and Serine 89 appears to be a target for a number of kinase cascades [21–24]. Although, the p300 S89A polymorphism has not been reported in humans, to the best of our knowledge, homozygous p300 S89A mice, albeit born at sub-Mendelian ratios and exhibiting slightly decreased body weights, were relatively normal and fertile. However, after insult/stress and with aging (to be published separately later), we have found a range of interesting phenotypes associated with this single point mutation. Herein we report our initial findings regarding the intestinal phenotype of the p300 S89A mice. We first investigated the interaction of β-catenin with p300 in intestinal crypts, a region associated with activated Wnt signaling and found as anticipated that it was significantly reduced in S89A mice (Figure 2, right panel). Decreased β-catenin/p300 interaction in the p300 S89A mice did not manifest itself in obvious defects in either ileal or colonic architecture under normal homeostatic conditions (Figure 3C). However, treatment with a mild insult (2% DSS), while having minimal effects on wild type C57BL/6 mice, had a striking effect on the p300 S89A mice as evidenced by the development in the p300 S89A mice of severe colitis, a significant risk factor predisposing to colorectal cancer [27,28] (Figure 3A,B).

Further investigation demonstrated cell intrinsic differences in the intestinal epithelium, based upon RNA-seq of intestinal organoids, as well as microbiome compositional differences and differential immune responses in the intestine. Interestingly, S89A mice separately housed were dramatically more sensitive (Figure 4A) than when co-housed with WT mice (Figure 4A, left panel and Figure 4B). However, when separated again for four weeks after being co-housed for four weeks, S89A mice demonstrated intermediate sensitivity to DSS treatment (Figure 4C). These results point to a complex interplay between host intrinsic differences in the epithelium and extrinsic interaction with the intestinal microbiome associated with differential microbiome colonization and metabolite production [65,66] the host immune response, both innate and adaptive [67], related to a single amino acid variance within the highly conserved and critical region of signal integration in p300 [1,68].

Global genomic and proteomic analysis showed a number of prominent pathway differences including lipid metabolism, oxidative stress response, mitochondrial function and oxidative phosphorylation. We have found in further analyses of other organ systems (liver, brain and adipose tissue) that these are fundamental differences generally associated with differential Kat3/β-catenin coactivator usage. Notably, sulfate transporter *SLC13A1*, an FXR transcriptional target was significantly down-regulated in both organoids and ileum of S89A mice, consistent with the importance of p300 Ser89 phosphorylation on p300/nuclear receptor interactions [44,45]. Sulfate insufficiency impedes detoxification, heightens the risk of xenobiotic toxicity, and modifies the activity and metabolism of numerous physiologic compounds, including proteoglycans, hormones, and neurotransmitters [69] and very recently integrated microbiota and metabolite profiles linked Crohn's disease with sulfur metabolism [65]. Interestingly, the levels of both *p*-cresol sulfate and phenol sulfate, potentially toxic intestinal bacterial fermentation products [70], were significantly upregulated in the liver metabolome of S89A mice (to be published separately). Given the importance of cellular metabolism and mitochondrial function in the regulation of the immune response [71,72], further investigations with regard to intestinal immunity in the p300 S89A mice and the effects in other organ systems are ongoing and will be reported in due course. Additionally, mitochondrial activity has been linked to maintaining a state of physiological hypoxia at the colonic surface. Limiting the amount of oxygen at the mucosal surface controls the aerobic growth of facultative anaerobic bacteria [73], whereas reduced mitochondrial bioenergetics decreases epithelial oxygen consumption, thereby increasing epithelial oxygenation and the diffusion of oxygen into the intestinal lumen [74–76]. Recent experimental evidence [77] has provided support for the hypothesis that an expansion of facultative anaerobic bacteria in IBD patients are secondary to changes in epithelial energy metabolism [78]. Furthermore, it was demonstrated that treating mitochondrial dysfunction in the colon using the PPAR agonist 5-ASA [79], consistent with the dysfunctional PPAR signaling associated with S89A mutation, ameliorated signs of disease in mice with pre-IBD and normalized the microbiota composition by restoring epithelial hypoxia [77].

The diverse array of effects on fundamental processes including epithelial differentiation, metabolism, immune response and microbiome colonization, all brought about by a single amino acid modification S89A, highlights the role of this region in the Kat3 coactivator p300 as a critical signaling nexus and the rationale for conservation of this residue and surrounding region for hundreds of million years of vertebrate evolution. Additional studies related to the fundamental regulation of metabolism via differential Kat3/β-catenin usage and its roles in development and disease will be reported in due course.

#### **5. Conclusions**

We describe the generation of novel p300 S89A knock-in mice carrying a single site directed amino acid mutation in p300, changing the highly evolutionarily conserved serine 89 to alanine, thus enhancing Wnt/CBP/catenin signaling (at the expense of Wnt/p300/catenin signaling). We show that S89A mice are extremely sensitive to intestinal insult resulting in colitis, which is known to significantly predispose to colorectal cancer. We demonstrate cell intrinsic differences, and microbiome compositional differences and differential immune responses, in the intestines of S89A versus wild type mice. Genomic and proteomic analyses reveal pathway differences, including lipid metabolism, oxidative stress response, mitochondrial function and oxidative phosphorylation. The diverse effects on fundamental processes including epithelial differentiation, metabolism, immune response and microbiome colonization, all brought about by a single amino acid modification S89A, highlights the critical role of this region in p300 as a signaling nexus in development and disease (e.g., inflammation and cancer) and the rationale for conservation of this residue and surrounding region for hundreds of million years of vertebrate evolution.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2072-6 694/13/6/1288/s1, Figure S1: Schematic of flip-excision switch construct used to create p300 S89A germ line mice, Figure S2: Body weight and blood count did not show any major abnormalities between p300 S89A mice and wild type mice, Figure S3: p300 S89A mice show differential usage of Kat3 coactivators CBP and p300 by β-catenin, Figure S4: Immunoblot analysis of Duox2 protein levels in intestinal tissue of S89A and WT mice, Figure S5: Bray–Curtis (beta) diversity between wild type and p300 S89A mutant mouse groups, Figure S6: Jaccard (beta) diversity between wild type and p300 S89A mutant mouse groups, Figure S7: Association of 14-3-3ε with p300 is decreased in intestinal tissue from p300 S89A mice, Table S1: Comparison of Blood Chemistry between p300 S89A and Wild Type Mice, Table S2: Organoids RNA-seq, Table S3: Intestine RNA-seq, Table S4: Intestine co-IP of CBP vs. p300 proteomics, Table S5: Intestine global proteomics, Table S6: Proteomic Identification of Differentially Expressed Proteins in p300 S89A vs. Wild Type Mice.

**Author Contributions:** K.K.Y.L.: formal analysis, funding acquisition, writing. X.H.: formal analysis, investigation, visualization, review and editing. K.C.: formal analysis, investigation, visualization, review and editing. C.N.: formal analysis, investigation, visualization, review and editing. D.P.L.: visualization, review and editing. K.K.L.: formal analysis, writing. N.K.: formal analysis, review and editing. Y.H.: formal analysis, review and editing. S.K.H.: formal analysis, review and editing. E.M.: investigation, review and editing. Y.E.: formal analysis, investigation, visualization, review and editing. P.T.F.: formal analysis, review and editing. A.J.O.: formal analysis, review and editing. N.-O.C.: formal analysis, visualization, writing. M.O.: formal analysis, investigation, writing. M.K.: conceptualization, formal analysis, funding acquisition, project administration, resources, supervision, writing. All authors have read and agreed to the published version of the manuscript.

**Funding:** K.K.Y.L. has been supported by NIH K08AA025112. M.K. has been supported by City of Hope Comprehensive Cancer Center Support Grant NIH P30CA033572, NIH R01CA166161, R21NS074392, R21AI105057, and R01HL112638, and the Rotary Coins for Alzheimer's Research Trust (CART).

**Institutional Review Board Statement:** Animal studies were approved by the University of Southern California Institutional Animal Care and Use Committee (IACUC) as per protocol #11023.

**Data Availability Statement:** Proteomics data have been deposited in the ProteomeXchange Consortium via the jPOST [35] partner repository with the data set identifier PXD021750. Additional genomics and proteomics data have been included in Supplementary Materials. All other data supporting the findings of this study are available from the corresponding author on reasonable request.

**Acknowledgments:** The authors thank Tomoyo Sasaki for her technical contributions to the project.

**Conflicts of Interest:** M.K. has an equity position in 3 + 2 Pharma. The other authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Article* **The Hepatitis B Virus Pre-Core Protein p22 Activates Wnt Signaling**

**Bang Manh Tran 1, Dustin James Flanagan 1,2, Gregor Ebert 3,4, Nadia Warner 5, Hoanh Tran 3,4, Theodora Fifis 6, Georgios Kastrappis 6, Christopher Christophi 6, Marc Pellegrini 3,4, Joseph Torresi 7, Toby James Phesse 1,8,\* and Elizabeth Vincan 1,5,9,\***


Received: 21 April 2020; Accepted: 27 May 2020; Published: 31 May 2020

**Abstract:** An emerging theme for Wnt-addicted cancers is that the pathway is regulated at multiple steps via various mechanisms. Infection with hepatitis B virus (HBV) is a major risk factor for liver cancer, as is deregulated Wnt signaling, however, the interaction between these two causes is poorly understood. To investigate this interaction, we screened the effect of the various HBV proteins for their effect on Wnt/β-catenin signaling and identified the pre-core protein p22 as a novel and potent activator of TCF/β-catenin transcription. The effect of p22 on TCF/β-catenin transcription was dose dependent and inhibited by dominant-negative TCF4. HBV p22 activated synthetic and native Wnt target gene promoter reporters, and TCF/β-catenin target gene expression in vivo. Importantly, HBV p22 activated Wnt signaling on its own and in addition to Wnt or β-catenin induced Wnt signaling. Furthermore, HBV p22 elevated TCF/β-catenin transcription above constitutive activation in colon cancer cells due to mutations in downstream genes of the Wnt pathway, namely *APC* and *CTNNB1*. Collectively, our data identifies a previously unappreciated role for the HBV pre-core protein p22 in elevating Wnt signaling. Understanding the molecular mechanisms of p22 activity will provide insight into how Wnt signaling is fine-tuned in cancer.

**Keywords:** Wnt signaling; hepatitis B virus; HBV; cancer; liver cancer; β-catenin; TCF/LEF

#### **1. Introduction**

Liver cancer is the second most common cause of cancer deaths worldwide and is projected to increase by ~40% by 2030 [1]. The most common type of liver cancer is hepatocellular carcinoma (HCC), which has very limited treatment options and a poor prognosis because it is usually diagnosed at a late stage [2]. The Wnt signal transduction pathway is aberrantly activated in most cases of HCC and mutations to the catenin beta 1 (*CTNNB1*) gene, the gene that codes for β-catenin, occurs in up to 40% of cases making it the most frequent mutation in HCC [3,4]. β-Catenin is the main effector of the canonical Wnt signaling pathway [5] and these mutations to *CTNNB1* lead to constitutive activation of Wnt signaling [6,7]. Liver cancer is also linked to chronic infection with the hepatitis B virus (HBV) that leads to cirrhosis and accounts for 50% of HCC cases [8]. Here, we investigated the oncogenic interplay between these two drivers of liver cancer, namely HBV and Wnt signaling.

Wnt/β-catenin signaling is activated by the coupling of Wnt to its cognate receptor, Frizzled (FZD), which initiates a series of events in the cytoplasm that leads to the activation of (TCF)/lymphoid enhancer factor (LEF)/β-catenin (referred to as TCF/β-catenin for simplicity from here on) mediated gene transcription. In the absence of Wnt, β-catenin is primarily engaged at cell-cell adherens junctions and any free β-catenin is cleared by a cytoplasmic destruction complex that contains several proteins, including Axin, adenomatous polyposis coli (APC), glycogen synthase kinase 3 (GSK3) and casein kinase 1 (CK1) [5]. Free, cytoplasmic β-catenin associates with the destruction complex and is sequentially phosphorylated by CK1 and GSK3 at its N-terminus, a post-translational modification that targets it for ubiquitylation and proteasomal degradation. However, upon activation of Wnt-FZD signaling, GSK3 enzyme activity is inhibited and β-catenin escapes phosphorylation and subsequent degradation, accumulates in the cytoplasm and translocates into the nucleus where it complexes with the enhanceosome to initiate the TCF/β-catenin target gene transcription [9]. In liver cancer, the phosphorylation sites of β-catenin are absent due to mutations to the *CTNNB1* gene, leading to the constitutive activation of Wnt/β-catenin signaling [3,4,10].

Another common etiologic factor in liver cancer is HBV infection [10,11]. HBV is an enveloped DNA virus whose genome codes for four overlapping genes, namely the envelope or surface (*S*) gene, the core (*C*) gene, the *X* gene and the polymerase (*P*) gene. The protein products include the surface antigens coded by the *S* gene, the capsid core proteins coded by the *C* gene and the HBx protein coded by the *X* gene. Post-translational processing of the HBV pre-core protein (p25) yields the HBV e antigen (HBeAg, p17) via a p22 intermediate [12]. The HBx protein has been extensively studied for its effects on Wnt/β-catenin signaling [13], however, much less is known about the potential oncogenic interplay with the other HBV proteins. Here, we performed a screen to determine the effects of HBV proteins on Wnt/β-catenin signaling and identified p22, the HBe precursor protein, as a potent activator on its own and in conjunction with active Wnt signaling. Importantly, p22 activated Wnt/β-catenin signaling in colon cancer cells that harbor mutations in intracellular components of the Wnt signaling cascade that result in constitutive activation of signaling. Concomitant regulation of Wnt signaling at multiple levels of the signaling cascade via various mechanisms (genetic, epigenetic, post-translational etc.) to achieve the "just right" level of Wnt signaling for a particular process is a common theme emerging for Wnt-addicted cancers [14–16] and here, we demonstrate that HBV p22 might contribute to our understanding of this fine tuning in cancer.

#### **2. Results**

#### *2.1. E*ff*ect of HBV Proteins on TCF-*β*-Catenin Transcription*

To investigate novel mechanisms of oncogenic interaction between HBV and Wnt signaling we screened the ability of various HBV proteins (Figure S1) for their effect of TCF/β-catenin transcription in the presence of Wnt stimulation (Wnt3a conditioned medium). TCF/β-catenin transcription was detected using the TCF reporter, super TOPflash (sTOPflash), which contains eight TCF response elements upstream of a minimal TK (Thymidine Kinase) promoter and sFOPflash, which has the TCF sites mutated [17,18]. The HBx protein activated TCF/β-catenin transcription above Wnt stimulation, however, the pre-core protein p22 was able to increase Wnt activity to a level markedly greater than the HBx protein (Figure 1a). The HBV envelope proteins did not activate reporter activity, nor did the pre-core precursor p25 or core p21, despite significant overlap in the amino acid sequence between the core/precore proteins (Figure S1). The precore contains the genetic sequence of two different proteins, the core protein HBc (p21) (183 amino acids) and precore polypeptide p25 (212 amino acids). They differ only by 29 amino acids at the N-terminus as p25 retains the signal sequence. The cleavage of 19 amino acids from this signal sequence releases cytosolic p22. P22 is further truncated, losing the arginine-rich C-terminal domain, to yield HBe (p17), which is secreted [19]. Expression of p22 was confirmed by immunoblot on whole cell lysates prepared from transfected Huh7 cells using an anti-HBc antibody and, as shown by others [19], neither p17 nor p25 were detected by immunoblot (Figures 1b and S2). HBV p17 and p25 were detected by confocal immunofluorescence in transfected Huh7 cells (Figure S3). Confocal microscopy of Huh7 cells transfected with pCI-p22 and the same anti-core antibody showed diffuse cytoplasmic, diffuse nuclear and, cytoplasmic puncta (Figures 1c and S3) placing p22 in the cellular compartments where Wnt signaling components are found [20].

**Figure 1.** Wnt signaling activation is induced by hepatitis B virus (HBV) precore protein p22. (**a**) Effect of various HBV proteins on TCF/β-catenin transcription activity in Huh7 cells, was determined by reporter activity (sTOPflash reporter) and is shown as fold change relative to empty vector (EV) (mean ± SEM, \* *p* < 0.05, \*\*\* *p* < 0.0001 Student *t*-test, *n* ≥ 3 independent experiments for each data point) (**b**) Expression of protein from the indicated plasmids transfected in Huh7 cells was confirmed by immunoblot. Lysates prepared from Huh7 cells transfected with EV and the parental, un-transfected cells served as negative controls. Lysate from HBV core p21 transfected Huh7 cells was used as a positive control. The membrane was stained with anti-HBc antibody first, then re-probed with anti α-tubulin antibody. (**c**) Huh7 cells were transfected with p22 plasmid and p22 protein expression (red) and localization detected with anti-HBV core antibody and confocal microscopy (nuclei are blue). Scale bars = 20 μM.

#### *2.2. HBV p22 Activates TCF-*β*-Catenin Transcription*

Next, we demonstrated that p22 activates Wnt signaling on its own and can increase Wnt signaling activity in cells, which are stimulated with either Wnt3a or ectopic over-expression of full length, wild type β-catenin (β-cat-WT) (Figure 2a). The stimulatory effect of p22 on reporter activity was dose-dependent (Figure 2b) and decreased at the higher levels of p22 in the presence of β-cat-WT (Figure 2c). Notably, the levels of transcriptionally active non-phosphorylated β-catenin (β-cat-ACT) [21,22] were increased above that seen with β-cat-WT when p22 was co-expressed (Figures 2d and S4). In the presence of active Wnt signaling, β-catenin escapes phosphorylation and subsequent degradation, and the elevated levels of β-cat-ACT confirm this mechanism for p22 activation of TCF/β-catenin transcription. Data to illustrate the comparative reporter activity between the different conditions is shown in Figure S5.

**Figure 2.** HBV p22 stimulates Wnt signaling in Huh7 cells. (**a**) The effect of HBV p22 alone or in addition to stimulation by Wnt 3a or wildtype β-catenin (β-cat-WT) on TCF/β-catenin transcription in Huh7 cells, was determined by reporter activity (sTOPflash reporter) and is shown as fold change relative to empty vector (EV) (mean±SEM, \*\*\* *p* < 0.0001 Student *t*-test, *n* = 8 independent experiments). (**b**) Huh7 cells were transfected with the indicated amounts of p22 expression plasmid. The figure shows the dose-dependent effect of HBV p22 on TCF/β-catenin transcription activity (sTOPflash reporter) (mean ± SEM, \* *p* < 0.05, \*\* *p* < 0.001 Student *t*-test, *n* = 4 independent experiments). (**c**) Huh7 cells were transfected with the indicated amounts of p22 and 100 ng of wild-type β-catenin expression plasmids. Co-expression of 5–50 ng p22 increased TCF/β-catenin transcription activity (sTOPflash reporter) mediated by wild-type β-catenin; reporter activity decreased when 100 or 200 ng p22 was co-transfected with wild-type β-catenin (mean ± SEM, \*\* *p* < 0.001, \*\*\* *p* < 0.0001 Student *t*-test, *n* = 3 independent experiments). (**d**) Immunoblot analysis for the transcriptionally active form of β-catenin (β-cat-ACT) on lysates prepared from Huh7 cells co-transfected with 100 ng wild-type β-catenin, 100 ng of p22 or equivalent EV expression plasmids. The membrane was stripped and re-probed with anti-actin antibody. The bar graph shows quantitative analysis for the levels of detected active β-catenin using Image Lab software and normalized for β-actin levels (mean ± SEM, \*\* *p* < 0.001 Student *t*-test, *n* = 3 samples).

During natural HBV infection, p22 is processed to p17 or HBV e antigen (HBeAg) and secreted into the extracellular space [19]. We confirmed that the transfected p22 is processed to p17 by detecting and quantifying HBeAg in the supernatant of transfected Huh7 cells (Figure S6). Notably, ectopically expressed p17 or p25 did not activate sTOPflash reporter activity above activation by β-catenin (Figure S7).

#### *2.3. HBV p22 Activates Native TCF*/β*-Catenin Promoters*

Next, we tested the ability of p22 to activate native TCF/β-catenin target gene promoters. First, we used our previously characterized Frizzled-7 (FZD7) promoter reporter, pFz7-prom [23]. FZD7 is a TCF/β-catenin target gene [23,24] and forms a positive feedback loop in various cancers, including HCC [25–27]. As shown above with the sTOPflash reporter (Figure 2a), HBV p22 activated the pFz7-prom on its own, and in the context of Wnt3a stimulation or β-cat-WT over-expression (Figure 3a).

Secondly, given that Wnt signaling is dependent on a three-dimentional tissue context [28], we tested the ability of p22 to activate native TCF/β-catenin target gene promoters in the liver in vivo. HBV is an exquisitely human hepatotropic virus and does not infect mouse hepatocytes. However, using hydrodynamic tail vein injection (HDI) plasmids can be introduced into mouse hepatocytes in live animals [29]. A large volume of plasmid containing saline was intravenously injected into mice. This volume overwhelms the heart and is shunted into the hepatic vein and the hepatocytes take up the injected solution (Figure 3b). The mice were culled 6 days and 20 days post HDI and their livers processed for mRNA gene expression analyses using quantitative RT-PCR (qRT-PCR). Expression of Wnt target genes (e.g., Fzd7, Glul) and those that are not target genes (e.g., SOCS3) was determined. At 6 days post-HDI, cyclin D2 and SOCS3 were upregulated by p22 (Figure S8a). Cyclin D2 is upregulated upon activation of Wnt signaling via truncating the *APC* gene and regulates proliferation in this setting [30], suggesting it is a Wnt target gene, however this may be indirect. Fzd7, a Wnt target gene [23,24] shows a trend in upregulation in response to p22 at 6 days post HDI, which was significantly different by 20 days post-HDI (Figures 3c and S8), whilst the expression of another TCF/β-catenin target gene glutamine synthetase (Glul, Figures 3c and S8b) was only upregulated by p22 at day 20, suggesting early and late regulation or signaling thresholds. There were trends towards increased expression of other TCF/β-catenin target genes but these changes did not reach significance (full qRT-PCR gene analyses are shown in Figure S8 and primer sequences in Table S1). Collectively, these data show p22 activates natural promoters of TCF/β-catenin target genes in the context of a human liver cancer cell line Huh7 (Figure 3a) and normal liver hepatocytes in vivo (Figures 3c and S8).

**Figure 3.** HBV p22 activates TCF/β-target gene native promoters. (**a**) Effect of HBV p22 on FZD7-native promoter reporter activity, with and without stimulation with Wnt3a or 100 ng wild-type β-catenin (β-cat-WT), in Huh7 cells was determined by luciferase activity (pFz7-prom reporter) and is shown as fold change relative to empty vector (EV) (mean ± SEM, \*\* *p* < 0.001, \*\*\* *p* < 0.0001 Student *t*-test, *n* = 6 independent experiments). (**b**) Schematic diagram of hydrodynamic tail-vein injection in mice (adapted from [31]). (**c**) Expression of TCF/β-target genes Fzd7 and glutamine synthase (Glul) was increased in mouse livers 20 days post HDI injection of p22. Gene expression was determined by qRT-PCR and is shown relative to empty vector (EV) (mean ± SEM, \* *p* < 0.05 Student *t*-test, *n* ≥ 4 mice).

#### *2.4. HBV p22 Activates TCF*/β*-Catenin Transcription in Addition to a Mutation to Downstream Wnt Pathway Components*

Thus far, we have demonstrated that p22 activates TCF/β-catenin transcription on its own and in the context of Wnt stimulation and β-cat-WT over-expression. This mimics one scenario of additional Wnt signaling in cancer i.e., signaling from the ligand-receptor complex. Next, we investigated p22 activity in other cancer contexts, namely in the context of mutant intracellular components that constitutively activate the Wnt pathway i.e., truncated APC and stabilized, mutant β-catenin.

The role of Wnt signaling in cancer has been most extensively studied in colon cancer as Wnt signaling is frequently deregulated in these cancers [32]. Thus, to investigate the effect of p22 in cancer cells with endogenous mutations to intracellular Wnt pathway components, we used colon cancer cell lines SW480 and HCT116 that harbor truncated APC and mutated β-catenin, respectively [18,33]. We also tested the effect of p22 in HEK293T cells that have no known mutations in the Wnt pathway and are known to respond to Wnt [34]. In each cell line (HEK293T, SW480 and HCT116) p22 activated TCF/β-catenin transcription (sTOPflash) above the basal level (Figure 4a).

**Figure 4.** HBV p22 increases TCF/β-catenin signaling in the context of oncogenic activation of the Wnt pathway. (**a**) Effect of 100 ng p22 expression plasmid on TCF/β-catenin transcription activity (sTOPflash reporter) in HEK293T cells with no known mutation or aberrant modulation of Wnt signaling; SW480 cells with truncated, mutant APC, rendering Wnt signaling constitutively active and HCT116 cells with mutation at the N-terminus of β-catenin, making Wnt signaling constitutively active (mean ± SEM, \* *p* < 0.05, \*\*\* *p* < 0.0001 Student *t*-test, *n* = 3, 5 and 3 experiments, respectively). Reporter activity is expressed relative to empty vector (EV). (**b**) HBV p22 upregulates TCF/β-catenin transcription (sTOPflash reporter) in the context of truncated APC and this upregulation is blocked by dnTCF4. SW480 cells were co-transfected with 100 ng of p22 and dnTCF4 expression plasmids and the reporter activity is expressed relative to EV (mean ± SEM, \*\*\* *p* < 0.0001 Student *t*-test, *n* = 5 experiments). (**c**) HBV p22 upregulates TCF/β-catenin transcription (sTOPflash reporter) in the context of mutant, oncogenic β-catenin in Huh7 cells. The effect of co-transfection of 100 ng p22 expression plasmid with 100 ng of wild-type or mutant β-catenin on TCF/β-catenin is shown relative to EV (mean ± SEM, \*\*\* *p* < 0.0001 *t*-test, *n* = 4 experiments).

There are four mammalian TCF genes and TCF4 is known to be expressed by SW480 cells [18]. Thus, we tested the ability of a dominant negative form of TCF4 (dnTCF4) [18] to inhibit TCF/β-catenin transcription (sTOPflash) in this cell line. As expected, dnTcf4 decreased constitutive Wnt signaling

in SW480 cells. HBV p22 increased Wnt signaling in SW480 cells and this increase was reduced by dnTcf4 (Figure 4b). Collectively, these data show p22 regulates Wnt/β-catenin signaling in the context of genetic mutations that initiate Wnt-addicted cancers.

Next, to further test p22 activity in the context of mutant β-catenin compared to β-cat-WT, we used the N-terminally truncated, oncogenic form of β-catenin (ΔN-β-cat) that lacks the regulatory domains [33]. ΔN-β-Cat increased TCF/β-catenin transcription (sTOPflash) above β-cat-WT to a similar level as p22, while ΔN-β-cat and p22 together elevated reporter activity above either alone (Figure 4c). Data to illustrate comparative reporter activity between some of these different conditions is shown in Figure S5.

#### **3. Discussion**

The emerging theme for Wnt-addicted cancers is that the pathway is regulated via multiple mechanisms [16]. This has been extensively investigated in colon cancer. Colon cancers frequently harbor truncating mutations to *APC* that yield proteins with impeded function in degrading β-catenin; or oncogenic mutations to the *CTNNB1* gene that remove the destruction complex phosphorylation sites in the N-terminus of β-catenin [35]. The end result of either mutation is the constitutive activation of Wnt signaling and adenoma formation [6,18,33,36,37]. However, Wnt signaling is also deregulated at the level of the ligand/receptor in colon cancer. Naturally occurring inhibitors of Wnt-FZD interaction are silenced by promoter hypermethylation, while Wnts and FZDs are over-expressed (reviewed in [15,25]). Thus, transcription of TCF/β-catenin target genes can be increased or decreased despite a mutation to downstream components of the pathway. Indeed, all Wnt-addicted cancers show concomitant deregulation to Wnt signaling via intracellular and cell surface mechanisms [16]. Consistent with this, a potent anti-tumor effect was demonstrated by blocking FZD7 function in gastric cancer cells with and without mutant *APC* [38].

Notably, liver cancer displays similar Wnt-addicted mechanisms to colon and gastric cancer [16]. Constitutive activation of Wnt signaling in HCC is primarily via mutations to the *CTNNB1* gene that remove the regulatory phosphorylation sites from the N-terminus of β-catenin [3]. However, as in colon and gastric cancer, there is additional regulation of the pathway via over-expression of Wnts and FZDs and epigenetic silencing of naturally occurring inhibitors of Wnt-FZD interaction, for example secreted frizzled related proteins (sFRP) [16,39]. Furthermore, most cases of HCC have a viral etiology and are the culmination of chronic infection with HBV leading to liver disease where HBV proteins, such as HBx, are hypothesized to exert their oncogenic activity, at least in part, through activation of Wnt/β-catenin signaling [8]. Here, we screened the various HBV proteins for their impact on Wnt signaling and demonstrated that another HBV protein, p22, was more potent than HBx. HBV surface proteins (small, middle or large) did not activate TCF/β-catenin transcription. Interestingly, the other pre-core/core proteins (p25, p21 or p17) also did not activate TCF/β-catenin transcription despite significant overlap in their amino acid sequence with p22. Clinical studies show HBe-positivity is a significant independent risk factor of HCC and fatality in chronic HBV-infected patients [40,41]. Furthermore, HBe is produced within the first week after HBV infection in experimental models [42], and thus p22 has the potential to contribute to early events in the transition to cancer. Here, we showed ectopically expressed p22 was localized diffusely in the cytoplasm and nucleus, and in cytoplasmic puncta, indicating potential co-localization with various levels of the Wnt signaling machinery [20]. We also demonstrated Fzd7 and GLUL are induced by p22 in vivo; this shows that genes associate with liver cancer (*Fzd7* [39]) and β-catenin-mediated liver zonation and regeneration (*GLUL,* [43]) are induced by p22 in normal hepatocytes. Furthermore, we demonstrated that p22 can increase TCF/β-catenin transcription on its own and in conjunction with ectopically expressed wild-type or mutant β-catenin; and in colon cancer cells with endogenous mutant *APC* (SW480 cells) or *CTNNB1* (HCT116 cells). Activation of TCF/β-catenin transcription in the SW480 cells by p22 was blocked by dnTCF4, confirming impact specifically on Wnt signaling.

Collectively, our data identifies HBV p22 as a novel regulator of Wnt signaling in the context of cancer and provides insight into the mechanisms of 'just right' Wnt signaling in cancer. Identifying the molecular interactors of p22 will not only be relevant to HCC but to all Wnt-addicted cancers as it is a new tool to investigate context-dependent Wnt signaling. Immunohistochemical studies in colon cancer carcinomas show variable β-catenin staining where β-catenin is primarily membrane-bound in central areas of the tumor, and intense cytoplasmic and nuclear staining in localized regions that are referred to as the invasive front associated with metastasis [44,45]. This implies that Wnt signaling is constrained in cancer cells allowing for bursts of intense signaling for various processes such as metastasis. It remains to be determined if this localized hyperactive Wnt signaling is due to loss of transcriptional repression or activation of transcription. Further investigation of the p22 mechanism of action in ex vivo models systems for example that do not have the limitations of continuous, transformed cell lines and mouse models with respect to human disease [27], might reveal novel avenues of research to help identify new components to selectively harness different aspects of Wnt signaling; for example, blocking oncogenic Wnt signaling while preserving the critical role Wnt signaling provides to ensure the correct regulation of stem cells and homeostasis of many epithelial tissues. Selective regulation of Wnt signaling is at the core of identifying druggable Wnt pathway targets, as the desired outcome for a cancer specific drug that inhibits Wnt is for the drug to allow normal physiological processes to proceed thus reducing the toxicity of a blanket approach of inhibiting Wnt signaling.

#### **4. Materials and Methods**

#### *4.1. Hydrodynamic Injection of Mice*

C57BL/6 mice used in experiments were between 6 and 10 weeks old, and age- and sex-matched (both sexes were used). Hydrodynamic injection (HDI) was performed as we previously described [29]. Briefly, unanesthetized mice were injected intravenously (iv) through the tail vein with 10 μg pCI-p22 or pCI-EV (pCI, Promega, Madison, WI, USA) in a volume of saline equivalent to 8% of the mouse body weight. The injection was performed within 5 s. Mice were killed 6- and 20-days post HDI, their liver resected and processed for analysis. The Walter and Eliza Hall Institute of Medical Research Animal Ethics Committee (AEC) reviewed and approved all animal experiments (AEC number 2017.016).

#### *4.2. RNA Extraction and Quantitative RT-PCR (qRT-PCR)*

Mouse liver tissues were homogenized in TRizol (Invitrogen, Carlsbad, CA, USA) and total RNA purified, DNAse treated and quantified as previously described [46]. cDNA was synthesized and subjected to qPCR using SYBR green (ABI). Gene expression was calculated relative to the housekeeping gene β2M (2−ΔΔCt) as described previously [46] and was expressed as fold change over empty vector (EV).

#### *4.3. Cell Lines and Wnt3a Conditioned Medium*

The human cell lines (SW480, HCT116, HEK293T and Huh7) were purchased from ATCC. SW480, HCT116 and HEK293T were maintained in RPMI-1640 supplemented with 20 mM HEPES, 10% (v/v) heat-inactivated fetal bovine serum (FBS), L-glutamine and antibiotics (penicillin and streptomycin). Wnt3a producing L-cells (L-3a) and the parental L-cells (L) were a generous gift from Prof Karl Willert [34]. L-3a, L and the Huh7 cells were maintained in DMEM, 10% (v/v) heat-inactivated FBS, supplemented with L-glutamine and antibiotics. Conditioned medium was prepared from L-3a and L cells in parallel as previously described [34].

#### *4.4. Transfection and Reporter Assays*

Cells were seeded into 24-well plates to reach 60–70% confluence overnight. Cells were transfected with 400 ng total plasmid (empty vector added to keep total plasmid constant) that included 100 ng sTOPflash or sFOPflash (generous gift from Prof Randall T Moon [17]); or 100 ng pGL or pGL-FZD7 promoter [23] and 2 ng *Renilla* luciferase plasmid (phRG-TK, Promega). The pDNA3.1 plasmids expressing β-catenin, ΔNβ-catenin and dnTCF4 were generous gifts from Professor Hans Clevers [18,22] and added at 100 ng/well. The pCI HBV protein expression plasmids were a generous gift from Professor Stephen Locarnini and added at 100 ng, unless indicated otherwise in the text. Cells were transfected using plasmids in OptiMEM (Life Technologies, Grand Island, NY, USA) and Lipofectamine LTX with Plus reagent (Invitrogen) according to the manufacturer's instructions. Cells were harvested 48 h later and analyzed using the dual luciferase reporter assay system (Promega). For Wnt3a stimulation, cells were treated with 200 μl L-3a or L conditioned medium for 6 h before harvesting in passive lysis buffer. Luciferase activity with control reporters sFOPflash and pGL, and L conditioned medium were negligible. Reporter activity was expressed relative to *Renilla* to the control for transfection efficiency and plotted as fold change over empty vector (EV) as previously described [38].

#### *4.5. Immunoblot Analysis*

Pre-cast 4–20% polyacrylamide gels (Mini-Protein TGX, Biorad, Hercules, CA, USA were used to separate proteins (Mini-Protein Tetra Cell, Biorad) and transferred onto nitrocellulose membranes using the Transblot-Turbo instrument (Biorad). The membranes were air-dried and blocked overnight in 1% skim milk at 4 ◦C. The following day, the membranes were incubated in primary antibody for 1 h and bound antibody detected with secondary antibody and ECL (Western Lightening Plus ECL, Perkin Elmer, Waltham, MA, USA). Primary antibody used were mouse anti-HBcAg [C1] (1:1000, Abcam ab8637, Cambridge, UK), mouse anti-αTubulin (1:1000, Abcam ab7291), mouse anti-active β-catenin (1:1000, Merck Millipore 05-665) and mouse anti-β-actin (1:5000, ThermoFisher AM4302, Waltham, MA, USA). Secondary antibody was rabbit anti-mouse polyclonal antibody HRP (1:10,000 Dako P0260, Glostrup, Denmark).

#### *4.6. Immunofluorescenc Confocal Microscopy*

Cells were seeded into two-well Nunc Lab-Tek (Thermofisher) chamber slides to reach 60–70% confluence overnight. Cells were transfected with 200 ng plasmid as described above. After 48 h, cells were fixed with 4% paraformaldehyde, permeabilized with 1% Triton-X100 and blocked with 1% FBS and stained with control antibody or anti-HBcAg [C1] (1:400, Abcam ab8637) primary antibody and detected with goat anti-mouse alexa fluor 488 (1:1000 Invitrogen A11029). DAPI (1:2000) was used for nuclear staining and the cells analyzed using Zeiss LSM700 as previously described [38].

#### *4.7. Statistical Analysis*

The data represent mean ± SEM, where *n* is at least three independent experiments with cell lines or tissue from at least three mice per cohort, unless stated otherwise. The Student *t*-test was used for comparisons and significance was defined as *p* < 0.5.

#### **5. Conclusions**

Mutations to *APC* and *CTNNB1* are the most frequent mutations in colon and liver cancer, respectively, and are thought to initiate cancer. Here we demonstrate that the HBV precore protein p22 can activate Wnt signaling in these cancer contexts. The ability of p22 to additionally activate Wnt signaling in the context of these mutations indicates oncogenic interplay between HBV infection and Wnt signaling in liver cancer. Furthermore, it is now clear that Wnt-addicted cancers harbor aberrations to Wnt signaling via both intracellular and cell-surface mechanisms [16], thus our findings identify HBV p22 as a novel tool to understand "additional" regulation and "fine-tuning" of Wnt signaling in the context of cancer [14,25]. Understanding the mechanisms that underly normal, wanted Wnt signaling and pathological, unwanted Wnt signaling is an important step for exploiting the Wnt pathway for anti-cancer treatment.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6694/12/6/1435/s1, Figure S1: Schematic of the HBV genome and the genes encoding various HBV proteins. The HBV genome, depicted as a long purple continuous strand, encodes 7 proteins from 4 open reading frames (ORFs) (surface [S], core [C], polymerase [P], and the X gene [X]), which are shown as large arrows in different colors, and 3 upstream regions [precore (preC), preS1, and preS2]. The transcripts, ORFs, gene regions and protein products are also shown on the right, Figure S2: Expression of protein from the indicated plasmids. Huh7 cells were transfected with the indicated plasmids and protein expression confirmed by immunoblot. Lysates prepared from Huh7 cells transfected with EV and the parental, un-transfected cells served as negative controls. Lysate from HBV core p21 transfected Huh7 cells was used as a positive control. (**a**) The membrane was stained with anti-HBc antibody first, then (**b**) re-probed with anti-tubulin antibody. The boxed areas were used for the cropped blots in Figure 1, Figure S3: Sub-cellular localization of HBV p22. The indicated expression plasmids were transfected into Huh7 and the cells subjected to confocal microscopy following staining with control anti-body and anti-HBV core antibody (red, while DAPI stained nuclei are blue). A higher magnification of the boxed area of the p22 transfected cells is also shown. Scale bars = 20 μM, Figure S4: HBV p22 stimulates Wnt signaling in Huh7 cells. Huh7 cells were co-transfected with 100 ng wild-type β-catenin and the indicated amounts of p22 plasmid and the cell lysates subjected to immunoblot for (**a**) active β-catenin. The membrane was stripped and re-probed with (**b**) anti-actin antibody. The boxed regions in (a) and (b) were used the cropped immunoblots in Figure 2d, Figure S5: Comparative reporter activity in Huh7 cells across the various conditions. The TOPflash and FOPflash reporter activities in Huh7 cells transfected with the indicated plasmids and treated with the indicated conditioned media [L-cell conditioned medium (CM) or L-cell-Wnt3a conditioned medium (Wnt3a CM)] are plotted on the same Y-axis to demonstrate the relative reporter activity between controls [(FOPflash, CM, empty vector (EV)] and test samples (TOPflash, expression plasmids, Wnt3a CM) and are shown as fold change reporter activity relative to FOPflash/EV (Mean ± SEM, Student t test, *n* = 3 experiments). Reporter activity in control samples was negligible, Figure S6: Quantitation of HBeAg levels in the supernatant of transfected Huh7 cells. HBeAg levels in the supernatant fluid of transfected cells were determined (**a**) two days and (**b**) three days after transfection using a commercial Roche anti-HBe kit and Cobas e411 instrument. Cells were transfected with increasing amounts of HBV p22-containing plasmid, from 0 - 200 ng per well, with or without co-transfected 100 ng wild type β-catenin (Mean ± SD, *n* = 3 replicate wells). Transfected p22 was processed to HBeAg and detected in the supernatant, confirming normal processing, Figure S7: Effect HBV p25 and p17 on Wnt signalling. Effect of increasing amounts of transfected HBV precore p17 (**a**) and p25 (**b**) expression plasmids on TCF/β-catenin transcription (sTOPflash reporter) in Huh7 cells co-transfected with 100 ng wild type β-catenin was determined and is shown relative to no p17 and p25, respectively (Mean ± SEM, \* *p* < 0.05, Student *t*-test, *n* = 3 experiments), Figure S8: HBV p22 upregulates gene expression in vivo. (**a**) Quantitative RT-PCR analysis of gene expression in livers of mice tail-vein-injected with EV or HBV p22 containing plasmids at 6 days post injection (mean ± SEM, \* *p* < 0.05, *n* = 7 and 8 for EV and p22 injected mice, respectively). (**b**) Quantitative RT-PCR analysis of gene expression in livers of mice tail-vein-injected with EV or p22 containing plasmids at 20 days post injection (mean ± SEM, \* *p* < 0.05, *n* = 4 and 5 for EV and p22 injected mice, respectively), Table S1: qRT-PCR Primer sequences.

**Author Contributions:** Conceptualization, B.M.T.; T.J.P.; and E.V.; formal analysis, B.M.T.; D.J.F.; T.J.P.; and E.V.; funding acquisition, D.J.F.; G.E.; N.W.; H.T.; T.F.; C.C.; J.T.; and E.V.; investigation, B.M.T.; D.J.F.; and G.E.; methodology, B.M.T.; D.J.F.; G.E.; T.J.P.; and E.V.; resources, N.W.; H.T.; T.F.; G.K.; C.C.; M.P.; and J.T.; supervision, E.V.; visualization, B.M.T.; D.J.F.; and E.V.; writing—original draft, B.M.T.; and E.V.; writing—review and editing, D.J.F.; N.W.; H.T.; T.F.; C.C.; M.P.; and T.J.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Melbourne Health through a project grant number PG-002-2016 awarded to E.V.; T.J.P.; G.E.; N.W.; T.F.; and C.C.; and a post-graduate scholarship to B.M.T.; E.V.; and T.J.P. were funded, in part, by grants from the National Health and Medical Research Council (NHMRC), project grant number APP1099302 and investigator grant number APP1181580. T.J.P. was funded by BLS/CMU Fellowship and MRC (MR/R026424/1). D.J.F.; was funded, in part, by a Cancer Council of Victoria fellowship and a Melbourne Health early career grant GIA-033-2016.

**Acknowledgments:** We thank Damian Neate, Danni Colledge and Jean Moselen for technical assistance. We also thank Randall T. Moon, Hans Clevers, Thomas Brabletz, Peter Revill, Stephen Locarnini and Karl Willert for gifting cell lines and plasmids; and the staff at the Walter and Eliza Hall Institute Biological Resource Facility (mice) and the Biological Optical Microscopy Platform (BOMP), University of Melbourne for their assistance.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review β***-Catenin Activation in Hepatocellular Cancer: Implications in Biology and Therapy**

**Yekaterina Krutsenko, Aatur D. Singhi and Satdarshan P. Monga \***

Department of Pathology and Pittsburgh Liver Research Center, University of Pittsburgh and University of Pittsburgh Medical Center, Pittsburgh, PA 15261, USA; yek14@pitt.edu (Y.K.); singhiad@upmc.edu (A.D.S.) **\*** Correspondence: smonga@pitt.edu; Tel.: +1-(412)-648-9966; Fax: +1-(412)-648-1916

**Simple Summary:** Liver cancer is a dreadful tumor which has gradually increased in incidence all around the world. One major driver of liver cancer is the Wnt–β-catenin pathway which is active in a subset of these tumors. While this pathway is normally important in liver development, regeneration and homeostasis, it's excessive activation due to mutations, is detrimental and leads to tumor cell growth, making it an important therapeutic target. There are also some unique characteristics of this pathway activation in liver cancer. It makes the tumor addicted to specific amino acids and in turn to mTOR signaling, which can be treated by certain existing therapies. In addition, activation of the Wnt–β-catenin in liver cancer appears to alter the immune cell landscape making it less likely to respond to the new immuno-oncology treatments. Thus, Wnt–β-catenin active tumors may need to be treated differently than non-Wnt–β-catenin active tumors.

**Abstract:** Hepatocellular cancer (HCC), the most common primary liver tumor, has been gradually growing in incidence globally. The whole-genome and whole-exome sequencing of HCC has led to an improved understanding of the molecular drivers of this tumor type. Activation of the Wnt signaling pathway, mostly due to stabilizing missense mutations in its downstream effector β-catenin (encoded by *CTNNB1*) or loss-of-function mutations in *AXIN1* (the gene which encodes for Axin-1, an essential protein for β-catenin degradation), are seen in a major subset of HCC. Because of the important role of β-catenin in liver pathobiology, its role in HCC has been extensively investigated. In fact, *CTNNB1* mutations have been shown to have a trunk role. β-Catenin has been shown to play an important role in regulating tumor cell proliferation and survival and in tumor angiogenesis, due to a host of target genes regulated by the β-catenin transactivation of its transcriptional factor TCF. Proof-of-concept preclinical studies have shown β-catenin to be a highly relevant therapeutic target in *CTNNB1*-mutated HCCs. More recently, studies have revealed a unique role of β-catenin activation in regulating both tumor metabolism as well as the tumor immune microenvironment. Both these roles have notable implications for the development of novel therapies for HCC. Thus, β-catenin has a pertinent role in driving HCC development and maintenance of this tumor-type, and could be a highly relevant therapeutic target in a subset of HCC cases.

**Keywords:** β-catenin mutations; tumor metabolism; tumor immunology; molecular therapeutics; precision medicine

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil-

**Citation:** Krutsenko, Y.; Singhi, A.D.; Monga, S.P. β-Catenin Activation in Hepatocellular Cancer: Implications in Biology and Therapy. *Cancers* **2021**, *13*, 1830. https://doi.org/10.3390/

Academic Editor: Claudio Tiribelli

cancers13081830

iations.

Received: 1 March 2021 Accepted: 9 April 2021 Published: 12 April 2021

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

### **1. The Wnt–**β**-Catenin Signaling Pathway**

The protein later termed Wnt1 was first identified almost 40 years ago in the context of its proto-oncogenic nature [1,2]. Subsequent studies have characterized Wnt1 itself, as well as other highly conserved components of Wnt signaling, as a key mediator involved not only in tumorigenesis, but also in the fundamental cellular processes governing embryonic development and adult tissue homeostasis [3,4]. Yet, the vital role of aberrant Wnt signaling in cancer initiation and progression remains one of the most intriguing and vital themes in the field. The Wnt pathway involves a multitude of components, including ligands, receptors, and co-receptors acting in autocrine, paracrine, and endocrine fashion to regulate the processes of cell fate determination, proliferation, and polarity, among others [2,4,5]. Structural and functional classification has indicated the existence of several distinct Wnt signaling pathways, which can be broadly subdivided based on the involvement of β-catenin. β-Catenin-dependent canonical Wnt signaling remains arguably the most investigated branch.

In the canonical pathway, the control of the Wnt-dependent cellular processes is achieved by a tight regulation of the amount of β-catenin—a transcriptional co-activator and a regulator of cell–cell adhesion. Normally, in the absence of Wnt signals, cytosolic levels of β-catenin remain low due to continuous proteasomal degradation of the protein, initiated by its destruction complex. The complex, composed of the scaffold Axin, tumor-suppressor adenomatous polyposis coli (*APC*) gene product, and diversin, also includes two kinases, casein kinase 1 (CK1), and glycogen synthase kinase 3 (GSK3), which sequentially phosphorylate β-catenin, priming it for recognition by the ubiquitin ligase β-TrCP [1]. In the absence of negative regulation, the glycosylation and palmitoylation of Wnt glycoproteins allows their biological activity to in turn activate the Wnt–β-catenin signaling. The cascade is induced by the binding of secreted Wnts to the seven transmembrane G-protein-coupled Frizzled (Fz) receptors located at the plasma membrane [5]. The binding initiates the formation of a multicomponent complex consisting of Wnt ligand, Frizzled, and its co-receptor LRP (low-density lipoprotein receptor-related protein) 6 or 5 [6]. This, in turn, signals for the recruitment of Dishevelled (Dvl), and results in the phosphorylation of LRP5/6, thereby providing a docking site for the Axin and tethering it to the cell membrane, which eventually renders the β-catenin destruction complex inactive. Thus, the presence of Wnt ligands interferes with the sequestration of β-catenin and its subsequent ubiquitination, thereby stabilizing the protein in cytoplasm. This allows for the nuclear translocation of β-catenin, where it triggers the expression of Wnt-induced genes (i.e., Cyclin D1, c-Myc, vascular endothelial growth factor (VEGF), interleukin-8 (IL-8), etc.) by acting as transcriptional co-activator in conjunction with T-cell factor (TCF) and lymphoid enhancer factor (LEF) DNA-binding proteins [7].

#### **2. Wnt–**β**-Catenin Signaling in Liver Pathophysiology**

The central role of the canonical Wnt–β-catenin signaling pathway in multiple aspects of normal cell functioning and in pathobiological processes is especially eminent in liver [8–11]. There, β-catenin orchestrates embryonic development, patterning, adult tissue metabolism, proliferation, and regeneration. While discussing the many facets of β-catenin signaling as a component of the Wnt pathway is outside the scope of the current review, we would like to remind the readers of a few pertinent concepts that are also relevant in hepatocellular cancer (HCC).

#### *2.1. Wnt–β-Catenin Signaling in Hepatic Development*

β-Catenin was first reported to be active in normal mouse and chick embryonic liver development almost two decades ago [12–14]. β-Catenin was seen to be active in stages of hepatic development which showed proliferating hepatoblasts and immature hepatocytes. When mouse embryonic liver cultures were propagated in the presence of antisense oligonucleotides against the β-catenin gene, there was a notable deficit in the resident cell proliferation. This was later verified by conditional deletion of the βcatenin gene or via activation of β-catenin through APC gene loss from mouse hepatoblasts in vivo [15,16]. In addition to these observations, both in vitro and in vivo studies showed a dramatic compromise in hepatocyte maturation. This was seen as the maintenance of hepatoblast markers in the hepatocytes in the β-catenin absent or knocked-down livers, as well as by deficient markers of mature fetal hepatocytes, including glycogen [16]. Thus, β-catenin plays a role in both the proliferation of immature hepatocytes and hepatoblasts during earlier stages of hepatic development, but plays an equally important role in the maturation of immature hepatocytes during later stages. These temporal targets of

β-catenin include c-myc and cyclin-D1 for proliferation, as well as CEBPα and as-yet unknown targets, which are likely distinct from its well-known zone-3 targets in adult liver [16]. It is also worth mentioning that, after birth, there is a postnatal growth spurt in livers from postnatal day 5 to about 25 days, after which the liver is mostly quiescent, showing minimal hepatocyte turnover [17]. β-Catenin signaling is also a major contributor of the postnatal wave of hepatocyte proliferation, and in its absence there is a decreased growth spurt which leaves liver-specific β-catenin knockout mice with around 15% lower liver-weight to body-weight ratio (LW/BW).

#### *2.2. Wnt–β-Catenin Signaling in Liver Regeneration*

Livers possess a unique feature of regeneration following surgical resection or toxicantinduced injury to regain its lost mass within days to weeks. The liver does so without any progenitor cell activation but via the replication of resident hepatocytes (and other cells) in the liver [18]. Wnt–β-Catenin signaling has been shown to be a key component of the normal molecular machinery of the liver following surgical resection [19]. Within hours of two-thirds hepatectomy, there is a nuclear translocation of β-catenin in hepatocytes and the appearance of β-catenin–TCF complex [20,21]. This is sustained for almost the first 48 h of regeneration. Using several genetic knockout mouse models, it appears that Wnt2 and Wnt9b are massively upregulated in hepatic sinusoidal endothelial cells and less so in monocytes/macrophages at 12 h after hepatectomy (earliest time point examined through individual cell-type isolation after surgery), followed by the engagement of Fzd-LRP5/6, resulting in the activation of β-catenin–TCF4 to regulate cyclin-D1 gene transcription [19]. The increased cyclin-D1 observed during this time allows for hepatocyte G1–S phase transition and eventually contributes to timely hepatocyte proliferation and the recovery of hepatic mass [22]. The absence of Wntless from endothelial cells (and less so macrophages) or the absence of LRP5 and 6 from hepatocytes or the absence of β-catenin from hepatocytes, all lead to a notable deficit in cyclin-D1 expression and a dramatically lower hepatocyte proliferation at 40–48 h after two-thirds hepatectomy [23–27]. Livers eventually recover in all models, despite a notable delay in restitution, and the mechanisms allowing for recovery in the absence of Wnt–β-catenin signaling remain unknown at this time. A similar role of the pathway during hepatocyte proliferation has also been reported after injury from acetaminophen, carbon-tetrachloride, diethoxycarbonyl dihydrocollidine, cholinedeficient ethionine supplemented diet, and in Mdr2 knockout mice, making Wnt–β-catenin signaling a global hepatic repair pathway [28–33].

Intriguingly, a recent study also showed an important role of the Wnt–β-catenin pathway in serving a dual role of not only inducing hepatocyte proliferation but also maintaining hepatocyte function during liver regeneration after surgical resection, as well as after acetaminophen-induced injury and repair. Using single-cell RNA-sequencing, Walesky et al. showed a clever "division of labor" by the hepatocytes in the remnant liver following surgery or toxicant injury [34]. This strategy allows liver to maintain function even while it is proliferating, as distinct subsets of hepatocytes acquire proliferative versus hepatocyte-function phenotype, as shown by gene expression studies. Intriguingly, both these functions are regulated by the Wnt–β-catenin pathway; the cell source of the Wnt for regulating the hepatocyte function by β-catenin appears to be macrophages and not sinusoidal endothelial cells, which are likely the source of Wnts for β-catenin activation in hepatocytes for proliferative function.

#### *2.3. Wnt–β-Catenin Signaling in Liver Zonation*

Another unique characteristic of the liver is the expression of unique genes by the hepatocytes based on their location within a microscopic hepatic lobule. This disparate gene expression allows for the hepatocytes to perform distinct functions that are necessary for the delivery of optimal hepatic output in terms of metabolism, synthesis, and detoxification, which are the broad categories of around 500 functions that hepatocytes perform to maintain health and homeostasis. Toward this end, Wnt–β-catenin signaling

is known to be the major regulator of the expression of genes in the zone-3 or pericentral region of the metabolic lobule [26,35,36]. These genes belong to the category of glutamine synthesis, glycolysis, lipogenesis, ketogenesis, bile acid synthesis, heme metabolism, and xenobiotic metabolism. Some of these target genes include Glul, which encodes glutamine synthetase (GS), and is specifically localized to 1–2 layers of hepatocytes around the central vein [37]. To prevent ammonia from leaving the liver, the zone-3 hepatocytes are efficient in its uptake and the high levels of GS in these cells are responsible for condensing ammonia to glutamate, leading to the formation of glutamine. Thus, intracellular levels of glutamine are highest in zone-3 hepatocytes. Some of the other key targets of β-catenin in zone-3 hepatocytes include Axin-2, Lect2, Cyp2e1, Cyp1a2, and others. Recently, choline transporter organic cation transporter 3 was also shown to be a target of the Wnt–β-catenin signaling, which led to the increased uptake of choline by HCC to promote phospholipid formation and DNA hypermethylation, and contributing to hepatocyte proliferation [38]. In fact, several of these β-catenin targets are upregulated in liver tumors where β-catenin signaling is highly activated in both preclinical models and in patients. Conversely, genetic knockout models that lack Wnt secretion from endothelial cells, lack LRP5 and 6 on hepatocytes, or lack β-catenin in hepatocytes, all lack zone-3 targets of the Wnt–β-catenin pathway [23,24,26,27,36]. Wnt2 and Wnt9b appear to be the major drivers of zonated β-catenin activation, and appear to be within the endothelial cells lining the central vein [39].

Thus, broadly, β-catenin seems to be playing a role in hepatocyte proliferation in physiological states including hepatic development (prenatal and postnatal) and liver regeneration (surgical and injury-driven), as well as in regulating hepatocyte functions including basally in the hepatocytes contained in zone-3 of the metabolic lobule. It is pertinent to mention the existence of regulators of the Wnt–β-catenin signaling that have been shown to play a role in the aforementioned hepatic processes. Factors like R-spondins and their receptors LGR4/5 have been shown to potentiate the effects of the Wnt–β-catenin pathways and have been specifically shown to positively impact the processes of both liver regeneration and metabolic zonation [40,41].

#### **3.** β**-Catenin as a Component of the Adherens Junctions in Liver Pathophysiology**

In addition to β-catenin being the major effector of Wnt signaling, it plays another evolutionarily conserved role at the adherens junctions (AJs), where it links the cytoplasmic tail of E-cadherin to α-catenin and F-actin [42]. Since the extracellular domain of E-cadherin of one cell binds to its counterpart on the next epithelial cell, the AJs are important mediators of intercellular adhesion. AJs are also present on hepatocytes, which are the predominant functioning epithelial cells of the liver. In fact, β-catenin and E-cadherin are mostly seen at the cell surface of hepatocytes. Immunohistochemistry is rarely sufficiently sensitive to detect β-catenin in cytoplasm or nuclei—even in zone-3 hepatocytes, where it is basally active. β-Catenin clearly associates with E-cadherin in the normal liver, and this association is likely part of maintaining junctional integrity, cell polarity, and epithelial identity, and plays a role in both cell adhesion in addition to providing some barrier function within this highly secretory and vascular organ.

#### *3.1. β-Catenin–E-Cadherin Complex in the Liver and Its Regulation*

The regulation of β-catenin at the AJs in the hepatocytes is not completely understood. There is an incomplete understanding of whether the same pool of β-catenin is allocated to Wnt signaling and AJs, of when and how this allocation occurs, and of how dynamic this process is [42]. The β-catenin–E-cadherin complex does not seem to be influenced by the Wnt signaling pathway. While liver-specific β-catenin knockout mice showed an absence of β-catenin–E-cadherin interactions, disruption of the Wnt–β-catenin signaling pathway did not impact this complex. This was evident when Wnt co-receptors LRP5/6 were conditionally deleted from hepatocytes, or when Wnt secretion was prevented from hepatic sinusoidal endothelial cells by loss of Wntless [24,27]. In both these models, βcatenin was intact at the AJs and was observed to be interacting with E-cadherin, thus maintaining cell–cell junctions and intact blood–bile barriers. This suggests that the absence of Wnt signaling does not impact the association. Interestingly, tyrosine phosphorylation of β-catenin, especially at tyrosine residue 654, has been shown to play an important role in negatively impacting β-catenin's association with E-cadherin [43]. Several receptor tyrosine kinases (RTKs), such as Src, EGFR, and Met, have been shown to phosphorylate β-catenin at these residues to negatively impact the AJ assembly, for which β-catenin tyrosine residues 654 and 670 have been shown to be important [44]. The fate of β-catenin following release from this complex is not completely clear, but may function as a coactivator for the TCF family, similar to its role in the Wnt signaling [45]. Indeed, RTKs like HGF and EGF can induce the nuclear translocation and activation of β-catenin signaling to cause liver growth, and can also be seen in a subset of tumors like hepatoblastomas and fibrolamellar HCCs [46–48]. Additionally, the cytoplasmic domain of E-cadherin in and around residues 685–699 contains several serine phosphorylation sites, and when these sites are phosphorylated, they interact extensively with armadillo repeats 3–4 of the β-catenin protein [49]. These phosphorylation events may be important in regulating β-catenin–E-cadherin interactions.

#### *3.2. γ-Catenin Compensates for β-Catenin at AJs in the Absence of β-Catenin*

Another important observation made in the livers of mice lacking β-catenin in hepatocytes was the maintenance of intact AJs. This coincided with an increase in γ-catenin or plakoglobin, a normal inhabitant of the desmosomes. Indeed, in the β-catenin knockouts, γcatenin was shown to co-precipitate and thus bind to E-cadherin in lieu of β-catenin [50,51]. This was also previously observed in skin and heart [52,53]. To demonstrate the true functionality of the γ-catenin interaction with E-cadherin in the absence of β-catenin, we conditionally knocked out both β- and γ-catenin from liver epithelia using albumin-cre. This led to a severe cholestatic disease, progressive fibrosis, and mortality, which was associated with perturbations in cell–cell junctions, paracellular leaks, and a decrease in E-cadherin [54]. Indeed, β-catenin binds to the region of E-cadherin which contains the PEST sequence motifs, which allow for the recognition of E-cadherin by ubiquitin ligases as well as proteasomal degradation [49]. The binding of β-catenin to E-cadherin masks these motifs and allows for uneventful trafficking of the complex to the AJs. It is likely that γ-catenin binds to the same region of E-cadherin when β-catenin is absent, preventing E-cadherin degradation and successful delivery of the E-cadherin–γ-catenin complex to the cell surface. This also explains the notable decrease in E-cadherin in the β-γ-catenin double-knockout livers [54].

#### **4. Hepatocellular Cancer**

#### *4.1. Alarming Trends in HCC Incidence*

The incidence of hepatocellular carcinoma (HCC) has risen steadily in the US and worldwide over last decades [55,56]. Analysis of the NCI's (National Cancer Institute's) Surveillance, Epidemiology and End Results (SEER) database reveals alarming trends in HCC incidence. The rates for new liver and intrahepatic bile duct cancer cases have been rising on average 2.7% each year over the last 10 years. Death rates have risen on average 2.6% each year from 2005 to 2014. In 2014, there were an estimated 66,771 people living with liver tumors in USA. In 2020, liver tumors represented 2.4% of all new cancer cases in the US, with around 42,810 new diagnosed cases [55]. Globally, HCC is the 5th most common malignancy in men, 9th most common cause of cancer in women, and the overall 6th most common cancer worldwide [56].

#### *4.2. Cellular and Molecular Pathogenesis of HCC*

Most HCCs are a consequence of years of hepatic damage and wound healing. The events leading up to HCC are complex and involve bouts of cell injury and death, immune cell infiltration, oxidative stress, and stellate cell activation [57]. The liver tries to replace the dying hepatocytes through chronic regeneration via hepatocyte proliferation. Proliferating hepatocytes are susceptible to DNA damage and mutations, and the associated activation of signaling pathways. Any such alterations that provide proliferative and survival advantage to a cell lead to the initiation of the neoplastic transformation. Transcriptomic and whole-genome sequencing has validated that subsets of HCC are "driven" by key oncogenic signaling pathways [58–60]. The whole-exome sequencing of a large number of HCC cohorts has revealed common mutations that are the basis of the molecular classification of HCC [59]. Such analysis has revealed that irrespective of etiology, chronic injury, and downstream cellular events, HCC is driven by a few common genetic aberrations and molecular pathway activation, with only some preferential signaling evident in a few etiologies [60]. One common pathway activated in HCC independent of etiology is Wnt–β-catenin signaling.

#### **5.** β**-Catenin and Hepatocellular Cancer**

#### *5.1. Mechanism of β-Catenin Activation in HCC*

It is important to emphasize the key phosphorylation sites located in exon-3 of βcatenin, which are important in its eventual degradation. When the Wnt signals are absent, β-catenin is sequentially phosphorylated at serine-45 (S45), S33, S37, and threonine-41 (T41) by casein kinase I (CKI) and glycogen synthase kinase 3β (GSK3β) [61]. Phosphorylated β-catenin is recognized by β-transducin repeat-containing protein for ubiquitination and proteasomal degradation, and requires intact D32 and G34 sites [62]. When Wnt signaling is on, it inactivates the β-catenin degradation complex consisting of Axin-1 and adenomatous polyposis coli gene product (APC) in addition to CKI and GSK3β. Around 26–37% of all HCCs display *CTNNB1* mutations [8,63]. These missense mutations are localized to exon-3 of CTNNB1, the gene encoding for β-catenin, and affect phosphorylation and ubiquitination sites in the β-catenin promoter, making it resistant to degradation. This leads to β-catenin stabilization, nuclear translocation, and activation of the downstream target genes, playing important and unique roles in tumor biology in several subsets of HCC cases. There are several targets of β-catenin reported in HCC [8]. Some highly relevant ones include glutamine synthetase (GS), cyclin-D1, VEGF-A, lect2, Axin-2, and others.

Loss-of-function mutations in *AXIN1* are another major contributor to HCC development. *AXIN1* is also among the top five mutated genes in HCC, seen in around 8% of human HCCs. This gene normally encodes for a protein essential for β-catenin degradation. In the absence of a functional Axin-1, β-catenin levels are increased and Wnt signaling is activated. Indeed, in preclinical models which used sleeping beauty transposon/transposase to express shRNA-*Axin1* along with Met proto-oncogene in either a hepatic β-catenin-sufficient or deficient liver, the requirement of β-catenin was unequivocally shown in this model [64]. Intriguingly, only a subset of targets of the β-catenin signaling are positive in these tumors, including cyclin-D1 and c-myc, and interestingly, these tumors are GS-negative.

Analysis of early HCC, multinodular HCC, and comparison of primary and metastatic HCC has also indicated that β-catenin has a trunk role in HCC similar to other major drivers such as mutations in *TERT* promoter or *TP53* [65].

#### *5.2. Animal Models to Study β-Catenin Activation in HCC*

The hepatic overexpression of β-catenin or the expression of mutated, constitutivelyactive β-catenin alone is insufficient for HCC development, as reported in many mouse models, suggesting cooperation with other pathways [37,66]. Indeed, *CTNNB1* mutations significantly correlate with the presence of other mutations such as in *TERT* promoter, *NFE2L2, MLL2, ARID2,* and *APOB* [59,67]. *CTNNB1* mutations also are seen to co-occur with the overexpression/activation of Met, Myc, or Nrf2 [63,68,69]. Using a reductionist approach, such concomitant alterations have been modeled in mice by the co-expression of various combinations in vivo using the sleeping beauty transposon/transposase or CRISPR/Cas9 approach and hydrodynamic tail vein injection [70]. For example, 11% of human HCCs show concomitant *CTNNB1* mutations and Met overexpression/activation, and their co-expression

in murine liver in the Met–β-catenin model leads to clinically relevant HCC [63,71]. Likewise, Myc–β-catenin represents 6% of human HCCs [68]; and Nrf2–β-catenin represents 9–12% of HCC [69]. The continued generation and characterization of these models for their clinical relevance, biology, and for testing therapies, is of high value.

#### **6. State of Therapies for HCC**

The five-year survival of liver tumors is 19.6%, attributable partially to lack of effective therapies [55]. For localized disease, partial hepatectomy or liver transplantation are most beneficial. Loco-regional therapies like radio frequency ablation and transarterial chemoembolization are palliative or useful as neoadjuvants. Until recently, sorafenib was the only FDA-approved agent for unresectable HCC, and this non-specific tyrosine kinase inhibitor (TKI) improved survival by 3 months [72]. Several agents have been approved for HCC treatment in the last 5 years. Regorafenib was approved as second-line treatment, showing improvement in survival to 10.6 months vs. 7.8 months for placebo [73]. In 2017, the immune checkpoint inhibitor (ICI) nivolumab was approved by the FDA as second-line treatment, almost doubling overall survival to 15 months in the Checkmate trial [74]. More recently, another TKI, lenvatinib, was approved as first-line therapy, showing non-inferiority to sorafenib [75]. Cabozantinib, a Met inhibitor, also got approval for second-line use in HCC [76]. Another ICI, pembrolizumab, was awarded an accelerated approval as second-line therapy for HCC based on the KEYNOTE-224 trial [77]. More recently, the results of a phase III clinical trial (IMbrave150) showing higher efficacy to sorafenib and a response rate of around 35% to atezolizumab (anti PD-L1) plus bevacizumab (anti-VEGFA) led to their FDA approval as first-line therapy [78]. Some major existing challenges include a lack of biomarker-based therapy to select a proper subset of patients for a specific treatment and to improve response rates to ICIs, which have revolutionized oncology in general.

#### **7. Targeting** β**-Catenin for HCC Treatment**

#### *7.1. Proof-of-Concept Studies*

Because β-catenin is active in a notable subset of HCCs, and is also considered a trunk mutation, its inhibition could have a major impact on the treatment of a subset of these tumors. Several proof-of-concept studies in HCC, both in vitro and in vivo, have demonstrated the relevance of inhibiting β-catenin as a treatment strategy for HCC. siRNA-mediated *CTNNB1* knockdown resulted in a marked decrease in the viability and proliferation of human hepatoma cells in vitro [79]. Similarly, suppressing β-catenin via gamma-guanidine-based peptide nucleic acid antisense also reduced the viability, proliferation, metabolism, and survival of cells of an HCC line [80]. Interestingly, inhibition of β-catenin signaling also resulted in the diminished secretion of angiogenic factors, implying the dual positive effect of such suppression [80]. The DsiRNAs-mediated knockdown of β-catenin mRNA led to a significant decrease of tumor burden in mice bearing ectopic tumors originating from either Hep3B or HepG2 cells [81]. Using a chemical carcinogen (diethylnitrosamine) and tumor promotion (phenobarbital) model which selectively leads to *Ctnnb1*-mutation-driven HCC, β-catenin inhibition using locked nucleic acid antisense (LNA) had a profound impact on tumor development [82]. More recently, using Kras–β-catenin-driven HCC (which highly resembles the Met–β-catenin model), β-catenin was inhibited using EnCore lipid nanoparticles loaded with a Dicer substrate small interfering RNA targeting *CTNNB1*. This led to a notable decrease in tumor burden, also demonstrating β-catenin to be a highly relevant target in HCC for cases driven by *CTNNB1* mutations.

#### *7.2. Where to Target Wnt–β-Catenin Signaling in HCC*

The most important mechanism of β-catenin activation in HCC are the mutations in *CTNNB1* and the mutations in *AXIN1*. While there have been several other mechanisms identified to modulate β-catenin signaling, including the upregulation of certain Wnt genes, Frizzled genes, and epigenetic loss of negative regulators like DKK and FRPs and others, their true relevance remains unclear since Wnt–β-catenin signaling, like other signaling

pathways, is able to regulate its overall activity via robust post-translational mechanisms. However, mutations in *CTNNB1* or *AXIN1* deem the β-catenin protein non-degradable and hence cannot be regulated by the normal mechanisms, which converge on β-catenin degradation to control the signaling pathway activity. This also suggests that several classes of Wnt inhibitors will not work in HCCs because they inhibit or impair Wnt activity upstream of the observed mutations in *CTNNB1* or *AXIN1.* Hence, the suppression of β-catenin itself using the RNA-based therapies discussed in the preceding section, or those impairing β-catenin nuclear translocation, impairing its interaction with TCF4 or preventing the β-catenin–TCF complex from transactivating target genes, would be most effective in treatment of some subsets of HCC. Finally, identifying unique opportunities related to β-catenin signaling in HCC is important, as it may help in selecting or excluding the right group of patients, and may help to identify innovative opportunities to target other mechanisms that are intimately related to β-catenin activation unique to HCC.

#### *7.3. How to Target β-Catenin in HCC*

Targeting β-catenin itself using RNA-based therapies is highly desirable. Several classes of siRNA- and antisense-based therapies have been described for use against βcatenin. The use of EnCore lipid nanoparticles along with Dicer substrate small interfering RNAs is especially innovative because it can be modified to specifically deliver the payload to liver tumors, and the safety of their use has been shown in patients [83]. Others such as peptide nucleic acid antisense, locked nucleic acid antisense, and other modalities have been reported, and may have eventual clinical use [80–82].

There may be an opportunity to identify the mechanisms of the nuclear transport or nuclear export of β-catenin. Targeting molecules that cargo β-catenin to the nucleus or activate its export out of the nucleus could have efficacy in the treatment of β-cateninmutated HCCs. Pegylated interferon-α2a (peg-IFN), previously a first-line therapy for hepatitis C virus (HCV) patients, was shown to induce the levels of Ran-binding protein 3 (RanBP3), which is known to export β-catenin out of the nucleus [84]. Peg-IFN treatment was also shown to induce association between RanBP3 and β-catenin, and led to decreased TopFlash reporter activity that was abrogated by siRNA-mediated RanBP3 knockdown. In vivo, peg-IFN treatment led to increased nuclear RanBP3, decreased nuclear β-catenin and cyclin D1, and decreased GS, and eventually led to decreased tumor cell proliferation.

The use of small-molecule inhibitors that interfere with its interactions with TCF or other relevant co-factors or components of the transcriptional complex would be highly desirable. However, a high specificity of the small-molecule inhibitors will be required because of the overlap of the β-catenin–TCF4 binding site, and with the binding sites for APC and E-cadherin [85]. Even though a number of the identified compounds showed selectivity of inhibition in vitro (e.g., PKF115-584, CGP049090, and PKF118-310), none of them has entered clinical trials [85]. PR1-724, the next-generation compound of the original small-molecule ICG-001, interferes with β-catenin–TCF4 interactions with CBP, a histone acetyltransferase essential for transcriptional function of the complex [86,87]. PRI-724 has been shown to be safe in patients with HCV-related cirrhosis, and may be of high relevance in the treatment of subsets of HCC with known mutations in *CTNNB1* [88].

#### *7.4. Unique and Exploitable Aspects of Targeting β-Catenin in Subsets of HCC*

In addition to a general role of β-catenin in regulating tumor cell proliferation, survival, and angiogenesis, there are specific and unique aspects of β-catenin activation due to mutations in HCC which can have notable biological and therapeutic implications—especially related to a step towards precision medicine.

#### 7.4.1. Role of β-Catenin in Tumor Immune Evasion

ICIs have revolutionized the treatment of many tumors, including HCC as can be seen by the FDA approval of nivolumab and pembrolizumab as second-line therapy and of atezolizumab (anti PD-L1) plus bevacizumab (anti-VEGFA), as first line treatment for

unresectable HCC [78]. However, there are no available biomarkers which predict either the efficacy or lack thereof to ICIs. Clinical response to ICIs, most of which are T-cellbased therapies, depend on the presence of a CD8+ T cell inflamed environment and chemokines and interferon signature within the tumor [89]. Intriguingly, activation of βcatenin signaling has been linked to immune evasion in tumors such as melanoma through T-cell exclusion from tumors [90]. This is shown in our own analysis as well (Figure 1). Several mechanisms underlie this observation, including the effect of β-catenin activation on CD8<sup>+</sup> T cell priming and infiltration by acting on Batf3-lineage CD103<sup>+</sup> dendritic cells (DCs) and decreasing CCL4 production by inducing the expression of transcription repressor ATF3 [91]; disruption of Foxp3 transcriptional activity, key for development and function of regulatory T cells [92]; and increased Treg survival, which can reduce CD8<sup>+</sup> T cell proliferation [93]. HCCs with β-catenin activation have been linked to immune cell exclusion [94,95]. We have shown that *CTNNB1*-mutated HCCs are resistant to anti-PD-1 [68], and hence may benefit from the inhibition of β-catenin or its downstream effectors to sensitize these tumors to ICIs.

**Figure 1.** β-Catenin activation in HCC reduces CD8 T cell infiltration in the tumor. The top panel shows histology of explanted liver for hepatocellular cancer (HCC) showing the presence of two distinct tumors (separated by a dotted line) which are otherwise difficult to distinguish and demarcate by hematoxylin and eosin (H&E) staining (100×). The middle panel shows the immunohistochemistry of the adjacent tissue section to the top panel, for glutamine synthetase (GS), a surrogate marker of β-catenin activation due to mutations in *CTNNB1*. The staining for GS shows the presence of uniform positive staining in the upper-right part which decorates a β-catenin-active HCC, whereas the lower-left tumor is negative for this stain. Immunohistochemistry for CD8 for a subset of T cells, in the section adjacent to those shown in the top and middle panels, shows a general dearth of positive cells in the top-right (β-catenin-active) tumor, while there are notably more CD8-positive cells in the lower left or in the non-β-catenin-active HCC. The two tumors are separated by a dotted line.

One additional relevant mechanism in the liver might be through a known interaction of β-catenin with NF-κB in the hepatocytes and liver tumor cells [96]. This inhibitory association between the p65 subunit of NF-κB and β-catenin prevents NF-κB activation even when appropriate upstream effectors of NF-κB are present. In this study, we also showed that this association led to reduced p65-luciferase reporter activity when constitutively active β-catenin was transfected in hepatoma cells. Furthermore, β-catenin

mutated HCCs showed decreased p65 nuclear translocation. Knowing that NF-κB signaling plays a major role in inducing inflammatory milieu [97], its suppression brought about by stable β-catenin due to mutations in *CTNNB1* may be one additional contributor of an immune-deficient tumor microenvironment which may in turn lead to resistance to ICIs.

#### 7.4.2. Role of β-Catenin in Regulating Tumor Metabolism Through mTORC1 in HCC

The suppression of β-catenin in *CTNNB1*-mutant liver tumors decreases tumor burden in many models [71,82]. We made a unique discovery of how this response was mediated by the regulation of mTORC1 by β-catenin [98]. The Wnt–β-catenin pathway transcriptionally regulates the expression of *Glul*, which encodes GS in hepatocytes in zone-3 of the hepatic lobule [37], and leads to the highest glutamine in zone-3 hepatocytes [99] (Figure 2A,B). Glutamine directly phosphorylates mTOR at serine-2448 in lysosomes [100]. We identified p-mTOR-S2448 (active mTORC1) [101] in zone-3 hepatocytes basally, which was absent in hepatocyte-specific knockout (KO) of β-catenin, Wnt co-receptors LRP5-6, and GS (Figure 2B). We also found by immunohistochemistry (IHC) that HCCs with *CTNNB1* mutations are simultaneously positive for GS and p-mTOR-S2448 in preclinical models and patients (Figure 2B). We also showed a dependence of the CTNNB1-mutated HCCs to mTORC1 by their susceptibility to mTOR inhibition by rapamycin in a preclinical model. This may be a novel way to target β-catenin mutated liver tumors in patients until anti-β-catenin therapies become a reality.

**Figure 2.** Unique mTORC1 addiction of *CTNNB1*-mutated HCCs due to glutamine. (**A**) The unique axis of mTORC1 activation in β-catenin gene mutated HCCs due to overexpression of *GLUL,* the gene encoding for glutamine synthetase (GS), which generates glutamine from ammonia and glutamate, and in turn glutamine activates mTORC1 in lysosomes. (**B**) The top panel shows immunohistochemistry for GS and p-mTOR-S2448 in adjacent sections from a normal mouse liver. Both proteins are localizing exclusively to zone-3 hepatocytes in the immediate proximity to the central vein (200×). The whole slide scans (middle row) of two adjacent tissue microarrays of human HCC samples stained for the same antibodies against GS and p-mTOR-S2448 also shows several HCCs to be simultaneously positive for GS and p-mTOR-S2448. A representative tissue array sample is magnified (400×) to show GS and p-mTOR-S2448-positive HCC (bottom panels).

**Funding:** This work was supported in part by NIH grants 1R01DK62277, 1R01DK116993, 1R01CA204586, 1R01CA251155, R01CA250227, and Endowed Chair for Experimental Pathology to S.P.M. The tissue samples for analysis were provided by the Clinical Biospecimen Repository and Processing Core of the Pittsburgh Liver Research Center and supported by 1P30DK120531.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of the University of Pittsburgh (STUDY19070068, STUDY20010114, and STUDY20040276 on 3/23/2021).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** S.P.M. has research funding from Vicero Inc., Revolution Medicines, and ALIGOS Therapeutics. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**

