**Targeting the CBP**/β**-Catenin Interaction to Suppress Activation of Cancer-Promoting Pancreatic Stellate Cells**

**Mingtian Che 1, Soo-Mi Kweon 1, Jia-Ling Teo 1, Yate-Ching Yuan 2, Laleh G. Melstrom 3, Richard T. Waldron 4,5, Aurelia Lugea 4,5, Raul A. Urrutia 6, Stephen J. Pandol 4,5 and Keane K. Y. Lai 7,8,\***


Received: 2 May 2020; Accepted: 2 June 2020; Published: 5 June 2020

**Abstract:** Background: Although cyclic AMP-response element binding protein-binding protein (CBP)/β-catenin signaling is known to promote proliferation and fibrosis in various organ systems, its role in the activation of pancreatic stellate cells (PSCs), the key effector cells of desmoplasia in pancreatic cancer and fibrosis in chronic pancreatitis, is largely unknown. Methods: To investigate the role of the CBP/β-catenin signaling pathway in the activation of PSCs, we have treated mouse and human PSCs with the small molecule specific CBP/β-catenin antagonist ICG-001 and examined the effects of treatment on parameters of activation. Results: We report for the first time that CBP/β-catenin antagonism suppresses activation of PSCs as evidenced by their decreased proliferation, down-regulation of "activation" markers, e.g., α-smooth muscle actin (α-SMA/Acta2), collagen type I alpha 1 (Col1a1), Prolyl 4-hydroxylase, and Survivin, up-regulation of peroxisome proliferator activated receptor gamma (Ppar-γ) which is associated with quiescence, and reduced migration; additionally, CBP/β-catenin antagonism also suppresses PSC-induced migration of cancer cells. Conclusion: CBP/β-catenin antagonism represents a novel therapeutic strategy for suppressing PSC activation and may be effective at countering PSC promotion of pancreatic cancer.

**Keywords:** pancreatic cancer; pancreatic stellate cells; Wnt signaling; CBP; p300; pancreatitis; fibrosis

#### **1. Introduction**

Pancreatic cancer, predominantly comprised of pancreatic ductal adenocarcinoma (PDAC), ranks as the 4th leading cause of cancer deaths in both men and women in the United States, with ~52% of pancreatic cancer patients being diagnosed at an advanced stage of disease for which 5-year survival is a dismal 3% [1]. As such, there is an urgent need for treatments that offer durable benefits to PDAC patients. Treatments for PDAC, which traditionally have focused on targeting pancreatic tumor

cells (i.e., parenchymal cells), have been insufficient or have failed for the most part [2,3]. More recently, it has been recognized that activated pancreatic stellate cells (PSCs) (i.e., stromal cells), which are characterized by increased proliferation, up-regulation of "activation" markers, and enhanced migration, promote PDAC progression [2–4]. Moreover, the "desmoplasia" of PDAC, i.e., the extensive pro-tumorigenic fibrosis effected by activated PSCs [2–5], has been found to correlate negatively with patient survival and to be present at similar levels in both primary tumors and metastatic lesions [6]. Thus, activated PSCs represent an attractive therapeutic target to aid in combating PDAC.

The Wnt signaling pathway is an incredibly complex and critical controller of a myriad of processes in mammals, intricately regulating cellular proliferation and cellular differentiation [7,8]. "Canonical" Wnt signaling (or Wnt/β-catenin signaling) is the arm of the pathway associated with β-catenin accumulation in the nucleus and β-catenin forming a complex with members of the TCF/LEF family of transcription factors to regulate target gene transcription. A previous study has shown that retinoic acid-mediated suppression of Wnt/β-catenin signaling suppresses PSC activation in mice with chronic pancreatitis, leading to amelioration of chronic pancreatitis (which itself is a predominant risk factor for PDAC [9,10]) and associated fibrosis [11]. It has been previously demonstrated that, in Wnt/β-catenin signaling, β-catenin recruits either of the Kat3 transcriptional coactivators, cyclic AMP-response element binding protein-binding protein (CBP) or its closely related homolog, p300, to effect transcription and expression of respective target genes [12–15]. CBP/β-catenin signaling is associated with symmetric non-differentiative proliferation in cancer and fibrosis, whereas p300/β-catenin signaling initiates differentiation and a decrease in cellular potency [12–15]. Recently, it has been reported that the small molecule specific CBP/β-catenin antagonist ICG-001 [16] suppresses the activation of hepatic stellate cells, which are developmentally and functionally analogous to PSCs [17,18], as well as suppressing associated fibrogenesis in an acute CCl4-induced liver injury mouse model [19]. However, the significance of the CBP/β-catenin signaling pathway in PSCs is largely unknown. Thus, the CBP/β-catenin signaling pathway represents a potentially viable, but not yet characterized therapeutic opportunity to target in activated PSCs.

Based on the aforementioned observations, we set out to investigate whether antagonizing the CBP/β-catenin signaling pathway would suppress activation of PSCs and may be useful for combating PDAC and chronic pancreatitis. Herein, we report for the first time that the small molecule specific CBP/β-catenin antagonist ICG-001 suppresses activation of PSCs as evidenced by their decreased proliferation, down-regulation of activation markers, e.g., α-smooth muscle actin (α-SMA/Acta2), collagen type I alpha 1 (Col1a1), Prolyl 4-hydroxylase, and Survivin, up-regulation of peroxisome proliferator activated receptor gamma (Ppar-γ), which is associated with quiescence, and reduced migration; and that migration of PDAC cells is reduced when co-cultured with PSCs which have been pre-treated with ICG-001. Hence, CBP/β-catenin antagonist ICG-001 represents a novel therapeutic option for suppressing PSC activation and promotion of PDAC.

#### **2. Results**

#### *2.1. CBP*/β*-Catenin Antagonism Suppresses Proliferation of Pancreatic Stellate Cells*

Activated pancreatic stellate cells (PSCs) are known to promote pancreatic ductal adenocarcinoma (PDAC) progression and are characterized by increased proliferation [2–4]. To investigate whether inhibition of CBP/β-catenin signaling would suppress proliferation of PSCs, the small molecule specific CBP/β-catenin antagonist ICG-001 [16] versus control (DMSO) was used to treat immortalized mouse PSC line (imPSC) and immortalized human PSC line (ihPSC), which were established as previously described [20,21]. We found that ICG-001 inhibited proliferation of imPSC and ihPSC, as assessed by CellTiter-Glo proliferation assay (Figure 1A,B), microscopy (Figure 1C,D), and cell counting (Figure 1E,F). In addition, ICG-001 IC50 of ~25 μM and ranging from ~5 to ~25 uM, for imPSC and ihPSC, respectively, were estimated based on CellTiter-Glo proliferation assay (Figure 1A,B). Furthermore, ICG-001 treatment induces imPSC and ihPSC to change from a more spread out morphology to a thinner

or more round, quiescent morphology (Figure 1C,D). Collectively, our results demonstrate that ICG-001 suppresses activation of PSC by inhibiting proliferation and inducing quiescent morphology.

**Figure 1.** Cyclic AMP-response element binding protein-binding protein (CBP)/β-catenin antagonism suppresses proliferation of pancreatic stellate cells. Effect of CBP/β-catenin antagonist ICG-001 versus control (DMSO) treatment for 48 h on proliferation of immortalized mouse pancreatic stellate cells (imPSC) (**A**) and immortalized human pancreatic stellate cells (ihPSC) (**B**) as assessed by CellTiter-Glo assay. imPSC, immortalized mouse pancreatic stellate cells; ihPSC, immortalized human pancreatic stellate cells. Effect of ICG-001 treatment for 48 h on proliferation of imPSC (**C**) and ihPSC (**D**) as assessed by microscopy. Scale bar: 150 μm. Effect of ICG-001 treatment for 24 h and 48 h on proliferation of imPSC (**E**) and ihPSC (**F**) as assessed by cell counting. *n* = 3, \* *p* < 0.05 when each ICG-001 group compared to control (DMSO) at respective time point.

#### *2.2. CBP*/β*-Catenin Antagonism Suppresses Activation Markers of PSCs*

We next tested whether inhibition of CBP/β-catenin signaling would suppress established activation markers of PSCs, such as *Acta2*, *Col1a1*, and *Survivin (Birc5)* [2–4,22–24] at the level of mRNA expression. We found that CBP/β-catenin antagonist ICG-001 versus control (DMSO) suppressed in a dose dependent manner, Acta2, Col1a1, and Survivin (Birc5) mRNA expression by up to ~60%, 70%, and 50%, respectively, in imPSC (Figure 2A–C), as assessed by qPCR. Consistent with these findings, we found that ICG-001 also induced mRNA expression of Ppar-γ, which is associated with PSC quiescence [4,22,25,26], up to ~2.1-fold in imPSC. ICG-001 also suppressed *COL1A1* and *SURVIVIN*

*(BIRC5)* mRNA expression by up to ~75% and 90%, respectively, in ihPSC, but interestingly *ACTA2* mRNA expression was induced up to ~1.9-fold, whereas PPAR-γ mRNA expression was not detected (Ct value > 35) (Figure 2E–G).

**Figure 2.** CBP/β-catenin antagonism suppresses gene expression of activation markers of pancreatic stellate cells. Effect of CBP/β-catenin antagonist ICG-001 versus control (DMSO) treatment for 48 h of immortalized mouse pancreatic stellate cells (imPSC) on mRNA expression of activation markers, Acta2 (**A**); Col1a1 (**B**); Survivin (**C**); and Ppar-γ which is associated with quiescence (**D**). Effect of ICG-001 treatment for 48 h of immortalized human pancreatic stellate cells (ihPSC) on mRNA expression of activation markers, ACTA2 (**E**); COL1A1 (**F**); and SURVIVIN (**G**). *n* = 3, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 compared to control (DMSO), # *p* < 0.05, ## *p* < 0.01, ### *p* < 0.001 compared to ICG-001 5 μM, & *p* < 0.05, && *p* < 0.01compared to ICG-001 10 μM.

We next assessed the effect of ICG-001 on PSC activation and quiescence markers at the level of protein expression, by immunofluorescence or immunoblot. We found that ICG-001 reduced the expression of Acta2 (α-SMA) and Survivin and induced the expression of Ppar-γ in imPSC (Figure 3A,B,D), and similar results were obtained for SURVIVIN and PPAR-γ in ihPSC (Figure 3C,E), as assayed by immunofluorescence. Interestingly, α-SMA was not detected by immunofluorescence in ihPSC, consistent with previous findings, which failed to detect α-SMA at the protein level in this cell line (data not shown) and another immortalized human PSC cell line [27]. Consistent with the immunofluorescence results, ICG-001 reduced the expression of α-SMA by up to ~40% in imPSC (Figure 4A). Given that Prolyl 4-hydroxylase is a central enzyme in the hydroxylation of proline residues in procollagen, serving as a functional indicator of collagen synthesis and thus as another PSC activation marker [26,28], we next tested the effect of ICG-001 on the expression of this marker at the protein level. Immunoblot shows that ICG-001 suppresses Prolyl 4-hydroxylase (P4HA2) by up to ~50% in imPSC (Figure 4B). Thus, based on these protein expression data, along with the aforementioned mRNA expression data, we conclude that CBP/β-catenin antagonism suppresses activation and induces quiescence markers of PSCs.

**Figure 3.** CBP/β-catenin antagonism suppresses protein expression of activation markers of pancreatic stellate cells as assessed by immunofluorescence. Effect of CBP/β-catenin antagonist ICG-001 versus control (DMSO) treatment for 72 h of immortalized mouse pancreatic stellate cells (imPSC) on protein expression of activation markers, Acta2 (α-SMA) (**A**); and Survivin (**B**); and Ppar-γ which is associated with quiescence (**D**); Effect of ICG-001 treatment for 72 h of immortalized human pancreatic stellate cells (ihPSC) on protein expression of activation markers, SURVIVIN (**C**); and PPAR-γ (**E**); Scale bar: 100 μm. Hoechst: Hoechst 33342.

**Figure 4.** CBP/β-catenin antagonism suppresses protein expression of activation markers of pancreatic stellate cells as assessed by immunoblot. Effect of CBP/β-catenin antagonist ICG-001 versus control (DMSO) treatment for 72 h of immortalized mouse pancreatic stellate cells (imPSC) on protein expression of activation markers, Acta2 (α-SMA) (**A**) and Prolyl 4-hydroxylase (P4HA2) (**B**). Numerical values below protein bands indicate densitometric quantitation normalized to Ponceau S or GAPDH as indicated and then to control (DMSO). Numerical values and associated horizontal marks to the left of protein bands indicate relative position of molecular weight (kDa) markers. (Whole immunoblots are presented in Figure S1.).

#### *2.3. CBP*/β*-Catenin Antagonism Suppresses Migration of PSCs and PSC-Induced Migration of Cancer Cells*

Next, we tested whether inhibition of CBP/β-catenin signaling would suppress PSC migration which is another established characteristic of activated PSCs [2–4]. To do so, we treated imPSC with CBP/β-catenin antagonist ICG-001 versus control (DMSO) and found that ICG-001 treatment suppressed migration by up to ~90%, as assessed by Transwell migration assay (Figure 5A,B). ICG-001 also suppressed migration of ihPSC by up to ~50% (Figure 5C,D). Activated PSCs are known to induce pancreatic cancer cell migration [2–4,29–31], possibly via PSC-mediated induction of epithelial-mesenchymal transition in cancer cells [31]. Accordingly, we reasoned that pancreatic cancer cells, co-cultured with PSCs, which have been pre-treated with and thus presumably "de-activated" by ICG-001, would exhibit decreased migration compared with pancreatic cancer cells co-cultured with PSCs pre-treated with vehicle control. To test this notion, we co-cultured mouse pancreatic cancer cell line Panc02 with imPSC, which had been pre-treated for 72 h with ICG-001 or control, and assessed Transwell migration of the pancreatic cancer cells. We found that ICG-001 pre-treatment of imPSC, which were subsequently co-cultured with Panc02 cancer cells, suppressed PSC-induced migration of Panc02 cancer cells by up to ~60% (Figure 5E,F). Similarly, we found that ICG-001 pre-treatment of ihPSC, which were subsequently co-cultured with human pancreatic cancer cell line PANC-1, suppressed PSC-induced migration of PANC-1 cancer cells by up to ~70% (Figure 5G,H).

**Figure 5.** CBP/β-catenin antagonism suppresses migration of pancreatic stellate cells and pancreatic stellate cell-induced migration of pancreatic cancer cells. Effect of CBP/β-catenin antagonist ICG-001 versus control (DMSO) on migration of immortalized mouse pancreatic stellate cells (imPSC) ((**A**), Crystal Violet staining of cells which migrated; and (**B**), relative migration) and on migration of immortalized human pancreatic stellate cells (ihPSC) ((**C**), Crystal Violet staining of cells which migrated; and (**D**), relative migration), as assessed by Transwell migration assay. Note: imPSC and ihPSC were treated with ICG-001 for 48 h, after which time cells were seeded in serum-free medium onto 8-μm Transwell insert, and the lower chamber was filled with 10% FBS medium. Cells were then allowed to migrate for 6 h (imPSC) or 24 h (ihPSC) and kept in corresponding concentrations of ICG-001 versus control (DMSO) during migration. Effect of ICG-001 on imPSC-induced migration of mouse pancreatic cancer cells Panc02 ((**E**), Crystal Violet staining of cells which migrated; and (**F**), relative migration) and on ihPSC-induced migration of human pancreatic cancer cells PANC-1 ((**G**), Crystal Violet staining of cells which migrated; and (**H**), relative migration), as assessed by Transwell migration assay. Note: For PSC-induced Panc02 and PANC-1 cell migration, imPSC and ihPSC were pre-treated with ICG-001 for 72 h, after which time respective PSCs were seeded in 10% FBS medium into the lower chamber and respective cancer cells were seeded in 10% FBS medium onto Transwell insert. Cells were then allowed to migrate for 24 h. Relative migration was determined by counting the number of cells which had migrated across the Transwell insert as assessed by Crystal Violet staining and then normalizing to control (DMSO). *n* = 3, \*\*\* *p* < 0.001 compared to control (DMSO), # *p* < 0.05, ## *p* < 0.01, ### *p* < 0.001 compared to ICG-001 5 μM, & *p* < 0.05, &&& *p* < 0.001compared to ICG-001 10 μM. Scale bar: 1 mm.

#### **3. Discussion**

Given that pancreatic cancer, predominantly comprised of pancreatic ductal adenocarcinoma (PDAC), ranks as the 4th leading cause of cancer deaths in the United States with a 5-year survival for advanced stage disease of only 3% [1], there is an urgent need for treatments that offer durable benefits to PDAC patients. With increasing recognition that PDAC treatments traditionally focused on targeting pancreatic tumor cells have been insufficient or failed [2,3] and that activated pancreatic stellate cells (PSCs) promote PDAC progression [2–4] and are the key effector cells of desmoplasia [2–5], which correlates negatively with patient survival [6], there has been an increasing focus on developing novel therapeutic strategies which target activated PSCs to aid in combatting PDAC [2–5].

We now report for the first time that the small molecule specific CBP/β-catenin antagonist ICG-001 suppresses activation of PSCs as evidenced by their decreased proliferation, down-regulation of activation markers, e.g., Acta2 (in imPSC but apparently not in ihPSC), Col1a1, Prolyl 4-hydroxylase, and Survivin, up-regulation of Ppar-γ which is associated with quiescence, and reduced migration of PSC, as well as by reduced PSC-induced migration of pancreatic cancer cells. Our results are consistent with those of a previous study showing that retinoic acid-mediated suppression of Wnt/β-catenin signaling suppresses PSC activation as evidenced by inhibition of PSC proliferation and Col1a1 expression in vitro and by amelioration of PSC-mediated chronic pancreatitis and associated fibrosis in mice [11]. Our results are also consistent with those of a recent report revealing that ICG-001 suppresses the activation of hepatic stellate cells (which are developmentally and functionally analogous to PSCs) as evidenced by inhibition of α-SMA and collagen-I expression and migration by hepatic stellate cells in vitro, as well as by suppression of associated fibrogenesis in an acute CCl4-induced liver injury mouse model [19]. Expression of Acta2, Col1a1, Prolyl 4-hydroxylase, and Survivin is associated with activated PSCs [2–4,22–24,26,28] which actively proliferate and migrate [4], whereas expression of Ppar-γ is associated with quiescent PSCs [4,22,25,26]. Activated PSCs are the key effector cells for producing the collagen stroma of PDAC, with the resulting fibrous stroma capable of impeding chemotherapeutics/drugs from reaching targets [3]. The interplay between activated PSCs and PDAC cells enhances cancer progression [3], e.g., activated PSCs induce PDAC cell migration which has previously been correlated with epithelial-mesenchymal transition (EMT) [3,31]. Hence, based on our results, we would expect that suppressing activation and inducing quiescence of PSCs by treatment with ICG-001 would have a therapeutically beneficial effect on PSCs and PSC-associated PDAC progression in vivo.

Interestingly, we found in our current study that Acta2 mRNA expression differed between imPSC and ihPSC in response to ICG-001 treatment, with imPSC showing down-regulation and ihPSC showing apparent up-regulation in expression. The exact cause of the observed apparent discrepancy is unknown. A possible explanation for the discrepancy is that imPSC originates from relatively normal tissue whereas ihPSC originates from cancer tissue. Additionally, the Ct value of Acta2 for imPSC treated with control (DMSO) was ~21, whereas the Ct value for ihPSC treated with control (DMSO) was ~27, suggesting that perhaps Acta2 is not substantially expressed at the mRNA level in ihPSC versus imPSC, so that the observed up-regulation in mRNA expression with ICG-001 treatment in ihPSC may not be entirely comparable to the down-regulation in imPSC, given that the baseline Ct values between the two cell lines are so different. Moreover, the immortalization process, together with the difference in tissue of origin, in addition to the inherent biological variability between the different cells, may explain the differential regulation of Acta2 mRNA expression. An analogous explanation could be offered for the observed difference in Ppar-γ mRNA expression between imPSC and ihPSC. Thus and as recently underscored by Lenggenhager et al., cognizance of differences in PSC origin, condition of the pancreas from which PSCs were derived, and whether PSC cultures were primary or immortalized, is important given that such differences may explain apparently contradictory results between experiments using different types of PSCs [27].

In a broader, more fundamental context, Acta2 [32,33], Col1a1 [34], and Survivin [35] have all been previously identified as direct or indirect targets of the CBP/β-catenin signaling pathway and associated with a proliferative and pro-fibrotic phenotype which characterizes activated PSCs [2–4,22–24]. Furthermore, Ppar-γ, a nuclear receptor which is associated with PSC quiescence [4,22,25,26], is known to have anti-inflammatory/anti-fibrotic activity [36] and should compete with β-catenin for binding to CBP's N-terminal region, thereby phenocopying ICG-001 antagonism of CBP/β-catenin binding and associated signaling [37]. As such, it is not surprising that various studies have shown that treatment with CBP/β-catenin antagonist ICG-001 (or the structurally related derivative PRI-724) is effective pre-clinically at ameliorating fibrosis in the peritoneum [38], endometrium [39], lung [13], kidney [32], skin [40], heart [41], and liver [19,42]. Importantly, a recent phase 1 trial utilizing PRI-724 demonstrated that treatment of patients safely improved liver histology and Child–Pugh scores for

cirrhosis [43], suggesting that CBP/β-catenin antagonism is a viable therapeutic for combating fibrotic diseases in general. Moreover, it is known that chronic pancreatitis is a pathological syndrome characterized by persistent fibrosis effected by activated PSCs and is itself a predominant risk factor for PDAC [9,10], conferring a ~8 to 12-fold increased risk of developing PDAC to chronic pancreatitis patients [44]. These observations, in conjunction with the results of our current study, have overarching therapeutic implications, namely: CBP/β-catenin antagonism would be expected not only to be effective at suppressing activation of PSCs and thereby ameliorating already existing PDAC, but also to be effective as a "PDAC prophylactic" by inhibiting activation of PSCs during the early pre-cancerous stage of pancreatic fibrosis/chronic pancreatitis.

Given the pressing need to develop and implement better treatment strategies for combatting PDAC, we now present in this Communication our novel results on the effectiveness of CBP/β-catenin antagonism in suppressing PSC activation, with broad therapeutic implications for treating PDAC and chronic pancreatitis, both of which are known to be promoted by activated PSCs. Because of the limitation in scope of our current study, future studies (e.g., using primary PSCs and in vivo models) would be required to further validate the effect of CBP/β-catenin antagonism on PSC biology/pathobiology, including the interaction between PSCs and PDAC cells.

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

#### *4.1. Cell Lines and Culture Conditions*

Immortalized mouse pancreatic stellate cell line (imPSC) and immortalized human pancreatic stellate cell line (ihPSC) were kindly provided by Richard T. Waldron, Aurelia Lugea, and Raul A. Urrutia and were established as previously described [20,21]. imPSC were grown and cultivated in Dulbecco's Modified Eagle Medium (DMEM) with low glucose (1000 mg/L) while ihPSC were grown and cultivated in DMEM with high glucose (4500 mg/L). All culture medium was supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin unless otherwise indicated. Cells were maintained in an incubator at 37 ◦C with 5% CO2.

#### *4.2. Pharmacologic Agents*

Small molecule specific CBP/β-catenin antagonist ICG-001 as previously described [16] was donated by Professor Michael Kahn and used at concentrations as indicated.

#### *4.3. Cell Proliferation Assays*

CellTiter-Glo assay (Promega) was performed according to the manufacturer's protocol. Cells were plated in triplicate in 96-well plates at 1 <sup>×</sup> 10<sup>4</sup> cells/well in 100 <sup>μ</sup>L of medium. Plates were incubated at 37 ◦C in 5% CO2. The next day, cells were treated with ICG-001 at 100 μM, 50 μM, 25 μM, 12.5 μM, 6.25 μM, 3.13 μM 1.56 μM or control (DMSO) and were incubated for an additional 48 h. Then, 50 μL CellTiter-Glo reagent and 50 μL of DMEM were added to the wells and incubated for 10 min protected from the light. Luminescence signal was assayed using EnVision Multilabel Plate Reader (Perkin-Elmer).

Cell proliferation was also assessed by cell counting using a hemocytometer. Briefly, imPSC or ihPSC were seeded in 6-well plates at 5 <sup>×</sup> 10<sup>4</sup> cells/well and incubated at 37 ◦C in 5% CO2. The next day, cells were treated with ICG-001 at 5 μM, 10 μM, 25 μM or control (DMSO) for 24 h or 48 h. Cell numbers were counted at 0, 24, and 48 h after treatment.

#### *4.4. Quantitative Polymerase Chain Reaction (qPCR)*

Cells were treated with ICG-001 for 48 h. Total mRNA was extracted by TRIzol reagent (Invitrogen) according to the manufacturer's protocol. cDNA was synthesized using qScript cDNA Synthesis Kit (Quantabio) and used as template for qPCR with SYBR Green detection method. The PCR primer sequences used for mouse cells were as follows: Acta2 (F: 5 -GTCCCAGACATCAGGGAGTAA-3 , R: 5 -TCGGATACTTCAGCGTCAGGA-3 ); Col1a1 (F: 5 -GCTCCTCTTAGGGGCCACT-3 , R: 5 -CCACGTCTCACCATTGGGG-3 ); Survivin (F: 5 -GAGGCTGGCTTCATCCACTG, R: 5 -ATGCTCCTCTATCGGGTTGTC-3 ); Ppar-γ (F: 5 -TTTTCCGAAGAACCATCCGATT-3 , R: 5 -ATGGCATTGTGAGACATCCCC-3 ). The PCR primer sequences used for human cells were as follows: ACTA2 (F: 5 -CTATGAGGGCTATGCCTTGCC-3 , R: 5 -GCTCAGCAGTAGTAACGAAGGA-3 ); COL1A1 (F: 5 -GAGGGCCAAGACGAAGACATC-3 , R: 5 -CAGATCACGTCATCGCACAAC-3 ); SURVIVIN (F: 5 -AGGACCACCGCATCTCTACAT-3 , R: 5 -AAGTCTGGCTCGTTCTCAGTG-3 ); PPAR-γ (F: 5 -CTATGGAGTTCATGCTTGTG-3 , R: 5 -GTACTGACATTTATTT-3 ). Housekeeping gene PCR primer sequences used were GAPDH for mouse cells (F: 5 -GGTGCTGAGTATGTCGTGGA-3 , R: 5 -ACAGTCTTCTGGGTGGCAGT-3 ) and GAPDH for human cells (F: 5 -AGAAGGCTGGGGCTCATTTG-3 , R: 5 AGGGGCCATCCACAGTCTTC-3 ).

#### *4.5. Immunofluorescence*

Cells were plated, and the next day cells were treated with ICG-001 or control (DMSO) for 72 h, followed by 4% PFA fixation for 10 min. After 3 times of PBS washing, 1% BSA with 0.1% Triton X-100 was used for blocking nonspecific binding. Primary antibodies for α-SMA (Cell Signaling Technology, #19245s; 1:100), Survivin (Cell Signaling Technology, #2808s; 1:250), and Ppar-γ (Affinity BioReagents, #PA3-821; 1:50) were used for overnight incubation at 4 ◦C. Secondary antibody anti-rabbit IgG-Alexa Fluor 488 (Invitrogen, #A11034; 1:1000) was incubated for 40 min at room temperature. Hoechst 33342 was used for nuclear staining for 10 min. A fluorescence microscope (Eclipse Ti2, Nikon) was used to observe target protein expression.

#### *4.6. Western Blot*

2 <sup>×</sup> 10<sup>5</sup> of imPSC were plated in 10 cm plates. The next day, cells were treated with ICG-001 or control (DMSO) and were incubated for an additional 72 h. Then, cells were collected, and total cellular proteins were extracted using M-PER Mammalian Protein Extraction Reagent (ThermoFisher). After protein quantification by Bradford assay method, protein samples were separated by 10% SDS PAGE, followed by transfer to nitrocellulose membrane. Next, the membranes were blocked with 5% milk in Tris-Buffered Saline with 0.1% Tween. The membranes were incubated with primary antibody overnight at 4 ◦C and incubated with secondary antibody for 1 h the following day. Primary antibodies for α-SMA (Cell Signaling Technology, #19245s), Prolyl 4-hydroxylase (P4HA2) (ThermoFisher, #PA5-96280), and GAPDH (Santa Cruz Biotechnologies, #sc-32233), and secondary antibody anti-rabbit IgG-HRP (Santa Cruz Biotechnologies, #sc-2357) were used. Protein bands were detected using ECL prime Western blotting detection reagent (Amersham) and visualized by Chemidoc Imaging System (Bio-Rad). Each protein band of interest was digitized by densitometry program ImageJ (NIH) or ImageLab (Bio-Rad). Densitometric quantitation of protein bands was normalized to Ponceau S or GAPDH and then to control (DMSO).

#### *4.7. Transwell Migration Assay*

For PSC Transwell migration assay, imPSC and ihPSC were treated with ICG-001 for 48 h. Then, <sup>1</sup> <sup>×</sup> 104 cells were seeded in serum-free DMEM onto 8-μm Transwell insert (Corning). The lower chamber was filled with 10% FBS supplemented DMEM. Cells were then allowed to migrate while incubated at 37 ◦C with 5% CO2 for 6 h (imPSC) or 24 h (ihPSC) and kept in corresponding concentrations of ICG-001 versus control (DMSO) during migration.

For PSC-induced Panc02 and PANC-1 cancer cell Transwell migration assay, imPSC and ihPSC were pre-treated with ICG-001 for 72 h, and then seeded in 10% FBS DMEM into the lower chamber of 24-well plate at a total number of 5 <sup>×</sup> 104 (imPSC) or 1 <sup>×</sup> 105 (ihPSC) cells per well. Panc02 or PANC-1 cells (1 <sup>×</sup> 104) were seeded in 10% FBS DMEM onto Transwell insert. Cells were then incubated and allowed to migrate for 24 h.

For both PSC Transwell migration assay and PSC-induced Panc02 and PANC-1 cancer cell Transwell migration assay, after incubation, cells were stained with 1% Crystal Violet solution (Sigma) for 30 min. The cells on the upper surface of the Transwell insert were gently removed by cotton swab, and the cells which had migrated to the bottom surface of the insert were counted under bright field microscopy. Relative migration was determined by normalizing the number of cells which had migrated with ICG-001 treatment to the number of cells which had migrated with control (DMSO) treatment.

#### *4.8. Statistical Analysis*

Numerical data were expressed as the means ± SD unless otherwise noted. Student's t-test was performed to assess the statistical significance between two sets of data as appropriate. One-way ANOVA followed by post-hoc Tukey test was performed for multiple comparisons when appropriate. *p* values less than 0.05 were considered significant.

#### **5. Conclusions**

We report for the first time that the small molecule specific CBP/β- ◦C antagonist ICG-001 suppresses activation of PSCs as evidenced by their decreased proliferation, down-regulation of activation markers, e.g., Acta2, Col1a1, Prolyl 4-hydroxylase, and Survivin, up-regulation of Ppar-γ which is associated with quiescence, and reduced migration; furthermore, migration of PDAC cells is reduced when co-cultured with PSCs which have been pre-treated with ICG-001. Hence, CBP/β-catenin antagonism represents a novel therapeutic strategy for suppressing PSC activation and may be effective at treating PDAC and "pre-cancerous" chronic pancreatitis, both of which are known to be promoted by activated PSCs.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6694/12/6/1476/s1, Figure S1: CBP/β-catenin antagonism suppresses protein expression of activation markers of pancreatic stellate cells as assessed by immunoblot (Whole immunoblots).

**Author Contributions:** M.C., methodology, formal analysis, investigation, writing—original draft preparation, visualization; S.-M.K., methodology, formal analysis, investigation, writing—original draft preparation, visualization; J.-L.T., methodology, investigation, writing—review and editing; Y.-C.Y., investigation, writing—review and editing; L.G.M., investigation, writing—review and editing; R.T.W., methodology, investigation, resources, writing—review and editing; A.L., methodology, investigation, resources, writing—review and editing; R.A.U., methodology, resources, writing—review and editing, funding acquisition; S.J.P., methodology, investigation, resources, writing—original draft preparation, funding acquisition; K.K.Y.L., conceptualization, methodology, formal analysis, investigation, resources, writing—original draft preparation, writing—review and editing, visualization, supervision, project administration, funding acquisition. All authors have read and agreed to the published version of the manuscript.

**Funding:** K.K.Y. Lai has been supported by NIH K08AA025112. S.J. Pandol has been supported by NIH U01DK108314, P01CA233452, and P01CA236585. R.A. Urrutia has been supported by NIH R01DK052913.

**Conflicts of Interest:** The 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**


© 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* **Feasibility of Targeting Traf2-and-Nck-Interacting Kinase in Synovial Sarcoma**

**Tetsuya Sekita 1,2, Tesshi Yamada 3,\*, Eisuke Kobayashi 4, Akihiko Yoshida 5, Toru Hirozane 2, Akira Kawai 4, Yuko Uno 6, Hideki Moriyama 6, Masaaki Sawa 6, Yuichi Nagakawa 3, Akihiko Tsuchida 3, Morio Matsumoto 2, Masaya Nakamura 2, Robert Nakayama <sup>2</sup> and Mari Masuda <sup>1</sup>**


Received: 22 April 2020; Accepted: 10 May 2020; Published: 16 May 2020

**Abstract:** Background: The treatment of patients with metastatic synovial sarcoma is still challenging, and the development of new molecular therapeutics is desirable. Dysregulation of Wnt signaling has been implicated in synovial sarcoma. Traf2-and-Nck-interacting kinase (TNIK) is an essential transcriptional co-regulator of Wnt target genes. We examined the efficacy of a small interfering RNA (siRNA) to *TNIK* and a small-molecule TNIK inhibitor, NCB-0846, for synovial sarcoma. Methods: The expression of TNIK was determined in 20 clinical samples of synovial sarcoma. The efficacy of NCB-0846 was evaluated in four synovial sarcoma cell lines and a mouse xenograft model. Results: We found that synovial sarcoma cell lines with Wnt activation were highly dependent upon the expression of *TNIK* for proliferation and survival. NCB-0846 induced apoptotic cell death in synovial sarcoma cells through blocking of Wnt target genes including *MYC*, and oral administration of NCB-846 induced regression of xenografts established by inoculation of synovial sarcoma cells. Discussion: It has become evident that activation of Wnt signaling is causatively involved in the pathogenesis of synovial sarcoma, but no molecular therapeutics targeting the pathway have been approved. This study revealed for the first time the therapeutic potential of TNIK inhibition in synovial sarcoma.

**Keywords:** Wnt signaling; synovial sarcoma; TNIK; NCB-0846; MYC

#### **1. Introduction**

Synovial sarcoma is a rare aggressive neoplasm that accounts for 10–20% of soft tissue sarcomas. It affects mainly adolescents and young adults [1,2], and 40–50% of patients are under the age of 30 at diagnosis [3]. The mainstay of treatment is wide surgical excision and conventional chemotherapy [4,5]. However, the disease tends to show early or late recurrence and often becomes resistant to cytotoxic agents. The 10 year disease-free survival rate of patients with distant metastases remains around

50% [6]. It is desirable to develop new molecular therapeutics targeting pathways essential for the growth and survival of synovial sarcoma. The fusion *SS18*-*SSX* (*SSX1*, *SSX2*, or *SSX4*) gene produced by a chromosomal translocation, t (X;18) (p11.2; q11.2), is detectable in ~95% of synovial sarcomas [7–9]. Although dysregulation of the BAF chromatin-remodeling complex has been shown to be involved in the oncogenic activity of SS18-SSX [10,11], no therapeutics that can target the product of SS18-SSX or the BAF complex have yet been developed.

The canonical (β-catenin-dependent) Wnt signaling pathway plays crucial roles in the regulation of diverse biological processes including cell proliferation, survival, migration, and polarity, specification of cell fate, and self-renewal of embryonic stem cells, and its dysregulation has been implicated in the generation and progression of various malignancies [12]. Wnt signaling is also implicated in the pathogenesis of synovial sarcoma; synovial sarcoma cells frequently show accumulation of β-catenin protein in the nucleus [13], and express Wnt target gene products such as AXIN2 (axis inhibition protein 2), DKK1 (dickkopf1), survivin, c-MYC, and cyclinD1 [14]. SS18-SSX is responsible for the nuclear translocation of β-catenin [15,16], and Wnt signaling is aberrantly activated by SS18-SSX in a transgenic mouse model; inhibition of Wnt signaling through genetic loss of β-catenin blocks synovial sarcoma tumor formation [17]. *SS18*-*SSX2*-specific small interfering RNA (siRNA) reduces the expression of Wnt target gene products [14]. Together, these studies have highlighted the Wnt signaling pathway as a potential therapeutic target for synovial sarcoma.

Through comprehensive mass spectrometry analysis of the nuclear proteins of colorectal cancer cells, we previously identified Traf2-and-Nck-interacting kinase (TNIK) as a component of the T-cell factor-4 (TCF4) and β-catenin transcriptional complex, the most downstream effector of the Wnt signaling pathway [18]. More than 80% of colorectal cancers carry inactive mutations in the *APC* tumor-suppressor gene, and Wnt signaling is activated downstream of it. We found that TNIK was essential for transactivation of Wnt target genes and that colorectal cancer cells were highly sensitive to TNIK inhibition [19,20]. We screened a compound library and identified a novel small-molecule TNIK inhibitor named NCB-0846. NCB-0846 suppresses the transcriptional co-regulator function of TNIK by modifying its conformational structure [21,22]. NCB-0846 exhibited marked anti-tumor and anti-stem-cell activities in colorectal cancer cells and patient-derived xenografts through blocking of Wnt target gene expression [21].

Based on these findings, we speculated that TNIK inhibition would be effective for treatment of synovial sarcoma. Here, we report the therapeutic potential of TNIK inhibition in synovial sarcoma.

#### **2. Results**

#### *2.1. Activation of Wnt Signaling and TNIK in Synovial Sarcoma*

To evaluate the activation of Wnt signaling, four synovial sarcoma cell lines were transfected with a pair of reporters (super-TOP and super-FOP luciferase reporter plasmids), and their luciferase activity was measured. Active transcription of T-cell factor (TCF)/lymphoid enhancer factor (LEF) was detected in two synovial sarcoma cell lines, HS-SY-II and SYO-1 (Figure 1A). Expression of a Wnt target gene product (AXIN2 protein) (Figure 1B) and nuclear expression of β-catenin (red, Figure 1C) were detected in these two cell lines. Nuclear translocation of TNIK is indicative of its active status [19]. Nuclear expression of TNIK was detected in all four cell lines examined (green, Figure 1C), and TNIK was co-localized with β-catenin in the nuclei of synovial sarcoma cell lines with Wnt activation (merge, Figure 1C). Using immunohistochemistry, the expression of β-catenin and TNIK was then examined in tissue specimens resected from 20 patients with synovial sarcoma. We detected nuclear staining of β-catenin in 90% (18/20) of the examined cases, and these tumors also exhibited nuclear expression of TNIK (Figure 1D and Table S1).

**Figure 1.** Wnt activation in synovial sarcoma. (**A**) T-cell factor (TCF)/lymphoid enhancer factor (LEF) transcriptional activity of synovial sarcoma cells. Four synovial sarcoma cell lines (HS-SY-II, SYO-1, Yamato, and Aska) were transfected with the super-TOP flash or super-FOP flash luciferase reporter, and their luciferase activity was measured 24 h later. Data represent the mean TOP/FOF ratio (± S.D.) of three replicates. (**B**) Expression of the axis inhibition protein 2 (AXIN2) and γ-tubulin (loading control) proteins determined by immunoblotting. (**C**) Dual immunofluorescence analysis of β-catenin and Traf2-and-Nck-interacting kinase (TNIK) protein expression in synovial sarcoma cells. Scale bar: 20 μm. (**D**) Immunohistochemical analysis of the β-catenin and TNIK proteins in clinical specimens of synovial sarcoma. Representative cases with strong positive (++) and negative (−) nuclear β-catenin expression are shown. Scale bars: 100 μm in low-power views (left) and 10 μm in high-power views (right).

#### *2.2. Growth Suppression of Synovial Sarcoma Cells Through Silencing of TNIK*

Transfection of three siRNA constructs targeting *TNIK* (siTNIK#1, #2, and #3) into HS-SY-II and SYO-1 synovial sarcoma cells was confirmed to reduce the levels of *TNIK* gene expression relative to cells transfected with control siRNA (Ctrl) (Figure 2A). Real-time monitoring revealed that knockdown of *TNIK* induced the almost complete growth arrest of HS-SY-II and SYO-1 cells (Figure 2B) and significantly reduced TCF/LEF transcription in HS-SY-II cells lentivirally engineered to stably carry a TOP-driven green fluorescent protein (GFP) reporter construct (Figure 2C), even after being normalized to cell viability (Figure 2D). The four synovial sarcoma cell lines were transfected with siRNA to *TNIK* (siTNIK#2) or control siRNA (siCtrl), and their viability was assessed 72 h later. *TNIK* knockdown significantly suppressed the viability of HS-SY-II, SYO-1, and Yamato cells, but not that of Aska cells (Figure 2E). Aska cells lack Wnt activation or *MYC* gene amplification (discussed later). *TNIK* knockdown induced cleavage of poly (ADP-ribose) polymerase-1 (PARP-1) in HS-SY-II cells (Figure 2F), indicating induction of apoptosis.

**Figure 2.** Growth suppression and apoptosis induction in synovial sarcoma cells by knockdown of *TNIK*. (**A**) HS-SY-II and SYO-1 cells were transfected with control small interfering RNA (siRNA) (siCtrl) and siRNA to *TNIK* (siTNIK#1, #2, and #3), and their relative expression of *TNIK* (normalized to *ACTB*) was quantified in triplicate by real-time RT-PCR 72 h after transfection. The expression level in cells

transfected with siCtrl was set at 1. \* *p* < 0.05, \*\*\* *p* < 0.0005, \*\*\*\* *p* < 0.0005 (multiple *t*-test corrected using the Holm–Sidak method). (**B**) Real-time growth monitoring of HS-SY-II and SYO-1 cells transfected with siCtrl and siRNA to TNIK (siTNIK#1, #2, and #3). Data represent the mean cell index (https://www. aceabio.com/products/icelligence/) ± S.D. of three replicates. (**C**,**D**) Suppression of TCF/LEF transcription by *TNIK* knockdown. HS-SY-II cells engineered to stably carry a TOP-driven green fluorescent protein (GFP) reporter were transfected with siCtrl or siTNIK#2. Average integrated intensity (summed fluorescence intensity per cell) (https://www.essenbioscience.com/media/uploads/files/8000-0193-A00\_ ZOOM\_Fluorescence\_Processing\_Tech\_Note.pdf#search=%27Average+Integrated+Intensity%27) was monitored every 6 h for 72 h (**C**). Total integrated intensity (total sum of fluorescence intensity per well) was normalized to ATP production 24 h after transfection (D). Data represent the mean ± S.D. of three replicates. \*\* *p* < 0.005, \*\*\*\* *p* < 0.0005 (multiple *t*-test corrected using the Holm–Sidak method). (**E**) Synovial sarcoma cells were transfected with siCtrl or siTNIK#2, and their expression of *TNIK* (normalized to *ACTB*) was quantified by real-time RT-PCR 72 h after transfection (left). Their relative viability to siCtrl (set to one) was assessed in terms of ATP production (right). \*\* *p* < 0.005, \*\*\* *p* < 0.0005, \*\*\*\* *p* < 0.0005, n.s. not significant (multiple *t*-test corrected using the Holm–Sidak method). Data represent the mean ± S.D. of three replicates. (**F**) Expression of the poly (ADP-ribose) polymerase-1 (PARP-1) and γ-tubulin (loading control) proteins determined by immunoblotting for 72 h.

#### *2.3. Sensitivity of Synovial Sarcoma to NCB-0846*

Based on the remarkable growth suppression and apoptosis induction in synovial sarcoma cells by silencing of the *TNIK* gene, the sensitivity of synovial sarcoma cell lines to a small-molecule TNIK inhibitor, NCB-0846, was then evaluated. Consistent with the siRNA to *TNIK*, NCB-0846 reduced the viability of HS-SY-II, SYO-1, and Yamato cells with a half maximal inhibitory concentration (IC50) of 339, 356, and 767 nM, respectively. Aska cells were insensitive to NCB-0846 and had an IC50 value exceeding 2.0 μM (Figure 3A). The water-soluble hydrochloride salt of NCB-0846 (named NCB-1055) [21] was administered orally to immune-deficient mice subcutaneously inoculated with HS-SY-II cells. The xenografts regressed below the baseline (before administration) even after the first administration of NCB-1055 and did not re-grow (Figure 3B). Real-time monitoring of cell-surface phosphatidylserine (PS) revealed that NCB-0846, but not its diastereomer (named NCB-0970), induced apoptotic cell death of HS-SY-II cells within 6 h after the start of drug treatment (Figure 3C). NCB-0970 was used as a negative control, i.e., a compound having the same chemical structure as NCB-0846 except for an opposite configuration of one terminal hydroxyl group [21]. An increase of the sub-G1 cell population (Figure 3D) and cleavage of PARP-1 (Figure 3E) confirmed the induction of apoptotic cell death by NCB-0846.

#### *2.4. Gene Expression Profiling*

We then examined the changes in gene expression associated with the early induction of apoptosis by NCB-0846. HS-SY-II cells were exposed to NCB-0846 or NCB-0970 for 6 h, and their relative RNA expression (FPKM, fragments per kilobase of exon per million mapped reads) was determined using a next-generation sequencer. We found that the expression of a large number (6710/14,611) of genes was suppressed more than 2-fold by treatment with NCB-0846 in comparison to that with NCB-0970 (Figure 4A,B), indicating that this compound had a large impact on gene transcription beyond the suppression of Wnt target gene expression. Gene set enrichment analysis (GSEA) (Table S2) revealed significantly concordant alteration of a group of genes annotated to the Wnt signaling pathway (Figure 4C). The differentially expressed genes were mapped to the Wnt signaling pathway deposited in the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (Figure 4D). We previously reported that TNIK was required for the tumor-initiating function of colorectal cancer stem cells [21,22]. Consistently, a significant proportion of downregulated genes were mapped to the signaling pathways regulating stem cell pluripotency (Figure S1). The entire RNA sequencing dataset has been deposited in the DNA Data Bank of Japan (DDBJ) Sequence Read Archive (SRA) database with the accession number DRA010051.

**Figure 3.** Sensitivity of synovial sarcoma to NCB-0846. (**A**) ATP production by four synovial sarcoma cell lines cultured with increasing doses of NCB-0846 for 72 h. Data represent the mean (relative to no treatment) of three replicates. (**B**) HS-SY-II cells were inoculated into the subcutaneous tissues of 6 week old female NOD.CB17-*Prkdc*scid/J (NOD/SCID) mice. When the average volume of the xenografts reached ~200 mm3, water (vehicle, *n* = 5) or 80 mg/kg (*n* = 5) NCB-0846 hydrochloride (NCB-1055) [21] was administered orally on the days indicated by -. Tumor volume was measured on the days of drug administration (left), and tumors were excised (lower right) and weighed (upper right) 7 days after the start of drug administration. \* *p* < 0.05, \*\* *p* < 0.005 (multiple *t*-test corrected using the Holm–Sidak method). Error bars represent S.E.M. (**C**) HS-SY-II cells were cultured with dimethyl sulfoxide (DMSO) (vehicle), NCB-0846 (3 μM) or NCB-0970 (3 μM) in the presence of the Real-time-Glo™ Annexin V Apoptosis Assay Reagent (Promega), and relative luminescence unit (URL) data were collected at every 2 h over a 10 h time course. \*\*\*\* *p* < 0.0005 (multiple *t*-test corrected using the Holm–Sidak method). Data represent the mean of three readings for each replicate ± S.D. (**D**) HS-SY-II cells were untreated (Ctrl) or treated with DMSO, NCB-0846 (3 μM), or NCB-0970 (3 μM) for 6 h. The percentage of cells in each cell cycle fraction was determined by flow cytometry. (**E**) HS-SY-II cells were treated with DMSO (control), NCB-0846 (3 μM), or NCB-0970 (3 μM) for 7 h. The expression levels of PARP-1 and γ-tubulin (loading control) were determined by immunoblotting.

**Figure 4.** Gene expression profiling of synovial sarcoma cells treated with NCB-0846. (**A**) Scatter plot of genes differentially expressed between cells treated with NCB-0846 and NCB-0970 (negative control). Red dots represent genes upregulated more than 2-fold, and blue dots represent genes downregulated more than 2-fold in cells treated with NCB-0846. (**B**) Heat map plot of genes differentially expressed between NCB-0846 and NCB-0970. The upper color bar represents the degree of differential expression. (**C**) Gene set enrichment analysis (GSEA) showing the significant enrichment of genes annotated to the gene ontology (GO) terms "Wnt signaling pathway" (*p* = 0.004) and "canonical Wnt signaling pathway" (*p* = 0.005) and to "Wnt signaling pathway" deposited in the KEGG database (*p* = 0.002). NES: normalized enrichment score (http://software.broadinstitute.org/gsea/index.jsp). (**D**) Mapping of differentially expressed genes onto the Wnt signaling pathway. Yellow boxes indicate genes downregulated (>2-fold) by NCB-0846.

#### *2.5. NCB-0846 Suppresses MYC Gene Expression*

Using real-time RT-PCR, we then confirmed the differential expression of Wnt target genes. The expression of 88% (78/88) of known Wnt target genes (https://web.stanford.edu/group/nusselab/cgibin/wnt/target\_genes) was found to be downregulated (Table S3). Among these genes, *MYC* showed the most significant degree of downregulation (Figure 5A). *MYC* encodes the c-MYC protein, a transcription factor that regulates as many as 10–15% of genes in the genome [23]. We confirmed the significant enrichment of c-MYC transcriptional targets among genes regulated by NCB-0846 (Figure 5B). This marked downregulation of *MYC* was also observed in other synovial sarcoma cell lines (Figure 5C).

**Figure 5.** NCB-0846 suppresses *MYC* gene expression. (**A**) Comparison of Wnt target gene expression (normalized to *GAPDH* and log-transformed) of HS-SY-II cells treated with NC-0846 and NCB-0970 for 6 h. (**B**) Significant enrichment of c-MYC target genes revealed by RNA sequencing and Gene set enrichment analysis (GSEA). (**C**) Four synovial sarcoma cell lines were treated with dimethyl sulfoxide (DMSO) (control), NCB-0846 (3 μM), or NCB-0970 (3 μM) for 6 h, and expression of the *MYC* gene (relative to DMSO) was quantified by real-time RT-PCR and normalized to that of *ACTB*. \*\*\* *p* < 0.0005, \*\*\*\* *p* < 0.0005 (multiple *t*-test corrected using the Holm–Sidak method). Data represent the mean ± S.D. of three replicates.

#### *2.6. Dependency of Synovial Sarcoma Cells on MYC*

*MYC* is one of the targets of TCF/LEF transcription factors [24], and Wnt signaling is known to exert its oncogenic activity primarily through transactivation of the *MYC* gene [25]. We found that synovial sarcoma HS-SY-II cells with active Wnt target gene expression were highly dependent on *MYC* gene expression for proliferation (Figure 6A). However, Yamato cells also expressed the c-MYC protein (Figure 6B) in spite of inactive Wnt signaling (Figure 1A–C). We found that an increase (2.2-fold) in the copy number of the *MYC* gene (Figure 6C) appeared to be responsible for the upregulation. A high degree (>2.0-fold) of *MYC* oncogene amplification is known to be infrequent in synovial sarcoma [26]. However, nuclear expression of c-MYC was detected in 85% (17/20) of clinical specimens and was frequent (≥30% of tumor cells) in 15% of them (3/20) (Figure S2 and Table S1). The Aska cell line carried the normal copy number (1.0-fold) of *MYC* (Figure 6C), and its level of c-MYC expression was

lower than in other cell lines (Figure 6B). Knockdown of *MYC* gene expression by siRNA reduced the viability of HS-SY-II, SYO-1, and Yamato cells, but Aska cells were insensitive to silencing of *MYC* (Figure 6D) and NCB-0846 (Figure 3A), suggesting that NCB-0846 induces apoptotic cell death of synovial sarcoma at least partially through transcriptional suppression of *MYC*.

**Figure 6.** Dependence of synovial sarcoma cells on *MYC*. (**A**) Relative *MYC* expression (left) and real-time growth monitoring (right) of HS-SY-II cells transfected with control small interfering (siRNA) (siCtrl) and siRNA to *MYC* (siMYC#1 and #2). Data represent the mean *MYC* expression (normalized to *ACTB*) (left) and cell index (right) ± S.D. of three replicates. \* *p* < 0.05 (multiple *t*-test corrected using the Holm–Sidak method). (**B**) The expression of c-MYC and γ-tubulin (loading control) in four synovial sarcoma cell lines was determined by immunoblotting. (**C**) Relative copy numbers of the *MYC* gene (normalized to the RNase P gene) in four synovial sarcoma cell lines determined by digital PCR. (**D**) Four synovial sarcoma cell lines were transfected with control siRNA (siCtrl) and siRNA to *MYC* (siMYC#2) in triplicate. Seventy-two hours later, their relative expression of *MYC* (normalized to *ACTB*) was quantified by real-time RT-PCR (left), and their relative viability was assessed in terms of ATP production (right). Data represent the mean ± S.D. of three replicates. \*\* *p* < 0.005, \*\*\* *p* < 0.0005, \*\*\*\* *p* < 0.0005, n.s. not significant (multiple *t*-test corrected using the Holm–Sidak method).

#### **3. Discussion**

Conventional cytotoxic chemotherapeutic agents including anthracycline, ifomide, and trabectedin have proven to be effective for the treatment of metastatic synovial sarcoma [27], but their usage and efficacy are often limited by the emergence of adverse events and drug resistance. Pazopanib is the first and only molecular therapeutic agent approved for the treatment of multiple histological subtypes of soft tissue sarcoma [28]. Pazopanib is a multi-tyrosine kinase inhibitor, and its main mode of action is believed to be inhibition of vascular endothelial growth factor receptor (VEGF)-mediated tumor angiogenesis [29]. The median survival of synovial sarcoma patients treated with pazopanib, however, was only 10.6 months, and the pazopanib treatment was associated with a high frequency of adverse events including hypertension, thrombocytopenia, and pneumothorax [28,30]. Early clinical trials of T lymphocytes genetically engineered to target the NY-ESO-1 cancer/testis antigen have yielded promising results [31,32], but this cancer immunotherapy is applicable only to patients with the human leukocyte antigen (HLA)-A\*0201 or -A\*0206 type as well as expression of NY-ESO-1 in tumors. Moreover, autologous lymphocyte cultivation is incurs significant costs and requires long-term discontinuation of ongoing treatment, potentially leading to fatal disease progression. Frizzled homolog 10 (FZD10) has attracted attention as a promising therapeutic target for synovial sarcoma [33], as its expression is limited to the cell membrane of synovial sarcoma and absent from vital organs [34]. A recent first-in-human clinical trial clarified the biodistribution, safety, and recommended dose of a radiolabeled humanized monoclonal antibody to FZD10 [35], but its efficacy has not been established.

Synovial sarcoma is uniquely characterized by the balanced chromosomal translocation t[X, 18; p11, q11], demonstrable in virtually all cases and not found in any other human neoplasms [2,8]. This translocation creates an in-frame fusion of SS18 to SSX1, SSX2, or SSX4, whereby all but the eight C-terminal amino acids of SS18 are replaced by the 78 C-terminal amino acids of the SSX partner. Kadoch and Crabtree observed that SS18-SSX was incorporated into the SWI/SNF (SWItch/Sucrose Non-Fermentable) complex [36]. Middeljans and colleagues reported that expression of the fusion oncogene induced depletion of the BAF47 (*SMARCB1*) subunit from the SWI/SNF complex [37]. Potential convergence may exist between the SWI/SNF complex and Wnt signaling, as loss of *SMARCB1* reportedly activates Wnt signaling [38]. Barham and colleagues [17] provided direct evidence for involvement of Wnt signaling in the SS18-SSX-mediated carcinogenesis of synovial sarcoma. The Wnt signaling pathway is aberrantly activated in an SS18-SSX2 transgenic mouse model, and genetic loss of β-catenin (*Ctnnb1*) blocks tumor formation in this model. Trautmann and colleagues [14] found that introduction of SS18-SSX into untransformed cells induced transactivation of Wnt target genes. Synovial sarcoma cell lines (SYO-1, CME-1, and HS-SY-II) showed sensitivity to three small-molecule inhibitors of the TCF/β-catenin complex (PKF115–584, CGP049090, and PKF118–310). β-Catenin stabilization in a transgenic animal model reportedly enhanced SS18-SSX-driven tumorigenesis and produced more dedifferentiated tumors [39]. Based on these findings, it is considered feasible to target a signaling molecule of the Wnt signaling pathway in synovial sarcoma.

TNIK is a component of the TCF4 and β-catenin transcriptional complex and functions as an essential co-regulator of Wnt target gene expression [19,40]. We screened a chemical library and identified a small-molecule TNIK-inhibitory compound named NCB-0846. This compound inhibited the expression of various Wnt target genes (such as *MYC*, *AXIN2*, and *CD44*) through conformational modification of TNIK and abrogated the stemness of colorectal cancer cells [21]. The *MYC* oncogene is a direct target of TCF/LEF family transcription factors [24] and centrally mediates the oncogenic activity of Wnt signaling [25]. In the present study, we revealed that Wnt signaling is activated in synovial sarcoma cells (Figure 1) and that siRNA-mediated or pharmacological TNIK inhibition reduced their viability and induced apoptosis (Figures 2 and 3). NCB-0846 suppressed the expression of *MYC* and other Wnt target genes (Figure 5). Synovial sarcoma cell lines with high c-MYC protein expression were sensitive to the compound (Figure 3) and to the gene silencing of *MYC* (Figure 6). These results suggest that transcriptional *MYC* gene suppression is the central mode of action (MOA) of NCB-0846.

c-MYC is a versatile transcription factor that regulates the expression of genes involved in various biological functions such as cell proliferation, apoptosis, differentiation, and metabolism [41], and its inhibition would be expected to have a huge impact on the cancer transcriptome [42]. Aberrant expression or gene amplification of *MYC* has been implicated in the aggressiveness of various malignancies [43,44]. Shen and colleagues examined 32 cases of limb synovial sarcoma immunohistochemically and revealed a significant association of c-MYC expression with poor patient prognosis [45]. Synovial sarcoma is histologically divided into monophasic, biphasic, and poorly differentiated subtypes. We previously revealed the significant association of poorly differentiated synovial sarcoma with the expression of *MYC* [46]. Patients with poorly differentiated synovial sarcoma

showed a high risk of recurrence [47]. NCB-0846 may be effective for the treatment of aggressive poorly differentiated synovial sarcoma.

In conclusion, we demonstrated for the first time that TNIK is a feasible drug target in synovial sarcoma. No effective molecular therapeutics have yet been approved for this lethal disease. We observed marked regression of xenografts even after the first oral administration of NCB-846, confirming its high efficacy. The compound is now under preclinical development aimed at investigational new drug (IND) application.

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

#### *4.1. Ethical Issues*

All of the animal experimental protocols in this study were reviewed and approved by the ethics and recombination safety committees of the National Cancer Center Research Institute (Tokyo, Japan) (T-17-022-m01, approved on 21 July 2017). The minimum number of animals necessary to obtain reliable results was used, and maximum attention was paid to animal rights and welfare protection. The use of human materials was reviewed and approved by the Institutional Review Board (IRB) of the National Cancer Center (Tokyo, Japan) (2004-050, approved on 30 October 2014 and revised on 7 November 2019). All patients gave their informed consent at the time. The IRB waived the requirement for obtaining new informed consent for this retrospective study. The investigations were carried out in accordance with the Declaration of Helsinki (https://www.wma.net/what-we-do/medical-ethics/declaration-of-helsinki/).

#### *4.2. Cell Lines*

Human synovial sarcoma HS-SY-II, Aska [48], and Yamato [48] cell lines were obtained from the Riken BioResource Center (Tsukuba, Japan). The SYO-1 cell line was established by one of the authors (A.K.) [49]. All cell lines were maintained in Dulbecco's modified Eagle medium (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10–20% fetal calf serum (Thermo Fisher Scientific). Absence of mycoplasma contamination was routinely confirmed using the e-Myco VALiD Mycoplasma PCR Detection Kit (iNtRon Biotechnology, Seoul, Korea).

#### *4.3. Luciferase Reporter Assay*

A pair of luciferase reporter constructs, super TOP-FLASH and super FOP-FLASH (Addgene, Watertown, MA, USA), was used to evaluate TCF/LEF transcriptional activity. Cells were transiently transfected in triplicate with one of the luciferase reporters and phRL-TK (Promega, Madison, WI, USA) (internal control) [18]. Luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega) and normalized to that of *Renilla reniformis*. Data are presented as the ratio of TOP-FLASH to FOP-FLASH (TOP/FOP ratio).

#### *4.4. Antibodies*

Antibodies used in this study are listed in Table S4.

#### *4.5. Immunoblot Analysis*

Protein samples were fractionated by SDS–PAGE and blotted onto Immobilon-P membranes (Millipore, Burlington, MA, USA) as described previously [50]. After incubation with the primary antibodies at 4 ◦C overnight, the blots were detected with the relevant horseradish-peroxidaseconjugated anti-mouse or anti-rabbit IgG antibody (Cell Signaling Technology, Danvers, MA, USA) and Western lighting ECL Pro (PerkinElmer, Waltham, MA, USA). Signals were visualized with the LAS-4010 system (GE Healthcare, Chicago, IL, USA) and quantified using the ImageJ software package [51]. The uncropped images and relative quantification of blots in Figures 1B, 2F, 3E and 6B are shown in Figures S3–S6, respectively.

#### *4.6. Immunofluorescence Microscopy*

Cells were fixed with 4% paraformaldehyde (PFA) for 10 min and permeabilized in 0.5% Triton X-100 for 3 min. The fixed cells were incubated with a primary antibody overnight at 4 ◦C and subsequently with a relevant secondary antibody (AlexaFluor 488-conjugated anti-rabbit IgG or AlexaFluor 568-conjugated anti-mouse IgG, Invitrogen, Waltham, MA, USA) for 1 h at 37 ◦C. The nuclei were stained with DAPI (Vectashield HardSet Mounting Medium with DAPI, Vector Laboratories, Burlingame, CA, USA). Images were captured using a TCS SP8 confocal microscope (Leica Microsystems, Wetzlar, Germany).

#### *4.7. Patients and Tumor Samples*

The study included tumor tissues surgically resected from 20 patients with synovial sarcoma (10 women and 10 men; median age at diagnosis 50 years, range 5–71 years). Clinicopathological characteristics are summarized in Table S1. The diagnosis of synovial sarcoma was made by a pathologist specialized in soft tissue sarcomas (A.Y.) and confirmed by the detection of *SS18* rearrangement by fluorescence in situ hybridization (FISH) or RT-PCR and/or the reduced expression of SMARCB1 [52,53].

#### *4.8. Immunohistochemistry*

Immunoperoxidase staining was performed using the Ventana DABMap detection kit and an automated slide stainer (Discovery XT, Ventana Medical Systems, Oro Valley, AZ, USA) [54]. The stained tissues were scored as strong positive (++, ≥30%), positive (+, <30%), or negative (−) according to the percentage of tumor cells with nuclear expression (Figure S7).

#### *4.9. Gene Silencing by RNA Interference*

Cells seeded at 50–70% confluency were transfected with siTNIK (s22905, s22906, and s22907; Thermo Fisher Scientific) and siMYC (s9129 and s9130; Thermo Fisher Scientific) at a final concentration of 50 nM in accordance with the manufacturer's instructions.

#### *4.10. Real-Time RT-PCR*

Total RNA was prepared with a RNeasy Plus Mini Kit and treated with RNase-free DNase (Qiagen, Hilden, Germany). The cDNA was synthesized using a High-Capacity cDNA reverse transcription kit (Thermo Fisher Scientific) and subjected to TaqMan gene expression assay using pre-designed primer and probe sets (listed in Table S5). Amplification data measured as an increase in reporter fluorescence were collected using the StepOne™ Real-Time PCR System (Thermo Fisher Scientific). The relative mRNA expression level normalized to the internal control (human β-actin (*ACTB*) gene) was calculated using the comparative threshold cycle (CT) method [18]. Experiments were performed in triplicate and repeated at least two times. Wnt Signaling Targets RT2 Profiler PCR Arrays (Qiagen) were used for pathway-focused gene expression analyses.

#### *4.11. Digital PCR*

Total DNA was extracted from 5 <sup>×</sup> 10<sup>6</sup> cells using the DNA Easy Blood and Tissue kit (Qiagen), in accordance with the manufacturer's instructions. Copy number variation (CNV) data were obtained by the QuantStudio 3D digital PCR system (Life Technologies, Carlsbad, CA, USA) using pre-designed primer and probe sets (listed in Table S6) and analyzed with the QuantStudio 3D Analysis Suite Cloud software (Thermo Fisher Scientific). RNase P (*RPPH1*) was selected as an internal standard gene (Table S6).

#### *4.12. Real-Time Cell Analysis (RTCA)*

Cells were seeded at 5000 cells per well in 96 well clusters one day before transfection with control RNA (siCtrl) or siRNA to *TNIK* (siTNIK) or *MYC* (siMYC) using Lipofectamine RNAiMAX (Invitrogen). Cell growth was monitored periodically by a real-time cell electronic sensing analyzer (xCELLigence, ACEA Biosciences, Santa Clara, CA, USA) for 108 h via calculation of cell index (https://www.aceabio.com/products/icelligence/). Experiments were performed in triplicate and repeated two times.

#### *4.13. Real-Time Monitoring of Transcriptional Activity*

Lentiviral reporter gene transfer was used to evaluate the TCF/LEF transcriptional activity of HS-SY-II after transfection with siCtrl or siTNIK. Cells were infected with TCF/LEF reporter lentiviral particles encoding the GFP gene under control of the TCF/LEF-responsive promoter (Signal Lenti TCF/LEF Reporter (GFP) (Qiagen)) at a multiplicity of infection of 10 in the presence of 4 μg/mL SureEntry Transduction Reagent (Qiagen) for 24 h. GFP-positive cells were cloned by limiting dilution in the presence of 2 μg/mL puromycin (Sigma-Aldrich, St. Louis, MO, USA) and sorted with an S3e cell sorter (BIO-RAD, Hercules, CA, USA). The cells were seeded at a density of 20,000 per well in 96 well plates (Corning, Corning, NY, USA) and transfected with siCtrl or siTNIK. The amount of fluorescence was measured using Incucyte ZOOM (Essen BioScience, Tokyo, Japan).

#### *4.14. Drug Sensitivity*

Cells were seeded at a density of 3000 per well in 96 well plates. Twenty-four hours after seeding, the cells were exposed to serially diluted compounds (0.003, 0.01, 0.03, 0.1, 0.3, 1, 3, and 10 μM) and incubated for 72 h. ATP production was measured using a Cell Titer-Glo Luminescent Cell Viability Assay kit (Promega).

#### *4.15. Xenografts*

Five million HS-SY-II cells suspended in PBS containing 25% Matrigel (BD Biosciences, Franklin Lakes, NJ, USA) were inoculated into the subcutaneous tissues of 6 week old female NOD/SCID (NOD.CB17-Prkdcscid/J) mice. When the average tumor volume reached ~200 mm3, the mice were randomized according to tumor volume (five mice/group) and administered water (vehicle alone) or 80 mg/kg (body weight) NCB-0846 HCl (NCB-1055) dissolved in water by oral gavage twice a day in a 7 day schedule of 5 days on and 2 days off.

#### *4.16. Real-Time Monitoring of Apoptosis Induction*

The Real-Time-Glo™ Annexin V Apoptosis and Necrosis Assay reagent (Promega) was prepared as instructed in its technical manual and added to culture media at the beginning of drug treatment. Luciferase activity was measured every 2 h using the GloMax Discover System (Promega).

#### *4.17. Cell Cycle Analysis*

Cells were dissociated with Accutase, fixed with 70% EtOH at 4 ◦C, stained with Guava Cell Cycle reagent (Merck-Millipore, Burlington, MA, USA) in accordance with the manufacturer's instructions, and analyzed using a Guava easy Cyte HT flow cytometer (Merck-Millipore). Cell doublets were eliminated by doublet discrimination gating. Data were analyzed using the FLOWJO version 10 software package (Treestar, Ashland, OR, USA).

#### *4.18. RNA Sequencing*

Total RNAs were extracted from HS-SY-II cells treated with 3 μM NCB-0846 or 3 μM NCB-0970 for 6 h. After confirming the absence of contamination with genomic DNA using a 2200 TapeStation (Agilent, Santa Clara, CA, USA), the TruSeq Stranded mRNA SamplePrep Kit was used to construct the sequencing library (Illumina, San Diego, CA, USA), and the libraries were sequenced using Illumina NovaSeq 6000 using a NovaSeq 6000 S4 Reagent Kit. Base calling was performed using the Illumina Basecall Software (bcl2fastq2 v2.20) with default parameters. Gene lists extracted from the transcriptome analyses were uploaded to the Database for Annotation, Visualization, and Integrated Discovery (DAVID) Bioinformatics database (https://david.ncifcrf.gov/), and the statistical significance of functional annotation was evaluated. The pathway analysis was performed by displaying the DAVID data on a pathway map of KEGG (Kyoto Encyclopedia of Genes and Genomes (http://www.genome.jp/kegg/). Clustering analysis was performed with MeV (http://mev.tm4.org). GSEA software was used to evaluate the statistical significance of pathway enrichment and to calculate the NES.

#### *4.19. Statistical Analysis*

All statistical analyses were performed using GraphPad Prism 8 (GraphPad, San Diego, CA, USA). Unless otherwise indicated, two-tailed Student's *t*-tests of two groups assuming equal variances were used to calculate *p* values. Differences at *p* < 0.05 were considered significant.

#### **5. Conclusions**

Synovial sarcoma is highly dependent upon the expression of TNIK for cell proliferation and survival, and a small-molecule TNIK inhibitor NCB-0846 induced rapid apoptotic death of synovial sarcoma cells. This study demonstrated for the first time the therapeutic potential of TNIK inhibition in synovial sarcoma.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6694/12/5/1258/s1, Figure S1: Mapping of differentially expressed genes onto the signaling pathways regulating pluripotency of stem cells. Yellow boxes indicate genes downregulated (>2-fold) by NCB-0846, Figure S2: Immunohistochemical analysis of the c-MYC proteins in clinical specimens of synovial sarcoma. Representative cases with strong positive (++) and negative (−) nuclear c-MYC expression are shown. Scale bars: 100 μm in low-power views (left) and 20 μm in high-power views (right), Figure S3. Uncropped immunoblots of Figure 1B. The expression levels of axis inhibition protein 2 (AXIN2) were normalized to those of γ-tubulin, and quantification relative to HS-SY-II is shown below the blots, Figure S4. Uncropped immunoblots of Figure 2F. The expression levels of Traf2-and-Nck-interacting kinase (TNIK) and cleaved poly (ADP-ribose) polymerase-1 (PARP-1) were normalized to those of γ-tubulin, and quantification relative to siCtrl is shown below the blots, Figure S5. Uncropped immunoblots of Figure 3E. The expression levels of cleaved PARP-1 were normalized to those of γ-tubulin, and quantification relative to the dimethyl sulfoxide (DMSO) control is shown below the blots, Figure S6. Uncropped immunoblots of Figure 6B. The expression levels of c-MYC were normalized to those of γ-tubulin, and quantification relative to HS-SY-II is shown below the blots, Figure S7. Scoring of immunohistochemistry. The 20 tissue samples of synovial sarcoma were scored as strong positive (++, ≥30%), positive (+, <30%), or negative (−) according to the percentage of tumor cells with nuclear β-catenin (top), TNIK (middle), and c-MYC (bottom) expression. Scale bars: 20 μm, Table S1: Expression of the β-catenin, TNIK, and c-MYC proteins in clinical specimens, Table S2: Pathway analysis of genes regulated by NCB-0846, Table S3: Regulation of Wnt target genes by NCB-0846, Table S4: List of antibodies used in this study, Table S5: Pre-designed primer and probe sets used for real-time RT-PCR, Table S6: Pre-designed primer and probe sets used for digital PCR.

**Author Contributions:** Conceptualization, T.S., T.Y., E.K. and M.M. (Mari Masuda); Material and clinical data provision, E.K., A.Y., A.K., Y.U., H.M., and M.S.; Methodology development, T.S. and M.M. (Mari Masuda); Experiments, T.S., T.H., and M.M. (Mari Masuda); Writing, T.S., T.Y., and E.K.; Supervision, Y.N., A.T., M.M. (Morio Matsumoto), M.N. and R.N.; Project Administration, M.M. (Mari Masuda); Funding Acquisition, T.S., T.Y. and M.M. (Mari Masuda). All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by the National Cancer Center Research and Development Fund (30-A-2 to M.M. (Mari Masuda))**,** the Acceleration Transformative Research for Medical Innovation (ACT-MS) program of the Japan Agency for Medical Research and Development (AMED) (16im0210804h0001 to T.Y.), the Kobayashi Foundation for Cancer Research (to T.Y.), a KAKENHI Grant-in-Aid for Challenging Research (16K14627 to M.M. (Mari Masuda) and 19H05566 to T.Y.), a Grant-in Aid for Scientific Research (B) (17H03603 to M.M. (Mari Masuda)) from the Japan Society for the Promotion of Science (JSPS), a Cancer Research Grant from the Foundation for Promotion of Cancer Research in Japan (to M.M. (Mari Masuda)), a Research Grant from the Princess Takamatsu Cancer Research Fund (to M.M. (Mari Masuda)), and a Grant-in-Aid for Early-Career Scientists (19J21415 to T.S.) of the Japan Society for the Promotion of Science (JSPS).

**Conflicts of Interest:** U.Y., H.M., and M.S. are employees of Carna Biosciences, Inc. T.Y. and M.M. (Mari Masuda) have received a research grant from Carna Biosciences, Inc. The remaining authors have no conflicts of interest to declare.

#### **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* **Wnt/**β**-Catenin Signaling and Immunotherapy Resistance: Lessons for the Treatment of Urothelial Carcinoma**

**Alexander Chehrazi-Raffle, Tanya B. Dorff, Sumanta K. Pal and Yung Lyou \***

Department of Medical Oncology & Experimental Therapeutics, City of Hope Comprehensive Cancer Center, Duarte, CA 91010, USA; achehraziraffle@coh.org (A.C.-R.); tdorff@coh.org (T.B.D.); spal@coh.org (S.K.P.) **\*** Correspondence: ylyou@coh.org; Tel.: +1-626-256-4673; Fax: +1-626-301-8233

**Simple Summary:** Metastatic urothelial cell carcinoma (UCC) is a significant public health burden with a median survival estimated at about 15 months. The use of immunotherapy with immune checkpoint inhibitors has greatly improved outcomes but only benefits a minority (~20%) of patients. In this review we discuss the evidence showing how a key molecular pathway known as Wnt/βcatenin signaling can be a driver of immunotherapy resistance and how these insights can serve as lessons for improving future treatment of urothelial carcinoma.

**Abstract:** Urothelial cell carcinoma (UCC) is a significant public health burden. It accounts for approximately 90 percent of all bladder cancers with an estimated 200,000 annual deaths globally. Platinum based cytotoxic chemotherapy combinations are the current standard of care in the frontline setting for metastatic UCC. Even with these treatments the median overall survival is estimated to be about 15 months. Recently, immune checkpoint inhibitors (ICIs) have demonstrated superior clinical benefits compared to second line chemotherapy in UCC treatment. However only a minority of patients (~20%) respond to ICIs, which highlights the need to better understand the mechanisms behind resistance. In this review, we (i) examine the pathophysiology of Wnt/β-catenin signaling, (ii) discuss pre-clinical evidence that supports the combination of Wnt/β-catenin inhibitors and ICI, and (iii) propose future combination treatments that could be investigated through clinical trials.

**Keywords:** Wnt; β-catenin; urothelial cancer; immune checkpoint inhibitor; immunotherapy resistance

#### **1. Introduction**

Urothelial cell carcinoma (UCC) is the most common malignancy of the urinary system. It accounts for approximately 90 percent of all bladder cancers with an estimated 200,000 annual deaths globally [1,2]. UCC is also an aggressive histology as 25% of patients who receive potentially curative treatment for localized disease will unfortunately succumb to tumor metastasis.

Cytotoxic chemotherapy is the current standard of care in the frontline setting for metastatic UCC. The median overall survival is estimated to be about 15 months with modern chemotherapy regimens containing platinum-based agents [3,4]. Once patients progress on first line chemotherapy treatments the second line chemotherapies have limited efficacy with median progression-free survival periods of 3–4 months (Figure 1) [5,6].

More recently, immune checkpoint inhibitors (ICIs) have demonstrated superior clinical benefits compared to second line chemotherapy in UCC treatment [7,8]. However, only a minority of patients (~20%) respond to ICIs in the treatment of UCCs and other malignancies [7–9]. It has also been noted that those patients who respond to ICIs can often maintain an impressive durable response lasting more than 14–15 months [7,8]. This phenomenon has been observed across numerous cancer subtypes [10], highlighting the need to better understand the mechanisms behind ICI resistance.

**Citation:** Chehrazi-Raffle, A.; Dorff, T.B.; Pal, S.K.; Lyou, Y. Wnt/β-Catenin Signaling and Immunotherapy Resistance: Lessons for the Treatment of Urothelial Carcinoma. *Cancers* **2021**, *13*, 889. https://doi.org/10.3390/ cancers13040889

Academic Editor: David Wong

Received: 30 November 2020 Accepted: 1 February 2021 Published: 20 February 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/).

**Figure 1.** Current systemic treatments in metastatic urothelial cell carcinoma.

Several mechanisms of ICI resistance in cancers have been reviewed extensively elsewhere [11,12]. Previously proposed resistance pathways include PTEN, FGF, MYC, TGFB, TP53, WNT, VEGF, and ANG2 [11,12]. The majority of studies investigating immunotherapy resistance mechanisms have been done in non-UCC studies; as a result, the proposals in this review extrapolate data derived from both urothelial and non-urothelial studies. In this review, we will (i) examine the pathophysiology of Wnt/β-catenin signaling, (ii) discuss pre-clinical evidence that supports the combination of Wnt/β-catenin inhibitors and ICI, and (iii) propose future combination treatments that could be investigated through clinical trials.

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

Wnt signaling is a highly coordinated and conserved signaling cascade that occurs at the cell surface and within the cytoplasm. This pathway mediates an array of biological functions, including cell fate decisions during embryonic development, stem cell equipoise, and immune system homeostasis [13–16]. Recent reviews published elsewhere provide a more exhaustive discussion on the β-catenin dependent and independent pathways [17–22]. For the purposes of this review (which is most relevant to ICI resistance) we will focus primarily on β-catenin-dependent Wnt signaling.

Canonical, or β-catenin-dependent, Wnt signaling is one of the primary sources of dysregulated transcription in cancer. In the "on-state", the signal cascade begins at the cell surface with Wnt ligands binding to the Frizzled:LRP5/LRP6 receptor complexes, and culminates in the nucleus with the formation of a transcription-activating complex [23]. The primary mediator of this cell surface-to-nucleus signal is β-catenin, a membrane/cytoplasmic armadillo repeat protein which lacks the ability to independently promote DNA transcription [17,20,24]. Instead, β-catenin is trafficked into the nucleus to DNA-binding T-cell factor (TCF)/lymphoid enhancer binding factor (LEF) transcription factors [24,25].

Once bound to DNA by TCF/LEFs, β-catenin recruits other co-activators and regulatory components that collectively activate transcription of the downstream genes known as the Wnt target genes. These sets of Wnt target genes drive cells to proliferate, self-renew, differentiate and survive in a variety of tissues and contexts. In normal cells, feedback inhibition results in this activity occurring only transiently, which in turn prevents overactivation of Wnt target gene transcription. Signal transduction is thus "turned off" in cells with low or absent Wnt because β-catenin becomes unstable by being tagged in the cytoplasm for ubiquitination by the destruction complex, which then leads to proteasome degradation.

However, in various cancers (i.e., colon cancer) mutations in the destruction complex components (e.g., *APC*, *AXIN2* and *FAM123B/WTX*) or regulators of the receptors/ligand (e.g., *RNF43/ZNRF3*, *RSPO2,* or *RSPO3*) components can lead to unchecked Wnt signaling. These mutations negate the cytoplasmic feedback controls and create cells with constitutive, high levels of β-catenin and aberrantly high levels of Wnt target gene transcription that can initiate carcinogenesis and immune suppression [20,26–29].

#### **3. Upregulation of Wnt/**β**-Catenin in Bladder Carcinogenesis**

Several correlative studies have shown conflicting evidence between upregulation of Wnt/β-catenin signaling and UCC carcinogenesis [30–34]. For instance, The Cancer Genome Atlas (TCGA) Research Network detected Wnt signaling alterations in 73% of UCC tumors [35]. However, Ahmad et al. noted Wnt signaling in only 33% of their clinical UCC samples [36–38]. The discrepancy could most likely be due to comparisons using different methods and patient populations. For example, one Ahmad et al. study used a tissue microarray array with core biopsy samples, whereas a TCGA study detected aberrations in Wnt signaling through genomics using RNA-seq and whole exome sequencing [35–38]. Also there was a difference in sample size with TCGA and Ahmad et al. studies using 131 and 60 patient samples respectively [35–38]. Additionally it was noted that β-catenin expression's correlation to tumor grade and muscle invasion has been inconsistent [39]. Despite these discrepancies between studies, it is evident that a substantial proportion of UCC develops in the context of Wnt signaling aberrations.

From a pathophysiologic perspective, numerous pre-clinical studies have implicated the silencing of endogenous Wnt inhibitors as potential oncogenic events. CpG hypermethylation of the WIF1 (Wnt inhibitory factor-1) promoter was found to lead to decreased transcription and increased Wnt signaling activity in human bladder cancer cell lines [40]. Knockdown of WIF1 by siRNA in bladder cancer cell lines led to increased activity in c-myc and cyclin D1 mRNA transcription and increased cell growth [40]. These results suggested that Wnt signaling via WIF1 could potentially promote development of UCC [40]. Another proposed mechanism involves aberrations in the oncogene activation-induced cytidine deaminase, which upregulates the Wnt/β-catenin pathway and thereby promotes UCC growth [41]. More studies are needed to better understand how Wnt signaling can drive urothelial carcinogenesis.

#### **4. Wnt/**β**-Catenin Induces Immune Cell Exclusion in Urothelial Cancer**

Due to the limited efficacy of ICI treatments, much effort is being dedicated to developing predictive biomarkers of response and understanding the biological mechanisms for resistance. One widely established predictive biomarker for ICI response is intratumoral enrichment of CD8+ T-cells prior to treatment [42,43]. Therefore, many studies have used the presence and quantity of CD8+ T-cell infiltration as a surrogate marker when performing correlative studies to determine if other molecular pathways may be involved in predicting the ICI response.

A recent study by Sweis et al. used a bioinformatics approach to correlate CD8+ T-cell infiltration with various signaling pathways [44]. The investigators analyzed the whole exome sequencing (WES) and RNA-seq transcriptional profile data from the 267 samples of urothelial bladder cancer collected for the TCGA study. The investigators stratified these tumors based on a 160-gene T-cell inflamed expression signature indicative of a T-cell inflamed and non-inflamed microenvironment. This T-cell inflamed gene signature was then validated by performing immunohistochemistry (IHC) staining for CD8+ T-cell infiltration on a sample of 19 tumors (7.1%).

Once stratified into inflamed vs. non-inflamed phenotypes, the investigators uncovered that 730 genes were preferentially expressed in the non-T-cell-inflamed tumors. Ingenuity pathway analysis then showed that one of the top upstream regulators for these groups of differentially expressed genes were those that were regulated by β-catenin/Wnt signaling. The authors then went back to the 19 samples which they had initially performed CD8+ T-cell IHC staining and co-stained for nuclear β-catenin as a marker for active β-catenin dependent Wnt signaling. The investigators found a statistically significant

inverse relationship between nuclear β-catenin and the density of CD8+ T cells infiltrating the tumor.

To further validate the Wnt signaling pathway as a mediator of non-T-cell-inflamed tumor microenvironments, a follow up study done by Luke et al. employed a similar approach (WES genomics and RNA-seq transcriptional profiling) and analyzed 9244 samples across 31 different types of cancers [45]. The investigators used their previously developed 160-gene T-cell inflamed expression signature to segregate the samples into T-cell inflamed, intermediate, or non-T-cell inflamed. The investigators defined Wnt/β-catenin signaling activation at three different levels: assessment of somatic mutations or copy number alterations in *CTNNB1* (gene for β-catenin) and other regulatory genes predicted to result in pathway activation, expression of downstream Wnt target genes, and β-catenin protein levels which were assessed through reverse phase protein array (RPPA). With respect to the 363 UCC samples included in this cohort, all three levels correlated with a non-T-cell inflamed tumor signature, the most pronounced of which was CTNNB1 protein level. Taken together, these findings suggest that there is a significant correlation between upregulation of Wnt signaling and a non-T-cell-inflamed microenvironment in UCC.

#### **5. ICI Attenuation via CCL4**

As previously discussed, translational studies have suggested that Wnt/β-catenin signaling may induce a non-T-cell-inflamed tumor phenotype thereby excluding immune cells from the tumor microenvironment and dampening the therapeutic effect of ICIs. To elucidate molecular mediators, the Gajewski group used a genetically engineered melanoma mouse model with active β-catenin signaling (BRAF/PTEN/CAT-STA) in the tumors [46].

In their mouse model, the authors found that β-catenin signaling activation was associated with low levels of tumor infiltrating CD8+ T-cells. Conversely, mice in which β-catenin signaling was absent contained a high density of CD8+ T-cell infiltration. In order to discern if this was due to differences in neo-antigens, the authors introduced a neoantigen (SIY) expressing construct genetically into the tumors and adoptively transferred T-cells with SIY T cell receptor. They found that the transferred T-cells accumulated in the BRAF/PTEN-STA tumors but not the β-catenin expressing BRAF/PTEN/Bcat-STA tumors despite both tumors now expressing the neo-antigen. Furthermore, anti-PD-1 and anti-CTLA-4 agents were rendered ineffective in the Wnt-activated (BRAF/PTEN/Bcat-STA) mice but remained effective in Wnt-inactivated (BRAF/PTEN-STA) mice. These results suggested that upregulation of Wnt/β-catenin may indeed induce resistance to immune checkpoint inhibition.

The investigators then queried whether this blunted response to ICIs could be dependent on antigen presentation from CD103+ dendritic cells (DC). Within Wnt/β-cateninactivated T-cell-depleted tumors, they found that CD103+ DCs were nearly absent and IFN-β cytokine expression was reduced. The investigators then found that intratumoral injection of CD103+ DCs led to restoration of T-cells infiltration within the tumor. This supported the role of CD103+ dendritic cells as key mediators of an antitumor immune response. To characterize the mechanism of failed recruitment of the CD103+ DCs, the investigators analyzed the gene expression of these two tumor types and found that four chemokines (CCL3, CXCL1, CXCL2, and CCL4) were lower in the non-T-cell inflamed BRAF/PTEN/Bcat-STA tumors. Of these four chemokines, only CCL4 was found on an in vitro DC migration assay to possess the ability to effectively modulate cell migration.

Furthermore, the investigators found that the Wnt signaling target gene ATF3–which also binds at the promoter region of the CCL4 gene—was expressed at higher levels in the β-catenin activated melanoma tumors. This negative feedback was substantiated by then demonstrating that gene knockdown of ATF3 and CTNNB1 in melanoma cell lines led to upregulation of CCL4 expression (Figure 2).

**Figure 2.** Wnt/β-catenin signaling can alter T-cell infiltration status and ICI response via CCL4. (Created with BioRender®).

#### **6. Wnt/**β**-Catenin Signaling Induces Immune Cell Exclusion by Affecting the Tumor Microenvironment (TME)**

Tumor-associated macrophages (TAMs) are amongst the most common tumor immune infiltrating cells in the tumor microenvironment (TME) [47]. TAMs are classically thought to exist in two polarized states with the activated M1 and M2 subtypes [47]. The M1 subtypes are thought to play a significant role in the anti-tumor immune response by producing reactive oxygen species (ROS) and pro-inflammatory cytokines [47]. The M2 subtype has been found to have an opposite immunosuppressive function by producing anti-inflammatory cytokines (i.e., IL1, IL-13, and TGF-β) which can promote tumor growth and ICI resistance [47]. These anti-inflammatory cytokines and chemokines can also induce the production of regulatory T-cells which directly inhibit cytotoxic T cells further driving immunosuppression [47–49].

It has been shown that Wnt/β-catenin signaling can modulate the TAMs population in the TME leading to a protumoral phenotype which may be ICI resistant [50,51]. In a study done by Kaler et al. using isogenic colon cancer cell lines (HCT116 and Hke-3 cells) with mutated active β-catenin, the investigators found that TAMs could further enhance the pre-existing Wnt/β-catenin signaling present and protect the cancer cells from TRAILinduced apoptosis [50]. In contrast, when HCT116 cancer cells with an inactive β-catenin allele were cultured with TAMs, the investigators noted that these cells were susceptible to TRAIL induced apoptosis and were unable to increase their Wnt/β-catenin signaling levels [50]. The investigators also found that the isogenic colon cancer cell lines (HCT116 and Hke-3 cells) with mutated active β-catenin when cultured with TAMs would produce more snail protein, which is a known Wnt/β-catenin signaling target gene and driver of tumor mesenchymal transition [50]. These results suggested that the increased Wnt/βcatenin signaling from the TAMs could induce snail gene expression and drive a tumor mesenchymal transition phenotype [50]. Of note this nail driven tumor mesenchymal transition has recently been reported to be a possible mechanism for ICI resistance [50–52].

Another potential mechanism for ICI resistance is through the tumor's ability to create a hostile TME that is acidic from increased lactic acid production which can lead to impaired cytotoxic T-cell function [53–55]. A detailed discussion on how tumors create a hostile hypoxic and acidic TME which leads to suppression of the T-cells' cytotoxic

function is beyond the scope of this manuscript. For a more comprehensive review on this topic there are many excellent reviews which can be found in the reference section of this manuscript [53,56,57]. Briefly, the oncogenic mutations that drive carcinogenesis (i.e., Akt/PI3k/mTOR and Wnt/β-catenin signaling) have also been shown to drive a metabolic reprogramming of cells from oxidative phosphorylation towards aerobic glycolysis [56,58]. This phenomenon wherein cancer cells prefer to undergo the more inefficient aerobic glycolysis even in the presence of oxygen has been known for almost 100 years since it was first described by Dr. Otto Heinrich Warburg [57,59]. It is thought that cancer cells have evolved this shift towards aerobic glycolysis as a way to produce metabolic byproducts which can then be converted to provide the needed biomass to use as building blocks for its rapid cell proliferation [56,57]. As the tumor grows larger in size its metabolic demands also increase in an unregulated manner which often outstrips the local oxygen and nutrient supply [53,56]. This imbalance in metabolic demand and available supply of local resources creates a hostile TME that is hypoxic, acidic (due to lactic acid build up), and nutrient deficient [53,56]. In addition, to the existing overactive oncogenic signaling pathways present in the cancer cells (i.e., Akt/PI3k/mTOR and Wnt/β-catenin signaling) these hostile TME conditions will further drive the tumors to adapt by increasing angiogenesis and glycolysis via the VEGF and HIF signaling pathways [53,56]. These same acidic and hypoxic TME conditions will then inhibit the oxidative phosphorylation that is needed by T-cells in order to perform their cytotoxic functions potentially leading to immunosuppression and ICI resistance [53]. In fact it has been shown that high lactate concentrations in the TME can impede the CD8+ T-cells ability to export lactate and suppress their natural cytotoxic function [60].

As discussed above, the Wnt/β-catenin signaling pathway was found to initially play a central role in carcinogenesis by driving cell proliferation [20]. More recently, it has also been found to play an additional role in cancer metabolism by metabolically reprogramming cancer cells to promote aerobic glycolysis and lactic acid production [58,61–63]. In a study done by Pate et al. the investigators found that by using genetically engineered human colon cancer cell lines that overactive Wnt/β-catenin signaling drives aerobic glycolysis and lactic acid production by upregulating the genes pyruvate dehydrogenase kinase 1 (PDK1) and monocarboxylate transporter 1 (MCT1/SLC16A1) [58,61]. They also found that when this metabolic shift towards glycolysis occurred that there was also a corresponding inhibition in the gene expression of pyruvate dehydrogenase (PDH) and oxidative phosphorylation [58,61]. Other independent studies have also provided further supporting evidence that Wnt/β-catenin signaling can drive the metabolic reprogramming of cancer cells towards lactic acid production and aerobic glycolysis [62,63].

It has also been shown that the lactic acid in the TME can play a role in immunosuppression and drive further tumor growth [53,54,64]. In a study done by Brand et al. the investigators found that patients who had melanoma tumors with increased LDHA gene expression and lactic acid levels were more likely to have findings of impaired T and NK cell infiltration consistent with an immunosuppressed or immune deficient tumor phenotype [54]. The investigators then used shRNA to create LDHlow murine melanoma and pancreatic cancer cell lines [54]. Through the use of various clever control experiments the investigators showed that knockdown of the LDHA gene resulted in a stable tumor cell phenotype that produced low levels of lactate with no effects on the other metabolic pathways analyzed [54]. They then proceeded to inject these murine melanoma and pancreatic cells lines which were either LDHhigh or LDHlow into syngeneic mice [54]. They found that the LDHlow had impaired tumor growth and higher T-cell and NK cell infiltration compared to the LDHhigh tumors [54]. These findings suggested that the acidic TME created by uncontrolled lactate production led to impaired immunosurveillance and T-cell and NK cell infiltration leading to an immune deficient TME [54]. In another independent study done by Harel et al. the investigators found that increased oxidative phosphorylation and lipid metabolism in melanoma tumors by proteomic analysis were more likely to have potentiated antigen presentation and response to anti-PD1 immune checkpoint inhibitor

or TIL-based immunotherapy [64]. The investigators of the Harel et al. study concluded that the tumors with increased oxidative phosphorylation were undergoing less glycolysis, secreting less lactate, and creating a more favorable TME for immune cells [64].

The above studies provide evidence supporting the hypothesis that the presence of lactic acid in the TME can be immunosuppressive by inhibiting the needed oxidative phosphorylation of cytotoxic T-cells. As a result, this has led to the proposal that targeting lactic acid production could be a potential way to overcome ICI resistance [55]. In summary, the above findings provide evidence that Wnt/β-catenin signaling can drive ICI resistance by modulating the TME through the interaction with TAMs or driving lactic acid production and creating a local immunosuppressive environment for cytotoxic T-cells (Figure 3) [50,53–58,61,64].

**Figure 3.** Wnt/β-catenin signaling can alter tumor microenvironment. (Adapted from "Tumor Microenvironment", by BioRender.com (2020). Retrieved from https://app.biorender.com/biorender-templates).

#### **7. Overcoming ICI Resistance with** β**-Catenin Inhibition**

The above mentioned studies provide strong evidence that the Wnt/β-catenin signaling pathway drives immune cell exclusion which can then lead to immune checkpoint inhibitor resistance in cancer treatment [44,45,65]. As a result, one could reason that combining a Wnt/β-catenin signaling inhibitor and ICI may lead to overcoming this resistance mechanism (Figure 4).

Early therapeutic efforts primarily centered on finding targets for Wnt inhibition [23,66–68]. However, one of the major hurdles that researchers encountered was developing a molecule small enough to penetrate the nuclear membranes yet robust enough to counteract the large β-catenin regulatory complex [23,68,69]. Another challenging adverse class effect was on-target bone toxicity, which ultimately led to the early termination of several Phase I studies [23,68,70–72].

**Figure 4.** Proposed model for overcoming Wnt signaling driven immune checkpoint inhibitor resistance. (**A**) Immune cell exclusion driven by Wnt signaling. (**B**) Combination of Wnt signaling inhibitor and immune checkpoint inhibitor can overcome resistance.

More recently, several studies have shifted focus toward downstream inhibition of the intranuclear transcriptional β-catenin complex to enhance immune cell infiltration within the tumor microenvironment. For instance, Ganesh et al. designed a β-catenin inhibitor (DCR-BCAT) that selectively silenced CTNNB1 (the gene which transcribes/βcatenin) in tumors using an RNAi oligonucleotide [73]. Using allografted B16F10 mouse melanoma cells on immunocompetent C57BL/6 mice, which are known to be refractory to ICI treatments through T-cell exclusion [73], Ganesh et al. found that treatment with DCR-BCAT significantly increased the intratumoral density of CD8+ T-cells compared to the placebo control. Quantitative analysis of tumor RNA detected a decrease in β-catenin gene expression as well as a concomitant increase in CCL4 expression. Furthermore, single-cell flow cytometry of the DCR-BCAT mouse tumors showed a significant increase in CD8+, CD3+, CD103+, and PD-1 positive cells, suggesting that these tumors were transitioning to a T-cell-inflamed phenotype.

Encouraged by these results, the investigators subsequently examined if their βcatenin inhibitor could reconstitute an immune response within their T-cell-excluded tumor model. Although monotherapy with either the DCR-BCAT or an ICI was minimally effective, the combination of DCR-BCAT plus ICI elicited a synergistic effect with reductions in tumor size by as much as 87% [73]. Moreover, the authors confirmed that this combination was effective in another model, the Neuro2A (neuroblastoma) cell lines, which are also non-T-cell-inflamed at baseline [73]. These findings suggest that a β-catenin inhibitor can effectively downregulate Wnt/β-catenin signaling and induce a T-cell-inflamed phenotype that can potentiate a response to immune checkpoint inhibitors [73].

#### **8. Ongoing Clinical Trials and Future Directions**

In recent years, several novel agents with varied mechanisms of action have attempted to mitigate the immunosuppressive tumor microenvironment through WNT/β-catenin inhibition (Figure 5, Table 1). One such therapeutic effort in development is DKN-01, an antibody that antagonizes the WNT/β-catenin pathway through inhibition of DKK1 [74]. Preliminary results from a Phase 1b/2a study of DKN-01 plus pembrolizumab (NCT02013154) demonstrated a disease control rate of 80% in patients who had tumors with high DKK1 expression as compared to a disease control rate of 20% in patients with low DKK1 expression [74].

**Figure 5.** Current WNT/β-catenin inhibitors being used in combination with ICI for human clinical trial. (Adapted from "Wnt//β-catenin signaling", by BioRender.com (2020). Retrieved from https: //app.biorender.com/biorender-templates).

**Table 1.** Current clinical trials combining Wnt inhibitor and immune checkpoint inhibitor.


Another class of WNT/β-catenin inhibitors disrupt PORCN, an enzyme that facilitates WNT secretion [75]. A recent Phase I study of the PORCN inhibitor WNT974 combined with the PD-1 monoclonal antibody spartalizumab (NCT01351103) reported impressive results across several solid tumors, including stable disease in 53% of patients who were previously refractory to ICIs [76]. Of note, neither one of these trials included urothelial carcinoma and focused on other malignancies such as GI cancers, melanoma, and NSCLC. However, seeing how the combination of ICIs with WNT/β-catenin inhibitors has produced some signal of efficacy even in the early phase clinical trials, this combination warrants further investigation for the treatment of UCC.

#### **9. Conclusions**

In summary, WNT/β-catenin signaling can drive immune cell exclusion and may be a resistance mechanism for immune checkpoint inhibitors. Several preclinical studies have shown that inhibition of the WNT/β-catenin pathway in conjunction with an ICI can effectively overcome this resistance mechanism. With respect to UCC, this combination is particularly promising given the high frequency of WNT/β-catenin aberrations in correlative studies as well as its potential role in upregulating urothelial oncogenesis. Thus, to complement ongoing clinical trials across other solid tumors, additional studies that validate the synergistic relationship of ICIs and WNT/β-catenin inhibitors in UCC are urgently needed.

**Author Contributions:** Y.L., A.C.-R., S.K.P. and T.B.D. wrote and revised the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** Y.L. is supported by the American Cancer Society.

**Conflicts of Interest:** Y.L., T.B.D. and A.C.-R. have no conflicts of interest that might be relevant to the contents of this manuscript. S.K.P.: Consulting or Advisory Role: Genentech, Bristol-Myers Squibb.

#### **References**


## *Article* **Methylation Patterns of** *DKK1***,** *DKK3* **and** *GSK3β* **Are Accompanied with Different Expression Levels in Human Astrocytoma**

**Anja Kafka 1,2,\*, Anja Bukovac 1,2, Emilija Brglez 1, Ana-Marija Jarmek 1, Karolina Poljak 1, Petar Brlek 1, Kamelija Žarkovi´c 3,4, Niko Njiri´c 1,5 and Nives Pe´cina-Šlaus 1,2**


**Simple Summary:** Astrocytomas are the most common type of primary brain tumor in adults. In this study, 64 astrocytoma samples of grades II–IV were analyzed for genetic and epigenetic changes as well as protein expression patterns in order to explore the roles of the Wnt pathway components, such as DKK1, DKK3, GSK3β, β-catenin, and APC in astrocytoma initiation and progression. Our findings on *DKK1* and *DKK3* show the importance of methylation in the regulation of Wnt signaling activity and also indicate pro-oncogenic effects of GSK3β on astrocytoma development and progression. Close connections between large deletions and mutations in the APC gene and increased β-catenin expression in glioblastoma were also established. Our results suggest that Wnt pathway related genes and proteins play an active role in the etiology of astrocytic brain tumors.

**Abstract:** In the present study, we investigated genetic and epigenetic changes and protein expression levels of negative regulators of Wnt signaling, *DKK1*, *DKK3*, and *APC* as well as glycogen synthase kinase 3 (GSK3β) and β-catenin in 64 human astrocytomas of grades II–IV. Methylation-specific PCR revealed promoter methylation of *DKK1*, *DKK3*, and *GSK3β* in 38%, 43%, and 18% of samples, respectively. Grade IV comprised the lowest number of methylated *GSK3β* cases and highest of *DKK3*. Evaluation of the immunostaining using H-score was performed for β-catenin, both total and unphosphorylated (active) forms. Additionally, active (pY216) and inactive (pS9) forms of GSK3β protein were also analyzed. Spearman's correlation confirmed the prevalence of β-catenin's active form (rs = 0.634, *p* < 0.001) in astrocytoma tumor cells. The Wilcoxon test revealed that astrocytoma with higher levels of the active pGSK3β-Y216 form had lower expression levels of its inactive form (*p* < 0.0001, Z = −5.332). Changes in *APC's* exon 11 were observed in 44.44% of samples by PCR/RFLP. Astrocytomas with changes of *APC* had higher H-score values of total β-catenin compared to the group without genetic changes (t = −2.264, *p* = 0.038). Furthermore, a positive correlation between samples with methylated *DKK3* promoter and the expression of active pGSK3β-Y216 (rs = 0.356, *p* = 0.011) was established. Our results emphasize the importance of methylation for the regulation of Wnt signaling. Large deletions of the *APC* gene associated with increased β-catenin levels, together with oncogenic effects of both β-catenin and GSK3β, are clearly involved in astrocytoma evolution. Our findings contribute to a better understanding of the etiology of gliomas. Further studies should elucidate the clinical and therapeutic relevance of the observed molecular alterations.

**Keywords:** astrocytic brain tumors; Wnt signaling; DKKs; GSK3β; APC; β-catenin

**Citation:** Kafka, A.; Bukovac, A.; Brglez, E.; Jarmek, A.-M.; Poljak, K.; Brlek, P.; Žarkovi´c, K.; Njiri´c, N.; Pe´cina-Šlaus, N. Methylation Patterns of *DKK1*, *DKK3* and *GSK3β* Are Accompanied with Different Expression Levels in Human Astrocytoma. *Cancers* **2021**, *13*, 2530. https://doi.org/10.3390/ cancers13112530

Academic Editors: Michael Kahn and Keane Lai

Received: 2 May 2021 Accepted: 19 May 2021 Published: 21 May 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/).

#### **1. Introduction**

Astrocytomas are glial cell tumors originating from astrocytes and account for nearly half of all primary brain tumors. According to the latest World Health Organization (WHO) classification, there are three different grades of astrocytoma, indicating their growth potential and aggressiveness [1,2]. Diffuse astrocytoma, defined as a grade II neoplasm, is a type of low-grade infiltrative glioma. Grade II astrocytomas have a tendency to progress toward high-grade malignancies called anaplastic astrocytomas (grade III) and eventually secondary glioblastomas (GBM, grade IV). The proliferative potential of diffuse astrocytomas and their growth rate are much lower than those of GBMs, which are a highly aggressive tumor with pronounced brain invasion and fast progression [1]. In addition to the biological behavior, an important criterion for the classification of diffuse glioma is the status of *IDH1* and *IDH2* gene mutations; astrocytomas are now defined as *IDH* mutant or *IDH* wild-type. Low-grade astrocytomas and secondary GBMs often carry *IDH* mutations, associated with younger age, as well as a much better prognosis [3,4]. *IDH* wild-type status refers to 90% of GBMs and indicates a primary tumor that arises de novo and carries a poorer prognosis than those classified as *IDH* mutant.

Despite new molecular findings that characterize tumors in the group of diffuse gliomas, the differences between individual pathohistological grades are still insufficiently investigated. For this reason, we decided to study the molecular characteristics of the WNT and AKT signaling pathway components in astrocytomas of different grades.

Signaling pathways form a complex network of molecular interaction in our cells, and a close connection between Wnt/β-catenin and PI3K/AKT/mTOR signaling has been described in many cancers [5]. One of the most prominent linking elements between these pathways is GSK3β (glycogen synthase kinase 3) [6]. The major mode of GSK3β activity regulation is through phosphorylation events. Activated Akt molecule phosphorylates GSK3β on the amino acid serine 9 (S9), leading to GSK3β inactivation. In contrast, GSK3β is activated by autophosphorylation or phosphorylation on tyrosine 216 (Y216) by other kinases [5,7]. In addition to S9 phosphorylation, promoter methylation may also be one of the mechanisms of GSK3β inactivation [8].

In the Wnt signaling pathway, GSK3β plays a key role in modulating β-catenin and TCF/LEF (T cell factor/lymphoid enhancer-binding factor) transcription factor activity [7]. Active GSK3β can act as a tumor suppressor as it participates, together with other members of the destruction complex including APC (Adenomatous Polyposis Coli), AXIN1 and CK1 (Casein Kinase 1), in phosphorylation and subsequent degradation of the oncogenic β-catenin protein. In the pathway's "off" state, TCF/LEF is inactive due to its interaction with the repressor Groucho. In contrast, inactive GSK3β stimulates cell proliferation in the pathway's "on" state. The pathway is activated upon binding of Wnt ligands to the Frizzled (Fz) receptor and the co-receptor lipoprotein receptor-related protein (LRP) 5/6, resulting in the disintegration of the destruction complex and β-catenin cytoplasmic accumulation. Afterward, unphosphorylated β-catenin enters the cell nucleus, where it interacts with transcription factors from the TCF/LEF family, leading to Wnt target gene transcription (*cyclin D1*, *c-myc*, *fra-1*, *c-jun*, etc.) thus stimulating tumor growth [9]. Except for phosphorylating β-catenin, GSK3β can phosphorylate LRP co-receptor, thus revealing a binding site for AXIN on LRP, which mimics pathway activation by a Wnt ligand [10] (Figure 1).

Wnt signaling activity is also regulated by evolutionarily conserved inhibitors and activators that antagonize Wnt signaling such as the Dickkopf (DKK) gene family. The family consists of four members (DKK 1–4) in humans that specifically inhibit the Wnt/βcatenin signaling cascade by preventing the Wnt ligand from binding to LRP 5/6 coreceptors. Some members of the DKK family interact with transmembrane proteins Kremen 1 and 2, also modulating the pathway's activity [11]. Not all DKK family members have consistent roles. Recent reports reveal that they can have dual agonistic and antagonistic functions, depending on the cellular context. Numerous studies report on changes in DKK protein expression within tumor tissues. DKK1 is differentially expressed in different types

of human cancers, and its expression affects cell invasion, proliferation, and tumor growth. Some authors have reported on DKK1 overexpression [12–21], while others have noted its downregulation in tumors [22–24]. On the other hand, DKK3 is omnipresent in normal human tissues, including the brain; however, it is significantly depleted in various cancer cell types. DKK3 silencing due to epigenetic alterations has also been reported in multiple cancers [12]. However, there are few studies investigating the expression of DKK1 and DKK3 in gliomas [25–28].

**Figure 1.** Overview of Wnt signaling pathway. (**a**) In the canonical Wnt pathway, DKK directly competes with Wnt for binding to LRP6. When DKK binds to the receptor, cytosolic pool of β-catenin is maintained at low levels through proteasomal degradation, due to its phosphorylation by the complex consisting of Axin/APC/CK1/GSK-3β. (**b**) Binding of Wnt to receptors Fz/LRP leads to the recruitment of components of the destruction complex to the membrane. This prevents phosphorylation and degradation of β-catenin, resulting in its accumulation in the cytoplasm. Stabilized β-catenin translocates into the nucleus and activates transcription of Wnt target genes.

These opposite reports indicate the need for further elucidation of the role of Wnt signaling molecules in cancer. Our study aims to contribute to the great efforts that are being made to clarify the genetic and epigenetic signatures in gliomas. Our goal was to clarify the behavior of *DKK1*, *DKK3*, and GSK3β and identify potential correlations to changes of *APC* and beta-catenin genes and proteins.

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

#### *2.1. Tissue Samples*

Sixty-four astrocytoma samples of different pathohistological types and grades, together with corresponding blood and formalin-fixed paraffin-embedded (FFPE) slides of brain tumor tissues, were collected with patients' consents from the Departments of Neurosurgery and Departments of Pathology University Hospital Centers "Zagreb" and "Sisters of Charity".

Chosen slides were reviewed by a certified pathologist (KŽ) to confirm the diagnosis (diffuse astrocytoma, anaplastic astrocytoma, glioblastoma). The diagnoses of astrocytic brain tumors were in concordance with the most recent WHO classification of the tumors of the central neural system [3]. In selected cases, additional immunohistochemical analyses (IDH1/2, ATRX, p53) were conducted in order to provide the correct diagnosis. The patients included in the study had no family history of brain tumors and did not undergo any cancer treatment prior to surgery, which could affect the results of molecular analyses. The sample consisted of 10 diffuse astrocytomas (grade II), 11 anaplastic astrocytomas (grade III), and 43 glioblastomas (grade IV). Twenty seven patients were female and 37 male. The age of patients varied from 6 to 83 (mean age = 50.31, median = 54 years). The mean age at diagnosis for females was 54.85 (median 56) and for males, 47 years (median 49).

The study was approved by the Ethical Committees, School of Medicine University of Zagreb (Case number: 380-59-10106-14-55/147; Class: 641-01/14-02/01) and University Hospital Centers "Sisters of Mercy" (number EP-7426/14-9) and "Zagreb" (number 02/21/JG, class: 8.1.-14/54-2). Patients gave their informed consent.

#### *2.2. DNA Extraction*

The genomic DNA extraction from unfixed frozen tumor tissue was performed according to the protocol by Green and Sambrook [29]. Briefly, approximately 0.5 g of tumor tissue was homogenized with 1 mL extraction buffer (10 mM Tris–HCl, pH 8.0; 0.1 M EDTA, pH 8.0; 0.5% sodium dodecyl sulfate) and incubated with proteinase K (100 μg/mL; Sigma-Aldrich, St. Louis, MO, USA) overnight at 37 ◦C. Organic (phenol–chloroform) extraction and ethanol precipitation followed.

The salting-out method was used to extract DNA from peripheral blood leucocytes [30]. Five milliliters of blood was lysed with 15 mL RCLB (red blood cell lysis buffer) (0.16 M NH4Cl; 10 mM KHCO3; 10 mM EDTA; pH 7.6), centrifuged (15 min/5000× *g*), and incubated overnight with 2 mL SE buffer (sodium-EDTA; 75 mM NaCl; 25 mM Na2EDTA; pH 8), 200 μL 10% SDS (sodium dodecyl sulphate) and 15 μL proteinase K (Sigma, Darmstadt, Germany) (20 mg/mL) at 37 ◦C. The salting-out method and isopropanol precipitation followed. The method is based on the principle that proteins and other cellular components, except DNA, will precipitate in a saturated salt solution (5M NaCl) due to their relative hydrophobicity.

The extracted DNA was successfully used for genetic (PCR/RFLP) and epigenetic (MS-PCR) analysis.

#### *2.3. Polymerase Chain Reaction (PCR), Restriction Fragment Length Polymorphism (RFLP), Loss of Heterozygosity (LOH)*

#### 2.3.1. Polymerase Chain Reaction

The PCR mixture (25 μL) for *APC*'s exon 11 amplification consisted of 10 pmol of each primer (5- -GGACTACAGGCCATTGCAGAA-3 and 5- -GGCTACATCTCCAAAAGTCAA-3- ), ~250 ng template DNA, 2.5 μL × 10x PicoMaxx reaction buffer, 2.5 mM of each dNTP, and 0.5 μL (1.25 U) of PicoMaxx high fidelity PCR system polymerase. PCR conditions were initial denaturation, 4 min/95 ◦C; denaturation, 1 min/94 ◦C; annealing, 2 min/58 ◦C; extension, 1.5 min/72 ◦C; for 35 cycles and final extension 7 min/72 ◦C. The PCR products were analyzed on 2% agarose gels.

#### 2.3.2. Restriction Fragment Length Polymorphism/Loss of Heterozygosity

Loss of heterozygosity of the *APC* gene was detected on the basis of restriction fragment length polymorphism (RFLP) of the PCR products. RFLP was performed by using restriction enzyme Rsa I, which recognizes a polymorphic site in exon 11 of the *APC* gene. PCR amplification of exon 11 generated a fragment of 133 bp that Rsa I cleaves into 85 bp and 48 bp fragments if the polymorphic site is present, or leaves uncleaved if the site is absent. LOH/Rsa I was demonstrated only in informative (heterozygous) samples when the tumor DNA showed loss of either the single uncut band (133 bp) or of the two cut bands (85 + 48 bp) compared to autologous blood tissue. PCR aliquots (20 μL) were digested with 6 U Rsa I (New England BioLabs, SAD) overnight at 37 ◦C and were electrophoresed on Spreadex EL 400 Mini gels (Elchrom Scientific, AL-Diagnostic GmbH, Amstetten, Austria) in the ORIGINS electrophoresis system (AL-Diagnostic GmbH, Amstetten, Austria) at 120 V and 55 ◦C.

#### 2.3.3. Methylation-Specific PCR (MSP)

After isolation from tumor tissue, DNA was treated with bisulfite using the MethylEdge Bisulfite Conversion System (Promega, Madison, WI, USA) following the manufacturer's instructions. Bisulfite-treated DNA was afterward used for methylation-specific PCR (MSP). Primer sequences of *DKK1*, *DKK3*, and *GSK3β* promoter region for MSP were synthesized according to [31–33], respectively (Table 1).


**Table 1.** PCR primers used for MSP.

MR-F and MR-R-primer set for methylated reaction; UMR-F and UMR-R-primer set for unmethylated reaction; bp-base pairs.

PCRs for bisulfite-treated DNA were performed using TaKaRa EpiTaq HS (TaKaRa Bio, USA): 1XEpiTaq PCR Buffer (Mg2 <sup>+</sup> free), 2.5 mM MgCl2, 0.3 mM dNTPs, 20 pmol of each primer (Sigma-Aldrich, USA), 50 ng of DNA, and 1.5 units of TaKaRa EpiTaq HS DNA Polymerase in a 25 μL final reaction volume. PCR cycling conditions are shown in Table 2.


**Table 2.** MSP conditions for amplification of promoter region of *DKK1*, *DKK3*, and *GSK3β* genes.

MR: methylated reaction; UMR: unmethylated reaction.

PCR products were separated on 2% agarose gel stained with Syber Safe nucleic acid stain (Invitrogen, Thermo Scientific, Waltham, MA, USA) and visualized on a UV transilluminator. Methylated Human Control (Promega, Madison, WI, USA) was used as a positive control for the methylated reaction, while unmethylated DNA EpiTect Control DNA (Qiagen, Hilden, Germany) served as a positive control for the unmethylated reaction.

#### *2.4. Immunohistochemistry (IHC)*

Immunohistochemical staining was performed on 4 μm thick paraffin embedded tissue sections placed on silanized glass slides (DakoCytomation, Glostrup, Denmark). Tissue sections were deparaffinized in xylene (3×, 5 min), rehydrated in a decreasing

ethanol series, (100%, 96%, and 70% ethanol; 2×, 3 min), and placed in water (30 s). Antigen retrieval was performed by heating the sections in microwave oven 2 times for 10 min at 400 W and 3 times for 5 min at 350 W in 6 M citrate buffer. Afterward, the endogenous peroxidase activity was blocked using 3% hydrogen peroxide for 10 min in dark. Non-specific binding was blocked by incubating samples with protein block serum-free ready-to-use (Agilent Technologies, Santa Clara, CA, USA) for 30 min at 4 ◦C. Next, sections were incubated with primary antibody Anti-GSK3β (phospho Y216) (rabbit polyclonal; ab75745, Abcam, Cambridge, MA, USA; dilution 1:100), Anti-GSK3β (phospho S9) (rabbit polyclonal; ab131097, Abcam, Cambridge, MA, USA; dilution 1:100), Active β-Catenin (rabbit monoclonal, non-phospho (Ser33/37/Thr41), D131A1, Cell Signalling Technology, Danvers, MA, USA; dilution 1:800) and Anti-Human Beta-Catenin (mouse monoclonal; Clone b-Catenin-1, M3539, Dako, Santa Clara, CA, USA; dilution 1:200) overnight at 4 ◦C. Dako REAL Envision detection system Peroxidase/DAB, Rabbit/Mouse, HRP (Agilent Technologies, Santa Clara, CA, USA) was used for visualization according to the manufacturer's instructions, and the sections were counterstained with hematoxylin.

The level of expression of examined proteins in the healthy brain was determined by using the cerebral cortex of the human brain (Amsbio, Oxfordshire, UK). It was found that the level of immunoreactivity of all examined proteins in the healthy brain tissue was generally low, and the signal was detected only in the cytoplasm. Human placenta (decidual cells) and colon cancer served as positive controls. Negative controls underwent the same staining procedure but without incubating samples with primary antibodies.

#### *2.5. Microscopic Analysis*

In the tumor hot-spot area, 200 cells were counted and the intensity of protein expression was determined using the computer program ImageJ (National Institutes of Health, Bethesda, MD, USA). Astrocytic brain tumors stained for pGSK3β-Y216, pGSK3β-S9, βcatenin (non-phospho Ser33/37/Thr41), and total β-catenin protein were interpreted by 5 independent observers, of which 2 were pathologists using the following criteria: score 0 (no staining), score 1 (<10% tumor cells), score 2 (10–50% of tumor cells), and score 3 (>50% of tumor cells).

Next, a histological score (H-score) was calculated as the sum of the percentages of positively-stained tumor cells multiplied by the weighted intensity of staining:

$$\text{H-score} = \left[1 \times \text{(\% of cells 1+)} + 2 \times \text{(\% of cells 2+)} + 3 \times \text{(\% of cells 3+)}\right].$$

The H-score, therefore, ranged from 0–300, where '% of cells' represents the percentage of stained cells for each intensity (1 = lack or weak expression, 2 = moderate expression, and 3 = strong expression).

Immunohistochemical results were interpreted blindly in regard to the genetic and epigenetic status.

#### *2.6. Statistical Analysis*

Statistical analysis was performed using SPSS v.19.0.1 (SPSS, Chicago, IL, USA) statistical program. The significance level was set at *p* < 0.05.

The distribution of the data was assessed by the Kolmogorov-Smirnov test and Shapiro-Wilk W-test. Depending on the results of the test of normality and the number of patients per group, differences in the values between the 3 grades were analyzed by one-way variance analysis (ANOVA) or Kruskal-Wallis test, while differences between the 2 groups were tested by Student's t-test or the Mann-Whitney test. Pearson χ2 and Spearman's correlation were used to test the relationships between *DKK1*, *DKK3*, and *GSK3β* methylation, *APC* genetic change, GSK3β and β-catenin protein expression levels, localization, and other clinical and demographic features.

#### **3. Results**

*3.1. Methylation Status of Promoter Regions of GSK3β, DKK,1 and DKK3*

Expression of *DKK1*, *DKK3*, and *GSK3β* genes is controlled, among other mechanisms, by DNA methylation (Figure 2).

**Figure 2.** Methylation-specific PCR analysis for (**A**) *GSK3β*, (**B**) *DKK1*, and (**C**) *DKK3* promoter in astrocytic brain tumors grade II–IV. The presence of a visible PCR product in lanes marked M indicates the presence of methylated promoters, the presence of a product in lanes marked U indicates the presence of unmethylated promoters; (**D**) methylated human control (MC) was used as positive control for methylated reaction, unmethylated human control (UMC) was used as positive control for unmethylated reaction, and water served as negative control. L–standard DNA 50 bp ladder (Invitrogen).

Out of 50 analyzed astrocytoma samples of different grades, 41 (82%) had an unmethylated *GSK3β* gene promoter, while methylation of the promoter region was detected in nine samples (18%), including three AII (30%), three AIII (27%), and three GBM (10.34%) (Figure 3A). Furthermore, methylation of *DKK1* promoter was detected in 19 of 50 tumors (38%), including three AII (30%), five AIII (45.45%), and eleven GBM (37.93%), respectively. Similarly, *DKK3* promoter was methylated in 21 of 49 tumors (42.86%), including four AII (44.44%), four AIII (36.36%), and thirteen GBM (44.83%), respectively (Figure 3B,C). Although it was obvious that grade IV comprised the lowest number of methylated cases for *GSK3β* and highest for *DKK3*, the Kruskal-Wallis test showed no significant association of methylation patterns of *GSK3β* (*p* = 0.235)*, DKK1* (*p* = 0.771), and *DKK3* (*p* = 0.723) promoter regions with the tumor malignancy grade.

Spearman test did not reveal a statistically significant association of promoter methylation between *DKK1* and *DKK3* (*p* = 0.429), *DKK1* and *GSK3β* (*p* = 0.775), or *DKK3* and *GSK3β* (*p* = 0.121) in our sample (Figures S1–S4).

**Figure 3.** Graph showing the percentage of samples with methylated (M) and unmethylated (UM) promoter of (**A**) *GSK3β*, (**B**) *DKK1*, and (**C**) *DKK3* in astrocytic brain tumors grade II–IV.

#### *3.2. pGSK3β-S9 and pGSK3β-Y216 Expression Levels*

The effect of epigenetic changes on the protein expression levels was investigated in the same group of patients. When analyzing active pGSK3β-Y216 in the total sample, low expression was observed in 4% (2/50), moderate expression in 26% (13/50), and strong expression in 70% (35/50). Low expression of the inactive form, pGSK3β-S9, was present in 36% (18/50), moderate in 50% (25/50), and high in 14% (7/50) of total astrocytoma samples (Figure 4).

When analyzing both forms of GSK3β expression in specific astrocytoma types, diffuse astrocytoma revealed 50% (5/10) of the samples with a lack or low levels of pGSK3β-S9 expression, while 70% (7/10) of the samples showed high levels of pGSK3β-Y216. In anaplastic astrocytoma, moderate expression was observed in 60% (6/10) of analyzed cases for pGSK3β-S9, while 70% (7/10) of samples showed a high level of pGSK3β-Y216 expression. The majority of glioblastoma samples analyzed for active pGSK3β-Y216 showed high levels of expression (70%), while for the inactive pGSK3β-S9, moderate expression was observed in 50% of samples. The signal was co-localized in the cytoplasm and cell nucleus in all of the analyzed samples (Figure 5).

**Figure 5.** Characteristic immunohistochemical staining of active pGSK3β-Y216 and inactive pGSK3β-S9 protein in astrocytoma. (**A**) astrocytic brain tumor grade II with unmethylated GSK3β promoter showing weak cytoplasmic staining of pGSK3β-S9; (**B**) same astrocytic brain tumor grade II with unmethylated GSK3β promoter showing strong cytoplasmic and nuclear staining of pGSK3β-Y216; (**C**) glioblastoma (grade IV) with unmethylated GSK3β promoter showing lack of cytoplasmic staining of pGSK3β-S9; (**D**) same glioblastoma with unmethylated GSK3β promoter showing strong cytoplasmic and nuclear staining of pGSK3β-Y216; (**E**) glioblastoma (grade IV) with methylated GSK3β promoter showing moderate cytoplasmic and strong nuclear staining of pGSK3β-S9; (**F**) glioblastoma (grade IV) with methylated GSK3β promoter showing weak cytoplasmic staining of pGSK3β-Y216. Scale bar 50 μm.

Protein expressions of both active and inactive forms of GSK3β (pGSK3β-S9 (*p* = 0.728) and pGSK3β-Y216 (*p* = 0.820)) showed no significant association with any specific astrocytoma grade. Wilcoxon test revealed a statistically significant difference between the expression of pGSK3β-S9 and pGSK3β-Y216 protein in investigated astrocytoma (*p* < 0.0001, Z = −5.332). This result indicates that samples with a higher level of expression of active pGSK3β-Y216 have a lower expression level of inactive pGSK3β-S9 protein.

#### *3.3. Total β-Catenin and Unphosphorylated β-Catenin Expression Levels in Glioblastoma*

Expression and localization of total β-catenin (detects both phosphorylated and unphosphorylated form) and active (unphosphorylated) β-catenin were further examined in glioblastoma.

H-score analysis for total β-catenin revealed 36.66% (11/30) samples with weak, 56.67% with moderate (17/30) and 6.67% (2/30) with strong protein expression when compared to levels of β-catenin in a healthy brain (Figure 4). Grouped together, elevated expressions (2+ and 3+) were detected in 63.34% of samples. Most samples showed cytoplasmic expression (86.67%), while co-localization of the signal in cytoplasm and nucleus was present in only 4 cases (13.33%).

Unphosphorylated β-catenin showed a similar distribution of signal strengths. Hscore analysis detected 33.33% (10/30) of samples with weak, 60% (18/30) with moderate, and 6.67% (2/30) with strong expression (Figure 4). When compared to cellular levels of β-catenin in a healthy brain, grouped elevated expressions (2+ and 3+) were observed in 66.67% of samples. Cytoplasmic and nuclear co-localization of unphosphorylated βcatenin was noticed in five samples (16.67%), and again in most samples, the expression was present exclusively in the cytoplasm (83.33%).

Spearman's correlation showed a statistically significant positive correlation between H-score values of total and unphosphorylated β-catenin (rs = 0.634, *p* < 0.001), which confirms the presence of an active form of β-catenin in tumor cells.

#### *3.4. APC Exon 11 Genetic Changes in Glioblastoma*

Genetic changes of *APC* exon 11 were analyzed in 27 glioblastoma samples that were available for the analysis, nine (33%) of which were homozygous i.e., uninformative. Taking into account informative (heterozygous) samples, 44.44% (8/18) showed one of the two observed genetic changes. More precisely, LOH was detected in 33% (6/18) and the introduction of the restriction site because of mutation in 11.11% (2/18) cases (Figure 6).

**Figure 6.** APC exon 11/RsaI/RFLP in glioblastoma samples is demonstrated. Lane M-standard DNA 50 bp ladder (Invitrogen); lanes 1, 2: heterozygous sample (tumor and blood), both alleles, cut and uncut, are visible; lane 3: possible restriction site introduced in tumor sample; lane 4: paired homozygous sample (blood); lanes 5, 6: homozygous sample (tumor and blood), uncut alleles are visible; lanes 7, 8: homozygous sample (tumor and blood), cut alleles are visible; lane 9: LOH, cut allele is missing; lane 10: corresponding informative blood sample, both alleles, cut and uncut, are visible.

Eight glioblastoma samples with genetic change in *APC* exon 11 showed moderate expression of total β-catenin in 62.5% and unphosphorylated β-catenin in 50% of samples. Student t-test revealed a significant difference of total β-catenin expression between groups with and without genetic changes in *APC's* exon 11. Samples with changes had higher H-score values for total β-catenin, compared to the group without changes (t = −2.264, *p* = 0.038). However, a significant association of unphosphorylated (active) β-catenin H-score values with groups with or without *APC* genetic changes (U = 54.500, *p* = 0.197) could not be established.

#### *3.5. The Correlations of Molecular Findings and Clinical Parameters*

Spearman test revealed a significant positive correlation between *DKK3* methylation status and expression of active pGSK3β-Y216 (rs = 0.356, *p* = 0.011), indicating that when the *DKK3* was epigenetically silenced, the expression of the active GSK3β was on the rise. Inactive pGSK3β-S9 protein was significantly positively correlated with the methylation of the GSK3β promoter region (rs = 0.278, *p* = 0.050). Additionally, a bivariate correlation between active pGSK3β-Y216 and active (unphosphorylated) β-catenin showed a trend of negative association of the two proteins (rs = −0.427, *p* = 0.088). Our analysis also showed that samples with genetic change in *APC* exon 11 were statistically significantly correlated with the increase of total β-catenin expression (rs = 0.542, *p* = 0.004). The upregulation of β-catenin expression was noticed in 66% of analyzed samples. Of note is that 55% of

the samples with upregulated β-catenin showed methylation in negative regulators of the signaling *DKK1* or *DKK3*.

Finally, no statistical significance was found between molecular findings and clinical parameters (*p* > 0.05), meaning that the analyzed molecular features were independent of the patients' age and sex.

#### **4. Discussion**

The Wnt signaling pathway is frequently implicated in the etiology of various cancers and plays important roles in tumor initiation and progression. As recent reports indicate, impairment of negative regulators of Wnt signalization, i.e., DKKs, is often involved in tumor formation and growth [34]. Expression of the *DKK1* and *DKK3* gene is controlled, among other mechanisms, by DNA methylation, a common epigenetic silencing tool, which is increased in many tumors and tumor cell lines. Most research has indicated a DKK1 inhibitory effect in tumors [35–38], but interestingly, some studies showed DKK1- s tumorpromoting role [18,39]. Similarly, DKK3 was discovered to be downregulated in various types of malignant tissue [36,40–44], but there are also reports of DKK3 overexpression in hepatocellular and esophageal cancer [45–47].

Our findings on 50 astrocytoma samples revealed promoter methylation of *DKK1* in 38% and *DKK3* in 42.86% of the samples. Methylation of *DKK1* and *DKK3* was relatively constant across different grades (*DKK1*-AII 30%, AIII 45.45%, GBM 37.93%; *DKK3*-AII 44.44%, AIII 36.36%, GBM 44.83%). The remaining portion of astrocytoma samples did not show *DKK1* and *DKK3* promoter methylation; downregulation could be explained by the existence of additional epigenetic regulatory events. In the case of DKK1 and DKK3, these may be post-translational modifications in the histone tails, which are associated with transcriptional repression [35,48]. We also point out other mechanisms here that may contribute to *DKK1* and *DKK3* gene (and consequently DKK1 and DKK3 protein) silencing, such as the mutational burden and various miRNAs, as recently shown in a case of melanoma [49] and colorectal cancer [38,50,51].

A study by Götze et al. [27] suggests that primary and secondary GBMs are characterized by different *DKK1* and *DKK3* gene methylation profiles, helpful to distinguish between glioblastoma subtypes. In their study, promoter methylation of *DKK1* was quite rare in lower-grade astrocytoma but frequent in glioblastoma. Additionally, *DKK1* methylation was more frequently observed in secondary GBM (5/10), whereas none of the primary GBMs showed methylation of the *DKK1* promoter region. These findings suggest that methylation of *DKK1* may be linked to glioma progression and thus might be a potential prognostic marker. Furthermore, this study indicates that methylation of *DKK3* is a rare event in glioma, with no obvious association with the tumor type or grade [27]. Our research showed no significant association between methylation of *DKK1* (*p* = 0.767) or *DKK3* (*p* = 0.885) and the tumor grade or type. However, the frequency of methylation of the two genes was overall substantial, and it was shown that methylation of *DKK1* was higher in pooled grades III and IV (40%) in comparison to AII (30%). No significant association of promoter methylation between *DKK1* and *DKK3* (*p* = 0.429) was found in our study. Mizobuchi et al. [25] showed that DKK3 plays a pivotal role in regulating cell survival in human malignant glioma, promotes apoptosis, and facilitates the degradation of β-catenin. Similarly, there are also reports about a DKK1 pro-apoptotic function in glioma [36].

Apart from epigenetic silencing of negative regulators, aberrant pathway activity may be a result of the mutations in downstream components. Modifications that cause change in GSK3β activity, *CTNNB1* gene mutations targeting sites phosphorylated by GSK3β on β-catenin (S33, S37, and S41), and mutations of proteins that form a destruction complex with GSK3β (for example, APC) may all cause Wnt pathway hyperactivity [9].

The influence of GSK3β on tumor formation and promotion is still controversial. GSK3β can display both pro-oncogenic and tumor-suppressive effects as it has diverse roles in numerous cellular processes that also differ among different cell types [5].

In order to determine the character of the GSK3β role in astrocytoma grades II–IV, we examined the promoter methylation of the *GSK3β* gene and level of expression of active (pY216) and inactive (pS9) form of GSK3β protein. Although the proportion of methylated samples was relatively small in each astrocytoma grade, our results showed that methylation of *GSK3β* decreased with grade. In astrocytoma grade II, 30% of samples had methylated *GSK3β* promoter, followed by 27% of astrocytoma grade III and 10% of glioblastoma. Of note is that the number of unmethylated samples increased with grade, meaning that GSK3β is upregulated in aggressive cases. GBM is primarily diagnosed at older ages, and recent reports suggest that epigenetics, especially DNA methylation, seem to be age-dependent [52].

Immunohistochemical analysis revealed low expression levels of active pGSK3β-Y216 in 4% (2/50), moderate levels in 26% (13/50), and strong levels in 70% (35/50) of samples, which is consistent with the findings on unmethylated promoter. Expression of inactive pGSK3β-S9 was weak in 36% (18/50), moderate in 50% (25/50), and strong in 14% (7/50) of astrocytoma samples in grades II–IV. Wilcoxon test showed significant opposite levels of expression between pGSK3β-S9 and pGSK3β-Y216 protein in astrocytoma cells (*p* < 0.001, Z = −5.332), indicating that samples with a higher level of expression of active pGSK3β-Y216 have a lower expression level of inactive pGSK3β-S9 protein. Inactive pGSK3β-S9 protein was significantly positively correlated with methylation of *GSK3β* promoter (rs = 0.278, *p* = 0.050), showing that in methylated cases, phosphorylation events also decrease protein expression. Similarly, Shakoori et al. [53,54] studied the expression of active and inactive forms of GSK3β in colorectal cancer. Although the study included a smaller number of samples, they proved that most patients had elevated expression of pGSK3β-Y216, whereas pGSK3β-S9 was mainly present in non-neoplastic tissues. Contrary to previously mentioned studies, high expression of inactive pGSK3β-S9 is found in skin [55], oral [56], and lung [57] cancers, which suggests tumor-suppressing effects of the enzyme in those malignant tissues.

The pro-oncogenic activity of GSK3β is based on the findings that deregulated GSK3β maintains tumor cell survival, proliferation, and invasion by enhancing machinery for cell motility and migration [58]. Finally, growing evidence marks GSK3β as a potential therapeutic target in cancer [59,60], thus encouraging the development of GSK3β inhibitors for cancer treatment [61]. In glioblastoma multiforme, such inhibitors facilitate apoptosis through inhibition of anti-apoptotic mechanisms in mitochondria and the NFkB pathway that is essential for cell survival [58,62].

Although GSK3β is generally considered a cytosolic protein, it can also be present in the nucleus. Our data showed an elevated level of expression of pGSK3β-Y216 in the cell nuclei of almost every sample of astrocytic brain tumors. It is known that GSK3β nuclear levels increase in response to apoptotic stimuli, and its major role is to affect gene expression by regulating the activity of many transcription factors [63].

The largest number of glioblastomas analyzed for both total and unphosphorylated β-catenin had moderate cytoplasmic expression (56.67% and 60%, respectively); weak expression was noted in 36.66% and 33.33%, respectively; while a strong signal was present in a smaller percentage of samples (6.67% and 6.67% respectively). Overall cytoplasmic accumulation of β-catenin predominated, whereas the expression of total and unphosphorylated β-catenin in the nucleus was observed in only four (13.33%) and five (16.67%) samples, respectively. It seems that strong expression and consequent transfer in the nucleus occurs in a smaller number of glioblastomas. Utsuki et al. [64] and Kahlert et al. [65] also found a small number of samples with nuclear expression, which can be partly explained by the Wnt pathway activity only in a small proportion of glioblastoma cells that have stem cell properties [66]. Another explanation is that the active form of beta-catenin that is transferred to the nucleus lacks specific epitopes and cannot be detected by this antibody. Spearman's correlation showed a statistically significant positive correlation between H-score values of total and unphosphorylated β-catenin (rs = 0.634, *p* < 0.001), thus confirming the presence of an active form of β-catenin in tumor cells. Phosphory-

lation status and localization of β-catenin are important indicators of the Wnt signaling pathway's activation. Liu et al. [67] studied β-catenin expression in different grades of astrocytoma and noticed significantly higher levels of β-catenin in glioblastoma compared to lower grades and control groups, thus suggesting a role for β-catenin in the progression of malignant gliomas. The study by Sareddy et al. [68] on astrocytoma grade II–IV reports on the positive correlation of β-catenin mRNA and protein levels with the increase of malignancy grades. They also noticed a nuclear and cytoplasmic accumulation of β-catenin in astrocytoma, which is the hallmark of active Wnt/β-catenin signaling. Previous research by our laboratory has shown that tumors of neuroepithelial origin have higher levels of β-catenin expression compared to β-catenin expression levels in healthy brain tissue [69]. Kafka et al. [70] revealed that DVL3, TCF1, and LEF1 expression significantly increased with astrocytoma malignancy grades, suggesting their cooperation with nuclear β-catenin and joint involvement in malignant progression. In the present investigation, the bivariate correlation between unphosphorylated active β-catenin and active pGSK3β-Y216 showed a trend of negative association of the two proteins (rs = −0.427, *p* = 0.088), confirming their mutually dependent relationship. Still, it seems that GSK3β activity toward β-catenin does not depend unambiguously on its phosphorylation status on S9, which appears to be a protective mechanism when GSK3β is aberrantly phosphorylated by some kinases [71,72]. It has been shown that highly active Akt does not fully inhibit GSK3β activity in some cancers and cancer cell lines [53]. However, a more recent study on human colorectal cancer cell lines shows that hyperactive Akt causes GSK3β inhibition and consequential β-catenin accumulation [73]. Recent studies demonstrated that active DKK3 is associated with reduced cytoplasmic and nuclear accumulation of β-catenin in different tumor types [74].

*APC* is a tumor-suppressor gene and an essential component of the beta-catenin complex that controls cytoplasmic beta-catenin levels. *APC* mutations occur early in gliomagenesis and result in increased beta-catenin levels that lead to the expression of Wnt responsive genes [9]. In our study, genetic change in *APC* exon 11 was present in 44.44% of the informative samples. Our laboratory group previously reported on *APC* exon 11 genetic changes in human brain tumors [75], brain metastases [76], and laryngeal squamous cell carcinoma [77] and found 33.3%, 58.8%, and 41% of samples with LOH or mutation of this gene, respectively. The present investigation also found that samples with changes of the *APC* gene had significantly higher values of total β-catenin H-score, compared to the group without genetic changes (t = −2.264, *p* = 0.038). Although the result for unphosphorylated β-catenin was not significant, its elevated expression in glioblastoma indicates the pathway's activity and its association with genetic changes in *APC*. The relatively rare expression of β-catenin in the nucleus may also be explained by work from Morgan et al. [78], where they showed that APC loss alone was insufficient to stimulate nuclear β-catenin translocation, and further dysregulation is required. Another explanation for the rare β-catenin nuclear expression is the finding that most of the C-terminal deletions show the predominant nuclear localization [79], and the antibody that we used in our study was raised against the C-terminal β-catenin epitope.

In conclusion, the results of this study undoubtedly indicate the activation of the Wnt signaling pathway in astrocytoma. Our findings on *DKK1* and *DKK3* demonstrate the importance of methylation in the regulation of Wnt signaling activity but also suggest that additional regulatory mechanisms may be involved. Our findings indicate pro-oncogenic effects of GSK3β on astrocytoma development and progression not necessarily connected to the Wnt destruction complex. It is also evident that large deletions and mutations in the *APC* gene increase the level of β-catenin expression in glioblastomas. This research can provide more data about astrocytoma pathogenesis and help to better understand and improve the management of gliomas.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/cancers13112530/s1, Figures S1–S4: M = 50 bp molecular marker; MK-MR = methylated control methylated reaction; MK-UMR = methylated control unmethylated reaction; UMK-MR = unmethylated control methylated reaction; UMK-UMR = unmethylated control unmethylated reaction; S(1–6)-MR = sample (1–6) methylated reaction; S(1–6)-UMR = sample (1–6) unmethylated reaction; NK-MR = negative control methylated reaction; NK-UMR = negative control unmethylated reaction.

**Author Contributions:** A.K. conceived the idea, designed the study, performed experimental work, contributed to data acquisition and analysis, wrote the manuscript, and revised it for important intellectual content. A.B. contributed to the data interpretation, manuscript editing, and revised the manuscript for important intellectual content. E.B. performed experimental work and participated in data collection, interpretation, and analysis. A.-M.J. performed experimental work and contributed to interpretation of the results. K.P. performed experimental work and participated in tumor sample analysis and results interpretation. P.B. contributed to data acquisition, the interpretation of the results, and manuscript editing. N.N. contributed to data acquisition and participated in tumor sample collection. K.Ž. participated in tumor sample analysis and revised the manuscript for important intellectual content. N.P.-Š. conceived the idea, designed the study, contributed to analysis and interpretation of the results, wrote the manuscript and revised it for important intellectual content, and approved the final version of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Scientific Centre of Excellence for Basic, Clinical and Translational Neuroscience (project "Experimental and clinical research of hypoxic-ischemic damage in perinatal and adult brain"; GA KK01.1.1.01.0007 funded by the European Union through the European Regional Development Fund).

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of School of Medicine University of Zagreb (Case number: 380-59-10106-17-100/98; Class: 641-01/17-02/01, 23 March 2017), Ethics Committee of University Hospital Center Zagreb (number 02/21/AG, class: 8.1-16/215-2, 02 February 2017), and Ethics Committee of University Hospital Center "Sisters of Charity" (number EP-5429/17- 5, 23 March 2017).

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

**Data Availability Statement:** Data supporting reported results are contained within the article. Some of the data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy issues.

**Conflicts of Interest:** All authors declare that they have no conflict of interest.

#### **References**


#### *Article*
