Antibodies against disheveled segment polarity protein (DVL)2 (#3224), DVL3 (#3218), protein kinase B (AKT; #9272), β-catenin (#8480), Axin2 (#5863), Cyclin D1 (#2978), c-Myc (#5605), c-Jun (#9165), low-density lipoprotein receptor-related protein (LRP)6 (#2560), survivin (#2803), E-cadherin (#3195), cleaved caspase 3 (#9661), cleaved poly (ADP-ribose) polymerase (PARP; #5625), and phosphorylation-specific antibodies against AKT Ser473 (#9271), glycogen synthase kinase (GSK)-3αβ (Ser21/9; #9331), RanBP3 Ser58 (#9380), non-p (Active) β-catenin Ser33/Ser37/Thr41 (#8814), phosphorylated (p)-histone 3 Ser10 (#9701), LRP6 Ser1490 (#2568), α-tubulin (#2144) and extracellular signal-regulated kinase (ERK)1/2 Thr202/Tyr204 (#4370) were obtained from Cell Signaling Technology, Inc. (Danvers, MA, USA). The antibodies against ERK1/2 (sc-94), glutamine synthetase (SC-74430) were obtained from Santa Cruz Biotechnology, Inc. (Dallas, TX, USA). Anti-mouse cluster of differentiation (CD)31 antibody (#102502) was from BioLegend, Inc. (San Diego, CA, USA). p-β-catenin Tyr142 (#CP1081) was purchased from ECM Biosciences LLC (Versailles, KY, USA).

L-cells (ATCC CRL-2648) and L-Wnt-3A cells (ATCC CRL-2647) cells were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). Control-conditioned medium from L-cells and Wnt-3A-conditioned medium from L-Wnt-3A cells was prepared according to the protocols provided by the ATCC. The conditioned medium was concentrated 2X and sterile-filtered.

Between January 2004 and September 2018, 280 primary HCCs were obtained during surgery at Department of General Surgery, Singapore General Hospital (Singapore) and used for establishment of HCC PDX models. Written informed consent was obtained from all patients prior to tissue collection and the present study received Ethics Committee approval from National Cancer Centre Singapore (NCCS; Singapore) as well as Singapore General Hospital. HCC PDX models were established as described previously (18). Primary HCC13-0109, HCC26-0808A, and HCC06-0606 cells were freshly isolated from PDX tumors as follows: HCC tumors were finely minced and washed 3 times with modified Eagle's medium (MEM; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany). The minced tissue was incubated with MEM containing 5% fetal bovine serum (FBS; GE Healthcare Life Sciences, Logan, UT, USA) and 5 mg/ml collagenase (Roche Diagnostics, Indianapolis, IN, USA) at 37°C for 12 h. Cells were harvested by centrifuging at 4°C at 800 × g for 10 min. The cell pellets were washed 3 times with serum-free MEM, and primary HCC cells were plated at a density of 5.0×10⁶ cells per well in MEM containing 10% FBS and 1% penicillin-streptomycin (growth medium), and incubated at 37°C and 5% CO₂ for 48 h. The primary cells were then used for subsequent experiments.

Primary HCC06-0606 or HCC07-0409 cells were treated with the vehicle, 0.5 µM refametinib, 2.5 µM sorafenib or 2.5 µM sorafenib plus 0.5 µM refametinib in growth medium at 37°C for 24 h then stimulated with concentrated conditioned medium prepared either from the control L-cells or L-Wnt-3A cells at 37°C for 2 h. The cells were harvested, and changes in the levels of proteins of interest were determined by western blotting.

Phosphorothioate oligonucleotides (ODN) directed against β-catenin antisense (5′-TAAGAGCTTAACCACAACTG-3′) and scrambled control (5′-CAGTAATCGAATAGCTACCA-3′) were purchased from TriLink Biotechnologies (San Diego, CA, USA). A further 5×10⁶ HCC06-0606 cells were transfected with 150 nM scrambled control or 150 nM ODN directed against β-catenin using Plus™ Reagent (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) at 37°C for 48 h. The cells were subsequently assayed for β-catenin protein and its downstream targets by western blotting.

The present study was approved by the Ethics Board at NCCS (Singapore). Mice were maintained at NCCS according to the 'Guide for the Care and Use of Laboratory Animals' (19). All animal experiments were performed at NCCS.

Sorafenib (Nexavar™; Bayer AG) and selumetinib (14) (Selleck Chemicals, Houston, TX, USA) were suspended in vehicle [30% captisol in water: 5% glucose (50% v:v)] at 1.25 and 3.125 mg/ml, respectively. Refametinib (Bayer AG) (15) was diluted in dimethyl sulfoxide at a concentration of 100 mg/ml as stock. Stock refametinib (0.150 ml) was further diluted in 3.925 ml 30% captisol in water followed by 3.925 ml 5% glucose to obtain an appropriate concentration.

Primary HCCs have previously been used to establish HCC PDX xenograft lines (18), of which the following 9 HCC lines (HCC01-0207; HCC25-0705A; HCC10-0505; HCC09-0913; HCC01-0708; HCC29-0909A; HCC06-0606; HCC26-0808A, and HCC07-0409), as well as one sorafenib-resistant HCC26-0808ASora54, were used to establish tumors in 450 male C.B-17 SCID mice aged 9-10 weeks and weighed 23-25 g (InVivos Pte. Ltd., Singapore) as described previously (18,20). Mice were provided with sterilized food and water ad libitum, and housed in negative pressure isolators, which were set at 23°C and 43% humidity, with 12-h light/dark cycles.

C.B-17 SCID mice bearing HCC26-0808A tumors (18) were treated once daily with 10 mg/kg sorafenib for 80 to 90 days. Following the initial response to sorafenib, HCC26-0808A mice gradually acquired resistance, leading to further tumor growth. Sorafenib-resistant tumors were harvested for serial transplantation when they reached the size of 1,500 mm³. The cycle was repeated until sorafenib had minimal impact on the growth of treated tumors.

Mice bearing the indicated tumors (8-10 mice per group) were orally administered either 200 µl vehicle, 10 mg/kg sorafenib, 15 mg/kg refametinib (or 25 mg/kg selumetinib) or 10 mg/kg sorafenib plus refametinib daily for 12-14 days. Treatment commenced when the tumors reached 150-175 mm³ in size. Bi-dimensional measurements were performed once every 2-3 days and tumor volumes were calculated based on the following formula: Tumor volume = [(length) x (width²) × (π/6)], and plotted as the means ± standard error of the mean for each treatment group vs. time, as previously described (18). Body and tumor weights were recorded at the time of sacrifice. The tumors were harvested 2 h following the final treatment and stored at -80°C for later biochemical analysis. The efficacy of each treatment was determined using the T/C ratio, where T and C are the median weight of drug-treated and vehicle-treated tumors, respectively, at the end of treatment.

Tumors from vehicle- and drug-treated mice were homogenized in buffer containing 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.5% NP-40, 1 mM EDTA, 25 mM NaF, supplemented with proteinase inhibitors and 10 mM Na₃VO₄. Protein concentration was determined by Bradford assay. In total, 80 µg protein/lane was resolved by either 8% or 14% SDS-PAGE, and western blotting was performed as previously described (21). All primary antibodies were diluted in 1% low-fat skimmed milk powder dissolved at a ratio of 1:1,000 to a final concentration of 1 µg/ml. Blots were incubated with primary antibodies for 14-16 h at 4°C. After 3 washes in TBST, blots were incubated with a secondary antibody [horseradish peroxidase (HRP)-conjugated goat anti-rabbit immunoglobulin (Ig)G (H+L; #31460; 1:5,000; Thermo Fisher Scientific, Inc.] for 60 min at room temperature. All primary antibodies were then visualized using WesternBight™ ECL chemiluminescent detection reagents (Advansta, Inc., Menlo Park, CA, USA). For quantification analysis, total density of the band corresponding to protein blotting with the indicated antibody was quantified using the GS-900 Calibrated Densitometer and Image Lab™ Software 6.0.1 (Bio-Rad Laboratories, Inc., Hercules, CA, USA), normalized to both the tubulin loading control and the appropriate phosphorylated/total protein where applicable, and expressed as the fold-change.

Tumor tissues were fixed in PBS buffer containing 4% formaldehyde (ICM Pharma, Pte. Ltd, Singapore) for 24 h at room temperature and embedded in paraffin. Tissue sections (5-µm) were blocked with 5% low-fat skimmed milk powder dissolved in TBS containing 0.1% Tween (TBST) for 1 h at room temperature, followed by 3 washes in TBS. Sections were then incubated with CD31 (1:100), p-histone 3 Ser10 (1:300), cleaved PARP (1:100), and total β-catenin (1:150) antibodies for 14-16 h at 4°C to assess micro-vessel density, cell proliferation, and apoptosis, cytoplasmic and nuclear β-catenin, respectively, as described (20). Following 3 washes in TBST (5 min each), sections were incubated with biotin-conjugated goat anti-rabbit IgG secondary antibody (#31820; 1:150; Thermo Fisher Scientific, Inc.) for 45 min at room temperature, followed by 3 washes in TBS and then incubated with streptavidin HRP-conjugated antibody (#21126; 1:500; Thermo Fisher Scientific, Inc.). Slides were counterstained with hematoxylin (Sigma Diagnostics, Inc., Livonia, MI, USA) for 10 sec at room temperature. Images were captured on an Olympus BX60 light microscope (Olympus Corporation, Tokyo, Japan). Five random 0.159-mm² fields at ×100 magnification were captured for each tumor. The number of p-histone 3 Ser10 and cleaved PARP-positive cells among at least 500 cells per field was counted and expressed as the number of positive cells per 1,000 cells. For the quantification of mean microvessel density in sections stained for CD31, 5 random fields at a magnification of ×100 were selected for each tumor.

The CTNNB1 (β-catenin) mutational status of 9 HCC models was determined via WES as described previously (22).

Data were presented as the mean ± standard error of the mean. Differences in the protein level, tumor weight at sacrifice, p-histone 3 Ser10 index, mean micro-vessel density, and the number of cleaved PARP-positive cells were compared. Student's t-test was used for comparisons between two groups. One-way analysis of variance followed by the Tukey-Kramer method post-hoc test was used when comparing more than two groups. P<0.05 was considered to indicate a statistically significant difference.

WES analysis revealed that HCC29-0909A harbored a T41A β-catenin mutation, Y126 p53 mutation and EZH2 intergenic deletion; HCC10-0505 harbored a S45P β-catenin mutation, p53 deletion, and 2108G>T JAK1 mutation. HCC01-0708 harbored a T41A β-catenin mutation, Y126 p53 mutation, and N-RAS amplification; HCC07-0409 harbored a T41A β-catenin mutation, Y126 p53 mutation, and CCND1 and N-RAS amplification; HCC26-0808A had a NF1 deletion, H168R p53 mutation, 3G>T VHL mutation, and loss of CDKN2A and CDKN2B; HCC06-0606 harbored a 407A>C p53 mutation and a MCL1 amplification; HCC01-0207 had a 341G>A RET mutation, R249S p53 mutation and LRP1B truncation; HCC25-0705A had a 341G>A RET mutation and R249S p53 mutation; HCC09-0913 harbored a R249S p53 mutation. HCC01-0708, HCC07-0409 and HCC29-0909A harbored a T41A β-catenin mutation, and HCC10-0505 had S45P β-catenin mutations. These observations are consistent with those of previous studies, indicating that mutations in the β-catenin gene account for 20-30% of all HCC cases (10,11).

As treatment with sorafenib led to the upregulation of p-ERK1/2 (21) and ERK kinase activated Wnt/β-catenin signaling (23), the combined effects of sorafenib/MEK inhibitor on tumor growth and the Wnt/β-catenin signaling pathway were evaluated. Consistent with our previous study (21), oral delivery of sorafenib led to significant inhibition of tumor growth as compared with the vehicle group (P<0.01). Addition of refametinib to sorafenib resulted in significantly improved anti-tumor activity in all HCC models tested when compared with either agent alone (P<0.01; Fig. 1). As compared with the vehicle group, refametinib significantly inhibited tumor growth of HCC29-0909A, HCC26-0808A, HCC06-0606 and HCC01-0207 (P<0.01; Fig. 1) but not HCC09-0913 and HCC01-0708. Tumor regression was observed in wild-type β-catenin HCC26-0808A, HCC06-0606, HCC01-0207 and HCC09-0913 models following sorafenib/refametinib treatment. The antitumor activity of sorafenib/refametinib was independent of β-catenin mutational status as both mutant and wild-type β-catenin xenografts were inhibited by this combination (Fig. 1). Refametinib monotherapy was not considered to be active in HCC01-0708 (mutant β-catenin) and HCC09-0913 (wild-type β-catenin) models (Fig. 1), suggesting that factors other than β-catenin were responsible for the responsiveness. As predicted, sorafenib/refametinib inhibited cell proliferation and caused increased apoptosis, as compared with sorafenib or refametinib alone (P<0.05; Fig. 2). Similar results were obtained when selumetinib was combined with sorafenib (data not shown).

The effects of sorafenib/refametinib combination on β-catenin and its downstream targets in mutant β-catenin HCC29-0909A and wild-type β-catenin HCC06-0606 xenografts were then investigated. As presented in Fig. 3, sorafenib alone had minimal effect on p-LRP6 Ser1490 and p-RanBP3 Ser58. Quantification of the immunoblots revealed that a decrease in these proteins was observed in refametinib-treated HCC06-0606 (Fig. 3A) and HCC29-0909A xenografts (Fig. 3B). Notably, sorafenib/refametinib resulted in a significant decrease in p-LRP6 Ser1490 (P<0.05; Fig. 3). As presented in Fig. 3, there was no significant difference in p-GSK-3αβ (Ser21/9) levels between the treatment groups. Although total β-catenin was modestly reduced, p-β-catenin Tyr142, p-RanBP-3 Ser58 and active β-catenin in sorafenib/refametinib-treated tumors were significantly reduced (P<0.05; Fig. 3). Significant reduction of β-catenin downstream targets (Axin-2, survivin, c-Myc, cyclin D1 and glutamine synthetase) in sorafenib/refametinib-treated tumors was observed (P<0.05; Fig. 3). Similar results were observed when wild-type HCC09-0913 and mutant β-catenin HCC01-0708 were treated with sorafenib/selumetinib (data not shown).

The tyrosine kinases have been demonstrated to induce phosphorylation of β-catenin at tyrosine 142 (24,25). As p-β-catenin Tyr142 (24) and p-RanBP3 (26) have been reported to be involved in β-catenin export, whether sorafenib/refametinib treatment caused nuclear export was examined. Tumor sections among different treatment groups were stained with total β-catenin antibody. Intense nuclear and cytoplasmic β-catenin staining was observed in vehicle-treated mutant β-catenin tumors (HCC29-0909A and HCC01-0708) (Fig. 4). The majority of β-catenin in sorafenib-treated tumors was located in the cytoplasm and nucleus (Fig. 4). Refametinib markedly reduced nuclear and cytoplasmic β-catenin concomitant with a marked increase in membranous β-catenin (Fig. 4). There was no clear difference in β-catenin relocalization between sorafenib- and refametinib-treated tumors. Notably, nuclear β-catenin staining was barely detectable in HCC29-0909A and HCC01-0708 tumors treated with sorafenib/refametinib (Fig. 4). The nuclear staining was less pronounced in the HCC10-0505 model with the S45P mutant of β-catenin, suggesting that there may be functional differences amongst the various mutations (data not shown). In HCC29-0909A and HCC01-0708 models, a membranous signal accumulated in areas of intercellular contact, likely representing β-catenin associated with membranous cadherins (24,27). These observations suggest that β-catenin export from the nucleus and/or the cytoplasm to the membrane following sorafenib/refametinib treatment likely occurs via the inhibition of p-β-catenin Tyr142 (25) and p-RanBP3 (26). β-catenin relocalization also occurred when HCC09-0913 tumors were treated with sorafenib/selumetinib (data not shown).

As presented in Fig. 5A, treatment with antisense, but not scrambled control, β-catenin resulted in decreased cell viability, β-catenin levels and those of its known downstream targets. Treatment of wild-type β-catenin HCC06-0606 cells with Wnt-3A-conditioned medium led to the significant upregulation of p-LRP6 Ser1490, p-ERK1/2 Thr202/Tyr204, p-RanBP3 Ser58 and p-β-catenin Tyr142 (P<0.01; Fig. 5B) as compared with control-conditioned medium. Marked increases in survivin, cyclin D1 and glutamine synthetase were also noted. Pre-treatment of HCC06-0606 cells with refametinib or sorafenib/refametinib, but not with sorafenib, for 24 h abolished Wnt-3A-induced upregulation of p-LRP6 Ser1490, p-β-catenin Tyr142, p-RanBP3 Ser58 and p-ERK1/2 Thr202/Tyr204 (P<0.05; Fig. 5B). These findings suggest that sorafenib/refametinib profoundly suppressed β-catenin activity, resulting in decreased proliferation and elevated levels of apoptosis. Similar observations were obtained when mutant β-catenin HCC07-0409 cells were treated with Wnt-3A-conditioned medium (data not shown).

While growth of the parental HCC26-0808A line was significantly reduced by sorafenib, growth of the HCC26-0808ASora54 line was unaffected (data not shown). This indicates that HCC26-0808ASora54 is resistant to sorafenib treatment. As presented in Fig. 6A, sorafenib-resistant HCC26-0808ASora54 tumors that continued to grow in the presence of sorafenib experienced a significant reduction in tumor growth when ERK activity was blocked by refametinib (P<0.001). The addition of refametinib to sorafenib abolished tumor growth (Fig. 6A) and the final tumor weight of sorafenib/refametinib group was significantly smaller than that of the refametinib or sorafenib groups (P<0.001; Fig. 6B). This was associated with inhibition of the β-catenin signaling pathway, as evidenced by a significant reduction in p-LRP6 Ser1490, p-β-catenin Tyr142, p-RanBP-3 Ser58, and active non-phospho-β-catenin, Axin2, survivin, cyclin D1, c-Myc, and glutamine synthetase levels (P<0.05; Fig. 6C). Nuclear exportation of β-catenin was also observed when HCC26-0808ASora54 was treated with sorafenib/refametinib (Fig. 6D).