Aberrant activation of Wnt/β-catenin signaling plays a key role in the onset and development of hepatocellular carcinomas (HCC), with about half of them acquiring mutations in either CTNNB1 or AXIN1. The serine/threonine kinase receptor-associated protein (STRAP), a scaffold protein, was recently shown to facilitate the aberrant activation of Wnt/β-catenin signaling in colorectal cancers. However, the function of STRAP in HCC remains completely unknown. Here, increased levels of STRAP were observed in human and mouse HCCs. RNA sequencing of STRAP knockout clones generated by gene editing of Huh6 and Huh7 HCC cells revealed a significant reduction in expression of various metabolic and cell-cycle–related transcripts, in line with their general slower growth observed during culture. Importantly, Wnt/β-catenin signaling was impaired in all STRAP knockout/down cell lines tested, regardless of the underlying CTNNB1 or AXIN1 mutation. In accordance with β-catenin's role in (cancer) stem cell maintenance, the expressions of various stem cell markers, such as AXIN2 and LGR5, were reduced and concomitantly differentiation-associated genes were increased. Together, these results show that the increased STRAP protein levels observed in HCC provide growth advantage among others by enhancing Wnt/β-catenin signaling. These observations also identify STRAP as a new player in regulating β-catenin signaling in hepatocellular cancers.
Elevated STRAP levels in hepatocellular cancers provide a growth advantage by enhancing Wnt/β-catenin signaling.
Hepatocellular carcinoma (HCC) is the sixth most prevalent cancer and the third leading cause for cancer-related deaths worldwide with around 500,000 new cases diagnosed each year (1, 2). Hepatocarcinogenesis initiates with the accumulation of aberrant genetic and epigenetic modifications leading to the dysregulation of signaling pathways, which transform the normal hepatocytes toward malignant phenotypes (3).
Inappropriate activation of Wnt/β-catenin signaling has been frequently reported in HCC (4). As the central component of Wnt/β-catenin signaling, the transcription factor β-catenin is tightly regulated by a multiprotein complex composed of the adenomatous polyposis coli (APC) tumor suppressor, scaffold proteins AXIN1 and AXIN2, and the kinases GSK3 and CK1α (4, 5). In the absence of Wnt ligands, β-catenin is constitutively phosphorylated and degraded to maintain a minimal level in the cytoplasm. On Wnt stimulation, the multiprotein complex dissociates causing the accumulation of cytosolic and nuclear β-catenin. The latter triggers the transcription of specific target genes. Aberrant activation of Wnt/β-catenin signaling in HCC has been attributed to activating mutations in CTNNB1 (20%–25%) or loss of function mutations in AXIN1 (10%), AXIN2 (3%–4%), and APC (1%–2%; refs. 4, 6, 7).
The serine/threonine kinase receptor-associated protein (STRAP) encoded by the STRAP gene harbors seven WD40-repeat domains (8). It is considered to be a scaffolding protein without enzymatic function that exerts regulatory functions on a variety of cellular processes ranging from signal transduction, transcriptional regulation, RNA processing, vesicular trafficking to cell-cycle progression (9). STRAP was shown to be overexpressed and exert oncogenic properties in breast cancer, colorectal cancer, and lung carcinomas (10–12). Originally, STRAP was shown to inhibit canonical TGFβ signaling (13). Later, it became apparent that STRAP modulates various other cellular processes and signaling pathways such as signaling through ASK1, P53, PI3K/PDK1, and P21Cip1 (9, 14, 15). More recently, Wnt/β-catenin signaling was demonstrated to be stimulated by increased STRAP in colorectal cancer through binding with GSK-3β around the catalytic domain, which diminished subsequent ubiquitin-dependent degradation of β-catenin (16). However, the function of STRAP in HCC progression remains elusive.
In this study, we investigated the expression level of STRAP in HCC tumor tissues and used clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9-mediated gene editing to knockout STRAP in HCC cell lines. Our results suggest that upregulation of STRAP protein provides growth advantage to HCC cells via enhancing Wnt/β-catenin signaling. These observations identify STRAP as a new player in regulating Wnt/β-catenin signaling in HCC.
Materials and Methods
Human HCC cell lines Hep3B, HepG2, HepaRG, Huh6, Huh7, PLC/PRF/5, SNU398, SNU182, and SNU449 were cultured as reported previously (7). Identity of all cell lines was confirmed by short tandem repeat genotyping. CTNNB1 and AXIN1 mutations reported in Supplementary Table S1 were confirmed in these HCC cell lines by Sanger sequencing and were in accordance with those reported at COSMIC, the Catalogue Of Somatic Mutations In Cancer (http://cancer.sanger.ac.uk; ref. 17). For the preparation of Huh7-conditioned medium, cells were cultured in complete DMEM medium for 3 days, followed by collection and filtration of medium according to standard procedures.
Tissue microarray (TMA) construction was described previously (18). Briefly, archived formalin-fixed paraffin-embedded tissue samples from 141 patients who underwent hepatic resection for HCC at the Erasmus MC University Medical Center, Rotterdam, between 2004 and 2013, were collected. Three or four 0.6 mm cores from the tumor area as well as two 0.6 mm cores from the corresponding tumor-free liver area of these patients were taken. The TMAs were made using an automated tissue arrayer ATA-27 (Beecher Instruments) or a manual tissue arrayer MTA-1 (Beecher Instruments). Clinicopathologic characteristics are presented in Supplementary Table S2.
The Cancer Genome Atlas (TCGA) LIHC illuminahiseq_rnaseqv2_RSEM_genes_normalized (MD5) data were obtained from the Broad Institute's Firehose (http://firebrowse.org/?cohort=LIHC&download_dialog=true). In this dataset, 373 HCC samples are available for which gene expression analysis was performed (19). In addition, 50 paired adjacent tumor-free tissues are also available for gene expression analysis. RNA sequencing (RNA-seq) levels of STRAP were obtained and matched to the available survival data.
DEN induction of liver tumors in mice
Mice of C57BL/6J background or mixed with C3H/HeOuJ or CD1 (all 3–4 weeks of age) were administrated weekly with diethylnitrosamine (DEN; i.p. injection; 100 mg/kg) for 6 to 17 weeks to induce liver tumor formation. Mice were sacrificed 3 to 16 months after the last DEN injection, after which livers were fixed in PBS-buffered formalin and embedded in paraffin according to routine procedures. All animal experiments were approved by the Committee on the Ethics of Animal Experiments of the Erasmus Medical Center.
The following antibodies were used for Western blot analysis or IHC staining: STRAP (611346, BD Transduction Laboratories and HPA027320, Atlas antibodies), β-catenin (610154, BD Transduction Laboratories), non-phospho (Active) β-catenin (Ser33/37/Thr41; #8814, Cell Signaling Technology), Tubulin (sc-8035, Santa Cruz Biotechnology), β-actin (sc-47778, Santa Cruz Biotechnology) and anti-rabbit or anti-mouse IRDye-conjugated secondary antibodies (LI-COR Biosciences), and horseradish peroxidase (HRP)–conjugated anti-mouse polymer secondary antibody (Envision, DAKO). Propidium iodide solution, diaminobenzidine (DAB), and crystal violet solution were purchased from Sigma.
Paraffin-embedded tumor slides were deparaffinized in xylene, rehydrated in graded alcohols, and then rinsed in PBS with 0.025% Trition. Antigen retrieval was performed in a microwave in Tris/EDTA (pH 8) for 10 minutes. Endogenous peroxidase activity was blocked by incubation in 1.5% H2O2 at room temperature for 15 minutes. After blocking by 5% nonfat dry milk in PBS, the sections were incubated with STRAP antibody (611346, BD Transduction Laboratories; 1:100) at 4°C overnight. HRP-conjugated anti-mouse polymer secondary antibody was then applied for 1 hour. Then reaction products were visualized using DAB and counterstained with hematoxylin. STRAP staining was scored by two independent observers. The intensity of STRAP staining was classified in three categories: 0, 1, and 2, respectively, correlating with weak, moderate, or strong staining, and scored by two independent investigators resulting in a Kappa test of 0.609 (for STRAP in HCC tumors), which was deemed acceptable. In our study, we generated STRAP knockout HCC cell lines that were used to test the specificity of the antibody (Supplementary Fig. S1).
β-Catenin reporter assays
The β-catenin reporter assays were basically performed as previously described (20, 21). In short, 20 hours before transfection, we plated 0.5 × 105 cells per well on 24-well plates. Each well was transfected with 250 ng Wnt-responsive element (WRE) or mutant-responsive element (MRE) vectors and 10 ng CMV-Renilla using FuGENE HD Transfection Reagent (E2311, Promega). We measured luciferase activities in a LumiStar Optima luminescence counter (BMG LabTech) and normalized the data for the transfection efficiency by using the Dual Luciferase Reporter Assay system (E1980; Promega) according to the manufacturer's instruction. Transfections were performed twice in duplicate, and the mean and SE were calculated for each condition. The β-catenin reporter activities are shown as WRE/MRE ratios.
Cells were lysed in Laemmli sample buffer with 0.1 mol/L DTT and heated for 10 minutes at 95°C, followed by loading and separation on a 10% SDS-PAGE. After 90-minute running at 120 V, proteins were electrophoretically transferred onto a polyvinylidene difluoride membrane (Invitrogen) for 1.5 hours with an electric current of 250 mA. The membrane was blocked with Odyssey Blocking Buffer followed by incubation with primary antibody (1:1,000) overnight at 4°C. Anti-rabbit or anti-mouse IRDye-conjugated secondary antibodies (1:5,000) were applied for 1 hour at room temperature. Blots were assayed for tubulin or β-actin content as standardization of sample loading, scanned, and quantified by Odyssey infrared imaging (Li-COR Biosciences). Results were visualized and quantified with Odyssey 3.0 software.
Gene knockdown by siRNA
Smartpool ON-TARGETplus siRNAs targeting STRAP were purchased from Dharmacon. The ON-TARGETplus Non-targeting siRNA #2 was used as negative control. Cells were reverse-transfected in a 24-well plate using a total of 0.8 μL DharmaFECT formulation 4 (Thermo Fischer Scientific) and 25 nmol/L of each siRNA per well. Following 72-hour incubation, the effect of knockdown was tested by Western blotting. Alternatively, 48 hours after reverse transfection, the cells were transfected with WRE or MRE vectors and CMV-Renilla for β-catenin reporter assay.
Construction of CRISPR/Cas9 STRAP-targeting vectors
Single guide RNAs (sgRNA) targeting exon 1 or 2 of human STRAP were designed using the following CRISPR design tool (http://crispr.mit.edu/). Supplementary Table S3 depicts the three selected sgRNAs, chosen because of lowest predicted potential exonic off-target sites. Oligos were dissolved at 100 pmol/μL and annealed by combining 10 μL of each with 2 μL of NEB buffer 3, heated in a PCR machine to 94°C for 4 minutes, removed, and allowed to cool down to room temperature. Annealed oligos were diluted 1,000x in water of which 1 μL was combined with 100 ng of BbsI-digested and purified pX330 in a total ligation volume of 20 μL using 1.5 units T4 DNA ligase. Next, ligated plasmids were electroporated into DH10B E. coli. After plating, correct plasmids were identified and sequence-verified using standard procedures.
Generation of STRAP knockout HCC cell lines
Huh6 and Huh7 cell lines were transfected in 6-well plates using 7.5 μL FuGENE HD Transfection Reagent (E2311; Promega) and 2 μg of each pX330 plasmid per well together with 0.2 μg GFP expression construct. GFP expression was used to select the cells that received high levels of the pX330 CRISPR/Cas9 constructs. After incubation at 37°C for 24 hours, single cells were prepared for fluorescence-activated cell sorting to a 96-well plate. After single-cell sorting, Huh7 cells were maintained in DMEM supplemented with either 20% FCS or 25% Huh7-conditioned medium. Huh6 was cultured in complete DMEM medium.
Clones grown successfully from single cells were first subjected to Western blotting with anti-STRAP antibody (611346; BD Transduction Laboratories). For each cell line, apparently successful STRAP knockout and control clones were selected for DNA sequence verification using oligos shown in Supplementary Table S4. For clones with complicated chromatograms, we also employed next-generation sequencing (NGS) on an Ion-Torrent device using the fusion method for amplicon library preparation. This method uses oligos designed to directly include barcodes and adaptors required for processing on the Ion-Torrent device (see Supplementary Table S5). PCR products were generated using Q5 proofreading polymerase (NEB) according to manufacturer's instructions, followed by purification and NGS according to routine protocols. All selected clones were retested for STRAP-loss using an additional STRAP antibody (HPA027320, Atlas antibodies). Western blot result and observed sequence alterations are depicted in Supplementary Fig. S2.
Colony formation assays
We performed two types of clonogenic assays, i.e., plating of single cells directly on cell culture surface or plating in soft agar. For the former, 1,000 cells for each clone were seeded in 6-well plates and were cultured in 2 mL complete DMEM medium per well. Two weeks later, the cells were washed with PBS, fixed in 4% PBS-buffered paraformaldehyde for 10 minutes, and stained with crystal violet solution. Number of colonies was counted under a microscope.
For the soft-agar assay, a base layer of 0.3 mL of complete DMEM/F12 medium (2% B27, 1% N-2, 20 ng/mL FGF, 20 ng/mL EGF, 100 μg/mL Primocin) containing 0.6% soft agar was allowed to settle in 24-well plates. Next, 0.6 mL of complete 0.6% soft-agar DMEM/F12 medium was added containing 1,000 (Huh6) or 2,000 (Huh7) single cells. After settling of the agar, 0.5 mL of liquid medium was added, which was replaced every other day. After 2 weeks of culturing, colonies were fixed and stained with 0.005% crystal violet in 10% PBS-buffered formalin. Pictures were taken of the complete 24-well, and colonies were automatically counted with ImageJ. All colony formation assays were performed in triplicates. The mean and SE were calculated for each condition.
Periodic acid Schiff staining of cultured cells
Cells were cultured in 6-well plates until they reached 60% to 70% confluency. Next, they were washed with PBS twice followed by fixation in 10% PBS-buffered formalin for 10 minutes. After two ddH2O washes, cells were incubated with 0.5% periodic acid solution for 10 minutes, followed by two ddH2O washes and incubation in Schiff's reagent (Sigma-Aldrich) for 15 minutes. Cells were washed in tap water and visualized under an inverted microscope.
Quantitative real-time PCR
RNA was isolated with the Machery-NucleoSpin RNA II kit (BIOKE) and quantified using a Nanodrop ND-1000. CDNA was prepared from total RNA using a random-primed cDNA Synthesis Kit (Takara Bio) and subjected to qRT-PCR analyses. Analyses were performed using the StepOne Real-Time PCR System and the StepOnev2.0 software (Applied Biosystem). All expression levels are depicted relative to the expression of GAPDH. Primer sequences are provided in Supplementary Table S6.
RNA extraction, Illumina library preparation, and sequencing
Total RNA was isolated with the Machery-NucleoSpin RNA II kit (BIOKE) and quantified using a Nanodrop ND-1000. RNA quality was checked using a RNA Pico chip on the Agilent Bioanalyzer. Library was constructed and sequenced with an Illumina HiSeqTM2000 (GATC Biotech). Briefly, the mRNA was enriched using oligo-dT magnetic beads, followed by fragmentation (about 200 bp). Then the first strand of cDNA was synthesized using random hexamer-primer, and the second strand was further synthesized in a reaction buffer including dNTPs, RNase H, and DNA polymerase I. Double-stranded cDNA was purified with magnetic beads. Then, the 3′-end single nucleotide A (adenine) was added, and adapters were ligated to the fragments which were enriched by PCR amplification.
The Illumina single-end reads were trimmed to remove the TrueSeq adapter sequences using Trimmomatic (v.0.33). Subsequently, the reads were mapped to the human reference genome build hg38 with the RNA-seq aligner STAR (v2.4.2a) and the Homo sapiens GENCODE v23 annotation. Raw counts were measured with summarizeOverlaps function from the Bioconductor GenomicAlignments package (v1.12.1) using the setting mode union. The differentially expressed genes were called with a generalized linear model using a negative binomial distribution and accounting for the different cell lines (Huh6 and Huh7). The calculations were performed by the DESeq2 package (v1.16.1). We applied a Wald test to identify statistical significant differently expressed genes with an FDR that was calculated using Benjamini–Hochberg correction and set a threshold value of 0.01. After blind variance stabilizing log2 transformation of the counts, the differentially expressed genes were used to calculate scaled gene-wise values (Z-score). The scaled values were clustered hierarchically with complete linkage using Euclidean distances and subsequently plotted in a heat map with pheatmap package(v1.0.8). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) gene enrichment analyses were carried out as described previously (22). We used R(v 3.4.0) for statistics and visualization of the data.
All results were presented as mean ± SD or mean ± SEM as described in the figure legends. Comparison of STRAP protein staining between HCC tumor group and adjacent normal groups was performed with test of proportion. The t test was used for statistical evaluation of number of colonies formed in the soft-agar colony assay. Differences were considered significant at a P value less than 0.05.
The RNA-seq data from this study have been submitted to the Gene Expression Omnibus (23) database under the accession number GSE101061.
STRAP protein is upregulated in HCC tumor tissues
In order to assess the expression level of STRAP protein in HCC lesions, we stained a TMA containing HCC tumors and patient-matched adjacent normal tissues. For 109 tumor samples, a STRAP intensity score could be obtained. In most normal samples, STRAP protein was expressed at low to moderate levels, whereas it was significantly elevated in the majority of HCC tumors (Fig. 1). Within the tumor cells, STRAP showed a predominant cytoplasmic location (Fig. 1D). Similar results were observed in DEN-induced liver tumors in mice, in which 22 of 28 tumor nodules showed increased STRAP expression relative to flanking normal liver tissue (Supplementary Fig. S3).
In our cohort, IHC-evaluated STRAP expression was not significantly associated with any of the available clinicopathologic characteristics presented in Supplementary Table S2 (data not shown), including tumor recurrence [HR, 1.10; 95% confidence interval (CI), 0.57–2.10, P = 0.785] or HCC specific mortality (HR, 0.79; 95% CI, 0.37–1.71, P = 0.557). Given that our semiquantitative IHC analysis may miss subtle differences in expression levels, we explored an independent dataset, that is the TCGA liver cancer cohort, which includes 373 HCC cases that were analyzed by RNA expression profiling (19). No significant difference in STRAP RNA expression was observed between tumors and 50 normal adjacent liver tissues. Survival analysis of the top and bottom 30% STRAP expressors revealed a significant trend (P = 0.011) of reduced survival in the high expressing group, suggesting that higher STRAP levels may contribute to tumor progression (Supplementary Fig. S4).
Overall, these analyses show that STRAP protein levels are increased in most liver tumors of man and mouse, but that its expression level is only modestly associated with patient characteristics.
Transcriptome analysis of STRAP knockout clones by RNA-seq
We tested the baseline expression of STRAP in our panel of nine liver cancer cell lines. All showed readily detectable STRAP protein and RNA, with little variation between cell lines (Supplementary Fig. S5). In order to determine the function of STRAP protein for supporting cell growth, we used the CRISPR/Cas9 technology to disrupt STRAP expression in Huh6 and Huh7 cell lines. As shown in Supplementary Fig. S2, the STRAP protein was completely lost in the selected knockout clones of both cell lines.
To investigate the genome-wide effects of STRAP in regulating gene expression in HCC cell lines, total RNA of selected STRAP knockout and control clones (n = 3 each) was subjected to RNA-seq. The hierarchical clustering results successfully distinguished the Huh6 from the Huh7 cell line. Importantly, the STRAP KO Huh6 clones preferentially clustered with Huh7 KO ones. Likewise, control Huh6 and Huh7 clones were clustered. According to STRAP genotype, 5,605 differentially expressed genes (threshold FDR < 0.01) were clustered in both Huh6 and Huh7 (Fig. 2A).
For validation of the differentially expressed genes identified from RNA-seq, a total of eight genes were selected for qRT-PCR in Huh6 and Huh7 cell lines, which were among the top genes either up- or downregulated. As shown in Fig. 2B, log2 fold change of these genes tested by qRT-PCR significantly correlated with those from RNA-seq (R = 0.998 in Huh6 and R = 0.913 in Huh7). Taken together, these data indicate that STRAP plays an important role, directly or indirectly, in the transcriptional regulation of many genes in HCC cell lines.
Loss of STRAP reduces expression of many metabolic and cell-cycle–related genes
Both KEGG pathway enrichment and GO enrichment analysis revealed that many metabolic processes were reduced in activity in the STRAP knockout clones. In addition, this analysis showed that cell-cycle progression was significantly affected, as a result of STRAP loss in both Huh6 and Huh7 (Fig. 3A and B). Both the reduced expression of metabolic and cell-cycle–related genes are in line with the general slower growth of all knockout clones observed during routine culture (data not shown). To assess the role of STRAP on the reproductive viability of HCC cells, a colony formation assay on regular culture plates was employed with the STRAP knockout clones and controls thereof. We observed that loss of STRAP dramatically decreased not only the number but also the size of Huh6 and Huh7 colonies (Fig. 4). These results indicate that STRAP is important for an efficient outgrowth of single HCC cells.
Loss of STRAP attenuates Wnt/β-catenin signaling activity
It has been reported that upregulation of STRAP correlates with increased Wnt/β-catenin signaling activity in colorectal cancer (16). We wondered whether this also holds true for HCC cells, most of which are known to also depend on β-catenin signaling for sustaining optimal growth (24). To this aim, we evaluated the expression change within the RNA-seq data of several (liver-specific) β-catenin signaling target genes reported previously, i.e., AXIN2, LGR5, MYC, CCND1, GLUL, RGN, and BIRC5 (also known as Survivin; ref. 25). In Huh7, all these genes were downregulated in the STRAP knockout clones with the exception of RGN and GLUL. In the CTNNB1-mutant Huh6 cells, the differences were less obvious, but most genes showed a trend toward lower expression (Fig. 5A). The effect of STRAP on Wnt/β-catenin signaling activity was confirmed using a more sensitive β-catenin reporter assay (Fig. 5B).
Using siRNA-mediated knockdown of STRAP, similar reductions in β-catenin reporter activity were observed in the parental Huh6 and Huh7 lines as well as in four additional HCC lines, indicating that STRAP is required to maintain optimal β-catenin signaling in most, if not all, HCC lines (Fig. 5C). This even is the case in the cell lines endogenously expressing a dominant acting β-catenin variant (Huh6, SNU398, and HepG2).
STRAP has been shown to bind to GSK3β around the catalytic domain, thereby reducing the N-terminal phosphorylation of β-catenin (16, 26) and subsequently increasing the active signaling pool of β-catenin, i.e., unphosphorylated at S33/S37/T41. Hence, we investigated the amount of the active β-catenin signaling pool present in the HCC clones. As shown in Fig. 5D, Huh6 showed reduced levels of active β-catenin in the STRAP knockout clones, whereas no change was observed in Huh7.
Besides N-terminal phosphorylation, the signaling strength of β-catenin is also regulated by phosphorylation at its C-terminus. For example, S675 phosphorylation by protein kinase A (PKA) has been shown to increase Wnt signaling by recruiting transcriptional coactivators (21, 27–29). To investigate whether STRAP promotes Wnt/β-catenin signaling through indirectly affecting the phosphorylation of S675, we tested its levels. As indicated in Fig. 5E, reduced levels of phosphorylated β-catenin at S675 were observed in the STRAP knockout clones of Huh6, whereas no clear change was seen in the Huh7 cell line. Thus, in Huh6, loss of STRAP is accompanied by lower levels of unphosphorylated N-terminal β-catenin and reduced phosphorylation at its C-terminus, both features that are associated with reduced signaling, whereas no alteration is observed in Huh7.
Together, these findings suggest that STRAP enhances Wnt/β-catenin signaling in all tested HCC cells. The exact molecular mechanism appears however to differ between cell lines.
Loss of STRAP associates with reduced stemness and increased differentiation markers
Wnt/β-catenin signaling is essential for the homeostatic self-renewal and proliferation of the hepatic stem/progenitor cells (30). In particular, Wnt/β-catenin–driven AXIN2+ (31) and LGR5+ (32) cells have been identified as stem cells that self-renew and give rise to mature hepatocytes. STRAP itself has also been linked to stemness in colorectal cancer (33). Therefore, loss of STRAP and the resulting reduction in β-catenin signaling may lead to reduced expression of stem/progenitor cell markers. We observed a clear reduction at the transcription level of AXIN2 and LGR5 and other liver progenitor markers (SOX9, CD44, and PROM1/CD133; ref. 34) in STRAP knockout Huh6 cells. In Huh7, expression of AXIN2, LGR5, and PROM1 but not SOX9 and CD44 was decreased (Supplementary Fig. S6A).
The colony formation assay shown in Fig. 4 partially supports the stem cell assumption, but may be biased by differences in plating efficiency associated with STRAP loss. Therefore, we performed a similar assay by plating the cells in soft agar. As shown in Fig. 6, a significant reduction in colonies is formed in the STRAP knockout clones of Huh6 (P = 0.001), whereas no clear difference is seen for Huh7.
Conversely, STRAP loss increased most liver differentiation related genes, such as ALB, AFP, and HNF4A (35, 36), in Huh6 and more obviously in Huh7 cells (Supplementary Fig. S6B). To corroborate this observation, we explored glycogen storage as a marker of differentiation in liver cancer cells using periodic acid Schiff (PAS) staining (Fig. 7). All knockout clones of Huh6 showed multiple cells with a prominent PAS staining, whereas this was rarely visible in the control clones. In Huh7, the difference was less obvious, although also here a trend toward more positivity was observed in the knockout clones.
STRAP has been identified as a scaffolding protein that is upregulated in breast, lung, and colorectal cancers, and was shown to promote their growth (10–12). Here, we show that also in most hepatocellular cancers, increased levels of STRAP protein can be observed. Furthermore, knockout experiments in two different HCC cell lines showed that its expression is required to support optimal growth. Mechanistically, we provide evidence that many signaling pathways and metabolic processes are affected following STRAP loss, including the Wnt/β-catenin signaling pathway.
Given the well-known importance of β-catenin signaling to sustain HCC growth, we have investigated this pathway in more detail (24, 25). Knockout/down of STRAP resulted in reduced β-catenin signaling in all six HCC lines investigated, regardless of the underlying CTNNB1 or AXIN1 mutation being present in these lines. This is largely in line with the observation of Yuan and colleagues for colorectal cancer cells (16). Mechanistically, STRAP has been shown to bind to the catalytic site of GSK3β, effectively resulting in reduced N-terminal phosphorylation and reduced breakdown of β-catenin (16, 26). Hence, loss of STRAP is expected to result in more GSK3-mediated phosphorylation of β-catenin, which is in accordance with the notable decrease in nonphosphorylated (active) β-catenin observed in Huh6. In Huh6, a second mechanism appears active by which STRAP increases overall β-catenin signaling. PKA-mediated phosphorylation of β-catenin at S675, previously shown to result in increased signaling (21, 27–29), is about 2-fold higher in the clones that have retained STRAP expression. This result suggests that STRAP may be involved in modulating PKA activity through yet unknown mechanisms. Nevertheless, Huh7 cells apparently share neither mechanism for STRAP to support Wnt/β-catenin signaling, despite being one of the more sensitive cell lines (see Fig. 5B and C), suggesting alternative routes involved. As Wnt/β-catenin signaling can be fine-tuned at multiple levels (37), uncovering the exact mechanism specifically for Huh7 is outside the scope of this current article. Whichever the exact mechanism, our results show that the increased STRAP expression observed in most HCCs will support their growth by increasing overall β-catenin signaling.
The Wnt/β-catenin signaling pathway is well known for its role in (cancer) stem cell maintenance (30–32, 38, 39). Accordingly, the reduced β-catenin signaling following knockout of STRAP is associated with lower expression of the β-catenin–regulated stem cell markers AXIN2 and LGR5. In addition, several other liver progenitor markers (SOX9, CD44, and PROM1/CD133) are reduced in expression, whereas differentiation markers are elevated. These results suggest that one mechanism by which elevated STRAP expression supports HCC growth is to shift the balance toward the induction of stem cell features. This observation is in line with the work of Jin and colleagues who showed that STRAP promotes stem cell features in several colorectal cancer cell lines (33).
Besides its role in fine-tuning β-catenin signaling, STRAP has also been linked to various other cellular processes likely contributing to HCC cell viability (9, 14). One of the first functions attributed to STRAP was its role in inhibiting TGFβ signaling (13). In the normal liver, this signaling pathway has a crucial role in limiting hepatocyte proliferation and inducing differentiation (40). Likewise, TGFβ signaling is considered to act as a tumor suppressor during the early stages of liver tumor formation by inducing cell-cycle arrest and apoptosis. As such, the elevated STRAP levels that we observe in HCC may contribute to tumor growth by restraining the tumor-suppressive effects of TGFβ, which is supported by the reactivation of several TGFβ antiproliferative (CDKN1A, CDKN2B, CDKN1C, and EIF4EBP1) and proapoptotic (BIK, BCL2L11, DAPK1, FAS, and GADD45B) target genes (41, 42) derived from RNA-seq data following STRAP loss in Huh6 and Huh7 cells (Supplementary Fig. S6C). However, the role of TGFβ signaling in liver cancer is complicated by the observation that a subset of tumors becomes resistant to the cytostatic and apoptotic effects of TGFβ and in fact exploits TGFβ to support their growth, migration, and invasion during later stages (40, 43). This indicates that the specific contribution of STRAP to TGFβ-mediated tumor suppression has to be evaluated on a case-by-case basis.
In summary, we show that most HCCs show upregulation of STRAP expression. Our in vitro analyses suggest that its elevated expression is important to support optimal growth by affecting a variety of metabolic processes and signaling pathways of known importance for liver cancer. Especially, its contribution to increase Wnt/β-catenin signaling is likely to be a major effector of its tumor-promoting role.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: W. Wang, M.P. Peppelenbosch, R. Smits
Development of methodology: K. Sideras, H.J.G. van de Werken, M. Lavrijsen
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W. Wang, K. Sideras, M. van der Heide, R. Smits
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Wang, P. Liu, K. Sideras, H.J.G. van de Werken, M. Lavrijsen, M.P. Peppelenbosch, R. Smits
Writing, review, and/or revision of the manuscript: W. Wang, K. Sideras, H.J.G. van de Werken, M.P. Peppelenbosch, M. Bruno, Q. Pan, R. Smits
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W. Wang, S. Li, H.J.G. van de Werken, W. Cao, Q. Pan
Study supervision: M.P. Peppelenbosch, R. Smits
Other (assistance with part of experiments): S. Li
This research was financially supported by a China Scholarship Council PhD fellowship to W. Wang (file no. 201306190123), S. Li (file no. 201408060053), and P. Liu (file no. 201408220029). We thank Manzhi Zhao for helping to score the TMA and Katya Mauff for help with statistical analyses.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.