Abstract
In colorectal cancer, APC-mediated induction of unregulated cell growth involves posttranslational mechanisms that prevent proteasomal degradation of proto-oncogene β-catenin (CTNNB1) and its eventual translocation to the nucleus. However, about 10% of colorectal tumors also exhibit increased CTNNB1 mRNA. Here, we show in colorectal cancer that increased expression of ZNF148, the gene coding for transcription factor ZBP-89, correlated with reduced patient survival. Tissue arrays showed that ZBP-89 protein was overexpressed in the early stages of colorectal cancer. Conditional deletion of Zfp148 in a mouse model of Apc-mediated intestinal polyps demonstrated that ZBP-89 was required for polyp formation due to induction of Ctnnb1 gene expression. Chromatin immunoprecipitation (ChIP) and EMSA identified a ZBP-89–binding site in the proximal promoter of CTNNB1. Reciprocally, siRNA-mediated reduction of CTNNB1 expression also decreased ZBP-89 protein. ChIP identified TCF DNA binding sites in the ZNF148 promoter through which Wnt signaling regulates ZNF148 gene expression. Suppression of either ZNF148 or CTNNB1 reduced colony formation in WNT-dependent, but not WNT-independent cell lines. Therefore, the increase in intracellular β-catenin protein initiated by APC mutations is sustained by ZBP-89–mediated feedforward induction of CTNNB1 mRNA. Cancer Res; 76(23); 6877–87. ©2016 AACR.
Introduction
Mutations in the adenomatous polyposis coli gene (APC) occur in about 80% of colorectal cancers (1, 2). Most of the mutations prevent interaction with the protein product of the CTNNB1 locus (β-catenin), which accumulates in the cytoplasm and then translocates to the nucleus to interact with DNA-binding proteins in the T-cell factor family (LEF/TCF) activating proliferative target genes (2, 3). Although some colorectal cancers exhibit mutations in downstream Wnt target genes, for example, AXIN2, CCND1, and MYC, Wnt-dependent proliferation has been attributed primarily to posttranslational alterations that increase β-catenin protein stability. As a result, therapy to disrupt this process has focused on targeting pathways that decrease nuclear β-catenin protein (4). Although prior studies have shown increased mRNA expression by in situ hybridization, especially in areas of tissue invasion (5, 6), few proposed therapies focus on the regulation of β-catenin gene expression. Recently, it has been shown that thyroid hormone suppresses β-catenin transcription (7). In colorectal cancer liver metastases, β-catenin increases its own expression through TCF-4 (8), suggesting that increasing CTNNB1 mRNA contributes to colorectal cancer progression. Dashwood and colleagues were the first to compare the presence of several regulatory elements within the rat and human CTNNB1 promoters and specifically identified a ZBP-89 DNA-binding site within the first 2,000 bp of the rat promoter (9). In contrast, the human CTNNB1 promoter is more GC rich and contains a TATA box and three putative GC-rich Sp1 sites within the first 150 bp of the transcriptional start site (10). However, the function of these proximal GC-rich elements was not functionally tested.
Zinc finger–binding protein-89 kDa (ZBP-89 protein encoded by the human ZNF148 or mouse Zfp148 locus) is a ubiquitously expressed Krüppel-type zinc finger transcription factor that binds GC-rich DNA elements frequently in concert with Sp1 (11–13). ZBP-89 is required for mucosal protection in the colon when challenged with a pathogen and mediates mucosal restitution mechanisms that work in concert with the Wnt–β-catenin pathway (14, 15). Furthermore, we found that mice carrying a conditional deletion of Zfp148 in intestinal epithelial cells (Zfp148ΔIEC) are more susceptible to mucosal damage from infectious agents in part due to reduced tissue levels of serotonin and antimicrobial peptides (15, 16). ZBP-89 synergizes with β-catenin to induce mucosal defense genes encoding tryptophan hydroxylase 1 and antimicrobial peptides called defensins (15). Having established that ZBP-89 and β-catenin functionally cooperate in the normal colon, we considered whether ZBP-89 regulates CTNNB1 gene expression not only to restore homeostasis but also in promoting neoplastic transformation.
In a study of 742 colorectal cancers, ZNF148 gene expression increased in the transition from normal to stage I colorectal cancer but then decreased at later stages of cancer progression (stages III and IV; ref. 17). This study was consistent with a prior report indicating that ZNF148 is downregulated in colorectal cancers that progress to stage III (18). We previously showed that ZBP-89 forms a protein–protein complex with the tumor suppressor p53 (19), and recent reports indicate that it suppresses p53 function, as genetic deletion of one Zfp148 allele increases p53 activity (20, 21). Moreover, other reports have shown that ZBP-89 protein expression promotes esophageal, renal cell, and hepatocellular carcinoma (22–24). As increased ZBP-89 mRNA and protein expression occurs in several cancers, including stomach (25) and colorectal cancer(17), we hypothesized that the transcription factor likely plays an essential role in tumorigenesis and examined whether ZBP-89 synergizes with Wnt signaling through its ability to regulate CTNNB1 gene expression.
Materials and Methods
Human tissue arrays
Three human tissue arrays (COC1501, COC1502, and COC1503) containing representative samples in duplicate 1.1-mm sections of human colon (4 normal colon tissue, 11 adenomas, 193 and colorectal cancers) were purchased from Pantomics (www.pantomics.com) and include the age, gender, pathology, tumor grade, and stage.
Kaplan–Meier survival curves
Survival curves were generated from Affymetrix microarray data deposited in the Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/). The search terms were “colon,” “cancer,” “GPL96,” “GPL570,” “GPL571,” “survival,” as well as any combination. Only the GEO files with survival data were used and are as follows: RFS total = 1,373: GSE12945 n = 51; GSE14333 n = 226; GSE17538 n = 232; GSE31595 n = 37; GSE33114 n = 90; GSE37892 n = 130; GSE39582 n = 519; GSE41258 n = 115. Twenty-seven of the 1,400 samples in the eight datasets did not report survival data and were therefore excluded from the analysis. Therefore, only files survival data were used. Gene expression and the “relapse-free survival” data were integrated simultaneously into one database. To analyze the prognostic value of the ZNF148 probes BERF-1 or BFCOL, the colon cancer patient samples were split into two groups according to high versus low levels of gene expression and then were used to generate Kaplan-Meier plots of “relapse-free survival.” To define a cut-off value, we computed a Cox regression analysis for each percentile between the lower and upper quartiles of gene expression, and the best threshold was used when drawing the Kaplan-Meier plot (26). The HRs with 95% confidence intervals and log-rank P value were calculated as described previously (27).
Animal models
Generation of Zfp148FL/FL mice on a C57BL/6 genetic background is described previously (15). ApcFL/+, Zfp148FL/+, and Zfp148FL/FL mice were bred to the VillinCre (VC) line to generate mice that were heterozygous for the Apc allele alone or were expressed with the Zfp148FL/+ and Zfp148FL/FL genotypes. All methods and procedures were approved by the University of Michigan Animal Care and Use Committee, which maintains an American Association of Assessment and Accreditation of Laboratory Animal Care facility.
Plasmids and cell lines
Upon receipt, frozen stocks were generated, checked for mycoplasma within 3 months of experiments, and used between 3 and 20 passages after receipt. HEK293 and SW480 colon cell lines were purchased from ATCC in 2015 and the DLD-1 line in 2016. The KM12 cell lines were purchased from the NCI Repository in 2013. The DLD-1 and RKO cell lines were kindly provided by Eric Fearon (University of Michigan, Ann Arbor, MI). All lines except DLD-1 and RKO were cultured in DMEM (Gibco, Life Technologies) supplemented with 10% FBS and antibiotics (100 U/mL penicillin plus 100 μg/mL streptomycin). RKO and DLD-1 cells were cultured in RPMI1640 with 10% FBS, glutamine, and antibiotics.
Transient transfections
Wild-type (WT) and mutant (Mut) CTNNB1-Luc reporters (75 ng per well) with 2 ng/well of Renilla reporter were transfected with Lipofectamine 2000 (Life Technologies) into SW480 cells cultured on 12-well plates for 24 hours. Firefly luciferase was normalized to Renilla. Each experiment was performed in duplicate three times. siRNA transfections were carried out using Lipofectamine RNAi-MAX in antibiotic-free medium. siRNA oligos for ZNF148 were a pooled mixture of four genomic sequences (siGENOME SMARTpool, Dharmacon #M-012658-01). Pooled CTNNB1 siRNAs were from Fisher Scientific (#AM16708).
Chromatin immunoprecipitation
SW480 cells were treated with 1% formaldehyde and then quenched with glycine for 5 minutes at room temperature. Cell lysates (500 μL) were sonicated to shear DNA into primarily 200 to 1,000 bp fragments (four cycles of 30-second intervals) using a 130W Sonics Vibracell (VCX130PB). An aliquot (5%) was removed from the sonicate for total input DNA, and the remainder (475 μL) was used to immunoprecipitate cross-linked protein with either ZBP-89 antibody or IgG (antibody control). Immune complexes were captured using Protein A/G agarose beads. After repeated washes, the bound proteins were eluted from the beads and diluted 1:15 in deionized water (EZ Magna ChIP A/G Kit, #17-10086, Millipore). After proteinase K and RNase A digestions, precipitated chromatin fragments were analyzed by PCR using NovaTaq DNA Polymerase (Millipore) and three sets of primers flanking the CTNNB1, ZNF148, or ChrA promoter regions (Supplementary Data).
To perform chromatin immunoprecipitation (ChIP) analysis of β-catenin binding to the ZNF148 promoter, the transcriptional start site was identified in silico using transcript alignment and the site of POL2A binding within the UCSC Genome browser (http://genome.ucsc.edu). ChIP–qPCR values were expressed as a percentage of input DNA after sequence confirmation. Primers for human chromogranin A (CHGA) were used as the nonrelated promoter control. PCR products were resolved on a 1% agarose gel, sequenced, and then quantified by qPCR for triplicate experiments.
qPCR
Cultured cells were harvested in TRIzol (Invitrogen) for RNA extraction and DNase treatment. Quantitative PCR analysis was carried out on triplicate cDNA samples using the Bio-Rad C1000 thermal cycler, analyzed with CFX ManagerV2 software, and normalized to HPRT. Experiments were performed at least three times in triplicate. Mouse PCR primer pairs for Zfp148 were: forward: 5′-GGC ATG TCT TCA TTC ATA GAG G; reverse: 5′-CTC ATA CCA CAT TCA TCA CAG C. Hprt primer pairs were: forward: 5′-AGT CCC AGC GTC GTG ATT AGC; reverse: 5′-ATA GCC CCC CTT GAG CAC ACA G.
Colony formation assay
DLD-1 and RKO cells were seeded onto 6-well plates in growth medium (10% FBS) at 1,000 cells per well. For KM12 L4 and SW480 cells, 100 cells per well were seeded onto 6-well plates in growth media containing 10% FBS. Sixteen hours after plating, the cells were transfected with 25 nmol/L siGenome SMARTpool siRNAs for ZNF148 (M-012658-01, Dharmacon), CTNNB1 (M-003482-00, Dharmacon), or nontargeted controls (D-001206-13, Dharmacon). Media were replenished every 3 days with siRNAs or nontargeted controls until colonies formed at week 3. Plating efficiency was 20%. The wells were rinsed with PBS, fixed in 10% formalin for 30 minutes, and then stained for counting with a solution of 0.1% crystal violet. The number of colonies formed for each treatment was expressed as a percentage of the nontargeted control siRNA.
MTT assay
The MTT assay was performed according to the manufacturer's instructions. Briefly, RKO, DLD-1, KM-12, and SW480 cells (1,000 cells/well) were transfected with the gene-specific and nontargeted siRNA oligos (25 nmol/L). Four days after transfection, 0.5 mg/mL of MTT labeling solution was added to each well and incubated for 4 hours at 37°C. Formazan crystals generated were dissolved by incubating with the solubilization reagent overnight, and the absorbance was measured on plate reader at A570 and reference wavelength of A660. Cell viability was determined using the equation:
The indicated cell lines were plated in 96-well (1,000 cells/well) plates. Twenty-four hours after plating, the cells were transfected with 25 nmol/L of a nontargeted, ZNF148 or CTNNB1 siRNA using PolyJet transfection reagent (SignaGen) according to the manufacturer's instructions. The S33Y cDNA expression vector was transfected 24 hours after transfection of the siRNA again using the PolyJet reagent. Two days after both transfections, 0.5 mg/mL of MTT labeling solution was added, incubated for 4 hours at 37°C, and then measured as described above.
Intestinal enteroid generation
The intestinal mucosa was minced into 5-mm pieces on ice followed by gland dissociation in 2 mmol/L EDTA/PBS at 4°C for 30 minutes. After filtration through a 100-μm nylon mesh, the suspension was centrifuged at 200 × g for 5 minutes, the glands were resuspended in 50 μL of Matrigel supplemented with Advanced DMEM/F12 (Invitrogen) containing 50% Wnt plus R-Spondin conditioned media generated from stably expressing L cells (28) with additional factors purchased from Invitrogen, that is, Noggin (100 ng/mL) and EGF (50 ng/mL), and then cultured at 37°C in 5% CO2. Two days after culturing the crypt Matrigel, suspension enteroids were observed in the WT crypt cultures. The presence of budding enteroids versus spheroids was enumerated over 8 days after the initial plating plus three additional passages.
Statistical analysis
Comparison among multiple histologic groups was performed using the Kruskal–Wallis test as detailed above. One-way ANOVA with Tukey test for multiple comparisons was used to compare polyp numbers, normalized luciferase values, MTT, and colony-forming assays. Two-way ANOVA was used to analyze the enteroid/spheroid time course.
Results
Increased ZNF148 expression correlates with poor survival in colorectal cancer
Although elevated ZBP-89 protein expression correlates with poor survival in colorectal and esophageal adenocarcinomas (17, 18, 22), The Cancer Genome Atlas databases show that the ZNF148 locus that encodes ZBP-89 is mutated in only 1.4% of colorectal cancers (total of 504 cases from three databases; Supplementary Table S1). To examine ZBP-89 protein expression in the colon at different histologic stages, immunohistochemical analysis was performed on three sets of colorectal cancer tissue arrays. ZBP-89 protein expression in normal human colon was low as observed for β-catenin protein expression (Fig. 1A and B). Although most of the ZBP-89 staining remained cytoplasmic and/or membranous (Fig. 1C, E, G, I, and K) and was completely negative in 2.6% of the samples, the intensity of ZBP-89 expression in adenomas increased significantly in parallel with increased β-catenin expression compared with normal colon (Fig. 1C and D). Moreover, the membranous and cytoplasmic staining pattern for both ZBP-89 and β-catenin in colorectal cancer was consistent with prior reports (http://www.proteinatlas.org). Nuclear β-catenin was initially observed in stage I and II colorectal cancers and coincided with approximately 60% of the tissues expressing ZBP-89 protein (Fig. 1E–L). The intensity of the signal for both proteins remained elevated in all four cancer stages compared with normal colon (Fig. 1M and N). In addition, we examined whether elevated ZNF148 mRNA expression correlated with survival. Microarray datasets for colorectal cancer were downloaded from the GEO public database for analysis using two probe sets (BFCOL and BERF1) that recognize sequences at the C-terminus and 3′UTR of ZNF148 (Supplementary Fig. S1). Elevated ZNF148 mRNA expression levels in colorectal cancer patients correlated with worse relapse-free survival compared with patients with low levels of expression (Fig. 1O and P). Therefore, we tested the hypothesis that elevated ZBP-89 expression contributes to adenoma progression to cancer after disruption of one Apc allele, which generates higher levels of β-catenin due to protein stabilization.
Conditional Zfp148 deletion reduced Apc+/ΔIEC-mediated polyp formation
To determine how ZBP-89 expression contributes to adenoma progression, we conditionally deleted one or both Zfp148 alleles in a mouse model of β-catenin–driven polyps in which one Apc allele was also conditionally deleted using the VillinCre mouse line (VC:Apc+/FL;Zfp148FL/FL, hereafter referred to as Apc+/ΔIEC; Zfp148ΔIEC). Deletion of one or both Zfp148 alleles alone does not alter intestinal growth (15), consistent with no significant difference in Ki67 staining in the crypt zone (Supplementary Fig. S2). Therefore, the Zfp148ΔIEC mice were placed on the mutant Apc genetic background (Apc+/ΔIEC) to stabilize the cellular levels of β-catenin and subsequently predispose the intestinal mucosa to hyperplasia and polyp formation. Deletion of one or both Zfp148 alleles significantly reduced β-catenin protein and mRNA tissue levels (Fig. 2A and B). Mice heterozygous for the Apc allele (Apc+/ΔIEC) generated numerous small intestinal polyps within 5 months after birth (Fig. 2C–E), consistent with prior reports (29). Moreover, deleting one or both Zfp148 alleles significantly reduced the number of intestinal polyps (Fig. 2F), but not the size or location of the Apc-dependent polyps (Fig. 2H and I). Insufficient numbers of polyps developed in the colon to assess the effect of deleting Zfp148 (Fig. 2G). In the context of the Apc deletion alone, β-catenin protein levels were higher than with intact Apc loci, and there was no significant effect on mRNA levels (Fig. 2A and B versus D and E). Elevated levels of β-catenin protein persisted even with the loss of one or both Zfp148 alleles (Fig. 2D) due to stabilization of β-catenin protein with deletion of one Apc allele (30, 31). Nevertheless, deleting both Zfp148 alleles significantly decreased both β-catenin protein and mRNA (Fig. 2D and E). Collectively, these results were consistent with direct transcriptional induction of the ctnnb1 promoter by ZBP-89 or indirectly through mRNA stability contributing to the cellular pool of β-catenin protein.
ZNF148 deletion reduces Wnt-dependent cell growth
To further test the functional dependence of β-catenin expression on ZBP-89, colony-forming assays were performed after knocking down ZNF148 expression by siRNA in β-catenin–dependent colorectal cancer cell lines (DLD-1, SW480), which produce truncated APC protein (32) or β-catenin–independent colorectal cancer cell lines (RKO, KM12), which produce wild-type APC (Fig. 3A and B; ref. 33). Reduced ZBP-89 expression decreased colony formation (Fig. 3A and B) and cell viability (Fig. 3C) in the β-catenin–dependent cell lines by 50% but had no effect on the β-catenin–independent cell lines. Reducing ZBP-89 expression decreased β-catenin protein expression, while siRNA-mediated knockdown of CTNNB1 also reduced ZBP-89 expression (Fig. 3D).
Reducing CTNNB1 expression depressed colony formation by approximately 90% in the β-catenin–dependent, but not the β-catenin–independent cell lines (Fig. 3A–D). Furthermore, reduced cell viability in the SW480 line was reversed by overexpressing constitutively active S33Y mutant β-catenin protein but had no effect on the viability of the β-catenin–independent KM12 cell line (Supplementary Fig. S3). Thus, we concluded that ZBP-89 modulation of cell growth in a Wnt-dependent cell line requires β-catenin. In addition, increasing ZBP-89 expression dose dependently increased β-catenin protein as well as colony growth in the DLD-1 cell line (Fig. 3E and F). Therefore, in the context of Wnt signaling, ZBP-89 was necessary and sufficient for Wnt-dependent cell growth.
Zfp148 deletion promotes enteroid budding
Wnt signaling is highest at the base of the intestinal crypt, where the stem cells reside (34). Formation and sustained propagation of intestinal enteroids requires cells from the stem cell niche (35). Deletion of Apc accompanied by elevated Wnt signaling promotes spheroid-shaped enteroids remarkable for the absence of crypt buds (35). As previously reported (36), we found that enteroids from wild-type mice form budding appendages indicative of cells from the intestinal crypt composed of a mixture of stem- and transit-amplifying cells (Fig. 4A). In contrast, glands isolated from ApcΔIEC/+ mice remain spheroid consistent with high Wnt activity and maintenance of the stem cell phenotype (Fig. 4B) and were readily quantified (Fig. 4C). Enteroids prepared from the Zfp148ΔIEC/ΔIEC mice did not exhibit significant differences in their ability to form budding enteroids compared with the Cre negative control enteroids (Fig. 4D and E). In contrast, glands from ApcΔIEC/+;Zfp148ΔIEC/+ mice over time exhibited some budding compared with the ApcΔIEC/+ enteroids, which generated small enteroids by day 8 (Fig. 4D and E). Therefore, the spheroid phenotype correlated with synergy between Zfp148 and elevated Wnt signaling in the stem cell niche after loss of one Apc allele, while budding occurred with deletion of one Zfp148 allele. Apparently, elevated Wnt signaling in the absence of even one Zfp148 allele was sufficient to induce budding.
ZBP-89 directly regulates the CTNNB1 promoter
Hypothesizing that ZBP-89 directly binds to the CTNNB1 promoter, we identified a consensus ZBP-89 site at −120 bp in silico upstream from the transcriptional start site of the human CTNNB1 gene (Fig. 5A). ChIP analysis revealed that the most proximal ZNF148 element at −120 bp (within the −209 to −1 segment) contained the highest affinity DNA-binding site compared with overlapping DNA segments at −422 to −261 and −539 to −391 (Fig. 5A). The putative GC-rich ZBP-89–binding site at −120 was examined further by EMSA and revealed several complexes. A portion of the upper complex contained Sp1, as the addition of Sp1 antibody supershifted the complex. There was decreased probe binding when ZBP-89 antibody was added, indicating that a component of the same complex contained ZBP-89 (Fig. 5B). We used DNA affinity precipitation using a biotinylated probe to further demonstrate that ZBP-89 binds specifically to the DNA element at −120 bp (Fig. 5C). Adding WT oligo competitively eliminated protein binding to the biotinylated probe (Fig. 5B, lane 3), whereas protein binding was retained with the addition of the 2-bp mutant oligo (Fig. 5B, lane 4). Next, we showed that full-length ZBP-89 expression induced a CTNNB1 luciferase reporter containing 591 bp of the promoter but did not induce the reporter in both SW480 and HEK293 cells if the 2-bp point mutation was introduced into the ZBP-89 DNA-binding site at −120 (Fig. 5D, E, and G). Cotransfecting ZBP-89 deletion mutants with the reporter plasmid revealed that induction required the presence of zinc fingers located within the amino terminal domain (Fig. 5D, F, and H). Therefore, ZBP-89 protein binds directly to the CTNNB1 promoter to induce transcription. This mechanism was also consistent with the observation that removal of the Zfp148 alleles decreased β-catenin mRNA and protein expression in the intestinal glands, which subsequently decreased polyp formation in Apc-mutant mice (see Fig. 2).
As the knockdown of CTNNB1 expression decreased ZBP-89 expression (Fig. 3C), we tested the possibility that Wnt–β-catenin signaling induces the ZNF148 locus, subsequently increasing the cellular levels of ZBP-89 protein. In silico analysis of the proximal ZNF148 promoter revealed putative TCF/LEF sites between −987 and −583 (Genomatix), which was confirmed by ChIP analysis using antibodies to β-catenin (Fig. 6A–C). Ectopic expression of WT or constitutively active β-catenin in SW480 cells induced ZNF148 mRNA most dramatically in the presence of Wnt3A-conditioned media (Fig. 6D). Furthermore, ectopic expression of ZBP-89 slightly induced ZNF148 mRNA expression, suggesting some degree of self-regulation (Fig. 6D). Thus, as both ZBP-89 and β-catenin reciprocally induced gene expression of each other, the results collectively suggested feedforward synergy between the CTNNB1 and ZNF148 loci (Fig. 7).
Discussion
Mutations in the APC gene locus increase cellular levels of β-catenin through a variety of pathways (2). However, the primary focus has generally been on posttranslational mechanisms resulting in β-catenin protein stabilization and its translocation to the nucleus after dissociation from wild-type APC protein, β-catenin phosphorylation, ubiquitination, and shuttling to the proteasome. More recently, reports have acknowledged that CTNNB1 gene expression also contributes to increased cellular levels of this proto-oncogene in the context of APC mutations and appears to be related especially to the degree of colorectal cancer invasiveness (8). Yet, the transcription factors inducing its gene expression have not been examined in detail. Comparative analysis of the CTNNB1 promoter among mammalian species indicates that it is GC-rich and contains several putative Sp1/ZBP-89 binding sites within the proximal promoter (10).
The ZBP-89 transcription factor family consists of 2 other genes, ZNF281 and ZNF740 (genecards.org). ZNF281 encodes ZBP-99 protein, which recognizes the same GC-rich binding site as ZBP-89, as their DNA-binding domains are 79% identical and 91% similar (37). We therefore considered that ZBP-99 and ZBP-89 might regulate some of the same target genes. Coincidentally, the role of ZBP-99 in intestinal stem cells has been studied in some detail with respect to regulating β-catenin and colorectal cancer progression (38, 39). Specifically, the transcription factor SNAIL, which promotes epithelial-to-mesenchymal transition directly induces ZNF281 gene expression. Moreover, quantitative ChIP analysis revealed that ZBP-99 occupies the promoter of the Wnt target gene LGR5, which ultimately increases β-catenin activity and the acquisition of stem cell traits (38). ZBP-99 binds the proximal CTNNB1 promoter and induces CTNNB1 mRNA in osteogenic stem cells, demonstrating that a ZBP-89 family member directly binds and regulates CTNNB1 gene expression (40). Thus, ZBP-99, a ZBP-89 family member, is able to indirectly regulate β-catenin activity in colorectal cancer cell lines and directly bind to the CTNNB1 promoter in specific cell types, suggesting that ZBP-89 might also regulate CTNNB1 gene expression.
Like ZBP-99, we show here that ZBP-89 also directly binds the CTNNB1 promoter. ZBP-89 binding contributed to elevated β-catenin activity, as deleting one or both Zfp148 alleles was sufficient to reduce β-catenin mRNA and protein expression, resulting in fewer polyps. Moreover, elevated β-catenin activity in Apc-mutant organoids lost its stem cell phenotype (spheroid shape), when Zfp148 was deleted, suggesting that ZBP-89 also correlates with stemness. A recent report of Zfp148 deficiency also found reduced polyp formation in an Apcmin model of intestinal polyps but did not detect changes in β-catenin (21), perhaps due to reduced Zfp148 in the immune compartment where ZBP-89 is also highly expressed (12).
At least 80% of colorectal cancers exhibit APC mutations and correspond to elevated levels of β-catenin protein, raising the question as to how ZNF148 gene expression increases under conditions of elevated β-catenin levels (31). Indeed, we found that knocking down CTNNB1 mRNA was sufficient to dramatically reduce ZBP-89 protein expression. Thus, we conclude that β-catenin protein accumulates in the absence of a functional APC allele and initiates a program of cell proliferation through its target genes, including induction of ZNF148 gene expression. Subsequently, sustained levels of CTNNB1 expression can be maintained through the binding of ZBP-89 protein to the CTNNB1 promoter, further expanding the pool of β-catenin protein. The absence of ZNF148 mutations, yet increased mRNA levels, in APC+/−-dependent tumors suggests that unregulated Wnt signaling initiates the increase in ZBP-89 gene expression. In this way, the two transcriptional regulators provide a “feedforward” mechanism to sustain unregulated β-catenin signaling in the stem cell niche (Fig. 7).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: B.E. Essien, M. Saqui-Salces, J.L. Merchant
Development of methodology: B.E. Essien, S. Sundaresan, R. Ocadiz-Ruiz, A. Chavis, M. Saqui-Salces, B. Győrffy, J.L. Merchant
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B.E. Essien, S. Sundaresan, R. Ocadiz-Ruiz, A. Chavis, A.C. Tsao, A.J. Tessier, M.M. Hayes, M. Saqui-Salces, A.J. Kang, Y.M. Shah
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B.E. Essien, S. Sundaresan, R. Ocadiz-Ruiz, M. Saqui-Salces, Y.M. Shah, B. Győrffy, J.L. Merchant
Writing, review, and/or revision of the manuscript: B.E. Essien, M. Saqui-Salces, Y.M. Shah, J.L. Merchant
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B.E. Essien, A.C. Tsao, M.M. Hayes, A. Photenhauer
Study supervision: B.E. Essien, J.L. Merchant
Acknowledgments
The authors thank the Molecular Biology Core of the UM DDRC P30 DK-034933 and Frederick L. McDonald III for assisting with primer development for the ZNF148 promoter and schematic figure design.
Grant Support
This study was supported by grants R01 DK55732 to J.L. Merchant and R01 CA148828 to Y.M. Shah.
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.