- Research article
- Open Access
Loss of DIP2C in RKO cells stimulates changes in DNA methylation and epithelial-mesenchymal transition
BMC Cancer volume 17, Article number: 487 (2017)
The disco-interacting protein 2 homolog C (DIP2C) gene is an uncharacterized gene found mutated in a subset of breast and lung cancers. To understand the role of DIP2C in tumour development we studied the gene in human cancer cells.
We engineered human DIP2C knockout cells by genome editing in cancer cells. The growth properties of the engineered cells were characterised and transcriptome and methylation analyses were carried out to identify pathways deregulated by inactivation of DIP2C. Effects on cell death pathways and epithelial-mesenchymal transition traits were studied based on the results from expression profiling.
Knockout of DIP2C in RKO cells resulted in cell enlargement and growth retardation. Expression profiling revealed 780 genes for which the expression level was affected by the loss of DIP2C, including the tumour-suppressor encoding CDKN2A gene, the epithelial-mesenchymal transition (EMT) regulator-encoding ZEB1, and CD44 and CD24 that encode breast cancer stem cell markers. Analysis of DNA methylation showed more than 30,000 sites affected by differential methylation, the majority of which were hypomethylated following loss of DIP2C. Changes in DNA methylation at promoter regions were strongly correlated to changes in gene expression, and genes involved with EMT and cell death were enriched among the differentially regulated genes. The DIP2C knockout cells had higher wound closing capacity and showed an increase in the proportion of cells positive for cellular senescence markers.
Loss of DIP2C triggers substantial DNA methylation and gene expression changes, cellular senescence and epithelial-mesenchymal transition in cancer cells.
The disco-interacting protein 2 homolog C (DIP2C), an uncharacterised gene expressed at high level in most human solid tissues and adult tumour types , was identified by us as a putative cancer gene in exome-wide mutational analyses of hormone-receptor negative breast tumours [2, 3]. Further studies have estimated the DIP2C somatic mutation prevalence at ~5% of breast cancer cases . Recently, DIP2C was also found mutated in 9-14% of small-cell lung cancers , strengthening the evidence for a role in tumorigenesis.
Conserved across species, the human DIP2 family proteins DIP2A, DIP2B and DIP2C are highly similar, with DIP2C and DIP2B sharing 72.2% amino acid identity . All three proteins are predicted to contain DMAP1 binding (pfam06464) and AMP binding (pfam00501) domains, which give properties of binding to the transcriptional co-repressor DNA methyltransferase 1 associated protein 1 (DMAP1), and acting enzymatically via an ATP-dependent covalent binding of AMP to their substrate, respectively. The most studied family member, DIP2A, is a potential cell membrane receptor for Follistatin-like 1 (FSTL1), a secreted protein with possible role in e.g. regulation of embryonic tissue formation, joint inflammation and allograft tolerance [7, 8]. Nervous-system specific expression of Dip2 protein has been shown in mouse and Drosophila during embryonic development , which is interesting considering that all three isoforms are associated with neurodevelopmental disorders. The DIP2A gene is a candidate for developmental dyslexia and autism [10, 11], DIP2B deficiency has been associated with mental retardation , and DIP2C has been implicated in developmental delay . While DIP2A lacks known association to cancer development, an SNP associated with DIP2B expression has been proposed to affect colorectal cancer risk . Thus far DIP2C is the only family member that has been identified as a candidate cancer gene through somatic mutation analysis.
Mutations found in breast cancers are predicted to inactivate DIP2C function . To investigate the role of DIP2C inactivation in human cancer and identify processes affected by the activity of this gene we engineered and characterised human DIP2C knockout cell lines which revealed that loss of DIP2C affects cell growth, cell cycle regulation, and migratory capacity, potentially through regulation of DNA methylation.
The DIP2C knockout construct was designed using CCDS7054.1. Primers are listed in Additional file 1: Tables S1 (PCR) and S2 (RT-qPCR). Exon 9 was chosen for deletion based on its location early in the transcript, as well as conforming to criteria for successful rAAV-mediated gene targeting as described in literature [14, 15]. Homology arm (HA) sequences were PCR amplified from RKO (ATCC, Manassas, VA, USA) gDNA using Platinum Taq DNA Polymerase High Fidelity (Invitrogen, Carlsbad, CA, USA) and a touchdown cycling protocol with restriction endonuclease-site tagged primers 1-4. The amplified HAs were then digested with the respective restriction endonucleases (Fermentas/Thermo Scientific, Waltham, MA, USA). The selection cassette, containing an IRES neo gene flanked by LoxP sites was excised from the pSEPT vector  by XbaI and XhoI (Fermentas) digestion. The AAV vector backbone with inverted terminal repeats (ITRs) and an ampicillin bacterial resistance marker was released from the pAAV-MCS vector (Stratagene, San Diego, CA, USA) by NotI digestion, and gel purified alongside the excised selection cassette and the digested HAs. The 5′ HA, selection cassette, and 3′ HA were ligated between the AAV vector backbone ITRs using T4 DNA ligase (Fermentas). Fragment cloning and orientation was confirmed by PCR (primers 5-8) and Sanger sequencing. The DIP2C-rAAV virus particles were produced by transfection of 70% confluent AAV-293 cells (Stratagene; cultured in DMEM, 10% FBS and 1% penicillin/streptomycin (PEST) (all from Gibco/Life Technologies, Carlsbad, CA, USA)) with Lipofectamine (Invitrogen) and 5 μg each of pAAV-RC, pHELPER (Stratagene) and the targeting construct, with harvesting of the cell lysate after 48 h as described .
Cell lines and targeting
The human colorectal cancer cell line RKO (ATCC, CRL-2577) was cultured in McCoy’s 5A (Gibco), 10% FBS and 1% PEST. Human immortalized mammary epithelial cell line MCF10a (ATCC, CRL-10317) was cultured in DMEM-F12 (Gibco), 5% horse serum (Gibco), 0.02 μg/ml EGF (PeproTech, Rocky Hill, NJ, USA), 10 μg/ml Insulin (Sigma-Aldrich, St. Louis, MO, USA), 0.5 μg/ml Hydrocortisone (Sigma-Aldrich), 0.1 μg/ml Cholera Toxin (Sigma-Aldrich) and 1% PEST. Cells were transfected with DIP2C-rAAV as described , and selected for 2 weeks at limiting dilution with 0.8 mg/ml (RKO) or 0.1 mg/ml (MCF10a) Geneticin (Gibco). Single-cell clones with site-specific construct integration were identified by PCR (primer pairs 9 + 10, 6 + 9, 7 + 10). The neo selection cassette was removed by Ad-Cre virus (Vector Biolabs, Malvern, PA, USA) infection . Single-cell clones identified by PCR (primers 11 + 12) to lack selection cassette were verified to be sensitive to Geneticin. A second targeting round was carried out as described above to generate homozygous knockouts. For overexpression, parental RKO cells were transfected with 2.5 μg Myc-DDK tagged DIP2C TrueORF Gold cDNA clone expression vector (RC209325, OriGene, Rockville, MD, USA) and Lipofectamine 2000 (Invitrogen) and enriched for stable integration in 0.8 mg/ml Geneticin. Single-cell clones overexpressing DIP2C were identified by RT-qPCR (primers DIP2C F and DIP2C R), and construct integration was verified in gDNA by PCR (primers 13-16). The RKO cells were authenticated by STR profiling at ATCC (June 2016). The DIP2C knockout and overexpression cells had 86-97% of their respective STR alleles in common with the parental RKO cells. The MSI status of RKO cells and establishment of clones from single cells are plausible sources for variation in alleles, as suggested by others . Cells were tested for mycoplasma using the MycoAlert mycoplasma detection kit (Lonza, Basel, Switzerland).
Cell morphology and growth
Cells were imaged with an IncuCyte HD (Essen BioScience, Ann Arbor, MI, USA) every 6-12 h during culturing, recording cell confluency for growth curves. Alternatively, since DIP2C KO cells differed in size, growth curves were generated by collection and counting of cells at set time points. For cell size comparison, cell diameter data was collected from the Cedex cell counter (Roche Innovatis, Switzerland) at eight occasions for a total of >5000 cells/cell line. For colony formation analyses, 400 cells plated in triplicate in 6-well plates were stained with 5% methylene blue in methanol after 10 days and colonies quantified. The plating efficiency was calculated as the number of obtained colonies divided by the number of seeded cells. For cell cycle analysis, equal numbers of cells fixed in ice cold 70% ethanol were stained with FxCycle PI/RNase staining solution (Molecular Probes, Eugene, OR, USA) for 15 min at room temperature, washed once in PBS, and analysed using a FlowSight flow cytometer (Amnis, Merck Millipore, Darmstadt, Germany).
Samples lysed in RIPA buffer (25 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1% NP40, 0.1% SDS) with protease inhibitors (Roche) were separated on NuPAGE Novex 4-12% Bis-Tris protein gels with 1× Novex Bis-Tris MOPS running buffer (Life Technologies), transferred onto Hybond C-Extra membranes (Amersham Biosciences, UK), and probed with mouse anti-DIP2C antibody (SAB1411930, Sigma Aldrich) diluted 1:300, rabbit anti-ZEB1 (HPA027524, Atlas Antibodies, Stockholm Sweden) diluted 1:300, mouse anti-p53 DO-1 antibody (sc-126) diluted 1:1000, and rabbit anti-p21 H-164 antibody (sc-756) diluted 1:300 (both from Santa Cruz Biotechnology, Dallas, TX, USA). Secondary antibodies Pierce goat-anti-mouse (#31430) and goat-anti-rabbit (#31460) (Thermo Scientific) were diluted at 1:10,000. Mouse anti-β-actin (A5441, Sigma Aldrich) was used as loading control. Immunoreactive proteins were visualized using SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific) on the ImageQuant LAS 4000 imaging system (GE Healthcare) (exposure p53 – 10-20 s, p21 – 1-3 min, β-actin - 0.5-1 s). Relative protein amounts were quantified by densitometric analysis using ImageJ .
Integrity and concentration of RNA was determined using a RNA 6000 nano chip on the Bioanalyzer 2100 instrument (Agilent, Santa Clara, CA, USA). Samples were sequenced on the Ion Proton system (Ion Torrent/Life Technologies) at the SciLife Lab NGI Uppsala platform. RNA-sequencing reads were aligned to the UCSC database hg19 human genome sequence (downloaded with the gene coordinate references via the Illumina iGenomes project ) using Tophat2 (version 2.0.4) . Gene expression level quantification and identification of differentially expressed genes was done using Cufflinks (version 2.1.1) . The ten most up- and downregulated genes across samples were selected for RT-qPCR validation, excluding genes without data in a DIP2C −/− clone, and genes with data in <2 knockout samples. The GSEA MSigDB v5.0  Hallmarks gene set  and the DAVID Bioinformatics Resources [24, 25] (v6.8, annotation category GOTERM_BP_FAT) were used to compute overlaps and identify enriched biological functions for regulated genes. The entire RNAseq data set has been deposited in the NCBI Gene Expression Omnibus (GEO) database  (accession number GSE80746).
Primers for qPCR were designed online using ProbeFinder (Roche), or retrieved from literature (IL13RA2, p14 ARF, p16 INK4a) [27, 28], and evaluated by a five-step 1:5 dilution standard curve. TATA-box binding protein (TBP) was selected as reference gene by stability evaluation across independent RKO and DIP2C knockout samples using the Cotton EST database RefFinder tool . The Maxima H minus First strand cDNA synthesis kit (Thermo Scientific) with random primers was used for cDNA synthesis. Technical triplicate 20 μl qPCR reactions with 1× Maxima SYBR Green/ROX qPCR Master Mix (Thermo Scientific) and 0.3 μM of each primer were run on the StepOne Real-Time PCR system (Applied Biosystems, Foster City, CA, USA), including controls for gDNA contamination, and no-template controls. Data was analysed by the ΔΔCt method using the StepOne software v2.1 (Applied Biosystems).
DNA methylation analysis
Genomic DNA was analysed on the Infinium HumanMethylation 450 K Bead Chip array (Illumina, San Diego, CA, USA) at the Uppsala SciLife Lab NGI SNP/SEQ technology platform. Raw data IDAT files were processed at the Uppsala Array and analysis facility. Color balance adjustment and background correction  was performed in the statistical computing language R , using the methylumi package from the Bioconductor project . Filtering was performed using the “pfilter” function with default settings from the R-package watermelon, available from the Bioconductor project , removing sites with bead count <3 in >5% of the samples, and sites where >1% of samples had a detection p value >0.05. Quantile normalization of the pooled signal intensities of methylated and unmethylated probes was done before calculation of β-values with the “nanet” method from the same R-package. The probe type bias in the Illumina Infinium technology was eliminated by beta mixture quantile dilation (BMIQ), as suggested by others . CpG sites were annotated to RefSeq genes according to the Human Methylation 450 k manifest file version 1.2, and selected according to gene context. The β-value median for sites in each RefSeq gene was calculated, using CpG sites annotated to more than one gene in the calculation of the median for all those genes, and used to calculate beta-diff values. Functional enrichment analysis was performed using the GSEA MSigDB Hallmarks gene set as described above. The DNA methylation data set was deposited in the NCBI GEO database  (accession number GSE86402).
Senescence and scratch assays
Sub-confluent cells were stained using the Senescence β-Galactosidase Staining Kit (Cell Signaling Technology, Danvers, MA, USA), and blue staining was visualised by light microscopy. Results were quantified using the Cell counter plugin for ImageJ. Confluent cell monolayers in 6- or 12-well plates were scratched with a 200 μl pipet tip, washed with fresh medium at least three times and imaged by light microscopy or in an IncuCyte HD instrument. For microscopy images, the TScratch software  was used to calculate the open wound area at defined time points. For IncuCyte images, the surface covered by cells calculated by the instrument was used to determine the open wound area in accordance with TScratch.
Generation of DIP2C knockout cells
The DIP2C missense and frameshift mutations identified in breast cancer in previous studies  are located predominantly in the first half of the transcript but outside the predicted DMAP1 binding domain (Fig 1a). By recombinant adeno-associated virus (rAAV) mediated gene targeting we generated DIP2C-deficient human cells containing a genomic 48 bp deletion in DIP2C exon 9 (Fig. 1b). We first targeted DIP2C in the breast epithelial cell line MCF10a, but failed to identify construct integration despite several attempts (not shown). We then targeted the human colorectal cancer cell line RKO, also well known to function with rAAV technology, and obtained three heterozygous knock-out clones (DIP2C +/− #1-3) following screening of 605 Geneticin-resistant clones, indicating <1% targeting efficiency. Two homozygous RKO DIP2C knock-out clones (DIP2C −/− #1-1 and #1-2) were confirmed following additional rounds of targeting in DIP2C +/− #1. Approximately four-fold and two-fold reductions in DIP2C mRNA levels were shown in homozygous and heterozygous RKO-derived knock-out clones, respectively (Fig. 1c). To investigate the effects of DIP2C overexpression in the same cell system we selected for stable integration of Myc-DDK-tagged DIP2C in transfected parental RKO cells and established clones with 3-7 fold overexpression of DIP2C mRNA (Fig. 1d-e). The engineered isogenic cell system and the stable overexpression clones were then used to study the phenotypes associated with DIP2C expression in cancer cells.
RNA expression analysis
To investigate effects on gene expression levels we performed RNA sequencing, obtaining data for the expression of >4500 genes in each isogenic cell line. We identified 402 and 378 genes with more than 4-fold change up or down, respectively, in the DIP2C knockout clones under normal growth conditions (Additional file 2: Table S3). Strikingly, DIP2C +/− clones in many cases exhibited gene expression changes to the same magnitude as DIP2C −/− clones. From the gene lists, twelve of the most up-regulated (RGS4, HGF, IL13RA2, CALB2, CDKN2A and DCDC2) and down-regulated (TNS4, SLC1A3, MAP1B, UCA1, GRPR and DCLK1) genes were assayed by RT-qPCR in independent DIP2C −/− samples, validating gene expression changes consistent with the RNA sequencing data for all tested genes (Fig. 2). Gene set enrichment analysis (GSEA)  indicated function in epithelial to mesenchymal transition (EMT), apoptosis, inflammation and angiogenesis, along with genes regulated by different signalling pathways, such as TNF and KRAS, among the differentially expressed genes (Additional file 2: Table S4). Similarly, functional enrichment analysis using the DAVID bioinformatical resources  showed enrichment of genes involved with cell migration, blood vessel development and cell death (Additional file 2: Table S5). In summary, expression profiling revealed hundreds of genes, many of which implicated in processes linked to cancer, for which expression was altered by DIP2C knockout.
Characterization of cell growth
Analysis of the growth of DIP2C knockout cells showed an approximately 50% decreased ability to form colonies and a 35-50% reduced growth rate for DIP2C −/− cells, whereas heterozygous DIP2C +/− cells grew slightly slower but did not show any statistically significant reduction in colony number and formed macroscopically larger colonies compared to parental RKO (Fig. 3a-b). Overexpression of DIP2C did not affect the growth rate of RKO cells (Fig. 3c). Cell cycle analysis showed an increased proportion of DIP2C −/− cells in G1 phase (Fig. 3d), indicating a possible G1 arrest. The RNA-seq data indicated altered mRNA expression of several known cell cycle regulators, including p21/CDKN1A, which is a known transcriptional target of p53. For the TP53 gene, a minor transcriptional downregulation was observed by RNA-seq (Ave Log2 fold change −0.78 in DIP2C −/− cells), however as p53 protein levels are regulated not only through transcription  we analysed p53 protein levels further. Consistent with transcriptional data, immunoblot for p53 and p21 showed upregulation of p21 whereas the p53 level was slightly decreased in DIP2C −/− cells compared to the parental RKO cells (Fig. 3e-f). Interestingly, the p53 level was unaltered in DIP2C +/− cells compared to parental RKO whereas the p21 level was upregulated almost 2-fold more in DIP2C +/− cells compared to DIP2C −/− cells. These results were consistent over repeated independent experiments. These results suggest that DIP2C activity affects cell growth by modulation of cell cycle regulation in the G1 phase through a p53-independent mechanism.
The cyclin dependent kinase inhibitor 2A (CDKN2A) locus encodes at least three related genes that serve as tumour suppressors, of which p14 ARF and p16 INK4a , which have distinct first exons but share exons two and three translated in different reading frames, are known cell cycle regulators with multiple links to human cancer . To discern if the upregulation of CDKN2A observed by RNA sequencing was due to isoform-specific expression from the CDKN2A locus, we performed qPCR and found p14 ARF and p16 INK4a upregulated to the same extent in DIP2C −/− cells (Fig. 3g). Overexpression of DIP2C in RKO cells did not lead to changes in CDKN2A expression (Fig. 3h). As p16 INK4a is a marker of cellular senescence we then stained cells for β-galactosidase activity at pH 6, which is another characteristic of senescent cells, detecting staining of up to 1.9% of DIP2C −/− cells, compared to 0.2% of parental RKO cells (Fig. 3i). Importantly, complete growth cessation was not detected for any engineered cell line during culturing for more than 2 months (not shown), suggesting that cellular senescence is a non-permanent state in these cells, or that only a subpopulation of cells express the markers. Cell enlargement is another sign of senescence and imaging of live cells revealed that DIP2C knock-out cells appeared stretched-out relative to parental RKO cells (Fig. 3j). Further examination of cell size in suspension revealed a 15.9% (SD = 8.9) increase in cell volume for knockout clones (P = 0.003, two-tailed Student’s t-test). Also DIP2C +/− cells were affected, displaying an intermediate morphological phenotype in relation to parental RKO and DIP2C −/− cells. In summary, these data show that altered levels of DIP2C triggers a senescence response in human RKO CRC cells.
Analysis of cell migration and EMT markers
Functions associated with migratory capacity were enriched among the differentially expressed genes in DIP2C knockout cells. To investigate association to EMT, a process believed to influence metastasis by modulating cell motility [37, 38], we used RT-qPCR to analyze expression of Zinc finger E-box binding homeobox 1 (ZEB1) and Vimentin (VIM), which both were found ~4-fold upregulated by RNA sequencing. We validated transcriptional upregulation of the EMT regulator ZEB1 in DIP2C −/− cells but could not validate upregulation of the EMT marker VIM (Fig 4a). We also investigated the frequently used EMT markers E-cadherin/cadherin 1 (CDH1) and N-cadherin/cadherin 2 (CDH2), detecting low or no expression in all cell clones, consistent with not being detected as expressed by RNA sequencing as well as previous reports of CDH1 not being expressed in RKO cells . Immunoblotting further validated the upregulation of ZEB1 in DIP2C KO clones (Fig. 4b). Epithelial-mesenchymal transition is associated with CD44 high/CD24 low-expressing breast cancer stem and stem-like cells [37, 38], and CD44 was part of the GSEA Hallmarks EMT gene set found enriched among the differentially expressed genes in DIP2C −/− cells. By RT-qPCR we detected CD44 and CD24 to be more than 9-fold up- and 2.5-fold downregulated, respectively, in DIP2C −/− cells, validating the RNA sequencing data for these genes (Fig. 4c). We then functionally assessed the cells ability for migration by a scratch assay and observed a 30% higher capacity to close wounds for DIP2C −/− cells (Fig. 4d). This demonstrates that DIP2C activity affects traits and gene expression associated with transition to a more mesenchymal state in RKO cells.
DNA methylation analysis
The DIP2 family proteins contain an N-terminal binding domain for DMAP1, a protein that participates in global maintenance DNA methylation by interaction with DNA methyltransferase 1 (DNMT1) [40, 41], and a SNP in DIP2B recently provided association to the DNA methylation process by being linked to DNA methylation variation in HapMap cell lines . We therefore investigated DNA methylation in the DIP2C knockouts by array hybridization, obtaining beta values reflecting the methylation level in the range of 0-1 (representing 0-100% methylation) at 482548 genomic cytosine positions. We identified 33,700 differentially methylated CpG sites, of which 62% were hypomethylated (methylation level in KO cells was decreased ≥0.3 units), and 38% were hypermethylated (methylation level in KO cells was increased ≥0.3 units). While hypomethylated sites were more common in isolated CpGs across the genome (49% of hypo- and 40% of hypermethylated sites), and in intergenic genomic regions (31% hypo and 26% hyper), hypermethylation events were particularly elevated in CpG islands (22% hyper and 12% hypo), and more frequently associated with proximal promoter regions (45% hyper and 34% hypo) (Fig. 5a-b). Calculation of median methylation levels within the proximal promoter regions and bodies of genes revealed differential methylation of promoter regions in 520 genes and of gene bodies in 580 genes (Additional file 2: Tables S6-S7). Investigating the correlation between DNA methylation and transcript abundance, we observed a weak negative correlation between DNA methylation change at promoter regions and change in RNA expression level, which grew to strong (Pearson’s r = −0.66, P = 2.196 × 10−10) when considering only those genes that exhibited differential DNA methylation (Fig. 5c-d, Additional file 1: Table S8). Conversely, no correlation to gene expression level was observed for changes in gene body methylation. With the exception of one gene (GRPR), all the 28 genes with differential promoter methylation and ≥4-fold differential expression showed correlation between DNA methylation and gene expression changes in DIP2C −/− cells (Additional file 1: Table S9). Gene set enrichment analysis for the genes with differential promoter methylation revealed association with immune response function such as inflammation and coagulation, EMT and apoptosis (Additional file 2: Table S10). In summary these results show that there is correlation between altered promoter DNA methylation and differential gene expression in DIP2C knockout cells. Further, there are similarities in functions enriched among differentially methylated genes and genes exhibiting altered expression which support the hypothesis that DIP2C knockout alters gene expression partly through affected DNA methylation patterns.
Genomic profiling has revealed a large subset of genes as likely drivers of breast tumorigenesis, including the hitherto non-characterized DIP2C gene [2,3,4] which is interesting in association to cancer development since it may belong to the growing number of epigenetic regulators implicated in cancer. Here we exploited rAAV-mediated gene targeting to knock out one or two alleles of DIP2C in human cancer cells, enabling DIP2C to be studied under control of its endogenous promoter. Gene editing was attempted in two cell lines well established for use with the rAAV-targeting method, with the low targeting efficiency (no viable clones for untransformed mammary MCF10a cells and <1% targeted RKO cells) potentially owing to the significant changes in growth and transcription depending on loss of DIP2C that were observed in targeted RKO cells. This could mean that inactivating DIP2C mutations are dependent on additional genetic perturbation for cells to be able to survive their introduction. In support of this theory, DIP2C mutation was reported as a late event when the timing of mutations and chromosome rearrangements were investigated in a breast cancer genome . If this is a general observation in affected tumours has however not been studied.
Even though the DIP2C mRNA level remained at ~50% of that of parental RKO, heterozygous DIP2C +/− cells also to some degree adopted the stretched morphology and decreased growth rate seen in DIP2C −/− cells, which suggests possible haploinsufficiency or a dominant negative effect of the damaged allele. The potentially functional DMAP1 binding domain spans amino acids 9-119 of DIP2C, encompassing sequence from exons 1-4. In the knockout cells a large part of DIP2C exon 9 was deleted, possibly generating a truncated protein with a preserved DMAP1 binding domain which could cause binding of non-functional DIP2C and blocking of normal DIP2C function in heterozygous clones. The predicted inactivating missense and frameshift mutations in breast cancer that motivated this study and the mutations later identified in lung cancer are all heterozygous and localized downstream of the DMAP1 binding domain  (Fig. 1a), suggesting that such a mechanism could be active also in patient tumours. Furthermore this observation is interesting as decreased expression levels of both DIP2C and DIP2B are associated with mental retardation [6, 12].
Disruption of DNA methylation patterns is a hallmark of cancer, and both promoter hypermethylation and global loss of DNA methylation is observed in cancers . In DIP2C knockout cells hypermethylation was the dominating effect at CpG islands and in sites located closely upstream of transcription start sites, which agrees with the DNA methylation pattern associated with gene silencing [44, 45]. Typically heavily methylated in normal tissue , isolated CpGs in the genome were instead preferentially hypomethylated. Differential gene promoter methylation was shown correlated to changes in gene expression, particularly for those genes with more than fourfold differential expression, suggesting that DIP2C KO methylation defects directly influence gene expression. In contrast, as expected [44, 45], gene expression was not correlated to differential methylation at gene bodies. Although a physical interaction has not been demonstrated between DIP2C and DMAP1, for which DIP2C has a putative binding site, based on these results we cannot rule out that DIP2C is involved in the regulation of DNA methylation through this pathway or by another mechanism. Although important in cancer development , DNMT regulation during tumorigenesis is poorly understood. Both DNMT1 and DMAP1 have multiple known interaction partners [41, 47, 48], and DMAP1 has suggested roles not only in DNA methylation but also in histone acetylation and DNA repair [47, 49], suggesting areas of investigation for future studies of DIP2C function.
In DIP2C knockout cells, 10-15% of the functionally assigned differentially expressed genes were involved with regulation of cell death processes (Additional file 2: Table S5). The DIP2C knockout cells did not show signs of apoptosis but displayed multiple markers for cellular senescence, a mechanism activated in ageing cells or by different forms of cellular stress, such as oncogenic signalling, as protection against inappropriate growth signals . Seemingly contradictory, senescent cells can secrete factors that e.g. promote EMT and inflammation, which could stimulate tumorigenic processes . Functional consequences of cellular senescence induction in DIP2C −/− cells cannot be determined from this cell system, but interestingly EMT and inflammation were among processes enriched for in the differentially expressed and/or methylated genes. Notably, DIP2C is among seven genes on chromosome 10p14-15 whose loss has been associated with the ability to escape from senescence in cervical cancer . Such ability is suggested to be an important mechanism in the progression from pre-malignant to malignant cells . Overexpression of DIP2C in CRC cells did not induce senescence markers in the present study, which is consistent with literature on overexpression in primary human fibroblasts and keratinocytes .
Epithelial-mesenchymal transition is a reversible spectrum of transitory cell states where cells express different levels of epithelial and mesenchymal markers [37, 53]. The RKO cell line has increased mesenchymal characteristics compared to several other colorectal cancer cell lines, with low expression of cytoskeletal structure and cell adhesion proteins and high migration and invasive capability . The EMT inducer ZEB1 is previously reported expressed in RKO cells , meaning that loss of DIP2C may augment the RKO EMT phenotype by further ZEB1 upregulation as suggested by the data presented herein. Furthermore, high CD44 and low CD24 expression, characteristics associated with the breast cancer stem cell phenotype and the EMT state, was revealed in DIP2C KO cells, with potential implications for treatment and metastasis ability [37, 38]. The epithelial and mesenchymal states impact the stages of tumorigenesis differently [37, 53], suggesting that timing may influence the effect of DIP2C mutations on tumour development. Here, DIP2C KO caused increased migration in the scratch assay, suggesting possible impact on e.g. the ability to metastasize.
Functional studies of the genes that are altered in cancer will increase the understanding of the changes that are induced as normal cells transform into cancer cells. In this project significant phenotypic and transcriptional alterations were induced by loss of the candidate breast and lung cancer gene DIP2C in cancer cells. Transcriptional changes were correlated to altered DNA methylation, suggesting that DIP2C activity has a role in regulation of this process. Results from functional assays indicate that inactivating DIP2C mutations may function to promote metastasis through EMT induction, but DIP2C knockout also triggered a senescence response in the cells that could either stimulate or inhibit tumorigenesis depending on the context. The function of DIP2C in normal cells still remains to be determined, but based on these results we cannot rule out an association to DNA methylation processes through the predicted N-terminal DMAP1 binding domain.
Cyclin dependent kinase inhibitor 2A
Disco-interacting protein 2 homolog C
DNA methyltransferase 1 associated protein 1
DNA methyltransferase 1, EMT, epithelial-mesenchymal transition
Gene set enrichment analysis
Inverted terminal repeat
Recombinant adeno-associated virus
Zinc finger E-box binding homeobox 1
IST Online, Medisapiens Inc., http://ist.medisapiens.com. Accessed 23 Feb 2016.
Sjöblom T, Jones S, Wood LD, Parsons DW, Lin J, Barber TD, Mandelker D, Leary RJ, Ptak J, Silliman N, et al. The consensus coding sequences of human breast and colorectal cancers. Science. 2006;314:268–74.
Wood LD, Parsons DW, Jones S, Lin J, Sjoblom T, Leary RJ, Shen D, Boca SM, Barber T, Ptak J, et al. The genomic landscapes of human breast and colorectal cancers. Science. 2007;318:1108–13.
Jiao X, Wood LD, Lindman M, Jones S, Buckhaults P, Polyak K, Sukumar S, Carter H, Kim D, Karchin R, Sjoblom T. Somatic mutations in the notch, NF-KB, PIK3CA, and hedgehog pathways in human breast cancers. Genes Chromosomes Cancer. 2012;51:480–9.
Rudin CM, Durinck S, Stawiski EW, Poirier JT, Modrusan Z, Shames DS, Bergbower EA, Guan Y, Shin J, Guillory J, et al. Comprehensive genomic analysis identifies SOX2 as a frequently amplified gene in small-cell lung cancer. Nat Genet. 2012;44:1111–6.
Winnepenninckx B, Debacker K, Ramsay J, Smeets D, Smits A, FitzPatrick DR, Kooy RF. CGG-repeat expansion in the DIP2B gene is associated with the fragile site FRA12A on chromosome 12q13.1. Am J Hum Genet. 2007;80:221–31.
Ouchi N, Asaumi Y, Ohashi K, Higuchi A, Sono-Romanelli S, Oshima Y, Walsh K. DIP2A functions as a FSTL1 receptor. J Biol Chem. 2010;285:7127–34.
Tanaka M, Murakami K, Ozaki S, Imura Y, Tong XP, Watanabe T, Sawaki T, Kawanami T, Kawabata D, Fujii T, et al. DIP2 disco-interacting protein 2 homolog a (drosophila) is a candidate receptor for follistatin-related protein/follistatin-like 1--analysis of their binding with TGF-beta superfamily proteins. FEBS J. 2010;277:4278–89.
Mukhopadhyay M, Pelka P, DeSousa D, Kablar B, Schindler A, Rudnicki MA, Campos AR. Cloning, genomic organization and expression pattern of a novel drosophila gene, the disco-interacting protein 2 (dip2), and its murine homolog. Gene. 2002;293:59–65.
Egger G, Roetzer KM, Noor A, Lionel AC, Mahmood H, Schwarzbraun T, Boright O, Mikhailov A, Marshall CR, Windpassinger C, et al. Identification of risk genes for autism spectrum disorder through copy number variation analysis in Austrian families. Neurogenetics. 2014;15:117–27.
Poelmans G, Engelen JJ, Van Lent-Albrechts J, Smeets HJ, Schoenmakers E, Franke B, Buitelaar JK, Wuisman-Frerker M, Erens W, Steyaert J, Schrander-Stumpel C. Identification of novel dyslexia candidate genes through the analysis of a chromosomal deletion. Am J Med Genet B Neuropsychiatr Genet. 2009;150B:140–7.
Descipio C, Conlin L, Rosenfeld J, Tepperberg J, Pasion R, Patel A, McDonald MT, Aradhya S, Ho D, Goldstein J, et al. Subtelomeric deletion of chromosome 10p15.3: clinical findings and molecular cytogenetic characterization. Am J Med Genet A. 2012;158A:2152–61.
Closa A, Cordero D, Sanz-Pamplona R, Sole X, Crous-Bou M, Pare-Brunet L, Berenguer A, Guino E, Lopez-Doriga A, Guardiola J, et al. Identification of candidate susceptibility genes for colorectal cancer through eQTL analysis. Carcinogenesis. 2014;35:2039–46.
Stoimenov I, Ali MA, Pandzic T, Sjoblom T. Computational and molecular tools for scalable rAAV-mediated genome editing. Nucleic Acids Res. 2015;43:e30.
Rago C, Vogelstein B, Bunz F. Genetic knockouts and knockins in human somatic cells. Nat Protoc. 2007;2:2734–46.
Topaloglu O, Hurley PJ, Yildirim O, Civin CI, Bunz F. Improved methods for the generation of human gene knockout and knockin cell lines. Nucleic Acids Res. 2005;33:e158.
Parson W, Kirchebner R, Muhlmann R, Renner K, Kofler A, Schmidt S, Kofler R. Cancer cell line identification by short tandem repeat profiling: power and limitations. FASEB J. 2005;19:434–6.
Rasband WS. Image J. U. S. National Institutes of Health, Bethesda, Maryland, USA. http://imagej.nih.gov/ij/, 1997-2016.
The Illumina iGenomes Project. http://support.illumina.com/sequencing/sequencing_software/igenome.ilmn. Accessed Mar 2014.
Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, Salzberg SL. TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol. 2013;14:R36.
Trapnell C, Williams BA, Pertea G, Mortazavi A, Kwan G, van Baren MJ, Salzberg SL, Wold BJ, Pachter L. Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nat Biotechnol. 2010;28:511–5.
Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES, Mesirov JP. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005;102:15545–50.
Liberzon A, Birger C, Thorvaldsdottir H, Ghandi M, Mesirov JP, Tamayo P. The molecular signatures database (MSigDB) hallmark gene set collection. Cell Syst. 2015;1:417–25.
Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4:44–57.
Huang da W, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009;37:1–13.
Edgar R, Domrachev M, Lash AE. Gene expression omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 2002;30:207–10.
Bakshi S, Zhang X, Godoy-Tundidor S, Cheng RY, Sartor MA, Medvedovic M, Ho SM. Transcriptome analyses in normal prostate epithelial cells exposed to low-dose cadmium: oncogenic and immunomodulations involving the action of tumor necrosis factor. Environ Health Perspect. 2008;116:769–76.
Kanellou P, Zaravinos A, Zioga M, Stratigos A, Baritaki S, Soufla G, Zoras O, Spandidos DA. Genomic instability, mutations and expression analysis of the tumour suppressor genes p14(ARF), p15(INK4b), p16(INK4a) and p53 in actinic keratosis. Cancer Lett. 2008;264:145–61.
Xie F, Xiao P, Chen D, Xu L, Zhang B. miRDeepFinder: a miRNA analysis tool for deep sequencing of plant small RNAs. Plant Mol Biol. 2012;80(1):75–84.
Triche TJ Jr, Weisenberger DJ, Van Den Berg D, Laird PW, Siegmund KD. Low-level processing of Illumina Infinium DNA Methylation BeadArrays. Nucleic Acids Res. 2013;41:e90.
R: A Language and Environment for Statistical Computing. R Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2014. https://www.R-project.org.
The Bioconductor Project. www.bioconductor.org. Accessed Nov 2014.
Marabita F, Almgren M, Lindholm ME, Ruhrmann S, Fagerstrom-Billai F, Jagodic M, Sundberg CJ, Ekstrom TJ, Teschendorff AE, Tegner J, Gomez-Cabrero D. An evaluation of analysis pipelines for DNA methylation profiling using the Illumina HumanMethylation450 BeadChip platform. Epigenetics. 2013;8:333–46.
Geback T, Schulz MM, Koumoutsakos P, Detmar M. TScratch: a novel and simple software tool for automated analysis of monolayer wound healing assays. BioTechniques. 2009;46:265–74.
Kumari R, Kohli S, Das S. p53 regulation upon genotoxic stress: intricacies and complexities. Mol Cell Oncol. 2014;1:e969653.
Kim WY, Sharpless NE. The regulation of INK4/ARF in cancer and aging. Cell. 2006;127:265–75.
Polyak K, Weinberg RA. Transitions between epithelial and mesenchymal states: acquisition of malignant and stem cell traits. Nat Rev Cancer. 2009;9:265–73.
May CD, Sphyris N, Evans KW, Werden SJ, Guo W, Mani SA. Epithelial-mesenchymal transition and cancer stem cells: a dangerously dynamic duo in breast cancer progression. Breast Cancer Res. 2011;13:202.
Buck E, Eyzaguirre A, Barr S, Thompson S, Sennello R, Young D, Iwata KK, Gibson NW, Cagnoni P, Haley JD. Loss of homotypic cell adhesion by epithelial-mesenchymal transition or mutation limits sensitivity to epidermal growth factor receptor inhibition. Mol Cancer Ther. 2007;6:532–41.
Lee GE, Kim JH, Taylor M, Muller MT. DNA methyltransferase 1-associated protein (DMAP1) is a co-repressor that stimulates DNA methylation globally and locally at sites of double strand break repair. J Biol Chem. 2010;285:37630–40.
Rountree MR, Bachman KE, Baylin SB. DNMT1 binds HDAC2 and a new co-repressor, DMAP1, to form a complex at replication foci. Nat Genet. 2000;25:269–77.
Bell JT, Pai AA, Pickrell JK, Gaffney DJ, Pique-Regi R, Degner JF, Gilad Y, Pritchard JK. DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biol. 2011;12:R10.
Newman S, Howarth KD, Greenman CD, Bignell GR, Tavare S, Edwards PA. The relative timing of mutations in a breast cancer genome. PLoS One. 2013;8:e64991.
Kulis M, Queiros AC, Beekman R, Martin-Subero JI. Intragenic DNA methylation in transcriptional regulation, normal differentiation and cancer. Biochim Biophys Acta. 1829;2013:1161–74.
Brenet F, Moh M, Funk P, Feierstein E, Viale AJ, Socci ND, Scandura JM. DNA methylation of the first exon is tightly linked to transcriptional silencing. PLoS One. 2011;6:e14524.
Girault I, Tozlu S, Lidereau R, Bieche I. Expression analysis of DNA methyltransferases 1, 3A, and 3B in sporadic breast carcinomas. Clin Cancer Res. 2003;9:4415–22.
Cai Y, Jin J, Tomomori-Sato C, Sato S, Sorokina I, Parmely TJ, Conaway RC, Conaway JW. Identification of new subunits of the multiprotein mammalian TRRAP/TIP60-containing histone acetyltransferase complex. J Biol Chem. 2003;278:42733–6.
Qin W, Leonhardt H, Pichler G. Regulation of DNA methyltransferase 1 by interactions and modifications. Nucleus. 2011;2:392–402.
Negishi M, Chiba T, Saraya A, Miyagi S, Iwama A. Dmap1 plays an essential role in the maintenance of genome integrity through the DNA repair process. Genes Cells. 2009;14:1347–57.
Romagosa C, Simonetti S, Lopez-Vicente L, Mazo A, Lleonart ME, Castellvi J, Ramon y Cajal S. p16(Ink4a) overexpression in cancer: a tumor suppressor gene associated with senescence and high-grade tumors. Oncogene. 2011;30:2087–97.
Campisi J. Aging, cellular senescence, and cancer. Annu Rev Physiol. 2013;75:685–705.
Liesenfeld M, Mosig S, Funke H, Jansen L, Runnebaum IB, Durst M, Backsch C. SORBS2 and TLR3 induce premature senescence in primary human fibroblasts and keratinocytes. BMC Cancer. 2013;13:507.
Tam WL, Weinberg RA. The epigenetics of epithelial-mesenchymal plasticity in cancer. Nat Med. 2013;19:1438–49.
Halvey PJ, Wang X, Wang J, Bhat AA, Dhawan P, Li M, Zhang B, Liebler DC, Slebos RJ. Proteogenomic analysis reveals unanticipated adaptations of colorectal tumor cells to deficiencies in DNA mismatch repair. Cancer Res. 2014;74:387–97.
NCBI Gene Expression Omnibus, https://www.ncbi.nlm.nih.gov/geo/. Accession number GSE80746.
NCBI Gene Expression Omnibus, https://www.ncbi.nlm.nih.gov/geo/. Accession number GSE86402.
We thank Tanzila Mahzabin and Maria Karoutsou for technical assistance. Methylation profiling was performed by the SNP&SEQ Technology Platform in Uppsala. The facility is part of the NGI Sweden and Science for Life Laboratory. The SNP&SEQ platform is also supported by the Swedish Research Council and the Knut and Alice Wallenberg Foundation. Array data was analysed by the Uppsala Array and Analysis facility.
Availability of data and materials
The RNA sequencing dataset generated and analysed during this study is available in the NCBI GEO data repository  with accession number GSE8074654 . The DNA methylation array dataset generated and analysed during the current study is available in the NCBI GEO data repository with accession number GSE86402 . All additional data generated and/or analysed during this study are included in this published article and its additional files.
This work was supported by Research grants 2006/2154, 2007/775, and 2012/834 from the Swedish Cancer Society, and Research grants F06-0050 and RBa08-0114 from the Swedish Foundation for Strategic Research to TS, and Postdoctoral grant 2012/1235 from the Swedish Cancer Society and a postdoctoral stipend from the Swedish Society for Medical Research to CL. There was no role of the funding bodies in the design of the study, in collection, analysis, and interpretation of data or in writing the manuscript.
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interest with the contents of this article.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Additional file 1: Table S1.
- Primers for generation and validation of isogenic DIP2C cell lines and DIP2C overexpressing cells. Primers are indicated by their number in the text. Primers were purchased from Sigma Aldrich. Table S2. - Primers for RT-qPCR. Primer pairs for amplification of the transcript indicated by the respective name. Primers were purchased from Sigma Aldrich. Table S8. - Summary of DNA methylation analysis in DIP2C −/− cells. Table S9. - Genes with change in methylation (beta diff ≥│0.3│) at promoter sites and gene expression (log2 fold change ≥│2│) in DIP2C −/− #1-1 compared to RKO. (PDF 199 kb)
Additional file 2: Table S3.
- Genes with ≥4 fold change in expression in DIP2C knockout cells. Table S4. - Gene set overlap results for differentially expressed genes (≥4-fold change up or down) investigated with GSEA MSigDB Hallmarks gene set. Table S5. - Functional annotation chart for differentially expressed genes (>4-fold change up or down) investigated with the DAVID Functional annotation tool using the GO Biological process (GO_BP_FAT) annotation category. Table S6. - Genes with ≥│0.3│ units change in median promoter DNA methylation in DIP2C knockout cells. Table S7. - Genes with ≥│0.3│ units change in median gene body DNA methylation in DIP2C knockout cells. Table S10. - Gene set overlap results for promoter differentially methylated genes (≥│0.3│ change in methylation level at promoter sites in DIP2C−/− #1-1) investigated with GSEA MSigDB Hallmarks gene set. (XLSX 152 kb)
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
About this article
Cite this article
Larsson, C., Ali, M.A., Pandzic, T. et al. Loss of DIP2C in RKO cells stimulates changes in DNA methylation and epithelial-mesenchymal transition. BMC Cancer 17, 487 (2017). https://doi.org/10.1186/s12885-017-3472-5
- Gene knockout
- rAAV-mediated gene targeting
- Tumour cell biology
- DNA methylation
- Epithelial-mesenchymal transition (EMT)