Coiled-coil domain containing 68 (CCDC68) demonstrates a tumor-suppressive role in pancreatic ductal adenocarcinoma.

Using integrative genomics and functional screening, we identified coiled-coil domain containing 68 (CCDC68) as a novel putative tumor suppressor gene (TSG) in pancreatic ductal adenocarcinoma (PDAC). CCDC68 allelic losses were documented in 48% of primary PDAC patient tumors, 50% of PDAC cell lines and 30% of primary patient derived xenografts. We also discovered a single nucleotide polymorphism (SNP) variant (SNP rs1344011) that leads to exon skipping and generation of an unstable protein isoform CCDC68Δ(69-114) in 31% of PDAC patients. Overexpression of full length CCDC68 (CCDC68(wt)) in PANC-1 and Hs.766T PDAC cell lines lacking CDCC68 expression decreased proliferation and tumorigenicity in scid mice. In contrast, the downregulation of endogenous CCDC68 in MIAPaca-2 cells increased tumor growth rate. These effects were not observed with the deletion-containing isoform, CCDC68Δ(69-114).

Introduction

Pancreatic adenocarcinoma (PDAC) is the fourth leading cause of cancer-related death in North America, with a poor overall 5-year survival rate of only 5% (1). To date, no chemotherapeutic treatments have been found to be effective against this lethal disease.

Aside from a small percentage of familial cases, PDAC is driven by the accumulation of somatic alterations. Both the loss of tumor suppressor genes (TSGs) and activation of oncogenes are involved in pancreatic carcinogenesis. Key events in pancreatic carcinogenesis that have been validated as high-frequency alterations include KRAS activating mutations and inactivation of CDKN2A, SMAD4 and p53 TSGs (2). However, recent genome-wide surveys have demonstrated significant genetic heterogeneity among PDAC patients, with the occurrences of rare somatic mutations in many genes (3–5). Although many of the catalogued genetic alterations can be linked to one of 12-core cancer signaling pathways, experimental evidence to verify these novel alterations as either drivers of pancreatic carcinogenesis or passengers is still lacking. In this context, we have combined integrative analysis of genome/transcriptome data with functional shRNA and ORFeome screens to identify CCDC68 as a putative TSG in PDAC.

Results

Integrative genomic analysis reveals CCDC68 as a putative TSG in PDAC

We have previously shown that the immortalized near-normal human pancreatic duct epithelial (HPDE) cell line H6c7, when transformed by KRASG12V oncogene (H6c7-Kr), gives rise to sporadic tumors when implanted into scid mice (6). However, H6c7-KrasT cell line re-derived from one of these tumors, could now produce tumors in scid mice with 100% efficiency. We hypothesized that additional genetic alterations arose in the H6c7-KrasT cell line, which synergized with KRASG12V oncogene to cause the full malignant transformation of H6c7 cells into invasive carcinoma. To identify genetic alterations in H6c7-KrasT cells, whole-genome tiling path array comparative genomic hybridization (aCGH) was used to compare DNA copy number between H6c7-KrasT and H6c7-Kr cells. To select for true events, only alterations with the same status in both replicates and encompassing two adjacent clones were considered. Alterations in 11 genomic regions on 9 chromosome arms were identified (Supplementary Fig. S1). The main regions included losses on chromosome 8, 15q and 18q. These regions mapped to 221 genes with gain and 2342 genes with loss.

To identify the most probable TSGs involved in the transformation of H6c7-Kr into H6c7-KrasT cells we utilized integrative genomic analysis (Fig. 1A). Firstly, aCGH data was integrated with aCGH and transcriptome data from 20 established PDAC cell lines (7). The aCGH data were segmented to determine the copy number status (gain, loss or neutral) for each gene in each cell line. Then, the expression of each gene was compared between the altered and neutral samples using the Mann-Whitney U test. Genes whose expression was in the direction predicted by the copy number change and had a p-value<0.05 were considered significant. This analysis identified 206 genes lost and underexpressed in 20 PDAC lines relative to H6c7 (Supplementary TableS1). Further analysis revealed 5/206 genes (CCDC68, ARHGEF10, POLI, ME2 and CLN8) with mutation or homozygous deletion in primary PDAC patients (3) and decreased mRNA expression in at least 30% of PDAC cell lines as compared to H6c7 cells (Fig. 1A). One of the identified genes, coiled-coil domain containing 68 (CCDC68) was further investigated as a putative TSG in PDAC (bottom panel: Fig. 1A).

Figure 1

In vitro proliferation screen identifies CCDC68 as a putative TSG in PDAC

A. Flow diagram of the integrative genomic analysis utilized for the identification of putative tumor suppressor genes in PDAC. Bottom panel is the HeatMapViewer displaying average CCDC68 mRNA expression values across PDAC cell lines relative to H6c7 as measured by three independent RT-qPCR experiments. The largest expression values are shown as red bars and the smallest values are displayed as blue bars. B. The changes in cell proliferation as a result of CCDC68 overexpression or knockdown were measured 72 hour post transfection using MTS proliferation assay. The bars represent mean±SD of four replicates. p<0.05; One-way ANOVA, Bonferroni. C. Stable knockdown of CCDC68 in MIAPaca-2 cells significantly increased their growth. D. Stable overexpression of CCDC68 in PANC-1 cell line significantly decreased their growth as measured by XCELLigence. The data points (C, D) represent mean±SD for two independent experiments. p<0.0001; ANOVA-mixed model, Bonferroni. E. Graph enumerating soft agar colony formation of PANC-1 EV and PANC-CCDC68 cells. Data denotes mean±SD of three independent experiments. p<0.0001 Student t test.

CCDC68 expression negatively affects the growth of PDAC cell lines

We assessed the proliferative effects of CCDC68 knockdown (H6c7-Kr and MIAPaca-2) and CCDC68 open reading frame (ORF) overexpression (PANC-1 cells) in PDAC cell lines using the MTS assay. While decreased CCDC68 levels significantly increased proliferation of H6c7-Kr and MIAPaca-2 cells, overexpression of CCDC68 significantly decreased proliferation of PANC-1 cells (Fig. 1B). Using the xCELLigence platform, we confirmed the increased growth rate of MIAPaca-2 cells upon CCDC68 knockdown (Fig. 1C) and decreased growth rate of PANC-1 cells expressing CCDC68ORF (Fig. 1D). Furthermore, the exogenous expression of CCDC68 significantly decreased soft agar colony formation of PANC-1 cells (Fig. 1E).

CCDC68 expression decreases tumorigenicity of PDAC cell lines

CCDC68 was overexpressed in H6c7-KrasT, PANC-1 and Hs.766T cell lines containing undetectable endogenous CCDC68 protein (top panel: Fig. 2A,B,C) and the resulting cell lines and those derived from vector-only controls were injected into subcutaneous tissue of scid mice to assess tumor growth rates (Fig. 2A,B,C). Overexpression of CCDC68ORF in H6c7-KrasT cells completely abrogated their ability to grow tumors in mice (Fig. 2A). Furthermore, our results revealed that overexpressing CCDC68 significantly attenuated tumor growth in PANC-1 and Hs.766T xenograft models (Fig. 2B,C). The decreased tumor sizes in PANC-1 and Hs.766T xenograft models overexpressing CCDC68 were also confirmed at necropsy (Supplementary Fig. S2).

Figure 2

CCDC68 mediates tumorigenicity in scid mice

CCDC68 was stably overexpressed in A. H6c7-KrasT, B. PANC-1 A. and C. Hs.766T, and downregulated in D. MIAPaca-2 cell lines. Resulting cell lines and their controls were then injected in scid mice and tumor growth was monitored over time. Data points represent mean±SD of tumor measurements in 5 animals. p<0.0001, Mixed-model ANOVA, Bonferroni.

We also downregulated CCDC68 in MIAPaca-2 cells which express CCDC68 endogenously, using two independent shRNAs targeting the 3′UTR region of the CCDC68 mRNA (top panel: Fig. 2C). The resulting cell lines were subjected to in vivo tumorigenicity assays. Downregulation of wild type CCDC68 by two independent shRNAs increased tumor growth rate of MIAPaca-2 cells (Fig. 2C). All MIAPaca-2 shCCDC68 xenografts showed significantly increased tumor volumes and weights compared with the shLuc2 (luciferase control) tumors (Supplementary Fig. S2). Together these results indicate that CCDC68 behaves as a TSG in PDAC xenografts.

CCDC68 protein expression is decreased in PDAC

We performed CCDC68 immunohistochemistry staining on tissue microarrays (TMA) created using 46 primary PDAC xenograft tissue specimens and several normal pancreas specimens. The normal interlobular and small duct epithelium consistently showed moderate cytoplasmic staining, but PDAC xenografts showed variable CCDC68 staining (Fig. 3A). Nearly 60% of PDAC cases showed lost/reduced CCDC68 protein expression as compared to normal ducts (Fig. 3A). CCDC68 was expressed in the supranuclear cytoplasm of normal duct epithelium (Fig. 3B) and well-differentiated PDAC (Fig. 3C), while significant loss of staining was evident in the moderately (Fig. 3D) and poorly differentiated PDAC (Fig. 3E). Some CCDC68 staining was observed in the PDX stroma, however, this staining was not seen in either normal pancreas or human PDAC (Supplementary Fig. S3) and likely represents unspecific binding of CCDC68 antibody to protein expressed by mouse fibroblasts.

Figure 3

CCDC68 protein expression is associated with differentiation status in PDAC

(A) IHC was performed on TMA for 46 PDAC xenografts and normal pancreas using a CCDC68 specific antibody optimized for IHC. The graph summarizes the scoring results of all TMAs as compared to expression in normal pancreatic ducts. p<0.0001, Mann-Whitney U. CCDC68 perinuclear expression was apparent on the apical side of the normal pancreatic ducts (B) and well-differentiated tumors (C). Moderately (D) and poorly (E) differentiated tumors showed random and low expression of CCDC68 protein.

CCDC68 copy number loss occurs frequently in PDAC

To evaluate the role of gene deletion as a mechanism for loss of CCDC68 expression, we assessed the copy number status of CCDC68 in 19 PDAC cell lines and 32 patient-derived xenografts (PDX) models using a combination of qPCR and FISH analysis (Fig. 4A,B). FISH revealed loss of CCDC68 DNA copy in 10/19 (50%) of PDAC cell lines analyzed (Supplementary Fig. S4). Since there was a correlation between qPCR and FISH analysis of CCDC68 DNA copy number changes (Fig. 4A), we next assessed CCDC68 DNA copy status in 32 PDX models using qPCR. DNA copy loss was defined as values below the mean+SD of qPCR derived DNA copy number in PDAC cell lines with two copies of CCDC68 DNA by FISH. We documented CCDC68 copy loss in 30% of PDX models (Supplementary Fig. S4).

Figure 4

CCDC68 copy number status and mRNA expression in PDAC

(A) The loss of CCDC68 DNA copy number was documented in 19 PDAC cell lines using FISH and qPCR analysis. Cell lines were divided into two groups based on the status of CCDC68 copy number as determined by FISH. Images are showing the DNA copy loss of CCDC68 in representative cell lines as revealed by FISH analysis. (B) Correlation analysis of CCDC68 DNA copy number and CCDC68 mRNA expression in 32 primary xenografts and 19 cell lines measured using qPCR. The data represent averages of three independent experiments for each sample. (C) Western blotting using CCDC68 specific antibody confirms the mRNA expression changes in selected PDAC cell lines. (D) Correlation of CCDC68 mRNA levels with CCDC68 protein levels in PDX models. Protein expression was scored by IHC while mRNA expression was measured using qPCR. Average data in A), B) and D) was plotted for each sample and represented by the Tukey boxplot. p value is calculated using Man-Whitney statistical test.

To corroborate our finding of widespread copy number loss of CCDC68 in PDAC, we analyzed copy number data from 125 primary PDAC donors made available by the International Cancer Genome Consortium (ICGC, project code PACA-AU). Nearly 50% of this cohort (60/125) exhibited copy number loss of the CCDC68 gene. This rate of loss corresponded to a rank in the 91st percentile among all genes on chromosome 18, pointing to the specificity of this particular alteration in PDAC (Supplementary Fig. S5).

We next investigated the influence of copy number loss on CCDC68 mRNA expression. mRNA expression was assessed by RT-qPCR covering two CCDC68 exons in 32 primary patient-derived mouse xenograft (PDX) models and 19 PDAC cell lines with RT-qPCR analysis encompassing two CCDC68 exons. There was no significant overall correlation between copy number and mRNA expression of CCDC68 in PDAC xenografts and cell lines (Fig. 4B). However, there was an overall correlation between CCDC68 mRNA and protein expression as determined in several PDAC cell lines (Fig. 4C) and PDX models (Fig. 4D).

Novel CCDC68 alternate splice variant lacking amino acids 69–114

Although CCDC68 copy number loss occurs in 50% of PDAC, only a small subset of those patients actually showed reduced CCDC68 mRNA expression, suggesting additional regulatory mechanisms in PDAC. We hypothesized that mutation of CCDC68 could account for loss of function in PDAC. Sequencing of CCDC68 in 19 PDAC cell lines did not reveal any somatic mutations in CCDC68. However, in the AsPC-1 cell line we discovered a CCDC68 transcript with deletion of exon 5 (Fig. 5A). To determine if this was caused by mutation in the splicing site, we amplified and sequenced the 5′donor splice site and indeed identified the c.620G>A substitution one nucleotide upstream of the exon 5 donor splice (SNP rs1344011) (Fig. 5B). PCR analysis revealed the presence of this SNP in 31% of PDX models (10/32) suggesting that its expression might be relevant to PDAC disease (data not shown). The translation analysis of rs1344011 CCDC68 variant predicted a protein CCDC68Δ69–114 with histidine 69-lysine 114 in-frame deletion. This smaller CCDC68Δ69–114 protein was confirmed by western blotting in AsPC-1 as compared to 293T cells expressing exogenous CCDC68wt and two cell lines, HPAF-II and MIAPaca-2 with high endogenous CCDC68wt expression (Fig. 5C).

Figure 5

Identification of novel CCDC68Δ69–114 splice variant

(A) Amplification of full length CCDC68 ORF by PCR shows truncated PCR fragment in AsPC-1 cells. (B) Chromatogram demonstrating the G-A substitution occurring in exon 5 donor splice site in AsPC-1 cell line DNA. (C) Endogenous protein expression of truncated AsPC-1 CCDC68 protein. 293T protein lysate was isolated 72 hours following transient transfection of 293T cells with either EV or pENTR-CCDC68 vector expressing full length CCDC68wt cDNA. 1=CCDC68, 2=actin loading control. (D) Minigene analysis confirms the exon 5 skipping as a result of SNP r.1344011 variation. Red stars show the position of SNP rs.1344011 in the minigene construct. E) CCDC68Δ69–114 specific mRNA expression was documented in AsPC1 cells using q-RT-PCR as outlined in the Materials and Methods. Data points represent mean±SD from two independent experiments. Variable ratios of CCDC68Δ69–114/CCDC68wt mRNA (F) and protein (G) levels were documented in PDX models. Data points in F) represent means of three independent experiments.

NetGene2 splice site prediction software revealed that G/A substitution in SNP rs1344011 decreases the strength of the donor splice site, suggesting that exon skipping may result from this substitution (http://www.cbs.dtu.dk/services/NetGene2/). To confirm this hypothesis, we subcloned minigene cassettes containing Exon5-Intron-Exon6 with (SNP) and without (wt) the rs1344011 substitution into the pET01 vector and overexpressed them in 293T and NIH3T3 cells. While cells transfected with either wt or SNP minigene Exon5-Intron-Exon6 cassette expressed the fused exon5-exon6 transcript, SNP minigene expressing cells also expressed a single exon6 transcript demonstrating the skipping of exon5 resulting from rs1344011 (Fig. 5D).

Our data indicates that while SNP rs1344011-containing AsPC-1 cells contain a single copy of CCDC68 gene, both CCDC68Δ69–114 and CCDC68wt are expressed in these cells. This suggests that the rs1344011 splice donor substitution does not result in the complete skipping of exon 5 in CCDC68 transcript. We hence sought to determine the ratio of the two CCDC68 splice variants in PDAC cell lines and PDX models. First, H6c7 and AsPC-1 cells were subjected to RT-qPCR analysis using CCDC68Δ69–114/CCDC68wt specific primers (Fig. 5E), revealing that the rs13440011 variant accounts for ~40% of CCDC68 transcripts in AsPC-1. We then tested 6 PDX models with detectable CCDC68Δ69–114 mRNA expression and documented variable transcript ratios of CCDC68Δ69–114/CCDC68wt (Fig. 5F). Although variable ratios of CCDC68Δ69–114/CCDC68wt protein levels were also apparent in 6 PDX models (Fig. 5G) the levels did not correlate well with the CCDC68Δ69–114/CCDC68wt transcript levels suggesting that the levels of deleted and wild type CCDC68 are regulated at the posttranscriptional level.

To address this discrepancy, we compared the protein stability of CCDC68wt and CCDC68Δ69–114 using a protein half-life assay in the presence of the protein translational inhibitor cyclohexamide. CCDC68Δ69–114 showed significantly decreased half-life compared to CCDC68wt, suggesting that loss of amino acids 69–114 impairs the protein stability of CCDC68 (Supplementary Fig. S6).

CCDC68 splice variant is non-functional as a tumor suppressor

We next investigated the “tumor suppressive” role of CCDC68Δ69–114 variant in PDAC by overexpressing CCDC68wt and CCDC68Δ69–114 in PANC-1 cells and examined the effects on cellular proliferation (Supplementary Fig. S7), as well as subcutaneous and orthotopic tumor growth in scid mice (Fig. 6A,B). While overexpression of CCDC68 significantly decreased in vivo and in vitro growth of PANC-1 cells compared to the empty vector controls, this tumor suppressive effect was absent in PANC-1 cells expressing the truncated CCDC68Δ69–114 (Fig. 6). These findings support the conclusion that the tumor suppressive properties of CCDC68 are specific to the full-length isoform.

Figure 6

Loss of amino acids 69-114 decreases the tumor suppressive function of CCDC68

PANC-1 cells expressing EV, CCDC68wt or CCDC68Δ69–114 were implanted (A) subcutaneously or (B) orthotopically into the pancreas of scid mice. (A) Data points represent mean±SE of tumor measurements in 5 animals over time. p<0.05; Mixed-model ANOVA, Bonferroni (B) Pancreatic tumors were removed 30 days post implantation and the data points represent mean+ SE of final tumor volumes in 5 mice. p<0.05;Two-way ANOVA; Bonferroni).

Discussion

Here we report the identification and demonstration of the tumor suppressive role of CCDC68 in PDAC. We initially identified CCDC68 in a screen for novel TSGs synergizing with KRAS oncogene to drive malignant transformation of a near-normal HPDE cell line and its tumor suppressive activity was confirmed in several PDAC cell line models. We showed that CCDC68 loss of function occurs through copy number loss and the expression of an unstable protein isoform, CCDC68Δ69–114, which lack a tumor suppressive function.

Our initial analysis revealed CCDC68 as a putative TSG whose loss of function enhanced tumor formation of H6c7-Kras cells in scid mice. Incomplete tumor penetrance of our in vitro PDAC H6c7 model (partly transformed by the KRAS oncogene) in scid mice prompted us to investigate the existence of additional genes that enhance the penetrance of this model as a result of acquired genomic alterations (6). Thus, we compared genomic profiles of partially penetrant H6c7-Kr cells with those of completely penetrant H6c7-KrasT cells to identify such candidates. aCGH analysis revealed prominent copy number losses on chromosomes 8p, 15p and 18q in H6c7-KrasT cell lines with 100% tumor penetrance. Integration of this genomic data with transcriptome data collected from both PDAC cell lines and patient samples identified CCDC68 as the most probable candidate for enhancing the tumorigenesis of H6c7-Kr cells. This hypothesis was further strengthened by in vitro shRNA and cDNA screen that revealed CCDC68 as a negative regulator of cell proliferation in H6c7-Kr, PANC-1 and MIAPaca-2 cell lines. This prompted us to continue investigation of CCDC68 as a novel TSG in PDAC.

CCDC68 is located on the 18q chromosome arm frequently lost in PDAC. SMAD4 has been recognized as a TSG in this region and is inactivated in 50% of PDAC by homozygous deletion or mutation. However, several studies have reported homozygous deletions (HDs) and loss of heterozygosity (LOH) in genes telomeric of the 18q21.1 locus (8–13). These include ME2, ELAC1 and MEX3C on 18q21.1 and DCC, SNORA37 and MBD2 on 18q21.2. More recently, SNP analysis and exome sequencing has identified LOH and copy neutral LOH (CN-LOH) at the 18q resident genes POLI and CCDC68 (3,13). Due to the high frequency of genetic alterations affecting genes located downstream of SMAD4, it is likely that other TSGs on 18q in PDAC remain to be identified. This hypothesis is strengthened by the observation that introduction of an additional copy of chromosome 18 into cultured PDAC cells decreases tumorigenic potential of these cells both in vitro and in vivo independently of SMAD4 inactivation (14).

Our results indicated that 60% of PDAC patients have decreased CCDC68 protein expression levels suggesting that the expression of CCDC68 has negative effect on PDAC tumor biology. We also observed that CCDC68 expression associates with well-differentiated tumors. However, since our cell line xenografts revealed no significant effect of CCDC68 on overall differentiation it is not likely that CCDC68 itself affects the differentiation status of PDAC tumors.

First line of evidence suggesting that CCDC68 might be a tumor suppressor was reported in colorectal adenocarcinoma where correlative copy number loss and CCDC68 underexpression was observed in majority of CRC patients (15). While, we documented copy loss of CCDC68 in half of the PDAC cell lines and PDAC patients, no significant correlation between CCDC68 copy number and mRNA expression was observed. Mismatch between mRNA and copy number variation has been reported in cancer (16,17) and it reflects the numerous transcriptional regulatory mechanisms, including epigenetic and/or micro-RNA silencing. For CCDC68, this requires further investigations.

An additional novel and significant finding of our study is the identification of a novel CCDC68 splice variant devoid of tumor suppressive function in PDAC. The truncated in-frame protein CCDC68Δ69–114 is a result of exon skipping in patients harboring a donor splice site variant SNP rs1344011. Specifically, we determined that: 1) wild type and splice variant transcripts are expressed in all SNP containing cell lines and tissues, independent of copy number; 2) PDAC patients carrying SNP rs1344011 exhibit a variable variant/wild type mRNA expression CCDC68 ratio; 3) The CCDC68wt/CCDC68Δ69–114 ratio appear to be regulated at the posttranscriptional level and 4) the protein isoform resulting from SNP rs1344011 has diminished tumor suppressive ability in PDAC cell lines. Disregulated alternative splicing plays a pivotal role in carcinogenesis. In the case of TSGs, induced overexpression of antagonistic variants in cancer is often phenotypically equivalent to loss of function and this has been shown for several tumor suppressors including PTEN, BRCA1 and TP53 (18–20). We demonstrated the absence of tumor suppressive activity of the CCDC68 variant in PANC-1 cells, further suggesting that CCDC68Δ69–114 may functionally oppose CCDC68wt. It remains to be investigated whether the CCDC68 tumor suppressive function is regulated by a critical balance between wt and truncated variant CCDC68 expression in patients carrying the rs1344011 SNP.

There was a mismatch between transcript and protein CCDC68Δ69–114/CCDC68wt ratios in tested PDX models suggestive of posttranscriptional regulation of CCDC68 variant levels. Several cellular mechanisms are in place to ensure the critical balance of particular proteins required for normal function including regulation of translation efficiency and protein turnover. In cancer including PDAC, this regulatory machinery is perturbed and would result in variant protein imbalances. Comparative experiments to investigate the protein stability CCDC68wt and CCDC68Δ69–114 show that the stability of CCDC68 protein is in part dependent on amino acids 69–114. Hence, although PDX models 135 and 110 expressed high CCDC68Δ69–114/CCDC68wt transcript ratio levels, protein stability could account for decreased ratios of translated CCDC68Δ69–114 proteins. The variable CCDC68Δ69–114/CCDC68wt ratio across PDAC could then be explained by differences in the regulation of protein turnover.

In addition to differences in regulation, loss of aa69–114 could also impact the function of CCDC68 protein. Functional properties of proteins can be dramatically altered by a series of post-translational modifications (PTMs) that ultimately affect the chemical properties of proteins. Using PTM prediction tools, we have identified several putative PTMs including lysine acetylation, SUMOylation, O-linked glycosylation, phosphorylation and ubiquitination sites are residing in the aa69–114 of CCDC68 protein (Supplementary Fig. S8). The loss of any of these sites could have significant consequences on the function and/or regulation of CCDC68. However, further exploration of the cellular function(s) of CCDC68 protein is needed to establish the impact of these PTMs on CCDC68 tumor suppressive ability.

This is the first study to describe a tumor suppressor role for CCDC68 in cancer. Previous studies of CCDC68 have been mainly descriptive. Originally named se57-1, CCDC68 has been identified as a putative tumor antigen in 21% of cutaneous T-cell lymphoma (21), 17% of renal cell (22) and 15% of colorectal carcinoma patients (23). Simultaneously and consistent with our PDAC findings, the same studies documented dramatic losses of CCDC68 expression in the majority of patients. Downregulation of CCDC68 has also been documented in 89% of primary colorectal patients and its expression was highly correlated with the associated gene copy number (15). This data also suggests the possibility that CCDC68 is also a novel candidate TSG in colorectal cancer. While this hypothesis requires further biological validation, the evidence of CCDC68 loss of function in human malignancies is accumulating. TCGA catalogues CCDC68 disruption through homozygous deletions, hypermethylation and somatic mutations across many human cancer types (http://www.cbioportal.org/). Our results on the role of CCDC68 in pancreatic carcinogenesis and the accumulating evidence of CCDC68 genetic alterations in cancer provide evidence that CCDC68 is a putative tumor suppressor.

Materials and Methods

Cell culture

PDAC cell lines used in this study were obtained and cultured as recommended by the American Type Culture Collection (Manassas, VA). H6c7 cells were cultured as described previously (24).

Array comparative genomic hybridization

Tiling path arrays were processed as previously described (25). CGH profiles were segmented to identify DNA copy number alterations using aCGH-Smooth. Duplicate profiles for H6c7-Kr and H6c7-KrasT were compared and clones were only considered if they were altered in the same direction in both profiles. Resulting clones were compared between parental and derivative cell lines. Specific regions of gain and loss that spanned two or more adjacent clones were compiled for each derivative and genes mapping to within these altered regions were determined.

Microarray analysis

Transcriptional profiling was done using the Illumina HumanHT-12 v4 array (Illumina, San Diego, CA) and the data was processed as previously described (26). Genes were considered aberrantly expressed if the fold change between samples and controls exceeded 1.5-fold and expression differences common to all cell lines were included in subsequent analyses. A network consisting of human protein-protein interaction pairs was generated using MIMI Plugin in Cytoscape. Protein clusters representing highly interconnected regions in the network were generated using the MCODE plugin in Cytoscape.

Fluorescent in situ hybridization (FISH)

Two probes were used: internal control centromeric probe CEP18 labeled with SpG (Abbott Molecular, Des Plaines, IL) and CCDC68 (RP11-108F19 BAC clone probe; TCAG, Toronto, ON). The CCDC68 probe was labeled with SpO using nick translation kit (Abbott Molecular) according to manufacturer’s protocol and hybridization was performed as described previously (27). Slides were scored at 63x magnification on an Imager M1 Zeiss microscope (Carl Zeiss Canada Limited, Toronto, CA) and analyzed using the MetaSystems Isis FISH Imaging v5.3 (MetaSystems, Newton, MA). A minimum of 100 non-overlapping intact interphase nuclei were scored per each sample. 40% cut-off value was applied to identify a heterozygous loss of CCDC68 in paraffin sections and 10% cut off value in cell suspensions.

Orfeome library and vector construction

Gateway CCDC68 ORF entry clones obtained from Human ORFeome library Version 1.1 (Fisher Scientific, Ottawa, ON) were subcloned into the pLD-puro-Ccf and pLD-puro-tGFP (28) as described previously (29). All cell lines were STR genotyped and tested for mycoplasma.

Lentiviral shRNA knockdown

CCDC68 knockdown in MiaPACA-2 was accomplished using a lentiviral shRNA method with RNAi Consortium clones: shCCDC68-1; NM_025214.1-2062s1c1 (TRCN0000129087), shCCDC68-2; NM_025214.2-1313s21c1 (TRCN0000412383). The shRNA control used was shLUC2 (GTGCCAGAGTCCTTCGATTCC). Lentiviral transduction was performed using protocols from TRC (http://www.broad.mit.edu/rnai/trc/lib).

Mutation analysis

CCDC68 ORF was amplified from cDNA transcribed from normal H6c7 cells, 19 PDAC cell lines and 31 primary xenografts using Touchdown PCR (TD-PCR) as described previously (30). Generated PCR products (P2-F+P2-R primers) were subjected to direct sequencing using sequencing primers P2-SF and P2-SR (Supplementary TableS2). SNP was further confirmed with a new set of primers specifically amplifying the region in question.

Minigene assay

Genomic DNA from either H6c7 (WT) or AsPC-1 (SNP) including CCDC68 intron-exon5-intron-exon6 was amplified using primers CCDC68DELFF and CCDC68151RR (Supplementary Table S2). The PCR fragment was subcloned into the pET01 exon trap vector (MoBiTec, Germany). After sequence confirmation, 293T and NIH3T3 cells were transfected with minigene constructs. RNA was isolated and the corresponding cDNA was amplified using pET01 specific forward primer (ETprim06) and CCDC68 reverse primer (P2SR). PCR products were examined on a 2% agarose gel.

Western analysis

Western analysis was described previously (31). Primary antibodies used in this study were: CCDC68 (S1852; Epitomics, Burlingame, CA), CCDC68 (SAB1103198;Sigma-Aldrich, St.Louis, MO) and β-ACTIN (A1978; Sigma Aldrich).

Immunohistochemistry (IHC)

IHC protocols were described previously (32). The staining intensity of CCD68 antibody (Sigma; SAB1103198,1:3000) was scored by a certified pathologist on a scale from 0–3 (0=absent; 1=low; 2=medium; 3=high).

DNA copy number analysis and RT-qPCR analysis

Quantitative polymerase chain reaction (qPCR) was performed as described previously (31) using two control genomic markers (G64212 and D4S1193) and two sets of CCDC68 specific primers (P1-F+P2-D; 68QPCRLCM-F+68QPCRLCM-R). CCDC68 copy number was estimated using comparative CT method relative to reference controls (n=3). Copy number changes were reported relative to median copy number changes across all the samples. Standard curve analysis was used or CCDC68Δ69–114 and CCDC68wt copy number estimation. cDNA was amplified using primers for CCDC68w (E6-7F1+R1) and CCDC68Δ69–114 (E4-6F1+R1). Standard curve was established using diluted vector DNA (pDONORCCDC68wt and pDONORCCDC68Δ69–114) ranging from 30–3 000 000 copies. RNA isolation and assay techniques were published previously (31). Relative quantification of qPCR data was performed using ΔΔCT method. The average Ct values for the duplicates were constructed separately for the target gene and two reference genes (RPS13 and β–ACTIN). All the primers used are listed in Supplementary Table S2. Heatmaps were created using Integrative Genomic Viewer software (33).

xCELLigence and MTS assays

MTS assays were performed as described previously (31). Growth curves were constructed using the xCELLigence platform (ACEA Biosciences, San Diego, CA). Briefly, 5 000 cells were seeded per well of E-plate. Impedance was measured every 15 minutes for 120–144 hours.

Soft agar and tumorigenic assay

Soft agar assay was described previously (34). Tumorigenicity assay was performed in scid mice using subcutaneous and orthotopic injections as described (26) in accordance with protocols approved by the Animal Care Committee of the Ontario Cancer Institute.

Meta-analysis of DNA copy number and mRNA expression of PDAC patients

Copy number and mRNA expression profiling data from the Australian pancreatic cancer project (project code PACA-AU) was downloaded from the ICGC data portal (https://dcc.icgc.org). This study was chosen due to its large cohort size. Relevant data was extracted with custom parser scripts and manipulated and visualized using the R programming language.

Statistical analysis

All statistical analysis was performed using GraphPad Prism 5 (GraphPad, La Jolla, CA). Statistical tests used are indicated in each figure. P values <0.05 were considered significant.