DRO1 Inactivation Drives Colorectal Carcinogenesis in Apc^Min/+ Mice Free

The inactivation of APC is an early step in the tumorigenesis of the majority of human colorectal cancers (1–3). APC inactivation in mice results in the development of intestinal adenoma and only rarely and then late in the development of cancers (4–6). The Apc^Min/+ mouse is the best studied and the most widely used animal model of human familial adenomatous polyposis (FAP). In Apc^Min/+ mice, spontaneous loss of the heterozygous wild-type Apc allele (7) leads to the formation of multiple benign adenomas throughout the intestinal tract, of which the vast majority is located in the small intestine. Consequently, Apc^Min/+ mice die at an age of approximately 150 days due to intestinal obstruction by the tumors and bleeding (4, 5). In contrast to the human disease, where affected individuals develop hundreds of adenoma mainly in the colon, which then progress to invasive cancer early in life (8), Apc^Min/+ mice commonly develop only few colon adenomas and only rarely colon carcinoma (4–6). The reason for this discrepancy is not known.

Previously, we identified DRO1 (CCDC80) to be a ubiquitously expressed candidate tumor suppressor gene (9). The expression of DRO1 has been found to be reduced in human cancers, including colorectal and thyroid tumors (9, 10). We found DRO1 to suppress anchorage-independent growth and to sensitize cancer cells to detachment-induced apoptosis (anoïkis).

On the basis of our previous findings, we reasoned that DRO1 (CCDC80) might act as a tumor suppressor gene of the colon in vivo. To test this hypothesis, we generated a mouse model (Dro1^−/−) lacking the second exon of Dro1, which contains the start codon and encodes for two thirds of the open reading frame and investigated the impact of DRO1 loss on colorectal carcinogenesis.

For generation of Dro1^−/− mice, mouse embryonic stem cells were targeted with a construct containing Dro1 exon 2 flanked with loxP sequences (Fig. 1A). Briefly, the essential exon 2 of the Dro1 gene and 5′ and 3′ homology arms were amplified by PCR (Expand High FidelityPLUS PCR System, Roche Diagnostics) from gDNA of E14 mouse embryonic stem cells (primer sequences available on request). In doing so, a loxP site was added to the 3′ end of the 5′ homology arm. A targeting vector was created by insertion of the PCR products into the multiple cloning site of the cloning vector pPNT4 (11). The construct was linearized by SalI restriction digest and electroporated into E14 mouse embryonic stem cells. Correctly targeted clones were injected into C57BL/6N blastocysts that were transplanted into the uteri of pseudopregnant NMRI foster mothers. Chimeric progeny were intercrossed with C57BL/6N mice to establish a mouse line with floxed Dro1 alleles (Dro1^fl/fl). Dro1^fl/fl females were mated to transgenic males expressing Cre recombinase under the control of the CMV promoter (12) to obtain Dro1^−/− mice. Dro1^fl/fl and Dro1^−/− mice were bred into the Apc^Min/+ background to obtain Dro1^−/−; Apc^Min/+ and Dro1^fl/fl;Apc^Min/+ mice (backcrossed in the C57BL/6 background for 3 generations). In all experiments, mice on the Dro1^fl/fl background (in the following referred to as Dro1^+/+ mice) were used as controls.

Apc^Min/+ mice were purchased from the Jackson Laboratory and maintained on a C57BL/6J background. Mice were inspected on a daily basis and sacrificed when moribund. Animals were housed under specific pathogen-free conditions in a closed barrier system. Experiments were carried out in accordance with the German Animal Welfare Act and were officially sanctioned by the local authorities.

Mice were genotyped by PCR analysis from genomic tail tip DNA. For purification of total RNA from microdissected FFPE tissue sections from human colorectal carcinoma specimens and from normal colon epithelium, the RNeasy FFPE kit (Qiagen) was used. For quantitative RT-PCR, RNA cleanup was performed using the RNeasy Mini kit (Qiagen). The gDNA was removed from the RNA preparation using DNase I, Amplification Grade (Invitrogen). First-strand cDNA was generated using the SuperScript First Strand cDNA Synthesis System (Invitrogen). Primer sequences are displayed in Supplementary Table S1.

Ten micrograms of liver gDNA was digested with PstI (MBI Fermentas), separated on a 0.9% agarose TAE gel, and blotted on a nylon membrane (Pall Corporation). A digoxigenin (DIG)-labeled hybridization probe was generated by PCR using primers Dro3′end#1 (5′-GCAAAGCTCTAGATAAGCCAG-3′) and Dro3′end#2 (5′-ACTCCTGTTCATAATGGCCAG-3′) and the DIG Probe Synthesis kit (Roche). For hybridization, washing, and detection, the DIG Block and Wash Buffer Set (Roche) and CSPD ready-to-use detection solution (Roche) were used. For information on the strategy used for Southern blot analysis, see Supplementary Fig. S1A.

Mice were killed by cervical dislocation, the intestine excised, rinsed with PBS (pH 7.4) to remove fecal material, fixed in 4% buffered formaldehyde solution in the form of “swiss rolls” (13), dehydrated, and embedded in paraffin. Serial sections were cut and stained with hematoxylin and eosin (H&E) or Periodic acid-Schiff reagent (PAS) according to standard protocols.

Immunohistochemical staining was performed on 4-μm sections of formalin-fixed, paraffin-embedded (FFPE) tissue samples. Slides were finally counterstained with hematoxylin (Vector Laboratories). Reagents used for immunohistochemistry are displayed in Supplementary Table S2.

The number of PAS-positive cells per total cells in the crypt–villus axis of the small intestine was counted in 20 crypts per mouse. To evaluate intestinal proliferation, the number of bromodeoxyuridine (BrdUrd)-stained/Ki-67–stained cells was counted in 20 crypts of the small intestine and 20 crypts of the colon per mouse. To evaluate intestinal apoptosis rate, the number of cleaved caspase-3–positive cells was counted in 50 crypt–villus units in the small intestine and in 100 crypts in the colon per mouse. Within neoplastic tissue, the percentage of nuclear β-catenin–positive cells was estimated.

Nuclear expression of β-catenin in human colorectal cancer samples was evaluated only on the basis of the quantity of stained tumor nuclei throughout the whole tumor, whereas intensity of staining was not considered. The score was as follows: 0, negative; +, <30%; ++, 30%–60%; +++, >60% positive cells. All experiments were carried out in a blinded manner.

The small intestine of sacrificed mice was cut into 3 equal segments and each intestinal section was placed on a piece of filter paper, opened longitudinally, laid open, and fixed in 4% buffered formaldehyde solution. Tumor number and their maximum diameter were determined under a dissecting microscope at 10× magnification. The colon and rectum were scored as “colon”. A quantity of small intestinal lesions and all colonic tumors sized ≥2 mm in diameter were resected, including adjacent normal tissue, dehydrated, and embedded in paraffin, 4-μm tissue sections cut in parallel with the mucosal surface, and stained with H&E. In a subset of 5- and 10-week-old mice, the colon was processed as “Swiss roll,” and a single well-oriented H&E-stained section was investigated. Histopathologic analysis of neoplastic lesions was performed in a blinded manner using standard criteria according to the classification of human adenomas of the colon and the assessment of the degree of dysplasia. The diagnosis of intramucosal adenocarcinoma was made for lesions with high-grade dysplasia/intraepithelial neoplasia (IEN) in combination with focal invasion of the lamina propria mucosae, cytologic features such as cribriform architecture with intraluminal accumulation of tumor and inflammatory cell debris (dirty necrosis) and desmoplastic stromal reaction. Adenocarcinomas invading through the lamina muscularis mucosae into the tela submucosa were classified as invasive adenocarcinoma (14). For subsequent analysis, invasive adenocarcinoma and intramucosal adenocarcinoma were summarized under the diagnosis adenocarcinoma.

Colon samples were stained with methylene blue as described before (15) and analyzed by light microscopy at 100-fold magnification and transillumination (Stemi SV 6, Zeiss).

Protein lysates from mouse tissue samples were generated using M-PER mammalian extraction reagent (Thermo Scientific) and separated by electrophoresis in discontinuous SDS-PAGE. The following antibodies were used for immunodetection: anti-ERK1/2 (Cell Signaling), anti-p-ERK1/2 (Cell Signaling), anti-c-MYC (Cell Signaling), and anti-ACTIN (MP Biomedicals). Densitometry was performed using ImageJ. To calculate relative band intensities, band intensity for p-ERK1/2 and c-MYC was normalized to ERK1/2 (for p-ERK1/2) and ACTIN (for c-MYC) band intensity.

To display the time to tumor mortality, Kaplan–Meier survival curves were used, and log-rank statistics were used to test for differences between genotype groups. To analyze significance of differences, 2-tailed Student t test or 2-tailed Mann–Whitney U test were performed (GraphPad Prism 4). P < 0.05 was considered to be statistically significant.

In Dro1^−/− mice, genetic deletion of Dro1 was confirmed by Southern blot and PCR analysis (Supplementary Fig. S1), and no Dro1 expression was detected by quantitative RT-PCR in various tissues as compared with controls (Fig. 1B). Dro1^−/− mice were viable and fertile and revealed no differences in litter size or mortality (data not shown). Upon autopsy of 13 Dro1^−/− and 16 Dro1^+/+ control mice, which had been carefully followed for 18 months, a single hepatocellular carcinoma was observed in a Dro1^−/− mouse (Fig. 1C). No intestinal tumor was identified in these mice. Detailed analysis of small intestine and colon revealed no changes in anatomy and distribution of the 4 intestinal cell lineages (Supplementary Fig. S2) and no significant differences in the rates of proliferation and apoptosis (Supplementary Fig. S3). Thus, DRO1 appears to be dispensable for the maintenance of the intestinal epithelium, and disruption of Dro1 does not result in the initiation of intestinal tumors.

To study the consequence of DRO1 deficiency on intestinal tumorigenesis in a tumor-prone background, Dro1^−/− mice were bred to Apc^Min/+ mice. Dro1^−/−;Apc^Min/+ mice exhibited a remarkable phenotype compared with Dro1^+/+; Apc^Min/+ controls. While the median survival of Dro1^+/+; Apc^Min/+ mice was 142 days, the median survival of Dro1^−/−; Apc^Min/+ mice was reduced to 94 days (HR, 0.35; 95% confidence interval, 0.12–0.55; P = 0.0004; Fig. 1D). There was no difference in number, distribution, size, and histology of tumors in the small intestine (Fig. 2A and B; Supplementary Fig. S4A and S4B), but we observed a striking difference in the colonic tumor burden between Dro1^+/+; Apc^Min/+ and Dro1^−/−; Apc^Min/+ mice (Fig. 2C and D). The mean number of colonic tumors was 11.9 in Dro1^−/−; Apc^Min/+ mice compared with 3.1 in controls (P = 0.002; Fig. 2D, α), with a significant increase in mean tumor multiplicity in both the proximal (P = 0.038) and distal (P = 0.002) colon in Dro1^−/−; Apc^Min/+ mice (Fig. 2D, β). No difference in the relative distribution of tumors throughout the colon or median tumor size was observed (Fig. 2D, γ). Although tumor number was increased about 3-fold in the colon of Dro1^−/−; Apc^Min/+ mice, the vast majority of polyps localized to the small intestine in both genotypes.

Macroscopically, no metastases to lymph nodes, liver, and lungs were identified. Upon histologic analysis, the 15 colon tumors isolated from 18 control mice revealed to be adenomas (Fig. 3A and Supplementary Fig. S4C). In contrast, analysis of the 83 tumors isolated from the colons of 19 Dro1^−/−;Apc^Min/+ mice revealed 29 (35%) adenocarcinoma and 54 (65%) adenoma (Fig. 3A and B and Supplementary Fig. S4D). Notably, 28 (97%) of these 29 carcinomas were located in the distal colon.

To further study the development of tumors, cohorts of mice were sacrificed at 5 or 10 weeks of age. No differences in the polyp load of the small intestine were observed (Supplementary Fig. S5A). At the age of 5 weeks, no colon tumors were observed in either genotype (Supplementary Fig. S5B). However, 10-week-old Dro1^−/−; Apc^Min/+ mice revealed significantly more tumors in the colon than Dro1^+/+; Apc^Min/+ controls (P = 0.007; Supplementary Fig. S5B). Four of 14 (29%) tumors in the Dro1^−/−; Apc^Min/+ mice and 1 of 6 (17%) tumors in the control mice were adenocarcinoma (Supplementary Fig. S5C–S5F and Supplementary Table S3). Moreover, there was no difference in the formation of preneoplastic aberrant crypt foci (ACF) between Dro1^−/−; Apc^Min/+ and Dro1^+/+; Apc^Min/+ mice (Supplementary Fig. S6), supporting the notion that loss of Dro1 expression accelerates tumor development and progression via an adenoma-to-carcinoma sequence.

We next used immunohistochemistry to study colon epithelium of 5-week-old mice derived from Dro1^−/−; Apc^Min/+ and control mice. Architecture, distribution of cell types, and proliferation and apoptosis rates in the colon were undistinguishable between both genotypes (Supplementary Fig. S7). We also observed no difference in the rate of apoptosis (Supplementary Fig. S8A–S8C) between colon tumors from Dro1^−/−; Apc^Min/+ and control mice. As tumors of Apc^Min/+ mice are known to display an activation of Wnt/β-catenin signaling (16), we tested whether loss of DRO1 might lead to the further deregulation of this pathway. However, no differences in nuclear protein expression of β-catenin were found (Supplementary Fig. S8D–S8G). Consistently, we found no correlation between DRO1 expression level and the degree of nuclear β-catenin accumulation in primary human colorectal cancer specimens (Supplementary Fig. S9). These results suggest that changes in colonic tumor multiplicity and neoplastic progression in Dro1^−/−; Apc^Min/+ mice are most probably mediated by pathways distinct of Wnt/β-catenin signaling.

There is growing evidence that activation of the MEK/ERK signaling pathway is implicated in the development and progression of colorectal cancer. In Apc^Min/+ mice, MEK/ERK signaling activation has been identified to be fundamental for intestinal tumor development (17). To elucidate whether changes in the phosphorylation status of ERK can be observed in colon epithelial samples and colon tumors after DRO1 loss in Apc^Min/+ mice, immunoblot analysis and immunohistochemistry were performed. This revealed that levels of p-ERK1/2 were higher in colon epithelium and in colon tumors of Dro1^−/−; Apc^Min/+ mice than in epithelium and adenoma of control mice (Fig. 3C and D). We also observed higher levels of p-ERK1/2 in adenoma than in normal epithelium of control mice albeit not as high as in Dro1^−/−; Apc^Min/+ mice (Fig. 3C and Supplementary Fig. S10). Therefore, we assume that additional mechanisms might contribute to the upregulation of the ERK–survival axis.

In Apc^Min/+ mice, p-ERK1/2 activation has been demonstrated to promote intestinal tumorigenesis by c-MYC oncoprotein stabilization (17). To investigate whether upregulation of p-ERK1/2 in Dro1^−/−; Apc^Min/+ mice is associated with c-MYC activation, immunoblot analysis was performed. We found higher levels of c-MYC in colon tumors than in normal colon epithelium in both Dro1^+/+; Apc^Min/+ and Dro1^−/−; Apc^Min/+ mice (Fig. 3C). Moreover, we found c-MYC upregulation in adenoma and adenocarcinoma from Dro1^−/−; Apc^Min/+ mice compared with adenoma from control mice (Fig. 3C and Supplementary Fig. S10). Thus, c-MYC oncogenic activation might facilitate colon carcinogenesis in Dro1^−/−; Apc^Min/+ mice.

Previous in vitro studies suggested DRO1 to be a putative tumor suppressor gene (9). Until today, investigation of the Dro1 tumor suppressor function in vivo was restricted to expression analysis, offering evidence for involvement in colorectal, mammary, and thyroid carcinogenesis (9, 10, 18, 19). The results of the present study clearly demonstrate Dro1 to be an important colonic tumor suppressor gene in vivo that suppresses polyp initiation and particularly prevents tumor progression toward malignancy in Apc^Min/+ mice.

Interestingly, DRO1 deficiency did not result in obvious effects on number, size, or histopathologic quality of polyps in the small intestine of Apc^Min/+ mice, indicating that the Dro1 tumor suppressor gene function in the intestinal tract is confined to the colon. Recent investigations on the gene expression in the murine intestinal tract have uncovered site-specific differences between the small intestine and colon (20). It is conceivable that differences in gene expression pattern between small and large intestine are responsible for the restriction of the tumor suppressor activity of Dro1 to one compartment.

To the current understanding, human colorectal cancer develops from adenomatous precursor lesions by the accumulation of multiple independent somatic alterations in oncogenes and tumor suppressor genes (21). We found that loss of DRO1 alone was not sufficient to initiate intestinal tumorigenesis. It is well known that inactivation of a tumor suppressor gene alone is often not sufficient to cause tumorigenesis, and additional mutational events that result in perturbation of other critical signaling pathways are necessary (22). Inactivating mutations of the APC tumor suppressor gene are viewed as an early if not initiating event in up to 80% of human sporadic colorectal cancers and have been implicated in the development of a multitude of adenomas in the colon and rectum of patients with FAP (2, 23). Our findings support a model in which loss of Dro1 expression drives the development of adenoma and the progression of adenoma to carcinoma in the colon of the Apc^Min/+ mouse. Loss of DRO1 appears to be required for phosphorylation of ERK. Therefore, ERK might represent the survival signal and deregulated Wnt/β-catenin signaling the growth signal required for colorectal tumorigenesis in this model. An important role of the ERK pathway in the establishment of the Min phenotype has previously been described (17), and activation of ERK has been demonstrated to predict poor prognosis in patients with colorectal cancer (24). Furthermore, in addition to the inactivation of APC, activation of ERK signaling either by loss of DRO1 as demonstrated here or by mutation of KRAS in another APC-mutant background (25) appears to act synergistically and results in an increase in tumor multiplicity and malignant transformation in the colon.

We also found DRO1 inactivation to increase c-MYC protein levels in colon tumors from Apc^Min/+ mice. Activation of the c-MYC oncoprotein has been identified to be essential for intestinal tumorigenesis in APC-deficient mice (26, 27). Previously, Lee and colleagues demonstrated β-catenin–independent c-MYC stabilization in Apc^Min/+ mice via phosphorylation by ERK1/2 (17). As nuclear β-catenin accumulation within neoplastic tissue was unaffected by DRO1 loss, the observed changes in c-MYC protein level might also be caused by p-ERK1/2 upregulation.

Our previous in vitro studies have identified DRO1 to sensitize colorectal cancer cells to detachment-induced apoptosis/anoïkis (9). In colon tumors from Apc^Min/+ mice, we found no effect of DRO1 inactivation on the rate of apoptosis. However, our analysis might not account specifically for anoïkis, as staining for cleaved caspase-3 targets apoptotic cells in general, not just the few that die upon detachment from the extracellular matrix. Therefore, the role of anoïkis in the observed phenotype is still unclear, and further studies are needed to elucidate this point.

The Dro1^−/−; Apc^Min/+ mouse demonstrates that DRO1 is not simply a modifier of the Min-phenotype but a potent suppressor of colonic tumorigenesis. This new mouse model may serve as a valuable tool for the study of colorectal carcinogenesis in the future. Further studies are required to elucidate the molecular mechanisms underlying the tumor suppressor gene function of Dro1 and to better understand its role in human colorectal carcinogenesis.

F. Hiltwein is the Head of Animal Care and Manager In Vivo Biology at Heidelberg Pharma GmbH. T. Kirchner reports receiving a commercial research grant from Merck and Amgen and is a consultant/advisory board member for Amgen, Roche, Pfizer, and Merck-Serono. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J.I. Grill, A. Herbst, T. Kirchner, M.R. Schneider, F.T. Kolligs

Development of methodology: J.I. Grill, J. Neumann, A. Ofner, M.K. Marschall, M.R. Schneider, F.T. Kolligs

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.I. Grill, J. Neumann, F. Hiltwein, A. Ofner, M.K. Marschall, M.R. Schneider

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.I. Grill, A. Herbst, A. Ofner, M.K. Marschall, B. Göke, M.R. Schneider, F.T. Kolligs

Writing, review, and/or revision of the manuscript: J.I. Grill, J. Neumann, A. Herbst, E. Wolf, B. Göke, M.R. Schneider, F.T. Kolligs

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Wolf, T. Kirchner

Study supervision: A. Herbst, M.R. Schneider, F.T. Kolligs

The authors thank Ingrid Renner-Müller and Petra Renner for help with animal care, Andrea Sendelhofert and Anja Heier for help with immunohistochemistry, Benjamin Hirschi and Maik Dahlhoff for help with microscopy, and Sabrina Porada for help with necropsy. They also thank Guido Bommer (Universite Catholique de Louvain, Bruxelles, Belgium) for providing materials and helpful discussions.

This work was supported by the German Cancer Consortium (DKTK), Heidelberg, Germany/German Cancer Research Center, Heidelberg, Germany and a grant of the German Research Foundation (KO1826/5-1 and KO1826/5-2) to F.T. Kolligs.

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.