Haploinsufficiency of Atp2a2, Encoding the Sarco(endo)plasmic Reticulum Ca-ATPase Isoform 2 Ca Pump, Predisposes Mice to Squamous Cell Tumors via a Novel Mode of Cancer Susceptibility

A null mutation in one copy of the Atp2a2 or ATP2A2 gene, encoding sarco(endo)plasmic reticulum Ca-ATPase isoform 2 (SERCA2), leads to squamous cell tumors in mice and to Darier disease in humans, a skin disorder that also involves keratinocytes. Here, we examined the time course and genetic mechanisms of tumor development in the mutant animals. Atp2a2 mice overexpressed keratins associated with keratinocyte hyperactivation in normal forestomachs as early as 2 months of age. By the age of 5 to 7 months, 22% of mutants had developed papillomas of the forestomach, and 89% of mutants older than 14 months had developed squamous cell papillomas and/or carcinomas, with a preponderance of the latter. Tumors occurred in regions that had keratinized epithelium and were subjected to repeated mechanical irritation. The genetic mechanism of tumorigenesis did not involve loss of heterozygosity, as tumor cells analyzed by laser capture microdissection contained the wild-type Atp2a2 allele. Furthermore, immunoblot and immunohistochemical analysis showed that tumor keratinocytes expressed the SERCA2 protein. Mutations were not observed in the ras proto-oncogenes; however, expression of wild-type ras was up-regulated, with particularly high levels of K-ras . Loss of the p53 tumor suppressor gene occurred in a single massive tumor, whereas other tumors had increased levels of p53 protein but no mutations in the p53 gene. These findings show that SERCA2 haploinsufficiency predisposes mice to tumor development via a novel mode of cancer susceptibility involving a global change in the tumorigenic potential of keratinized epithelium in Atp2a2 mice. (Cancer Res 2005; 65(19): 8655-61) Introduction The endoplasmic reticulum (ER) is the principal Ca storage organelle in cells, and one of the most critical determinants of Ca levels in the ER of most cell types is the activity of sarco(endo)plasmic reticulum Ca-ATPase isoform 2 (SERCA2; refs. 1, 2). Encoded by the Atp2a2 gene, this pump sequesters Ca in the ER, and for this reason, its activity is an important regulator of normal Ca homeostasis and signaling (3). Alterations in Ca handling are associated with cell proliferation and differentiation (4, 5), and perturbations of Ca homeostasis have been suggested to contribute to the development of cancer (6, 7). Reduced levels of SERCA2, caused by a null mutation in one allele of the ATP2A2 gene, leads to Darier disease in humans, an autosomal dominant skin disorder characterized by multiple keratotic papules (8). Although loss of one copy of the Atp2a2 allele in mice did not cause symptoms characteristic of Darier disease, it did lead to a very high incidence of squamous cell tumors (9), involving keratinocytes, the same cell type affected in humans. Atp2a2 heterozygous mutant (Atp2a2 ) mice developed hyperkeratinized tumors in regions of stratified squamous epithelia that included the oral mucosa, tongue, palate, esophagus, nonglandular mucosa of the stomach, skin, and genitalia. SERCA2 protein expression was reduced in the affected tissues, consistent with the hypothesis that the tumors were due to SERCA2 haploinsufficiency (9). This is an exciting possibility, as it would represent a novel mode of tumor susceptibility. However, an alternative possibility is that the tumors arise in rare cells in which the loss of Atp2a2 heterozygosity has occurred, which would be similar to mechanisms involving known tumor suppressor genes. Thapsigargin, an irreversible inhibitor of SERCA2 activity, serves as a tumor promoter in multistage mouse skin carcinogenesis (10) and has been shown to induce DNA synthesis (11) and cause growth stimulation (12) in cultured keratinocytes. It is unclear, however, whether these effects are due to partial inhibition of SERCA2, which would resemble haploinsufficiency, to complete inhibition, which would resemble loss of heterozygosity (LOH), or to some other effect that is independent of SERCA2. The primary genetic lesions in a wide variety of tumors are gainof-function mutations in the ras family of proto-oncogenes (13). Ras oncogene expression alone has been considered insufficient for tumor progression, with malignant conversion requiring additional lesions resulting in loss of tumor suppressor activity (14); the p53 gene is the most commonly mutated tumor suppressor gene. Thus, tumorigenesis often involves the accrual of a succession of stochastic, genetic mutations that occur over time within a specific cell, allowing that cell to progress toward malignancy (14). However, tumors induced by nongenotoxic carcinogens typically have revealed mutations in neither the H-ras nor the p53 gene (15). The generation of these tumors was associated with irritant damage and with hyperproliferation, with early effects characterized by hyperplasia and hyperkeratosis, progressing to papillomas and squamous cell carcinomas (16). These observations are Requests for reprints: Gary E. Shull, Department of Molecular Genetics, Biochemistry and Microbiology, University of Cincinnati College of Medicine, 231 Albert Sabin Way, Cincinnati, OH 45267-0524. Phone: 513-588-0056; E-mail: gary.shull@ uc.edu. I2005 American Association for Cancer Research. doi:10.1158/0008-5472.CAN-05-0026 www.aacrjournals.org 8655 Cancer Res 2005; 65: (19). October 1, 2005 Research Article Research. on April 14, 2017. © 2005 American Association for Cancer cancerres.aacrjournals.org Downloaded from consistent with the proposal that enforced cell proliferation can lead to alternative pathways of tumor development (17). In this report, we present our analyses of the kinetics of tumor development in Atp2a2 mice and of possible genetic deficiencies involving the ras and p53 genes. We present evidence that Atp2a2 haploinsufficiency itself, rather than LOH, is responsible for the cancer phenotype of these mice, with enhanced tumor susceptibility occurring in keratinized epithelia. Lesions, such as hyperplasia and hyperkeratosis, occur in the early stages followed by tumor initiation and progression via nonclassic pathways involving elevated expression of wild-type H-ras , K-ras , and, surprisingly, p53. Materials and Methods Mice. The generation of the Atp2a2 mice has been reported earlier (1). The mice were maintained on both a mixed 129/Svj and Black Swiss background and an inbred FVB/N background (back-crossed 10 generations). Genotypes were determined by PCR analysis of tail DNA as described before (1). All animals were maintained in accordance with institutional guidelines. Age-matched wild-type and mutant mice, usually siblings, were paired and observed as they aged; >90 pairs of mice were euthanized at different ages [5-7 or 8-13 and z14 months (when morbidity set in)] and tissues observed for any overt tumors. This allowed us (a) to study the kinetics of tumor development and (b) to determine genetic deficiencies, especially in the early tumors. At necropsy, the oral cavity and various organs were examined for tumors or abnormalities; tissues were fixed in 10% neutral buffered formalin, embedded in paraffin, sectioned, and stained with H&E. Histologic diagnosis was carried out to identify lesions, which included squamous cell papillomas and squamous cell carcinomas. Tumor tissue isolated for DNA or protein analysis was frozen in liquid nitrogen and stored at 80jC until further use. Morphometry. Forestomach samples were fixed in 2.5% glutaraldehyde/ 2% paraformaldehyde, dehydrated, and embedded in Spurr’s resin. Sections (2 Am) were stained with toluidine blue and used for morphometry. Avoiding areas of tangential cut, the thickness of the forestomach epidermis was measured with a line beginning at the basement membrane drawn to the lumen of the stomach (f10 measurements per animal) using SigmaScan Pro software (Jandel Scientific, Palo Alto, CA) and Excel. In the epidermis, mitotic cells and apoptotic bodies were counted and relative percentages were obtained. Data were analyzed using SAS version 8.0 (Cary, NC), and means and SEs of themeanswere determined by genotype, gender, and age using the General Linear Model. Results were considered significant when P < 0.05. When a variable in wild-type mice showed no significant differences among the categories of age and/or gender, these data were pooled. Apoptotic index was also determined using the Klenow FragEL DNA Fragmentation Detector kit (Calbiochem Biochemicals, La Jolla, CA). Laser capture microdissection and PCR analysis. The protocol was adapted from the National Institute of Child Health and Human Resources LCM Research Resources. Tumors or tissue sections with hyperplasia (5 Am thickness) were stained with H&E and stored in a desiccator until microdissection. Cells were captured on CapSure Macro LCM Caps (Arcturus Bioscience, Inc., Mountain View, CA) using a PixCell II laser capture microscope (Arcturus Bioscience) at the Cincinnati Children’s Hospital Laser Capture Core. Captured cells were digested in DNA extraction buffer [100 mmol/L Tris (pH 8.5), 10 mmol/L EDTA, 200 mmol/L NaCl, 0.2% SDS + 0.6 mg/mL proteinase K) for f20 hours at 55jC. Protein and SDS were precipitated using the Puregene DNA Isolation kit (Gentra Systems, Minneapolis, MN). Genomic DNA was then isolated by isopropanol precipitation in the presence of glycogen and probed for the presence of the wildtype and mutant Atp2a2 alleles using PCR analysis as described earlier (1). PCR amplification and sequencing analysis. Genomic DNA was isolated from tumor samples by phenol/chloroform extraction. The Ensembl Genome Browser, developed and maintained by the European Molecular Biology Laboratory and the Sanger Institute, was used to design primers for the various exons in the H-ras , K-ras , and p53 genes; for the p53 gene, exons 3 to 9 were sequenced with primers located in the intronic regions to permit analysis of all intron-exon boundaries. Primers were synthesized by Sigma-Genosys (The Woodlands, TX); primer sequences and PCR protocols can be obtained by contacting the authors. PCR-amplified fragments were resolved on a 2.5% agarose gel and extracted using the QIAQuick gel extraction kit (Qiagen, Valencia, CA). The University of Cincinnati DNA Core Facility using a 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA) conducted sequencing of both strands, and the Basic Local Alignment Search Tool at the National Center for Biotechnology Information was used for alignments. Immunoblot and immunohistochemical analysis. Tissues were pulverized in liquid nitrogen and suspended in homogenization buffer [10 mmol/L NaCl, 20 mmol/L PIPES (pH 7.0), 0.5% NP40, 0.05% hmercaptoethanol, 5 mmol/L EDTA, 50 mmol/L NaF, and a protease inhibitor cocktail (Sigma, St. Louis, MO)]. Tissue was homogenized manually using a glass homogenizer and solubilized on ice for f2 hours. Protein concentration was estimated using the Coomassie Plus protein assay reagent (Pierce, Rockford, IL). Proteins were separated by reducing SDS-PAGE and transferred to nitrocellulose membranes. The various antibodies used were polyclonal anti-SERCA2 (antibody N1; ref. 18; provided by Jonathan Lytton, University of Calgary, Calgary, Alberta, Canada), monoclonal anti-H-ras (clone 7D7.2, Chemicon International, Temecula, CA), monoclonal anti-K-ras (clone F234, Santa Cruz Biotechnology, Santa Cruz, CA), monoclonal anti-p53 (clone PAb 240, Novocastra Laboratories, Newcastle upon Tyne, United Kingdom), and polyclonal anti-mouse keratin 6 (K6) and keratin 10 (K10; Covance Research Products, Berkeley, CA). Horseradish peroxidase–conjugated secondary antibodies were from KPL, Inc. (Gaithersburg, MD). Chemiluminescence was developed using the LumiGlo Chemiluminescent substrate kit (KPL), and autoradiograms were developed using BioMax Films ML and MR (Kodak, Rochester, NY). For immunohistochemistry, tumor tissue was excised, fixed in 10% buffered formalin, and embedded in paraffin. Sections (5 Am thickness) were deparaffinized, and on rehydration, antigen retrieval was carried out in 10 mmol/L citric acid (pH 3.0) at 37jC for 30 minutes. Sections were stained for SERCA2 protein using the N1 antibody in conjunction with the Cell and Tissue Staining kit (R&D Systems, Minneapolis, MN). Stained sections were dehydrated and mounted using Fluoromount G (Southern Biotechnology Associates, Inc., Birmingham, AL). Sections were analyzed using the Axioskop and Axiovision 4.0 software (Carl Zeiss, Germany). Results Atp2a2 mice develop squamous cell tumors. We have reported previously that Atp2a2 mice on the mixed 129/Svj and Black Swiss background spontaneously developed squamous cell tumors on aging (9). In the current study, the cancer phenotype was found to persist even when the mice were transferred onto the FVB/N background, and there seemed to be no difference in penetrance of the phenotype for either background. As shown in Tables 1 and 2, 22% (6 of 27 mice) of 5to 7-month-old mutant animals had papillomas in the forestomach. In addition, 11 other mutants in this group had foci of hyperplasia in the forestomach, 4 of which were associated with hyperkeratosis. Carcinomas were not observed in any of the 5to 7-month-old mutants and no squamous cell tumors were observed in the 27 wild-type controls. Thirty-three pairs of mice (and 11 additional unpaired mutants) were studied between 8 and 13 months of age. Of these 44 mutants, 29 animals developed squamous cell tumors, which included 21 carcinomas and 19 papillomas. The forestomach was the most affected tissue with 13 papillomas and 9 carcinomas. None of the 33 wild-type controls developed tumors of any type (Tables 1 and 2). When Atp2a2 mice were aged to z14 months, the 5 http://www.ensembl.org Cancer Research Cancer Res 2005; 65: (19). October 1, 2005 8656 www.aacrjournals.org Research. on April 14, 2017. © 2005 American Association for Cancer cancerres.aacrjournals.org Downloaded from cancer phenotype exhibited 89% penetrance (32 of 36 mice). Of the 55 mutants ( from 36 pairs plus 19 unpaired mutants) studied at this age, only one 15-month-old female had no abnormalities, whereas 5 other animals developed moderate to severe hyperplasia of keratinized epithelia but no tumors. Although the most frequently affected tissue was forestomach, with 22 papillomas and 14 carcinomas, the tissue exhibiting the highest frequency of carcinomas was skin (17 of 55 mice, with 3 mice having multiple skin tumors). Interestingly, there were no papillomas in the skin, suggestive of a higher rate of tumor progression and in sharp contrast to forestomach, oral cavity, and esophagus, where papillomas persisted in mice older than 14 months of age. All of the tumors were well differentiated and had extensive keratinization (Fig. 1). Of the 36 wild-type mice aged to at least 14 months, only 2 animals developed squamous cell tumors, a papilloma on the lip and a carcinoma on the penis, at 18 and 17 months, respectively. Tumorigenesis in Atp2a2 mice does not involve loss of heterozygosity or loss of expression of the remaining wild-type allele. The susceptibility of keratinized epithelial cells in Atp2a2 mice to neoplastic transformation could result from either a cell-specific loss of wild-type Atp2a2 expression or Atp2a2 haploinsufficiency. In cases of a preexisting null mutation in one allele, the major genetic mechanism for loss of expression of the wild-type allele is LOH (19, 20). To test this hypothesis, we did laser capture microdissection, which allows selective isolation of tumor cells free of stromal cells, followed by analysis of the Atp2a2 alleles present in the tumor cells. The tumors were selected from a spectrum of affected tissues that included skin, palate, esophagus, forestomach, penis, and vagina. PCR analysis of genomic DNA from the laser-captured tumor cells revealed the presence of the wild-type allele in all of the samples (Fig. 2A). Given the number of tumors analyzed, this result rules out loss of the remaining wild-type allele as the primary genetic mechanism of tumorigenesis. To determine whether SERCA2 protein expression from the wildtype allele may have been down-regulated via epigenetic mechanisms, immunoblot and immunohistochemical analyses were done. Tumors analyzed by immunoblotting included squamous cell papillomas and/or squamous cell carcinomas from forestomach, tongue, and palate. This experiment revealed that expression of the SERCA2 protein, when normalized to actin (Fig. 2B), tended to exceed levels seen in normal mutant tissues. Furthermore, immunohistochemical analysis of sections from squamous cell papillomas and squamous cell carcinomas revealed robust expression of the SERCA2 protein in tumor keratinocytes (Fig. 2C and D). These results indicate that loss or inactivation of the single, wild type Atp2a2 allele was therefore not the primary genetic lesion driving tumor development. An exciting alternative would be that the single wild-type Atp2a2 allele, unable to compensate for the loss of the targeted allele, causes a tissue-level alteration in keratinized epithelia, making it susceptible to tumorigenesis. Atp2a2 forestomachs with increased levels of keratin associated with keratinocyte hyperactivation. To elucidate the effects, if any, of reduced SERCA2 expression on keratinized epithelia before development of lesions, protein lysates of forestomach from 2-month-old Atp2a2 and Atp2a2 animals were analyzed by immunoblotting to determine expression levels of keratins. This tissue, which is free of lesions at this age, was chosen because stomach was the most commonly affected organ. Keratinocytes express highly specific keratin markers depending on their state of proliferation and/or differentiation. Keratin 1 and K10 are expressed by postmitotic keratinocytes in the suprabasal layers during normal differentiation, whereas expression of alternative keratins, such as K6, is often indicative of enforced proliferation and activation, typically occurring during wound healing and inflammation. Expression of K6 was elevated 76% in histologically normal mutant forestomachs when compared with wild-type controls (Fig. 3). In contrast, there was no significant alteration in expression of K10 (Fig. 3). Despite the frequent association of higher K6 levels with increased proliferation, the keratinized epithelium of Atp2a2 forestomach did not exhibit evidence of increased proliferation. Morphometric analysis of forestomachs revealed no significant differences in epidermal thickness (Atp2a2, 40.4 F 2.7 Am; Atp2a2 , 36.9 F 2.1 Am; P = 0.35) or mitotic index (percent mitosis: Atp2a2, 0.57 F 0.3; Atp2a2 , 0.70 F 0.3; P = 0.78) between wild-type and mutant tissues. Similarly, there was no significant difference in the apoptotic index as determined by morphometry and an in situ end-labeling assay, similar to the TUNEL assay (data not shown). These results suggest that tissue level alterations occurring in prelesional Atp2a2 keratinized epithelia, as revealed by aberrant K6 expression in forestomach, are not due to alterations in keratinocyte proliferation or apoptosis but rather reflect altered differentiation. Atp2a2 tumors lacked mutations in the ras and p53 genes, but ras and p53 protein levels were increased. Initiation of squamous cell tumors is typically associated with oncogene expression, with gain-of-function mutations occurring in one or more of codons 12, 13, or 61 of the ras genes. Furthermore, tumor progression is believed to require additional genetic lesions targeting tumor suppressor genes, such as p53 , which may otherwise play a role in ras-mediated apoptosis or cell cycle arrest. To determine whether these mutations were occurring, sequence analysis of H-ras , K-ras , and p53 exons was carried out to identify any such point mutations. Genomic DNA was isolated from five squamous cell papillomas and six squamous cell carcinomas from various tissues, including palate, esophagus, forestomach, and penis. For analysis of H-ras and K-ras genes, primers were designed to span exons 1 and 2, which include codons 12, 13, and 61. No mutations were found at these codons in any of the tumors. For mutational analysis of the p53 gene, primers were designed to span exons 3 to 9 because of increasing evidence that mutations can occur outside the gene segment traditionally analyzed (exons 5-8). Table 1. Penetrance of cancer phenotype: animals with tumors of keratinized epithelia/total number of animals