High Incidence of Breast and Endometrial Neoplasia Resembling Human Cowden Syndrome in pten^+/− Mice

PTEN was identified as a candidate tumor suppressor gene frequently deleted at chromosome 10q23 in a number of advanced tumors such as glioblastoma, prostate, kidney, and breast carcinoma(1, 2) and independently discovered in a search for novel tyrosine phosphatases (3). A significant rate of PTEN mutations were also reported in high-grade glioblastomas (4, 5, 6, 7, 8, 9) and sporadic cancers of the prostate(10), thyroid (1, 2, 11, 12), and endometrium(13, 14, 15).

The human PTEN gene encodes a 403-amino-acid polypeptide with a high degree of homology to protein phosphatases and tensin(1, 2). The importance of PTEN phosphatase activity for its tumor suppressor function is highlighted by the fact that most of tumor-associated PTEN mutations disrupt the integrity of its phosphatase domain (6, 16). Despite its homology to protein phosphatases, PTEN is able to dephosphorylate PI³(3,4,5) trisphosphate [PI(3,4,5)P₃], the primary product of PI3′K activity (17) and to negatively regulate PI3′K-mediated cell survival (18). Expression of high levels of PTEN in certain cell lines leads to apoptosis (18, 19), whereas in others, overexpression of PTEN causes G₁ arrest (20, 21) or both G₁arrest and apoptosis (22). PTEN has also been shown to interact directly with focal adhesion kinase and reduce its tyrosine phosphorylation (23).

In addition to frequent mutations in sporadic tumors, germ-line mutations of PTEN are believed to cause two related autosomal-dominant hamartoma syndromes: CS (OMIM 158350) (24, 25) and BZS (OMIM 153480) (16, 24, 26, 27, 28). Although each of the conditions is characterized by discrete clinical symptoms, the affected patients share high susceptibility for development of benign hamartomas throughout the body in early life(24, 26, 27). Furthermore, patients with CS have increased incidence of cancers of the breast, thyroid, and brain(25), whereas the incidence of cancers in BZS is not well documented (26). On the basis of the PTENmutation spectrum and phenotypic similarities between the two conditions, it has recently been suggested that PTENmutation-positive cases of CS and BZS should be considered distinct presentations of a single disease (26).

Mice null for pten die during embryogenesis between gestation day E6.5 and E9.5 (29, 30) from an apparent failure to form chorioallantoic fusion (29). Mice heterozygous for pten are highly susceptible to tumors. The predominant type of malignancies in pten^+/−mice at a young age is of lymphoid origin. Fifteen to 20% of all mice develop thymic and peripheral lymphomas, with infiltration into multiple organs and tissues (29, 30). Recently, a polyclonal autoimmune disorder, similar to the one that develops in fas-deficient mice has also been described in pten^+/− mice (31).

Interestingly, young pten^+/− mice fail to exhibit symptoms found in patients with CS or BZS. Here, we present evidence that, when older than six months, pten heterozygous mice develop a range of tumors, some of which represent hallmark features of PTEN-associated hamartoma syndromes. The majority of pten^+/− female mice develop breast and endometrial neoplasia, whereas males have increased incidence of neoplastic transformation of prostate epithelia. Furthermore, increased frequency of gastrointestinal tract hamartomas and adrenal gland neoplasia was observed. Significantly, a large proportion of tumors that develop in pten heterozygous mice are associated with LOH at the pten locus and manifest hyperphosphorylation of protein kinase B, directly implicating PTEN/PKB regulated pathway(s) in the development of these tumors.

After the mice were sacrificed, a full necropsy was performed on each animal. Organs were fixed in 10% buffered formaldehyde, and representative tissue from each organ was submitted for paraffin embedding and preparation of H&E-stained sections. Immunohistochemistry was performed on 3-μm sections using microwave antigen retrieval technique and standard peroxidase-antiperoxidase staining. The rabbit polyclonal antibodies against PKB and phosphorylated PKB (anti-P473)(New England Biolabs, Beverly, MA) were used at 1:200 and 1:50 dilutions, respectively.

Southern blot analysis was performed as described previously(29).

Western blot analysis was performed as described previously(18). Rabbit polyclonal serum was used at a 1:1000 dilution.

We and others have previously reported on the incidence of tumors in young pten^+/− mice and characterized their formation on both a histological and a molecular level (29, 30). To assess PTEN-related tumorigenesis in relation to age, a group of pten^+/− mice were monitored for tumor formation for up to 65 weeks of life. The investigated group included 65 female and 16 male PTEN^+/− mice in C57BI6/129Jv background. Fifty pten^+/− mice were sacrificed when they showed signs of illness between the ages of 26 and 65 weeks. The remaining mice, which showed no external signs of illness, were sacrificed between the ages of 60 and 65 weeks. The control group of wild-type mice, which included 12 females and 8 males, showed no signs of illness during the observation period. One wild-type female mouse was sacrificed at the age of 26 weeks, whereas the remaining control mice were sacrificed between 55 and 65 weeks of age. The tissues from all of the mice were fixed, and their histopathology was examined. The predominant pathology in pten^+/− mice at necropsy involved the breast, endometrium, gastrointestinal tract,prostate, adrenal glands, and lymphoid organs.

One half [32 of 65 (49%)] of all of the pten^+/− female mice developed breast tumors. Tumors ranged from 2 to 30 mm in diameter, and they occurred in thoracic as well as in abdominal/inguinal mammary glands. All of the primary tumors were relatively well differentiated. In contrast to normal breast histology (Fig. 1,A), tumors displayed prominent epithelial and stromal cell proliferation(Fig. 1,B). The epithelial cells formed acinar or tubular glands, but the proliferating spindle stromal cells often blended imperceptibly with the glandular epithelial cells (Fig. 1,C). The tumor cells displayed low-grade nuclear pleomorphism and relatively low mitotic counts but commonly showed infiltrating growth (Fig. 1,D). Some tumors with marked stromal component were reminiscent of carcinosarcoma. The majority of breast tumors in pten^+/− animals could be classified as Dunn’s adenocarcinoma type C (32), except for the small tumors(1–2 mm) with relatively uniform histology that were diagnosed as fibroadenoma (Fig. 1,E). Approximately one-third of the tumors demonstrated prominent papillary and papillomatous appearance consistent with Dunn’s type B adenocarcinoma (Fig. 1,F), but most of them also had discernible stromal proliferation. In view of the size of these tumors relative to the size of the mouse, they were considered as adenocarcinoma. Consistent with such a notion, despite the fibroadenoma-like appearance of the primary tumor, one animal had a metastatic tumor in a regional lymph node (Fig. 1,G), while three other mice had lung metastases (Fig. 1,H). The metastatic lesions retained a similar morphological appearance as the primary tumors. Only one mouse (26 weeks old) younger than 30 weeks developed a breast tumor. However, the incidence of breast neoplasia dramatically increased with age as 61% of mice between the ages of 30 and 49 weeks and 83% of animals between 50 and 65 weeks had tumors in this organ (Table 1).

The uterus of a wild-type female mouse was normal (Fig. 2,A). However, all of the pten^+/− female mice that were 26 weeks of age or older showed endometrial hyperplasia. Thirteen(20%) of sixty-five showed simple hyperplasia with an increased number of endometrial glands but without nuclear atypia (Fig. 2,B),whereas the remaining 80% of the animals exhibited various degrees of atypical hyperplasia. Thirty (46%) of 65 were classified as hyperplasia with low-grade atypia and had endometrial glands with scant intervening stroma and/or branching tubules (Fig. 2,C). The other 34% (22 of 65) of uteri displayed hyperplasia with high-grade atypia characterized by intraglandular epithelial cell proliferation that was papillary or cribriform in appearance (Fig. 2, D and E). Both focal and diffuse atypical endometrial hyperplasia were detected. High-grade atypical hyperplasia occurred more commonly in mice older than 30 weeks than in those 29 weeks or younger (45 versus 13%). Fourteen (22%) of 65 of the female mice developed endometrial carcinoma, with the earliest case found in a 30-week old mouse. Unlike the appearance of endometrial hyperplasia, the incidence of endometrial carcinoma in pten^+/− mice could not be correlated with increasing age (Table 1). The carcinomas were poorly differentiated and composed of sheets of large pleomorphic epithelial cells with central necrosis and an expanding growth pattern (Fig. 2 F). All of the carcinomas occurred in a background of diffuse atypical hyperplasia with high-grade atypia. Local or distant metastasis was not detected in any of the animals.

Among 16 male mice ranging in age from 26 to 65 weeks, one animal had prostatic adenocarcinoma (Fig. 3,A) characterized by the proliferation of dysplastic glandular epithelial cells associated with necrosis and infiltration into the surrounding stroma. This animal was 65 weeks old. Four other mice (40–65 weeks old) showed focal papillary or cribform proliferation of the prostatic epithelial cells with nuclear atypia, hence, resembling the histological appearance of high-grade prostatic intraepithelial neoplasia found in human prostate (Fig. 3 B). Two other animals were found to have focal hyperplasia but without atypia.

Nineteen (23%) of 81 mice (1 male and 18 females) showed marked expansion and proliferation of chromaffin cells of the adrenal medulla,consistent with medullary hyperplasia or pheochromocytoma. The tumors measured 3–15 mm in diameter. The smaller tumors showed atrophic compression of the cortex (Fig. 3,C), whereas the larger ones exhibited invasion through the capsule into the surrounding fat (Fig. 3 D). All of the tumors developed in mice older than 35 weeks. A majority developed unilaterally, but two animals showed bilateral lesions. One male mouse developed an adrenal cortical carcinoma with metastasis to the lungs.

For each studied animal, random segments of the bowel and stomach were also examined. Sections from 7 (9%) of 81 animals showed hamartomatous lesions involving the mucosa of stomach, duodenum, small bowel, colon,or anorectal junction. These lesions were generally composed of a polypoid proliferation of the mucosal epithelium and stroma, associated with irregular microcystic dilatation of the glands (Fig. 3, E and F). Gastrointestinal lesions in pten^+/− mice were reminiscent of the juvenile polyps of the human gastrointestinal tracts. None of the polyps showed evidence of invasive growth into the muscularis mucosa or submucosa,although some of them demonstrated focal dysplastic changes of glandular epithelial cells. Many animals also displayed enlarged Peyer’s patches composed of lymphohistiocytic infiltrates.

As seen in the younger mice and previously reported (29, 30), older pten^+/− mice also developed a range of lymphomas during their lifetime. A total of 43 (53%) mice had enlarged cervical, mediastinal, or abdominal lymph nodes and/or thymus. In many mice, splenomegaly was also apparent. Histologically, lymph nodes and thymi demonstrated an effacement of normal corticomedullary architecture by mixed inflammatory cell infiltrates that included atypical lymphocytes. Fifteen mice (18%) developed malignant lymphoma,characterized by a monotonous infiltrate composed of small/large cleaved lymphocytes and lymphomatous infiltration of solid organs including kidney, lung, and liver. Lymphoma occurred in mice of all ages and with similar frequency in both males and females.

To investigate whether somatic mutations of the ptenwild-type allele contribute to the development of tumors in pten^+/− mice, total genomic DNA from representative endometrial, breast, and prostate tumors was isolated and subjected to Southern blot analysis together with tail DNA from a pten^+/− mouse as control, using a probe that distinguishes between the wild-type and the mutant allele (Fig. 4,A) (29). A number of examined tumors displayed either a complete loss of the wild-type allele or a reduction of its intensity. Residual hybridization in these instances most likely resulted from contamination of tumor DNA with that from the surrounding normal tissue. Relative intensities of radioactive signals corresponding to the mutant and wild-type band, respectively, were quantified, and mutant:wild-type ratio greater than 2:1 scored as LOH for pten in a given tumor (Fig. 4 B). On the basis of such criteria, nearly all of the investigated tumors displayed LOH at the pten locus, which indicated an importance for loss of PTEN function in the formation of these neoplasms.

Our previous work, as well as that of others (18, 21, 29, 33), has implicated PKB/Akt hyperphosphorylation and activation in the development of PTEN-related tumors. In order to investigate the role of PKB/Akt in the etiology of tumors in pten^+/− animals, immunohistochemistry for native and phosphorylated form of PKB, respectively, were performed on representative adjacent sections of the uterine lesions and breast tumors. The antibody to native (unphosphorylated) PKB detects immunoreactivity in the breast ductal epithelium (Fig. 5,A) as well as in normal glandular epithelial and stromal cells of the uterus (Fig. 5,B). In contrast, the antibody to phosphorylated PKB only stained the breast (Fig. 5,C) and uterine (Fig. 5,D) carcinoma cells, as well as the hyperplastic endometrial glands (not shown). Consistent with the immunochemical data, Western blot analysis of tumors also revealed hyperphosphorylation of PKB in tumors that have lost expression of PTEN(Fig. 5 E). Thus, in addition to causing lymphomas(34), activation of PKB/Akt is also implicated in the development of breast and endometrial neoplasia in mice.

The development of mouse models for studying cancer predisposition syndromes has had limited success thus far. Here, we demonstrate that mice heterozygous for the pten gene are affected by neoplasia of multiple tissues resembling the tumors that develop in patients with PTEN hamartoma-tumor syndromes, CS and BZS. Endometrial and breast hyperplasia and breast fibroadenoma, as well as hamartomatous intestinal polyps, identified in pten^+/− mice closely resemble the early manifestations of CS (25, 35). Furthermore, breast and endometrial carcinoma seen in pten^+/− mice are often found in Cowden’s kindreds later in life (25, 27, 36).⁴Mutations in tumor suppressor genes are generally recessive in nature. In patients suffering from familial cancer predisposition syndromes, a mutation of a tumor suppressor gene is transmitted through the germ line, whereas LOH at the second allele is thought to be a crucial step in tumor formation. Significantly, LOH at the pten locus is frequently identified in tumors arising in pten^+/− mice, which emphasizes the necessity of full PTEN inactivation for tumor formation. Interestingly, thyroid,skin, and central nervous system hamartomas, often found in carriers of PTEN mutations, could not be detected in our mice. Thus, pten^+/− mice represent a model system for studying certain aspects of the PTEN-associated hamartoma syndromes in the laboratory.

In addition to the association of PTEN mutations with breast neoplasia as part of CS, reduction or complete loss of PTEN expression was detected in 33% of sporadic breast carcinoma (37). Here, we show that 49% of pten^+/− mice develop breast tumors, with most of them having characteristics of well-differentiated adenocarcinoma. Similar to breast tumors in CS patients (36), breast lesions in pten^+/− mice displayed prominent stromal proliferation. We noted an increase of breast tumor incidence with age in pten^+/− mice, potentially indicating a requirement for the accumulation of additional mutations for tumor progression in this organ. It would be informative to determine whether the incidence and the time of appearance of tumors in pten^+/− mice can be accelerated by different genetic backgrounds.

Analysis in human tumors have implicated PTEN mutations in the advanced stages of the disease. However, mutations of PTEN can be found at all stages of endometrial neoplasia in humans (13, 14, 15). Similarly, we have identified all of the stages of endometrial hyperplasia in pten^+/−mice, accompanied by LOH at the pten locus. Thus, the endometrial carcinogenesis in pten^+/− mice mimicks the endometrial neoplasia that develops in women with unopposed estrogen stimulation. Similar to the breast tumors, incidence of endometrial hyperplasia in pten^+/− animals increased with age. Even though all endometrial carcinomas in pten^+/− mice arise in a background of atypical hyperplasia, correlation between their formation and the age of the animal is not apparent, further lending support to an early role of PTEN mutations in endometrial tumorigenesis.

Unlike the humans with germ-line mutations of PTEN,pten^+/− mice form tumors of prostate, adrenal gland,and lymphoid tissues. Significantly, even though prostate tumors are not a clinical component of PTEN-associated hamartoma syndromes, PTEN mutations are frequently found in both human prostate cancer samples and cell lines (10, 38, 39). In view of the fact that the incidence of prostate cancer increases with age in humans as well as in our mice, a study of a larger group of pten^+/− mice is needed to establish these mice as a model system for this type of neoplasia. In addition, it would be informative to systematically examine the possible involvement of PTEN mutations in lymphoid and adrenal neoplasia in humans.

We and others have previously demonstrated the ability of PTEN to negatively influence cellular-survival signaling via regulation of PKB/Akt, and we have implicated the deregulation of this kinase as a feature of pten-associated lymphoid tumors (18, 29, 40). Here, we show hyperphosphorylation of PKB/Akt in pten-associated endometrial and breast tumors. Thus, in situ assessment of PKB/Akt phosphorylation might represent a good indicator of loss of PTEN function, by both genetic and epigenetic means and could potentially be used for diagnostic purposes. Significantly, in humans, amplification of PKB/Akt has been documented in breast, prostate, ovarian, and pancreatic cancers and cell lines (41, 42, 43, 44).

Certain similarities as well as differences between our pten^+/− mice and those generated by other groups exist. The predominant pathological features in our pten^+/− mice include breast, endometrial,prostate, adrenal, gastrointestinal, and lymphoid tissues. The high incidence of breast hyperplasia and breast carcinoma seems to be exclusive to our mice. Moreover, formation of endometrial carcinoma, as well as the complete penetrance of endometrial disease in pten^+/− mice, together with the LOH at the pten locus in both breast and endometrial lesions have not been previously described. Both our mice and those generated by Podsypanina et al. (30) exhibit lymphoproliferation, development of lymphomas, appearance of benign polyps in the intestine as well as endometrial and prostate hyperplasia. (29, 30), (this report). Recently, an autoimmune disorder, similar to the one seen in fas-deficient mice has been reported in ptenheterozygous mice (31). In view of the fact that many of the histological features of lymphoproliferation resemble those of autoimmunity, additional investigations are needed to determine whether our mice suffer from a similar autoimmune condition. Given the similarity between the genetic backgrounds utilized by different groups, it is unlikely that a difference in modifying alleles might be a reason for the discrepancy in observed phenotypes. Even though the gene-targeting strategies between the groups differ, all of the three mutations appear to be null. An alternative explanation is that the severity of the reported phenotypes could be different. Whereas all of the tumors in our mice and the lymphoid tumors described by Posypanina et al. (30) are associated with LOH at the pten locus, the defects described by Di Cristofano et al. (31) appear in a haploinsufficient background.

High penetrance of breast and endometrial neoplasia in pten^+/− mice described here and their phenotypic correlation with tumors arising in humans suffering from CS and BZS, together with their predictability and high incidence, offer a unique opportunity to further study the role of PTEN in the formation of these tumors. Strategies aimed at evaluating the influence of specific genetic backgrounds, as well as the effect of environmental factors and humoral signals on tumorigenesis in pten^+/− mice, should provide further insight into the function of PTEN in tumor suppression and carcinogenesis.

We would like to thank James Ho for help with histology and Malte Peters for critical reading of the manuscript.