p21Cip1 is a cyclin-dependent kinase inhibitor whose abundance increases in cells exposed to radiation or other DNA-damaging agents. Such increases activate a G1 checkpoint, which allows time for DNA repair before S phase entry. By inhibiting cell cycle progression, p21Cip1 potentially suppresses tumorigenesis, and in support, we show that p21Cip1 heterozygous and nullizygous mice develop more tumors than do wild-type mice when exposed to a single dose of γ-irradiation. Importantly, we also show that p21Cip1 nullizygosity increases the incidence of metastatic tumors in irradiated mice. We suggest that p21Cip1 is haploinsufficient for tumor suppression and functions as an antimetastatic agent.
Although the p21Cip1 gene is rarely mutated in human tumors (1), p21Cip1 is expressed at subnormal amounts or is dysfunctional in some human tumors. Cancer-related events that reduce p21Cip1 expression include the functional inactivation of the tumor suppressor and transcription factor p53, an event that occurs frequently in human tumors (2, 3, 4), and the hypermethylation and consequent silencing of the p21Cip1 promoter, which occurs in leukemic bone marrow cells (5). The Her-2/neu gene is often amplified in breast cancers (6), and events downstream of Her-2/neu may limit the accessibility of p21Cip1 to cyclin-dependent kinases (7). Importantly, inefficient expression of p21Cip1 is indicative of short survival times for patients with acute lymphoblastic leukemia or colorectal cancer (5, 8). Although p21Cip1-null mice do not spontaneously develop tumors, at least ≤1.5 years of age (9, 10), they are often more susceptible to tumor formation initiated by oncoproteins, carcinogens, or loss of tumor suppressors (11, 12, 13, 14). However, the role of p21Cip1 in tumor suppression remains unclear (15, 16, 17), and our studies examined in detail the effects of p21Cip1 gene dosage on tumor formation in irradiated mice. We show that p21Cip1 heterozygosity and nullizygosity increase tumor incidence in irradiated mice, whereas p21Cip1 nullizygosity (but not heterozygosity) increases tumor metastasis.
Materials and Methods
Mice and Treatment.
Mice with inactivating targeted mutations of both p21Cip1 alleles (obtained from Dr. Tyler Jacks, Massachusetts Institute of Technology; Ref. 18) were crossed with wild-type mice (The Jackson Laboratory, Bar Harbor, ME), which have a comparable mixed C57BL/6J × 129S1/SvImJ (B6;129S) genetic background. p21+/− progeny were sibling mated to produce p21+/+, p21+/−, and p21−/− mice. Genotyping was performed as described previously (11). At 14 ± 2 days of age, mice received a single 4-Gy dose of whole body γ-irradiation from a 137Cs source. After irradiation, mice were monitored daily by visual inspection and sacrificed when tumors of ≥1 cm were noted, when morbidity was noted, or at 14 months of age. Euthanasia was performed in accordance with the procedures outlined in the 2000 Report of the American Veterinary Medical Association Panel on Euthanasia. All experimental and surgical procedures performed on mice were in accordance with the guidelines of the University of South Florida and the NIH guidelines outlined in the “Guide for Care and Use of Laboratory Animals” (NIH publication 85-23).
Necropsies were done in the following stepwise manner. The skin, subcutis, skeletal muscle, inguinal lymph node and mammary fat pad, cervical lymph node, and salivary glands were inspected visually for tumors or other abnormalities and collected. The entire alimentary tract, including the esophagus, stomach, small intestine, cecum, and colon, was inflated with 10% neutral-buffered formalin and rolled in segments to fit into cassettes for histological processing. The reproductive tract and associated glands and organs, urinary bladder, spleen, pancreas, kidneys, adrenal glands, liver, and gall bladder were evaluated and collected. The larynx, trachea with attached thyroid and parathyroid glands, and lungs were inflated with 10% neutral-buffered formalin, fixed with the heart and thymus, and trimmed and placed in cassettes for processing. The brain cut in three cross-sections, portions of the spinal column in lateral and longitudinal sections, pituitary gland, eyes, and Harderian gland were inspected and collected. Tumors and tissues were fixed in 10% neutral-buffered formalin, dehydrated, embedded in paraffin, sectioned at 3 μm, and stained with H&E. Histological sections were masked and interpreted independently by R. W. E. and D. C. Discrepancies were resolved by consultation with other pathologists. Over 3200 histological preparations were examined. Staining of selected tumor sections for p21Cip1 expression was done as described previously (13).
Statistical comparisons of proportions in all three groups and between pairs of groups were based on Fisher’s exact test. Numbers of tumors per animal were compared using the Kruskal-Wallis Test, a nonparametric analogue of the pooled t test, and one-way ANOVA. Survival rates over time were compared using the Log-rank test. All reported Ps are two sided. All statistical analyses were performed using SAS (version 8.2). Ps < 0.05 are considered significant, and Ps between 0.05 and 0.1 are considered suggestive.
Results and Discussion
Approximately 200 mice (p21+/+, p21+/−, and p21−/− genotypes) received a 4-Gy dose of whole body γ-irradiation at 2 weeks of age. Irradiated mice were monitored daily and necropsied when signs of illness or palpable tumors were seen. Mice that did not develop overt symptoms of disease were considered “symptom free” and necropsied 13.5 months after γ-irradiation. Necropsies were done on 31 p21+/+ mice, 105 p21+/− mice, and 61 p21−/− mice, and all organs were inspected macroscopically and microscopically. p21Cip1 gene dosage had no significant effect on the symptom-free survival of γ-irradiated mice (data not shown). All mice were symptom free at 4 months of age, and 70% of all mice were symptom free at 14 months of age. However, as determined by necropsy and histological analysis, most mice were tumor bearing. The percentages of p21+/+, p21+/−, and p21−/− mice with tumors were 68, 79, and 80%, respectively, and not significantly different. On the other hand, p21+/− and p21−/− mice had significantly more tumors per mouse than did p21+/+ mice (P = 0.0277 in a three-way comparison; Table 1). Average numbers of tumors per mouse were 2.23 for p21+/− mice (P = 0.016) and 2.22 for p21−/− mice (P = 0.01), as compared with 1.1 for p21+/+ mice. Moreover, 52% of p21+/− mice (P = 0.0032) and 54% of p21−/− mice (P = 0.0045) had more than one tumor per mouse, as compared with only 23% of p21+/+ mice. Immunohistochemistry confirmed the presence of p21Cip1 in tumors of p21+/− mice (Fig. 1). These findings suggest that the p21Cip1 gene is haploinsufficient for tumor suppression in γ-irradiated mice.
Tumors were classified as benign or malignant based on morphology. Tumors were considered benign if they contained few mitotic figures, lacked cellular pleomorphisms, were localized and well demarcated, and consisted of expansible masses of well-differentiated cells. Tumors were considered malignant if they had infiltrating advancing margins, large areas of necrosis, numerous and often atypical mitotic figures, and pleomorphic cells. p21+/− and p21−/− mice developed 2.2-fold more benign tumors (P = 0.0299 in a three-way comparison) and 1.7-fold more malignant tumors per mouse than did p21+/+ mice (Table 1). The latter difference was not significant perhaps because of small sample numbers. However, p21+/− mice (11%; P = 0.0934) and p21−/− mice (10%; P = 0.0687) developed multiple malignant tumors per mouse (in separate organ systems), whereas p21+/+ mice did not. These findings show that a greater incidence of benign and perhaps also malignant tumors accounts for the greater incidence of total tumors in p21Cip1 heterozygous and nullizygous mice, as compared with wild-type mice.
All three groups of irradiated mice developed tumors in several organs, most notably the Harderian gland, lung, ovary, and small intestine (Table 1). p21+/− and p21−/− mice developed significantly more Harderian gland and ovarian tumors per mouse than did p21+/+ mice (see Ps in the legend to Table 1). Moreover, the spectrum of tumor types was broader in p21+/− and p21−/− mice than in p21+/+ mice. Tumor types present in p21+/− and p21−/− mice but not in p21+/+ mice include adrenal gland pheochromocytomas, Harderian gland adenocarcinomas, hepatocellular carcinomas, and ovarian tubulostromal adenomas and granulosa cell tumors. Other than thymic lymphomas, which were present in all three cohorts, p21+/+ mice did not develop tumors of mesenchymal cell origin. In contrast, p21+/− and p21−/− mice developed a variety of mesenchymal tumors, including hemangiomas, hemangiosarcomas, and leiomyosarcomas. Thus, loss of one or both copies of the p21Cip1 gene increases the incidence of Harderian gland and ovarian tumors and the diversity of benign and malignant tumors in irradiated mice.
Despite similar numbers of malignant tumors per mouse, p21Cip1 nullizygous mice developed 2.7-fold more metastatic tumors per mouse than did p21Cip1 heterozygous mice (Fig. 2,A and Table 2, P = 0.018). Numbers of metastatic tumors per mouse were 0.1, 0.09, and 0.24 for p21+/+, p21+/−, and p21−/− mice, respectively (P = 0.0366 in a three-way comparison). Thymic lymphomas, which invariably were disseminated, accounted for all of the metastatic tumors in p21+/+ mice and >50% of the metastatic tumors in p21+/− mice. On the other hand, 70% of the metastatic tumors in p21−/− mice arose from organs other than the thymus; e.g., the Harderian gland, ovary, and spleen. Examples of distant metastases in p21Cip1-null mice are shown in Fig. 2 B–D. When thymic lymphomas were excluded from the analysis, the numbers of metastatic tumors per mouse were 0.18 for p21−/− mice, as compared with zero for p21+/+ mice (P = 0.0078) and 0.04 for p21+/− mice (P = 0.0147; P = 0.0035 in a three-way comparison). Moreover, 26% of nonlymphoma malignant tumors in p21−/− mice metastasized as compared with 0% in p21+/+ mice and 6% in p21+/− mice (P = 0.0086 in a three-way comparison). These findings clearly show that deletion of both copies of the p21Cip1 gene promotes tumor metastasis in γ-irradiated mice.
Our study provides the most in-depth analysis to date of the combined effects of p21Cip1 gene dosage and radiation on whole animals. We show that p21+/− and p21−/− mice develop more tumors when irradiated than do p21+/+ mice. A greater incidence of Harderian gland and ovarian tumors and a wider spectrum of tumor types contributed to the overall increase in tumor incidence in p21+/− and p21−/− mice, as compared with p21+/+ mice. Because p21Cip1 was expressed in a large sampling of tumors from p21+/− mice, we suggest that the p21Cip1 gene is haploinsufficient for tumor suppression in irradiated mice. Our results are in accordance with previous reports showing enhanced susceptibility of p21Cip1-null mice to chemically induced carcinogenesis (11, 12) and haploinsufficient tumor suppression by p21Cip1 in mice expressing a mutant allele of the APC tumor suppressor gene (14) or the v-Ras oncoprotein (13). Our results differ, however, from those of Martin-Caballero et al. (16), who observed a protective effect of p21Cip1 nullizygosity on the survival of γ-irradiated mice. In the study of Martin-Caballero et al. (16), 35 mice received a 1.75-Gy dose of radiation once a week for 4 weeks. As a result, all of the wild-type mice developed T-cell lymphomas and did not survive >10 months. In contrast, in our study, ∼200 mice received a single 4-Gy dose of radiation, and most mice were lymphoma free and viable at 14 months. Thus, our study allowed us to detect a wide variety of tumor types in irradiated p21+/+, p21+/−, and p21−/− mice and to accurately assess tumor burden.
The most intriguing finding of our study is the increased tendency of p21−/− tumors to metastasize, as compared with p21+/+ and p21+/− tumors. Invasive, malignant tumors give rise to metastatic tumors by a multistep process in which tumor subclones detach from the primary tumor, invade surrounding tissue, disseminate through the bloodstream to distant organs, and grow and vascularize in new locations (19, 20). The inability of irradiated p21Cip1-null cells to arrest in G1 and the consequent replication of damaged DNA may result in the aberrant expression of gene products that induce or inhibit metastasis. Alternatively, loss of p21Cip1 and consequent dysregulated cyclin-dependent kinase activity and cell cycle progression may directly facilitate the growth of new tumors at sites of metastasis. Determination of how p21Cip1 suppresses metastasis represents an exciting challenge and may facilitate the design of efficacious anticancer strategies.
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.
Supported by the Cortner-Couch Endowed Chair for Cancer Research (to W. J. P.), NIH Grants CA78038 and CA67360 (to W. J. P), and grants from the American Cancer Society (to R. J. J. and D. C.) and the Eleanor Naylor Dana Charitable Trust (to R. W. E.).
a For the ovary, uterus, and vagina, tumors per mouse was determined by dividing tumor number by the number of female mice: 15, 48, and 32 for the p21+/+, p21+/−, and p21−/− cohorts, respectively.
b For the prepucial gland, seminal vesicle, and testes, tumors per mouse was determined by dividing tumor number by the number of male mice: 16, 57, and 29 for the p21+/+, p21+/−, and p21−/− cohorts, respectively.
We thank Sandra Livingston for assistance with histopathology, Nancy Olashaw for manuscript preparation, and Anita Parker and Janelle Simon of the Vivarium Core Facility at the Moffitt Cancer Center. We also thank the helpful service of the Molecular Imaging Core at the Moffitt Cancer Center.