Genetic prevention of lymphoma in p53 knockout mice allows the early development of p53-related sarcomas.

Homozygous knockout of p53 in mice leads to early mortality from lymphoma, with almost complete penetrance, thus hampering studies of other tumor histotypes related to p53 alterations. To avoid lymphoma development, we crossed p53 knockout mice (BALB-p53 mice) with alymphocytic BALB/c Rag2-/-;Il2rg-/- (RGKO) mice. We compared the tumor spectrum of homozygous (BALB-p53-/-) and heterozygous (BALB-p53+/-) mice with alymphocytic mice (RGKO-p53-/- and RGKO-p53+/-). Lymphoma incidence in BALB-p53-/- mice exceeded 80%, whereas in RGKO-p53-/- it was strongly reduced. The prevalent tumor of RGKO-p53-/- mice was hemangiosarcoma (incidence over 65% in both sexes, mean latency 18 weeks), other tumors included soft tissue sarcomas (incidence ~10%), lung and mammary carcinomas. Tumor spectrum changes occurred also in p53 heterozygotes, in which lymphomas are relatively rare (~20%). RGKO-p53+/- had an increased incidence of hemangiosarcomas, reaching ~30%, and females had an increased incidence of osteosarcomas, reaching ~20%. Osteosarcomas shared with the corresponding human tumors the involvement of limbs and a high metastatic ability, mainly to the lungs. Specific alterations in the expression of p53-related genes (p16Ink4a, p19Arf, p15Ink4b, p21Cip1) were observed. Genetic prevention of lymphoma in p53 knockout mice led to new models of sarcoma development, available for studies on hemangiosarcoma and osteosarcoma onset and metastatization.


INTRODUCTION
The expression of the tumor suppressor gene p53 is altered by mutation and other mechanisms in a large proportion and in a wide variety of human cancers [1,2].
p53 knockout mice are a precious tool to investigate the mechanisms of carcinogenesis in the absence of p53. However, p53 knockout mice are prone to the early development of lymphomas, in particular of thymic origin [3][4][5], hence the study of other solid tumors is greatly hampered in these mice. This is a general problem in mice harboring cancer syndromes. Early onset and lethality of a given tumor type can preempt the study of other tumors arising later in life. It is a problem mostly unsolvable, even though empirical solutions can be found in some instances by changing the genetic background, or through direct or indirect modifications of the gene of interest, for example using tissue-specific Cre-lox systems [4].
In the case of p53, variations in tumor spectrum were obtained either using mutant p53 genes with gain of function properties [6][7][8][9] or through conditional www.impactjournals.com/oncotarget inactivation of p53 in specific tissues [4,[10][11][12][13][14]. Gain of function mutants are certainly informative for what concerns the corresponding human mutations, but not for those conditions, such as the Li-Fraumeni syndrome, in which biallelic losses are frequent lesions [4]. On the other hand, inactivation of p53 in prespecified tissues leaves open the question of which tumors could result from an organism-wide lack of p53.
For what concerns the predominant lymphoma of p53 knockout mice, the cell population at risk of neoplastic transformation, i.e. lymphocytes, is not indispensable for life, at least under sterile conditions, and various alymphocytic mouse lines were obtained through genetic manipulation. Therefore, we reasoned that cross-breeding of p53 knockout mice with alymphocytic mice could avoid lymphoma development. We show here that p53 knockout mice crossed with Rag2 −/− ;Il2rg −/− mice (RGKO-p53 −/− mice), which lack T, B and NK cells [15], had a strong impairment of lymphoma development, thus allowing the study of solid tumors, mainly sarcomas, caused by the lack of p53.
All p53 −/− mice developed progressive tumors by 30 weeks of age ( Fig. 2A). Tumor onset was slightly, but significantly, delayed in RGKO-p53 mice: BALB-p53 −/− mice had a median tumor-free survival time of 16 weeks, while RGKO-p53 −/− mice had a median of 19.5 weeks, however, after 20 weeks of age, the tumor-free survival curves overlapped.

Prevention of lymphoma development
More then three quarters of BALB-p53 −/− mice developed lymphomas, mostly of thymic origin. A minority of mice developed sarcomas or carcinomas (Fig. 2B), frequently in combination with lymphoid tumors.
The incidence of lymphomas in RGKO-p53 −/− mice was strongly reduced and became a sporadic event, possibly resulting from the neoplastic transformation of immature precursors (see below).
Tumor-free survival curves of individual tumor histotypes showed that lymphoma-free survival and hemangiosarcoma-free survival differed significantly between BALB-p53 −/− and RGKO-p53 −/− , the former developing early lymphomas and only few hemangiosarcomas and the latter showing the opposite (Fig. 6A). Tumor-free survival curves for osteosarcoma, soft tissue sarcoma and mammary carcinoma did not differ significantly between the two strains.
Interestingly, in RGKO-p53 −/− mice, the global incidence of carcinomas (34%) was significantly lower than that of sarcomas (82%), thus suggesting that neoplastic transformation of epithelial cells was slower than that of mesenchymal cells, or required additional somatic events.
Finally we addressed the question of the origin of lymphomas in RGKO-p53 −/− mice. It must be kept in mind that these "alymphocytic" mice actually lack lymphocytes with a rearranged antigen receptor, but bone marrow lymphopoiesis and migration proceed normally. The rare lymphomas of RGKO-p53 −/− mice arose predominantly in the thymus. One case studied by flow cytometry consisted of blast cells with large forward and side scatter and an immature single-positive (ISP)-like phenotype (CD45R/ B220 dim , CD3 − , CD4 − , CD8 + ) [16][17][18][19]. The residual incidence of lymphomas in RGKO-p53 −/− mice suggests that lymphomas of p53 −/− mice should be re-examined to assess the role of p53 before and after rearrangement.
Female RGKO-p53 +/− showed a distinctive increase, up to 20%, in the incidence of osteosarcoma, which became the third most common tumor in these mice (Fig. 7B). Osteosarcomas (Fig. 3 D-F) mainly affected the limbs (90% of cases) and gave rise to spontaneous metastatic dissemination, in particular to the lungs (25% of cases).

Hemangiosarcoma and osteosarcoma cell lines
To further investigate the nature of tumors arising in RGKO-p53 −/− mice, we derived in vitro cultures and established stable cell lines from hemangiosarcomas and osteosarcomas. Hemangiosarcoma cell lines, which in vitro expressed typical endothelial markers, such as VEGF-R2, CD31 and CD146, were poorly tumorigenic, giving rise to local tumors only after very long latency times (Tab. 1).
In contrast, osteosarcoma cell lines rapidly gave rise to ossified local tumors (Tab. 1). In both cases, growth in immunocompetent mice was less efficient and slower than in immunodeficient mice, thus indicating immune recognition of cells deriving from tumors arisen in immunodeficient hosts. Distant metastases to the lungs were found in a sizeable proportion (>40%) of mice bearing local tumors, in particular in immunodeficient mice. Intravenous injections of tumor cells confirmed the faster growth and the higher malignancy of osteosarcoma cells (Tab. 2).

Expression of p53 and p53-related tumor suppressor genes
We analyzed the expression of p53 in normal tissues, tumors and cell lines obtained from mice carrying different p53 genotypes. As expected, p53 was absent in all samples from p53 −/− mice (Fig. 8). It is worth noting that osteosarcomas of p53 +/mice retained p53 heterozygosity, which was instead lost in cell lines, a phenomenon known to occur also in human tumors [20].

DISCUSSION
Carcinogenesis in p53 knockout mice recapitulates the natural history of tumor onset in humans, starting with lymphoid malignancies, followed by sarcomas, then by carcinomas. However, in p53 knockout mice, the high prevalence and fast growth of lymphomas obscured all other tumor histotypes.
Thank to the fact that life without mature lymphocytes is possible, if the host is not exposed to deadly infections, we could strongly reduce the onset of lymphomas in p53 knockout mice by crossing them with alymphocytic Rag2 −/− ;Il2rg −/− mice. This revealed a tumor landscape dominated by soft tissue sarcomas, in particular of vascular origin. It is interesting to note that the global incidence of carcinomas remained far below the incidence of sarcomas, thus suggesting that, as sarcomas were masked by lymphomas, now carcinomas are masked by sarcomas. If it were possible to genetically delete the mesenchymal cells that give rise to sarcomas, as we did with lymphoma precursors, all mice would probably develop carcinomas at a later age. It is worth mentioning that in p53 knockout mice a shift from lymphomas to sarcomas, and an increase in survival was obtained also by treatment with rapamycin the inhibitor of mTOR (mammalian Target Of Rapanycin) [29]. The activity of    the mTOR pathway, usually inhibited by p53, seems to be increased in some tissues of p53 −/− mice with a tendency to elevated IGF-1 levels, but the effects of absence of p53 can be partially attenuated by rapamycin, leading to changes in tumor spectrum [29,30].
The tumor spectrum of p53 +/− heterozygotes is completely different from that of p53 −/− homozygotes. Firstly, tumors arise much later in the life of heterozygotes. As these tumors usually show loss of heterozygosity at p53, the longer latency can be equaled to the time required for the loss of the wild-type p53 allele. Secondly, carcinomas dominate the tumor spectrum of heterozygotes. There could be several explanations for this fact, however the simplest one connects long latency with tumor histotype: loss of p53 heterozygosity occurs when mesenchymal cell populations are mostly quiescent, after the burst of proliferation required for growth to adult size, thus neoplastic transformation to sarcoma becomes relatively rare. For what concerns the tumor spectrum modifications introduced by the alymphocytic Rag2 -/-;Il2rg -/genotype, as in p53 −/− mice, also in p53 +/− mice, this resulted in a reduction of lymphomas in favor of sarcomas.
The sizeable incidence of osteosarcomas in female RGKO-p53 +/− mice makes an interesting model system for the study of bone tumors. It is worth noting that epidemiological studies of human osteosarcoma show a typical bimodal age distribution of tumor onset, with a first peak in early adolescence and a second peak in adults over the age of 65. Osteosarcoma among the elderly is frequently related to Paget's disease or represents a second malignancy, in some instances caused by previous radiation therapy for the first tumor. The incidence of osteosarcoma generally is higher in males than in females, but in the case of osteosarcoma as a second malignancy some studies reported a slightly higher frequency in females [31]. The increased incidence of osteosarcoma in immunodeficient female p53 +/− mice could reflect the higher incidence of osteosarcoma as second tumor in women and, when compared to the lower incidence in female immunocompetent p53 +/− mice, could suggest a role for the immune response.
RGKO mice lack adaptive immune responses and innate responses are severely impaired, not only because they lack NK cells, but also because all the cytokines secreted by the various lymphoid subpopulations are absent. The fact that the profound immune deficiency of RGKO mice caused only minor variations in tumor latency of p53 knockout mice is a testament to the potency of the carcinogenic risk (or better, carcinogenic certainty) triggered by p53 alterations. In fact, the only detectable variations ascribable to the lack of immune surveillance were the slightly faster onset of tumors, in particular sarcomas, in RGKO-p53 +/− mice and the slower growth of tumor cells derived from the same tumors in immunocompetent versus immunodeficient hosts.
The latter result could be attributed to the immunogenicity of tumors arising in the absence of immunoediting, i.e. the negative selection of immunogenic cell variants actuated by the immune system in immunocompetent mice [32].
RGKO-p53 −/− mice are a novel model system for the preclinical development of targeted therapies of human cancer with p53 alterations, which represent a prevalent molecular alteration of human tumors in need of novel targeted therapeutic approaches. A challenging set of new molecular targets can derive directly from the lack of p53, which frequently results in the derangement both of upstream p53 regulators and of downstream p53-regulated genes [22,[33][34][35][36][37][38][39]. In fact, various tumor suppressor genes, such as p21Cip1, p16Ink4a, p15Ink4b and p19Arf were up-regulated in sarcomas of RGKO-p53 mice, thus offering novel opportunities for the development of synthetic lethal therapeutic approaches.
In conclusion, we have shown that it is possible to modify the spectrum of tumors of p53 knockout mice which, in the absence of lymphocytes, develop a variety of solid tumors, in particular soft tissue sarcomas and osteosarcomas. The alymphocytic p53 knockout mice are a novel model system for studies of sarcoma development and for preclinical studies of targeted therapies.

Mice
BALB/c p53 +/mice (BALB/cJ-Trp53tm1Tyj, originally obtained from The Jackson Laboratory, Bar Harbor, MI) have one p53 allele inactivated by the insertion of a neomycin resistance cassette in the region spanning exons 2 to 6 [5]. Rag2 -/-;Il2rg -/breeders were kindly given by Drs. T. Nomura and M. Ito of the Central Institute for Experimental Animals (Kawasaki, Japan) [40,41]. Mice were bred in our animal facilities under sterile conditions. BALB-p53 +/− mice were crossed with BALB Rag2 -/-;Il2rg -/and the offspring were genotyped by PCR analysis of genomic DNA (see below for PCR conditions and primers). Selected progeny was used for subsequent crossings. Starting with the F2 generation, mice carrying genotypes of interest were weekly inspected for early signs of tumor onset (dyspnea, weight loss or suffering) in addition to meticulous palpation of limbs, mammary glands and peritoneum internal organs. Mice were sacrificed according to the criteria for humane endpoints approved by the Institutional Animal Care and Use Committee of the University of Bologna. At sacrifice, an accurate necropsy was performed. Brain, lung, heart, liver, adrenals, kidneys, spleen and lymph nodes, genitourinary tract, salivary glands, subcutis, muscles, mammary glands and limbs were macroscopically inspected and samples were collected of all visible tumors. Part of each tumor was fixed in 10% formalin, when necessary decalcified, and embedded in paraffin. The remaining tumor was divided into small samples and snap frozen. Two-three-micrometer paraffin sections were stained with hematoxylin and eosin and evaluated by an experienced pathologist.

Genotyping and real-time polymerase chain reaction
DNA for genotyping was isolated by digestion of tail tissue in a buffer containing proteinase K. To distinguish between the wild-type and mutant allele, two sets of primers were used for each gene: Wild-type Il2rg gene, forward CTG CTC AGA ATG CCT CCA ATT CC, reverse GAT CCA GAT TGC CAA GGT GAG TAG; mutated IL2rg gene, same forward primer, reverse CCT GCG TGC AAT CCA TCT TGT TCA AT (neo cassette) [42]. Wild-type RAG2 gene, forward GGG AGG ACA CTC ACT TGC CAG TA, reverse AGT CAC TCT ACC TCT GAG GAC TGA; mutated RAG2 gene, same reverse primer and forward CGG CCG GAG AAC CTG CGT GCA A (neo cassette) [43,44]. Wild-Type Trp53 gene, forward CCC GAG TAT CTG GAA GAC AG, reverse TAC TTG GCG GCT GGA TA; mutated Trp53 gene same reverse primer and forward CTT GGG TGG AGA GGC TAT TC (neo cassette).