Insights into Wild-Type and Mutant p53 Functions Provided by Genetically Engineered Mice

Germline Mutant Trp53 Transgenes

A number of mutant p53 transgenic lines have been generated using zygote injection methods in which multiple copies of a mutant Trp53 transgene were introduced into the mouse germline. In these models, the randomly integrated mutant transgenes are expressed either globally or in a tissue-specific manner in the presence of the two endogenous wild-type Trp53 alleles. Thus, tumors arising in these mice result in part from dominant-negative suppression of the wild-type p53 protein forms by overexpressed mutant p53 protein. The first genetically engineered p53 mutant models were of this type, in which globally expressed mutant p53 transgenes (originally isolated from murine cancers) resulted in an array of different tumor types, including osteosarcomas, lung adenocarcinomas, and lymphomas [Lavigueur et al., 1989] (Table 1). The broad spectrum of tumor susceptibility in these mutant Trp53 transgenic mice is also reflected in the diverse array of cancer types occurring prematurely in Li–Fraumeni cancer predisposition patients, who often inherit a mutant germline TP53 allele [Malkin et al., 1990].

More frequently, p53 mutant transgenes have been expressed in a tissue-specific manner using appropriate tissue-specific promoters driving the mutant p53 transgene. The transgenes have generally, though not always, been murine versions of human TP53 hotspot mutations in human cancers. For example, albumin gene promoter driven liver-specific expression of Trp53 transgene R246S (equivalent to human TP53 mutation R249S that is a liver cancer hotspot mutation) led to increased formation of liver tumors in response to treatment with liver carcinogens [Ghebranious and Sell, 1998]. Similarly, mammary gland-specific expression of Trp53 missense mutant R172H, analogous to human hotspot mutation R175H, resulted in accelerated mammary tumor formation following mammary carcinogen administration [Adams and Horton, 1998]. Interestingly, in this model and a epidermis-specific Trp53 R172H model, expression of the mutant transgene was associated with increased tumor-associated chromosomal instability relative to carcinogen-induced tumors that lacked the transgene, indicating one potential gain-of-function activity of missense mutant p53 [Adams and Horton, 1998; Wang et al., 1998]. Not all such mutant Trp53 transgenic models exhibit accelerated tumorigenesis, however. Prostate-specific and mammary gland-specific expression of a mutant Trp53 R172L actually showed resistance to carcinogen-induced tumorigenesis compared with controls, despite having a mutation observed in some human cancers [Armstrong et al., 1995; Hernandez et al., 2003; Petitjean et al., 2007]. Such observations indicate that individual missense mutations in p53 may preserve partial wild-type function and there may be large variations in tumor susceptibility induced by individual missense mutants.

Germline Knockout Trp53 Deletions

Gene targeting and mouse embryonic stem cell (ESC) methods were exploited to generate the first Trp53 knockout mice. Several laboratories generated these mice in the early 1990s and all knockout alleles were null for p53 activities through deletion of one or more Trp53 exons [Donehower et al., 1992; Tsukada et al., 1993; Jacks et al., 1994; Purdie et al., 1994]. Unlike many tumor-suppressor gene knockout mice, most homozygous p53 null (Trp53−/−) mice were developmentally viable, suggesting a modest role for p53 in embryonic development and a more central role as a stress response protein and tumor suppressor. Trp53 knockout mice were globally susceptible to a wide array of tumors, though lymphomas dominated in the Trp53−/− mice and lymphomas, soft tissue sarcomas, and osteosarcomas in the Trp53+/− mice. The cancer types observed in the Trp53+/− mice displayed some similarity to the high frequency of osteosarcomas, soft tissue sarcomas, and leukemias diagnosed in Li–Fraumeni patients, for which the Trp53+/− mice serve as a genetic model. Breast cancers are very frequent among Li–Fraumeni lineages, but were not observed in the standard strains of Trp53+/− mice (C57BL/6 and 129/Sv) [Donehower et al., 1995; Malkin, 2011]. However, when the Trp53 null allele was crossed into the mammary cancer susceptible BALB/c strain, mammary cancers were shown to be very frequent, illustrating the importance of strain background in the cancer spectrum of Trp53-deficient mice. The Trp53+/− mice also showed differences in loss or retention of the remaining wild-type Trp53 allele in developing tumors. About half of the C57BL/6 Trp53+/− mice exhibited retention of the wild-type Trp53 allele in tumors, whereas less than 10% of BALB/c tumors displayed retention of the wild-type Trp53 allele. Thus, in the BALB/c model, Trp53 allele loss conformed to the classic Knudson “two hit” model for tumor suppressors, whereas in C57BL/6 mice, a haploinsufficiency model for Trp53 might be invoked [Knudson, 1971; Berger et al., 2011].

As one of the first Trp53 genetically engineered mice, the Trp53 knockout mice have been workhorse models for applied and basic cancer studies for decades [Donehower, 1996]. These mice have been crossed to hundreds of other genetically engineered mice to investigate various gene interactions in the context of cancer phenotypes. The first reported cross of p53-deficient mice to Rb-deficient mice showed that bi-deficient progeny mice had accelerated tumorigenesis and a novel cancer spectrum not exhibited by either parental line [Williams et al., 1994; Harvey et al., 1995]. Often, though not always, Trp53 deficiency has accelerated tumorigenesis in mice with mutations in other tumor susceptibility genes. Likewise, carcinogens are generally more potent tumor accelerators in the context of p53 deficiency. The p53 knockout mice have been widely used as test organisms for carcinogenicity assay and the p53+/− have been approved by the US Food and Drug Administration as suitable models for rodent carcinogenicity assays on candidate drugs [MacDonald et al., 2004]. In comprehensive testing, p53+/− mice have shown to correctly identify 84% of genotoxic carcinogens [Pritchard et al., 2003]. Because p53-deficient mice have also been shown to be more sensitive to developmental abnormalities, such as neural tube defects, testing of teratogens on p53+/− embryos has increased our understanding of the effects of p53 signaling on development [Armstrong et al., 1995; Nicol et al., 1995].

Germline Knockin Trp53 Missense Mutations

The evolution of mouse ESC gene-targeting methods soon allowed the efficient introduction of single missense mutations into the mouse germline. The first two reported germline missense mutations were Trp53 R172H and R172P, both generated by the Lozano laboratory [Liu et al., 2000; Lang et al., 2004; Liu et al., 2004]. The R172H mutation, mimicking a hotspot TP53 mutation in human cancers, R175H, resulted in tumor phenotypes very similar to those arising in mice with Trp53 null alleles described above. Both tumor-free survival curves and spectrum of tumor types were similar for Trp53+/− versus Trp53+/R172H mice and Trp53−/− versus Trp53R172H/R172H mice. However, rates of metastases were consistently higher in the missense mutant Trp53 mice compared with the null allele-containing mice. These results confirmed that some missense mutant forms of p53 may have gain-of-function properties not possessed by null alleles of Trp53 that merely exhibit loss-of-function. Importantly, it was shown that one likely component of missense p53 variant gain-of-function was the functional inactivation of p53 family members p63 and p73 [Lang et al., 2004].

The mice with germline Trp53 R172P mutations mimicked a TP53 mutation (R175P) that occurs in human cancers but at a lower frequency than the R172H mutation. This mutation only partially inactivated p53 function, in that the ability to induce cell cycle arrest was retained, whereas the induction of apoptosis was absent [Liu et al., 2004]. Mice homozygous for the Trp53 R172P mutation display delayed tumorigenesis rates compared with Trp53 null mice, though tumor types for both groups were largely lymphomas. Moreover, many of the R172P tumors maintained diploid karyotypes and chromosomal stability, unlike the aneuploidy frequently seen in Trp53 null mouse tumors [Liu et al., 2004]. Such results indicate that wild-type p53 likely inhibits tumor initiation and progression both through induction of apoptosis as well as regulation of cell cycle checkpoints. Additionally, p53 may prevent chromosomal instability through its checkpoint regulatory functions.

Conditional Knockin Trp53 Mutant Alleles

Conditional missense Trp53 alleles have also been introduced into the mouse germline. The first such model was the “floxed p53” mouse [Marino et al., 2000; Jonkers et al., 2001]. This model contains LoxP recombination sites placed in introns 1 and 10 of the mouse Trp53 gene. When Cre recombinase is expressed in tissues of these mice, Trp53 exons 2–10 are deleted and a null allele is generated. This model has also been widely used to generate tissue-specific cancer models. Tissue or cell type-specific Cre recombinase-expressing mice or inducible Cre mice are generally crossed to the floxed p53 mice and the biallelic offspring are monitored for tumor formation in the tissue(s) of interest. In some cases, both Trp53 along with another tumor-suppressor allele (frequently Rb) are inactivated to accelerate tumor formation. Cre-LoxP approaches have been used to generate tissue-specific cancer models for medulloblastoma, soft tissue sarcomas, osteosarcomas, ovarian tumors, squamous cell carcinomas, mammary cancers, melanomas, gastric cancer, and pheochromocytomas [Marino et al., 2000; Zindy et al., 2007; Berman et al., 2008; Martinez-Cruz et al., 2008; Quinn et al., 2009; Artandi and DePinho, 2010; Choi et al., 2010; Fishler et al., 2010; Monahan et al., 2010; Tonks et al., 2010; Shimada et al., 2012]. In my own laboratory, we have crossed the floxed Trp53 mice to globally expressed tamoxifen-inducible CreERT2 mice [Hinkal et al., 2009]. We globally deleted Trp53 in mice at different times after carcinogenic radiation by treating the mice with tamoxifen. Deletion of Trp53 before, during, or after radiation for up to 4 weeks resulted in similar accelerated lymphomagenesis. Only mice that continuously retained wild-type Trp53 had delayed radiation-induced tumors. Thus, long-term maintenance of intact p53 beyond initial carcinogenic event was critical for p53-mediated tumor suppression [Hinkal et al., 2009].

Conditional missense Trp53 alleles have also been generated and studied extensively. First reported in the Jacks laboratory were mice with Cre-activatable Trp53 mutations R172H and R270H, mimicking human cancer hot-spot mutations R175H and R273H [Olive et al., 2004]. Each of these mice contains a LoxP–STOP–LoxP (LSL) cassette in intron 1 of the two germline mutant Trp53 alleles. In the presence of Cre recombinase, the STOP cassette is excised and the mutant Trp53 allele is efficiently transcribed and expressed. As with the floxed Trp53 mice, these conditional mutants are sometimes crossed to tissue-specific Cre mice to determine the effects of mutant p53 on tumorigenic phenotypes. In the first use of these models, the LSL–R172H and LSL–R270H mice were crossed to protamine-Cre mice to generate offspring that would express Cre recominase in haploid sperm. Subsequent matings resulted in global germline Trp53 R172H and R270H mice. These mice developed tumors at approximately the same rate as Trp53+/− mice, but more carcinomas were observed in the missense mutant mice, suggesting gain-of-function effects in epithelial cells. Gain-of-function effects for missense versions of p53 were corroborated in vitro when cells expressing the mutant Trp53 variants were inhibited in growth after downregulation of the mutant p53 by shRNA vectors [Olive et al., 2004].

The conditional nature of these two mutant Trp53 alleles has also been exploited in other studies by crossing tissue-specific Cre mice with the Trp53 LSL–R172H and LSL–R270H mice [Jackson et al., 2005; Wijnhoven et al., 2005; Doyle et al., 2010]. In an example of such an experiment, mice with a Cre-inducible K-Ras^G12D oncogenic allele and either a floxed Trp53, or Trp53 LSL-R172H, or LSL–R270H allele were nasally instilled with Cre-expressing adenoviruses [Jackson et al., 2005]. These adenoviruses resulted in lung adenomas and carcinomas as early as 6 weeks after infection. Interestingly, the mice with the R270H allele showed higher lung lesion progression than the floxed p53 or the R172H mice, indicating only some missense mutations exhibit observable gain-of-function in this context. Interestingly, only the R270H mice developed sinonasal tumors in this protocol, suggesting a cell type-specific gain-of-function role for this Trp53 variant.

Human TP53 Knockin Alleles

To facilitate experimental studies of the role of the human p53 in carcinogenesis studies, Hollstein and coworkers knocked in a wild-type human TP53 gene into the endogenous murine Trp53 locus [Luo et al., 2001]. In this chimeric, germline locus exons 4–9 of human TP53 (the primary sites of mutation in human tumors) have replaced the corresponding exons of the mouse Trp53 locus. The resulting model was designated the humanized p53 knockin (Hupki) mouse. The chimeric p53 was properly expressed and appeared to have all of the normal wild-type p53 functions in mouse tissues. The authors proposed that this would be a particularly useful tool for examining carcinogen and drug effects on p53 sequences and functions. In fact, treatment of Hupki mouse fibroblasts treated with benzo(a)pyrene, a tobacco smoke carcinogen, resulted in a similar spectrum of TP53 mutations as that observed in human lung tumors [Liu et al., 2005]. In follow-up studies, the hotspot mutations R248W and R273H were knocked into the chimeric TP53 allele and these mice were monitored for tumors. Homozygous Hupki R248 mice developed tumors at roughly the same rate as Trp53 null mice [Song et al., 2007]. The missense mutant Hupki mice did exhibit much higher frequencies of interchromosomal translocations that were attributed of gain-of-function effects on mutant p53 on the Mre11–Rad50–Nbs1 DNA double-strand break repair complex.

Recently, the Hollstein laboratory has developed a powerful approach for efficiently and rapidly generating novel knockin p53 mutations in ES cells (and p53 mutations in somatic cells) through an approach called recombinase-mediated cassette exchange [Wei et al., 2011, 2012]. Such technological advancements should result in an accelerated development of novel p53 germline mutant mice.

Phosphorylation Site Trp53 Knockin Mutant Alleles

Phosphorylation of the p53 protein represents a critical posttranslational modification mechanism in the regulation of p53 function. Roughly 25 phosphorylation sites have been identified across the length of the 393 amino acid protein and at least 30 serine/threonine kinases have been shown to mediate these phosphorylations. Generally, though not always, phosphorylation of p53 is associated with stabilization and activation of p53 activities as a transcription factor and inducer of apoptosis in response to a variety of stresses [Appella and Anderson, 2001; Jenkins et al., 2012; Loughery, 2013]. Cell culture studies on a number of p53 phosphorylation sites have shown significant effects on downstream p53 functions. Thus, there have been significant incentives to show that such phosphorylation site alterations also have important phenotypic effects within the whole organism.

The most extensively studied p53 phosphorylation site in human p53 is serine 15. In response to DNA damage, damage-inducible kinases such as ATM and ATR directly phosphorylate p53 at serine 15 (reviewed in Jenkins et al., 2012; Shiloh and Ziv, 2013]. This is a strong marker of p53 activation and stabilization and enhances p53 activities as a transcription factor and cell cycle inhibitor. The first Trp53 phosphorylation site mutant mouse was produced by the Jones laboratory, converting ATM/ATR target site Ser18 to Ala18 (Ser15 in human p53) [Sluss et al., 2004] (Table 2). As expected, mice with the Trp53 S18A mutation did show reduced apoptotic responses in thymocytes and splenocytes. Surprisingly, this mutation had little or no effect on fibroblast proliferation or DNA damage-induced cell cycle arrest responses. The mice also exhibited a modestly increased tumor susceptibility (late onset lymphomas and sarcomas) but much delayed relative to mice with a null Trp53 allele [Armata et al., 2007]. Crossing the S18A allele into c-myc transgenic lymphoma-susceptible background led to accelerated lymphomagenesis, confirming that this phosphorylation site on p53 plays a significant role in tumor suppression [Sluss et al., 2010].

A second key phosphorylation site mutated in mice was Ser23 (Ser20 in humans) [MacPherson et al., 2004]. Phosphorylation of this site by Chk1 or Chk2 damage-responsive kinases stabilizes p53 by inhibiting interactions with E3 ubiquitin ligase Mdm2. Fibroblasts from the Trp53 Ser23 mutant mice did not show major changes in damage-induced p53 stability or cell cycle arrest functions. However, IR-induced p53 stabilization and resistance to apoptosis were more significantly affected in thymocytes. The homozygous Ser23 mutant mice developed B-cell tumors between 1 and 2 years of age (significantly delayed relative to p53 null mice) [MacPherson et al., 2004]. Interestingly, a subsequent study described the characterization of double mutant S18A, Ser23A mutant mice [Chao et al., 2006]. Cells from these mice retained partial p53 apoptotic and cell cycle inhibitory functions, and developed lymphomas at the rate roughly similar to the S18A mutant mice. These studies again show the biological importance of p53 N-terminal phosphorylation sites that regulate Mdm2–p53 binding and p53 stabilization.

Phosphorylation of human p53 Ser315 by kinases such as CDK2 and aurora kinase A has been noted in the literature [Loughery, 2013], and these events may either negatively or positively regulate p53. Two corresponding S312A knockin mice have been reported [Slee et al., 2010; Lee et al., 2011]. These mice were phenotypically normal and exhibited no premature tumorigenesis. One study did not detect any differences between wild-type mice and S312A mutant mice [Lee et al., 2011], whereas the second described more stabilization of p53 following oncogenic stimulation, partial rescue of embryonic lethality of Mdm4−/− embryos, and accelerated lymphomas and liver tumors in response to ionizing radiation (IR) [Slee et al., 2010]. The S315A mutation was also knocked into the Hupki allele, and significant deficiencies in ESC differentiation were noted in Hupki S315A ESCs relative to their wild-type counterparts [Lin et al., 2005]. Part of this differentiation deficiency in the mutant ESCs was ascribed to an inability of S315A p53 to recruit the corepressor mSin3a to the Nanog promoter. Nanog promotes ESC self-renewal, and this p53 mutation may prevent normal differentiation in ESCs.

A prominent C-terminal phosphorylation site on p53 is Ser389 (Ser392 in humans). This site is specifically phosphorylated following UV damage. Mice generated with homozygous TP53 S389A mutations show no increased cancer susceptibility, but skin treatment of wild-type and S389A mice with UV-B irradiation demonstrated acceleration in skin tumorigenesis in the mutant mice relative to the wild-type mice [Bruins et al., 2004]. Subsequent carcinogen testing with this model also showed that it was more susceptible than control mice to 2-acetylaminofluorene-induced bladder cancers due in part to a reduction in expression of proapoptotic genes and other p53-dependent genes [Hoogervorst et al., 2005; Bruins et al., 2007; Bruins et al., 2008]. Thus, in response to specific stresses (e.g., UV irradiation and bulky carcinogenic adducts) Ser389 phosphorylation can facilitate the tumor-suppressor effects of p53, but this posttranslational modification is unlikely to play a significant role in p53-mediated tumor suppression initiated by other oncogenic events.

Finally, Ser46, a phosphorylation site closely associated with p53-regulated apoptosis activities, has been mutated in the Hupki background [Feng et al., 2006]. ESCs and embryo fibroblasts from these mice showed slight reductions in p53 stabilization following damage and partial impairment of the p53 apoptotic response. Moreover, some p53 proapoptotic genes were reduced in expression following DNA damage in mutant thymocytes relative to wild-type thymocytes. There were also modest increases in Ser46 mutant mouse embryo fibroblast (MEF) escape from senescence relative to wild-type MEFs, indicating effects of the mutation on regulation of p53-mediated senescence as well as apoptosis.

Acetylation Site Trp53 Knockin Mutant Alleles

Acetylation of p53, like phosphorylation, has been shown to regulate p53 functional activities [Brooks and Gu, 2011]. For example, DNA damage and other stresses increase p53 acetylation levels and facilitate p53 transcriptional activities. At least 13 acetylation sites have been identified on p53 and eight of these have been deemed indispensible for p53 activation [Brooks and Gu, 2011]. The first acetylation site mutants were reported in 2005 by two groups, one changing six lysine acetylation sites to arginine and one changing seven lysine sites to arginine so as to render them nonacetylatable [Feng et al., 2005a; Krummel et al., 2005]. The Trp53^6KR altered six acetylation sites in the C-terminal part of p53 shown to be important for transcriptional activation. These sites also play a role in p53 ubiquitination. Surprisingly, the Trp53^6KR mutant cells exhibited largely normal p53 stabilization after DNA damage, though p53-dependent gene expression was moderately defective in ESCs and thymocytes following damage [Feng et al., 2005b]. The Trp53^7KR model contained alteration of seven C-terminal p53 residues and resulted in mice with normal phenotypes. Like the Trp53^6KR mice, the Trp53^7KR mice had relatively normal p53 damage-induced stabilization characteristics and normal cell cycle arrest and apoptosis functions [Krummel et al., 2005]. The Trp53^7KR thymocytes did exhibit higher levels of p53 target activation, and Trp53^7KR MEFs displayed less immortalization potential than their wild-type counterparts, suggesting slightly hypermorphic phenotypes for this variant p53.

Like the Trp53^6KR and Trp53^7KR mutants, a single-acetylation site mutant, Trp53^K317R (K320R in humans), which alters a site N-terminal to the two previously discussed C-terminal multisite acetylation mutants, showed modest differences with wild-type p53 responses. Thymocytes and MEFs from Trp53^K317R mice showed normal p53-dependent gene expression, but increased IR-induced p53-dependent apoptosis was observed in multiple cell types in the mutants relative to wild-type mice. Moreover, p53 target gene upregulation was modestly more increased in the mutant cells following DNA damage, as was observed for the Trp53^7KR mice.

A TIP60/MOF-targeted acetylation site within the DNA-binding domain, K117 (K120 in humans), was also mutated to arginine in mice [Li et al., 2012]. The Trp53^K117R mutants displayed normal cell cycle arrest and senescence functions, but p53-induced apoptotic functions were completely lost in several radiation-treated tissues. Despite this loss of p53 apoptotic functions, these mice showed no enhanced tumor susceptibility for up to 70 weeks, indicating that K117-related p53-induced apoptotic functions play a relatively insignificant role in overall tumor suppression. The authors of the Trp53 K117R study also generated and characterized a triple acetylation site mutant, Trp53^3KR. In this mutant, sites K161 and K162 were changed to arginine along with K117. In the triple site mutant, both p53 apoptotic and cell cycle functions were abrogated. Yet these mice displayed only a modest later life tumor incidence despite these profound p53 functional deficiencies. Importantly, the authors did show that the triple site mutant retained the ability to activate antioxidant genes and regulate energy metabolism, indicating that these functions are more important in maintaining p53 tumor-suppressor function than previously understood [Li et al., 2012].

Transactivation Domain Trp53 Knockin Mutant Alleles

A central function of p53 is transcriptional regulation of hundreds of target genes that are associated with cell cycle control, apoptosis, autophagy, DNA repair, antioxidant defenses, and metabolism. While amino acid residues 100–300 roughly comprise the DNA-binding domain of the protein, the critical transactivation domains of p53 are localized near the N-terminus and comprise amino acids 1–40 and 41–83. These domains are critical for the various protein–protein interactions that facilitate p53 transactivation functions. Two laboratories reported the development of transactivation-deficient L25Q, W26S [Johnson et al., 2005; Nister et al., 2005] mutant mice. As expected, these mutants were defective for most transactivation functions of p53. However, this mutant did retain the ability to transactivate the proapoptotic Bax gene and was only partially defective for induction of apoptosis, depending on the stressor. DNA damage was ineffective in inducing apoptosis, but serum deprivation and hypoxia were capable of inducing an apoptotic response in the mutant cells. Because this particular mutation in p53 is incapable of binding to Mdm2, the E3 ubiquitin ligase that mediates p53 stability, germline expression of this mutant resulted in early embryonic lethality, presumably due to hyperstabilization of the mutant p53.

The Attardi laboratory followed up this study with another set of mutants to more fully characterize the roles of the transactivation domains in tumor suppression and p53 transcriptional functions [Brady et al., 2011]. Conditional Trp53 mutants at amino acids L25Q,W26S (transactivation domain 1), at F53Q,F54S (transactivation domain 2), and at L25Q,W26S,F53Q,F54S (both transactivation domains) were generated and further characterized. Again, the Trp53^25,26 mutant mice exhibited defects in transcriptional activation of p53 proapoptotic target genes and cell cycle arrest genes, but were not as defective for transactivation as p53 null mice. The Trp53^53,54 mutant mice displayed relatively wild-type levels of p53 target gene expression. The double mutant Trp53^25,26,53,54 mice showed p53 target-expression profiles similar to p53 null mice, suggesting p53 transactivation was completely knocked out. As might be expected, Trp53^53,54 mice retained largely intact cell cycle arrest, senescence, and apoptotic responses, whereas Trp53^25,26 mice showed loss of apoptosis but some retention of p53-mediated senescence. Trp53^25,26,53,54 mutants showed virtually complete loss of senescence induction. When the transactivation domain mutations were tested in a mutant K-Ras-driven lung cancer model, high numbers of lung tumors appeared in the p53 null background and the Trp53^25,26,53,54 background, though total lesion numbers were marginally less in the latter mice, suggesting a residual tumor-suppressor activity. K-Ras-driven tumors in Trp53^53,54 background were as low as in the wild-type p53 background, consistent with its largely intact p53 transactivation functions. Surprisingly, however, K-Ras-driven lung tumors in the Trp53^25,26 background were almost as infrequent as in the wild-type background. Testing of the Trp53^25,26 and Trp53^25,26,53,54 mice in several tissue lineages confirmed that the Trp53^25,26 mutation retains significant p53 tumor-suppressor function in mesenhchymal, central nervous system, and lymphoid lineages (unlike its Trp53^25,26,53,54 counterpart) [Jiang et al., 2011]. This result indicates that despite its highly crippled p53-mediated cell cycle arrest and apoptosis functions, this Trp53^25,26 mutation still retains high levels of tumor-suppressor function. The authors went on to identify novel p53 target genes that could still be induced by mutant p53^25,26 and showed that some of these genes could affect tumor growth, and did not have functions directly related to apoptosis or cell cycle control.

The transactivation domains at the N-terminus of p53 regulate many functions aside from transactivation, such as transcriptional repression, DNA replication, recombination, centrosome duplication, and mitochondrial function. Point mutations in the transactivation domain could affect these other functions. To separate these functions from transcriptional activation, the Attardi laboratory generated a mouse with an inducible chimeric VP16-p53 gene, p53(VP16), that replaced the p53 transactivation domains with the Herpes Simplex virus VP16 transactivation domain. This chimeric p53 protein transactivated a broad range of p53 target genes and was hyperefficient at inducing senescence in MEFs, even in the absence of stress. Nevertheless, it was completely nonfunctional in the ability to induce apoptosis, despite the ability to strongly transactivate proapoptotic genes such as Perp, Bax, and Noxa. Thus, these mice were important in showing that apoptotic gene transactivation by p53 is insufficient in generating an apoptotic phenotype and that other p53-associated functions (not present in the chimeric mutant) are required for full implementation of the cell death program.

Proline-Rich Domain Knockin Trp53 Mutant Alleles

The p53 protein contains a proline-rich domain extending from amino acids 58–98 that contain five repeats of the PXXP motif. The functional role of this domain is not fully understood, though it plays a role in p53 stabilization, transactivation, and apoptosis. It may also have tumor-suppressor function as deletion of residues 62–91 were shown to be deficient in growth suppression while maintaining transactivation function [Walker and Levine, 1996]. The Wahl laboratory generated a germline mutant Trp53 allele that had deleted six prolines in the proline-rich domain (p53^∆P) [Toledo et al., 2006]. This mutation results in a moderate to severe defect in p53 transactivation of p53 target genes. The p53^∆P mouse cells displayed an absence of G1 cell cycle arrest but appeared to maintain normal or only modestly deficient proapoptotic functions. Interestingly, crossing of this mutant mouse into an Mdm2 or Mdm4 null background rescued Mdm4 null embryos but not Mdm2 embryos from embryonic lethality. The p53^∆P mutant mice revealed only a slight enhancement of tumorigenesis relative to wild-type mice. However, p53^∆P showed little tumor-suppression activity in an E1A–Ras-induced tumor model.

In a subsequent paper from the Wahl laboratory, the investigators generated additional mouse lines with germline point mutations within the p53 proline-rich domain [Toledo et al., 2007]. The mutant p53^TTAA removed two prolyl isomerase Pin1 target sites that might affect Mdm2 binding and regulate protein stabilization. The second mutant p53^AXXA replaced four key proline sites with alanine, effectively removing all of the PXXP motifs in this region. In both models, mutant p53 accumulation in response to DNA damage was normal and induction of both mutants resulted in relatively wild-type transactivation of p53 target genes. In addition, both mutants displayed normal cell cycle arrest, apoptotic, and tumor-suppressor functions, indicating that Pin1 target sites and PXXP motifs do not have critical roles in the tumor-suppressor functions of wild-type p53.

Mice with a germline mutation in nuclear localization signal 1 were generated by mutating lysines 316–318 to alanine [Regeling et al., 2011]. As a transcription factor, nuclear localization plays a critical role in p53 functionality. MEFs from these mutant mice show proliferation rates similar to p53 null MEFs and reduced transactivation and apoptosis functions as well as defects in nuclear localization, as expected. Surprisingly, the Trp53 nuclear localization signal (NLS)1 mutant conferred even more resistance to UV damage in E1A transformed MEFs than in the same cells of a p53 null background. Equally surprising was the increased induction of embryonic and perinatal lethality caused by the neural tube developmental defect exencephaly in these mice.

DNA Binding Cooperativity Domain Knockin Mutant Allele

Optimized function of transactivation by p53 requires formation of a tetramer. Interference with the interactivity of the DNA-binding domains in the p53 tetramer can reduce p53 activity on its target genes. To investigate the effects of tetramer cooperativity, Stiewe and coworkers engineered a germline knockin mutant Trp53 allele E177R [Timofeev et al., 2013]. Cells in the mutant mouse exhibited deficiencies in apoptosis, whereas cell cycle arrest, senescence, and p53-associated metabolism functions were retained. Interestingly, the homozygous mutant mice were tumor-prone (angiosarcomas, carcinomas, and lymphomas), but did not succumb as rapidly as p53 null mice, behaving in some ways similarly to the Trp53 R172P mice described earlier.

Wild-Type Trp53 Transgenic Alleles

The first model to assess the effects of extra wild-type p53 expression was described in 1996. A mouse mammary tumor virus (MMTV) promoter driving a mouse p53 cDNA was introduced into the mouse germline [Godley et al., 1996] (Table 3). Surprisingly, rather than mammary gland phenotypes, significant kidney developmental phenotypes were observed. During embryogenesis, kidney metanephric cells displayed high rates of apoptosis and kidneys grew to only half of normal size by adulthood. The authors concluded that high levels of expression of transgene-derived p53 led to major differentiation defects in developing kidneys, resulting in apoptosis and defects in epithelial cell conversion. A second wild-type p53 transgenic model contained a liver-specific antithrombin III-wild-type Trp53 transgene [Gillet et al., 2000]. These mice showed elevated p53 and p53 target gene expression, but did not affect liver development. The extra p53 in these mice did reduce early SV40 T antigen-induced liver dysplasia, but ultimately had no effect on progression to hepatocellular carcinoma.

In 2002, Serrano and coworkers described engineered mice with globally expressed p53 transgenes [Garcia-Cao et al., 2002]. The generated lines contained one or two additional genomic copies of the Trp53 gene and flanking regions so that the transgenes would have normal Trp53 regulatory regions. These “super p53” mice did not show increased p53 protein levels or p53 target gene expression in normal tissues, but displayed increased responses to DNA damage in multiple tissues and exhibited increased resistance to carcinogen-induced tumors relative to normal mice. Importantly, these mice also aged normally, had a normal lifespan, yet developed fewer spontaneous tumors compared with wild-type mice. Thus, it appears that if p53 dosage can be globally increased in a manner that maintains its normal stress-responsive regulatory properties, augmented cancer resistance may be obtained without any deleterious pathologies or lifespan effects. Following up on the success of the super p53 mice, the Serrano and Blasco laboratories generated super Arf/p53 mice and even super Arf/p16/Tert/p53 mice that not only retained the enhanced cancer resistance of the super p53 mice, but actually exhibited extended median lifespans relative their wild-type counterparts. Such results indicated that enhanced normally regulated Trp53 expression could increase lifespan in the appropriate genetic contexts [Matheu et al., 2007; Tomas-Loba et al., 2008].

Inducible Wild-Type Trp53 Knockin Alleles

The generation of mice with conditional wild-type Trp53 alleles has been instrumental in furthering our understanding of the mechanisms by which p53-suppresses tumor initiation and progression. A good example of this type of model was the p53ER(TAM) line, which contained a wild-type Trp53 gene fused to a modified tamoxifen-sensitive estrogen receptor domain and knocked in to the endogenous Trp53 locus [Christophorou et al., 2005]. Mice homozygous for this allele were essentially null for p53 in the absence of tamoxifen (p53 remains cytoplasmic and nonfunctional) and developed mostly thymic lymphomas at almost the same frequency as Trp53 null mice. Treatment of the mice with tamoxifen restored the p53-mediated DNA damage response, normal p53 transcriptional regulation, and the normal p53 tumor-suppressor responses to oncogene activation [Christophorou et al., 2005]. With this model, Evan and coworkers then performed an elegant set of experiments in which the p53 fusion gene was reactivated in lymphoma-bearing mice. Tamoxifen treatment of these mice resulted in immediate activation of p53 functions, extensive apoptosis of lymphoma cells, and a significant survival extension [Martins et al., 2006]. However, restoration of p53 in the lymphomas eventually led to emergence of p53-resistant tumors due to loss of p19^ARF, a key activator of p53. In another set of experiments, IR treatment of the p53ER(TAM) mice induced early lymphomas in the absence of tamoxifen [Christophorou et al., 2006]. When mice were injected with tamoxifen at the same time as IR treatment, high levels of apoptosis were seen in the irradiated tissues, due to tamoxifen-activated p53. However, this p53 activation did not delay IR-induced lymphoma formation relative to IR-treated, nontamoxifen-treated mice. Only when p53 was restored 8 days after IR treatment was a significant delay in lymphoma onset observed. Such findings indicated that the immediate p53 damage response was irrelevant to tumor suppression. Only p53 activation considerably after the initial damage was important for suppressing tumor development. Moreover, this effect was p19^ARF-dependent. Thus, the primary mechanism of tumor suppression by p53 is not through apoptotic elimination of cells with damage, but rather through p19^ARF-facilitated suppression of oncogenically activated clones.

A second model with inducible wild-type Trp53 knockin allele was described by Jacks and coworkers, using a Cre recombinase LoxP flanked stop cassette engineered into intron 1 of the endogenous Trp53 allele [Ventura et al., 2007]. Mice homozygous for this allele are null for p53 function in the absence of Cre. However, when the mice were crossed with Cre-ERT2 mice that globally express a tamoxifen-inducible Cre recombinase, the biallelic progeny contained a global tamoxifen-inducible wild-type Trp53 allele. These mice were allowed to develop lymphomas and sarcomas in the absence of tamoxifen. The tumor-bearing animals were then treated with tamoxifen and p53 was shown to be activated. P53 activation resulted in regression of both lymphomas and sarcomas, with the major mechanisms of regression being apoptosis in lymphomas and senescence in sarcomas [Ventura et al., 2007]. Thus, the experiments with mice-containing inducible wild-type Trp53 mutations have demonstrated that restoration of p53 activity in an intact tumor environment is an effective therapeutic approach.

Hypermorphic Trp53 Alleles

Besides addition or restoration of wild-type Trp53 to the mouse, genetically engineered mice with hyperactive p53 activity have also been described [Tyner et al., 2002]. My laboratory in 2002 described a serendipitously generated p53 mutant mouse that resulted in hypermorphic p53 phenotypes. This mutant allele of p53 was deleted in the 5′ part of the p53 gene as well as upstream genes, but produced a truncated p53 protein containing only exons 7–11 of p53. These mice, dubbed p53^+/m mice, had average lifespans 23% shorter than their wild-type littermates and displayed a wide variety of accelerated aging phenotypes, including sarcopenia, osteoporosis, skin atrophy, delayed hair regrowth, and reduced tolerance to stresses. Moreover, these developed virtually no tumors and were cancer-resistant. Experiments on the cells from these mice showed that the truncated form of p53 drove wild-type p53 into the nucleus and created a low-level constitutive activation of p53 [Moore et al., 2007]. Experiments on hematopoietic stem cells (HSC) showed an accelerated decline in HSC functionality with age in the Trp53^+/m mice relative to their wild-type counterparts [Dumble et al., 2007].

A second hypermorphic p53 mouse reported by Scrable and coworkers globally expressed a 44 kD natural-truncated p53 isoform as a transgene and these mice also displayed accelerated aging phenotypes and a shortened median lifespan similar to those observed in the above described p53^+/m model [Maier et al., 2004]. These mice also were cancer-resistant, consistent with a hyperactive p53 response. Thus, two mice expressing truncated forms of p53 can mediate augmented cancer resistance in addition to shortened longevity and accelerated aging phenotypes. These phenotypes differ from the super p53 and super Arf/p53 mice that display cancer resistance but not early aging phenotypes. The most likely explanation for these differences in aging phenotypes lies with the nature of the Trp53 allele involved. The super p53 and super ARF/p53 mice maintained normal regulation of p53, whereas the Trp53^+/m and p44 TG mice were truncated forms that appeared to induce constitutive activation of p53. Constitutive p53 activation may be more deleterious over the lifespan of the animal by inducing a continuous antiproliferative stress response in all cells, particularly stem and progenitor cells, that may inhibit their normal functions and thus induce late life atrophy phenotypes [Dumble et al., 2004; Serrano and Blasco, 2007].

Finally, a p53 hypermorphic model was generated by Xu and coworkers in which Trp53 phosphorylated amino acid sites Ser21 and Ser23 (corresponding to human Ser18 and Ser20) were converted to aspartic acids, thus mimicking phosphorylated serine sites [Liu et al., 2010]. When these two sites are phosphorylated, binding of the p53 inhibitors Mdm2 and Mdm4 to p53 is reduced and p53 stability is increased. In fact, mice with a knockin allele of these mutations were embryonically lethal. Only when a selectable marker cassette that reduced expression of the knockin allele was retained could mice with this allele survive. These mice were further crossed to Trp53 null mice to generate p53^TSD/− mice that were further characterized for a variety of phenotypes. These mice showed runting and died by 6 weeks of age, displaying a variety of premature aging phenotypes, including lymphopenia, anemia, and progressive depletion of the hematopoietic system, neural stem cells, and spermatogonial stem cells [Liu et al., 2010]. Overall p53-dependent target gene expression was increased in multiple tissues of the mutant mice, indicating elevated p53 functional activities. Interestingly, expression of the p53 proapoptotic gene Puma was dramatically upregulated in these mice, and many of the segmental aging defects were ameliorated when the p53 mutation was crossed into a Puma null background. These studies confirmed that constitutively hyperactive p53 can deplete stem cell functions through enhanced apoptosis and thus affect aging-related phenotypes.