The C terminus of p53 regulates gene expression by multiple mechanisms in a target- and tissue-specific manner in vivo.

The p53 tumor suppressor is a transcription factor that mediates varied cellular responses. The C terminus of p53 is subjected to multiple and diverse post-translational modifications. An attractive hypothesis is that differing sets of combinatorial modifications therein determine distinct cellular outcomes. To address this in vivo, a Trp53(ΔCTD/ΔCTD) mouse was generated in which the endogenous p53 is targeted and replaced with a truncated mutant lacking the C-terminal 24 amino acids. These Trp53(ΔCTD/ΔCTD) mice die within 2 wk post-partum with hematopoietic failure and impaired cerebellar development. Intriguingly, the C terminus acts via three distinct mechanisms to control p53-dependent gene expression depending on the tissue. First, in the bone marrow and thymus, the C terminus dampens p53 activity. Increased senescence in the Trp53(ΔCTD/ΔCTD) bone marrow is accompanied by up-regulation of Cdkn1 (p21). In the thymus, the C-terminal domain negatively regulates p53-dependent gene expression by inhibiting promoter occupancy. Here, the hyperactive p53(ΔCTD) induces apoptosis via enhanced expression of the proapoptotic Bbc3 (Puma) and Pmaip1 (Noxa). In the liver, a second mechanism prevails, since p53(ΔCTD) has wild-type DNA binding but impaired gene expression. Thus, the C terminus of p53 is needed in liver cells at a step subsequent to DNA binding. Finally, in the spleen, the C terminus controls p53 protein levels, with the overexpressed p53(ΔCTD) showing hyperactivity for gene expression. Thus, the C terminus of p53 regulates gene expression via multiple mechanisms depending on the tissue and target, and this leads to specific phenotypic effects in vivo.

The p53 tumor suppressor is a transcription factor that mediates a variety of cellular responses. The best characterized of these include cell cycle arrest, which, if sustained, leads to senescence and apoptosis (Vousden and Prives 2009). The p53 protein is kept inactive through interactions with several negative regulators, most notably Mdm2 (Luo et al. 2004; Kruse and Gu 2009; Vousden and Prives 2009; Manfredi 2010). It has been proposed that post-translational modifications of p53 are critical to disrupt its binding to Mdm2 and facilitate its interaction with cofactors, resulting in a p53 protein that is robust in its ability to transcriptionally regulate target gene expression (Kruse and Gu 2009; Manfredi 2010). The C terminus of p53 in particular is subjected to multiple and diverse post-translational modifications (Kruse and Gu 2009; Carvajal and Manfredi 2013) quite similar to that seen with histone tails (Kouzarides 2007). An attractive hypothesis is that differing sets of combinatorial modifications therein determine distinct cellular outcomes (Carvajal and Manfredi 2013).

The C terminus was originally characterized as playing a key role in sequence-specific DNA binding by p53 (Hupp et al. 1992). Its excess of basic residues contributes to its ability to bind to DNA, albeit nonspecifically. It had been proposed that this region of p53 interferes with the ability of the core domain to bind response elements with sequence specificity and therefore has a negative regulatory function (Anderson et al. 1997; Ahn and Prives 2001). This notion was discredited by studies suggesting that this may be an artifact of the particular in vitro DNA-binding assays being used (Espinosa and Emerson 2001), although this remains controversial (Luo et al. 2004; Kruse and Gu 2009; Vousden and Prives 2009; Carvajal and Manfredi 2013). Experiments in cell culture were more in line with a positive role for the C terminus (Chen et al. 1996; Kruse and Gu 2009; Hamard et al. 2012; Carvajal and Manfredi 2013). These supported a role in tracking of p53 on genomic DNA (McKinney et al. 2004; Kruse and Gu 2009; Carvajal and Manfredi 2013) as well as serving as recruitment sites for transcriptional cofactors (Barlev et al. 2001; Mujtaba et al. 2004). The challenge in interpreting such experiments is that they rely on ectopic expression of an exogenously introduced p53, often overexpressed at a nonphysiologically relevant level.

To gain more definitive insight into the role of this region of p53, a Trp53 mouse was generated that replaces the endogenous p53 with a truncated form of p53 that lacks the C-terminal 24 amino acids. In spite of presenting a wild-type phenotype immediately after birth, Trp53 mice quickly degenerate and die within 2 wk post-partum. Two main phenotypes are observed: a hematopoietic failure and an abnormal cerebellar development. The C terminus of p53 is shown to regulate gene expression at several levels: p53 accumulation, genomic site occupancy, or transcriptional activation. The precise mechanism depends on the tissue and target, thereby conferring specific phenotypic effects in vivo.

Results

Generation of the Trp53ΔCTD/ΔCTD mouse

To generate a Trp53 mouse that lacks the C-terminal 24 amino acids of p53, a knock-in strategy was used (Fig. 1A). To enable future studies with targeted expression of the deletion mutant, a Trp53 mouse was first generated by means of a targeting vector harboring two exon 11s separated by a NEO selection cassette (see the Supplemental Material for details on the construction of the targeting vector). The resulting targeted Trp53 allele contains a wild-type exon 11 in frame with the rest of the upstream gene and enables the expression of a full-length (FL), wild-type p53 protein along with the shorter mouse-specific alternative spliced form, p53AS (Bienz et al. 1984). Two loxP recombination sites within intron 10 and downstream from the NEO cassette, respectively, enable the excision of the first wild-type exon 11 along with the NEO cassette after recombination with the CRE recombinase. The resulting Trp53 allele contains the mutated exon 11 in frame with the rest of the gene and enables the expression of a truncated p53 protein (p53ΔCTD) along with the shorter mouse-specific alternative spliced form (p53AS) as well. Two FRT recombination sites located within intron 10 and in the 3′ untranslated region (UTR) have been introduced for future studies using murine embryonic fibroblasts (MEFs) derived from Trp53 or Trp53 animals.

Figure 1.

Generation of mice expressing a p53 protein devoid of its C-terminal domain. (A) Schematic of the targeting construct harboring two exon 11s separated by a NEO selection cassette. The resulting targeted Trp53 allele contains a wild-type exon 11 in frame with the rest of the gene and enables for the expression of a FL, wild-type p53 protein along with the shorter mouse-specific alternative spliced form (p53AS). After CRE recombination, the first wild-type exon 11 and the NEO cassette are excised, resulting in a Trp53 allele that contains the second exon 11, which includes a STOP mutation at position 367. The Trp53 allele encodes for a truncated p53 protein (p53ΔCTD) that lacks the last 24 amino acids as well as the p53AS isoform. The genotyping strategies are detailed in the Supplemental Material and Supplemental Figure 1. (Yellow triangles) loxP sites; (green triangles) FRT sites; (H) HindIII; (X) XhoI; (black bar) Southern blot probe; (a & b, c, and d) primers used for genotyping; (TK) thymidine kinase. (B) Immunoblot shows p53 protein expression in MEFs derived from Trp53 and Trp53 animals before or after CRE recombination. The cells were infected with an adenovirus expressing the CRE recombinase at MOI = 200 to induce the expression of the truncated form. Twenty-four hours later, cells were treated with DOX at the final concentration of 0.2 μg/mL to reach appreciable levels of p53, and 24 h after treatment, cells were harvested, and immunoblotting analysis was conducted. β-Actin was used as a loading control. p53FL and p53ΔCTD are almost indistinguishable in size, and long running times and large gels were necessary to observe the difference in migration. (C) Immunoprecipitation of p53 in MEFs derived from Trp53 and Trp53 animals with the p53CTD-specific monoclonal antibody PAb421. The cells were treated for 6 h with the proteasome inhibitor MG132 at a final concentration of 40 μM to reach appreciable levels of p53. The PAb421 antibody fails at precipitating the truncated p53 protein from Trp53 cells. A 5% input was immunoblotted for p53, and β-actin was used as a loading control.

Trp53 heterozygous mice were viable and appeared phenotypically normal. They were intercrossed to generate Trp53 mice, which were also viable and indistinguishable from Trp53 animals, and were born at the expected Mendelian ratio (Supplemental Fig. 1A). To ascertain that the CRE-driven recombination was functional, MEFs were derived from embryonic day 14.5 (E14.5) embryos and treated with increasing doses of a CRE-expressing adenovirus (Ad-CRE) (Supplemental Fig. 1B). Using genomic DNA from MEFs of +/+, NEO/+, and NEO/NEO genotypes, recombination was observed at a multiplicity of infection (MOI) as low as 100, and the nonrecombined DNA was undetectable in Trp53 MEFs at MOI = 500. To confirm that the recombination occurring at the genomic DNA level was giving rise to the shorter p53ΔCTD protein, MEFs were treated with Ad-CRE for 24 h and then treated with the DNA-damaging agent doxorubicin (DOX) for another 24 h at the indicated dose (Fig. 1B). Long running times were necessary to observe the difference in size between wild-type and p53ΔCTD proteins. Trp53 mice were crossed with Protamine-CRE (PrmCre) mice to generate Trp53-expressing males. These males were intercrossed with wild-type C57BL/6J females to generate Trp53 heterozygous mice, which were born at the expected Mendelian ratio (data not shown), with no phenotypic difference from wild-type mice. Ultimately, these heterozygous mice were bred to obtain Trp53 homozygous animals. To confirm that these animals were expressing the truncated form of p53, MEFs derived from E14.5 Trp53 and Trp53 embryos were treated with the proteasome inhibitor MG132 to attain discernable amounts of p53 protein and analyzed by immunoblotting. Similar to what was observed in NEO/NEO MEFs, a shorter form of p53 could be seen in ΔCTD/ΔCTD MEFs (Fig. 1C). The PAb421 monoclonal antibody has an epitope that is located within the C terminus of p53. An immunoprecipitation using this antibody confirmed that the protein expressed in the ΔCTD/ΔCTD cells lacked the PAb421 epitope (Fig. 1C).

Deletion of the C-terminal 24 amino acids from p53 leads to postnatal developmental defects and death within 2 wk post-partum

Mice homozygous for deletion of the p53 C terminus displayed a striking phenotype after birth. Although the mutant pups did not show any significant phenotypic difference from their +/+ and heterozygous littermates at day 1 post-partum (P1) (Fig. 2A), these animals were markedly reduced in size and weight by P10 (Fig. 2A,B) and died within 2 wk post-partum (Fig. 2C). They exhibited several developmental defects, including kinked tails, abnormal tail tip, and digit pigmentation, and significant ataxia suggestive of underlying neurological defects. Their organs were isolated and weighed at P1 and P10. The tibia, thymus, and spleen were dramatically reduced in size, while the liver and kidney were unaffected, and the heart was enlarged (Fig. 2D; Supplemental Fig. 2A,D). The observed cardiomegaly is reminiscent of that seen in Trp53 mutants after irradiation (Wang et al. 2011) and is likely a compensatory consequence of the severe anemia these mutant mice present at P10. Mutant tibias were reduced in size and weight, and the red color associated with the bone marrow was not observed (Fig. 2D). Indeed, when assessed by complete blood count, Trp53 animals showed significant anemia, neutropenia, and thrombocytopenia (Fig. 2E). None of these defects were observed in the Trp53 heterozygous animals.

Figure 2.

Deletion of the C-terminal 24 amino acids from p53 leads to postnatal developmental defects and death within 2 wk post-partum. (A) Pictures of whole animals at P1 show no difference between littermates of different genotypes (Trp53, Trp53, and Trp53). At P10, Trp53 animals are smaller than their wild-type counterparts and exhibit several developmental defects, including kinked tails, abnormal tail tip, and digit pigmentation. Bar, 1 cm. (B) Weight curve shows marked growth retardation for Trp53 animals but no significant difference between Trp53 and Trp53 mice. (C) Survival curves spanning 30 d post-partum illustrate 100% death within 2 wk for Trp53 animals, whereas all other genotypes survive. (D) Pictures of P1 and P10 whole organs demonstrate no significant difference between all genotypes at P1 but marked anemia and reduction in size for most Trp53 organs at P10. Weight graphs of P10 whole organs show reduced weight for Trp53 spleen, thymus, and tibias but not the liver and no significant difference between Trp53 and Trp53 organs. Bar, 0.4 cm. (E) Complete blood counts reveal a marked reduction of white and red blood cells, platelets, hematocrit percentage, and hemoglobin concentration in Trp53 animals at P10.

The terminal 24 amino acids of p53 do not include the tetramerization domain (amino acids 307–355). It is therefore likely that wild-type p53 and p53ΔCTD are capable of co-oligomerization. In order to discern whether the FL p53 protein was playing a dominant-negative role in Trp53 mice, these animals were crossed with Trp53 mice. The Trp53− offspring were phenotypically indistinguishable from Trp53, Trp53−, and Trp53 animals up to 10 mo of age (data not shown). In contrast to Trp53 animals, which succumb to spontaneous lymphomas or sarcomas 6 mo after birth, Trp53 and Trp53− mice were all tumor-free up to the age of 6 mo (data not shown). Thus, the truncated mutant p53 retains tumor suppressor activity. The rescue of Trp53 mice by deletion of one of the mutant alleles (Fig. 2C) also supports the idea that the level and/or activity of the p53 protein devoid of its C terminus (p53ΔCTD) are central to the observed phenotype.

Deletion of the C-terminal 24 amino acids from p53 induces hematopoietic failure post-partum

To investigate this further, histopathology studies using paraffin sections of P10 pups were conducted. No significant differences were observed in kidney, bladder, skin, or lungs at P10 (Supplemental Fig. 2C; data not shown). The hematopoietic compartment, however, was strongly affected by the deletion of the C terminus. Although no significant differences between genotypes were observed immediately before birth (E18.5) in bone marrow (Fig. 3A) and liver (Fig. 3B), Trp53 bone marrow cellularity started to decline as early as P1 and was virtually aplastic at P10 or P12 compared with its wild-type and heterozygous counterparts. No difference was noted between the bone marrow of Trp53 and Trp53 animals at any age.

Figure 3.

Deletion of the C-terminal 24 amino acids from p53 induces hematopoietic failure post-partum. (A) Bone sections stained with haematoxylin and eosin (H&E) from Trp53, Trp53, and Trp53 animals of the indicated ages. No difference is observed at E18.5 among all phenotypes, but a progressive hematopoietic failure is seen as soon as P1 only in Trp53 animals. (B) Liver sections stained with H&E show no difference between E18.5 Trp53, Trp53, and Trp53 animals and an impaired EMH (indicated by black arrowheads) starting as soon as P1 only in Trp53 animals. Centrilobular degeneration (yellow arrowheads) is observed over the regions surrounding the central vein (black asterisks). Bar, 200 μm.

Mice normally experience extramedullary hematopoiesis (EMH) in the liver at birth (Wolber et al. 2002), and it ceases around P9–P13. At E18.5, livers from all three genotypes exhibited abundant EMH, characterized by aggregates of mature and immature hematopoietic cells from all three lineages (erythrocytic, granulocytic, and megakaryocytic) (Fig. 3B). At P1, abundant EMH was still apparent in the Trp53 mouse, with decreased amounts in the Trp53 mouse (Fig. 3B). At P10 and P12, scattered EMH was still present in the Trp53 and Trp53 mice, but EMH was virtually absent in the Trp53 mice. There was also centrilobular vacuolar degeneration in the Trp53 mice at P10 and P12 that was not apparent in the Trp53 and Trp53 mice (Fig. 3B). Centrilobular areas are the last to receive oxygenated blood in the liver and are more susceptible to hypoxic injury. The Trp53 mice are significantly anemic, and their attendant hypoxia might contribute to the observed centrilobular vacuolar degeneration.

The remaining hematopoietic organs were also affected by the deletion of the C terminus of p53. The spleens of Trp53 and Trp53 mice were identical in size and cellularity (Figs. 2D, 9C, below) and exhibited abundant EMH in the red pulp and cellular white pulp. The spleens of the Trp53 mice, in contrast, were dramatically reduced in size and cellularity (Figs. 2D, 9C, below) and exhibited sparse EMH and less cellular white pulp (Fig. 9C, below). Likewise, although no differences were noted between the Trp53 and Trp53 mice, the thymus was markedly less cellular in the Trp53 mice, especially prominent in the cortex (Fig. 6C, below). Gut-associated lymphoid tissue (GALT) and lymph nodes were not apparent in most of the Trp53 mice, and when present, these tissues were much less cellular than their Trp53 and Trp53 counterparts (Supplemental Fig. 2C). Taken together, these data argue that the deletion of the C terminus of p53 has profound consequences for both medullary hematopoiesis and EMH.

Figure 9.

In the spleen, p53 lacking the C-terminal domain is overexpressed and hyperactive in a target gene-selective manner. (A) Relative expression of five p53 direct target genes in P10 spleens from Trp53, Trp53, Trp53, and Trp53 animals by qRT–PCR demonstrates target-specific regulation. Expression is normalized to Gapdh. In all panels, n = 5; bars indicate SEM; (*) P < 0.05; (***) P < 0.001; (N.S.) nonsignificant (Student's t-test). (B, top) Immunoblot shows increased p53 expression in the mutant spleen compared with wild type at P10. β-Actin was used as the loading control. (Bottom) Quantitation of p53 levels (relative intensity of p53 bands vs. actin bands); n = 5; bars indicate SEM; (***) P < 0.001 (Student's t-test). (C) Histology of the spleen of P10 animals shows no difference between Trp53 and Trp53 organs but reduced size and cellularity in Trp53 animals. IHC on the same tissues reveals a significant and dose-dependent increase of p53 protein expression. Bar, 200 μm. (D, top) SA-β-Gal activity in P10 splenocytes isolated from Trp53, Trp53, and Trp53 animals reveals no difference in senescent cell numbers between wild-type, heterozygous, or mutant thymi. (Bottom) IHC on spleen sections from P10 Trp53, Trp53, and Trp53 animals reveals no significant increase of activated caspase 3 staining in the mutant spleen.

Figure 6.

The C-terminal domain is a negative regulator of p53 activity in the thymus. (A) Relative expression of five p53 direct target genes in P10 thymi from Trp53, Trp53, Trp53, and Trp53 animals by qRT–PCR demonstrates tissue-specific up-regulation of the two proapoptotic targets Bbc3 (Puma) and Pmaip1 (Noxa). Expression is normalized to Gapdh. In all panels, n = 5; bars indicate SEM; (*) P < 0.05 (Student's t-test). (B, top) Immunoblot shows comparable p53 expression between the wild-type and mutant thymi at P10. p21 expression is not changed, whereas Puma protein levels are increased in the mutant thymus. β-Actin was used as the loading control. (Bottom) Quantitation of p53 levels (relative intensity of p53 bands vs. actin bands), n = 5; bars indicate SEM; (P) Student's t-test; (N.S.) nonsignificant. (C) P10 Trp53 thymi display reduced size and cellularity but no difference in basal p53 levels, as assessed by H&E staining and IHC, respectively. Bar, 200 μm. (D) ChIP on P10 thymus protein extracts followed by qPCR demonstrates enhanced binding of p53ΔCTD on its response elements within Bbc3 (Puma) and Pmaip1 (Noxa) genes.

Hematopoietic failure can have multiple causes but is often associated with hematopoietic stem cell (HSC) defects. To determine whether the C terminus of p53 played a role in HSC functions, the relative abundance of lineage-negative Lin− Sca1+ cKit+ (LSK) cells (in prenatal livers and P10 whole bone marrow) of Trp53, Trp53, Trp53, and Trp53 mice was measured by flow cytometry. Although no significant difference was observed in the relative frequency of LSK cells between the Trp53, Trp53, and Trp53 mice in the fetal liver at E14.5 (Fig. 4A; Supplemental Fig. 3A), a statistically significant reduction in the absolute number of fetal liver LSK cells was noted in the Trp53 mice. In accordance with previous studies, the relative numbers of LSK cells were higher in Trp53 mice. By comparison, in the whole bone marrow at P10, both the relative frequency and absolute numbers of LSK cells were significantly decreased in the Trp53 and Trp53 mice in a dose-dependent manner (Fig. 4A; Supplemental Fig. 3B). In fact, the LSK population in the bone marrow of the Trp53 mice was nearly completely depleted. (Fig. 4A; Supplemental Fig. 3B). To compare the differentiation capacity of stem cells from the wild-type and mutant fetal liver and P10 bone marrow, a colony-forming unit assay was conducted. Interestingly, in accordance with the measurable presence of LSK cells by flow cytometry in the livers of the Trp53 mice, the liver cells cultured from the Trp53, Trp53, and Trp53 mice were capable of forming colonies (Fig. 4B), implying that the differentiation capacity of the Trp53 liver LSK cells is intact in vitro. Conversely, no colonies were observed using whole bone marrow cells from Trp53 mice (Fig. 4B; Supplemental Fig. 3C), in accordance with the observed ablation of LSK cells in Trp53 mouse bone marrow measured by flow cytometry. HSCs are highly sensitive to reactive oxygen species (ROS), which limit and impair their functions (He et al. 2009). It has also been shown that ROS can induce the p53 pathway. Nevertheless, cultivating the fetal liver cells in 5% oxygen did not rescue the phenotype observed at 20% oxygen (data not shown).

Figure 4.

HSC homeostasis is perturbed in Trp53 mice. (A) Percentages of LSK cells from E14.5 fetal liver and P10 whole bone marrow show decreased numbers of HSCs in the bone marrow. Bars indicate SEM. (*) P < 0.05; (**) P < 0.01; (***) P < 0.001 (one-way ANOVA); (N.S.) nonsignificant. (B) Number of colonies formed by E14.5 fetal liver and P10 whole bone marrow cells from Trp53, Trp53, Trp53, and Trp53 animals cultivated on methylcellulose-based medium. Colonies originating from three types of progenitors were counted: erythroid progenitors (burst-forming unit-erythroid [BFU-E]), granulocyte and macrophage progenitors (colony-forming unit-granulocyte and macrophage [CFU-GM]), and multipotential granulocyte, erythroid, macrophage, and megakaryocyte progenitors (CFU-granulocyte, erythroid, macrophage, and megakaryocyte [CFU-GEMM]). (C) The percentage survival of lethally irradiated mice of mixed genetic background (BL6/129Sv) transplanted with E14.5 fetal liver cells from matching BL6/129Sv Trp53, Trp53, and Trp53 embryos shows the decreased ability of ΔCTD/ΔCTD cells to recolonize the recipient bone marrow. Animals receiving no transplant were used as negative control. (D) Relative expression of two transcription factors essential for hematopoiesis (Gfi1 and Gfi1b) in P10 livers, bone marrow, spleens, and thymi from Trp53, Trp53, Trp53, and Trp53 animals by qRT–PCR reveals impaired activity for the p53ΔCTD mutant. Expression is normalized to Gapdh. In all panels, n = 5; bars indicate SEM; (*) P < 0.05; (**) P < 0.01 (Student's t-test).

To address whether the wild-type or mutant HSC repopulating capabilities were different, fetal liver cells from donor mice with varying genotypes (+/+, ΔCTD/+, or ΔCTD/ΔCTD) were transplanted into lethally irradiated recipient mice of matching mixed genetic background (BL6/129Sv). Less than 70% of the recipients who received either ΔCTD/ΔCTD cells survived, (Fig. 4C), demonstrating that the HSCs from the Trp53 fetal livers are capable of reconstituting the bone marrow of the irradiated recipient, although not as consistently as their wild-type counterparts. This capacity to reconstitute the wild-type recipient mice is in accordance with the in vitro colony-forming capacity of the Trp53 fetal liver cells (Fig. 4B). The reduced ability to reconstitute all irradiated recipients may indicate a cell-intrinsic defect in the Trp53 HSCs or simply reflect the reduced absolute LSK cell numbers among mutant fetal liver cells. These data collectively demonstrate that the deletion of the C terminus of p53 disturbs critical HSC functions, including differentiation and repopulating capacities.

Recent studies have shown that p53 plays a pivotal role in HSC self-renewal (Milyavsky et al. 2010) and quiescence (Liu et al. 2009), in part by activating a subset of specific and relevant target genes. Notably, p53 transcriptionally activates expression of the transcription factor Gfi1, which has been shown with its homolog, Gfi1b, to be one of the key regulators of hematopoiesis (Cellot and Sauvageau 2005; Duan and Horwitz 2005; Liu et al. 2009; van der Meer et al. 2010). Gfi1 and Gfi1b are both expressed in a tissue-specific manner. Although both are expressed in the bone marrow, Gfi1 is expressed exclusively in the thymus, while Gfi1b is expressed in the spleen (Tong et al. 1998). The expression levels of these genes in hematopoietic tissues across different genotypes were measured by quantitative RT–PCR (qRT–PCR) (Fig. 4D). Interestingly, basal levels of Gfi1 were controlled by p53 only in the thymus (Fig. 4D) and to a lesser extent in the spleen, even though the overall expression in the latter was significantly diminished compared with the thymus, as expected. In the bone marrow, on the other hand, Gfi1 basal levels were identical between Trp53, Trp53, and Trp53 genotypes but were significantly reduced in the Trp53 animals. The same pattern was observed for Gfi1b, with p53 controlling basal levels of Gfi1b in the spleen but not in the bone marrow (Fig. 4D). Thus, in the bone marrow, the dominant source of hematopoiesis in newborn and adult mice, the deletion of the C terminus of p53 resulted in reduced expression of Gfi1 and Gfi1b, two of the most important regulators of hematopoietic homeostasis. These low levels of Gfi1 and Gfi1b might indicate that the pool of cells in which these two genes are normally expressed is absent in the mutant bone marrow. Furthermore, flow cytometric analysis of CD3+, CD19+, and CD11b+ populations revealed an aberrant cellular subset composition in the mutant bone marrow, with an accumulation of CD3-positive cells and no change in CD19+ or CD11b+ cells' relative frequency among all genotypes (Supplemental Fig. 7). It is also possible that abnormal expression of p53 target genes within different hematopoietic lineages may drive shifts in cell populations within organs, which may account for differences in whole-tissue gene expression readouts. This could explain, at least in part, the phenotypic difference observed between Trp53 and Trp53 mice and the hematopoietic failure observed in the latter after birth.

Deletion of the C-terminal 24 amino acids from p53 induces senescence in bone marrow cells

To gain further insight into the mechanism of hematopoietic failure in Trp53 mice, p53-dependent senescence and apoptosis were assessed in whole bone marrow cells isolated from P10 animals. First, whole bone marrow cells from P10 animals were stained for senescence-associated β-galactosidase (SA-β-Gal) activity (Fig. 5A,B). The number of SA-β-Gal-positive cells was greatly increased in Trp53 animals, while little was observed between the other genotypes. mRNA levels of several bona fide senescence markers were measured by qRT–PCR. Cdkn1a (p21) and Cdkn2b (p15INK4b), two potent cell cycle inhibitors that have been shown to be overexpressed in senescent cells (Collado and Serrano 2006), were up-regulated in the mutant bone marrow cells (Fig. 5C). Expression of Cdkn1a (p21) is clearly p53-dependent, as its level is markedly reduced in p53-null tissue. The increase in its expression is thus likely due to hyperactivity of p53ΔCTD. This was not the case for several other p53 target genes, including Mdm2, Bbc3 (Puma), Pmaip1 (Noxa), or Tigar (Supplemental Fig. 4). Expression of Cdkn2b (p15INK4b) is also similar between wild-type and null animals. Thus, in the case of Cdkn2b (p15INK4b), the increased levels that are observed (Fig. 5C) are likely an indicator of the enhanced senescence. Two other transcripts originating from the Cdkn2a locus (p19ARF and p16INK4a) did not show any statistically significant difference between Trp53, Trp53, Trp53, and Trp53 bone marrow. In addition, bone marrow cells were stained for activated caspase 3, an apoptotic marker (Fig. 5D). MEFs derived from Trp53, Trp53, Trp53, and Trp53 E14.5 embryos and treated with the DNA-damaging agent DOX for 24 h were used as positive and negative controls (Fig. 5D, inserts). Although the p53ΔCTD mutant protein retained its ability to induce apoptosis after DNA damage in these cells, no apoptosis was observed in bone marrow cells from all of the genotypes tested (Fig. 5D). These data suggest that although p53ΔCTD retains its ability to induce both senescence and apoptosis, only senescence is involved in the observed bone marrow phenotype.

Figure 5.

Deletion of the C-terminal 24 amino acids from p53 induces senescence in bone marrow cells. (A) Quantitative assay of SA-β-Gal activity in P10 bone marrow cells from Trp53, Trp53, Trp53, and Trp53 animals reveals a marked increase in senescent cells in the mutant bone marrow. n = 5; bars indicate SEM; (***) P < 0.001 (Student's t-test). (B) Representative pictures of A. (C) Relative expression of several senescence markers in P10 whole bone marrow from Trp53, Trp53, Trp53, and Trp53 animals by qRT–PCR. The p53 direct target gene Cdkn1a (p21) is a bona fide senescence marker and is up-regulated in ΔCTD/ΔCTD bone marrow. Cdkn2b but not the two transcripts encoded by Cdkn2a is significantly up-regulated in the ΔCTD/ΔCTD bone marrow. Expression is normalized to Gapdh. In all panels, n = 5; bars indicate SEM; (*) P < 0.05; (**) P < 0.01 (Student's t-test); (N.S.) nonsignificant. (D) Whole bone marrow cells from P10 animals of the genotypes Trp53, Trp53, Trp53, and Trp53 stained with an activated caspase 3 antibody, a bona fide marker for apoptotic cells, did not show enhanced apoptosis. Insets represent MEFs derived from E14.5 Trp53, Trp53, and Trp53 embryos and treated with the DNA-damaging agent DOX at 0.2 μg/mL. Twenty-four hours after treatment, cells were fixed and stained with an anti-activated caspase 3. Unlike p53−/− cells, p53ΔCTD/ΔCTD MEFs are capable of undergoing apoptosis after DNA damage to the same extent as p53+/+ or p53ΔCTD/+ MEFs.

In the thymus, the C-terminal domain negatively regulates p53-dependent gene expression by inhibiting DNA binding

The thymi in the Trp53 mice are reduced in size (Fig. 2D) and show severe morphological differences at the microscopic level (Fig. 6C). To examine p53-dependent effects in this tissue, the gene expression of a set of well-characterized p53 targets was compared between wild-type and p53-null animals (Fig. 6A). All targets examined showed clear p53 dependence in their gene expression: Cdkn1a (p21), Mdm2, Bbc3 (Puma), Pmaip1 (Noxa), and Tigar (Fig. 6A). Of these, the wild-type and the p53ΔCTD mice showed little difference in expression of either Cdkn1a (p21), Mdm2, or Tigar. In contrast, both Bbc3 (Puma) and Pmaip1 (Noxa) showed significantly increased expression in the presence of the p53ΔCTD. The effects on p21 and Puma protein expression were confirmed by immunoblotting (Fig. 6B). These differences in gene expression were not due to altered p53 levels, determined by either immunoblotting of tissue extracts (Fig. 6B) or immunohistochemistry (IHC) (Fig. 6C). To gain mechanistic insight into the difference in gene expression, chromatin immunoprecipitation (ChIP) analysis was performed on selected genomic sites. Occupancy of p53 on all sites examined was only seen in the wild-type but not the null tissues (Fig. 6D). Although wild-type and p53ΔCTD were found equivalently associated with the two response elements in the Cdkn1a (p21) promoter, the p53ΔCTD showed markedly increased occupancy on the elements in the Bbc3 (Puma) and Pmaip1 (Noxa) genes (Fig. 6D). Thus, the enhanced Bbc3 (Puma) and Pmaip1 (Noxa) expression seen with p53ΔCTD is accompanied by increased occupancy of the relevant genomic sites. This suggests that in the thymus, the C terminus of p53 negatively regulates DNA binding at least on a subset of target genes.

Hyperactive p53 lacking the C-terminal domain induces apoptosis but not senescence in thymocytes

To determine whether this enhanced expression of proapoptotic targets has a corresponding phenotypic outcome, thymocytes were stained for either SA-β-Gal activity (as a marker for senescence) or activated (cleaved) caspase 3 (as an indicator of apoptosis) (Fig. 7). Although little SA-β-Gal activity could be detected (Fig. 7B), there is clear evidence of activated caspase 3 in the thymi from ΔCTD/ΔCTD mice but not those harvested from either the wild-type or heterozygous animals (Fig. 7A). Taken together, this indicated that in the thymus, the C terminus of p53 prevents DNA binding to Bbc3 (Puma) and Pmaip1 (Noxa) but not Cdkn1a (p21). This in turn leads to differential gene expression that is consistent with an enhanced apoptosis in these tissues. It is likely, then, that this increased cell death is responsible for the smaller thymus size in the homozygous mutant mice.

Figure 7.

Hyperactive p53 lacking the C-terminal domain induces apoptosis but not senescence in thymocytes. (A) IHC on thymus sections from P10 Trp53, Trp53, and Trp53 animals reveals a significant increase of activated caspase 3 staining in the mutant thymus. Black arrowheads show activated caspase 3-positive cells. (B) SA-β-Gal activity in P10 thymocytes isolated from Trp53, Trp53 and Trp53 animals reveals no difference in the number of senescent cells between wild-type, heterozygous, or homozygous mutant thymi.

In the liver, the C-terminal domain of p53 is required for target gene expression at a step subsequent to DNA binding

Livers from the Trp53 mice are unaffected in size (Fig. 2D) but show altered colorization, most likely due to altered blood cell production (Fig. 2E). To examine p53-dependent effects in this tissue, p53 target gene expression was again compared between wild-type and p53-null animals (Fig. 8A). Cdkn1a (p21), Mdm2, Pmaip1 (Noxa), and Tigar showed clear p53 dependence in their gene expression, while Bbc3 (Puma) did not (Fig. 8A). As in the thymus, comparison of wild-type and p53ΔCTD mice showed little difference in expression of Cdkn1a (p21). In contrast, Mdm2, Pmaip1 (Noxa), and Tigar all showed significantly reduced expression in the presence of the p53ΔCTD. The effects on p21 protein expression were confirmed by immunoblotting (Fig. 8B). These differences in gene expression were not due to altered p53 levels, determined by either immunoblotting of tissue extracts (Fig. 8B) or IHC (Fig. 8C). ChIP analysis showed that occupancy of p53 on all sites examined was only seen in the wild-type but not the null tissues (Fig. 8D). There were no significant differences in promoter occupancy by wild type or p53ΔCTD for either the Cdkn1a (p21), Mdm2, or Pmaip1 (Noxa) genes (Fig. 8D). Intriguingly, the p53ΔCTD showed markedly reduced occupancy on the Tigar gene (Fig. 8D). Thus, the reduced gene expression seen with p53ΔCTD is not reflected in changes in occupancy of the relevant genomic sites. This argues that in the liver, the C terminus of p53 is needed for a subsequent step in transcriptional regulation beyond DNA binding, perhaps at the level of coactivator recruitment.

Figure 8.

In the liver, p53 lacking the C-terminal domain is expressed at wild-type levels but is defective in gene expression. (A) Relative expression of five p53 direct target genes in P10 livers from Trp53, Trp53, Trp53, and Trp53 animals by qRT–PCR demonstrates tissue-specific down-regulation of all targets except p21. Expression is normalized to Gapdh. In all panels, n = 5; bars indicate SEM; (*) P < 0.05; (**) P < 0.01; (***) P < 0.001; (N.S) nonsignificant (Student's t-test). (B, top) Immunoblot shows comparable p53 expression between wild-type and mutant livers at P10. β-Actin was used as the loading control. (Bottom) Quantitation of p53 levels (relative intensity of p53 bands vs. actin bands); n = 5; bars indicate SEM; (P) Student's t-test; (N.S.) nonsignificant. (C) P10 Trp53 and Trp53 livers display comparable size and cellularity and no difference in basal p53 levels, as assessed by H&E staining and IHC, respectively. Bar, 200 μm. (D) ChIP on P10 liver protein extracts followed by qPCR. In all panels, n = 4; bars indicate SEM; (*) P < 0.05; (**) P < 0.01; (N.S.) nonsignificant (Student's t-test).

In the spleen, p53 lacking the C-terminal domain is hyperactive due to its overexpression

Spleens from the Trp53 mice are reduced in size (Fig. 2D) and show morphological differences at the microscopic level (Fig. 9C). In this tissue, p53-dependent expression of Cdkn1a (p21), Pmaip1 (Noxa), and Tigar is seen, but not for Mdm2 or Bbc3 (Puma) (Fig. 9A). Mice expressing p53ΔCTD showed enhanced expression of Cdkn1a (p21) and Pmaip1 (Noxa) but reduced mRNA levels of Tigar (Fig. 9A). The effects on p21 protein expression were confirmed by immunoblotting (Fig. 9B). In contrast to the thymus and liver, p53ΔCTD levels were substantially higher, as seen by immunoblotting of tissue extracts (Fig. 9B) or IHC (Fig. 9C). There was no detectable evidence of senescence or apoptosis using appropriate assays (Fig. 9D). Thus, the increased protein levels of p53ΔCTD are accompanied by enhanced gene expression on some (Cdkn1a and Pmaip1) but not other (Tigar) genes. The molecular basis for this striking difference remains to be determined. p53ΔCTD is similarly overexpressed in fetal livers (Supplemental Fig. 5A,B). As in the spleen, this is accompanied by enhanced gene expression of Cdkn1a (p21) and Pmaip1 (Noxa) and reduced levels of Tigar mRNA (Supplemental Fig. 5C). In contrast to the thymus and adult liver, in the spleen and fetal liver, the C terminus is needed to maintain appropriate p53 protein levels.

The p53 C-terminal domain is required for proper cerebellum development in vivo

Aside from hematopoietic failure, another major phenotype observed in Trp53 mice was pronounced ataxia, which suggested the presence of neurological defects. Thus, the brains of Trp53 and Trp53 animals were compared. Overall, the brains of homozygous mutant mice were smaller than those of their wild-type or heterozygous counterparts (Supplemental Fig. 6A; data not shown). The cerebellar vermis in the homozygous mutant brain was virtually nonexistent, revealing more of the colliculi (Supplemental Fig. 6A). In addition, the cerebellar folia appeared shallower with a perturbed foliation pattern, while no significant differences were noted in the cerebrum or brainstem (Supplemental Fig. 6B). Previous studies have shown that p53 levels are critical in the development of the external granular layer (EGL) and internal granular layer (IGL) of the cerebellum (Malek et al. 2011). Accordingly, there was an apparent decrease in the thickness and cellularity of both the EGL and IGL of the Trp53 cerebellum compared with the Trp53 and Trp53 counterparts (Supplemental Fig. 6B,C). Of note, the Purkinje cells of the homozygous mutant cerebellum failed to form a uniform monolayer, as observed in the wild-type or heterozygous counterparts (Supplemental Fig. 6C). Interestingly, when assessed by immunoblot, p53 protein levels were increased in the cortex of both Trp53 and Trp53 animals compared with Trp53, whereas no difference was observed in the cerebellum of these three genotypes (Supplemental Fig. 6D). This finding argues that the C-terminal domain regulates p53 functions in the cerebellum without interfering with its level of expression. These data collectively demonstrate the importance of C-terminal regulation of p53 in cerebellar development. Further studies will be needed to determine the precise mechanism by which p53 intervenes in this process.

Discussion

The p53 tumor suppressor is a critical mediator of the cellular response to stress and is mutated in >50% of human tumors (Brosh and Rotter 2009). Trp53 mice were originally described as developmentally normal but rapidly succumb to spontaneous cancers (mainly lymphomas and sarcomas) within 6–9 mo after birth (Donehower et al. 1992; Jacks et al. 1994). p53 has also been implicated in stem cell maintenance and homeostasis (Zheng et al. 2008; Cicalese et al. 2009; Liu et al. 2009; Spike and Wahl 2011; Bonizzi et al. 2012). These p53 functions may be of particular relevance to tissues that are known to contain a significant pool of adult stem cells (Uccelli et al. 2008; Doulatov et al. 2012;). As p53 has also been shown to regulate reprogramming of somatic cells into induced pluripotent stem cells (iPSCs), these findings argue in support of the idea that one of the roles of p53 is to tightly regulate homeostatic adult tissues. Similarly, p53 has been implicated in brain development (Liu et al. 2007; Terzian et al. 2007; Malek et al. 2011; Mendrysa et al. 2011). Although originally described as developmentally normal, Trp53 mice were later shown to have development defects, including impaired maternal reproduction (Hu et al. 2007), aberrant mesenchymal differentiation programs (Molchadsky et al. 2008), and exencephaly (Armstrong et al. 1995; Sah et al. 1995). The molecular basis for these developmental functions of p53 clearly is an important area for further study.

In the present study, it was shown that the deletion of the C-terminal domain of the p53 protein is sufficient to provoke striking postnatal developmental phenotypes in the absence of induced DNA damage, including hematopoietic failure and brain defects. Previous reports of C-terminal Trp53 and Trp53 mice in which six or seven lysines have been mutated to arginines failed to show any significant phenotype compared with wild type (Feng et al. 2005; Krummel et al. 2005). Recently, however, a re-evaluation of the Trp53 mouse revealed that the C-terminal lysines may be important for HSC survival, although this is only revealed after DNA damage-dependent activation (Wang et al. 2011). While the present study was under consideration, Simeonova et al. (2013) reported that mice expressing a truncated p53 lacking the C-terminal 31 amino acids presented a severe aplastic anemia, leading to death within weeks of birth, similar to what was observed with the p53ΔCTD-expressing mice reported here. These findings and the current data support the hypothesis that p53 C-terminal post-translational modifications are likely to be indispensable for proper p53 function and strongly argue for a central role of p53 in the maintenance and homeostasis of the adult hematopoietic compartment through its basic C-terminal domain. This central and novel role for the C terminus is supported by the observation that the hematopoietic phenotype is triggered by the mere deletion of the C terminus of p53 in the absence of genotoxic stress. Moreover, the data presented here shed light on the mechanisms leading to the hematopoietic failure in Trp53 animals.

According to these data, p53 activity is critical in regulating hematopoietic homeostasis in vivo, in agreement with several previous studies (Liu et al. 2007, 2009; Abbas et al. 2010; Wang et al. 2011; Ceccaldi et al. 2012). Both fetal and postnatal hematopoietic tissues are affected by the C-terminal deletion, although to differing extents. Although the absolute number of fetal liver HSCs is reduced in the mutant animals, these cells are still capable of properly differentiating and, to a certain extent, reconstituting the bone marrow of lethally irradiated recipients. Flow cytometry analysis of gross CD3+, CD19+, and CD11b+ populations in P10 organs revealed aberrant cellular subset composition in postnatal hematopoietic tissues, with an overall increase in CD3+ T-cell populations and decrease in CD19+ B-cell populations of mutant p53 mice (Supplemental Fig. 7). Consistent with their observed colony-forming capacity (Fig. 4B), p53 mutant HSCs were capable of differentiating toward major lymphoid and myeloid blood lineages, yet subsequent lineage development and/or homeostasis is somehow disrupted. Hematopoiesis fails by P10 in these mutant p53 mice, leading to death within a time frame coincident with the transition from liver to bone marrow-predominated hematopoiesis. HSCs migrate from the fetal liver to the bone marrow niche, which provides the appropriate cellular and molecular microenvironment for their self-renewal and differentiation (Suda et al. 2005). While the bone marrow of the Trp53 mice is virtually devoid of HSCs, fetal liver cells from Trp53 mice are capable of engrafting and rescuing lethally irradiated recipient mice, although not as efficiently as wild-type cells. These data raise the intriguing possibility that p53 might also play important roles in such non-cell-autonomous functions as formation or maintenance of the bone marrow stem cell niche. The migration of HSCs to the bone marrow as well as the maintenance of the bone marrow stem cell niche are critically dependent on the chemokine CXCL12 (Suda et al. 2005; Ding and Morrison 2013; Greenbaum et al. 2013). As p53 activation has been demonstrated to attenuate cancer cell migration through repression of CXCL12 (Moskovits et al. 2006), it is tempting to speculate that p53ΔCTD hyperactivity within the bone marrow niche could cause CXCL12 repression, thereby ablating the CXCL12 chemokine gradient and undermining the bone marrow stem cell niche in the Trp53 mice. Further studies will be needed to determine the extent of p53 contribution in these processes.

Mdm2 negatively regulates p53 protein levels through ubiquitination and subsequent proteasomal degradation. Cell-based studies initially demonstrated that the p53 C-terminal lysines are required for this process (Kubbutat et al. 1997, 1998). Studies in vivo later contradicted these results, however, by showing that mutation of these C-terminal lysines that had been shown to be targeted by ubiquitination had no effect on p53 stability (Feng et al. 2005; Krummel et al. 2005). This could reflect species differences between humans and mice as well. Here it is shown that p53ΔCTD protein levels are differentially regulated in a tissue-specific manner. In the thymus, cerebellum, and adult liver, the wild-type and truncated p53 proteins are expressed at comparable levels. In contrast, in the spleen, brain cortex, and fetal liver, the p53ΔCTD protein is found at significantly higher levels. This is reminiscent of previous studies showing that in the setting of a hypomorphic Mdm2 (Malek et al. 2011) or haploinsufficiency for Mdm2 and Mdm4 (Terzian et al. 2007), p53 shows tissue-specific hyperactivity in the hematopoietic compartment and the cerebellum. These findings and the current data support a model in which p53 activity is tightly regulated in specific organs through both its C-terminal domain and its negative regulators, Mdm2 and Mdm4. This is corroborated by one study in which the deletion of the C-terminal domain of p53 decreased the Mdm2–p53 interaction (Poyurovsky et al. 2010). Further investigations will clearly be needed to clarify the relationship between p53 and its negative regulators in Trp53 animals.

Abnormal levels of several relevant markers (including the cell cycle-dependent kinase inhibitors Cdkn2b/p15 and Cdkn1a/p21, the latter being a direct p53 target gene) are correlated with high levels of senescence in the bone marrow of mutant animals. Previous studies have shown that HSC quiescence is maintained by Cdkn1a (p21) (Cheng et al. 2000). These results are substantiated by other studies that have proposed that p53 counteracts stem cell reprogramming by activating a senescence-promoting stress response (Hong et al. 2009; Utikal et al. 2009). Indeed, a similar phenotype was found in mice that express an endogenous mutant p53 (172P) in the absence of Mdm2 (Liu et al. 2007). This particular mutant p53 protein lacks the ability to mediate an apoptotic response but can still up-regulate p21 and cause cell cycle arrest (Ryan and Vousden 1998). This is consistent with the hematopoietic phenotype being mediated through hyperactivity of the p53-dependent cycle arrest pathway via p21. Interestingly, Cdkn2b (p15) has been shown to be repressed by Gfi1, and its knockout in mice leads to monocytosis and predisposition to myeloid leukemia (Basu et al. 2009; Bies et al. 2010). In the Trp53 mice, down-regulation of Gfi1 in bone marrow cells is correlated with higher expression of Cdkn2b (p15) and senescence. It is tempting to speculate that the senescent cells observed in the mutant bone marrow might correspond to HSCs. In fact, the percentage of senescent cells in the Trp53 bone marrow roughly equals the commonly admitted percentage of stem cells among whole bone marrow cells in mice (∼0.05%) (Fig. 5A). Careful analysis of the bone marrow cell composition will be needed to substantiate this hypothesis.

In addition to controlling levels of expression, the C-terminal domain of p53 has been extensively studied in transcriptional regulation (Kruse and Gu 2009). It has alternatively been shown to exert either a negative or positive effect depending on the particular study and conditions (Kruse and Gu 2009; Carvajal and Manfredi 2013). Its highly basic amino acid content is consistent with an interaction with DNA. Indeed, studies have suggested that it can negatively regulate by interfering with sequence-specific DNA binding by the core domain (Hupp et al. 1992; Anderson et al. 1997; Ahn and Prives 2001). Alternatively, it is required to serve as a means for p53 to track along DNA (McKinney et al. 2004). Post-translational modifications within this domain have also been implicated in the binding of specific transcriptional cofactors (Barlev et al. 2001; Mujtaba et al. 2004). Studies with the Trp53 serve to reconcile these apparently contradictory findings in the literature in that the C terminus appears to mediate each of these effects but in a tissue-specific and target gene-specific manner.

In the thymus, wild-type and truncated p53 are not expressed at significantly different levels. Here, the expression of Bbc3 (Puma) and Pmaip1 (Noxa) is enhanced in the homozygous mutant mice (Fig. 6A). This correlates with an increased occupancy on the corresponding genomic sites by ChIP (Fig. 6D). Such a finding is consistent with the idea that in this tissue, the C terminus somehow interferes with sequence-specific DNA binding and thereby attenuates gene expression. This is selective for Bbc3 (Puma) and Pmaip1 (Noxa), as this effect is not seen with Cdkn1a (p21) (Fig. 6). In contrast, in the liver, where, again, wild-type and mutant p53 are not expressed at appreciablely different levels, the mRNA levels for Tigar are significantly decreased (Fig. 8A), and this is associated with less occupancy on the endogenous Tigar gene by the mutant p53 as compared with the FL wild-type protein (Fig. 8D). This latter finding supports a mechanism in which the C terminus is required for sequence-specific DNA binding and subsequent gene activation. Finally, in the liver, Mdm2 shows reduced gene expression in mutant tissues (Fig. 8A), although gene occupancy is comparable between wild-type and truncated p53 proteins (Fig. 8D). This suggests that the differential regulation is at a step subsequent to DNA binding, most likely because of impaired cofactor recruitment. How does one reconcile such tissue- and target-specific effects? It is possible that existing epigenetic landscapes are established in a p53-independent but tissue-specific manner. This combined with the efficiency of post-translational modifications of the C terminus being different between tissues may thus lead to distinct requirements for target gene expression.

It is intriguing that the overt phenotypic effects of the deletion of the C terminus were restricted primarily to the hematopoietic compartment in homozygous mutant mice. (Figs. 2–4). Since many p53 activities are mediated through the modulation of the expression of target genes, an attractive hypothesis is that such tissue-specific phenotypes rely on cell type-specific target genes. The expression pattern of Gfi1 and Gfi1b, two essential regulators of hematopoiesis (Hock et al. 2004; van der Meer et al. 2010), is highly restricted to hematopoietic organs (Tong et al. 1998). Both genes are down-regulated in Trp53 hematopoietic tissues (Fig. 4D). One of these regulators (Gfi1) was previously shown to be a direct p53 target gene involved in the maintenance of HSC quiescence in a p53-dependent manner (Liu et al. 2009). More recently, Gfi1 was shown to directly interact with p53 and impair its proapoptotic functions in thymocytes, suggesting a possible negative feedback loop (Khandanpour et al. 2013). This would support a model in which Gfi1 interacts with the C terminus of p53, thereby impairing p53 functions. Further characterization of the p53/Gfi1b relationship would be informative. In fact, a rapid survey of the promoter and the first intron of this gene revealed several potential bona fide p53-responsive elements in these regions (data not shown).

Recently, small peptides derived from the p53 C terminus have been proposed as anti-tumorigenic therapeutic agents (Snyder et al. 2004). The data here raise concerns about the use of therapeutic approaches based on small molecules that disrupt the Mdm2–p53 interaction (Vassilev 2007). Increases in p53 protein activity may have unintended consequences on the normal hematopoietic tissues of patients.

In summary, the C-terminal domain of p53 was shown to play key roles in the postnatal homeostasis of the hematopoietic compartment and development of the brain in vivo. Further studies are required to establish the precise molecular pathways at play. Nevertheless, the data indicate that HSCs are highly sensitive to p53 activity and that deletion of the C terminus of p53 can affect key functions and ultimately induce abnormal senescence in mutant bone marrow. Furthermore, the exquisite tissue and target specificity that is observed in these mutant p53 mice have important implications for the role of therapies leading to p53 activation in treatment of tumors of differing origins.

Materials and methods

Generation of the Trp53NEO/NEO and the Trp53ΔCTD/ΔCTD mice

The Trp53 mouse was generated by means of a targeting vector harboring two exon 11s separated by a NEO selection cassette (see Fig. 1A; Supplemental Material). Trp53 mice were crossed with PrmCre mice (a gift from Dr. Philippe Soriano) (O'Gorman et al. 1997) to generate Trp53-expressing males. These males were intercrossed with wild-type C57BL/6J females to generate Trp53 heterozygous mice. Ultimately, these heterozygous mice were bred to obtain Trp53 homozygous animals.

Histopathology and IHC

Mice were placed in 10% formalin for 48 h and then decalcified before sectioning into ∼3- to 4-mm coronal sections. Sections were processed and embedded en bloc and stained with haematoxylin and eosin (H&E). One femur from each genotype (Trp53, Trp53, and Trp53) was used to harvest bone marrow for bone marrow brush smears and stained with Diffquik for cytologic evaluation. IHC was performed using the anti-p53 CM5 antibody (Leica Microsystems).

qRT–PCR and immunoblotting

Whole organs were harvested at the indicated times and processed as follows. For RNA preparation, a small sample was excised and submerged in RNAlater RNA stabilization reagent (Qiagen) for later use. The tissue was disrupted and homogenized (PowerGen 125 homogenizer, Fisher Scientific). Total RNA and cDNA were prepared as previously described (Hamard et al. 2012). For protein preparation, a small sample was excised and homogenized into a lysis buffer composed of 50 mM HEPES (pH 7.5), 1% Triton X-100, 150 mM NaCl, 1 mM MgCl2, 1 mM phenylmethylsulfonyl fluoride (PMSF), 5 mg/mL leupeptin, and 50 mg/mL aprotinin. One-hundred micrograms of total protein was resolved by SDS-PAGE. Immunoblot analysis was conducted using the following antibodies: anti-p53 (CM5, Leica Microsystems) and anti-β-actin (Sigma).

Flow cytometry and HSC and fetal liver cell isolation

For determining the percentage of LSK cells, total bone marrow cells were preincubated with 5% rat serum, followed by incubation with c-Kit-FITC, Sca-1-PE, and a biotin-conjugated cocktail containing anti-CD3, anti-CD11b, anti-CD45R/B220, anti-Ly6G, and anti-TER119 (all from BD Biosciences), followed by APC-conjugated streptavidin labeling (eBiosciences). At the end of incubation, the stained cells were treated with DAPI to exclude dead cells and were acquired on either a LSR-II or Fortessa (BD Biosciences). For fetal liver cell staining, pregnant mice were sacrificed at E13.5, fetal liver cells were isolated, and single-cell suspensions were prepared by passing through a cell strainer (70 μM). For fetal liver LSK analysis, anti-CD11b was excluded from the multilineage cocktail. All of the data were analyzed with FlowJo 9.5 software (Tree Star).

Colony-forming unit assay

About 2 × 105 fetal liver cells from E14.5 fetuses of different genotypes were grown on methylcellulose-based medium (MethoCult M3434, Stem Cell Technologies) and maintained in minihumidity chambers in 20% or 5% oxygen tissue culture incubators. After 2 wk, colonies were classified and enumerated based on their morphology. Three types of colonies were counted: colony-forming unit-granulocyte and macrophage (CFU-GM), burst-forming unit-erythroid (BFU-E), and CFU-granulocyte, erythroid, macrophage, and megakaryocyte (CFU-GEMM).

SA-β-Gal assay

Bone marrow cells from P10 whole bone marrow were collected and cytospun on slides. Cells were stained with Senescence-Associated β-Galactosidase Staining kit (Cell Signaling) according to the manufacturer's protocol. Using an inverted microscope and a camera, the number of blue cells were counted and averaged, and a representative picture was taken.

Activated caspase 3 assay

Bone marrow cells from P10 whole bone marrow were collected and cytospun on slides. MEFs derived from E14.5 Trp53, Trp53, Trp53, or Trp53 embryos were grown on coverslips and treated with DOX (Sigma) for 24 h. Both cell types were stained with an anti-cleaved caspase 3 antibody (Cell Signaling, no. 9661S), and immunofluorescence analysis was conducted as previously described (Hamard et al. 2012).

Transplantation experiments

C57/B6 mice were irradiated with a split dose of 6 Gy, 4 h apart. Three hours after the second dose, 0.5 × 106 fetal liver cells from E14.5 Trp53, Trp53, Trp53, or C57/B6 embryos were injected retro-orbitally into ketamine-sedated mice. Mice were monitored for survival on a daily basis. For irradiation control, four mice were irradiated without transplantation for each experiment.

Statistics

Data are represented as means and SEM. Unless otherwise indicated, all experiments were performed in triplicate. A two-tailed student's t-test was used for comparison between two groups. P < 0.05 was considered significant. One-way ANOVA was used for comparison between more than two groups. P < 0.05 was considered significant.

ChIP assay on tissues

Briefly, the organs were dissected from P10 animals, and a single-cell suspension was prepared by crushing them through a 70-μm nylon cell strainer (BD Falcon, catalog no. 352350). Cells were cross-linked in 10 mL of 1% formaldehyde (EM Science, no. FX0415-5) for 10 min at room temperature. The cross-linking reaction was stopped by adding 0.5 mL of 2.5 M glycine to a final concentration of 125 mM for 5 min at room temperature. Cells were spun at 2000 rpm for 5 min, washed once in 1× PBS, and spun again. The remainder of the protocol was previously described (Carvajal et al. 2012) and adapted from Espinosa et al. (2003) using Protein A Dynabeads (Invitrogen) for the pull-down.