p53 Activity Dominates That of p73 upon Mdm4 Loss in Development and Tumorigenesis

This article is featured in Highlights of This Issue, p. 1

The tumor-suppressing function of p53 is dampened in the majority of human cancers (1). This alteration can be through deletion or mutation of the TP53 gene itself, or its functional inactivation via upregulation of its negative regulators, MDM2 and MDM4 (2). Several studies and multiple mouse models have elucidated the importance of Mdm2 and Mdm4 as bona fide regulators of p53 in embryonic development, normal physiology, DNA damage, and tumorigenesis. Mdm2-null mice die at embryonic day (E) 3.5 prior to implantation, whereas Mdm4-deficient mice arrest at pregastrulation and die at E7.5. These embryonic lethal phenotypes are completely rescued by deletion of p53 (3–5). Moreover, deletion of either Mdm2 or Mdm4 in neural progenitor cells of mouse is neonatal lethal. Mdm2 loss in the brain results in hydranencephaly due to apoptosis, whereas Mdm4 deficiency causes porencephaly due to both apoptosis and cell cycle arrest. Both phenotypes are rescued by deletion of p53. Interestingly, concomitant deletion of Mdm2 and Mdm4 in developing central nervous system (CNS) resulted in a more severe phenotype compared with loss of each gene alone, revealing that these proteins are not functionally redundant (6). In addition, although Mdm2^+/− or Mdm4^+/− mice have a normal life span, both are radiosensitive and, upon DNA damage, succumbed to p53-dependent early death (7). Furthermore, Mdm2^+/− Mdm4^+/− double-heterozygous mice are not viable due to neural and hematopoietic defects. This phenotype is rescued by loss of a single p53 allele, underscoring the fine-tuning in the p53/Mdm2/Mdm4 regulatory network and the cell-type–specific sensitivity (7). Lastly, conditional deletion of Mdm2 in adult mice is also lethal due to extensive p53-induced irreversible tissue damage (8, 9), whereas deletion of Mdm4 under similar condition causes reversible pathologies and is compatible with life (10). Many human tumors have high levels of MDM2 and MDM4 due to either gene amplification or overexpression. For instance, elevated levels of MDM4 are detected in retinoblastoma (11, 12), melanoma (13), breast cancer (14, 15), glioblastoma (16), and hematopoietic neoplasms, such as pre-B acute lymphoblastic leukemia in adults (17) and acute myeloid leukemia (18). The major role of these proteins in tumor development and progression is thought to be inhibition of p53 activity, as a large number of tumors with high MDM4 levels retain wild-type p53. Mdm4 exerts its inhibitory effect via physical masking of the transactivation (TA) domain of p53 protein, and, unlike Mdm2, appears to have no direct role in controlling the level of p53 protein by degradation.

As a member of the p53 family, p73 is a transcription factor with substantial structural and functional homology to p53. Having the highest homology in the DNA-binding domain (63%), as well as identical residues that interact with DNA, allows p73 to bind to the consensus p53-binding site, activates common target genes, and mediates apoptosis and cell-cycle arrest (19). However, unlike p53, p73 is mutated in less than 0.5% of human tumors (20). Other studies have shown that p73 is altered in various tumor types through LOH or epigenetic silencing in tumors, such as lymphoma and leukemia (21–24), glioma (25), neuroblastoma (26), and breast cancer (27). Such observations imply p73 functions as a tumor suppressor. In fact, genetic studies clearly highlight this function as p73^+/− mice develop spontaneous tumors (28). Also, the combination of p73 loss with loss of other tumor-suppressor genes, such as p53, or presence of an oncogene, such as Myc, enhances tumor burden and/or increases tumor dissemination and metastasis (28–30). Although both p53 and p73 belong to the same family, distinct differences between them have been identified. For instance, genetic studies revealed that although p53-null mice are for the most part viable without prominent developmental defects, p73-deficient mice have serious developmental abnormalities, such as chronic infections and inflammation, infertility due to abnormalities in pheromone signaling, and neurological defects, including hippocampal dysgenesis, cortical hypoplasia, and hydrocephaly (31).

Given the significance of Mdm2 and Mdm4 in regulating p53 activity in vivo and the high structural homology of p53 and p73 in the TA domain, several studies examined the interaction of Mdm proteins with p73. Coexpression of Mdm2 and p73 in human embryonic kidney (HEK-293) cells with wild-type p53 as well as p53-deficient human osteosarcoma (Saos-2) and lung adenocarcinoma (H1299) cells showed that similar to the p53–Mdm2 interaction, p73 and Mdm2 bind via their N-termini (32–34). Such binding reduces p73 transcriptional activity (33, 35) and, in turn, p73-induced apoptosis (32). However, p53 and p73 bind to separate regions on Mdm2 (35), and, unlike p53, p73 is not destabilized or degraded by Mdm2 (32, 33). Moreover, recently, an in vivo study proposed that Mdm2 exerts its oncogenic activity through inhibition of both p53 and p73 functions (36). Mdm4, like Mdm2, also similarly binds p73 (34); yet, whether such interaction represses p73 transcriptional activity remains to be elucidated. Interestingly, a detailed quantitative characterization of interaction of Mdm proteins with p53-family proteins has revealed that Mdm4 has higher affinity for p73 than p53 (37). To interrogate the biologic significance of the Mdm4–p73 interaction in vivo, here, we present a comprehensive genetic characterization of the potential Mdm4 and p73 interaction during both development and tumorigenesis.

p73^+/− mice (ref. 28; from Dr. Elsa Flores, MD Anderson Cancer Center, Houston, TX), Mdm4^Δ2/+ mice (5), and p53^+/− (38) mice on a mixed background (129/SvJae and C57BL/6) were used for our studies. p73^+/− mice (28), Mdm4^fx/+ mice (5), and Nes-Cre transgenic mice (from The Jackson laboratory) on C57BL/6 background were used for studying CNS development. Mdm4^Tg15 mice (39) and p73^+/− mice (28), both on C57BL/6 background, were crossed to generate the cohort for tumor study. All mice were maintained in the mouse facility at the University of Texas M.D. Anderson Cancer Center in compliance with the Institutional Animal Care and Use Committee guidelines. Mice were sacrificed upon morbidity or after reaching the 2-year endpoint of the study.

Whole embryos were fixed in 4% paraformaldehyde and paraffin embedded for sagittal sectioning. Collected organs were fixed in 10% formalin and paraffin-embedded for sectioning. Sections were dewaxed and rehydrated according to standard protocols and stained with hematoxylin and eosin (H&E). Antibodies used for immunohistochemistry were as follows: Ki67, 1:100 (ab16667; Abcam); caspase-3, 1:200 (9661S; Cell Signaling Technology); CD45R/B220, 1:50 [Clone RA3-6B2 (RUO); BD Pharmingen]; CD3, 1:100 (ab5690; Abcam); and Lysozyme, 1:1,000 (NBP1-95509; Novus). Subsequently, stained sections were detected with VECTASTAIN Elite ABC Reagent and Vector DAB substrate (Vector Laboratories) and counterstained with nuclear fast red (Vector Laboratories).

Total RNA was isolated from embryonic brain at E14.5 using TRIzol reagent (Ambion), treated with DNase I (Roche Life Science), and then reverse transcribed by using the First-Strand cDNA Synthesis Kit (GE Healthcare). Quantitative real-time PCR was performed according to the manufacturer's instructions (Bio-Rad). Expression was normalized to Rplp0.

All statistical analyses were performed using GraphPad Prism 6 software, and a P value of <0.05 was considered statistically significant. The difference between observed and expected frequencies of embryos was determined by χ² test. The multiple comparisons test for gene expression analyses of embryonic brain was performed by ANOVA followed by the Newman–Keuls multiple comparison test. The difference between survival curves was determined by using the log-rank (Mantel–Cox) test, and the difference of the tumor spectrum of each group of mice was analyzed by the Fisher exact test.

Given that Mdm4 is essential for dampening p53 activity during embryogenesis and that Mdm4 binds to p73 with relatively higher affinity compared with p53 in in vitro studies, we first examined whether the Mdm4–p73 axis is critical for embryonic development. We intercrossed Mdm4^Δ2/+ (the Mdm4^Δ2 allele represents a true null allele as opposed to the Mdm4⁻ allele, which deletes exons 3 to 5 and leads to production of a small protein; see ref. 2) and p73^+/− mice to obtain Mdm4^Δ2/Δ2 p73^−/− embryos and examine whether Mdm4^Δ2/Δ2 embryonic lethality at E7.5 could be rescued in a p73-null background. To overcome the issue of p73^−/− infertility, we used p73 heterozygous mice as breeders, and consequently, the frequency of Mdm4^Δ2/Δ2 p73^−/− embryos was expected to be 6.25%. At E11.5, we collected 36 embryos with normal morphology, 2 resorbed embryos, and 16 empty deciduae. We did not obtain any Mdm4^Δ2/Δ2 p73^−/− embryo with normal morphology; however, one of two resorbed embryos was Mdm4^Δ2/Δ2 p73^−/− and the other was Mdm4^Δ2/Δ2. The phenotypes of both were comparable, indicating that p73 loss does not rescue Mdm4^Δ2/Δ2 phenotype at this developmental stage. Further, we studied the rescue at an earlier time point, E9.5, and obtained 37 embryos with normal morphology as well as 5 resorbed embryos. The normal embryos were wild-type or heterozygous for Mdm4, whereas the resorbed embryos were Mdm4-null with various p73 alleles [Mdm4^Δ2/Δ2 (1), Mdm4^Δ2/Δ2 p73^+/− (1), and Mdm4^Δ2/Δ2 p73^−/− (3) embryos]. In addition, we observed no significant difference in the morphology of Mdm4^Δ2/Δ2 p73^−/− compared with Mdm4^Δ2/Δ2. Combined, Mdm4^Δ2/Δ2 embryos, regardless of the p73 genotype, were remnants of resorbed embryo with no recognizable morphology (Table 1; χ² = 30; df = 5; P value < 0.0001), indicating that loss of p73, in the presence of unrestricted p53 activity, does not rescue the Mdm4-null phenotype.

We next examined the contribution of both p53 and p73 to the rescue of Mdm4^Δ2/Δ2 early embryonic lethality. We examined the role of p53 gene dosage in the rescue of Mdm4^Δ2/Δ2 early embryonic lethality and the effect of p73 loss on the phenotype of Mdm4^Δ2/Δ2 p53^+/− embryos. We crossed Mdm4^Δ2/Δ2 p53^−/− with Mdm4^Δ2/+ p73^+/− mice to test whether 50% reduction in p53 gene dosage extends the lifespan of Mdm4^Δ2/Δ2 embryos and concomitantly whether deletion of one allele of p73 suffices to further advance embryonic development. At E11.5, we observed 10 Mdm4^Δ2/+ p53^+/− embryos with normal morphology as well as 17 Mdm4^Δ2/Δ2 p53^+/− embryos with an aberrant developmental phenotype, suggesting that deletion of one allele of p53 cannot completely rescue the Mdm4^Δ2/Δ2 phenotype. However, decreased p53 dosage in Mdm4^Δ2/Δ2 p53^+/− embryos did delay the Mdm4-null phenotype. Although these embryos were still severely runted compared with Mdm4^Δ2/+ p53^+/− embryos (Supplementary Fig. S1A–S1C), they overcame pregastrulation arrest, as evidenced by the formation of neural groove with neural folds on each side as well as allantois, a conspicuous landmark of gastrulation (40). The observation of a partial rescue due to haploid loss of p53 is in agreement with findings from another Mdm4 knockout mouse model (41). Moreover, Mdm4^Δ2/Δ2 p53^+/− and Mdm4^Δ2/Δ2 p53^+/− p73^+/− embryos had comparable phenotypes, indicating that loss of one allele of p73 is not sufficient to rescue Mdm4^Δ2/Δ2 p53^+/− phenotype (Supplementary Table S1; χ² = 10; df = 3; P value = 0.0186).

To determine whether the complete loss of p73 can extend the survival of Mdm4^Δ2/Δ2 p53^+/− embryos, we crossed Mdm4^Δ2/Δ2 p53^−/− p73^+/− with Mdm4^Δ2/+ p73^+/− mice and analyzed embryos at E9.5. We obtained 18 Mdm4^Δ2/+ p53^+/− embryos, which were morphologically comparable with wild-type, and 28 Mdm4^Δ2/Δ2 p53^+/− embryos, which were abnormally small (Table 2; χ² = 30.000; df = 5; P value < 0.0001). The Mdm4^Δ2/Δ2 p53^+/− embryos in the background of wild-type or null p73 had the same phenotype (Supplementary Fig. S1D–S1F), suggesting that p73 loss has no impact on the Mdm4^Δ2/Δ2 phenotype.

Given the strong dependence of Mdm4 loss on p53 activity in early development, we next focused on the possible cooperation of Mdm4 and p73 during development of the brain, an organ wherein both Mdm4 and p73 have well-established roles during embryogenesis. Mdm4 plays an important role in neural development by regulating p53 (41). Conditional deletion of Mdm4 in the developing CNS by Nestin-driven Cre recombinase (Nes-cre) starts as early as E10.5 (6, 42) and results in porencephaly, a cavity in the cerebrum, and pups are not viable after birth due to severe brain-tissue deficiency. This phenotype is a gradually progressive neuronal loss due to both cell-cycle arrest and apoptosis and rescued by deletion of p53 (6, 43). On the other hand, p73 is well known as a multifunctional protein in the field of neurobiology (44) and crucial for the maintenance of neural stem cells (45, 46). p73 is expressed sparsely in preplate, the first stage in corticogenesis, as early as E10.5, and it is highly expressed throughout the cortical hem and in Cajal–Retzius (CR) cells at E12.5 (31, 47). However, the phenotype of p73 deficiency in the brain is not evident during prenatal life, except for mild cortical hypoplasia after E14.5, and the core triad, including cortical hypoplasia, hippocampal dysgenesis, and ventriculomegaly, is only prominent in postnatal and adult mice (48).

Given that both Mdm4 and p73 have established roles in embryologic development of brain, we tested whether Mdm4 is a negative regulator of p73 during CNS development and examined if the phenotype of Mdm4 deficiency in the CNS could be rescued by p73 loss. For this study, we used the Mdm4^fx conditional allele (5). We crossed Mdm4^fx/fx p73^+/− mice with Mdm4^fx/+ p73^+/− Nes-cre mice to generate Mdm4^fx/fx p73^−/− Nes-cre embryos. Embryos were collected at E14.5, the earliest time point when apoptosis and cell cycle arrest due to lack of Mdm4 are apparent in the embryonic brain (6). In total, 121 embryos were collected (Supplementary Table S2). The histopathologic studies showed that the phenotype of porencephaly in Mdm4^fx/fx Nes-cre (n = 3) and Mdm4^fx/fx p73^−/− Nes-cre (n = 5) embryonic brains was comparable (Fig. 1A–C). Moreover, Mdm4^fx/fx p73^−/− Nes-cre brains, similar to Mdm4^fx/fx Nes-cre, had decreased proliferation and increased apoptosis rates, as evidenced by Ki67 and cleaved caspase-3 (CC3) staining, and there was no difference in the pattern of proliferation and apoptosis between the two genotypes (Fig. 1D–F and 1G–I). These results clearly show that even in a tissue wherein p73 has prominent functions, deletion of p73 does not rescue the Mdm4-null phenotype.

Given the strong evidence of the Mdm4–p73 interaction in vitro, we decided to interrogate the inhibitory regulation of Mdm4 on p73 activity at the molecular level in our system. Several studies have been performed to identify p73-specific target genes in various cell types or cancer cell lines under various conditions. In each set of experiments, different target genes have been identified (46, 49, 50). Thus, we decided to examine, in our experimental model system, whether p73 target genes known to play a role in the CNS were aberrantly expressed. Thus, the embryos with the following genotypes were used for gene expression analysis by real-time RT-qPCR: Mdm4^fx/+ and Mdm4^fx/fx (wild-type embryonic brain), Mdm4^fx/+ p73^−/− and Mdm4^fx/fx p73^−/− (p73^−/− embryonic brain), Mdm4^fx/fx Nes-cre (Mdm4^Δ2/Δ2 embryonic brain), Mdm4^fx/fx p73^−/− Nes-cre (Mdm4^Δ2/Δ2 p73^−/− embryonic brain). The Mdm4^Δ2/Δ2 p53^−/− embryonic brains at E14.5 were obtained from intercrossing Mdm4^Δ2/Δ2 p53^−/− mice.

14-3-3σ is a p53 target gene that inhibits the G₂–M progression and induces cell-cycle arrest (51). Overexpression of p73 upregulates the transcript of 14-3-3σ two to six times higher than does p53 overexpression (49, 50). p73 also directly binds to the consensus p53/p73-binding site in the regulatory region of 14-3-3σ gene (49), suggesting that 14-3-3σ might be a bona fide target of p73. More importantly, a comprehensive study on the differential expression of 14-3-3 protein isoforms showed that 14-3-3σ is located in the nuclei of developing rat hippocampus neurons (52) as well as human hippocampus (53), corresponding to the region that p73^−/− and TAp73^−/− mice display brain defects. In our analyses, 14-3-3σ was highly expressed in Mdm4^Δ2/Δ2 embryonic brain compared with wild-type (P value < 0.01), and deletion of p73 or p53 abrogates such transcriptional upregulation (P value < 0.01 and <0.05, respectively; Fig. 2A), indicating that both p53 and p73 regulate 14-3-3σ.

Perp is another p53 target gene, which mediates apoptosis (54) in a cell context–dependent manner (55). More specifically, studies on Perp-deficient mice have shown that Perp exerts its p53-dependent apoptosis in developing CNS as well as thymocytes; however, it is dispensable for apoptosis in E1A-expressing mouse embryo fibroblasts (55). Four p53-binding sites have been recognized within the promoter and the first intron of the murine Perp gene (56), and it had been suggested that p73 might bind to these sites as well (57). Thus, we examined the effect of Mdm4 loss on Perp expression in the presence and absence of p53 or p73. Perp was significantly upregulated in Mdm4^Δ2/Δ2 embryonic brain compared with wild type (P value < 0.01), whereas loss of either p53 or p73 abolished its expression (P value < 0.01 and <0.05, respectively; Fig. 2B).

p21 is the well-characterized canonical transcriptional target of p53 (58, 59) and a major mediator of p53-dependent cell-cycle arrest. It was also recognized as one of the first p73 target genes in cells overexpressing wild-type p73; yet, unlike p53, p73 does not induce p21 in response to DNA damage (60). Moreover, the level of p21 induced by p53 was three to six times higher than p73 (50). In our system, p21 was only upregulated in Mdm4-deficient embryonic brain in a p53-dependent manner (P value < 0.01; Fig. 2C).

We also examined a number of p73-specific target genes, which modulate neurogenesis, including p57 (61, 62), Jag1 (63), Jag2 (46, 49, 63), Hes5 (45, 46), and Notch2 (46); however, we did not observe any statistically significant increase in expression of these targets as a result of Mdm4 loss (data not shown). Combined, these observations highlight the dominance of p53 activity, which results in extensive cell-cycle arrest and apoptosis in Mdm4-deficient embryonic brain and compromises other pathways in the developing CNS.

Given that overexpression of Mdm4 and loss of p73 have been implicated in tumor development, we explored whether these two alterations cooperate during tumorigenesis. We generated a cohort of Mdm4^Tg15 p73^+/−, Mdm4^Tg15, p73^+/−, and wild-type mice in C57BL/6 background. Because p73^−/− mice die at an early age due to malnutrition and growth retardation rather than tumor development, we did not include Mdm4^Tg15 p73^−/− and p73^−/− mice in the cohort. Mdm4^Tg15 is one of three generated Mdm4 transgenic mouse lines which develop spontaneous tumors (39). At the 2-year endpoint of this study, 18 of 23 (78.3%) Mdm4^Tg15 p73^+/− mice had been sacrificed, compared with 15 of 17 (88.2%) Mdm4^Tg15 and 1 of 13 (7.7%) p73^+/− mice. Both Mdm4^Tg15 p73^+/− and Mdm4^Tg15 mice had significantly shorter life span compared with p73^+/− and wild-type littermates (P value < 0.0001), and their median survival was 593 and 665 days, respectively. However, the difference between the survival curves of Mdm4^Tg15 p73^+/− and Mdm4^Tg15 mice was not statistically significant (Fig. 3A). We further examined the tumor frequency and spectrum of each group. Note that 78.2% (18 of 23) of Mdm4^Tg15 p73^+/− and 82.3% (14 of 17) Mdm4^Tg15 mice develop tumors by 2 years of age, compared with 23% (3 of 13) of p73^+/− mice. Moreover, Mdm4^Tg15 p73^+/− and Mdm4^Tg15 mice had comparable tumor phenotypes, mainly B-cell lymphoma, histiocytic or dendritic cell sarcoma, and brain tumors, whereas p73^+/− mice developed B-cell lymphoma, histiocytic or dendritic cell sarcoma, and angiosarcoma (Fig. 3B; Table 3). Unlike p73^+/− tumors that show approximately 40% LOH (36), none of the tumors from Mdm4^Tg15 p73^+/− mice underwent LOH (data not shown), suggesting that overexpressed Mdm4 might be inhibiting the p73 wild-type allele in tumorigenesis. Intriguingly, 4 Mdm4^Tg15 p73^+/− mice had unclassified large cell brain tumor compared with 1 Mdm4^Tg15 mouse (20% of Mdm4^Tg15 p73^+/− mice compared with 6.6% of Mdm4^Tg15 mice). Moreover, 5 Mdm4^Tg15 p73^+/− mice had B-cell lymphoma compared with only 1 Mdm4^Tg15 mouse (21.7% of Mdm4^Tg15 p73^+/− mice compared with 5.8% of Mdm4^Tg15 mice). These Mdm4^Tg15 p73^+/− mice also had several disseminated foci of lymphoma in lymphoid and nonlymphoid organs and/or multiple primary tumors (Fig. 4). On the other hand, lymphoma in the Mdm4^Tg15 mouse was localized in only one lymphoid organ, the mesenteric lymph node. Since several studies have suggested a role of p73 in lymphomagenesis (21, 22, 30), increased lymphomagenesis in Mdm4^Tg15 p73^+/− mice indicates a possible cooperation between Mdm4 and p73. Yet, these differences were not statistically significant perhaps due to the small sample size (Fig. 3C).

The observations that Mdm4 is overexpressed in many tumors and most tumors lack p73 mutations suggest that overexpression of Mdm4 may be a mechanism for inactivation of p73 in human cancer. Despite the strong in vitro evidence of Mdm4–p73 binding in mammalian cells and the higher affinity of Mdm4 for p73 rather than p53 in biochemical studies, thus far no studies have examined the biologic significance of this interaction. In this study, using genetic mouse models, we provide the first comprehensive in vivo characterization of the Mdm4 and p73 interaction during development and tumorigenesis.

Our results reveal that loss of p73 does not rescue Mdm4-deficient early embryonic lethality, indicating that in the presence of enhanced p53 activity, such interaction does not have biologic importance in embryogenesis. We further examined the consequences of the Mmd4–p73 interaction in the developing CNS, a more relevant biologic system. We found that Mdm4 inhibits p73-mediated upregulation of 14-3-3σ and Perp, indicating that p73 contributes to the Mdm4-deficient brain phenotypes. However, the effect of unrestricted p53 and, subsequently, the activation of a plethora of target genes are stronger than the effect of p73 and dominate the Mdm4^Δ2/Δ2 p73^−/− brain phenotype. Another explanation for the dominant effect of p53 activity is its expression pattern compared with p73 during CNS development. At E10.5, the expression of p53 is ubiquitous in the mouse brain (64), whereas p73 is expressed sparsely and mostly in the telencephalon (47). At later embryonic stages, p53 expression is more heterogeneous with high levels of p53 mRNA in the ventricular zone neuroepithelial progenitors, telencephalon, and mesencephalon, and low levels in the cortical plate (64), whereas p73 is highly expressed in cortical hem, a structure close to the hippocampus, and more specifically in CR cells (47). Thus, it seems reasonable to postulate that upon deletion of Mdm4, both p53 and p73 have increased activity as evidenced by the upregulation of their target genes, yet each affects distinct parts of embryonic brain. Consequently, p53 expression in more regions of brain dictates the Mdm4^Δ2/Δ2 p73^−/− brain phenotype.

Congenital porencephaly is a rare anomaly in the neonate's brain due to aberrant development of CNS (65). It is characterized as a cavity in cerebral hemispheres and causes severe seizures and mental retardation in patients (65). It is mostly sporadic and has been related to trauma and ischemic injury at mid-gestation (66), but the observations of familial porencephaly suggested that genetics could also be involved in a subset of patients (67). Thus far, dysregulation of the Mdm4–p53 pathway has not been studied directly. However, COL4A1 is among the limited mutations that have been found to be associated with this anomaly (68–70). Surprisingly, this gene is regulated by p53 as well as p73 (71, 72), which indicates that p53 and p73 pathway could be activated in porencephaly. Interestingly, hippocampal atrophy has been reported in 95% of cases with congenital porencephaly (73). Such striking coexistence implies a common mechanism underlying both pathologies; however, it remains to be elucidated. Despite the prominent role of p73 during CNS development and in the maintenance of neural stem cells, there is still no report on the role of this gene in the pathogenesis of congenital brain defects. Together, the observation of porencephaly in Mdm4-null brains, the Mdm4–p73 interaction in developing CNS, and the prominent role of p73 in the formation of the hippocampus suggest that it is worth examining the genetic alterations of Mdm4 and p73 in the subset of patients with porencephaly and hippocampal atrophy.

Our tumor study revealed that either Mdm4 overexpression or p73 heterozygosity is tumorigenic, which is consistent with previous studies (28, 39). Mdm4^Tg15 p73^+/− mice also developed spontaneous tumors, a higher frequency of lymphoma and brain tumors, as well as a strikingly highly disseminated lymphoma compared with each individual alteration. Mdm4^Tg15 mice in our cohort had longer survival and developed tumors with increased latency compared with our published Mdm4^Tg15 cohort study (39). Such difference between two cohorts of one transgenic mouse line might result from the altered pattern of transgene expression over serial breeding and across generations (39). On the other hand, p73^+/− mice in our cohort had prolonged survival along with a very low frequency of tumorigenesis. Such a discrepancy between p73^+/− mice in our cohort and published studies could be due to the genetic background, a factor that has already been shown to affect the rate and spectrum of tumor development in other mouse models, such as p53^−/− mice (74, 75). The p73^+/− mice in our cohort were in C57BL/6, whereas in the previous study, these mice had mixed genetic background of C57BL/6 and 129/SvJ (28). Together, p73 heterozygosity in C57BL/6 background develops tumors with lower frequency compared with mixed background, whereas Mdm4 overexpression develops tumor, yet the rates of lymphoma and brain tumor are very low. Interestingly, the combination of both alterations results in more aggressive phenotype, higher incidence of severely disseminated lymphoma, and higher incidence of brain tumor. The presence of wild-type p73 allele, as evidenced by lack of LOH, in the Mdm4^Tg15 p73^+/− tumors also indicates that either Mdm4 overexpression dampens the tumor suppressor activity of remaining p73 allele or loss of only one allele of p73 in the context of high levels of Mdm4 does suffice for tumor development due to their collaboration through cancer signaling pathway(s). However, the differences in survival curve and tumor spectrum of Mdm4^Tg15 p73^+/− and Mdm4^Tg15 mice are not statistically significant, implying that the dampened p53 activity due to high levels of Mdm4 is the major driver of spontaneous tumorigenesis in these mice and compromises the potential Mdm4–p73 cooperation. Our published study on Mdm4 transgenic mouse lines clearly demonstrates that Mdm4^Tg1 p53^+/− mice have significantly shorter survival compared with Mdm4^Tg1 and p53^+/− mice while they have a longer life span than p53^−/− mice (39). In addition, tumors from Mdm4^Tg1 p53^+/− mice showed a very low rate of LOH (less than 8%) compared with p53^+/− tumors with approximately 50% rate of LOH (76), signifying that the observed difference in the tumorigenesis of Mdm4^Tg1 p53^+/− mice results from cooperation of Mdm4 and p53 and ultimately dampened p53 activity (39). The comparison of Mdm4^Tg15 p73^+/−, Mdm4^Tg1 p53^+/−, Mdm4^Tg15, and Mdm4^Tg1 survival curves of current and previously published studies (39) clearly shows that the Mdm4^Tg1 p53^+/− survival curve is significantly shifted to the left compared with Mdm4^Tg15 and Mdm4^Tg1 survival curves, whereas the survival curve of Mdm4^Tg15 p73^+/− is barely shifted to the left (Fig. 3C). Because the survival and tumor spectrum of Mdm4^Tg15 p73^+/− mice are not significantly different compared with those of Mdm4^Tg15, our data underscore the strength of dampened p53 activity in driving the tumor phenotype in this model, which dominates the possible cooperation between Mdm4 and p73 during tumorigenesis.

No potential conflicts of interest were disclosed.

Conception and design: M. Tashakori, Y. Zhang, G. Lozano

Development of methodology: Y. Zhang, S. Xiong

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Tashakori

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Tashakori, Y. Zhang, S. Xiong, M.J. You, G. Lozano

Writing, review, and/or revision of the manuscript: M. Tashakori, Y. Zhang, S. Xiong, M.J. You, G. Lozano

Study supervision: G. Lozano

This work is supported by NIH grant CA47296 (to G. Lozano), a Cancer Prevention Research Institute of Texas (CPRIT) research training award (RP140106, to M. Tashakori), and Julia Jones Matthews Cancer Research Scholar award (to M. Tashakori). M.J. You is supported in part by NIH/NCI R01 CA164346, CPRIT RP140402, and Center for Genetics and Genomics at The University of Texas MD Anderson Cancer Center.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.