ABSTRACT
The function of the p53 tumor suppressor to inhibit proliferation or initiate apoptosis is often abrogated in tumor cells. Mdm2 and its homolog, Mdm4, are critical inhibitors of p53 that are often overexpressed in human tumors. In mice, loss of Mdm2 or Mdm4 leads to embryonic lethal phenotypes that are completely rescued by concomitant loss of p53. To examine the role of Mdm2 and Mdm4 in a temporal and tissue-specific manner and to determine the relationships of these inhibitors to each other, we generated conditional alleles. We deleted Mdm2 and Mdm4 in cardiomyocytes, since proliferation and apoptosis are important processes in heart development. Mice lacking Mdm2 in the heart were embryonic lethal and showed defects at the time recombination occurred. A critical number of cardiomyocytes were lost by embryonic day 13.5, resulting in heart failure. This phenotype was completely rescued by deletion of p53. Mice lacking Mdm4 in the heart were born at the correct ratio and appeared to be normal. Our studies provide the first direct evidence that Mdm2 can function in the absence of Mdm4 to regulate p53 activity in a tissue-specific manner. Moreover, Mdm4 cannot compensate for the loss of Mdm2 in heart development.
p53, which acts as a guardian of genome integrity, is activated in response to genotoxic stress, directing the cell to undergo cell cycle arrest, DNA repair, or apoptosis (34). The activation of these pathways prevents the proliferation of errors in the genome during replication and cell division. Conversely, mutation or deletion of p53 allows uncontrolled proliferation and the perpetration of genetic errors. The fact that approximately 50% of human tumors have a mutation in the p53 tumor suppressor gene supports this role (14).
The activity of p53 is negatively regulated by numerous proteins (22). The functional significance of two of these, Mdm2 and Mdm4, has been examined with mouse models. Mice lacking Mdm2 die early in development (18, 25). Embryonic death occurs before implantation as a result of the activation of the p53-dependent apoptotic pathway in blastocysts (3). Concomitant deletion of p53 completely rescues this lethal phenotype. Loss of Mdm4, the gene encoding a second p53 inhibitor, also results in an embryonic lethal phenotype that is rescued by loss of p53 (7, 23, 28).
While clearly Mdm2 and Mdm4 are both potent p53 inhibitors, the relationship between them is complex and entangled. Mdm2 is an E3 ubiquitin ligase that catalyzes ubiquitination of itself and p53 (13, 15, 20). Mdm4, on the other hand, does not appear to ubiquitinate p53 (17). Both Mdm2 and Mdm4 bind the same p53 domain with similar affinities (2). The relationship between Mdm2 and Mdm4 is more complex, as Mdm4 was identified with yeast two-hybrid screens by using Mdm2 as bait (30, 32). In transient-transfection experiments, Mdm2 and Mdm4 interact through their RING domains. This interaction has two major effects: (i) it pulls Mdm4, a cytoplasmic protein, into the nucleus, and (ii) it inhibits the E3 ligase activity of Mdm2, allowing stabilization of p53 and Mdm2. Since Mdm2 and Mdm4 interact, bind the same domain of p53, and yet have different effects on p53, the ratio of these two proteins to each other should determine the outcome of p53 regulation. Indeed, when cells have higher levels of Mdm2 than of Mdm4, p53 is ubiquitinated and unstable. When cells have higher levels of Mdm4 than of Mdm2, stable but inactive p53 is present (12). One must keep in mind, though, that all of these data are from overexpression experiments producing supraphysiological levels of Mdm2 and Mdm4 and that most studies use tagged versions of the proteins, which may affect their activities as well. Nevertheless, these in vitro studies raised questions as to the ratios of Mdm2 and Mdm4 in normal development and in specific tissues in the regulation of p53 activity.
A delicate balance, controlled by multiple pathways, is critical to maintaining p53 at appropriate levels. While absence of p53 results for the most part in a normal mouse embryo, too much p53 results in developmental abnormalities. The Mdm2 and Mdm4 null phenotypes in mice are classic examples of developmental defects due to constitutive p53 activity. Mdm2 null mice die by initiating p53-dependent apoptosis at embryonic day 3.5 (E3.5), while two different alleles of Mdm4 initiate p53-dependent cell cycle arrest and/or apoptosis at later developmental time points (3, 23, 28). Interestingly, these results suggest that the role of the p53 inhibitors in regulating p53 function in apoptosis and cell cycle arrest during embryonic development may vary in a temporal and tissue-specific manner. To examine this possibility, we generated Mdm2 (11) and Mdm4 conditional alleles (this study).
To probe the functional significance of Mdm2 and Mdm4 on apoptosis and proliferation, we used the α-myosin heavy chain promoter driving Cre expression in the developing heart. During embryonic development, mononucleated contractile cardiomyocytes proliferate (29). Differentiation of cardiomyocytes occurs shortly after birth, and eventually these cells withdraw from the cell cycle. Shaping of the embryonic heart involves a balance of apoptosis and proliferation, which continue postnatally in cardiomyocytes but cease by adulthood (5, 19). We chose the α-myosin heavy chain because it is cardiac specific, it is expressed beginning at E8.5, and its expression lasts throughout development (6, 21, 27). We deleted Mdm2 and Mdm4 in cardiomyocytes by crossing the α-myosin heavy chain promoter-driven Cre mouse (αMyHC-Cre) (1) with Mdm2 and Mdm4 conditional alleles, and we show tissue-specific differences for the roles of Mdm2 and Mdm4 in heart development.
MATERIALS AND METHODS
Generation of Mdm4 conditional allele.The pLG plasmid (8) with modifications contains two frt sites flanking a pgk-neo cassette; two loxP sites were added outside of the pgk-neo cassette. A 0.9-kb EcoRI fragment containing Mdm4 exon 2 was cloned into the vector, as was a 1.3-kb intron 1 EcoRI/SalI fragment as the left arm and a 5.2-kb intron 2 EcoRI/SphI fragment as the right arm. The M. D. Anderson Genetically Engineered Mouse Facility conducted targeting of embryonic stem (ES) cells. DNA isolated from ES cells was digested with EcoRI and probed with a 3′ external probe. The positive clones from the mini-Southern blot were further confirmed by PCR using a 5′ external primer. Chimeric mice were mated to C57BL6/J mice. To delete the pgk-neo cassette, Mdm4+/neo mice were mated to Rosa26-Flipper mice. Mating of Mdm4+/FX and CMV-Cre mice (Jackson Laboratories) generated Mdm4 +/ Δ 2 mice.
PCR genotyping.ES clones were amplified using primer F1, 5′-TTAACCAGTTGGGTCCATCTCTTCTGG-3′, and a primer within neo cassette neo19, 5′-GCTATCAGGACATAGCGTTGGC-3′. The following primers were used to distinguish between the Mdm4 conditional and null alleles: F4, 5′-TAGAATCTGGAATTACAGACAG-3′; In1re, 5′-TGTCTTTAGCATTTACTAAGAGCT-3′; and In2re, 5′-TATCCAGTGTCCTCTTCTGGCTT-3′. Another set of primers was used to distinguish Mdm4 wild-type and conditional alleles: F3, 5′-GGTGTCCTTGAACTTGCTGTGTAGAA-3′, and E2re, 5′-CTGGGCCGAGGTGGAATGTGATGT-3′. The Cre transgene was genotyped using the following primers: cre-up, 5′-TCCAATTTACTGACCGTACACCAA-3′, and cre-dn, 5′-CCTGATCCTGGCAATTTCGGCTA-3′. The p53 wild-type and mutant alleles were genotyped as previously described (16). Primers A, B, and C were previously described and amplify the following alleles: wild-type Mdm2, Mdm2Δexons 5-6, and Mdm2FM (11).
β-Galactosidase (β-gal) staining of embryos.Embryos at various stages of development were fixed for 1 h, rinsed, and stained overnight as described previously (26). Embryos were embedded in paraffin, and sections (7 μm thick) were mounted on Superfrost Plus slides (Fisher Scientific, Pittsburgh, PA). Slides were counterstained with eosin.
Histology, TUNEL staining, and immunohistochemistry.Embryos at E9.5 were fixed overnight in 4% (vol/vol) paraformaldehyde and were embedded in paraffin. Sections (7 μm thick) were mounted on Superfrost Plus slides and were stained with eosin and hematoxylin. Terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) assays were carried out on paraffin-embedded embryo sections as described previously (9), modified by using ABC and DAB kits from Vector Laboratories (Burlingame, CA). Slides were counterstained with methyl green.
The mouse sections were subjected to immunohistochemistry by using commercial monoclonal antiactin, α smooth-muscle immunoglobulin G (1:200) (Sigma, St. Louis, MO), CM-5 (1:200) (Vector Laboratories, Burlingame, CA), and Ki-67 (1:200) (Santa Cruz Biotechnology, Santa Cruz, CA). Visualization was performed using the ABC and DAB kits named above. Counterstaining was performed with nuclear fast red.
Western blot analysis.Cell lysates (1or 2 mg) from wild-type and p53 −/− Mdm4 Δ2/Δ2 mouse embryo fibroblasts were precleared with protein A agarose (Sigma, St. Louis, MO) and immunoprecipitated with an Mdm4 polyclonal antibody (J. A. Barboza and G. Lozano, unpublished data) or a green fluorescent protein antibody (Santa Cruz Biotechnology, Santa Cruz, CA) as the control, overnight. Western blot analysis was performed with the same Mdm4 antibody.
RESULTS
Generation of Mdm2 and Mdm4 conditional alleles.The generation of the Mdm2 conditional allele (Mdm2 FM) has been described previously (11). It was designed to delete Mdm2 exons 5 to 6, which encode the p53-binding domain (Fig. 1A). Mdm2 +/FM mice were mated to CMV-Cre mice to generate the Mdm2 Δ exons 5 - 6 allele that results from Cre recombination in the germ line. When heterozygous Mdm2 Δ exons 5 - 6 mice were mated with each other, no homozygous Mdm2 Δ exons 5 - 6 mice were born, indicating that this allele also resulted in embryonic lethality. Dissection of embryos at E8.5 showed approximately 25% empty deciduas, a phenotype identical to that of the Mdm2 null embryo. Loss of p53 also rescued the Mdm2 Δ exons 5 - 6 homozygous phenotype. Thus, loss of Mdm2 from the conditional allele led to the same embryonic lethal phenotype as was previously described for an Mdm2 null allele (25).
Mdm2 and Mdm4 conditional alleles. (A) The Mdm2 conditional allele was designed to delete exons 5 and 6, which encode part of the p53-binding domain (11). (B) Targeting strategy for the Mdm4 conditional allele deletes the first coding exon. Exons are numbered and represented as black boxes. Arrowheads denote locations and directions of PCR primers. Frt sites are shown as filled circles, while loxP sites are represented as diamonds. R1, EcoRI; NEO, neomycin resistance gene; TK, herpes simplex virus thymidine kinase gene. (C) Southern blot analysis of targeted ES cell lines. DNA from ES cells was digested with EcoRI. Blots were hybridized with an exon 4 probe and show the presence of a 10-kb fragment in correctly targeted ES cells. (D) PCR analysis of targeted ES cell lines. PCRs were performed using primer F1 located outside of the left arm of homology and primer neo19 within the neomycin gene. Positive clones had a 2.2-kb diagnostic PCR product. M, 1-kb-plus DNA ladder; nc, negative control. (E) Immunoprecipitation (IP) and Western blot analysis of lysates from Mdm4 Δ 2/ Δ 2 p53 − / − mouse embryonic fibroblasts. Cell lysates were immunoprecipitated with the indicated antibodies followed by Western blot analysis with the Mdm4 antibody. WT, wild type; DN1 and DN2, Mdm4 Δ 2/ Δ 2 p53 − / − cell lines 1 and 2, respectively; GFP, green fluorescent protein; IgG, immunoglobulin G. The arrow indicates Mdm4.
The Mdm4 conditional allele was generated using both loxP and frt sequences (Fig. 1B). The neo gene was flanked by frt sequences. loxP sites were placed in introns 1 and 2 to generate a deletion of exon 2, the first coding exon. ES cells were examined for homologous recombination by using Southern blot analysis and PCR amplification (Fig. 1C and D). Thirteen of 96 ES clones analyzed were positive, and two different ES cell clones contributed to the germ line of mice. One was studied in detail. Mice inheriting the targeted Mdm4 +/neo gene were crossed with the Rosa-Flipper mouse (4) to remove the neo cassette. Double mutant mice were mated back to C57BL6/J females to generate the conditional allele Mdm4 +/FX. Approximately 90% of the progeny from the backcrosses had the neo cassette completely deleted. Mice homozygous for the Mdm4 conditional allele (Mdm4 FX/FX) were generated and showed no obvious defects.
To generate mice with a deletion of Mdm4 from the conditional allele and examine their phenotype, we crossed the Mdm4 +/FX mouse to a CMV-Cre mouse. Mice heterozygous for this Mdm4 Δ 2 allele were normal. Mdm4 +/ Δ 2 mice were intercrossed to examine the phenotype of Mdm4 Δ 2/Δ 2 mice. None of the 42 offspring born were Mdm4 Δ 2/Δ 2, indicating that the deletion of exon 2 of Mdm4 results in an embryonic lethal phenotype. Dissection of embryos at E9.5 showed that all Mdm4 Δ 2 / Δ 2 embryos were abnormal, similar to the phenotype for the original Mdm4 null mutation (28). We also tested whether the embryonic lethality of Mdm4 Δ 2 / Δ 2 mice can be rescued by loss of p53. The Mdm4 Δ 2 allele was bred into a p53 null background. The Mdm4 Δ 2/ Δ 2 p53 − / − mice were viable and born at approximately the expected ratio. We also performed Western blot analysis using mouse embryo fibroblasts from Mdm4 Δ 2/ Δ 2 p53 − / − mice and detected no Mdm4 protein (Fig. 1E). Thus, extensive analyses of the Mdm2 and Mdm4 conditional alleles indicated that deletion with Cre recombinase resulted in null alleles that had phenotypes identical to those of the original null alleles.
Loss of Mdm2 in embryonic heart is lethal.Since shaping of the embryonic heart involves a balance of apoptosis and proliferation, we decided to examine the roles of the p53 inhibitors Mdm2 and Mdm4 in heart development. We deleted Mdm2 and Mdm4 in the mouse embryonic heart by using the αMyHC-Cre mouse (1). αMyHC-Cre mice contain the α-myosin heavy-chain promoter driving expression of Cre recombinase in mature cardiomyocytes. To formally test whether the αMyHC-Cre transgene expressed Cre in the heart during embryogenesis, we crossed the αMyHC-Cre mouse to the ROSA26 reporter line that contains a lacZ gene that is not expressed unless Cre recombination occurs (31). αMyHC-Cre ROSA26 embryos were collected at various stages and stained for β-gal activity. LacZ staining was first observed at E9.5 and was specific for the heart (Fig. 2A and B). Staining became more prominent at later embryonic stages but remained heart specific (Fig. 2C and D). Thus, the αMyHC-Cre transgene expressed Cre beginning at E9.5 and expression was cardiac specific.
Characterization of the αMyHC-Cre transgenic line during embryogenesis. αMyHC-Cre mice were mated with the ROSA26 reporter line. Embryos at different developmental stages were stained with X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) solution. (A) Histological section of E8.5 embryo. Arrowheads indicate heart structures. (B) Histological section of blue-stained E9.5 embryo. V, ventricle. (C) Histological section of blue-stained E12.5 embryo. (D) Whole-mount blue-stained E12.5 embryo.
Mice homozygous for the Mdm2 FM allele were crossed to Mdm2 +/ − mice (25) that contained the αMyHC-Cre gene. From this cross, 38 total progeny were genotyped and none was Mdm2 −/FM αMyHC-Cre (Table 1). If viable, 25% of the mice (approximately 10 mice) from this cross should have been Mdm2 −/FM αMyHC-Cre. Thus, loss of Mdm2 in the heart resulted in an embryonic lethal phenotype.
Loss of Mdm2 in the heart is lethala
Embryos at various embryonic stages were obtained by crossing an Mdm2 FM/FM female to an Mdm2 +/ − αMyHC-Cre male to determine the timing of embryonic lethality. All Mdm2 −/FM αMyHC-Cre embryos died by E13.5, and most lacked any visible signs of a heart (data not shown). All embryos at E9 were histologically normal (see Fig. S1 in the supplemental material). Abnormalities appeared only in Mdm2 −/FM αMyHC-Cre embryos as early as E9.5, the same time that heart-specific Cre recombination was first detected. Embryos at E9.5 showed a significant thinning of the myocardial layer in the ventricles, which eventually led to heart failure and embryonic death (Fig. 3A and B). Immunohistochemistry with an α smooth-muscle actin antibody, a marker that stains cardiomyocytes at this embryonic stage, showed the presence of cardiomyocytes in the mutant embryo at E9.5, but the heart mass was smaller in size (Fig. 3C and D). The embryonic cardiomyocytes in the ventricles were affected more than those in the atrium of the heart, as shown by LacZ staining with the ROSA26 reporter (data not shown). Careful review of these data showed Cre recombination was strongest in the ventricles. Many of the mutant embryos were also smaller than wild-type embryos, probably due to lack of blood flow. However, mutant embryos developed to the correct developmental stage based on somite count.
Analysis of sections of normal and Mdm2 −/FM αMyHC-Cre mutant embryos. Normal (A, C, E, and G) and mutant (B, D, F, and H to J) embryos are shown at E9.5 (A to D) and E13.5 (E to J). All images are hematoxylin and eosin-stained sections, except for images in panels C and D, which were immunostained with α smooth-muscle actin antibody to label the cardiomyocytes. Yellow bars indicate widths of ventricles (V) and atria (A) in normal and mutant hearts. Arrows in panels E and F indicate the location of the heart, and the arrow in panel H shows abnormal blood deposition. (I) View of Mdm2 −/FM αMyHC-Cre mutant atrium. (J) View of Mdm2 −/FM αMyHC-Cre mutant liver.
One of the few Mdm2 −/FM αMyHC-Cre embryos at E13.5 that still had a heart showed abnormal trabeculation in the ventricle and abnormal ventricular wall structure (Fig. 3E to H). The mutant heart failed to function properly, as indicated by blood leaking outside the heart and the backup of blood in the atrium (Fig. 3H and I). In addition, the decreased blood flow resulted in congestion and backup of blood in other organs (Fig. 3J).
To examine mechanistically the loss of cardiomyocytes, we performed apoptosis and proliferation assays with embryos at E9.5, the earliest developmental stage that showed defects. To determine if the abnormal heart phenotype was due to apoptosis, TUNEL assays were performed. A rare apoptotic cell appeared in the normal heart as expected, but apoptosis was much higher in the Mdm2 mutant embryo (Fig. 4A and B). To examine proliferation, sections of normal and abnormal hearts were stained with Ki-67 (Fig. 4C and D). Even though fewer cells were present in the Mdm2 −/FM αMyHC-Cre embryo hearts, they retained the ability to proliferate. Immunohistochemistry was also used to examine the presence of p53. Increased p53 levels were detected in some cells of the mutant heart but not in the normal heart (Fig. 4E and F). The presence of only a small number of p53-positive cells was probably due to the fact that cells were dying. In conclusion, the absence of Mdm2 in the heart results in increased p53 levels, leading to the loss of myocardial cells, particularly in the ventricles. The loss of these myocardial cells results in an abnormal heart structure that eventually leads to embryonic lethality.
Analysis of apoptosis and proliferation in heart sections of normal and Mdm2 −/FM αMyHC-Cre mutant embryos at E9.5. Normal (A, C, and E) and mutant (B, D, and F) hearts are shown at E9.5. (A and B) TUNEL assay. (C and D) Immunohistochemistry for Ki-67. (E and F) Immunohistochemistry for mouse p53 protein. A, atrium; V, ventricle.
To determine if recombination of the Mdm2 FM allele is heart specific, DNA was first collected from the tissues of a normal 3-week-old Mdm2 +/FM αMyHC-Cre mouse and analyzed by PCR amplification to detect recombination. Recombination was observed only in the heart (Fig. 5A). Recombination appeared to be incomplete, but that is because cardiomyocytes represent only 15% of the cells at this stage. To examine recombination during embryogenesis, we collected DNA from E9.5 embryos. The recombined Mdm2 Δ exons 5 - 6 allele was present in the Mdm2 +/FM αMyHC-Cre embryo (Fig. 5B). Taken together, the data indicate that loss of Mdm2 specifically in the heart of a developing embryo results in a lethal phenotype.
Heart-specific deletion of Mdm2. (A) PCR of DNA isolated from the tissues of an Mdm2 +/FM αMyHC-Cre mouse at 3 weeks of age. Primers A, B, and C are depicted in Fig. 1A. M, 1-kb-plus DNA ladder; ht, heart; lu, lung; li, liver; tl, tail; sk mu, skeletal muscle; wt, wild type. (B) Same PCR as described above, performed on DNA isolated from E9.5 embryos. C, control Mdm2 +/FM sample; R, recombined Mdm2 +/FM αMyHC-Cre sample.
To determine if the lethal phenotype is p53 dependent, the p53 null allele (16) was crossed into the Mdm2 FM background. Of 30 progeny genotyped from a cross between Mdm2 FM/FM p53 − / − and Mdm2 +/− p53 + / − αMyHC-Cre mice, 6 were Mdm2 −/FM p53 − / − αMyHC-Cre (Table 2). These mice had the same life span as p53 − / − and p53 − / − Mdm2−/− mice (data not shown). Importantly, haploinsufficiency at the p53 locus did not rescue the phenotype (Table 2). Thus, the absence of p53 rescued the heart lethal phenotype.
Loss of p53 rescues heart lethalitya
Loss of Mdm4 in heart does not result in embryonic lethality.Mice homozygous for the conditional Mdm4 FX allele were crossed to Mdm4 +/ Δ 2 αMyHC-Cre mice. Progeny from these crosses showed the correct ratio of Mdm4 Δ 2/FX αMyHC-Cre mice at weaning (Table 3). To confirm recombination of the Mdm4 FX conditional allele by the Cre transgene, DNA was purified from the organs of Mdm4 FX/FX αMyHC-Cre mice. As shown in Fig. 6, the null specific Mdm4 product was present only in the heart but not in the lung, liver, tail, or skeletal muscle. These data indicate that recombination occurred and was specific to the hearts of double mutant mice. Again, recombination did not appear to be very robust, but only 15% of the cells at this developmental stage are cardiomyocytes. Additionally, since we generated mice with a floxed Mdm4 allele in the presence of a null Mdm4 allele, any recombination event would result in an Mdm4 null cell. No discernible differences in the embryonic hearts of normal and Mdm4 mutant embryos were observed (Fig. 7). Thus, loss of Mdm4 in the heart was not embryonic lethal. Some Mdm4 Δ 2/FX αMyHC-Cre mice died within 1 year. The reason for this shortened life span is unknown at the moment.
Deletion of Mdm4 in the heart produced normal morphology at E9.5. Whole-mount β-gal staining was performed on E9.5 mouse embryos. (A) Mdm4 +/ − αMyHC-Cre ROSA26 mouse. (B) Mdm4 −/FX αMyHC-Cre ROSA26 mouse. V, ventricle.
Absence of a phenotype in hearts lacking Mdm4 a
DISCUSSION
Studies published to date have not examined the role of Mdm2 or Mdm4 in adult tissues or in a tissue-specific manner. Toward this goal, we deleted Mdm2 and Mdm4 in cardiomyocytes specifically to address their importance in a developmental system that undergoes apoptosis and proliferation. The loss of Mdm2 resulted in lethality, while surprisingly loss of Mdm4 did not. Since we used crosses in which one Mdm2 (or Mdm4) null allele was combined with the floxed conditional allele, every recombination event yields an Mdm2 (or Mdm4) null cell. One possible explanation for the absence of a phenotype with loss of Mdm4 is that Mdm4 is not expressed in cardiomyocytes. However, in situ hybridization with Mdm4 probes shows clear expression of Mdm4 mRNA in the heart of an E11.5 embryo (23). Thus, these data provide the first indication that Mdm2 and Mdm4 have different roles in a specific cell type.
At least four p53 inhibitors have been identified thus far, and we do not yet understand the need for so many p53 inhibitors. We do, however, know that the levels of active p53 must be tightly regulated to avoid cell lethality. Several in vivo experiments support this hypothesis. Ectopic expression of wild-type p53 in the kidneys, mostly during embryogenesis, hinders the differentiation of the ureteric bud, leading to smaller kidneys and subsequent renal failure (10). Mice heterozygous for the p53 gene that produced a truncated active form of p53 displayed increased resistance to spontaneous tumors and early onset of aging-associated phenotypes (33). Additionally, loss of either Mdm2 or Mdm4 results in embryonic lethal phenotypes that are rescued by loss of p53 (7, 18, 23, 25, 28). The p53 dependency suggests that the inability to down-modulate p53 activity in the absence of Mdm2 or Mdm4 is lethal. Thus, in several examples, too much p53 results in cell death.
Biochemical experiments have indicated that Mdm2 and Mdm4 form a heterocomplex (30, 32) and that this complex is critical for inactivation of p53 (12). In fact, in vitro studies by Gu et al. (12) suggest that Mdm2 and Mdm4 are mutually dependent. Mdm2 is functionally inefficient in the absence of Mdm4 due to the targeting of itself for degradation. Heterocomplex formation between the two proteins results in an increase in Mdm2 stability by hindering autoubiquination, thus enhancing Mdm2's ability to target p53 for degradation. In some studies, Mdm4, which lacks a nuclear localization signal, is dependent on Mdm2 to take it to the nucleus (24). The studies described here indicate that Mdm2 can, in the absence of Mdm4, function as a perfectly good p53 inhibitor. The reverse was not true; Mdm4 could not compensate for loss of Mdm2.
Marine and Jochemsen (22) described various models for the roles of Mdm2 and Mdm4 in negatively regulating p53 in vivo. Among them was the possibility that Mdm2 was the main regulator of p53 and Mdm4 acted as a cofactor at particular tissues or stages of differentiation, such as during the massive expansion of progenitor cells. Mdm4 would assist Mdm2 in the rapidly dividing cells to reduce p53 activity. Another proposed model is that Mdm2 and Mdm4 function independently in different cell types, although this was believed to be less likely. The analyses presented here, which indicate that loss of Mdm4 has no effect on heart development while that of Mdm2 results in abnormal heart structure and embryonic lethality, support the model that Mdm2 and Mdm4 are independent regulators of p53 in heart development. Thus, we provide the first direct evidence that Mdm2 and Mdm4 may have independent roles in a temporal and tissue-specific manner.
ACKNOWLEDGMENTS
We thank Michael Schneider for the αMyHC-Cre mice.
This study was supported by a grant from the National Institutes of Health (CA47296) to G.L. and a training grant (CA009299) supporting J.D.G.
FOOTNOTES
- Received 20 June 2005.
- Returned for modification 31 July 2005.
- Accepted 12 October 2005.
- Copyright © 2006 American Society for Microbiology
