Distinct roles of Mdm2 and Mdm4 in red cell production

The p53 protein plays an important role in preventing cancer development by inducing apoptosis or inhibiting the proliferation of cells undergoing cellular stress.¹  p53 performs these functions mainly by acting as a transcription factor, affecting the transcription of a plethora of genes.²  Hence, it is essential to restrain p53 activities during normal growth and development. Central to this process are the 2 structurally related proteins, Mdm2 and Mdm4.³  Germline inactivation of either Mdm2 or Mdm4 leads to embryonic mortality that is completely rescued by concomitant deletion of p53.^4–8  Interestingly, Mdm2-null mice die before implantation due to excessive cell death,⁹  whereas Mdm4-null mice die at midgestation. Mdm4-null mice exhibit cell proliferation defects, particularly in the fetal liver erythroid progenitor cells, and apoptosis specifically in the neuroepithelium,⁸  suggesting that Mdm4 has cell-intrinsic functions in early brain development and erythropoiesis. In contrast, recent data indicate that Mdm2 might be dispensable for red cell production. The red blood cell count was indeed nearly normal in mice with one null and one hypomorphic Mdm2 allele.¹⁰  However, frequency of p53-dependent apoptosis was increased in a subset of homeostatic tissues in these mice. These data combined suggest that Mdm2 and Mdm4 are nonoverlapping critical regulators of p53 that may exert their functions in a temporal and tissue-specific manner. To examine this possibility we have specifically inactivated Mdm2 and Mdm4 in the erythroid lineage in vivo

The Mdm2 and Mdm4 floxed alleles, and the GFPcre knock-in allele, controlled by the endogenous erythropoietin receptor (EpoR) promoter, were genotyped as described.^11–13 

Pregnant females were injected intraperitoneally with 100 μg of BrdU per gram of body weight and killed 2 hours later. Assays were performed as previously described.¹⁴  Detection of phosphorylated histone H3 (1:30; Calbiochem, San Diego, CA) and BrdU-positive cells (1:200; Boehringer Mannheim, Mannheim, Germany) was performed as described.^8,15  Light microscopy was performed using a BX51 Discussion microscope (BX51, Olympus) equipped with the following objectives: 10 ×/0.25 numerical aperture (NA), 20 ×/0.4 NA, 40 ×/0.65 NA, and 100 ×/1.25 NA (oil). Images were captured with a color camera cooled interline transfer CCD (Coolsnap color camera, Roper Scientific, Tucson, AZ) (1392 × 1040 pixel resolution; Olympus, Hamburg, Germany). RSI image software (Roper Scientific, Tucson, AZ) version 1.7.3 was used for image acquisition.

Embryos were fixed overnight in 0.2% paraformaldehyde/1 × PBS, incubated in 2 mM MgCl₂/30% sucrose/1 × PBS overnight, and embedded in OCT; 6 μm cryosections were prepared. Immunohistologic analysis was performed as described¹⁶  using anti-p53 (1:100, CM5; Novocastra Laboratories, Newcastle upon Tyne, United Kingdom), anti–cleaved caspase-3 (1:100; Cell Signaling Technology, Beverly, MA), and anti–Ter-119 (1:100, clone Ter-119; eBioscience, San Diego, CA). Microscopy was performed with an Axiophot fluorescence microscope (Axiophot; Zeiss, Jena, Germany) equipped with a 40 ×/1.3 NA oil objective and a color camera (RTE/CCD-1300-Y/HS; Roper Scientific) cooled interline transfer CCD (1030 × 1300 pixel resolution; Carl Zeiss, Jena, Germany). Metamorph online software (Molecular Devices, Downingtown, PA) version 5.0 was used for image acquisition.

Mice were injected subcutaneously on days 0, 1, and 3 with 35 mg/kg of phenylhydrazine in PBS. Blood was collected on days 0, 3, 6, and 9 for measurement of hematocrit and reticulocyte count as described.¹⁷ 

Erythroid cells expressing the cellular marker Ter119 were purified from bone marrow by magnetic-activated cell sorting (MACS) as described.¹³ 

Cells were plated and the colonies were scored as described.^18,19 

Whereas no Mdm2^lox/lox; EpoR^GFP-Cre/+ pups were found at weaning age, about 60% of the expected Mdm4^lox/lox; EpoR^GFP-Cre/+ mice were viable (Tables S1–S2, available on the Blood website; see the Supplemental Materials link at the top of the online article). Examination of embryos at successive stages of embryonic development revealed that Mdm2^lox/lox; EpoR^GFP-Cre/+ embryos die at about embryonic day (E) 13. This phenotype is p53 dependent, because Mdm2^lox/lox; EpoR^GFP-Cre/+; p53^−/− mice showed no erythropoietic defects and were viable (Figure 1A; Figure S1). At E12, Mdm2^lox/lox; EpoR^GFP-Cre/+ embryos appeared very pale (Figure 1A). Only rare erythroid cells were found in the heart, dorsal aorta, and liver (not shown) of these embryos, and most of them (> 90%) exhibited pyknotic nuclei, characteristic of apoptotic cells (Figure 1C). In contrast, numerous erythropoietic foci are found in the heart and liver of E12 Mdm4^lox/lox; EpoR^GFP-Cre/+ embryos (not shown), and at E13, these embryos are undistinguishable from control littermates (Figure 2A). Consistent with the anatomic and histologic analyses, the number of nucleated cells in livers was lower in Mdm2^lox/lox; EpoR^GFP-Cre/+ embryos, but not in Mdm4^lox/lox; EpoR^GFP-Cre/+ embryos, compared with control embryos.

The number of erythrocytes and erythroid colony-forming units (CFU-Es) in the blood vessels of the yolk sac of Mdm2^lox/lox; EpoR^GFP-Cre/+ embryos was dramatically reduced compared with control embryos (Figure 1B). In contrast, primitive erythropoiesis appears unaffected in Mdm4^lox/lox; EpoR^GFP-Cre/+ mutants (Figure 1B). Importantly, the efficiency of recombination was comparable at the Mdm2 and Mdm4 loci of the EpoR^GFP-Cre/+ mutants (Figure S2).

Most erythroid cells were apoptotic in mdm2^lox/lox; EpoR^GFP-Cre/+ embryos, both at E10.5 in the blood islands of the yolk sac and at E12 in the heart and dorsal aorta (Figure 1C). In contrast, apoptosis was not detected in the erythroid lineage of mdm4^lox/lox; EpoR^GFP-Cre/+ mice (not shown). Importantly, whereas no increase in p53 protein levels was detected in erythrocytes from mdm4^lox/lox; EpoR^GFP-Cre/+ embryos (not shown), more than 90% of the nuclei from mdm2^lox/lox; EpoR^GFP-Cre/+ erythrocytes showed marked p53 immunoreactivity (Figure 1D).

Whereas Mdm4 is dispensable for primitive erythropoiesis, it is required for the high erythropoietic rate during definitive erythropoiesis. Mdm4^lox/lox; EpoR^GFP-Cre/+ fetal livers contain fewer CFU-Es (Figure 2B) and erythroid burst-forming units (BFU-Es; not shown) erythroid progenitors than controls. E14.5 mdm4^lox/lox; EpoR^GFP-Cre/+ embryos were paler (Figure 2A), and their hematocrits were 50% lower than that of controls (Figure 2C). This anemia did not cause embryonic death before E12.5 (Table S2); however, about 40% of mutants die, presumably from anemia, between E12.5 and day 21 after birth (P). Importantly, mdm4^lox/lox; EpoR^GFP-Cre/+; p53^−/− did not show any erythropoietic defects (Figure 2A, C). Together, Mdm4 controls, in a p53-dependent manner, a rate-determining step during fetal liver erythropoiesis.

We next examined the cause of fetal anemia in Mdm4^lox/lox; EpoR^GFP-Cre/+ embryos. Because there was no significant increase in apoptosis in Mdm4-deficient erythroid progenitors, we examined their ability to proliferate. The number of phosphorylated histone H3-positive erythroid progenitors was strikingly reduced in Mdm4^lox/lox; EpoR^GFP-Cre/+ embryos (Figure 2D). Moreover, a statistically significant reduction in the percentage of S-phase cells was observed in the dorsal aorta of Mdm4^lox/lox; EpoR^GFP-Cre/+ embryos (Figure 2E). The data indicate that the anemia observed in these embryos is a consequence of decreased cell proliferation. In agreement, expression of p21 and Ptprv, 2 key mediators of p53-induced cell-cycle exit, was slightly but reproducibly increased in fetal liver cells from Mdm4^lox/lox; EpoR^GFP-Cre/+ embryos (not shown).

To determine whether Mdm4 continues to function beyond fetal life, we measured hematocrit, reticulocyte count, and the number of bone marrow–erythroid progenitors in surviving adult Mdm4^lox/lox; EpoR^GFP-Cre/+ mice. Surprisingly, these mice have normal hematocrit (Figure 2F), and similar numbers of bone marrow CFU-E and BFU-E erythroid progenitors compared with control mice (Figure S3). Moreover, after administration of phenylhydrazine, an agent that induces hemolytic anemia,²⁰  the magnitudes and kinetics of reticulocyte counts and hematocrit fluctuations were similar in controls and Mdm4 mutants (Figure 2F). Thus, adult Mdm4^lox/lox; EpoR^GFP-Cre/+ mice are not deficient in generating high erythropoietic rates in response to erythropoietic stress. Importantly, the deleted (loxP) allele is largely predominant over the nondeleted (floxed) allele in Ter119-positive cells sorted from Mdm4^lox/lox; EpoR^GFP-Cre/+ bone marrow (Figure S3D). This result strongly argues against a specific selection of Mdm4-competent erythoid cells in these mice. Together, these data indicate that Mdm4 is specifically required for fetal erythropoiesis, but not for production of adult red cells.

We show in this study that Mdm2 has an essential function during primitive erythropoiesis in rescuing erythroid progenitors from p53-mediated apoptosis. In contrast, Mdm4 is dispensable for primitive erythropoiesis, but is required for generating high erythropoietic rates during definitive fetal erythropoiesis. Several recent reports have described the conditional inactivation of Mdm2 and Mdm4 in other cell types, including cardiomyocytes, progenitor and postmitotic neuronal cells, and smooth muscles of the gastrointestinal (GI) tract.^12,14,21,22  As in our present study, inactivation of Mdm2 consistently led to a more pronounced activation of p53 function and consequently to phenotypes more remarkable than those caused by loss of Mdm4. Depending on the cell type, loss of Mdm4 leads to phenotypes ranging from not obvious, as in cardiomyocytes or smooth muscle cells of the GI tract, to minor, as in proliferating and postmitotic neuronal cells.²³  So far, the data have been consistent with a model in which Mdm2 is the master regulator of p53, with Mdm4 functioning to assist Mdm2 only in specific cell types. The data herein are partly consistent with this model, but they also highlight the ability of Mdm4 to function, in a given tissue, specifically at a particular phase of the differentiation program. Mdm4 is indeed required only for the massive expansion phase of erythroid progenitor cells during definitive erythropoiesis. These cells undergo rapid, multiple DNA replication cycles with little time for DNA repair mechanisms to operate, conditions that favor p53 activation. In this context, it is not surprising that additional brakes on p53, such as that provided by Mdm4, are active in these cells. Notably, in adult mice Mdm4 is not required for steady-state erythropoiesis or for the increase in erythropoietic rate in response to stress. This observation further supports the existence of molecular differences between fetal and adult erythropoiesis.^24,25 

Conflict-of-interest statement: The authors declare no competing financial interests.

Correspondence: Jean-Christophe Marine, Laboratory for Molecular Cancer Biology, VIB Technologiepark, 927, B-9052 Ghent, Belgium; e-mail: chris.marine@dmbr.ugent.be.

We thank Dieter Defever and Ines Bonk for excellent technical assistance. S.F. and P.F. were supported by grants from FNRS-Télévie and Fondation pour la Recherche Médicale, respectively. This work was supported in part by grants from the Association for International Cancer Research, Belgian Foundation against Cancer (nonprofit organization), and by EC FP6 funding (ACTIVEP53, contract 503576). This publication reflects only the authors' views. The commission is not liable for any use that may be made of the information herein.