Mdm2/p53 levels in bone marrow mesenchymal stromal cells Is essential for maintaining the hematopoietic niche in response to DNA damage

Mesenchymal stromal cells (MSCs) are a key component of the bone marrow (BM) niche, providing essential support required for maintenance of hematopoietic stem cells. To advance our understanding of physiological functions of p53 and Mdm2 in BM-MSCs, we developed traceable conditional mouse models targeting Mdm2 and/or Trp53 in vivo. We demonstrate that Mdm2 is essential for the emergence, maintenance and hematopoietic support of BM-MSCs. Mdm2 haploinsufficiency in BM-MSCs resulted in genotoxic stress-associated thrombocytopenia, suggesting a functional role for Mdm2 in hematopoiesis. In a syngeneic mouse model of acute myeloid leukemia (AML), Trp53 deletion in BM-MSCs improved survival, and protected BM against hematopoietic toxicity from a murine Mdm2i, DS-5272. The transcriptional changes were associated with dysregulation of glycolysis, gluconeogenesis, and Hif-1α in BM-MSCs. Our results reveal a physiologic function of Mdm2 in BM-MSC, identify a previously unknown role of p53 pathway in BM-MSC–mediated support in AML and expand our understanding of the mechanism of hematopoietic toxicity of MDM2is.


Introduction
Wild-type p53 functions are frequently suppressed by murine double minute 2 (MDM2) protein, an E3 ubiquitin ligase that targets p53 for proteasome degradation 1,2 . Genetic studies by global or tissuespeci c Mdm2 deletion have established the primary function of Mdm2 in regulating p53 and have shown that the lethal phenotype of Mdm2 loss was attributed to increased p53 activity 3 . We previously showed that Nutlin-3a-mediated p53 activation in BM-MSCs downregulates Cxcl12 expression via HIF-1α pathway 4 . Cxcl12, also called stromal cell-derived factor 1 or SDF-1, is secreted by perivascular MSCs and regulates homing, proliferation, and differentiation of hematopoietic stem cells 5,6,7,8 . In addition, while the restoration of p53 activity by MDM2 inhibitors (MDM2i) represents a promising approach in cancer therapy, little information regarding the effect of these drugs on the BM microenvironment is available.
Homozygous deletion of Mdm2 in mature osteoblasts using Col3.6-Cre resulted in skeletal defects, reduced bone formation and lethality before birth 9 . p53-null mice displayed a signi cant increase in the osteogenic differentiation potential of MSCs, mediated in part by upregulation of osteoprotegerin 10 . Mdm2 haploinsu cient mice typically were not distinguishable from wild-type mice except for their hematopoietic failure following ionizing radiation (IR), although it could not be determined whether BM failure after IR was caused directly by cytotoxic effects on hematopoietic stem cells or indirectly through alterations in MSCs-mediated hematopoietic support 11,12 .
In addition to the role of p53 in cell cycle arrest and apoptosis, a growing body of evidence suggests that activation of p53 in the tumor microenvironment following MDM2i treatment could result in stromal senescence and an immunosuppressive microenvironment 13,14,15 . MSCs derived from patients with acute myeloid leukemia (AML), but not normal MSCs, highly express p53, suggesting that the p53 pathway is active in the leukemia microenvironment 16 , although the extent to which p53 activity in MSCs is relevant to AML progression is poorly understood.
Here we investigated whether Mdm2/p53 levels in MSCs contribute to MSCs survival, differentiation, and hematopoietic support in vivo. Several in vivo systems were used to identify Mdm2 functions in MSCs and their support in hematopoietic maintenance. We demonstrated that the hematopoietic toxicity of MDM2i therapy is in part due to p53 activation in MSCs and that deletion of Trp53 in MSCs can prevent such deleterious side effects and improve response to therapy. These observations illustrate a previously unappreciated function of the Mdm2/p53 pathway in differentiation of MSCs and their role in hematopoietic support.

Mdm2 levels are essential for the survival of MSCs
We rst sought to generate a BM-speci c MSC reporter mouse to mark MSCs and perivascular cells in vivo. Because of its previously characterized expression in other settings, we used the Osx-Cre allele 17,18,19 combined with mTmG allele 20 to generate double-transgenic Osx-Cre;mTmG reporter mice. The mTmG allele expresses membrane-localized red uorescence globally in the absence of Cre recombinase and green uorescence speci cally in Cre recombinase-expressing cells. Consistent with previous reports 17,18 , GFP-expressing cells were exclusively present in MSCs that give rise to BM stromal cells and eventually differentiate to osteoblasts and adipocytes (Supplementary Figure S1A). In adult BM, GFP marked the osteoblast lineages including trabecular, endosteal, and periosteal cells. The Lepr+ MSCs and perivascular cells within the BM were green uorescent protein positive (GFP+) (Supplementary Figure   S1B). Flow cytometric analysis of markers for MSCs in GFP+ cells derived from BM showed that the GFP+ cells were partially positive for CD73, CD44, and CD90, suggesting that the Osx-Cre;mTmG reporter accurately marked the population of MSCs (Supplementary Figure S1C). Thus, the population of GFP+ cells marked by Osx-Cre represent the major characteristics of MSCs in postnatal mice.
Next, we evaluated the role of Mdm2 in MSCs by conditional deletion of Mdm2 using Osx-Cre as the driver. We interbred progeny of Osx-Cre;mTmG mice with the Mdm2 allele 21 so that we were able to delete Mdm2 speci cally in MSCs and trace them with GFP ( Figure 1A). Then, we evaluated the bone and hematopoietic phenotype of littermate embryos derived from crossing of the Mdm2 / mice with the Osx-Mdm2 /+ mice. Mice with homozygous deletion of Mdm2, Osx-Cre;Mdm2 / , died shortly after birth due to skeletal malformation and compromised breathing ( Figure 1D). Developing bones in the Osx-Mdm2 / mice displayed enlarged chondrocytes without ossi cation and trabecular bone formation ( Figure 1B).
However, the Osx-Mdm2 /+ mice did not show any abnormalities in trabecular bone formation and the structure of the growth plate ( Figure 1C). Importantly, hematopoiesis was eliminated in the metaphysis of developing bones in Osx-Mdm2 / mice, suggesting that Mdm2 is essential for hematopoietic support by MSCs ( Figure 1E). The hematopoietic cells were distributed throughout the metaphysis of developing bones in Osx-Mdm2 /+ mice ( Figure 1F).
Since mice with homozygous deletion of Mdm2 in MSCs were not viable, we focused on Osx-Mdm2 /+ mice and further characterized their bone development. We used terminal deoxynucleotidyl transferasemediated dUTP nick-end labeling (TUNEL) staining to estimate DNA strand breaks as a marker of cell apoptosis in bone cells. Many MSCs, but not mature osteocytes, in Osx-Mdm2 /+ femurs were TUNEL positive, whereas few TUNEL positive cells were seen in the femurs of Osx-Mdm2 +/+ mice (Figures 1G and 1H). To evaluate the osteoblast differentiation potential of MSCs in Osx-Mdm2 /+ mice, we performed bone densitometry. The trabecular bone volume in Osx-Mdm2 /+ mice was signi cantly decreased signi cantly more than that of similarly-aged Osx-Mdm2 wt mice ( Figure 1I and 1J), suggesting that osteoblast differentiation of MSCs was also attenuated in Osx-Mdm2 /+ mice. Thus, genetic depletion of Mdm2 in MSCs was associated with increased MSCS apoptosis levels and osteopenia in adult mice.
Mdm2 haploinsu ciency in MSCs promotes thrombocytopenia after IR-induced cellular stress We examined whether the Mdm2 level in MSCs was important in hematopoiesis. The population of hematopoietic stem cells de ned as CD150 + Lin -/c-Kit + /Sca-1 + was not signi cantly altered in the BM of Osx-Mdm2 /+ mice compared with control Osx-Mdm2 +/+ mice, suggesting that the hematopoietic support of MSCs was not functionally compromised in Osx-Mdm2 /+ mice (Supplementary Figure S2A).
To determine whether the Mdm2 level in MSCs was important in hematopoietic recovery after cellular damage by irradiation, the Osx-Mdm2 /+ and control mice were irradiated (6 Gy), and peripheral blood was analyzed by cell counter. We included a cohort of hematopoietic-speci c Mdm2 haploinsu cient (Vav-Cre;Mdm2 /+ ) mice, known to be IR-sensitive, as positive controls ( Figure 2A). Analysis of the peripheral blood one week after IR showed a signi cant decrease in peripheral platelet counts after IR in all mice ( Figure 2B). There were no signi cant differences between Osx-Mdm2 /+ (Mdm2 haploinsu cient MSCs) and Vav-Cre;Mdm2 /+ (Mdm2 haploinsu cient hematopoietic stem cells) in response to IR, as both groups developed severe thrombocytopenia. However, platelet counts were signi cantly lower in Osx-Mdm2 /+ mice than in the Mdm2 wt controls (p < 0.001), suggesting that Mdm2 haploinsu ciency in MSCs contributed to platelet production by megakaryocytes during the recovery phase. Analysis of BM sections derived from irradiated mice revealed a striking pancytopenia in Osx-Mdm2 /+ as well as Vav-Cre;Mdm2 /+ mice compared with controls ( Figure 2C). The megakaryocytes were absent in Vav-Cre;Mdm2 /+ mice ( Figure 2D). However, the number of megakaryocytes were comparable between Osx-Mdm2 /+ and the controls, suggesting that the thrombocytopenia after irradiation was due to the functional impairment of megakaryocytes in platelet production rather than their depletion (Figures 2D and 2E). In contrast to Osx-Mdm2 /+ mice, which survived after irradiation, all Vav-Cre;Mdm2 /+ mice died around 2 weeks after IR, likely due to eradication of the hematopoietic stem cells ( Figure 2F). TUNEL staining of BM samples derived from Osx-Mdm2 /+ mice after IR showed apoptosis of hematopoietic cells, whereas the MSCs were completely devoid of TUNEL positivity, indicating that Mdm2 haploinsu cient MSCs survived the DNA damage following irradiation, and that the thrombocytopenia observed in these mice was not due to loss of MSCs ( Figure 2G). In addition, immunostaining showed strong p53 accumulation in MSCs of Osx-Mdm2 /+ mice ( Figure 2H), suggesting that Mdm2 haploinsu ciency results in accumulation of p53 after IR stress. Thus, inhibition of Mdm2 in MSCs contributes to thrombocytopenia after IR.  Figure   3C). Importantly, the BM of mice with deletion of Trp53 in MSCs (Osx-Cre;Trp53 / ) displayed areas of active hematopoiesis and high cellularity ( Figure 3C). The megakaryocytes were completely depleted in Osx-Mdm2 /+ mice, whereas Osx-Cre;Trp53 / mice displayed a signi cantly higher number of megakaryocytes in the BM (p < 0.01, n = 3). Collectively, these data demonstrate that p53 levels in MSCs are functionally important in hematopoietic failure after MDM2i therapy. p53 in MSCs contributes to response to murine MDM2i

Deletion of Trp53 in
Previously, we reported that MSCs derived from AML patients display signi cantly higher p53 protein levels 16 . To determine whether p53 levels in MSCs contribute to the response to MDM2i, we established a traceable syngeneic leukemia model in Osx-Cre;mTmG andOsx-Cre;mTmG;Trp53 / mice and analyzed their survival. Mice were transplanted with leukemia cells originally derived from p53-null mice and transformed by lentivirus-mediated delivery of an oncogenic AML-ETO fusion gene 23 as well as a uorescent transgene to express turquoise uorescence ( Figure 4A). We validated the Cre activity and engraftment of leukemia cells by uorescent microscopic analysis of BM isolated from syngeneic AML mice 10 days after transplant ( Figure 4B). Mice were treated with vehicle or DS-5272 starting on day 3 after transplant for 10 days. Deletion of p53 in MSCs signi cantly prolonged mice's survival beyond discontinuation of therapy (p < 0.003), suggesting that survival of AML cells in response to DS-5272 might depend on p53 levels in MSCs ( Figure 4C). Of note, Mdm2 inhibition with nutlin-3a has been shown to disrupt p73-MDM2 interaction in p53-null cells 24 . To assess the early molecular pathways in p53-null MSCs after Mdm2 inhibition, we isolated GFP+ cells from Osx-Cre;Trp53 / mice (p53 null) and determined the gene expression pro les of vehicle-or nutlin-treated cells after 24 hours ( Figure 4D). RNA-Seq and subsequent Ingenuity Pathway Analysis of p53null MSCs revealed upregulation of genes involved in glycolysis as well as Hif-1α signaling ( Figures 4E-4G). The top seven pathways identi ed by the upregulated differentially expressed genes are illustrated in Figure 4G. Mdm2 inhibition induced Slc2a1 (Glut1), Pdk1, and Fgfr3 genes, known to promote osteoblast differentiation 25,26,27 . Scd2 (stearoyl-CoA desaturase 2) plays a role in the regulation of energy metabolism and lipid synthesis and was signi cantly upregulated by Mdm2 inhibition. The top upstream activated regulator was CD38, a key cellular metabolic driver of aging 28 ( Figure 4H). In addition to CD38, Hif-1, Egln, IL5, and IL15 were enriched as upstream regulators of genes induced in p53-null MSCs treated with Mdm2i. These expression changes suggest that Mdm2 inhibition might promote osteoblast differentiation of p53-null MSCs partly through metabolic pathways.

Heterozygous deletion of Mdm2 in Osx-Trp53 / mice results in osteosclerosis and myelo brosis
To determine the direct effect of Mdm2 inhibition in Osx-Trp53 / mice, we interbred progeny from Osx-Mdm2 +/ and Trp53 mice with mTmG reporter mice to generate Osx-Cre;Mdm2 +/ ;Trp53 / ;mTmG mice (referred to hereafter as Osx-Mdm2 +/ ;Trp53 / ). This enabled us to image the population of MSCs (GFP+) irrespective of their cell surface markers that change upon their differentiation. As expected, homozygous deletion of Trp53 reversed the prenatal lethal phenotype of Osx-Mdm2 / mice, and pups were born at Mendelian ratios without an obvious bone phenotype (data not shown). Unlike Osx-Mdm2 +/ mice, in which we had observed less trabecular bone volume, Osx-Mdm2 +/ ;Trp53 / mice displayed trabecular bone formation mainly due to trabecular ossi cations, resulting in a sponge-like network of trabecular bone (Figures 5A-5C). Bone densitometric analysis con rmed a sclerotic trabecular bone phenotype in the Osx-Mdm2 +/ ;Trp53 / mice, accompanied by signi cant increase in trabecular bone volume ( Figures 5C-5D). Bone histomorphometric further con rmed the ossi cation of trabecular bones in Osx-Cre;Mdm2 +/ ;Trp53 / mice ( Figure 5E). Osx-Mdm2 +/ ;Trp53 / mice became moribund as they aged and died of BM failure. Histopathologic analysis of the BM in Osx-Mdm2 +/ ;Trp53 / mice revealed massive endochondral bone formation that reduced the effective marrow space by approximately 80%. The architecture of the growth plate displayed typical chondrocytes with regular proliferating and hypertrophic zones. However, osseous trabeculae with new bone formation were present throughout the epiphysis, metaphysis, and diaphysis of the long bones as well as vertebrae ( Figure 5F). The diaphyseal cortical bone diameter was increased and coalesced with the subjacent trabecular bone. Histologic examination by reticulin staining showed widespread reticulin positivity in the BM reminiscent of myelo brosis ( Figure 5G). Of note, the trabecular bone volume in mice with deletion of p53 in MSCs, Osxcre;Trp53 / , was comparable with that of p53 wild-type mice, suggesting that the observed phenotype in Osx-Mdm2 +/ ;Trp53 / mice was due to decreased levels of Mdm2 ( Supplementary Figures S2B and  S2C).
Next, we sought to determine whether the sclerotic BM was derived from MSCs. Since the sclerotic bones were densely mineralized, we decided to perform immunohistochemistry for GFP in the BM sections isolated from Osx-Cre;mTmG;Mdm2 +/ ;Trp53 / mice. As shown in Supplementary Figure S2D, the sclerotic BM was GFP positive, suggesting that the sclerotic trabecular bones were derived from MSCs. Collectively, these data demonstrate that genetic depletion of Mdm2 in MSCs lacking p53 promotes osteoblast differentiation leading to lethal osteosclerosis.

Discussion
Several new selective MDM2is have been developed and advanced into early phase clinical trials in different cancers, with promising results 29,30,31 . However, dose-limiting hematopoietic toxicities such as thrombocytopenia often compromise treatment e cacy, and therefore ineffective treatment is common. We investigated the effects of Mdm2 de ciency on MSCs and explored the role of p53 in MSCs in drugrelated cytopenia. First, by using traceable conditional animal models, we demonstrated that heterozygous deletion of Mdm2 resulted in osteopenia. Second, we presented genetic evidence that Mdm2 levels are crucial in the differentiation of MSCs, particularly in the absence of p53. Third, we demonstrated that p53 levels in MSCs are important in Mdm2i-associated cytopenia. Together, these data identify an important role for Mdm2/p53 in the homeostasis of MSCs and their hematopoietic support.
We developed a genetic model enabling the identi cation and imaging of multipotent stromal cells in vivo. The population of MSCs marked by Osx-Cre were CD45 negative, as previously described 18 . Other studies have reported MSCs in reporter mice with use of Gli1-cre mice 32 , transgenic CD73-EGFP BAC mice 33 , and nestin-Cre mice 34 . Gli1-cre and CD73-EGFP reporter mice enabled labeling of presinusoidal endothelial cells, whereas expression of the reporter was minimal in LepR+ cells.
A previous report of mice bearing a different Mdm2oxed allele driven by Col3.6-Cre revealed multiple skeletal defects and reduced bone length; however, osteoblasts deleted for Mdm2 did not undergo apoptosis even though they exhibited elevated p53 activity 9 . Our data indicate that homozygous deletion of Mdm2 completely blocks osteogenesis and that Mdm2 is essential for osteoblast differentiation.
Another intriguing phenotype in Osx-Mdm2 /+ mice is the tness selection of emerging MSCs in early development. Osx-Mdm2 /+ mice displayed apoptosis of MSCs in early developing bones at E18.5; however, the population of MSCs was compensated and was able to reconstitute the developing bones. These data suggest that the Mdm2/p53 pathway maybe involved in the cell competition process during early bone development, as previously reported in other tissues 35 .
Mice with loss of one copy of Mdm2 in their MSCs displayed osteopenia mainly in their trabecular bone compartment, which is known to be derived from MSCs 36 . Deletion of Trp53 reversed the phenotype, and robust osteogenesis occurred with perturbation of Mdm2 and p53. The observed hyperostosis phenotype was not present in Osx-Cre;Trp53 / mice, indicating a role of Mdm2 in the differentiation of MSCs to osteoblasts and broblasts. In addition, our data show that myelo brosis observed in Osx-Mdm2 /+ ;Trp53 / mice originated from MSCs. Recently, Leptin receptor-expressing MSCs were identi ed as the source of myo broblasts in primary myelo brosis 37 .
We present genetic evidence identifying a previously unappreciated role of Mdm2 in the differentiation of MSCs lacking p53. Mdm2 is commonly thought of as an essential gene only when p53 is competent and thus is regarded erroneously to be unnecessary when p53 is absent. Considering that Mdm2 levels are regulated by p53, our data support the hypothesis that downregulation of Mdm2 upon p53 deletion may be important for cellular differentiation. Mechanistically, we demonstrate that the glycolysis and Hif-1α pathways are upregulated in p53-null MSCs upon MDM2i treatment. Mdm2 inhibition has been reported to actively downregulate Hif-1α through p53 activation 38 . We previously showed that Hif-1α plays a role in the nutlin-mediated downregulation of Cxcl12 4 . In addition, stabilization of HIF-1α in Osx-positive cells in postnatal mice was shown to stimulate trabecular bone formation and glycolysis by upregulating pyruvate dehydrogenase kinase 1 (PDK1), a key glycolytic enzyme 39 . We previously reported that PDK1 is signi cantly upregulated in MSCs derived from patients with AML 16 . In addition, Glut1 expression is increased in osteoblasts to switch their metabolic pathway to glycolysis during differentiation 40 . We observed that inhibition of Mdm2 in cells lacking p53 can induce several glycolytic enzyme-coding genes other than Slc2a1 (Glut1), including Hk1, PfkI, and Slc16a3, to convert the metabolism to glycolysis. The p53-independent role of Mdm2 in cellular metabolism and its role in the regulation of cell differentiation remain to be explored in future studies.
The hematopoietic toxicities associated with MDM2is such as nutlin-3 have been attributed to the involvement of MDM2 in hematopoiesis 41 . We observed that the level of Mdm2 in MSCs is important in hematopoietic failure due to MDM2i therapy. Importantly, deletion of p53 in MSCs completely blocked the hematopoietic failure upon MDM2i treatment, suggesting that targeting p53 activity in MSCs can be therapeutically bene cial in the context of MDM2i-associated hematopoietic toxicity. Thus, MDM2i can affect the BM microenvironment to induce various cellular responses. It remains to be explored whether the increased p53 levels in MSCs upon challenge with genotoxic drugs may contribute to hematopoietic failure after therapy in the same way.
In summary, our study sheds light on the importance of the Mdm2/p53 pathway in the differentiation of MSCs and BM hemostasis. The balance between Mdm2 and p53 is crucial in maintaining MSCs as well as to the survival of hematopoietic cells, which are dependent on MSCs. Thus, the role of the Mdm2/p53 pathway in homeostasis of the BM microenvironment could have important therapeutic rami cations.
Cell culture MSCS cultures derived from mice were established by intrafemoral ushing with subsequent crushing of bone. Bone fractions were cultured in MesenCult Expansion Kit medium (Stemcell, 5513) in a T-75 ask, continuously refreshed, selected, and expanded for adherent cells to approximately 80% con uence. To isolate Osx+ MSCs, the GFP+ population was sorted using an Aria II Cell Sorter. GFP+ MSCs were subsequently returned to culture in Minimum Essential Medium α (Corning, catalogue number 15-012-CV) with 10% fetal bovine serum.
RNA sequencing MSCS cultures were plated in a six-well plate. After 48 hours, the cells were treated with 10 μM nutlin-3a for 24 hours. RNA was extracted by using the Direct-Zol Microprep kit (Zymo Research, R2060). RNA sequencing was performed as previously described 43 .
Fluorescence microscopy Isolated tissues were xed in 4% paraformaldehyde in PBS overnight. Next, bones were washed in PBS and decalci ed in 14% EDTA solution for 10 days at 4°C. Bone samples were subsequently washed in PBS and immersed in 30% sucrose PBS solution overnight. Bones and soft tissues were transferred and submerged in optimal cutting temperature (OCT) compound (Tissue-Tek, 4583), after which the embedded tissue was cut into 5-µm sections with use of a cryostat. Slides were either stored as frozen slides at -80°C or further processed. Tissue slides were washed with PBS and then incubated in 3% bovine serum albumin for 1 hour at room temperature. After three subsequent 5-minute washes in PBS, primary antibody incubation against perilipin (Cell Signaling 9349S), leptin receptor (R&D Systems AF497), and p53 (CM5, Leica P53-CM5P-L) was performed at antibody-speci c dilutions in a lightprotected hydration chamber at 4°C. Secondary antibody staining against the primary antibody host species (Chicken anti-Goat IgG AF647, Invitrogen A-21469, and Donkey anti-Rabbit IgG AF647, Invitrogen A-31573) was performed at dilutions of 1:500 for 1 hour at room temperature. Tissue slides were washed thrice in PBS for 5 minutes, stained with 4′,6-diamidino-2-phenylindole (DAPI, Life technologies, D3571) at 1:750 for 5 minutes, and rinsed with PBS. Slides were mounted with VECTASHIELD Mounting Medium (Fisher, H-1000), and the cover slides were sealed with nail polish.

Imaging and spectral deconvolution
Frozen sections and immuno uorescence-stained slides were imaged by using Vectra multispectral imaging system, version 2 (Akoya Biosciences). All samples were scanned at 40× magni cation and visualized with a Phenochart slide viewer (Akoya Biosciences).

Tissue preparation and immunohistochemical analysis
Immunohistochemical analysis was performed at room temperature by using the VECTASTAIN Elite ABC HRP Kit (Vector Laboratories, Pk-6101). Formalin-xed para n-embedded slides were depara nized in two changes of xylene, followed by two changes of 100% ethanol, and then once through 95%, 75%, and 50% ethanol for 5 minutes per change. Endogenous peroxidase activity was subsequently inhibited by immersion in 3% H 2 O 2 solution in methanol for 10 minutes. After two 5-minute washes in Tris-buffered saline (TBS), slides were immersed in Coplin jars with antigen retrieval solution and placed in an IHC-Tek Epitope Retrieval Steamer (IHC World, IW-1102) for 10 minutes. After a cool-down period of 20 minutes at room temperature, the slides were blocked for 60 minutes with 10% serum in 0.1% TBST. Primary antibody incubation for p53 (CM5, Leica P53-CM5P-L) was done overnight in 2% FBS in 0.1% TBST in a light-protected hydration chamber at 4°C. Upon primary antibody incubation, in addition to two 5-minute washes in TBS, slides were incubated in biotinylated secondary antibody solution for 30 minutes at room temperature. Next, after two 5-minute washes in TBS, slides were incubated for 30 minutes in ABC solution. Two 5-minute washes in PBS followed, after which the slides were incubated in 3,3′diaminobenzidine solution until the desired signal intensity was reached. The diaminobenzidine solution was washed off with deionized H 2 O, and slides were next counterstained with hematoxylin for 80 seconds. Incubation of slides in lithium carbonate solution for 60 seconds followed, after which the slides were immersed in deionized H 2 O for 10 minutes. Slides were next immersed for two changes, respectively, of 95% ethanol and 100% ethanol for 90 seconds per change. Next slides were immersed three times for 5 minutes in xylene. Finally, the slides were mounted with Richard-Allan Scienti c Mounting Medium (Thermo Scienti c, 4112).
TUNEL staining TUNEL staining of frozen tissue slides was performed using the Click-iT Plus TUNEL Assay for In Situ Apoptosis Detection kit (Thermo Fisher Scienti c, C10619). Brie y, solutions outlined by the kit were prepared before starting the staining procedure. Frozen tissue sections were brought to room temperature in PBS for 15 min, and the tissue was encircled with a PAP pen. Tissues were xed in 4% formaldehyde in PBS at 37° C, washed twice for 5 min in PBS, and subsequently incubated for 15 min with proteinase K solution at room temperature. Next, slides were washed in PBS for 5 min, xed again in 4% formaldehyde in PBS for 5 min at 37° C, and rinsed in deionized H2O. Incubation of slides with 100 μL of compound A for 10 min at 37°C followed, after which the tissue was incubated with 50 μL of TdT reaction mixture for 60 min at 37° C. After a deionized H2O rinse, slides were washed in 0.2 μm ltered 3% bovine serum albumin 0.1% PBST for 5 min, rinsed once more in PBS, and incubated with 50 μL of reaction cocktail for 30 min at 37° C. Tissue slides were rinsed in PBS and stained with DAPI at 1:750 for 5 min. Slides were mounted as described above.

Skeletal morphometry
Trabecular and cortical bone parameters were characterized with use of micro-computed tomography (μCT) imaging. Brie y, age-sorted mice with genotypes of interest were scanned in vivo. Scans were conducted with a 13-μm pixel size in a Bruker Sky Scan 1276 μCT scanner. Femur data were further analyzed by using the CT Analyzer (CTan) program to assess trabecular and cortical bone microarchitecture.

Statistical analyses
The mean and standard deviation for the indicated number of samples were calculated using GraphPad Prism6 software. The Student's t test was used for comparative analysis between two groups. Analysis of variance was used to compare multiple groups. pvalues ≤ 0.05 were considered statistically signi cant.

Disclosure statement
The authors declare no potential con icts of interest.

ACKNOWLEDGMENTS
We sincerely thank the staff of the Animal Facility, Wendy Schober and Nalini Patel, in the Flow Cytometry Core, Kiersten Maldonado in the Imaging Facility. We also thank Gigi Lozano for providing Mdm2-ox mice. This work was funded by Haas chair in Genetics (to M.A.), CPRIT MIRA (to M.A.), Leukemia SPORE (to M.A.), SPORE Career Enhancement Award (to R.P.), and TRIUMPH fellowship (to E.A.). Editorial support was provided by Bryan Tutt, Scienti c Editor, Research Medical Library. This research was performed in the Flow Cytometry and Cellular Imaging Core Facility, which is supported in part by the National Institutes of Health through MD Anderson's Cancer Center Support Grant P30CA016672. AUTHOR CONTRIBUTIONS R.P. and M.A. designed experiments, analyzed data, and wrote the manuscript; R.H.M. and R.P. performed the experiments and analyzed data. E.A., Z.A. and L.O. performed research, and analyzed data; P.P.L. helped edit the manuscript; B.L. and J.K.B. shared expertise and analyzed data.

Availability of data and materials
The data and materials used and/or analyzed for the current study are available upon request from the corresponding author. showing the trabecular bone density in indicated mice (n = 3, mean ± SD); ***p < 0. 001, Student's t test.