The Dysfunctional MDM2–p53 Axis in Adipocytes Contributes to Aging-Related Metabolic Complications by Induction of Lipodystrophy

Adipose tissue is an active and dynamic endocrine organ that modulates systemic metabolism in response to nutrient availability, lifestyle, and environmental changes. Adipose tissues undergo redistribution, remodeling, and deterioration during aging (1). In general, adipose mass increases with age, but reaches a peak in middle age and declines in extreme old age (2). The age of onset of aging-specific reduction of adipose tissue (lipodystrophy) varies across different ethnic groups (2). Noticeably, good adipose tissues, like subcutaneous white adipose tissue (sWAT) and brown adipose tissue (BAT), are markedly reduced in extreme old age, accompanied with diminished lipid-handling ability, aberrant secretory profile, defective adaptive thermogenesis, and de novo adipogenesis (1,3–5). Senescent cells are gradually increased in aged fat, which disrupt the microenvironment and hence impede adipose tissue regeneration (1). Removal of senescent cells in adipose tissue delays the fat loss and its related metabolic dysfunctions in the aging mouse models (6,7). Several premature aging mouse models suffer from a dramatic loss of sWAT (8–10). These findings suggest that proper adipose tissue function and mass are crucial for metabolic health in aging; however, the underlying mechanism of adipose tissue dysfunction, in particular fat loss, with aging remains poorly understood.

Chronic activation of the tumor suppressor p53 is detrimental to metabolic health and exacerbates premature aging in certain circumstances, especially when its activation cannot be antagonized by its negative regulator, murine double minute 2 (MDM2) (11–13). Expression of the active form of p53, which lacks of MDM2-binding domain, reduces life span (9). Transgenic overexpression of the truncated and MDM2-insensitive p53 isoform (i.e., p44 isoform) accelerates aging (14). These studies suggest that tight regulation of p53 is crucial for the aging process (11,15), but the role of MDM2 in this context remains enigmatic. In this study, we show that MDM2 is selectively downregulated in aged sWAT and BAT. We used an adipocyte-specific MDM2-knockout (Adipo-MDM2-KO) mouse model to mimic this change in aging and demonstrate that adipocyte MDM2 deficiency triggers a cluster of aging-related metabolic disorders via p53. Our results highlight the crucial role of the MDM2–p53 axis in the maintenance of adipose tissue health during aging.

All of the mice were on the C57BL6/J genetic background. The 12-week-old and 2-year-old C57BL/6J male mice were obtained from The Laboratory Animal Unit of The University of Hong Kong (HKU). Homozygous or heterozygous Adipo-MDM2-KO mice were generated by crossing MDM2^{floxed/floxed} mice (16) with Adipoq-Cre mice (expressing Cre recombinase under the control of the adiponectin promoter; The Jackson Laboratory). To obtain adipocyte-specific MDM2-p53 double-KO (DKO) mice, Adipo-MDM2-KO mice were crossed with p53^{floxed/floxed} mice (The Jackson Laboratory) (17). Genotyping for mice carrying the MDM2^floxed allele, p53^floxed allele, and adiponectin Cre was previously described (13,18). All animals were allocated to their experimental group according to their genotypes, sex, and age and fed with standard chow (catalog #5053; LabDiet). The animals were housed in a room with temperature (23°C) and light (12-h light/dark cycle) control and had free access to water and diet.

For the glucose tolerance test (GTT), the mice were fasted for 16 h before intraperitoneal (IP) injection with d-glucose (2 g/kg). For the insulin tolerance test (ITT), the mice were fasted for 6 h followed by IP injection of human insulin (0.75 units/kg; Novo Nordisk). For ZVAD treatment, ZVAD (6 mg/kg; catalog #187389-52-2; Calbiochem) was administrated IP every 2 days for 5 weeks as previously described (19). For fat transplantation, 12-week-old wild-type (WT) donor mice were sacrificed with an overdose of pentobarbital (150 mg/kg body weight), and sWATs were collected for transplantation. sWATs (∼250 mg) from WT donors were implanted subcutaneously through small incisions in the shaved skin of the back of 12-week-old Adipo-MDM2-KO mice and their WT littermates as described in our previous study (20). The mice were allowed to recover for 2 weeks before metabolic characterization.

Total RNA was extracted using TRIzol (catalog #15596026; Invitrogen). The cDNA was prepared by a reverse-transcription kit according to the manufacturer’s manual (catalog #A5001; Promega), and real-time quantitative PCR was performed in the Step One Plus Real-time PCR System (Applied Biosystems) using SYBR Green (catalog #21966100; Roche) with the gene-specific primers (Supplementary Table 1).

Fat depots were digested with collagenase type I (catalog #17100017; Thermo Fisher Scientific) at 37°C for 30 min. The digested tissue was filtered through a 70-μm cell strainer (catalog #352350; Corning) and then centrifuged at 500g for 5 min. The cell pellet contained the stromal vascular fraction (SVF), and the supernatant contained adipocytes. The SVFs were cultured in DMEM (catalog #12800082; Thermo Fisher Scientific) with 10% FBS (catalog #10270; Thermo Fisher Scientific) and 1% penicillin-streptomycin (catalog #15140122; Thermo Fisher Scientific) until 100% confluence. Differentiation of white and brown adipocytes from SVFs was described as we previously reported (21). To collect conditional medium from mature adipocytes, mature adipocytes after 7-day differentiation were washed with PBS three times and cultured in DMEM containing 10% FBS for 24 h. The conditional medium was collected and mixed with the adipocyte differentiation medium in a ratio of 50:50.

Fat depots were digested and filtered as above, followed by staining with LipidTOX (catalog #H34476; Invitrogen), Annexin V, and propidium iodide (catalog #V13242; Invitrogen). The stained cells were subjected to flow cytometry analysis using a BD LSR Fortessa Analyzer.

Tissues or cells were homogenized in a RIPA lysis buffer (catalog #9803; Cell Signaling Technology) with phosphatase inhibitors and protease inhibitor cocktails (catalog #11697498001; Sigma-Aldrich; pH 8.0). Proteins were separated by SDS-PAGE, transferred onto polyvinylidene difluoride membrane, and probed with primary antibodies against MDM2 (catalog #04-1530; EMD Millipore), p53 (catalog #2524; Cell Signaling Technology), p21 (catalog #ab7960; Abcam), β-tubulin (catalog #2128; Cell Signaling Technology), GAPDH (catalog #5174; Cell Signaling Technology), or β-actin (catalog #sc-1616; Santa Cruz Biotechnology), followed by incubation with horseradish peroxidase–conjugated secondary antibodies (Cell Signaling Technology). The protein bands were visualized by enhanced chemiluminescence reagents (GE Healthcare) and quantified using the ImageJ software.

The tissues were fixed in 4% formaldehyde in PBS, subjected to tissue processing, and cut into 5-μm sections, followed by hematoxylin and eosin staining as described in our previous study (22). Adipocyte size and number were determined using ImageJ. For Oil Red O staining in cells, cultured adipocytes were washed with PBS and then fixed in 4% formaldehyde for 1 h at room temperature. After fixation, the cells were washed with PBS and incubated with 0.3% Oil Red O solution (catalog #O0625; Sigma-Aldrich) for 10 min at room temperature. β-Galactosidase staining in freshly collected adipose tissues was performed using a β-galactosidase staining kit (catalog #11674; Cell Signaling Technology).

Adipose tissue sections were subjected to antigen retrieval by boiling in sodium citrate buffer (10 mmol/L, pH 4.5), followed by blocking with PBS containing 10% FBS and 3% BSA for 1 h. The sections were incubated with an antiperilipin antibody (catalog #9349; Cell Signaling Technology) overnight at 4°C. The slides were incubated with an anti-rabbit IgG conjugated with red fluorescent dye at room temperature for 1 h, followed by TUNEL staining (catalog #11684795910; Roche) according to the manufacturer’s instructions. TUNEL-positive and perilipin double-positive adipocytes were counted in 10 random fields per sample.

Metabolic parameters in mice were measured in an environmental-controlled Comprehensive Lab Animal Monitoring System (Columbus Instruments) as we previously described (23). For the cold challenge experiment, the mice were subjected to a 4°C environment for 3 h. Rectal temperature was measured with a thermometer (model 4610 Precision Thermometer; Measurement Specialties).

Serum insulin levels were measured using a mouse high-sensitivity insulin ELISA kit (catalog #32270; Antibody and Immunoassay Services, HKU). Serum levels of leptin and adiponectin were measured using a mouse leptin ELISA kit (catalog #RD291001200R; BioVendor) and mouse adiponectin ELISA kit (catalog #32010; Antibody and Immunoassay Services, HKU), respectively. Plasma glucose was determined using an Accu-Chek glucose meter. Serum triglyceride, cholesterol, β-hydroxybutyrate, and free fatty acid levels were measured using Liquicolor Triglyceride (catalog #2100; Stanbio Laboratory), cholesterol (catalog #1010; Stanbio Laboratory), β-hydroxybutyrate (catalog #2440; Stanbio Laboratory), and Half Micro Test Kit (catalog #11383175001; Roche), respectively. Serum levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were determined using the ALT/SGPT Liqui-UV test (catalog #2930/430; Stanbio Laboratory) and AST/SGOT Liqui-UV test kits (catalog #2930/2920; Stanbio Laboratory). Activity of caspase-3 was determined by a Caspase-3 Fluorometric assay kit (catalog #K105-100; BioVision) according to the manufacturer’s protocol.

The mice were acclimated to treadmill running (20 cm/s) using an animal treadmill (Columbus Instruments) every other day for 1 week. After acclimation, the mice were subjected to an exercise ability assessment. The speed was set as 20 cm/s. Time is running time to exhaustion; distance is distance traveled at time of exhaustion. Exhaustion is defined by refusal to resume running despite electronic shock plus splayed posture (24).

The data are expressed as mean ± SD. All statistical analyses were performed using SPSS or GraphPad Prism 5.0. Equality of variances was assessed by Levene test. Statistical significance was assessed by Student t test, one-way ANOVA with Bonferroni correction for multiple comparisons, or log-rank test. The Welch t test was used for data with unequal variance. A value of P < 0.05 was considered statistically significant.

With advanced age (2 years old), the weight of sWAT and interscapular BAT in C57BL/6 mice was markedly reduced when compared with young controls (12 weeks old) (Fig. 1A). Adipose senescence and augmented adipocyte apoptosis were observed in aged adipose tissues (Supplementary Fig. 1A and B). Adipose tissue functions were deteriorated in aged mice, as exemplified by cold-sensitive phenotype and aberrant expression of pro- and anti-inflammatory adipokines/cytokines (including tumor necrosis factor-α [TNF-α], interleukin-6 [IL-6], and adiponectin) and increased level of p21 (the senescent marker) (Supplementary Fig. 1C–E).

Protein levels of p53 and p21 were increased in sWAT and BAT of 2-year-old mice compared with the young counterparts (Fig. 1B). Quantitative PCR (QPCR) analysis revealed that the mRNA level of MDM2 but no other p53-negative regulators, including MDM4, WW domain containing E3 ubiquitin protein ligase 1 (WWP1), the transcriptional factor E4F1, and p53-induced protein with a RING-H2 domain (PIRH2), was reduced in aged sWAT and BAT (Fig. 1C and D). Consistently, the protein level of MDM2 was reduced in sWAT and BAT in advanced age (Fig. 1B). A similar change in MDM2 and p53 protein level was also observed in epididymal WAT (eWAT) of the old mice (Supplementary Fig. 2). These findings indicate that the aberrant MDM2–p53 axis is linked to adipose tissue dysfunction in aging.

We hypothesized that chronic activation of p53 by MDM2 deletion in adipocytes would recapitulate age-related adipose tissue dysfunction. To test this, we performed metabolic phenotyping in Adipo-MDM2-KO mice and their WT littermates. MDM2 protein was dramatically reduced in fractionated adipocytes from sWAT of Adipo-MDM2-KO mice, but not in the SVF of sWAT (Supplementary Fig. 3). p53 was induced in adipocytes of sWAT from Adipo-MDM2-KO mice (Supplementary Fig. 3).

Three-week-old Adipo-MDM2-KO mice had a similar amount of WAT compared with their WT littermates (Fig. 2A and Supplementary Table 2). A trend of reduction in interscapular BAT was seen in 3-week-old Adipo-MDM2-KO mice (Supplementary Table 2). Adipo-MDM2-KO mice displayed a remarkable and progressive reduction in sWAT, eWAT, and BAT: ∼30–40% reduction at 6 weeks and 100% reduction at 12 weeks and 24 weeks (Fig. 2A and B and Supplementary Table 2). Adipocyte number was decreased, and more stromovascular cells were present in sWAT, eWAT, and BAT of 4-week-old Adipo-MDM2-KO mice (Fig. 2C and D). Circulating levels of leptin and adiponectin started to decline in Adipo-MDM2-KO mice at the age of 6 weeks and were nearly undetectable at 24 weeks (Fig. 2E and F). These data show an early onset of lipodystrophy in Adipo-MDM2-KO mice.

Adipo-MDM2-KO mice displayed normal glucose homeostasis at the age of 3 and 6 weeks, but they suffered from hyperglycemia and hyperinsulinemia at 12 and 24 weeks (Fig. 3A and B). Genetic deletion of adipocyte MDM2 caused insulin resistance and glucose intolerance, as revealed by ITT and GTT at the age of 13 and 12 weeks, respectively (Fig. 3C–F). Augmented glucose-stimulated insulin secretion and increased β-cell mass were observed in Adipo-MDM2-KO mice (Fig. 3G and H). The lipodystrophy in Adipo-MDM2-KO mice was also associated with severe nonalcoholic fatty liver, increased hepatic triglyceride accumulation, and elevated levels of circulating ALT and AST (Supplementary Table 2 and Fig. 3I–L). Next, we examined whether heterozygous deletion of MDM2 in adipocytes has any effect on glucose metabolism. GTT and ITT revealed that heterozygous Adipo-MDM2-KO mice had normal glucose homeostasis (Supplementary Fig. 4A–D).

Adipo-MDM2-KO mice exhibited normal food intake at the age of 3 and 6 weeks, whereas they showed hyperphagia at the age of 12 and 24 weeks (Supplementary Fig. 5A), accompanied by increased body weight and lean mass (Supplementary Table 2). Furthermore, 12-week-old Adipo-MDM2-KO mice displayed normal energy expenditure, respiratory exchange ratio, locomotor activity, and core body temperature under ad libitum feeding; however, they were unable to maintain energy balance upon food deprivation (Supplementary Fig. 5B–E). Fasting-induced lipolysis and ketogenesis were impeded in 12-week-old Adipo-MDM2-KO mice (Supplementary Fig. 5F and G). Adipo-MDM2-KO mice were sensitive to cold challenge at the age of 16 weeks but not 3 weeks (Supplementary Fig. 5H).

To determine whether activation of p53 in adipocytes is responsible for the MDM2-null phenotypes, we generated DKO mice by crossing Adipo-MDM2-KO mice with p53^{floxed/floxed} mice (17). Lack of p53 normalized the elevated p21 level in adipocytes of Adipo-MDM2-KO (Fig. 4A). Lipodystrophy (scored by circulating adiponectin level and fat mass) (Fig. 4B and C) and its associated dysregulation of glucose homeostasis (hyperglycemia, hyperinsulinemia, glucose intolerance, and insulin resistance) (Fig. 4D–I), lipid metabolism (hyperlipidemia and impaired lipolysis) (Supplementary Fig. 6), and energy metabolism (defective adaptation to fasting and cold) (Supplementary Fig. 7) were completely rescued by concomitant deletion of p53. These findings demonstrate that activation of p53 is responsible for the MDM2-null phenotypes.

We next elucidated the underlying cause of lipodystrophy in Adipo-MDM2-KO mice. In vitro, SVFs isolated from sWAT and BAT of 3-week-old Adipo-MDM2-KO mice and WT controls were able to differentiate into mature adipocytes with lipid droplets (Supplementary Fig. 8A). Expression of brown and white adipocyte–specific markers were similar between the two groups (Supplementary Fig. 8B and C), excluding the possibility that the lipodystrophy is due to defective cell-autonomous adipogenesis.

In contrast, the differentiated brown and white adipocytes lacking MDM2 underwent apoptosis spontaneously, as indicated by elevated levels of proapoptotic genes such as PUMA, Bax, and cleaved caspase-3 and reduced expression of the antiapoptotic gene Bcl2 (Fig. 5A and B). The proapoptotic and antiapoptotic genes were also upregulated and downregulated, respectively, in sWAT and BAT of Adipo-MDM2-KO mice when compared with their WT controls (Fig. 5C). Consistently, more apoptotic adipocytes were detected in sWAT and BAT of Adipo-MDM2-KO mice, as revealed by TUNEL assay and flow cytometry analysis using Annexin V and propidium iodide labeling (Fig. 5D and E). Augmented adipocyte apoptosis in Adipo-MDM2-KO mice was rescued by inactivation of p53 (Supplementary Fig. 9).

To verify whether apoptosis is the major cause of lipodystrophy, we treated 3-week-old Adipo-MDM2-KO mice with the caspase inhibitor ZVAD. Adipocyte apoptosis in sWAT but not in BAT of Adipo-MDM2-KO mice was suppressed by ZVAD treatment, accompanied by a partial restoration of WAT mass but not BAT mass (Fig. 6A–C). Treatment with ZVAD had no effect on circulating leptin levels in WT controls but partially reversed the reduction of leptin in Adipo-MDM2-KO mice (Fig. 6D). Of note, treatment of ZVAD was unable to rescue the cold sensitivity in Adipo-MDM2-KO mice (Fig. 6E). We speculated that the ineffectiveness of ZVAD treatment on restoration of BAT mass and function may be due to lower reactivity and poor penetrability of ZVAD in BAT, and/or the dysfunction is not merely due to increased brown adipocyte apoptosis. Although the adipocyte apoptosis was completely reversed by ZVAD treatment in WAT of Adipo-MDM2-KO mice, the adipose tissue mass remained significantly lower than that in WT controls, suggesting that an additional pathway contributes to the lipodystrophy phenotype.

Activation of p53 is known to induce senescence, which has been shown to impede adipogenesis (25,26). Therefore, we hypothesized that MDM2 deletion triggers p53-induced cellular senescence, which in turn disrupts adipose tissue microenvironment, leading to defective adipogenesis. β-Galactosidase staining revealed that more senescent cells were identified in BAT and WAT of Adipo-MDM2-KO mice (Fig. 7A). The genes related to senescence-associated secretory phenotype, including p21, TNF-α, IL-6, and MCP-1, were upregulated in adipose tissues and adipocytes lacking MDM2 (Fig. 7B–E), and such upregulation was completely reversed by inactivation of p53 (Supplementary Fig. 10A–C). In contrast, gene expressions of TNF-α, IL-6, adiponectin, and p21 in sWAT were not altered by heterozygous deletion of MDM2 (Supplementary Fig. 4E).

To investigate whether MDM2-null senescent adipocytes interfere with adipogenesis, we collected conditional medium from mature adipocytes differentiated from the SVF of Adipo-MDM2-KO mice, which are supposed to contain the secretory factors impairing adipogenesis. MDM2-null adipocytes displayed increased expression of senescent markers (Supplementary Fig. 10D). Next, we incubated the SVF isolated from WT mice with the conditional medium during their differentiation. Oil Red O staining and QPCR analysis revealed that adipogenesis was blocked by the conditional medium obtained from MDM2-null adipocytes when compared with those from WT controls (Supplementary Fig. 10E and F). These data indicate that the lipodystrophy in Adipo-MDM2-KO mice, at least in part, is due to the inhibition of de novo adipogenesis by the senescent microenvironment in adipose tissues.

To assess whether aberrant MDM2–p53 axis and its associated lipodystrophy contribute to aging-associated physiological decline, we compared exercise ability in Adipo-MDM2-KO mice, DKO mice, and their WT littermates. Although voluntary physical activity was similar between Adipo-MDM2-KO mice and their WT controls (Supplementary Fig. 7C), involuntary exercise capacity was reduced in 12-week-old Adipo-MDM2-KO mice (Fig. 8A). In addition, the life expectancy of Adipo-MDM2-KO mice but not DKO mice was shorter than their WT controls (Fig. 8B). Apart from adipose tissues, senescence was also detected in skeletal muscle, heart, and liver of Adipo-MDM2-KO mice, and such senescence phenotypes did not occur in DKO mice (Fig. 8C–E). These data collectively suggest that lipodystrophy induced by the aberrant adipose MDM2–p53 axis causes multiorgan senescence and accelerates aging.

To determine whether fat transplantation is able to alleviate aging-associated phenotypes, we transplanted sWAT from WT mice into 12-week-old Adipo-MDM2-KO mice. Sham operations were also performed in MDM2-null mice and their WT controls. Hypoleptinemia and hyperphagia in Adipo-MDM2-KO mice were partially reversed by the fat transplantation (Supplementary Fig. 11A and B). Also, fat transplantation largely reversed hyperglycemia, hyperinsulinemia, glucose intolerance, insulin resistance, and hyperlipidemia in Adipo-MDM2-KO mice (Supplementary Fig. 11C–K). Reduced exercise ability was largely rescued by fat transplantation (Supplementary Fig. 12A). At the molecular level, lipodystrophy-induced premature senescence in liver and skeletal muscles was also partially attenuated (Supplementary Fig. 12B and C). Taken together, these data show that replenishing a healthy fat is able to improve metabolic health in aging.

Deterioration of adipose tissues in aging drives systemic metabolic disorders including insulin resistance, hyperlipidemia, and energy imbalance (1); however, the underlying cause remains vaguely understood. The current study shows that the MDM2–p53 axis restricts adipocyte apoptosis and senescence, two hallmarks of the aging cell (15,27). Dysfunction of this axis induces progressive lipodystrophy, leading to multiple metabolic complications, multiorgan senescence, and shorter life span.

Chronic activation of p53 has been unequivocally shown to cause metabolic diseases and aging (12,28–30). In adipose tissues, p53 and its downstream senescent and apoptotic programs are activated in extreme old age (Fig. 1) and conditions of obesity and diabetes (31,32). Minamino et al. (31) attempted to mimic this p53 activation by overexpressing p53 in adipose tissue using an AP2 promoter (so-called AP2-p53 transgenic mice). AP2-p53 transgenic mice displayed adipose tissue senescence but surprisingly had normal adiposity (31). On the contrary, we show in this study that chronic activation of p53 by deleting MDM2 (which resembles the changes in aged adipose tissue) not only elicits adipocyte senescence but also apoptosis, leading to progressive lipodystrophy. Like the accelerated aging mouse models (10), this lipodystrophy mouse exhibited numerous metabolic deficits, reduced exercise capacity, multiorgan senescence, and shorter life span. In line with our study, global activation of p53 and the dysfunctional MDM2–p53 feedback loop results in accelerated aging, including loss of subcutaneous fat in rodents (9,14). Therefore, the functional MDM2–p53 axis is crucial to prevent aging-induced lipodystrophy.

Senescent cells are gradually accumulated in adipose tissue during aging and interfere with both the local microenvironment and distant tissues by secreting a wide array of biologically active factors including cytokines, growth factors, matrix metalloproteinases, and metabolites (1,33). Numerous secretory factors from senescent adipocytes, such as activin A, TNF-α, and IL-6, have been shown to impair adipogenesis (26,34). Consistent with previous studies (6,26), our data indicate that the senescent microenvironment in adipose tissue or senescent factors from adipocytes do not favor adipocyte progenitor differentiation. Further investigation to identify which secretory factors from MDM2-null adipocytes contribute to the inhibitory effect on adipogenesis is warranted.

Apart from adipose tissues, we observed senescence in skeletal muscle, liver, and heart of Adipo-MDM2-KO mice, which were accompanied by decline of exercise ability and shorter life span. Similar to Adipo-MDM2-KO mice, a lipodystrophy mouse model caused by deletion of DICER has increased premature mortality risk (35). Further studies to investigate whether the global senescent phenotype is also present in other lipodystrophic mouse models and the underlying mechanism by which lipodystrophy and/or adipocyte senescence mediates these aging-related physiological declines are warranted.

Although there are several mouse models with either senescence or apoptosis in adipose tissue, none of them has both defects, and few of them develop severe lipodystrophy. For instance, activation of adipocyte apoptosis by genetic deletion of focal adhesion kinase or receptor-interacting protein kinase 3 has no obvious effect on fat mass under lean or obese condition, but causes adipose tissue inflammation and insulin resistance (19,36). Targeted activation of caspase-8 or coactivation of adipocyte apoptosis and lipolysis by deletion of both insulin receptor and IGF-1 receptor causes lipodystrophy (37,38). On the contrary, adipose senescence caused by abrogation of DNA polymerase η increases adiposity in mice (39). Our study is the first to show that simultaneous activation of apoptosis and senescence leads to severe lipodystrophy. These data also indicate that removal of not only senescent adipocyte progenitors but also senescent and apoptotic adipocytes is essential to rejuvenate adipose tissue function in aging (26,40).

The role of MDM2 in adipocyte function has first been demonstrated in cell experiments, in which MDM2-p53 double-null mouse embryonic fibroblasts are unable to differentiate into mature adipocytes (41,42). In contrast, we showed that adiponectin-Cre–mediated deletion of MDM2 has no effect on cell-autonomous adipogenesis. In addition, DKO mice also showed normal adiposity. These data argue that MDM2 exerts its proadipogenic effect at early differentiation only as adiponectin is expressed during the late stage of adipogenesis. We proposed that the differential role of MDM2 on adipocyte development and functions via p53-dependent and -independent manners: at the initiation of adipogenesis, MDM2 enhances the binding of CREB coactivator, CREB-regulated transcription coactivator (Crtc2), and TORC2 to the promoter of C/EBPδ, a key transcription factor responsible for adipocyte differentiation, in a p53-independent manner (41). In the mature adipocytes, MDM2 prevents p53-induced apoptosis and senescence, thereby maintaining adipose tissue mass. Thus, MDM2 plays multiple roles in the adipocyte life cycle from birth to senescence and apoptosis and hence represents a key regulator of adipocyte dynamics during aging.

In summary, our work suggests that the dysfunctional MDM2–p53 axis in adipocytes contributes to lipodystrophy and aging-associated disorders. Furthermore, adipocyte apoptosis, adipocyte senescence, and/or defective adipogenesis contributes to fat loss and dysfunction in Adipo-MDM2-KO mice (Supplementary Fig. 13). Given that inactivation of MDM2 causes severe damage in adipose tissues, insulin resistance (43), and pancreatic β-cell dysfunctions (13), the use of MDM2 inhibitors (several are in clinical trials) for cancer therapy requires caution. As MDM2 expression is substantially reduced in aged adipose tissue, restoration of MDM2 might represent a new approach to revitalize adipose tissue health. In addition, the proper function of the MDM2–p53 axis not only prevents tumorigenesis but also assures metabolic health during aging.

Acknowledgments. The authors thank Guillermina Lozano (The University of Texas MD Anderson Cancer Center) for providing MDM2^{floxed/floxed} mice.

Funding. This work was supported by the French National Research Agency, Research Grant Council of Hong Kong Joint Scheme (A-HKU705/13), Research Grant Council of Hong Kong Collaborative Fund (C7055-14G), General Research Fund (17100717), and Health Medical Research Fund (04151756).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Z.L. performed most of the experiments and wrote the manuscript. L.J. generated data showing the accelerated physiological decline and multi-organ senescence of Adipo-MDM2-KO mice. J.-K.Y. supervised the study. B.W. and K.K.L.W. contributed the data showing the diabetic phenotypes of the Adipo-MDM2-KO mice (GTT and ITT experiments). P.H. initiated the project and edited the manuscript. A.X. and K.K.Y.C. contributed to experimental design, supervised to the work, and wrote the manuscript. Z.L., L.J., J.-K.Y., B.W., K.K.L.W., P.H., A.X., and K.K.Y.C. contributed to data analysis and discussion. Z.L. and K.K.Y.C. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 78th Scientific Sessions of the American Diabetes Association, Orlando, FL, 22–26 June 2018.