Research ArticleCancer

Targeting MDM2-dependent serine metabolism as a therapeutic strategy for liposarcoma

See allHide authors and affiliations

Science Translational Medicine  10 Jun 2020:
Vol. 12, Issue 547, eaay2163
DOI: 10.1126/scitranslmed.aay2163

Two faces of MDM2

The oncogene MDM2 is best known for encoding a protein that helps degrade p53, a key tumor suppressor. However, it also has other oncogenic effects, and Cissé et al. have uncovered one that affects serine metabolism in liposarcoma, a type of soft tissue cancer. The authors showed that in this tumor type, MDM2 is recruited to chromatin independent of any interaction of p53. Inhibiting MDM2 impairs serine synthesis and increases the cells’ reliance on exogenous serine, which is often in short supply in tumors. The authors then demonstrated that a small-molecule inhibitor of MDM2 impairs liposarcoma growth both in vitro and in patient-derived xenograft models.

Abstract

Well-differentiated and dedifferentiated liposarcomas (LPSs) are characterized by a systematic amplification of the MDM2 oncogene, which encodes a key negative regulator of the p53 pathway. The molecular mechanisms underlying MDM2 overexpression while sparing wild-type p53 in LPS remain poorly understood. Here, we show that the p53-independent metabolic functions of chromatin-bound MDM2 are exacerbated in LPS and mediate an addiction to serine metabolism that sustains nucleotide synthesis and tumor growth. Treatment of LPS cells with Nutlin-3A, a pharmacological inhibitor of the MDM2-p53 interaction, stabilized p53 but unexpectedly enhanced MDM2-mediated control of serine metabolism by increasing its recruitment to chromatin, likely explaining the poor clinical efficacy of this class of MDM2 inhibitors. In contrast, genetic or pharmacological inhibition of chromatin-bound MDM2 by SP141, a distinct MDM2 inhibitor triggering its degradation, or interfering with de novo serine synthesis, impaired LPS growth both in vitro and in clinically relevant patient-derived xenograft models. Our data indicate that targeting MDM2 functions in serine metabolism represents a potential therapeutic strategy for LPS.

INTRODUCTION

Sarcomas, which represent about 1% of all cancers, are malignant tumors of mesenchymal origin that develop from soft tissues or bone. Among the 100 different histological subtypes of sarcomas, liposarcomas (LPSs) are the second most frequent subtype after gastrointestinal stromal tumors, accounting for 15 to 20% of all sarcomas. The prognosis of LPS is very heterogeneous and depends on the tumor location, the histological subtype, and the grade/size of the tumor at diagnosis (1). The risks of recurrence and metastatic dissemination of advanced LPS vary between 20 and 40% in case of localized tumors and depend mainly on the quality of surgical resection, which remains the most effective therapeutic strategy to date. LPSs are poorly responsive to classical chemotherapies, and although targeted therapy with the tyrosine kinase inhibitor pazopanib has demonstrated efficacy in patients with several types of sarcomas, it provides no benefit to patients with LPS (2). Moreover, there is currently no cure available for metastatic or nonresectable LPS, and the median overall survival for these patients is around 15 months after diagnosis (3). Hence, additional therapeutic strategies for LPS are urgently needed.

The most common LPS subtypes, well-differentiated and dedifferentiated LPS (WD-LPS and DD-LPS, respectively), are characterized by the systematic amplification (nearly 100% of cases) of the q13-15 region of chromosome 12 encompassing the MDM2 oncogene, which encodes a major negative regulator of the p53 tumor suppressor. MDM2 amplification is currently used as a diagnostic biomarker to distinguish WD/DD-LPS from other sarcoma subtypes that more commonly harbor p53 mutations (49). However, the molecular mechanisms underlying the systematic amplification of MDM2 in LPS remain unknown.

Oncogenic function of MDM2 has been, so far, mainly attributed to its ubiquitin E3 ligase activity that targets p53 for proteasomal degradation. However, growing evidence indicates that its oncogenic activities extend beyond the regulation of p53 (1012). MDM2 is involved in a complex network of protein-protein interactions that bestow MDM2 with p53-independent functions that can contribute to oncogenesis (13, 14). Thus, MDM2-mediated ubiquitylation of other targets including members of the activating transcription factor (ATF), the forkhead box O (FOXO), the small mothers against decapentaplegic (SMAD), the E2F families of transcription factors, the retinoblastoma protein (pRB), and the histones H2A and H2B can influence cell proliferation, DNA repair, transcription, ribosome biosynthesis, and cell fate (10, 1517). Moreover, chromatin-associated MDM2 influences transcriptional repression of genes involved in cell identity and pluripotency through its direct interaction with the polycomb repressor complex 2 (PRC2) (18). MDM2 can also have a cell survival function, independently of p53, by interacting with proteins involved in DNA repair (13, 18). In addition, we recently characterized an additional function of MDM2, independent of p53 and of its ubiquitin E3 ligase activity, which plays a key role in serine metabolism (19). Thus, chromatin-bound MDM2 controls the transcription of genes implicated in de novo serine synthesis, amino acid transport, and glutathione metabolism. These results suggest that the role of the p53 pathway in serine metabolism is, at least in part, directly regulated by MDM2 (20). Here, we show that MDM2-associated functions in serine metabolism are exacerbated in LPS cells, rendering these cells highly sensitive to pharmacological manipulation of MDM2 expression. Hence, our findings open promising therapeutic opportunities for patients with LPS.

RESULTS

MDM2 is recruited to chromatin independently of p53 in LPS cells

To investigate the importance of MDM2 in LPS, we first analyzed MDM2 localization in human LPS using a tissue microarray (TMA) including 38 samples of DD/WD-LPS and 29 samples of other sarcoma subtypes. Immunohistochemical (IHC) analysis of MDM2 subcellular localization in these samples indicated that MDM2 predominantly localized in the nucleus of DD/WD-LPS cells, whereas it presented a weaker and more cytoplasmic expression pattern in other sarcomas (Fig. 1, A and B, and fig. S1). Immunofluorescence (IF) analysis of MDM2 localization in a panel of six cell lines representing different sarcoma subtypes that were cultured in similar conditions showed that MDM2 was nuclear in 60 to 95% of DD-LPS cells, whereas it exhibited a mainly cytoplasmic localization in other sarcoma subtypes (Fig. 1, C and D, and fig. S2A). The nuclear localization of MDM2 in WD/DD-LPS cells prompted us to evaluate whether nuclear MDM2 is associated with chromatin. Chromatin fractionation assays performed on distinct sarcoma cell lines indicated that about 50% of endogenous MDM2 was detected in the fraction enriched in chromatin-bound proteins in LPS cells, whereas no more than 5% of total MDM2 was recruited to chromatin in cell lines originating from other sarcoma subtypes (Fig. 1, E and F, and fig. S2B). Furthermore, a high quantity of endogenous chromatin-bound MDM2 was detected in primary naive tumor samples of human WD/DD-LPS, ranging from 60 to 95% of total MDM2, whereas MDM2 was mainly present in the fraction containing cytoplasmic and nucleosoluble proteins in other sarcoma subtypes (Fig. 1, G and H). MDM2 was similarly detected in the nuclei of parental IB115 LPS cells and in several independent clones derived from this DD-LPS cell line in which TP53 was genetically inactivated by the CRISPR-Cas9 gene-editing technology, indicating that MDM2 localizes to the nucleus independently of p53 in LPS cells (Fig. 1, I and J, and fig. S2, C to G). Chromatin fractionation assays showed that the quantity of chromatin-bound MDM2 was identical in p53-proficient and p53-deficient IB115 cells (Fig. 1, K and L, and fig. S2H). Together, these data indicate that MDM2 exhibits a specific subcellular localization and is recruited to chromatin independently of p53 in human DD-LPS.

Fig. 1 MDM2 is preferentially localized in chromatin in LPS cells.

(A and B) MDM2 subcellular localization (nuclear or cytoplasmic) was determined by immunohistochemistry (IHC) analysis of a TMA composed of 38 DD/WD-LPS and 29 other sarcoma subtypes. (A) Microphotographs of a representative DD-LPS and myxofibrosarcoma. Scale bars, 100 μm. (B) Percentages of cells exhibiting detectable MDM2 nuclear or cytoplasmic staining in each sample (n = 100 cells analyzed per sample). (C) Subcellular localization of endogenous MDM2 in human cell lines derived from DD-LPS or leiomyosarcoma was determined by IF analysis. MDM2 (green), vimentin (red), and nuclei (blue). Scale bars, 50 μm. (D) Percentages of cells exhibiting cytoplasmic or nuclear localization of MDM2 in a panel of cell lines derived from different sarcomas (mean ± SD of n = 3 independent experiments with 100 to 500 cells analyzed in each experiment). (E) Localization of endogenous MDM2 was determined by immunoblotting in a representative panel of human sarcoma cell lines by subcellular fractionation assays. C indicates the fraction enriched in chromatin-associated proteins. S indicates the pool of nuclear and cytoplasmic soluble proteins. The quality of the fractionation and equal loading were verified by TATA-binding protein (TBP) and tubulin (TUB) amounts. (F) Amounts of chromatin-bound versus soluble (cytoplasmic and nuclear soluble proteins) MDM2, relative to total MDM2, as determined by immunoblotting in a representative panel of human sarcoma cell lines (mean ± SD; n = 4 independent experiments). (G) Subcellular localization of endogenous MDM2 was determined in fresh sarcoma samples by fractionation assays. Immunoblot analysis of MDM2, TBP, and tubulin in C and S fractions in a representative panel of WD/DD-LPS or sarcomas of other subtypes. (H) Quantification of immunoblots on C and S fractions prepared from n = 8 WD/DD-LPS and n = 3 sarcomas of different subtypes. Statistical significance was evaluated using nonparametric Mann-Whitney U tests. ***P ≤ 0.001. (I) IF analysis of endogenous MDM2 localization in parental or p53-KO IB115 LPS cell lines. MDM2 (green), vimentin (red), and nuclei (blue). Scale bars, 50 μm. (J) Quantification of the intracellular localization of MDM2 in parental or p53-KO IB115 cells (mean ± SD; 100 to 500 cells were counted per cell line in three independent experiments). (K) Localization of endogenous MDM2 in parental or p53-KO IB115 DD-LPS cells was determined by subcellular fractionation as in (E). (L) Quantification of immunoblots assessing the quantity of MDM2 protein in the soluble and chromatin fractions in parental or p53-KO IB115 DD-LPS cells (mean ± SD; n = 5 independent experiments).

MDM2 is a master regulator of serine metabolism in LPS

We previously showed that MDM2 regulates a transcriptional program implicated in the metabolism of the non-essential amino acids serine and glycine (19). To further investigate the role of chromatin-bound MDM2 in LPS, we examined the expression of several MDM2 direct target genes involved in this metabolic pathway upon short hairpin RNA (shRNA)–mediated depletion of endogenous MDM2, including 3-phosphoglycerate dehydrogenase (PHGDH), phosphoserine aminotransferase 1 (PSAT1), and phosphoserine phosphatase (PSPH), three genes encoding enzymes involved in de novo serine synthesis, and solute carrier family 1 member 4 (SLC1A4), a gene encoding a neutral amino acid transporter implicated in serine uptake. To distinguish between p53-dependent and p53-independent effects of MDM2 on metabolism in LPS cells, we performed experiments in parallel in parental and p53-KO IB115 DD-LPS cells. The quantity of PHGDH, PSAT1, PSPH, and SLC1A4 mRNAs decreased in LPS cells expressing two independent shRNAs targeting MDM2, whereas that of another metabolic gene, glutathione synthetase (GSS), or another transporter of the SLC family, solute carrier family 30 member 1 (SLC30A1), used as controls, remained unchanged upon MDM2 depletion (Fig. 2A and fig. S3A). Immunoblotting of total protein extracts prepared from these cells showed that MDM2 depletion decreased the amounts of PHGDH, PSAT, PSPH, and SLC1A4 proteins (Fig. 2B and fig. S3, B and C). Next, we investigated the molecular mechanisms by which MDM2 controls serine metabolism in LPS cells. Consistent with our previous findings supporting the role of ATF4 in MDM2 recruitment to chromatin (19), shRNA-mediated depletion of ATF4 did not affect the total amount of MDM2 protein but abolished its recruitment to chromatin and favored its localization in the soluble fraction of the nucleus and in the cytoplasm (Fig. 2C and fig. S3D). Comparable results were obtained in parental and p53-KO IB115 cells, confirming that the MDM2-ATF4 complex is recruited to chromatin independently of p53 (fig. S3E). Furthermore, quantitative chromatin immunoprecipitation (qChIP) assays indicated that depletion of ATF4 decreased the recruitment of endogenous MDM2 to the promoter region of PSAT1 and another previously characterized MDM2 target gene, Dickkopf WNT signaling pathway inhibitor 1 (DKK1) (19). Consistent with the close interplay between ATF4 and MDM2, MDM2 depletion also impaired ATF4 recruitment to the promoter region of these two MDM2 target genes (Fig. 2, D and E). Because previous reports showed that some cancer cells are highly dependent on serine and glycine for proliferation, we hypothesized that interfering with MDM2/ATF4-mediated regulation of serine/glycine metabolism in LPS cells would influence their growth (2124). Regardless of their p53 status, the expansion of IB115 DD-LPS cells in vitro decreased upon MDM2 or ATF4 depletion, an effect that was further enhanced when these cells were cultured in a serine and glycine-deprived (-Ser/Gly) medium (Fig. 2, F and G, and fig. S3, F to H). The importance of serine/glycine metabolism for the growth of LPS cells was confirmed in cultured p53-KO IB115 cells in which PHGDH, PSAT1, or PSPH was silenced by shRNA or by pharmacological inhibition of PHGDH by CBR-5884 (Fig. 2H and fig. S4, A to E). Next, we evaluated the importance of MDM2-mediated control of serine/glycine metabolism for tumor growth in vivo. To answer this question, IB115 DD-LPS cells stably expressing p53-shRNA together with a control or an MDM2 shRNA were subcutaneously injected into nude mice on normal chow or fed a serine- and glycine-deprived diet (-Ser/Gly) of equivalent caloric value and equal content in total amino acids. The decreased availability of exogenous serine and glycine combined with the depletion of endogenous MDM2 in LPS cells additively decreased tumor growth (Fig. 2I and fig. S4, F and G). Moreover, interfering with endogenous PHGDH also affected the tumorigenic potential of LPS, an effect that was further increased in mice fed with a -Ser/Gly diet (Fig. 2J and fig. S4H). Hence, these data support the notion that MDM2-mediated control of serine/glycine metabolism is a driving event that sustains tumor growth of LPS.

Fig. 2 Serine metabolism is regulated by chromatin-bound MDM2 in LPS cells.

(A) Relative quantity of MDM2 mRNA and that of its direct target genes PHGDH, PSAT1, PSPH, and SLC1A4 and of two control metabolic genes SLC30A1 and GSS were determined by RT-qPCR in p53-KO IB115 cells 64 hours after shRNA-mediated depletion of MDM2 (shMDM2). Data were normalized to the corresponding control samples prepared from p53-KO IB115 cells expressing a control shRNA (mean ± SD; n = 7 independent experiments). Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (B) Amounts of MDM2, PHGDH, PSAT, PSPH, and SLC1A4 proteins were determined by quantitative immunoblotting of whole-cell extracts prepared from p53-KO IB115 cells 64 hours after transduction with lentiviruses encoding control- or MDM2-shRNAs. (C) Localization of endogenous MDM2 was determined by immunoblotting of chromatin (C) or soluble (S) fractions prepared from p53-KO IB115 cells expressing shRNAs targeting MDM2 or ATF4. The quality of the fractionation and equal loading were verified by TBP and TUB immunoblotting. (D and E) ChIP-qPCR experiments showing the relative recruitment of MDM2 (D) and ATF4 (E) on the PSAT1 and DKK1 promoters in p53-KO IB115 cells expressing shRNAs targeting MDM2 or ATF4. Results are represented as the relative ratio between the mean value of immunoprecipitated chromatin (calculated as a percentage of the input) with the indicated antibodies and the one obtained with a control irrelevant antibody (mean ± SEM; n = 4 independent experiments). Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (F to H) Cell proliferation was monitored in real time using the xCELLigence Real-Time Cell Analysis (RTCA) assay (mean ± SD; n = 5 independent experiments performed in quadruplicates). Bars represent the numbers of cells in p53-KO IB115 cells cultured in complete or serine- and glycine-deprived (-Ser/Gly) medium 4 days after transduction with lentiviruses expressing a control-shRNA or (F) ATF4-shRNA, (G) MDM2-shRNA, or (H) PHGDH-shRNA. Statistical significance was evaluated using Student’s t tests. (I) Nude mice were subcutaneously xenografted with IB115 shp53 cells stably expressing control- or MDM2-shRNAs, and fed with a complete or a serine- and glycine-deprived diet (-Ser/Gly). Box and whisker plots represent the tumor volume (mean ± SD, n = 9 tumors per group) in each experimental group measured when the first animal reached the ethical endpoint. Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (J) Nude mice were subcutaneously xenografted with IB115 shp53 cells stably expressing control- or PHGDH-shRNAs, and fed with a complete or -Ser/Gly diet. Box and whisker plots represent the tumor volume (mean ± SD, n = 10 tumors per group) in each experimental group measured when the first animal reached the ethical endpoint. Statistical significance was evaluated using nonparametric Mann-Whitney U tests. *P ≤ 0.05, **P ≤ 0.01, and *** P ≤ 0.001. NS, not significant.

Nutlin-3A treatment enhances MDM2-associated metabolic functions in LPS cells

The effect of the ubiquitin E3 ligase activity of MDM2 on p53 degradation had been the rationale for the clinical use of compounds such as Nutlins, which interfere with MDM2-p53 protein interaction and reactivate p53 tumor suppressor functions in tumors harboring wild-type (WT)–p53 (2528). However, most clinical trials performed with this class of pharmacological inhibitors have unexpectedly demonstrated poor, if any, therapeutic benefit for LPS patients (25). This prompted us to investigate in more detail the effects of Nutlin-3A on LPS cells. As in all other cell types harboring WT-p53, p53 was rapidly stabilized in response to Nutlin-3A treatment in a dose-dependent manner in a panel of LPS cells (Fig. 3A and fig. S5A). Nutlin-3A also increased the quantities of MDM2 mRNA and total protein, likely as a result of the induction of the well-described p53-MDM2 positive feedback loop (Fig. 3, A and B, and fig. S5A) (29). The quantity of chromatin-bound MDM2 increased in two distinct Nutlin-treated LPS cell lines (Fig. 3C), an effect that correlated with increased expression of MDM2 target genes involved in serine/glycine metabolism (Fig. 3, D and E, and fig. S5B). Our data indicate that Nutlin-3A stabilizes the amount of chromatin-bound MDM2 in LPS cells and potentiates its p53-independent functions in serine/glycine metabolism.

Fig. 3 Nutlin-3A treatment enhances the metabolic functions of MDM2 in LPS cells.

(A) Amounts of MDM2 and p53 proteins were determined by immunoblotting of whole-cell extracts prepared from IB115 and IB111 LPS cells upon Nutlin-3A treatment for 24 hours (2.5 to 10 μM). Equal loading was verified by actin immunoblotting. (B) The relative amounts of MDM2 mRNA were determined in mock or Nutlin-3A–treated IB115 or IB111 DD-LPS cells by RT-qPCR 24 hours after Nutlin-3A treatment (10 μM) (mean ± SD; n = 7 independent experiments). Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (C) Localization of endogenous MDM2 was assessed in IB115 and IB111 cells treated with Nutlin-3A (10 μM) for 24 hours by subcellular fractionation. Fractions enriched in chromatin-associated proteins (C) and pooled fractions containing nuclear and cytoplasmic-soluble proteins (S) were analyzed by immunoblotting. The quality of the fractionation and equal loading were verified by TBP and TUB immunoblotting. (D) The relative amounts of PHGDH, PSAT1, PSPH, and SLC1A4 mRNAs were determined in IB115 and IB111 cells by RT-qPCR 24 hours after Nutlin-3A treatment (10 μM). Data were normalized to the corresponding control samples prepared from IB115 cells treated with dimethyl sulfoxide (DMSO) (mean ± SD; n = 7 independent experiments). Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (E) PHGDH, PSAT, PSPH, and SLC1A4 proteins were quantified by immunoblotting of whole-cell extracts prepared from IB115 and IB111 cells 24 hours after Nutlin-3A treatment. **P ≤ 0.01 and ***P ≤ 0.001.

The pharmacological inhibitor SP141 inhibits MDM2-associated functions in serine metabolism and affects nucleotide biosynthesis in LPS cells

To circumvent the effects of Nutlin-3A on MDM2-associated metabolic functions, we evaluated the efficacy of SP141, the lead compound of a class of pyrido-indole molecules that triggers MDM2 self-ubiquitylation and its proteasomal degradation (30). As expected, treatment of both parental and p53-KO IB115 DD-LPS cells with 1 μM SP141 decreased the amount of MDM2 protein, and consistent with its p53-independent role in serine metabolism in LPS cells, treatment of both p53-proficient and p53-deficient IB115 cells with SP141 decreased expression of the MDM2 target genes PHGDH, PSAT1, PSPH, and SLC1A4 (Fig. 4, A and B, and fig. S6, A and B). To confirm the specificity of this MDM2 inhibitor in LPS cells, we transduced p53-KO IB115 cells with lentiviruses encoding a previously characterized MDM2 mutant (ΔAD-C464A) that predominantly localizes to chromatin due to the deletion of its central acidic domain and harbors a mutation in its E3 ligase domain (C464A) that confers resistance to SP141 by abolishing its auto-ubiquitylation (31). The expected subcellular localization of this ΔAD-C464A MDM2 mutant in p53-KO IB115 was confirmed by chromatin fractionation assays (Fig. 4C). Expression of MDM2 target genes implicated in serine metabolism in SP141-treated p53-KO IB115 LPS cells expressing the ΔAD-C464A MDM2 mutant was indistinguishable from that in control mock–treated cells transduced with an empty vector, demonstrating that SP141 specifically affects serine metabolism by interfering with the transcriptional activities of chromatin-bound MDM2 (Fig. 4B).

Fig. 4 MDM2-associated metabolic functions in nucleotide biosynthesis are inhibited by the pharmacological inhibitor SP141 in DD-LPS cells.

(A) The quantity of MDM2, PHGDH, PSAT, PSPH, SLC1A4, and actin (loading control) proteins was determined by quantitative immunoblotting of whole-cell extracts prepared from p53-KO IB115 cells 20 hours after treatment with the MDM2 inhibitor SP141 (1 μM). (B) The relative amounts of PHGDH, PSAT1, PSPH, and SLC1A4 mRNAs were determined by RT-qPCR in p53-KO IB115 cells transduced with lentiviruses encoding empty vector or the MDM2 ΔAD-C464A mutant 20 hours after treatment with 1 μM SP141. Data were normalized to the corresponding control samples prepared from p53-KO IB115 cells treated with DMSO (mean ± SD; n = 5 independent experiments). Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (C) The subcellular localization of endogenous MDM2 and of the ectopic MDM2 ΔAD-C464A mutant was assessed by subcellular fractionation in p53-KO IB115 LPS cells upon transduction with empty control or MDM2 ΔAD-C464A lentiviruses. Fractions enriched in chromatin-associated proteins (C) and pooled fractions containing nuclear- and cytoplasmic-soluble proteins (S) were analyzed by immunoblotting. The quality of the fractionation and equal loading were verified by TBP and TUB immunoblotting. (D) Stable isotope tracing experiments in p53-KO IB115 cells treated for 24 hours with DMSO or SP141 (0.25 μM) cultured for the last 12 hours in the presence of uniformly labeled [U-13C]Glu in -Ser/Gly medium. LC-MS was used to detect the relative amount of the 13C-labeled m+3 isotopolog of intracellular serine. (E) Schematic representation of 13C incorporation from [U-13C]Glu into purines through the pentose phosphate and de novo SSPs. (F) Stable isotope tracing experiments in p53-KO IB115 cells transduced with a control empty vector or lentiviruses encoding the MDM2 ΔAD-C464A mutant and treated for 24 hours with DMSO or SP141 (0.25 μM). Cells were cultured for the last 12 hours in the presence of uniformly labeled [U-13C]Glu in -Ser/Gly medium. LC-MS was used to detect the relative amounts of 13C-labeled (m+6 to m+9) isotopologs detected in intracellular purines that derive from 13C-labeled serine. (G) Stable isotope tracing experiments in p53-KO IB115 cells transduced with control empty vector or lentiviruses encoding the MDM2 ΔAD-C464A mutant treated for 24 hours with DMSO or SP141 (0.25 μM). Cells were cultured for 24 hours in the presence of uniformly labeled [U-13C]Ser. LC-MS was used to detect the 13C-enrichment in the indicated purines. (H) Relative amounts of 13C-labeled (m+1 to m+4) isotopologs of intracellular purines that derive from 13C-labeled serine. Bars represent the percentages of the indicated 13C-labeled isotopologs in AMP and GMP (±SD). **P ≤ 0.01. NS, no significant difference.

We next performed stable isotope tracing experiments to determine the metabolic fate of exogenous glucose or serine upon pharmacological or genetic inhibition of MDM2 in LPS cells. First, to confirm the role of MDM2 in de novo serine synthesis, we incubated parental or p53-KO IB115 DD-LPS cells expressing control or MDM2 shRNAs or treated with SP141 (0.25 μM) with uniformly 13C-labeled glucose ([U-13C]Glu) for different amounts of time (between 0 and 24 hours) before harvesting these cells for analysis of 13C-labeled intracellular metabolites by liquid chromatography–mass spectrometry (LC-MS). Of note, these cells were cultured in serine- and glycine-deprived medium to enhance the flux into the serine synthesis pathway (SSP). Quantitative analysis of carbon isotopolog distribution (CID) indicated that the amount of intracellular 13C-labeled (m+3) serine derived from [U-13C]Glu decreased upon genetic or pharmacological inactivation of MDM2, confirming its role in the SSP (Fig. 4D and fig. S6C). Next, we investigated how MDM2-depleted LPS cells use glucose-derived serine. CID indicated a reduction of the relative amounts of the different 13C-labeled m+6 to m+9 isotopologs of the purine nucleotides adenosine and guanosine triphosphates (ATP and GTP, respectively) that are generated upon incorporation of glucose-derived carbons from serine and glycine into the purine ring, as well as of their mono/diphosphate precursors (AMP/ADP and GMP/GDP). The impact of MDM2 inhibition on the channeling of [U-13C]Glu into purine biosynthesis, obtained by SP141 treatment or upon shRNA-mediated depletion of endogenous MDM2, was comparable in parental and p53-KO IB115 DD-LPS cells (Fig. 4, E and F, and fig. S6D). Similarly to MDM2 inhibition and consistent with previous data (32), shRNA-mediated depletion of PHGDH also decreased the incorporation of 13C-labeled glucose-derived carbons into purines (fig. S6, E and F). Expression of ectopic ΔAD-C464A MDM2 mutant in SP141-treated p53-KO IB115 LPS cells rescued these metabolic defects (Fig. 4F). Moreover, stable isotope tracing experiments performed with 13C-uniformaly labeled serine ([U-13C]Ser) in p53-KO IB115 LPS cells cultured in complete medium indicated that pharmacological or genetic inhibition of MDM2 decreased its channeling into purine synthesis through the folate pathway (m+2 to m+4 isotopologs of AMP and GMP) (Fig. 4, G and H, and fig. S7, A and B). Similar results were obtained in p53-proficient and p53-deficient IB115 cells, and 13C-enrichment in AMP and GMP was partly rescued upon expression of the ΔAD-C464A MDM2 mutant in SP141-treated cells, confirming that this MDM2 inhibitor efficiently targeted the p53-independent metabolic functions of chromatin-bound MDM2 in LPS cells (Fig. 4, G and H). Serine uptake was not affected in these conditions, but the 13C-enrichment in the m+1 isotopolog of glycine and in the m+1 and m+2 isotopologs of serine in MDM2-deficient p53-KO IB115 cells incubated with [U-13C]Ser supported the notion that the channeling of serine into purine synthesis was impaired upon MDM2 inhibition and favored the reverse conversion of 13C-labeled (m+2) glycine, and its derived 1-C unit, back to serine (Fig. 5, A to C, and fig. S7, C to E).

Fig. 5 Stable isotope tracing experiments showing that SP141 inhibits MDM2-associated metabolic functions in DD-LPS cells.

(A) p53-KO IB115 cells were treated for 24 hours with DMSO or SP141 (0.25 μM) and cultured for the last 12 hours in the presence of uniformly labeled [U-13C]Ser. LC-MS was used to detect the relative amount of the 13C-labeled m+3 isotopolog of intracellular serine. (B) Relative amount of the m+2 isotopolog of intracellular glycine upon incubation with [U-13C]Ser in the same cells as in (A). (C) Relative amounts of the m+1 and m+2 isotopologs of intracellular serine upon incubation with [U-13C]Ser in the same cells as in (A). (D and E) 13C-enrichment in aspartate (D) and pyrimidines (E) upon incubation of p53-KO IB115 cells treated for 24 hours with DMSO or SP141 (0.25 μM) and cultured with [U-13C]Glu for 24 hours. Bars represent the percentages of 13C-enrichment in the indicated metabolites (± SD).

The SSP has recently been shown to coordinate biosynthesis of purines and pyrimidines by maintaining proper mass balance within central carbon metabolism (33). To evaluate whether such mechanism also contributed to the metabolic defects observed in MDM2-deficient LPS cells, we evaluated the incorporation of glucose-derived carbons in aspartate, an amino acid that contributes to the pyrimidine ring. Consistent with this hypothesis, 13C-enrichment in aspartate in p53-KO IB115 DD-LPS cells incubated with [U-13C]Glu decreased upon SP141 treatment, a metabolic defect that was rescued in cells stably expressing the ΔAD-C464A MDM2 mutant (Fig. 5D). Moreover, the global 13C-enrichment in pyrimidines was impaired in these SP141-treated p53-KO IB115 DD-LPS cells (Fig. 5E). Similar defects were observed upon expression of an shRNA targeting MDM2, confirming its role in pyrimidine synthesis (fig. S7, F and G). Together, these data indicate that pharmacological and genetic inhibition of MDM2 impairs de novo serine synthesis and affects nucleotide metabolism in LPS cells through several mechanisms.

Inhibition of MDM2-mediated control of nucleotide synthesis by SP141 impairs both cell proliferation and cell survival

To further delineate the metabolic functions of MDM2 in LPS cells, we next examined cell viability and the cell cycle profile of IB115 DD-LPS cells upon treatment with SP141. We observed a rapid loss of cell viability of IB115 LPS cells as soon as 24 hours after treatment with 2 μM SP141. SP141-induced cell death was observed both in parental and in p53-KO ID115 DD-LPS cells, indicating that this effect was mainly p53 independent (Fig. 6A and fig. S8A). The effect of SP141 on the growth of LPS cells was confirmed on a panel of 10 different DD-LPS cell lines. The IC50 (median inhibitory concentration) of SP141 was lower in LPS cells than in cell lines isolated from other sarcoma subtypes, ranging from 200 nM to 3.2 μM and from 5 to 40 μM, respectively (Fig. 6B). The enhanced sensitivity of DD-LPS cells to SP141 compared to other sarcoma subtypes was also confirmed when these cells were cultured in anchorage-independent conditions. Thus, massive cell death was evidenced in spheroids generated from both p53-proficient or p53-deficient IB115 DD-LPS cells upon treatment with 2 μM SP141, but not in spheroids generated from cells (IB140 and IB143) originating from two different leiomyosarcomas (Fig. 6C and fig. S8B). shRNAs directed against MDM2 or PHGDH also decreased cell viability of IB115 DD-LPS cells cultured in three-dimensional (3D) conditions, irrespectively of their p53 status, suggesting that the impact of MDM2 inhibition on cell survival of LPS cells resulted from impaired serine metabolism that occurred independently of p53 (Fig. 6D and fig. S8C). Stable expression of MDM2 ΔAD-C464A rescued the cell viability and the expansion of p53-KO IB115 DD-LPS cells under SP141 treatment, as well as the expression of MDM2 target genes implicated in serine metabolism (Fig. 6E and fig. S8, D and E).

Fig. 6 MDM2-mediated control of nucleotide synthesis is inhibited by SP141, which impairs both cell proliferation and cell survival.

(A) Percentages of viable cells in parental or p53-KO IB115 DD-LPS cells cultured for 82 hours in the presence of SP141 (1 μM). Measurements were made using the xCELLigence RTCA assay (mean ± SD; n = 7 independent experiments performed in quadruplicates). Statistical significance was evaluated using Student’s t tests. (B) IC50 for SP141 in 10 DD-LPS and 6 sarcoma cell lines of different subtypes. Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (C) The impact of SP141 on cell viability was determined in 3D culture conditions. Twenty-four hours after seeding LPS or sarcoma cells of other subtypes, the resulting spheroids were treated with SP141 (2 μM) or DMSO and photographed after 6 days in culture (scale bars, 500 μm). Live (blue) and dead (green) cells were detected using a fluorescent viability kit. (D) p53-KO IB115 cells transduced with lentiviruses encoding control, PHGDH, or MDM2 shRNAs were cultured for 6 days as spheroids, and live and dead cells were detected as in (C) by IF microscopy at the indicated time points (scale bars, 500 μm). (E) Cell viability of p53-KO IB115 LPS cells transduced with lentiviruses encoding the MDM2 ΔAD-C464A mutant or with the corresponding control empty vector upon SP141 (2 μM) treatment. Bars represent the relative numbers of live cells after 48 hours of SP141 treatment in 2D cultures (mean ± SD, n = 7 independent experiments performed in triplicates). Statistical significance was evaluated using Student’s t tests. (F) Viability of p53-KO IB115 LPS cells upon SP141 (2 μM) treatment in the presence or absence of exogenous nucleosides (EmbryoMax Nucleosides, 1:100 dilution). Bars represent the numbers of live cells after 48 hours of SP141 treatment (1 μM) in 2D cultures (mean ± SD, n = 7 independent experiments performed in triplicates). Statistical significance was evaluated using Student’s t tests. (G) MDM2, PSPH, phospho-γH2AX (P-γH2AX), and actin (loading control) proteins were quantified by immunoblotting of whole-cell extracts prepared from p53-KO IB115 cells 24 hours after treatment with SP141 (0.25 and 0.5 μM) or DMSO. (H) IF analysis of P-γH2AX (red) and DAPI (4′,6-diamidino-2-phenylindole) (blue) in p53-KO IB115 cells 24 hours after treatment with SP141 (0.5 μM) or DMSO. Scale bars, 50 μm. (I) FACS analysis of the cell cycle distribution of p53-KO IB115 cells 16 hours after treatment with 0.5 μM SP141 or DMSO (mean ± SEM; n = 4 independent experiments). (J) Relative numbers of viable p53-KO IB115 cells 24 hours after treatment with DMSO or SP141 in the presence of different cell death inhibitors [DPI (1 μM), ZVAD (30 μM), PJ34 (50 μM), IM54 (10 μM), and ferrostatin-1 (1 μM)] (mean ± SD; n = 8 independent experiments performed in triplicates). Statistical significance was evaluated using Student’s t tests. (K) Endogenous PARP1, AIF, TUB, and TFAM proteins were quantified by immunoblotting, using nuclear (Nuc) extracts or purified mitochondria (Mito) prepared from p53-KO IB115 cells, 24 hours after treatment with DMSO or SP141 (1 μM). Equal loading was verified by TBP and TFAM immunoblotting. *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001. NS, no significant difference.

Next, we evaluated whether the impairment of serine metabolism and nucleotide synthesis observed upon MDM2 inhibition in LPS cells contributed to their growth defects. Accordingly, addition of a mix of nucleosides in the culture medium of SP141-treated p53-KO IB115 DD-LPS cells rescued their viability (Fig. 6F). Because depletion of nucleotide pools is known to induce replication stress and DNA damage, we then evaluated whether MDM2-depleted LPS cells displayed evidence of DNA damage. Increased phosphorylation of histone H2AX on Ser139 (γH2AX), a hallmark of DNA damage (32), was detected by immunoblotting and IF as soon as 12 hours after treatment of LPS cells with lower doses (0.25 and 0.5 μM) of SP141 (Fig. 6, G and H, and fig. S8F). Consistent with these findings, SP141-treated p53-KO IB115 DD-LPS cells exhibited an altered cell cycle profile, with cells accumulating in the S and G2-M phases of the cell cycle (Fig. 6I and fig. S8G).

These data suggested that perturbations of chromatin-bound MDM2 upon SP141 treatment resulted in nucleotide biosynthesis and replication defects that ultimately turned on a cell death program. In an attempt to identify the cell death pathway activated after MDM2 inhibition, we used several drugs that interfere with different cell death programs, including the apoptosis inhibitor ZVAD, the necrosis inhibitor IM54, the ferroptosis inhibitor ferrostatin-1, and the poly(ADP-ribose) polymerase 1 (PARP1) inhibitor PJ34 (34). Ferrostatin-1 and IM54 treatment failed to rescue LPS cells from SP141-induced cell death. Addition of ZVAD or the PARP inhibitor PJ34 partly rescued the viability of SP141-treated IB115 DD-LPS cells. Similar effects were obtained in both parental and p53-KO IB115 DD-LPS cells, indicating that this cell death program occurred independently of p53 (Fig. 6J and fig. S8H). Depletion of nucleotide pools has been shown to induce parthanatos, an atypical cell death program mediated by PARP and AIF1 (35). Upon SP141-mediated inhibition of MDM2 in LPS cells, we observed an increased amount of PARP protein and the release of AIF from mitochondria (Fig. 6K and fig. S8I). Moreover, addition of DPI, a pharmacological inhibitor of AIF1, also reduced cell death to a comparable extent to PJ34, confirming the induction of parthanatos (Fig. 6J) (36). Together, our results show that MDM2 inhibition by SP141 affects nucleotide synthesis in LPS cells, resulting in DNA damage and subsequently cell death, in part, through parthanatos.

Targeting MDM2-associated metabolic functions is an effective therapeutic strategy for LPS

To evaluate the clinical relevance of our findings and the applicability of therapeutic strategies aiming at targeting MDM2-associated metabolic activities, we compared the in vivo efficacy of SP141 and doxorubicin, a chemotherapy used as a first-line treatment for primary LPS, in a xenograft model in which IB115 DD-LPS cells were subcutaneously injected into nude mice. When tumors reached 150 mm3, xenografted mice were treated with SP141 (40 mg/kg daily) or doxorubicin (2 mg/kg, two times per week), and tumor growth was followed over a period of 3 weeks. SP141 treatment decreased tumor growth, whereas doxorubicin had no statistically significant effect (P = 0.31) in this xenograft model (Fig. 7A and fig. S9A). In contrast to doxorubicin, a 3-week treatment with SP141 did not result in obvious side effects in this xenograft model, as illustrated by the lack of weight loss or major histological defects in peripheral tissues (fig. S9, B and C). Consistent with the low toxicity of this MDM2 inhibitor, the IC50 of SP141 was 50- to 100-fold higher in various untransformed cells from murine or human origins compared to that determined in our panel of LPS cells (fig. S9D). The same SP141 regimen resulted in similar antitumor effects in xenograft models generated with three distinct DD-LPS cell lines, including parental and p53-deficient IB115 cells (Fig. 7, B and C, and fig. S10A). Consistent with the role of MDM2 in serine metabolism, immunoblotting analysis of total protein extracts prepared from control or SP141-treated tumors showed that MDM2 inhibition in vivo resulted in the decreased expression of PHGDH and PSAT (Fig. 7D).

Fig. 7 Pharmacological inhibition of MDM2-associated metabolic function inhibits LPS growth.

(A) IB115 cells were subcutaneously injected into nude mice. Tumors were treated with SP141 (40 mg/kg, daily injections) or doxorubicin (2 mg/kg, twice a week) when they reached 150 mm3. Box and whisker plots represent tumor volumes in each experimental group measured when the first animal reached the ethical end-point (mean ± SD, n = 9 to 34 tumors per group). Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (B) Different LPS cell lines (LPS A, LPS B, IB115, IB115 shp53, and p53-KO-IB115 cells) were subcutaneously injected into nude mice. Tumors were treated with SP141 when they reached 150 mm3 (40 mg/kg, daily injections). Box and whisker plots represent tumor volumes in each experimental group (mean ± SD, n = 6 to 12 tumors per group). Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (C) Photographs of representative mock- or SP141-treated mice at the end of the experiment. (D) MDM2, PHGDH, PSAT, and actin (loading control) proteins were quantified by immunoblotting using extracts prepared from mock- or SP141-treated IB115 DD-LPS–derived tumors. (E) Nude mice were subcutaneously xenografted with IB115 cells stably expressing the MDM2 ΔAD-C464A mutant or the corresponding empty lentiviral vector. When they reached 150 mm3, tumors were treated with SP141 (40 mg/kg, daily injections). Box and whisker plots represent tumor volumes in each experimental group (mean ± SD, n = 9 tumors per group). Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (F) Photographs of representative mock- or SP141-treated tumors generated from control or IB115 cells stably expressing MDM2 ΔAD-C464A mutant. (G) Relative amounts of PHGDH, PSAT1, PSPH, and SLC1A4 mRNAs were determined in samples prepared from the same tumors as in (F) by RT-qPCR. Data were normalized to the corresponding control mock–treated tumor samples (mean ± SD; n = 10 tumors per group). Statistical significance was evaluated using nonparametric Mann-Whitney U tests. (H and I) SP141 treatment of PDX sarcoma models. PDXs were generated in nude mice by subcutaneous injection of freshly resected tumor samples from three different patients with LPS (H) or from one patient each with leiomyosarcoma or with undifferentiated sarcoma (I). Experiments were performed on established PDXs at serial passage 5 or 7. Animals were treated with SP141 (40 mg/kg, daily injections) when tumors reached 150 mm3. Box and whisker plots represent tumor volumes (mean ± SD, n = 5 tumors per group) in each experimental group measured when the first animal reached the ethical end-point. Statistical significance was evaluated using nonparametric Mann-Whitney U tests. *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001. NS, no significant difference.

To confirm that the effects observed upon SP141 treatment resulted from inhibition of chromatin-bound MDM2, we next measured tumor growth upon injection of IB115 DD-LPS cells that stably expressed the MDM2 ΔAD-MDM2-C464A mutant. Consistent with our in vitro data, tumors generated with these cells grew as efficiently as control tumors despite the daily administration of SP141, and they exhibited comparable expression of MDM2 target genes implicated in serine metabolism when compared to control cells (Fig. 7, E to G, and fig. S10B). These results show that inhibition of chromatin-bound MDM2 by SP141 influences serine metabolism in vivo and impairs tumor growth. Last, to confirm the effect of MDM2 inhibition in clinically relevant murine models of human LPS, we developed several patient-derived tumor xenograft (PDX) models generated upon transfer of freshly resected human tumor samples of primary naive LPS and other sarcoma subtypes into immunodeficient nude mice. In these preclinical models, administration of SP141 decreased tumor growth in three independent DD-LPS PDX models, whereas it had no significant effect (P = 0.84) on two different PDXs originating from other sarcoma subtypes (Fig. 7, H and I, and figs. S10C and S11, A to C). These data support the notion that pharmacological inhibition of MDM2 by SP141 is a suitable strategy to treat LPS.

DISCUSSION

WD- and DD-LPS, the most frequent LPS subtypes, are poorly responsive to classical chemotherapies, and there is currently no cure available for metastatic or nonresectable LPS. The first-line treatment for LPS patients includes high doses of doxorubicin, but it provides limited clinical benefit and commonly results in severe side effects (7). Hence, finding additional therapeutic strategies for LPS remains an urgent clinical challenge. The specific genetic signature of WD/DD-LPS, which harbor WT p53 but are characterized by a systematic amplification of MDM2, has been the scientific rationale for the clinical evaluation of compounds that disrupt MDM2-p53 interaction, hoping that these molecules would unleash p53-associated tumor suppressor activities. Inhibition of MDM2-p53 interaction by Nutlins, a class of pharmacological compounds that bind to a well-defined hydrophobic pocket of MDM2 involved in p53 binding, results in the accumulation of active p53 and causes cell cycle arrest or cell death in most p53-proficient tumor cells in vitro (37). Unfortunately, clinical data showed that compounds of the Nutlin family have little, if any, efficacy in LPS and result in drug resistance mainly attributed to p53 mutations (25). Our data suggest that the poor clinical response of LPS to Nutlins likely reflects the paradoxical stabilization of MDM2 that stems from p53-mediated activation of MDM2 transcription. Moreover, we show that Nutlin-3A increases the amount of chromatin-bound MDM2 in LPS cells and promotes the expression of its target genes implicated in amino acid metabolism, a process that may further enhance its pro-oncogenic activities. Thus, as exemplified by Nutlins, therapies aiming at interfering with MDM2-p53 interaction but resulting in MDM2 stabilization are likely to provide a poor clinical response in LPS. In contrast, we provide genetic and pharmacological evidence that targeting MDM2-associated metabolic functions represents an efficient alternative strategy for LPS. In that context, MDM2 degraders, such as SP141, should be considered as a potential strategy for targeting p53-independent metabolic functions of MDM2 that should synergize with p53-mediated effects. It is noteworthy that LPSs appear to be more sensitive to pharmacological inhibition of MDM2 by SP141 than other sarcoma subtypes due to the addiction of WD/DD-LPS cells to MDM2. Daily treatment with this inhibitor resulted in no detectable side effects in xenograft models. Moreover, the lower IC50 observed in LPS cells compared to that determined in the panel of untransformed cells in which we evaluated the efficacy of SP141 opens a window of opportunity for clinical applications. Further work will be required to evaluate the efficacy of this MDM2 inhibitor in other cancer types harboring MDM2 amplification.

Metabolic rewiring is considered a hallmark of cancer cells, and many oncogenes or tumor suppressors control multiple metabolic pathways (38). Among those, the p53 protein has been shown to be a central regulator of metabolism (39). MDM2 also plays important roles in cellular metabolism independently of p53, by regulating serine metabolism, the folate cycle, redox homeostasis, mitochondrial functions, and respiration (19, 31, 4042). Different metabolic pathways have been suggested to contribute to LPS development, including the nucleoside salvage pathway and de novo long-chain fatty acid synthesis (43, 44). Here, we demonstrate that interfering with MDM2-mediated control of serine and glycine metabolism, or inhibiting PHGDH, has anticancer effects both in vitro and in clinically relevant preclinical murine models of LPS. Moreover, our observation showing that these effects can be further increased in conditions where an exogenous source of serine and glycine is limited raises important questions about the potential interest of using serine- and glycine-poor diets to exacerbate the effects of pharmacological inhibitors of this metabolic pathway (fig. S12).

Serine and glycine metabolism supports the growth of cancer cells by contributing to their anabolic demands and their epigenome as well as by regulating their redox state (22). We found that MDM2 controls serine and glycine metabolism and sustains LPS growth by promoting de novo nucleotide synthesis. Interfering with MDM2 functions in purine and pyrimidine synthesis compromises the proliferative and survival capacities of LPS cells, thereby impinging on their tumorigenic potential. Our findings are intriguing in light of a previous report showing that high-grade LPS cells exhibit an up-regulation of the nucleoside salvage pathway as a consequence of increased deoxycytidine kinase activity (44). Together, these data suggest that LPS cells use several metabolic pathways in a nonmutually exclusive manner to sustain their high demand for nucleotides, a notion that is consistent with their high-proliferative rate and rapid growth in patients. Validation of this hypothesis will require further investigations, but these results open potential avenues for combinatorial therapies based on MDM2 inhibitors and the nucleoside analog gemcitabine.

Our data indicate that pharmacological inhibition of MDM2 by SP141 impinges on proliferation and induces cell death in LPS cells, at least partly through parthanatos. Parthanatos is a caspase-independent cell death pathway that is distinct from apoptosis, necrosis, or other identified forms of regulated cell death (45). Parthanatos can be induced by toxic stimuli that induce nuclear PARP-1 excessively, causing the accumulation of poly(ADP-ribosyl)ated proteins in mitochondria and the translocation of AIF from the mitochondria to the nucleus, where it induces large-scale DNA fragmentation and ultimately causes cell death. Parthanatos can be triggered by multiple cues, including severe DNA damage and oxidative stress. Although we identified massive DNA damage in SP141-treated LPS cells, we do not exclude the possibility that other defects, such as increased production of reactive oxygen species that can result from MDM2 inhibition (19), also contributed to the induction of parthanatos, and likely of other forms of cell death, in these cells. Although parthanatos has been mainly reported so far to contribute to cardiovascular and renal disorders, diabetes, and neurological diseases (45), our data suggest that strategies aiming at triggering parthanatos may be suitable for certain cancer types, in particular for cancers harboring MDM2 overexpression. Together, our data showing that the control of cellular metabolism by chromatin-bound MDM2 is an important driver of liposarcomagenesis pave the way for potential strategies for treating this cancer type, for which there are currently few therapeutic options.

MATERIALS AND METHODS

Study design

This study aimed at better understanding the role of chromatin-bound MDM2 in the metabolic reprogramming of LPS cells and at evaluating the potential of MDM2 degraders as a therapeutic strategy for LPS. We assessed the amounts of chromatin-bound MDM2 in LPS cell lines and in patient-derived tumor samples. Reverse transcription quantitative polymerase chain reaction (RT-qPCR), immunoblot, ChIP, and xenograft experiments allowed us to demonstrate that MDM2 is a master regulator of serine metabolism in LPS. We combined pharmacological and genetic approaches to demonstrate the importance of MDM2-mediated regulation of serine metabolism in LPS development both in vitro and in vivo. Stable isotope tracing experiments performed upon pharmacological or genetic inhibition of MDM2 indicated that MDM2 controls the channeling of serine into de novo nucleotide synthesis to sustain cell proliferation and cell survival. Last, the clinical potential of the MDM2 degrader SP141 was confirmed in several LPS mouse models.

All data from in vitro experiments were integrated from at least three biological replicates, with the different cell lines indicated in table S1. The researchers were blinded for the measurement of tumor sizes in vivo and for the analysis of in vitro results.

Human samples were obtained from patients with histologically confirmed sarcoma (table S2) upon validation by board-certified pathologists. Samples were included in our clinical biobank after the written agreement of patients and approval by the local Ethics Committee and the national agency for drug safety [Agence Nationale de Sécurité du Médicament (ANSM); ANSM number: 2016-A00638-43].

For mouse xenograft essays, the size of the experimental groups was calculated on the basis of previous experiments to reach a statistical test power of 0.80 and a significance of 0.05. Groups of mice were randomized by the size of the tumor at the beginning of the experiment. Experiments were performed in accordance with protocols approved by the French Council of Animal Care guidelines and national ethical guidelines of the Institut National de la Santé et de la Recherche Médicale (INSERM) Animal Care Committee (authorization CEEA-LR-12067).

Data collection for each experiment is detailed in the respective figures, figure legends, and methods. No data were excluded from studies in this manuscript. Statistical tests were chosen based on the nature of variables, assumption of data distribution, and size. In general, Student’s t tests or nonparametric Mann-Whitney U tests were used.

IC50 measurement

Growth inhibition after SP141 treatment (IC50) was determined using the sulforhodamine B (SRB) assay. Cells were seeded in 96-well plates (5000 cells per well) in complete medium for assays in triplicates. Twenty-four hours later, serial dilutions of the indicated compounds were added to the cells. After 48 hours, cells were fixed by adding a trichloroacetic acid solution (10% final concentration) and stained with a 0.4% SRB solution in 1% acetic acid. Fixed SRB was dissolved in 10 mM tris-HCl solution, and absorbance at 560 nm was read using an MRX spectrophotometer.

13C stable isotope tracing experiments

Measurement of 13C incorporation in intracellular metabolites upon incubation with 13C-labeled glucose or serine was performed as previously described (20). For more details, see Supplementary Materials and Methods.

Xenografts

Bilateral subcutaneous injections of 5 × 106 LPS cells were carried out on 8-week-old CD-1-Foxn1nu mice (Charles River). One week before injection, mice were fed with control diet (amino acid diet, TD99366, Harlan) or the same diet lacking serine and glycine (Harlan). The diets had equal caloric value (3.9 kcal/g) and equal amounts of total amino acids (179.6 g/kg). Total food intake was controlled to be identical in all experimental groups. Mice were housed in a pathogen-free barrier facility in accordance with the regional ethics committee for animal welfare (n°CEEA-LR-12067). Volumetric measurements of xenografted tumors were performed every 3 days by the same person using a manual caliper [volume = (length × width2)/2]. All animals were euthanized when the first animal reached the ethical endpoint (volume = 1500 cm3 or ulceration), and xenograft tumors and other organs were removed and snap-frozen for Western blot analysis, IHC, and hematoxylin and eosin (H&E) staining. SP-141 was dissolved in polyethylene glycol 400 (PEG400):ethanol:saline (57.1:14.3:28.6, v/v/v) and administered by intraperitoneal injection at a dose of 40 mg/kg per day, 5 days per week. LPS and sarcoma PDX mouse models were established in collaboration with the surgical and pathology departments of the Institut du Cancer de Montpellier (ICM) by inserting a tumor fragment of about 40 mm3 in the interscapular region of nude mice (Swiss nude, Charles River).

Statistical analysis

Data are expressed as mean ± SD. Statistical significance was evaluated using Student’s t tests or nonparametric Mann-Whitney U tests. P values less than 0.05 were considered to be statistically significant. *P < 0.05, **P < 0.01, and ***P < 0.001. Original data are in data file S1 for sample sizes (n < 20 per group).

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/12/547/eaay2163/DC1

Materials and Methods

Fig. S1. MDM2 is preferentially localized in the nucleus in LPS cells.

Fig. S2. MDM2 is preferentially recruited to chromatin in LPS cells.

Fig. S3. The MDM2-ATF4 complex regulates serine metabolism in LPS cells.

Fig. S4. MDM2 regulates serine metabolism in LPS.

Fig. S5. Nutlin-3A enhances serine metabolism in LPS.

Fig. S6. The pharmacological inhibitor SP141 inhibits MDM2-associated metabolic functions in nucleotide synthesis in DD-LPS cells.

Fig. S7. Stable isotope tracing experiments showing inhibition of MDM2-associated metabolic functions by SP141 in DD-LPS cells.

Fig. S8. Cell proliferation and survival defects upon pharmacological inhibition of MDM2 by SP141.

Fig. S9. MDM2 degrader as a potential therapeutic strategy for LPS.

Fig. S10. MDM2 metabolic function as a potential therapeutic strategy for LPS.

Fig. S11. Characterization of PDX models of sarcomas.

Fig. S12. Inhibition of MDM2-mediated control of serine metabolism as a potential therapeutic strategy for LPS.

Table S1. Sarcoma cell lines.

Table S2. Characterization of PDX models.

Table S3. Primers for qPCR.

Data file S1. Individual data from sample sizes (n < 20 per group).

REFERENCES AND NOTES

Acknowledgments: We thank the CRB-ICM (BB-033-0059) and CRB-IUCT for the tumor samples supplied for this study. We thank M. Lacroix for input and critical reading of the manuscript. We thank all members of the animal, imaging, cytometry, and histology facilities (Unité Mixte de Service, UMS3426 Montpellier BioCampus) for technical help. We thank the Cancéropole Grand Sud Ouest (cGSO) for their support for metabolomic analyses. Funding: This research was supported by grants from the Fondation ARC, the Ligue contre le Cancer (Equipe labelisée 2016), INSERM, and Montpellier comprehensive cancer center (SIRIC, grant 12553). G.A. was supported by a fellowship from the Fondation de France. M.Y.C. was supported by a fellowship from the French ministry of research and the FRM foundation. Author contributions: A.M.-G and M.-C.C provided LPS samples and contributed to their histological analysis. M.Y.C., L.L.C., and L.K.L., designed the studies, interpreted the data, and wrote the manuscript; M.Y.C., S.P., L.G., R.R., L.K.L., M.F., M.H., G.A., and F.B. performed the experiments; M.Y.C., L.G., L.K.L., C.T., and H.D. contributed to the in vivo experiments. N.F., S.C., and S.L.G. provided medical expertise, and J.-C.P. and F.C. assisted with data interpretation. Competing interests: L.K.L., L.L.C., M.Y.C., R.R., N.F., and S.C. are inventors on patent application EP17306681 submitted by INSERM Transfert that covers “Methods for the diagnosis and treatment of liposarcoma.” All other authors declare that they have no competing interests. Data and materials availability: All the data used for this study are present in the paper or the Supplementary Materials.

Stay Connected to Science Translational Medicine

Navigate This Article