DIMP53‐1: a novel small‐molecule dual inhibitor of p53–MDM2/X interactions with multifunctional p53‐dependent anticancer properties

The transcription factor p53 plays a crucial role in cancer development and dissemination, and thus, p53‐targeted therapies are among the most encouraging anticancer strategies. In human cancers with wild‐type (wt) p53, its inactivation by interaction with murine double minute (MDM)2 and MDMX is a common event. Simultaneous inhibition of the p53 interaction with both MDMs is crucial to restore the tumor suppressor activity of p53. Here, we describe the synthesis of the new tryptophanol‐derived oxazoloisoindolinone DIMP53‐1 and identify its activity as a dual inhibitor of the p53–MDM2/X interactions using a yeast‐based assay. DIMP53‐1 caused growth inhibition, mediated by p53 stabilization and upregulation of p53 transcriptional targets involved in cell cycle arrest and apoptosis, in wt p53‐expressing tumor cells, including MDM2‐ or MDMX‐overexpressing cells. Importantly, DIMP53‐1 inhibits the p53–MDM2/X interactions by potentially binding to p53, in human colon adenocarcinoma HCT116 cells. DIMP53‐1 also inhibited the migration and invasion of HCT116 cells, and the migration and tube formation of HMVEC‐D endothelial cells. Notably, in human tumor xenograft mice models, DIMP53‐1 showed a p53‐dependent antitumor activity through induction of apoptosis and inhibition of proliferation and angiogenesis. Finally, no genotoxicity or undesirable toxic effects were observed with DIMP53‐1. In conclusion, DIMP53‐1 is a novel p53 activator, which potentially binds to p53 inhibiting its interaction with MDM2 and MDMX. Although target‐directed, DIMP53‐1 has a multifunctional activity, targeting major hallmarks of cancer through its antiproliferative, proapoptotic, antiangiogenic, anti‐invasive, and antimigratory properties. DIMP53‐1 is a promising anticancer drug candidate and an encouraging starting point to develop improved derivatives for clinical application.

The transcription factor p53 plays a crucial role in cancer development and dissemination, and thus, p53-targeted therapies are among the most encouraging anticancer strategies. In human cancers with wild-type (wt) p53, its inactivation by interaction with murine double minute (MDM)2 and MDMX is a common event. Simultaneous inhibition of the p53 interaction with both MDMs is crucial to restore the tumor suppressor activity of p53. Here, we describe the synthesis of the new tryptophanol-derived oxazoloisoindolinone DIMP53-1 and identify its activity as a dual inhibitor of the p53-MDM2/X interactions using a yeast-based assay. DIMP53-1 caused growth inhibition, mediated by p53 stabilization and upregulation of p53 transcriptional targets involved in cell cycle arrest and apoptosis, in wt p53-expressing tumor cells, including MDM2-or MDMX-overexpressing cells. Importantly, DIMP53-1 inhibits the p53-MDM2/X interactions by potentially binding to p53, in human colon adenocarcinoma HCT116 cells. DIMP53-1 also inhibited the migration and invasion of HCT116 cells, and the migration and tube formation of HMVEC-D endothelial cells. Notably, in human tumor xenograft mice models, DIMP53-1 showed a p53-dependent antitumor activity through induction of apoptosis and inhibition of proliferation and angiogenesis. Finally, no genotoxicity or undesirable toxic effects were observed with DIMP53-1. In conclusion, DIMP53-1 is a novel p53 activator, which potentially binds to p53 inhibiting its interaction with MDM2 and MDMX. Although target-directed, DIMP53-1 has a multifunctional activity, targeting major hallmarks of cancer through its antiproliferative, proapoptotic, antiangiogenic, anti-invasive, and antimigratory properties. DIMP53-1 is a promising anticancer drug candidate and an encouraging starting point to develop improved derivatives for clinical application.

Introduction
The sequence-specific transcription factor p53 regulates a plethora of genes involved in crucial cellular processes, including cell cycle arrest, cell death, and DNA repair (Hong et al., 2014). Inactivation of the p53 tumor suppressor function is a common event in human cancers with a dramatic impact on tumor development and dissemination (Burgess et al., 2016;Hong et al., 2014;Wade et al., 2013). Although a substantial proportion of cancers harbor wild-type (wt) p53, its function is found inactivated or at least inhibited (Burgess et al., 2016;Wade et al., 2013), mainly by the murine double minute (MDM) proteins, MDM2 and MDMX (or MDM4). Mechanistically, both MDMs bind to p53 inhibiting its transcriptional activity. Additionally, the E3 ligase MDM2 triggers p53 ubiquitin-proteasome degradation. Although MDMX has no E3 ligase activity, the MDM2-MDMX heterodimer ubiquitinates p53 with higher efficiency than MDM2 homodimers (Burgess et al., 2016;Gomes et al., 2016;Wade et al., 2013). Therefore, both MDMs are powerful oncogenes, commonly overexpressed in several human cancers (Burgess et al., 2016).
Accumulating data demonstrate that wt p53 is a valuable therapeutic target and that its activation through inhibition of the p53-MDM interactions is a promising anticancer strategy. Many p53-MDM2 interaction inhibitors have been identified, several of which are currently under clinical trials (Burgess et al., 2016;Gomes et al., 2016;Wade et al., 2013). Nonetheless, given the distinct and cooperative function of both MDMs on p53 inactivation (Burgess et al., 2016;Gomes et al., 2016;Wade et al., 2013), and the resistance of MDMX-overexpressing cells to MDM2-only inhibitors (e.g., Nutlin-3a) (Li and Lozano, 2013), small molecules that suppress the inhibitory effect of both MDMs represent the ideal strategy for full p53 reactivation (Burgess et al., 2016;Gomes et al., 2016;Wade et al., 2013). However, the availability of such compounds is still limited (Graves et al., 2012;Lee et al., 2011;Soares et al., 2015a).
Here, we report the identification of DIMP53-1 as a new p53 activator, which potentially binds to p53 inhibiting its interaction with MDM2 and MDMX. Additionally, DIMP53-1 has in vitro and in vivo p53-dependent antitumor properties, involving antiproliferative, proapoptotic, antiangiogenic, anti-invasive, and antimigratory activities.

Yeast-based screening assay
Saccharomyces cerevisiae cells expressing human wt p53 alone or co-expressed with human MDM2/MDMX were used, as described (Soares et al., 2015a). Briefly, cells were grown in galactose-selective medium with compounds or 0.1% DMSO for 42 h; cell growth was analyzed by colony-forming unit counts with the determination of EC 50 (concentration that causes 50% of effect) values.

Cell cycle and apoptosis
HCT116, MCF-7, and SJSA-1 cells were seeded in sixwell plates at 1.5 9 10 5 cells/well density. Cells were thereafter treated with DIMP53-1 or solvent for 24 h. For cell cycle, cells were stained with propidium iodide (PI; Sigma-Aldrich) followed by flow cytometric analysis (Soares et al., 2015a). For apoptosis, cells were analyzed by flow cytometry using the Annexin V-FITC Apoptosis Detection Kit I (BD Biosciences, Enzifarma, Porto, Portugal) according to the manufacturer's instructions (Soares et al., 2015a).

RNA extraction and RT-qPCR
Total RNA, from HCT116 cells treated with DIMP53-1 or solvent for 24 h, was extracted using the Illustra TM RNAspin Mini RNA Isolation Kit (GE Healthcare Life Sciences, Milan, Italy). One microgram of RNA was used for cDNA synthesis using the M-MuLV reverse transcriptase and RevertAid cDNA Synthesis kit (ThermoFisher, Monza and Brianza, Italy) in 20 lL final volume, following the manufacturer's instructions. RT-qPCR assays were performed in a 384-well plate on a CFX Touch Real-Time PCR Detection System (Bio-Rad, Milan, Italy), starting with 25 ng of cDNA (Lion et al., 2013). The 2X KAPA SYBR Ò FAST qPCR Kit (Kapa Biosystems, Rome, Italy) and specific primers for BAX and CDKN1A (p21), Eurofins (MWG, Milan, Italy), were used; GAPDH and B2M were used as reference genes.

Western blot
HCT116, MCF-7, and SJSA-1 cells were seeded in sixwell plates at 1.5 9 10 5 cells/well density. After treatment with compounds or solvent, cells were lysed and the protein fractions were analyzed by western blot, as described (Soares et al., 2015b). Antibodies are described in 2.1.

Cellular Thermal Shift Assay (CETSA)
To evaluate drug target interactions in cells, the cellular thermal shift assay (CETSA) was performed as described (Tan et al., 2015). Briefly, HCT116p53 +/+ cells were lysed in appropriate buffer (25 mM Tris pH 7.4, 10 mM MgCl 2 , 2 mM DTT) by Dounce homogenization. HCT116p53 +/+ cell lysates were incubated with DIMP53-1 or solvent for 1 h at room temperature and then heated to the indicated temperatures for 3 min, cooled to room temperature for 3 min, and placed on ice. Insoluble protein was separated by centrifugation, and soluble protein was detected by western blot. In the experiments at different heating temperatures, the signal intensity was normalized to the intensity at 25°C (GAPDH denaturation with heating temperatures unable to use as loading control). At a constant temperature (39°C), the increase in nondenatured p53 was calculated setting the signal obtained with DMSO at 39°C as 0, and the signal obtained with DMSO at 25°C (considered the maximum amount of nondenatured p53) as 1.

In vitro migration and invasion assays
Cell migration was analyzed using the wound healing assay and the QCM 24-Well Fluorimetric Chemotaxis Cell Migration Kit (8 lm; Merck Millipore, VWR), as described . In the wound healing assay, confluent HCT116p53 +/+ and HMVEC-D cells, with a wound in the middle of the well, were treated with 3 lM DIMP53-1 or solvent. Cells were photographed using the Moticam 5.0MP camera with Motic's AE2000 inverted microscope with 4009 magnification at different time-points of treatment until complete closure of the wound. Wound closure was calculated by subtracting the 'wound' area (measured using IMAGE J software, Bethesda, MD, USA) at the indicated time periods after treatment from the initial (0 h) 'wound' area. In Chemotaxis Cell Migration Kit, 0.5 9 10 6 cellsÁmL À1 of HCT116p53 +/+ cells were prepared in serum-free RPMI 1640 and treated with 3 lM of DIMP53-1 or solvent, for 24 h. The prepared cell suspensions were distributed in 24-well plates (300 lL/insert), followed by the addition of 500 lL medium containing 10% FBS to the lower chamber. Cells that migrated through the 8-lm pore membranes were eluted, lysed, and stained with a green fluorescence dye that binds to cellular nucleic acids. Cell invasion was analyzed using the QCM 24-Well Fluorimetric Cell Invasion Kit (Merck Millipore, VWR), according to the manufacturer's instructions. This assay consists in the evaluation of the capacity of cell migration through the ECMatrix layer of the upper chamber of the system, and it was performed as chemotaxis cell migration assay, except the incubation time with DIMP53-1 or solvent of 48 h. In both assays, the number of migrated cells was proportional to the fluorescence signal measured using the Bio-Tek Synergy HT plate reader (Izasa, Gondomar, Portugal) at 480/520 nm (excitation/emission).

Micronucleus assay
Genotoxicity was analyzed by the cytokinesis-block micronucleus assay in human lymphocytes, as described . Briefly, fresh peripheral blood samples were collected from healthy volunteers into heparinized vacutainers. Blood samples, suspended in RPMI medium supplemented with 10% FBS, were treated with 7, 14, and 21 lM of DIMP53-1, 1 lgÁmL À1 cyclophosphamide (positive control), or solvent for 44 h. Cells were thereafter treated with 3 lgÁmL À1 cytochalasin B (cytokinesis preventive) for 28 h. Lymphocytes were isolated by density gradient separation (Histopaque-1077 and -1119), fixed in 3 : 1 methanol/glacial acetic acid, and stained with Wright stain. For each sample, 1000 binucleated lymphocytes were blindly scored using a Leica light optical microscope (Wetzlar, Germany); the number of micronuclei per 1000 binucleated lymphocytes was recorded.

In vivo antitumor and toxicity assays
Animal experiments were conducted according to the EU Directive 2010/63/EU and to the National Authorities. The BALB/c nude mice and Wistar rats (Charles-River Laboratories, Barcelona, Spain) were housed under pathogen-free conditions in individual ventilated cages. For toxicity assays, Wistar rats were treated with 50 mgÁkg À1 DIMP53-1, vehicle (DMSO), or saline solution (control) by intraperitoneal injection, twice a week, for two weeks. After four administrations, blood samples and organs (kidneys, spleen, heart, and liver) were collected for toxicological analysis. Each group was composed of four animals. Xenograft tumor assays were performed with HCT116p53 +/ + and HCT116p53 À/À tumor cells. Briefly, 1 9 10 6 HCT116 cells (in PBS) were inoculated subcutaneously in the mice dorsal flank. Tumor dimensions were assessed by caliper measurement, and their volumes were calculated [tumor volume = (L 9 W 2 )/2], where L and W represent the longest and shortest axis of the tumor, respectively. Treatment started when tumors reached approximately 100 mm 3 volume (14 days after the grafts). Mice were thereafter treated twice a week with 50 mgÁkg À1 DIMP53-1 or vehicle by intraperitoneal injection for two weeks. Tumor volumes and body weights were monitored twice a week until the end of the treatment. Animals were sacrificed by cervical dislocation at the end of the study, when tumors reached 1500 mm 3 or if the animals presented any signs of morbidity. Each group was composed of six animals.

Immunohistochemistry
Tumor tissues were fixed in 10% formalin, embedded in paraffin, sectioned at 4 lm, and stained with hematoxylin and eosin (H&E) or antibodies, as described . Briefly, antigen retrieval was performed by boiling the sections for 20 min in 10 mM citrate buffer (pH 6.0) for staining with all antibodies, except for anti-VEGF for which tissues were treated with 10 mM EDTA buffer (pH 8.0). Antibodies are described in 2.1. Immunostaining was carried out using the UltraVision Quanto Detection System HRP DAB Kit, from Lab Vision Thermo Scientific, according to the manufacturer's instructions. Evaluation of DAB (3,3 0 -diaminobenzidine) intensity and quantification of marked cells were performed using IMAGE J software. Microvessel densities were determined by counting CD34-positive vessels, as described (Maeda et al., 2007;Weidner et al., 1991).

TUNEL assay
TUNEL assay was performed using the In Situ Cell Death Detection Kit Fluorescein (Roche, Sigma-Aldrich), according to the manufacturer's instructions, as described .

Flow cytometric data acquisition and analysis
The Accuri TM C6 flow cytometer and the CELLQUEST software (BD Biosciences, Enzifarma) were used. The FLOWJO software (Ashland, OR, USA) was used to identify and quantify cell cycle phases.

Results
3.1. Identification of DIMP53-1 as a potential dual inhibitor of the p53-MDM2/X interactions using a yeast-based assay Derivatives of SLMP53-1, a tryptophanol-derived oxazoloisoindolinone p53 activator  [International Patent (Soares et al., 2014)], containing different protective groups in the nitrogen of the indole moiety, were synthesized. The effect of this new chemical library on p53-MDM2 and p53-MDMX interactions was thereafter investigated, using the reported yeast-based screening assay (Soares et al., 2015a). In this assay, the expression of human wt p53 in yeast causes growth arrest that is inhibited by human MDM2 or MDMX. The effect of 0.1-50 lM compounds was evaluated, and DIMP53-1 (Fig. 1A) was identified as a potential dual inhibitor of the p53-MDM2/X interactions (Fig. 1B). Nutlin-3a and SJ-172550 were used as positive controls because, as in human cells (Reed et al., 2010;Vassilev et al., 2004), they inhibit the negative effect of MDM2 and MDMX, respectively, having no impact on the other MDM (Fig. 1B). Contrary to Nutlin-3a and SJ-172550, DIMP53-1 relieved the negative effect of both MDMs on p53 (Fig. 1B). Based on EC 50 values, DIMP53-1 was less effective than Nutlin-3a on p53-MDM2 interaction, but more effective than SJ-172550 on p53-MDMX interaction (Fig. 1C). Additionally, 0.1-50 lM DIMP53-1 did not interfere with the growth of control yeast (data not shown), corroborating its selectivity toward the p53-MDM2/X interactions.
The ability of DIMP53-1 to block the p53-MDM2/X interactions in yeast was further demonstrated by Co-IP (Fig. 1D). Actually, 10 and 20 lM DIMP53-1 led to a visible decrease in the amount of MDM2 or MDMX co-immunoprecipitated with p53 in yeast cells coexpressing p53 and MDM2 or MDMX, respectively.
3.2. DIMP53-1 causes p53 stabilization and upregulation of p53 transcriptional targets through potential binding to p53, inhibiting its interaction with MDM2 and MDMX, in human tumor cells To confirm the molecular mechanism of action of DIMP53-1 as a dual inhibitor of the p53-MDM2/X interactions, its activity was ascertained in p53 +/+ and p53 À/À HCT116 cells. The SRB assay revealed a significant reduction in the DIMP53-1 growth inhibitory effect in the absence of p53, during 24-and 48-h treatment ( Fig. 2A). Despite the selectivity of DIMP53-1 to the p53-pathway, this compound also inhibited the growth of HCT116p53 À/À cells. This may indicate an alternative mechanism of action of DIMP53-1 independent of p53, for longer incubation times and higher concentrations of compound.
hypothesized that DIMP53-1 might bind to p53. To confirm this hypothesis, the potential interaction of DIMP53-1 with p53 was checked by CETSA. In this assay, we analyzed the impact of DIMP53-1 on p53 thermal stabilization, measured by the amount of soluble p53 upon heating. From 39°C to 42°C, 10 lM DIMP53-1 caused significant p53 thermal stabilization (Fig. 3A). Additionally, DIMP53-1 induced a concentration-dependent p53 thermal stabilization, at 39°C (Fig. 3B), reestablishing the levels of nondenatured p53 protein observed at 25°C in DMSO-treated sample. Furthermore, 100 lM DIMP53-1 (the highest concentration tested to demonstrate the interaction of DIMP53-1 with p53) did not interfere with MDM2 and MDMX thermal stabilization at different heating temperatures (Fig. S2).
Altogether, these results demonstrate that DIMP53-1 is a selective activator of the p53-pathway, suppressing the MDM2 and MDMX inhibitory effect in human tumor cells due to a potential interaction with p53.

DIMP53-1 is nongenotoxic in tumor and normal cells and has low cytotoxicity against normal cells
The genotoxicity of DIMP53-1 was evaluated in tumor and normal cells. In HCT116p53 +/+ tumor cells, the impact of DIMP53-1 on DNA damage was analyzed by checking comet-positive cells and histone H2AX phosphorylated on serine 139 (cH2AX; phosphorylation of histone H2AX marks the first step in cellular response to DNA double-strand breaks, and its visualization allows the assessment of DNA damage). The results obtained showed that, unlike etoposide (positive control), 7, 14, and 21 lM DIMP53-1 did not increase the percentage of comet-positive cells after 48h treatment (Fig. 4A,B), or the levels of cH2AX after 12-h treatment (Fig. 4C). Furthermore, in peripheral lymphocytes of normal individuals, 7, 14, and 21 lM DIMP53-1 did not increase the number of micronuclei compared to solvent (Fig. 4D,E). DIMP53-1 cytotoxicity was also checked in normal cells by assessing its growth inhibitory effect on breast epithelial MCF10A cells, through the SRB assay (Fig. 4F). The IC 50 value of DIMP53-1 in these cells was higher than 75 lM, supporting its selective cytotoxicity toward tumor cells.
Altogether, DIMP53-1 is nongenotoxic in human tumor and normal cells and has low cytotoxic effects against human normal cells.

DIMP53-1 reduces in vitro angiogenesis and tumor cell migration and invasion
The impact of DIMP53-1 on HCT116p53 +/+ cell migration and invasion was investigated. With such purpose, the concentration of 3 lM (IC 10 ) of DIMP53-1, which does not significantly interfere with HCT116p53 +/+ cell growth, was used. In the wound healing assay, 3 lM of DIMP53-1 significantly inhibited HCT116p53 +/+ cell migration, and the subsequent wound closure, when compared to solvent (Fig. 5A,B). These results were confirmed using the chemotaxis cell migration assay, in which 24-h treatment with 3 lM of DIMP53-1 led to a significant reduction in HCT116p53 +/+ cell migration compared to solvent (Fig. 5C). Additionally, 3 lM of DIMP53-1 inhibited HCT116p53 +/+ cell invasion through a Matrigel Ò matrix, evaluated by the cell invasion assay after 48-h treatment (Fig. 5D).
The antiangiogenic potential of DIMP53-1 was also investigated. Initially, its antiproliferative effect on HMVEC-D endothelial cells was assessed, and an IC 50 higher than 50 lM was obtained (data not shown), indicating low toxicity of DIMP53-1 toward endothelial cells. Thereafter, the wound healing assay was performed to evaluate the effect of 14 lM of DIMP53-1 (IC 10 value in HMVEC-D) on HMVEC-D cell migration. At this concentration, a significant decrease in endothelial cell migration was observed (Fig. 5E,F). Moreover, using an in vitro angiogenesis assay, a significant antiangiogenic effect was observed after 16-h treatment with 10 and 14 lM of DIMP53-1. In fact, DIMP53-1 led to a dose-dependent decrease in HMVEC-D tube formation (Fig. 5G,H).

DIMP53-1 has in vivo antitumor activity without apparent toxic side effects
To evaluate some primary toxicity signs, 50 mgÁkg À1 DIMP53-1 was tested in Wistar rats. Following the same administration procedure conducted in tumor , 10 lM of DIMP53-1 was used and lysate samples were heated at different temperatures; plot represents the signal intensity normalized to the intensity at 25°C. In (B), lysate samples were treated with increasing DIMP53-1 concentrations and heated at 39°C; plot represents the increase in nondenatured p53 calculated setting the signal obtained with DMSO at 39°C as 0, and the signal obtained with DMSO at 25°C (considered the maximum amount of nondenatured p53) as 1. Results are mean AE SEM of three independent experiments. (C) DIMP53-1 concentration-response growth curves in SJSA-1 and MCF-7 cells, after 48-h treatment; data are mean AE SEM of four independent experiments; incubation with DMSO, in equivalent % of DIMP53-1, was used to normalize the results. (D,E) Cell cycle arrest (D) and apoptosis (E) were determined in SJSA-1 and MCF-7 cells at IC 50 and 2 9 IC 50 (2 9 DIMP53-1) concentrations, after 24-h treatment; data are mean AE SEM of three independent experiments; values were significantly different from DMSO (*P < 0.05; **P < 0.01; ***P < 0.001). (F) Western blot analysis was performed in SJSA-1 and MCF-7 cells, after 24-h (p21) and 48-h (PARP, p53, MDM2, BAX) treatments with the IC 50 of DIMP53-1 or DMSO. In (A), (B), and (F), immunoblots are representative of three independent experiments; in (B) and (F), GAPDH was used as loading control. xenograft mice models, organs' relative weight (trophism) and biochemical and hematological data were analyzed for saline, vehicle, and DIMP53-1 groups (Table 1). No differences between the three groups on relative weight of liver, kidneys, heart, and spleen were observed. Concerning biochemical data, only a slight decrease in urea in the vehicle group compared to the saline group, and a slight increase in uric acid in the DIMP53-1 group compared to controls (saline and vehicle groups) were observed. These results indicated no apparent liver or kidney toxicity. Regarding hematological data, just a small increase in reticulocyte number was observed in the vehicle group compared to the saline group, with no alterations between DIMP53-1 and the control groups. Overall, no apparent toxic side effects were observed for DIMP53-1 on the tissues most commonly affected by conventional chemotherapeutics. Fig. 4. DIMP53-1 is nongenotoxic in normal and tumor cells and has low cytotoxicity against normal cells. (A-C) DNA damage was measured in HCT116p53 +/+ cells by comet assay (A and B) and by analysis of cH2AX expression levels (C) after treatment with etoposide (ETOP; positive control) or DIMP53-1. In (A), scale bar = 20 lm; magnification = 200 9. In (B), quantification of comet-positive cells (containing more than 5% of DNA in the tail; assessed by OPEN COMET/IMAGEJ); 100 cells were analyzed in each group. In (C), cH2AX levels were determined by western blot; immunoblots are representative of three independent experiments; GAPDH was used as loading control. (D and E) Genotoxicity of 7, 14, and 21 lM DIMP53-1 by cytokinesis-block micronucleus (MN) assay after 72-h treatment in human lymphocyte cells; 5 lgÁmL À1 cyclophosphamide (CP; positive control). In (D), scale bar = 20 lm; magnification = 1000 9. In (E), the number of MN per 1000 binucleated lymphocytes was recorded. (F) DIMP53-1 concentration-response growth curve in MCF10A cells, after 48-h treatment; incubation with DMSO, in equivalent % of DIMP53-1, was used to normalize the results. In (B), (E), and (F), data are mean AE SEM of three to four independent experiments; in (B) and (E), values were significantly different from DMSO (**P < 0.01; ***P < 0.001).
The in vivo antitumor potential of DIMP53-1 was evaluated using human tumor xenograft mice models of p53 +/+ and p53 À/À HCT116 cells. Four intraperitoneal administrations of 50 mgÁkg À1 DIMP53-1 inhibited the growth of p53-expressing HCT116 tumor compared to vehicle (Fig. 6A). Conversely, for the same conditions, DIMP53-1 did not interfere with the growth of p53-null HCT116 tumors, further reinforcing its p53-dependent antitumor activity (Fig. 6A). Furthermore, no significant body weight loss or morbidity signs were observed in DIMP53-1treated mice compared to vehicle (Fig. 6B).
The subsequent analysis of tumor tissues was performed to check in vivo p53-dependent antitumor events promoted by DIMP53-1. Proliferation, apoptosis, and angiogenesis markers were checked in p53 +/+ and p53 À/À HCT116 tumor tissues by immunohistochemistry and TUNEL staining (Fig. 6C-F). In p53-expressing tumor tissues, DIMP53-1 reduced proliferation (decrease in Ki-67-positive staining) and stimulated apoptosis (increase in BAX expression and DNA fragmentation demonstrated by TUNEL-positive staining), compared to vehicle (Fig. 6C-E). To study the angiogenic profile of tumor tissues, the vascular endothelial growth factor (VEGF; angiogenesisinducing factor) and the microvessel density (MVD; determined using the marker of newly formed vessels CD34) were determined. The results obtained revealed Data were analyzed for saline, vehicle, and 50 mgÁkg À1 DIMP53-1 (treated) rat groups, after four intraperitoneal administrations (twice a week). Results are mean AE SEM of four independent experiments; *P < 0.05 (comparison was made between saline and vehicle groups, and between vehicle and treated groups). ALT, alanine aminotransferase; AST, aspartate aminotransferase; BW, body weight; CK, creatine kinase; HCT, hematocrit; HGB, hemoglobin concentration; PCT, plateletcrit; PLT, platelet; RBC, red blood cell count; RET, reticulocytes; WBC, white blood cells. lower levels of VEGF and MVD in p53-expressing tumor tissues treated with DIMP53-1, compared to vehicle (Fig. 6C,D,F). Particularly, an overall fivefold reduction in MVD was observed in p53-expressing tumor tissue treated with DIMP53-1 (Fig. 6F). Conversely, no apparent differences in these markers were observed between DIMP53-1 and vehicle in p53-null tumors (Fig. 6C-F). Altogether, these results support an in vivo p53dependent antiproliferative, proapoptotic, and antiangiogenic activity of DIMP53-1.

Discussion
The complexity underlying cancer highlights the need for the development of strategies that would target the intricate cancer network, in order to modulate the different hallmarks of cancer and hence treat cancer as a complex disease. The key role of p53 in cancer hallmarks renders p53-targeted anticancer therapies highly encouraging. Actually, an effective inhibition of cancer development and progression has been achieved with strategies devised to rectify a dysfunctional p53 pathway, particularly due to MDM2 and MDMX overexpression (Hong et al., 2014;Li and Lozano, 2013;Wade et al., 2013). Consistently, the dual inhibition of the p53-MDM2/X interactions, for full p53 reactivation, has gained strength to treat wt p53-expressing tumors, particularly MDMX-overexpressing tumors commonly resistant to only MDM2 inhibitors (Burgess et al., 2016;Li and Lozano, 2013).
Here, we report a novel tryptophanol-derived oxazoloisoindolinone, DIMP53-1, identified as a new dual inhibitor of the p53-MDM2/X interactions. The molecular mechanism of action of DIMP53-1, identified in yeast, was validated in human tumor cells with and without p53. In fact, DIMP53-1 caused tumor cell growth inhibition mediated by p53 stabilization and upregulation of p53 transcriptional targets involved in cell cycle arrest and apoptosis, in wt p53-expressing tumor cells, including MDM2-or MDMX-overexpressing cells. Notably, DIMP53-1 inhibited the p53-MDM2/X interactions by potentially binding to p53 in human tumor cells.
The loss of p53 function has been related to the development of a metastatic phenotype (Powell et al., 2014), the most frequent cause of mortality in patients with cancer (Cordani et al., 2016;Spano et al., 2012). Actually, p53 stimulates the transcription of repressors of cell migration and invasion (Powell et al., 2014). It is therefore expected that the restoration of p53 function may suppress cancer dissemination. Thus, an effective antimigratory and anti-invasive activity of DIMP53-1 was demonstrated in wt p53-expressing tumor cells.
Angiogenesis is a major hallmark of cancer as proliferation and metastatic spread of cancer cells depend on the adequate supply of oxygen and nutrients (Baeriswyl and Christofori, 2009). Actually, several antiangiogenic agents have been explored in cancer treatment, particularly in combination with conventional chemotherapeutic agents (Vasudev and Reynolds, 2014). The p53 activity has been negatively correlated with this process through indirect inhibition of key proteins, such as VEGF. Particularly, it was demonstrated that p53 indirectly represses VEGF expression by inhibiting transcription factors such as SP1 and E2F (Pal et al., 2001;Qin et al., 2006). Here, the therapeutic potential of DIMP53-1 was further reinforced through confirmation of in vivo antiangiogenic activity through depletion of VEGF in tumors. Interestingly, despite the antiangiogenic activity of DIMP53-1 in an in vitro endothelial cell system (without tumor cells), the results obtained in vivo showed that this antiangiogenic effect is highly dependent on tumor environment, particularly of the p53 status in these tumors. In fact, the antiangiogenic activity of DIMP53-1 was suppressed in p53-null tumor xenografts. These results emphasize a strong connection between the activation of the p53 pathway by DIMP53-1 and its antiangiogenic activity in tumors. Further studies are required to clarify the molecular pathways involved in DIMP53-1 antiangiogenic activity.
Additionally, DIMP53-1 is nongenotoxic in both normal and tumor cells and presents no significant toxicity both in normal cells and in rats. Most importantly, in human tumor xenograft mice models, a p53dependent antitumor activity of DIMP53-1 was observed. Actually, DIMP53-1 suppressed the growth of wt p53-expressing tumors, through inhibition of proliferation and induction of apoptosis, without interfering with the growth of p53-null tumor xenografts.
Interestingly, to date, only the furan derivative RITA was reported as a small-molecule inhibitor of the p53-MDM2 interaction by binding to p53 instead of MDM2 Hong et al., 2014;Wade et al., 2013). However, conversely to RITA, DIMP53-1 has no genotoxic effects, also acts on the p53-MDMX interaction, and exhibits selectivity to the p53 pathway, particularly highlighted in in vivo mice models.

Conclusions
This work reports the identification of a new potential p53 ligand, which activates the p53 function through inhibition of its interaction with MDM2/X. DIMP53-1 has in vivo p53-dependent antitumor properties with no apparent toxic side effects, and exhibits antiproliferative, proapoptotic, antiangiogenic, anti-invasive, and antimigratory activities. Collectively, although DIMP53-1 is a targeted agent, it also presents a multifunctional activity, interfering with several hallmarks of cancer. Besides its great promise as an anticancer drug candidate, DIMP53-1 is also an encouraging starting point for further development of dual inhibitors of the p53-MDM2/X interactions with improved therapeutic potential for clinical translation.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Scheme S1. Synthesis of DIMP53-1. Fig. S1. 1 H NMR and 13 C NMR data for compound DIMP53-1. Fig. S2. CETSA experiments of MDM2 and MDMX.