Discovery of a new small-molecule inhibitor of p53–MDM2 interaction using a yeast-based approach

, a known small-molecule inhibitor of p53–MDM2 interaction, pyranoxanthone 1 binds to the p53-binding site of MDM2. Overall, in this work, a novel small-molecule inhibitor of p53–MDM2 interaction with a xanthone scaffold was identiﬁed for the ﬁrst time. Besides its potential use as molecular probe and possible lead to develop anticancer agents, the pyranoxanthone 1 will pave the way for the structure-based design of a new class of p53–MDM2 inhibitors.


Introduction
Despite the huge diversity of genes implicated in tumorigenesis, the p53 tumor suppressor protein stands out as a master regulator of various signaling pathways in this process. The many roles of p53 as a tumor suppressor include the ability to induce cell cycle arrest, DNA repair, senescence, and apoptosis, among others [1][2][3][4]. The p53 activity is ubiquitously lost in cancers either by mutation in the TP53 gene, or by inactivation of its protein, thereby indicating its relevance as a therapeutic target in cancer. About half of all human cancers express a wildtype (wt) p53 form, which is inactivated due to the overexpression of the main endogenous negative regulator, murine double minute 2 (MDM2). The oncoprotein MDM2 binds p53 and negatively regulates p53 activity by direct inhibition of p53 transcriptional activity and enhancement of p53 degradation via the ubiquitin-proteasome pathway. Restoration of p53 activity by inhibiting the p53-MDM2 interaction represents an appealing therapeutic strategy for many wt p53 tumors with overexpressed MDM2, and has therefore been the focus of a large effort in drug discovery [1][2][3][4].
The virtual screening of a library of xanthone derivatives led us to the identification of potential novel MDM2 ligands. The activity of these compounds as inhibitors of p53-MDM2 interaction was investigated using a yeast phenotypic assay, herein developed for the initial screening. Using this approach, in association with a yeast p53 transactivation assay, the pyranoxanthone (3,4-dihydro-12hydroxy-2,2-dimethyl-2H,6H-pyrano [3,2-b]xanthen-6-one) (1) was identified as a putative smallmolecule inhibitor of p53-MDM2 interaction.
The activity of the pyranoxanthone 1 as inhibitor of p53-MDM2 interaction was further investigated in human tumor cells with wild-type p53 and overexpressed MDM2. Notably, the pyranoxanthone 1 mimicked the activity of known p53 activators, leading to p53 stabilization and activation of p53dependent transcriptional activity. Additionally, it led to increased protein levels of p21 and Bax, and to caspase-7 cleavage. By computational docking studies, it was predicted that, like nutlin-3a, a known small-molecule inhibitor of p53-MDM2 interaction, pyranoxanthone 1 binds to the p53-binding site of MDM2.
Overall, in this work, a novel small-molecule inhibitor of p53-MDM2 interaction with a xanthone scaffold was identified for the first time. Besides its potential use as molecular probe and possible lead to develop anticancer agents, the pyranoxanthone 1 will pave the way for the structure-based design of a new class of p53-MDM2 inhibitors.
The successful development of inhibitors of protein-protein interactions remains a remarkable challenge for medicinal chemists owing to issues that include general lack of smallmolecule scaffolds for drug design, the typical flatness of the interface, the difficulty of distinguishing real from artifactual binding, and the size and character of typical small-molecule libraries [5,6]. In spite of this, an increasing number of smallmolecule inhibitors of p53-MDM2 interaction have been developed during the last years. Most of these inhibitors have been obtained using techniques such as computational chemistry, which enables to provide increased specificity and affinity for the target [7]. Although many of these small-molecules have shown potent in vitro activity, only a limited number of compounds have demonstrated to possess desirable pharmacokinetic properties and acceptable toxicity profiles in in vivo evaluations. To date, the most studied chemotypes have been cis-imidazolines (such as nutlins), benzodiazepines, and spirooxindoles. Proofs of concept studies with some of these compounds have demonstrated that inhibitors of p53-MDM2 interaction have therapeutic potential against tumors with a wt p53 form [1][2][3][4]. Efforts are in progress to identify new scaffolds of inhibitors of p53-MDM2 interaction with improved biological activities. These molecules are particularly required as molecular probes and have the potential to be developed as anticancer drug candidates [1][2][3][4].
Xanthone derivatives have been widely reported as potential anticancer agents [reviewed in [8,9]]. In fact, in the last years, molecular modifications developed by our research group on the tricyclic xanthone scaffold led to potent inhibitors of the growth of several human tumor cell lines [10][11][12][13][14]. Particularly, prenylated xanthone derivatives have drawn attention due to their potency and selectivity against breast adenocarcinoma MCF-7 tumor cells, with wt p53 and overexpressed MDM2 [12]. This was in fact corroborated by a recent work, in which we showed that one of the tested prenylated xanthones exhibited a higher potency against tumor cells with a wt p53 form (MCF-7) than against tumor cells with a mutant p53 form (breast MDA-MB-231) [15]. In addition, reports from other authors have strictly related the antitumor activities of two prenylated xanthones, gambogic acid and amangostin, with the activation of a p53-dependent pathway [16][17][18]. Together, these works raised the hypothesis that xanthone derivatives may represent a promising chemical scaffold to look for new inhibitors of p53-MDM2 interaction.
In the present work, with the virtual screening of a library of xanthone derivatives, potential novel MDM2 ligands were identified, and their activities as inhibitors of p53-MDM2 interaction were investigated using yeast cell-based screening assays. Using this approach, the pyranoxanthone 3,4-dihydro-12-hydroxy-2,2dimethyl-2H,6H-pyrano[3,2-b]xanthen-6-one (1; Fig. 1), was identified as a promising inhibitor of p53-MDM2 interaction, which successfully activated p53 and downstream cell signaling in human tumor cells. With our findings, a new small-molecule inhibitor of p53-MDM2 interaction with a xanthone scaffold was here identified for the first time.

Plasmids
For the yeast phenotypic assay, the yeast expression vectors pGADT7-(LEU2) encoding human MDM2 (kindly provided by Dr. Xue-Min Zhang from National Center of Biomedical Analysis, China) with ADH1 constitutive promoter, and pLS89-(TRP1) encoding human wt p53 (kindly provided by Dr. Richard Iggo from Swiss Institute for Experimental Cancer Research, Switzerland) with GAL1-10 inducible promoter were used. For the yeast transactivation assay, the yeast expression vectors pTSG-(TRP) encoding human wt p53 under the control of a GAL1 inducible promoter [21] and pRB24-(HIS3) encoding human MDM2 under the control of a PGK1 constitutive promoter (kindly provided by Dr. Rainer Brachmann from Irvine University, CA, USA), were used.

Yeast strain, transformation, and growth conditions
For the yeast growth-inhibition assay, Saccharomyces cerevisiae CG379 was co-transformed using the standard lithium acetate method. For selection of co-transformed yeasts, cells were routinely grown in a minimal selective medium with 2% (w/w) glucose, 0.7% (w/w) yeast nitrogen base without amino acids from Difco (Quilaban, Sintra, Portugal) and all the amino acids required for yeast growth (50 mg mL À1 ) except leucine and tryptophan, to approximately 1 optical density at 600 nm (OD 600 ). To induce expression of wt p53 and MDM2, yeasts were diluted to 0.05 OD 600 into selective induction medium with 2% (w/w) galactose and 2% (w/w) raffinose (instead of glucose), and incubated at 30 8C under continuous orbital shaking (200 r.p.m.) for approximately 42 h (time required by control yeast, co-transformed with the empty vectors pLS239 and pGADT7, to achieve 0.45 OD 600 ). Yeast growth was analyzed by counting the number of colony-forming units (CFU) per mL (CFU mL À1 ) after 2 days incubation at 30 8C on Sabouraud Dextrose Agar plates from Liofilchem (Frilabo, Porto, Portugal).

Effects of compounds on yeast cell growth
Co-transformed yeast cells were incubated in selective induction medium in the presence of 10 mM nutlin-3a, 10 mM xanthones 1-12 or 0.1% DMSO only, for approximately 42 h, at 30 8C, under continuous orbital shaking. Yeast cell growth was analyzed as described in Section 2.3. Results were estimated considering as 100% growth the number of CFU obtained with the control yeast (co-transformed with the empty vectors).

Yeast cell cycle analysis
Flow cytometric analysis of DNA content was performed using Sytox Green Nucleic Acid, as described [22]. Briefly, 1 Â 10 7 cells were fixed in 70% (v/v) ethanol, incubated with 250 mg mL À1 RNase A (Sigma-Aldrich, Sintra, Portugal) and 1 mg mL À1 Proteinase K (Sigma-Aldrich, Sintra, Portugal), and further incubated with 10 mM Sytox Green Nucleic Acid from Invitrogen (Alfagene, Carcavelos, Portugal). For the flow cytometric analysis the FACSCalibur flow cytometer from BD Biosciences (Enzifarma, Porto, Portugal) and the CellQuest software from BD Biosciences (Enzifarma, Porto, Portugal) were used. Yeast cell cycle phases were identified and quantified using ModFit LT software (Verity Software House Inc., Topsham, USA).

Dual-luciferase yeast p53 transactivation assay
Dual-luciferase yeast p53 transactivation assay was performed basically as reported [23]. Briefly, a diploid yeast reporter strain was constructed by mating yLFM-PUMA, RFM-M2 strain with BY4704 strain. For selection of co-transformed yeasts, cells were routinely grown in minimal selective medium with all the amino acids required for yeast growth (50 mg mL À1 ) except for tryptophan and histidine. To induce expression of human wt p53, yeast cultures were diluted to approximately 0.1 OD 600 in selective induction medium containing 0.048% (w/w) galactose and 2% (w/ w) raffinose. Yeast cells, diluted in selective induction medium as referred above, were seeded into 96-well plates (120 mL/well) and incubated in the presence of 10 mM nutlin-3a, 10  Promega, Milan, Italy), followed by the addition of 5 mL of the Stop&Glow buffer (Promega, Milan, Italy), and the light units were measured again with a luminometer. In this assay, since the expression of luciferase occurs in a p53-dependent manner, the p53 transcriptional activity is reflected by the luciferase activity, which was quantified in the presence or absence of the tested compounds.

Human tumor cell lines and growth conditions
The human breast adenocarcinoma-derived MCF-7 cell line was obtained from the InterLab Cell Line Collection, ICLC (Genoa, Italy). The human colon adenocarcinoma HCT116 cell line harboring a wt p53 form (HCT116 p53 +/+ ) and its isogenic derivative, in which p53 has been knocked out (HCT116 p53 À/À ) were kindly provided by Dr. B. Vogelstein (The Johns Hopkins Kimmel Cancer Center, Baltimore, MD, USA). Cell lines were routinely cultured in RPMI with ultraglutamine medium from Lonza (VWR, Carnaxide, Portugal) supplemented with 10% fetal bovine serum from Gibco (Alfagene, Carcavelos, Portugal) and maintained in a humidified incubator at 37 8C with 5% CO 2 in air.

Dual-luciferase reporter assay in human tumor cell lines
Dual-luciferase reporter assay in human tumor cell lines was performed basically as reported in [24]. Briefly, 5 Â 10 4 cells/well were seeded into 24-well plates and incubated for 24 h before transfection. Cells were transfected at approximately 80% confluence using the Myrus LT-1 reagent (Tema Ricerca, Milan, Italy), and according to the manufacturer's instructions. Specifically, 350 ng of the pG13-luc reporter vector was used along with 50 ng of the control pRLSV40 plasmid introduced to normalize the transfection efficiency. After transfection, cells were treated with 1.5 mM doxorubicin, 10 mM nutlin-3a, 10 mM xanthones or DMSO only, for 16 h. Cells were harvested and the luciferase assay was carried out using the dual-luciferase reagent as described in Section 2.6.
Docking simulations in MDM2 (pdb code: 1YCR) were undertaken in AutoDock Vina [29]. AutoDock Vina considered the target conformation as a rigid unit, while the ligands were allowed to be flexible and adaptable to the target. A search exhaustiveness of 8 was employed. The 9 lowest energy conformations for each ligand were retrieved. Chimera 1.6.1 [30] and Pymol 0.99 [31] were used for visual inspection of results and graphical representations.

Statistical analysis
Data were analyzed statistically using the SigmaStat 3.5 software. Differences between means were tested for significance using Student's t-test (P < 0.05).

Development of a yeast phenotypic assay for the screening of inhibitors of p53-MDM2 interaction
A previous work performed by our group showed that expression of human wt p53 in S. cerevisiae induced growth inhibition associated with S-phase cell cycle arrest [22]. Here, it is shown that although the single expression of human MDM2 in yeast did not interfere with the cell growth (therefore represented by the control yeast, transformed with the empty vectors, treated with DMSO only in Figs. 2B and 3A) and cell cycle progression, its co-expression with wt p53 significantly reduced the p53-induced growth inhibition and cell cycle arrest ( Fig. 2A-C). These results strengthens the previously reported conservation in yeast of the negative effect of MDM2 on p53 activity [reviewed in [32]], and gave rise to a possible use of these cells in the development of a simple growth-inhibition screening assay to search for inhibitors of p53-MDM2 interaction. In this yeast assay, inhibitors of p53-MDM2 interaction would abolish or abate the negative effect of MDM2 on p53, restoring the p53induced growth inhibition and S-phase cell cycle arrest. The assay was, in fact, validated testing nutlin-3a, the known inhibitor of p53-MDM2 interaction [33,34]. The treatment of yeast cells coexpressing p53 and MDM2 with 10 mM nutlin-3a for 42 h markedly reduced the negative effect of MDM2 on p53-induced growth inhibition and cell cycle arrest, without interfering with the activity of p53 or MDM2 when expressed alone ( Fig. 2B and C; in Fig. 2B similar effects of nutlin-3a were obtained on control yeast and yeast cells expressing only MDM2). In fact, in the presence of nutlin-3a approximately 68% and 74.6% of the p53induced growth inhibition and S-phase cell cycle arrest, respectively, were re-established (Table 1; Fig. 2B and C). Concentrations of nutlin-3a lower than 10 mM did not significantly reduce the negative effect of MDM2 on p53-induced growth inhibition and cell cycle arrest. Moreover, for concentrations of nutlin-3a higher than 10 mM, a cytotoxic effect on control yeast was observed (data not shown). Altogether, with the obtained results a yeast phenotypic assay, based on simple measurements of yeast cell growth and analysis of yeast cell cycle, was developed that could enable us to search for inhibitors of p53-MDM2 interaction.
MDM2 residues Gly58, Asp68, Val75, and Cys77 are critical for the interaction with p53 [35]. In the virtual screening against MDM2 (PDB code: 1YCR), to study only the interactions established by a ligand in this binding site, the grid box was placed in a position so that the ligand could not interact elsewhere with the MDM2 (Fig. A1). The binding affinity values found for the best scoring conformations of the well-characterized inhibitors of p53-MDM2 interaction, which are known to establish interactions with MDM2 protein, were between À6.9 (bzd) and À5.6 (nutlin-3a) kcal mol À1 . A total of 46 xanthone derivatives showed binding affinity values in this range (Table A1), with the pyranoxanthone 1 (Fig. 1) revealing the highest binding affinity (À7.6 kcal mol À1 ). However, limiting the docking studies to the above mentioned grid, the nutlin-3a showed a binding affinity lower than the expected [36]. Nutlin-3a was previously predicted to contact with nine MDM2 amino acids [36] and, by limiting the interaction site using the grid box, nutlin-3a was not allowed to contact with all the MDM2 amino acids. Subsequently, the study of the interactions established by the ligands with the entire MDM2 protein (1YCR) was performed removing the grid box. In this case the absolute binding affinity value of nutlin-3a rose to À7.4 kcal mol À1 binding at a different site in the MDM2 protein (Table A2). However, the binding affinity of the hit pyranoxanthone 1 was the same (À7.6 kcal mol À1 ). A possible explanation for these results is that there are no other interaction sites other than the 4 amino acids selected with the grid box in the MDM2 protein that would lead to higher pyranoxanthone 1 binding affinity.
Based on the overall in silico results and on the IC 50 (concentration required to cause 50% growth inhibition) values against human tumor cells with wt p53 [10,11,13,15,19], the xanthone derivatives 1-14 were selected for the yeast cell-based screening assay (Table 1; xanthones 13 and 14 were not tested due to the low solubility in DMSO).

Identification of pyranoxanthone 1 as a promising inhibitor of p53-MDM2 interaction using the yeast approach
Using the developed yeast phenotypic assay, the activity of the most promising xanthone derivatives 1-12 on p53-MDM2 interaction was investigated. Since xanthones 10-12 were cytotoxic in control yeast, their effects on p53-MDM2 interaction were not evaluated. Among the tested xanthones, only the pyranoxanthone 1 significantly reduced the MDM2 inhibitory effect on p53 activity (Table 1 56% and 87% of the p53-induced growth inhibition and S-phase cell cycle arrest, respectively, were re-established (Table 1; Fig. 3A and B). Dose-response curves for the effects of 1-100 mM pyranoxanthone 1 on the reduction of the MDM2 negative effect on p53induced growth inhibition in yeast co-expressing p53 and MDM2 and on the inhibition of growth of control yeast were obtained (Fig. 3C). The concentration of 10 mM pyranoxanthone 1 was selected as the lower concentration for which a significant reduction of the negative effect of MDM2 on p53-induced growth inhibition was obtained without cytotoxic effects on control yeast. In opposition, xanthones 2-9 did not interfere with the negative effect of MDM2 on p53-induced yeast growth inhibition (Table 1; Fig. 3A; represented by xanthone 2) and cell cycle arrest ( Fig. 3B; represented by xanthone 2).
The effect of the xanthone derivatives on the negative regulation of p53-dependent transcriptional activity by MDM2 was also analyzed in yeast using a dual-luciferase p53 transactivation assay previously reported in [23]. This assay was carried out in yeast cells co-expressing p53 and MDM2, and used the p53 response element derived from the BBC3/PUMA gene fused to a luciferase reporter. The results showed that, likewise to 10 mM nutlin-3a, 10 mM pyranoxanthone 1 significantly increased the luciferase activity. This indicated a reversion of the MDM2 inhibitory effect on p53 transcriptional activity by pyranoxanthone 1 (Fig. 3D). As obtained in the yeast phenotypic assay, xanthones 2-9 did not interfere with the p53 transcriptional activity (Fig. 3D; represented by xanthone 2).
Altogether, the results obtained in yeast strongly supported that the pyranoxanthone 1 was a promising inhibitor of p53-MDM2 interaction.

Pyranoxanthone 1 reactivates the p53 activity and downstream cell signaling in human tumor cells
The molecular mechanism of action of pyranoxanthone 1 as inhibitor of p53-MDM2 interaction was further ascertained in human tumor cells harboring wt p53 and overexpressed MDM2, and was compared to that obtained with known small-molecule activators of p53 activity, 1.5 mM doxorubicin and 10 mM nutlin-3a. In order to analyze the effect of pyranoxanthone 1 on p53dependent transcriptional activity, a p53 reporter assay was performed in human tumor cells with wt p53, HCT116 p53 +/+ (Fig. 4A) and MCF-7 (Fig. 4C). In parallel, the activity of pyranoxanthone 1 was evaluated in p53 null HCT116 cells (HCT116 p53 À/À ; Fig. 4B) and MCF-7 cells transfected with an empty reporter vector, which served as negative controls (Fig. 4D). In this assay, the synthetic p53 response element pG13 fused to the luciferase reporter gene was used. The obtained results showed that, likewise doxorubicin (in HCT116 p53 +/+ ), and doxorubicin and nutlin-3a (in MCF-7), 10 mM pyranoxanthone 1 markedly increased the luciferase activity in HCT116 p53 +/+ (Fig. 4A) and MCF-7 (Fig. 4C) tumor cells. On the contrary, it did not interfere with the luciferase activity in p53-negative control cell lines exposed to the same conditions ( Fig. 4B and D). These results demonstrated that the pyranoxanthone 1 successfully activated the p53-dependent transcriptional activity in human tumor cell lines.
The effect of pyranoxanthone 1 on the stabilization of p53 protein levels was also analyzed in HCT116 tumor cells. As expected, similarly to positive controls, 10 mM pyranoxanthone 1 increased the p53 baseline levels upon 4, 8 and 16 h treatments in HCT116 p53 +/+ cells, suggesting an inhibition of the p53 degradation by MDM2 in these cells (Fig. 5A). However, contrarily to nutlin-3a, for the different tested time points, no increase on MDM2 protein levels was observed in these tumor cells after treatment with 10 mM pyranoxanthone 1 (Fig. 5B). In spite of this, likewise positive controls, 10 mM pyranoxanthone 1 increased the expression levels of other proteins encoded by p53 target genes, namely p21 (Fig. 5C) and Bax (Fig. 5D) after 16 and 8 h treatments, respectively, in HCT116 p53 +/+ but not in HCT116 p53 À/À tumor cells. Similarly, it enhanced the procaspase-7 cleavage to the active caspase-7 (LS) form at the 8 h treatment in HCT116 p53 +/+ but not in HCT116 p53 À/À tumor cells (Fig. 5E).
Altogether, these results confirmed the effectiveness of the pyranoxanthone 1 in human tumor cells with wt p53. In fact, the pyranoxanthone 1 activated the p53-dependent transcriptional activity, and increased the protein levels of p53, p21, Bax and cleaved caspase-7.
3.5. Analysis of the predicted binding model of pyranoxanthone 1 to MDM2 supports that pyranoxanthone 1 binds to MDM2 Based on the obtained results, a careful visual inspection of the pyranoxanthone 1 on the limited interaction site of MDM2 was performed by computational docking. As shown in Fig. 6A, the best fitting molecule (pyranoxanthone 1) adopted a pose within the p53-binding site, filling the space supposedly occupied by the p53 helix. Fig. 6B highlights the polar interaction established between pyranoxanthone 1 and MDM2. In contrast, nutlin-3a-MDM2 interaction revealed no hydrogen bonding (Fig. 6C). This obtained result is in accordance to [36], in which was verified that nutlin-3a binds to MDM2 mainly by hydrophobic interactions, where it is predicted to contact nine MDM2 amino acids, but all without hydrogen bonds.

Discussion
To search for new inhibitors of p53-MDM2 interaction several assays have been developed in the last years. However, most of them are actually quite expensive when applied to the screening of large libraries of compounds, limiting their use in drug discovery. In the present work, a simple, selective, and reliable yeast growthinhibition assay was developed to search for novel inhibitors of p53-MDM2 interaction. This assay can be easily adapted to the high-throughput screening of large chemical libraries based on simple measurements of the yeast cell growth in a cost-effective manner [reviewed in [32]]. In this study, the use of this yeast assay led us to the identification of a potential lead compound (pyranoxanthone 1), thus providing a proof of concept for its effectiveness in the discovery of new inhibitors of p53-MDM2 interaction. The yeast growth-inhibition assay proves to be greatly useful in the initial screening for a first selection of most promising compounds to be tested in more complex cell models.
In this work, using the virtual screening based on multiple binding modes, a set of putative MDM2 ligands with a xanthone scaffold was identified. Based on the generated ranking docking scores, 14 of the potential ligands, exhibiting in our previous studies antiproliferative activity against tumor cells with wt p53 [11,13,15], were selected. The activity of these 14 compounds as inhibitors of p53-MDM2 interaction was subsequently investigated using the developed yeast phenotypic screening assay followed by a yeast p53 transactivation assay [described in [23]]. Using this yeast assay, the pyranoxanthone 1 (3,4-dihydro-12-hydroxy-2,2-dimethyl-2H,6Hpyrano [3,2-b]xanthen-6-one) was identified as a promising inhibitor of p53-MDM2 interaction. Interestingly, the active hit identified in yeast also exhibited the highest binding affinity (À7.6 kcal mol À1 ) to the MDM2 local of interaction with p53 in the virtual screening. Nevertheless, a correlation between docking scores and activity was not observed and future studies would require considering not only the ligands, but also the target as flexible to improve the accuracy of the docking results.
To further ascertain whether the pyranoxanthone 1 had effects on tumor cells consistent with an inhibition of p53-MDM2 interaction, the activity of this compound was analyzed in human tumor-derived cell lines with wt p53 and overexpressed MDM2. In a recent work, pyranoxanthone 1 revealed a higher potency against tumor cells with wt p53 (MCF-7; GI 50 = 5.3 AE 0.7) than against tumor cells with mutant p53 (MDA-MB-231; GI 50 = 29.0 AE 3.6) [15]. Supporting previous results from our group [11,13,15], this work confirmed the effectiveness of pyranoxanthone 1 in tumor cells with wt p53. Most importantly, herein it is shown that, in conformity to what obtained in yeast, the pyranoxanthone 1 mimicked the activity of known smallmolecule activators of p53 activity in human tumor cells, leading to the successful activation of p53 and downstream cell signaling.
By computational docking studies it was possible to ascertain that similarly to nutlin-3a, pyranoxanthone 1 binds to MDM2, and therefore activating p53 through inhibition of MDM2 function in tumor cells. However, distinct types of interactions were detected for these two compounds. While interaction of pyranoxanthone 1 with MDM2 involves hydrogen interaction with Gly58, a residue critical for the interaction of MDM2 with p53, nutlin-3a-MDM2 interaction mainly involves hydrophobic interactions.
As a whole, in this work, pyranoxanthone 1 (3,4-dihydro-12hydroxy-2,2-dimethyl-2H,6H-pyrano [3,2-b]xanthen-6-one) was identified as a new inhibitor of p53-MDM2 interaction. Our finding thus adds, for the first time, the xanthone scaffold to the list of chemotypes of small-molecule inhibitors of p53-MDM2 interaction. In contrast to nutlin-3a, the pyranoxanthone 1 represents a very promising achiral compound with feasible synthesis and herein different binding modes with MDM2 were predicted. Besides its potential use as molecular probe and possible anticancer agent, pyranoxanthone 1 mainly represents a useful lead compound for the structure-based design of more potent drug-like analogs.

Table A1
Results of docking simulations for MDM2 (with grid box) performed in PyRx/AutoDockVina.

Table A2
Results of docking simulations for MDM2 (total protein) done in PyRx/AutoDockVina.