Structures of low molecular weight inhibitors bound to MDMX and ...

Report

Cell Cycle 9:5, 1104-1111; March 15, 2010; © 2010 Landes Bioscience

Structures of low molecular weight inhibitors bound to MDMX and MDM2 reveal new approaches for p53-MDMX/MDM2 antagonist drug discovery Grzegorz M. Popowicz,1 Anna Czarna,1 Siglinde Wolf,1 Kan Wang,2 Wei Wang,2 Alexander Dömling2 and Tad A. Holak1,* Max Planck Institute for Biochemistry; Martinsried, Germany; 2University of Pittsburgh; Departments of Pharmaceutical Sciences and Chemistry; Pittsburgh, PA USA

1

Key words: MDMX, MDM2, p53, structure, drug-design, cancer

Intensive anticancer drug discovery efforts have been made to develop small molecule inhibitors of the p53-MDM2 and p53-MDMX interactions. We present here the structures of the most potent inhibitors bound to MDM2 and MDMX that are based on the new imidazo-indole scaffold. In addition, the structure of the recently reported spiro-oxindole inhibitor bound to MDM2 is described. The structures indicate how the substituents of a small molecule that bind to the three subpockets of the MDM2/X-p53 interaction should be optimized for effective binding to MDM2 and/or MDMX. While the spiro-oxindole inhibitor triggers significant ligand-induced changes in MDM2, the imidazo-indoles share similar binding modes for MDMX and MDM2, but cause only minimal induced-fit changes in the structures of both proteins. Our study includes the first structure of the complex between MDMX and a small molecule and should aid in developing efficient scaffolds for binding to MDMX and/or MDM2.

Introduction The tumor suppressor p53 protein, “the guardian of the genome,” plays a central role in maintaining the integrity of the genome and is crucial for protecting the organism from cancer.1-3 In cancer cells the p53 tumor suppression is compromised either through mutations affecting the p53 locus directly (approximately 50% of all malignancies) or through aberration of its normal regulation.1-6 The latter group of tumors retains the wildtype p53, but the p53 pathway is mostly inactivated by its negative regulators, the MDM2 and MDMX proteins. MDM2 and MDMX bind directly to p53 and inhibit its transactivation activity.5,7 MDM2, in addition, induces p53 ubiquitination and proteasomal degradation.2,8,9 The restoration of the impaired function of a single gene, p53, by directly disrupting the MDM2-p53/MDMX-p53 interactions, offers an attractive new avenue for anticancer therapy across a broad spectrum of cancers.10-20 Cancer cells have been shown to be extremely sensitive to restoration of p53 function, verifying the expectation of highly effective therapies from this approach 21-24 and thus an intensive anticancer drug discovery effort is ongoing to develop antagonists of this interaction. While many MDM2 antagonists have been described in patent and scientific literature, no small molecule is known to tightly bind to MDMX.14,25-27 Recent data, suggesting a distinct and complementary mode of action of MDM2 and MDMX in

the regulation of the pro-apoptotic activity of p53, have raised the notion that the development of dual inhibitors of the two oncogenic proteins should result in more effective antitumor strategies.2,4,19,28 As of today, there are three classes of small molecule inhibitors of MDM2 that are able to disrupt MDM2-p53 binding with high (nM) affinity and specificity.15,29,30 These compounds are, however, only weak inhibitors of the MDMX-p53 interaction (Ki’s 30–70 µM; Table 1).25,31 The first and best-documented compound in the p53-MDM2 area is Nutlin-3. It is a cis-imidazolidine derivative and has been discovered during a HTS, followed by medicinal chemistry optimization of the initial hit.10,29 The second class of potent and highly selective inhibitors of the p53MDM2 interaction are derivatives of spiro-oxindoles.15,16,32 The best optimized derivatives of this spiro-oxindoles group, MI-219 and MI-63, bind to MDM2 >1,000-fold better than the p53 wildtype peptide (Ki value of 5–6 nM). Derivatives of MI-219/ MI-63 and Nutlin-3 showed the desired cellular downstream effects and have progressed to advanced preclinical development or early phase clinical trials.15 The third group of MDM2-p53 antagonists relies on a benzodiazepinedione core.30 The optimized compounds of this series could suppress the growth of wild-type p53 cells with IC50 in the 7–30 µM range and 3–9-fold selectivity for cells with functional p53 but low cellular potency and selectivity of these agents makes it difficult to evaluate their antitumoral potential.33

*Correspondence to: Tad A. Holak; Email: [email protected] Submitted: 12/14/09; Accepted: 12/15/09 Previously published online: www.landesbioscience.com/journals/cc/article/10956 1104

Cell Cycle

Volume 9 Issue 6

Report

Report

Table 1. Binding constants of the MDM2 and MDMX ligands used for structural studies; determined by the fluorescence polarization binding assay MDM2

*

Results

MDMX

IC50 [uM]

Ki [nM]

IC50 [uM]

Ki [uM]

Nutlin-3

0.27 ± 0.0483

69

221.8 ± 19.3

70

MI-63

0.18 ± 0.015

36

WK23

1.71 ± 1.1027

916

44.5 ± 1.9

36

WK298

0.19 ± 0.0768

109

19.7 ± 3.8

11

55*

Value given by Shangary et al.16 No measurable binding in our assay.

We and others have recently discovered that compounds based on the imidazo-indole scaffold can bind to MDM2 in the nM range.27,34-36 In addition, an optimized compound of Boettcher et al.36 (designated as WK298 or Novartis-101 hereafter) is the first antagonist of the MDMX-p53 interaction with the reported single digit µM Ki potency. No cell activity is available for these compounds yet. The binding of the wild-type p53 based peptides to MDM2 and MDMX is by now well established in structural biology.5,31,37,38 There are also recently structures of a number of highaffinity mutant peptides, which bind with high affinity to both MDMX as well as MDM2.20,39-42 However, there are to date only two crystallographic structures of small-molecule inhibitors in complex with MDM2 and none for the MDMX-p53.29,30 The binding mode of peptides and small molecules into the same protein-protein interface, however, is different with respect to topology and binding interactions and therefore it is essential to have structural information of the latter in order to be able to optimize small molecular weight compound series. The binding mode of the Nutlin has been elucidated in atomic detail (Nutlin-2, PDB: 1RV1)29 and the structure of the complex of a benzodiazepinedione inhibitor with MDM2 has been published (PDB: 1T4E).30 Despite of several models for spiro-oxindoles MI-219 and MI-63, there are no direct structural data on this family of compounds. Moreover, it was found that small differences in the structures of MDM2 and MDMX lead to dramatic differences in their affinities to small binding molecules.38,43 Here we describe three high-resolution structures of small molecule inhibitors bound to MDM2 or MDMX that are representative of the class of the spiro-oxindole inhibitors and those based on the 3-imidazoyl-indoles scaffold. Our experimental data demonstrate the exact positioning of each substituent of the scaffolds in the p53-binding sites of MDM2 and MDMX, needed to obtain high potency inhibitors. The structures explain the cause of the generally observed large differences in potencies against MDM2 and MDMX. A significant induced-fit adaptation of the MDM2 structure has been found upon binding of the MI-63-analog. A number of additional interactions were found for each complex that increase inhibitors’ affinities. The experimental data we report here should greatly increase the structural knowledge of MDM2 inhibitors and provide a first detailed insight into a small molecule interaction with MDMX. Our data should thus allow for the design of future high-potency, mono and dual-action inhibitors of MDM2 and MDMX and for

www.landesbioscience.com

the structure-based optimization of those that have been reported until now.

Binding of the 6-chloroindole-based ligands to MDM2 and MDMX. We have measured binding Ki constants of all small molecule ligands used in our study with the fluorescence polarization (FP) binding assay. Results are presented in Table 1, Figure 1, and for some additional compounds, for which the structures were not obtained, in the Supporting Table 1. All inhibitors have nanomolar affinity towards MDM2 and 100–1,000 lower affinity towards MDMX. Presumably all compounds were optimized for binding to MDM2 with disregard of MDMX. Nevertheless, the affinity of WK298 was considered as suitable for the MDMX complex crystallization. It should be noted that MI-63 and Nutlin-3 show much higher selectivity towards MDM2 than the WK23 and WK298 compounds (the chemical structures of the compounds are shown in Fig. 1). Clearly, the data suggest that WK23 could serve as the starting point for the optimization into the nanomolar binding towards MDMX. Structure of the MDMX protein with the WK298 inhibitor. The interfaces of p53-MDMX and p53-MDM2 rely on the steric complementarity between the MDMX/2 clefts and the hydrophobic face of the p53 α-helix31,37,38 and in particular, on a triad of p53 amino acids Phe19, Trp23 and Leu26, which deeply insert into the MDMX and MDM2 clefts (structural characteristics of the p53-MDM2 and p53-MDMX complexes are shown in Supporting Fig. 1 of the Supporting Information and the domain arrangements of p53, MDM2 and MDMX are shown in Supporting Fig. 2). Thus, these p53 amino acids comprise the “hot spot” of the p53-MDMX/2 interaction. Noteworthy is the cross section dimension of the p53 binding site in MDMX/2 (∼18 Å) that is the size of small organic molecules thus indicating the possibility of small molecule interference. The WK298 (Novartis-101) compound binds to MDMX in a way that mimics parts of the binding of the native p53 peptide (Figs. 2A and 3). The Trp23 pocket is filled with the 6-chloroindole substituent, whose NH forms a hydrogen bond with the carbonyl atom of Met53, remarkably similar to the native p53 residue. The plane of the 6-chloroindole is shifted by 0.63 Å towards Leu26 compared to the native Trp23 side-chain. The second key substituent in WK298, the 4-chlorobenzyl ring, penetrates the Leu26 pocket similarly to the 4-(4-chlorophenyl) substituent in the Nutlin-MDM2 complex. Finally the third principal MDMX interacting substituent of WK298, the phenyl ring, fills the Phe19 pocket; however, the plane of this phenyl ring is nearly perpendicular to the plane of the Phe19 ring of p53. The amide oxygen attached to the 2-position of the indole moiety forms a hydrogen bond (of 3.29 Å) to the His54 Nδ atom of MDMX. Interestingly, the N,N-dimethylpropylamine part of the inhibitor folds over the Phe19 pocket and shields from solvent the hydrophobic region formed by Met61 and Tyr66 of MDMX. Since WK298 has approximately a three times lower Ki constant to MDMX than WK23, which is devoid of this N,N-dimethylpropylamine side-chain, it turns

Cell Cycle

1105

Figure 1. MDM2 and MDMX inhibitors used for structural study. (A) WK23 and WK298 differ only by an additional amide moiety attached to the 2-position of the indole ring. MI-63-analog is a spiro-oxindole compound. (B) The compounds interact with both MDM2 and MDMX when tested by our fluorescence polarization assay. All compounds show much lower affinity towards MDMX. All molecules studied are based on the 6-chloroindole substructure.

out that this additional and unexpected interaction improves binding substantially (assuming that WK23 binds the same way to MDMX). Upon binding of the compound, the general fold of MDMX remains virtually unchanged when compared to the complex with p53 peptide.31,38 The MDMX binding pocket undergoes, however, substantial induced-fit changes to accommodate this non-native ligand. The presence of a large chlorine atom at the 6-position of the indole ring, that is absent in the native Trp23 of p53, causes the MDMX surrounding residues to retract from their positions creating sufficient space for accommodating the chlorine. This effect is visible in the whole Trp23 pocket and propagates into the Leu26 pocket. It is most pronounced at the Leu56 side-chain (the Trp23 pocket), which is retracted by 0.81 Å relative to the native p53-MDMX complex of Popowicz et al.38 The other residues affected are Met53 (0.71 Å), Leu98 (0.69 Å) and Leu102 (1.14 Å). Moreover, the main chains and the whole helices forming this region are noticeably shifted to enlarge the binding pocket: the α1 helix is shifted by 0.56 Å at the Met53 Cα atom and the α2' helix 1.03 Å at Leu98. A noticeable displacement of the α2' helix is also forced by the 4-chlorobenzyl group in the Leu26 pocket. To make room suitable to accommodate the

1106

4-chlorobenzyl substituent, the Cα of Tyr99 is shifted by 1.27 Å and its side-chain is moved outside the binding pocket by 5.34 Å at phenol-O atoms adopting only partially an “open” conformation, as defined by31,38 but still points inwards the binding pocket (Fig. 3). Taken together, the above described changes resemble the behavior of MDMX seen in the structures with the bound 6-chloro-tryptophan-based peptide reported by Kallen et al. (PDB: 2FEA).39 A major difference arises at the end of helix α2', where residues Pro95-Tyr99 of the MDMX-WK298 complex are closer to the native p53 complex (1 Å at Pro95 and 1.27 Å at Tyr99 Cα) than the structure of the peptidomimetic inhibitor (1.92 Å at Pro95 and 1.38 Å at Tyr99 Cα). Strikingly, the Tyr99 side-chain is flipped outside the binding pocket (the “open” conformation) in the 2FEA structure but remains in an intermediate, “open/closed” conformation in our structure. This contradicts the theory of the cross-talk between Trp23 and Leu26 pockets,39 as the presence of the 6-chloroindole ring in the Trp23 pocket, and additionally, a bigger than the Leu26, of the 4-chlorobenzyl moiety in the Leu26 pocket for WK298, do not cause the “open” conformation of Tyr99 nor excessive conformational changes in the α2' region of the Leu26 subpocket.

Cell Cycle

Volume 9 Issue 6

Figure 2. Inhibitors bound to MDM2 and MDMX proteins. The inhibitors are shown in stick models with carbon atoms colored yellow, nitrogen blue and oxygen red. Key side-chains of the proteins are labeled. Hydrogen bonds are depicted as yellow dashed lines. The numbering of substituents in WK23 and WK298 is provided in the Supporting Information. (A) Compound WK298 binds to MDMX by filling its Trp23 subpocket with the 6-chloroindole group. The 4-phenyl group is located in the Phe19 and 1-(4-chlorobenzyl) group in the Leu26 pockets, respectively. Two hydrogen bonds to Met53 and His54 are formed. The N,N-dimethylpropylamine part of the WK298 molecule folds over Gly57 and Met61, forming additional hydrophobic protection of the binding cleft. Tyr99 closes the Leu26 subpocket. (B) Similarly, the panel shows inhibitor WK23 bound to the MDM2 protein. A very similar binding pose as in (A) is formed despite of different shapes of the p53 binding sites in MDM2 and MDMX. The His96-Tyr100 region has the most pronounced differences in the shape of the Leu26 pocket, but the position of 1-(4-chlorobenzyl) is not altered. A hydrogen bond is formed between the indole of WK23 and the Leu54 carbonyl oxygen of MDM2. (C) The MI-63-analog inhibitor binds to MDM2 also by nesting the chlorophenyl substructure of the 6-chlorooxindole into the Trp23 subpocket. The Leu26 subpocket is filled by the 2-fluoro-3-chlorophenyl ring. This ring is located as in WK298 and WK23 (A and B) but its plane is rotated to allow the phenyl substituent atoms to fill the bottom of the MDM2 pocket. The neopentyl fragment fills the Phe19 pocket and is a cause of a substantial induced-fit reshaping of the binding cleft. The Tyr67 side-chain is rotated to form a much steeper wall “closing” the binding region. The whole Tyr67-His73 region acquires a different fold to allow the Tyr67 movement. The compound forms two hydrogen bonds with Leu54 and His96. The ethyl-morpholino part of the compound is not taking part in the binding and is not seen in the electron density.

It is clear that those large differences between the structures of the WK298/MDMX complex and that of the native p53-MDMX are energetically unfavorable and lead to the weaker binding of WK298 to MDMX than to MDM2. The chloroindole-based peptide can achieve much higher affinity towards MDMX than the WK298 compound (and cause much smaller disturbance in the structure of the native MDMX conformation). Since both the 2FEA structure and our complex contain chloroindole as a main entity for filling the Tpr23 pocket, it could be expected that also the non-peptidic 6-chloroindole-containing compound may achieve high affinity towards MDMX; the weaker binding of WK298 to MDMX is due to its non-optimal occupation of the Phe19 and Leu26 pockets by chlorophenyl and phenyl rings, respectively. Structure of MDM2 with the WK23 inhibitor. WK23 represents the smallest, ∼1 µM-binding inhibitor of the MDM2-p53 interaction described so far (Table 1 and Fig. 1). Despite of its low molecular weight (462 Dalton) and only four aromatic groups, it is able to efficiently fill the whole p53 binding pocket. With an exception of one hydrogen bond, between Leu54 and the chloroindole nitrogen, all interactions have purely hydrophobic character. The structure of MDM2 with the bound WK23 inhibitor appears to be barely different from the native p53-MDM2 model (PDB: 1YCR)37 (Figs. 2B and 4). With the exception of solventexposed residues located far from the binding site, there are no major differences arising upon the binding of this inhibitor. There are, however, a few side-chains that differ from the Nutlin-MDM2 structure: The Tyr100 side-chain is situated in the “closed” conformation, nearly identical to that seen in the Nutlin-2-MDM2


Cell Cycle

1107

Figure 3. The MDMX inhibitor WK298 compared to the native p53 binding. The inhibitor is shown with yellow carbon atoms. MDMX bound to the native p53 peptide is presented as a dark blue ribbon plot with most important residues shown in stick model. MDMX bound to the WK298 inhibitor is similarly shown in light grey-blue. For readability purposes only side-chains of Phe19, Trp23 and Leu26 of p53 are shown (green, labeled in italics). The Trp23 side-chain and the 6-chloroindole part of the inhibitor are bound in a nearly identical position both forming hydrogen bonds with the Met53 carbonyl oxygen. The Trp23 subpocket undergoes induced-fit changes to accommodate the larger 6-chloroindole group. Leu56, Leu98 and Leu102 are retracted from the native peptide complex position to make space for an inhibitor. The position of the Tyr99 side-chain is altered but remains in a “closed” orientation. An additional hydrogen bond is formed between the inhibitor and His54.

structure (PDB: 1RV1),29 but with the plane of the phenyl ring rotated by approx. 75°. Additionally, the Met62 side-chain points outside the binding pocket, resembling the configuration seen in the Nutlin-MDM2 structure, while in the p53-peptide/MDM2 model it directs itself towards the binding pocket. Ile99 and Leu57 in the native p53-MDM2 structure are located 0.76 Å and 1.03 Å, respectively, outwards the centre of the binding pocket compared to those in the MDM2-WK23 and Nutlin-MDM2 structures. These residues have identical positions and conformations in both the MDM2-WK23 and Nutlin-MDM2 structures. On the other hand, the MDM2 region that surrounds the Phe19 pocket, from Tyr67 to Gln72, shows significant divergence of both main- and side-chains compared the structure of Nutlin-MDM2. It is difficult, however, to ascertain whether this is caused by protein-ligand interactions or crystal contacts. The 6-chloroindole substituent of WK23 is situated in the identical position as the side-chain of the native p53 tryptophan with the chlorine atom penetrating the very bottom of the binding pocket. Position of this chlorine atom is only 0.74 Å apart from the position of the Nutlin-2 bromine atom. The nitrogen of the

1108

Figure 4. MDMX-WK298 and MDM2-WK23 bind in an identical mode. WK298 bound to MDMX is shown with yellow carbon atoms; WK23 carbon atoms, in salmon. MDM2 is shown as a red ribbon plot with important residues shown as stick model. MDMX is shown similarly in light blue. Important residues are labelled in italics, if different from MDMX, or their numbers given in parentheses, if of the same type as in MDMX. Despite significant structural differences between MDM2 and MDMX, especially in the Pro95-Tyr99 region of MDMX and the corresponding His96-Tyr100 of MDM2, the ligands are bound in a nearly identical way. Only the position of the chlorophenyl substituent in the Leu26 pocket is shifted between the two structures to adjust to different proteins environment. Clearly, weaker binding of the compounds towards MDMX compared to MDM2 is caused by the sub-optimal interaction with the Leu26 subpocket.

6-chloroindole group forms a hydrogen bond to the Leu54 carbonyl oxygen, also seen in the native p53 peptide. The Leu26 pocket is filled with the 4-chlorobenzyl substituent, with the chlorine atom positioned 2.16 Å towards Tyr100 than the corresponding Cγ of the native Leu26. Its position is also very similar to the bromophenyl ring of Nutlin-2, with the distance between the Cl and Br atoms of 1.06 Å and the angle between aromatic ring planes being 13°. In general, the Leu26 pocket is filled more completely in the WK23-MDM2 and to a greater depth by the 4-chlorobenzyl substituent than by the aliphatic Leu26 side-chain of p53, with only minor induced-fit changes in the MDM2 structure. The Phe19 pocket is filled by the phenyl ring of the WK23 compound, the centre of the WK23 phenyl ring is located close to the centre of the p53 Phe19 ring, but its plane is oriented perpendicular to it. The Phe19 ring is also able to submerge deeper into the binding pocket by approx 2 Å. It is likely that in the case of the Phe19 pocket a large area of hydrophobic interactions is required rather than filling of the cavity into its depth. Nutlin-2, WK23, as well as the benzodiazepinedione-based inhibitor (PDB: 1T4E),30 occupy most of the Phe19 pocket aperture, while not penetrating into its very bottom. The aromatic imidazole

Cell Cycle

Volume 9 Issue 6

Figure 5. The MI-63-analog compound causes a significant induced fit readjustment of the MDM2 structure. MDM2 structures are shown as ribbon plots with important residues visualized as stick models, MDM2 bound to WK23 is shown in red and bound to MI-63-analog in green (ethyl-morpholino part as not defined in the electron density is not shown). MI-63-analog is drawn as stick model with yellow carbon atoms while WK23 has carbon atoms colored dark teal. While the Trp23 and Leu26 subpockets are filled in similar fashion, the neopentyl moiety of MI-63-analog requires readjustment of the Tyr67-His73 region. The Tyr67 side-chain is rotated to fill remaining space in the Phe19 subpocket. To allow its movement Gln73 and, to lesser extent His73, are moved away from the binding site.

ring, whose primary purpose in the WK23 molecule is to provide a scaffold to which all substituents are attached, provides also an additional hydrophobic contact with Val93 of MDM2. The center of the non-aromatic heterocyclic scaffold of Nutlin-2, the imidazoline, is in the same position as the WK23 central scaffold, however the planes of the two scaffolds are tilled and the WK23 imidazole is shifted only 1.03 Å towards the α2 helix. This behavior can be explained by the different nature of the two central scaffolds of WK23 and Nutlin. Whereas the WK23 imidazole is aromatic and the three substituents are in plane, the imidazoline of Nutlin is non aromatic and the 4- and 5-substituents exit the scaffold in a cis-configuration. Thus, the Nutlin 4and 5-substituents have the optimal angle to penetrate the Phe19 and Leu26 pockets, whereas the WK23 molecule can only penetrate deeply into the Leu26 pocket with a benzylic methylene group used to accomplish the optimal penetration angle. Additionally, there is a water molecule trapped in the network of hydrogen bonds that involves the carboxylic group at position 2 of the indole ring of the inhibitor and side-chains of MDMX (the Leu54 carbonyl oxygen and the side-chain of Gln59). In a similar fashion, an amide oxygen of the benzodiazepindione inhibitor resides at almost the same position as the carboxyl-oxygen of WK23.


In summary, WK23 represents an extremely efficient, featurerich inhibitor of the MDM2-p53 interaction. It does not cause excessive induced fit changes in MDM2, and should allow for additional modifications to improve its pharmacological properties, like, for example solubility. Structure of MDM2 with the MI-63-analog inhibitor. The complex of MDM2 with the MI-63-analog inhibitor gave crystals with the asymmetric unit that contained three separate MI-63analog-MDM2 complexes. Despite different crystal contacts and environments, the three complexes show no differences in structures with exception of the extreme N-terminus, the Tyr100 side-chain (in proximity of it), and the Glu69-Gln71 loop. The MDM2 fold is very similar to the native one, with the mainchain RMSD of 0.74 Å. The MI-63-analog molecule is found to bind to MDM2 in a nearly the exact mirror image of the mode of the binding suggested by Shangary et al.16 (Figs. 2C and 5). This might be due the fact that their studies focused on a different diastereoisomer than that in our crystal. The attempts to crystalize other diastereoisomers were unsuccesful. Interestingly, the measured affinity of “our” diastereoisomer is equal to the one reported in Shangary et al.16 The 6-chlorooxindole ring of MI-63-analog is, as expected, located inside the Trp23 pocket. It forms a 2.87 Å hydrogen bond with the Leu54 carbonyl oxygen. When compared to the native peptide or WK23 compound, this substituent’s plane is rotated by approx 10° so that the five-member ring is closer to the Leu26 pocket. The 6-chlorooxindole is also descending 0.69 Å towards the bottom of the pocket than the 6-chloroindole substituent in WK23. Surprisingly, this does not cause any induced fit changes of the MDM2 residues at the bottom of the cleft. The Leu26 pocket is filled by a 2-fluoro-3-chlorophenyl ring with the chlorine atom located similarly to the bromine of Nutlin-2 and the chlorine of WK23. This MI-63-analog chlorine atom is also located 0.52 Å closer to the α2' helix (which forms part of the pocket) than in its corresponding atoms in WK23, Nutlin-2, and the benzodiazepinedione, respectively. Again, no significant induced fit changes can be found. Tyr100 remains in an “open” conformation, this residue’s side-chain, however, has different conformations in each molecule present in the asymmetric unit. This indicates that the side-chain of Tyr100 in the MDM2inhibitor complex is prone to increased flexibility and does not form a rigid binding pocket boundary. The Phe19 subpocket is filled by the neopentyl group of MI-63-analog. This group occupies approximately the same space as the phenyl ring of WK23 and reaches approximately the same depth of the pocket as the Phe19 ring. This leaves some unoccupied space in the Phe19 pocket free, which is efficiently filled by the Tyr67 side-chain of MDM2. To achieve this, the entire Tyr67 main-chain is rotated and shifted by 0.83 Å towards the inside of the binding pocket and its side-chain rotated into the binding pocket, whereas in the native complex it points outwards. While the whole Glu69Gln71 region shows significant differences among chains present in the asymmetric unit, the position of Tyr67 remains identical among the three structures. To allow the Tyr67 ring to enter the binding site, the main chain of His73 is retracted by 2.02 Å (at Cα position) outward from the binding pocket. Altogether, these

Cell Cycle

1109

changes extensively reshape the Phe19 pocket, forming a “massive” induced fit rearrangement (Figs. 2C and 5). The pyrrolidine ring of MI-63-analog extends the hydrophobic interaction over Val93 in a way similar to the imidazole ring of the WK23 inhibitor. The amide group attached to it forms a short (2.23 Å) hydrogen bond between its carbonyl oxygen and His96 Nε. This hydrogen bond, although located in a solvent accessible region, is likely to be beneficial to the binding energy. It should be noted that the 6-chlorooxindole substituent is rigidly tied-up to the pyrrolidine core by a spiro connection, while compounds WK23 and WK298 have a rotatable bond between their chloroindole and imidazole rings. Considering a different shape of the p53 binding cleft of MDMX, the lack of this degree of freedom may disable the adjustment of the ligand to the different shape of the MDMX pocket, which may be necessary for the efficient binding to MDMX. On the other hand these reduced degrees of rotational freedom might also contribute the very high affinity of MI-63-analog to MDM2 by entropy reduction. The remaining part of the MI-63-analog molecule, including the morpholine ring and the aliphatic fragment attached to the amide group, is not visible in the electron density map. This indicates high flexibility of this segment of the MI-63-analog molecule and the lack of any direct interaction with MDM2. Clearly this substituent can be used to improve other drug-required properties. In summary MI-63-analog fulfils every requirement of a potent MDM2 inhibitor. It also causes significant ligand-binding induced changes in the shape of the Phe19 pocket. This and the rigid connection between the chloroindole and pyrrolidine rings may well explain its high specificity in binding to MDM2 and a dramatic loss of its potency towards MDMX, as neither the protein nor the ligand seem to be able to undergo necessary structural rearrangements for effective binding to MDMX. Imidazoyl-indole MDMX-p53 and MDM2-p53 inhibitors share similar binding modes. The MDM2 protein was crystallized with the WK23 inhibitor and MDMX with WK298. The two inhibitors differ by the presence of the ((3-(dimethylamino) propyl)(methyl)amino)pyrrolidin-1-yl tag in the WK298 compound. The portions of the two compounds that are responsible for the primary interaction inside the p53 binding pockets are identical (Figs. 2A and B, 3 and 4). Interestingly, despite of the differences between MDM2 and MDMX pockets (RMSD for main chain atoms 1.22 Å) seen in the native and inhibitor bound structures, the modes of the binding of those inhibitors to MDM2 and MDMX, remain virtually identical. The plane of the chloroindole ring in the MDMX structure, when viewed along the imidazole-chloroindole connecting bond, is rotated by only 14.8° relative to that in the MDM2 structure. The 4-chlorobenzyl substituents of these two compounds fill the Leu26 subpockets (for MDMX the 4-chlorobenzyl group is located 0.53 Å towards the α2' helix compared to that in MDM2). The 6-chloroindole ring, substituting the native tryptophan p53 side-chain, appears to be an efficient, interacting scaffold well suited for both proteins. While it requires a substantial, induced-fit adaptation of MDMX, it can effectively contribute to the MDMX binding by overcoming the energy cost of the protein

1110

adaptation, as is also shown by the even higher affinity of the 6-chloro-Trp peptide inhibitor.39 The Phe19 binding subpocket of both MDM2 and MDMX is similar, with the biggest differences located at the rim of the p53-binding cleft. As it was stated above, an effective inhibitor does not need to penetrate this space to the very bottom. Rather, it seems, that an enlarged hydrophobic surface is required for optimal interaction. Therefore, it should suffice to fill this subpocket with a nonpolar, aliphatic, but preferably aromatic moiety, which is universal to both proteins’ pockets. The Leu26 pocket seems to present the major challenge for developing a dual action, and potent MDM2 and MDMX inhibitor. The major differences, clustered around the α2' helix, are most probably the cause for much weaker binding of the known MDM2 inhibitors to MDMX. Apparently, the chlorophenyl group, present in all known MDM2 inhibitors so far, is not optimal for the p53-binding pocket of MDMX. While position of Tyr100 in MDM2 varies greatly from structure to structure, the equivalent Tyr99 of MDMX remains in most cases in the “closed” conformation and this configuration must be accounted for as an important factor when optimizing the inhibitors towards MDMX. Certainly, this subpocket still requires the optimization of binding modules to obtain the nanomolar small molecular weight MDMX antagonists. Discussion We have solved the first structure of MDMX bound to a small molecule inhibitor. A closely similar molecule was also crystallized with MDM2 to analyze structural differences. Additionally, the structure of an analog of the recently published MI-63 inhibitor was solved to characterize binding properties of this and related spirooxindolic compounds. As expected, the occupancy of the three subpockets of Phe19, Trp23 and Leu26 appear to be a necessary and sufficient condition to achieve good inhibitor binding. It is clear that the 6-chloroindole group in the Trp23 subpocket is well suited to bind both MDM2 and MDMX. Additionally, the Phe19 subpocket has to be filled with large-surface hydrophobic entity. The Leu26 subpocket poses the greatest challenge for the design of dual MDM2, MDMX inhibitors, as MDM2 and MDMX differ significantly in the shape and properties of residues surrounding this pocket. Nevertheless, our structures should allow for rational development of the known MDM2-p53 inhibitory scaffolds towards greater affinity to MDMX. Materials and Methods Ligand synthesis. MI-63-analogs were a generous gift from Dr. Bruno Schoentjes (Ortho-Biotech Oncology Research & Development, 2340 Beerse-Belgium). The syntheses of the remainder of the compounds used in this study are described in the Supplemental Information. Protein production, purification, crystallography and fluorescence experiments. Protein production and purification was performed as described previously.43 Detailed structure solution procedure and fluorescence polarization assay methodology are described in Supplemental Information.

Cell Cycle

Volume 9 Issue 6

Note

Acknowledgements

Supplementary materials can be found at: www.landesbioscience.com/supplement/PopowiczCC9-6-Sup. pdf

This work was supported the Deutsche Krebshilfe, Grant 108354 and the RAND program (NIH). We thank Thorsten Berg for his kind help and expertise with the FP assay.

References 1. 2.

3. 4. 5. 6.

7.

8.

9. 10. 11.

12.

13.

14.

15.

Vogelstein B, Lane D, Levine AJ. Surfing the p53 network. Nature 2000; 408:307-10. Toledo F, Wahl GM. Regulating the p53 pathway: in vitro hypotheses, in vivo veritas. Nat Rev Cancer 2006; 6:909-23. Vousden KH, Lane DP. p53 in health and disease. Nat Rev Mol Cell Biot 2006; 8:275-83. Marine JCW, Dyer MA, Jochemsen AG. MDMX: from bench to bedside. J Cell Sci 2007; 120:371-8. Joerger AC, Fersht AR. Structural biology of the tumor suppressor p53. Ann Rev Bioch 2008; 77:557-82. Vazquez A, Bond EE, Levine AJ, Bond GL. The genetics of the p53 pathway, apoptosis and cancer therapy. Nat Rev Drug Disc 2008; 7:979-87. Momand J, Zambetti GP, Olson DC, George D, Levine AJ. The mdm-2 oncogene product forms a complex with the p53 protein and inhibits p53-mediated transactivation. Cell 1992; 69:1237-45. Haupt Y, Maya R, Kazaz A, Oren M. Mdm2 promotes the rapid degradation of p53. Nature 1997; 387:2969. Kubbutat MH, Jones SN, Vousden KH. Regulation of p53 stability by Mdm2. Nature 1997; 387:299-303. Vassilev LT. MDM2 inhibitors for cancer therapy. Trends Mol Med 2007; 13:23-31. Murray JK, Gellman SH. Targeting protein-protein interactions: Lessons from p53/MDM2. Biopolymers 2007; 88:657-86. Dömling A. Small molecular weight protein-protein interaction antagonists—an insurmountable challenge? Curr Opin Chem Biol 2008; 12:281-91. Bixby D, Kujawski L, Wang S, Sami N, Malek SN. The pre-clinical development of MDM2 inhibitors in chronic lymphocytic leukemia uncovers a central role for p53 status in sensitivity to MDM2 inhibitormediated apoptosis. Cell Cycle 2008; 7:971-9. Cheok CF, Lane DP. New developments in small molecules targeting p53 pathways in anticancer therapy. Drug Dev Res 2008; 69:289-96. Shangary S, Wang SM. Small-molecule inhibitors of the MDM2-p53 protein-protein interaction to reactivate p53 function: a novel approach for cancer therapy. Annu Rev Pharmacol Toxicol 2008; 49:223-41.


16. Shangary S, et al. Temporal activation of p53 by a specific MDM2 inhibitor is selectively toxic to tumors and leads to complete tumor growth inhibition. Proc Natl Acad Sci USA 2008; 105:3933-8. 17. Meulmeester E, Jochemsen AG. p53: A guide to apoptosis. Curr Cancer Drug Targ 2008; 8:87-97. 18. Lehman JA, Eitel JA, Batuello CN, Mayo LD. Therapeutic considerations for Mdm2: not just a one trick pony. Expert Opin Drug Dicov 2008; 3:130921. 19. Wade M, Wahl GM. Targeting Mdm2 and Mdmx in cancer therapy: better living through medicinal chemistry? Mol Cancer Res 2009; 7:1-11. 20. Macchiarulo A, Pellicciari R. MDM2/MDMX inhibitor peptide: WO2008106507. Expert Opin Therap Pat 2009; 19:721-6. 21. Marx J. Recruiting the cell’s own guardian for cancer therapy. Science 2007; 315:1211-3. 22. Martins CP, Brown-Swigart L, Evan GI. Modeling the therapeutic efficacy of p53 restoration in tumors. Cell 2006; 127:1323-34. 23. Ventura A, et al. Restoration of p53 function leads to tumour regression in vivo. Nature 2007; 445:661-5. 24. Xue W, et al. Senescence and tumour clearance is triggered by p53 restoration in murine liver carcinomas. Nature 2007; 445:656-60. 25. Laurie NA, et al. Inactivation of the p53 pathway in retinoblastoma. Nature 2006; 444:61-6. 26. Wade M, Wong ET, Tang M, Stommel JM, Wahl GM. Hdmx modulates the outcome of p53 activation in human tumor cells. J Biol Chem 2006; 281:3303644. 27. Doemling A. Selective and dual-action p53/mdm2/ mdm4 antagonists. WO 2008130614. 28. Bassett EA, Wang W, Rastinejad F, El-Deiry WS. Structural and functional basis for therapeutic modulation of p53 signaling. Clin Cancer Res 2008; 14:637686. 29. Vassilev LT, et al. In vivo activation of the p53 pathway by small-molecular antagonists of Mdm2. Science 2004; 303:844-8. 30. Grasberger BL, et al. Discovery and cocrystal structure of benzodiazepinedione HDM2 antagonists that activate p53 in cells. J Med Chem 2005; 48:909-12.

Cell Cycle

31. Popowicz GM, et al. Molecular basis for the inhibition of p53 by Mdmx. Cell Cycle 2007; 6:2386-92. 32. Shangary S, Wang SM. Targeting the MDM2-p53 interaction for cancer therapy. Clin Cancer Res 2008; 14:5318-24. 33. Koblish HK, et al. Benzodiazepinedione inhibitors of the Hdm2: p53 complex suppress human tumor cell proliferation in vitro and sensitize tumors to doxorubicin in vivo. Mol Cancer Therap 2006; 5:160-9. 34. Dömling A, Beck B. Preparation of pyrrolylimidazoles as antibiotics and antitumor agents. WO 2001025213. 35. Beck B, Leppert CA, Mueller BK, Dömling A. Discovery of pyrroloimidazoles as agents stimulating neurite outgrowth. QSAR Combin Sci 2006; 25:52735. 36. Boettcher A, et al. 3-Imidazoyl-indoles for the treatment of proliferative diseases. WO 2008119741. 37. Kussie PH, et al. Structure of the MDM2 oncoprotein bound to the p53 tumor suppressor transactivation domain. Science 1996; 274:948-53. 38. Popowicz GM, Czarna A, Holak TA. Structure of the human Mdmx protein bound to the p53 tumor suppressor transactivation domain. Cell Cycle 2008; 7:2441-3. 39. Kallen J, et al. Crystal structures of human MdmX HdmX; in complex with p53 peptide-analogues reveal surprising conformational changes. J Biol Chem 2009; 284:8812-21. 40. Hu B, Gilkes DM, Chen J. Efficient p53 activation and apoptosis by simultaneous disruption of binding to MDM2 and MDMX. Cancer Res 2006; 67:8810-7. 41. Pazgier M, et al. Structural basis for high-affinity peptide inhibition of p53 interactions with MDM2 and MDMX. Proc Natl Acad Sci USA 2009; 106:466570. 42. Zondlo SC, Lee AE, Zondlo NJ. Determinants of specificity of MDM2 for the activation domains of p53 and p65: proline27 disrupts the MDM2-binding motif of p53. Biochemistry 2006; 45:11945-57. 43. Czarna A, et al. High affinity interaction of the p53 peptide-analogue with human Mdm2 and Mdmx. Cell Cycle 2009; 8:1176-84.

1111