Computer-Aided Identification and Lead Optimization of Dual Murine Double Minute 2 and 4 Binders: Structure-Activity Relationship Studies and Pharmacological Activity

Article abstract of DOI:10.1021/acs.jmedchem.7b00912

The function of p53 protein, also known as "genome guardian", might be impaired by the overexpression of its primary cellular inhibitor, the murine double minute 2 protein (MDM2). However, the recent finding that MDM2-selective inhibitors induce high leve

Full text of DOI:10.1021/acs.jmedchem.7b00912

Subscriber access provided by UNIV OF CAMBRIDGE
Article
Computer-aided Identification and Lead Optimization of
Dual Murine Double Minute 2 and 4 Binders: Structure
Activity Relationship Studies, and Pharmacological Activity.
Mariateresa Giustiniano, Simona Daniele, Sveva Pelliccia, Valeria La Pietra, Deborah
Pietrobono, Diego Brancaccio, Sandro Cosconati, Anna Messere, Stefano Giuntini, Linda
Cerofolini, Marco Fragai, Claudio Luchinat, Sabrina Taliani, Giuseppe La Regina, Federico
Da Settimo, Romano Silvestri, Claudia Martini, Ettore Novellino, and Luciana Marinelli
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b00912 • Publication Date (Web): 18 Sep 2017
Downloaded from http://pubs.acs.org on September 19, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Computerꢀaided Identification and Lead Optimization
of Dual Murine Double Minute 2 and 4 Binders:
Structure Activity Relationship Studies, and
Pharmacological Activity.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
§
2§
1
1
Mariateresa Giustiniano, Simona Daniele, Sveva Pelliccia, Valeria La Pietra, * Deborah
2
1
5
5
3,4
Pietrobono, Diego Brancaccio, Sandro Cosconati, Anna Messere, Stefano Giuntini, Linda
3
3,4
3,4
2
6
Cerofolini, Marco Fragai, Claudio Luchinat, Sabrina Taliani, Giuseppe La Regina, Federico
2
6
2
1
1
Da Settimo, Romano Silvestri, Claudia Martini, * Ettore Novellino, Luciana Marinelli *
1
Dipartimento di Farmacia, Università degli Studi di Napoli Federico II, via D. Montesano 49,
2
3
8
0131, Napoli, Italy. Dipartimento di Farmacia, Università di Pisa, Italy. Laboratory of Food
3
Chemistry, Dipartimento di Agraria (QuaSic.A.Tec.), Magnetic Resonance Center (CERM)
4
University of Florence, Via L. Sacconi 6, 50019 Sesto Fiorentino (FI), Italy. Department of
Chemistry "Ugo Schiff", University of Florence, Via della Lastruccia 3ꢀ13, 50019 Sesto Fiorentino
5
6
(
FI), Italy. DiSTABiF, Second University of Naples, Caserta 81100, Italy. Dipartimento di
Chimica e Tecnologie del Farmaco, Università La Sapienza, Piazzale Aldo Moro 5, 00185 Roma.
KEYWORDS: MDM2, MDM4, dual inhibitors, proteinꢀprotein interaction, Glioblastoma
Multiforme, Molecular Docking, Virtual Screening.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT
The function of p53 protein, also known as “genome guardian”, might be impaired by the overꢀ
expression of its primary cellular inhibitor, the murine double minute 2 protein (MDM2). However,
the recent finding that MDM2ꢀselective inhibitors induce high levels of its homolog MDM4,
prompt us to identify, through a receptorꢀbased virtual screening on an in house dababase, dual
MDM2/MDM4 binders. Compound 1 turned out to possess an IC of 93.7 nM and of 4.6 nM on
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
50
MDM2 and MDM4 respectively. A series of compounds were synthetized to optimize its activity
on MDM2. As a result, compound 12 showed low nanomolar IC50 for both targets. NMR studies
confirmed the pocket of binding of 12 as predicted by the Glide docking software. Notably, 12 was
able to cause concentrationꢀdependent inhibition of cell proliferation, yielding an IC50 value of 356
±
21 nM in neuroblastoma SHSY5Y cells, and proved even to efficiently block CSC growth.
INTRODUCTION
The tumor suppressor protein p53, commonly referred as “guardian of the genome”, is one of the
most widely studied regulators of cell fate when the integrity of the genome is damaged. It is
generally activated in response to stress signals such as oncogene activation or hypoxia, and triggers
the expression of downstream targets thus inducing cellular apoptosis, senescence, and cell cycle
1
arrest, in order to block the proliferation of damaged cells. The importance of p53 integrity in
cancer genesis has been extremely clear when it has been shown that the TP53 gene is mutated or
deleted in nearly 50% of all human tumors and that p53ꢀnull mice are prone to develop spontaneous
2
cancers. Although p53 retains wildꢀtype status in the remaining 50% of human malignancies, its
function might be impaired by multiple mechanisms, where the main one is mediated by the
3
overexpression of its primary cellular inhibitor, the murine double minute 2 protein (MDM2). In
fact, in physiological conditions, MDM2 and p53 regulate each other in a feedback mechanism
2
ACS Paragon Plus Environment

Page 3 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
where, upon activation, p53 transcribes MDM2 gene, whose protein binds to p53, thus hampering
p53 binding to targeted DNA. Additionally, MDM2 causes p53 proteasomal degradation through its
E3 ubiquitin ligase activity, and promotes the p53 export out of the cell nucleus, where it will be no
4
longer accessible to target DNA. Deregulation of this autoꢀregulatory loop can provoke malignant
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
transformation of normal cells. In fact, overexpression of MDM2 can lead to neutralization of p53
tumor suppressor function. Accordingly, MDM2 gene amplification has been found in at least 19
tumor types as soft tissue tumors (20%), osteosarcomas (16%), esophageal carcinomas (13%), and
neuroꢀblastomas, where overexpression of MDM2 correlates with poor clinical prognosis and poor
5
treatment response to current cancer therapies. Thus, in this scenario, three possible anticancer
therapeutic approaches are viable: a) the targeting of mutated p53 through drugs that, acting as
6
chaperones of mutant p53, restore its function b) the gene therapy and c) the inhibition of MDM2,
7
which has been the most explored route in last decades. In this regard, the availability of Xꢀray
8
crystallographic structural complexes of MDM2 bound to p53 and of MDM2 in complex with p53ꢀ
mimicking peptides or nonpeptidic molecules laid the foundation for the design of highꢀaffinity
9
ꢀ18
MDM2 binders. Notwithstanding, these substantial advances made in the field, Wahl G. M. et al.
1
9
highlighted how MDM2ꢀselective inhibitors may induce high levels of MDM4, thus affecting the
20
clinical response of these drugs. In fact MDM4, also referred to as MDMX, is the MDM2
homolog, and binds to p53 at its transꢀactivation domain mainly repressing its transcriptional
21
activity. It is overexpressed in 18% to 19% of breast, lung, and colon cancers, in 50% of head and
2
2
23
neck squamous carcinomas and in 65% of retinoblastomas. To date, a number of biological
investigations and genetic evidence brought to light the importance of reciprocal levels of MDM2
and MDM4, and it seems that an optimal p53 reactivation and in turn a lasting, effective clinical
2
4
response can only be achieved by targeting both MDM proteins simultaneously. Therefore, the
identification of high affinity MDM2/MDM4 inhibitors is an emerging field of interest and claims
2
5
more efforts for medicinal chemists. Recently, attempts to restore the wildꢀtype p53 tumor
suppressor function has been achieved by interfering with MDM2/MDM4 heteroꢀdimerization,
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
6
which is required for an efficient inactivation of p53 function. This approach led to a peptide
(
Pep3), mimicking the MDM4 Cꢀterminus tail, which has been proved to be mainly effective
26
towards a specific subcellular pool of MDM2/MDM4/p53. Different complexity is instead posed
by the simultaneous targeting of p53 binding domain of both MDM2 and MDM4. In fact, despite
the 50% of amino acids sequence identity, computational and Xꢀray studies provided evidence that
these domains in the two MDM proteins are sufficiently dissimilar to explain the lack of activity of
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
7
the MDM2 inhibitors towards MDM4. Nonetheless, the same studies also highlight that the
binding domains of human MDM2 and MDM4 share some similarities thus offering promise for the
finding of smallꢀmolecule compounds that can simultaneously target the two proteins. In fact,
2
8ꢀ30
several peptide and nonpeptide dual inhibitors have been recently developed.
However,
therapeutic peptides have well known drawbacks such as chemical and physical instability,
susceptibility to proteases and oxidation, so that organic small molecules are generally preferred.
On the other hand, the organic dual inhibitors reported so far possess IC50 falling in micromolar
3
1ꢀ34
range.
Thus, nowadays the call for potent and effective modulators of p53ꢀ MDM2/MDM4 is
echoing and fosters the search for new chemical entities.
Comparing the Xꢀray crystallographic structural complexes of p53/MDM2 (PDB: 1YCR) and
p53/MDM4 (PDB: 3DAC), it is evident that the three critical residues of p53 transactivation
domain bind the MDMs Nꢀterminal domains in wellꢀdefined surface pockets characterized by three
hydrophobic clefts hereon named: Phe19, Trp23 and Leu26 pockets. The binding key residues of
p53 interact with L54, L57, I61, M62, Y67, Q72, V75, F86, F91, V93, H96, I99, Y100, I103 from
MDM2 and interact with M53, L56, I60, M61, Y66, Q71, V74, L85, F90, V92, P95, L98, Y99,
L102 from MDM4. Nine out of these fourteen residues are completely conserved for both proteins,
therefore clefts at the surface of MDM2 and MDM4 are similar to each other, a fact that in turn
35
enhances the possibility to find dual inhibitor.As part of our ongoing efforts in identifying novel
modulators of p53 pathway, useful against Glioblastoma Multiforme (GBM), we here report the
4
ACS Paragon Plus Environment

Page 5 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
identification, through a virtual screening campaign, of a small molecule (1) as dual MDM2/MDM4
binder and its potent derivative 12, obtained through a computerꢀaided optimization step. NMR
studies have been used to confirm the pocket of binding in MDM2. p53 dissociation from MDM2/4
has been quantified through an immuneꢀenzymatic assay. The restoration of the p53 pathway
together with the ability of 12 to induce both early and late apoptosis and to inhibit cancer cell
proliferation of different glioblastoma (U87MG) and neuroblastoma (SHSYꢀ5Y) cell lines, and of
their stem cells (CSC) subpopulation was assessed. Notably, 12 was able to cause concentrationꢀ
dependent inhibition of cell proliferation, yielding an IC50 value of 356 ± 21 nM in neuroblastoma
SHSY5Y cells, and proved to efficiently block CSC growth.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
RESULTS AND DISCUSSION
Virtual Screening Campaign. With the aim of finding smallꢀmolecules that can target both MDM
proteins simultaneously, a Virtual Screening (VS) campaign, by means of Glide5.5 (Glide, version
5
.5; Schrodinger, LLC: New York, 2009) was accomplished. First, we screened an inꢀhouse
compound library of ~4000 small molecules on a MDM2 3D structure (PDB code: 3LBL, see
Methods section for structure choice criteria). VS results were then sorted on the basis of the
docking scores, which ranged from −11.083 to −0.681. Solutions with a docking score higher than
the average docking score calculated for the known active compounds (−6.00) were discarded (see
Methods and Table SIꢀ1 for details). On the basis of this criterion, about the 30% of the entries were
retained (825 molecules) and visually inspected into the MDM2 binding site. Only the molecules
able to occupy at least two of the three MDM2 subpockets (Leu26, Trp23 and Phe19) have been
then used (145 entries) in a second round of screening against the p53ꢀbinding site of MDM4 (PDB
code: 4N5T). Giving the docking score (ꢀ4.017) of the known selective MDM4 inhibitor 2ꢀ[2ꢀ
Chloroꢀ4ꢀ[(1,5ꢀdihydroꢀ3ꢀmethylꢀ5ꢀoxoꢀ1ꢀphenylꢀ4Hꢀpyrazolꢀ4ꢀylidene)methyl]ꢀ6ꢀ
36,37
ethoxyphenoxy]ꢀacetic acid methyl ester (SJꢀ172550, from now called 1.1),
only the molecules
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
with a docking score lower than ꢀ4.00 were retained (64). Again, the visual inspection of these bestꢀ
ranked molecules binding modes took in consideration the possibility of these ligands to establish
hydrophobic contacts into the three MDM4 subpockets and to reach the Nꢀterminus region. Only
eight compounds were selected for in vitro tests with the aim to evaluate their ability to inhibit both
MDM2 and MDM4 (see Table SIꢀ2 for structures and activities). Seven molecules displayed a good
inhibitory potency on MDM2 (IC in the nanomolar range); however, only compound 1 proved to
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
50
potently induce the dissociation of p53 from MDM4 (IC = 4.6 ± 0.7 nM), although it was less
50
effective toward MDM2 (IC50 = 93.7 ± 10.3 nM) (see paragraph “Biological Evaluation” for
details). With these data in hand, we decided to test its synthetic progenitor, herein named 2, already
available in our inꢀhouse database, although not highly ranked in the VS. As a result, 2 displayed a
better activity on MDM4 (IC = 10.4 ± 1.2 nM) with respect to 1, retaining a fair potency on
50
MDM2 (IC = 35.4 ± 3.9 nM, Table 1).
50
Worth noticing, these nine selected ligands demonstrated to be structurally unrelated to the MDM2
ligands described so far. In fact, structural similarities between the 65 best ranked ligands and
38
known MDM2 inhibitors (taken from the BindingDB databank) were calculated based upon
39
40
JChem fingerprints, with the Tanimoto similarity index between each hit and the most similar
known inhibitor (see Table SIꢀ1) ranging from 0.173 to 0.450, indicating that a simple 2D similarity
search would not have allowed to cherryꢀpick the 9 ligands as MDM2 antagonists.
Given these results, a lead optimization campaign was undertaken simultaneously on 1 and on the
more synthetically accessible 2, in order to improve the activity on MDM2 while retaining that on
MDM4, that was already in a low nanomaolar range for both hits (IC = 4.6 ± 0.7 nM and 10.4 ±
50
1
.2 nM, for 1 and 2 respectively)
6
ACS Paragon Plus Environment

Page 7 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Lead Optimization Design toward MDM2
The lead optimization process was designed by taking into account the binding poses of 1 and 2
within the MDM2 receptor. As for compound 1 in MDM2 (Figure 1a), the triphenyl moiety
occupies both the Trp23 and Phe19 lipophilic clefts, while the third phenyl ring is too far to
properly occupy the Leu26 pocket. In addition, from docking results analysis it seems as if the
bulky triphenyl moiety does not allow an accommodation of the pyridinylaminopropoxy chain into
the canonical binding site forcing it to flip toward the L2 loop where contacts with the M62 and E69
side chains could be detected. Based on this observation, compounds 12 and 13 with less bulky
groups were synthesized as first derivatives.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
As regards the lead 2, it occupies, similarly to 1, the Trp23 and Phe19 clefts with its triphenyl
group, whereas its pꢀphenol group can well fit into the Nꢀterminus region establishing hydrophobic
interactions with the Q24, I19, Y100, M50 side chains and Hꢀbonding the Y100 hydroxyl group,
while the methyl ester can partially occupy the Leu 26 subpocket (1b). The proper accommodation
into the Nꢀterminus region together with these additional interactions might explain the better
activity of 2 with respect to 1 on MDM2 (2IC =35.4 nM; 1IC =93.7 nM).
50
50
Figure 1. Docking poses of 1 (a) and 2 (b) in MDM2 binding site. 1 and 2 are depicted as golden
and coral sticks, respectively. The protein surfaces are shown as transparent cyan while the
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
interacting residues as cornflower sticks. For clarity reason, MDM2 binding subpockets are defined
in magenta (Leu 26), blue (Trp 23) and red (Phe 19) dots, respectively (in accordance with the p53
interacting side chains).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
For 2, which is synthetically more accessible than 1, a series of analogues with diphenyl moiety
instead of the triphenyl one were designed and, in order to improve the solubility, the methyl ester
was hydrolyzed (3ꢀ11). As regards the central chain of 2, we tried to invert the chirality of the Sꢀ
tyrosine scaffold (5). Furthermore, to investigate the effect of different substituents on the phenol
ring, involved in both hydrophobic and polar contacts, we removed the hydroxyl group (6), and
inserted, in position 4, electronꢀwithdrawing (7ꢀ8), polar (9) or alkyl (10-11) groups.
Synthesis
Compounds 4-11 were synthesized in two steps starting from readily available enantiopure amino
acids methyl ester and phenyl propionic acid derivatives, as shown in Scheme 1. In a first step the
carboxylic acid was activated with EDC and HOBt in dry chloroform, and the amino acid methyl
ester was added at 0°C. The reaction was stirred overnight at room temperature and purified to give
intermediates 2, 3, 14-22 in good yields (61ꢀ98%). The hydrolysis of the ester function mediated by
lithium hydroxide in a mixture of THF/H O gave compounds 4-11 with yields ranging from 62 to
2
9
7% (see also Table SIꢀ3 in Supplementary Information). Compound 15 has been obtained from its
ether precursor 14 by means of trifluoroacetic acid promoted cleavage of tertꢀbutyl group (Scheme
). Reduction of the nitroꢀgroup of compound 17 gave compound 18 in 97% yield (Scheme 3).
Compounds 1, 12, and 13 were synthesized from intermediates 2, 3, and 22, respectively, through a
Mitsunobu reaction with intermediate 23, where the pyridine nitrogen has to be protected as
2
N
ꢀ
oxide (Scheme 4). In the case of triphenylpropionic acid derivative 2, sonication was needed to
accomplish 20% yield of the desired ether 26. Intermediates 24, 25, and 26 were then reacted with
ammonium formate in order to restore the pyridine nitrogen, and finally hydrolysis of the ester
8
ACS Paragon Plus Environment

Page 9 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
function gave compounds 1, 12, and 13 in quite good yields (53ꢀ72%). According to Scheme 5,
intermediate 23 was obtained in a four steps synthetic strategy, starting from 2ꢀbromopyridine and
3
ꢀaminopropanol. After a neat reaction for twelve hours at 150° C, intermediate 30 was retrieved in
2% yield. The alcohol function was then protected as tertꢀbutylꢀdimethylsilyl (TBDMS) ether 31,
8
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
and the pyridine nitrogen was oxidized to
Finally, TBDMS protection was cleaved with TBAF to give the desired alcohol intermediate 23 in
6% yield.
Nꢀoxide using metaꢀchloroperbenzoic acid (mꢀCPBA).
7
Scheme 1. Synthesis of 2-(3-substitutedpropanamido)-3-substitutedpropanoic acids (4-11).
Scheme
2.
Synthesis
of
(R)-methyl-2-(3,3-diphenylpropanamido)-3-(4-
hydroxyphenyl)propanoate (15).
Scheme
3.
Synthesis
of
(S)-methyl
2-(3,3-diphenylpropanamido)-3-(4-
aminophenyl)propanoate (18)
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme
4.
Synthesis
of
(S)-2-(3-substitutedpropanamido)-3-(4-(3-(pyridin-2-
ylamino)propoxy)phenyl)propanoic acids (1, 12-13)
1
0
ACS Paragon Plus Environment

Page 11 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 5. Synthesis of 2-(3-hydroxypropylamino)pyridine-1-oxide (23)
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Biological Evaluation
MDM2/p53 complex dissociation. The ability of the new synthetized analogues to bind MDM2
was assessed by an immuneꢀenzymatic assay on cell lysates containing the native p53/MDM2
35
complex. Briefly, in this assay, cell lysates obtained from U87MG cells were preꢀincubated for 10
min with the tested compounds, and then incubated on wells preꢀcoated with a MDM2 antibody (for
details see Methods section).
The compounds selected from the VS results (see Supporting Information) showed a moderate
affinity toward MDM2 (Table 1ꢀSI), with the exception of N1ꢀ[5ꢀchloroꢀ6ꢀmethylꢀ2ꢀ(2ꢀ
4
1
pyridyl)pyrimidinꢀ4ꢀyl]ꢀN2ꢀ[6ꢀ(trifluoromethyl)ꢀ2ꢀpyridyl]ethaneꢀ1,2ꢀdiamine (RDR03871) and
1
2
ACS Paragon Plus Environment

Page 13 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
our selected lead compounds 1 and 2, which dissociated the MDM2/p53 with high to very low
nanomolar potencies (IC50 values, Table 1ꢀSI and Figure S3). Considering the synthesized
derivatives, compound 5, where the chirality of the Sꢀtyrosine scaffold was inverted (see Supporting
Information for docking results and explanations), and compound 11 showed a very good inhibitor
potency on MDM2 (IC50= 9.6 nM and 9.2 nM, respectively). However, it was the derivative 12 that
showed the higher activity with respect to the leads 1 and 2 (IC MDM2= 7.85 ± 1.07 nM, Figure 2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
0
and Table 1).
MDM4/p53 complex dissociation. An immuneꢀenzymatic assay on SHSYꢀ5Y cell lysates was set
35
up to quantify p53 dissociation from MDM4, retracing the ELISA used for p53/MDM2. Specific
absorbance at 450 nm was found to proportionally increase with protein concentration of SHSYꢀ5Y
cells (Suppl. Fig. 1A). Furthermore, blank samples, obtained in the absence of the primary p53
antibody, were under the 20ꢀ25% of total values. To further validate the assay, the selective MDM4
inhibitor 1.1 and the MDM2 inhibitor 4ꢀ((4R,5S)ꢀ4,5ꢀbis(4ꢀchlorophenyl)ꢀ2ꢀ(2ꢀisopropoxyꢀ4ꢀ
methoxyphenyl)ꢀ4,5ꢀdihydroꢀ1Hꢀimidazoleꢀ1ꢀcarbonyl)piperazinꢀ2ꢀone (Nutlinꢀ3, named hereafter
42,43
1
.2)
were selected as reference compounds (see Chart 1). 1.1 displaced the p53/MDM4
complex, with IC value of 847.4 ± 90.8 nM (Figure SIꢀ1 and Table 1), comparable to the affinity
5
0
36,37
value reported in literature.
As expected, 1.2 was not able to completely dissociate p53 from
MDM4 (Figure S1ꢀB), showing a maximal percentage of inhibition of 44 ± 2 (Table 1).
Among the compounds selected from the VS, only 1, and then its progenitor 2 (Table 1), showed a
good ability to bind MDM4, while the others (as reported in Table S1), were not able to disrupt
significantly the MDM4/p53 complex. Conversely, our synthesized derivatives dissociated the
MDM4/p53 complexes with high to very low nanomolar potencies (Table 1). In particular, almost
all our derivatives, including compound 5 that showed good results for MDM2 (see Figure S4 in SI
for explanations), were less potent than the lead compounds, while 12 and 13 showed lower or
comparable IC50 values (6.10 ± 0.85 and 3.3 ± 0.6 nM, respectively) with respect to 1 and 2 .
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Ultimately, these preliminary results show that compound 12 presents a higher activity with respect
to the leads 1 and 2 toward both MDM2 and MDM4, and effectively induces the dissociation of p53
from MDM2 and MDM4 with comparable potency.
Cl
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Cl
O
O
O
O
N
O
O
O
N
N
N
N
O
NH
Cl
O
SJꢀ1722550
Nutlinꢀ3
1.1
1.2
Chart 1. Chemical structures of the MDM2 and MDM4 reference inhibitors.
Table 1. Effect of new compounds on the dissociation of human p53/MDM2 or p53/MDM4
1
complex. Concentration (nM) leading to halfꢀmaximal inhibition of p53/MDM2 complex.
2
Concentration (nM) leading to halfꢀmaximal inhibition of p53/MDM4 complex. Data represent the
mean values (±SEM) of three independent determinations.
1
2
Compound
Structures
IC50
IC50
MDM2/p53
nM)
MDM4/p53
(nM)
(
1
4
ACS Paragon Plus Environment

Page 15 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
2
93.7 ± 10.3
4.6 ± 0,7
35.4 ± 3.9
18.9 ± 2.0
10.4 ± 1,2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3
4
5
6
22.8 ± 2.6
23.4 ± 2.7
9.6 ± 1.1
43.0 ± 4.3
214.5 ± 23.1
136.4 ± 15.5
140.0± 15.0
7
8
65.3 ± 5.5
13.8 ± 1.8
10.7 ± 1.6
87.9 ± 5.6
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
9
50.3 ± 4.7
207.4± 18.1
1
0
1
115 ± 16
9.2 ± 0.8
401.0 ± 39.9
55.8 ± 6.3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
1
1
2
3
7.85 ± 1.07
6.1 ± 0.9
3.3 ± 0.6
141.5 ± 18.5
1
1
.1
.2
> 1000
847.4 ± 90.8
> 1000
108 ± 5 nM
1
6
ACS Paragon Plus Environment

Page 17 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. Dissociation of human p53/MDMs complex. U87MG (for p53/MDM2) or SHSYꢀ5Y
(for p53/MDM4) cell lysates were incubated with 12 for 10 min, and the percentage of residual
p53/MDMs complex was quantified as reported in the Experimental Section. The data are expressed
as a percentage with respect to control cells (mean ± SEM, N = 3).
Identification of the binding site for 12 on MDM2 through NMR studies. In order to verify the
correctness of the in silico calculations about the pocket of binding for this series of compounds on
MDM2, an NMR study was performed. We did not investigate also the MDM4 as we expected
similar results. Thus, as regards the MDM2, the amino acids forming the binding site for 12 have
1
15
been identified by monitoring the chemical shift perturbation in the 2D Hꢀ N HSQC NMR
1
5
spectrum of the N isotopically enriched protein upon the addition of increasing amounts of the
44
ligand (Figure 3). Aliquots of a DMSO solution of 12 were added to a buffered solution of MDM2
protein at the concentration of 200 µM. The residues E25, M50, K51, E52, F55, Y56, G58, Q59,
I61, M62, L66, D68, H73, I74, D84, F91, S92, V93, K94, Y100, and I103 experience the largest
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
chemical shift variation in the presence of 12 at the theoretical concentration of 400 µM (see Figure
4
a). A comparison between the residues exhibiting the largest chemical shift perturbations in the
presence of 12 and those resulting from the binding of p53, according to the Xꢀray structure of
4
5
MDM2 with p53 transactivation domain clearly suggest that 12 binds the canonical p53 binding
site and no allosteric pockets are engaged. However, the ligand precipitation, observed during the
titration, prevents an exact estimation of its concentration in the buffered solution of the protein and
thus of its potency.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
15
Figure 3. 2D Hꢀ N HSQC spectrum of MDM2 (0.2 mM) in the absence (red) and in the presence
(blue) of 12 (0.4 mM).
1
8
ACS Paragon Plus Environment

Page 19 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. a) chemical shift perturbation (CSP) between the uninhibited MDM2 (0.2 mM) and
MDM2 in the presence of 12 (0.4 mM) evaluated according to the formula
ꢄ
ꢄ
∆δꢇ
1
∆δ ꢀ ꢁ ꢂ∆δ ꢅ ꢆ
ꢁ ꢈ ; the residues exhibiting the largest CSP are highlighted in red; b)
ꢃ
2
5
CSP mapping on the structure of the protein (PDB code: 3LBL) in the presence of 12 bound as
resulted from docking; the residues with the largest perturbation are highlighted in red while 12 is in
green.
Binding Mode of 12 in MDM2 and MDM4. Once NMR studies unambiguously confirmed that 12
binds the canonical binding site of p53/MDM2, docking calculations, by means of Glide5.5
software, were performed to unveil the theoretical binding mode of 12 to MDM2. Docking
calculation on MDM4 protein were also accomplished.
1
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
In particular, Figure 5a shows the lowest energy binding pose found for 12 in MDM2, where the
diphenyl moiety occupies the Phe19 (interacting residues: I61, M62, Y67, V93) and the Trp23
(interacting residues: L54, L57, F86, F91, I99, I103) pockets, and the tyrosine central scaffold
stretches along the α2 helix and establishes Van der Waals interactions with the carbon chains of
the K51 and F55 residues. The propoxyꢀamino linker interacts with the Q24, I19, M50, Y100, side
chains of the Nꢀterminus region, while the pyridinyl ring stacks with the H96 within the Leu26
pocket. Clearly, all these interactions endowed 12 with a low nanomolar inhibitory activity toward
MDM2. The fact that the full occupancy of the three binding clefts (especially Phe19 and Trp23),
together with an enlarged area of hydrophobic interactions, is important to reach a high inhibitor
potency toward this enzyme, is wellꢀknown and it is herein further highlighted by the IC50 values of
compounds 11 and 13. In fact, although owing a shorter chain, 11 can still completely occupy the
whole binding site of MDM2, displaying an IC comparable to that of 12, while a lower IC is
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
50
found for 13 that, possessing an phenylꢀethyl group in place of the diꢀphenyl one, is no longer able
to properly fulfill the deep and hydrophobic Phe19 or the Trp23 subpockets. Notably, the docking
results are in line with the NMR data aboveꢀpresented. In fact, as shown in Figure 4b, in the
described docking pose, 12 is surrounded by the residues exhibiting the largest CSP.
On the other hand, when 12 is docked into the MDM4 binding site (see Figure 5b) a different pose,
with respect to that found in MDM2 is observed. In both receptors the diphenyl group occupies the
Trp23 and the Phe19 subꢀpockets but the differences in the overall shape of the two proteins, with
the MDM4 pocket appearing more compact, makes the central linker of 12 to arrange differently.
Differences between the two proteins also arise in the Nꢀterminus region and especially in the Nꢀ
terminal coil conformation. The cleft shape of MDM4 allows the flexible linker of 12 to plunge
towards the Nꢀterminus region so that the central aryl group is sandwiched between the Y96 phenol
and the K51 carbonyl side chain, and the aminoꢀpyridine group Hꢀbonds with the P18 backbone and
establishes hydrophobic contacts with the Q20, V46 and Y96 side chains. This mode of interaction
2
0
ACS Paragon Plus Environment

Page 21 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
that stretches all along the enzymatic cavity evidently is responsible for the low nanomolar activity
toward MDM4. In fact, ligands endowed with different core bulkiness (triphenylꢀpropanamido,
diphenylꢀpropanamido, phenylbutanamido) but possessing the same long propoxyꢀamino linker can
be equally accommodated without affecting the inhibitory potency (1: IC50=4,6 nM; 12: IC50=6,1
nM; 13: IC50=3,3 nM), while shorter derivatives such as 11, are no longer able to fulfil the whole
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
cavity resulting endowed with a lower IC50
.
Noteworthy, it has to be said that although molecular modeling studies presented herein surely
provide explanations for the ligand activity, and are in good accordance with the NMR data, the
flexibility of the Nꢀterminal coil and the poor amount of information about the Nꢀterminus region
especially as regards the MDM4, make the Xꢀray crystallography needed to unambiguously prove
these predictions.
Figure 5. Docking poses of compound 12 in MDM2 (a) and MDM4 (b) binding sites. The ligand is
shown as magenta sticks, the protein surface as transparent cyan (MDM2) and transparent pale
green (MDM4), while the interacting residues as blue and green sticks, respectively.
Reactivation of the p53 pathway and antiproliferative activities in human neuroblastoma
cells. Based on the results obtained in the immunoenzymatic assay, derivative 12 was used in
further assays aimed at exploring the effects of a cellular treatment with a dual MDM2/4 inhibitor.
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
To this purpose, human neuroblastoma cells (SHSYꢀ5Y) were chosen as representative tumour cell
46,47
line, because they express a wildꢀtype p53, and present a high expression of MDM2/4.
The
36,37
42,43
selective MDM4 inhibitor 1.1
and the MDM2 inhibitor 1.2
were tested in parallel
experiments.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
First, the reactivation of p53 pathway was assessed, by evaluating transcription induction of p53
target genes and accumulation of p53 protein.
As expected, 1.2 induced a significant accumulation of p53 protein (Fig. 6A and B). Similar results
were observed with 1.1 (Fig. 6A). When cells were incubated with the two drugs, a significant
higher p53 accumulation with respect to cells treated with 1.2 or 1.1 alone (Fig. 6A and B) was
observed. The dual inhibitor 12 caused a significant enhancement of p53 protein levels (Fig. 6A and
B), even greater with respect to that elicited by the coꢀtreatment 1.1+1.2. This effect may be
explained considering the affinities of 12 toward MDM2 and MDM4, which are greater with
respect to those elicited by 1.2 and 1.1 toward MDM2 or MDM4, respectively. Anyway, such
results suggest that blocking both MDM2 and MDM4 can induce a better p53 protein reactivation.
To confirm this, we measured the transcriptional levels of the following p53 target genes: MDM2,
physiological inhibitor of p53 and its main transcriptional target; PUMA, a gene product required
4
7
for p53ꢀmediated apoptosis; p21, a cell cycle inhibitor. Challenging neuroblastoma cells with the
MDM2 inhibitor 1.2 for 24 h caused a significant induction of MDM2 and p21 (Figure 6C),
43
consistent with the data obtained in different tumor cell lines. No significant changes were
observed in PUMA mRNA levels (Figure 6C), consistent with the lack of cellular apoptosis
4
3
exhibited by 1.2 after 24 h of cell treatment The MDM4 blocker 1.1 significantly increased
MDM2 mRNA levels, whereas it did not affect p21 and PUMA transcription (Figure 6C), thus
suggesting that MDM4 inhibition is not enough to fully reactivate p53 transcriptional activity in the
analyzed time of cell treatment. When the two compounds were combined together, a significant
increase in p21 and PUMA transcriptional levels was evidenced with respect to singleꢀtreated cells
(Figure 6C). Such results are consistent with those obtained in western blotting analysis, and
2
2
ACS Paragon Plus Environment

Page 23 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
confirm that the combined MDM2/4 inhibition could reactivate p53 function with an efficacy
compared with single therapy. Compound 12 caused a significant enhancement of the three
analyzed p53 target genes (i.e., MDM2, p21 and PUMA), with greater efficacy with respect to cells
incubated with 1.1+1.2 (Figure 6C). These results suggest that the new compound is able to induce
an efficient reactivation of p53 pathway.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6. Reactivation of p53 pathway. (A, B) SHSYꢀ5Y cells were treated with 12 (1 µM), or the
MDM2 inhibitor 1.2 (1 µM), or the MDM4 inhibitor 1.1 (1 µM), alone or in combination, for 24 h.
Following treatment, p53 protein expression was evaluated by western blot analysis, as reported in
the Experimental Section, using GAPDH as the loading control. A representative western blot (A)
and the quantitative analysis of immunoreactive bands (B) are shown. The data are expressed as the
percentage of optical density (OD) versus control (mean ± SEM, N = 2). (C) The cells were treated
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
as in (A), mRNA of p53 target genes (MDM2, PUMA, and p21) was quantified by realꢀtime RTꢀ
PCR as reported in the Experimental Section. The data are expressed as the fold change versus
control (mean ± SEM, N = 3). *P < 0.05, **P < 0.01, ***P < 0.001 versus Control; ## P < 0.01
versus cells treated with 1.2 alone; §§§ P < 0.001 versus cells treated with 1.1 alone.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Next, the antiproliferative effects of a simultaneous MDM2/4 inhibition were examined by
incubating SHSYꢀ5Y cells with the selected compounds. As depicted in Figure 7 (panel A), both
1
.2 (1 µM) and 1.1 (1 µM) exhibited a significant and comparable inhibition of neuroblastoma cell
proliferation (22.9 ± 1.3 and 23.7 ± 2.6 % of inhibition, respectively). When combined together, the
two compounds exhibited a significant increase in their antiproliferative effects (47.1 ± 2.3 % of
inhibition at 1 µM). Notably, derivative 12, at the same tested concentration (1 µM), caused an
inhibition of cell proliferation of 57.5 ± 2.5 (Fig. 7A). The new derivative caused concentrationꢀ
dependent effects, yielding an IC50 value of 356 ± 21 nM (Fig. 7B). Altogether, these data
demonstrate that the simultaneous MDM2/4 blocking is able to induce greater antiproliferative
effects in human neuroblastoma cells.
As a comparison, the same experiments were repeated in human U87MG cells, exhibiting a wildꢀ
48
type p53 but not overexpressing MDM4. In these cells, the MDM4 inhibitor alone did not affect
cell proliferation after 24 h of treatment (Figure 7C). Conversely, 1.2 and 12 caused a significant
inhibition of cell proliferation (Figure 7C). When used simultaneously, 1.2 was able to sensitize
SHSYꢀ5Y to the MDM4 inhibitor (Figure 7C). Of note, the new derivative was less efficacious in
U87MG cells with respect to SHSYꢀ5Y cells (30.1 ± 4.4 % in U87MG cells). These results are
consistent with the minor MDM4 protein expression of U87MG cells.
To consolidate these data, the effect of the lead compound was assessed also by Neutral Red Cell
Proliferation, which is based on the ability of viable cells to incorporate and bind the supravital dye
4
9
neutral red in the lysosomes. As depicted in the Figure S5 in SI, compound 12 was confirmed to
2
4
ACS Paragon Plus Environment

Page 25 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
inhibit SHSYꢀ5Y and U87MG cell proliferation, showing comparable percentages of inhibition
with respect to those obtained in the MTS assay.
In addition, preliminary experiments were performed isolating the cancer stem cell subpopulation of
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
50,51
the tumor cell lines, which is responsible for cancer recurrence and metastasis.
When CSCs,
isolated from U87MG or SHSYꢀ5Y were incubated with compound 12 (100 nMꢀ10µM), a
significant and concentrationꢀdependent inhibition of cell proliferation was evidenced (Figure 7D).
The effects elicited by the new derivative were significantly higher in CSCs isolated from
neuroblastoma cells with respect to U87MG cells (% of inhibition at 100 µM: 78.4 ± 5.9 and 99.2 ±
0
.4 in U87MG and SHSY5Y cells, respectively). These results confirm that compound 12 is more
efficacious in cells overexpressing MDM4.
Finally, we verified if the antiproliferative effects of 12 were associated to cellular apoptosis.
Challenging SHSYꢀ5Y with the new derivative (1 µM) caused a significant externalization of
phosphatidylserine, in the absence or in the presence of 7ꢀaminoꢀactinomycin binding to DNA, thus
denoting the induction of both early and late apoptosis (Figure 8A and B).
2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. Antiproliferative effects of 12 on tumor cells and in their stem cell counterpart. SHSYꢀ5Y
(A, B), U87MG (C), or CSCs (D) were challenged for 24 h (AꢀC) or seven days (D) with the
indicated concentrations of 12, or 1.2 (1 µM), or 1.1 (1 µM), alone or in combination. Following
incubation, cell proliferation was determined by MTS assay as reported in the Experimental
Section. The data are reported as percentage with respect to control, set to 100% (mean ± SEM, N =
3
). A sigmoid dose−response analysis (GraphPad Prism 5 software) was derived in SHSYꢀ5Y cells
(B). **P < 0.01, ***P < 0.001 vs Control; ## P < 0.01 versus cells treated with 1.2 alone; §P < 0.05,
§§ P < 0.01 versus cells treated with 1.1 alone.
2
6
ACS Paragon Plus Environment

Page 27 of 53
Journal of Medicinal Chemistry
1
2
A
B
3
4
5
6
7
8
9
1
00
9
8
7
6
0
0
0
0
earlyapoptosis
late apoptosis/death
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
50
40
30
***
2
0
***
Control
1
2
10
0
Figure 8. Apoptotic effects of derivative 12. (A, B) SHSYꢀ5Y cells were challenged for 24 h with
ꢁM 12. Following treatment, cells were collected and the percentage of early/late apoptosis was
1
determined as described in the Experimental Section. (A) Representative plots of Annexin assay (B)
Quantification of Annexin staining. The data are reported as the percentage of apoptotic cells
versus the total number of cells (mean ± SEM, N = 3). *** P < 0.001 versus control.
CONCLUSIONS
The high importance of the impaired functioning of p53 in human malignancies genesis and
progression, is still nowadays a stimulus for finding effective novel anticancer agents. In this
respect, a number of small molecule MDM2 inhibitors are undergoing clinicalꢀstage studies for the
treatment of diverse human cancers. However, it is recently emerging that MDM2ꢀselective
inhibitors may induce high levels of MDM4, thus affecting the final clinical response. Thus, it
seems that an optimal and lasting p53 reactivation can more efficiently be achieved by targeting
both MDM proteins simultaneously. Herein, through a “sequential” virtual screening performed
first on MDM2 and then on MDM4 the lead compound 1 was identified. It turned out to possess an
IC50 of 93.7 nM and of 4.6 nM on MDM2 and MDM4 respectively. A series of compounds were
synthesized to optimize its activity on MDM2. As a result, compound 12 showed low nanomolar
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
IC50 for both targets (IC50 of 7.8 nM and of 6.1 nM towards MDM2 and MDM4 respectively).
Taken together NMR experiments and molecular docking, allow to identify a pocket of binding in
MDM2 and a possible binding mode. Notably, 12 induced a significant higher p53 accumulation
with respect to cells treated with 1.2 or 1.1 alone and with respect to that elicited by the coꢀ
treatment 1.1+1.2. Along the same lines 12, caused a concentrationꢀdependent inhibition of
neuroblastoma cell proliferation more efficiently than both 1.2 and 1.1 alone and in combination.
Notably, 12 showed a significant and concentrationꢀdependent inhibition of CSCs cell proliferation,
isolated from both U87MG and SHSYꢀ5Y. Finally, challenging SHSYꢀ5Y with 12 (1 µM) a
significant externalization of phosphatidylserine was observed, thus denoting the induction of both
early and late apoptosis.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In perfect line with its activity profile, the effects elicited by the new derivative were significantly
higher in CSCs isolated from neuroblastoma cells, overexpressing MDM4, with respect to U87MG
cells. Taking in mind that the cancer stem cell (CSCs) subpopulation of the tumor cell lines are
responsible for cancer recurrence and metastasis, the latter data are worth of note. All in all, the
results herein presented hold new promises for a better understanding of the structural requirements
for dual disruptors of MDM2/MDM4/p53 interaction and for preclinical studies of our compound
EXPERIMENTAL SECTION
Virtual Screening. For our study, our inꢀhouse database of ~4000 compounds was used. Such
database was prepared using LigPrep (LigPrep, version 2.5, Schrödinger, LLC, New York, NY,
2011) generating all possible tautomeric, enantiomeric, and protonation states and keeping only
those possessing good ADME properties (calculated by means of QikProp). The final database was
composed of 4333 molecules. Compound 12 and 5 were prepared in the same way, considering all
protomeric states.
2
8
ACS Paragon Plus Environment

Page 29 of 53
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
As per the MDM2 Xꢀray selection, multiple 3D structures of MDM2 can be found in the Protein
Data Bank (PDB). Among them, we took into consideration just the ones containing the Nꢀterminus
residues 16−24 (e.g., 3LBL, 4HBM, 4DIJ, 4JVR, 4JVE, 1T4E, 4ERF, etc.) which folds into an
ordered helix that changes shape and size of the catalytic pocket and provides additional interaction
points. So, among the human nonꢀtruncated Xꢀray structures, only those coꢀcrystallized with an
organic compound were considered, and the one with the highest resolution (1.60 Å) (PDB code:
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3
LBL) containing a spirooxindole derivative was chosen for our Virtual Screening campaign.
As for the MDM4 Xꢀray selection, the same criteria were followed. Only the nonꢀtruncated human
and humanized Xꢀray structures were considered and among them, the one with a highest resolution
(1.70 Å) was chosen (PDB code: 4N5T). This protein is a humanized Zebrafish MDM4, where the
two mutations L46V/V95L were generated by DNA siteꢀdirected mutagenesis in order to make the
binding site identical to that of a human MDM4. The further mutation E20Q, which is located on
the Nꢀterminus region, was manually produced before preparing the protein for the Virtual
Screening campaign. In this way, the whole cavity of 4N5T, including the Nꢀterminus region, was
identical to that of a human MDM4.
The two proteins were prepared using the Protein Preparation Wizard implemented in Maestro Suite
2
011 (Maestro, version 9.0.211; Schrodinger, LLC: New York, 2009). During the preparation, all
water molecules were deleted, hydrogen atoms added, the complex minimized. The receptor grids
were generated using the grid generation in Glide 5.5 (Glide, version 5.5; Schrodinger, LLC: New
York, 2009.) centred around the crystallized ligand using default settings. For the VS on the MDM2
structure, the SP mode in Glide was used first, retaining only the best 50% hits that were redocked
in XP mode, letting all other settings at the default. Solutions with a docking score higher than the
average docking score of the known active compounds (ꢀ6.00) were discarded. Structures and
inhibition data for known inhibitors (~30 hits including compound 1.2) were downloaded from the
38
BindingDB database. The molecules were prepared using LigPrep, considering all the protonation
2
9
ACS Paragon Plus Environment