Efficient reactivation of p53 in cancer cells by a dual MdmX/Mdm2 inhibitor

Article abstract of DOI:10.1021/ja509223m

The aberrant interaction between p53 and Mdm2/MdmX is an attractive target for cancer drug discovery because the overexpression of Mdm2 and/or MdmX ultimately impairs the function of p53 in approximately half of all human cancers. Recent studies have show

Full text of DOI:10.1021/ja509223m

Subscriber access provided by UNIV NEW ORLEANS
Article
Efficient reactivation of p53 in cancer cells by a dual MdmX/Mdm2 inhibitor
Lingyun Qin, Fei Yang, Cindy Zhou, Yao Chen, Huashan Zhang, and Zhengding Su
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/ja509223m • Publication Date (Web): 02 Dec 2014
Downloaded from http://pubs.acs.org on December 10, 2014
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 the American Chemical Society 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 37
Journal of the American Chemical Society
1
2
3
4
Efficient reactivation of p53 in cancer cells by a dual MdmX/Mdm2 inhibitor
5
6
7
Lingyun Qin1, Fei Yang1, Cindy Zhou2, Yao Chen1, Huashan Zhang1 and Zhengding Su1,2,3,*
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
1Department of Biotechnology and Collaborative Innovation Center for Industrial Fermentation,
Hubei University of Technology, Wuhan 430068, China
2Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada and
3Department of Structural Biology, St Jude Children’s Research Hospital, Memphis, TN 38105,
USA
*Author to whom correspondence should be addressed: Z. Su, Email: z2su@uwaterloo.ca or
zhengding.su@stjude.org
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 37
1
2
3
4
Abstract:
5
6
7
8
9
The aberrant interaction between p53 and Mdm2/MdmX is an attractive target for cancer
drug discovery because the overexpression of Mdm2 and/or MdmX ultimately impairs the
function of p53 in approximately half of all human cancers. Recent studies have shown that
inhibition of both Mdm2 and MdmX is more efficient than that of a single target in promoting
cellular apoptosis in cancers. In this study, we demonstrate that a dual smallꢀmolecule antagonist
of Mdm2/MdmX can efficiently reactivate the p53 pathway in model cancer cells overexpressing
MdmX and/or Mdm2. The dual antagonist was rationally designed based on segmental
mutational analysis of the Nꢀterminal domain of MdmX and the crystal structure of the Nꢀ
terminal domain of Mdm2 in complex with nutlinꢀ3a (an Mdm2ꢀspecific inhibitor). The current
work establishes a small molecule therapeutic candidate that targets cancers overexpressing
Mdm2 and/or MdmX.
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
2
ACS Paragon Plus Environment

Page 3 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Direct gene alterations of p53 or the binding of Mdm2 and MdmX to the p53
transactivation domain are two alternative mechanisms causing the inactivation of p53 function
and tumor survival1ꢀ3. The elevated expression of Mdm2 and/or MdmX not only accounts for
nearly half of all cancers that retain wildꢀtype p53, but also highly correlates with highꢀgrade,
lateꢀstage, and more resistant tumors3;4. Therefore, reactivation of p53 in wildꢀtype p53ꢀ
expressing tumors through the release of its physical interaction with Mdm2 and MdmX has
become the focus of cancer drug discovery3;5.
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
Thus far, many smallꢀmolecule inhibitors of Mdm2 have been discovered based on
combinatorial chemistry and medicinal chemistry6. These inhibitors promote p53ꢀmediated cell
cycle arrest, apoptosis and tumor regression in cancer cells overexpressing Mdm2. One of these
promising Mdm2 inhibitors is Nutlinꢀ3a, which is currently under clinical investigation1;7.
Although the structures of MdmX and Mdm2 are highly similar, the existing Mdm2 inhibitors
exhibit relatively weak binding affinity for the MdmX protein, leading to only marginal
induction of apoptosis in tumor cell lines overexpressing MdmX8ꢀ10. Conversely, recent studies
have confirmed that the inhibition of both Mdm2 and MdmX in cancer cells expressing wildꢀ
type p53 can activate p53 more strongly than agents that only antagonize Mdm2 activity11ꢀ13
.
Thus, to achieve full reactivation of p53 in cancer cells, it is necessary to apply a therapeutic
treatment that is dualꢀspecific to disrupt the interaction between p53 and its negative regulators
Mdm2 and MdmX. It has recently been made progress for designing dual inhibitors of Mdm2
and MdmX with different rationales that either inhibit both proteins14ꢀ15 or induce the
dimerization of Mdm2 and MdmX12.
In a cancer cell expressing wildꢀtype p53, p53 activity is often impaired by the interaction
of its 15ꢀamino acid transactivation domain (p53p) with the Nꢀterminal domains of Mdm2 and
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 37
1
2
3
4
5
6
7
8
9
MdmX (referred to hereafter as NꢀMdm2 and NꢀMdmX, respectively). In this study, we
compared the ligand binding behaviors of NꢀMdm2 and NꢀMdmX to p53p and nutlinꢀ3a using
segmental mutagenesis, NMR spectrometry and Xꢀray crystallography, with the aim of designing
a dual inhibitor of Mdm2/MdmX that utilizes the nutlinꢀ3a scaffold.
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
Segmental mutation analysis of N-MdmX
NꢀMdmX and NꢀMdm2 are highly homologous in terms of their amino acid sequences,
secondary structural elements, and tertiary conformations (Fig. 1a & 1b). It is notable that there
are nine segments of NꢀMdmX that are significantly different from their counterparts in Nꢀ
Mdm2 with regards to their physicochemical properties including polarity, hydrophobicity and
backbone flexibility. The isolated p53 transactivation domain (p53p) can bind to NꢀMdmX and
NꢀMdm2 with Kd, values of 0.42 µM and 0.13 µM, respectively (Table 1). In contrast, NꢀMdmX
and NꢀMdm2 exhibit distinct binding affinities to nutlinꢀ3a with Kd values of 6.5 µM and 0.019
µM, respectively. Interestingly, even though nutlinꢀ3a shows a weak affinity for MdmX, this
molecule was shown to be effective at killing cancer cells in models of cancer caused by
MdmX16.
We sought to investigate how the nine segments of NꢀMdmX differentiate the binding
affinities of NꢀMdmX from NꢀMdm2 for p53p and nutlinꢀ3a. A series of mutants was
constructed by substituting each segment in NꢀMdmX with its counterpart from NꢀMdm2, one at
a time. The resulting nine singleꢀsegmental mutants were evaluated for their binding affinities for
p53p and nutlinꢀ3a by isothermal titration calorimetry (ITC). As summarized in Table 1, the nine
mutants exhibited different affinities for p53p and nutlinꢀ3a. Two of these singleꢀsegmental
4
ACS Paragon Plus Environment

Page 5 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
replacements yielded significant effects on p53p binding, whereas the others showed no effects.
The replacement of Segment F (NꢀMdmXF) significantly enhanced p53p binding compared with
the wildꢀtype protein, whereas the replacement of Segment H (NꢀMdmXH) dramatically reduced
p53p binding. It is likely that the local conformation of the amino acid sequence around Segment
F and Segment H have affected the binding of p53p, as the NꢀMdmXF mutant adopted a
relatively rigid sequence from NꢀMdm2, whereas in the NꢀMdmXH mutant, a rigid sequence was
replaced with a flexible one.
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
To determine why the mutant NꢀMdmXH had a reduced binding affinity for p53p, which
was unexpected, we replaced the adjacent segment on helix 4 in NꢀMdmXH with its counterpart
from NꢀMdm2 to generate a double segmental mutant (NꢀMdmXHI). As indicated in Table 1,
this double mutation in NꢀMdmXHI was able to compensate for the affinity loss in the NꢀMdmXH
mutant, and showed almost comparable binding affinity to that of NꢀMdmX. This result suggests
that the amino acid sequences from segments F, H, and I might play an important role in
determining the p53p binding affinity of NꢀMdmX and NꢀMdm2.
Compared with p53p binding, these segmental replacements showed completely different
effects on nutlinꢀ3a binding. The mutations in NꢀMdmXF and NꢀMdmXH enhanced nutlinꢀ3a
binding in 4.3ꢀfold and 2.4ꢀfold, respectively, compared with NꢀMdmX. Notably, the mutation in
segment G (NꢀMdmXG), which resulted in no effect on p53p binding, enhanced nutlinꢀ3a
binding more than 100ꢀfold. Like p53p, nutlinꢀ3a exhibited enhanced binding affinity in the
double mutant NꢀMdmXHI. Based on these observations, it is likely that NꢀMdmX and NꢀMdm2
employ different mechanisms for recruiting nutlinꢀ3a and p53p, which are related to the
structural conformations of segment F, segment G and segment H.
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 37
1
2
3
4
Structural comparison between N-Mdm2/ nutlin-3a complexes and N-MdmX/p53p complexes
5
6
7
8
9
To identify a nutlinꢀ3a scaffoldꢀbased analog that targets both Mdm2 and MdmX, we
sought to determine the structures of NꢀMdm2 and NꢀMdmX complexed with nutlinꢀ3a. After
highꢀthroughput screening for crystallization conditions and further optimization, we were able
to obtain a highꢀresolution Xꢀray crystal structure of the NꢀMdm2/nutlinꢀ3a complex at 1.3 Å
(Fig. 2a), although our NꢀMdmX/nutlinꢀ3a complex suffered difficulties in crystallization.
Nevertheless, we determined the NMR structure of NꢀMdmX complexed with a nutlinꢀ3a analog
(SJ298)17. These structural data should provide valuable information to understand the structural
mechanism of the interactions of NꢀMdmX and NꢀMdm2 with small molecules.
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
Our crystal structure revealed that the principal features of the p53p binding surface on
NꢀMdm2 (Fig. 2b, 1ycr.pdb18) were preserved during nutlinꢀ3a binding (Fig. 2c). The protein
surface had a relatively deep hydrophobic pocket that was primarily filled by three p53p side
chains (Phe19, Trp23 and Leu26), which were therefore called the F19, W23 and L26 binding
pockets, respectively. However, the surface of NꢀMdm2 was obviously altered when it was
bound to nutlinꢀ3a rather than p53p. These major changes occurred at the F19 and L26 binding
pockets (Fig. 2c vs. Fig. 2b). The F19 binding pocket widened, whereas the L26 binding pocket
narrowed, forming a hydrophobic surface for docking a chloroꢀ group of nutlinꢀ3a. We also
observed that the His95 residue of NꢀMdm2 interacted with nutlinꢀ3a (data not shown),
mimicking the interaction between p53p and the His95 residue of NꢀMdm219.
The interaction of nutlinꢀ3a with the NꢀMdm2 surface provides a valuable model to
predict the interaction of small molecules with NꢀMdmX. The NꢀMdm2 conformation in the Nꢀ
Mdm2/nutlinꢀ3a complex showed very similar secondary and tertiary structural characteristics to
6
ACS Paragon Plus Environment

Page 7 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
those of NꢀMdmX complexed with p53p (3dba.pdb, Fig. 2d). One interesting observation was
that the F19 binding pocket for nutlinꢀ3a on NꢀMdm2 was very similar to that observed on Nꢀ
MdmX complexed with p53p (Fig. 2c vs. Fig. 2e). Conversely, when comparing p53p binding
on NꢀMdmX and NꢀMdm2, one found that two of the three key residues of p53p, Phe19 and
Trp23 are positioned in the same configuration, suggesting that the F19 and W23 binding
pockets on NꢀMdmX should be able to accommodate the corresponding functional groups of
nutlinꢀ3a. However, the configuration of Leu26 residue of p53p is shifted approximately 0.6 Å
(Fig. 2e), caused by the open and closed positions of Tyr99 on NꢀMdm2 and NꢀMdmX,
respectively19, when the p53p binds to NꢀMdmX and NꢀMdm2, respectively. This observation
indicated that the L26 binding pockets are flexible in NꢀMdmX and NꢀMdm2. The W23 binding
pocket also becomes shallower in the NꢀMdmX/p53p complex than NꢀMdm2 complexed with
p53p (Fig. 2b & 2e). Using a computational docking analysis, we found that neither paraꢀchloro
group of nutlinꢀ3a fit into the W23 and L26 binding pockets on the p53p binding surface (Fig.
2f). Therefore, it is likely that the W23 and L26 binding pockets on MdmX are too flexible to
have sufficient interactions with proven Mdm2 inhibitors, particularly the nutlin compounds, as
observed using standard docking procedures for mimicking nutlin/MdmX interactions20. This
flexibility could also explain why nutlinꢀ3a exhibited a weak binding affinity for NꢀMdmX. Thus,
nutlinꢀ3a, as a specific inhibitor of the Mdm2ꢀp53 interaction, has not been optimized for binding
to NꢀMdmX.
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
Dynamic characteristics of the binding pockets in N-Mdm2 and N-MdmX
The hydrophobic residues from the three helical elements (H2, H3 and H4, Fig. 3a)
heavily contribute to the formation of W23 and L26 binding pockets in NꢀMdm2 and NꢀMdmX.
These binding pockets are closed by the hydrophobic interactions between the Cꢀterminal
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 37
1
2
3
4
5
6
7
8
9
residues of the H4 helix (i.e., Y’99, Y’103, L’101 and L’106) and the H2 helix (Fig. 3a). A salt
bridge between D100 and R103 on the H4 helix of NꢀMdmX exposes its binding surface to the
solvent environment, unlike NꢀMdm2 (Fig. 3b). In NMR experiments, we found that when Nꢀ
MdmX was complexed with nutlinꢀ3a, its H2 and H4 helices were much more flexible than their
counterparts in the NꢀMdm2/nutlinꢀ3a complex. All the resonance peaks could be assigned in the
N15ꢀH1 HSQC NMR spectrum for NꢀMdm2 in the free form, whereas only 62 out of 89 residues
could be assigned in the NꢀMdmX spectrum. The missing peaks were mainly located in the
region between Lys64 and Lys94 in between H2 and H4 helices. Many NMR resonance peaks in
the 15Nꢀ1H HSQC spectrum of NꢀMdmX remained missing from these regions, even in the
complex with nutlinꢀ3a (Fig. 3c). Upon nutlinꢀ3a binding, only three new resonance peaks
appeared for residues Met61, Asp67 and Leu81. Furthermore, our heteronuclear overhauser
effect (NOE) experiments on the complexes confirmed that the region in between Lys64 and
Lys94 of NꢀMdmX was very flexible, even when complexed with nutlinꢀ3a, compared with the
NꢀMdm2/nutlinꢀ3a complex (Fig. 3d), which was related to segments D, E, F and G (Fig. 1a).
Considering current and previous structural investigations on NꢀMdmX and NꢀMdm2 interacting
with p53p and nutlinꢀ3a, the interactions between p53p and nutlinꢀ3a differ from those of Nꢀ
MdmX and NꢀMdm2 with respect to segments G, H and I. It was likely that the hydrophobic
packing by H2, H3 and H4 helices as well as the rigidity of S2 and S3 sheets and H3 helix
cooperated to form different binding surfaces on NꢀMdmX and NꢀMdm2, which determined their
binding affinities for p53p and nutlinꢀ3a. Our observations suggested that the development of
potent dual inhibitors for MdmX and Mdm2 should be feasible when starting from the nutlinꢀ3a
scaffold; in particular, these efforts should focus on enhancement of the hydrophobic packing of
H2, H3 and H4 helices as well as the rigidity of S2, S3 sheets and H3 helix.
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
8
ACS Paragon Plus Environment

Page 9 of 37
Journal of the American Chemical Society
1
2
3
4
Rational design of MdmX/Mdm2 dual inhibitors
5
6
7
8
9
We next optimized the nutlinꢀ3a scaffold for NꢀMdmX binding to obtain a dual inhibitor
targeting both Mdm2 and MdmX. To achieve this goal, we considered the following combined
strategies: 1) enhancing the hydrophobic interactions of the designed compounds with the F19
and W23 binding pockets; 2) reducing the volume of the pharmacophores for the L26 binding
pocket and W23 binding pocket; 3) introducing hydrogen bonding between the designed
compounds and the binding pockets, as hydrogen bonding has been found to enhance the binding
of p53 peptidomimetics and small molecules19;21; 4) enhancing the rigidity of S2, S3 sheets and
H3 helix as the region between S2 and S3 in the NꢀMdmX/nutlinꢀ3a complex is significantly
more flexible than that in the NꢀMdm2/nutlinꢀ3a complex. Therefore, modification of nutlinꢀ3a
focused on its three main pharmacophores (R1, R2, and R3 in Fig. 4a) to develop nutlin3aꢀ
analogs to fit into the p53p binding surface on NꢀMdmX. The modification of these functional
groups in nutlinꢀ3a was strictly limited to the pharmacophores that existed among FDAꢀapproved
smallꢀmolecule drugs, as assessed through chemoinformatics22. Accordingly, a structureꢀoriented
nutlinꢀ3a analog library containing 216 compounds was prepared using combinatory chemistry
(Fig. 4a).
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
These compounds were screened for their binding affinities using a fluorescence
polarization (FP) assay (see Methods). Six compounds from the library were found to displace
p53p from the NꢀMdmX/p53p complex with median inhibitory concentration (IC50) values less
than 300 nM. The binding affinities of the six compounds for NꢀMdmX were further evaluated
with ITC to obtain the Kd values (Fig. 4b). Three of the six compounds were confirmed to
maintain high affinity for NꢀMdm2. One of these compounds, compound H109, exhibited high
9
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 37
1
2
3
4
5
6
affinity for both NꢀMdmX and NꢀMdm2 with Kd values of 27 nM for NꢀMdmX and 5.7 nM for
NꢀMdm2 (Fig. 4b & 4d).
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
Specificity and dose response of the potent inhibitor
Through molecular docking, compound H109 nicely fits the p53p binding surface on Nꢀ
15
MdmX (Fig. 5a). A Nꢀ1H HSQC spectrum of NꢀMdmX in complex with the compound in a
molar ratio of 1:2, respectively, generated sharp resonances, compared with those of NꢀMdmX in
the free form (Fig. 5b). Upon H109 binding, the flexible region in the NꢀMdmX backbone
became significantly perturbed and rigid (Fig. 5c). It was likely that H109 enhanced binding
affinity to NꢀMdmX through multiple modes, including the enhancement of hydrophobic
interactions in the W23 and L26 binding pockets and the reduction in backbone dynamics in the
region around Gln71 and Val92 residues via the hydrogen bonds introduced.
Compound H109 should be effective only in cells with wildꢀtype p53 and not in cells
with transcriptionally inactive p53. To determine whether the inhibition of p53ꢀMdm2/MdmX
binding by H109 could be translated into the activation of the cellular p53 pathway, we treated
three cancer cell lines with wildꢀtype p53, HCT116, RKO and H460a, which were previously
used to evaluate nutlinꢀ3a7, with H109 for 8 hours and monitored the expression of the p53 and
p21 genes via realꢀtime polymerase chain reaction (RTꢀPCR). As indicted in Fig. 6a, the
transcription of p21 increased in a doseꢀdependent manner in all cell lines, consistent with the
accumulation of its transcriptional activator, p53. By contrast, transcription of the p53 gene itself
was unaffected by H109. These data indicated that the compound upꢀregulated p53 via a
posttranslational mechanism functioning like nutlinꢀ3a.
10
ACS Paragon Plus Environment

Page 11 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
To confirm the specificity of the effect of H109 on MdmX, a GSTꢀNꢀMdmX fusion
protein was used to pull down p53 in vitro in the presence of H109, nutlinꢀ3a or nutlinꢀ3b. As
shown in Fig. 6b, H109 significantly inhibited the interaction of NꢀMdmX with the p53 protein
at a concentration of greater than 0.3 ꢀM, whereas nutlinꢀ3a showed a marginal inhibition of the
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
interaction of NꢀMdmX with p53 until 20 ꢀM of ligand was applied. As expected, nutlinꢀ3b
showed no inhibition of the same interaction at the tested ligand concentrations.
To test in vivo whether H109 could induce the p53 pathway by inhibiting MdmX binding
to p53, we utilized an Mdm2 or MdmXꢀdeficient mouse embryonic fibroblast (MEF) growth
assay. The p53LSL/ꢀ MEFs with the indicated genotypes were infected with adenoꢀGFPꢀCre for 12
h to allow Creꢀmediated excision of the stop element and subsequent p53 expression according
to methods reported previously16;23;24. The cells were then exposed to H109, nutlinꢀ3a (the
positive control) or nutlinꢀ3b (the negative control) for 48 h, and cell growth/viability was
measured using an MTT assay. As a result, reduced cell viability was observed in Creꢀinfected
cells expressing both Mdm2 and MdmX (Fig. 6c), and a significant and doseꢀdependent decrease
in cell number was observed in H109ꢀtreated cells lacking Mdm2 (Fig. 6d) or MdmX (Fig. 6e).
On the contrary, nutlinꢀ3a exhibited strong inhibition of the cells expressing Mdm2 (Fig. 6e), but
not these expressing MdmX (Fig. 6d). Cells lacking both Mdm2 and MdmX were largely
unaffected by this treatment (Fig. 6f). Taken together, these results suggested that the nutlin
analogꢀmediated decrease in cell viability was strictly p53ꢀdependent, and that compound H109
more specifically affected MdmX than nutlinꢀ3a.
Discussion
11
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 37
1
2
3
4
5
6
7
8
9
In this study, our segmental mutagenesis experiments showed that different amino acid
sequences differentiated the binding affinity of nutlinꢀ3a and p53p for NꢀMdmX and NꢀMdm2,
which aided our design of a compound library to obtain potent inhibitors of Mdm2/MdmX.
Previously, two crystal structures of NꢀMdm2 in complex with nutlin compounds were reported,
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
one of which was obtained using a surface entropy reduction mutant (NꢀMdm2E68A+K69A
,
4hg7.pdb)25 and another obtained from NꢀMdm2 complexed with a nultinꢀ3a analog (two
chlorophenyl groups were replaced with bromophenyl groups, 1RV1)7. The mutant Mdm2
structure and the structures of the wildꢀtype Mdm2 complexed with nutlinꢀ3a share essentially
the same features, indicating a compact binding pocket for nutlinꢀ3a. Although detailed structural
information on NꢀMdmX/nutlinꢀ3a complexes is not currently available, comparison of the
structures of NꢀMdmX/p53p and NꢀMdm2/p53p complexes as well as the Mdm2/nutlinꢀ3a
structures revealed that the binding pocket of nutlinꢀ3a on NꢀMdm2 is very similar to the p53ꢀ
binding pocket on NꢀMdmX. However, some differences exist between the two pockets. One
obvious observation is that in the ligand binding pocket in NꢀMdmX, the residue Tyr99 (from H2
helix) and the residue Met54 (from H4 helix) block ligand binding when interacting with p53p.
Additionally, the Nꢀterminus of NꢀMdmX H4 helix widens the binding pockets, resulting in two
shallow surfaces in the W23 and L26 pockets of NꢀMdmX that are different from those in Nꢀ
Mdm2. These features were also observed in the crystallographic structure of NꢀMdmX
complexed with WW298, a weak inhibitor of MdmX designed by Popowicz et al.26. Taken
together, these results indicate possible “crossꢀtalk” between the W23 and L26 binding pockets27.
Conversely, our ITC data also confirmed that the binding of nutlinꢀ3a to these surfaces on Mdm2
and MdmX exhibited different thermodynamic behaviors. As indicated previously by Popowicz
et al.,28 our data also suggest that there is a high probability that a MdmXꢀspecific inhibitor
12
ACS Paragon Plus Environment

Page 13 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
would exhibit strong affinity for Mdm2. Therefore, our mutational and structural analysis not
only explain why antagonists of Mdm2 that mimic the p53 peptide, such as nutlinꢀ3a,
ineffectively bind to MdmX, but also provides a rationale for designing inhibitors of both
Mdm2/MdmX based on the nutlinꢀ3a scaffold. In this study, we developed a dualꢀspecific
MdmX/Mdm2 inhibitor by accounting for the unique structural properties of MdmX as well as
the common features of Mdm2/MdmX.
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
When considering nutlinꢀ3a as a scaffold, our data suggested that an MdmX inhibitor
should focus on the enhancement of its interaction with the W23 and L26 surfaces as well as the
adjacent region around the Nꢀterminus of helix H4 in NꢀMdmX. In particular, mimicking the
interaction of His95 residue with small molecules should be considered, as it has recently been
used to successfully design a new nutlin analog21. Importantly, the introduction of hydrogen
bonding was determined to be essential for enhancing compound binding in this work. Notably,
that the comparison of a previous NꢀMdm2E68A+K69A/nutlinꢀ3a structure with the current structure
of NꢀMdm2/nutlinꢀ3a suggests that the mutation region of Segment E in NꢀMdmX contains sites
for introducing hydrogen bonding between small molecules and NꢀMdmX. Accordingly, a
nutlinꢀ3aꢀbased compound library of the dual smallꢀmolecule inhibitors of the Mdm2/MdmXꢀ
p53 interaction was designed for initial screening, and as a result, a few highꢀaffinity dual
inhibitors were identified (Fig. 4). Compared with a previous dual Mdm2/MdmX inhibitor that
can efficiently inhibit the p53ꢀMdm2/MdmX interaction by inducing Mdm2/MdmX
dimerization12, two other publications, the current inhibitors bind to two targets individually;
however, the detailed mechanism needs to be verified through structural investigation.
Inhibition of both Mdm2 and MdmX expression by dual peptide inhibitors and antiꢀsense
oligonucleotides has been shown to kill tumors caused by overexpressed MdmX11;12;29. The dual
13
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 37
1
2
3
4
5
6
7
8
9
smallꢀmolecule inhibitors of the Mdm2/MdmXꢀp53 interaction obtained in this work have also
been confirmed to exhibit p53ꢀdependent activity in multiple cellular models, implying that their
antitumor mechanism is derived from the activation of the p53 pathway. Our work strengthens
the notion that reactivating the powerful growthꢀsuppressive and proꢀapoptotic activity of p53 by
inhibiting Mdm2/MdmX is a promising and valuable strategy for treating cancer, but further
investigation is needed to address its true therapeutic potential.
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
Materials and Methods
Cell lines and reagents
HCT116, KAO and H164a cells (ATCC) were cultured in a 37°C incubator with 5% CO2
according to ATCC protocols. Nutlinꢀ3a and nutlinꢀ3b were purchased from Sigma (St. Louis,
USA). Antiꢀp53, antiꢀMdmX, antiꢀβꢀactin, antiꢀrabbit IgGꢀHRP, antiꢀmouse IgGꢀHRP, and
normal mouse/rabbit IgG were products of Santa Cruz Biotechnology (Santa Cruz, USA).
Lipofectamine™ 2000 was a product of Invitrogen.
Protein preparation
The genes encoding the p53ꢀbinding domain of human MdmX (NꢀMdmX, amino acids
22–110) and human Mdm2 (NꢀMdm2, amino acids 22–110) were synthesized with E. coli
optimized codons and subꢀcloned into a modified pET28 plasmid (Novagen), in which the
thrombin cleavage site was replaced with a TEV protease site. The same NꢀMdmX DNA
fragment was also subcloned into a pGEXꢀ5Pꢀ1 vector (GE) to produce a GST fusion protein.
14
ACS Paragon Plus Environment

Page 15 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
The full length of wildꢀtype p53 DNA was subcloned into the aforementioned the pET vector to
produce a HISꢀtag fusion protein.
7
8
9
To construct the segmental mutations, the pET28ꢀMdmX plasmid was used as a PCR
template. For each mutation, a pair of primers was designed so that the 5’ꢀend sequences
matched those in the template. The first batch of PCR reactions was performed individually for
each side of the targeted mutation site with one of the T7 universal primers paring with one of
the designed mutational primers, followed by PCR purification with a PCR purification kit
(Qiagen). The PCR products from the first batch of reactions were used as templates for the
second batch of PCR reactions, for which only two T7 universal primers were used.
Subsequently, the DNA fragments were treated with restriction enzymes and subꢀcloned into the
pET28 vector, which has been preꢀtreated with the same restriction enzymes. All of the
mutations were confirmed with DNA sequencing.
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
Recombinant proteins were prepared in E. coli BL21 (DE3) cells. Cells were grown in
Luria broth and protein expression was induced with 0.4 mM IPTG at 20 °C for 12 h. For
heteronuclear NMR experiments, protein samples were uniformly labeled with 15N in M9
minimal medium with BME vitamins (Sigma) containing 2 g/L (15NH4)2SO4 and 2 g/L[13C6]
glucose as the sole sources of nitrogen and carbon, respectively. Cells were harvested from a 1 L
culture by centrifugation at 8000 x g for 30 min, resuspended in a lysis buffer, and subjected to
periodic sonication for 5 ꢀ 6 min. The lysates were cleared by spinning at 100,000 x g, and the
supernatant was loaded onto a 20 mL NiꢀNTA agarose column (Qiagen). If necessary, the His6ꢀ
tags of the proteins were removed with TEV protease digestion at 4 °C overnight, followed by a
second NiꢀNTA agarose chromatographic purification. The flowꢀthrough was concentrated to
15
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 37
1
2
3
4
10ꢀ12 ml, followed by purification with a S200 gel filtration column (GE). Peak fractions were
5
6
7
combined and concentrated to 1 mg/mL, and freshly frozen in liquid nitrogen, and kept at ꢀ80 °C.
8
9
GSTꢀNꢀMdmX was purified with GST beads and gel exclusion chromatography. Peak
fractions were combined and concentrated to 1 mg/mL, freshly frozen in liquid nitrogen, kept at ꢀ
80 °C.
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
ITC
The proteinꢀligand interactions were characterized using an isothermal titration
microcalorimeter, ITC200 (GE Healthcare/MicroCal), at 25°C. A typical experiment included
the injection of 19 aliquots (2.1 ꢁL each) containing approximately 0.2 mM ligand solution into
a protein solution of approximately 10 ꢀ 20 µM in the ITC cell (volume approximately 200 ꢁL).
An additional set of injections was run in a separate experiment with buffer in the cell instead of
the protein solution as a control. Before data analysis, the control values were subtracted from
the main experimental data.
The binding isotherms were integrated to give the enthalpy change (ꢁH) plotted as a
function of the molar ratio of the ligand. When necessary, prior to the integration procedures, the
baseline was manually adjusted to minimize the background noise. The initial titration point was
always discarded. The ꢁH ∕ molar ratio plot was a sigmoidal, representing the fractional
saturation of the binding sites by the ligand. The Origin 7.0ꢀbased software provided by
GE/MicroCal was used for data analysis, and the one set of sites model was used as the basic
option. The association constant Ka (1/Kd) was determined from the slope of the central linear
part of the fractional saturation curve. The Gibbs free energy change (ꢂG) and the entropy
16
ACS Paragon Plus Environment

Page 17 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
change (ꢂS) were calculated based on the following equations: ꢂG = −RTlnKa = ꢂH − TꢂS,
where ꢂH was derived from the original ꢂH∕molar ratio plots.
7
8
9
Crystallization and protein structure determination
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
To prepare protein/ligand complexes for crystallization, frozen NꢀMdm2 was thawed and
bufferꢀexchanged using a Centricon concentrator (3000 MW cutoff) to 20 mM Tris, pH 7.0, 200
mM NaCl, 2 mM DTT. The complex was then formed by combining the protein with inhibitor at
a molar ratio of 1:3ꢀ5, respectively, and allowing overnight incubation at 4ºC, followed by
concentrating to 12ꢀ15 mg/mL. Crystallization were assessed with Qiagen kits and optimized
using the hanging drop vapor diffusion method. The reservoir solution contained 18% saturated
ammonium sulfate, 100 mM Tris (pH 7.0). Rectangular crystals grew in 1 week and could be
cryopreserved by brief exposure to a solution containing equal volumes of the crystal growth
reservoir and 25% ethylene glycol, followed by introduction into liquid nitrogen.
Cryopreserved crystals were transported to beamline X8C at the National Synchrotron
Light Source at Brookhaven National Laboratory. The data were collected at 100ºK with a
Quantum 4 detector (ADSC) at resolution of 1.3 Å and processed using HKL2000. The crystal
used for structure solution was in space group C2 with cell dimensions a = 121.21 Å, b = 38.43
Å, c = 67.87 Å and ß = 112.34 º. Overall, the dataset was 98.2% complete, had a redundancy of
3.4 and merged well with R = 0.055. Structure solution was achieved using the program
MOLREP from the CCP4 suite and a search model based on the published NꢀMdm2 structure
(1ycr.pdb)18. Structure presentations were prepared using PyMol software.
NMR experiments
17
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 37
1
2
3
4
All NMR spectra were collected at 25°C on a Bruker Avance 600 MHz spectrometer
5
6
7
15
equipped with a tripleꢀresonance pulsedꢀfield gradient probe. Nꢀ1H HSQC NMR spectra were
recorded in the StatesꢀTPPI mode for quadrature detection30. All the NMR samples were
prepared in a buffer containing 20 mM sodium phosphate, 200 mM NaCl, 2 mM DTT, 0.02%
NaN3, and 95% H2O/5% D2O at pH 6.0. Protein samples were prepared through buffer exchange
with the NMR buffer. The final concentration of the protein complex was approximately 0.4 mM.
All datasets were acquired with 2,048 complex points in t2 and 128 complex points in t1. All
datasets were processed using NMRPipe. Spectral display, assignments, and analysis were
performed using the NMRView software package. Time domain data were zeroꢀfilled once and
apodized with a 90ꢀshifted cosineꢀsquared window function. Solvent suppression was achieved
by postꢀacquisition convolution of the time domain data.
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
The heteronuclear NOE values were measured at 800 MHz for 15Nꢀlabeled protein
complexed with nutlinꢀ3a using the pulse sequences described by Farrow et al31. A 5s recycle
delay was used for the heteronuclear NOE experiment. Steadyꢀstate heteronuclear NOE data
were obtained in an interleaved manner with and without proton presaturation (3s). Errors in the
heteronuclear NOE values were estimated from the rootꢀmeanꢀsquare variation of noise in empty
regions of the two spectra as previously described 32.
Preparation of the combinatorial library
Synthesis of nutlin analogs was performed using the procedure developed by Davis et al33
via the development of a diastereoꢀ and enantioselective bisamidineꢀcatalyzed aryl nitromethane
addition to an azomethine. A general procedure to synthesize nutlinꢀ3a analogues is described as
follows.
18
ACS Paragon Plus Environment

Page 19 of 37
Journal of the American Chemical Society
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
All reactions used 1.1 equivalents of nitroalkane in toluene (0.1 M) with a 18–26 h
reaction time unless otherwise noted. If the R1, R2, or R3 group contained a carboxyl or amine
subgroup, it was protected and deprotected with standard procedures.
The reaction was monitored by LCꢀMS until conversion was complete. The reaction was
quenched with water. The aqueous layer was extracted with dichloromethane. The organic layers
were combined, dried over sodium sulfate, and concentrated to give the crude product.
Purification was performed on a Waters reverseꢀphase HPLC (C18 column, mobile phase: water
with 0.1 % formic acid, and methanol with 0.1% formic acid) and was further separated by SFC
(ODꢀH column) in order to give pure enantiomers.
FP assays
19
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 20 of 37
1
2
3
4
5
6
7
8
9
FP assays were conducted in assay buffer containing 10 mM Tris (pH 8.0), 200 mM
NaCl, and 0.01% Tweenꢀ20. The p53p peptide (amino acids 15–29) was labeled with fluorescein,
(fluoresceinꢀGSGSSQETFSDLWKLLPEN, Fluꢀp53p), and a mutant p53 peptide (fluoresceinꢀ
GSGSSQETASDLAKLAPEN, Fluꢀp53pAAA) was used as the negative control.
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
The FP assays were performed with 15 nM fluorescein and 1 µM NꢀMdmX. For the Nꢀ
MdmX/peptide inhibitor assay, nutlin analogs were preincubated with the protein for 30 min.
Each labeled peptide was then added and incubated for 30 min. FP assays were conducted in
384ꢀwell black microplates (Corning). The FP assays were analyzed using an EnVision
multilabel plate reader with a 555 nm excitation filter and a 632 nm static and polarized filter.
The unlabeled competitor peptide and nutlinꢀ3a were used as positive controls, and the alanineꢀ
substituted p53 peptide (p53pAAA) was used as a negative control.
Computational docking
Potent dual inhibitors selected from the library were generated with ChemBioDraw 13
and further energetically minimized with ChemBio3D (Perkin Elmer). The minimized structures
were exported in pdb format and manually evaluated using the AutoDock 4 program34. The
receptor grid box was set directly on the Mdm2 crystal structure obtained in this study after
extraction of the included nutlinꢀ3a molecule.
Real-time PCR
Cancer cells with wildꢀtype p53 (HCT116, RKO, and H460a) were treated with H109 for
8 h, and the change in the level of transcription was measured by quantitative PCR and expressed
as fold induction compared with the untreated control. Total RNA was prepared, reverse
20
ACS Paragon Plus Environment

Page 21 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
transcribed and subjected to realꢀtime PCR assays with the primers for p21 and p53. RTꢀPCR
experiments were performed using the ABI7900HT system (Applied Biosystems). Primers and
probes were designed using Primer Express software (Applied Biosystems). TaqMan probes
were synthesized with 5′ 6ꢀcarboxyfluorescein and 3′ black hole quencher. RNA was prepared
using TRIzol reagent, and cDNA was synthesized using the SuperScript system (Invitrogen,
Carlsbad, CA).
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
GST-pull-down assays and western blotting
A GST pullꢀdown experiment was employed to reconstitute the twoꢀcomponent
molecular complexes formed by NꢀMdmX and p53. GSTꢀNꢀMdmX bound to glutathione
agarose beads incubated with the wildꢀtype p53 protein (1:3 molar ratio) in the binding buffer
(50 mM sodium phosphate, pH 6.8) for 2 h at 4°C in the absence or presence of compound. The
beads were washed extensively with the binding buffer, and the samples were boiled for 10 min
in the SDSꢀPAGE sample buffer.
Proteins were separated by SDSꢀPAGE and electrically transferred to a polyvinylidene
difluoride membrane (Millipore, USA). After blocking in 5% skim milk in tris buffered saline
(TBS), the membranes were probed with specific primary antibodies overnight at 4°C, washed
three times with TBSꢀTween 20, and incubated with HRPꢀconjugated secondary antibodies at
room temperature for 1 h. Then, the membranes were washed with TBSꢀTween 20 and the
protein bands were visualized using enhanced chemiluminescence (ECL) reagents (Pierce
Chemical Co., USA).
MTT assays
21
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 22 of 37
1
2
3
4
5
6
7
8
9
MEF preparation and adenoviral infections were performed in a mouse model according
to the procedures described by Laurie et al16. All animal studies were performed in accordance
with the guidelines approved by the Institutional Animal Care and Use Committee
(IACUC). Briefly, the p53 gene in mice was flanked by loxP sites by use of a transcriptional stop
element (LSL). The allele was transferred into Mdm2 and/or MdmXꢀnull background. MEFs
were prepared from 13ꢀdayꢀold embryos and infected with adenoꢀCreꢀGFP (Vector Development
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
Lab) for 12 h to allow for efficient infection and LSL excision. Cells were grown at 37°C with 5%
CO2 in Dulbecco’s modified Eagle medium supplemented with 10% serum, penicillin, and
streptomycin. For the proliferation assay, cells were plated at a density of 1x104/cm2. After 2 h,
1 ꢁL of 100× nutlinꢀ3a analog, nutlinꢀ3a or nutlinꢀ3b in d6ꢀdimethyl sulfoxide was added and
cells continued to culture for 48 h at 37°C. The plate was set out for 20 min to reach room
temperature, and then 100 ꢁL CellꢀTiterꢀGlo (Promega, USA) was added. The plate was
immediately mixed for 2 min and then was placed in a dark drawer for 15 min. The plate
contents were read on an Envision 2103 multiꢀlabel plate reader (Perkin Elmer). Data were
processed with MicroCal Origin software (v8.0, MicroCal).
Acknowledgments
We thank Dr. David Duda for his excellent technical assistance. This work was partially
funded by the Key Laboratory of Fermentation Engineering (Ministry of Education).
Competing financial interests
The authors declare no competing financial interests.
22
ACS Paragon Plus Environment

Page 23 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
7
Supporting Information Available:
8
9
There are additional results and discussion on thermodynamic characteristics of
segmental mutants binding to p53p and nutlinꢀ3a, as well as thermodynamic profiles of Nꢀ
MdmX mutants in comparison with NꢀMdmX and NꢀMdm2, titrated with p53p and nutlinꢀ3a.
This information is available free of charge via the internet at http://pubs.acs.org.
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
References
(1) Muller, P. A.; Vousden, K. H. Nat.Cell Biol. 2013, 15, 2ꢀ8.
(2) Cheok, C. F.; Verma, C. S.; Baselga, J.; Lane, D. P. Nat.Rev.Clin.Oncol. 2011, 8, 25ꢀ37.
(3) Wade, M.; Li, Y. C.; Wahl, G. M. Nat.Rev.Cancer 2013, 13, 83ꢀ96.
(4) Macchiarulo, A.; Giacche, N.; Carotti, A.; Moretti, F.; Pellicciari, R. Med.Chem.Commun.
2011, 2, 455ꢀ465.
(5) Popowicz, G. M.; Domling, A.; Holak, T. A. Angew.Chem.Int.Ed Engl. 2011, 50, 2680ꢀ
2688.
(6) Zak, K.; Pecak, A.; Rys, B.; Wladyka, B.; Domling, A.; Weber, L.; Holak, T. A.; Dubin,
G. Expert.Opin.Ther.Pat 2013, 23, 425ꢀ448.
(7) Vassilev, L. T.; Vu, B. T.; Graves, B.; Carvajal, D.; Podlaski, F.; Filipovic, Z.; Kong, N.;
Kammlott, U.; Lukacs, C.; Klein, C.; Fotouhi, N.; Liu, E. A. Science 2004, 303, 844ꢀ848.
(8) Carrillo, A. M.; Bouska, A.; Arrate, M. P.; Eischen, C. M. Oncogene 2014,
doi:10.1038/onc.2014.27.
(9) Kruse, J. P.; Gu, W. Cell 2009, 137, 609ꢀ622.
(10) Huang, L.; Yan, Z.; Liao, X.; Li, Y.; Yang, J.; Wang, Z. G.; Zuo, Y.; Kawai, H.; Shadfan,
M.; Ganapathy, S.; Yuan, Z. M. Proc.Natl.Acad.Sci.U.S.A 2011, 108, 12001ꢀ12006.
23
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 24 of 37
1
2
3
4
5
6
(11) Graves, B.; Thompson, T.; Xia, M.; Janson, C.; Lukacs, C.; Deo, D.; Di Lello, P.; Fry, D.;
Garvie, C.; Huang, K. S.; Gao, L.; Tovar, C.; Lovey, A.; Wanner, J.; Vassilev, L. T.
Proc.Natl.Acad.Sci.U.S.A 2012, 109, 11788ꢀ11793.
7
8
9
(12) Chang, Y. S.; Graves, B.; Guerlavais, V.; Tovar, C.; Packman, K.; To, K. H.; Olson, K.
A.; Kesavan, K.; Gangurde, P.; Mukherjee, A.; Baker, T.; Darlak, K.; Elkin, C.; Filipovic,
Z.; Qureshi, F. Z.; Cai, H.; Berry, P.; Feyfant, E.; Shi, X. E.; Horstick, J.; Annis, D. A.;
Manning, A. M.; Fotouhi, N.; Nash, H.; Vassilev, L. T.; Sawyer, T. K.
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
Proc.Natl.Acad.Sci.U.S.A 2013, 110, E3445ꢀE3454.
(13) Garcia, D.; Warr, M. R.; Martins, C. P.; Brown, S. L.; Passegue, E.; Evan, G. I. Genes
Dev. 2011, 25, 1746ꢀ1757.
(14) Zhuang, C.; Miao, Z.; Zhu, L.; Dong, G.; Guo, Z.; Wang, S.; Zhang, Y.; Wu, Y.; Yao, J.;
Sheng, C.; Zhang, W. J Med Chem. 2012 55, 9630ꢀ9642
(15) Blackburn, T. J.; Ahmed, S.; Coxon, C. R; Liu, J.; Lu, X.; Golding, B. T.; Griffin, R. J.;
Hutton, C.; Newell, D. R.; Ojo, S.; Watson, A. F.; Zaytzev, A.; Zhao, Y.; Lunec, J.;
Hardcastle, I. R. Medchemcomm. 2013, 4, 1297ꢀ1304.
(16) Laurie, N. A.; Donovan, S. L.; Shih, C. S.; Zhang, J.; Mills, N.; Fuller, C.; Teunisse, A.;
Lam, S.; Ramos, Y.; Mohan, A.; Johnson, D.; Wilson, M.; RodriguezꢀGalindo, C.;
Quarto, M.; Francoz, S.; Mendrysa, S. M.; Guy, R. K.; Marine, J. C.; Jochemsen, A. G.;
Dyer, M. A. Nature 2006, 444, 61ꢀ66.
(17) Su, Z. D. Symposium on Biomolecular Strucuture, Function, and Dynamics, St.Jude
Children's Research Hospital, Memphis TN 2012.
(18) Kussie, P. H.; Gorina, S.; Marechal, V.; Elenbaas, B.; Moreau, J.; Levine, A. J.; Pavletich,
N. P. Science 1996, 274, 948ꢀ953.
(19) Pazgier, M.; Liu, M.; Zou, G.; Yuan, W.; Li, C.; Li, C.; Li, J.; Monbo, J.; Zella, D.;
Tarasov, S. G.; Lu, W. Proc.Natl.Acad.Sci.U.S.A 2009, 106, 4665ꢀ4670.
(20) Bharatham, N.; Bharatham, K.; Shelat, A. A.; Bashford, D. J.Chem.Inf.Model. 2014, 54,
648ꢀ659.
(21) Michelsen, K.; Jordan, J. B.; Lewis, J.; Long, A. M.; Yang, E.; Rew, Y.; Zhou, J.;
Yakowec, P.; Schnier, P. D.; Huang, X.; Poppe, L. J.Am.Chem.Soc. 2012, 134, 17059ꢀ
17067.
(22) Knox, C.; Law, V.; Jewison, T.; Liu, P.; Ly, S.; Frolkis, A.; Pon, A.; Banco, K.; Mak, C.;
Neveu, V.; Djoumbou, Y.; Eisner, R.; Guo, A. C.; Wishart, D. S. Nucleic Acids Research
2011, 39, D1035ꢀD1041.
(23) Ventura, A.; Kirsch, D. G.; McLaughlin, M. E.; Tuveson, D. A.; Grimm, J.; Lintault, L.;
Newman, J.; Reczek, E. E.; Weissleder, R.; Jacks, T. Nature 2007, 445, 661ꢀ665.
24
ACS Paragon Plus Environment

Page 25 of 37
Journal of the American Chemical Society
1
2
3
4
5
(24) Olive, K. P.; Tuveson, D. A.; Ruhe, Z. C.; Yin, B.; Willis, N. A.; Bronson, R. T.;
Crowley, D.; Jacks, T. Cell 2004, 119, 847ꢀ860.
6
7
8
9
(25) Anil, B.; Riedinger, C.; Endicott, J. A.; Noble, M. E. Acta Crystallogr.D.Biol.Crystallogr.
2013, 69, 1358ꢀ1366.
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
(26) Popowicz, G. M.; Czarna, A.; Holak, T. A. Cell Cycle 2008, 7, 2441ꢀ2443.
(27) Kallen, J.; Goepfert, A.; Blechschmidt, A.; Izaac, A.; Geiser, M.; Tavares, G.; Ramage,
P.; Furet, P.; Masuya, K.; Lisztwan, J. J.Biol.Chem. 2009, 284, 8812ꢀ8821.
(28) Popowicz, G. M.; Czarna, A.; Rothweiler, U.; Szwagierczak, A.; Krajewski, M.; Weber,
L.; Holak, T. A. Cell Cycle 2007, 6, 2386ꢀ2392.
(29) Wade, M.; Wahl, G. M. Mol.Cancer Res. 2009, 7, 1ꢀ11.
(30) Marion, D.; Driscoll, P. C.; Kay, L. E.; Wingfield, P. T.; Bax, A.; Gronenborn, A. M.;
Clore, G. M. Biochemistry 1989, 28, 6150ꢀ6156.
(31) Farrow, N. A.; Zhang, O.; FormanꢀKay, J. D.; Kay, L. E. J.Biomol.NMR 1994, 4, 727ꢀ
734.
(32) Dutta, K.; Shi, H.; CruzꢀChu, E. R.; Kami, K.; Ghose, R. Biochemistry 2004, 43, 8094ꢀ
8106.
(33) Davis, T. A.; Johnston, J. N. Chem.Sci. 2011, 2, 1076ꢀ1079.
(34) Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.; Hart, W. E.; Belew, R. K.;
Olson, A. J. Journal of computational chemistry 1998, 19, 1639ꢀ1662.
25
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 26 of 37
1
2
3
4
Legends for Figures and Tables
5
6
7
8
9
Fig. 1. Segmental mutation analysis of the binding affinity of N-MdmX and N-Mdm2 for
p53p a) Amino acid sequence alignment of NꢀMdmX and NꢀMdm2. The sequences were
annotated according to the translated products of their cDNAs. Their secondary elements are
depicted based on the structure of the NꢀMdmX/p53p complexes (3dab.pdb). The symbols S and
H denote βꢀsheet and αꢀhelical structures, respectively. Nine different segments between Nꢀ
MdmX and NꢀMdm2 are marked alphabetically from A to I. b) NꢀMdmX and NꢀMdm2 showed
similar tertiary structures (green: NꢀMdm2, 1ycr.pdb, and magenta: NꢀMdmX, 3dab.pdb). c) The
thermodynamic signatures of NꢀMdmX, NꢀMdm2 and NꢀMdmX mutants binding to p53p (left)
and nutlinꢀ3a (right) were compared in pairs.
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
Fig. 2. Comparison of the binding surfaces of N-Mdm2 and N-MdmX to p53p and nutlin-
3a. a) An Xꢀray crystal structure of NꢀMdm2 in complex with nutlinꢀ3a was obtained at 1.3 Å in
this work. b) Three binding pockets on NꢀMdm2 were defined by adapting to three key residues
of p53p, L19, W23 and F26 (1ycr.pdb). c) The p53pꢀbinding surface on NꢀMdm2 changed upon
nutlinꢀ3a binding. d) Nutlinꢀ3a on NꢀMdm2 was compared with p53p on NꢀMdmX. Magenta:
nutlinꢀ3a; cyan: p53p. e) The L26 residue of p53p showed a different configurations on the
surfaces of NꢀMdm2 and NꢀMdmX. Cyan: p53p on Mdm2; orange: p53p on NꢀMdmX. f)
Nutlinꢀ3a was docked on the p53p binding surface on NꢀMdmX.
26
ACS Paragon Plus Environment

Page 27 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Fig. 3. The conformational difference between N-MdmX and N-Mdm2 lies in their H4
helices. a) The ligand binding surfaces of NꢀMdmX (green) are formed by the hydrophobic
packing from helices H2, H3 and H4. The side chains of segment H (95HR97K) of NꢀMdm2 were
presented as thin sticks, whereas segment H in mutant NꢀMdmXH is presented as a thick stick
(cyan). The mutation in segment H in the NꢀMdmXH mutant may result in an inability to the Nꢀ
terminal conformation of helical H4 into that of its counterpart in NꢀMdm2. b) The hydrophobic
interaction between the Cꢀterminus of helix H4 (Y’99, Y’103, L’101 and L’106) with helix H2
tightened the NꢀMdm2 conformation, whereas the salt bridge between D100 and R103 in Nꢀ
MdmX may cause the p53p binding surface to become more open to the solvent environment. c)
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
15
The Nꢀ1H HSQC NMR spectrum of NꢀMdmX in complex with nutlinꢀ3a (red) was compared
with that of NꢀMdm2 in complex with nutlinꢀ3a (black). d) The 15Nꢀ1H heteronuclear NOE
experiments revealed that the backbone dynamics of NꢀMdmX (top panel) and NꢀMdm2 (bottom
panel) were significantly different from each other when complexed with nutlinꢀ3a.
Fig. 4. A dual Mdm2/MdmX inhibitor was obtained by screening a structural biology-
based combinatorial library of nutlin-3a analogs. a) A combinatorial library was constructed
by focusing on three key functional groups, R1, R2 and R3, of the nutlinꢀ3a scaffold. The
annotation of each functional group originated from the nutlinꢀ3a moiety. b) Six primary
compounds were selected by screening the library at IC50 < 300 nM, and compound H109 was
identified as a promising dual inhibitor that bound to both MdmX and Mdm2. c)
Thermodynamic profiles describing the heat flow over the course of NꢀMdmX (left) and Nꢀ
Mdm2 (right) titrated with compound H109, were fitted to a oneꢀsite binding model to obtain the
values for stoichiometry (n), the binding constant (Kd) and enthalpy (ꢂH).
27
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 28 of 37
1
2
3
4
5
6
7
8
9
Fig. 5. Molecular docking and NMR characterization of the interaction between compound
H109 and N-MdmX. a) Molecular docking experiments suggested that compound H109 could
bind to the p53 binding pocket of NꢀMdmX. b) The 15Nꢀ1H HSQC NMR spectrum of NꢀMdmX
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
titrated with the compound in a ratio of 1:2 (red) compared with NꢀMdmX in free form (black). c)
The compound significantly perturbed the HSQC resonance peaks of NꢀMdmX (top panel),
especially for the region between residues 60 and 97. Additionally, the 15Nꢀ1H heteronuclear
NOE data (bottom panel) revealed that the backbone dynamics of NꢀMdmX in the same region
became significantly more rigid upon compound binding.
Fig. 6. Inhibition of Mdm2/MdmX-p53 binding by compound H109 activates the p53
pathway in cancer cells with wild-type p53. a) Treatment with compound H109 induced the
expression of the p21 gene but not the p53 gene. Solid and open symbols indicate p21 and p53
transcription levels, respectively. Triangle: H460a cell; circle: RKO cell and square: HCT116
cell. Data were averaged from three replicates. b) Inhibition of the interaction between NꢀMdmX
and p53 by H109 was assayed using GST pullꢀdown experiments (top panel) in comparison with
nutlinꢀ3a (middle panel) and nutlinꢀ3b (bottom panel). NꢀMdmX and p53 proteins were detected
with their antibodies (Ab) on SDSꢀPAGE gels. c-f) Cell viability of Creꢀinfected cells expressing
Mdm2 and/or MdmX was significantly influenced by H109 compared with nutlinꢀ3a and nutlinꢀ
3b. Empty bar: nutlinꢀ3a (a positive control); grey bar: nutlinꢀ3b (an inactive nutinꢀ3a analog as
a negative control); and, dark bar: H109. Reduced cell viability following the addition of H109
was observed in cells expressing both Mdm2 and MdmX (panel c). H109 induced apoptosis in
cells lacking either Mdm2 or MdmX. Like nutlinꢀ3a, a significant and doseꢀdependent decrease
in cell number was observed in H109ꢀtreated cells lacking MdmX (panel d), and importantly,
similar phenomena were also observed in cells lacking Mdm2 (panel e). In contrast, nutlinꢀ3b
28
ACS Paragon Plus Environment

Page 29 of 37
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
exhibited no effect on the cell viability for all the tested cells. Cells lacking both Mdm2 and
MdmX were largely unaffected by treatment with H109 (panel f). The data represent the mean (±
standard deviation) of three independent experiments.
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
29
ACS Paragon Plus Environment