Identification of Noncompetitive Inhibitors of Cytosolic 5′-Nucleotidase II Using a Fragment-Based Approach

Article abstract of DOI:10.1021/acs.jmedchem.5b01616

We used a combined approach based on fragment-based drug design (FBDD) and in silico methods to design potential inhibitors of the cytosolic 5′-nucleotidase II (cN-II), which has been recognized as an important therapeutic target in hematological cancers. Two subgroups of small compounds (including adenine and biaryl moieties) were identified as cN-II binders and a fragment growing strategy guided by molecular docking was considered. Five compounds induced a strong inhibition of the 5′-nucleotidase activity in vitro, and the most potent ones were characterized as noncompetitive inhibitors. Biological evaluation in cancer cell lines showed synergic effect with selected anticancer drugs. Structural studies using X-ray crystallography lead to the identification of new binding sites for two derivatives and of a new crystal form showing important domain swapping. Altogether, the strategy developed herein allowed identifying new original noncompetitive inhibitors against cN-II that act in a synergistic manner with well-known antitumoral agents.

Full text of DOI:10.1021/acs.jmedchem.5b01616

Subscriber access provided by The University of Liverpool
Article
Identification of non-competitive inhibitors of cytosolic
'-nucleotidase II using a fragment-based approach
5
Zsuzsanna Marton, Rémi Guillon, Isabelle Krimm, Preeti Preeti, Rahila Rahimova,
David Egron, Lars Petter Jordheim, Nushin Aghajari, Charles Dumontet,
Christian Périgaud, Corinne Lionne, Suzanne Peyrottes, and Laurent Chaloin
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.5b01616 • Publication Date (Web): 24 Nov 2015
Downloaded from http://pubs.acs.org on November 27, 2015
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 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Identification of non‐competitive inhibitors of cytosolic 5'‐nucleotidase II using a
fragment‐based approach
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
†,¶
‡,¶
§
†
‡
ꢀ
Zsuzsanna Marton , Rémi Guillon , Isabelle Krimm , Preeti , Rahila Rahimova , David Egron ,


‡
†
ꢀ
Lars P. Jordheim , Nushin Aghajari , Charles Dumontet , Christian Périgaud , Corinne Lionne ,
‡
†,*
Suzanne Peyrottes and Laurent Chaloin
†
Centre d'études d'agents Pathogènes et Biotechnologies pour la Santé (CPBS), FRE 3689 CNRS ‐
Université Montpellier, 1919 route de Mende, 34293 Montpellier cedex 5, France.
‡
Institut des Biomolécules Max Mousseron (IBMM), UMR 5247 CNRS – Université Montpellier ‐ ENSCM,
Campus Triolet, cc1705, Place Eugène Bataillon, 34095 Montpellier cedex 5, France.
§
Institut des Sciences Analytiques, UMR 5280 CNRS, Université Lyon 1, ENS de Lyon, 5 rue de la Doua,
6
9100 Villeurbanne, France.
ꢀ
Institut de Biologie et Chimie des Protéines FR3302, Molecular and Structural Bases of Infectious
Diseases UMR 5086 CNRS – Université Lyon 1, 7 Passage du Vercors, 69367 Lyon, France.

Université Lyon 1, Centre de Recherche en Cancérologie de Lyon, INSERM U1052, CNRS UMR 5286,
Centre Léon Bérard, 69008 Lyon, France.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 59
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
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
We used a combined approach based on Fragment‐Based Drug Design (FBDD) and in silico
methods to design potential inhibitors of the cytosolic 5'‐nucleotidase II (cN‐II), which has been
recognized as an important therapeutic target in hematological cancers. Two sub‐groups of
small compounds (including adenine and bi‐aryl moieties) were identified as cN‐II binders and a
fragment growing strategy guided by molecular docking was considered. Five compounds
induced a strong inhibition of the 5’‐nucleotidase activity in vitro, and the most potent ones
were characterized as non‐competitive inhibitors. Biological evaluation in cancer cell lines
showed synergic effect with selected anticancer drugs. Structural studies using X‐ray
crystallography lead to the identification of new binding sites for two derivatives and of a new
crystal form showing important domain swapping. Altogether, the strategy developed herein
allowed identifying new original non‐competitive inhibitors against cN‐II that act in a synergistic
manner with well‐known antitumoral agents.
INTRODUCTION
To efficiently block an enzymatic function, one approach consists in the development of
structure‐guided inhibitors that will fit in the active binding site or known regulatory sites.
Taking into account that substrate binding sites of enzymes exhibit quite complex shapes,
Fragment‐Based Drug Design (FBDD) represents a powerful strategy for the identification of
1‐5
6
such compounds (for a detailed overview, see the recent review by Chen et al.). Thus,
fragment screening has been increasingly used for generating novel starting‐points in drug
2
ACS Paragon Plus Environment

Page 3 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
discovery programs. The use of small chemical entities (molecular weights beneath 300 Da),
called fragments, facilitates the discovery of drug candidates by taking the advantage of the
high quality interactions of the fragments within binding pockets of a protein, leading to high
ligand efficiencies. In this respect, NMR and X‐ray crystallography are the most frequently used
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
7
screening tools, even though surface plasmon resonance (SPR) and isothermal titration
8
calorimetry (ITC) are also employed. Initially applied to target soluble proteins, the FBBD
9,10
11
approach has now been successfully extended to membrane proteins, and nucleic acids.
Here, we applied the FBDD approach to the design of inhibitors against the 5'‐cytosolic
nucleotidase II (cN‐II). This protein catalyzes the hydrolysis of purine nucleoside 5'‐
monophosphates (NMP) into nucleosides and inorganic phosphate in order to regulate
intracellular nucleotide pools. cN‐II belongs to the large halo‐acid deshydrogenase (HAD) family
including eight 5’‐nucleotidases with various cellular localization and substrate specificity. 5’‐
nucleotidases are also present in bacteria making them attractive targets for drug design as
12,13
previously described using a similar approach.
The human cN‐II is also involved in the
1
4
resistance mechanism of some anti‐cancer treatments based on cytotoxic nucleoside analogs.
Indeed, a high level of cN‐II expression in blasts was found to be predictive of a worse patient
15,16
outcome after cytarabine‐based treatment of acute myeloid leukemia (AML).
Recently,
mutants of cN‐II, with hyperactive enzymatic activity, have been shown to be associated with
17,18
resistance to chemotherapy in relapsed patients with acute lymphoblastic leukemia (ALL).
cN‐II is also known as high‐K nucleotidase II and therefore the inhibitors derived by analogy to
M
1
9‐21
the substrate exhibited elevated inhibition constants (millimolar range).
This enzyme
represents a difficult target for classical approaches of drug design due to both a low affinity for
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 59
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
its substrate and a complex allosteric regulation. At least, two effector sites were identified by
2
2,23
kinetics and crystallography,
showing that cN‐II can be regulated through the binding 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
small regulators such as ATP, BPG or Ap A. In this respect, FBDD approach was applied to cN‐II
4
using the NMR screening of a chemical library of fragments, the building of their assemblies
using molecular modeling, followed by synthesis, in vitro inhibition, biological assays in cancer
cell models and structural studies.
RESULTS
NMR screening of the fragment library
A chemical library composed of 300 fragments was screened by NMR using recombinant
purified human cN‐II. Two NMR experiments were performed (STD and Waterlogsy) with
mixtures of five fragments. From the positive pools, the fragment hits were then tested alone
and only those showing binding signals in both experiments were confirmed as hits (Figure S1,
Supporting information). A relatively high hit rate (10%) was obtained with this ligand‐observed
NMR screening approach, which is known to identify very weak binders. The fragment hits could
be divided into two groups and representative structures are shown in Table 1. The first group
contains aromatic and nitrogen‐containing heterocylces, while the second one contains
scaffolds reported in the literature as frequent‐hitter privileged binding motives, such as the bi‐
phenyl motif.
4
ACS Paragon Plus Environment

Page 5 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 1: Representative fragments from the two groups identified as cN‐II binders by NMR
screening. FSTD indicates the amplification factor measured by STD and fragment F4 is shown as
a 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
Inhibition at
Inhibition at
1 mM (%)
Fragment #
Structure
F
STD
Fragment #
Structure
Group n°2
F
STD
1
mM (%)
Group n°1
Cl
N
HO
OH
N
N
F1
0.5
23 ± 5
F8
H2
N
NH2
<0.5
23 ± 4
N
H
NH
2
NH
2
HO
OH
F2
F3
F4
F5
F6
3
6
30 ± 17
35 ± 3
0.5 ± 1
11 ± 9
36 ± 5
F9
6.6
2.7
4.2
6.4
3.1
40 ± 6
19 ± 3
36 ± 5
19 ± 1
2 ± 1
N
O
N
NH
2
F10
F11
F12
F13
N
H
2
N
NH
2
2
H N
N
HN
N
0
N
HO
OH
5.5
4.3
HO
N
H
HN
N
NH
2
N
N
HO
OH
F7
<0.5
19 ± 1
F14
7.2
8 ± 2
N
H
N
The latter are highly hydrophobic and no consensus was observed for the chemical functions
attached to the aromatic rings. The first group contained purine‐based fragments with low STD
factors (derived from F1 or F7) and a wide range of nitrogen‐based heterocycles, such as
quinoline (F2), “phenyl‐pyrazinyl” (F3) and “phenyl‐imidazole” (F6) scaffolds with FSTD values
ranging from 3 to 6 (Table 1). Furthermore, the heterocyclic ring was preferentially aromatic, as
fragment F4 was not detected as a ligand in contrast to F3 (therefore, F4 was not selected for
the next steps). Interestingly, 1‐H‐benzimidazol‐2‐yl‐methanol (F5), structurally closed to a
purine scaffold, was also found to bind cN‐II. The second group is composed of hydrophobic
derivatives, mainly containing bi‐aryl (for instance, F10, F12 and F14) and naphthalene (F9)
scaffolds with FSTD values between 2.7 and 7.2. This suggests the presence of a hydrophobic
pocket where they may bind. In addition, we noticed that the presence of a polar group as
substituent drastically changes the STD signal and therefore, the binding strength for these bi‐
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 59
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
aryl fragments. Indeed, hydroxyl groups were more favored (F12 or F14) than amino groups
(F10 or F11) and even more preferable than the combination of these (F8). Competition
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
experiments with fragments could not be performed due to the lack of an identified competitor,
as even IMP and inosine appear as very weak ligands (IMP is rapidly hydrolyzed into inosine,
which is known to have a very weak affinity of 3‐4 mM). Nevertheless, INPHARMA experiments
were performed to assess whether the fragments share the same binding site as AdiS, an
24
anthraquinone‐like cN‐II inhibitor previously identified by virtual screening. Its affinity for cN‐II
was determined by kinetics by means of its K value of 2 ± 0.2 mM. As illustrated in Figure S2
i
(
Supporting information), interligand NOEs were observed between fragments and AdiS. This
result constitutes an experimental evidence that these two fragments (F3 and F10) are able to
bind into the IMP binding site of the enzyme since the AdiS inhibitor was previously
characterized as a competitive inhibitor.
cN‐II inhibition by hit fragments
Fragments identified by ligand‐observed NMR experiments were further evaluated by
enzymatic inhibition assays using the purified recombinant enzyme in order to validate the most
interesting fragments. Most of them were able to inhibit hydrolysis of IMP when used at high
concentrations. However, due to solubility issues for concentrations higher than 2 mM, the IC50
could not be determined and a comparison of their effect at 1 mM was used as indicator of
inhibitory activity (Table 1). The most efficient fragment (40 ± 6% of inhibition) was 2,7‐
dihydroxynaphthalen (F9), followed by bi‐aryl‐related compounds harboring two six‐membered
aromatic rings (≥ 35% of inhibition) such as 5‐phenyl‐2‐pyrazinylamine (F3) and 2‐biphenyl‐
6
ACS Paragon Plus Environment

Page 7 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
amine (F11). This result was confirmed by the moderate effect (~20% of inhibition) observed
with structurally related fragments including a biphenyl or a benzophenone motif (F8, F12, F14
and F10, respectively). Amongst the first group, several fragments were able to significantly
inhibit cN‐II such as adenine (F7) and 2‐amino‐6‐chloropurine (F1). Similarly, inhibition was
observed for the other nitrogen‐containing aromatic rings (F2) and with a weaker effect for F5.
Another interesting substructure in this group was also detected as a cN‐II inhibitor, and
corresponds to the association of a phenyl group and an imidazole moiety (F6), thus displaying a
common chemical feature with the bi‐aryl subgroup. Interestingly, when the two phenyl groups
were separated by freely rotatable bonds such as for compound F13, it was completely inactive.
This suggests that the flexibility and/or the distance between the two rings may be important.
Finally, the most promising fragment in this group was the biphenyl amine (F11) with a single
bond between both aromatic rings. In agreement with its low FSTD value, the fragment F4 was
found to be totally inactive, whereas fragment F8 was partially able to inhibit cN‐II despite its
low FSTD value.
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
In silico binding prediction of the hit fragments
Based on the results of NMR experiments and of the in vitro inhibition assays, the most
promising fragments were composed of the following scaffolds: “C‐6‐substituted purine”,
“naphtalene”, “bi‐aryl”, “phenyl‐imidazole” and “quinoline”. In order to guide the assembly,
merging or growing process, a molecular docking study (Figure 1) was performed in the enzyme
active site (as INPHARMA experiments indicated that the fragments bind to the IMP site).
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 59
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 1: Molecular docking of the most representative hit fragments on the cN‐II crystal
structure (2XCW, shown as solvent accessible surface with the magnesium ion depicted as a
green sphere). Superimposition of docking poses obtained with adenine (or F7 in stick
8
ACS Paragon Plus Environment

Page 9 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
representation and carbon atoms in magenta) and (A) F2 (quinoline, green) or (B) F9
(naphthalene, light grey) showing overlapping areas. Overlay of the binding poses obtained with
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
adenine (magenta) and (C) F6 (light green) or (D) F11 (dark blue) fragments showing two distinct
binding areas. (E) N‐6‐benzyl‐adenine (F15, yellow) and (F) N‐6‐benzoyl‐adenine (F16, orange)
binding poses.
Steric hindrance was predicted between adenine (F7) and quinoline (F2) or naphthalene
(F9) fragments (Figure 1A and B). A partial overlap was also observed with the 4,4'‐diamino‐
benzophenone (F10, data not shown). In contrast, phenyl‐imidazole (F6) or bi‐aryl (F11)
compounds (Figure 1C and D) were predicted to bind to a different cavity near the adenine site.
The purine fragment was repeatedly found to be deeply inserted near the magnesium ion in
contrast to phenyl or bi‐aryl containing fragments which appeared to bind to a more exposed
area (usually occupied by inosine). From the docking results, two sub‐pockets were identified in
which adenine and phenyl groups can be accommodated. Therefore, we searched for
commercially available and structurally related compounds that combine these two scaffolds
and evaluated their binding mode. As shown in Figure 1E and F, the binding mode for N‐6‐
benzyl‐adenine (F15) and N‐6‐benzoyl‐adenine (F16) was found to be compatible with the ones
of adenine and phenyl moieties (see structure of F15‐F16 in Figure 2). Moreover, the nature of
the linker between these two scaffolds seems to have a direct effect on the adenine location
(more buried in the case of N‐6‐benzoyl‐adenine). Thus, the most interesting association
appeared to be N‐6‐benzoyl‐adenine (F16). These two compounds were assayed by NMR and
both were detected as cN‐II ligands using STD and Waterlogsy experiments as illustrated in
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 59
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
Figure 2. The FSTD values for the protons are related to their solvent accessibility: the lower the
F
STD value, the higher the solvent accessibility of the proton upon binding to the protein target.
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 2: Per‐proton FSTD percentages for the two larger fragments, N‐6‐benzyl‐adenine or F15
A) and N‐6‐benzoyl‐adenine or F16 (B) showing the different solvent accessibility of the protons
(
(highest STD percentages indicate protons deeply buried in the protein binding site).
If we analyze the docking pose of F16 (Figure 1F), and compare with FSTD values, the
proton in position C8 (68%) of the adenine moiety is solvent exposed as compared to the proton
in position C2 (100%). Regarding the phenyl ring, the proton in para position (100%) is more
buried than the other ones (80 or 88%). These comparisons between docking and STD data are
in perfect agreement. For compound F15, due to spectral overlap, the difference of accessibility
of the various phenyl protons could not be measured accurately but the proton in position C8
(68%) of the adenine moiety is observed as solvent exposed while the proton in position C2
(100%) is buried (Figure 2). Thus, N‐6‐benzyl‐adenine (F15) and N‐6‐benzoyl‐adenine (F16) were
10
ACS Paragon Plus Environment

Page 11 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
further evaluated for their cN‐II inhibition capacity. Both compounds exhibited a moderated
inhibition of cN‐II activity (20 ± 6 and 28 ± 2% of inhibition at 1 mM for F15 and F16,
respectively), confirming our docking‐guided hypothesis.
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
Synthesis of the targeted compounds
According to all data obtained from NMR binding, in vitro activity, molecular docking and
group epitope mapping, a generic chemical structure was proposed based on the most
interesting fragments (Figure 3). We explore the chemical diversity through the relative position
of the two phenyl groups (ortho, meta and para or fused), as well as the position and the nature
of the five‐membered nitrogen‐containing heterocycle.
H N
2
NH2
N
N
F11
N
N
O
N
X
H
X
F7
X X
HN
X
N
X
NH
N
H
N
O
X= N or C
N
N
F6
NH2
HN
N
N
N
F3
N
N
H
F16
Figure 3: Retained strategy for fragment growing starting from adenine and linking with two
merged structural scaffolds (bi‐aryl and phenyl‐imidazole) as shown by the three circles.
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 59
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
This strategy for chemical synthesis allowed both to focus on the major sub‐groups
previously identified and to include diversity in the merging process. In this respect, final
compounds may incorporate N‐ or C‐branched imidazole as well as pyrrole moieties (Table 2).
This latter five‐membered aromatic ring is a well known pharmacophore with remarkable
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
25
biological properties, and is representative of the numerous nitrogen‐containing heterocycles
found as hit fragments. Thus, the proposed synthetic pathway (Figure S3, Supporting
Information) would allow the sequential introduction of the various scaffolds previously
suggested. The aryl‐aryl bond formation was performed by traditional palladium catalyzed
Suzuki cross‐coupling reaction. The introduction of the pyrrole or the imidazole was envisaged
26,27
through N‐arylation,
whereas the C4‐imidazole analogue could be obtained in a single step.
Thus, bi‐aryl carboxamide derivatives (compounds 1a‐e, 2a‐f, 3a‐d and 4a‐d, Scheme 1A) were
obtained by coupling adenine with commercially available biphenyl carboxylic acid 5a, carbonyl
chlorides 5b‐e or substituted biphenyl carboxylic acids 6a‐f, 7a‐d and 8a‐d. For the carboxylic
acids coupling step, an excess of N,N’‐carbonyldiimidazole (CDI) was used as activation reagent
in presence of a catalytic amount of 4‐dimethylaminopyridine (DMAP) in anhydrous DMF at 100
°C until starting material was consumed (HPLC monitoring). In the case of the imidazole
derivatives 8a‐d, these coupling conditions led to simultaneous loss of the monomethoxytrityl
(MMTr) protecting group. Substituted biphenyl carboxylic acids 6a‐f (Scheme 1B), 7a‐d and 8a‐d
(Scheme 2) were obtained beforehand using Suzuki coupling of the corresponding boronic acid
pinacol esters 11a,b or commercially available boronic acid ethyl esters with substituted aryl
bromides or derivatives 13a,b and 17a,b.
12
ACS Paragon Plus Environment

Page 13 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
O
HN
N
O
A
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
(i)
N
R1
1a-c
N
H
N
5
a ortho R = OH
1
5
b meta R = Cl
O
1
1
5c para R = Cl
NH2
N
HN
O
N
(
i)
N
N
+
Cl
1
d-e
N
H
N
N
H
N
5
d ortho
e meta
5
O
X
Y
HN
N
Z
O
X
Y
(i)
N
N
HO
Z
2a-f X = N ; Y = Z = C
3a-d X = Z = N ; Y = C
4a-d X = C ; Y = N ; Z = NH
N
H
6
7
a-f X = N ; Y = Z = C
a-d X = Z = N ; Y = C
8a-d X = C ; Y = N ; Z = N-MMTr
B
N
N
I
N
N
O
(
ii)
(iii)
(iv)
(v)
6a-f
Br
Br
B
O
CO Me
CO
H
2
9
9
a meta
b para
10a meta
10b para
2
1
1
1
1
1
2a CO Me ortho / pyrrole meta
2
2b CO Me ortho / pyrrole para
2
1
1a meta
11b para
2c CO Me meta / pyrrole meta
2
2d CO Me meta / pyrrole para
2e CO Me para / pyrrole meta
12f CO Me para / pyrrole para
2
2
2
Scheme 1. Synthesis of targeted derivatives 1a‐e, 2a‐f, 3a‐d and 4a‐d (A) and intermediates 6a‐f
(B). Reagents and conditions: (i) CDI, DMAP, DMF, 100 °C, 39 h‐6.6 days (for R = OH); or
pyridine, 100 °C, 3 h (for R = Cl); (ii) pyrrole, CuCl (5 mol%), nBu NOH (40% aq.), 80 °C, 36 h; (iii)
1
1
4
bis(pinacolato)diboron, Pd (dba) , 2‐(dicyclohexylphosphino)‐2',4',6'‐triisopropylbiphenyl, AcOK,
2
3
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 59
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
DMF, 110 °C, 2 h; (iv) methylbromobenzoate, Pd
2
(dba)
3
, K
2
CO
3
, DMF, 100 °C, 18 h; (v) NaOH 2M,
MeOH, 1,4‐dioxane, 50 °C, 1‐4 h (for 12a‐e); t‐BuOK, Et
2
O, H
2
O, rt, 3 h (for 12f).
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
28,29
The conditions described by Buchwald and co‐workers,
were applied using potassium
carbonate as base and tris(dibenzylideneacetone)dipalladium or tetrakis(triphenylphosphine)
palladium as catalyst. A final treatment with aqueous NaOH in ethanol led to the saponification
of the ester functions. For compounds containing the C4‐imidazole heterocycle (Scheme 2), the
nitrogen atom in position 1 of the 4‐(bromophenyl)‐1H‐imidazole derivatives 16a and 16b was
protected with a momomethoxytrityl group, leading in 96‐98% yields to derivatives 17a and
1
7b, respectively. In contrast to the 4‐(4‐bromophenyl)‐1H‐imidazole 16b which is commercially
available, the imidazole moiety in meta position of compound 16a was constructed from ‐
bromophenone derivative 15 treated by formamide according to an adapted procedure from
3
0,31
the literature.
14
ACS Paragon Plus Environment

Page 15 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A
N
N
N
N
N
I
N
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
(
i)
(ii)
(iii)
Br
Br
9
a meta
b para
13a meta
13b para
CO Et
CO H
9
2
2
1
1
4a CO Et ortho : imidazole meta
7a CO
2
H ortho : imidazole meta
2
2
2
2
4b CO Et ortho : imidazole para
7b CO
2
H ortho : imidazole para
H meta : imidazole meta
2
14c CO Et meta : imidazole meta
4d CO Et meta : imidazole para
7c CO
7d CO
1
2
H meta : imidazole para
B
MMTr
MMTr
N
N
MMTr
N
H
N
N
N
O
Br
Br
N
N
(
iv)
(v)
(vi)
(vii)
Br
Br
CO
Et
CO H
2
2
1
5
16a meta
6b para
commercialy
available)
17a meta 18a CO Et ortho : imidazole meta 8a CO H ortho ; imidazole meta
2
2
2
1
17b para 18b CO Et ortho : imidazole para 8b CO
H ortho ; imidazole para
H meta ; imidazole meta
d CO H meta ; imidazole para
2
(
18c CO
Et meta : imidazole meta 8c CO
2
2
18d CO
Et meta : imidazole para
8
2
2
Scheme 2. Synthesis of intermediates 7a‐d and 8a‐d containing imidazole as heterocycle.
Reagents and conditions: (i) Imidazole, CuI (10 mol%), benzotriazole (20 mol%), t‐BuOK, DMSO,
1
00 °C, 14 h; (ii) ethoxycarbonylbenzeneboronic acid, Pd(PPh
NaOH 2M, EtOH, 1,4‐dioxane, 50 °C, 2‐6 h; (iv) NH CHO, 170 °C, 14 h; (v) 4‐monomethoxytrityl
chloride, Et N, DMF, rt, 2 h; (vi) ethoxycarbonylbenzeneboronic acid, Pd(PPh , K CO , DMF,
00 °C, 2‐3 h; (vii) NaOH 2M, EtOH, 1,4‐dioxane, 50 °C, 1‐10 h.
3
)
4
, K
2
CO , DMF, 100°C, 3‐5 h; (iii)
3
2
3
3
)
4
2
3
1
cN‐II inhibition by newly synthesized compounds
The chemical diversity obtained by using innovative synthesis pathways allowed us to evaluate
most of the possible assemblies according to before‐mentioned criteria. All new compounds
were evaluated for their capacity to inhibit nucleotidase activity using the rapid green malachite
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 59
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
assay (Table 2). Interestingly, as soon as two fragments were linked together (bi‐aryl and purine
moieties) a strong inhibition was observed such as for the meta‐compound 1b with 68% at 0.2
mM. This result is in agreement with the docking prediction and the STD spectrum of this
compound (Figure S4, Supporting Information). Indeed, compound 1b was found to be deeply
inserted into the IMP binding site and this prediction was confirmed by STD data showing that
aromatic protons of the second phenyl are more buried (protons at position C2’, C4’ and C6’).
The relative orientation of the second phenyl ring was found to be decisive as compound 1a
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
(
ortho) was almost inactive at this concentration (17% inhibition) while para (1c) position was
more advantageous with 57% inhibition. The effect of the addition of a third group (either
pyrrole or imidazole) was also dependent on the orientation related to the second phenyl ring.
As some compounds showed inhibitory activity after addition of the pyrrole group in positions
meta (2a, 73% inhibition) and para (2b, 85% inhibition) whereas the corresponding
unsubstituted compound was inactive (1a, phenyl in ortho). In contrast, when the second
phenyl ring was oriented in meta or para position, the addition of either N‐pyrrole or N‐/C4‐
imidazole did not improve the efficacy of the final compound when compared to compound 1b
(assembly of two fragments), except for derivative 2c which inhibited 56% of the cN‐II activity.
Compound 2f, for which the orientation of phenyl and pyrrole rings is in para/para, was
depicted as the less efficient compound with no activity detected at 0.2 mM. This result may
indicate the limited space or volume available in the enzyme binding site as this particular
derivative may be viewed as an extended scaffold. For N‐ or C4‐imidazole‐derived compounds
(3a‐d and 4a‐d), most of them were found to be inactive or weak inhibitors against cN‐II (3c, 4a,
4
c and 4d with 39%, 35%, 25% and 26% inhibition, respectively). This result was surprising
16
ACS Paragon Plus Environment

Page 17 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
according to the high structural similarity between imidazole and pyrrole. However, these two
five‐membered nitrogen‐containing heterocycles share common features but differ in water
solubility and acidity (pKa). Focusing on the most active compounds, the inhibition was further
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
confirmed by steady state kinetics (Figure 4). Inhibition constants (K ) were only determined for
i
soluble compounds and were in agreement with the inhibition values observed at low
concentration (0.2 mM). Inhibition constants ranged from 0.5 mM for 2b to 2.8 mM for 1c and
are promising for further lead optimization, as previously identified competitive inhibitors
showed higher K
c) were characterized as competitive inhibitors, derivatives based on the three‐fragment
assembly (2a‐c) exhibited a non‐competitive inhibition mode. The most potent compound (2b),
with a K value of 0.46 mM, displayed a typical non‐competitive inhibition kinetic. To illustrate
i
(≥ 1‐2 mM). While the compounds based on a two‐fragment assembly (1b and
1
i
such difference in inhibition mode, the kinetics of compounds 1b, 2a and 2b are presented in
Figure 4 with secondary and double‐reciprocal plots. The initial prediction was carried out
assuming that IMP was the targeted binding site for these inhibitors. However, structural
analogies between this site and the regulatory sites (also binding nucleotides) may be
responsible of this selection of alternative binding sites for compounds 2a‐c.
Table 2: Structures and inhibitory activities of bi‐aryl carboxamide‐N‐(9H‐purin‐6‐yl) derivatives
(
AdiS, a previously identified compound is shown as positive control).
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 59
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
O
O
N
O
HN
HN
N
HN
N
Het
N
N
N
N
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
N
H
N
HN
N
H
N
H
N
N
N
1a-c
1d-e
N
N
N-pyrrole
4
N-imidazole C -imidazole
2a-f
3a-d
4a-d
Compound Adenine /
Substituent /
Inhibition (%)
at 0.2 mM
20 ± 2%
K (mM)
i
#
orientation
orientation
(Inhibition mode)*
2 ± 0.2
AdiS
‐
‐
‐
‐
1
a
ortho
meta
17 ± 1%
n.d.**
1b
68 ± 5%
0.95 ± 0.17
(
(
Competitive)
2.78 ± 0.61
Competitive)
n.d.
1
c
para
‐
57 ± 3%
1d
ortho
meta
ortho
‐
‐
40 ± 5%
30 ± 5%
73 ± 5%
1
e
n.d.
2a
N‐pyrrole / meta
N‐pyrrole / para
N‐pyrrole / meta
1.65 ± 0.10
(
(
(
Non‐competitive)
0.46 ± 0.03
2
b
ortho
meta
85 ± 4%
Non‐competitive)
0.79 ± 0.05
2c
56 ± 13%
Non‐competitive)
n.d.
2
d
meta
para
N‐pyrrole / para
N‐pyrrole / meta
28 ± 1%
0%
2e
n.d.
18
ACS Paragon Plus Environment

Page 19 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
f
para
ortho
ortho
meta
meta
ortho
ortho
meta
meta
N‐pyrrole / para
N‐imidazole / meta
N‐imidazole / para
N‐imidazole / meta
N‐imidazole / para
0%
n.s.***
n.d.
n.s.
3
a
b
13 ± 3%
0%
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
3
c
39 ± 7%
8 ± 8%
35 ± 5%
3 ± 3%
25 ± 5%
26 ± 5%
n.d.
n.s.
3d
4a
4b
C
4
‐imidazole / meta
n.d.
n.d.
n.s.
C
4
‐imidazole / para
4
c
C
4
‐imidazole / meta
4
d
C
4
‐imidazole / para
n.s.
*
i
Values of inhibition and K are means from three independent experiments at different inhibitor concentrations ±
SD. **n.d., not determined. ***n.s., not soluble at concentrations above 0.2 mM)
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 59
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 4: Secondary plots and double reciprocal representations of steady state rate constants
as a function of substrate concentrations (IMP), for the most interesting compounds. (A) in the
absence () or with 1b at 0.5 mM () or 1 mM (). (B) in the absence () or in the presence
20
ACS Paragon Plus Environment

Page 21 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
of 2a at 0.4 mM () or 0.8 mM (). (C) in the absence () or in the presence of 2b at 0.2 mM
(
) or 0.8 mM ().
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 of newly synthesized compounds
All derivatives were evaluated for their antiproliferative activity on cancer cells. IC50
values as determined by the MTT assay on human follicular lymphoma cells (RL) ranged from 4.7
to 215 µM (Table 3). For compounds with IC50 lower than 100 µM, their synergistic activities
with nucleoside analogues were evaluated. This is based on the assumption that cN‐II inhibitors
should act synergistically with cytotoxic nucleoside analogues as was shown with inhibitors
issued from virtual screening and by sensitization of cancer cells with low expression of cN‐
2
4,32,33
II.
Synergy was observed with several combinations with cladribine and clofarabine as for
example compound 1b (CI95 = 0.8 and 0.8), compound 2d (CI95 = 0.8 and 0.7), compound 2f
(CI95 = 0.5 and 0.7), compound 3c (CI95 = 0.8 and 0.7) and compound 4c (CI95 = 0.9 and 0.6).
However, no synergy was detected with fludarabine (compound 2d being excepted with the
limit value of 0.9), suggesting a distinct metabolism for this compound. This difference with
fludarabine was also observed with AdiS, a first‐generation cN‐II inhibitor identified by virtual
24
screening.
Table 3: Biological activity of the newly synthesized compounds on RL cells. IC50 values (µM) and
CI95 values are mean values of at least three independent experiments ± SEM. Compounds for
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 59
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
which IC50 value was below 100 µM are highlighted in light grey and their synergistic activities in
dark grey (AdiS is a previously identified compound by virtual screening).
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
Compound # /
IC50
CI95
Fludarabine Cladribine Clofarabine
0.4 ± 0.1 0.24 ±.0.03
CI95
CI95
AdiS 1036 ± 267 4.2 ± 1.7
1
a
165 ± 41
25 ± 4
n.d.
1.2 ± 0.2
n.d.
1.1 ± 0.2
0.8 ± 0.1
n.d.
n.d.
0.8 ± 0.1
n.d.
1
b
1
c
157 ± 38
>300
1d
n.d.
n.d.
n.d.
1
e
a
240 ± 31
23 ± 9
n.d.
n.d.
n.d.
2
n.d.
1.2 ± 0.8
1.4 ± 0.1
1.4 ± 0.2
n.d.
2b
35 ± 3
1.5 ± 0.2
1.4 ± 0.3
1.0 ± 0.1
2.9 ± 1.3
2
c
11 ± 2
2
d
e
8 ± 2
0.9 ± 0.1 0.8 ± 0.1 0.7 ± 0.2
2
398 ± 62
87 ± 22
168 ± 32
51 ± 9
n.d.
1.3 ± 0.3
n.d.
n.d.
0.5 ± 0.2
n.d.
n.d.
0.7 ± 0.1
n.d.
2
f
3a
3
b
1.1 ± 0.1
1.9 ± 0.8
n.d.
1.2 ± 0.2
0.8 ± 0.2
n.d.
1.8 ± 1.0
0.7 ± 0.2
n.d.
3
c
34 ± 5
3d
203 ± 20
215 ± 8
107 ± 3
4
a
n.d.
n.d.
n.d.
4
b
n.d.
n.d.
n.d.
22
ACS Paragon Plus Environment

Page 23 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
4
c
30 ± 3
5 ± 1
2.9 ± 1.7
3.7 ± 2.5
0.9 ± 0.2
1.2 ± 0.3
0.6 ± 0.3
0.4 ± 0.2
4
d
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
Binding sites of two active FBDD assemblies
For the inhibitors which displayed highest biology activity on RL cells (see Table 3) and
which were water‐soluble (1b, 2c, 3c and 4c), co‐crystallization and soaking experiments were
performed. Crystals grew in the presence of ligands 2c and 3c, and the crystal structure, hereof
a new form crystallizing under conditions never reported so far for the complex with ligand 2c
(
Table S1, Supporting Information), displays an overall same fold as those previously
2
4,34
reported,
with an α/β core‐domain and a cap‐domain (Figure 5A). The two complexes
display an RMSD of 0.88 Å (2c) and 0.78 Å (3c) compared to PDB‐ID 2J2C, respectively. Two
molecules are present in the asymmetric unit for both ligand complexes, for which electron
density in chain A is visible for residues 26‐493 with both compounds, but with two disordered
helix regions (317‐321 and 357‐360) for 2c and a short chain break (357‐358) for 3c. As concerns
chain B, electron density is visible for residues 26‐493 (2c) and 26‐492 (3c), with disordered
regions being 314‐324, 359‐361 (2c) and 315‐322, 357‐362 (3c). As seen in Figure 5B (with ligand
2
c), an important amount of domain swapping is observed for these protein complexes. This
particular feature suggests a strong influence of the ligand binding on the structural
organization of the enzyme. Analysis of the dimer interfaces between molecules A and B in the
35
two structures employing the PISA online server, revealed the presence of 39/35 hydrogen
bonds and 9/10 salt‐bridge interactions at the subunit interface for cN‐II structures with ligands
2c/3c, respectively.
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 59
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
Interestingly, a region reported disordered (401‐416) in the previously determined
24,34,36
structures,
due to flexibility, is visible and well ordered in the two structures with ligands
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
c and 3c, as is an additional part of the C‐terminal (Figure 5C). The position of 2c in the
structure is unique and has not been observed in previous solved structures (Figure 5C and D).
Indeed the ligand is stacked between α‐helices formed by residues 228‐240 and 371‐401 in a
cleft in the α/β core‐domain at the back of the active site, at a distance of ~10 Å of this latter,
where the N‐pyrrole part of ligand 2c is positioned at the place of helix α1 from the apo‐
36
structure (2XCX, ). When comparing the apo‐ and ligand bound structures, the long α‐helix
formed by residues 376‐401 displays a kink at residue 383, and the last part of the helix (383‐
4
01) has been tilted by 15°. In molecule A, electron density corresponding to the entire ligand
was clearly present, whereas in chain B only the electron density corresponding to the
adenosine moiety was clearly visible. For this reason, in molecule B only the adenosine moiety
was refined in the structure. In both molecules interactions between ligand 2c and the enzyme
are made via hydrogen‐bonds (S464) and hydrophobic (V59, V227, I466) interactions or via
water molecules (D229, A463) to the adenosine part of the ligand. The position of 3c in the
structure has not been observed in earlier solved structures (Figure 5E). Indeed, electron
densities clearly indicated that two molecules of 3c are stacked between two cap domains at
the dimer interface where parts of the N‐ and C‐termini join. The ligand interacts with residues
S40, L476 and R478 from molecule A and with residues Q420, S445 and R446 from molecule B
(
Figure 5F).
24
ACS Paragon Plus Environment

Page 25 of 59
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
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 59
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
Figure 5: (A) Dimer structure in the cN‐II‐2c complex. For molecule A the cap‐domain is shown
in light‐cyan and the core‐domain in brown, whereas for molecule B these are shown in dark‐
blue and orange, respectively. Ligand 2c (pink) is binding to the cap‐domain in both molecules
but only one was fully visible. (B) Surface presentation of the cN‐II‐2c complex displaying
domain swapping between dimers, and binding cavities for ligand 2c. Color codes are as in panel
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
(
A). (C) cN‐II‐2c complex in a secondary structure presentation showing molecule A, and
molecule B as a surface presentation. Color codes are as in panel (A), with the active site region
in red, effector site 1 in yellow and effector site 2 in purple. The region corresponding to the
loop which has not been observed in any previously solved crystal structure of cN‐II as well as
ligand 2c are shown in pink, and the C‐terminal part of the molecule is displayed in dark blue
and (D) Close‐up view of the newly identified binding site of ligand 2c in which interacting
residues are highlighted in orange. (E) Surface presentation of the cN‐II‐3c complex displaying
domain swapping between dimers, and ligand binding cavities with ligand 3c binding at the
interface of molecules A and B. The arrow indicates the binding site of the second ligand 3c
molecule and color codes are as in panel (a). (F) Close‐up on the 3c ligand binding site at the
interface of molecules A and B.
DISCUSSION AND CONCLUSIONS
Using a small library of 300 fragments and targeting a particularly difficult enzyme, two
sub‐groups of fragments were identified by NMR studies and most of them were able to inhibit
nucleotidase activity of cN‐II. Molecular docking studies guided the merging process and the in
26
ACS Paragon Plus Environment

Page 27 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
vitro evaluation of two more compounds (F15 and F16) gave additional experimental evidences
of the relevance of our approach. Thus, the synthesis of 19 novel compounds was achieved in
order to enhance the success rate as the degree of freedom increases with the size of the
molecule. However, negative results can appear as for compounds 1d and 1e that turned out to
be no better than the initial fragment, i.e. the naphthyl moiety (40% of inhibition for both
fragment F9 and compound 1d). Interestingly, we noticed that when two fragments were
combined (for instance, compounds 1a to 1e), the orientation of the second one (bi‐aryl or
naphthyl) was highly sensitive and governed the inhibition efficiency (when comparing
compounds 1a and 1b, for instance). With the two‐fragment assemblies, a competitive
inhibition mode was observed against cN‐II activity. This result was expected according to the
INPHARMA and docking results obtained earlier. However and surprisingly, enlarging the size of
the compounds (i.e. the association of three fragments) resulted in a change of the inhibition
mode (as shown with compounds 2a, 2b and 2c). Although the initial strategy was targeting the
active site, the discovery of more potent and non‐competitive inhibitors is highly encouraging.
Definitely, such inhibitors may behave more selective and cannot be displaced by a high
substrate concentration as competitive inhibitors.
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
Five compounds were found to be cytotoxic when tested on cancer cell lines (RL) as
shown by their IC50 values in the low micromolar range. Although this result is very promising,
one may speculate about the specificity of the observed inhibition. In order to get further
insights, synergy experiments were performed in presence of well‐known cytotoxic nucleoside
analogues used in anticancerous therapies. Several molecules (1b, 2d, 2f, 3c and 4c) that
induced an antiproliferative activity on cancer cells were also linked to a synergistic effect,
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 59
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
mainly with clofarabine and cladribine but not with fludarabine. Clarifying the absence of
3
7
synergy with fludarabine is complex since fludarabine monophosphate is a substrate of cN‐II,
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
and fludarabine was recently described as a mixed cN‐II inhibitor through its binding to the
21
active site and a regulatory site. Moreover, a recent publication showed that human
32
glioblastoma cells with decreased expression of cN‐II were resistant to fludarabine. This clearly
indicates a different metabolism of fludarabine as compared to clofarabine and cladribine and
consequently disfavoring the synergistic effect. Nevertheless, a synergy was observed with
some compounds indicating that the inhibition of cN‐II is beneficial for the metabolism of these
two cytotoxic nucleoside analogs. Our results also pointed out the non‐selectivity of one
derivative (2f) which did not inhibit the cN‐II activity in vitro but promoted a synergy with
cladribine and clofarabine. The discrepancy observed between the in vitro inhibitory activity and
antiproliferative effects of some compounds may be explained by the low target specificity for
few compounds in a cell context, as shown by the low IC values and may also be due to their
50
poor water‐solubility leading to a weak cellular uptake (loss of compound in membrane
compartments). When synergy is observed for a compound without detecting any cN‐II
inhibition, this clearly denotes the presence of another target, likely another enzyme in charge
of the anticancerous drug metabolism or activation.
Therefore, to better understand the mechanism by which our compounds actually alter
the enzyme activity, X‐ray crystallography was successfully achieved with two of them (2c and
3
c). As indicated in the results section, the ligand 2c occupies the position of α‐helix 1 in the
apo‐enzyme structure. This position constitutes a new binding site that was not previously
identified, and underlines the highly dynamic system observed for this regulatory enzyme. The
28
ACS Paragon Plus Environment

Page 29 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
binding of ligand 2c may induce dramatic changes in the functional dynamics of the enzyme and
may explain the non‐competitive inhibition mode observed in kinetics. Inspection of the crystal
structure of cN‐II‐3c, indicated that ligand 3c also bound to a previously unidentified binding site
located at the dimer interface between the cap domains. Thus, one may speculate that the
effect of ligand 3c is due to a perturbation of the “gathering point” at this interface and thereby
may potentially affect the oligomerization state of the enzyme. Indeed, both enzyme complexes
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
display a new crystal form in which a dimer is formed via domain swapping, a feature which has
38,39
been reported for other systems as being essential for oligomer assembly.
Therefore,
oligomerization and thereby domain swapping of this enzyme putatively is highly physiologically
relevant and may be involved in the enzyme activity or regulation. Interestingly, both ligand
compounds in the solved structure differ in terms of inhibition efficiency as 2c was found more
active in vitro without any synergy on cell models and 3c was less active in vitro but with synergy
on cell models. The chemical structure of both compounds is highly similar especially in the
orientation of the phenyl and imidazole or pyrrole groups (meta/meta orientation in each case).
According to the in vitro and in cellulo data, it appears advantageous for these inhibitors to
incorporate a pyrrole group instead of an imidazole. Interestingly, structurally similar fragments
(bi‐aryl, naphthyl and pyrrole) were found in the search of inhibitors against an analogous
13,40
enzyme, the human purple acid phosphatase.
In this case, naphthalene derivatives (inactive
against cN‐II) were active in the low micromolar range indicating that although these enzymes
share common structural features, they differ in their active site and enzymatic reaction
mechanism. The inhibition constants determined for the most active compounds are still in the
low millimolar range, that is insufficient for drug development and the next challenging step will
29
ACS Paragon Plus Environment