Heterocyclic Diamidine DNA Ligands as HOXA9 Transcription Factor Inhibitors: Design, Molecular Evaluation, and Cellular Consequences in a HOXA9-Dependant Leukemia Cell Model

Article abstract of DOI:10.1021/acs.jmedchem.8b01448

Most transcription factors were for a long time considered as undruggable targets because of the absence of binding pockets for direct targeting. HOXA9, implicated in acute myeloid leukemia, is one of them. To date, only indirect targeting of HOXA9 expres

Full text of DOI:10.1021/acs.jmedchem.8b01448

Subscriber access provided by TULANE UNIVERSITY
Article
Heterocyclic Diamidine DNA ligands as HOXA9 Transcription
Factor Inhibitors: Design, Molecular Evaluation and Cellular
Consequences in HOXA9-Dependant Leukemia Cell Model
Sabine Depauw, MELANIE LAMBERT, Samy Jambon, Ananya Paul, Paul Peixoto, Raja Nhili,
Laura Marongiu, Martin Figeac, Christelle Dassi, Charles Paul-Constant, Benjamin Billoré,
Arvind Kumar, Abdelbasset A. Farahat, Mohamed A Ismail, Ekaterina Mineva, Daniel P. Sweat,
Chad E Stephens, David W. Boykin, W. David Wilson, and Marie-Hélène David-Cordonnier
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b01448 • Publication Date (Web): 15 Jan 2019
Downloaded from http://pubs.acs.org on January 16, 2019
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 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 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.
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 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Wilson, W.; Georgia State University, Department of Chemistry
David-Cordonnier, Marie-Hélène; Université Lille-Nord de France, Institut
de Recherches sur le Cancer de Lille
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
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 83
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
Heterocyclic Diamidine DNA ligands as HOXA9
Transcription Factor Inhibitors: Design, Molecular
Evaluation and Cellular Consequences in HOXA9-
Dependant Leukemia Cell Model.
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
Sabine Depauw,‡,1 Mélanie Lambert,‡,1 Samy Jambon,‡,1 Ananya Paul, Paul Peixoto, Raja
2
†,1
1
1
3
1
1
Nhili, Laura Marongiu, Martin Figeac, Christelle Dassi, Charles Paul-Constant, Benjamin
Billoré, Arvind Kumar, Abdelbasset A. Farahat, Mohamed A. Ismail,2,5 Ekaterina Mineva,2
1
2
2,4
6
6
2
2
Daniel P. Sweat, Chad E. Stephens, David W.Boykin, W. David Wilson, and Marie-Hélène
David-Cordonnier*,1
1
UMR-S1172-JPARC (Jean-Pierre Aubert Research Center), INSERM, University of Lille,
Centre Hospitalier Universitaire de Lille, Institut pour la recherché sur le Cancer de Lille (IRCL),
F-59045 Lille, France.
2
Department of Chemistry, Georgia State University, Atlanta, GA 30303, United States.
3
Functional and Structural Genomic Platform, Lille University, F-59000 Lille, France.
4
Department of Pharmaceutical Organic Chemistry, Faculty of Pharmacy, Mansoura University,
Mansoura 35516, Egypt.
ACS Paragon Plus Environment
1

Page 3 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
5
6
Department of Chemistry, Faculty of Science, Mansoura University, Mansoura 35516, Egypt.
Department of Chemistry and Physics, Augusta University, Augusta, GA 30904, United States
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
* To whom correspondence should be addressed. Tel: +33 320 169 223; Fax: +33 320 538 562;
Email: marie-helene.david@inserm.fr, ORCID ID: 0000-0001-9831-5577.
‡
The authors wish it to be known that, in their opinion, these authors should be regarded as joint
First Authors
†
Present Address: Peixoto Paul, EA3922, University of Bourgogne Franche-Comté, Besançon,
France.
ABSTRACT:
Most transcription factors were for long time considered as undruggable targets due to the
absence of binding pockets for direct targeting. HOXA9, implicated in acute myeloid leukemia,
is one of them. To date, only indirect targeting of HOXA9 expression or multi-target HOX/PBX
protein/protein interaction inhibitors has been developed. As an attractive alternative by
inhibiting the DNA binding, we selected a series of heterocyclic diamidines as efficient
competitors for HOXA9/DNA interaction through binding as minor groove DNA ligands on the
HOXA9 cognate sequence. Selected DB818 and DB1055 compounds altered HOXA9-mediated
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 83
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
transcription in luciferase assays, cell survival and cell cycle, but increased cell death and
granulocyte/monocyte differentiation, two main HOXA9 functions also highlighted using
transcriptomic analysis of DB818-treated murine Hoxa9-transformed hematopoietic cells.
Altogether, these data demonstrate for the first time the propensity of sequence-selective DNA
ligands to inhibit HOXA9/DNA binding both in vitro and in a murine Hoxa9-dependent
leukemic cell model.
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
INTRODUCTION
Transcription factors are often associated with different phases of the oncogenic process
(initiation, development, invasion, metastasis…) and represent ~20% of the established targets in
cancers. However, only a few of them are targeted in cancer therapies. The different approaches
for targeting transcription factors are i) to directly target the protein via a functional pocket, ii) to
target protein-protein interactions and iii) to interfere with the transcription factor's ability to
interact with its cognate binding site on a targeted DNA sequence.1–4 Only a few of them can be
targeted through their ligand binding pocket. Hormone/retinoid receptors are a notable example,
and more precisely the retinoid receptor RARα, which can be targeted by all-trans retinoic acid
(ATRA). This compound has been used in human therapy for three decades in PML-RARα-
5
rearranged acute promyelocytic leukemia (APL) through the induction of differentiation. Other
6
transcription factors families have been considered as undruggable targets. Alternatives are
protein/protein inhibitors, that can for example target the p53 and mdm2/mdmx interface7,
8
9
10,11
or CBP/βcatenin12 interactions, or the
BCL6 and STAT3 homodimers , YAP/TAZ
1
3
14
development of drugs from phenotypic screening as against NRF2 or HSF1 . Only a limited
ACS Paragon Plus Environment
3

Page 5 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
number of strategies are dedicated to direct targeting of transcription factor-DNA binding
domains and a recent example is the PAX2 transcription factor DNA binding domain.15
Alternatively, another manner to target a transcription factor at the DNA binding level is to
develop molecules that target the DNA promoter sequence and compete with transcription factor
binding to its cognate sequence. This is the case for a variety of minor groove DNA binding
structures. Many different series of DNA minor groove binders are also in preclinical
development such as mithramycines, polyamides and diamidine heterocyclic derivatives.1,4,16,17
Among this group, mithramycin A and some of its derivatives inhibits SP1/DNA,18 which is
associated with the deregulation of many SP1-driven genes such as MYC, FOXM1, p21, Ki-67,
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
4
CRABP1, BAK or KLF5. Mithramycin A and derivatives also block EWS-Fli-1/DNA
interaction19 and have entered clinical trials against refractory Ewing sarcoma . Interestingly,
two other minor groove DNA binders recently showed interesting anti-leukemic activities by
targeting transcription factors associated with acute myeloid leukemia (AML): the alkylating
agent–conjugated pyrrole-imidazole polyamide Chb-M against RUNX121 and DB2313 against
20
2
2
PU.1 .
HOXA9 is a well-known transcription factor implicated in oncogenic and leukemic processes.
Besides its implication in some solid tumors,23–27 HOXA9 is clearly associated with different
hematologic malignancies.28–32 Indeed, HOXA9 is over-expressed in ~70 % of AML or fused to
NUP98 in the t(7;11)(p15;p15) translocation leading to the hybrid protein NUP98-HOXA9
responsible for 2-5 % of AML33,34 and 5 % of myelodysplasic syndromes (MDS) . Functionally,
HOXA9 is involved in the differentiation blockade, resulting in the accumulation of leukemic
blasts in patients.28,29,32,36 Such leukemic activity is directly attributable to the HOXA9 DNA
binding activity. Indeed, deletion of its homeodomain abolishes the propensity of HOXA9 to
35
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 83
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
induce leukemia in mice.37–39 Furthermore, HOXA9 was found quite recently as one of the genes
associated with leukemic stem cell signature and resulting in poor prognosis.40 The HOXA9
transcription factor is therefore a good candidate for the development of new targeted therapies
for differentiation that would be intended for a large number of patients suffering from AML.
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
HOXA9 binds to it regulatory DNA sequence at a 5’-tgATTTAt site (JASPAR MA0594.1), as
4
1
defined by ChIP-seq in a murine model expressing a tagged-Hoxa9 protein. DNA recognition
occurs through the interaction between the third α-helix of its homeodomain and the DNA major
groove, and between the amino-terminal unfolded module of its homeodomain which makes
4
2
deep contacts in the minor groove of the AT-rich segment of the consensus binding site. In the
present study, we evaluated a focused library of heterocyclic diamidine DNA ligands as
functional inhibitors of human HOXA9/DNA binding activity. We evaluated their mode of
binding to DNA in vitro and the cellular consequences of the protein-DNA complex inhibition.
We further selected DB818 and DB1055 for additional investigation and evidenced their
inhibitory effect on HOXA9-induced transcription control by using HOXA9-driven luciferase
gene expression. The compound effects on cell survival and clonogenic activity of Hoxa9-
transformed murine bone marrow cells argues for cell differentiation process, as highlighted by
transcriptomic analyses of DB818-treated leukemic cells. These results clearly show for the first
time the selection of direct functional inhibitors of HOXA9/DNA binding through the targeting
of the HOXA9 consensus binding site by DNA ligands.
RESULTS AND DISCUSSION
Design of the evaluated compounds
ACS Paragon Plus Environment
5

Page 7 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In order to develop inhibitors for the HOXA9 transcription factor at its DNA binding level, we
took advantage of the HOXA9 DNA binding domain particularities. Indeed HOXA9 binds
through a direct interaction of the third α-helix and the N-terminal portion of its homeodomain
with the major and minor grooves, respectively. The key interaction are with the 5’-ATT(A/T)
sequence of the 5’-tgATTTAt defined consensus site.41,42 We were specifically interested in
heterocyclic diamidine compounds that have been previously reported to be potent in vitro
transcription factor DNA binding inhibitors: DB293 against Pit-1 and Brn-343 and DB1255
against the ERG transcription factor44,45. The AT specific minor groove binding diamidines
evaluated as HOXA9 inhibitors were obtained from our large library of heterocyclic diamidines
synthesized as potential anti-parasitic agents. The selections were made from a number of
structurally diverse diamidine types with many selected for their binding to AT-rich, ATT(A/T)
or ATGA binding to DNA 46,47 and new structures were designed on the basis of three central
aromatic groups R , R and R flanked by an amidine moiety at each extremity (Table 1). A
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
1
2
3
series of 39 compounds are classified based on the presence of R as a 5- or 6-membered ring, on
2
the nature or R rings (general with a phenyl or a benzimidazole). Modifications of the R ring
3
1
and the amidine extremities (R and R ) offer a large diversity of molecules to expand
4
5
structure/function relationships (Table 1).
Approximately two-thirds of those selected have been previously reported and references to their
synthesis are provided in Supplementary Information. The syntheses of those previously
unreported are described below.
The synthesis of three furan benzimidazole analogues (4a-c; DB915, DB928, DB995) is
outlined in Scheme 1. The 2-cyanoaryl-5-formyl furans were prepared employing standard Stille
cross-coupling conditions allowing 2-(tri-n-butylstannyl) - 5-(dimethoxymethyl) furan- to react
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 83
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
with the various cyanoaryl bromides in the presence of tetrakis (triphenylphospine) palladium.
The required bis-nitriles were obtained by oxidative coupling of 3, 4-diaminobenzonitrile with
the various 2-cyanoaryl-5-formyl furans using benzoquinone as the oxidant. The diamidines
were prepared by first converting the bis-nitrile into the corresponding amidoximes, which were
acetylated, and subsequently subjected to hydrogenolysis to yield the desired diamidines. In the
case of DB928, the hydrogenolysis in addition to cleaving the N-OAc bond also cleaved the
benzyl group.
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
Br
OCH3
(a)
(b)
+
CHO
2 X1
Bu3Sn
O
O
NC
NC
X
X
X3
OCH3
2 X1
NC
X
X3 1a-c
N
HN
a-c
(c)
N
O
O
HON
2 X1
X1
HN
X3
X2
X3
NOH
CN
H N
2
2
3a-c
H N
2
(d)
X1
C
X2
C
X3
N
a
b
c
OBn (OH in case of DB928)
O
HN
X1
HN
a-c
X2
X3
NH
N
C
Nil
Nil
H N
2
4
H N
2
C
N
a DB928
b DB915
c DB995
Scheme 1. Reagents and conditions: (a) i. Pd(PPh3)4, 1,4-dioxane,ii. 2.0 M HCl; (b) 3,4-diaminobenzonitrile, 1,4-
benzoquinone, EtOH; (c) NH OH-HCl/KO-t-Bu, DMSO; (d) i. AcOH/Ac O, ii. H /Pd-C, AcOH.
2
2
2
The route employed for synthesis of the thiazole benzimidazole (7; DB2651) is shown in
Scheme 2. In this case Suzuki cross-coupling between 5-bromothiazole-2-carbaldhyde and 4-
cyanophenylboronic acid was utilized to obtain 5-(4-cyanophenyl) thiazole-2-carbaldehyde. In
this example, the bis-nitrile was achieved by oxidative coupling of 3, 4-diaminobenzonitrile and
the thiazole aldehyde using sodium metabisulfite as the oxidant. The diamidine was obtained
from the bis-nitrile using classical Pinner methodology.
ACS Paragon Plus Environment
7

Page 9 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
N
N
(
a)
(b)
OHC
CN
Br
(
HO) B
2
OHC
S
S
CN
5
H
N
H
N
N
N
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
(c)
S
S
N
N
NH2
NH
CN
HN
NC
7
6
NH2
DB 2651
Scheme 2. Reagents and conditions: (a) Pd(PPh ) , Na CO /H O, dioxane, reflux; (b) 3,4-
3
4
2
o
3
2
diaminobenzonitrile, Na S O / DMSO, 140 C; (c) i HCl(g), EtOH, ii NH (g), EtOH.
2
2
3
3
In Scheme 3 the synthesis approach to five (11a-e; DB1478, DB1481, DB1879, DB2005,
DB2529) heteroaryl indole diamidines is presented. The 2-arylcyanofurans and thiophenes were
obtained using Stille coupling between 2-(tri-n-butylstannyl) furan and 2-(tri-n-butylstannyl)
thiophene with the indicated cyanoaryl bromides. NBS bromination of the 2-aryl furans and
thiophenes at room temperature yielded the corresponding 5-bromo-2-aryl furans and
thiophenes. Reaction of either the Boc-protected 5 or 6-cyano-2-(trimethlystannyl) indole with
the 5-bromo-2-aryl furans or thiophenes yield the Boc-protected bis-nitriles. The bis-nitriles were
converted into the diamidines by reaction with lithium bis(trimethylsilyl) amide.
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 83
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
X3
X3
X3
(
a)
(b)
X1
Br
CN
X1
SnBu3
Br
X1
X2
X2
CN
HN
X2
CN
8a-e
9a-e
X3
X3
(a)
(c)
SnMe3
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
+
NC
X1
X1
N
NH
NH2
X2
N
O
CN
X2
NC
NH
O
O
O
H N
2
10a-e
11a-e
X1
X2
X3
a DB1478
b DB1481
c DB1879
d DB2529
e DB2005
CN/Am
a
5
O
C
H
b
c
d
5
6
6
O
S
S
N
H
H
H
C
N
e
6
O
C
OMe
-
Scheme 3: Reagents and conditions; (a) Pd(PPh ) , Dioxane; (b) NBS,DMF, rt; (c) i LiN(TMS) /THF, ii HCl(g), EtOH.
3
4
2
The synthesis of the bis-N-(4-nitrophenylethyl) analogue (12a; DB421) of furamidine and the
tellurophene analogue (12b; DB1751) of furamidine is presented in Scheme 4. The former was
obtained from the bis-nitrile by conversion into the corresponding imidate ester, which on
reaction with 4-nitrophenethyl amine yielded DB421. Reaction of 2, 5-bis-(4-cynophenyl)
tellurophene with lithium bis(trimethylsilyl)amide provided DB1751.
a) for DB421
HN
X
X
NH
NC
CN
b) for DB1751
RHN
12a-b
NHR
X = O, Te
a
DB421 X = O, R = CH CH2
b DB1751 X = Te, R = H
NO2
2
Scheme 4. Reagents and conditions; (a) i HCl(g), EtOH ii 4-nitrophenethylamine, EtOH; (b) i LiN(TMS) ,THF, rt,
2
ii HCl(g), EtOH.
The synthesis of the N-methylbenzimidazole analogue (14; DB1314) is shown in Scheme 5.
The required bis-nitrile was obtained by oxidative coupling of 3-(methylamino)-4-
ACS Paragon Plus Environment
9

Page 11 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
aminobenzonitrile with 4-cyano-3’-formyl-1, 1’-biphenyl using sodium metabisulfite as the
oxidant. The mentioned lithium bis(trimethylsilyl)amide method was used to convert the bis-
nitrile into the diamidine.
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
(a)
N
(b)
N
CHO
N
N
HN
NC
NC
13
14
CN
NH2
NH
H N
2
DB 1314
Scheme 5. Reagents and conditions: (a) 4-amino-3-(methylamino)benzonitrile, Na S O , DMF, reflux.
2
2
3
(
b) i LiN(TMS) , THF, rt ii HCl(g), EtOH.
2
The route utilized for synthesis of the quinoline analog (18; DB1504) is outlined in Scheme 6.
The Pfitzinger reaction between 2-acetylthiophene and 6-bromoisatin provided six-bromo-2-
(thiophen-2-yl)quinoline-4-carboxylic acid which on heating in the presence of copper(I)cyanide
results in both decarboxylation and replacement of the bromo group with a cyano group. Once
again, NBS bromination of the 2-aryl thiophene at room temperature yielded the desired 5-
bromo-2-aryl thiophene analogue. Reaction of 4-cyanphenylboronic acid with the 5-bromo-2-
aryl thiophene derivative under standard Suzuki coupling conditions provided the needed bis-
nitrile. Again, the lithium bis(trimethylsilyl)amide methodology was employed to obtain the
diamidine DB1504.
ACS Paragon Plus Environment
1
0

Journal of Medicinal Chemistry
Page 12 of 83
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
COOH
N
O
Br
Br
a)
Me
+
O
S
S
N
O
H
c)
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
b)
N
N
Br
S
S
CN
CN
15
16
N
N
e)
HN
H N
NH
NH2
S
d)
S
CN
NC
2
18
17
DB 1504
o
o
Scheme 6. Reagents and conditions: (a) NaOH, H O,100 C, overnight; (b) CuCN, DMF, 140 C (c) NBS, DMF,
2
o
rt ;(d) 4-cyanophenylboronic acid, 1,4-dioxane, Pd(PPh ) , 2M K CO , 70 C, overnight;
3
4
2
3
(
e) i LiN(TMS) , THF, rt, 48 h, ii HCl(g), EtOH.
2
Scheme 7 shows the approach used for the synthesis of the bis-N-methylbenzimidazole
analogue (19; DB2572). In this case, the bis-aldehyde was oxidatively coupled, mediated by
benzoquinone, with 3-amino-4-(methylamino)benzamidine to directly yield DB2572.
N
(a)
OHC
S
S
N
N
CHO
H N
2
N
19
DB 2572
NH
NH
H N
2
Scheme 7. Reagents and conditions: (a) 3-amino-4-(methylamino)benzamidine, benzoquinone, EtOH, reflux.
Selection of HOXA9/DNA binding inhibitors
An ELISA-derived DNA binding inhibition assay was first used to evaluate this series of
compounds for the inhibition of the HOXA9 protein interaction with its cognate DNA binding
ACS Paragon Plus Environment
1
1

Page 13 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
sequence 5’-aTGATTTA (HBS for HOXA9-binding site) which is 5’-biotinylated in order to be
immobilized on streptavidin-coated 96-wells plates. Specific (HBS) and nonspecific (NS)
oligonucleotides (sequences in Supplementary Table S1) were used as competitor DNAs to
validate the experimental conditions for selective protein/DNA inhibition experiments. The
previously characterized AT-rich DNA ligand DB75 and the ATGA-site DNA binding
compound DB293 were used as controls. Among those evaluated, ten compounds showed
significant activity on HOXA9/DNA binding (grey and white box for at least 50% and 75%
inhibition, respectively) (Figure 1A) and were further evaluated using EMSA (representative gels
presented in Figure 1B and Supplementary Figure S1). HOXA9 protein expressed using a
reticulocyte lysate system was incubated with radiolabelled HBS-double stranded
oligonucleotide (sequence in supplementary Table S1) and increasing concentrations of the
tested compounds. The retarded band on the gel shows the formation of the HOXA9/HBS
complex (bound, b), separated from the free HBS DNA (f). This conclusion was further
validated by addition of HOXA9-directed antibody (Ab) and by the absence of the retarded band
using reticulocyte lysate prepared in the absence of HOXA9-expressing vector (first and last
lanes “0” from the upper left panel in Figure 1B). Quantification of the HOXA9/DNA inhibition
from EMSA is presented in Figure 1C. This result demonstrated that DB818, DB1055, DB884,
DB1879 and DB2529 are highly potent inhibitors of HOXA9/DNA binding with more that 90%
inhibition at 2.5 µM (“High” panel). DB1177, DB883 and DB2223, on the other hand, are less
efficient (“Medium” panel) with 90% inhibition reached using 7.5 µM. By contrast, DB1620,
DB1878 and DB911 are poor inhibitors (“Low” panel) while others are considered to be inactive
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
(right panel, “Inactive”).
ACS Paragon Plus Environment
1
2

Journal of Medicinal Chemistry
Page 14 of 83
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
In terms of structure activity relationship, DB818, DB1879 and DB2529 are structurally related
compounds that share a common central thiophene ring but differ by the presence or absence of a
nitrogen atom that converts (i) the indole ring of DB1879 to a benzimidazole ring in the DB818
for R position or (ii) a phenyl ring in R position of DB1879 and to a pyridine ring in DB2529.
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
3
1
Neither modification alters their propensity to inhibit the HOXA9/DNA interaction. By contrast,
changing the thiophene ring of DB818, DB1879 or DB2529 to a furan (DB293, DB1878 or
DB915, respectively) or a thiazole ring (DB2651) in the benzimidazole series strongly reduces
HOXA9 inhibition efficacy whereas a pyrazole ring modification still confers some weaker
inhibitory activity to DB2223. Surprisingly, such modification of the thiophene ring to a thiazole
ring in the di-phenyl series changes an inactive compound (DB351) to an effective (but low)
inhibitor of HOXA9/DNA binding (DB1620). The addition of a methyl group on the
benzimidazole ring (DB818) at the R position or changing the DB818 benzimidazole ring to a
3
quinoline ring or to a N-isopropyl-benzimidazole-phenyl moiety strongly alters its propensity to
inhibit the HOXA9/DNA interaction (see DB2429, DB1504 or DB2708) relative to DB818. A
methyl group on the phenyl ring in R also alters the inhibitory effect of DB1878 (unmethylated)
1
as shown using DB2005 (methylated) (Figure 1A). Moreover, the addition of a methyl group to
the benzimidazole ring in the R position alters the HOXA9 inhibition efficiency as observed
3
with the highly active unmethylated compounds DB818 and DB2529, relative to their inactive
methylated counterparts DB2429 and DB2430 (of note, this latter also differs by the presence of
a furan 5-membered ring in R position instead of a thiophene). Finally, the two pyrrole
2
derivatives bearing an additional ring (phenyl for DB884 or pyridine for DB883) at both the
R /R positions are identified as high (DB884) and medium (DB883) inhibitors.
1
5
ACS Paragon Plus Environment
1
3

Page 15 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The results obtained from ELISA-derived DNA binding inhibition assay and EMSA are in good
agreement and highlight DB818, DB1055, DB883, DB1879 and DB2529 as interesting inhibitors
of HOXA9/DNA binding. By contrast, DB828, that failed to inhibit HOXA9/DNA binding, was
selected as a negative control for further analyses. It is also worth noting that neither DB818 nor
DB1055 inhibits the DNA binding of the HMG-box protein, a protein known to interact with
DNA on AT-rich sites (Supplementary Figures 1B-C).
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
Direct DNA binding of the selected HOXA9/DNA inhibitors into the groove of DNA
Thermal melting experiments allow a robust screening for the relative binding affinity of the
ligands with DNA sequences. In order to understand ligand-DNA complex stability, we
conducted thermal melting studies of different diamidine compounds with cognate HBS
sequences. Ligand and HBS duplex DNA sequences were prepared at a ratio of 1:1 and 2:1
(Figures 2A). At R=1:1, DB818, DB1055 and DB2529 showed Tm values of 7-10°C. The
thermal stability of the complex increase at R=2:1 ligand-DNA ratio (Tm >10°C) for DB818,
DB1055, DB2529, DB1879 and DB1878. On the other hand, DB884 and DB1177 showed
moderate binding with the HBS duplex (Tm = 8-9 °C). However, DB883, DB884, DB2223 and
DB1620 bind very weakly with the cognate HBS sequence.
To evaluate the binding mode of diamidine compounds with the same HOXA9-binding
site the circular dichroism (CD) spectral experiment was performed (Figure 2B). Neither the
duplex DNA nor the free diamidines compound exhibits CD signals in the compound absorption
region above 300 nm. Adding increasing amount of diamidine compounds that evidenced in
vitro positive results, leads to a substantial positive induced CD signals (ICD) for the complex
formation with DNA at the wavelengths between 300 and 400 nm. This observation indicates
ACS Paragon Plus Environment
1
4

Journal of Medicinal Chemistry
Page 16 of 83
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
that the diamidine compounds bind within the groove of the HBS DNA sequence, more likely
within the minor groove as strongly evidenced in the crystal and/or NMR structure for DB818,
DB1055, DB884 or DB293 to 5’-AATT-containing DNA (PDB: 1VZK, 2I5A, 2I2I, 2GYX)47–49
.
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
On the other hand, DB828 induces only weak CD changes at the highest drug concentrations in
agreement with the low Tm. All tested compounds except DB1177 and DB828 changed the
positive ellipticity band of the DNA helix (positive CD band near 275 nm), suggesting that they
induce modification of the stacking of the successive bases of the DNA.
HOXA9/DNA inhibitors directly bind to HOXA9-cognate sequence
To define the specific drug/DNA binding sequence, DNaseI footprinting experiments were used
for selected compounds from the “High” (DB818, DB1055, DB1879, DB2529), “Medium”
(DB1177, DB883), “low” (DB1878) and “inactive” (DB828) subgroups for HOXA9/DNA
binding inhibition. As can be seen from Figure 3A, DNase I generated DNA fragments were
separated upon electrophoretic migration from top to bottom. Some of the cleavage sites were
protected from DNaseI digestion around the 50bp position with addition of increasing
concentrations of selected diamidine compounds. The protected site was clearly evidenced on the
corresponding densitometric analysis located on the HOXA9 binding site 5’-ATTTA for
compounds from the “High” subgroup DB818, DB1055, DB1879 and DB2529 (Figure 3B). By
contrast, DB828 and DB883 failed to recognize the 5’-ATTTA sequence, whereas DB1878 and
DB1177 has a much weaker binding activity of the 5’-ATTTA site (Figure 3B). These results are
in agreement with the DNA melting temperature studies on the HBS sequence (Figures 2A). We
used the same UV/vis DNA melting approach to evaluate the binding of DB818, DB1055 and
DB828 on the mutated HBS-oligonucleotide sequences. We have mutated the sequence at the 5’-
ATGA (HBS-ATGAm) or the AT-rich (HBS-ATm) or containing both mutations (HBS-2mut)
ACS Paragon Plus Environment
1
5

Page 17 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
oligonucleotide sequences are presented in Supplementary Table S1). For the HBS-ATGAm
sequence, the Tm values obtained with DB818 and DB1055 were similar to that obtained with
wild-type HBS oligonucleotide, which suggests that the diamidine compounds do not bind at the
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
5
’-ATG binding site of HBS (Figure 3C). By contrast, when the 5’-ATTTA site was mutated
alone for HBS-ATm or together with the ATGA portion for HBS-2mut, Tm values decreased
by 4°C, which is similar to the observed value for the DB828. From the previous experiment
(Figure 2A) it has been observed that DB828 binds with HBS as a weak nonspecific binder. This
result is in agreement with the DNase I footprinting experiments which suggest that, among
other, DB818 and DB1055 interact at the 5’-ATTTA-sequence of the HOXA9 binding site
(Figure 2) but did not bind to an HBS-mutated DNA as shown in Supplementary Figure S2. This
sequence specific binding evidenced DB818 and DB1055 as competitors for HOXA9 binding to
its cognate sequence on the DNA helix.
Surface plasmon resonance (SPR) experiments were applied to measure the dissociation
constants K for DB818 or DB1055 on HBS or HBS-2mut sequences (Figure 4). DB818 and
D
DB1055 have similar strong K values on HBS sequence (4.6 nM and 5.5 nM, respectively). The
D
dissociation constant is much greater on the mutated sequence (0.35 µM and 1.7 µM,
respectively), highlighting that DB818 is ~75-fold more selective for HBS than to HBS-mutated
sequence whereas DB1055 is even more selective with >300-fold difference in K values (Figure
D
4).
To gain structural insight into the mechanism by which the DB818 and related diamidines
compounds inhibit HOXA9 binding to the cognate HBS, we performed a 100 ns molecular
dynamics simulation (MD) of the DB818 binding to ATGATTTA sequence (Figure 5,
ACS Paragon Plus Environment
1
6

Journal of Medicinal Chemistry
Page 18 of 83
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
Supplementary Figure S3, Supplementary Table S2). The HBS-bound complex showed that
DB818 is selectively bound to the minor groove of AT-rich 5’-A T T T A /5’-T A A A T
1
2
3
4
5
5’ 4’ 3’ 2’ 1’
binding site and makes excellent van der Waals interactions with the walls of the minor groove
Figure 5A-B). Two amidine groups, −N−Hs form H-bond either with T=O of dT base or A-N3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
1
’
of dA base (Figure 5C-D) that are an average of 2.3 Å in length. The third strong H-bond is
5
formed with the T=O of dT base to the H-N-benzimidazole of DB818 with 2.0 Å bonding
4
length, to account for much of the binding selectivity of DB818 (Figure 5D). Additional
selectivity in binding is provided by the −CH group of the six-member ring of benzimidazole
that points deeply into the minor groove and forms a dynamic close interaction with A-N3 of the
dA4’ base of the A•T bp. Interaction of the phenyl ring with dT base further stabilizes
DB818/DNA interaction and increases sequence selectivity. The sum of these numerous
relatively stabilizing interactions, marks DB818 a strong ATTTA sequence binder within the
minor groove which may explain how DB818 inhibits the binding of the HOXA9 protein at the
major groove and minor groove of its cognate binding site (see conclusions section).
HOXA9/DNA binding inhibitors are effective at entering cells and localizing to the
nucleus.
In order to address the cellular consequences of HOXA9 inhibitors, we used MigA9 cells,
resulting from retroviral transduction of the murine Hoxa9 gene in murine bone marrow cells
that were injected in sub-lethally irradiated mice and then collected from the bone marrow after
leukemia burden and cultured in appropriate medium as described.26 We first evaluated the
cytotoxic effect of a series of diamidine compounds. Cell survival activity (IC ) on MigA9 cells
5
0
was measured at 72 hours of treatment by using a colorimetric assay based on the reduction of a
MTS tetrazolium dye. The IC values of diamidine compounds are at micromolar range, except
5
0
ACS Paragon Plus Environment
1
7

Page 19 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
DB828 which has very high IC50, and supports previous in vitro observations. Surprisingly,
DB2223, as an effective in vitro inhibitor of the HOXA9/DNA binding complex, failed to alter
MigA9 cell survival (Table 2). We also evaluated their susceptibilities to enter cells and localize
in the nucleus by using fluorescence microscopy as previously described for DB75 and its
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
5
0
derivatives. Most of the diamidines have intrinsic fluorescence properties in blue wavelengths.
Compounds were evaluated either on MigA9 murine leukemia cell line (Figure 6A and
Supplementary Figure S4) and/or HT-29 adherent cell line (Supplementary Figure S5). All
evaluated compounds entered the nucleus except DB911 which could not be detected due to the
lack of intrinsic fluorescence properties. DB828, which also localized to the nucleus but failed to
inhibit HOXA9 DNA binding, was used as a negative control in the following cellular studies.
DB818 and DB1055, but not DB828, inhibit HOXA9-dependant transcription control
In order to address the inhibitory effect of DB818 and DB1055 on the HOXA9-mediated
transcriptional control, we performed luciferase assays. We used a pGL3-promoter vector in
which the minimal promoter is under the control of 6-repeated HBS sequences, either the wild-
type or mutated sequence at the 5’-ATTTA binding site (HBS and HBS-mut oligonucleotides,
Supplementary Table S1). Cell transfection was performed in the K562 cell line as a leukemic
5
1
cell model that is known for its lack of internal HOXA9 expression. The absence of HOXA9
expression has been validated by using qRT-PCR and western-blotting, personal information and
Supplementary Figure S6A, respectively. In this cell line, both DB818, DB1055 and DB828
compounds have no effect on cell survival and clonogenic propensities (Supplementary Figure
S5B, Table 4, top lanes). Extrinsic HOXA9 protein was expressed by transfecting the pCMV-
AD-HOXA9 vector in the K562 HOXA9-negative cells (Supplementary Figure S6A). HOXA9
protein has only low propensity to transactivate by itself in the absence of co-factors. Therefore,
ACS Paragon Plus Environment
1
8

Journal of Medicinal Chemistry
Page 20 of 83
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
the region encoding the transactivation domain (AD) of NF-κB was fused in frame to the
HOXA9 full length open-reading frame in order to express the HOXA9-AD fused protein that
presents increased transactivation properties relatively to HOXA9 alone (personal information,
see also Material and Methods section). Luciferase assays were performed in the presence versus
absence of evaluated compounds of DB818 and DB1055 as HOXA9/DNA binding inhibitors, or
DB828 as a negative control. The results showed that both DB818 and DB1055, but not DB828,
were able to inhibit HOXA9-mediated transactivation in the presence of wild-type HBS
sequences by controlling luciferase expression but not in the presence of HBS-mut sequences
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 6B). These results also suggested that the HBS binding site is required for selective
inhibition of HOXA9 transcriptional activity.
Gene expression analysis of MigA9 cells treated with DB818 highlights induced cell death
processes.
In order to investigate the cellular mechanistic action of selected diamidine HOXA9/DNA
inhibitors, we initially used a global gene expression analysis of MigA9 cells which were treated
with DB818 (2.5µM) for 6, 12 and 24 hours. Co-deregulated genes with |mean fold changes| ≥
1
.5 are presented as heatmap in Figure 7A (|mean fold changes| ≥ 1.5). GSEA analyses on
MSigDB C2-Chemical and Genetic Perturbation gene sets (Figure 7B) evidenced that the
significant genes up-regulated in DB818-treated MigA9 cells at the different time correspond to
genes that were down-regulated in human CD34+ cells transduced with NUP98-HOXA9 fusion
(TAKEDA TARGETS OF NUP98 HOXA9 FUSION 8D DN), Figure 7B. Similarly, the list of
significant commonly down-regulated genes in DB818-treated MigA9 cells are highlighted in
geneset enrichments corresponding to genes up-regulated in murine HoxA9+Meis1 transformed
cells (WANG IMMORTALIZED BY HOXA9 AND MEIS1 UP). Both results of CGP genesets
ACS Paragon Plus Environment
1
9

Page 21 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
enrichments suggest that the gene deregulation in opposite way between the transformation of
hematopoietic progenitors with HoxA9+Meis1 or NUP98-HOXA9 and the inhibition of HOXA9
activity by DB818 in MigA9. Other genesets associated with HOXA9 expression control
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
NUYTTEN EZH2 TARGETS UP),52 genes controlled by HOXA9 (GABRIELY MIR21
(
5
3
TARGETS), leukemic stem cell inhibition (GAL LEUKEMIC STEM CELL DN) and AML-
induced differentiation (MARTENS TRETINOIN RESPONSE DN) were also found as
significantly enriched in our dataset (Figure 7B). Ingenuity Pathway Analysis (IPA) of
deregulated genes highlights the activation of macrophage differentiation, a reduced quantity of
hematopoietic cells and an increase in apoptosis (Figure 7C). Analysis of the BloodSpot database
(www.bloodspot.eu) found multiple commonly deregulated genes (identified as arrows in Figure
7
A) which are significantly associated with the differentiation of granulocyte-monocyte
progenitors (GMP) to granulocytes and monocytes (Supplementary Figure S7). Among them, the
well-known monocyte/macrophage expressing the integrin Itgam (CD11b) is expressed upon
differentiation of AML blasts54,55 and the Oasl2 enzyme that is also associated with the
5
6
differentiation of monocytes and dendritic cells . Such over-expression of Itgam and Oasl2 were
validated by qRT-PCR after a 48 hours of treatment with DB818 (5 or 10µM), as were validated
the over-expression of Akap13 and Arfgef1 (Figure 7D) as examples of genes associated with
granulocyte and/or monocyte differentiation (Supplementary Figure S7). Similar results of
overexpression of these genes were also observed for DB1055, (Figure 7D) but DB828 failed to
achieve the similar results. Recently, Sun et al. (2018) identified murine Hoxa9 DNA binding
sites by ChIP-seq analysis of a murine leukemia model obtained by transformation of murine
progenitor cells by co-expression of hemagglutinin (HA)-tagged HOXA9 and MEIS1 using an
anti-HA antibody.57 Among Hoxa9 ChIP-peaks, the sequence identified as “DE-651 peak”
ACS Paragon Plus Environment
2
0

Journal of Medicinal Chemistry
Page 22 of 83
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
located at position chr7:82861417-82863417 (from NCBI37/mm9 mouse reference genome) is
positioned within Akap13 gene. Interestingly, we identified a cognate HOXA9 binding sequence
within DE-651 peak sequence (Figure 8A). An oligonucleotide overlapping this HBS site
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
(Akap13-HBS, Supplementary Table S1) was synthesized and radiolabeled to then be evaluated
using EMSA for HOXA9 binding and DB818 or DB1055 inhibitory effect. As evidenced in
Figure 8B, HOXA9 interacts with this sequence and such HOXA9/DNA binding is inhibited by
the addition of increasing concentrations of DB818 or DB1055. Such interaction and its
inhibition are possible option to explain HOXA9-mediated control of Akap13 expression,
assuming that HOXA9 binding is associated with an inhibition of Akap13 expression. DB818
and DB1055 would abolish such negative control of Akap13 expression by HOXA9, leading to
an overexpression of Akap13 as evidenced by qRT-PCR (Figure 7D) for this gene which over-
expression is associated with granulocyte and monocyte differentiation (Supplementary Figure
S7).
As highlighted above using the global list of deregulated genes, IPA verified two main
biological processes that were altered following DB818 treatment on MigA9 cells: cell
differentiation (up-regulated) and cell proliferation (down-regulated). Based on this in silico
transcriptomic analysis, cellular analyses were then performed to address the impact of DB818
and its derivatives on the cell proliferation and differentiation in MigA9 cells.
Inhibition of MigA9 cell proliferation
MTS measurements (Table 2) showed that most of the diamidines (except DB828 and DB2223)
reduced the cell survival of the HoxA9-transformed cell line MigA9. Comprehensive changes in
cell survival, determined by using MTS, may be due to multiple cellular processes such as cell
ACS Paragon Plus Environment
2
1

Page 23 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
death, cell cycle arrest and cell differentiation processes among others. Consequently, we then
further addressed the impact of diamidine compound treatment on each of those cellular
processes to have a better insight into the mechanism of action of those molecules. First, we
validated the induction of cell death by DB818, as highlighted in transcriptomic analysis, using
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
+
both the quantification of the proportion of propidium iodide-positive cells (PI ) and of subG1
population, two markers of cell death addressed from cell cycle analysis. The results in Table 2
show an increase in the PI-positive population for both 24 and 48 hours treatment and for the
subG1 proportion at 48 hours treatment with DB818. These results suggest that DB818 induces
apoptotic cell death, in agreement with transcriptomic analyses that highlighted expression of
gene sets associated with a decrease of the quantity of blood cells upon treatment with DB818
(Figure 7). Using the same experimental approaches, we showed that DB1878, DB1879 and
DB2529, as close derivatives of DB818, also strongly induce cell death (Table 2). Interestingly,
the five ring containing compounds DB883 and DB884 were the most active in terms of cell
death, suggesting that cell death is the main cellular process that is responsible for the decrease
of cell survival by DB883 and DB884. By contrast, DB1055 does not induce cell death after 24
or 48 hours treatment to explain its effect on global cell survival. The impact of drug treatment
on cell cycle resulted in a strong reduction in the proportion of MigA9 cells in S-phase upon
treatment with DB818 and derivatives (DB1878/1879/2529), as well as with DB1055 and
DB1620 (Table 3). This reduction of cycling cells is associated with a strong increase in the
G0/G1 phase (DB1055, DB1878, DB2529), in the G2/M phase (DB818) or in both phases
(DB1879, DB1620) at the tested concentrations and at 48H treatment.
Anti-clonogenic activities were then performed by using DB818, DB1055 and DB828 to address
cell death and cell differentiation processes. This experiment first highlights the impact of both
ACS Paragon Plus Environment
2
2

Journal of Medicinal Chemistry
Page 24 of 83
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
DB818 and DB1055 on global clonogenic propensity of the HOXA9-positive MigA9 cell line,
but not on the HOXA9-negative K562 cell line (Table 4 and Supplementary Figure S6C-D).
Surprisingly, the intrinsic fluorescence of DB818 and DB828 were still visible even after 10 days
of culture in Methocult using fluorescence microscopy with DAPI-filters, which indicates very
high stability of the diamidine compounds within the cells (Supplementary Figure S6D). Results
for clonogenic assays clearly show reduction in the number of colony sub-types obtained in the
presence of DB818 or DB1055, but not DB828, for the untreated wells (CTR) (Table 4).
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
Induction of granulocyte/monocyte differentiation.
Finally, using the same anti-clonogenic experiment, we proved that the global decrease in the
colony number is associated with a reduced proportion of CFU-GM (granulocyte and monocyte)
in favor of an increase of CFU-M (monocyte only) and CFU-G (granulocyte only). These results
suggest that both DB818 and DB1055 induced a decrease in the number of the most immature
hematopoietic progenitors cells (CFU-GM) and favored the development of colonies that are
more engaged in granulocytic or monocytic differentiation (CFU-G, CFU-M). This result is in
agreement with gene expression analyses with DB818 (GSEA, IPA analyses), confirmed by
qRT-PCR for both DB818 and DB1055 (but not for DB828) on Itgam (CD11b) as well as on
CD14 expression as two markers of monocytic to macrophagic differentiation. Altogether these
molecular and cellular experiments highlight the leukemic cell differentiation process associated
with treatment with the HOXA9/DNA binding inhibitors DB818 and DB1055.
CONCLUSIONS
ACS Paragon Plus Environment
2
3

Page 25 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In the present work, we selected a series of heterocyclic diamidines and derivatives as
sequence-selective DNA ligands that directly inhibit HOXA9/DNA binding (Figure 1, Figure
2A) through groove binding (Figure 2B) on the HOXA9 cognate DNA sequence (Figure 3).
Particularly, we identified DB818, DB1055, DB884, DB1879 and DB2529 as strong inhibitors
of HOXA9/DNA binding using both an ELISA-derived DNA binding inhibition assay (Figure
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
1
A) and EMSA with more than 90% inhibition at 2.5 µM (Figure 1B-C). Those compounds also
strongly stabilize the DNA helix based on DNA melting temperature results (Figure 2A).
DB818, DB1055 and DB2529 were the most efficient compounds based on results obtained at
the lower drug/DNA ratio (R=1:1). The recognition occurs at the HOXA9 cognate DNA
sequence as demonstrated using DNase I footprinting experiments (Figure 3), through groove
binding, as shown using circular dichroism (Figure 2B), pointing deeply in the minor groove and
making close contacts with AT bases of the HBS sequence as evidenced using molecular
modeling studies (Figure 5). DNase I footprinting experiments also evidenced that DB818 and
DB1055 appeared to be very active and selective compounds on binding to the HOXA9-binding
site (Figure 3) but not to the mutated HBS sequence (Supplementary Figure S2). Moreover,
DB818 and DB1055 failed to inhibit DNA binding of the HMG-box protein used as a specificity
control (Supplementary Figure S1B-C). We therefore selected DB818 and DB1055 as highly
potent inhibitors that compete with HOXA9 for binding to its cognate sequence. DB818 and
DB1055 were previously shown to interact with AT-rich DNA sequences45 with binding KD
values of 9 nM and 17 nM respectively. Our K measurements obtained with the HBS-
D
containing DNA sequence highlight stronger interaction with values of 4.6 and 5.5 nM,
respectively (Figure 4). The binding of DB818 and DB1055 to HBS requires the presence of the
5’-ATTTA portion of HBS since mutation of the HBS site on this AT-rich site abolished binding
ACS Paragon Plus Environment
2
4

Journal of Medicinal Chemistry
Page 26 of 83
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
of DB818 and DB1055 to the DNA. More precisely, the interaction of DB818 or DB1055 in the
minor groove of DNA (Figures 2 and 5) is an interesting point regarding the mode of DNA
binding of the HOXA9 transcription factor that implies the correct orientation of three alpha-
helices of HOXA9 homeodomain. The HOXA9 helix-3 binds in the major groove of the DNA
helix, whereas the amino-terminal part of its homeodomain deeply bound in the minor groove of
the AT-rich portion of the HOXA9-binding site.42 This interaction with the minor groove is
associated with direct contact of Arg-5 with A on the lower strand and T on upper strand of the
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
’
3
HOXA9/DNA binding site (upper strand/lower strand: A T T T A /5’-T A A A T ). Water-
1
2
3
4
5
5’ 4’ 3’ 2’ 1’
mediated additional contacts of Arg-5 with A of the lower strand, and water-mediated contacts
3’
4
2
of Arg-2 with both A and T on lower strand reinforce the DNA binding. This AT-rich
4
’
5’
portion is the exact position where DB818 and DB1055 are lying in the minor groove, which
allows them to block the correct positioning of HOXA9 homeodomain to its cognate site.
HOXA9 binding to the DNA is associated with a DNA bending of ~20° resulting from minor
groove compression in order to allow deep contacts of helix-3 to an enlarged major groove on
the opposite side. Such DNA distortion towards the minor groove results from interaction of
amino acids of the amino-terminal arm (Pro-7 and Tyr-8) with helix-1 of the homeodomain of
HOXA9 (Thr-13). Binding of DB818 and DB1055 to the HOXA9 binding site may also not be
compatible with minor groove compression by helix-3 binding in the major groove and may
enhance HOXA9/DNA binding inhibition by DB818/1055. At the cellular level, both DB818 and
DB1055 control HOXA9-directed transcription based on luciferase gene expression assays
(Figure 6B) and induce, among other derivatives, cell death, cell cycle arrest and cell
differentiation in a model of Hoxa9-transformed murine MigA9 leukemia cell line (Tables 2-4,
Figure 7 and Supplementary Figure S6). Interestingly, DB828, used here as a control as an
ACS Paragon Plus Environment
2
5

Page 27 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
unselective minor groove DNA binding drug that failed to interact with HOXA9 cognate
sequence (Figure 3) and to inhibit HOXA9/DNA interaction in vitro (based on ELISA and
EMSA experiments, Figure 1) and in vivo (based on luciferase assays, Figure 6B). DB828 also
failed to alter cell survival (MTS, Table 2), to induce cell death (PI-positive or sub-G1
populations analysis, Table 2) and MigA9 cell differentiation (anti-clonogenic activities and
CD11b/CD14 differentiation markers expression, Table 4 and Figure 7).
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
Up to now, no direct HOXA9/DNA binding inhibitors have been discovered. Only indirect
inhibition strategies have been developed. The first approach is by targeting proteins of the MLL
complex that controls HOXA9 gene expression. This is the case with inhibitors of DOT1L such
as EPZ-5676 (pinometostat, currently in a phase I clinical trial) that reduce HOXA9 gene
expression and induce cell differentiation58,59 or of inhibitors of the menin/MLL interaction60,61
.
The second approach is the targeting of the HOX/PBX interface of multiple HOX proteins with
their co-factors of the PBX family by the HXR9 peptidomimetic developed against a variety of
solid tumors and hematopoietic malignancies.62
Some other heterocyclic diamidines were previously identified as transcription factors
inhibitors, DB293 against Pit-1 and Brn-3,43 DB1255 against ERG44,45 and DB2113 against
PU.122,63 but none were previously evaluated against the HOXA9 transcription factor.
Identification of DB818 and DB1055 as HOXA9/DNA binding competitors and inducers of
granulocyte/monocyte differentiation offers an exciting new opportunity to target the HOXA9
transcription factor. HOXA9 is one of the master regulators of the differentiation blockade in
acute myeloid leukemia28,29,32,36, being at the crossroads of many cell signaling perturbations
resulting from well-characterized mutations or translocations defining AML subgroups such as
ACS Paragon Plus Environment
2
6

Journal of Medicinal Chemistry
Page 28 of 83
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
for instance MLL translocations, MYST3-CREBBP or NUP98-NSD1 fusion proteins and NPM1
or EZH2 mutations.30,31 Therefore, treatment of such AML subtypes with DB818/DB1055 needs
to be further evaluated in the future to offer new potential differentiation therapies. HOXA9 is
also associated with different solid tumor models such as glioblastoma, gastric and prostate
cancers23–26 against which DB818, DB1055 and/or derivatives might also be evaluated to
propose future therapeutic strategies. This DNA targeting approach to inhibit HOXA9/DNA
binding and function is a very promising strategy with the potential to develop new anti-cancer
treatments and particularly new differentiation therapies in leukemia.
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
Experimental Section
1
. Chemistry
References for previously reported diamidines are provided in the Supplementary Information.
The synthesis of the previously unreported diamidines follows below and in the Result and
1
Discussion section. Representatives HNMR spectra for the diamidines are provided in the
Supplementary Information. The purity (>95%) of all compounds evaluated in this study was
1
established by elemental analysis and H high resolution NMR spectroscopy. Elemental
Analysis was performed by Atlantic Microlab Inc., Norcross, GA.
General procedure for the synthesis of compounds 1a-c
6
4
To a stirred solution of the bromo compounds (10 mmol), 5-(dimethoxymethyl)furan-2-yl-tri-
n-butylstannane65 (10 mmol) in 30 mL anhydrous 1,4-dioxane
was added tetrakis-
o
triphenylphosphine palladium (300 mg). The stirred mixture was warmed to 90–100 C for 16 h.
ACS Paragon Plus Environment
2
7

Page 29 of 83
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The solvent was removed under reduced pressure; the residue was extracted using 200 mL of
dichloromethane from aqueous solution containing 3 mL of concentrated ammonia, then washed
with water, and passed through celite. The organic layer was evaporated and the oily residue was
mixed with 15 mL 2.0M HCl, and stirred for 2 h, the separated precipitate was filtered off and
washed with ethanol. The product was crystallized from ethanol.
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
o
1
3
-Benzyloxy-4-(5-formylfuran-2-yl)-benzonitrile (1a).73% yield, mp 164-165 C. H NMR
(
DMSO-d );  5.38 (s, 2H), 7.16 (d, J = 3.6 Hz, 1H), 7.40-7.46 (m, 3H), 7.51-7.56 (m, 3H), 7.61
6
(d, J = 3.6 Hz, 1H), 7.80 (s, 1H), 8.00 (d, J = 8.1 Hz, 1H), 9.63 (s, 1H). MS (m/z, rel.int.); 303
+
(M , 40), 275 (25), 198 (10), 91 (100).
o
1
6
-(5-Formylfuran-2-yl)nicotinonitrile (1b). Yield 60%, mp 204-205 C (EtOH). H NMR
(
DMSO-d );  7.56 (d, J = 3.6 Hz, 1H), 7.71 (d, J = 3.6 Hz, 1H), 8.07 (d, J = 8.4 Hz, 1H), 8.44
6
+
(dd, J = 8.4, 2.1 Hz, 1H), 9.09 (d, J = 2.1 Hz, 1H), 9.72 (s, 1H). MS (m/z, rel.int.); 199 (M +1,
1
00).
o
1
5
-(5-Formylfuran-2-yl)picolinonitrile (1c). Yield 62%, mp 191-192 C (EtOH). H NMR
(DMSO-d );  7.66 (d, J = 3.6 Hz, 1H), 7.74 (d, J = 3.6 Hz, 1H), 8.17 (d, J = 8.1 Hz, 1H), 8.46
6
+
(
dd, J = 8.1, 2.1 Hz, 1H), 9.25 (d, J = 2.1 Hz, 1H), 9.71 (s, 1H). MS (m/z, rel.int.); 198 (M , 100),
1
97 (70), 141 (30).
2-{5-[(2-Benzyloxy)-4-cyanophenyl]-furan-2-yl}-1H-benzimidazole-6-carbonitrile(2a).
Prepared adopting reported methodology.66 Yield 81%, mp 247-248.5 C (DMF). H NMR
o
1
(
DMSO-d );  5.39 (s, 2H), 7.18 (d, J = 3.6 Hz, 1H), 7.40-7.49 (m, 4H), 7.55-7.67 (m, 4H), 7.74-
6
ACS Paragon Plus Environment
2
8