Discovery and development of extreme selective inhibitors of the ITD and D835Y mutant FLT3 kinases

Article abstract of DOI:10.1016/j.ejmech.2019.111710

Aberrant activation of FMS-like tyrosine receptor kinase 3 (FLT3) is implicated in the pathogenesis of acute myeloid leukemia (AML) in 20–30% of patients. In this study we identified a highly selective (phenylethenyl)quinazoline compound family as novel potent inhibitors of the FLT3-ITD and FLT3-D835Y kinases. Their prominent effects were confirmed by biochemical and cellular proliferation assays followed by mice xenograft studies. Our modelling experiments and the chemical structures of the compounds predict the possibility of covalent inhibition. The most effective compounds triggered apoptosis in FLT3-ITD AML cells but had either weak or no effect in FLT3-independent leukemic and non-leukemic cell lines. Our results strongly suggest that our compounds may become therapeutics in relapsing and refractory AML disease harboring various ITD and tyrosine kinase domain mutations, by their ability to overcome drug resistance.

Full text of DOI:10.1016/j.ejmech.2019.111710

Journal Pre-proof
Discovery and development of extreme selective inhibitors of the ITD and D835Y
mutant FLT3 kinases
Ferenc Baska, Anna Sipos, Zoltán Őrfi, Zoltán Nemes, Judit Dobos, Csaba Szántai-
Kis, Eszter Szabó, Gábor Szénási, László Dézsi, Péter Hamar, Mihály T. Cserepes,
József Tóvári, Rita Garamvölgyi, Marcell Krekó, László Őrfi
PII:
S0223-5234(19)30862-1
DOI:
https://doi.org/10.1016/j.ejmech.2019.111710
EJMECH 111710
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 15 July 2019
Revised Date: 2 September 2019
Accepted Date: 16 September 2019
Please cite this article as: F. Baska, A. Sipos, Zoltá. Őrfi, Zoltá. Nemes, J. Dobos, C. Szántai-Kis,
E. Szabó, Gá. Szénási, Láó. Dézsi, Pé. Hamar, Mihá.T. Cserepes, Jó. Tóvári, R. Garamvölgyi, M.
Krekó, Láó. Őrfi, Discovery and development of extreme selective inhibitors of the ITD and D835Y
mutant FLT3 kinases, European Journal of Medicinal Chemistry (2019), doi: https://doi.org/10.1016/
j.ejmech.2019.111710.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
©
2019 Elsevier Masson SAS. All rights reserved.

Discovery and development of extreme selective inhibitors of the
ITD and D835Y mutant FLT3 kinases
1
, ┴
1, ┴
2 ┴
1
1
1
Ferenc Baska , Anna Sipos , Zoltán Őrfi , Zoltán Nemes , Judit Dobos , Csaba Szántai-Kis ,
3
4
4,5
4
6
Eszter Szabó , Gábor Szénási , László Dézsi , Péter Hamar , Mihály T. Cserepes , József
6
1
7
1,7,8 *
Tóvári , Rita Garamvölgyi , Marcell Krekó , László Őrfi
1
Vichem Chemie Research Ltd., 1022 Budapest, Hungary
2
Department of Molecular Biology, Max Planck Institute of Biochemistry, 82152 Martinsried,
Germany
3
4
5
st
1
Department of Paediatrics, Semmelweis University, 1083 Budapest, Hungary
Institute of Pathophysiology, Semmelweis University, 1089 Budapest, Hungary
Nanomedicine Research and Education Center, Semmelweis University, 1089 Budapest,
Hungary
6
,
Department of Experimental Pharmacology, National Institute of Oncology 1122 Budapest,
Hungary
7
Department of Pharmaceutical Chemistry, Semmelweis University, 1085 Budapest, Hungary
8
Drug Research Co. 1161 Budapest, Batthyány u. 92.
┴
These authors contributed equally to this work.
*
Corresponding author
László Őrfi associate professor
Department of Pharmaceutical Chemistry, Semmelweis University
Hőgyes Endre u. 9, 1092 Budapest, Hungary
Tel: + 36 30 231 6885 Fax: + 36 1 217 0891
e-mail: orfi.laszlo@pharma.semmelweis-univ.hu

ABSTRACT
Aberrant activation of FMS-like tyrosine receptor kinase 3 (FLT3) is implicated in the
pathogenesis of acute myeloid leukemia (AML) in 20-30% of patients. In this study we identified
a highly selective (phenylethenyl)quinazoline compound family as novel potent inhibitors of the
FLT3-ITD and FLT3-D835Y kinases. Their prominent effects were confirmed by biochemical
and cellular proliferation assays followed by mice xenograft studies. Our modelling experiments
and the chemical structures of the compounds predict the possibility of covalent inhibition. The
most effective compounds triggered apoptosis in FLT3-ITD AML cells but had either weak or no
effect in FLT3-independent leukemic and non-leukemic cell lines. Our results strongly suggest
that our compounds may become therapeutics in relapsing and refractory AML disease harboring
various ITD and tyrosine kinase domain mutations, by their ability to overcome drug resistance.
INTRODUCTION
The FMS-like tyrosine receptor kinase 3 is a cell surface receptor belonging to the third class of
tyrosine kinase family receptors[1] In the human body FLT3 is expressed by immature blood
.
cells to the highest extent and plays a pivotal role in the differentiation and survival of
hematopoietic stem cells in the bone marrow[2–4]. Like other receptor kinases, it consists of
three main structural elements: the extracellular, transmembrane and intracellular domains. The
intracellular part contains the juxtamembrane domain and the two kinase domains which are
separated by the small kinase insert domain[5,6]. The ATP-binding site of the kinase is located
between the N- and C-terminal lobes surrounded by the common structural elements (hinge
region, DFG motif, activation loop and αC-helix; Figure 1A). The juxtamembrane domain plays
2

an important role in the autoinhibition of FLT3 and mutations often occur in this part of the
protein[7].
Under physiological conditions the protein can be found in the cell membrane in a monomeric
form. The binding of the FLT3 ligand (FL or FLT3L) leads to its activation, causing
conformation changes (dimerization) and phosphorylation of the intracellular part[8]. The
activated FLT3 kinase turns on various signal transduction pathways (Figure 1B) to convey the
information to the nucleus, which pathways control transcription, translation, differentiation and
apoptosis. The two most important pathways are the PI3K/Akt and RAS/RAF/MEK/ERK [9].
Due to its function, overexpression and/or mutation of the FLT3 kinase are one of the most
commonly observed phenomena in malignant tumors of the hematopoietic system and they are
potential risk factors in AML patients[10–12]. The most frequent activating mutation of FLT3
(~23% in AML) is the internal tandem duplication (ITD) which extends the length of the
juxtamembrane domain and interrupts the autoinhibition mechanism of the kinase[13,14]. The
length of the ITD mutation varies from patient to patient but it is usually 3-400 base pairs[15].
The second most common mutation (~7% in AML) of FLT3 is the D835Y point mutation, which
alters the amino acid sequence of the activation loop causing enhanced protein activation[16,17].
FLT3-activating mutations critically regulate leukemic transformation by accelerating
proliferation and suppressing apoptosis and the FLT3-ITD mutations are known to confer poor
prognosis, while the prognostic impact of D835Y mutation is controversial. Moreover FLT3 and
its mutations were identified as the most important driver genes not only in AML, but in ALL
and CML as well[18,19]. These findings highlight FLT3 as a highly attractive target for drug
development.
3

Most of the known FLT3 inhibitors are multikinase inhibitors, which demonstrated promising
efficiency in clinical trials but they have unfavorable side effects and/or poor pharmacokinetic
profiles[20].
Although the quinazoline-based tandutinib (Figure 2) inhibits FLT3 in the nanomolar range
(
IC50= 220 nM), it also binds to other receptor kinases (PDGFRß, FGFR, VEGFR, c-Kit) in the
same concentration range (IC = 170-200 nM)[21]. Despite its poor pharmacokinetic properties,
5
0
tandutinib was investigated in clinical trials in combination therapy for other cancer types as
well[22–24].
Compound KW-2449 (Figure 2) was developed as an FLT3 kinase inhibitor (IC = 6.6 nM) but
5
0
further profiling revealed that the compound is active also against many other kinase
enzymes[25]. Because of its poor pharmacokinetic profile KW-2449 showed weaker effect in
clinical trials than expected and under physiological conditions the inhibition of FLT3 was
insignificant[26]. Sunitinib, an oxindole based multikinase inhibitor, got the approval of the U.S.
Food and Drug Administration (FDA) against renal cell carcinoma and gastrointestinal stromal
tumor in 2006[27]. According to selectivity studies sunitinib inhibits approximately 150 kinases
including FLT3 (wt FLT3 IC = 250 nM, FLT3-ITD IC = 50 nM, FLT3-D835Y IC = 30 nM)
5
0
50
50
and showed remarkable results in phase I trials[28–30]. In recent years sunitinib has also been
tested in combination with standard chemotherapy in patients with FLT3-mutated AML[31].
The natural product staurosporine was the initial lead compound for many semi-synthesized
inhibitors including midostaurin and lestaurtinib (Figure 2). Both compounds inhibit a broad
spectrum of kinases[32,33]. In preclinical experiments midostaurin showed an IC value of 30
5
0
nM on FLT3-ITD enzyme and induced apoptosis in high proportions of FLT3-mutant cells[34].
Midostaurin results were promising and it has been approved by the FDA for the treatment of
4

FLT3 mutation-positive adult AML patients in 2017 April [35–37]. Lestaurtinib (developed by
Teva/Cephalon) showed not only stronger inhibitory effect on FLT3 kinase (IC50= 2-3 nM) but
also improved cytotoxicity on cell lines expressing wild type and mutant FLT3[38]. Results of
lestaurtinib suggest a well-tolerated compound with mild side-effects, and it is still in clinical
development[39,40].
Although the styryl-pyrimidine derivative ENMD-2076 (Figure 2) was originally published as an
Aurora inhibitor (IC = 14 nM), additional targets have been identified with kinase selectivity
5
0
profiling. ENMD-2076 inhibits RET, Src, VEGFR2, PDGFRα, FMS with the same efficacy
IC50= 10-60 nM), but it is a ten-fold better FLT3 inhibitor (IC50= 1.86 nM)[41].
(
Quizartinib (Figure 2), developed by Ambit Biosciences Corporation, is an inhibitor which was
originally designed against FLT3. Notably, quizartinib displayed better selectivity over kinases
than the previously described inhibitors. The IC of quizartinib is 1.1 nM on FLT3-ITD and 4.2
5
0
nM on the wild type kinase[42]. In clinical trials the compound showed favorable and very
encouraging effect, especially in the case of FLT3-ITD patients[43]. However, acquired drug
resistance reduces the applicability of the compound. It was previously published that
quizartinib-treated FLT3-ITD patients developed D835Y and F691L point mutations in the
activation loop[44] and in the gatekeeper residue respectively, which decreased affinity of the
compound. Quizartinib is a type II kinase inhibitor which binds to the DFG-out inactive
conformation of the protein. The D835Y point mutation destabilizes the inactive state and
quizartinib is not able to bind to FLT3 [45]. Although other FLT3 inhibitors like midostaurin and
lestaurtinib are type I inhibitors and can bind to the active conformation, they are staurosporine
analogues with low selectivity. Thus, there is a high demand for a selective FLT3-ITD/FLT3-
D835Y inhibitor.
5

PLX3397 (Figure 2), a triple kinase inhibitor of CSFR1R (enzymatic IC =13nM), c-KIT
5
0
(enzymatic IC50=27nM) and FLT3-ITD (enzymatic IC50=11nM) inhibits FLT3 signaling in
FLT3-ITD mutant cells (e.g. biochemical IC50=18nM in MV4-11 cells)[46]. PLX3397 is a novel
FLT3 inhibitor that overrides F691L. Based upon its encouraging preclinical activity, a phase 1/2
safety and efficacy study of orally administered PLX3397 was initiated in adults with relapsed or
refractory FLT3-ITD acute myeloid leukemia (AML) [47].
The investigational type I inhibitor, crenolanib (Figure 2), is a highly selective and potent FLT3
tyrosine kinase inhibitor (TKI) with activity against the FLT3-ITD mutants as well as against the
FLT3-D835Y point mutants that are resistant to quizartinib; however, it shows a loss of potency
against the gatekeeper mutation F691L and shows moderate activity against c-KIT[48,49].
Correlative data from an ongoing clinical trial demonstrate that crenolanib can achieve sufficient
plasma level to inhibit both FLT3-ITD and resistance-conferring FLT3-D835 mutants in AML
patients, in vivo[50].
The development of drug resistance during treatment of hematologic malignancies has been a
challenging issue for TKIs[19,51,52]. Various factors have been identified in the background of
resistance against FLT3 inhibitors including point mutations, plasma inhibitory activity (PIA),
protective effect on bone marrow stromal cells and high levels of FLT3 ligand (FL). Point
mutations within the kinase domain of FLT3-ITD especially at positions N676, F691 and D835
lead to substantial resistance to quizartinib and midostaurin[19,53]. Additional mutations causing
quizartinib-resistance have been described using in vitro models[54].
Myelosuppression can be a challenging clinical parameter to evaluate in any trial that involves
AML patients. Most AML patients enrolled on early phase studies are already myelosuppressed
from the burden of their disease. Nonetheless, suppression of bone marrow function to some
degree is almost always a feature of certain TKIs, which have activity against c-KIT[49], as c-
6

KIT is essential for normal erythropoiesis and megakaryocyte function[55]. For this reason it is
important to determine a therapeutic index between the targeted receptor and c-KIT for TKIs
used to treat hematologic malignancies in order to maintain normal hematopoiesis and improve
the outcomes of treatments[56]. Based on these observations there is an unmet need for next-
generation FLT3 inhibitors that overcome drug resistance and can reduce myelosuppression.
Here we report the development of novel and selective FLT3-ITD and FLT3-D835Y inhibitors
having high affinity in vitro for the quizartinib-resistant FLT3-ITD/D835Y double mutant
enzyme. Another unique advantage of our compounds is that they have no or negligible activity
against c-KIT. Thus, we believe that compound III is a promising drug candidate with strong
therapeutic potential to overcome drug resistance and to prolong disease-free survival of AML
patients.
RESULTS AND DISCUSSION
TM
The Nested Chemical Library [57] of Vichem was used to discover hit compounds against the
FLT3 kinase. The phenylethenylpyrimidine compound ENMD-2076 (Figure 2) was described as
a low nanomolar FLT3 inhibitor[41]. The kinase inhibitory profile of a similar quinazoline-based
compound (CP-31398, Figure 3A) had not been investigated before. Previously it was published
as a p53 activator molecule[58]. Since structural similarity can be found between ENMD-2076
and CP-31398 we measured the FLT3-ITD inhibitory activity of CP-31398, which was found to
be 3.81 ± 0.87 (IC ). Therefore, we chose our CP-31398-related phenylethenylquinazoline
5
0
derivative compound XIII (Figure 3A) for target identification and selectivity profiling study.
®
The kinase inhibitory profile of compound XIII was determined by the KINOMEscan binding
assay (DiscoveRx Corporation) and the FLT3 kinase was identified as the primary target (wt
FLT3 = 81% displacement) of the molecule (Figure 3C). Moreover, in the case of mutant FLT3
7

kinases we found > 90% displacement: FLT3-D835H = 97%, FLT3-ITD = 96%, FLT3-D835Y =
5%, FLT3-N842I = 95%, at 5µM concentration. Only lower affinity was detected against a few
9
kinases (CSNK1A1 77%, GAK 65% and RIOK1 63%), but it is important to note that the hit
compound did not bind either to the members of the third class of the receptor tyrosine kinase
family (c-Kit, CSFR, PDGFR), nor to any other receptor kinases. This is a very promising result
considering the fact that some marketed kinase inhibitors hit more than 30% of the whole
kinome[59].
Chemistry
Using the synthetic route (Figure 4) developed for the preparation of CP-31398 (see
Supplementary data) we synthesized additional 122 (phenylethenyl)quinazoline derivatives in
order to establish a detailed structure-activity relationship.
First, the 2-methylquinazolone intermediates were prepared in good overall yields using the
appropriate anthranilic acids, acetic anhydride and concentrated ammonium hydroxide solution.
These intermediates were converted into phenylethenylquinazolones with various benzaldehydes
via aldol-type condensation on the 2-methyl group. Using a microwave reactor not only increased
the yield of these reactions, but also shortened the reaction time to 1.5 – 2 hours. Next, these
phenylethenylquinazolones were mixed with phosphoryl chloride and stirred at 90 °C in the
presence of catalytic amount of dimethylformamide. The formed 4-chloro-
phenylethenylquinazolines are unstable: therefore all derivatives were freshly synthesized and
used in the last step without any analytical confirmation.
In the end, 4-amino-phenylethenylquinazolines were prepared from the imidoyl chlorides via
nucleophilic aromatic substitution (S Ar). The instability of the 4-chloro derivatives resulted in a
N
mixture of the 4-hydroxyl- and 4-alkylamino derivatives, even under dry reaction conditions.
8

Therefore the crude products were purified by column chromatography to separate the desired
phenylethenyl)quinazoline compounds with 10-60% yield
(
Biochemical properties and selectivity of the compounds
In the first instance all compounds were tested with fluorescence polarization based IMAP™
(
Immobilized Metal Assay for Phosphochemicals) kinase assay which was suitable for MTS/HTS
screening at one concentration (10 µM), then only the best compounds were selected for the
further IC measurements. The best FLT3 inhibitors synthesized are summarized in Table 1.
5
0
Based on the results of the biochemical assays we concluded that the best molecules are low-
nanomolar inhibitors of the FLT3-ITD and FLT3-D835Y kinases, moreover, they showed
selectivity over the wild-type FLT3 (Table 1). The significance of this finding is that the
compounds also have a verified high inhibitory effect on D835Y mutant protein. Although the
D835Y point mutation occurs less frequently (~7%) in AML, this mutation has been identified to
play a role in acquired resistance against FLT3 inhibitors. Therefore, we also analyzed the
inhibitory effect of tandutinib and quizartinib on the D835Y mutant FLT3. As expected,
quizartinib showed an activity of only 1.33 ± 0.28 µM (IC ) against the D835Y mutant enzyme
5
0
which confirmed the previously described discovery that the mutation destabilizes the inactive
(DFG-out) conformation and obstructs the binding of quizartinib[43]. Our lead molecules
(indicated in grey in Table 1) are highly effective on FLT3-D835Y in the sub-micromolar range,
not to mention the improved selectivity of the compound family over quizartinib[59].
Structure-activity relationship (SAR) study
The large number of the synthesized molecules gave us the opportunity to establish a detailed
structure-activity relationship, based on the FLT3-ITD biochemical assay results and molecular
9

docking experiments. The measured IC values for all prepared compounds are shown in the
5
0
Supplementary data, Table 1. The quinazoline scaffold (Figure 4) is essential for the FLT3
1
inhibitory effect and the core ring can be unsubstituted or monosubstituted (R ), while the
1
2
1
th
disubstituted (R , R ) compounds were ineffective. Incorporation of a halogen atom (R ) at the 6
position did not alter the effect substantially. However, the fluorinated derivatives had lower
activities than the molecules containing bromine or chlorine at this position. Compounds having
th
th
substituents in the 7 and 8 positions were ineffective, possibly due to reasons described in the
molecular modelling paragraph. The benzo[g]quinazoline-based molecules condensed with
3
benzaldehydes were also ineffective. The R group can be substituted with a phenyl or a
thiophene ring. In the case of the unsubstituted quinazolines, the substituted phenyl derivatives
were effective but the thiophene analogues did not show any activity. However, in case of the
tricyclic compounds, the thiophene derivatives were active while the substituted phenyl
3
derivatives were completely inactive. The R group has to be a monosubstituted aromatic ring for
the strongest inhibition. Derivatives containing a 3,4-disubstituted phenyl groups showed reduced
effect and the 3,4,5-trisubstituted phenyl containing molecules were completely ineffective. The
3
para substitution of the R phenyl ring was the most preferred; substituents in the meta position
greatly reduced the inhibitory effect. These observations are explained well by the inspection of
the size and shape of the binding site that shows unfavourable properties for the binding of
th
analogues bearing substituents other than the 4 position. The best substituents were the
4
isopropyl, methylsulfonyl and methylsulfanyl functional groups. The R group has to be a side
chain that contains a tertiary amino group. The shorter chains (N,N-dimethylethane-1,2-diamine,
N,N-dimethylpropane-1,3-diamine) were much better than the longer and larger (N,N-
1
1
diethylpropane-1,3-diamine, N ,N -diethylpentane-1,4-diamine, morpholine, N-methylpiperazine
1
0

or pyrrolidine) ones. The presence of the side-chain is essential for the activity because
compounds without it had no inhibitory effect (see Supplementary data, Table 2).
Antiproliferative activity in leukemia cells
After the selective and potent inhibitory effect of the compounds was demonstrated in different
biochemical assays (FLT3-WT, FLT3-ITD and FLT3-D835Y), the cell proliferation inhibitory
activity was tested in a panel of leukemic and non-leukemic cells. The compounds inhibited the
cellular proliferation of the FLT3-ITD bearing leukemic (MV4-11 FLT3-dependent) cell line
with an IC50 ranging from 0.01 to 2.34 µM. FLT3-ITD mutation was identified as a driver
mutation in this cell line[29] and the cells acquired growth advantage due to the ITD mutation.
FLT3 signal independent leukemic cell lines (U937[60] and K562[61]) and other non-leukemic
cell line were either weakly or not inhibited at all by the compounds (Table 2). Overall, these
compounds are potent and highly selective inhibitors of the proliferation of FLT3-driven cells.
Because some of our molecules demonstrated strong inhibition both in biochemical and in
cellular assays, the selectivity of these lead compounds had to be verified. The 4-isopropyl
derivative compound III demonstrated good solubility and membrane penetration values (Table
3
), therefore, this molecule was chosen for the selectivity validation by KINOMEscan™ binding
assay.
The remarkable selectivity profile of compound III (Figure 5) did not change due to the
structural modifications (compared to compound XIII - Figure 3A and 3B) and even improved
the FLT3 binding affinity. The results also confirmed the high binding affinity for the double
mutant FLT3-ITD/D835Y quizartinib resistant enzyme (Figure 5).
Next, we determined the mechanism of the inhibition. Most of the quinazoline-based kinase
inhibitors are ATP-competitive, therefore, we hypothesized that our compounds are ATP-
1
1

competitive as well. In order to prove this hypothesis, the IC values of the lead compounds
5
0
m
were measured at three different ATP concentrations: at the K ATP concentration (5.1 µM for
the FLT3-ITD enzyme), and at a lower (0.51 µM) and at one higher (51 µM) concentration.
According to the Cheng−Prusoff equation[62], the increasing ATP concentration will increase the
IC value of the compounds if the they are ATP-competitive inhibitors.
5
0
The results (Figure 6) demonstrated that our compounds have higher IC50 at higher ATP
concentration so compounds III, VII and IX are ATP-competitive.
Due to the high homology of the ATP binding sites, the ATP-competitive inhibitors are generally
not selective for their target proteins; therefore there is no particular explanation for the extreme
selectivity of the compounds. We suppose that the cause of the selectivity may be attributable to
favorable combination of non-covalent and covalent binding to FLT3. Covalent kinase inhibitors
can target a nucleophilic group at a special position in the binding site (e.g. cystein residue).
Other kinases or family members will not react, because they do not have the targeted
nucleophilic residue in the appropriate position[63]. It is important to note that, of the class III
receptor tyrosine kinases, only FLT3 has an adequately positioned cysteine residue in the hinge
region for the covalent reaction to occur. The Aurora kinase isoforms do not contain a similarly
positioned cysteine either [64], which may also explain why our compounds have no such
inhibitory activity on Aurora described for ENMD-2076, despite structural similarities.
Our developed FLT3 inhibitors contain a double bond between the two aromatic rings and this
ethenyl group can act as a Michael acceptor. To prove this hypothesis we used 2-mercaptoethanol
(ME) and N-acetylcysteine (AcCys) as a Michael donor and performed chemical experiments to
+
model this reaction. The addition product ([M+H] 439.3) was detected first after 24 hours of
incubation with ME (ratio 1:1) at room temperature (Supplementary data, Figure 1). By using
1
2

excessive amount of ME (ratio 1:50), the addition product appeared after 60 min
Supplementary data, Figure 2). These measurements were repeated with AcCys as well and
(
resulted in very similar results. In case of equivalent quantity (inhibitor and AcCys 1:1), the
+
addition product ([M+H] 524.3) could be detected after 18 hours (Supplementary data, Figure
3
). By using AcCys in excess, 30 min incubation was enough for the reaction to occur
(Supplementary data, Figure 4).
Based on these measurements, our compounds may likewise form covalent bonds with the ATP
binding site of the FLT3 kinase and this could partly answer the extreme selectivity. According to
a recent report on kinase inhibitors, they can show ATP-competitive kinetics while binding
covalently to the active site[65].
In silico modelling of the covalent binding mode
Molecular modelling was applied to further investigate the binding mode of the compounds.
Reversible glide docking was conducted on all of the derivatives. Covalent docking was
performed on the ligands using a custom made covalent docking mechanism protocol[66].
Our theory is that a covalent reaction occurs between the ethenyl group and Cys694 of the hinge
region after initial binding of the molecule. Such mode of action was reported by Nijmeijer,
Engelhardt et al. of a hH4 partial agonist compound with a 2-ethenylpyrimidine structure, that is
able to form covalent adducts with thiol containing molecules glutathione and cysteine ethyl
ester[67]. No similar reaction was observed in the case of the 2-methylpyrimidine derivative,
suggesting a covalent mechanism of the ethenyl compound. Alkenyl and alkynyl substituted
heteroarene compounds and their ability to form covalent bonds with cysteines were described in
a recent overview of covalent warheads by Matthias Gehringer and Stefan A. Laufer[68].
1
3

Our glide docking models indicate a minimal distance of 3 Å between the sulfhydryl group of
Cys694 and the ethenyl carbon of our compounds. The reversible ligand-enzyme complexes
show π-π stacking interactions of the phenyl ring with Phe830 and, to a lesser extent, Phe691.
Either a hydrogen bond or an ionic bond is also observed between the protonated amino group of
the side chain and the carboxylate group of Asp698.
The supposed covalent reaction takes place on the β-carbon of the ethenyl group, analogous to
α,β-unsaturated carbonyl compounds[69] and acrylamide moiety-containing covalent
inhibitors[63]. Covalent docking of the compounds indicate a uniform binding mode of the active
derivatives (Figure 7), with minimal displacement relative to the reversible docking model.
Interactions described above with Phe830 and Asp698 are also present in the covalent models,
while a small movement of the molecules outwards the pocket, towards the solvent is observed.
Docking scores also confirm the favourable energy balance of the covalent reaction
th
(Supplementary Data, Table 3). Models of inactive compounds bearing substituents in the 8
position of the quinazoline ring show a substantial displacement compared to the actives. The
also inactive 6,7-dimethoxy substituted analogues are represented as protonated both on the
amino side chain and on the pyrimidine nitrogen at pH 7.4 simulations, possibly disrupting the
active binding conformation. Benzo[g]quinazoline derivatives mostly overlap with their
quinazoline analogues, although steric clashes between the core and the hinge region prevent
them from tighter binding and properly interacting with the phenylalanines nearby through their
phenyl or thienyl rings. This is in accordance with their reduced inhibitory activity. The models
also suggest that only molecules with para positioned substituents have adequate space inside the
binding site when covalently docked; analogues with multiple phenyl ring substituents do not fit
well in the pocket, which explains their inactivity.
1
4

Compounds induce apoptosis in FLT3-ITD expressing MV4-11 cells
With the reproducible anti-proliferative effect observed for compound III, compound VII and
compound IX against the MV4-11 cell line, their mechanism of action was investigated by flow
cytometry. Apoptosis and necrosis were demonstrated by staining the cells with Annexin V and
propidium iodide. The cells were treated with increasing concentrations of compound III,
compound VII and compound IX to demonstrate the dose-dependent effect on the induced
apoptosis.
The majority of the untreated cells (DMSO was used as control) was viable (82.6 ± 3.6%) and
showed neither Annexin V-FITC nor PI positivity. As expected a dose-dependent increase was
+
-
+
+
observed in early (AN PI ) and late apoptotic cells (AN PI ) (Figure 8A) following treatment
with compound III, compound VII and compound IX.
Caspase-3 is a key downstream effector in the apoptosis pathway therefore Caspase-Glo® 3/7
assay was used for measuring caspase 3/7 activity. Using the assay as a determinant, a significant
elevation of caspase activity was detected after exposure to as little as 0.185 µM compound III,
VII and IX in the FLT3-ITD mutant MV4-11 cells (Figure 8B). This effect was as pronounced
as the effect of cyclophosphamide used at higher concentrations and verified the early stage of
apoptosis induced by the compounds in the MV4-11 cell line in a dose-dependent manner.
Together, the data indicate that compound III, compound VII and compound IX effectively
induce apoptosis.
Determination of ADME properties
The solubility of the compounds was measured at two pH values using HPLC (Table 3). The
compounds showed good solubility at pH 7.4 but some of them were considerably more soluble
at acidic pH of 2.0, which is not surprising due to the secondary and tertiary amino groups in
1
5

their structures. Some molecules and one of the lead compounds (III) showed acceptable
solubility (108 µM) at physiological pH. These results suggest that special compound formulation
is not required for early preclinical development.
Most of the tested phenylethenylquinazoline compounds demonstrated good penetration values
-6
(
e
Table 3). According to our results (P values = 0.36 – 9.61*10 cm/s), we can conclude that
some of these phenylethenylquinazoline derivatives can cross the cell membranes by passive
diffusion.
Pharmacokinetic parameters and in vivo studies of compound III
Compound III has a large volume of distribution (Vss_obs = 4.26 L/kg) and a plasma half-life
(T ½) of 50.6 min (Table 4). Although oral bioavailability (F (%) = 16) was relatively poor,
suggesting high first pass metabolism and/or poor gastrointestinal absorption, compound III
reached sufficiently high plasma concentrations (16.6 ng/ml ± 1.9) after a single oral
administration to exert an effect in a xenograft model.
SCID mice were subcutaneously inoculated with MV4-11 cells. 19 days after the inoculation,
compound III was administered intraperitoneally at a dose of 17 mg/kg on every other day for 3
weeks. By the end of this period, the MV4-11 tumors have reached an average volume of
3
1
393.13 mm . Compared to the vehicle control, the treatment with compound III reduced the
tumor volume and weight significantly, by 48% and 49% in average, respectively (Figure 9).
During the treatment period no additional body weight loss and no death was observed (data
shown in Table 5). These results demonstrated that compound III effectively reduced the tumor
size and was well-tolerated at the dose we employed.
1
6

CONCLUSIONS
In summary, a novel series of phenylethenylquinazoline derivatives were designed and
synthesized by general and affordable methods. The exploration and optimization of the different
substituents positioned on the quinazoline scaffold resulted in the discovery of a new, highly
selective FLT3 inhibitor compound family. To our current knowledge these are selective FLT3
inhibitors, which showed an impressive in vitro efficacy on single mutant enzymes (FLT3-ITD,
FLT3-D835Y) and also had a high affinity for the quizartinib-resistant double mutant (FLT3-
ITD/D835Y) enzyme. We demonstrated with chemical reactions and in silico modelling
experiments that the compounds can form a covalent bond with the thiol side chains of cysteines
and we assume that this mechanism is present when binding to the FLT3 kinase as well. This
phenomenon can probably contribute to the high degree of selectivity. Besides these observations
we detected an ATP-competitive characteristic for compound III.
In addition, the selectivity and specificity were demonstrated in cellular assays and in xenograft
leukemia model. Compound III showed effective antitumor activity in vivo, by significantly
reducing tumor volume and weight after treatment on every other day for 3 weeks. In conclusion,
these molecules should be evaluated in further drug discovery and development studies to treat
FLT3-ITD bearing, quizartinib-resistant AML.
ACKNOWLEDGMENTS
The authors would like to thank Eszter Illyés, Ildikó Szilágyi and Sándor Boros the HPLC-MS
and NMR measurements and structure elucidations. We would also like to thank to Nora Breza,
Péter Bánhegyi and Peter Markó for their technical support. JChem for Excel was used for
structure based property calculation, JChem for Excel 16.10.1000.1215, 2016, ChemAxon
1
7

(http://www.chemaxon.com) This research was funded by the National Research, Development
and Innovation Office ( K116295 and KFI_16-1-2017-0439, J.T.), Hungary.
AUTHOR CONTRIBUTIONS
FB designed and synthesized the compounds. AS set and performed the FLT3 biochemical assays,
cellular and solubility experiments. ZŐ discovered the high selectivity to FLT3 of an early hit
compound and accomplished the flow cytometric and caspase 3/7 assay measurements on MV4-
1
1 cells. ZN measured the solubility of the molecules. JD performed cellular experiments. CsSzK
supervised the in vitro FLT3 kinase assays and ESz did the instrumental analysis in the PK
studies. GSZ, LD and PH performed the PK studies on mice. MTC and JT managed and
performed the in vivo xenograft model. RG helped in the chemistry and designed the figures for
the paper. MK accomplished the in silico molecular modelling experiments and covalent docking.
LŐ initiated, supervised and directed the project and prepared the paper with FB, AS and ZŐ.
CONFLICT OF INTEREST
The authors declare no competing financial interest.
ASSOCIATED CONTENT
Supporting information
1
Preparation, measured melting points, LCMS, H NMR, selectivity profiling and in vitro assay
data of the compounds.
EXPERIMENTAL SECTION
Chemistry (Detailed chemistry of the highlighted compounds)
General synthesis method of 2-methyl-4H-3,1-benzoxazin-4-one derivatives (Figure 4, a)
1
8

The 2-aminobenzoic acid derivatives (1 mmol) were dissolved in acetic anhydride (1 ml) and the
reaction mixture was stirred at reflux temperature (140 °C) for 4 hours. The solvent was then
evaporated and the crude product was washed with hexane/ether 2:1 and filtered out. These
products were used without further purification or analytical measurements in the next reaction.
General synthesis method of 2-methylquinazolin-4(3H)-one derivatives (Figure 4, b)
The benzoxazin-4-one intermediates (1 mmol) were suspended in concentrated ammonium
hydroxide (1 ml) and were stirred at room temperature overnight. 10% sodium hydoxide solution
was added to the reaction mixtures until resulting transparent solutions. The pH was adjusted to 7
with acetic acid. The products precipitated from the solution, were filtered out and were washed
with water. The solid compounds were dried under vacuum overnight.
6
-chloro-2-methylquinazolin-4(3H)-one (1)
Starting material: 10.00 g (58.3 mmol) 2-amino-5-chlorobenzoic acid. Product: 8.76 g white
powder. Yield: 77 %. Mp: 271-272 °C
1
H NMR (300 MHz, DMSO-d ) δ ppm 7.98 (s, 1H); 7.76 (d, J = 8.6 Hz, 1H); 7.57 (d, J = 8.7 Hz,
6
1
H); 2.34 (s, 3H)
LCMS (ESI) m/z 195.0 (M+H)+; 192.9 (M-H)-; calculated: 194.0 Rt: 2.43 min.
-fluoro-2-methylquinazolin-4(3H)-one (2)
6
Starting material: 5.00 g (32.2 mmol) 2-amino-5-fluorobenzoic acid. Product: 3.50 g white
powder. Yield: 61 %. Mp: 258-259 °C
1
H NMR (300 MHz, DMSO-d ) δ ppm 12.29 (bs, 1H); 7.68 (m, 3H); 2.34 (s, 3H)
6
+
-
LCMS (ESI) m/z 179.2 (M+H) ; 176.9 (M-H) ; calculated: 178.1 Rt: 2.01 min.
-bromo-2-methylquinazolin-4(3H)-one (3)
6
1
9

Starting material: 6.00 g (27.7 mmol) 2-amino-5-bromobenzoic acid. Product: 5.69 g white
powder. Yield: 86 %. Mp: 297-298 °C
1
1
2
H NMR (300 MHz, DMSO-d
6
) δ ppm 8.13 (d, J = 2.3 Hz, 1H); 7.88 (dd, J = 8.7 Hz, J = 2.3 Hz,
1
H); 7.50 (d, J = 8.7 Hz, 1H); 2.33 (s, 1H)
+
-
LCMS (ESI) m/z 239.1 (M+H) ; 236.8 (M-H) ; calculated: 238.0 Rt: 2.52 min.
General synthesis method of 2-[(E)-2-phenylethenyl]quinazolin-4(3H)-one intermediates (Figure
4
, c)
The 2-methylquinazolin-4(3H)-ones (1 mmol) were mixed with benzaldehydes (1.5 mmol) and 1
drop concentrated sulphuric acid was added to the mixtures. The reaction was carried out in
microwave set (Personal Chemistry, Emrys™ Creator, heating power: 150 W, measured pressure
1
-7 bar, reaction time: 1.5 hours) at 190 °C. The crude product was washed with 5% sodium
hydrogen carbonate and filtered out. The products were crystallised from dimethyl formamide
and dried under vacuum.
2
-[(E)-2-(4-fluorophenyl)ethenyl]quinazolin-4(3H)-one (4)
Starting material: 3.00 g (18.7 mmol) 2-methylquinazolin-4(3H)-one; 3.48 g (28.0 mmol) 4-
fluorobenzaldehyde. Product: 3.27 g brown powder. Yield: 66 %. Mp: 286-287 °C (DMF)
1
H NMR (300 MHz, DMSO-d ) δ ppm 12.33 (bs, 1H); 8.11 (d, J = 8.1 Hz, 1H); 7.95 (d, J = 16.1
6
Hz, 1H); 7.83 (t, J = 7.1 Hz, 1H); 7.75 (m, 3H); 7.48 (t, J = 7.2 Hz, 1H); 7.30 (t, J = 8.7 Hz, 2H);
6
.99 (d, J = 16.2 Hz, 1H)
LCMS (ESI) m/z 267.2 (M+H) ; 264.9 (M-H) ; calculated: 266.1 Rt: 3.57 min.
-{(E)-2-[4-(methylsulfanyl)phenyl]ethenyl}quinazolin-4(3H)-one (5)
+
-
2
2
0

Starting material: 3.00 g (18.7 mmol) 2-methylquinazolin-4(3H)-one; 4.27 g (28.0 mmol) 4-
methylsulfanyl) benzaldehyde. Product: 4.58 g dark yellow powder. Yield: 83 %. Mp: 269-
(
2
70 °C (DMF)
1
H NMR (300 MHz, DMSO-d ) δ ppm 12.27 (bs, 1H); 8.13 (d, J = 7.5 Hz, 1H); 7.90 (d, J = 16.1
6
Hz, 1H); 7.79 (t, J = 7.3 Hz, 1H); 7.63 (m, 3H); 7.38 (t, J = 7.3 Hz, 1H); 7.13 (d, J = 7.5 Hz, 2H);
6
.92 (d, J = 16.2 Hz, 1H); 2.51 (s, 3H)
+
-
LCMS (ESI) m/z 295.1 (M+H) ; 293.1 (M-H) ; calculated: 294.1 Rt: 3.82 min.
2
-{(E)-2-[4-(propan-2-yl)phenyl]ethenyl}quinazolin-4(3H)-one (6)
Starting material: 1.20 g (7.5 mmol) 2-methylquinazolin-4(3H)-one; 1.66 g (11.2 mmol) 4-
(propan-2-yl)benzaldehyde. Product: 0.90 g yellow powder. Yield: 41 %. Mp: 240-242 °C
(DMF)
1
H NMR (300 MHz, DMSO-d ) δ ppm 12.32 (bs, 1H); 8.10 (d, J = 7.5 Hz, 1H); 7.93 (d, J = 15.9
6
Hz, 1H); 7.80 (t, J = 6.0 Hz, 1H); 7.67 (d, J = 7.8 Hz, 1H); 7.59 (d, J = 7.6 Hz, 2H); 7.47 (t, J =
.9 Hz, 1H); 7.34 (d, J = 7.9 Hz, 2H); 7.00 (d, J = 16.1 Hz, 1H); 2.90 (m, 1H); 1.23 (m, 6H).
6
+
-
LCMS (ESI) m/z 291.2 (M+H) ; 289.2 (M-H) ; calculated: 290.1 Rt: 4.31 min
2
-{(E)-2-[4-(methylsulfonyl)phenyl]ethenyl}quinazolin-4(3H)-one (7)
Starting material: 1.00 g (6.2 mmol) 2-methylquinazolin-4(3H)-one; 1.71g (9.3 mmol) 4-
(methylsulfonyl)benzaldehyde. Product: 1.30 glightbrown powder. Yield: 64 %. Mp: 310-311 °C
(DMF)
1
H NMR (300 MHz, DMSO-d
6
) δ ppm 12.42 (bs, 1H); 8.15 (d, J = 3.0 Hz, 1H); 8.01 (m, 3H);
.91 (d, J = 8.4 Hz, 2H), 7.83 (t, J = 7.7 Hz, 1H); 7.71 (d, J = 7.9 Hz, 1H); 7.52 (t, J = 7.1 Hz,
H); 7.19 (d, J = 16.2 Hz, 1H); 3.26 (s, 3H)
7
1
2
1

+
-
LCMS (ESI) m/z 327.1 (M+H) ; 325.0 (M-H) ; calculated: 326.1 Rt: 2.96 min.
6
-chloro-2-[(E)-2-(4-fluorophenyl)ethenyl]quinazolin-4(3H)-on (8)
Starting material: 1.10 g (5.6 mmol) 1; 1.04 g (8.4 mmol) 4-fluorobenzaldehyde. Product: 1.07 g
yellow powder. Yield: 64 %. Mp: 334-336 °C (DMF)
1
H NMR (300 MHz, DMSO-d ) δ ppm 12.47 (bs, 1H); 8.04 (d, J = 2.4 Hz, 1H); 7.95 (d, J = 16.2
6
1
2
Hz, 1H); 7.82 (dd, J = 8.7 Hz, J = 2.4 Hz, 1H); 7.72 (m, 3H); 7.30 (t, J = 8.7 Hz, 2H); 6.95 (d, J
16.2 Hz, 1H)
LCMS (ESI) m/z 301.2 (M+H) ; 298.9 (M-H) ; calculated: 300.1 Rt: 4.13 min.
-chloro-2-{(E)-2-[4-(methylsulfanyl)phenyl]ethenyl}quinazolin-4(3H)-one (9)
=
+
-
6
Starting material: 1.10 g (5.6 mmol) 1; 1.27 g (8.4 mmol) 4-(methylsulfanyl)benzaldehyde.
Product: 1.45 g yellow powder. Yield: 79 %. Mp: 317-318 °C (DMF)
1
6
H NMR (300 MHz, DMSO-d ) δ ppm 12.38 (bs, 1H); 8.02 (s, 1H); 7.91 (d, J = 16.1 Hz 1H);
7
2
.80 (d, J = 8.6 Hz, 1H); 7.67 (d, J = 8.7 Hz, 1H); 7.59 (d, J = 8.0 Hz, 2H); 7.33 (d, J = 7.9 Hz,
H); 6.95 (d, J = 16.0 Hz, 1H); 2.52 (s, 3H)
+
-
LCMS (ESI) m/z 329.1 (M+H) ; 327.1 (M-H) ; calculated: 328.0 Rt: 4.38 min.
6
-chloro-2-{(E)-2-[4-(propan-2-yl)phenyl]ethenyl}quinazolin-4(3H)-one (10)
Starting material: 1.00 g (5.1 mmol) 1; 1.13 g (7.6 mmol) 4-(propan-2-yl)benzaldehyde. Product:
1
.19 g yellow powder. Yield: 72 %. Mp: 277-278 °C (DMF)
1
H NMR (300 MHz, DMSO-d ) δ ppm 12.25 (bs, 1H); 8.03 (d, J = 2.4 Hz, 1H); 7.93 (d, J = 16.2
6
1
2
Hz, 1H); 7.82 (dd, J = 8.8 Hz, J = 2.5 Hz, 1H); 7.68 (d, J = 8.8 Hz, 1 H); 7.57 (d, J = 8.1 Hz,
H); 7.34 (d, J = 8.1 Hz, 2H); 6.94 (d, J = 16.1 Hz, 1H); 2.92 (m, 1H); 1.22 (d, J = 6.9 Hz, 6H)
2
+
-
LCMS (ESI) m/z 325.1 (M+H) ; 323.1 (M-H) ; calculated: 324.1 Rt: 4.77 min.
2
2

6
-chloro-2-{(E)-2-[4-(methylsulfonyl)phenyl]ethenyl}quinazolin-4(3H)-one (11)
Starting material: 1.00 g (5.1 mmol) 1; 1.41 g (7.6 mmol) 4-(methylsulfonyl)benzaldehyde.
Product: 1.00 g yellow powder. Yield: 55 %. Mp: 313-315 °C (DMF)
1
H NMR (300 MHz, DMSO-d ) δ ppm 12.48 (bs, 1H); 8.05 (d, J = 2.4 Hz, 1H); 7.99 (m, 3H);
6
1
2
7
.91 (m, 2H); 7.85 (dd, J = 8.8 Hz, J = 2.5 Hz, 1H); 7.72 (d, J = 8.7 Hz, 1H); 7.16 (d, J = 16.2
Hz, 1H); 3.25 (s, 3H)
LCMS (ESI) m/z 361.0 (M+H) ; 359.0 (M-H) ; calculated: 360.0 Rt: 3.42 min.
-fluoro-2-{(E)-2-[4-(propan-2-yl)phenyl]ethenyl}quinazolin-4(3H)-one (12)
Starting material: 1.00 g (5.6 mmol) 2; 1.24 g (8.4 mmol) 4-(propan-2-yl)benzaldehyde. Product:
+
-
6
1
.28 g yellow powder. Yield: 74 %. Mp: 244-245 °C (DMF)
1
H NMR (300 MHz, DMSO-d ) δ ppm 12.40 (bs, 1H); 7.91 (d, J = 16.2 Hz, 1H); 7.71 (m, 3H);
6
7
.58 (d, J = 8.1 Hz, 2H); 7.33 (d, J = 8.1 Hz, 2H); 6.94 (d, J = 15.9 Hz, 1H); 2.92 (m, 1H); 1.22
(d, J = 6.9 Hz, 6H)
+
-
LCMS (ESI) m/z 309.1 (M+H) ; 307.1 (M-H) ; calculated: 308.1 Rt: 4.45 min.
6
-bromo-2-[(E)-2-(4-methoxyphenyl)ethenyl]quinazolin-4(3H)-on (13)
Starting material: 1.50 g (6.3 mmol) 3; 1.27 g (9.4 mmol) 4-methoxybenzaldehyde.
Product: 1.80 g yellow powder. Yield: 80 %. Mp: 324-325 °C (DMF)
1
H NMR (300 MHz, DMSO-d ) δ ppm 12.39 (bs, 1H); 8.16 (s, 1H); 7.95 (s, 1H); 7.90 (m, 1H);
6
7
.61 (m, 3H); 7.03 (d, J = 8.4 Hz, 2H); 6.85 (d, J = 15.9 Hz, 1H); 3.82 (s, 3H)
+
-
LCMS (ESI) m/z 357.2 (M+H) ; 355.3 (M-H) ; calculated: 356.0 Rt: 4.15 min.
General synthesis of 4-chloro-2-[(E)-2-phenylethenyl]quinazoline derivatives (Figure 4, d)
The synthesized phenylethenylquinazolin intermediates (1 mmol) were suspended in
phosphorous oxychloride (2 ml) and stirred at 90 °C overnight. The reaction mixtures were
2
3

cooled down to room temperature and the phosphorous oxychloride was distilled off. The crude
products were dissolved in chloroform, washed with cold water and with 5% sodium hydrogen
4
carbonate. The organic layers were dried with MgSO , filtered, concentrated and dried in a
desiccator. Due to the instability of imidoyl chlorides, these products were used without further
purification or analytical measurements in the next reaction.
4
-chloro-2-[(E)-2-(4-fluorophenyl)ethenyl]quinazoline (14)
Starting material: 1.60 g (6.0 mmol) 4. Product: 1.71 g orange powder. Yield: 100 %.
-chloro-2-{(E)-2-[4-(methylsulfanyl)phenyl]ethenyl}quinazoline (15)
Starting material: 1.10 g (3.7 mmol) 5. Product: 1.16 g orange powder. Yield: 100 %.
-chloro-2-{(E)-2-[4-(propan-2-yl)phenyl]ethenyl}quinazoline (16)
Starting material: 0.90 g (3.1 mmol) 6. Product: 0.84 g orange powder. Yield: 88 %.
-chloro-2-{(E)-2-[4-(methylsulfonyl)phenyl]ethenyl}quinazoline (17)
Starting material: 1.30 g (4.0 mmol) 7. Product: 0.89 g brown powder. Yield: 65 %.
,6-dichloro-2-[(E)-2-(4-fluorophenyl)ethenyl]quinazoline (18)
Starting material: 1.49 g (5.3 mmol) 8. Product: 1.60 g orange powder. Yield: 100 %.
,6-dichloro-2-{(E)-2-[4-(methylsulfanyl)phenyl]ethenyl}quinazoline (19)
Starting material: 1.45 g (4.4 mmol) 9. Product: 1.50 g orange powder. Yield: 98 %.
,6-dichloro-2-{(E)-2-[4-(propan-2-yl)phenyl]ethenyl}quinazoline (20)
Starting material: 1.10 g (3.4 mmol) 10. Product: 1.14 g orange powder. Yield: 98 %.
,6-dichloro-2-{(E)-2-[4-(methylsulfonyl)phenyl]ethenyl}quinazoline (21)
Starting material: 1.00 g (2.8 mmol) 11. Product: 0.95 g brown powder. Yield: 89 %.
-chloro-6-fluoro-2-{(E)-2-[4-(propan-2-yl)phenyl]ethenyl}quinazoline (22)
4
4
4
4
4
4
4
4
2
4

Starting material: 1.30 g (4.2 mmol) 12. Product: 1.00 g orange powder. Yield: 73 %.
-bromo-4-chloro-2-[(E)-2-(4-methoxyphenyl)ethenyl]quinazoline (23)
6
Starting material: 1.00 g (2.8 mmol) 13. Product: 0.99 g orange powder. Yield: 94 %.
General synthesis method of N-alkyl-2-[(E)-2-phenylethenyl]quinazolin-4-amine final products
(Figure 4, e)
The 4-chloro-2-[(E)-2-phenylethenyl]quinazoline derivatives (1 mmol) were solved in dry
dioxane (10 ml). Diisopropyl-ethylamine (1 mmol) and the appropriate primary amines (2 mmol)
were added to the solutions. The reactions were stirred under argon atmosphere at 80 °C for 12
hours. The solvent was evaporated and the compounds were purified by column chromatography
or preparative TLC using chloroform/methanol 9:1 as eluent. The products were prepared as free
bases or salts (hydrochloride, oxalate, fumarate or citrate).
N'-{2-[(E)-2-(4-fluorophenyl)ethenyl]quinazolin-4-yl}-N,N-dimethylethane-1,2-diamine (I)
Starting material: 0.33 g (1.2 mmol) 14; 0.21 g (2.4 mmol) N,N-dimethyl-1,2-ethanediamine.
Product: 0.185 g light yellow powder (free base). Yield: 46 %. Mp: 152-154 °C
1
H NMR (300 MHz, DMSO-d ) δ ppm 8.17 (bs, 1H); 8.16 (s, 1H); 7.90 (d, J = 15.9 Hz, 1H);
6
7
2
.73 (m, 3H); 7.68 (m, 1H); 7.45 (m, 1H); 7.27 (m, 2H); 7.10 (d, J = 15.8 Hz, 1H); 3.75 (m, 2H);
.60 (m, 2H); 2.28 (s, 6H)
+
-
LCMS (ESI) m/z 337.3 (M+H) ; 335.0 (M-H) ; calculated: 336.2 Rt: 0.45; 2.03 min.
N,N-dimethyl-N'-(2-{(E)-2-[4-(methylsulfanyl)phenyl]ethenyl}quinazolin-4-yl)ethane-1,2-
diamine (II)
2
5

Starting material: 0.20 g (0.6 mmol) 15; 0.11 g (1.2 mmol) N,N-dimethyl-1,2-ethanediamine.
Product: 0.083 g yellow powder (1.0 eq. fumaric acid salt). Yield: 29 %. Mp: 196-200 °C
1
6
H NMR (300 MHz, DMSO-d ) δ ppm 8,28 (bs, 1H); 8.17 (d, J = 7.8 Hz, 1H); 7.88 (d, J = 15.3
Hz, 1H); 7.66 (m, 4H); 7.44 (t, J = 7.0 Hz, 1H); 7.29 (d, J = 7.1 Hz, 2H), 7.10 (d, J = 16.1 Hz,
H); 6.57 (s, 2H); 3.85 (m, 2H); 3.90 (m, 2H); 2.48 (s, 9H)
1
+
-
LCMS (ESI) m/z 365.2 (M+H) ; 363.3 (M-H) ; calculated: 364.2 Rt: 0.45; 2.25 min.
N'-{2-[(E)-2-(4-(propan-2-yl)phenyl)ethenyl]quinazolin-4-yl}-N,N-dimethylethane-1,2-
diamine (III)
Starting material: 0.20 g (0.6 mmol) 16; 0.11 g (1.2 mmol) N,N-dimethyl-1,2-ethanediamine.
Product: 0.085 g light yellow powder (1.7 eq. fumaric acid salt). Yield: 25 %.
Mp: 164-168 °C
1
H NMR (300 MHz, DMSO-d ) δ ppm 8.33 (bs, 1H); 8.17 (d, J = 8.1 Hz, 1H); 7.79 (d, J = 15.9
6
Hz, 1H); 7.72 (m, 2H); 7.61 (d, J = 7.9 Hz, 2H); 7.44 (t, J = 7.4 Hz, 1H); 7.29 (d, J = 7.8 Hz,
2
6
H); 7.09 (d, J = 15.8 Hz, 1H); 6.58 (s, 3.5H); 3.87 (m, 2H); 2.96 (m, 2H); 2.91 (m, 1H); 2.55 (s,
H); 1.22 (d, J = 6.9 Hz, 6H).
+
-
LCMS (ESI) m/z 361.4 (M+H) ; 359.4 (M-H) ; calculated: 360.2 Rt: 0.45; 2.18; 2.45 min.
N,N-dimethyl-N'-(2-{(E)-2-[4-(methylsulfonyl)phenyl]ethenyl}quinazolin-4-yl)ethane-1,2-
diamine (IV)
Starting material: 0.30 g (0.9 mmol) 17; 0.16 g (1.8 mmol) N,N-dimethyl-1,2-ethanediamine.
Product: 0.108 g light yellow powder (free base). Yield: 30 %. Mp: 186-188 °C
1
H NMR (300 MHz, DMSO-d
.49 (t, J = 7.69 Hz, 1H); 7.33 (d, J = 16.0 Hz, 1H); 3.77 (q, J = 6.0 Hz, 2H); 3.25 (s, 3H); 2.60 (t,
J = 6.8 Hz, 2H); 2.27 (s, 6H)
6
) δ ppm 8.22 (bs, 1H); 8.19 (s, 1H); 7.97 (m, 5H); 7.73 (m, 2H);
7
2
6

+
-
LCMS (ESI) m/z 397.1 (M+H) ; 395.1 (M-H) ; calculated: 396.2 Rt: 0.39; 1.78 min.
N'-{6-chloro-2-[(E)-2-(4-fluorophenyl)ethenyl]quinazolin-4-yl}-N,N-dimethylpropane-1,3-
diamine (V)
Starting material: 3.30 g (10.3 mmol) 18; 2.10 g (20.6 mmol) N,N-dimethyl-1,3-propanediamine.
Product: 1.7 g light yellow powder (1.0 eq. fumaric acid salt). Yield: 33 %.
Mp: 195-199 °C
1
H NMR (300 MHz, DMSO-d
.68 (d, J = 8.9 Hz, 1H); 7.25 (t, J = 9.0 Hz, 2H); 7.09 (d, J = 15.9 Hz, 1H); 6.55 (s, 2H); 3.67 (q,
J = 6.3 Hz, 2H); 2.71 (t, J = 6.9 Hz, 2H); 2.42 (s, 6H); 1.95 (m, 2H)
6
) δ ppm 8.37 (bs, 2H); 7.91 (d, J = 15.8 Hz, 1H); 7.75 (m, 3H);
7
+
-
LCMS (ESI) m/z 381.3 (M+H) ; 379.0 (M-H) ; calculated: 380.2 Rt: 0.46; 1.96; 2.29 min.
N'-(6-chloro-2-{(E)-2-[4-(methylsulfanyl)phenyl]ethenyl}quinazolin-4-yl)-N,N-
dimethylethane-1,2-diamine (VI)
Starting material: 0.20 g (0.6 mmol) 19; 0.11 g (1.2 mmol) N,N-dimethyl-1,2-ethanediamine.
Product: 0.085 g yellow powder (1.3 eq. fumaric acid salt). Yield: 26 %. Mp: 201-204 °C
1
H NMR (300 MHz, DMSO-d ) δ ppm 8.40 (bs, 1H); 8.34 (s, 1H); 7.90 (d, J = 15.9 Hz, 1H);
6
7
2
.69 (m, 4H); 7.29 (d, J = 7.8 Hz, 2H); 7.09 (d, J = 16.1 Hz, 1H); 6.57 (s, 2,6H); 3.83 (m, 2H);
.89 (m, 2H); 2.50 (s, 9H)
+
-
LCMS (ESI) m/z 399.2 (M+H) ; 397.2 (M-H) ; calculated: 398.1 Rt: 0.46; 2.01; 2.41 min.
N'-(6-chloro-2-{(E)-2-[4-(propan-2-yl)phenyl]ethenyl}quinazolin-4-yl)-N,N-dimethylethane-
1
,2-diamine (VII)
Starting material: 0.25 g (0.7 mmol) 20; 0.12 g (1.4 mmol) N,N-dimethyl-1,2-ethanediamine.
Product: 0.089 g yellow powder (1.4 eq. fumaric acid salt). Yield: 23 %. Mp: 222-227 °C
2
7

1
H NMR (300 MHz, DMSO-d ) δ ppm 8.37 (t, J = 5.30 Hz, 1H); 8.33 (d, J = 2.3 Hz, 1H); 7.90 (d,
6
1
2
J = 15.9 Hz, 1H); 7.74 (dd, J = 8.9 Hz, J = 2.3 Hz, 1H); 7.68 (d, J = 8.9 Hz, 1H); 7.61 (d, J = 8.2
Hz, 2H); 7.29 (d, J = 8.2 Hz, 2H); 7.08 (d, J = 15.9 Hz, 1H); 6.57 (s, 2.8H); 3.83 (q, J = 5.7 Hz,
2
H), 2.92 (m, 1H); 2.88 (m, 2H); 2.47 (s, 6H); 1.22 (d, J = 6.9 Hz, 6H)
+
-
LCMS (ESI) m/z 395.3 (M+H) ; 393.4 (M-H) ; calculated: 394.2 Rt: 2.73 min.
N'-(6-chloro-2-{(E)-2-[4-(methylsulfonyl)phenyl]ethenyl}quinazolin-4-yl)-N,N-
dimethylpropane-1,3-diamine (VIII)
Starting material: 0.20 g (0.5 mmol) 21; 0.10 g (1.0 mmol) N,N-dimethyl-1,3-propanediamine.
Product: 0.138 g yellow powder (free base). Yield: 62 %. Mp: 168-171 °C
1
H NMR (300 MHz, DMSO-d ) δ ppm 8.37 (bs, 2H); 7.96 (m, 5H); 7.77 (d J = 9.0Hz, 1H); 7.69
6
(d, J = 9.0 Hz, 1H); 7.31 (d, J = 16.1 Hz, 1H); 3.66 (q, J = 5.4 Hz, 2H); 3.24 (s, 3H); 2.43 (m,
2
H); 2.23 (s, 6H); 1.86 (m, 2H)
+
-
LCMS (ESI) m/z 445.2 (M+H) ; 443.2 (M-H) ; calculated: 444.1 Rt: 0.44; 1.89; 2.07 min.
N'-(6-chloro-2-{(E)-2-[4-(methylsulfonyl)phenyl]ethenyl}quinazolin-4-yl)-N,N-
dimethylethane-1,2-diamine (IX)
Starting material: 0.20 g (0.5 mmol) 21; 0.09 g (1.0 mmol) N,N-dimethyl-1,2-ethanediamine.
Product: 0.122 g yellow powder (free base). Yield: 57 %. Mp: 228-231 °C
1
H NMR (300 MHz, DMSO-d ) δ ppm 8.39 (d, J = 1.9 Hz, 1H); 8.29 (t, J = 5.0 Hz, 1H); 7.95 (m,
6
1
2
5
3
H); 7.77 (dd, J = 8.9 Hz, J = 2.2 Hz, 1H); 7.70 (d, J = 8.9 Hz, 1H); 7.31 (d, J = 15.9 Hz, 1H);
.75 (q, J = 5.5 Hz, 2H); 3.24 (s, 3H); 2.59 (t, J = 6.3 Hz, 2H); 2.26 (s, 6H)
+
-
LCMS (ESI) m/z 431.1 (M+H) ; 429.2 (M-H) ; calculated: 430.1 Rt: 0.44; 1.91; 2.06 min.
N'-(6-chloro-2-{(E)-2-[4-(methylsulfonyl)phenyl]ethenyl}quinazolin-4-yl)-N,N-
diethylethane-1,2-diamine (X)
2
8