Inhibitors to Overcome Secondary Mutations in the Stem Cell Factor Receptor KIT

Article abstract of DOI:10.1021/acs.jmedchem.7b00841

In modern cancer therapy, the use of small organic molecules against receptor tyrosine kinases (RTKs) has been shown to be a valuable strategy. The association of cancer cells with dysregulated signaling pathways linked to RTKs represents a key element in

Full text of DOI:10.1021/acs.jmedchem.7b00841

Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES
Article
Inhibitors to overcome secondary mutations
in the stem cell factor receptor KIT
Helena Kaitsiotou, Marina Keul, Julia Hardick, Thomas Mühlenberg, Julia Ketzer, Christiane Ehrt,
Jasmin Krüll, Federico Medda, Oliver Koch, Fabrizio Giordanetto, Sebastian Bauer, and Daniel Rauh
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b00841 • Publication Date (Web): 09 Oct 2017
Downloaded from http://pubs.acs.org on October 9, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Inhibitors to overcome secondary mutations in the
stem cell factor receptor KIT
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
1
1
2, 3
2, 3
Helena Kaitsiotou , Marina Keul , Julia Hardick , Thomas Mühlenberg , Julia Ketzer ,
1
1§
4‡
1
4†
Christiane Ehrt , Jasmin Krüll , Federico Medda , Oliver Koch , Fabrizio Giordanetto ,
2, 3
1,
Sebastian Bauer *, Daniel Rauh *
KEYWORDS: KIT; mast/stem cell growth factor receptor; GIST; ponatinib
1
Faculty of Chemistry and Chemical Biology, TU Dortmund University, OttoꢀHahnꢀStraße 4a,
Dꢀ44227 Dortmund, Germany
2
Department of Medical Oncology, Sarcoma Centre West German Cancer Centre University
3
DuisburgꢀEssen, Medical School, Hufelandstraße 55, Dꢀ45122 Essen, Germany and German
Cancer Consortium (DKTK), Partner site University Hospital Essen, Essen, Germany
4
Taros Chemicals GmbH & Co. KG, EmilꢀFiggeꢀStraße 76a, Dꢀ44227 Dortmund, Germany
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT
In modern cancer therapy, the use of small organic molecules against receptor tyrosine kinases
RTKs) has been shown to be a valuable strategy. The association of cancer cells with
(
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
dysregulated signaling pathways linked to RTKs represents a key element in targeted cancer
therapies. The tyrosine kinase mast/stem cell growth factor receptor KIT is an example of a
clinically relevant RTK. KIT is targeted for cancer therapy in gastrointestinal stromal tumors
(GISTs) and chronic myelogenous leukemia (CML). However, acquired resistance mutations
within the catalytic domain decrease the efficacy of this strategy and are the most common cause
of failed therapy. Here, we present the structureꢀbased design and synthesis of novel type II
kinase inhibitors to overcome these mutations in KIT. Biochemical and cellular studies revealed
promising molecules for the inhibition of mutated KIT.
ACS Paragon Plus Environment
2

Page 3 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
KIT is a type III receptor tyrosine kinase that is an important signaling protein for the
development of melanocytes, erythrocytes, germ cells, mast cells, and interstitial cells of cajal
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1ꢀ3
(
gastrointestinal pacemaker cells). Upon binding of its ligand, stem cell factor (SCF), KIT
activates downstream signaling pathways that promote cell survival and cell proliferation and
4
inhibit apoptosis. Constitutive activation of KIT as an oncogenic driver has first been described
in feline sarcomas (“kitten”ꢀKIT), but was later also found in a variety of human cancers
5
, 6
including melanomas , seminomas, acute myeloid leukemias, systemic mastocytosis, sinonasal
7, 8
lymphoma, and gastrointestinal stromal tumors (GIST).
Particularly, GISTs have since then
become a paradigm of successful targeted treatment for cancer.
9
ꢀ11
GIST represents the most common mesenchymal tumor of the gastrointestinal tract
and
approximately 85 % of GISTs harbor oncogenic gainꢀofꢀfunction mutations of KIT or the
1
2, 13
plateletꢀderived growth factor receptor (PDGFR).
Activating mutations of KIT represent an
early oncogenic event, but KIT also remains the central oncogenic driver in patients with highly
7, 14, 15
advanced GIST disease.
Primary, activating mutations of KIT in GIST most commonly
occur within the juxtamembrane (exon 11) and the extracellular regions (exon 9), and only very
1
6ꢀ19
rarely in other exons (8, 13, 17).
Three inhibitors of KIT have until now been approved for
the treatment of GIST. Of note, these three inhibitors had been primarily developed as inhibitors
20
21
of kinases other than KIT such as BCRꢀABL (1 (imatinib) ) and VEGFR (2 (sunitinib) , 3
22
23ꢀ26
(regorafenib) ) (Figure 1).
The possibility of inhibiting dysregulated tyrosine kinases in kinase mutationꢀdriven cancer
has led to the development of tyrosine kinase inhibitor (TKI) therapies. In many cases, TKIs
have been shown to be more beneficial than traditional cancer treatments in terms of side effects
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 60
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
17, 18
and the overall clinical outcomes for patients with GIST.
1, a 2ꢀphenylaminopyrimidine
derivative, is a rapidly absorbed oral kinase inhibitor, that effectively inhibits exon 9 and exon 11
KIT mutants, and whose response rates, as well as duration of clinical benefit, correlate with
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
27, 28
genotype.
Long lasting disease control (median progressionꢀfree survival [PFS] exon 11: 2.3
2
9
years; exon 9: 1.6 years) paired with a favorable side effect profile have made 1 the standard
firstꢀline treatment. Nonetheless, the majority of patients eventually progress and secondary
mutations within the ATPꢀbinding pocket (exon 13: V654A; exon 14: T670I) and the activation
3
0, 31
loop (affecting codons 816ꢀ829) represent the major mechanisms of resistance.
Both 2 and 3
show better activity against KIT mutants with secondary resistance mutations, and approval was
3
2
granted based on an improved median PFS of 4ꢀ5 months over placebo in secondꢀ and
3
3
thirdꢀ line treatment. However, response rates for both drugs are low (<8 %). Most patients
progressing on 1 exhibit multiple, polyclonal resistance mutations (genomic heterogeneity) and
neither drug inhibits the full spectrum of resistant mutants which may explain the moderate
3
4
clinical activity. Peculiarly, inhibitors had not yet been developed for the needs of patients with
GIST, but had rather been byproducts of drug development for other diseases. 2 and 3 were
optimized as potent inhibitors of VEGFR, but very little evidence suggests an added therapeutic
value of inhibiting VEGFR in KITꢀmutant GIST. A randomized phase III trial that compared 1
with 1 plus bevacizumab, a VEGF antibody, was prematurely terminated for futility and lack of
3
5
evidence for clinical activity. In addition, both drugs inhibit a vastly broader spectrum of
kinases compared to 1, particularly angiogenic kinases such as VEGFR 1ꢀ3 and TIE2, which
adds substantial skin and gastrointestinal toxicity, as well as hypertensionꢀrelated side effects.
Compared to the substantial benefits obtained with 1 as firstꢀline treatment, the clinical benefits
of 2 and 3 are moderate, and the side effect profiles of both inhibitors are much less favorable
ACS Paragon Plus Environment
4

Page 5 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
35ꢀ38
than those of 1.
Currently, (S)ꢀ1ꢀ(4ꢀfluorophenyl)ꢀ1ꢀ(2ꢀ(4ꢀ(6ꢀ(1ꢀmethylꢀ1Hꢀpyrazolꢀ4ꢀ
3
9
yl)pyrrolo[2,1ꢀf][1,2,4]triazinꢀ4ꢀyl)piperazinꢀ1ꢀyl)pyrimidinꢀ5ꢀyl)ethanꢀ1ꢀamine
((BLUꢀ285)
(
NCT02508532)), a potent inhibitor of PDGFRAꢀD842 and KIT exon 17 mutations and 1ꢀN'ꢀ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
[
2,5ꢀdifluoroꢀ4ꢀ[2ꢀ(1ꢀmethylpyrazolꢀ4ꢀ
4
0
yl)pyridineꢀ4ꢀyl]oxyphenyl]ꢀ1ꢀNꢀphenylcyclopropaneꢀ1,1ꢀdicarboxamide ((DCCꢀ2618) (NCT
4
1
0
2571036)), a type III inhibitor of KIT and PDGFRA, and PLX9486/PLX3397
(NCT02401815) represent the first wave of novel compounds, that were designed especially for
KIT to exhibit strong activity against exon 13/14 and 17 which is not only important for GIST,
42
but also for KITꢀdriven leukemias and melanomas. 4 (ponatinib) is a nextꢀgeneration, highly
potent ATPꢀcompetitive inhibitor of BCRꢀABL which was rationally designed to inhibit the
4
3, 44
notoriously resistant T315I mutation.
Approval was granted for the treatment of T315Iꢀ
positive BCRꢀABL leukemias (chronic myelogenous and acute lymphoblastic) based on
significant antileukemic activity across categories of disease stage and mutation status within the
4
5
PACE trial. Notably, 4 also inhibits a broad spectrum of imatinibꢀresistant KIT mutants,
including those affecting codon Asp816 of exon 17, which confers resistance to all approved
3
4
drugs in GIST, but is less potent to secondary mutant V654A.
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. (A) Structural alignment of kinase inhibitors, approved by the FDA for the treatment
of GIST, bound to wild type KIT (pdbꢀcode: 4u0i) (1 (green, pdbꢀcode: 4bkj), 2 (yellow, pdbꢀ
code: 3g0e), 3 (blue, modeled) and 4 (red, pdbꢀcode: 4u0i)). Unique substitution patterns and
ACS Paragon Plus Environment
6

Page 7 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
hinge binding elements of the inhibitors highlighted (right). (B) Xꢀray crystal structure of 4 in
complex with wild type KIT (pdbꢀcode: 4u0i) with modeled secondary mutations (sticks and
surfaces). Hydrogen bonds are depicted by blue dotted lines.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In patients with GIST, 4 has already shown encouraging activity with clinical benefit rates of
up to 55 % in a small cohort of heavily pretreated patients with primary exon 11 mutation. A
current trial (POETIG, NCT03171389) investigates the clinical benefit of 4 in secondꢀline and
lastꢀline patients. 4 is also a multikinase inhibitor that targets protein kinases of the VEGFR,
PDGFR, and SRC families. Most common side effects include skin rash, fatigue, myalgia, dry
skin, headache, and mucositis, which is usually perceived less intense than with 2 or 3. A more
severe side effect observed in 11.8 % of patients with BCRꢀABLꢀdriven leukemia are arterial
3
5,
thrombotic events which appear to be doseꢀdependent and less frequent at lower dose intensity.
4
6
The role of inhibition of angiogenesisꢀrelated kinases (VEGFR1–3 and TIE2) by 4 as related
3
4, 47
to the observed thrombotic events has yet to be explored.
Against the background of the
broad inhibitory spectrum of 4 and its unique ability to inhibit D816H mutant KIT, we compared
the scaffolds of currently available GIST therapeutics and 4 to potentially generate a novel
compound with activity on KIT exon 17, 13 and 14 mutants, including secondary mutant
3
4
V654A, which is the most likely escape mutation of 4. Basically, the back pocketꢀbinding
element of 4 was preserved because this part is analogous to 1, except for an inverted amide
bond, which is already optimized to generate a more favorable interaction pattern and the
trifluoromethyl substituent. For the latter, we decided to gradually revert to the substitution
pattern of 1 to determine if the selectivity profile of 1 is retained. The triple bond connecting the
back pocketꢀbinding element with the hingeꢀbinding element was also retained because it
bypasses the challenging gatekeeper mutation T670I. The gatekeeper position in kinases controls
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 60
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
access to the ATPꢀbinding site, playing an important role in ATPꢀcompetitive inhibitor
4
8, 49
selectivity.
In the case of KIT, the primary mutation V559D in the juxtamembrane region
leads to constitutive activation of the kinase, whereas the bulkier and more lipophilic residue
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
18, 50
T670I modulates the conformation of the catalytic domain for less favorable binding of 1.
The crystal structure of wild type KIT in complex with 4 (pdbꢀcode: 4u0i) guided our structureꢀ
3
4
based design regarding the hingeꢀbinding element (Figure 1C). In the present study, the main
objective was to evaluate the potential for additional interactions with the hinge region, as these
should compensate for the loss of Van der Waals interactions with most of the known inhibitors
34
due to the second gatekeeperꢀmutation V654A. To support our structureꢀactivity relationship
(SAR) studies in the development of optimized compounds, we performed molecular dynamics
(MD) simulations to gain insights into the conformational changes of the kinase domain in the
presence of the secondary mutation. Based on the information obtained from the MD simulation
and its combination with rational structureꢀbased design on docking studies, we designed,
synthesized, and tested a focused library of KIT mutantꢀselective inhibitors to overcome
acquired drug resistance in GISTs.
RESULTS
Structural analysis of wild type KIT, KIT V559D/V654A and KIT V559D/T670I mutant
variants
We used molecular modeling and MD simulations based on the sequence of KIT and its crystal
structure in complex with 4 (pdbꢀcode: 4u0i) to generate models of two KIT variants.
Subsequent docking studies were performed to gain insight into the binding mode of 4 with
ACS Paragon Plus Environment
8

Page 9 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
respect to the V654A and T670I variants. Docking indicated that 4 could bind to the mutants in
an analogous manner to that observed for wild type KIT. A detailed visual inspection showed
that the highest ranked poses for all docking studies show only negligible deviations from the
native binding mode of 4 in the original crystal structure for wild type KIT, and the T670I and
V654A variants. We selected the poses with the smallest RMSD to the binding mode observed
for the Xꢀray structure (pdbꢀcode: 4u0i) (For details see SI) to perform 300 ns MD simulations
with all three KIT variants in the absence and presence of 4 to get further insight into possible
binding mode differences. The corresponding RMSD profiles for MD simulations are given in
Figure S1.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. (A) Comparison of the ligand binding surfaces using representative structures of the
MD simulation trajectory derived from a residueꢀbased RMSDꢀclustering. The alignment and
representations of 4 bound to wild type KIT (gray), apo KIT T670I (brown) and apo KIT V654A
5
1
(
blue), structures were generated with SiteHopper , respectively. (B) Results of the binding site
comparison of mutant KIT T670I bound to 4 and apo mutant KIT T670I to wild type KIT with
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 60
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
respect to shape and physicochemical properties (hydrogen bond donor, hydrogen bond acceptor,
and hydrophobic features). The pockets of the representative structures of the cluster with the
highest occupancy for the simulations of the apo enzymes (apo) and the enzymes bound to 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
(ponatinib) were used. They were compared to the wildtype KIT crystal structure bound to
ponatinib (pdbꢀcode: 4u0i). (C) Results of the binding site comparison of ligandꢀbound mutant
KIT V654A and apo mutant KIT V654A to wild type KIT with respect to shape and
physicochemical properties. The comparison was performed as outlined in (B).
The ligandꢀbound KIT variants exhibited nearly identical behavior during a 300 ns simulation;
the most prominent ligandꢀprotein hydrogen bond interactions were stable in all three trajectories
(Table S6). Additionally, molecular mechanics PoissonꢀBoltzmann (Generalized Born) solventꢀ
accessible surface area (MMꢀPB(GB)SA) calculations (Table S7) show similar results for KIT
wild type and the variants under investigation. The differences observed in the binding free
energies do not reflect the outstanding differences of the enzymes in the binding of 4 (Table 2).
5
2
Given the uncertainties of those methods , we investigated additional factors which might have
an impact on the binding affinity of 4. In contrast to the trajectories in the presence of 4, the MD
simulations of wild type KIT apo structures and the variants elucidated major differences.
Particularly variant V654A exhibited higher flexibility with respect to the residues involved in
the binding of 4 (Figure S2). An mdpocket analysis of the trajectories in the absence of ligands
did not reveal major differences in the pocket volume of all three enzymes (Figure S5).
Nevertheless, a clustering of the conformations from MD trajectories with respect to the binding
5
1
site residues (Table S5) of 4 and subsequent SiteHopper ꢀbased binding site comparison of
representatives of the clusters with highest occupancy showed that the wild type and V654A
binding sites are different with respect to shape and physicochemical properties, whereas the
ACS Paragon Plus Environment
10

Page 11 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
differences between wild type and T670I were less pronounced (Figure 2, Table S8 for more
details).
Interestingly, these differences were not observed for the MD simulations in the presence of
bound ligand. Both variants show a higher similarity to the wild type when bound to 4. In
contrast, the decreasing similarity in the unbound state hints toward major deviations of the
variants’ binding site conformation in the absence of ligands. Therefore, we speculate that the
difference in binding and activity of 4 with respect to the investigated variant might result from
different binding site conformations of the kinase in the apo state (especially for V654A)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(Figure 2), as well as shifts in the equilibrium between the active and inactive conformation of
5
3, 54
the kinase that are not observable for the proteinꢀligand complexes.
As no clear SAR for the
inhibitors and their activity on V654A was available, we analyzed whether this variant might
show different activation properties as compared to the wild type and variant T670I. Intriguingly,
in the MD simulation of the V654A apo structure, we observed high flexibility of the alpha helix
5
3
C, a structural feature of KIT important for activation of the kinase, as well as in the binding
pocket itself (Figure S2). As this helix is a crucial determinant of KIT activation and
5
4
inactivation, we analyzed the conformational variation of the helix along the simulated
trajectory by monitoring the angle of the Nꢀterminal axis of the alpha helix C (residues Leu631
to Leu637) and the unit cell axis x (Figure S3). The data suggest that the V654A mutation
perturbs the dynamics of the alpha helix C and renders KIT in a conformation less suited for
binding of Type II inhibitors when compared to wild type KIT. The uniqueness of this behavior
as compared to the wild type enzyme and variant T670I could be shown by Principle Component
Analysis (PCA, Figure S6). The modes, i.e. the most prominent directions of structural changes,
indicate a bending motion of the alpha helix C solely for variant V654A.
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 60
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
Also, we hypothesize that, similar to the activating mechanism of most gatekeeper mutations
in protein kinases (T670I in KIT) as shown by Azam et al for cꢀABL und cꢀSRC, the V654A
mutation in KIT induces a stabilization of the regulatory Rꢀspine and therefore the active
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
48, 55
conformation of the kinase domain.
Overall, this hampers the rational design of molecules
targeting KIT variant V654A. We, therefore, focused our optimization strategy on the
development of derivatives of 4 to more effectively target KIT V654A. Such an approach proved
4
9, 56, 57
successful for example in second and third generation BCRꢀABL inhibitors.
Structure-based design of type II inhibitors against KIT mutants V654A and T670I
Based on our MD simulations and the crystal structure of 4 in complex with wild type KIT
(pdbꢀcode: 4u0i) as well as the crystal structures of wild type KIT in the active (pdbꢀcode:
58
58
1
t46) and autoꢀinhibited (pdbꢀcode: 1t45) state, we decided to maintain the overall potent
5
9
1
4
1
scaffold of 4 and change the substitution patterns in R ꢀR (Figure 3). R was optimized with
respect to V654A mutation, keeping the alkyne linker to maximize the effect on the T670I
2
mutation. R was gradually modified to the substitution pattern of 1 to afford the improved
kinase selectivity profile of 1. Modeling of mutant V654A revealed an altered surface in this part
of the ATP binding pocket due to the size of the mutated amino acid. This slight change might
offer the possibility of targeting V654A selectively within the KIT mutant family. Thus, we
3
4
designed larger hydrophobic substituents at R while reducing the size of the R substituents in
4
line with results from the MD simulations, that suggest that the methyl group in the R position
might not be as well tolerated in the mutant forms (T670I and V654A) when compared to wild
type KIT (Figure 2A). Afterwards, we performed molecular docking to examine the feasibility of
the anticipated interactions. An ensemble docking approach using the representative structures of
ACS Paragon Plus Environment
12

Page 13 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a clustering with respect to the binding site of 4 was applied for all three enzymes. 4 was
redocked to evaluate and control the docking parameters. Given that our compounds are quite
similar to 4, we predicted high similarities in the binding modes in KIT, suggesting inhibitory
activity towards wild type KIT and the secondary variants V654A and, T670I, and D816H.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Coreꢀscaffold of synthesized compounds modeled into the crystal structure of wild
type KIT (pdbꢀcode: 4u0i). The general scaffold and the substitution patterns (red) of the
designed compounds are shown (right).
Given the similarity to 4, our modeling studies suggested a comparable binding mode for
compounds 5 and 7ꢀ10. The compounds occupy three subpockets within the catalytic domain of
KIT: the ATP pocket, gatekeeper pocket, and back pocket (Figure 3). The hingeꢀbinding
moieties form hydrogen bonds to the backbone of the residues within the hinge region. The
allosteric or back pocket, which is only accessible when the kinase is in the inactive
conformation, is accessed via the substituted phenylbenzamide element. We suggest that the
amide’s NH and carbonyl moieties of the compounds establish hydrogen bonds with the side
chain carboxylate group of Glu640 in the alpha helix C and the backbone nitrogen of Asp810 in
the DFG motif, respectively. The methylated nitrogen of the methylpiperazine moiety is able to
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 60
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
interact with carbonyl groups in the backbone (Ile789/His790) of the activation loop. We focused
our optimization efforts on the hingeꢀbinding element, gradually altered the substitution pattern
of the back pocketꢀbinding element, and tried to target the slight changes of the binding site
surface observed in KIT mutant V559D/V654A. Optimization of the hingeꢀbinding element was
mainly focused on compensating for the loss of Van der Waals interactions caused by the V654A
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
34
mutation. We presumed that a sixꢀmembered ring system incorporating a substitution pattern
capable of interacting with the hinge region would be favorable compared to the imidazoloꢀ
pyridazine moiety of 4. This imidazoloꢀpyridazine moiety forms one single hydrogen bond via
N7 with Cys673 in the hinge region. As MD simulation and docking studies suggested that the
V654A mutation disrupts not only the network of hydrophobic interactions related to the
compounds but also the conformation of the entire catalytic domain, we designed a hingeꢀ
binding element consisting of a single ring system, which was then progressively extended. The
substitution pattern of this single ring system was developed to maximize possible interactions
with the hinge region. Scaffolds based on pyridine and pyrimidine seemed to be a good choice to
achieve this objective. Starting with the hingeꢀbinding element with the least predicted number
of interactions with the hinge, we gradually extended the number of possible interactions in
different directions of the ATPꢀbinding site. This process allowed us to iteratively test the
structureꢀbased hypotheses via matched molecular pair analysis and establish an SAR in order to
develop optimized small molecules that are selective and potent against the majority of KIT
mutants.
Synthesis of a focused structure-based small molecule library
ACS Paragon Plus Environment
14

Page 15 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
We synthesized a focused library of type II inhibitors based on 4 and paid particular attention
to the derivatization of the pyridine/pyrimidineꢀbased hingeꢀbinding element with various
substituents in different positions to evaluate interactions across the ATPꢀbinding pocket. We
expected a distinctive SAR regarding the potency towards the KIT mutants and explored
substitution patterns that are able to engage in favorable interactions. We maintained the alkyne
linker of 4 connecting the hingeꢀbinding element with the back pocket moiety, due to its
favorable overall orientation and function (interaction with gatekeeper residue Ile670).
Exploration of the conformational space in the back pocket required the design of a different
compound series. Herein the benzoic acid moiety forming the amide bond was maintained,
whereas the trifluoromethyl substituent of the ((Nꢀmethylpiperazinꢀ1ꢀyl)methyl)ꢀ3ꢀ
(trifluoromethyl)aniline moiety was systematically changed to fluorine and later to hydrogen. We
reasoned that this approach would have provided insight in the relevance of this substituent in
terms of binding affinity and selectivity. Importantly, this new substitution pattern represents one
of the main differences from 1.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Considering the V654A mutation, we also designed inhibitors targeting the altered binding site
in the mutant kinase domain to study the newly created chemical space and examine the impact
of this structural motif on selectivity and inhibitory activity. Molecular modeling indicated that
derivatization of the benzoic acid moiety’s 2ꢀposition would be ideal to deploy such strategy.
Scheme 1. Synthesis of derivatives (5a-d, 7a-l, 8a-d, 9a-b and 10) based on the structure of 4.
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Reagents and conditions: (i) NꢀBromosuccinimide, AIBN, DCE, 80 °C, 12 h (60ꢀ66 %); (ii)
1
NꢀMethylpiperazine, K CO , DCM, rt, 3 h (80ꢀ91 %); (iii) 1NꢀMethylpiperazine, NaBH(OAc) ,
THF, rt, 16 h (60ꢀ70 %); (iv) Pd/C, H , EtOH, 12 h (98 %); (v) HOBt, EDC, DIPEA, DMF,
DMAP, carboxylic acidꢀderivative rt, 16 h (49ꢀ88 %); (vi) Pd(PPh ) Cl , CuI, (iꢀPr) NH, R ꢀBr,
MeCN, 80 °C, 16 h; (vii) Pd(PPh ) Cl , CuI, (iꢀPr) NH, TBAF, MeCN, 80 °C, 16 h (1ꢀ22 %).
2
3
3
2
1
3 2
2
2
3
2
2
2
Compounds 5a-d, 6a, 7a-l, 8a-d, 9a-b, and 10 (Table 1) were synthesized using the general
route depicted in Scheme 1. Here, benzaldehyde derivatives 2dꢀf were transformed to the
corresponding benzylpiperazines 3aꢀc via reductive amination. Alternatively, toluene derivatives
1
aꢀc were first selectively brominated and then subjected to an S 2 substitution with 1Nꢀ
N
methylpiprerazine. Anilines 4aꢀc were obtained by reduction of the arylnitro intermediates 3aꢀc
via palladium (Pd/C)ꢀcatalyzed hydrogenation. The iodobenzamide precursors 5a-f were
prepared by amide coupling of anilines 4aꢀc with the corresponding iodobenzoic acid. The
reaction conditions were altered to comply with the characteristics of the starting material, but
the overall outcome of the modifications remained the same. The aromatic silylꢀprotected
alkynes were either sourced directly from commercial sources or synthesized from the
corresponding bromide precursors using Sonogashira reaction conditions. Purification of the
ACS Paragon Plus Environment
16

Page 17 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
protected alkynes was challenging because of several impurities and the relatively unstable
protecting group on the one hand, and because of the low boiling point of the silylꢀprotected
alkynes on the other hand. Complete evaporation of the solvent was not possible. Thus, the
protected alkynes were used without further purification. The second Sonogashira coupling
between the silylꢀprotected alkynes and precursors 5a-f were also performed under Pdꢀcatalyzed
Sonogashira reaction conditions Here, in order to minimize the occurrence of dimer and polymer
formation, the precursors were heated with the reactants to 80 °C, then cooled to 40 °C, followed
by addition of the silylꢀprotected alkynes. As the reaction mixture was heated to 80 °C again, the
silyl protection group was removed in situ by subsequent addition of TBAF. Compounds 7a-l,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
8a-d, 9a-b, and 10 were obtained using this procedure (Table 1).
Activity-based in vitro characterization of compounds
We characterized our focused compound library with respect to the inhibition of various
clinically relevant KIT mutants, including T670I, V654A, and D816H, as well as wild type KIT
Table 2), using an activityꢀbased biochemical assay. KITꢀmediated phosphorylation of an
(
artificial substrate was quantified through homogeneous timeꢀresolved FRET measurements.
Starting with the SAR studies for the improvement of the hingeꢀbinding element supported by
our modeling studies, we decided on sixꢀmembered heterocycles and also tested precursors 5a-d
and the alkyne substituted precursor 6a. Precursors 5a-d had IC values of 6.8 nM, 555 nM,
5
0
1039 nM, and 184 nM, respectively, for wild type KIT, but did not inhibit the KIT mutants
tested. Notably, the original precursor of 4 (5a, 6.8 nM) was the most potent inhibitor of wild
type KIT, whereas the alkyne functionalized precursor (6a) exhibited decreased activity
(231 nM). According to our modeling studies, 2ꢀ and 3ꢀpyridyl substituents were the most
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 60
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
probable hingeꢀbinding moieties to form hydrogen bonds with the hinge, mimicking the
imidazoleꢀpyridazine hingeꢀbinding element of 4. The comparison of compounds 7a and 7d
indicated that the 3ꢀpyridyl hingeꢀbinding element of 7d was more potent in all KIT mutants.
The IC values of both 7a and 7d were in the same nanomolar range for wild type KIT (3.5 nM
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
and 2.1 nM, respectively), but this changed dramatically for the mutant forms (Table 2). To
further evaluate our SAR study, we used 2ꢀpyridyl decorated with electronꢀrich substituents in
the 5ꢀ (7b) and 3ꢀposition (7c). As 2ꢀpyridyl was the less potent pyridine tested, we expected a
more visible increase in potency if the substituents formed additional favorable interactions. 7b
presented an overall decreased potency, especially for KIT T670I, compared to unsubstituted 2ꢀ
pyridyl 7a. Compound 7c substituted in the 3ꢀposition with a methoxy group maintained an
overall potency profile comparable to 7a. As the extension of the sixꢀmembered heterocycle did
not exhibit the anticipated increase, we evaluated a different strategy.
Our docking studies showed that an aptly placed amino group in the paraꢀposition to the
alkyne linker should be able to secure an additional hydrogen bond with the hinge region
(Figure 4). In order to test this hypothesis and simultaneously evaluate heterocyclic diversity, we
synthesized 2ꢀaminopyrimidineꢀ and 3ꢀaminopyridazineꢀcontaining derivatives (7e and 7f,
respectively). This selection was made to determine whether the addition of the amine
substituent in the paraꢀposition leads to increased potency and then whether the nitrogen in the
orthoꢀposition leads to decreased potency even though the nitrogen in the metaꢀposition is
present. As shown in Table 2, the pyrimidine scaffold 7f demonstrated a significantly increased
inhibitory effect in all tested KIT mutants, as well as wild type KIT in a low nanomolar range
(2ꢀ247 nM). Interestingly, the 3ꢀaminopyridazine analog 7e also demonstrates inhibition of wild
type KIT in a low nanomolar range (5.8 nM) but a significant decrease in potency for all KIT
ACS Paragon Plus Environment
18

Page 19 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
mutant forms. Our SAR study demonstrated that the hingeꢀbinding element is important for the
overall inhibitory effect of this series of scaffolds based on 4. Here, a reduction in the size of the
hingeꢀbinding element and the addition of an amino substituent postulated to form an additional
hydrogen bond with the hinge region led to 7f, a small molecule inhibitor of comparable potency
to 4. With 7f in hand, we opted for a few structural analogs to further expand our SAR studies.
Addition of a 2ꢀmethoxyethyl sidechain to the aminopryrimidine nitrogen atom (7g) resulted in
an overall loss of potency for all mutant forms, especially for secondary mutant V654A
compared to 7f. As hingeꢀbinding moieties, 2ꢀaminopyridinꢀ5ꢀyl (7k) and 2ꢀaminoꢀ3ꢀ
methylpyridinꢀ5ꢀyl motifs (7i) exhibit significantly reduced activity for mutant D816H and
secondary mutant T670I. The same behavior was observed for compound 7l. Loss of a nitrogen
atom in the pyrimidine scaffold of 7f and extension at the same site did not seem to be favored.
An obvious change in the scaffold of 4 compared to 1, in addition to the hingeꢀbinding motif, is
the presence of a trifluoromethyl substituent in the back pocketꢀbinding element. This structural
difference prompted us to use the most potent hingeꢀbinding heterocycles of compounds 7a-l and
synthesize the analogs 8a-d without the trifluoromethyl group. Surprisingly, compounds 8a-d
presented no inhibition of the KIT variants we investigated without apparent trends when the
relevant hydrogenꢀtrifluoromethyl matched molecular pairs were considered, as summarized in
Table 2. This result prompted us to investigate a substituent in the back pocketꢀbinding element
that is able to form similar interactions as the trifluoromethyl group. For compounds 9a-b, we
decided on fluoroꢀsubstitution, but this modification had the same overall result as observed for
compounds 8a-d. Our observations of compounds 8a-d and 9a-b strongly indicate that the
trifluoromethyl substituent is crucially important for the potency of these inhibitors, as its
replacement to both hydrogen and fluorine atoms induces a severe loss of potency in all
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 60
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
clinically relevant KIT mutations. Notably, when we repositioned the methyl group decorating
the phenylcarboxamide element of 7f from its original 4ꢀposition on the phenyl to the 2 position
(10) to target the secondary mutation V654A, activity on the V654A and T670I mutants was
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
significantly reduced. We next investigated the selectivity of 7f and 10 over other protein kinase
families. 90 different protein kinases were assessed employing the SelectScreen® Kinase
Profiling Services from Life Technologies (Figure S4, Table S1). The kinase profiling data
underlines the improved selectivity of 10.
Figure 4. (A) Docking pose of 7f in complex with wild type KIT. (B) Close up of the back
pocket. Depicted are the residues spanning the hydrophobic pocket in wild type KIT.
Activity of inhibitors against GIST cells in culture
In addition to biochemical testing, we investigated the inhibitory effects of compounds 5a-d, 6a,
7a-l, 8a-d, 9a-b, and 10 on a human metastatic GIST cell line harboring a 57ꢀbp deletion of exon
11 (GISTꢀT1) that results in the expression of constitutively active KIT. We also examined the
associated gatekeeper mutant T670I (GISTꢀT1ꢀT670I), the second gatekeeper mutant V654A
ACS Paragon Plus Environment
20

Page 21 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(GIST430ꢀV654A), and the independent KIT activation loop mutation D816E (GISTꢀT1ꢀD816E)
(Table 3). Each cell line was treated for 72 h with compounds 5a-d, 6a, 7a-l, 8a-d, 9a-b, and 10
(0ꢀ20 µM), with 4 and 1 used as controls. The tested compounds reduced the number of viable
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
GISTꢀT1 cells, especially compounds 7a-l with a trifluoromethyl substituent in the back pocket
element, with IC50 values in the range of 20ꢀ120 nM. A significant difference between the
compounds was found in the investigation of secondary mutants. Basically, all compounds with
a different substitution pattern in the back pocketꢀbinding element (8a-d, 9a-b) significantly lost
activity in all mutant forms of KIT. In contrast to the biochemical data, the nuances in the effect
of the different hingeꢀbinding motifs were observable. Compounds 8d and 9b, which share the 2ꢀ
aminopyrimidine as a hingeꢀbinding element, exhibit the best activity within this compound
class. Compounds 7a-l had a strong effect on imatinibꢀresistant GISTꢀT1ꢀT670I cells due to their
similarity to 4 and the favorable interactions between the gatekeeper and alkyne linker. The
treatment of GISTꢀT1ꢀD816E and GIST430ꢀV654A cells followed the same trend observed for
the isolated proteins in the biochemical assays. In line with this, compound 7f and structurally
related analogs 7i and 7k exhibited a consequential reduction in cell viability in GISTꢀT1ꢀD816E
in line with the observed biochemical data (7f: IC 42.2 nM, GI 141 nM; 7i: IC 99.3 nM,
5
0
50
50
GI50 474 nM; 7i: IC50 140.4 nM, GI 221 nM). Interestingly, the cellular effect for most
50
compounds of the trifluoromethylꢀsubstituted series resulted in an improved antiproliferative
effect comparing well to the biochemical data for GIST430ꢀV654A cells. In particular,
compound 7f showed the strongest inhibitory effect across all KIT variants both at protein and
cellular level. The 2ꢀaminopyrimidin hingeꢀbinding moiety of compound 7f exhibited improved
efficacy in GIST430ꢀV654A cells compared to 4 (51 nM vs. 149 nM). A KITꢀnegative
GISTꢀ48B was used as a diseaseꢀspecific negative control since it does not express and lost its
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 60
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
oncogenic dependency on KIT. Overall, the tested molecules of the trifluoromethylꢀsubstituted
compound series exhibited an offꢀtarget profile comparable to that of 4, with GI values in the
50
range of 1.8ꢀ6.2 µM indicating a KITꢀspecific mechanism of action. In contrast, compounds 8a-
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
9b resulted in neither a significant decrease in the proliferation of GISTꢀnegative GISTꢀ48B cells
nor a strong inhibitory effect in all KIT variants. An interesting inhibition profile was observed
for compound 10. Though related precursors 5c and 5d exhibited no inhibition of wildꢀtype KIT
or the KIT mutants, it was apparent that small molecule 10, with the reversed substitution pattern
3
4
in positions R and R , causes very low inhibition of KITꢀnegative cells (16 µM) while retaining
good activity with mutant GIST430ꢀV654A (308 nM) and GISTꢀT1ꢀD816E cells (381 nM).
ACS Paragon Plus Environment
22

Page 23 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. (A), (C) Western blot analysis after incubation (3 h) with doses of 7e, 7f, 10, and 4 in
GISTꢀT1ꢀD816E and GIST430ꢀV654A. (B), (D), (F) Corresponding doseꢀresponse curves of
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 60
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
compounds 7e, 7f, 10, and 4 in GIST cells expressing (A) constitutively active KIT with
secondary mutant V654A (GIST430ꢀV654A), (C) constitutively active KIT with secondary
mutant D816E (GISTꢀT1ꢀD816E) and (F) KIT negative cell line GISTꢀ48B after 72 h incubation
with the respective inhibitors. (E) GI values on GISTꢀ48B, GISTꢀT1ꢀD816E, and GIST430ꢀ
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
V654A.
For further characterization of target inhibition in vitro, we then selected the three most potent
compounds from this study (7e, 7f, and 10) and performed western blot analysis in the most
relevant cell lines (GISTꢀT1ꢀD816E and GIST430ꢀV654A; Figure 5). By interrogating KITꢀ
phosphorylation and KITꢀrelated signaling (KIT Y703 (autophosphorylation site)) and related
downstream targets Akt (pS473) and MAPK (pT202/Y204) we observed 7f to be the most potent
compound with strong and almost complete signaling inhibition at 100 nM in GIST430ꢀV654A
and GISTꢀT1ꢀD816E, respectively. Selected compounds 7f, 7e, and 10 reduced the cellular
amounts of pKIT. pAkt, and pMAPK in GISTꢀT1 and GISTꢀT1ꢀD816E cells in a concentrationꢀ
dependent manner. The reduction in phosphorylated proteins in GIST430ꢀV654A and
GISTꢀT1ꢀD816E cells was most distinct with compound 7f, nicely corroborating our
biochemical and cellular results. The tested compounds were the most potent in the series of
trifluoromethylꢀsubstituted molecules. Therefore, the western blot analysis emphasized our
findings in the cellular and biochemical systems. The different substitution pattern of 7e
compared to 7f illustrates that, in addition to the trifluoromethylꢀsubstituent in the back
pocketꢀbinding element, the substitution pattern of the hingeꢀbinding element is crucially
important for inhibitory activity. The western blot analysis points to a favorable nitrogen in the
heterocycle at the 3ꢀposition and an additional hydrogen bond donor at the paraꢀposition,
whereas the nitrogen in the heterocycle at the 2ꢀposition is less favorable.
ACS Paragon Plus Environment
24

Page 25 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Pharmacokinetic characterization of compounds 7f and 10 compared to FDA approved
KIT inhibitors
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We evaluated the pharmacokinetic characteristics of 7f and 10 regarding microsomal stability
(Table S2 and S3, Diagram S1 and S2) plasma stability, cellular absorption (Cacoꢀ2 cell assay)
as well as compound plasma protein binding (Table S4). We compared the values to 1, 3 and 4.
The stability in mouse liver microsomes showed an intrinsic clearance (CL ) of
int
ꢀ1
ꢀ1
ꢀ1
ꢀ1
36 µL min mg for 10, which is in the range of the control compounds (18ꢀ36 µL min mg ).
In contrast, 7f showed three times higher intrinsic clearance. Interestingly, the CL in human
int
ꢀ1
ꢀ1
liver microsomes was much lower for 7f (1 µL min mg ) when compared to 10
ꢀ1
ꢀ1
ꢀ1
ꢀ1
(9 µL min mg ) and well in the range of 4 (1 µL min mg ). We also investigated the plasma
stability of 7f and 10. 7f showed a human plasma stability of 91 % while 10 showed a plasma
stability of 59 %. The reversed behavior was determined for mouse plasma stability (50 % for 7f
and 84 % for 10) both compounds are in the range of the FDA approved inhibitors 2, 3 and 4
(Table S4). Additional to that our compounds show a plasma protein binding of 98.4 % to
99.6 % which are well in the range of 4 (99.8 %) (Table S4). We also investigated the cellular
absorption by determining influx and efflux across a layer of Cacoꢀ2 cells. 7f and 10 show a
better permeability when compared with 4 (Table S4). With a ratio under 1, 7f and 10 show no
significant efflux and are in the same range as 4.
DISCUSSION AND CONCLUSION
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 60
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
Drug resistance to kinase inhibitors due to secondary mutations within the catalytic pocket
6
0
represents a common mechanism and a grand challenge in targeted cancer therapy. In the case
of GIST, attempts to harness progression due to secondary mutations have had moderate success
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
32, 36
as most patients only transiently respond to second and thirdꢀline treatments (2 and 3).
The
improved activity against exon 13/14 (2) and exon 17 mutant KIT (3) is paid for with side effects
resulting from the promiscuous kinase inhibition profile of both drugs. Unfortunately, neither
drug is able to inhibit all mutants nor are drug combinations feasible due to the additive
3
5
toxicity.
Identifying a compound with strong activity against exon 13/14 and 17 is therefore of utmost
clinical relevance not only for GIST, but also for KITꢀdriven leukemias and melanomas.
Here, we present the structureꢀbased design and synthesis of a focused library targeting
clinically relevant KIT variants. We used the general scaffold of 4 as structural basis for a
rational design to advance said scaffold in terms of reduced toxicity and increased potency for
resistant KIT mutation V654A. Therefore, starting from already published Xꢀray
crystallography, we performed molecular mechanicsꢀbased MD simulations of the V654A and
T670I variants, analyzed their differential structural behavior and devised structureꢀbased
hypotheses for our inhibitor design. Using synthetic chemistry, we produced a focused
compound library, paying special attention to variation in the hingeꢀbinding element (resulting in
compounds 7a-7l), variation in the substitution pattern of the back pocketꢀbinding element (8a-
3
4
9
b), and modification of substituents R and R (5c-d and 10), targeting specifically the
secondary mutant V654A. Our SAR studies showed that 7a-l exhibit similar activities and
selectivity profiles toward wild type KIT analogously to the parent compound 4. However, the
observed inhibitory profiles were markedly different for the mutant forms, culminating in the
ACS Paragon Plus Environment
26

Page 27 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
identification of 7d, 7i, 7f, and 10 as the most potent in the series. These results demonstrate that,
in a sixꢀmembered aromatic heterocyclic ring system as the hingeꢀbinding element, the
substitution of the meta and paraꢀposition to the alkyne linker with a protonꢀacceptor and
protonꢀdonor functionality (7f), respectively, results in favorable interactions with all KIT
variants. The enhanced potency of 7f (2ꢀ247 nM for all KIT variants) compared to its related
compounds was due to the formation of an extra hydrogen bond to the kinases’ hinge region. Our
in silico studies further support these findings (Figure 4). The results of our MD simulations
conform quite well to our biochemical and cellular data. The generally longer incubation time in
the cellular context resulted in increased potency, as well as the expected observable nuances in
the effectivity of the inhibitors. Based on the observed higher flexibility of the binding site of
KIT V654A in the apo state (Figure S2), as derived from our MD simulations, a slower binding
of the inhibitors is probable, resulting in an increased energy barrier for complex formation.
Surprisingly, regression of the substitution pattern of the back pocketꢀbinding element to an
imatinibꢀbased substitution pattern with the trifluoromethyl group in lieu of fluorine and
hydrogen atoms, led to a fatal loss of activity for 8a-9b in terms of the mutant forms of KIT, as
well as the anticipated reduction in toxicity. The comparison of modeled compounds 7f, 8d, and
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
9b pointed to a crucial importance of the trifluoromethyl substituent. According to our docking
studies, the trifluoromethyl substituent seems to act as an anchor perfectly embedded in a small
4
8, 59
sub pocket of the back pocket (Figure 4).
This sub pocket is highly conserved in many
55
kinases due to its role in forming the regulatory Rꢀspine . Therefore, in many cases, targeting
5
9, 61, 62
this pocket leads to enhanced potency, but also to poor selectivity and potential toxicity.
Our promising in vitro results prompted us to extend our SAR studies at a cellular level. We
addressed the clinically relevant KIT variants in human, untreated, metastatic GISTꢀT1, ꢀT670I,
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 60
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
and ꢀD816E cells, as well as GIST430ꢀV654A cells. Cellular characterization revealed a reduced
viability of all GIST cell lines in the same range as our in vitro characterizations. 7f turned out to
be the most potent in this series of compounds, with a significant impact on the reduction of
viable GIST430ꢀV654A cells (GI 51 nM) compared to FDAꢀapproved compound 4 (149 nM).
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
This observation proves our approach in altering the hingeꢀbinding element to introduce motifs
with more varied substitution patterns capable of forming additional favorable interactions with
KIT to compensate for the loss of interactions generated by secondary mutation V654A.
3
4
Alteration of substituents R and R in 10 leads to a promising inhibition profile with next to no
activity in GISTꢀnegative cells. In addition, these inhibitors suppress the phosphorylation of KIT,
its mutant forms, and its downstream cascade.
In summary, using structureꢀbased design, we developed inhibitors for various KIT mutants,
paying special attention to secondary mutant V654A. We observed increased inhibitory activity
(
compound 7f) and an optimized inhibition profile concerning toxicity in KITꢀnegative cells
compound 10), which could also be observed in a conducted kinase profiling (Table S1 and
(
Figure S4). The pharmacokinetic characterization showed that the inhibitors 7f and 10 are in the
range of the FDA approved inhibitors concerning the investigated parameters (Table S4). Taken
together, these results will aid in the development of next generation inhibitors to target
clinically relevant KIT mutants.
ACS Paragon Plus Environment
28

Page 29 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
EXPERIMENTAL SECTION
CHEMISTRY.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Unless otherwise noted, all reagents and solvents were purchased from Acros, Fluka, Sigmaꢀ
Aldrich, Alfa Aesar, Combiblocks, Apollo, or Merck and used without further purification. Dry
1
13
solvents were purchased as anhydrous reagents from commercial suppliers. H and C NMR
spectra were recorded on a Bruker Avance DRX 400 spectrometer at 500 and 600 and 126 and
1
151 MHz, respectively. H chemical shifts are reported in δ (ppm) as s (singlet), d (doublet), dd
(doublet of doublet), t (triplet), q (quartet), m (multiplet), and bs (broad singlet) and are
1
3
referenced to the residual solvent signal: CDCl (7.26), DMSOꢀd (2.50) or MeODꢀd (3.34). C
3
6
4
spectra are referenced to the residual solvent signal: CDCl (77.1), DMSOꢀd (39.52) or MeODꢀ
3
6
d4 (49.86). Highꢀresolution electrospray ionization mass spectra (ESIꢀFTMS) were recorded on a
Thermo LTQ Orbitrap (highꢀresolution mass spectrometer from Thermo Electron) coupled to an
“Accela”HPLC system supplied with a “Hypersil GOLD”column (Thermo Electron). Analytical
TLC was carried out on Merck 60 F245 aluminumꢀbacked silica gel plates. Compounds were
purified by column chromatography using Baker silica gel (40−70ꢁm particle size). Preparative
HPLC was conducted on an Agilent HPLC system. The purification of all final compounds to
>95 % purity was conducted by highꢀperformance liquid chromatography (HPLC). The purity
was determined using Agilent 1200 series HPLC systems with UV detection at 210 nm (system:
Agilent Eclipse XDBꢀC18 4.6 mm × 150 mm, 5 ꢀM, 10ꢀ100 % CH CN in H O, with 0.1 % TFA,
3
2
for 15 min at 1.0 mL/min).
ACS Paragon Plus Environment
29