Comparison of the Kinase Profile of Midostaurin (Rydapt) with That of Its Predominant Metabolites and the Potential Relevance of Some Newly Identified Targets to Leukemia Therapy

Article abstract of DOI:10.1021/acs.biochem.8b00727

The multitargeted protein kinase inhibitor midostaurin is approved for the treatment of both newly diagnosed FLT3-mutated acute myeloid leukemia (AML) and KIT-driven advanced systemic mastocytosis. AML is a heterogeneous malignancy, and investigational drugs targeting FLT3 have shown disparate effects in patients with FLT3-mutated AML, probably as a result of their inhibiting different targets and pathways at the administered doses. However, the efficacy and side effects of drugs do not just reflect the biochemical and pharmacodynamic properties of the parent compound but are often comprised of complex cooperative effects between the properties of the parent and active metabolites. Following chronic dosing, two midostaurin metabolites attain steady-state plasma trough levels greater than that of the parent drug. In this study, we characterized these metabolites and determined their profiles as kinase inhibitors using radiometric transphosphorylation assays. Like midostaurin, the metabolites potently inhibit mutant forms of FLT3 and KIT and several additional kinases that either are directly involved in the deregulated signaling pathways or have been implicated as playing a role in AML via stromal support, such as IGF1R, LYN, PDPK1, RET, SYK, TRKA, and VEGFR2. Consequently, a complex interplay between the kinase activities of midostaurin and its metabolites is likely to contribute to the efficacy of midostaurin in AML and helps to engender the distinctive effects of the drug compared to those of other FLT3 inhibitors in this malignancy.

Full text of DOI:10.1021/acs.biochem.8b00727

Subscriber access provided by University of South Dakota
Article
Comparison of the kinase profile of midostaurin
(Rydapt®) with that of its predominant metabolites
Paul William Manley, Giorgio Caravatti, Pascal Furet, Johannes
Roesel, Phi Tran, Trixie Wagner, and Markus Wartmann
Biochemistry, Just Accepted Manuscript • DOI: 10.1021/acs.biochem.8b00727 • Publication Date (Web): 27 Aug 2018
Downloaded from http://pubs.acs.org on August 28, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
Comparison of the kinase profile of midostaurin (Rydapt®) with that of its predominant
metabolites and the potential relevance of some newly identified targets to leukemia
therapy
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
*,†
†
†
‡
§
Paul W. Manley, Giorgio Caravatti, Pascal Furet, Johannes Roesel, Phi Tran, Trixie
†
‡
Wagner, and Markus Wartmann
†
Global Discovery Chemistry, Novartis Institutes for Biomedical Research, Novartis
International AG, CHꢀ4002 Basel, Switzerland
‡
Oncology Disease Area, Novartis Institutes for Biomedical Research, Novartis International
AG, CHꢀ4002 Basel, Switzerland
§
Department of Drug Metabolism and Pharmacokinetics, Novartis Institutes for Biomedical
Research, East Hanover, New Jersey 07936, United States
ACS Paragon Plus Environment

Biochemistry
Page 2 of 53
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: The multiꢀtargeted protein kinase inhibitor midostaurin is approved for the
treatment of both newlyꢀdiagnosed FLT3ꢀmutated acute myeloid leukemia (AML) and KITꢀ
driven advanced systemic mastocytosis (SM). AML is a heterogeneous malignancy and
investigational drugs targeting FLT3 have shown disparate effects in patients with FLT3ꢀmutated
AML, probably as a result of their inhibiting different targets and pathways at the administered
doses. However, the efficacy and sideꢀeffects of drugs do not just reflect the biochemical and
pharmacodynamic properties of the parent compound, but are often comprised of complex
cooperative effects between the properties of the parent and active metabolites. Following
chronic dosing, two midostaurin metabolites attain steadyꢀstate plasma trough levels greater than
that of the parent drug. In this study we characterised these metabolites and determined their
profiles as kinase inhibitors using radiometric transphosphorylation assays. Like midostaurin the
metabolites potently inhibit mutant forms of FLT3 and KIT, as well as several additional kinases
that are either directly involved in the deregulated signaling pathways or which have been
implicated as playing a role in AML via stromal support, such as IGF1R, LYN, PDPK1, RET,
SYK, TRKA and VEGFR2. Consequently, a complex interplay between the kinase activities of
midostaurin and its metabolites is likely to contribute to the efficacy of midostaurin in AML and
helps to engender the distinctive effects of the drug compared to other FLT3 inhibitors in this
malignancy.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 3 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
Often the efficacy and sideꢀeffects of drugs do not only reflect the biochemical and
pharmacodynamic properties of the parent molecule, but result from complex cooperative effects
between the properties of the parent and its active metabolites. This aspect is often overlooked
when attempts are made to assign class effects for both efficacy and adverseꢀevents to a group
of drugs, which is particularly the case for protein kinase inhibitors. The human kinome has a
full complement 538 genes [http://kinase.com/web/current/], most (≥478) of which encode
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
,2
protein kinases with catalytic domains whose sequences are closelyꢀrelated. These protein
kinases, which can be clustered into groups based upon sequence similarity and biochemical
function, catalyse the transfer of the terminal phosphate group of adenosine triphosphate
(ATP) onto the sideꢀchain hydroxy groups of either serine, threonine or tyrosine residues in
substrate proteins, thereby playing a crucial role in the multitude of signal transduction
pathways that regulate the functions of all eukaryotic cells. Drugs designed to target one or
more particular protein kinases in order to elicit a desired pharmacological effect are rarely
specific, but have distinct inhibition profiles both within and across the different groups of
protein kinases. Consequently, it is extremely difficult to assign class effects to drugs acting
upon protein kinases.
ACS Paragon Plus Environment

Biochemistry
Page 4 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 1. Structures of midostaurin (1), staurosporine (2) and the primary metabolites of
midostaurin: CGP62221 (3), e1 (4) and e2 (5). Whereas the 1:1 mixture of e1 and e2 of was
previously designated CGP52421, the definitive stereochemistry of the two epimeric C3ꢀ
hydroxylated compounds has been established in the present study.
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
This situation applies to midostaurin (1; Rydapt®; Figure 1), a drug recently approved by
health authorities for the treatment of two malignancies: (i) acute myeloid leukemia (AML) in
newly diagnosed patients who are FMSꢀlike tyrosine kinase 3 (FLT3) mutationꢀpositive, in
3
combination with chemotherapy; (ii) as monotherapy in advanced systemic mastocytosis
(SM), which includes aggressive systemic mastocytosis, systemic mastocytosis with
4
associated haematological neoplasm and mast cell leukemia. Although the activity of
midostaurin in SM most likely results from targeting the oncogenic D816V mutant form of
5
the stem cell factor receptor tyrosine kinase (KIT), as will be discussed in this article, the
efficacy of midostaurin in AML patients is probably a consequence of the inhibition of
multiple kinases including FLT3, by both the parent compound and several of its metabolites.
6
ꢀ9
Among the first natural products to be discovered as inhibitors of protein kinases,
the
bacterial alkaloid staurosporine (2; Figure 1) was initially found to be a potent inhibitor of the
1
0
PKC family of phospholipaseꢀdependent kinases, although it was later shown to inhibit
9,11,12
and/or bind to many members of the human kinome.
With this knowledge staurosporine
13
served as a lead compound for a drug discovery programme for PKC inhibitors, which
14
culminated in 1986 with the discovery of midostaurin. Based upon the then known
biochemical and pharmacological profile of the drug as both an antiꢀtumour and antiꢀ
15,16
angiogenic agent,
midostaurin was advanced into Phase 1 clinical trials as an antiꢀcancer
17,18
agent and as a treatment for diabetic macular edema.
During the course of these studies it was recognised that the presence of mutant forms of the
FLT3 transmembrane receptor kinase in the leukemic cells of AML patients was associated
ACS Paragon Plus Environment

Page 5 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
1
9ꢀ25
with poor prognosis.
These mutations comprise of either internal tandem duplications
(ITD) at various positions and of varying length (up to 100 aminoꢀacids) within the
25
juxtamembrane domain that compromise the autoregulatory mechanism of FLT3, or of
aminoꢀacid substitutions of the activation loop Asp835 residue that destabilise the inactive
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
26
conformation of the kinase domain. . These mutations render the receptor kinase
constitutively active, such that it is no longer dependent upon extracellular engagement of the
FLT3L growth factor to drive the cell cycle progression and proliferation of the transformed
24
haematopoietic cells. These findings prompted the search for FLT3 kinase inhibitors as
2
7,28
potential therapies for AML.
Midostaurin was duly found to be a potent inhibitor of wildꢀ
type as well as ITD and D835Y mutant forms of FLT3, and this activity translated into
29
efficacy in FLT3ꢀdependent myeloproliferative disease models in mice. Based upon these
findings, midostaurin was advanced into clinical trials in both AML and myelodysplastic
syndrome (MDS) patients who harboured either wildꢀtype or mutated FLT3, where it was
30
demonstrated to have single agent activity. Subsequently a Phase 3 trial in patients aged 18ꢀ
60 years having AML harbouring activating FLT3 mutations, showed that addition of
midostaurin (50 mg, BID) to standard chemotherapy significantly improved eventꢀfree
3,31
survival and overall survival compared to standard chemotherapy,
health authority approvals of the drug in 2017.
thereby leading to
From clinical studies it has now been established that midostaurin undergoes extensive
metabolism by the hepatic CYP3A4 enzyme into three primary metabolites: the 10ꢀOꢀ
demethylated compound CGP62221 (3) and two epimeric C3ꢀhydroxylated compounds, e1
(4) and e2 (5) (Figure 1), with the definitive stereochemistry of these being established in the
present study (previously the e1 + e2 mixture of was designated CGP52421). Thus, following
14
the administration of a single 50 mg dose of [ C]ꢀlabelled drug to fasted, healthy adults (n =
6
), midostaurin, CGP62221, e1 and e2 accounted for 22%, 28%, 5.3% and 33% of the total
3
2
drugꢀrelated mean plasma AUC0ꢀ96hr respectively, and in diabetes patients (n = 9), following
ACS Paragon Plus Environment

Biochemistry
Page 6 of 53
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
5
0 mg BID the mean plasma trough levels on day 28 of midostaurin, CGP62221 and e2 were
3
3
0.82, 1.48 and 6.73 µM. These findings have been further substantiated in AML patients,
where following 50 mg BID the mean steadyꢀstate plasma trough levels of midostaurin,
CGP62221 and [e1 + e2] were 1.25 ± 0.35, 2.08 ± 0.29 and 9.61 ± 1.16 µM, confirming
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
29,34
substantial accumulation of e2 but, as indicated from a single patient, not e1.
In addition to midostaurin, other indolocarbazoleꢀtype panꢀkinase inhibitors, as well as
structurally diverse agents with varying degrees of selectivity towards FLT3 have been
evaluated in clinical trials (Figure 2), and these have been found to elicit quite disparate
35ꢀ37
effects.
Therefore it is important to investigate the potential mechanisms of actions of
midostaurin that contribute to its efficacy in AML. In this study we compare the kinase profile
of midostaurin with that of its predominant metabolites and discuss our findings.
H
H
N
N
HO
O
O
O
O
N
N
N
N
N
N
N
N
O
O
H
H
NH2
O
HO
HO
HN
Crenolanib
(Arog)
UCN-01
(Kyowa Hakko Kirin)
Lestauritinib; CEP701
(Kyowo Hakko Kirin / Cephalon)
F
H N
O
F
F
2
N
H
H
N
O
N
N
N
N
N
NH
N
Cl
O
N
NH
N
Gilteritinib; ASP2215
Pexidartinib; PLX3397
(Plexxikon)
(Kotobuki / Astellas)
H
N
H
N
N
O
O
HN
N
O
N
N
S
H
H
N
N
O
O
O
Cl
N
F
F
F
O
Quizartinib; AC220
(Ambit / Daiichi Sankyo)
Sorafenib; BAY43-9006
(Bayer)
ACS Paragon Plus Environment

Page 7 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
Figure 2. Structures of investigational ATPꢀcompetitive FLT3 kinase inhibitors studied in
AML patients. Like midostaurin, UCNꢀ01, lestauritinib and crenolanib are typeꢀ1 inhibitors
3
8
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
binding to the active conformation of FLT3, whereas quizartinib, pexidartinib and sorafenib
bind to an inactive conformation in a typeꢀ2 fashion.
EXPERIMENTAL DETAILS
Preparation and characterization of compounds:
N-[(9S,10R,11R,13R)-2,3,10,11,12,13-hexahydro-10-methoxy-9-methyl-1-oxo-9,13-epoxy-
1
H,9H-diindolo[1,2,3-gh:3',2',1'-lm]pyrrolo[3,4-j][1,7]benzodiazonin-11-yl]-N-
methylbenzamide (1; midostaurin) was prepared as an amorphous solid by benzoylation of 2
39
as described; 1H NMR (600 MHz, DMSOꢀd6) δ 1.98 (br s, 1H), 2.41 (br s, 1H), 2.29 ꢀ 2.44
(m, 1H), 2.46 (br s, 1H), 2.51 ꢀ 2.62 (m, 1H), 2.75 (br s, 2H), 2.82 (br s, 2H), 2.83 ꢀ 2.90 (m,
2H), 4.51 (br s, 1H), 5.00 (br s, 2H), 7.08 ꢀ 7.12 (m, 1H), 7.30 (t, J=7.76 Hz, 1H), 7.37 (t,
J=7.62 Hz, 1H), 7.43 (br s, 1H), 7.47 (br s, 1H), 7.47 ꢀ 7.57 (m, 3H), 7.63 (br s, 2H), 7.69 (br
s, 1H), 8.07 (d, J=7.89 Hz, 2H), 8.63 (s, 1H), 9.29 (d, J=8.04 Hz, 1H). Solubility (shakeꢀflask
o
method) at pH 1.0 and 6.8 (25 C): 0.3 and 0.1 µg/L respectively.
N-[(9S,10R,11R,13R)-2,3,10,11,12,13-hexahydro-10-hydroxy-9-methyl-1-oxo-9,13-epoxy-
1
H,9H-diindolo[1,2,3-gh:3',2',1'-lm]pyrrolo[3,4-j][1,7]benzodiazonin-11-yl]-N-
methylbenzamide (3; CGP62221). A mixture of (9S,10R,11R,13R)ꢀ2,3,10,11,12,13ꢀ
hexahydroꢀ10ꢀhydroxyꢀ9ꢀmethylꢀ11ꢀ(methylamino)ꢀ9,13ꢀepoxyꢀ1H,9Hꢀdiindolo[1,2,3ꢀgh:3',
40
2',1'ꢀlm]pyrrolo[3,4ꢀj][1,7]benzodiazoninꢀ1ꢀone (2.47 g, 5.4 mmol) and benzoic anhydride
o
(1.22 g, 5.4 mmol) in EtOH (65 mL of 95%) stirred at 70 C for 45 min. The solvent was
evaporated off under reduced pressure and the residue was dissolved in EtOAc. The solution
was washed with HCl (1 M), water, saturated aq. NaHCO
3
and saturated aq. NaCl, dried
(
Na SO ) and the solvent was evaporated off under reduced pressure. The crude product was
2
4
ACS Paragon Plus Environment

Biochemistry
Page 8 of 53
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
purified by column chromatography (silica gel, 5% EtOH in CH
2
Cl
2
) and recrystallised from
o
EtOHꢀCH Cl to afford 3 as a colourless crystalline solid (1.41 g, 33%): m.p. 246ꢀ257 C
2
2
(decomp.); 1H NMR (600 MHz, DMSOꢀd6) δ 9.27 (d, J=8.0 Hz, 1H), 8.60 (s, 1H), 8.04 (d,
J=7.5 Hz, 1H), 7.97 (s, 1H), 7.75 (s, 1H), 7.60 (s, 2H), 7.52 – 7.42 (m, 5H), 7.34 (t, J=7.3 Hz,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
7
H), 7.29 (t, J=7.2 Hz, 1H), 7.12 (s, 1H), 5.78 (d, J=6.9 Hz, 1H), 4.99 (s, 3H), 4.80 (s, 1H),
.72 (d, J=29.6 Hz, 4H), 2.38 (s, 1H), 2.26 (s, 3H). Anal. Cald. for C34
2.20; H, 5.17; N, 9.90. Found: C, 72.00; H, 5.48; N, 9.62.
28 4 4 2
H N O .0.5H O: C,
N-[(3S,9S,10R,11R,13R)-2,3,10,11,12,13-hexahydro-3-hydroxy-10-methoxy-9-methyl-1-
oxo-9,13-epoxy-1H,9H-diindolo[1,2,3-gh:3',2',1'-lm]pyrrolo[3,4-j][1,7]benzodiazonin-11-
yl]-N-methylbenzamide (4). A solution of 2,3ꢀdichloroꢀ5,6ꢀdicyanoꢀ1,4ꢀbenzoquinone
(
DDQ; 0.406 g, 1.75 mmol) in CH
3
CN – H
2
0 (3.5 mL of 1:1) was added to a solution of 1
o
(
1.00 g, 0.75 mmol) in CH Cl (23 mL) at 20 C in the dark. The resulting green solution was
2
2
stirred for 10 min and then extracted with CH
2 2
Cl . The combined extracts were dried
(
Na SO ) and solvent was evaporated off under reduced pressure. The crude mixture was
2
4
purified by column chromatography (silica gel, 0ꢀ3% EtOH in EtOAc) and crystallized from
dioxaneꢀtoluene to afford the less polar epimer, 4 as a colourless crystalline solid (0.32 g,
o
4
2%): m.p. 298ꢀ301 C; 1H NMR (600 MHz, DMSOꢀd6) δ 9.25 (dd, J=7.9, 1.1 Hz, 1H), 8.91
8.81 (m, 1H), 8.47 (dd, J=7.9, 1.2 Hz, 1H), 8.00 (s, 1H), 7.76 – 7.40 (m, 8H), 7.35 (t, J=7.5
–
Hz, 1H), 7.32 (t, J=7.8 Hz, 1H), 7.09 (s, 1H), 6.84 (s, 0H), 6.52 – 6.39 (m, 2H), 5.10 (d,
J=58.7 Hz, 1H), 4.52 (s, 1H), 4.27 (d, J=90.8 Hz, 1H), 3.03 – 2.67 (m, 6H), 2.64 – 2.21 (m,
5H), 1.98 (s, 1H).
Attempts to isolate the more polar epimer, 5 resulted in epimerisation and / or formation of
the 3ꢀmethoxyꢀderivative.
Crystal structure determination and refinement of compound 4: Diffraction data were
collected with a Bruker AXS SMART 6000 CCD detector on a threeꢀcircle platform
ACS Paragon Plus Environment

Page 9 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
goniometer with Cu(Ka) radiation (λ = 1.54178 Å) from a rotating anode generator equipped
with Helios multilayer mirrors. A semiꢀempirical absorption correction (SADABS) was
applied, based on the intensities of symmetryꢀrelated reflections measured at different angular
settings (maximum and minimum transmission 0.6495 and 0.7531). The structure was solved
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
2
by dualꢀspace recycling methods and refined on F with the SHELXTL suite of programs.
Anisotropic displacement parameters were used for all nonꢀhydrogen atoms. All hydrogen
atoms were calculated in idealized positions and refined using a riding model. The final cif
41
file was generated with shelxl 2013/4. The absolute structure was determined to be C2R,
42
C4R, C5R, C6S, C21S; Flack x = 0.06(8), based on 2479 quotients.
1
Compound 4 crystallized in the hexagonal space group P6 with one ordered equivalent of
dioxane. Additional residual electron density in a channel parallel to the crystallographic c
axis (0, 0, z) could not be interpreted unambiguously. The contribution of the disordered
43
solvent to the scattering factors was taken into account with PLATON/SQUEEZE. A total
of 145 electrons was found in the cell, corresponding to approximately three molecules of
dioxane or toluene. The compound 4 to solvent ratio could then be assumed to be
approximately 1:1.5. Where relevant, the crystal data reported (chemical formula, formula
weight M , absorption coefficient µ, F(000), and density D ) are given without the
r
x
contribution of the disordered solvent.
3
Final data: C35
H
30
N
4
O
5
·C
4
H
8
O
2;
M
r
= 674.73, crystal size 0.36·0.03·0.02 mm , hexagonal,
3
space group P 6
1
(No. 169) with a = 19.342(5), c = 16.050(5) Å, V = 5200(3) Å , Z = 6, D
c
=
ꢀ3
ꢀ1
1
.293 gꢁcm , µ = 0.733 mm , F(000) = 2136, 78990 reflections measured, 6061 independent,
R = 0.0324, 2.64º < θ < 66.53º, T = 100(2) K, 455 parameters, 1 restraint, R = 0.0311, wR
int
1
2
= 0.0696 for 5640 reflections with I > 2σ(I), R
1
= 0.0351, wR
2
= 0.0709 for all 6061 data,
3
GoF = 1.051, restrained GoF = 1.051, res. el. dens. = +0.12 / ꢀ0.14 eꢁÅ .
ACS Paragon Plus Environment

Biochemistry
Page 10 of 53
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
Structure assignment of midostaurin metabolites e1 and e2. Midostaurin was incubated with
thawed, cryopreserved human hepatocytes and the incubate was characterised by ultraꢀhigh
performance liquid chromatography, coupled with timeꢀofꢀflight mass spectrometry (UHPLCꢀ
MS). Briefly, vials containing cryopreserved, mixed gender hepatocytes (BiorecalmationIVT,
Baltimore, MD) were thawed at 37°C for 1.5 min and the contents were suspended in 48 mL of
buffer (InVitroGRO HT, cat. #Z99019; BiorecalmationIVT). The cells were recovered by
centrifugation (80 × g for 10 min) and resuspended in a 2% mixture of modified Krebsꢀ
Henseleit buffer for hepatocyte incubation (In VitroGRO KHB, cat. # Z99074;
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
BiorecalmationIVT) in fetal calf serum (FCS; Invitrogen, Walkersville, MD). An aliquot of the
final suspension was removed for cell counting and for cell viability estimation (Nucleoconuter
6
NCꢀ3000). A 500 µL aliquot of the final cell suspension (~2 × 10 cells/mL) was added to one
well of a 12ꢀwell plate predispensed with 0.5 mL of In VitroGRO KHB containing 2% FCS and
midostaurin (10 ꢂM) and incubated for 24 h without shaking, at 37°C in a humidified cell culture
2 3
incubator (5% CO , 95% air). The well was emptied and washed with CH CN (2 mL). Aliquots
(1 mL) of the combined contents and washings were centrifuged (13500 rpm for 10 min) at 10°C
and evaporated to dryness (GeneVac, LTD Ipswich, UK) at 30°C. The residues were
reconstituted in aqueous MeOH (200 µL of 50%) and 10 µL aliquots of the cell lysate mixture
were injected into the UHPLCꢀMS system for analysis.
ACS Paragon Plus Environment

Page 11 of 53
Biochemistry
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 3. Ultraꢀhigh performance liquid chromatography coupled with timeꢀofꢀflight mass
spectrometry (UPLCꢀMS) evidence confirming that the less polar midostaurin metabolite e1,
formed from human hepatocytes, has 3ꢀ(R)ꢀstereochemistry. Top panel: Product ion mass
spectra of midostaurin metabolite e2, which eluted at 24.9 minute, characterized by m/z 569
2
(corresponding to loss of H O). Middle panel: Product ion mass spectra of midostaurin
metabolite e1, which eluted at 22.5 minute. Bottom panel: product ion mas spectrum
synthesised standard sample of 3ꢀ(S)ꢀhydroxylated compound (4).
The hydroxylated epimer e1 of the CGP52421 mixture eluted at 22.5 minute and epimer e2
eluted at 24.9 minute. Consistent with the ease of methoxylation discussed above, under the
UPLCꢀMS conditions, source fragmentation (loss of water) was pronounced for e2 (Figure 3).
The authentic C3ꢀhydroxylated compound 4 had a retention time and product ion mass spectrum
which exactly matched that previously identified as epimer e1 (Figure 3).
Biochemical evaluation of effects on kinase activity: The kinase profiles of compounds were
determined at ProQinase GmbH (Freiburg, Germany). All kinases were produced from human
cDNAs either as fullꢀlength or enzymatically active fragments, expressed in Sf9 insect cells or in
ACS Paragon Plus Environment

Biochemistry
Page 12 of 53
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
Escherichia coli as recombinant glutathione Sꢀtransferase (GST) fusion proteins or Hisꢀtagged
proteins, and purified by either GSH affinity chromatography or immobilized metal affinity
chromatography (affinity tags were removed from a number of kinases during purification). The
purity and identity of each kinase was checked by SDSꢀPAGE/silver staining and by western
blot analysis with specific antibodies or, in the case of lipid kinases by mass spectrometry.
Kinases provided by vendors (Carna Biosciences, Inc.; Invitrogen Corp.; Millipore Corp.) were
expressed, purified and quality controlled based on vendorꢀsupplied information.
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
The effects of compounds on human protein kinases were assessed in radiometric assays
33
( PanQinase®), performed using a BeckmanCoulter Biomek 2000/SL robotic system, with
96ꢀwell FlashPlates (PerkinꢀElmer, Boston,MA) in a 50 µL reaction volume. The reaction
2
cocktail was pipetted in four steps: (i) 10 µL of nonꢀradioactive ATP solution (in H O); (ii) 25
33
µL of assay buffer–[γꢀ P]ATP mixture; (iii) 5 µL of test compound in 10% DMSO; (iv) 10 µL
of enzymeꢀsubstrate mixture. The assay for all enzymes contained 70 mM HEPESꢀNaOH (pH
33
7
2
.5), 3 mM MgCl , 3 µM sodium orthovanadate, 1.2 mM dithiothreitol, ATP–[ γꢀ P]ꢀATP
5
(variable amounts, corresponding to the apparent ATP Km of the respective kinase; ≈ 8 x 10
cpm/well), and purified protein kinase and substrate (both variable amounts). The concentrations
of ATP, enzymes and substrates employed are shown in Supplementary Table 1. Additional
chemicals incorporated in specific assays were as follows: All PKC assays (except for PKCꢀµ
and ꢀν) also contained 1mM CaCl
dioleylꢀglycerol; the CAMKꢀ1D, ꢀ2A, ꢀ2B, ꢀ2D, ꢀ4, ꢀK1, ꢀK2, DAPK2, EEF2K, MYLK,
MYLK2 and MYLK3 assays included 1 µg/mL calmodulin and 0.5 mM CaCl ; the PRKGꢀ1 and
2
, 4 mM EDTA, 5 µg/mL phosphatidylserine and 1 µg/mL 1,2ꢀ
2
ꢀ2 assays contained 1 µM cGMP; the DNA pharmacokinetic (PK) assay contained 2.5 µg/mL
DNA. The reaction cocktails were incubated at 30°C for 60 min and reactions then stopped by
adding 50 µL of 2 % (v/v) H
3
PO
4
. Plates were then aspirated, washed with NaCl (2 x 200 µL
33
0
.9 % w/v) and the incorporation of Pi (counting of ‘cpm’) was determined with a
microplate scintillation counter (Microbeta, Wallac). For each kinase, the median value of the
ACS Paragon Plus Environment

Page 13 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
cpm of three wells was defined as ‘low control’ (n=3), which reflected unspecific binding of
radioactivity to the plate in the absence of enzyme, but in the presence of substrate. In addition,
the median value of the cpm of three other wells was taken as the ‘high control’, corresponding
to full activity in the absence of any inhibitor (n=3). The difference between high and low control
of each enzyme was taken as 100% activity. For data evaluation, the low control of each kinase
was subtracted from the high control value as well as from their corresponding ‘compound
values’ and the residual activity (in %) for each compound well was calculated by using the
formula: Residual Activity (%) = 100 x [(signal of compound – low control) / (high control –
low control)]. To measure IC50 values against selected enzymes, 10 concentrations of each
compound in the 0.3 nM ꢀ 10 µM range were used and values were calculated using Quattro
Workflow V3.1.1 (Quattro Research GmbH, Munich, Germany). The fitting method was a
leastꢀsquares fit based upon the ‘sigmoidal response (variable slope)’ with parameters ‘top’
fixed at 100% and ‘bottom’ at 0%.
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
Effects on lipid kinase activity were assessed using a nonꢀradiometric ADPꢀGlo™ Assay
(Promega, Madison, Wi, USA) performed in 96ꢀwell halfꢀarea microtiter plates (Greiner Bioꢀ
One, Frickenhausen, Germany) in a 25 ꢂL reaction volume. The reaction cocktail was
pipetted sequentially: (i)10 ꢂL of ATP solution (variable concentrations, corresponding to the
apparent ATPꢀKm of the respective kinase) in assay buffer (50 mM HEPESꢀNaOH, pH 7.5, 1
mM EGTA, 100 mM NaCl, 0.03% CHAPS, 2 mM DTT; assays for PI4KB, PIK3C2A,
PIK3C2B, PIK3C3, PIK3CA/PIK3R1, PIK3CD/PIK3R1 and PIK3CG additionally contained
3
2
mM MgCl ); (ii) 5 ꢂL of test compound (10 ꢂM) in 10% DMSO; (iii) 10 ꢂL of variable
amounts of enzymeꢀsubstrate mixtures; the concentrations of ATP, enzymes and substrates
employed are shown in Supplementary Table 1. After incubation at 30°C for 40 min,
reactions were stopped with 25 ꢂL ADPꢀGlo reagent per well. Plates were then incubated for
40 min at room temperature, followed by addition of 50 ꢂL kinase detection reagent per well
and incubated for a further 60 min at room temperature. Signals were measured as ‘counts per
ACS Paragon Plus Environment

Biochemistry
Page 14 of 53
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
second’ (cps), using a Victor2 microplate multilabel reader (Perkin Elmer, Boston, Ma, USA)
in luminescence mode. For each kinase, the median cps value of three wells with complete
reaction cocktails, but without kinase, was defined as ‘low control’. Additionally, for each
kinase the median value of the cps of three other wells with the complete reaction cocktail,
but without any compound, was taken as the ‘high control’, i.e. full activity in the absence of
any inhibitor. The difference between high and low control of was taken as 100% activity for
each kinase. As part of the data evaluation the low control of each kinase was subtracted from
the high control value as well as from their corresponding ‘compound values’. The residual
activity (in %) for each compound well was calculated by using the formula: Residual
Activity (%) = 100 x [(cps of compound – low control) / (high control – low control)].
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
Evaluation of effects on proliferation / viability of human cancer cell lines: The cellular
effects of midostaurin were determined on 469 commercially available human cancer cell
44
lines (including 69 hematologic) using the cancer cell line encyclopoedia (CCLE) screen.
Cellꢀline identities were confirmed by singleꢀnucleotide polymorphism genotyping and they
were all shown to be free of mycoplasma by PCR. The screen employed the CellTiterꢀGlo®
Luminescent Cell Viability Assay (Promega, Madison, USA) based upon quantitation of ATP
4
4
as an indicator of metabolically active cells according to the published method, to assess the
number of viable cells in culture (appropriate media supplemented with 10% FCS) following
72 h incubation with drug in a 1536ꢀwell format. The maximum effect level (Amax) and the
inflection point were taken from 8ꢀpoint doseꢀresponse curves, several of which are illustrated
in Supplemental Figure S1, and the results are summarised in Figure 4.
Evaluation of effects on proliferation / viability of kinase dependent cell lines: The
potential for compounds to inhibit particular kinases in a cellular context was further
quantified in the BaF3 assay system consisting of wildꢀtype ILꢀ3ꢀdependent hematopoietic
BaF3 cell models rendered ILꢀ3 independent by transduction with various constitutively
ACS Paragon Plus Environment

Page 15 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
4
5
active tyrosine kinases, and in human cell lines carrying oncogenic kinases. BaF3ꢀBCRꢀ
ABL1 cells were obtained by transfecting the interleukinꢀ3ꢀdependent murine hematopoietic
BaF3 cell line with a pGD vector containing wildꢀtype (p210 kD) BCRꢀABL1 (B2A2)
26,61,62
cDNA.
BALB/c 3T3 A31 mouse embryonic fibroblasts (ATCC Cat. CCLꢀ163),
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
expressing PDGFRꢀA and ꢀB were obtained from B. J. Druker (Oregon Health and Science
University, Portland, Oregon). BaF3ꢀTelꢀPDGFRβ,28 transduced with a fusion protein
comprised of the dimerising portion of TEL and the transmembrane and kinase domain of the
PDGFRβ, were provided by D. G. Gilliland (Brigham and Women’s Hospital and Harvard
Medical School, Boston, Massachusetts). GIST882, a human gastrointestinal stromal tumour
(
GIST) cell line expressing an activating KIT mutation (exon 13, K642E) was provided by J.
Fletcher (MIT Cancer Center and Department of Biology, Cambridge, Massachusetts).
MOLM13 were obtained from Yoshinobu Matsuo, Fujisaki cell center, Hayashibara
Biochemical Labs Inc. 675ꢀ14 Fujisaki, Okayama 702ꢀ8006, Japan. MV4ꢀ11 were obtained
from J. D. Griffin, DanaꢀFarber Cancer Institute, Boston, Massachusetts.
Cells were cultured in RPMIꢀ1640 (Amimed cat. # 1ꢀ14F01ꢀI) supplemented with 2% Lꢀ
glutamine (Amimed cat. # 5ꢀ10K50ꢀH) and 10% FCS (Amimed cat. # 2ꢀ01F16ꢀI). Wildꢀtype,
parental BaF3 cells were maintained in the above medium plus 10 U/mL recombinant mouse
ILꢀ3 (Roche # 1380745). The resazurin sodium salt dye reduction assay kit (Alamar Blue™;
cat. # DAL1100, BioSource International Inc.) was generally used to measure cell
proliferation according to supplier instructions. Briefly, 15000 cells were seeded in 190 ꢂL
fresh medium into 96ꢀwell plates, followed by addition of 10 ꢂL medium containing
compound dilutions at 20ꢀfold their final intended concentration. Cells treated with vehicle
(
DMSO, 0.1% FCS) only served as controls. Doseꢀresponse effects were determined by 3ꢀ
fold serial dilutions of the test compound, starting at 10 ꢂM. Following incubation of the cells
for 48 h at 37°C and 5% CO , the effect of inhibitors on cell viability was assessed following
2
addition of 20 ꢂL resazurin sodium salt solution (130 ꢂg/mL PBS) and incubated for an
ACS Paragon Plus Environment

Biochemistry
Page 16 of 53
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
4
6
2
additional 6 h at 37°C and 5% CO . The levels of resorufin, were quantified using a
SaphireII 96ꢀwell fluorometer (TECAN, Männedorf, Switzerland) with excitation and
emission wavelengths set at 544 and 590 nm, respectively. In addition, a plate blank value
was determined in a well containing only 100 ꢂL of medium and no cells. Acquired raw data
were exported to Excelꢀfile format. For data analysis, the plate blank value was subtracted
from all data points. The effect of a particular test compound concentration on cell
proliferation and viability was then expressed as percentage of the corrected fluorescence
reading obtained for cells treated with vehicle only, which was set as 100%. GI50 values were
determined using XLfit (V4.2) curveꢀfitting software, applying standard four parameter
logistic model #205 (IDBS, Guilford, UK). The luminescent ATP detection assay kit,
ATPLite™ (Perkin Elmer Life Sciences; cat. # 6016947), based upon the production of light
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
(
luminescence) caused by the reaction of ATP with added luciferase and Dꢀluciferin, was
used to measure the proliferation of GIST882 cells following 70 h incubation with
47
midostaurin according to our published method.
RESULTS
Biochemistry. Kinase inhibition was determined using recombinant enzymes in
transphosphorylation assays. Initially the effects of compounds (two determinations in all cases)
at a fixed concentration of 10 µM (solubility became limiting at higher concentrations) was
assessed against 320 wildꢀtype protein kinases and 13 phospatidylꢀinositol lipid kinases. In this
preliminary screen, midostaurin, 3 and 4 gave residual activity of <50% (n = 2) for 159, 178 and
1
03 protein kinases respectively (corresponding to selectivity scores at 10 µM of 0.534, 0.575
12
and 0.347), whereas none of the lipid kinases were substantially inhibited (Table 1).
Table 1: Comparison of the overall selectivity profiles of midostaurin and major metabolites
against 94 wildꢀtype protein kinases.
Number of protein kinases inhibited
Compound
IC50 ≤ 10 μM
IC50 ≤ 1.0 μM
IC50 ≤ 0.1 μM
IC50 ≤ 0.01 μM
Midostaurin (1)
93
81
22
0
ACS Paragon Plus Environment

Page 17 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
CGP62221 (3)
4 (e1)
94
85
86
75
24
27
16
1
1
0
0
1
4
+ 5 (e1 + e2)
2
Based upon the extent of percentage residual phosphorylation activity compared to control in the
full panel and the pharmacological importance of the target, doseꢀresponse curves were then
generated and IC50 values determined for midostaurin and the metabolites 3, 4 (e1) and
CGP52421 (a 1:1 mixture of e1 + e2) against 94 wildꢀtype enzymes, together with two mutant
forms of FLT3 and nine mutant forms of KIT (Tables 1 and 2). In all cases investigated, where
the residual activity was >50% in the preliminary screen, the measured IC50 value was >5 µM.
Most of the wildꢀtype kinases showing substantial sensitivity (IC50 < 3 µM) to the four
compounds were members of the tyrosine kinase (TK) group (40 of 76 evaluated; full
complement 90), although at a concentration of 10 µM none of the compounds reduced the
residual activity of any of the ephrin TKs by >50%. Many serineꢀthreonine kinases of the AGC
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
(
protein kinase A, G and C kinases), CAMK (Caꢀ and calmodulinꢀregulated kinases) and STE
(homologues of the yeast sterile kinases) families were also potently inhibited, although those of
the CMGC (cyclinꢀdependent kinases, MAP kinases, glycogen synthase kinases, casein kinase 2)
and CK1 (casein kinase) groups were insensitive at concentrations < 10 µM.
Table 2: Comparison of the concentrations of midostaurin and major metabolites required to
inhibit the transphosphorylation activity of selected protein kinases (IC50 values in ꢂM).
Kinases are tabulated according to group.
1,2
Kinase
3
4
(e1)
4 + 5
(e1 + e2)
Midostaurin
1
(listed by group)
(CGP62221)
TK: Tyrosine kinase group
ACK1
0.079
0.645
0.183
0.578
0.669
0.190
0.708
0.566
0.687
0.253
0.989
0.143
0.846
0.576
0.240
1.39
0.227
2.25
0.259
0.879
0.344
0.929
2.60
ALK (GST-HIS-tag)
BLK
0.100
1.37
BMX
BTK
2.17
CSF1R
CSK
0.192
1.30
0.523
1.28
0.668
1.38
DDR2
EGFR
FER
0.670
1.48
0.976
1.69
1.72
1.43
0.056
1.01
1.78
0.878
2.90
FES
4.75
ACS Paragon Plus Environment

Biochemistry
Page 18 of 53
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
FGFR2
0.225
0.048
0.026
0.014
0.263
0.392
0.220
6.32
0.241
0.062
0.024
0.014
0.858
1.54
1.87
1.19
0.554
1.66
FLT3 wild-type
FLT3 ITD
FLT3 D835Y
FYN
0.337
0.186
0.863
1.28
0.361
0.277
1.24
HCK
1.04
IGF-1R
0.179
2.93
0.154
7.58
0.130
6.48
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
INSR
INSR-R
2.05
1.37
4.47
3.96
JAK2
0.229
0.056
2.17
0.415
0.062
4.46
1.80
1.71
JAK3
0.459
3.79
0.300
6.23
KIT wild-type
KIT A829P
KIT D816H
KIT D816V
KIT T670I
KIT V559D
KIT V559D / T670I
KIT V559D / V454A
KIT V560G
KIT V654A
LCK
0.533
0.045
0.066
0.058
0.392
0.083
0.432
0.101
1.21
1.38
2.54
2.55
0.084
0.080
0.115
0.673
0.218
1.21
0.229
0.197
2.42
0.663
0.615
2.21
0.920
1.66
1.39
1.84
2.15
3.21
0.110
1.23
0.426
4.21
0.392
4.18
0.144
0.180
0.210
1.07
2.22
1.09
0.968
0.307
0.429
4.69
LTK
0.292
0.810
0.691
0.330
0.335
0.107
0.068
0.044
1.94
2.00
LYN
0.383
5.22
MERTK
MUSK
0.275
0.304
0.081
0.026
0.128
0.469
0.194
0.074
0.050
0.123
0.055
0.357
0.015
0.683
3.05
2.77
PDGFR-α
PDGFR-β
RET
0.811
0.300
0.940
1.62
1.04
0.451
1.39
ROS
3.45
SRC (GST-HIS-tag)
SYK (aa1-635)
TNK1
2.82
2.07
0.126
0.224
0.016
0.047
0.079
0.568
0.029
0.577
1.87
0.983
1.56
1.18
TRK-A
1.20
0.489
0.489
1.43
TRK-B
1.71
TRK-C
1.22
TYK2
3.27
4.53
VEGFR-2
ZAP70
0.057
3.02
0.062
2.11
AGC: Protein kinase A, G and C family of serine-threonine kinases
GRK7
PDPK1
PKA
0.187
0.078
0.409
0.220
0.419
0.210
0.225
0.109
0.223
0.100
0.048
0.080
0.071
0.896
0.646
0.944
0.454
0.692
1.49
2.06
0.385
1.11
4.38
7.67
4.60
3.50
3.32
7.95
0.908
1.56
2.09
0.097
1.79
1.53
2.71
1.44
1.07
1.70
7.16
1.15
1.86
PKC-α
PKC-β1
PKC-β2
PKC-γ
PKC-δ
PKC-ε
PRK1
2.13
0.333
0.296
PRK2
ACS Paragon Plus Environment

Page 19 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
PRKG1
0.181
0.018
0.878
0.051
0.070
0.033
0.214
0.007
1.00
1.25
0.118
6.87
1.56
0.174
9.13
PRKG2
PRKX
RPS6KA2
RPS6KA3 (alias RSK2)
RPS6KA6
0.043
0.070
0.044
1.05
0.969
1.04
0.898
0.974
1.02
CAMK: Calcium and calmodulin-regulated group of serine-threonine kinases
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
BRSK2
CAMK2A
CAMK2D
CAMK2G
CAMKK2
CHK1
1.66
0.445
0.170
0.681
0.136
1.80
0.780
0.820
0.251
1.56
> 10
8.58
5.18
8.89
1.90
8.76
> 10
> 10
9.11
6.49
1.23
> 10
1.44
> 10
0.787
0.296
3.85
5.34
5.60
> 10
6.99
4.85
8.90
1.31
2.31
> 10
> 10
> 10
8.50
2.20
> 10
1.17
4.64
1.08
0.700
5.07
7.38
6.92
1.08
0.261
1.94
DAPK1
DAPK3
MELK
8.60
5.16
1.14
0.563
1.18
0.341
0.593
0.120
0.315
0.281
0.294
0.675
0.125
0.598
0.914
0.521
MYLK
PHKG1
PHKG2
PIM1
0.056
2.30
0.388
0.704
0.235
0.111
0.294
1.78
PIM3
SIK1
SIK2
SNRK
STK17A
STK33
0.385
STE: Homologues of the yeast sterile kinase family of serine-threonine kinase
MAP3K7 / MAP3K71P1
MAP3K9 (alias MLK1)
MAP3K10
0.355
0.073
0.063
0.029
0.290
0.470
0.032
0.256
0.309
1.31
0.224
0.361
0.195
0.966
1.76
3.32
0.527
0.902
0.484
2.84
2.81
0.221
0.425
0.278
3.03
MAP3K11
MAP4K2
MAP4K5
> 10
> 10
MST1 (alias STK4)
MST2
0.137
0.978
0.959
0.648
2.11
0.602
2.48
SLK
7.36
6.36
TKL: Tyrosine kinase-like family
LRRK2
0.102
0.163
4.17
5.11
Other
AURK-A
AURK-B
AURK-C
GSG2
0.164
0.247
0.305
7.86
0.247
0.192
0.459
0.863
0.587
0.662
0.630
0.358
1.99
2.08
0.757
4.65
> 10
1.82
2.27
1.63
> 10
> 10
3.78
0.992
8.51
> 10
1.07
8.38
0.896
> 10
8.96
IKK-ε
0.100
0.157
0.135
0.436
> 10
SAK (alias PLK4)
TKB1
TSF1
WEE1
1
Apparent IC for 1:1 mixture.
50
ACS Paragon Plus Environment

Biochemistry
Page 20 of 53
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
Midostaurin, in addition to potently inhibiting wildꢀtype FLT3, as well as mutant forms of FLT3
and KIT, inhibited 21 additional kinases with IC50 values ≤ 100 nM and a total of 81 wildꢀtype
kinases with IC50 values ≤ 1000 nM. CGP62221 (3) showed a similar pattern of selectivity
against the 320 protein kinases, as well as similar degrees of potency to that of midostaurin,
although it was slightly more potent (≈ 3ꢀfold) against CHK1, FER, PRKG2, ROS, TRKꢀA / –B,
and markedly less active (> 6ꢀfold) against LCK, PKCꢀδ / ꢀε, PRK2, CAMKK2 and MAPK11.
Both 4 (e1) and e2 maintained potency against VEGFR2 (neither midostaurin nor the
metabolites inhibited VEGFR1 by >50% at 10 µM), but in general the presence of a hydroxylꢀ
group in the lactamꢀring resulted in reduced activity. Metabolite e1 (4) only inhibited 13 kinases
with IC50 values < 400 nM, all of which were potently inhibited by both midostaurin and
CGP62221. In comparison to e1 (4), e2 (5) was apparently more active (IC [e1+e2] ≤ 3ꢀfold
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
50
IC50 [e1]) against FGFR2, LTK, TRKB and PDPK1, and clearly much less active (IC50 [e1+e2]
≥ 3ꢀfold IC50 [e1]) against BLK and the D816 mutant forms of KIT.
Cell proliferation /viability assays
ACS Paragon Plus Environment

Page 21 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. Antiproliferative profile of midostaurin across a large panel of human cancer cell
lines. The scatter plot shows the maximun effect level, Amax (%), versus inflection point (µM)
of midostaurin in cell viability assays assessed on 469 cell lines. Cell lines with Amax ≤30%
are typically classified as nonꢀresponding. Cells carrying mutated FLT3 (MOLM13 and
MONOMAC1) are depicted as red triangles, with 65 other hematopoietic cell lines depicted
as yellow triangles and solid tumor cell lines as blue circles.
Although midostaurin impacted the viability of many cell lines in the CCLE screen at
concentrations > 400 nM (Figure 4), it showed considerable selectivity at lower
concentrations (Amax levels < 30% were not regarded as being of significance). The greatest
activity was seen towards the human AML cell lines MOLM13 and MONOMAC1, the
viability of which was substantially curtailed at concentrations ≤ 100 nM (due to the curve
fitting algorithm employed the MOLM13 Amax is underestimated and should be ≈70%;
Supplemental Figure S1). The proliferation of OCIꢀAML2 AML, MOLPꢀ8 multiple myeloma
ACS Paragon Plus Environment

Biochemistry
Page 22 of 53
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 LNꢀ405 glioma cell lines were also substantially impacted at concentrations in the range
of 300ꢀ400 nM.
Table 3: Comparison of the effects of midostaurin, CGP62221 and [4 + 5] (the CGP524211:1
mixture of e1 and e2) on cell lines dependent and independent upon midostaurinꢀsensitive
(IC50 < 500 nM; no fill) and –insensitive kinases (grey fill) and untransduced BaF3 cells
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
(mean GI50 nM ± SEM; n ≥ 3).
1
Effect on cell proliferation
Kinase
Cell line
Midostaurin CGP62221 (3)
4 + 5
FLT3-ITD
FLT3-ITD
FLT3-ITD
IGF-1R
BaF3-FLT3-ITD
MOLM13 (heterozygous)
MV4-11 (homozygous)
BaF3-Tel-IGF-1R
BaF3-KIT-D816V
BaF3-Tel-PDGFRβ
BaF3-PTC3-RET
39 ± 2
48.4 ± 6.9
26.3 ± 7.1
319 ± 38
88 ± 6
28 ± 6
656 ± 155
n.d.
2
n.d.
n.d.
n.d.
189 ± 31
50 ± 7
1315 ± 269
319 ± 28
63 ± 9
KIT (D816V)
PDGFRβ
RET
3
19 ± 3
< 12
96 ± 7
167 ± 27
1218 ± 85
196 ± 54
497 ± 82
152 ± 15
n.d.
1216 ± 180
> 10000
2316 ± 630
5071 ± 323
1046 ± 201
n.d.
4
ABL1
BaF3-BCR-ABL1
BaF3-NPM-ALK
655 ± 91
364 ± 8
373 ± 41
253 ± 25
ALK
FGFR3
INSR
BaF3-Tel-FGFR3
BaF3-Tel-INSR
5
KIT (K642E)
GIST882
399 ± 171
388 ± 11
Untransduced BaF3 wild-type
327 ± 89
3657 ± 459
1
The Alamar Blue™ assay was used for all cell lines with the exception of GIST882, where
2
the ATPLite™ assay was employed (see Experimental section for details); n.d.: not
3
4
determined; three determinations gave IC50 values of 12, < 5 and < 5 nM; none of the
compounds reduced the residual transphosphorylation activity of ABL1 by >35% at 10 µM;
midostaurin inhibited K642E KIT autophosphorylation with an IC50 value of 4523 ± 1805
5
4
8
(n=5) in GIST882 cells as determined by capture ELISA (J. Mestan, unpublished results).
The effects of compounds on cell viability was further tested on a small panel of human cancer
cells and engineered murine hematopoietic cell lines. Murine interleukin (IL)ꢀ3ꢀdependent proꢀ
B lymphoma cells (wildꢀtype BaF3) were used to generate subꢀlines whose proliferation and
survival was rendered ILꢀ3 independent by stable transduction with individual tyrosine
45,48
kinases either activated by mutation or fusion with a dimerizing protein partner.
The
effects of compounds on cell viability were generally assessed using the resazurin assay in
which viable cells reduce resazurin to the highly fluorescent resorufin, whereas nonꢀviable
46
cells rapidly lose their reductive capacity and fail to produce a fluorescent signal, although
in the case of GIST882 the ATPLite™ assay was employed to measure ATP concentrations
ACS Paragon Plus Environment

Page 23 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
4
7
which rapidly decline when cells undergo necrosis or apoptosis. The effects on cell viability
are compared in Table 3. In general, biochemical kinase inhibition translated into
antiproliferative activity. Midostaurin and CGP62221 (3) potently inhibited (GI50 < 100 nM)
the proliferation of those cell lines driven by FLT3, D816V KIT, PDGFRβ and RET.
However both compounds, also inhibited the viability of wildꢀtype BaF3 cells (GI50 300ꢀ400
nM), and midostaurin in particular reduced the viability of BaF3ꢀBCRꢀABL1, BaF3ꢀTelꢀ
INSR and human gastrointestinal stromal GIST882 cells. In contrast, the epimeric mixture of
metabolites [e1+e2] substantially inhibited the proliferation of only the TelꢀPDGFRβ (GI50 63
nM), D816V KIT (GI50 320 nM) and FLT3ꢀITD (GI50 650 nM) BaF3 cell lines, while the
wildꢀtype cells were relatively insensitive.
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
DISCUSSION
Collectively AML is the most common form of adult leukemia, having an incidence of 4.2 per
1
00,000 population with an overall 5ꢀyear survival rate of 27% in the U.S.A. in the period
49
2007ꢀ2013. However, AML is a heterogeneous malignancy that arises from hematopoietic
progenitor cells which progressively acquire large, diverse sets of coexisting cytogenetic and
epigenetic lesions leading to the activation of proꢀproliferative pathways and impaired normal
50,51
hematopoietic differentiation.
Consequently, as suitable targeted therapies become
37
available, different AML genotypes should mandate different therapeutic interventions.
Gainꢀofꢀfunction mutations in the FLT3 gene are detected in about 30% of AML patients,
with the majority (≈ 66%) of these being ITD mutations that confer an adverse prognosis
compared to FLT3 kinase domain mutations, or mutations in other genes. Consequently,
much hope has been engendered that inhibiting aberrant FLT3 signalling in leukemic cells
will provide therapeutic benefit to some groups of AML patients. However, assessing the
potential for kinase inhibitors that target FLT3 was confounded in early clinical trials, because
the different drugs investigated (Figure 2) had very different target profiles which made
ACS Paragon Plus Environment

Biochemistry
Page 24 of 53
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
3
5
attributing therapeutic effects to biochemical mechanisms problematic. Midostaurin
(Rydapt®) has been developed as an FLT3 inhibitor and is the first targeted therapy to receive
3
1
Health Authority approval for the treatment of AML, but as presented here its efficacy is
probably the result of a complex interplay between the kinase activities of the drug and its
metabolites.
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
In order to compare the kinase profiles of midostaurin with that of the three major
metabolites, in addition to CGP62221 (3) which was readily prepared by benzoylation of the
methylamine precursor available from the fermentation broth of a mutant strain of
40
Streptomyces longisporoflavus, it was necessary to have a discrete sample of at least one of
the two epimers of the CGP52421 mixture (Figure 1). Using the method of Kasai and coꢀ
52
workers, oxidation of 1 with DDQ afforded a 1:1 mixture of the two epimeric C3ꢀhydroxylated
derivatives, e1 and e2, as detected in midostaurinꢀtreated patients. Whereas the more lipophilic
epimer, e1 was sufficiently soluble to be purified by column chromatography, epimer e2 defied
attempts at purification and, when dissolved in protic solvents, such as methanol afforded an
epimeric mixture of 3ꢀmethoxy derivatives. Singleꢀcrystal Xꢀray diffraction studies of e1
showed it to correspond to 4, having (3S,9S,10R,11R,13R) absolute stereochemistry, thus
inferring that e2 had (R)ꢀstereochemistry at the corresponding 3ꢀposition of 5. The instability
53
of 5 is analogous to that reported for the hydroxylated staurosporine, UCNꢀ01 (Figure 2)
and such epimerisations probably stem from the ease of formation of an intermediate 1Hꢀ
5
4ꢀ56
isoindolinꢀ1ꢀone.
Liquid chromatography ꢀ mass spectroscopy studies unambiguously
confirmed that 4 corresponds to epimer e1, thus allowing in vitro characterization of the
metabolites.
2
9
Weisberg and coworkers first reported that midostaurin and the epimeric 1:1 mixture of
metabolites [e1 + e2] (CGP52421), inhibited the transphosphorylation activity of recombinant
GSTꢀFLT3 in a radiometric assay (8 µM ATP; substrate: poly{GluTyr}4:1) with IC50 values
ACS Paragon Plus Environment

Page 25 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
of 528 and 643 nM respectively. Without invoking intracellular drug accumulation, this
relatively weak activity was inconsistent with their findings in murine BaF3 cells transfected
to express either ITDꢀ or D835Yꢀmutant FLT, where midostaurin abbrogated FLT3
autophosphorylation at concentrations substantially below 1 µM. The BaF3 results were
subsequently supported by studies employing FTL3ꢀdependent human leukemia cell lines,
where midostaurin potently inhibited the autophosphorylation of wildꢀtype (RS4;11 cells) and
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
57
ITD FLT3 (MV4ꢀ11 cells) with IC values of 15 and 13 nM respectively. This
50
autophosphorylation data is consistent with results from the present biochemical study, where
midostaurin potently inhibited the transphosphorylation activity of the wildꢀtype, as well as
the ITDꢀ and D835Yꢀmutant forms of FLT3 with IC50 values of 48, 26 and 14 nM
58
respectively, and was further corroborated using the InVitrogen SelectScreen™, where the
drug inhibited wildꢀtype and D835Y FLT3 with IC values of 20 and 3.6 nM respectively.
50
Here we show that the major metabolites also inhibit the kinase activity of FLT3, with
CGP62221 (3) possessing activities comparable to those of the parent drug, while e1 (4) and
e2 (5) have IC50 values in the range of 200ꢀ400 nM against the ITD and D835Y mutants, and
low micromolar activity against the wildꢀtype enzyme. However, when taking into
consideration that following chronic dosing, steadyꢀstate plasma trough levels of CGP62221
(3) are slightly higher than those of midostaurin and those of e2 (5) are about 8ꢀfold higher, it
is expected both metabolites make important contributions to the inhibition of FLT3ꢀcatalysed
phosphorylation in the leukemic cells of AML patients.
In addition to targeting FLT3, together with kinases such as those of the PKC family and
1
5,31
VEGFR2 which had supported the initial clinical development of midostaurin,
here we
show that midosaurin inhibits a large number of additional tyrosine and serineꢀthreonine
kinases, with IC50 values substantially below 1 µM. The major metabolite CGP62221 (3) is
also a multitargetedꢀkinase inhibitor, possessing a similar potency and selectivity profile to
that of midostaurin, although the 4ꢀfold greater inhibition of FER is perhaps of significance.
ACS Paragon Plus Environment

Biochemistry
Page 26 of 53
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The activities of midostaurin and CGP62221 against the mutant forms of KIT are of particular
5
relevance since in addition to being involved in SM, a high expression level of this protoꢀ
oncogene is a poor prognostic marker in AML and activating KIT mutations, which mainly
occur in the activation loop of the kinase domain (exon 17) resulting in Asp816Val and
Asn822Lys substitutions, are found in 25 ꢀ 30% of cases of coreꢀbinding factor (CBF)ꢀ
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
9,60
AML.
In contrast to CGP62221 (3), the 3ꢀhydroxylated metabolites 4 and 5 are generally less active
and have somewhat different selectivity profiles, resulting in an increased selectivity towards
VEGFR2. Since 4 is only a minor metabolite, given its kinase profile it is unlikely to play a
substantial contribution to the pharmacology of midostaurin in patients. However, because 5
accumulates to become the major circulating component, the kinase targets of this metabolite,
and PDPK1 in particular, considering that it is overꢀexpressed in AML and promotes PKCꢀ
61,62
mediated survival of leukemic blasts,
parent drug.
are likely to contribute to the pharmacology of the
Figure 5. Representations of (a) midostaurin docked into a homology model of an active
65
conformation of FLT3 (based upon the staurosporineꢀLCK cocrystal structure; pdb# 1QPJ)
and (b) epimer e2 (5) docked into PDPK1 (based upon the UCNꢀ01 ꢀ PDPK1 cocrystal
64
structure; pdb# 1OKZ). The carbon atoms of ligands and kinases are depicted in green and
ACS Paragon Plus Environment

Page 27 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
grey respectively, with nitrogens, oxygens and sulphurs shown in blue, red and yellow
respectively; Hꢀbonds are depicted as dotted black lines. The lactam moieties of both ligands
make bidentate Hꢀbond interactions with backbone amides of the hinge region of the kinase
SH1 domains, mimicking the amidine group of ATP. The selectivity of midostaurin over
staurosporine can be attributed to (i) the absence of a basic MeNHꢀgroup capable of
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
6
4
interacting with acidic residues (e.g. Glu91 in CHK1) in the riboseꢀpocket, and (ii) the bulky
benzoyl group impeding binding, except to kinases where it can make favourable
hydrophobic contacts with the hydrocarbon part of the sideꢀchains of adjacent residues: FLT3
(Arg815), JAK3 (Arg953), RET (Arg624), VEGFR2 (Arg1030) and PDPK1 (Glu209;
6
2
analogous to that with Glu291 reported between midostaurin and DYRK1A. Although 4 and
5
are generally a much less potent kinase inhibitors, 5 maintains activity against PDPK1,
which is attributable to the 3ꢀRꢀhydroxy group Hꢀbonding to Thr222, analogous to that
observed with UCNꢀ01.
Although coꢀcrystallization studies with midostaurin are impeded by the poor solubility of the
63
ligand, modelling studies based upon the only available coꢀcrystal structure with DYRK1A,
can in part explain the observed kinase selectivities (Figure 5). In general, the 3ꢀhydroxy
groups of 4 and 5 lead to reduced kinase activity due to the lack of favorable interactions to
overcome desolvation, although the potent inhibition of PDPK1 by 5 can be rationalised by an
interaction between the 3ꢀRꢀhydroxy group and the sideꢀchain hydroxyl of the Thr222 residue
6
4
(
DFGꢀ1), analogous to that observed with the hydroxystaurosporine UCNꢀ01. However,
unlike UCNꢀ01 which also potently inhibits CHK1, the benzoyl groups of midostaurin and its
metabolites make electrostatically and sterically unfavorable contact with the Glu91 residue
of this kinase.
To investigate the effects of inhibiting many kinases in cells, midostaurin was profiled against
a large panel of human cancer cell lines in the CCLE. The drug showed greatest
ACS Paragon Plus Environment

Biochemistry
Page 28 of 53
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
antiproliferative activity against the MOLM13 and MONOMAC1 AML cells which harbour
66,67
ITD and V592A FLT3 mutations respectively.
It also substantially reduced the viability of
the OCIꢀAML2 AML and the MOLPꢀ8 multiple myeloma cell lines. The effect on OCIꢀ
AML2 cells (Supplementary Figure S1) is notable since these cells carry a point mutation in
the DNMT3A gene, which encodes an R635W aminoꢀacid replacement in the DNA
methyltransferase domain, and DNMT3A is frequently mutated in AML patients and is
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
1,68ꢀ70
associated with a poor outcome.
MOLPꢀ8 cellꢀline (Supplementary Figure S2), the genotype of which shows loss of CDKN2A
a tumor suppressor gene), loss of PTEN, together with a gainꢀofꢀfunction mutation in NRAS.
Also of possible relevance is the activity against the
(
However, at present it is not known which target(s) of midostaurin underlie the
antiproliferative effects of the drug in these two cell lines.
Further insight into the cellular activity of the compounds against nonꢀFLT3ꢀdependent cells
comes from a comparison of their effects on a panel of cells derived from either human
tumors or murine hematopoietic BaF3 cells rendered growthꢀfactor independent by
transfection with constructs encoding constitutively active protein kinases (Table 3). The
inhibition of FLT3ꢀITD kinase biochemical activity translated into potent antiproliferative
activity in FLT3ꢀITD dependent BaF3, MOLM13 and MV4ꢀ11 cells, with potencies
comparable with those reported to inhibit the phosporylation of FLT3 and its signalling in a
29,71,72
variety of FLT3ꢀdependent AML cells.
As expected, potent activity leading to a high
degree of selectivity was also evident towards those cells dependent upon other kinases that
were strongly inhibited in the biochemical screen (D816V KIT, PDGFRβ and RET).
However, unlike 4 and 5 [e1+e2], both midostaurin and CGP62221 (3) also inhibited the
proliferation of wildꢀtype BaF3 cells (GI50 values 388 and 327 nM respectively), and
midostaurin in particular reduced the viability of BaF3ꢀBCRꢀABL1 (GI50 655 nM), BaF3ꢀTelꢀ
INSR (GI50 253 nM) and human gastrointestinal stromal GIST882 cells (GI50 399 nM) that
are dependent upon constitutively activated kinases not targeted by the drug (ABL1, INSR
ACS Paragon Plus Environment

Page 29 of 53
Biochemistry
1
2
3
4
5
6
7
8
9
and K642E KIT). This pattern of activity is consistent with midostaurin and CGP62221
inhibiting one or more elements of the ILꢀ3 signalling pathway in wildꢀtype BaF3 cells and
elements downstream of the oncogenic kinases that drive the proliferation of the other cell
lines.
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
An effect on the ILꢀ3 signalling pathway is potentially of particular relevance to the treatment
of AML, since ILꢀ3 regulates the production of hematopoietic cells and overexpression of the
ILꢀ3 receptor αꢀchain has been shown to provide a survival and growth advantage to leukemic
7
3
cells, and is associated with poor prognosis. However, as for the OCIꢀAML2 and the
MOLPꢀ8 cells, deconvoluting the kinase target(s) responsible for inhibiting downstream
signalling and causing the general cytotoxicity seen in the CCLE screen at concentrations of
≥4 µM is not straightforward. Several protein kinases have been reported to be elements in
multiple ILꢀ3 signal transduction pathways, including JAKꢀ, MAPKꢀ and SRCꢀfamily
74
kinases, and a number of kinases (AURKꢀA, CAMK2D, FLT3, GRK7, LRRK2, PHKG1,
PRKG2, RET, ROS, RPS6KA, TRKꢀA, TYK2) are shown in this study to be potently
inhibited by midostaurin and CGP62221 with IC50 values <300 nM, while remaining
relatively insensitive (≥10ꢀfold less active) to 4 and 5 (Table 2). In addition to FLT3 and KIT,
a number of other kinases including several of those mentioned above, have been implicated
in playing a role in various genotypes of AML, either as elements of signaling pathways in
AML cells or implicated by providing stromal support (Table 4). Several of these are
substantially inhibited by midostaurin and its metabolites at physiologically relevant
concentrations and consequently activity against these targets probably contributes to the
pharmacological effects of the drug in AML patients carrying FLT3 mutations.
Table 4: Kinases other than FLT3 or KIT that have been implicated as playing a role in AML
and their sensitivity towards midostaurin and metabolites.
1
Sensitivity
Kinase
Role
AKT / PI3K
FLT3-ITD in AML is associated with activation
Insensitive
ACS Paragon Plus Environment