Aromatic interactions with phenylalanine 691 and cysteine 828: A concept for FMS-like tyrosine kinase-3 inhibition. Application to the discovery of a new class of potential antileukemia agents

Article abstract of DOI:10.1021/jm060368s

FLT3 kinase inhibitors are currently under investigation as a new treatment for acute myeloid leukemia. We report here a molecular concept invoking interactions between an aromatic ring and the side chains of Phe691 and Cys828, two residues of the ATP poc

Full text of DOI:10.1021/jm060368s

J. Med. Chem. 2006, 49, 4451-4454
4451
Aromatic Interactions with Phenylalanine 691
Chart 1
and Cysteine 828: A Concept for FMS-like
Tyrosine Kinase-3 Inhibition. Application to the
Discovery of a New Class of Potential
Antileukemia Agents
Pascal Furet,* Guido Bold, Thomas Meyer,
Johannes Roesel, and Vito Guagnano*
NoVartis Pharma AG, NoVartis Institutes for Biomedical Research,
CH-4002 Basel, Switzerland
ReceiVed March 29, 2006
common to these molecules was conferring them high inhibitory
activity against the FLT3 kinase.
Abstract: FLT3 kinase inhibitors are currently under investigation as
a new treatment for acute myeloid leukemia. We report here a molecular
concept invoking interactions between an aromatic ring and the side
chains of Phe691 and Cys828, two residues of the ATP pocket, to obtain
potent and specific inhibitors of this kinase. The hypothesis has been
validated by the successful design of a new inhibitor prototype showing
promising antiproliferative activity in cellular assays.
The majority of known kinase inhibitors possess chemical
moieties interacting with three crucial parts of the ATP binding
site that we term the adenine region and hydrophobic regions I
16
and II (Figure 1).
According to our binding models (Figure 2), 1-3 conform
to this rule in their interactions with the FLT3 kinase. The central
1
7
portions of the inhibitors occupy the adenine region where
they form a hydrophobic sandwich with residues Ala642 and
Leu818. In the same region, they are engaged in the typical
hydrogen bonds with the backbone of the hinge segment,
Inhibition of the tyrosine kinase activity of the FLT3 (FMS-
like tyrosine kinase-3) receptor is a therapeutic concept of
current interest in antileukemia drug research.1-4 This new
approach to the treatment of acute myeloid leukemia (AML)
has emerged following evidence that constitutive activation of
the FLT3 receptor plays an important role in the development
of this aggressive hematological malignancy for which no
effective cure exists at the moment.5
18
precisely at residues Cys694 and Glu692. The inhibitors also
19
contain a moiety occupying hydrophobic region II, the
lipophilic slot opened to solvent constituted by residues Leu616
and Gly697 in the FLT3 kinase.
-7
However, what most attracted our attention is the way,
according to the models, 1-3 interact with hydrophobic region
I, the back pocket of the ATP site formed by residues Met665,
Val675, Leu689, Phe691, and Cys828. The three inhibitors have
in common a phenyl ring (shown in green in Figure 2) that,
based on interatomic distance criteria, can establish a S-H/π
interaction with the side chain of Cys828 and at the same time
The first generation of FLT3 kinase inhibitors was obtained
by screening compounds well established as inhibitors of other
8
protein kinases. Several compounds of this type, exemplified
by the staurosporine derivative PKC412, have entered clinical
9
trials in AML patients. Medicinal chemistry efforts in the field
are now focusing on identifying follow-up clinical candidates
with improved properties, in particular with regard to specificity.
In this respect, to support our own efforts in this direction, we
initiated molecular modeling studies aimed at understanding the
structural determinants of inhibition of the FLT3 kinase by small
molecules interacting with its ATP binding site. We present here
a concept for FLT3 kinase inhibition resulting from this work.
Its validation by a first successful application to the design of
the prototype of a new class of potent and specific FLT3
inhibitors is also reported.
make an aromatic-aromatic edge to face C-H/π interaction
20-22
with the side chain of Phe691.
The residues corresponding
to positions Phe691 and Cys828 in the sequence of the FLT3
23
kinase domain are not conserved in the protein kinase family.
These residues are located at the entrance of the hydrophobic
region I pocket. For this reason the residue corresponding to
Phe691 has been named the gate keeper in the literature, and
To gain insight into the structural determinants of FLT3
kinase inhibition, we constructed by homology a model of the
three-dimensional atomic structure of the kinase domain of the
FLT3 receptor and used it to perform docking studies of
compounds reported as inhibitors of this kinase in the litera-
ture.1 In particular, we docked 1 (SU5416), 2 (AG1295), and
0,11
3
(D64406) into the ATP binding site of the model. 1-3 were
prominent FLT3 kinase inhibitors in the literature at the time
we initiated our work.1
2-15
The compounds were reported to
inhibit the autophosphorylation of the FLT3 receptor in the
submicromolar range in cellular assays. As described below,
the resulting binding modes of 1-3 in the ATP pocket of the
homology model suggested that a specific structural feature
*
To whom correspondence should be addressed. For P.F.: phone, 41
6
1 696 79 90; fax, 41 61 696 15 67; e-mail, pascal.furet@novartis.com.
For V.G.: phone, 41 61 696 24 60; fax, 41 61 696 62 46; e-mail,
vito.guagnano@novartis.com.
Figure 1. Schematic representation of the ATP binding site of protein
kinases (active conformation).
1
0.1021/jm060368s CCC: $33.50 © 2006 American Chemical Society
Published on Web 06/22/2006

4
452 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 15
Letters
Scheme 1a
a
(i) (a) (CH3)3SiBr, DMSO, CH3CN, 0 °C, room temp; (b) 5, reflux,
34%; (ii) NaOH, DMF, 2 h, 75 °C, 92%; (iii) CSCl2, CHCl3, saturated
NaHCO3, 1 h, room temp, 98%; (iv) 2 M NH3 in MeOH, 1 h, 60 °C, 99%.
Figure 2. Binding models of reference compounds 1-3 (A-C) and
designed prototype 4 (D) in the ATP site of the FLT3 kinase. Hydrogen
bonds to the hinge segment are indicated by magenta lines. The phenyl
ring interacting with Phe691 and Cys828 is shown in green.
bromide and dimethyl sulfoxide in acetonitrile. Thiazole 4 was
prepared according to this procedure and obtained in 34%
overall yield (Scheme 1A). The corresponding thiourea 5 was
synthesized in three steps starting from p-aminophenol, as shown
in Scheme 1B. O-Alkylation of p-aminophenol with 1-(2-
chloroethyl)pyrrolidine hydrochloride is performed in DMF and
in the presence of finely powdered NaOH. Reaction of the
resulting aniline with thiophosgene affords the corresponding
isothiocyanate, which is then converted into the desired thiourea
upon treatment with ammonia in methanol (Scheme 1B). It is
noteworthy that the synthesis of thiazole derivatives devoid of
a basic moiety (the cyclopentylamino group for thiazole 4)
requires a further step, namely, the addition of diisopropylethyl-
the role it plays in determining the selectivity of kinase inhibitors
is well established.24,25 In the case of the FLT3 kinase, the
concomitant presence at these positions of two amino acids able
to form specific and particularly stable interactions with aromatic
rings is remarkable. On the basis of the above analysis, we
postulated that a sufficient condition for a molecule to potently
inhibit the FLT3 kinase was to possess an aryl ring able to make
a S-H/π interaction with Cys828 and an aromatic-aromatic
edge to face interaction with Phe691 and this in addition to
possessing two chemical moieties conforming to the more
generic hydrophobic region II and adenine region pharmaco-
phores.
To test this hypothesis, we designed by interactive molecular
modeling a prototype compound (4 shown in Figure 2D) with
the three postulated important structural features. In this
molecule, the 2-aminothiazole moiety was chosen to interact
with the adenine region. Once this moiety was anchored in the
binding site by formation of hydrogen bonds with the backbone
of Cys694, we realized that the 5-position of the thiazole ring
was very well suited to attach a phenyl ring targeting Cys828
and Phe691 to form the desired aromatic interactions. Another
phenyl ring attached to the 2-amino position was adequate to
fill the hydrophobic region II slot by formation of a hydrophobic
sandwich with Leu616 and Gly697. The design was completed
by attachment of a solubilizing cyclopentylamino side chain
extending into the solvent from the para position of the latter
phenyl ring.
2
6
amine (three-step, one-pot protocol). As observed by HPLC
and mass spectrometry, it appears that the tertiary amine
facilitates the dehydration of a reaction intermediate to provide
the final product.
The designed prototype 4 was tested in biochemical assays
measuring its capacity to inhibit the catalytic activity of the
FLT3 receptor kinase domain and that of various other
recombinant kinases. The resulting experimental data are
reported in Table 1. As can be seen, 4 turned out to inhibit the
FLT3 kinase with an IC50 of 50 nM while most of the other
kinases were not significantly inhibited at a concentration as
high as 10 µM. Thus, to our satisfaction, potent and selective
inhibition of the FLT3 kinase was achieved, thereby validating
the design concept.
Two other kinases in Table 1, c-Kit and KDR, were inhibited
in the submicromolar range by 4, although less potently than
FLT3. Interestingly, both kinases also present a cysteine residue
at the position corresponding to Cys828 in FLT3. Along the
same line, CDK1, the only kinase in Table 1 sharing a
phenylalanine gate keeper residue with FLT3, is significantly
inhibited by 4 (IC50 ) 2.1 µM). These observations give support
to our hypothesis that aromatic interactions with Cys828 and
Phe691 provide significant binding affinity for the ATP site of
the FLT3 kinase. From a comparison of the data reported in
Table 1, it is in principle difficult to make a precise quantitative
assessment of the contribution of these interactions to the overall
5
-Aryl-N-aryl-2-aminothiazoles can be prepared by reacting
thioureas with 2-aryl-2-bromoacetaldehydes. A drawback as-
sociated with such a synthetic approach is the purification and
handling of the R-bromoaldehydes that appear prone to undergo
degradation. Thus, we have developed an efficient, two-step,
one-pot protocol that involves the in situ preparation of the
2
-aryl-2-bromoacetaldehydes followed by cyclocondensation
with the thiourea (Scheme 1A). The bromination is ac-
complished by treatment of the aldehyde with trimethylsilyl-

Letters
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 15 4453
Table 1. IC50 Values of Compound 4 in Enzymatic Assays and Nature
of Gate Keeper Residues
order of magnitude in potency compared to the inhibition
observed with the BaF3-ITD cells having the wild-type gate
keeper residue (IC50 ) 0.24 µM) clearly supports the existence
of the postulated aromatic-aromatic interaction with Phe691.
In conclusion, modeling of the binding modes of reported
FLT3 kinase inhibitors led us to propose that aromatic interac-
tions with residues Phe681 and Cys828 of the ATP pocket can
be exploited to obtain potent and specific inhibitors of this
kinase. The discovery of 4 on the basis of this concept is very
encouraging for its further use in the search for new classes of
FLT3 kinase inhibitors as potential antileukemia agents.
a
gate keeper 1b
gate keeper 2c
kinase
IC50, µM
FLT3
c-Src
c-Abl
c-Kit
c-Met
KDR
Tie-2
EGF-R
IGF1-R
EphB4
JAK2
CDK1
PKA
0.050
2.8
1.2
Phe
Thr
Thr
Thr
Leu
Val
Ile
Thr
Met
Thr
Met
Phe
Met
Met
Val
Met
Thr
Leu
Val
Cys
Ala
Ala
Cys
Ala
Cys
Ala
Thr
Gly
Ser
Gly
Ala
Thr
Thr
Ala
Gly
Gly
Ala
Ser
0.26
>10
0.91
8.0
>10
>10
>10
>10
2.1
>10
>10
>10
>10
>10
>10
>10
Acknowledgment. We thank Maria D’Addio for technical
assistance.
PKB
Supporting Information Available: Experimental procedures
and analytical data for all intermediate and final compounds and
description of biochemical and cellular assays. This material is
available free of charge via the Internet at http://pubs.acs.org.
FGFR-1
Ins-R
c-Raf-1
Axl
Ret
References
a
All IC50 values represent averages of at least three experimental
(
(
(
1) Sternberg, D. W.; Licht, J. D. Therapeutic intervention in leukemias
that express the activated fms-like tyrosine kinase 3 (FLT3):
opportunities and challenges. Curr. Opin. Hematol. 2005, 12, 7-13.
2) Advani, A. S. FLT3 and acute myelogenous leukemia: biology,
clinical significance and therapeutic applications. Curr. Pharm. Des.
determinations. The reference compound PKC412 has an IC50 of 0.50 µM
b
in the FLT3 inhibition assay used in this study. Residue corresponding to
c
Phe691 in the FLT3 kinase. Residue corresponding to Cys828 in the FLT3
kinase.
2005, 11, 3449-3457.
3) Heinrich, M. C. Targeting FLT3 kinase in acute myelogenous
leukemia: progress, perils and prospects. Mini-ReV. Med. Chem.
2004, 4, 255-271.
binding affinity because 4 interacts with other parts of the ATP
site (the adenine region and hydrophobic region II) where amino
2
7
acid differences also exist within the protein kinase family.
(4) Levis, M.; Small, D. FLT3: ITDoes matter in leukemia. Leukemia
2
003, 17, 1738-1752.
However, some of the kinases differ, in terms of the residues
in close contact with the inhibitor, only by the nature of one of
the gate keeper residues. Comparing the data in such cases can
provide a rough estimate of these contributions. Thus, all the
residues in close contact with 4 but corresponding to Cys828
in FLT3 are identical in c-Kit and c-Src. This residue is also a
cysteine in c-Kit, while it is an alanine in c-Src. Comparison of
the inhibition data for these two kinases, assuming that the
difference in activity is mainly due to the cysteine-alanine
change, suggests that the proposed S-H/π interaction contrib-
utes approximately 1 order of magnitude to potency. From a
similar comparison for FLT3 and c-Kit, which only differ by
the nature of the gate keeper residue Phe691, it can be concluded
that the modeled aromatic-aromatic edge to face interaction
provides a 5-fold improvement in potency compared to the
corresponding interaction between a threonine side chain and a
phenyl ring.
Encouraged by the potent inhibition of the FLT3 kinase
displayed by 4 at the enzymatic level, the compound was tested
in cellular assays measuring its ability to inhibit the proliferation
of cells whose growth is driven by a constitutively activated
FLT3 receptor. These included BaF3 cell lines transfected with
FLT3 constructs bearing two forms of activating mutations, the
ITD (internal tandem duplication) and D835/Y mutations, as
well as the MV4:11 cell line derived from AML patients
carrying the ITD mutation. Consistent with the high potency
observed in the enzymatic assay, 4 was able to block, in a dose-
dependent manner, the proliferation of these cell lines at low
concentrations. This is testified by IC50 values of 0.24, 0.76,
(
5) Gilliland, D. G.; Griffin, J. D. The roles of FLT3 in hematopoiesis
and leukemia. Blood 2002, 100, 1532-1542.
(6) Stirewalt, D. L.; Radich, J. P. The role of FLT3 in haematopoietic
malignancies. Nat. ReV. Cancer 2003, 3, 650-665.
(
7) Nakao, M.; Yokota, S.; Iwai, T.; Kaneko, H.; Horiike, S.; Kashima,
K.; Sonoda, Y.; Fujimoto, T.; Misawa, S. Internal tandem duplication
of the FLT3 gene found in acute myeloid leukemia. Leukemia 1996,
10, 1911-1918.
(
8) Levis, M.; Small, D. Novel FLT3 tyrosine kinase inhibitors. Expert
Opin. InVest. Drugs 2003, 12, 1951-1962.
9) Weisberg, E.; Boulton, C.; Kelly, L. M.; Manley, P.; Fabbro, D.;
Meyer, T.; Gilliland, D. G.; Griffin, J. D. Inhibition of mutant FLT3
receptors in leukemia cells by the small molecule tyrosine kinase
inhibitor PKC412. Cancer Cell 2002, 1, 433-443.
(
(
10) At the time we initiated this work, no X-ray crystal structure of the
FLT3 kinase was available. We constructed a homology model of
the FLT3 kinase using the coordinates of the Lck kinase in complex
with AMP-PNP (PDB code 1QPC). This template was chosen
because it was the highest resolution structure of a tyrosine kinase
adopting an active conformation (conformational state appropriate
for docking the type of inhibitor discussed here) available at that
time. In the meantime, the crystal structure of the FLT3 kinase in an
autoinhibited conformation was reported.11 In this structure, the ATP
binding site is obstructed by residue Phe830 of the activation loop
(“DFG out” conformation), which in principle precludes its use for
docking the inhibitors of our study. However, we could derive a
model of the ATP binding site corresponding to an active conforma-
tion from this structure by changing the conformation of the DGF
motif to a “DFG in” conformation. Docking the inhibitors using this
model gives the same results as with the homology model. Details
of protein model construction and inhibitor docking are given in the
Supporting Information.
(
11) Griffith, J.; Black, J.; Faerman, C.; Swenson, L.; Wynn, M.; Lu, F.;
Lippke, J.; Saxena, K. The structural basis for autoinhibition of FLT3
by the juxtamembrane domain. Mol. Cell 2004, 13, 169-178.
12) Yee, K. W. H.; O’Farrell, A. M.; Smolich, B. D.; Cherrington, J.
M.; McMahon, G.; Wait, C. L.; McGreevey, L. S.; Griffith, D. J.;
Heinrich, M. C. SU5416 and SU5614 inhibit kinase activity of wild-
type and mutant FLT3 receptor tyrosine kinase. Blood 2002, 100,
(
and 0.052 µM obtained respectively in the BaF3-ITD, BaF3-
D835/Y, and MV4:11 assays.2
8,29
Hence, with 4 we had an entry
2
941-2949.
in a new class of FLT3 kinase inhibitors active at the cellular
level. To gain additional evidence that 4 is a specific inhibitor
of the FLT3 kinase interacting with its gate keeper residue, the
activity of the compound in inhibiting the proliferation of BaF3
cells expressing FLT3-ITD with Phe691 mutated to isoleucine
was measured. An IC50 of 4.1 µM was obtained. The loss of 1
(
13) Levis, M.; Tse, K.-F.; Smith, B. D.; Garrett, E.; Small, D. A FLT3
tyrosine kinase inhibitor is selectively cytotoxic to acute myeloid
leukemia blasts harboring FLT3 internal tandem duplication muta-
tions. Blood 2001, 98, 885-887.
14) Tse, K. F.; Novelli, E.; Civin, C. I.; Bohmer, F. D.; Small, D.
Inhibition of FLT3 mediated transformation by use of a tyrosine
kinase inhibitor. Leukemia 2001, 15, 1001-1010.
(

4
454 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 15
Letters
(
15) Teller, S.; Kraemer, D.; Boehmer, S.-A.; Tse, K. F.; Small, D.;
Mahboobi, S.; Wallrapp, C.; Beckers, T.; Kratz-Albers, K.; Schwae-
ble, J.; Serve, H.; Boehmer, F.-D. Bis(1H-2-indolyl)-1-methanones
as inhibitors of the hematopoietic tyrosine kinase Flt3. Leukemia
(22) Duan, G.; Smith, V. H., Jr.; Weaver, D. F. Characterization of
aromatic-thiol π-type hydrogen bonding and phenylalanine-cysteine
side chain interactions through ab initio calculations and protein
database analyses. Mol. Phys. 2001, 99, 1689-1699.
(23) Hanks, S. K.; Quinn, A.-M. Protein kinase catalytic domain sequence
database: identification of conserved features of primary structure
and classification of family members. Methods Enzymol. 1991, 200,
38-62.
2002, 16, 1528-1534.
(16) Traxler, P.; Furet, P. Strategies towards the design of novel and
selective protein tyrosine kinase inhibitors. Pharmacol. Ther. 1999,
82, 195-206.
(17) The lactam ring in 1, the pyrazine ring in 2, and the bis-pyrrolyl
methanone moiety in 3.
(24) Vieth, M.; Higgs, R. E.; Robertson, D. H.; Shapiro, M.; Gragg, E.
A.; Hemmerle, H. Kinomicssstructural biology and chemogenomics
of kinase inhibitors and targets. Biochim. Biophys. Acta 2004, 1697,
243-257.
(
18) The three inhibitors accept a hydrogen bond from the backbone NH
group of Cys694. 1 and 3 donate a hydrogen bond to the backbone
carbonyl group of Glu692. 3 gives in addition a hydrogen bond to
the backbone carbonyl group of Cys694.
(25) Cherry, M.; Williams, D. H. Recent kinase and kinase inhibitor X-ray
structures: mechanisms of inhibition and selectivity insights. Curr.
Med. Chem. 2004, 11, 663-673.
(19) The pyrrole ring in 1, the phenyl ring fused to the pyrazine ring in
2, and the phenol ring in 3.
(26) Manuscript in preparation.
(
20) For a detailed description of these two types of intermolecular
interactions see ref 21 and references therein. By S-H/π interaction,
it is meant an interaction between the thiol group of Cys828 and a
phenyl ring of the inhibitor in which the side chain sulfur atom is
positioned 3.5-4.0 Å below the plane of the phenyl ring with the
thiol proton pointing towards its center. Such an interaction, also
termed aromatic-thiol π hydrogen bond, has been calculated to
provide 2.6 kcal/mol of stabilization energy (gas phase) in ref 22.
Aromatic-aromatic edge to face interactions are more common and
are very well reviewed in ref 21.
(27) For instance, 4 is expected to have a different interaction energy in
the hydrophobic region II for CDK1 compared to FLT3 because
CDK1 has a deletion in this region; the residue corresponding to
Gly697 does not exist in CDK1.
(28) The reference compound PKC412 has IC50 values of 0.040, 0.006,
and 0.024 µM in the BaF3-ITD, BaF3-D835/Y, and MV4:11 assays,
respectively.
(29) As control, 4 was also tested on the wild-type BaF3 cell line. No
significant inhibition of the proliferation of these cells, whose growth
is independent of FLT3 signaling, was obtained at concentrations
up to 3 µM.
(
21) Meyer, E. A.; Castellano, R. K.; Diederich, F. Interactions with
aromatic rings in chemical and biological recognition. Angew. Chem.,
Int. Ed. 2003, 42, 1210-1250.
JM060368S