Discovery of a novel series of quinoxalines as inhibitors of c-Met kinase

Article abstract of DOI:10.1016/j.bmcl.2008.11.062

A series of quinoxaline inhibitors of c-Met kinase is described. The postulated binding mode was confirmed by an X-ray crystal structure and optimisation of the series was performed on the basis of this structure. Future directions for development of the

Full text of DOI:10.1016/j.bmcl.2008.11.062

Bioorganic & Medicinal Chemistry Letters 19 (2009) 397–400
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of a novel series of quinoxalines as inhibitors of c-Met kinase
*
John Porter , Simon Lumb, Fabien Lecomte, James Reuberson, Anne Foley, Mark Calmiano, Kelly le Riche,
Helen Edwards, Jean Delgado, Richard J. Franklin, Jose M. Gascon-Simorte, Alison Maloney,
Christoph Meier, Mark Batchelor
UCB Celltech, 216 Bath Road, Slough SL1 3WE, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of quinoxaline inhibitors of c-Met kinase is described. The postulated binding mode was con-
firmed by an X-ray crystal structure and optimisation of the series was performed on the basis of this
structure. Future directions for development of the series are discussed together with the identification
of a novel quinoline scaffold.
Received 24 October 2008
Revised 17 November 2008
Accepted 18 November 2008
Available online 21 November 2008
Ó 2008 Elsevier Ltd. All rights reserved.
Keywords:
Kinase Inhibitor
c-Met
Quinoxaline
Members of the receptor tyrosine kinase (RTK) family are
attractive targets for cancer therapy as inhibition can disrupt sig-
naling pathways that mediate tumour formation and growth. c-
Met kinase is a member of this family that, together with its ligand,
hepatocyte growth factor (HGF), is important for normal mamma-
lian development. However, c-Met has been shown to be deregu-
lated and associated with high tumour grade and poor prognosis
in a number of human cancers.1 c-Met can become activated by a
variety of mechanisms, including gene amplification and mutation
inducing motility, invasiveness and tumourgenicity into the trans-
formed cells.2 Activation leads to receptor dimerisation and
recruitment of several SH2 domain containing signal transducers
that activate a number of pathways including the Raf-Mek-Erk
and PI3k-Akt cascades. Targeting the ATP binding site of c-Met is
a popular strategy for inhibition of the kinase, with a number of
drug candidates reaching the clinical testing phase.3 We now wish
to report our efforts in this area.
A HTS campaign identified the quinoxaline 1a as a promising
starting point. This low molecular weight, relatively soluble, ATP-
competitive compound had an IC50 of 1.3 lM in the c-Met
biochemical assay.4 Cross-screening against a panel of over 200 ki-
nases showed that 1a was highly selective for c-Met.5 Docking 1a
into the published crystal structure of c-Met (pdb:1R0P)6 sug-
gested that the N-4 atom of the quinoxaline ring formed a H-bond
to Met1160 in the hinge region of the kinase, whilst the trifluoro-
methyl and phenyl groups occupied hydrophobic pockets, and the
piperazine group was directed towards solvent. With this binding
mode in mind we embarked on an exploration of the SAR.
The quinoxalines were prepared as shown in Scheme 1.
Commercially available diamine 2 was cyclised to give a 1:1
mixture of regioisomeric quinoxalinones 3 and 4 from which the
desired isomer 3 was isolated by trituration in 45% yield. Chlorina-
tion at the 2-position allowed the substitution of a range of amines
and was followed by Suzuki coupling to give the desired products.
Variation of substitution at the 6-position was achieved by appro-
priate choice of starting material. A related series of quinolines
were prepared as shown in Scheme 2. Bromination, reduction and
a Sandmeyer reaction7 transformed the commercially available
quinoline 7 to the quinolinol 10. Ether formation under mildly basic
conditions was followed by a Buchwald–Hartwig reaction8 to give
the quinoline 12.
N
1
X
N
7
8
5
2
6
4
3
CF3
N
1a X=N
Replacement of the N-methyl piperazine with other groups
showed that it is important to have a basic, preferably tertiary,
nitrogen substituent at the quinoxaline 2-position, although the
O-linked piperidine, 18, retained some potency, Table 1. We postu-
late that the presence of an exocyclic heteroatom counteracts the
electron withdrawing properties of the nitrogen atom at the 1-po-
1b X=CH
* Corresponding author. Tel.: +44 1753 534655.
E-mail address: john.porter@ucb-group.com (J. Porter).
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.11.062

398
J. Porter et al. / Bioorg. Med. Chem. Lett. 19 (2009) 397–400
H
Br
Br
2
CF3
N
O
H
N
NH2
NH2
i
O
+
N
CF3
CF3
N
3
Br
4
ii
Br
N
Br
N
N
N
Cl
N
N
N
iv
iii
N
N
N
CF3
CF3
CF3
1
5
6
Scheme 1. Preparation of the quinoxalines. Reagents and conditions: (i) HCOCO2Et, EtOH, reflux; (ii) POCl3; (iii) N-methylpiperazine, THF, reflux; (iv) PhB(OH)2, Pd(PPh3)4,
dioxane, Na2CO3, reflux.
Realising that there was scope for optimisation of both potency
and solubility by appropriate substitution at the 2-position, we
now turned our attention to the 8-position where a study of the
published X-ray crystal structures led us to believe that there
was the greatest scope for enhancement of potency. A wide range
of substituted phenyl analogues were prepared, a representative
selection are shown in Table 2. Although bulky functionality was
not accommodated, for example, 26, most smaller substituents
did not have any significant effects on potency. However, indoli-
nones 33 and 34 and the benzimidazoline 32 did show promising
activity
NO2
NO2
8
NH2
9
Br
Br
i
ii
iii
iv
N
N
N
N
7
NO2
NO2
OH
Br
vi
v
O
O
N
Br
N
10
In an attempt to rationalise these observations, we solved the
X-ray crystal structure of the c-Met-28 complex9, Figure 1a.
Although the binding mode was revealed to be as we had predicted,
N
N
12
11
the complex co-crystallised with a molecule of
c-butyrolactone
Scheme 2. Preparation of the quinolines. Reagents and conditions: (i) NBS, AcOH,
reflux; (ii) Fe, FeCl3, EtOH, AcOH, reflux; (iii) NaNO2, AcOH, H2SO4, water, 0 °C; (iv)
10% aq H2SO4, reflux; (v) 3-nitrobenzylbromide, Cs2CO3, THF, 25 °C; (vi)
N-methylpiperazine, Pd2(dba)3, BINAP, NaOtBu, toluene, reflux.
Table 2
SAR of substituents at the quinoxaline 8-position
N
R
Table 1
N
N
N
SAR of substituents in the 2- and 6-positions of 8-phenylquinoxaline
CF3
N
N
R1
Compound
R
c-Met IC50 (lM)
R2
24
25
26
27
28
29
30
3,4-Dimethoxyphenyl
2,3-Dichlorophenyl
4-Benzyloxyphenyl
3-Carbamoylphenyl
3-Hydroxyphenyl
4-Hydroxyphenyl
3-Nitrophenyl
1.5
1.8
0% @ 100
5.0
lM
Compound
R1
R2
c-Met IC50
1.3
(lM)
0.9
0.8
12.9
1a
13
14
15
16
17
18
19
20
21
22
23
N-Methyl piperazinyl
H
CF3
CF3
CF3
CF3
CF3
CF3
CF3
H
F
Cl
Me
CN
0% @ 100
20.1
5.3
l
M
M
Amino
Dimethylamino
Acetamido
N-Acetyl piperazinyl
H
N
0% @ 100
1.3
l
31
32
33
34
2.0
N-Methyl piperidin-4-oxy
N-Methyl piperazinyl
N-Methyl piperazinyl
N-Methyl piperazinyl
N-Methyl piperazinyl
N-Methyl piperazinyl
19.6
28.3
3.9
2.8
5.0
H
N
0.39
0.17
0.33
O
N
H
60.2
O
sition, strengthening the putative H-bond between N-4 and the
hinge residue Met1160. This hypothesis was strengthened by the
observation that the quinoline 1b (IC50 0.65 lM) was nearly twice
as potent as 1a. The trifluoromethyl group at the 6-position ap-
pears to be preferred, presumably optimally occupying a hydro-
phobic pocket that can also accommodate fluoro, 20, chloro, 21,
or methyl, 22, groups, though not for example, a nitrile group, 23.
N
H
O
N

J. Porter et al. / Bioorg. Med. Chem. Lett. 19 (2009) 397–400
399
Figure 1. (a) X-ray crystal structure of the c-Met-28-c-butyrolactone complex showing key interactions. The protein residues are in green with the positions of Met1229-
Tyr1230 from PDB:1ROP shown in pink for comparison. (b) Overlay of a model of 38 (yellow) with the structure of 28 (grey) showing postulated interaction of the nitro group
with Asp1222 and the phenyl ring with Tyr 1230.
from the buffer solution, which appeared to cause a conformational
change to Met1229 and Tyr1230 (compared to pdb:1ROP6). This
conformational change is similar to that observed in the crystal
structure of SU11274 (pdb:2RFS) and has been suggested to be
responsible for the difference in profiles of c-Met inhibitors.10 The
Table 3
SAR of homologated substituents at the quinoxaline 8-position
N
R
N
N
N
c
-butyrolactone also appeared to be forming a H-bond to the back-
bone NH of Asp1222. Consequently, we attempted to both incorpo-
rate this feature and overlap the -cloud of the displaced Tyr1230 in
CF3
p
Compound
Substituent
c-Met IC50 (lM)
our molecules by moving the aromatic 8-substituent further from
the quinoxaline scaffold.
Incorporation of N- or O-linked substituents onto the bromide
6, Scheme 1 using Buchwald–Hartwig chemistry11 allowed us to
prepare the desired compounds, some of which are outlined in Ta-
ble 3. A two atom linking chain appears to be optimum, but the
most significant observation is the effect of the 3-nitro substituent,
38, which produces a 30-fold increase in potency. We believe that
35
36
1.9
1.1
N
H
H
N
the electron withdrawing effect of the nitro group enhances
p-
37
38
7.1
N
H
stacking with the phenyl ring of Tyr1230 and forms a H-bond with
Asp1222, Figure 1b. Compound 38 retained the selectivity profile
of the parent 1a in the kinase screening panel.5 Interestingly, the
analogous 30, Table 2, was only weakly active. Replacing the ami-
nomethyl linker with sulfonamide, for example 42, improved po-
tency still further. As the nitro group can confer mutagenic and
carcinogenic properties12 we attempted to identify isosteric groups
to overcome these potential issues. The benzoxadiazole group has
been reported as an isosteric replacement for the nitro group in a
PDE4D inhibitor13 and the calcium antagonist isradipine.14 Incor-
poration of this moiety into our scaffold gave 44 whose equipoten-
cy with 38 suggests a similar mode of binding. However, the
regioisomeric 45 was 10-fold less active than the corresponding
4-nitro analogue 40.
H
N
0.035
NO2
H
N
39
40
41
0.54
0.055
0.32
NO2
NO2
H
N
O
NO2
NO2
Attempts to identify robust routes to the appropriately substi-
tuted 3,5,7-quinolines have proved unsuccessful; a survey of the
literature suggests this is a relatively unexplored substitution pat-
tern. However, we have prepared the analogous 3,5-disubstituted
quinoline 12, scheme 2, and were pleased to observe an improved
H
N
42
43
44
0.017
0.73
S
O
O
potency in the biochemical assay (IC50 0.057 lM) suggesting that
H
N
3,5,7-trisubstituted quinolines could form a series of c-Met inhib-
itors with a similar binding mode to the quinoxalines.
In conclusion we have identified novel quinoxaline and quino-
line inhibitors of c-Met kinase and have rationalised the SAR by
reference to an X-ray crystal structure.
CONH2
N
H
N
0.031
N
O
References and notes
N
N
45
H
N
0.38
O
1. Birchmeier, C.; Birchmeier, W.; Gherardi, E.; van de Woude, G. F. Nat. Rev. Mol.
Cell Biol. 2003, 4, 915.

400
J. Porter et al. / Bioorg. Med. Chem. Lett. 19 (2009) 397–400
remaining @10 M) against nine kinases including Flt family members and
2. (a) Christensen, J. G.; Burrows, J.; Salgia, R. Cancer Lett. 2005, 225, 1; (b) Ma, P.
l
C.; Maulik, G.; Christensen, J.; Salgia, R. Cancer Metastasis Rev. 2003, 22, 309.
3. (a) Cui, J. J. Expert Opin. Ther. Patents 2007, 17, 1035; (b) Comoglio, S. G.;
Giordano, S.; Trusolino, L. Nat. Rev. Drug Disc. 2008, 7, 504.
4. IC50 values for inhibitors of c-Met were determined using an IMAP time
resolved fluorescence resonance energy transfer (TR-FRET) assay. 50 nM 6 His-
tagged recombinant human c-Met residues 974-end (Millipore) was incubated
in 20 mM Tris, 10 mM MgCl2, 2.5 mM MnCl2, 0.01% Tween 20 and 2 mM DTT
KDR. Similarly, compound 38 showed weak activity against 15 kinases,
including Flts, Ron, Abl, Axl and TrkA.
6. Schiering, N.; Knapp, S.; Marconi, M.; Flocco, M. M.; Cui, J.; Perego, R.; Rusconi,
L.; Cristiani, C. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 12654.
7. Hartley, C. S.; Elliott, E. L.; Moore, J. S. J. Am. Chem. Soc. 2007, 129, 4512.
8. Wolfe, J. P.; Buchwald, S. L. J. Org. Chem. 2000, 65, 1144.
9. Crystallisation and solution of structure performed by Proteros Biostructures
GmbH, Am Klopferspitz 19, D-82152, Martinsried, Germany, http://
www.proteros.com. Crystallographic data for the structure in this Letter have
been deposited with the PDB (pdb code 3F66).
10. Bellon, S. F.; Kaplan-Lefko, P.; Yang, Y.; Zhang, Y.; Moriguchi, J.; Rex, K.;
Johnson, C. W.; Rose, P. E.; Long, A. M.; O’Connor, A. B.; Gu, Y.; Coxon, A.; Kim,
T.-S.; Tasker, A.; Burgess, T. L.; Dussault, I. J. Biol. Chem. 2008, 283, 2675.
11. (a) Wolfe, J. P.; Wagaw, S.; Marcoux, J.-F.; Buchwald, S. L. Acc. Chem. Res. 1998,
31, 805; (b) Shen, Q.-L.; Hartwig, J. F. J. Am. Chem. Soc. 2007, 129, 7734; (c)
Surry, D. S.; Buchwald, S. L. J. Am. Chem. Soc. 2007, 129, 10354.
12. (a) Purohit, V.; Basu, A. K. Chem. Res. Toxicol. 2000, 13, 673; (b) Ashby, J.;
Tennant, R. W. Mutat. Res. 1991, 257, 229.
13. Hersberger, R.; Dawson, J.; Meuller, T. Bioorg. Med. Chem. Lett. 2002, 12, 233.
14. Man I’nt Veld, A. J. Am. J. Hypertension 1991, 4, 96S.
with 5
25 l for 60 min at ambient temperature. Inhibitors were tested in duplicate at
10 concentrations starting from 20 M at a final concentration of 1% DMSO.
The reaction was stopped by addition of 50 l of IMAP stop solution containing
lM ATP and 200 nM 5FAM-KKKSPGEYVNIGFG-NH2 in a total volume of
l
l
l
60%Buffer A:40%Buffer B and a 1in 400 dilution of beads and Terbium reagent.
Plates were read after an overnight incubation at +4 °C on an Analyst HT reader.
Reported IC50’s are from a minimum of two experiments (n = 2). Data analysis
was carried out using a four parameter curve fit. The standard errors of the
mean were calculated and expressed as a percentage of the mean IC50. The
average for this value was 12%.
5. Millipore Bioscience Division, Millipore UK Ltd, Gemini Crescent, Dundee
Technology Park, Dundee, DD2 1SW, United Kingdom. http://
www.millipore.com. Compound 1a showed weak activity (<30% activity