Design of novel quinazoline derivatives and related analogues as potent and selective ALK5 inhibitors

Article abstract of DOI:10.1016/j.bmcl.2009.02.087

Starting from quinazoline 3a, we designed potent and selective ALK5 inhibitors over p38MAP kinase from a rational drug design approach based on co-crystal structures in the human ALK5 kinase domain. The quinazoline 3d exhibited also in vivo activity in an acute rat model of DMN-induced liver fibrosis when administered orally at 5 mg/kg (bid).

Full text of DOI:10.1016/j.bmcl.2009.02.087

Bioorganic & Medicinal Chemistry Letters 19 (2009) 2277–2281
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design of novel quinazoline derivatives and related analogues as potent
and selective ALK5 inhibitors
a
a
b
a
a
a
a
F. Gellibert a, , M.-H. Fouchet , V.-L. Nguyen , R. Wang , G. Krysa , A.-C. de Gouville , S. Huet , N. Dodic
*
a GlaxoSmithKline, 25-27 Avenue du Québec, 91951 Les Ulis, France
b GlaxoSmithKline, Research Triangle Park, NC 27703, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Starting from quinazoline 3a, we designed potent and selective ALK5 inhibitors over p38MAP kinase from
a rational drug design approach based on co-crystal structures in the human ALK5 kinase domain. The
quinazoline 3d exhibited also in vivo activity in an acute rat model of DMN-induced liver fibrosis when
administered orally at 5 mg/kg (bid).
Received 29 January 2009
Revised 20 February 2009
Accepted 22 February 2009
Available online 26 February 2009
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
Transforming growth factor-b
ALK5
P38MAP Kinase
Liver fibrosis
Transforming growth factor-b1 (TGF-b1) is a key mediator in
progressive fibrosis in the kidney, liver, heart, lung, bone marrow
and skin, which enhances extracellular matrix production by both
increasing the transcription of matrix proteins, for example, fibro-
nectin and collagen, and inhibiting enzymes responsible for matrix
degradation.1 More recently the role of TGF-b in cancer biology
was described in the literature.2 TGF-b1 signals through a family
of transmembrane serine/threonine kinase receptors. These recep-
tors can be divided into two classes, the type I or activin like kinase
(ALK) receptors and type II receptors. Specifically, the binding of
TGF-b1 to the type II receptor causes phosphorylation of the GS do-
main of the TGF-b type I receptor, ALK5. The ALK5 receptor, in turn,
phosphorylates the cytoplasmic proteins smad2 and smad3 at two
carboxyl terminal serines. The phosphorylated smad proteins form
heteromeric complexes with smad4, after which the complex
translocates into the nucleus to affect gene transcription.3
Therefore identification of small-molecule inhibitors of the ki-
nase activity of the TGF-b type I receptor (also named activin-like
kinase or ALK5) to block the pro-fibrotic effect of overexpression
of TGF-b1 represents an attractive target for the treatment of fibro-
tic diseases4 and cancer.5
activities, with a binding affinity of 194 nM and 1270 nM, respec-
tively. In this paper, we describe the rationale and exploration of
structure–activity relationship (SAR) leading from 3a to potent
ALK5 inhibitors with improved selectivity over p38. Among the
most potent analogues identified, the quinazoline 3d was orally ac-
tive in a rat model of liver-induced fibrosis when administered at
5 mg/kg bid.
The general synthetic route for the quinazoline compounds is
shown in Scheme 1, and is illustrated for compounds 3a–d and
8a–g. Coupling of 4-aminopyridine with 4-chloroquinazoline
intermediates 7a–d using either standard Buchwald conditions
(Pd2(dba)3, BINAP) in presence of sodium tert-butylate, or in DMF
at 150 °C, afforded the quinazolines 3a–d. The analogues 8a–g
were prepared from 4-chloro-2-(6-methyl-2-pyridinyl)quinazoline
(7d) and the corresponding amines using the same reaction condi-
tions as for 3a–d (Scheme 1). Reaction of 6-methyl-2-pyridinecarb-
oximidamide (10) with methyl acetoacetate and sodium ethylate
in ethanol produced 6-methyl-2-(6-methyl-2-pyridinyl)-4(1H)-
pyrimidinone (11) (Scheme 2). Subsequent activation of C-4 posi-
tion with POCl3, followed by the reaction of 4-aminopyridine with
the 4-chloro intermediate 12 led to pyrimidine 13. Coupling of the
2-amino-4-methyl-3-thiophenecarboxamide (14) and 3-amino-2-
thiophenecarboxamide (18) with 6-methyl-2-pyridinecarboxylic
acid using standard peptide coupling procedure (HOBT, EDCI,
Et3N), followed by cyclization with NaOH, gave, respectively, the
thieno[2,3-d]pyrimidin-4(1H)-one (15) and thieno[3,2-d]pyrimi-
din-4(1H)-one (19). Chlorination followed by reaction with 4-ami-
nopyridine led to the thienopyrimidine compounds 17 and 21,
respectively. The quinoline compound 25 was synthesized by
To date several chemical series have been described in the liter-
ature with potent small-molecule ALK5 inhibitors (Fig. 1).6 In order
to identify new templates we decided to screen in-house kinase-
focused libraries. Through this approach, we identified
a
quinazoline compound 3a presenting both ALK5 and p38 moderate
* Corresponding author.
E-mail address: francoise.gellibert@gsk.com (F. Gellibert).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.02.087

2278
F. Gellibert et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2277–2281
H
N
H
N
N
N
N
N
N
N
N
N
O
O
N
N
O
N
1
O
H
GW788388
SB-505124
N
F
N
N
HN
N
N
N
N
N
2 (SCIOS)
LY550410
Figure 1. Small-molecules inhibitors of ALK-5 kinase domain.
O
Cl
N
O
O
c
a
NH2
NH
b
N
NH
R1
NH2
R1
N
NH2
R1
O
7a-d
6a-d
4
5a-d
N
R2
HN
HN
N
N
d
N
N
or
N
R1
a, R1=phenyl
8a-d
8a, R2=3-methyl-4-pyridinyl
8b, R2=3-phenylcarboxamide
8c, R2=4-pyrimidinyl
3a-d
b, R1=3-chloro-phenyl
c, R1=4-methyl-1,3-thiazol-2-yl
d, R1=6-methyl-2-pyridinyl
8d, R2=1H-indazol-5-yl
8e, R2=1-methyl-1H-indazol-5-yl
8f, R2=1,3-benzothiazol-6-yl
8g, R2=1H-1,2,3-benzotriazol-5-yl
Scheme 1. Reagents and conditions: (a) R1COOH, HOBT, EDCI, Et3N, DMF; (b) NaOH, EtOH, reflux, 1 h; (c) POCl3, toluene, reflux; (d) 4-aminopyridine or R2NH2, DMF, 150 °C
or 4-aminopyridine, Pd2(dba)3, BINAP, NaOtBu, toluene, reflux.
condensation of 2-aminobenzonitrile 22 with 1-(6-methyl-2-
pyridinyl)ethanone, followed by cyclization of the ketimide 23
and coupling of the resulting 4-aminoquinoline 24 with 4-bromo-
pyridine (Scheme 3).
oline series (Table 1), leads to compound 3d displaying higher
ALK5 binding affinity and approximately 40-fold increase in selec-
tivity over p38 compared to 3a. Interestingly 3d is also 26-fold
more potent in the cell-based assay than the phenyl 3a. Although
the 3-chlorophenyl 3b and the 4-methyl-1,3-thiazol-2-yl 3c deriv-
atives are also potent ALK5 inhibitors with no significant p38 activ-
ity, these derivatives exhibit only low to moderate cellular activity
compared to 3d, respectively. Having established for this new
chemical series that the 6-methyl-pyridin-2-yl group was impor-
tant not only for the selectivity over p38 but also for ALK5 inhibi-
tion, we decided to explore the SAR for the two other heterocycles.
Interestingly, the pyrimidin-4-yl 8c and N-1H-indazol-5-yl 8d
derivatives are potent ALK5 inhibitors in both the binding and
cell-based assay with ALK5 inhibition comparable to 3d (Table
2). It appears that substitution of the 4-aminopyridine with a
All the compounds (Tables 1–3) were tested in a kinase assay
using purified human ALK5 kinase domain produced in Sf9 insect
cells.7 Binding to p38MAP kinase was tested on purified recombi-
nant GST-p38 a 8
These compounds were also evaluated in a
.
TGF-b dependent transcriptional luciferase assay in HepG2 cells.9
We previously demonstrated in the naphthyridine series,6a that
p38 activity could be reduced by replacing a 2-phenyl group with a
pyridin-2-yl or a 6-methyl-pyridin-2-yl. Indeed, the 6-methylpyri-
dine moiety is engaged in the ‘selectivity pocket’ of ALK5 contain-
ing Ser280 while for p38 the corresponding residue is Thr106. As
expected, this structural modification when applied to the quinaz-

F. Gellibert et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2277–2281
2279
N
N
O
N
Cl
HN
NH2
NH
N
N
c
e
d
a
NC
N
N
N
N
N
N
N
b
13
12
11
10
9
N
Cl
HN
N
O
N
O
N
N
NH
NH2
NH2
h
d
f
N
N
N
S
S
S
N
S
g
16
15
17
14
N
N
Cl
N
O
N
HN
O
S
S
S
N
NH
N
S
NH2
NH2
f
d
h
N
N
N
g
18
19
20
21
Scheme 2. Reagents and conditions: (a) Na, EtOH; (b) NH4Cl, EtOH, reflux; (c) methyl acetoacetate, EtONa, EtOH, reflux; (d) POCl3, toluene, reflux; (e) 4-aminopyridine,
BINAP, Pd2(dba)3, tBuONa, toluene, reflux; (f) 6-methyl-2-pyridinecarboxylic acid, HOBT, EDCI, Et3N, DMF; (g) NaOH, EtOH, reflux, 1 h; (h) 4-aminopyridine, DMF, 150 °C.
N
NH2
N
HN
N
CN
CN
N
b
c
a
N
N
NH2
N
24
25
23
22
Scheme 3. Reagents and conditions: (a) 1-(6-methyl-2-pyridinyl)ethanone, para-toluene sulfonic acid, toluene, reflux; (b) LDA, ether, À78 °C to rt; (c) 4-bromopyridine,
BINAP, Pd2(dba)3, tBuONa, toluene, reflux.
methyl in the ortho position 8a, or N-methylation of the indazole
ring 8e decreases significantly the binding affinity. Replacement
of the indazole core with benzothiazole 8f or benzotriazole 8g
leads also to a drop of the binding affinity for ALK5 by 26- to
149-fold over 8c, respectively. All these quinazoline derivatives
action with a water molecule is only observed for 3d, involving
the NH spacer at C-4 position of the quinazoline.
From this model and docking studies (data not shown), one
explanation for the reduction in the binding to ALK5 observed for
8e, 8f and 8g could be attributed to unfavourable interaction with
the hinge region resulting from the substitution or replacement of
the NH indazole.
Replacement of the quinazoline core with quinoline or thieno-
pyrimidines was very well tolerated (Table 3) which is consistent
with the proposed binding mode of these derivatives where the
main interactions with the binding site are maintained. These com-
pounds can be differentiated by their cellular activity, with only
the thienopyrimidine 21 and the quinoline 25 being equipotent
to 3d. The selectivity against p38 was also maintained for these
new derivatives (13, 17, 21 and 25) with activity at concentrations
8a–g are poor p38 inhibitors with activity higher than 10 lM.
In an attempt to better understand the binding mode of these
derivatives, the X-ray structures of compounds 3d and 8d bound
to the ATP binding pocket of human ALK5 were solved10 (Fig. 2).
Both compounds form hydrogen bonds with the hinge region as re-
ported previously for the other ALK5 inhibitors. The 4-pyridine
nitrogen of 3d is involved in a hydrogen bond with His 283 back-
bone NH, while the nitrogens N5 and N6 of 8d form H-bond with
His 283 backbone NH and CO, respectively. The second additional
interaction observed for 3d and 8d involves the 6-methyl-pyri-
din-2-yl nitrogen that forms a water-mediated network of hydro-
gen bonds with the enzyme. The water molecule donates
hydrogen bonds to the pyridyl nitrogen of 3d and 8d and the car-
boxyl oxygen of Glu245, and accepts hydrogen bonds from the
phenol of Tyr249 and the backbone NH of Asp351. Another inter-
higher than 10 lM (data not shown).
Among the most potent ALK5 inhibitors described above with
IC50 < 100 nM, only quinazoline 3d exhibited adequate pharmaco-
kinetic parameters in rat for dosing orally (e.g., a bioavailability of
40% and oral exposure with an AUC of 524 ng h/ml at 5 mg/kg). The

2280
F. Gellibert et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2277–2281
Table 1
Table 2 (continued)
Replacement of the phenyl ring
Compd
R2
ALK5 binding
TGF-b cellular assay
IC50a
(
lM)
IC50b
(lM)
N
H
N
HN
N
8g
3.715
>5
N
N
N
R1
a
Values are the mean of two or more separate experiments, pIC50 were calcu-
lated and mean pIC50 was converted in IC50
.
3a-d
b
Averaged values of two or more separate experiments.
Compd
R1
ALK5 binding
p38
a
binding
M)
TGF-b cellular assay
IC50a
(
lM)
IC50a
(
l
IC50b
(lM)
Table 3
3a
3b
0.194
0.049
1.27
>15
2.465
1.612
Replacement of the quinazoline core with other heterocycles
Cl
N
HN
N
N
3c
0.020
0.025
>15
0.342
0.095
X
S
N
N
Het.
3d
6.72
a
Values are the mean of two or more separate experiments, pIC50 were calculated
and mean pIC50 was converted in IC50
.
b
S
Averaged values of two or more separate experiments.
N
N
N
S
N
N
N
N
25
13
17
21
Table 2
Study of the diversity at position 4
Compd
ALK5 binding
TGF-b cellular assay
IC50a
(l
M)
IC50b
(lM)
R2
13
17
21
25
0.093
0.016
0.022
0.022
0.745
0.318
0.060
0.058
HN
4
N
N
N
a
Values are the mean of two or more separate experiments, pIC50 were calcu-
lated and mean pIC50 was converted in IC50
.
b
8a-g
Averaged values of two or more separate experiments.
Compd
R2
ALK5 binding
TGF-b cellular assay
IC50a
(lM)
IC50b
(lM)
in vivo efficacy of 3d was tested in an acute model of dimethylni-
trosamine (DMN)-induced liver disease.11 DMN given for three
consecutive days to rats increased collagen IA1 mRNA expression
in the liver by about 10-fold at day 8. In this model, 3d inhibits
in a dose-dependent manner the DMN-induced collagen IA1 mRNA
overexpression with 66% and 73% inhibition at the doses of 5 and
10 mg/kg po bid, respectively (Fig. 3).
8a
8b
0.199
1.017
N
O
0.263
0.545
NH2
The quinazoline 3d was tested against a panel of more than 15
kinases12 (Table 4). Although good selectivity was obtained against
most of them, significant percentage of inhibition (93%) was ob-
N
8c
0.040
0.025
0.060
0.090
N
served for ROCK1 at 10 lM.
In summary, a series of quinazoline was optimized and led to
novel and potent ALK5 inhibitors. The improvement of the selectiv-
ity over p38 was achieved by targeting the selectivity pocket with a
6-methyl-2-pyridine. Among these compounds, 3d, exhibited anti-
fibrotic activity when administered orally at 5 mg/kg bid in a mod-
el of DMN-induced liver fibrosis in the rat despite a modest oral
exposure at this low dose. Due to the pleiotropic effects of TGF-b,
one concern is now to demonstrate that chronic administration
of a ALK5 inhibitor is devoid of toxicity, therefore as a representa-
tive compound of a novel chemical series 3d has been progressed
in toxicity studies that will be reported in due time.
N
8d
N
H
N
8e
8f
0.588
0.646
0.794
>5
N
S
N

F. Gellibert et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2277–2281
2281
Figure 2. X-ray crystal structure of quinazoline compounds 3d and 8d bound respectively to the ATP binding site of the kinase domain of human ALK-5.
Table 4
Selectivity of compound 3d against a panel of kinasesa
Compd AMPK CDK2 CSK1 CSK2 GSK3 JNK1 LCK MKK1 MSK1 MAPK2 P70S6K P38 b P38
c
P38 d PKC PDK1 Phos.K ROCK SGK
3d
26
2
25
À2 67
38
1
5
1
15
34
À5
1
À11
28
7
À6
93
5
a
Values are %inhibition at 10
lM using 100 lM ATP. For kinases used and assay details see Ref. 12.
30
25
20
15
10
5
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2009.02.087.
References and notes
1. (a) Massague, J.; Blain, S. A.; Lo, R. S. Cell 2000, 103, 295; (b) Border, W. A.;
Noble, N. A. N. Eng. J. Med. 1994, 331, 1286; (c) Blobe, G. C.; Schiemann, W. P.;
Lodish, H. F. N. Eng. J. Med. 2000, 342, 1350.
2. (a) Roberts, A.; Mishra, L. Oncogene 2005, 24, 5667; (b) Lahn, M.; Kloeker, S.;
Berry, B. S. Exp. Opin. Invest. Drugs 2005, 14, 629; Dumont, N.; Arteaga, C. L.
Cancer Cell 2003, 3, 531.
3. Derynck, R.; Zhang, Y.; Feng, X.-H. Cell 1998, 95, 737.
4. (a) Grygielko, E. T.; Martin, V. M.; Tweed, C. J. Pharmacol. Exp. Ther. 2005, 313,
943; (b) de Gouville, A.-C.; Boullay, V.; Krysa, G., et al Br. J. Pharmacol. 2005, 145,
166; (c) Bonniaud, P.; Margetts, P. J.; Kolb, M., et al Am. J. Resp. Crit. Care Med.
2005, 171, 889; (d) Laping, N. J. Curr. Opin. Pharmacol. 2003, 3, 204.
5. Downing, J. R. N. Eng. J. Med. 2004, 351, 528.
0
6. (a) Gellibert, F.; Woolven, J.; Fouchet, M. H.; Mathews, N.; Goodland, H.;
Lovegrove, V.; Laroze, A.; Nguyen, V. L.; Sautet, S.; Wang, R.; Janson, C.; Smith,
W.; Krysa, G.; Boullay, V.; de Gouville, A.-C.; Huet, S.; Hartley, D. J. Med. Chem.
2004, 47, 4494; (b) Gellibert, F. et al J. Med. Chem. 2006, 49, 2210; (c) Byfield, S.
D.; Major, C.; Laping, N. J.; Roberts, A. B. Mol. Pharmacol. 2004, 65, 744; (d)
Sawyer, J. S. et al Bioorg. Med. Chem. Lett. 2004, 14, 3581; (e) Munchhof, M. J.
WO 2004147574.
control
DMN + vehicle
DMN + 3d
5mg/kg
DMN + 3d
10mg/kg
Figure 3. Effect of compound 3d administered orally (bid) in the acute DMN rat
model. Experiments were performed using 5–8 animals per group. Difference in
Col1A1 mRNA between control and DMN + vehicle, DMN + 3d was statistically
significantly (p < 0.0005 and p < 0.05, respectively).
7. Description of assay conditions for ALK5 binding assay can be found in
Supplementary data.
8. Description of assay conditions for p38alpha Fluorescence Polarization assay
can be found in Supplementary data.
9. Description of assay condition for TGF-b cellular assay can be found in
Supplementary data.
Acknowledgements
10. The coordinates for compounds 3d and 8d have been deposited with the RCSB
Protein Data Bank, PDB codes are 1SWZE and 1JYJE, respectively.
11. Description of assay conditions for acute DMN model can be found in
Supplementary data.
We acknowledge Géraldine Poulain and Christelle Sury-Martin
for their contribution to the synthesis of these compounds, and
the DMPK team for performing the pharmacokinetic study.
12. Davies, S. P.; Reddy, H.; Caivano, M.; Cohen, P. Biochem. J. 2000, 351, 95.