Identification and optimisation of a series of: N -(4-anilino-2-pyridyl)acetamide activin receptor-like kinase 1 (ALK1) inhibitors

Article abstract of DOI:10.1039/c6md00039h

A novel class of N-(4-anilino-2-pyridyl)amide based activin receptor-like kinase (ALK1) inhibitors are disclosed, which were rapidly optimised to a ligand efficient probe compound 21 with good potency in enzyme (4 nM) and cell (45 nM) assays and favourable physical and pharmacokinetic properties (24 h free cover over cell IC₅₀ after a single 50 mg kg^-1 dose in nude mice). This was achieved by identifying a small, ligand efficient group in the solvent channel (C2) whilst optimising the selectivity pocket (C4) group for enzyme and cell potency, using related SAR that has been observed previously for Src inhibitors.

Full text of DOI:10.1039/c6md00039h

MedChemComm
View Article Online
View Journal
RESEARCH ARTICLE
Identification and optimisation of a series of N-(4-
anilino-2-pyridyl)acetamide activin receptor-like
kinase 1 (ALK1) inhibitors†‡
Cite this: DOI: 10.1039/
c6md00039h
a
Frederick W. Goldberg, Paula Daunt,b Stuart E. Pearson,a Ryan Greenwood,a
*
Matthew Gristb and Judit É. Debreczenib
A novel class of N-(4-anilino-2-pyridyl)amide based activin receptor-like kinase (ALK1) inhibitors are
disclosed, which were rapidly optimised to a ligand efficient probe compound 21 with good potency in en-
zyme (4 nM) and cell (45 nM) assays and favourable physical and pharmacokinetic properties (24 h free
cover over cell IC50 after a single 50 mg kg−1 dose in nude mice). This was achieved by identifying a small,
ligand efficient group in the solvent channel (C2) whilst optimising the selectivity pocket (C4) group for en-
zyme and cell potency, using related SAR that has been observed previously for Src inhibitors.
Received 19th January 2016,
Accepted 20th April 2016
DOI: 10.1039/c6md00039h
www.rsc.org/medchemcomm
eneca collection was tested for ALK1 inhibition in a fluores-
cence polarization (FP) enzyme assay, based on selecting
Introduction
Activin receptor-like kinase (ALK1, also referred to as ACVRL1)
is a serine/threonine kinase of interest for oncology, due to
its reported linkage to angiogenesis in vivo.1,2 It is structurally
related to the TGF-β type I receptor (also known as activin
receptor-like kinase 5, ALK5) for which numerous small mole-
cule inhibitors have been published. By contrast, only one
series of small molecule inhibitors of ALK1 have been
disclosed3 (Fig. 1), although the mTOR inhibitor Panulisib
(P-7170, Piramal, phase I) has been reported to have ALK1
activity,4–6 monoclonal antibodies have also been described
(PF-3446962, Pfizer)7 and Acceleron are in phase II with
Dalantercept (ACE-041),8 a fusion protein comprising a por-
tion of the human ALK1 protein and IgG1 Fc region. Further-
more, a series of ALK2 inhibitors has recently been reported,
some of which are also reported to have ALK1 activity,9 which
is not surprising given the close structural homology between
ALK1 and ALK2.
compounds we knew to be structurally related to inhibitors
of ALK5. This work identified compound 3 (Fig. 2), which
was potent (48 nM) in the enzyme assay. Although we later
demonstrated that this compound only had modest activity
in an ALK1 cell assay (3.6 μM, measuring translocation of
pSMAD1/5 in HMEC-1 cell line), we were attracted to this
compound as a start point due to its excellent ligand effi-
ciency11 (LE) of 0.48. Interestingly, the C4 group, which was
assumed to bind into the selectivity pocket12 of the ATP bind-
ing site, is the same selectivity pocket group as in Src inhibi-
tor Saracatinib 4 (AZD0530),13 although the 2-amidopyridine
hinge binder and acetamide solvent channel group differ
significantly. Although the related 2,4-bisanilino-pyrimidine
hinge binder is well known in the kinase field,14–16 2-amido-
4-anilinopyridine hinge binders were unknown to us at the
time, although since this work was performed two examples
with similar hinge binders have been reported within a
quinoline-based series of ALK5 inhibitors.17
In a previous publication10 we reported the discovery of a
series of ALK5 (TGFβR1) inhibitors. This series did not have
significant ALK1 activity however (compounds were typically
∼100× selective for ALK5 vs. ALK1) so a new series was re-
quired to develop probe compounds for ALK1 inhibition. To
identify hits a small subset (1000 compounds) of the AstraZ-
a AstraZeneca, 310 Cambridge Science Park, 319 Milton Road, Cambridge,
CB4 0WG, UK. E-mail: frederick.goldberg@astrazeneca.com
b AstraZeneca, Mereside, Alderley Park, Macclesfield, Cheshire, SK10 4TG, UK
† The authors declare no competing interests.
‡ Electronic supplementary information (ESI) available: Additional experimental
protocols are given in the supporting information. See DOI: 10.1039/
c6md00039h
Fig. 1 An example from a Bayer-Schering ALK1 patent 1 and Panulisib 2.
This journal is © The Royal Society of Chemistry 2016
Med. Chem. Commun.

View Article Online
Research Article
MedChemComm
pound 13 had good cell potency (0.24 μM). Increasing the
size of the cycloalkyl group further to cyclobutyl or cyclo-
pentyl led to decreased potency, as did changing the group to
phenyl. The physical properties of compound 13 (measured
log D7.4 = 3.3, aq. solubility = 46 μM) were acceptable, so 13
was selected for further optimisation.
Switching focus onto the C4 group we started by investi-
gating whether the chlorine and dioxole groups were both re-
quired for potency (Table 2) with simple C4 substituents to
identify a more efficient compound. We found that removing
either the dioxole (17) or the chlorine (18), or both (19), sig-
nificantly reduced potency. However, we also observed that a
4-pyridyl group 20 was more potent (at lower clog P) than the
corresponding phenyl 19. Thus we investigated adding in the
4-pyridyl nitrogen to the 5-chloro-1,3-benzodioxol-4-amine
group in 13 to give 21. Compound 21 has excellent potency
in both biochemical (4 nM) and cell (45 nM) assays with
moderate lipophilicity (log D7.4 = 2.7), and therefore an im-
proved LLE (ligand lipophilicity efficiency, defined as enzyme
pIC50 − log D7.4) of 5.7 vs. 4.5 for 13. Methylating the NH
group as in 22 was tolerated but decreased potency, so was
not explored further. Achieving such high levels of potency
with low MW (332) and moderate lipophilicity gave us confi-
dence that this was an efficient scaffold for further optimisa-
tion, and that compound 21 should be profiled further as a
potential in vitro and in vivo probe compound for the effects
of ALK1 inhibition.
Unfortunately we did not have access to an ALK1 crystal
structure, so in order to confirm the predicted binding mode
of 21 we obtained a crystal structure bound to ALK5. As
expected the amidopyridine picks up the familiar donor–ac-
ceptor pair interaction to the hinge residues Tyr-282 (3.0 Å)
and His-283 (2.8 Å), which places the cyclopropyl group into
the solvent-exposed channel (Fig. 3a). The chloro-
dioxolopyridine orientates itself into the hydrophobic “selec-
tivity pocket” where it makes a water-mediated hydrogen
bond to the protein through residues Tyr-249 and Glu-245
Fig. 2 Structures of initial ALK1 hit compound 3 and Saracatinib 4.
Chemistry
The initial hit 3 and variations of the solvent channel aniline
were accessed via Buchwald–Hartwig amination of 4-chloro-
pyridine 5 with 5-chloro-1,3-benzodioxol-4-amine 6. The amide
group could then be varied by removal of the acetyl group
and HATU-mediated amide coupling with the relevant carbox-
ylic acid to form 7 (Scheme 1). In order to assess the SAR of
the C4 group, the order of the steps could be reversed to put
the C4 group in last. Thus, a Buchwald–Hartwig amination
was performed on 4-chloropyridine
8 with a range of
arylamines to give 9, which generally worked well although a
very low yield was obtained for addition of 4-aminopyridine
(for compound 20). The C4 aniline could then be selectively
methylated on the C4 nitrogen if desired with Cs2CO3/MeI to
give 10.
Results and discussion
Compound 3 was optimised for cell potency whilst keeping
molecular weight low and maintaining log D7.4 < 4. We felt
that this strategy would give us the best chance of identifying
a probe compound that would be suitable for studies to es-
tablish the effects on an ALK1 inhibitor on angiogenesis
in vivo. There was no advantage for increasing the size to
ethyl (11, Table 1) or iso-propyl (12), but cyclopropyl 13 led to
increased enzyme potency. Encouragingly, cell potency also
improved (albeit with a drop-off from enzyme to cell) so com-
Scheme 1 Reagents and conditions: (i) AcCl, pyridine, 83%; (ii) PdIJOAc)2, xantphos, Cs2CO3, DMA, chlorobenzo[d][1,3]dioxol-4-amine, 83%
(iii) NaOH, EtOH/H2O, 100%; (iv) HATU, RCO2H, iPr2NEt, DMA, 31–82%; (v) cPrCOCl, pyridine, 53%; (vi) Pd(OAc)2, xantphos, Cs2CO3, DMA,
RNH2, 3–51%; (vii) MeI, Cs2CO3, DMF, 61%.
Med. Chem. Commun.
This journal is © The Royal Society of Chemistry 2016

View Article Online
MedChemComm
Research Article
Table 1 Optimisation of the amide group
The general kinase selectivity of 21 was assessed by sub-
mission to a panel of 55 representative kinases18 at Dundee
University at 1 μM (see ESI‡ for complete dataset). At that
concentration none showed inhibition >80%, and only two
kinases showed >50% inhibition (EPHA2 and YES1). In sub-
sequent testing we also found that 21 was selective vs. Src
(4.4 μM in a HTRF enzyme assay, and inactive at 1 μM in the
Dundee university panel) despite the structural similarity to 4.
However, in addition to ALK1 it also inhibited ALK5 (0.12 μM
in an ALK5 cell assay, measuring translocation of GFP-smad2
in a recombinant MDA-MB-468 cell line). To assess ALK family
selectivity more generally, compound 21 was submitted to
a panel of FRET enzyme assays at Thermofisher, where it
showed similarly high levels of potency against ALK1 (1.1
nM, n = 3), ALK2 (1.2 nM, n = 2), ALK3 (2.4 nM, n = 2) and
ALK5 (1.9 nM, n = 3), but had some degree of selectivity
against ALK4 (31 nM, n = 2) and ALK6 (1.3 μM, n = 2).
ALK1 enz
ALK1 cell
a
Ex.
R
IC50 (μM) IC50 (μM) LEb
clog P log Dc
a
3
11
12
13
14
15
16
Methyl
Ethyl
Iso-propyl
Cyclopropyl 0.017
Cyclobutyl 0.23
Cyclopentyl 0.94
Phenyl 0.34
0.048
0.040
0.80
3.6
0.98
0.48 3.7
2.7
0.46 4.2
0.36 4.5
0.46 4.2
0.38 4.6
0.33 5.1
0.34 5.2
0.24
3.3
a
b
Geometric mean of at least 2 independent measurements. Ligand
efficiency, defined in this article as LE = (enzyme pIC50/heavy atom
count) × 1.37. c Measured log D at pH 7.4.
The in vivo PK data from Han Wistar rats (AUC = 3.8 μM h
at 5 μmol kg−1, CL = 8.4 ml min−1 kg−1, F = 38%) were prom-
ising so 21 was assessed in mouse PK. In nude mice, dosing
orally at 50 mg kg−1, 21 provided >10× free cover over the cell
IC50 24 h after a single dose (free Cmax = 31.7 μM, free con-
centration after 24 h = 0.37 μM, cell IC50 = 0.045 μM, protein
(Fig. 3b), similar water-mediated interactions as described in
our previous publication with the ALK5 protein.10
Table 2 Optimisation of the C4 group (selectivity pocket group)
LLEb
4.5
clog P
log Dc
Solubilityd (μM)
a
a
Ex.
R1
R2
H
ALK1 enz. IC50 (μM)
ALK1 cell IC50 (μM)
13
0.017
0.24
4.2
3.3
46
17
18
H
H
4.6
1.6
3.8
3.5
4.4
3.1
3.7
49
0.12 (n = 1)
2.3
19
20
21
H
H
H
2.3 (n = 1)
1.0
2.3
5.7
3.5
2.4
3.9
3.3
2.7
3.8
0.004
0.045
63
22
Me
0.026
0.32
3.7
a
b
Geometric mean of at least 2 independent measurements, unless otherwise stated. Ligand lipophilicity efficiency (LLE), defined in this
c
d
article as LLE = enzyme pIC50 − log D7.4
.
Measured log D at pH 7.4. Solubility performed under thermodynamic conditions from a solid
sample.
This journal is © The Royal Society of Chemistry 2016
Med. Chem. Commun.

View Article Online
Research Article
MedChemComm
1% formic acid or 1% aq. NH4OH) and MeCN. All reactions
were performed under nitrogen unless otherwise stated. All
in vivo experiments were performed in compliance with the
relevant laws and institutional guidelines.
Synthesis of compound 21 (see ESI‡ for other compounds)
N-(4-Chloropyridin-2-yl)cyclopropanecarboxamide 8. To a
solution of 4-chloropyridin-2-amine (25 g, 194.5 mmol) in pyr-
idine (200 mL) at 0 °C, cyclopropanecarbonyl chloride (20.33
g, 194.5 mmol) was added dropwise. The reaction mixture
was stirred for 1 h at 0 °C then was allowed to warm to rt.
The reaction mixture was stirred for 1 h at rt before being
concentrated in vacuo. The resulting residue was dissolved in
DCM (50 mL), washed with water and dried over MgSO4. The
resulting crude product was triturated with 9 : 1 iso-hexane/
diethyl ether to afford the title compound 8 (20.41 g, 53%) as
1
a cream solid; H NMR (300 MHz, DMSO) 0.78–0.90 (4H, m),
1.97–2.07 (1H, m), 7.20 (1H, dd), 8.16 (1H, s), 8.31 (1H, d),
11.07 (1H, s); m/z MH+ 197.
N-[4-[(6-Chloro-[1,3]dioxoloij4,5-b]pyridin-7-yl)amino]-2-
pyridyl]cyclopropanecarboxamide 21. To a stirred solution of
N-(4-chloropyridin-2-yl)cyclopropanecarboxamide 8 (15 g, 76.3
mmol) in 1,4-dioxane (200 mL) was added 6-chloro-
[1,3]dioxoloij4,5-b]pyridin-7-amine (13.16 g, 76.3 mmol), ce-
sium carbonate (62.1 g, 190.7 mmol), (9,9-dimethyl-9H-
xanthene-4,5-diyl)bisIJdiphenylphosphine) (4.86 g, 8.39 mmol)
and diacetoxypalladium (1.37 g, 6.10 mmol). The reaction
mixture was degassed with N2 and heated at reflux for 4 h,
then was allowed to cool to rt, diluted with DCM (200 mL)
and filtered. The filtrate was concentrated and the residue
was purified by flash silica chromatography, eluting with 2%
MeOH in DCM. The resulting residue was recrystallised from
DCM and diethyl ether, washed with diethyl ether and dried
in vacuo to afford the title compound 21 (15.00 g, 59%) as a
white powder, m.p. 206–207 °C; 1H NMR (400 MHz, DMSO)
0.76–0.79 (4H, m), 1.97–2.01 (1H, m), 6.15 (2H, s), 6.57–6.59
(1H, m), 7.62 (1H, d, J 2.3), 7.75 (1H, s), 7.99 (1H, d, J 5.7),
8.99 (1H, s), 10.59 (1H, s); 13C NMR (175 MHz, DMSO) 8.5
(2C), 16.0, 101.1, 102.1, 108.8, 117.8, 126.5, 129.6, 138.8,
148.1, 148.7, 152.6, 158.5, 172.6; HRMS (ESI) calc. for
C15H14N4O3Cl (MH+) 333.0754, found 333.0744.
Fig. 3 Crystal structure of 21 bound to ALK5, pdb code 5FRI, showing
(a) the hinge interactions and (b) the water-mediated hydrogen bond
in the selectivity pocket. The ATP binding sites of ALK1 and ALK5 are
similar, although one notable difference is that the gatekeeper residue
(Ser280 in ALK5) is Thr in ALK1.
binding in murine plasma = 27% free) so 21 is a suitable
probe compound to study the effects of ALK1 inhibition in
both in vitro and in vivo model systems.
Experimental section
General
All solvents and chemicals used were reagent grade. Flash
column chromatography was carried out using prepacked sil-
ica cartridges from Redisep, Biotage, or Crawford and eluted
using an Isco Companion system. Purity and characterization
of compounds were established by a combination of liquid
chromatography-mass spectroscopy (LC-MS) and NMR analyt-
ical techniques and was >95% for all compounds. Chemical
shifts are reported in ppm relative to tetramethylsilane (TMS)
(0 ppm) or solvent peaks as the internal reference, J values
are given in Hz. Merck precoated thin layer chromatography
(TLC) plates (silica gel 60 F254, 0.25 mm, art. 5715) were
used for TLC analysis. Preparative HPLC was performed on
C18 reversed-phase silica on a Waters or Phenomenex col-
umn using decreasingly polar mixtures of water (containing
Acknowledgements
The authors would like to thank Richard Ward for his assis-
tance with the figures and for general discussion, Steve Powell
for bioscience leadership, Adina Hughes for generation of
ALK cell data, Ieuan Roberts for generation of ALK family se-
lectivity data and Chris Phillips for deposition of the crystal
structure into the PDB.
References
1 F. Bertolini, P. Marighetti, I. Martin-Padura, P. Mancuso,
D. D. Hu-Lowe, Y. Shaked and A. D'Onofrio, Drug Discovery
Today, 2011, 16, 1052–1060.
Med. Chem. Commun.
This journal is © The Royal Society of Chemistry 2016

View Article Online
MedChemComm
Research Article
2 L. J. A. C. Hawinkels, A. G. de Vinuesa and P. ten Dijke,
Expert Opin. Invest. Drugs, 2013, 22, 1371–1383.
11 A. L. Hopkins, C. R. Groom and A. Alex, Drug Discovery
Today, 2004, 9, 430–431.
3 S. Ince, O. Prien, S. Lu, H. Yu, M. Husemann and K. Schuck,
WO2007147647, Bayer Schering Pharma Aktiengesellschaft,
Germany, 2007.
4 V. R. Agarwal, A. Joshi, M. Venkataraman, D. Bhatia, J. Bose,
L. S. Kolla, P. Gill and S. Sharma, AACR, Chicago, 2012.
5 V. A. Venkatesha, A. Joshi, M. Venkataraman, V. Sonawane,
D. Bhatia, P. Tannu, J. Bose, S. Choudhari, A. Srivastava,
P. K. Pandey, V. J. Lad, R. Sangana, T. Ahmed, A. Damre, V.
Deore, B. Sahu, S. Kumar, S. Sharma and V. R. Agarwal, Mol.
Cancer, 2014, 13, 259.
6 A. Jalota-Badhwar, D. R. Bhatia, S. Boreddy, A. Joshi, M.
Venkatraman, N. Desai, S. Chaudhari, J. Bose, L. S. Kolla, V.
Deore, N. Yewalkar, S. Kumar, R. Sharma, A. Damre, A.
More, S. Sharma and V. R. Agarwal, Mol. Cancer Ther.,
2015, 14, 1095–1106.
12 A. K. Ghose, T. Herbertz, D. A. Pippin, J. M. Salvino and J. P.
Mallamo, J. Med. Chem., 2008, 51, 5149–5171.
13 L. F. Hennequin, J. Allen, J. Breed, J. Curwen, M. Fennell,
T. P. Green, D. B. Lambert van, R. Morgentin, R. A. Norman,
A. Olivier, L. Otterbein, P. A. Ple, N. Warin and G. Costello,
J. Med. Chem., 2006, 49, 6465–6488.
14 T. H. Marsilje, W. Pei, B. Chen, W. Lu, T. Uno, Y. Jin, T.
Jiang, S. Kim, N. Li, M. Warmuth, Y. Sarkisova, F. Sun, A.
Steffy, A. C. Pferdekamper, A. G. Li, S. B. Joseph, Y. Kim, B.
Liu, T. Tuntland, X. Cui, N. S. Gray, R. Steensma, Y. Wan, J.
Jiang, G. Chopiuk, J. Li, W. P. Gordon, W. Richmond, K.
Johnson, J. Chang, T. Groessl, Y. He, A. Phimister, A.
Aycinena, C. C. Lee, B. Bursulaya, D. S. Karanewsky, H. M.
Seidel, J. L. Harris and P. Michellys, J. Med. Chem., 2013, 56,
5675–5690.
7 A. Necchi, P. Giannatempo, L. Mariani, E. Fare, D. Raggi, M.
Pennati, N. Zaffaroni, F. Crippa, A. Marchiano, N. Nicolai, M.
Maffezzini, E. Togliardi, M. G. Daidone, A. M. Gianni, R.
Salvioni and F. De Braud, Invest. New Drugs, 2014, 32, 555–560.
8 J. C. Bendell, M. S. Gordon, H. I. Hurwitz, S. F. Jones, D. S.
Mendelson, G. C. Blobe, N. Agarwal, C. H. Condon, D.
Wilson, A. E. Pearsall, Y. Yang, T. McClure, K. M. Attie, M. L.
Sherman and S. Sharma, Clin. Cancer Res., 2014, 20, 480–489.
9 A. H. Mohedas, Y. Wang, C. E. Sanvitale, P. Canning, S.
Choi, X. Xing, A. N. Bullock, G. D. Cuny and P. B. Yu, J. Med.
Chem., 2014, 57, 7900–7915.
15 I. Aliagas-Martin, D. Burdick, L. Corson, J. Dotson, J.
Drummond, C. Fields, O. W. Huang, T. Hunsaker, T.
Kleinheinz, E. Krueger, J. Liang, J. Moffat, G. Phillips, R.
Pulk, T. E. Rawson, M. Ultsch, L. Walker, C. Wiesmann, B.
Zhang, B. Zhu and A. G. Cochran, J. Med. Chem., 2009, 52,
3300–3307.
16 C. Bardelle, D. Cross, S. Davenport, J. G. Kettle, E. J. Ko,
A. G. Leach, A. Mortlock, J. Read, N. J. Roberts, P. Robins
and E. J. Williams, Bioorg. Med. Chem. Lett., 2008, 18,
2776–2780.
17 A. Jonczyk, C. Amendt and F. Zenke, WO2011054433, Merck
Patent GmbH, Germany, 2011.
18 J. Bain, L. Plater, M. Elliott, N. Shpiro, C. J. Hastie, H.
McLauchlan, I. Klevernic, J. S. Arthur, D. R. Alessi and P.
Cohen, Biochem. J., 2007, 408, 297–315.
10 F. W. Goldberg, R. A. Ward, S. J. Powell, J. E. Debreczeni,
R. A. Norman, N. J. Roberts, A. P. Dishington, H. J. Gingell,
K. F. Wickson and A. L. Roberts, J. Med. Chem., 2009, 52,
7901–7905.
This journal is © The Royal Society of Chemistry 2016
Med. Chem. Commun.