Triazolopyridines as selective JAK1 inhibitors: From hit identification to GLPG0634

Article abstract of DOI:10.1021/jm501262q

Janus kinases (JAK1, JAK2, JAK3, and TYK2) are involved in the signaling of multiple cytokines important in cellular function. Blockade of the JAK-STAT pathway with a small molecule has been shown to provide therapeutic immunomodulation. Having identified

Full text of DOI:10.1021/jm501262q

Subscriber access provided by Brought to you by ST ANDREWS UNIVERSITY LIBRARY
Article
Triazolopyridines as selective JAK1
inhibitors: from hit identification to GLPG0634
Christel Jeanne Menet, Stephen Robert Fletcher, Guy Van Lommen, Raphael geney,
Javier Blanc, Koen Smits, Nolwenn Jouannigot, Ellen Van der Aar, Philippe Clement-
Lacroix, Liên Lepescheux, René Galien, Béatrice Vayssiere, Luc Nelles, Thierry Christophe,
Reginald Brys, Luc Van Rompey, Fabrice Ciesielski, Murielle Uhring, and Pierre Deprez
J. Med. Chem., Just Accepted Manuscript • Publication Date (Web): 04 Nov 2014
Downloaded from http://pubs.acs.org on November 5, 2014
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 free 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 accessible to all
readers and 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.
Journal of Medicinal Chemistry 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 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Triazolopyridines as selective JAK1 inhibitors: From hit identification to
GLPG0634
1
2
1
3
1
Christel J. Menet* , Stephen R, Fletcher , Guy Van Lommen , Raphael Geney , Javier Blanc ,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
1
3
1
Koen Smits , Nolwenn Jouannigot , Pierre Deprez , Ellen M. van der Aar , Philippe Clementꢀ
3
3
3
3
1
Lacroix , Liên Lepescheux , René Galien , Béatrice Vayssiere , Luc Nelles , Thierry
1
1
4
4
1
Christophe , Reginald Brys , Muriel Uhring , Fabrice Ciesielski , Luc Van Rompaey
1
Galapagos NV, Generaal de Wittelaan L11A3, 2800 Mechelen, Belgium
2
BioFocus, Chesterford Research Park, Saffron Walden, Essex, CB10 1XL, United Kingdom
Galapagos SASU, 102 Avenue Gaston Roussel, 93230 Romainville, France
NovAliX, Bld Sébastien Brant, BP 30170, Fꢀ67405 Illkirch CEDEX, France
3
4
*Corresponding author: email: christel.menet@glpg.com, tel : +32 15 342 938
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 70
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: Janus kinases (JAK1, JAK2, JAK3, TYK2) are involved in the signaling of multiple
cytokines important in cellular function. Blockade of the JAKꢀSTAT pathway with a small
molecule has been shown to provide therapeutic immunomodulation. Having identified JAK1 as
a possible new target for arthritis at Galapagos, the compound library was screened against JAK1
resulting in the identification of a triazolopyridineꢀbased series of inhibitors represented by 3.
Optimization within this chemical series led to identification of GLPG0634 (65, filgotinib), a
selective JAK1 inhibitor currently in Phase 2B development for RA and Phase 2A development
for Crohn’s Disease (CD).
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
ACS Paragon Plus Environment

Page 3 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Janus kinases (JAKs) are cytoplasmic nonꢀreceptor tyrosine kinases. Four human JAK family
members are known: JAK1, JAK2, JAK3 and TYK2 (tyrosine kinase 2). It is well documented
that JAK kinases play a central role in cytokine receptor signaling through phosphorylation and
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
activation of signal transducer and activator of transcription (STAT) proteins. JAKs are
constitutively bound to the cytoplasmic tail of the cytokine receptors. Upon binding of the
cytokine, JAK family members autoꢀ and/or transꢀphosphorylate, leading to phosphorylation of
the receptors at specific tyrosine (Tyr) residues. The resulting activated STATs migrate to the
nucleus and modulate transcription. JAK1, JAK2, and TYK2 kinases are ubiquitously
expressed, whereas JAK3 is limited to the lymphoid lineage. Many cytokines relying on JAKs
for intracellular signal transduction have key pathoꢀphysiological roles in autoꢀimmune and
inflammatory diseases. Such cytokines include ILꢀ2 (interleukine), ILꢀ6, ILꢀ12, ILꢀ23, IFNꢀα,
1
IFNꢀγ (interferon) and GMꢀCSF (Granulocyte macrophage colonyꢀstimulating factor).
Therapeutic benefit from JAK kinase inhibition has been established in rheumatoid arthritis (RA)
TM
with tofacitinib, 1 (Xeljanz ) and in the treatment of myelofibrosis with ruxolitinib, 2
TM
(
Jakafi ). Tofacitinib generally reported as a JAK1/JAK3 inhibitor is found in house (figure 1)
TM
to have a JAK1/2/3selectivity profile. Ruxolitinib, 2 (Jakafi ) generally known as a
JAK1/JAK2 inhibitor has been shown in different literature references to have a
2
,3
ꢀ/ꢀ
JAK1/JAK2/TYK2 profile (figure 1). In 1998, Schreiber et al. showed using JAK1 cells that
although responsive to many cytokines JAK1 is necessary for cells to respond effectively to
multiple immuneꢀrelevant cytokines; these include all class II cytokine receptors, receptors that
utilize the γ subunit for signaling and the family of cytokine receptors that depend on the gp130
c
4
subunit signaling. Moreover, in 2001 Yoshimura et al. conducted in vivo CIA (Collagenꢀ
Induced Arthritis
)
experiments by periarticular injection of adenovirus carrying
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 70
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
CIS3/SOCS3/SSI3 to inhibit JAK tyrosine kinase activity. This study showed drastic reduction
of the severity of the CIA disease, leading them to conclude that a JAK1 inhibitor may be a
useful therapeutic reagent for RA (rheumatoid arthritis). The work carried out by these two
groups validated the Galapagos approach to the identification of JAK1 as a possible new target
in arthritis using viral siRNA libraries in functional screening. Additional recent findings suggest
that JAK1 dominates ILꢀ2 induced JAK1/JAK3/STAT5 signaling and ILꢀ6 induced
JAK1/JAK2/TYK2/STAT1 signaling. This implies that JAK1 inhibition alone can provide in
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
,7,8,4,9,10,11
vivo efficacy towards immuneꢀinflammatory diseases.
Since JAK2 inhibition leads to
unwanted side effects such as anemia, thrombocytopenia and neutropenia, selective JAK1
inhibition could provide an increased therapeutic window, allowing higher dosing and efficacy,
while avoiding doseꢀlimited pharmacology observed with pan JAK inhibitors (i.e. compounds
12,13,14,15
inhibiting multiple JAKs).
Herein we describe the finding and characterization of a triazolopyridine series as JAK inhibitor
along with our initial efforts at achieving the desired JAK1 selectivity profile within this
chemotype.
Figure 1: tofacitinib, 1; ruxolitinib, 2: *Biochemical IC50 determination: recombinant JAK1 (Invitrogen), JAK2
(
Carna Biosciences), The ATP concentration was equivalent to Km value.
CN
N N
Compound
JAK1, IC50 (nM)*
JAK2, IC50 (nM)*
1
2
0.8
1.2
0.6
0.2
N
N
N
CN
N
O
N
N
N
H
N
H
1
2
4
ACS Paragon Plus Environment

Page 5 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To identify chemical starting points for JAK1 inhibitors, the BioFocus kinaseꢀfocused library
collection was screened against the kinase domain of JAK1 in an in vitro biochemical assay at 10
µM initially. All compounds having a percentage inhibition above 75% were further evaluated in
dose response tests with measurement of IC50 values. Triazolo[1,5ꢀa]pyridine 3 exhibited
encouraging JAK1 inhibitory activity (70 nM), and owing to its modular nature, emerged as a
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
6
good platform to initiate chemistry (low molecular weight: 308, appropriate xLogP3 : 2.35).
Furthermore, 3 appeared to be slightly more potent on JAK1 than on the other JAKs in the
biochemical assays (Figure 2). A similar selectivity profile has since been reported by Siu et al.
17
on the same triazoloꢀpyridine scaffold. The catalytically active JH1 (JAK Homology) kinase
domain carries out the phosphorylation events responsible for the physiological function of the
JAK kinases, and this selective series of molecules appeared to be ATP competitive inhibitors
binding within this region. However, compound 3 was not optimal, showing a moderate potency
on JAK1 kinase and low metabolic stability (6% of compound remaining after 60 min in human
microsomes).
Figure 2. Compound 3, hit identified during our HTS; *LiPE (lipophilic efficiency) = pIC ꢀxlogP3
50
Hit: compound 3
N
N
hJAK1 IC = 70 ± 14 nM
50
NH
N
O
hJAK2 IC = 138 ± 22 nM
50
hJAK3 IC = 528 ± 82 nM
50
O
hTYK2 IC = 519 ± 55 nM
50
LiPE* = 4.65
Exploiting the subtle sequence differences in the active sites of the JAK family kinases via a
combination of structureꢀactivity relationships (SAR) and structureꢀbased design, we report a
series of modifications to the exocyclic amine and the 5ꢀposition of compound 3. To the best of
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 70
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
our knowledge, at that time no JAK1 selective inhibitor was yet described and no help from
known SAR could be used. Nevertheless sequential improvements in potency together with
achieving a balance of drugꢀlike features, led to the discovery of compound 65 (GLPG0634,
filgotinib), a potent, selective, and orally active JAK1 inhibitor with an excellent PK profile in
two preclinical species (rat and dog). In addition, compound 65 showed good in vivo efficacy in
a rat CIA model.
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
Chemistry: Analogs of 3 were synthesized according to the procedure depicted in Scheme 1,
starting from the commercially available 2ꢀaminoꢀ6ꢀbromoꢀpyridine 4. The central scaffold was
constructed by reaction of ethoxycarbonyl isothiocyanate with 4 to form the thiourea 5 which
1
8
was then reacted with hydroxylamine to yield, after heating, the desired intermediate 6.
Scheme 1. Synthesis of the triazoloꢀpyridine scaffold
8
1
N
7
i
S
OEt
ii
NH2
N
2
6
Br
N
NH2
Br
N
N
N
O
N
H
H
5
3
4
5
Br
6
i) ethoxycarbonyl isothiocyanate, DCM, RT; ii) hydroxylamine hydrochloride, DIPEA, EtOH/MeOH, reflux
Next, paraꢀmethoxyphenyl boronic acid was first coupled via a Suzuki reaction to the bromoꢀ
aminoꢀtriazolopyridine scaffold to provide 7. Derivatives of 7 with various groups in the 2ꢀ
position were easily accessible using different functionalization of the free amine. Acylation was
carried out with the acyl chlorides and the initial double acylation adducts easily hydrolyzed with
an ammonia solution in methanol to provide the monoꢀamides (Scheme 2). Palladium coupling
was carried out by first transforming the free amine to iodine by a Sandmeyer reaction, followed
by Buchwald or Suzuki reactions (Scheme 2). To vary the 5ꢀposition, the triazoloꢀpyridine
scaffold 6 was acylated with cyclopropanecarboxyl chloride to give 9, followed by a Suzuki
reaction (Scheme 2).
6
ACS Paragon Plus Environment

Page 7 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 2. Synthesis of the triazoloꢀpyridine derivatives
N
iia,b,c
N
N
i
N
N
6
NH2
NH
A
R
NH2
N
N
N
W n
N
vi
Br
7
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
W = O
n = 0, 1
O
O
A= C, CH2
2
5, 28, 30, 32, 33, 36
N
N
iii
NH
N
N
O
NH
R
N
N
iv
Br
9
N
N
I
N
vii
27, 35
O
N
N
N
O
v
NH
8
N
N
N
N
O
Ar
34
3
, 37- 47, 55 - 65
O
i) paraꢀmethoxyphenyl boronic acid, Pd(dppf)Cl , K CO , 1,4ꢀdioxane/water, 100°C; ii) a. acid chloride, Et N,
2
2
3
3
CH CN; b. NH (7N) in MeOH; b. 4ꢀnitrophenyl chloroformate, DIPEA, methylamine, DCM; c. aldehyde,
3
3
19
Ti(OiPr) Na(CN)BH , EtOH; iii) HI (57% aqueous solution); NaNO , DMSO; iv) Pd(OAc) , Cs CO , BINAP,
4,
3
2
2
2
3
aniline, 1,4ꢀdioxane or toluene; v) Pd(dppf)Cl , K CO , boronic acid, 1,4ꢀdioxane, vi) Cyclopropanecarbonyl
2
2
3
chloride, Et N, CH CN; vii) paraꢀmethoxyphenyl boronic acid, Pd(dppf)Cl , K CO , 1,4ꢀdioxane/water, 100°C
3
3
2
2
3
Examples such as 49 with additional substitution on the scaffold were synthesized similarly,
starting from the aminoꢀpyridine with the desired substitution, followed by sequential coupling,
acylation and Suzuki reaction (Scheme 3).
Scheme 3. Synthesis of the compound 49
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 70
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
H
H
N
NH2
N
O
N
ii
i
NH2
N
N
S
O
N
O
N
O
O
Cl
12
1
0
Cl
11
Cl
NH
N
N
iii
NH
iv
N
N
N
N
O
O
O
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
O
Cl
4
9
1
3
O
i) ethoxycarbonyl isothiocyanate, DCM, RT; ii) hydroxylamine hydrochloride, DIPEA, EtOH/MeOH, reflux; iii)
acid chloride, Et N, CH CN; iv) paraꢀethoxyphenyl boronic acid, Pd(dppf)Cl , K CO , 1,4ꢀdioxane/water, 100°C
3
3
2
2
3
The isomeric scaffold with substitution at the 8ꢀposition (compounds 50 and 52) was obtained
starting from the isomeric pyridine with or without substitution. Compound 54 was made by the
same sequence, starting from aminoꢀpyrazine.
For the synthesis of the triazoloꢀpyrimidine scaffold, compound 51, it was necessary to start the
synthesis via a Suzuki coupling on 6ꢀaminoꢀ2ꢀchloropyrimidine, as the cyclisation (step ii)
would not proceed in the presence of the chlorine (Scheme 4).
Scheme 4. Synthesis of compound 51, triazoloꢀpyrimidine scaffold
NH2
H
H
NH2
N
N
O
N
N
ii
i
N
N
N
N
S
O
Cl
14
15
1
6
O
O
N
N
N
iv
iii
NH
O
NH2
N
N
N
N
N
17
51
O
O
8
ACS Paragon Plus Environment

Page 9 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
i) paraꢀethoxyphenyl boronic acid, Pd(dppf)Cl , K CO , 1,4ꢀdioxane/water, 100 °C, ii) ethoxycarbonyl
2
2
3
isothiocyanate, DCM, RT; iii) hydroxylamine hydrochloride, DIPEA, EtOH/MeOH, reflux; iv) acid chloride, Et N,
3
CH CN
3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
An alternative approach to the synthesis of compound 53 was required as the methods above
failed to give the desired product. Instead, acetophenone 18 was reacted with dimethylformamide
dimethyl acetal 19 to give compound 20, which was condensed with diꢀaminotriazole 21 to yield
22. Acylation of 22 gave the required product 53 (Scheme 5).
Scheme 5. Synthesis of compound 53
O
O
H
N
O
N
i
N
+
NH2
+
O
N
N
H N
2
18
19
20
21
N
N
N
N
N
H
N
iii
NH2
ii
N
N
N
O
22
53
i) 150°C, MW (Wmax:300; Pmax:150PSI); ii) EtOH, 100 °C; iii) acid chloride, Et N, CH CN
3
3
Results and Discussion
The goal of this work was to develop a potent and selective JAK1 inhibitor possessing
appropriate ADME/DMPK properties to enable clinical development via oral dosing for the
treatment of immuneꢀinflammatory disorders. Moreover it was reasoned that, in order to be a
potential drug candidate, any new compounds should reach levels of potency in the low
nanomolar range like the existing drug 1 or 2 with more importantly a greater JAK1 selectivity.
Indeed, the accepted dose by the FDA of 1 is low (5 mg), but some observation of reduced
hemoglobin levels at higher doses (15 mg BID) during clinical trials was attributed to possible
effects on hematopoiesis through inhibition of JAK2 and consequently its lack of selectivity.
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 70
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
This led us to hypothesise that an increased degree of selectivity for JAK1 over JAK2 may be
sufficient to provide a suitable therapeutic window.
Analogues of 3 were primarily screened in JAK1 and JAK2 biochemical assays. Selectivity
against JAK3 and TYK2 biochemical assay was subsequently evaluated for selected compounds.
Biochemical JAK2 inhibition was assessed early in the screening cascade, due to the essential
role of JAK2 in hematopoiesis, including EPO receptor signaling and red blood cell homeostasis
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
3,14
as mentioned above.
Assessment against a selectivity panel of 14 kinases was also employed
for compounds of interest. The series appeared to be selective towards the JAK kinases, as none
of an initial target list of 14 kinases were inhibited by the compound 3 at an IC < 2 µM (~>30
5
0
fold of selectivity) (Table 1).
Table 1. Selectivity towards 14 kinase panel (biochemical assay)
Kinase
FGFR3
GSK3β
IKKβ
Cpd 3 IC (nM)
Kinase
PYK2
RIPK2
ROCK1
SYK
Cpd 3 IC (nM)
50
50
>5,741
7,114
3,184
50,978
>10,000
2,272
24,689
4,440
LCK
MAPKAPK5
p38α
>10,000
>10,000
39,986
TAK1
TBK1
CDK2
>10,000
>4,000
>10000
PKA
From Hit to potent JAK inhibitors (SAR)
1
0
ACS Paragon Plus Environment

Page 11 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
For initial SAR development towards enhancing JAK1 selectivity, both the cyclopropyl and
paraꢀmethoxy phenyl positions were explored via carefully designed parallel synthesis efforts.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 2. Substitution on exocyclic NH. N/A: compound not active against JAK2; NT: not tested against JAK2.
Compounds were tested at least twice unless indicated otherwise (n=1); * (JAK2 IC )/(JAK1 IC ). Biochemical
50
50
IC50 determination: recombinant JAK1 (Invitrogen), JAK2 (Carna Biosciences), the ATP concentration was
equivalent to Km value.
N
R
N
N
O
Compound
R
JAK1 IC50 (nM) ± s.e.m
JAK2 selectivity index)*
Compound
R
JAK1 IC50 (nM) ± s.e.m
(JAK2 selectivity index)*
(
O
O
7
0 ± 14 (2)
3
N
H
30
N
>10,000
H
>
10,000 (N/A)
7
NH2
3,055 ± 716 (0.6)
31
N
N
2
2
4
5
N
>10,000 (N/A)
>10,000 (NT)
32
33
>10,000 (N/A)
413 (0.7) (n=1)
H
O
O
N
H
N
H
N
H
O
2
6
>10,000 (N/A)
34
35
36
>10,000 (N/A)
2.8 ± 0.27 (0.9)
1,979 ± 691 (0.5)
N
N
H
N
O
O
S
27
5,740 (N/A) (n=1)
O
N
N
H
H
O
O
2
2
8
9
>10,000 (N/A)
>4,000 (N/A)
N
H
N
H
O
N
H
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 70
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
We first started investigating the 2ꢀposition of the triazoloꢀpyridine through replacement of the
cyclopropylamide of the original hit. SAR rationalization efforts were supported by modeling
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
0
using a proprietary docking model based on the JAK1 coꢀcrystal structure incorporating 1. The
validity of this docking model was later confirmed by the crystal structure of JAK2 with
compound 65, Figure 4. Synthesized compounds without a free hydrogen atom on the amine at
the 2ꢀposition (24, 26 and 31) essentially lost all activity on JAK1. Modeling studies indicate that
an Hꢀbond is formed with the main chain carbonyl of Leu959 of the hinge region, as shown on
Figure 3A.
Figure 3. A. Model of compound 3 docked in the JAK1 JH1 kinase domain; B. Model of compound 3 docked in the
JAK1 JH1 kinase domain. B. Molecular surface and side chains of residues within 5 ångströms of compound 3.
A
B
From our early investigations on the JAK1 binding mode (Figure 3A/B), it was initially expected
that the core could be substituted with various solubilizing groups at the 2ꢀposition which would
drive potency and improve the ADME profile, as this portion of the molecule would be directed
toward the solvent region. However, replacement of the cyclopropyl amide moiety displayed in
1
2
ACS Paragon Plus Environment

Page 13 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 2 showed the importance of this position on the activity and selectivity of the compounds.
Surprisingly, close amide analogs, such as isopropylamide 28 or acetamide 29 displayed much
reduced potency against JAK1. Only the crystal structure obtained later (Figure 4), gave a
possible explanation for this SAR cliff. It was hypothesized that the superior ability of the
cyclopropane ring to donate an Hꢀbond could explain the SAR observations. Indeed, the
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
65/JAK2 crystal structure shows a putative Hꢀbond between the CH carbon atom of the
cyclopropylamide and the carbonyl oxygen of Leu932 and of Pro933 in JAK2 (Pro960 and
Leu959 in JAK1): bond length 3.3ꢀ3.4 Å. This type of interaction has been previously described
21
in the literature. Cyclopropanes and other hydrocarbons containing threeꢀmembered rings have
a certain sp2 character and gain substantial Hꢀbond acidity. Furthermore compound 25 (methylꢀ
cyclopropyl) and compound 28 (isopropylamide) support this hypothesis by having little or no
activity and being unable to form this nonꢀclassical hydrogen bond. Urea 33 bearing an
additional Hꢀdonor is more potent than the ethylamide analog 36, but it should be noted that 33
loses selective against JAK2.
Directly linking alkyl groups to the exoꢀnitrogen (e.g. 32) is not tolerated. Thus, potency on
JAK1 is lost. This is postulated to be due to the difference in pKa of the NH, preventing a strong
Hꢀbond with the carbonyl of the hinge and/or maybe due to a higher flexibility of these
substituents. In addition, the carbonyl in the amide functionality is directly involved in Hꢀbond
interactions with neighboring water molecules as observed in the Xꢀray structure of the JAK2/65
crystal (Figure 4). The loss of such interactions in the alkyl analogue could be detrimental for
potent binding.
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 70
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
Larger replacement of the cyclopropyl like benzylamides or replacement of the carbonyl by
sulfonamides resulted in poorly active or inactive compounds towards JAK1, possibly due to
subꢀoptimal interaction with the hinge.
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 only suitable replacement of the cyclopropylamide group was the aniline 35, increasing
hugely the IC to 2.8 nM, however the selectivity profile shifted from JAK1 to JAK2 (further
5
0
1
7
details on this JAK2 selective series have been described by others). Some other groups
reported that bulkier groups in the hinge region were better tolerated in JAK2 than in JAK1 due
to a larger available space in that area for JAK2. In summary the cyclopropyl moiety appears to
25
be a preferred substituent for JAK1 selectivity over JAK2.
Table 3. Investigation on 5ꢀposition; NT = not tested; Compounds were tested at least twice unless indicated; *
(
JAK2 IC )/(JAK1 IC ), (JAK3 IC )/(JAK1 IC ), (TYK2 IC )/(JAK1 IC ). Biochemical IC determination:
5
0
50
50
50
50
50
50
recombinant JAK1, TYK2 (Invitrogen), JAK2, and JAK3 (Carna Biosciences)
N
NH
N
N
O
Ar
Selectivity Index*
Compound
Ar
JAK1
IC (nM)
JAK2
2.0
JAK3
TYK2
7.4
50
3
70 ± 14
184 ± 30
371 ± 68
7.5
7.1
2.0
O
3
3
7
8
2.3
1.5
4.1
2.8
O
1
4
ACS Paragon Plus Environment

Page 15 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Cl
3
4
9
0
142 ± 29
156 (n=1)
1.1
3.1
6.7
7.4
2.4
28.2
Cl
N
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
41
42
2,554 ± 532
547 ± 136
0.7
1.5
1.5
4.4
>1.6
6.2
O
N
O
N
N
4
4
4
3
4
5
975 ± 469
1,351
2.1
1.1
0.9
>4.1
2.1
>4.1
>3.0
2.8
N
N
H
O
572 (n=1)
3.8
O
46
103 ± 39
>10,000
1.0
NT
3.0
NT
18
4
7
NT
O
We next turned our attention to the SAR around the phenyl group at the 5ꢀposition of the
triazolo[1,5ꢀa]pyridine core (Table 3). Replacement of phenyl by more polar heterocycles
(compounds 42 to 44) resulted in a substantial (8 to 20 fold) reduction in potency. This is
consistent with modeling studies from which it was evident that the phenyl ring sits in a
hydrophobic subꢀpocket of JAK1, forming hydrophobic contacts with Leu881 (Leu855 in
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 70
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
JAK2), Val938 (Val863 in JAK2), Ser963 (Ser936 in JAK2), Leu1010 (Leu983 in JAK2) and
Gly1020 (Gly993 in JAK2) (Figure 3B) .
Relative to the unsubstituted phenyl derivative 37, substituents in the orthoꢀposition caused a
twoꢀfold decrease in potency for small substituents. This drop in potency was even higher for
larger groups (for example compound 47). This may be explained by a steric clash observed with
the triazolopyridine scaffold, forcing an unfavorable offꢀplanar conformation of the phenyl ring.
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 comparison, adding a chloroꢀsubstituent at the metaꢀposition of the phenyl ring (compound
39) has no effect on the potency of JAK1 inhibition, but seems to be more favorable towards
JAK2 inhibition. Consequently, it was observed that the most potent and most selective
compounds were those with the 5ꢀphenyl substituted at the paraꢀposition. Modeling studies
confirmed that substituents in this position point toward the glycineꢀrich loop where it is possible
to make stabilizing interactions or have a positive impact on the water network. Indeed, this loop
has been reported in numerous kinase crystal structures to be involved in a network of
interactions with inhibitors, and displays remarkable conformational flexibility allowing
accommodation of a vast array of ligands.
Table 4. Changes on scaffold; N/A= not Active; * (JAK2 IC )/(JAK1 IC ). Biochemical IC determination:
5
0
50
50
recombinant JAK1 (Invitrogen), JAK2 (Carna Biosciences)
N
N
NH
O
O
Compound
structure
JAK1 IC50 (nM)
JAK2 selectivity
Compound
structure
JAK1 IC50 (nM)
(JAK2 selectivity
(
1
6
ACS Paragon Plus Environment

Page 17 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
index)*
index)*
N
N
O
NH
O
N
4
4
8
9
27 ± 9 (2)
52
53
N/A
N
N
F
NH
O
F
N
O
F
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
N
N
NH
N
O
O
3,012 ± 492 (>1)
N
N
N/A
N
NH
O
N
O
N
N
N
O
NH
N
O
5
5
0
1
67 ± 17 (2.2)
54
3,098 (1)
N
N
NH
O
N
N
O
N
N
NH
N
O
418 ± 125 (1.5)
O
Substitution at the 6ꢀposition of the scaffold appears not to be tolerated for JAK1 inhibition, as
shown for compound 49 (Table 4). The reason lies in the skewed torsion of the phenyl ring at
the 5ꢀposition provoked by the adjacent methoxy group. An isomeric scaffold was made by
repositioning the ring junction nitrogen. Analogs of this isomeric scaffold, 8ꢀsubstituted 2ꢀ
aminoꢀ[1,2,4]triazolo[1,5ꢀa]pyridine, had similar or slightly reduced potency (48 vs 50), and
similar ADME properties (data not presented). On this scaffold, additional substitution also led
to loss of potency (compound 52). Finally, modification of the scaffold by addition of an extra
nitrogen atom showed no inhibition of JAK1. It appears that the electronics of the pyrazoloꢀ
pyridine scaffold are optimal for binding in the JAK ATP pocket.
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 70
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
Encouraged by these preliminary results of the 5ꢀphenyl substituent, we then focused the SAR
exploration on the paraꢀposition of this substitution.
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
Table 5. SAR of the paraꢀposition; *Not measured due to precipitation; * ASOL = kinetic solubility measure with
2
% DMSO; PPB = Plasma Protein Binding. Biochemical IC50 determination: recombinant JAK1, TYK2
(
Invitrogen), JAK2 and JAK3 (Carna Biosciences); ** LiPE (lipophilic efficiency) = pIC ꢀxlogP3
50
N
N
NH
N
O
R
Index selectivity
JAK2 JAK3 TYK2
PPB
%
Compound
R
JAK1 IC50
nM)
xlogP3 ASOL* LiPE
(
pH=7.4
**
(
rat)
HN
S
55
O
O
23±8
1.3
37
3.0
90
1.89
>80
5.74
N
N
5
6
O
N
0.6±0.2
5.3
1.7
21
ꢀ
66
ꢀ
99
ꢀ
3.6
4
5.62
4.47
H
57
OH
8.3 (n=1)
3.61
28
OH
HN
5
8
1.4 (n=1)
11.4
26
9
99
3.5
9.4
5.35
CN
N
O
5
6
9
0
147 (n=1)
92 ± 12
7.5
1.8
ꢀ
ꢀ
ꢀ
3.63
1.84
ꢀ
3.20
5.19
CN
N
9.5
6.1
71
>38
O
1
8
ACS Paragon Plus Environment

Page 19 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
An illustrative selection of the synthesized substitutions at the paraꢀposition is presented in
Table 5. Substitution seems to increase the potency when large enough to contact the glycine
rich loop. Sulfonamides in the paraꢀposition (e.g., 55) gave high potencies. Unfortunately,
permeability was found to be compromised for this subseries with high efflux observed (Caco2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ꢀ
6
ꢀ1
assay: permeability apparent (A2B) 3.7 cm.10 sec ; efflux ratio: 16.5). The best potency and
selectivity for JAK1 was obtained with groups consisting of a long chain terminated by an
aromatic group such as compound 56 or 58 or 59. JAK1 has a histidine in the pꢀloop that can
have a favorable interaction with aromatic moieties. In the number of substitutions that were
explored, high selectivity was achieved for compound 58 with 11 fold selectivity for JAK1 over
JAK2 inhibition (the phenyl can directly interact with the histidine and the nitrile with the
glycine rich loop) and low selectivity was obtained if the aromatic was substituted by an alkoxy
like compound 57 (less favorable in the interaction with the histidine). This lipophilic motif,
unfortunately, was not compatible with good ADME properties, giving low solubility, low LiPE
(
<6) and often high intrinsic clearance (data not presented).
Moderate activity was regained with the introduction of the polar, moderately basic,
morpholinomethyl group in compound 60. Moreover promising in vitro properties surfaced in
this series: T human microsome > 90 min; no efflux with high permeability and good solubility
1/2
and PPB of 70% in rat.
Good aqueous solubility and PPB with improved potency.
Encouraged by the good ADME properties, we sought alternative of the morpholine moiety. 60
was used as a reference compound for probing the structureꢀactivity relationships of a diverse
subset of amines to assess if identification of a compound with high potency and improved
1
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 70
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
physical properties could be attained. In this small library, all compounds conserve the JAK1ꢀ
selectivity, but only few examples show attractive potency on JAK1.
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
Table 6. SAR of the paraꢀphenyl substitution. *RLM: Rat Liver Microsome (%remaining after 60 min): *** Caco2:
ꢀ6
ꢀ1
permeability (A2B) 0.3 cm.10 sec , efflux ratio, 69. NT = not tested (racemic mixture). Biochemical IC
50
determination: recombinant JAK1, TYK2 (Invitrogen), JAK2, and JAK3 (Carna Biosciences); ** LiPE (lipophilic
+
efficiency) = pIC ꢀxlogP3; inꢀhouse data
50
N
N
NH
N
O
index JAK1-selectivity
Compound
Ar
JAK1 IC50
nM)
ASOL
pH_7.4
RLM
% *
PPB %
(rat)
LiPE**
5.19
(
JAK2 JAK3 TYK2
N
6
0
92 ± 12
>4,000
1.8
ꢀ
9.5
ꢀ
6.1
ꢀ
>38
ꢀ
38
ꢀ
71
ꢀ
O
61
N
3.38
N
NH
F
6
6
2
3
16 ± 4
27 ± 4
2.6
1.1
9.5
8.7
6.2
3.2
>80
27
80
26
83
70
5.19
4.20
F
O
F
N
F
F
NH
6
4***
41 ± 14
2.1
8
9.6
>85
85
NT
6.07
S
O
O
2
0
ACS Paragon Plus Environment

Page 21 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
N
O
6
5
S
10 ± 0.8
0.8
2.8
0.8
81
11.6
42
>85
ꢀ
65
47.5
6.64
7.77
O
+
+
1
cf Figure 1
0.4
27
33
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The nature of the para-substitution on the 5ꢀphenyl substituent, in particular lipophilic groups,
was found to have a pronounced effect on potency, as shown above. To extend the
understanding of this effect, compound 60 bearing a morpholinoꢀbenzyl group was used as a
starting point for the optimization of the LiPE and ADME properties of the series (Table 6).
Extended amines with strong basicity such as piperazine 61 appeared to be completely inactive
against JAK1. The less basic piperidine analogue, 4ꢀdifluoropiperidine 63 and the nonꢀbasic
trifluoroꢀamide gave rise to good potency but loss of selectivity and again decreased LiPE.
After extensive studies, the cyclic sulfone moiety was found to offer a good balance of
selectivity and with a LiPE slightly lower than compound 1, but still higher than 6 (Table 6).
Thus compounds 65 and 64 reached a JAK1 potency below 50 nM with an improved selectivity
of ~2/3ꢀfold over JAK2 in contrast to the potent but unselective 1. Compound 64 was not
profiled further due to the very low permeability observed in Caco2 and less attractive selectivity
toward the other JAKꢀmembers (the two enantiomers were not separated). Compound 65 showed
the best in vitro stability in rat; however compounds 60, 63 and 65 were all three advanced into
rat PK studies.
Table 7. Pharmacokinetics in rats; formulation was in methyl cellulose (MC) 0.5%
IV
PO
IV
PO
PO
5 mg/kg
63
1
mg/kg
5 mg/kg
1 mg/kg
5 mg/kg
Compound
60
65
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 70
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
C0 or Cmax (ng/mL)
Tmax (h)
863
_
1,320
0.33
1,437
_
1,407
_
310
2.2
1,681
_
547
0.25
690
_
AUC(0ꢀz) (ng.h/mL)
Cl (L/h/kg)
Vss (L/kg)
470
2.12
1.46
0.74
739
1.35
1.76
1.6
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
_
_
_
T1/2 (h)
0.92
3.9
0.92
N/A
F (%)
62
45
The PK properties of the 3 selected compounds are summarized in Table 7. Both 60 and 65
showed good oral bioavailability with compound 65 displaying a longer halfꢀlife with a lower
total plasma clearance than compound 60 as expected from in vitro data (Table 6). Therefore
compound 65 was profiled further in a dog PK demonstrating good oral bioavailability of 67%
(Table 8).
Table 8: Pharmacokinetics in dog; formulation was in methyl cellulose (MC) 0.5%
IV
PO
1
mg/kg
5 mg/kg
Compound
C0 or Cmax (ng/mL)
Tmax (h)
65
1,143
_
1,807
1.5
AUC(0ꢀz) (ng.h/mL)
Cl (L/h/kg)
4,098
0.25
1.7
13,908
_
Vss (L/kg)
_
T1/2 (h)
7.5
5.2
F (%)
67
2
2
ACS Paragon Plus Environment

Page 23 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The biochemical potency of the described compounds was measured at the Km of ATP
whereupon compound 65 showed only a selectivity of 2.8 fold against JAK2, lower than for
2
2,23
recently reported JAK1 selective inhibitors.
Discrepancies between enzyme and cellular data
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
for JAK kinases has been recognized for some time, but only recently have potential
2
4
explanations supported by experimentation been published to address these inconsistencies. A
first explanation stipulates that in the cell, the concentration of ATP is accepted to be in the 1 to
5
mM range. Consequently, neither a K nor an IC measurement at ATP Km in an enzymatic
i 50
3
assay represents a good evaluation of cellular activity for ATP competitive inhibitors. The
second explanation is due to the role of the different JAK within the cells. In recent years the
6
,11
importance of the JAKs in the heterodimeric pair has been the source of intensive research.
Thus different groups have demonstrated the hierarchical role of the JAKs in the cytokine
pathway, in particularly the dominant role of JAK1 in the IL2, IL22 or IL6 pathways. These later
observations show that different selectivity profiles of inhibitors is obtained when looking at
different cytokine pathways. Finally, shifts in biochemical potency and cellular potency between
JAK2 and JAK1 have since been reported on other compound series by a number of research
2
3,24,25
groups.
Consequently, to assess the profile of compound 65 in potentially more relevant cellular settings,
its potency and selectivity was determined in cellular and whole blood assays. Cellular assays
were developed for cytokines employing different combinations of JAK heteroꢀ and homoꢀ
dimers for signaling.
Cell lines were preꢀincubated with 65 and treated with cytokines that employ different JAK
heterodimeric or JAK2 homodimeric complexes for signaling (Table 9). Compound 65 inhibited
JAK1/JAK3/γ and JAK1/TYK2 signaling pathways most potently (IC values from 150 to 760
c
50
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 70
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
nM), whereas JAK1/JAK2 signaling mediated by type II and gp130 receptor complexes or JAK2
signaling were inhibited with low micromolar potencies. Compound 65 does not inhibit JAK2
homodimer–mediated signaling induced by EPO or PRL (IC > 10 µM). Due to the selectivity
5
0
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
of compound 65 against JAK3 and TYK2 in the biochemical assay and precedent literature that
6
,11
showed JAK1 dominance,
it can be concluded that inhibition of JAK/STAT signaling
involving JAK1, JAK3 and TYK2 is certainly due to JAK1 inhibition. Hence, compound 65
preferentially inhibits JAK/STAT signaling involving JAK1 over JAK2 kinase in a cellular
context. This observation was recently confirmed elsewhere with a selectivity of >14 reported for
compound 65 in cell assays using ILꢀ6/pSTAT1 for JAK1 pathway and EPO/pSTAT5 for
3
JAK2.
8,26
Table 9: Cell assays employing different JAK heterodimeric or homodimeric complex.
Copyright 2013. The
American Association of Immunologists, Inc.
JAK involved
Cell type
Trigger
Read-out
IC50, nM (pIC50 ± s.e.m)
JAK1ꢀJAK3
JAK1ꢀJAK3
JAK1ꢀTYK2
JAK1ꢀJAK2
JAK1ꢀJAK2
JAK2
THPꢀ1
NKꢀ92
U2OS
HeLa
U2OS
TFꢀ1
ILꢀ4
ILꢀ2
pSTAT6
pSTAT5
154, 203
148, 757, 367
494, 436
1,045
IFNꢀαB2
OSM
INFγ
pSTAT1
STAT1 reporter
pSTAT1
3,364
ILꢀ3
pSTAT5
3,524
JAK2
BaF3
UT7ꢀ
EPO
ILꢀ3
Proliferation
4,546
JAK2
JAK2
EPO
PRL
pSTAT5
pSTAT5
>10,000
>10,000
22Rv1
2
4
ACS Paragon Plus Environment

Page 25 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Human blood was preꢀincubated with the different compounds, and ILꢀ6 or GMꢀCSF was added
to stimulate respectively JAK1 or JAK2 signaling pathways (Table 10). STAT1 and STAT5
+
+
phosphorylation levels respectively were measured in CD4 lymphocytes and CD33 monocytes.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 10. Human whole blood potency on JAK1 and JAK2 signaling pathway. Compound 65 displayed a high
selectivity for JAK1 over JAK2 in a cellular environment
IL6-induced pSTAT1
IC (nM)
GM-CSF-induced pSTAT5
Compounds
IC (nM)
50
50
6
6
6
5
0
2
629 (n =7)
5,590 (n=3)
4,467 (n=3)
6,152 (n=3)
2,749 (n=3)
74 (n =16)
17,453 (n =7)
>30,000 (n=3)
>30,000 (n=3)
64
63
1
>30,000 (n=3)
>10,000 (n=2), >30,000 (n=1)
740 (n=7)
2
1,053 (n=5)
4,982 (n=2)
Once again, in this setting compound 65 showed a higher selectivity (> 27 fold) for JAK1 over
JAK2 (Table 10). We also confirmed that compound 65 was the most potent compound on JAK1
signaling within the series of compounds with improved ADME properties (60, 62, 63 and 64
IC > 2,000). In the same assay compounds 1 and 2 do not reach the same level of selectivity as
5
0
compound 65, with 65 being 3 fold more selective than 1. This key finding provided optimism
for the first time that JAK1 selectivity could be achieved in vivo within this series. Remarkably,
compound 2 only showed a weak potency in this assay, being 2 fold less potent than compound
65, which was not predicted from its biochemical data. This apparent discrepancy was also
3
shown elsewhere.
2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 70
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
8
Extensive ADME profiling of 65 : The in vitro ADME and safety pharmacology properties of
compound 65 were assessed. Metabolic stability in vitro is favorable with 65%, 45% and 87% of
compound left respectively in rat, dog and human liver microsomes after 60 min incubation and
a halfꢀlife of more than 200 min in hepatocytes for all species. Permeability in a Cacoꢀ2 assay
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
was found to be 3.5 10 cm/s (Papp (A2B)) with an efflux ratio of 16 indicating that 65 is a
substrate of Pꢀgp (permeability increased in presence of verapamil, Pꢀgp inhibitor). However,
despite the high efflux, good solubility of GLPG0634 allows high absorption in animal species
(Table 7 and Table 8). Thermodynamic solubilities of the compound in aqueous buffer solutions
of pH 3 and pH 7.4 are 243 µg/mL and 40 µg/mL respectively. Plasma protein bindings are
4
7.5% in rat, 28.9% in dog and 31.8% in human. An initial assessment of potential
cardiovascular toxicity was performed using a hERG patch clamp assay, in which compound 65
displayed a safe profile (IC50 >100 ꢁM). Inhibition of the major cytochrome P450 (CYP)
isosymes (1A2, 2C9, 2C19, 2D6, and 3A4) is notably low with an IC50 >70 ꢁM for each
CYP450 tested.
Kinase profiling: Following the identification of 65 as a lead molecule, in vitro selectivity was
studied extensively. The compound was tested at 1 µM against a panel of 170 kinases. Of the
1
70 kinases tested, only weak inhibition was observed for hFLT3, hFLT4 and hCSF1R, with
IC ’s of 338 nM, 274 nM and 489 nM, respectively. No offꢀtarget liabilities were identified
5
0
upon screening a diverse panel of 99 GPCRs, ion channels, transporters and nonꢀkinase enzymes
data not shown).
(
Crystallography
To explore the binding mode of our selected inhibitor to the enzyme, 65 was coꢀcrystallized with
JAK2 JH1 domain crystals, as JAK1 coꢀcrystal was not readily available at that time and since
2
6
ACS Paragon Plus Environment

Page 27 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
our modeling predicted a similar binding mode. As shown in Figure 4, 65 binds to the ATP
binding pocket of JAK2 with the amino triazolopyridine core interacting with the hinge region of
the kinase. The triazolopyridine ring is positioned against the hinge, and forms via the exocyclic
NH and the triazolo nitrogen N3 two hydrogen bonds with the main chain atoms of Leu932. The
phenyl ring sits within a hydrophobic environment. The terminal thiomorpholine dioxide group
packs against the glycine rich loop, forming polar interactions with main chain atoms of this
flexible loop (Gly861, Ser862), the side chain of Val863, the catalytic Lys882 and Asp994 of the
DFG segment. This same binding should occur in all JAK ATP pockets and may explain the low
selectivity observed in biochemical assay. In conclusion, the crystal structure does not explain
the high selectivity observed in the whole blood assay.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4: Crystal structure of JAK2 with 65; A) Crystal structure of JAK2 with compound 65; B) Detailed view of
the Xꢀray crystal structure of compound 65 bound to the ATPꢀbinding site of JAK2. Presumed hydrogen bonds
involving compound 65 are represented as yellow lines with interatomic distances indicated in angstroms (PDB:
4P7E)
B
A
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 70
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
8
In vivo efficacy assessment: The rat collagenꢀinduced arthritis (CIA) model was used to assess
the potential therapeutic utility of 65 for the treatment of autoimmune arthritis, as it was shown
2
3
previously that selective JAK1 inhibitor can be efficacious. We measured disease progression
following daily oral administration of 65 at doses of 0.1, 0.3, 1, 3, 10 and 30 mg/kg in 3 separate
studies. Disease severity was assessed periodically, scoring clinical signs of disease and
comparing to a vehicle and an etanerceptꢀtreated group as positive control (Enbrel® 10 mg/kg, 3
times/week). Data obtained from the 3 studies were normalized to the vehicle data for each
experiment, allowing a metaꢀanalysis of the 3 studies (Figure 5). A doseꢀdependent effect was
observed in all readouts. A statistically significant reduction of the clinical score was measured,
surprisingly even at doses as low as 0.1 mg/kg and 0.3 mg/kg.
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
8,27
Figure 5. CIA rat model. Difference between the clinical score at day 0 and day 15 (Meta-analysis of combined
data obtained for the three studies after 15 d of treatment. E, etanercept; V, vehicle. 65 doses are indicated on the x-
axis. The numbers of animals used to obtain a meta-analysis score for V, E, 0.1, 0.3, 1, 3, 10, and 30 mg/kg were 29,
1
0, 20, 20, 29, 20, 10, and 29, respectively. *p , 0.05, **p , 0.01, ***p , 0.001 versus vehicle, Student t test).
Copyright 2013. The American Association of Immunologists, Inc.
Only at 10 mg/kg the concentration of 65 in blood covered whole blood IC50 for almost 8 hrs
(
Figure 6). At doses up to 3 mg/kg, the in vivo exposure of the compound barely reaches the IC50
2
8
ACS Paragon Plus Environment

Page 29 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
in the whole blood assay (ILꢀ6/pSTAT1). Note that the potency of in the rat biochemical assay is
in the same range as human (29 nM ± 8.9)). In particular, showing efficacy for 65 in the rat CIA
model at doses as low as 0.1 and 0.3 mg/kg, the latter confirmed in two separate experiments,
was surprising. It is difficult to understand the strong efficacy of the compound at these low
doses. As discussed by Van Rompaey et al. we cannot provide a definitive explanation for the
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
8
exceptional efficacy of compound 65. One can surmise that the very low protein binding of 65
allows for a rapid and extensive penetration into tissues, including the inflamed joints. FLT3 and
FLT4 are the closest offꢀtarget kinases and their inhibition could contribute to the antiꢀ
inflammatory effects, but the >25 fold selectivity for JAK1 inhibition does not support this
8
hypothesis. The compound is also known to form an active metabolite (compound 66, Scheme
2
8
6). This metabolite is formed by simple hydrolysis of the electron deficient aniline in position 2
of the triazoloꢀpyridine and was found to be less potent than compound 65, while retaining a
good JAK1 selectivity. The PK profile of 66 in humans show a high and sustained exposure, well
exceeding the exposure of parent 65, and this metabolite is understood to contribute to the
28
overall clinical efficacy. However, at doses below 3 mg/kg in rats, plasma levels would still
remain well below its IC for JAK1 inhibition. Thus, none of the current hypotheses are truly
50
satisfactory, and future work will need to identify more likely explanations.
Figure 6: Steady state PK (day 9) of 65 in rat CIA. WBA IC pSAT1/IL6 and GM_CSF/pSTAT5 are indicated
5
0
2
9
ACS Paragon Plus Environment