Strategic use of conformational bias and structure based design to identify potent JAK3 inhibitors with improved selectivity against the JAK family and the kinome

Article abstract of DOI:10.1016/j.bmcl.2013.02.012

Using a structure based design approach we have identified a series of indazole substituted pyrrolopyrazines, which are potent inhibitors of JAK3. Intramolecular electronic repulsion was used as a strategy to induce a strong conformational bias within the ligand. Compounds bearing this conformation participated in a favorable hydrophobic interaction with a cysteine residue in the JAK3 binding pocket, which imparted high selectivity versus the kinome and improved selectivity within the JAK family.

Full text of DOI:10.1016/j.bmcl.2013.02.012

Bioorganic & Medicinal Chemistry Letters 23 (2013) 2793–2800
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Strategic use of conformational bias and structure based design
to identify potent JAK3 inhibitors with improved selectivity against
the JAK family and the kinome
a
a
a
b
Stephen M. Lynch a, , Javier DeVicente , Johannes C. Hermann , Saul Jaime-Figueroa , Sue Jin ,
Andreas Kuglstatter c, Hongju Li a, Allen Lovey a, John Menke b, Linghao Niu b, Vaishali Patel b, Douglas Roy a,
Michael Soth a, Sandra Steiner a, Parcharee Tivitmahaisoon a, Minh Diem Vu b, Calvin Yee a
⇑
a Discovery Chemistry, Hoffmann-La Roche, pRED, Pharma Research & Early Development, 340 Kingsland Street, Nutley, NJ 07110, United States
b Inflammation Biology, Hoffmann-La Roche, pRED, Pharma Research & Early Development, 340 Kingsland Street, Nutley, NJ 07110, United States
c Discovery Technologies, Hoffmann-La Roche, pRED, Pharma Research & Early Development, 340 Kingsland Street, Nutley, NJ 07110, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Using a structure based design approach we have identified a series of indazole substituted pyrrolopyr-
azines, which are potent inhibitors of JAK3. Intramolecular electronic repulsion was used as a strategy to
induce a strong conformational bias within the ligand. Compounds bearing this conformation partici-
pated in a favorable hydrophobic interaction with a cysteine residue in the JAK3 binding pocket, which
imparted high selectivity versus the kinome and improved selectivity within the JAK family.
Ó 2013 Published by Elsevier Ltd.
Received 20 November 2012
Revised 24 January 2013
Accepted 1 February 2013
Available online 13 February 2013
Keywords:
Kinase inhibitors
Janus kinase
JAK
Structure based design
Kinase selectivity
Conformational bias
Janus kinase 3 (JAK3) is a critical component in the pathway of
all cytokines that use the common chain receptor for signal
chemotypes.12,13 As biological exploration of these molecules has
progressed, a certain level of disagreement has emerged in the lit-
erature as to consequences of selective JAK3 inhibition.14 While
one study has concluded that JAK1 inhibition is not required for
efficacy in mouse models of inflammatory disease, another group
has questioned whether selective inhibition of JAK3 over JAK1 will
effectively modulate immunologically relevant pathways.15–17 Re-
sults from our in-house JAK inhibitor program, which targeted
c
transduction and its deficiency has been associated with the severe
combined immunodeficiency (SCID) phenotype.1–3 JAK3 is known
to operate in concert with JAK1 in order to facilitate phosphoryla-
tion of signal transducers and activators of transcription proteins
(STATs) and propagate cellular signaling.4,5 Recent clinical trial
data for the JAK inhibitors tofacitinib (Fig. 1, 1a) and ruxolitinib
(1b) have offered compelling proof of concept for JAK inhibition
for multiple inflammatory indications including rheumatoid
arthritis (RA)6–10 and psoriasis,9,11 but also revealed side effects
that could be attributable to inhibition of JAK2 and/or JAK1.7,8
While JAK2 and JAK1 are widely expressed, JAK3 is found predom-
inantly in cells of hematopoietic origin.1 Therefore selective inhibi-
tion of JAK3 may effectively suppress inflammatory signal
transduction while avoiding potential side effects.
N
N
N
N
N
N
CN
O
Since the JAK family of protein kinases have been clinically val-
idated as therapeutic targets, a tremendous amount of research has
been focused on identification of inhibitors derived from novel
N
N
N
H
N
N
1a
1b
⇑
Corresponding author. Tel.: +1 973 235 7315; fax: +1 973 235 6263.
E-mail address: stephen.lynch@roche.com (S.M. Lynch).
Figure 1. Structures of tofacitinib (1a) and ruxolitinib (1b).
0960-894X/$ - see front matter Ó 2013 Published by Elsevier Ltd.
http://dx.doi.org/10.1016/j.bmcl.2013.02.012

2794
S. M. Lynch et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2793–2800
Table 1
Enzyme potency, selectivity and solubility data for tofacitinib, ruxolitinib and preliminary pyrrolopyrazine front pocket groups22
O
R1
CN
2
1
3
N
HN
N
H
R2
=
O
N
Cl
O
N
N
N
N 4
R2
7
O
O
5
6
2
3
4
5
R1
IC50 (nM)
JAK2
Selectivity
LYSAb
(lg/mL)
a
Compound
JAK3
JAK1
JAK2/JAK3
JAK1/JAK3
1a
1b
N/A
N/A
2.2 0.6
10.7 2.2
4.0 0.7
<0.3
1.6 0.1
0.8 0.2
2
<1
1
<1
461
ND
2a
0.3 0.1
8.2 1.1
0.8 0.2
3.2 1.1
3
11
33
2b
3
37
0.8 0.03
18
3.3 0.6
5
47 16
5
2
3
4
6
18
4
277
ND
366
<1
0.4 0.02
5.2 1.4
0.8 0.1
7.4 0.2
4
4
23
4
5
1.0 0.2
1
a
Mean SEM (standard error of the mean), n P 3 except for 3 (n = 2).
LYSA = lyophilized solubility assay, ND = not determined.
b
inflammatory indications, suggested that compounds with im-
proved selectivity for JAK3 over JAK1 can inhibit IL-2 signaling
(which is dependent on JAK3 and JAK1) with similar potencies as
less selective analogs.18
Our JAK inhibitor program relied upon a structure based ap-
proach centered around a pyrrolopyrazine hinge binding motif in
which strategic substitutions at the 3-position were found to im-
pact the observed selectivity. This exercise culminated in the iden-
tification of potent JAK family inhibitors such as the
pyrrolopyrazine bisamides 2 (Table 1) which possessed good to
moderate selectivity for JAK3.18 Ligands of this type were found
to impart a degree of selectivity by exploiting the unique vectors
found in the bisamide sidechain to satisfy size-sensitive hydropho-
bic interactions in the upper and lower portions of the JAK binding
pocket.
In thinking about how to further increase both the JAK family
and overall pan-kinase selectivity within this promising scaffold
we decided to focus on specifically targeting position 909/936,
Figure 2. Superimposed crystal structures of Jak3 (carbon atoms and cartoon colored in green, PDB Code 1YVJ)19 and Jak2 (carbon atoms and cartoon colored in orange, PDB
code 3JY9).20 Oxygen atoms are colored in red, nitrogen atoms in blue and sulfur atoms are colored in yellow. Residue labeling corresponds to Jak2 and Jak3, respectively. All
visualizations were captured with MOE.21

S. M. Lynch et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2793–2800
2795
160
120
80
CN
O
CN
O
N
N
N
HN
N
N
HN
N
H
H
O
O
N
N
Cl
40
N
N
5-A
5-B
0
D
S
E
N
T
A
C
P
R
Q
V
L
H
Cl
Amino Acid
CN
O
CN
O
N
N
Figure 3. Prevalence of aligned amino acid residues at position 909/936 across the
kinome (490 kinases). Red denotes polar, hydrophilic residues; blue denotes
hydrophobic residues.
HN
N
N
HN
N
N
H
H
O
O
N
N
N
N
Cl
N
N
one of the two adenosine triphosphate (ATP)-binding site residues
which are not conserved among JAK1, JAK2, and JAK3 (Fig. 2).
While JAK3 has a cysteine in position 909, both JAK1 and JAK2 pos-
sess a more polar serine moiety at the corresponding position 936.
We hypothesized that introduction of a hydrophobic substituent in
this region of the binding pocket would create a favorable interac-
tion with Cys909 but would negatively interact with Ser936 thus
increasing our JAK family selectivity. Looking across the kinome,
position 909/936 is typically occupied by polar residues such as
aspartate, glutamate, and serine (Fig. 3). With the exception of ser-
ine, most amino acids occupying this location are more bulky than
cysteine. Therefore, the combination of hydrophobic and steric
interactions at this position might offer improvement in pan-ki-
nase selectivity as well.
Our lead molecule 2a possessed a small cyclopropyl group in
the front portion of the binding pocket most close in proximity
to Cys909. Our previous experience told us that analogs containing
flat, aryl groups in this region such as 3 were very potent JAK3
inhibitors, but poor physicochemical properties coupled with ob-
served JAK2 promiscuity rendered them unattractive. In an effort
to identify more drug-like molecules with defined vectors for
growth we became interested in small, polar heterocycles. For in-
stance, pyrazole 4 maintained modest JAK family selectivity and
offered a potential vector from N1 for interaction with cysteine.
We also explored biaryl systems such as indole 5 which, although
potent, did not provide the desired levels of selectivity (Table 1).
According to our molecular modeling, these compounds—in
particular 5—showed compelling potential to interact with
Cys909. As a possible explanation for the observed lack of selectiv-
ity, we reasoned that free rotation around the front pocket
aryl–aryl bond of conformer 5-A could result in an equally potent
conformer 5-B (Fig. 4). This latter conformer would orient the
6a-A
6a-B
Cl
Figure 4. Conformational preferences of 5 and 6a.
indole away from Cys909 and toward solvent where it could be
accommodated equally well by all members of the JAK family.
One strategy for limiting this undesired rotation was to intro-
duce a conformational constraint within the scaffold. Conforma-
tional constraint is a well-known medicinal chemistry strategy
for preorganization of a flexible ligand into a biologically active
conformation.23 Although the constraint typically takes the form
of a rigid linker, we hoped to use a strong electronic repulsion to
accomplish a similar task. For instance, the N–N lone pair repulsion
present in the indazole analog 6a of indole 5 was predicted compu-
tationally to bias the molecule toward the desired conformation A
(Fig. 4). A rotational scan of a simplified pyrazine–indazole linkage
displayed a strong preference (4.7 kcal/mol) for conformer A in
which the nitrogen lone pairs have adopted an anti-orientation
(Fig. 5). When subjected to the same rotational scan, a simplified
pyrazine–indole linkage showed no preference between conforma-
tions A and B.
We prepared indazole 6a and were gratified that this compound
showed good potency and a much improved JAK family selectivity
profile. To probe this scaffold further, analogs bearing a variety of
substitution patterns on the indazole were subsequently prepared
(Table 2). The 6-Cl (6a) clearly showed better selectivity than the
5-Cl analog (6b) or the unsubstituted indazole (6c). Polar, planar
substituents at C6 showed reasonable potency and selectivity as
evidenced by compounds 6d–f. On the other hand, substitutions
indole
indazole
8
7
6
5
4
3
2
1
0
N
N
N
X
N
A
N
Φ
B
X
N
A
B
X = C, N
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
Torsion (
Φ
)
Figure 5. Rotational scan of heteroaryl-linked pyrazines. Energies are calculated at the B3LYP/6-31+G(d,p) level with Jaguar.24 The torsion (/) is defined as the dihedral angle
between the pyrazine nitrogen and the indole/indazole 2-position. All atoms were relaxed and energy minimized freely with only / fixed at each sampled angle.

2796
S. M. Lynch et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2793–2800
Table 2
Enzyme potency, selectivity and solubility data for substituted indazoles22
R1
CN
O
N
N
H
HN
N
O
N
R2
N
N
R1
R2
IC50 (nM)
JAK2
Selectivity
LYSAb
(lg/mL)
a
Compound
JAK3
JAK1
JAK2/JAK3
JAK1/JAK3
6a
6b
6c
6d
6e
6f
6-Cl
1.1 0.2
3.0 0.1
2.9 1.4
0.6 0.3
2.3 1.3
1.8 0.5
3.1 0.3
10 0.7
1.8 0.3
22 0.2
31 0.4
20
9
28
20
8
<1
5-Cl
28
20
12
2
3
1
61
22
17
3
3
1
ND
ND
<1
H
7
6-F
20
17
9
28
22
21
3
6-CN
6-OMe
6-Cyclo-Pr
6-t-Bu
4-F,6-Cl
38 20
51 41
<1
16
15
3
4
38
10
4
2
<1
6g
6h
6i
5
ND
ND
<1
13 1.6
56
12 0.6
64
1
1
3
2
31
36
6j
6-Cl
6-Cl
<0.4
4.2 0.6
1.1 0.1
4.6 0.1
0.8 0.1
>11
4
>12
3
<1
6k
0.3 0.06
ND
6l
6-Cl
0.8 0.1
1.3 0.1
5.5 0.6
2
7
ND
a
Mean SEM (standard error of the mean), n P 3 except for 6b (n = 2).
LYSA = lyophilized solubility assay, ND = not determined.
b
Figure 6. Crystal structure of 6d (PDB accession number 3ZC6) in complex with Jak3 at 2.4 Å resolution.26 Carbon atoms of the protein are colored in pink and ligand carbon
atoms colored in gold. Nitrogen atoms are colored in blue, fluorine atoms in green, sulfur atoms in yellow and oxygen atoms including water molecules are colored in red.
Displayed are also water molecule oxygen atoms close to the location of Cys-909 stemming from a superimposed Jak1 crystal structure (colored magenta, PDB code 3EYG)
and a superimposed Jak2 crystal structure (colored cyan, PDB code 3FUP).27

S. M. Lynch et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2793–2800
2797
Table 3
Table 5
Enzyme potency, selectivity and solubility data for constrained heterobicyclic
Caliper profiling data for selected compounds
analogs22
CN
O
N
N
H
HN
N
O
N
R
a
Compound R
IC50 (nM)
JAK2
Selectivity LYSAb
(lg/
mL)
JAK3
10.5
JAK1
JAK2/ JAK1/
JAK3 JAK3
N
7
4
83 10 370 38
8
35
53
<1
N
Cl
F
8
0.3 0.07 5.9 0.7 16 0.6 20
39
31
ND
N
N
F
9
1.0 0.05 37
3.4 0.9 57
4
9
73 10 37
73
18
N
N
N
F
10
61
8
17
N
Cl
a
Mean SEM (standard error of the mean), n P 3.
LYSA = lyophilized solubility assay.
b
Table 4
PBMC potency and selectivity data for select compounds
a
Compound
PBMC IC50
GM-CSFc
(lM)
Selectivity
GM- IFN-
CSF/IL-
IL-2b
IFN-cc
c
/IL-
2
2
1a
1b
2a
3
0.028 0.001 0.184 0.02
0.023 0.003 0.026 0.005 0.031 0.008
0.031 0.003 0.19 0.06
0.170 0.01
7
1
6
2
3
6
1
9
5
5
0.29 0.08
0.49 0.10
0.23 0.07
>30d
0.10 0.02
0.05 0.01
0.24 0.04
0.15 0.07
0.18 0.10
0.43 0.06
0.30 0.20
1.29 0.35
0.20 0.02
0.17 0.04
>30d
0.74 0.46
0.88 0.59
>30d
5
6a
6d
6f
6i
8
1.35 0.94
5
5
9
15
2.61e
>30d
Values represent percent inhibition at 10 lM concentration (green <50%; red >95%).
4.23 1.47
10.8 2.5
2.86 0.18
12.9 7.6
14
8
10
10
9
a
Mean SEM (standard error of the mean), n P 3 except for 3, 6d, 6f, 8 and 9
(n = 2).
binding interaction. The bisamide sidechain also adopted the antic-
ipated conformation with the methyl group shallowly filling the
back lower region of the pocket and the cyanoazetidine angled up-
ward under the glycine rich loop. The indazole occupied the front
region of the pocket and displayed an approximate 15° tilt down-
ward to rest over top of Cys909 and make a favorable hydrophobic
contact. Substituents at the indazole 6-position appear to be in
closest proximity to residues 909/936 and would be expected to
displace conserved waters associated with the polar network sur-
rounding Ser936 in JAK2 and JAK1.25 While displacement of these
waters would require a larger energetic penalty for JAK2 and JAK1,
the hydrophobic cysteine interactions present in JAK3 could com-
pensate for the water loss, thus contributing to the observed selec-
tivity. Excessive steric bulk in this region presumably interferes
b
Gated on CD3 T-cells.
Gated on CD14 monocytes.
Low solubility was a potentially limiting factor for IC50 determination.
n = 1.
c
d
e
residing out-of-plane such as cyclopropyl (6g) or tert-butyl (6h)
were detrimental to potency and almost completely destroyed
selectivity. This suggested a limited amount of space in this region
of the binding pocket. An additional fluorine at C4 (6i) further im-
proved selectivity over both JAK2 and JAK1. Larger substitutions at
this position were not tolerated.
We obtained a crystal structure of 6d bound to JAK3 (Fig. 6). The
structure showed the pyrrolopyrazine making the expected hinge

2798
S. M. Lynch et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2793–2800
with optimal cysteine hydrophobic contact causing a loss in JAK3
potency. The 4-fluorine present in analog 6i likely favors a non-pla-
nar orientation of the indazole sidechain relative to the pyrrolopyr-
azine core, thus enforcing a downward tilt of the indazole towards
Cys909.
improved solubility compared to the indazoles 6 and 7 as mea-
sured by the LYSA (lyophilized solubility) assay.
In order to assess the JAK family selectivity of these compounds
in a more biologically relevant system, select examples were tested
in a cellular functional assay. This assay measured inhibition of
phosphorylation of downstream STAT proteins in peripheral blood
mononuclear cells (PBMCs) upon treatment with varying stimuli.
We knew from our previous investigations that placement of
large, hydrophobic substituents in the back, lower binding pocket
generally increased both potency and selectivity toward JAK3
(compare 2a and 2b). To examine possible synergy between the
front pocket indazole and the lower pocket group, analogs 6j–l
bearing progressively larger R1 substituents were prepared.
Although these three compounds were exceedingly potent toward
JAK3, a dramatic decrease in JAK family selectivity was observed
suggesting an intricate interplay between the indazole in the front
pocket and the alkyl lower pocket group. Steric clash between the
indazole and bulky lower pocket groups may alter the torsion of
the indazole–pyrrolopyrazine bond, tilting the indazole away from
position 909/936. This new conformation might no longer disfavor
JAK2 and JAK1 and account for the observed increase in potency to-
ward both of these enzymes. Presumably the lower pocket methyl
group is optimal for both shallow space filling and allowing an
ideal indazole torsion.
Thus, stimulation with IL-2, GM-CSF, or IFN-c induces JAK3/1,
JAK2, or JAK2/1 mediated phosphorylation of STAT5a, STAT5a, or
STAT1, respectively (Table 4). Initial lead compound 2a showed
good potency in these assays with preference observed for inhibi-
tion of IL-2 signaling versus GM-CSF and IFN-c signaling. Similar
cellular potencies and selectivities were observed for tofacitinib
1a, and slightly lower cellular selectivities for trimethoxyphenyl
analog 3 and indole analog 5. In contrast, ruxolitinib 1b, while po-
tent in our cellular assays, showed no cellular selectivities. Ruxolit-
inib 1b is a JAK1/2 inhibitor with a relatively large selectivity
window (>10-fold) versus JAK3 in our enzyme assays, while 1a,
2a, 3, and 5 are all pan-JAK or modestly JAK3-selective inhibitors.
We therefore conclude that more potent JAK3 inhibition relative
to JAK1 and/or JAK2 inhibition can improve selectivity for inhibi-
tion of our IL-2 cellular assay versus our GM-CSF and IFN-c assays.
In light of the successful results obtained with the indazole
front pocket group we decided to explore several additional het-
erocycles at this position (Table 3). In principal, any 5,6-heterobi-
The effects of further improvements in JAK3 versus JAK2 and
JAK1 enzyme selectivities upon cellular selectivities are subtle
and equivocal. While the results for indazoles 6a and 6i suggest
a greatly improved cellular selectivity in our assays, the more mod-
est cellular selectivities observed for other compounds with similar
or better JAK3 enzyme selectivities temper that initial conclusion.
The indazoles in general exhibited poor solubilities (Table 2), and
this issue may have been a limiting factor for accurate IC50 deter-
cyclic system in the front pocket could impart
a similar
conformational bias, however diversified distribution of electronics
in the region of Cys909/Ser909 might be a factor in the overall JAK
family selectivity. Although less potent, N-1 linked indazoles such
as 7 were determined to be JAK3 selective inhibitors. Of greater
interest were the imidazo[1,5]pyridines such as 8 and 9. The di-
fluoro analog 9 was particularly selective within this series. Of note
was the imidazo[1,2]pyridine 10 which was also observed to have
reasonable potency and good JAK3 selectivity even though it
lacked the key nitrogen ortho to the biaryl linkage. Rotational scan
of this scaffold (as described in Fig. 5) reinforced this result by pre-
dicting a >4 kcal/mol energy difference favoring the desired confor-
mation. One benefit of the imidazo[1,5] pyridines 8 and 9 was their
mination for 6a and 6i in the GM-CSF and IFN-c assays. However,
we were able to determine definitive IC50 curves for all other
examples in Table 4 and will focus our discussion on those
compounds.
All of the JAK3 versus JAK2 and JAK1 enzyme selective examples
show cellular selectivities inhibiting IL-2 versus GM-CSF and IFN-
c
signaling, and the overall trend appears to be that improving
selectivity for inhibition of JAK3 versus JAK2 and JAK1 results in
Figure 7. Kinomescan dendrogram of 6a versus 442 kinases at 10 lM. JAK3 is shaded in blue.

S. M. Lynch et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2793–2800
2799
an improved cellular selectivity for inhibiting IL-2 versus GM-CSF
and IFN- signaling. For example, the most JAK3-selective exam-
pocket alkyl substituent not only for JAK family selectivity but
for pan-kinase selectivity as well. Imidazo[1,5]pyridine 9 also
showed a reasonable selectivity profile against this panel of ki-
nases, which confirms the observed trend with a more soluble ana-
log. For more comprehensive profiling 6a was assayed using the
c
ples, compounds 8 and 9, demonstrate approximately 10-fold im-
proved cellular selectivities compared to the least JAK3-selective
example, the JAK1/2 inhibitor ruxolitinib 1b. However, it is difficult
to define an exact relationship between enzyme and cellular selec-
tivities; for example, the cellular selectivities observed for indazole
6d are similar to those seen for the less enzyme-selective examples
2a and tofacitinib 1a. This continuing uncertainty underscores the
complexity of JAK/STAT signaling.
To evaluate kinome selectivity, we profiled compounds by Cal-
iper screening against a 48 kinase panel. Caliper profiling measures
percent inhibition of phosphorylation of a peptide substrate.28 The
data in Table 5 is represented by a heat map, assigning <50% inhi-
bition as completely green and >95% inhibition as completely red.
The Caliper data clearly indicate that indazoles 6a and 6i demon-
strated greater pan-kinase selectivity than indole 5. Indazoles 6j
and 6k, which possessed larger lower pocket functionality lost
selectivity versus the kinome. This result highlights the signifi-
cance of interplay between the front pocket and lower
Kinomescan platform.29 Data generated at 10
lM showed that this
compound was quite selective against a panel of 442 wild-type ki-
nases, displacing >90% of ligand in only 29 cases (Fig. 7). Kd values
against the kinase domains of JAK3 (1.4 nM), JAK2 (11 nM), JAK1
(73 nM), and Tyk2 (350 nM) were also determined for 6a. Although
these Kds are not exactly correlated with our in-house results, they
overall confirm our observed JAK family selectivities and also show
that this ligand discriminates against Tyk2, the final member of the
JAK family.
The indazole substituted pyrrolopyrazines described were syn-
thesized using the convergent approach outlined in Scheme 1.
Amine salts 12 were prepared by coupling of the appropriate
Boc-protected amino acid with 4-cyanoazetidine followed by Boc
deprotection. The SEM protected pyrrolopyrazine acid 11 was then
coupled to the amine salt 12 to provide bromine containing
R1
CN
O
O
R1
CN
N
O
SEM
SEM
OH
N
N
H
N
a
N
+
H2N
R1
CN
O
N
N
N
N
O
TFA
N
N
H
HN
N
Br
Br
O
e, f
11
14
12
13
N
R2
I
Bu3Sn
N
N
b, c
d
N
R2
R2
R2
N
N
N
N
N
6a-l
H
16
15
Scheme 1. Synthesis of indazoles 6a–l. Reagents and conditions: (a) HATU, i-Pr2NEt, DMF, rt; (b) I2, KOH, DMF, rt; (c) KOt-Bu, MeI, 0 °C to rt; (d) i-PrMgCl, THF, ꢀ10 °C then
Bu3SnCl, ꢀ10 °C to rt; (e) Pd(PPh3)4, CuI, DMF, 85 °C, 2 h; (f) TFA, CH2Cl2 then ethylenediamine, CH2Cl2.
CN
O
CN
N
O
N
H
N
H
O
N
SEM
O
O
N
H
N
B
N
N
a-c
O
+
N
N
N
Cl
Br
Cl
Boc
N
17
18
5
d,e,c
CN
O
N
N
H
N
H
N
O
N
N
N
Cl
7
Scheme 2. Synthesis of indole 5 and indazole 7. Reagents and conditions: (a) Pd(PPh3)4, Na2CO3, DME, H2O, 90 °C, overnight, 52%; (b) NaH, DMF, 0 °C then MeI, 61%; (c) TFA,
CH2Cl2 then ethylenediamine, CH2Cl2, 58–63%; (d) CuI, NaI, trans-N,N0-dimethylcyclohexane-1,2-diamine, 1,4-dioxane, 110 °C, 48 h; (e) CuI, K3PO4, trans-N,N0-dimethyl-
cyclohexane-1,2-diamine, 5-chloro-1H-indazole, toluene, 110 °C, 24 h, 58% (2 steps).

2800
S. M. Lynch et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2793–2800
N
References and notes
a, b
c
HN
R
R
N
R
1. O’Shea, J. J.; Pesu, M.; Borie, D. C.; Changelian, P. S. Nat. Rev. Drug Disc. 2004, 3,
555.
N
N
N
H
O
2. Macchi, P.; Villa, A.; Giliani, S.; Sacco, M. G.; Frattini, A.; Porta, F.; Ugazio, A. G.;
Johnston, J. A.; Candotti, F.; O’Shea, J. J.; Vezzoni, P.; Notarangelo, L. D. Nature
1995, 377, 65.
3. Russell, S. M.; Tayebi, N.; Nakajima, H.; Riedy, M. C.; Roberts, J. L.; Aman, M. J.;
Migone, T. S.; Noguchi, M.; Markert, M. L.; Buckley, R. H.; O’Shea, J. J.; Leonard,
W. J. Science 1995, 270, 797.
19
20
21
I
Bu3Sn
N
d
e
N
R
R
N
N
22
23
4. Ghoreschi, K.; Laurence, A.; O’Shea, J. J. Immunol. Rev. 2009, 228, 273.
5. Pesu, M.; Laurence, A.; Kishore, N.; Zwillich, S. H.; Chan, G.; O’Shea, J. J.
Immunol. Rev. 2008, 223, 132.
6. Flanagan, M. E.; Blumenkopf, T. A.; Brissette, W. H.; Brown, M. F.; Casavant, J.
M.; Chang, S.-P.; Doty, J. L.; Elliott, E. A.; Fisher, M. B.; Hines, M.; Kent, C.;
Kudlacz, E. M.; Lillie, B. M.; Magnuson, K. S.; McCurdy, S. P.; Munchhof, M. J.;
Perry, B. D.; Sawyer, P. S.; Strelevitz, T. J.; Subramanyam, C.; Sun, J.; Whipple, D.
A.; Changelian, P. S. J. Med. Chem. 2010, 53, 8468.
I
Bu3Sn
f
N
e
N
N
N
N
Cl
N
Cl
Cl
7. Riese, R. J.; Krishnaswami, S.; Kremer, J. Best Pract. Res. Clin. Rheumatol. 2010,
24, 513.
24
25
26
8. Kremer, J. M.; Bloom, B. J.; Breedveld, F. C.; Coombs, J. H.; Fletcher, M. P.;
Gruben, D.; Krishnaswami, S.; Burgos-Vargas, R.; Wilkinson, B.; Zerbini, C. A.;
Zwillich, S. H. Arthritis Rheum. 2009, 60, 1895.
9. West, K. Curr. Opin. Invest. Drugs 2009, 10, 491.
10. Incyte’s JAK inhibitor demonstrates rapid and marked clinical improvement in
rheumatoid arthritis patients. Incyte Corporation press release, October 26,
Scheme 3. Synthesis of imidazo[1,5]pyridinyl- and imidazo[1,2]pyridinyl stann-
anes. Reagents and conditions: (a) H2 (45 psi), Pd/C, concd HCl, EtOH, 3.5 h; (b)
HCO2H, 110 °C, overnight; (c) POCl3, toluene, 110 °C, 24–29% (3 steps); (d) I2,
NaHCO3, EtOH/H2O, rt, 27–38%; (e) i-PrMgCl, THF, ꢀ10 °C then Bu3SnCl, ꢀ10 °C to
rt, 83–90%; (f) n-BuLi, THF, ꢀ50 °C then N-iodosuccinimide, ꢀ50 to 0 °C.
2008.
http://investor.incyte.com/phoenix.zhtml?c=69764&p=irol-
newsArticle&ID=1217246&highlight (accessed January 23, 2013).
11. Punwani, N.; Scherle, P.; Flores, R.; Shi, J.; Liang, J.; Yeleswaram, S.; Levy, R.;
Williams, W.; Gottlieb, A. J. Am. Acad. Dermatol. 2012, 67, 658.
12. Wilson, L. J. Expert Opin. Ther. Pat. 2010, 20, 609.
13. Wrobleski, S. T.; Pitts, W. J. Ann. Rep. Med. Chem. 2009, 44, 247.
14. Ghoreschi, K.; Laurence, A.; O’Shea, J. J. Nat. Immunol. 2009, 10, 356.
15. Lin, T. H.; Hegen, M.; Quadros, E.; Nickerson-Nutter, C. L.; Appell, K. C.; Cole, A.
G.; Shao, Y.; Tam, S.; Ohlmeyer, M.; Wang, B.; Goodwin, D. G.; Kimble, E. F.;
Quintero, J.; Gao, M.; Symanowicz, P.; Wrocklage, C.; Lussier, J.; Schelling, S. H.;
Hewet, A. G.; Xuan, D.; Krykbaev, R.; Togias, J.; Xu, X.; Harrison, R.; Mansour, T.;
Collins, M.; Clark, J. D.; Webb, M. L.; Seidl, K. J. Arthritis Rheum. 2010, 62(8),
2283.
bisamide 13. Substituted indazoles 14 were iodinated and selec-
tively methylated at N-1 under basic conditions (N1:N2 ratio
ꢁ3:1). Grignard exchange at low temperature followed by quench-
ing with tributyltin chloride provided the stannanes 16. Stille cou-
pling and a two-stage SEM deprotection afforded the target
molecules 6a–l.
Indole 5 was prepared as described in Scheme 2. Suzuki cou-
pling of bromide 17 and indole boronic ester 18 proceeded with
concomitant loss of the Boc protecting group. Selective N-methyl-
ation of the indole nitrogen followed by SEM deprotection afforded
5. The N1-linked indazole derivative 7 was prepared from bromide
17 using the copper catalyzed halide exchange/N-arylation se-
quence developed by Buchwald.30
Imidazo[1,5]pyridines were synthesized according to the route
illustrated in Scheme 3. Hydrogenation of 2-cyanopyridines 19 fol-
lowed by formylation in refluxing formic acid afforded the for-
mates 20. Treatment with phosphorous oxychloride followed by
heating induced cyclization to the appropriately substituted imi-
dazo[1,5]pyridines 21 in moderate yield for the three step se-
quence. Iodination followed by Grignard exchange and quenching
with tributyltin chloride provided the stannanes 23 required for
synthesis of 8 and 9 through Stille coupling and deprotection as de-
scribed above. Likewise, commercially available 7-chloroimi-
dazo[1,2]pyridine 24 was converted to 10 via the stannane 26.
In conclusion, using a structure based approach, we have dis-
covered a potent series of indazole-substituted pyrrolopyrazine
JAK3 inhibitors which show excellent pan-kinase and improved
JAK family selectivity at an enzyme level. We attribute our ob-
served selectivity to hydrophobic interactions with a cysteine res-
idue rare to the JAK3 binding pocket. Electronic repulsion was used
as a tool to induce conformational bias within the ligand which
favored the desired hydrophobic interaction and imparted the ob-
served improvement in kinase selectivity. This strategy was found
to be general in nature and was subsequently applied to additional
preclinical kinase inhibitor discovery programs.
16. Thoma, G.; Nuninger, F.; Falchetto, R.; Hermes, E.; Tavares, G. A.;
Vangrevelinghe, E.; Zerwes, H.-G. J. Med. Chem. 2011, 54, 284.
17. Haan, C.; Rolvering, C.; Raulf, F.; Kapp, M.; Druckes, P.; Thoma, G.; Behrmann, I.;
Zerwes, H.-G. Chem. Biol. 2011, 18, 314.
18. Soth, M.; Hermann, J.; Yee, C.; Alam, M.; Barnett, J. W.; Berry, P.; Browner, M. F.;
Harris, S.; Hu, D.-Q.; Jaime-Figueroa, S.; Jahangir, A.; Jin, S.; Frank, K.;
Frauchiger, S.; Hamilton, S.; He, Y.; Hendricks, T.; Hilgenkamp, R.; Ho, H.;
Kutach, A. K.; Hekmat-Nejad, M.; Henningsen, R.; Hoffman, A.; Hsu, J.; Itano, A.;
Kuglstatter, A.; Liao, C.; Lynch, S.; Menke, J.; Niu, L.; Patel, V.; Railkar, A.; Roy,
D.; Shao, A.; Shaw, D.; Steiner, S.; Sun, Y.; Tan, S.-L.; Wang, S.; Vu, M. D. J. Med.
Chem. 2013, 56, 345.
19. Boggon, T. J.; Li, Y.; Manley, P. W.; Eck, M. J. Blood 2005, 106, 996.
20. Wang, T.; Duffy, J. P.; Wang, J.; Halas, S.; Salituro, F. G.; Pierce, A. C.; Zuccola, H.
J.; Black, J. R.; Hogan, J. K.; Jepson, S.; Shlyakter, D.; Mahajan, S.; Gu, Y.; Hoock,
T.; Wood, M.; Furey, B. F.; Frantz, J. D.; Dauffenbach, L. M.; Germann, U. A.; Fan,
B.; Namchuk, M.; Bennani, Y. L.; Ledeboer, M. W. J. Med. Chem. 2009, 52, 7938.
21. Molecular Operating Environment (MOE), 2011.10; Chemical Computing Group
Inc., 1010 Sherbrooke St. West, Suite #910, Montreal, QC, Canada H3A 2R7,
2011.
22. Enzyme assay measured inhibition of phosphorylation of
synthetic peptide catalyzed by JAK1, JAK2 or JAK3. All enzymes reactions
were run at adenosine triphosphate (ATP) concentrations of 1.5 M. Km’s of
these enzymes for ATP were determined to be 1.5 M (JAK3), 6 M (JAK2), and
20 M (JAK1).
23. Martin, S. F. Pure Appl. Chem. 2007, 79, 193.
24. Jaguar, version 7.8, Schrödinger, LLC, New York, NY, 2011
a biotinylated
l
l
l
l
25. Michel, J.; Tirado-Rives, J.; Jorgensen, W. L. J. Am. Chem. Soc. 2009, 131, 15403.
26. X-ray crystal structure determined by Proteros Biostructures GmbH
(Martinsried, Germany).
27. Williams, N. K.; Bamert, R. S.; Patel, O.; Wang, C.; Walden, P. M.; Wilks, A. F.;
Fantino, E.; Rossjohn, J.; Lucet, I. S. J. Mol. Biol. 2009, 387, 219.
28. Dunne, J.; Reardon, H.; Trinh, V.; Li, E.; Farinas, J. Assay Drug Dev. Technol. 2004,
2, 121.
29. Fabian, M. A. B.; William, H.; Treiber, D. K.; Atteridge, C. E.; Azimioara, M. D.;
Benedetti, M. G.; Carter, T. A.; Ciceri, P.; Edeen, P. T.; Floyd, M.; Ford, J. M.;
Galvin, M.; Gerlach Jay, L.; Grotzfeld, R. M.; Herrgard, S.; Insko, D. E.; Insko, M.
A.; Lai, A. G.; Lelias, J.; Mehta, S. A.; Milanov, Z. V.; Velasco, A. M.; Wodicka, L.
M.; Patel, H. K.; Zarrinkar, P. P.; Lockhart, D. J. Nat. Biotechnol. 2005, 23, 329.
30. Antilla, J. C.; Baskin, J. M.; Barder, T. E.; Buchwald, S. L. J. Org. Chem. 2004, 69,
5578.
Acknowledgment
The authors wish to thank David Goldstein for his support of
this work.