Discovery of a class of highly potent Janus Kinase 1/2 (JAK1/2) inhibitors demonstrating effective cell-based blockade of IL-13 signaling

Article abstract of DOI:10.1016/j.bmcl.2019.04.008

Disruption of interleukin-13 (IL-13) signaling with large molecule antibody therapies has shown promise in diseases of allergic inflammation. Given that IL-13 recruits several members of the Janus Kinase family (JAK1, JAK2, and TYK2) to its receptor complex, JAK inhibition may offer an alternate small molecule approach to disrupting IL-13 signaling. Herein we demonstrate that JAK1 is likely the isoform most important to IL-13 signaling. Structure-based design was then used to improve the JAK1 potency of a series of previously reported JAK2 inhibitors. The ability to impede IL-13 signaling was thereby significantly improved, with the best compounds exhibiting single digit nM IC₅₀’s in cell-based assays dependent upon IL-13 signaling. Appropriate substitution was further found to influence inhibition of a key off-target, LRRK2. Finally, the most potent compounds were found to be metabolically labile, which makes them ideal scaffolds for further development as topical agents for IL-13 mediated diseases of the lungs and skin (for example asthma and atopic dermatitis, respectively).

Full text of DOI:10.1016/j.bmcl.2019.04.008

Accepted Manuscript
Discovery of a Class of Highly Potent Janus Kinase 1/2 (JAK1/2) Inhibitors
Demonstrating Effective Cell-Based Blockade of IL-13 Signaling
Mark Zak, Emily J. Hanan, Patrick Lupardus, David G. Brown, Colin Robinson,
Michael Siu, Joseph P. Lyssikatos, F. Anthony Romero, Guiling Zhao, Terry
Kellar, Rohan Mendonca, Nicholas C. Ray, Simon C. Goodacre, Peter H.
Crackett, Neville McLean, Christopher A. Hurley, Po-wai Yuen, Yun-Xing
Cheng, Xiongcai Liu, Marya Liimatta, Pawan Bir Kohli, Jim Nonomiya, Gary
Salmon, Gerry Buckley, Julia Lloyd, Paul Gibbons, Nico Ghilardi, Jane R.
Kenny, Adam Johnson
PII:
S0960-894X(19)30217-3
https://doi.org/10.1016/j.bmcl.2019.04.008
BMCL 26375
DOI:
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
11 November 2018
27 March 2019
3 April 2019
Please cite this article as: Zak, M., Hanan, E.J., Lupardus, P., Brown, D.G., Robinson, C., Siu, M., Lyssikatos, J.P.,
Romero, F.A., Zhao, G., Kellar, T., Mendonca, R., Ray, N.C., Goodacre, S.C., Crackett, P.H., McLean, N., Hurley,
C.A., Yuen, P-w., Cheng, Y-X., Liu, X., Liimatta, M., Kohli, P.B., Nonomiya, J., Salmon, G., Buckley, G., Lloyd,
J., Gibbons, P., Ghilardi, N., Kenny, J.R., Johnson, A., Discovery of a Class of Highly Potent Janus Kinase 1/2
(JAK1/2) Inhibitors Demonstrating Effective Cell-Based Blockade of IL-13 Signaling, Bioorganic & Medicinal
Chemistry Letters (2019), doi: https://doi.org/10.1016/j.bmcl.2019.04.008
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Discovery of a Class of Highly Potent Janus Kinase 1/2 (JAK1/2)
Inhibitors Demonstrating Effective Cell-Based Blockade of IL-13
Signaling.
Mark Zaka,*, Emily J. Hanana, Patrick Lupardusa, David G. Brownb, Colin Robinsonb, Michael
Siua, Joseph P. Lyssikatosa, F. Anthony Romeroa, Guiling Zhaoa, Terry Kellara, Rohan Mendoncaa,
Nicholas C. Rayb, Simon C. Goodacreb, Peter H. Crackettb, Neville McLeanb, Christopher A.
Hurleyb, Po-wai Yuenc, Yun-Xing Chengc, Xiongcai Liuc, Marya Liimattaa, Pawan Bir Kohlia, Jim
Nonomiyaa, Gary Salmonb, Gerry Buckleyb, Julia Lloydb, Paul Gibbonsa, Nico Ghilardia, Jane R.
Kennya, Adam Johnsona.
a Genentech Inc., 1 DNA Way, South San Francisco, California 94080, USA
b
Charles River Laboratories, 8-9 Spire Green Centre, Harlow, Essex, CM19 5TR, United
Kingdom
c Pharmaron Beijing Co. Ltd., 6 Taihe Road, BDA, Beijing 100176, PR China
* Corresponding Author. E-mail address: mzak@gene.com. Phone: 650-467-4533
Abstract
Disruption of interleukin-13 (IL-13) signaling with large molecule antibody therapies has shown
promise in diseases of allergic inflammation. Given that IL-13 recruits several members of the
Janus Kinase family (JAK1, JAK2, and TYK2) to its receptor complex, JAK inhibition may offer
an alternate small molecule approach to disrupting IL-13 signaling. Herein we demonstrate that
JAK1 is likely the isoform most important to IL-13 signaling. Structure-based design was then
used to improve the JAK1 potency of a series of previously reported JAK2 inhibitors. The ability
to impede IL-13 signaling was thereby significantly improved, with the best compounds exhibiting
single digit nM IC50’s in cell-based assays dependent upon IL-13 signaling. Appropriate
substitution was further found to influence inhibition of a key off-target, LRRK2. Finally, the most
potent compounds were found to be metabolically labile, which makes them ideal scaffolds for
further development as topical agents for IL-13 mediated diseases of the lungs and skin (for
example asthma and atopic dermatitis, respectively).
1

Keywords
Janus Kinase, JAK, JAK1, JAK2, interleukin-13, IL-13, structure-based design, asthma, atopic
dermatitis
2

The Janus kinases (JAKs) are a family of four intracellular tyrosine kinases (JAK1, JAK2, JAK3,
and TYK2) that regulate the JAK/STAT signaling pathway.1 The JAK/STAT cascade is initiated
when signaling molecules known as cytokines bind to the extracellular domains of their
transmembrane receptor complexes. The JAKs associate with the intracellular domains of the
cytokine receptors and become phosphorylated and activated upon cytokine binding.
Subsequently, an additional family of intracellular proteins known as the signal transducers and
activators of transcription (STATs) become phosphorylated. Once phosphorylated the STATs
dimerize and translocate to the cell nucleus where they control genes relevant to a range of
biological functions including immunity, inflammation, and hematopoiesis. The JAKs are
attractive drug targets because they can be effectively inhibited by small molecules, and because
inhibiting JAKs can block the cellular effects of the cytokines that signal through them.
One important cytokine signaling through JAK/STAT is interleukin-13 (IL-13). IL-13 is
implicated in allergic inflammation, thus interrupting its signaling via JAK inhibition may provide
benefit in associated diseases such as severe asthma and atopic dermatitis.2 Many cytokines recruit
and activate multiple JAK isoforms in their receptor complexes. In the case of IL-13 three
isoforms, namely JAK1, JAK2, and TYK2 are activated.3 However, it is known that not all the
JAKs associated with a given cytokine’s receptor complex play an equal role in controlling the
downstream signaling. IL-6, for example, is another cytokine that recruits and activates JAK1,
JAK2, and TYK2. Notably, selective JAK1 inhibition can effectively block IL-6 signaling,
whereas inhibition of JAK2 and TYK2 appear to play only a minor role.4
To assess the relative importance of JAK1, JAK2, and TYK2 kinase activity on IL-13 signaling,
we tested a panel of inhibitors with different JAK family selectivity profiles in an IL-13 triggered
cell-based assay (Table 1). IL-13 signals through its receptor complex and associated JAKs,
3

converting inactive and unphosphorylated STAT6 to the active and phosphorylated form
(pSTAT6). Thus, our IL-13-pSTAT6 cell-based assay involved stimulating BEAS-2B cells5 with
IL-13, and monitoring for pSTAT6 formation in the presence or absence of JAK inhibitor.6 During
assay benchmarking we wished to minimize the potential for poor cell penetrance to limit cell-
based potency of test compounds. As such we, chose highly characterized JAK inhibitors (1 –
5)7,8,9,10,11 with both potent activity in cell-based assays dependent on the JAK isoforms they
inhibit, and moderate to high permeability in MDCK transwell assays (Table 1). Compound 1 is a
highly biochemically potent and selective JAK2 inhibitor that also shows potent inhibition in a
JAK2-dependent cell-based assay (SET-2 IC50 = 60 nM). Despite potent cellular engagement of
JAK2, we found that 1 showed negligible inhibition in our IL-13-pSTAT6 assay, indicating that
JAK2 inhibition is not likely to be effective in disrupting IL-13 signaling. Similarly, 2 is a potent
and selective TYK2 inhibitor that previously showed strong inhibition in a TYK2-dependent cell-
based assay (IL-23-pSTAT3 IC50 = 66 nM). Despite potent cellular TYK2 engagement, compound
2 exhibited only very weak cellular inhibition in our IL-13-pSTAT6 assay, indicating that TYK2
inhibition is also not likely to be effective in disrupting IL-13 signaling. Also notable is the fact
that both compounds 1 and 2 are highly permeable in the MDCK transwell assay, making it
unlikely that poor cell penetrance is responsible for the poor IL-13-pSTAT6 potency. In contrast
with the selective JAK2 (1) and TYK2 (2) inhibitors, compounds with potent biochemical JAK1
inhibition (3, 4, 5) exhibited correspondingly potent inhibition in our IL-13-pSTAT6 assay.
Indeed, disruption of cell-based IL-13 signaling appeared to track consistently with JAK1
biochemical potency regardless of potency against the other JAK isoforms. Notably, compounds
2 – 5 possess markedly different selectivity profiles for JAK1 relative to the other JAK family
members, yet in each case the IL-13-pSTAT6 IC50 was remarkably proportional to the biochemical
4

JAK1 Ki (IL-13-pSTAT6 IC50 ÷ JAK1 Ki = 50x – 71x). The consistent correlation between IL-13-
pSTAT6 IC50 and JAK1 Ki, coupled with the negligible disruption in IL-13 signaling by potent
JAK2 (1) or TYK2 (2) inhibitors led us to believe that JAK1 inhibition likely played the dominant
role in disrupting IL-13 signaling, while JAK2 and TYK2 were likely less important.
Table 1. Biochemical and cell-based JAK inhibition, and MDCK permeability of previously
reported JAK inhibitors.
N
O
O
Cl
F
N
N
N
HO
F
H
N
N
Cl
F
N
O
N
N
N
N
N
N
N
NH
N
N
O
N
N
N
N
N
N
N
N
H
N
N
N
H
N
N
N
H
H
H
1
2
G-989 (3)
4
5
Tofacitinib ( )
Compound name:
BMS-911543 ( )
G-444 ( )
Upadacitinib ( )
JAK isoform(s)
most potently inhibited:
JAK2
TYK2
JAK1
JAK1
JAK1, JAK3, JAK2
Reported
progression status:
Ph1/2a
clinical trials
Preclinical in vivo
tool compound
Preclinical in vivo
tool compound
Ph3 clinical
trials
Approved
drug
aLiterature Data
bData Generated in This Work
Cell-
Based
IC50
Primary
JAK
Involved
cMDCK
A:B,
B:A÷A:B
IL-13-
pSTAT6 pSTAT6 IC50
hIL-13-
dJAK1 Ki eJAK2: fJAK3: gTYK2:
Ex
Cell-Based
Assay Name
(nM)
JAK1
JAK1 JAK1
IC50 (nM) ÷ JAK1 Ki
(nM)
1
2
SET-2
JAK2
TYK2
60
66
16, 0.94
16, 0.81
360 0.0013x i0.31x j0.11x k>1000
lNC
56x
IL-23-
pSTAT3
23
1.0x
26x
2.4x 0.028x 1300
IL-6-
pSTAT3
3
4
5
JAK1
JAK1
JAK1
110
9
6.2, 0.81
2.9
39x
i16x
10x
j18x
11x
210
45
71x
50x
61x
IL-6-
pSTAT3
3.6, 0.80 0.89
2.7x
1.1x
IL-6-
pSTAT3
53
3.4, 1.2
0.64
0.50x
39
aCell-based data for each compound was obtained from the indicated references: 17, 28, 39, 410, 59.
bSee Supplementary Data for assay details.
cPermeability in transwell MDCK assays. Units: cm/s x 10-6. A:B classifications: <1 = low, 1 – 10
= moderate, >10 = high.
dBiochemical JAK1 Ki. Genentech assay.
5

eRatio of biochemical JAK2 Ki / JAK1 Ki. Genentech assays.
fRatio of biochemical JAK3 Ki / JAK1 Ki. Genentech assays unless otherwise indicated.
gRatio of biochemical TYK2 Ki / JAK1 Ki. Genentech assays unless otherwise indicated.
hBiochemical to cell shift from the JAK1 biochemical assays to the IL-13-pSTAT6 cell-based
assay.
iRatio of biochemical JAK3 IC50 / JAK1 IC50. SelectScreen Kinase Profiling Services, Thermo
Fisher Scientific, Madison, WI.
jRatio of biochemical TYK2 IC50 / JAK1 IC50. SelectScreen Kinase Profiling Services, Thermo
Fisher Scientific, Madison, WI.
kIC50 not reported since 50% inhibition not achieved at top concentration tested (1 µM).
lNot calculated since an IL-13-pSTAT6 IC50 was not reported.
Genentech has previously disclosed a series of pyrazolopyrimidine JAK inhibitors.12 Although
optimized to be most potent against JAK2, these compounds retained residual levels of JAK1
inhibition. Selected SAR highlights are shown in Table 2. The JAK potency of an early example
(6) was improved by incorporating a regioisomeric central pyrazole moiety (7). Further elaboration
with a 2,5-dichlorophenyl group (8) led to additional JAK potency enhancement. Translocation of
the methyl group to the alternate N of the central pyrazole (9) was deleterious to potency.
Consistent with the data generated with the compounds in Table 1, the relatively weak JAK1
biochemical potencies of compounds 7, 8, and 9 resulted in correspondingly weak cell-based
disruption of IL-13 signaling. In order to more effectively block IL-13 signaling, we wished to
discover related analogs with improved JAK1 inhibition.
aTable 2. Biochemical and cell-based JAK inhibition of previously reported12 pyrazolopyrimidine
JAK inhibitors.
N
N
N
N
N
N
N
N
N
N
N
N
O
O
O
O
NH
N
NH
Cl
Cl
NH
NH
Cl
Cl
N
N
N
N
N
N
N
Cl
Cl
7
6
8
9
bIL-13-
pSTAT6
IC50 (nM)
bJAK1 Ki bJAK2 Ki
(nM) (nM)
Ex
6

6
7
8
9
110
46
12
6.4
>1000
230
2.8
0.42
39
3.5
>1000
aBlank space indicates data not generated.
bGenentech assays. All values represent the mean of at least two independent runs. See
Supplementary Data for assay details.
To generate hypotheses for additional JAK1 potency optimization we examined JAK-
complexed crystal structures of ADP13, and compound 6.12 As expected, ADP forms multiple
favorable interactions with JAK1, including engaging Glu966 and Arg1007 in hydrogen bond
(H-bond) interactions via the ribose hydroxyl groups (Figure 1a). The binding mode of
compound 6 relative to ADP is shown in Figure 1b, and these structures guided further
optimization of compound 8. For example, as shown in Figure 1c, appropriate substitution at
positions R1 and R2 could offer productive vectors to engage Glu966 and Arg1007.
7

Figure 1. X-ray crystal structures (a) ADP bound to JAK1 (PDB Accession code: 5KHW13). (b)
Overlay of ADP and 6. Ligands extracted from PDB Accession codes: 5KHW13 and 4HGE12.
(c) Binding mode of pyrazolopyrimidine scaffold in JAK1.
Since substitutions at R1 and R2 were expected to approach Glu966 and Arg1007, respectively,
we wished to study whether modifications at those positions would impact JAK1 potency (Table
3). Additionally, since compound 8 was already a relatively lipophilic starting point we wished
to improve potency while maintaining or lowering cLogP. Excessive cLogP is linked to a
number of undesirable properties such as promiscuity and toxicity, thus lipophilic ligand
efficiency (LLE) was used as a convenient metric to optimize for maximum potency with
minimum cLogP.14 Analogs 10 – 12 (Table 3) were synthesized in an attempt to engage Arg1007
with a H-bond, but a large JAK1 potency loss was observed in each case relative to 8. Analogs
13 – 23 were designed to optimally fill the ribose pocket by substituting at R2. Notably, JAK1
potency and LLE could be significantly improved by appropriate substitution at R2 into the
ribose pocket. For example, compound 15 was ~4.5-fold more potent against JAK1 than
compound 8 with a substantial improvement in LLE. Groups larger than the methyl ether present
in 15 were generally not beneficial to potency (compounds 16 – 19). However, incorporation of a
difluoromethoxy group was a notable exception, with compounds 20 and 21 exhibiting improved
biochemical and cell-based potency relative to their methoxy matched pairs 14 and 15. Although
not immediately obvious, x-ray crystal structures ultimately revealed that the difluoromethoxy
group not only optimally filled the ribose pocket, but also engaged Arg1007 in a H-bonding
interaction (see Figures 2c and 2d). The final ribose pocket modifications surveyed were analogs
22 and 23. Although the thiomethyl ether 22 was more potent than its methyl ether matched pair
15, potency was not further improved by incorporation of the difluoromethyl moiety (compound
8

23). Thus, the difluoromethoxy R2 group was chosen as the favored substitution for further SAR
exploration.
Several of the analogs in Table 3 (14, 16, 17, 19, 20, 23) were designed to H-bond to Glu966
via the hydrogen atom at R1 of the central pyrazole. However, despite the additional H-bond
interaction (see Figure 2), potency was not significantly improved, as evidenced by the R1
H→CH3 matched pairs 14→15 and 20→21.
aTable 3. Influence of groups targeting ribose pocket / Arg1007 and Glu966 on biochemical and
cell-based JAK inhibition.
N
N
N
O
NH
Cl
N
N
R1
R2
cIL-13-
pSTAT6
IC50 (nM)
bJAK1 cJAK1 Ki cJAK2 Ki
Ex
R1
R2
Cl
cLogP
LLE
(nM)
(nM)
8
CH3
3.1
2.1
1.6
5.5
5.4
5.7
2.8
32
45
0.42
13
230
10 CH3
11 CH3
HO
8.6
OH
12 CH3
13 CH3
2.0
5.9
12
4.0
HO
3.2
2.2
2.3
5.8
6.9
6.9
0.91
0.89
0.57
0.28
0.36
0.35
14
15 CH3
16
H
23
25
O
O
O
H
2.6
6.3
1.4
0.80
54
9

O
O
17
H
2.8
3.1
3.4
6.1
5.0
5.3
1.3
7.9
2.1
0.45
2.0
21
18 CH3
O
F
19
20
H
H
0.49
64
F
F
F
F
O
F
2.5
7.2
0.21
0.071
5.2
O
F
21 CH3
22 CH3
2.7
3.1
3.6
7.0
6.2
5.5
0.21
0.46
0.85
0.088
0.23
0.23
4.7
11
26
S
S
F
23
H
F
aBlank spaces indicate data not generated.
bLLE = JAK1 pKi – cLogP
cGenentech assays. All values represent the mean of at least two independent runs. See
Supplementary Data for assay details.
Figure 2 summarizes x-ray structural observations related to key SAR described thus far. As
expected from the JAK2 crystal structure of 612, compounds 15, 20, and 21 bind via their
pyrazolopyrimidine motifs to the JAK1 hinge residues Leu959 and Glu957, and interact with the
gatekeeper Met956 sidechain (Figures 2b, 2c, 2d). The chlorophenyl group present in each of 15,
20, and 21 participates in favorable Van der Waals interactions with the JAK1 P-Loop. In the
case of 15, the phenyl methoxy group extends into the base of the ribose pocket and makes a
favorable Van der Waals contact with the sidechain of Leu1010 (Figure 2b). As shown in Figure
2a, the x-ray crystal structures of compound 15, and the more potent analogs 20 and 21, are
similar, with only a minor downward movement of the P-Loops and ligands observed for 20 and
21. These subtle shifts are likely a result of the additional interactions observed between the
difluoromethoxy groups present in 20 and 21 and the ribose pocket of the JAK1 protein. Similar
10

to the methoxy group present in 15 the potency-enhancing difluoromethoxy groups in 20 and 21
participate in favorable Van der Waals interactions with Leu1010. Additionally, 20 and 21
participate in favorable Van der Waals interactions with the sidechain methylene of Ser963, as
well as dipolar interactions with the backbone carbonyl carbon of Gly1020 (both interactions via
the fluorine atoms); while the polarized H resident in the difluoromethoxy group forms a non-
classical hydrogen bond with the backbone carbonyl of Arg1007. Notably, ADP in complex with
JAK1 was also observed to hydrogen bond with the Arg1007 backbone carbonyl (see Figure 1a).
We believe these additional ribose pocket interactions are the source of the enhanced potency of
the difluoromethoxy analogs 20 and 21 relative to the methoxy analogs 14 and 15. A final
important structural observation is the apparent lack of benefit of interacting with the sidechain
of Glu966. Although ADP does form a hydrogen bond with the carboxylate sidechain of Glu966
(Figure 1a), this residue is solvent exposed and conformationally mobile (compare Figures 2b,
2c, and 2d). Thus, although 20 forms a hydrogen bond with Glu966 while compound 21 does not
(Figure 2c vs. Figure 2d), the JAK1 potency of the two compounds is virtually identical.
11

Figure 2. X-ray crystal structures of 15 (yellow), 21 (pink), and 20 (cyan) in complex with JAK1.
(a) Overlay of all three structures. (b), (c), (d): Individual structures of 15 (PDB Accession code:
6N77), 21 (PDB Accession code: 6N78), and 20 (PDB Accession code: 6N79), respectively.
Where indicated, distances refer to heavy atom to heavy atom.
Having optimized the ribose pocket group, we then surveyed additional heterocycles as
potential replacements for the central pyrazole ring. The primary design hypothesis was to
enforce the co-planarity between the central heterocycle and the proximal amide carbonyl with
either a NH→O H-bond or related S→O interaction (see Table 4). Unfortunately, both imidazole
24 and thiazole 25 lost potency compared to the corresponding pyrazole 21.
aTable 4. Influence of central pyrazole replacements on biochemical and cell-based JAK
inhibition.
12

N
N
N
N
N
N
N
N
N
O
O
O
NH
NH
NH
Cl
Cl
Cl
HN
S
N
N
N
N
O
O
O
F
F
21
24
25
F
F
cIL-13-
pSTAT6
IC50 (nM)
bJAK1
LLE
cJAK1 Ki cJAK2 Ki
Ex cLogP
(nM)
(nM)
24
25
3.0
3.1
5.4
5.9
3.9
1.1
0.99
0.45
58
aBlank spaces indicate data not generated.
bLLE = JAK1 pKi – cLogP
cGenentech assays. All values represent the mean of at least two independent runs.
The next region of optimization was the bicyclic hinge binder. Table 5 summarizes the key
hypotheses guiding design of the hinge binders, as well as SAR of the final compounds.
Compounds 26 (extra hinge-binding contact) and 27 (regioisomeric hinge binder) were
approximately equipotent to the matched pair 20. Substitution on the 6-membered pyrimidine or
pyridazine ring was detrimental to potency (28 – 30), as was incorporation of an
isothiazolopyridine as the hinge binder (31). A 6,6-hinge binder was detrimental to potency (32),
as were a number of related 6,5-systems (33 – 37). One promising alternate hinge binder was
pyrazolopyridine 38 which displayed potency similar to the matched pair pyrazolopyrimidine 15.
Methylation of the hinge binding pyrazole moiety to enforce the tautomer required for an
intramolecular H-bond with the pendant amide led to a loss in potency (39). Examination of
crystal structures of pyrazolopyrimidine 15 and pyrazolopyridines 38 and 39 (Figure 3) reveals
that the scaffolds bind to JAK1 with similar interactions to the hinge and gatekeeper residues,
13

and a virtually identical placement of the P-Loop and ribose pocket moieties. Notably, the
pyrazole hinge binder moiety of 38 approaches within H-bonding distance (2.69 Å) to a
crystallographic water molecule already tightly bound to Met956, Gly1020, and Asp1021
(Figure 3b). The methylated pyrazole hinge binder (39), while enforcing the low energy binding
conformation, also disrupts the tightly bound crystallographic water, presumably leading to the
loss in potency (Figure 3c). Incorporation of the difluoromethoxy ribose pocket group with the
optimal 6,5 hinge binder led to a minor increase in potency (38→40).
aTable 5. Influence of hinge-binding groups on biochemical and cell-based JAK inhibition.
Other Hinge
Acceptor
Donor
Key Features
of Hinge
binding group
binding groups
A
B
enforce
planarity
R
( )
N
N
R
R
enforce
planarity
N
O
O
NH
NH
Cl
Cl
O
NH
Cl
HN
N
N
N
N
N
O
F
O
O
F
F
21
F
cIL-13-
pSTAT6
IC50 (nM)
bJAK1 cJAK1 Ki cJAK2 Ki
Ex A or B
R
cLogP
LLE
(nM)
(nM)
N
N
N
H2N
26
27
28
29
30
A
A
A
A
A
2.0
3.0
2.8
2.9
3.3
7.7
0.19
0.042
4.0
N
F
N
N
6.7
5.3
6.1
6.2
0.21
7.8
0.071
2.4
5.0
N
N
N
N
N
0.99
0.29
0.10
0.078
34
13
N
N
N
N
14

N
S
N
31
32
33
A
B
A
3.0
2.8
2.9
6.3
5.9
5.3
0.47
1.9
0.085
0.62
0.96
20
38
42
N
N
H
N
N
N
5.6
H
N
N
34
B
2.3
5.9
5.7
2.6
N
S
N
N
N
35
36
A
A
3.0
3.4
6.3
5.3
0.51
1.8
0.17
0.50
9.2
36
O
H
N
N
N
37
B
2.1
4.3
370
730
N
N
NH
38
39
40
B
B
A
2.0
2.1
2.2
7.2
6.3
7.1
0.60
4.3
0.78
4.7
20
93
18
N
N
N
N
N
NH
0.56
0.49
N
aBlank spaces indicate data not generated.
bLLE = JAK1 pKi – cLogP
cGenentech assays. All values represent the mean of at least two independent runs. See
Supplementary Data for assay details.
15

Figure 3. X-ray crystal structures of (a) 15 (PDB Accession code: 6N77), (b) 38 (PDB Accession
code: 6N7B), and (c) 39 (PDB Accession code: 6N7A) in complex with JAK1.
As shown in Table 6, a final round of SAR exploration was conducted on the phenyl ring
extending beneath the P-Loop (see compound 41 for numbering convention). Substitution at the
5-position of the phenyl was important (41 – 47), and the most lipophilic groups were typically
most potent (46, 47). Fluorine incorporation was not beneficial at the 6-position (48, 49), and led
to a potency loss at the 3- and 4-positions (50, 51). Notably, however, installation of H-bond
donors at the 4-position (52 – 55) led to highly biochemically potent compounds, with indazole
55 in particular displaying exceptional levels of biochemical JAK1 inhibition (Ki = 75 pM) and
LLE (8.4). Despite the outstanding JAK1 biochemical potency, a large cell shift was observed
for 55 presumably due to permeability limitations (see Table 8). Replacement of the indazole
ring with a naphthalene moiety led to a reduction in biochemical potency, but a smaller cell shift
(56 and 57). As a result, compound 57 exhibited biochemical and cell-based potency similar to
compound 20.
As shown in Figure 4, the chlorophenyl (15), indazole (54), and naphthyl (56) containing
compounds bind in a similar orientation to JAK1 (Figure 4a), with a small degree of P-loop
collapse around indazole 54 likely due to its smaller size. Additionally, the indazole participates
16

in a H-bonding interaction with the sidechain of Asp1021, likely leading to the increase in LLE
and biochemical JAK1 inhibition (Figure 4b). By comparison, the naphthyl group does not make
any specific H-bonds and, as suggested by its lower LLE, its affinity is driven by hydrophobic
interactions (Figure 4c).
aTable 6. Influence of P-Loop groups on biochemical and cell-based JAK inhibition.
N
N
N
O
NH
R
HN
N
cIL-13-
pSTAT6 IC50
(nM)
bJAK1 cJAK1 Ki
cJAK2 Ki
Ex
41
42
R
cLogP
1.8
LLE
(nM)
(nM)
6
5
3
1
4
2
O
6.7
3.3
0.86
0.27
F
F
CN
1.5
7.5
0.90
22
O
F
F
F
43
44
45
2.0
2.2
2.6
7.1
6.6
6.7
0.71
1.4
0.19
0.34
0.19
34
27
13
O
F
F
O
F
F
0.53
O
F
F
17

Br
46
2.8
3.0
2.2
3.0
2.3
7.1
6.8
6.8
6.3
5.7
0.12
0.15
0.95
0.56
9.2
0.047
0.045
0.29
0.16
1.2
3.3
2.8
15
O
F
F
I
47
48
O
F
F
F
Cl
O
F
Cl
49
50
14
O
F
F
Cl
F
O
Cl
51
52
F
2.2
1.8
6.5
7.7
2.2
0.79
0.16
42
O
Cl
Cl
OH
OH
0.31
140
O
53
2.1
7.9
0.11
0.057
36
O
F
F
N
NH
54
55
1.3
1.7
8.2
8.4
0.30
0.20
370
40
O
N
NH
0.075
0.052
O
F
F
56
57
2.6
3.0
6.6
0.67
0.31
0.34
0.14
14
O
6.5
6.4
O
F
F
aBlank spaces indicate data not generated.
bLLE = JAK1 pKi – cLogP
18

cGenentech assays. All values represent the mean of at least two independent runs. See
Supplementary Data for assay details.
Figure 4. X-ray crystal structures of 15 (yellow), 54 (cyan), and 56 (pink) in complex with JAK1.
(a) Overlay of all three structures. (b), (c): Individual structures of 54 (PDB Accession code:
6N7D), and 56 (PDB Accession code: 6N7C), respectively.
To assess selectivity for the JAK family relative to off-target kinases we screened several
optimized inhibitors against a panel of kinases. As shown in Figure 5, compound 20 was found
to be relatively selective for the JAK family, with LRRK2 and FYN being the only non-JAK
kinases significantly inhibited in a 71-membered panel. LRRK2 was believed to be an off-target
kinase of concern since potent inhibition has been implicated in lung toxicity.15 In an effort to
identify compounds with reduced LRRK2 potency, multiple analogs were screened against
LRRK2 at a single concentration of 0.1 µM, with a smaller number of compounds subsequently
advanced to a detailed selectivity measurement comparing the LRRK2 IC50 relative to the JAK1
IC50 (Table 7). In determining the selectivity for JAK1 relative to LRRK2 we believed it was
important to use the same assay format for both assays. Since the LRRK2 assay was conducted
19

at Thermo Fisher, we also generated a JAK1 IC50 at Thermo Fisher, and used that value for the
selectivity calculations (see Table 7). Notably, reference compound 8 was significantly less
potent against LRRK2 than 20 when screened at the single 0.1 µM concentration. However, 8
was also significantly less potent against JAK1, thus the selectivity relative to LRRK2 (LRRK2
IC50 ÷ JAK1 IC50) for compound 8 (13x) was not improved relative to compound 20 (15x).
Compounds 26, 46, 47, and 55 all had ~equal or greater inhibition of LRRK2 at 0.1 µM than did
20, thus, they were not advanced to the IC50 assays since we concluded they would be unlikely to
have significantly improved selectivity. In contrast, compounds 21, 40, and 57 were found to
have reduced LRRK2 inhibition at 0.1 µM than did 20, thus, were advanced to the IC50 assays. In
each case selectivity for JAK1 relative to LRRK2 was improved compared to compound 20 (20:
15x, 21: 47x, 40: 41x, 57: 83x). The two compounds (21 and 57) with the best combination of
JAK1 biochemical inhibition, potent blockade of cell-based IL-13 signaling, and selectivity for
JAK1 vs LRRK2 were further profiled in the broader kinase panel (Figure 5) and, as desired,
were found to only weakly inhibit non-JAK family kinases other than Fyn.16
Figure 5. Extended off-target kinase inhibition of 20 and 21, and 57.a
aSelectScreen Kinase Profiling Services, Thermo Fisher Scientific, Madison, WI. See
Supplementary Data for assay details.
aTable 7. LRRK2 inhibition of selected analogs.
20

b% LRRK2 bLRRK2
bJAK1
IC50
(nM)
LRRK2 IC50
÷
JAK1 IC50
Ex
Inhibition
@ 0.1 µM
IC50
(nM)
8
14
87
61
99
52
86
91
89
31
400
18
32
1.2
1.7
13x
15x
47x
20
21
26
40
46
47
55
57
81
79
1.9
41x
180
2.2
83x
aBlank spaces indicate data not generated.
bSelectScreen Kinase Profiling Services, Thermo Fisher Scientific, Madison, WI. See
Supplementary Data for assay details.
Finally, the majority of compounds were found to have moderate to poor metabolic stability
when incubated with human liver microsomes (Table 8). The only metabolically stable example
was 55, a molecule with a markedly lower LogD7.4 than the other compounds in Table 8.
Unfortunately, 55 also possessed dramatically reduced membrane permeability as measured in
MDCK transwell assays, leading to impaired cell potency (see Table 6) and concern that it would
suffer from permeability-limited intestinal absorption if dosed orally. This series of compounds
may, thus, be most suited for optimization as locally delivered topical agents, where poor
metabolic stability is a desirable feature.
Table 8. Metabolic stability, lipophilicity, and permeability of selected analogs.
21

cMDCK A:B,
B:A÷A:B
aHLM
CL
Ex
bLogD7.4
8
12
14
15
11
15
15
15
4.4
17
2.9
3.2
3.0
3.3
2.7
3.2
3.4
1.6
3.5
20, 0.88
14, 1.1
14, 0.55
15, 0.83
9.5, 1.9
17, 1.2
15, 1.0
1.6, 1.0
13, 1.2
20
21
26
40
46
47
55
57
aClearance predicted from human liver microsomes. Units: mL/min/kg. See Supplementary Data
for assay details.
bMeasured LogD7.4.17
cPermeability in transwell MDCK assays. Units: cm/s x 10-6. See Supplementary Data for assay
details. A:B classifications: <1 = low, 1 – 10 = moderate, >10 = high.
In conclusion, a series of highly potent JAK1/2 inhibitors has been identified. Notably, the
significantly improved JAK1 potency also led to effective blockade of cell-based IL-13
signaling. Compounds 21 and 57 showed an optimum balance between potent JAK1 inhibition
and minimal off target kinase inhibition. All of the analogs exhibiting the best cell-based potency
were also found to be poorly stable in the presence of human liver microsomes. Exquisite
potency, favorable kinome selectivity, and metabolic lability are preferred features in compounds
intended for local delivery as topical agents. Given IL-13’s role in diseases of the lungs and
skin2, additional optimization of this series for topical delivery to those organs may be warranted.
Appendix A. Supplementary Data
22

Experimental procedures and analytical data for compounds 20 and 21. Synthetic schemes and
general reaction conditions for all compounds 10-57. Crystallography statistics for compounds
15, 20, 21, 38, 39, 54, and 56 in complex with JAK1. Procedures for Genentech biochemical and
cell-based assays, and Thermo Fisher biochemical kinase assays. Procedures for MDCK and
liver microsome stability assays.
Acknowledgements
This work was funded by Genentech.
References and Notes
N
N
N
N
N
N
N
N
N
O
O
O
NH
Cl
NH
NH
Cl
N
HN
N
N
N
N
Cl
O
O
F
F
8
21
57
F
F
0.21
4.7
2.8
230
13x
0.31
6.4
Biochemical: JAK1 Ki (nM)
Cell: IL-13-pSTAT6 IC50 (nM)
Off-target: Selectivity vs. LRRK2
47x
83x
1
Shuai K, Liu B. Regulation of JAK-STAT signalling in the immune system. Nat Rev Immunol.
2003;3(11): 900-911.
2 (a) Wills-Karp M, Luyimbazi J, Xu X, et al. Interleukin-13: Central Mediator of Allergic
Asthma. Science. 1998;282(5397): 2258-2261. DOI: 10.1126/science.282.5397.2258. (b) Corren
J. Role of interleukin-13 in asthma. Curr Allergy Asthma Rep. 2013;13(5): 415-420. DOI:
10.1007/s11882-013-0373-9. (c) Wollenberg A, Howell MD, Guttman-Yassky E, et al.
Treatment of atopic dermatitis with tralokinumab, an anti-IL-13 mAb. J Allergy Clin Immunol.
2018. DOI: 10.1016/j.jaci.2018.05.029. (d) Simpson EL, Flohr C, Eichenfield LF, et al. Efficacy
23

and safety of lebrikizumab (an anti-IL-13 monoclonal antibody) in adults with moderate-to-
severe atopic dermatitis inadequately controlled by topical corticosteroids: A randomized,
placebo-controlled phase II trial (TREBLE). J Am Acad Dermatol. 2018;78(5): 863-871 e811.
DOI:10.1016/j.jaad.2018.01.017.
3 Clark JD, Flanagan ME, Telliez JB. Discovery and development of Janus kinase (JAK) inhibitors
for inflammatory diseases. J Med Chem. 2014;57(12): 5023-5038. DOI: 10.1021/jm401490p.
4
Sohn SJ, Barrett K, Van Abbema A, et al. A restricted role for TYK2 catalytic activity in human
cytokine responses revealed by novel TYK2-selective inhibitors. J Immunol. 2013;191(5): 2205-
2216. DOI: 10.4049/jimmunol.1202859.
5
Reddel RR, Ke Y, Gerwin BI, et al. Transformation of Human Bronchial Epithelial Cells by
Infection with SV40 or Adenovirus-12 SV40 Hybrid Virus, or Transfection via Strontium
Phosphate Coprecipitation with a Plasmid Containing SV40 Early Region Genes. Cancer
Research. 1988;48(7): 1904-1909.
6 Details on the IL-13-pSTAT6 assay may be found in the Supplementary Data section.
7
Wan H, Schroeder GM, Hart AC, et al. Discovery of a Highly Selective JAK2 Inhibitor, BMS-
911543, for the Treatment of Myeloproliferative Neoplasms. ACS Med Chem Lett. 2015;6(8): 850-
855. DOI: 10.1021/acsmedchemlett.5b00226.
8
Liang J, Van Abbema A, Balazs M, et al. Identification of an imidazopyridine scaffold to generate
potent and selective TYK2 inhibitors that demonstrate activity in an in vivo psoriasis model.
Bioorg Med Chem Lett. 2017;27(18): 4370-4376. DOI: 10.1016/j.bmcl.2017.08.022.
9
Zak M, Hurley CA, Ward SI, et al. Identification of C-2 hydroxyethyl imidazopyrrolopyridines
as potent JAK1 inhibitors with favorable physicochemical properties and high selectivity over
JAK2. J Med Chem. 2013;56(11): 4764-4785. DOI: 10.1021/jm4004895.
10
WO2015061665. Jak1 selective inhibitor and uses thereof. Voss JW, Camp HS, Padley RJ.
11
Flanagan ME, Blumenkopf TA, Brissette WH, et al. Discovery of CP-690,550: a potent and
selective Janus kinase (JAK) inhibitor for the treatment of autoimmune diseases and organ
transplant rejection. J Med Chem. 2010;53(24): 8468-8484.
12 Hanan EJ, van Abbema A, Barrett K, et al. Discovery of potent and selective pyrazolopyrimidine
janus kinase 2 inhibitors. J Med Chem. 2012;55(22): 10090-10107. DOI: 10.1021/jm3012239.
13
Caspers NL, Han S, Rajamohan F, et al. Development of a high-throughput crystal structure-
determination platform for JAK1 using a novel metal-chelator soaking system. Acta Crystallogr F
Struct Biol Commun. 2016;72(Pt 11): 840-845. DOI: 10.1107/S2053230X16016356.
14
Freeman-Cook KD, Hoffman RL, Johnson TW. Lipophilic efficiency: the most important
efficiency metric in medicinal chemistry. Future Medicinal Chemistry. 2013;5(2): 113-115. DOI:
10.4155/FMC.12.208
24

15 Fuji RN, Flagella M, Baca M, et al. Effect of selective LRRK2 kinase inhibition on nonhuman
primate lung. Science Translational Medicine. 2015;7(273): 273ra15. DOI:
10.1126/scitranslmed.aaa3634.
16
Thermo Fisher JAK3 and TYK2 IC50 values for 21 and 57 were also determined to enable
selectivity calculations relative to JAK1. The following selectivity ratios were measured.
Compound 21: JAK3 IC50 ÷ JAK1 IC50 = 2.1x. TYK2 IC50 ÷ JAK1 IC50 = 3.0x. Compound 57:
JAK3 IC50 ÷ JAK1 IC50 = 2.7x. TYK2 IC50 ÷ JAK1 IC50 = 5.9x.
17
Lin B, Pease JH. A novel method for high throughput lipophilicity determination by microscale
shake flask and liquid chromatography tandem mass spectrometry. Comb. Chem. High Throughput
Screen. 2013;16(10): 817-825. DOI: 10.2174/1386207311301010007.
25