The Discovery of 3-((4-Chloro-3-methoxyphenyl)amino)-1-((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-1H-pyrazole-4-carboxamide, a Highly Ligand Efficient and Efficacious Janus Kinase 1 Selective Inhibitor with Favorable Pharmacokinetic Properties

Article abstract of DOI:10.1021/acs.jmedchem.7b01135

The discovery of a potent selective low dose Janus kinase 1 (JAK1) inhibitor suitable for clinical evaluation is described. As part of an overall goal to minimize dose, we pursued a medicinal chemistry strategy focused on optimization of key parameters th

Full text of DOI:10.1021/acs.jmedchem.7b01135

Subscriber access provided by READING UNIV
Article
The Discovery of 3-((4-Chloro-3-methoxyphenyl)amino)-1-
((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-1H-pyrazole-4-
carboxamide, a Highly Ligand Efficient and Efficacious JAK1
Selective Inhibitor with Favorable Pharmacokinetic Properties
Tony Siu, Jason Brubaker, Peter Fuller, Luis Torres, Hongbo Zeng, Joshua Close, Dawn
M Mampreian, Feng Shi, Duan Liu, Xavier Fradera, Kevin Johnson, Nathan Bays, Elma
Kadic, Fang He, Peter Goldenblatt, Lynsey Shaffer, Sangita B Patel, Charles A Lesburg,
Carla Alpert, Lauren Dorosh, Sujal V Deskmukh, Hongshi Yu, Joel Klappenbach, Fiona
Elwood, Christopher J. Dinsmore, Rafael Fernandez, Lily Moy, and Jonathan R Young
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01135 • Publication Date (Web): 20 Nov 2017
Downloaded from http://pubs.acs.org on November 20, 2017
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 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
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
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 45
1
2
3
4
5
6
The Discovery of 3-((4-Chloro-3-methoxyphenyl)amino)-1-((3R,4S)-4-cyanotetrahydro-2H-
pyran-3-yl)-1H-pyrazole-4-carboxamide, a Highly Ligand Efficient and Efficacious JAK1
7
8
9
Selective Inhibitor with Favorable Pharmacokinetic Properties
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
Tony Siua*, Jason Brubakera, Peter Fullera, Luis Torresa, Hongbo Zenga, Joshua Closea, Dawn M.
Mampreiana, Feng Shik, Duan Liub, Xavier Fraderac, Kevin Johnsonh, Nathan Baysd, Elma Kadicd, Fang
Hel, Peter Goldenblattd, Lynsey Shafferd, Sangita B. Patele, Charles A. Lesburge, Carla Alpertd, Lauren
Doroshd, Sujal V. Deskmukhf, Hongshi Yug, Joel Klappenbachh, Fiona Elwoodj, Christopher J. Dinsmorea,
Rafael Fernandezd, Lily Moyi, and Jonathan R. Younga
a
d
Departments of Medicinal Chemistry, bDiscovery Process Chemistry, cModeling & Informatics, In Vitro
e
f
Pharmacology, Structural Chemistry, Pharmacokinetics Pharmacodynamics and Drug Metabolism,
j
gDiscovery Pharmaceutical Sciences, hMolecular Biomarkers, iIn Vivo Pharmacology, Respiratory and
Immunology, Merck & Co., Inc., 33 Avenue Louis Pasteur, Boston, MA 02115 USA
kDepartment of Chemistry, Pharmaron Beijing Co. Ltd.
6 Taihe Road
BDA, Beijing, 100176
P.R. China
lDepartment of Biology, Pharmaron Beijing Co. Ltd.
6 Taihe Road
BDA, Beijing, 100176
P.R. China
Abstract
The discovery of a potent selective low dose JAK1 inhibitor suitable for clinical evaluation is described. As
ACS Paragon Plus Environment

Page 3 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
part of an overall goal to minimize dose, we pursued a medicinal chemistry strategy focused on
optimization of key parameters that influence dose size, including lowering human Clint and increasing
intrinsic potency, bioavailability, and solubility. To impact these multiple parameters simultaneously, we
used lipophilic ligand efficiency as a key metric to track changes in the physicochemical properties of our
analogs, which led to improvements in overall compound quality. In parallel, structural information guided
advancements in JAK1 selectivity by informing on new vector space, which enabled the discovery of a
unique key amino acid difference between JAK1 (Glu966) and JAK2 (Asp939). This difference was
exploited to consistently produce analogs with the best balance of JAK1 selectivity, efficacy, and projected
human dose, ultimately culminating in the discovery of compound 28.
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
Introduction
The Janus KinaseꢀSignal Transducer and Activator of Transcription (JAKꢀSTAT) pathway is a wellꢀ
studied signaling pathway that regulates cytokine and growth factor production.1 Signaling in this pathway
is mediated by four tyrosine kinases including JAK 1, 2, and 3, as well as nonꢀreceptor tyrosineꢀprotein
kinase 2 (TYK2). These cytoplasmic JAK isoforms are associated with cytokine receptors as homo and
hetero dimers and trimers. Upon activation through cross phosphorylation, they recruit and phosphorylate
STAT transcription factors, which translocate to the nucleus and activate gene transcription. Impressively,
only four JAK isoforms are needed to regulate the vast number of signaling cytokines including interferons,
interleukins, endocrine factors, and growth factors. These cytokines regulate a host of biological functions
including T and B cell activation and differentiation, inflammation, and hemopoiesis. The JAKꢀSTAT
pathway has a vast role in regulating crucial cellular events and is a key contributor to and extensively
linked with immune and malignantꢀrelated diseases.2
Rheumatoid arthritis (RA) is an autoimmune disease characterized by inflammation of the synovial
membrane of joints.3 RA associated inflammation leads to severe joint pain along with stiffness and over
time chronic inflammation leads to destruction of the joint and loss of function. Central to the pathogenesis
of RA is the inflammatory function of immune cells that are coordinated and maintained by cytokines
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 45
1
2
3
4
5
6
7
8
9
regulated by the JAK family of kinases. To combat the inflammatory etiology of the disease, there are
commercial therapies that target both the cytokines (e.g. antiꢀTNFα agents, infliximab and entanercept) and
the JAK pathway (e.g. panꢀJAK inhibitor, tofacitinib (1), Figure 1).4 Despite the success of these safe and
efficacious treatments, an unmet medical need still remains for safe small molecule disease modifying antiꢀ
rheumatic drugs (DMARDS) that can achieve greater efficacy and provide more convenience than the
current therapies.5 However, a limiting factor for achieving greater efficacy with tofacitinib is the dose
limiting anemia observed in clinical trials which is associated with JAK2 inhibition and its effects on
erythropoiesis.4a These data suggest that an inhibitor selective for JAK1 over JAK2 may offer greater
clinical benefit by increasing the therapeutic margin over erythropoietin (EPO) inhibition. A key assumption
in this approach is that JAK1 inhibition alone is sufficient to achieve similar antiꢀinflammatory efficacy as
panꢀJAK inhibition. A distinct challenge to develop JAK1 selective inhibitors is the high sequence
homology between JAK1 and JAK2 in the catalytic domain. Despite these challenges, numerous research
groups have reported their strategies and progress in this area.6
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
In this paper we describe our strategy towards the identification of a small molecule JAK1 selective
inhibitor with drugꢀlike properties and low projected human dose. Due to the differences in KM values for
ATP in the various JAK isoforms, the most relevant measurement of isoform selectivity is to differentiate
against tofacitinib on key pathways of efficacy (e.g. interleukinꢀ6 (ILꢀ6)) relative to anemia (e.g. EPO). By
differentiating compound selectivity using cellꢀbased assays, we sought to improve the potential for clinical
translation of efficacy over anemia. Recently, we disclosed pyrazole carboxamide 2 as a potent JAK1
selective tool compound with a modest 10ꢀfold selectivity over JAK2 in both enzymatic and pathway cellꢀ
based assays (Figure 1).7 As a proof of concept, compound 2 demonstrated a larger therapeutic window over
tofacitinib in the collagen induced arthritis (CIA) efficacy and reticulocyte anemia models.8 These initial
efforts supported the strategy that selectivity for the ILꢀ6 pathway through JAK1 inhibition is sufficient to
achieve efficacy while sparing anemia bioꢀmarkers.
ACS Paragon Plus Environment

Page 5 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
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
Compound
1 (tofacitinib)
2
1.5
19
JAK1 IC50 (nM)a
JAK2 IC50 (nM)a
JAK3 IC50 (nM)a
TYK2 IC50 (nM)a
Cell IL6 IC50 (nM)b
Cell EPO IC50 (nM)b
Ratio EPO/IL6
1.3
1.4
0.34
12
1226
13
124
107
0.86
64
692
11
a
Figure 1. Selectivity profile for tofacitinib 1 and 2. Values in this table were determined by the HTRF
assay and are the mean of n = 2 experiments. bValues in this table are determined by GeneBLAzer assay and
are the mean of n = 2 experiments.
While achieving our goal of a pathway JAK1 selective inhibitor, extensive characterization of 2
revealed a high human predicted daily dose (343 mg QD) and a potential hERG liability (Figure 2).9
Minimizing the daily dose reduces body burden and the risk for idiosyncratic toxicities as well as improves
patient compliance.10 Furthermore, the high dose of 2 resulted in a relatively high dose number (D0 = 80),
which often complicates formulation options and risk assessment for clinical compounds.11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 45
1
2
3
4
5
6
hERG Activity of Compound 2
Predicted Human Dose of Compound 2
7
8
9
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
MK499 IC50 = 12,200 nM
Patch Clamp IC50 = 3000 nM
CV Guinea Pig QT NOEL:
8x over predicted clinical Cmax
r; h Hept Clint (mL/min/kg) = 109; 26
Rat Clp = 41 mL/min/kg
Vdss = 11 L/kg
F% = 16
Dog Clp = 21 mL/min/kg
Vdss = 8 L/kg
F% = 42
Predicted Human Dose = 343 mg QD
D0 = 80
Figure 2. Characterization and liabilities of compound 2.
Considering the potential issues with a high daily dose drug and the need for enabled formulation,
our goal in the current effort was to identify a lead candidate with predicted daily dose ≤ 100 mg while
decreasing D0. To optimize for a lower dose of an AUC driven target, we focused on the key parameters of
intrinsic clearance (Clint) and unbound AUC (AUCu), fraction absorbed (fa), and fraction escaping gut
metabolism (fg), whose relations is expressed by the equation in Figure 3.12 Thus, reduction of the dose of 2
required decreasing human Clint and AUCu (by increasing intrinsic potency), as well as increasing fa through
improvements in solubility and permeability.
ACS Paragon Plus Environment

Page 7 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
ꢀꢁꢂ × ꢂꢄꢅꢆꢇ
ꢃ
Dose
=
ꢈꢉ ∗ ꢈꢊ
7
8
9
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 3. Parameters that affect human dose calculation for AUC driven protein target. AUC unbound
(AUCu), intrinsic clearance (Clint), fraction absorbed (fa), fraction escape from gut metabolism (fg).
In addition to the high daily dose of compound 2, modest affinity for the hERG channel (MK499
IC50 = 12,200 nM) was confirmed in the functional patch clamp hERG blockade assay (hERG IC50 = 3,000
nM).13 Ultimately, the unacceptable safety margin for QTc prolongation shown in a guinea pig study (8 fold
margin over predicted human Cmax) combined with its high daily dose precluded 2 from further
progression.14 Blockade of the hERG channel has the potential to alter the QT interval and trigger life
threatening tosades des pointes, therefore further optimization was necessary to identify a suitable clinical
candidate.15
Results and Discussion
Initial strategies for reducing hERG channel binding focused on applying destabilizing SAR
derived from the 3D QSAR model for hERG channel activity.16 This strategy included structural
modifications including removing or attenuating the basicity of the most basic amine in 2 as well as
increasing overall polarity of the inhibitor. These approaches successfully led to multiple SAR advances
resulting in weakened hERG binding, but unfortunately negatively impacted other dose related parameters
which hindered progress (Figure 4). Replacing the basic amine with a hydroxyl group (e.g. compound 3) led
to improved MK499 (IC50 = 26,000 nM) while maintaining intrinsic potency (JAK1 IC50 = 1.2 nM), but the
bioavailability was compromised (F% =10). A similar effect on hERG binding was achieved through steric
hindrance by shielding the basic amine with methyl oxetane (e.g. compound 4) (MK499 IC50 = 46,000 nM,
JAK1 IC50 = 0.6 nM) but did not improve the Clint compared to 2, presumably due to the increased lipophilic
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 45
1
2
3
4
5
6
7
8
9
substituents compared to 2. Maintaining the basic amine while increasing the overall polarity of the
compound (e.g. compound 5 (HPLC Log D = 1.54 vs. 2.15 for compound 2) MK499 IC50 = 32,000 nM,
JAK1 IC50 = 0.3 nM) improved the hERG signal, albeit with negative impacts on cellular permeability and
overall fraction absorbed when dosed orally. The team explored many SAR permutations to balance hERG
binding, Clint, and bioavailability, but ultimately failed to deliver compounds with the required balance
profile within the compound 2 chemotype.
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
3
4
5
Basic Amine Replacement
JAK1 IC50 = 1.2 nM
Hindered Amine
JAK1 IC50 = 0.6 nM
MK499 IC50 = 46000 nM
Increased Polarity
JAK1 IC50 = 0.3 nM
MK499 IC50 = 32000 nM
MK499 IC50 = 26000 nM
Poor Bioavailability
Rat F% = 10
Low Permeability
Papp = 3.3 x 10ꢀ6 cm/sec
Rat F% = 0
High Intrinsic Clint
r;h Hept Clint = 80;32 mL/min/kg
ACS Paragon Plus Environment

Page 9 of 45
Journal of Medicinal Chemistry
1
2
3
4
Figure 4. Strategies to address hERG binding negatively impacted doseꢀrelate parameters
5
6
7
8
9
Although compound 2 possessed adequate properties to achieve proof of concept and demonstrate
in vivo functional selectivity, the challenges around modification of the basic amine to attenuate hERG and
lower dose motivated the team to remove the amine entirely. This led to the identification of truncated
compound 6 with high ligand binding efficiency (LBE = 0.50).17 Removal of the basic amine completely
ablated hERG binding activity and provided an encouraging starting point for optimization despite
relatively high clearance parameters (Figure 5). As such, the team embarked on a design strategy to
simultaneously optimize multiple parameters by improving the physicochemical properties while
maintaining efficient drugꢀenzyme interactions.18 To that end, lipophilic ligand efficiency (LLE) was
employed as a key parameter to aid in the design process and track compound progression.19 Through this
approach, tetrahydropyran 7 was discovered with the high LLE (7.9) that reflected the significant
improvement in potency through the introduction of a polar oxygen heteroatom (Figure 5). The subtle
decrease in polarity (HPLC Log D = 2.08) also translated to improved Clint in both rat and human (r; h Hept
Clint = 70; 6 mL/min) and a large decrease in unbound in vivo rat clearance. This improved approach
focused on physicochemical property considerations to simultaneously address offꢀtarget liabilities as well
as refine key parameters central to drive down dose.
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
JAK1 IC50 (nM)
LBE
3
0.4
0.50
0.56
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 45
1
2
3
4
LLE
5.5
7.9
5
6
7
8
9
HPLC Log D pH 7.0
MK499 IC50 (nM)
Rat Clp (mL/min/kg)
2.66
>60000
484
2.08
>60000
20
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
Unbound Rat Cl (mL/min/kg 12100
110
r; h Hept Clint (mL/min/kg)
160; 25
70; 6
Figure 5. Minimal pharmacophore has high LBE and LLE
Tetrahydropyran pyrazole analogs can efficiently be synthesized in a few short steps (Scheme 1).
Tetrahydroꢀ2Hꢀpyranꢀ4ꢀone (8) was converted to carbonitrile (9) by first forming the cyanohydrin with
TMSꢀCN and TMSꢀOTf followed by subsequent treatment of the cyanohydrin with POCl3 and elimination.
Conjugate addition with pyrazole carboxamides induced by DBU afforded racemate 10, which was
separated by chiral SFC to afford enantiomerically pure 10 in 14% overall yield. With enantiomerically pure
intermediate 10 in hand, Buchwald coupling with various halogenated arenes using Pd2(dba)3 and tꢀBu Xꢀ
Phos smoothly and selectively arylated the 3ꢀamino pyrazole nitrogen.20
Scheme 1. Synthesis of tetrahydropyran pyrazole analogs: a) i) TMSꢀCN, TMSꢀOTf, 0 °C ii) POCl3,
ACS Paragon Plus Environment

Page 11 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pyridine, 0 °C, CH2Cl2, 60% (two steps); b) i) 3ꢀaminoꢀ4ꢀpyrazole carboxamide, DBU, 70 °C, EtOH ii)
chiral SFC separation of active enantiomer 14% (two steps); c) tꢀBu XPhos, KOAc, Pd2(dba)3, 60 °C,
iPrOH, 30ꢀ80%. Trimethylsilyl cyanide (TMSꢀCN), Trimethyl silyl trifluoromethansulfonate (TMSꢀOTf),
Phosphoryl chloride (POCl3), 1,8ꢀDiazabicycloundecꢀ7ꢀene (DBU), 2ꢀDiꢀtertꢀbutylphosphinoꢀ2’,4’,6’ꢀ
triisoporopylbiphenyl (tꢀBu XPhos).
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
To confirm that the new chemotype was indeed a superior series, the analogs in the tetrahydropyran
chemotype (11) were compared to those in the earlier cylohexylamine chemotype (12) using a pie chart
visualization analysis (Figure 6).21 Consistent with the comparison in Figure 5, the tetrahydropyran
chemotype generated a higher percentage of analogs with LLE >0.5 (87.7%) compared to the
cyclohexylamine chemotype (14.5%). The increase in LLE translated to an enrichment of tetrahydropyran
analogs with a cleaner hERG profile (79% with MK499 IC50 > 20,000 nM vs. 41% cyclohexylamine series)
and increased metabolic stability (64% with Clint <20 mL/min/kg vs. 32% in the cyclohexylamine series).
Bioavailability did not improve between the two series. Taken together, these analyses highlighted the
enrichment of overall compound quality toward our goals of reduced hERG liability and lower projected
human dose. With an efficient template in hand, we next undertook a systematic exploration of the solvent
exposed aromatic ring to further optimize on LLE and understand its impact on the PK parameters,
specifically rat and human Clint, bioavailability in rat, and FASSIF solubility.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 45
1
2
3
4
5
6
7
8
9
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 6. Binned analysis comparing LLE, MK499 IC50 distribution, human hepatocytes Clint, and F%
between tetrahydropyran and cyclohexylamine chemotypes. Filter settings: JAK1 Enzyme IC50 <100 nM,
Cell IL6 IC50 <1000 nM, 897 samples for LLE and MK499, 350 samples for human hepatocytes Clint, 88
samples for F%.
From the data shown in Table 1, lipophilic substituents on the aromatic ring such as compounds 14,
15, 16, and 17 possessed good LLE and acceptable bioavailability; however, the increase in lipophilicity
was associated with lower FASSIF solubility and higher rat and human intrinsic clearance. On the other
hand, polar substituents containing heteroatoms (18, 19, 20, 21) generally improved rat and human intrinsic
clearance (19 r; h Hept Clint = 4.8; 2.5 mL/min/kg, 20 r; h Hept Clint = 5; 1.5 mL/min/kg, and 21 r; h Clint =
<40; <20 mL/min/kg). Optimizing clearance by reducing lipophilicity was consistent with established
ACS Paragon Plus Environment

Page 13 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
medicinal chemistry strategies.22 Additionally, the added heteroatom polarity contributed to the increased
FASSIF solubility as demonstrated by compound 20 (FASSIF = 159 uM). Despite the benefits of increased
polarity in improving intrinsic clearance and solubility, the significant lower Log D and polarity negatively
impacted the rat bioavailability with compounds 18, 19, 20, and 21 showing F% <20. Ultimately,
optimizing for PK properties proved to be a significant challenge since the desired physicochemical profile
for low Clint and high F% did not align.
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
To overcome the PK obstacle, we visualized the issue by plotting the impact of HPLC Log D and
bioavailability and hClint. Figure 7 highlights the challenge of balancing the conflicting relationship of
lipophilicity and polarity. As HPLC Log D increased, the frequency of compounds with F% >20 increased
at the expense of compounds with hClint <20 mL/min/kg. The plot also revealed that HPLC Log D 1.50ꢀ
2.50 was the optimal range for both bioavailability and hClint properties. This analysis provided a targeted
range to enable design of inhibitors in the desired property space.
Table 1: Structure activity relationship of tetrahydropyrans
Compound
R
JAK1; 2; 3;
Ratio
JAK 2/
1
Cell IL6;EPO
IP (nM)b
LLEc
MK499
Hepatocyte
F (%)f HPLC
LogD
FASSIF(uM)
TYK2 IC50(nM)
IC50(nM)d ClintmL/min/
a
kg
pH7.0g
(rat;human)g
13
0.4; 3; 361;
2.9
8X
33; 209 (6X)
7.90
>60000
70; 6
79
2.08
18
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 45
1
2
3
4
5
6
7
8
9
14
15
16
0.13; 1.5; 256;
2
12X
14; 140 (10X) 7.72
40000
40000
35000
9; 85
71
2.57
2.18
2.58
39
0.3; 2; 304;
1.8
7X
31; 334 (11X) 7.82
<40;
40
42
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
<20
0.5; 1.9; 280;
3.9
4X
37; 200 (5X)
6.98
262; 41
NA
NA
17
0.1; 0.8; 196;
0.7
11X
5; 54 (11X)
7.69
25000
59; 28
42
2.27
NA
18
0.02; 0.5; 91;
0.2
23X
6; 39 (7X)
8.03
>60000
55; <20
18
1.48
42
19
0.06; 0.7; 116;
0.5
11X
23; 254 (11X) 8.96
13200
4.8; 2.5
3
1.00
96
20
0.24; 2.8; 511;
3
10X
42; 414 (10X) 8.68
>60000
5; 1.5
19
1.18
159
ACS Paragon Plus Environment

Page 15 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
21
0.05; 0.47;
100; 0.4
9X
14; 48 (3X)
8.37
>60000
<40;
<20
9
1.93
NA
6
7
8
9
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
22
0.15; 1.2;
230; 1.2
8X
13; 89 (7X)
8.44
40000
6; 3
47
1.77
130
aValues in this table are determined by the HTRF assay and are the means from at least n = 2 experiments.
bValues in this table are determined by GeneBLAzer assay and are the means from at least n = 2
experiments. cValues are calculated from pIC50 – ALogP98. dhERG binding assay, ref 13. Dose; rat iv 0.5
f
mg/kg as a solution in PEG400:H2O (60:40 (v/v)), po: 1 mg/kg as a solution in PEG400:H2O (60:40 (v/v)).
gSee supporting information. NA = not available.
With a target physicochemical range in mind, we turned our attention to identifying SAR that fit
within the desired HPLC Log D range. The introduction of a nitrile is a useful functional group that
embodies polar features without significantly reducing Log D.23 To explore this concept, 22 was prepared
and exhibited a balanced HPLC Log D (1.77) without compromising LLE (8). This balanced profile
translated to a low r; h intrinsic clearance (r; h Hept Clint = 6; 3 mL/min/kg) with a desired aqueous
solubility (FASSIF = 130 uM) without compromising the bioavailability (F% = 47).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 45
1
2
3
4
5
6
7
8
9
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 7. Analysis of %F and Clint frequency plotted against HPLC Log D. HPLC Log D between 1.50ꢀ
2.50 is the optimal range. Filter settings: JAK1 Enzyme IC50 <100 nM, Cell IL6 IC50 <1000 nM.
The favorable profile of 22 warranted further animal and human ADME data as well as human
whole blood potency to inform on human dose prediction (Table 2).9 Gratifyingly, in vitro Clint correlated
well with the in vivo unbound clearance (Clu) across species, allowing for confidence in human in vitro in
vivo extrapolation (IVIVE). Allometric scaling and translatable exposure to higher order species combined
with the human whole blood potency (HWB IL7 IC50 = 220 nM) informed to project the predicted human
dose for 22 at 15 mg QD with an excellent D0 = 2. The excellent dose and dose number for 22 validated our
approach of optimizing on intrinsic potency, Clint, solubility, and bioavailability in order to identify a low
dose lead molecule.
ACS Paragon Plus Environment

Page 17 of 45
Journal of Medicinal Chemistry
1
2
3
4
Table 2. Pharmacokinetic parameters of compounds 22 and 28 in preclinical species
5
6
7
Rata
Dogb
Cynoc
Predicted Humand
8
9
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
22
19
6
28
21
22
12
20
82
67
28
14
22
2.4
6
28
19
22
0.9
3
28
0.6
6
Clp (mL/min/kg)
e
Hept Clint
35
122
NA
NA
36
PPB %
Cl unbound
(mL/min/kg)
Vdss (L/kg)
t1/2 (hr)
81
100
94
79
12
96.7
581
78
4
97.8
27
350
2.6
1.4
47
2.7
1.6
47
2.7
2.6
51
1.2
1.1
29
1
2.4
2.8
32
2.3
30
72
1
5.6
18
20
56
F (%)
Human Wholee
Blood (nM)
220
1760
Dose (QD) (mg)d
Dose Numberf
15
2
81
25
aDose; rat iv 0.5 mg/kg as a solution in PEG400:H2O (60:40 (v/v)), po: 1 mg/kg as a solution in
b
PEG400:H2O (60:40 (v/v)). Dose dog iv 0.25 mg/kg as a solution in DMSO:PEG400:water(20:60:20),
c
cyno PO 0.5 mg/kg as a solution in PEG400/TWEEN80/H2O(20:60:20) Dose; cyno iv 0.25 mg/kg as a
d
solution in DMSO:PEG400:water(20:60:20), cyno PO 0.5 mg/kg as a solution in captisol 30%. See
reference 9 eSee supporting information. f See reference 11a.
Despite the successful approach of focusing on physicochemical properties and dose to identify 22
with favorable drug like properties, the optimization process eroded the JAK1 pathway selectivity of 22 (7ꢀ
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 45
1
2
3
4
5
6
7
8
9
fold), precluding further development of the molecule. During the rapid SAR screening process of the
aromatic vector of the inhibitor, compound 23 with 50ꢀfold selectivity for JAK1 over JAK2 was identified.
Examining the structural uniqueness of 23, we hypothesized that projecting into the southeast region of the
aromatic ring might account for the enhanced selectivity. To further test this hypothesis and exploit this
potential unique interaction, we combined the findings from Table 1 with SAR at the southwest region of
the aromatic ring (Table 3). Consistent with the same strategy as in Table 1, we were mindful of optimizing
for selectivity while maintaining desirable LLE in order to refine PK parameters that related to lowering
dose. Using 22 as a baseline with 8ꢀfold selectivity in enzyme activity and 7ꢀfold in cellular activity,
introduction of a meta methoxy group (24) enhanced the JAK2/JAK1 selectivity to 46ꢀfold in the enzyme
and 23ꢀfold in the pathway assay. Additional changes including Cl, CN, and cyclopropyl as exemplified by
25, 26, 27 led to modest improvements in selectivity which suggest the unique positioning of a methoxy
group was critical for selectivity. Additionally, this SAR was translatable to other promising aromatic
analogs such 28, 29, 30. Notably, accessing the same vector by constraining small substituents using fused
rings led to even greater selectivity as demonstrated by 31, 32, and 33, achieving >50ꢀfold in enzyme and
>80ꢀfold selectivity in cell assays. Despite the exquisite selectivity from the fused ring analogs, these
compounds did not possess the desired in vitro and in vivo PK properties to progress further.
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 new meta substituent discovery expanded our SAR resulting in numerous highly selective
compounds with desirable PK properties. Notably, the balanced properties of compound 28 with a para
chloro and meta methoxy not only imparted improved JAK1 selectivity (39ꢀfold enzyme and 25ꢀfold cell)
but also retained favorable rat Clu (350 mL/min/kg) with low hept hClint (6 mL/min/kg) and desirable
bioavailability (F% = 47). The overall selectivity and dose parameters for 28 were superior to other analogs
and consequently progressed to full PK determinations (Table 2). Evaluation of 28 in multiple species
across a panel of studies revealed a good correlation between in vitro hept Clint and in vivo Clu, leading to a
confident low human predicted total clearance of 0.6 mL/min/kg with a halfꢀlife of 20 hr and bioavailability
of 56%. Furthermore, the predicted human PK properties combined with the human whole blood (IC50 =
1760 nM) led to the predicted human dose of 81 mg QD with a D0 = 25.
ACS Paragon Plus Environment

Page 19 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
Table 3. Structure activity relationship of tetrahydropyrans
8
9
Cell IL6; EPOb
IP (nM)b
LLEc
Hepatocyte
ClintmL/min
kg (rat;
F (%)d HPLC Log
d
Compound
R
Rat Clu
JAK1; 2; 3; TYK
2 IC50 (nM)a
Ratio JAK2;
JAK1
pH 7.0e
mL/min/kg
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
human)e
23
0.04; 2;
315; 1.7
50X
16; 224 (14X)
9.65
535
62; 62
NA
3.15
24
25
26
0.07; 3.2;
391; 9
46X
11X
6X
19; 430 (23X)
8.79
7.95
7.79
310
NA
NA
<40; 24
2
1.87
2.34
1.99
1.4; 16;
72; 744 (10X)
NA
NA
NA
>1496; 40
0.9; 5; 656;
11
77; 1192 (15
X)
NA
27
28
0.4; 2.7; 556;
4
7X
26; 261 (10X)
7.35
7.85
268
350
NA
23
47
2.18
2.34
0.1; 3.9; 397;
12
39X
13; 320 (25X)
35; 6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 45
1
2
3
4
5
29
0.18; 3.4; 701
; 13
19X
28; 530 (19X)
8.11
522
9; 4
4
2.07
6
7
F
N
N
8
9
30
31
0.4; 6; 1130;
13
15X
53X
39; 600 (15X)
7.94
7.85
108
500
14; 2
49
1.89
1.40
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
0.19; 10;
>1496; 8
24; 2114
(88x)
15; 15
NA
OMe
F
N
32
33
0.24; 7; 1012;
12
29X
43X
14; 608 (43X)
7.06
8.45
1212
52; 25
19
1.81
1.02
0.16; 7; 783;
14
24; 1635
(68X)
NA
NA
NA
aValues in this table are determined by the HTRF assay and are the means from n = 2 experiments. Values
in this table are determined by GeneBLAzer assay and are the means from at least n = 2 experiments.
cValues are calculated from pIC50 – ALogP98. dDose; rat iv 0.5 mg/kg as a solution in PEG400:H2O (60:40
b
e
(v/v)), po: 1 mg/kg as a solution in PEG400:H2O (60:40 (v/v)). See supporting information. NA = not
available.
Having significantly altered the structure from starting point 2, the off target kinase profile of 28
was assessed in a broad kinase panel consisting of 265 kinases.24 Kinase selectivity of 2, 22, and 28 is
depicted using a radar plot where the distance from the center is proportional to the fold selectivity over
JAK1 in log10 units. As shown in Figure 8, 28 was more selective and differentiated from 2 since its curve
ACS Paragon Plus Environment

Page 21 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
lies outside that of 2. In addition to the radar plot, several additional metrics were also used to analyze
kinase selectivity from different perspectives. While the percentage of kinases with at least 100ꢀfold
selectivity is a common metric used to understand selectivity it can be misleading if the few kinases under
100ꢀfold are equally potent or more potent than the target kinase. Alternatively, the partition index method
can be used to quantify the fraction of the inhibitor that would be bound to the target of interest at
thermodynamic equilibrium, with values closer to one indicating that the ligand is highly selective and
binds only to the desired target.25 Lead compound 28 is a highly selective for JAK1 inhibitor demonstrating
greater than 100ꢀfold selectivity over 99% of the 265 kinases tested with a partition index of 0.94 (Figure 8).
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
Compounds
% Kinases >100 X Partition Index
Over JAK1
98
98
99
0.72
0.78
0.94
2
22
28
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 45
1
2
3
4
Figure 8. Analysis of kinase selectivity profile of compounds 2, 22, 28 against 265 kinases using a radar
5
6
plot, percent of kinases at least100ꢀfold selectivity over JAK1 and partition index.
7
8
9
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 9. Coꢀcrystal structure of 28 in JAK1 kinase domain showing ligand protein interactions (PDB
5WO4)
In order to gain greater clarity on the proteinꢀligand molecular interactions and the nature of the
JAK isoform selectivity, crystallographic studies of 28 were examined. Figure 9 shows compound 28
binding to the active site of the JAK1 kinase domain.26 The primary amide anchors the ligand by making
hydrogen bonds to the backbone carbonyl oxygen of Glu957 and the backbone amide NH of Leu959 in the
hinge region of the kinase. Additionally, the primary amide is involved in a water molecule (H2Oꢀ412 in
PDB 5WO4) mediated hydrogen bond interaction with the backbone carbonyl oxygen atom of Gly1020 in
the back pocket. The para position of the aniline projects out to solvent, consistent with the flexible SAR
ACS Paragon Plus Environment

Page 23 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
observed in this region of the molecule. Furthermore, the cyano substituted tetrahydropyran ring forces an
increased dihedral angle resulting in hydrophobic packing in two deep pockets. The tetrahydropyran oxygen
is buried under the glycine rich loop (omitted for clarity in Figure 9); however, this interaction is poorly
understood because simple van der Waals interactions do not satisfyingly explain the 10ꢀfold decrease in
potency when lipophilic cyclohexane is substituted for the tetrahydropyran. Furthermore, the twisted
conformation of the tetrahydropyran projects the nitrile group into a small hydrophobic backꢀpocket created
by Leu1010 and Gly1020 unique to JAK family kinases. The combination of these unique interactions in
conjunction with the conserved water molecule H2Oꢀ412 mediated hydrogen bond confers the excellent
JAK family kinase selectivity.
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 10. Comparison of 28 contacts with residues Glu966 in JAK1 and Asp939 in JAK2. In the crystal
structure of 28 bound to JAK1, the dashed lines correspond to van der Waals contacts between the ligand
and Glu966 identified with the ViewContacts software. No van der Waals interactions were identified when
28 was modeled into the active site of JAK2 (PDB 3LPB).
To determine the nature of the JAK isoform selectivity, comparison of the xꢀray crystal structures of
28 bound to human JAK1 and overlaid on a JAK2 crystal structure revealed a unique 3ꢀmethoxy group
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 45
1
2
3
4
5
6
7
8
9
interaction with a single amino acid difference between JAK1 and JAK2.27 In particular, the 3ꢀmethoxy
group interacts with the longer side residue Glu966 in JAK1 through van der Waals interactions, whereas in
JAK2 the equivalent residue (Asp939) is positioned further away from any direct interaction (Figure 10).28
Interestingly, this finding is consistent with the observations of Kim et al and Zak et al, which led to the
elegant designs of selective JAK1 inhibitors.6d,f Overall, our findings along with others suggest that
exploiting the Glu966 residue in JAK1 is a general and consistent strategy to optimize for JAK1 selectivity
over JAK2.
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 11. Compound 22 dosed PO BID at 0.1, 1, 10 mg/kg and 28 PO BID at 0.3, 3, 10, 30 mg/kg were
efficacious in reducing paw swelling in a 30 day rat CIA model. Dexamethasone was included as a positive
control. .*p < 0.005, statistically different from nonꢀarthritic controls, ^p < 0.005, statistically different from
CIA vehicle group.
On the basis of the favorable lead compound profile, in vivo properties, and kinase selectivity,
compounds 22 and 28 were studied in a rat collagen induced arthritis model (CIA)29 using a therapeutic
dosing paradigm (Figure 11).30 Rats were immunized and boosted with collagen intradermally at the base of
the tail on days one and seven. Animals were stratified into treatment groups with comparable levels of paw
ACS Paragon Plus Environment

Page 25 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
inflammation on day 16 and treatment with compounds 22 and 28 (p.o. BID) and dexamethasone (p.o QD)
was initiated on day 17. Both compounds exhibited dose dependent inhibition of inflammation. The
maximum attenuation of paw swelling with compounds 22 and 28 was achieved with 10 and 30 mg/kg
respectively, which was comparable to dexamethasone treated CIA non arthritic control groups. Compound
28 demonstrated that selectively inhibiting the JAK1 pathway is sufficient to drive efficacy in a preclinical
inflammation model and suggests that JAK1 selective compounds may offer an advantage by offsetting the
potential anemia side effects of inhibiting JAK2.
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
Conclusion
In conclusion, while the lead compound 2 in the cyclohexylamine series served as a useful tool to
probe the effects of JAK1 selectivity, further progression of that chemotype was challenging due to hERG
activity and high predicted human dose. To address these issues, we attempted to optimize and differentiate
analogs in that series based on the parameters associated with predicted human dose including intrinsic
potency, Clint, bioavailability, and solubility. Multiparameter optimization on compound 2 proved to be a
difficult endeavor due to opposing SAR between decreasing hERG and PK parameters. A shift in strategy
toward a new core with the tetrahydropyran chemotype followed by LLE guided optimization produced an
enriched cohort of analogs with reduced hERG signal and improved PK parameters. In conjunction with
these advances, structural chemistry information provided key insights in guiding modifications to improve
JAK1 selectivity, in particular revealing a key amino acid difference between JAK1 and JAK2. By
exploiting selectivity advancements in combination with the improved tetrahydropyran core, a significant
enrichment of JAK1 selective compounds with the desired drug like properties and low predicted human
dose were synthesized. These efforts culminated in the discovery of JAK1 selective compound 28, which
has an excellent predicted human dose projection of 81 mg QD and is highly active in our CIA
inflammation model. Furthermore, the pathway selectivity of 28 allowed for further pharmacological
investigation into delineating an improved therapeutic index with a JAK1 selective compound. This data
will be reported in a future communication.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 45
1
2
3
4
5
Experimental Section
6
7
8
9
General experimental methods. Commercial reagents were obtained from reputable suppliers and used as
received. All solvents were purchased in septum−sealed bottles stored under an inert atmosphere. All
reactions were sealed with septa through which an argon atmosphere was introduced unless otherwise noted.
Liquid reagents and solvents were transferred under a positive pressure of nitrogen via syringe. Reactions
were conducted in microwave vials or round bottomed flasks containing Teflon−coated magnetic stir bars.
Microwave reactions were performed with a Biotage Initiator Series Microwave (fixed hold time setting;
reaction temperatures monitored by the internal infrared sensor). Reactions were monitored by thin layer
chromatography (TLC) on pre−coated TLC glass plates (silica gel 60 F254, 250 µm thickness) or by
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
LC/MS (30 mm x 2 mm 2 ꢀm column + guard; 2 ꢀL injection; 3% to 98% MeCN/water + 0.05% TFA
gradient over 2.3 minutes; 0.9 mL/min flow; ESI; positive ion mode; UV detection at 254 nm).
Visualization of the developed TLC chromatogram was performed by fluorescence quenching. Flash
chromatography was performed on an automated purification system using pre−packed silica gel columns.
1H NMR spectra were recorded on either a 500 or a 600 MHz Varian spectrometer; chemical shifts (δ) are
reported relative to residual proton solvent signals. Data for NMR spectra are reported as follows: chemical
shift (δ ppm), multiplicity (s = singlet, brs = broad singlet, d = doublet, t = triplet, q = quartet, dd = doublet
of doublets, td = triplet of doublets, m = multiplet), coupling constant (Hz), integration. All final
compounds possess a purity of at least 95% as determined by 1H NMR, LCMS, HRMS.
3,6-dihydro-2H-pyran-4-carbonitrile (9)
To the solution of trimethylsilyl cyanide (105 g, 1080 mmol) in DCM (370 mL) were added tetrahydroꢀ4Hꢀ
pyranꢀ4ꢀone (90 g, 900 mmol) and trimethylsilyl triflate (6 g, 27 mmol) at 0 oC. The resulting mixture was
o
stirred at 0 C for 1 hour before the addition of pyridine (1120 mL) and phosphoryl chloride (413 g, 2700
mmol). The mixture was refluxed for 12 hours, and then poured into the mixture of 2N hydrochloric acid
ACS Paragon Plus Environment

Page 27 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
aqueous solutions (1.5 L), and extracted with EtOAc (3*2 L). All the organic solutions were washed with
brine (2*1000 mL), dried over sodium sulfate, filtered and concentrated under vacuum. The crude residue
was purified by column chromatography (eluted with PE / EtOAc = 10 / 1) to give 3,6ꢀdihydroꢀ2Hꢀpyranꢀ4ꢀ
carbonitrile (59 g, 60.1 %) as a yellow oil. 1H NMR (300 MHz, CDCl3) δ 6.62 – 6.59 (m, 1H), 4.29 – 4.21
(m, 2H), 3.78 (t, J = 5.4 Hz, 2H), 2.34 – 2.30 (m, 2H).
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
3-amino-1-((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-1H-pyrazole-4-carboxamide (10)
The solution of 3ꢀaminoꢀ1Hꢀpyrazoleꢀ4ꢀcarboxamide (85 g, 671 mmol), 3,6ꢀdihydroꢀ2Hꢀpyranꢀ4ꢀ
o
carbonitrile (205 g, 1.88 mol) and DBU (234 g, 1.54 mol) in ethanol (850 mL) was stirred at 70 C
overnight under nitrogen and then concentrated under vacuum. The crude residue was purified by silica gel
flash column chromatography (eluted with DCM / Methanol = 30 / 1) to yield racemic compound 10 (47 g,
29.8 %) as a yellow solid. The chiral separation of 380 g of the racemic compound was accomplished by
dissolving in ACN/MeOH (1:1) to a concentration of 25 mg/mL. Injections of 16 mL were made on a Thar
350 preparative SFC (Column: ChiralPak ICꢀ10 ꢁM, 300x50 mm; Mobile phase: 45% 2ꢀpropanol, 55%
CO2; Flow rate: 220 mL/min; Column temperature: 38 °C). After separation, the fractions were dried by
rotary evaporation to give 3ꢀaminoꢀ1ꢀ((3S,4R)ꢀ4ꢀcyanotetrahydroꢀ2Hꢀpyranꢀ3ꢀyl)ꢀ1Hꢀpyrazoleꢀ4ꢀ
carboxamide (174 g, 48.8 %, 99.6 % ee) from the first (faster eluting) peak and 3ꢀaminoꢀ1ꢀ((3R,4S)ꢀ4ꢀ
cyanotetrahydroꢀ2Hꢀpyranꢀ3ꢀyl)ꢀ1Hꢀpyrazoleꢀ4ꢀcarboxamide (181 g, 47.6%, 99.6% ee) from the second
(slower eluting) peak. The absolute configuration of stereochemistry was determined from Xꢀray
crystallography of compound 28 with JAK1 protein. The second (slower eluting) peak was carried on to to
generate SAR. 1H NMR (500 MHz, DMSOꢀd6) δ 8.03 (s, 1H), 7.36 (brs, 1H), 6.80 (brs, 1H), 5.36 (s, 2H),
4.86ꢀ4.31 (td, J = 10.5, 4.5 Hz, 1H), 3.91ꢀ3.88 (dd, J = 11.5, 4.5 Hz 1H), 3.86 – 3.83 (m, 1H), 3.53ꢀ3.50
13
(m, 2H), 3.39ꢀ3.33 (td, J = 11.5, 2 Hz, 1H), 2.10ꢀ2.07 (m, 1H), 1.95ꢀ1.87 (m, 1H). C NMR (500 MHz,
DMSOꢀd6): δ 166.3, 157.2, 131.6, 120.2, 101.2, 69.8, 65.8, 57.8, 31.9, 28.5. LRMS (ESI) calc'd for
C10H14N5O2 [M+H]+: 236, Found: 236.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 45
1
2
3
4
1-((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-3-(phenylamino)-1H-pyrazole-4-carboxamide (13).
5
6
7
8
Prepared following similar procedure described for 28 in 57% yield. LRMS (ESI) calcd for (C16H17N5O2)
[M + H]+ 312.1, found 312.1. HRMS: calcd for (C16H17N5O2) [M + H]+ 312.1460, found 312.1451;
difference 2.9 ppm. 1H NMR (500 MHz, CD3OD) δ 8.17 (s, 1H), 7.51ꢀ7.50 (d, J = 8.7 Hz, 2H). 7.26ꢀ7.23 (t,
J = 16.1, 8.1 Hz, 2H), 6.87ꢀ6.84 (t, J = 14.5, 7.4 Hz, 1H), 4.44ꢀ4.39 (m, 1H), 4.12ꢀ4.08 (m, 1H), 4.01ꢀ3.97
(m, 1H), 3.90ꢀ3.86 (m, 1H), 3.72ꢀ3.66 (m, 1H), 3.60ꢀ3.55 (m, 1H), 2.23ꢀ2.18 (m, 1H), 2.09ꢀ2.01 (m, 1H). 13
C NMR (500 MHz, DMSOꢀd6): δ 166.4, 153.05, 141.7, 131.6, 129.4 (2C), 120.23, 120.03, 116.4 (2C), 101.
9
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
20
3, 69.4, 65.8, 58.0, 31.9, 28.4. [α]D +145 (c 0.83, MeOH/DMSO 2:1).
3-((4-chlorophenyl)amino)-1-((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-1H-pyrazole-4-carboxamide
(14). Prepared following similar procedure described for 28 in 45% yield. LRMS (ESI) calcd for
(C16H16ClN5O2) [M + H]+ 346.1, found 346.1. HRMS: calcd for (C16H16ClN5O2) [M + H]+ 346.1071, found
1
346.1063; difference 2.3 ppm. H NMR (500 MHz, CDCl3) δ 8.83 (s, 1H), 7.77 (s, 1H), 7.47ꢀ7.45 (d, J =
8.7 Hz, 2H), 7.25ꢀ7.23 (d, J = 9 Hz, 2H), 5.48 (brs, 2H), 4.25ꢀ4.21 (m, 1H), 4.19ꢀ4.15 (m, 1H), 4.06ꢀ4.02
(m, 2H), 3.68ꢀ3.64 (m, 1H), 3.60ꢀ3.56 (m, 1H), 2.17ꢀ2.14 (m, 1H), 2.07ꢀ1.99 (m, 1H).
1-((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-3-((4-fluorophenyl)amino)-1H-pyrazole-4-carboxamide
(15). Prepared following similar procedure described for 28 in 40% yield. LRMS (ESI) calcd for
1
(C16H16FN5O2) [M + H]+ 330.1, found 330.1. H NMR (600 MHz, DMSOꢀd6) δ 9.10 (s, 1H), 8.25 (s, 1H),
7.67 (brs, 1H), 7.52ꢀ7.48 (m, 2H), 7.13 (brs, 1H), 7.06ꢀ7.03 (t, J = 8.9 Hz, 2H), 4.54ꢀ4.49 (m, 1H), 3.99ꢀ
3.96 (m, 1H), 3.88ꢀ3.86 (m, 1H), 3.66ꢀ3.61 (m, 2H), 3.47ꢀ3.42 (m, 1H), 2.12ꢀ2.09 (m, 1H), 1.97ꢀ1.91 (m,
1H).
1-((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-3-((3-cyclopropylphenyl)amino)-1H-pyrazole-4-
carboxamide (16).Prepared following similar procedure described for 28 in 17% yield. LRMS (ESI) calcd
for (C19H21N5O2) [M + H]+ 352.2, found 352.2. HRMS: calcd for (C19H21N5O2) [M + H]+ 352.1773, found
ACS Paragon Plus Environment

Page 29 of 45
Journal of Medicinal Chemistry
1
2
3
1
352.1767; difference 1.7 ppm. H NMR (500 MHz, DMSOꢀd6) δ9.08 (s, 1H), 8.28 (s, 1H), 7.71 (brs, 1H),
7.33ꢀ7.32 (d, J = 7.8 Hz, 1H), 7.16 (brs, 1H), 7.13ꢀ7.10 (t, J = 7.8 Hz, 2H), 6.57ꢀ6.56 (d, J = 7.3 Hz, 1H),
4.57ꢀ4.53 (m, 1H), 4.05ꢀ4.02 (m, 1H), 3.92ꢀ3.90 (m, 1H), 3.68ꢀ3.61 (m, 2H), 3.48ꢀ3.43 (m, 1H), 2.17ꢀ2.15
(m, 1H), 2.01ꢀ1.95 (m, 1H), 1.91ꢀ1.86 (m, 1H), 0.93ꢀ0.92 (d, J = 7.44 Hz, 2H), 0.67 (brs, 2H).
4
5
6
7
8
9
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-((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-3-((4-(difluoromethoxy)phenyl)amino)-1H-pyrazole-4-
carboxamide (17). Prepared following similar procedure described for 28 in 17% yield. LRMS (ESI) calcd
for (C17H17F2N5O3) [M + H]+ 378.2, found 378.2. HRMS: calcd for (C17H17F2N5O3) [M + H]+ 378.1377,
found 378.1379; difference ꢀ0.5 ppm. 1H NMR (500 MHz, DMSOꢀd6) δ 9.16 (s, 1H), 8.27 (s, 1H), 7.72 (brs,
1H), 7.56ꢀ7.54 (d, J = 7.8 Hz, 2H), 7.18 (brs, 1H), 7.22ꢀ6.92 (dd, J = 73, 6.8 Hz, 1H), 7.07ꢀ7.06 (d, J = 6.8
Hz, 2H), 4.57ꢀ4.52 (m, 1H), 4.02ꢀ3.99 (m, 1H), 3.90ꢀ3.88 (m, 1H), 3.69ꢀ3.63 (m, 2H), 3.49ꢀ3.44 (m, 1H),
2.14ꢀ2.12 (m, 1H), 2.00ꢀ1.92 (m, 1H).
1-((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-3-((4-(1-methyl-1H-pyrazol-4-yl)phenyl)amino)-1H-
pyrazole-4-carboxamide (18). Prepared following similar procedure described for 28 in 10% yield. LRMS
(ESI) calcd for (C20H21N7O2) [M + H]+ 392.2, found 392.2. 1H NMR (500 MHz, DMSOꢀd6) δ 9.55 (s, 1H),
8.94ꢀ8.93 (d, J = 2.6 Hz, 1H), 8.68ꢀ8.67 (d, J = 2.3 Hz, 1H), 8.30 (s, 1H), 7.95ꢀ7.93 (d, J = 8.8 Hz, 1H),
7.92ꢀ7.91 (d, J = 1.9 Hz, 1H), 7.76 (brs, 1H), 7.55ꢀ7.52 (dd, J = 8.7, 2.1 Hz, 1H), 7.26 (brs, 1H), 4.46ꢀ4.41
(m, 1H), 3.44 (s, 3H), 2.19ꢀ2.17 (m, 1H), 2.03ꢀ1.99 (m, 1H), 1.94ꢀ1.86 (m, 1H), 1.83ꢀ1.79 (m, 1H), 1.76ꢀ
1.71 (m, 2H), 1.48ꢀ1.38 (m, 2H).
3-((4-(1H-1,2,3-triazol-1-yl)phenyl)amino)-1-((3R,4S)-4-cyanotetrahydro-2H-pyran-3-yl)-1H-
pyrazole-4-carboxamide (19). Prepared following similar procedure described for 28 in 9% yield. LRMS
(ESI) calcd for (C18H18N8O2) [M + H]+ 379.1, found 379.1. 1H NMR (500 MHz, DMSOꢀd6) δ 9.39 (s, 1H),
8.70 (d, J = 0.93 Hz, 1H), 8.31 (s, 1H), 7.91 (d, J = 0.93 Hz, 1H), 7.76 (brs, 1H), 7.77ꢀ7.75 (d, J = 9.1 Hz,
2H), 7.73ꢀ7.71 (d, J = 9.1 Hz, 2H), 7.24 (brs, 1H), 4.61ꢀ4.56 (m, 1H), 4.05ꢀ4.02 (m, 1H), 3.93ꢀ3.89 (m, 1H),
ACS Paragon Plus Environment