Design and optimization of a series of 4-(3-azabicyclo[3.1.0]hexan-3-yl)pyrimidin-2-amines: Dual inhibitors of TYK2 and JAK1

Article abstract of DOI:10.1016/j.bmc.2020.115481

Herein, we disclose a new series of TYK2/ JAK1 inhibitors based upon a 3.1.0 azabicyclic substituted pyrimidine scaffold. We illustrate the use of structure-based drug design for the initial design and subsequent optimization of this series of compounds. One advanced example 19 met program objectives for potency, selectivity and ADME, and demonstrated oral activity in the adjuvant-induced arthritis rat model.

Full text of DOI:10.1016/j.bmc.2020.115481

Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Design and optimization of a series of 4-(3-azabicyclo[3.1.0]hexan-3-yl)
pyrimidin-2-amines: Dual inhibitors of TYK2 and JAK1
⁎
Andrew Fensomea, , Catherine M. Amblerb, Eric Arnoldb, Mary Ellen Bankerb, James D. Clarkc,
Martin E. Dowtya, Ivan V. Efremova, Andrew Flickb, Brian S. Gerstenbergera, Roger S. Giffordc,
Ariamala Gopalsamya, Martin Hegenc, Jason Jussifc, David C. Limburgb, Tsung H. Linc,
Betsy S. Pierceb, Raman Sharmab, John I. Trujillob, Felix F. Vajdosb, Fabien Vincentb,
Zhao-Kui Wana, Li Xinga, Xiaojing Yanga, Xin Yangb
a Medicine Design, Pfizer Inc, 1 Portland Street, Cambridge, MA 02139, United States
b Medicine Design, Pfizer Inc, Eastern Point Road, Groton, CT 06340, United States
c Inflammation and Immunology, Pfizer Inc, 1 Portland Street, Cambridge, MA 02139, United States
A R T I C L E I N F O
A B S T R A C T
Keywords:
TYK2
Herein, we disclose a new series of TYK2/ JAK1 inhibitors based upon a 3.1.0 azabicyclic substituted pyrimidine
scaffold. We illustrate the use of structure-based drug design for the initial design and subsequent optimization of
this series of compounds. One advanced example 19 met program objectives for potency, selectivity and ADME,
and demonstrated oral activity in the adjuvant-induced arthritis rat model.
JAK1
Kinase
Autoimmune
Inflammation
Psoriasis
Arthritis
1. Introduction
dermatitis, alopecia areata, psoriatic arthritis, lupus, ulcerative colitis
and Crohn's disease.7,8 Bristol Myers-Squibb are studying a TYK2 in-
The Janus kinases (JAK) are signaling molecules in the JAK/STAT
(signal transducer and activator of transcription) pathways, which play
a key role in intracellular signaling to extracellular cytokine signals.1
Inhibition of the JAK kinases (TYK2, JAK1, JAK2 and JAK3) has re-
ceived considerable interest for therapeutic intervention in auto-im-
mune disease, and this area has been reviewed recently.2 Whilst mul-
tiple compounds are in clinical development, only four compounds
have been approved by the US FDA, Fig. 1. Xeljanz™ (tofacitinib) 1, a
potent inhibitor of JAK1 which also inhibits JAK2, JAK3 and with
weaker inhibition of TYK2, was approved for rheumatoid arthritis, ul-
cerative colitis and psoriatic arthritis. The JAK1/JAK2 inhibitor Olu-
miant™ (baricitinib) 2 is approved for rheumatoid arthritis and Jakafi™
(ruxolitinib) 3 (JAK1/JAK2 inhibitor) has approvals for myelofibrosis,
polycythemia vera and acute graft versus host disease.3–5 The JAK1
selective inhibitor Rinvoq™ (upadacitinib) 4 was approved in 2019 for
the treatment of rheumatoid arthritis.6 We have previously reported
upon PF-06700841 5, a TYK2/JAK1 inhibitor currently in Phase 2
clinical studies for psoriasis, hidradenitis suppurativa, vitiligo, atopic
hibitor, BMS-986165 6, that binds to the pseudokinase JH2 domain of
TYK2 in multiple indications.9,10
In the current work, we were interested in developing a dual in-
hibitor of both JAK1 and TYK2. The JAK1 activity would inhibit the γ-
common chain cytokines (e.g. interleukin (IL-)2, IL-4, IL-7, IL-9, IL-15,
IL-21), IL-6 and Type I and II interferons, precedented by the clinically
approved agents. The γ -common chain cytokines play an important
role in both innate and adaptive immune systems, and have important
function in the development of key inflammatory cell types (T, B, NK
and innate lymphoid cells), important in many autoimmune condi-
tions.11
The TYK2 component would introduce to the profile inhibition of
IL-23 and IL-12 signaling, supported by clinical data from monoclonal
antibodies to IL-12 and IL-23 and associated pathway proteins.12 In
genome-wide association studies, TYK2 loss of function polymorphisms
are shown to provide protection from multiple auto-immune diseases,
including psoriasis, rheumatoid arthritis, systemic lupus erythematosus
and inflammatory bowel disease.13 Inhibition of both JAK1 and TYK2 is
⁎ Corresponding author.
E-mail address: Andrew.Fensome@Pfizer.Com (A. Fensome).
https://doi.org/10.1016/j.bmc.2020.115481
Received 5 February 2020; Received in revised form 25 March 2020; Accepted 27 March 2020
0968-0896/©2020ElsevierLtd.Allrightsreserved.
Pleasecitethisarticleas:AndrewFensome,etal.,Bioorganic&MedicinalChemistry,https://doi.org/10.1016/j.bmc.2020.115481

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
Fig. 1. Reference JAK inhibitors 1–6.
hypothesized to be useful in multiple autoimmune diseases including,
including psoriasis, psoriatic arthritis, lupus erythematosus and in-
flammatory bowel disease, by inhibiting type I interferon signaling
(JAK1/TYK2), together with JAK1 driven γ -common chain cytokine
and IL6 inhibition in combination with IL12 and IL23 inhibition in one
molecule.
IFNα IC80, whilst maximizing the window for EPO inhibition, with a
target EPO ICxx* ≤ IC30. As a reference For PF-06700841 5, the EPO
7
ICxx = IC17
.
As part of the earlier work we reported on molecules such as the
[3.2.1] ring systems 7 and 8, Fig. 2, Table 1.7 We also investigated
smaller bridged systems such as 9 (IC50 = 209 and 3900 nM, for TYK2
and JAK1 respectively). This latter compound was less potent than the
[3.2.1] series molecules and lacked the balanced TYK2/JAK1 potency
profile we were seeking.
In an earlier manuscript we disclosed the discovery and initial
characterization of PF-06700841 5.7 In that paper we discuss our pro-
gram objectives, derived from clinical and preclinical PK:PD relation-
ships for tofacitinib (Xeljanz TM) 1, and the utility of human whole
blood assays to study cytokine inhibition in a standardized set of pri-
mary cell assays. With this understanding we projected that efficacy
across autoimmune diseases would be driven by average (Cav) IC80
coverage of interferon alpha (IFNα), the primary target of TYK2 and
JAK1 inhibition. To reduce JAK2 mediated effects on red blood cell
differentiation, EPO inhibition (CD34+ cells spiked into human whole
blood) would be limited to Cav ≤ IC30. For candidate level molecules a
predicted human PK model would be established to calculate projected
cytokine inhibition levels, relative to IFNα IC80. for the expedience of
daily SAR and to allow us to understand selectivity against EPO sig-
naling, a calculated Cav inhibition or ICXX* was used based upon the
IFNα IC80 and EPO IC50 from whole blood assays (ICXX* = 100*((IC80
IFNα)/(IC80 IFNα + IC50 EPO)). Our objective was to drive potency on
To increase potency, we were interested in a strategy where we
could form a hydrogen bond interaction with the catalytic aspartate,
Asp-1041. We built a model of 9 docked into the TYK2 catalytic do-
main, from which the proximity (3.4 Å) between the bridge head
carbon and the carboxylate of Asp-1041 can be seen. We hypothesized
that the endocyclic amide nitrogen of the [3.1.1]diaza-bicyclic ring
system (i) could be moved into an exocyclic position, to interact with
ASP-1041. In earlier work we had found that a bridged ring system was
optimal for potency, due to reduced degrees of freedom, and the ability
to project into the lower lipophilic portion of the binding site.7 To this
end we built a new CeC bond to form a [3.1.0] ring system (ii), com-
pounds 10–12 (Table 2) the amide NeH of which would be available to
form the desired H-bond, Fig. 3.
Functional enzymatic activity against the four JAK kinases was
2

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
Fig. 2. Structures of compounds 7–9.
compound 8, that the 5-fluoro pyrimidine derivative was slightly more
potent than the 5-H pyrimidine congener, and as such in the current
work elected to retain the 5-fluorine substituent.7 As such we prepared
10 which proved to be the eutomer (IC50 = 139, 774, 5663 and >
10,000 nM respectively for TYK2, JAK1, JAK2 and JAK3). In human
whole blood the compound was active but with modest potency against
IFNα signaling. The distomer, compound 11, proved to be much weaker
(TYK2 IC50 = 5303 nM, JAK1 IC50 > 7359 nM).
Table 1
Enzymatic potencies and efficiency data for compounds 7 – 9.
IC50 (nM)a
Compound
TYK2
JAK1
JAK2
JAK3
LogDb
TYK2 LipEc
7
8
9
32
122
81
329
> 10,000
2531
1.8
1.8
1.7
5.9
5.8
5.2
55
260
209
3900
9644
> 10,000
Compound 10 was crystalized with the catalytic JH1 domains of
both TYK2 and JAK1, Fig. 5a and b respectively. Gratifyingly, the
structure confirmed our design hypothesis, with a hydrogen bond being
formed between the amide NeH and the catalytic Asp-1041 in TYK2
and Asp-1021 in JAK1, with the S-stereochemical configuration at this
center. This stereochemical feature was consistent throughout the
series. The structures also show the two differential residues between
the two ATP binding sites: Ile-960 in TYK2 which is Val-938 in JAK1.
Val-981 in Tyk2 makes the typical hinge donor/acceptor hydrogen
bonds with the ligand, whereas in Jak1 this residue is Leu-959. A sol-
vent front hydrogen bond is also made between the ethanol side chain
of the molecule and the backbone NeH of either Ser-985 (TYK2) or Ser-
963 (JAK1). A lipophilic pocket below the cyclopropane of the [3.1.0]
ring system is also evident in the structure. The ADME characterization
of the compound encouraged us to explore this chemotype further: the
compound had reasonable kinetic aqueous solubility (278 μM at pH
a
Compounds were assayed at least twice, and the IC50 reported as the
geometric mean.
b
c
Shake flask logD at pH 7.4.
LipE = pIC50 – LogD.
obtained using a PerkinElmer LabChip® EZ Reader assay, under pseudo-
physiological conditions, with ATP at a concentration of 1 mM, as re-
ported previously.14 Under these conditions we are able to compare
JAK enzyme potencies directly without taking into account the KM for
ATP for each target. Moreover, as previously reported, the pseudo
physiological conditions more closely correlate with cellular functional
readouts with minimum potency shifts.14 Cellular data was obtained
from human whole blood assays where we studied inhibition of STAT
phosphorylation via flow cytometry, experimental conditions were re-
ported previously.7 Each signaling pathway was activated with a near
maximal concentration of the desired cytokine. EPO signaling was de-
termined in human whole blood spiked with CD34+ cells.7 Whole blood
data was corrected for protein binding to provide an estimate of un-
bound potency (assuming blood: plasma ratio = 1), thereby allowing
comparisons to be made between compounds, independent of plasma
binding. Protein binding data was either measured or calculated using a
QSAR model.
7.4),
moderate
passive
permeability15
(MDCK-II
mean
PAPP = 4.4 × 10-6 cm/sec) and good stability in human liver micro-
somes (< 8 μL/min/mg), consistent with a LogD = 1.75.
In efforts to increase potency, we first turned our attention to the
lipophilic pocket at the bottom of the binding site. The addition of an
endo methyl group to the cyclopropane 12, gave a small boost in po-
tency (IC50 = 57 and 157 nM for TYK2 and JAK1 respectively), but
with no net gain in lipophilic efficiency (LipE = 5.41 and 5.45 for 10
and 12 respectively). The increase in potency translated into the human
whole blood assay (IFNα IC50(u) = 443 (1 1 1) nM) with good
Following on from our earlier work we decided to make the [3.1.0]
analog of 8 - and allow us to make matched pair comparisons, Fig. 4,
Table 2. In earlier work we had shown in molecules similar to
Fig. 3. Initial design strategy for [3.1.0] ring system: (a) docked binding mode of compound 9 in TYK2 ATP binding site; (b) break-a-ring/ make-a-ring design of
[3.1.0] ring system.
3

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
Fig. 4. Structures of compounds 10–12.
Table 2
JAK Enzymatic potencies, human whole blood data and efficiencies for compounds 10 – 12.
Compd.
IC50 (nM)
HWB IC50 (nM)
HWB IFNα IC80 (nM)a
EPO ICXX*
TYK2a
JAK1a
JAK2a
JAK3a
LogDb
Fuc (%)
0.41
Tyk2 LipEd
IFNαa
IFNα(u)
EPOa
e
10
11
12
139
5303
57
774
5663
> 10000
> 10000
> 8458
1.74
1.75
2.10
5.4
3.8
5.4
1523
ND
621
ND
> 20000
ND
6092
ND
< 23
ND
> 7359
157
> 10000
1617
0.31
443
111
15,264
1772
10
a
b
c
d
e
Compounds were assayed at least twice, and the IC50 reported as the geometric mean.
Shake flask logD at pH 7.4.
Human plasma free fraction measured at 2 μM compound concentration.
LipE = pIC50 – LogD.
Free IC50 = IFNα HWB IC50* human plasma free fraction.
Fig. 5. Compound 10 crystallized with TYK2 (a) and JAK1 (b) showing H-bond to Asp-1041 and Asp-1021 respectively. The structures also illustrate the differential
residues between TYK2 and JAK1: Val-981 to Leu-959 and Ile-960 to Val-938, respectively.
selectivity over EPO signaling (EPO HWB IC50 = 15,264 nM, ICXX* =
IC10).
with R,R-cyclopropyl stereochemistry, compound 13, was overall less
selective for JAK2 and JAK3 (13 IC50 = 340 and 949 nM for JAK2 and
JAK3; 14 IC50 = 2167 and > 10,000 nM for JAK2 and JAK3). The
increase in enzymatic potency for this modification translated into cells
for the more potent nitrile isomer. In the human whole blood IFNα
assay, 13 was more active than the parent 10 (IFNα IC50(u) = 153 (73)
nM). Unbound potency values are needed when comparing compounds
whereby plasma protein binding is accounted for. This improved un-
bound whole blood potency is a consequence of improved potency for
both TYK2 and JAK1, bearing in mind that IFNα utilizes both TYK2 and
JAK1 for signal transduction. For the less active isomer 14, the same
whole blood potency gains were not realized compared with analog 10.
We next combined the cyano-cyclopropane and the endo methyl
group to provide compounds 15 and 16. The expected additive increase
in TYK2 potency was not achieved, although the JAK1 potency did
improve. Compound 15, with R,R-stereochemistry, was more active on
TYK2 and JAK1 but again lacked the desired family selectivity. The S,S-
cyano-cyclopropane 16 was more interesting, showing balanced ac-
tivity against the primary enzymatic targets (TYK2 and JAK1 IC50 = 32
and 41 nM, respectively). However, in the IFNα human whole blood
assay the compound was relatively weak (IC50(u) = 971 (3 7 4) nM). We
To better guide efforts to increase potency further, we turned to
computational analysis. Starting with the crystal structure of 10 bound
to TYK2, we simulated water energetics for the apo protein, Fig. 6,
Table 3.16 From the simulation, three water molecules were shown to
be proximal to the cyclopropane. Simulated apo waters W1 and W2
would be displaced upon ligand binding. W1 is a moderately unstable
water, whilst W2 was the highest energy water in the simulation, dis-
placement of which could theoretically improve potency. The third
water, W3, was also predicted to be unstable (ΔG = 3.04, ΔH = 1.23
and –TΔS = 1.8), displacement of this water presented an additional
opportunity to increase potency.
We first prepared the diastereomeric trans-cyano cyclopropyl
amides 13 and 14, Fig. 7, Table 3. Crystallography showed that both
isomers would be able to displace the high energy simulated water W3
as well as W2, Fig. 8. Gratifyingly, both compounds were more effi-
cient, potent inhibitors of TYK2 and JAK1 than the parent 10 (Com-
pound 13 IC50 = 29 and 18 nM for TYK2 and JAK1, TYK2 LipE = 6.44;
Compound 14 IC50 = 67 and 304 nM for TYK2 and JAK1, TYK2
LipE = 6.04, respectively). The more potent of the two compounds
4

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
Fig. 6. Compound 10 overlaid onto simulated apo water molecules in the TYK2 ATP binding site. Table 3. Energetics for simulated apo waters W1 – W3.
hypothesized that the cellular potency was hampered by a lack of
permeability (MDCKII mean Papp = 1.2 × 10-6 cm/sec).
pyrazole of the above compounds to address this issue. As would have
been expected from the previous data, the R,R-cyclopropane 17 was
indeed the more active isomer on TYK2 and JAK1, however again
showed potent activity on JAK2 and in particular JAK3 (IC50 = 32, 11,
228 and 225 nM for TYK2, JAK1, JAK2, JAK3 respectively). The less
potent S,S-isomer 18 had similar enzymatic potency to the hydro-
xyethyl congener 16. As would be expected, the LogD of these two
Whilst the N-hydroxyethyl substituent on the pyrazole did not pose
a secondary metabolism liability (human hepatocyte Clint < 6 μL/
min/million cells for 10 and 12) for this series of compounds, it did
present additional polarity, and an H-bond donor/acceptor which could
influence permeability. To this end we made the simple N-methyl
Fig. 7. Structures of compounds 13–20.
Table 3
JAK Enzymatic potencies, human whole blood data and efficiencies for compounds 13 – 20.
Compd. R Group
IC50 (nM)
HWB IC50 (nM)
HWB IFNα IC80 (nM)a EPO ICXX
*
TYK2a JAK1a JAK2a JAK3a
LogDb Fuc (%) Tyk2 LipEe IFNαa IFNα(u)
EPOa
f
1
2
3
13
14
15
16
17
18
19
20
a
b
a
b
a
b
c
e
e
f
g
g
g
g
h
h
h
h
29
67
37
32
32
49
29
56
18
304
6
340
949
1.40
0.47d
0.42
6.4
6.0
6.1
6.0
5.4
5.3
5.2
4.6
153
1812
148
971
87
73
13,017
612
4
2167
128
> 10000 1.43
888
57
> 20000 7248
3
194
1.71
1.75
2.35
2.32
2.65
2.92
1492
592
28
23
15
15
6
f
41
11
81
41
113
676
2282
255
0.37d
0.29
0.35
0.24
0.15
374
21
12,850
1942
3884
348
f
228
f
1168
652
3225
4267
4023
421
191
1733
104
42
9368
1684
764
f
11,440
18,837
d
f
1132
414
6932
27
a
b
c
d
e
f
Compounds were assayed at least twice, and the IC50 reported as the geometric mean.
Shake flask logD at pH 7.4.
Human plasma free fraction measured at 2 μM compound concentration.
Calculated free fraction.
LipE = pIC50 – LogD.
Free IC50 = IFNα HWB IC50* human plasma free fraction.
5

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
Fig. 8. Crystal structure of (a) Compound 13/ TYK2 (2.09 Å) and (b) Compound 14/ TYK2 (2.15 Å), overlaid with simulated apo waters W1, W2 and W3.
compounds was approximately 0.6 units higher, but most importantly
the passive permeability improved noticeably (MDCKII mean
PAPP = 12.9 and 13.8 × 10-6 cm/sec for 17 and 18). This improved
permeability brought with it an increase in cellular activity for the more
selective compound 18 (IFNα HWB IC50(u) = 421 (104) nM), however
this compound still did not meet the program goals for cellular potency.
The EPO selectivity was in line with the JAK2 potency (EPO IC50
9,368 nM, ICXX* = IC15).
protein. Consistent with our apo water hypothesis, water W2 would be
displaced by the compound. The distal water W3 is also within proxi-
mity to the fluorine atom and may also be expected to be displaced.
Compound 19 was selected for further profiling against an ex-
panded set of cytokines in human whole blood assays, the values are
reported as unbound IC50. Against the TYK2/JAK2 pathways IL-23
(pSTAT3) and IL-12 (pSTAT4) the compound had IC50(u) = 207 nM and
108 nM, respectively. Compound 19 inhibited the γ -common signaling
In the cyno cyclopropyl amide series, the trend for the R,R-en-
antiomer being less selective (increased potency) than the S,S-isomer
can be rationalized from the structure if we extrapolate across the JAK
kinase family, Fig. 8. From the structure of compound 13 (R,R-en-
antiomer) with TYK2, we can see that the nitrile projects into the P-loop
of the kinase, making interactions with the protein, Fig. 8a. In the case
of the S,S-enantiomer, the nitrile functionality is projecting away from
the P-loop, towards solvent, and hence has less interaction with the
protein, Fig. 8b, which would engender greater selectivity.
pathways
(JAK1/JAK3)
represented
by
IL-21/pSTAT3
(IC50(u) = 319 nM), and IL-15/pSTAT5 (IC50(u) = 265 nM). Interferon-
γ /pSTAT1 was inhibited by compound 19 with a whole blood
IC50(u) = 265 nM. IL-10/pSTAT3 and IL-27/pSTAT3 signaling was in-
hibited by 19 with IC50(u) = 659 nM and 155 nM, respectively. Kinase
selectivity was determined at both the KM for ATP for each kinase, or at
1 mM in the Carna Bioscience panel, at a compound test concentration
of 1 μM. Under the [ATP] at KM conditions the compound had activity
at 15 out of 307 kinases above 60% inhibition, and 20 targets between
30 and 60% inhibition, Supplementary Table 1. Under the 1 mM ATP
conditions, compound 19 was active against 5 kinases at or above 16%
inhibition: JAK3 (16.0%), TNK1 (50.6%), JAK2 (62.9%), JAK1 (87.0%)
and TYK2 (90.2%). Against a promiscuity panel of 94 targets (internal
and Cerep) compound 19 only showed activity for EGFR kinase
([ATP] = KM, IC50 4,110 nM, Cerep) and VEGFR2 ([ATP] = KM, IC50
284 nM, Cerep). Follow-up cellular assays for both EGFR and VEGFR2
were negative (Caliper Biosciences), supported by data from the Carna
Biosciences panel for EGFR (1.3% inhibition at a test concentration of
1 μM ([ATP] = 1 mM)) and KDR/VEGFR2 (5.0% inhibition at a test
concentration of 1 μM ([ATP] = 1 mM).
Amongst the cyclopropane substitution patterns we tried in the
program, the most fruitful was the difluorocyclopropane.7 We prepared
the difluoro cyclopropyl derivatives with the goal of increased cellular
potency with a balanced JAK enzyme profile. Compound 19, with S-
stereochemistry at the cyclopropane chiral center, proved to have good
potency against TYK2 and JAK1 (IC50 = 29 and 41 nM respectively,
TYK2 LipE = 5.2), with promising selectivity against JAK2 and JAK3
(JAK2 and JAK3 IC50 = 652 and 4356 nM, respectively). In the IFNα
human whole blood assay, the compound had good potency (IFNα
HWB IC50(u) = 191 (42) nM; EPO HWB IC50 = 11,440 nM, ICXX* =
IC6). In contrast, the isomeric cyclopropane 20 was a little less active on
both TYK2 and JAK1 (IC50 = 56 and 113 nM respectively), which was
inadequate to drive high cellular activity (IFNα HWB IC50(u) = 1733
(4 1 4) nM). The binding mode of compound 19 was consistent with
compounds discussed thus far, Fig. 9. Of note in the structure is the
projection of the two fluorine atoms up towards the P-loop of the
The in vitro metabolism was studied in human liver microsomes
(Clint,u < 8 µL/min/mg protein) and hepatocytes (Clint,u 3.45 µL/min/
million cells), which indicated low metabolic turnover. Plasma protein
binding was moderate in rat (fraction unbound (fu) 0.19) and human (fu
0.24), with no significant blood to plasma partitioning (human Crbc/Cp
Fig. 9. Crystal structure of compound 19 bound to TYK2 ATP binding site (2.18 Å), overlaid with apo simulated waters W1 – W3.
6

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
1.15).
The physical properties of compound 19 were favorable for oral
dosing with a molecular weight of 407.4, a measured LogD (pH 7.4) of
2.7, and a calculated TPSA of 88 Å2. Compound 19 also showed high in
vitro passive permeability (MDCKII mean PAPP = 17 × 10-6 cm/s).
Aqueous solubility was 0.14 mg/mL at pH 6.5 and > 15 mg/mL at pH 2
from crystalline material. The pharmacokinetics of 19 were studied in
Sprague − Dawley rats following intravenous and oral administration
(1 and 3 mg/kg respectively), where the compound showed a plasma
clearance of 13 mL/min/kg, a volume of distribution of 1.2 L/kg, and
half life of 1.4 h. Following the 3 mg/kg oral dose, the observed Cmax
was 614 ng/mL, the AUC∞ was 2030 ng·h/mL with oral bioavailability
of 54%.
Based on the physicochemical properties 19 was predicted by the
Extended Clearance Classification System (ECCS) to be a class 2 com-
pound with rate determining pathway of metabolic clearance.17 Oxi-
dative metabolites of 19 accounted for the primary routes of bio-
transformation in rat and human in vitro hepatic systems, consistent
with CYP450 as the primary clearance route. Metabolic pathways in-
cluded oxidation of the N-methyl pyrazole, N-demethylation (pyrazole),
and N-dealkylation (loss of pyrazole). Metabolic clearance phenotyping
of 19 indicated that CYP3A4 was likely the predominant mediator of
metabolism. As the clearance mechanisms of 19 were determined to be
mediated primarily by CYP450 metabolism (principally through
CYP3A4), with limited renal and biliary clearance expected, human
pharmacokinetic predictions were made using human liver microsomes
and hepatocytes, as well as standard single species allometric scaling
techniques, Table 4.
Fig. 10. Radial plot of cytokine inhibition ICXX relative to the IFNα IC80 for
compound 19, derived from potency in human whole blood assays and pre-
dicted average daily human plasma drug concentration.
Table 5
Human whole blood IC50 values for each cytokine and associated JAK signaling
pair.
Cytokine
JAK kinase pair
STAT
HWB IC50(u) (nM)a
Clb: blood clearance; Vss: volume of distribution at steady state;
Fraction unbound plasma protein binding in rat and human were 0.19
and 0.24 respectively. Human clearance predictions from in vitro sys-
tems were scaled based on the well-stirred model or single species
scaling from allometry.18,19
IFNα
IL6
JAK1/TYK2
JAK1/Jak2
JAK1/JAK2
JAK1/JAK3
JAK1/JAK3
JAK1/TYK2
JAK1/JAK2
TYK2/JAK2
TYK2/JAK2
JAK2/JAK2
STAT3
STAT1
STAT1
STAT5
STAT3
STAT3
STAT3
STAT4
STAT3
STAT5
42
183
265
265
319
659
155
108
207
2519
IFN γ
IL15
IL21
IL10
IL27
IL12
IL23
EPO
The overall average daily % cytokine inhibition profile (ICXX) was
calculated for 19 based upon the modelled Cav exposure of approxi-
mately 89 nM (unbound) at an expected efficacious dose of 75 mg QD,
Fig. 10, Table 5.20,21 Under these conditions JAK1/TYK2 driven IFNα
inhibition is set at a target of IC80. The IL-12 and IL-23 target cover
(TYK2/JAK2) was IC61 and IC45, respectively. IL-6 inhibition (JAK1/
JAK2 or TYK2) via pSTAT1 in CD34+ cells was IC48. Contribution to the
profile from the γ -common chain cytokines (JAK1/JAK3), represented
by IL-15 and IL-21 were lower at IC39 and IL35, respectively. IL-27
(JAK1/JAK2 or TYK2) and IL-10 (JAK1/TYK2) target cover was esti-
mated to be at IC52 and IC20, respectively. Importantly, the compound is
expected to spare JAK2/JAK2-driven EPO signaling (IC06) (see Fig 11.).
Compound 19 was studied in vivo in a rat adjuvant-induced arthritis
(AIA) disease model.22 Female Lewis rats immunized with complete
Freund’s adjuvant were orally dosed with 19 or vehicle for seven con-
secutive days after disease onset as measured by hind paw volume using
plethysmography. Paw volumes and body weights were followed
throughout the study. At the end of 7 days of dosing, plasma con-
centrations of 19 were assessed. Two studies were completed.
In the initial study, immunized rats were dosed QD with 3, 10, 30 or
100 mg/kg of 19 or vehicle (PO) (see supplemental information).
Because a significant reduction in paw swelling was observed for all
a
Compounds were assayed at least twice, and the IC50 reported as the
geometric mean.
doses, a second study was conducted focused on lower doses (1, 5, 15 or
30 mg/kg 19 or vehicle control), Fig. 10. A significant reduction of hind
paw swelling was observed with doses as low as to 3 mg/kg, Supple-
mental Fig. 1. Collective pharmacodynamics modeling of both studies
using an Emax model fit indicated an unbound Cav50 of approximately
199 nM (95% CI 134 – 294 nM), Supplemental Fig. 2.
In conclusion we designed a new series of [3.1.0] bicyclic TYK2/
JAK1 inhibitors based upon a rational design starting from earlier work
in our program. This new series was further optimized based on
emerging SAR and relying upon crystal structures of key analogs and
computational modelling and simulation. The lead analogs from this
series, compound 19, was demonstrated to have good in vitro and in vivo
ADME properties. The predicted human PK and dose, and the projected
JAK target cover successfully met the program objectives for a clinical
candidate.
Table 4
1.1. Synthesis
Summary of predicted human blood pharmacokinetics of compound 19 from in
vivo and in vitro systems
We have previously described the synthesis of compounds 7 − 9a
and we will focus on the synthetic replacement of the functionalized
diamine core of 8 with 3-azabicyclo[3.1.0]hexan-1-amine as shown in
Scheme 1. Commercially available 2,4-dichloro-5-fluoropyrimidine was
substituted via direct displacement of the 4-chloride with ethanethiol as
the nucleophile. At room temperature, the reaction was highly selective
for displacement at the pyrimidyl 4-chloride driven in part by the
Method
Human Prediction
Clb (mL/min/kg)
Vss (L/kg)
Human Liver Microsomes
Human Hepatocytes
1.4
1.6
3.5
Rat Single Species Scaling
1.5
7

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
Fig. 11. Effect of compound 19 in the rat adjuvant-induced arthritis model following oral dosing of 1–30 mg/kg.
pyrimidine 5-fluoro presumably due to increased electrophilicity at this
position. The pyrazole-2-amino pyrimidine intermediate 22 was con-
structed via SNAr under acidic conditions using 21 and 2-(4-amino-1H-
pyrazol-1-yl)ethan-1-ol. Oxidation to the sulfone 23 followed by dis-
placement with commercially sourced t-butyl (3-azabicyclo[3.1.0]
hexan-1-yl)carbamate afforded a mixture of compounds 24 and 25.
Compound 24 and 25 were separated by chiral supercritical fluid -
chromatography (SFC). The protecting groups were removed under
standard acidic conditions followed by amide coupling with either cy-
clopropane carboxylic acid, or trans-cyano cyclopropane carboxylic
acid using HATU providing acetylated analogs 10, 11 and 28. Chiral
SFC separation of compound 28 provided analogs 13 and 14.
hexan-1-amine core.23 The key intermediate 29 was made through al-
kylation of benzyl amine with 2-bromoacetonitrile followed by a second
alkylation using 1-bromobut-2-yne. Subsequent hydrogenation with
Lindlar catalyst set the required (Z) configuration of the olefin 31. The
stereospecific intramolecular cyclopropanation reaction was achieved
using Ti(OiPr)4 and cyclohexyl Grignard reagent to provide the trans-
racemic mixture 32. This mixture was purified by preparative SFC to
afford intermediates 33 and 34. Manipulation of the protecting groups
would prove useful in our modular approach to final target synthesis.
Therefore, Boc protection of 33 using standard conditions followed by
hydrogenation (50 psi) with Pearlman’s catalyst afforded key building
block 36.
The synthesis of the 6-methyl-3-azabicyclo[3.1.0]hexan-1-amine
cores are described in Scheme 2. We utilized an intramolecular cyclo-
propanation reaction of a nitrile and di-substituted alkene described by
Cha and coworkers to achieve the 6-substituted 3-azabicyclo[3.1.0]
Assembly of the final compounds was performed in a modular
fashion by routes described in Scheme 3 and Scheme 4. The route il-
lustrated in Scheme 3 incorporates the acetylated portion of the mo-
lecule first followed by construction of the pyrazole-2-amino
Scheme 1. Reagents and conditions. (a) ethanethiol, potassium t-butoxide, DME, 0 °C – RT, 36 h (76%); (b) 2-(4-amino-1H-pyrazol-1-yl)ethan-1-ol hydrochloride,
12 M HCl, n-butanol, 130 °C, 16 h (56%); (c) oxone, THF, 15 °C, 16 h (38%); (d) t-butyl (3-azabicyclo[3.1.0]hexan-1-yl)carbamate, TEA, MeCN, RT – 90 °C, 16 h
(48%); (e) chiral SFC (45%); (f) 24 or 25, con. HCl, dioxane, MeOH, RT, 3 h (100%); (g) 26 or 27, R1 carboxylic acid, HATU, TEA, DMF, RT, 3 h (49–63%); (h) chiral
SFC (24–32%).
8

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
Scheme 2. Reagents and conditions. (a) 2-bromoacetonitrile, DIPEA, ACN, RT, 3 h (94%); (b) 1-bromobut-2-yne, K2CO3, ACN, 40 °C, 18 h (77%); (c) Lindlar catalyst,
H2, MeOH, 30 °C, 24 h; (d) Ti(OiPr)4, cyclohexylmagnesium chloride (2 M in ether), THF, 20–30 °C, 3.5 h (36%); (e) chiral SFC; (f) ditertbutyldicarbonate, TEA, DCM,
RT, 18 h (91%); (g) Pd(OH)2, H2 (50 psi), MeOH, RT, 48 h (100%).
pyrimidine moiety. Beginning with the trans-racemic intermediates 32,
the amides 37 were made through standard coupling conditions or in
this case using the acid chloride. Benzyl deprotection using hydro-
genation conditions previously described afforded 38 and subsequent
SNAr with 2,4-dichloro-5-fluoropyrimidine provided 39. Purification
under chiral SFC conditions provided the single enantiomers (40 and
41) and the hinge binding moiety of 40 was completed using acidic
SNAr conditions to provide compound 12.
Scheme 4 describes the construction of the hinge binding motif
followed by the late stage variation of the amide moiety. 36 underwent
Scheme 3. Reagents and conditions. (a) cyclopropanecarbonyl chloride, TEA, CH2Cl2, RT, 1 h (77%); (b) Pd(OH)2, H2 (50 psi), MeOH, RT, 16 h; (c) 2,4-dichloro-5-
fluoropyrimidine, TEA, MeOH, RT, 1 h (70% over 2 steps); (d) chiral SFC (36%); (e) 2-(4-amino-1H-pyrazol-1-yl)ethan-1-ol hydrochloride, IPA, 140 °C mw, 1 h
(38%).
9

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
Scheme 4. Reagents and conditions. (a) 2,4-dichloro-5-fluoropyrimidine, TEA, MeOH, RT, 18 h (69%); (b) 1-methyl-1H-pyrazol-4ylamine hydrochloride or 2-(4-
amino-1H-pyrazol-1-yl)ethan-1-ol hydrochloride (0.239 g, 1.46 mmol), IPA, 140 °C mw, 1 h (86–94%); (c) R2 carboxylic acid, HATU, TEA or DIPEA, DCM or DMA,
RT, 1.5–18 h (75–93%); (h) chiral SFC.
selective SNAr with 2,4-dichloro-5-fluoropyrimidine at the 4-position to
provide the substituted product 42 in 69% yield. The substituted pyr-
azole-2-amino pyrimidine moiety was made through a second SNAr
reaction under acidic conditions. The reaction conditions also resulted
in the loss of the Boc group on the [3.1.0] azabicycle to provide 43 or
44 in excellent yields. Amide formation using 43 or 44 and various
carboxylic acidsc under standard HATU conditions provided 15–18 and
45. Chiral SFC purification of 45, provided 19 and 20 as single en-
antiomers.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Acknowledgements
Funding for this work was provided by Pfizer, Inc.
The authors would like to thank David Hepworth for helpful com-
ments during preparation of the manuscript. The authors would also
like to acknowledge the contributions of Matthew Griffor and John
Knafels (Pfizer) for provision of TYK2 protein reagents and crystal-
lization expertise.
2. Notes
2.1. Disclosures
This research used resources at the Industrial Macromolecular
Crystallography Association Collaborative Access Team (IMCA-CAT)
beamline 17-ID, supported by the companies of the Industrial
Macromolecular Crystallography Association through a contract with
Hauptman-Woodward Medical Research Institute. This research used
resources of the Advanced Photon Source, a U.S. Department of Energy
(DOE) Office of Science User Facility operated for the DOE Office of
Science by Argonne National Laboratory under Contract No. DE-AC02-
06CH11357.
The authors were employees of Pfizer Inc. during this work and
declare no competing interest.
All activities involving laboratory animals were carried out in ac-
cordance with federal, state, local and institutional guidelines gov-
erning the use of laboratory animals in research and were reviewed and
approved by Pfizer (or other) Institutional Animal Care and Use
Committee. Pfizer animal care facilities that supported this work are
fully accredited by AAALAC International.
Appendix A. Supplementary material
3. PDB ID code
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.bmc.2020.11l5481.
The atomic coordinates for the X-ray structure of JAK1 with com-
pound 10 can be accessed from the RCSB Protein Data Bank (www.rcsb.
org) using PDB code 6W8L. Atomic coordinates for the TYK2 X-ray
structures with compounds 10, 13, 14 and 19 can be accessed using
PDB codes 6NVY, 6VNS, 6VNV and 6VNX respectively. Authors will
release the atomic coordinates upon article publication.
References
1. O'Sullivan LA, Liongue C, Lewis RS, Stephenson SE, Ward AC. Cytokine receptor
10

A. Fensome, et al.
Bioorganic&MedicinalChemistryxxx(xxxx)xxxx
signaling through the Jak-Stat-Socs pathway in disease. Mol Immunol.
Biologic Agents for the Treatment of Immune-Mediated Inflammatory Conditions.
2007;44:2497–2506.
Clin Pharmacol Ther. 2018;103:88–101.
2. Virtanen AT, Haikarainen T, Raivola J, Silvennoinen O. Selective JAKinibs: Prospects
in Inflammatory and Autoimmune Diseases. BioDrugs: Clin Immunotherapeutics,
Biopharmaceut Gene Therapy. 2019;33:15–32.
13. Dendrou CA, Cortes A, Shipman L, et al. Resolving TYK2 locus genotype-to-pheno-
type differences in autoimmunity. Sci Transl Med. 2016;8 363ra149.
14. Thorarensen A, Banker ME, Fensome A, et al. ATP-mediated kinome selectivity: the
missing link in understanding the contribution of individual JAK Kinase isoforms to
cellular signaling. ACS Chem Biol. 2014;9:1552–1558.
3. Clark JD, Flanagan ME, Telliez JB. Discovery and development of Janus kinase (JAK)
inhibitors for inflammatory diseases. J Med Chem. 2014;57:5023–5038.
4. Fridman JS, Scherle PA, Collins R, et al. Selective inhibition of JAK1 and JAK2 is
efficacious in rodent models of arthritis: preclinical characterization of INCB028050.
J Immunol (Baltimore, Md: 1950). 2010;184:5298–5307.
15. Di L, Whitney-Pickett C, Umland JP, et al. Development of a new permeability assay
using low-efflux MDCKII cells. J Pharm Sci. 2011;100:4974–4985.
16. Zhu, H. Water's Role in Compound Design for Drug Discovery. CCG UGM and
Conference, Montreal, Canada.
5. Quintas-Cardama A, Vaddi K, Liu P, et al. Preclinical characterization of the selective
JAK1/2 inhibitor INCB018424: therapeutic implications for the treatment of mye-
loproliferative neoplasms. Blood. 2010;115:3109–3117.
17. Varma MV, Steyn SJ, Allerton C, El-Kattan AF. Predicting Clearance Mechanism in
Drug Discovery: Extended Clearance Classification System (ECCS). Pharm Res.
2015;32:3785–3802.
6. Parmentier JM, Voss J, Graff C, et al. In vitro and in vivo characterization of the JAK1
selectivity of upadacitinib (ABT-494). BMC Rheumatol. 2018;2:23.
18. Hosea NA, Collard WT, Cole S, et al. Prediction of human pharmacokinetics from
preclinical information: comparative accuracy of quantitative prediction approaches.
J Clin Pharmacol. 2009;49:513–533.
7. Fensome A, Ambler CM, Arnold E, et al. Dual Inhibition of TYK2 and JAK1 for the
Treatment of Autoimmune Diseases: Discovery of ((S)-2,2-Difluorocyclopropyl)((1 R,
5 S)-3-(2-((1-methyl-1 H-pyrazol-4-yl)amino)pyrimidin-4-yl)-3,8-diazabicyclo[3.2.1]
octan-8-yl)methanone (PF-06700841). J Med Chem 2018;61:8597–612.
8. Banfield C, Scaramozza M, Zhang W, et al. The Safety, Tolerability,
Pharmacokinetics, and Pharmacodynamics of a TYK2/JAK1 Inhibitor (PF-06700841)
in Healthy Subjects and Patients With Plaque Psoriasis. J Clin Pharmacol. 2017.
9. Weinstein DM, Ryan M. Advances in the discovery and development of selective
tyrosine kinase 2 (TYK2) inhibitors. Med Chem Rev. 2018;53:177–200.
10. Wrobleski ST, Moslin R, Lin S, et al. Highly Selective Inhibition of Tyrosine Kinase 2
(TYK2) for the Treatment of Autoimmune Diseases: Discovery of the Allosteric
Inhibitor BMS-986165. J Med Chem. 2019;62:8973–8995.
19. Obach RS. Predicting clearance in humans from in vitro data. Curr Top Med Chem.
2011;11:334–339.
20. Dowty ME, Jesson MI, Ghosh S, et al. Preclinical to clinical translation of tofacitinib,
a Janus kinase inhibitor, in rheumatoid arthritis. J Pharmacol Exp Ther.
2014;348:165–173.
21. Dowty MEL, Lin TH, Wang L, et al. An Updated Analysis of Invitro Cytokine
Inhibition Profiles of a Number of Janus Kinase Inhibitors at Clinically Maningful
Concentrations. United Eur Gastroenterol J. 2017;5.
22. Hegen M, Keith Jr JC, Collins M, et al. Utility of animal models for identification of
potential therapeutics for rheumatoid arthritis. Ann Rheumatic Dis.
2008;67:1505–1515.
11. Warren J. Leonard, J.-X. L., John J. O'Shea. The γc Family of Cytokines: Basic Biology
to Therapeutic Ramifications. Immunity 2019;50:832–50.
23. Astashko D, Lee HG, Bobrov DN, Cha JK. Stereochemistry of the Kulinkovich
Cyclopropanation of Nitriles. J Organic Chem. 2009;74:5528–5532.
12. Frieder J, Kivelevitch D, Haugh I, Watson I, Menter A. Anti-IL-23 and Anti-IL-17
11