Examining the chirality, conformation and selective kinase inhibition of 3-((3R,4R)-4-methyl-3-(methyl(7H-pyrrolo[2,3-d]pyrimidin-4-yl)amino) piperidin-1-yl)-3-oxopropanenitrile (CP-690,550)

Article abstract of DOI:10.1021/jm801142b

Here, we examine the significance that stereochemistry plays within the clinically relevant Janus kinase 3 (Jak3) inhibitor 1 (CP-690,550). A synthesis of all four enantiopure stereoisomers of the drug was carried out and an examination of each compound revealed that only the enantiopure 3R,4R isomer was capable of blocking Stat5 phosphorylation (Jak3 dependent). Each compound was profiled across a panel of over 350 kinases, which revealed a high level of selectivity for the Jak family kinases for these related compounds. Each stereoisomer retained a degree of binding to Jak3 and Jak2 and the 3R,4S and 3S,4R stereoisomers were further revealed to have binding affinity for selected members of the STE7 and STE20 subfamily of kinases. Finally, an appraisal of the minimum energy conformation of each stereoisomer and molecular docking at Jak3 was performed in an effort to better understand each compounds selectivity and potency profiles.

Full text of DOI:10.1021/jm801142b

8012
J. Med. Chem. 2008, 51, 8012–8018
Examining the Chirality, Conformation and Selective Kinase Inhibition of
3-((3R,4R)-4-methyl-3-(methyl(7H-pyrrolo[2,3-d]pyrimidin-4-yl)amino)piperidin-1-yl)-3-
oxopropanenitrile (CP-690,550)
Jian-kang Jiang,† Kamran Ghoreschi,‡ Francesca Deflorian,§ Zhi Chen,‡ Melissa Perreira,† Marko Pesu,‡ Jeremy Smith,
Dac-Trung Nguyen,† Eric H. Liu,| William Leister,† Stefano Costanzi,§ John J. O’Shea,‡ and Craig J. Thomas*,†
NIH Chemical Genomics Center, National Human Genome Research Institute, National Institutes of Health, 9800 Medical Center DriVe,
RockVille, Maryland 20850, Molecular Immunology and Inflammation Branch, National Institute of Arthritis and Musculoskelatal and
Skin Diseases, National Institutes of Health, Bethesda, Maryland 20850, Laboratory of Biological Modeling and Transplantation and
Autoimmunity Branch, National Institute of Diabetes and DigestiVe and Kidney Diseases, National Institutes of Health, Bethesda,
Maryland 20892, Kelly SerVices, 6101 ExecutiVe BouleVard, Room 392, Mail Stop Code 9307, RockVille, Maryland 20852
ReceiVed September 12, 2008
Here, we examine the significance that stereochemistry plays within the clinically relevant Janus kinase 3
(Jak3) inhibitor 1 (CP-690,550). A synthesis of all four enantiopure stereoisomers of the drug was carried
out and an examination of each compound revealed that only the enantiopure 3R,4R isomer was capable of
blocking Stat5 phosphorylation (Jak3 dependent). Each compound was profiled across a panel of over 350
kinases, which revealed a high level of selectivity for the Jak family kinases for these related compounds.
Each stereoisomer retained a degree of binding to Jak3 and Jak2 and the 3R,4S and 3S,4R stereoisomers
were further revealed to have binding affinity for selected members of the STE7 and STE20 subfamily of
kinases. Finally, an appraisal of the minimum energy conformation of each stereoisomer and molecular
docking at Jak3 was performed in an effort to better understand each compounds selectivity and potency
profiles.
Introduction
The potent and selective Janus kinase 3 (Jak3a) inhibitor
3-((3R,4R)-4-methyl-3-(methyl(7H-pyrrolo[2,3-d]pyrimidin-4-
yl)amino)piperidin-1-yl)-3-oxopropanenitrile (1) (CP-690,550)
was reported in 2003 as an orally active immunosuppressant
for autoimmune disease and transplant patients.1 The structure
was revealed as a substituted piperidine linked to a deazapurine
core (1) (Figure 1). Interestingly, the initial report did not
designate the stereochemistry at the 3 and 4 positions of the
substituted piperidine ring. Reports within the patent literature2-4
and subsequent manuscripts5,6 have denoted the structure as the
enantiopure (3R,4R) analogue 1. At the time of writing, 1 has
demonstrated efficacy in phase 2 clinical evaluation as an
immunosuppressive for renal transplant rejection7 and for
treatment of rheumatoid arthritis.8 Undoubtedly, a major
Figure 1. The chemical structure of 1 and related stereoisomers 2, 3,
foundation for the clinical success of this agent is the potent
and selective Jak3 inhibition. The original report provided
evidence that 1 inhibited Jak3 with an IC50 value of 1 nM while
inhibiting Jak2, Jak1, Rock-II, and Lck with IC50 values of 20,
and 4.
112, 3400, and 3870 nM, respectively.1 A panel of 28 other
kinases did not demonstrate any relevant inhibition. Recently,
Karaman et al. presented the interaction maps for 38 clinically
relevant kinase inhibitors across a panel of 317 kinases.9 The
manuscript included 1 and reported the binding potential at Jak3
and Jak2 as 2.2 and 5 nM (Kd) [the primary kinase domain of
Jak1 was not incorporated in this screen], respectively. The
report included additional binding for 1 at Camk1 (Kd of 5000
nM), DCamkL3 (Kd of 4.5 nM), Mst2 (Kd of 4300 nM), Pkn1
(Kd of 200 nM), Rps6ka2 (Kin.Dom.2-C-terminal) (Kd of 1400
nM), Rps6ka6 (Kin.Dom.2-C-terminal) (Kd of 1200 nM), Snark
(Kd of 420 nM), Tnk1 (Kd of 640 nM), and Tyk2 (Kd of 620
nM).
* To whom correspondence should be addressed. Phone: 301-217-4079.
Fax: 301-217-5736. E-mail: craigt@nhgri.nih.gov. Address: Dr. Craig J.
Thomas, NIH Chemical Genomics Center, NHGRI, National Institutes of
Health 9800 Medical Center Drive, Building B, Room 3005, Mail Stop
Code: 3370, Bethesda, MD 20892-3370.
†
NIH Chemical Genomics Center, National Human Genome Research
Institute, National Institutes of Health.
‡
Molecular Immunology and Inflammation Branch, National Institute
of Arthritis and Musculoskelatal and Skin Diseases, National Institutes of
Health.
§
Laboratory of Biological Modeling, National Institute of Diabetes and
Digestive and Kidney Diseases, National Institutes of Health.
|
Despite these additional activities, 1 remains a remarkably
selective kinase inhibitor. In a recent report, Changelian et al.
related the clinical success of Jak3 inhibitors directly to their
selectivity.10 As the binding of any small molecule to a protein
target is inextricably linked to its structure, we found the
Transplantation and Autoimmunity Branch, National Institute of
Diabetes and Digestive and Kidney Diseases, National Institutes of Health.
Kelly Services.
a Abbreviations: Jak, Janus kinase; Tyk, Tyrosine kinase; IL, interleukin;
Stat, signal transducers and activators of transcription; MCMM, Monte Carlo
multiple minimum.
10.1021/jm801142b This article not subject to U.S. Copyright. Published 2008 by the American Chemical Society
Published on Web 11/19/2008

Janus Kinase Inhibitor CP-690,550
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 24 8013
Scheme 1a
a Note: strategy and yields reflect those achieved and reported in Org. Process Res. DeV. 2005, 9, 51-56.
Scheme 2
stereospecific nature of 1 and its selectivity against over 300
kinases to be of interest. Hoping to explore this facet of the
molecule, we first set out to synthesize 1 and its three related
stereoisomeric derivatives (analogues 2, 3, and 4) (Figure 1).
Inhibition of Stat5 Phosphorylation by 1, 2, 3, and 4. With
1 and its three related stereoisomeric derivatives (2, 3, and 4)
in hand, we set out to ascertain each compounds ability to
effectively inhibit Jak3. The Jak-Stat signaling pathway is a
major regulatory element for gene transcription and plays a key
role in processes such as immunoregulation and cellular
proliferation and differentiation.13 Jak3 natively associates with
the common gamma chain γc forming a shared receptor for
selected cytokines [including interleukin (IL)-2, IL-4, IL-7, IL-
9, IL-15, and IL-21].14 Upon cytokine binding, Jak3 is phos-
phorylated (presumably autophosphorylation), allowing signal
transducers and activators of transcription (Stats) to bind to the
cognate cytokine receptors via conserved Src homology 2 (SH2)
domains.15 Receptor-bound Stats are phosphorylated, dimerize
(both homo and heterodimerization), and translocate to the
nucleus to trigger gene transcription. To examine cellular Jak3
activity directly, we analyzed enriched, human CD4+ T cells
isolated from PBMC’s incubated with each compound at
relevant concentrations (5, 50, and 500 nM) and a DMSO
control prior to stimulation with IL-2. The degree of Stat5
phosphorylation was analyzed from cell lysates via immunob-
lotting with an antiphospho-Stat5 mAb (antiactin mAb was
performed as a control to confirm equal protein loading) (Figure
2, left blot). From this experiment, it was clear that only 1
maintained the ability to affect Stat5 phosphorylation at the
concentrations tested, highly suggesting that the alternate
stereochemical configurations of the molecule had deleterious
effects on Jak3 inhibition.
Results
Synthesis of 1, 2, 3, and 4. The synthetic route undertaken
by Pfizer has evolved to ultimately rely upon a 4-step trans-
formation yielding the requisite (3R,4R)-1-benzyl-N,4-dimeth-
ylpiperidin-3-amine from 4-methylpyridin-3-amine (Scheme 1).5
Crystallization with a di-p-toluoyltartrate salt was utilized to
achieve enantiopurity following reduction of the substituted
pyridine derivative. This route provides an elegant and efficient
means to yield kilograms of the enantiomerically pure material
needed for efficient production of 1. It does not, however,
provide a means to investigate 3,4-trans analogues of the
piperidine ring. To explore the desired alternate stereochemical
possibilities, we expanded upon a method described by Ledous-
sal and co-workers that relies upon the stereocenter that is set
within Garner’s aldehyde and a key step involving the ring-
closing metathesis reaction (Scheme 2).11 Here, the ultimate
stereocenter at C3 of the piperidine ring is set by the choice of
L-serine and utilizes precedented chemistry12 to arrive at (R)-
tert-butyl 2,2-dimethyl-4-(prop-1-en-2-yl)oxazolidine-3-carbox-
ylate (6) (see Supporting Information for full synthetic details).
Although several deviations from the reported work by Ledous-
sal and co-workers11 were necessary, the general strategy
provided (R)-tert-butyl 1-(allyl(benzyl)amino)-3-methylbut-3-
en-2-ylcarbamate (7) in good yields. Application of the Grubbs
second generation catalyst in refluxing dichloromethane afforded
the requisite piperidine derivative 8 in yields typically exceeding
90%. Hydrogenation of the 3,4-alkene moiety resulted in the
chromatographically separable piperidines 9 and 10. Following
separation, the remainder of the synthesis followed the synthetic
strategy validated by White and co-workers to arrive at both 1
and 2.5 Utilizing D-serine as the starting material and following
the same route allowed synthetic elaboration of 3 and 4.
Diastereomeric purity was examined via reverse phase HPLC
analysis and enantiomeric purity was verified via chiral HPLC
methods (see Supporting Information for details).
IL-12 is another important immunoregulatory cytokine. The
IL-12 receptor comprises two subunits that associate with Jak2
and Tyk2 and activates Stat4.16,17 A primary selectivity issue
for 1 is its reported downregulation of Jak2. We examined the
ability of each compound to block the phosphorylation of Stat4
within IL-12 stimulated cells (Figure 2, right blot). The results
demonstrate no clear inhibition by 1 or its related stereoisomers.
This suggests that 1 is capable of selectively inhibiting Jak3
without disrupting the functions of Jak2 or Tyk2 in a cellular
environment at the concentrations tested.
Analysis of Kinase Selectivity. To fully understand these
compounds potential, we pursued a direct analysis of each

8014 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 24
Jiang et al.
Figure 2. Gel analysis showing the inhibitory ability of 1, 2, 3, and 4 to disrupt Stat5 phosphorylation (left blot) or Stat4 phosphorylation (right
blot) in human CD4+ cells. CD4+ T cells were rested and incubated with DMSO control (lane 2), 1 (lane 3-5), 2 (lane 6-8), 3 (lane 9-11), or
4 (lane 12-14) at increasing concentrations (5, 50, and 500 nM, respectively) for 1 h before being stimulated with IL-2 (1000 µg/mL; left blot) or
IL-12 (100 ng/mL; right blot) for 15 min or incubated in medium without cytokine stimulation (lane 1; both gels). Cell lysates were partitioned,
stripped, and immunoblotted with antiphospho-Stat5 (rabbit) mAb (left blot, top gel) or antiphospho-Stat4 (rabbit) mAb (right blot, top gel). Antiactin
mAb (mouse) was used to confirm equal loading of protein across lanes (bottom gels).
Figure 3. Dendrogram representation of the selectivity profile for kinase inhibition by 1 and stereoisomers 2, 3, and 4 within a panel of 354
kinases. Activity for 1: DCamKL3 ) 4.5 nM, Jak1 ) 3 nM, Jak2 ) 2 nM, Jak3 ) 0.7 nM, Tyc2 ) 250 nM, Pkn1 ) 200 nM, Snark ) 420 nM,
Tnk1 ) 640 nM. Activity for 2: Jak2 ) 600 nM, Jak3 ) 190 nM, Mst1 ) 250 nM, Mst2 ) 450 nM. Activity for 3: Jak2 ) 270 nM, Jak3 ) 180
nM. Activity for 4: Jak2 ) 420 nM, Jak3 ) 150 nM, Map4K3 ) 460 nM, Map4K5 ) 790 nM.
stereoisomer against purified Jak3. Further, 1 represents a novel
and unique chemotype for kinase inhibition and it was of interest
to profile each stereoisomer across a panel of kinases. Recently,
Ambit Biosciences reported the aforementioned quantitative
analysis of 38 known kinase inhibitors (including 1) across a
panel of 317 kinases.9 We submitted 1 and the stereoisomeric
analogues 2, 3, and 4 across the same panel (at present
containing 354 kinases). The initial profile provides activity as
a percentage of DMSO control (for the full report, see the
Supporting Information). Activities beyond a selected threshold
were submitted for Kd determinations, and the results are shown
as a dendrogram representation in Figure 3. The profile of 1
closely matched the published data [with 2 notable exceptions:
Mst2, “no hit” (reported IC50 of 4.2 µM), and Tyk2, “no hit”
(reported IC50 of 620 nM)]. The profile additionally found a
Kd of 210 nM for 1 at Rock (Rock was not part of the original
317 kinases screened in ref 9). Full Kd determinations for 1
were pursued for the four related Jak targets as well as the Jak1
(JH2domain-pseudokinase). These results confirmed that 1 binds
Jak3 and Jak2 nearly equipotently (Kd of 0.75 nM for Jak3 and
a Kd of 1.8 nM for Jak2). The disassociation constants for 1 at
Jak1 and Tyk2 were recorded at 1.7 and 260 nM, respectively
(as compared to the earlier report of 640 nM binding at Tyk2).
No affinity was observed for 1 at the Jak1 (JH2domain-
pseudokinase). These data contrast sharply with the original
report denoting a higher degree of selectivity for Jak3 over Jak2
and Jak1. Interestingly, these results conflict with the cell-based
study showing little or no inhibition of Stat4 phosphorylation
by 1.
The profile results for 2, 3, and 4 indicate that each ste-
reoisomer retains a degree of affinity for Jak3 and Jak2, although
the potency of the interaction drops significantly. The profile
for 3 (the enantiomer of 1) showed solitary activity at Jak3 and
Jak2 (Kd’s of 180 and 270 nM, respectively). Enantiomers 2
and 4 had similar Kd’s for Jak3 and Jak2 but also maintained
several novel interactions. For instance, 2 was found to have
modest binding potential for Mst1 and Mst2 (Kd’s of 320 and
450 nM, respectively). Analogue 4 was found to have modest
binding at Map4K3 and Map4K5 (Kd’s of 460 and 790 nM,
respectively). Mst and Map4K kinase subfamilies reside on the
related STE20 and STE7 branches of the kinome. That enan-
tiomers 2 and 4 show activity at these related targets, suggesting
that this chemotype may represent a novel starting point for
the development of selective inhibitors of these important kinase
classes.
Minimum Energy Conformations of Unbound 1, 2, 3,
and 4 in Water. Chirality, pharmacology, and drug discovery
are intertwining subjects dating back to the early use of quinine,
atropine, and opiates to several of today’s best known drugs.
In each instance, the chiral nature of these small molecules plays

Janus Kinase Inhibitor CP-690,550
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 24 8015
in the ultimate three-dimensional structures of these agents. This
broadly accepted phenomenon is intensified when placing chiral
substituents on five- and six-membered ring structures due to
hypersensitivity in ring conformations.
Docking of 1, 2, 3, and 4 at Jak3. There are four members
of the Jak family of kinases; Jak1, Jak2, Jak3, and Tyrosine
kinase 2 (Tyk2).15 Each member of this family retains seven
conserved sequence regions (or Jak homology domains): the
JH1 domain (or kinase domain), the JH2 domain (or pseudoki-
nase domain), the JH3 and JH4 domains (both are SH2
domains), and JH6 and JH7 domains (collectively, the FERM
domain or four-point-one, ezrin, radixin, and moesin homology
domain).13,15 In 2005, Boggon et al. reported the crystal structure
for the Jak3 kinase domain bound to a staurosporine analogue
23 (AFN941).19
Utilizing this structure as a template, the four stereoisomers
1-4 were docked at the Jak3 catalytic cleft using Glide 4.5 in
order to shed light on the mechanistic preference for the binding
of 1.20 In particular, on the basis of the crystallographic
coordinates of the Jak3-23 complex, the inhibitors were docked
at the ATP-binding site, lined by residues from the N-terminal
lobe on the roof of the pocket (Leu828, Gly829, Val836, and
Ala853), the C-terminal lobe on the floor of the pocket (Leu956,
Gly908, and Cys909), and the hinge region (from Met902 to
Leu905). The opening of the cleft is defined by hydrophilic
residues like Arg953, Asn954, Asp949, and Gln988. Interactions
with residue backbones of the hinge region define the binding
motif of many kinase inhibitors (including 23 with Jak3). We,
therefore, utilized specified hydrogen bonds between Glu903
and Leu905 and each stereoisomer as a criterion for retrieving
the ligand poses from the docking results along with the docking
score and the energetic contributes to the binding interactions.
Figure 4. Molecular models of the lowest energy conformers of 1-4
generated through Monte Carlo conformational searches. The model
of 1 is colored by atom type (gray: carbon; red: oxygen; blue; nitrogen),
while the models of 2, 3, and 4 are colored orange, red, and yellow,
respectively. Compounds 2, 3, and 4 are overlapped to 1 by superim-
position of the six member ring atoms. All compounds prefer the chair
conformation; however, 1 and 4 show a θ of ∼180°, while 3 and 2
adopt the diametrically opposite chair conformation, with a θ of ∼0°.
a role in their biochemical efficacy. With a deeper understanding
of the chiral nature of 1 and its kinase selectivity profile, we
explored the role of the (4R)-methyl substituent and the (3R)-
deazapurine moiety in defining its minimum energy conforma-
tion and how this probable conformation facilitates binding to
Jak3.
The conformational space of the unbound inhibitors 1-4 was
studied by subjecting the molecules to two consecutive Monte
Carlo multiple minimum (MCMM) conformational searches
(initially in vacuo, subsequently in implicit water). The resulting
minimum energy models are shown in Figure 4 and can be
discussed utilizing the truncated Fourier (TF) series-based
coordinates for the description of six-membered ring puckering
established by Haasnoot18 (plots representing the azimuthal θ
angles, the two torsional angles describing the orientation of
the base, and percentage energy increase of the various
conformers obtained for 1-4 and are provided in the Supporting
Information). The six-membered ring of all the compounds can
adopt two diametrically opposite chair conformations, repre-
sented by θ angles of 0° and 180°. Enantiomers 1 and 3, which
have the methyl substituent and the base on the same side of
the ring plane, show a clear preference for having the methyl
substituent in an equatorial position and the deazapurine moiety
in an axial position (preferred θ of ∼180° for 1 and ∼0° for 3).
Enantiomers 2 and 4 position these substituents on opposing
sides of the plane of the piperidine ring conferring a stronger
preference for having the two substituents in equatorial positions
(preferred θ of ∼0° and ∼180° for 2 and 4, respectively).
Interestingly, the signal for piperidine ring C3-H of 1 was noted
at 4.78 ppm, while the C3-H of 2 was found at 4.32 ppm
(assigned via COSY analysis). The relative downfield shift in
1 highly suggests a more equatorial character for the C3-H of
1 and relative axial character for the C3-H of 2, which is
consistent with the results from the MCMM searches.
The results from the highest scoring Jak3-1 docking complex
are shown in Figure 5 and illustrate that the N3 and N9 nitrogens
of the 7-deazapurine moiety participate in key hydrogen bonds
with residues Glu903 and Leu905. These interactions mimic
hydrogen bonds found within the crystal structure of Jak3 with
23. Another significant interaction involves hydrogen bonds
formed between the nitrile function and Arg953 at the opening
of the cleft. This docking pose further validates the notion that
the 4R-methyl group occupies an equatorial position while the
3R-base moiety is directed into an axial position in the chair
conformation of the piperidine ring.
Comparing the docking poses for 1, 2, 3, and 4 found in the
highest scoring Jak3 docking complexes to the minimum energy
structures of the unbound 1, 2, 3, and 4 from the conformational
analyses provides valuable insight into the superior binding
associated with the (3R,4R) stereochemical configuration of 1.
Figure 6 shows the predicted unbound conformation for each
compound overlaid with the conformation associated with
docking at Jak3. From this rendering, it is clear that only 1 docks
with Jak3 in a conformation that extensively resembles the
compounds minimum energy conformation (Figure 6a). For 2,
the six-membered ring assumes a half-chair conformation with
both the substituent in equatorial position (Figure 6b). Com-
pound 3 docked with the six-membered ring in a chair
conformation and, contrary to the conformational preferences
revealed by the MCMM search, the methyl and base substituents
were found in the axial and equatorial position, respectively
(Figure 6c). Finally, compound 4 docked with the six-membered
ring in a twist-boat conformation with both methyl and base
substituents in the equatorial position (Figure 6d). These data
indicate that compounds 2, 3, and 4 are forced to adopt unlikely
Using the deazapurine base as the anchor point for discussion
(and likely for binding to Jak3), it is clear that even the fairly
“minor” change of the stereochemical configuration of the
methyl group in structures 1 and 2 results in significant changes

8016 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 24
Jiang et al.
Figure 5. Proposed binding mode of 1 in the catalytic site of Jak3. The protein is shown with the ribbons representation with the N-lobe colored
in cyan, the C-lobe in orange, and the hinge region in yellow. The residues in the binding pocket are colored by atom type. Compound 1 is shown
as ball-and-stick model with carbon atoms colored in green. The major polar interactions between compound 1 and the residues in the pocket are
highlighted as dotted lines.
Comparison between the catalytic pockets of crystal structures
of Jak3 and Jak2 revealed conformational differences in the
glycine-rich loop and the activation loop that result in a rather
tighter pocket for Jak2. Docking of 1 within the crystal structure
of the catalytic cleft of Jak225 (see Supporting Information)
suggests that the complexes of 1 with both Jak3 and Jak2 are
decidedly similar. Only three residues in spatial proximity to
the binding site of 1 at Jak3 and Jak2 are divergent: Jak3
Ala966-Jak2 Gly993, in proximity of the DFG motif, Jak3
Cys909-Jak2 Ser936, at the end of the hinge region, and Jak3
Gln988-Jak2 Glu1015, in the activation loop. Cycles of
MCMM conformational search performed on the Jak3-1
complex, granting flexibility to the ligand and the residues within
a 4 Å radius allow for a potential hydrogen bond between the
nitrile function and Gln988, an interaction that would be missing
in Jak2. However, the docking pose of 1 in Jak2 does retain
the key hydrogen bond with Arg980 (Arg953 in Jak3).
It is unclear how this lone deviation may affect binding, but
Figure 6. Superposition of the six-membered ring between the lowest
given the relative Kd and IC50 values reported for 1 at both
energy conformation of compounds 1-4 (colored in green, orange, red,
targets, the difference is presumably negligible. This is also
and yellow) and the respective best scored docking poses (colored by
consistent with the fact that, due to the different conformation
of the portion of the activation loop located immediately prior
to the APE motif, in Jak2 Glu1015 points away from the binding
site and would not be in proximity with the nitrile moiety.
From the docking comparisons, the similar disassociation
constants for 1 at Jak3 and Jak2 are not surprising. Early results
from the clinical use of 1 demonstrate efficacy but also unwanted
anemia and neutropenia.26 This suggests that unwelcome
downregulation of Jak2 is occurring to an appreciable extent.
Nonetheless, phase 1 clinical evaluations demonstrated a reason-
able safety profile and numerous phase 2 evaluations are
currently underway. The IC50 values reported by Changelian et
al. indicate a small degree of selectivity between Jak3 and Jak2
(1 nM for Jak3 versus 20 nM for Jak2). It is unclear if the
selectivity differences noted in these studies are the result of
the relative Km (ATP) of each kinase. Nonetheless, whether 1
atom type).
high energy conformations in order to bind effectively at the
Jak3 catalytic site.
Discussion
Inhibition of Jak3 and Jak2 by 1. Jak3 represents an
intriguing therapeutic target.21 Jak3 is primarily expressed within
T cells and NK cells and specific mutations to Jak3 result in
T-B+NK- severe combined immunodeficiency (SCID).22 Un-
surprisingly, the knockout phenotype for Jak3 is a viable but
immunocompromised animal.23 Conversely, Jak2 is ubiquitously
expressed and knockouts are embryonic lethal.24 Given these
data, substantial effort has been invested in the search for highly
selective Jak3 inhibitors.
Jak2 possesses a high degree of homology to Jak3 (47%)
and is particularly homologous at the kinase active site.19

Janus Kinase Inhibitor CP-690,550
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 24 8017
utilizing KINOMEscan. Activity is recorded via a competition
binding assay of selected kinases that are fused to a proprietary
tag. Measurements of the amount of kinase bound to an im-
mobilized, active-site directed ligand in the presence and absence
of the test compound provide a % of DMSO control for binding of
ligand. Activities between 0 and 10 were selected for Kd determina-
tions. Dendrogram representations were generated by an in-house
visualization tool designated PhyloChem. Dendrogram clustering
and apexes are based on the human phylogenetic kinase data
available at http://kinase.com/human/kinome.
Analysis of Stat5 and Stat4 Phosphorylation. Human CD4+
positive cells were enriched from peripheral blood mononuclear
cells obtained from a healthy donor by magnetic separation (CD4+
MACS beads). CD4+ cells were activated for 3 days with plate
bound anti-CD3 and anti CD28 antibodies (5 ug/mL each) and then
expanded for another 4 days in the presence of IL-2 (50 U/mL).
Cells were rested overnight in 1% RPMI and preincubated with 1,
2, 3, 4 or DMSO control for 1 h at indicated concentrations [(5,
50, 500 nM) DMSO concentration was equal in all preparations]
and then activated with IL-2 (1000 µg/mL) or IL-12 (100 ng/mL)
for 15 min. Cells (10 × 106/condition) were lysed in 1% Triton-X
lysis buffer and equal amounts of cell lysate were run in NuPage
Bis-Tris gel (4-12% gradient). Proteins were transferred onto
nitrocellulose membrane. Detection was done with indicated
antibodies using Odyssey Western blotting system according to
manufacturer’s instructions. [http://www.licor.com/bio/Quantita-
tiveWesterns/index.jsp] Primary antibodies used: antiactin mouse
mAb (Chemicon), 1:5000, antiphospho-Stat5 rabbit mAb (Cell
Signaling), antiphospho-Stat4 (Zymed) 1:1000, and secondary goat-
antimouse IgG Ab (Rockland) and goat-antirabbit IgG (Molecular
Probes).
binds/inhibits Jak2 at 1 or 20 nM concentrations, it is likely
that the physiological levels of the drug will surpass the amount
needed for effective downregulation of Jak2. The more compel-
ling experiments, however, are cell-based studies such as the
assessment of inhibition of Stat4 phosphorylation by 1 (Figure
2) and the previous report that 1 effectively inhibits IL-2
stimulated cell proliferation (Jak3 dependent) while having much
weaker effect on granulocyte macrophage-colony stimulation
factor induced proliferation (Jak2 dependent). It remains to be
seen if these results provide clues into the method by which
cytokine receptor/Jak pairs initiate signaling cascades.
Conclusion
Kinases are among the most intriguing therapeutic targets in
the human proteome and kinase inhibitors are becoming staples
of the pharmacopeia. A primary doctrine of drug design is to
limit the number of chiral centers placed into small molecules
intended for clinical use for a myriad of reasons. 1 goes against
convention and incorporates not one, but two chiral centers.
Using a combination of molecular modeling, target profiling
and cell-based analyses we have shown that the chiral nature
of 1 is a key facet that defines its ability to bind and inhibit its
primary target. Additionally, discrete stereoisomers of 1 may
prove useful starting points for novel small molecules targeting
alternate branches of the kinome. Finally, the divergence of
activity for 1 in purified protein assays versus cell-based assays
remains an intriguing characteristic of this compound and should
be explored further.
Monte Carlo Conformational Searches. Compounds 1-4 were
sketched in Maestro and subjected to 100 steps of Monte Carlo
multiple minimum (MCMM) conformational search performed in
vacuo by means of MacroModel (Schro¨dinger LLC, New York,
NY, 2007).27,28 The lowest energy conformer was subsequently
used as the starting point for additional 1000 steps of MCMM
search, this time performed using water as implicit solvent (surface
generalized Born model). All calculations were conducted with the
OPLS_2005 force field.
Methods
Synthesis of 1, 2, 3, and 4. Complete procedures for the synthesis
of 1, 2, 3, and 4 are presented in the Supporting Information. The
general strategy followed precedented chemistry from refs, 5, 11,
and 12. Analysis of diastereopurity and enantiopurity were deter-
mined through reverse phase and chiral phase HPLC methods.
Proton NMR for all enantiomers was identical.
Protein Preparation. The X-ray crystallographic structure of
the human Jak3 kinase domain in a catalytically active state and in
complex with the staurosporine derivative 23 was retrieved from
the Protein Data Bank (PDB code: 1YVJ, atomic resolution 2.55
Å).19 The protein structure was prepared for the docking studies
using the Protein Preparation Wizard tool implemented in Maestro
(Schro¨dinger LLC, New York, NY, 2007). All crystallographic
water molecules and other chemical components were deleted, the
right bond orders were assigned, and the hydrogen atoms were
added to the protein. Arginine and lysine side chains were
considered as cationic at the guanidine and ammonium groups, and
the aspartic and glutamic residues were considered as anionic at
the carboxylate groups. The hydrogen atoms were subsequently
minimized employing the Polak-Ribiere conjugate gradient (PRCG)
method until a convergence to the gradient threshold of 0.05 kJ/
(mol Å). The atomic charges were computed using the OPLS_2005
force field.
3-((3R,4R)-3-((1H-Indol-4-yl)(methyl)amino)-4-methylpiperi-
din-1-yl)-3-oxopropan-enitrile (1). [R]D ) +10.4 (c 0.64,
25
MeOH); NMR spectra are complicated due to amide rotomers. 1H
NMR (400 MHz, CDCl3) δ 12.67 and 12.63 (s, 1H), 8.39 and 8.37
(s, 1H), 7.40-7.44 (m, 1H), 6.80-6.90 (m, 1H), 4.78 (brs, 1H),
4.08-4.20 and 3.96-4.06 (m, 2H), 3.68-3.86 (m, 2H), 3.37-3.46
and 3.14-3.22 (m, 2H), 3.35 (s, 3H), 2.34-2.46 (m, 1H),
1.76-1.95 (m, 1H), 1.52-1.67 (m, 1H), 1.07 and 1.04 (d, J ) 7.4
Hz, 3H). 13C NMR (100 MHz, CDCl3) δ 162.52 and 162.46, 159.3,
159.0, 145.4 and 145.1, 124.0 (br), 117.0 and 116.9, 105.2 and
104.9, 102.7, 56.8 and 55.7, 45.2, 43.1 and 42.1, 38.7 and 35.8,
31.8 and 31.4, 30.7, 25.8 and 25.7, 14.4 and 13.9. IR (neat) 3130,
2965, 1619, 1535, 1452, 1197, 1134, 1053, 906, 838, 720 cm-1
.
HRMS (FAB) Calcd for C16H21N6O (M + H) 313.1777, found
313.1778.
3-((3R,4S)-3-((1H-Indol-4-yl)(methyl)amino)-4-methylpiperi-
25
din-1-yl)-3-oxopropane- nitrile (2). [R]D ) +74.8 (c 0.60,
Molecular Docking. All compounds were docked inside the
active site of Jak3 using Glide 4.5,20 the automated docking program
implemented in the Schro¨dinger package (Schro¨dinger LLC, New
York, NY, 2007). The binding site was defined around the position
occupied by the cocrystallized ligand in the Jak3 complex structure
1YVJ. In the receptor grid generation, a 10 × 10 × 10 Å3 cubic
docking box was generated and the known H-bond interactions
between most of the kinase inhibitors and the backbone of the hinge
segment were enforced, defining the backbone amino groups of
Leu905 and the backbone carboxylic groups of Glu903 as potential
H-bond donor and acceptor respectively. The XP mode of Glide
was utilized. The obtained complexes between Jak3 and the best
scored pose of each compound were then submitted to 1000 steps
of MCMM conformational search performed with the OPLS_2005
MeOH); the NMR are complicated due to the rotamers of the amide.
1H NMR (400 MHz, CDCl3) δ 12.58 (brs, 1H), 8.38 and 8.37 (s,
1H), 7.42 (s, 1H), 6.96 and 6.87 (d, J ) 2.4 Hz, 1H), 4.32 (brs,
1H), 3.96-4.16 (m, 2H), 3.62-3.72 (m, 1H), 3.37 and 3.31 (s,
3H), 3.05-3.16 (m, 1H), 2.86-2.98 (m, 1H), 2.62-2.76 (m, 1H),
2.06-2.22 (m, 1H), 1.78-1.87 (m, 1H), 1.18-1.32 and 1.34-1.48
(m, 1H), 0.85 and 0.80 (d, J ) 6.4 Hz, 3H). 13C NMR (100 MHz,
CDCl3) δ 162.6 and 162.4, 159.3, 159.0, 145.7 (br), 124.0 (br),
117.0 and 116.9, 105.1 and 104.7, 102.5, 47.3 and 46.1, 43.7, 42.5,
41.3, 33.6, 33.0 and 32.9, 25.9 and 25.8, 18.5. IR (neat) 3123, 2963,
1599, 1541, 1452, 1198, 1137, 1056, 907, 830, 720 cm-1. HRMS
(FAB) Calcd for C16H21N6O (M + H) 313.1777, found 313.1775.
Profiles of 1, 2, 3, and 4. Kinase profiles were performed by
Ambit Biosciences (San Diego, CA: http://www.ambitbio.com/)

8018 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 24
Jiang et al.
P. T.; Faraoni, R.; Floyd, M.; Hunt, J. P.; Lockhart, D. J.; Milanov,
Z. V.; Morrison, M. J.; Pallares, G.; Patel, H. K.; Pritchard, S.;
Wodicka, L. M.; Zarrinkar, P. P. A quantitative analysis of kinase
inhibitor selectivity. Nat. Biotechnol. 2008, 26, 127–132.
(10) Changelian, P. S.; Moshinsky, D.; Kuhn, C. F.; Flanagan, M. E.;
Munchhof, M. J.; Harris, T. M.; Doty, J. L.; Sun, J.; Kent, C. R.;
Magnuson, K. S.; Perregaux, D. G.; Sawyer, P. S.; Kudlacz, E. M.
The specificity of Jak3 kinase inhibitors. Blood 2008, 111, 2155–2157.
(11) Hu, X. E.; Kim, N. K.; Ledoussal, B. Synthesis of trans-(3S)-Amino-
(4R)-alkyl-and-(4S)-Aryl-piperidines via Ring-Closing Metathesis
Reaction. Org. Lett. 2002, 4, 4499–4502.
(12) Williams, L.; Zhang, Z.; Shao, F.; Carroll, P. J.; Joullie´, M. M.
Grignard Reactions to Chiral Oxazolidine Aldehydes. Tetrahedron
1996, 52, 11673–11694.
(13) Yamaoka, K.; Saharinen, P.; Pesu, M.; Holt, V. E. T., III; Silvennoinen,
O.; O’Shea, J. J. The Janus kinases (Jaks). Genome Biol. 2004, 5,
253.1253.6.
force field. The energy minimization was employed with PRCG
procedure until convergence to the gradient threshold of 0.05 kJ/
(mol Å). The reproduction of the binding mode of 23 in the catalytic
site of Jak3 as in the crystallographic structure 1YVJ (rmsd value
about 0.43 Å) validated the docking and MCMM search protocol
used for this study.
Acknowledgment. We thank Dr. David Maloney and Dr.
Douglas Auld for advice and help during the preparation of this
manuscript. We thank Allison Peck for critical reading of this
manuscript. This research was supported by the Molecular
Libraries Initiative of the National Institutes of Health Roadmap
for Medical Research, the Intramural Research Program of the
National Human Genome Research Institute, the National
Institute of Arthritis and Musculoskeletal and Skin Diseases,
and the National Institute of Diabetes and Digestive and Kidney
Diseases, National Institutes of Health.
(14) O’Shea, J. J.; Park, H.; Pesu, M.; Borie, D.; Changelian, P. New
strategies for immunosuppression: interfering with cytokines by
targeting the Jak/Stat pathway. Curr. Opin. Rheumatol. 2005, 17, 305–
311.
(15) Bhandari, R.; Kuriyan, J. Jak-Stat signaling. In Handbook of Cell
Signaling, 1st ed.; Bradshaw, R., Dennis, E., Eds.; Academic Press:
New York, 2003; 343-347.
(16) Bacon, C. M.; McVicar, D. W.; Ortaldo, J. R.; Rees, R. C.; O’Shea,
J. J.; Johnston, J. A. Interleukin 12 (IL12) Induces Tyrosine Phos-
phorylation of JAK2 and TYK2: Differential Use of Janus Family
Tyrosine Kinases by IL-2 and IL-12. J. Exp. Med. 1995, 181, 399–
404.
Supporting Information Available: Synthetic procedures,
selected spectra, expanded experiments and complete profile results.
This material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Changelian, P. S.; Flanagan, M. E.; Ball, D. J.; Kent, C. R.; Magnuson,
K. S.; Martin, W. H.; Rizzuti, B. J.; Sawyer, P. S.; Perry, B. D.;
Brissette, W. H.; McCurdy, S. P.; Kudlacz, E. M.; Conklyn, M. J.;
Elliot, E. A.; Koslov, E. R.; Fisher, M. B.; Strelevitz, T. J.; Yoon, K.;
Whipple, D. A.; Sun, J.; Munchhof, M. J.; Doty, J. L.; Casavant, J. M.;
Blumenkopf, T. A.; Hines, M.; Brown, M. F.; Lillie, B. M.;
Subramanyam, C.; Shang-Poa, C.; Milici, A. J.; Beckius, G. E.; Moyer,
J. D.; Su, C.; Woodworth, T. G.; Gaweco, A. S.; Beals, C. R.; Littman,
B. H.; Fisher, D. A.; Smith, J. F.; Zagouras, P.; Magna, H. A.;
Saltarelli, M. J.; Johnson, K. S.; Nelms, L. F.; Des Etages, S. G.; Hayes,
L. S.; Kawabata, T. T.; Finco-Kent, D.; Baker, D. L.; Larson, M.; Si,
M.-S.; Paniagua, R.; Higgins, J.; Holm, B.; Reitz, B.; Zhoa, Y.-J.;
Morris, R. E.; O’Shea, J. J.; Borie, D. C. Prevention of Organ Allograft
Rejection by a Specific Janus Kinase 3 Inhibitor. Science 2003, 302,
875–878.
(2) Blumenkopf, T. A.; Flanagan, M. E.; Munchhof, M. J. Patent EP
1235830, 2004; Japanese Patent JP 2003516405, 2003; U.S. Patent
US 2001053782 2001; U.S. Patent US6627754, 2003; Patent WO
0142246, 2001.
(3) Ruggeri, S. G.; Hawkins, J. M.; Makowski, T. M.; Rutherford, J. L.;
Urban, F. J. Japanese Patent JP 2007039455, 2007; Patent WO
2007012953, 2007.
(4) Flanagan, M. E.; Li, Z. J. Japanese Patent JP 2005511696, 2005; U.S.
Patent US 2003130292, 2003; Patent WO 03048162, 2003.
(5) Cai, W.; Colony, J. L.; Frost, H.; Hudspeth, J. P.; Kendall, P. M.;
Krishnan, A. M.; Makowski, T.; Mazur, D. J.; Phillips, J.; Ripin,
D. H. B.; Ruggeri, S. G.; Stearns, J. F.; White, T. D. Investigation of
Practical Routes for the Kilogram-Scale Production of cis-3-Methyl-
amino-4-methylpiperidines. Org. Process Res. DeV. 2005, 9, 51–56.
(6) Sorbera, L. A.; Serradell, N.; Bolo´s, J.; Rosa, E.; Bozzo, J. CP-690,550.
Drugs Future 2005, 32, 674–680.
(7) Busque, S.; Leventhal, J.; Brennan, D.; Klintmalm, G.; Steinberg, S.;
Shah, T.; Lawendy, N.; Wang, C.; Chan, G. CP-690,550 (CP), a Jak3
Inhibitor, in de Novo Kidney Transplant (KT) Recipients: 6-Month
Results of a Phase 2 Trial. Am. J. Transplant 2007, (s2), 304.
(8) Kremer, J. M.; Bloom, B. J.; Breedveld, F. C.; Coombs, J.; Fletcher,
M. P.; Gruben, D.; Krishnaswami, S.; Burgos-Vargas, R.; Wilkinson,
B.; Zerbini, C. A. F.; Zwillich, S. H. A randomized, double-blind,
placebo-controlled trial of 3 dose levels of CP-690,550 versus placebo
in the treatment of acute rheumatoid arthritis, 70th Annu. Sci. Meeting
Am. Coll. Rheumatol. 2007, Abstract L40.
(17) Bacon, C. M.; Petricoin, E. F., III; Ortaldo, J. R.; Rees, R. C.; Larner,
A. C.; Johnston, J. A.; O’Shea, J. J. Interleukin 12 induces tyrosine
phosphorylation and activation of STAT4 in human lymphocytes. Proc.
Natl. Acad. Sci. U.S.A. 1995, 92, 7307–7311.
(18) Haasnoot, C. A. G. The Conformation of Six-Membered Rings
Described by Puckering Coordinates Derived from Endocyclic Torsion
Angles. J. Am. Chem. Soc. 1992, 114, 882–887.
(19) Boggon, T. J.; Li, Y.; Manley, P. W.; Eck, M. J. Crystal Structure of
the Jak3 kinase domain in complex with a staurosporine analog. Blood
2005, 106, 996–1002.
(20) Glide 4.5; Schro¨dinger, LLC: New York, 2007.
(21) Podder, H.; Kahan, B. D. Janus kinase 3: a novel target for selective
transplant immunosuppression. Expert Opin. Ther. Targets 2004, 8,
613–629.
(22) Russell, S. M.; Tayebi, N.; Nakajima, H.; Riedy, M. C.; Roberts, J. L.;
Aman, M. J.; Migone, T.-S.; Noguchi, M.; Markert, L.; Buckley, R. H.;
O’Shea, J. J.; Leonard, W. J. Mutations of Jak3 in a Patient with SCID:
Essential Role of Jak3 in Lymphoid Development. Science 1995, 270,
797–800.
(23) Thomis, D. C.; Gurniak, C. B.; Tivol, E.; Sharpe, A. H.; Berg, L. J.
Defects in B lymphocyte maturation and T lymphocyte activation in
mice lacking Jak3. Science 1995, 270, 794–797.
(24) Neubauer, H.; Cumano, A.; Muller, M.; Wu, H.; Huffstadt, U.; Pfeffer,
K. Jak2 deficiency defines an essential developmental check-point in
definitive hematopoiesis. Cell 1998, 93, 397–409.
(25) Lucet, I. S.; Fantino, E.; Styles, M.; Bamert, R.; Patel, O.; Broughton,
S. E.; Walter, M.; Burns, C. J.; Treutlein, H.; Qilks, A. F.; Rossjohn,
J. The structural basis of Janus kinase 2 inhibition by a potent and
specific pan-Janus kinase inhibitor. Blood 2006, 107, 176–183.
(26) Liu, E. H.; Siegel, R. M.; Harlan, D. M.; O’Shea, J. J. T cell-directed
therapies: lessons learned and future prospects. Nat. Immunol. 2007,
8, 25–30.
(27) Chang, G.; Guida, W. C.; Still, W. C. An Internal Coordinate Monte
Carlo Method for Searching Conformational Space. J. Am. Chem. Soc.
1998, 111, 4379–4386.
(28) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton,
M.;Caufield,C.;Chang,G.;Hendrickson,T.;Still,W.C.MacromodelsAn
Integrated Software System for Modelling Organic and Bioorganic
Molecules Using Molecular Mechanics. J. Comput. Chem. 1990, 11,
440–467.
(9) Karaman, M. W.; Herrgard, S.; Treiber, D. K.; Gallant, P.; Atteridge,
C. E.; Campbell, B. T.; Chan, K. W.; Ciceri, P.; Davis, M. I.; Edeen,
JM801142B