Discovery of potent and selective PKC-θ inhibitors

Article abstract of DOI:10.1016/j.bmcl.2006.09.056

An uHTS campaign was performed to identify selective inhibitors of PKC-θ. Initial triaging of the hit set based on selectivity and historical analysis led to the identification of 2,4-diamino-5-nitropyrimidines as potent and selective PKC-θ inhibitors. A homology model and initial SAR is presented demonstrating that a 2-arylalkylamino substituent in conjunction with suitable 4-diamino substituent are essential for achieving selectivity over many kinases. Additional hit to lead profiling is presented on selected compounds.

Full text of DOI:10.1016/j.bmcl.2006.09.056

Bioorganic & Medicinal Chemistry Letters 17 (2007) 225–230
Discovery of potent and selective PKC-h inhibitors
Charles L. Cywin,a,* Georg Dahmann,d Anthony S. Prokopowicz, III,a
Erick R. R. Young,a Ronald L. Magolda,a Mario G. Cardozo,a Derek A. Cogan,a
Darren DiSalvo,a John D. Ginn,a Mohammed A. Kashem,a John P. Wolak,a
Carol A. Homon,a Thomas M. Farrell,c Heather Grbic,c Hanbo Hu,c Paul V. Kaplita,c
Lisa H. Liu,c Denice M. Spero,c Deborah D. Jeanfavre,b Kathy M. O’Shea,b
Della M. White,b Joseph R. Woska, Jr.b and Maryanne L. Brownb
aBoehringer Ingelheim Pharmaceuticals, Inc., Department of Medicinal Chemistry, 900 Ridgebury Road,
Ridgefield, CT 06877-0368, USA
bBoehringer Ingelheim Pharmaceuticals, Inc., Department of Immunology and Inflammation, 900 Ridgebury Road,
Ridgefield, CT 06877-0368, USA
cBoehringer Ingelheim Pharmaceuticals, Inc., Department of Drug Discovery Support, 900 Ridgebury Road,
Ridgefield, CT 06877-0368, USA
dBoehringer Ingelheim Pharma GmbH and Co. KG, Department of Medicinal Chemistry, Birkendorfer Strasse 65,
D-88397 Biberach an der Riss, Germany
Received 25 July 2006; revised 18 September 2006; accepted 20 September 2006
Available online 19 October 2006
Abstract—An uHTS campaign was performed to identify selective inhibitors of PKC-h. Initial triaging of the hit set based on selec-
tivity and historical analysis led to the identification of 2,4-diamino-5-nitropyrimidines as potent and selective PKC-h inhibitors.
A homology model and initial SAR is presented demonstrating that a 2-arylalkylamino substituent in conjunction with suitable
4-diamino substituent are essential for achieving selectivity over many kinases. Additional hit to lead profiling is presented on
selected compounds.
ꢀ 2006 Elsevier Ltd. All rights reserved.
Protein kinase C theta (PKC-h) is a member of the pro-
tein kinase C serine/threonine kinase family that has
recently been shown to play a critical role in T cell sig-
naling. The novel PKCs of which PKC-h is a member
(nPKC-d,e,g,h) require DAG and PS but are calcium
independent. PKC-h, unlike most other PKCs, has
restricted expression, found only in T cells and skeletal
muscle.1 The murine knockout phenotype was specific
to defective NF-kB signaling in mature T cells.2 TCR-
mediated NF-kB activation was absent from PKC theta
À/À mature T cells but intact in thymocytes. NF-kB
activation by IL1 and TNF alpha was equivalent to
wild-type animals. T cell levels were normal. However,
T cell activation was severely depressed, as determined
by IL2 secretion and T cell proliferation. Upon T cell
activation, PKC theta is the isoform that translocates
upon antigenic stimulation of TCR.3 The phenotype of
PKC-h knockout mice is specific to T cell activation
and is a calcium independent pathway. The knockout
phenotype indicates that a PKC-h inhibitor would be
an immunosuppressive agent since it renders mature T
cells unable to proliferate and release IL2. Unlike de-
scribed immunosuppressive agents, a PKC-h inhibitor
would not alter T cell development. Autoimmune dis-
eases, such as psoriasis, and complications of trans-
plants are characterized by pathogenic responses as a
result of inappropriate T cell activation. Present treat-
ment includes cyclosporine, FK506, and steroids that re-
duce disease severity by reducing T cell activation. There
is an unmet medical need for an immunosuppressive
drug with a safer side-effect profile. This potential led
to the development of an uHTS screen in hopes of iden-
tifying novel and selective small molecule inhibitors of
PKC-h. The screen and the hit to lead process which fol-
lowed resulted in the discovery of a potent and selective
Keywords: PKC-h; Kinase; Inhibitor; Hit to lead.
*
Corresponding author. Tel.: +1 203 791 6401; fax: +1 203 837
4358; e-mail: ccywin@rdg.boehringer-ingelheim.com
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.09.056

226
C. L. Cywin et al. / Bioorg. Med. Chem. Lett. 17 (2007) 225–230
series of PKC-h inhibitors. The process of triaging the
hits and the initial SAR studies will be the focus of this
report.
selectivity could be achieved (Figs. 1 and 2).4 From these
data the pyrimidine class (highlighted with green tabs)
had a trend toward selectivity with one compound (3)
clearly showing very good selectivity.5 In contrast the
indolinones active for PKC-h (no tabs) appeared to have
no exploitable SAR for achieving selectivity.6 We there-
fore focused our attention on developing the pyrimidine
class into viable leads with good selectivity, drug like
properties, and cellular activity.
The uHTS screen yielded a dose responsive hit set
with activities that ranged from low nanomolar to
high micromolar. The compound classes in the hit
set were known to have members with cross reactivity
to other kinases. Analysis of the hit set comparing his-
torical data of the various classes in a panel of kinases
indicated possible class related cross reactivity with
VEGFR1, LYN, IR, and SYK. As these kinases have
the potential for undesirable side effects in a PKC-h
inhibitor, selectivity was the primary driving factor
in the initial triaging of the hit set. PKC-a and
PKA counterscreens were performed on all hits with
minimum of 10· selectivity required. Two classes, dia-
minopyrimidine (1) and indolinones (2), emerged as
the focus based on potency, selectivity in the count-
erscreens, and available SAR.
The pyrimidines were synthesized as shown in Scheme 1.
The 2,4-dichloro-5-nitropyrimidine was treated with
potassium thiocynanate in AcOH which reacts selective-
ly at the 4-position and allows for clean differentiation
of the 2- versus 4-positions.7 Displacement of 2-chloro
with arylalkylamines proceeded smoothly followed by
displacement of the thiocyanate with an appropriate
amine gave the desired compounds.8 An excess of diam-
ines was necessary to avoid diaddition or alternatively
diamines could be mono-protected and used in an addi-
tion-deprotection sequence.5
A representative set of these compounds from exisiting
analogs covering the major SAR elements of the 2 series
and spanning a potency range from low nM to 1 lM
were tested against IR, LYN, SYK, and VEGFR1 to
determine if either series offered SAR that indicated
Initial homology modeling of PKC-h based on CDK2
followed by docking of 3 suggested that the pyrimi-
dine binds to the hinge region at the ATP binding site
via a 2-point binding of the 2-amino substituent and
nitrogen of the pyrimidine at the 1 position with
Leu461. Another prediction was the key interaction
of the distal amine of the 4-substituent with Asp508.
The model suggested that modifications of the 2-sub-
stituent may lead to additional hydrogen bond inter-
actions with Tyr460 as well as affect selectivity by
steric interactions with the region near Leu386 proxi-
mal to the glycine-rich loop and the specificity surface
near Tyr468 (Fig. 3).9
Figure 1. Two structural classes of interest from primary uHTS screen.
0.005 μM
O
N+
O
N
N
N
NH
H
Cl
3
NH2
0.5 μM
Green tabs =
Pyrimidines
No tabs =
Indolinones
1 μM
100x
0x
10x
Figure 2. Heat map of selectivity.

C. L. Cywin et al. / Bioorg. Med. Chem. Lett. 17 (2007) 225–230
Table 1. Preference for the 5-nitro series
227
Compound
R2
R5
PKC-h Inhibition
IC50 (lM)
4
5
6
7
3,5-ClPh
3,5-ClPh
3-ClPh
NO2
CF3
NO2
CF3
0.68
4.20
0.06
0.64
3-ClPh
Scheme 1. General synthetic method.
Table 2. 2-Position SAR
Steric Hindrance
LEU386
H-bond
TYR460
Compound
R
PKC-h Inhibition
IC50 (lM)
LEU461
8
3
H
o-Cl
0.50
0.046
0.085
0.17
TYR468
9
m-Cl
p-Cl
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
ASP508
o-F
m-F
0.048
0.083
0.21
p-F
o-OCH3
m- OCH3
p- OCH3
o-CH3
m- CH3
o-CF3
0.017
0.21
12
0.034
0.039
0.012
0.018
0.015
0.006
0.005
0.005
0.039
0.011
0.022
0.10
Figure 3. Homology model of PKC-h kinase domain bound with 3.
o-OCF3
o-CHF2
o-SCF3
o-SCH3
o-Br
The pyrimidine hits from the uHTS campaign had either
5-nitro or 5-trifluoromethyl substituents. The ꢀ10-fold
increase in potency in the nitro series (Table 1) led us
to focus on the 5-nitro series for our initial SAR surveys
at the 2- and 4-positions. The evaluation of the data in
Figure 2 showed a trend for the 2-benzylamino deriva-
tives, and especially 3, to be more selective than the
2-anilino derivatives for PKC-h however the 2-benzyla-
mino series was comparatively under represented.
o-Ph
o-NO2
2,3-diF
2,4-diF
2,5-diF
2,6-diF
2,3-diCl
2,4-diCl
2,5-diCl
2-CH3-3-Cl
2,3-diCH3
2,5-diCH3
3,5-diCl
3,5-diOCH3
0.035
0.042
0.008
0.027
0.007
0.046
0.013
0.91
The data from the initial hits, as well as the homology
model, suggested that the 2-position would render
exploitable SAR. Table 2 shows the results of our early
exploration of the 2-position.10 There was a preference
for ortho over meta substitution but both were beneficial
compared to the unsubstituted benzyl (3, 8–18). Substi-
tution at the para position led to neutral or detrimental
effects as the size of the substituent was increased, but
also led to decreased selectivity (vide infra). The varia-
tions in potency based on the ortho-substituent were
affected less by electronics than by sterics. Compare
compounds 3, 11, 14, and 17 where electronic nature
0.40
1.5
of the substituent was varied with compounds 20, 22,
23, and 24 where the steric demands are more dominant.
Disubstitution was permitted and appeared to be
additive with dichloro substituents in the 2,3- and

228
C. L. Cywin et al. / Bioorg. Med. Chem. Lett. 17 (2007) 225–230
Table 3. Linker variations
Table 4. 4-Position SAR
Compound R4
PKC-h
Inhibition
IC50 (lM)
Compound
R2
PKC-h Inhibition
IC50 (lM)
53
54
55
56
57
58
59
60
61
62
63
64
Cyclohexylmethyl
>10
39
40
41
42
43
44
45
46
47
48
49
50
51
52
2-Br-Bz
2.2
2.5
cis-4-(H2NCH2)-Cyclohexylmethyl
trans-4-(H2NCH2)-Cyclohexylmethyl
4-( H2NCH2)-Phenylmethyl
4-((CH3)2NCH2)-Cyclohexylmethyl
4-(AcNHCH2)-Cyclohexylmethyl
3-(H2NCH2)-Cyclohexylmethyl
3-( H2NCH2)-Phenylmethyl
Piperdin-4-ylmethyl
0.018
0.029
0.16
PhC(CH3)2–
(S)-PhCH(CH3)–
(R)-PhCH(CH3)–
(S)-Indan-1-yl
(R)-Indan-1-yl
Indan-2-yl
3.7
5.7
0.054
>10
4.4
0.46
0.031
0.094
0.27
0.090
0.18
0.13
0.054
0.077
0.003
0.20
Ph(CH2)2–
Ph(CH2)3–
Ph(CH2)4–
0.090
>10
0.45
H2N(CH2)3–
H2N(CH2)5–
H2N(CH2)7–
2-CH3-Ph(CH2)2–
2-F-Ph(CH2)2–
3-F-Ph(CH2)2–
4-F-Ph(CH2)2–
0.11
substituent (54 and 55). The basicity of the amine was
important as dimethyl analog 57 was tolerated. Howev-
er, acylation resulted in the inactive analog 58.
2,5-disubstitution patterns (31 and 33) as contrasted
with the difluoro analogs 27 and 29. However there were
size constraints and the dimethyl analog 36 showed de-
creased activity relative to the parent 8. Comparatively,
3,5-disubstitution led to weaker analogs (37 and 38).
Table 3 highlights SAR studies of the linker at the 2-po-
sition. Amide analog 39 displayed significantly reduced
potency compared to 24. Mono- or dimethylation of
the methylene of the benzyl group was also detrimental
to potency (40–42). Interestingly, 44 showed that the
detrimental effects could be partially alleviated by ring
constraining the molecule. This may be accounted for
by the substitution on the aryl ring, increased linker
length or restricted rotation leading to preferential ori-
entation of the substituent. Orientation appears to be
at least partially contributing as the optical antipode
43 had no increase in activity relative to 41. The linker
length did not appear to be a strong contributor to
potency as the phenethyl, phenpropyl, and phenbutyl
analogs all had similar values (46–48). The ring con-
strained analog 45 had a significant boost in potency
reinforcing a preferred orientation versus the effect of
substitution on the aryl moiety (compare to 49). The
ortho substitution on the aryl may be less preferred than
meta in the phenethyl series (50–52).
Cpd
59
Cpd
23
22
19
24
26
35
14
21
54
27
55
17
25
30
34
57
9
32
45
29
18
3
11
12
48
28
46
64
56
15
10
13
44
63
47
37
36
From the model as well as SAR of the uHTS hits we
expected the diamine at the 4-position would be impor-
tant for activity. The role of the terminal amine in 3 was
confirmed by deleting it as in 53 which resulted in loss of
activity (Table 4). The spacing of the diamine was criti-
cal as was seen with analogs 62–64. Restricted rotation
by incorporation of a ring improved potency as seen
with 3 and 59. However, the more rigid and planar
phenyl analogs 56 and 60 were significantly less potent
than the saturated versions. There was no stereochem-
ical preference for the 1,4-cyclohexylbis(methylamine)
100x
Figure 4. Heat map of selectivity ratio for representative analogs.
0x
10x

C. L. Cywin et al. / Bioorg. Med. Chem. Lett. 17 (2007) 225–230
Table 5. Cellular and early in vitro ADME profiling
229
Compound
IL-2 Inhibitiona IC50 (lM)
CYP Inhibition IC50 (lM)
HLM, T1/2 (min)
Caco-2 AB/BA (10À6 cm/s)
2C9
2D6
3A4 (BFC/BQ)
3
23
57
61
0.084
0.14
0.19
0.72
1.0
1.4
4.3
4.7
1.7
>30
0.7/1.4
6.9/>30
1.7/13
48
65
40
66
4.0/6.9
1.6/1.9
8.4
17.6
10.6/10.2
8.1/10.6
28/>30
a Inhibition of IL-2 production in human CD4+ T cells stimulated with anti-CD3anti-/CD28 mAb.
To address the issue of selectivity we tested analogs in
a panel of kinases and found them to be highly selec-
Acknowledgments
tive. Figure 4 shows selectivity data for the 38 repre-
sentative analogs, with IC50 values for PKC-h
<0.5 lM, against a panel of 13 kinases representing
data from 494 individual dose responses. Potencies in-
crease upward from the bottom of the figure. Some
trends could be visualized even amongst this highly
selective set. The analogs where the linker was extend-
ed tended to show reduced selectivity (46–48) as did the
o-Ph analog 25. The observed high selectivity of these
analogs can be rationalized by the tight requirement
for a properly positioned amino group to interact with
Asp508 coupled with the 2-benzylamino substituent
which controls specificity through its interactions at
the glycine-rich loop, the specificity surface, and the
hinge regions.
The authors acknowledge George R. Rogers and
Cynthia Sledziona for their technical assistance in gener-
ating the uHTS data.
References and notes
1. Baier, G.; Telford, D.; Giampa, L.; Coggeshall, K. M.;
Baier-Bitterlich, G.; Isakov, N.; Altman, A. J. Biol. Chem.
1993, 268, 4997.
2. Sun, Z.; Arendt, C. W.; Ellmeier, W.; Schaeffer, E. M.;
Sunshine, M. J.; Gandhi, L.; Annes, J.; Petrilka, D.;
Kupfer, A.; Schwartzberg, P. L.; Littman, D. R. Nature
2000, 404, 402.
3. (a) Monks, C. R. F.; Freiberg, B. A.; Kupfer, H.; Sciaky,
N.; Kupfer, A. Nature 1998, 395, 82; (b) Monks, C. R. F.;
Kupfer, H.; Tamir, I.; Barlow, A.; Kupfer, A. Nature
1997, 385, 83.
In addition to selectivity, other hit to lead criteria
included activity in cells as well as acceptable drug-like
properties. The compounds were also shown to be
ATP competitive.11 Cellular activity was assessed by
measuring inhibition of IL-2 production in human
CD4+ T cells activated by costimulation with anti-
CD3 and anti-CD28 mAbs. We also assessed represen-
tative compounds for inhibition of CYP’s, stability to
human liver microsomes (HLM), and permeability in
Caco-2 cells (Table 5).12,13 Although these compounds
contain a nitro group which is potential structural alert,
the overall profile was acceptable for further advance-
ment in the hit to lead process.
4. Heat maps were prepared using Spotfireꢁ Decision Site
8.2ꢁ from Spotfire, Inc., 212 Elm St., Somerville, MA,
02144 USA.
5. (a) For recent patents on pyrimidines as kinase inhibitors:
Dahmann, G. US Patent Application US 2003/171359;
Chem. Abstr. 2003, 319721; (b) Cardozo, M. G. U.S.
Patent Applications U.S. 2005/124640 and U.S. 2004/
242613; Chem. Abstr. 2005, 497492; (c) Baudler, M. U.S.
Patent Application U.S. 2005/0222186; Chem. Abstr. 2003,
1006977.
6. For recent patent on indolinones as kinase inhibitors:
Roth, G. J. U.S. Patent Application U.S. 2005/043389;
Chem. Abstr. 2004, 80649.
7. Takahashi, T.; Naito, T.; Inoue, S. Chem. Pharm. Bull.
1958, 6, 334.
It has been shown that suitably substituted 5-nitro-2,4-
diaminopyrimidines are highly potent and selective
inhibitors of PKC-h. The potencies ranged from micro-
molar to low nanomolar. An arylalkyl moiety is pre-
ferred for selectivity at the 2-position but there is a
high degree of flexibility in the substitution on the aryl
ring. Sterics at the 2-position are also limiting for retain-
ing potency and selectivity. There is a requirement for a
suitably spaced basic diamine at the 4-position. Substi-
tution of the terminal amine is permitted provided
basicity remains intact. The SAR presented shows that
selectivity can be achieved in this known class of kinase
inhibitors. We demonstrated that these compounds pos-
sess acceptable in vitro metabolic profiles. These com-
pounds were of further interest and additional studies
to gain better understanding of the binding mode and
the observed selectivity, to expand SAR including 5-ni-
tro replacements, as well as advancing the overall pro-
file of these compounds will be the subject of future
reports.
8. Preparation of 24 is representative: compound I (12 g
62 mmol) was dissolved in AcOH and cooled to 0 ꢂC.
KSCN (6.3 g, 65 mmol) was added portionwise and the
mixture was stirred for 2 h. The mixture was diluted with
H2O and filtered. The solid was washed with Et2O to yield
II (9.5 g 71%). Compound II (100 mg, 0.46 mmol) was
suspended in 1 mL abs EtOH. o-BromobenzylamineÆHCl
(103 mg, 0.46 mmol) was dissolved in 1 mL abs EtOH and
added to the reaction mixture quickly at rt. Then Et3N
(129 lL, 0.92 mmol) was added dropwise. The reaction
mixture was stirred 16 h. The resulting precipitate was
filtered providing 136 mg of the desired (2-bromo-benzyl)-
(5-nitro-4-thiocyanato-pyrimidin-2-yl)-amine as
a white
solid. 1,4-Cyclohexanebis(methylamine) (65 mg, 0.46 mmol)
was dissolved in 1 mL CH2Cl2 and added to a suspension of
(2-bromo-benzyl)-(5-nitro-4-thiocyanato-pyrimidin-2-yl)-
amine (40 mg, 0.11 mmol) in 1 mL CH2Cl2. The reaction
was shaken 16 h at rt. The reaction mixture was then
loaded directly onto a 4.5 g pre-packed disposable flash
cartridge and eluted with a gradient solvent system
(CH2Cl2/10% NH4OH in MeOH). Compound 24,

230
C. L. Cywin et al. / Bioorg. Med. Chem. Lett. 17 (2007) 225–230
43 mg, 89% was isolated as
MS(M+H) = 449 mp 96–100.
a
yellow solid;
are added to reaction mixture and incubated for 90 min at
rt. FP is measured on a LJL Analyst using fluorescein
filters (485 nm excitation, 530 nm emission, 505 nm
dichroic).
9. Cherry, M.; Williams, D. H. Curr. Med. Chem. 2004, 11, 663.
10. Inhibition of enzymatic activity of PKC-h was measured in
a two-step competition fluorescence polarization (FP)
assay utilizing Panvera’s protein kinase assay kit (PanVera
P2940). In the kinase reaction step, 50 nM PKC-h
(Panvera P2996) diluted in assay buffer (20 mM Hepes,
pH 7.6 , 0.1 mM CaCl2, 10 mM MgCl2, 0.01% Chaps,
200 lM TCEP (Pierce # 77720), 100 lM Na orthovana-
date, and Protease Inhibitor Cocktail (Boehringer Mann-
heim 1836153; 1 tablet to 50 mL buffer)) is pre-incubated
with compound dilutions at rt for 10 min. The reaction is
11. Data not shown; Compound 3 IC50 = 0.29 lM at 50 lM
ATP and IC50 = 0.63 lM at 100 lM ATP.
12. Inhibition of IL-2 production was determined by isolating
human CD4+ T cells from whole blood by positive
selection. Purified T cells are activated through the TCR
and CD28 via anti-CD3 and anti-CD28 mAbs. Com-
pounds are diluted in 5% DMSO to a final concentration
of 0.125% DMSO at every dose. 50,000 cells are added in
100 lL of media/well followed by 100 lL of compound or
DMSO alone. Cells are incubated overnight at 37 ꢂC and
then the supernatants are analyzed for IL-2 with the R&D
Systems IL-2 ELISA kit (Cat#02050) following a 1:10
dilution.
started with
a
mixture of peptide substrate
(RFARKGSLRQKNV, Panvera P2760) and ATP (final
assay concentrations are 1 lM peptide and 10 lM ATP).
The assay plates are incubated for 60 min at rt. In the FP
detection step, fluorescein-labeled phosphopeptide tracer
and anti-phosphoserine antibody diluted in quench buffer
13. Kaplita, P. V.; Hu, H.; Liu, L.; Farrell, T. M.; Grbic, H.;
Spero, D. M. JALA 2005, 10, 140.