Identification and optimization of indolo[2,3-c]quinoline inhibitors of IRAK4

Article abstract of DOI:10.1016/j.bmcl.2014.03.056

IRAK4 is responsible for initiating signaling from Toll-like receptors (TLRs) and members of the IL-1/18 receptor family. Kinase-inactive knock-ins and targeted deletions of IRAK4 in mice cause reductions in TLR induced pro-inflammatory cytokines and these mice are resistant to various models of arthritis. Herein we report the identification and optimization of a series of potent IRAK4 inhibitors. Representative examples from this series showed excellent selectivity over a panel of kinases, including the kinases known to play a role in TLR-mediated signaling. The compounds exhibited low nM potency in LPS- and R848-induced cytokine assays indicating that they are blocking the TLR signaling pathway. A key compound (26) from this series was profiled in more detail and found to have an excellent pharmaceutical profile as measured by predictive assays such as microsomal stability, TPSA, solubility, and c log P. However, this compound was found to afford poor exposure in mouse upon IP or IV administration. We found that removal of the ionizable solubilizing group (32) led to increased exposure, presumably due to increased permeability. Compounds 26 and 32, when dosed to plasma levels corresponding to ex vivo whole blood potency, were shown to inhibit LPS-induced TNFα in an in vivo murine model. To our knowledge, this is the first published in vivo demonstration that inhibition of the IRAK4 pathway by a small molecule can recapitulate the phenotype of IRAK4 knockout mice.

Full text of DOI:10.1016/j.bmcl.2014.03.056

Bioorganic & Medicinal Chemistry Letters 24 (2014) 2066–2072
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Identification and optimization of indolo[2,3-c]quinoline inhibitors
of IRAK4
a
a
a
b
L. Nathan Tumey a, , Diane H. Boschelli , Niala Bhagirath , Jaechul Shim , Elizabeth A. Murphy ,
Deborah Goodwin b, Eric M. Bennett c, Mengmeng Wang d, Lih-Ling Lin b, Barry Press a, Marina Shen b,
Richard K. Frisbie a, Paul Morgan b, Shashi Mohan b, Julia Shin b, Vikram R. Rao b
⇑
a Pfizer Global R&D, 445 Eastern Point Rd., Groton, CT 06340, USA
b Pfizer Global R&D, 200 Cambridge Park Dr., Cambridge, MA 02140, USA
c Pfizer Global R&D, 87 Cambridgepark Dr., Cambridge, MA 02140, USA
d Pfizer Global R&D, 1 Burtt Rd., Andover, MA 01810, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
IRAK4 is responsible for initiating signaling from Toll-like receptors (TLRs) and members of the IL-1/18
receptor family. Kinase-inactive knock-ins and targeted deletions of IRAK4 in mice cause reductions in
TLR induced pro-inflammatory cytokines and these mice are resistant to various models of arthritis.
Herein we report the identification and optimization of a series of potent IRAK4 inhibitors. Representative
examples from this series showed excellent selectivity over a panel of kinases, including the kinases
known to play a role in TLR-mediated signaling. The compounds exhibited low nM potency in LPS-
and R848-induced cytokine assays indicating that they are blocking the TLR signaling pathway. A key
compound (26) from this series was profiled in more detail and found to have an excellent pharmaceu-
tical profile as measured by predictive assays such as microsomal stability, TPSA, solubility, and clogP.
However, this compound was found to afford poor exposure in mouse upon IP or IV administration.
We found that removal of the ionizable solubilizing group (32) led to increased exposure, presumably
due to increased permeability. Compounds 26 and 32, when dosed to plasma levels corresponding to
Received 10 February 2014
Revised 17 March 2014
Accepted 18 March 2014
Available online 29 March 2014
Keywords:
TLR signaling
Indoloquinoline
IRAK4
Kinase inhibitor
Inflammation
ex vivo whole blood potency, were shown to inhibit LPS-induced TNF
a in an in vivo murine model. To
our knowledge, this is the first published in vivo demonstration that inhibition of the IRAK4 pathway
by a small molecule can recapitulate the phenotype of IRAK4 knockout mice.
Ó 2014 Elsevier Ltd. All rights reserved.
In the past decade there has been considerable interest in tar-
geting the innate immune system in the treatment of autoimmune
diseases and sterile inflammation. Receptors of the innate immune
system provide the first line of defense against bacterial and viral
insults. These receptors recognize bacterial and viral products as
well as pro-inflammatory cytokines and thereby initiate a signaling
cascade that ultimately results in the up-regulation of inflamma-
regulation of innate immunity.4 IRAK4 is responsible for initiating
signaling from TLRs and members of the IL-1/18 receptor family.
Kinase-inactive knock-ins and targeted deletions of IRAK4 in mice
lead to reductions in TLR and IL-1 induced pro-inflammatory cyto-
kines.5–7 IRAK-4 kinase-dead knock-in mice have been shown to be
resistant to induced joint inflammation in the antigen-induced-
arthritis (AIA) and serum transfer-induced (K/BxN) arthritis mod-
els.8 Likewise, humans deficient in IRAK4 also display the inability
to respond to challenge by TLR ligands and IL-1.9 However, the
immunodeficient phenotype of IRAK4-null individuals is narrowly
restricted to challenge by gram positive bacteria, but not gram neg-
ative bacteria, viruses or fungi. This gram positive sensitivity also
lessens with age implying redundant or compensatory mecha-
nisms for innate immunity in the absence of IRAK4.10 These data
suggest that inhibitors of IRAK4 kinase activity will have therapeutic
value in treating cytokine driven autoimmune diseases while having
minimal immunosuppressive side effects. Additional recent studies
suggest that targeting IRAK4 may be a viable strategy for the treat-
ment of other inflammatory pathologies such as atherosclerosis.11
tory cytokines such as TNFa, IL6, and interferons. Recently it has
become apparent that self-generated ligands such as nucleic acids
and products of inflammation such as HMGB1 and Advanced
Glycated End-products (AGE) are ligands for Toll-like receptors
(TLRs) which are key receptors of the innate immune system.1–3
This demonstrates the role of TLRs in the initiation and perpetua-
tion of inflammation due to autoimmunity.
Interleukin-1 receptor associated kinase (IRAK4) is a ubiqui-
tously expressed serine/threonine kinase involved in the
⇑
Corresponding author. Tel.: +1 860 715 6440.
E-mail address: Nathan.tumey@pfizer.com (L.N. Tumey).
http://dx.doi.org/10.1016/j.bmcl.2014.03.056
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

L. N. Tumey et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2066–2072
2067
Indeed, the therapeutic potential of IRAK4 inhibitors has been
recognized by others within the drug-discovery community as evi-
denced by the variety of IRAK4 inhibitors have been reported to-
date.12–16 However, limited data has been published about these
compounds and they appear to suffer from a variety of issues such
as poor kinase selectivity and poor whole-blood potency that pre-
clude their advancement into the pre-clinical models. To the best
of our knowledge, no in vivo studies of IRAK4 inhibitors have been
reported to-date in the literature. Herein we report a new class of
IRAK4 inhibitors that are shown to recapitulate the phenotype ob-
served in IRAK4 knockout and kinase-dead mice.
An aggressive high-throughput screening approach was under-
taken in order to identify lead compounds that could be advanced
quickly into a medicinal chemistry program. A Caliper high
throughput screen was run using the kinase domain of IRAK4
and an in-house optimized peptide substrate. Approximately
350,000 compounds were evaluated resulting in 1935 hits that
were reconfirmed in a 5-point dose–response assay and further
evaluated using a DELFIA-based assay. The hits were prioritized
based on factors such as potency, kinase selectivity, and synthetic
tractability. These efforts led to the identification of multiple lead
series that will be described in due course. One lead series that
was identified contained a 3-substituted-4-aminoquinoline, as
illustrated by compound 3.
The early SAR around this series was quite erratic and difficult
to interpret. For example, compound 3 initially was thought to
be a 95 nM inhibitor of IRAK4. However, resynthesis of 3 resulted
in a batch that was found to be nearly 1000-fold less potent. After
assay variability and compound stability issues were ruled out, a
closer look at the analytical data for compound 3 revealed a trace
amount (ꢀ3% by UV) of a contaminant with a molecular mass con-
sistent with indolo[2,3-c]quinoline 4 and/or 5. This compound pre-
sumably arises from an intramolecular direct arylation, a reaction
with significant precedent in literature.17 We observed that the
less active batch of 3 contained far smaller amounts of the impu-
rity. Together these results strongly implied that the activity of
compound 3 may be coming from the trace indolo[2,3-c]quinoline
4 or 5.
A more robust synthesis of compound 4/5 was required in order
to test the above hypothesis. The initial synthesis was accom-
plished by treating compound 2 under standard Suzuki conditions,
but in the absence of a boronic acid coupling partner and resulted
in a mixture of the regiomers 4 and 5 along with significant
amounts of des-iodo 2. The 9-cyano-indolo[2,3-c]quinoline 4 was
purified from the mixture by reverse-phase HPLC and found to
be a 7.4 nM inhibitor of IRAK4. The corresponding 7-isomer (5)
was found to be at least 15-fold less active, although the exact
IC50 value is questionable due to the inability to completely purify
it from residual amounts 4. The exceptional potency of compound
4 confirmed our suspicions that the observed activity of the
3-cyclopropylquinoline (3) was likely due to trace contamination
of the sample with the isomeric cyano-indolo[2,3-c]quinolines 4/5.
In order to plan and prioritize SAR studies of compound 5, it was
necessary to have some understanding of the binding mode and
key interactions that this compound might make with IRAK4.
Therefore, a docking study of this compound was performed using
a 100,000 step MCMM/LMCS Monte Carlo procedure in MacroMod-
el 9.6 (MMFFs and GB/SA energetics) with the published IRAK4
crystal structure complex with staurosporine.18 The procedure
found 20 binding modes within 10 kcal/mol of the global mini-
mum, which was consistent with the SAR. Higher energy modes
either did not form hydrogen bonds with the hinge or exhibited de-
fects such as poor shape complementarity near the gatekeeper or
inconsistencies with accumulated SAR. Several of the 20 modes dif-
fered from the global minimum only by minor changes to side
chain rotamers for residues (197, 200, 213, 229, 233, 262, 264,
313, 329) which were unrestrained along with the inhibitor during
the energy minimization. The global minimum is illustrated in
Figure 1. The model strongly suggests that the quinoline nitrogen
makes a key interaction with the ‘hinge’ methionine (Met265), in
a similar manner to a variety of related quinoline and quinazoline
kinase inhibitors. The cyanophenyl ‘headpiece’ reaches deep into
the binding pocket, with the nitrile forming a hydrogen-bond with
the catalytic lysine (Lys213). The 7 and 8 position of the indolo[2,3-
c]quinoline rest snuggly against the unusually large gatekeeper
residue, Tyr262. This large gatekeeper residue is unique to the IRAK
family of kinases, of which only IRAK1 and IRAK4 are kinase-active
proteins. Finally, the dimethoxy moiety points out toward a sol-
vent exposed region.
With the above binding mode in mind, a series of analogs were
made to probe the SAR around the cyanophenyl ‘headpiece’ and
the quinoline core. Generally, mixtures of 7 and 9-substituted
indolo[2,3-c]quinolines were made by taking advantage of the oxi-
dative cyclization described above in Scheme 1. However, the yield
of this synthesis was low and the separation of the two isomers
proved to be quite challenging. Scheme 2 illustrates a regiospecific
synthesis of 9-substituted indolo[2,3-c]quinolines that relies on a
2-aminoboronic acid that participates in a tandem Suzuki and
intramolecular SNAr reaction. Using these two routes, a variety of
analogs (6–15) were made in order to probe the hypothesized
binding mode. As predicted, the observed SAR around the tetracy-
clic core is consistent with the docking studies outlined above
(Table 1). Movement of the cyano functionality to the 7, 8 and 10
position of the indolo[2,3-c]quinoline resulted in a precipitous loss
of activity (compounds 5, 6, and 7, respectively). This is consistent
with the 7 and 8 position of molecule sitting in very close proxim-
ity to the tyrosine gatekeeper. Likewise, the addition of a methyl
group to the quinoline core (8) resulted in a dramatic loss in activ-
ity, likely disrupting the hinge-binding interaction with Met265
due to unfavorable steric contacts between the methyl group and
the backbone of residues 263–264. The incorporation of a methyl
group at the analogous position in related quinazoline and quino-
line-based kinase inhibitors also results in dramatic loss of activ-
ity.19,20 A variety of functionality was incorporated in order to
replace the cyano functionality on the headpiece. (9–15) Elimina-
tion of the cyano group (15) resulted in a 100-fold loss in activity
while replacement with a nitro (13) or carboxamide (12) retained
significant potency. A variety of attempts to open up the tetracyclic
ring system resulted in loss of activity (16–19).
Figure 1. The proposed binding mode of compound 4 with IRAK4. The surface of
the indoloquinoline is represented in grey while the protein surface is represented
by the blue grid.

2068
L. N. Tumey et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2066–2072
CN
CN
9
CN
Cl
N
HN
N
NH
N
b or c
NH
N
a
O
O
I
O
O
O
O
I
O
O
7
1
4 (9-isomer)
5 (7-isomer)
3
2
(major product)
Scheme 1. Reagents and conditions: (a) 3-Amino-benzonitrile, EtOH, reflux; (b) cyclopropyl boronic acid, Pd(PPh3)2Cl2, Cs2CO3, THF/H2O, 85 °C; (c) Pd(PPh3)2Cl2, Cs2CO3, THF/
H2O, 85 °C.
X
H2N
X
HN
Cl
N
(HO)2B
MeO
MeO
MeO
MeO
I
a
N
4
9 15
,
-
1
CN
CN
H2N
X
Cl
N
HN
N
MeO
F
I
HN
(HO)2B
MeO
F
MeO
b
a
R
O
N
20
21
Scheme 2. Reagents and conditions: (a) Pd(PPh3)4, Cs2CO3, 3:1 DME/H2O, 100 °C,16 h; (b) ROH, NaH (4 equiv), DMF, 110 °C, 16 h.
Table 1
SAR around the cyanophenyl ‘headpiece’
CN
9
10
R1
8
CN
CN
H2N
N
HN
N
HN
N
O
O
O
O
N
7
O
O
R2
N
N
Cl
17
18
16
(3-CN)
19 (4-CN)
4-15
Compd#
R1
R2
IRAK4 IC50 (nM)
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
9-CN
7-CN
8-CN
10-CN
9/7-CN
9-CF3
9-OMe
9-OCF3
9-CONH2
9-NO2
9-NHAc
H
H
H
H
H
Me
H
H
H
H
H
H
H
7.4
130#
30,000
5100
14,000§
92
615
2480
26
5.3
23,600
630
19,000
>100,000
>100,000
>100,000
NA
NA
NA
NA
#
Compound 5 was contaminated with ꢀ5% of compound 4. It is believed (though unconfirmed) that the observed activity of compound 5 is due entirely to this
contamination.
§
Compound 8 was made as an inseparable mixture of the 9 and 7 isomers.
Despite its excellent potency in biochemical assays, it was quite
prepared that incorporated amine and ether containing tailpieces.
(22–32) The synthesis of these analogs is outlined in Scheme 2 and
generally relied on an SNAr reaction of an alkoxide with 3-fluoro-
indolo[2,3-c]quinoline 21. To our delight, the addition of solubiliz-
ing tailpieces generally resulted in a further boost in potency,
yielding sub-nM inhibitors of IRAK4 (Table 2). The 3-carbon tail
generally provided additional potency as compared to the 2-carbon
tails, as observed by comparing 22 and 24 to their longer-chain
counterparts, 23 and 26. All compounds with the amine-containing
tails (22–29) were observed to have excellent solubility (generally
apparent that compound 4 would not be suitable for in vivo studies
due largely to its exceptionally poor solubility. Attempts to im-
prove the solubility by breaking up the tetracyclic core failed to
give compounds with meaningful inhibition of IRAK4 (16–19,
above). Therefore, we next turned our attention to the addition
of polar tailpieces that would point into the solvent exposed area
of the binding pocket. Modeling studies (above) suggested that
the 3-methoxy group is oriented in a favorable direction to reach
into bulk solvent. With this in mind, a series of analogs was

L. N. Tumey et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2066–2072
2069
Table 2
SAR around the solubilizing tailpiece
CN
HN
O
N
Compd#
Structure
MeO–
IRAK4 IC50 (nM)
7.4
IRAK1 IC50 (nM)
NA
hPBMC R848 induced TNF
NA
a
(nM)
hWB R848-induced IL6 (nM)
NA
Solubility (pH 7.4)
0 lg/mL
4
N
N
22
23
2.6
120
130
41
24
2700
4300
>100
>100
l
l
g/mL
g/mL
O
O
0.29
N
24
25
26
27
28
29
0.75
2.6
170
120
65
9.8
850
>100
l
g/mL
N
O
HN
N
134
85
14,000
950
69 lg/mL
O
N
O
0.094
0.10
6.2
>100
l
g/mL
N
O
O
68
44
1400
2700
>10,000
68
68
71
l
l
l
g/mL
g/mL
g/mL
N
N
310
310
890
1800
HN
6.3
O
N
O
29
30
0.11
0.46
86
200
36
960
>100
O
O
210
3600
0
lg/mL
O
31
32
0.55
1.3
260
290
4700
8300
0
0
l
g/mL
g/mL
O
O
18
l
O
O
>70
l
g/mL) while those with ether-containing tails (30–32) were
results from functional assays and efficacy studies. One persistent
issue in the development of IRAK4 inhibitors is the importance of
the contribution of activity of IRAK1 to the cellular potency of
the compounds. Several studies have postulated that inhibition
of IRAK4 and IRAK1 may be required for the inhibition of proin-
flammatory cytokines.22 Therefore we strove for selectivity over
the related isoform, IRAK1, in order to determine whether dual
inhibition of both isoforms would be necessary for blockage of
the TLR pathway (Table 2). Encouragingly, most compounds
showed >100-fold selectivity over this kinase. Both IRAK1 and
IRAK4 inhibition were evaluated in the presence of ATP at their
found to have poor solubility.
Having now established excellent enzymatic potency against
IRAK4, we next turned our attention to the functional activity of this
novel class of inhibitor. Compounds of interest were screened for
their ability to inhibit R848 induced TNF in primary human mono-
cytes (hPBMCs).21 R848 activation of TLR7 and TLR8 on monocytes
results in the production of pro-inflammatory cytokines such as
IL-6 and TNFa. The compounds shown in Table 2 were screened in
this cellular assay and found to potently inhibit this activity.
Compounds were generally 10 to 100-fold less potent in the cellular
assay as compared to the enzymatic assay, possibly indicating a
permeability or efflux issue (vide infra). Compounds containing
secondary amines (e.g., 25) were found to be very weakly active in
the cellular assay, presumably due to poor permeability.
Additionally, compounds were screened for their ability to
block R848-induced release of IL6 in human whole blood (hWB).
As can be seen in Table 2, the most potent compounds in the
hPBMC assay were found to have low micromolar activity in hu-
man whole blood. The discrepancy between the cellular and
whole-blood potencies is most likely attributable to plasma pro-
tein binding, which is >99% for two compounds tested (see Table 4
and related discussion).
Km (35 lM and 600 l
M, respectively).31 The tighter binding of
IRAK1 to ATP suggests that the high levels of cellular ATP
(1–2 mM) will limit the functional potency of these inhibitors
against IRAK1 even more than the IC50 values in Table 2 suggest.
Therefore, it can be inferred that most of the cellular and hWB
activity described in Table 2 is derived from the inhibition of
IRAK4.
Compounds of key interest were additionally evaluated against
a panel of kinases selected based on sequence diversity and safety
concerns. The summarized data is shown in Table 3. All kinases
(including IRAK4) were tested with the ATP concentration at the
Km. The desired selectivity profile for drug discovery programs var-
ies greatly depending upon the indication being pursued and the
stage of the program. While there is great debate in the field about
the desirable selectivity profile for oncology applications,23 it is
relatively well-agreed that pursuit of a non-oncology indication
requires very stringent criteria for compound advancement. Our
A key hurdle in all kinase drug-discovery programs is the estab-
lishment of appropriate levels of selectivity over off-target
enzymes that may lead to toxicity. Additionally, most signaling
cascades consist of numerous kinases and the inhibition of
multiple pathway kinases can complicate the interpretation of

2070
L. N. Tumey et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2066–2072
Table 3
Table 4
Kinase selectivity for selected IRAK4 inhibitors (shown as fold-selectivity compared to
the IRAK4 IC50 in Table 2)
Pharmaceutical profile of two advanced leads
CN
HN
N
O
O
O
N
O
O
26
32
N
hIRAK4 IC50
hPBMC R848 induced IL6 IC50
THP cell LPS-induced TNF IC50
hWB R848 induced IL6 IC50
hWB LPS induced TNFa IC50
MW
0.094 nM
85 nM
56 nM
950 nM
5500 nM
430
1.3 nM
18 nM
29 nM
8300 nM
15,000 nM
390
a
clogP
TPSA
3.2
3.5
77 Å2
89 Å2
Microsomal stability (t1/2
)
Rat
Mouse
Human
>30 min
>30 min
>30 min
23 min
10 min
20 min
Cyp,% inhibition @3 lM
2C9
2D6
3A4
11%
0%
15%
0.8%
11%
8%
À7%
Human plasma binding (% free)
MDCK permeability
A to B (cm/s)
0.9%
0.10 Â 10À6
0.6 Â 10À6
Efflux ratio
78
71
10 mpk mouse PK (IP)
AUC (h ng/mL)
540
0.75
6500
5.5
Cmax
(
lM)
100 mpk mouse PK (IP)
AUC (h ng/mL)
Cmax (lM)
10,000
14
NA
NA
2 mpk mouse PK (IV)
Cl (mL/min kg)
Vd (L/kg)
AUC (h ng/mL)
10 mpk mouse PK (PO)
AUC (h ng/mL)
%F
37
34
900
29
1.2
580
80
2%
650
11%
primary goal at this stage of the program, however, was to achieve
proof-of-mechanism in vivo in order to justify continued invest-
ment. With this in mind, the selectivity for most of the compounds
shown in Table 2 was thought to be adequate. Indeed, the selectiv-
ity ratios for the vast majority of the kinases tested were >>100-
fold. Most compounds showed poor selectivity against only three
Compounds of interest were routinely screened for microsome
stability and CYP inhibition. Indolo[2,3-c]quinolines with amine
tails (such as 26) generally showed excellent microsome stability
across 3 species and showed minimal inhibition of Cyp2C9, 2D6,
and 3A4 (Table 4). However, this compound exhibited unexpect-
edly poor exposure in BALB/c mice by IP or PO administration.
Permeability for this compound was quite poor, as measured by
A to B flux across MDR1-transfected Madine-Darby canine kidney
(MDCK) monolayer. Moreover, the efflux ratio was high, strongly
suggesting that this compound is a substrate for PGP.26 (Any ratio
>2 is suggestive of PGP efflux.) Since both IP and PO dosing are sub-
ject to ‘first pass’ metabolism in the liver, we speculated that the
poor exposure may be due to either rapid metabolism or to PGP-
mediated biliary excretion. However, compound 26 and related
compounds (such as 23) had excellent microsomal stability sug-
gesting that metabolism was not a major route of elimination.
Moreover, compound 23 was evaluated in a hepatocyte stability
kinases: CHK1, PDGFRa, and VEGFR2. The selectivity of these com-
pounds is fascinating in light of the fact that the proposed key
interactions with the hinge (Met265) and the catalytic lysine
(Lys213) should be common to nearly all kinases. We believe that
the observed selectivity for this class of compounds partially arises
from the tight association of the ‘southern’ face of the indolo[2,3-
c]quinoline with the gatekeeper (Tyr262). Tyrosine gatekeepers
are unique to the IRAK family of kinases, of which only IRAK-1
and IRAK-4 are kinase-active. Only approximately 15% of kinases
have aromatic gatekeepers, and the vast majority of these are
phenylalanine.24 The relative rarity of aromatic gatekeepers com-
bined with the very tight SAR around the ‘gatekeeper’ region of
the core lead us to believe that this interaction may play an key
role in the observed selectivity.
assay and found to be exceptionally stable (Clint = 0.1–0.3 lL/
While the modest off-target activity (particularly against VEG-
FR2 and PDGFR) was certainly a concern, it was gratifying to see
that the compounds in Table 3 possessed minimal activity against
min/106 cells), only showing two minor metabolites: Demethyla-
tion and oxidation of the tertiary amine. Based on this evidence,
we reasoned that the poor PK exposure of 26 must be due to
PGP-mediated biliary excretion. This immediately led to two po-
tential strategies for improved exposure. First, by ‘swamping’ the
PGP pumps, the exposure may be improved in a non-linear man-
ner. (In other words, the PGP pumps have a finite capacity for ef-
flux. Once that capacity is reached, additional compound will be
free to enter circulation.) Secondly, the exposure may be improved
by finding a compound with a decreased efflux ratio or increased
permeability.
kinases involved in the TLR-signaling pathway, namely LCK, P38a,
and MK2.25 This gave us increased confidence that the inhibition of
R848-induced cytokine activity (Table 2) could, in fact, be attrib-
uted to IRAK4 inhibition. Armed with this information, the com-
pounds were examined in further depth to identify one or more
that could be advanced into in vivo proof-of-concept studies to
show that IRAK4 inhibition can block TLR signaling in an animal
model.

L. N. Tumey et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2066–2072
2071
The strategy of swamping the PGP pumps in order to increase
exposure was employed successfully. Increasing the IP dose of
compound 26 by 10-fold (100 mpk) increased the exposure and
the Cmax by approximately 20-fold, strongly indicating that our
strategy of swamping the PGP pumps had been successful (Table 4).
The measured blood levels of 26 (Cmax = 14
lM) were now found to
be in excess of the whole blood IC50 (5.5 M) for approximately 3 h
l
following the initial dose, paving the way for an in vivo proof-of-
mechanism study. (vide infra).
It has been noted that PGP substrates are disproportionately
hydrophobic and amphipathic, frequently containing aromatic
moieties linked to strong H-bond acceptors and positively charged
residues.27,28 We therefore hypothesized that by reducing the
polarity and H-bonding ability of the flexible ‘tail’ we may be able
to reduce the affinity of the compounds towards PGP. Moreover, a
non-ionized tailpiece may improve the permeability of the com-
pound thereby decreasing the efflux ratio. Therefore, in order to
find inhibitors with a decreased efflux ratio and improved perme-
ability, a subset of the previously described compounds was tested
for MDCK permeability (22–33, Table 5). As expected, compounds
with tertiary amines, such as 26–29 were found to have very poor
permeability (A–B) and high efflux ratios. Perhaps not surprisingly,
the least permeable compounds (26 and 29) also displayed the
poorest exposure upon IP administration. Ether -based solubilizing
tails afforded improved permeability (A–B) but only marginally
improved the efflux ratio. Interestingly, IP exposure seems to be
correlated reasonably well with permeability but not with efflux
ratio. This suggests that improvement in permeability can over-
come some PGP liabilities. Compound 32, with a di-ethylene glycol
tail, was found to provide the best balance of plasma exposure and
whole blood potency. This compound was profiled in more detail,
as outlined in Table 4.
Interestingly, compound 32 was observed to be slightly more
potent than compound 26 in the cellular assays (PBMCs and THPs)
in spite of being approximately 10-fold less potent against
IRAK4. This is consistent with its improved permeability
(0.60 Â 10À6 cm/s vs 0.10 Â 10À6 cm/s). The improved permeabil-
ity is also, presumably, largely responsible for the increased IP
and PO exposure of compound 32 compared to compound 26. This
improved exposure is in spite of poorer microsomal stability
(10–23 min) and increased polar surface area. Like compound 26,
compound 32 is highly protein-bound (>99%) thereby accounting
for the discrepancy between the cellular potency (29 nM) and
Figure 2. Stimulation of female C57Bl/6 mice with 8 mg of D-galactosamine (D-Gal)
followed by 500 ng of LPS.
compound 26 (536 h ng/mL vs 6500 h ng/mL). Unfortunately, the
blood levels at Cmax (5.5 M) were still below the targeted blood
levels based on the whole-blood potency (15 M). However, based
l
l
on the experience with compound 26, we had confidence that in-
creased doses of 32 should result in a proportional (or even
super-proportional) increase in exposure. Therefore, this com-
pound (dosed at 30 mpk by IP) was selected for advancement into
an in vivo proof-of-mechanism study.
The proof-of-mechanism study was performed on female
C57Bl/6 mice stimulated with 8 mg of D-galactosamine (D-Gal) fol-
lowed by 500 ng of LPS (Fig. 2). Animals were treated with test
compound or vehicle 2 h prior to stimulation with LPS. The animals
were harvested 1.5 h after stimulation and the blood levels of a
variety of cytokines were measured (Fig. 2). Vehicle-treated ani-
mals stimulated with LPS/
as indicated by levels of TNF
D
-Gal had a robust cytokine response
and IL-12. However, animals treated
a
with compound 26 (100 mpk) and compound 32 (30 mpk) showed
very low circulating cytokine levels, essentially equal to naïve
(non-stimulated) animals. The Cmax unbound concentration of 26
and 32 at the administered doses are estimated to be 110 and
49 nM, respectively (based on the plasma protein binding in
Table 4). These values are below the concentration necessary for
inhibition of most of the kinases described in Table 3 but above
the cellular IC50 reported in Table 2. These results strongly suggest
that these compounds are blocking the TLR signaling cascade
in vivo through an IRAK4 dependent mechanism.
the whole-blood potency (15 lM). As mentioned previously, a
10 mg/kg dose (IP) of compound 32 resulted in approximately
10-fold higher exposure as compared to the same dose of
LPS signals through the TLR-4 pathway, a signaling cascade that
has been shown by numerous studies to be dependent upon IRAK4
kinase activity. Mice deficient in functional IRAK4 have been
Table 5
Permeability and AUC for selected compounds
Compd#
R
MDCK, A–B (cm/s)
MDCK efflux ratio
216
Mouse 10 mpk IP AUC (h ng/mL)
1167
N
N
O
29
0.06 Â 10À6
N
O
26
27
0.10 Â 10À6
0.13 Â 10À6
78
1364
2680
N
N
O
O
263
28
30
0.14 Â 10À6
0.8 Â 10À6
167
40
O
O
2877
0.68 Â 10À6
0.60 Â 10À6
45
73
2912
6506
O
O
31
32
O
O
O

2072
L. N. Tumey et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2066–2072
8. Koziczak-Holbro, M.; Littlewood-Evans, A.; Pollinger, B.; Kovarik, J.; Dawson, J.;
Zenke, G.; Burkhart, C.; Muller, M.; Gram, H. Arthritis Rheum. 2009, 60, 1661.
9. Hernandez, M.; Bastian, J. F. Curr. Allergy Asthma Rep. 2006, 6, 468.
10. Lavine, E.; Somech, R.; Zhang, J. Y.; Puel, A.; Bossuyt, X.; Picard, C.; Casanova, J.
L.; Roifman, C. M. J. Allergy Clin. Immunol. 2007, 120, 948.
11. Rekhter, M.; Staschke, K.; Estridge, T.; Rutherford, P.; Jackson, N.; Gifford-
Moore, D.; Foxworthy, P.; Reidy, C.; Huang, X.-d.; Kalbfleisch, M.; Hui, K.; Kuo,
M.-S.; Gilmour, R.; Vlahos, C. J. Biochem. Biophys. Res. Commun. 2008, 367, 642.
12. Wang, Z.; Wesche, H.; Stevens, T.; Walker, N.; Yeh, W.-C. Curr. Top. Med. Chem.
2009, 9, 724.
13. Buckley, G. M.; Fosbeary, R.; Fraser, J. L.; Gowers, L.; Higueruelo, A. P.; James, L.
A.; Jenkins, K.; Mack, S. R.; Morgan, T.; Parry, D. M.; Pitt, W. R.; Rausch, O.;
Richard, M. D.; Sabin, V. Bioorg. Med. Chem. Lett. 2008, 18, 3656.
14. Buckley, G. M.; Ceska, T. A.; Fraser, J. L.; Gowers, L.; Groom, C. R.; Higueruelo, A.
P.; Jenkins, K.; Mack, S. R.; Morgan, T.; Parry, D. M.; Pitt, W. R.; Rausch, O.;
Richard, M. D.; Sabin, V. Bioorg. Med. Chem. Lett. 2008, 18, 3291.
15. Buckley, G. M.; Gowers, L.; Higueruelo, A. P.; Jenkins, K.; Mack, S. R.; Morgan, T.;
Parry, D. M.; Pitt, W. R.; Rausch, O.; Richard, M. D.; Sabin, V.; Fraser, J. L. Bioorg.
Med. Chem. Lett. 2008, 18, 3211.
shown to have an attenuated response to LPS stimulation, showing
dramatically lower levels of IL-6, IL-12, and TNF than their wild-
a
type counterparts.5,29 The two inhibitors described above (32 and
26) were shown to essentially recapitulate this phenotype in
non-mutant mice. Therefore, it is hoped that the resistance to
joint-inflammation that has been reported for IRAK4 deficient mice
will also be observed for these compounds.8 Moreover, based on
several recently reported studies of IRAK4-deficient patients, we
believe that the actions exhibited by these compounds may trans-
late into human efficacy. Whole blood and PBMCs isolated from
IRAK4-deficient patents have been shown to be nonresponsive to
LPS stimulation and other TLR-ligands (except for TLR3
ligands).10,30 Therefore, we are optimistic that blockage of IRAK4
signaling by inhibitors such as those described in this Letter will
ultimately be therapeutically useful in human disease.
16. Powers, J. P.; Li, S.; Jaen, J. C.; Liu, J.; Walker, N. P. C.; Wang, Z.; Wesche, H.
Bioorg. Med. Chem. Lett. 2006, 16, 2842.
17. Campeau, L.-C.; Parisien, M.; Jean, A.; Fagnou, K. J. Am. Chem. Soc. 2006, 128,
581.
Acknowledgments
18. Wang, Z.; Liu, J.; Sudom, A.; Ayres, M.; Li, S.; Wesche, H.; Powers, J. P.; Walker,
N. P. C. Structure 2006, 14, 1835.
19. Hu, Y.; Green, N.; Gavrin, L. K.; Janz, K.; Kaila, N.; Li, H.-Q.; Thomason, J. R.;
Cuozzo, J. W.; Hall, J. P.; Hsu, S.; Nickerson-Nutter, C.; Telliez, J.-B.; Lin, L.-L.;
Tam, S. Bioorg. Med. Chem. Lett. 2006, 16, 6067.
20. Bridges, A. J.; Zhou, H.; Cody, D. R.; Rewcastle, G. W.; McMichael, A.; Showalter,
H. D. H.; Fry, D. W.; Kraker, A. J.; Denny, W. A. J. Med. Chem. 1996, 39, 267.
21. Gorden, K. B.; Gorski, K. S.; Gibson, S. J.; Kedl, R. M.; Kieper, W. C.; Qiu, X.;
Tomai, M. A.; Alkan, S. S.; Vasilakos, J. P. J. Immunol. 2005, 174, 1259.
22. Qin, J.; Jiang, Z.; Qian, Y.; Casanova, J.-L.; Li, X. J. Biol. Chem. 2004, 279, 26748.
23. Morphy, R. J. Med. Chem. 2010, 53, 1413–1437.
We would like to thank Katherine Lee for advice and useful dis-
cussion during the project and during drafting of this report. We
also thank Mala Gopalsamy and the exploratory chemistry team
for identifying this lead series. We are greatful to the members
of the Wyeth Chemical Technologies (CT) and Drug-Safety and
Metabolism (DSM) team for performing the pharmaceutical profil-
ing. Finally, we would also like to thank Li Di for helpful discus-
sions and advice about pharmaceutical properties and PK results.
24. Zuccotto, F.; Ardini, E.; Casale, E.; Angiolini, M. J. Med. Chem. 2010, 53, 2681.
25. Gaestel, M.; Kotlyarov, A.; Kracht, M. Nat. Rev. Drug Disc. 2009, 8, 480.
26. del Amo Eva, M.; Heikkinen Aki, T.; Monkkonen, J. Eur. J. Pharm. Sci. 2009, 36,
200.
Supplementary data
27. Penzotti, J. E.; Lamb, M. L.; Evensen, E.; Grootenhuis, P. D. J. J. Med. Chem. 2002,
45, 1737.
28. Eckford, P. D. W.; Sharom, F. J. Chem. Rev. 2009, 109, 2989.
29. Koziczak-Holbro, M.; Gluck, A.; Tschopp, C.; Mathison, J. C.; Gram, H. Eur. J.
Immunol. 2008, 38, 788.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2014.03.
056.
30. Ku, C.-L.; von Bernuth, H.; Picard, C.; Zhang, S.-Y.; Chang, H.-H.; Yang, K.;
Chrabieh, M.; Issekutz, A. C.; Cunningham, C. K.; Gallin, J.; Holland, S. M.;
Roifman, C.; Ehl, S.; Smart, J.; Tang, M.; Barrat, F. J.; Levy, O.; McDonald, D.;
Day-Good, N. K.; Miller, R.; Takada, H.; Hara, T.; Al-Hajjar, S.; Al-Ghonaium, A.;
Speert, D.; Sanlaville, D.; Li, X.; Geissmann, F.; Vivier, E.; Marodi, L.; Garty, B.-Z.;
Chapel, H.; Rodriguez-Gallego, C.; Bossuyt, X.; Abel, L.; Puel, A.; Casanova, J.-L. J.
Exp. Med. 2007, 204, 2407.
31. Cushing, L.; Stochaj, W.; Siegel, M.; Czerwinski, R.; Dower, K.; Wright, Q.;
Hirschfield, M.; Casanova, J.L.; Picard, C.; Puel, A.; Lin, L.L.; Rao, V. J. Biol. Chem.
2014, in press.
References and notes
1. O’Neill, L. A. J. Curr. Opin. Pharmacol. 2003, 3, 396.
2. Kanzler, H.; Barrat, F. J.; Hessel, E. M.; Coffman, R. L. Nat. Med. 2007, 13, 552.
3. Wagner, H. Adv. Immunol. 2006, 91, 159.
4. Suzuki, N.; Saito, T. Trends Immunol. 2006, 27, 566.
5. Kawagoe, T.; Sato, S.; Jung, A.; Yamamoto, M.; Matsui, K.; Kato, H.; Uematsu, S.;
Takeuchi, O.; Akira, S. J. Exp. Med. 2007, 204, 1013.
6. Fraczek, J.; Kim, T. W.; Xiao, H.; Yao, J.; Wen, Q.; Li, Y.; Casanova, J.-L.; Pryjma, J.;
Li, X. J. Biol. Chem. 2008, 283, 31697.
7. Kim, T. W.; Staschke, K.; Bulek, K.; Yao, J.; Peters, K.; Oh, K.-H.; Vandenburg, Y.;
Xiao, H.; Qian, W.; Hamilton, T.; Min, B.; Sen, G.; Gilmour, R.; Li, X. J. Exp. Med.
2007, 204, 1025.