Optimization of the physicochemical and pharmacokinetic attributes in a 6-azauracil series of P2X7 receptor antagonists leading to the discovery of the clinical candidate CE-224,535

Article abstract of DOI:10.1016/j.bmcl.2011.04.077

High throughput screening (HTS) of our compound file provided an attractive lead compound with modest P2X₇ receptor antagonist potency and high selectivity against a panel of receptors and channels, but also with high human plasma protein bindi

Full text of DOI:10.1016/j.bmcl.2011.04.077

Bioorganic & Medicinal Chemistry Letters 21 (2011) 3708–3711
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Optimization of the physicochemical and pharmacokinetic attributes
in a 6-azauracil series of P2X7 receptor antagonists leading to the discovery
of the clinical candidate CE-224,535
⇑
Allen J. Duplantier , Mark A. Dombroski, Chakrapani Subramanyam, Aimee M. Beaulieu, Shang-Poa Chang,
Christopher A. Gabel, Crystal Jordan, Amit S. Kalgutkar, Kenneth G. Kraus, Jeff M. Labasi,
Christopher Mussari, David G. Perregaux, Rick Shepard, Timothy J. Taylor, Kristen A. Trevena,
Carrie Whitney-Pickett, Kwansik Yoon
Pfizer Global Research and Development, Groton Laboratories, Eastern Point Road, Groton, CT 06340, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
High throughput screening (HTS) of our compound file provided an attractive lead compound with mod-
est P2X7 receptor antagonist potency and high selectivity against a panel of receptors and channels, but
also with high human plasma protein binding and a predicted short half-life in humans. Multi-parameter
optimization was used to address the potency, physicochemical and pharmacokinetic properties which
led to potent P2X7R antagonists with good disposition properties. Compound 33 (CE-224,535) was
advanced to clinical studies for the treatment of rheumatoid arthritis.
Received 7 March 2011
Revised 15 April 2011
Accepted 19 April 2011
Available online 24 April 2011
Keywords:
P2X7
Ó 2011 Elsevier Ltd. All rights reserved.
Antagonist
Rheumatoid arthritis
6-Azauracil
Pharmacokinetic
Log P
The P2X family of purinoreceptors are ligand gated ion channels
of which there are seven known receptors (P2X1–P2X7) with 30 –
50% amino acid sequence identity.1 The largest of the group, the
P2X7 receptor (P2X7R), is expressed in the periphery on immune
cells and in the CNS on microglia and astrocytes and has been
shown to be linked to glutamate release.2 Because of its role in
interleukin-1 (IL-1) posttranslational processing,3 this receptor
has been the target of drug discovery efforts in the treatment of
inflammatory diseases, pain4 and neurodegenerative diseases such
as Alzheimer’s disease (AD).5
Activation of the P2X7R on leukocytes leads to opening of a non-
selective cation channel which in turn can lead to complete depo-
larization of the membrane potential.6 Within seconds of opening,
the P2X7R channel transitions to a pore state that is permeable to
large organic molecules, such as YoPro Yellow.7 This feature, which
is a hallmark of the P2X7R, creates a state of osmotic imbalance and
can cause LPS-activated monocytes and macrophages to die.
Recent studies have shown that P2X7R activation is associated with
activation of the inflammasome, an assembly of cytoplasmic pro-
teins that facilitates caspase-1 activation and, in turn, proteolytic
processing of proIL-1b and release of the mature cytokine to the
medium. This unique type of programmed cellular death response
resulting in the release of proinflammatory mediators has been
termed pyroptosis8 to distinguish it from a non-inflammatory
apoptotic process. Thus, a P2X7R antagonist should block IL-1 post-
translational processing and lead to a reduction in the release of IL-
1b. In support of this hypothesis, studies using P2X7R knockout
mice have shown that absence of the P2X7R leads to an inability
of macrophages and monocytes to generate and release mature
IL-1b in response to ATP challenge.9 In vivo, P2X7R-deficient mice
demonstrated less severe outcomes in anticollagen-induced arthri-
tis9 and formalin-induced pain disease models.4 In efforts to test
the hypothesis of whether a P2X7R antagonist would improve the
signs and symptoms of rheumatoid arthritis (RA) in humans, we
sought to discover an orally active clinical candidate that could
be safely administered to humans at doses capable of achieving
free systemic concentrations above its IC90 for blocking IL-1b re-
lease. The high bar of achieving >90% inhibition of IL-1b in the
clinic was set in efforts to improve upon the efficacy of anakinra,
a natural IL-1 receptor antagonist for the treatment of RA.
While several small molecule P2X7R antagonists have been
reported in the literature,10 our efforts began with high throughput
screening of our compound file against the P2X7R using a YOPRO-1
⇑
Corresponding author.
E-mail address: allen.j.duplantier@pfizer.com (A.J. Duplantier).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.04.077

A. J. Duplantier et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3708–3711
3709
Cl
O
O
ment with hydrogen (10) greatly diminished the blocking of IL-1b
release. The 2-ethyl-1-hexyl substituent of 1 could be reduced in
size to 2-ethyl-1-butyl (11), n-hexyl (12) or n-pentyl (13) without
loss of activity, but the n-butyl group (14) and other 1-4 carbon al-
kyl groups (data not shown) did not appear large enough to ade-
P2X7R Yo-Pro IC50 = 1.1 μM
IL-1β IC50 = 0.8 μM
MW = 379
cLog P = 4.45
N
H
N
O
N
quately fill the lipophilic pocket (IL-1b IC50 >5 lM) (Table 2).
NH
Tying the alkyl side chain of 11 into a cyclohexyl ring (15) did
not affect potency, but surprisingly the analogous cyclopentyl ana-
log (16) was much less active, and direct attachment of carbocycles
to the amide nitrogen led to inactivity (data not shown). In efforts
to reduce lipophilicity, an oxygen atom was walked around the
aforementioned active analogs (17–19), but this change led to loss
of activity. Next, the amide side chain was replaced with benzyl
(20) and then homologated out to phenethyl (21) and phenylpro-
pyl (22). Of these, only compound 21 had an IL-1b IC50 under
1
Figure 1. Properties of compound 1.
fluorescence assay11 and uncovered compound 1 as a novel hit.
Although compound 1 had moderate P2X7R antagonist activity,
and relatively low MW it did have moderately high c Log P (
Fig. 1), and thus our first goal was to improve potency and lower
c Log P simultaneously. Initial efforts towards this goal were to ex-
plore the amide region via parallel chemistry. Thus acid 5 was cou-
pled with a variety of readily available amines (Scheme 1) and
biological testing11 of the resulting amides (6) provided a wealth
of SAR information.
5
l
M. A variety of subtituents were substituted around the phen-
ethyl moiety (data not shown) and 2-chlorophenethyl (23)
emerged as a viable lead (IL-1b IC50 = 0.29 M). Compound 23
was equipotent in the YOPRO assay (IC50 = 0.3 M), but was inac-
tive in the presence of human blood (IL-1b IC50 >10 M), likely
l
l
l
The synthesis of 1-phenyl-6-azauracils has been reported, and
using appropriate modifications of the literature procedures,12
the preparation of compound 1 and its analogs began with diazo-
tization of 2-chloro-5-aminobenzoic acid (2) (Scheme 1). Addition
of the diethylcarbamate of malonic acid (3) to the diazonium salt of
2 at room temperature followed by heating sequentially in the
presence of sodium acetate and then sulfuric acid provided the
1-aryl-6-azauracil-5-carboxylate 4 in moderate yield (46%). Subse-
quent decarboxylation with sodium hydroxide in mercapto-acetic
acid provided intermediate 5. The carboxylic acid moiety of 5
was readily esterified or converted to a variety of amides (6) via
acylation, and the 3-position of the 6-azauracil ring could then
be alkylated with either alkyl halides in the presence of base, or
by epoxides by heating in DMF without the addition of base to pro-
vide analogs 7 (Scheme 2).
due to high human plasma protein binding of > 99.5%. However,
the in vitro ADME profile of 23 was favorable [little to no turnover
in human liver microsomes and human hepatocytes, excellent
absorptive permeability (CACO Papp AB = 10 Â 10À6 cm/sec,
BA = 21 Â 10À6 cm/sec)], and an in vivo rat pharmacokinetic (PK)
study showed decent oral bioavailability (F = 44%), consistent with
the observed low plasma clearance (CLp = 9.5 mL/min/kg).
Our follow-up strategy for compound 23 was to further improve
potency through amide side chain manipulation while reducing
plasma protein binding through further reduction of c Log P. In ef-
forts toward that end, the systematic insertion of nitrogen atoms
around the phenethyl moiety of 21 lowered lipophilicity, but
unfortunately led to a substantial loss in activity (compounds
24–26). However, sytematically inserting polar groups, such as hy-
droxyl and cyano around the cyclohexyl ring of 15 gave some
promising results. While hydroxylation at the 2-, 3- and 4-position
of the cyclohexyl ring led to a significant loss in activity (data not
shown), quaternary substitution at the 1-position provided an in-
crease in activity and a decrease in c Log P (e.g., compound 27,
The necessity of the amide NH moiety of compound 1 was con-
firmed by the dramatic loss in activity when it was either methyl-
ated (8) or replaced with oxygen (9) (Table 1). Likewise, the ortho-
chloro substituent of 1 was also shown to be important as replace-
O
N
O
NH2
HN
OH
OH
O
O
O
O
N
i
ii
O
EtO
N
H
N
H
OEt
+
OH
Cl
O
3
2
Cl
O
4
O
O
H
N
HN
N
iii, iv
N
O
N
O
N
H
N
OH
R1
Cl
O
6
Cl
O
5
Scheme 1. Reagents and conditions: (i) AcOH, 12 N HCl (3.3 equiv), NaNO2, 10 ꢀ 15 °C, 30 min; then NaOAc (1 equiv), addition of compound 3, 10 °C–rt, 80 min; then NaOAc
(2 equiv), reflux, 6 h; then 50% H2SO4, reflux, 2 h (one pot, isolated by filtration, purified by ether trituration, 46% yield); (ii) mercaptoacetic acid (1.1 equiv), NaOH (2.2 equiv),
water, 175 °C, isolated by filtration, recrystallized from methanol/water, 75%; (iii) oxalyl chloride, DCM, 92%; (iv) amine, 1 N NaOH (2 equiv).

3710
A. J. Duplantier et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3708–3711
Table 2
O
O
N
SAR around the amide moiety of compound 1
R2
O
N
R1
O
N
H
O
HN
N
N
N
N
i
N
O
HN
O
R2X or
R
H
N
H
N
O
Cl
M or % inh @ 5
R1
R1
7
Compd
R1
IL-1b IC50,
l
l
Ma,b
c Log p
6
Cl
O
Cl
O
Scheme 2. Reagents and conditions: (i) alkyl bromide, Cs2CO3, acetone; or
substituted epoxide, DMF, 80 °C.
11
1.1
3.4
12
13
14
3.6
3.5
3.0
2.5
Table 1
2.8
Importance of NH and Cl groups in compound 1
0% inh
15
16
1.3
3.5
3.0
O
X
N
N
0% inh
H
N
O
17
18
2% inh
>5
1.8
2.0
O
O
R
O
Compd
R
X
IL-1b IC50
,
l
M or % inh @ 5 l
Ma,b
1
8
9
Cl
Cl
Cl
H
NH
NMe
O
0.8
19
20
>5
1.1
2.8
O
0% inh
19% inh
0% inh
10
NH
15% inh
a
Inhibition of the release of lL-1b from monocytes stimulated by ATP.
IC50 value is an average of n = 2.
b
21
22
0.94
3.0
3.3
IC50 = 0.4 lM, c Log P = 2.3). Furthermore, the 1-hydroxycyclohep-
0% inh
tyl homolog 28 was 4-fold more potent than 27 and was signifi-
cantly less protein bound compared to 23 (human plasma
Fu = 0.18 vs 0.005), had sub-micromolar potency in the presence
of human blood (IC50 = 564 nM, IC90 = 1000 nM), low lipophilicity
(c Log P = 2.8) and low predicted clearance from human liver
microsomal and hepatocyte studies.
Cl
23
0.29
3.7
N
24
25
26
15% inh
22% inh
22% inh
1.5
1.5
1.5
Attempts to improve potency by replacement of the ortho-Cl
group of 28 with methyl (IC50 = 0.7 M), fluoro (20% inhibition at
l
N
1
lM) or cyano (0% inhibition at 1 lM) were unsuccessful. Subse-
quently, we held the 1-hydroxycycloheptyl group of compound
28 constant and explored substitution off of the 3-position of the
6-azauracil ring (Table 3). Methyl substitution (29) revealed that
the N3 proton was not necessary for activity. Furthermore,
although the ethyl analog (30) was equipotent with 28 in the IL-
1b assay, it did not lose significant potency in the presence of hu-
man blood (IC50 = 0.185 lM). Many close-in analogs were prepared
varying the substituent at the 3-position (e.g., 31–33), and of these,
the (R)-1-hydroxy-2-methoxyethyl group (33) proved optimal as
N
HO
HO
27
28
0.40
2.3
2.9
0.083 (n = 67)
its low potency in the IL-1b assay (IC50 = 0.001
lower plasma protein binding (Fup = 0.27) translated into excep-
tional potency in the presence of human blood (IC50 = 0.001 M,
IC90 = 0.0047 M).
lM) coupled with
a
Inhibition of the release of lL-1b from monocytes stimulated by ATP.
IC50 values are an average of n = 2 unless otherwise noted.
b
l
l
The pharmacology and PK profile of compounds 23, 28 and 33 is
shown in Table 4. In rats, compound 33 had low CLp (11 mL/min/
kg) and a large Vdss of 7.6 L/kg, which resulted in a half-life of
2.4 h. Upon oral administration to rats at 5 mg/kg, compound 33
provided maximal plasma exposure (Cmax) that was ꢀ90 fold over
interaction with the major human CYP isozymes (CYP-1A2, 2C19,
2C9, 2D6 or 3A4, IC50’s >30 M). From a safety point of view, 33
was negative in our biolumAmes and micronucleus assays, had
no effect on hERG at 1 M, and did not show toxicity in rats when
dosed daily at 500 mg/kg, p.o. over a 4 day period. Unfortunately,
we were unable to obtain meaningful data for 33 in a rodent dis-
ease model because of its weak potency against the mouse
l
l
its IC90 in human blood (Cmax = 0.21
l
g/mL or 0.44
lM). The oral
bioavailability of 33 was low in rats (F = 2.6%), but this was be-
lieved to be a rat specific phenomenon since corresponding oral
bioavailability in both dog (59%) and monkey (22%) was adequate.
P2X7R. For example, 10 lM concentrations of 33 yielded minimal
At 10
lM compound 33 had no effect on a CEREP panel of 107
inhibitions of ATP-induced IL-1b posttranslational processing
receptors, channels and transporters, nor did it have any
when using LPS-activated murine peritoneal macrophages.

A. J. Duplantier et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3708–3711
3711
Table 3
Compound 33, also referred to as CE-224,535, was recently
evaluated in the clinic for the treatment of rheumatoid arthritis
(RA).13 Although CE-224,535 (500 mg, BID) was generally well tol-
erated over a 12 week study where median trough plasma concen-
trations reached ꢀ250 ng/mL (519 nM), it was not effective in
treating RA in patients with an inadequate response to
methotrexate.13
In conclusion, through a combination of parallel chemistry and
property focused medicinal chemistry, we optimized an HTS hit
P2X7R antagonist (1) to identify a clinical candidate 33 (CE-
224,535). The excellent PK and safety profile of compound 33 al-
lowed for us to adequately test the effect of the P2X7R antagonist
mechanism in the clinic as a potential treatment for RA. Although
ineffective for the treatment of RA at adequate drug exposures,
compound 33 may still hold promise for the treatment of other dis-
eases such as pain and AD.
Substitution on N3 of 6-azauracil moiety and its effect on P2X7R potency in the
presence of human blood
OH
O
HN
N
N
N
O
R2
O
Cl
a,b
Compd
R2
IL-1b IC50
M)
IL-1b (human blood)
a,b
(l
IC50
(lM)
28
29
30
H
Me
Et
0.083 (n = 67)
0.042
0.447 (n = 7)
NTc
0.185 (n = 1)
0.086
31
32
0.034
0.019
0.200 (n = 1
O
0.071
References and notes
OH
33
0.0014 (n = 5)
0.0008 (n = 13)
O
1. Koles, L.; Furst, S.; Illes, P. Drug News Perspect. 2005, 18, 85.
2. Papp, L.; Vizi, E. S.; Sperlagh, B. NeuroReport 2004, 15, 2387.
3. Solle, M.; Labasi, J.; Perregaux, D. G.; Stam, E.; Petrushova, N.; Koller, B. H.;
Griffiths, R. J.; Gabel, C. A. J. Biol. Chem. 2001, 276, 125.
4. Chessell, I. P.; Hatcher, J. P.; Bountra, C.; Michel, A. D.; Hughes, J. P.; Green, P.;
Egerton, J.; Murfin, M.; Richardson, J.; Peck, W. L.; Grahames, C. B.; Casula, M.
A.; Yiangou, Y.; Birch, R.; Anand, P.; Buell, G. N. Pain 2005, 114, 386.
5. Parvathenani, L. K.; Tertyshnikova, S.; Greco, C. R.; Roberts, S. B.; Robertson, B.;
Posmantur, R. J. Biol. Chem. 2003, 278, 13309.
a
b
c
Inhibition of the release of lL-1b from monocytes stimulated by ATP.
IC50 values are an average of n = 2 unless otherwise noted.
NT = not tested.
Table 4
6. Perregaux, D. G.; McNiff, P.; Laliberte, R.; Conklyn, M.; Gabel, C. A. J. Immunol.
2000, 165, 4615.
7. Gunosewoyo, H.; Coster, M. J.; Bennett, M. R.; Kassiou, M. Bioorg. Med. Chem.
2009, 17, 4861.
8. Bergsbaken, T.; Fink, S. L.; Cookson, B. T. Nat. Rev. Microbiol. 2009, 7, 99.
9. Labasi, J. M.; Petrushova, N.; Donovan, C.; McCurdy, S.; Lira, P.; Payette, M. M.;
Brissette, W.; Wicks, J. R.; Audoly, L.; Gabel, C. A. J. Immunol. 2002, 168, 6436.
10. (a) Guile, S. D.; Alcaraz, L.; Birkinshaw, T. N.; Bowers, K. C.; Ebden, M. R.; Furber,
M.; Stocks, M. J. J. Med. Chem. 2009, 52, 3123; (b) Carroll, W. A.; Donnelly-
Roberts, D.; Jarvis, M. F. Purinergic Signalling 2009, 5, 63; (c) Donnelly-Roberts,
D. L.; Jarvis, M. F. Br. J. Pharmacol. 2007, 151, 571; (d) Romagnoli, R.; Baraldi, P.
G.; Cruz-Lopez, O.; Lopez-Cara, C.; Preti, D.; Borea, P. A.; Gessi, S. Exp. Opin. Ther.
Targets 2008, 12, 647.
Pharmacology and pharmacokinetics of key P2X7R antagonists.
Compd 23 Compd 28 Compd 33
Yo-Pro IC50 (nM)
IL-1b IC50 (nM)
300
285
15
83
4
1
1
Human blood IC50 (nM) >10,000
Human blood IC90 (nM)
447
1000
4.7
0.27
2.6
11
7.6
2.4
0.21
2.6
Human plasma Fu
c Log P
CLp (mL/min/kg)
Vdss (L/kg)
0.005
3.7
9.5
0.3
3
0.18
2.9
20
8.5
3.2
1.3
Rat PK:
Dog PK:
T1/2 (h)
11. Analogs of compound
1 were evaluated as antagonists of the P2X7R by
Cmax
(
lg/mL)
2
assessing whether they inhibited the release of IL-1b from LPS-activated
human monocytes maintained in the presence of low serum. Promising
compounds were re-evaluated in an analogous assay, employing human whole
blood samples sequentially treated with LPS and ATP. The final concentration
of ATP was 6 mM. Selected compounds were also tested for their ability to
inhibit uptake of YOPRO-1 by ATP-treated P2X7R over-expressing HEK293 cells
to provide direct evidence of P2X7R impact. For assay protocols, see:
Duplantier, A. J. and Subramanyam, C. WO 2003042191, 2003.
F (%)
CLp (mL/min/kg)
Vdss (L/kg)
T1/2 (h)
Cmax (lg/mL)
44
75
3.8
0.6
3.8
7.4
12
0.61
0.46
3.5
59
16
0.68
0.77
0.55
22
F (%)
84
21
1.5
Monkey PK: CLp (mL/min/kg)
Vdss (L/kg)
12. Miller, M. W.; Mylari, B. L.; Howes, H. L., Jr.; Lynch, J. E.; Lynch, M. J.; Koch, R. C.
J. Med. Chem. 1979, 22, 1483.
13. Stock, T.; Bloom, B. J.; Wang, X.; Gupta, P. and Mebus, C. Eur. League Against
Rheumatism, Rome, Italy, June 16–19, 2010; THU0217.
T1/2 (h)
Cmax (lg/mL)
3.4
0.6
F (%)
72