Identification and SAR of novel diaminopyrimidines. Part 1: The discovery of RO-4, a dual P2X3/P2X2/3 antagonist for the treatment of pain

Article abstract of DOI:10.1016/j.bmcl.2009.02.003

P2X purinoceptors are ligand-gated ion channels whose endogenous ligand is ATP. Both the P2X₃ and P2X_2/3 receptor subtypes have been shown to play an important role in the regulation of sensory function and dual P2X₃/P2X_2/3 antagonists offer significant potential for the treatment of pain. A high-throughput screen of the Roche compound collection resulted in the identification of a novel series of diaminopyrimidines; subsequent optimization resulted in the discovery of RO-4, a potent, selective and drug-like dual P2X₃/P2X_2/3 antagonist.

Full text of DOI:10.1016/j.bmcl.2009.02.003

Bioorganic & Medicinal Chemistry Letters 19 (2009) 1628–1631
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Identification and SAR of novel diaminopyrimidines. Part 1: The discovery
of RO-4, a dual P2X3/P2X2/3 antagonist for the treatment of pain
a
a
a
b
b
David S. Carter a, , Muzaffar Alam , Haiying Cai , Michael P. Dillon , Anthony P. D. W. Ford , Joel R. Gever ,
*
Alam Jahangir a, Clara Lin a, Amy G. Moore a, Paul J. Wagner a, Yansheng Zhai a
a Department of Medicinal Chemistry, Roche Palo Alto LLC, 3431 Hillview Avenue, Palo Alto, CA 94304, USA
b Department Biochemical Pharmacology, Roche Palo Alto LLC, 3431 Hillview Avenue, Palo Alto, CA 94304, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
P2X purinoceptors are ligand-gated ion channels whose endogenous ligand is ATP. Both the P2X3 and
P2X2/3 receptor subtypes have been shown to play an important role in the regulation of sensory function
and dual P2X3/P2X2/3 antagonists offer significant potential for the treatment of pain. A high-throughput
screen of the Roche compound collection resulted in the identification of a novel series of diaminopyrim-
idines; subsequent optimization resulted in the discovery of RO-4, a potent, selective and drug-like dual
P2X3/P2X2/3 antagonist.
Received 17 December 2008
Revised 30 January 2009
Accepted 2 February 2009
Available online 7 February 2009
Keywords:
P2X3
Ó 2009 Elsevier Ltd. All rights reserved.
Purinoceptors
ATP
RO-4
Chronic pain
Inflammatory pain
Neuropathic pain
Drug-like
P2X3 antagonist
P2X purinoceptors are ligand-gated ion channels activated by
adenosine 50-triphosphate (ATP). Seven P2X receptor subunits have
been identified (P2X1–7) and each channel shown to be assembled
from three subunits.1 Understanding the pharmacology is complex
since each subunit has the ability to form heteromeric channels. A
total of seven heteromeric P2X family members have been identi-
fied.2 Homomeric P2X3, and the closely related heteromultimeric
P2X2/3, receptors are predominantly localized on small to medium
diameter sensory afferent neurons and have become increasingly
recognized as playing a major role in mediating the primary sen-
sory effects of ATP. P2X3 receptor expression is upregulated in
DRG neurons following ligation of the sciatic nerve in the chronic
constriction injury (CCI) model.3 P2X3-KO mice demonstrate a re-
duced sensitivity to thermal stimuli and decreased pain behaviors.4
Furthermore reduction of P2X3 expression through intrathecal
administration of P2X3-selective antisense oligonucleotide,5 and
more recently siRNA,6 also causes a significant decrease in pain
behaviors in mice.
nyl)adenosine 50-triphosphate (TNP-ATP) (1) was the only reported
dual P2X3/P2X2/3 antagonist though its nucleotide nature primarily
limits its use to in vitro experiments.7 In a more recent report the
peptide spinorphin has been reported as a selective P2X3 antago-
nist but no data at P2X2/3 is described8 (Fig. 1).
The discovery of A-317491 (2) marked a significant break-
through; for the first time a truly selective, non-nucleotide, small
molecule dual P2X3/P2X2/3 antagonist was reported.9 The impres-
sive in vivo efficacy profile in multiple chronic inflammatory and
neuropathic pain models validated the importance of these recep-
tors in pain signaling pathways. More recent data demonstrating
the efficacy of A-317491 in a cyclophosphamide induced cystitis
model also signals the potential of targeting P2X3-containing recep-
tors for the treatment of overactive bladder.10 While displaying an
impressive array of in vivo activity, the potential of A-317491 as a
therapeutic agent is limited by the poly-acidic nature it shares with
many of the non-selective antagonists. The presence of the three
carboxylic acid groups significantly limits both the oral bioavail-
ability and the CNS penetration. In this Letter we describe the dis-
covery and structure–activity relationships of a novel series of
diaminopyrimidines resulting in the discovery of RO-4; a potent,
selective11 and drug-like12 dual P2X3/P2X2/3 receptor antagonist.
A high-throughput screening campaign of the Roche compound
collection using the rat recombinant P2X3 receptor was performed.
Despite the mounting evidence for the importance of these
receptors, the field was hampered by the lack of potent and
selective small molecule antagonists. 20,30-O-(2,4,6-Trinitrophe-
* Corresponding author. Tel.: +1 650 855 5615; fax: +1 650 855 5238.
E-mail address: david.carter@roche.com (D.S. Carter).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.02.003

D. S. Carter et al. / Bioorg. Med. Chem. Lett. 19 (2009) 1628–1631
1629
NH2
N
OMe
N
O
OMe
N
NH2
N
a,b
c,d,e
O
P
O
P
O
P
N
Ar
Ar
O
N
OH
O
O
O
OMe
NH2
N
N
OMe
N
N
OH OH
OH
TNP-ATP (1)
Scheme 1. Preparation of C-linked diaminopyrimidines. Reagents and conditions:
(a) LiTMP, THF, RCHO, 0 °C; (b) MnO2, toluene, reflux; (c) 7 M NH3, MeOH, 80 °C,
sealed tube; (d) LiAlH4, THF, reflux; (e) TFA, CH2Cl2, Et3SiH.
O
O
O2N
NO2
NO2
CO2H
CO2H
Table 1
Optimization of the arene side-chain
NH2
N
R
HO2C
O
O
N
N
NH2
O
A-317491 (2)
O
a,c
b,c
Entry
Compound
R
P2X3
P2X2/3
1
2
3
4
5
6
7
8
5
3
6
Me
Et
< 5.0
5.8
<5.0
6.9
<5.0
<5.0
<5.0
5.7
<5.0
5.0
Figure 1.
n-Pr
i-Pr
c-Pr
i-Bu
t-Bu
c-Bu
7 (RO-3)
8
9
10
11
In addition to a very low overall hit rate of 0.01%, many of the more
potent hits contained poly-acidic functionality, similar to the
known antagonists, with the implied poor drug-like properties
associated with these functional groups. These were not deemed
suitable starting points for a medicinal chemistry program aimed
at producing potential therapeutic agents so a careful analysis of
the less-potent hits was undertaken. The low molecular weight
diaminopyrimidine 3 was identified with a pIC50 of 5.8 at P2X3
5.9
<5.0
<5.0
<5.0
<5.0
<5.0
a
FLIPR: mean pIC50, rP2X3 CHO cell.
FLIPR: mean pIC50, hP2X2/3 1321N1 (astrocytoma) cells.
pIC50 values are the mean of at least three experiments performed in triplicates,
b
c
standard deviation 20%.
but with no activity at P2X2/3 below 10 lM (pIC50 < 5). This mole-
cule was originally synthesized as an analog of the anti-bacterial
diaminopyrimidine trimethoprim (4) and shares many of its struc-
tural features (Fig. 2).
The lead optimization strategy for the diaminopyrimidine series
focused on three key features; the small alkyl side-chain, the
arene–pyrimidine linker and the 3,4-disubstituted arene. Although
necessary for activity, the nature of the diaminopyrimidine SAR
will be the subject of a subsequent communication.
(6) analogs were devoid of all P2X3 activity. However, isopropyl
arene 7 (RO-3) was 10-fold more potent than ethyl arene 3. Fur-
thermore, incorporation of one additional methyl group, either at
the benzylic carbon (tert-butyl 10) or the terminal carbon (iso-bu-
tyl 9) resulted in complete loss of activity. Also, the c-butyl analog
was not active. Together these results suggest that although there
seems to be a requirement for a small side-chain, a major contrib-
uting force for P2X3 binding must be the torsional angle of the side-
chain. Specifically, analogs such as i-propyl arene 7 can adopt a
low-energy conformation which places the side-chain C–H in plane
(toward the diaminopyrimidine). This orients a methyl group into
a region of space inaccessible to inactive analogs such as tert-butyl
arene 10.
Trimethoprim (4) is prepared via the on the metric ton scale.13
This synthesis is based on an aldol condensation of an aldehyde
with cyanomethoxyethane to build in the precursor of the diami-
nopyrimidine ring. Because this method failed to give appreciable
amounts of aldol adduct with hindered aromatic aldehydes, we
chose to develop a new route based on lithiated 2,4-dimethoxy-
pyrimidine.14 This five step synthesis is outlined in Scheme 1.
Ortholithiation of 2,4-dimethoxypyrimidine was accomplished
with lithium 2,2,6,6-tetramethylpiperidine (LTMP) in THF at 0 °C.
Addition of a substituted aromatic aldehyde15 gave a secondary
alcohol which was oxidized with MnO2 to a ketone. Treatment of
the ketone with NH3/MeOH at 80 °C gave a 5-acyl-2,4-diaminopyr-
imidine. Reduction of the ketone via LiAlH4 followed by treatment
with TFA/Et3SiH gave the target diaminopyrimidines (typically 15%
for 5 steps).
Next the role of arene–diaminopyrimidine linker was examined
(Table 2). Both ketone 13 and alcohol 12 were completely inactive.
Extended straight-chain aliphatic analogs such as phenethyl 14
Table 2
Ether containing linker leads to an impressive boost in potency
NH2
N
R
X
Initial optimization efforts focused on preparation of side-chain
analogs (Table 1). Screening hit 3, which contained an ethyl side-
chain, was moderately active at P2X3. Methyl (5) and n-propyl
N
NH2
O
O
a,c
b,c
Entry
Compound
Structure
P2X3
P2X2/3
1
2
3
4
5
12
13
14
15
16
R = i-Pr, X = CHOH
R = i-Pr, X = C = O
R = Et, X = CH2CH2
R = Et, X = O
<5.0
<5.0
<5.0
6.8
<5.0
<5.0
<5.0
5.7
NH2
N
NH2
N
O
O
R = i-Pr, X = O
7.6
6.3
N
NH2
NH2
N
O
a
O
FLIPR: mean pIC50, rP2X3 CHO cell.
FLIPR: mean pIC50, hP2X2/3 1321n1c cell.
pIC50 values are the mean of at least three experiments performed in triplicates,
O
b
c
(3)
(4)
standard deviation 20%.
Figure 2.

1630
D. S. Carter et al. / Bioorg. Med. Chem. Lett. 19 (2009) 1628–1631
pIC50 = 5.5). In fact, this compound, which lacks the key 3-methoxy
NH2
N
CN
O
group, was found to be fivefold less potent than 4-methoxy ben-
zene 19 (P2X3 pIC50 = 5.8), suggesting that size alone is not influ-
encing activity. Both chloroanisole 30 and bromoanisole 31 were
similarly potent to dimethoxy benzene 16. Interestingly, iodoani-
sole 32 (RO-4), was fourfold more potent than dimethoxy benzene
16 (P2X3 pIC50 = 8.0). These results together suggest that the sub-
stituent at the 3-position is not only occupying a hydrophobic
pocket but may be fulfilling a key hydrogen bond interaction. Since
hydrogen bond strength is dominated by distance and basicity,
fluorine would be expected to be a very weak acceptor (inactive),17
O
a,b
Ar
c
OH
Ar
Ar
OMe
NH2
N
Scheme 2. Preparation of O-linked diaminopyrimidines. Reagents and conditions:
(a) iodoacetonitrile, K2CO3, DMF, 50 °C; (b) NaH, DME, Ethyl formate, 80 °C, MeI; (c)
quanidine carbonate, NaOMe, DMSO, 110 °C.
were also inactive. However, replacing the methylene linker of
ethyl arene 3 with oxygen gave a 10-fold boost in P2X3 activity.
Additionally, the cooperative effect of the oxygen linker with the
isopropyl side-chain of ether 16 led to an impressive increase in
P2X3 potency (18-fold increase). Molecular modeling studies sug-
gest that replacement of the bridging methylene by oxygen re-
stricts the rotation of the diaminopyrimidine by the formation of
an intramolecular hydrogen bond between it and the NH2 group
of the diaminopyrimidine.
Table 3
3-Methoxy group replacements lead to potent drug-like analogs
NH2
O
N
4
N
NH2
R2
The preparation of oxygen-linked analogs began with an alkyl-
ation of a 2-alkyl phenol (Scheme 2). This transformation was
accomplished in DMF with iodoacetonitrile in the presence of
K2CO3 at 50 °C. The resulting cyanomethylphenyl ether was treated
with NaH in DME and refluxed for 3 h in the presence of excess
ethyl formate. The ensuing unstable enol was alkylated with
methyl iodide to furnish an enol ether as a mixture of E/Z isomers
(typically 10:1). The final diaminopyrimidine was prepared by con-
densing the enol ether with guanidine carbonate in DMSO at
110 °C.
Having determined the optimal side-chain and linker we shifted
our attention to modifying the methoxy groups (Table 3). To deter-
mine the individual binding contributions of each methoxy group,
we prepared three deletion analogs: unsubstituted benzene 17, 3-
methoxy benzene 18 and 4-methoxy benzene 19 (entries 1–3).
Unsubstituted benzene 17 was almost completely inactive (P2X3
pIC50 = 5.4). But 3-methoxy arene 18 was 10-fold more potent
while 4-methoxy arene 19 was only fourfold more potent than
unsubstituted analog (17). Because of the differing contributions
on binding, we suspect that the two methoxy groups occupy sepa-
rate hydrophobic pockets.16
Because demethylation of the 4-methoxy group was the pri-
mary metabolite observed in this series, we attempted to find a
suitable methoxy group replacement (R1; Table 3, entries 4–7).
Although slightly less active, bromobenzene 20 and iodobenzene
21 were found to be suitable methoxy replacements. However, flu-
oro-phenylbenzene 6 and other small heterocyclic groups resulted
in a significant loss of activity. Furthermore, the highly electron-
withdrawing cyano benzene 23 was completely inactive at P2X3.
Thus the hydrophobic pocket occupied by substituents in the 4-po-
sition (R2) is small and tolerates groups that are lipophilic in
nature.
Based on the data listed in Table 3, we believed that optimiza-
tion of the 3-position (R1) could have a dramatic impact on po-
tency. We began with the preparation of simple alkyl analogs.
Small chain analogs such as toluene 24 and ethyl benzene 25,
which are similar in size to a methoxy group, were nearly equipo-
tent to dimethoxy benzene 16. However, incorporation of a small
electron-rich substituent in alkynyl benzene 28 afforded an
impressive boost in potency (P2X3 pIC50 = 8.2). In contrast, larger
alkyl replacements, such as isopropyl benzene 26 and biphenyl
27, were 10-fold less potent at P2X3.
3
R1
a,c
b,c
Entry
Compound
R1
H
OMe
H
OMe
OMe
R2
P2X3
P2X2/3
1
2
3
4
5
17
18
19
20
21
H
H
5.4
6.4
5.8
7.1
7.4
5.7
6.8
6.4
6.5
6.4
OMe
Br
I
6
7
22
23
OMe
OMe
5.6
5.1
F
<5.0
<5.0
N
8
9
10
11
24
25
26
27
Me
Et
i-Pr
Ph
OMe
OMe
OMe
OMe
7.0
7.3
6.3
6.0
6.4
6.8
5.7
5.6
12
28
OMe
8.2
7.9
13
14
15
16
29
30
31
32 (RO-4)
F
OMe
OMe
OMe
OMe
5.5
7.6
7.7
8.0
5.8
7.0
7.1
7.1
Cl
Br
I
17
18
33
34
OMe
OMe
6.2
7.0
5.6
6.0
N
O
S
O
O
19
20
21
22
35
36
37
38
OMe
OMe
OMe
OMe
7.0
5.8
6.5
6.2
<5.0
NH2
O
6.8
6.8
<5.0
OH
H
N
N
H
N
N
H
N
N
N
23
39
OMe
7.4
6.5
Since the lipophilic isopropyl group is a potential site of
metabolic oxidation, we decided to search for (R1) replacements
that would lower logP and thereby improve metabolic stability.
We began by preparing a series of four halogenated analogs (Table
3: entries 13–16). Fluoroanisole 29 was nearly inactive (P2X3
N
a
FLIPR: mean pIC50, rP2X3 CHO cell.
b
c
FLIPR: mean pIC50, hP2X2/3 1321N1 (astrocytoma) cells.
pIC50 values are the mean of at least three experiments performed in triplicates,
standard deviation 20%.

D. S. Carter et al. / Bioorg. Med. Chem. Lett. 19 (2009) 1628–1631
1631
5. Barclay, J.; Patel, S.; Dorn, G.; Wotherspoon, G.; Moffatt, S.; Eunson, L.; Abdel’al,
S.; Natt, F.; Hall, J.; Winter, J.; Bevan, S.; Wishart, W.; Fox, A.; Ganju, P. J.
Neurosci. 2002, 22, 8139.
6. Dorn, G.; Patel, S.; Wotherspoon, G.; Hemmings-Mieszczak, M.; Barclay, J.; Natt,
F. J. C.; Martin, P.; Bevan, S.; Fox, A.; Ganju, P.; Wishart, W.; Hall, J. Nucl. Acids
Res. 2004, 32, e49.
7. Virginio, C.; Robertson, G.; Surprenant, A.; North, R. A. Mol. Pharmacol. 1998, 53,
969.
8. Jung, K.; Moon, H. D.; Lee, G. E.; Lim, H.; Park, C.; Kim, Y. J. Med. Chem. 2007, 50,
4543.
9. Jarvis, M. F.; Burgard, E. C.; McGaraughty, S.; Honore, P.; Lynch, K.; Brennan, T.
J.; Subieta, A.; Van Biesen, T.; Cartmell, J.; Bianchi, B.; Niforatos, W.; Kage, K.;
Yu, H.; Mikusa, J.; Wismer, C. T.; Zhu, C. Z.; Chu, K.; Lee, C.; Stewart, A. O.;
Polakowski, J.; Cox, B. F.; Kowaluk, E.; Williams, M.; Sullivan, J.; Faltynek, C.
Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 17179.
10. Ito, K.; Iwami, A.; Katsura, H.; Ikeda, M. Naunyn-Schmiedeberg’s Arch. Pharmacol.
2008, 377, 483.
11. RO-4 is selective for P2X3/P2X2/3 over other P2X purinoceptors (P2X1, P2X2,
while the other larger halogens have a much greater propensity to
act as a hydrogen bond acceptor (more active).18 This is an impor-
tant finding because functional groups that have the ability to act
as hydrogen bond acceptors can also help to improve the PK prop-
erties by lowering logP and decreasing polar surface area.19
With results suggesting a possible hydrogen bond interaction
with the groups occupying the 3-position (R1), analogs that act as
hydrogen bond donors or acceptors were prepared (Table 3; en-
tries 18–23). These analogs contained neutral, basic and acidic
functional groups. Analogs that contained neutral polar functional
groups such as cyano benzene 33, sulfone 34 and carboxamide 35
were moderately active. The acidic functionality of carboxylic acid
36 was moderately active while bioisosteric tetrazole 39 was equi-
potent with dimethoxy benzene 16. Finally, moderately basic imid-
azole 37 was active (P2X3 pIC50 = 6.8) while the much more basic
dihydroimidazole 38 was completely inactive.
In conclusion, a high-throughput screening campaign identified
a diaminopyrimidine-based series of mixed P2X3/P2X2/3 antago-
nists. Optimization of this series resulted in an impressive boost
in potency due to the cooperative effect of an isopropyl side-chain
and an ether linker. Replacements for the potentially labile 4-
methoxy group were explored. Substitution of the 3-methoxy
group with various hydrogen bond acceptors yielded active ana-
logs with reduced lipophilicity. Finally, RO-4 (32) a potent drug-
like dual P2X3/P2X2/3 antagonist was selected for further
evaluation.
P2X4, P2X5, P2X7 IC50 > 10
lM). Also, the CEREP profile of RO-4 established
selectivity across a wide range of targets (weakly active at five receptors: ML1,
H2, 5HT2A, 5HT3, 5HT6 pKi = 1–6 lM) with no activity (>50% at 10 lM) for the
remaining receptors/enzymes.
12. (a) Lipinski, C. A.; Lombardo, B. W.; Dominy, P. J. F. Adv. Drug. Del. Rev. 1997, 46,
3; (b) This series meets all the criteria defined by Lipinski for drug-likeness. As
a result RO-4 exhibited good oral bioavailability (dog SDPK: CL 13 mL/min/kg,
72%F, T1/2 1.5 h, Caco-2 permeability: AB/BA = 1).
13. (a) Manchand, P. S.; Rosen, P.; Belica, P. S.; Oliva, G. V.; Perrotta, A. V.;
Wong, HS. J. Org. Chem. 1992, 57, 3531; (b) Manchand, P. S. US Patent
4,115,650, 1978.
14. Wada, A.; Yamamoto, J.; Hamaoka, K. O.; Nagai, S.; Kanamoto, K. J. Heterocycl.
Chem. 1990, 27, 1831.
15. (a) Flippin, L. A.; Carter, D. S.; Dupree, N. J. P. D. Tetrahedron Lett. 1993, 34,
3255; (b) Flippin, L. A.; Muchowski, J. M.; Carter, D. S. J. Org. Chem. 1993, 58,
2463.
16. Further evidence in support of two separate hydrophobic pockets comes from
the ground-state conformation of dimethoxybenzene 16 (both methoxy groups
should be oriented away from each other in the plane of the aromatic ring) and
from the 3,4-methylenedioxy and 3,4-ethylenedioxy analogs of benzene 16,
which were inactive.
17. (a) Dunitz, J. D.; Taylor, R. Chem. Eur. J. 1997, 3, 89; (b) Howard, J. A.; Hoy, V. J.;
O’Hagen, D.; Smith, G. T. Tetrahedron 1996, 52, 38.
18. (a) Kovács, A.; Varga, Z. Coord. Chem. Rev. 2006, 250, 710; (b) Sony, S. M. M.;
Ponnuswamy, M. N. Bull. Chem. Soc. Jpn. 2006, 79, 1766.
References and notes
1. Gever, J. R.; Cockayne, D. A.; Dillon, M. P.; Burnstock, G.; Ford, A. P. D. W.
Pfluegers Archiv. 2006, 452, 513.
2. Dubyak, G. R. Mol. Pharmacol. 2007, 72, 1402.
3. Novakovic, S. D.; Kassotakis, L. C.; Oglesby, I. B.; Smith, J. A. M.; Eglen, R. M.;
Ford, A. P. D. W.; Hunter, J. C. Pain 1999, 80, 273.
4. Cockayne, D. A.; Hamilton, S. G.; Zhu, Q.; Dunn, P. M.; Zhong, Y.; Novakovic, S.;
Malmberg, A. B.; Cain, G.; Berson, A.; Kassotakis, L.; Hedley, L.; Lachnit, W. G.;
Burnstock, G.; McMahon, S. B.; Ford, A. P. D. W. Nature 2000, 407, 1011.
19. (a) Refsgaard, H. H. F.; Jensen, B. F.; Guldbrandt, M.; Christensen, M. S. J. Med.
Chem. 2005, 48, 805; (b) Veber, D. F.; Johnson, S. R.; Cheng, H. Y.; Smith, B. R.;
Ward, K. W.; Kopple, K. D. J. Med. Chem. 2002, 45, 2615.