5-Phosphonomethylquinoxalinediones as competitive NMDA receptor antagonists with a preference for the human 1A/2A, rather than 1A/2B receptor composition

Article abstract of DOI:10.1016/S0960-894X(02)00074-4

NMDA antagonists derived from 5-phosphonomethyl-1,4-dihydroquinoxaline-2,3-dione (3a) are potent anticonvulsant agents, and display strong protective effects in the electroshock-induced convulsion assay in mice. Their preference for the human NMDAR 1A/2A over 1A/2B subunit composition was optimized, leading to (1RS,1′S)-PEAQX (9r), which shows a >100-fold selectivity.

Full text of DOI:10.1016/S0960-894X(02)00074-4

Bioorganic & Medicinal Chemistry Letters 12 (2002) 1099–1102
5-Phosphonomethylquinoxalinediones as Competitive NMDA
Receptor Antagonists with a Preference for the Human 1A/2A,
Rather than 1A/2B Receptor Composition
Yves P. Auberson,* Hans Allgeier, Serge Bischoff, Kurt Lingenhoehl, Robert Morettiy
and Markus Schmutz
Novartis Pharma AG, Klybeckstrasse 141, 4002 Basel, Switzerland
Received 12 December 2001; accepted 23 January 2002
Abstract—NMDA antagonists derived from5-phosphonomethyl-1,4-dihydroquinoxaline-2,3-dione ( 3a) are potent anticonvulsant
agents, and display strong protective effects in the electroshock-induced convulsion assay in mice. Their preference for the human
NMDAR 1A/2A over 1A/2B subunit composition was optimized, leading to (1RS,10S)-PEAQX (9r), which shows a >100-fold
selectivity. # 2002 Elsevier Science Ltd. All rights reserved.
In the course of our studies on substituted 5-methyl-
quinoxalinediones, we identified 3a as a potent, achiral,
orally active NMDA antagonist with a 15-fold pref-
erence for human NMDA receptors with the 1A/2A,
rather than 1A/2B, subunit composition. As no selective
NMDA 1A/2A receptor antagonist has been described
to date, we were interested in improving this preference,
measured by electrophysiological recordings in Xenopus
laevis oocytes expressing either 1A/2A or 1A/2B het-
eromeric human NMDA (hNMDA) receptors.1 Three
series of 5-phosphonomethylquinoxalinediones were
synthesized for this purpose (Fig. 1), ultimately leading
to 9r ((1RS,10S)-PEAQX), which shows a >100-fold
preference for hNMDA 1A/2A receptors (Fig. 2).
eromeric hNMDA 1A/2A or 1A/2B receptors, and their
anticonvulsant properties in the mouse electroshock-
induced seizure test.
Chemistry
a-Unsubstituted 5-phosphonomethylquinoxalinediones
were obtained from5-bromomethylquinoxalines, for
example 13a (Scheme 1), which reacted with trimethyl-
phosphite to yield dimethyl phosphonate 2, and after
acidic hydrolysis 5-phosphonomethylquinoxalinedione
3d. Alternatively, hydrogenolysis or replacement of the
bromine atom in a Suzuki reaction yielded, after
hydrolysis, 3a or 3e. The 7-chloro and 7-fluoro analo-
gues (3b and 3c) were prepared fromthe corresponding
bromomethylquinoxalines.3b,c
In contrast to previously described competitive NMDA
antagonists,2 compounds of the first series show that
chirality is not required for subnanomolar affinity at
NMDA receptors. The addition of a chiral side chain,
however, markedly improves the distinction betweeen
hNMDA receptors composed of 1A/2A or 1A/2B sub-
units.
a-Hydroxy-, methoxy- or acetoxy-5-phosphonomethyl-
quinoxalinediones were obtained from5-carbaldehydes
(Scheme 2). For instance, 43b reacted with phosphinic
acid dimethyl trimethylsilyl ester4 to yield a-hydroxy-
phosphonate 5b, which was deprotected to quinox-
alinedione 6b in concentrated HCl, or methylated before
deprotection to yield 6c. Acylation of 6b with acetic
anhydride yielded 6d.
This article describes the binding affinity of these novel
5-phosphonomethylquinoxalinediones, as well as their
functional selectivity in Xenopus oocytes expressing het-
a-Amino-5-phosphonomethylquinoxalinediones were
obtained by addition of phosphinic acid dimethyl tri-
methylsilyl ester to an imine, as exemplified in Scheme 3
for aldehyde 7 and methylamine, yielding 8b. This
*Corresponding author. Tel.: +41-61-696-8404; fax: +41-61-696-
8676; e-mail: yves.auberson@pharma.novartis.com
yPresent address: Firmenich SA, CP 239, rte des Jeunes 1, 1211 Geneve
8, Switzerland.
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00074-4

1100
Y. P. Auberson et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1099–1102
Figure 1. 5-Phosphonomethyl-1,4-dihydroquinoxaline-2,3-diones.
Scheme 2. a-Hydroxy-5-phosphonomethylquinoxalinediones. Reagents
and conditions: (a) (MeO)2PHO, TMSCl, Et3N, CH2Cl2, 0 ꢀC, 1 h, then
4, rt, 4 h, 56%; (b) NaH, THF, then Me2SO4, 42%; (c) 2 N HCl, 100 ꢀC,
18 h, 33–100%; (d) Ac2O, 70 ꢀC, 18 h, 46%.
Figure 2. Effect of 9r ((1RS,10S)-PEAQX) on NMDA-induced cur-
rents in Xenopus oocytes expressing hNMDAR 1A/2A or 1A/2B. Data
show mean valueꢁSEM (N=3 or more).
Scheme 1. 5-Phosphonomethylquinoxalinediones. Reagents and con-
ditions: (a) P(OR)3, reflux, 98% (3a: R=Me; 3e: R=Et); (b) H2, 10%
Pd/C, Et3N, MeOH; (c) AcOH, 12 N HCl, reflux, 20 h, 64%; (d)
p-ClPhB(OH)2, (Ph3P)4Pd, Na2CO3, MeOCH2CH2OMe, H2O, 24 h,
reflux, quant; (e) (i) TMSBr, CH2Cl2 then EtOH; (ii) HBr/AcOH,
66%.
Scheme 3. a-Aminoxy-5-phosphonomethylquinoxalinediones. Reagents
and conditions: (a) (i) MeNH2 HCl, MgSO4, K2CO3, CH2Cl2; (ii)
.
(MeO)2PHO, TMSCl, Et3N, 0 ꢀC to rt, 20 h; (b) BnBr, MeCN, iPr2Net,
60%; (c) Ac2O, quant; (d) 4 N HCl, reflux, 18 h, or 33% HBr in
AcOH, 40 ꢀC, 12 h, 75–95%; (e) (MeO)2PHO, MeOOCNH2, AcOH,
SOCl2, 86%; (f) 4 N HCl, 100 ꢀC, 20 h, 44%; (g) 4 N HCl, 100 ꢀC, 24 h,
68%.
intermediate can be benzylated (8j) or acylated (8c)
before deprotection under acidic conditions. Primary
a-amino phosphonates were prepared by condensation
with methyl carbamate and dimethylphosphite,5 which
gave with aldehyde 4 the protected aminophosphonate
10 in good yield. A partial deprotection was obtained
after a 20 h treatment with 4 N HCl at 100 ꢀC, leaving
the methylcarbamate mostly intact (9e). A prolonged
treatment led to the fully deprotected a-aminophos-
phonate 9a.
Finally, 6-phosphonomethylquinoxalinedione (15) was
obtained by reaction of ethyl oxalyl chloride on diethyl
p-aminobenzylphosphonate, followed by nitration and
reductive cyclization (Scheme 4).
All new compounds showed satisfactory 1H NMR
(400 MHz) and MS analyses.
The methyl- and benzylphosphinates 11a–c (Table 1)
were prepared like the corresponding phosphonates,
using benzyl- or methylphosphinic acid ethyl ester in
place of dimethylphosphite. 5-Carboxymethyl-1,4-dihy-
droquinoxaline-2,3-dione (12) was synthesized from 1 in
15% yield by reaction with Bu4NCN in CH2Cl2/H2O at
roomtemperature, hydrolyzed and deprotection under
acidic conditions (AcOH/H2SO4/H2O 1:1:1), followed
by catalytic debromination on 10% Pd/C in the pre-
sence of triethylamine.
Scheme 4. 6-Phosphonomethylquinoxalinedione. Reagents and con-
ditions: (a) EtOOC–COCl, pyridine, 37%; (b) NH4NO3, (CF3CO)2O,
88%; (c) TiCl3, acetone, water.

Y. P. Auberson et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1099–1102
1101
Table 1. In vitro affinities, functional selectivity in Xenopus oocytes and anticonvulsant effect in the MES test
Compd
X
Y
CGP 39653a
Oocytes: hNMDARb 1A/2A 1A/2B ratio
MESc ED50
3a
3b
3c
3d
3e
H
F
H
H
H
H
H
0.5
1.0
1.7
1.5
0.8
2
30
—
—
—
1.8
1:15
0.7
n.t.
2.4
2.5
0.7
—
—
—
0.2
—
—
Cl
Br
—
1:9
p-ClPh
6a
6b
6c
6d
H
HO
HO
1.5
1.2
17
1.1
—
—
—
0.7
—
—
—
1:1
—
—
—
0.5
1.2
n.t.
5
NO2
NO2
NO2
MeO
AcO
8.1
9ae
9b
NO2
H
H2N
MeNH
40
1100
860
65
—
790
2800
90
—
6200
24,000
930
—
1:8
11.6
n.t.
n.t.
2.8
9c
H
Me(Ac)N
PrNH
MeOCONH
1:9
9de
9e
NO2
NO2
1:10
1:13
370
110
1400
8.8
9fd
NO2
4.1
—
—
—
3.7
9gd
9hd
H
NO2
BnNH
BnNH
2
1.5
4.2
—
34
—
1:8
—
2
1.5
9ie
NO2
H
3
—
—
—
4
9j
160
600
30,000
1:50
40%
9k
9l
H
H
PhNH
PhCH2CH2NH
130
7.4
490
40
6200
1600
1:8
1:40
0%
10
9m
H
38
170
1500
1:9
2.3
9n
9o
9p
9q
9r
H
H
H
H
H
31
9.7
120
27
50
18
570
900
1:11
1:50
60%
6.7
2
60
820
1:14
1700
270
>30,000
29,600
>1:17
1:111
15
8
23
11ad
11bd
11c
NO2
NO2
NO2
BnNH
BnNH
HO
60
80
50
200
—
1300
1300
—
1:26
1:7
—
20%
0%
42%
0%
aIC50 (nM) or % inhibition at 1 mM in the [3H]CGP39653 binding assay,6 average of at least two experiments run in triplicate.
bIC50 (nM) based on at least six concentrations of test compound; n=3 or more.
cMouse electroshock-induced seizure test, ED50 (mg/kg) or % protection at 50 mg/kg, 1 h after ip administration; n=5.
dHBr salt.
eHCl salt.

1102
Y. P. Auberson et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1099–1102
Biological Activity
The in vivo potency in the MES test appears to be
independent of the affinity for hNMDA receptors com-
posed of 1A/2B subunits. For instance, 3a and 6a,
which have similar affinities in the radioligand binding
assay, display identical potencies in the MES test,
despite very different IC50 values at hNMDA 1A/2B
receptors. The most selective compound, (1RS,10S)-
PEAQX (9r), which is practically inactive in Xenopus
oocytes expressing hNMDA 1A/2B receptors, displays
an ED50 value of 23 mg/kg in the MES test.
The parent structure, 5-phosphonomethylquinoxaline-
dione (3a), is a highly potent antagonist with sub-
nanomolar affinity in the [3H]CGP39653 binding
assay6a for NMDA receptors. In vivo, it shows a very
long duration of action, with an ED50 value in the
mouse electroshock-induced seizure test (MES)7 of 1.1
mg/kg, 8 h after ip administration, and 6.7 mg/kg, 8 h
after oral administration. Variation of the aromatic
substitution on C(7) (3b–e) shows that this position is
largely insensitive to steric as well as to electronic
effects, a pattern not seen with quinoxalinediones acting
at AMPA receptors3a,e or at the glycine-binding site of
NMDA receptors.3d,e Replacement of the phosphonic
acid by a carboxylate (12) strongly decreased in vitro
affinity (IC50=250 nM), and abolished activity in the
electroshock-induced convulsion model in mice up to 50
mg/kg (ip, 1 h). Of note, 6-phosphonomethyl-quinox-
alinedione (15), was completely inactive at hNMDA
receptors. Compounds in Table 1 have no binding affi-
nity for AMPA receptors,8 and at least a 250-fold
selectivity with regards to the glycine-binding site of
NMDA receptors9 (data not shown).
In contrast to their phosphonate analogues, phosphi-
nates (11a–c) have weaker binding affinity to hNMDA
receptors. 11a shows a 26-fold preference for the 1A/2A
subunit composition, and is like 11b and 11c inactive in
the MES test.
Acknowledgements
The authors would like to thank C. Boesch, R. Boesch
and P. Schmid for skillful synthetic support. Thanks
also to P. Martin and N. Reymann for running the
radioligand binding assays, to C. Portet and A. Jeker
for the anticonvulsion tests, and to R. Bromfor the
oocyte work.
Introduction of a hydroxy group in the benzylic
position (6a,b) maintains both affinity for hNMDA
receptors and in vivo potency, with O-acylation (6d)
or O-methylation (6c) having only minor effects on
affinity. Compound 6a does not differentiate between
hNMDA 1A/2A and 1A/2B receptors subunit compo-
sitions.
References and Notes
1. Hess, S. D.; Daggett, L. P.; Crona, J.; Deal, C.; Lu, C.-C.;
Urrutia, A.; Chavez-Noriega, L.; Ellis, S. B.; Johnson, E. C.;
Velicelebi, G. J. Pharmacol. Exp. Ther. 1996, 278, 808.
2. (a) See, for example: (a) Herrling, P. L.; Emre, M.; Wat-
kins, J. C. In Excitatory Amino Acids: Clinical Results with
Antagonists; P. L. Herrling, Ed.; Academic: London, 1997; pp
7 and 129. (b) Lehmann, J.; Hutchison, A. J.; McPherson,
S. E.; Mondadori, C.; Schmutz, M.; Sinton, C. M.; Tsai, C.;
Murphy, D. E.; Steel, D. J.; Williams, M.; Cheney, D. L.;
Wood, P. L. J. Pharmacol. Exp. Ther. 1988, 246, 65. (c) Orn-
stein, P.; Monn, J. A.; Schoepp, D. D. Drugs News Persp.
1994, 7, 5.
In the a-aminophosphonic acid series, N-substitution by
a relatively bulky group like benzyl (9g) or cyclohexyl-
methyl (9i) appears necessary to reach low nanomolar
binding affinity. An additional methyl group on the
nitrogen strongly decreased in vitro and in vivo potency
(9j), while allowing a 50-fold discrimination between
hNMDA 1A/2A and 1A/2B subunit combinations.
3. (a) Auberson, Y. P.; Bischoff, S.; Moretti, R.; Schmutz, M.;
Veenstra, S. J. Bioorg. Med. Chem. Lett. 1998, 8, 65. (b)
Auberson, Y. P.; Acklin, P.; Allgeier, H.; Biollaz, M.; Bischoff,
S.; Ofner, S.; Veenstra, S. J. Bioorg. Med. Chem. Lett. 1998, 8,
71. (c) Acklin, P.; Allgeier, H.; Auberson, Y. P.; Ofner, S.;
Veenstra, S. J. WO 9817672 A1, 1998; Chem. Abstr. 1998, 128,
321753. (d) Acklin, P.; Allgeier, H.; Auberson, Y. P.; Bischoff,
S.; Ofner, S.; Sauer, D.; Schmutz, M. Bioorg. Med. Chem.
Lett. 1998, 8, 493. (e) Auberson, Y. P.; Acklin, P.; Bischoff, S.;
Moretti, R.; Ofner, S.; Schmutz, M.; Veenstra, S. J. Bioorg.
Med. Chem. Lett. 1999, 9, 249.
In terms of receptor affinity, the position of the phenyl
ring seems optimal in the benzylamine derivative 9g,
which is more potent than the aniline (9k) and phe-
nethyl analogues (9l). Introduction of an additional
alkyl group in the benzylic position decreased affinity 5-
to 15-fold (9m–o).
In terms of subunit selectivity however, introduction of
a methyl group in this position has a significant effect.
The (R)-phenethyl derivative (9n) shows no increase in
selectivity as compared to 9g, but the (S)-phenethyl ste-
4. Afarinkia, K.; Rees, C. W.; Cadogan, J. I. G. Tetrahedron
1990, 46, 7175.
reoisomer (9o) displays
a
50-fold preference for
5. Yuan, C. Y.; Wang, G. H.; Chen, S. J. Synthesis—Stuttgart
1990, 6, 522.
6. Sills, M. A.; Fagg, G.; Angst, C.; Pozza, M. F.; Brundish,
D. E.; Hurt, S. D.; Wilusz, E. J.; Williams, M. Eur. J. Phar-
macol. 1991, 192, 19.
7. Schmutz, M.; Portet, C.; Jeker, A.; Klebs, K.; Vassout, A.;
Allgeier, H.; Heckendorn, R.; Fagg, G. E.; Olpe, H. R.; van Rie-
zen, H. Naunyn-Schmiedeberg’s Arch. Pharmacol. 1990, 342, 61.
8. Honore, T.; Lauridsen, J.; Krogsgaard-Larsen, P. J. Neu-
´
rochem. 1982, 38, 173.
hNMDA receptors composed of the 1A/2A subunit
combination. Finally, introduction of a bromine atom
in the para position of the phenethyl group led to 9r
[(1RS,10S)-PEAQX], which has a high binding affinity
for NMDA receptors (IC50=8 nM), and a functional
preference in excess of 100-fold for hNMDA 1A/2A
(IC50=of 270 nM) over 1A/2B receptors (IC50=29,600,
Fig. 2). The separation of the (RS)- and (RR)-diastere-
omers and their pharmacological properties will be
described in detail elsewhere.
9. Baron, B. M.; Siegel, B. W.; Harrison, B. L.; Gross, R. S.;
Hawes, C.; Towers, P. J. Parmacol. Exp. Ther. 1996, 279, 62.