Imidazo[1,2-a]pyrimidines as functionally selective GABAA ligands

Article abstract of DOI:10.1016/j.bmcl.2005.11.112

Imidazo[1,2-a]pyrimidines are GABA_A receptor benzodiazepine binding site ligands which can exhibit functional selectivity for the α₃ subtype over the α₁ subtype. SAR studies to optimize this functional selectivity are desc

Full text of DOI:10.1016/j.bmcl.2005.11.112

Bioorganic & Medicinal Chemistry Letters 16 (2006) 1175–1179
Imidazo[1,2-a]pyrimidines as functionally selective GABAA ligands
*
´
Wesley P. Blackaby, John R. Atack, Frances Bromidge, Jose L. Castro,
Simon C. Goodacre, David J. Hallett, Richard T. Lewis, George R. Marshall,
Andrew Pike, Alison J. Smith, Leslie J. Street, David F. D. Tattersall and Keith A. Wafford
Neuroscience Research Centre, Merck Sharp and Dohme Research Laboratories, Terlings Park,
Eastwick Road, Harlow, Essex CM20 2QR, UK
Received 25 October 2005; revised 23 November 2005; accepted 24 November 2005
Available online 6 January 2006
Abstract—Imidazo[1,2-a]pyrimidines are GABAA receptor benzodiazepine binding site ligands which can exhibit functional selec-
tivity for the a3 subtype over the a1 subtype. SAR studies to optimize this functional selectivity are described.
ꢀ 2006 Elsevier Ltd. All rights reserved.
c-Aminobutyric acid (GABA) is the major inhibitory
neurotransmitter in the central nervous system. It acts
at the GABAA receptor, a pentameric supramolecular
complex that acts as a ligand-gated chloride ion channel.
A functional receptor is formed by the co-assembly of
subunits selected from the family of 19 gene products
(a1–6, b1–3, c1–3, d, e, p, h, and q1–3), which are differen-
tially expressed throughout the brain.1,2 The most abun-
dant GABAA receptor subtypes contain a, b, and c
subunits in a 2:2:1 stoichiometry. The benzodiazepine
class of drugs,3,4 widely used as therapy for anxiety
and panic disorders, are generally non-selective agonists
in that they allosterically modulate the GABA-mediated
chloride ion flux through the channel of GABAA recep-
tors containing b, c2, and either a1, a2, a3 or a5 sub-
units.1,2 However, the side effect profile associated with
the classical benzodiazepines, such as their sedative
properties, is far from ideal.5 Studies with transgenic
mice and with subtype selective compounds suggest that
a1-containing receptors are responsible for mediating
the sedative/muscle relaxant properties of benzodiaze-
pines and that a3- and/or a2-containing receptors are
important for anxiety.6 The goal of our research was
to identify subtype selective ligands, be that functionally
selective or binding selective, with the expectation of
gaining an improved side effect profile over currently
used benzodiazepines.7
Benzimidazole 1 (NS-2710)8 (Fig. 1) exhibits anxiolytic
activity in several animal behavioral models with an im-
proved side effect profile. We attributed this profile to
the functional selectivity of this compound, which in
our hands showed full efficacy at a3 and a2-containing
receptors (comparable with the full agonist chlorodia-
zepoxide [CDZ]) and somewhat reduced efficacy at the
a1 subtype. To test our hypothesis the ideal compound
would be an antagonist at a1 with significant efficacy
at a3- and/or a2-containing receptors. We have recently
disclosed the details of SAR studies around 1 which led
to the discovery of imidazopyrimidine 2.9 This paper
describes the SAR of a related series of compounds in
which we have examined heterocyclic replacements for
the biaryl moiety (rings A and B).
Attempts to prepare an imidazopyrimidine 3-zincate
species (via lithium halogen exchange and transmetalla-
tion with zinc chloride) or 3-boronate ester (via palladi-
um catalysis10) from bromide 411 proved unsuccessful.
However, as shown in Scheme 1 bromide 4 cleanly
underwent magnesium bromine exchange12 using
i-PrMgCl and the resulting Grignard quenched with
tributyltin chloride to give the key stannane intermedi-
ate 5. Although the stannane could be chromatographed
on silica gel (using aprotic solvents as eluent) it was
more conveniently kept as a THF stock solution which
was stable for extended periods when stored in the
refrigerator. The stannane solution was subsequently
used to couple to a variety of heterocyclic halides under
standard Stille conditions to give the imidazopyrimi-
dines 3a–h. Best yields were obtained with heteroaryl
bromides or 2-chloropyridines. 4-Chloropyridines were
Keywords: Imidazo[1,2-a]pyrimidines; GABAA receptor ligands.
*
Corresponding author. Tel.: +44 1279 440404; fax: +44 1279
440187; e-mail: Wesley_Blackaby@Merck.com
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.11.112

1176
W. P. Blackaby et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1175–1179
EtO
F3C
N
N
N
N
N
N
F3C
N
5
N
NC
2
N
6
3
Het
N
1
2
3
Ring A
NS-2710
Ring B
Figure 1.
iii
F3C
N
N
F3C
N
i, ii
F3C
N
during the Suzuki–Miyaura coupling reaction of bro-
mide 4 with boronic acids.
N
N
N
N
N
SnBu3
Br
Het
3a-h
4
5
see Table 1
45-75% from 4
Binding and efficacy data for the imidazopyrimidines 3
are shown in Table 1. Although the parent pyridin-2-yl
compound 3a shows similar affinity to 1 at a3-containing
receptors, additional substitution at the 6-position as in
compounds 3b–e gives a marked increase in affinity. The
pyrimidine analogs were more sensitive to substitution
effects, with the 6-chloro compound 3g losing affinity,
whereas the 2-chloropyrimidine 3h gave high affinity.
In our high-throughput functional efficacy assay based
Scheme 1. Reagents and conditions: (i) i-PrMgCl, THF, À78 ꢁC;
(ii) Bu3SnCl, À78 ꢁC to rt; (iii) aryl halide, Pd(PPh3)4, reflux.
poorer coupling partners requiring longer reaction
times. The mild Stille conditions negated the formation
of products arising from Dimroth rearrangement13 of
the imidazopyrimidine core which has been observed
Table 1. Binding affinity and efficacy for imidazo[1,2-a]pyrimidines at GABAA receptor subtypes
Compound
Het
Kia (nM)
Efficacyb
a1
a3
a1
a3
1
—
1.9
13
[0.4]
[0.99]
N
N
N
N
3a
b
9.1
0.31
0.31
1.2
11
À0.4
0.11
À0.07
0.02
0.68
0.88
1.7
Br
c
À0.24
À0.4
CF3
OMe
d
À0.22
À0.09
[À41.4%]
[À51.6%]
NC
N
e
f
0.36
>33
>33
0.93
0.46
>33
>33
2.0
À0.08
—
0.1
—
—
N
N
Cl
Cl
Cl
g
h
—
N
N
N
À0.22
À0.19
[À24.9%]
a Affinity was determined by the inhibition of [3H]Ro 15-1788 (flumazenil) binding to human recombinant GABAA receptors containing b3c2 plus
either a1, a3 or a5 stably expressed in L(tkÀ) cells. Values are means of 3–10 separate determinations.15
b Modulation of chloride ion flux in cells expressing b3c2 plus either a1 or a3 produced by an EC20 equivalent concentration of GABA in the presence
of an approximate 1000 · Ki concentration of test compound. Efficacy is expressed relative to the full agonist chlorodiazepoxide (relative effica-
cy = 1.0). Values are means of at least seven independent experiments.16 Negative values reflect a reduction in the GABA EC20 were induced
chloride flux and reflect inverse agonist responses.
Efficacy data given in square brackets [] were measured at GABAA receptors stably expressed in L(tkÀ) cells using whole cell patch-clamp recording
and represent the effect of the test compound on the current produced by an EC20-equivalent of GABA relative to the full agonist chlorodiazepoxide
(relative efficacy = 1.0).17 Values shown as negative percentages reflect the fact that the GABA EC20 currents were attenuated rather than potentiated
(i.e., responses were inverse agonist rather than agonist responses).

W. P. Blackaby et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1175–1179
1177
on chloride ion flux,13 all the 2-aza heterocycles showed
low efficacy at cells containing a3b3c2 receptors. For
compounds 3d and 3h patch-clamp measurements on
L(tkÀ) cells expressing a1- and a3-containing GABAA
receptors confirm them to be unselective inverse agonists
compared with the highly efficacious benzimidazole 1. In
the pentylenetetrazole (PTZ) assay carried out in Swiss
Webster (SW) mice, compound 3h was proconvulsant
when dosed at 10 mg/kg i.p. (which is indicative of
inverse agonists at the benzodiazepine binding site14).
tions 2,4-dichloropyrimidine gave a 1:2 mixture of the 2-
coupled isomer (9) and 4-isomer (8) which, fortuitously,
were readily separable by flash chromatography on sili-
ca. The heterocyclic halides were then coupled to stann-
ane 5 as previously described. Pyridine 10e was available
via a Suzuki coupling of 3f in refluxing THF, use of
higher temperatures or microwave irradiation led to
significant amounts of Dimroth products.
The data shown in Table 2 show that a nitrogen for car-
bon switch is well tolerated at the 2 and 4 positions of
ring A; pyridines 3e, 10b and pyrimidine 10d show excel-
lent binding affinity at a1-, a3-, and a5-containing sub-
types. Nitrogen at position 6 was not tolerated in
either the pyridine (10e) or pyrimidine (10c) case. The
position of the nitrogen also has a profound effect in
terms of efficacy. Whereas compound 3e shows low effi-
cacy at both a1 and a3 receptors, compound 10b shows
excellent functional selectivity with low efficacy at
a1-containing receptors and moderate efficacy at the
To examine the effect upon efficacy in biaryl-containing
system 2 (Fig. 1) more closely, ring B was kept constant
as 2-cyanophenyl whilst carbon was switched for nitro-
gen around ring A. The required halides were readily
prepared by Suzuki coupling of the pyridine or pyrimi-
dine dihalide with 2-cyanophenyl boronic acid as shown
in Scheme 2. Coupling with 2,4-dichloropyridine took
place predominantly at the 2-position as expected based
on literature precedent18 to give 7. Under similar condi-
Cl
Br
Cl
i
Br
Br
i
CN
CN
N
62%
54%
N
N
N
N
Cl
7
6
Cl
N
Cl
Cl
N
i
ii
N
CN
N
N
CN
+
6,7,8,9
10a-d
85%
see Table 2
24-72% from 4
Cl
8
9
F3C
N
N
F3C
N
N
N
i
N
NC
35%
N
3f
10e
N
Cl
Scheme 2. Reagents and conditions: (i) 2-Cyanophenylboronic acid, Pd(PPh3)4, K2CO3 (aq), THF, reflux; (ii) 5, Pd(PPh3)4, THF reflux.
Table 2. Binding affinity, efficacy and DLM turnover data for imidazo[1,2-a]pyrimidines
F3C
N
N
N
6
NC
2
N
10a-e
5
4
Compound
N position
Kia (nM)
Efficacyb
Turnoverc % DLM
a1
a3
a1
a3
2
—
2
0.66
0.33
0.10
[0.19]
À0.08
0.53
[0.59]
0.1
100
15
3e
0.36
0.46
[À22.0%]
À0.39
0.43
10a
10b
5
4
18
2.5
>33
2.9
À0.55
0.04
[0.15]
—
—
4
[0.45]
—
10c
10d
10e
2.6
2.4
6
>33
1.0
>33
1.4
—
56
—
À0.29
—
À0.15
—
>33
>33
a As for Table 1.
b As for Table 1.
c Figure represents % of test compound metabolized when incubated with dog liver miocrosomes [DLM] at a drug concentration of 1 lM and a
protein concentration of 0.4 mg/ml for 15 min at 37 ꢁC.

1178
W. P. Blackaby et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1175–1179
Table 3. Binding affinity and efficacy for imidazo[1,2-a]pyrimidines at
GABAA receptor subtypes
a3 subtype. The stability of the compounds was also
examined in dog liver microsome [DLM] preparations.
Gratifyingly the turnover for the N-4 nitrogen com-
pound 10b was dramatically reduced compared with
the parent biphenyl compound 2, suggesting that plasma
clearance might also be reduced by this modification.
F3C
N
N
N
11a-e
R
N
To achieve the required antagonism at a1-containing
receptors, our attention shifted to ring B modification
whilst keeping the functionally selective pyridin-4-yl
(ring A) of compound 10b constant. As shown in
Scheme 3, chloropyridine 13 was prepared as a late stage
intermediate. Coupling of 4-bromo-2-methoxypyridine
with stannane 5 proceeded smoothly to give 11a. Hydro-
lysis with TMS iodide followed by chlorination with
thionyl chloride gave 13. Suzuki couplings with com-
pound 13 were surprisingly sluggish and the more forc-
ing conditions and lengthy reaction times required led to
substantial amounts of Dimroth by-products, which
were difficult to separate. Although fluorophenyl com-
pound 11b was accessed via this route, in general it
was easier to prepare the biaryl coupling precursor. Su-
zuki coupling of 2-chloro-4-methoxypyridine with 3-flu-
orophenylboronic acid under Suzuki conditions gave 14,
which on treatment with PBr3 gave the coupling precur-
sor bromide 15. Suzuki coupling of Boc-protected
2-chloro-4-aminopyridine followed by TFA treatment
gave aminopyridine 16. Diazotization in concentrated
HBr then gave bromide 17, the coupling precursor.
Chloropyridine 18 was accessed from 2,4-dichloropyri-
dine as described for compound 10b above. The biaryl
halide precursors were then coupled to stannane 5.
Compound
R
Kia (nM)
Efficacyb
a1
a3
a1
a3
CN
10b
2.5
2.9
0.04
[0.15]
0.43
[0.45]
11a
11b
OMe
29
>33
>33
—
—
—
—
F
>33
OMe
11c
5.8
6.1
0.35
0.54
CN
F
11d
11e
0.99
8.3
2.2
14
0.09
[0.0]
0.21
[0.22]
F
0.06
0.23
a As for Table 1.
b As for Table 1.
11d retained good affinity and the desired efficacy profile
being an antagonist at a1-containing receptors and a par-
tial agonist at a3 receptors. The stability of compound
11d was also examined in dog and human liver micro-
some preparations; less than 5% turnover was observed
in both species, suggesting that plasma clearance might
be low for this compound. Compound 11d was then pro-
gressed into our in vivo occupancy assay in rat which
As shown in Table 3 replacement of the cyanophenyl
with methoxy (11a) (cf. 6-methoxypyridine 3d) or 3-flu-
orophenyl (11b) is not well tolerated both modifications
losing an order of magnitude in affinity. 2-Methoxyphe-
nyl (11c) and 2-fluorophenyl (11e) were better tolerated
in terms of affinity and also retained functional selectiv-
ity for a3-containing receptors over a1-containing recep-
tors. However, the 2-cyano-4-fluorophenyl compound
F3C
N
F3C
N
N
N
Br
N
i
N
N
iv
14%
F
86%
11a R=OMe
12 R=OH
13 R=Cl
ii 65%
iii 92%
OMe
R
N
N
11b
R
R
N
OMe
N
NBoc2
v
v, vii
OMe
F
67%
N
Cl
N
Cl
16 R=NH2
17 R=Br
14 R=OMe
viii 71%
vi 52%
15 R=Br
Cl
Cl
N
i
v
CN
15, 17, 18
11c-e
see Table 3
21-88% from 4
38%
N
Cl
18
F
Scheme 3. Reagents and conditions: (i) 5, Pd(PPh3)4, THF, reflux; (ii) TMSI, DCM; (iii) thionyl chloride; (iv) ArB(OH)2, PdCl2(dppf), Na2CO3
dioxane, 50 ꢁC; (v) ArB(OH)2, PdCl2(dppf), Na2CO3 dioxane, 80 ꢁC; (vi) PBr3; (vii) TFA, DCM; (viii) NaNO2, Cu(I)Br, HBr.

W. P. Blackaby et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1175–1179
1179
uses displacement of [3H]Ro 15-1788 to measure the ex-
tent to which the compound penetrates the CNS and
occupies GABAA receptor benzodiazepine binding sites.
When dosed at 1 mg/kg po as a suspension in 0.5%
methocel, 11d gave 50% receptor occupancy with a plas-
ma Occ50 of 0.15 lM and brain to plasma ratio of 0.73.
G.; Macaulay, A.; Brown, N.; Howell, O.; Moore, K. W.;
Carling, R. W.; Street, L. J.; Castro, J. L.; Ragan, C. I.;
Dawson, G. R.; Whiting, P. J. Nat. Neurosci. 2000, 3, 587.
7. Atack, J. R. Expert Opin. Invest. Drugs 2005, 14, 601.
8. Teuber, L.; Watjen, F.; Fukuda, Y.; Ichimaru, Y.
WO Patent 99/19323, 1999.
9. Hallett, D. J.; Atack, J. R.; Castro, J. L.; Cook, S.;
Carling, R. W.; Dias, R.; Dawson, G. R.; Goodacre, S.;
Humphries, A.; Kelly, S.; Lincoln, R. J.; Marshall, G. R.;
McKernan, R. M.; Street, L. J.; Reynolds, D. S.; Russell,
M. G. N.; Smith, A.; Sheppard, W. F. A.; Stanley, J. L.;
Tye, S.; Wafford, K. A.; Whiting, P. J. Abstracts of Papers,
230th National Meeting of the American Chemical Soci-
ety, Washington, DC, 2005.
10. Ishiyama, T.; Murata, M.; Miyaura, N. J. Org. Chem.
1995, 60, 7508.
11. Blackaby, W. P.; Castro Pineiro, J. L.; Chambers, M. S.;
Goodacre, S. C.; Hallett, D. J.; Jones, P.; Lewis, R. T.;
MacLeod, A. M.; Maxey, R. J.; Moore, K. W.; Street, L.
J. WO Patent 02/076983, 2002.
In conclusion, the SAR of heterocyclic replacements for
the biaryl moiety of 2 has been investigated. Introducing
nitrogen into ring A of compound 2 has a profound ef-
fect on affinity, efficacy and metabolic stability. 6-Substi-
tuted pyridin-2-yls such as 3b–e showed reduced efficacy
at a3 receptors. Pyridines with the nitrogen at the 4-po-
sition such as 10b, 11c, and 11e showed higher efficacy at
a3 and functional selectivity over a1 receptors. Modifica-
tion of rings A and B in tandem enabled further modu-
lation of the efficacy profile which led to compound 11d
with high affinity at GABAA receptors and an attractive
efficacy profile, being an antagonist at a1b3c2 receptors
and a partial agonist at a3b3c2 receptors which exhibits
good receptor occupancy in the rat on oral dosing.
12. Abarbri, M.; Dehmel, F.; Knochel, P. Tetrahedron Lett.
1999, 40, 7449.
13. Guerret, P.; Jacquier, R.; Maury, G. Bull. Soc. Chim. Fr.
1972, 9, 3503.
14. Kubova, H.; Mikulecka, A.; Haugvicova, R.; Mares, P.
Naunyn-Schmiedeberg’s Arch. Pharmacol. 1999, 360,
565.
References and notes
1. Simon, J.; Wakimoto, H.; Fujita, N.; Lalande, M.;
Barnard, E. A. J. Biol. Chem. 2004, 279, 41422.
2. Sieghart, W. Pharmacol. Rev. 1995, 47, 181.
3. Bowery, G., Enna, S. J.; The GABA Receptors, 2nd ed.;
1997; p 332.
4. Gupta, S. P. Prog. Drug Res. 1995, 45, 67.
5. Atack, J. R. Curr. Drug Targets CNS Neurol. Disord.
2003, 2, 213.
15. Hadingham, K. L.; Wingrove, P.; Le Bourdelles, B.;
Palmer, K. J.; Ragan, C. I.; Whiting, P. J. Mol. Pharma-
col. 1993, 43, 970.
16. Smith, A. J.; Alder, L.; Silk, J.; Adkins, C.; Fletcher, A. E.;
Scales, T.; Kerby, J.; Marshall, G.; Wafford, K. A.;
Mckernan, R. M.; Atack, J. R. Mol. Pharmacol. 2001, 59,
1108.
17. Brown, N.; Kerby, J.; Bonnert, T. P.; Whiting, P. J.;
Wafford, K. A. Br. J. Pharmacol. 2002, 136, 965.
18. Cruskie, M. P., Jr.; Zoltewicz, J. A.; Abboud, K. A.
J. Org. Chem. 1995, 60, 7491.
6. McKernan, R. M.; Rosahl, T. W.; Reynolds, D. S.; Sur,
C.; Wafford, K. A.; Atack, J. R.; Farrar, S.; Myers, J.;
Cook, G.; Ferris, P.; Garrett, L.; Bristow, L.; Marshall,