8-Fluoroimidazo[1,2-a]pyridine: Synthesis, physicochemical properties and evaluation as a bioisosteric replacement for imidazo[1,2-a]pyrimidine in an allosteric modulator ligand of the GABAA receptor

Article abstract of DOI:10.1016/j.bmcl.2005.12.037

8-Fluoroimidazo[1,2-a]pyridine has been established as a physicochemical mimic of imidazo[1,2-a]pyrimidine, using both in silico and traditional techniques. Furthermore, a novel synthesis of a 3,7-disubstituted-8- fluoroimidazopyridine 3 has been developed and the utility of the physicochemical mimicry has been demonstrated in an in vitro system. Here, the 8-fluoroimidazopyridine ring contained in ligand 3 acts as a bioisosteric replacement for imidazopyrimidine in the GABA_A receptor modulator 2.

Full text of DOI:10.1016/j.bmcl.2005.12.037

Bioorganic & Medicinal Chemistry Letters 16 (2006) 1518–1522
8-Fluoroimidazo[1,2-a]pyridine: Synthesis,
physicochemical properties and evaluation
as a bioisosteric replacement for imidazo[1,2-a]pyrimidine
in an allosteric modulator ligand of the GABAA receptor
Alexander C. Humphries,* Emanuela Gancia, Myra T. Gilligan, Simon Goodacre,
David Hallett, Kevin J. Merchant and Steve R. Thomas
The Neuroscience Research Centre, Merck Sharp and Dohme, Terlings Park, Harlow, Essex, CM20 2QR, UK
Received 1 November 2005; revised 8 December 2005; accepted 11 December 2005
Available online 4 January 2006
Abstract—8-Fluoroimidazo[1,2-a]pyridine has been established as a physicochemical mimic of imidazo[1,2-a]pyrimidine, using both
in silico and traditional techniques. Furthermore, a novel synthesis of a 3,7-disubstituted-8-fluoroimidazopyridine 3 has been devel-
oped and the utility of the physicochemical mimicry has been demonstrated in an in vitro system. Here, the 8-fluoroimidazopyridine
ring contained in ligand 3 acts as a bioisosteric replacement for imidazopyrimidine in the GABAA receptor modulator 2.
ꢀ 2006 Elsevier Ltd. All rights reserved.
Nitrogen-bridgehead fused heterocycles containing an
imidazole ring are a common structural motif in phar-
macologically important molecules, with activities span-
ning a diverse range of targets. Probably the most widely
used heterocyclic system from this group is imidazo[1,2-
a]pyridine, which is contained in marketed drugs such as
N
N
N
NC
O
N
Me
O
N
Me
H
N
Zolpidem
Olprinone
the benzodiazepine agonist Zolpidemꢁ and the PDE 3
1
MeO
Et
N
O
N
inhibitor Olprinoneꢁ,2 as well as other experimental
molecules.3 However, alternative derivatives, such ꢁas
the closely related imidazo[1,2-a]pyrimidine Divaplon ,4
are also prevalent (Fig. 1).
N
Me
Divaplon
During the course of a medicinal chemistry programme
aimed at discovering novel allosteric modulators of the
GABAA receptor, it was established that the imidazo-
pyridine compound 1 (Fig. 2) has good affinity at multi-
ple subtypes of the receptor (Table 1), with a degree of
functional selectivity in vitro.5 Subsequent work demon-
strated that a 10-fold increase in affinity is achieved by
replacement of the 8-position C–H with an azine nitro-
gen atom, giving the analogous imidazo[1,2-a]pyrimi-
dine 2. However, the strategy adopted in the Merck
Figure 1. Some literature examples of imidazo[1,2-a]pyridine-type
drugs.
programme has been to minimise the functional effects
at GABAA a1 (seek an a1 antagonist),6 whilst maximis-
ing positive modulation at GABAA a3 (agonist activi-
ty).7 Since compound 2 still possesses significant
functional activity at the a1 subtype, there remained a
desire to pursue further structural modifications of the
heterocyclic core, with the aim of producing a second
generation lead. Such a compound should retain the
attractive sub-nanomolar activity of 2, but possess lower
functional activity at GABAA a1 and increased func-
tional selectivity. This paper details one of the strategies
adopted.
Keywords: Imidazo[1,2-a]pyridine; Imidazo[1,2-a]pyrimidine; Bioiso-
sterism; Physicochemical; GABAA; Basicity; Lipophilicity.
*
Corresponding author. Tel.: +44 1279 440239; fax: +44 1279
440390; e-mail: alexander_humphries@merck.com
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.12.037

A. C. Humphries et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1518–1522
1519
OH
using semi-empirical calculations on MMFF-minimised
structures.14 The results are displayed visually in Figure
3, with the magnitude of the dipoles being 3.37 D (4),
5.10 D (5) and 4.52 D (6). Consistent with our hypothe-
sis, it is clear that the direction and strength of the dipole
in the 8-fluoroimidazopyridine 6, together with its
overall electrostatic surface, show a much closer resem-
blance to the imidazopyrimidine 5 than to the des-F
compound 4.
8
X
7
N
N
3
NC
1 (X = C-H)
2 (X = N)
3 (X = C-F)
F
Figure 2. Allosteric modulators of the GABAA receptor.
Encouraged by the molecular modelling results, we set
out to synthesise the 8-fluoroimidazopyridines 11 and
12, in order to compare with the known5 compounds
7–10 (Fig. 4).
Table 1. Binding affinities and in vitro efficacy data for 1, 2, and 3 at
GABAA a1 and a3 receptors
Compound
Affinity, Kia (nM)
Efficacy (% CDZ)b
a1
a3
a1
a3
35c
60d
34d
61c
In common with the general synthesis of imidazo[1,2-
a]pyridines, 8-fluoroimidazopyridines can be prepared
by the reaction of a 2-aminopyridine with bromoacetal-
dehyde, used as the diethyl or dimethyl acetal (Scheme
1).11 However, we wished to incorporate an isopropanol
substituent at the 7-position of the final ring system 11
(corresponding to the 4-position of the starting pyridine
13)—giving a substitution pattern which had not been
previously exemplified. To achieve our aim, we envis-
aged using Queguiner’s procedure,15 whereby lithiation
followed by an electrophilic quench can be used to
1
2
3
4.35
0.71
0.20
5.16
0.47
0.32
79c
104c
a Ki values for benzodiazepine sites on stably expressed human
GABAA axb3c2 receptors (x = 1 or 3) in LTK mouse fibroblast cells.
Inhibition curves were carried out using receptors labelled with
[3H]Ro15-1788 at a concentration of twice the Kd. Ki values were
calculated according to the Cheng-Prussof equation. Data shown are
means of three determinations (see Ref. 6).
b In vitro efficacy expressed relative to the maximal response of the full
agonist chlordiazepoxide (CDZ) in stably expressed human GABAA
axb3c2 receptors (x = 1 or 3) in LTK mouse fibroblast cells
(see Ref. 6).
c Measured using a 36Clꢁ flux assay at 1000· Ki (drug) and GABA
EC20. Data shown are means of at least 7 determinations.
d Maximum modulation of the GABA EC20 measured by electro-
physiology. Data shown are means of at least 5 determinations.
The increased affinity resulting from replacement of the
8-position C–H group (in 1) to an azine nitrogen (in 2)
represents a relatively modest difference in binding ener-
gy (ꢀ6 kJ molꢁ1).8 Hence, we speculated that it most
likely resulted from subtle changes to properties such
as the electrostatic surface, the dipole moment and
pKa, which relate to the heterocycle as a whole, rather
than a significantly altered receptor–ligand interaction
in the locality of the 8-position, for example, a hydrogen
bond.9 We surmised that replacement of the same C–H
unit by C–F would achieve a similar perturbation of
electrostatic surface, dipole moment and pKa, based on
some of the known effects of fluorine in bioactive mole-
cules.10 However, incorporation of a C–F unit should be
complementary to using an azine N, as other properties,
including the lipophilicity and localised steric and elec-
tronic interactions, should be altered in quite different
ways. Although a small number of compounds contain-
ing an 8-fluoroimidazopyridine are known in the litera-
ture,11,12 no study has been carried out to compare the
heterocycle’s physicochemical properties with those of
either the parent des-F imidazopyridine or the analo-
gous imidazopyrimidine. To the best of our knowledge,
there are also no examples of its use as a bioisostere for
imidazopyrimidine in any biological system.13 There-
fore, we set out to make this comparison within the con-
text of our known GABAA receptor modulators (Fig. 2).
Figure 3. Dipole moments (A) and electron density surfaces (B) for the
simplified GABAA ligands 4, 5 and 6. Red and blue colourations
indicate high and low electron density, respectively.14
F
R
R
N
R
N
N
N
N
N
N
7 (R = H)
9 (R = H)
11 (R = H)
OH
OH
OH
8 (R =
)
10 (R =
)
12 (R =
)
Our first approach was to model and compare a series of
simplified 3,7-disubstituted compounds (4–6) in silico
Figure 4. Structures 7–12 for physicochemical parameter comparison.

1520
A. C. Humphries et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1518–1522
F
F
F
H
Cl
N
N
a
b
X
N
N
N
OMe
MeO
14
16
11
Scheme 1. Reagents and conditions:11 (a) 2-bromoacetaldehyde
diethylacetal, 24% hydrobromic acid, 100 ꢂC, then room temperature,
EtOH, NaHCO3, then filter, then 13, reflux, 52%.
Scheme 3. Reagents and conditions: (a) 2-aminoacetaldehyde dime-
thylacetal, NaOtBu, BINAP, Pd(OAc)2, THF, reflux, 79%; (b) 24%
hydrobromic acid, 90 ꢂC, then room temperature, IPA, NaHCO3, then
filter, then 50 ꢂC.
install a substituent ortho to a fluorine atom in a pyridyl
system. However, a 2-amino substituent (as in 13) is
incompatible with C-lithiation chemistry and hence,
we adopted an alternative building block, that is, 2-chlo-
ro-3-fluoropyridine 14 (Scheme 2). As expected, 14
undergoes facile ortho-lithiation and an acetone quench
afforded the required derivative 15 in high yield. This
compound as well as the unfunctionalised pyridine 14
were then converted to the corresponding imidazopyri-
dines 12 and 11, respectively, using a novel two-step
sequence commencing with replacement of the 2-chloro
substituent by benzophenone imine under Buchwald
conditions.16 Rather than deprotecting the ‘masked’
aminopyridine in a stepwise fashion, imine hydrolysis
was carried out concomitantly with deprotection of the
bromoacetaldehyde diethyl acetal fragment, using hot
hydrobromic acid.5 On cooling, the solution was neutra-
lised with sodium carbonate and heated again to afford
11 and 12 in satisfactory yield. We also examined
whether it would be possible to functionalise the 7-posi-
tion of the fused heterocycle 11 directly, using chemistry
analogous to functionalising the pyridine 4-position.
However, this strategy proved unsuccessful using any
of Queguiner’s conditions (i.e., lithium diisopropyl-
amide, butyllithium/TMEDA or butyllithium/DABCO
complexes),15 with 3-substituted material being pro-
duced in each case. To try and overcome this, strategies
involving dilithiation or site-blocking of the 3-position
using trimethylsilane were also attempted, but again
without success.
Since this chemistry was first disclosed,17 it has been
established that Buchwald aminations of 3-fluoro-2-
chloropyridine are general for a series of primary alkyl
amines.18 This led us to investigate an alternative ‘re-
versed-synthon’ approach to adding the two-carbon unit
to the pyridine ring (Scheme 3).19 In this case, 14 was
reacted with aminoacetaldehyde dimethylacetal to form
the aminopyridine 16. However, this compound failed to
cyclise to 11 under either acidic or neutral conditions
after acid-catalysed aldehyde deprotection.
Physicochemical descriptors (pKa and logD7.4) for the
three unsubstituted cores (7, 9 and 11), together with
those for the three 7-substituted analogues (8, 10 and
12) were measured and are presented in Table 2. The
pKa values were determined by analysing the UV chro-
mophores during a pH titration (using Sirius SGA tech-
nology) and, for the compounds 7 and 9, the values
obtained are very close to those previously published
in the literature.20,21 Interestingly, in both the unsubsti-
tuted and 7-isopropanol series, it is clear that the C–F
unit (in 11 and 12) alters the pKa in an identical manner
to an azine nitrogen (in 9 and 10), that is, by approxi-
mately two units. LogD7.4 values were determined using
the traditional ‘shake-flask’ method with partitioning
between octanol and a pH7.4 buffer. In this case, 8-fluoro
derivative 11 shows closer similarity to the des-fluoro
imidazopyridine 7, rather than the imidazopyrimidine
9. However, the effect of adding the isopropanol substi-
tuent is interesting, in that, for both 10 and 12, the
logD7.4 value has risen slightly (by 0.3 and 0.2, respec-
tively), as a result of adding the lipophilic group while,
on the other hand, the value for 8 is 0.2 lower than
for the parent 7. We believe this arises from the fact that
Table 2. Physicochemical data for compounds 7–12
R
X
N
N
a
b
Compound
X
R
pKa
LogD7.4
7
8
C–H
C–H
N
–H
–C(OH)Me2
–H
6.9 (6.8)c
7.2
0.8
0.6
9
4.9 (4.8)d
5.4
ꢁ0.2
0.1
10
11
12
N
–C(OH)Me2
–H
–C(OH)Me2
C–F
C–F
4.9
5.4
0.9
1.1
Scheme 2. Reagents and conditions: (a) LDA, THF, ꢁ78 ꢂC, then
˚
acetone (pre-dried over 4 A molecular sieves), to room temperature,
95%; (b) benzophenone imine, Cs2CO3, BINAP (6 mol%), Pd(OAc)2
(4 mol%), PhMe, 95 ꢂC; (c) 2-bromoacetaldehyde diethylacetal, 24%
hydrobromic acid, 90 ꢂC, then room temperature, IPA, NaHCO3, then
filter, then 50 ꢂC, 76% (11, 2 steps), 34% (12, 2 steps); (d) BuLi,
TMEDA or DABCO, ꢁ78 ꢂC, Et2O or THF, then 11, then acetone
a Measured by analysis of the UV chromophore during a pH titration
(Sirius SGA).
b Ratio of concentrations in octanol and pH7.4 aqueous buffer after
shaking for 20 min.
c Literature value (see Ref. 21).
d Literature value (see Ref. 22).
˚
(pre-dried over 4 A molecular sieves), to room temperature.

A. C. Humphries et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1518–1522
1521
in going from 7 to 8, the pKa changes from 6.9 to 7.2,
which is almost identical to the test pH. As a result,
the ratio of protonated to non-protonated material rises
by almost 50%, accounting for the net reduction in the
logD7.4. On the other hand, the pKa changes for the
other two cores (4.9–5.4) represent a very minor change
in the degree of protonation at pH 7.4.
ther optimisation, for example, by small variations of
the 3- and 7-substituents, which are known to influence
functional efficacy.5
In summary, we have demonstrated that 8-fluoroimi-
dazo[1,2-a]pyridine closely resembles imidazo[1,2-a]py-
rimidine in respect of a number of physicochemical
parameters, such as electrostatic surface, dipole moment
and pKa, while, at the same time, it possesses a signifi-
cantly higher logD7.4, making it a complementary
replacement for a C–H unit. Furthermore, the signifi-
cance of the resemblance has been established in a
GABAA system, where 8-fluoroimidazopyridine has
been shown to be a bioisostere of imidazopyrimidine,
with the ligands 3 and 2 both affording ꢀ6 kJ molꢁ1
greater binding energy than the parent imidazopyridine
1. However, the effects of 2 and 3 on in vitro efficacy at
the GABAA a1 and a3 receptor subtypes have been
shown to be different, with 3 producing a smaller
functional effect at receptors containing the a1-subtype,
accompanied by a degree of functional selectivity for
a3-containing receptors. We therefore propose 3 as a
lead for optimisation in the search for a pure ‘a3-ago-
nist’. The results of optimisation studies to achieve this
goal will be published in due course.
In order to determine the significance of the gathered
physicochemical information within the biological sys-
tem of interest (GABAA), we needed to functionalise
the 3-position to afford compound 3. It is known that
imidazopyridine-type structures can be readily arylated
at the 3-position via electrophilic bromination, allowing
Suzuki or Stille couplings,5 or by direct Heck coupling
of the unbrominated heterocycle.22 Unsurprisingly, both
of these methods worked well for the preparation of 3
(Scheme 4), although the direct Heck coupling proved
to be considerably more efficient as there was no need
for temporary alcohol protection.
8-Fluoroimidazopyridine 3 exhibits high affinity at both
GABAA a1 and a3 receptors with Ki values of 0.20 and
0.32 nM, respectively (Table 1).23 Comparison of this
data with those previously established for the imidazo-
pyridine 1 and imidazopyrimidine 2 demonstrates that
replacement of the C–H unit by a C–F yields the same
6 kJ molꢁ1 increase in binding energy as replacement
of the C–H with an azine N.24 Hence, the replacements
can be correctly termed bioisosteric, at least within this
biological system.11 However, the functional effects of
the C–F and N replacements are not identical, with
compound 3 giving a significantly lower in vitro efficacy
at GABAA a1-receptors (maximal response = 34% of
CDZ, cf. 60% for 2).23 The compound also shows rea-
sonable functional selectivity for the GABAA a3 subtype
(maximal response = 104% of CDZ, cf. 79% for 2).
These data makes 3 a promising starting point for fur-
Acknowledgments
We thank Remya Muralikuttan for assistance in obtain-
ing logD7.4 values, Alison Smith for Ki determinations
and George Marshall for determining the in vitro effica-
cy of compound 3.
References and notes
1. Harrison, T. S.; Keating, G. M. CNS Drugs 2005, 19, 65.
2. Ueda, T.; Mizusgige, K.; Yukiiri, K.; Takahashi, T.;
Kohno, M. Cerebrovasc. Dis. 2003, 16, 396.
3. Zuang, Z.-P.; Kung, M.-P.; Wilson, A.; Lee, C.-W.; Plo¨ssl,
K.; Hou, C.; Holzman, D. M.; Kung, H. F. J. Med. Chem.
2003, 46, 237; Colletti, S. L.; Frie, J. L.; Dixon, E. C.;
Singh, S. B.; Choi, B. K.; Scapin, G.; Fitzgerald, C. E.;
Kumar, S.; Nichols, E. A.; O’Keefe, S. J.; O’Neill, E. A.;
Porter, G.; Samual, K.; Schmatz, D. M.; Scwartz, C. D.;
Shoop, W. L.; Thompson, C. M.; Thompson, J. E.; Wang,
R.; Woods, A.; Zaller, D. M.; Doherty, J. B. J. Med.
Chem. 2003, 46, 349.
OH
F
TESO
F
N
N
a, b
N
N
c
Br
12
e
4. Clements-Jewery, S.; Danswan, G.; Gardner, C. R.;
Matharu, S. S.; Murdoch, R.; Tully, W. R.; Westwood,
W. J. Med. Chem. 1988, 31, 1220.
5. Goodacre, S.; Street, L. J.; Hallett, D. J.; Kelly, S.;
Blackaby, W. P.; Lewis, R.; McKernan, R. M.; Stanley, J.;
Reynolds, D.; Sohal, B.; Cook, S.; Pike, A.; Wafford, K.
A.; Marshall, G.; Castro, J. L.; Atack, J. R. J. Med. Chem.
2006, 49, 35.
OH
F
TESO
F
N
N
d
N
N
NC
NC
3
6. GABAA a1 agonist activity is generally accepted to be
responsible for sedation in vivo: 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, G.; Macauley, A.;
Brown, N.; Howell, O.; Moore, K. W.; Carling, R. W.;
Street, L. J.; Castro, J. L.; Ragan, I.; Dawson, G. R.;
Whiting, P. J. Nat. Neurosci. 2000, 3, 587.
F
F
Scheme 4. Reagents and conditions: (a) Br2, KBr, NaOAc, MeOH,
69%; (b) TESOTf, EtNi-Pr2, CH2Cl2, ꢁ78 ꢂC, to room temperature,
83%; (c) 20-fluoro-50-(pinacolataborolan-2-yl)biphenyl-2-carbonitrile,5
Pd(PPh3)4 (5 mol%), 2 M Na2CO3, THF, 75 ꢂC, 60%; (d) catalytic
concd HCl (aq), EtOH, >95%; (e) 50-bromo-20-fluorobiphenyl-2-
5
carbonitrile, Cs2CO3, Pd(PPh3)4 (6 mol%), 1,4-dioxane, 35%.

1522
A. C. Humphries et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1518–1522
the software program Spartanꢁ (version 04; Wavefunction
Inc.).
7. The significance of GABAA a3 activity in vivo is contro-
versial, although researchers at Merck believe it to be
responsible for anxiolytic effects: Lo¨w, K.; Crestani, F.;
Keist, R.; Benke, D.; Brunig, I.; Benson, J. A.; Fritschy,
15. Marsais, F.; Queguiner, G. Tetrahedron 1983, 39, 2009.
16. Wolf, J. P.; Ahman, J.; Sadighi, J. P.; Singer, R. A.;
Buchwald, S. L. Tetrahedron Lett. 1997, 38, 6367.
17. Goodacre, S. C.; Hallett, D. J.; Humphries, A. C.; Jones,
P.; Kelly, S.; Merchant, K. J.; Moore, K. W.; Reader, M.
PCT Int. Appl. WO 03/099816; Chem. Abstr. 2003, 140,
16738.
¨
J.-M.; Rulick, T.; Bluethmann, H.; Mo¨hler, H.; Rudolph,
¨
U. Science 2000, 290, 131; Reynolds, D. S.; McKernan, R.
M.; Dawson, G. Trends. Pharm. Sci. 2001, 22, 402.
8. Under physiological conditions (T = 310 K), DG = 5.85
logK.
9. Andrews, P. In The Practice of Medicinal Chemistry;
Wermuth, C. G., Ed.; Academic Press: London, 2003; pp
327–339.
18. Munson, P. M.; Thompson, W. J. Synthetic Commun.
2004, 34, 759.
19. Normally, the carbonyl unit becomes linked to the amino
group.
10. Edwards, P. N. In Organofluorine Chemistry: Principles
and Commercial Applications; Banks, R. E., Smart, B. E.,
Tatlow, J. C., Eds.; Plenum: New York, 1994; pp 501–541.
11. Nishimura, T.; Yoshimura, Y.; Yamaoka, M. J. Antibiot.
1991, 44, 1371; Yamanaka, M.; Miyake, K.; Suda, S.;
Ohhara, H.; Ogawa, T. Chem. Pharm. Bull. 1991, 39, 1556.
12. Podhorez, D. E. J. Heterocycl. Chem. 1991, 28, 971.
13. ‘Bioisosteres are groups or molecules which have chemical
and physical similarities producing broadly similar bio-
logical effects’: Thornber, C. W. Chem. Soc. Rev. 1979, 8,
563.
20. Li, W.; Nelson, D. P.; Jensen, M. S.; Hoerrner, R. S.; Javadi,
G. J.; Cai, D.; Larson, R. D. Org. Lett. 2003, 5, 4835.
21. Starr, D. F.; Bulbrook, H.; Hixon, R. M. J. Am. Chem.
Soc 1932, 54, 3971; Catalan, J.; Elguero, J. J. Heterocycl.
Chem. 1984, 21, 269.
22. Armarego, W. L. F. J. Chem. Soc. 1965, 2778.
23. Affinity and in vitro efficacy values were determined as
described in Ref. 6. Affinity values are the mean of three
determinations; efficacy values are the mean of five
determinations.
14. Structures were built and energy minimised using the
Merck Molecular Mechanics Forcefield (MMFF). The
electron density maps and dipoles are extracted from an
AM1 semi-empirical calculation. Images are generated by
24. This is the basis of the ‘Anchor Principal’, whereby with
all else being invariant, the differences in energy are
assumed to arise from the single point change only: Page,
M. I. Angew. Chem. Int. Ed. Engl. 1977, 16, 449.